UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Molecular mechanism of cyclic nucleotide action on HCN pacemaker channels Ng, Leo Chun Ting 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_may_ng_leo.pdf [ 7.98MB ]
Metadata
JSON: 24-1.0229565.json
JSON-LD: 24-1.0229565-ld.json
RDF/XML (Pretty): 24-1.0229565-rdf.xml
RDF/JSON: 24-1.0229565-rdf.json
Turtle: 24-1.0229565-turtle.txt
N-Triples: 24-1.0229565-rdf-ntriples.txt
Original Record: 24-1.0229565-source.json
Full Text
24-1.0229565-fulltext.txt
Citation
24-1.0229565.ris

Full Text

MOLECULAR MECHANISM OF CYCLIC NUCLEOTIDE ACTION ON  HCN PACEMAKER CHANNELS by  Leo Chun Ting Ng  B.Sc. (Biochemistry), The University of British Columbia, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Cell and Developmental Biology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2016  © Leo Chun Ting Ng, 2016 ii  Abstract  HCN pacemaker channels generate the “funny” current that is responsible for initiation and regulation of electrical activity in the heart and the nervous system. Activated near resting membrane potentials, the channel conducts a net inward current that maintains action potential. The direct binding of cyclic AMP upon adrenaline release also enhances channel opening, leading to an increased frequency of action potential. The cytoplasmic cAMP-binding domain is coupled to the C-linker found between the binding domain and the transmembrane domain, subsequently triggering the opening of the pore. Despite extensive research, the mechanism that links the binding event to channel facilitation is still unclear. In this thesis, we looked into the different aspects of the missing link. In addition to basic understanding, the significance of learning about HCN channel modulation by cAMP is that there may be therapeutic advantage of controlling heart rate via drug interaction with the cyclic nucleotide binding pocket. Using the isolated C-linker/binding domain, we pinpointed residues in the binding pocket that contribute to strong affinity and ligand specificity by single residue alanine-scanning and isothermal titration calorimetry for measurement of binding affinity. By comparing our binding data to functional measurements of potency in full-length channels containing the same single substitutions, we found that two residues, L633 and I636, reduced potency more severely than affinity when mutated, and proposed that these residues are involved in a post-binding transition event. We also found that two partial agonists, cCMP and cIMP, bound to the canonical site, but failed to fully promote tetrameric gating ring which is found on the inner side of the pore and hypothesized to facilitate its opening. We proposed that the weakened interactions between the partial agonists and the C-helix of the binding domain limit the formation of the gating ring and iii  led to reduced facilitation of opening in the full-length channel. Finally, we elucidated the mechanism behind two disease-associated mutations found in the cytosolic C-linker/binding domain portion of the HCN2 and HCN4 channels to gain a better understanding of how they influence cAMP binding and channel opening, and cause epilepsy and profound bradycardia, respectively.   iv  Preface  This thesis contains work that is in preparation for publication in peer-reviewed journals. I was the lead investigator on these projects, summarized and included as part of the thesis in Chapter Two, Three, and Four. Chapter Two and Three are presented in the original form, except for elaborated experimental procedures in Chapter Two.  Chapter 2: Molecular determinants of cyclic nucleotide binding and coupling of binding to HCN2 channel opening Leo Ng, Meiying Zhuang, Filip Van Petegem, Eric Accili I was responsible for designing and performing experiments, as well as data analysis and figures preparation. M. Zhuang was involved in protein purification and in obtaining and analyzing ITC data. E. Accili and I prepared the manuscript, and M. Zhuang and F. Van Petegem edited the draft. F. Van Petegem provided guidance in developing the experiment. E. Accili was the supervisory author and oversaw the entire project. The manuscript will be sent.  Chapter 3: Oligomeric interactions by C-terminal domains promote cyclic nucleotide facilitation of opening of the HCN2 channel Leo Ng, Igor Putrenko, Victoria Baronas, Filip Van Petegem, Eric Accili E. Accili and I were responsible for all major area of project formation, in designing the experiments and composing the manuscript. I. Putrenko assisted in the design of the functional assays and collected all the patch-clamping data, and V. Baronas collected data for binding of cUMP using isothermal titration calorimetry; both of them also contributed to editing the v  manuscript. I collected data for the remaining ligands and solved the crystal structures. I also analyzed and prepared the final figures. F. Van Petegem was involved with providing expert opinions throughout the project and with manuscript edits. The manuscript will be sent out shortly. Acknowledgement: S. Wong and B. Gardill provided feedback and validated the five crystal structures. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357.  Chapter 4: Two disease mutations in the carboxy-terminal region impact ligand binding and effect Leo Ng, Sarah Chow, Filip Van Petegem, Eric Accili Along with E. Accili and S. Chow, we developed and designed the project. S. Chow was involved in collecting data and creating figures for the thermal stability assay, and obtaining the crystals and collecting the raw crystallographic data. I collected the ITC and DLS data, and solved the crystal structure for the epilepsy-associated mutation, and made the final figures. The same set of ITC and DLS data was used in Chapter Three of S. Chow’s PhD dissertation in 2013.  I was the lead investigator for the bradycardia mutation project, responsible for designing and performing the experiments, analyzing data, and producing figures. Data from this chapter will be used in preparing a manuscript. vi  Table of Contents  Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iv Table of Contents ......................................................................................................................... vi List of Tables ................................................................................................................................xv List of Figures ............................................................................................................................. xvi List of Abbreviations ................................................................................................................. xix Acknowledgements ................................................................................................................... xxii Dedication ................................................................................................................................. xxiii Chapter 1: Introduction ................................................................................................................1 1.1 Overview ......................................................................................................................... 1 1.2 The hyperpolarization-activated mixed cation or “funny” current ................................. 2 1.2.1 History of the hyperpolarization-activated mixed-cation current ............................... 2 1.2.2 Biophysical properties of If / Ih ................................................................................... 3 1.2.3 Modulation of If by the autonomic nervous system .................................................... 4 1.2.4 Role of If in cardiac pacemaking ................................................................................ 6 1.3 Hyperpolarization-activated cyclic nucleotide-gated “HCN” channels underlie the "funny" current ............................................................................................................................ 8 1.3.1 Cloning the HCN genes .............................................................................................. 8 1.3.2 Structure and function of HCN channels .................................................................... 9 1.3.2.1 Responses to voltage and voltage-sensing elements of the HCN channel .......... 9 1.3.2.2 The HCN channel pore and cation permeation ................................................. 13 vii  1.3.2.3 Facilitation of channel opening by cAMP and the structural components involved ........................................................................................................................... 14 1.3.2.4 Heteromeric assembly of HCN subunits ........................................................... 15 1.3.2.5 Regulation by other factors ............................................................................... 16 1.3.2.6 Cellular trafficking of the HCN channel ........................................................... 18 1.3.3 Tissue expression of the four isoforms ..................................................................... 19 1.3.4 Genetic disruption of specific HCN genes demonstrates the physiological role of HCN channels ....................................................................................................................... 19 1.4 HCN channel in human disease .................................................................................... 20 1.4.1 Disrupting HCN expression leads to heart failure .................................................... 20 1.4.2 HCN inhibition as treatment for cardiac disease ...................................................... 21 1.4.3 Disease-associated mutations in HCN channels ....................................................... 22 1.4.3.1 Disease-associated mutations in the HCN2 channel isoform ........................... 22 1.4.3.2 Disease-associated mutations in the HCN1 channel isoform ........................... 24 1.4.3.3 Disease-associated mutations in the HCN4 channel isoform ........................... 25 1.5 Regulation of HCN function by cyclic nucleotide binding .......................................... 30 1.5.1 Functional contribution of C-linker and CNBD ....................................................... 30 1.5.1.1 The C-linker couples to the CNBD and transduces the signal to the pore ....... 30 1.5.1.2 The CNBD interacts with cAMP directly and influences channel activity ...... 31 1.5.2 Homologous proteins containing an HCN-like CNBD ............................................ 33 1.5.3 Structure of the C-terminal domains in HCN channels ............................................ 39 1.5.3.1 First crystal structure reveals the binding pocket ............................................. 39 1.5.3.2 All published HCN carboxy-terminus are structurally similar ......................... 41 viii  1.5.3.3 Cyclic GMP binds to the HCN2 isoform in the syn configuration ................... 43 1.5.3.4 Selectivity for cGMP ........................................................................................ 44 1.5.4 Formation of the cytosolic gating ring ...................................................................... 45 1.5.5 Dynamic movement of the carboxy terminal domains ............................................. 46 1.5.5.1 C-helix of the CNBD moves upon ligand binding............................................ 46 1.5.5.2 Subsequent C-linker rearrangement .................................................................. 49 1.5.6 Functional assessment of ligand binding to the CNBD ............................................ 51 1.5.6.1 Residues important for binding interactions ..................................................... 51 1.5.6.2 Cyclic AMP binds to the tetrameric form of the C-linker and CNBD with negative cooperativity ....................................................................................................... 55 1.5.6.3 Contributions of individual binding sites to the effect of cAMP ...................... 56 1.6 Summary and rationale for my studies ......................................................................... 57 Chapter 2: Molecular Determinants of Cyclic Nucleotide Binding and Coupling of Binding to HCN2 Channel Opening .........................................................................................................62 2.1 Introduction ................................................................................................................... 62 2.2 Experimental procedures .............................................................................................. 64 2.2.1 Molecular biology and cloning ................................................................................. 64 2.2.2 Protein purification ................................................................................................... 66 2.2.3 Ligand preparation .................................................................................................... 67 2.2.4 Isothermal titration calorimetry ................................................................................ 67 2.2.5 Dynamic light scattering ........................................................................................... 69 2.3 Results ........................................................................................................................... 70 ix  2.3.1 Cyclic GMP binds to the HCN2 C-linker/CNBD with negative cooperativity and with lower affinity than cAMP ............................................................................................. 70 2.3.2 Single alanine substitutions of CNBD residues close to cAMP and cGMP mainly reduce binding affinity .......................................................................................................... 74 2.3.3 Isoleucine 636 in the C-helix confers some selectivity for cAMP binding .............. 77 2.3.4 Comparing cyclic nucleotide potency in the full-length channel with binding affinity to the C-linker/CNBD ........................................................................................................... 79 2.3.5 Large differences in the pattern of thermodynamics accompany large mismatches between cyclic nucleotide binding affinity and potency ....................................................... 83 2.4 Discussion ..................................................................................................................... 86 Chapter 3: Oligomeric Interactions by C-terminal Domains Promote Cyclic Nucleotide Facilitation of Opening of the HCN2 Channel ..........................................................................93 3.1 Introduction ................................................................................................................... 93 3.2 Experimental procedures .............................................................................................. 95 3.2.1 HCN protein purification and mutagenesis............................................................... 95 3.2.2 Ligand preparation .................................................................................................... 96 3.2.3 Ligand-induced tetramerization by dynamic light scattering (DLS) ........................ 96 3.2.4 Direct measurements of cyclic nucleotide binding by isothermal titration calorimetry (ITC) ................................................................................................................................... 97 3.2.5 Whole-cell patch clamp electrophysiology ............................................................... 97 3.2.6 Crystallization of HCN2-ligand complexes .............................................................. 98 3.3 Results ........................................................................................................................... 99 x  3.3.1 Cyclic CMP and cIMP do not promote self-association of the HCN2 C-linker/CNBD when measured by dynamic light scattering .................................................. 99 3.3.2 Cyclic CMP and cIMP bind to the HCN2 C-linker/CNBD without negative cooperativity ....................................................................................................................... 102 3.3.3 Cyclic CMP and cIMP are less effective facilitators of HCN2 opening than the other cyclic nucleotides examined ............................................................................................... 109 3.3.4 Cyclic UMP makes contacts with the HCN2 C-helix that differ from those made by cCMP ................................................................................................................................. 115 3.3.5 Crystal structures show that cAMP-cGMP intermediates bind in the anti configuration and identify unique contacts made by cIMP with the CNBD ...................... 124 3.4 Discussion ................................................................................................................... 128 Chapter 4: Two Disease Mutations in the Carboxy-Terminal Region Impact Ligand Binding and Effect .....................................................................................................................131 4.1 Introduction ................................................................................................................. 131 4.2 Experimental procedures ............................................................................................ 134 4.2.1 Cloning, site-directed mutagenesis, expression and purification ............................ 134 4.2.2 Crystallization, data collection and structure determination ................................... 135 4.2.3 Isothermal titration calorimetry (ITC) .................................................................... 136 4.2.4 Dynamic light scattering (DLS) .............................................................................. 136 4.2.5 Thermal melt analysis ............................................................................................. 137 4.3 Results: The epilepsy-associated mutation E488K in the HCN2 channel .................. 138 4.3.1 Modified interactions originating at the E488K residue reduce self-association of the mutant HCN2 C-linker/CNBD ........................................................................................... 138 xi  4.3.2 Modified interactions originating at the E488K residue eliminate negatively cooperative binding of cAMP to the mutant HCN2 C-linker/CNBD ................................. 140 4.3.3 The crystal structure of the epilepsy mutant exhibits an abolished intersubunit interaction ........................................................................................................................... 142 4.3.4 The hydrogen bond acceptor of the intersubunit interaction also affects tetramerization and binding ................................................................................................ 145 4.3.5 The E488K mutation does not impact the stability of the C-linker/CNBD ............ 145 4.4 Results: The S672R bradycardia-associated mutation in the HCN4 channel ............. 147 4.4.1 Location of the S672R mutation in the HCN4 C-linker/CNBD ............................. 147 4.4.2 Cyclic AMP and cGMP promote oligomerization of both the wild type and mutant HCN4 C-linker/CNBD........................................................................................................ 148 4.4.3 Cyclic AMP and cGMP bind to the HCN4 C-linker/CNBD with negative cooperativity and lower affinity than to the wild type HCN4 C-linker/CNBD .................. 150 4.4.4 The thermal stability is equal between the wild type and mutant HCN4 C-linker/CNBD ....................................................................................................................... 152 4.5 Discussion ................................................................................................................... 154 4.5.1 The impact of the epilepsy-associated mutation HCN2 E488K on the C-linker/CNBD ....................................................................................................................... 154 4.5.2 Impact of the bradycardia-associated mutation HCN4 S672R on the C-linker/CNBD   ................................................................................................................................. 157 4.5.3 Concluding remarks ................................................................................................ 159 Chapter 5: Discussion ................................................................................................................160 5.1 Overview ..................................................................................................................... 160 xii  5.2 Chapter summaries ...................................................................................................... 161 5.3 Molecular determinants of ligand binding in the CNBD ............................................ 163 5.3.1 Ligand determinants of anti and syn configuration in the CNBD .......................... 163 5.3.2 Interactions between cyclic nucleotide and HCN2 binding domain are identified by comparing results from alanine substitutions and analogues .............................................. 166 5.3.2.1 The PBC contributes to the strength of binding as well as potency ............... 166 5.3.2.2 Re-positioning of R635 assists cUMP and cIMP binding .............................. 167 5.3.2.3 Interaction with carbonyl oxygen of R632 has a small contribution to affinity ...   ......................................................................................................................... 168 5.3.3 A CNBD salt bridge reduces stability of the HCN2 C-terminal domain ................ 169 5.4 Gating model ............................................................................................................... 170 5.4.1 Equilibrium between bound and unbound states .................................................... 171 5.4.2 Binding to CNBD promotes activation of the C-linker .......................................... 172 5.4.3 Equilibrium between active and resting C-linker ................................................... 173 5.4.4 A C-linker gating ring promotes the pore opening ................................................. 174 5.5 Physiological relevance of the data in this thesis ........................................................ 176 5.5.1 Cellular level of cAMP and cGMP ......................................................................... 176 5.5.2 Fluctuation in cAMP/cGMP levels ......................................................................... 177 5.5.3 Physiological relevance of other cyclic nucleotides ............................................... 179 5.6 Limitations to the methods.......................................................................................... 180 5.6.1 Advantages and disadvantages of using the soluble C-linker/CNBD .................... 180 5.6.2 X-ray crystallography does not capture the active C-linker ................................... 182 xiii  5.6.3 Biochemical and cellular experiments in the thesis do not fully mimic the cellular environment of heart cell or neuron .................................................................................... 183 5.7 Future directions ......................................................................................................... 184 5.7.1 Structural approach for studying the full-length construct ..................................... 184 5.7.1.1 Resolving the two gating ring theories ........................................................... 185 5.7.2 Fitting our data in a mathematical model ............................................................... 187 5.7.3 Screening drugs using dynamic light scattering ..................................................... 188 5.7.4 Identifying isoform-specific agonists and antagonists for therapy and correcting dysfunctional mutant channels ............................................................................................ 188 Bibliography ...............................................................................................................................190 Appendices ..................................................................................................................................215 Appendix 1 A schematic diagram of the protein used in the thesis ........................................ 215 Appendix 2 Sets of forward and reverse primer for performing site-directed mutagenesis ... 216 Appendix 3 All nine mutations from Chapter Two allow ligand-induced oligomerization ... 217 Appendix 4 Cyclic AMP and cGMP have extremely low affinity after introduction of E582A and R632A .............................................................................................................................. 219 Appendix 5 Effect of cyclic nucleotides from Chapter Three on oligomerization of the HCN2 C-linker/CNBD as function of ligand concentration .............................................................. 220 Appendix 6 Omit maps of the five co-crystal structures in Chapter Three ............................ 221 Appendix 7 Certain intersubunit interactions abolished by mutation shows reduced ligand-induced oligomerization .......................................................................................................... 222 Appendix 8 Characterization of the effects of analogues on the HCN2 binding domain ....... 225 Appendix 9 Mutations in HCN4 binds responds to ligand normally ...................................... 230 xiv  Appendix 10 The Drosophila isoform C-terminus bind with one binding ............................. 232 Appendix 11 Compilation of ITC data from cAMP binding to various HCN2 mutations ..... 234 Appendix 12 Compilation of ITC data from cGMP binding to various HCN2 mutations ..... 237 Appendix 13 Compilation of ITC data from cAMP or cGMP binding to wild type or mutated HCN4 C-linker/CNBD............................................................................................................ 239 Appendix 14 Compilation of ITC data from cAMP or cGMP in the Drosophila isoform ..... 240 Appendix 15 : Compilation of ITC data from various ligands binding to HCN2 C-linker/CNBD ........................................................................................................................... 241  xv  List of Tables  Table 1.1 CNBD-containing proteins. .......................................................................................... 39 Table 3.1 Values for binding affinity and biophysical parameters describing voltage-dependent activation with or without ligand. ............................................................................................... 111 Table 3.2 Values for half activation voltage and fraction of channel activation at three voltages...................................................................................................................................................... 113 Table 3.3 Data collection and refinement statistics for mHCN2 C-linker/CNBD co-crystallized with five different ligands presented in the chapter. ................................................................... 117 Table 3.4 The polar and non-polar interactions between the HCN2 binding pocket and ligands...................................................................................................................................................... 124 Table 4.1 Data collection and refinement statistics for E488K mHCN2 C-linker/CNBD co-crystallized with cAMP............................................................................................................... 144 Table 5.1 Configuration of cAMP and cGMP in solution and in binding pockets of three proteins. ....................................................................................................................................... 165  xvi  List of Figures  Figure 1.1 Effect of isoprenaline and acetylcholine on action potential of the SAN. .................... 8 Figure 1.2 The topology of the HCN channel subunit. ................................................................. 13 Figure 1.3 Shifts in voltage dependence induced by cAMP. ........................................................ 16 Figure 1.4 The positions of the disease mutations found in human HCN1, 2, and 4 isoforms. ... 25 Figure 1.5 The multiple sequence alignment of proteins containing a CNBD. ............................ 34 Figure 1.6 Sequence alignment of the C-terminal region in the four HCN human isoforms. ...... 41 Figure 1.7 Structural similarities between the C-linker/CNBD segment in solution. .................. 43 Figure 2.1 Location of residues of the cyclic nucleotide binding region that lie near to the ligand and were chosen for mutagenesis ................................................................................................. 64 Figure 2.2 Cyclic AMP and cGMP bind to the HCN2 C-linker/CNBD with a similar pattern of thermodynamics but with different affinities. ............................................................................... 72 Figure 2.3 Single point alanine substitutions produce mild to moderate effects on the heat released upon cAMP binding. ....................................................................................................... 74 Figure 2.4 Single point alanine substitutions produce mild to moderate effects on the heat released upon cGMP binding. ....................................................................................................... 77 Figure 2.5 Opposite effects of aspartate substitution of isoleucine 636 in the C-helix of the HCN2 C-linker/CNBD on binding affinity of cAMP and cGMP. ........................................................... 78 Figure 2.6 Most values for binding affinity of cAMP and cGMP to the HCN2 C-linker/CNBD are correlated with their potency in the full-length channel. ........................................................ 80 Figure 2.7 The thermodynamics of cAMP and cGMP binding to the wild type and mutant HCN2 C-linker/CNBD. ............................................................................................................................ 86 xvii  Figure 2.8 Impact of residues on cAMP and cGMP binding affinity and potency mapped onto the structure of the HCN2 C-linker/CNBD. ....................................................................................... 90 Figure 3.1 Oligomerization of the HCN2 C-linker/CNBD is promoted by cAMP, cGMP and a sub-set of structurally related analogues. .................................................................................... 101 Figure 3.2 The cyclic purine nucleotides cAMP and cGMP bind to the HCN2 C-linker/CNBD with negative cooperativity. ........................................................................................................ 104 Figure 3.3 The cyclic pyrimidine nucleotide cUMP, but not cCMP,  binds to the HCN2 C-linker/CNBD with negative cooperativity. ................................................................................. 107 Figure 3.4 The cyclic purine nucleotides cPUMP and 2-NH2-cPUMP but not cIMP bind to the HCN2 C-linker/CNBD with negative cooperativity. .................................................................. 109 Figure 3.5 Cyclic AMP and analogues variably shift the HCN2 Ih activation curve to less negative potentials. ..................................................................................................................... 114 Figure 3.6 Crystal structures of five analogues bound to the HCN2 C-linker/CNBD reveal their ligand-binding configuration and important interactions. .......................................................... 122 Figure 3.7 Effect of cUMP on the binding and oligomerization R635A mutation in HCN2 C-linker/CNBD. .............................................................................................................................. 126 Figure 4.1 Self-association and ligand-induced oligomerization are eliminated by mutations that abolish the E488-Y459 intersubunit interaction. ........................................................................ 139 Figure 4.2 Negative cooperativity is eliminated when E488-Y459 interaction is abolished. ..... 141 Figure 4.3 Crystal structure of E488K-HCN2 C-linker/CNBD in the presence of cAMP. ........ 143 Figure 4.4 Thermal stability is not greatly modified by E488K mutations in the C-linker or by the binding of cAMP. ........................................................................................................................ 146 Figure 4.5 Location of S672 in HCN4 C-linker/CNBD. ............................................................ 148 xviii  Figure 4.6 cAMP and cGMP promote oligomerization of the mutant and wild type HCN4 C-linker/CNBD. .............................................................................................................................. 149 Figure 4.7 cAMP and cGMP bind to the mutant C-linker/CNBD with negative cooperativity but lower binding affinity than to the wild type. ............................................................................... 152 Figure 4.8 Thermofluor analysis shows that the mutant and wild type HCN4 C-termini have similar thermal stability. ............................................................................................................. 153 Figure 5.1 The proposed gating model during cAMP modulation. ............................................ 176    xix  List of Abbreviations  Amino Acid One Letter Code  A  Ala  Alanine  C  Cys  Cysteine  D  Asp Aspartate  E  Glu  Glutamate F   Phe  Phenylalanine  G  Gly  Glycine  H  His  Histidine  I  Ile  Isoleucine  K   Lys  Lysine  L  Leu  Leucine  M  Met Methionine  N  Asn  Asparagine  P  Pro  Proline  Q  Gln  Glutamine  R  Arg  Arginine  S  Ser  Serine T  Thr  Threonine V  Val  Valine  W  Trp  Tryptophan   ACh  acetylcholine AF  atrial fibrillation AFM  atomic force microscopy AMIH  HCN homologue from Apis mellifera APS  Advanced Photon Source ATP  adenosine-5’ -triphosphate BLAST basic local alignment search tool βME  beta-mercaptoethanol bpm  beats per minute C-linker carboxy-linker cAMP  adenosine-3',5'- cyclic  monophosphate CAP  catabolite gene activator protein  cCMP  cytidine-3’, 5’- cyclic monophosphate CCP4  collaborative computational project number 4 xx  cGMP  guanosine-3’, 5'-cyclic monophosphate CHO  Chinese hamster ovary (cell) cIMP  inosine-3’, 5’- cyclic monophosphate CNBD  cyclic nucleotide binding domain CNG  cyclic nucleotide-gated (channel) Coot  crystallographic object-oriented toolkit cPuMP purine riboside-3’, 5’- cyclic monophosphate cTMP  thymidine- 3’, 5’- cyclic monophosphate cUMP  uridine-3’, 5’- cyclic monophosphate DEER  double electron-electron resonance DEP  dishevelled, EGL-10 and pleckstrin (domain) ΔG  Gibb’s free energy ΔH  enthalpy DLS  dynamic light scattering DMIH  HCN homologue from Drosophila melanogaster DMSO  dimethyl sulfoxide dNTP  deoxy-nucleoside triphosphate ΔS  entropy EAG  ether à-go-go  EC50  effective concentration (K1/2) EDTA  ethylenediaminetetraacetic acid ELK   ether-à-go-go-Like  Epac  exchange factor directly activated by cAMP EST  expressed sequence tag F  Faraday constant (96, 485 C/mol) FRET   fluorescence resonance energy transfer GEF  guanine nucleotide exchange factors GFP  green fluorescent protein HCN  hyperpolarization-activated cyclic nucleotide-gated (channel) HEK  human embryonic kidney (cell) HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid hERG  (human) ether-à-go-go-related gene HMT  His6-MBP-TEV protease cleave site (tag) HPLC  high performance liquid chromatography i.  residue number If  funny current (Ih in the heart)  Ih  hyperpolarization-activated current ITC  isothermal titration calorimetry k  slope factor xxi  K1/2  concentration of substrate that produces a half maximal shift (EC50) KCNH  potassium channel, voltage gated eag related subfamily H Kd  dissociation constant Kv  voltage-gated potassium (channel) LQTS  long QT syndrome LVNC  left ventricular noncompaction cardiomyopathy MBP  maltose binding protein MD  molecular dynamics MES  2-(N-morpholino)ethanesulfonic acid MlotiK potassium channel from Mesorhizobium loti MthK  potassium channel from Methanobacterium thermoautotrophicum MWCO molecular weight cut off N3A  N-terminal helix, an eight residue loop and the A-helix NO  nitric oxide OD  optical density PBC  phosphate binding cassette PCR  polymerase chain reaction PHENIX python-based hierarchical environment for integrated xtallography PIP2  phosphatidylinositol 4,5-bisphosphate PKA  cAMP-dependent protein kinase (protein kinase A) PKG  cGMP-dependent protein kinase  (protein kinase G) rpm  revolutions per minute RT-PCR real-time PCR SAN  sinoatrial node SCAM  substituted cysteine accessibility method s.e.m.  standard error of mean SpIH  HCN homolog from testis of Strongylocentrotus purpuratus Src  sarcoma (tyrosine kinase)  SthK  potassium channel from Spirochaeta thermophila  TEV  tobacco etch virus Tm  mid-unfolding temperature TRIP8b tetratricopeptide repeat-containing Rab8b–interacting protein V1/2  mid-point voltage of activation fitted by Boltzmann function Vmax  maximal voltage shift at saturating cNMP concentrations   xxii  Acknowledgements  The journey to a PhD has not been an easy one, but I am extremely grateful for the people around me who have made the trip more enjoyable. Not only did I obtain a degree and lab experience, but I also developed interpersonal, leadership, and organizational skills in the process.  I'd like to first give my warmest gratitude to Dr. Eric Accili, from adopting me into the lab, persuading me to transfer, and helping me finish the degree. There were so many times when I would whine about the projects or the stress, and he would always reassure of my progress and my ability. Dr. Sarah Chow, a former PhD in the lab, also gave me lots of encouragements as a mentor, entrusted me with a great project and prepared me to fill her big shoe. A big shoutout goes to AC, AH, AW, JX, KL, MZ, VB, and VC, or my trainees in the lab collectively known as the Accili Army, who provided some of the data in my thesis and kept the lab youthful. I am also thankful for the lab's friendly neighbour, Dr. Filip van Petegem and the lab members for including me in their lab events and for engaging with me many intellectual conversations (and silly chitchats).  I'd like to thank my friends for being supportive and pushing me towards the goal of graduation, and for tolerating with my busy schedules. Finally, I would not have gone this far without my family. Even though they are not science-people, they are awesome at taking care and spoiling me, so that I can solely focus on work. xxiii  Dedication     To Terrance, my nephew.  Here's your bed-time storybook in ten years.1  Chapter 1: Introduction  1.1 Overview Heart rate regulation is critical and malfunction in cardiac pacemaking can often precipitate heart disease. It was found that patients with a fast heart rate of 100 bpm or greater are at 3-fold higher risk of heart disease, and patients with slow heart rate of 60 bpm or less have a 4-fold higher risk of sudden death due to myocardial infarction (Benetos, Rudnichi, Thomas, Safar, & Guize, 1999; Chang et al., 2003; Jouven et al., 2005).  These conditions can be fatal if untreated. The current clinical approach is by correction via medication or electronic pacemaker implantation (Ferrari, 2002; Shattock & Camm, 2006). Our goal in the lab is to study the natural pacemaker, the sinoatrial node (SAN), how the heart beat is initiated and maintained, and how the knowledge can be applied in drug development.   The frequency of a beating heart depends on the rate of diastolic depolarization during an action potential in pacemaker cells of the sinoatrial node. We study the hyperpolarization-activated current (Ih), which is one of the currents that contribute to this part of the action potential. This current, generated by a flow of inward sodium ions through hyperpolarization-activated cyclic nucleotide-gated (HCN) channels, can also be modulated by β-adrenergic stimulation, increasing channel activity and subsequently the heart rate. HCN channels are similar to voltage-gated potassium (Kv) channels. One special feature is their ability to bind to cyclic nucleotide, and hence the ability to respond to cellular level of adrenaline. Cyclic AMP (cAMP) binds to the intracellular cyclic nucleotide binding domain (CNBD), which triggers a series of conformational changes that leads to gating ring formation and channel opening. The first part of the introduction briefly describes the history of the channel 2  and the biophysical differences between the channel isoforms. The remaining sections will mainly focus on what is currently known about cAMP modulation. This includes structural comparison of HCN with other CNBD-containing proteins, quantitative analyses on cAMP binding, and the conformational changes and dynamics upon ligand binding. The previous findings and understanding of HCN channels build up the rationale of my projects.  1.2 The hyperpolarization-activated mixed cation or “funny” current 1.2.1 History of the hyperpolarization-activated mixed-cation current Cardiac automaticity was observed by Claudius Galen, and later by Leonardo da Vinci in their anatomical investigations, and they both discovered the rhythmic and spontaneous nature of cardiac muscle, despite removing the heart from the body (D. DiFrancesco, 1993, 2010). The isolated beating heart indicated that its activity was independent from the nerves. The explanation of this mysterious and continuous electrical conductance was not proposed until centuries later when the introduction of voltage-clamp provided better understanding of the ionic mechanism behind the oscillating pattern (Brown, Giles, & Noble, 1977; Noma & Irisawa, 1976; Vassalle, 1966). It was suggested that the regularity in action potential firing in the pacemaker of the heart was due to a number of electrical currents (Brown, 1982).  This pacemaking mechanism in the heart was first thought to be associated with a decaying potassium current observed in Purkinje fibers. Upon depolarization, it was thought that an outward potassium current (IK2) diminishes as an onset to firing a subsequent action potential. This current seemed to associate with sympathetic innervations of heart rate because it was modulated by β-adrenergic stimulation (Hauswirth, Noble, & Tsien, 1968; Noble & Tsien, 1968). A decade later, in the late 1970s, Brown et al. applied adrenaline and found an increase in 3  an inward current in rabbit SAN tissues. This new current was called the pacemaker current because it is involved in pacemaking and adrenaline-promoted acceleration of beating frequency (Brown, DiFrancesco, & Noble, 1979). However, in contrast to the IK2 current, the pacemaker current is carried by a mix of sodium and potassium ions and is activated upon hyperpolarization. The pacemaker current is also known as the “funny” current (If) because its properties were found to be odd as compared to more well-known potassium, sodium and calcium voltage-activated currents that activate upon membrane depolarization and conduct mainly one ion. It turned out that IK2 and If were in fact the same current (D. DiFrancesco & Ojeda, 1980). Upon reinterpretation on the ionic properties and kinetics of IK2 from Purkinje fibers, it was a blend of activation of If and the deactivation of IK during hyperpolarization, giving rise to a mirage of a pure K+ current with a reversal potential close to K+ equilibrium potential (D. DiFrancesco, 1981a, 1981b). By blocking potassium currents with barium and cesium, DiFrancesco was able to identify the If current in Purkinje fibers and characterize its unique properties (D. DiFrancesco, 1981a).  1.2.2 Biophysical properties of If / Ih The unique biophysical properties of the If suggest an important contribution to the generation of diastolic depolarization and spontaneous action potentials. The current is inactive at positive voltages and is switched on upon hyperpolarization near the resting membrane potential, at the -40 to -50 mV threshold. The range of activation of pacemaking activity in the SAN is -60 to -75 mV (Yanagihara & Irisawa, 1980b), while it is -50 to -60 mV in purkinje fibers (D. DiFrancesco, 1981b; D. DiFrancesco & Ojeda, 1980), allowing inward If. The reverse potential lies between ENa and EK at about -20 mV, indicating the inward current consists of 4  permeability of both ions (Yanagihara & Irisawa, 1980a). In the rising phase of an action potential, If is rapidly deactivated.   The pioneering single channel recordings were identified in rabbit sinoatrial node cells. The mixed inward current was found to activate at less than -40mV and the single If-channel conductance was reported for the first time to be 1pS, which is one of the smallest measured among the cation channels in the voltage-gated family (D. DiFrancesco, 1986).  Other hallmark features of If are the voltage-dependent block by cesium, which allowed for its separation from other currents in Purkinje fibers and sinoatrial node myocytes (D. DiFrancesco, 1981b, 1982; D. DiFrancesco, Ferroni, Mazzanti, & Tromba, 1986), and its response to acetylcholine and noradrenaline (D. DiFrancesco et al., 1986; Ishii, Takano, Xie, Noma, & Ohmori, 1999; Mangoni, 2001).   1.2.3 Modulation of If by the autonomic nervous system A remarkable feature of If is its sensitivity to autonomic stimulation. In the body, the parasympathetic nervous system, via the action of the neurotransmitter acetylcholine (ACh), dampens heart rate whereas the opposing sympathetic nervous system acts by releasing noradrenaline from nerve endings in the heart and adrenaline from the adrenal gland which accelerates the rate (Brown et al., 1979; D. DiFrancesco, 1986; D. DiFrancesco et al., 1986; Hauswirth et al., 1968). In the sinoatrial node, β-adrenergic stimulation shifts the range over which If activates to less negative potentials whereas muscarinic stimulation shifts the activation range to more negative potentials (D. DiFrancesco et al., 1986; D. DiFrancesco & Tromba, 1988a, 1988b).  5  In the heart, adrenaline binds to a G-protein coupled receptor and triggers the Gα s stimulating subunit to activate adenylyl cyclase (Krupinski et al., 1989; Premont et al., 1996). This membrane-bound enzyme converts ATP into cAMP, which then serves as an intracellular messenger. On the other hand, ACh binds to muscarinic receptors in the heart and the Gα i inhibitory receptor decreases cAMP level in the cell (Simon, Strathmann, & Gautam, 1991). Mimicking the effect of β-adrenergic stimulation, a positive shift in the activation range of If  was found to occur by the direct action of cAMP on the cytoplasmic side of the channel with a half-maximal cAMP concentration (EC50) of 0.2 µM (D. DiFrancesco & Tortora, 1991). Both cAMP and cGMP bind to sinoatrial HCN, and facilitate opening to the same extent, but the former  has an EC50 that is >thirty-fold lower (0.2 versus 7.85 µM) (D. DiFrancesco & Tortora, 1991). Adrenaline and intracellular cAMP increase single If-channel activity upon hyperpolarization without modifying the single channel conductance (D. DiFrancesco, 1986; D. DiFrancesco & Mangoni, 1994).  Downstream of adrenaline stimulation, cAMP also activates protein kinase A (PKA) in the heart which contributes to regulation of electrical and contractile activity by phosphorylating the L-type calcium channel, and increasing its activity, as well as by phosphorylating other proteins involved in excitation-contraction coupling (Ferrier, Zhu, Redondo, & Howlett, 1998; Igami, Yamaguchi, & Kasai, 1999). However, excised patch experiments on cardiac pacemaker cells showed that direct application of PKA to the cytoplasmic side had no effect on If (D. DiFrancesco & Tortora, 1991). The direct action of cAMP is marked by a depolarizing shift in the range of voltages over which the channel opens, ranging from ~11 to 14 mV in the SAN and accounting for most of the additive shift of ~18  mV seen upon maximal stimulation with β-adrenergic and muscarinic agonists (Figure 1.1) (Accili, Redaelli, & DiFrancesco, 1997; D. 6  DiFrancesco & Mangoni, 1994). Furthermore, ACh also activates an inwardly rectifying potassium current (IK,ACh) in the pacemaker cell which contributes to slowing of beating rate; however, the activation of IK,ACh requires a 20-fold higher concentration of ACh compared to that required for If inhibition, which suggests that an effect on If predominates at lower levels of ACh in the body (D. DiFrancesco, Ducouret, & Robinson, 1989).   Similar properties are also found in electrical activity in neurons (Halliwell & Adams, 1982; Maccaferri, Mangoni, Lazzari, & DiFrancesco, 1993), aiding in signal transmission and controlling the rhythmic activity within the nervous system (Leresche, Jassik-Gerschenfeld, Haby, Soltesz, & Crunelli, 1990; McCormick & Pape, 1990b).  In the central nervous system, Ih is generated at voltages near resting membrane potential, and by allowing inward current, allows depolarization of the cell (Aponte, Lien, Reisinger, & Jonas, 2006; Nolan et al., 2003). Constitutively active Ih serves to stabilize fluctuation in membrane potentials (Nolan, Dudman, Dodson, & Santoro, 2007). Cyclic AMP level is altered according to hormones and neurotransmitters to facilitate activation of Ih (Frere & Luthi, 2004; McCormick & Pape, 1990a; Pape & McCormick, 1989; Tokimasa & Akasu, 1990) .   1.2.4 Role of If in cardiac pacemaking The unusual properties of If make it suitable for acting as a pacemaker current (D. DiFrancesco, 1993). In the SAN, the If is constitutively active at voltages near the resting membrane potential and, because its reversal potential is approximately -20 mV in physiological conditions, has the ability to initiate spontaneous activity. In addition, during action potential of SAN cells, outward potassium currents lead to hyperpolarization of the membrane, which also activates inward movement of cations because of the reversal potential. The inward ionic current 7  contributes to depolarization of the membrane in the activation range during diastole (D. DiFrancesco, 1981b; D. DiFrancesco, Ohba, & Ojeda, 1979). Depolarization leads to action potential firing, which is mainly due to the activation of the L-type calcium current, and the cycle is repeated. The If deactivates at more depolarized voltages, and re-activates upon repolarization when the membrane potential reaches -40/-45 mV in the SAN (D. DiFrancesco, 1991; D. DiFrancesco et al., 1986; van Ginneken & Giles, 1991). The If opposes the contribution of the delayed rectifier potassium current, which diminishes during the repolarization phase. Thus, the contribution of inward If coincides with the diastolic depolarization in SAN cells and is thought to be a major contributor to this phase of the action potential (Figure 1.1A). The diastolic depolarization is a distinguishing property of conducting cells, such as SAN cells, and is lacking in working muscle, such as ventricular myocytes, in the adult mammal. Support for this comes from using the drug ivabradine. This drug is a If-specific blocker at low concentrations slows the rate of diastolic depolarization in isolated cells and to slow heart rate in animals and people (Bois, Bescond, Renaudon, & Lenfant, 1996; Ekman et al., 2011; Fox, Ford, Steg, Tendera, & Ferrari, 2008; Heusch et al., 2008; Mulder et al., 2004) The changes in the action potential and heart rate by autonomic stimulation also likely involve If. Adrenaline or isoproterenol, at low levels, added to SAN cells accelerates the spontaneous rate by steepening the upward slope of diastolic depolarization, without large effects on the shape, magnitude, and duration of rest of action potential (Brown et al., 1979; Zaza, Robinson, & DiFrancesco, 1996). Similarly, muscarinic stimulation decreases diastolic depolarization without drastically changing the other parameters (Bucchi, Baruscotti, Robinson, & DiFrancesco, 2007; Zaza et al., 1996). These findings suggest a major contribution of If to the changes in heart rate by autonomic input (Figure 1.1A). Despite the contribution of If, other 8  currents and cellular systems are probable contributors to pacemaking and its modulation in the SAN by the autonomic nervous system (Bogdanov, Vinogradova, & Lakatta, 2001; Huser, Blatter, & Lipsius, 2000; Irisawa, Brown, & Giles, 1993; Rubenstein & Lipsius, 1989).    Figure 1.1 Effect of isoprenaline and acetylcholine on action potential of the SAN.  (A) The If is responsible for the recovery and re-firing of action potential during diastolic depolarization. The rate of recovery and the frequency of AP firing can be modified by 0.01 µM isoprenaline or 0.003 µM acetylcholine, leading to an increase or decrease in the heart rate, respectively. The duration of diastolic depolarization is depicted with a dash line, and the duration changes with the modulation. (B) Tested in rabbit sinoatrial myocytes, the activation curves show the steady state fraction of open channels (Y-axis) with 1 µM perfusion of either chemical. The rate change is due to a shift in the voltage dependence for activation, contributing to an increase (isoprenaline) or decrease (acetylcholine) in fraction of open channels at a particular voltage. Modified from (Accili, Proenza, Baruscotti, & DiFrancesco, 2002)  1.3 Hyperpolarization-activated cyclic nucleotide-gated “HCN” channels underlie the “funny” current 1.3.1 Cloning the HCN genes The HCN channel was originally identified in 1997 when a new Src-interacting protein was identified in mouse brain using yeast two-hybrid screening (Santoro, Grant, Bartsch, & A B 9  Kandel, 1997), but the function of the protein and its relationship to If/Ih was discovered later (Santoro et al., 1998). Also in 1998, two other groups identified the HCN proteins. One group used BLAST searches of the EST databases using the cAMP-binding region from the cyclic nucleotide-gated (CNG) family (Ludwig, Zong, Jeglitsch, Hofmann, & Biel, 1998; Zagotta & Siegelbaum, 1996). A second group cloned a channel from sea urchin testis that exhibited the hallmark features of Ih/If (Gauss, Seifert, & Kaupp, 1998). These proteins were found to be similar to those of the voltage-gated potassium family, especially the CNG channels and the so-called ether-a-go-go related channels, and were eventually named Hyperpolarization-activated Cyclic nucleotide-gated or “HCN” channels (Clapham, 1998). At least four isoforms, from four different genes, are found in vertebrates while most invertebrates possess one HCN gene (Jackson, Marshall, & Accili, 2007).   