Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Molecular recognition by genetically engineered myoglobins Hunter, Christie Lynne 1997

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

Item Metadata

Download

Media
831-ubc_1997-195953.pdf [ 8.11MB ]
Metadata
JSON: 831-1.0088272.json
JSON-LD: 831-1.0088272-ld.json
RDF/XML (Pretty): 831-1.0088272-rdf.xml
RDF/JSON: 831-1.0088272-rdf.json
Turtle: 831-1.0088272-turtle.txt
N-Triples: 831-1.0088272-rdf-ntriples.txt
Original Record: 831-1.0088272-source.json
Full Text
831-1.0088272-fulltext.txt
Citation
831-1.0088272.ris

Full Text

MOLECULAR RECOGNITION BY GENETICALLY ENGINEERED MYOGLOBINS by CHRISTIE L Y N N E HUNTER B.Sc, The University of Calgary, 1989 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF M O L E C U L A R BIOLOGY AND BIOCHEMISTRY We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA March 1997 © Chri stie Lynne Hunter, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of fh f y / The University of British Columbia Vancouver, Canada Date DE-6 (2/88) ABSTRACT Specific formation of non-covalently associated complexes between biomolecules is a phenomenon of fundamental importance in biology. The present study examines a variety of myoglobin variants that have been designed as models of such "molecular recognition" processes. Myoglobin (Mb) has been selected for this purpose because it is a small (17.6 kDa) protein with a non-covalently bound heme prosthetic group that can bind dioxygen and other ligands reversibly and because recombinant Mbs can be expressed efficiently in high yield. As summarized below, three types of molecular recognition processes have been considered: (a) the interaction of divalent metal ions with a metal binding site that has been introduced on the surface of the protein near the partially exposed heme edge, (b) the contribution of the hydrogen bonds formed between the heme propionate groups and protein residues to the stabilization of the interaction of heme with apoMb, and (c) the influence of amino acid residues in the active site of the protein on the ability of the heme iron to bind azide. As part of this work, relevant spectroscopic and electrochemical properties of the variants involved have also been considered. A binding site for metal ions was engineered on the surface of horse Mb near the heme 6-propionate, similar to the site present in the manganese peroxidase enzyme produced by the white-rot fungus, Phanearochaete chrysosporium. The equilibrium constant for Mn(II) binding to engineered site was determined to be 1.3 x 104 M ' 1 (pH 7.0, 25 °C, 1=17.2 mM), and the affinity was shown to decrease at lower pH and higher ionic strength. A low level of oxidation of Mn(II) to Mn(III) in the presence of H 2 0 2 was observed in the variants of Mb possessing enhanced metal binding. Other metal ions were shown to bind with varying affinities to this engineered site, Cd(II) > Co(II) > Cu(II) * Mn(II). ii To study the importance of the hydrogen bonding interactions between the heme propionates and the protein matrix on heme binding dynamics and stability, a series of variants and derivatives of horse heart Mb and bovine liver cytochrome b5 was constructed in which the amino acid residues that form hydrogen bonds with the heme propionate groups were systematically substituted. These hydrogen bonding interactions play a partial role in stabilizing the heme within the binding pocket as suggested by the observed decreased thermal stability, the increased rate constants for heme dissociation, and the lower activation energies for heme dissociation of the Mb and cytochrome b5 variants. These hydrogen bonding interactions play a more significant role in the heme reorientation kinetics as indicated by the greatly increased rate constants for heme reorientation observed for the Mb variants. The correlation between the stability of protein-small molecule complexes in the gas phase and their stability in aqueous solution was evaluated using electrospray mass spectrometry through the investigation of the series of horse heart Mb and bovine liver cytochrome b5 variants described above. A good correlation between the stability of heme binding in the gas phase and in solution was obtained which suggests the hydrogen bonding interactions present in solution are maintained in the gas phase under these mild conditions and that the protein is not grossly misfolded in the gas phase on the time scale of these measurements. Charge reversal mutations close to the heme pocket had significant effects on the properties of the distal heme binding pocket, specifically, the binding of the anionic azide ligand, the p^TA of the distal water molecule and the reduction potential of the heme iron. Metal binding to the surface of the protein near the heme affects the properties of the distal heme binding pocket, exhibiting linkage to the binding affinity for anionic ligands and to the reduction potential of the heme iron. iii T A B L E OF CONTENTS ABSTRACT ii T A B L E OF CONTENTS iv LIST OF TABLES ix LIST OF FIGURES xi ABBREVIATIONS xiv ACKNOWLEDGMENTS xvi I. INTRODUCTION A. Molecular recognition in biology 1 1. Metal Binding Sites in Proteins 1 a. Manganese-containing proteins 2 b. Manganese peroxidase . . , 3 c. Engineered metal binding sites 5 2. Heme-apoprotein interactions 7 a. Heme orientation heterogeneity 8 3. Ligand binding by Myoglobin 10 a. Distal bound water in ferric Mb 11 b. Azide binding to ferric Mb 12 B. Investigating Molecular Recognition 13 1. Myoglobin 14 2. Creation of a metal binding site 15 iv 3. Investigation of heme-protein interactions 19 4. Ligand binding to myoglobin 22 5. Electrospray Mass Spectrometry: An emerging method for studying protein-ligand interactions 22 II. E X P E R I M E N T A L PROCEDURES A. Preparation of proteins 1. Oligonucleotide-directed site-specific mutagenesis 25 2. Protein expression and purification 26 B. EPR spectroscopy 1. Mn(II)-myoglobin titrations 27 C. ^ - N M R experiments 1. Mn(II)-myoglobin titrations 28 2. Reorientation kinetics 29 3. pH titrations 30 D. Potentiometric titrations 30 E. Enzyme assays 1. Manganese peroxidase activity measurements 31 2. Peroxidase activity measurements 31 F. Heme dissociation assays 1. Heme dissociation rates measurements 32 2. Activation energy measurements 33 G. Mass spectrometry v 1. Gas phase dissociation voltage determination 33 2. Tandem mass spectrometry 34 3. Data simulation 36 H. Azide binding studies 1. Equilibrium dissociation constant determination . 36 2. FTIR spectroscopy 37 I. Spectrophotometry determination of heme pKA 38 J. Spectroelectrochemistry 38 K. Circular dichroism spectroscopy 40 III. RESULTS A. Characterization of Mn(II) binding to Mb variants 1. EPR spectroscopy 41 2. Potentiometric Mn(II) titrations 44 3. 'H-NMR Mn(II) titrations 48 4. Enzymatic activities 51 B. Effects of hydrogen bonding interactions on heme binding to Mb 1. Heme binding disorder 54 2. Kinetics of heme reorientation 54 3. Kinetics of heme dissociation 58 4. Activation energy of heme dissociation 64 5. Thermal stability 68 6. Mass spectrometry 71 vi C. Effects of amino acid substitutions on ligand binding 1. Azide binding 81 a. Linkage of Mn(II) binding and azide binding 84 2. HemepK A 87 D. Other characterizations 1. Electrochemistry 89 a. Linkage of Mn(II) binding and reduction potential 91 2. Circular dichroism 92 3. Electronic absorption spectra 93 4. ' H - N M R measurements 93 IV. DISCUSSION A. Development of a new metal binding site in Mb 98 1. Structural characterizations 99 2. Functional characterizations 105 B. Importance of the heme propionate interactions on heme protein stability 107 1. Kinetics of Heme Reorientation 109 2. Kinetics of Heme Dissociation 110 3. Mechanism of Heme Reorientation 112 4. Activation energy measurements 113 5. Thermal stability measurements 114 6. Modulation of proximal histidine character 115 vii 7. ES-MS as a technique for studying protein complex stability . . . 118 a. Dissociation voltages of the various charge states 118 b. Complex stability in the gas phase 119 C. Summary 122 C. Functional consequences of electrostatic modification of the active site of Mb 1. Azide binding 124 2. H e m e p £ A 127 3. Reduction potentials 128 4. Modulation of function by metal binding 129 D. Concluding Remarks 131 V. REFERENCES 132 VI. APPENDIX A. Map of the p G Y M vector 147 B. Construction of the Optically-Transparent Thin Layer Electrode (OTTLE) cell 148 C. Extinction coefficients of the myoglobin variants 149 D. Absorption maxima of the myoglobin variants 150 E. Assignments of Heme Resonances of the Myoglobin Variants 151 F. Metal-ligand distances of Mn(II) binding sites in Mb and MnP 153 G. Calculation of Equilibrium Mn(II) Binding Constant for Ferrous Lys45Glu/Lys63Glu Mb 154 viii LIST OF T A B L E S 1. Mutagenic oligonucleotides 25 2. Mn(II) equilibrium binding constants of the Mb variants from EPR titrations . . 42 3. Mn(II) equilibrium binding constants of the Mb variants from potentiometric titrations 46 4. Manganese peroxidase activity measurements for the Mb variants 52 5. Rate constants for the formation of Mb compound I-like species 52 6. Heme reorientation rate constants for wild-type and variant Mbs 58 7. Heme dissociation rate constants for wild-type and variant Mbs 61 8. Heme dissociation rate constants for wild-type and variant cytochrome b5. . . . 63 9. Activation energies and pre-exponential factors for wild-type and variant of Mb and cytochrome b5 66 10. Melting temperatures of the Mb and cytochrome b5 variants 69 11. Dissociation voltages, charge * voltage, and number of atoms for wild-type and variants of Mb and cytochrome b5 75 12. Azide equilibrium dissociation constants of wild-type and variants Mb 82 13. FTrR absorption maxima and equilibrium binding constants of the Mn(II)-Mb azide derivatives 86 14. Heme pK A values of the Mb variants 88 15. Reduction potentials of the Mb variants 90 16. Reduction potentials of wild-type and Lys45Glu/Lys63Glu Mb in the presence and absence of Mn(II) 91 ix 17. Molar extinction coefficients of the Mb variants ; . .• 149 18. Absorption spectra of the various derivatives of the Mb variants 150 19. Heme methyl hyperfine shifts and mean heme methyl shifts for the met-aquo Mb variants 151 20. Heme resonance assignments in the met derivatives of the Mb variants 151 21. Heme resonance assignments in the met-CN derivatives of the Mb variants . . 152 22. Mn(II)-ligand distances in metal binding sites in Mb and MnP 153 x LIST O F F I G U R E S 1. The catalytic cycle of manganese peroxidase (MnP) 4 2. The structure of protoheme IX 9 3. Ribbon representation of horse heart Mb 16 4. Designed metal binding site in horse heart Mb 18 5. Mn(II) binding site of MnP 19 6. Hydrogen bonding interactions to heme in horse heart Mb and bovine liver cytochrome b5 21 7. Diagram of the triple quadrupole mass spectrometer 35 8. EPR spectra of Mn(II) with and without Lys45Glu Mb 41 9. pH dependence of Mn(II) binding to the Lys45Glu/Lys63Glu Mb var iant . . . . 44 10. Potentiometric Mn(II) binding titration curve for Lys45Glu/Lys63Glu Mb . . . 45 11. X H-NMR spectra of the Lys45Glu/Lys63Glu Mb with increasing concentrations of Mn(II) 50 12. pH dependence of the ABTS turnover of wild-type, Lys45Glu Mb and Lys45Glu/Lys63Glu Mb 53 13. ^ - N M R spectra of the reconstitution of the Lys45Glu Mb with heme 56 14. Heme reorientation kinetics of the Lys45Glu Mb 57 15. Absorption spectra of wild-type and His64Tyr/Val68Phe variant metMb . . . . 59 16. Heme dissociation from Mb variants monitored spectrophotometrically 60 17. Temperature dependence of heme dissociation rate from His97Leu/Lys45Glu Mb 65 xi 18. Dependence of Arhennius activation energy of heme dissociation on the number of H-bonds for Mb and cytochrome b5 67 19. Dependence of melting temperature (TM) on the number of H-bonds for Mb variants 70 20. Electrospray mass spectrum of Ser92Asp/Lys45Glu Mb 73 21. Mass spectrum of wild-type Mb at various A V o s values Linear dependence of complex dissociation on A V o s 74 22. MS/MS experiment, daughter scan of the+10 holoMb peak 76 23. Dependence of experimental and simulated dissociation voltages on Mb charge state 77 24. Gas phase dissociation voltage vs the solution phase Arhennius activation energy for heme dissociation 80 25. Titration of Ser92Asp Mb variant with azide 81 26. Infrared spectrum of Lys45Glu/Lys63Glu M b - N 3 complex 84 27. Spectrophotometric pH titration of Lys45Glu/Lys63Glu Mb 88 28. Spectroelectrochemical titration of the Lys45Glu Mb 89 29. CD spectra of the wild-type and Lys45Glu Mb 92 30. pH dependence of the chemical shift of the 6-a' heme resonance of the Lys45Glu Mb 95 31. X-ray crystal structure of the Lys45Glu/Lys63Glu Mb in the presence of Mn(II) 102 32. Electrostatic surface potential diagram of wild-type and Lys45Glu/Lys63Glu Mb 103 xii 33. X-ray crystal structure of the Lys45Glu/Lys63Glu Mb in the presence of Cd(II) 104 34. Schematic representation of the reaction of apoMb with heme 108 35. Correlation of melting temperature with activation energy for Mb variants . . . 116 36. Map o f thepGYM vector 147 37. Diagram of the OTTLE cell construction 148 38. Thermodynamic cycle of Mn(II) binding to the Lys45Glu/Lys63Glu Mb . . 154 xiii A B B R E V I A T I O N S ABTS 2,2-azinobis-(3-ethyl benzthiazoline-6-sulphonic acid) apoMb myoglobin lacking the heme prosthetic group CAD collisionally-activated dissociation CcP cytochrome c peroxidase CD circular dichroism D M E dimethylester heme E A Arrhenius energy of activation E. coli Escherichia coli EPR electron paramagnetic resonance spectroscopy ES-MS electrospray mass spectrometry FPLC fast protein liquid chromatography FT-rR fourier transform infrared spectroscopy Hb hemoglobin FTEPES N-2-hydroxyethylpiperazine-N' -2-ethane-sulfonic acid holoMb myoglobin containing the heme prosthetic group HRP horseradish peroxidase I ionic strength Mb myoglobin MES 2-(N-Morpholino)-ethane-sulfonic acid metMb oxidized myoglobin MnP manganese peroxidase xiv MOPS 3 -(N-morpholino)-propane-sulfonic acid OTTLE optically transparent thin layer electrode N M R nuclear magnetic resonance spectroscopy SCE standard calomel electrode SHE standard hydrogen electrode TAPS N-tris(hydroxymethyl)-methyl-3 -amino^propane-sulfonic acid Tris tris(hydroxymethyl)aminomethane v D average dissociation voltage of four charge states A V o s voltage difference between the orifice and skimmer plates XV A C K N O W L E D G M E N T S Thank you to my family, friends and colleagues. xvi I. INTRODUCTION A . Molecular Recognition in Biology. Non-covalent, specific interactions between biomolecules are central to all life processes. The regulation of biological activities involves specific interactions between proteins and other molecules. Precise interactions are observed in enzyme-substrate, receptor-ligand, antibody-antigen, and protein-nucleic acid complexes. For example, the replication, transcription and translation of genetic material involve a plethora of specific interactions that are essential in ensuring these processes are accurate and highly regulated. The ability of antibodies to bind unique antigens allows the immune system to recognize and eliminate foreign substances and is another well-recognized example of a biological process involving specific protein recognition events. The present study involves the investigation of a variety of relatively simple molecular recognition processes using variants of myoglobin (Mb) created by site-directed mutagenesis. 1. Metal Binding Sites in Proteins Binding of metal ions by proteins is one of the better understood examples of protein molecular recognition (Tainer et al., 1991). The understanding of metal binding is an important goal in the study of protein structure and function as about one-third of all proteins require a metal ion (Ibers & Holm, 1980). The specific properties of metal ions allow metalloproteins to perform a wider range of functions that are not possible using the naturally-occurring amino acids alone. Metal ions in proteins are involved in ligand binding, transport and storage, substrate recognition, signaling, protein folding and stability, electron transfer and catalysis. The transition metals most commonly found in biology are Fe, Cu, Zn, and Mn from the first 1 row transition series. The alkali earth metals, Ca and Mg, are also found in many protein structures. (Tainer et al., 1991; Glusker, 1991) Affinity of a protein binding site for a specific metal ion depends on metal ion charge and radius (polarizability), and the type and geometry of ligands provided by the protein. Potential ligands from the protein typically include the side chains with electron donating character (aspartic acid, glutamic acid, cysteine, histidine, and to a lesser extent, lysine, serine, tyrosine, methionine) and backbone carbonyl oxygens. In addition, metal ions often bind in a shell of hydrophillic ligands surrounded by a hydrophobic region (Yamashita et al., 1990). Iron-containing proteins, the most abundant type of metalloprotein, can be classified as either heme proteins (e.g. myoglobin, cytochrome c) or non-heme Fe proteins (e.g. hemerythrin, ferredoxin). Electron transfer metalloproteins require redox-active metal ions such as iron and copper. Iron and copper-containing proteins are also essential in the transport of oxygen (e.g. hemoglobin (Fib), hemocyanin) and in oxidative metabolism (e.g. cytochrome P450). Intracellularly, calcium is important as a second messenger. This role requires that the affinity of the Ca2+-binding proteins be fine tuned to allow initiation of biological responses at the appropriate intracellular concentrations of Ca 2 + (McPhalen et al., 1991). Zinc ions are found in proteins in both structural and catalytic situations. A well-known function of zinc is in the maintenance of structure of D N A binding domains such as zinc fingers and zinc clusters in transcription factors (Berg, 1990; O'Halloran, 1993; Berg & Shi, 1996). Ribozymes are another recently discovered class of metalloenzymes that require divalent metal cations for their chemistry and often stability. These enzymes are made entirely of ribonucleic acid (RNA) and can catalyze transformations on R N A molecules (Pyle, 1993). Manganese-containing proteins. Manganese is an essential trace element for humans, plants and many microorganisms (Archibald, 1986). Production of oxygen by oxygenic green plants and algae is an essential process in the ecosphere and is intimately dependent on manganese. The water-2 oxidizing complex of photosystem II in chioroplasts, which produces dioxygen from water, utilizes a Mn cluster. This Mn cluster consists of at least two and probably four Mn atoms and appears to accumulate oxidizing equivalents and act as a catalytic site (Dismukes et al., 1994; Nugent, 1996). M n superoxide dismutase, found in prokaryotes and the mitochondrial matrix of eukaryotes, is an essential protein in aerobic cells. This enzyme catalyzes the dismutation of superoxide radicals that are formed during normal oxygen metabolism to oxygen and hydrogen peroxide (Fridovich, 1986; Hassan, 1989). Manganese acts as a relatively strong Lewis acid in its T and 3 + oxidation states and, therefore, prefers hard Lewis bases such as carboxylate groups as ligands (Glusker, 1991; Larson & Pecoraro, 1992). In the Mn-containing metalloproteins for which the three-dimensional structure has been determined, carboxylate groups are frequently observed as ligands to the metal ions. Examples of such proteins include concanavalin A (Hardmann et al., 1982) and xylose isomerase (Carrell et al., 1989; Collyer et al., 1990). Manganese peroxidase. Another important manganese-containing enzyme is manganese peroxidase (MnP). A large proportion of the earth's biomass consists of lignin, a heterogeneous, highly crosslinked, phenyl propane polymer, which is highly resistant to enzymatic degradation due to its complex structure (Sarkanen, 1971). Lignin biodegradation is most effectively performed by the white-rot fungi. MnP (a 46000 Da glycoprotein) is one of the two extracellular heme enzymes produced by the lignolytic white rot fungus Phanerochaete chrysosporium (Kuwahara et al., 1984). This peroxidase can, in the presence of H 2 0 2 , oxidize Mn(II) to Mn(III) (Glenn & Gold, 1985; Paszczynski et al., 1986). The Mn(III) ion then functions as the diffusible oxidizing species that can subsequently oxidize plant cell wall polymer lignin (Kirk & Farrell, 1987; Gold et al., 1989) and a variety of aromatic pollutants (Bumpus & Aust, 1987; Hammel, 1989; Joshi & Gold, 1993). 3 The catalytic cycle of MnP is similar to that of horseradish peroxidase (HRP) and cytochrome c peroxidase (CcP) where compound I, formed by reaction with H 2 0 2 , oxidizes two molecules of substrate (Wariishi et al., 1989; Dunford, 1991; Bosshard et al., 1991; English & Tsaprailis, 1995). The catalytic cycle (Figure 1) begins with the binding of hydrogen peroxide to the resting Fe(III) form of the peroxidase enzyme. Reduction of H 2 0 2 to H 2 0 results in the formation of a highly oxidized ferryl intermediate, Fe(IV)=0, and a porphyrin n cation radical; this doubly oxidized enzyme intermediate is referred to as Compound I (MnPI) (Wariishi et al., 1988; Wariishi et al., 1989). Two molecules of substrate, Mn(n), are oxidized by MnPI to return the enzyme to its resting state, Fe(III) (Wariishi et al., 1988) with compound U (MnPU) as an intermediate. The two equivalents of Mn(III) are then able to diffuse away and oxidize various organic substrates. Organic diacids such as oxalate and malonate are secreted by the fungus into the medium to facilitate Mn(III) dissociation from the enzyme and stabilize Mn(IH) in aqueous solution through complex formation (Wariishi et al., 1992). MnP + H 2 0 2 *' > MnPI + H 2 0 MnPI + M n 2 + *' > MnPII + M n 3 + *3 MnPII + M n 2 + >" MnP + M n 3 + K 2 M n 3 + + 2 A H • 2 M n 2 + + 2A-Figure 1. The catalytic cycle of manganese peroxidase. MnPI and MnPII refer to compound I and compound II of MnP, respectively. A H and A- represent the reduced and oxidized forms of the organic substrates, respectively. While the present investigation was in progress, the crystal structure of the MnP enzyme from P. chrysosporium was determined (Sundaramoorthy et al., 1994). Not surprisingly, the active site of 4 the MnP enzyme is structurally similar to that of CcP (Sundaramoorthy et al., 1994). Five important active site residues, highly conserved in plant and fungal peroxidases (English & Tsaprailis, 1995), were identified in the MnP structure (Sundaramoorthy et al., 1994). On the proximal side of the heme, His 173 is coordinated to the heme iron and is hydrogen bonded to a buried aspartyl residue at position 242. The key catalytic residues in the distal heme pocket, His46 and Arg43, are critical in the reaction of the enzyme with peroxide. Asn80, also in the distal heme pocket, influences the reaction with peroxide by forming a hydrogen bond with His46. The importance of these conserved residues has been most extensively studied in CcP. On the proximal side of the heme in CcP, the hydrogen bond formed by Asp235 and His 175 increases the histidinate character of the proximal ligand (His 175) and strengthens the His-Fe bond, which maintains the pentacoordinate character of the iron and lowers the reduction potential (Poulos & Finzel, 1984; Wang et al., 1990; Goodin & McRee, 1993; Ferrer et al., 1994). The distal histidine, His52, has been shown to function as an acid-base catalyst (Poulos & Kraut, 1980; Erman et al., 1992), and the nearby Arg48 residue is critical to the stability of compound I (Vitello et al., 1993; Bujons et al., 1996). Asn82 forms a hydrogen bond with the distal histidine residue, and thereby ensures that the second imidazole nitrogen is ready to accept a proton from the H 2 0 2 substrate (Satterlee et al., 1994; English & Tsaprailis, 1995). Finally, CcP possesses the unusual characteristic of having the radical centre reside on an amino acid residue, Trpl91, in close proximity to the heme (Sivaraja et al., 1989; Scholes et al., 1989). Engineered metal binding sites. The large number of metalloprotein crystal structures now available permits the elucidation of general principles for metal binding. Binding motifs common to specific metal ions can be determined to aid in the design of new metalloproteins. Molecular model building methods and computer programs are emerging to assist the introduction of new metal 5 binding sites into proteins of known structure (Hellinga & Richards, 1991) or to search protein structures for groups of residues with appropriate geometric arrangements for metal binding (Gregory et al., 1993). In recent years, a number of metal binding sites have been successfully designed and constructed in non-metalloproteins. For example, a buried transition metal ion binding site similar to the type I copper sites found in azurin and plastocyanin has been introduced into the hydrophobic core of thioredoxin (Hellinga et al., 1991). High affinity metal binding sites have been engineered into recombinant proteins for a number of purposes (for review, Arnold & Haymore, 1991; Tainer et al., 1991; Regan, 1993; Matthews, 1995; Hellinga, 1996). For example, an exposed metal binding site created on a protein surface has been used to facilitate protein purification (Arnold & Haymore, 1991). Metal chelating peptides have been attached to the N - and C-termini of proteins to create metal binding sites that also facilitate purification by immobilized metal ion affinity chromatography (Smith et al., 1988; Hochuli et al., 1988). In other studies, metal affinity sites were constructed on the surface of a bovine somatotropin (Suh et al., 1991) and cytochrome c (Todd et al., 1991) by creating a His-X 3 -His motif on a surface a-helix. This motif places two solvent exposed histidine ligands across a single turn of the oc-helix that increases the affinity of the protein for Cu(II) and thus the retention of the protein on a Cu(II)-containing metal-affinity column. Several natural metalloproteins, including "zinc-finger" D N A binding proteins (Berg, 1988; Parraga et al., 1988), possess this His-X 3 -His amino acid motif as part of the metal binding site. Metal binding sites have been engineered into a number of enzymes for the purpose of metalloregulation (Higaki et a l , 1992). For example, the creation of a Cu(II) binding site in trypsin involving one of the active site histidines allowed the activity of this protease to be regulated by the presence of metal ions (Higaki et al., 1990). A metal-dependent regulatory site was created in the 6 active site of Staphylococcal nuclease by introducing a cysteine residue into the hydrophobic active site (Corey & Schultz, 1989). Specificity for histidine-containing peptide substrates has been engineered into trypsin by creating metal binding sites in the substrate binding pocket of the protein (Willett et al., 1995). Engineered metal binding sites in the transmembrane channels, Staphylococcal a-hemolysin and Shaker-A K + channel, allows the activity of these pore channels to be modulated by the presence or absence of metal ion (Walker et al., 1994; Yellen et al., 1994). The use of metal ions to stabilize protein structures has also been reported. Increased structural stability through creating or enhancing a Ca 2 + binding site was achieved in lysozyme (Kuroki et al., 1989) and subtilisin BPN'fPantoliano et al., 1988; Braxton & Wells, 1992). The binding of Zn(II) to the metal binding site engineered into a four helix bundle protein enhanced the stability of this protein to guanidine-HCl denaturation (Handel & DeGrado, 1990). Finally, metal ion binding sites have also been incorporated into the light chains of antibodies through site-specific mutagenesis to create antibodies with potentially catalytic metal cofactor sites (Iverson et al., 1990; Roberts et al., 1990). 2. Heme-apoprotein interactions The binding of the iron-protoporphyrin IX (heme) prosthetic group by proteins is an interesting problem in molecular recognition. The heme group is required for function in a large range of proteins from electron transfer proteins, to oxygen binding proteins, to heme enzymes. This diverse range of heme chemical reactivity is modulated by the surrounding protein matrix and the manner in which the protein interacts with the heme (Poulos & Finzel, 1984; Poulos, 1988; English, 1994). The structures of myoglobin (Mb) and hemoglobin (Hb) are optimal for oxygen transport. Cytochromes play a central role in the electron transfer chains of respiration and photosynthesis. The heme 7 enzymes, like the peroxidases and cytochromes P450, are larger in size than cytochromes and Mb and promote formation of activated iron-oxo intermediates that are able to oxidize organic substrates (Li & Poulos, 1994). Structural factors controlling the binding of heme to heme proteins include coordination of the heme iron by the axial ligands, van der Waals interactions between the polarizable heme % system and hydrophobic amino acids, small steric constraints imposed by the protein on the vinyl substituents, and the hydrogen bonds formed by amino acid residues with the heme propionates. Although the determination of the affinity constants of proteins for heme is difficult owing to the highly aggregated nature of heme in aqueous solution, horse heart Mb has been estimated to bind heme with high affinity (K D ~10"15 M) (Banerjee, 1962). Through heme exchange reactions with horse heart Mb, cytochrome b2 from yeast was found to possess a similar affinity for heme (Pajot, 1971) and Aplysia limacina Mb was found to exhibit lower affinity (Rossi-Fanelli & Antonini, 1960). Heme orientation heterogeneity. 'H -NMR spectroscopy has revealed that the interaction of heme with apomyoglobin (apoMb) and apocytochrome b5 is more complex than previously appreciated (Keller et al., 1976; La Mar et al., 1978b). The N M R spectra of these proteins revealed a heterogeneity consistent with the existence of two protein populations that differ in the orientation of heme binding. Specifically, the two populations differ by 180° rotation of the heme about the afy-meso heme axis (Figure 2) (La Mar et al., 1978b; Keller & Wuthrich, 1980). For sperm whale and horse heart Mb, 92% of the protein binds heme in the "major" orientation and 8% binds heme in the "minor" orientation (La Mar et al., 1983). In pig and rabbit Mb and bovine liver cytochrome b5, - 9 0 % of the protein binds heme in the "major" orientation (Keller et al., 1976; Keller & Wuthrich, 1980; La Mar et a l , 1981; McLachlan et a l , 1986). At equilibrium, Hb A contains - 2 % and 10% of the minor heme orientation in the a- and P-subunits, respectively (Yamamoto & La Mar, 1986). In 8 yellowfin tuna Mb, the difference in the relative stabilities of the two orientational isomers is reduced as only 67% of the heme is in the major orientation at equilibrium (Levy et a l , 1985). Interestingly, upon initial addition of heme to apoMb, the two orientational isomers are formed in equal abundance, and with time an equilibrium is established in which one orientational isomer dominates (Jue et al., 1983). Circular dichroism has also emerged as an alternative technique for measuring heme disorder, yielding similar results to 'H -NMR for sperm whale and yellowfin tuna Mb (Aojula et al., 1986; BelleUi et al., 1987; Aojula et al., 1988) and cytochrome b5 (Singh & Wilson, 1990). Monomeric Hb from the marine annelid Glycera dibranchiata possesses a heme moiety bound predominantly in the minor orientation, as observed by 'H -NMR (Constantinidis et al., 1988) and CD measurements (Santucci et al., 1988). In contrast to these proteins, CcP exhibits no heterogenity of heme binding (Satterlee et al., 1983). HOOC C O O H HOOC C O O H Figure 2. The structure of iron-protoporphyrin LX with the nomenclature used in this work indicated. The positions of the heme side chains after 180 0 rotation about the aly-meso heme axis are illustrated. 