Arginine Kinase: a Crystallographic Investigation of Essential Substrate Structure Shawn A

Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2006 Arginine Kinase: A Crystallographic Investigation of Essential Substrate Structure Shawn A. (Shawn Adam) Clark

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected]

THE FLORIDA STATE UNIVERSITY

COLLEGE OF ARTS AND SCIENCES

ARGININE KINASE; A CRYSTALLOGRAPHIC INVESTIGATION OF ESSENTIAL SUBSTRATE STRUCTURE

SHAWN A. CLARK

A Dissertation submitted to the Department of Chemistry and Biochemistry In partial fulfillment of the requirements for the Degree of Doctor of Philosophy

Degree Awarded: Fall Semester, 2006

The members of committee approve the dissertation of Shawn A. Clark defended on 11/2/2006

______Michael Chapman Professor Co-Directing Dissertation

______Tim Logan Professor Co-Directing Dissertation

______Ross Ellington Outside Committee Member

______Al Stiegman Committee Member

______Tim Cross Committee Member

Approved:

______Naresh Dalal, Department Chair, Department of Chemistry & Biochemistry

The Office of Graduate Studies has verified and approved the above named committee members.

ii ACKNOWLEDGEMENTS

The voyage through academic maturity is full of numerous challenges that present themselves at many levels: physically, mentally, and emotionally. These hurdles can only be overcome with dedication, guidance, patience and above all support. It is for these reasons that I would like to express my sincerest and deepest admiration and love for my family; my wife Bobbi, and my two children Allexis and Brytney. Their encouragement, sacrifice, patience, loving support, and inquisitive nature have been the pillar of my strength, the beacon of my determination, and the spark of my intuition. Their luminous personalities and quick wit have been a daily inspiration giving me the drive to push forward even during the most challenging of tasks or times. Without them I would have never been successful in my endeavor for scientific enlightenment and the pursuit of the career I so dearly love. I would also like to convey my deepest love and respect for my mother Vicki Clark. Her relentless work ethic and moral fortitude instilled at a very young age the virtues of perseverance and understanding that have guided my perspective on life, my approach to humanity, and my unwavering judgment to never compromise my scientific views. I would further like to extend my appreciation to Dr. Michael Chapman who was a constant source of inspiration, advice and encouragement. I am grateful for his guidance, knowledge and continued patience during my struggle to finish this research. I would also like to thank my supervisory committee members particularly; Dr. Tim Logan who’s relationship with my former teacher, Dr. Linda Luck, and overall dedication were the impetus for me attending graduate school and ultimately being successful. No young scientist can succeed as a single entity. The input, support, and diverse views of my peers were integral to my success and desire to succeed. For this I am eternally grateful and give a special thanks to one Dr. Donald “Jeff” Bush. I would not be the scientist that I am today without his influence of spirit and boundless understanding of scientific principles. He has gone above and beyond the call of duty on several occasions to help me and my family and I will consider him a true friend and scientific peer for life. I would further like to give thanks to my many laboratory colleagues that provided support in my research; Ms. Eliza Ruben, Mr. Omar Davulcu, Dr. Mohammad Yousef.

iii

TABLE OF CONTENTS LIST OF TABLES………………………………………………………………………... vii

LIST OF FIGURES………………………………………………………………………. viii

ABREVIATIONS...... ix

ABSTRACT………………………………………………………………………………..x

1. INTRODUCTION……………………………………………………………………... 1 1.1 High Energy Molecule…………………………………………………………. 1 1.2 Phosphagen Kinases and Substrate Diversity………………………………….. 3 1.3 Clinical Relevance……………………………………………………………... 5 1.4 Induced Fit……………………………………………………………………... 6 1.5 Mechanistic Studies……………………………………………………………. 9 1.6 Substrates Alignment…………………………………………………………... 11 1.7 Alternate Substrates and Enzyme Investigations………………………………. 13