1.3.2 Structure and function of HCN channels Sequence homology suggests that the HCN channels resemble voltage-gated potassium channels in structure and function (Santoro et al., 1997). Thus, HCN channels are predicted to exist as tetramers with intracellular N and C-termini, and that each subunit has a predicted topology that is similar to that for a voltage-gated potassium channel.  1.3.2.1  Responses to voltage and voltage-sensing elements of the HCN channel When expressed in heterologous cells, functionally characterized HCN isoforms are activated upon hyperpolarization of membrane potential (Accili et al., 2002; Robinson & Siegelbaum, 2003; Wahl-Schott & Biel, 2009). The voltage dependence of activation is a determination of the range over which the channels activate and is plotted as the degree of 10  activation versus membrane voltage (Figure 1.1B). This S-shaped activation curve can be fit with a single-order Boltzmann function, revealing the half-maximal activation potential (V1/2) and inverse slope factor for If. There is a large variability in the value, depending on isoform, cell type and experimental conditions. The usual range of V1/2 for cloned channels is from -70 to -100 mV. When expressed in the same cells and compared, HCN4 has the most negative V1/2 (-140 mV) and HCN1 has the least negative V1/2 (-70 to -90 mV); HCN2 and HCN3, are in between, at -70 to -100 mV and -80 to -95 mV, respectively (Altomare et al., 2003; Baruscotti, Bucchi, & DiFrancesco, 2005; Stieber, Stockl, Herrmann, Hassfurth, & Hofmann, 2005). As comparison, in SAN cells, the activation range of If is between ~-45 to  -90 mV, with a half-activation voltage of -70mV (Accili et al., 1997; D. DiFrancesco & Tromba, 1988b). The positive shift in SAN cells may be due, in part, to the availability of basal intracellular cAMP pre-activating the channel (D. DiFrancesco & Mangoni, 1994). In ventricular cardiomyocytes, the activation range and threshold are in a more negative range of voltages, with half maximal activation of -110mV (Fares, Bois, Lenfant, & Potreau, 1998; Robinson, Yu, Chang, & Cohen, 1997; H. Yu, Chang, & Cohen, 1995). This variation in the V1/2 and in the cAMP-induced shift of V1/2 could arise from several factors. The variation could be simply due to experimental conditions. This includes difference in pH, ionic concentration, and configuration of the patch-clamp experiments (Biel, Michalakis, & Zong, 2009). The cellular context may also influence the activation range of HCN voltage dependence, which has been observed when comparing the channels expressed in native myocytes versus mammalian cell lines (Qu, Altomare, Bucchi, DiFrancesco, & Robinson, 2002).  The similarity in sequence between HCN and Kv channels suggests that the structure of the transmembrane domain of HCN channel is similar to that of depolarization-activated Kv 11  channel and that this domain contributes to voltage-sensing. The predicted topology suggests that each HCN channel subunit contains six alpha helical segments (S1-S6) in the transmembrane domain, and that four subunits interact to form the channel (Figure 1.2). The positively-charged S4 helix is lined with basic residues spaced out along the helix. When these charges are neutralized by mutations in HCN2, the voltage dependence is shifted by 20 mV in the hyperpolarized direction (Vaca et al., 2000). These charges are involved in gating or protein folding and trafficking (J. Chen, Mitcheson, Lin, & Sanguinetti, 2000). It was suggested that the S4 segment moves through the membrane electric field in a way that is similar to Kv channels (Broomand, Männikkö, Larsson, & Elinder, 2003), except the direction of S4 movement is inverted where inward movement leads to channel opening (Elinder, Männikkö, & Larsson, 2001; Mannikko, Elinder, & Larsson, 2002; Vemana, Pandey, & Larsson, 2004). After testing cysteine accessibility in open and closed states, the results suggest that S4 helix lies in a narrow water-filled canal in the open state and does not undergo large translational, rotational, or paddle movement between the two states. Instead, it has been proposed that the S4 segment itself is relatively static and voltage sensing is achieved by rearrangement of surrounding environment. Based on voltage-gated sodium channels, the movement of surrounding helices leads to the collapse of the crevice, which is coupled to closing of channel (Bell, Yao, Saenger, Riley, & Siegelbaum, 2004; Yang, George, & Horn, 1996). A suggested mechanism involves the voltage sensor driving a twisting movement of the S5 helix upon hyperpolarization, which leads to an outward tilting of the S6 helix from the central axis of the channel (Kwan, Prole, & Yellen, 2012) and opens the activation gate in the distal S6 helix (B. S. Rothberg, Shin, Phale, & Yellen, 2002; K. S.; Shin, Rothberg, & Yellen, 2001). Although the exact mechanism is still unknown, it has been implicated that the S4-S5 linker, S6, and post-S6/C-linker region are important for 12  voltage-dependent gating (J. Chen, Mitcheson, Tristani-Firouzi, Lin, & Sanguinetti, 2001; Decher, Chen, & Sanguinetti, 2004; Prole & Yellen, 2006; B. S.; Rothberg, Shin, & Yellen, 2003). Among the four mammalian HCN isoforms, there are also differences in the rates of activation and deactivation in response to membrane potential. The HCN1 isoform has the fastest activation rate, followed by HCN2, HCN3, and HCN4, respectively. Studies with HCN1 and HCN4, the fastest and slowest activated isoform respectively, showed that differences in the transmembrane domain contribute to differences in activation kinetics; single mutations or chimeras of the S1, the S1-S2 linker and S6 components from HCN1 to that of HCN4 lead to slowing of activation gating of HCN channels (Ishii, Takano, & Ohmori, 2001).   13   Figure 1.2 The topology of the HCN channel subunit. Each subunit consists of intracellular N- and C- termini, 6 transmembrane helical segments that resemble the voltage-gated potassium channel, and a ligand binding region in the C-terminus that resembles the CNG channel. Each channel is made up of 4 homo- or hetero-subunits. Inward mixed cation movement generates the If.  Adapted from (Biel et al., 2009)  1.3.2.2 The HCN channel pore and cation permeation Like If/Ih, the cloned HCN channels have mixed permeability and select for K+ over Na+ around 5:1; thus, the reversal potential in physiological solutions dictates an inward flow of sodium ions, activated near the resting membrane potential (Ishii et al., 1999; Ludwig et al., 1998; Macri, Angoli, & Accili, 2012). If/Ih and HCN channels are nearly impermeable to lithium ions (Wollmuth & Hille, 1992) (Ho, Brown, & Noble, 1994) and are inhibited by cesium ions 14  (D. DiFrancesco, 1982; Ludwig et al., 1998). In agreement between cloned and intact cells, the size of the recorded single channel were very small, 1.5 pS and 1 pS respectively (Dekker, 2006; D. DiFrancesco, 1986; Michels et al., 2005).  As with voltage-gated potassium channels, the S5-S6 helices of HCN channels have been proposed to form the central pore (Biggin, Roosild, & Choe, 2000; Kuang, Purhonen, & Hebert, 2015; Lu, Klem, & Ramu, 2001). The mammalian HCN isoforms possess a GYG motif in the outer pore (Figure 1.2), which is also found in potassium channels and was thought to be responsible for potassium selectivity (Doyle et al., 1998; Y. Zhou, Morais-Cabral, Kaufman, & MacKinnon, 2001). Clearly, the presence of the motif in HCN channels is not sufficient to confer potassium selectivity, given their significant permeability to sodium ions (Macri et al., 2012). The inner pore of the HCN channels is thought to house the gate of the pore as it does on Kv channels (MacKinnon, 2003; Macri et al., 2012; B. S. Rothberg et al., 2002; Yellen, 1998).   1.3.2.3 Facilitation of channel opening by cAMP and the structural components involved The cloned HCN channels are impacted by the binding of cAMP but different isoforms are impacted variably. The mammalian HCN isoforms exhibit large differences in the magnitude of cAMP effects. A small depolarizing shift in the range of HCN1 activation is produced by cAMP, in the range of +2 to +7 mV, while the HCN2 and HCN4 isoforms are shifted by much larger amounts, +15 to +25 mV (Elinder, Mannikko, Pandey, & Larsson, 2006; Ludwig, Zong, Stieber, et al., 1999; Seifert et al., 1999; Viscomi et al., 2001; Wang, Chen, & Siegelbaum, 2001). Cyclic AMP has little effect on the human HCN3 isoform (Cao-Ehlker et al., 2013; Stieber et al., 2005) and the mouse HCN3 channel responds by shifting a small amount in the hyperpolarizing direction (Mistrik et al., 2005) (Figure 1.3). 15  The opening of HCN channels is also sensitive to cGMP, which produces a comparable depolarizing shift in the activation curve to cAMP in both the HCN2 and HCN4 channel (Ludwig et al., 1998) (Figure 1.3A). However, HCN channels are less selective for cGMP, shown by a lower potency (EC50) values and true for HCN1, 2, and 4 (S. Chen, Wang, & Siegelbaum, 2001; Möller et al., 2014; Xu et al., 2012; L. Zhou & Siegelbaum, 2007). Cyclic AMP is 10 to 60 times more potent than cGMP in the HCN2 isoform (Ludwig et al., 1998; Zagotta et al., 2003; L. Zhou & Siegelbaum, 2007). The mechanism by which cAMP modulates HCN channels is discussed in more detail below.   1.3.2.4 Heteromeric assembly of HCN subunits One reason for the discrepancy in biophysical properties between Ih measured in native and HCN-expressing cell lines may be the intrinsic differences in activation among the mammalian isoforms and the presence of multiple isoforms in various tissues. The ability of HCN channels to form heterotetramers further complicates If properties, where, for example, heterotetramers of HCN1 and HCN2 would have intermediate channel kinetics properties and cAMP sensitivity as compared to respective individual homotetramers (Brewster, Bernard, Gall, & Baram, 2005; S. Chen et al., 2001). The heteromeric assembly of the two cardiac isoforms, HCN2 and HCN4, displays activation properties similar to endogenous myocardial If currents (Ye & Nerbonne, 2009). Heteromerization of different isoform combination seems to be possible except for HCN2 and HCN3 (Much et al., 2003). The various combinations in different stoichiometries of heteromeric channels give rise to many possibilities and thus diversify the channel behaviour and function (Lewis, Estep, & Chetkovich, 2010).  16   Figure 1.3 Shifts in voltage dependence induced by cAMP.  (A) The half-activation voltage shows the difference in basal properties of each isoform, where HCN1 and HCN3 have less negative V1/2 and are proposed to be pre-activated. In the inside-out patch experiment, addition of ligand to the bath solution allows a shift in voltage dependence. *** p  < 0.0001, * p < 0.05. cGMP has a smaller effect at this ligand concentration due to a lower potency. (B) HCN2 and 4, the cardiac isoforms, are greatly modulated by cAMP, whereas the pre-activated isoforms have less effect. Adapted from (Stieber et al., 2005)  1.3.2.5 Regulation by other factors HCN channels are regulated by factors which could be responsible for the differences in function observed between different cell types normally in the body and/or in heterologous expression systems. HCN channels are regulated by serine/threonine kinases or tyrosine (Src) kinases, which assisted in the identification of the HCN clones (Santoro et al., 1997). Application of a kinase inhibitor in oocytes causes a negative shift in the activation of HCN2 (H. G. Yu, Lu, Pan, & Cohen, 2004), but co-expressing a constitutively active Src kinase shifts in the opposite direction (Arinsburg, Cohen, & Yu, 2006). A particular example was found in Y476 in HCN2, 17  located in the C-linker, where when phosphorylated, channel activation is accelerated and shifts the activation to the right as application of cAMP would (Zong et al., 2005). Phosphorylation could be involved in aiding cAMP binding as well, but inhibition of kinases did not abolish the effect of cAMP (Zong et al., 2005). Thus, cAMP modulation is independent, and possibly a different mechanism, from phosphorylation (Accili et al., 1997). A more recent study has shown that PKA is involved in shifting the voltage dependence in cells with heterologously expressed HCN4 channels and found phosphorylation sites in the N- and C- terminus (Liao, Lockhead, Larson, & Proenza, 2010). HCN channels are regulated by auxiliary proteins. The cytoplasmic accessory protein called tetratricopeptide repeat-containing Rab8b-interacting protein (TRIP8b) is involved in the regulation HCN cell surface expression and the sensitivity to cAMP in neurons (Lewis et al., 2009; Zolles et al., 2009). One of the binding sites for TRIP8b is thought to be in the CNBD which interferes with cAMP modulation (Santoro et al., 2011; Zolles et al., 2009). The NMR structure suggests that TRIP8b does not directly bind to nor compete for the cAMP binding site, but it stabilizes the apo conformation (Saponaro et al., 2014). Recent double electron-electron resonance (DEER) experiment shows a resonance change in the phosphate binding cassette (PBC) when the carboxy-terminal region interacts with TRIP8b, and thus reducing its affinity for cAMP. The movement of the C-helix that couples binding to channel facilitation is also impeded (DeBerg et al., 2015; Saponaro et al., 2014). This coupling will be explained in more detail in later sections.  Other known examples of modulatory peptides for HCN channels include mink-related peptide (MiRP1), which serves as a beta subunit that assists with protein expression and increases activation kinetics of the channel (H. Yu et al., 2001) via a “leaky” instantaneous current (Proenza, Angoli, Agranovich, Macri, & Accili, 2002); KCNE2 has the opposite effect 18  and slows activation by a hyperpolarizing shift in the voltage dependence (Decher, Bundis, Vajna, & Steinmeyer, 2003); the adapter protein SAP97 interacts with the C-terminus and alters the If behaviour (Peters et al., 2009).    Phosphatidylinositol-4,5-bisphosphate (PIP2) and other negatively charged phospholipids play a physiological role in controlling the If. PIP2 serves as an allosteric regulator that causes a depolarizing shift by about 20mV to the Ih activation (Pian, Bucchi, Decostanzo, Robinson, & Siegelbaum, 2007; Zolles et al., 2006). The effect is independent to that of cAMP. The depolarizing effect is still present in constructs lacking the CNBD (Pian et al., 2007; Pian, Bucchi, Robinson, & Siegelbaum, 2006) . The effect of PIP2 is equally effective in all four mammalian isoforms (Ying et al., 2011; Zolles et al., 2006). PIP2 has similar facilitating effects on HCN channel opening endogenously in rabbit SAN cells (Pian et al., 2006) and embryonic cardiomyocytes (Zolles et al., 2006).   1.3.2.6 Cellular trafficking of the HCN channel  Like potassium channels, N-termini of HCN1 and HCN2 interact and are also necessary for expression of functional tetrameric channels. Using yeast two-hybrid system, the N-termini of the HCN1 and HCN2 isoforms were found to interact with themselves and with each other and to be necessary for functional expression of either homo- or heterotetramers. GFP-tagged N-terminus-truncated mutation shows that channels remain in the intracellular compartments rather than on the cell surface and eliminates channel assembly. On the other hand, the carboxy-terminus does not interact with itself in the yeast two  hybrid assay and is not required for functional channel expression (Proenza, Tran, et al., 2002; Tran et al., 2002). 19  Immunohistochemistry staining shows a mutation in the N-terminus, P257S, stops localization to the cell membrane and instead retains localization to the cytoplasm (Macri et al., 2014).  1.3.3 Tissue expression of the four isoforms The HCN isoforms are expressed and distributed non-ubiquitously in the body. HCN channel expression is prominent in the heart and nervous system. All four isoforms are expressed in the heart but at different levels depending on the region. The SAN is mainly composed of the HCN4 isoform (Ishii et al., 1999; Shi et al., 1999), although there is evidence for lower levels of HCN1 and HCN2 (Moroni et al., 2001; Whitaker, Angoli, Nazzari, Shigemoto, & Accili, 2007). HCN4 are also found in the atrioventricular node and Purkinje fibers while HCN2 is the predominant isoform in atrial and ventricular myocytes. HCN2 is found to be distributed ubiquitously in the heart. (Moosmang, Biel, Hofmann, & Ludwig, 1999; Shi et al., 1999).  Although present in the heart, HCN1 and HCN3 are more generally considered as contributors in the nervous system along with the HCN2 and HCN4 isoforms (Biel et al., 2009; Robinson & Siegelbaum, 2003).  1.3.4 Genetic disruption of specific HCN genes demonstrates the physiological role of HCN channels Genetic disruption of HCN genes provides evidence for physiological roles of Ih. Among the isoforms, HCN4 contributes to 80% and HCN2 to 20% of the If current in the SAN, shown by knockout experiments (Harzheim et al., 2008; Herrmann, Stieber, Hofmann, & Ludwig, 2006; Stieber et al., 2003). When the gene is knocked out globally or specifically in the heart, HCN4-deficient mice showed a reduction in cardiac beating by 40% before embryonic day 10. Even 20  though heart beat is not completely abolished, the mice were no longer sensitive to β-adrenergic response (Stieber et al., 2003). Overexpression of an HCN4 subunit, with a mutation in the CNBD (R669Q) that eliminated strong binding of cyclic nucleotides, also stopped the action of adrenaline in homozygotic mice (Harzheim et al., 2008). Both of these HCN4- deficient mice died in utero around embryonic day 11 (Harzheim et al., 2008; Stieber et al., 2003). Therefore, HCN4 channels are crucial for responding to β-adrenergic agents, as well as developing the adult pacemaker cells and driving the beating of the heart. Tissue-specific knockouts reinforce the expression data and confirm the cardiac function of If in vivo. A temporal knockout of HCN2 or HCN4 isoform in the SAN does not show a fatal phenotype. Adult HCN4-deficient mice, like embryonically knocked-out mice, displayed an 80% reduction If; the basal heart rate was normal except these mice exhibited episodes of recurrent sinus pauses and beta-adrenergic stimulation of heart rate was unaffected (Herrmann, Stieber, Stockl, Hofmann, & Ludwig, 2007). The knockout caused a hyperpolarizing shift in pacemaker cells and thus, HCN channels were less likely to open. HCN2 knockout resulted in sinus dysrhythmia but sinus rhythm and autonomous regulation was unchanged. HCN2 knockout is also not embryonically lethal (Ludwig et al., 2003). Knockout experiments on HCN3 conclude that the isoform regulates the ventricular action potential. The knockout reduces If by 30% and shifts the resting membrane potential to more hyperpolarized potentials (Fenske et al., 2011).  1.4 HCN channel in human disease  1.4.1 Disrupting HCN expression leads to heart failure Gene expression in the body is tightly regulated. An increased expression of HCN channels resulting in increased activity of If is associated with arrhythmia (Kuwabara et al., 21  2013). The up-regulation of HCN2 and HCN4 are prominent in hypertrophic and failing hearts. It was reported that transcription of these genes was driven by a ubiquitous Sp1 protein and was inhibited by muscle-specific mRNAs. An abnormal balance of the two factors leads to activation of the cardiac genes (Lin, Xiao, Luo, Chen, & Wang, 2009). A reduced expression of HCN4, on the other hand, is associated with heart failure and aging (X. Huang, Yang, Du, Zhang, & Ma, 2007; Zicha, Fernandez-Velasco, Lonardo, L'Heureux, & Nattel, 2005). It also increases susceptibility to sick sinus syndrome and atrial fibrillation (Ellinor et al., 2012).    1.4.2 HCN inhibition as treatment for cardiac disease  HCN channels have a unique pacemaking role in the heart, and its regulation can alter the heart rate and thus serves as a therapeutic target. For example, the inhibition of HCN channels has been proposed as a treatment option for diseases because this approach should selectively decrease heart rate without directly compromising contractility. Changing the heart rate could reduce stress on cardiomyocytes or satisfy oxygen demands for cells, and would improve conditions such as angina, ischemic heart diseases, or heart failure. It has been a goal pursued by pharmaceutical companies for decades to control heart rate through administered drugs because it is a safer approach than the surgical procedure. External cesium or rubidium inhibit If, but these ions are not specific as they block or interfere with other ion channels (D. DiFrancesco, 1982). Pure bradycardic agents were later discovered to reduce diastolic depolarization specifically without changing other action potential parameters, such as ivabradine mentioned earlier (Bois et al., 1996; Scicchitano et al., 2014). The 22  drug binds on the intracellular side upon channel opening (Bucchi, Baruscotti, & DiFrancesco, 2002).  Ivabradine has been clinically proven to inhibit HCN channels specifically and to treat angina pectoris (Riccioni, Vitulana, & D'Orazio, 2009; J. C. Tardif, Ford, Tendera, Bourassa, & Fox, 2005). The drug works in vitro with perfusion of isolated SAN cells, and also in vivo with delivery via the bloodstream (Thollon et al., 1997). Ivabradine is effective for various cardiovascular conditions with lowered heart rate or left ventricular dysfunction (Ekman et al., 2011; Fox et al., 2008; J. C. Tardif et al., 2005; J. C. P. Tardif, P., Kahan, & Investigators, 2009) and was recently approved by the U.S. Food and Drug Administration (2015), under the trade name Corlanor to reduce hospitalization from worsening heart failure. The downside to ivabradine are side effects, including visual disturbances, which are probably caused by Ih block of HCN channels in the retina (Cervetto, Demontis, & Gargini, 2007), profound bradycardia, and atrial fibrillation. Due to side effects, the search for an ideal agent to control heart rate continues, which also leads to the rationale of this thesis of targeting the cAMP binding site instead as a therapeutic solution. The advantage of using a drug that acts on the binding site is due to the intrinsic ligand specificity between isoforms.  1.4.3 Disease-associated mutations in HCN channels  1.4.3.1 Disease-associated mutations in the HCN2 channel isoform Clinical studies have shown HCN channels play a critical role in the pathogenesis of epilepsy. Defects in ion channels have been linked to the cause of inheritable idiopathic epilepsy (Catterall, Dib-Hajj, Meisler, & Pietrobon, 2008; Mulley, Scheffer, Petrou, & Berkovic, 2003). 23  Dysfunction in HCN channels, also found expressed in neurons, could lead to such pathology (Poolos, 2005; Tang, Sander, Craven, Hempelmann, & Escayg, 2008) (Figure 1.4).  A gain-of-function mutation was located in the post-CNBD region, where a triple proline deletion in positions 719-721, and produced 35% more If compared to wild type. The deletions were found more commonly in children with febrile seizure (FS) (Dibbens et al., 2010). Later, a heterozygous missense mutation in the intracellular N-terminus of HCN2 (S126L) was found in 2 unrelated children diagnosed with FS. Mutated channels were sensitive to temperature and had faster kinetics at higher temperature, increasing availability of the HCN current leading to FS (Nakamura et al., 2013).  In the C-terminus, two mutations were found that are linked by idiopathic generalized epilepsy. The first is a mutation of a highly conserved residue (R527Q), located in the D’ helix of the C-linker, two residues away from the intersubunit salt bridge that dictates channel opening. This variant was spotted during a mutation analysis of HCN1 and HCN2 in 84 patients with epilepsy. The mutated channel showed no significant differences in biophysical properties between the mutant and wild type. However, it did show a shallower GV slope, suggesting the possibility of a difference in the gating mechanism (Tang et al., 2008). Although the mechanism of cAMP modulation seems unaffected, only a saturating amount of cAMP was tested rather than in dose-response manner. In this thesis, we showed that cAMP affinity to this mutant CNBD is unaffected (unpublished data, Appendix 7F). The shallow GV slope could be an indicator of a different protein conformation (Craven & Zagotta, 2004; Decher et al., 2004; L. Zhou, Olivier, Yao, Young, & Siegelbaum, 2004) altering channel assembly.  Another epilepsy-associated mutation is found in the C’ helix of the HCN2 C-linker (E515K in human, E488K in mouse). The homomeric mutant channel shifts the activation curve 24  to the left, but the cAMP-stimulated rightward shift is unaffected. Because this is thought to be a loss-of-function mutation, the current proposal is that HCN currents are involved in an inhibitory pathway for neuronal excitability (J. C. DiFrancesco et al., 2011). Thus, down-regulation of Ih causes epilepsy, and increasing Ih by lamotrigine can remedy epilepsy (Peng, Justice, Zhang, He, & Sanchez, 2010; Poolos, Migliore, & Johnston, 2002). We looked further into this mutation and speculated how the single substitution leads to the negative shift in the absence of cAMP. The data was reported in Chapter Four.  1.4.3.2 Disease-associated mutations in the HCN1 channel isoform Found localized in neuronal dendrites, the HCN1 isoform plays an important role in stabilizing the neuronal membrane potential and its defect could lead to epilepsy (Kase & Imoto, 2012; Poolos, 2010). An international collaborative project involved whole-exome sequencing on children diagnosed with early infantile epileptic encephalopathies (Nava et al., 2014). They suffer from seizures beginning from 4 to 13 months of age with possible impairment in cognitive and motor development. Several mutations, located in the cytoplasmic side of the channel especially in the S4-S5 linker, were identified in the HCN1 channel. These mutations associated with epilepsy were tested using patch-clamp electrophysiology on CHO cells. Three mutations, S100F in the N-terminal segment, H279Y in the S4-S5 linker, and D401H in the C-linker, caused a depolarized shift in the half-activation voltage by 27 mV, 17 mV, and 46 mV, respectively. These mutations also led to faster activation and slower deactivation, giving them a gain-of-function phenotype. At the same time, S272P and R297T from the S4-S5 linker are loss-of-function mutations. Funny current was not detectable in CHO cells expressing these mutations, suggesting a reduction in protein expression or in protein stability. Although the study reinforced 25  the relationship between epilepsy and HCN1 channels, the cause remains unknown, whether it is the lacking or the hyperactivity of If that results in seizure.   Figure 1.4 The positions of the disease mutations found in human HCN1, 2, and 4 isoforms.  In general, the mutations in HCN2 (green) or HCN1 (purple) cause epilepsy and those in HCN4 (red) cause sinus bradycardia. The residue number is based on respective isoform in human. The disease hotspots seem to locate in the selectivity filter and the C-terminal region. The red dots with no labels indicate other mutations also discovered in HCN4 that are associated with AF but are omitted here. Those found in the C-linker and CNBD are studied more in depth in this thesis. Modified from (Verkerk & Wilders, 2015)  1.4.3.3 Disease-associated mutations in the HCN4 channel isoform A number of disease-associated mutations have been identified in the HCN4 channel. Some of the mutated sites are found in the pore, and a significant portion is found in the 26  cytoplasmic C-terminus (Figure 1.4). These mutations are associated with sinus node dysfunctions including bradycardia, left ventricular non-compaction (LVNC), AV block, atrial fibrillation (AF), and/or chronotropic incompetence (Verkerk & Wilders, 2014).  The mutation often causes a shift in the voltage dependence of the channel, and sometimes abolishes the β-adrenergic response (Verkerk & Wilders, 2014).These studies provide evidence for the importance of HCN channels and its links to pacemaking and cardiac arrhythmia.  Four mutations are in the extracellular S5-S6 linker of the transmembrane segments which includes the pore-loop. This loop is highly conserved GYG motif for potassium selectivity. G480R of HCN4 reduced the membrane expression of channels and therefore led to a reduction in current density, hyperpolarizing shift in voltage dependence, and slower activation kinetics. Eight members in the family were diagnosed with asymptomatic sinus bradycardia starting at a young age (Nof et al., 2007). Likewise, A485V mutation was reported with a reduction in current density and negative shift in activation curve (Laish-Farkash et al., 2010). Both Y481H and G482R were associated with sinus bradycardia and LVNC. Y481H mutated channel resulted in more than 40mV hyperpolarizing shift in the voltage dependence (Milano et al., 2014). G482R mutants revealed a 65% reduction in current density (Milano et al., 2014) and strong hyperpolarizing shift as well. Both would have minimal contribution to HCN4 current density during diastolic depolarization (P. A. Schweizer et al., 2014). Mutation in the S4-S5 linker (A414G) has the same disease phenotype found in three members of the same family and same shift in voltage dependence (Milano et al., 2014). There are two disease mutations that truncate the CNBD. The first disease discovered to be linked to idiopathic sinus bradycardia, where the patient had a heart rate of 41 bpm.  This mutation (L573X) produced a pre-matured truncation after the C’-helix and causes the removal 27  of the CNBD and part of C-linker. Without a CNBD, it makes sense that the mutated channel is no longer sensitive to cAMP. Patch-clamp experiment also showed a hyperpolarized shift in channel activation (Schulze-Bahr et al., 2003). A similar phenotype was observed in mice with this mutation (Alig et al., 2009). This is surprising because previous studies have shown when CNBDs of HCN1 and HCN2 were truncated, the inhibitory action was also abolished, shifting the dependence to more depolarized potentials (Viscomi et al., 2001; Wainger, DeGennaro, Santoro, Siegelbaum, & Tibbs, 2001). Another contradicting result was the fact that heteromeric channels were also cAMP insensitive, even though it was shown before that cotransfection of channels with active CNBD and inactive CNBD produced an intermediate response to cAMP; the examples were with either the cAMP-responsive HCN4 and cAMP-unresponsive HCN1 (Altomare et al., 2003), or the wild type HCN2 and CNBD-defected R591E mutant (Harzheim et al., 2008; Ulens & Siegelbaum, 2003). Such contradiction could be due to isoform differences and an inability to form the tetrameric gating ring.  The other truncation mutation (E617X) was found in a German family, and 8 members had sinus bradycardia with heart rate of ~46 bpm. An insertion in the HCN4 gene led to a frame shift that truncated at the end of the B-helix, taking out from the C-helix onwards. Patch-clamp experiments showed no difference in the shift in half-maximal voltage dependence, but the channel is insensitive to cAMP. The phenotype is similar to L573X, where both homomeric and heteromeric channels also fail to respond to cAMP. It could be a dominant-negative mutation that alters cAMP responsiveness and gating ring formation (P. Schweizer et al., 2010). This phenotype is consistent with removing the C-helix of HCN2 channels (Wainger et al., 2001). Also in the cytoplasm, the mutation K530N in the A’ helix of the C-linker was discovered in 6 members of a family, where patients had tachycardia-bradycardia syndrome and 28  atrial fibrillation (Duhme et al., 2013). The proband also had sinus bradycardia of 50-60 bpm. Electrophysiological studies showed the loss-of-function mutation caused a hyperpolarized shift in the activation curve only in the heteromeric (mut/wt) channel but surprisingly, had no effect in the homomeric mutant channel as compared to the wild type channel. The authors speculated it was due to a more stable arrangement between identical subunits. This also supports that the tetrameric channel behaves in a dimer of dimers configuration. Response to cAMP was unaffected for homomeric channels, and the shift was about 7 mV greater for heteromeric channels. This residue is one of the tripeptide residues involved in reversing the polarity of cAMP (L. Zhou et al., 2004). The binding affinity is comparable to that of wild type HCN4 (unpublished data, Appendix 9).  The D553N mutation from the B’ helix is associated with cardiac arrhythmia including bradycardia, QT prolongation, and Torsade de Pointes. It was first suggested that the dominant-negative defect was due to poor protein trafficking to the plasma membrane, shown by reduction in membrane expression, leading to reduced current (Ueda et al., 2004). A later group provided contradicting results where surface expression was comparable to wild type, and it was the reduction in current amplitude that leads to the observed phenotype (Netter, Zuzarte, Schlichthörl, Klöcker, & Decher, 2012). Surprisingly, despite the distance of the mutation to the CNBD, the mutation abolished response to β-adrenergic stimulation. The mutation could be due to the removal of an electrostatic interaction (from the original negatively charged aspartate) with the transmembrane loops, similar to S4-S5 linker with D443 of HCN2 (Decher et al., 2004). The mutated residue comes immediately before a conserved phosphorylation site in both mice and human and is important for controlling the activation kinetics, where dephosphorylation would increase If in adult ventricular myocytes (Arinsburg et al., 2006; Zong et al., 2005). We 29  have shown that this mutation in the isolated C-terminal fragment does not abolish cAMP binding (unpublished data, Appendix 9), suggesting that the mutation interferes with the channel in another way.  A mutation in the cyclic nucleotide-binding domain, S672R, is associated with asymptomatic sinus bradycardia and was discovered in 15 members of an Italian family. In HEK cells, the S672R mutant channel demonstrated a hyperpolarized shift in activation voltage range in the absence of cAMP, while the potency, measured as the EC50, of cAMP was unaffected (Milanesi, Baruscotti, Gnecchi-Ruscone, & DiFrancesco, 2006). The hyperpolarized shift in HCN4 activation would be expected to reduce the contribution of If to the diastolic depolarization, and thus, heart rate would be slowed. Another report confirmed the leftward shift, but also claimed it was due to reduced cAMP binding and not to an intrinsically defected modification (Xu et al., 2012). The cause of bradycardia is thus still unknown at the moment, but it was claimed that the second report was based on a HCN2-HCN4 chimera which may have led to the discrepancies (D. DiFrancesco, 2013). We examined this mutation biochemically in Chapter Four of this thesis. Since many of the mutations reside in the cytoplasmic C-terminal end in the C-linker and CNBD, many following studies revolved around the structural and functional contribution of this region. Also, many of these mutations cause a hyperpolarizing shift that leads to bradycardia, but how the single residues change the basal channel properties remains unknown.  30  1.5 Regulation of HCN function by cyclic nucleotide binding 1.5.1 Functional contribution of C-linker and CNBD Distal to the pore is an intracellular region which consists of the CNBD and the C-linker, a piece that connects the pore to the CNBD. It is known that cAMP binds to the cytoplasmic CNBD and modulates the cloned channel by shifting the voltage dependence to the left. The same shift can also be achieved when the C-terminus is truncated, showing that this region provides tonic inhibition of the pore which is relieved by cAMP (Wainger et al., 2001). Studying this region helps our understanding on how binding is coupled to pore opening.  1.5.1.1 The C-linker couples to the CNBD and transduces the signal to the pore In CNG channels, it is proposed that C-linker is coupled to the activation of the transmembrane core and also enhanced ligand binding to the open channel (Gordon & Zagotta, 1995; Paoletti, Young, & Siegelbaum, 1999). Intersubunit disulphide bonds between A’ helices of neighbouring subunits show they are in close proximity during channel modulation (Gushchin, Gordeliy, & Grudinin, 2012; Hua & Gordon, 2005).  The C-linker is the coupler between ligand binding and channel activation. It has been proposed that ligand binding causes a centrifugal rearrangement in the C-linker, which translates into a tetrameric “gating ring” and increases in open probability of the channel. Changing the C-linker can therefore affect HCN channel activities. For example, a tripeptide sequence (FPN) in the C-linker, normally found in fCNGA4 channel, was substituted in HCN1 and HCN2 channels (QEK) to force the channels into a constitutively activated state, shifting the basal activation to more depolarizing potentials as though they were pre-activated by cAMP. The sequence also reverses the polarity of cAMP effect. The FPN substitution favours the apo tetrameric form and 31  cAMP promote dissociations, possibly due to proline acting as a helix breaker (L. Zhou et al., 2004).  Another set of mutations (K472E/E502K/D542E) in the HCN2 isoform disrupts salt bridges formation in the C-linker, which also results in channel activation (Craven & Zagotta, 2004).  Disease mutations were also found in the C-linker, namely R500Q and E488K which are epilepsy mutations. E488K is a loss-of-function mutation where there is a hyperpolarized shift in voltage dependence, making the channel more difficult to open (J. C. DiFrancesco et al., 2011; Tang et al., 2008). Finally, an additional cGMP binding site in C-linker of HCN4 was characterized and has been proposed to limit effect of cAMP-induced channel facilitation (Lolicato et al., 2014).  1.5.1.2 The CNBD interacts with cAMP directly and influences channel activity In a HCN1-HCN2 chimera experiment where individual domains were swapped between the two isoforms, the C-linker and CNBD have great effect on the voltage dependence, but the CNBD is the predominant contributor to the shift induced by cAMP. HCN1 and HCN2 have different gating properties and different sensitivity to cAMP. The uniqueness allows the characterization of the different domains on gating properties. In the absence of cAMP, the voltage dependence is more negative for HCN2 compared to HCN1, and swapping the C-linker and CNBD portion changes the gating activity (S. Chen et al., 2001). The CNBD normally inhibits the formation of a cAMP-induced gating ring, and the inhibitory action is more severe in HCN2 and HCN4 (P. Schweizer et al., 2010). An even greater shift in the activation gating of HCN2 is achieved if the CNBD is truncated, suggesting that cAMP normally relieves tonic inhibition of channel opening (Wainger et al., 2001). The general belief is that the CNBD of HCN1 has less inhibitory regulation by the CNBD, leading to a smaller maximum shift in 32  channel opening by cAMP. We and others have hypothesized that the gating ring of HCN1 that facilitates opening may be more stable in the absence of cAMP as compared to the HCN2 and HCN4 channel, which correlates with a less negative half-maximal potential (Chow, Van Petegem, & Accili, 2012; Lolicato et al., 2011).  The C-helix, found at the distal end of the CNBD is critical for the effect of cAMP; deleting this region abolishes the contribution of cAMP. However, deleting the C-helix does not alter the voltage dependence of HCN2 activation basal voltage, suggesting that this region is not involved in tonic inhibition of channel opening (Wainger et al., 2001). The rearrangement of this helix after ligand binding has been studied, in HCN channels as well as related CNBD-containing proteins. Most of these proteins are modulated by cyclic nucleotides except for KCNH channels. These proteins are useful models when comparing with the structure of the CNBD of HCN channels. In the four isoforms, only HCN3 is insensitive to cAMP. A chimera between HCN4 with the CNBD of HCN3 reports that its CNBD is able to bind to cAMP and produces the activating effect. The finding suggests that the usual HCN3 CNBD makes a handicapped interaction with the C-linker, leading to the impaired effect (Stieber et al., 2005). This could be analogous to EAG potassium channel which also contains a CNBD. However, the effect is silent because of an intrinsic ligand distal to the C-terminus that blocks ligand binding (Brelidze, Carlson, Sankaran, & Zagotta, 2012; Li, Ng, Yoon, & Kang, 2014). Besides ligand binding, a motif in the B-helix of CNBD is also deemed important for exporting the channel from the endoplasmic reticulum to the cell surface (Nazzari et al., 2008).  33  1.5.2 Homologous proteins containing an HCN-like CNBD The HCN CNBD is similar to that for a number of other proteins (Figure 1.5, Table 1.1). The domain consists of a mixture of alpha-helices and an 8-strand β-roll. The β-roll contains a highly conserved PBC that anchors the ligand in the hydrophobic binding pocket. The hydrophobic residues around the pocket helps stabilize and shield the ligand. The binding of cAMP creates new interactions in the binding domain and triggers the movement of the most distal C-helix. This initiates a cascade of conformational changes that modifies interactions within and downstream of the domain. These cAMP-binding regulatory domains are linked to parts of the protein that cause catalytic or functional effects (Berman et al., 2005; Kannan et al., 2007). In our case, binding of cAMP to HCN triggers channel opening in the pore region. The homologous structures provide insights on the mechanics of ligand binding and subsequent structural changes.  34   Figure 1.5 The multiple sequence alignment of proteins containing a CNBD. There are great similarities between the conserved domain. In cases where protein has tandem CNBD repeats as PKA and PKG, CNBD-B was aligned because it was shown to bind to cAMP first. Sequence cut-offs were based on SMART. The percent homology of each protein was compared to CNBD of hHCN2 and generated by ALIGN at the GENESTREAM network server (Pearson, Wood, Zhang, & Miller, 1997). The two asterisks show two key residues, numbered as E582 for ribose binding and R591 for phosphate binding.    The proteins are similar in structure but demonstrate differences in cyclic nucleotide selectivity and configuration (syn versus anti configuration), which are likely due to the differences in primary sequence at key locations. In solution, free cAMP and cGMP prefers to be in the anti and syn configuration, respectively (Alderete, 2000; Das et al., 2009). Table 1.1 describes the similarities and uniqueness in the different structures and also their ability to bind to ligand.   35  Homologous CNBD-containing proteins Ligand modulation and configuration The CAP protein is found in bacteria and is the first CNBD structure crystallized. It functions as a transcription factor that regulates gene expression via the fluctuation of cAMP. The protein exists as a homodimer, and each subunit consists of an N-terminal regulatory domain that binds to cAMP and a downstream helix-turn-helix motif that acts as a DNA-binding domain. Cyclic binding provide allosteric transition that switches on the protein to allow strong and specific binding of DNA and to interact with RNA polymerase (CAP Botsford & Harman, 1992). The C-helix is found to undergo significant rearrangement, rigidifying the distal coil and extending the helix by three turns at the dimerization interface. This causes rotation and translation of the DNA-binding domain to position the helices, allowing DNA recognition (Dong, Malecki, Lee, Carpenter, & Lee, 2002; Passner, Schultz, & Steitz, 2000; Schultz, Shields, & Steitz, 1991; Weber & Steitz, 1987).   Cyclic AMP binds in the anti- configuration (Weber & Steitz, 1987). cGMP binds in the syn configuration but does not stimulate DNA binding due to its failure to make key contacts and elicit conformational change (Ebright, Le Grice, Miller, & Krakow, 1985; Popovych, Tzeng, Tonelli, Ebright, & Kalodimos, 2009)  CNG channels produce the primary current in photoreceptors and olfactory sensory neurons, and are open in response to cellular cGMP (CNG T. Y. Chen et al., 1993; Finn, Grunwald, & Yau, 1996; Kaupp et al., 1989). This pathway is regulated by phosphodiesterase to degrade cGMP, which is activated upon light altering the phototransduction cascade (Burns & Baylor, 2001); alternatively, binding an odorant to receptor could trigger production of cAMP in the adenylyl cyclase cascade in olfactory neurons (Frings, 2001). CNG and HCN channels are often compared because of their similarities in the carboxy-terminal domain and in the formation of heterotetramers (Bradley, Li, Davidson, Lester, & Zinn, 1994; Craven & Zagotta, 2006; M. D. Z. Varnum, W. N., 1996). A major difference is that CNG channels are insensitive to a change in membrane potential (Mazzolini, Marchesi, Giorgetti, & Torre, 2010). However, the ligand-binding process is highly relatable and serves as an excellent model for studying allosteric activation. The binding of cAMP is cooperative and it triggers a change that leads to the opening of pore via C-linker movement (Biskup et al., 2007; Craven & Zagotta, 2006; Puljung & Zagotta, 2013; Ruiz, 1999) . The post-CNBD region may be important for subunit interaction in CNG channels (Zhong, Lai, & Yau, 2003; Zhong, Molday, Molday, & Yau, 2002).  CNG is 40 times more selective for cGMP than for cAMP (Altenhofen et al., 1991).  cAMP and cIMP are found to be partial agonists whereas cGMP is a full agonist (M. D. Varnum, Black, & Zagotta, 1995). Like HCN channels, cAMP binds in the anti, while cGMP binds in the syn configuration (Altenhofen et al., 1991; Flynn, Black, Islas, Sankaran, & Zagotta, 2007; M. D. Varnum et al., 1995) Guanine nucleotide exchange factors (GEFs) regulate the activation of Rap proteins, which in turns act as molecular switches. A type of GEF is Epac, an exchange protein that is activated by cAMP (Epac de Rooij et al., 1998). The protein consists of an N-terminal regulatory domain and a C-terminal catalytic domain Cyclic AMP binds in the syn and cGMP in the anti configuration (Das et al., 2009). Both with micromolar 36  Homologous CNBD-containing proteins Ligand modulation and configuration for nucleotide exchange. The regulatory domain of Epac2 contains two cAMP-binding domains, with a DEP hairpin domain in between. Only the binding domain (CNBD-B) near to the catalytic domain is necessary for cAMP-mediated activation (de Rooij et al., 2000). The redundant CNBD-A also binds with lower affinity 84µM, whereas the essential one binds with 1µM. CNBD-B also interferes with the catalytic domain from interacting with Rap1. Cyclic AMP triggers a conformational change to the orientation of the PBC, and induces a hinge and lid movement that would relieve the hindrance of the CNBD. Unlike other CNBDs, the C-helix does not move towards the ligand upon binding. Instead, three β-strands form a lid to enclose the binding cavity. This hinge-lid region involves structural rearrangement upon binding liberates the catalytic domain from steric inhibition (Rehmann, Prakash, et al., 2003).  affinities, cAMP is an agonist but cGMP is an antagonist where it binds but fails to promote GEF activity (Rehmann, Schwede, Døskeland, Wittinghofer, & Bos, 2003). KCNH familyBrelidze, Carlson, & Zagotta, 2009  The KCNH ion channel family is comprised of EAG, ERG, and ELK channels involved with repolarization of cardiac conduction and neural excitability. Although the structural topology appears the same as carboxy-terminal of HCN channels, there are crucial differences that make these channels insensitive to cAMP ( ). Unlike the canonical CNBD, the electrostatic profile of these binding pockets in the β-roll appears to be negative, and thus repulsive to the phosphate group in cyclic nucleotides, impeding direct binding. There is also a ninth β-strand immediately following the C-helix which would occupy the usual cAMP binding site, acting as an intrinsic ligand. This “ligand” contains an aromatic residue that goes in place of adenine from cAMP (Brelidze et al., 2012; Brelidze, Gianulis, DiMaio, Trudeau, & Zagotta, 2013). The intrinsic ligand seems to have different functional effects because mutations of the tripeptide “intrinsic ligand” in hERG channels increase deactivation but no effect on half-maximal activation voltage (Brelidze et al., 2013). On the other hand, mutations in zebrafish ELK causes a depolarizing shift (Brelidze et al., 2012), where that in EAG1 channels shift the GV curve to a more hyperpolarizing potential (Marques-Carvalho et al., 2012). Mutations of the “intrinsic ligand” in hERG channels are associated with LQT syndrome. Another surprising feature is in the C-linker region and the “elbow-to-shoulder” binding motif. HCN carboxy terminal subunits are held together in a four-fold symmetry, where zELK is by a two-fold, with significant rotation around the A’ helix. The “elbow” of ERG channel, to a greater surprise, rests on the shoulder of the same subunit due to a longer loop between the B’ and C’ helices and a longer D’ helix. The intrasubunit interface means ERG channels are monomeric (Brelidze et al., 2013).   Because of the intrinsic ligand in place of where cyclic nucleotide would bind, cAMP and cGMP are not found in the crystal structures and they do not exhibit any effect on the channel. 37  Homologous CNBD-containing proteins Ligand modulation and configuration MlotiK channel has great resemblance with HCN channels because not only does it contain a CNBD, it also contains similar transmembrane domain and the signature GYG potassium-selective sequence. As others, B- and C-helix move towards each other in the bound form, but they stay as rigid bodies in the transition with no changes in secondary structures (MlotiK1 potassium channel Schünke, Stoldt, Lecher, Kaupp, & Willbold, 2011). Atomic force microscopy (AFM) imaging has shown the two states of the channel: the CNBD is more well-defined and ordered in the bound state, and clearly more structurally varied in the unbound state (Mari et al., 2011). A significant difference is in the length of the C-linker because MlotiK C-linker is only 20 residues long, compared to 80 in HCN or CNG channels. Since C-linker is important for intersubunit contacts, these channels do not form oligomers (Clayton, Silverman, Heginbotham, & Morais-Cabral, 2004). Although the crystal structure of the CNBD fragment shows a dimer, linked by the short C-linker, the domain is found to exist in an independent manner and that cAMP binding is non-cooperative. The cryoelectron microscopy image confirms that even when the full channel arranges in a four-fold symmetry as potassium channels, the CNBDs from each subunit are distinguished with discrete gaps (Chiu et al., 2007); the NMR study also confirms a monomeric subunit at high cAMP concentration (Schünke et al., 2011). Using this channel, the gating mechanism of cAMP binding was proposed, where binding triggers CNBD to move closer to the membrane, tilts the VSD, and twist the S5-P-S6 pore into an open conformation (Kowal et al., 2014).  Cyclic AMP is around 10 times more potent than cGMP (Altieri et al., 2008).Introducing three mutations in the C-helix and β4 -β5 loop can invert the selectivity from cAMP to cGMP (Pessoa, Fonseca, Furini, & Morais-Cabral, 2014). Cyclic AMP binds in the anti (Clayton et al., 2004), while cGMP binds in the syn in both wild type and the triple mutant (Altieri et al., 2008).  PKAC. Kim, Cheng, Saldanha, & Taylor, 2007  Protein kinase A is ubiquitous in mammalian cells and a common receptor for cAMP, and phosphorylates many cellular processes. The C-terminus contains two tandem cAMP-binding domains like Epac, CNBD-A and CNBD-B, and a connecting linker region that blocks the active site of the catalytic subunit (; C. Kim, Vigil, Anand, & Taylor, 2006). The PBC is thought to shield the ligand from degradation by phosphodiesterases (Diller, Xuong Madhusudan, & Taylor, 2001). The helical structures are largely re-organized upon cAMP docking to the β-roll. Ligand binding is highly cooperative and occurs in a sequential order, where cAMP binds to CNBD-B first, and its C-helix re-orients into a hydrophobic cap that favours binding to CNBD-A (J. B. Wu, S.; Xuong, N.H.; Taylor, S.S., 2004). This locks the adenine into the β-roll and the new position favours the release of the inhibition on the catalytic subunit, activating kinase activity (Vigil et al., 2006).   While both agonists, PKA has 200 times higher selectivity for cAMP than cGMP (Shabb, Ng, & Corbin, 1990). Both ligands bind in the syn configuration (Su et al., 1995). The local steric clashes of cGMP reduces binding affinity (Das et al., 2009). 38  Homologous CNBD-containing proteins Ligand modulation and configuration Protein kinase G is the main downstream modulator of the nitric oxide-cGMP signalling pathway, and it phosphorylates substrates that are, for example, involved in smooth muscle tone and platelet activation. The protein contains a leucine zipper for homodimer formation, an auto-inhibitory sequence that blocks the catalytic cleft, and two tandem CNBDs followed by the catalytic domain (PKG J. J. Kim et al., 2011). The protozoan isoform shows a series of structural changes upon cGMP binding, which includes formation of the P-helix towards the β-roll, migration of B-helix towards the PBC, departure of the N3A (E’-F’-A helices) motif from the PBC, and capping of the pocket by the C-helix. The last movement disrupt the interaction between the regulatory and catalytic domain (J. J. Kim et al., 2015).   