9 A variety o fNMR studies has established that the interaction of heme substituent groups with the protein is important in stabilizing the heme-protein complex. Most of these studies have addressed this issue by studying the interaction of wild-type proteins with synthetic analogues of iron-protoporphyrin LX. In the case of Mb and cytochrome b5, the heme vinyl groups (Keller et al., 1976; La Mar etal., 1981; La Mar etal., 1984; La Mar etal., 1986; Lee et al., 1990; Lee et al., 1991a) and the heme propionate groups (La Mar et al., 1985; La Mar et al., 1986; La Mar et al., 1989; Hauksson et al., 1990; Lee et al., 1991b; Santucci et al., 1993) have been suggested to be of particular importance to the dynamics of heme binding to both apoproteins. Also, the formation of the proximal histidine-iron bond has been shown to have an influence on heme binding to apoMb (Kawamura-Konishi et al., 1988; La Mar et al., 1989; Yee & Peyton, 1991). 3. Ligand binding by myoglobin Exogenous ligands can bind to Mb and Hb at the sixth coordination site of the heme iron on the distal side of the heme in both the oxidized and reduced state. In the reduced state (Fe2+), Mb binds the diatomic gases, NO, 0 2 and CO, although the distal pocket has specifically evolved to favor reversible binding of dioxygen. In the oxidized state (Fe3+), a water molecule occupies the sixth coordination position of the heme iron at acidic pH. Anions such as N 3 ' , CN", OH" and F' can displace the water and occupy this binding site. A comprehensive review of kinetic and equilibrium studies on the binding of the diatomic gases, 0 2 and CO, to the five-coordinate, reduced Mb (deoxyMb) has appeared recently (Springer et al., 1994). The importance of various structural factors such as steric hindrance, hydrogen bonding and local polarity in controlling ligand binding has been investigated through the use of variant proteins. The highly conserved amino acids, His64, Val68, Phe43, Phe46 and Leu29, in the distal 10 binding pocket have been the focus of much of this type of work. According to Springer and colleagues (1996), the primary function of the distal histidine, His64, appears to be to stabilize 0 2 bound in the heme pocket preferentially to CO through hydrogen bonding to the second oxygen atom of the bound 0 2 ligand. Pentacoordinate model hemes in organic solvent bind CO with 30 000 to 100 000 times greater affinity than 02, however, within the protein matrix of Mb, binding of CO to heme is only favored 30-fold (Springer et al., 1994). The crystal structure of deoxyMb has indicated the presence of a non-coordinated water in the distal pocket that forms a hydrogen bond with the distal histidine (Takano, 1977a; Quillin et al., 1993). The displacement of this bound water from the distal heme pocket is required for binding of both 0 2 and CO ligands (Springer et al., 1994). Distal bound water in ferric Mb. In sperm whale and horse heart metMb, the water in the distal pocket is bound directly to the heme iron to result in a six-coordinate iron. The acid-alkaline transition observed in these Mbs involves the titration of this coordinated H 2 0 to an OH" ion and is accompanied by a change from a high spin Fe(III) to a spin equilibrium (Antonini & Brunori, 1971). Hydrogen bonding interactions between this water molecule and the distal histidine are important in stabilizing the coordination of this ligand. A. limacina Mb, in which the distal histidine is replaced by a valine, possesses a pentacoordinate iron at neutral pH that has been attributed to the inability of the valine to stabilize binding of a distal water molecule through hydrogen bond formation (Bolognesi et al., 1989). The importance of hydrogen bonding by the distal histidine in the stabilization of the coordinated water ligand is also illustrated in Mb variants in which the distal histidine is replaced. For example, the His64Leu and His64Val variants in human Mb (Ikeda-Saito et al., 1992) and the His64De and His64Val variants in sperm whale Mb (Quillin et al., 1993) are five-coordinate due to the lack of stabilization of the distal water ligand. Alternatively, the coordinated water ligand is 11 retained in the His64Gln Mb variant of human Mb (Dceda-Saito et al., 1992) where the glutamine can presumably form a hydrogen bond with the bound ligand. In variants possessing a tyrosyl residue at position 64, the distal water molecule is displaced by the direct ligation of the tyrosine to the heme iron (Egeberg et al., 1990; Pin et al., 1994; Tang et al., 1994a; Maurus et al., 1994). Coordination of this tyrosine residue to the heme iron alters the electronic absorption spectrum of the variant. Hargrove and co-workers (1994a) have developed an assay for measuring the rate of heme dissociation from heme proteins by taking advantage of the altered spectroscopic properties of the His64Tyr/Val68Phe variant Mb. In this assay, an excess of the His64Tyr/Val68Phe apoprotein is incubated with the heme protein of interest, and heme dissociation is followed spectrophotometrically as the heme is transferred from the holoprotein to the excess variant apoMb. Azide binding to ferric Mb. Azide binding to ferric Mb provides a sensitive probe of the characteristics of the distal heme pocket and information that is complementary to that provided by measurements of oxygen binding equilibria and kinetics. Binding of both 0 2 and azide requires positive dipoles in the distal heme pocket and the displacement of a distal water molecule (Brancaccio et al., 1994). The proposed polar channel for azide entry into the distal pocket is formed by Arg(Lys)45, His64, Val68, Thr67 and the heme 6-propionate (Brancaccio et al., 1994). This channel is blocked by the side chain of the distal histidine, His64, and transient rotations of the side chain of this residue allow approach of the azide ion to the heme iron (Brancaccio et al., 1994). This channel was originally proposed (Perutz & Matthews, 1966) as the ligand entry path into the distal pocket of Hb and Mb; however, experimental evidence does not support this pathway for entry of the diatomic gases (Carver et al., 1990; Gibson et al., 1992; Quillin et al., 1993). Simulations have suggested that these apolar diatomic gases enter the heme binding pocket of Mb through multiple 12 hydrophobic channels (Elber & Karplus, 1990; Czerminski & Elber, 1991). A number of factors are important in determining the stability of binding of azide to ferric Mb. Displacement of the water molecule bound in the distal or sixth position of the heme iron provides a kinetic barrier to azide binding. The hydrogen bonding capability of the distal ligand is another important determinant. The crystal structure of the azide bound form of wild-type horse heart Mb indicates that the distal histidine forms a hydrogen bond with the bound azide molecule (R. Maurus, unpublished). As azide is a negatively charged molecule, the electrostatic character of the heme binding pocket will also be a factor. Substitutions with positively or negatively charged amino acids at the entrance to the heme binding pocket in sperm whale Mb variants caused increases and decreases, respectively, in both the rate constants for azide association and the affinity for azide (Brancaccio et al., 1994; Smerdon et al., 1995). B. Investigating Molecular Recognition. With the development of site-directed mutagenesis (and peptide synthesis for lower molecular weight proteins), specific modifications of proteins can be made with relative ease. By altering single residues in protein active sites or at protein interfaces, the involvement of individual amino acid residues in molecular recognition can be determined. Understanding the function of these specific interactions permits modulation of the stability and specificity of these non-covalent complexes. The focus of the work presented in this thesis has been to study a variety of types of molecular recognition by the protein, Mb. Included in this work is consideration of structural factors related to the interaction of metal ions with a binding site engineered on the protein surface, the binding of heme to apoMb, and the binding of azide to the heme iron. 13 1. Myoglobin Mb consists of a single polypeptide with 153 amino acid residues, and it possesses a non-covalently bound heme prosthetic group. Mb occurs in skeletal and heart muscle where it functions to store and transport dioxygen. The first three-dimensional structure of sperm whale Mb was determined in 1958 (Kendrew et al., 1958; Kendrew et al., 1960), and high resolution structures of the met (Takano, 1977b), deoxy (Takano, 1977a), and oxy (Phillips, 1978) forms of sperm whale Mb have been reported subsequently. Recently, the high resolution structure of horse heart Mb, in which 88% of the protein sequence is identical to sperm whale Mb, was determined by Evans and Brayer (1990). Horse heart Mb is small (17570 Da), stable, and well-studied and can be expressed readily in large quantities (-25 mg/L culture) in E. coli from a synthetic gene (Guillemette et al., 1991). The secondary structure of horse heart Mb consists of eight a-helices (A to H) that form the binding pocket for the heme prosthetic group between helices B, C, E and F (Figure 3). The majority of the amino acid residues (more than 80%) are involved in helices (Evans & Brayer, 1990). The iron atom is coordinated to the porphyrin equatorially through the four pyrrole nitrogens and to the protein axially through the proximal histidine. The exchangeable sixth coordination site of the heme iron forms the ligand binding site and this side of the heme plane is referred to as the distal side. The side of the heme opposite the ligand binding site is termed the proximal side. The heme moiety in Mb is positioned with the propionate groups extending out of the heme crevice to the protein surface and is almost completely inaccessible to solvent (Stellwagen, 1978). Another example of a small heme protein that possesses a non-covalently bound heme is cytochrome b5. This membrane-bound protein can be solubilized by digestion with trypsin or protease-contaminated lipase to produce a soluble heme binding domain of 82 and 93 amino acids, respectively (Ozols & Strittmatter, 1969; Funk et al., 1990). The three-dimensional structure of the 14 soluble domain of cytochrome b5 from bovine liver reveals that the heme binding crevice consists of four antiparallel helices with a P-pleated sheet forming the base of the pocket (Mathews et al., 1971a; Durley & Mathews, 1996). The heme iron atom is ligated axially to two histidine residues at positions 39 and 63. As observed in Mb, the heme propionate groups are exposed to solvent, and most of the heme group is enveloped by the protein (Stellwagen, 1978). 2. Creation of a metal binding site. The question addressed in this portion of the study was whether Mb could be modified to oxidize Mn(II) to Mn(III) in the presence of hydrogen peroxide, and thereby mimic the activity of the native manganese peroxidase enzyme (MnP) from P. chrysosporium. The goal was to create a manganese binding site on the surface of horse heart Mb that could participate in electron transfer reactions with the heme iron. Mb provides a reasonable starting point for this work as it already possesses a low level of peroxidase activity (George & Irvine, 1956; King & Winfield, 1963). Mb has also been shown to catalyze a variety of peroxide-dependent reactions such as the epoxidation of styrene (Ortiz de Montellano & Catalano, 1985; Choe et al., 1994), the oxidation of low-density lipoproteins (Hogg et al., 1994), and the peroxidation of unsaturated lipids (Galaris et al., 1990; Rao et al., 1994; Kanner & Harel, 1995). 15 Figure 3. Ribbon diagram of the helical and non-helical regions in horse heart myoglobin. The eight oc-helices are labelled on the figure. 16 Mb has been used previously as a model protein for the study of electron transfer reactions and metal binding. Pentaammineruthenium(III), after photoreduction, has been shown by Gray and coworkers to reduce the heme iron of sperm whale Mb when the Ru(III) complex was covalently attached to surface histidines on metMb (Mayo et al., 1986; Cowan et al., 1988; Therien et al., 1991). Sperm whale Mb has been shown to bind and reduce Cu(II) site-specifically through the complexation to a number of histidyl residues on the protein surface (Hegetschweiler et al., 1987; Van Dyke et al., 1992). The Mn(IT) binding site developed here in horse heart Mb was designed to possess a variety of characteristics. It was important that the site be freely solvent accessible for ease of turnover. The site should bind Mn(II) with sufficient stability and be sufficiently close to the heme to promote efficient electron transfer. As manganese prefers oxygen ligands, residues possessing carboxylate functional groups were the ligands of choice. The heme 6-propionate was chosen to provide a carboxylate ligand and create a continuous o bonded path from the bound metal to the heme edge. Glutamyl residues were introduced into positions 45, 60, and 63 to place additional potential carboxylate ligands adjacent to the heme 6-propionate and the partially solvent-exposed edge of the heme group (Figure 4). Mn(II) binding to the Mb variants in solution was characterized by nuclear magnetic resonance spectroscopy ('H-NMR), electron paramagnetic resonance spectroscopy (EPR), and potentiometric titrations. In addition to creating the transition metal binding site and assessing the metal binding characteristics, preliminary enzymatic assays were performed to assess the ability of Mn(II) bound at this site to transfer electrons to the heme iron. 17 Figure 4. The putative Mn(II) binding site designed for wild-type Mb. Lys45, Lys60 and Lys63 have been replaced with glutamyl residues. Minimization of the distances between the carboxylate groups and the Mn(II) ion was achieved by rotation of the side chains and the heme 6-propionate. 18 During the course of this work, the structure of the manganese peroxidase (MnP) from P. chrysosporium was reported (Sundaramoorthy et al., 1994), and the manganese binding site in this protein was found to be similar to the metal binding site designed here for Mb. The Mn(II) binding site of MnP (Figure 5) is located near the surface of the protein and is comprised of acidic amino acid residues (Asp 179, Glu35, and Glu39) and the heme 6-propionate group. Two water molecules complete the hexa-coordinate Mn(II) binding site, and one of these water molecules is hydrogen bonded to the heme 7-propionate. The coordination of Mn(II) at this site is octahedral, which is typical of Mn(II) coordination complexes (Demmer et al., 1980). Figure 5. The Mn(II) binding site of the manganese peroxidase enzyme from P. chrysosporium (Sundaramoorthy et al., 1994). 3. Investigation of heme-protein interactions. As indicated above, much of the work reported to date concerning the binding of heme to apoprotein and the stability of the resulting heme-protein complex has employed synthetic heme groups with altered substituents. In the present study, site-directed mutagenesis was employed to 19 alter individual heme-protein interactions to permit assessment of their contributions to the stability of heme binding. The focus of this work was the characterization of the contributions of hydrogen bonds formed between the heme propionate groups and amino acid residues to the stability of heme binding to Mb and cytochrome b5. In addition to altering the amino acid residues involved in hydrogen bonding interactions, these interactions were modified by reconstitution of both proteins with dimethylester heme (DME-heme, heme in which both propionate groups have been converted to the corresponding methyl esters) to reduce hydrogen bonding interactions with the propionates (Reid et al, 1984; Lim, 1990). The residues involved in hydrogen bonding interactions with the heme propionate groups of horse heart Mb and cytochrome b5 are illustrated in Figure 6. In Mb, Leu89, His93, His97, Ser92 and the heme 7-propionate form a hydrogen-bonding lattice on the proximal side of the heme that stabilizes the interaction of the heme with the apoprotein. Specifically, the main chain carbonyl of Leu89 and the hydroxyl group of Ser92 form hydrogen bonds with the proximal histidine, His93. His97 and the hydroxyl group of Ser92 interact with the heme 7-propionate. In addition, Lys45 on the distal side of the heme interacts with the heme 6-propionate. (Evans & Brayer, 1990). In sperm whale Mb, residue 45 is an arginine, which can interact with the distal histidine through a water molecule and with the heme 6-propionate (Takano, 1977b). In bovine liver cytochrome b5, the heme 7-propionate forms two hydrogen bonds with Ser64, one with the side chain hydroxyl group and one with the main chain amide. The heme 6-propionate is fully exposed to solvent and does not participate in any interactions with the protein (Mathews et al, 1971b; Durley & Mathews, 1996). A number of experiments have been employed to evaluate the contributions of these hydrogen bonding interactions to the stability of heme-apoprotein interaction. The effects of these modifications on the heme orientational equilibrium and the kinetics by which equilibration is achieved following 20 reconstitution of the apoprotein with heme were studied with an ^ - N M R method described previously (La Mar et al., 1984). The rate constants for heme dissociation from the variant proteins were also evaluated using a method described by Hargrove and co-workers (1994a). From the temperature dependence of these heme dissociation rate constants, the Arrhenius activation energy for heme dissociation from each variant was calculated. Finally, the thermal stability of each variant was determined by circular dichroism by monitoring the change in secondary structure with temperature. Figure 6. Illustration of the hydrogen bonding interactions formed by the heme propionate groups with amino acid residues of horse heart myoglobin and bovine liver cytochrome bs. 21 4. Ligand binding to myoglobin. Site-directed mutagenesis allows the role of specific amino acid residues surrounding the heme binding site in the recognition and binding of ligands to be examined. The role of the distal Ffis64 residue in ligand binding to both oxidation states of sperm whale, human and pig Mb has been studied extensively through replacement with variety of amino acid residues (Springer et al., 1989; Rohlfs et al., 1990; Carver et al., 1990; Cutruzzola et al., 1991; Ikeda-Saito et al., 1992; Biram et al., 1993; Quillin et al., 1993). The effects of changing charged residues near the heme binding site have also been studied for the same Mb species (Smerdon et al., 1991; Brancaccio et al., 1994; Smerdon et al., 1995). Similar studies have not been performed with horse heart Mb and consideration of azide binding to Mb has been limited. In this work, the effects of changing the electrostatic character of residues close to the heme on the binding of azide and water in horse heart Mb were investigated and the importance of the distal histidine in azide binding was assessed. The existence of a polar channel for ligand entry in horse heart Mb is discussed. 5. Electrospray Mass Spectrometry: an emerging method for studying protein-ligand interactions. The advent of electrospray ionization in the late 1980's has revolutionized biological mass spectrometry. This technique allows the production of intact, multiply charged ions of large, non-volatile, labile biomolecules in the gas phase (Meng et al., 1988; Fenn et al., 1989; Smith et al., 1991; Chait & Kent, 1992). With electrospray ionization, accurate molecular weights of large proteins and oligonucleotides can be determined readily and sequences of proteins and peptides can be determined (Biemann, 1992; Mann & Wilm, 1995). Electrospray ionization is also sufficiently gentle to permit the observation of non-covalently associated complexes in the gas phase. Since the first observation of these types of complexes in the gas phase (Ganem et al., 1991; Katta& Chait, 1991), agreat deal 22 of attention has been focused on their characterization. At present, protein-protein, protein-nucleic acid, enzyme-substrate, and enzyme-inhibitor complexes, and other examples have been detected in the gas phase by this method (Smith & Zhang, 1994; Loo, 1995). Several reports have demonstrated that the intermolecular interactions observed with ESI-MS are related to the interactions of these compounds in solution (reviewed in Loo, 1995). The structure and properties of proteins in vacuo are issues of current debate. The principal forces that determine protein structure and stability are hydrogen bonds, hydrophobic interactions and packing forces. In addition, disulfide bonds are stabilizing interactions in some proteins. As both hydrogen bonding and hydrophobic interactions will be strongly influenced by the presence or absence of water molecules, there is concern whether the amino acid sequence of a protein dictates the same tertiary structure in vacuo as in solution. (Wolynes, 1995) Collisionally-activated dissociation (CAD) has been used to dissociate clusters (Shahin, 1966) and break covalent bonds (Caldecourt et al., 1983; Aleksandrov et al., 1984) of ions formed by atmospheric pressure ion sources. Ions are accelerated by electric fields through a region containing a locally high density of gas. Resulting collisions with the gas transfer translational energy of the ions to internal energy and lead to bond dissociation (Bruins, 1991). The total energy that can be transferred is proportional to the potential difference applied across the acceleration region. This method of fragmenting ions can be used at the ion sampling interface or in a collision cell within the mass spectrometer. As complexes of greater stability generally require a greater voltage for dissociation, CAD has been used to measure ion or complex stability. Non-covalently associated complexes observed in the gas phase from electro spray ion sources can be dissociated by CAD in the sampling interface by increasing the declustering potential, which is the voltage difference between the orifice and skimmer plates in the sampling interface (Smith & 23 Light-Wahl, 1993; Tang et al., 1994b; Lim et al., 1995; Loo, 1995). For example, the complex formed by the HTV protease dimer and a substrate-based inhibitor was observed in the gas phase and shown to dissociate at higher orifice voltages (Baca & Kent, 1992). Similarly, heme can be dissociated from Mb by increasing the orifice voltage (Konishi & Feng, 1994). Also, the quantity of DTN-y dimers observed in the gas phase was found to be linearly dependent on the voltage applied to the orifice plate (Huang et al., 1993). Finally, this method has been used to compare the relative stabilities of ras-GDP and interferon-y dimers (Huang et al., 1993) and of tetramers of concanavilin A and Hb (Light-Wahl et al., 1994). Two important issues that have not yet been addressed are (I) whether small differences in the affinity of proteins for ligands can be detected in the gas phase and (ii) whether the stability of these non-covalently associated complexes in the gas phase correlates with the stability of such complexes in solution. The strategy used in the present study to address these questions was to compare the contributions of individual hydrogen bonds toward the stability of a series of protein-heme complexes in both the gas and aqueous phases. In solution, the Arrhenius activation energy (EA) for heme dissociation from Mb and cytochrome b5 was determined from the temperature dependence of the rate constants for heme dissociation (Hargrove et al., 1994a). The relative stability of heme binding to the apoMb and apocytochrome bs variants in the gas phase was determined by CAD in the differentially pumped ion sampling interface of the electrospray mass spectrometer. A comparison of the solution properties of these complexes with their properties in the gas phase should shed some light on the structure of these protein complexes in the gas phase. A good correlation between the stability of such complexes in the solution and gas phase would suggest that this method may be generally useful in the study of molecular recognition. 24 H. EXPERIMENTAL PROCEDURES A. Preparation of proteins L Oligodeoxyribonucleotide-directed site-specific mutagenesis A synthetic gene coding for horse heart Mb (Guillemette et al., 1991) was used for the preparation of the Mb variants (pGYM vector, Figure 36, Appendix A). Mutagenic oligodeoxyribonucleotides were synthesized with a modified Applied Biosystems 3 80A D N A Synthesizer at the UBC Nucleic Acid and Protein Synthesis laboratory and purified using a Millipore C18 Sep Pak column. The procedure used for oligonucleotide-directed mutagenesis has been described previously (Kunkel, 1985; Zoller & Smith, 1987). The single variants (Lys45Glu and Ffis97Leu) were prepared by Ff. Lee and S. Rafferty, respectively. The multiple variants (FIis97Leu/Lys45Glu, His97Leu/Lys45Glu/Lys63Glu and Ser92Asp/Lys45Glu) were prepared by L. Eltis by sub-cloning D N A fragments containing one mutation into the gene containing the other mutation using the restriction sites available in the wild-type gene sequence. Preparation of the lipase-solubilized bovine liver cytochrome b5 Ser64Ala mutant has been described previously (Funk et al., 1990). Table 1. Oligodeoxyribonucleotides used for site-directed mutagenesis of Mb. Mutations Sequence Lys63Glu 5'-C A A C GGT ACC A T G TTC TTT C A G ATC TTC AG-3' Lys63Ser 5'-C A A C GGT ACC A T G A G A TTT C A G ATC TTC AG-3' Asp60Glu/Lys63Glu 5'-C GGT ACC A T G TTC TTT C A G TTC TTC A G A CGC C-3' Ser92Ala 5'-T AGT AGC A T G AGC TTG CGC A A G C G G TTT G A G C-3' 25 2. Protein expression and purification Commercial horse heart Mb (Sigma M l 882) was purified chromatographically according to a previously reported method (Tomoda et al., 1981). The recombinant wild-type and variant horse heart Mb proteins were expressed in E. coli LE392 cells. Cells were grown in a 20 L fermentor in Superbroth media [tryptone (10 g/L), yeast extract (8 g/L), NaCl (5 g/L) and ampicillin (100 mg/L)] and harvested with a tangential flow centrifuge at the U B C Biotechnology Fermentation facility. The proteins were purified as described (Lloyd & Mauk, 1994), however, the variants were not further purified by FPLC as the HR 10/10 Mono-Q anion exchange resin (Pharmacia) removed the heme from the protein. Final A4 0g/A2 8 0 ratios greater than 5.5 and 5.0 were obtained for the wild-type and variant Mb proteins, respectively. Molar absorption extinction coefficients of the variant proteins (Appendix C) were determined by the pyridine hemochromogen method (de Duve, 1948). The recombinant proteins isolated from E. coli were reconstituted with heme to eliminate sulffnyoglobin formed during fermentation (Lloyd & Mauk, 1994). ApoMb was prepared on ice by the acid-butanone method (Teale, 1959) in the following manner. The pH of the Mb solution was lowered to <2.0 and the heme was extracted from the aqueous layer with 2-butanone (Sigma-Aldrich). The extraction was repeated until the butanone layer was colorless. The aqueous apoMb solution was then dialyzed at 4 °C for 12 hours against a series of buffers: 0.6 mM NaHC0 3 /l mM EDTA, 0.6 mMNaHCOj, and 100 mM sodium phosphate buffer, pH 7.0. ApoMb was reconstituted with 1.2 equivalents of iron-protoporphyrin IX (10 mg/ml in 0.2 M NaOH) (Porphyrin Products, Logan, Utah) and the excess heme was removed from the holoMb by elution over a Sephadex G-75 Superfine column (2.5 x 75 cm). Reconstitution of wild-type apoMb with DME-heme has been reported previously (Lim, 1990). The purification of recombinant wild-type lipase-solubilized bovine liver cytochrome b5 and 26 the Ser64Ala variant has been described (Funk et al., 1990). Recombinant human erythrocyte cytochrome b5 (Lloyd et al., 1994) and trypsin-solubilized bovine liver cytochrome b5 (Reid & Mauk, 1982) have been purified previously. Trypsin-solubilized cytochrome b5 reconstituted with D M E -heme was also prepared previously (Reid et al., 1984); B. Electron Paramagnetic Resonance spectroscopy 1. Mn(II)-myoglobin titrations Binding of the paramagnetic Mn(II) metal ion to Mb was studied with a Bruker Model ESP 300E spectrometer at X-band frequencies (ca. 9.5 GHz) at 25 °C in 1.5 x 90 mm glass capillaries (Baxter-Canlab). The EPR parameters used were as follows: a modulation frequency of 100 kHz, a modulation amplitude of 12.8 G, a microwave frequency of 9.45 GHz, and a microwave power of 3.16 mW. Mb samples were exchanged into 50 mM HEPES, pH 7.0 (ionic strength (I) = 17.2 mM) prior to titration. Titrations were performed using a series of samples (-6-7 samples) with a fixed MnS0 4 concentration (0.125 mM), varying protein concentrations (0.1 - 2 mM) and a final volume of 40 pX for each sample. To determine the amount of Mn(II) bound at each protein concentration, the intensity of the third line of the Mn(II) six-line signal was ascertained and compared to the intensity of the signal in the absence of protein. The binding constants (Kf) were calculated from a least squares fit of the data to the Hughes-Klotz equation (1) for a single set of n non-interacting metal binding sites (Hughes & Klotz, 1956) (MTNSQ Version 4.02, MicroMath, Inc.). [Mb] t o t a l / [Mn(II)]b o u n d = ( [Mn(II)] f c e KA n )"' + n1 (1) Similar methods have been used to measure Mn(II) binding to sperm whale Mb (Gersonde & Netter, 1966), apotransferrin (Harris & Chen, 1994) and bacteriophage T7 R N A polymerase (Moon Woody et al., 1996). 27 The pH dependence of Mn(II) binding to Lys45Glu/Lys63Glu Mb was also studied. The buffer system for these measurements, consisting of 5 mM MES, 5 mM MOPS, 5 mM TAPS and 95 mMNaCl , allows the pH to be adjusted over the pH range of 5.0 to 7.5 by addition of NaOH with little change in ionic strength (Barker & Mauk, 1992). The relative affinities of Cu(II), Cd(II), and Co(U) for the engineered metal binding site were also determined by EPR through competition with bound Mn(n). In these experiments, the intensity of the EPR signal of 0.125 mM M n S 0 4 in 50 mM HEPES buffer, pH 7.0 was determined first. Then, the decrease in intensity of the Mn(II) signal in the presence of a molar equivalent of protein was determined. Finally, an equivalent of the competing metal ion was added, as CuS0 4 , CdS0 4 , or CoCl 2 solutions, and the return of the Mn(H 20) 2 + signal was measured. The amount of Mn(II) displaced provides an estimate of the relative affinities of the competing metals under the conditions of the measurement. C. 'H-Nuclear Magnetic Resonance spectroscopy 1. Mn(Il)-myoglobin titrations 'H-NMR spectra was recorded at 20 °C with a Bruker MSL-200 spectrometer operating in quadrature detection mode at 200 MHz. A total of 20000 transients were collected over a 62.5 kHz bandwidth with 8192 data points. A superWEFT pulse sequence was used for solvent suppression (Inubushi & Becher, 1983). Chemical shifts were referenced to DSS (2,2-dimethyl-2-silapentane-5-sulfonate) through the residual water resonance. MetMb samples were exchanged into 50 mM deuterated sodium phosphate buffer (Merck and Co., Inc), pH 7.0 (uncorrected pH-meter reading). The protein solutions (400 uL of 1 mM Mb) were then titrated with 0 to 1.8 equivalents of M n S 0 4 solution (0.1 M) and the N M R spectra were recorded. Before Fourier transformation, the free 28 induction decay was multiplied by an exponential function to introduce 20 Hz line broadening. Equilibrium constants were not calculated do to the limited resolution of the instrument. 