Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The nature and role of oxidation state dependent conformational differences in cytochrome c Berghuis, Albert M. 1993

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

Item Metadata

Download

Media
831-ubc_1993_spring_phd_berghuis_albert.pdf [ 8.42MB ]
Metadata
JSON: 831-1.0086349.json
JSON-LD: 831-1.0086349-ld.json
RDF/XML (Pretty): 831-1.0086349-rdf.xml
RDF/JSON: 831-1.0086349-rdf.json
Turtle: 831-1.0086349-turtle.txt
N-Triples: 831-1.0086349-rdf-ntriples.txt
Original Record: 831-1.0086349-source.json
Full Text
831-1.0086349-fulltext.txt
Citation
831-1.0086349.ris

Full Text

THE NATURE AND ROLE OF OXIDATION STATE DEPENDENTCONFORMATIONAL DIFFERENCES IN CYTOCHROME CByAlbert Marinus BerghuisB. Sc. (Chemistry) Rijks Universiteit Groningen, The Netherlands, 1983M. Sc. (Chemistry) Rijks Universiteit Groningen, The Netherlands, 1987A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESTHE DEPARTMENT OF BIOCHEMISTRYWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril 1993© Albert Marinus Berghuis, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of Biochemistry The University of British ColumbiaVancouver, CanadaDate^April 19 1993DE-6 (2/88)AbstractThe objective of the work described in this thesis was to study the nature and role of conforma-tional differences between the oxidation states of cytochrome c. Using x-ray crystallographictechniques, the oxidized form of yeast iso-l-cytochrome c was solved and compared to the pre-viously determined reduced state. The following differences between the oxidation states wereidentified. Three segments of polypeptide chain, located for the most part on the Met80 sideof the protein, were shown to display an increase in mobility in the oxidized state. A conservedinternal water molecule, Wat166, was observed to shift 1.7 A towards the heme iron atom andreorient its dipole moment in the oxidized state. As part of this movement several hydrogenbonds were broken including the interaction between Tyr67 OH and the Met80 SD heme ligand.Finally, differences between the two oxidation states were also observed for the conformationof the pyrrole A propionate and its associated hydrogen bond network, the distortion of theporphyrin ring plane, and the orientation of the imidazole plane of the His18 ligand.In order to assess the function of the observed conformational differences between the twooxidation states of cytochrome c, the three dimensional structures of five mutants (N52A, N521,Y67F, N52I-Y67F and 175M) were determined, of which three were completed in both oxidationstates. Correlation of wild-type and variant protein structures with functional studies suggestedthat Wat166 was a central feature in oxidation state dependent differences, and three roles forthis water molecule could be identified. First, the oxidation state dependent positioning andorientation of Wat166 appears to be particularly important for modulating the interactionbetween Tyr67 OH and Met80 SD. This hydrogen bond was shown to influence the electronwithdrawing power of the Met80 ligand and therefore is a factor in controlling the midpointreduction potential of cytochrome c. Secondly, the presence of Wat166 is necessary to maintainthe spatial and hydrogen bonding relationships between residues in this region of the protein.iiFinally, Wat166 also appears to mediate the oxidation state dependent flexibility of selectedpolypeptide chain segments. The biological function of this phenomenon is still unclear, butour results suggests that it might play a role in interactions between cytochrome c and its redoxpartners.In conclusion, the work described in this thesis gives insight into the structure-functionrelationships in cytochrome c and provides a basis for future studies aimed at understandingthe mechanism of electron transfer carried out by this protein.iiiTable of ContentsAbstract^ iiTable of Contents^ ivList of Tables ixList of Figures^ xiList of Abbreviations^ xvAcknowledgments xvii1 Introduction 11.1 The Biological Role of Eukaryotic Cytochromes c ^ 11.2 The Structure of Eukaryotic Cytochromes c 31.2.1^Amino acid sequences ^ 31.2.2^Three dimensional structures 41.3 Oxidation State Dependent Conformational Differences in Cytochrome c ^ 101.4 The Midpoint Reduction Potential of Cytochromes ^ 141.4.1^Introduction ^ 141.4.2^Theories for the control of midpoint reduction potential ^ 141.5 Cytochrome c Mutant Studies ^ 171.5.1^Introduction ^ 171.5 .2^Water-switch mutants ^ 181.6 Thesis Objectives ^ 21iv2 General Overview of Experimental Methods 232.1 Crystallization ^ 232.1.1^The hanging drop method ^ 232.1.2^The hair-seeding technique 242.1.3^Changing oxidation states of crystals ^ 252.2 Data Collection and Data Processing 252.2.1^Methods of data collection ^ 252.2.2^Processing of diffraction intensity data ^ 262.2.3^Scaling of data sets ^ 312.3 Structure Refinement 312.3.1^Construction of the starting model ^ 312.3.2^Reciprocal-space refinement 332.3.3^Manual interventions ^ 332.3.4^Accuracy of Structures 343 Oxidized Yeast Iso-1-Cytochrome c 363.1 Experimental Procedures ^ 363.1.1^Crystallization 363.1.2^Data collection and data processing ^ 373.1.3^Refinement and analyses ^ 373.2 Results 403.2.1^Polypeptide chain conformation ^ 403.2.2^Heme structure ^ 453.2.3^Internal water structure ^ 493.3 Discussion 553.3.1^Focal points for oxidation state dependent structural alterations ^ 553.3.2^Mechanistic implications ^ 644 Reduced and Oxidized Yeast Iso-1-Cytochrome c Y67F Mutant 684.1 Experimental Procedures ^ 684.1.1^Crystallization 684.1.2^Data collection and data processing ^ 684.1.3^Refinement and analyses ^ 694.2 Results 724.2.1^Polypeptide chain conformation^ 724.2.2^Heme structure ^ 764.2.3^Mutation site: the reduced state ^ 824.2.4^Mutation site: the oxidized state 854.3 Discussion 864.3.1^Conformational effects of the Y67F mutation ^ 864.3.2^Effect of the Y67F mutation on midpoint reduction potential ^ 904.3.3^Role of tyrosine 67 in eukaryotic cytochromes c ^ 925 Yeast Iso-l-Cytochrome c N52I and N52I-Y67F Mutants 945.1 Experimental Procedures ^ 945.1.1^Crystallization 945.1.2^Data collection and data processing ^ 955.1.3^Refinement and analyses ^ 965.2 Results 995.2.1^Polypeptide chain conformation ^ 995.2.2^Heme structure ^ 1035.2.3^Mutation site region 1065.3 Discussion 1085.3.1^Structural effects of mutations ^ 1085.3.2^Importance of the hydrogen bond network around Wat166 ^ 112vi6 Reduced Yeast Iso-1-Cytochrome c N52A and 175M Mutants^1166.1 Experimental Procedures ^  1166.1.1 Crystallization  1166.1.2 Data collection and data processing ^  1176.1.3 Refinement and analyses ^  1176.2 Results ^  1206.2.1 Polypeptide chain conformation ^  1206.2.2 Heme structure ^  1236.2.3 The N52A mutation site ^  1246.2.4 The I75M mutation site  1286.3 Discussion ^  1296.3.1 Plasticity of the pyrrole A propionate region ^  1296.3.2 Functional role of the internal water molecule Wat166 ^ 132Summary^ 136Bibliography^ 140Appendices^ 151A Determination of the Absolute Scale Factor^ 151A.1 Introduction ^  151A.2 Initial Estimate for the Absolute Scale Factor ^  151A.2.1 Linear rescale ^  151A.2.2 Least-squares rescale  152A.2.3 Wilson plot method ^  152A.3 Refinement of the Absolute Scale Factor ^  154B Theory of Crystallographic Refinement^ 157B.1 Introduction ^  157viiB.2 Reciprocal-space Refinement ^  158B.2.1 The non-linearity of the equations ^  159B.2.2 The observations to parameters ratio  161B.2.3 The size of the matrix ^  163C Estimating Coordinate Errors in Macromolecular Structures^166C.1 Introduction ^  166C.2 Empirical error estimates ^  167C.3 Luzzati error estimates  168C.4 Cruickshank error estimates ^  169viiiList of Tables1.1 Sequence alignment for yeast iso-1, yeast iso-2, tuna, horse and rice cytochromes c 51.2 Eukaryotic cytochrome c structures determined by x-ray crystallography ^ 61.3 Secondary structural elements present in yeast iso-l-cytochrome c ^ 101.4 Properties of water-switch mutants ^  201.5 Thermodynamic properties for the midpoint reduction potential of wild-type andmutant yeast iso-1-cytochromes c ^  212.6 Typical stereochemistry for refined structures of yeast iso-l-cytochrome c . . . ^ 353.7 Final stereochemistry of oxidized yeast iso-l-cytochrome c at 1.9 A resolution . . 383.8 Heme conformation and ligand geometry in the two oxidation states of yeastiso-l-cytochrome c   463.9 Heme solvent accessibility in the two oxidation states of yeast iso-l-cytochrome c 483.10 Structural changes observed on going from the reduced to the oxidized state inyeast iso-1-cytochrome c ^  564.11 Final stereochemistry for reduced and oxidized Y67F yeast iso-l-cytochrome c^704.12 Heme geometry of wild-type and Y67F mutant proteins ^ 774.13 Heme solvent accessibility of wild-type and Y67F mutant proteins ^ 794.14 Heme propionate hydrogen bond interactions in wild-type and Y67F mutantyeast iso-l-cytochrome c   814.15 Hydrogen bond interactions at the mutation site in the wild-type and Y67F proteins 844.16 Structural differences observed in Y67F mutant structures when compared totheir wild-type yeast iso-l-cytochrome c counterparts ^  87ix5.17 Data collection statistics for the N52I and N52I-Y67F yeast iso-l-cytochromes c 955.18 Refinement results and stereochemistry for the final models of the yeast iso-1-cytochrome c N52I and N52I-Y67F mutants in both oxidation states ^ 975.19 Average deviations of polypeptide chain atoms in the N52I and N52I-Y67F mu-tants with respect to reduced wild-type yeast iso-l-cytochrome c ^ 1005.20 Heme conformation and ligand geometry in wild-type, N52I, Y67F and N52I-Y67F mutant yeast iso-l-cytochromes c ^  1045.21 Heme propionate hydrogen bond interactions in wild-type, N52I, Y67F and N52I-Y67F mutant yeast iso-l-cytochromes c ^  1055.22 Structural differences observed in the N52I and N52I-Y67F mutant structureswhen compared to wild-type yeast iso-l-cytochrome c ^  1096.23 Final stereochemistry for N52I and I75M yeast iso-l-cytochromes c ^ 1186.24 Heme conformation and ligand geometry in wild-type, N52A and I75M yeastiso-1-cytochromes c  1246.25 Heme solvent accessibility for wild-type, N52A and I75M yeast iso-l-cytochromes c 1256.26 Heme propionate hydrogen bond interactions in wild-type, N52A and I75M yeastiso-l-cytochromes c ^  1276.27 Structural differences observed in the reduced N52A and I75M mutant structureswhen compared to reduced wild-type yeast iso-l-cytochrome c^ 130B.28 Typical numbers of stereochemical restraints of different classes used in the re-finement of yeast iso-l-cytochrome c structures ^  164xList of Figures1.1 Physiological redox partners of cytochrome c ^  21.2 Heme atom labelling convention ^  31.3 Positional deviations between eukaryotic cytochromes c ^  71.4 Schematic representation of yeast iso-l-cytochrome c  81.5 Structure of reduced yeast iso-1-cytochrome c ^  91.6 Space-filled representation of yeast iso-l-cytochrome c ^  111.7 Oxidation state dependent changes in tuna cytochrome c  121.8 Oxidation state dependent changes in cytochrome c ^  131.9 Mutation site area for "water-switch" mutants  192.10 Crystallization setup using the hanging drop method ^  232.11 Graphical representation of the completeness of diffraction data ^ 262.12 Completeness of diffraction data as a function of sigma cutoff  272.13 Absorption curve from a typical cytochrome c crystal ^  292.14 Decay profile during data collection for a typical cytochrome c crystal ^ 292.15 Fourier maps of a mutant cytochrome c structure ^  322.16 Statistics for the refinement of a mutant yeast iso-l-cytochrome c structure^343.17 Luzzati plot of oxidized yeast iso-l-cytochrome c ^  393.18 Structure of oxidized yeast iso-l-cytochrome c  403.19 Positional deviations between reduced and oxidized yeast iso-l-cytochrome c 413.20 Region around Arg13 - G1y84 in the two oxidation states of yeast iso-1-cytochrome c 423.21 Area around Trp59 in reduced and oxidized yeast iso-l-cytochrome c   42xi3.22 Comparison of the thermal factors between reduced and oxidized yeast iso-l-cytochrome c ^  433.23 Matrix comparison of the thermal factors between the two oxidation states ofyeast iso-l-cytochrome c ^  443.24 Space-filled representation of the oxidation state dependent flexible regions inyeast iso-l-cytochrome c ^  473.25 Region about the pyrrole A propionate in reduced and oxidized yeast iso-l-cyto-chrome c ^  493.26 Region about the internal water molecule Wat166 in the two oxidation states ofyeast is o-1- cyt ochrome c ^  503.27 The internal cavity occupied by Wat166 in reduced and oxidized yeast iso-l-cyto-chrome c ^  513.28 Oxidation state dependent dipole orientation of Wat166 in reduced and oxidizedyeast iso-1 and oxidized horse cytochrome c ^  533.29 Position of the internal water molecule Wat166 in eukaryotic cytochromes c . . ^ 603.30 Region about the internal water molecule Wat166 in the oxidized structures ofyeast iso-1 and horse cytochrome c ^  613.31 Push-Button mechanism for electron transfer in yeast iso-l-cytochrome c^664.32 Luzzati plot of reduced and oxidized yeast iso-l-cytochrome c Y67F mutantstructures ^  714.33 Structure of the reduced yeast iso-l-cytochrome c Y67F mutant protein ^ 724.34 Positional deviations between wild-type and Y67F mutant proteins ^ 734.35 Comparison of the thermal factors between wild-type and Y67F mutant proteins 754.36 Region about the pyrrole A propionate in wild-type and Y67F mutant proteins . 804.37 Region about the mutation site in wild-type and Y67F mutant yeast iso-l-cyto-chrome c ^  83xii5.38 Luzzati plots of reduced and oxidized yeast iso-l-cytochrome c N52I and N52I-Y67F mutant structures ^  985.39 a-carbon tracings of the reduced and oxidized structures of wild-type, N52I,Y67F and N52I-Y67F yeast iso-1-cytochromes c   995.40 Positional deviations between wild-type and N52I and N52I-Y67F mutant proteins1015.41 Comparison of the thermal factors between wild-type, N52I and N52I-Y67F mu-tant proteins ^  1025.42 Area around the pyrrole A propionate in wild-type and N52I-Y67F mutant yeastiso-l-cytochrome c in the reduced state ^  1065.43 Region about the mutation site in wild-type yeast iso-1-cytochrome c and N52I,Y67F and N52I-Y67F mutants ^  1076.44 Luzzati plots of N52A and I75M yeast iso-l-cytochrome c mutant structures . . . 1196.45 a-carbon tracings of the wild-type, N52A and I75M mutant yeast iso-l-cyto-chrome c structures ^  1206.46 Positional deviations of N52A and I75M mutant proteins with wild-type yeastiso-l-cytochrome c ^  1216.47 Comparison of the thermal factors between wild-type, N52A and I75M mutantproteins ^  1226.48 Region about the mutation site in the N52A mutant yeast iso-l-cytochrome c ^ 1256.49 Area around the pyrrole A propionate in N52A mutant protein ^ 126^6.50 Region about the mutation site in I75M mutant yeast iso-l-cytochrome c   1286.51 Area around the pyrrole A propionate in the I75M mutant protein ^ 1296.52 Water-switch mechanism for stabilization of the oxidation states of yeast iso-l-cytochrome c ^  133A.53 Wilson plot for the reduced N52I-Y67F mutant data set ^ 154A.54 Modified Wilson plot for the reduced N52I-Y67F mutant data set ^ 155A.55 Comparison of structure factor amplitudes ^  156B.56 Overdeterminancy for the refinement of yeast iso-l-cytochrome c ^ 162B.57 Sparseness of the matrix used in refinement ^  165C.58 Luzzati plot for estimating coordinate errors ^  169C.59 Cruickshank error estimates for different atom types in oxidized yeast iso-l-cyto-chrome c ^  171xivList of AbbreviationsB ^ Thermal factorCCP  Cytochrome c peroxidaseCD ^ Circular dichroismDTT  Dithiothreitol^ Midpoint reduction potentialEXAFS ^ Extended x-ray absorption fine structureF0,   Observed and calculated structure factorsI ^ Intensity of a reflection Absolute scale factorNMR ^ Nuclear magnetic resonanceSHE  Standard hydrogen electrodeUV ^ Ultraviolet lighta, b, c ^ Crystallographic unit cell axes, or axis lengthse.u.  entropy units; 1 e.u.^1 cal mor lfo ^  Atomic scattering factorh, k, 1  ^Miller indicesT.M.S. ^  Root mean squaredx, y, z ^ Positional parametersa ^ PhaseA  Wavelength of the radiation used in the diffraction experiment.All the experiments described in this thesis are done with CuKa radiation, A = 1.54184 Axv^ One of the rotational axes which define the orientation of thecrystal in the diffraction experiment. This angle is varied toobtain an estimate of the absorption effect when using a diffrac-tometer for data collectionSo ^ Generic symbol for a mathematical functionCis ^  Standard deviation of x or estimated standard deviation of quan-tity x9 ^ Diffraction angleA ^ Angstrom (0.1 nm)The conventions of the IUPAC-IUB Combined Commissions on Biochemical Nomenclature arefollowed for both three letter and one letter abbreviations for amino acids [J. Biol. Chem. 241,527-533 (1966); J. Biol. Chem. 243, 3557-3559 (1968)]; for designating atoms and for describ-ing the conformational torsional angles of the polypeptide chain [J. Biol. Chem. 245, 6489-6497(1970)]. Designations for atoms of the protoheme IX group are according to the BrookhavenNational Laboratory Protein Data Bank (Bernstein et al., 1977; see also Figure 1.2).The amino acid numbering scheme used for the yeast iso-l-cytochrome c wild-type andmutant structures described in this thesis is based on an alignment to the sequences of vertebratecytochromes c. The N-terminal residue is numbered -5 and the C-terminal residue is numbered103 (see Table 1.1). The numbering of solvent molecules for these structures is identical tothat of the reduced yeast iso-l-cytochrome c wild-type protein (Louie & Brayer, 1990; ProteinData Bank entry lYCC). Water molecules not found in reduced yeast iso-l-cytochrome c havebeen assigned numbers starting with 300.xviAcknowledgmentsI would like to take this opportunity to thank all the people who contributed to the workdescribed in this thesis. A few people should be mentioned here specifically. First and foremostis my supervisor Gary Brayer who encouraged my scientific curiosity as well as my independenceas a researcher during the last five years. Connie Leung, Teresa Yang, Teresa Hii and YuoguangLuo were invaluable for their technical assistance. My fellow grad-students, Gordon Louie,Michael Murphy, Terry Lo and David Burk were a constant source of information and a soundingboard for ideas. Specifically, I would like to thank Mike, who not only was the driving forcebehind the development of some of the graphics software used in this thesis (Preras3d/Raster3dpackage), but also helped develop some of the ideas expressed in Appendix A. Moreover webecame close friends in the process. I also like to express my gratitude to our collaboratorsGrant Mauk and Michael Smith here at U.B.C., George McLendon and Fred Sherman at theUniversity of Rochester, and Arieh Warshel at U.S.C., who were very generous in providingprotein samples and sharing their results. This acknowledgment page would be incompletewithout mentioning Alfred Gartner, Guy Guillemette and Steve Rafferty, without whom thecontents of this thesis would be quite different. I would also like to acknowledge the two othermembers of my supervisory committee Phil Bragg and Ian Clark-Lewis. Finally, I would liketo thank my wife Lida, without whose support I might have not entered graduate school, andwould surely have not finished it.xviiChapter 1Introduction1.1 The Biological Role of Eukaryotic Cytochromes cEukaryotic cytochromes c are small ( ,,,13,000 Daltons) soluble proteins which are found in theinter-membrane space of mitochondria. They form part of the respiratory chain and primar-ily shuttle electrons between the membrane bound cytochrome c reductase and cytochrome coxidase complexes (see Figure 1.1). In order to accomplish this task cytochrome c possessesa covalently attached heme as a prosthetic group (see Figure 1.2). Positioned in the center ofthe heme group is an iron atom. By varying the valence state of the heme iron atom between2+ and 3+, the reduced and oxidized form respectively, cytochrome c can donate and acceptelectrons and thus function as an electron transport protein.Cytochrome c[Fe(III)] + e — 4=L Cytochrome c[Fe(II)]oxidized^ reducedCytochrome c reductase and cytochrome c oxidase are not the only redox partners forcytochrome c (Figure 1.1). In yeast, cytochrome c peroxidase (CCP) and cytochrome b 2 havebeen identified as accepting electrons from or donating electrons to cytochrome c ( Altschulet al., 1940; Abrams et al., 1942; Bach et al., 1942a,b). Cytochrome c also functions as anelectron acceptor from cytochrome b5 and sulfite oxidase in animal systems ( McLeod et al.,1961; Cohen & Fridovich, 1971; Ito, 1980a,b; Lederer et al., 1983). The cytochrome c/CCPand cytochrome c/cytochrome b 5 redox couples deserve special mention since they have servedas model systems for studying biological electron transfer (McLendon & Miller, 1985; Cokic &Erman, 1987; Liang et al., 1987; Mauk et al., 1991). Studies of these two complexes have notonly focussed on the thermodynamics and kinetics of biological electron transfer, but have also1Chapter 1. Introduction^ 2Outer^Inter-membrane space^Innermembrane membraneFigure 1.1: A schematic representation of the mitochondrial inter-membrane space showingthe location of cytochrome c and its physiological redox partners. The arrows in the diagramindicate the direction of the flow of electrons. The abbreviations used are: Cyt c - cytochrome c;Cyt b2 - (flavo)cytochrome b 2 ; Cyt b5 - cytochrome b 5 ; Sul Ox - sulfite oxidase; CCP - cyto-chrome c peroxidase; Cyt c Reductase - cytochrome c reductase (complex III); Cyt c Oxidase -cytochrome c oxidase (complex IV). The interaction between cytochrome c and cytochrome b2and CCP is only observed in yeast. The interaction between cytochrome c and sulfite oxidaseand cytochrome b 5 is only observed in animal systems.explored the structural aspects of complex formation (Salemme, 1976; Poulos & Kraut, 1980;Lum et al., 1987; Wendoloski et al., 1987; Pelletier & Kraut, 1992).Chapter 1. Introduction^ 3Figure 1.2: A schematic representation of the atomic skeleton of the protoheme IX group ofcytochrome c and the atom and pyrrole ring labelling convention used herein (Bernstein et al.,1977). In eukaryotic cytochromes c the porphyrin ring is covalently attached to the proteinthrough two thioether linkages (Cysl4 SG - heme CAB; Cys17 SG - heme CAC). The ironatom is six coordinated with His18 NE2 and Met80 SD providing the remaining ligands. Inthe view presented, the solvent exposed heme edge observed in cytochromes c is located at theright, the histidine ligand is behind the porphyrin ring and the methionine ligand is positionedin front.1.2 The Structure of Eukaryotic Cytochromes c1.2.1 Amino acid sequencesCytochromes c from eukaryotes have long been the subject of extensive study. As a consequence,a large number of amino acid sequences are presently known. This wealth of information hasChapter I. Introduction^ 4been the basis for phylogenetic studies as well as functional studies (Dayhoff et al., 1976;Dickerson & Timkovich, 1975; Amati et al., 1988; Moore & Pettigrew, 1990). Moore andPettigrew (1990) compiled sequences from 96 eukaryotic species. Examination of this datashowed that within the eukaryotes the amino acid sequence is highly conserved, the lengthof the polypeptide chain varies between 103 and 113 residues and alignment of the sequencesfrom 94 species can be accomplished without creating any gaps (the two remaining sequencesare from Euglina gracilis and Tetrahymena pyrzformis). In these 94 sequences, 27 residues areinvariant and an additional 16 amino acids are conserved in at least 90 of the sequences (seeTable 1.1).In many cases the reason for the high degree of sequence homology in eukaryotic cyto-chromes c is puzzling. Although several residues are critical for maintaining the fold andfunction of cytochrome c, a large number of invariant residues can be mutated without abol-ishing biological activity (Hampsey et al., 1986). Furthermore, by mutating the conservedresidues 52 and 67, variants of cytochrome c have been made which appear not to be affectedin their function but which are actually more stable than the wild-type protein (Das et al.,1989; McLendon et al., 1991; Hickey et al., 1991; Luntz et al., 1989). This has not only raisedquestions regarding the specific role of these conserved residues but also instigated a debate onthe evolutionary development of cytochrome c (Margoliash, 1990).1.2.2 Three dimensional structuresThe high degree of homology seen in sequence alignments is reflected in the tertiary structuresof the cytochromes c. To date a total of six structures, two of which are in both oxidation states,have been determined by x-ray crystallographic techniques (see Table 1.2). Analysis of the fivehigh resolution structures available shows that the polypeptide fold is virtually identical (seeFigure 1.3). Pair-wise comparison gives a maximum average deviation for main-chain atomsof only 0.57 A (between horse and yeast iso-2 cytochrome c, sequence identity is 58% ). Athorough analysis of the subtle differences in conformation between the different eukaryoticChapter I. Introduction^ 5Table 1.1: Sequence alignment for yeast iso-i, yeast iso-2, tuna, horse and rice cytochromes c-9^1^10^ 20• •^• o^• • • o^0^0^•Iso-1 - - - - TEFKAGSAKKGATLFKTRTLQCHTVEKGGPHYVIso-2 AKESTGFKPGS AKKGATLFKTRCQQCHT I EEGGPNKV GTuna ^ GDVAKGKKTFVQKCAQCHTVENGGKHKV GHorse GDVEKGKK I FVQKCAQCHTVEKGGKHKT GRice - ASFSEAPPGNPKAGEK I FKTKCAQCHTVDKGAGHKQ G30• o • • o • 40^ 50• 0^• 0 • 60•Iso-1 PNL H I FGR HS QAE YS Y TD AN IKKNVL DENNM S EIso-2 PNL H G I FGR HS G QVK G YS Y TD AN I NKNVK w DEDSMS ETuna PNL w G L FGR KT G QAE G YS Y TD AN KSKGIV w NNDT L MEHorse PNL H G L FGR KT G QAP G FT Y TD AN KNKGIT w KEET L MERice PNL N G L FGR QS G TT P G YS Y ST AN KDMAV I w EENT L YD70 80 90^100•• 0•••00..0•• • •^• oo o oIso-1 Y L T NPJ KYIPGTKM ATGGLKKEKDINDL I TYL KKACE -Iso-2 Y L T NPJ KYIPGTKM AFAGLKKEKDRNDL I TYMTKAAK -Tuna Y L E NPKKYIPGTKM I FAGIKKKGERQDLVAYL K SATS -Horse Y L E NPKKYIPGTKM I FAG IKKKTEREDL I AY L KKATNERice Y L L NPJ KYIPGTKM VFPGLKJ PQERADL I SYL KEATS -The sequences of yeast iso-1 (Smith et al., 1979), yeast iso-2 (Montgomery et al., 1980), tuna(Kreil, 1965), horse (Margoliash et al., 1961) and rice (Mori & Morita, 1980) cytochromes chave been aligned so as to maximize the structural homology present. The single letter code isused to identify the amino acids and the residue numbering is based on the sequence of tunacytochrome c. The amidation states of residues 52 and 54 in rice are the opposite of that givenby Mori and Morita (1980). The present amide assignment fits the chemical data as well as theoriginal assignment and is more consistent with other sequences (Moore & Pettigrew, 1990).Note the single-letter code J is used to denote E-N-trimethyl lysines. Those amino acid residuesidentical in all five protein sequences are enclosed in boxes. Residues which are observed to beinvariant in 94 eukaryotic sequences are marked with a bullet (•), those which are conserved inat least 90 sequences are marked with a circle (o).cytochrome c structures can be found in Brayer and Murphy (1993).Given the high degree of similarity between structures, reduced yeast iso-l-cytochrome cwill be used here as the archetypal cytochrome c for describing structural conformations. Thisstructure has been determined to the highest resolution and forms the basis of the structuralChapter I. Introduction^ 6Table 1.2: Eukaryotic cytochrome c structures determined by x-ray crystallographySourceOxidationstateResolution(A)R-factor(%)PDBcode t ReferenceYeast iso-1 reduced 1.2 19.2 lYCC (Louie & Brayer, 1990)Yeast iso-1 oxidized 1.9 19.7 2YCC (Berghuis & Brayer, 1992)Yeast iso-2 reduced 1.9 19.0 1YEA (Murphy et al., 1992)Tuna reduced 1.5 15.9 5CYT (Takano & Dickerson, 1981a)Tuna oxidized 1.8 20.8 3CYT (Takano & Dickerson, 1981b)Horse oxidized 1.9 17 (Bushnell et al., 1990)Rice oxidized 1.5 19 1CCR (Ochi et al., 1983)Bonito § reduced 2.3 1 CYC (Tanaka et al., 1975)Brookhaven protein data bank (PDB) codes (Bernstein et al., 1977) are given to specify thecoordinates used for comparison studies. The horse cytochrome c coordinates were obtained asthe result of other studies ongoing in this laboratory.t The oxidized yeast iso-1 structure is included in this table for completeness but will not bediscussed until Chapter 3.§ The coordinates available for the bonito structure have not undergone crystallographic re-finement. Therefore bonito cytochrome c is not used in further comparisons.work described in this thesis.Yeast iso-l-cytochrome c is a globular protein roughly 35 x 35 x 30 A in size, consisting offive a-helixes and a number of /3-loops and turns which wrap the polypeptide chain around theprotoheme IX group (see Figures 1.4 and 1.5, and Table 1.3). The porphyrin ring is covalentlyattached to the polypeptide chain through two thioether linkages with cysteines 14 and 17. Thetwo residues His18 and Met80 form the fifth and sixth ligands to the heme iron atom.The heme group is almost completely buried inside the protein matrix leaving less than10% of its surface exposed to solvent (see Figure 1.6). The two heme propionates, each ofwhich probably carries a negative charge (Moore, 1983) are also buried inside the protein.The propionate carboxyl groups form extensive hydrogen bond interactions with proton donorgroups of nearby polypeptide chain and with two internal water molecules. Examination of theother cytochrome c structures and the available amino acid sequences suggests that most ofChapter 1. Introduction^ 710^20^30^40^50^60^70^80^90^100Residue NumberFigure 1.3: A plot of the overall average deviations of main-chain atoms of yeast iso-2, tuna,horse and rice cytochromes c from those of yeast iso-l-cytochrome c along the course of thepolypeptide chain. Only residues 1 to 103, which are common to all five cytochromes c, arerepresented. The vertical bars represent the range of individual pairwise average deviations ofmain-chain atoms between yeast iso-1 and the other cytochromes c. Structures with pairwisedeviations both larger than 0.7 A and 30% greater than the overall average deviations of allstructures are labelled at the particular residues involved (H = horse, R = rice, T = tuna andY = yeast iso-2). (Figure reproduced with permission of M.E.P. Murphy; Brayer & Murphy(1993))these hydrogen bonds are conserved in all eukaryotic cytochromes c.Located around the exposed heme edge is a "ring" of positively charged residues which arefunctionally conserved in most eukaryotic cytochromes c (see Figure 1.6). Chemical modifica-tion studies have implicated these residues as being involved in complexation with complemen-tary negatively charged residues on the surfaces of redox partners (for a review see Margoliash& Bosshard, 1983). This has led to the proposal that the region around the exposed heme edgeis the binding site of cytochrome c (Dickerson & Timkovich, 1975). Model-building studies ofcytochrome c with cytochrome b 5 as well as the recent structure of the complex of cytochrome cwith CCP reaffirm this hypothesis (Salemme, 1976; Pelletier & Kraut, 1992).Chapter 1. Introduction^ 8Figure 1.4: A schematic representation of the structure of yeast iso-1-cytochrome c in the stan-dard orientation used herein, i.e. heme group is viewed edge on, propionates point downwardsand the His18 and Met80 ligands are located right and left of the porphyrin ring, respectively.To highlight the cytochrome c fold, a-helixes are represented as cylinders and the remainingpeptide chain as a ribbon. Also shown are the two cysteine residues Cys14 and Cys17 whichare covalently attached to the heme group.Further comparisons between the five high resolution cytochrome c structures availablehave identified four conserved water molecules (Brayer & Murphy, 1993). Two of these are ofparticular interest since they are located inside the protein matrix. The first is Wat121 which ispositioned adjacent to Arg38 and the propionate A carboxyl group. It forms hydrogen bonds toboth groups as well as to backbone atoms of residues 39 and 42. The second conserved internalwater molecule (Wat166) is located in a cavity on the left side of the cytochrome c molecule inthe standard orientation (see Figure 1.4). It forms three hydrogen bonds with the side-chainsof the highly conserved residues Asn52, Tyr67 and Thr78 (see Table 1.1). Structural studieshave implicated this water molecule in the function of cytochrome c (Takano & Dickerson,Chapter 1. Introduction^ 9Figure 1.5: A stereo-drawing of the conformations of all main-chain (thick lines) and side-chain(thin lines) atoms in reduced yeast iso-1-cytochrome c in (a) the standard view and (b) a viewwhich is rotated 90°. The heme group has also been drawn with thick lines. Also drawn arethe two heme ligand bonds to His18 and Met80, and the two covalent thioether linkages tocysteines 14 and 17. For clarity, every 5th a-carbon atom has been labelled with its one-letteramino acid designation and sequence number.1981b; Bushnell et al., 1990). Other studies suggest, however, that Wat166 is not essential formaintaining the function of cytochrome c (McLendon et al., 1991), and possibly serves only todestabilize the protein (Luntz et al., 1989).Chapter 1. Introduction^ 10Table 1.3: Secondary structural elements present in yeast iso-l-cytochrome cSecondary structure Residueselement^involveda-Helix^2-14/3-Turn (type II)^21-247-Turn^27-29/3-Turn (type II)^32-35/3-Turn (type II)^35-38/3-Turn (type II)^43-46^(distortedt)a-Helix^49-55^(residue 55 is distorted)a-Helix 60-70a-Helix^70-75/3-Turn (type II)^75-78a-Helix^87-102t Mediated through a water molecule.1.3 Oxidation State Dependent Conformational Differences in Cytochrome cDespite all that is now known about the structure of eukaryotic cytochromes c, a fundamen-tal controversy still remains regarding the nature and role of oxidation state conformationalchanges. On the one hand, a number of techniques give strong indications that the proteinundergoes a dramatic alteration in its conformation upon oxidation. For example, studiessuggest that the oxidized state has a significantly increased adiabatic compressibility (Edenet al., 1982). In addition, the radius of gyration as observed by small-angle X-ray scattering islarger for the oxidized protein (Trewhella et al., 1988). The oxidized state of cytochrome c isalso more susceptible to proteolytic digestion (Nozaki et al., 1958) and thermal and chemicaldenaturation (Butt & Keilin, 1962; Dickerson & Timkovich, 1975). An increase in hydrogenexchange rates for the oxidized state of cytochrome c has been measured as well (Ulmer & Kagi,1968; Liu et al., 1989). Second derivative amide I infrared spectroscopy also suggests that thereare differences in the secondary structure between the oxidation states of cytochrome c (Donget al., 1992). And finally, a differential chemical modification technique (Bosshard & Zurrer,Chapter 1. Introduction^ 11Figure 1.6: A space-filled representation of the structure of yeast iso-l-cytochrome c in thestandard orientation. Heme group atoms are drawn in black with the solvent exposed edgefacing the viewer. Dark grey atoms are associated with positively charged lysine or arginineresidues that encircle the exposed heme edge and are believed to be involved in complexationwith redox partners.1980) and one-dimensional NMR studies (Moore, 1983; Williams et al., 1985b) provide fur-ther indications that the reduced and oxidized forms of cytochrome c have distinctly differentconformations.On the other hand, examination of the three-dimensional structures of eukaryotic cyto-chromes c in differing oxidation states as revealed by x-ray crystallography does not show largedifferences between the two forms (Takano & Dickerson, 1981a,b; Bushnell et al., 1990). Forthe reduced and oxidized states of tuna cytochrome c, only a few small changes in conformationand a rearrangement of hydrogen bonds near the site of a conserved internal water moleculeChapter 1. Introduction^ 126718^'.^H ..H' ,'  OH^N\^..H H "NH^/FINI\0^^52 J ^ H\41Figure 1.7: A schematic representation of the oxidation state dependent changes in tuna cyto-chrome c as described by Takano and Dickerson (1980,1981b). The reduced state of the proteinis shown in thick lines, the oxidized form is displayed in thin lines, and hydrogen bonds arerepresented by dashed lines. Distances from the internal water molecule (equivalent to Wat166in yeast iso-l-cytochrome c) to the heme iron atom are also indicated on the figure. The distancefor the oxidized state is an average for the two molecules in the crystallographic asymmetricunit.(equivalent to Wat166 in yeast iso-l-cytochrome c) were noted (see Figure 1.7). Further, noloss of hydrogen bonds upon oxidation was observed, nor was a change in the main or side-chain flexibility between the oxidation states observed (Takano & Dickerson, 1980, 1981b).Essentially the same results are obtained when comparing cytochrome c structures from differentspecies (Bushnell et al., 1990). In this context it should be mentioned that there has beensome confusion about the exact assignments of hydrogen bonds around the internal water site.Originally it was considered that residues 41 and 52 hydrogen bond to each other, but basedon geometric constraints this cannot be the case. Actually, both G1y41 NH and Asn52 ND2form hydrogen bonds to the nearby pyrrole A propionate group. Therefore, Figure 1.8 whichis based on structures of five eukaryotic cytochromes c represents a more correct description of`O■3078OH,NH6.96718.OH.^4%.4 ■'H...0 NH /H /NH0pyrrole Apropionate/^/^H/ OHI. • #0.^. 