Cyclic GMP binds in syn and cAMP binds in either configuration, where the variability is indicated by high B-factors (J. J. Kim et al., 2011). The selectivity for both ligands are similar at CNBD-A, but affinity for cGMP is 240 times higher at CNBD-B (G. Y. Huang et al., 2014; J. J. Kim et al., 2011). A recent structure of a prokaryotic CNBD-containing channel was published. It is found in Spirochaeta thermophile and binds to cAMP and cGMP. Its functional properties closely resemble HCN or CNG channels in eukaryotes, gated by intracellular cAMP to produce current that conducts K+ and Na+ ions. It is insensitive to changes in voltage. Like eukaryotic channels, it has a C-linker and a CNBD. The unseen feature is that cGMP is an antagonist in this channel, where ligand binding induces no effect to channel activation (SthK channel Brams, Kusch, Spurny, Benndorf, & Ulens, 2014). The advantage of this model is in two co-crystal structures, one with each ligand. The co-crystal with cGMP is similar to the published HCN crystal structures in the resting state. The co-crystal with cAMP is captured in the active conformation, which is not possible previously due to the favouring of the stabilized resting but bound form (Kesters et al., 2015).  Cyclic GMP is an antagonist of the channel whereas cAMP is an agonist. Both of the ligands bind in the anti configuration. cGMP in anti- configuration is unable to establish contacts with the C-helix and to bind to the conserved threonine in PBC (Kesters et al., 2015).   The related channel found in sea urchin sperm was one of the first cloned HCN channels (SpIH Gauss et al., 1998). The electrophysiological properties are similar to that of HCN channels where they are activated by hyperpolarizing potentials and modulated by cAMP. The biggest difference is that in the absence of cAMP, Ih from SpIH undergoes inactivation. The binding of cAMP removes inactivation and increases open probability, but the signature cAMP-induced shift is absent (Gauss et al., 1998). Truncating the C-terminal region abolishes both the inactivating current and sensitivity to cAMP, implying that, like mammalian HCN, cAMP relieves the auto-inhibitory mechanism imposed by the CNBD (Flynn & Zagotta, 2011; Wainger et al., 2001). The intracellular gate is still thought to be coupled to both voltage sensor and CNBD, but the In SpIH, cAMP is a full agonist while cGMP is a partial agonist. The crystal structure shows cAMP binds in the anti configuration, but that with cGMP was not solved yet (Flynn et al., 2007; Flynn & Zagotta, 2011). 39  Homologous CNBD-containing proteins Ligand modulation and configuration inactivation is caused by an uncoupling slip between the gate and the voltage sensor that can be stabilized by cAMP (K. S. Shin, Maertens, Proenza, Rothberg, & Yellen, 2004). SpIH is a model used to describe partial agonism and on ligand selectivity. Two mutations that mimic CNGA1 channels can convert cGMP to a full agonist and facilitate maximal opening. Cyclic GMP mildly stabilizes the open state of SpIH while cAMP provides greater stabilization (Craven & Zagotta, 2006; Flynn et al., 2007).  Table 1.1 CNBD-containing proteins. CNBD-containing proteins are compared to HCN channels in terms of their structures, their binding configuration and conformational changes upon ligand binding. The list of proteins is ordered alphabetically.  1.5.3 Structure of the C-terminal domains in HCN channels 1.5.3.1 First crystal structure reveals the binding pocket  To focus on characterizing the effect of cAMP, many biochemical tests were performed on an isolated piece located in the C-terminus, starting immediately after the transmembrane segments up to the end of the CNBD. It resides in the cytoplasm and therefore is soluble in physiologically-relevant buffer solutions. The soluble piece is named C-linker/CNBD here in the thesis. The purified C-linker/CNBD of HCN2 channel was first visualized by x-ray crystallography in 2003. The protein is 200 amino acid residues in total, consisting of the C-linker (80 residues, i.443-533) and CNBD (120 residues, i.534-645) (Figure 1.6). The C-linker contains six alpha helices labelled A’ to F’, whereas the CNBD is comprised of 4 alpha helices (A, P, B, C) and 8 beta strands forming a jelly-roll topology between the A helix and B helix. C-helix ends the crystal structure (Zagotta et al., 2003). The CNBD of HCN2 folds structurally similar to that of other previously solved CNBD-containing proteins such as CAP or PKA. 40  Similar to CAP, the binding domain contains eight anti-parallel β -strands and orients themselves to form a β-roll (Weber & Steitz, 1987). Similar to the folding scheme of PKA, the pocket contains an additional P helix embedded in the β -roll motif between strand 6 and 7 (C. Kim et al., 2006). Other published structures from CNBD-containing channels are listed above where the CNBD are mostly similar, and they differ more in the post-binding transition (Table 1.1). Oligomerization of individual subunits and post-binding movement of the C-helix seem to be common themes among these homologous proteins. The crystal structure of HCN2 C-linker/CNBD appears to have a four-fold symmetry, consistent with the fact that the full channel is probably a tetramer (a structure of the full channel has not been reported). Most of the intersubunit interactions are located in the C-linker. Such interaction resembles an elbow (A’ + B’ helices) of one subunit resting on top of a shoulder (C’ + D’ helices) of the neighbouring subunit, until four individual subunits are connected. This helix-turn-helix motif involves a mixture of hydrophobic interactions (isoleucine and leucine residues), hydrogen bonds (tyrosine residues), and a salt bridge (K472 and E502) (Zagotta et al., 2003).  Cyclic AMP makes 7 interactions with the β -roll and C-helix according to the crystal structure and molecular simulations (Zagotta et al., 2003; L. Zhou & Siegelbaum, 2007). The phosphoribose is buried deep in the binding pocket and the nitrogenous base is near the entry site. The phosphoribose interacts only with the β-roll, while the nitrogenous base has hydrophobic interactions with both the β -roll and C-helix (Figure 1.7). The most conserved feature inside the β -roll is the PBC, consisting of the P helix and the β6 and β7 strands. In the β-roll, the 3 residues in contact are R591, T592 and E582. The 4 residues in the C-helix are R632, 41  R635, I636, and K638. These residues are also important for cAMP selectivity over cGMP (Zagotta et al., 2003).    Figure 1.6 Sequence alignment of the C-terminal region in the four HCN human isoforms. The secondary structure is assigned based on the crystal structure (Zagotta et al., 2003). The light grey indicates the C-linker whereas the dark grey is the CNBD. The primary sequences are very similar between isoforms, especially between the two cardiac isoforms.   1.5.3.2 All published HCN carboxy-terminus are structurally similar   Following HCN2, the crystal structures of the C-linker/CNBD piece of HCN1 and HCN4 isoform were solved. It is also no surprise that the C-linker/CNBD of HCN4 also turned out to be structurally similar to that of HCN2 since both isoforms are sensitive to cAMP. HCN4 is unique in its long C-terminus that comes after the binding domain, but the structure is not yet known (Xu, Vysotskaya, Liu, & Zhou, 2010). HCN1 has a weak dependence on cyclic nucleotides in the full-length channel. HCN1 channels seem to be pre-activated, reducing the amount of CNBD-basal inhibition and cAMP-induced shift. Surprisingly, the global structures of C-linker/CNBD 42  of the three isoforms are almost identical (Lolicato et al., 2011) (Figure 1.7). The C-linker/CNBD of HCN3 has not yet been solved.  The apo structure of the HCN2 C-linker/CNBD was also crystalized, but was not informative. Since the binding of cAMP favours the open, active state of the channel, the unbound structure was expected to show a different and inactivated state which is not seen in the holo structure, illustrating potential conformational changes upon binding. The cAMP-free structure was solved with the addition of bromine ions in the crystallizing condition; unfortunately, this negative ion resides in the pocket and binds to R591 of PBC, and hence, mimicking the anionic phosphate group of cAMP. Thus, the overall structure remains unchanged compared to the holo structure because of the bound bromine or it could be because of crystal packing artefacts. The apo structure does show some subtle differences from the holo structure. For example, it does show an increase in the flexibility of the distal C-helix. There is a lack of electron density after residue L633 in the apo structure which suggests that the distal end of the C-helix takes on a looser coiled conformation in the absence of cAMP. The ligand stabilizes and favours the formation of the secondary structure (Taraska, Puljung, Olivier, Flynn, & Zagotta, 2009). Below, we summarize different experiments which suggest that the C-helix moves and undergoes a coil-to-helix transition upon cAMP binding. 43     Figure 1.7 Structural similarities between the C-linker/CNBD segment in solution.  The three published co-crystal structures of HCN channels and cAMP were aligned with PyMOL. The subunits are structurally identical and cAMP binds in the pocket with the same orientation. The biggest discrepancy seems to be in the flexible β4-β5 loop. The structure files were obtained from protein database: 3U0Z (green) (Lolicato et al., 2011), 1Q5O (blue) (Zagotta et al., 2003), and 3OTF (pink) (Xu et al.).  1.5.3.3 Cyclic GMP binds to the HCN2 isoform in the syn configuration The crystal structure of HCN2 C-linker/CNBD with cGMP was also solved, and, the overall topology was identical to the topology with cAMP. In the binding pocket, however, cAMP binds in the anti- configuration while cGMP binds in syn. The functional data suggests that the different configurations and different intermolecular interactions change the potency but they do not alter the maximal effect on channel opening (L. Zhou & Siegelbaum, 2007). The rotation about the glycosidic bond allows cGMP to make additional contact in the cyclic 44  nucleotide pocket. The 2-NH2 group on the guanine ring forms an extra hydrogen bond with hydroxyl group of T592 (Zagotta et al., 2003). In HCN4, a similar structure was solved, closed but resting state with cGMP (Lolicato et al., 2014). In HCN4, the structure of the C-linker/CNBD bound to cGMP was similar to that of HCN2. Unique to HCN4 C-terminus is that cGMP also binds to the C-linker between B’ and C’ helices, near the intrasubunit interface, and interacts with β8 of the jelly roll. At this site, cGMP also binds in the syn conformation, forming hydrogen bonds with Y559, K562, F564, E566, and R680. F564 with the 6-keto and Y559 with 2-NH2 in the guanine ring explain why this site is unique to cGMP and not cAMP. This could serve as an additional regulatory mechanism on channel activation. Cyclic dinucleotide binds in the non-canonical site and antagonizes binding of cAMP to the canonical site, and limits the effect of cAMP modulation of beating frequency in sinoatrial node myocytes (Lolicato et al., 2014).   1.5.3.4 Selectivity for cGMP Cyclic AMP is regarded as the natural ligand for HCN channels because the EC50 is sub-micromolar and it is at least 10 times more potent than cGMP in facilitating If in the sinoatrial node and in cells expressing the HCN2 channel isoform (D. DiFrancesco & Tortora, 1991; Ludwig et al., 1998; L. Zhou & Siegelbaum, 2007). Selectivity of cAMP and cGMP in HCN isoforms is dictated by the residues and preferable contacts in the binding pocket, similar to that in PKA and PKG. The specificity was first speculated from cAMP-bound structure with CAP (Weber & Steitz, 1987), and later from cGMP-bound structures for HCN2 and SpIH, the HCN isoform found in sperm of Strongylocentrotus purpuratus  (Flynn et al., 2007; Gauss et al., 1998; Zagotta et al., 2003). In SpIH, cGMP is a partial agonist, where the amount of current is reduced 45  even with saturating amount of cGMP. The carboxy-terminal segment of SpIH did not co-crystallize with cGMP, possibly due to the inhomogeneity of the protein as the ligand does not stabilize the conformational changes as a full agonist would. The homologous position T592 in HCN2 is a valine (V632 in SpIH), denying hydrogen bond capability from the side chain. Mutating the residue into a threonine improves binding but is not sufficient to promote full agonism. Another mutation I665D is also introduced since aspartate residue interacts with the ligand in cGMP-favoured CNG channels (M. D. Varnum et al., 1995). Aspartate is able to make hydrogen bonds with N1 in the ring as well as 2-NH2 group. This mutation alone does not invert selectivity; however, together with V633T, cGMP is able to shift the voltage dependence to the same extent as cAMP. The structural model suggests the threonine helps initial binding of cGMP in the syn conformation and aspartate interacts with the ligand during allosteric conformational change for channel opening (Flynn et al., 2007). The same aspartate mutation is performed in HCN2 (I636D) and it improves the potency for cGMP by 6 times (L. Zhou & Siegelbaum, 2007).   1.5.4 Formation of the cytosolic gating ring The gating ring is thought to form in the cytosolic domains distal to the S6 transmembrane helix, and is promoted upon adding cAMP.  The idea of a gating ring arose from the increased population of C-linker/CNBD dimers and tetramers upon adding cAMP, measured by analytical ultracentrifugation (Zagotta et al., 2003). This ring in the full-length channel is responsible for relieving the inhibitory stress caused by the CNBD and allowing channel facilitation (Wainger et al., 2001). The ability to promote tetramer formation upon adding cAMP was reinforced by data obtained by dynamic light scattering (DLS), which measures the apparent 46  radius of the protein. The radius of the C-linker/CNBD piece increased by four-fold with the addition of cAMP (Chow et al., 2012). Removal of the proximal helices of the C-linker abolishes tetramer formation.  A second theory suggests that interactions found in the shoulder-elbow motif have to be broken to achieve the cAMP-induced effect. The bound HCN2 C-linker/CNBD structures may represent a bound but inactivated state as suggested by a functional and mutagenesis study on two salt bridges that exist in that crystal structure. The salt bridges were between lysine 472 (B’ helix) and glutamate 502 of the D’ helix in the neighboring subunit and between K472 and aspartate 542 (β roll) of the same subunit. When charge was reversed by mutation, the absence of both salt bridges (K472E) abolished the effect of cAMP on the activation curve and significantly shifts the voltage dependence to more depolarized potential, mimicking the effect of cAMP (Craven & Zagotta, 2004). Thus, it was proposed that cAMP binding breaks these salt bridges, disrupting intersubunit interactions, and facilitates pore opening by hyperpolarization (Craven, Olivier, & Zagotta, 2008; Gushchin et al., 2012). It is still unclear which gating ring theory is correct. In the thesis, evidence is provided in favour of the notion that interactions between subunits and formation of a gating ring are promoted by cyclic nucleotide binding.  1.5.5 Dynamic movement of the carboxy terminal domains 1.5.5.1 C-helix of the CNBD moves upon ligand binding  The dynamic model of the C-linker and cyclic nucleotide binding domain is supported by many biochemical experiments. Truncation of the C-helix or mutation in one residue (R632A) in this region is sufficient to abolish the effect of cAMP in the HCN2 isoform (Wainger et al., 2001; L. Zhou & Siegelbaum, 2007). This helix is first tested in CNGA1 channel of retinal 47  photoreceptors and they proposed that the β-roll and C-helix comes close together during the opening allosteric transition (M. D. Varnum et al., 1995). This movement and the convergence of the two pieces are also modeled in SpIH, the sea urchin HCN channel, which is required in upgrading cGMP from a partial to a full agonist (Flynn et al., 2007).  The capping of C-helix upon ligand binding is also observed by large structural rearrangement in PKA, EPAC (exchange factor directly activated by cAMP), and MlotiK (potassium channel from Mesorhizobium loti) proteins. In fact, there were six conformations of the C-helix in the apo-MlotiK1 CNBD solved by X-ray crystallography, suggesting the distal part of the C-helix is flexible and can take on different orientations (Altieri et al., 2008).  Using biochemical methods, the conformational dynamics of the C-helix in CNG channels was confirmed. Using disulfide crosslinks in the C-helix, the added bridges inhibit channel activation and keep the C-terminus in the closed state. These interacting regions move apart during normal opening mechanism (Matulef & Zagotta, 2002; Taraska & Zagotta, 2007). Similarly, substituted cysteine accessibility method (SCAM) experiment located a residue that is accessible only in the unbound state, and another that is accessible only in closed state. These experiments were simple proofs that cyclic AMP binds to the β-roll, and that an allosteric transition involves the relative movement of C-helix towards the β-roll (Matulef, Flynn, & Zagotta, 1999).  The movement is further observed using more sensitive methods, transition metal ion fluorescence resonance energy transfer (FRET) and DEER. FRET is a powerful system transferring energy from the donor to its partner, and thereby measuring the change in molecular distances. Although there is no direct visualization of the interaction, the method effectively measures the degree of movement in real time, with no restriction to the size of the protein. 48  Using transition metals, even shorter distances can be detected. FRET, unlike crystallography, is able to measure the changes in solution, mimicking the physiological behaviour of the protein. Upon ligand binding, the proximal (near N-terminal end) C-helix moves around 5Ǻ towards the β-roll at the core of the CNBD, consistent with SCAM experiments on CNG channels. The movement also appears to be coupled to the movement of the F’ helix of the C-linker, and the movement could propagate to the pore to relieve inhibitory stress (Puljung & Zagotta, 2013; Taraska et al., 2009). To consolidate that finding, DEER spectroscopy uses electron paramagnetic resonance on a pair of location. The coupling between the two sites, with spin-labelled at mutated cysteine residues, can be detected via excitation and probing of the partners, and the oscillation signal from the microwave pulses would measure the time traces, which are later converted to distance distributions. This technique shows the proximal C-helix moves up to 9Ǻ closer to the β-roll. The distal end also moves closer to the β-roll, sealing the pocket. The β-roll itself did not move upon ligand binding, consistent among crystal structures of related proteins (Tibbs, Liu, Leypold, & Siegelbaum, 1998; L. Zhou & Siegelbaum, 2007). Residues in the β-roll have a small role in ligand efficacy and remain relatively stationary. On the other hand, the C-helix was found to be mobile in the absence of cAMP, from the heterogeneity from DEER studies (Puljung, DeBerg, Zagotta, & Stoll, 2014). This was also observed in flexible, unresolved region in the apo crystal structure as well as the dynamic behaviour in NMR; the helix is restricted in conformation when ligand is introduced (Taraska et al., 2009).   These movements have been observed in similar proteins, where there is big C-helix dynamics upon ligand binding in MlotiK1 potassium channels (Schünke et al., 2011; Schunke, Stoldt, Novak, Kaupp, & Willbold, 2009). The same movement was also proposed for PKA (J. B. Wu, S.; Xuong, N.H.; Taylor, S.S., 2004) and Epac (de Rooij et al., 2000). Although PKA and 49  Epac are not ion channels, the C-helix rearrangement also leads to protein activation, where it relieves the otherwise auto-inhibited conformation.  Similarly, it was proposed that this transition of C-helix in channels, from a less ordered state to a helix, is coupled to channel opening (Puljung & Zagotta, 2013).  1.5.5.2 Subsequent C-linker rearrangement Upon binding, the C-helix is thought to re-organize and extend the helical structure, and move towards the binding pocket. This dynamic conformational change can be more clearly illustrated with performing nuclear magnetic resonance (NMR) on the C-linker/CNBD piece of HCN channel. Crystallography only captures the structure at an energetically stable form under favourable buffer solutions. Despite being a useful technique, there is still the caveat of whether the structure is relevant in vivo. For HCN channels, the caveat was also the inability to capture the active conformation of the C-linker. An independent visualizing technique is NMR. The advantage of NMR is its ability to observe dynamic motion in the structure during binding event from the difference in chemical shifts. So far, the experiments were only conducted with the soluble C-linker/CNBD rather than the full channel due to size constraint. Portion of the C-linker was also cleaved off to abolish intersubunit interactions and minimize unnecessary noise. This also brings down the protein size to a suitable range for NMR (Akimoto et al., 2014; Saponaro et al., 2014).   The earlier NMR solution to HCN4 C-linker/CNBD segment provided insights on the dynamic mechanism between binding and tetramerization. The main transition is between the N3A (E’, F’ and A helices) motif and the B- and C- helices. These individual bodies remain rigid 50  but the relative position dictates the state of the C-linker. In the auto-inhibited (inactive) state, the B- and C- helix are flipped away from the core of the β-roll, which allows the N3A to remain in close proximity to the β-domain. Movement of N3A motif is also proposed in protein kinase structures (J. J. Kim et al., 2015; Kornev, Taylor, & Ten Eyck, 2008). In this state, the tertiary structure creates a steric clash between the β-subdomain and the shoulder-elbow motif and destabilizes tetrameric form. In the active state when cAMP binds, the orientations of the two motifs are swapped. C-helix moves closer as it caps and shields the ligand (Akimoto et al., 2014). The capping was observed in other related proteins. During the capping process, the N3A motif is flipped out, and the steric hindrance in the shoulder-elbow motif is relieved to allow a stably bound tetramer, supporting the fact that cAMP promotes tetramerization (Lolicato et al., 2011; Zagotta et al., 2003). The model is consistent with previous findings. Removal of C-helix makes the channel constitutively auto-inhibitory because the C-helix movement is required to initiate the activation process (P. Schweizer et al., 2010; Wainger et al., 2001). Complete truncation of the CNBD abolishes auto-inhibition because the steric clash from β-subdomain is removed (Wainger et al., 2001). The stationary β-roll and dynamic F’ and C-helix are consistent with findings from DEER spectroscopy (Puljung et al., 2014; Puljung & Zagotta, 2013).  The NMR solution of HCN2 C-linker/CNBD demonstrated the conformational changes within the CNBD and C-linker upon ligand binding. The series of conformational changes starts with the folding of the P-helix between β6 and β7 upon ligand docking in the PBC and the rearrangement of hydrophobic residues in the β –roll. The C-helix undergoes a coil-to-helix transition and elongates from the distal end, and both B- and C- helix moves towards the β-roll cavity. Finally, the F’ helix forms and displaces the N3A (or N-helical bundle with E’ and F’ helices). The upward movement of the N-helical bundle is the critical movement in cAMP 51  modulation and in signalling pore opening (Saponaro et al., 2014). The upward translation of the bundle has been a conserved mechanism among CNBD-containing proteins (Rehmann, Wittinghofer, & Bos, 2007). For example, in MlotiK1 channels, an upward movement is observed using AFM and the C-linker moves closer to the cell membrane (Mari et al., 2011). This series of movement is consistent as observed in the open conformation for SthK channels, a CNBD-containing homologue from Spirochaeta thermophile that captured the C-linker in the bound-and-active state in the presence of agonist cAMP. The widening of the “gate” in the C-terminus confers the activated form.  The cAMP-bound structure measures 28.6Ǻ, compared to the resting structure with cGMP that is only 23.8 Ǻ. This is also comparable to the measurement of the open gate in Kv1.2 channels (29.7 Ǻ) and MthK channel (30.3 Ǻ) (Long, Campbell, & Mackinnon, 2005). The widening is due to outward movement of the P-, B- and C- helices in the CNBD, but the β-roll shows no conformational difference. In the C-linker, D’, E’, and F’ helices appears the same as the closed structure, but the C’ helix shifts outwards away from the pore and the A’ helix bends towards the membrane (Kesters et al., 2015). The outward movement of the C-linker was also seen in HCN4 (Akimoto et al., 2014), and the bending might be similar to the twist from MlotiK1 channels (Kowal et al., 2014). A bend of the A’ helix is also seen in the proline breaker of the tripeptide mutation, which also leads to a depolarizing shift in the activation curve (L. Zhou et al., 2004).  1.5.6 Functional assessment of ligand binding to the CNBD 1.5.6.1 Residues important for binding interactions To functionally identify residues important for cyclic nucleotide-binding and effect, single point mutations of residues have been made and their effect on function has been 52  ascertained (L. Zhou & Siegelbaum, 2007). In this study, residues were chosen based on their proximity to the ligand in the crystal structure of the bound HCN2 C-linker/CNBD and by molecular dynamics (MD) simulation using this structure. Importantly, none of the reported single mutations made in this region of the HCN2 structure affected gating in the absence of ligand. These mutations, however, reduced the potency of cAMP and cGMP, suggesting that these residues contribute to ligand modulation. The extent to which binding affinity influences potency is not clear. In Chapter Two of the thesis, we tested the strength of direct binding of ligand to the C-linker/CNBD with the same alanine substitutions to isolate the role of the residues on affinity.   (a) Phosphate binding cassette  According to MD simulation, the strongest interaction between the ligand and CNBD is the electrostatic interaction between R591/T592 and the cyclic phosphate moiety of the ligand. Arginine 591 is found in many HCN homologues, as well as the CNBD-containing proteins described (Berman et al., 2005; Jackson et al., 2007). It resides in the β6-β7 linker, the buried core of the binding pocket, and directly interacts with the equatorial oxygen of the phosphate moiety. Molecular dynamics shows this electrostatic interaction with the ligand is the strongest among residues. A charge-reversal mutation (R591E) creates repulsion with the cyclic nucleotide and is sufficient to abolish any cAMP-induced response (S. Chen et al., 2001; Tibbs et al., 1998; Ulens & Siegelbaum, 2003). The alanine substitution of this arginine (R591A) causes a 20-fold reduction in potency (L. Zhou & Siegelbaum, 2007). Using the analogue Rp-cAMP, which replaces the equatorial oxygen of the cyclic phosphate group with a sulphur atom, also disrupts the electrostatic interaction and reduces the binding affinity by 130 times (Möller et al., 2014), . 53  Threonine 592 is conserved in the β6-β7 linker of HCN channels, as well as in PKG and CNG channels, which select for cGMP over cAMP (Altenhofen et al., 1991; Shabb et al., 1990). A single substitution at this position (Ala into Thr) changes the preference of PKA for cGMP (Shabb et al., 1990). The crystal structure of HCN2 shows that the residue forms two hydrogen bonds with the axial oxygen in the phosphate moiety. The residue is important for cGMP selectivity because it makes an extra hydrogen bond with the exocyclic 2-NH2 in syn guanosine ring. In the alanine substitution (T592A), the ligand-induced shift of cAMP remains the same while it reduces by 25% for cGMP. The potency is slightly reduced for cAMP but dramatically reduced for cGMP (L. Zhou & Siegelbaum, 2007). The analogue Sp-cAMP, which replaces the axial oxygen of the phosphate group with a sulphur atom, lowers the binding affinity by 16 times (Möller et al., 2014). In lower organism, for example for sea urchin or Drosophila, this residue position is a valine which denies hydrogen bond capacity in the side chain. As a result, cGMP becomes a partial agonist in SpIH because the syn conformation is no longer stabilized. Threonine mutation helps stabilize cGMP binding (Flynn et al., 2007).   (b) β6-β7 linker for ribose binding   A glutamate (E582) side chain forms a highly conserved interaction with the 2’OH in the ribose moiety of cAMP (Berman et al., 2005). Alanine substitution of this residue (E582A) reduces the potency of cAMP by 430 times (L. Zhou & Siegelbaum, 2007). The hydrogen bond can also be disrupted by capping the 2’-hydroxyl via methylation using the analogue 2’-O-Me-cAMP (Möller et al., 2014) (Appendix 8C). The analogue does not bind to any of the HCN isoforms, indicating that the absence of the bond or the new steric clash eliminates binding. 54  Methylation allows specificity of Epac because whereas the methylated cAMP interferes binding to PKA and HCN channels, it acts as an activator both in vitro and in vivo (Enserink et al., 2002).  The side chains of two residues, E582 in the β-roll and R632 in the C-helix, form an internal salt bridge within the binding pocket. The salt bridge itself is not essential for gating because although the R632A mutation completely abolishes the effect of cyclic nucleotides, the E582A mutation still responds to cNMP. The glutamate would cause repulsion with the negatively charged phosphate, but its side chain stabilizes 2’OH in the ribose moiety. E582A mutation reduces the potency tremendously.    (c) C-helix interacts with the nitrogenous base  Residues in the C-helix of CNG channels are thought to move during the gating event (Mazzolini, Punta, & Torre, 2002), are also involved in cNMP efficacy and selectivity in related CNG channels (Goulding, Tibbs, & Siegelbaum, 1994; M. D. Varnum et al., 1995). In the HCN2 channel, alanine substitutions of single residues in the C-helix were examined for their effects on cAMP facilitation of opening (L. Zhou & Siegelbaum, 2007). Most of the single alanine substitutions in the region compromised potency without altering the maximum effect of cAMP. However, alanine substitution of arginine 632 (R632A) abolishes the cAMP-induced shift in activation curve, This, R632, which makes key interactions with the ligand, is critical for mediating the full allosteric effect of cAMP on channel opening. Alanine substitutions of R635 and K638 reduced the potency of cAMP to a greater extent than cGMP, while alanine substitution of I636 enhanced the potency of cGMP and reduced the potency of cAMP. Thus, these residues influence selectivity for cAMP over cGMP.  55  Isoleucine 636 in the HCN2 CNBD is conserved in the sea urchin HCN called SpIH, in which cGMP acts as a partial agonist; cGMP is a full agonist in the HCN2 channel. Mutation of this isoleucine to an aspartate residue in SpIH converts cGMP into a full agonist (Flynn et al., 2007). Mutation of isoleucine 636 to an aspartate increases the potency of cGMP in the HCN2 channel while a crystal structure of the mutant HCN2 C-linker CNBD shows that a stabilizing hydrogen bond is introduced by aspartate, which may be responsible for enhancing binding (L. Zhou & Siegelbaum, 2007).   1.5.6.2 Cyclic AMP binds to the tetrameric form of the C-linker and CNBD with negative cooperativity  Using isothermal titration calorimetry (ITC), our laboratory has shown that cAMP binds to the HCN2 and HCN4 C-linker/CNBD tetramer with negative cooperativity, with a stoichiometry that suggests one site binds with high affinity (0.1 µM) and three sites bind with lower affinity (~1.5 µM) (Chow et al., 2012). Cyclic AMP binds to a monomeric version of the C-linker/CNBD, which lacks part of the C-linker region, with a single binding site and an affinity that is comparable to the lower affinity measured in the tetrameric construct (~1-2 µM). The thermodynamics parameters generated from fitting the ITC isotherms suggest that both enthalpy and entropy are favoured in the high affinity binding event, while entropy is energetically unfavoured despite a greater enthalpy in the low affinity binding event. The greater entropic cost of the second event leads to lower binding as well as the emergence of negative cooperativity. These data are similar to those obtained for cAMP binding to the Catabolite Activating Protein (CAP), a homologous CNBD-containing protein dimer which is also 56  negatively cooperative and  due to a bigger entropic compensation cost (Popovych, Sun, Ebright, & Kalodimos, 2006). Negative cooperativity of cAMP binding was also suggested subsequently in the full-length HCN2 channel (Kusch et al., 2011). Using results from confocal patch-clamp fluorimetry, the kinetics of cAMP binding was measured and modelled to determine rates and equilibrium constants of binding. Global fitting to a ten-state allosteric model suggests that cAMP binding to two sites is sufficient for the full effect and that binding to the third site is negatively cooperative. However, it is currently not clear how negative cooperativity and the values of binding obtained by fluorescence or ITC data fit with known gating models and known values of potency obtained by functional experiments.   1.5.6.3 Contributions of individual binding sites to the effect of cAMP Although four cAMP binding sites are present in the full-length HCN channels, whether all four sites are necessary to trigger the maximal effect is still elusive. One of the earlier studies used tandem tetramers and introduced the R591E mutation in various stoichiometries in the tetramer. The mutation abolished strong affinity to cAMP, and thus, ligand occupancy can be controlled by the number of subunits with the mutation. It was found that each additional functional subunit, hence each additional cAMP molecule bound, builds onto the maximal shift (V1/2), until all four sites are occupied (Ulens & Siegelbaum, 2003). The EC50 remains unchanged with binding. The study showed that binding of the first molecule of cAMP significantly facilitates channel opening, and each subsequent binding event has a smaller and non-linear increase in the maximum shift in channel activation produced by cAMP. 57  A later study reported that only two cAMP molecules have to be bound to get the maximal effect. The study used confocal patch-clamp fluorometry to simultaneously measure the conductance as well as the fluorescence based on the amount of fcAMP (fluorescent cAMP) bound. When 60% of fcAMP was bound the conductance increased to a maximum; an increase in conductance is also noted in the HCN2 isoform in addition to a positive shift in the activation curve. However, the shift of activation curve (V1/2) had not reached a maximum, unlike the increase in conductance, until all the binding sites were occupied (Kusch, Biskup, et al., 2010); the contradiction between these observations was not commented upon in that paper. Nevertheless, the data appear to be otherwise consistent between the paper by Kusch et al. and Ulens & Siegelbaum, namely that all four subunits must be occupied in order for a complete shift in channel activation.  1.6 Summary and rationale for my studies   The HCN channels play an essential role in pacemaking of the heart as they depolarize and restart the cycles of action potentials. The uniqueness in gating properties allows these channels to open during resting and hyperpolarizing membrane potentials. Their response to cAMP contributes to basal heart rate and changes in rate during sympathetic (adrenaline) and parasympathetic (acetylcholine) stimulations. Four HCN isoforms are found in mammals, each coded for by a separate gene. Based on sequence, HCN channels are structurally similar to voltage-gated potassium channels, likely being formed by four subunits coming together around a central pore. Each of the subunits contains a CNBD, which is able to bind cAMP, which facilitates channel opening. This adds another level of regulations for these channels, and opens up a new door to therapeutically controlling the heart rate using HCN channel modulators.  58  However, the mechanism underlying the effect of cAMP on the HCN channels remains unclear. So far, it has been proposed that cyclic nucleotide binds and triggers a series of conformational rearrangement in the CNBD and C-linker, which leads to pore opening. It is believed that the C-helix of the CNBD initiates by capping the cyclic nucleotide, and the rearrangement leads to a downstream movement of the C-linker, which relieves the inhibitory stress on the pore. Current knowledge is combined new data to assess and elucidate the key residues of CNBD for binding, the conformational changes involved during agonistic modulation.   In this thesis, three main issues are addressed. First, which residues in the HCN2 binding region influence binding affinity? Mutating key residues near the ligand individually, or the use of cyclic nucleotide analogues with specific chemical modifications, allows for the determination of residues that impact binding affinity and how this compares with its role in potency. The importance of individual residues in determining selectivity between cAMP and cGMP is also examined. Mapping the CNBD is especially useful for drug design to achieve strong binding and specificity.  Using analogues also allows us to mimic the effect of the mutation to confirm the effect is not due to protein defect. The main technique is isothermal titration calorimetry (ITC), which measures the heat generated by the bimolecular interaction between the protein and ligand. The machine is composed of two cells made from Hastelloy alloy to minimize heat loss and chemical resistance. The sample cell contains the purified protein and the reference contains filtered water. A mechanical syringe filled with the ligand is injected into the sample cell, and when the program begins, it progressively titrates in a small volume of the ligand at a time so that the binding site would gradually saturate. Each injection measures the energy required to 59  compensate for the temperature difference between the two cells, and the magnitude of heat is plotted. The area under each peak is then integrated to obtain the total heat from each amount of ligand. In Chapter Two, the binding affinity of cAMP to the HCN2 C-linker/CNBD with single amino-acid substituted mutations at the binding pocket were tested. We pinpointed residues that are especially important for achieving high affinity. We also drew correlation between the reductions in potency to the reduction in affinity. It is still not clear how cAMP is more potent than cGMP. In Chapter Three, the use of cAMP analogues closes in on constituents on the purine ring that are responsible for strong binding. It also solves the determinants for the anti-syn transition between cAMP and cGMP using X-ray crystallography.  Second, the relationship between gating ring formation and facilitation of opening in the HCN2 channel isoform is examined. The knowledge on the coupling between the two processes is very limited. It has been proposed that the tetrameric gating ring is allowed when cAMP binds and triggers the formation of intersubunit linkages in the C-linker, but the proposal is still contradictory, as a study also showed breaking intersubunit salt bridges facilitates channel opening. Does facilitation by cAMP correlate with the degree of oligomer formation? Alternatively, could it be the disruption of intersubunit interactions that lead to channel facilitation? One important technique used is dynamic light scattering (DLS), which indirectly measures the oligomerization state of the C-linker/CNBD. Protein molecules in solution move in Brownian motion, where bigger molecules move slower and smaller molecules move faster. When laser light is shone onto the molecule, the moving particles will scatter at different intensities. The fluctuation of the intensities yields the velocity of the Brownian motion, and can be translated into particle radius based on Stokes-Einstein relationship. The radius assumes the 60  protein is spherical or globular, and it is interpreted as the hydrodynamic radius, where water molecules surround the protein particle. The radius, after autocorrection and regularization, is manipulated into the molecular weight of the particle, based on the assumption, which in turn, determines the oligomerization states. This technique is especially efficient for the isolated C-linker/CNBD since it has the propensity to tetramerize at the C-linker interface in the presence of cAMP. In Chapter Three, we tested the effect of partial agonists on the ability to oligomerize. We also speculated the cause of partial agonism by comparing the structures of the CNBD/binding site to that of a full agonist. Thirdly, we focus on understanding the impact of disease-associated mutations of the C-linker and CNBD of the HCN2 and HCN4 channel isoforms. In Chapter Four, the two mutations we looked at are an epilepsy-associated mutation, E488K in HCN2 and a bradycardia-associated mutation, S672R in HCN4. Both of these mutations cause a hyperpolarized shift in the basal voltage dependence, but cAMP has normal maximal effect.  Furthermore, the crystal structures for both of the mutations allow the speculation on the structural differences caused by a change in amino acid. The purified protein is kept in a sealed environment and the vapour exchange with different conditions allows it to precipitate in a metastable state. The crystals are cryoprotected to minimize thermo-expansion or contraction. At X-ray light sources, a beam of electrons is directed at the crystals and based on the packing of the protein, the beam will produce a diffraction pattern with different intensity and amplitude. The structural factors can be translated into electron density using Fourier transformation. The phase problem of the structures in these chapters was solved by molecular replacement of the HCN2-cAMP co-crystal, using Phaser from the CCP4 suite. The structure (based on observed and model density) was refined with Refmac from the CCP4 suite.  61  In summary, the goal of the thesis is to investigate on the progressive events that occur during channel facilitation by cyclic nucleotides. Our findings map the binding site of full and partial agonists, reinforce the importance of the role of the C-linker and its oligomerization, and elucidate how the transitions of ligand binding and C-linker oligomerization lead to channel opening. 62  Chapter 2: Molecular Determinants of Cyclic Nucleotide Binding and Coupling of Binding to HCN2 Channel Opening  2.1 Introduction HCN channels are activated by hyperpolarization of the membrane potential and most isoforms open more easily when cAMP binds to a CNBD located in the intracellular C-terminus (D. DiFrancesco & Tortora, 1991; Gauss et al., 1998; Ludwig et al., 1998; Santoro et al., 1998). For the mammalian HCN2 isoform, EC50 values, a measure of potency, were found to be ~0.1 µM for cAMP and 10 times or more for cGMP, while both cyclic nucleotides produce the same maximum effect, a depolarizing shift of ~18 mV in the range of channel opening (Flynn et al., 2007; Wang et al., 2001; Zagotta et al., 2003; L. Zhou & Siegelbaum, 2007). However, it is not clear if cAMP and cGMP differ in binding affinity to the channel or in how strongly they couple to channel opening; both factors contribute to potency (Colquhoun, 1998). Understanding the basis of this difference is important because cGMP could effectively compete with cAMP for binding in vivo if the difference between these cyclic nucleotides were mainly in how binding is coupled to opening rather than in the binding process itself.  We have a poor understsanding of the residues in the binding pocket of HCN or CNG channels that control the coupling of ligand binding to opening. Such residues are important for understanding how ligand binding ultimately leads to channel opening because current structural and kinetic models suggest that interactions with ligand are dynamic and promote binding to the open state (S. Chen, Wang, Zhou, George, & Siegelbaum, 2007; Kusch, Biskup, et al., 2010; Kusch et al., 2011; Monod, Wyman, & Changeux, 1965; Tibbs, Goulding, & Siegelbaum, 1997; 63  Wang, Chen, Nolan, & Siegelbaum, 2002; L. Zhou & Siegelbaum, 2007). In the HCN2 channel, a previous study combined molecular dynamics simulations and single point alanine substitutions with patch clamp electrophysiology and kinetic modeling to identify eleven residues in the cyclic nucleotide binding pocket of the HCN2 isoform and determine their contributions to potency of cAMP and cGMP and their coupling to opening (L. Zhou & Siegelbaum, 2007). Alanine substitution of R632 in the C-helix almost completely eliminated the maximum effect of cyclic nucleotide and thus was identified as a residue that was important for strong coupling of the ligand to opening. In contrast, alanine substitutions of the ten other residues reduced potency of cAMP and cGMP on HCN2 channel opening with largely modest impact on maximum effect.  Here, we apply a simple experimental approach to identify residues that are important for the coupling of cyclic nucleotide binding to HCN2 opening. Importantly, we directly measure the binding affinity of cAMP and cGMP to the isolated segment containing the C-linker and CNBD of the wild type HCN2 channel as well as to mutants of this region containing single point alanine substitutions of the nine aforementioned residues (Figure 2.1). By comparing cyclic nucleotide binding affinity to the HCN2 C-terminal segment with potency in the equivalent full-length channel, we are able to unmask residues that contribute to allosteric coupling of binding to opening. Our data also suggest that potency depends upon the full-length HCN2 channel maintaining high affinity for the cyclic nucleotide throughout the gating process, which in turn depends upon residues in the C-helix that move relative to the ligand.   64   Figure 2.1 Location of residues of the cyclic nucleotide binding region that lie near to the ligand and were chosen for mutagenesis Representations of the HCN2 C-linker/CNBD which was previously co-crystallized with (A) cAMP (pdb: 1Q5O) and (B) cGMP (pdb: 1Q3E). Image generated by PYMOL. The zoomed-in ribbon representation of the binding domain depicts the relative positions of the ligands to the residues studied in this paper. Seven residues are found in the C-helix and two residues are found in the β-roll between the P helix and the 7th beta sheet. The hydrogen bonds between the β-roll and the phosphate moiety are shown as dashes.  2.2 Experimental procedures  2.2.1 Molecular biology and cloning The mouse HCN2 (i. 443-646) C-linker/CNBD template (Appendix 1B) was amplified and inserted into a modified pET28 vector via ligation-independent cloning. The HCN fragment was first screened as a stabilized and soluble construct by Zagotta et al., and was used in crystallography and other biochemical tests. The cloning was completed by a previous PhD student in the lab. In the inserted vector, the translated region contains a HMT (hexahistidine, maltose binding protein, and tobacco-etch virus (TEV) protease cleavage site ENLYFQSNA) tag. The C-linker/CNBD of interest comes after the cleavage site, with a tripeptide (serine-65  asparagine-alanine) cloning artefact sequence in the N-terminus of the C-linker/CNBD. Refer to Appendix 1A for a schematic diagram of the protein. Hexahistidine and maltose binding protein are for purification using affinity chromatography. The cleavage site is specific to TEV protease, which frees the tag from construct of interest. Single point mutations were made using polymerase chain reaction (PCR) and specific sets of primers following the Quikchange protocol (Strategene). The forward and reverse primers were manufactured by IDT, and dissolved in water to a final concentration of 100 ng/µL. Refer to Appendix 2A for the sets of primers used to flank the mutation site and alter the genetic code. Polymerization was performed by Pfu Turbo polymerase (Sigma). 5% of DMSO was added to the polymerizing reaction to destabilize double-stranded DNA. After the PCR program, the reaction products were subjected to DpnI restriction enzyme to digest the methylated template DNA at 37oC for 3 hours. The digested products were transformed in CaCl2 competent DH5α strain of Escherichia coli cells (Invitrogen) and plated for colonies. The mutated constructs were confirmed by sequencing (Eurofins Scientific) before transforming into Rosetta DE3 pLacI cells (Novagen). The cells of interest were selected with antibiotics, kanamycin (from the original vector) and chloramphenicol (from Rosetta cells). Viable cells were cultured in 2xYT media (16 g/L biotryptone, 10 g/L yeast extract, and 5 g/L sodium chloride) at 37oC, shaking at 250 rpm. Protein synthesis was induced by 0.4 M IPTG at optical density (OD) of 0.6. Cells were allowed 3-4 hours to grow and express before harvesting by centrifugation. The cell pellets were frozen at -40oC until purification. A glycerol stock (800 µL of culture at OD 0.6 and 200 µL of filtered glycerol) was prepared before induction for preserving and re-culturing the cells, and was frozen at -80oC.The glycerol stocks were again sequenced (Eurofins Scientific) to ensure the presence of the mutation.  66  2.2.2 Protein purification Harvested cells were resolubilized in 40 mL lysis buffer containing buffer “A” (250 mM KCl, 10 mM HEPES at pH 7.4), 5% (vol/vol) glycerol for stabilization, 25 µg/ml DNase I and lysozyme for cellular digestion, 375 µM EDTA at pH 8.0 as an ion chelator, and 1 mM PMSF as a protease inhibitor. The mixture is lysed by sonication in ice for 12 minutes (1 sec pulse on, 1 sec pulse off), and centrifugation at 35000x g separated soluble cell material from cellular debris. The fusion protein (HMT- C-linker/CNBD) in the supernatant/lysate was trapped in a cobalt affinity column (Talon, GE Healthcare) and could be eluted by 2 column volumes of 250 mM KCl + 500mM imidazole at pH 7.4. The HMT tag was cleaved by TEV protease inside a 3.5 kDa dialysis membrane containing 80 µL of 0.5 M EDTA (pH 8.0) and 1:1000 βME for optimal protease activity. The dialysis was done in 2 L of buffer A and 10mM βME overnight, to get rid of imidazole. The cleaved products were subjected to cobalt column again, where the tag would be trapped and the isolated C-linker/CNBD was in the flowthrough. The C-linker/CNBD has low affinity for cobalt, so an extra column volume of 6% imidazole allowed complete elution. Lastly, the C-linker/CNBD was purified by a cation exchanger (ResourceS, GE Healthcare). The protein was first charged by dialyzing in a low salt buffer (10 mM KCl, 20 mM MES, 10 mM βME) at pH 6.0 for 2 hours, and later eluted with an increasing [KCl] gradient up to 1 mM KCl, at 4oC.  Protein was dialyzed in “ITC buffer” (150 mM KCl, 20 mM Hepes pH 7.0, and 10 mM βME) overnight. Purity was confirmed with SDS PAGE. Amicon concentrators with 10 kDa cut-off were used to concentrate the protein to at least 200 µM. The concentration was determined by spectrophotometry with absorbance at 280 nm and the Edelhoch method.  67  2.2.3 Ligand preparation Cyclic AMP and cGMP were obtained in powder form from Sigma. Ligands were dissolved in water to make a stock of 10 or 20 mM. The concentration was measured by spectrophotometry, where the nitrogenous base absorbs at particular wavelength.  2.2.4 Isothermal titration calorimetry Cyclic AMP or cGMP (2 mM) was titrated, 1 µl at a time, into the sample cell containing 200 µM of respective purified protein constructs, for a total of 40 injections with ~120 seconds in between injection for equilibration. In the ITC200 instrument (Malvern), the reference cell contained filtered water.  The experiments were maintained at 25oC and the heat difference between the sample and reference cell was recorded upon each injection. With the injector spinning at 750 rpm, the heat difference at each injection interval was integrated to generate the binding isotherm. At least three replicates were compiled for each protein-ligand combination. Origin 7.0 (with MicroCal ITC add-on) was used to fit the isotherm with either a one or two-site independent binding model, depending on which gave a better fit. The affinity and thermodynamics parameters were calculated based on the fit. Ligands were titrated into buffer as negative controls, and the background was determined to be negligible. In cases where the ligands generated substantial heat, the heats at each injection were subtracted. Origin 7.0 processes the data. The change in heat from one injection to the next is calculated by  𝛥𝑄(𝑖) = 𝑄(𝑖) + 𝑑𝑉𝑖𝑉𝑜 �𝑄(𝑖) + 𝑄(𝑖 − 1)2 � − 𝑄(𝑖 − 1) 68  where Q(i) is the heat of the injection, and Q(i-1) is the heat from previous injection. dVi is the volume added per injection, into Vo, or the changing cell volume. The ΔQ(i) at each injection is plotted against the total of ligand added so far, which generates the integrated curve (bottom curve). The heat at each injection, Q(i), is measured by the instrument, and the total amount of heat is Q, or a sum of the individual heats per injection ΣΔQ. The association constant (Ka) and enthalpy (ΔH) can be determined by   𝑄 = 𝑛𝛥𝐻𝑃𝑇𝑉𝑜2 �1 +  𝐿𝑇𝑛𝑃𝑇 + 1𝑛𝐾𝑎𝑃𝑇 −  ��1 + 𝐿𝑡𝑛𝑃𝑇 +  1𝑛𝐾𝑎𝑃𝑇�2 −  4𝐿𝑇𝑛𝑃𝑇�  where PT is the total protein concentration in the sample cell, corrected for the dilution by the injecting ligand. LT is the total corrected ligand concentration. n is the stoichiometry. Finally, the laws of thermodynamics allow the determination of the other parameters using  𝛥𝐺 = −𝑅𝑇 ln(𝐾𝑎) ,𝑤ℎ𝑒𝑟𝑒 � 1𝐾𝑎�  𝑖𝑠 𝑎𝑙𝑠𝑜 𝐾𝑑  𝛥𝐺 =  𝛥𝐻 − 𝑇𝛥𝑆  so that entropy (ΔS) and Gibb’s free energy (ΔG) are also reported. As we mentioned previously (Chow et al., 2012), binding isotherms are described by the equation c = Ka[M]N, where Ka is the association constant, [M] is the total protein concentration in the cell, and N is the stoichiometry of interaction. The value for c was between 1 and 1000 for 69  most of the data used for this study, which is generally considered to yield data that is accurate (Wiseman, Williston, Brandts, & Lin, 1989). In our study, cGMP binding to the HCN2 R591A and T592A C-linker/CNBD mutants yielded c values less than 1.     