2. MATERIALS AND METHODS……………………………………………………… 15 2.1 Materials……………………………………………………………………….. 15 2.2 Methods………………………………………………………………………... 15 2.2.1 Bacterial Cultures……………………………………………………. 15 2.2.2 Protein Overexpression in Bacteria………………………………….. 15 2.2.3 Protein Purification………………………………………………….. 16 2.2.3.1 Preparation of Lysate……………………………………… 16 2.2.3.2 Recovery of Crude Inclusion Bodies……………………… 16 2.2.3.3 Unfolding of Inclusion Bodies…………………………….. 16 2.2.3.4 Chromatography and Protein Refolding…………………... 16 2.2.4 SDS-PAGE (reducing conditions)…………………………………... 18

iv 2.2.4.1 Polyacrylamide Gel………………………………………... 18 2.2.4.2 Sample Preparation and Electrophoresis…………………... 18 2.2.4.3 Coomassie Staining………………………………………... 18 2.2.4.4 Silver Staining……………………………………………... 19 2.3 Protein Crystallization…………………………………………………………. 20 2.3.1 Crystal Growth………………………………………………………. 20 2.3.2 Crystal Freezing and Cryoprotection………………………………... 23 2.4 Diffraction Measurements…………………………………………………….. 23 2.4.1 Data Collection……………………………………………………… 23 2.5 X-Ray Crystallography Processing………………………………………….... 23 2.5.1 Indexing, Integration and Scaling…………………………………… 23 2.5.2 Conversion to Structure Factor Amplitudes…………………………. 24 2.5.3 Molecular Replacement……………………………………………… 25 2.6 Graphical Rendering and Structural Analysis…………………………………. 26

3. RESULTS……………………………………………………………………………… 40 3.1 Expression and Purification…………………………………………………… 28 3.2 Optimization of Crystallization……………………………………………….. 28 3.3 Data Collection………………………………………………………………... 34 3.4 Molecular Replacement and Refinement……………………………………… 37 3.5 Induced Fit……………………………………………………………………... 40 3.6 Active Site Coordination………………………………………………………. 45

4. DISCUSSION………………………………………………………………………….. 50 4.1 Protein Purification……………………………………………………………. 50 4.2 Crystallization…………………………………………………………………. 50 4.3 Induced Fit and Dynamic Domains…………………………………………… 51 4.4 Enzyme Stabilization………………………………………………………….. 53 4.5 Active Site Coordination………………………………………………………. 55

v 4.5.1 Active Site Pre-organization………………………………………… 55 4.5.2 Overlay of Wild Type and Analogue Complexes…………………… 59 4.5.3 Inter-residue Distances in the Active Site…………………………… 60 4.6 Ligand Binding………………………………………………………………… 61 4.6.1 Phosphagen Bonding Restraints……………………………………... 61 4.6.2 Mg+2 and Nitrate Coordination………………………………………. 61 4.7 Reactive Alignment……………………………………………………………. 63

5. CONCLUSION………………………………………………………………………... 66

APPENDICIES A Supplies…………………………………………………………………………. 68 B Chemicals………………………………………………………………………... 69 C Solutions…………... ……………………………………………………………. 70 D Buffers…………………………………………………………………………… 71 E Instruments………………………………………………………………………. 72 F Computational Software………………………………………………………… 73

REFERENCES………………………………………………………………………….... 74

BIOGRAPHICAL SKETCH…………………………………………………………….. 86

vi LIST OF TABLES

Table 3.3.1. Data processing and model statistics………………………………………… 36

Table 3.5.1. Residue content of dynamic domains………………………………………... 41

vii

LIST OF FIGURES

Figure 1.1.1 Graphical representation of ATP……………………………………………. 1

Figure 1.2.1 Chemical representation of known phosphagens…………………………… 3

Figure 1.2.2 Secondary structure of arginine kinase……………………………………... 5

Figure 2.3.1 Graphical representation of arginine analogues…………………………….. 22

Figure 3.1.1 Silver stained SDS-PAGE of arginine kinase purification steps……………. 28

Figure 3.2.1 Room temperature arginine kinase crystals (non-optimized conditions)…… 29

Figure 3.2.2 Room temperature arginine kinase crystals, first optimization……………... 30

Figure 3.2.3 Room temperature arginine kinase crystals, fianl optimization…………….. 31

Figure 3.2.4 Graph of the refractive index………………………………………………... 32

Figure 3.2.5 Schematic saturation diagram of arginine kinase (TSAC)………………….. 33

Figure 3.3.1 Diffraction pattern of arginine kinase (TSAC)……………………………… 35

Figure 3.4.1 Progress of molecular refinement (Rfree)……………………………………. 38

Figure 3.4.2 Ramachandran plot of the final model for ILO……………………………... 39

Figure 3.5.1 Ribbon diagram of DynDom analysis………………………………………. 42

Figure 3.6.1 Wild type arginine kinase superimposed on analogue structures…………… 47