2. Reorientation kinetics The measurement of the rate constants for heme reorientation following reconstitution of apoMb with heme was performed as described previously (La Mar et al., 1984). As these variants have significantly increased rates of heme reorientation, the K C N quench method (La Mar et al., 1984) was employed to take advantage of the fact that heme reorientation occurs much more slowly in the metMbCN derivative. ApoMb was prepared as described above and exchanged into 50 mM deuterated sodium phosphate buffer (0.8 mM Mb, pH 7.0, uncorrected pH meter reading). Heme (0.9 equivalents), dissolved in NaOH (0.2 M), was added to the apoMb solution (25 °C) to start the reaction. The reconstitution mixture was sampled (400 uL) at various time intervals and quenched with 20 equivalents of KCN. The T l -NMR spectrum of each sample was recorded immediately. Each spectrum represents an average of 15000 transients with a spectral bandwidth of 38.5 kHz and 4096 data points. For the metMbCN derivative, apodization of the free induction decay introduced 5 Hz line broadening. The equilibrium constant, for heme disorder (ATEQ) can be defined as KEQ = k{/kb = {Ml]J[m1]„ (2) where and are the forward and reverse rate constants for heme reorientation, respectively, and M j and ml are the intensities of the 5-methyl resonance of the major orientation or the 8-methyl resonance of the minor orientation, respectively (La Mar et al., 1984). As the resonances of the two orientations in some Mb variants were not well resolved, the intensities of the two peaks were determined with the curve fitting function of the program Grams/386 (Galactic Industries) utilizing a combination of Gaussian and Lorentzian line shapes. A plot of ln{(A, - A»)/(A 0 - A J } vs time (t) 29 has a slope equal to -kohsd where A, = M,/(Mi + n^), and where A„ and A„ are the initial and equilibrium ratios of the two orientations, respectively (La Mar et al., 1984). The rate constants for the forward (£ f) and reverse (kb) reactions were then calculated from the relationship kobsA = kf + &b. 3. pH titrations The pH dependence of the paramagnetically shifted resonances in the met and cyano forms of the wild-type, Lys45Glu and Lys45Glu/Lys63Glu variants was measured. The parameters used to collect the 'H-NMR spectra for the metMb and metMbCN samples are given above. The proteins (~1 mM) were exchanged into 0.1 M NaCl in 100% D 2 0 or 10% D2O/90% H 2 0 , and the pH was varied from pH 5.0 to pH 9.0 by adding HC1 or NaOH. To produce metMbCN, 10 equivalents of K C N was added. D. Potentiometric titrations Binding of Mn(II) to the wild-type and variant forms of Mb was studied by monitoring the change in proton binding to the.proteins that accompanies metal binding. The titration equipment has been described previously (Mauk et al., 1991). Standardized solutions of MnCl 2 (9988-2, E. Merck, NJ) or CoCl 2 (9986, E. Merck, NX), sold as standards for atomic absorption spectroscopy, were diluted gravimetrically to between 3.8 and 5.2 mM with KC1 solutions to give a final ionic strength of 17.2 or 100 mM. The diluted metal ion solution and the Mb solution (30 - 40 p M Mb in 17.2 mM or 100 mM KC1) were initially adjusted to the same pH. The protein solution was then titrated with the metal ion, and the resulting change in proton binding was quantified by recording the volume of standardized acid (HC1) or base (NaOH) solution required to return the solution to the initial pH. The digitized titration curves were fitted to equations for either a one-site (Mauk et al., 1991) or two-site 30 model (Mauk et al., 1994) by a non-linear least squares procedure (Scientist, MicroMath, Inc.) to obtain the binding constants (ATX and K2) and the stoichiometric number of protons released per mole of metal ion bound (q). Titrations were performed at pH values in the range of pH 5.5 to 7.0. E. Enzyme assays 1. Manganese peroxidase activity measurements The manganese peroxidase activities of wild-type and variant forms of Mb were determined by monitoring the rate of formation of the Mn(III)-malonate complex at 270 nm (Wariishi et al., 1992). The measurements were performed with a Cary-219 UV-visible spectrophotometer equipped with a thermostatted cuvette holder and a Lauda Model RC-3 circulating water bath. The spectrophotometer was interfaced to a microcomputer for data collection (OLIS, Bogart, GA). Because of the low levels of activity and very small absorbance changes, only limited kinetics measurements were performed. The initial rate of reaction (units/umol) was determined for each Mb variant under identical conditions (10 mM MnS0 4 , 0.1 mM H 2 0 2 , 0.2 u M Mb) in 0.1 sodium malonate buffer (Sigma), pH 6.2 at 25 °C. A unit of activity is defined as the increase in absorbance at 270 nm per minute. 2. Peroxidase activity measurements The peroxidase activity of each Mb variant was determined by measuring the rate of oxidation of ABTS (2,2-azinobis-(3-ethylbenzthiazoline-6-sulphonic acid) (Boehringer-Mannheim) in the presence of H 2 0 2 . Assays were performed in MES buffer (1=0.1 M , pH 6.0) at 25 °C, with a protein concentration of 2 u M and an ABTS concentration of 0.24 mM. The H 2 0 2 concentration was varied from 0.25 to 4.0 mM. The increase in absorbance at 414 nm (e = 36 mM^cm"1 ) in the first 30 31 seconds of the reaction was used to determine the initial velocity. The initial velocities were analyzed in terms of the peroxidase ping-pong mechanism (3) proposed by Dunford (1991), and the values for kx were obtained by a least squares fit of the data (MINSQ version 4.02, MicroMath, Inc.) to equation 3. 2 [Mb]0 / v = ( A3 [AH] 2 V1 + ( [ H 2 0 2 ] )-' (3) The dependence of peroxidase activity on pH was also studied over the pH range 5.0 to 8.0 with the buffer system containing 5 mM MES, 5 mM MOPS, 5 mM TAPS and 95 mM NaCl described above. Here, the rate of ABTS oxidation was measured using 0.98 mM H 2 0 2 and 0.19 mM ABTS at 25 °C. F. Heme dissociation assays 1. Heme dissociation rate measurements Kinetic analysis of heme dissociation from the wild-type and modified forms of Mb and cytochrome b5 was performed as described previously (Hargrove et al., 1994a) except that the EGs64Tyr/Val68Phe double variant of horse heart Mb was used rather than the corresponding variant of sperm whale Mb. Rate data were collected with a Cary Model 3 spectrophotometer that was equipped with a water-jacketed cell holder and a Lauda Model RC3 circulating, thermostatted water bath and interfaced to a microcomputer. The dissociation of heme from Mb and cytochrome b5 derivatives was monitored at 410 and 600 nm, respectively. Heme dissociation was monitored under pseudo-first order conditions with a His64Tyr/Val68Phe variant apoMb (90 uM) to Mb or cytochrome b5 (3 uM) ratio of 30:1 so that kobsi = k.H , where k.H is the rate constant for hemin dissociation (Hargrove et al., 1994a). The resulting data were fitted to a single exponential equation by a non-linear least squares procedure (Scientist, MicroMath, Inc.). Sodium acetate buffer (0.15 M) was used at pH 5.0 and 5.5, and sodium phosphate buffer (0.15 M) was used at pH 7.0. All assay 32 solutions contained 0.45 M sucrose to stabilize the His64Tyr/Val68Phe apoMb variant (Hargrove et al., 1994a). 2. Activation energy measurements The Arrhenius energy of activation of heme dissociation (E^ for each of the Mb and cytochrome b5 variants was determined from the temperature dependence of the heme dissociation rate constant. The dissociation rate constants were measured as described above over the temperature range of20 - 37 °C. The activation energy was determined by fitting the rate data to the Arrhenius equation (4) (MTNSQ, version 4.02, MicroMath, Inc.). In £ = In A - EA/(RT) (4) Kinetic data were collected at pH 5 to minimize the assay times. Heme dissociates from these proteins much more slowly at higher values of pH complicating the precise determination of rate constants due to the lengthy exposure of apoMb to elevated temperatures. G. Mass spectrometry 1. Gas phase dissociation voltage determination The stability of heme binding to Mb and cytochrome b5 in the gas phase was determined by coUisionally-activated dissociation (CAD) of the heme-protein ions in the ion sampling region of an electrospray triple quadrupole mass spectrometer (ES-MS) (Figure 7). Gas phase protein ions, formed by pneumatically-assisted electrospray, passed through a dry nitrogen curtain gas, a 0.25 mm sampling orifice and a skimmer before reaching a radio frequency only quadrupole (Q0). The voltage difference across the orifice-skimmer region (AV o s ) was varied from 30 to 150 V by changing the DC voltage on the orifice plate to dissociate heme-protein ions by CAD. The ions were then analyzed 33 for mass in quadrupole one (QI). The legend for Figure 7 provides the voltage settings used for this experiment. The relative abundance of the holoprotein (n)+ and the corresponding apoprotein (n-l)+ was measured, and the A V 0 S that produced 50% dissociation for each charge state was determined (charge states 14+ to 11+ for Mb and 10+ to T for cytochrome 65). The average of these four voltages (VD) for each variant protein was taken as the measure of the voltage required to dissociate the heme-protein complex. As the voltage required for heme dissociation was sensitive to sprayer position, the position was kept constant for all the measurements. The Mb samples were 5 pM protein in 40 uM Tris-HCl buffer, pH 8.0/10 % methanol, and the cytochrome b5 samples were 5 pM protein in 40 pM sodium phosphate buffer, pH 7.0/10% methanol. 2. Tandem mass spectrometry In the tandem mass spectrometry (MS/MS) experiments, holoMb precursor ions were mass selected in the first quadrupole (QI) and dissociated in the collision cell (Q2). The resulting fragment ions were then analyzed for mass in the third quadrupole (Q3). The argon gas pressure in the collision cell (2.2 mTorr) was monitored with a precision capacitance manometer. The legend of Figure 7 provides the experimental voltages used for this experiment. 34 . Sprayer (5500 V) ^Curtain(1000 V) 1 1 1 N 2 , i i . Orifice (140-250 V) Skimmer (110 V) Collision cell Exit plate Ar i Q0/Q1 Q1/Q2 Q2/Q3 ! t \ 1 / \ QO SR QI Q2 det. Q3 turbo pump turbo pump yi 6xl0"3torr 2 x 10"5 ton-rotary pump 0.8 torr Figure 7. Diagram of the triple quadrupole mass spectrometer. The DC voltages used at the sample interface and the pressures in the various compartments are marked on the diagram. The voltages used for the dissociation experiments on the other components were as follows: R0, 103 V; Q0/Q1, 100 V; SRO, 87.9 V; Rl , 102 V; Q1/Q2, 0 V; R2, 93 V; Q2/Q3, 90 V; R3, 93 V; Exit, 0 V. The voltages altered for the MS/MS experiment were: orifice, 180V; Q0/Q1, 102.3 V; SRO, 90.1 V;R2, 73 V; Q2/Q3, 35 V. 35 3. Data simulation As the determination of the dissociation voltage (V D) described above assumes loss of only a heme(l)+ species, a data simulation was undertaken to evaluate the effects of competing charged and uncharged heme loss on the calculation of V D . The experiment was simulated numerically starting with the initial experimental charge state distribution of holoMb and assuming a voltage difference for 50% dissociation of the holoMb ions (VyJ. The extent of reaction ( r j (the fraction of holoMb converted to apoMb) at each orifice-skimmer voltage difference (AV o s ) was calculated from the linear relation rx = ( 0 . 6 / V , / l ) x A V o s - 0 . 1 (5) This dependence assumes a minimum voltage is required to get any reaction and is therefore offset (-0.1). The intensity of each holoMb peak and each resulting apoMb peak was calculated by assuming 20 to 50% of the heme is lost as a neutral ion. Model peak distributions were generated for voltages covering the experimental range, and an "experimental" V D was calculated from the simulated data using the same method used in the experiment (Section Gl ) . The calculated V D was then compared to the voltage difference (V,/2) that gave 50% dissociation of the protein in the model. H. Azide binding studies I. Equilibrium dissociation constant determination The equilibrium dissociation constants for azide binding to wild-type and variant Mbs were determined by spectrophotometric titrations with the Cary 219 spectrophotometer. Mb solutions (10 uMMb, 20 °C in 0.1 M sodium phosphate buffer, pH 7.0, 50 m M HEPES/50 mM MnS0 4 , pH 7.0, or 50 mM HEPES/0.2 M NaCl, pH 7.0) were titrated with azide solutions and the decrease of the Soret band intensity was measured after equilibrium was acheived (-10 mins). The volume of azide 36 solution added (2 mM, 10 mM, 200 mM or 1 M azide) was kept to a minimum (0.5 - 2.0 pL/addition) to avoid significant sample dilution. The equilibrium dissociation constants (Kdapp) for most variants were determined by fitting the data to equation 6 (Ikeda-Saito et al., 1992). [MbN3] / [Mb] t o t = [N 3 - k e / (Kr +[N3-]free) (6) The fractional saturation, [MbN3] / [Mb] t o t , was determined from the absorbance change caused by azide binding to Mb divided by the total absorbance change observed at saturating concentrations of azide. The variants with reduced azide affinity, His64Thr and His64Ile, were not titrated to saturation due to the large changes in ionic strength that would occur at high azide concentrations. An alternative equation was derived assuming [N 3 ']fr«= [N 3 '] t ot • Abs = ( Ab S i + [N3-] t o t Abs f ) / ( KT + [N^L,) 0) In this equation, Abs; is the initial absorption in the absence of azide, Abs f is the final absorbance at azide saturation and Abs is the absorbance at each point. Equilibrium binding data were fitted with either equation 6 or 7 with the program MINSQ. 2. FTIR spectroscopy FTIR spectra were recorded with a Perkin-Elmer System 2000 FTIR spectrometer equipped with a liquid nitrogen cooled mercury cadmium telluride detector. An average of 250 scans was collected for each protein at 2 cm"1 resolution in the range 1900-2200 cm"1. Concentrated Mb solutions (~5 mM) in buffer were mixed with 2 pL of a 40 mM sodium azide solution (BDH Inc.) in the same buffer to give a final azide concentration of ~ 3.6 mM. Less than a molar equivalent of azide was used to minimize the intensity of the unbound azide peak. Buffers used were 0.1 M potassium phosphate, pH 7.0, 50 mM HEPES/50 mM MnS0 4 , pH 7.0, and 50 m M HEPES/0.2 M NaCl, pH 7.0. The protein solutions were measured in a cell (CaF2 windows and a 0.05 mm spacer) 37 thermostated at 20 °C with a water-jacketed cell holder (Specac, Inc.) connected to a circulating water bath (Lauda Model RS3). The absorption spectra were obtained from the difference between the spectra of the MbN 3 complexes and the spectrum of wild-type Mb in the absence of azide. In addition, the absorption spectra of free azide with and without Mn(II) were measured. The resulting infrared absorption spectra were fitted using the curve fitting function in Grams/386 (Galactic Industries). A linear combination of both Gaussian and Lorentzian line shapes and the minimum number of bands required to give an acceptable fit were used for analysis of the spectra. The and the linewidth of the unbound azide absorption band were fixed during the fitting process. I. Spectrophotometric determination of heme pKA The pKA of the water coordinated to the iron in the distal sixth position in metMb was determined by spectrophotometric pH titration. The Mb samples were exchanged into 0.1 M NaCl. NaOH (0.1 M) was used to vary the pH over the pH range of 6.0 - 11.5 and the resulting changes in the visible spectrum from 700 to 450 nm were monitored. The change in absorbance at 583 nm was plotted vs pH and fitted to a form of the Henderson-Hasselbach equation (8) with the program MTNSQ (Version 4.02, MicroMath, Inc.). Abs5 8 3 = (A + B x lO*"-^) / ( i o * " - ^ +1) (8) where A = [AT] , B = [AT] eA_ and [AT] = [A"] + [AH]. J. Spectroelectrochemistry The reduction potentials of Mb variants were determined at equilibrium by potentiometric titrations with an optically-transparent thin layer electrode (OTTLE) cell (Figure 37, Appendix B) 38 similar to that described previously (Heineman et al., 1975; Reid et a l , 1982; Lim, 1990). The OTTLE cell was machined from Lucite plexiglass, with two quartz plates forming the window (path length of -0 .2 mm). The working electrode was a semi-transparent gold mesh (500 lines/inch) (Buckbee-Mears Co., Minneapolis, MN). A saturated calomel electrode (Radiometer, REF401), standardized with quinhydrone, was used along with a platinum wire counter electrode. The temperature of the OTTLE cell was monitored with a copper-constantan subminiature thermocouple and a F L U K E Model 2175A digital thermometer. The water-jacketed OTTLE cell holder was constructed from aluminium and thermostated at 25 °C with a Lauda Model RC-3 circulating water bath to 25 °C. The applied potential was controlled by a E G & G Model 173 Potentiostat/Galvanostat. Protein concentrations between 100 - 150 p M were used in the following buffers: sodium phosphate buffer (1=0.1 M , pH 7.0), 50 mM HEPES buffer (pH 7.0) containing 50 mM MnS0 4 , or 50 mM HEPES buffer (pH 7.0) containing 0.2 M N a C l Ru(NH 3) 6 and 2-hydroxy-l,4-naphthoquinone were used as mediators at concentrations o f - 6 pM. The Ru(NH 3) 6 was recrystallized from commercial material (Alfa) (Pladziewicz et a l , 1973). Trace amounts of Rhus vernicifera laccase (a gift of Dr. David Thackeray) were used to minimize dissolved oxygen and catalase (300 units) (Sigma) was included to remove any hydrogen peroxide generated during the experiment. The absorption spectrum at each potential was recorded with the Cary-219 spectrophotometer described previously, and the change in the Soret band intensity of the reduced protein was monitored. The midpoint reduction potential (EM) was calculated by fitting the data to the Nernst equation (9) with the program MTNSQ (Version 4.02, MicroMath, Inc.). In this relationship, E° is the applied potential, A R and A„ are the absorbances of the fully reduced and fully oxidized protein, respectively, and A ; is the absorbance at the applied potential. E° = E M + 0.059 log{(AR - A, )/(Ai - A„)} (9) 39 The slope of a Nernst plot is a constant at a given temperature (0.059 V at 25 °C) and was used as a criterion for Nernstian behavior. The calculated reduction potentials were adjusted to the standard hydrogen electrode (SHE) scale as described by Dutton (1978). K. Circular dichroism spectroscopy Circular dichroism spectra were recorded with a Jasco Model J-720 spectropolarimeter (Jasco, Japan) equipped with a Neslab Model RTE-110 circulating water bath and a Neslab Model RS-2 remote sensor (Neslab Instruments Inc., New Hampshire). Both the spectropolarimeter and water bath were operated under computer control. The spectropolarimeter was calibrated with ammonium-tf-camphor-10-sulphonate (Aldrich Chem. Co.). Protein samples (10 uM Mb; 10 mM sodium phosphate buffer, pH 7.0) were placed into a cylindrical, water-jacketed quartz cell (0.1 cm path length), and the spectra were recorded from 190 to 250 nm at room temperature. The thermal stability of the proteins was measured by monitoring the ellipticity at 222 nm over the temperature range 40 to 85 °C with a heating rate of 50 °C/hr. The thermal unfolding curves were smoothed with the filter function of the Jasco 720 software. The midpoint melting temperature (TM) was determined from the first derivative of the resulting plot of ellipiticity vs temperature. 40 m. RESULTS A. Characterization of Mn(II) Binding to Myoglobin Variants 1. EPR Spectroscopy The isotropic six line EPR pattern of a freely rotating Mh(H 20) 6 2 + complex (55Mn 1=5/2) at room temperature is illustrated in Figure 8 A. The decrease in the intensity of the Mn(H 20) 6 2 + signal upon addition of protein can be used to determine the amount of Mn(II) bound to the protein. The broadening of the EPR signal of Mn(H20)6 2 + upon addition of 10 equivalents of the Lys45Glu Mb variant is illustrated in Figure 8B. 300 340 380 Magnetic Field (mT) Figure 8. EPR spectra of the paramagnetic Mn(H20)6 2 + ion (0.125 mM) in the (A) absence and (B) presence of 10 equivalents of Lys45Glu Mb (50 mM HEPES, pH 7.0,1=0.17 mM, 25 °C). 41 Table 2. Equilibrium binding constants (K^) for Mn(II), determined by EPR spectroscopy, of the Mb variants [50 m M HEPES, pH 7.0 (1=0.17 mM), 25 °C]. Protein KA ( M 1 ) Relative to W T wild-type 180 ±40 1 Lys45Glu 370 ±50 2.1 Lys45Glu/Lys63Glu 1730 ±300 9.6 Lys45Glu/Lys63Ser 410 ±50 2.3 Lys45 Glu/Lys63 Glu/Asp60Glu 1700 ± 200 9.4 His97Leu/Lys45Glu 370 ± 60 2.1 His97Leu/Lys45Glu/Lys63 Glu 1400 ± 200 7.8 Lvs45Glu/Lvs63 Glu/His97Leu/Ser92Ala 1900 ± 200 10.6 The equilibrium constant (K^) for Mn(IT) binding to each Mb variant (Table 2) was calculated from the EPR titrations using the Hughes-Klotz equation (1), assuming one binding site per Mb molecule (n=l). Wild-type horse heart Mb has an affinity constant for Mn(II) of 180 M"1. A 2-fold enhancement in the Mn(II) binding constant relative to wild-type Mb is observed in the variants possessing the Lys45Glu substitution. When the Lys45Glu and Lys63Glu substitutions are combined, a 10-fold increase in binding constant is observed. The other double variant, Lys45Glu/Lys63Ser, does not exhibit enhanced Mn(II) binding relative to Lys45Glu Mb, suggesting that a glutamyl residue is necessary at position 63 for enhanced Mn(II) binding. Substituting the aspartyl residue at position 60 with a glutamyl residue also does not increase the binding of manganese to Mb. The His97Leu substitution, located on the proximal side of the heme, has little effect on metal binding. The quadruple variant was constructed to remove all interactions between the heme 7-propionate and the apoprotein to evaluate the possibility that the propionate would rotate to the distal side and participate in metal binding. However, Mn(II) binding is not further enhanced in this variant. 42 Therefore, glutamyl residues at positions 45 and 63 are identified as the primary substitutions necessary to enhance Mn(II) ion binding to the surface of Mb. A competition experiment was performed with the Lys45Glu/Lys63Glu variant of Mb in order to determine the relative affinity of this protein for a number of different metal ions. Addition of one equivalent of the variant to the MnS04 solution (0.125 mM in 50 mM HEPES, pH 7.0, 25 °C) caused a decrease in the intensity of the free Mn(H 20) 6 2 + signal to ~77% of the original signal intensity. When one equivalent of CuS04 is added, the free Mn(II) ion signal increased to -83% of the original signal intensity. An increase in the EPR signal indicates the Cu(II) ion displaces the Mn(II) ion and binds to similar location. Addition of equivalents of either CoCl2 or CdS04 to the Mn(IT)/protein solution resulted in signal increases to -92% and 100%, respectively, of the original free Mn(n) ion signal intensity. Therefore, the relative stability of metal ion binding to this engineered site is Mn 2 + * Cu 2 + < Co 2 + < Cd 2 +. The pH dependence of Mn(II) binding to the Lys45Glu/Lys63Glu variant was also assessed over the pH range of 5.5 - 7.5 (1= 0.1 M, 25 °C). The affinity of this variant for Mn(II) decreases below pH 6.5 (Figure 9). The titration curve was fitted to a Henderson-Hasselbach function for one titratable species and a pATA value of 5.6 was calculated. This observation suggests that one of the carboxylate ligands involved in the Mn(II) binding site, either Glu45, Glu63, or the heme 6-propionate is protonating and possesses a relatively high pKA. The usual range of pKA values for carboxylate groups within proteins is 2 - 5.5 (Fersht, 1977). The 6-a' proton resonance of the metLys45Glu/Lys63Glu variant exhibits a pH dependent chemical shift as monitored by 'H-NMR spectroscopy (Results section D4) which also suggests the presence of a titratable group near this proton with a pKA value of 5.2. The Mn(II) binding constants measured in the pH titration were 43 lower than those measured above owing to the greater ionic strength used in the pH dependence experiments. 5.6 6.0 6.4 6.8 7.2 pH Figure 9. pH dependence of Mn(II) binding to the Lys45Glu/Lys63Glu Mb variant (5 mM MES/ 5 m M MOPS/5 m M TAPS/95 mM NaCl, 25 °C) as determined by EPR titrations. 2. Potentiometric titrations Mb variants demonstrating the greatest affinity for Mn(II) in the EPR experiments were further analyzed by potentiometric titration (Table 3). The resulting titration curves were fitted to either a one- or two-site model (e.g. Figure 10A and B). The data fit best with the two-site model indicating the presence of two Mn(JJ) binding sites on the surface of the Lys45Glu/Lys63Glu variant at pH 7.0 and an ionic strength of 17 mM. The equilibrium Mn(II) binding constants of the variants calculated from the titration curve (Table 3) confirm that a metal binding site has been created (KA on the surface of Mb and indicate the presence of an additional pre-existing metal binding site (K2). 44 [MnCLJ / [Mb] Figure 10. Potentiometric titration curve for the Lys45Glu/Lys63Glu variant (pH 7:0,1=0.017 mM, 25 °C) fitted to (A) one-site and (B) two-site models. 45 Table 3. Equilibrium constants for Mn(II) binding to the Mb variants at selected values of ionic strength (I) and pH (25 °C) as determined by potentiometric titration. p H I (mM) K\ ( M 1 ) K2 (Mr1) Wild-type 17.2 _ 690 ± 20 6.92 q=0.226 ± 0.003 Lys45Glu/Lys63Glu 6.96 17.2 12800 ± 400 q=0.233 ± 0.005 620 ± 20 q=0.780 ± 0.007 5.99 17.2 3100 ±500 q=0.19±0.04 380 ±80 q=0.66 ± 0.02 5.49 17.2 522 ± 40 q=0.512± 0.003 5.98 100 300 ± 10 q=0.66 ± 0.02 Co(II) 5.97 100 10000 ± 3000 q=0.032 ±0.004 350 ±20 q=0.98 ± 0.02 Lys45Glu/Lys63Glu/Ser92Ala/His97Leu 6.92 17.2 14800 ± 600 q=0.229 ± 0.006 720 ± 30 q=0.815± 0.006 46 This pre-existing lower affinity Mn(II) binding site was also observed in wild-type Mb. The affinity of the genetically engineered site is - 2 0 fold greater than that of the site inherent to the wild-type protein. Both of these metal binding sites exhibit decreased affinity for Mn(II) when the ionic strength is increased or the pH is decreased. The titration curves for the Lys45Glu/Lys63Glu variant obtained at pH 5.5 or in 100 mM KC1 fit best to the one-site model. Presumably, the affinity of the endogenous metal site under these conditions is too low to detect and binding is only observed at the engineered site. Titration of the Lys45Glu/Lys63Glu variant with Co(II) indicated the affinity of the engineered site for Co(U) was -30 fold greater than for Mn(II) which is in agreement with the relative affinities determined in the EPR competition experiment (Results section A l ) . At pH 7.0, the stoichiometric number of protons released per mole of metal ion bound (q) is low for the engineered site (Table 3), presumably due to a small number of protonated carboxylates. At lower pH, however, increased proton release is observed due to the increased amount of protonation of the carboxylate ligands involved in the binding site. The apparent discrepancy between results derived from the potentiometric and EPR titrations merit comment. First, one binding site per protein was sufficient for fitting the binding data obtained from the EPR experiments, although the potentiometric titrations later unambiguously established the presence of two metal binding sites per Mb molecule at low ionic strength. Agreement to within one order of magnitude was obtained with the EPR experiment when one site was assumed in the calculation from the potentiometric titration data. Analysis of the EPR data in terms of more than one binding site was not warranted owing to the small portion of the binding curve measured in the EPR titration. This feature of the EPR experiment means that the second site is not adequately represented in the data because the Mn(II)/Mb ratios used in the experiment were all less than one. Mn(II) ions will only start binding to the low affinity site in an appreciable amount when the Mn(II)/Mb ratio is 47 greater, as used in the potentiometric titration. Secondly, during the potentiometric titration, the protein concentration and ionic strength were held constant throughout the experiment. In the case of the EPR titration, the protein concentration increased at least 10 fold. The contribution of protein concentration and charge to the effective ionic strength of a solution is not well understood. Mb has a large number of charged groups (21 lysyl and arginyl residues and 22 aspartyl and glutamyl residues), yet the net charge on the protein is low. For example, at pH 7.0, the net charge on horse heart Mb is < 1 based on Tanford-Kirkwood calculations, and the net charge on sperm whale Mb is ~2 based on experimental titration curves (Shire et al., 1974). Good agreement between the two methods was found at higher ionic strengths where the ionic strength changes due to increasing protein concentration in the EPR experiments would be less significant. Calculations and experiments done previously (Linse et al., 1995) show that the stability of Ca 2 + binding to calmodulin and calbindin can vary significantly as protein concentrations range between 0.1 to 7 mM. Although accurate determination of a binding constant using the EPR method is complicated by the high protein concentration required for measuring the Mn(II) binding constant of Mb, this method does allow for the assessment of the relative binding affinities of a series of variants. The potentiometric titrations provide a more accurate method for assessing the binding of metal ions, as a much greater portion of the binding curve is monitored, and the protein concentration is held constant. 3. 'H-NMR Mn(Il) titrations The location of the Mn(II) binding site on the surface of the Mb variants was determined by 'H -NMR spectroscopy. The paramagnetic Mn(II) ion causes broadening of the proton resonances in proximity to the metal binding site through dipolar coupling between the Mn(II) ion and the adjacent nuclei (Bertini & Luchinat, 1986; Banci et al., 1993). The paramagnetic center affects the 48 transverse relaxation rate (T2) and thus the linewidth of a proton resonance in a fashion that is inversely proportional to the distance (r6) between the Mn(II) and the proton. The paramagnetically shifted heme resonances of native metMb have been assigned previously for sperm whale Mb (La Mar et al., 1980) and have been assigned here to horse heart Mb as the 'H -NMR spectra of both proteins are very similar. The low field region of the 'H -NMR spectrum of the Lys45Glu/Lys63Glu Mb variant with various concentrations of Mn(II) is shown in Figure 11. The greatest broadening is observed for the resonance assigned to the 6-a proton of the heme 6-propionate, indicating Mn(II) binds close to this proton. Broadening is also observed for the 6-a ' proton resonance; however, it is difficult to quantify the magnitude of the broadening due to the similar chemical shift of the 4-a proton resonance of the 4-vinyl heme substituent. The 5-CH 3 heme resonance also experienced some minor broadening. The structure of the heme prosthetic group is shown in Figure 2 (page 9). Similar patterns of broadening were observed when the following variants were titrated with up to 1 equivalent of Mn(II): Lys45Glu, His97Leu/Lys45Glu, His97Leu/Lys45Glu/Lys63Glu, and Lys45Glu/Lys63Glu/His97Leu/Ser92Ala Mb. Evidently these variants bind Mn(II) in a similar location. In similar experiments, the N M R spectrum of wild-type Mb exhibits minimal broadening of the heme resonances upon addition of Mn(II). Therefore, the replacement of Lys45 and Lys63 with glutamyl residues created a site for Mn(II) binding near pyrrole III of the heme that was not present in the wild-type protein. This site involves the heme 6-propionate as indicated by the broadening of the 6-a and 6-a ' proton resonances. 