17.005278Chapter I. Introduction^ 13Figure 1.8: A schematic representation of the oxidation state dependent changes found in allfive eukaryotic cytochromes c whose structures are known as described by Bushnell et al. (1990).As in Figure 1.7, the reduced state of the protein is plotted in thick lines, the oxidized formis drawn in thin lines, and hydrogen bonds are represented by dashed lines. Distances fromthe internal water molecule to the heme iron atom are average values based on the availablehigh resolution structures. The discrepancies with the description of Takano and Dickerson (seeFigure 1.7) are due to a more careful analysis of the hydrogen bonds made by Asn 52.the oxidation state dependent conformational differences than Figure 1.7.To rationalize the dramatically different experimental results obtained by x-ray crystallog-raphy and other biophysical techniques it has been suggested that crystal contacts and high saltconcentrations might mask the conformational differences between oxidation states (Williamset al., 1985b; Trewhella et al., 1988; Liu et al., 1989; Dong et al., 1992). Recent two-dimensionalNMR experiments, however, disagree with this view and conclude that the solution structureis very similar to that seen by x-ray crystallography (Wand et al., 1986; Feng et al., 1990;Feng & Englander, 1990; Gao et al., 1991). Thus the question of the nature of oxidation stateChapter 1. Introduction^ 14dependent conformational differences remains to be resolved.1.4 The Midpoint Reduction Potential of Cytochromes1.4.1 IntroductionThe driving force for biological electron transfer is the difference in the midpoint reductionpotentials of the reacting proteins. Among the eukaryotic cytochromes c the value of thisproperty appears to be extremely conserved. Although the precise value of reduction potentialsis difficult to compare due to their sensitivity to experimental conditions, it appears that thevariation in Em for eukaryotic cytochromes c is on the order of only 10-20 mV (Pettigrew &Moore, 1987; McLendon et al., 1991). For example, at 25°C, pH 7.0 and p=0.01 M, the midpointreduction potentials for such diverse species as yeast iso-1, tuna and turkey cytochromes c are261, 265 and 260 respectively (Margalit Schejter, 1973).The small range of midpoint reduction potentials observed suggests that maintenance of thisvalue is critical for the functioning of cytochrome c. It is to be expected that if the reductionpotential is raised or lowered by a certain amount the driving force will be too small or large forcontrolled and efficient electron transfer to occur between cytochrome c and its redox partners.However, in vivo studies on a yeast mutant cytochrome c which has an Em that is 50 mV lowerbut does not inhibit growth has raised questions regarding this assumption (McLendon et al.,1991).1.4.2 Theories for the control of midpoint reduction potentialBecause of the importance of the midpoint reduction potential to biological electron transfer,it is not surprising that much research has been done to understand the factors that determinethese values. A large number of these studies have focused on the many types of cytochromesthat have thus far been isolated. The reduction midpoint potentials of these span a range ofnearly 800 mV (Churg Warshel, 1986). Marchon et al. (1982) showed that a variation of,150 mV can be attributed to different axial ligands to the heme iron atom. Substitutions toChapter 1. Introduction^ 15the porphyrin ring such as the thioether linkages in c-type cytochromes can cause an alterationin the midpoint potential of an additional ,150 mV (Margoliash Schejter, 1966; Mashikoet al., 1981; Marchon et al., 1982). However, a variation of nearly 500 mV can still be observedwhen both the heme ligands and porphyrin ring substitutions are kept constant (Cusanovichet al., 1988). An explanation for this latter observation must be sought in the differencesimposed on the heme by the surrounding protein matrix.Polarity of the heme environmentKassner showed that the midpoint reduction potential of iron porphyrin ring groups is highlydependent on the dielectric constant of the solution the experiment is carried out in. Heproposed that the reduction potential of cytochromes may, therefore, be controlled by thepolarity of the heme environment (Kassner, 1972; Kassner, 1973). An examination of sevenheme containing proteins by Stellwagen (1978), however, revealed that while the midpointpotential varies over a range of 300 mV for these proteins the polarity of the heme environmentis nearly constant. It was observed that the amount of heme solvent exposure did correlatereasonably well with the variation in midpoint potential. Stellwagen concluded that hemesolvent exposure rather than the polarity of the heme environment determines the value of themidpoint reduction potential. It should be mentioned that in this study the heme groups ofthe seven proteins had different axial ligands and varying substituents to the porphyrin ringsuggesting caution should be used in the interpretation of these results.Axial ligand effectsBased on differences in the chemical shift of the methionine ligand in a number of His/Metligated cytochromes c, Moore and Williams (1977) proposed that the midpoint potential canbe altered by regulating the iron-sulfur ligand bond length. In this view the conformation of theprotein constrains the position of the methionine ligand so as to result in a specific midpointreduction potential. For example, shortening of the iron-sulfur ligand bond by ,0.1 A shouldChapter 1. Introduction^ 16result in a drop of 400 mV for the reduction potential. However, it is unlikely that this is themechanism used by nature to modulate midpoint reduction potential. EXAFS studies haveshown that iron-sulfur bond lengths in a variety of cytochromes c prove to be independent ofthe midpoint potential (Korzun et al., 1982).Another possible mechanism for influencing the midpoint potential is variation of the orien-tation of the ligand histidine imidazole plane. Korzun et al. (1982) remark that upon reductionthe added electron will occupy the non-bonding ds, or dy, orbitals of the iron. Steric interfer-ence of these orbitals with the histidine ligand imidazole plane will alter the energy levels ofthe orbitals and thus affect the midpoint potential of the protein. In tuna cytochrome c theorientation of the imidazole ring is dependent on oxidation state (Takano & Dickerson, 1981b).Electrostatic effectsIn its simplest form oxidation of cytochrome c is essentially a change in the charge of the hemeiron atom, and it is expected that electrostatic factors will play an important role in determiningreduction potential. The relationship between electrostatic properties and midpoint reductionpotential is, however, not a simple one. It is clear from a comparison of related cytochromes cthat the overall net charge of the protein does not correlate with the midpoint reduction po-tential (Moore et al., 1986; Cusanovich et al., 1988). Nonetheless, it has been shown that bychemically modifying lysine residues the reduction potential of horse cytochrome c and Eu-glina cytochrome c-552 can be altered in a computationally predictable manner (Schejter et al.,1982).Moore (1983) proposed that midpoint potential variations among homologous cytochromes care caused by differences in the heme propionate charge. One electrostatic interaction that hasspecifically been implicated in affecting midpoint potential is that between the propionate Acarboxyl group and Arg38. This salt-bridge is the only one to the heme propionates andtherefore could be important for regulating the midpoint reduction potential. Mutation ofArg38 to an alanine does in fact result in a drop of 50 mV in the reduction potential (CutlerChapter 1. Introduction^ 17et al., 1989).Conformational stabilizationThe midpoint reduction potential reflects the relative stability of the two oxidation states.Stabilization of either state by rearrangements in protein conformation will, therefore, influencethe midpoint potential. In the past this factor has been considered to play a small role inestablishing the reduction potential of eukaryotic cytochromes c (Moore et al., 1986). Based onthe oxidation state differences seen in tuna cytochrome c, computational methods estimate theconformational stabilization to be ,40 mV (Churg & Warshel, 1983; Warshel, 1983). This mightbe an underestimation since as discussed above there is conflicting data regarding the amountand nature of conformational adjustments upon oxidation. Based on the pH dependence of thereduction potential it has been estimated that the conformational stabilization may accountfor 70-110 mV (Rogers & Moore, 1988). Measurements of the reorganization energy in cyto-chrome c/cytochrome b 5 kinetic experiments arrive at similar values for the conformationalstabilization (,80 mV) (McLendon & Miller, 1985).1.5 Cytochrome c Mutant Studies1.5.1 IntroductionSite-directed mutagenesis techniques have had a profound impact on the study of eukaryoticcytochromes c. Thanks to the ability to alter specific residues, theories regarding the propertiesof cytochrome c and the role certain residues play in establishing these properties can be tested.A good example of this is the study on the role of the invariantly conserved residue Phe82.Phenylalanine 82 in yeast iso-l-cytochrome c was the first residue to be targeted for structure-function studies using site-directed mutagenesis (Pielak et al., 1985). The reason for investi-gating this residue was that it was implicated in being directly involved in the electron transferprocess with CCP (Poulos & Kraut, 1980) and cytochrome b 5 (Wendoloski et al., 1987). Dueto its location adjacent to the porphyrin ring, Phe 82 was also considered to be important inChapter 1. Introduction^ 18maintaining the midpoint reduction potential (Kassner, 1972, 1973). Studies of several mu-tants made at the 82 position confirm that this residue is important for both biological electrontransfer (Liang et al., 1987, 1988) and regulation of midpoint potential (Rafferty et al., 1990).Structural studies have also shown that the lowered midpoint potentials observed for the serineand glycine mutants can be explained in terms of a decreased hydrophobic environment and anincreased heme exposure (Louie et al., 1988b; Louie & Brayer, 1989).1.5.2 Water-switch mutantsOften mutations at certain residues raise questions rather than provide answers. Recentlyattention has focused on a group of mutants of highly conserved residues which possess someunusual properties. The location of these mutation sites adjacent to a conserved internal watermolecule that moves upon changes in oxidation state has prompted the name "water-switch"mutants for this family of variant cytochromes c (see Figure 1.9).The two most studied mutations of this class of variants are N52I and Y67F. The N52Ivariant was discovered as a second site revertant to two nonfunctional mutants (Das et al., 1989).Subsequent studies concentrated on this mutant's midpoint reduction potential (Burrows et al.,1991; Langen et al., 1992), electron transfer kinetics (Whitford et al., 1991), in vivo growth rates(McLendon et al., 1991), and thermodynamic stability (Hickey et al., 1991). The replacementof Tyr67 to a phenylalanine was made by semi-synthesis techniques more than ten years ago(Koul et al., 1979), but its unusual physiological properties were not recognized until recently.Most studies on Y67F have described the stability of this protein and its dramatically lowermidpoint reduction potential (Luntz et al., 1989; Wallace et al., 1989; Rafferty, 1992). But thisvariant also exhibits altered pH dependent behaviour and electron transfer kinetics (Barkeret al., 1991; J.G. Guillemette, private communication).Table 1.4 gives a comprehensive list of the properties of water-switch mutants and associatedreferences. A number of unique results have been obtained. For example, replacement ofan internal hydrophilic amino acid for a hydrophobic one should result in an increase of theChapter I. Introduction^ 19Figure 1.9: A drawing showing the region about an internally bound water molecule (Wat166)in reduced yeast iso-l-cytochrome c (Louie & Brayer, 1990). Mutation of residues at positions52, 67, 75 and 78 results in variants with several unusual properties. In this drawing the courseof the polypeptide chain is shown with a ribbon, heme-ligand interactions are indicated by thinwhite bonds and hydrogen bonds are shown by thin black dotted lines.midpoint reduction potential of cytochrome c (Kassner, 1972,1973). The opposite is, however,observed for all such water-switch mutants, the most striking of which is the N52I-Y67F doublemutant which has a 56 mV lower midpoint reduction potential (Table 1.5). In spite of thedramatically lower midpoint potentials of several of the variants, most of them do not appearto be affected in terms of biological activity. Also the stability of four of the water-switchmutants as measured by denaturation studies and alkaline isomerization is increased. In fact,the isoleucine replacement at position 52 is one of the most stabilizing single site mutationsmade to date (Pace, 1990).Chapter 1. Introduction^ 20Table 1.4: Properties of water-switch mutantsfMutant^Properties^ ReferenceN52A^- Ern is 30 mV lower- 4° increase in Tn,- similar reorganization energy- does not interfere in complexation- similar pKa- lower reorganization energyN52I^- Ern is 50 mV lower- 13-17° increase in Tin- similar reorganization energylower reorganization energydoes not interfere in complexationGrowth rates unaffected or even betterpKa of alkaline isomerization increases1.5 pH unitsY67F^- Ern is 40-55 mV lower- more resistant to extreme pH- stronger Fe-S bond- complicated pH curve56% biological activitybiological activity unaffectedlower reorganization energy- pKa of alkaline isomerization increases1.5 pH units(Burrows et al., 1991; Rafferty, 1992)(Hickey et al., 1991)(Whitford et al., 1991)(Whitford et al., 1991)(Rafferty, 1992)(Rafferty, 1992)(Burrows et al., 1991; McLendon et al., 1991;Langen et al., 1992)(Das et al., 1989; Hickey et al., 1991)(Whitford et al., 1991)J.G. Guillemette, private communication(Whitford et al., 1991)(McLendon et al., 1991)J.G. Guillemette, private communication(Luntz et al., 1989; Wallace et al., 1989;Frauenhoff & Scott, 1992; Rafferty, 1992)(Luntz et al., 1989; Wallace et al., 1989)(Luntz et al., 1989)(Barker et al., 1991)(Koul et al., 1979)(Margoliash, 1990)(Rafferty, 1992)J.G. Guillemette, private communicationN52I-Y67F - Em is 55 mV lower^J.G. Guillemette, private communicationlower reorganization energy^J.G. Guillemette, private communicationpKa of alkaline isomerization increases J.G. Guillemette, private communication2.5 pH unitsI75M^Em is 45 mV lower^(Rafferty, 1992)- similar reorganization energy^(Rafferty, 1992)- similar pKa of alkaline isomerization^(Rafferty, 1992)t Most of the variants discussed in the references are of yeast iso-l-cytochrome c created by site-directedmutagenesis. However, Luntz et al., (1989) studied a mutant of rat cytochrome c and Koul et al.,(1979), Wallace et al., (1989) and Frauenhoff & Scott (1992) studied Y67F mutants made by semi-synthesis techniques from horse cytochrome c.t This observation is not supported by more recent studies (Margoliash, 1990; McLendon et al., 1991).Chapter 1. Introduction^ 21Table 1.5: Thermodynamic properties for the midpoint reduction potential of wild-type andmutant yeast iso-1-cytochromes cErn f(mV)AG°(kcal/mole)AS,?,(e.u.)AH°(kcal/mole)Wild-type 290 -6.7 -9.1 -14.0N52A 257 (-33) -5.9 (+0.8) -8.0 (+1.1) -12.9 (+1.1)N52I 232 (-58) -5.3 (+1.4) -8.2 (+0.9) -12.4 (+1.6)Y67F 234 (-56) -5.4 (+1.3) -7.5 (+1.6) -12.3 (+1.7)N52I-Y67F 234 (-56) -5.4 (+1.3) -8.5 (+0.6) -12.6 (+1.4)I75M 245 (-45) -5.7 (+1.0) 11.7 (-2.6) -13.8 (+0.2)T78G 249 (-41) -5.7 (+1.0) -6.3 (+2.8) -12.3 (+1.7)This data is from S.P. Rafferty and J.G. Guillemette (University of British Columbia). Valuesgiven in brackets indicate the difference between the mutant and wild-type thermodynamicparameters.1. 25°C, pH 6.0, p=0.1 M, and SHE referenceSeveral fundamental questions are posed by these results. Why are residues surroundingthis internal water molecule so highly conserved while they do not appear to be essential forbiological activity? What is the reason for retaining the midpoint potential for eukaryotic cyto-chromes c at such a precise value? How can single (or double) site mutations have such a largeeffect on the midpoint reduction potential of cytochrome c and the stability of this protein?1.6 Thesis ObjectivesIt is clear from the above overview that, although much is known about cytochrome c, manyquestions still remain. The nature of the conformational changes that occur upon change inoxidation state and the function of these changes in the electron transfer event are particularlyunclear. The main objective of the work in this thesis is to identify the structural differencesoccurring between the two oxidation states in eukaryotic cytochromes c and evaluate the rolethey play in biological electron transfer.Chapter 1. Introduction^ 22Identification of the conformational differences between oxidation states has been accom-plished by solving the three-dimensional structure of the oxidized form of yeast iso-l-cyto-chrome c to high resolution using x-ray crystallographic techniques, and comparing this struc-ture to that of the reduced form (Louie & Brayer, 1990). Further assessment of these structuraldifferences has been done by studying mutants of yeast iso-1-cytochrome c with amino acidsaltered at one of the focal points of oxidation state dependent conformational changes, namelythe internal water molecule Wat166. In total, the structures of five mutants have been solved(N52A, N521, Y67F, N52I-Y67F and I75M). Of these, three have been studied in both oxidationstates. These results provide insight into the nature of a number of functional properties ofcytochrome c and their relationship to the structural framework on which they must operate.Chapter 2General Overview of Experimental Methods2.1 Crystallization2.1.1 The hanging drop methodA wide variety of techniques have been developed for growing protein crystals (Giege Mikol,1989; McPherson, 1990). For the growth of wild-type and mutant yeast iso-1-cytochrome ccrystals the method of vapor diffusion in hanging drops has proven very successful (Sherwood& Brayer, 1985; Brayer & Murphy, 1993).In a typical setup 5 pl of protein solution is suspended from a siliconized cover slip over1 ml of mother-liquor (Figure 2.10). The mother-liquor used is 0.1 M sodium phosphate bufferin the pH range of 5-8 which is 80-95% saturated with ammonium sulfate. For the purpose ofkeeping the protein in a specific oxidation state 10-70 mM of a reducing or oxidizing agent isadded. As reducing agents DTT and sodium dithionite have been used, whereas sodium nitriteprotein solution^cover slipmother-liquorFigure 2.10: Experimental setup for vapor diffusion in hanging drops used in crystallizing yeastiso-l-cytochrome c and mutants thereof.23Chapter 2. General Overview of Experimental Methods^ 24and potassium ferricyanide have been added to the mother-liquor to maintain an oxidizingenvironment. The solution in the hanging droplet is identical to that of the reservoir with theexception that it is less saturated with ammonium sulfate (70-85%) and there is protein present(,20 mg/ml).The ammonium sulfate concentration in the hanging drop is chosen such that cytochrome cis just soluble. The concentration of ammonium sulfate in the mother-liquor is made slightlyhigher. Over time the protein solution and mother-liquor equilibrate through vapor diffusion.The result is that the protein solution will gradually attain the same ammonium sulfate con-centration as that of the mother-liquor and the protein droplet will become supersaturated. Ifconditions are favorable nucleation sites form from which protein crystals grow.2.1.2 The hair-seeding techniqueUnfortunately spontaneous crystal formation has rarely been observed for wild-type or mutantforms of yeast iso-l-cytochrome c. This has been attributed to a lack of nucleation sites presentin the protein droplet (Louie, 1990). To overcome this deficiency a hair-seeding technique hasbeen used (Leung et al., 1989). In this method, miniscule remnants of crushed cytochrome ccrystals are inserted into the protein droplet using a hair. These remnants form the nucleationsites for crystal growth. A major advantage of using crushed yeast iso-1-cytochrome c crystalsas seeds, is that the resulting crystals grown are found to be isomorphous with those of the wild-type protein. This property not only greatly simplifies the structure solution process for mutantproteins, but also allows for careful comparisons to be made between structures without theadded complexity of differing crystal contacts. Thus, all the crystals discussed in this thesis havethe same space group as wild-type yeast iso-l-cytochrome c, namely P43 21 2, with comparablecell dimensions a = b = 36.46-36.56 A and c = 137.02-139.12 A (cell dimensions for reducedwild-type yeast iso-1-cytochrome c are a = b = 36.46 A and c = 136.86 A).Chapter 2. General Overview of Experimental Methods^ 252.1.3 Changing oxidation states of crystalsExcept for the oxidized wild-type yeast iso-1-cytochrome c structure, all crystals used for de-termining cytochrome c structures in the oxidized form were initially grown in a reducing en-vironment. To obtain crystals containing cytochrome c in the oxidized state, reduced crystalswere transferred to mother-liquor which contained potassium ferricyanide instead of a reducingagent. The transfer was done gradually, i.e. crystals were washed with a series of solutionscontaining mother-liquor with decreasing levels of reducing agent and increasing levels of theoxidizing agent. During this process crystals were observed to change color from red to deepbrown-red, signifying the oxidation of the heme iron atom from Fe 2+ to Fe3+ . After the trans-fer was completed crystals were soaked in mother-liquor containing only potassium ferricyanidefor at least one hour before mounting.2.2 Data Collection and Data Processing2.2.1 Methods of data collectionFor the structures described in this thesis two methods of data collection have been employed.Six structures were solved with data sets obtained using an Enraf-Nonius CAD4-F11 diffrac-tometer. The remaining three data sets were collected using the Rigaku R-Axis II imaging platearea detector. The reason that two data collection methods were used is historical. During mostof the period in which the work described was done the laboratory was equipped with only theCAD4-F11 diffractometer for collecting diffraction data. Recently the laboratory acquired anarea detector and thus the latest data sets were all measured using this new technology.The approach taken by the two data collection methods is substantially different. With theEnraf-Nonius diffractometer each reflection is measured individually, whereas the Rigaku areadetector measures many reflections simultaneously in a manner similar to the photographicrotation method (Arndt & Wonacott, 1977). The result is that a complete data set can becollected much more rapidly using the area detector. For example, collection of a 2.0 A dataChapter 2. General Overview of Experimental Methods^ 26Figure 2.11: Graphical representation of the completeness of the measured diffraction dataobtained from two mutant cytochrome c crystals using two different data collection tools (seealso Figure 2.12). Displayed in (a) is a view of reciprocal space showing the 100% complete2.0 A N52A data set obtained from an Enraf-Nonius CAD4-F11 diffractometer. Each pointin this figure represents a measured reflection. As shown, it is necessary to only collect is ofreciprocal space since this is the unique part for the tetragonal space group P43 21 2. Shown in(b) is an identical representation for a 2.0 A data set from an N52I-Y67F crystal collected onthe R-Axis II imaging plate area detector under unfavorable geometric conditions. While thediffractometer data set is 100% complete, the cusp region around the c* axis is absent in thedata set obtained with the area detector.set takes about two weeks on a diffractometer versus one day on the Rigaku area detector.However, due to geometrical reasons, it is not always possible to collect 100% of the availabledata with an area detector, while this is not a problem with a diffractometer (Figure 2.11).Nonetheless, by using multiple scans on the area detector at alternative settings, almost all theavailable data is usually accessible. A further advantage of area detector data is the quality ofthe data collected (as measured by counting statistics) which is far superior to that obtainedfrom a diffractometer (see Figure 2.12). This allows for the collection of data sets to higherresolution and from much smaller crystals than is possible with a diffractometer.2.2.2 Processing of diffraction intensity dataIn order to convert measured intensities into structure factors, a number corrections have to beapplied to the data set. First, they must be corrected for background scatter which is mainly10080-Na) 60-C4a5 400 20 —-C.)0.075A300.150SIN(0) /0.225100 —S 800 60a,^-a) 400 2000.075^0.150^0.225SiN(9) /3510BChapter 2. General Overview of Experimental Methods^ 27Figure 2.12: A plot of the completeness of the diffraction data obtained from two mutantcytochrome c crystals using two different data collection tools as a function of F(see alsoFigure 2.11). Displayed in (a) is the 2.0 A N52A data set obtained from an Enraf-NoniusCAD4-F11 diffractometer. Shown in (b) is a 2.0 A data set from an N52I-Y67F crystal collectedon an R-Axis II imaging plate area detector under unfavorable geometric conditions. As canbe seen, the number of reflections above 3 sigma (thick lines) is larger in the area detector dataset compared to the one obtained with the diffractometer, despite the fact that the data set isless complete overall. This is the result of the large number of multiple reflections measured bythis method and the improved counting statistics obtained as a result.caused by non-crystalline material in the beam path. A second correction is for absorption.This arises since crystals are generally not spherical in shape resulting in unequal path-lengthsthrough the crystal for different reflections. As a consequence, the amount of absorption ofx-rays by the crystal will also vary for different reflections. A further correction is for decayoccurring due to crystal degradation during x-ray exposure. The final two corrections to beapplied are those for Lorentz and polarization effects. The Lorentz correction is required becausereflection intensity is dependent on the geometry of the data collection method. The polarizationcorrection adjusts for the variation of reflection intensity with the amount of refracted x-raybeam polarization. This latter correction factor is dependent on the source of the x-ray beamas well as on the geometry of the diffraction experiment. Once these corrections have beenmade reflection intensities can be converted into structure factors by taking the square root ofthe intensity.Chapter 2. General Overview of Experimental Methods^ 28Diffractometer data processingWith a diffraction intensity data set obtained using a diffractometer, background correctionis relatively easy. The measurement of each reflection is divided into three parts: the initialand finals of the scan are taken as the background radiation; the intermediary i of the scanrepresents the intensity of the reflection. The measured background for each reflection can thusbe used to correct for the background radiation of that reflection. To improve the backgroundcorrection, background averaging can be used. In this context background averaging means thatan estimate of the background radiation for each reflection is not only based on the backgroundmeasurements of that reflection but also on the backgrounds of reflections which are close inreciprocal space. For the processing of yeast cytochrome c diffraction data it was found thatusing reflections within a radius of 0.0005-0.0010 A -1 (,6-18 neighbouring reflections) gavegood results.The absorption correction is applied using the empirical method of North et al., (1968).In this method a reflection which is aligned along the 0-axis of data collection is used. Themeasured intensity of this "0-independent" reflection varies with rotation of the 0-axis due toabsorption and this variation can be used to estimate the amount of correction required for ageneral reflection. Measurements conducted on several mutant cytochrome c crystals show thatabsorption is not only dependent on 0, but also on the resolution of the reflection (Figure 2.13).By measuring several 0-independent reflections at different 0 values an estimate can be madeof the resolution component of the absorption correction.Correction for intensity decay is performed by measuring a number of standard reflectionsrepeatedly over the course of the full exposure time. Generally the standard reflections aregrouped according to resolution so that the decay correction can be applied on this basisas well since high resolution reflections are much more sensitive to crystal decay than lowresolution reflections (Figure 2.14). For a detailed description of the applied corrections andtheir implementation in the diffractometer data processing program ICP, which was used hereinfor structure determinations, see Murphy (1993).IO 0.8 8oA^A.'0T)  0.6ceA 10,A.•2ci 0.7A • • A:•0.50.9Chapter 2. General Overview of Experimental Methods^ 290.40^45^90^135^180^225^270^315^360Azimuthal angle coFigure 2.13: A plot of a typical absorption curve for a cytochrome c crystal (N52A mutant yeastiso-l-cytochrome c). Shown are the relative transmissions as a function of the azimuthal angle for threeq independent reflections with different resolutions (0 8.7 A, ^ 3.9 A and 6. 2.2 A). Also shown arefitted curves for the three resolutions (solid line 8.7 A, dashed line 3.9 A and dotted line 2.2 A) whichwere used to correct for absorption by the method of North et al., (1968).1.1••o^oo • 0^o •o• Ss o^0on^o o • ^BO,..., •^000• .•-4'^ •5 0.8 • ° • • ....og 'Ir.....i •C4=0^24^48^72^96^120^144^168^192^216Exposure Time (hours)0.9 • • cisi 0.a)^ .:13^li. . s> • . •- li . a . en •0.7a Dlr., .• a...,^. 0^• .-..0.60.50.4• • 0 ;'?-1'...e. 4:93 A•• • •••IP■TAO 264 288 312 336Figure 2.14: A plot of the decay profile for four reflections which were periodically measured duringdiffractometer data collection for a representative cytochrome c crystal (175M mutant yeast iso-l-cyto-chrome c). For decay correction these four reflections were divided in two groups depending on theirresolution (group 1: 0 4.3 A and • 3.5 A ; group 2: ^ 3.4 A and ■ 2.8 A) to which curves werefitted (solid line - group 1, dashed line - group 2). The fitted curves were then subsequently used for aresolution dependent decay correction of all general reflections measured.Chapter 2. General Overview of Experimental Methods^ 30Area detector data processingSince the method of data collection on the Rigaku R-Axis II is essentially identical to thescreen-less oscillation or rotation method (Arndt & Wonacott, 1977), processing of collecteddata resembles the procedure described by Rossmann et al., (1979). This means that forbackground corrections the measured intensity directly surrounding a peak is used for estimatingthe background radiation for that reflection. The correction for decay is performed by scalingbetween data collection frames. Two frames are scaled with respect to each other by minimizingthe function:=Dial - Ke-Bor,i19 )2 rbhklhklIn this function la, and IL I are identical reflections present on the two frames a and b. Sincethe scale factor is of the form Ke-B ( siTA,eB )2 a resolution dependent decay correction is applied.There is no explicit absorption correction, but absorption effects are corrected for by the abovementioned inter-frame scaling and by merging of symmetry related reflections.An indicator of the quality of the merged diffraction data is the agreement between re-flections which are measured more than once (i.e. duplicates and symmetry mates). Thisagreement is generally expressed in the form of the merging R-factor:Ehk1E7=01-filikt — Lad' Rmerge =Ehk1 ELO Iihkl (2.2)In this expression h is one of n identical reflections and I is the average intensity for thesen reflections. For the three data sets collected on the Rigaku R-Axis II area detector anddiscussed in this thesis, values of 6.3-8.4% were obtained. In contrast, merging R-factors of13.4% and higher were obtained for diffractometer data. This difference in merging R-factorsis, however, not completely caused by a difference in quality of the diffraction data. For datasets collected on a diffractometer the merging R-factor is intrinsically higher due to the limitednumber of repeat measurements.(2.1)Chapter 2. General Overview of Experimental Methods^ 312.2.3 Scaling of data setsThe structure factor data obtained from a diffraction experiment is on an arbitrary scale andmust therefore be scaled such that it can be compared to calculated structure factors. Thereare several methods available to obtain the necessary absolute scale factor. For the structuresdiscussed in this thesis a method analogous to that described by Wilson (1942) was used fordetermining an initial absolute scale factor. The value of this scale factor was subsequentlyimproved upon during structure refinement. For a detailed description of the methods used todetermine the absolute scale factor see Appendix A.2.3 Structure RefinementAn immediate advantage in growing crystals of mutant cytochromes c isomorphous to thoseof the reduced wild-type structure is that it facilitates the process of structure solution sincedifference Fourier techniques can be used to build an initial structural model.2.3.1 Construction of the starting modelThe first step in constructing a starting model for refinement is the examination of a differenceFourier map. The coefficients for this map are (Fat/ — Fa  t \ ,) where Ft' is the observedstructure factor of the wild-type yeast iso-l-cytochrome c structure (reduced state) and Frnutis the structure factor from the mutant cytochrome c protein that is being studied. The phasesused are those of the refined wild-type structure. This Fourier map shows positive electrondensity peaks where there are atoms in the wild-type protein, which are absent in the to berefined structure, and vice versa for the negative peaks (see Figure 2.15a).However, such maps are generally insufficient for building a complete starting model andtherefore, for the structures discussed in this thesis a more conservative approach was taken.Starting with the wild-type yeast iso-l-cytochrome c structure, atoms which were close to eithernegative or positive peaks in the difference Fourier map were removed from this structure. Inpractice this means that mutated residues were represented by alanine residues and waterChapter 2. General Overview of Experimental Methods^ 32Figure 2.15: Stereo-diagrams of electron density maps in the region of the mutation site of the N521mutant yeast iso-1-cytochrome c (oxidized) at different stages of structural refinement. In (a) is shownthe initial difference electron density map used for constructing a starting model. Positive peaks arerepresented by the cages drawn with solid lines, whereas the dashed line cages indicate negative peaks.A drawing of the (F0 — F,) difference electron density map after 11 cycles of refinement prior to fittingthe missing atoms of the mutated residue is shown in (b). The (2F 0 — Fc) difference map of the finalrefined structure is displayed in (c).Chapter 2. General Overview of Experimental Methods^ 33molecules near mutation sites were deleted from the model.2.3.2 Reciprocal-space refinementThe refinement of the cytochrome c structures examined herein was carried out with the re-strained parameter least-squares reciprocal-space refinement program PROLSQ (Hendrickson& Konnert, 1980, 1981; Hendrickson, 1985 ). A discussion of the theory behind this structurerefinement approach is given in Appendix B. Refinements were terminated when overall shiftswere small (overall r.m.s. shift smaller than 0.05 A) and when no further improvements could bemade by inspection of difference electron density maps (Figure 2.16). The total number of re-finement cycles necessary for each structure was highly dependent on the amount of diffractiondata available. In general, cytochrome c structures for which more than 4000 unique reflectionshad been collected could be refined in less than 60 cycles. When fewer reflections were available,refinement required at least 70 cycles to reach convergence. The structures obtained all hadgood agreement between the observed and calculated structure factors (R-factor range 17-22%)and good stereochemistry (see Table 2.6).2.3.3 Manual interventionsInspection of (F0 — Fc ) and (2F, — Fc ) difference electron density maps calculated over the courseof refinement was used to identify missing atoms excluded from the original starting model.This applies not only to mutated residues (see Figure 2.15b) but also to solvent molecules. Newsolvent molecules were only included in the refinement model if they had reasonable hydrogenbond partners and were consistently observed in (2F, — Fe) difference electron density maps.In latter stages of refinement (3F0 — 2F,) difference maps were also examined and proved to bevery helpful in repositioning the side-chains of surface residues.1'''■.-2L-"----____"3L__,4 5I 1,,^I I, II\ ...-I I,^I 1, 11 ,r^I II----1I10^15^20^25^30Refinement Cycle3.53.02.52.01.50^50cf.-N.o0 ....-.•-10 cL oo .1.-4z.65• AccC.' 30.0L.-.6 25.04I II 20 0ce^.15.0...1-e0.12..-...-.O0 0.086.0.04cr0.00-04?025c ......,^.o ,co 06.,...0— 0.020'5.c rS° 0.0150.010Chapter 2. General Overview of Experimental Methods^ 34Figure 2.16: A plot of the variation of the R-factor, overall r.m.s. shifts, and r.m.s. bondlength and torsional angle deviations from ideal values, during the course of refinement for theoxidized yeast iso-1-cytochrome c N52I mutant. Five manual interventions were carried outduring refinement as indicated by the numbered vertical breaks. In the first intervention watermolecules with thermal factors over 40 A2 were deleted and the absolute scale of the observedstructure factors was adjusted. In the second intervention the Ile52 side-chain was fitted (seeFigure 2.15b) and further adjustments to the solvent structure and the absolute scale weremade. The remaining three manual interventions were used to further adjust solvent structureand position solvent exposed side-chain groups with high degrees of mobility.2.3.4 Accuracy of StructuresSeveral methods have been devised to estimate the errors in the atomic coordinates of macro-molecules determined using x-ray diffraction methods. For the structures described in this thesisChapter 2. General Overview of Experimental Methods^ 35Table 2.6: Typical stereochemistry for refined structures of yeast iso-1-cytochrome cStereochemicalrefinement parametersBond distances (A)1-2 bond distance1-3 bond distance1-4 bond distancespecial distancestPlanar restraints (A)Chiral volume (A 3 )Non-bonded contactst (A)single-torsionmulti-torsionpossible hydrogen bondsTorsion angles ( 0 )planar (0°or 180°)staggered (±60°,180°)orthonormal (±90°)r.m.s. deviation fromideal valuesTypical refinement restraintweighting values0.017-0.024 0.0200.044-0.050 0.0300.052-0.067 0.0500.048-0.101 0.0600.016-0.018 0.0200.197-0.286 0.1500.218-0.238 0.2500.179-0.242 0.2500.212-0.262 0.2502.3-2.7 2.522.3-26.7 20.018.3-24.6 15.0t The special distances define the bonds between the heme iron and the His18 and Met80ligands.t The r.m.s. deviations from ideality for this class of restraint incorporates a reduction of 0.2 Afrom the radius of each atom involved in a contact.two methods are used: the Luzzati plot (Luzzati, 1952) and Cruickshank formula (Cruickshank,1949, 1954, 1985). Both methods give similar estimates for the accuracy of the cytochrome cstructures described herein with overall r.m.s. coordinate errors in the range of 0.15-0.25 A. Athorough discussion of coordinate error determination is given in Appendix C.Chapter 3Oxidation State-dependent Conformational Changes in Cytochrome c3.1 Experimental Procedures3.1.1 CrystallizationYeast iso-l-cytochrome c was isolated from Saccharomyces cerevisiae as previously described(Pielak et al., 1985, 1986). Crystals of the oxidized form of this protein were grown usingthe hanging drop method and employing a hair seeding technique (Leung et al., 1989; see alsoSection 2.1). Crystal growth was observed under similar conditions to those previously reportedfor the reduced form of yeast iso-l-cytochrome c (Sherwood & Brayer, 1985; Louie et al., 1988a).The buffer used was 0.1 M sodium phosphate, pH 6.2 which contained 30 mM sodium nitrateinstead of a reducing agent such as DTT. As precipitant a 90% saturated ammonium sulfatesolution was used. In this manner small crystals (0.2 x 0.1 x 0.05 mm) were grown in 2 to3 days. To obtain larger crystals, macro seeding was performed. After 7 to 10 days, crystalssuitable for high resolution x-ray diffraction (0.5 x 0.4 x 0.2 mm) were obtained. In order toensure complete oxidation of the protein, crystals were transferred into a solution containing20 mM potassium ferricyanide prior to mounting for diffraction studies. Spectroscopic analysesof crystalline material confirmed that the iso-1-cytochrome c present was in the oxidized form.Precession camera photography and further diffractometry measurements showed crystals ofoxidized iso-l-cytochrome c were isomorphous with those grown for the reduced form of thisprotein. The space group is P4 3 2 1 2 with cell dimensions of a = b = 36.47 A and c = 137.24 A(cell dimensions of reduced iso-l-cytochrome c are a = b = 36.46 A and c = 136.86 A).36Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 373.1.2 Data collection and data processingDiffraction data was collected to 1.9 A resolution from a single crystal on an Enraf-NoniusCAD4-F11 diffractometer, with a 36.8 cm crystal to counter distance and a helium purged pathfor the diffracted beam. The radiation used was nickel filtered and generated from a coppertarget X-ray tube operating at 26 mA and 40 kV. A total of 8632 intensities were measuredusing continuous w scans of 0.6° in width at a scan speed of 0.55° min -1 , with backgroundstaken as the terminal one-sixth of the total scan width on either side of each reflection. Threestandard reflections were measured every 2.8 h of X-ray exposure time to monitor crystal decayand slippage. The ambient temperature during data collection was maintained at 15°C.To obtain structure factors, diffraction intensities were first corrected for backgrounds. Inthis process the measured backgrounds for each reflection, as well as those of neighboring re-flections, were used to estimate the correction factor. Intensities were further corrected for ab-sorption using an empirical curve (North et al., 1968) obtained by measuring a phi-independentreflection in 72 consecutive 5° steps. The correction for crystal decay was performed using apolynomial fit to the decay profiles of the three monitor reflections. Diffraction intensities werethen merged, corrected for Lorentz and polarization effects and converted into structure factoramplitudes (7925 in total; see also Section 2.2). These were put on an absolute scale using theWilson plot statistical method (Wilson, 1942; see also Appendix A).3.1.3 Refinement and analysesInitially, a reduced minus oxidized iso-1-cytochrome c difference electron density map wascalculated. The reduced structure used to phase difference map coefficients had previouslybeen refined to 1.7 A resolution with an R-factor of 19.7% (Louie & Brayer, 1990). Since thismap indicated that no large positional shifts were present between the two oxidation states,restrained parameter refinement (Hendrickson & Konnert, 1981; see also Appendix B) of theoxidized protein was initiated using the reduced protein structure as a starting model. Includedwere 65 water molecules from the reduced structure with thermal factors less than 30 A2 . AlsoChapter 3. Oxidized Yeast Iso-1-Cytochrome c^ 38Table 3.7: Final stereochemistry of oxidized yeast iso-1-cytochrome c at 1.9 A resolutionStereochemicalrefinement parametersr.m.s. deviation fromideal valuesRefinement restraintweighting valuesBond distances (A)1-2 bond distance 0.022 0.0201-3 bond distance 0.046 0.0301-4 bond distance 0.057 0.050Planar restraints (A) 0.018 0.020Chiral volume (A 3 ) 0.231 0.150Non-bonded contactst (A)single-torsion 0.226 0.250multi-torsion 0.227 0.250possible hydrogen bonds 0.254 0.250Torsion angles (°)planar (0°or 180°) 2.8 2.5staggered (±60°,180°) 26.7 19.0f The r.m.s. deviations from ideality for this class of restraint incorporates a reduction of 0.2 A. fromthe radius of each atom involved in a contact.included were four internal water molecules and a sulfate anion. Refinement utilized 3929structure factors which had a F/a(F) ratio > 2.0 and were within the resolution range of6.0-1.9 A. During the course of refinement three manual interventions were carried out basedon F, — I', difference and fragments maps, and 2F0 — I', and 3F0 — 2F, difference maps. Inthe first, four more water molecules were removed from the model, several water moleculeswere repositioned, and some surface lysine and arginine side-chains were placed in better fittingdensity. During the second manual intervention special effort was put into the positioning ofthe Asn52 side-chain which had ill-defined density. In the last manual intervention only theterminal side-chain group of a number of asparagine and glutamine residues were adjusted tooptimize electron density fits and hydrogen bonding interactions.After a total of 117 cycles of refinement, convergence was reached to give a final crystal-lographic R-value of 19.7%. As documented in Table 3.7, the final structure determined foroxidized yeast iso-l-cytochrome c has good stereochemistry. From a plot of R-factor versus^to.