Figure 3.6.2 Schematic representation of inter-residue distances in the active site………. 49

Figure 4.3.1 Ribbon diagram of the non-cognate structures superimposed on the native arginine kinase TSAC structure…………………………………………………….. 53

Figure 4.5.1.1 Active site water-bonding network…………………………………………57

Figure 4.5.3.1 Surface representation of the binding pocket for the analogue complexes.. 60

Figure 4.7.1 ADP bonding angles…………………………………………………………. 64

viii

ABBREVIATIONS

AAC-alternate analogue complex LARG-L-arginine ADP-adenosine diphosphate L.B.-Luria broth AK-arginine kinase L.B.-Amp-Luria broth + ampicillin APS-ammonium persulfate MI-myocardial infarction ATP-adenosine triphosphate NMP-nucleoside monophosphate CIT-L-citrulline NMR-nuclear magnetic resonance CK-creatine kinase NMWCO-nominal molecular weight cutoff DARG-D-arginine NO-nitric oxide DEAE-diethylaminoethyl ORN-L-ornithine DNA-Deoxyribonucleic acid PAGE-polyacrylamide gel electrophoresis DTT-dithiothreitol PEG-polyethylene glycol EDTA-ethylenediaminetetraacetic acid PMSF-phenylmethylsulfonyl fluoride EPR- electron paramagnetic resonance PPR-proton paramagnetic resonance HEPES-(N-[2-Hydroxyethyl]piperazine-N’- R.M.S.D-root-mean-square-deviation [2-ethanesulfonic acid]) SDS-sodium dodecyl sulfate HR-high-resolution TSAC-transition state analog complex ILO-imino-ethyl-ornihtine (nitrate, MgADP, arginine) IPTG-isopropyl-beta-D- WT-wild type thiogalactopyranoside

ix ABSTRACT:

Phosphagen kinases are a family of enzymes that play a role in high-energy cells by buffering the ATP concentration. Arginine kinase (AK) catalyzes the magnesium-dependent reversible transphosphorylation between N-phospho-L-arginine (PArg) and ATP. Previous investigations of arginine kinase have suggested that precise substrate alignment may play an important role in this bimolecular reaction and that phosphagen specificity is mediated by an induced fit mechanism. In order to uncover the mechanism of catalysis we present the crystallographic structures of arginine kinase with four arginine homologues: L-imino-ethyl-ornithine (ILO), L- citrulline (CIT), L-ornithine (ORN), and D-arginine (DARG) replacing the cognate arginine substrate. Each homologue was co-crystallized with MgADP- and nitrate as in the transition state analog complex (TSAC). The models were refined at 2.4Å, 2.0Å, 2.0Å and 2.8Å with Rfree values of 23%, 25%, 24% and 23%, respectively. These structures were used to study the basis of substrate specificity and the role of substrate alignment in catalysis. The non-cognate ligands bind and induce the substrate-bound enzyme conformation but are inactive. These data show that although AK is highly specific a number of different phosphagen mimics can bind, but are non-reactive. Comparison of the structures indicates that substrate specificity might be mediated by a mechanism that allows only the cognate substrate to be held rigidly in the proper orientation for efficient catalysis. These data also suggest that substrate alignment may play, at least, a role in catalysis. Here it is suggested that the induced fit movements of the enzyme most likely do not play a role in substrate discrimination but do play a role in substrate alignment. The structures also show that the guanidinium and α-amine moieties contribute little to substrate binding relative to the carboxylate, which appears to be key to substrate binding.

x CHAPTER 1 INTRODUCTION:

1.1 High-Energy Molecules When a cell needs to make a reaction go in an energetically unfavorable (endergonic) direction, it uses a mechanism termed coupling: it converts the substrate to something with higher chemical potential by using another reaction that is energetically favorable (exergonic). The chemical potential of the modified substrate is higher than that of the product so that the reaction will now tend to proceed spontaneously in the desired direction. There are many high-energy molecules in nature: phosphoenolpyruvate, acetyl phosphate, carbamyl phosphate, pyrophosphate, p-Nitrophenyl acetate, phosphocreatine, phosphoarginine and 1, 3-bisphoglycerate are just a few molecules that are used in biological processes. ATP is the most widely distributed high-energy compound within the human body (Ritter, 1996). ATP is an abbreviation for adenosine triphosphate, a complex molecule that contains the nucleoside adenosine (which consists of the purine base adenine and the sugar ribose) and a tail consisting of three phosphates (figure 1.1.1).