49 X X X X u u u u I I I I 100 80 60 40 20 Chemical Shift (ppm) Figure 11. The 200 MHz Ti -NMR spectra (low field region) of the Lys45Glu/Lys63Glu Mb variant with increasing concentrations of Mn(II). The heme resonance experiencing the most severe broadening is labeled with a solid arrow. The resonances experiencing minor broadening are labeled with an open arrow. 50 4. Enzymatic activities The ability of the bound Mn(IT) ions to transfer electrons to the heme iron of Mb was assessed by measuring the manganese peroxidase activity of the Mb variants. Manganese peroxidase activity was assayed in the presence of malonate, as this diacid facilitates Mn(III) dissociation from the protein and stabilizes Mn(III) in aqueous solution through formation of the Mn(III)-malonate complex (Wariishi et al., 1992). Because of the low level of activity of these variants, only limited kinetics were carried out. The initial rate of formation of the Mn(III)-malonate complex by each of the Mb variants in the presence of 0.1 mM H 2 0 2 and 10 mM MnS0 4 was determined (Table 4). While wild-type Mb oxidizes Mn(II) relatively slowly under these conditions, variants containing the Lys45Glu substitution were found to have increased rates of Mn(III)-malonate formation, ~3 fold. Similarily, the His97Leu variant also showed an ~4-fold enhancement in rate. Combining these two substitutions in the same protein yields a variant with ~6-fold greater rate of Mn(III)-malonate formation than the wild-type protein. An essential feature of the mechanism of this reaction is the oxidation of Mb by H 2 0 2 to produce a Fe(IV)=0 species and an unstable aromatic protein radical (King & Winfield, 1963; Yonetani & Schleyer, 1967). The kinetics of the reaction of Mb with H 2 0 2 in the presence of ABTS were investigated and analyzed with the peroxidase ping pong mechanism (3) (Table 5). ABTS, functioning as a substrate, undergoes a one-electron oxidation to form a radical cation with an absorption maxima at 414 nm. The radical cation disproportionates in a time dependent manner, but this process is too slow to affect the initial rate measurements (Childs & Bardsley, 1975). As seen for the manganese peroxidase activity measurements, the Lys45Glu and His97Leu variants show small enhancements (2-3 fold) in the rate constant (kA over wild-type Mb for the reaction of Mb with H 2 0 2 . The combination of the two substitutions are additive and yield a double 51 Table 4. Rate of Mn(III)-malonate formation for the Mb variants measured in 0.1 M malonate, pH 6.2 (25 °C, 10 mM MnS0 4 , 0.1 mM H 2 0 2 , and 2 u M Mb). Variant Rate (units/umol) Relative to wild-type wild-type 14 ± 2 1 Lys45Glu 44 ± 7 3 Lys45Glu/Lys63Glu 51 ±8 3.5 Lys45Glu/Lys63Glu/Asp60Glu 41 ± 2 3 His97Leua 59 ±5 4 His97Leu/Lys45Glua 81 ± 2 6 His97Leu/Lys45Glu/Lvs63Glua 97 ±5 7 a measured by L. Eltis. Table 5. Rate constants for the formation of compound I (kt) for the Mb variants in MES buffer, pH 6.0 (25 °C, 1=0.1 M , 0.24 mM ABTS, 2 u M Mb, and 0.25 - 4.0 mM H 2 0 2 ) . Variant ( M 1 sec1) Relative to wild-type wild-type 430 ± 20 1 Lys45Glu 1270 ±50 3 Lys45Glu/Lys63Glu 1050 ±70 2.5 Lys45 Glu/Lys63 Glu/Asp60Glu 920 ± 30 2 His97Leua 770 ± 70 2 His97Leu/Lys45Glua 2180 ±70 5 His97Leu/Lys45Glu/Lys63Glua 2400 ± 100 6 a measured by L. Eltis. 52 variant with a further increase in the rate constant for compound I formation. Increases in the rate constants (£ t) determined in the presence of ABTS show a rough correlation with the increase in manganese peroxidase activity of the Mb variants. The pH dependence of ABTS turnover in the presence of H 2 0 2 for wild-type Mb and the Lys45Glu and Lys45Glu/Lys63Glu variants was determined over the pH range 4.7 - 7.5 (Figure 12). The extinction coefficient of the ABTS radical cation does not vary with pH over this range. For all proteins, the rate of ABTS turnover increases as pH decreases (Figure 12). The pH optimum for wild-type Mb was determined to be pH 5.0 while the optimal pH for the variants is <4.7. 200 B S 100 ti Figure 12. pH dependence of the ABTS turnover by wild-type Mb (•) and the Lys45Glu (o) and Lys45Glu/Lys63Glu (•) variants (0.98 mM H 2 0 2 , 0.19 mM ABTS, 25 °C). 53 B. Effects of hydrogen bonding interactions on heme binding to myoglobin 1. Heme binding disorder Comparison of the 'H -NMR spectra obtained at equilibrium (pH 7.0, uncorrected pH meter reading) for the metMb-CN derivatives of the Mb variants (spectra not shown) indicates that the ratios of the major and minor heme orientational isomers estimated from the intensity of the methyl resonances (M, and nij, respectively) are similar to those previously observed for sperm whale and horse heart Mb (La Mar et al., 1983). The equilibrium constant for the heme disorder equilibrium (A'EQ) was found to be -12 for all of the Mb variants included in this study. Addition of K C N to each of the freshly reconstituted heme/apoMb mixtures establishes in each case that equal amounts of the two orientational isomers are present immediately after addition of heme for the wild-type and variants of horse heart Mb (spectra not shown) as found for the wild-type sperm whale protein (La Mar et al., 1984). 2. Kinetics of heme reorientation The rate constants for heme reorientation in sperm whale Mb have been shown previously to be highly pH dependent and to exhibit a minimum near neutral pH (La Mar et al., 1984). For this reason, heme reorientation kinetic studies were most conveniently performed at pH 7.0. The 'H -NMR spectrum of the metMb-CN derivative of the Lys45Glu variant was monitored as a function of time immediately following reconstitution with heme (Figures 13 and 14). The resonances labelled M x and m1 correspond to the heme methyl peaks previously assigned in the spectrum of sperm whale metMb-C N to the major and minor heme orientations, respectively (La Mar et al., 1983) and are identified here by analogy. The ratio of major to minor forms is approximately 1:1 immediately after reconstitution (Figure 13 A). With time, the intensity of M b the 5-methyl heme resonance of the major 54 heme orientation, increases at the expense of m l 5 the 8-methyl heme resonance of the minor heme orientation, as the minor form converts into the major form, and equilibrium is achieved in -21 hours (Figure 13D). The other resonances that change in intensity with heme reorientation are also indicated in Figure 13. The rate constants for heme reorientation determined from these experiments for the Mb variants are shown in Table 6. Wild-type horse heart Mb exhibits a value for k{ that is slightly greater than that previously reported for sperm whale Mb reconstituted with heme at pH 6.4 (La Mar et al., 1984). For the variants in which an amino acid substitution was expected to eliminate one of the hydrogen bonds formed with the heme propionate groups (Lys45Glu, Lys45Glu/Lys63Glu, His97Leu), both k{ and k^ increased 11-18 fold. Nonspecific electrostatic effects are not a factor in this process as replacement of Lys63, which is not involved in hydrogen bonding interactions with the heme, does not exhibit any significant influence on the kinetics of heme reorientation. The Ser92Asp variant shows a 3-4 fold enhancement in rate constant over the other single variants. The somewhat greater influence of this substitution may result from the disruption of the additional hydrogen bond normally formed by Ser92 with the proximal His93 ligand (Shiro et al., 1994; Lloyd et al., 1996). Elimination of two of the propionate hydrogen bonds in the double variants, Lys45Glu/His97Leu and Ser92Asp/Lys45Glu results in increases in rate constant of-120 and 390-fold, respectively. The largest increase in rate constant (-1100-fold) is seen for the quadruple variant, Lys45Glu/Lys63Glu/His97Leu/Ser92Ala, which achieves equilibrium in <20 minutes. The rate constant for heme reorientation in the His64Tyr/Val68Phe double variant is similar to that of the wild-type protein. Similar measurements were not attempted with the cytochrome b$ derivatives due to the faster rate of heme reorientation (Singh & Wilson, 1990) in this protein and to limited amounts of material. 55 i i 1 1 1 30 20 10 0 -10 Chemical Shift (ppm) Figure 13. The ^ - N M R spectra (200 MHz) of the Lys45Glu variant of horse heart cyano-metMb (0.5 mM) in deuterated sodium phosphate buffer (p = 0.1 M , 20 °C, pH 7.0 (uncorrected pH meter reading)). The apoMb was reconstituted with heme (0.9 equivalents) at 25 °C, samples were removed, and the reaction was stopped with K C N at (A) 1.5 mins, (B) 31.5 mins, (C) 81.5 mins, and (D) 21 hr. The peaks corresponding to the two heme orientational isomers are labeled as follows: M , denotes the 5-methyl heme resonance of the major orientation (28.0 ppm) and m, denotes the 8-methyl heme resonance of the minor orientation (28.4 ppm). 56 0 Time (min) Figure 14. Kinetics of heme reorientation of the Lys45Glu variant of horse heart Mb with heme at 25 °C, pH 7.0 as monitored by 'H -NMR spectroscopy. Substitutions of Lys45, Ser92 and Ffis97 with residues unable to form hydrogen bonds with the heme propionates have significant effects on the rate constants for heme reorientation due to elimination of the hydrogen bonding interactions to the heme propionate substituents. As each residue is replaced, large increases in the rate constant for heme reorientation (~ 10-fold) are observed, and the effects of each substitution are multiplicative. Replacing any of these three residues with residues unable to interact with a heme propionate allows the heme to equilibrate much more quickly within the heme binding pocket. Therefore, the hydrogen bonding interactions appear to play a role in the initial trapping of the minor orientation. The increases in reorientation rate constant are greater for position 92 variants, presumably because such substitutions eliminate a hydrogen bond to the proximal histidine residue in addition to the hydrogen bond to the heme propionate. 57 Table 6. Rate constants for heme reorientation in wild-type and variant forms of metMb (pH 7.0, 25 °C). Shown in parenthesis are rates normalized to wild-type Mb. Protein Acr(x 104 min 1 ) kb (x 104 min 1 ) sperm whale Mb wild-type 3.1 a horse heart Mb wild-type 6.2 ±0.7(1) 0.522 ±0.04(1) Lys45Glu 85 ± 4 (14) 7.0 ±0.2 (14) Lys45Glu/Lys63Glu 110 ± 10(18) 10 ± 1 (18) His97Leu 70 ± 10(11) 6.0 ±0.7 (11) Ser92Aspb 250 ± 20 (40) 21 ± 1 (40) His97Leu/Lys45Glu 750 ±50 (120) 62 ± 3 (120) Ser92Asp/Lys45Glu 2400 ± 200 (340) 200 ± 10 (340) Lys45Glu/Lys63 Glu/His97Leu/Ser92Ala 7000 ± 1000(1100) 580 ±60 (1100) His64Tyr/Val68Phe 15 ± 1 (2) 1.3 ±0.1 (2) a pH 6.4 (La Mar et al., 1984). b measured by E. Lloyd. 3. Kinetics of heme dissociation The electronic absorption spectrum of the His64Tyr/Val68Phe double variant of horse heart metMb is essentially identical to that of the corresponding variant of sperm whale Mb (Hargrove et al., 1994a) and is consistent with coordination of Tyr64 as the sixth ligand to the heme iron. Wild-type apoMb was found to exhibit a melting temperature (T^ of 63.2 °C while the corresponding value for the His64Tyr/Val68Phe variant was found to be 63.9 °C. The molar absorptivities of the Mb variants and derivatives studied were greater than that of the His64Tyr/Val68Phe variant, allowing the heme dissociation assays to be monitored at the Soret maxima of the variant or derivative (Figure 15). The absorbance changes observed (pH 5.0) for the dissociation of heme from 58 some of the Mb variants studied are shown in Figure 16. In each case, the data were adequately described by a single exponential function. The rate constants for heme dissociation determined in this manner at pH 5.0, 5.5 and 7.0 are set out in Table 7. 400 500 600 Wavelength (nm) Figure 15. Electronic absorption spectra of wild-type and E£is64Tyr/Val68Phe variant metMb at pH 7.0, 25 °C. At all values of pFL the rate constants for heme dissociation from the variant Mbs are greater than those observed for the wild-type protein (Table 7). The rate constants for heme dissociation increase at low pH due to the protonation of the proximal histidine and subsequent loss of the Fe-histidine bond (Giacometti et al., 1977, Coletta et al., 1985). These increases in the heme dissociation rate constants are quite dramatic as the pH is lowered below pH 5.5. The effect of pH on the behavior of the variants is difficult to interpret in detail, owing to the electrostatic nature of the substitutions involved and the potentially complex electrostatic consequences that each of these substitutions could exhibit. The rate constants measured at pH 5.0 were used for a quantitative assessment of the variants because of the difficulty in precise determination of the rate constants for 59 the slow heme dissociation processes observed at higher pH. Extended exposure of the apoprotein to elevated temperature (37 °C) during to the long assay times could result in problems of denaturation of the apoprotein. 0 10 20 30 40 50 Time (min) Figure 16. Absorbance changes at 410 nm observed upon mixing Mb variants (3pM) with the His64Tyr/Val68Phe variant apoMb (90pM) (sodium acetate buffer (0.15 M), 0.45 M sucrose, pH 5.0, 37 °C). Absorbance changes were normalized for comparison. At pH 5.0, the rate constant for heme dissociation from wild-type horse heart Mb is greater than that reported for sperm whale Mb under similar conditions (Hargrove et al., 1994a) but similar to recently reported values for horse heart Mb (Hargrove et al., 1996b). For the horse heart protein, elimination of a single hydrogen bond formed by either heme propionate group results in ~3-fold increase in the rate constant for heme dissociation as demonstrated by the results obtained for the Lys45Glu, Lys45Glu/Lys63Glu, Lys45Glu/Lys63Ser, Lys45Glu/Lys63Glu/Asp60Glu, and His97Leu variants. As observed for the heme reorientation kinetics, these effects are specific for hydrogen bonding interactions because the replacement of Lys63Glu, Lys63Ser or Asp60Glu does not produce 60 Table 7. Rate constants for heme dissociation (h.^) from wild-type, variant, and heme-substituted forms of Mb (37 °C). Shown in parentheses are rates normalized to wild-type Mb. Myoglobin Rate, k.H (x 102 min 1 ) p H 5.0 p H 5.5 p H 7.0 Sperm whale wild-type o.or 0.0002a R45E 0.9a 0.003a Horse heart wild-type 0.042" wild-type 3.6±0.1 (1) 0.6 ±0.1 0.10 ±0.02 K45E 10 ± 2 (3) 0.8 ±0.1 0.37 ±0.05 K45E/K63E 11 ± 1 (3) 00.74 ±0.01 0.23 ± 0.04 K45E/K63S 12 ± 1 (3) 0.95 ± 0.02 0.37 ±0.03 K45E/K63E/D60E 9.8 ±0.8 (3) 0.68 ±0.01 -H97L 12 ± 2 (3) 1.2 ± 0.1 0.5 ±0.1 S92D 24.1 ±0.4 (7) 1.3 ±0.1 3.0 ±0.01 H97L/K45E 35.5 ±0.9(10) 6.0 ±0.3 2.72 ±0.03 S92D/K45E 75.3 ±0.5(21) 7.8 ±0.1 2.5 ±0.3 DME-Mb 150 ±20 (41) 11.9 ± 0.1 1.8 ±0.1 K45E/K63E/H97L/S92A 240 ± 6 (66) 37.8 ±0.9 28.2 ±0.7 a Hargrove et al., 1994a. b Hargrove et al., 1996b 61 any change in rate constant relative to that observed for the Lys45Glu variant. The Ser92Asp variant exhibited a rate constant for heme dissociation that is ~ 7-fold greater than that of the wild-type protein, a greater increase than observed for the other two single variants. Again, the greater perturbation introduced by replacement of this residue may be attributed to the elimination of the hydrogen bond formed by Ser92 with the proximal His93 ligand to the heme iron as well as the hydrogen bond this residue forms with the heme propionate. As observed for the heme reorientation kinetics, the effects of removing individual heme propionate-apoprotein hydrogen bonds on the rate constant for heme dissociation appear to be multiplicative. This relationship is illustrated by the observation that the heme dissociation rate constant for the Lys45Glu/His97Leu double variant is ~3-fold greater than that for each of the single variants, Lys45Glu and His97Leu. Similarly, the Lys45Glu/Ser92Asp variant exhibits a ~3-fold greater rate constant for heme dissociation relative to that for the single variant, Ser92Asp, and the quadruple variant shows a -3-fold increase in rate constant over the Lys45Glu/Ser92Asp variant. Removal of each hydrogen bond involving a propionate group results in an ~3-fold increase in the rate constant for heme dissociation at pH 5.0. Although the esterification of the heme propionate groups in the DME-Mb derivative perturbs the electrostatics of the solvent-exposed heme edge significantly, esterification does not rule out the possible formation of a single hydrogen bond with an adjacent amino acid residue. In the absence of structural information, therefore, it is difficult to predict what hydrogen bonding interactions may be retained in this derivative. From the apparent quantitative relationship between heme propionate hydrogen bonding interactions and heme dissociation rate constants outlined above, however, it seems reasonable to infer that the -41-fold greater rate constant observed for DME -Mb relative to 62 Table 8. Rate constants for heme dissociation (k.^) from wild-type, variant and heme-substituted forms of cytochromes b5 (37 °C). Shown in parentheses are rates normalized to wild-type cytochrome b5. Cytochrome bs Rate, *. H (x 102 min *) pH 5.0 p H 5.5 p H 7.0 lipase-solubilized wild-type 7.1 ±0.9(1) 2.3 ±0.3 0.5 ±0.2 Ser64Ala 72 ± 5 (10) 47.7 ±0.5 22 ±3 trypsin-solubilized native 11.4 ±0.9(1) 3.2 ±0.1 0.38 ±0.03 DME-heme 61 ± 1 (6) 17.3 ±0.5 3.2±0.1 human erythrocyte native 30.3 ±0.2 5.2 ±0.1 0.28 ±0.01 wild-type represents a destabilization of heme binding equivalent to the loss of an equivalent of three hydrogen bonds. The molar absorptivity of cytochrome bs in the Soret region is similar to that of the His64Tyr/Val68Phe double variant of Mb with heme bound, so the dissociation of heme from cytochrome b5 was more conveniently monitored at 600 nm. The rate constants for heme dissociation from the various forms of cytochrome bs (Table 8) are uniformly greater than those observed for metMb at all three pH values. Cytochrome b5 exhibits a pH dependence of heme dissociation similar to Mb, again due to the protonation of the coordinating histidines. Replacement of Ser64, which hydrogen bonds to 7-propionate, with an alanyl residue increased the rate constant for heme dissociation ~10-fold relative to that observed for wild-type lipase-solubilized cytochrome b5. On the basis of the observations described above for the Mb variants, the magnitude of this increase suggests that neither of the hydrogen bonds normally formed by the propionate group are present in this 63 variant. The rate constant for heme dissociation from DME-substituted trypsin-solubilized cytochrome bs was ~6-fold greater than that observed for wild-type trypsin-solubilized cytochrome b5. In this case, the increase in rate constant is intermediate between the increase expected from the elimination of one or two hydrogen bonds on the basis of the results described above for Mb. 4. Activation energy of heme dissociation The stability of the heme-globin complex was also quantified by determining the energy of activation for heme dissociation. The rate constants for heme dissociation were measured at pH 5.0 over a temperature range of 20 - 37 °C with the method described earlier (Hargrove et al., 1994a). As previously observed, all kinetic data were described well by a single exponential function in this temperature range. The dependence of the rate constant for heme dissociation on temperature is illustrated in Figure 17 for the His97Leu/Lys45Glu variant. Corresponding results obtained for the other proteins used in this study also produced linear Arhennius plots over this temperature range (Figure 17, inset). The activation energies for heme dissociation (EJ and the pre-exponential factors (A) for the variant proteins are listed in Table 9. The E A values for wild-type Mb and cytochrome 65 were found to be 106 ± 7 and 116 ± 7 U/mol, respectively, at pH 5.0. As the heme-globin complex is stabilized at higher pH, it is expected that the activation energy of dissociation will increase. The activation energies for heme dissociation of horseradish peroxidase and leghemoglobin were previously determined to be 139 ± 10 and 132 ± 6 U/mol, respectively, at pH 7.0 (Smith et al., 1991), so the E A values determined here are comparable. 64 0 20 40 60 80 Time (mins) Figure 17. Temperature dependence of the rates of heme dissociation from His97Leu/Lys45Glu Mb (20 - 37 °C, pH 5.0). The original data shown here were measured at (a) 37 °C, (b) 34 °C, (c) 31 °C, (d) 26 °C, (e) 21 °C. The activation energy, E A , was derived from these data (inset) with the Arhennius equation, In k = In A - EA/(RT). The energies of activation of heme dissociation from the variant proteins (Table 9, Figure 18) decrease as the amino acid residues that form hydrogen bonds with the heme propionates groups are replaced. The Mb variants exhibited a decrease in activation energy of ~8 - 10 kJ/mol for each hydrogen bond that is lost. Surface hydrogen bonding interactions tend to be relatively weak owing to exposure to solvent, so these results are consistent with the energy of stabilization normally expected from a hydrogen bond, -10-40 kJ/mol (Fersht, 1977). In total, these three hydrogen bonds account for - 3 0 % of the activation energy for heme dissociation, as illustrated by the 30 kJ/mol decrease observed for DME-Mb. As observed in both the heme reorientation and heme dissociation kinetic experiments, variants in which Ser92 has been replaced exhibit lower activation energies than 65 expected (Table 9, Figure 18). Heme propionate-protein interactions are also important in stabilizing heme binding to cytochrome bs as reconstitution with the DME-heme also decreased the activation energy of heme dissociation (15 kJ/mol) (Table 9, Figure 18). The lipase-solubilized and trypsin-solubilized forms of cytochrome b5 exhibit similar activation energies as these proteins differ in size by just 11 amino acid residues at the N- and C-termini and are identical in the regions responsible for heme binding. Table 9. Activation energies (E^j and pre-exponential factors (A) for heme dissociation from wild-type and modified forms of horse heart myoglobin and bovine liver cytochrome b5. Protein E A (kJ/mol) In A (s-1) Myoglobin wild-type 106 ± 7 34 ± 3 Lys45Glu 97 ± 4 31 ± 2 His97Leu 93 ± 4 30 ± 1 Lys45Glu/His97Leu 87 ± 3 29 ± 1 Ser92Asp/Lys45Glu 81 ± 4 27 ± 2 DME-Mb 77 ± 5 26 ± 2 Lys45Glu/Lys63 Glu/His97Leu/Ser92Ala 73 ± 3 25 ± 1 Cytochrome b5 lipase-solubilized 116 ± 7 38 ± 3 trypsin-solubilized 115 ± 4 38 ± 2 trypsin-solubilized, DME 101 ± 4 34 ± 2 66 Number of Hydrogen Bonds Figure 18. Dependence of Arhennius activation energy for heme dissociation (E^ on the number of hydrogen bonds formed between the heme propionates and the protein. The Mb variants are labeled as follows: wild-type (•), Lys45Glu (*), His97Leu (•), Lys45Glu/His97Leu (•), Ser92Asp/Lys45Glu (*), DME-Mb (•), and Lys45Glu/Lys63Glu/His97Leu/Ser92A la (#) Mb. The cytochrome b5 variants are labeled as follows: lipase-solubilized (O), trypsin-solubilized (A) and DME-trypsin-solubilized (•) cytochrome b5. The number of hydrogen bonds (1 or 0) in DME-cytochrome b5 is still uncertain but is represented here as 1 for the purpose of the figure. 67 5. Thermal stability The thermal stability (7M) of the wild-type and variant Mbs was assessed by monitoring the loss of secondary structure by circular dichroism (Table 10). Wild-type Mb is a thermally stable protein that exhibits a TM of nearly 80 °C. All of the variants were slightly destabilized to thermal denaturation relative to the wild-type protein with the variant containing four substitutions destabilized the most (-13 °C). Nevertheless, all of the variants studied here retain a high degree of thermal stability as the amino acid substitutions involved disturb the overall protein structure very little (Results section D2). The interactions between the heme propionate groups and the protein residues evidently play a small role in maintaining the thermal stability of Mb because successive elimination of these interactions results in incremental decreases in melting temperatures (Figure 19). The melting temperatures observed for lipase-solubilized and trypsin-solubilized bovine liver cytochrome b5 (Table 10) are lower than that for Mb. Replacement of Ser64 with Ala produces a 14 °C decrease in TM relative to the wild-type protein in lipase-solubilized cytochrome b5. Esterification of the heme propionate groups in the trypsin-solubilized protein also decreases the thermal stability by 9 °C. As observed for Mb, the interactions between the heme propionates and the protein residues play a relatively small role in determining the thermal stability of cytochrome b5. 68 Table 10. Melting temperatures ( 7 ^ of the myoglobin and cytochrome b jderivatives at pH 7.0 (sodium phosphate buffer, 1=0.01 M) determined by monitoring the Cotton effect in the CD spectrum at 222 nm. Protein TM(°Q Myoglobin wild-type 79 ± 1 Lys45Glu 74.7 ±0.5 Lys45Glu/Lys63Glu 73 ± 1 Lys45Glu/Lys63Ser 73.8 ±0.1 Lys45Glu/Lys63Glu/Asp60Glu 74 ± 1 His97Leu 76.5 ±0.5 His97Leu/Lys45Glu 72.0 ±0.3 His97Leu/Lys45Glu/Lys63Glu 70 ± 1 Ser92Asp/Lys45Glu 71 ± 1 Lys45Glu/Lys63 Glu/His97Leu/Ser92Ala 65.4 ±0.5 Cytochrome b5 lipase-solubilized 66 ± 1 Ser64Ala lipase-solubilized 52 ± 1 trypsin-solubilized 63 ± 2 DME-trypsin-solubilized 54 ± 1 69 Figure 19. The dependence of melting temperature (T^ on the number of potential hydrogen bonding interactions formed between the apoMb and the heme propionate groups: wild-type (•), Lys45Glu (*), Lys45Glu/Lys63Glu (A), Lys45Glu/Lys63Ser (o), Lys45Glu/Lys63Glu/Asp60Glu (•), Ffis97Leu (•), His97Leu/Lys45Glu (•), His97Leu/Lys45Glu/Lys63Glu (+), Ser92Asp/Lys45Glu (*), and Lys45Glu/Lys63Glu/His97Leu/Ser92Ala (#) Mb. 70 6. Mass spectrometry Mass analysis of the Mb and cytochrome b5 variants was performed with an electrospray triple quadrupole mass spectrometer. The mass spectrum of the Ser92Asp/Lys45Glu variant (Figure 20) is representative of the results obtained. For the recombinant Mb proteins, two peaks, differing by 131 amu, were identified for each charge state (Figure 20, inset). This difference corresponds to the mass of a methionyl residue and indicates incomplete processing of the JV-terminal methionyl residue when Mb is over-expressed in E. coli. The extent of processing varied from preparation to preparation. The region of the spectrum containing the heme peaks (centered around 616 amu) consists of a series of peaks that correspond to the isotopic peaks of carbon and iron. Highly charged protein peaks, centered around 800 amu (Figure 20), indicate a small amount of denatured protein was present in solution. Denatured proteins have been shown to be more highly charged in the gas phase than folded proteins (Chowdhury et a l , 1990; Katta & Chait, 1991; Feng & Konishi, 1993). Interestingly, no shift in the charge state distribution of the variants was observed despite the removal of as many as three basic residues. The stability of heme binding to Mb and cytochrome b5 in the gas phase was measured by increasing the voltage difference (AV o s ) across the orifice-skimmer region (Figure 7, Methods section G) of the ion sampling interface to increase the collisionally-activated dissociation (CAD) of the non-covalently associated heme-protein complexes. The mass spectra of wild-type Mb measured at a several A V o s values are shown in Figure 21. No significant changes in the charge state distribution were observed at higher A V o s other than the shift resulting from the loss of heme (1)+. As the voltage difference is increased, the apoprotein peaks increase at the expense of the holoprotein peaks. Also, the intensity of the free heme (1)+ peak increases as heme dissociation increases. The holoprotein of charge state (n)+ loses heme principally as a singly charged positive ion and produces an apoprotein 71 with charge (n-l)+. Due to the difference in ion transmission through the interface at different A V o s , the relative abundance of each charge state pair was used. The relative abundance of the holoprotein (n)+ and apoprotein (n-l) + for all charge states was found to be linearly dependent on the A V o s (Figure 2 ID). The voltage difference required for 50% dissociation of the heme-protein complex for each of the four most abundant charge states of the Mb and cytochrome b5 variants (A V,/2) was determined from plots similar to that depicted in Figure 21D. The AV,/3 values for each charge state of wild-type and Lys45Glu/His97Leu Mb are shown in Figure 23 A. The A V M for the Lys45Glu/Ffis97Leu variant decreases -13 V relative to that of wild-type Mb for each charge state. As the AV,/2 values decrease with increasing charge in a similar fashion in all the variants, the average voltage for all four charge states (V D) was taken as a measure of the relative energy required to dissociate the heme-protein complex. The gas phase dissociation voltages (V D) for each variant are given in Table 11, along with the products of charge state and dissociation voltage (AV,/2). 72 Figure 20. Electrospray mass spectrum of the Ser92Asp/Lys45Glu Mb variant in 10% MeOH with a orifice-skimmer voltage difference of 60 V. The inset expands the region of the spectrum for the holoprotein peak with a charge state of+13. 73 d a Pi heme (+1) all a!2 alO 600 1000 1400 1800 m / z J L J L J L 0 45 65 85 105 Orifice-Skimmer Voltage Difference (V) Figure 21. Mass spectrum of wild-type Mb at A V o s values of (A) 40 V, (B) 100 V, and (C) 150 V. The holoprotein peaks (hl4-hl 1) decrease and the apoprotein peaks (al3-al0) increase as the voltage difference between the orifice and skimmer is increased. (D) Linear dependence of heme-protein complex dissociation on the orifice-skimmer voltage difference (AV o s ) for the +13 holo (•) / +12 apo (o) charge states for wild-type Mb. 74 Table 11. Average dissociation voltages (V D), charge state (q) * dissociation voltage (AV H ) and the number of atoms for wild-type and modified forms of horse heart myoglobin and bovine liver cytochrome bs. Protein V D (V) q x AV* (V) # of atoms Myoglobin a wild-type 85 ± 2 1040 ±30 2486 Lys45Glu 80 ± 1 990 ± 10 2481 His97Leu 81 ± 1 1000±10 2488 Lys45Glu/His97Leu 72 ± 2 890 ± 30 2478 Ser92Asp/Lys45Glu 89 ±3 1100 ±40 2483 DME-Mb 66 ± 1 830 ± 10 2514 Lys45Glu/Lys63Glu/His97Leu/Ser92Ala 78 ± 4 970 ± 60 2477 Cytochrome bs b lipase-solubilized 65 ± 1 550 ± 10 1556 trypsin-solubilized 65 ± 1 540 ± 10 1386 trypsin-solubilized, D M E 58 ± 2 480 ± 20 1414 a 5 p M Mb in 40 p M Tris-HCl buffer, pH 8.0/10% methanol b 5 p M cytochrome bs in 40 p M sodium phosphate buffer, pH 7.0/10% methanol This procedure for determination of the A V o s required for %50 heme dissociation from the holoprotein is exact only when heme is dissociated from the complex entirely as a +1 charged species. MS/MS experiments, where the heme-protein complex was dissociated by CAD in the collision cell, indicated that -20-30% of heme is dissociated as an uncharged species for each of the charge states (Figure 22) (Li et al., 1993). Loss of uncharged heme from the (n)+ charge state produces the same molecular ion for apoMb asloss of heme (1)+ from the (n+l)+ charge state. Because of the different abundances of the (n)+ and (n+l)+ holospecies, this dissociation of uncharged heme can cause 75 differences in the apparent A V o s for individual charge states that do not derive from differences in the binding energy of heme. 0 s -CO ti to > 100 75 50 25 0 heme +1 600 apo +9 apo +10 holo -trio yH Ii i l n l l l II,.nil] 1000 1400 1800 m/z Figure 22. MS/MS spectrum of the +10 holoprotein precursor ions peak dissociated by CAD in the collision cell (Q2) containing argon gas (2.2 mTorr). To assess the potential contribution of dissociation of uncharged heme to the dissociation voltage for the individual charge states (AV,/2) and to the experimentally determined average dissociation voltage (V D) for the 11+ - 14+ charge states of Mb, the experiment was simulated numerically. The results of the simulation, assuming dissociation of 30% of the heme in an uncharged state and initial dissociation voltages (V D) of 88 and 75 V, are shown in Figure 23B. The apparent decrease in AV,/2 for the higher charge states in the data simulation (Figure 23B) is similar to the decrease observed experimentally (Figure 23 A). This decrease reflects the shape of the initial intensity distribution of the holoMb peaks and the dissociation of neutral heme. The importance of using the 76 Charge State Figure 23. Dependence of dissociation voltage (AV,/2) on charge state. (A) The experimental dissociation voltages of the charge states for wild-type (•) and His97Leu/Lys45Glu ( A ) Mb. (B) The dissociation voltages for the charge states of simulated data. The two sets of distributions were generated by assuming the initial experimental charge state distribution of Mb, average dissociation voltages of 88 V (o) and 75 V (A), and the dissociation of 30% of the heme in a neutral state. 