^al>-0.8 t. In-0.6 0.^CO-0.4-^20.20.30—92 0.20—_Ice0.15—0.12^0.16^0.20SIN(9) / A0.100.08 0.24 0.28Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 39Figure 3.17: Plots of the dependence on resolution of both the R-factor agreement betweencalculated and observed structure factors (A — A), and the fraction of data used (C) — 0, axisat top right), for the final refined model of oxidized yeast iso-l-cytochrome c. For this analysisreciprocal space was divided into shells according to sin(0)/A with each containing at least260 reflections. For the purpose of assessing the accuracy of the atomic coordinates, curvesrepresenting the theoretical dependence of R-factor on resolution assuming various levels ofr.m.s. error in the atomic positions of the model (Luzzati, 1952) are also drawn (dashed lines).This analysis suggests an overall r.m.s. coordinate error of ,0.22 A.resolution (Luzzati, 1952) one can estimate the r.m.s. error in the coordinates of the structureto be about 0.22 A (see Figure 3.17). A separate estimate of the atomic coordinate error canbe obtained from an unrestrained refinement of the final model (Read et al., 1983). Four cyclesof unrestrained refinement caused the R-factor to drop to 14.5%. The resulting model had anr.m.s. deviation of 0.21 A from the final structure. A third method for estimating coordinate er-rors is that of Cruickshank (1949,1954). Using this method an overall r.m.s. error for all proteinatoms of 0.23 A is obtained. One has to realize that these coordinate error estimates are aver-age values, and those parts of the structure that are well defined will have smaller coordinateerrors, whereas the uncertainty in the positions of mobile surface side-chains is undoubtedlymuch higher.Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 40Figure 3.18: A stereo-drawing of the conformations of all main-chain (thick lines) and side-chain(thin lines) atoms in yeast iso-l-cytochrome c in the oxidized state. The heme group has alsobeen drawn with thick lines. Also drawn are the two heme ligand bonds to His18 and Met80,and the two covalent thioether linkages to cysteines 14 and 17. For clarity, every 5th a-carbonatom has been labelled with its one-letter amino acid designation and sequence number.3.2 Results3.2.1 Polypeptide chain conformationFor the comparison with the reduced state of this protein the oxidized structure was superim-posed on the basis of all main-chain and heme atoms on the 1.2 A reduced iso-l-cytochrome cstructure using a least-squares algorithm. A stereo-drawing of the structure of oxidized yeastiso-1-cytochrome c is presented in Figure 3.18. A similar illustration for the reduced protein isshown in Figure 1.5 as reproduced from the data of Louie & Brayer (1990). A plot of the ob-served average positional deviations between corresponding main-chain atoms in oxidized andreduced yeast iso-1-cytochrome c is shown in Figure 3.19. The conformational changes observedare small and it is notable that none of the hydrogen bond interactions between main-chainatoms are lost upon change in oxidation state. The overall average value for main-chain atomdifferences in all residues is 0.31 A. Note that Thr(-5) and Glu(-4) are substantially disorderedin both structures of yeast iso-l-cytochrome c and the differences observed are therefore likelyChapter 3. Oxidized Yeast Iso-/-Cytochrome c^ 41I ^ I ^ I ^ I ^ I ^ I ^ I ^ I ^ I ^ IsI^I5 15^25i^I^ r35 45 55^65Residue Number75 85 95Figure 3.19: A comparison of the average positional deviations between main-chain (thick line)and side-chain (thin line) atoms of the oxidized and reduced forms of yeast iso-l-cytochrome c.The horizontal dashed line represents the overall value of 0.31 A observed for the averagedeviation between all main-chain atoms. The filled dark circle at position 104 represents theoverall average deviation for all heme atoms of 0.24 A.due to differential fits to the same poor electron density rather than a reflection of oxidationstate. This disorder is apparent from the large thermal B values ( ,50-55 A 2 ) assigned to theseresidues in both structures.Two conformational changes do appear to result from differing oxidation states. The largestinvolves G1y84 (average deviation of 0.75 A) and results in the formation of a new hydrogenbond in the oxidized protein (length 2.6 A) between the side-chain of Arg13 (NH1 moves ,-1.0 A)and the carboxyl oxygen atom of G1y84 (moves , 1.1 A; see Figure 3.20). In the reduced proteinboth of these groups are found associated with surface solvent molecules (Louie & Brayer, 1990).As Figure 3.21 illustrates, a second conformational change in the main-chain of Trp59 (averagedeviation 0.64 A) is probably the result of a lengthening in the oxidized protein of the hydro-gen bond between the side-chain of this residue and the nearby heme propionate group (from 3.1Wet 107- .%<•• . ,,Gly 84Gly 89Ala 81Arg 19Phe 82Met 80Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 42Figure 3.20: A composite stereo-drawing showing the course of the polypeptide chain aboutArg13 and G1y84 in the oxidized (thick lines) and reduced (thin lines) states of yeast iso-l-cyto-chrome c. Upon oxidation these two residues move closer together to form a new hydrogen bond(dashed line). Wat 107 also undergoes a comparable shift in position and is hydrogen bondedto the carboxyl group of G1y84 in both oxidation states. Also drawn are the relative positionsof the heme moiety and the side-chain of Phe82.Wet 124 Wat 124Figure 3.21: A composite stereo-drawing showing a portion of the heme group and the adjacentpolypeptide chain about Trp59 in the oxidized (thick lines) and reduced (thin lines) states ofyeast iso-l-cytochrome c. In the oxidized protein the hydrogen bond (dashed lines) betweenthe side-chain of Trp59 and the nearby heme propionate lengthens leading to a shift in themain-chain atoms of this amino acid. Corresponding shifts are observed for Wat124 and theside-chain of Asp60.I I Ai 04 YK,•,^ Ai —4^„AAA- 'r/^I'II/ "I ► ywyr Pr vf\ fi\^,t 4.k,,i ,,‘44\ Aro sof,1^/110C5 go0-5 135040(-1§-15 3020< 100Chapter 3. Oxidized Yeast Iso-I-Cytochrome c^ 43—5^5^15^25^35^45^55^65^75^65^95^103Residue NumberFigure 3.22: Plots of the average thermal factors of main-chain atoms along the polypeptidechain in oxidized (thick line) and reduced (thin line) yeast iso-l-cytochrome c. The upperplotted line (axis designation at top right) shows the observed differences between the meanmain-chain temperature factors in the two oxidation states. Three regions of polypeptide chainwith significantly higher thermal factors in the oxidized protein (residues 47-59 65-72 and81-85) have been highlighted with cross hatched boxes.to 3.4 A). This is reflected in an increase in the average thermal parameters for the side-chainof Trp59 from 15 to 21 A2 . Associated with this movement is the reorientation of the side-chain of Asp60. Other large side-chain displacements in Figure 3.19 are associated with poorlydefined surface residues and would appear to result from positional uncertainty rather than asa consequence of change in oxidation state.Between oxidation states, comparable values are observed for the overall average thermalfactor for all atoms in the polypeptide chain of yeast iso-l-cytochrome c (16.4 and 16.5 A2 forreduced and oxidized, respectively). However, as evident from Figure 3.22, if comparisons aremade of thermal factors along the course of the polypeptide chain, substantive differences arepresent. These differences are further illustrated in the matrix representation of Figure 3.23. InLegend15.0 and greaterRON 10.0 to 15.05.0 to 10.096 -86-76 -66 -56 -46 -36 -26 -16-6--5^6^16^26 36^46^56Residue number66^76^86^96Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 44Figure 3.23: A matrix representation of the differences in average main-chain thermal fac-tors between the oxidized and reduced states of yeast iso-l-cytochrome c. Each matrixpoint Px ,y represents an amino acid pairing (x, y) and was calculated using the equation:Px,y = (-13s — By)oxidized (Bx By)reduced where B is the average main-chain thermal factorof a given amino acid. Positive matrix values are displayed as squares of different levels ofblackness according to the scale on the right. As this matrix has inverse symmetry across thediagonal line drawn, negative values are redundant and are omitted for clarity. The advantageof this approach is that displayed values are not affected by differences in overall thermal factorbetween the two structures. Within the matrix, amino acids which have significantly higheraverage main-chain thermal factors in the oxidized structure produce vertical streaks. These in-clude residues 47-59, 65-72 and 81-85, with maximal differences observed for Asn52, Tyr67 andPhe82. As discussed in the text, the N-terminal region is not considered in this analysis due tothe presence of positional disorder in both oxidation states. Amino acids producing horizontalstreaks indicate the presence of significantly larger thermal factors for their main-chain atomsin the reduced state of yeast iso-l-cytochrome c. Residues of this type are fewer in number andare centered about Arg38, Thr96 and Leu98.Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 45terms of lower thermal factors in the oxidized protein, the most significant main-chain differenceinvolves Arg38, a residue having a water mediated interaction with the adjacent pyrrole ring Apropionate group. Also affected are the two immediately adjacent residues Gly37 and His39.The side-chain atoms of Arg38 exhibit an even larger , 14 A2 lowering of average thermal Bvalues in the oxidized state.In total, four regions of polypeptide chain have significantly higher thermal factors in theoxidized form of yeast iso-l-cytochrome c. The significance of the first, which involves the N-terminus, is questionable since this region is disordered and poorly resolved in electron densitymaps. The three other segments affected (residues 47-59, 65-72 and 81-85) are particularlyinteresting since they appear to be related to oxidation state dependent conformational changesin internal water structure and the heme group. As Figures 3.22 and 3.23 illustrate, thoseportions of polypeptide chain are sharply delineated, with maximal increases in thermal factorsin the oxidized state being observed for Asn52, Tyr67 and Phe82. Figure 3.24 shows that thesethree segments are located to the Met80 ligand side of the heme group.3.2.2 Heme structureHigh resolution studies of reduced yeast iso-l-cytochrome c have shown that the heme groupis substantially distorted from planarity into a saddle shape (Louie & Brayer, 1990). In theoxidized protein the type of distortion observed is similar, but considerably more pronounced,suggesting the degree of heme planarity is dependent on oxidation state (Table 3.8). None ofthe heme iron coordinate bonds is significantly different between oxidation states, although thelargest deviation observed is a lengthening of the Met80 ligand bond. Overall, the averagedisplacement of side-chain atoms in Met80 is 0.36 A between redox states, with the largestshifts observed for the CB (0.55 A) and CE (0.47 A) carbon atoms. These displacementsresult in a small 7° change in the torsion angle around the Met80 - heme iron ligand bond.It is notable that a substantial increase in thermal factors for the side-chain of Met80 is alsoobserved. In reduced yeast iso-l-cytochrome c the average side-chain thermal factor is 5 A2Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 46Table 3.8: Heme conformation and ligand geometry in the two oxidation states of yeastiso-l-cytochrome ctI. Angular deviations between both the plane normals of individual pyrrolerings and the heme coordinate bonds and both the pyrrole nitrogen planeand the porphyrin ring plane ( 0 ) ta)^Pyrrole ring^Pyrrole N plane^Porphyrin ring planeA 12.6 (9.4) 8.3 (6.7)B 14.1 (11.1)^12.5 (11.9)C^9.6 (8.8) 13.8 (9.8)D 12.8 (8.1)^10.0 (6.0)b) Heme coordinate bondsFe - His18 NE2^7.2 (2.1)^6.5 (3.3)Fe - Met80 SD 3.3 (4.9) 5.4 (7.5)II. Heme iron coordinate bond distances (A)His18 NE2^2.01 (1.99)Met80 SD 2.43 (2.35)Heme NA^1.97 (1.97)Heme NB 1.98 (2.00)Heme NC^2.01 (1.99)Heme ND 2.05 (2.00)III. Heme propionate hydrogen bond interactions (distances in A) §Heme atom^Hydrogen bond partners and distances01A Tyr48 OH 2.83 (2.83), Wat121 2.85 (2.81),Wat168 2.87 (2.85)02A^G1y41 N 2.60 (3.21), Asn52 ND2 3.54 (3.34),Trp59 NE1 3.43 (3.09), Wat121 3.34 (4.01)01B^^Thr49 OG1 2.79 (2.64), Thr78 OG1 3.07 (2.90),Lys79 N 2.67 (3.17)02B^Thr49 N 2.75 (2.94)t Values for reduced yeast iso-l-cytochrome c are shown in parentheses.t Each pyrrole ring is defined by nine atoms, which include the five ring atoms plus the first carbonatom bonded to each ring carbon. The porphyrin ring plane is defined by the five atoms in each ofthe four pyrrole rings, the four bridging methine carbons, the first carbon atom of each of the eightside-chains and the heme iron (33 atoms in total). The pyrrole nitrogen plane is defined by only the 4pyrrole nitrogens (see Figure 1.2 for the heme atom labeling convention used).§ Distances are provided for the heme 02A to Wat121 (reduced state) and Asn52 ND2 (oxidized state)interactions for the sake of comparison, even though these are too long to represent hydrogen bonds.Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 47Figure 3.24: A space-filling representation of yeast iso-l-cytochrome c showing the locationof the three segments of polypeptide chain (drawn in gray spheres) which have main-chainthermal factors which are significantly larger in the oxidized protein. A general indication ofthe location of the individual affected segments is provided by the letters A—C, corresponding toresidues 47-59, 65-72 and 81-85, respectively. Heme atoms are shown with black spheres, whilewhite spheres indicate regions of polypeptide chain with comparable or lower thermal factorsin oxidized yeast iso-l-cytochrome c. Although contiguously linked through interactions in theprotein interior, the three segments having higher thermal factors are subdivided at the proteinsurface by the polypeptide strand composed of residues 73-80 (labelled D).whereas it increases to 12 A2 in the oxidized protein. Coupled with the observed shifts inbond orientation, these thermal factors suggest a weakening of the Met80 SD - heme ironligand bond in the oxidized state. In contrast, no significant change in ligand bond distanceor thermal parameters is observed for His18 between oxidation states. However, the preciseorientation of the His18 side-chain may be a function of oxidation state in that a 46.7° angleis found between the imidazole ring plane and a vector drawn through the NA and NC pyrroleChapter 3. Oxidized Yeast Iso-1-Cytochrome c^ 48Table 3.9: Heme solvent accessibility in the two oxidation states of yeast iso-l-cytochrome cOxidized Reduced1. Solvent accessible heme atoms andsurface area exposed (A 2 )CHD 0.0 2.6CMC 9.5 9.4CAC 5.1 3.3CBC 19.2 17.4CMD 11.7 9.82. Total heme exposure (A 2 ) 45.5 42.53. Total heme surface (A 2 ) 503.3 495.74. Heme surface area exposed (%) 9.0 8.6Computations were done using the method of Connolly (1983) and the results represent theaccessible molecular surfaces of the atoms listed (see Figure 1.2 for heme atom nomenclature)The probe sphere used had a radius of 1.4 A.nitrogen atoms in reduced yeast iso-l-cytochrome c, whereas this value is 55.8° in the oxidizedprotein.An analysis of the heme solvent accessibility in the two oxidation states is documented inTable 3.9. These results show that total heme accessibility is comparable, although slightlyhigher in the oxidized protein.Another aspect of heme structure differing between oxidation states is the positioning ofthe pyrrole ring A propionate group (see Figure 3.25). This group's alternative placement inoxidized yeast iso-l-cytochrome c can be ascribed to changes in three torsion angles. The firstinvolving the C2A—CAA bond which rotates ,,,20°, the second involving the CAA—CBA bondwhich rotates ,30°, the third involving the CBA—CGA bond which is rotated -- ,45°. The netresult is that the 02A oxygen atom of the propionate carboxyl group moves ,-,-,0.6 A in a directiontowards G1y41, with the 01A oxygen atom showing a more modest 0.2 A displacement.As evident from Table 3.8 and Figure 3.25, not all groups that hydrogen bond to this pro-pionate can accommodate the observed positional shifts in the oxidized state. In particular,Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 49pyrrole A propionate^ pyrrole A propionateA BFigure 3.25: Drawings of the region about the pyrrole A propionate group in (a) reduced and(b) oxidized yeast iso-1-cytochrome c, illustrating the positional shifts and altered hydrogenbonding patterns observed. The pyrrole ring A propionate group is highlighted with darkshaded balls. Hydrogen bonds are indicated by thin dashed lines. The two internally boundwater molecules Wat121 and Wat168, which mediate the interaction of Arg38 with this hemepropionate, are shown with larger spheres.much weaker interactions are made to Asn52 and Trp59, and both of these side-chains haveincreased thermal factors. Also affected is the position of the internally bound water moleculeWat121, which is conserved in all eukaryotic cytochrome c structures determined to date (Bush-nell et al., 1990). In reduced yeast iso-l-cytochrome c, Wat121 forms a hydrogen bond to theO1A oxygen of the pyrrole ring A propionate group. As can be seen in Figure 3.25, a ,0.5 Ashift of Wat121 in the oxidized protein allows the formation of bifurcated hydrogen bonds toboth the propionate O1A and 02A carboxyl oxygen atoms. Notably, the side-chain of Arg38,whose interaction with the propionate group is mediated by Wat121 and Wat168 shows nosignificant positional displacement in the oxidized protein.3.2.3 Internal water structureThe internally bound water molecule, Wat166 undergoes a large shift in position in responseto oxidation state. In oxidized yeast iso-l-cytochrome c, Wat166 moves 1.7 A almost directlyTyr 67Wat 166.. *******Thr78Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 50A BFigure 3.26: Heme and polypeptide chain structure about the internally bound water moleculeWat 166 in (a) reduced and (b) oxidized yeast iso-l-cytochrome c. Heme ligand interactionsare indicated by thin white bonds, whereas hydrogen bonds are shown by dotted lines.towards the heme iron atom, closing the distance between these groups from 6.6 A to 5.0 A(Figure 3.26). Most affected by this movement is Asn52, to which Wat166 no longer forms ahydrogen bond. As mentioned before, the hydrogen bond between Asn52 and the pyrrole ringA propionate group is also lost (Table 3.8 and Figure 3.25). Despite the large shift in Wat166,hydrogen bonds to both Tyr67 and Thr78 are retained and the distances between the hydrogenbonding atoms involved are similar in both oxidation states.As illustrated in Figure 3.27a, in reduced iso-1-cytochrome c Wat166 is very tightly con-strained into a small spherical cavity having a volume of , 10 A3 . To accommodate Wat166movement, the densely packed protein matrix must undergo small concerted conformationaladjustments, the largest of these involving the side-chain of Asn52 (average side-chain displace-ment is 0.47 A). As shown in Figure 3.27b, the resultant cavity surrounding Wat166 in theChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 51ABFigure 3.27: Stereo-drawings of the internal cavity occupied by Wat166 in (a) the reducedand (b) the oxidized form of yeast iso-l-cytochrome c. Also drawn are the nearby residuesto which Wat166 is hydrogen bonded and the adjacent heme group and heme ligands. In thereduced protein the volume of the cavity is -40 A3 , which expands to ,25 A3 in oxidized yeastiso-l-cytochrome c.oxidized protein is significantly expanded and is now ,25 A3 in volume and highly asymmetri-cal, with Wat166 occupying only that portion of the available volume closest to the heme ironatom. This leaves considerable free volume available to amino acids adjacent to Wat166 andthey are therefore conformationally less constrained.The looser packing evident about Wat166 in oxidized yeast iso-l-cytochrome c could providean explanation for the observed increased thermal factors of the nearby polypeptide segmentsChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 5247-59 and 65-72. For the 47-59 segment this trend is undoubtedly accentuated by the loss ofhydrogen bonds from Wat166 and the pyrrole A propionate to the Asn52 side-chain, as well asthe lengthening of the hydrogen bond from the same propionate to the side-chain of Trp59. AsFigures 3.22 and 3.23 demonstrate, the largest average main-chain thermal factor increase inthe 47-59 segment is localized at Asn52, with the side-chain of this residue also experiencinga similar increase (average increase +16 A 2 ). Another potentially important aspect of the lossof the Wat166 to Asn52 ND2 hydrogen bond, is the removal of a structural link between the47-59 and 65-72 polypeptide segments. Despite retention of the Wat166 to Tyr67 OH hydrogenbond, the main-chain of residues 65-72 show increased thermal factors in the oxidized state.This increase is maximal at Tyr67 and is mirrored in a substantive overall mean increase in itsoverall side-chain thermal factor from 13 A 2 to 27 A2 on going from the reduced to oxidizedstates. The primary factor involved in the mobility changes in the 65-72 region again appearsto be the increased conformational volume accessible to these residues.Although the hydrogen atoms of Wat166 cannot be resolved experimentally, the nature ofthe surrounding hydrogen bonding network makes it possible to speculate on the orientationof the dipole moment of this internal water molecule in both the reduced protein where theheme iron atom is formally uncharged, and in the oxidized state where the iron atom has apositive charge. As illustrated in Figure 3.26a, the complex of hydrogen bond interactionsinvolving Wat166, Asn52, Tyr67, Thr78 and Met80 in reduced yeast iso-l-cytochrome c, re-strictively defines the orientation of Wat166 and thus its dipole moment. Based on Tyr67 OHand Thr78 0G1 being able to function both as hydrogen bond donor and acceptor groups,three potential orientations for Wat166 can be delineated. Two of these orientations cannot berealized without breaking hydrogen bonds between protein groups and therefore seem unlikely.The third alternative orientation maximizes hydrogen bond interactions (see Figure 3.28a). Inthis Wat166 orientation the positive end of the dipole of this group is pointing in the directionof the heme iron atom with a ,50° declination. A nearly identical dipole moment orientationfor Wat166 can be modelled in reduced yeast iso-2 and tuna cytochromes c, the only other twoAm 52Chapter 3. Oxidized Yeast Iso-1-Cytochrome c^ 53Figure 3.28: Stereo-drawings showing the immediate vicinity about Wat166 in (a) reduced and(b) oxidized yeast iso-l-cytochrome c, as well as in (c) oxidized horse cytochrome c. Alsoshown is the hypothetical placement of hydrogen atoms for the purpose of discerning the dipolemoment of Wat166 in the presence and absence of a positive charge at the heme iron atom.Chapter 3. Oxidized Yeast Iso-I-Cytochrome c^ 54reduced cytochromes c whose structures are known (Murphy et al., 1992; Takano & Dickerson,1981a).The placement of the dipole moment of Wat166 in the oxidized state is less clearly definablesince the interaction formed to the side-chain of Asn52 is not present, allowing for severalorientational possibilities. However, given that the heme iron atom in the oxidized state ispositively charged, it is reasonable to expect that the resultant electrostatic field will be animportant factor in the orientation of the dipole of Wat166. Based on a simple model, wherepropionate oxygen atoms each have a charge of —1, the heme iron atom a charge of +1 anda water molecule has a dipole moment of 1.85 Debye, the following equation can be used tocalculate the potential energies for the various possible orientations for Wat166:U = — 1113114 cos 0 (3.3)where U is the potential energy of the dipole, ii. is the dipole moment, P is the intensity ofthe electric field caused by the surrounding charges and 0 is the angle between fi and E. Theelectric field strength can be evaluated using Coulomb's law.From this analysis two energetically favorable orientations can be derived. In one, which isillustrated in Figure 3.28b, the negative end of the dipole of Wat 166 is pointing almost directlyat the heme iron atom, but requires the Tyr67 OH to Met80 SD hydrogen bond to be broken. Inthe second possible orientation, the hydrogen bond between the side-chains of Tyr67 and Met80is retained, but as a consequence the dipole moment is less favorably oriented. The differencein potential energy of these two water dipole orientations is ,55 KJ/mole, with the first beingmore favorable. Clearly which dipole orientation will dominate depends on the strength of theTyr67 OH to Met80 SD hydrogen bond which is broken in one orientation and not in the other.Suggested values for the energy of a hydrogen bond range from —15 to —25 KJ/mole (Schultz &Shirmer, 1979; Creighton, 1984). Since the energy gain in optimally orienting the dipole in theelectric field is more than twice the energy lost by breaking a hydrogen bond, it is reasonableto expect the dipole orientation shown in Figure 3.28b to predominate in the oxidized yeastiso-l-cytochrome c structure.Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 55The loss of the Tyr67 OH to Met80 SD hydrogen bond in the oxidized state could beexpected to be an additional stabilizing feature in two ways. Firstly, it would tend to make theMet80 heme ligand less electron withdrawing, a situation favoring stabilization of the positivelycharged heme iron atom. Secondly, the new hydrogen bond interaction from Wat166 to the side-chain of Tyr67 assists in properly orienting the dipole moment of Wat166 adjacent to the hemegroup in order to maximally stabilize the positive charge resident there. That the hydrogenbond to the side-chain of Met80 is lost, is supported by the observed increase in thermal factorsfor both the side-chains of Tyr67 and Met80 in the oxidized state.Further confirmation of the dipole orientation of Wat166 in oxidized yeast iso-l-cytochrome ccan be derived from the available three-dimensional structures of oxidized horse (Bushnellet al., 1990), tuna (Takano & Dickerson, 1981b) and rice (Ochi et al., 1983) cytochromes c.The presence and position of Wat166 is conserved in all these proteins (Bushnell et al., 1990).In addition, the determination of the direction of the dipole moment of Wat166 has greatercertainty, since for these proteins the hydrogen bond to Asn52 is retained in the oxidized state.In each of these structures, the most likely dipole orientation for Wat 166 is similar to thatproposed herein for oxidized yeast iso-l-cytochrome c. This can be seen in Figure 3.28c, whichshows the dipole orientation of Wat166 in horse cytochrome c. Collectively, these results clearlysuggest a major role for Wat166 in stabilizing the oxidized as well as the reduced forms of cyto-chrome c through the differential orientation of dipole moment, shift in distance to the hemeiron atom and alterations in the surrounding hydrogen bond network.3.3 Discussion3.3.1 Focal points for oxidation state dependent structural alterationsA compilation of significant structural differences observed on going from the reduced to oxidizedstates of yeast iso-l-cytochrome c is presented in Table 3.10. It is clear from this data thatoxidation state dependent conformational differences are for the most part expressed in terms ofchanges in thermal parameters of specific regions, adjustments to heme structure, movement ofChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 56Table 3.10: Structural changes observed on going from the reduced to the oxidized state inyeast iso-l-cytochrome c1. Positional displacements of polypeptide chain (see Figure 3.19)(a) Movement of Arg13 and Gly84 to form a hydrogen bond (see Figure 3.20)(b) Lengthening of the interaction between Trp59 and the heme pyrrole ring A propionategroup (see Figure 3.21)2. Thermal factor parameters of main-chain atoms (see Figures 3.22 and 3.23)(a) Lower values observed for residues 37-39, focussed at Arg38; side-chain of Arg38 alsohas reduced values(b) Higher values found for three polypeptide chain segmentsi. residues 47-59, focussed at Asn52ii. residues 65-72, focussed at Tyr67iii. residues 81-85, focussed at Phe82All the side-chains of Asn52, Tyr67 and Phe82 show higher mobility3. Heme structure and ligands(a) Increased distortion of heme planarity (see Table 3.8)(b) Readjustment of the pyrrole A propionate group with a realignment of hydrogen bond-ing interactions (see Figure 3.25)(c) Higher thermal parameters for the Met80 side-chain.4. Internal water structure(a) Large displacement of Wat166 towards the heme iron atom coupled with a change inhydrogen bonding interactions (see Figure 3.26)(b) Wat166 movement is facilitated by shifts in the protein matrix to enlarge the availableinternal cavity space (see Figure 3.27)(c) Reorientation of the dipole of Wat166 to favor stabilization of the charged heme ironatom (see Figure 3.28)5. Hydrogen bond interactions (see Figures 3.20, 3.21, 3.25 and 3.26)(a) Stronger : G1y41 N - Heme 02A(b) Weaker: Trp59 NE1 - Heme 02A(c) Lost: Asn52 ND2 - Heme 02AAsn52 ND2 - Wat 166Tyr67 OH - Met80 SD(d) New: Arg13 NH1 - G1y84 0Wat121 - Heme 02AChapter 3. Oxidized Yeast Iso-1-Cytochrome c^ 57internal water molecules and the realignment of hydrogen bonding patterns, rather than in termsof explicit shifts in the polypeptide chain which are found to be minimal. In total there appear tobe three points about which the observed conformational differences are focussed. These involvethe pyrrole ring A propionate group, the internal water molecule Wat166 and the Met80 hemeligand. All three of these features are structurally linked together through a hydrogen bondingnetwork. They are furthermore localized for the most part on the Met80 ligand side of the hemegroup, adjacent to the solvent exposed edge of this latter group (see Figure 3.24). As a whole,the amino acids involved also form a large portion of the protein surface implicated in beingcentral to the complexation interactions formed with electron transfer partners (Salemme, 1976;Margoliash & Bosshard, 1983; Pelletier & Kraut, 1992).It is instructive to examine each focal point for structural changes in oxidized yeast iso-l-cytochrome c in light of the conformational shifts observed and their probable role in stabilizingthis state of the protein.Heme pyrrole ring A propionate groupUpon oxidation there is an apparent change in the nature of the relationship between the hemeiron atom and the pyrrole ring A propionate, leading to a conformational adjustment in thislatter group (see Figure 3.25). As a consequence a realignment of the hydrogen bonds to thisgroup takes place (Table 3.8). Arg38 whose interaction with this propionate group is mediatedby two internal water molecules (Wat121 and Wat168) is also affected by a change in oxidationstate. In the oxidized state Arg38 becomes positionally more stable as can be deduced fromthe observed decrease in thermal factors for this residue.Precisely what triggers conformational changes about the pyrrole A propionate group andhow these serve to stabilize the oxidized state of the protein is uncertain. Nonetheless thereare two conformational states in this region corresponding to the different oxidation states ofthe heme iron atom. It is likely that the intricate network of hydrogen bonds present in thisregion facilitates these changes, perhaps by oxidation state dependent differential delocalizationChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 58of the negative charge of the ionized heme propionate group (Barlow & Thornton, 1983; Singhet al., 1987). Alternatively, direct through space interactions of the positively charged hemeiron atom and the negatively charged pyrrole A propionate may be the primary factor involvedand the observed conformational changes serve to enhance this interaction (Moore, 1983). Alsopossible are electrostatic interactions between the positively charged side-chain of Arg38 andthe heme iron atom, as well as the nearby negatively charged propionate group. Replacementof this arginine by other amino acids is known to substantially lower the reduction potentialof cytochrome c (Proudfoot & Wallace, 1987; Cutler et al., 1987) indicating the ability of thisamino acid to directly affect the functional properties of the heme. Finally it is conceivablethat the observed increased distortion of the heme from planarity in the oxidized state leadsto a need for conformational adjustments that are focussed in the area of the pyrrole ringA propionate group. In this view, the adjustments to the surrounding protein matrix maybe simply designed to accommodate the alternative positions of this propionate group. It isnotable that structurally similar changes have been observed in horse (Feng et al., 1990) andyeast iso-l-cytochrome c (Gao et al., 1991) using two-dimensional NMR, and in yeast iso-l-cyto-chrome c mutants having lowered reduction potentials and apparently hybrid oxidation stateconformations using x-ray crystallographic techniques (Louie et al., 1988b; Louie & Brayer,1989).The internal water molecule Wat166A second focal point for oxidation state dependent structural changes is Wat166. This internallyburied water molecule is directly linked to shifts about the pyrrole ring A propionate groupthrough interactions with the side-chain of Asn52 (see Figure 3.26a). In the oxidized state, theWat166 to Asn52 hydrogen bond is lost and this water molecule moves closer to the heme ironatom. These changes are accommodated by the enlargement of the cavity containing Wat166due to small concerted movements of the surrounding protein matrix. The result of theseevents in conjunction with the changes around the pyrrole A propionate group, is an increasedChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 59flexibility for the two protein segments consisting of residues 47-59 and 65-72. It should benoted that the Asn52 residue would appear to play a central role in mediating these flexibilitychanges through breakage of its hydrogen bond to both the propionate group and the internalwater molecule in the oxidized state.The apparent, more open and flexible nature of the structure of oxidized cytochrome c hasbeen documented in many other studies. For example, reduced cytochrome c has a greaterresistance to ligand displacement by exogenous substituents, to proteolytic digestion and tounfolding by both chemical denaturants and by heat (see reviews by Margoliash & Schecter,1966; Salemme, 1977). It also has a lower viscosity in solution (Fisher et al., 1973), a lowercompressibility (Eden et al., 1982) and a smaller radius of gyration as indicated by small-angleX-ray scattering (Trewhella et al., 1988). It is notable that elimination of the Wat166 cavityby replacement of Asn52 with an isoleucine residue leads to significantly increased stabilizationof the oxidized form of yeast iso-l-cytochrome c (Das et al., 1989; Hickey et al., 1991).The trigger for Wat166 movement in the oxidized state of cytochrome c and the subsequentcascade of related conformational changes, appears to be an electric field induced responseto the oxidation state of the heme. The shift of Wat166 could serve several functions. Itresults in direct stabilization of the positively charged heme iron by the dipole orientation ofWat166 (see Figure 3.28). Through the side-chain of Asn52, Wat166 is also physically linkedto conformational changes about the pyrrole A propionate group and may help mediate theseadjustments. Wat166 movement also appears to be a primary factor in the increased mobility ofnearby polypeptide chain segments which form a part of the putative surface binding site withelectron transfer partners and in this way could potentially influence binding in an oxidationstate dependent manner (see Figure 3.24). Finally, as discussed below, its positioning mayresult in modification of the strength of the Met80 SD heme iron atom ligand interaction.The oxidation state dependent positioning of Wat166 has also been observed in other mi-tochondrial cytochrome c structures in which this internal water molecule is invariably present(Bushnell et al., 1990; see also Section 1.3). As can be seen from Figure 3.29, there are clearlyChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 60Figure 3.29: A stereo-drawing of the internal water site as found in all of the high resolutionstructures of eukaryotic cytochromes c thus far solved. In thin lines are displayed: tuna, yeastiso-1 and yeast iso-2-cytochrome c all in the reduced form The oxidized structures of tuna (boththe inner and outer molecule), rice and horse are displayed in medium thick lines. The yeastiso-l-cytochrome c in the oxidized state is shown in heavy lines.two areas over which the equivalent internal water molecules in yeast iso-1 and iso-2, horse,rice, and the tuna cytochromes c are distributed. The first area, furthest away from the heme,contains the water molecules of reduced yeast iso-1 and iso-2, and tuna cytochromes c. Thearea closer to the heme holds those water molecules from the oxidized structures of yeast iso-1,tuna, rice and horse cytochromes c.Despite the obvious similarities among the different oxidized structures, there are differencesbetween the oxidized yeast iso-l-cytochrome c and the oxidized tuna, horse and rice structures(see Figure 3.30). First, the shift made by the internal water in yeast iso-l-cytochrome c isroughly twice that seen in tuna. Furthermore, its distance to the heme iron is smaller than thatobserved in any of the other oxidized cytochrome c structures. The second difference is in theside-chain conformation of Asn52. As mentioned before, in oxidized yeast iso-l-cytochrome c,hydrogen bonds from Asn52 ND2 to both the heme and the internal Wat166 are lost. In thecase of tuna, rice and horse cytochromes c, the Asn52 side-chain amide has flipped over 180°in the oxidized state. In these proteins, the ND2 group is still hydrogen bonded to the hemepyrrole A propionate and the nearby internal water is now hydrogen bonded to the OD1 atomMet 80Tyr 67Wat 166Thr 78His 18Asn 52Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 61A^BFigure 3.30: Heme and polypeptide chain structure about the internally bound water moleculeWat 166 in (a) oxidized yeast iso-l-cytochrome c and (b) oxidized horse cytochrome c.Heme-ligand interactions are indicated by thin white bonds, whereas hydrogen bonds are shownby thin black dotted lines. An identical hydrogen bonding network as present in horse cyto-chrome c is also seen in the oxidized structures of rice and tuna.of Asn52. This rearrangement allows the internal water molecule to move closer to the hemeiron without the loss of a hydrogen bond to the side-chain of residue 52. Notably, the retentionof a hydrogen bond to Asn52 does not affect the orientation of the dipole moment of the internalwater molecule (Figure 3.28c), nor does it interfere with the possible role this water moleculehas in modulating the strength of the iron sulfur ligand bond, as discussed below.The Met80 ligand regionThe region about the Met80 heme ligand and Phe82 forms a third focus for oxidation statedependent conformational changes. Coincident with observations showing that oxidized cyto-chrome