Fig. 1.1.1. A graphical representation of ATP.

This ubiquitous molecule is used to build complex molecules, contract muscles, and generate electricity in nerves. Nearly all fuel sources can be used to produce ATP, which in turn powers virtually every activity of the cell and organism. As far as known, all organisms from the simplest bacteria to humans use ATP as their primary energy currency. A steady supply of ATP is so critical that a poison that attacks any of the proteins used in ATP production can have detrimental effects on an organism. Certain cyanide compounds, for example, are poisonous because they bind to the copper atom in cytochrome oxidase. This binding blocks the electron

1 transport system in the mitochondria where ATP manufacture occurs (Goodsell, 1996, p.74). Mitochondria produce on the order of 30 molecules of ATP from each molecule of glucose that is oxidized. ATP is converted to phosphocreatine by the mitochondrial creatine kinase (MiCK), for shipment to the energy-requiring systems in the cytosol and other targets in the cell such as ion transport associated ATPases. In general the “burning” of an ATP molecule is coupled to another reaction (the two reactions occur at the same time and at the same place), usually utilizing the same enzyme complex. The release of a phosphate from ATP is exothermic and the reaction it is connected to is endothermic. The terminal phosphate group is transferred to another compound, a process called phosphorylation, producing ADP, a phosphate (Pi) analogue, and energy. The high- energy bonds of ATP are actually rather unstable bonds. Because they are unstable, the energy of ATP is readily released when ATP is hydrolyzed in cellular reactions. One should realize that ATP is an energy-coupling agent (not a fuel) and it is not a storehouse of energy maintained for future use. It should also be noted that the energy of the phosphoanhydride bonds between the terminal phosphates are not necessarily “high”, just the free energy of hydrolysis. ATP is produced by one set of reactions and immediately consumed by another set of reactions. ATP is formed as needed, primarily by oxidative processes in the mitochondria. ATP is not kinetically unstable but rather is structured so that its hydrolysis is slow in the absence of a catalyst. This insures that its energy is utilized only in the presence of the appropriate enzyme (McMurry and Castellion, 1996). The hydrolysis of one mole of ATP yields 7.3 kcal of energy under standard conditions but the free energy change for the hydrolysis of ATP in the cell may be much greater. It is used for many cellular functions including transport (moving substances across cell membranes) and mechanical work (supplying the energy needed for muscle contraction). It supplies energy not only to heart and skeletal muscle but also to many other cellular processes enabling them to carry out their many functions. The amount of free energy available for different cellular processes is dependent on the ATP/ADP ratio as defined by the following equation (Kammermeier, H. 1987; Kammermeier, H. 1993; Ellington, W.R. 2001):

-1 ∆GATP= Gº´+ R · T · ln{(ATP) · [(ADP) (Pi)]} (Equation 1)

Hence, when ATP is consumed for cellular processes it must be rapidly regenerated to provide

2 for further energy requirements. ATP is regenerated by glycolysis but at a rate much slower than what is required to maintain a steady concentration of ATP (McGilvery, 1979). The cell can maintain a constant amount of ATP by transferring a phosphate group from a phosphorylated guanidine “storage molecule”, such as creatine or arginine, to ADP. The enzymes of the phosphagen kinase family catalyze this reaction.

1.2 Phosphagen Kinases and Substrate Diversity Phosphagen kinases are a family of enzymes that play a role in regulating the dynamic energy needs of high-energy cells by maintaining the concentration of ATP (Schlattner et al., 2000). Phosphagen kinases are found in diverse systems throughout nature and are exemplified by creatine kinase (CK) in mammals (Watts, 1973) and arginine kinase (AK) in invertebrates (Morrison, 1973). Over the last several decades these enzymes have been studied by kinetic, mutagenic, biochemical, and structural methods.

Fig. 1.2.1. A chemical representation of known phosphagen molecules.

Phosphagen kinases show varying specificity for substrates (phosphagens) that differ in size and electrostatic potential (Ellington, 2001; Kenyon and Reed, 1983; Kurby et al., 1954). All phosphagens contain a guanidinium group but the substituents of the α-carbon atom are different (figure 1.2.1). Of all phosphagens, only arginine and lombricine have an α-amino group.