77 same charge states for calculation of the average dissociation is also illustrated. Although there are significant differences in the dissociation voltages for individual charge states, the average dissociation voltages (VD) for the 11+ to 14+ charge states calculated from the two simulated data sets are 86 and 73 V, within ±2 V of the initial dissociation voltages (V D) assumed in the simulations. Additional simulations for dissociation of 0 to 50% uncharged heme resulted in average voltages within ±4 V of the V D value assumed initially. While the loss of uncharged heme causes the various charge states to exhibit significantly different AV,/2 values, the voltage (V D) averaged over individual charge states is remarkably insensitive to the extent of uncharged heme loss and so can be regarded as a measure of the voltage required to dissociate the complex. The average dissociation voltages (V D) determined experimentally for the forms of Mb and cytochrome b5 are reported in Table 11. In comparing the V D for each Mb variant with that of wild-type, it was observed that the elimination of a single hydrogen bond to a heme propionate group in Mb lowered V D by 4-10 V. Similarly, the average voltage required to dissociate 50% of the heme from trypsin-solubilized DME-cytochrome b5 is 7 V lower than that required for trypsin-solubilized cytochrome bs The dissociation voltage (V D) for each Mb and cytochrome b5 variant is a measure of the relative internal energy required to cause the complex to dissociate within the transit time to the mass analyzing quadrupole, QI (~2 ms). Therefore, differences in stability of a non-covalently associated complex in the gas phase that result from eliminating a single hydrogen bonding interaction can be observed with this method. As seen in Figure 24, the dissociation voltages obtained in the gas phase (solid symbols) correlated well with the activation energies observed in solution and decrease in the following order: wild-type > Lys45Glu > His97Leu > Lys45Glu/His97Leu > DME-Mb. A correlation suggests that these protein-propionate interactions that occur in solution are retained upon ionization and provide 78 the same relative degree of stabilization to the heme-globin complex in both phases. Both lipase- and trypsin-solubilized cytochrome bs exhibited similar dissociation voltages in the gas phase and similar activation energies for heme dissociation in solution. Esterification of heme propionate groups of cytochrome b5 decreases the activation energy for heme dissociation in solution and also decreases the voltage required to dissociate heme from the holoprotein in the gas phase (Figure 24). Again, this behavior is consistent with the persistence in the gas phase of the known protein solution structure. Two of the Mb variants, Ser92Asp/Lys45Glu and Lys45Glu/Lys63Glu/His97Leu/Ser92Ala, both of which involve replacement of Ser92, exhibit a dissociation voltage greater than expected (Figure 24, stars), suggesting that this substitution actually stabilizes the complex in the gas phase. In solution, however, the variants possessing a substitution at position 92 are destabilized. Additional work is required to understand the effects of other types of bonding interactions, such as axial ligation, on gas phase stability. 79 95 -Myoglobin S92D/K45E(1) WT(3V 97L(2) ' K45E(2) K45E/H97L(1) 80 90 E A (kJ/mol) 70 65 60 55 95 Cytochrome b5 Tryptic(2) | 1 r V V | 1 Lipase(2) DME(l/0) i i 105 E A (kJ/mol) 115 Figure 24. The average dissociation voltage (V D) measured in the gas phase is plotted against the activation energy for heme dissociation (EA) measured in solution. The numbers in paraenthesis represent the number of hydrogen bonds present between the heme propionates and the protein. 80 C. Effects of amino acid substitutions on ligand binding 1. Azide binding The equilibrium dissociation constants (X d a p p) for binding of the anionic azide ligand, N3", to the distal coordination position of the heme iron of the Mb variants were determined by titrating each protein with azide and monitoring the change in the absorption spectrum spectrophotometrically (Figure 25). The absorbance change at the Soret maximum of the oxidized protein was used to determine the fractional saturation, |M)N 3]/[Mb] t o t, of the protein with azide. The equilibrium dissociation constants (Table 12) were then determined by fitting the data to equation 6 or 7. Wavelength (nm) Figure 25. Electronic absorption spectra of the Ser92Asp myoglobin variant in the presence of varying concentrations of azide (0.1 M sodium phosphate buffer, pH 7.0, 20 °C). The fit of the data to equation 6 is shown in the inset. 81 Table 12. Equilibrium dissociation constants (Xd a p p) for azide binding to wild-type and variants of oxidized myoglobin (0.1 M sodium phosphate buffer, pH 7.0, 20 °C). Variant Kdapp (uM) A d a p p (protein) (wild-type) wild-type 20 ± 1 1 Sigma wild-type 26.0 ±0.5 1.3 Lys45Glu 99 ±3 5 Lys45Glu/Lys63Glu 280 ± 10 14 Lys45Glu/Lys63Ser 160 ± 10 8 His97Leu 27.4 ±0.7 1.4 His97Phe 28 ± 1 1.4 His97Tyr 24 ±22 1.2 Val67Arg 4.0 ±0.5 0.2 Ser92Asp 28.8 ±0.5 1.4 His64Lys 74 ± 1 3.7 His64Thr 440 ± 20 20 His64Ile 4200 ± 200 209 A stoichiometry of 1:1 was observed for azide binding to all the Mb variants. The equilibrium dissociation constants determined here for azide binding to wild-type and commercial horse heart Mb agree well with the values reported previously for human (25 uM; Ikeda-Saito et al., 1992) and sperm whale Mb (27 uM; Cutruzzola et al., 1991). Substitutions on the proximal side of the heme binding pocket at positions 92 and 97 have little effect on the stability of azide binding. On the distal side of the heme pocket, changes in electrostatic character of the residues significantly influence binding of this negatively charged azide ligand. For example, addition of a positively charged arginine at position 67 increases the affinity of the protein for azide by ~5-fold, and the replacement of lysine with a glutamic acid at position 45 decreases the affinity by ~5-fold. In the case of the double variant, 82 Lys45Glu/Lys63Glu, the additional substitution of lysine with glutamic acid at position 63 decreases the stability of azide binding relative to wild-type to a greater extent than the single variant, Lys45Glu. A binding constant of intermediate magnitude is observed for the other double variant, Lys45Glu/Lys63Ser, where Lys63 is replaced with a serine instead of a glutamic acid. The FTIR spectrum of the Lys45Glu/Lys63Glu M b - N 3 complex in the region of the asymmetric stretching vibration of the azide ion is shown in Figure 26. The two bands at 2022.8 and 2045.6 cm"1 can be assigned to the low-spin and high-spin forms of the bound azide, respectively. Unbound azide gives rise to a broad band at 2048.3 cm"1. FTIR spectra of the azide complexes formed by these distal heme pocket variants (Table 13) are similar to the spectra observed for wild-type Mb, thus indicating the azide binds in a similar orientation in all of these proteins. The variants show small shifts in the low spin band maxima from wild-type ( v ^ = 2023.2 cm"1) to higher energy for the Val67Arg variant (2024.3 cm"1) and to lower energy for Lys45Glu Mb (2022.8 cm"1). The Lys45Glu/Lys63Glu variant (2022.8 cm"1) has the same band maximum as the single variant. The small shifts in band maxima of the low spin form have been attributed to the interaction of a charged group with the transition dipole moment of the intra-azide bonds (Bogumil et al., 1994). These results suggest that residues in positions 45 and 67 are near the azide binding site and have a direct electrostatic influence on azide binding. No further shift in band maximum is observed for the Lys45Glu/Lys63Glu variant as position 63 is located a greater distance away from the heme binding site. In Mb, the distal histidine residue is located near to the NI atom of the bound azide ligand and can stabilize binding of this ligand through hydrogen bonding (R. Maurus, unpublished). The distal histidine variants His64Thr and His64Ue exhibit 20 and 209 fold decreases in azide affinity, respectively (Table 12). The distal residue in these variants is unable to hydrogen bond to the bound 83 Low-spin 2080 2040 2000 Wavenumber (cm*1) Figure 26. Infrared spectrum ofLys45Glu/Lys63Glu Mb-N 3 complex (5 mM Mb, 3.6 m M azide, 0.1 M potassium phosphate, pH 7.0). Shown are the experimental spectrum (upper line), the absorption bands identified from the curve fitting analysis and the residuals of the fit. azide and cannot stabilize binding. The azide complex of the His64Ile variant is less stable than the azide complex of the His64Thr variant owing to the increased hydrophobic character of the distal heme pocket of the Ffis64Ile variant. This additional destabilization could also involve increased steric hindrance to azide binding by the large side chain of isoleucine. The FJis64Lys variant exhibits only a 4-fold decrease in affinity for azide. Although a lysyl residue at position 64 could potentially hydrogen bond to the bound azide and stabilize binding, other factors appear to dominate and compromise the stability of the azide complex. Linkage of Mn(II) binding and azide binding. As substitutions at position 45 and 63 exert an electrostatic influence on ligands bound in the distal pocket, it was conceivable that a metal ion bound to this negatively charged protein surface near the heme propionates could also influence a 84 bound ligand. As shown earlier (Results section A1-A2), the Lys45Glu and Lys45Glu/Lys63Glu Mb variants bind Mn(II) in this region of the protein surface. For this reason, the FTIR spectra of the azide complexes and the equilibrium azide binding constants of these variants were determined in the presence of Mn(n). The infrared spectra of the variants were measured in the presence of M n S 0 4 at a concentration where the binding sites on the variant proteins should be >85% occupied. The spectrum of azide and M n S 0 4 in the absence of protein exhibits a weak feature at -2077 cm"1 that represents the Mn(II)-azide complex and corresponds to ~ 6 % of the free azide peak (data not shown). Therefore, the amount of Mn(II)-azide complex formed in the experiment is negligible. The absorption maxima obtained from curve fitting analysis of the FTIR spectra of the Mn(TJ)-Mb azide complexes are listed in Table 13. The band maxima for both the high- and low-spin azide derivatives in the presence of Mn(II) ions are quite similar to those of the wild-type Mb azide complex, indicating no large structural pertebations caused by metal binding. In the absence of Mn(n), it was found that the replacement of a positive charge with a negative charge in either variant results in a small shift in the low-spin band to lower energy relative to wild-type Mb. When Mn(II) ions are added to the variants, the absorption maxima of the low-spin bands shift to slightly higher energy relative to wild-type Mb. Evidently, the Mn(II) ion bound to the negative protein surface created by the replacement of Lys45 and Lys63 with glutamyl residues is sufficiently close to the bound azide ligand to influence binding. The band maximum for the low-spin band of wild-type Mb does not shift in the presence of Mn(II) ions. Additional control spectra of the Mb-azide complexes in an isoionic NaCl buffer verified that the azide stretching frequency changes in the presence of Mn(II) were not a consequence of differing ionic strength. 85 Table 13. FTIR absorption maxima derived from curve fitting analysis and the equilibrium binding constants (Xdapp), for the Mn(TJ)-Mb azide derivatives at 20 °C. Buffers used were (a) 0.1 M KPi , pH 7.0, (b) 50 mM HEPES, pH 7.0/50 mM M n S 0 4 and (c) 50mM HEPES, pH 7.0/200 mM NaCl. Protein Low-spin component vm a i («""') High-spin component "max ( C H I 1 ) tfdapp (uM) Wild-type Mb + N 3 K P i a 2023.2 2045.6 20 ± 1 NaCl 2023.2 2045.7 28 ± 1 Mn(II) 2023.2 2045.7 26 ± 1 Lys45Glu Mb + N 3 K P i a 2022.8 2045.6 99 ±3 NaCl 2023 2045.5 130 ± 10 Mn(II) 2023.5 2045.6 49 ±3 Lys45Glu/Lys63Glu + N 3 K P i a 2022.8 2045.6 280 ± 10 NaCl 2022.8 2045.3 277 ± 9 Mn(II) 2023.7 2045.7 32 ±3 (Bogumil et al., 1994). The effect of metal binding on the equilibrium constants for azide binding (£ d a p p ) was determined for these variants using similar solution conditions as used for the FTIR experiments (Table 13). Addition ofMnS0 4 to the Lys45Glu and Lys45Glu/Lys63Glu Mb variants increased the azide binding affinity of these proteins 2- and 9-fold, respectively, over the binding observed in NaCl at a similar ionic strength. No change was observed in Kdm for wild-type Mb in the presence of Mn(TJ). Therefore, the binding of MntTI) to the engineered metal binding site stabilizes azide binding in the distal heme pocket of these Mb variants. 86 2. HemepKA At neutral pH, a water molecule is coordinated to the exchangeable ligand binding site of the heme iron in the distal heme pocket. At alkaline pH, this distal ligand loses a proton to form a hydroxyl ion. The pKA of this distal H 2 0 ligand (the heme pK^ in horse heart Mb has been determined previously to be 8.93 at 20 °C (Antonini & Brunori, 1971). Titration of this group can be monitored by electronic absorption spectroscopy as the heme iron changes from high spin (at low pH) to a spin equilibrium (at alkaline pH) (Figure 27). The dependence of the absorption spectrum on pH fits well to a Henderson-Hasselbach function (8) for one titratable species. Al l the Mb variants measured showed similar spectral changes with pH as wild-type Mb, indicating all variants still possess a distal water ligand. The pKA values of the water bound in the distal pocket of the Mb variants at 25 °C are listed in Table 14. The replacement of Lys45 or Lys63 with glutamyl residues results in a large increase in the heme pKA of the variant and the increases due to these two substutitions are additive. These substitutions remove a positive charge and situate a negatively charged carboxylate group in proximity to the entrance to the distal pocket. The negative charge destabilizes the OH" ligand bound in the heme pocket, and thereby increases the pKA of the bound H 2 0 . In the double variant with a serine at position 63, the increase in heme pKA was intermediate between the increases observed for the Lys45Glu and Lys45Glu/Lys63Glu variants. No change in the heme pKA was observed for variants containing substitutions at positions 92 and 97 on the proximal side of the heme. The effect of Mn(II) binding on the heme pKA could not be determined due to the formation of Mn(OH) 2 precipitates at higher pH values. 87 < 500 600 700 Wavelength (nm) Figure 27. Spectrophotometric pH titration of the Lys45Glu/Lys63Glu Mb variant in 0.1 M NaCl at 25 °C. The pH values are 7.50, 8.48, 8.91,9.88,10.07, 10.41, 10.84, 11.50. Table 14. The heme pKA values for the myoglobin variants at 25 °C in 0.1 M NaCl. Variant P^A wild-type 8.99 ±0.04 Lys45Glu 9.64 ± 0.02 Lys45Glu/Lys63Glu 10.06 ±0.02 Lys45Glu/Lys63Ser 9.94 ± 0.02 Lys45Glu/Lys63Glu/Asp60Glu 10.04 ±0.02 His97Leua 9.01 ±0.05 F£is97Leu/Lys45Glua 9.62 ± 0.06 His97Leu/Lys45Glu/Lys63 Glu a 10.01 ±0.02 Ser92Asp/Lys45Glua 9.54 ±0.05 Lvs45Glu/Lvs63Glu/F£is97Leu/Ser92Ala 9.65 ± 0.04 a L. Eltis, unpublished. 88 D. Other characterizations 1. Electrochemistry The family of spectra collected during the spectroelectrochemical titration of the Lys45Glu Mb variant is shown in Figure 28. The reduction potential for this variant was calculated from the Nernst plot (10) shown in the inset. Sharp isosbestic points (366, 419, 462, 522, 607, 670 nm) are consistent with the presence of just two chromophoric species and indicate minimal difficulty arising from protein instability or residual oxygen contamination. Similar spectra were obtained for each of the variant Mbs and the reduction potentials and slopes calculated from the Nernst plot (10) of each titration are indicated in Table 15. -0.6 -0.4 0.0 0.4 0.8 log[(A r-A i)/(A i-A 0)] 400 500 600 Wavelength (nm) Figure 28. Spectroelectrochemical titration of the Lys45Glu myoglobin variant at 25 °C (sodium phosphate buffer, 1= 0.1 M , pH 7.0). The absorbance spectra show the varying ratios of metMb (O) and deoxyMb (R) generated by the applied potential ( E J . The applied potential (mV vs SFfE) were: -237.1 (R), -30.2, -11.2, 7.4, 26.3, 45.2, 65.1, 444.5 (O). 89 Table 15. Reduction potentials ( E M ) for the myoglobin variants used in this study (sodium phosphate buffer, 1=0.1 M pH 7 .0 , 2 5 ° C ) . Protein E M (mV vs SHE) Nernst slope (mV) wild-type 52 ± 2 63 ±3 Lys45Glu 26 ±3 60 ± 1 Lys45Glu/Lys63Glu 0 ± 6 65 ±5 Lys45Glu/Ly63Ser 20 ± 2 60 ± 1 Lys45Glu/Lys63Glu/Asp60Glu -5.2 ± 2 63 ± 1 His97Leu 63 ±3 68 ± 6 His97Leu/Lys45Glu 40 ±5 67 ±3 His97Leu/Lys45Glu/Lys63 Glu 15 ± 2 64 ±3 Ser92Asp/Lys45Glu 29 ± 7 61 ± 2 His97Leu/Lys45Glu/Lys63 Glu/Ser92Ala 4 ± 3 63 ± 2 As observed in the azide binding and heme pKA studies, changes in the electrostatic character of residues on the distal side of the heme also affect the reduction potential of Mb. The decrease in the midpoint potentials observed for the Lys45Glu, Lys45Glu/Lys63Glu, and Lys45Glu/Lys63Glu/Asp60Glu variants can be attributed to the reversal in charge of these residues. Replacing a positively charged lysyl residue with a negatively charged glutamyl residue will stabilize the Fe(III) oxidation state over the Fe(II) oxidation state and subsequently decrease the midpoint potential. The Lys45Glu/Lys63Ser variant exhibits a reduction potential intermediate between the Lys45Glu and Lys45Glu/Lys63Glu variants. Therefore, the electrostatic character of the amino acid residues at both positions 45 and 63 influence the Fe(II)/Fe(III) equilibrium of the heme iron. The His97Leu substitution on the proximal side of the heme results in a small increase in reduction potential (-10-15 mV) in all variants which possess this substitution. The effect of changing Ser92 is difficult to ascertain from the data presented here. 90 Linkage of Mn(II) binding and reduction potential The ability of a metal ion bound to the surface of Mb to influence the stability of azide binding within the distal heme pocket of the double variant and the dependence of the reduction potential on the electrostatic character of nearby protein residues suggested that metal binding may also influence the midpoint potential of the Fe(II)/Fe(III) equilibrium of the heme iron. To investigate this possibility, the reduction potentials of wild-type Mb and the Lys45Glu/Lys63Glu variant were determined in the presence and absence of Mn(II) (Table 16). As shown earlier, the Lys45Glu/Lys63Glu variant binds Mn(II) at an engineered site near the heme propionates on the protein surface (Results, Section A1-A3). The presence of Mn(II) in solution has very little effect on the of wild-type Mb as the metal ion does not bind near the heme (Results, Section A1-A3). However, the binding of Mn(II) to the surface of the Lys45Glu/Lys63Glu variant causes an increase of ~30 mV in the reduction potential of the heme iron. This increase is expected as the binding of the metal ion will increase the amount of positive charge near the heme which will destabilize the Fe(III) form. Control measurements performed in an isoionic NaCl buffer Table 16. Reduction potentials (E j^) of wild-type and Lys45Glu/Lys63Glu Mb variant in the presence and absence of Mn(JJ). Buffers used were (a) 50 m M HEPES, pH 7.0/200 m M NaCl and (b) 50 m M HEPES, pH 7.0/50 mM MnS0 4 . Protein E M (mV vs SHE) Nernst slope (mV) wild-type NaCl* 45 ± 2 64 ± 7 M n S 0 4 b 50 ± 4 63 ±5 Lys45Glu/Lys63Glu NaCP 9 ± 2 64 ±5 MnSO„ b 40 ±5 62 ±3 91 verified the reduction potential changes in the presence of Mn(II) were not a consequence of changing ionic strength. 2. Circular dichroism Circular dichroism has been used widely to study the conformation and conformational changes of proteins in solution (Chen et al., 1972). The CD spectra of the wild-type and variant Mbs were measured in the far-UV region at pH 7.0. Comparison of the resulting spectra indicated no large perturbations in the secondary structure of the variant proteins. Representative spectra are shown in Figure 29 for wild-type Mb and the Lys45Glu variant. Figure 29. Far U V CD spectra of wild-type Mb (—) and the Lys45Glu variant (- - -) (0.01 M sodium phosphate buffer, pH 7.0 and 25 °C). 92 3. Electronic absorption spectra The absorption maxima for several derivatives (Fe(III)-aquo, Fe(III)-CN, Fe(II), and Fe(II)-CO) of the Mb variants (Appendix D, Table 18) are similar to those observed for corresponding derivatives of wild-type Mb. This observation indicates that these substitutions do not produce any significant changes in the electronic properties of the heme. One exception is the Ffis64Tyr/Val68Phe Mb variant which was constructed for the heme dissociation assays (Results section B3). The spectrum of the oxidized form of this variant exhibits a decrease in extinction coefficient (Appendix C, Table 17), a small red-shift of the Soret maximum and an absorption band at 599.5 nm. These charactersitics are indicative of axial coordination of the distal tyrosyl residue to the heme iron (Egeberg et al., 1990). The single variant, His64Tyr, of horse heart Mb has previously been shown to coordinate the heme iron atom (Tang et al., 1994a; Maurus et al., 1994). 4. 'H-NMR measurements To characterize the structural properties of these Mb variants further, the ^ - N M R spectrum for each of the met-aquoderivatives in 100% D 2 0 was determined (a representative spectrum for Lys45Glu/Lys63Glu Mb is shown in Figure 11 A). Most of the hyperfine shifted resonances of the variants could be assigned tentatively by comparison to the spectrum of sperm whale Mb (La Mar et a l , 1980, Unger et a l , 1985). Some of the single proton heme resonances are not assigned in several of the variants because they could not be identified from the limited information obtained here. The chemical shifts of the heme methyl resonances for the Mb variants (Appendix E, Table 19) were found to be similar to those of the wild-type protein. This observation indicates that the unpaired spin distribution of the heme in all the variants must be similar (La Mar et a l , 1993) and, 93 therefore, that no large structural perturbations result from the amino acid substitutions. The mean chemical shift of the heme methyl groups observed for all of the Mb variants, s(M l i 3 i5 i g), is -75 ppm. The broad resonance centered at - 4 0 ppm downfield and previously assigned to the heme meso protons (La Mar et al., 1980) was observed in all the variants. Both of these observations indicate the presence of a six-coordinate heme iron (Rajarathnam et al., 1991), and with the results of the heme pKA experiments (Results section C2), confirm the presence of a water ligand in the sixth position of all of these Mb variants. Small changes in the chemical shifts of the single proton heme resonances were observed in some of the variants (Appendix E, Table 20). The 7-a proton heme resonance in variants possessing substitutions at position 92 and 97 demonstrates the greatest change in chemical shift (-15 ppm and -10 ppm upfield, respectively). This change in chemical shift suggests that reorientation of the heme 7-propionate may result from the removal of the hydrogen bonds involving the propionate and these protein residues (Lloyd et al., 1996). Both sets of doublets (6-a' and 4-a, and 7-a' and 2-cc) in the wild-type spectrum are resolved to some extent in the spectra of variants possessing the Lys45Glu substitution and were identified on the basis of their pH dependent behavior and proximity to the site of the substitution. The identity of the 6-a' proton resonance is further confirmed by the broadening of this resonance observed in the Mn(II)-Mb titrations (Figure 11). N M R spectra of the oxidized derivatives of the wild-type, Lys45Glu and Lys45Glu/Lys63Glu variants were collected over the pH range of 4.5 to 8.0. The behavior of the heme 6-propionate in this experiment was of particular interest because this group is adjacent to residue 45. The 6-a ' heme resonance exhibited little pH dependent behavior in N M R spectra of wild-type Mb but this resonance titrated with a pK& of 5.7 and 5.2 in the Lys45Glu and Lys45Glu/Lys63Glu variants, respectively. The pH dependent chemical shift data of this resonance for both variants was fitted to a function for 94 a single titrating group (Figure 30). The titrating group responsible for this behavior is probably either the heme 6-propionate or Glu45. These values are similar to the pK& of the metal binding site (pKa » 5.5) determined from the pH dependence of Mn(II) binding to the Lys45Glu/Lys63Glu (Results section A l ) . 5.0 6.0 7.0 pH Figure 30. The chemical shift of the 6-a ' heme resonance is monitored as a function of pH in N M R pH titrations of the oxidized derivative of the Lys45Glu variant. Tentative assignments for some of the heme resonances in the metMb-CN derivatives of the variants (Appendix E, Table 21) have been made here by comparison to the spectra of the metMb-CN derivatives of sperm whale (Emerson & La Mar, 1990) and horse heart Mb (Lecomte & La Mar, 1985). A representative spectrum in 100% D 2 0 is shown in Figure 13D for metMb-CN derivative of the Lys45Glu variant Mb. The heme methyl hyperfine shifts of variants containing substitutions at positions 45 and 97 are similar to the shifts observed in the spectrum of the wild-type protein. In the spectra of variants possessing substitutions for Ser92, the heme methyl chemical shifts are shifted 95 upfield. This observation has been attributed previously to a small rotation of the His93 proximal ligand resulting from the loss of the hydrogen bond with Ser92 (Lloyd et al., 1996). The correlation of the heme methyl hyperfine shifts with the angle formed by the plane of the axial histidine imidazole ring and the line defined by the nitrogen atoms of pyrroles II and IV has been observed (Yamamoto et al., 1990, Soltis & Strouse, 1988). This orientation difference affects the unpaired spin density pattern of the heme, which in turn influences the contact shifts of the heme substitutents. Another noteworthy difference in the spectra of the variants containing the Lys45Glu substitution is the apparent disappearance of the resonance assigned to the exchangeable His64N €H proton resonance. The exchangeable proton resonances, His64N€H and His93N 8H, are only observed when the proteins are measured in a 10% D2O/90% H 2 0 solution. As the pH of the solution is increased above 8.5, the His64N £H heme resonance of the Lys45Glu variant reappears with a chemical shift (23.56 ppm) similar to that observed in the spectrum of wild-type Mb (23.54 ppm) (data not shown). The linewidth of the exchangeable proton peak is related to the rate of chemical exchange, where rapid exchange of a proton broadens the peak beyond detection (Lecomte & La Mar, 1985). Therefore, the replacement of Lys45 with a glutamyl residue alters the pH dependence of the exchange behavior of this proton. Efficient acid-catalyzed proton exchange occurs at a more basic pH in the Lys45Glu variant than for wild-type Mb suggesting greater access of protons to the heme pocket in the variant. The identity of the residue at position 45 has been shown previously to play a role in the exchange behavior of this residue (Lecomte & La Mar, 1985). Lecomte and coworkers have demonstrated that horse and dog Mb possess a lysyl residue at position 45 which stabilizes the open pocket conformation of Mb and increases base-catalyzed exchange at higher pH relative to sperm whale Mb with an arginine at position 45. The removal of the hydrogen bond between the heme 6-propionate and Lys45 in the Lys45Glu-containing variants also appears to favor 96 a more open heme pocket conformation. Finally, previous work has demonstrated that the pH dependence of the 5-CH 3 proton resonance chemical shift in the spectrum of metMb-CN derivative of sperm whale Mb results from titration of His97 (p^ A of 5.3) (La Mar et al., 1978a; Krishnamoorthi & La Mar, 1984). In horse heart Mb, His97 was shown to have a p^ A of 5.4 (Lloyd et al., 1996). The imidazole group of the proximally located His97 lies parallel to the plane of the heme, directly below the carbon atom of pyrrole HI at heme position 5. Titration of the metMb-CN derivative of Lys45Glu Mb in 100% D 2 0 reveals a pH-dependent chemical shift of the 5-CFf3 resonance with a pKA of 5.2 (data not shown). Therefore, replacement of Lys45 with Glu has little effect on the pKA of His97. 