2.2.5 Dynamic light scattering In order to determine the oligomerization state of the HCN2 C-terminus, we used dynamic light scattering. The protein was diluted to 12.5, 25. 50, 100, or 200 µM with ITC buffer into a 50 µL mixture, in the absence or presence of ligand (at a 10:1 ratio of ligand to protein).  The mixture was spun for 2 minutes at 6000 rpm to get rid of dust debris, before applying to a 384-multiwell microtiter plate. The hydrodynamic radii were measured at 22oC with DynaPro plate reader II (Wyatt Technology), and autocorrected and converted to molecular weights based on parameters of a globular protein. Five readings were averaged per acquisitions, and ten acquisitions were again averaged per well. Acquisitions with polydispersity over 30% were discarded. The proportion of protein at the tetrameric state is correlated from the molecular weight. Particles move in Brownian motion in solution, and large particles diffuse more slowly. Likewise, the fluctuation in light intensity is slower for larger particles. The program Dynamics 7.0 plots the average overall change in intensity over time using the autocorrelation function:  𝑔2(𝜏) = 1 +  𝛽𝑒−2𝐷𝑡𝑞2𝜏  where τ is the time interval, β is the amplitude of the correlation function dependent on the instrument, Dt is the translational diffusion coefficient, and q is a scattering vector defined by the 70  optical setup. 𝑞 = 4𝜋𝑛0𝜆0𝑠𝑖𝑛𝜃2  n0 is the refractive index, λ0 is the wavelength of the laser, and θ is the scattering angle. The size of the globular protein is a function of the hydrodynamic radius (Rh), which depends on the physical size (not mass) of the protein and its behaviour in solution such as diffusion and viscosity properties. The radius is determined by the Stokes-Einstein relation  𝐷𝑡 =  𝑘𝐵𝑇6𝜋𝜂𝑅ℎ  where kB is the Boltzmann’s constant, T is temperature in Kelvin, and η is dynamic viscosity of water at that temperature. The relationship assumes the protein is a hard sphere traveling in random motion in a non-interacting and viscosity-stable environment. Finally the distribution of Rh is represented as a histogram by regularization, smoothing out the distribution. Polydispersities are calculated for the radius distribution, which is the standard deviation of the distribution from the mean values.  2.3 Results 2.3.1 Cyclic GMP binds to the HCN2 C-linker/CNBD with negative cooperativity and with lower affinity than cAMP Cyclic GMP and cAMP facilitate opening by shifting the activation range of the HCN2 channel by the same maximum amount towards the less negative range of voltages, by about 18 71  mV, but the former is less potent by ten times or more (Zagotta et al., 2003; L. Zhou & Siegelbaum, 2007). To determine if the difference in potency between them can be explained by a difference in binding affinity to the CNBD, isothermal titration calorimetry (ITC) was used to measure and compare the binding affinities of cGMP and cAMP to the isolated C-linker/CNBD of the HCN2 channel. The HCN2 C-linker/CNBD piece we used is comprised of the six α helices in the C-linker and the four α helices and eight β-roll strands that constitute the CNBD. It has been previously shown that this piece forms predominantly tetramers when present in high enough concentrations (Chow et al., 2012). Using 200 µM of HCN2 C-linker/CNBD, we found that cGMP, like cAMP, produced a two-phase pattern in the binding isotherm that could be best fitted with a two-independent binding site model (Figure 2.2A); this yielded high and low affinity binding values of ~0.36 µM and 6.13 µM, respectively. The values of binding affinity for cGMP are lower than those for cAMP, which were 0.09 µM and 1.39 µM.  Like those for cAMP binding, the energetics of cGMP binding are different for each binding event (Figure 2.2B). The low affinity binding event is driven by a favourable change in enthalpy and an unfavourable change in entropy, whereas changes in both enthalpy and entropy are favourable for the high affinity binding event. However, high affinity binding of cGMP shows a more favourable change in enthalpy and a less favourable change in entropy than for high affinity binding of cAMP. The energetics of low affinity binding for cGMP also show a more favourable change in enthalpy, but a more unfavourable change in entropy than cAMP, yielding a smaller change in free energy of binding for the former ligand. So for both binding events, there seems to be an entropic penalty for binding of cGMP relative to cAMP, which underlies the overall decreased affinity. 72    Figure 2.2 Cyclic AMP and cGMP bind to the HCN2 C-linker/CNBD with a similar pattern of thermodynamics but with different affinities.  (A) Plots of heat produced upon progressive injections of cAMP (left) and cGMP (right) to 200 μM HCN2 C-linker/CNBD, measured by ITC. The inflections in the top plot arise from injections of cAMP where each inverted peak shows the heat difference between the sample and reference compartment. The peaks decrease in magnitude as binding sites become saturated. The lower plot shows values determined by integration of the area under the peaks from the upper plot versus the ratio of injected ligand to protein. The solid line through the values represents a two independent binding site model, which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for binding affinity are shown in Appendix 11 (cAMP) and /12 (cGMP). Both cAMP and cGMP bind with negative cooperativity. (B) Bar graph showing the thermodynamics of binding for the two binding events, a high affinity and a low affinity binding event, which were determined from the fit in A. Values in the graph represent means ± s.e.m.   73    74  Figure 2.3 Single point alanine substitutions produce mild to moderate effects on the heat released upon cAMP binding. Shown are plots of heat produced by progressive injections of cAMP to the 200 μM wild type and mutant HCN2 C-linker/CNBD. The inflections in the top plot arise from injections of cAMP where each inverted peak shows the heat difference between the sample and reference compartment. The peaks decrease in magnitude as binding sites become saturated. The lower plot shows values determined by integration of the area under the peaks from the upper plot versus the ratio of injected ligand to protein. The solid line through the values represents a two independent binding site model, which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for the binding affinity and energetics of binding are found in Figure 2.7 and Appendix 11.   2.3.2 Single alanine substitutions of CNBD residues close to cAMP and cGMP mainly reduce binding affinity A previous study (L. Zhou & Siegelbaum, 2007) showed that individual alanine substitution of nine residues lying near the cAMP binding site in the HCN2 CNBD produced mild to moderate reductions in cAMP potency without greatly altering its maximum functional effect, nor did they affect gating in the absence of cAMP. In that study, the same alanine substitutions produced both reductions and increases in cGMP potency. It is not known if these alanine substitutions affect cyclic nucleotide binding affinity or how binding is coupled to channel opening, both of which could explain the changes in potency. In order to assess the contributions of these residues to binding affinity, we again used ITC to measure binding of cAMP and cGMP to the C-linker/CNBD of HCN2 as well as to mutant versions of this protein containing the same single alanine substitutions that were previously examined by patch clamp electrophysiology.   Cyclic AMP binding to all the mutant proteins produced a two-phase pattern of binding that was best fitted with a two-independent binding site model (Figure 2.3). The values for Kd 75  are variably reduced by the alanine substitutions, suggesting that the individual side chains of these residues variably enhance binding.  Cyclic GMP produced a two-phase pattern of binding for a sub-set of the mutants, whereas the others were best fitted with a single independent binding site model (Figure 2.4). For the mutants fitted with a two independent site model, the values for Kd are moderately reduced and increased by the alanine substitutions, suggesting that the individual side chains of these residues variably enhance and reduce binding affinity. For the mutants fitted with a single binding site model, the Kd values are reduced to a greater extent than those which exhibit negative cooperativity. The lack of negative cooperativity could mean that there is a single binding event, or that the two binding events are too similar in their thermodynamic profile to be resolved in the isotherms. Among the single alanine mutants, the residues arginine 591 (R591) and threonine 592 (T592) in the β-roll have the greatest reduction in binding for both cAMP and cGMP. These two residues are found in a region known as the phosphate binding cassette and they form hydrogen bonds with oxygen atoms of the cyclic phosphate group of the cyclic nucleotides (Zagotta et al., 2003).    76    77  Figure 2.4 Single point alanine substitutions produce mild to moderate effects on the heat released upon cGMP binding. Plots of heat produced upon progressive injections of cGMP to the 200 μM wild type and mutant HCN2 C-linker/CNBD. The inflections in the top plot arise from injections of cGMP where each inverted peak shows the heat difference between the sample and reference compartment. The peaks decrease in magnitude as binding sites become saturated. The lower plot shows values determined by integration of the area under the peaks from the upper plot versus the ratio of injected ligand to protein. The solid line through the values represents a one or a two independent binding site model, which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for the binding affinity and energetics of binding are found in Figure 2.7 and Appendix 12.   2.3.3 Isoleucine 636 in the C-helix confers some selectivity for cAMP binding It was previously suggested that the isoleucine 636 (I636) residue of the C-helix of the HCN2 channel may confer some of the selectivity cAMP over cGMP (Flynn et al., 2007; L. Zhou & Siegelbaum, 2007). Consistent with this, we found that alanine substitution of I636 in the C-helix decreases the affinity for cAMP but mildly increases the affinity for cGMP. To further investigate this residue, we substituted an aspartate residue for I636 (I636D), which was shown previously to greatly enhance the potency of cGMP on the HCN2 channel (Flynn et al., 2007; L. Zhou & Siegelbaum, 2007). By ITC, we found that cGMP bound more strongly to the I636D HCN2 Clinker/CNBD than to the wild type whereas the binding affinity of cAMP for this mutant was decreased (Figure 2.5A). The difference in binding between cAMP and cGMP for the I636D mutant supports the interpretation provided by the crystal structure of the mutant HCN2 C-terminus bound to cGMP, which suggests that the added aspartate makes an additional hydrogen bond with the purine ring of cGMP in the syn conformation, strengthening binding (Flynn et al., 2007) (Figure 2.5B).   78   Figure 2.5 Opposite effects of aspartate substitution of isoleucine 636 in the C-helix of the HCN2 C-linker/CNBD on binding affinity of cAMP and cGMP. (A) Plots of heat produced by progressive injections of cAMP and cGMP to the 200 μM I636D mutant HCN2 C-linker/CNBD. The solid line through the values represents a two independent binding site model, which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for binding affinity are shown in Appendix 11 (cAMP) and 12 (cGMP) and the values for the energetics of binding are found in Figure 2.7 and the Appendices. The I636D mutation lowers and raises the binding affinity of the HCN2 C-linker/CNBD for cAMP and cGMP, respectively. (B) The binding pocket of HCN2 I636D-cGMP co-crystal (2Q0A). The dashed lines represent additional hydrogen bonds of the side chain carboxylate oxygens of aspartate with a nitrogen atom in the guanine ring (N1) and another with the amine group of the guanine ring, which occur in the syn configuration of cGMP. 79  2.3.4 Comparing cyclic nucleotide potency in the full-length channel with binding affinity to the C-linker/CNBD In order to directly compare the effects of the mutations on binding affinity and potency, we plotted the Kd values for each HCN2 C-terminal construct against the K1/2 values for the equivalent full-length channel, which were obtained previously (L. Zhou & Siegelbaum, 2007). The plots also show a line with a slope that represents a linear relationship between binding affinity and potency for the wild type HCN2 channel (Figure 2.6). For cAMP, the slope of this relationship is 0.88 for high affinity binding, where Kd and K1/2 are close in magnitude. The relationship between Kd and K1/2 for most of the alanine mutants lies close to this line, suggesting that the decrease in potency could be explained mainly by a proportional decrease in binding affinity (Figure 2.6A).  The distance of the values from the line represents the approximate contribution of the residue to the coupling of binding to channel opening. This approach can identify a contribution of coupling to potency even when the maximum effect produced by the ligand, as measured by patch clamp electrophysiology, is not greatly altered, which can occur according to some models of ligand-dependent gating (Colquhoun, 1998).  80   Figure 2.6 Most values for binding affinity of cAMP and cGMP to the HCN2 C-linker/CNBD are correlated with their potency in the full-length channel. Plots of cAMP binding affinity (Kd) versus cAMP potency (K1/2) for wild type HCN2 and single point HCN2 mutants (solid circles). The potencies of each construct were obtained previously by Zhou et al (L. Zhou & Siegelbaum, 2007). The straight line represents the relationship between binding affinity and potency in the wild type channel; Kd = 0.88(K1/2) for high affinity binding and Kd=13.9(K1/2) for low affinity binding. The open square represents the values for cGMP binding and potency for the wild type channel. The plot on the right is a magnification of the region within the dotted rectangle in the plot on the left.   81    Plots of binding cGMP affinity (Kd) versus cGMP potency (K1/2) for wild type HCN2 and single point HCN2 mutants. (A) Plots for cGMP with values of Kd that were determined from fits with a two independent binding site model. The straight line represents the relationship between binding affinity and potency in the wild type channel; Kd=0.056(K1/2) for high affinity binding and Kd=0.96(K1/2) for low affinity binding. (B) The same plot for cGMP with Kd values that were determined from fits with a single binding site model (symbols coloured blue and red). The plot on the bottom right is a magnification of the region within the dotted rectangle in the plot on the bottom left, and contains only those values coloured in blue. Lines represent the slopes for high and low affinity binding in the wild type as in the above plot.    82  Among the alanine mutants, leucine 633 (L633) stands out and shows a much larger reduction in potency of cAMP in the full-length HCN2 channels than in binding affinity of cAMP for the C-linker/CNBD piece. The comparatively larger decrease in potency suggests that L633 has a large impact on how binding is coupled to opening. The lower potency also suggests that, unlike the wild type channel, the full-length L633A channel does not maintain high affinity for the cyclic nucleotide during the gating process.  For cGMP binding to the wild type HCN2 channel, the Kd is smaller than K1/2; the slope of the linear relationship is 0.06 for high affinity binding and 0.96 for low affinity binding (Figure 2.6B).  The potency of cGMP on the wild type HCN2 channel is lower than the high binding affinity of this ligand for the C-linker/CNBD and closer to the low binding affinity suggesting that it does not maintain high affinity for the channel during gating. The relationship between Kd and K1/2 for all alanine mutants with two binding events lie close to this line, suggesting that the alterations in potency can be explained mainly by proportional changes in binding affinity. As for cAMP, the approximate contribution of the residue to coupling of cGMP binding to opening is represented by the distance of the values from the line; a shift to the right suggests that the residue contributes positively to coupling whereas a shift to the left suggests that the residue contributes negatively to coupling.  The alanine mutants with a single binding affinity were divided into two groups. In red are T592A and R591A, which produced the largest effects on binding affinity and potency and lie to the left of both lines. In blue are K638A and L633A, which lie to the left of both lines, and R635A, which lies in between both lines. These three mutations also produce large effects on binding affinity and potency, but not as large as those mutants in red. However, as mentioned above, the lack of negative cooperativity probably means that there is only a single binding 83  event, or that the two binding events are similar in affinity and thermodynamics (Appendix 3). Thus, while these mutants appear to have a strong impact on binding affinity and potency, their contribution to the coupling of binding to opening is not clear.  Cyclic AMP binds to the I636D mutant channel less strongly than to the wild type channel but, like L633A, the reduction of potency in the full-length I636D channel is much larger than the associated decrease in binding affinity. In contrast, cGMP binds to the I636D mutant C-linker/CNBD more strongly than to the wild type channel but the increase in potency is larger than the increase in binding affinity. Thus, the I636D mutation reduces both binding affinity and efficient coupling of cAMP binding to opening and does the opposite for cGMP.   2.3.5 Large differences in the pattern of thermodynamics accompany large mismatches between cyclic nucleotide binding affinity and potency As mentioned above, a reproducible and similar thermodynamic pattern accompanies the binding of cAMP and cGMP to the HCN2 C-linker/CNBD (see Figure 2.2B). The pattern of thermodynamics is generally preserved for cAMP and cGMP binding to the mutant HCN2 C-termini when negative cooperativity is preserved. The pattern of thermodynamics of the low affinity cAMP binding for L633A and the I636D mutations, which produce the largest discrepancies between binding affinity and potency (Figure 2.7), is notable. A change in enthalpy is less favourable and the entropy difference is less unfavourable as compared to the binding event of other mutants and wild type. Because the pattern of thermodynamics of the low affinity binding event is indicative of the nature of negative cooperativity observed, the difference in thermodynamics for this event between these two mutants and the other constructs supports a difference in the allosteric effect of cAMP on the HCN2 C-linker/CNBD.    84  For the L633A, R635A and K638A, which only show a single binding event, the thermodynamic parameters are compared to both the high and low affinity events of the other mutants (Figure 2.7). This shows that the thermodynamic profiles for L633A and R635A were most similar to those for the other high affinity binding events whereas the pattern of thermodynamics for K638A was more similar to those for the low affinity binding events. Because the c values are outside of the generally acceptable range (see Experimental Procedures), the values for thermodynamic parameters for cGMP binding to R591A and T592A HCN2 C-linker/CNBD are not plotted, but are found in the caption for Figure 2.7.    85   86  Figure 2.7 The thermodynamics of cAMP and cGMP binding to the wild type and mutant HCN2 C-linker/CNBD.  Bar graphs of the thermodynamics of binding, ΔG, ΔH, and -TΔS, which were derived from the fits in Figures 2.2, 2.3, and 2.5. Values of thermodynamics for mutants with only one binding event for cGMP are presented on both plots for comparison. Values plotted represent the means ± s.e.m. The thermodynamics for the binding of cGMP to the R591A and T592A mutants are not shown on the graph. R591A values for ∆H = -427700 kcal/mol, -T∆S = 426140 kcal/mol, ∆G= -1560 kcal/mol. T592A values for ∆H = -20.71 kcal/mol, -T∆S = 11.98 kcal/mol, ∆G = -5.10 kcal/mol. For other values, refer to Appendix 11 (cAMP) and Appendix 12 (cGMP).  2.4 Discussion In this chapter, we compared binding affinity of cAMP and cGMP to the HCN2 C-linker/CNBD piece with previously published data on their potency in the full-length channel (L. Zhou & Siegelbaum, 2007) and, by making this comparison, we were able to identify residues that are important for coupling ligand binding to opening. For the wild type HCN2 channel, the values for high affinity binding (0.09 µM) and potency (0.1 µM) of cAMP are very close in magnitude, suggesting that binding is very strongly coupled to channel opening. However, the Kd value of cGMP for high affinity binding (0.36 µM) is much lower than its potency (6.4 µM) in this channel, suggesting binding is less strongly coupled to channel opening. Because cGMP also binds to the HCN2 CNBD with lower affinity than cAMP, we suggest that the greater potency of cAMP as compared to cGMP is due to both higher binding affinity and stronger coupling of binding to channel opening. Our data suggest that efficient coupling of binding to channel opening, and thus high potency, depends upon the full-length HCN2 channel maintaining high affinity for the cyclic nucleotide throughout the gating process.    Importantly, as was shown previously with cAMP (Chow et al., 2012), the binding of cGMP to the HCN2 C-linker/CNBD exhibits negative cooperativity and the overall thermodynamic pattern of cGMP binding to the HCN2 C-linker/CNBD was similar to that of 87  cAMP binding. Binding of cyclic nucleotides with negative cooperativity is a novel finding for HCN channels and is supported by studies using a fluorescent cAMP analogue in the full-length HCN2 channel (Kusch et al., 2011). However, for cGMP binding, the favourable contribution of enthalpy was larger for both high and low affinity binding events, while the contribution of entropy was either less favourable (high affinity binding) or more unfavourable (low affinity binding). The greater change in enthalpy could be due to extra hydrogen bonds between cGMP and the HCN2 CNBD, apparent in the co-crystal structure (Zagotta et al., 2003). Cyclic GMP binds to the CNBD in the syn configuration, unlike cAMP which binds in the anti configuration, and forms one extra hydrogen bond that is found between the 2-NH2 of cGMP and side chain hydroxyl group of T592. It is also apparent from the structure that a water molecule is trapped and forms hydrogen bonds between N1 of guanine ring and main chain carbonyl of R632, which is not found in the cAMP HCN2 structure. Less favoured, or more unfavoured, changes in entropy could be due to the extra hydrogen bonds that reduce the mobility of cGMP upon binding as well as motion of T592, which is indicated by a lower relative B-factor for this residue in the cGMP-HCN2 structure than in the cAMP-HCN2 structure (Flynn & Zagotta, 2011). Finally, the differences in the change in entropy of binding between cAMP and cGMP could be due to differences in desolvation upon binding. Previously, computational methods suggested that cGMP is more difficult to desolvate, thereby reducing potency and, presumably, binding affinity (L. Zhou & Siegelbaum, 2008). Thus, less favoured, or more unfavoured, changes in entropy could be due to the greater energy required to desolvate cGMP and the binding cavity upon binding.  We found that single point alanine substitutions of residues near the cyclic nucleotide binding site mainly reduced binding affinity of cAMP and cGMP to the HCN2 CNBD, in 88  parallel with reductions in potency by the equivalent single point mutations in the full-length channel. We observe that the residues in which alanine substitution produced roughly parallel effects on binding affinity and potency mainly control binding affinity and, thus, they are likely to remain static relative to the ligand during gating. Of those residues, as can be appreciated in Figure 2.8 (left), R591 and T592 contribute to binding and potency of cAMP to the greatest extent. Arginine 591, which is found deep within the phosphate binding cassette (PBC) in the beta roll, makes contact with the equatorial oxygen of the cyclic phosphate group of cAMP and cGMP (Akimoto et al., 2014; Zagotta et al., 2003) and is highly conserved not only among the HCN isoforms (Jackson et al., 2007) but also in many proteins containing this family of cyclic nucleotide binding domains (Shabb & Corbin, 1992; Tibbs et al., 1998; Weber & Steitz, 1987). Threonine T592, which is also found in the PBC and is conserved in  many of the proteins in this family (Jackson et al., 2007; Shabb & Corbin, 1992), makes contact with the axial oxygen of the cyclic phosphate group of cAMP and cGMP (Akimoto et al., 2014; Zagotta et al., 2003).  One residue, L633 in the distal C-helix of the CNBD, stood out in our analysis because its replacement by alanine produced a smaller reduction of binding affinity of cAMP than of potency, the largest mismatch found among residues examined by alanine substitution (Figure 2.8). This large mismatch suggests that L633 is critical for efficient coupling of binding to opening. Thus, during HCN2 channel gating, L633 probably undergoes dynamic interactions with cAMP binding, in contrast to R591 and T592 that most likely remain static relative to cAMP. This scenario is consistent with the recent model proposed based upon NMR of the cAMP-sensitive HCN4 C-linker/CNBD (Akimoto et al., 2014), which has since been supported by NMR studies of the HCN2 C-linker/CNBD (Saponaro et al., 2014). This model suggests that cAMP binding orders the phosphate binding cassette and distal C-helix, and shifts the B- and C-89  helices toward the β-roll, with movement of the N3A domain (helices that form the most distal part of the C-linker domain and proximal part of the CNBD) away from the β-roll. The model is consistent with studies on the HCN2 C-linker/CNBD using transition metal FRET and double electron-electron resonance, which also suggest that the C-helix structure is stabilized and  moves toward the β-roll when cAMP binds (Puljung et al., 2014; Puljung & Zagotta, 2013; Taraska et al., 2009).   According to the cAMP-bound HCN2 C-linker/CNBD crystal structure, L633 does not make direct contact with cAMP or cGMP, as the residue is not within the 4Ǻ van der Waals radius of the ligand, but it is next to R632 which, according to the crystal structure and molecular simulations, makes contact with the cyclic nucleotide and forms a salt bridge with glutamate 582 (E582) in the P-helix. Substitution of R632 with alanine almost completely eliminates the cyclic nucleotide-induced depolarizing shift in channel activation, in support of its role in allostery (L. Zhou & Siegelbaum, 2007). It is possible that L633 makes its contribution by supporting strong interactions of R632 with the cyclic nucleotide and with E582 during gating, or by supporting interactions of the ligand with other key regions of the binding domain during gating, and helping to maintain high affinity for the cyclic nucleotide.    90   Figure 2.8 Impact of residues on cAMP and cGMP binding affinity and potency mapped onto the structure of the HCN2 C-linker/CNBD. Crystal structures of HCN2 C-linker/CNBD with cAMP (1Q5O, top) and with cGMP (1Q3E, bottom) were zoomed into where the ligands bind. The left panels show the impact on binding affinity whereas the right panel shows the impact on potency, for both ligands. The impact is based on the effect of single point alanine substitutions of the corresponding residue. The impact is colour-coded according to the legend, where red indicates a strong impact and blue indicates a lesser impact. The I636 and K638 residues were not resolved in the cGMP-HCN2 structure.   Another residue, I636 in the distal C-helix, also stood out in our analysis of both cAMP and cGMP binding. Alanine substitution of I636 had a small inhibitory effect on cAMP binding and potency as shown previously (L. Zhou & Siegelbaum, 2007), but it moderately increased the binding affinity and potency of cGMP. For cAMP, potency was reduced to a greater extent than 91  binding affinity in the I636A mutant, suggesting that I636 predominantly influences coupling rather than binding of cAMP. For cGMP, enhancement of binding affinity and potency were of similar magnitude in the I636A mutant, suggesting that I636 mainly reduces cGMP binding affinity and that this reduction explains the reduction in potency. According to the cAMP-bound HCN2 crystal structure and molecular simulations, I636 makes contact with cAMP, but is not resolved in the cGMP-HCN2 structure (Zagotta et al., 2003; L. Zhou & Siegelbaum, 2007). Nevertheless, we propose that I636 promotes cAMP binding and hinders cGMP binding, and moves relative to the ligand to help maintain high binding affinity for cAMP during gating. Overall, the promotion of cGMP binding and potency by replacing I636 with aspartate (I636D) was larger than that resulting from replacement of this residue by alanine, with Kd and EC50 values approaching those for cAMP in the wild type channel. Thus, the addition of aspartate likely produced effects over and above those resulting from the loss of isoleucine. Like L633A, the mutation I636D had a smaller inhibitory effect on the binding affinity of cAMP than on its potency in the mutant HCN2 channel, producing a large mismatch in these values. Thus, replacement of I636 by aspartate inhibited efficient coupling of cAMP binding to opening. In stark contrast, the binding affinity of the HCN2 C-linker/CNBD for cGMP is increased by I636D but not to the same extent as potency, suggesting that this mutation increases both efficient coupling of cGMP binding to opening and binding affinity. A structure of the I636D mutant HCN2 C-linker/CNBD, solved by X-ray crystallography, suggests that the added aspartate can form an additional hydrogen bond with cGMP (Flynn et al., 2007) (Figure 2.5B), which would explain the greater binding affinity. However, alanine substitution of I636 also improved cGMP binding affinity to the CNBD to almost the same extent as the aspartate substitution but without the same large effect on potency. Thus, over and above the effect of removing isoleucine, the 92  aspartate contributes mainly to coupling of cGMP binding to opening with a minimal contribution to binding affinity.  In summary, our data suggest that high potency depends upon the full-length HCN2 channel maintaining high affinity for the cyclic nucleotide throughout the gating process, which in turn depends upon residues in the C-helix. Movement of the C-helix of the CNBD and other alterations in the structure of this region that have been proposed upon binding of cAMP (Akimoto et al., 2014; DeBerg et al., 2015; Puljung et al., 2014; Saponaro et al., 2014; Taraska et al., 2009) could adjust the binding interactions to maintain high binding affinity as well as to provide an allosteric link to downstream elements that stabilize the open state and facilitate opening. Cyclic AMP is not only better able to bind to the CNBD than cGMP, but it is also better able to maintain high binding affinity during gating. Leucine 633 is a site that likely moves and helps to maintain high affinity binding for cAMP during gating, while I636 acts to favour and disfavour cAMP and cGMP binding strength, respectively, as well as to move relative to the ligand and help to maintain high binding affinity for cAMP during gating. Aspartate substitution of I636 emphasizes that even though cAMP and cGMP bind and promote alterations in C-terminal structure in a similar manner, they do so through incompletely overlapping static and dynamic interactions with the binding site.         93  Chapter 3: Oligomeric Interactions by C-terminal Domains Promote Cyclic Nucleotide Facilitation of Opening of the HCN2 Channel  3.1 Introduction Hyperpolarization-activated Cyclic Nucleotide-gated ‘HCN’ channels are activated by hyperpolarization of the membrane potential and some isoforms open more easily when cAMP binds to the CNBD of the intracellular C-terminus(D. DiFrancesco & Tortora, 1991; Gauss et al., 1998; Ludwig et al., 1998; Santoro et al., 1998). HCN isoforms have significant sequence homology with channels in the voltage-gated potassium channel superfamily(Gauss et al., 1998; Ludwig et al., 1998; Santoro et al., 1998) for which several crystal structures have been solved(Doyle et al., 1998; Lee, Lee, Chen, & MacKinnon, 2005; Long et al., 2005). By extrapolation, HCN subunits are predicted to have six transmembrane helices (S1-S6) with a cytosolic N- and C-terminus, and to combine as tetramers to form the ion-conducting channel. A tetrameric structure predicts that each individual HCN channel has four cAMP binding sites.    A C-terminal region of HCN2 provides tonic inhibition on pore opening which is relieved by binding of cAMP to the CNBD (Wainger et al., 2001). A study of a purified version of this HCN2 C-linker/CNBD using analytical ultracentrifugation shows that its tetramerization in solution is promoted by cAMP, which has led to the proposal that ligand binding promotes intersubunit interactions and a gating ring in the full-length channel that relieves the inhibition and facilitates opening(Zagotta et al., 2003). The correspondence between facilitation of opening and the degree of in vitro tetramer formation is supported by a comparison of HCN1 and HCN2 channels; in the absence of cAMP, the C-linker/CNBD of HCN2 form oligomers less efficiently 94  (Chow et al., 2012; Lolicato et al., 2011) and HCN2 opening is more inhibited than opening of HCN1(Wainger et al., 2001).  Correlation between facilitation of opening and degree of tetramerization is also supported by experiments showing that replacement of a tripeptide sequence in the C-linker domain of HCN2 with that of a cyclic nucleotide gated channel reverses the action of cAMP, causing it to inhibit both channel opening and tetramerization (L. Zhou et al., 2004).   An alternative structural view of cAMP facilitation has also been proposed which suggests that disruption of interaction between subunits, rather than promotion of interaction between them, facilitates the opening of the HCN2 channel (Craven et al., 2008; Craven & Zagotta, 2004). A salt bridge triad was identified in the crystal structure of the HCN2 C-linker/CNBD piece, in the C-linker region that connects the cAMP binding domain to the pore. One salt bridge is formed by residues within one subunit and another between residues in adjacent subunits. Mutations that break the salt bridges favour channel opening, which suggests that these salt bridges normally help to keep the channel closed. Together, the data suggest that the C-linker is in a bound but resting or closed conformation in the crystal structure (Zagotta et al., 2003) and that cAMP binding re-arranges the C-linker, which disrupts the salt bridges and facilitates opening of the HCN2 channel.   Here, we provide evidence in support of the view that facilitation of the opening in the HCN2 channel by cAMP occurs by promoting intersubunit interactions and formation of a gating ring. We examined a series of cAMP analogues and identified partial agonists of HCN2 based on their ability to promote the oligomerization of the isolated HCN2 C-linker/CNBD domains in solution. Most of the ligands tested, including both purine and pyrimidine cyclic nucleotides, bind to the CNBD, promote oligomerization of this C-linker/CNBD domain in solution and 95  produce a depolarizing shift of the HCN2 activation curve which is as large as that produced by cAMP. In contrast, cCMP and cIMP produce a smaller depolarizing shift of the HCN2 activation curve and do not promote oligomerization of the C-linker/CNBD in solution. Our data support a model in which the degree of facilitation of opening in the full-length HCN2 channel by a given ligand depends upon its ability to promote oligomerization of the C-linker/CNBD in solution.   3.2 Experimental procedures  3.2.1 HCN protein purification and mutagenesis The HCN2 C-linker/CNBD (i. 443-645, Appendix 1B) was sub-cloned in a modified pET28 vector (Novagen), containing a His6-tag, maltose binding protein, and a cleavage site for the tobacco etch virus (TEV) protease preceding the construct of interest (Appendix 1A). The mutation R635A was made by standard Quikchange protocol (Stratagene). Refer to Appendix 2A for the primers. The protein was expressed in Escherichia coli Rosetta (DE3) pLacI cells grown in 2xYT media at 37ºC for 4 hours, after induction with 0.4 M IPTG at optical density of 0.6. The cells were harvested by centrifugation, and resolubilized in lysis buffer. Cell debris was removed by centrifugation for 30 minutes at 35000x g. Purification protocol is similar to (Chow et al., 2012). The tagged proteins were separated by cobalt affinity columns (Talon, Clontech) and eluted with 500 mM imidazole. The cleavage of the HMT tag was performed by incubating the purified protein with TEV protease overnight at 4oC. The tag was removed by another round of Talon resin, and further purified on a ResourceS cation exchanger (GE Healthcare). Purified HCN2 C-linker/CNBD was dialyzed in “ITC buffer” (150 mM KCl, 20 mM Hepes pH 7.4, 10 mM βME). Concentration was determined by spectrophotometry, using the Edelhoch method. Protein purity 96  was confirmed by SDS-PAGE. Refer to section 2.2.2 for a more elaborated protocol on protein purification.  3.2.2 Ligand preparation Cyclic AMP and cGMP were purchased from Sigma. The other analogues were from Biolog Life Science Institute (Bremen, Germany). Ligands were dissolved in water to make a stock of 10 mM. They were further diluted to 2 mM with “ITC buffer”. The exact concentrations of these analogues were determined by spectrophotometry with their respective extinction coefficients. A negative background test was done on the ITC for each ligand to make sure heat is not generated by water-buffer dilution or by ligand-solvent interaction. The background produced minimal noise for each of the ligand (data not shown).  3.2.3 Ligand-induced tetramerization by dynamic light scattering (DLS) The proteins and ligands were mixed and diluted to the correct ratio. Samples were centrifuged for 1 minute to remove debris and 50 μL were loaded into each well in the 384-well microtiter plate. DLS experiments were performed with a DynaPro Plate reader (Wyatt Technology) to measure the hydrodynamic radii. Ten acquisitions (5 seconds per scan) were performed at 22oC, and the radii were averaged and regularized, with Dynamics 7.0, to produce the apparent molecular weights. Only readings with polydispersity values lower than 30% were used. More detailed steps are outlined in section 2.2.5.    97  3.2.4 Direct measurements of cyclic nucleotide binding by isothermal titration calorimetry (ITC) Titrations were performed in ITC200 (MicroCal, Malvern) at 25oC in an adiabatic environment. 200 µM (diluted in ITC buffer) of HCN2 Clinker/CNBD was added to the sample cell. 2mM of the ligand was loaded on the syringe and increments of 1 μL were injected into the sample cell. Heat difference was recorded, between the reference cell with ddH2O and the sample cell with ligand-protein mixture. Calorimetric data were analyzed and binding isotherms were generated with the software Origin 7.0 for MicroCal. Refer to section 2.2.4 for equations and a more thorough protocol.  3.2.5 Whole-cell patch clamp electrophysiology  Cells expressing GFP were chosen for whole-cell patch clamp recordings 24-48 hours post transfection. The pipette solution contained (in mM): 130 K-Asp, 10 NaCl, 0.5 MgCl2, 1 EGTA, and 5 HEPES with pH adjusted to 7.4 using KOH. For experiments at saturating levels of cAMP, 2 mM cAMP (Na salt) was added to the pipette solution. Extracellular recording solution contained (in mM): KCl 5.4, NaCl 135, MgCl2 0.5, CaCl2 1.8, HEPES 5, pH 7.4. Whole-cell currents were recorded using an Axopatch 200B amplifier and Clampex software (Axon Instruments) at room temperature. Patch clamp pipettes were pulled from borosilicate glass and fire polished before use (pipette R= 2.5-4.5 MΩ).  Data were filtered at 2 kHz and were analyzed using Clampfit (Axon Instruments) and Origin 8.0 (Microcal) software. The Ih activation curves were determined from time-dependent tail currents at a 2 s pulse to -35 mV following 3 s test pulses ranging from -150 mV to -10 mV, in 20 mV steps. Single tail current test pulses were followed by a 500 ms pulse to +5 mV to 98  ensure complete channel deactivation. The resting current was allowed to return to its baseline value before subsequent voltage pulses.  Normalized tail current amplitudes were plotted as a function of test voltage and values from each cell were fit with a single order Boltzmann function (equation 1),   𝑓(𝑉) = 𝐼𝑚𝑎𝑥�1 +  𝑒�𝑉1/2−𝑉�/ 𝑘 ��   to determine the midpoint of activation (V½) and slope factor (k). The effective charge (Z) was calculated using the equation 𝑍 = 𝑅𝑇 𝑘𝐹�  , where T = 295K, and R is the gas constant and F is the Faraday’s constant.  3.2.6 Crystallization of HCN2-ligand complexes Protein complexes, 200 μM (~10 mg/ml) of HCN2 C-linker/CNBD with 5 mM of respective analogue, were diluted with ITC buffer. Crystals were obtained using the hanging drop vapour diffusion method, after one to two weeks at 4oC. The well solution contained 10-26% PEG400, 0.1 M sodium citrate pH 4.6-5.5, 0.2 M NaCl with reagents obtained from FLUKA and Sigma. The crystals selected were from these well conditions:  cCMP + HCN2 C-linker/CNBD 200 mM NaCl, 0.1 mM sodium citrate pH 5.0, 16% PEG  cUMP + HCN2 CLCNBD  200 mM NaCl, 0.1 mM sodium citrate pH 4.6, 13% PEG  cIMP + HCN2 CLCNBD  200 mM NaCl, 0.1 mM sodium citrate pH 5.5, 14.5% PEG  cPuMP + HCN2 CLCNBD  200 mM NaCl, 0.1 mM sodium citrate pH 5.5, 18% PEG  2-NH2-cPuMP + HCN2 CLCNBD 200 mM NaCl, 0.1 mM sodium citrate pH 4.6, 12% PEG  99  The crystals were cryoprotected in identical conditions from the crystallization mix, with the exception of the PEG400 concentration, which was at 26%. After flash cooling in liquid nitrogen, data sets were collected at either beamline 23-ID-B or 23-ID-D (GM/CA-XSD) of Advanced Photon Source, and processed using XDS (Kabsch, 2010) or HKL2000 (Otwinowski & Minor, 1997). The structures were solved via molecular replacement using Phaser with PDB entry 1Q5O as search model.  Structure files for the ligands were generated with JLigand in the CCP4 suite (Lebedev et al., 2012; Winn et al., 2011). The structures were refined through alternating cycles of manual building with COOT (Emsley, Lohkamp, Scott, & Cowtan, 2010) and automated refinement in PHENIX (Adams et al., 2010). All structure images were prepared using PYMOL (Version 1.3 Schrödinger, LLC). The final structures will be available in the PDB upon publication. Intersubunit interactions were analyzed with PISA (Krissinel & Henrick, 2007) and HBplus (McDonald & Thornton, 1994), and ligand interactions were determined by Contact in the CCP4 suite.  3.3 Results 3.3.1 Cyclic CMP and cIMP do not promote self-association of the HCN2 C-linker/CNBD when measured by dynamic light scattering In solution, cAMP promotes oligomerization of isolated HCN2 C-linker/CNBD domains, suggesting that intersubunit interactions are part of the alterations in structure induced by cAMP which lead to facilitation of opening in the full-length HCN2 channel (Chow et al., 2012; Zagotta et al., 2003). Because the propensity of the HCN2 C-terminus to oligomerize in solution appears to correlate with facilitation of opening, we used an oligomerization assay to identify cAMP analogues that act as antagonists or partial agonists; we hypothesized that partial agonists or 100  antagonists would bind to the HCN2 C-linker/CNBD but not promote oligomerization to the same extent as cAMP.   The C-linker/CNBD of HCN2 that was used for this study is comprised of six α helices in the C-linker and the four α helices and eight β-strands that make up the CNBD. We initially compared the extent that cCMP and cUMP, both cyclic pyrimidine nucleotides, promote self-association of the HCN2 C-linker/CNBD with that produced by cAMP, using Dynamic Light Scattering (DLS) (Figure 3.1). Cyclic CMP is known to facilitate HCN2 channel opening (Zong et al., 2012) to a lesser extent than cAMP and, therefore, we expected that cCMP would also promote oligomerization of the HCN2 C-linker/CNBD to a lesser extent than cAMP. As it was found previously, the apparent molecular weight of the HCN2 C-terminus is increased with increasing amount of protein even without the involvement of ligand, showing that tetramers are formed at high protein concentrations (Chow et al., 2012). By adding cAMP, the average weight is further increased, corresponding to an increased proportion of larger oligomeric forms, which is consistent with previous studies using DLS and ultracentrifugation (Chow et al., 2012; Zagotta et al., 2003; L. Zhou et al., 2004). Cyclic UMP produced an effect on molecular weight that was similar to that produced by cAMP, although larger amounts of the former were required (Figure 3.1). In contrast, cCMP had no effect on the apparent molecular weight of HCN2 C-linker/CNBD, even when present at very high concentrations (Appendix 5). The lack of effect of cCMP on oligomerization of the HCN2 C-linker/CNBD in solution suggests that its ability to promote a gating ring in the full-length channel is likewise limited, consistent with its smaller effect on HCN2 channel opening (Zong et al., 2012).   By DLS, we also examined cGMP, which produces the same maximum effect on the full-length HCN2 channel opening as does cAMP but with lower potency (Zagotta et al., 2003; L. 101  Zhou & Siegelbaum, 2007), and three cyclic nucleotide analogs that are intermediate in structure between cAMP and cGMP; cPuMP (no exocyclic constituent), 2-NH2-cPUMP (2-NH2) and cIMP (6-keto). Of these three analogues, only cIMP was unable to promote oligomerization, even when present in concentrations of up to 10 mM, which is 200 times higher than the Kd (Figure 3.1, Appendix 5).   Figure 3.1 Oligomerization of the HCN2 C-linker/CNBD is promoted by cAMP, cGMP and a sub-set of structurally related analogues. Plots of estimated average molecular weight versus concentration of purified HCN2 C-linker/CNBD protein are shown. The estimated molecular weight was determined by Dynamic Light Scattering and is proportional to the population of oligomers in solution. An increased in oligomerization occurs with higher protein concentration and further increased in the presence of cAMP, cGMP (top), cUMP (right), cPuMP and 2-NH2-cPuMP (left). Cyclic CMP (right) and cIMP (left), have no effect on the apparent molecular weight. Values represent means ± s.e.m, determined from three separate preparations of purified protein measured on different days.   102  3.3.2 Cyclic CMP and cIMP bind to the HCN2 C-linker/CNBD without negative cooperativity  We next compared the binding of cyclic pyrimidine nucleotides and cyclic purine nucleotides to the purified HCN2 C-linker/CNBD using isothermal titration calorimetry (ITC). The protein was tested at 200 µM, a concentration at which the C-linker/CNBD is predominately in a tetrameric form in solution as shown by DLS. We found that each cyclic nucleotide produced a clear and distinctive pattern of heat release (Figure 3.2-3.4B). Cyclic AMP, as we showed previously (Chow et al., 2012), produced two clear transitions, which were best fit with a two independent binding site model and yielded two Kd values of ~0.13 µM and ~1.8 µM (Figure 3.2B). These two affinities have previously been reported to indicate negative cooperativity (Chow et al., 2012). We found that cGMP binds to the HCN2 C-linker/CNBD also with negative cooperativity, like cAMP, but with overall lower affinity (Kd values of ~0.4 and ~8.5 µM) (Figure 3.2B).  103    104  Figure 3.2 The cyclic purine nucleotides cAMP and cGMP bind to the HCN2 C-linker/CNBD with negative cooperativity. (A) Chemical structure of the two ligands. The atoms in the purine ring are labelled as a reference. (B) Plots of heat and binding isotherms produced by progressive injections of cAMP (left) and cGMP (right) to 200 μM HCN2 C-linker/CNBD as indicated, measured by ITC. The solid line through the values represents a two independent binding site model, which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for binding affinity are shown in Table 3.1. (C) Bar graph showing the affinity and thermodynamics for the two binding events, a high affinity and a low affinity binding event which were determined from the fit in B. Values in the graph represent means ± s.e.m., and can be found in Appendix 15.  Cyclic UMP produced a pattern similar to cAMP, but, again, with lower values for affinity; high affinity binding Kd of ~1.3 µM and a low affinity binding Kd of ~23.7 µM (Figure 3.3B). In contrast to cAMP and cUMP, the heat released by cCMP could be fit best with a single binding site model, yielding an affinity of approximately 20 µM. We also found that cPuMP (~0.2 μM and ~2.8 μM) and 2-NH2-cPuMP (~0.4 μM and ~3 μM) bound with negative cooperativity and in similar range as the binding of cAMP (Figure 3.4B). In contrast, the heat released by cIMP could be fit best with a single binding site model, yielding an affinity of ~49 μM (Figure 3.4B).   The binding of cCMP and cIMP was driven by both favourable enthalpy and entropy, similar to the high affinity binding event for cAMP and the other ligands (Figure 3.2-3.4C). However, the affinity is lower, which seems to be result of a less favourable entropic contribution for both cCMP and cIMP. Binding of the second cAMP or cUMP molecule occurs with lower affinity but more favourable enthalpy than binding of the first molecule, indicating that the entropic contribution is an important determinant of affinity and may underlie negative cooperativity. Together, the data suggest that cCMP and cIMP bind to the high affinity binding site of the tetrameric C-linker/CNBD but that, upon binding, they are poor at inducing 105  conformational changes and negative cooperativity.  Interestingly, these are the same two molecules that failed to induce oligomerization (Figure 3.1), suggesting a direct correlation between negative cooperativity and the ability to induce stronger tetramers.  106     107  Figure 3.3 The cyclic pyrimidine nucleotide cUMP, but not cCMP, binds to the HCN2 C-linker/CNBD with negative cooperativity. (A) Chemical structure of the two ligands. The atoms in the pyrimidine ring are labelled as a reference. (B) Plots of heat and binding isotherms produced upon progressive injections of cCMP and cUMP to 200 μM HCN2 C-linker/CNBD measured by ITC. The solid line through the values represents either a single binding site model (cCMP) or a two independent binding site model, which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for binding affinity are shown in Table 3.1. Only cUMP binds with negative cooperativity. (C) Bar graph showing the affinity and thermodynamics of binding, which were determined from the fit in B. Values in the graph represent means ± s.e.m., and can be found in Appendix 15.   108    109  Figure 3.4 The cyclic purine nucleotides cPuMP and 2-NH2-cPuMP but not cIMP bind to the HCN2 C-linker/CNBD with negative cooperativity.  (A) Chemical structure of the three ligands. The atoms in the purine ring are labelled as a reference. (B) Plots of heat and binding isotherms produced upon progressive injections of cPuMP, 2-NH2-cPuMP and cIMP to 200 μM HCN2 C-linker/CNBD measured by ITC. The solid line through the values represents either a single binding site model (cCMP) or a two independent binding site model, which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for binding affinity are shown in Table 3.1. Only cIMP binds without negative cooperativity.  (C) Bar graph showing the affinity and thermodynamics of binding, which were determined from the fit in B. Values represent means ± s.e.m., and can be found in Appendix 15.  3.3.3 Cyclic CMP and cIMP are less effective facilitators of HCN2 opening than the other cyclic nucleotides examined The results from the oligomerization and binding experiments suggest that cCMP and cIMP have weaker effects than cAMP on the full-length HCN2 channel. A weaker facilitation of HCN2 opening has been noted previously for cCMP (Zong et al., 2012). We compared the effects of all analogs examined biochemically above on the HCN2 channel expressed in Chinese hamster ovary (CHO) cells using the whole-cell patch clamp approach with 1 or 2 mM of cyclic nucleotides in the pipette solution or no cyclic nucleotide at all. The concentrations chosen are well above the maximum concentration that is required to produce a maximum effect on HCN2 channels expressed in HEK cells (Zong et al., 2012) or on sinoatrial HCN (D. DiFrancesco & Tortora, 1991), and well above the binding affinities determined by isothermal titration calorimetry (see section 3.3.2).   The Ih activation curves were determined by plotting normalized tail current amplitudes versus test voltage (Figure 3.5), which were then fit with Equation 1 to obtain values for half-activation voltage (V½) and slope factor (k) (Table 3.1). Compared to the control cells, those exposed to saturating concentrations of cCMP and cIMP produced the smallest shifts in V½ of 110  activation and the smallest increase in conductance at three different voltages (Table 3.2), i.e. they are weaker facilitators of opening than the other cyclic nucleotides examined. These data support a direct relationship between agonist-induced structural changes and oligomerization of the HCN2 C-linker/CNBD domains and facilitation of opening in the full-length HCN2 channel.    