3 Despite the structural variation in this region of the substrate, there is a lack of corresponding amino acid sequence variability in the corresponding sub-site of the enzyme active site. In fact, for a diverse family of enzymes that catalyze different reactions, the phosphagen kinases are remarkably well conserved, sharing approximately 40% sequence identity (Babbitt et al., 1986; Mühlebach et al., 1994; Dumas & Camonis 1993). During the last decade several structures of several phosphagen kinases have been determined. Below is a partial list of representative structures: octameric mitochondrial CK: substrate-free……………………....(Forstner et al., 1996) chicken sarcomeric mitochondrial CK: with and without ATP…(Fritz-Wolf et al., 1996) arginine kinase (TSAC):………………………...(Zhou et al., 1998; Yousef et al., 2003) arginine kinase: substrate-free………………………………………(Yousef et al., 2003) mutant arginine kinase (TSAC): E314D, E225Q…………………….(Pruett et al., 2003) mutant arginine kinase (TSAC): C271A……………………………..(Gattis et al., 2004) rabbit muscle CK:………………………………………………………(Rao et al., 1998) chicken cytosolic brain-type CK:……………………………………...(Eder et al., 1999) cytosolic bovine retinal CK:……………………………………………(Tisi et al., 2001) human ubiquitous mitochondrial CK:…………………………………(Eder et al., 2000) human mitochondrial CK:…………………………………………...... (Shen et al., 2001) As a general rule, AK’s are predominantly monomeric whereas CK’s are either dimeric (cytosolic) or octameric (mitochondrial). The enzymes show very similar overall structures. Specifically, arginine kinase consists of one large domain and one small domain. The smaller N- terminal domain consists of four alpha helices and two beta sheets while the larger C-terminal domain consists of an eight-stranded anti-parallel beta sheet and five alpha helices (Figure 1.2.2). These structural investigations have produced a significant amount of data relevant to the catalytic mechanism of phosphagen kinases and provide a paradigm for addressing fundamental questions about the nature of bimolecular reactions.

Fig. 1.2.2. Ribbon projection of the secondary structure of arginine kinase in the TSAC form. Image generated with Web-lab Viewer Pro.

1.3 Clinical Relevance Phosphagen kinases are found in many different organisms where they serve as major regulators of intracellular energy exchange (Ellington, 1989). Lower chordates and the deuterostome invertebrates express both AK and CK, whereas CK is the only phosphagen kinase found in vertebrates (reviewed in; Ellington, 2001). The absence of AK in higher organisms suggests it may serve as a target for selective eradication of several pathogens. For example, in 2000 an AK was cloned from the protozoan parasite Trypanosoma cruzi (Pereira et al., 2000). More recently molecular and biochemical characterizations of AK’s in both Trypanosoma cruzi and Trypanosoma brucei, the etiological agents of Chagas’ disease and human sleeping sickness respectively, indicate that these enzymes may serve as targets for the development of inhibitor compounds that are able to destroy the parasite while sparing host tissues (Alonso et al., 2003; Pereira et al., 2004). Interestingly, AK has also recently become an attractive target for the control of cockroach populations (Brown and Grossman, 2004).

5 AK is believed to be the ancestral phosphagen kinase from which CK evolved. The two enzymes share many properties although CK’s display a more complex polymeric state. In man, two mitochondrial CK’s occur mainly as octamers (acidic, Mia- and basic, Mib-CK), with Mib-

CK being the sarcomeric muscle specific isoform, and Mia-CK being the form found in brain and most other tissues (Walliman et al., 1992). In addition three cytosolic isoenzymes exist in homo- or heterodimeric form (BB-, MM-, and MB-CK). MM is dominant in skeletal muscle, BB in the central nervous system, and MB constitutes a significant portion (15%) of cardiac muscle. The distribution of these isoforms among different tissues has been used in clinical diagnosis of various disorders, including acute myocardial infarction (MB isoform) (Pasternak and Braunwald, 1991), and muscular dystrophy (MM isoform) (Griggs et al., 1991). The potential role of CK’s in the pathogenesis of cardiomyopathy, muscular dystrophy, and diseases of cellular energy homeostasis are in the early stages of investigation (Ramelli et al., 2006; Khuchua et al., 1992; Saks et al., 1991). Nevertheless, an understanding of AK’s reactive mechanism is very likely to contribute to the future understanding of the role of CK in human disease.