c has a more open structure and the present study which suggests this is likely localizedon the Met80 side of the heme group, are data indicating a substantive weakening of the Met80Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 62ligand bond in the oxidized state (Dickerson & Timkovich, 1975; Williams et al., 1985b). Onlyin the oxidized state is cytochrome c known to undergo a conformational change at alkaline pHin which this bond is lost (Pearce et al., 1989) and a new ligand is formed by some other groupwhose identity has not yet been established (Gadsby et al., 1987). The presence of a weakerligand bond in the oxidized state is supported by our observation that the side-chain of Met80has an increased overall thermal factor in oxidized yeast iso-1-cytochrome c. This change inthermal factor values is very localized in that neither the heme group average thermal factornor that of the His18 ligand is affected. Also apparent is some displacement of side-chain atomsand a small ,--, 7°change in the rotation bond angle between the Met80 side-chain and the hemeiron atom. However bond weakening does not appear to cause a lengthening of the Met80 SDheme iron bond (Korzun et al., 1982). This observation is supported by the current study,where although this bond distance shows the largest deviation amongst heme iron coordinatebonds (Table 3.8), this deviation is within experimental error (see Appendix C).An essential parameter in the regulation of the relatively high reduction potential of cyto-chrome c is the strong withdrawing power that the Met80 ligand exerts on the heme iron atom(Marchon et al., 1982; Mathews, 1985). Our results and those of others suggest that relativeto the reduced state, in the oxidized protein the Met80 ligand is weakened and exerts less ofan electron withdrawing effect, a situation favoring stabilization of the positively charged hemeiron atom. An important factor in modulating Met80 ligand bond strength is likely to be theinteraction this group has with the side-chain of Tyr67. In reduced yeast iso-l-cytochrome c theMet80 SD sulfur atom is the acceptor in a hydrogen bond to the Tyr67 OH (see Figure 3.28).This hydrogen bond could be expected to increase the electron withdrawing strength of theMet80 ligand and thereby serve to further increase the heme reduction potential. This isclearly illustrated by the -.55 mV drop in reduction potential observed when this hydrogenbond is eliminated by mutation of Tyr67 to a phenylalanine (Luntz et al., 1989; Wallace et al.,1989; see also Section 1.5 and Chapters 4 and 5).As discussed earlier, in the oxidized state it appears that the Met80 SD to Tyr67 OHChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 63hydrogen bond is broken (see Figure 3.28). This change would facilitate stabilization of oxidizedcytochrome c through two mechanisms. First, breakage of this bond would diminish the electronwithdrawing power of the Met80 ligand. Second, the side-chain of Tyr67 could interact to allowthe internal water molecule Wat166 to reorient its dipole moment to maximally stabilize thepositive charge resident on the heme iron atom. It should be stressed here that these featuresare not unique to yeast iso-l-cytochrome c. Model-building of the dipole of the internal watermolecule in other eukaryotic cytochrome c structures provide the same results (see Figure 3.28c).Thus, stabilization of the oxidized state through differential orientation of the dipole momentof the internal water molecule, which not only interacts favorably with the altered electrostaticfield, but also regulates the electron withdrawing power of the Met80 sulfur ligand, appears tobe a universal feature among eukaryotic cytochromes c.Perturbations about the Met80 ligand also appear to be associated with the greater mobilityobserved in the adjacent polypeptide chain segment composed of residues 81-85. Maximalincreases in thermal factors are localized at Phe82, a residue implicated in the electron transfermechanism of cytochrome c (Liang et al., 1987, 1988; Louie et al., 1988b; Louie & Brayer, 1989;Nocek et al., 1991 ). The side-chain of this residue is positioned parallel to the plane of theheme, near its solvent exposed edge and is in van der Waals contact with the side-chain of Met80(see Figure 3.20). It is also positioned in that portion of the cytochrome c surface believed toform the contact face in complexes with electron transfer partners (Salemme, 1976; Lum et al.,1987; Pelletier & Kraut, 1992). A small change in the orientation of the phenyl group of Phe82with respect to the heme plane is observed in the oxidized state of yeast iso-l-cytochrome c. Ithas been postulated that the side-chain of Phe82 may undergo very large motions in complexeswith electron transfer partners (Wendoloski et al., 1987), however this suggestion has not beenproven experimentally (Burch et al., 1990).Another conformational change in this region is the formation of a new hydrogen bondbetween G1y84 0 and Arg38 NH1. Arg13 has also been implicated in being an important com-ponent in electron transfer complexation interactions (Margoliash & Bosshard, 1983; SatterleeChapter 3. Oxidized Yeast Iso-1-Cytochrome c^ 64et al., 1987; Lum et al., 1987). This new hydrogen bond has also been observed in mutantyeast iso-l-cytochrome c structures which mimic the oxidized state (Louie et al., 1988b; Louie& Brayer, 1989). Given its apparent importance to the electron transfer process, it is notunexpected that the region about Phe82 is sensitive to oxidation state changes.3.3.2 Mechanistic implicationsTaken together, our results suggest that there is a concerted and cooperative deployment of in-terconvertible structural changes which facilitate stabilization of the alternative oxidation statesof the heme group of cytochrome c. Based on these observations and the surface presentationof oxidation state dependent conformational changes, it is possible to propose a trigger mecha-nism for electron transfer events mediated by cytochrome c. As illustrated in Figure 3.24, thoseresidues about the three focal points of conformation readjustments are localized to the Met80ligand side of the heme, adjacent to the solvent exposed edge of this group. In terms of the outersurface of cytochrome c, these changes are centrally positioned in that region of the surfacebelieved to form complementary complexation interactions with redox partners (Margoliash &Bosshard, 1983; Poulos & Kraut, 1980; Pelletier & Kraut, 1992). One rather unusual aspect ofthe surface representation of those residues conformationally adjusted according to oxidationstate is the unperturbed nature of residues 73-80 which bisects this region.It is notable that the 73-80 segment in cytochrome c has the most highly conserved sequence,with almost complete absence of natural variance at the amino acids involved (Hampsey et al.,1986; Louie et al., 1988a). The major secondary structural feature formed by this portion ofpolypeptide chain is a type II /3-turn composed of residues 75-78 (Louie & Brayer, 1990). Itis also the side-chain of Thr78 that forms a hydrogen bond to the oxidation state sensitiveWat 166.Given its central location within the putative contact zone with redox partners, the highdegree of sequence conservation, its unique response to oxidation state and the linkage throughThr78 to the three focal points of conformational change, it seems possible that the 73-80Chapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 65segment acts to trigger the alternative structural changes required to stabilize the two oxidationstates of cytochrome c. This peptide segment is appropriately positioned to carry out thistriggering function by virtue of its surface location next to the solvent exposed edge of theheme, a likely route for an electron to travel to and from the heme group. One might think ofresidues 73-80, or perhaps more specifically the 75-78 0-turn, as a push-button contact triggeroperated by protein-protein contacts with redox partners, leading to the generation of thenecessary conditions required to facilitate electron transfer.Precisely how this proposed push-button trigger might function cannot be determined indetail on the basis of our current studies. This arises from only being able to resolve the twoendpoints of the oxidation reaction, the reduced and oxidized cytochrome c structures, notthe intermediary transition state conformation (or possibly conformations). Nonetheless, thecurrent results suggest that the proposed push-button trigger would retain the same spatialpositioning in the two end oxidation states, but once pushed by protein-protein contacts ina complex with a redox partner, would act to promote changes in protein conformation thatwould facilitate electron transfer in either direction. One role of differential mobility in theregion about residues 73-80 could be that of mediating these changes by providing the necessaryconformational flexibility. This proposal is analogous to the type of switching carried out ina conventional electrical circuit having a push-button contact switch that reversibly alters thedirection of current flow.It must be emphasized, as diagrammatically illustrated in Figure 3.31, that the structuraldetails of the intermediary conformational states through which cytochrome c must pass ongoing between alternative oxidation states are missing from our understanding of this process.Also missing is any information on the potential role played by redox partner interactions beyondthe proposed recognition and triggering events. While it seems likely that the intermediarystate is some hybrid of the fully oxidized and reduced cytochrome c structures, it is entirelypossible that redox partner complexation results in transient conformations (of either residues73-80 or elsewhere) that are not evident in the two end states. Such surface transitions haveChapter 3. Oxidized Yeast Iso-l-Cytochrome c^ 66Intermediate OxidizedFigure 3.31: A schematic representation of the proposed push-button mechanism for electrontransfer events in yeast iso-l-cytochrome c. Upon complexation with redox partners residues73-80, or perhaps more specifically 75-78, alter their conformation so as to facilitate electrontransfer and the necessary concomitant structural readjustments. After electron transfer hastaken place and the oxidation state dependent conformational changes have occurred, cyto-chrome c disengages from the complexed redox partner and residues 75-78 return to theiroriginal conformation. The exposed heme edge is shown in black in this figure, whereas areaslabelled A, B, and C correspond to those segments of polypeptide chain which become moremobile in the oxidized state of cytochrome c (see Figure 3.24). Residues 73-80 are representedby the push-button trigger.been suggested from the results of other studies (Koloczek et al., 1987; Weber et al., 1987;Hildebrandt & Stockburger, 1989). Indeed, it could very well be that our minimalist three-state scheme may be too simple and that multiple transition states span the gap betweenoxidized and reduced cytochrome c.Our current results do suggest a number of approaches to investigating the intermediarystate that is proposed to occur during electron transfer. For example, an obvious target forfuture structure-function-mutagenesis studies is the protein segment composed of residues 73-80. Of particular interest here would be the effect of three classes of replacements. Thoseinvolving residues that outwardly face into the surrounding solvent and presumably contact thesurfaces of complexed redox proteins (Lys73, Tyr74, Pro76, G1y77, Lys79), those residues thatcontact Wat166 or line its internal cavity (I1e75, Thr78, Met80) and those residues likely to beinstrumental in setting the folded conformations (Pro76, G1y77). In this latter group, Gly77 isa particularly interesting residue, since it is not only at the third position of the type II 0-turn,Chapter 3. Oxidized Yeast Iso-1-Cytochrome c^ 67but it also adopts 0,0 angles not accessible to other amino acids with side-chains.At this point only limited attention has been directed at the role of the 73-80 segment.The available studies show that mutation of residues in this region leads to destabilization ofcytochrome c and to low biological activity (Boon et al., 1979; Wood et al., 1988; ten Kortenaaret al., 1985; Wallace et al., 1989). These results are in accord with the mechanistic proposalsdiscussed herein.Chapter 4Mutation of Tyrosine-67 in Cytochrome c Significantly Alters The Local HemeEnvironment4.1 Experimental Procedures4.1.1 CrystallizationProtein samples for these studies were obtained from J.G. Guillemette (University of BritishColumbia). Crystals of the Y67F mutant protein in the reduced state were grown in a 0.1 Msodium phosphate buffer (pH 5.8) containing 94% saturated ammonium sulfate and 40 mMDTT, using the hanging drop method and employing a hair seeding technique (Leung et al.,1989). In most instances crystals did not grow large enough using this method so that supple-mentary macro-seeding techniques were used. To obtain protein in the oxidized state, crystalswere soaked for a day in 20 mM ferricyanide. Both the reduced and oxidized mutant proteincrystals were isomorphous with those grown for wild-type yeast iso-l-cytochrome c (Louie &Brayer, 1990; Chapter 3). The space group was P432 1 2 and unit cell dimensions for the reducedand oxidized mutant protein crystals were a = b = 36.46 A, c = 137.71 A and a = b = 36.47 A,c = 139.08 A, respectively (unit cell dimensions for reduced and oxidized wild-type proteincrystals are a = b = 36.46 A, c = 136.86 A and a = b = 36.47 A, c = 137.24 A, respectively).4.1.2 Data collection and data processingDiffraction data was collected on an Enraf-Nonius CAD4-F11 diffractometer with a 36.8 cmcrystal to counter distance and a helium purged path for the diffracted beam. The radiationused was generated from a copper target x-ray tube operating at 26 mA and 40 kV and wasnickel filtered. For each reduced and oxidized protein data collection only one crystal was68Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^69used. A total of 9612 reflections to 1.9 A resolution were measured for the Y67F mutant in thereduced state, and for its oxidized counterpart a total of 9211 reflections to 2.2 A resolutionwere collected. The lower resolution of the oxidized Y67F dataset was due to the smaller crystalsize in this case. The ambient temperature during data collection was maintained at 15°C.The two data sets were converted to structure factors as previously described for the oxi-dized form of yeast iso-l-cytochrome c (see Section 3.2) except for the decay and absorptioncorrections. To correct for decay, the six intensity control reflections periodically measuredduring data collection, were sorted into two groups so that a resolution dependent correctioncould be applied. A similar approach was taken for the absorption correction in that two phicurves were collected at different theta angles and used to estimate the resolution dependenceof the absorption profile (see Section 2.2). After merging of duplicate reflections the reducedY67F data set had 7882 structure factors and the oxidized data set contained 5274 structurefactors.4.1.3 Refinement and analysesBefore restrained parameter refinement (Hendrickson & Konnert, 1981) was initiated, eachstructure factor data set was put on an absolute scale using the method described by Wilson(1942). The starting model used for refinement was the 1.2 A resolution refined structure of re-duced yeast iso-l-cytochrome c (Louie & Brayer, 1990) with Tyr67 replaced by a phenylalanineand Cys102 by a threonine. In total, 64 well defined water molecules present in the wild-typereduced structure were used in the starting model. These included the internal water moleculesWat110, Wat121, and Wat168, but excluded Wat166 which is located adjacent to the mutationsite. Also included was a sulfate ion which is observed in both the reduced and the oxidizedforms of the wild-type protein. Refinement of the reduced mutant protein structure utilized3565 structure factors that had a F/a(F) ratio > 2.0 and were within the resolution range of6.0 to 1.9 A; for the oxidized form 3207 structure factors were employed with a F/a(F) ratio> 1.5 and were in the resolution range 6.0 to 2.2 A.Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^70Table 4.11: Final stereochemistry for reduced and oxidized Y67F yeast iso-l-cytochrome cStereochemicalrefinement parametersr.m.s. deviation fromideal valuesReduced^OxidizedweightingtparametersBond distances (A)1-2 bond distance 0.024 0.020 0.0201-3 bond distance 0.046 0.046 0.0301-4 bond distance 0.061 0.058 0.050Planar restraints (A) 0.017 0.016 0.200Chiral volume (A3 ) 0.223 0.220 0.150Non-bonded contactst (A)single-torsion 0.238 0.225 0.250multi-torsion 0.223 0.222 0.250possible hydrogen bonds 0.237 0.251 0.250Torsion angles (°)planar (0°or 180°) 2.3 2.4 2.5staggered (±60°,180°) 24.8 25.4 19.0t The weighting parameters used in the final cycles of refinement were identical for the reduced andoxidized yeast iso-l-cytochrome c Y67F mutant.t The r.m.s. deviations from ideality for this class of restraint incorporates a reduction of 0.2 A from theradius of each atom involved in a contact.During the course of refinement several manual interventions were carried out for both thereduced and the oxidized mutant structures. Based on Fo — F, difference and fragment maps,and 2F0 —11 and 3F0 — 211 difference maps, a number of side chains of surface residues wererepositioned and several water molecules were added and deleted from the model. At the end ofrefinement the reduced Y67F structure contained 48 water molecules and 1 sulfate ion, while theoxidized structure had 30 water molecules plus the sulfate ion. Each refinement was concludedwhen positional shifts became small (r.m.s. shifts < 0.03 A), indicating that convergence hadbeen reached.The final mutant protein structures display excellent stereochemistry (see Table 4.11) andfinal crystallographic R-factors were 20.1% and 19.6% for the reduced and oxidized structures,respectively. By plotting R-factor versus resolution an estimate can be made of the r.m.s.0.30-0.25-•.1a-0.5 00-0.00.05 0.15^0.20SIN(0) / x0.10 0.25 0.300.20-0.15OChapter 4. Reduced and Oxidized Yeast Iso-1-Cytochrome c Y67F Mutant^71Resolution (A)5.0^3.3^2.5^2.0Figure 4.32: Plots of the dependence on resolution of both the R-factor agreement betweencalculated and observed structure factors (^ — ^), and the fraction of data used (Q — Q, axisat top right), for the final refined structures of reduced (filled symbols) and oxidized (opensymbols) Y67F yeast iso-l-cytochrome c. For this analysis reciprocal space was divided intoshells according to sin(9)/A, with each containing at least 260 reflections. For the purpose ofassessing the accuracy of the atomic coordinates of the mutant structures, curves representingthe theoretical dependence of R-factor on resolution assuming various levels of r.m.s. error inthe atomic positions of the model (Luzzati, 1952) are also drawn (dashed lines). This analysissuggest an overall r.m.s. coordinate error in the range of 0.22 — 0.24 A for both structures.coordinate error in each mutant structure (Luzzati, 1952; see also Appendix C). Inspection ofFigure 4.32 shows this error to be in the range of 0.22-0.24 A for both the reduced and oxidizedstructures. A separate estimate of coordinate errors was made using the method developedby Cruickshank (1949, 1954, 1985). Evaluating Cruickshank's formulas for all protein atomsindicates the overall r.m.s. coordinate errors are 0.21 A and 0.22 A for the reduced and oxidizedY67F mutant proteins. These values are in good agreement with those determined by Luzzati'smethod.Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^72Figure 4.33: A stereo-drawing of the conformations of all main-chain (thick lines) and side-chain(thin lines) atoms in the reduced yeast iso-1-cytochrome c Y67F mutant protein. The hemegroup has been drawn with thick lines. Also drawn are the two heme ligand bonds to His18and Met80, and the two covalent thioether linkages to cysteines 14 and 17. For clarity, every5th a-carbon atom has been labelled with its one-letter amino acid designation and sequencenumber.4.2 Results4.2.1 Polypeptide chain conformationA stereo-drawing of the structure of reduced yeast iso-l-cytochrome c in which Tyr67 has beenreplaced by a phenylalanine residue is shown in Figure 4.33. Note the amino acid numberingused is based on the alignment in Table 1.1. The overall fold of the oxidized form of thisprotein is identical to its reduced counterpart. The related reduced and oxidized structuresof wild-type yeast iso-1-cytochrome c have been previously described and comparable figuresdisplaying these structures are shown in Figures 1.5 and 3.18.As illustrated in Figure 4.34b, a comparison of the reduced forms of the Y67F mutant andwild-type yeast iso-1-cytochrome c reveals little perturbation in the placement of main-chainatoms (overall average deviation is 0.29 A), with the exception of the first two N-terminal5 15 25 35 45 55 65 75 85 95"'""' I^I"'^... ." I^I" .... I.'" nn,5 15 25 35 45 55 65 75 85 95: Y67F(red) vs. WT(red)1.1 ^ It .....^I ,   I ^ I ^ I^I ^ I^I4-C0•-5 15 25 35 45 55 65 75 85 95, ..... ....„ ..... ....,... ..... ,Residue Number5o- 4-- 3-- 2-- 1-0^..... —, ...5 15 25 35 45 55 65 75 85 95Residue Numberleouto 11,mo In do ..... Id ..... am!A: WT(red) vs. WT(ox)5 5- 4--o 0Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^73Figure 4.34: A comparison of the average positional deviations between main-chain (thicklines) and side-chain (thin lines) atoms of the reduced and oxidized forms of both the yeastiso-l-cytochrome c wild-type and Y67F proteins. In (a) wild-type reduced and oxidized arecompared, (b) shows the deviations between reduced Y67F mutant and reduced wild-type, (c)is the comparison between the the oxidized mutant and wild-type structures, and (d) showsthe average deviations between the two oxidation states of the Y67F mutant. The horizontaldashed lines represent the average deviations between all main-chain atoms. The filled darkcircles at position 104 represent the overall average deviation for heme atoms.residues. However, these two amino acids (Thr(-5) and Glu(-4)) have been found to be sub-stantially disordered in all yeast iso-l-cytochrome c structures resolved thus far and thereforethe differences observed are undoubtedly due to differential fits to poor electron density in thisregion rather than the result of the introduced mutation. Between the two reduced proteins,the next two largest deviations occur at Gly41 (0.5 A) and Trp59 (0.5 A). These displacements,however, are less than two times the overall average deviation between all main chain atoms andthus they do not appear to be of marked significance. It is notable that all of the main-chainto main-chain hydrogen bonds present in the wild-type protein are also observed in the Y67Fmutant. A detailed listing of these hydrogen bonds can be found in Table 8 of Louie & BrayerChapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^74(1990).Comparison of the oxidized wild-type and Y67F yeast mutant iso-l-cytochromes c (seeFigure 4.34c) indicates an overall average deviation of 0.35 A for main-chain atoms. Twoamino acids have displacements greater than two times this value. The first is G1y84 whichmoves , 1 A in the oxidized state of the wild-type protein to form a hydrogen bond between itscarbonyl group and the side-chain of Arg13 (see Chapter 3). The formation of this hydrogenbond does not occur in the oxidized state of the Y67F mutant. The second largest positionaldeviation (0.7 A) involves the main-chain atoms of Met80, one of the two residues forming anaxial heme ligand. This residue is immediately adjacent to the mutation site and the observedshift may be related to the incorporation of an additional water molecule in this region of theY67F protein. As with the reduced wild-type and Y67F mutant yeast iso-l-cytochromes c acomparable complement of main-chain to main-chain hydrogen bonds is retained in the twooxidized proteins.A detailed analysis of the conformational changes between the reduced and oxidized formsof the Y67F mutant protein (see Figure 4.34d), reveals an average main-chain displacement of0.34 A. One major difference is observed. In the oxidized Y67F protein, G1u44 shifts 0.9 Asuch that its carbonyl oxygen can form a new hydrogen bond with the side chain of His26. Thenext largest residue shift involves Met80, although this is only slightly greater than two timesthe overall average observed main-chain displacements. One factor in the shift of Met80 maybe the loss of the hydrogen bond from Met80 N to Thr78 0G1 in the reduced Y67F protein(see below). This hydrogen bond, which is present in the oxidized protein, appears to be lostin the reduced form in favor of a new interaction with an additional water molecule bound inthis region.With respect to side-chain conformations between the reduced and oxidized Y67F mutantstructures, many side-chains at solvent exposed locations appear to have altered conformations.As Figure 4.34d shows, these include virtually all lysine (-2, 5, 11, 22, 55, 86, 89, 100) andseveral glutamic acid residues (21, 61, 66), as well as Ser47 and Leu9. However, these apparent5 15 25 35 45 55 65—— 4030— 20--0^75 85 95^5 15 25 35 45 55 65 75 85 95^J.^ I^I^I^ I^ 1.........1"°.- 50s_0▪ 40-0—r~^-0rn 10—•0 .1^' ^I^I^I0O 40-OO 30—.c 20 —9^I^"I^I^I5 15 25 35 45 55 65 75 85 95Residue Number5 15 25 35 45 55 65 75 85 95Residue Number0O 10 -4)<>Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^75Figure 4.35: Plots of the average thermal factors of main-chain atoms along the polypeptidechain in the reduced and oxidized structures of Y67F and wild-type yeast iso-1-cytochrome c.The frames show (a) the wild-type reduced (thick line) and oxidized (thin line) mean thermalfactors, (b) the values for reduced Y67F (thick line) and wild-type (thin line), (c) the oxidizedstructures of Y67F (thick line) and wild-type (thin line), and (d) the values for the reduced(thick line) and oxidized (thin line) Y67F mutant structures.differences likely result from the mobile nature of these side-chains rather than being relatedto oxidation state. A detailed analysis of conformation changes in main-chain and side-chainatoms between reduced and oxidized wild-type iso-1-cytochrome c (Figure 4.34a), can be foundin Chapter 3.An analysis of main-chain thermal factors along the course of the polypeptide chain in thereduced and oxidized Y67F mutant, as well as comparisons to the related wild-type structuresare shown in Figure 4.35. As shown in Figure 4.35a, a number of differences are observedbetween thermal factor profiles in the two oxidation states of the wild-type iso-1-cytochrome c.These have been proposed to play a role in promoting electron transfer and stabilizing theprotein in its alternative oxidation states (see Section 3.3). Between the reduced Y67F andChapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^76wild-type proteins (see Figure 4.35b), thermal factor variability along the polypeptide chainis quite similar. The only notable difference involves a small increase in mobility for residues56-59 in the mutant protein. This change is maximal at Trp59, a residue which has an averagemain-chain thermal factor of 20 A2 in the reduced wild-type protein and 28 A2 in the mutant.As Figure 4.35c shows, the oxidized Y67F structure has a very different thermal factorprofile when compared to the oxidized wild-type protein. Thus, the nature of changes observedfor thermal factors between oxidation states in the wild-type protein and those found in theY67F protein are substantially different (see Figures 4.35a, d). In particular, the oxidizedY67F mutant appears to lack the increased polypeptide chain mobility changes observed inthe oxidized wild-type protein. As Figure 4.35d shows, two portions of polypeptide chain havemarked lower average thermal factor values. These include residues 51-61 and 70-77. Althoughthere is a general trend to lower thermal factor values in the oxidized state of the Y67F protein,two small regions composed of residues 37-39 and 42-46, do exhibit greater mobility. Thus whilethe mutation of Tyr67 to a phenylalanine appears to have minimal impact on polypeptide chainpositioning, it strongly affects the profile of average thermal factors observed, especially in theoxidized state.4.2.2 Heme structureThe heme groups of the reduced and oxidized Y67F mutant are distorted from planarityinto a saddle shape (see Table 4.12). This distortion is more pronounced for both mutant formsin comparison to that of reduced wild-type protein, but comparable to that seen in the caseof the oxidized wild-type protein. Thus the increase in heme planar distortion observed uponoxidation of wild-type iso-l-cytochrome c (Section 3.2) is not seen between the two oxidationstates of the mutant protein. The average angular deviations of pyrrole rings from the pyrroleN plane are 11.5° and 11.6° for reduced and oxidized Y67F respectively, and 9.4° and 12.2° forthe reduced and oxidized forms of the wild-type protein.As Table 4.12 shows heme iron coordinate bond lengths are comparable in both oxidationChapter 4. Reduced and Oxidized Yeast Iso-1-Cytochrome c Y67F Mutant^77Table 4.12: Heme geometry of wild-type and Y67F mutant proteinsReduced Reduced Oxidized OxidizedWild-type Y67F^Wild-type^Y67Fa) Angular deviations (°) between the pyrrole nitrogen planenormal and the four individual pyrrole ring plane normalsand the heme coordinate bonds.A 9.4 13.7 12.6 13.4B 11.1 9.1 14.1 10.7C 8.8 11.7 9.6 12.0D 8.1 11.3 12.8 10.2Fe - His18 NE2 2.1 3.9 7.2 1.9Fe - Met80 SD 4.9 3.9 3.3 6.3b) Angular deviations (°) between the porphyrin ring planenormal and the four pyrrole ring normals, the pyrrole nitro-gen plane normal and the heme coordinate bonds.A 6.7 9.0 8.3 7.4B 11.9 12.3 12.4 12.4C 9.8 11.6 13.8 10.8D 6.0 6.4 10.0 4.4NNNN 2.6 4.9 4.3 6.0Fe - His18 NE2 3.3 5.1 6.5 5.8Fe - Met80 SD 7.5 6.9 5.4 7.6c) Heme iron ligand bond distances (A).His18 NE2 1.99 1.96 2.01 2.05Met80 SD 2.35 2.36 2.43 2.44Heme NA 1.97 2.03 1.97 2.01Heme NB 2.00 1.98 1.98 2.01Heme NC 1.99 1.97 2.02 2.01Heme ND 2.00 2.03 2.05 2.02Each pyrrole ring is defined by nine atoms, which include the five ring atoms plus the firstcarbon atom bonded to each ring carbon. The porphyrin ring plane is defined by the five atomsin each of the four pyrrole rings, the four bridging methine carbons, the first carbon atom ofeach of the eight side chains and the heme iron (33 atoms in total). The pyrrole nitrogen planeis defined by only the 4 pyrrole nitrogens. Heme atom nomenclature used in this table andelsewhere follows that of the Protein Data Bank as illustrated in Figure 1.2.Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^78states, and in both wild-type and mutant proteins. In contrast further analysis shows the anglebetween the pyrrole bonds to the heme and the imidazole ring plane of the heme ligand His18appears to be dependent upon oxidation state. This effect was previously noted in the wild-typeprotein (Section 3.2). In reduced Y67F protein, the His18 ring plane and a vector throughthe NA and NC heme atoms have a relative angle of 44.8° (in reduced wild-type this value is46.7°). In the oxidation form the plane of the His18 side-chain is rotated , 10° towards the Aand C pyrrole rings of the heme, making an angle of 54.6°. A similar rotation is observed inthe oxidized wild-type protein where this angle becomes 55.8°.A comparison of the Met80 ligand in the reduced wild-type and Y67F mutant proteinsshows the structural differences present are minimal, even though this residue is adjacent tothe mutation site and has substantially altered interactions with nearby residues in the mutantprotein. Particularly notable is the loss of the hydrogen bond from the Met80 SD sulfur atomto the hydroxyl group of Tyr67 present in the wild-type protein and its substitution for ahydrogen bond to Wat300 in the Y67F mutant. Also the hydrogen bond between Met80 Nand Thr78 0G1 is absent in the mutant structure. In contrast a comparison of the oxidizedwild-type and Y67F mutant proteins shows that more substantial conformational differencesare present. The average positional shift of Met80 side-chain atoms is rs ,0.8 A (compared with,0.3 A between the reduced proteins) with the terminal Met80 CE methyl group showingthe largest displacement (r.1.8 A). In particular there is a ,40° rotation of this methyl groupabout the ligand bond. As will be discussed herein, this conformation may be the result of thedisordering of water molecules in the nearby internal protein cavity. In the oxidized wild-typeprotein a well ordered water molecule (Wat166) is located adjacent to Met80 and believed tostabilize this conformation of the protein (Chapter 3).An analysis of the solvent accessibility of the heme group in the two oxidation states of boththe wild-type and mutant iso-1-cytochromes c is presented in Table 4.13. This data shows thatheme solvent exposure is not dependent upon the oxidation state of either protein. However, theaverage heme exposure to solvent in the two states of the mutant protein ( , 7.7%) is somewhatChapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^79Table 4.13: Heme solvent accessibility of wild-type and Y67F mutant proteinsReducedWild-typeReducedY67FOxidizedWild-typeOxidizedY67F1. Solvent accessible heme atoms andsurface area exposed (A 2 )CHD 2.6 0.0 0.0 0.0CMC 9.4 9.9 9.5 9.0CAC 3.3 0.0 5.1 2.9CBC 17.4 22.1 19.2 15.8CMD 9.8 7.5 11.7 9.42. Total heme exposure (A 2 ) 42.5 39.5 45.5 37.13. Total heme surface (A 2 ) 495.7 500.0 503.3 499.94. Heme surface area exposed (% ) 8.6 7.9 9.0 7.4Computations were done using the method of Connolly (1983) and the results represent theaccessible molecular surfaces of the atoms listed. The probe sphere used had a radius of 1.4 A.smaller than in the wild-type protein ( ,8.8%).Another factor which is clearly dependent on whether the protein is in the reduced oroxidized state is the positioning of the pyrrole A propionate group (Figure 4.36). The confor-mation of this propionate in the Y67F reduced structure is essentially identical to that seen inthe wild-type reduced form. In oxidized wild-type protein, the pyrrole A propionate exhibitsan alternate conformation where the three bonds C2A-CAA, CAA-CBA, and CBA-CGA haverotated by ,20°, ,30°, and ,45°, respectively (Section 3.2). In the Y67F oxidized proteinthese same bonds have changed by ,20°, ,25°, and ,35°, respectively, resulting in a virtuallyidentical positioning of the propionate compared to that seen in oxidized wild-type protein. Asin the wild-type protein, oxidation of the mutant cytochrome c also results in the movement ofan internal water molecule (Wat121) such that it can form a bifurcated hydrogen bond to bothoxygens of the propionate (Section 3.2).The hydrogen bond network around the pyrrole A propionate in the two mutant structuresChapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^80Figure 4.36: Drawings of the pyrrole A propionate group region as observed in (a) reduced and(b) oxidized wild-type yeast iso-1-cytochrome c, and in the (c) reduced and (d) oxidized Y67Fmutant protein. The pyrrole A propionate group is highlighted with dark shaded balls. Thetwo internally bound water molecules, Wat121 and Wat168, which mediate the interaction ofArg38 with this heme propionate, are shown as larger spheres. Hydrogen bonds are indicatedby thin dashed lines.Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^81Table 4.14: Heme propionate hydrogen bond interactions in wild-type and Y67F mutant yeastiso-l-cytochrome cHydrogen bondpartnersDistances (A)ReducedWild-typeReducedY67FOxidizedWild-typeOxidizedY67F01A Tyr48 OH 2.83 2.97 2.83 3.51Wat121 2.81 2.57 2.85 2.69Wat168 2.85 3.41 2.87 (3.62)02A G1y41 N 3.21 2.82 2.60 2.88Asn52 ND2 3.34 2.67 (3.54) 3.41Trp59 NE1 3.09 3.21 3.43 2.89Wat121 (4.01) (3.74) 3.34 3.51OlD Thr49 OG1 2.64 2.60 2.79 2.71Thr78 OG1 2.90 2.75 3.07 3.08Lys79 N 3.17 (3.39) 2.67 2.9602D Thr49 N 2.94 2.77 2.75 2.95Interactions were accepted as hydrogen bonds only if they met all of the following criteria: aII- • -A distance < 2.6 A, a D-H• • •A angle > 120°, and a C-A• • •1 angle > 90°. Valuesgiven in brackets are not considered to be hydrogen bonds by this criteria, but are listed forcomparison.closely resembles that of the reduced and oxidized wild-type structures (see Table 4.14). How-ever, a few differences in this hydrogen bond network are observed (see Figure 4.36). In thereduced Y67F mutant structure Wat121 is shifted by ,-,, 1 A and as a result the hydrogen bondbetween Wat121 and Arg38 NE, as well as the hydrogen bond between this water molecule andHis39 0 are absent. Because of the absence of this latter interaction, the His39 CA moves by0.25 A and this shift is propagated to the side chain (shift in CB is 0.50 A), culminating in a30° rotation of the imidazole ring. In addition to these differences the hydrogen bond betweenTrp59 and the propionate A carboxyl group is weakened as judged by the increased hydrogenbond length (see Table 4.14) and the increased thermal factors for the tryptophan side-chain(average increase +15 A2).Chapter 4. Reduced and Oxidized Yeast Iso-I-Cytochrome c Y67F Mutant^82In the oxidized mutant structure the interaction of Arg38 with the propionate, which ismediated through Wat121 and Wat168, is severely weakened as indicated by the absence of theWat121 - Arg38 NE and Wat168 - heme 01A hydrogen bonds, as well as the increase in thermalfactors for residues 37-39 (see Figure 4.35). This is in sharp contrast to the wild-type proteinwhere the interaction between Arg38 and the pyrrole A propionate becomes stronger in theoxidized state as deduced from the observed thermal factors (see Chapter 3). In addition, thehydrogen bond interaction between Asn52 ND2 and the propionate is not lost on going fromthe reduced to the oxidized state in the Y67F mutant as it is in the wild-type protein. Thisis evident not only from the distance between these two groups, but also from the absence ofan increase in thermal factors for the Asn52 side-chain in the mutant protein. In the wild-typeprotein the Asn52 side-chain becomes substantially more mobile in the oxidized state (averageincrease +16 A 2 ).4.2.3 Mutation site: the reduced stateIn the reduced state of wild-type yeast iso-1-cytochrome c, the side chain hydroxyl group ofTyr67 is involved in an extensive hydrogen bonding network (Chapter 3). As Figure 4.37ashows, this involves the nearby side-chains of Asn52, Thr78 and Met80, as well as an internallybound water molecule Wat166. This water molecule is a conserved structural feature foundin all eukaryotic cytochromes c for which three-dimensional structures have been determined(Bushnell et al., 1990; Brayer & Murphy, 1993). The exchange of tyrosine for phenylalanine atposition 67 leads to the creation of an internal cavity due to the absence of the tyrosyl hydroxylgroup. As Figure 4.37c illustrates, this extra internal volume is filled by an additional watermolecule (Wat300) positioned in a location comparable to that of the original hydroxyl group.However, since Wat300 is not covalently bound to Phe67 it is further displaced from this side-chain than a tyrosyl hydroxyl group. In order to accommodate Wat300, a shift in position ofWat166 is required and this group moves downwards ( ,--1 A) towards the side-chain of Asn52.Thus an important consequence of the tyrosine to phenylalanine mutation at position 67 is toChapter 4. Reduced and Oxidized Yeast Iso-I-Cytochrome c Y67F Mutant^83Figure 4.37: Drawings showing the region about residue 67 in the (a) reduced and (b) oxidizedwild-type and, (c) reduced and (d) oxidized Y67F proteins. In each case the heme group isshown with dark shaded balls and internal water molecules are represented by larger spheres.Heme ligand interactions are indicated by thin white bonds whereas hydrogen bonds are shownby thin dashed lines.Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^84Table 4.15: Hydrogen bond interactions at the mutation site in the wild-type and Y67F proteinsHydrogen bond partnersDistances (A)ReducedWild-typeReducedY67FOxidizedWild-typeOxidizedY67FAsn52 ND2 Heme 02A 3.34 2.67 (3.53) 3.41Asn52 ND2 Wat166 3.14 3.15 (4.26)Tyr67 OH/Wat300t Thr78 OG1 (4.17) 2.94 (4.42)Tyr67 OH/Wat300t Met80 SD 3.25 3.22 (3.12)Tyr67 011/Wat300t Wat166 2.62 2.87 2.63Thr78 OG1 Heme 02D 2.90 2.75 3.07 3.08Thr78 OG1 Wat166t 2.72 3.25 2.70 -Met80 N Thr78 OG1 3.32 (3.69) 2.73 2.81Interactions were accepted as hydrogen bonds only if they met all of the following criteria: aH• • •A distance < 2.6 A, a D—H• • •A angle > 120°, and a C—A• • •H angle > 90°. Distancesgiven in brackets are not considered to be hydrogen bonds by this criteria, but are listed forthe sake of comparison.t In the reduced Y67F mutant protein, the hydroxyl group of Tyr67 is replaced by a newwater molecule (Wat300) positioned in approximately the same location. In the oxidized Y67Fstructure, neither Wat300 or Wat166 can be definitively located and both are likely undergoinglarge dynamic motions in the enlarged internal protein cavity available.increase the number of water molecules in the nearby internal cavity from one to two.Surprisingly, the presence of Wat300 leads to relatively little change in the orientationof nearby side-chains. The ring atoms of the side-chain of Phe67 show no significant shiftsin position (average deviation 0.1 A), average thermal factors (average decrease ,-1 A 2 ) or inoverall ring orientation (angular deviation ,1°), when compared to its tyrosyl counterpart in thewild-type protein. The largest shift occurs in the side-chain of Asn52 (average deviation 0.6 A),which moves to accommodate the displaced Wat166. Movement of Asn52 is accomplished forthe most part by a 15° rotation in the torsional angle between its CB and CG atoms.Insertion of Wat300 does result in a substantial reorganization of the hydrogen bond networkin this region of the protein (Table 4.15, Figures 4.37a,c). The most prominent change is theloss of the Tyr67 OH to Met80 SD hydrogen bond, which is replaced by a new interactionChapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^85involving Wat300. This latter group is in turn hydrogen bonded to Wat166 and the side-chainof Thr78. Note that in the wild-type protein the hydroxyl group of Tyr67 is too far removedfrom the OG1 of Thr78 to form a similar interaction. The positioning of the side-chain ofThr78 is largely unaffected by these changes, although a loss of the hydrogen bond betweenthis side-chain and the main-chain amide group of Met80 is observed.4.2.4 Mutation site: the oxidized stateNeither Wat300 nor Wat166 can be resolved in the Y67F oxidized protein structure, presumablybecause they are undergoing greater dynamic motion than in the reduced mutant protein wherethe thermal factor of these water molecules are 30 A 2 and 27 A2 , respectively. This increasedmobility likely results from two factors, both of which involve the side chain of Asn52. AsFigure 4.37d shows, in the oxidized state this side-chain shifts further away from the internalcavity (average deviation 0.9 A) and is unlikely to be in a position to form a hydrogen bondto an internal water molecule. Furthermore, the shift of this side chain is primarily responsiblefor opening up the size of the internal cavity containing Wat300 and Wat166 from a volume of35 A3 in the reduced mutant protein, to 60 A3 in the oxidized state. This increased volumewould allow considerably more freedom of movement to both water molecules.Despite the increased size of the cavity, the shifts in side-chain atoms of residues lining themutation site are small between the reduced and oxidized Y67F proteins, with the exceptionof Asn52. Further comparisons show side-chain differences between the oxidized wild-type andY67F cytochromes c are also small. Again, the largest deviation between the two oxidizedproteins involves Asn52 (average deviation 0.9 A). Although in each case the general directionof shifts is similar, the two Asn52 side-chains do adopt different local conformations. In themutant protein the Asn52 ND2 group can still hydrogen bond to the pyrrole A propionate inthe oxidized state, while in the wild-type structure this linkage is broken. One further differencebetween oxidation states of the Y67F mutant protein is the presence of a Thr78 OG1 to Met80 Nhydrogen bond in the oxidized state (see Table 4.15 and Figure 4.37c). This hydrogen bond isChapter 4. Reduced and Oxidized Yeast Iso-I-Cytochrome c Y67F Mutant^86observed in both the reduced and oxidized states of wild-type yeast iso-l-cytochrome c.In contrast to the oxidized wild-type protein, the side-chains of Asn52, Phe67 and Met80 allhave lower average thermal factors in the oxidized Y67F protein. This is particularly prominentfor the side chain of Phe67 for which the average thermal factor is reduced by .'23 A2 . Thisobservation coincides with other results (Figure 4.35) indicating that the polypeptide chain ofthe oxidized Y67F mutant protein does not exhibit the regional increases in mobility observedfor the oxidized wild-type protein.4.3 Discussion4.3.1 Conformational effects of the Y67F mutationShown in Table 4.16 is an overview of the structural differences observed between the reducedand oxidized Y67F yeast iso-1-cytochrome c mutant structures and their wild-type counterparts.The conformational differences observed are localized for the most part in two regions, themutation site and the pyrrole A propionate region (see Figures 4.36 and 4.37). These two areashave previously been shown to be sensitive to the oxidation state of cytochrome c (Chapter 3).Mutation siteIt has been proposed that by mutating tyrosine 67 to a phenylalanine, and thereby removingthe hydroxyl group from this residue, the heme environment would become more hydrophobic(Koul et al., 1979; ten Kortenaar et al., 1985; Wallace et al., 1989). It was also suggested thatthis mutation would affect the internal water molecule Wat166 which is located adjacent to themutation site and is important in mediating oxidation state dependent conformational changes(Chapter 3). Indeed, Luntz et al. (1989) proposed that the Y67F mutation would excludeWat166 from the structure of cytochrome c. This proposal was based on NMR experimentsand stability studies, and referred to work performed by Rashin et al. (1986) which indicatedthat a water molecule buried inside a protein would be energetically unfavorable unless it couldform at least three hydrogen bonds. In this line of reasoning, the loss of a hydrogen bondChapter 4. Reduced and Oxidized Yeast Iso-1-Cytochrome c Y67F Mutant^87Table 4.16: Structural differences observed in Y67F mutant structures when compared to theirwild-type yeast iso-1-cytochrome c counterpartsReduced Y67F Mutant^ Oxidized Y67F MutantA. Positional displacements of polypeptide chain (see Figure 4.34)1. No significant differences in main-chain^1. Movement of Gly84 seen in oxidized wild-placement observed^ type is not observed2. Small shift in Met80 main-chain atomsB. Thermal factor parameters of main-chain atoms (see Figure 4.35)1. Small increase in flexibility for residues56-59, focussed at Trp591. Regional increases in flexibility seen inwild-type are not observed2. Increase of thermal factors observed forresidues 37-39 and 42-463. Substantially lower thermal factors for theside-chains of residues lining the internalwater cavityC. Heme structure (see Figure 4.36 and Tables 4.12 and 4. 14)1. Heme plane more distorted2. Pyrrole A propionate hydrogen bond net-work weakened1. Pyrrole A propionate to Arg38 interactionis lostD. Mutation site (see Figure 4.37 and Table 4.15).1. Additional water molecule (Wat300)present in internal cavity2. Realignment of hydrogen bonds1. Internal cavity size increased2. Large dynamic motion of Wat166 andWat3003. Movement of Asn52 results in interactionwith pyrrole A propionateE. Hydrogen bond interactions (see Figures 4.36 and 4.37 and Tables 4.14 and .1.15).Weaker: Trp59 NE1 - Heme 02ALost:^Tyr670H- Met80 SDMet80 N - Thr78 OG1Arg38 NE - Wat121Wat121- His39 0New:^Wat300- Thr78 OG1Wat300- Met80 SDWat300- Wat166Stronger: Trp59 NE1 - Heme 02ALost:^Arg13 NH 1- Gly84 0Arg38 NE- Wat121Wat168 - Heme 01AWat166, Wat300tNew:^Asn52 ND2- Heme 02At The positions of Wat166 and Wat300 cannot be defined in the oxidized Y67F mutant structure andtheir hydrogen bonding interactions are correspondingly unknown.Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^88to Wat166 in the Y67F mutant protein would be expected to result in the loss of this watermolecule.Our results show that Wat166 is retained in the Y67F mutant cytochrome c. Further, thehydrophobicity of this region is not increased and may even be decreased due to the presenceof an additional water molecule (Wat300) which is found to occupy a position very similar tothe hydroxyl group of Tyr67 in the wild-type protein (see Figure 4.37). In the reduced statethese two water molecules are part of an extensive hydrogen bond network in this region (seeTable 4.15) and each forms three hydrogen bonds, meeting the minimal requirements suggestedby Rashin et al. (1986). Therefore this work has clearly shown the interpretation presented byLuntz et al. (1989) to be incorrect.In the oxidized wild-type structure Wat166 reorients its dipole so as to stabilize the positivecharge residing on the heme iron atom (see Figure 3.28). In this process several hydrogen bondsinvolving Asn52 ND2 and Tyr67 OH are broken or realigned. Furthermore, three segments ofpolypeptide chain located on the Met80 side of the heme group become more mobile (Chap-ter 3). This increase in flexibility and the resulting decrease in stability of the oxidized stateis a well documented property of eukaryotic cytochromes c (see Section 1.3). In the reducedY67F mutant protein the normally resident hydrogen bond network that mediates oxidationstate dependent shifts is substantially modified. In addition, the ordered hydrogen bond inter-actions observed in the oxidized state of the wild-type protein are absent in the Y67F variant.One important consequence of these changes is the absence of regional increases in thermalfactors for specific segments of polypeptide chain in the oxidized state of the Y67F mutant (seeFigure 4.35). In fact, the segments involved (residues 47-59, 65-72 and 81-85), as well as theside-chains in the vicinity of Wat166 that show large thermal factor increases in the oxidizedform of the wild-type protein retain comparable thermal factors in both oxidation states. Anadditional factor that may contribute to this effect is that in the oxidized state of the mutantprotein a hydrogen bond between Asn52 and the pyrrole A propionate is maintained whichcould serve to prevent the increased flexibility in the region of residues 47-59 observed in theChapter 4. Reduced and Oxidized Yeast Iso-1-Cytochrome c Y67F Mutant^89oxidized wild-type structure (see Figure 4.35). These data correlate well with observations madeby other investigators that the oxidized state of the Y67F variant cytochrome c is more stablethan wild-type (Luntz et al., 1989; Wallace et al., 1989; Margoliash, 1990; Frauenhoff & Scott,1992). This work suggests that the hydrogen bond network about Wat166 in wild-type cyto-chrome c is critical for modulating the flexibility of nearby polypeptide chain segments betweenoxidation states, a feature that disappears when this hydrogen bond network is disrupted.Our results show the two internal water molecules, Wat166 and Wat300, to have greaterdynamic motion in the oxidized state of the Y67F mutant protein. Two factors that may explainthis behaviour are the increased size of the internal cavity available in the oxidized state dueto the shift of Asn52, and the loss of the hydrogen bond between Wat166 and Asn52 ND2.Another potential factor is related to the positive charge residing on the heme iron atom inthe oxidized form of the protein. In this state the dipole moments of Wat166 and Wat300 willbe strongly influenced to realign in response to the electrostatic field created by the positivelycharged iron atom. Model-building studies suggest that in this process, several hydrogen bondsobserved in the reduced state of the protein will be broken. Thus in the oxidized state neitherWat166 nor Wat300 are likely to form discrete hydrogen bonds with protein groups, a factorundoubtedly contributing to the observed increase in mobility of these water molecules.Pyrrole A propionate regionComparison of the reduced forms of the Y67F mutant and wild-type yeast iso-l-cytochromes cshows that a change in the nature of the hydrogen bond network around the pyrrole A propi-onate group has occurred (see Figure 4.36 and Table 4.14). The observed differences involve theinvariant residues Trp59 and Arg38, and the internal water molecule Wat121, which is presentin all high resolution structures of eukaryotic cytochromes c determined thus far (Brayer &Murphy, 1993). As in the reduced state, the hydrogen bond network of the Y67F protein isalso perturbed in the oxidized state relative to the wild-type structure. These changes involvenot only those observed between the reduced proteins, but in addition Asn52, and a secondChapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^90internal water molecule Wat168. Surprisingly, despite these observed differences the pyrrole Apropionate group of the Y67F protein undergoes an oxidation state dependent conformationalchange which is virtually identical to that seen in wild-type yeast iso-l-cytochrome c (Fig-ure 4.36; Chapter 3).Our studies suggest that one consequence of the Y67F mutation is that the charge distri-bution on the pyrrole A propionate carboxyl group is altered, thereby affecting its hydrogenbonding characteristics. A possible source for this perturbation is that the introduced muta-tion results in modification of the delocalized ir-electron system on the porphyrin ring. Thatdelocalized electrons in the porphyrin ring can have an effect on pyrrole A propionate functionhas been proposed by Moore (1983). Analysis of the electronic spectra of the Y67F proteinclearly suggests that the conjugated 7r-electron system has been disturbed (J.G. Guillemette,private communication). UV-visible spectra of horse cytochrome c in which the same muta-tion has been introduced suggests a similar effect (Wallace et al., 1989). Further evidence forthis relationship can been seen in a structural study of reduced F82S yeast iso-l-cytochrome c(Louie et al., 1988b). In this mutant protein the 7r-electron system of the heme is affected byintroduction of a solvent channel adjacent to the heme plane. The resultant conformationalchanges about the pyrrole A propionate are similar to those observed upon change in oxidationstate in the Y67F mutant and wild-type yeast iso-l-cytochromes c.4.3.2 Effect of the Y67F mutation on midpoint reduction potentialAs shown in Table 1.5 mutation of Tyr67 to a phenylalanine in yeast iso-l-cytochrome c causesa drop of r.56 mV in midpoint reduction potential. This same mutation shows a comparablelowering of midpoint reduction potential in both horse and rat cytochromes c (Wallace et al.,1989; Luntz et al., 1989; Frauenhoff & Scott, 1992). Some confusion as to the source ofthis effect is evident in the literature. Conventional theories with regard to the factors thatcontrol reduction potential would predict that the Y67F mutant should have a higher midpointreduction potential since it would increase the hydrophobicity of the heme pocket (Kassner,Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^911972, 1973; see also Section 1.4). To reconcile this apparent paradox between theory andexperiment, Wallace et al. (1989) proposed that this mutation must cause a dramatic change inthe electronic structure of the heme group. It should be noted that an increase in heme solventexposure is not responsible for the lower reduction potential of the Y67F mutant as has beensuggested by Stellwagen (1978). As Table 4.13 shows, the wild-type and mutant proteins havecomparable heme solvent exposures.Our studies provide an alternative explanation for the lower midpoint reduction potentialobserved. As discussed in Chapter 3, the hydrogen bond between the hydroxyl group of Tyr67and the sulfur atom of the Met80 ligand appears to be present or absent depending on theoxidation state of the protein (see Figures 3.28 and 4.37). By modulating this hydrogen bond,stabilization of both oxidation states can be achieved. According to this view, the hydrogenbond between Tyr67 OH and Met80 SD is also an important determinant in setting the value ofthe midpoint reduction potential by virtue of increasing the electron withdrawing power of theMet80 ligand. The electron withdrawing strength of the Met80 ligand is recognized as a centraldeterminant in setting the reduction potential of cytochrome c (Marchon et al., 1982; Moore &Pettigrew, 1990). Thus, weakening or removal of the Tyr67 OH to Met80 SD hydrogen bondwould be expected to lead to a decrease in the midpoint reduction potential. This is preciselythe effect observed in the Y67F mutant protein.One factor that might affect the observed value for the reduction potential in the Y67Fmutant protein is the presence of Wat300 which fills the space formerly occupied by the hydroxylgroup of Tyr67. However, this effect is likely to be limited by the greater mobility and less idealpositioning of Wat300, as well as the absence of a conjugated 7r-electron system from the phenylgroup of residue 67 which would serve to further stabilize a hydrogen bond to the Met80 SDsulfur atom as is found in the wild-type protein. This conclusion is supported by mutationsintroduced into cytochrome c which eliminate both the interaction between Tyr67 and Met80and all water molecules from the nearby internal cavity (Chapter 5). In these mutant proteins avalue for the midpoint reduction potential comparable to that of the Y67F variant is observed.Chapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^924.3.3 Role of tyrosine 67 in eukaryotic cytochromes cOur structural studies suggest a dual role for tyrosine 67 in cytochrome c. One role is to serveas part of an oxidation state dependent hydrogen bond network that stabilizes the alternativeoxidation states of cytochrome c by optimally positioning and orienting the dipole moment ofWat166. As part of this process Tyr67 participates in modifying the local flexibility of a nearbypolypeptide chain segment (residues 65-72) depending on the oxidation state of the protein.A second role for Tyr67 is in setting the value of the midpoint reduction potential througha hydrogen bond interaction with the sulfur atom of the Met80 ligand. In the Y67F mutantprotein both these structural features are affected, resulting in a substantially lower midpointreduction potential and a structurally more rigid and stable form of the oxidized protein.Some questions remain as to how important these factors are to biological electron transferactivity. On one hand, the high sequence conservation of Tyr67 would suggest it is essential forbiological function. In a compilation of 96 eukaryotic cytochrome c sequences, only the proteinfrom Euglina gracilis does not have a tyrosine at position 67 (Moore & Pettigrew, 1990). In thisorganism residue 67 is a phenylalanine instead. On the other hand, the yeast strains used in thisstudy and those which possess the Y67F rat cytochrome c mutant do not exhibit diminishedgrowth rates. In fact, it has been reported that these actually grow at a faster rate than thenormal wild-type strain (Luntz et al., 1989; Margoliash, 1990).These observations are rather surprising, especially in the light of the conserved presenceof Tyr67 and the properties of the protein with which it is associated. For example, midpointreduction potential is an extremely conserved property exhibiting limited variation among eu-karyotic cytochromes c (+20 mV, with the exception of the Euglina gracilis protein; Pettigrew& Moore, 1987). In the absence of other data one would expect the ,56 mV drop in mid-point reduction potential of the Y67F mutant protein to have a clear impact on the biologicalfunction of cytochrome c. However, as shown by McLendon and co-workers (Komar-Panicucciet al., 1992) a drop in midpoint reduction potential by as much as 120 mV does not prohibitin vivo respiration. Similar questions remain as to the importance of Tyr67 modulation ofChapter 4. Reduced and Oxidized Yeast Iso-l-Cytochrome c Y67F Mutant^93nearby polypeptide chain flexibility and in assisting the reorientation of Wat166 in response tooxidation state. While our results serve to show some of the roles played by Tyr67, they cannotprovide a clear indication of the evolutionary pressures that have lead to the present structureof cytochrome c, nor the allowable variance in these features that will maintain sufficientlyefficient biological electron transfer.Chapter 5Perturbation of a Conserved Internal Water Molecule and its AssociatedHydrogen Bond Network in Cytochrome c by the Mutations N52I and N52I-Y67F5.1 Experimental Procedures5.1.1 CrystallizationProtein samples of the N52I variant of yeast iso-l-cytochrome c were provided by F. Sherman(University of Rochester) and used for analysis of the reduced state of the protein. For workwith the oxidized protein the additional mutation of C102T was introduced to prevent protein-protein dimerization (Cutler et al., 1987). This protein was provided by J.G. Guillemette(University of British Columbia). For study of the N52I-Y67F protein the C102T mutationwas present in both oxidation states. These two latter proteins were also provided by J.G.Guillemette.Crystals of the reduced N52I and N52I-Y67F mutant proteins were grown in a 0.1 M sodiumphosphate buffer (pH 6.0-6.4) containing 86-94% saturated ammonium sulfate, using a com-bination of the hanging drop method and a hair-seeding technique (Leung et al., 1989). Inorder to maintain the reduced state of the protein during crystallization either 20 mM DTT or70 mM sodium dithionite was added to the crystallization buffer. To obtain crystals in whichthe protein was in the oxidized state, crystals were soaked for one hour in mother-liquor contain-ing 20-40 mM potassium ferricyanide prior to mounting (see Section 2.1). All mutant proteincrystals proved to be isomorphous with those grown for wild-type yeast iso-l-cytochrome c (seeTable 5.17).94Chapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^ 95Table 5.17: Data collection statistics for the N52I and N52I-Y67F yeast iso-l-cytochromes cN52IReduced^OxidizedN52I-Y67FReduced^OxidizedCell dimensions (1)fa, b 36.46 36.47 36.56 36.52c 137.82 137.02 137.53 139.12No. of reflections collected 8449 27230 33491 12907No. of unique reflections 7910 5790 5700 4771Merging R-factor (%)t 13.4 8.4 7.0 6.3Resolution (A) 1.9 2.0 2.05 2.0t The unit cell dimensions for the reduced and oxidized wild-type proteins are a = b = 36.46 Aand c = 136.86 A, and a = b = 36.47 A and c = 137.24 A, respectively. The space group isP432 1 2 for all proteins.t The merging R-factor is defined as: R merge  7' I/ thk1 --7hkil • The calculation of the1=0 It hk1merging R includes all reflections measured more than once (i.e. duplicates and symmetrymates). For the reduced N52I data set which was collected on a diffractometer this value ishigher due to the limited number of repeat measurements. The merging R-factor between thereduced and oxidized data sets of the N52I and N52I-Y67F variants was 12.4% in both cases.5.1.2 Data collection and data processingDiffraction data for the reduced state of the yeast iso-l-cytochrome c N52I mutant was collectedfrom one crystal on an Enraf-Nonius CAD4-F11 diffractometer, with a 36.8 cm crystal tocounter distance and a helium purged path for the diffracted beam. The radiation used wasnickel filtered and generated from a copper target X-ray tube operating at 26 mA and 40 kV.Data collection and processing methodology for this cytochrome c variant were identical to thatdescribed for the oxidized structure of wild-type yeast iso-l-cytochrome c (see Section 2.2 and3.1).Diffraction data for the oxidized form of the N52I mutant protein, and the reduced andoxidized N52I-Y67F mutant proteins, were collected on a Rigaku R-Axis II imaging plate areadetector system. The incident radiation consisted of CuK, X-rays from a rotating anodeChapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^ 96generator operating at 80-90 mA and 50-60 kV. Individual data collection frames were exposedfor 20-25 minutes. Crystals were oscillated through a 0 angle of 1.0° for each frame. For eachof these three mutant proteins only one crystal was used to collect diffraction data. Processingof area detector diffraction data sets was done as described in Section 2.2. All diffraction datasets were put on an absolute scale using the Wilson plot statistical method (Wilson, 1942).Data collection and processing statistics are shown in Table 5.17.5.1.3 Refinement and analysesInspection of mutant protein minus reduced wild-type yeast iso-l-cytochrome c difference elec-tron density maps revealed that structural changes were restricted to those regions directlyaround mutation sites. Therefore, least-squares restrained parameter refinement (Hendrickson& Konnert, 1981) was initiated using as a starting model the reduced wild-type yeast iso-l-cytochrome c structure in which mutated residues were represented by alanines. Included ineach model was a selection of well-determined water molecules and a sulfate ion. Excluded fromstarting models was Wat166 which is located adjacent to the side-chains of residues 52 and 67.Refinement of mutant structures utilized structure factors with a resolution greater than 6.0 Aand that had a F/a(F) ratio > 2.0. During the course of refinement the remaining side-chainatoms for residues 52 and 67 were unambiguously located and added to the refinement models.In addition, based on F, — F, difference and fragment maps, and 2F0 —11 and 3F, — 211 dif-ference electron density maps, the side-chains of some surface residues were repositioned andwater molecules were added or deleted according to how well they could be resolved. Finalrefinement statistics are given in Table 5.18.The final refined structures of the reduced and oxidized yeast iso-l-cytochrome c N52I andN52I-Y67F mutants display good stereochemistry (see Table 5.18). The coordinate error inthese structures was assessed by inspection of Luzzati plots (see Figure 5.38) and by the methodof Cruickshank (Luzzati, 1952; Cruickshank, 1985; see also Appendix C). In Luzzati plots theoverall r.m.s. coordinate errors for the mutant structures was estimated to be ,--, 0.20-0.22 A,Chapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^ 97Table 5.18: Refinement results and stereochemistry for the final models of the yeast iso-l-cyto-chrome c N52I and N52I-Y67F mutants in both oxidation statesN52IReduced^OxidizedN52I-Y67FReduced^OxidizedI. Refinement resultsR-factor (%) 18.5 18.2 17.9 18.1No. of reflections in refinement 4142 4898 5406 4538No. of atoms 956 944 950 937No. of solvent molecules 59 46 53 40II. Stereochemistry of final modelsr.m.s. deviation from ideal valuesBond distances (A)1-2 bond distance1-3 bond distance1-4 bond distancePlanar restraints (A)Chiral volume (A3 )Non-bonded contactst (A)single-torsionmulti-torsionpossible hydrogen bondsTorsion angles (°)planar (0° or 180°)staggered (±60°,180°)0.017 0.021 0.022 0.0220.044 0.046 0.044 0.0410.062 0.060 0.061 0.0580.017 0.018 0.017 0.0170.217 0.219 0.215 0.2350.226 0.218 0.219 0.2130.224 0.179 0.181 0.1820.233 0.230 0.212 0.2252.6 2.9 2.7 2.523.7 24.0 22.3 21.5t The r.m.s. deviations from ideality for this class of restraint incorporates a reduction of 0.2 Afrom the radius of each atom involved in a contact.while values of '0.15 A were obtained for these structures with Cruickshank's method.5.0^3.3 2.5^2.00.05 0.10 0.15^0.20SIN(8) /0.25 0.30^ 1.00- 0.75 &130- 0.50 .2aL1_-1 0.250.30 -0.25 -OO 0.20 -cc0.15 -0.10Chapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^ 98Resolution (A)Figure 5.38: Plots of the dependence on resolution of both the R-factor agreement betweencalculated and observed structure factors (axis lower left), and the fraction of data used (axisat top right), for the final refined structures of reduced (m ) and oxidized (0) N521, and reduced(11) and oxidized (0) N521-Y67F mutant yeast iso-1-cytochromes c. For this analysis reciprocalspace was divided into shells according to sin(9)/A, with each containing at least 230 reflections.For the purpose of assessing the accuracy of the atomic coordinates of the mutant structures,curves representing the theoretical dependence of R-factor on resolution assuming various levelsof r.m.s. error in the atomic positions of the model (Luzzati, 1952) are also drawn (dashedlines). This analysis suggests an overall r.m.s. coordinate error for the four mutant structuresof between rs-,0.20 and 0.22 A.Chapter 5. Yeast Iso-l-Cytochrome c N521 and N52I-Y67F Mutants^995.2 Results5.2.1 Polypeptide chain conformationBoth the reduced and oxidized structures of the wild-type (Louie & Brayer, 1990; Chapter 3)and Y67F mutant (Chapter 4) yeast iso-l-cytochromes c have been determined. To allowfor a comprehensive analysis of the effects of mutations in the N52I and N52I-Y67F proteins,all the available coordinate sets were superimposed using a least-squares procedure, based onall main-chain and heme atoms. This comparison reveals that the polypeptide chain fold isunaffected by these position 52 and 67 mutations (see Figure 5.39). As indicated in Table 5.19average deviations for main-chain atoms between the reduced and oxidized N52I and N52I-Y67F structures, and the reduced wild-type protein are of the order of 0.17-0.25 A, which isslightly less than that observed for the reduced and oxidized Y67F proteins (0.29 and 0.32 A,respectively). The only region where substantially different conformations of the polypeptidechain are observed is at the N-terminus (see Figure 5.40). However, the first three residues of theH5^H5Figure 5.39: A stereo-drawing of the a-carbon backbones of the reduced and oxidized structuresof wild-type, N52I, Y67F and N52I-Y67F yeast iso-l-cytochromes c. The heme group has alsobeen drawn as well as the two heme ligands, Hisl8 and Met80, and cysteines 14 and 17, whichform thioether linkages to the heme porphyrin ring. In addition the side-chain atoms of residues52 and 67, which are replaced in the mutant proteins, are also displayed. For clarity, every 5thresidue has been labelled with its one-letter amino acid designation and sequence number.Chapter 5. Yeast Iso-I-Cytochrome c N521 and N521-Y67F Mutants^100Table 5.19: Average deviations of polypeptide chain atoms in the N52I and N52I-Y67F mutantswith respect to reduced wild-type yeast iso-l-cytochrome cN52I^N52I-Y67FReduced Oxidized Reduced OxidizedAverage deviations (A)main-chain atoms 0.25 0.20 0.17 0.20buried side-chain atoms 0.30 0.27 0.24 0.24exposed side-chain atoms 0.65 0.65 0.52 0.51A side-chain was considered buried if less than 20% of its surface area was solvent exposedrelative to the unfolded state in any one of the four mutant protein structures (Shrake Rupley,1973; Perry et al., 1990).polypeptide chain (Thr(-5), Glu(-4) and Phe(-3)) are disordered in electron density maps andthe differing conformations observed reflect alternate fits to the same poor electron density. Inno other instances are shifts of main-chain atoms more than twice the overall average deviationobserved. Given this result it is not surprising that all of the main-chain to main-chain hydrogenbonds present in the reduced yeast iso-l-cytochrome c wild-type structure are also observed inthe four mutant structures (for a complete list of these hydrogen bonds see Table 8 in Louie &Brayer (1990)).Examination of side-chain conformations reveals that all buried residues have average devi-ations less than 0.8 A, except for Leu9 (see Figure 5.40). In the reduced and oxidized N52I andN52I-Y67F structures the side-chain of Leu9 has average deviations of 0.5 A, 1.0 A, 0.8 A and1.0 A, respectively. This positional flexibility in Leu9 side-chain placement is also reflected inthe multiple conformations observed in the reduced wild-type yeast iso-1-cytochrome c struc-ture (Louie, 1990). Overall, average deviations for buried side-chains are comparable to thoseobserved for main-chain atoms (Table 5.19). As evident from Figure 5.40 many solvent exposedside-chains display a variety of different conformations among the various structures, indicativeof their high degree of mobility. This is not only reflected in the overall average deviations-- •1^I^1^15 15 25 35I^ t ^I , , ■ ■ ,,,,, t95715^850.00.02.0 ^ 1 ^ 45 55^65Main—chain 1.0-I ,1111, 1.11C00tv^1.0 7cnE2.0 73.0 -4.0-5.0Chapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^101Residue NumberFigure 5.40: A plot of the overall average positional deviations of the reduced and oxidizedN52I and N52I-Y67F mutant proteins when compared to the reduced yeast iso-1-cytochrome cwild-type structure. In the top frame overall average deviations for main-chain atoms are shown,while in the bottom frame the overall average deviations for side-chain atoms are displayed.The vertical bars represent the range of individual pairwise average deviations. In the bottomframe, dotted vertical bars refer to solvent exposed residues.observed, which are approximately double that for buried side-chain atoms (see Table 5.19),but also in the higher thermal factors for this group of atoms.It has been proposed that differences in polypeptide chain flexibility are an essential elementin the structural differences between oxidation states of cytochrome c (see Section 3.3). Analysisof the thermal factors of main-chain atoms for the reduced and oxidized N52I and N52I-Y67Fmutants using a difference matrix method (Section 3.2) provides the following results (see Fig-ure 5.41). For all four mutant proteins the overall thermal factor profile is remarkably similar tothat of the reduced wild-type protein. However, two differences appear to be significant. In the-s 26^36^46^58^66^76^86^88 26^38 66^78^86^66Residue number Residue number16Chapter 5. Yeast Iso-l-Cytochrome c N521 and N52I-Y67F Mutants^102Legend15.0 and greaterMa 10.0 to 15.05.0 to 10.0Figure 5.41: A matrix representation of the differences in average main-chain thermal fac-tors between the reduced and oxidized N52I and N52I-Y67F mutant proteins, and the reducedwild-type protein. Each matrix point Px , y represents an amino acid pairing (x, y) and wascalculated using the equation: Px ,y (Bx — ity )mut ant — (Bx — By ) wild —type where B is theaverage main-chain thermal factor of a given amino acid. Positive matrix values are displayedas squares of different levels of blackness according to the scale on the right. As this matrixhas inverse symmetry across the diagonal line drawn, negative values are redundant and areomitted for clarity. The advantage of this approach is that displayed values are not affectedby differences in overall thermal factor between the two structures. Within the matrix, aminoacids which have significantly higher average main-chain thermal factors in the mutant struc-tures produce vertical streaks. These include residues 65-68 in the reduced N52I protein withmaximal differences observed for Tyr67. Amino acids producing horizontal streaks indicate thepresence of significantly lower thermal factors for their main-chain atoms in the mutant protein.An example of this are residues 54-55 in the oxidized N52I-Y67F variant.Chapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^103reduced N52I mutant protein a marked increase in main-chain thermal factors is observed forresidues 65-68. This effect is focussed at Tyr67 where the average increase is ,+10 A 2 . A cor-responding increase in side-chain thermal factors is also observed (average increase ,+10 A 2 ).The second difference is seen in the oxidized N52I-Y67F mutant protein. Here residues 54-55,focussed at Lys54, show a significant drop in thermal factors (average decrease ,,,-10 A 2 ). Thisdecrease in flexibility is not only restricted to main-chain atoms but is also observed for theside-chains of these amino acids.5.2.2 Heme structureIn iso-1-cytochrome c the porphyrin ring of the heme group is found to be distorted fromplanarity into a saddle shape. As evident in Table 5.20, the degree of distortion is oxidationstate dependent and most pronounced in the oxidized form. The N52I-Y67F protein shows asimilar trend, although overall heme plane distortion is smaller in the oxidized state. However,both the N52I and Y67F mutant proteins in either the reduced or oxidized state show distortionssimilar to those seen in the oxidized wild-type protein.Inspection of heme iron coordinate bond lengths between the different reduced and oxidizedstructures show these to be similar within expected atomic coordinate errors. However, theorientation of the His18 imidazole plane does appear to be a function of oxidation state (seeTable 5.20). For the reduced wild-type and mutant cytochromes c the average angle between theimidazole plane normal and a line drawn through the NA and NC heme atoms is ,46°. In theoxidized state the His18 side-chain rotates towards the A and C pyrrole rings, resulting in valuesin the range of 50-55° for this angle. In contrast, the Met80 ligand does not exhibit oxidationstate dependent shifts and virtually identical conformations are observed in the wild-type, N52Iand N52I-Y67F cytochrome c structures (average deviations 0.2-0.3 A).Both heme propionates form extensive hydrogen bond interactions with protein groups(see Table 5.21). Between the different reduced and oxidized structures, the hydrogen bondsformed by the pyrrole ring D propionate show particularly high conservation. Nonetheless, thisChapter 5. Yeast Iso-l-Cytochrome c N521 and N52I-Y67F Mutants^104Table 5.20: Heme conformation and ligand geometry in wild-type, N52I, Y67F and N52I-Y67Fmutant yeast iso-l-cytochromes cReduced^ OxidizedPyrrole Wild-type N52I Y67F N52I-Y67F^Wild-type N52I Y67F N52I-Y67F I. Angular deviations between the pyrrole nitrogen plane normal and thefour pyrrole ring plane normals (°)A^9.4^12.2^13.7^10.7^12.6^13.4^13.4^8.9B 11.1^15.4^9.1^8.9 14.1^10.3^10.7^11.6C^8.8^10.2^11.7^9.4^9.6^9.2^12.0^11.6D 8.1^11.3^11.3^7.5 12.8^10.7^10.2^8.5Average^9.3^12.7^11.5^9.1^12.3^10.9^11.6^10.2II. Angular deviations between the porphyrin ring plane normal and the fourpyrrole ring plane normals, plus the pyrrole nitrogen plane normal (°)A^6.7^8.0^9.0^7.0^8.3^9.2^7.4^5.4B 11.9^14.3^12.3^10.5 12.4^11.8^12.4^12.8C^9.8^13.6^11.6^9.5^13.8^10.2^10.8^11.2D 6.0^8.0^6.4^4.2 10.0^7.2^4.4^5.0Average^8.6^11.0^9.8^7.8^11.1^9.6^8.8^8.6NNNN^2.6^4.2^4.9^3.7 4.3^4.3^6.0^3.8III. Angles between the normal of the His18 imidazole plane and a vectordrawn through the NA and NC heme atoms (°)46.7^45.4^44.8^46.9^55.8^49.7^54.6^49.7Each pyrrole ring is defined by nine atoms, which include the five ring atoms plus the firstcarbon atom bonded to each ring carbon. The porphyrin ring plane is defined by the five atomsin each of the four pyrrole rings, the four bridging methine carbons, the first carbon atom ofeach of the eight side-chains and the heme iron (33 atoms in total). The pyrrole nitrogen planeis defined by only the 4 pyrrole nitrogens (see Figure 1.2 for heme atom labeling convention).propionate group does have more than a single conformation. For example, when comparingthe C2D-C3D-CAD-CBD torsion angle (102° in the reduced wild-type protein), a range ofvalues are observed, the largest and smallest coming from the reduced N52I and N52I-Y67Fstructures (113° and 76°, respectively). However, this torsion angle variability does not affectthe position of the two pyrrole D propionate oxygen atoms which in the reduced wild-type andthe four mutant structures are all located within 0.3 A of each other. This is due to smallerChapter 5. Yeast Iso-I-Cytochrome c N521 and N52I-Y67F Mutants^105Table 5.21: Heme propionate hydrogen bond interactions in wild-type, N52I, Y67F andN52I-Y67F mutant yeast iso-l-cytochromes cInteractionDistances (A)ReducedWild-type N52I Y67F N52I-Y67FOxidizedWild-type N52I Y67F N52I-Y67FO1A Tyr48 OH 2.83 2.63 2.97 2.53 2.83 2.67 3.51 2.79Wat121 2.81 3.47 2.57 2.83 2.85 3.20 2.69 2.45Wat168 2.85 3.50 3.41 3.00 2.87 3.31 (3.62) 2.9102A G1y41 N 3.21 3.07 2.82 3.25 2.60 3.33 2.88 3.34Asn52 ND2 3.34 - 2.67 (3.54) 3.41 -Trp59 NE1 3.09 2.84 3.21 3.05 3.43 2.80 2.89 2.86Wat121 (4.01) (3.90) (3.74) (3.68) 3.34 (3.89) 3.51 (3.71)OlD Thr49 OG1 2.64 2.50 2.60 2.68 2.79 2.73 2.71 2.52Thr78 OG1 2.90 2.75 2.75 3.03 3.07 3.03 3.08 2.95Lys79 N 3.17 3.03 (3.39) 3.26 2.67 3.10 2.96 3.1202D Thr49 N 2.94 2.79 2.77 2.97 2.75 3.17 2.95 2.95Interactions were accepted as hydrogen bonds only if they met all of the following criteria: aH• • •A distance < 2.6 A, a D-H• • •A angle > 120°, and a C-A• • •1 angle > 90°. Valuesgiven in brackets are not considered to be hydrogen bonds by this criteria, but are listed forcomparison.compensatory angular changes in other pyrrole D propionate torsional angles.In the reduced and oxidized N52I and N52I-Y67F mutant structures the conformation ofthe pyrrole A propionate is essentially the same as that observed in the reduced wild-typestructure (see Figure 5.42). The largest difference is an ,-, 15° rotation of the carboxyl group ofthe pyrrole A propionate around the CBA-CGA bond in both mutant proteins. This causesthe O1A oxygen atom to move towards Tyr48 OH and the 02A oxygen atom to move towardsTrp59 NE2. This conformational readjustment appears to occur in response to the absence ofthe Asn52 side-chain in these mutants. In reduced wild-type and Y67F mutant yeast iso-l-cytochromes c Asn52 is positioned adjacent to the heme pyrrole A propionate to which it formsChapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^106Figure 5.42: A stereo-drawing showing the region around the pyrrole ring A propionate inreduced N52I-Y67F (thick lines) and wild-type (thin lines) yeast iso-1-cytochromes c. Mutationof residue 52 from an asparagine to an isoleucine results in removal of a hydrogen bond (dashedlines) to