111   Binding affinity Electrophysiology DLS  N High Kd  (μM) Low Kd (μM) N V1/2 (mV) Slope (z) P value  Control (no ligand) -- --- --- 7 -97.15 ± 3.55 11.11 ± 0.37 --- --- + cAMP 4 0.13 ± 0.03  1.83 ± 0.21  7 -87.14 ± 3.77 11.46 ± 0.89 0.039 *  + cGMP 4 0.43 ± 0.07  8.53  ± 0.70  6 -87.45 ± 2.73 12.61 ± 1.00 0.029 *  cAMP-cGMP intermediates  + cPuMP 3 0.17 ± 0.03  2.76 ± 0.19  7 -88.57 ± 4.55 12.61 ± 1.01 0.081  + 2NH2-cPuMP 3 0.41 ± 0.04  3.08 ±   0.29  6 -87.96 ± 2.81 14.09 ± 1.05 0.037 *  + cIMP 4 49.32 ±  6.45   7 -92.73 ± 2.69 10.72 ± 0.69 0.170  Cyclic pyrimidine nucleotides a Control (no ligand) -- --- --- 6 -114.89  ± 3.76 13.52 ± 2.54 --- --- + cCMP  3 20.23 ± 0.66   6 -111.96 ± 4.77 20.06 ± 3.08 0.320   + cUMP  3 1.33 ± 0.17  23.70 ± 1.82  6 -104.72 ± 2.30 14.78 ± 1.26 0.022 *  C-helix mutation R635A Control R635A (no ligand) -- --- --- 6 -101.21 ± 2.59 10.88 ± 0.34 --- --- + cAMP 3 0.19 ± 0.01 1.89 ± 0.05 6 -87.75 ± 1.97 11.46 ± 0.49 0.001 *  + cUMP 3 183.8 ± 85.7 a 6 -88.82 ± 2.24 11.44 ± 1.17 0.002 *   Table 3.1 Values for binding affinity and biophysical parameters describing voltage-dependent activation with or without ligand. The table summarizes values for binding affinity, half-activation voltage (V1/2) and slope factor determined from fits of the data obtained from individual cells by equation 1. For binding affinity, values for mean ± s.e.m are shown as determined from separate experiments (N = number of separate experiments). Kd values determined for cCMP and cIMP are italicized because only one binding site was observed. For the biophysical parameters, the values are means determined from individual cells (N = number of cells) ± s.e.m. The p values were calculated using one-tailed unpaired student t-test comparing data obtained with cyclic nucleotide in the pipette solution with respective control data obtained from a separate set of cells. p < 0.05 are highlighted with an asterisk.  A one-tailed test was used because all of the treatments were expected to produce a depolarizing shift in the activation curve. The DLS column 112  shows whether the ligand is able to promote oligomerization (or an increased in apparent molecular weight of HCN2 C-linker/CNBD) upon adding ligand. The data shows a correlation where a ligand that does not promote oligomerization in the soluble C-linker/CNBD also causes a smaller shift in the activation curve in HCN2 full-length channel. aA different patch-clamp protocol was used for cyclic pyrimidine nucleotides.   Wild type (HCN2) data with modified protocol (lower KCl internal solution and higher ligand concentration)    Mutant R635A-HCN2 data with modified protocol   n V1/2 Conductance at  -75mV Conductance at  -90mV Conductance at  -105mV Control 6 -101.21 ± 2.59 0.0615 ± 0.0227  0.2703 ± 0.0543 0.5964 ± 0.0419 +2mM cAMP 6 -87.75 ± 1.97 (0.0010)* 0.2233 ± 0.0331 (0.0012)* 0.5445 ± 0.0379 (0.0010)* 0.8050 ± 0.0277 (0.0010)* +2mM cUMP 6 -88.82 ± 2.244 (0.0024)* 0.1954 ± 0.0413 (0.0087)* 0.5145 ± 0.0484 (0.0036)* 0.7844 ± 0.0228 (0.0014)*       n V1/2 Conductance at  -75mV Conductance at  -90mV Conductance at  -105mV Control 7 -97.15 ± 3.55  0.0752 ± 0.0229 0.3513 ± 0.0660 0.6424 ± 0.0653 +2mM cAMP 7 -87.14 ± 3.77 (0.0386)* 0.2251 ± 0.0730 (0.0368)* 0.5371 ± 0.0574 (0.0275)* 0.7902 ± 0.287 (0.0302)* +2mM cGMP 6 -87.45 ± 2.73 (0.0294)* 0.2230 ± 0.0478 (0.0069)* 0.5289 ± 0.0416 (0.0257)* 0.7769 ± 0.0188 (0.0462)* +2mM 2-NH2-cPuMP 6 -87.96 ± 2.82 (0.0368)* 0.2389 ± 0.0464 (0.0035)* 0.5295 ± 0.0468 (0.0282)* 0.7550 ± 0.0757 (0.0815) + 2mM cIMP 7 -92.73 ± 2.69 (0.1701) 0.1463 ± 0.0350 (0.0575) 0.4338 ± 0.0582 (0.1835) 0.7451 ± 0.0426 (0.1061) +2mM cPuMP 7 -88.57 ± 4.55 (0.0814) 0.2574 ± 0.0582 (0.0065)* 0.5389 ± 0.0651 (0.0329)* 0.7633 ± 0.05118 (0.0852) 113  Wild type (HCN2) data before modification (higher [KCl] in internal solution)   n V1/2 Conductance at  -90mV Conductance at  -105mV Conductance at  -120mV Control 6 -114.89 ± 3.76  0.1084 ± 0.0224 0.3559 ± 0.0473 0.6528 ± 0.0365 +1mM cCMP 6 -111.96 ± 4.77 (0.3199) 0.2355 ± 0.0478 (0.0184) * 0.4429 ± 0.0437 (0.1033) 0.6556 ± 0.0339 (0.4780) +1mM cUMP 6 -104.72 ± 2.30 (0.0219) * 0.2611 ± 0.398 (0.0037) * 0.5068 ± 0.0459 (0.0226) * 0.7579 ± 0.0272 (0.0218) *  Table 3.2 Values for half activation voltage and fraction of channel activation at three voltages. The table summarizes the values for half-activation voltages and compares the fraction of activation at three voltages for each condition. This method compares the change in effect of ligand independent of the slope factor. The p values were calculated using one-tailed unpaired student t-test comparing data obtained with cyclic nucleotide in the pipette solution with respective control data obtained from a separate set of cells. p < 0.05 are highlighted with an asterisk.    114   Figure 3.5 Cyclic AMP and analogues variably shift the HCN2 Ih activation curve to less negative potentials. Plots of degree of current activation, determined by normalizing tail current amplitudes to their maximum value (I/Imax), versus test voltages. The solid lines through the values represent fitting a Boltzmann curve by equation 1 (methods), which yielded values for V1/2 and k (Table 3.1 and 3.2). The shift in V1/2 is smaller for cCMP and cIMP, classifying them as partial agonists. 115  3.3.4 Cyclic UMP makes contacts with the HCN2 C-helix that differ from those made by cCMP To directly determine how cCMP and cUMP bind to the C-linker/CNBD of the HCN2 channel, we used X-ray crystallography and compared the solved structures (Table 3.3). The refined structure of HCN2 C-terminus with cCMP and cUMP show that each forms a fully-liganded tetramer with the cyclic pyrimidine nucleotide bound in the anti configuration as does cAMP (Figure 3.6; Appendix 6). The different complexes have near-identical folds, but the specific interactions in the nucleotide-binding pocket differ substantially, and readily explain the ability of cUMP to bind with higher affinity than cCMP. This is particularly the case for residues E582, C584 and A593 found in a region containing the P-helix and between beta strands 6 and 7, as well as at residues R632, R635 and I636 found in the C-helix near the end of the CNBD (Figure 3.6; Table 3.4). In both cCMP and cUMP, the side chain of residue E582 forms two hydrogen bonds with the OH group of the ribose and forms a salt bridge with the side chain of R632. A non-polar interaction of R632 with the pyrimidine ring is also found in both cCMP and cUMP structures. However, the nitrogen of the E582 main chain interacts more closely (within 4 Ǻ) with the equatorial oxygen of the cyclic phosphate and with the 2’-OH of the ribose of cCMP. The nitrogen of the main chain of C584 also forms a hydrogen bond with the axial oxygen of the cyclic phosphate group of cCMP but not cUMP. The side chain of R632 forms two hydrogen bonds with the 2-keto group of cCMP but only one with the 2-keto group of cUMP. The carbonyl group of the R632 main chain forms a hydrogen bind with the 4-NH2 group of cCMP whereas this group is also close to the 4-keto group of cUMP and may experience some repulsive electrostatic interaction between the negative partial charges. The 4-keto group of cUMP is also close to the side chain of R635, 116  which is positioned near enough to form hydrogen bond in the cUMP bound structure but not in the cCMP bound structure. The pyrimidine ring of cUMP, but not cCMP, makes a non-polar interaction with I636, found adjacent to R635 in the C-helix and A593, located just before the seventh beta sheet. In summary, the distal C-helix and the region between beta strands 6 and 7 which is distal to the P-helix are more closely associated with cUMP than with cCMP.   To assess the contribution of R635 to the binding affinity and maximum effect of cUMP, we mutated this residue to an alanine (R635A) in the C-linker/CNBD as well as the full channel to abolish the newly observed bond. The alanine substitution had little effect on the binding affinity of cAMP but produced a more drastic effect on the binding of cUMP, eliminating negative cooperativity and reducing binding affinity. Nevertheless, alanine substitution of arginine 635 did not alter the ability of cUMP, as well as cAMP, to promote oligomerization of the HCN2 C-linker/CNBD or their maximum effect on opening (Figure 3.7). Because cUMP retains its full ability to promote oligomerization of the R635A HCN2 C-linker/CNBD and to facilitate opening of the full-length R635A channel, it is possible that negative cooperativity is still present, but simply unresolved in the ITC because of insufficient differences in affinity.         117   mHCN2 C-linker/CNBD + cCMP mHCN2 C-linker/CNBD + cUMP Data collection   Space group P 4 21 2 P 2 21 21 Cell dimensions        a, b, c (Å) 96.204, 96.204, 46.68 46.486, 89.925, 98.421      α, β, γ (°) 90, 90, 90 90, 90, 90 Resolution (Å) 30.42 – 2.24 (2.32 – 2.24) 28.68 – 2.01 (2.08 – 2.01) Rmeas (%) 7.1 (47.6) 8.8 (49.7) I/ σI 12.74 (2.54) 7.35 (2.00) Completeness (%) 96.19 (87.56) 95.1 (94.3) Redundancy 5.34 (5.24) 1.54 (1.52)    Refinement   Resolution  30.42 - 2.24 28.68 – 2.01  No. reflections 10939 (1657) 50521 (8055) Rwork 0.2213 0.1978 Rfree 0.2553 0.2440 No. atoms 1651 3533      Protein 1611 3214      Ligand 20 40      Water 20 279 B-factors (Average) 73.30 26.10      Protein 73.40 25.50      Ligand 76.80 20.30      Water 65.10 33.60 r.m.s.d.        Bond lengths (Å) 0.012 0.011      Bond angles (o) 1.14 1.14    Ramachandran outliers (%) 0.0 0.0 Ramachandran favored (%) 96.0 98.2  Table 3.3 Data collection and refinement statistics for mHCN2 C-linker/CNBD co-crystallized with five different ligands presented in the chapter. Data, each set from a single crystal of HCN2 C-linker/CNBD liganded with a cyclic pyrimidine nucleotide, was integrated with HKL2000 and XDS, and refinement was completed with Refmac5 of CCP4 suite and phenix.refine of PHENIX. Values in parentheses are for the highest resolution shell. Data quality is indicated by Rmeas which looks at the unique reflection 118  independent of data redundancy, calculated by   𝑅𝑚𝑒𝑎𝑠 =  ∑ � 𝑛𝑛 − 1ℎ𝑘𝑙 ∑ |𝐼ℎ𝑘𝑙,𝑗 − 〈𝐼ℎ𝑘𝑙〉|𝑛𝑗=1∑ ∑ 𝐼ℎ𝑘𝑙,𝑗𝑗ℎ𝑘𝑙    where Ihkl is the mean intensity of the j symmetry-related observations of reflections hkl. The model quality is indicated by Rwork or Rfree calculated by   𝑅 = ∑ |𝐹ℎ𝑘𝑙𝑜𝑏𝑠 − 𝐹ℎ𝑘𝑙𝑐𝑎𝑙𝑐|ℎ𝑘𝑙∑ 𝐹ℎ𝑘𝑙𝑜𝑏𝑠ℎ𝑘𝑙   where Fcalc represents the calculated protein structure factor based on the atomic model. Rfree is calculated only using 5% of isolated reflections.   119   mHCN2 C-linker/CNBD + cPuMP mHCN2 C-linker/CNBD + 2NH2-cPuMP mHCN2 C-linker/CNBD + cIMP Data collection    Space group P 4 21 2 P 4 21 2 I 4 2 2 Cell dimensions         a, b, c (Å) 96.278, 96.278, 46.763 96.429, 96.429 46.423 96.037, 96.037, 115.913      α, β, γ (°) 90, 90, 90 90, 90, 90 90, 90, 90 Resolution (Å) 31.79-2.10 (2.18-2.10) 38.40 – 2.09 (2.14 – 2.07) 50 – 1.77 (1.88 – 1.77) Rmeas (%) 6.3 (89.3) 7.0 (79.7) 7.4 (120.9) I/ σI 18.60 (2.68) 26.99 (4.08) 19.06 (2.10) Completeness (%) 99.5 (99.0) 94.4 (99.1) 97.5 (95.2) Redundancy 7.04 (7.09) 13.86 (13.18) 9.83 (9.92)     Refinement    Resolution  31.79-2.10 38.40 – 2.07 44.12 – 1.771 No. reflections 13266 (2090) 13521 (2028) 26044 (4035) Rwork 0.2006 0.2161 0.2144 Rfree 0.2489 0.2614 0.2628 No. atoms  1683 1721 1804      Protein 1623 1647 1616      Ligand 21 22 22      Water 39 52 166 B-factors (Average) 59.58 49.40 39.37      Protein 59.78 49.40 37.21      Ligand 54.83 50.50 33.65      Water 53.68 47.00 44.84 r.m.s.d.         Bond lengths (Å) 0.0113 0.016 0.011      Bond angles (o) 1.437 1.77 1.50     Ramachandran outliers (%) 0.0 0.98 0.5 Ramachandran favored (%) 96.5 96 98.0  Data set was obtained from a single crystal of HCN2 C-linker/CNBD liganded with a cyclic purine nucleotide.   120   121     122  Figure 3.6 Crystal structures of five analogues bound to the HCN2 C-linker/CNBD reveal their ligand-binding configuration and important interactions. The binding behaviour of the five analogues to HCN2 C-linker/CNBD is visualized using X-ray crystallography with PYMOL. The PYMOL structures consist of the protein represented with ribbon backbone and the ligand represented as sticks; residues that interact with the ligand are also represented as sticks. The blue electron density is generated by the 2mFo-DFc map of the structure, sigma cut-off at 1.0, showing how each ligand fits. Refer to Appendix 6 for Fo-Fc omit map. The interactions inside the five binding pocket are also presented in a simplified 2-D schematics, and the pocket of cAMP (1Q5O) and cGMP (1Q3E) are added as a reference. Dashed lines are hydrogen bonds determined by PISA; double-headed arrows are salt bridges, and curve lines are hydrophobic interactions. The numbers categorize where the bonds are found and is corresponded to Table 3.4 to get intermolecular distances.    123   cAMP cGMP cCMP cUMP cIMP cPuMP 2-NH2-cPuMP ②Axial oxygen of cyclic phosphate T592 (N) 2.81 2.73 2.89 2.92 2.88 2.96 2.79 T592 (SC) 3.03 2.75 2.82 2.77 2.92 2.59 2.34 C584 (N) 3.7 (5.0) 3.76 (4.5) (4.1) (4.1) (4.1) ③Equatorial oxygen of cyclic phosphate R591 (SC) 3 2.89 2.65 3.11 2.83 3.02 3.01 G581 (O) 3.14 3.34 3.22 3.21 3.37 3.23 3.27 E582 (N) 3.83 (4.3) 3.97 (4.1) (4.2) 3.76 (4.0) I583 (N) 3.74 3.78 3.85 3.51 3.73 3.54 3.72 C584 (N) 3.11 3.06 3.31 2.90 2.98 2.99 2.96 ④2’OH in ribose G581 (N) 2.69 2.77 3.08 2.94 2.94 2.86 2.90 E582 (N) 3.94 3.77 3.85 (4.0) (4.1) 3.91 3.96 E582 (SC) 2.67 3.98 2.60 3.02 2.36 3.03 2.81 3.18 2.83 3.14 2.47 3.78 2.56 3.57 R632 (N) 3.52 3.37 3.27 3.09 3.23 3.38 3.52 Group at 6-purine (cAMP, cGMP, and cIMP) or 4-pyrimidine (cCMP and cUMP) R632 (O) ⑤ 3.0 (3.77)* ⑤ 2.68 (3.76)* (3.95)* no group no group R635 (SC) (5.0) truncated (6.1) ⑦ 2.78 ⑦ 3.54 no group no group 2-NH2 (cGMP and 2-NH2-cPuMP) or 2-keto (cCMP and cUMP) group T592 (O) no group ⑥3.98 (7.2) (6.9) no group no group (8.9) R632 (SC) no group (8.1) (9.4) ⑧ 3.89        3.90 ⑧ 3.88 no group no group ⑨3.85 Non-polar interactions with the nitrogenous base, within 4Å  V564 M572 R632 I636 V564 M572 R632  V564  R632 V564 A593 R632 I636 V564 M572 R632 I636 V564  R632 V564 M572 I583 R632 I636     124  Table 3.4 The polar and non-polar interactions between the HCN2 binding pocket and ligands. The numbers are hydrogen bond distances, in Angstroms, between the different ligands in the first row and the different CNBD residues in the first column, obtained from the different crystal structures using PYMOL or Contacts from CCP4 suite. The abbreviation beside the residue number denotes the atom in the protein making the interactions: (N) for main chain nitrogen; (O) for main chain carbonyl, and (SC) for side chain atom(s). The distances in parentheses show that the hydrogen bond donor and acceptor are outside of the van der Waals radii, denying bond formation. The asterisks are keto-carbonyl repulsion between the ligand and protein, which show true atomic distance but the bond does not exist. The numbers refer to specific bonds that correspond to Figure 3.6.  3.3.5 Crystal structures show that cAMP-cGMP intermediates bind in the anti configuration and identify unique contacts made by cIMP with the CNBD We also examined the structure of the HCN2 C-linker/CNBD liganded to the cAMP-cGMP intermediates, knowing that cAMP binds in the anti configuration while cGMP binds in the syn configuration (Zagotta et al., 2003). Crystals of HCN2 saturated with cPuMP or 2-NH2-cPUMP diffracted at a resolution of ~2 Ǻ and both ligands bind in the anti-configuration as for cAMP (Figure 3.6; Table 3.3; Appendix 6). Cyclic IMP also bound in the anti configuration, which was somewhat surprising because none of the cAMP-cGMP intermediates bound in the syn configuration. Together, these data suggest that both the NH2 group at C2 and the keto-oxygen at N6 are required for cGMP binding in the syn configuration.  125      126   Figure 3.7 Effect of cUMP on the binding and oligomerization R635A mutation in HCN2 C-linker/CNBD. (A) Plots of heat and binding isotherms produced upon progressive injections of cAMP and cUMP to 200 μM of R635A HCN2 C-linker/CNBD measured by ITC. The solid line through the values represents either a single binding site model (cUMP) or a two independent binding site model (cAMP), which yielded values for affinity and energetics (ΔG, ΔH, and ΔS). Values for binding affinity are shown in Table 3.1. Values for thermodynamics are found in Appendix 15. (B) Bar graph showing the thermodynamics of binding, which were determined from the fit in A. Values in the graph represent means ± s.e.m. (C) Shown are plots of estimated average molecular weight versus concentration of purified HCN2 C-linker/CNBD protein. The estimated molecular weight was determined by DLS and is proportional to the population of oligomers in solution. An increased in oligomerization occurs with increasing protein concentration and further increased in the presence of cAMP and cUMP. Values represent means ± s.e.m, determined from three separate preparations of purified protein measured on different days. (D) Activation curves determined by plotting tail current amplitudes, which were normalized to their maximum value (I/Imax), versus test voltage. The solid lines through the values represent fitting of a Boltzmann curve by equation 1 (section 3.2.5). Fits yielded values for V1/2 and k, which are found in Table 3.1 along with other parameters. Both cAMP and cUMP produce a shift in the activation curve that is comparable to the shift they produce in the wild type channel.  127  Because the analogues bind in the anti-configuration, the structure of the C-linker/CNBD bound with cIMP, cPuMP and 2-NH2 cPuMP can be compared directly with the cAMP-bound structure.  Knowing the binding configurations, the binding pocket of cIMP can be compared to full agonists, cAMP, cPuMP, or 2-NH2-cPuMP, to understand its low binding affinity and reduced ability to promote tetramerization and facilitate opening. Although many of the specific interactions are similar between cIMP and the other cyclic nucleotides, major differences in interactions occur, again, at residues E582 and C584 found in the P-helix between beta strands 6 and 7, and at residues R632 and R635 found in the C-helix (Figure 3.6; Table 3.4). The nitrogen of the backbone of C584 is farther from the axial oxygen of the cyclic phosphate group of cIMP, 2-NH2-cPuMP and cPuMP than that of cAMP and does not form a hydrogen bond. The nitrogen of the backbone of E582 is farther from the equatorial oxygen of the cyclic phosphate group of cIMP and 2-NH2-cPuMP than that of cAMP and cPuMP, again reducing the probability of forming a hydrogen bond. The E582 backbone nitrogen also is farther from the 2’-OH group of the cIMP ribose and, unlike this interaction for the other three cyclic nucleotides, will also not form a hydrogen bond. The main chain carbonyl of R632 is close and forms a hydrogen bond with the 6-NH2 group of cAMP but it creates an electrostatic repulsion with the 6-keto group of cIMP instead due to the partial negative charges; this main chain carbonyl does not interact with the other two ligands, which lack an exocyclic group at position 6 of the purine ring. The side chain of R635 forms a close interaction with the 6-keto group of cIMP, likely forming a hydrogen bond not found in the other cyclic nucleotides. In summary, the keto-carbonyl repulsion causes a greater distance between cIMP and the P-helix, while distal C-helix, especially the R635 side chain, is more closely associated with cIMP than with the other three cyclic nucleotides.  128  3.4 Discussion To better understand the nature of cyclic nucleotide agonism in the HCN2 channel and to identify partial agonists, we examined the ability of cAMP analogues to induce changes in the structure of the C-linker/CNBD and correlated these changes with the degree of facilitation of opening. We find that the degree of oligomerization correlates with degree of facilitation of opening. Unlike the other analogues, cCMP and cIMP were unable to promote oligomerization of the HCN2 C-linker/CNBD in vitro. Because oligomerization in solution is thought to represent the formation of a gating ring in the full-length channel to facilitate opening (Chow et al., 2012; Lolicato et al., 2011; Zagotta et al., 2003), the inability of cCMP and cIMP to do so suggests that a gating ring is inefficiently promoted in the full-length HCN2 channel by these molecules.  Both cCMP and cIMP are less effective facilitators of HCN2 opening as compared to the other ligands. Cyclic CMP is a pyrimidine cyclic nucleotide that has been shown previously to act as a partial agonist of the HCN2 and HCN4 channels and of sinoatrial HCN channels (D. DiFrancesco & Tortora, 1991; Zong et al., 2012). We found that the effect of cCMP on the HCN2 channel was indeed limited as compared to most of the other agonists. In contrast, cUMP, another pyrimidine cyclic nucleotide, was able to promote oligomerization of the HCN2 C-linker/CNBD and bind with negative cooperativity, and it produced stronger facilitation of opening. Unlike cIMP, cPuMP and 2-NH2-cPUMP were able to promote oligomerization of the HCN2 C-linker/CNBD and bind with negative cooperativity, and they also produced greater facilitation of HCN2 opening than cIMP. Together with the actions of cAMP and cGMP, these data suggest that the degree of facilitation for a given ligand depends, at least in part, by its ability to promote a gating ring in the full-length channel.  129  The three cAMP-cGMP intermediates bound to the HCN2 C-linker/CNBD in the anti configuration like cAMP and not like cGMP, which is known to bind in the syn configuration (Zagotta et al., 2003). Because the three cAMP-cGMP intermediates do not bind to the HCN2 C-linker/CNBD in the syn configuration, less information can be gleaned from the binding data of these analogues about the atoms important for controlling cGMP binding affinity and effect. An important finding that does arise from these data is that both exocyclic groups, the 2-keto and 6-NH2 groups, are required in order for cGMP to bind in the syn conformation. These data appear to follow the syn/anti configurations of the cyclic nucleotides in water with cGMP existing in syn configuration to a greater extent than cAMP (L. Zhou & Siegelbaum, 2008).     The resolved crystal structures of the HCN2 C-linker/CNBD in complex with the ligands suggest that interactions in regions within the sixth and seventh beta strands and in the distal C-helix of the CNBD contribute to the allosteric changes which lead to facilitation of opening. These interactions are notable at or near residue R632 in the distal C-helix, which has been proposed as a residue critical for allosteric signalling between ligand binding and channel opening (L. Zhou & Siegelbaum, 2007); differences were found near this residue when the cCMP-bound and cUMP structures were compared as well as when the cIMP bound structure was compared to the cAMP, cPuMP and 2-NH2-cPuMP bound structures. Our data fit with previous data in HCN2 and HCN4 using NMR, FRET and DEER suggesting that cAMP binding shifts the B- and C-helices toward the β-roll (Akimoto et al., 2014; Puljung et al., 2014; Saponaro et al., 2014; Taraska et al., 2009; VanSchouwen, Akimoto, Sayadi, Fogolari, & Melacini, 2015), and orders the P-helix and distal C-helix, and that the shifts induced by cCMP are less stable than those produced by cAMP (Akimoto et al., 2014). Finally, as demonstrated by the binding and action of cPuMP, the interactions made by exocyclic groups are not required for 130  purine cyclic nucleotide promotion of tetramerization of the C-linker/CNBD and strong facilitation, and make only minor contributions to binding affinity. However, as shown by cIMP, the addition of a keto group at position 6 is sufficient to weaken binding, limit tetramerization of the C-linker/CNBD and render facilitation of channel opening less effective.  An understanding of how cAMP and its analogs act on HCN channels is important for several reasons. First, cCMP, cIMP and cUMP are experiencing a re-birth as potential second messengers in vivo (Beste & Seifert, 2013) and knowing their mechanism of action is a necessary prerequisite for understanding their role in vivo. Second, understanding how these molecules act will help to understand cyclic nucleotide agonism in general in HCN channels, which may be useful for development of therapeutic molecules that control their activity. Currently, the drug ivabradine, an HCN inhibitor, is available in the clinic for treatment of angina and for heart failure due to its ability to limit increases in heart rate that exacerbates those conditions (D. DiFrancesco & Borer, 2007; Roubille & Tardif, 2013). However, ivabradine, a pore blocker (Bucchi et al., 2002; Bucchi et al., 2013; Bucchi, Tognati, Milanesi, Baruscotti, & DiFrancesco, 2006), is associated with side effects such as compromised vision and profound bradycardia (Fox et al., 2013), rendering it unavailable for a sub-set of patients. Another potential application for HCN inhibition is in the control of inflammatory and neuropathic pain, which is reduced or eliminated in mice that have had the HCN2 gene deleted from neurons of the dorsal root ganglia (Emery, Young, Berrocoso, Chen, & McNaughton, 2011). In these and other examples, keys for the development of HCN modulators may well be isoform specificity (Emery, Young, & McNaughton, 2012) and more subtle effects that could potentially be attained by targeting the CNBD, which modulates ion flow, rather than by targeting the pore, which is required for ion flow.    131  Chapter 4: Two Disease Mutations in the Carboxy-Terminal Region Impact Ligand Binding and Effect  4.1 Introduction HCN channels, coded by four mammalian genes (HCN1-4), underlie the hyperpolarization-activated current, so designated Ih, which is known for its contribution to repetitive action potential firing in neurons and conduction tissue of the heart  (D. DiFrancesco, 1993; Pape, 1996). This current has a number of “funny” properties that contribute to the spontaneous, regular beating rhythm of electrical excitable cells and thus Ih in the heart is often referred to as If (D. DiFrancesco, 1993). In these cells hyperpolarization of the membrane potential opens HCN channels and permits the inward flow of sodium, and lesser outflow of potassium, which slowly depolarizes the membrane potential towards threshold for action potential firing. During repolarization of the action potential, HCN channels are once again activated and the cycle repeats itself.   In the sinoatrial node, Ih is modulated by β-adrenergic and muscarinic stimulation, which respectively stimulates or inhibits cAMP production by adenylyl cyclase (Brown et al., 1979; D. DiFrancesco et al., 1989; D. DiFrancesco et al., 1986; D. DiFrancesco & Tortora, 1991). Additional facilitation of Ih channel opening occurs by the direct action of cAMP on the cytoplasmic side of the channel with a half-maximal cAMP concentration (EC50) of 0.2 µM (D. DiFrancesco & Tortora, 1991). Facilitation by cAMP is marked by a depolarizing shift in the range of voltages over which the channel opens, ranging from ~11 to 14 mV in the SAN and accounting for most of the additive shift of ~18 mV seen upon maximal stimulation with ß-132  adrenergic and muscarinic agonists (Accili et al., 1997; D. DiFrancesco & Mangoni, 1994). The effect of cAMP is largest in HCN2 and HCN4 channel isoforms (Stieber et al., 2005). The direct action of cAMP on the cytoplasmic side of patches excised from SAN myocytes (D. DiFrancesco & Tortora, 1991) suggested that the channel itself contained a region to which this cyclic nucleotide binds, instead of indirectly via PKA phosphorylation. Indeed, the initial cloning of subunits underlying Ih showed that this is the case. The primary structure of HCN subunits are similar to potassium channels and, thus, are predicted to have six transmembrane helices (S1-S6) with a cytosolic N- and C-terminus, and to combine as tetramers to form the ion-conducting channel (Siu, Lieu, & Li, 2006). The C-terminus contains a cyclic nucleotide-binding domain (CNBD) and a C-linker that connects this domain to the pore (S6). A tetrameric structure predicts that each individual HCN channel has four cAMP binding sites.  As an essential modulator in the thalamus, the HCN2 channel contributes to neuronal excitability and rhythmicity (Robinson & Siegelbaum, 2003; Wahl-Schott & Biel, 2009). Consistent with a role in the brain, HCN2 dysfunction has been implicated as an underlying cause of diseases such as epilepsy (Bender & Baram, 2008; Chung et al., 2009; Dibbens et al., 2010; Dubé et al., 2010; Ludwig et al., 2003; Reid, Phillips, & Petrou, 2012; Tang et al., 2008).  The physiological role of the channel is tested by knocking out the gene in mice. The homozygous HCN2-deficient mouse shows prominent reduction of Ih in thalamocortical neurons, leading to a hyperpolarized shift in the resting membrane potential of these cells and to the cause of reduced locomotor activity and spontaneous absence epilepsy (D. Kim et al., 2001; Ludwig et al., 2003; McCormick & Bal, 1997). Recently, a point mutation, E515K (E488K in mouse, used in this chapter) in the C-linker of human HCN2 channel was associated with generalized idiopathic epilepsy (J. C. DiFrancesco et al., 2011). Compared to the wild type HCN2 channel, 133  hyperpolarization-induced opening of this mutant HCN2 channel is strongly inhibited, but the maximum effect of cAMP is not affected. The molecular mechanism underlying the inhibition of hyperpolarization-induced opening is not known.  On the other hand, the HCN4 channel is the predominant isoform in the sinoatrial node (SAN) of the heart and plays an integral role in pacemaking. Knockouts of the HCN4 isoform, either globally or specifically in cardiac tissues, have resulted in embryonic death, profound bradycardia, and an 85% reduction in Ih in the SAN (Stieber et al., 2003). HCN4 channels have a significant role in the maturation of embryonic heart and pacemaker cells; the gene is expressed during the progression of development in the SAN (Garcia-Frigola, Shi, & Evans, 2003; Vicente-Steijn et al., 2011) . In the temporally-controlled knockout of HCN4 in mature adult heart, Ih is reduced by 80% but despite the reduction, the cardiac electrical activity is normal. The arrhythmic phenotype arises from frequent sinus pauses (Herrmann et al., 2007; Hoesl et al., 2008). The importance of the HCN4 CNBD to pacemaking is strongly supported by unstable and/or low heart rates in mice with overexpression of this channel that are insensitive to cAMP due to either a mutation within the CNBD (Harzheim et al., 2008) or a deletion of the CNBD altogether (Alig et al., 2009).  A mutation in the human HCN4 CNBD, S672R, was shown to associate with a bradycardia phenotype in an Italian family (Milanesi et al., 2006). The HCN4 mutation did not greatly affect the potency (EC50) or maximum effect of cAMP, but it shifted the range of mutant channel activation to more negative voltages in the absence of cAMP. This hyperpolarizing shift in mutant channel activation is proposed to mimic the effects of acetylcholine and reduce the amount of depolarizing current flowing during diastole and, thus, to be the underlying mechanism for the bradycardia phenotype. A subsequent study (Xu et al., 2012) suggested that 134  there was reduced binding of cAMP to the mutant HCN4 channel and that that the potency of cAMP was substantially reduced; however, analysis of potency was carried out in a HCN2-HCN4 chimera, unlike in the original study where potency was determined in the wild type HCN4 channel. Furthermore, such a large reduction in cAMP potency, as pointed out in a recent review of HCN4 mutations, would cause chronotropic incompetence, a feature that was not observed in the bradycardia patients.  Here, we examine these two disease-associated mutations. Based on our data, we offer a molecular explanation for how these mutations affect channel function.   4.2 Experimental procedures 4.2.1 Cloning, site-directed mutagenesis, expression and purification The C-linker/CNBD segment of mouse HCN2 (i. 443-645) and rabbit HCN4 (i. 522-725, Appendix 1B) were cloned into a modified pET28 vector (Novagen). A His-tag, maltose binding protein, and TEV protease cleavage site was inserted in the N-terminal side of the construct of interest (Appendix 1A). Site-directed mutagenesis was performed using the Quikchange protocol (Stratagene). In Section 4.3, glutamate 488 was mutated to a lysine, while tyrosine 459 was mutated to an alanine in mouse HCN2. In Section 4.4, serine 672 was mutated to an arginine in rabbit HCN4. Refer to Appendix 2A/B for the primers. Protein are expressed at 37oC in E. coli Rosetta (DE3) pLacI cells (Novagen), induced at OD600 of 0.6 with 0.4 M IPTG, and inoculated for another 4 hours before harvesting. The cell pellets were resolubilized and lysed with sonication in 250 mM KCl and 10 mM Hepes at 7.4 (buffer A), and the addition of glycerol, 25 µg/ml DNase I and lysozyme, 0.4 mM EDTA, and 10 mM PMSF. The constructs were applied to Talon (cobalt-bound-resin) columns, and eluted with 135  25 mM KCl and 500mM imidazole at pH 7.4. The elute was dialyzed with TEV protease in buffer A for two hours. The content was run on another Talon column and the flowthrough was collected. The protein was then charged in dialysis buffer with 10 mM KCl + 20 mM MES at pH 6.0 for two hours and applied to a ResourceS column (GE Healthcare) at an increasing KCl concentration from 0% to 50%. The protein was confirmed on a SDSPAGE gel, and was dialyzed in 150 mM KCl and 10 mM HEPES at pH 7.4. The protein was concentrated to 20 mg/ml using 10,000 MWCO concentrator (Amicon, Millipore), and determined with spectrophotometry and the Edelhoch method. Refer to Section 2.2.2 for a more detailed outline on the purification protocol.  4.2.2 Crystallization, data collection and structure determination Crystals were obtained using the hanging-drop method at 4°C. 1 μL of protein solution was mixed with 1 μL of reservoir solution of 200 mM NaCl, 0.1 mM citric acid and 15% PEG 400. Crystals were cryoprotected in reservoir solution supplemented with 20% glycerol and flash frozen in liquid nitrogen. Diffraction data sets were collected at the Advance Photon Source (APS) beamline 23-ID-D-GM/CA. Data was processed with XDS (Kabsch, 2010), phased by molecular replacement with 1Q5O, and then refined using phenix.refine with TLS restraints and successive rounds of manual modeling in COOT (Adams et al., 2010; Emsley et al., 2010). Data collection and refinement statistics are available in (Table 4.1). Analysis of interactions between intersubunit interfaces was done by PDBePISA (Krissinel & Henrick, 2007) and final diagram was prepared by PYMOL (Version 1.8 Schrödinger, LLC).  136  4.2.3 Isothermal titration calorimetry (ITC) The HCN2 constructs were prepared to 400 µM while HCN4 constructs to 200 µM, and the protein was inserted in the sample cell. Filtered water was added into the reference cell. The syringe was loaded with 4 mM or 2 mM of cAMP, which was 10x the concentration of the HCN C-linker/CNBD. Titration experiments were performed in an ITC200 (GE Healthcare) at 25ºC. The ligand injected into the sample cell 1 µL at a time, for a total of 40 injections. The heat needed to compensate for the temperature difference between the sample and reference cell was recorded for each injection, and all heat was integrated to generate the binding isotherm. Fit was done with a two-independent-site binding model on Origin 7.0 (MicroCal ITC add-on), which generated the values of affinity and energetics. Refer to Section 2.2.4 for the equations behind the numerical values.  4.2.4 Dynamic light scattering (DLS) The protein was diluted to 12.5, 25. 50, 100, 200, or 400 µM into a 50 µL mixture either in the absence or presence of ligand (at a 10:1 ratio of ligand to protein). Samples were centrifuged for 2 minutes at 6000 rpm to remove dust particles before adding to a 384-multiwell microtiter plate. DLS measures the fluctuation in Brownian motion on a DynaPro Plate Reader (Wyatt Technology). Acquisitions were performed at 22ºC with Dynamics 7.0 with 10 scans of 5 seconds each. Acquisitions with polydispersity over 30% were discarded. The values were autocorrected and converted to apparent molecular weights using Dynamic 7.0 and statistics were performed with Origin 7.0. Molecular weights of oligomeric forms of protein were easily separated from aggregated protein, which was identified by its relatively immense value. Refer to Section 2.2.5 for the equations behind the numerical values. 137   4.2.5 Thermal melt analysis  Melting curves were done either in the absence, or presence of cAMP or cGMP. The protein was diluted to 50 µM, and the ligand to 500 µM (if any). Samples for the experiments were 50 µL in total, containing the protein, ligand, and 1x SYPRO Orange solution (Invitrogen) according to manufacturer’s instructions. The samples were inserted and fluorescence was detected in a DNA engine opticon 2 real-time PCR machine (Bio-Rad) with SYBR green filter. The dye fluoresced upon binding to hydrophobic residues, which were exposed during thermal denaturation. The temperature increased from 25oC to 95oC at 0.5oC intervals, and each temperature was held for 15 seconds. The Y-axis of the melting curve was the normalized fluorescence, related to the fraction of unfolded protein. Negative control (dye in buffer) is subtracted from individual trials. Maximal fluorescence was when protein was completely unfolded, and it decreased when the dye dissociated from the protein. Residual signal could be due to the dye binding to aggregated protein. The inflection point on the melting curve corresponds to Tm (mid-unfolding temperature), and was alternatively determined with the first derivative (dRFU/dT) of the fluorescence intensities.    138  4.3 Results: The epilepsy-associated mutation E488K in the HCN2 channel 4.3.1 Modified interactions originating at the E488K residue reduce self-association of the mutant HCN2 C-linker/CNBD Experimental evidence suggests that oligomerization of the HCN2 C-linker/CNBD is triggered in solution, where the C-linker is in an active, rather than resting, conformation (Craven & Zagotta, 2004; Zagotta et al., 2003). Based on the proposed gating ring theory, the propensity of the HCN2 C-linker/CNBD to self-associate is correlated to channel opening (Chapter Three of the thesis). Since the opening is inhibited in the epilepsy mutant channel (J. C. DiFrancesco et al., 2011), a reduced ability of E488K HCN2 C-linker/CNBD to oligomerize might also be expected.  The oligomeric state of the E488K HCN2 C-linker/CNBD as a function of protein concentration is determined by Dynamic Light Scattering (DLS), which measures the scattered wave pattern based on apparent radius of molecules in solution. The apparent weight of the wild type HCN2 C-terminus increased with higher protein concentrations supporting that the C-terminus can self-associate into a gating ring even in the absence of cAMP (Figure 3.1, Figure 4.1A); the effect of cAMP is consistent with the proposal, where tetrameric ring is promoted and hence larger molecular weights were observed especially at lower protein concentration. The mutation has substantial effects on both the ability to self-associate and to tetramerize. The mutated HCN2 C-linker/CNBD requires higher local concentration of protein to reach the same degree of oligomerization, and the effect of cAMP is greatly reduced (Figure 4.1B). The reduced tendency of E488K-HCN2J to oligomerize is consistent with the resting C-linker being favoured. Similar result was obtained with an alanine substitution in position 488 (data not shown). The 139  limitation of active C-linker is due to a loss of glutamate rather than new interactions with the lysine side chain.  Figure 4.1 Self-association and ligand-induced oligomerization are eliminated by mutations that abolish the E488-Y459 intersubunit interaction. Plots of apparent molecular weight versus concentration of (A) wild type HCN2 C-linker/CNBD, (B) E488K C-linker/CNBD, or (C) Y459A C-linker/CNBD without cAMP (white) and with excess cAMP (closed) as determined by DLS. The wild type data is comparable to Figure 3.1 as previously conducted where increased protein concentration promotes self-association and addition of cAMP promotes further oligomerization. Both abilities are limited when the hydrogen bond E488-Y459 is abolished. Data plotted are mean values ± s.e.m.    140  4.3.2 Modified interactions originating at the E488K residue eliminate negatively cooperative binding of cAMP to the mutant HCN2 C-linker/CNBD   We next examined the ability of cAMP to bind to the tetrameric wild type or mutant C-linker/CNBD. The failure to bind may explain the small changes in ligand-induced tetramerization. Also, the intersubunit communication via the C-linker determines oligomerization, as well as negative cooperative binding of cAMP, which can be measured using isothermal titration calorimetry (ITC). Negative cooperativity was observed previously by ITC, and the tetrameric C-linker/CNBD complex was achieved at 200 µM without aid of cAMP. Due to the reduced effect in self-association, we used 400 µM of C-linker/CNBD instead to ensure tetrameric form. As found previously, the binding of cAMP to the C-linker/CNBD produces a pattern of heat release that is fitted with two independent binding sites model. The stoichiometry and the dissociation constants of the fit for both events imply that one molecule of cAMP binds to one of the subunits in the tetramer with higher affinity, which results in a conformational change that reduces the subsequent three binding sites. This negative cooperative binding is again observed here for wild type, with 0.12 and 0.86 µM for the two events, respectively (Figure 4.2A). In contrast, the addition of cAMP to tetrameric E488K-HCN2 C-linker/CNBD elicited a pattern of heat production that was fit best with a single binding site model, yielding a binding affinity of ~2.56 μM (Figure 4.2A). The energetics of binding for the single cAMP binding event in E488K-HCN2 C-linker/CNBD were not similar to those for either the low or high affinity binding event in the wild type (Figure 4.2B), but were more similar to those we found previously for cAMP binding to the monomeric form of HCN2 C-linker/CNBD lacking the first two helices of the C-linker (Chow et al., 2012). The similarity in binding energetics 141  between the mutant tetramer and the monomeric form of wild type HCN2 suggest that C-linker-mediated intersubunit interactions are also lacking in the mutant C-linker/CNBD.    Figure 4.2 Negative cooperativity is eliminated when E488-Y459 interaction is abolished. (A) Plots of heat produced upon progressive injections of cAMP into 400 µM of wild type HCN2, E488K, and Y459A C-linker/CNBD, as indicated. The wild type data is comparable to Figure 3.2 from previous experiment, and was fitted with a two-independent site binding model. The solid line through the values of the lower plots represents a single binding model in the two mutants, suggesting that cAMP does not bind with negative cooperativity. The fits yielded values for affinity and energetics (ΔG, ΔH, and -TΔS). (B) Bar graph comparing the energetics of binding of cAMP to wild type HCN2 (left) and the two mutant C-linker/CNBD (right), as determined from fitting the data in (A). Numerical values can be found in Appendix 11. Data plotted are mean values +/- s.e.m.  142  4.3.3 The crystal structure of the epilepsy mutant exhibits an abolished intersubunit interaction Rather than a global effect, the reduced effect to tetramerize could be local between the site of the mutation and nearby residues. We solved the structure of E488K HCN2 C-linker/CNBD by x-ray crystallography to 3.5Ǻ, using molecular replacement of 1Q5O. Like that structure of wild type HCN2 C-linker/CNBD, the structure is in the same space group and also has a four-fold symmetry (Figure 4.3A, B). Aligning the previously solved structure and the new mutant structure, the cyclic nucleotide binding domain has very little difference with a RMS of 0.324. There is complete overlap between cAMP of the two structures, as well as three residues (R591, T592, and R632) that are known to make important contacts with the ligand (Figure 4.3C). The C-linker region overlaps with a RMS of 0.581 (and a global RMS of 0.498), where the biggest difference is at the site of the mutation. Glutamate 488 in the C’ helix of the C-linker normally makes a hydrogen bond with tyrosine 469 in the A’ helix of the neighbouring subunit (Figure 4.3D). This is one of the intersubunit bonds that hold the tetramer together. Without the bonding partner, the neighbouring Y459 recedes away from the adjacent subunit by 1.1A. Looking at the published structures in protein database (1Q5O, 1Q3E, 1Q43, 3BPZ, 3FFQ), this hydrogen bond is conserved, signifying the importance of this bond. Other hydrogen bonds were compared as well using PISA, but no new hydrogen bonds were created. The global structure remains as a tetramer, which is consistent with our DLS data that self-association is not affected. However, the lack of this bond could be involved in limiting that ability.  143   Figure 4.3 Crystal structure of E488K-HCN2 C-linker/CNBD in the presence of cAMP.  (A) Ribbon representation of the E488K HCN2 C-terminus monomer (pink), overlaid with the wild type C-terminus (grey). The binding pocket is bordered by a blue box, and magnified in Figure 4.3C. The mutation site, distant from the binding pocket, is represented by spheres. (B)  Ribbon representation of the tetramer, for both the mutant (shades of pink) and wild type (shades of grey). The mutation site is bordered by a blue box, and magnified in Figure 4.3D. The spheres are located in an intersubunit region. (C) Close-up view on the binding pocket, with the ligand and three interacting residues. The residues are nicely overlaid between the mutant and the wild type C-linker/CNBD. (D) Close-up view on the mutation site and the intersubunit interaction. Glutamate 488 normally interacts with tyrosine 459 from the adjacent neighbour (black dash). The side chain of the mutation E488K swings away from the tyrosine residue and the intersubunit interaction is abolished. 144   E488K mHCN2 C-linker/CNBD + cAMP Data collection  Space group P 4 21 2 Cell dimensions       a, b, c (Å) 97.087, 97.087, 46.243      α, β, γ (°) 90, 90, 90 Resolution (Å) 48.54 – 2.298 (2.38-2.298) Rsym (%) 8.7 (94.9) I/ σI 14.11 (2.13) Completeness (%) 99.71 (100)    Refinement  Resolution  48.54 – 2.30 No. reflections 10297 (999) Rwork/Rfree 0.1775/0.2265 No. atoms 1667      Protein 1610      Ligand 12      Water 45 B-factors (Average) 48.30      Protein 48.10      Ligand 69.30      Water 49.60 r.m.s.d.       Bond lengths (Å) 0.007      Bond angles (o) 0.999   Ramachandran outliers (%) 0.0 Ramachandran favored (%) 99.0  Table 4.1 Data collection and refinement statistics for E488K mHCN2 C-linker/CNBD co-crystallized with cAMP.  Data from a single crystal was integrated with HKL2000, and refinement was performed using phenix.refine of PHENIX. Values in parentheses are for the highest resolution shell. Refer to caption of Table 3.3 for formulas of the R factors.  145  4.3.4 The hydrogen bond acceptor of the intersubunit interaction also affects tetramerization and binding Unlike glutamate 488 in the crystal structure of the wild type HCN2 C-linker/CNBD, lysine 488 in the mutant crystal structure does not interact with tyrosine at position 459 of the adjacent subunit. To determine if this interaction contributes to the relative preference for the resting conformation of the C-linker, we mutated Y459 to an alanine in the wild type HCN2 C-terminus and examined oligomerization by DLS. We found that Y459A HCN2 C-linker/CNBD, like E488K mutation, must be present in comparatively higher concentrations in order to form oligomers in solution (Figure 4.1C). The reduced tendency of Y459A- and E488K-HCN2 C-linker/CNBD to oligomerize suggests that this specific intersubunit interaction contributes to shifting the preference towards the resting C-linker and limiting the formation of a gating ring.  Since the C-linker interaction also contributes to cooperative binding, we also tested Y459A mutation with ITC, and again found a single apparent binding event with a dissociation constant of ~8 µM (Figure 4.2A). The energetics is also similar to that of E488K mutation, solidifying that both hydrogen-bond partners resemble the monomeric form.  4.3.5 The E488K mutation does not impact the stability of the C-linker/CNBD We next carried out thermal melting measurements to determine if the mutation of glutamate 488 to lysine alters the stability of the HCN2 C-linker/CNBD. A change in stability could cause the destabilizing of the gating ring, explaining the shift in equilibrium and leading to the intrinsic defects. These measurements showed minimal changes in the mid-unfolding temperature (Tm), determined by the temperature at which the greatest change in fluorescence is observed. Based on our hypothesis, the C-linker in the absence of cAMP has higher tendency to 146  be in the resting state, whereas cAMP increases the population of active state in solution. The experiment was carried at 50 µM of protein, which, according to DLS, is mainly monomeric without cAMP and tetrameric with cAMP, depicting the two different states respectively. The data suggests that the two conformations do not differ in stability. In addition, the E488K mutation has approximately the same Tm as the wild type in both the absence and presence of cAMP at ~36oC (Figure 4.4). The single amino acid change did not destabilize the global structure of the C-linker/CNBD, and thus eliminating the connection between destabilization and intrinsic defects.  Figure 4.4 Thermal stability is not greatly modified by E488K mutations in the C-linker or by the binding of cAMP.  Plots of normalized fluorescence versus temperature of indicated HCN2 C-linker/CNBD constructs, in the presence or absence of cAMP. SYPRO orange is a hydrophobic dye that binds to the protein hydrophobic core upon heat denaturation, thus increasing the fluorescence significantly. Maximal fluorescence intensity is observed when the protein is completely unfolded, followed by dye-protein dissociation when intensity decreases. The mid-unfolding temperature (Tm) is determined by the inflection point, or the global maximum of the first derivative of the curve. The mid-unfolding temperatures (oC) are 36.3 ± 1.0 and 36.0 ± 1.3 for HCN2, 33.3 ± 2.6 and 36.2 ± 0.8 for HCN2 E488K, without and with cAMP, respectively. The stability is not significantly affected by mutation or addition of cAMP.   147  4.4 Results: The S672R bradycardia-associated mutation in the HCN4 channel 4.4.1 Location of the S672R mutation in the HCN4 C-linker/CNBD The serine 672 residue is highly conserved among HCN isoforms and located in the seventh beta sheet of the CNBD. Its location is shown in Figure 4.5, with and without the mutation (Xu et al., 2012; Xu et al., 2010). Both the wild type serine and mutated arginine side chain point in the opposite direction of the cAMP binding pocket. Although not in direct contact with the ligand, S672 is close to a well conserved residue, arginine 669, which resides in the phosphate binding cassette and makes a critical contact with the phosphate group of the cyclic nucleotide (Lolicato et al., 2011; Weber & Steitz, 1987).         148   Figure 4.5 Location of S672 in HCN4 C-linker/CNBD.  (A) The fragment of HCN4 C-linker/CNBD (i. 521-739) of HCN4 co-crystallized with cAMP (pdb: 3OTF, green) and represented with ribbon backbone. This is overlaid with the same ligand but for the S672R mutant construct (pdb: 4HBN, blue), after aligning the peptide main chain atoms. In both cases, cAMP was displayed in sticks. The residue is positioned in the seventh β-sheet of the CNBD. (B) The inset shows a close-up at the binding pocket and the serine or arginine residue is displayed in sticks, in corresponding colours. These residues point away from where cAMP binds.  4.4.2 Cyclic AMP and cGMP promote oligomerization of both the wild type and mutant HCN4 C-linker/CNBD We compared the oligomeric nature of the wild type and S672R HCN4 C-linker/CNBD, and the effect of cAMP and cGMP on oligomerization, using DLS. We found that saturating concentrations of both cyclic nucleotides increased the estimated molecular weight of the C-terminus for both the wild type and mutant HCN4 channel (Figure 4.6). Because oligomerization by cAMP is thought to correlate with facilitation of opening, our data are consistent with the original study showing that the maximum effect of cAMP on the HCN4 149  channel, a depolarizing shift of the activation curve of about 15 mV, was unaltered by the S672R mutation (Milanesi et al., 2006). Oligomerization as a function of protein concentration or in response to cyclic nucleotide was not altered by the mutation.    Figure 4.6 cAMP and cGMP promote oligomerization of the mutant and wild type HCN4 C-linker/CNBD. (A) Plot of average apparent molecular mass versus concentration of HCN4 C-linker/CNBD, without (open square), or with a saturating (10x excess over concentration of protein) concentration of cAMP (solid black square) or cGMP (solid grey square), determined by DLS.  The average molecular weight is correlated with the proportion of each oligomerization states. The population of tetramers (~90 kDa) increases with increasing amount of proteins. The addition of the ligand promotes formation of tetramers at lower concentration, indicated by an increase in average MW on top of self-oligomerization. Error bars indicate s.e.m.  (B) Same plot of average MW over a range of increasing concentration with the mutant C-terminus, without (open circle), with 10x cAMP (solid black circle), or 10x cGMP (solid grey circle). The mutant also self-associates, and tetramerization can be induced by ligands.      150  4.4.3 Cyclic AMP and cGMP bind to the HCN4 C-linker/CNBD with negative cooperativity and lower affinity than to the wild type HCN4 C-linker/CNBD Using 200 µM of wild type protein, we found that cAMP produced a two-phase pattern in the binding isotherm that could be best fitted with a two-independent binding site model (Figure 4.7A); this yielded high and low affinity binding values of 0.06 µM and 0.69 µM, respectively, similar to what was found previously (Chow et al., 2012). For the mutant C-linker/CNBD, cAMP also produced a two-phase binding pattern with negative cooperativity, but lower values for binding affinity were obtained, 0.42 µM and 4.81 µM for the high and low binding affinity, respectively. It is important to note that the biphasic pattern is apparent because the release of heat for each of two binding events differs in amount and saturates over separate ranges of cyclic nucleotide concentration.  