97 IV. DISCUSSION In this study, a series of Mb variants has been designed to examine several types of molecular recognition. Specifically, this work has concerned the binding of metal ions, the heme prosthetic group and anionic ligands to the protein. Ultimately, the results of such work should provide insight that will in time permit the a priori design of proteins with specific functional properties for creation of novel catalysts and therapeutic agents. Only in the absence of global structural changes in the protein can the effects of individual amino acid substitutions on molecular recognition by the protein be assessed with confidence. A number of spectroscopic techniques were used here to determine whether or not the substitutions introduced into Mb perturbed the overall structure of the protein. The CD spectra of each of the Mb variants indicated that the amino acid replacements involved had minimal effects on the secondary structure of Mb. The structural similarity in the heme binding sites was verified by the absence of large changes in the electronic absorption spectra of the Mb variants. Similar mean heme methyl chemical shifts were observed in the ^ - N M R spectra of all the metMb variants confirming the lack of any large changes in the heme binding site. Minor reorientations of some of the protein side chains and heme substituents were identified from the 'H -NMR spectra of the metMb and metMbCN derivatives. A. Development of a new metal binding site in myoglobin. The metal binding site on Mb was designed to satisfy two requirements. The first requirement was that the specific site for metal binding on the surface of the protein would bind Mn(II) with reasonable affinity. The second requirement was that metal ions bound to this site would transfer electrons to the heme iron. As Mb is well known to possess a peroxidase activity (George & Irvine, 98 1956; King & Winfield, 1963; Allentoffet al., 1992), the goal was to generate a metal binding site that would permit the use of MntTT) as the source of reducing equivalents in the peroxidase reaction. The design and construction of a suitable metal binding site in a protein involves a number of steps. First, the structural requirements of the site must be defined. The present goals were best met by a site with good solvent accessibility, a location near the heme to permit efficient electron transfer to the heme iron, carboxylate ligands, and retention of the structural integrity of the distal heme binding pocket. Second, a group of adjacent amino acid residues must be identified and replaced with potential metal ligands. The residues selected in this study were Lys45, Lys63, and Asp60. Single, double and triple variants with these substitutions were constructed by site-directed mutagenesis. Third, an initial screen for variants with increased metal binding affinity must be performed to identify the amino acid changes that are effective. The Mb variants were screened for relative Mn(II) binding affinity with an EPR titration technique and the Lys45Glu and Lys63Glu substitutions were identified as critical for Mn(II) binding. A glutamic acid at position 63 was essential for increased metal binding affinity as a serine residue at the same position did not increase metal binding stability. Substitutions in the proximal heme pocket at positions 92 and 97 did not affect metal binding to the site. Finally, detailed structural and functional characterization can then be performed to determine the properties of the most successful variants. Structural characterizations. The binding of Mn(II) to Mb in solution was characterized by ' H - N M R spectroscopy. Dipolar coupling between the Mn(II) ion and the nearby nuclei causes broadening of the paramagnetic resonances, and the change in linewidth of these resonances in the presence of Mn(II) is dependent on the metal-proton distance (Bertini & Luchinat, 1986; Banci et al., 1993). From the 'H-NMR experiments, all the Mb variants were shown to bind Mn(II) at the site introduced near the solvent accessible heme edge. Broadening of the heme 6-propionate resonance 99 in the presence of Mn(II) strongly suggests the involvement of this heme substituent in Mn(II) binding. No broadening of this resonance was observed in the corresponding spectrum of wild-type Mb in the presence of Mn(II) ions. Similar ^ - N M R experiments were employed to identify the site of Mn(II) binding near the heme edge of an authentic manganese peroxidase from the white rot fungus, P. chrysosporium (Band et al., 1993). To further characterize the Mn(n) binding properties of the Lys45Glu/Lys63Glu variant, the three-dimensional structure of this protein in the presence of Mn(II) was determined by X-ray crystallography (R. Maurus, unpublished). The structure (Figure 31) indicates that the Mn(II) ion binds the protein with the heme 6-propionate and the glutamyl residue at position 45 providing one and two oxygen ligands, respectively. One of the coordination positions in the structure is occupied by Hisl 13 from an adjacent molecule in the crystalline lattice. Presumably, this position is occupied by a water molecule in solution. Two water molecules complete the coordination environment of this site. The Mn(II) ion is located 10.4 A from the heme iron. This structure confirms the conclusion from the 'H -NMR experiments that the Mn(II) ion binds near the heme 6-propionate in the site created by site-directed mutagenesis. The second, low affinity Mn(II) binding site, identified in the potentiometric titration of Mb with Mn(II), was not seen in this structure, presumably owing to low occupancy by Mn(II). The three dimensional structure of the Lys45Glu/Lys63Glu variant shows that the glutamyl residue at position 63 is not directly coordinated to the Mn(II) ion. However, EPR results indicate a negatively charged residue at this position is required for increased stability of metal binding as Ser63 does not enhance Mn(II) binding. The electrostatic surface potential map, calculated with GRASP (Version 1.2, Nicholls et al., 1991) and the coordinates for wild-type Mb in the absence of Mn(II), shows a positive region (blue) on the surface of the protein in the vicinity of the heme 6-100 propionate (Figure 32A). The corresponding electrostatic surface of the Lys45Glu/Lys63Glu variant in the absence of Mn(II) demonstrates that this same region possesses significantly greater negative character (red) (Figure 32B). These calculations suggest that enhanced metal binding results from the additional negative charge provided by Glu63 which increases the attraction between the positively charged Mn(II) ion and the negative protein surface of the Lys45Glu/Lys63Glu Mb. In addition to providing potential metal ligands, the creation of a negative electrostatic surface surrounding the binding site is also an important consideration in the design of a metal binding site. Competition studies demonstrated that other divalent metal ions such as Cu(II), Co(II) and Cd(II) can also bind to this site. Therefore, the three-dimensional structure of the Lys45Glu/Lys63Glu variant in the presence of Cd(II) was determined by R. Maurus (unpublished). Two metal binding sites are observed in this structure (Figure 33). One of these sites involves the heme 6-propionate and the Glu45 residue and is similar to the Mn(II) site shown in Figure 31. The second site involves Hisl 19 and Aspl22 and is remote from the heme binding site of the protein. The Cd(H) ion bound at the heme edge is located 10.2 A from the heme iron. Comparison of Cd(II)- and Mn(II)-bound structures reveals small differences in the locations of the metal ions and the two coordinating residues, Glu45 and the heme 6-propionate. These small differences may represent a small structural adaptations to allow for complex formation. 101 Figure 31. Three-dimensional structure of the engineered Mn(II) binding site of the Lys45Glu/Lys63Glu Mb variant. Water molecules (shown in blue) and the carboxylate group of Glu45 directly coordinate to the Mn(II) ion. Glu63 does not coordinate but has been shown to be important for increased metal binding affinity. 102 Figure 32. Electrostatic surface potential diagram of (A) wild-type and (B) Lys45Glu/Lys63Glu Mb. Regions of positive, neutral, and negative potential are shown in blue, white and red, respectively. The potential of the surface near the heme 6-propionate is much more negative in the Lys45Glu/Lys63Glu variant than in wild-type Mb due to the replacement of two basic residues with acidic residues. 103 Figure 33. Three-dimensional structure of the Lys45Glu/Lys63Glu Mb variant containing Cd(II). The engineered Cd(II) binding site utilizes the heme 6-propionate and Glu45 as ligands and occupies a site similar to that observed for Mn(II) binding (Figure 31). The naturally occuring site located remote from the heme binding site involves His l 19 and Asp 122. 104 Binding of metal ions to the surface of Mb has been investigated previously. Sperm whale Mb has been shown to bind Mn(II) by means of an EPR technique similar to that used in this work (Gersonde and Netter, 1966). Binding of Mn(II) and Cu(II) to Aplysia brasiliana Mb was demonstrated with the EPR titration technique, and these metal ions were shown to bind to related sites through competition experiments (Baffa et al., 1986). The three dimensional structure of sperm whale Mb has been determined with Cu(TI) or Zn(IJ) ions bound, and in these structures, both metals bound in the same region (Banaszak et al., 1965). The Zn(II) ion was bound in the region of Hisl 19, Asnl22 and Lysl6, and the Cu(II) ion bound nearby utilizing Asnl22, His 12 and Lysl6 as metal ligands. The corresponding region of horse heart Mb is involved in binding Cd(II) (Figure 33). In addition, as the coordination requirements of Zn(II) are similar to those of Co(II), it is possible that the second Co(U) ion binding site on horse heart Mb, observed by potentiometric titration, is in the same region. Whether the second Mn(II) ion binds in this region of the protein also is difficult to determine from the data presented here, however, this site does possess the nitrogen and oxygen ligands preferred by Mn(II). Functional characterizations. The potentiometric titrations of the Lys45Glu/Lys63Glu variant identified two metal binding sites: one naturally-occurring site with low affinity and a second engineered site with significantly greater affinity. Stability of binding at the engineered site is - 2 0 fold greater than binding at the naturally-occurring site. The stability of binding of Mn(II) to the Lys45Glu/Lys63Glu variant decreases at lower pH, owing to the protonation of the heme 6-propionate or the Glu45 residue of the site (pKA of -5 .2 - 5.5). The stability of Mn(II) binding is inversely related to ionic strength as expected from the electrostatic nature of such interactions. Mn(H) binding also shows a small dependence on the oxidation state of the protein as the calculated Mn(II) binding constant for the reduced form of the Lys45Glu/Lys63Glu variant is ~3-fold greater 105 than for the oxidized form. This binding constant was calculated from the reduction potentials of the variant in the presence and absence of Mn(II) (Appendix G). Competition EPR studies have shown the engineered binding site to have affinity for other divalent metal ions, Cu(II), Co(II), and Cd(II). Potentiometric titrations of Lys45Glu/Lys63Glu Mb with Co(II) identified two Co(II) binding sites, both sites possessing greater binding affinity for Co(II) than observed for Mn(II). The affinity of Mn(II) for the metal binding site introduced into Mb in the current study is substantially lower than that observed for the Mn(II) binding site is MnP obtained from P. chtysosporium (M. R. Mauk, unpublished results). In the structure of MnP, three surface acidic amino acid residues and the heme 6-propionate are involved in Mn(II) binding (Sundaramoorthy et al., 1994). The availability of two additional protein ligands could account for the ~500-fold greater stability of Mn(II) binding to MnP relative to the Mb variants. Removal of one of these ligands (Asp 179) in MnP by site directed mutagenesis resulted in a drastic decreases in the rate of oxidation of Mn(U) to Mh(in) due to reduced affinity of the enzyme for Mn(II) (Kusters-van Someron et al., 1995). The Mn(II)-protein ligand distances in the Mn(II)-complexes of Mb and MnP are comparable (Appendix F, Table 22). Although Mb possesses the same prosthetic group and axial ligands as CcP, FfRP, and MnP, it is lacking key active site residues that confer efficient peroxidase activity in these proteins (Poulos & Kraut, 1980). Nevertheless, Mb does exhibit a limited peroxidase activity (George & Irvine, 1956; King & Winfield, 1963; Allentoff et al., 1992) and small increases in this activity were observed for variants possessing the Lys45Glu substitution. This increase could in part result from the small decrease in reduction potential observed for this variant. Peroxidases such as CcP (Conroy et al., 1978) and MnP H3 and H4 (Millis et al., 1989) exhibit reduction potentials for the Fe(II)/Fe(III) couple of -194, -88 and -93 mV vs SHE, respectively. It is also possible that greater negative charge 106 near the heme prosthetic group could stabilize the Fe(IV) intermediate that forms during the peroxidase catalytic cycle. Limited kinetic analysis was performed due to the unstable and transient nature of Mb compound I (Catalano et al., 1989; Wilks & Ortiz de Montellano, 1992) The increases in manganese peroxidase activity observed for the series of Mb variants correlate with the increases in the peroxidase activity observed with ABTS as a substrate (Table 4 and 5). The manganese peroxidase activities do not correlate well with the affinities of the variants for Mn(TJ) determined by the EPR titrations. This lack of correlation suggests that the limiting factor in the ability of these variants to oxidize Mn(II) to Mn(III) is the reactivity with H 2 0 2 and not the binding of the Mn(JJ) substrate. Greater manganese peroxidase activity of Mb might be possible with additional substitutions in the heme binding cavity to improve the peroxidase activity. In Mb and MnP, the activities for the oxidation of Mn(II) and the Mn(II) binding constants exhibit different pH dependencies. The activity of MnP is optimal at pH 4.5 (Glenn & Gold, 1985; Glenn et al., 1986) and this activity varies inversely with the pH dependence of the stability of Mn(II) binding (M. Mauk, unpublished). In the case of the Mb variants, peroxidase activity is optimal at pH 5.0, but Mn(H) binds with greatest affinity above pH 6.5. Finally, at the pH optimum for activity of MnP (pH 4.5), the stability of Mn(H) binding to the peroxidase is greater than the stability of Mn(II) binding to any of these Mb variants at any pH. B. Importance of the heme propionate interactions on heme protein stability. An ensemble of proteins has been prepared in which the number of hydrogen bonds formed between the heme propionate groups and specific amino acid residues was systematically altered. Availability of these proteins permits a comprehensive evaluation of the contribution of these hydrogen bonding interactions to the dynamics of heme-protein interactions for horse heart Mb and 107 cytochrome b5. The mechanism of heme association with apoMb proposed by La Mar and colleagues is depicted in Figure 34 (adapted from La Mar et a l , 1984) and is presumably applicable to cytochrome b5 and other proteins that possess non-covalently bound heme prosthetic groups. Figure 34. Schematic representation of the reaction of apoMb with heme. The major (M) and minor (m) orientations are represented by the unshaded and shaded circles, respectively and differ by 180° rotation about the aJy-meso carbon heme axis. (La Mar et a l , 1984) Rate constant k.H is dominated by Jc™ as only 8% of the protein binds heme in the minor orientation. This scheme depicts the initial binding of heme to the apoprotein in one of two orientations that differ in rotation of the heme about the a/y-meso carbon heme axis. Subsequently, the relative amounts of these two forms equilibrate, and the thermodynamically more stable orientation predominates. In the present study, the heme reorientation kinetics of the Mb variants have been investigated to characterize the first two processes. To characterize the stability of the major isomer 108 at equilibrium, the kinetics of heme dissociation, the activation energy for heme dissociation, and the thermal stability of the Mb and cytochrome b5 variants were studied. Kinetics of Heme Reorientation. For all of the Mb variants studied here, the 'H -NMR spectra obtained following addition of cyanide immediately after reconstitution with heme indicate formation of equal amounts of the two heme orientational isomers (i.e., the initial ratio of major (M) to minor (m) orientational forms was -1:1). Therefore, the rates for the binding of heme in both orientations (k™ and &2m) must be equal and independent of the protein interactions with the heme propionates. La Mar and co-workers (1984) also found these rates to be independent of the nature of the heme substituents at positions 2 and 4. The ^ - N M R spectra of the metMb-CN derivative for each of the Mb variants at equilibrium showed the equilibrium ratio (ATEQ) of major to minor heme orientation also to be unchanged from that of wild-type horse and sperm whale Mb (La Mar et al., 1983). As pyrroles III and IV are symmetrical about the aly-meso axis (Figure 2), the binding of heme in these two orientations does not change the positions of the two heme propionates relative to the proteins residues they interact with. Altering the residues near the heme propionates will affect both orientational isomers to the same extent, so the initial and equilibrium heme disorder ratios should be unaltered. Nevertheless, substitution of the Mb residues 45, 92, and 97, all of which form hydrogen bonds with the heme propionates, does have a significant effect on the rate of heme reorientation. As each hydrogen bond is removed, the heme is able to equilibrate much more rapidly within the heme binding pocket, and significant increases in the rate constants for heme reorientation (k{ and k^) are observed. Interaction of the heme propionates with these residues, therefore, appears to be critical in the kinetic capture of the minor orientation. The contributions of hydrogen bonding by heme propionate groups in the dynamics of heme 109 binding to Mb have been studied previously through the use of synthetic hemes with modified substituent groups. For example, sperm whale Mb reconstituted with hemins lacking the 6- or 7-propionate (6-methyl-6-despropionate-hemin and 7-methyl-7-despropionate-hemin) exhibited the same heme orientational equilibrium as Mb possessing iron-protoporphyrin IX (Takano, 1977; La Mar et al., 1986; La Mar et al., 1989). Similar results were subsequently reported for interaction of the same heme derivatives with horse heart Mb as monitored by CD spectroscopy (Santucci et al., 1993). However, the initial complex formed upon reconstitution with either propionate-deficient heme in excess CN" resulted in preferential placement of the remaining propionate near Arg45 (La Mar et al., 1989). These results indicate that removal of individual heme propionate groups does not affect the heme orientational equilibrium but does influence the relative amounts of the two orientational isomers formed immediately following reconstitution of the apoprotein, which is consistent with the conclusions drawn from the present study. Kinetics of Heme Dissociation. The interactions between the propionates and the protein surface residues were also found to have an effect on the rate constant for heme dissociation (k.H ) (k,H is dominated by k™ (Figure 34) as only 8% of the protein binds heme in the minor orientation at equilibrium). Inspection of the results reveals that as individual hydrogen bonding interactions involving the heme propionates are eliminated, the resulting increases in k.H are multiplicative (Table 7). These interactions play a role in inhibiting heme dissociation from the protein at equilibrium. Similar increases in heme dissociation rate constants were also observed in related forms of cytochrome b5 where the hydrogen bonds involving the propionates were eliminated and suggest that this functional role of heme propionate groups may apply to heme proteins in general. Recently, the association rate constants for CO-heme binding to apoMb were found to be relatively insensitive to substitutions within the heme binding pocket, indicating that the equilibrium constant for heme 110 binding is determined primarily by the heme dissociation rate constants (Hargrove et al., 1996a). Therefore, these propionate-protein interactions play a partial role in the affinity of Mb for heme. Olson and colleagues, through the study of heme dissociation in a large number of single variants of sperm whale Mb, have addressed other important factors in heme binding. The distal histidine inhibits heme loss from the binding pocket by forming a hydrogen bond with the water coordinated to the heme iron (Hargrove et al., 1994a, 1994b, 1996b). In addition, the Val68Thr variant was found to stabilize heme binding over wild-type Mb as the threonine residue can form an additional hydrogen bond to the coordinated distal water (Hargrove et al., 1994b). Replacement of the proximal histidine (His93Gly Mb) increases heme dissociation >100-fold (Hargrove et al., 1996b). Hydrophobic contacts and the maintenance of the apolar heme binding pocket are also very important in heme binding as substitutions at position 29, 43, 46, 89, 99, and 103 result in large increases in heme dissociation from Mb (Hargrove et al., 1996b). From the relative magnitudes of these increases in heme dissociation, the major factors in the equilibrium binding of the major heme orientation in myoglobin are the hydrophobic interactions and the apolarity of the heme pocket for the binding of this amphipathic prosthetic group. The strength of the proximal histidine-iron bond plays a large role in retaining heme with the remaining electrostatic interactions with the coordinated water and the propionates assuming a more minor role. (Hargrove et al., 1996b) Heme dissociation from similar variants in other Mb species have been studied and can be compared with the results obtained here for horse heart Mb, although the differences in species can also account for some variation. The rate constant for heme dissociation measured for the Arg45Glu variant of sperm whale Mb was found to be ~7-fold greater than that of the wild-type protein (Hargrove et al., 1994), a larger increase than that found here for horse heart Mb. This arginyl residue in sperm whale Mb, in addition to forming a hydrogen bond with the heme 6-propionate, interacts 111 with the distal histidine residue and is involved in a more extensive hydrogen bonding network than the lysyl residue in horse heart Mb. Ser92Ala and Ser92Leu variants also possess increased rate constants for heme dissociation over wild-type pig Mb (Smerdon et al., 1993). Mechanism of Heme Reorientation. The mechanism by which the heme prosthetic group bound to apoMb achieves the rotation required for interconversion of heme orientational isomers is an interesting problem. Conversion from the minor (m) form to the major (M) form could occur through one of two limiting mechanisms, an intramolecular rearrangement (m «* M) or an intermolecular pathway (m ** A «* M) (Figure 34). An intramolecular mechanism must involve a considerable degree of unfolding of the protein to allow the large, planar heme group to rotate 180° within the heme binding pocket. An intermolecular mechanism requires complete dissociation of the heme prosthetic group from the heme pocket. One means of evaluating the likelihood of these two mechanisms is through examination of the rate constants for the two opposing mechanisms. The rate constants for heme dissociation (k.Hor ) from two Mb variants (5 x 10"3 min"1 for Lys45Glu/His97Leu and 2 * 10"3 min"1 for Lys45Glu/Ser92Asp) were determined at 25 °C and pH 7.0 for comparison with the reorientation rate constants. These variants were chosen for this purpose because they exhibit relatively large values for k.H and are, therefore, more amenable to accurate kinetic measurements under these solution conditions. Because the heme disorder equilibrium constants (A^ -k{l kf) and the heme association rate constants for both orientations (A^M = k2m) are the same for all the variants, the ratios of the rate constants for heme dissociation from both orientational isomers must also remain the same. Therefore, the rate constants for heme dissociation from the minor orientation (k™) can be calculated to be 6 x 10"2 min"1 and 2.4 * 10"2 min"1 for Lys45Glu/F£is97Leu and Lys45Glu/Ser92Asp, respectively. Comparison of these values to those of the forward reorientation rate constants (£f) displayed in Table 6 reveals that the dissociation of 112 heme from the heme cavity is slightly slower than the forward reorientation rate constant. These rate constants can be compared in the case of wild-type Mb on the basis of published values. The value of #!M determined for sperm whale Mb (Smith et al., 1982) yields a = 1.3 * 10"4 min"1, which is also only slightly lower than the forward reorientation rate constant (3.1 x 10"4 min"1; La Mar et al. 1984). Previous experiments by La Mar et al. (1984) indicate that heme reorientation occurs more rapidly than heme displacement which in turn suggests that heme reorientation does not involve exchange of heme with the bulk solvent and implies that it occurs through an intramolecular mechanism. This observation is seemingly at odds with the observation here of the small increase in the rate constant for heme reorientation (£f) over heme dissociation (k™). However, preliminary kinetic simulations show that although kf is only slightly greater than k™, the intramolecular mechanism is the dominant pathway. In other words, the reaction mechanism for this process is sufficiently complex that simply comparing the magnitudes of the rate constants does not provide an accurate picture of the reorientation process. Therefore, our results are not in disagreement with previous results of La Mar et al. (1984) Activation energy measurements. Determination of the Arrhenius activation energies for heme dissociation from the Mb variants provides a measurement of the energy barrier that must be overcome for the heme to dissociate from the protein and gives an indication of the stability of the protein-heme complex. The energies of activation for dissociation of heme from the Mb variants decreases in an additive manner when the amino acid residues that form hydrogen bonds with the heme propionates and nearby protein residues are sequentially replaced. In combination, these three electrostatic interactions are responsible for - 3 0 % of the total activation energy for heme dissociation from Mb. In the case of cytochrome b5, a decrease in activation energy is also observed when the propionate-protein interactions are disrupted. These observations provide another indication of the 113 importance of these interactions in the stability of the heme-protein complex. Determination of the equilibrium binding constants from the heme association and dissociation rates allowed Hargrove and coworkers (1996b) to estimate the relative contributions of the various types of interactions to overall heme affinity in sperm whale Mb. The individual contributions to the affinity of apoMb for heme were found to be - 5 0 % for the hydrophobic interactions between the apolar heme pocket and the heme, - 2 5 % for the proximal histidine-iron bond and - 2 5 % for the specific electrostatic interactions between the protein and the heme (Hargrove et al., 1996a). The estimation of - 2 5 % for the electrostatic interactions is in agreement with the - 3 0 % contribution of the heme propionate-protein interactions to the total activation energy for heme dissociation determined here for horse heart Mb. Thermal stability measurements. The importance of the interactions between the protein and the heme propionate substituents on the thermal stability of the holoMb structure was evaluated by monitoring the decrease in oc-helical secondary structure as a function of temperature. As each of the propionate interactions was removed in the Mb variants, the melting temperature of the protein decreased - 3 - 5 °C. Nevertheless, with all three of the hydrogen bonds removed in the quadruple variant, the thermal stability of the protein was not severely compromised. Cytochrome bs was also destabilized by the removal of the protein-propionate interactions. Therefore, these interactions between the protein and the heme propionates play a small but observable role in the stabilization of the secondary structure of these two heme proteins. Recently, Hargrove and Olson (1996c) also showed that the resistance of sperm whale holoMb to denaturation is determined primarily by the heme affinity. This observation is consistent with the results here in horse heart Mb where systematic removal of the propionate-protein interactions increased heme dissociation (decreased heme affinity) and decreased the thermal stability of the holoprotein. Interpretation of small changes in rate constants, as observed in effects of single amino acid 1 1 4 substitutions on the kinetics of heme dissociation, is difficult. However, the dependence of the rate constants for heme dissociation, the activation energies for heme dissociation and thermal stability on the number of hydrogen bonds formed by the heme propionates with the apoprotein supports the conclusion that these interactions are important to the stability of the protein-heme complex. The good correlation observed between the temperature required for thermal denaturation and the activation energy required for heme dissociation (Figure 35) provides additional support for these conclusions. These results suggest a method for subtle modifications in protein-heme affinity, without causing large changes in the heme binding pocket, through modulating the number of protein-propionate interactions. Modulation of proximal histidine character. For both Mb and cytochrome b5, the stability of the heme-protein complex is dependent on the characteristics of the histidyl residues that coordinate to the heme iron atom. In Fe(II)-Mb, the Fe2+-His93 bond is effectively covalent in nature, consistent with slow heme dissociation (Hargrove et al., 1994) and heme reorientation (Jue et al., 1983). Reconstitution of Mb with protoporphyrin IX, heme lacking the iron atom, results in very rapid equilibration to the major heme orientation suggesting coordination of the proximal histidine residue to the heme iron atom is also important in the initial trapping of the minor orientation (La Mar et al., 1989). In Fe(III)-Mb, the proximal histidine residue protonates as the pH is decreased below pH 5.0, the Fe3+-histidine bond is disrupted, and the rate constant for heme dissociation increases. Cytochrome b5 exhibits a pH dependence of heme dissociation kinetics that is similar to and has the same origin as that of Mb. 115 o < 64 68 72 76 80 T m ( ° C ) Figure 35. Correlation between the activation energy for heme dissociation and the thermal stability (TM) of the myoglobin variants. The activation energies were determined from the temperature dependence of heme dissociation at pH 5.0, and the melting temperatures were determined from the CD measurements at pH 7.0. The Ser92 hydroxyl group and the Leu89 main chain carbonyl group, form a three-center hydrogen bond with the proximal histidine N3 proton in wild-type Mb (Evans & Brayer, 1990). Substitution of Ser92 in Mb removes one hydrogen bond to the proximal histidine (His 93) as well one to the heme 7-propionate group. The inability to observe electron density for the Asp92 in the electron density map of Ser92Asp Mb crystals (Lloyd et al., 1996) is consistent with residue mobility that would prevent or at least compromise the hydrogen bonding interaction of this residue with both His93 and the heme 7-propionate. The observed effect of the Ser92 substitutions on both heme reorientation and heme dissociation was greater than expected. The additional destabilization of the heme-protein complex resulting from the Ser92Asp or Ser92Ala substitutions in Mb could be 116 attributed to the removal of the hydrogen bond between the proximal histidine and the Ser92 residue. Removal of this hydrogen bond results in decreased imidazolate character of the proximal histidine which weakens the iron-histidine bond (Goodin and McRee, 1993). Recently, hydrogen bonding to the proximal histidine has been shown to be a key determinant of the length of the Fe-His bond in Mb and CcP (Sinclair et al., 1996). Alternatively, the destabilization could be due to increased solvent accessibility on the proximal side of the heme binding pocket (Smerdon et al., 1993). This additional destabilizing effect is also reflected in the finding that the activation energy for heme dissociation from the Ser92Asp/Lys45Glu and Lys45Glu/Lys63Glu/His97Leu/Ser92Ala variants is 4-6 kJ/mol less than that observed for