this propionate group and as a consequence it rotates , 15°. Similar rotations of thepyrrole A propionate group are also seen in the oxidized state of the N52I-Y67F mutant, aswell as in the reduced and oxidized states of the N52I mutant protein.a hydrogen bond interaction (see Chapters 3 and 4). Notably, the rearrangement of pyrrole Apropionate conformation and hydrogen bond interactions observed in the wild-type and Y67Fproteins upon a change in oxidation state (Chapters 3 and 4) does not occur in the N52I andN52I-Y67F variants.5.2.3 Mutation site regionIn reduced wild-type yeast iso-1-cytochrome c the side-chains of Asn52, Tyr67 and Thr78 formhydrogen bonds to a structurally conserved internal water molecule (Wat166; see Figure 5.43a).This water molecule shifts and reorients its dipole moment upon oxidation so as to stabilizethe positive charge on the heme iron atom (Chapter 3). Previous studies have shown thatmutation of Tyr67 to a phenylalanine results in an additional water molecule being introducedinto this internal water cavity (Chapter 4; Figure 5.43c). In contrast, our current studiesshow that substitution of Asn52 for an isoleucine results in the displacement of Wat166 (seeFigure 5.43b). In the reduced wild-type protein the size of the cavity occupied by Wat166 is,--10 A3 . As a consequence of the 11e52 substitution this cavity is absent in both the reducedBHe 52.416Met 80'4-His 18Thr 78DPhe 67Asn 52CChapter 5. Yeast Iso-1-Cytochrome c N521 and N521-Y67F Mutants^107Figure 5.43: Drawings showing the region about residues 52 and 67 in (a) wild-type, (b) N52I,(c) Y67F and (d) N52I-Y67F yeast iso-1-cytochromes c in the reduced state. In each case theheme group is shown with dark shaded balls and internal water molecules are represented bylarger spheres. Heme ligand interactions are indicated by thin white bonds whereas hydrogenbonds are shown by thin dashed lines.Chapter 5. Yeast Iso-l-Cytochrome c N521 and N52I-Y67F Mutants^108and oxidized states of the N52I variant thereby excluding a water molecule at this location.The absence of Wat166 in the N52I mutant protein disrupts the normal hydrogen bondnetwork in this region of the protein. One result of this is that in the reduced N52I protein theside-chain of Tyr67 shifts (average deviation 0.6 A) towards the side-chain of Thr78 to which itshydroxyl group forms a hydrogen bond (distance 3.4 A). In the oxidized N52I structure a similarsituation is seen. Here the side-chain of Tyr67 has a more modest shift (average deviation 0.3 A)and according to our normal criteria the interaction between the Tyr67 OH and Thr78 0G1 istoo long (3.6 A) to be considered a hydrogen bond. However, in both oxidation states of theN52I mutant, as a result of the Tyr67 side-chain shift, the hydrogen bond between Tyr67 OHand Met80 SD is absent. The loss of this hydrogen bond is further demonstrated in the reducedN52I protein by an increase in thermal factors for Tyr67.In the N52I-Y67F reduced and oxidized structures even fewer hydrogen bonds are formed(see Figure 5.43d). In both these structures the size of the internal cavity originally occupiedby Wat166 in wild-type yeast iso-l-cytochrome c has been reduced to ,8 A3 . This is too smalla volume to allow for water binding and as a consequence Wat166 is also excluded from thestructure of the N52I-Y67F mutant protein. Like the N52I variant this is the result of thepositioning of the side-chain of isoleucine 52. No significant positional shifts or rearrangementsof hydrogen bonds are seen to compensate for the substituted amino acids.5.3 Discussion5.3.1 Structural effects of mutationsA summary of the differences observed in the reduced and oxidized N52I and N52I-Y67F mutantstructures compared to their respective wild-type structures is given in Table 5.22. It is clearfrom this comparison that the mutations at position 52 and 67 primarily affect the conformationof two regions of the protein, these being the mutation site and the pyrrole A propionate regions.Differences are not only expressed in terms of atom shifts and changes in hydrogen bonding, butalso in the altered flexibilities of polypeptide chain segments that make up these two regions ofChapter 5. Yeast Iso-1-Cytochrome c N521 and N52I-Y67F Mutants^ 109Table 5.22: Structural differences observed in the N52I and N52I-Y67F mutant structures whencompared to wild-type yeast iso-l-cytochrome cN52I^ N52I-Y67FHeme structure (see Figure 5.42 and Tables 5.20A.1. Both reduced and oxidized mutants dis-play a similar thermal factor profile asseen in the reduced wild-type protein withthe exception of increased flexibility forresidues 65-68 (maximal at residue 67) inthe reduced proteinB.1. Distortion of the heme plane in the re-duced and oxidized state are similar tothat observed in the oxidized wild-typestructure2. Pyrrole A propionate hydrogen bond net-work does not display oxidation state de-pendent conformational differences3. Absence of asparagine side-chain at po-sition 52 causes small rotation of propi-onate A carboxyl groupC. Mutation site region (see Figure 5.43).1. Displacement of Wat166 from internalcavity2. Loss of Tyr67 OH to Met80 SD hydrogenbond interaction3. New hydrogen bond interaction formedbetween Tyr67 and Thr78 in the reducedstate1. Both reduced and oxidized mutants dis-play a similar thermal factor profile asseen for reduced wild-type protein, exceptfor decreased flexibility of residues 54-55(focused at Lys54) in the oxidized variantand 5.21)1. Distortion of the heme plane is less pro-nounced in oxidized state2. Pyrrole A propionate hydrogen bond net-work does not display oxidation state de-pendent conformational differences3. Absence of asparagine side-chain at po-sition 52 causes small rotation of propi-onate A carboxyl group1. Displacement of Wat166 from internalcavity2. Loss of Tyr67 OH to Met80 SD hydrogenbond interactionThermal factors for main- chain atoms (see Figure 5.41)the protein.Chapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^110Mutation site regionComparisons of mutation site regions show that the hydrogen bond network about Wat166 isaltered in both mutant proteins (see Figure 5.43). In the N52I mutant the hydrogen bondbetween Tyr67 OH and Met80 SD is severed, Wat166 is displaced, a new hydrogen bond isformed between Tyr67 OH and Thr78 OG1, and the link from Asn52 ND2 to the pyrrole Apropionate is lost. In the N52I-Y67F variant only the hydrogen bond between Thr78 OG1 andthe heme pyrrole D propionate remains of the original hydrogen bond network in this region.Surprisingly, the breakdown of the hydrogen bond network about the internal water molecule,Wat166, does not adversely affect the stability of cytochrome c, rather the opposite appearsto be the case. A dramatic increase of thermodynamic stability for the N52I yeast iso-l-cyto-chrome c mutant has been reported (Das et al., 1989; Hickey et al., 1991). Also, this mutantis less sensitive to alkaline pH. This can be seen in the oxidized wild-type protein where theMet80 SD - heme iron ligand bond is broken at pH 8.5 (Pearce et al., 1989) whereas this alkalinetransition occurs at pH 10.0 (J.G. Guillemette, private communication) when Asn52 is mutatedto an isoleucine. Resistance to this alkaline transition has been taken as a measure of proteinstability (Saigo, 1986). The N52I-Y67F mutant protein has not been as extensively studied,but measurements of its alkaline transition indicates this occurs at pH 11.0 (J.G. Guillemette,private communication) suggesting even greater stability than that found for the N52I mutant.These stability studies (all performed with the oxidized protein) correlate well with observationsthat neither of these mutant proteins display an increase in polypeptide chain flexibility in theoxidized state as is observed in the wild-type protein (see Figure 5.41). Indeed, in the N52Imutant residues 65-68 show a decrease in flexibility when this protein is in the oxidized state,and for the N52I-Y67F variant residues 54-55 appear to be less mobile under these conditions.It is noteworthy that both these polypeptide chain segments display an increase in flexibilityin the oxidized wild-type structure (Chapter 3).A similar situation is observed for the Y67F mutant cytochrome c, which is more resistantto high temperature, pH and denaturation by urea than the wild-type protein (Luntz et al.,Chapter 5. Yeast Iso-l-Cytochrome c N52I and N52I-Y67F Mutants^ 1111989). The Y67F variant also has a higher alkaline transition (pH 10.1; J.G. Guillemette, privatecommunication) that is similar to that of the N52I mutant protein. Furthermore, structuralstudies of the Y67F mutant protein have shown a breakdown of the hydrogen bond networkinvolving Wat166, particularly in the oxidized state (Chapter 4). This mutant protein alsoappears to be more rigid in the oxidized state as inferred from its thermal factor profile (seeFigure 4.35).Taken together, these observations suggest that the hydrogen bond network involving Wat166destabilizes the tertiary structure of cytochrome c, since alteration of this network or removalof Wat166 altogether, leads to considerable structural stabilization. It also appears that onefunction of Wat166 and the interactions it forms is to specifically increase the flexibility of threenearby segments of polypeptide chain in the oxidized state. This feature is abolished in all threeof the N52I, N52I-Y67F and Y67F variant proteins, suggesting much of the observed increasein protein stability may be due to this factor.Pyrrole A propionate regionThe region about the pyrrole A propionate has been previously shown to be sensitive to oxida-tion state (Chapter 3). In wild-type yeast iso-l-cytochrome c this group displays two distinctlydifferent conformations depending on the valence state of the heme iron atom (see Figure 3.25).In the Y67F variant nearly identical conformational changes are observed (see Figure 4.36).Other mutant cytochromes c are observed to have the pyrrole A propionate conformation foundin the oxidized wild-type protein even though they are in the reduced state (Louie et al., 1988b;Louie & Brayer, 1989).The origin of the driving force behind oxidation state dependent conformational changesabout the pyrrole A propionate remains unclear. Several factors such as, differential delocal-ization of the negative charge on the heme propionate group (Barlow & Thornton, 1983; Singhet al., 1987); electrostatic interactions between this propionate, the heme iron atom and the side-chain of Arg38 (Moore, 1983); and oxidation state dependent distortions of the heme porphyrinChapter 5. Yeast Iso-l-Cytochrome c N521 and N521-Y67F Mutants^ 112ring (Chapter 3) have all been suggested to affect the pyrrole A propionate conformation.Neither the N52I nor the N52I-Y67F mutant proteins have oxidation state dependent con-formational rearrangements about the pyrrole A propionate. This absence may provide someinsight into the factors behind this rearrangement in the wild-type protein. In the first instance,the charge on the heme iron does not appear to be a factor since this conformational changeis not elicited between oxidation states in these mutant proteins. Secondly, changes aboutthe pyrrole A propionate are not a consequence of increased heme plane distortion since, forexample, both N52I and Y67F display comparable distortions in both oxidation states whilethey do not exhibit similar pyrrole A propionate conformations (Tables 5.20 and 5.21). Theseobservations clearly suggest that the chemical character of residue 52 is critical in mediatingthe changes about the pyrrole A propionate group between oxidation states and could be drivenby the change in conformation that the Asn52 experiences as the result of the oxidation statedependent movement of Wat166 (see Figure 3.28).5.3.2 Importance of the hydrogen bond network around Wat166Control of midpoint reduction potentialBoth the N52I and N52I-Y67F variant proteins have midpoint reduction potentials which arers-,56 mV lower than that of wild-type yeast iso-l-cytochrome c (see Tables 1.4 and 1.5; Burrowset al., 1991; J.G. Guillemette, private communication). In this regard these mutant proteinsare similar to the Y67F variant (Chapter 4). In this latter case the observed lower reductionpotential can be attributed to the breakage of the hydrogen bond from the hydroxyl group ofTyr67 to the side-chain of Met80 which forms a heme ligand interaction. In the N52I mutantprotein the Tyr67 OH is oriented towards Thr78 (see Figure 5.43) strongly suggesting that theabsence of a hydrogen bond interaction to Met80 SD is also responsible for the lower midpointreduction potential of this mutant protein. Langen et al. (1992) have proposed that the removalof the Asn52 side-chain and its associated dipole might be a factor in the observed drop inmidpoint reduction potential. However, similarities in midpoint reduction potential betweenChapter 5. Yeast Iso-l-Cytochrome c N521 and N52I-Y67F Mutants^113the Y67F and N52I mutant proteins suggest this is unlikely to be a major contributor to thisphenomena (see Table 1.5). For the N52I-Y67F protein a similar situation exists. That is, theinteraction to the Met80 side-chain has been deleted leading to a drop in midpoint reductionpotential comparable to the Y67F and N52I variants.Beyond demonstrating the effects of the Tyr67 OH - Met80 SD interaction on the midpointreduction potential of cytochrome c, our collective results clearly delineate the role Wat166has in this area. As has been shown, this internal water molecule appears to play an integralpart in helping to stabilize the alternative oxidation states of this protein. In addition, it isclear that Wat166 is essential to the maintenance of the hydrogen bond network in this area.In particular, the N52I mutant protein shows the presence of Wat166 provides for the properhydrogen bond interactions between Tyr67 and Thr78. Its absence leads to a disruption inthe hydrogen bond network of which it is the central feature, the shift of Tyr67 to interactwith Thr78 and a subsequent drop in midpoint reduction potential. Thus these studies alsoshow that Wat166 is a critical determinant in maintaining the value of the midpoint reductionpotential of cytochrome c.Modulation of flexible regionsBesides regulation of midpoint reduction potential another possible role of Wat 166 and thehydrogen bond network that it forms an integral part of, is in the modulation of flexible regionsin cytochrome c between oxidation states. In wild-type yeast iso-1-cytochrome c Wat166 shiftsand reorients upon heme iron oxidation so that its dipole moment is positioned to stabilizethe positive charge residing on the heme iron atom. During this process a rearrangement ofthe hydrogen bond network about Wat166 takes place and several hydrogen bonds are broken,resulting in residues 47-59, 65-72 and 81-85 exhibiting increased mobility (see Chapter 3).The more open and flexible nature of the oxidized form of cytochrome c is a property welldocumented using a variety of techniques ranging from NMR (Williams et al., 1985a) to smallangle X-ray scattering (Trewhella et al., 1988) and infrared spectroscopy (Dong et al., 1992;Chapter 5. Yeast Iso-l-Cytochrome c N52I and N521-Y67F Mutants^114see also Section 1.3).The N521 and N521-Y67F mutations result in proteins which do not exhibit flexible regions inthe oxidized state. The same is also observed for the Y67F variant (Chapter 4). As mentionedabove, examination of the three dimensional structures of these three mutants reveals thatthe hydrogen bond network involving Wat166 is substantially altered, particularly when theseproteins are in the oxidized state. These observations strongly suggest that Wat166 is a centralcomponent in the mechanism which mediates the oxidation state dependent mobility of selectedpolypeptide chain segments.The biological role of oxidation state dependent polypeptide chain flexibility in cytochrome cremains unclear, but several theories have been suggested. For example, it has been proposedthat this factor is important for complexation and dissociation with redox partners (Bosshard& Zurrer, 1980; Rackovsky & Goldstein, 1984; Zhang et al., 1990; Dong et al., 1992). This isplausible since the determination of the oxidized structure of wild-type yeast iso-l-cytochrome chas shown that mobility differences are localized in the area implicated to be the binding sitewith redox partners (see Chapter 3). These flexibility differences might also be important forthe kinetics of biological electron transfer, either through changing the effective heme solventaccessible surface area (Schlauder & Kassner, 1979; Zheng et al., 1990), or through modulationof the reorganization energy (Churg & Warshel, 1983; Warshel, 1983; Marcus & Sutin, 1985;Williams et al., 1985b). Another possibility is that the more open and flexible nature of theoxidized state is a means to regulate the thermodynamic properties of the midpoint reductionpotential (Watt & Sturtevan, 1969).To summarize, our studies suggest that the function of the internal water molecule, Wat166,and its associated hydrogen bond network is three fold. First, the presence of Wat166 providesfor a convenient mechanism to modify the hydrogen bond network involving several key residuesnear the Met80 ligand, depending on the oxidation state of the heme. This appears to be par-ticularly important for modulating the hydrogen bond between Tyr67 OH and Met80 SD. Asso-ciated with this is the fact that the presence of Wat166 is necessary to maintain the spatial andChapter 5. Yeast Iso-I-Cytochrome c N52I and N521-Y67F Mutants^115hydrogen bonding relationships between residues in this area and in this way is also an impor-tant element in setting the value of the midpoint reduction potential of cytochrome c. Finally,Wat166 also appears to mediate oxidation state dependent mobility differences of polypeptidechain segments which might play a role in redox partner recognition (Chapter 3).Chapter 6Importance of the Dipole Orientation of a Conserved Internal Water Moleculein the Biological Function of Cytochrome c as Revealed bythe Mutations N52A and I75M6.1 Experimental Procedures6.1.1 CrystallizationProtein samples for these studies were obtained from J.G. Guillemette (University of BritishColumbia). Crystals of the N52A mutant of yeast iso-l-cytochrome c were initially grownin a 0.1 M sodium phosphate buffer (pH 6.4) containing 90% saturated ammonium sulfateand 60 mM DTT (to maintain the reduced state of the protein), using a combination of thehanging drop method and a hair-seeding technique (Leung et al., 1989; see also Section 2.1).However, these crystals were of insufficient size for high resolution x-ray diffraction analyses andadditional macro-seeding had to be employed to increase their dimensions. The I75M mutantprotein was crystallized in a 0.1 M sodium phosphate buffer (pH 5.3) containing 95% saturatedammonium sulfate and 70 mM sodium dithionite using similar methods. Prior to mounting,mutant crystals were transferred to fresh mother-liquor, with the exception that for the I75Mprotein 40 mM DTT was used instead of sodium dithionite. Mutant protein crystals proved tobe isomorphous to those of wild-type yeast iso-l-cytochrome c, having the same space group(P43212) and similar unit cell dimensions, a = b = 36.52 A, c = 137.89 A, and a = b = 36.55 A,c = 138.59 A for the N52A and I75M crystals, respectively. Wild-type yeast iso-1-cytochrome ccrystals have unit cell dimensions of a = b = 36.46 A, c = 136.86 A.116Chapter 6. Reduced Yeast Iso-1-Cytochrome c N52A and I75M Mutants^1176.1.2 Data collection and data processingDiffraction data for each mutant protein was collected from one crystal on an Enraf-NoniusCAD4-F11 diffractometer. The radiation used for the diffraction experiment was nickel filteredand generated from a copper target x-ray tube operating at 26 mA and 40 kV. Reflectionswere measured using continuous Sr/ scans. The ambient temperature during data collection wasmaintained at 15°C. To monitor slippage and decay, six reflections were measured periodically.In this manner, a total of 10779 reflections were collected to 2.0 A resolution for the N52Aprotein, and 10308 reflections to 1.9 A resolution for the I75M protein. Diffraction intensitieswere corrected for background, absorption, decay, Lorentz and polarization effects as previouslydescribed (see Section 2.2 and 3.1). The resultant structure factors were put on an absolutescale using a Wilson (1942) plot.6.1.3 Refinement and analysesMutant minus wild-type difference Fourier maps revealed that in the N52A and I75M proteinsstructural changes were restricted to the direct vicinity of each mutation site. The startingmodels for least-squares restrained parameter refinement (Hendrickson & Konnert, 1981) forboth mutant proteins was the wild-type yeast iso-l-cytochrome c structure (Louie & Brayer,1990) in which mutated residues were replaced by alanines. A subset of well determined solventmolecules was also included in the starting models, with the notable exception of the internalwater molecule, Wat166, which is located adjacent to the side-chains of both residues 52 and75. Several manual interventions were carried out during the course of refinement based oninspection of F, — Fc , 2110 — and 3F0 — 21-', difference electron density maps. These involvedadjustments to solvent structure and disordered side-chains in both structures, as well as theaddition of the remaining atoms to the side-chain of Met75 in the I75M structure.As shown in Table 6.23 the final refined models for the N52A and I75M yeast iso-l-cyto-chrome c mutants have good stereochemistry. The final standard crystallographic R-factorsfor these structures are 18.5% and 22.1%, respectively. The higher value obtained for theChapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and 175M Mutants^118Table 6.23: Final stereochemistry for N52I and I75M yeast iso-l-cytochromes cStereochemicalrefinement parametersr.m.s. deviation fromideal valuesN52A^I75MBond distances (A)1-2 bond distance 0.020 0.0241-3 bond distance 0.044 0.0501-4 bond distance 0.052 0.064Planar restraints (A) 0.016 0.018Chiral volume (A 3 ) 0.196 0.286Non-bonded contacts' (A)single-torsion 0.219 0.236multi-torsion 0.207 0.242possible hydrogen bonds 0.245 0.262Torsion angles (°)planar (0°or 180°)^2.6^2.6staggered (±60°,180°)^25.9^26.3f The r.m.s. deviations from ideality for this class of restraint incorporates a reduction of 0.2 Afrom the radius of each atom involved in a contact.I75M mutant reflects the smaller crystals available for this protein and the lower quality ofthe resultant diffraction data (Figure 6.44). The accuracy of the two structures was assessedby inspection of a Luzzati plot (Luzzati, 1952; Appendix C; see Figure 6.44) and suggests anr.m.s. coordinate error of '4.20 A for the N52A structure and ,-4.24 A for the I75M structure.A separate estimate for the overall coordinate error can be obtained using the method ofCruickshank (1949, 1954, 1985; see also Appendix C) and gave an r.m.s. coordinate error of0.18 A and 0.25 A for the N52A and I75M mutant structures, respectively. This is in goodagreement with Luzzati's method.5;0^3.3^ 1.002.5^2.00.05 0.250.10^0.15^0.20SINKS) / A0.30>L-0.75 g0-^C-0.50 .0130-0.250.35-0.30-0.25-C0CC 0.20-0.150.10^Chapter 6. Reduced Yeast Iso-1-Cytochrome c N52A and I75M Mutants^119Resolution (A)Figure 6.44: Plots of the dependence on resolution of both the R-factor agreement betweencalculated and observed structure factors (axis at bottom left) and the fraction of data used(axis at top right), for the final refined structures of the N52A (1 and solid lines) and I75M(lb and dashed lines) mutant yeast iso-l-cytochromes c. For this analysis reciprocal space wasdivided into shells according to sin(0)/A, with each containing at least 280 reflections. Toassess the accuracy of the atomic coordinates of the mutant structures, curves representingthe theoretical dependence of R-factor on resolution assuming various levels of r.m.s. error inthe atomic positions of the model (Luzzati, 1952) are also drawn (dashed lines). This analysissuggests an overall r.m.s. coordinate error of ,0.20 A for the N52A structure and ,0.24 A forthe I75M coordinates.' 4D 90 11 ^10L 15L 85 X 100V 201170P 26G 45X 66Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^1206.2 Results6.2.1 Polypeptide chain conformationAs can be seen in Figure 6.45 the mutations N52A and I75M in yeast iso-l-cytochrome c do notdramatically affect the overall polypeptide chain fold, even though these residues are buriedwithin the protein and completely inaccessible to solvent. Overall average deviations for allmain-chain atoms when compared to wild-type yeast iso-1-cytochrome c are 0.24 A and 0.31 Afor the N52A and I75M structures, respectively (Figure 6.46). These values are comparable tothose observed for other yeast iso-1-cytochrome c variants (Chapters 4 and 5). However, tworegions do display shifts greater than two times the overall average deviation. The first regionis at the N-terminal end of the polypeptide chain. As previously discussed these residues aredisordered in the wild-type protein and all variant structures determined thus far. Therefore,the differing conformations observed likely represent only alternate fits to the same poorlydefined electron density (Chapters 3, 4 and 5). A second region of conformational shifts isfocussed at G1y41. In the N52A structure the main-chain atoms of G1y41 and G1n42 haveX 5^IC 5Figure 6.45: A stereo-drawing of the a-carbon backbones of the wild-type (thin lines), N52A(thick lines) and I75M (medium lines) yeast iso-l-cytochromes c. The individual heme groupshave also been drawn along with the two heme ligands, His18 and Met80, and cysteines 14 and17, which form thioether linkages to the heme porphyrin ring. In addition, the side-chain atomsof residues 52 and 75 are drawn. Every 5th residue of the wild-type protein has been labelledwith its one-letter amino acid designation and sequence number.4.0-Side—chain I.....^" ' I ^ I ^ I ^ I ^ I ^ I5.05^15^25^35^45^55^65^75^85^95 ^......... t 0.00•a)^1.0 -rne2.0-3.0-Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^121Residue NumberFigure 6.46: A plot of the average positional deviations between the N52A (thick lines) andI75M (thin lines) variants and the wild-type structure of yeast iso-l-cytochrome c. The topframe shows the observed deviations for main-chain atoms, while the bottom frame displaysthe deviations observed for side-chain atoms. The dashed lines in the top frame represent theoverall average deviations of all main-chain atoms for the two mutants which are 0.24 A and0.31 A for the N52A and I75M structures, respectively.average deviations of 0.6 A and 0.5 A, respectively. Interestingly, these same two residues alsodisplay nearly identical shifts in the I75M structure.Examination of side-chain conformations reveal that the 55 residues whose side-chains areburied inside the protein matrix have virtually identical conformations in both the mutant andwild-type structures; average deviations for these side-chain atoms are 0.32 A and 0.43 A forthe N52A and I75M proteins, respectively. The remaining 53 solvent exposed side-chains areobserved to have conformations which vary considerably, reflecting their positioning on theprotein surface and their high degree of mobility at this location. The average deviations forResidue number Residue number36^46^56^66^76 66^9616Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^122Legendkm 15.0 and greater10.0 to 15.05.0 to 10.0Figure 6.47: A matrix representation of the differences in average main-chain thermal factorsbetween the reduced N52A and I75M mutant proteins, and wild-type yeast iso-l-cytochrome c.Each matrix point Px ,y represents an amino acid pairing (x, y) and was calculated using theequation: Pxy = (Bx — Byy )mutant (Bx By )wild—type where B is the average main-chain thermalfactor of a given amino acid. Positive matrix values are displayed as squares of different levelsof blackness according to the scale on the right. As this matrix has inverse symmetry across thediagonal line drawn, negative values are redundant and are omitted for clarity. The advantageof this approach is that displayed values are not affected by differences in overall thermal factorbetween the two structures. Within the matrix, amino acids which have significantly higheraverage main-chain thermal factors in the mutant structures produce vertical streaks. Theseinclude residues 57-59 and 66-67 in the I75M protein. Amino acids producing horizontal streaksindicate the presence of significantly lower thermal factors for their main-chain atoms in themutant protein. No residues are observed in either mutant protein to fall in this latter category.this class of side-chain atoms are 0.64 A and 0.97 A for the N52A and I75M mutant proteins,respectively. This high degree of mobility is also reflected in the thermal factors for the solventexposed side-chains which have average values of 21.4 A 2 and 22.5 A2 for the N52A and I75Mvariants, respectively. This can be compared to average thermal factor values of 14.9 A 2 and16.0 A2 for the buried side-chain groups in these two mutant proteins.An analysis of the differences in polypeptide chain flexibility between reduced wild-typeyeast iso-l-cytochrome c and the N52A and I75M mutant structures is shown in Figure 6.47.Although, overall the mutant and wild-type structures have a similar thermal factor profile afew differences are observed in the I75M mutant structure. These include residues 57-59 whichChapter 6. Reduced Yeast Iso-I-Cytochrome c N52A and I75M Mutants^123display an increase of 13 A 2 for main-chain atoms (average thermal factor of 36 A2 versus 23 A2in the wild-type protein). Corresponding increases in side-chain thermal factors are observedas well. A smaller difference involves residues 66-67 which display an increase of 11 A 2 formain-chain atoms (average thermal factor of 23 A 2 versus 12 A2 in the wild-type protein).Here again, comparable increases in side-chain thermal factors are observed. Comparison ofthe N52A mutant protein and wild-type yeast iso-1-cytochrome c does not indicate the presenceof any significant differences in polypeptide chain flexibility.6.2.2 Heme structureThe heme group of wild-type yeast iso-l-cytochrome c is substantially distorted from planarityin both oxidation states (see Tables 6.24 and Table 3.8). The degree of heme distortion is morepronounced for the two reduced N52A and I75M mutant structures compared to the reducedwild-type protein, with average angular deviations of the four pyrrole rings from the pyrrolenitrogen plane being 3-4° higher. This increased distortion is similar to that observed for theN521 and Y67F variants (Chapters 4 and 5), as well as for the oxidized state of the wild-typeprotein (Chapter 3).Examination of heme coordinate bond lengths reveal that in the two mutant proteins thesebonds are the same within expected atomic coordinate errors. The orientation of the His18imidazole plane, which is sensitive to the oxidation state of the protein (Chapter 5), is similarin both the reduced N52A mutant and wild-type protein. The angle between the imidazoleplane normal and a line drawn through the NA and NC atoms is 48.7° and 46.7°, respectively.However, in the reduced I75M mutant structure this angle is 54.5° and therefore more like thatfound in the oxidized wild-type protein structure where it is 55.8°. An analysis of the hemesolvent accessibility of the two mutant proteins shows this to be comparable to wild-type yeastiso-l-cytochrome c (see Table 6.25).Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and 175M Mutants^124Table 6.24: Heme conformation and ligand geometry in wild-type, N52A and I75M yeastiso-l-cytochromes cPyrrole^Wild-type N52A^I75MI. Angular deviations between the pyrrole nitrogenplane normal and the four pyrrole ring normals (°)A 9.4 11.5 16.3B 11.1 14.0 10.1C 8.8 8.6 15.7D 8.1 13.2 12.2Average 9.3 11.8 13.6II. Angular deviations between the porphyrin ringplane normal and the four pyrrole ring plane nor-mals, plus the pyrrole nitrogen plane normal (°)A 6.7 7.4 10.4B 11.9 12.8 14.8C 9.8 12.6 13.6D 6.0 9.8 6.3Average 8.6 10.7 11.3NNNN 2.6 4.3 6.0III. Average deviations of porphyrin ring atoms fromthe least squares porphyrin ring plane (A)0.178^0.200^0.223Each pyrrole ring is defined by nine atoms, which include the five ring atoms plus the firstcarbon atom bonded to each ring carbon. The porphyrin ring plane is defined by the five atomsin each of the four pyrrole rings, the four bridging methine carbons, the first carbon atom ofeach of the eight side-chains and the heme iron (33 atoms in total). The pyrrole nitrogen planeis defined by only the 4 pyrrole nitrogens (see Figure 1.2 for heme atom labeling convention).6.2.3 The N52A mutation siteIn wild-type yeast iso-l-cytochrome c, Asn52 is part of two elaborate hydrogen bond networks,one involving Tyr67, Thr78, Met80 and an internal water molecule Wat166, and the second in-volving the pyrrole A propionate group, Gly41 and Trp59 (see Figures 6.48 and 6.49). Mutationof Asn52 to an alanine residue results in the enlargement of the cavity occupied by Wat166from , 10 A3 to ,70 A3 . This increase in volume is more than sufficient to allow for a secondChapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^125Table 6.25: Heme solvent accessibility for wild-type, N52A and I75M yeast iso-l-cytochromes etWild-type N52A I75M1. Solvent accessible heme atoms andsurface area exposed (A 2 )CHD 2.6 2.4 2.3CMC 9.4 10.4 9.3CAC 3.3 4.2 3.8CBC 17.4 20.6 17.6CMD 9.8 10.5 8.22. Total heme exposure (A 2 ) 42.5 48.1 41.23. Total heme surface (A 2 ) 495.7 504.3 503.04. Heme surface area exposed (% ) 8.6 9.5 8.2t Computations were done using the method of Connolly (1983) and the results represent theaccessible molecular surfaces of the atoms listed. The probe sphere used had a radius of 1.4 A.Figure 6.48: Stereo-drawing showing the region about residue 52 in the N52A (thick lines)and wild-type (thin lines) yeast iso-1-cytochromes c. This mutation substantially increases thevolume of an internal water cavity leading to the inclusion of an additional water molecule(Wat300). The hydrogen bonds (dashed lines) formed by Wat300 are similar to those formedby Asn52 ND2 in the wild-type structure.Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^126pyrrole Apropionatepyrrole ApropionateFigure 6.49: A stereo-drawing showing the region around the pyrrole A propionate group in theN52A (thick lines) and wild-type (thin lines) yeast iso-l-cytochromes c. Hydrogen bonds areindicated by dashed lines. In the N52A structure Wat300 is located in roughly the same positionas the Asn52 ND2 group in wild-type yeast iso-l-cytochrome c. However, unlike Asn52 ND2,Wat300 does not form a hydrogen bond to the pyrrole A propionate according to the standardcriteria used herein (see Table 6.26).water molecule to be positioned in this region and an additional water molecule (Wat300) isindeed observed in electron density maps. The location of Wat166 in the N52A structure isvirtually identical to that found in wild-type yeast iso-l-cytochrome c while the position ofWat300 corresponds roughly to that of the Asn52 ND2 group in the wild-type protein (dis-placement of 0.3 A ; see Figure 6.48). Both Wat166 and Wat300 have relatively high thermalfactors ( ,,,50 A2 ) indicating a fair degree of mobility in their positioning.The effect of this mutation on the hydrogen bond network about Wat166 is subtle (seeFigure 6.48). Wat300 forms a similar interaction with Wat166 in the N52A mutant structure asdid Asn52 ND2 in wild-type yeast iso-1-cytochrome c. Other hydrogen bonds formed by Wat166do not appear to be significantly perturbed by the presence of Wat300. However, unlike theamide group of Asn52, Wat300 could function either as a hydrogen bond donor or acceptor toWat166. This implies that the dipole orientation of Wat166 is likely to be less clearly definedin the mutant structure compared to that of the wild-type protein (Chapter 3).The effect of the N52A mutation on the hydrogen bond network about the pyrrole A pro-pionate is illustrated in Figure 6.49. In the wild-type protein Asn52 ND2 forms a hydrogenChapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and 175M Mutants^127Table 6.26: Heme propionate hydrogen bond interactions in wild-type, N52A and I75M yeastiso-l-cytochromes cHydrogen bond partnersDistances ()OfWild-type N52A I75M01A Tyr48 OH 2.83 2.89 2.82Wat121 2.81 2.97 3.37Wat 168 2.85 3.24 3.0302A G1y41 N 3.21 2.69 2.63Asn52 ND2 (Wat300)t 3.34 (3.56) (4.15)Trp59 NE1 3.09 2.82 3.04OlD Thr49 OG1 2.64 2.44 2.79Thr78 OG1 2.90 3.16 2.97Lys79 N 3.17 3.14 2.9002D Thr49 N 2.94 2.74 2.55Interactions were accepted as hydrogen bonds only if they met all of the following criteria:a H• • •A distance < 2.6 A, a D-H• • •A angle > 120°, and a C-A• • .11 angle > 90°. Valuesgiven in brackets are not considered to be hydrogen bonds by this criteria, but are listed forcomparison.t In the N52A mutant structure Asn52 ND2 is replaced by Wat300.bond to the 02A oxygen atom of this propionate. In the N52A variant Wat300 is positionedfurther from this group (3.56 A). This factor, coupled with its high thermal factor suggeststhat the interaction of Wat300 with the 02A oxygen atom of the pyrrole A propionate is weak.(Table 6.26). A consequence of this weaker interaction appears to be a stronger interactionbetween G1y41 N and the 02A heme propionate oxygen atom. Movement of G1y41 (averagedeviation 0.6 A; see Figure 6.46) shortens the hydrogen bond between these two groups by0.5 A to an overall length of 2.7 A. A similar situation is observed in the oxidized state of thewild-type protein where the Asn52 ND2 interaction to the heme is also lost (Chapter 3).Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^128Figure 6.50: Stereo-drawing showing the region about residue 75 in I75M (thick lines) andwild-type (thin lines) yeast iso-1-cytochromes c. The presence of the Met75 SD sulfur atomadds another group to the hydrogen bond network (dashed lines) around Wat166.6.2.4 The I75M mutation siteIn wild-type yeast iso-1-cytochrome c the side-chain of I1e75 forms part of one side of theinternal cavity occupied by Wat166. Substitution of this residue for a methionine effectivelyadds another group to the hydrogen bond network around Wat166 (see Figure 6.50). Thispermits new interactions to be made between Met75 SD and Wat166, Asn52 ND2, as well asThr78 0G1. Thus, one effect of this substitution is that the effective dipole orientation ofWat166 is likely modified.One group positionally affected by these new interactions is the side-chain of Asn52 whichshifts 0.6 A towards the Met75 sulfur atom. This shift breaks the hydrogen bond link betweenAsn52 ND2 and the nearby pyrrole A propionate, causing a conformational change in thislatter group (see Table 6.26 and Figure 6.51). In comparison to the reduced wild-type protein,the pyrrole A propionate torsion angles C2A—CAA, CAA—CBA and CBA—CGA in the I75Mvariant are rotated by ,25°, ,20° and ,20°, respectively. A similar alteration is also seenin the oxidized form of wild-type yeast iso-l-cytochrome c where these three torsion angleschange by ,20°, ,s-,30° and ,45°, respectively (Chapter 3). Another consequence of the lossof the interaction between Asn52 ND2 and the pyrrole A propionate is strengthening of theChapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and 175M Mutants^129pyrrole •^ pyrrole •ProPlanate ProPlanateFigure 6.51: A stereo-drawing showing the region around the pyrrole A propionate in I75M(thick lines) and wild-type (thin lines) yeast iso-1-cytochrome c. Hydrogen bonds are repre-sented by dashed lines. In the pyrrole A propionate region the I75M mutant structure showssimilarities to that of the oxidized form of the wild-type protein in that the hydrogen bondbetween the propionate group and Asn52 is lost and the propionate group has undergone aconformational change.interaction between Gly41 N and the 02A heme propionate oxygen atom similar to that in theN52A variant and oxidized wild-type structures (Chapter 3). As part of this process residue 41moves 0.6 A towards the propionate group decreasing the distance between them from 3.2 A to2.6 A (Table 6.26).6.3 Discussion6.3.1 Plasticity of the pyrrole A propionate regionA summary of the structural differences observed between the N52A and I75M mutantstructures and wild-type yeast iso-l-cytochrome c is given in Table 6.27. From this overviewit can be seen that most of the structural changes present are localized around the pyrrole Apropionate group. Previous structural studies of wild-type and mutant forms of yeast iso-l-cytochrome c have also shown structural differences in this area. For example, replacement ofresidues 67 (Chapter 4; Figure 4.36) and 82 (Louie et al., 1988b; Louie & Brayer, 1989) resultin such conformational changes, as does going to the oxidized state in the case of the wild-typeN52A^I75MA. Positional displacements of polypeptide chain (see Figure 6.46)1. Movement of G1y41 towards the pyrrole A^1. Movement of Gly41 towards the pyrrole Apropionate^ propionateB. Thermal factor parameters of main-chain atoms (see Figure 6.47)1. No major differences^ 1. Increased thermal factors for residues 57-59 and 66-67C. Heme structure (see Figures 6.49 and 6.51 and Tables 6.24 and 6.26)1. Increased distortion of the heme plane^1. Increased distortion of the heme plane2. Small readjustments in the pyrrole A pro-^2. Rotation of the His18 imidazole planepionate hydrogen bond network^comparable to that of the oxidized wild-type structure3. Pyrrole A propionate conformationalchanges similar to those of the oxidizedwild-type structureD. Mutation site (see Figures 6.48 and 6.50).Additional water molecule (Wat300) takesthe place of original Asn52 ND22. Link lost to pyrrole A propionate1. Addition of an extra hydrogen bondinggroup to the Wat166 hydrogen bond net-work2. Link lost to pyrrole A propionateE. Hydrogen bond interactions (see Figures 6.48, 6.49, 6.50 and 6.51 and Table 6.26).Stronger: G1y41 N - Heme 02A Stronger: G1y41 N- Heme 02ALost: Asn52 ND2- Heme 02A Weaker: Wat121- Heme 01AAsn52 ND2- Wat166 Lost: Asn52 ND2- Heme 02ANew: Wat300- Wat166 New: Met75 SD- Asn52 ND2Met75 SD- Thr78 0G1Met75 SD- Wat166Chapter 6. Reduced Yeast Iso-1-Cytochrome c N52A and I75M Mutants^130Table 6.27: Structural differences observed in the reduced N52A and I75M mutant structureswhen compared to reduced wild-type yeast iso-l-cytochrome cprotein (see Figure 3.25). It is of interest that mutation of Asn52 to an isoleucine inhibits theconformational changes occurring upon heme oxidation and locks the hydrogen bond networkaround the pyrrole A propionate in the reduced state (see Figure 5.42 and Table 5.21).The N52A mutant studied here displays some of the same differences as seen in the oxidizedChapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and 175M Mutants^131wild-type structure. The hydrogen bond between the heme 02A oxygen atom and residue 52is lost and a strengthening of the hydrogen bond between this oxygen atom and G1y41 N isobserved (see Figure 6.49 and Table 6.26). Similar conformational differences are observed inthe I75M mutant structure, with the additional feature of increased thermal motion for Trp59.As observed in the oxidized wild-type structure, altered torsional angles for the pyrrole Apropionate group are also present in the I75M mutant structure (see Figure 6.51).Apparently the pyrrole A propionate region can adopt a number of stable conformations.Which of these conformations has the lowest energy, and thus predominates, would appear todepend on a number of factors. One factor may be the polarity of the heme environment (Louieet al., 1988; Chapter 4). Mutant structures in the reduced state such as F82S, F82G and toa lesser degree Y67F, which increase the hydrophilicity of the environment around the heme,exhibit pyrrole A propionate conformations which are a hybrid of those found in the reducedand oxidized wild-type structures. Further, the oxidized wild-type structure is also an exampleof the effect a change in polarity of the heme environment has on the pyrrole A propionateregion. On the other hand, mutations which create a more hydrophobic heme environmentsuch as in the N521 and N52I-Y67F proteins result in conformations of the pyrrole A propionatewhich are very similar to that seen in the wild-type reduced state, even when the heme ironatoms of these proteins are oxidized (Chapter 5). Collectively these studies show pyrrole Apropionate conformation could potentially be used as a rough barometer of the polarity of theheme environment.The structural mechanism linking the polarity of heme environment and pyrrole A pro-pionate conformation could potentially be electrostatic or mechanical. Louie et al., (1988)proposed that a change in the polarity of the heme environment might cause a redistribution ofthe electrons within the delocalized 7-electron system of the heme (see also Section 4.3; Wallaceet al., 1989). This in turn would alter the electrostatic interaction between the porphyrin ringand the propionate carboxyl group and result in a conformational change in the latter group ashas been suggested by Moore (1983). A more indirect explanation is also possible. DifferencesChapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and 175M Mutants^132in the polarity of the heme environment might be reflected in different degrees of distortionfrom planarity of the heme porphyrin ring. The need to accommodate heme plane distortionwithin the surrounding protein matrix could necessitate a change in pyrrole A propionate con-formation. Our studies show that a change in pyrrole A propionate conformation is alwaysaccompanied by an increase in the distortion of the heme plane (see Tables 3.8, 4.12, 5.20 and6.24).Although polarity of the heme environment appears to be an important factor in determin-ing the exact conformation of the pyrrole A propionate group and associated hydrogen bondnetwork, the N52A and I75M mutant structures presented here show that this is clearly not theonly factor. Inspection of these two mutant structures does not reveal an increase in either thepolarity of the heme pocket or the solvent exposure of the heme porphyrin ring (Table 6.25).In these cases it must be concluded that the observed differences in pyrrole A propionate con-formation are caused by other factors. For example, it is possible that these rearrangementsoccur as a result of breaking of the hydrogen bond between residue 52 and the heme pyrrole Apropionate group in the two mutant structures. Clearly the factors governing pyrrole A propi-onate conformation are complex and cannot be assigned to a single parameter such as polarityof the heme environment.6.3.2 Functional role of the internal water molecule Wat166The effect of the mutations, N52A and I75M, on the positions of atoms involved in the hydrogenbond network about the internal water molecule, Wat166, appears minimal (see Figures 6.48 and6.50). However, an analysis of the hydrogen bonds present does indicate that some subtle butimportant changes have occurred. In wild-type yeast iso-l-cytochrome c the Asn52 ND2 groupfunctions as a hydrogen bond donor to Wat166. As a result of the mutation of this residue to analanine the function of the Asn52 ND2 group is taken over by a new water molecule (Wat300).This water molecule can both donate or accept a hydrogen bond from Wat166, effectively bothlowering the hydrogen bond donor capacity, and at the same time increasing the hydrogenMet 80\'0--(104Tyr 67Wat 166Fe3+z.,Met 80Fe'Tyr 67t(VWat 166Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^133Reduced^OxidizedFigure 6.52: A schematic representation of the proposed water-switch mechanism for stabiliza-tion of the alternate oxidation states of yeast iso-l-cytochrome c.bond acceptor capacity on the side of Wat166 furthest away from Tyr67 (Figure 6.48). Thereplacement of 11e75 with a methionine residue produces similar results but only increases thehydrogen bond acceptor capacity on the side of Wat166 that is away from Tyr67 (Figure 6.50).These results suggest that as a result of the N52A and I75M mutations the precise orientationof Wat166 is likely to be affected.The orientation of Wat166 has been proposed to be crucial for the stabilization of the twooxidation states of cytochrome c (Chapter 3). In the reduced state Wat166 is oriented so thatthe maximum number of hydrogen bonds are formed in this region. In the oxidized form itis proposed that the dipole moment of Wat166 realigns in the electrostatic field generated bythe more positively charged heme iron atom. As a consequence of this change in orientationthe hydrogen bond between Tyr67 OH and the Met80 SD ligand is broken. The loss of thishydrogen bond is an additional stabilizing feature in the oxidized state of the protein since itlessens the electron withdrawing power of the Met80 ligand (Figure 6.52; see also Figure 3.28).If this proposed "water-switch" mechanism for the stabilization of the two oxidation statesof cytochrome c is valid then alteration of the orientation of Wat166 should affect the equilib-rium constant for oxidation of yeast iso-l-cytochrome c. This is equivalent to modifying theChapter 6. Reduced Yeast Iso-1-Cytochrome c N52A and 175M Mutants^134observed midpoint reduction potential of the protein. The two mutants discussed here test thishypothesis. As has been described above, both the N52A and I75M mutations increase thehydrogen bond acceptor capacity in a region adjacent to Wat166 and furthest away from theTyr67 hydroxyl group. This should result in Tyr67 OH functioning less as a hydrogen acceptorand more as a hydrogen donor group to Wat166 since the hydrogen atom of Wat166 interactingwith the Tyr67 side-chain can also interact with Wat300 in the N52A mutant or with Met75 SDin the I75M mutant. As a consequence, the hydrogen bond between Tyr67 OH and Met80 SDis weakened and the Wat166 hydrogen bond network will tend more towards the conformationseen in the oxidized state of the wild-type protein. Accordingly, the proposed water-switchmechanism suggests the net effect is that the midpoint reduction potential for these mutantproteins should be lower.As predicted by this mechanism, both mutant proteins do indeed show a drop in midpointreduction potential. The N52A protein has a midpoint reduction potential 33 mV lower thanthat of wild-type yeast iso-l-cytochrome c and for the I75M variant the midpoint reductionpotential is 45 mV lower (Table 1.5; Rafferty, 1992). The magnitude of the drop in midpointreduction potential observed is also in the range expected. Previous studies have shown thatcomplete removal of the Tyr67 OH - Met80 SD hydrogen bond results in an overall drop of,,,56 mV (Chapters 4 and 5), and this therefore represents the maximum limit expected formodifications affecting this hydrogen bond.These results support the proposed role of Wat166 in stabilizing the two oxidation statesof yeast iso-l-cytochrome c. However, it is likely that this water-switch mechanism is notonly restricted to yeast iso-l-cytochrome c. Model-building studies with the five other highresolution eukaryotic cytochrome c structures solved to date [reduced and oxidized tuna (Takano& Dickerson, 1981a,b); oxidized rice (Ochi et al., 1983); oxidized horse (Bushnell et al., 1990);and reduced yeast iso-2 (Murphy et al., 1992)] show that the proposed oxidation state dependentorientation of Wat166 is likely a common feature of all these proteins (Chapter 3; see alsoFigure 3.28). Further, comparison of 94 eukaryotic cytochrome c sequences (Moore & Pettigrew,Chapter 6. Reduced Yeast Iso-l-Cytochrome c N52A and I75M Mutants^1351990; see also Section 1.2) has shown that the four key residues involved, Asn52, Tyr67, Thr78and Met80 are invariant except for one case where Thr78 is replaced by an asparagine residue(Amati et al., 1988). It can be concluded that in most eukaryotic cytochromes c a comparablehydrogen bond network is present to that in yeast iso-l-cytochrome c and that stabilization ofthe different oxidation states for these cytochromes c is similar to the water-switch mechanismdescribed herein.SummaryThe first objective of the work described in this thesis was to identify the structural differ-ences that exist between the two oxidation states of cytochrome c. This was accomplishedby determining the structure of the oxidized state of yeast iso-l-cytochrome c using a crystalform isomorphous to that of the reduced protein whose structure had been previously solvedto high resolution (Louie Si Brayer, 1990). The use of isomorphous crystalline material andsimilar structure refinement methodologies for both oxidation states of yeast iso-1-cytochrome callowed for a comparative analysis to be made in the absence of possible systematic errors intro-duced due to differing structure determination approaches. These results showed that oxidationstate differences are expressed for the most part as increased mobility for selected segments ofpolypeptide chain in the oxidized state, rather than as discrete positional shifts of atoms. Thiscorrelates well with a large body of data from a wide variety of techniques which suggest thatthe oxidized form of cytochrome c is more open and less rigid than that of the reduced state(see Section 1.3 for references).Specifically, three regions of polypeptide chain in yeast iso-l-cytochrome c displayed anincrease in flexibility in the oxidized state of the protein, and include residues 47-59, 65-72 and81-85, with maximal increases being observed for Asn52, Tyr67 and Phe82. The side-chainsof two of these residues are hydrogen bonded to the internal water molecule, Wat166, whichshowed a 1.7 A shift towards the heme iron atom in the oxidized state of the protein. Furtherstudy of this water molecule suggested that it might be a major factor in stabilizing bothoxidation states through differential orientation of its dipole moment, shift in distance to theheme iron atom and alteration of the surrounding hydrogen bond network. Comparison betweenoxidation states also revealed some subtle differences in the conformation of the heme pyrrole Apropionate and the surrounding hydrogen bond network that had been previously postulated136Summary^ 137to occur (Moore, 1983). It also demonstrated that heme planar distortion and the orientationof the imidazole plane of the Hisl8 ligand is dependent on oxidation state (Chapter 3).Once the structural differences between the two oxidation states of cytochrome c wereidentified, the next objective was to determine the role of these observed differences in thebiological function of cytochrome c. The method of choice to assess the importance of thesedifferences was to study the structures of yeast iso-1-cytochrome c variants with mutations inthose regions of the protein that demonstrated oxidation state dependent conformations. Fivemutants were selected for this study: N52A, N521, Y67F, N52I-Y67F and 175M. All of thesemutations are located directly adjacent to the internal water molecule, Wat166, which studies ofthe wild-type protein had suggested was likely to play a central role in stabilizing the alternateoxidation states of cytochrome c. A large body of functional data was also available for thesemutants (see Section 1.5) which allowed for a careful examination of the impact of structuralalterations on functional properties.In total eight structures were determined to high resolution. The N52I, Y67F and N52I-Y67F yeast iso-l-cytochrome c variants were determined in both oxidation states, while theN52A and I75M mutant proteins were resolved in the reduced form of the protein. All thesemutations were observed to cause perturbations in the hydrogen bond network about Wat166.In the N52I and N52I-Y67F variants the internal cavity occupied by Wat166 was filled inresulting in the exclusion of this internal water molecule from these structures. The N52Aand Y67F mutations affected the hydrogen bond network in a more subtle way in that theAsn52 ND2 and Tyr67 OH groups which form hydrogen bonds to Wat166 were replaced bywater molecules. The exchange of 11e75 for a methionine residue resulted in an extra hydrogenbond acceptor group being added to the hydrogen bond network about Wat166. For the threemutants whose structures were determined in both oxidation states (N52I, Y67F and N52I-Y67F), none demonstrated the increased polypeptide chain flexibility observed in the oxidizedstate of the wild-type protein. Besides these changes, all mutant structures also revealed alteredconformations about the pyrrole A propionate group.Summary^ 138Correlation of the structural data provided by the wild-type and mutant structures withfunctional studies suggests the role of Wat166 is three fold. First, the presence of Wat166provides a convenient mechanism to modify the hydrogen bond network involving several keyresidues near the Met80 ligand, depending on the oxidation state of the protein. In the reducedstate this water molecule is oriented so that the maximum number of hydrogen bonds are formedin this region. In the oxidized form Wat166 realigns in the electrostatic field generated by themore positively charged heme iron atom. As a consequence of this change in orientation severalhydrogen bonds are broken, including the interaction between Tyr67 OH and the Met80 SDligand. This latter interaction has also been shown to be a factor in controlling the midpointreduction potential of cytochrome c by influencing the electron withdrawing power of the Met80ligand. In the N52I, Y67F and N52I-Y67F variants where this hydrogen bond is absent, themidpoint reduction potential has decreased by ,56 mV (Chapters 4 and 5). Examination of theN52A and I75M mutant proteins has shown that the strength of the Tyr67 OH to Met80 SDhydrogen bond is strongly correlated to the dipole orientation of Wat166 (Chapter 6). In asecond role, related to the first, Wat166 is required to maintain the spatial and hydrogen bondrelationships in this vicinity of cytochrome c. Absence of this internal water molecule, asdemonstrated in the N52I mutant protein, leads to shifts in such critical residues as Tyr67 andsubsequent rearrangements in the hydrogen bond network (Chapter 5).Finally, Wat166 also appears to modulate the oxidation state dependent mobility differencesin polypeptide chain segments that have been observed between the reduced and oxidizedstructures of cytochrome c. Although the importance of this phenomena is not fully understood,our studies suggest that it might play a role in the complexation and dissociation of cyto-chrome c with redox partners. An unusual aspect of the outer solvent surface presentationof those segments of polypeptide chain having greater conformational mobility in the oxidizedstate of the protein is that they are bisected by an unperturbed stretch of polypeptide chain(residues 73-80). This has led to the proposal that this highly conserved segment of polypeptidechain could act as a push-button contact trigger operated by protein-protein contacts with redoxSummary^ 139partners. The postulated function of this trigger is to initiate the necessary structural changesrequired to switch between oxidation states and in this way facilitate electron transfer.In conclusion, these studies provide a solid foundation for further analyses of structure-function relationships in cytochrome c targeted at gaining a deeper understanding of the electrontransfer reaction mediated by this protein.BibliographyAbrams, R., Altshul, A.M. & Hogness, T.R. (1942). Cytochrome c peroxidase II. Theperoxidase-hydrogen peroxidase complex. J. Biol. Chem., 142, 303-316.Altschul, A.M., Abrams, R. & Hogness, T.R. (1940). Cytochrome c peroxidase. J. Biol. Chem.,136, 777-794.Amati, B.B., Goldsmidth-Clermont, M., Wallace, C.J.A. & Rochaix, J. (1988). cDNA and de-duced amino acid sequences of cytochrome c from Chlamydomonas reinhardtii: unexpectedfunctional and phylogenetic implications. J. Mol. Evol, 28, 151-160.Arndt, U.W. & Wonacott, A.J., eds. (1977). The Rotation Method in Crystallography. North-Holland Publishing Company, Amsterdam, New York, Oxford.Bach, S.J., Dixon, M. & Keilin, D. (1942a). A new soluble cytochrome component from yeast.Nature (London), 149, 21.Bach, S.J., Dixon, M. & Zerfas, L.G. (1942b). Lactic dehydrogenase of yeast. Nature (London),149, 48-49.Barker, P.D., Mauk, M.R. & Mauk, A.G. (1991). Proton titration curve of yeast iso-l-ferricytochrome c. Electrostatic and conformational effects of point mutations. Biochem-istry, 30, 2377-2383.Barlow, D.J. & Thornton, J.M. (1983). Ion-pairs in proteins. J. Mol. Biol., 168, 867-885.Berghuis, A.M. & Brayer, G.D. (1992). Oxidation state-dependent conformational changes incytochrome c. J. Mol. Biol., 223, 959-976.Bernstein, F.C., Koetzle, T.F., Williams, G.J.B, Meyer, E.F., Bruce, M.D., Rodgers, J.R.,Kennard, 0., Shimanouchi, T. & Tasmui, M. (1977). The protein databank• a computer-based archival file for macromolecular structures. J. Mol. Biol., 112, 535-542.Blundell, T.L. & Johnson, L.N. (1976). Protein Crystallography. Academic Press, New York,London, San Francisco.Boon, P.J., Van Raay, A.J.M., Tesser, G.I. & Nivard, R.J.F. (1979). Semi-synthesis, confor-mation and cytochrome c oxidase activity of eight cytochrome c analogues. FEBS Letters,108, 131-135.Bosshard, H.R. & Zurrer, M. (1980). The conformation of cytochrome c in solution. J. Biol.Chem., 255, 6694-6699.140BIBLIOGRAPHY^ 141Bott, R. (1991). Assessing the accuracy of subtilisin BPN. In Abstracts of the Annual meetingof the American Crystallographic Association, vol. 19 of 2, p. 36, Toledo, Ohio. AmericanCrystallographic Association.Brayer, G.D. & Murphy, M.E.P. (1993). Structural studies of eukaryotic cytochromes c. In TheCytochrome c Handbook (Mauk, A.G. & Scott, R.A., eds.). Plenum Press, New York. Inpress.Burch, A.M., Rigby, S.E.J., Funk, W.D., MacGillivray, R.T.A., Mauk, M.R., Mauk, A.G. &Moore, G.R. (1990). NMR characterization of surface interactions in the cytochrome b 5-cytochrome c complex. Science, 247, 831-833.Burrows, A.L., Guo, L-H., Hill, A.O., McLendon, G. & Sherman, F. (1991). Direct electro-chemistry of proteins; Investigations of yeast cytochrome c mutants and their complexeswith cytochrome b 5 . Eur. J. Biochem., 202, 543-549.Bushnell, G.W., Louie, G.V. & Brayer, G.D. (1990). High-resolution three-dimensional struc-ture of horse heart cytochrome c. J. Mol. Biol., 214, 585-595.Butt, W.D. & Keilin, D. (1962). Absorption spectra and some other properties of cytochromec and of its compounds with with ligands. Proc. Royal Soc. London, B152, 429-458.Chambers, J.L. & Stroud, R.M. (1979). The accuracy of refined protein structures: comparisonof two independently refined models of bovine trypsin. Acta Crystallogr. sect. B, 35, 1861-1874Churg, A.K. & Warshel, A. (1986). Control of the redox potential of cytochrome c and micro-scopic dielectric effects in proteins. Biochemistry, 25, 1675-1681.Churg, A.K. Weiss, R.M. & Warshel, A. (1983). On the action of cytochrome c: correlatinggeometry changes upon oxidation with activation energies of electron transfer. J. Phys.Chem., 87, 1683-1694.Cohen, H.J. & Fridovich, I. (1971). Hepatic sulfite oxidase: purification and properties. J. Biol.Chem., 246, 359-366.Cokic, P. & Erman, J.E. (1987). The Effect of complex formation upon the reduction rates ofcytochrome c and cytochrome c peroxidase compound II. Biochim. Biophys. Acta, 913,257-271.Connolly, M.L. (1983). Solvent-accessible surfaces of proteins and nucleic acids. Science, 221,709-713.Creighton, T.E. (1984). Proteins: Structure and Molecular Principles. Freeman and Company,New York.Cruickshank, D.W.J. (1949). The accuracy of electron-density maps in x-ray analysis withspecial reference to dibenzyl. Acta Crystallogr., 2, 65-82.BIBLIOGRAPHY^ 142Cruickshank, D.W.J. (1954). The accuracy of electron-density maps in x-ray analysis: correc-tion. Acta Crystallogr., 7,519.Cruickshank, D.W.J. (1985). Fourier synthesis and structure factors. In International Tablesfor X-ray Crystallography (Kasper, J.S. & Lonsdale, K., eds.), vol. 2, pp. 317-339. D.Reidel Publishing Company, Dordrecht, 3rd edition.Cusanovich, M.A., Meyer, T.E. & G., Tollin. (1988). c-Type cytochromes: oxidation-reductionproperties. In Herne Proteins (Eichhorn, G.L. & Marzilli, J.G., eds.), vol. 7 of Advancesin Inorganic Biochemistry, pp. 37-92. Elsevier, New York.Cutler, R.L., Davies, A.M., Creighton, S., Warshel, A., Moore, G.R., Smith, M. & Mauk,A.G. (1989). Role of arginine-38 in regulation of the cytochrome c oxidation- reductionequilibrium. Biochemistry, 28, 3188-3197.Cutler, R.L., Pielak, G.J., Mauk, A.G. & Smith, M. (1987). Replacement of cysteine-107of Saccharomyces cerevisiae iso-l-cytochrome c with threonine: improved stability of themutant protein. Protein Eng., 1, 95-99.Das, G., Hickey, D.R., McLendon, D., McLendon, G. & Sherman, F. (1989). Dramatic ther-mostabilization of yeast iso-l-cytochrome c by an asparagine isoleucine replacement atposition 57. Proc. Nat. Acad. Sci., U.S.A., 86, 496-499.Dayhoff, M.O., Schwartz, R.M. & Orcutt, B.C. (1976). In Atlas of Protein Sequence andStructure (Dayoff, M.O., ed.), vol. 5, pp. 20-49. National Biomedical Research Foundation,Washington, D.C.Dickerson, R.E. & Timkovich, R. (1975). Cytochromes c. In The Enzymes (Boyer, P.D., ed.),vol. 11, part A, pp. 397-547. Academic Press, New York, 3rd edition.Dong, A., Huang, P. & Caughey, W.S. (1992). Redox-dependent changes in 13-extended chainand turn structures of cytochrome c in water solution determined by second derivativeamide I infrared spectra. Biochemistry, 31, 182-189.Eden, D., Matthew, J.B., Rosa, J.J. & Richards, F.M. (1982). Increase in apparent compress-ibility of cytochrome c upon oxidation. Proc. Nat. Acad. Sci., U.S.A., 79, 815-819.Fauman, E.B. & Stroud, R.M. (1991). An empirical approach to analysis of errors in x-rayprotein structures by reference to atomic B factors. In Abstracts of the Annual meetingof the American Crystallographic Association, vol. 19 of 2, p. 34, Toledo, Ohio. AmericanCrystallographic Association.Feng, Y. & Englander, S.W. (1990). Salt-dependent structure change and ion binding in cy-tochrome c studied by two-dimensional proton NMR. Biochemistry, 29, 3505-3509.Feng, Y., Roder, H. & Englander, S.W. (1990). Redox-dependent structure change and hyper-fine nuclear magnetic resonance shifts in cytochrome c. Biochemistry, 29, 3494-3504.BIBLIOGRAPHY^ 143Fisher, W.R., Tanuichi, H. & Anfinsen, C.B. (1973). On the role of heme in the formation ofthe structure of cytochrome c. J. Biol. Chem., 248, 3188-3195.Frauenhoff, M.M. & Scott, R.A. (1992). The role of tyrosine 67 in the cytochrome c hemecrevice structure studied by semisynthesis. Proteins: Struct. Funt. Genet., 14, 202-212.Gadsby, P.M.A., Peterson, J., Foote, N., Greenwood, C. & Thomson, A.J. (1987). Identifica-tion of the ligand-exchange process in the alkaline transition of horse heart cytochrome c.Biochem. J., 246, 43-54.Gao, Y., Boyd, J., Pielak, G.J. & Williams, R.J.P. (1991). Comparison of reduced and oxidizedyeast iso-l-cytochrome c using proton paramagnetic shifts. Biochemistry, 30, 1928-1934.Giege, R. & Mikol, V. (1989). Crystallogenesis of proteins. Trends Biochem. Sci., 7, 277-282.Hampsey, D.M., Das, G. & Sherman, F. (1986). Amino acid replacements in yeast iso-l-cytochrome c, comparison with the phylogenetic series and the tertiary structures of relatedcytochromes c. J. Biol. Chem., 261, 3259-3271.Hendrickson, W.A. (1985). Stereochemically restrained refinement of macromolecular struc-tures. In Methods in Enzymology (Wyckoff, H.W., Hirs, C.H.W. & Timasheff, S.N., eds.),vol. 115 B, pp. 252-270. Academic Press, Inc., Orlando.Hendrickson, W.A. & Konnert, J. (1980). Incorporation of stereochemical information intocrystallographic refinement. In Computing in Crystallography (Diamond, R., Ramaseshan,S. & Venkatesan, K., eds.), pp. 13.01-13.26. Indian academy of sciences, Bangalore.Hendrickson, W.A. & Konnert, J. (1981). Stereochemically restrained crystallographic least-squares refinement of macromolecular structures. In Biomolecular Structure, Function,Conformation and Evolution (Sunivasan, R., ed.), vol. 1, pp. 43-57. Pergamon Press,Oxford.Hickey, D.R., Berghuis, A.M., Lafond, G., Jaeger, J.A., Cardillo, T.S., McLendon, D., Das, G.,Sherman, F., Brayer, G.D. & McLendon, G. (1991). Enhanced thermodynamic stabilitiesof yeast iso-l-cytochrome c with amino acid replacements at position 52 and 102. J. Biol.Chem., 266, 11686-11694.Hildebrandt, P. & Stockburger, M. (1989). Cytochrome c at charged interfaces. 2. Com-plexes with negatively charged macromolecular systems studied by resonance Raman spec-troscopy. Biochemistry, 28, 6722-6728.Ito, A. (1980a). Cytochrome b 5-like hemoprotein of outer mitochondrial membrane; OM Cy-tochrome b I: Purification of OM cytochrome b from rat liver mitochondria and comparisonof its molecular properties with those of cytochrome b 5 . J. Biochem. (Japan), 87, 63-71.Ito, A. (1980b). Cytochrome b5-like hemoprotein of outer mitochondrial membrane; OM Cy-tochrome b II. Contribution of OM cytochrome b to rotenone insensitive NADH-cyt creductase activity. J. Biochem. (Japan), 87, 73-80.BIBLIOGRAPHY^ 144Kassner, R.J. (1972). Effects of nonpolar environments on the redox potentials of heme com-plexes. Proc. Nat. Acad. Sci., U.S.A., 69, 2263-2267.Kassner, R.J. (1973). A theoretical model for the effects of local nonpolar heme environmentson the redox potentials of cytochromes. J. Amer. Chem. Soc., 95, 2674-2677.Koloczek, H., Horie, T., Yonetani, T., Anni, H., Maniara, G. & Vanderkooi, J.M. (1987).Interaction between cytochrome c and cytochrome c peroxidase: excited-state reactions ofzinc- and tin-substituted derivatives. Biochemistry, 26, 3142-3148.Komar-Panicucci, S., Bixler, J., Bakker, G., Sherman, F. & McLendon, G. (1992). Tuning theredox potential of cytochrome c through synergistic site replacements. J. Amer. Chem.Soc., 114, 5443-5445.Korzun, Z.R., Moffat, K., Frank, K. & Cusanovich, M.A. (1982). Extended X-ray absorption finestructure studies of cytochrome c: structural aspects of oxidation-reduction. Biochemistry,21, 2253-2258.Koul, A.K., Folena-Wasserman, G. & Warme, P.K. (1979). Semi-synthetic analogs of cy-tochrome c at positions 67 and 74. Biochem. Biophys. Res. Commoun., 89, 1253-1259.Kreil, G. (1965). Die C-terminale Aminosauresequenz des Tunfischcytochroms c. Z. Physiol.Chemie., 340, 86-87.Langen, R., Brayer, G.D., Berghuis, A.M., McLendon, G., Sherman, F. & Warshel, A. (1992).Effect of the Asn52-41e mutation on the redox potential of yeast cytochrome c: Theoryand experiment. J. Mol. Biol., 224, 589-600.Lederer, F., Ghrir, R., Guiarad, B., Cortial, S. & Ito, A. (1983). Two homologous cytochromesb5 in a single cell. Eur. J. Biochem., 132, 95-102.Leung, C.J., Nall, B.T. & Brayer, G.D. (1989). Crystallization of yeast iso-2-cytochrome cusing a novel hair seeding technique. J. Mol. Biol., 206, 783-785.Liang, N., Mauk, A.G., Pielak, G.J., Johnson, J.A., Smith, M. & Hoffman, B. (1988). Regu-lation of interprotein electron transfer by residue 82 of yeast cytochrome c. Science, 240,311-313.Liang, N., Pielak, G.J., Mauk, A.G., Smith, M. & Hoffman, B.M. (1987). Yeast cytochromec with phenylalanine or tyrosine at position 87 transfers electrons to (zinc cytochrome cperoxidase)+ at a rate ten thousand times that of the serine-87 or glycine-87 variants. Proc.Nat. Acad. Sci., U.S.A., 84, 1249-1252.Liu, G., Crygon, C.A. & Spiro, T.G. (1989). Ionic strength dependence of cytochrome c struc-ture and ultraviolet resonance Raman spectroscopy. Biochemistry, 28, 5046-5050.Louie, G.V. (1990). Structural studies of wild-type and variant yeast iso-1-cytochromes c. PhDthesis, University of British Columbia, Vancouver, Canada.BIBLIOGRAPHY^ 145Louie, G.V. & Brayer, G.D. (1989). A polypeptide chain-refolding event occurs in the Gly82variant of yeast iso-l-cytochrome c. J. Mol. Biol., 210, 313-322.Louie, G.V. & Brayer, G.D. (1990). High-resolution refinement of yeast iso-l-cytochrome c andcomparisons with other eukaryotic cytochromes c. J. Mol. Biol., 214, 527-555.Louie, G.V., Hutcheon, W.L.B. & Brayer, G.D. (1988a). Yeast iso-l-cytochrome c: a 2.8Aresolution three-dimensional structure determination. J. Mol. Biol., 199, 295-314.Louie, G.V., Pielak, G.J., Smith, M. & Brayer, G.D. (1988b). The role of Phe-82 in yeastiso-l-cytochrome c and remote conformational changes induced by a serine residue at thisposition. Biochemistry, 27, 7870-7876.Lum, V.R., Brayer, G.D., Louie, G.V., Smith, M. & Mauk, A.G. (1987). Computer modellingof yeast iso-l-cytochrome c - yeast cytochrome c peroxidase complexes. Protein Struct.,Folding Design, 2, 143-150.Luntz, T.L., Schejter, A., Garber, E.A.E. & Margoliash, E. (1989). Structural significanceof an internal water molecule studied by site-directed mutagenesis of tyrosine-67 in ratcytochrome c. Proc. Nat. Acad. Sci., U.S.A., 86, 3524-3528.Luzzati, V. (1952). Traitement statistique des erreurs dans la determination des structurescristallines. Acta Crystallogr., 5, 802-810.Marchon, J.C., Mashiko, T. & Reed, C.A. (1982). How does nature control cytochrome redoxpotentials. In Electron Transport and Oxygen Utilization (Ho, C., ed.), pp. 67-72. Elsevier,Amsterdam.Marcus, R.A. & Sutin, N. (1985). Electron transfer in chemistry and biology. Biochim. Biophys.Acta, 811, 265-322.Margalit, R. & Schejter, A. (1973). Cytochrome c: a thermodynamic study of the relationshipsamong oxidation state, ion-binding and structural parameters; 1. The effects of tempera-ture, pH and electrostatic media on the standard redox potential of cytochrome c. Eur. J.Biochem., 32, 492-499.Margoliash, E. (1990). Was cytochrome c properly engineered in the first place. In Abstractsof the C5 Symposium; Control of charge transfer in cytochrome and chlorophyll complexes,p. 8, Montreal, Quebec, Canada. International Union for Pure and Applied Biophysics.Margoliash, E. & Bosshard, H.R. (1983). Guided by electrostatics, a textbook protein comesof age. Trends Biochem. Sci., 8, 316-320.Margoliash, E. & Schejter, A. (1966). Cytochrome c. Advan. Protein Chem., 21, 113-286.Margoliash, E., Smith, E.L., Kreil, G. & Tuppy, H. (1961). Amino Acid Sequence of HorseHeart Cytochrome c. Nature (London), 192, 1125-1127.BIBLIOGRAPHY^ 146Mashiko, T., Reed, C.A., Haller, K.J., Kastner, M. E. & Scheidt, W.R. (1981). Tioetherligation in iron-porphyrin complexes: models for cytochrome c. J. Amer. Chem. Soc., 103,5758-5767.Mathews, F.S. (1985). The structure, function and evolution of cytochromes. In Progress inBiophysics and Molecular Biology (Noble, D. & Blundell, T.L., eds.), vol. 45, pp. 1-56.Pergamon Press, Oxford.Mauk, M.R., Barker, P.D. & Mauk, A.G. (1991). Proton linkage of complex formation betweencytochrome c and cytochrome b 5 : electrostatic consequences of protein-protein interactions.Biochemistry, 30, 9873-9881.McLendon, G., Hickey, D.R., Berghuis, A.M., Sherman, F. & Brayer, G.D. (1991). Effects ofreaction free energy in biological electron transfer In Vitro and In Vivo. In Electron Transferin Inorganic, Organic and Biological Systems (Bolton, J.R., Mataga, N. & McLendon, G.,eds.), vol. 228 of Advances in Chemistry Series, pp. 179-189. American Chemical Society.McLendon, G. & Miller, J.R. (1985). The dependence of biological electron transfer rateson exothermicity: the cytochrome c/cytochrome b 5 couple. J. Amer. Chem. Soc., 107,7811-7816.McLeod, R.M., Farkas, W., Fridovich, I. & Handler, P. (1961). Purification and properties ofhepatic sulfite oxidase. J. Biol. Chem., 236, 1841-1846.McPherson, A (1990). Current approaches to macromolecular crystallization. Eur. J. Biochem.,189, 1-23.Montgomery, D., Leung, D., Smith, M., Shalit, P. Faye, G. & Hall, B (1980). Isolation andsequence of a gene for iso-2-cytochrome c in Saccharomyces cerevisiae. Proc. Nat. Acad.Sci., U.S.A., 77, 541-545.Moore, G.R. (1983). Control of redox properties of cytochrome c by special electrostatic inter-actions. FEBS Letters, 161, 171-175.Moore, G.R. & Pettigrew, G.W. (1990). Cytochrome c. Evolution, Structural and Physiochem-ical Aspects. Springer-Verlag, Berlin.Moore, G.R., Pettigrew, G.W. & Rogers, N.K. (1986). Factors influencing redox potentials ofelectron transfer proteins. Proc. Nat. Acad. Sci., U.S.A., 83, 4998-4999.Moore, G.R. & Williams, R.J.P. (1977). Structural basis for the variation in redox potential ofcytochromes. FEBS Letters, 79, 229-232.Mori, E. & Morita, Y. (1980). Amino Acid Sequence of Cytochrome c from Rice. J. Biochem.(Tokyo), 87, 249-266.Murphy, M.E.P (1993). Structural constraints for the folding stability and function of cy-tochrome c. PhD thesis, University of British Columbia, Vancouver, Canada.BIBLIOGRAPHY^ 147Murphy, M.E.P., Nall, B.T. & Brayer, G.D. (1992). Structure determination and analysis ofyeast iso-2-cytochrome c and a composite mutant protein. J. Mol. Biol., 227, 160-176.Nocek, J.M., Stemp, E.D.A., Finnegan, M.G., Koshy, T.I., Johnson, M.K., Margoliash, E.,Mauk, A.G., Smith, M. & Hoffman, B.M. (1991). Low-temperature, cooperative conforma-tional transition within [Zn-cytochrome c peroxidase, cytochrome c] complexes: variationwith cytochrome c. J. Amer. Chem. Soc., 113, 6822-6831.North, A.C.T., Phillips, D.C. & Mathews, F.S. (1968). A semi-empirical method of absorptioncorrection. Acta Crystallogr. sect. A, 24, 351-359.Nozaki, M., Mitzushima, H., Horio, T. & Okunuki, K. (1958). Further studies on proteinasedigestion of bakers yeast cytochrome c. J. Biochem., 256, 673-676.Ochi, H., Hata, Y., Tanaka, N., Kakudo, M., Sakkurai, T., Aihara, S. & Morita, Y. (1983).Structure of rice ferricytochrome c at 2.0A resolution. J. Mol. Biol., 166, 407-418.Ohlendorf, D.H., Treharne, A., Weber, P.C., Wendoloski, J.J.