For cGMP binding to wild-type HCN4, a two-phase pattern in the binding isotherm was again best fitted with a two-independent binding site model; this yielded high and low affinity binding values of 0.16 µM and 1.74 µM, respectively. For the mutant C-linker/CNBD, cGMP also produced a two-phase binding pattern with negative cooperativity, but lower values for binding affinity, 0.81 µM and 9.29 µM (Figure 4.7A).  Like those for cAMP binding, the energetics of cGMP binding is different for each binding event (Figure 4.7B). The low affinity binding event is driven by favorable enthalpy and unfavorable entropy, whereas both enthalpy and entropy are favourable for the high affinity binding event. This pattern is conserved between the two ligands, and between the two constructs.  151     152  Figure 4.7 cAMP and cGMP bind to the mutant C-linker/CNBD with negative cooperativity but lower binding affinity than to the wild type. (A) The top is the raw binding isotherm, plotting the magnitude of heat difference between the sample and reference cell upon progressive addition of 2 mM cAMP or cGMP into the sample cell with 200 µM HCN4 C-linker/CNBD wild type or S672R mutation. The bottom plots the integrated heat per injection as a function of the amount of added ligand. This plot is fitted with a two-independent binding site model (solid line), and the fit yields the binding affinity and thermodynamic parameters (ΔH, ΔS, and ΔG) for each of the binding site. The first binding site binds more strongly followed by a weaker site signifies negative cooperativity. (B) Histogram of the binding affinities, or (C) of energetics comparing the two cooperative sites, as well as between wild type and the mutant. The high affinity site and the low affinity site are compared separately, as indicated. Numerical values can be found in Appendix 13. Error bars indicate s.e.m. between replicates.  4.4.4 The thermal stability is equal between the wild type and mutant HCN4 C-linker/CNBD Thermal melting experiments were performed to determine and compare the mid-unfolding temperature between constructs. We found that there is no difference between the thermal stability between the wild type and mutant C-linker/CNBD, similar to the findings in E488K mutation. The unfolding temperatures were ~39 oC for both constructs. Furthermore, the addition of either cAMP or cGMP did not alter the unfolding. Different concentration of proteins and ligands were examined to rule out the effect of protein:ligand ratio, and the temperature was always in the same range regardless of the combination. (Figure 4.8).  153   Figure 4.8 Thermofluor analysis shows that the mutant and wild type HCN4 C-termini have similar thermal stability. Plot of normalized fluorescence over a temperature range, from the temperature with the minimum up to that with the maximum fluorescence. The melting curve is sigmoidal and is correlated to the fraction unfolded. The mid-unfolding temperature (Tm) is determined at the inflection point; the values are included in parentheses below. Wild type data is presented in squares, without (open, 39.56 ± 0.32 oC), with cAMP (solid black, 39.26 ± 0.20 oC), or with cGMP (solid grey, 38.95 ± 0.22 oC). Similarly, S672R data is presented in circles, without (open, 39.21 ± 0.22 oC), with cAMP (solid black, 39.70 ± 0.23 oC), or with cGMP (solid grey, 39.38 ± 0.19 oC). The errors are in s.e.m. with n=12. The Tm is not changed significantly by introducing the mutation or by adding ligand.    154  4.5  Discussion The Ih generated by HCN channels are involved in regulating the rhythmic electrical activity in the brain and in the heart, and channel disruption by mutation leads to a disruption in this activity. Here, we have characterized two of these mutations in the HCN channels that limit channel opening by shifting the activation curve to more negative potentials, leading to the disease phenotype (DiFrancesco 2011, Milanesi, Baruscotti et al. 2006). The E488K is an epilepsy mutant located in the C-linker of the HCN2 channel, a region connecting the cyclic nucleotide-binding domain to the pore of the channel. Similarly, the S672R mutation in the CNBD of the HCN4 channel was associated with bradycardia in an Italian family. A subsequent study suggested that the binding affinity cAMP to the HCN4 channel was reduced by the mutation and that cAMP was less potent in a mutated HCN2-HCN4 chimera. Here, by introducing either mutation into the isolated carboxy-terminal C-linker/CNBD of the HCN channel, we were able to assess in detail its effect on structure, oligomerization and cAMP binding, and to determine the molecular mechanism of the underlying defect.   4.5.1 The impact of the epilepsy-associated mutation HCN2 E488K on the C-linker/CNBD The mutation E488K was reproduced in the soluble C-linker/CNBD. While normally the C-linker/CNBD of wild type would self-oligomerize with increased local protein concentration and would exhibit ligand-induced oligomerization in the presence of cAMP (Chow et al., 2012), both of these properties were limited in the E488K mutation. We looked for the cause of the limited by visualizing the crystal structure. The resolved crystal structure of the E488K HCN2 C-linker/CNBD, like the wild type structure, demonstrated a C-linker region that is likely in its 155  resting state (Craven & Zagotta, 2004; Zagotta et al., 2003). The interactions summarized by PISA are very similar to those in the previously published structures. The only difference, not surprisingly, is the abolishing of a hydrogen bond between E488 and its partner, Y459 from the adjacent subunit. We proposed that the missing intersubunit interaction is, at least partially, the cause of the hyperpolarizing shift. Another observation is that cAMP binds to E488K pocket with no negative cooperativity, in contrast to its binding to wild type HCN2 C-linker/CNBD. The energetics of cAMP binding to the mutant are more similar to those obtained from cAMP binding to a monomeric form of the HCN2 C-linker/CNBD, which lacks the A’ and B’ alpha helices of the C-linker, remains monomeric even at high protein concentrations and shows no negative cooperativity upon cAMP binding (Chow et al., 2012). Thus, specific residues of the proximal C-linker are important for oligomerization of the C-linker/CNBD and intersubunit interactions that lead to negative cooperative binding of cAMP, as well as for channel gating (Craven & Zagotta, 2004; J. C. DiFrancesco et al., 2011). The absence of negative cooperativity in the mutant is consistent with the reduced intersubunit communication. On the other hand, the reduced affinity may reduce the potency of cAMP on this mutant channel, although this has not yet been examined. The energetics of cAMP binding to the mutant are more similar to those obtained from cAMP binding to a monomeric form of the HCN2 C-terminus, which lacks the A’ and B’ alpha helices of the C-linker, remains monomeric even at high protein concentrations and shows no negative cooperativity upon cAMP binding (Chow et al., 2012). Reduced potency of basal cAMP could also contribute a negative shift in mutant HCN2 channel activation in intact neurons.  Although counter-intuitive, a reduction of HCN channel activity and of Ih can lead to increase excitability in epileptic episodes. Inhibition of HCN opening by point mutation is the 156  predominant effect associated with cardiac disease and, recently, has been proposed as a defect that is expected to increase in excitability in the hippocampus and cortex in two ways, potentially leading to epilepsies (Poolos, 2012). First, the excitability of pyramidal and neocortical neurons is limited by HCN channel activity by increasing input resistance in dendrites and reducing EPSPs that are produced by synaptic inputs. Second, HCN channel activity augments excitability in non-pyramidal inhibitory interneurons of cortex and hippocampus because of their somatic localization and depolarizing influence on membrane potential. By disrupting HCN activity, pyramidal and neocortical neurons would become more active, and non-pyramidal interneurons become less active, increasing the excitability of the hippocampus and cortex. Another mechanism was proposed according to the phenotype HCN2-knockout mice. The lack of Ih in thalamocortical neurons removes inactivation of T-type calcium channels and lowers the threshold of firing upon the depolarizing signal, resulting in more rhythmic activity (Ludwig et al., 2003). We have shown a correlation between the ability to tetramerize and the shift in voltage dependence during gating modulation using partial agonists in Chapter Three. Similarly with the E488K mutation, favouring of the resting C-terminus conformation by disrupting the formation of the gating ring and inhibiting channel opening could lead to the observed negative shift and explain a common disease causing mechanism.  157  4.5.2 Impact of the bradycardia-associated mutation HCN4 S672R on the C-linker/CNBD The crystal structure for the S672R mutation was also similar (Xu et al., 2012). These structures are all deemed to be in a bound but resting state. Crystallography captures the protein in its most thermodynamically favoured form. The C-linker, when activated, may be more flexible and thus is reverted to the resting state during crystal packing. The mutant C-linker/CNBD and the wild type C-linker/CNBD are equal in thermal stability, regardless of whether cAMP or cGMP was present or not. This suggests that the intrinsic properties of the C-terminus are unperturbed, and that ligands do not further stabilize the conformation. The stability might not differ between these conformations in solution or the effect might only manifest when the mutant gating ring interacts with the transmembrane channel and pore. It could also be due to the thermal melt assay not being sensitive enough to distinguish the small changes in stability of individual constructs. Cyclic AMP and cGMP bind to both the wild type and mutant C-linker/CNBD with negative cooperativity, but that the affinity for both cyclic nucleotides is reduced by the mutation. A reduction in binding affinity is not surprising given the proximity of the residue to the phosphate binding cassette, a region that binds strongly to the phosphate group of the cyclic nucleotide. This residue might also contribute to allosteric changes that contribute to efficacy which, along with binding affinity, control potency.  The reduction in cAMP binding affinity for the mutant channel is consistent with data from original study (Milanesi et al., 2006). Although the potency of cAMP, which was determined by fitting the shift in activation range of HCN4 with a Hill equation, was unchanged, there is a notable and large decrease in the effect of cAMP on the mutant channel in the sub-158  micromolar range of concentrations. The difference in effect of cAMP between the wild type and mutant channel is reflected by the increase in slope of the Hill plot for the mutant channel. Indeed, the difference between high affinity binding affinity of cAMP to the wild type (~0.06µM) and mutant HCN4 channel (0.4µM) could reasonably explain the differences in effect of cAMP at sub-micromolar concentrations.   Nevertheless, we found that saturating concentrations of cAMP and cGMP were able to promote oligomerization of the HCN4 C-terminus. These data are consistent with the promotion of a gating ring by cAMP just under the pore of the full-length channel which facilitates opening and are consistent with the observation that the maximum effect of cAMP was unaltered by the mutation.  Our data suggest that the stability of the C-linker/CNBD and tetramerization (in the absence of cAMP) are not modified by the mutations, even though the position of the activation curve of the full-length HCN4 channel in the absence of cAMP is shifted to more negative potentials (Milanesi et al., 2006). Together, these data suggest that an interaction of the C-linker/CNBD with other parts of the channel is modified by the mutation.   The interaction could be with the intracellular S4-S5 linkers that has been shown involved in voltage gating (Kwan et al., 2012; Macri & Accili, 2004), or it could directly interact with the pore enhancing the inhibition or interfering with the relief of inhibition by the C-terminus. This mechanism was studied in insect HCN channels. Insects such as Apis mellifera (bees) and Drosophila melanogaster (fruit flies) only have one isoform coding for the channel, and to make up for the molecular diversity. In bees, when the S4-S5 linker is 32 amino acid residues longer, the half-maximal potential is shifted by 25 mV in the depolarized direction compared to other variants. The channel is still sensitive to cAMP, which shifts the activation curve even more (Gisselmann, Wetzel, Warnstedt, & Hatt, 2004). The extra clustered of 159  positively charged residues in the linker could alter the interaction with the C-linker, which was previously proposed to couple voltage sensing and channel activation (Decher et al., 2004). The mutation arginine could act via the same mechanism to produce its functional consequence, a hyperpolarizing shift of the activation curve, lessening the contribution of If in sinoatrial pacemaker and resulting in bradycardia. It has also been proposed that the C-terminus could interact with the N-terminus via electrostatic interactions to modulate channel function (Liu & Aldrich, 2011).   4.5.3 Concluding remarks  While comparing the mutations, although both E488K in HCN2 and S672R in HCN4 cause hyperpolarizing shift in the activation curve, they seem to act via different mechanisms during the gating process. The epilepsy-associated mutation appears to act by limiting gating ring formation and thus shifting channel activation to more negative potentials. The bradycardia-associated mutant is less sensitive to low levels of cAMP and cGMP, which reduce its contribution to the diastolic depolarization and heart rate under basal conditions. We propose that the effect of the S672R mutation on the channel in the absence of cAMP occurs because the mutation modifies an interaction with other parts of the channel such as the S4-S5 linker.        160  Chapter 5: Discussion  5.1 Overview The thesis focuses on the regulation of HCN activity by direct cAMP binding and by the C-terminal domains containing the binding site. We studied the direct interaction between cyclic nucleotides and the HCN binding domain, and how this binding leads to conformational changes that would explain channel facilitation. In addition to gaining insight into facilitation by cAMP, my studies may act as a guide for drug design to regulate the HCN channels therapeutically and to treat cardiac and neurological diseases. When the binding pocket is characterized, we can design a drug to achieve high affinity to outcompete the natural ligand and high selectivity to avoid side effects. The drug can enhance or inhibit the necessary changes to increase or decrease heart rate respectively if the mechanism behind ligand modulation is understood. We also looked at naturally occurring mutations in the binding domain of HCN channels to gain an understanding into the molecular mechanism underlying the disrupted channel, and in turn, to develop methods to correct or regain normal function. Here, a brief summary from each chapter will be followed by insights on the molecular determinants of ligand binding and gating ring formation. These insights lead to a modular model to explain the progressive events behind channel modulation by cyclic nucleotides. The Discussion ends with a section on the physiological relevance of my findings, the limitations of our approach, and future directions.    161  5.2 Chapter summaries Chapters 2-4 delved into mechanics of how cyclic nucleotides influence channel activity, from cyclic nucleotide binding to CNBD, to intersubunit interactions in C-linker, to channel activation. We used mutations in various locations in the C-linker/CNBD fragment or cAMP analogues to test and compare the effects on binding, oligomeric, structural, and functional properties. In the second chapter of the thesis, residues of the cyclic nucleotide binding domain which contribute to binding affinity were identified. Single substitutions of residues with alanine were introduced and binding affinity was measured by isothermal titration calorimetry. The effects of the alanine substitutions on binding affinity were compared to effects of the same substitutions on potency measured in the full-length channel previously (L. Zhou & Siegelbaum, 2007). They showed that the mutations produced no significant changes in the basal activation by voltage or in the magnitude of depolarizing shift of the activation curve induced by cyclic nucleotides in the full-length channel. We found that most of the alanine substitutions produced effects on binding affinity which paralleled their effects on potency. Since the potency reflects both the contribution from the affinity and subsequent conformational changes, the novel correlation dissects out the unknown post-binding event due to a site mutation. In general, high affinity leads to high potency in the channel. The residues that showed a non-proportional correlation between affinity and potency are proposed to be involved in post-binding conformational changes. These residues, L633 and I636, along with the already proven R632, reside in the C-helix. This is consistent with the movement of C-helix after cAMP binds (Puljung & Zagotta, 2013; Taraska et al., 2009). By comparing cAMP and cGMP, we can also conclude 162  cGMP is less potent due to a weaker affinity, and that I636 is an important determinant of cAMP binding preference as well as potency.   In the third chapter, we identified cyclic nucleotide analogues that bound to CNBD and produced partial and full maximum effects. The functional data separates analogues into full agonists (cAMP, cGMP, 2-NH2-cPuMP, cPuMP, and cUMP) and partial agonists (cCMP and cIMP). Their biochemical properties on the C-linker/CNBD also demonstrated the distinction between full and partial agonists. We found that there is a strong correlation between the ligand-induced tetramerization and the agonistic property. Partial agonists are less able to promote oligomerization of the HCN2 C-linker/CNBD in the presence of very high ligand concentration. We argue that the lack of oligomerization of the C-linker/CNBD leads to reduced formation of a stable gating ring, which ultimately leads to the cause of partial agonism. In the full channel, a conformational change allows gating ring formation, and a partial agonist is inefficient at stabilizing this change. Unfortunately, these changes were not captured by the crystal structures, possibly because the changes are subtle or that the resting state of the C-linker predominates for crystal contacts and masks over any structural changes. The changes in the C-linker and in the pore are not explained by C-linker/CNBD crystal structures, but the interactions and binding configurations of the ligand in the bound CNBD state are illustrated.   The fourth chapter focuses on two examples of disease mutations, each found in the C-linker/CNBD of the HCN2 and HCN4 channels, respectively, and they serve as models to test the notion of a gating ring. The epilepsy-associated mutation, E488K in HCN2 C-linker takes out an intersubunit interaction and leads to the observed reduction in oligomerization. This is consistent with the findings in the third chapter, where oligomerization is linked to gating ring formation. In this case, the missing hydrogen bond limits the population of active C-linker, 163  which correlates with the negative shift in channel activation (J. C. DiFrancesco et al., 2011). On the other hand, the bradycardia-associated mutation, S672R in HCN4 CNBD causes a negative shift in channel activation as well, but shows no change in oligomer formation. Thus, we argue that the mutation affects activation via another way, due to aggravating the inhibition by C-terminus or limiting the relief of that tonic inhibition (Wainger et al., 2001). In addition, the reduced affinity, similar to mutations in Chapter Two, changes the efficacy at concentrations in the physiological range, as depicted by the concentration-response curve (Milanesi et al., 2006). Even though both mutations change the basal gating properties, they have different mechanism of action and involve in different processes during gating.  5.3 Molecular determinants of ligand binding in the CNBD 5.3.1 Ligand determinants of anti and syn configuration in the CNBD In the first structures of the HCN2 isoform, two endogenous cyclic nucleotides were individually co-crystallized with the C-terminus binding pocket. Cyclic AMP and cyclic GMP were both found to bind to the same pocket, but in different configurations: anti and syn, respectively (Zagotta et al., 2003). Despite occupying the same binding site and both being full agonists in HCN channels, cAMP is more potent than cGMP in the cell (D. DiFrancesco & Tortora, 1991; Ludwig et al., 1998). In addition, using isothermal titration calorimetry, cAMP binds with higher affinity in both the high and low affinity binding events (Chapter Two). We questioned whether the difference in affinity and potency can be explained by difference in configurations. The two configurations are distinguished by the rotation about the N-glycosidic bond between the ribose and nitrogenous base. The rotation is possible for purines nucleotide, where 164  both configurations are found endogenously in DNA and RNA structures. In contrast, pyrimidine nucleotides do not take the syn configuration because the close proximity between the 2-keto of pyrimidine and its ribose ring makes the rotation sterically unfavourable. In solution, cAMP is found to favour the anti configuration, whereas cGMP favours the syn (Das et al., 2009). The syn is energetically preferred in solution because the 2-NH2 in the guanine could form a hydrogen bond with the axial phosphate oxygen of the same molecule.  Depending on the binding pocket, the composition and spatial orientation of amino acid residues could change the preference in configuration. For example, despite structural homology, both of the ligands bind in the syn configuration in PKA (Su et al., 1995), while cAMP binds in syn and cGMP binds in anti for Epac (Das et al., 2009). In the binding pocket of HCN2 and 4 mammalian channels, cAMP binds in the anti and cGMP in the syn, as in the solution.  Cyclic GMP prefers the syn configuration in HCN2 CNBD, which may occur because the rotation allows the exocyclic 2-NH2 of guanine ring to form a hydrogen bond with the side chain hydroxyl group of T592. Interestingly, this threonine is naturally a valine in sea urchin HCN (SpIH) orthologue and the lack of this hydrogen bond is proposed as one of the reasons why cGMP is a partial agonist (Flynn et al., 2007). With our data, we find that the difference in binding configuration does not explain the partial agonistic effect and the syn configuration is not necessarily the disfavoured form. In the example of Epac, syn cAMP is able to promote functional effect whereas anti cGMP does not.  Also, our result from Chapter Three shows that cIMP binds in the anti configuration and is a partial agonist.     165       Table 5.1 Configuration of cAMP and cGMP in solution and in binding pockets of three proteins. The same ligand can take on different configurations depending on the stabilization of the interactions in the binding site.  Cyclic IMP binds in the anti and has lower affinity than cGMP. In addition, we have also tested the binding of 8-Br-cAMP and 8-Br-cGMP, which are constitutively syn analogues. Due to the bulky bromine atom, the 8th position is forced to swing away from the ribose to avoid steric clashes, constricting the ligands in the syn configuration. Both of these ligands bind more strongly, even better than cAMP, to HCN2 C-linker/CNBD (Appendix 8E). This is presumably due to rearranging the electronegativity within the ring, and hence altering the interactions with the local environment. The 8-bromo derivatives of both cAMP and cGMP were previously found as potent agonists in optic nerves and sensory ganglia (Ingram & Williams, 1996). Using these ligands answers the question that syn conformation does not necessarily lead to weakened binding. In summary, specific features, the 2-NH2 and 6-keto group, of cGMP determine the binding configuration of this ligand. Both of these features must be present to achieve the syn configuration. The2-NH2 group alone (as in 2-NH2-cPuMP) does not determine the syn configuration. The intramolecular bond between the 2-NH2 and the cyclic phosphate oxygen of the syn cGMP in solution also does not help determine the syn configuration because the bond distance is 5.2Ǻ. Similarly, the 6-keto alone (as in cIMP) is not the sole determinant for the binding configuration of cGMP.   cAMP cGMP Free Anti Syn PKA-bound Syn Syn Epac-bound Syn Anti HCN-bound Anti Syn 166  5.3.2 Interactions between cyclic nucleotide and HCN2 binding domain are identified by comparing results from alanine substitutions and analogues 5.3.2.1 The PBC contributes to the strength of binding as well as potency The phosphate binding cassette provides interactions which are important for binding affinity and potency. Arginine 591 interacts with the equatorial oxygen of the phosphate group and threonine 592 interacts with the axial oxygen. Substitutions of these residues by alanine reduce the binding affinity for both cAMP and cGMP (Chapter Three). Binding experiments were also performed using Rp- and Sp- cyclic nucleotide analogues, in which a sulphur atom replaces the equatorial and axial oxygen of the phosphate group, respectively (Appendix 8AB). The binding affinity to wild type HCN2 C-linker/CNBD is reduced for both analogues, which is consistent with the importance of the phosphate group in promoting binding. We tested the functional effect of Rp- and Sp- cAMP analogues and both ligands still produce a maximal shift on the activation curve comparable to that of cAMP (Appendix 8AB). Consistent with mutational analysis, the PBC is involved in binding but not in coupling. The equatorial oxygen in the phosphate moiety is substituted with a sulphur atom in the Rp- analogue. We tested the effect of the substitution on both cAMP (Rp-cAMP) and cGMP (Rp-cGMP). Rp-cAMP binds with no apparent negative cooperativity and with a low affinity of ~60 µM and Rp-cGMP also binds weakly with 46 µM (Appendix 8A). The consequence of the sulphur-substitution is more severe than the R591A mutation, the PBC residue that interacts with that particular oxygen. This is because the analogue does not fully mimic the effect of PBC mutation. The equatorial oxygen atom makes more contacts with residues nearby, namely residues 581-584, and the O→S substitution may affect the local environment of the binding pocket. In addition, the sulphur-containing ligands are not perfect models because the sulphur 167  atom is less electronegative than oxygen atom, and thus it may cause secondary effect on partial charges of the ligand. The sulphur-substitution of the axial oxygen in cAMP (Sp-cAMP) is able to retain negative cooperativity and binds to CNBD with 0.53 µM and 14.22 µM for the two binding events. The binding values show that the effect of the substitution and T592A mutation are similar (Appendix 8B). In contrary, Sp-cGMP has a more severe effect on binding affinity to wild type C-linker/CNBD. The axial sulphur in Sp-cGMP may interfere the ligand from binding in the syn configuration.  5.3.2.2 Re-positioning of R635 assists cUMP and cIMP binding Like cIMP, the pyrimidines cUMP and cTMP also have a keto group in a location near residue R632 (position 6 in cIMP and position 4 in cUMP or cTMP). Cyclic TMP does not appear to bind to the isolated C-linker/CNBD as tested by ITC (Appendix 8D) whereas cUMP binds with negative cooperativity. Cyclic TMP does not bind to the CNBD possibly due to the extra 5-methyl group clashing with the binding pocket. The affinity for cUMP could be compensated by the re-positioning of R635 guanidinium side chain towards its 4-keto group, forming a new hydrogen bond, as shown in the crystal structure (Figure 3.6).  Cyclic pyrimidine nucleotides bind less strongly compared to ligands with two fused rings (purines). Lower affinity could be due to smaller surface area and reduced occupancy in the pocket, creating fewer van der Waals interactions with residues in the proximity. The position of the cyclic phosphate and ribose moiety do not vary among ligands, so a smaller surface area in the pyrimidine ring would create greater distances with the nearby C-helix, forming a looser-fit cap. The C-helix is important for cAMP modulation. The differences in the π-electrons and in 168  electronegativity between the purine and pyrimidine rings are also factors that affect the interaction with the C-helix.  It was proposed that cCMP has low affinity due to a difference in the delocalized π-electrons in the base with hydrophobic residues in the pocket (Zong et al., 2012). However, the movement of R635 towards cUMP could provide a tighter pocket for the ligand, which is absent in cCMP. We propose that this movement is similar to the induced-fit model in enzyme catalysis to favour substrate binding. The proposal is supported by a drastic reduction in binding affinity of cUMP when R635 is mutated to alanine. Cyclic UMP binds to the mutated protein with Kd of ~200µM, around 10 times lower than cUMP in wild type CNBD. The same R635A mutation did not affect the binding affinity nearly as much for cAMP. The Kd is reduced by two-fold but remains in the sub-micromolar range. The bigger adenine ring does not require additional stabilization, and hence the crystal structure shows R635 facing farther from the ligand at 5.1Ǻ.  Functional data suggests that the residue R635 in the C-helix has a small contribution to the affinity and potency of cAMP and cGMP (L. Zhou & Siegelbaum, 2007) and a more severe influence on the binding of cUMP (Chapter Three). However, this residue is not required for achieving the ligand-induced facilitation of channel opening.   5.3.2.3 Interaction with carbonyl oxygen of R632 has a small contribution to affinity The main chain carbonyl group of R632 makes a hydrogen bond with the 6-exocyclic NH2 of adenine of cAMP. Ligands with a 6-keto in the purine ring instead (for example, cIMP in the anti configuration) lack that particular hydrogen bond and the binding affinity to HCN2 C-linker/CNBD is reduced. Reduced affinity may be due to the electrostatic repulsion of the partial negative dipoles produced by the two carbonyl groups. Ligands without any exocyclic 169  component in the 6-position of the purine ring (cPuMP and 2-NH2-cPuMP) have only modest reduction in affinity to the CNBD compared to cAMP. The combined findings suggest that the bond between the R632 carbonyl and the 6-position improves affinity and the repulsion of the two groups would reduce affinity. To further investigate on the electrostatic repulsion with R632 carbonyl, we used another analogue, 6-Cl-cPuMP, where the 6-NH2 group of cAMP is replaced with an electronegative chlorine atom. The analogue also has no hydrogen-bond capability with the nearby R632 carbonyl. The binding affinity of 6-Cl-cPuMP to HCN2 C-linker/CNBD is 0.4 and 8µM, for high and low affinity event, respectively (Appendix 8F). The ligand binds ~4-fold weaker compared to cAMP but binds more strongly than cIMP. The phenomenon could be due to the difference in electronegativity between chlorine (3.16) and oxygen (3.44). The more electronegative oxygen has a stronger negative partial charge, causing a greater repulsion with R632 in the CNBD.   5.3.3 A CNBD salt bridge reduces stability of the HCN2 C-terminal domain Side chains of residue E582 in the β-roll of the CNBD and R632 from the C-helix form a salt bridge within the binding pocket (Zagotta et al., 2003; L. Zhou & Siegelbaum, 2007). Alanine substitution of R632 abolishes the ligand-induced facilitation of HCN2 channel opening whereas alanine substitution of E582 reduces cAMP potency significantly but still allows for full facilitation. Since the E582A mutation is still responsive to cAMP, the salt bridge is not absolutely essential for cAMP modulation. We tried to purify the isolated C-linker/CNBD with either the E582A or R632A mutation to test direct binding. However, the yield for both mutants was very low and insufficient to perform binding experiments. We propose that the salt bridge is important for stabilizing the folding of the CNBD in its tertiary structure.  170  To understand how R632 and E582 influence binding, we purified the mutated C-linker/CNBD with HMT tag on its N-terminus. Due to the internal favourable folding of the maltose binding protein in the tag, the fusion protein can be obtained. It was previously found that HMT-fused C-linker/CNBD of HCN2 can reproduce comparable binding affinity as untagged C-linker/CNBD, showing that HMT does not interfere with the binding site (Chow et al., 2012). Cyclic AMP binds to E582A-HMT protein with a Kd of 308µM whereas no observable binding is detected to R632A-HMT protein. We conclude that the internal salt bridge is responsible not only for keeping the CNBD structure stable but for maintaining high affinity for ligand (Appendix 4).  5.4 Gating model   Our data can be summarized with a modified modular model, initially proposed by Craven & Zagotta (Craven & Zagotta, 2004). The modular model assumes the CNBD, C-linker, and the pore act as individual modules (three individual boxes in Figure 5.1), which are also connected via conformational transitions. The modular model arises when it was discovered that the crystal structure, bound to cAMP or cGMP, contains salt bridges in the C-linker that only exist in the resting state. When those salt bridges were broken by mutations, the active state of the C-linker was favoured and the activation was promoted to the same extent as applying cAMP. Hence, it was possible for bound-state CNBD and resting-state C-linker to coexist in the crystal structure and for these two modules to act independently (Craven & Zagotta, 2004).  The individual modules include the CNBD, which exists in an equilibrium of bound and unbound state (Figure 5.1A); the C-linker, which exists in an equilibrium of resting and activated state (Figure 5.1C); and the pore, which exists in an equilibrium of open and closed 171  state (Figure 5.1E). A transition event connects and couples the modules and likely involves some conformational rearrangements for delivering the signal from one module to the next. Thus, a transition exists from CNBD to C-linker activation (Figure 5.1B), and another from C-linker rearrangement to pore opening (Figure 5.1D). The thesis has discussed the different equilibria or transitions in the model, providing examples to support how each module plays a role in ligand binding to facilitated opening.  5.4.1 Equilibrium between bound and unbound states Cyclic nucleotide modulation starts with binding of the cyclic nucleotide to the CNBD (Figure 5.1A).  The shift between the CNBDunbound to the CNBDbound state is determined by on and off rates and the affinity of the ligand. Individual mutations in Chapter Two draw attention to the binding site and residues that affect the affinity and potency. The mutations, mainly alanine substitutions, were shown to not affect the maximum response of cAMP and cGMP nor did they affect the gating properties in the absence of cNMP. In most cases, the mutations create less favourable and weakened interactions that modify the ability of cNMP to bind, and weaker binding shifts the favourability towards the unbound state. The higher Kd values quantitatively indicate that more ligand is required to achieve the bound state.  The mutations in the CNBD we tested reduce the binding affinity for cAMP, but they do not affect the downstream modules in the gating model. These mutations, including those in Chapter Two and S672R in Chapter Four, have lower affinity but no change on cAMP-induced oligomerization. Oligomerization is a measurement of C-linker activation. Despite the increased favourability towards the unbound state in some mutations, an increase in ligand concentration is 172  sufficient to overcome the reduced affinity, pushes the equilibrium towards the bound state, and trigger the following transition in the C-linker (Appendix 3).   5.4.2 Binding to CNBD promotes activation of the C-linker The binding to the CNBD triggers a conformational change that leads to the activation of the C-linker (Figure 5.1B). The activation involves forming linkages in the shoulder-elbow motif to create tetramers, which would stabilize the formation of a gating ring. Ligand binding leading to tetramer formation of C-linker/CNBD piece has been characterized previously by analytical ultracentrifugation (Zagotta et al., 2003). We suspect that the binding of partial agonists in Chapter Three promotes activation of the C-linker to a lesser extent than a full agonist. Reduced binding affinity alone does not explain the observed defect because adding excess partial agonists to both DLS and patch clamping experiments does not restore phenotype of a full agonist. The gating ring is hampered, leading to less tetrameric C-linker/CNBD in vitro and a smaller functional effect on full length channel. The smaller effect could be because the gating ring is less stable, or because the C-linker undergoes a different conformational change that lead to a different (weaker) gating ring.  To understand the mechanism behind partial agonists, we compared the crystal structures of two pyrimidine cyclic nucleotides, cCMP the partial agonist and cUMP the full agonist. We observed that cCMP makes closer contacts with the β-roll and at the same time, less with the C-helix. This is supported by a difference in distance distribution of the distal C-helix from DEER experiment between cAMP and cCMP (DeBerg et al., 2015). The correlation curve from Chapter Two also pointed out mutated residues that affect the potency disproportionally to affinity and are involved in a conformational change. These residues include R632, L633, and I636 which 173  reside along the C-helix and interact with the nitrogenous base. Together, the data shows that the C-helix movement is critical for facilitation, which fits with what was found already by FRET, DEER, and NMR experiments (Akimoto et al., 2014; Puljung et al., 2014; Saponaro et al., 2014; Taraska et al., 2009). The coupling between CNBD and C-linker involves the extension of the C-helix and the movement of the B- and C-helix to form the cap towards the core of the β-roll. The movement of the helices relieves the steric hindrance from the N3A motif (Akimoto et al., 2014), allowing it to go from the N3A-in to N3A-out conformation and favouring tetramer formation. The partial agonist cCMP, however, results in a smaller conformational rearrangement in the N3A motif, where the NMR chemical shifts of residues in the motif lie between that of the apo and the cAMP-bound state (Akimoto et al., 2014). The data together conclude that cCMP traps the C-linker in a handicapped state that does not allow proper activation of the tetrameric domain.  5.4.3 Equilibrium between active and resting C-linker The gating ring forms and unforms dynamically at a steady rate (Figure 5.1C), and this could be influenced by the equilibrium between bound and unbound states in the previous module. We propose that the gating ring is a product of tetramerization in the C-linker/CNBD segment, explaining why cAMP increases tetrameric population of the C-terminus region in vitro and shifts voltage dependence in vivo. We found that the epilepsy-associated mutation E488K limits the tetrameric gating ring formation, leading to shifting the equilibrium towards the resting C-linker. This mutation disrupts an intersubunit interaction needed to hold the subunits in a stable state. Reducing stable gating ring and favouring the resting state of the C-linker lead to the negative shift in the basal activation curve. The equilibrium is changed by an intrinsic property 174  of the channel, and so the addition of cAMP does not restore and undo the intrinsic defect. Cyclic AMP can still exhibit the same magnitude of functional shift as in wild type channel.  The equilibrium shifts following the previous transition (Figure 5.1B). In the bound state, the capping movement of the distal CNBD causes the N3A motif to flip outwards, thereby freeing the inhibition on the shoulder-elbow motif (Akimoto et al., 2014). Exposing the tetramer-forming interface promotes association between subunits and the equilibrium is shifted to more active C-linker.  5.4.4 A C-linker gating ring promotes the pore opening Another transition event occurs to couple C-linker activation and pore opening (Figure 5.1E). Since the C-linker is directly connected to the S6 segment, it is likely the signal is transduced near the junction. This transition is not well studied because the protein fragment for structural analyses using X-ray crystallography, NMR, or DEER are truncated and the fragment does not contain the S6 segment. It was proposed that cAMP binding causes a downward movement of the C-linker, relieving the intrinsic inhibition caused by the C-terminus (Akimoto et al., 2014). The transition could also involve the breaking of the salt bridges K472 with intersubunit E502 and intrasubunit D542. By breaking these bonds, the channels are fully activated even in absence of ligand. The breaking of these bonds is equivalent to binding to cAMP and relieving the stress brought upon by the C-linker (Craven & Zagotta, 2004). What exactly happens during the transition cannot be confirmed without isolating and visualizing the open state of the pore. The bradycardia-associated mutation may disrupt the relieving of tonic inhibition of the CNBD and limit the opening of the pore. This mutation shifts the basal activation dependence to 175  more negative, but our data shows that the binding in the CNBD and the activation of the C-linker are mostly unaffected. Instead, the S672R mutation affects an event that comes after the C-linker activation. A possible interaction underlying the gating mechanism could involve contacts between the mutated site and the pore or cytosolic linker regions of the transmembrane segments. The voltage sensor can play a role to influence the dynamic events. Each equilibrium shown (Figure 5.1A, C, E) can exist in the resting closed, active closed, resting open, and active open states (S. Chen et al., 2007) depending on voltage. The state of the pore also has an influence on ligand binding because it was found that cAMP is more likely to bind to the open channel tested by patch-clamp fluorometry (S. Wu et al., 2011).   176   Figure 5.1 The proposed gating model during cAMP modulation. The model is adopted from the modular model by Craven et al, separating the dynamic events in the CNBD, C-linker, and the pore. The initial binding (unbound to bound CNBD) depends on residues for selectivity and high affinity. The transition involves the elongation of the distal C-helix and swinging towards the ligand binding site. This new position allows N3A to swing out to form tetrameric contacts (inactive to active CL). To facilitate channel opening, the last transition carries the signal to the pore, where the C-linker will undergo a last transformation to relieve its tonic inhibition. The different states and transitions are supported by our data, and the model adopts results from previous studies (Akimoto et al., 2014; Craven & Zagotta, 2004; Saponaro et al., 2014; Zagotta et al., 2003).  5.5 Physiological relevance of the data in this thesis 5.5.1 Cellular level of cAMP and cGMP The cellular level of cAMP in pacemaker myocytes is believed to be in the sub-micromolar range based on indirect electrophysiological evidence. It was established that the 177  EC50 of cAMP, the concentration at which 50% of the channels in the cell are open, in rabbit SAN cells was 0.2 µM. Under hyperpolarizing conditions these measurements were taken, the contribution from other channels in the cell was minimized. In another set of experiment, acetylcholine or isoprenaline was introduced to SAN cells, causing a shift of dependence to the left or the right, respectively, from the activation curve generated under basal, cellular level of cAMP. Acetylcholine would deplete the supply of cAMP and isoprenaline would stimulate the release of cAMP to account for the corresponding changes in voltage dependence. The fact that the leftward and rightward shifts by the two stimulations were around the same in magnitude suggests that at basal level, around 50% of channels are already open without external factors. Combining the two pieces of information, the basal level of cAMP is approximately 0. 2 µM. FRET-based cAMP/cGMP sensors are being developed for detecting the localized concentrations of ligands (Sprenger & Nikolaev, 2013). The strong potency and affinity of both ligands for HCN channel makes them physiologically relevant. Knowing the high affinity of cAMP on HCN2 and HCN4 isoforms, we learn that the channels are very sensitive with binding to the first site. HCN channels exert great influence on the resting potential and input resistance, especially during resting potential when many channels are closed. Even a small change in ligand modulation could have a significant effect.  5.5.2 Fluctuation in cAMP/cGMP levels The cardiac HCN channels have higher preference for cAMP over cGMP. Thus, during cAMP surge from β-adrenergic stimulation, it is most likely cAMP that acts as the modulator. Only when cAMP level is low, both ligands could be candidate modulators. In fact, HCN 178  channels in SAN are responsive to nitric oxide (NO), which shifts the activation curve to the right, acting via the NO-cGMP pathway (Barbuti et al., 2004; Herring, Rigg, Terrar, & Paterson, 2001).  The small molecule was discovered to act as a signalling agent in the cardiovascular system, which led to a Nobel Prize in Physiology/Medicine in 1998. In the pathway, nitric oxide stimulates guanylyl cyclase activity near the channels. Local guanylyl cyclase near the channels could produce up to 100 µM of cGMP per second evoked by NO (Mo, Amin, Bianco, & Garthwaite, 2004).  Moreover, the unconventional ligand cGMP could be in contention during increase in endogenous ACh from vagal stimulation, causing a reduced level of its competitor, cAMP. ACh release can be promoted by black widow spider venom, and the symptoms are more severe in children or the elderly. Excess release was also correlated with sudden cardiac death in elderly and diabetic patients (Oberhauser, Schwertfeger, Rutz, Beyersdorf, & Rump, 2001). Besides acetylcholine, reduction of intracellular cAMP can also be caused by atrial natriuretic peptides (ANP), which significantly inhibit adenylyl cyclase (Anand-Srivastava & Cantin, 1986). These peptide hormones are linked to guanylyl cyclase receptors, and thus, ANP was found to increase cGMP while decreasing cAMP levels in rats (Anand-Srivastava, Sairam, & Cantin, 1990). Despite being a less potent ligand, cGMP could play a role in these circumstances when cAMP level diminishes. The cGMP-pathway could be a security measure to ensure pacemaking activity in healthy individuals. The less-efficient cGMP is upregulated to maintain basal activation of HCN channels and basal cardiac activity. It has been shown in protein kinases that the cyclic nucleotides can “cross-talk” where cGMP binds to PKA or cAMP binds to PKG, resulting in less activation of both kinases due to a reduced potency from the less-favoured ligand (Cornwell, Arnold, Boerth, & Lincoln, 1994; 179  Jiang, Colbran, Francis, & Corbin, 1992). Similarly, cGMP could compete for the binding site of HCN channels, and lowers the opening probability at low ligand concentration. Cross-talk proposes a new realm of channel regulation. However, naturally, cross-talk is prevented via selective control of cAMP- or cGMP- dependent signalling pathways, which includes the synthesis of cNMP, the degradation by phosphodiesterases, and intracellular compartmentalization of the components in respective pathways (Das et al., 2009; Pelligrino & Wang, 1998).  5.5.3 Physiological relevance of other cyclic nucleotides The pyrimidine cyclic nucleotides, cCMP, cUMP, cTMP, as well as the purine cyclic nucleotide cIMP, have been extracted from rat liver, heart, spleen and lung tissues by co-chromatography and analyzed by mass spectroscopy (Newton et al., 1986). Cyclic CMP and cUMP were found in mammalian cell lines in micromolar range, including in human urine sample, (Burhenne, 2011) using the highly sensitive method of HPLC-MS/MS. In some cell lines, the detected levels of cCMP and cUMP were comparable or higher than that of cGMP, implying a regulatory role for these pyrimidine cyclic nucleotides (Burhenne, 2011).  The broad specificity soluble guanylyl cyclase, almost ubiquitous in all cells, allows the conversion of the starting material (CTP or UTP) into respective cyclic nucleotides (Beste, Burhenne, Kaever, Stasch, & Seifert, 2012; Beste & Seifert, 2013). Thus, it is a growing research interest since these molecules have capability of being potential secondary messengers (Beste & Seifert, 2013; Schneider & Seifert, 2015). Cyclic CMP can bind to PKA and HCN channels, but is less efficacious and only partially activate the proteins (Wolter, Golombek, & Seifert, 2011; Zong et al., 2012). From our binding studies, it is unlikely that cCMP would outcompete with the 180  natural agonists given its weak binding. Cyclic UMP, however, has strong binding for the first high-affinity site, which creates very weak subsequent binding sites. Even though still agonistic, if cUMP binds to the first site, it is more difficult for other cyclic nucleotides to further bind to achieve full occupancy. It is also to note that it is possible for cCMP to convert into cUMP spontaneously via deamination. Such a spontaneous reaction occurs naturally to DNA and RNA, leading to complementary mismatch. In the case of HCN2 channels, the conversion allows a partial agonist to have full functional effects.  5.6 Limitations to the methods 5.6.1 Advantages and disadvantages of using the soluble C-linker/CNBD In this thesis, almost all the work was performed using the carboxy-terminal segment containing the C-linker and CNBD, starting immediately after the S6 transmembrane helix and ending after the CNBD. The advantage of using the isolated fragment is that it is soluble and easy to purify, and the binding and effect of ligand is narrowed down to sites available in this region. Furthermore, the values for cyclic nucleotide binding which we obtained by ITC are similar to values of EC50 (potency) obtained using patch clamp electrophysiology on the full-length channel. When combined with electrophysiological assays in the full-length channel, we have been able to garner a significant amount of information.  The disadvantage of the approach is that the C-linker/CNBD is studied in the absence of the rest of the channel and correlated with electrophysiological evidence from the full-length channel. An interaction of a gating ring with the pore has been proposed but direct evidence for cAMP-induced gating ring formation and its interaction with the pore have not been observed. For example, the C-linker was proposed to interact with the S4-S5 linker in both CNG (Kusch, 181  Zimmer, et al., 2010) and HCN channels (Decher et al., 2004), impacting the voltage sensing and activation of the channels (J. Chen et al., 2001; Kwan et al., 2012; Macri & Accili, 2004).  C-terminal portion distal to the CNBD could influence the effect of cAMP. The HCN2 C-linker/CNBD used in our experiments ends at residue 645 because the expression of various lengths was previously examined and this length yielded stable products (Zagotta 2003). Otherwise, elongating the C-terminus could introduce flexible regions that make folding difficult during the purification process. In CHO cells and for the HCN4 isoform specifically, it was found that the distal C-terminus is involved in the pre-relieving of tonic inhibition, thus shifting the voltage dependence to more depolarized potential and mimicking cAMP-induced facilitation  (Liao et al., 2012). In addition, the distal region is also found to be where the accessory protein TRIP8b binds to the channel, which would reduce the channel dependence on cyclic nucleotide upon binding (DeBerg et al., 2015; Saponaro et al., 2014). We have tested the effect of the distal region using the isoform from Drosophila melanogaster. Unlike the mammalian forms, the distal region of the CNBD is shorter and the longer version is also stable upon purification. We have created both versions of C-linker/CNBD, where they both start with the C-linker, but one truncates after the CNBD while the other to the very end of the wild type channel (Appendix 2C). Both constructs likely fold normally because they bind to cyclic nucleotides. We found that there is a small reduction with ligand binding when the C-terminus is longer. The Kd is 2.86 µM for the longer version (Appendix 10B), and 0.96 µM for the truncated version (Appendix 10A). It could be added steric effect that interferes with cyclic nucleotide entrance in the longer version. The N-terminal region in the cytosol is absent in the C-linker/CNBD. Among the mammalian isoforms, HCN4 has the longest stretch of conserved sequence preceding the 182  transmembrane segments. Different variants or mutations in the HCN4 N-terminus causes distinct voltage dependence and kinetics of channel activation, tested by patch-clamp experiments (Liu & Aldrich, 2011). The functional difference is believed to be caused by positively charged residues on the N-terminus, which can interact with the C-terminus or the transmembrane domain via electrostatic interactions. It was also found to change channel kinetics and trafficking (Ishii, Nakashima, Takatsuka, & Ohmori, 2007; Ludwig, Zong, Hofmann, & Biel, 1999). The N-terminus may modulate channel properties by interacting with the C-terminus or with other associating proteins (Allen, Fakler, Maylie, & Adelman, 2007; Liu & Aldrich, 2011; Mohapatra et al., 2009). The fragmented C-linker/CNBD denies any interaction with or between the pore, N-terminal, and distal post-CNBD regions.   