the His97Leu/Lys45Glu and DME-Mb variants possessing the same number of propionate-protein hydrogen bonds. This difference in activation energy is somewhat less than the activation energy (-10 kJ/mol) required for disruption of the His-Fe(III) bond of leghemoglobin at pH 7 (Smith et al., 1991; Paul et al., 1991). As indicated above, the results of the current study complement and extend those of previous reports concerning the influence of heme substituent groups on the interaction of heme with apoMb and apocytochrome b5 that employed structural analogues of heme. Through use of modified heme derivatives reconstituted into apoMb and apocytochrome b5, the heme vinyl groups (Keller et al., 1976; La Mar et al., 1981, 1984, 1986), the heme propionate groups (La Mar et al., 1985, 1986, 1989; Hauksson et al., 1990; Santucci et al., 1993), and axial ligands (La Mar et al., 1989; Yee & Peyton, 1991) have been suggested to be important to the dynamics of heme binding to both apoMb and apocytochrome b5. Our studies represent a comprehensive evaluation of the role the propionate-protein hydrogen bonding interactions play in the binding of heme to proteins. These interactions play a partial role in stabilizing the heme-protein complex along with other factors such as hydrophobic interactions and axial ligation. The hydrogen bonding interactions of the heme propionates are more 117 influential on the kinetics of heme reorientation. Results also suggest a more intramolecular mechanism for heme reorientation in which the heme rotates in association with the protein then rapidly rebinds. The heme partially dissociates from the protein cavity so rotation can occur in a less hindered manner but does not fully dissociate into free solution. ES-MS as a technique for assessing stability of non-covalent complexes. Previous studies have shown that the non-covalent complexes formed by proteins can be dissociated by collisionally activated dissociation (CAD) in the ion sampling interface (Huang et al., 1993; L i et al., 1993; Light-Wahl et al., 1994). However, there have been no reports of a systematic study in which the stability of such complexes has been compared in both the gas and solution phases. The ensemble of proteins created here has been shown to vary systematically in the stability of the heme-protein complex and provides an ideal system for evaluating the potential of ES-MS in the analysis of the stability of non-covalent complexes in the gas phase. In this study, the stability of the heme-protein complex in the gas phase was assessed by CAD in the ion sampling interface. The collisional energy resulting from collisional activation is transferred proportionally into internal energy of the complexes and results in heme dissociation. Consequently, the voltage difference applied at the sampling interface (V D) can be used as an indication of the energy required to dissociate the complex in the gas phase. Dissociation voltages of the various charge states. In this experiment, the relative ratio of holoprotein to apoprotein for each charge state pair was used to determine the extent of heme dissociation. Variation in the voltage difference (AV o s ) required for 50% dissociation for each charge state necessitated the consideration of the average observed for the four charge states (VD). The decrease in A V o s required for 50% dissociation of heme with increasing charge state (Figure 23) may derive from several sources. First, molecular ions of higher charge state experience a greater acceleration in the electric field between the orifice and skimmer and thus can acquire a given internal 118 energy at proportionately lower voltages. This effect should result in a constant product (charge x voltage) for a given protein. The data obtained in this study are consistent with this mechanism; all proteins show a constant charge * voltage with standard deviations < 6% (Table 11). Second, molecular ions with higher charge states may have different structures that may be destabilized by Coulomb repulsion. The differences in A V o s for dissociation of heme from the range of variants are nearly constant over the various charge states which indicates that any destabilization caused by increasing charge state is the same for all proteins studied or that the effects are small relative to the changes caused by the substitutions. Third, dissociation of neutral heme from the holoprotein may introduce a systematic variation of A V o s that is not related to the energy required to dissociate the complex. The dissociation of a neutral heme species from all charge states of Mb was observed in the MS/MS experiments where the heme-protein complex was dissociated in the collision cell. In these experiments, the ions are activated by CAD through a few relatively high energy collisions. Conversely, in the CAD experiment in the sampling interface, ions are activated by many relatively low energy collisions. Therefore, the relative dissociation of neutral and charged heme can differ in these types of experiments. It is also possible that neutral heme loss does not occur during CAD in the sampling interface. However, the effects of neutral heme dissociation on the interpretation of the results were considered in detail. The data simulation established that even with a significant degree of neutral heme dissociation, the method for data analysis was still appropriate. While the origin of the variation of dissociation voltage with charge state is uncertain in these analyses, the average voltage remains a useful measure of the voltage required to dissociate heme from a given protein. Complex stability in the gas phase. Within the series of Mb and cytochrome b5 variants, differences of 4-10 V were observed in the average dissociation voltage (V D) required to dissociate 119 heme from these non-covalently associated complexes. This observation suggests that the ES-MS method used here is sufficiently sensitive to detect subtle differences in the stabilities of protein-small molecule complexes. The specific interactions studied here must be present in the wild-type protein in the gas phase for differences in the stability of heme binding to the variant proteins to be observed. The observation that the destabilizing effects of a single substitution that removes a specific hydrogen bond in a non-covalently associated complex can be detected in the gas phase molecular ions with this method is particularly encouraging. The energy required for complex disruption in the gas phase was compared to the Arrhenius energy of activation for heme dissociation from the variants in solution. A good correlation between these two properties was obtained for the forms of Mb and cytochrome b5 included in this study (Figure 24, Results section B6), so the same relative stabilities for the interaction of heme and apoprotein must exist in both the gas and solution phases. Both the observations of small differences in stability of the heme-protein interaction in the gas phase and the correlation with the stability of these same complexes in solution strongly suggest that the structures of the molecular ions in the gas phase are similar to the structures of the proteins in solution. If the protein was grossly misfolded in the gas phase, a correlation with the behavior of the variants in solution would not be obtained nor would the effects of disrupting individual hydrogen bonds be observed. As the hydrogen bonding interactions in these complexes were of primary interest here, further studies are required to see if correlations can be observed between behavior in solution and the gas phase as other types of bonding interactions are varied. Another technique, blackbody infrared radiative dissociation (BIRD), has recently been used for the determination of the Arrhenius activation energies (E^ for the dissociation of large biomolecular non-covalent complexes in the gas phase (Williams et a l , 1996; Schnier et a l , 1996). 120 Both Mb and Hb exhibited similar trends in activation energy for heme dissociation in the gas phase and in solution (Williams et al., 1996). The values determined for the E A for heme dissociation from Mb in the gas phase compare reasonably well with the values determined here for the E A for heme dissociation in solution. These new results in combination with the findings of the current study provide further evidence for similarity in structure of the non-covalently associated complexes in the gas and solution phases. The one exception to the correlation between the gas phase and solution phase stabilities was the difference observed between phases in the variants containing substitutions at position 92. Replacement of Ser92 not only eliminates a hydrogen bond to heme 7-propionate, it also eliminates a hydrogen bond to the proximal histidine. This latter interaction is believed to modulate the character of the proximal histidine and therefore the strength of the heme Fe(III)-histidine bond (Goodin and McRee, 1993). Replacing Ser92 was shown to decrease the affinity of the protein for heme in solution while increasing the affinity of the protein for heme in the gas phase. Analysis of additional Mb variants in which the proximal ligand has been modified must be studied to understand the role this type of interaction plays in determining the stability of the heme-protein complex in the gas phase. The activation energy required for heme dissociation from Mb in solution is less than that required for lipase- and trypsin-solubilized cytochrome bs, yet a greater voltage is required to dissociate heme from Mb (MW -17600) than from cytochrome b5 (MW -10000) in the gas phase. This behavior is not surprising as larger ions are generally expected to require greater internal energies for heme dissociation on the time scale of the mass spectrometry experiment (Marzluff et al., 1994). The present results are in qualitative agreement with this expectation, although it is clear that a comprehensive understanding of this effect will require consideration of additional factors such as relative mass effects and collision cross sections in the ion activation process. In addition, as the 121 difference in the number of atoms present in wild-type and variant proteins is small (Table IT), no difference in the mechanism of activation is expected for a particular protein and its variants. Another interesting observation of the current study was the absence of any change in the charge state distribution of the Mb variants in which up to three basic residues were replaced. The process by which the protein ions acquire charge during the electrospray process is not well understood. Recent work suggests that a large decrease in the pH of the protein solution (up to 4 pH units) occurs as the result of redox reactions induced by the highly charged metal capillary of the electrospray ion source (van Berkel et al., 1996). A large decrease in pH of the injected sample would explain the large number of charges (14+-10+) seen when Mb is sprayed from solution at pH 8.0. Mb contains a large number of surface histidyl residues that titrate in this pH range and a drop in solution pH would result in an increase in positive surface charge (Shire et al., 1974). In addition, the ability to dissociate these heme-protein complexes with relatively mild sample interface conditions is more consistent with a destabilized heme-protein complex at lower pH than with the stability of heme binding expected at higher pH (KD ~10"15 M) (Banerjee, 1962). Finally, the activation energies determined in solution were obtained at a lower pH (pH 5.0) so a decrease in pH during the sample injection process will reduce the difference in solution conditions employed in the two types of experiments. Summary. The work described here represents the first study in which the contributions of individual hydrogen bonding interactions to the stability of a non-covalently associated protein-small molecule complex have been characterized in both the gas and solution phases through analysis of a series of modified proteins that differ systematically in their hydrogen bonding characteristics. The implications of these results are twofold. First, the linear correlation of dissociation voltages (V D) determined in the gas phase with the activation energies for heme dissociation in solution requires 122 considerable similarity in the structures of the protein in the gas and solution phases. This conclusion is consistent with the observation that the gas-phase heme-protein complex is more compact than that of the apoprotein and suggests that the gas phase protein is folded around the heme group as it is in solution (Douglas & Collings, 1996). The evidence presented here demonstrates that the hydrogen bonds known to stabilize the heme-protein complex in solution remain intact in the gas phase ion and contribute to the stability of the gas phase complexes in a similar fashion. Consequently, the gas phase ion must retain much of the solution structure for at least the millisecond time scale of this experiment. Secondly, the ability to observe small differences in protein-small molecule stability in two separate proteins suggests that ES-MS may be a generally applicable method for characterizing non-covalently associated protein-ligand complexes in the gas phase. One notable benefit of ES -MS is the small amount of sample required. The conventional electrospray source used here operated at 1 pL/min so the -160 pmol of each protein was consumed in the 40 minutes of data acquisition per protein. Developments of low flow electrospray sources promise to lower this sample requirement at least 100-fold (Wilm & Mann, 1994; Valaskovic et al., 1996; Wilm & Mann, 1996). By comparison, the method used here for determination of the activation energy for heme dissociation in solution required -300 times more protein and a minimum of five hours per protein. Solution methods often require additional indirect factors to indicate the presence of complex (e.g., spectroscopic probes, indicators of loss/gain of enzymatic activity). Finally, automation could further reduce the mass spectrometric data acquisition time and sample requirements and permit development of a rapid screening technique for evaluating stability of non-covalently associated complexes. 123 C. Functional Consequences of Electrostatic Modification of the Active Site of Myoglobin. Azide binding. Analysis of azide binding to the ferric form of Mb provides a sensitive probe of distal pocket structure and character and provides information complementary to characterization of 0 2 binding to the ferrous form of the protein. Both ligands have anionic character and are stabilized in the distal pocket by interaction with hydrogen bond donors or positive dipoles (Brancaccio et al., 1994). In addition, binding of azide and 0 2 both require the displacement of a water molecule from the distal heme pocket (Brancaccio et al., 1994). Azide complexes of wild-type and variant sperm whale, pig and human Mbs have been characterized previously (Cutruzzola et al., 1991; Biram et al., 1993; Brancaccio et al., 1994; Smerdon et al., 1995), however, work had not been reported for horse heart Mb-azide complexes. Azide binding to horse heart Mb possessing distal and proximal heme pocket substitutions have been investigated here to provide some insight into the character of the heme binding pocket of these variants and, more generally, to horse Mb. Replacement of the charged residues near the distal heme pocket changes the equilibrium binding constants of Mb for azide. Positive charge, introduced at position 67 in the Val67Arg variant, stabilizes binding of the anionic azide ligand and increases azide binding affinity. Alternatively, the substitution in the Lys45Glu variant destabilizes azide binding through charge repulsion between the nearby glutamyl residue and the azide. The Lys45Glu/Lys63Ser and Lys45Glu/Lys63Glu variant, which also possess greater negative charge near the distal pocket, exhibit azide binding affinity that is decreased by ~8 and 14-fold, respectively, relative to that of the wild-type protein. The first reported infrared spectra ofMbN 3 complexes (McCoy & Coughey, 1970) identified two infrared bands, that were assigned to a low-spin (2023 cm'1) and a high spin (2046 cm"1) species. These two species were later shown to be in thermal equilibrium at room temperature (Alben & Fager, 1972). The azide complexes of the Lys45Glu, Lys45Glu/Lys63Glu, and Val67Arg horse heart 124 Mb variants exhibit FTIR spectra with high spin and low spin band maxima similar to that observed for wild-type MbN 3 , indicating similar binding conformations of the azide anion. The small changes seen in the vm a x of the low-spin azide infrared bands, observed for the Lys45Glu and Val67Arg variants are opposite in direction which suggests these small shifts are due to changes in the electrostatic character at these positions. These shifts in have been attributed to the interaction of charge with the transition dipole moment of the intra-azide bonds (Bogumil et al., 1994). These substitutions change the electrostatic character of the residues near the distal heme pocket and this change is sensed by the bound azide. The Lys45Glu/Lys63Glu variant exhibits a shift similar to that of the single variant, Lys45Glu, which indicates that the additional negative charge at position 63 is too distant to influence the bound azide. FTER analysis of MbCO complexes of sperm whale and pig Mb variants also indicated that the major factor governing the v c o is the electrostatic potential surrounding the bound CO ligand (Li et al., 1994). The effects of altering the electrostatic properties of the active site on the affinity of azide for Mb has been studied for other species, and the results are consistent with those observed here. For example, replacing Lys45 of pig Mb or Arg45 of sperm whale Mb with a glutamyl residue produced a 3-fold decrease in azide binding affinity (Brancaccio et al., 1994). An intermediate effect was observed for the Lys45Ser variant of pig Mb (Brancaccio et al., 1994): In addition, the charge of the residue at position 67 affects azide binding affinity as observed in the increased affinity of the Thr67Arg variant and the decreased affinity of the Thr67Glu pig Mb variant for azide (Brancaccio et al., 1994). The replacement of the distal histidine residue with nonpolar residues dramatically decreases the affinity of Mb for azide. The distal histidine has been shown to form a hydrogen bond to the bound azide in the three-dimensional structure of the horse heart Mb azide complex (R. Maurus, 125 unpublished, Dr. G. D. Brayer lab). One of the principal reasons for the reduced affinity of the His64Thr and His64Ile Mb variants for azide is the inability of these distal residues to form a hydrogen bond with and thereby stabilize the bound azide ligand. The azide equilibrium constant for azide dissociation from the His64Lys variant is only four times weaker than that observed for wild-type Mb and much greater than that of the other distal ligand variants. This behavior is attributed to the potential of the lysine to form a hydrogen bond with the bound azide. In a sperm whale Mb variant designed to model the active site of A. limacina Mb, the decreased azide affinity of the His64Val variant was counterbalanced by the inclusion of a Thr67Arg substitution which was proposed to stabilize the bound azide through a hydrogen bonding interaction (Cutruzzola et al., 1991). Another important determinant in azide binding is the size of the distal ligand and, therefore, the steric barrier to the entry and exit of azide. Brancaccio et al. (1994) demonstrated a rough inverse correlation between the rate constants for the association and dissociation of azide and the size of residue 64. This steric hindrance could account for the further destabilization of the bound azide in the His64Ile variant relative to that of the His64Thr variant on the basis of the greater size of the isoleucine side chain. The further reduction in polarity of the distal pocket introduced by the His64Ile substitution could also contribute to the decreased binding of this anionic azide ligand. The results presented here are consistent with the proposed distal histidine 'trap door' theory for azide binding to horse heart Mb. First proposed by Perutz and Matthews in 1966 for the binding of ligands to Mb and Fib, it is thought that azide enters the distal pocket through a polar channel formed by Arg(Lys)45, Val68 and Thr67 and that the distal FIis64 residue blocks the end of the channel (Brancaccio et al., 1994). The histidine side chain can rotate and swing out toward solvent to open the channel. Changing the electrostatic character of this channel can repulse or attract the 126 anionic ligand as seen in the Lys45Glu and Val67Arg variants, respectively. Additionally, changing the size of the residue blocking the end of the channel (His64) can also affect azide binding. Proximal heme pocket variants (Ser92 and His97) have very little effect on azide binding as they are removed from the polar channel and the region of the active site involved in ligand binding. HemepKA The pH dependent changes that occur in the visible spectrum of metMb between pH 6 and 10 are due to the titration of the water molecule that is coordinated to the exchangeable ligand binding site of the heme iron. The titration of the molecular water bound to the heme iron is modulated by the protein environment surrounding the heme as the pKA for this equilibrium of wild-type horse Mb (8.93) (Antonini & Brunori, 1971) is significantly higher than the pKA of free heme (7.6) (Shack & Clark, 1947). Changes in the electrostatic character of the residues near the heme binding pocket were observed to affect the pKA of this coordinated water molecule. The Lys45Glu and Lys45Glu/Lys63Glu variants displayed increases in p^TA over the wild-type protein due to the destabilization of the anionic OH" ligand resulting from the proximity of the negatively charged side chains. Intermediate effects were observed in the Lys45Glu/Lys63Ser variant. Position 97 and 92 variants exhibited little change in the heme pKA as these residues reside on the proximal side of the heme. The effects of altering the electrostatic character of the heme binding pocket have been observed in other horse heart Mb variants as well. The heme pKA of DME -Mb (-8.5) is lower than that of wild-type Mb because the two negatively charged carboxylates are replaced with two uncharged methyl esters (Lim, 1990). Also, an increase in pKA of the bound water molecule was observed when the two aliphatic residues at positions 67 and 68 in the distal heme binding pocket were replaced with more polar residues (Val67Ala/Val68Ser, pKA = 9.3) (Hildebrand et al., 1996). A variant in which an aliphatic residue on the distal side of the heme was replaced with a positive 127 residue produced a decrease in heme pKA (Val67Arg Mb, pKA = 8.0) as expected on the basis of electrostatic considerations (E. Lloyd, unpublished). Therefore, the electrostatic character of the residues near the entrance to the distal pocket that destabilize azide binding also destabilize the binding of OH' relative to H 2 0 . Reduction potentials. Not surprisingly, the changes in the charged residues surrounding the distal heme pocket also influence the reduction potential of the heme iron. As the residues were systematically replaced with more acidic residues, the reduction potential of the variants decreased (Table 15). This result is presumably attributable to a stabilization of the Fe(III) form by the increased negative character of the distal pocket. Electrostatics have been argued previously to be one of the variables that influence the reduction potentials of metalloproteins (Aviram et al., 1981; Rodgers & Sligar, 1991; Caffrey & Cusanovich, 1991; Rivera et al., 1994). A large increase in the midpoint potential of DME-Mb has been observed previously ( A E M = 40 mV under similar solution conditions) (Lim, 1990). Esterification of the heme propionates decreases the negative character of these substituents which would be expected to increase the midpoint potential. A similar increase in reduction potential was observed in the DME-reconstituted form of cytochrome b5 (Reid et al., 1984). An increase in potential is also observed in the variant, Val67Arg, where the positive charge near the heme on the distal side is increased (E. Lloyd, unpublished). In a variant designed to increase the polarity of the distal pocket, Val67Ala/Val68Ser, the reduction potential was observed to decrease (Hildebrand et al., 1996). Other factors are known to influence the reduction potential of heme proteins are as follows: the nature and orientation of the axial ligands (Walker et al., 1986; Safo et al., 1991), the heme pocket hydrophobicity (Kassner, 1972), heme orientation (Walker et al., 1988), and the nature and 128 orientation of the heme substituents (Reid et al., 1986; Lee et al., 1991). The variants of Mb possessing substitutions on the distal side of the heme studied here were shown to have the same equilibrium heme orientation as wild-type Mb, ruling out this factor as a contribution to the altered reduction potentials of the variants. A number of factors could be responsible for the change in reduction potential of the proximal heme pocket variants. The 'H -NMR spectra of the variants possessing the His97Leu substitutions suggest a small reorientation of the heme 7-propionate. This substitution also increased the hydrophobicity in this region. As indicated previously, the small changes in the reduction potentials of the variants possessing the Ser92 mutations could result from a combination of factors that cannot be completely resolved (Lloyd et al., 1996). Modulation of function by metal binding. The ability of the Lys45Glu and Lys45Glu/Lys63Glu Mb variants to bind Mn(II) ions provides another method for modulating the electrostatic character of this distal region of the protein in a fashion that influences functional characteristics. The distance between the bound Mn(II) ion and the heme iron determined from the three dimensional structure of the Lys45Glu/Lys63Glu variant in the presence of Mn(II) is 10.4 A (R. Maurus, unpublished, Dr. G. D. Brayer lab). Therefore, the Mn(II) ion binds to the protein in sufficient proximity to the azide binding site to influence the bound azide ligand. In the FTIR experiment, adding Mn(II) to both the Lys45Glu and Lys45Glu/Lys63Glu variants shifts the v,,,^  of the low-spin azide band to higher energy from the v,^ measured in the absence of metal, as expected when the electrostatic potential is made more positive. The electrostatic influence of metal binding was also observed in the equilibrium binding constants for azide as these variants exhibited greater azide binding affinity in the presence of Mn(II) than in the absence of metal ion. The binding of Mn(II) to this region of the protein masks the negative charges of the carboxylate groups of the glutamyl side chains and the heme propionates and confers a more positive potential to the region that 129 is more electrostatically favorable for azide binding (Figure 32). Neither of these effects were observed with wild-type Mb, indicating they are a consequence of the binding of metal ions. The reduction potential of the Fe(n)/Fe(III) redox couple in Mb was shown to be influenced by the electrostatic environment surrounding the heme. Again, the binding of Mn(II) to the Lys45Glu/Lys63Glu Mb variant provides a convenient method for modulating the reduction potential of this protein. The binding of Mn(II) ions to the Lys45Glu/Lys63Glu variant results in a - 3 0 mV increase in the reduction potential, while similar treatment of wild-type Mb has no effect. The increase in positive charge due to the association of the metal destabilizes the Fe(III) form of the protein and increases the midpoint potential. The change observed in the reduction potential due to binding of the Mn(II) is similar in magnitude but opposite in direction to the change in potential observed for each lysine to glutamic acid substitution in the Lys45Glu and Lys45Glu/Lys63Glu variants (-25 mV). In summary, a number of variants have been prepared which alter protein residues within the heme binding pocket. The variants in which the electrostatic character of the residues surrounding the distal pocket have been altered exhibit pronounced effects on distal pocket properties. The increased negative charge at positions 45 and 63 destabilizes the binding of anionic ligands in the distal pocket and stabilizes additional positive charge on the heme iron. These effects are manifested over a relatively long range as the negatively charged functional groups of Glu45 and Glu63 are -13 and 15 A away from the heme iron, respectively, based on the three dimensional structure of the Lys45Glu/Lys63Glu Mb variant in the absence of metal (R. Maurus, unpublished, Dr. G. D. Brayer lab). In addition, MnfTJ) binding to protein surface residues can also modulate anionic ligand binding and the reduction potential of the heme iron by changing the electrostatic potential of the environment surrounding the solvent exposed heme edge. 130 D. Concluding Remarks The central role of molecular recognition in all life processes makes a detailed understanding of the specific non-covalent interactions involved in such processes important. The work presented here considers a number of specific examples of molecular recognition that include metal ion binding, heme-apoprotein interactions, and binding of small anionic ligands. The major achievements of this work can been summarized in the following manner. A transition metal ion binding site, capable of participating in low levels of electron transfer with the heme prosthetic group, has been introduced on the surface of Mb and characterized both structurally and functionally. The role of the hydrogen bonding interactions between the heme propionate groups and specific protein residues in heme binding dynamics and stability has been fully characterized. A promising method for studying non-covalent complex stability in the gas phase has been developed using electrospray mass spectrometry. The importance of electrospray mass spectrometry as a tool in the study of molecular recognition in biology has been further recognized by the observation here of strong correlations between the solution and gas phase non-covalent protein-heme complex stability. Also, the electrostatic nature of the protein residues near the heme binding site have been shown to influence the properties of the distal heme pocket, specifically ligand binding and reduction potentials. Finally, the modulation of the electrostatic character of the heme binding site through the use of metal ions and the subsequent effects on ligand binding and the reduction potential of the protein has been demonstrated. 131 V. R E F E R E N C E S Alben, J. O., & Fager, L. Y. (1972) Biochemistry 11, 842-847. Aleksandrov, M. L., Gall', L. N., Krasnov, N. V., Nikolaev, V. I., Paulenko, V. A., & Shkurov, V. A. (1984) Dokl. Adak. Nauk. SSSR 277, 379. Allentoff, A. J., Bolton, J. L., Wilks, A., Thompson, J. A., & Ortiz de Montellano, P. R. (1992) J. Am. Chem. Soc. 114, 9744-9749. Antonini, E., & Brunori, M. (1971) Hemoglobin and Myoglobin in their Reactions with Ligands, Vol. 21, American Elsevier Publishing Co., Inc., New York. Aojula, H. S., Wilson, M. T., & Drake, A. (1986) Biochem. J. 237, 613-616. Aojula, H. S., Wilson, M . T., Moore, G. R., & Williamson, D. J. (1988) Biochem. J. 250, 853-858. Archibald, F. (1986) CRC Crit. Rev. Micro. 