^Salemme, F.R. (1991). Ac-curacy of structures: a comparison of independently refined models of human IL-@. InAbstracts of the Annual meeting of the American Crystallographic Association, vol. 19 of2, p. 36, Toledo, Ohio. American Crystallographic Association.Pace, C.N. (1990). Measuring and increasing protein stability. Trends Biochem. Sci., 8, 93-98.Pearce, L.L., Gartner, A.L., Smith, M. & Mauk, A.G. (1989). Mutation-induced perturbationof the cytochrome c alkaline transition. Biochemistry, 28, 3152-3156.Pelletier, H. & Kraut, J. (1992). Crystal structure of a complex between electron transferpartners, cytochrome c peroxidase and cytochrome c. Science, 258, 1748-1755.Perry, K.M., Fauman, E.B., Finner-Moore, J.S., Montfort, W.R., Maley, G.F., Meley, F. &Stroud, R.M. (1990). Plastic adaptation toward mutations in proteins: structural compar-ison of thymidylate synthases. Proteins: Struct. Funt. Genet., 8, 315-333.Pettigrew, G.W. & Moore, G.R. (1987). Cytochrome c. Biological Aspects. Springer-Verlag,Berlin.Pielak, G.J., Mauk, A.G. & Smith, M. (1985). Site-directed mutagenesis of cytochrome c showsthat an invariant Phe is not essential for function. Nature (London), 313, 152-153.Pielak, G.J., Oikawa, K., Mauk, A.G., Smith, M. & Kay, C.M. (1986). Elimination of thenegative Soret Cotton effect of cytochrome c by replacement of the invariant phenylalanineusing site-directed mutagenesis. J. Amer. Chem. Soc., 188, 2724-2727.Poulos, T.L. & Kraut, J. (1980). A hypothetical model of the cytochrome c peroxidase-cytochrome c electron transfer complex. J. Biol. Chem., 255, 10322-10330.Press, W.H., Flannery, B.P., Teukolsky, S.A. & Vetterling, W.T. (1988). Numerical recipes.Cambridge University Press, Cambridge.BIBLIOGRAPHY^ 148Proudfoot, A.E.I. & Wallace, C.J.A. (1987). Semisynthesis of cytochrome c: the effect ofmodifying the conserved residues 38 and 39. Biochem. J., 248, 965-967.Rackovsky, S. & Goldstein, D.A. (1984). On the redox conformational change in cytochrome c.Proc. Nat. Acad. Sci., U.S.A., 81, 5901-5905.Rafferty, S.P. (1992). Functional properties of active site variants of yeast cytochrome c. PhDthesis, University of British Columbia, Vancouver, Canada.Rafferty, S.P., Pearce, L.L., Barker, P.D., Guillemette, J.G., Kay, C.M., Smith, M. & Mauk,A.G. (1990). Electrochemical, kinetic, and circular dichroic consequences of mutations atposition 82 of yeast iso-1-cytochrome c. Biochemistry, 29, 9365-9369.Rashin, A.A., Iofin, M. & Honig, B. (1986). Internal cavities and buried waters in globularproteins. Biochemistry, 25, 3619-3625.Read, R.J. (1986). Improved Fourier coefficients for maps using phases from partial structureswith errors. Acta Crystallogr. sect. A, 42, 140-149.Read, R.J. (1990). Structure-factor probabilities for related structures. Acta Crystallogr. sect.A, 46, 900-912.Read, R.J., Fujinaga, M., Sielecki, A.R. & James, M.N.G. (1983). Structure of the complex ofStreptomyces griseus protease B and the third domain of the turkey ovomucoid inhibitorat 1.8-A resolution. Biochemistry, 22, 4420-4433.Rogers, N.K. & Moore, G.R. (1988). On the energetics of conformational changes and pHdependent redox behaviour of electron transfer proteins. FEBS Letters, 228, 69-73.Rossmann, M.G., Leslie, A.G.W., Abdel-Meguid, S.S. & Tsukihara, T. (1979). Processing andpost-refinement of oscillation camera data. J. Appl. Cryst., 12, 570-581.Saigo, S. (1986). Isomerization and denaturation of homologous cytochromes c: correlationbetween local and gross conformational changes. J. Biochem, 100, 157-165.Salemme, F.R. (1976). A hypothetical structure for an intermolecular electron transfer complexof cytochrome c and b5 . J. Mol. Biol., 102, 563-568.Salemme, F.R. (1977). Structure and function of cytochrome c. Annu. Rev. Biochem., 46,299-329.Satterlee, J.D., Moench, S.J. & Erman, J.E. (1987). A proton NMR study of the non-covalentcomplex of horse cytochrome c and yeast cytochrome c peroxidase and its comparison withother interacting protein complexes. Biochim. Biophys. Acta, 912, 87-97.Schejter, A., Aviram, I. & Goldkorn, T. (1982). The contribution of electrostatic factors to theoxidation-reduction potentials of c-type cytochromes. In Electron Transport and OxygenUtilization (Ho, C., ed.), pp. 95-99. Elsevier, Amsterdam.BIBLIOGRAPHY^ 149Schlauder, G.G. Si Kassner, R.J. (1979). Comparative solvent perturbation of horse heartcytochrome c and Rhodospirillum rubrum cytochrome c2. J. Biol. Chem., 254, 4110-4113.Schultz, G.E. & Shirmer, R.H. (1979). Principles of Protein Structure (Cantor, C.R., ed.).Springer-Verlag, New York.Sherwood, C. Si Brayer, G.D. (1985). Crystallization and preliminary diffraction data for iso-1-cytochrome c from yeast. J. Mol. Biol., 185, 209-210.Shrake, A. & Rupley, J.A. (1973). Environment and exposure to solvent of protein atoms.Lysozyme and insulin. J. Mol. Biol., 79, 351-371.Singh, J., Thornton, J.M. Snarey, M. & Campbell, S.F. (1987). The geometries of interactingarginine-carboxyls in proteins. FEBS Letters, 224, 161-171.Smith, M., Leung, D.W., Gillam, S., Astell, C.R., Montgomery, D.L. Si Hall, B.D. (1979).Sequence of the gene for iso-1-cytochrome c in Saccharomyces cerevisiae. Cell, 16, 753-761Stellwagen, E. (1978). Haem exposure as the determinate of oxidation-reduction potential ofhaem proteins. Nature (London), 330, 86-88.Takano, T. & Dickerson, R.E. (1980). Redox conformational changes in refined tuna cytochromec. Proc. Nat. Acad. Sci., U.S.A., 77, 6371-6375.Takano, T. & Dickerson, R.E. (1981a). Conformation change of cytochrome c: I. Ferrocy-tochrome c structure refined at 1.5A resolution. J. Mol. Biol., 153, 79-94.Takano, T. & Dickerson, R.E. (1981b). Conformation change of cytochrome c: II. Ferricy-tochrome c refinement at 1.8A and comparison with the ferricytochrome structure. J. Mol.Biol., 153, 95-115.Tanaka, N., Yamane, T., Tsukihara, T., Ashida, T. & Kakudo, M. (1975). The crystal structureof bonito ferrocytochrome c at 2.3 A resolution. J. Biochem, 77, 147-162.ten Kortenaar, P.B.W., Adams, P.S.H.M. Si Tesser, G.I. (1985). Semisynthesis of horse heartcytochrome c analogues from two or three fragments. Proc. Nat. Acad. Sci., U.S.A., 82,8279-8283.Trewhella, J., Carlson, V.A.P., Curtis, E.H. & Heidorn, D.B. (1988). Differences in the solutionstructures of oxidized and reduced cytochrome c measured by small-angle X-ray scattering.Biochemistry, 27, 1121-1125.Ulmer, D.D. & Kagi, J.H.R. (1968). Hydrogen-deuterium exchange of cytochrome c: I. Effectof oxidation state. Biochemistry, 7, 2710-2717.Wallace, C.J.A., Mascagni, P., Chait, B.T., Collawn, J.F., Paterson, Y., Proudfoot, A.E.I. SiKent, S.B.H. (1989). Substitutions engineered by chemical synthesis at three conservedsites in mitochondrial cytochrome c. J. Biol. Chem., 264, 15199-15209.BIBLIOGRAPHY^ 150Wand, A.J., Roder, H. Si Englander, S.W. (1986). Two-dimensional 1 11 NMR studies of cy-tochrome c: hydrogen exchange in the N-terminal helix. Biochemistry, 25, 1107-1114.Warshel, A. Churg, A.K. (1983). Converting structural changes upon oxidation of cytochromec to electrostatic reorganization energy. J. Mol. Biol., 168, 693-697.Watt, G.D. & Sturtevan, J.M. (1969). The enthalpy change accompanying the oxidation offerrocytochrome c in the pH range 6-11 at 25°. Biochemistry, 8, 4567-4571.Weber, C., Michel, B. & Bosshard, H.R. (1987). Spectroscopic analysis of the cytochrome coxidase-cytochrome c complex: circular dichroism and magnetic circular dichroism mea-surements reveal change of cytochrome c heme geometry imposed by complex formation.Proc. Nat. Acad. Sci., U.S.A., 84, 6687-6691.Wendoloski, J.J., Matthew, J.B., Weber, P.C. & Salemme, F.R. (1987). Molecular dynamics ofa cytochrome c-cytochrome b 5 electron transfer complex. Science, 238, 794-796.Whitford, D., Gao, Y., Pielak, G.J., Williams, R.J.P., McLendon, G.L. Si Sherman, F. (1991).The role of the internal hydrogen bond network in first-order protein electron transferbetween Saccharomyces cerevisiae iso-l-cytochrome c and bovine microsomal cytochromeb5. Eur. J. Biochem., pp. 359-367.Williams, G., Clayden, N.J. Moore, G.R. Si Williams, R.J.P. (1985a). Comparison of thesolution and crystal structures of mitochondrial cytochrome c. Analysis of paramagneticshifts in the nuclear magnetic resonance spectrum of Ferricytochrome c. J. Mol. Biol.,183, 447-460.Williams, G., Moore, G.R. Si Williams, R.J.P. (1985b). Biological electron transfer: the struc-ture, dynamics and reactivity of cytochrome c. Comments Inorg. Chem., 4, 55-98.Wilson, A.J.C. (1942). Determination of absolute from relative X-ray intensity data. Nature(London), 150, 151-152.Wood, L.C., Muthukrishnan, K., White, T.B., Ramadas, L. & Nall, B.T. (1988). Constructionand Characterization of Mutant Iso-2-cytochrome c with replacement of conserved prolines.Biochemistry, 27, 8554-8561.Zhang, Q.P., Marohn, J. & McLendon, G. (1990). Macromolecular recognition in the cy-tochrome c - cytochrome c peroxidase complex involves fast two-dimensional diffusion. J.Phys. Chem., 94, 8628-8630.Zheng, C., Wong, C.F. Si McCammon, J.A. (1990). Fluctuation of the solvent-accessible surfacearea of tuna ferrocytochrome c. Biopolymers, 29, 1877-1883.Appendix ADetermination of the Absolute Scale FactorA.1 IntroductionThe structure factors initially obtained after data processing are on a relative scale. In order toperform a meaningful atomic coordinate refinement these structure factors must be multipliedby a scale factor so that they can be compared to calculated structure factors. This scalefactor is appropriately called the absolute scale factor. A common approach to determiningthe absolute scale factor is to first obtain an initial estimate by statistical means which islater improved upon during subsequent structure refinement. In total, three methods havecommonly been used for obtaining initial estimates for the absolute scale factor: linear rescale,least-squares rescale and the Wilson plot method.A.2 Initial Estimate for the Absolute Scale FactorA.2.1 Linear rescaleLinear rescale requires a structure factor data set from a related isomorphous crystalline formof the protein which is already on an absolute scale. The scale factor K between this structurefactor data set and a set of data just collected can be computed as follows:K = Ehk1 FR/1kt Ehk1 Fonki(A.4)where FR and F, are the structure factors from the highly related previously scaled data setand the data set to be scaled, respectively. The sum is taken over all the reflections which arecommon between the two data sets. This linear rescale method assumes that the total amountof scattering material in the unit cell is identical in the two related crystals. One drawback151Appendix A. Determination of the Absolute Scale Factor^ 152to this method is that the scale factor obtained is largely determined by the strong reflectionspresent.A.2.2 Least -squares rescaleAn improvement on the linear rescale approach is to introduce a least-squares analysis wherethe scale factor is chosen so as to minimize the difference between the two data sets. Thistranslates mathematically into minimizing the following function:= E(K Fohkt FRhkt) 2hklSolving a 0 gives the following expression for the scale factor:K  >hkl FRial Fohkt Ehkl(Fonkl) 2A drawback of the least-squares rescale method is that it assumes the overall thermal factor,which is related to the fall off in structure factor magnitudes with increasing resolution, is similarfor the two data sets. This is not necessarily the case even for highly related data sets. On theother hand, the linear rescale is less sensitive to differences in overall thermal factors since it isbiased towards strong reflections which occur mostly at low resolution.A.2.3 Wilson plot methodAn ab initio method for estimating absolute scale factor requiring only a knowledge of thechemical composition of the unit cell contents was developed by Wilson (1942). Assumingthe atomic contents of the unit cell are randomly distributed, the theoretical mean value ofdiffraction intensities as a function of resolution is given by:< IT > = E(foie —B1(42)2 ) 2^(A.7)j=iIn this equation N is the total number of atoms in the unit cell, and foi and Bi are the atomicscattering factor and isotropic thermal factor for atom i. If the individual thermal factors(A.5)(A.6)Appendix A. Determination of the Absolute Scale Factor^ 153are replaced by an overall thermal factor, the following expression can be formulated for theabsolute scale factor:NK2 < 7 > =< IT > e-2B(--5-1 ,2)2^fo2ii=1Note that the scale factor is squared in this equation since intensities instead of structure factorsare compared.Traditionally this equation has been evaluated by rewriting it into the following form:< 7 >2 = ln 2K 2B( sin 0 ) 2 (A.9)fo i2A plot of ln(< 7 >/Eliv i Poi) versus (sin 0/A) 2 should give a straight line with a slope of —2Band an intercept of — ln 2K. Figure A.53 gives an example of such a plot. As can be seen,at medium and high resolution (> 4.5 A) the data does follow a straight line as predicted byWilson, but at low resolution (< 5 A) the data points present are scattered. The reason for theanomalous behaviour observed at low resolution is the non-randomness of the unit cell contentsat this resolution. In determining the scale factor, resolution cutoffs are therefore first applied,after which a least squares fit to the theoretical line is calculated.An alternative approach to solving equation A.8 is to apply a non-linear least-squares pro-cedure (Press et al., 1988). Graphically this can be represented by plottingversus (sin 9/)) 2 and determining the best exponential curve of the form y = e-2Bx that fitsthe data points using a non-linear least-squares algorithm (see Figure A.54). As was the casefor the Wilson plot, because of the non-randomness of the unit cell contents at low resolution,resolution cutoffs are first applied after which the least-squares fit to the exponential curveis calculated. Effectively, the major difference between this method and the original Wilsonplot is the weighting of the data points in the least-squares procedure. As can be seen fromFigures A.53 and A.54, in the Wilson plot method the theoretical curves agree better at highresolution while the non-linear approach forces a better agreement for lower resolution datapoints. The results obtained with these two fitting methods therefore give an indication of thespread in possible values for the absolute scale factor. Experience from the studies herein has(A.8)< 7 >/Efv=i fo2- 2.4-- 2.8-- 3.2-- 3.6-- 4.0-Appendix A. Determination of the Absolute Scale Factor^ 154- 2.0-4.4 I^1^ll^I^I^I 10.00^0.01^0.02^0.03^0204^0.05(sin(0) / A)0.06Figure A.53: Wilson plot for determining the absolute scale factor for the reduced N52I-Y67Fmutant cytochrome c data set. For this analysis reciprocal space was divided into shells ac-cording to ( siAn 0 \ 2) with each containing at least 200 reflections. To determine the absolute scalefactor a least-squares fit of the theoretical line (solid) to the data points was calculated usingonly data points between 4.5-2.0 A as indicated by the vertical dotted lines. This analysissuggested that the absolute scale factor for this data set was 2.8. For comparison, a theoreticalline (dashed) is also shown representing the absolute scale factor determined using a non-linearmethod (see Figure A.54). This approach suggested a value of 2.3 for the absolute scale factor.shown that the final absolute scale factor resulting from structure refinement falls between thetwo values obtained using the linear and non-linear versions of the Wilson plot. Therefore asan initial estimate for the scale factor the average of these values was used.A.3 Refinement of the Absolute Scale FactorBeyond the usual scale refinement, a method found useful for estimating the shift in scale factor,and which has been used for the structure determinations presented herein, is the following. Inthis method the agreement between Fo and 11 is described as a function of both the absoluteAppendix A. Determination of the Absolute Scale Factor^ 1550.110.09.134-Z^0.070.05V0.030.010.00^0.01^0.02^0.03^Op4(sin(0) / A)Figure A.54: Absolute scale factor plot for the reduced N52I-Y67F mutant cytochrome c dataset. For this analysis reciprocal space was divided into shells according to (9 6e ) 2 with eachcontaining at least 200 reflections. To determine the absolute scale factor a least-squares fitof the theoretical exponential curve (solid line) to the data points was calculated using onlydata points between 4.5-2.0 A as indicated by the vertical dotted lines. This analysis suggesteda value of 2.3 for the absolute scale factor. Also shown is the theoretical curve (dashed line)derived from the absolute scale factor (2.8) determined using the Wilson plot method (seeFigure A.53). The difference in the values obtained with these two methods reflect the accuracyof the absolute scale factor determinations. In practice the average of the two values was usedas an initial estimate for the absolute scale factor.scale factor and the discrepancy in the overall thermal factor between data and model.KFohki = Fchki e °13 ( W )2^(A.10)Here AB represents the discrepancy between the actual overall thermal factor and that of theatomic model. To evaluate this equation one can rewrite it into the form:In^nFohk1 = In K + AB( si B)2AFchki (A.11)A plot of ln ;hi,/ versus (w)2 should give a straight line with a slope of AB and intercept ofhkl0.05^0.06— ln K (see Figure A.55). This is analogous to the Wilson plot method described above with0.280.240.200.16• 0.120.080.040.00—0.04—0.08—0.120.00^0.01^0.02^0.03^0.04(sin(0) / A) 20.05 0.06Appendix A. Determination of the Absolute Scale Factor^ 156Figure A.55: A comparison of the amplitudes of the observed and calculated structure factorsof the oxidized N52I structure after only six cycles of least-squares refinement. For this analysisreciprocal space was divided into shells according to ( 7 0 N2) with each shell containing at least100 structure factors. As can be seen, at high resolution (less than 4 A; (W) 2 > 0.015) theamplitudes of Fe 's are systematically underestimated due to a high value for the overall thermalfactor of the model. By fitting a straight line to the data points an estimate can be obtainedfor the discrepancy between the actual overall thermal factor and that of the model as well asfor the absolute scale factor of the data set. This analysis suggests that the overall thermalfactor for the oxidized N52I model after six cycles of refinement is ,3.5 A2 too high and thatthe absolute scale factor can be improved by multiplication of 1.07.the main difference being the substitution of Er_ i L i for Fe . Alternatively, one can minimizethe function:AB ( sin 6' )2 , 2=^ (KFohki F e^A )hkl(A.12)using non-linear least-squares procedures similar to what has been described above for themodified Wilson plot method. It is only necessary to apply the scale factor correction to theobserved structure factors. Subsequent structure refinement will improve the individual atomicthermal factors so that the discrepancy between the actual overall thermal factor and that ofthe model will disappear.Appendix BTheory of Crystallographic RefinementB.1 IntroductionThe foundation of protein structure refinement is that both the structural model and the ob-served intensities can be expressed in terms of structure factors. Structure factors for the modelcan be calculated using the following equation:N2iri(hxi+kyiChki^( sin )2 eF = K Efoi e P k Ai=1(B.13)Here Fchki is the calculated structure factor (Miller indices h,k,l), K is an overall scale factorand N is the total number of atoms in the unit cell. The atomic scattering factor, isotropicthermal factor and the position of atom i are represented by foi, Bi and (xi, yi, zi), respectively.The diffraction angle, which is dependent on the Miller indices, is represented by 0, and A isthe wavelength of the x-ray radiation.To obtain structure factors from the experimental diffraction data the square-root of themeasured intensities is taken.IFohkii = V Ihkl^ (B.14)Note that while Fc is a complex number only the amplitude of Fo can be measured experimen-tally. This reflects the fundamental phase problem in crystallography.A common way of expressing the agreement between the refinement model and the observedstructure factors is the crystallographic R-factor:R = Ehkl Il Fohkil^IFchkillEhk1lFohktiThe R-factor is often expressed as a percentage by multiplication by 100. In the case of aperfectly matching refinement model the R-factor would be 0% while with a randomly incorrect(B.15)157Appendix B. Theory of Crystallographic Refinement^ 158model this value would be ,59% for acentric data (Blundell & Johnson, 1976). Typical R-factorsfor small molecule structures are less than 5%, whereas protein structure determinations yieldR-factors in the range of 10 to 25%. The higher R-factors of protein structures can be attributedto the intrinsically greater flexibility of these molecules and the higher solvent content of thecrystals (from 30-95% of the unit cell volume). Both these properties lead to large thermalvibrations in these molecules, dynamic disorder among possible local conformations and staticvariations in the structures within different unit cells (Hendrickson & Konnert, 1981).B.2 Reciprocal-space RefinementThe three-dimensional structures of the proteins described in this thesis were refined using thereciprocal-space refinement method. In this method the object is to minimize the differencesbetween F, and using a least squares approach. Therefore, refinement is aimed at findingthe absolute minimum of the following function:^co = E whici(IFohk, — KiFchk, 1) 2^(B.16)hklIn this equation K is the overall scale factor from equation B.13 and CJJhki is a weighting factor.This weighting factor can reflect the accuracy of measured structure factor, or can be used togive structure factors different weights depending on their resolution.The standard way of determining the minimum of the function co is to take the partialderivatives with respect to all the parameters and set these to zero:Oxi^0 •^for all parameters xi^ (B.17)Thus for a function co with N parameters the problem is reduced to solving N equations withN unknowns.In the case where co is a linear function a solution can be obtained using a matrix approachsince the partial derivatives can also be written as linear equations with the general form of:(9(p„ = ailxl ai2x2 -I- • • • + aiixi " • + aiNxN — bi = 0oxi (B.18)Appendix B. Theory of Crystallographic Refinement^ 159In matrix form the N equations to solve become:/ an a12 • • aiN / x i \a21 a22 • • a2i • a2N x2 b2(B.19)ail ail • • • aii • aiN xi biaNi aN2 • • • aNi • • • aNN \ xN / bN /Or in abbreviated notation:AxE=To solve this, one has to invert matrix A.= A -1 x b^ (B.20)Three problems are encountered when one uses this approach for the refinement of macro-molecular structures. These are:1. the non-linearity of the equations2. the ratio of the number of observations to the number of parameters to be refined3. the size of the matrixB.2.1 The non-linearity of the equationsThe least-squares method described above is straight-forward but uses the fundamental as-sumption that the function co is linear. The function represented in equation B.16 is not linear.To solve this problem the function co or its derivatives are approximated by a linear functionusing a Taylor series:f(m)(a) f(z) = ^ (zm=O m.Here f(m)(a) is the in order derivative evaluated at a.(B.21)Appendix B. Theory of Crystallographic Refinement^ 160Applied to the structure refinement problem the procedure used is as follows. First simplifyequation B.16 by explicitly stating that (,o and 11 are a function of the variables xl, x 2 , • • (invector notation g) and treating the overall scale factor K as one of the variables of Fc .(p(i) = E coma [Fahk, — Fchkt(A 2^ (B.22)hklNote that the absolute value bars are deleted in this equation and those that follow for read-ability. Analogous to above, this function is minimal when all derivatives are zero:(9(,o^ )1 0Fchki (i)^0 ;= —2 Ekl whk, [F0 hkl FChkl°xi °xihfor all parameters x,^(B.23)At this point the problem can be linearized by expressing Fc(i) as a Taylor series leavingout all terms after the first derivative.Fc(i) = Fc(x1))+^01:9,(x-6) ( 3 x.?)3(B.24)Substituting this into equation B.23 and rearranging:;0,azNE E i (xi^j) =^WhklF0hk1 FChk1 (0-Fchki (x49 .9-Fchki(i)) ^[ ^kx0)1 4 9 FChk1k — / °xi^(B.25)j hkl^ hklSince there are N of these equations they can be written in the following matrix form.A x = 1.*^ (B.26)In this equationE aFchk , (x1 ) aFchki(xb )hkl^OXi^°xiaFchk(?)bi^E Whkl[Fohkt FchkikX \iJJ as ihklbi = xi — x iThis equation can be solved by inverting matrix A as discussed previously.In taking this approach to structure refinement, two factors must be remembered. First,solving for the above linear equations will only provide shifts (6i) for the parameters. HowaiiAppendix B. Theory of Crystallographic Refinement^ 161optimal these shifts are will depend on the validity of the Taylor series approximation. Inpractice, the updated parameters will be used in a new Taylor expansion and the process willbe iterated until the shifts are smaller than their estimated errors. Reducing the number ofiterations necessary to reach convergence is possible if more terms of the Taylor series areincorporated. However, most crystallographic refinement programs only use first order or atthe most second order derivatives because the extra computing time required for calculatingadditional terms in the Taylor expansion is far more than will be saved in reducing the numberof iterations.Secondly, the above method of minimization does not guarantee that the overall globalminimum of function (,o will be found. It is more likely that the minimization will converge to alocal minimum. To overcome this problem one can evaluate the agreement between F, and 11in real-space by inspection of difference electron density maps which can suggest improvementsto the structure and thus to the calculated structure factors. After adjustments to the structurehave been made reciprocal-space minimization can be continued.B.2.2 The ratio of the number of observations to the number of parameters to berefinedIn the least-squares refinement method the ratio of the number of data points to the numberof refinement parameters, also known as the overdeterminancy, is an important indicator ofthe reliability and feasibility of the refinement. The minimum overdeterminancy required fora least-squares refinement is 1, assuming that the data can be perfectly represented by theparameters. This can never be the case for experimental situations since the diffraction datacannot be perfectly measured and the refinement model cannot be perfectly represented. Tocompensate for this situation more data points than parameters are necessary. The higher theoverdeterminancy the more robust the refinement will be and the more precise the resultingmodel will become. In the case of small molecule structures the overdeterminancy can be ashigh as 10 resulting in extremely robust refinements and accurate structures.A1.7 ALa A1.0 A2.0 A2.1 A2.2 Ailk5.07:4.54'4.043.543.04'2.542.041.541.040.54.1.71he At.2 A2.0 A2.1 A2.2 AilkAppendix B. Theory of Crystallographic Refinement^ 1625.0 —4.5 —>..c.) 4.0—Cc3 3.5-]C•^3.0-EN 2.54(i)a) 2.0-730)" 1.54>0 1.0-:0.5^ I ^ I^I ^ 1^1^i^I40 50 60 70 80 90 100Completeness (%)i^1^II ^ I^ 1^ 140 50 60 70 80 90 100Completeness (%)Figure B.56: The overdeterminancy for the refinement of a typical cytochrome c structure withrespect to the completeness of the diffraction data used in the refinement and the resolutionof the data collected. In graph (a) the overdeterminancy is given for unrestrained refinement.In this case only structure factors are considered as observations, and as such the ratio of thenumber of observations to the number of parameters is generally not greater than 2. Whenstereochemical information is added as restraints, the observation to parameter ratio can bedramatically improved. In (b) is shown the overdeterminancy for a refinement using PROLSQ(Hendrickson & Konnert, 1981) with standard restraints ( ,6750 restraints; see Table B.28).This approach would theoretically improve the overdeterminancy by 1.75. However, since thestereochemical restraints are highly correlated and can not be counted as independent obser-vations, the actual improvement of the number of observations to the number of parametersratio is less than is shown in (b). How much less can, unfortunately, not be determined sincethis would require inversion of matrix A from equation B.26 (as modified for the restrainedparameter refinement approach) which is not feasible as discussed in the text.Unfortunately, for macromolecular structure determinations the number of observations(diffraction intensities) that can be measured is limited due to resolution dependent falloff inthe diffraction data. Figure B.56a shows that for a typical cytochrome c structure refinementone must have at least a complete 2.3 A. data set or a 50% complete 1.9 A data set to obtainan overdeterminancy greater than 1. For an overdeterminancy of 2, which is still insufficientfor a stable refinement, one needs to have at least a complete 1.9 A data set.Clearly, to make refinement of macromolecular structures feasible, the overdeterminancy ra-tio must be improved. A method to accomplish this is by increasing the number of observationsusing an approach known as restrained refinement. Here stereochemical knowledge is includedAppendix B. Theory of Crystallographic Refinement^ 163as data points by treating ideal bond distances, torsional angles, etc. as observations (Hendrick-son & Konnert, 1981). Mathematically this is done by minimizing an extended equation B.16as represented below:= Pxray (Pstereocherni Pstereochem2 + • • • (B.27)Here Coxray corresponds to the right hand side of equation B.16 and (Pstereochem, describes differentstereochemical restraints. An example of a stereochemical restrain is the function that controlsthe bond lengths:(B.28)In this equation di and dideal are the actual and ideal bond lengths for bond i and cob ond, is aweighting factor. In this method the structural model is allowed to deviate from ideal geometryby adjustments of the various weighting factors and is therefore called restrained refinement.The increase in observations can be quite significant using this approach. For a typical cyto-chrome c refinement, using the least-squares restrained parameter refinement program PROLSQ(Hendrickson & Konnert, 1981), with approximately 3800 variables (950 atoms), a total of about6600-6900 restraints can be added (depending on the exact conformation of the protein andthe number and position of solvent molecules; see Figure B.56 and Table B.28). However sincestructural restraints are correlated, this does not imply that the overdeterminancy is improvedby 38700 1.75.B.2.3 The size of the matrixThe normal matrix A in equation B.26 as modified for the restrained parameter refinementapproach can be large for macromolecular refinements. In the case of a typical cytochrome crefinement with isotropic temperature factors and fixed occupancies, the number of variablesis about 3800. This means that matrix A will contain 3800 2 or more than 14 million ele-ments making the inversion of matrix A computationally impractical. Therefore, an alternativemethod for the solution of equation B.26 is employed. In the implementation of PROLSQ, theNo. of bonds(Pbond lengths =^E^wbond, (di — dId eal ) 2Appendix B. Theory of Crystallographic Refinement^ 164Table B.28: Typical numbers of stereochemical restraints of different classes used in the refine-ment of yeast iso-l-cytochrome c structuresStereochemicalrefinement parametersBond distances1-2 bond distance1-3 bond distance1-4 bond distancespecial distancestPlanar restraintsChiral volumeNon-bonded contactssingle-torsionmulti-torsionpossible hydrogen bondsTorsion anglesplanar (0°or 180°)staggered (±60°,180°)orthonormal (±90°)Isotropic thermal factorsmain-chain bondmain-chain angleside-chain bondside-chain angleTypical number ofrestraintsTypical refinement restraintweighting values925 0.020 A1250 0.030 A340 0.050 A14 0.060 A780 0.020 A120 0.150 A3400 0.250 A90 0.250 A70 0.250 A110 2.5°170 20.0°15 15.0°500 1.5 A2650 2.5 A 2430 2.0 A 2620 3.0 A 2t The special distances define the bonds between the heme iron and the His18 and Met80ligands. See Hendrickson (1985) for a thorough description of the restraint definitions.method of conjugate gradients is employed since it efficiently utilizes the properties of matrixA (Hendrickson & Konnert, 1980). For example, matrix A is symmetrical around the diagonal,leading to a reduction of --, -12- in the number of elements required for the evaluation of matrixA. Additionally, not all elements of the matrix are equally important, with many having valuesclose to zero. It appears that the stability, speed and radius of convergence of refinement arenot seriously affected if only those elements of the matrix that are non-zero for stereochemicalAppendix B. Theory of Crystallographic Refinement^ 165Figure B.57: A diagram of the normal matrix A showing the elements evaluated by the re-strained parameter refinement program PROLSQ (Hendrickson & Konnert, 1981). The darkelements are in "self blocks" containing the six elements related to the atomic coordinates plusan additional element for the thermal factor and one element for refinement of the scale fac-tor (K). The hatched blocks are elements that relate parameters that are correlated throughstereochemical restraints. Blank elements are assumed to have a value of zero. In the aboveexample the atom pairs 1 & 2, 1 & 3, 2 & (n-1), and 3 & n are correlated through stereochemicalrestraints.restraints are retained (see Figure B.57). The result is that generally less than 1% of the el-ements of the matrix are evaluated. In the case of cytochrome c, approximately 0.2% of thematrix elements are retained in a typical refinement cycle. The conjugate gradient method asimplemented in PROLSQ also incorporates knowledge from previous iterations which increasesits convergence rate. Theoretically convergence will be reached in N steps (number of parame-ters), but experience has shown that a minimum can be reached in much fewer (typically 10-30)steps.Appendix CEstimating Coordinate Errors in Macromolecular StructuresC.1 IntroductionAny quantitative determination is incomplete without a reliable measure of the associateduncertainty. Without this knowledge interpretation of the obtained results can easily lead toerroneous conclusions. When applied to the study of mutant structures in a protein-engineeringenvironment, without an estimate of the accuracy of the three-dimensional structures involved,differences between structures can be either over or under estimated. Such a situation seri-ously hampers the interpretation of the relationships between protein structure and expressedbiological activity.Determining the accuracy of a three dimensional atomic structure solved by x-ray diffrac-tions methods is in theory straight-forward. One first determines the errors in the observedstructure factors and evaluates how these errors are propagated through subsequent calcula-tions. This will result in estimates for the accuracy of the parameters that define the structureand can be calculated for the case of small molecules as follows:Cri bii Ehk1 Whkl(IFohkiI^IFChk11) 2M - N (C.29)Here ai is the standard deviation of parameter i, bi, is the ith diagonal element of the inversematrix A -1 , A being the matrix from equation B.26, whki is the weighting factor used in theleast-squares refinement and M and N are the number of structure factors (observations) andparameters, respectively.From the discussion on refinement presented in Appendix B it can be seen that this proce-dure is generally not practical in the case of macromolecular crystallography due to the size of166Appendix C. Estimating Coordinate Errors in Macromolecular Structures^167matrix A. As a result, several alternatives for estimating the accuracy of macromolecular struc-tures have been developed. These include those based on empirical rules and those that have astatistical origin (Luzzati, 1952; Cruickshank, 1949,1954). However, the program SHELXL-92(by G. Sheldrick), which is able to calculate e.s.d.'s for macromolecules using equation C.29,has recently become available.C.2 Empirical error estimatesRecently a number of studies have been carried out in which the structures of highly relatedor identical proteins have been compared (Perry et al., 1990; Bott, 1991; Fauman & Stroud,1991; Ohlendorf et al., 1991). From an analysis of the deviations between these structures,which should represent random errors, empirical rules can be derived. For example, Perry et al.(1990) have proposed the following empirical equation:3art = —4 SR (0.0015B? — 0.0203Bi + 0.359) (C.30)In this formula the radial coordinate error of atom i is dependent on its thermal factor 13„ theresolution S, of the structure and the crystallographic R-factor represented as R. This methodonly provides information regarding the coordinate errors and does not give estimates for thestandard deviations of thermal factor parameters. This is a general feature of all the methodsfor estimating errors in structures.Clearly there are some drawbacks in the use of empirical rules. First the validity of thedatabase of structures must be considered. If related but not identical structures are used, thedifferences between these structures not only reflect random errors, but also real differences.Further, real differences between identical or nearly identical structures can exist due to differentcrystallization conditions or crystal contacts. These situations will result in an overestimation ofthe real coordinate errors. A second criticism relates to the possibility that not all the variablesare present in the equation representing the empirical rule. This will again result in false errorestimates in those cases were variables are missing or perhaps a particular set of structures isAppendix C. Estimating Coordinate Errors in Macromolecular Structures^168dependent on a different mix of variables. For example, in the empirical equation C.30 theatom type is not a variable, while it is to be expected that the position of an iron atom with acertain thermal factor will be more accurate than a carbon atom with the same thermal factorbecause of the greater scattering power of the iron atom. Thus empirical rules for estimatingerrors must be used with care particularly in their application to new structures.C.3 Luzzati error estimatesThe method developed by Luzzati (1952) is the one most often used for estimating the co-ordinate errors in macromolecular structures. In its derivation random errors in the atomiccoordinates are expressed in terms of probability distributions. These can be combined into anoverall probability distribution which will have a normal Gaussian form. Thus even if errors indifferent parts of the protein structure do not follow a normal distribution, the overall distribu-tion will still have a normal Gaussian form due to the central limit theorem. By applying themathematics of the Fourier transform to this normal distribution, a probability distribution forthe structure factors can be obtained. Evaluation of this probability distribution gives infor-mation on the overall coordinate error in the structure, although the accuracy for individualparts of the structure can not be determined.Luzzati expressed the probability distribution for the structure factors in the form of R-factors dependent on resolution and assuming that the errors in atomic positions are the onlysource for the differences between Fa and Fc . Therefore, a Luzzati plot (see Figure C.58) showsthe theoretical dependence of R-factor versus resolution (expressed as W) for different overallerror estimates of the structure. By inserting the R-factors of different resolution shells for aparticular experimentally determined structure an estimate of its overall accuracy can be made.Read (1986) extended this work by allowing for the fact that the difference between F, and11 is caused not only by coordinate errors, but also by an incomplete model. However, in asubsequent paper Read (1990) described a number of flaws in some of the assumptions made byLuzzati for deriving the structure factor probability distributions and how these distributionsAppendix C. Estimating Coordinate Errors in Macromolecular Structures^1690.550.500.450.400.350 0.30U0I^0.25CY0.200.150.100.050.000.00^0.05^0.10^0.15^0.20^0.25^0.30SIN(6) / xFigure C.58: Plot of the theoretical dependence of the crystallographic R-factor factor as afunction of resolution assuming various levels of r.m.s. coordinate errors (Luzzati, 1952). Alsoshown (dashed line and .A, symbols) is the dependence of the R-factor versus resolution for theexperimentally determined wild-type yeast iso-l-cytochrome c structure in the oxidized state.The analysis suggests that the r.m.s. coordinate error for this structure is ,,-,0.22 A.relate to coordinate errors. His main critique is that the Luzzati method does not take intoaccount the fact that a certain portion of the coordinate errors will be absorbed into the thermalfactor parameters, thereby skewing the relationship between the structure factor probabilitydistributions and coordinate errors. However, this effect will probably be small at the end ofrefinement. Nevertheless, the existence of such systematic errors suggests overall coordinateerrors obtained with the Luzzati method should be used comparatively (Read, 1990).C.4 Cruickshank error estimatesWhereas Luzzati's method starts from the assumption that there is a certain error in the coor-dinates of the model and evaluates how this affects the structure factors, a method developedby Cruickshank (1949,1954) begins by assuming errors are present in the observed structureAppendix C. Estimating Coordinate Errors in Macromolecular Structures^170factors and determines how this error is reflected in the electron density map. Taking the dif-ferences between Fo 's and Fe 's as an estimate for the errors in the observed structure factorsand assuming well resolved spherically symmetric atoms, Cruickshank estimated the standarddeviation of a peak position to be (for orthorhombic, tetragonal and cubic cells):27riEhki h 2 (1Fohkii 1Fchki1) 2aV g24Similar equations for ay and az can be derived. In this equation a is the length of the crystal-lographic unit cell a axis, h is one of the Miller indices, and V is the volume of the unit cell.The term 3-2,4 represents the curvature of the electron density at the peak position which canbe estimated directly from an F, map. An alternative method for estimating this curvature isto use the thermal factor of the atom at the peak position (Chambers & Stroud, 1979; Readet al., 1983; Cruickshank, 1985). When this is done the above equation transforms into:CIVEhki h2 (1 F0hkil^IFChk11)227r Ehm(C.32)Here m is 2 for centric reflections, otherwise it is 1. The expression for the radial error in thetetragonal space group P4 32 1 2 is then given by:ar i = V2ax i 2 + zi 2 (C.33)since ax, = ay, (a = b and Fhki Fkhl) and the angles between the unit cell axes are 90°. Thusthe modified Cruickshank equation relates the standard deviation of an atom position to theatom type by virtue of the scattering factor foi and the thermal factor (see Figure C.59).Chambers and Stroud (1979) remark that ax is equal to the estimated r.m.s. shift in x duringa single cycle of real-space difference Fourier refinement. This implies that for cases where therefinement of a model has not yet converged, Cruickshank's method will underestimate thecoordinate errors. To counteract this problem, Chambers and Stroud propose that comparisonsbetween refinement models that are separated by more than one cycle of difference Fourierrefinement will give more realistic estimates for the coordinate errors. Read et al. (1983) adaptedcIT (C.31)Appendix C. Estimating Coordinate Errors in Macromolecular Structures^1711.000.750s-0 0.500000.250.000^10^20^30^40 2 soThermal factor (A )Figure C.59: Plot of the dependence of the radial coordinate errors on the value of the thermalfactor for different atom types in oxidized yeast iso-l-cytochrome c as estimated by the Cruick-shank (1949, 1954) method. Also shown are the r.m.s. shifts (using the numerical values on thevertical axis) after unrestrained refinement as a function of the thermal factor for carbon (Q),nitrogen (^) and oxygen (A) atoms.this idea by suggesting that instead of difference Fourier refinement, shifts from unrestrainedleast-squares reciprocal-space refinement based on a fully converged structure could be used.When unrestrained refinement is applied to the oxidized form of wild-type cytochrome c (seeFigure C.59), it is observed that the r.m.s. shifts correspond closely to the radial error estimatesderived with equation C.32. On this basis it can be concluded than that this cytochrome cstructure is well refined and that reliable atomic error estimates can be obtained using themethod developed by Cruickshank. However, care must be taken that the assumptions made inderiving equation C.32 are valid for the structure whose coordinate errors are being estimated.Specifically, in the derivation it is assumed that thermal factors correctly reflect the curvatureof the electron density at atomic positions. This assumption is generally true, but fails forsurface side-chains with high degrees of mobility when thermal factor restraints are applied inthe refinement.60^70

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.831.1-0086349/manifest

Comment

Related Items