5.6.2 X-ray crystallography does not capture the active C-linker Crystallography has also proven to be limited in understanding the changes induced by cAMP binding because the C-linker remains in the same conformation regardless of the mutations, analogues, or isoforms attempted  (Craven et al., 2008; Lolicato et al., 2011; Möller et al., 2014; Xu et al., 2010; Zagotta et al., 2003). It may be that the energetic cost to maintain the active C-linker state is high and unfavourable, so only the inactive form is crystallized.   One solution is to use NMR, which allows the studying of protein dynamics during conformational changes during ligand binding or C-linker activation. Using NMR has been useful for identifying changes in structure that are induced by cAMP binding (Akimoto et al., 2014; DeBerg et al., 2015; Saponaro et al., 2014). The limitation of NMR is size and, to date, a full-length channel or even a tetrameric C-terminal subunit has not been reported.  183  5.6.3 Biochemical and cellular experiments in the thesis do not fully mimic the cellular environment of heart cell or neuron The patch-clamp experiments in CHO cells were useful in determining the shift induced by various ligands and distinguishing full agonists from partial agonists. However, the cells we used were not native cardiomyocytes. In the heart, HCN channels may be modulated by other proteins or molecules beside cyclic nucleotide. The native channels can also form heteromers composing of different isoforms. The biochemical experiments including ITC or DLS are useful because the protein we used is pure. At the same time, the high purity does not mimic the cellular environment, which contains other interacting or interfering proteins and contaminants. The HEPES buffer we used for protein purification mimics the physiological pH but not the electrolyte content. Finally, the experiments were performed at 22-25 oC, not the physiological temperature.  In both type of experiments, the results have neglected the presence of other signalling molecules or modification that can influence gating or ligand binding. For example, PIP2 is found to be a regulator of HCN channels and TRIP8b acts an allosteric inhibitor to cAMP binding. Post-translational modifications such as phosphorylation and glycosylation are also factors in changing HCN currents but are not tested in our experiments. Even in the C-terminal segment alone, there are 3 phosphorylation sites found in human. The lack of signalling molecules and modifications may not reflect the realistic affinity between the ligands and in vivo channels.  184  5.7 Future directions 5.7.1 Structural approach for studying the full-length construct   Much of the data and conclusions in my thesis came from experiments were done on the isolated C-terminus.  A useful next step for this work is to extend some of the findings to the full-length channel. It would be advantageous to test the affinity of cyclic nucleotide with the rest of the channel attached to the C-linker/CNBD piece to see whether the missing components – the N-terminus, the transmembrane domain, and distal C-terminus – play a part in binding affinity. The channel has to be first purified in a detergent-containing buffer to solubilize the transmembrane section, and use similar techniques for testing the binding affinity and comparing with the isolated fragment. If the values are similar, it would validate the testing protocol on the C-terminus piece which is an easier system to work with. The challenge in using the full length channel is its biphasic property and the difficulty in keeping both phases in solution. Also, the detergent in the buffer would affect the heat readings from the ITC (Heerklotz, Binder, Lantzsch, Klose, & Blume, 1997; Thapa et al., 2013). Another goal is to capture the full-length channel at different states using X-ray crystallography. The comparison between the open state and closed state of the channel would explain the structural dynamics during the gating process. Also, cryo-electron microscopy is emerging as a transformative approach with atomic resolution for full-length channels captured in multiple conformations (Binshtein, 2015; Saibil, 2000). The changes in the CNBD, the C-linker, and the pore can be accounted for comparing the apo and holo structures. The mechanism of action of how disease mutations lead to their phenotypes can also be better illustrated when the channel is investigated as a whole.  185  5.7.1.1 Resolving the two gating ring theories In the thesis, our data supports the theory where the subunits must associate when the gating ring is formed. We found that the subunit interactions are important, so disrupting the intersubunit linkages must lead to a demoting gating ring formation, causing negative consequences on gating properties. Partial agonists from Chapter Two causes limited oligomerization, resulting in a smaller shift in the activation curve. Another example is the epilepsy-associated mutation tested in Chapter Four. The electrophysiological data of E488K mutation showed a large negative shift in the activation curve and slower activation kinetics (J. C. DiFrancesco et al., 2011). The ligand-induced tetramerization of C-linker/CNBD was abolished despite the increased ratio of ligand. From our solved crystal structure, we concluded that the observation was due to the disappearance of an intersubunit interaction between the side chains of E488 and Y459. The reverse mutation Y459A showed the same biochemical properties in vitro, confirming that this intersubunit hydrogen bond helps hold the monomeric C-linker/CNBD together. Finally, the reversal double mutant Y459E/E488K also showed the same defected trend on the binding and oligomerization state as the single mutations (Appendix 7E). This may suggest that the double mutant does not restore the hydrogen bond because the new local geometry disallows bond formation. We have not obtained functional data for Y459A and double mutations, but we would expect they behave similar to E488K. To build upon the importance of forming intersubunit interactions, we have found that another epilepsy-associated mutation in the HCN2 C-linker (R500Q, Tang, 2008) also disrupts tetramerization (Appendix 7F). Functional studies have suggested that this mutation does not impact HCN2 gating in the absence of cAMP, and that the maximum effect of cAMP is also not altered. We created the same mutation in HCN2 C-linker/CNBD to test for the effect on binding 186  and oligomerization due to cAMP. Surprisingly, our preliminary data show that cAMP binds to the mutant C-linker/CNBD without negative cooperativity and only limitedly promoted oligomerization of this protein. The effect of the R500Q mutation is similar to the effect of the E488K epilepsy-associated mutation. However, the crystal structure does not show a direct interaction between R500Q and residues of the neighbouring subunit. The mutation could be affecting the local tertiary folding or the positively charged arginine could play an important role in tetramer formation. We will test the electrophysiological data again. We have also found another intersubunit hydrogen bond between the hydroxyl of tyrosine 477 on C’ helix and the backbone carbonyl of isoleucine 491 on B’ helix, which is found in many of the published HCN crystal structures, does not appear to be important for tetramer formation. We made the Y477F mutation to eliminate hydrogen bond formation. The mutation showed preliminary results similar to wild type in terms of binding and ability to oligomerize (Appendix 7G), suggesting that the particular intersubunit hydrogen bond is not required for gating ring formation. Thus, not all intersubunit bonds are necessary to stabilize tetramer formation. We will also correlate this finding with electrophysiological data. The opposing argument proposes that the formation of a gating ring involves breaking intersubunit bonds upon binding to cAMP in order to facilitate opening. Intrasubunit and intersubunit salt bridges, found in the crystal structure, were previously eliminated by mutations and were found to cause a depolarizing shift to the activation curve in the absence of cAMP (Craven & Zagotta, 2004). It was argued that the salt-bridge interactions between K472, E502, and D542 are broken to increase favorability of channel opening. We created the same mutations to disrupt the formation of these salt bridges in the soluble C-linker/CNBD and obtained preliminary binding and oligomerization data. Our DLS data shows that K472E and D542K 187  mutations cause a reduction in ligand-induced oligomerization (Appendix 7ABCD). Our data confirms that the mutations disrupt the association of subunits in the presence of cAMP. Together with the functional data, it appears that channel opening can be facilitated without tetramerization and that facilitation was achieved via a different mechanism. To conclusively resolve the debate between the two theories, we must explain the dynamics of gating ring during channel modulation by visualizing the gating ring in its inactivated and activated states.  5.7.2 Fitting our data in a mathematical model Currently, there is no model that accurately relates the binding affinity and negative cooperativity obtained by our calorimetry approach to measured values of potency. Existing allosteric models of the HCN2 channel are able to explain most of the current functional data. These models suggest that binding to the closed state is weaker than binding to the open state, which explains why the open state is favoured by bound ligand. However, the theoretical values for binding affinity to the open and closed state generated by these models, which jive with the experimentally determined values of potency, are much higher or lower, respectively, than the binding values obtained by our approach.   Our high affinity binding data is almost identical to the values of potency determined by experiment, suggesting that a different model is required to explain gating. We are collaborating with Dr. Yue Xian Li in the mathematics department at UBC to explore such a model and apply it to the data obtained in Chapter Two.   188  5.7.3 Screening drugs using dynamic light scattering  In the thesis, we proposed that the ability for soluble C-linker/CNBD to oligomerize in the presence of ligand is correlated with gating ring formation and agonism. In Chapter Three, we found that partial agonists did not promote oligomerization of C-linker/CNBD the way full agonists could. We have shown the degree of oligomerization at 100 µM C-linker/CNBD is proportional to the depolarizing shift on the activation curve using maximal ligand concentration. Thus, DLS can be a quick method for screening for potential agonists and antagonists. A disadvantage of the screen is that a lack of tetramerization could be due to the ligand not binding at all, and thus, a binding assay is still required. The same method could be translated to other systems if ligand-induced oligomerization is also linked to channel facilitation.  5.7.4 Identifying isoform-specific agonists and antagonists for therapy and correcting dysfunctional mutant channels   To search for therapeutics, it is important to understand how the mechanism of cAMP-induced facilitation differs among isoforms. Isoform specificity is important for therapeutics to improve on the current Ih-specific blocker, ivabradine. Although effective in controlling the heart rate (predominantly HCN2 and HCN4), it also blocks HCN1 channels in the retina, causing visual disturbance. Isoform-specific agonists can also remedy the effect of mutant channels. For example, to counteract the hyperpolarized shift caused by the S672R mutant channel, HCN4-specific agonist can be added to facilitate HCN4 channel opening without affecting the other isoforms.  A challenge is finding analogs which are selective for specific HCN isoforms. Because HCN2 and HCN4 respond most strongly to cAMP, we have carried out initial trials using HCN2 189  partial agonists discovered in Chapter Three on the HCN4 C-terminus.  Surprisingly, unlike HCN2 C-terminus, the partial agonists actually promoted enhanced tetramerization and ITC data showed negative cooperativity. According to our theory, this finding may suggest that these HCN2 partial agonists are able to allow gating ring formation, which means they may be full agonists in the HCN4 isoform. To test isoform specificity, functional data must be obtained. This could mean these ligands selectively affect HCN2 isoform, or that the gating ring equilibrium is set at different positions. Thus, ligand-induced gating is easier to achieve, which is consistent with HCN4 isoform being the most sensitive to cAMP. The search must continue from here. 190  Bibliography  Accili, E. A., Proenza, C., Baruscotti, M., & DiFrancesco, D. (2002). From funny current to HCN channels: 20 years of excitation. News Physiol Sci, 17, 32-37.  Accili, E. A., Redaelli, G., & DiFrancesco, D. (1997). Differential control of the hyperpolarization-activated current (i(f)) by cAMP gating and phosphatase inhibition in rabbit sino-atrial node myocytes. J Physiol, 500 ( Pt 3), 643-651.  Adams, P. D., Afonine, P. V., Bunkoczi, G., Chen, V. B., Davis, I. W., Echols, N., . . . Zwart, P. H. (2010). PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr, 66(Pt 2), 213-221.  Akimoto, M., Zhang, Z., Boulton, S., Selvaratnam, R., VanSchouwen, B., Gloyd, M., . . . Melacini, G. (2014). A mechanism for the auto-inhibition of hyperpolarization-activated cyclic nucleotide-gated (HCN) channel opening and its relief by cAMP. J Biol Chem, 289(32), 22205-22220.  Alderete, J. (2000). The Syn-Anti Equilibrium of Guanosine Cyclic 3',5'- Monophosphate and 8-Sustituted Derivatives: A Theoretical Study. Bol Soc Chil Quim, 45, 355-362.  Alig, J., Marger, L., Mesirca, P., Ehmke, H., Mangoni, M. E., & Isbrandt, D. (2009). Control of heart rate by cAMP sensitivity of HCN channels. Proc Natl Acad Sci U S A, 106(29), 12189-12194.  Allen, D., Fakler, B., Maylie, J., & Adelman, J. P. (2007). Organization and regulation of small conductance Ca2+-activated K+ channel multiprotein complexes. J Neurosci., 27(9), 2369-2376.  Altenhofen, W., Ludwig, J., Eismann, E., Kraus, W., Bönigk, W., & Kaupp, U. B. (1991). Control of ligand specificity in cyclic nucleotide-gated channels from rod photoreceptors and olfactory epithelium. Proc Natl Acad Sci USA, 88(21), 9868-9872.  Altieri, S. L., Clayton, G. M., Silverman, W. R., Olivares, A. O., De la Cruz, E. M., Thomas, L. R., & Morais-Cabral, J. H. (2008). Structural and energetic analysis of activation by a cyclic nucleotide binding domain. J Mol Biol., 381(3), 655-669.  Altomare, C., Terragni, B., Brioschi, C., Milanesi, R., Pagliuca, C., Viscomi, C., . . . DiFrancesco, D. (2003). Heteromeric HCN1-HCN4 channels: a comparison with native pacemaker channels from the rabbit sinoatrial node. J Physiol, 549(Pt 2), 347-359.  Anand-Srivastava, M. B., & Cantin, M. (1986). Atrial natriuretic factor receptors are negatively coupled to adenylate cyclase in cultured atrial and ventricular cardiocytes. Biochem Biophys Res Commun., 138(1), 427-436.  Anand-Srivastava, M. B., Sairam, M. R., & Cantin, M. (1990). Ring-deleted analogs of atrial natriuretic factor inhibit adenylate cyclase/cAMP system. Possible coupling of clearance atrial natriuretic factor receptors to adenylate cyclase/cAMP signal transduction system. J Biol Chem., 265(15), 8566-8572.  191  Aponte, Y., Lien, C. C., Reisinger, E., & Jonas, P. (2006). Hyperpolarization-activated cation channels in fast-spiking interneurons of rat hippocampus. J Physiol, 574(Pt 1), 229-243.  Arinsburg, S. S., Cohen, I. S., & Yu, H. G. (2006). Constitutively active Src tyrosine kinase changes gating of HCN4 channels through direct binding to the channel proteins. J Cardiovasc Pharmacol, 47, 578-586.  Barbuti, A., Gravante, B., Riolfo, M., Milanesi, R., Terragni, B., & DiFrancesco, D. (2004). Localization of pacemaker channels in lipid rafts regulates channel kinetics. Circ Res, 94(10), 1325-1331.  Baruscotti, M., Bucchi, A., & DiFrancesco, D. (2005). Physiology and pharmacology of the cardiac pacemaker (“funny”) current. . Pharmacol Ther 107, 59-79.  Bell, D. C., Yao, H., Saenger, R. C., Riley, J. H., & Siegelbaum, S. A. (2004). Changes in Local S4 Environment Provide a Voltage-sensing Mechanism for Mammalian Hyperpolarization–activated HCN Channels. J Gen Physiol., 123(1), 5-20.  Bender, R. A., & Baram, T. Z. (2008). Hyperpolarization activated cyclic-nucleotide gated (HCN) channels in developing neuronal networks. Prog Neurobiol, 86(3), 129-140.  Benetos, A., Rudnichi, A., Thomas, F., Safar, M., & Guize, L. (1999). Influence of heart rate on mortality in a French population: role of age, gender, and blood pressure. Hypertension, 33(1), 44-52.  Berman, H. M., Ten Eyck, L. F., Goodsell, D. S., Haste, N. M., Kornev, A., & Taylor, S. S. (2005). The cAMP binding domain: an ancient signaling module. Proc Natl Acad Sci USA, 102(1), 45-50.  Beste, K. Y., Burhenne, H., Kaever, V., Stasch, J. P., & Seifert, R. (2012). Nucleotidyl cyclase activity of soluble guanylyl cyclase α1β1. Biochemistry, 51(1), 192-204.  Beste, K. Y., & Seifert, R. (2013). cCMP, cUMP, cTMP, cIMP and cXMP as possible second messengers: development of a hypothesis based on studies with soluble guanylyl cyclase α(1)β(1). Biol Chem, 394(2), 261-270.  Biel, M. W.-S., C., Michalakis, S., & Zong, X. (2009). Hyperpolarization-activated cation channels: from genes to function. Physiol Rev, 89(3), 847-885.  Biggin, P. C., Roosild, T., & Choe, S. (2000). Potassium channel structure: domain by domain. Curr Opin. Struct. Biol., 10(4), 456-461.  Binshtein, E. O., M. D. (2015). Cryo-electron microscopy and the amazing race to atomic resolution. Biochemistry, 54(20), 3133-3141.  Biskup, C., Kusch, J., Schulz, E., Nache, V., Schwede, F., Lehmann, F., . . . Benndorf, K. (2007). Relating ligand binding to activation gating in CNGA2 channels. Nature, 446(7134), 440-443.  Bogdanov, K. Y., Vinogradova, T. M., & Lakatta, E. G. (2001). Sinoatrial nodal cell ryanodine receptor and Na(+)-Ca(2+) exchanger: molecular partners in pacemaker regulation. Circ Res, 88(12), 1254-1258.  192  Bois, P., Bescond, J., Renaudon, B., & Lenfant, J. (1996). Mode of action of bradycardic agent, S 16257, on ionic currents of rabbit sinoatrial node cells. Br J Pharmacol, 118, 1051-1057.  Botsford, J. L., & Harman, J. G. (1992). Cyclic AMP in prokaryotes. Microbiol Rev, 56(1), 100-122.  Bradley, J., Li, J., Davidson, N., Lester, H. A., & Zinn, K. (1994). Heteromeric olfactory cyclic nucleotide–gated channels: a subunit that confers increased sensitivity to cAMP. Proc Natl Acad Sci USA, 91, 8890-8894.  Brams, M., Kusch, J., Spurny, R., Benndorf, K., & Ulens, C. (2014). Family of prokaryote cyclic nucleotide-modulated ion channels. Proc Natl Acad Sci USA, 111(21), 7855-7860.  Brelidze, T. I., Carlson, A. E., Sankaran, B., & Zagotta, W. N. (2012). Structure of the carboxy-terminal region of a KCNH channel. Nature, 481(7382), 530-533.  Brelidze, T. I., Carlson, A. E., & Zagotta, W. N. (2009). Absence of direct cyclic nucleotide modulation of mEAG1 and hERG1 channels revealed with fluorescence and electrophysiological methods. J Biol Chem.  Brelidze, T. I., Gianulis, E. C., DiMaio, F., Trudeau, M. C., & Zagotta, W. N. (2013). Structure of the C-terminal region of an ERG channel and functional implications. Proc Natl Acad Sci USA, 110(28), 11648-11653.  Brewster, A. L., Bernard, J. A., Gall, C. M., & Baram, T. Z. (2005). Formation of heteromeric hyperpolarization-activated cyclic nucleotide-gated (HCN) channels in the hippocampus is regulated by developmental seizures. . Neurobiol Dis., 19, 200-207.  Broomand, A., Männikkö, R., Larsson, H. P., & Elinder, F. (2003). Molecular movement of the voltage sensor in a K channel. J Gen Physiol, 122(6), 741-748.  Brown, H. F. (1982). Electrophysiology of the sinoatrial node. Physiol. Rev., 62, 505-530.  Brown, H. F., DiFrancesco, D., & Noble, S. J. (1979). How does adrenaline accelerate the heart? Nature, 280(5719), 235-236.  Brown, H. F., Giles, W., & Noble, S. J. (1977). Membrane currents underlying activity in frog sinus venosus. J Physiol., 271, 783-816.  Bucchi, A., Baruscotti, M., & DiFrancesco, D. (2002). Current-dependent block of rabbit sino-atrial node I(f) channels by ivabradine. J Gen Physiol, 120(1), 1-13.  Bucchi, A., Baruscotti, M., Nardini, M., Barbuti, A., Micheloni, S., Bolognesi, M., & DiFrancesco, D. (2013). Identification of the molecular site of ivabradine binding to HCN4 channels. PLoS One, 8(1), e53132.  Bucchi, A., Baruscotti, M., Robinson, R. B., & DiFrancesco, D. (2007). Modulation of rate by autonomic agonists in SAN cells involves changes in diastolic depolarization and the pacemaker current. J Mol Cell Cardiol, 43(1), 39-48.  Bucchi, A., Tognati, A., Milanesi, R., Baruscotti, M., & DiFrancesco, D. (2006). Properties of ivabradine-induced block of HCN1 and HCN4 pacemaker channels. J Physiol, 572(Pt 2), 335-346.  193  Burhenne, H. B., KY.; Spangler, CM.; Voigt, U.; Kaever, V.; Seifert, R. . (2011). Determination of cytidine 3′,5 ′-cyclic monophosphate and uridine 3′,5 ′-cyclic monophosphate in mammalian cell systems and in human urine by high-performance liquid chromatography/mass spectrometry. Naunyn-Schmiedebergs Arch Pharmacol, 383, 33.  Burns, M. E., & Baylor, D. A. (2001). Activation, deactivation, and adaptation in vertebrate photoreceptor cells. Annu . Rev. Neurosci., 24, 779-805.  Cao-Ehlker, X., Zong, X., Hammelmann, V., Gruner, C., Fenske, S., Michalakis, S., . . . Biel, M. (2013). Up-regulation of hyperpolarization-activated cyclic nucleotide-gated channel 3 (HCN3) by specific interaction with K+ channel tetramerization domain-containing protein 3 (KCTD3). J Biol Chem, 288(11), 7580-7589.  Catterall, W. A., Dib-Hajj, S., Meisler, M. H., & Pietrobon, D. (2008). Inherited neuronal ion channelopathies: new windows on complex neurological diseases. J Neurosci, 28, 11768-11777.  Cervetto, L., Demontis, G. C., & Gargini, C. (2007). Cellular mechanisms underlying the pharmacological induction of phosphenes. Br J Pharmacol, 150, 383-390.  Chang, M., Havlik, R. J., Corti, M. C., Chaves, P. H., Fried, L. P., & Guralnik, J. M. (2003). Relation of heart rate at rest and mortality in the Women's Health and Aging Study. Am J Cardiol., 92(11), 1294-1299.  Chen, J., Mitcheson, J. S., Lin, M., & Sanguinetti, M. C. (2000). Functional roles of charged residues in the putative voltage sensor of the HCN2 pacemaker channel. J Biol Chem, 275(46), 36465-36471.  Chen, J., Mitcheson, J. S., Tristani-Firouzi, M., Lin, M., & Sanguinetti, M. C. (2001). The S4-S5 linker couples voltage sensing and activation of pacemaker channels. Proc Natl Acad Sci U S A, 98(20), 11277-11282.  Chen, S., Wang, J., & Siegelbaum, S. A. (2001). Properties of hyperpolarization-activated pacemaker current defined by coassembly of HCN1 and HCN2 subunits and basal modulation by cyclic nucleotide. J Gen Physiol, 117(5), 491-504.  Chen, S., Wang, J., Zhou, L., George, M. S., & Siegelbaum, S. A. (2007). Voltage sensor movement and cAMP binding allosterically regulate an inherently voltage-independent closed-open transition in HCN channels. J Gen Physiol, 129(2), 175-188.  Chen, T. Y., Peng, Y. W., Dhallan, R. S., Ahamed, B., Reed, R. R., & Yau, K. W. (1993). A new subunit of the cyclic nucleotide-gated cation channel in retinal rods. Nature, 362(6422), 764-767.  Chiu, P. L., Pagel, M. D., Evans, J., Chou, H. T., Zeng, X., Gipson, B., . . . Nimigean, C. M. (2007). The structure of the prokaryotic cyclic nucleotide-modulated potassium channel MloK1 at 16 A resolution. Structure, 15(9), 1053-1064.  Chow, S. S., Van Petegem, F., & Accili, E. A. (2012). Energetics of cyclic AMP binding to HCN channel C terminus reveal negative cooperativity. J Biol Chem, 287(1), 600-606.  194  Chung, W. K., Shin, M., Jaramillo, T. C., Leibel, R. L., LeDuc, C. A., Fischer, S. G., . . . Chetkovich, D. M. (2009). Absence epilepsy in apathetic, a spontaneous mutant mouse lacking the h channel subunit, HCN2. Neurobiol Dis, 33(3), 499-508.  Clapham, D. E. (1998). Not so funny anymore: pacing channels are cloned. Neuron, 21(1), 5-7.  Clayton, G. M., Silverman, W. R., Heginbotham, L., & Morais-Cabral, J. H. (2004). Structural basis of ligand activation in a cyclic nucleotide regulated potassium channel. Cell, 119(5), 615-627.  Colquhoun, D. (1998). Binding, gating, affinity and efficacy: the interpretation of structure-activity relationships for agonists and of the effects of mutating receptors. Br J Pharmacol, 125(5), 924-947.  Cornwell, T. L., Arnold, E., Boerth, N. J., & Lincoln, T. M. (1994). Inhibition of smooth muscle cell growth by nitric oxide and activation of cAMP-dependent protein kinase by cGMP. Am J Physiol, 267(5 Pt 1), C1405-1413.  Craven, K. B., Olivier, N. B., & Zagotta, W. N. (2008). C-terminal movement during gating in cyclic nucleotide-modulated channels. J Biol Chem, 283(21), 14728-14738.  Craven, K. B., & Zagotta, W. N. (2004). Salt bridges and gating in the COOH-terminal region of HCN2 and CNGA1 channels. J Gen Physiol, 124(6), 663-677.  Craven, K. B., & Zagotta, W. N. (2006). CNG and HCN channels: two peas, one pod. Annu Rev Physiol, 68, 375-401.  Das, R., Chowdhury, S., Mazhab-Jafari, M. T., Sildas, S., Selvaratnam, R., & Melacini, G. (2009). Dynamically driven ligand selectivity in cyclic nucleotide binding domains. J Biol Chem, 284(35), 23682-23696.  de Rooij, J., Rehmann, H., van Triest, M., Cool, R. H., Wittinghofer, A., & Bos, J. L. (2000). Mechanism of regulation of the Epac family of cAMP-dependent RapGEFs. J Biol Chem, 275(27), 20829-20836.  de Rooij, J., Zwartkruis, F. J., Verheijen, M. H., Cool, R. H., Nijman, S. M., Wittinghofer, A., & Bos, J. L. (1998). Epac is a Rap1 guanine-nucleotide-exchange factor directly activated by cyclic AMP. Nature, 396(6710), 474-477.  DeBerg, H. A., Bankston, J. R., Rosenbaum, J. C., Brzovic, P. S., Zagotta, W. N., & Stoll, S. (2015). Structural mechanism for the regulation of HCN ion channels by the accessory protein TRIP8b. Structure, 23(4), 734-744.  Decher, N., Bundis, F., Vajna, R., & Steinmeyer, K. (2003). KCNE2 modulates current amplitudes and activation kinetics of HCN4: influence of KCNE family members on HCN4 currents. Pflügers Arch, 446(6), 633-640.  Decher, N., Chen, J., & Sanguinetti, M. C. (2004). Voltage-dependent gating of hyperpolarization-activated, cyclic nucleotide-gated pacemaker channels: molecular coupling between the S4-S5 and C-linkers. J Biol Chem, 279(14), 13859-13865.  Dekker, J., Yellen, G. (2006). Cooperative Gating between Single HCN Pacemaker Channels. J Gen Physiol, 128(5), 561-567.  195  Dibbens, L. M., Reid, C. A., Hodgson, B., Thomas, E. A., Phillips, A. M., Gazina, E., . . . Petrou, S. (2010). Augmented currents of an HCN2 variant in patients with febrile seizure syndromes. Ann Neurol, 67(4), 542-546.  DiFrancesco, D. (1981a). A new interpretation of the pace-maker current in calf Purkinje fibres. J Physiol, 314, 359-376.  DiFrancesco, D. (1981b). A study of the ionic nature of the pace-maker current in calf Purkinje fibres. J Physiol, 314, 377-393.  DiFrancesco, D. (1982). Block and activation of the pace-maker channel in calf purkinje fibres: effects of potassium, caesium and rubidium. J Physiol, 329, 485-507.  DiFrancesco, D. (1986). Characterization of single pacemaker channels in cardiac sino-atrial node cells. Nature, 324(6096), 470-473.  DiFrancesco, D. (1991). The contribution of the hyperpolarization-activated current (If) to the generatiion of spontaneous activity in rabbit sino-atrial node myocytes. J Physiol., 434, 23-40.  DiFrancesco, D. (1993). Pacemaker mechanisms in cardiac tissue. Annu Rev Physiol, 55, 455-472.  DiFrancesco, D. (2010). The Role of the Funny Current in Pacemaker Activity. Circulation Research, 106, 434-446.  DiFrancesco, D. (2013). Funny channel gene mutations associated with arrhythmias. J Physiol., 591(Pt 17), 4117-4124.  DiFrancesco, D., & Borer, J. S. (2007). The funny current: cellular basis for the control of heart rate. Drugs, 67 Suppl 2, 15-24.  DiFrancesco, D., Ducouret, P., & Robinson, R. B. (1989). Muscarinic modulation of cardiac rate at low acetylcholine concentrations. Science, 243(4891), 669-671.  DiFrancesco, D., Ferroni, A., Mazzanti, M., & Tromba, C. (1986). Properties of the hyperpolarizing-activated current (if) in cells isolated from the rabbit sino-atrial node. J Physiol, 377, 61-88.  DiFrancesco, D., & Mangoni, M. (1994). Modulation of single hyperpolarization-activated channels (i(f)) by cAMP in the rabbit sino-atrial node. J Physiol, 474(3), 473-482.  DiFrancesco, D., Ohba, M., & Ojeda, C. (1979). Measurement and significance of the reversal potential for the pacemaker current (iK2) in sheep Purkinje fibers. J Physiol., 297, 135-162.  DiFrancesco, D., & Ojeda, C. (1980). Properties of the current if in the sino-atrial node of the rabbit compared with those of the current iK, in Purkinje fibres. J Physiol, 308, 353-367.  DiFrancesco, D., & Tortora, P. (1991). Direct activation of cardiac pacemaker channels by intracellular cyclic AMP. Nature, 351(6322), 145-147.  DiFrancesco, D., & Tromba, C. (1988a). Inhibition of the hyperpolarization-activated current (if) induced by acetylcholine in rabbit sino-atrial node myocytes. J Physiol, 405, 477-491.  196  DiFrancesco, D., & Tromba, C. (1988b). Muscarinic control of the hyperpolarization-activated current (if) in rabbit sino-atrial node myocytes. J Physiol, 405, 493-510.  DiFrancesco, J. C., Barbuti, A., Milanesi, R., Coco, S., Bucchi, A., Bottelli, G., . . . DiFrancesco, D. (2011). Recessive loss-of-function mutation in the pacemaker HCN2 channel causing increased neuronal excitability in a patient with idiopathic generalized epilepsy. J Neurosci, 31(48), 17327-17337.  Diller, T. C., Xuong Madhusudan, N. H., & Taylor, S. S. (2001). Molecular basis for regulatory subunit diversity in cAMP-dependent protein kinase: Crystal structure of the type II β regulatory subunit. . Structure, 9, 73-82.  Dong, A., Malecki, J. M., Lee, L., Carpenter, J. F., & Lee, J. C. (2002). Ligand-induced conformational and structural dynamics changes in Escherichia coli cyclic AMP receptor protein. Biochemistry, 41(21), 6660-6667.  Doyle, D. A., Morais Cabral, J., Pfuetzner, R. A., Kuo, A., Gulbis, J. M., Cohen, S. L., . . . MacKinnon, R. (1998). The structure of the potassium channel: molecular basis of K+ conduction and selectivity. Science, 280(5360), 69-77.  Dubé, C. M., McClelland, S., Choy, M. K., Brewster, A. L., Noam, Y., & Baram, T. Z. (2010). Fever, febrile seizures and epileptogenesis. Epilepsia, 51, 33.  Duhme, N., Schweizer, P. A., Thomas, D., Becker, R., Schroter, J., Barends, T. R., . . . Koenen, M. (2013). Altered HCN4 channel C-linker interaction is associated with familial tachycardia-bradycardia syndrome and atrial fibrillation. Eur Heart J.  Ebright, R. H., Le Grice, S. F., Miller, J. P., & Krakow, J. S. (1985). Analogs of cyclic AMP that elicit the biochemically defined conformational change in catabolite gene activator protein (CAP) but do not stimulate binding to DNA. J Mol Biol., 182, 91-107.  Ekman, I., Chassany, O., Komajda, M., Böhm, M., Borer, J. S., Ford, I., . . . Swedberg, K. (2011). Heart rate reduction with ivabradine and health related quality of life in patients with chronic heart failure: results from the SHIFT study. Eur Heart J., 32, 2395-2404.  Elinder, F., Männikkö, R., & Larsson, H. P. (2001). S4 charges move close to residues in the pore domain during activation in a K channel. J Gen Physiol, 118(1), 1-10.  Elinder, F., Mannikko, R., Pandey, S., & Larsson, H. P. (2006). Mode shifts in the voltage gating of the mouse and human HCN2 and HCN4 channels. J Physiol.  Ellinor, P. T., Lunetta, K. L., Albert, C. M., Glazer, N. L., Ritchie, M. D., Smith, A. V., . . . Kaab, S. (2012). Meta-analysis identifies six new susceptibility loci for atrial fibrillation. Nat Genet, 44(6), 670-675.  Emery, E. C., Young, G. T., Berrocoso, E. M., Chen, L., & McNaughton, P. A. (2011). HCN2 ion channels play a central role in inflammatory and neuropathic pain. Science, 333(6048), 1462-1466.  Emery, E. C., Young, G. T., & McNaughton, P. A. (2012). HCN2 ion channels: an emerging role as the pacemakers of pain. Trends Pharmacol Sci, 33(8), 456-463.  197  Emsley, P., Lohkamp, B., Scott, W. G., & Cowtan, K. (2010). Features and development of Coot. Acta Crystallogr D Biol Crystallogr, 66(Pt 4), 486-501.  Enserink, J. M., Christensen, A. E., de Rooij, J., van Triest, M., Schwede, F., Genieser, H. G., . . . Bos, J. L. (2002). A novel Epac-specific cAMP analogue demonstrates independent regulation of Rap1 and ERK. Nat Cell Biol, 4(11), 901-906.  Fares, N., Bois, P., Lenfant, J., & Potreau, D. (1998). Characterization of a hyperpolarization-activated current in dedifferentiated adult rat ventricular cells in primary culture. J Physiol, 506 ( Pt 1), 73-82.  Fenske, S., Mader, R., Scharr, A., Paparizos, C., Cao-Ehlker, X., Michalakis, S., . . . Biel, M. (2011). HCN3 contributes to the ventricular action potential waveform in the murine heart. Circ Res, 109(9), 1015-1023.  Ferrari, R. (2002). Selective heart rate reduction: a new approach to the management of angina pectoris. . Medicographia, 24, 201-210.  Ferrier, G. R., Zhu, J., Redondo, I. M., & Howlett, S. E. (1998). Role of cAMP-dependent protein kinase A in activation of a voltage-sensitive release mechanism for cardiac contraction in guinea-pig myocytes. J Physiol, 513 ( Pt 1), 185-201.  Finn, J. T., Grunwald, M. E., & Yau, K. W. (1996). Cyclic nucleotide– gated ion channels: an extended family with diverse functions. Annu Rev Physiol., 58, 395-426.  Flynn, G. E., Black, K. D., Islas, L. D., Sankaran, B., & Zagotta, W. N. (2007). Structure and rearrangements in the carboxy-terminal region of SpIH channels. Structure, 15(6), 671-682.  Flynn, G. E., & Zagotta, W. N. (2011). Molecular mechanism underlying phosphatidylinositol 4,5-bisphosphate-induced inhibition of SpIH channels. J Biol Chem, 286(17), 15535-15542.  Fox, K., Ford, I., Steg, P. G., Tendera, M., & Ferrari, R. (2008). Ivabradine for patients with stable coronary artery disease and left-ventricular systolic dysfunction (BEAUTIFUL): a randomised, double-blind, placebo-controlled trial. Lancet, 372(9641), 807-816.  Fox, K., Komajda, M., Ford, I., Robertson, M., Bohm, M., Borer, J. S., . . . Swedberg, K. (2013). Effect of ivabradine in patients with left-ventricular systolic dysfunction: a pooled analysis of individual patient data from the BEAUTIFUL and SHIFT trials. Eur Heart J, 34(29), 2263-2270.  Frere, S. G., & Luthi, A. (2004). Pacemaker channels in mouse thalamocortical neurones are regulated by distinct pathways of cAMP synthesis. J Physiol, 554(Pt 1), 111-125.  Frings, S. (2001). Chemoelectrical signal transduction in olfactory sensory neurons of air-breathing vertebrates. Cell Mol Life Sci, 58, 510-519.  Garcia-Frigola, C., Shi, Y., & Evans, S. M. (2003). Expression of the hyperpolarization-activated cyclic nucleotide-gated cation channel HCN4 during mouse heart development. Gene Expr Patterns, 3(6), 777-783.  198  Gauss, R., Seifert, R., & Kaupp, U. B. (1998). Molecular identification of a hyperpolarization-activated channel in sea urchin sperm. Nature, 393(6685), 583-587.  Gisselmann, G., Wetzel, C. H., Warnstedt, M., & Hatt, H. (2004). Functional characterization of Ih-channel splice variants from Apis mellifera. FEBS Lett., 575(1-3), 99-104.  Gordon, S., & Zagotta, W. N. (1995). Subunit interactions in coordination of Ni2+ in cyclic nucleotide-gated channels. . Proc Natl Acad Sci USA, 92, 10222-10226.  Goulding, E. H., Tibbs, G. R., & Siegelbaum, S. A. (1994). Molecular mechanism of cyclic-nucleotide-gated channel activation. Nature, 372(6504), 369-374.  Gushchin, I. Y., Gordeliy, V. I., & Grudinin, S. (2012). A novel dimerization interface of cyclic nucleotide binding domain, which is disrupted in presence of cAMP: implications for CNG channels gating. J Mol Model., 18(9), 4053-4060.  Halliwell, J. V., & Adams, P. R. (1982). Voltage-clamp analysis of muscarinic excitation in hippocampal neurons. Brain Res, 250(1), 71-92.  Harzheim, D., Pfeiffer, K. H., Fabritz, L., Kremmer, E., Buch, T., Waisman, A., . . . Seifert, R. (2008). Cardiac pacemaker function of HCN4 channels in mice is confined to embryonic development and requires cyclic AMP. Embo J, 27(4), 692-703.  Hauswirth, O., Noble, D., & Tsien, R. W. (1968). Adrenaline: mechanism of action on the pacemaker potential in cardiac Purkinje fibers. Science, 162(856), 916-917.  Heerklotz, H., Binder, H., Lantzsch, G., Klose, G., & Blume, A. (1997). Lipid/Detergent Interaction Thermodynamics as a Function of Molecular Shape. The Journal of Physical Chemistry B, 101(4), 639-645.  Herring, N., Rigg, L., Terrar, D. A., & Paterson, D. J. (2001). NO-cGMP pathway increases the hyperpolarisation-activated current, I(f), and heart rate during adrenergic stimulation. Cardiovasc Res, 52(3), 446-453.  Herrmann, S., Stieber, J., Hofmann, F., & Ludwig, A. (2006). Temporally controlled deletion of the pacemaker channel HCN4 in the sinoatrial node. Naunyn-Schmiedebergs Arch Pharmacol, 372, 313.  Herrmann, S., Stieber, J., Stockl, G., Hofmann, F., & Ludwig, A. (2007). HCN4 provides a 'depolarization reserve' and is not required for heart rate acceleration in mice. Embo J, 26(21), 4423-4432.  Heusch, G., Skyschally, A., Gres, P., van Caster, P., Schilawa, D., & Schulz, R. (2008). Improvement of regional myocardial blood flow and function and reduction of infarct size with ivabradine: Protection beyond heart rate reduction. . Eur Heart J., 29, 2265-2275.  Ho, W. K., Brown, H. F., & Noble, D. (1994). High selectivity of the i(f) channel to Na+ and K+ in rabbit isolated sinoatrial node cells. Pflügers Arch, 426, 68-74.  Hoesl, E., Stieber, J., Herrmann, S., Feil, S., Tybl, E., Hofmann, F., . . . Ludwig, A. (2008). Tamoxifen-inducible gene deletion in the cardiac conduction system. J Mol Cell Cardiol, 45(1), 62-69.  199  Hua, L., & Gordon, S. E. (2005). Functional Interactions Between A' Helices in the C-linker of Open CNG Channels. J Gen Physiol, 125(3), 335-344.  Huang, G. Y., Kim, J. J., Reger, A. S., Lorenz, R., Moon, E. W., Zhao, C., . . . Kim, C. (2014). Structural basis for cyclic-nucleotide selectivity and cGMP-selective activation of PKG I. Structure, 22(1), 116-124.  Huang, X., Yang, P., Du, Y., Zhang, J., & Ma, A. (2007). Age-related down-regulation of HCN channels in rat sinoatrial node. Basic Res Cardiol, 102(5), 429-435.  Huser, J., Blatter, L. A., & Lipsius, S. L. (2000). Intracellular Ca2+ release contributes to automaticity in cat atrial pacemaker cells. J Physiol, 524 Pt 2, 415-422.  Igami, K., Yamaguchi, N., & Kasai, M. (1999). Regulation of depolarization-induced calcium release from skeletal muscle triads by cyclic AMP-dependent protein kinase. Jpn J Physiol, 49(1), 81-87.  Ingram, S. L., & Williams, J. T. (1996). Modulation of the hyperpolarization-activated current (Ih) by cyclic nucleotides in guinea-pig primary afferent neurons. J Physiol, 492 ( Pt 1), 97-106.  Irisawa, H., Brown, H. F., & Giles, W. (1993). Cardiac pacemaking in the sinoatrial node. Physiol Rev, 73(1), 197-227.  Ishii, T. M., Nakashima, N., Takatsuka, K., & Ohmori, H. (2007). Peripheral N- and C-terminal domains determine deactivation kinetics of HCN channels. Biochem Biophys Res Commun., 359(3), 592-598.  Ishii, T. M., Takano, M., & Ohmori, H. (2001). Determinants of activation kinetics in mammalian hyperpolarization-activated cation channels. J Physiol, 537(Pt 1), 93-100.  Ishii, T. M., Takano, M., Xie, L. H., Noma, A., & Ohmori, H. (1999). Molecular characterization of the hyperpolarization-activated cation channel in rabbit heart sinoatrial node. J Biol Chem, 274(18), 12835-12839.  Jackson, H. A., Marshall, C. R., & Accili, E. A. (2007). Evolution and structural diversification of hyperpolarization-activated cyclic nucleotide-gated channel genes. Physiol Genomics, 29(3), 231-245.  Jiang, H., Colbran, J. L., Francis, S. H., & Corbin, J. D. (1992). Direct evidence for cross-activation of cGMP-dependent protein kinase by cAMP in pig coronary arteries. J Biol Chem, 267(2), 1015-1019.  Jouven, X., Empana, J. P., Schwartz, P. J., Desnos, M., Courbon, D., & Ducimetière, P. (2005). Heart-rate profile during exercise as a predictor of sudden death. N Engl J Med, 352(19), 1951-1958.  Kabsch, W. (2010). XDS. Acta Crystallogr D Biol Crystallogr, 66(Pt 2), 125-132.  Kannan, N., Wu, J., Anand, G. S., Yooseph, S., Neuwald, A. F., Venter, J. C., & Taylor, S. S. (2007). Evolution of allostery in the cyclic nucleotide binding module. Genome Biol., 8(12), R264.  200  Kase, D., & Imoto, K. (2012). The Role of HCN Channels on Membrane Excitability in the Nervous System. J Signal Transduct., 2012, 619747.  Kaupp, U. B., Niidome, T., Tanabe, T., Terada, S., Bönigk, W., Stühmer, W., . . . Numma, S. (1989). Primary structure and functional expression from complementary DNA of the rod photoreceptor cyclic GMP-gated channel. Nature, 342(6251), 762-766.  Kesters, D., Brams, M., Nys, M., Wijckmans, E., Spurny, R., Voets, T., . . . Ulens, C. (2015). Structure of the SthK Carboxy-Terminal Region Reveals a Gating Mechanism for Cyclic Nucleotide-Modulated Ion Channels. PLoS One, 10(1), e0116369.  Kim, C., Cheng, C. Y., Saldanha, S. A., & Taylor, S. S. (2007). PKA-I holoenzyme structure reveals a mechanism for cAMP-dependent activation. Cell, 130(6), 1032-1043.  Kim, C., Vigil, D., Anand, G., & Taylor, S. S. (2006). Structure and dynamics of PKA signaling proteins. Eur J Cell Biol., 85(7), 651-654.  Kim, D., Song, I., Keum, S., Lee, T., Jeong, M. J., Kim, S. S., . . . Shin, H. S. (2001). Lack of the burst firing of thalamocortical relay neurons and resistance to absence seizures in mice lacking alpha(1G) T-type Ca(2+) channels. Neuron, 31(1), 35-45.  Kim, J. J., Casteel, D. E., Huang, G., Kwon, T. H., Ren, R. K., Zwart, P., . . . Kim, C. (2011). Co-crystal structures of PKG Iβ (92-227) with cGMP and cAMP reveal the molecular details of cyclic-nucleotide binding. PLoS One, 6(4), e18413.  Kim, J. J., Flueck, C., Franz, E., Sanabria-Figueroa, E., Thompson, E., Lorenz, R., . . . Kim, C. (2015). Crystal structures of the carboxyl cGMP binding domain of the Plasmodium falciparum cGMP-dependent protein kinase reveal a novel capping triad crucial for merozoite egress. PLoS Pathog., 11(2), e1004639.  Kornev, A. P., Taylor, S. S., & Ten Eyck, L. F. (2008). A Generalized Allosteric Mechanism for cis-Regulated Cyclic Nucleotide Binding Domains. PloS Comput Biol., 4(4), e1000056.  Kowal, J., Chami, M., Baumgartner, P., Arheit, M., Chiu, P. L., Rangl, M., . . . Stahlberg, H. (2014). Ligand-induced structural changes in the cyclic nucleotide-modulated potassium channel MloK1. Nat Commun., 5, 3106.  Krissinel, E., & Henrick, K. (2007). Inference of macromolecular assemblies from crystalline state. J Mol Biol, 372(3), 774-797.  Krupinski, J., Coussen, F., Bakalyar, H. A., Tang, W. J., Feinstein, P. G., Orth, K., . . . Gilman, A. G. (1989). Adenylyl cyclase amino acid sequence: possible channel- or transporter-like structure. Science, 244(4912), 1558-1564.  Kuang, Q., Purhonen, P., & Hebert, H. (2015). Structure of potassium channels. Cell Mol Life Sci, 72(19), 3677-3693.  Kusch, J., Biskup, C., Thon, S., Schulz, E., Nache, V., Zimmer, T., . . . Benndorf, K. (2010). Interdependence of receptor activation and ligand binding in HCN2 pacemaker channels. Neuron, 67(1), 75-85.  201  Kusch, J., Thon, S., Schulz, E., Biskup, C., Nache, V., Zimmer, T., . . . Benndorf, K. (2011). How subunits cooperate in cAMP-induced activation of homotetrameric HCN2 channels. Nat Chem Biol, 8(2), 162-169.  Kusch, J., Zimmer, T., Holschuh, J., Biskup, C., Schulz, E., Nache, V., & Benndorf, K. (2010). Role of the S4-S5 linker in CNG channel activation. Biophys J, 99(8), 2488-2496.  Kuwabara, Y., Kuwahara, K., Takano, M., Kinoshita, H., Arai, Y., Yasuno, S., . . . Nakao, K. (2013). Increased expression of HCN channels in the ventricular myocardium contributes to enhanced arrhythmicity in mouse failing hearts. J Am Heart Assoc., 2(3), e000150.  Kwan, D. C., Prole, D. L., & Yellen, G. (2012). Structural changes during HCN channel gating defined by high affinity metal bridges. J Gen Physiol, 140(3), 279-291.  Laish-Farkash, A., Glikson, M., Brass, D., Marek-Yagel, D., Pras, E., Dascal, N., . . . Luria, D. (2010). A novel mutation in the HCN4 gene causes symptomatic sinus bradycardia in Moroccan Jews. J Cardiovasc Electrophysiol, 21(12), 1365-1372.  Lebedev, A. A., Young, P., Isupov, M. N., Moroz, O. V., Vagin, A. A., & Murshudov, G. N. (2012). JLigand: a graphical tool for the CCP4 template-restraint library. Acta Crystallogr D Biol Crystallogr, 68(Pt 4), 431-440.  Lee, S. Y., Lee, A., Chen, J., & MacKinnon, R. (2005). Structure of the KvAP voltage-dependent K+ channel and its dependence on the lipid membrane. Proc Natl Acad Sci U S A, 102(43), 15441-15446.  Leresche, N., Jassik-Gerschenfeld, D., Haby, M., Soltesz, I., & Crunelli, V. (1990). Pacemaker-like and other types of spontaneous membrane potential oscillations of thalamocortical cells. Neurosci Lett, 113(1), 72-77.  Lewis, A. S., Estep, C. M., & Chetkovich, D. M. (2010). The fast and slow ups and downs of HCN channel regulation. Channels (Austin), 4(3), 215-231.  Lewis, A. S., Schwartz, E., Chan, C. S., Noam, Y., Shin, M., Wadman, W. J., . . . Chetkovich, D. M. (2009). Alternatively spliced isoforms of TRIP8b differentially control h channel trafficking and function. J Neurosci, 29(19), 6250-6265.  Li, Q., Ng, H. Q., Yoon, H. S., & Kang, C. (2014). Insight into the molecular interaction between the cyclic nucleotide-binding homology domain and the eag domain of the hERG channel. FEBS Lett., 588(17), 2782-2788.  Liao, Z., Lockhead, D., Larson, E. D., & Proenza, C. (2010). Phosphorylation and modulation of hyperpolarization-activated HCN4 channels by protein kinase A in the mouse sinoatrial node. J Gen Physiol., 136(3), 247-258.  Liao, Z., Lockhead, D., St Clair, J. R., Larson, E. D., Wilson, C. E., & Proenza, C. (2012). Cellular context and multiple channel domains determine cAMP sensitivity of HCN4 channels: ligand-independent relief of autoinhibition in HCN4. J Gen Physiol., 140(5), 557-566.  202  Lin, H., Xiao, J., Luo, X., Chen, G., & Wang, Z. (2009). Transcriptional control of pacemaker channel genes HCN2 and HCN4 by Sp1 and implications in re-expression of these genes in hypertrophied myocytes. Cell Physiol Biochem, 23(4-6), 317-326.  Liu, H., & Aldrich, R. W. (2011). Tissue-specific N Terminus of the HCN4 Channel Affects Channel Activation. J Biol Chem, 286, 14209-14214.  Lolicato, M., Bucchi, A., Arrigoni, C., Zucca, S., Nardini, M., Schroeder, I., . . . Moroni, A. (2014). Cyclic dinucleotides bind the C-linker of HCN4 to control channel cAMP responsiveness. Nat Chem Biol, 10(6), 457-462.  Lolicato, M., Nardini, M., Gazzarrini, S., Moller, S., Bertinetti, D., Herberg, F. W., . . . Moroni, A. (2011). Tetramerization dynamics of C-terminal domain underlies isoform-specific cAMP gating in hyperpolarization-activated cyclic nucleotide-gated channels. J Biol Chem, 286(52), 44811-44820.  Long, S. B., Campbell, E. B., & Mackinnon, R. (2005). Crystal structure of a mammalian voltage-dependent Shaker family K+ channel. Science, 309(5736), 897-903.  Lu, Z., Klem, A. M., & Ramu, Y. (2001). Ion conduction pore is conserved among potassium channels. . Nature Genetics, 413(6858), 809-813.  Ludwig, A., Budde, T., Stieber, J., Moosmang, S., Wahl, C., Holthoff, K., . . . Hofmann, F. (2003). Absence epilepsy and sinus dysrhythmia in mice lacking the pacemaker channel HCN2. Embo J, 22(2), 216-224.  Ludwig, A., Zong, X., Hofmann, F., & Biel, M. (1999). Structure and function of cardiac pacemaker channels. Cell Physiol Biochem, 9(4-5), 179-186.  Ludwig, A., Zong, X., Jeglitsch, M., Hofmann, F., & Biel, M. (1998). A family of hyperpolarization-activated mammalian cation channels. Nature, 393(6685), 587-591.  Ludwig, A., Zong, X., Stieber, J., Hullin, R., Hofmann, F., & Biel, M. (1999). Two pacemaker channels from human heart with profoundly different activation kinetics. Embo J, 18(9), 2323-2329.  Maccaferri, G., Mangoni, M., Lazzari, A., & DiFrancesco, D. (1993). Properties of the hyperpolarization-activated current in rat hippocampal CA1 pyramidal cells. J Neurophysiol, 69(6), 2129-2136.  MacKinnon, R. (2003). Potassium Channels. FEBS Lett., 555(1), 62-65.  Macri, V., & Accili, E. A. (2004). Structural elements of instantaneous and slow gating in hyperpolarization-activated cyclic nucleotide-gated channels. J Biol Chem, 279(16), 16832-16846.  Macri, V., Angoli, D., & Accili, E. A. (2012). Architecture of the HCN selectivity filter and control of cation permeation. Sci Rep, 2, 894.  Macri, V., Mahida, S. N., Zhang, M. L., Sinner, M. F., Dolmatova, E. V., Tucker, N. R., . . . Ellinor, P. T. (2014). A novel trafficking-defective HCN4 mutation is associated with early-onset atrial fibrillation. Heart Rhythm, 11(6), 1055-1062.  203  Mangoni, M. E. N., J. (2001). Properties of the hyperpolarization-activated current (If) in isolated mouse sino-atrial cells. Cardiovascular Research, 52(1), 51-64.  Mannikko, R., Elinder, F., & Larsson, H. P. (2002). Voltage-sensing mechanism is conserved among ion channels gated by opposite voltages. Nature, 419(6909), 837-841.  Mari, S. A., Pessoa, J., Altieri, S., Hensen, U., Thomas, L., Morais-Cabral, J. H., & Müller, D. J. (2011). Gating of the MlotiK1 potassium channel involves large rearrangements of the cyclic nucleotide-binding domains. Proc Natl Acad Sci USA, 108(51), 20802-20807.  Marques-Carvalho, M. J., Sahoo, N., Muskett, F. W., Vieira-Pires, R. S., Gabant, G., Cadene, M., . . . Morais-Cabral, J. H. (2012). Structural, biochemical, and functional characterization of the cyclic nucleotide binding homology domain from the mouse EAG1 potassium channel. J Mol Biol., 423(1), 34-46.  Matulef, K., Flynn, G. E., & Zagotta, W. N. (1999). Molecular rearrangements in the ligand-binding domain of cyclic nucleotide-gated channels. Neuron, 24(2), 443-452.  Matulef, K., & Zagotta, W. N. (2002). Multimerization of the ligand binding domains of cyclic nucleotide-gated channels. Neuron, 36(1), 93-103.  Mazzolini, M., Marchesi, A., Giorgetti, A., & Torre, V. (2010). Gating in CNGA1 channels. Pflügers Arch, 459(4), 547-555.  Mazzolini, M., Punta, M., & Torre, V. (2002). Movement of the C-helix during the gating of cyclic nucleotide-gated channels. Biophys J., 83(6), 3283-3295.  McCormick, D. A., & Bal, T. (1997). Sleep and arousal: thalamocortical mechanisms. Annu Rev Neurosci, 20, 185-215.  McCormick, D. A., & Pape, H. C. (1990a). Noradrenergic and serotonergic modulation of a hyperpolarization-activated cation current in thalamic relay neurones. J Physiol, 431, 319-342.  McCormick, D. A., & Pape, H. C. (1990b). Properties of a hyperpolarization-activated cation current and its role in rhythmic oscillation in thalamic relay neurones. J Physiol, 431, 291-318.  McDonald, I. K., & Thornton, J. M. (1994). Satisfying hydrogen bonding potential in proteins. J Mol Biol, 238(5), 777-793. doi: 10.1006/jmbi.1994.1334 Michels, G., Er, F., Khan, I., Sudkamp, M., Herzig, S., & Hoppe, U. C. (2005). Single-channel properties support a potential contribution of hyperpolarization-activated cyclic nucleotide-gated channels and If to cardiac arrhythmias. Circulation, 111(4), 399-404.  Milanesi, R., Baruscotti, M., Gnecchi-Ruscone, T., & DiFrancesco, D. (2006). Familial sinus bradycardia associated with a mutation in the cardiac pacemaker channel. N Engl J Med, 354(2), 151-157.  Milano, A., Vermeer, A. M., Lodder, E. M., Barc, J., Verkerk, A. O., Postma, A. V., . . . Bezzina, C. R. (2014). HCN4 mutations in multiple families with bradycardia and left ventricular noncompaction cardiomyopathy. J Am Coll Cardiol, 64(8), 745-756.  204  Mistrik, P., Mader, R., Michalakis, S., Weidinger, M., Pfeifer, A., & Biel, M. (2005). The Murine HCN3 Gene Encodes a Hyperpolarization-activated Cation Channel with Slow Kinetics and Unique Response to Cyclic Nucleotides. J. Biol. Chem., 280(29), 27056-27061.  Mo, E., Amin, H., Bianco, I. H., & Garthwaite, J. (2004). Kinetics of a cellular nitric oxide/cGMP/phosphodiesterase-5 pathway. J Biol Chem, 279(25), 26149-26158.  Mohapatra, D. P., Misonou, H., Pan, S. J., Held, J. E., Surmeier, D. J., & Trimmer, J. S. (2009). Regulation of intrinsic excitability in hippocampal neurons by activity-dependent modulation of the KV2.1 potassium channel. Channels (Austin), 3(1), 46-56.  Möller, S., Alfieri, A., Bertinetti, D., Aquila, M., Schwede, F., Lolicato, M., . . . Herberg, F. W. (2014). Cyclic nucleotide mapping of hyperpolarization-activated cyclic nucleotide-gated (HCN) channels. ACS Chem Biol., 9(5), 1128-1137.  Monod, J., Wyman, J., & Changeux, J. P. (1965). On the nature of allosteric transitions: A plausible model. J Mol Biol., 12, 88-118.  Moosmang, S., Biel, M., Hofmann, F., & Ludwig, A. (1999). Differential distribution of four hyperpolarization-activated cation channels in mouse brain. Biol Chem, 380(7-8), 975-980.  Moroni, A., Gorza, L., Beltrame, M., Gravante, B., Vaccari, T., Bianchi, M. E., . . . DiFrancesco, D. (2001). Hyperpolarization-activated cyclic nucleotide-gated channel 1 is a molecular determinant of the cardiac pacemaker current I(f). J Biol Chem, 276(31), 29233-29241.  Much, B., Wahl-Schott, C., Zong, X., Schneider, A., Baumann, L., Moosmang, S., . . . Biel, M. (2003). Role of subunit heteromerization and N-linked glycosylation in the formation of functional hyperpolarization-activated cyclic nucleotide-gated channels. J Biol Chem, 278(44), 43781-43786.  Mulder, P., Barbier, S., Chagraoui, A., Richard, V., Henry, J. P., Lallemand, F., . . . Thuillez, C. (2004). Long-term heart rate reduction induced by the selective If current inhibitor ivabradine improves left ventricular function and intrinsic myocardial structure in congestive heart failure. . Circulation., 109, 1674-1679.  Mulley, J. C., Scheffer, I. E., Petrou, S., & Berkovic, S. F. (2003). Channelopathies as a genetic cause of epilepsy. Curr Opin Neurol, 16, 171-176.  Nakamura, Y., Shi, X., Numata, T., Mori, Y., Inoue, R., Lossin, C., . . . Hirose, S. (2013). Novel HCN2 mutation contributes to febrile seizures by shifting the channel's kinetics in a temperature-dependent manner. PLoS One, 8(12), e80376.  Nava, C., Dalle, C., Rastetter, A., Striano, P., de Kove, C. G. F., Nabbout, R., . . . Depienne, C. (2014). De novo mutations in HCN1 cause early infantile epileptic encephalopathy. Nature Genetics, 46, 640+.  