13, 63-109. Arnold, F. H., & Haymore, B. L. (1991) Science 252, 1796-1797. Aviram, I., Myer, Y. P., & Schejter, A. (1981) J . Biol. Chem. 256, 5540-5544. Baca, M. , & Kent, S. B. H. (1992) J. Am. Chem. Soc. 114, 3992-3993. BafFa, O., Say, J. C , Tabak, M. , & Nascimento, O. R. (1986) J. Inorg. Chem. 26, 117-125. Banaszak, L. J., Watson, H. C , & Kendrew, J. C. (1965)/. Mol. Biol. 12, 130-137. Banci, L., Bertini, I., Bird, T., Tien, M. , & Turano, P. (1993) Biochemistry 32, 5825-5831. Banerjee, R. (1962) Biochim. Biophys. Acta 64, 368-384. Barker, P. D. & Mauk, A. G. (1992) J. Am. Chem. Soc. 114, 3619-3624. Bellelli, A., Foon, R., Ascoli, F., & Brunori, M. (1987) Biochem. J. 246, 787-789. Berg, J. M. (1988) Pre. Natl. Acad. Sci. USA 85, 99-102. Berg, J. M . (1990) J . Biol. Chem. 265, 6513-6516. Berg, J. M. , & Shi, Y. (1996) Science 271, 1081-1085. 132 Bertini, I., & Luchinat, C. (1986) NMR of Paramagnetic Molecules in Biological Systems, Benjamin/Cummings, Merlo Park, CA. Biemann, K. (1992) Annu. Rev. Biochem. 61, 977-1010. Biram, D., Garratt, C. J., & Hester, R. E. (1993) Biochim. Biophys. Acta 1163, 67-74. Bogumil, R., Hunter, C. L., Maurus, R., Tang, H.-L., Lloyd, E., Brayer, G. D., Smith, M. , & Mauk, A. G. (1994) Biochemistry 33, 7600-7608. Bolognesi, M. , Onesti, S., Gatti, G., & Coda, A. (1989)/. Mol. Biol. 205, 529-544. Bosshard, H. R., Anni, H., & Yonetani, T. (1991) in CRC Peroxidases in Chemistry and Biology (Everse, J., Everse, K. E., & Grisham, M B . , Eds.) pp 51-84, CRC Press, Boston. Brancaccio, A., Cutruzzola, F., Allocatelli, C. T., Brunori, M. , Smerdon, S. J., Wilkinson, A. J., Dou, Y., Keenan, D., Ikeda-Saito, M. , Brantley, R. E., & Olson, J. S. (1994)7. Biol. Chem. 269, 13843-13853. Braxton, S., & Wells, J. A. (1992) Biochemistry 31, 7796-7801. Bruins, A. P. (1991) Mass Spec. Rev. 10, 53-77. Bujons, J., Dikiy, A., Ferrer, J. C , Banci, L., & Mauk, A. G. (1996) Eur. J. Biochem. 243, 72-84. Bumpus, J. A., & Aust, S. D. (1987) BioEssays 6, 166-179. Caffrey, M. S., & Cusanovich, M. A. (1991) Arch. Biochem. Biophys. 285, 227-230. Caldecourt, V. J., Zakett, D., & Tou, J. C. (1983) Int. J. Mass Spec. Ion Phys. 49, 233. Carrell, J. L., Gluster, J. P., Burger, V., Mafre, F., Tritsch, D., & Biellmann, J.-F. (1989) Proc. Nat. Acad. Sci., USA 86, 4440-4444. Carver, T. E., Rohlfs, R. J., Olson, J. S., Gibson, Q. H., Blackmore, R. S., Springer, B. A., & Sligar, S. G. (1990) J. Biol. Chem. 265, 20007-20020. Catalano, C. E., Choe, Y. S., & Ortiz de Montellano, P. R. (1989)7. Biol. Chem. 264, 10534-10541. Chait, B., & Kent, S. B. H. (1992) Science 257, 1885-1894. Chen, Y . -H . , Yang, J. T., & Martinez, H. M. (1972) Biochemistry 11, 4120-4131. Childs, R. E., & Bardsley, W. G. (1975) Biochem. J. 145, 93-103. 133 Choe, Y. S., Rao, S. I., & Ortiz de Montellano, P. R. (1994) Arch. Biochem. Biophys. 314, 126-131. Chowdhury, S. K., Katta, V., & Chait, B. T. (1990) J. Am. Chem. Soc. 112, 9012-9013. Clark, W. M . (1960) Oxidation-Reduction Potentials of Organic Systems, Waverly Press, Inc, Baltimore. Coletta, M. , Ascenzi, P., Traylor, T. G., & Brunori, M. (1985) J. Biol. Chem. 260, 4151-4155. Collyer, C. A., Henrick, K., & Blow, D. M. (1990) Moi. Biol. 212, 212-235. Conroy, C. W., Tyma, P., Daum, P. H., & Erman, J. E. (1978) Biochim. Biophys. Acta 537, 62-69. Constantinidis, I., Satterlee, J. D., Pandey, R. K., Leung, H. K., & Smith, K. M . (1988) Biochemistry 27, 3069-3076. Corey, D. R., & Schultz, P. G. (1989) J. Biol. Chem. 264, 3666-3669. Cowan, J. A., Upmacis, R. K., Beraton, D. N , Onuchic, J. N., & Gray, H. B. (1988) Ann. N. Y. Acad. Sci. 550, 68-84. Cutruzzola, F., Allocatelli, C. T., Ascenzi, P., Bolognesi, M. , Sligar, S. G., & Brunori, M. (1991) FEBSLett. 282, 281-284. Czerminski, R., & Elber, R. (1991) Proteins 10, 70-80. de Duve, C. (1948) Acta Chem. Scand. 2, 264-289. Demmer, H., Hinz, I., Keller-Rudex, H., Koeber, K., Kottelesch, H., & Schneider, D. (1980) in Coordination compounds of manganese (Schleitzer-Rust, E., Ed.) pp 1-186, Springer-Verlag, New York. Dismukes, G. C , Zheng, M. , Hutchins, R., & Philo, J. S. (1994) Bioc. Soc. Trans. 22, 323-327. Douglas, D. J., & Collings, B. A. (1996)./. Am. Chem. Soc. 118, 4488-4489. Dunford, H. B. (1991) in CRC Peroxidases in Chemistry and Biology (Everse, L, Everse, K. E., & Grisham, M . B., Eds.) pp 1-24, CRC Press, Boston. Durley, R. C. E., & Mathews, F. S. (1996) Acta Cryst. D 52, 65-76. Dutton, P. L. (\91Z)Meth. Enzmol. 54, 411-435. 134 Egeberg, K. D., Springer, B. A., Martinis, S. A., Sligar, S. G., Morikis, D., Champion, P. M. (1990) Biochemistry 29, 9783-9791. Elber, R., & Karplus, M. (1990)7 Am. Chem. Soc. 112, 9161-9175 Emerson, S. D., & La Mar, G. N. (1990) Biochemistry 29, 1545-1556. English, A. M. (1994) in Encyclopedia of Inorganic Chemsitry: Iron: Heme Proteins, Peroxidases, & Catalases (King, R. B., Ed.) pp 1682-1697, John Wiley &Sons, Chichester. English, A. M. , & Tsaprailis, G. (1995) Advances Inorg. Chem. 43, 79-119. Erman, J. E., Vitello, L. B., Miller, M. A., & Kraut, J. (1992) J. Am. Chem. Soc. 114, 6592-6593. Evans, S. V., & Brayer, G. D. (1990)7 Mol. Biol. 213, 885-897. Feng, R., & Konishi, Y. (1993) 7. Am. Soc. Mass Spectrom. 4, 638-645. Fenn, J. B., Mann, M. , Meng, C. K., Wong, S. F., & Whitehouse, C. M . (1989) Science 246, 64-71. Ferrer, J. C , Turano, P., Banci, L., Bertini, I., Morris, I. K., Smith, K. M. , Smith, M. , & Mauk, A. G. (1994) Biochemistry 33, 7819-7829. Fersht, A. (1977) Enzyme Structure and Mechanism, 2 ed., W. H. Freeman and Company., 296-299. Fridovich, I. (1986) Afv. Enz. Rel. Areas. Mol. Biol. 58, 61-97. Funk, W. D., Lo, T. P., Mauk, M . R., Brayer, G. D., MacGillivray, R. T. A., & Mauk, A. G. (1990) Biochemistry 29, 5500-5508. Galaris, D., Sevanian, A., Cadenas, E., & Hochstein, P. (1990) Arch. Biochem. Biophys. 281, 163-169. Ganem, B., L i , Y.-T., & Henion, J. D. (1991) 7. Am. Chem. Soc. 113, 6294-6296. George, P., & Irvine, D. H. (1956)7. Colloid Interface Sci. 11, 327-329. Gersonde, K., & Netter, H. (1966) 7 Mol. Biol. 22, 211-221. Giacometti, G. M. , Traylor, T. G., Ascenzi, P., Brunori, M. , & Antonini, E. (1977)7. Biol. Chem. 252, 1AA1-1AA%. 135 Gibson, Q. H., Regan, R., Elber, R., Olson, J. S., & Carver, T. E. (1992)/. Biol. Chem. 267, 22022-22034. Glenn, J. K., & Gold, M. H. (1985) Arch. Biochem. Biophys. 242, 329-341. Glenn, J. K., Akileswaran, L., & Gold, M. H. (1986) Arch. Biochem. Biophys. 251, 688-696. Glusker, J. P. (1991) Adv. Prot. Chem. 42, 1-76. Gold, M. H., Wariishi, H., & Valli, K. (1989) ACS Symp. Ser. 389, 127-140. Goodin, D. B., & McRee, D. M. (1993) Biochemistry 32, 3313-3324. Gregory, D. S., Martin, A. C. R., Cheetham, J. C , & Rees, A. R. (1993) Protein Eng. 6, 29-35. Guillemette, J. G., Matsushima-Hibiya, Y., Atkinson, T., & Smith, M. (1991) Protein Eng. 4, 585-592. Hammel, K. E. (1989) Enzyme Microb. Technol. 11,116-111. Handel, T., & DeGrado, W..F. (1990)7. Am. Chem. Soc. 112, 6710-6711. Hardmann, K. D., Agarwal, R. C , & Freiser, M . J. (1982) J. Moi. Biol. 157, 69-86. Hargrove, M. S., Singleton, E. W., Quillin, M. L., Ortiz, L. A., Phillips, Jr., G. N., Olson, J. S., & Mathews, A. J. (1994a) J. Biol. Chem 269, 4207-4214. Hargrove, M. S., Krzywda, S. K., Wilkinson, A. J., Dou, Y., Ikeda-Saito, M. , Olson, J. S. (1994b) Biochemistry 33, 11767-11775. Hargrove, M . S., Barrick, D., & Olson, J. S. (1996a) Biochemistry 35, 11293-11299. Hargrove, M. S., Wilkinson, A. J., & Olson, J. S. (1996b) Biochemistry 35, 11300-11309. Hargrove, M. S., & Olson, J. S. (1996) Biochemistry 35, 11310-11318. Harris, W. R., & Chen, Y. (1994)/. Inorg. Biochem. 54, 1-19. Hassan, H. M. (1989) Adv. in Gen. 26, 65-97. Hauksson, J. B., La Mar, G. N., Pandey, R. K., Rezzano, I. N., & Smith, K. M. (1990) J. Am. Chem. Soc. 112, 6198-6205. Hegetschweiler, K., Saltman, P., Dalvit, C , & Wright, P. E. (1987) Biochim. Biophys. Acta 912, 384-397. 136 Heineman, W. R., Norris, B. J., & Goelz, J. F. (1975) Anal. Chem. 47, 79-84. Hellinga, H. W., Caradonna, J. P., & Richards, F. M. (1991) J. Moi. Biol. 222, 787-803. Hellinga, H. W., & Richards, F. M. (1991)/. Moi. Biol. 222, 763-785. Hellinga, H. W. (1996) in Protein Engineering: Principles and Practice (Cleland, J. L., & Craik, C. S., Eds.) pp 369-398, John Wiley and Sons, Inc., New York. Higaki, J. N., Haymore, B. L., Chen, S., Fletterick, R. J., & Craik, C. S. (1990) Biochemistry 29, 8582-8586. Higaki, J. N., Fletterick, R. J., & Craik, C. S. (1992) TIBS 17, 100-104. Hildebrand, D. P., Tang, H.-L., Luo, Y., Hunter, C. L., Smith, M. , Brayer, G. D., & Mauk, A. G. (1996)./. Am. Chem. Soc. 118, 12909-12915. Hochuli, E., Bannwarth, W., Dobeli, H., Gentz, R., & Stuber, D. (1988) Bio/Technology 6, 1321-1325. Hogg, N., Rice-Evans, C , Darley-Usmar, V., & Wilson, M . T. (1994) Arch. Biochem. Biophys. 314, 39-44. Huang, E. C , Pramanik, B. N., Tsarbopoulos, A., Reichert, P., Ganguly, A. K., Trotta, P. P., Nagabhushan, T. L., & Covey, T. R. (1993) J. Am. Soc. Mass Spectrom. 4, 624-630. Hughes, T. R., & Klotz, I. M. (1956) Meth. Biochem. Anal. Ill, 265-297. Ibers, J. A., & Holm, R. H. (1980) Science 209, 223-235. Ikeda-Saito, M. , Hori, H., Andersson, L. A., Prince, R. C , Pickering, I. J., George, G. N., Sanders, C. R., Lutz, R. S., McKelvey, E. J., & Mattera, R. (1992) J. Biol. Chem. 267, 22843-22852. Inubushi, T., & Becher, E. D. (1983) J. Mag. Reson. 51, 128-133. Iverson, B. L., Iverson, S. A., Roberts, V. A., Getzoff, E. D., Tainer, J. A., Benkovic, S. J., & Lerner, R. A. (1990) Science 249, 659-662. Joshi, D., & Gold, M. H. (1993) Appl. Environ. Microbiol. 59, 1779-1785. Jue, T., Krishnamoorthi, R., & La Mar, G. N. (1983)7. Am. Chem. Soc. 105, 5701-5703. Kanner, J., & Harel, S. (1995) Arch. Biochem. Biophys. 237, 314-321. 137 Kassner, R. J. (1972) Proc. Nat. Acad. Sci. USA 69, 2263-2267. Katta, V., & Chait, B. T. (1991) J. Am. Chem. Soc. 113, 8534-8535. Kawamura-Konishi, Y., Kihara, J., & Suzuki, H. (1988) Eur. J. Biochem. 170, 589-595. Keller, R., Groudinsky, O., & Wuthrich, K. (1976) Biochim. Biophys. Acta 427, 497-511. Keller, R. M. , & Wuthrich, K. (1980) Biochim. Biophys. Acta 621, 204-217. Kendrew, J. C , Bodo, G., Dintzis, H. M. , Parrish, R. G., Wycoff, H., & Phillips, D. C. (1958) Nature 181, 662-666. Kendrew, J. C , Dickerson, R. E., Strandberg, B. E., Hart, R. G., Davies, D. R., Phillips, D. C , & Shore, V. C. (1960) Nature 185, 422-427. King, N. K , & Winfield, M. E. (1963) J. Biol. Chem. 238, 1520-1528. Kirk, T. K., & Farrell, R. L. (1987) Annu. Rev. Microbiol. 41, 465-505. Konishi, Y., & Feng, R. (1994) Biochemistry 33, 9706-9711. Krishnamoorthi, R., & La Mar, G. N. (1984) Eur. J. Biochem. 138, 135-140. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. USA 82, 488-492. Kuroki, R., Taniyama, y., Seko, C , Nakamura, H., Kikuchi, M. , & Ikehara, M. (1989) Proc. Natl. Acad. Sci. USA 86, 6903-6907. Kusters-van Someron, M. , Kishi, K., Lundell, T., & Gold, M . H. (1995) Biochemistry 34, 10620-10627. Kuwahara, M. , Glenn, J. K , Morgan, M. A., & Gold, M . H. (1984) FEBS Lett. 169, 247-250. La Mar, G. N., Budd, D. L., Sick, H., & Gersonde, K. (1978a) Biochim. Biophys. Acta 537, 270-283. La Mar, G. N., Budd, D. L., Viscio, D. B., Smith, K. M. , & Langry, K. C. (1978b) Proc. Natl. Acad. Sci. USA 75, 5755-5759. La Mar, G. N., Budd, D. L., Smith, K. M., & Langry, K. C. (1980) J. Am. Chem. Soc. 102, 1822-1827. La Mar, G. N., Burns, P. D., Jackson, J. T., Smith, K. M. , Langry, K. C , & Strittmatter, P. (1981) J. Biol. Chem. 256, 6075-6079. 138 La Mar, G. N., Davis, N. L., Parish, K. W., & Smith, K. M . (1983) J. Moi. Biol. 168, 887-896. La Mar, G. N., Toi, H., & Krishnamoorthi, R. (1984) J. Am. Chem. Soc. 106, 6395-6401. La Mar, G. N., Yamamoto, Y. , Jue, T., Smith, K. M. , & Pandey, R. K. (1985) Biochemistry 24, 3826-3831. La Mar, G. N., Emerson, S. D., Lecomte, J. T. L, Pande, U., Smith, K. M. , Craig, G. W., & Kehres, L. A. (1986) J . Am. Chem. Soc. 108, 5568-5573. La Mar, G. N., Pande, U., Hauksson, J. B., Pandey, R. K., Smith, K. M . (1989) J. Am. Chem. Soc. Ill, 485-491. La Mar, G. N., Davis, N. L., Johnson, R. D., Smith, W. S., Hauksson, J. B., Budd, D. L., Dalichow, F., Langry, K. C , Morris, I. K., & Smith, K. M. (1993) J. Am. Chem. Soc. 115, 3869-3876. Larson, E. J., & Pacoraro, V. L. (1992) Introduction to Manganese Enzymes, V C H Publishers, Inc., New York. Lee, K.-B., La Mar, G. N., Kehres, L. A., Fujimari, E. M. , Smith, K. M. , Pochapsky, T. C , & Sligar, S. G. (1990) Biochemistry 29, 9623-9631. Lee, K.-B., Jun, E., La Mar, G. N., Rezzano, I. N., Pandey, R. K., Smith, K. M. , Walker, F. A., & Buttlaire, D. H. (1991a) J. Am. Chem. Soc. 113, 3576-3583. Lee, K.-B., La Mar, G. N., Pandey, R. K., Rezzano, I. N., Mansfield, K. E., & Smith, K. M. (1991b) Biochemistry 30, 1878-1887. Lecomte, J. T. J., & La Mar, G. N. (1985) Biochemistry 24, 7388-7395. Levy, M . J., La Mar, G. N., Jue, T., Smith, K. M. , Pandey, R. K., Smith, W. S., Livingston, D. J., & Brown, W. D. (1985) J. Biol. Chem. 260, 13694-13698. L i , Y.-T., Hsieh, Y . -L . , Henion, J. D., & Ganem, B. (1993)/. Am. Soc. Mass Spectrom. 4, 631-637. L i , H., & Poulos, T. L. (1994) Structure 15, 461-464. L i , T., Quillin, M . L., Phillips Jr., G. N., & Olson, J. S. (1994) Biochemistry 33, 1433-1446. Light-Wahl, K. J., Schwartz, B. L., & Smith, R. D. (1994)7. Am. Chem. Soc. 116, 5271-52278. Lim, A. R. (1990) in Biochemistry, University of British Columbia, Vancouver, BC. 139 Lim, H.-K., Hsieh, Y. L., Ganem, B., & Henion, J. (1995) J. Mass Spectrom. 30, 708-714. Linse, S., Jonsson, B., & Chazin, W. J. (1995) Proc. Natl. Acad. Sci. USA 92, 4748-4752. Lloyd, E., Ferrer, J. C , Funk, W. D., Mauk, M. R., & Mauk, A. G. (1994) Biochemistry 33, 11432-11437. Lloyd, E., & Mauk, A. G. (1994) FEBS Lett. 340, 281-286. Lloyd, E., Burke, D. L., Ferrer, J. C , Maurus, R., Doran, J., Carey, P. R., Brayer, G. D., & Mauk, A. G. (1996) Biochemistry 35, 11901-11912. Loo, J. A. (1995) Bioconjugate Chem. 6, 644-665. Mann, M. , & Wilm, M . (1995) TIBS 20, 219-224. MarzlufF, E. M. , Campbell, S., Rodgers, M. T., & Beauchamp, J. L. (1994)7. Am. Chem. Soc. 116, 6947-6948. Mathews, F. S., Argos, P., & Levine, M. (1971a) Cold Spring Harbor Symp. Quant. Biol. 36, 387-395. Mathews, F. S., Levine, M. , & Argos, P. (1971b) Nature 223, 15-16. Matthews, D. J. (1995) Curr. Opin. Biotech. 6, 419-424. Mauk, M . R., Barker, P. D., & Mauk, A. G. (1991) Biochemistry 30, 9873-9881. Mauk, M. R., Ferrer, J. C , & Mauk, A. G. (1994) Biochemistry 33, 12609-12614. Maurus, R., Bogumil, R., Luo, Y., Tang, H-L., Smith, M. , Mauk, A. G., Brayer, G. D. (1994) J . Biol. Chem. 269, 12606-12610. Mayo, S. L., Ellis, W. R., Crutchley, R. J., & Gray, H. B. (1986) Science 233, 948-952. McCoy, S., & Coughey, W. S. (1970) Biochemistry 9, 2387-2393. McLachlan, S. J., La Mar, G. N., & Sletten, E. (1986)7. Am. Chem. Soc. 108, 1285-1291. McPhalen, C. A., Strynadka, N. C. J., & James, M. N. G. (\99\) Adv. Prot. Chem. 422, ll-XAA. Meng, C. K., Mann, M. , & Fenn, J. B. (1988) Zeitschrift fur Physik D: Atoms, Molecules and Clusters 361-368. Millis, C. D., Cai, D., Stankovich, M. T., & Tien, M. (1989) Biochemistry 28, 8484-8489. 140 Moon Woody, A. -Y. , Eaton, S. S., Osumi-Davis, P. A., & Woody, R. W. (1996) Biochemistry 35, 144-152. Nicholls, A., Sharp, K., & Honig, B. (1991) Proteins: Struc. Fund. Genet. 11, 281-296. Nugent, J. H. A. (1996) Eur. J. Biochem. 237, 519-531. O'Halloran, T. V. (1993) Science 261, 715-725. Ortiz de Montellano, P. R., & Catalano, C. E. (1985) 7 Biol. Chem. 260, 9265-9271. Ozols, J., & Strittmatter, P. (1969) J. Biol. Chem. 244, 6617-6618. Pajot, P. (1971) Bioc. Biophys. Res. Comm. 45, 887-892. Pantoliano, M. W., Whitlow, M. , Wood, J. F., Rollence, M. L., Finzel, B. C , Gilliland, G. L., Poulos, T. L., & Bryan, P. N. (1988) Biochemistry 27, 8311-8317. Parraga, G., Horvath, S. J., Eisen, A., Taylor, W. E., Hood, L., Young, E. T., & Klevit, R. E. (1988) Science 241, 1489-1492. Paszczynski, A., Huynh, V. -B. , & Crawford, R. (1986) Arch. Biochem. Biophys. 244, 750-765. Paul, J., Smith, M. L., & Paul, K. G. (1991) Biochim. Biophys. Acta 1079, 330-334. Perutz, M. F., & Matthews, F. S. (1966)7 Moi. Biol. 21, 199-202. Phillips, S. E. V. (1978) Nature 273, 247-248. Pin, S., Alpert, B., Cortes, R., Ascone, I., Chiu, M. L., & Sligar, S. G. (1994) Biochemistry 33, 11618-11623. Pladziewicz, J. R., Meyer, T. J., Broomhead, J. A., & Taube, H. (1973) Inorg. Chem 12, 639. Poulos, T. L , & Kraut, J. (1980)7 Biol. Chem. 255, 8199-8205. Poulos, T. L., & Finzel, B. C. (1984) in Peptide and Protein Reviews (Hearn, M. T. W., Ed.) pp 115-171, Marcel Dekker, Inc., New York. Poulos, T. L. (1988) in Advances in Inorganic Biochemistry: Heme Proteins (Eichborn, G. L., & Marzzilli, L. G., Eds.) pp 1-36, Elsevier, New York. Pyle, A. M . (1993) Science 261, 709-714. Quillin, M . L., Arduini, R. M. , Olson, R. S., & Phillips Jr., G. N. (1993) 7. Moi. Biol. 234, 140-155. 141 Rajarathnam, K., La Mar, G. N., Chiu, M. L., Sligar, S. G., Singh, J. P., & Smith, K. M. (1991) 7. Am. Chem. Soc. 113, 7886-7892. Rao, S. I., Wilks, A., Hamberg, M. , & Ortiz de Montellano, P. R. (1994) 7. Biol. Chem. 269, 7210-7216. Regan, L. (1993) Annu. Rev. Biophys. Biomol. Struct. 22, 257-81. Reid, L. S., & Mauk, A. G. (1982)7 Am. Chem. Soc. 104, 841-845. Reid, L. S., Taniguchi, V. T., Gray, H. B., & Mauk, A. G. (1982)7. Am. Chem. Soc. 104, 7516-7519. Reid, L. S., Mauk, M. R., & Mauk, A. G. (1984)7. Am. Chem. Soc. 106, 2182-2185. Reid, L. S., Lim, A. R., & Mauk, A. G. (1986)7. Am. Chem. Soc. 108, 8197-8201. Rivera, M. , Wells, M. A., & Walker, F. A. (1994) Biochemistry 33, 2161-2170. Roberts, V. A., Iverson, B. L., Iverson, S. A., Benkovic, S. J., Lerner, R. A., GetzofF, E. D., & Tainer, J. A. (1990) Proc. Natl. Acad. Sci. USA 87, 6654-6658. Rodgers, K. K., & Sligar, S. G. (1991)7. Am. Chem. Soc. 113, 9419-9421. Rohlfs, R. J., Mathews, A. J., Carver, T. E., Olson, J. S., Springer, B. A., Egeberg, K. D., & Sligar, S. G. (1990)7. Biol. Chem. 265, 3168-3176. Rossi-Fanelli, A., & Antonini, E. (1960)7. Biol. Chem. 235, PC 4-PC 5. Safo, M . K., Gupta, G. P., Walker, F. A., & Scheidt, W. R. (1991)7. Am. Chem. Soc. 113, 5497-5510. Santucci, R., Mintorovitch, J., Constantinidis, I., Satterlee, J. D., & Ascoli, F. (1988) Biochim. Biophys. Acta 953, 201 -204. Santucci, R., Ascoli, F., La Mar, G. N., Pandey, R. K., & Smith, K. M . (1993) Biochim. Biophys. Acta 1164, 133-137. Sarkanen, K. V. (1971) mLignins: Occurence, Formation, Structure, and Reactions (Sarkanen, K. V., & Ludwig, C. H., Eds.) pp 95-195, Wiley-Interscience, New York. Satterlee, J. D., Erman, J. E., La Mar, G. N., Smith, K. M. , Langry, K. C. (1983) Biochim. Biophys. Acta, 743, 246-255. Satterlee, J. D., Alam, S. L., Mauro, J. M. , Erman, J. E., & Poulos, T. L. (1994) Eur. J. Biochem. 224, 81-87. 142 Schnier, P. D., Gross, D. S., Price, W. D., Zhao, Y., Williams, E. R. (1996) in 44 ASMS Conference on MS and Allied Topics, Portland, OR. Scholes, C. P., Liu, Y , Fishel, L. A., Farnum, F. M. , Mauro, J. M. , & Kraut, J. (1989) Isr. J. Chem. 29, 85-92. Shack, J., & Clark, W. M. (1947)7. Biol. Chem. 171, 143-187. Shahin, M. M . (1966) J. Chem. Phys. 45, 2600. Shire, S. J., Hanania, G. I. H., & Gurd, F. R. N. (1974) Biochemistry 13, 2967-2974. Shiro, Y. , Iizuki, T., Marubayashi, K., Ogura, T., Kitagawa, T., Balasubramanian, S., & Boxer, S. G. (1994) Biochemistry 33, 14985-14992. Silva, K. E., Elgren, T. E., Que Jr., L., & Stankovich, M. T. (1995) Biochemistry 34, 14093-14103. Sinclair, R., Hallam, S., Chen, M. Chance, B., & Powers, L. (1996) Biochemistry 35, 15120-15128. Singh, H. K., & Wilson, M. T. (1990) Bioc. Soc. Trans. 18, 1272-1273. Sivaraja, M. , Goodin, D. B., Smith, M. , & Hoffman, B. M. (1989) Science 245, 738-740. Sligar, S. G. & Gunsalus, I. C. (1976) Proc. Nat. Acad. Sci. USA 73, 1078-1082. Smerdon, S. J., Dodson, G. G., Wilkinson, A. J., Gibson, Q. H., Blackmore, R. S., Carver, T. E., & Olson, J. S. (1991) Biochemistry 30, 6252-6260. Smerdon, S. J., Krzywda, S., Wilkinson, A. J., Brantley, R. E., Carver, T. E., Hargrove, M . S., & Olson, J. S. (1993) Biochemistry 32, 5132-5138. Smerdon, S. J., Krzywda, S., Brzozowski, A. M. , Davies, G. J., Wilkinson, A. J., Brancaccio, A., Cutruzzola, F., Allocatelli, C. T., Brunori, M. , L i , T., Brantley Jr., R. E., Carver, T. E., Eich, R. F., Singleton, E., & Olson, J. S. (1995) Biochemistry 34, 8715-8725. Smith, D. L., & Zhang, Z. (1994) Mass Spec. Rev. 13, 411-429. Smith, M . C , Furman, T. C , Ingolia, T. D., & Pidgeon, C. (1988) J. Biol. Chem. 263, 7211-7215. Smith, M . L., Ohlsson, P. I., & Paul, K. G. (1982) in Cytochrome P450: Biochemistry, Biophysics and Environmental Implications (Hielanem, E., Laitenen, M. , & Hanninen, O., Eds.) pp 601-605, Elsevier Biomedical, Amsterdam. 143 Smith, M . L., Paul, J., Ohlsson, P. I., Hjortsberg, K., & Paul, K. G. (1991) Proc. Natl. Acad. Sci. USA 88, 882-886. Smith, R. D., Loo, J. A., Loo, R. R. O., Busman, M. , & Udseth, H. R. (1991) Mass Spectrom. Rev. 10, 359-452. Smith, R. D., & Light-Wahl, K. J. (1993) Biol. Mass Spec. 22, 493. Springer, B. A., Egeberg, K. D., Sligar, S. G., Rohlfs, R. J., Mathews, A. J., & Olson, J. S. (1989)7 Biol. Chem. 264, 3057-3060. Springer, B. A., Sligar, S. G., Olson, J. S., & Phillips Jr., G. N. (1994) Chem. Rev. 94, 699-714. Soltis, S. M. , & Strouse, C. E. (1988) J. Am. Chem. Soc. 110, 2824-2829. Stellwagen, E. (1978) Nature 275,13-1'4. Suh, S.-S., Haymore, B. L., & Arnold, F. H. (1991) Protein Eng. 4, 301-305. Sundaramoorthy, M. , Kishi, K., Gold, M. FL, & Poulos, T. L. (1994)7. Biol. Chem. 269, 32759-32767. Tainer, J. A., Roberts, V. A., & GetzofF, E. D. (1991) Curr. Opin. Biotech. 2, 582-591. Takano, T. J. (1977a) 7. Mol. Biol. 110, 569-584. Takano, T. J. (1977b) 7. Mol. Biol. 110, 537-568. Tang, H-L., Chance, B., Mauk, A. G., Powers, L., Reddy, K. S., Smith, M. (1994a) Biochim. Biophys. Acta 1206, 90-96. Tang, X.-J . , Brewer, C. F., Saha, S., Chernushevich, I., Ens, W., & Standing, K. G. (1994b) Rapid Commun. Mass Spec. 8, 750-754. Teale, F. W. J. (1959) Biochim. Biophys. Acta 35, 543. Therien, M. J., Chang, J., Raphael, A. L., Bowler, B. E., & Gray, H. B. (1991) Structure and Bonding 75, 109-129. Tomoda, T., Takizawa, T., Tsuji, A., & Yoneyama, Y. (1981) Biochem. J. 193, 181-185. Todd, R. J., Van Dam, M. E., Casimiro, D., Haymore, B. L., & Arnold, F. H. (1991) Proteins 10, 156-161. Unger, S. W., Lecomte, J. T. J., & La Mar, G. N. (1985) 7. Mag. Res. 64, 521-526. 144 Valaskovic, G. A., Kelleher, N. L., Little, D. P., Aaserud, D. L, & McLafferty, F. W. (1996) Anal. Chem. 67, 3802. van Berkel, G. J., Zhou, F., & Aronson, J. T. (1996) Int. J. Mass Spec. Ion Proc. in press. Van Dyke, B. R., Bakan, D. A , Glover, K. A. M. , Hegenauer, J. C , Saltman, P., Springer, B. A., & Sligar, S. G. (1992) Proc. Natl. Acad. Sci. USA 89, 8016-8019. Vitello, L. B., Erman, J. E., Miller, M . A., Wang, L, & Kraut, J. (1993) Biochemistry 32, 9807-9818. Walker, F. A., Huynh, B. H., Scheidt, W. R., & Osvath, S. R. (1986) J. Am. Chem. Soc. 108, 5288-5297. Walker, R. A., Emrick, D., Rivera, J. E., Hanquet, B. J., & Buttlaire, D. H. (1988) J. Am. Chem. Soc. 110, 6234-6240. Walker, B., Kasianowicz, J., Krishnasastry, M. , & Bayley, H. (1994) Protein Eng. 7, 655-662. Wang, J , Mauro, M. , Edwards, S. L., Oatley, S. G., Fishel, L. A., Ashford, V. A., Xuong, N., & Kraut, J. (1990) Biochemistry 29, 7160-7173. Wariishi, H., Akileswaran, L., & Gold, M. H. (1988) Biochemistry 27, 5365-5370. Wariishi, H., Dunford, H. B., MacDonald, I. D., & Gold, M . H. (1989) J . Biol. Chem. 264, 3335-3340. Wariishi, H., Valli, K., & Gold, M. H. (1992) J . Biol. Chem. 267, 23688-23695. Wilks, A. & Ortiz de Montellano, P. R. (1992) J. Biol. Chem. 267, 8827-8833. Willett, W. S., Gillmor, S. A., Perona, J. J., Fletterick, R. J., & Craik, C. S. (1995) Biochemistry 34, 2172-2180. Williams, E. R., Schnier, P. D., Price, W. D., Gross, D. S., Zhao, Y., & Jockush, R. A. (1996) in 44 ASMS Conference on MS and Allied Topics, Portland, OR. Wilm, M . S., & Mann, M. (1994) Int. J. Mass Spectrom. Ion Proc. 135, 167. Wilm, M. S., & Mann, M. (1996) Anal. Chem. 68, 1-8. Wolynes, P. G. (1995) Proc. Natl. Acad. Sci. USA 92, 2426-2427. Yamamoto, Y., & La Mar, G. N. (1986) Biochemistry 25, 5288-5297. Yamamoto, Y., Nanai, N., Chujo, R., & Suzuki, T. (1990) FEBSLett. 264, 113-116. 145 Yamashita, M . M. , Wesson, L., Eisenman, G., & Eisenberg, D . (1990) Proc. Natl. Acad. Sci. USA 87, 5648-5652. Yee, S., & Peyton, D . H. (1991) FEBS lett. 290, 119-122. Yellen, G., Sodickson, D . , Chen, T.-Y., & Jurman, M. E. (1994) Biophys. J. 66, 1068-1075. Yonetani, T., & Schleyer, H. (1967)7. Biol. Chem. 242, 1974-1979. Zoller, M . J., & Smith, S. (1987)Meth. Enzymol. 154, 329-350. 146 VI. A P P E N D I X A . Map of the p G Y M vector. Figure 36. The Mb gene, containing a portion of the ribosome binding site of phage T7 gene 10 leader sequence located directly upstream, was inserted into the multiple cloning site of the pEMBL 18+ vector. The multiple cloning site is located within a short coding region (lacZ) for the a-peptide of P-galactosidase. This vector contains a selectable ampicillin resistance marker (AmpR), along with the origin (ori) required for replication of double stranded D N A in E. coli. The intergenic region of phage FI (FI) contains all the genes required for viral D N A replication and morphogenesis of viral particles upon superinfection with phage FI. 147 B. Construction of the optically-transparent thin layer electrode (OTTLE) cell. Reference electrode Counter electrode Gold electrode Figure 37. Schematic diagram of the optically-transparent thin layer electrode (OTTLE) cell with exploded view of the observation cell. 148 C. Extinction coefficients of the myoglobin variants. Table 17. Molar extinction coefficients of the met-aquo derivatives of the Mb variants (25 °C) determined by the pyridine hemochromagen method (de Duve, 1948). Variant pH Xmas (nm) e (mM 1 cm'1) Lys45Glu 7.0 409 162 Lys45Glu/Lys63Glu 7.0 409 159 Lys45Glu/Lys63Ser 7.0 409 163 Lys45Glu/Lys63/Asp60Glu 7.0 409 159 His97Leu 7.0 408 165 His97Leu/Lys45Glua 6.0 408 176 His97Leu/Lys45Glu/Lys63Glua 6.0 408 169 Ser92Asp/Lys45Glua 6.0 409 174 Lys45Glu/Lys63 Glu/His97Leu/Ser92Ala 7.0 407.5 162 His63Tyr/Val68Phe 7.0 412 95 "measured by L. Eltis. 149 D. Absorption maxima of the myoglobin variants. Table 18. Absorption spectra of the various derivatives of the Mb variants (0.1 M sodium phosphate buffer, pH 6.0 at 25 °C). Variant met CN deoxy CO Wild-type 409 423 433.5 423 504 540.5 557 541 631.5 577.5 K45E 409 422 434 423 503.5 541 556 541 632 576.5 K45E/K63E 409 422 433.5 423.5 502 540 558 542 630.5 575.5 K45E/K63S 409 422.5 434 423.5 503.5 540.5 556 541.5 631.5 574 K45E/K63E/D60E 409 422 434 423.5 503.5 540 556.5 541 632 575.7 H97L 408 422 434.5 422.5 503 541 556.5 541 633 575 H97L/K45E 408 422 434.5 423 502.5 543 556.5 541 632.5 574 H97L/K45E/K63E 408 422 434.5 423 504 541.5 556.5 541.5 633.5 574 S92D/K45E 408.5 422 433.5 422.5 504 540 557.5 540.5 633.5 574.5 K45E/K63E/H97L/S92A 407.5 420.5 433 422.5 503 541 556.5 540 632.5 571.5 H64Y/V68F (pH 7.0) 412 422 436 425.5 486 541.5 554.5 542.5 541.5 575 599.5 150 £ . Assignments of heme resonances of the myoglobin variants. Table 19. Assignments of the heme methyl hyperfine shifts and the mean heme methyl shift, s(M L 3 - S g ) , for the met-aquo derivatives of the Mb variants in deuterated sodium phosphate buffer (pH 7.0, 20 °C). Protein 8 -CH 3 5-CH3 3-CH3 I-CH3 s(M,.3.S8) wild-type 92.94 85.08 73.05 52.72 75.9 K45E 93.35 86.27 72.95 52.27 76.2 K45E/K63E 93.04 86.05 72.66 52.01 75.9 K45E/K63E/D60E 92.93 85.89 72.38 51.82 75.8 H97L 90.20 81.42 72.58 52.44 74.2 H97L/K45E 90.47 81.44 72.15 51.62 73.9 H97L/K45E/K63E 90.48 81.40 72.05 51.54 73.9 . S92D/K45E 92.87 89.28 72.20 52.57 76.7 K45E/K63E/H97L/S92A 91.48 86.71 71.81 53.31 75.8 Table 20. Assignments of the single proton hyperfine shifts for the met-aquo derivatives of the Mb variants in deuterated sodium phosphate buffer (pH 7.0, 20 °C). Protein 7-a 6-a 6-a' 4-a 7-a' 2-a wild-type 75.40 59.43 45.11 45.11 31.38 31.38 K45E 75.72 59.99 47.67 45.81 32.85 31.12 K45E/K63E 75.80 60.43 46.82 45.64 32.76 30.78 K45E/K63E/D60E 75.58 60.19 46.54 45.80 32.62 30.78 H97L 65.01 58.22 H97L/K45E 65.87 57.14 H97L/K45E/K63E 66.82 57.63 S92D/K45E 60.02 60.02 K45E/K63E/H97L/S92A 56.70 58.76 151 Table 21. Assignments of the heme resonances in the met-CN derivatives of the Mb variants (0.1 sodium phosphate buffer in 10% D 2 0 , pH 8.0, 20 °C). Protein 5 -CH 3 His64 N e H His93 N j H 1 C H 3 8-CH3 De99 C Y H wild-type 27.49 23.54 21.27 18.41 13.28 -10.13 K45E 27.94 - 21.20 18.43 13.36 -10.14 K45E/K63E 27.95 - 21.21 18.46 13.55 -10.14 K45E/K63E/D60E 27.90 - 21.19 18.44 13.54 -10.10 H97L 27.23 23.40 20.29 19.46 13.92 -8.88 H97L/K45E 27.80 - 20.06 19.44 14.36 -8.58 H97L/K45E/K63E 27.70 - 20.06 19.44 14.53 -8.55 S92D/K45E 25.27 - 21.15 16.46 12.79 -8.36 K45E/K63E/H97L/S92A 24.51 - 21.04 15.93 12.47 -6.89 152 F. Metal-ligand distances of Mn(II) binding sites in Mb and MnP. Table 22. The manganese ligands and bond distances for Mb and MnP from the Mn(II)-containing three-dimensional structures. Ligand Distance (A) Lys45Glu/Lys63Glu Mb Glu45 side chain carboxyl 2.35 Glu45 side chain carboxyl 2.86 Heme propionate-6 2.52 H i s l l 3 a 2.45 Water 2.25 Manganese peroxidase, MnP b Heme propionate-6 2.34 Glu35 side chain carboxyl 2.69 Glu39 side chain carboxyl 2.82 Asp 179 side chain carboxyl 2.57 Water 520 2.35 Water 441 2.57 a. the Hisl 13 interaction comes from an adjacent molecule in the crystalline lattice and is expected to be a water molecule in solution. b. Sundaramoorthy et al., (1994). 153 G. Calculation of equilibrium Mn(IT) binding constant for ferrous Lys45Glu/Lys63Glu Mb. K D (ox) Mb»Fe Mb»Fe3+-Mn 2+ E M (free) E M (bound) Mb-Fe 2+ Mb-Fe2+-Mn K D (red) 2+ Figure 38. The thermodynamic cycle of Mn(II) binding to the Lys45Glu/Lys63Glu variant in the reduced and oxidized states consists of the equilibrium Mn(II) binding constants of the oxidized (KD(ox)) and reduced species (K (red)) and the reduction potentials of the Mn(II) bound ( E x p o u n d ) ) and unbound (EM(free)) forms. Reduction potentials (E^) and equilibrium Mn(II) binding constants (K D ) are related to free energy (AG) through the equations AG=-nFE and AG=RT In K D . Due to the reversibility of redox and binding equilibria, the free energy changes around this cycle must equal zero (Clark, 1960; Sligar & Gunsalus, 1976; Silva et al., 1995). From these relations, the following equation arises (10): The reduction potentials of the Lys45Glu/Lys63Glu variant (pH 7.0,1=0.2 M , 25 °C) in the presence and absence of Mn(II) were determined to be 40 and 9 mV, respectively (Results section C4). The equilibrium binding constant of Mn(II) to this variant in the oxidized form was determined with the EPR titration technique (Results section A l ) to be 1.8 x 10 - 3 M (pH 7.0,1=0.1 M , 25 °C). From these values, the equilibrium binding constant of Mn(II) to the reduced form of this variant was calculated to be 5.5 * 10"4 M. Therefore, the Lys45Glu/Lys63Glu variant has a ~3-fold greater affinity for Mn(II) in the reduced form than in the oxidized form. Clearly, this calculation is (10) 154 approximate insofar as the EPR and electrochemical measurements were performed at different ionic strengths. To ensure that all MnfTJ) binding sites on the variant were occupied in the electrochemical experiments, a high concentration of metal ion was required. However, measurement of Mn(II) binding to Mb at an equivalent ionic strength would be difficult as metal binding would be very low. Nevertheless, it is useful to note that this approach can in principle provide information that is otherwise difficult to obtain directly. 155 

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.831.1-0088272/manifest

Comment

Related Items