Nazzari, H., Angoli, D., Chow, S. S., Whitaker, G., Leclair, L., McDonald, E., . . . Accili, E. A. (2008). Regulation of cell surface expression of functional pacemaker channels by a motif in the B-helix of the cyclic nucleotide-binding domain. Am J Physiol Cell Physiol, 295(3), C642-652.  205  Netter, M. F., Zuzarte, M., Schlichthörl, G., Klöcker, N., & Decher, N. (2012). The HCN4 Channel Mutation D553N Associated With Bradycardia Has a C-linker Mediated Gating Defect. Cell Physiol Biochem, 30, 1227-1240.  Newton, R. P., Kingston, E. E., Hakeem, N. A., Salih, S. G., Beynon, J. H., & Moyse, C. D. (1986). Extraction, purification, identification and metabolism of 3',5'-cyclic UMP, 3',5'-cyclic IMP and 3',5'-cyclic dTMP from rat tissues. Biochem J, 236(2), 431-439.  Noble, D., & Tsien, R. W. (1968). The kinetics and rectifier properties of the slow potassium current in cardiac Purkinje fibres. J Physiol, 195(1), 185-214.  Nof, E., Luria, D., Brass, D., Marek, D., Lahat, H., Reznik-Wolf, H., . . . Glikson, M. (2007). Point Mutation in the HCN4 Cardiac Ion Channel Pore Affecting Synthesis, Trafficking, and Functional Expression Is Associated With Familial Asymptomatic Sinus Bradycardia. Circulation, 116(5), 463-470.  Nolan, M. F., Dudman, J. T., Dodson, P. D., & Santoro, B. (2007). HCN1 channels control resting and active integrative properties of stellate cells from layer II of the entorhinal cortex. J Neurosci, 27(46), 12440-12451.  Nolan, M. F., Malleret, G., Lee, K. H., Gibbs, E., Dudman, J. T., Santoro, B., . . . Morozov, A. (2003). The hyperpolarization-activated HCN1 channel is important for motor learning and neuronal integration by cerebellar Purkinje cells. Cell, 115(5), 551-564.  Noma, A., & Irisawa, H. (1976). Membrane currents in the rabbit sinoatrial node cell as studied by the double microelectrode method. Pflügers Arch, 364(1), 45-52.  Oberhauser, V., Schwertfeger, E., Rutz, T., Beyersdorf, F., & Rump, L. C. (2001). Acetylcholine release in human heart atrium: influence of muscarinic autoreceptors, diabetes, and age. Circulation., 103(12), 1638-1643.  Otwinowski, Z., & Minor, W. (1997). Processing of X-ray Diffraction Data Collected in Oscillation Mode. Methods in Enzymology, 276, 307-326.  Paoletti, P., Young, E. C., & Siegelbaum, S. A. (1999). C-Linker of cyclic nucleotide-gated channels controls coupling of ligand binding to channel gating. J Gen Physiol, 113(1), 17-34.  Pape, H. C. (1996). Queer current and pacemaker: the hyperpolarization-activated cation current in neurons. Annu Rev Physiol, 58, 299-327.  Pape, H. C., & McCormick, D. A. (1989). Noradrenaline and serotonin selectively modulate thalamic burst firing by enhancing a hyperpolarization-activated cation current. Nature, 340, 715-718.  Passner, J. M., Schultz, S. C., & Steitz, T. A. (2000). Modeling the cAMP-induced Allosteric Transition Using the Crystal Structure of CAP-cAMP at 2.1 Å Resolution. J Mol Biol., 304(5), 847-859.  Pearson, W. R., Wood, T., Zhang, Z., & Miller, W. (1997). Comparison of DNA sequences with protein sequences. Genomics, 46(1), 24-36.  206  Pelligrino, D. A., & Wang, Q. (1998). Cyclic nucleotide crosstalk and the regulation of cerebral vasodilation. Prog Neurobiol, 56(1), 1-18.  Peng, B. W., Justice, J. A., Zhang, K., He, X. H., & Sanchez, R. M. (2010). Increased basal synaptic inhibition of hippocampal area CA1 pyramidal neurons by an antiepileptic drug that enhances I(h). Neuropsychopharmacology, 35, 464-472.  Pessoa, J., Fonseca, F., Furini, S., & Morais-Cabral, J. H. (2014). Determinants of ligand selectivity in a cyclic nucleotide-regulated potassium channel. J Gen Physiol, 144(1), 41-54.  Peters, C. J., Chow, S. S., Angoli, D., Nazzari, H., Cayabyab, F. S., Morshedian, A., & Accili, E. A. (2009). In situ co-distribution and functional interactions of SAP97 with sinoatrial isoforms of HCN channels. J Mol Cell Cardiol, 46(5), 636-643.  Pian, P., Bucchi, A., Decostanzo, A., Robinson, R. B., & Siegelbaum, S. A. (2007). Modulation of cyclic nucleotide-regulated HCN channels by PIP(2) and receptors coupled to phospholipase C. Pflügers Arch, 455, 125-145.  Pian, P., Bucchi, A., Robinson, R. B., & Siegelbaum, S. A. (2006). Regulation of gating and rundown of HCN hyperpolarization-activated channels by exogenous and endogenous PIP2. J Gen Physiol, 128(5), 593-604.  Poolos, N. P. (2005). The h-channel: a potential channelopathy in epilepsy? Epilepsy Behav., 7(1), 51-56.  Poolos, N. P. (2010). Genetic loss of HCN1 channels is exciting, but is it epileptic? Epilepsy Curr., 10(2), 49-51.  Poolos, N. P. (2012). Hyperpolarization-Activated Cyclic Nucleotide-Gated (HCN) Ion Channelopathy in Epilepsy.  Poolos, N. P., Migliore, M., & Johnston, D. (2002). Pharmacological upregulation of h-channels reduces the excitability of pyramidal neuron dendrites. Nat Neurosci, 5, 767-774.  Popovych, N., Sun, S., Ebright, R. H., & Kalodimos, C. G. (2006). Dynamically driven protein allostery. Nat Struct Mol Biol, 13(9), 831-838.  Popovych, N., Tzeng, S. R., Tonelli, M., Ebright, R. H., & Kalodimos, C. G. (2009). Structural basis for cAMP-mediated allosteric control of the catabolite activator protein. Proc Natl Acad Sci USA, 106(17), 6927-6932.  Premont, R. T., Matsuoka, I., Mattei, M. G., Pouille, Y., Defer, N., & Hanoune, J. (1996). Identification and characterization of a widely expressed form of adenylyl cyclase. J Biol Chem, 271(23), 13900-13907.  Proenza, C., Angoli, D., Agranovich, E., Macri, V., & Accili, E. A. (2002). Pacemaker channels produce an instantaneous current. J Biol Chem, 277(7), 5101-5109.  Proenza, C., Tran, N., Angoli, D., Zahynacz, K., Balcar, P., & Accili, E. A. (2002). Different roles for the cyclic nucleotide binding domain and amino terminus in assembly and expression of hyperpolarization-activated, cyclic nucleotide-gated channels. J Biol Chem, 277(33), 29634-29642.  207  Prole, D. L., & Yellen, G. (2006). Reversal of HCN Channel Voltage Dependence via Bridging of the S4-S5 Linker and Post-S6. J Gen Physiol.  Puljung, M. C., DeBerg, H. A., Zagotta, W. N., & Stoll, S. (2014). Double electron-electron resonance reveals cAMP-induced conformational change in HCN channels. Proc Natl Acad Sci USA, 111(27), 9816-9821.  Puljung, M. C., & Zagotta, W. N. (2013). A Secondary Structural Transition in the C-helix Promotes Gating of Cyclic Nucleotide-regulated Ion Channels. J Biol Chem, 288(18), 12944-12956.  Qu, J., Altomare, C., Bucchi, A., DiFrancesco, D., & Robinson, R. B. (2002). Functional comparison of HCN isoforms expressed in ventricular and HEK 293 cells. Pflügers Arch, 444(5), 597-601.  Rehmann, H., Prakash, B., Wolf, E., Rueppel, A., de Rooij, J., Bos, J. L., & Wittinghofer, A. (2003). Structure and regulation of the cAMP-binding domains of Epac2. Nat Struct Biol., 10(1), 26-32.  Rehmann, H., Schwede, F., Døskeland, S. O., Wittinghofer, A., & Bos, J. L. (2003). Ligand-mediated activation of the cAMP-responsive guanine nucleotide exchange factor Epac. J Biol Chem, 278(40), 38548-38556.  Rehmann, H., Wittinghofer, A., & Bos, J. L. (2007). Capturing cyclic nucleotides in action: snapshots from crystallographic studies. Nat Rev Mol Cell Biol, 8(1), 63-73.  Reid, C. A., Phillips, A. M., & Petrou, S. (2012). HCN channelopathies: pathophysiology in genetic epilepsy and therapeutic implications. Br J Pharmacol, 165(1), 49-56.  Riccioni, G., Vitulana, N., & D'Orazio, Z. (2009). Ivabradine: beyond heart rate control. Adv Ther, 26(1), 12-24.  Robinson, R. B., & Siegelbaum, S. A. (2003). Hyperpolarization-activated cation currents: from molecules to physiological function. Annu Rev Physiol, 65, 453-480.  Robinson, R. B., Yu, H., Chang, F., & Cohen, I. S. (1997). Developmental change in the voltage-dependence of the pacemaker current, if, in rat ventricle cells. Pflügers Arch, 433(4), 533-535.  Rothberg, B. S., Shin, K. S., Phale, P. S., & Yellen, G. (2002). Voltage-controlled gating at the intracellular entrance to a hyperpolarization-activated cation channel. J Gen Physiol, 119(1), 83-91.  Rothberg, B. S., Shin, K. S., & Yellen, G. (2003). Movements near the gate of a hyperpolarization-activated cation channel. J Gen Physiol., 122(5), 501-510.  Roubille, F., & Tardif, J. C. (2013). New therapeutic targets in cardiology: heart failure and arrhythmia: HCN channels. Circulation, 127(19), 1986-1996.  Rubenstein, D. S., & Lipsius, S. L. (1989). Mechanisms of automaticity in subsidiary pacemakers from cat right atrium. Circ Res, 64(4), 648-657.  208  Ruiz, M. L. K., J. W. (1999). Opening Mechanism of a Cyclic Nucleotide–gated Channel Based on Analysis of Single Channels Locked in Each Liganded State. J Gen Physiol, 113(6), 873-895.  Saibil, H. R. (2000). Conformational changes studied by cryo-electron microscopy. Nature Structural Biology, 7, 711-714.  Santoro, B., Grant, S. G., Bartsch, D., & Kandel, E. R. (1997). Interactive cloning with the SH3 domain of N-src identifies a new brain specific ion channel protein, with homology to eag and cyclic nucleotide-gated channels. Proc Natl Acad Sci U S A, 94(26), 14815-14820.  Santoro, B., Hu, L., Liu, H., Saponaro, A., Pian, P., Piskorowski, R. A., . . . Siegelbaum, S. A. (2011). TRIP8b regulates HCN1 channel trafficking and gating through two distinct C-terminal interaction sites. J Neurosci, 31(11), 4074-4086.  Santoro, B., Liu, D. T., Yao, H., Bartsch, D., Kandel, E. R., Siegelbaum, S. A., & Tibbs, G. R. (1998). Identification of a gene encoding a hyperpolarization-activated pacemaker channel of brain. Cell, 93(5), 717-729.  Saponaro, A., Pauleta, S. R., Cantini, F., Matzapetakis, M., Hammann, C., Donadoni, C., . . . Moroni, A. (2014). Structural basis for the mutual antagonism of cAMP and TRIP8b in regulating HCN channel function. Proc Natl Acad Sci USA, 111(40), 14577-14582.  Schneider, E. H., & Seifert, R. (2015). Report on the Third Symposium "cCMP and cUMP as New Second Messengers". Naunyn Schmiedebergs Arch Pharmacol, 388(1), 1-3.  Schultz, S. C., Shields, G. C., & Steitz, T. A. (1991). Crystal structure of a CAP-DNA complex: the DNA is bent by 90°. Science, 253(5023), 1001-1007.  Schulze-Bahr, E., Neu, A., Friederich, P., Kaupp, U. B., Breithardt, G., Pongs, O., & Isbrandt, D. (2003). Pacemaker channel dysfunction in a patient with sinus node disease. J Clin Invest, 111(10), 1537-1545.  Schünke, S., Stoldt, M., Lecher, J., Kaupp, U. B., & Willbold, D. (2011). Structural insights into conformational changes of a cyclic nucleotide-binding domain in solution from Mesorhizobium loti K1 channel. Proc Natl Acad Sci USA, 108(15), 6121-6126.  Schunke, S., Stoldt, M., Novak, K., Kaupp, U. B., & Willbold, D. (2009). Solution structure of the Mesorhizobium loti K1 channel cyclic nucleotide-binding domain in complex with cAMP. EMBO Rep, 10(7), 729-735.  Schweizer, P., Duhme, N., Thomas, D., Becker, R., Zehelein, J., Draguhn, A., . . . Koenen, M. (2010). cAMP sensitivity of HCN pacemaker channels determines basal heart rate but is not critical for autonomic rate control. Circ Arrhythm Electrophysiol., 3(5), 542-552.  Schweizer, P. A., Schroter, J., Greiner, S., Haas, J., Yampolsky, P., Mereles, D., . . . Thomas, D. (2014). The symptom complex of familial sinus node dysfunction and myocardial noncompaction is associated with mutations in the HCN4 channel. J Am Coll Cardiol, 64(8), 757-767.  209  Scicchitano, P., Cortese, F., Ricci, G., Carbonara, S., Moncelli, M., Iacoviello, M., . . . Ciccone, M. (2014). Ivabradine, coronary artery disease, and heart failure: beyond rhythm control. Drug Des Devel Ther, 8, 689-700.  Seifert, R., Scholten, A., Gauss, R., Mincheva, A., Lichter, P., & Kaupp, U. B. (1999). Molecular characterization of a slowly gating human hyperpolarization-activated channel predominantly expressed in thalamus, heart, and testis. Proc Natl Acad Sci U S A, 96(16), 9391-9396.  Shabb, J. B., & Corbin, J. D. (1992). Cyclic nucleotide-binding domains in proteins having diverse functions. J Biol Chem, 267(9), 5723-5726.  Shabb, J. B., Ng, L., & Corbin, J. D. (1990). One amino acid change produces a high affinity cGMP-binding site in cAMP-dependent protein kinase. J Biol Chem, 265, 16031-16034.  Shattock, M., & Camm, J. A. (2006). Pure heart rate reduction: the If channels from discovery to therapeutic target. . Br J Cardiol., 13(1), 27-35.  Shi, W., Wymore, R., Yu, H., Wu, J., Wymore, R. T., Pan, Z., . . . Cohen, I. S. (1999). Distribution and prevalence of hyperpolarization-activated cation channel (HCN) mRNA expression in cardiac tissues. Circ Res, 85(1), e1-6.  Shin, K. S., Maertens, C., Proenza, C., Rothberg, B. S., & Yellen, G. (2004). Inactivation in HCN channels results from reclosure of the activation gate: desensitization to voltage. Neuron, 41(5), 737-744.  Shin, K. S., Rothberg, B. S., & Yellen, G. (2001). Blocker state dependence and trapping in hyperpolarization-activated cation channels: evidence for an intracellular activation gate. J Gen Physiol., 117(2), 91-101.  Simon, M. I., Strathmann, M. P., & Gautam, N. (1991). Diversity of G proteins in signal transduction. Science, 252(5007), 802-808.  Siu, C. W., Lieu, D. K., & Li, R. A. (2006). HCN-encoded pacemaker channels: from physiology and biophysics to bioengineering. J Membr Biol., 214(3), 115-122.  Sprenger, J. U., & Nikolaev, V. O. (2013). Biophysical techniques for detection of cAMP and cGMP in living cells. Int J Mol Sci, 14(4), 8025-8046.  Stieber, J., Herrmann, S., Feil, S., Loster, J., Feil, R., Biel, M., . . . Ludwig, A. (2003). The hyperpolarization-activated channel HCN4 is required for the generation of pacemaker action potentials in the embryonic heart. Proc Natl Acad Sci U S A, 100(25), 15235-15240.  Stieber, J., Stockl, G., Herrmann, S., Hassfurth, B., & Hofmann, F. (2005). Functional expression of the human HCN3 channel. J Biol Chem, 280(41), 34635-34643.  Su, Y., Dostmann, W. R., Herberg, F. W., Durick, K., Xuong, N. H., Ten Eyck, L., . . . Varughese, K. I. (1995). Regulatory subunit of protein kinase A: structure of deletion mutant with cAMP binding domains. Science, 269(5225), 807-813.  210  Tang, B., Sander, T., Craven, K. B., Hempelmann, A., & Escayg, A. (2008). Mutation analysis of the hyperpolarization-activated cyclic nucleotide-gated channels HCN1 and HCN2 in idiopathic generalized epilepsy. Neurobiol Dis, 29(1), 59-70.  Taraska, J. W., Puljung, M. C., Olivier, N. B., Flynn, G. E., & Zagotta, W. N. (2009). Mapping the structure and conformational movements of proteins with transition metal ion FRET. Nat Methods, 6(7), 532-537.  Taraska, J. W., & Zagotta, W. N. (2007). Structural dynamics in the gating ring of cyclic nucleotide-gated ion channels. Nat Struct Mol Biol., 14(9), 854-860.  Tardif, J. C., Ford, I., Tendera, M., Bourassa, M. G., & Fox, K. (2005). Efficacy of ivabradine, a new selective I(f) inhibitor, compared with atenolol in patients with chronic stable angina. Eur Heart J, 26(23), 2529-2536.  Tardif, J. C. P., P., Kahan, T., & Investigators, A. S. (2009). Efficacy of the I(f) current inhibitor ivabradine in patients with chronic stable angina receiving beta-blocker therapy: a 4-month, randomized, placebo-controlled trial. Eur Heart J., 30(5), 540-548.  Thapa, A., Han, W., Simons, R. H., Chilkoti, A., Chi, E. Y., & Lopez, G. P. (2013). Effect of detergents on the thermal behavior of elastin-like polypeptides. Biopolymers, 99(1), 55-62.  Thollon, C., Bidouard, J. P., Cambarrat, C., Lesage, L., Reure, H., Delescluse, I., . . . Vilaine, J. P. (1997). Stereospecific in vitro and in vivo effects of the new sinus node inhibitor (+)-S 16257. Eur J Pharmacol., 339(1), 43-51.  Tibbs, G. R., Goulding, E. H., & Siegelbaum, S. A. (1997). Allosteric activation and tuning of ligand efficacy in cyclic-nucleotide-gated channels. Nature, 386, 612-615.  Tibbs, G. R., Liu, D. T., Leypold, B. G., & Siegelbaum, S. A. (1998). A state-independent interaction between ligand and a conserved arginine residue in cyclic nucleotide-gated channels reveals a functional polarity of the cyclic nucleotide binding site. J Biol Chem, 273(8), 4497-4505.  Tokimasa, T., & Akasu, T. (1990). Cyclic AMP regulates an inward rectifying sodium-potassium current in dissociated bull-frog sympathetic neurones. J Physiol, 420, 409-429.  Tran, N., Proenza, C., Macri, V., Petigara, F., Sloan, E., Samler, S., & Accili, E. A. (2002). A conserved domain in the NH2 terminus important for assembly and functional expression of pacemaker channels. J Biol Chem, 277(46), 43588-43592.  Ueda, K., Nakamura, K., Hayashi, T., Inagaki, N., Takahashi, M., Arimura, T., . . . Kimura, A. (2004). Functional characterization of a trafficking-defective HCN4 mutation, D553N, associated with cardiac arrhythmia. J Biol Chem, 279(26), 27194-27198.  Ulens, C., & Siegelbaum, S. A. (2003). Regulation of hyperpolarization-activated HCN channels by cAMP through a gating switch in binding domain symmetry. Neuron, 40(5), 959-970.  Vaca, L., Stieber, J., Zong, X., Ludwig, A., Hofmann, F., & Biel, M. (2000). Mutations in the S4 domain of a pacemaker channel alter its voltage dependence. FEBS Lett, 479(1-2), 35-40.  211  van Ginneken, A. C., & Giles, W. (1991). Voltage clamp measurements of the hyperpolarization-activated inward current I(f) in single cells from rabbit sino-atrial node. J Physiol, 434, 57-83.  VanSchouwen, B., Akimoto, M., Sayadi, M., Fogolari, F., & Melacini, G. (2015). Role of Dynamics in the Autoinhibition and Activation of the Hyperpolarization-activated Cyclic Nucleotide-modulated (HCN) Ion Channels. J Biol Chem, 290(29), 17642-17654.  Varnum, M. D., Black, K. D., & Zagotta, W. N. (1995). Molecular mechanism for ligand discrimination of cyclic nucleotide-gated channels. Neuron, 15(3), 619-625.  Varnum, M. D. Z., W. N. (1996). Subunit interactions in the activation of cyclic nucleotide–gated ion channels. Biophys. J, 70, 2667-2679.  Vassalle, M. (1966). Analysis of cardiac pacemaker potential using a "voltage clamp" technique. Am J Physiol, 210(6), 1335-1341.  Vemana, S., Pandey, S., & Larsson, H. P. (2004). S4 movement in a mammalian HCN channel. J Gen Physiol, 123(1), 21-32.  Verkerk, A. O., & Wilders, R. (2014). Pacemaker activity of the human sinoatrial node: effects of HCN4 mutations on the hyperpolarization-activated current. Europace, 16(3), 384-395.  Verkerk, A. O., & Wilders, R. (2015). Pacemaker activity of the human sinoatrial node: an update on the effects of mutations in HCN4 on the hyperpolarization-activated current. Int J Mol Sci, 16(2), 3071-3094.  Vicente-Steijn, R., Passier, R., Wisse, L. J., Schalij, M. J., Poelmann, R. E., Gittenberger-de Groot, A. C., & Jongbloed, M. R. (2011). Funny current channel HCN4 delineates the developing cardiac conduction system in chicken heart. Heart Rhythm, 8(8), 1254-1263.  Vigil, D., Lin, J. H., Sotriffer, C. A., Pennypacker, J. K., McCammon, J. A., & Taylor, S. S. (2006). A simple electrostatic switch important in the activation of type I protein kinase A by cyclic AMP. Protein Sci., 15(1), 113-121.  Viscomi, C., Altomare, C., Bucchi, A., Camatini, E., Baruscotti, M., Moroni, A., & DiFrancesco, D. (2001). C terminus-mediated control of voltage and cAMP gating of hyperpolarization-activated cyclic nucleotide-gated channels. J Biol Chem, 276(32), 29930-29934.  Wahl-Schott, C., & Biel, M. (2009). HCN channels: structure, cellular regulation and physiological function. Cell Mol Life Sci, 66(3), 470-494.  Wainger, B. J., DeGennaro, M., Santoro, B., Siegelbaum, S. A., & Tibbs, G. R. (2001). Molecular mechanism of cAMP modulation of HCN pacemaker channels. Nature, 411(6839), 805-810.  Wang, J., Chen, S., Nolan, M. F., & Siegelbaum, S. A. (2002). Activity-dependent regulation of HCN pacemaker channels by cyclic AMP: signaling through dynamic allosteric coupling. Neuron, 36(3), 451-461.  212  Wang, J., Chen, S., & Siegelbaum, S. A. (2001). Regulation of hyperpolarization-activated HCN channel gating and cAMP modulation due to interactions of COOH terminus and core transmembrane regions. J Gen Physiol, 118(3), 237-250.  Weber, I. T., & Steitz, T. A. (1987). Structure of a complex of catabolite gene activator protein and cyclic AMP refined at 2.5 A resolution. J Mol Biol, 198(2), 311-326.  Whitaker, G. M., Angoli, D., Nazzari, H., Shigemoto, R., & Accili, E. A. (2007). HCN2 and HCN4 isoforms self-assemble and co-assemble with equal preference to form functional pacemaker channels. J Biol Chem.  Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., . . . Wilson, K. S. (2011). Overview of the CCP4 suite and current developments. Acta Crystallogr D Biol Crystallogr, 67(Pt 4), 235-242.  Wiseman, T., Williston, S., Brandts, J. F., & Lin, L. N. (1989). Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal Biochem, 179(1), 131-137.  Wollmuth, L. P., & Hille, B. (1992). Ionic selectivity of Ih channels of rod photoreceptors in tiger salamanders. J Gen Physiol, 100(5), 749-765.  Wolter, S., Golombek, M., & Seifert, R. (2011). Differential activation of cAMP- and cGMP-dependent protein kinases by cyclic purine and pyrimidine nucleotides. Biochem Biophys Res Commun, 415(4), 563-566.  Wu, J. B., S.; Xuong, N.H.; Taylor, S.S. (2004). RIα subunit of PKA: A cAMP-free structure reveals a hydrophobic capping mechanism for docking cAMP into site B. Structure, 12, 1057-1065.  Wu, S., Vysotskaya, Z. V., Xu, X., Xie, C., Liu, Q., & Zhou, L. (2011). State-dependent cAMP binding to functioning HCN channels studied by patch-clamp fluorometry. Biophys J, 100(5), 1226-1232.  Xu, X., Marni, F., Wu, S., Su, Z., Musayev, F., Shrestha, S., . . . Zhou, L. (2012). Local and global interpretations of a disease-causing mutation near the ligand entry path in hyperpolarization-activated cAMP-gated channel . Structure, 20, 2116-2123.  Xu, X., Vysotskaya, Z. V., Liu, Q., & Zhou, L. (2010). Structural basis for the cAMP-dependent gating in the human HCN4 channel. J Biol Chem, 285(47), 37082-37091.  Yanagihara, K., & Irisawa, H. (1980a). Inward current activated during hyperpolarization in the rabbit sinoatrial node cell. Pflügers Arch, 385(1), 11-19.  Yanagihara, K., & Irisawa, H. (1980b). Potassium current during the pacemaker depolarization in rabbit sinoatrial node cell. Pflügers Arch, 388(3), 255-260.  Yang, N., George, A. J., & Horn, R. (1996). Molecular basis of charge movement in voltage-gated sodium channels. Neuron, 16(113-22), 113-122.  Ye, B., & Nerbonne, J. M. (2009). Proteolytic processing of HCN2 and co-assembly with HCN4 in the generation of cardiac pacemaker channels. J Biol Chem, 284, 25553-25559.  213  Yellen, G. (1998). The moving parts of voltage-gated ion channels. Q Rev Biophys, 31(3), 239-295.  Ying, S. W., Tibbs, G. R., Picollo, A., Abbas, S. Y., Sanford, R. L., Accardi, A., . . . Goldstein, P. A. (2011). PIP2-mediated HCN3 channel gating is crucial for rhythmic burst firing in thalamic intergeniculate leaflet neurons. J Neurosci, 31(28), 10412-10423.  Yu, H., Chang, F., & Cohen, I. S. (1995). Pacemaker current i(f) in adult canine cardiac ventricular myocytes. J Physiol, 485 ( Pt 2), 469-483.  Yu, H., Wu, J., Potapova, I., Wymore, R. T., Holmes, B., Zuckerman, J., . . . Wymore, R. (2001). MinK-related peptide 1: A beta subunit for the HCN ion channel subunit family enhances expression and speeds activation. Circ Res, 88(12), E84-87.  Yu, H. G., Lu, Z., Pan, Z., & Cohen, I. S. (2004). Tyrosine kinase inhibition differentially regulates heterologously expressed HCN channels. Pflügers Arch, 447(4), 392-400.  Zagotta, W. N., Olivier, N. B., Black, K. D., Young, E. C., Olson, R., & Gouaux, E. (2003). Structural basis for modulation and agonist specificity of HCN pacemaker channels. Nature, 425(6954), 200-205.  Zagotta, W. N., & Siegelbaum, S. A. (1996). Structure and function of cyclic nucleotide-gated channels. Annu . Rev. Neurosci., 19, 253-263.  Zaza, A., Robinson, R. B., & DiFrancesco, D. (1996). Basal responses of the L-type Ca2+ and hyperpolarization-activated currents to autonomic agonists in the rabbit sino-atrial node. J Physiol, 491 ( Pt 2), 347-355.  Zhong, H., Lai, J., & Yau, K. W. (2003). Selective heteromeric assembly of cyclic nucleotide-gated channels. Proc Natl Acad Sci USA, 100, 5509-5513.  Zhong, H., Molday, L. L., Molday, R. S., & Yau, K. W. (2002). The heteromeric cyclic nucleotide-gated channel adopts a 3A:1B stoichiometry. Nature, 420(6912), 193-198.  Zhou, L., Olivier, N. B., Yao, H., Young, E. C., & Siegelbaum, S. A. (2004). A conserved tripeptide in CNG and HCN channels regulates ligand gating by controlling C-terminal oligomerization. Neuron, 44(5), 823-834.  Zhou, L., & Siegelbaum, S. A. (2007). Gating of HCN channels by cyclic nucleotides: residue contacts that underlie ligand binding, selectivity, and efficacy. Structure, 15(6), 655-670.  Zhou, L., & Siegelbaum, S. A. (2008). Pathway and endpoint free energy calculations for cyclic nucleotide binding to HCN channels. Biophys J, 94(12), L90-92.  Zhou, Y., Morais-Cabral, J. H., Kaufman, A., & MacKinnon, R. (2001). Chemistry of ion coordination and hydration revealed by a K+ channel-Fab complex at 2.0 A resolution. Nature, 414(6859), 43-48.  Zicha, S., Fernandez-Velasco, M., Lonardo, G., L'Heureux, N., & Nattel, S. (2005). Sinus node dysfunction and hyperpolarization-activated (HCN) channel subunit remodeling in a canine heart failure model. Cardiovasc Res, 66(3), 472-481.  214  Zolles, G., Klocker, N., Wenzel, D., Weisser-Thomas, J., Fleischmann, B. K., Roeper, J., & Fakler, B. (2006). Pacemaking by HCN channels requires interaction with phosphoinositides. Neuron, 52(6), 1027-1036.  Zolles, G., Wenzel, D., Bildl, W., Schulte, U., Hofmann, A., Müller, C., . . . Klöcker, N. (2009). Association with the auxiliary subunit PEX5R/Trip8b controls responsiveness of HCN channels to cAMP and adrenergic stimulation. Neuron, 62(6), 814-825.  Zong, X., Eckert, C., Yuan, H., Wahl-Schott, C., Abicht, H., Fang, L., . . . Biel, M. (2005). A novel mechanism of modulation of hyperpolarization-activated cyclic nucleotide-gated channels by Src kinase. J Biol Chem, 280, 34224-34232.  Zong, X., Krause, S., Chen, C. C., Kruger, J., Gruner, C., Cao-Ehlker, X., . . . Biel, M. (2012). Regulation of hyperpolarization-activated cyclic nucleotide-gated (HCN) channel activity by cCMP. J Biol Chem, 287(32), 26506-26512.   215  Appendices  Appendix 1 A schematic diagram of the protein used in the thesis A     B HCN2 (Mus musculus) i. 443-646 DSSRRQYQEKYKQVEQYMSFHKLPADFRQKIHDYYEHRYQGKMFDEDSILGELNGPLREEIVNFNCRKLVASMPLFANADPNFVTAMLTKLKFEVFQPGDYIIREGTIGKKMYFIQHGVVSVLTKGNKEMKLSDGSYFGEICLLTRGRRTASVRADTYCRLYSLSVDNFNEVLEEYPMMRRAFETVAIDRLDRIGKKNSILLHK  HCN4 (Oryctolagus cuniulus) i. 522-725 DSSRRQYQEKYKQVEQYMSFHKLPPDTRQRIHDYYEHRYQGKMFDEESILGELSEPLREEIINFNCRKLVASMPLFANADPNFVTSMLTKLRFEVFQPGDYIIREGTIGKKMYFIQHGVVSVLTKGNKETKLADGSYFGEICLLTRGRRTASVRADTYCRLYSLSVDNFNEVLEEYPMMRRAFETVALDRLDRIGKKNSILLHK  DmIH, isoform B (Drosophila melanogaster) long version i. 1003-1290 short version i. 1003-1217 DSSRRQYREKVKQVEEYMAYRKLPRDMRQRITEYFEHRYQGKFFDEELILGELSEKLREDVINYNCRSLVASVPFFANADSNFVSDVVTKLKYEVFQPGDIIIKEGTIGTKMYFIQEGVVDIVMANGEVATSLSDGSYFGEICLLTNARRVASVRAETYCNLFSLSVDHFNCVLDQYPLMRKTMETVAAERLNKIGKNPNIMHQK|DEQLSNPESNTITAVVNALAAEADDCKDDDMDLKENLLHGSESSIAEPVQTIREGLPRPRSGEFRALFEGNTP  (A) The translated region includes the HMT tag followed by the construct of interest. The red portion represents the hexahistadine tag that binds to cobalt in the Talon affinity column. The blue is maltose binding protein for stabilizing the 3D protein folding in solution and for purifying using amylose affinity column. The green is a specific cleavage site for TEV protease. During purification, the tag is cleaved by TEV protease (see scissors), leaving the C-linker/CNBD region (and SNA tripeptide in the N-terminus) for biochemical assays. The stop codon in the construct of interest terminates translation. Refer to Section 2.2.2 for purification protocol.  (B) The sequences of the constructs used, including the mouse HCN2 C-linker/CNBD, the rabbit HCN4 C-linker/CNBD, and the Drosophila post-S6 segment.   N  SNA  C MBP MGSSHHHHH ENLYF construct of interest  216  Appendix 2 Sets of forward and reverse primer for performing site-directed mutagenesis   A. Site-directed mutagenesis in HCN2 Mutations  Primers Y459A 5’ TTTGTGGAAGGACATGGCTTGCTCTACTTGCTT 3’ 5’ AAGCAAGTAGAGCAAGCCATGTCCTTCCACAAA 3’ Y459E 5’ TTTGTGGAAGGACATCTCTTGCTCTACTTGCTT 3’ 5’ AAGCAAGTAGAGCAAGAGATGTCCTTCCACAAA 3’ K472E 5’ ATAGTAATCGTGGATCTCCTGGCGGAAGTCAGC 3’ 5’ GCTGACTTCCGCCAGGAGATCCACGATTACTAT 3’ Y477F 5’ AAAATCCACGATTACTTTGAACACCGGTACCAA 3’ 5’ TTGGTACCGGTGTTCAAAGTAATCGTGGATCTT 3’ E488A 5’ GGGAAGATGTTTGATGCGGACAGCATCCTTGGG 3’ 5’ CCCAAGGATGCTGTCCGCATCAAACATCTTCCC 3’ E488K 5’ GGGAAGATGTTTGATAAGGACAGCATCCTTGGG 3’ 5’ CCCAAGGATGCTGTCCTTATCAAACATCTTCCC 3’ E488Y 5’ GGGAAGATGTTTGATTACGACAGCATCCTTGGG 3’ 5’ CCCAAGGATGCTGTCGTAATCAAACATCTTCCC 3’ R500Q 5’ GTTCACAATCTCCTCTTGCAGTGGCCCGTTGAG 3’ 5’ CTCAACGGGCCACTGCAAGAGGAGATTGTGAAC 3’ E502K 5’ GTTGAAGTTCACAATCTTCTCACGCAGTGGCCC 3’ 5’ GGGCCACTGCGTGAGAAGATTGTGAACTTCAAC 3’ D542K 5’ CTCTCGGATGATGTATTTTCCAGGCTGGAAGAC 3’ 5’ GTCTTCCAGCCTGGAAAATACATCATCCGAGAG 3’ E582A 5’ GCTCCTATTTCGGGGCGATCTGCTTGCTCACG 3’ 5’ CGTGAGCAAGCAGATCGCCCCGAAATAGGAGC 3’ R591A 5’ GCTCACGAGGGGCCGGGCTACGGCCAGCGTGCGAGC 3’ 5’ GCTCGCACGCTGGCCGTAGCCCGGCCCCTCGTGAGC 3’ T592A 5’ CGAGGGGCCGGCGTGCGGCCAGCGTGCGAGC 3’ 5’ GCTCGCACGCTGGCCGCACGCCGGCCCCTCG 3’ I630A 5’ TTTGAGACTGTGGCTGCTGACCGGCTAGATCGC 3’ 5’ GCGATCTAGCCGGTCAGCAGCCACAGTCTCAAA 3’ D631A 5’ GAGACTGTGGCTATTGCCCGGCTAGATCGCATA 3’ 5’ TATGCGATCTAGCCGGGCAATAGCCACAGTCTC 3’ R632A 5’ CTGTGGCTATTGACGCGCTAGATCGCATAGGC 3’ 5’ GCCTATGCGATCTAGCGCGTCAATAGCCACAG 3’ L633A 5’ GTGGCTATTGACCGGGCAGATCGCATAGGCAAG 3’ 5’ CTTGCCTATGCGATCTGCCCGGTCAATAGCCAC 3’ D634A 5’ GCTATTGACCGGCTAGCTCGCATAGGCAAGAAG 3’ 5’ CTTCTTGCCTATGCGAGCTAGCCGGTCAATAGC 3’ R635A 5’ ATTGACCGGCTAGATGCCATAGGCAAGAAGAAC 3’ 5’ GTTCTTCTTGCCTATGGCATCTAGCCGGTCAAT 3’ I636A 5’ GACCGGCTAGATCGCGCAGGCAAGAAGAACTCC 3’ 5’ GGAGTTCTTCTTGCCTGCGCGATCTAGCCGGTC 3’ I636D 5’ GACCGGCTAGATCGCGACGGCAAGAAGAACTCC 3’ 5’ GGAGTTCTTCTTGCCGTCGCGATCTAGCCGGTC 3’ 217  K638A 5’ CTAGATCGCATAGGCGCGAAGAACTCCATCTTG 3’ 5’ CAAGATGGAGTTCTTCGCGAATATGCGATCTAG 3’  B. Site-directed mutagenesis in HCN4 Mutations Primers K530N 5’ CGCCAGTACCAGGAGAACTACAAGCAGGTGGAG 3’ 5’ CTCCACCTGCTTGTAGTTCTCCTGGTACTGGCG 3’ D554N 5’ CGCCAGCGCATCCACAACTACTACGAGCACCGG 3’ 5’ CCGGTGCTCGTAGTAGTTGTGGATGCGCTGGCG 3’ S672R 5’ GGCCGGCGCACGGCCCGCGTGCGGGCCGACACC 3’ 5’ GGTGTCGGCCCGCACGCGGGCCGTGCGCCGGCC 3’  C. Ligation-independent cloning for DmIH Construct Primers DmIH (long) 5’ TACTTCCAATCCAATGCAGACTCCAGCCGGCGCCAGTAT 3’ 5’ TTATCCACTTCCAATGTTATTATGGAGTGTTACCCTCGAACAA 3’ DmIH (short) 5’ TACTTCCAATCCAATGCAGACTCCAGCCGGCGCCAGTAT 3’ 5’ TTATCCACTTCCAATGTTATTACTTCTGATGCATTATGTTTGG 3’  The DNA sequences used were based on mouse HCN2 and rabbit HCN4 gene. All primers were ordered from Integrated DNA Technologies and diluted to a stock of 1 mg/µL with autoclaved water. (A/B)  The mutagenesis protocol is outlined in Section 2.2.1. (C) The DmIH sequence was cloned into pET28 vector containing HMT by LIC. The primers add a complementary sequence that matches with the vector to the insert (gene of interest). The vector and insert are treated with T4 DNA polymerase which exhibits 3’ to 5’ exonuclease activity to create a complementary overhang on either side of the fragments. The annealing at the long overhang is sufficient to hold the pieces together without the ligating process.    Appendix 3 All nine mutations from Chapter Two allow ligand-induced oligomerization  218   Plots of apparent molecular weight versus concentration of respective mutant C-linker/CNBD subunits in solution are shown. The ability to oligomerize for these mutations was proportional to the size of the globular protein measured by DLS. All nine mutants can self-associate at higher subunit concentrations. Ten-fold cAMP or cGMP was added to C-linker/CNBD to trigger oligomerization at lower subunit concentrations.  All mutations responded to both ligands, suggesting that both ligands can trigger the formation of a gating ring and allowing the same maximal shift to the activation curve. Error bars show s.e.m. 219  Appendix 4 Cyclic AMP and cGMP have extremely low affinity after introduction of E582A and R632A  These individual mutations could not be purified in substantial concentrations for biochemical experiments. Instead, these mutations were tested with the tag intact because MBP helps protein stabilization. 200 µM of the fusion mutant protein (HMT + C-linker/CNBD) was inserted to the sample cell of the ITC machine, and the program ran with slow injections of cAMP and cGMP. E582A has very weak affinity for both ligands, and R632A does not bind to the ligands at all. The data is limited to the sensitivity of the technique when the binding is weak. The data is consistent with low potency of the two ligands. 220  Appendix 5 Effect of cyclic nucleotides from Chapter Three on oligomerization of the HCN2 C-linker/CNBD as function of ligand concentration   Plots of estimated average molecular weight versus concentration of ligand are shown. The concentration of purified HCN2 C-terminus protein used for this study was 50 µM. The estimated molecular weight was determined by Dynamic Light Scattering and is proportional to the population of oligomers in solution. An increased in oligomerization occurs with increasing concentrations of cAMP, cGMP, cUMP, cPuMP and 2-NH2-cPuMP, but not cCMP and cIMP. The experiment rules out the dependence on concentration and the weaker binding of cCMP and cIMP, and the lack of oligomerization is consistent with the insufficient stabilization of an active gating ring, leading to partial agonism. Error bars show s.e.m., determined from three separate preparations of purified protein measured on different days.      221  Appendix 6 Omit maps of the five co-crystal structures in Chapter Three   Ribbon representation of HCN2 C-linker/CNBD as Figure 3.6. The residues are those that make contacts with the ligand. The red electron density depicts the Fo-Fc map, with sigma cut-off of 3.0, showing the missing density if refinement were performed without inserting the ligand. The red density confirms the docking presence and orientation of the ligand. 222  Appendix 7 Certain intersubunit interactions abolished by mutation shows reduced ligand-induced oligomerization 223  An intersubunit and intrasubunit salt bridge was found that when broken, the effect of cAMP is mimicked and channel opening is triggered. The same mutations were attained, and ITC and DLS experiments were performed on them.  ITC measures the affinity and the presence of cooperate binding. DLS measures the change in molecular weight by [C-linker/CNBD] or cAMP. The collaborative experiments show strong correlation between the presence of cooperative binding and ligand-induced oligomerization.  (A/B) D542 and K472 form an intracellular bond. Both single mutations are able to abolish ligand-induced tetramerization.  (C) Single substitutions E502K did not abolish ligand-induced oligomerization.  (D) Tetramerization can be rescued by a double mutant.    224   (E) To reinforce the fact that the epilepsy-associated mutation interferes with Y459 from the neighbouring subunit. As shown in Chapter Four already, the alanine substitutions for both residues eliminated the effect of ligand on oligomerization. Here, the rescue construct was made. The rescue (E488Y/Y459E) does not restore oligomer formation. This could be due to changing local orientation to deny hydrogen bond access. (F) Another epilepsy mutation, R500Q in HCN2 (R527Q in human), also discourages oligomer formation induced by cAMP. The biochemical results are very similar to that of E488K, showing no cooperative binding and no oligomer formation upon adding cAMP. The hampered effect of cAMP on the mutant could explain the disease phenotype. (G) The crystal structure shows another intersubunit hydrogen bond between the backbone of I491 and the side chain hydroxyl of Y477. The Y477F mutation, where the phenyl ring is conserved for local hydrophobic partners, behaves much like wild type. This hydrogen bond is not involved in activation of the C-linker.  225  Appendix 8 Characterization of the effects of analogues on the HCN2 binding domain      226   (A) Rp-cAMP or Rp-cGMP substitutes the equatorial oxygen with a sulphur atom and limits its interaction with the PBC. The experiment involves 2 mM of respective ligands titrating into 200 µM purified HCN2 C-linker/CNBD. (All the titrations in this appendix were done using the same protein/ligand concentrations.) Its normal bond with R591 is crucial for high affinity binding, explaining why both analogues bind with significantly weaker affinity, consistent with R591A mutation.  In the DLS experiment, it appears that 10-fold (compared to [C-linker/CNBD]) of Rp-cAMP and Rp-cGMP (red circles) do not promote tetramerization. However, a high dose (8mM) of Rp-cAMP (red triangles) was added to 100 and 200 µM of protein and higher molecular weights were observed. Therefore, the oligomerizing ability was hidden due to the low affinity.  Whole-cell configuration patch-clamp experiment with Rp-cAMP in pipette solution shows the same rightward shift of 17.21mV. (B) Sp-cAMP or Sp-cGMP substitutes the axial oxygen with a sulphur atom and limits its interaction with T592 of HCN2 C-linker/CNBD. The affinity is more severely affected by Sp-cGMP, but is weaker than that of cAMP, consistent with the T592A mutation. 10-fold Sp-cAMP (green circles) promotes tetramerization but Sp-cGMP caused aggregation of the protein which interfered with DLS readings. Patch-clamp experiment with Sp-cAMP shows a full shift of 11.77 mV.         227   (C) 2’-OMe-cAMP blocks the 2’ hydroxyl group from the ribose moiety from forming a hydrogen bond with the carboxylate group of E582. The small heats from the top curve and the flat integrated trend on the bottom suggest that cAMP no longer binds to the CNBD. This is consistent with the greatly weakened binding in E582A mutation. (D) Another cyclic pyrimidine nucleotide, cTMP, shows no binding to the C-linker/CNBD. Due to the structural similarities between cUMP and cTMP, we speculate that the extra 5-methyl group in cTMP causes steric clashes with the binding pocket 228   (E) 8-bromo derivatives (8-Br-cAMP or 8-Br-cGMP) are constitutively in the syn configuration due to the steric hindrance of the bulky bromine group with the ribose ring. Both ligands bind better than cAMP shown by ITC. Negative cooperativity is retained, and so is induced oligomerization. The results imply that the configuration of the ligand does not determine the strength of binding in HCN channels.  (F) 6-Cl-cPuMP changes the interaction between the 6th position consistent of the purine ring and the side chain of R632. The position normally holds a –NH2 group in cAMP and a keto group in cIMP. The affinity shows a trend where binding favours –NH2 group, followed by –Cl and –keto.  This suggests the interaction with R632 residue depends on the electronegativity of its partner.    229    (G) 7-CH-cAMP replaces the ring nitrogen with a carbon, thereby decreasing the solubility and changing the electron distribution in the ring. The analogue was found to be very potent, and from the ITC data, the high potency is due to high affinity to the binding domain. The strong binding is apparent by the steep transition on the bottom curve. (H) Ivabradine is a specific drug that targets HCN channels. The binding site for ivabradine is still unknown. The ITC data shows the heat is very small and is integrated into a flat line, indicating ivabradine does not bind to the C-linker or the CNBD, or it binds too weakly to be detected by the instrument. (I) Adenine, without the ribose and phosphate moiety, does not bind to the CNBD. The heat shown on top is similar to that in the negative control, and after subtraction, the isotherm shows a flat line. This is consistent with the general belief that the strongest binding occurs between the protein and the equatorial and 2’OH oxygens. 230  Appendix 9 Mutations in HCN4 binds responds to ligand normally     231   Two other mutations were found in the C-linker/CNBD of HCN4 that are associated with bradycardia. They are found in C-linker and have slightly weaker affinity compared to wild type. Both mutations retain negative cooperativity as in wild type. Both mutants are able to promote tetramerization in the presence of cAMP or cGMP. The binding data contradicts with the functional data of D553N because the β-adrenergic stimulation in the cell was abolished. This means the mutation does not change the binding site of the ligand, and instead, affects a post-binding event. The DLS data for K530N suggests that the ability to tetramerize is normal, so the hyperpolarized shift and the bradycardia phenotype could be due to a post-C-linker activation event. Both of these mutations could be similar to S672R, where the gating ring is formed, but the transitional conformational change to the pore is affected.      232  Appendix 10 The Drosophila isoform C-terminus bind with one binding   The Drosophila isoform does not show cooperative binding. The HCN homologue was also subcloned and expressed in the same vector. There is higher internal stability in the construct, which allows the purification of the C-linker/CNBD and the entire C-terminus.  (A) The C-linker/CNBD (also abbreviated as ‘s’ for shortened fragment), or the same C-linker/CNBD segment tested in mammalian isoforms, shows no cooperative binding. The two DLS graphs both support the fact that the short segment does not tetramerize by the addition of cAMP or cGMP.  The left curve shows no effect on different concentrations of protein when 10x ligand was added. On the right, 12.5µM of protein with addition of significantly higher ligand concentration also shows no difference in the apparent molecular weight. The absence of tetramerization is independent of ligand concentration. Cyclic GMP has very weak binding comparing to cAMP. Values of binding affinity and thermodynamic parameters can be found in Appendix 14.  233    (B) The longer version (abbreviated as ‘l’ in the figure) includes the C-linker, CNBD, and the remaining of the C-terminus. Refer to Appendix 1B for the sequence. The longer version also binds to cAMP, but has weaker affinity compared to the short version. The C-terminus extension could interfere with cAMP entrance to the binding pocket. Cyclic GMP also binds more weakly than cAMP. This is because the analogue position of T592, which binds to the phosphate moiety, is naturally a valine. The absence of the bond explains the weak binding. DLS experiments did not give consistent molecular weights. We propose the extension is more flexible in solution, giving big fluctuations in the scattering wave. Values of binding affinity and thermodynamic parameters can be found in Appendix 14.   234  Appendix 11 Compilation of ITC data from cAMP binding to various HCN2 mutations   high affinity low affinity  ΔBasal activation from wt (Δ ΔV1/2) Δligand shift from wt (Δ ΔVmax)  Kd  ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N Wild type  (ch 2) 0.11 ± 0.02 -6.21 ± 1.02 -3.31 ± 0.97 -9.53 ± 0.12 0.34 ± 0.02 1.53 ± 0.16 -16.59 ± 0.92 8.63 ± 0.86 -7.96 ± 0.07 0.66 ± 0.06 4   Tetramerization into gating ring R500Q 3.21 ± 0.35 -7.65 ± 0.45 0.15 ± 0.48 -7.50 ± 0.06 0.77 ± 0.13      3 2.0 ± 2.8 -0.1 ± 0.58 E488K 2.56 ± 0.17 -7.93 ± 0.15 0.29 ± 0.15 -7.63 ± 0.04 0.83 ± 0.05      6 -37.6 ± 3.0 0.1 ± 4.6 E488A 2.43  ± 0.37 -8.14  ± 0.77 0.49  ± 0.78 -7.65  ± 0.07 0.90  ± 0.04      4 / / Y459A 7.29 ± 0.61 -6.63 ± 0.13 -0.39 ± 0.12 -7.02 ± 0.05 1.05 ± 0.04      5 / / E488Y 3.42 ±  0.10 -6.94 ±  0.06 -0.52 ±  0.08 -7.46 ±  0.02 0.66 ±  0.01      2 / / Y459E 2.67 ±  0.04 -8.99 ±  0.13 0.12 ±  1.38 -8.86 ±  1.25 0.72 ±  0.01      2 / / E488Y-Y459E 2.61 ±  0.03 -7.63 ±  0.06 0.01 ±  0.05 -7.62 ±  0.01 0.80 ±  0.05      2 / / Y477F 0.22  ± 0.04 -6.61  ± 0.39 -2.48  ± 0.42 -9.09  ± 0.10 0.51  ± 0.04 2.73  ± 0.24 -15.11  ± 0.64 7.52  ± 0.60 -7.59  ± 0.05 0.58  ± 0.06 3 / / Salt bridge into activation K472E 3.37 ± 0.14 -8.29 ± 0.08 0.83 ±  0.10 -7.46 ± 0.02 0.45 ± 0.07      2 12.0 ± 3.7 -13.6 ± 2.2 E502K 0.05 ± 0.02 -4.75 ± 1.00 5.29 ± 0.80 -10.05 ± 0.21 0.18 ± 0.02 0.95 ± 0.27 -22.32 ± 4.54 14.07 ± 4.40 -8.226 ± 0.16 0.78 ± 0.17 3 -3.0 ± 3.5 -2.0 ± 2.0 D542K 3.01 ± -7.78 ± 0.20 ± -7.57 0.78 ±      7 1.0 ± 2.8 -4.2 ± 2.3 235   high affinity low affinity  ΔBasal activation from wt (Δ ΔV1/2) Δligand shift from wt (Δ ΔVmax)  Kd  ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N 0.45 0.54 0.46 ± 0.09 0.06 E502K-K472E 0.24 ± 0.06 -7.29 ± 2.1 -1.79 ± 1.92 -9.08 ± 0.19 0.27 ± 0.06 1.89 ± 0.45 -19.76 ± 3.05 11.91 ± 2.91 -7.85 ± 0.15 0.31 ± 0.06 3 0 ± 4.3 0.6 ± 2.4 Mapping the binding pocket E582A HMT 334.83 ± 26.18 -18.80 ± 9.52 14.05 ± 9.46 -4.75 ± 0.06 0.52 ± 0.04      2 -2.5 ± 1.0 -2.9 ± 0.6 R591A 1.65 ± 0.57 -6.19 ± 0.74 -1.78 ± 0.91 -7.97 ± 0.24 0.35 ± 0.02 33.35 ± 5.58 -15.32 ± 0.79 9.20 ± 0.86 -6.12 ± 0.11 0.70 ± 0.09 3 -3.5 ± 1.6 0.9 ± 0.6 T592A 0.61 ± 0.25 -7.46 ± 0.82 -1.20 ± 0.53 -8.66 ± 0.37 0.23 ± 0.01 11.83 ± 4.37 -20.52 ± 1.55 13.72 ± 1.36 -6.81± 0.21 0.56 ± 0.06 3 -2.0 ± 1.7 -0.7 ± 0.8 I630A 0.19 ± 0.06 -2.71 ± 1.03 -6.57 ± 0.93 -9.28 ± 0.25 0.28± 0.03 3.10 ± 0.74  -14.43 ± 1.98 6.85 ± 1.92 -7.59 ± 0.17 0.69 ± 0.07 4 -0.4 ± 2.1 -2.1 ± 0.4 D631A 0.11 ± 0.01 -7.76 ± 0.97 -1.76 ± 0.94 -9.52 ± 0.07 0.21 ± 0.02 1.90 ± 0.27 -18.38 ± 2.14 10.56 ± 2.06 -7.82 ± 0.10 0.56 ± 0.04 3 -4.0 ± 1.2 0.2 ± 2.1 R632A HMT  no binding      2 -1.0 ± 1.2 -18.1 ± 0.7 L633A 1.51 ± 0.31 -2.37 ± 0.23 -5.63 ± 0.26 -7.99 ± 0.13 0.43 ± 0.05 24.67 ± 2.94 -8.41 ± 0.68 2.10 ± 0.61 -6.31 ± 0.08 0.66 ± 0.07 5 -2.4 ± 1.1 -4.3 ± 0.7 D634A 0.10 ± 0.03 -4.92 ± 0.65 -4.80 ± 0.46 -9.72 ± 0.23 0.26 ± 0.03 1.39 ± 0.46 -15.94 ± 1.01 7.89 ± 0.81 -8.05 ± 0.22 0.81 ± 0.03 5 0.1 ± 1.4 -1.7 ± 0.5 R635A 0.20 ± 0.06 -8.07 ± 0.29 -1.12 ± 0.18 -9.19 ± 0.18 0.33 ± 0.03 2.19 ± 0.40 -15.15 ± 0.02 7.42 ± 0.11 7.73 ± 0.10 0.45 ± 0.02 3 -1.4 ± 1.3 1.9 ± 0.9 I636A 0.47 ± 0.04 -2.39 ± 0.12 -6.24 ± 0.09 -8.63 ± 0.05 0.36 ± 0.01 6.94 ± 0.31 -13.39 ± 0.40 6.35 ± 0.40 -7.04 ± 0.02 0.60 ± 0.03 4 -1.9 ± 1.7 1.3 ± 1.0 I636D 1.20 ± 0.51 -3.09 ± 0.55 -5.16 ± 0.41 -8.25 ± 0.26 0.56 ± 0.07 17.16 ± 2.34 -6.11 ± 0.86 -0.41 ± 0.77 ;-6.52 ± 0.08 0.82 ± 0.02 4 -3.4 ± 1.3 0.2 ± 0.8 K638A 0.31 ± 0.15 -5.14 ± 0.84 -3.97 ± 1.07 -9.11 ± 0.33 0.38 ± 0.04 8.00 ± 2.48 -14.48 ± 1.91 7.43 ± 1.73 -7.05 ± 0.21 0.50 ± 0.03 4 -1.3 ± 1.5 0.9 ± 1.1   All of the data was obtained from the fit (Origin 7.0) generated by the heat from 40 progressive injections of 2 mM cAMP into 200 µM mutant 236  HCN2 C-linker/CNBD. Mutations where only one binding event was observed are reported under “high affinity” and the second event is left blank. The last columns summarize the change in V1/2 (basal activation) or change in Vmax (cAMP modulation) comparing wild type of each respective mutant. Negative value is a hyperpolarized shift. Most cases are considered negligible changes. The values in red are the ones to focus. Kd in µM while ΔH, TΔS and ΔG in kcal/mol. Backslash (/) means the electrophysiological data was not collected. Uncertainties in s.e.m.    237  Appendix 12 Compilation of ITC data from cGMP binding to various HCN2 mutations   high affinity low affinity ΔBasal activation from wt (Δ ΔV1/2) Δligand shift from wt (Δ ΔVmax)  Kd ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N Wild type  (ch 2) 0.3 ± 0.04 -6.64 ± 0.57 -2.27 ± 0.61 -8.91 ± 0.07 0.29 ± 0.05 6.75 ± 0.39 -18.60 ± 3.92 11.54 ± 3.90 -7.06 ± 0.03 0.39 ± 0.07 4   Tetramerization into gating ring R500Q  6.06 -10.92 3.78 -7.14 0.381      1 / / E488K 6.46 ± 0.49 -6.24 ± 0.10 -0.85 ± 0.08 -7.09 ± 0.04 0.83 ±  0.05       6 / / E488A 9.44  ± 0.27 -4.21  ± 0.08 -2.65  ± 0.07 -6.85  ± 0.02 1.06  ± 0.01      2 / / Y459A  8.92 ± 1.10 -6.90 ± 0.65 -0.003 ± 0.60 -6.90 ± 0.08 0.55 ± 0.02      3 / / Y477F  0.51 -4.04 -4.53 -8.57 0.57 13.79 -6.08 -0.55 -6.63 0.5 1 / / Mapping the binding pocket E582A HMT  689.65 -43.08 38.74 -4.34 0.052      1 -2.5 ±  1.0 -6.6 ± 6.4 R591A 2.15E5 ± 9.20E6 321.71 ± 26.72 -3.09E5 ± 1.45E5 3.10E5 ± 1.46E5 468 ± 472      2 -3.5 ± 1.3 -0.4 ± 0.6 T592A 183.53 ± 2.69 -20.72 ± 3.63 15.62 ± 3.64 -5.10 ± 0.01 0.20 ± 0.03      2 -2.0 ± 2.5 -4.7 ± 0.8 I630A 0.49 ± 0.11 -3.01 ± 0.99 -5.63 ± 1.12 -8.64 ± 0.13 0.20 ± 7.78 ± 0.31 -8.16 ± 1.80 1.17 ± 1.78 -6.99 ± 0.03 0.43 ± 0.07 3 -0.4 ± 2.1 -1.2 ± 1.1 238   high affinity low affinity ΔBasal activation from wt (Δ ΔV1/2) Δligand shift from wt (Δ ΔVmax)  Kd ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N 0.03 D631A 0.20 ± 0.06 -5.38 ± 0.57 -3.83 ± 0.53 -9.22 ± 0.17 0.20 ± 0.02 4.78 ± 0.88 -18.45 ± 5.27 11.13 ± 5.14 -7.32 ± 0.13 0.63 ± 0.14 5 -4.0 ± 1.2 -0.1 ± 0.9 R632A HMT  no binding      4 -1.0 ± 1.2 -15.4 ± 0.7 L633A 25.27 ± 3.77 -3.50 ± 0.81 -2.81 ± 0.77 -6.30 ± 0.08 1.08 ± 0.21      4 -2.4 ± 1.1 -9.3 ± 0.8 D634A 0.54 ± 0.10 -0.71 ± 0.46 -7.96 ± 0.34 -8.67 ± 0.13 0.23 ± 0.04 8.04 ± 1.32 15.23 ± 1.89 8.24 ± 1.82 -6.99 ± 0.12 0.87 ± 0.14 5 0.1 ± 1.3 -2.7 ± 1.2 R635A 4.61 ± 0.16 -6.00 ± 0.86 -1.28 ± 0.89 -7.28 ± 0.02 0.67 ± 0.09      4 -1.4 ± 1.3 -2.0 ± 0.6 I636A 0.24 ± 0.02 -2.88 ± 0.32 -6.13 ± 0.29 -9.02 ± 0.04 0.23 ± 0.01 3.23 ± 0.15 -21.47 ± 0.67 13.96 ± 0.65 -7.51 ± 0.03 0.44 ± 0.02 4 -1.9 ± 1.7 2.8 ± 0.9 I636D 0.21 ± 0.07 -5.46 ± 1.45 -3.87 ± 1.33 -9.40 ± 0.11 0.18 ± 0.04 1.76 ± 0.51 -22.99 ± 0.19 15.23 ± 0.29 -7.99 ± 0.25 0.56 ± 0.08 4 -3.4 ± 1.3 5.5 ± 0.6 K638A 14.69 ± 2.04 -13.69 ± 4.38 6.89 ± 4.14 -6.80 ± 0.24 0.47 ± 0.09      3 -1.3 ± 1.5 -0.7 ± 0.6  As Appendix 11, the values are generated by ITC data from titration of 2 mM cGMP into 200 µM of HCN2 C-linker/CNBD. Fits with single binding model are reported under “high affinity” and the “low affinity” columns are left blank. The change in basal activation and cAMP modulation are summarized from literature. The only significant difference is the effect of cGMP on R632A mutant. Kd in µM while ΔH, TΔS and ΔG in kcal/mol. Uncertainties in s.e.m. 239  Appendix 13 Compilation of ITC data from cAMP or cGMP binding to wild type or mutated HCN4 C-linker/CNBD  high affinity low affinity    Kd ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N with cAMP Wild type 0.06 ± 0.02 -7.64 ± 0.66 -2.36± 0.66 -10.00 ± 0.17 0.11 ± 0.02 0.69 ± 0.10 -14.30 ± 0.86 5.86 ± 0.78 -8.44 ± 0.09 0.60 ± 0.04 7 K530N 0.27 ± 0.08 -4.72 ± 0.19 -4.32 ± 0.14 -9.04 ± 0.15 0.34 ± 0.05 4.02 ± 2.03 -11.12 ± 1.55 3.57 ± 1.80 -7.55 ± 0.24 0.49 ± 0.02 4 D553N 0.15 ± 0.01 -2.18 ± 0.36 -7.16 ±0.34 -9.34 ± 0.06 0.24 ± 0.01 1.22 ± 0.17 -13.16 ± 0.49 5.07 ± 0.43 -8.09 ± 0.08 0.73 ± 0.03 5 S672R 0.44 ± 0.06 -3.83 ± 0.98 -4.81 ± 0.90 8.64 ± 0.11 0.15 ± 0.01 5.05 ± 0.56 -16.60 ± 0.60 9.34 ± 0.61 -7.25 ± 0.07 0.85 ± 0.04 8 with cGMP Wild type 0.16 ± 0.02 -4.94 ± 0.61 -4.35 ± 0.61 -9.29 ± 0.05 0.14 ± 0.01 1.74 ± 0.23 -12.52 ± 1.00 4.64 ± 0.95 -7.89 ± 0.07 0.64 ± 0.05 6 K530N 0.50 ± 0.20 -3.48 ± 0.37 -5.17 ± 0.13 -8.65 ± 0.24 0.38 ± 0.03 6.93 ± 1.01 -8.19 ± 0.48 1.15 ± 0.40 -7.04 ± 0.09 0.38 ± 0.02 2 D553N 0.32 ± 0.09 -1.80 ± 0.51 -7.08 ± 0.34 -0.88 ± 0.16 0.24 ± 0.01 2.95 ± 0.23 -14.00 ± 0.91 6.45 ± 0.94 -7.54 ± 0.03 0.57 ± 0.01 2 S672R 0.64 ± 0.17 -4.55 ± 0.60 -3.81 ± 0.59 -8.36 ± 0.18 0.19 ± 0.01 9.95 ± 0.58 -19.03 ± 2.84 12.19 ± 2.85 -6.83 ± 0.04 0.85 ± 0.16 5  As Appendix 11, the values are generated by ITC data from titration of 2 mM cAMP or cGMP into 200 µM of HCN4 C-linker/CNBD. Kd in µM while ΔH, TΔS and ΔG in kcal/mol. Uncertainties in s.e.m.   240  Appendix 14 Compilation of ITC data from cAMP or cGMP in the Drosophila isoform  high affinity low affinity    Kd ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N with cAMP DmIH  (long) 2.87 ± 0.71 -7.38 ± 0.46 -0.26 ± 0.49 -7.64 ± 0.16 0.34 ± 0.04      5 DmIH (short) 0.78 ± 0.19 -10.67 ± 0.16 2.25 ± 0.34 -8.42 ± 0.21 0.30 ± 0.01      4 with cGMP DmIH  (long) 57.48 ± 0.66 -4.17 ± 0.31 -1.61 ± 0.30 -5.78 ± 0.01 0.27 ± 0.05      2 DmIH (short) 34.58  ± 1.78 -7.44  ± 1.15 -7.44  ± 1.15 1.36  ± 1.16 0.25  ± 0.04      4  As Appendix 11, the values are generated by ITC data from titration of 2 mM cAMP or cGMP into 200 µM of the Drosophila constructs. The short isoform contains the C-linker and the CNBD, just like the C-linker/CNBD from mammalian isoforms. The long isoform contains the C-linker, the CNBD, as well as the section distal to the CNBD to the end. Kd in µM while ΔH, TΔS and ΔG in kcal/mol. Uncertainties in s.e.m.   241  Appendix 15 : Compilation of ITC data from various ligands binding to HCN2 C-linker/CNBD   high affinity low affinity  CH   Kd ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N 3 cAMP  0.13 ± 0.03 -6.12 ± 0.43 -3.31 ± 0.25 -9.43 ± 0.20 0.28 ± 0.05 1.83 ± 0.21 -16.77 ± 1.34 8.78 ± 1.26 -7.99 ± 0.09 0.63 ± 0.03 4 3 cGMP  0.43 ± 0.08 -4.69 ± 0.71 -3.95 ± 0.63 -8.64 ± 0.12 0.36 ± 0.03 8.53 ± 0.70 -11.77 ± 0.97 4.71 ± 0.95 -7.07 ± 0.09 0.67 ± 0.09 4 3 cCMP 20.23 ± 0.66 -2.94 ± 0.10 -3.46 ± 0.12 -6.40 ± 0.02 1.23 ± 0.01      3 3 cIMP 39.79 ± 5.89 -4.50 ± 0.23 -1.55 ± 0.31 -6.04 ± 0.09 0.98 ± 0.05      7 3 cPuMP 0.17 ± 0.01 -2.24 ± 0.58 -7.00 ± 0.58 -9.25 ± 0.05 0.17 ± 0.01 2.92 ± 0.23 -18.98 ± 0.36 11.41 ± 0.32 -7.56 ± 0.05 0.58 ± 0.06 5 3 cUMP 1.33 ± 0.17 -1.24 ± 0.50 -6.79 ± 0.52 -8.03 ± 0.07 0.36 ± 0.03 23.7 ± 1.82 -10.78 ± 0.71 4.47 ± 0.75 -6.31 ± 0.04 0.73 ± 0.01 3 3 cTMP no binding       2 3 2NH2cPuMP 0.31 ± 0.06 -3.45 ± 0.69 -5.44 ± 0.62 -8.90 ± 0.13 0.21 ± 0.03 3.12 ± 0.26 -19.97 ± 2.51 12.45 ± 2.55 -7.52 ± 0.04 0.47 ± 0.04 3 5 6-Cl-cPuMP 0.54 ± 0.17 -0.39 ± 0.16 -8.22 ± 0.09 -8.61 ± 0.16 0.33 ± 0.02 10.37 ± 0.88 -9.41 ± 1.18 2.60 ± 1.20 -6.81 ± 0.05 0.72 ± 0.05 4 1 2’ OMe-cAMP no binding      2 5 8-Br-cAMP 0.05 ± 0.01 -6.72 ± 0.40 -3.28 ± 0.33 -10.00 ± 0.07 0.43 ± 0.10 0.61 ± 0.27 -14.31 ± 0.91 5.72 ± 0.67 -8.59 ± 0.24 0.45 ± 0.02 3 5 8-Br-cGMP 0.80 ± 0.27  -1.93 ± 0.41 -6.48 ± 0.53 -8.41 ± 0.26 0.26 ± 0.03 3.58 ± 1.14 -5.97 ± 0.60 -1.51 ± 0.74 -7.48 ± 0.17 0.70 ± 0.12 3 5 Rp-cAMP 59.43 ± 1.77 -6.96 ± 0.25 1.19 ± 0.26 -5.77 ± 0.02 0.83 ±  0.09       6 5 Rp-cGMP 46.34 ±  2.45 -6.96 ±  0.83 -0.08 ±  0.24 -7.04 ±  0.59 0.92 ±  0.14      3 5 Sp-cAMP 0.53 ± 0.11 -5.52 ± 0.40 -3.07 ± 0.31 -8.58 ± 0.12 0.39 ± 0.01 14.22 ± 0.21 -10.93 ± 1.46 4.32 ± 1.47 -6.61 ± 0.01 0.61 ± 0.02 3 5 Sp-cGMP 24.38 ±  5.64 ±  -0.50 ±  6.14 ±  0.79 ±       3 242    high affinity low affinity  CH   Kd ΔH -TΔS ΔG n Kd ΔH -TΔS ΔG n N 2.30 0.33 0.41 0.21 0.05 A adenine no binding       2 A ivabradine no binding       2  As Appendix 11, the values are generated by ITC data from titration of 2 mM respective ligand to 200 µM HCN2 C-linker/CNBD. The CH refers to which chapter the particular ligand has been mentioned. Kd in µM while ΔH, TΔS and ΔG in kcal/mol. Uncertainties in s.e.m.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0229565/manifest

Comment

Related Items