1.4 Induced Fit A common link between all proposed enzyme catalyzed reactions is the formation of an enzyme- substrate complex from which catalysis takes place. This concept was independently proposed in 1902: first by Brown and then by Henri. The concept extends the “lock-and-key” hypothesis which Emil Fischer proposed (1894) that an enzyme is like a lock and the substrate is like a key that fits into the lock. This hypothesis accounts for the high degree of specificity in some enzymes, but doesn’t explain other observations. For example, compounds whose structures are related to that of the natural substrate, but posses less bulky substituent, often fail to serve as substrates even though they fit the binding pocket. On the other hand compounds with bulkier substituent may bind more tightly to the enzyme than the natural substrate. Also, some enzymes that catalyze reactions between two substrates will not configure the catalytic residues until the other substrate is bound to the enzyme. The latter led to the proposal of an “induced-fit” hypothesis (Koshland., 1958; 1971). This hypothesis states that when a substrate begins to bind to an enzyme, interactions of various groups on the substrate with particular enzyme functional groups are initiated, and these interactions provoke a conformational transformation in the

6 enzyme itself. This rearrangement of the enzyme from a low to a high catalytic form may result in the proper alignment of catalytic groups. Compounds akin to the natural substrate, except with smaller or larger substituents, may bind to the enzyme but may not induce the conformational changes necessary for catalysis. Also, different substrates may induce non- identical forms of the active site. In this way conformational changes could serve as a basis for substrate specificity. It should be noted that the induced fit mechanism could be disadvantageous. In some cases the energy required to make the protein change conformation adds to the enthalpy of the activation barrier and reduces catalysis (Fersht, 1985). Nevertheless, these conformational changes may be necessary in some cases to exclude water from the active site thereby reducing unwanted free hydrolysis of the high-energy phosphoryl moiety. To date a significant amount of insight into domain motions and induced fit mechanisms have come from analyzing substrate bound and substrate-free structures by crystallography. The results of early investigations were reviewed by Janin and Wodak (1983). Domain motions can be classified into two general types, shear and hinge (Gerstine et al., 1994; Wriggers and Schulten, 1997). In addition to global domain motions localized movements can occur as noted in many cases in which a flexible loop folds over the active site upon substrate binding (Joseph et al., 1990; Schreuder et al., 1993; Zhou et al., 2000). Several proteins have been analyzed for induced-fit structural changes (reviewed by Gerstein et al., 1994 and Hayward, 1999). One of the most studied kinase families is the nucleoside monophosphate (NMP) kinase family which catalyzes the addition of a phosphate group to a monophosphate core. This structure has been investigated in many forms (substrate- free, binary and ternary) and these investigations provide a foundation for the catalytic mechanism (reviewed by Yan and Tsai, 1999). These enzymes consist of a LID, a core, and an NMP-binding domain. The binding of either substrate induces conformational changes in the LID and NMP-binding domains that coordinate the active site. In phosphoglycerate kinase binding of the triose sugar induces larger conformational changes than binding of the nucleotide substrate (Pickover et al., 1997; Davis et al., 1994), but both substrates are required to generate the conformation that is essential for catalysis (Cheung and Mas, 1996; Bernstein and Hol, 1998). Conformational changes in AK and CK are seen through the evaluation of the substrate- free and substrate-bound crystal structures (Yousef et al., 2003). Evidence supporting substrate-

7 induced conformational changes in AK and CK include spectroscoptic studies of the active site (Reed and Cohn, 1972), tryptic susceptibility differences (Lui and Cunningham, 1966), and small-angle x-ray scattering (Dunmas and Janin, 1983; Forstner et al., 1998). These data show that in AK and CK significant structural changes occur when the Mg-nucleotide or the transition state analogue complex (Mg+2, ADP, nitrate, and arginine or creatine) bind to the enzyme. However, these changes do not occur with the nucleotide or phosphagen alone. The decrease in the radius of gyration seen during these investigations suggests that the conformational changes noted in the crystal structures occur in solution and are not a crystal artifact. A Dynamic Domain (DynDom) computational analysis assigns a rotational axis and a translation vector for each residue of a protein proceeding between two conformational states, such as substrate-free to substrate bound form. A clustering algorithm matches residues that have similar rotation axis and translational magnitude and defines them as dynamic domains. A DynDom analysis of the substrate-free versus substrate bound forms of AK (Yousef et al., 2003) showed that the dynamic domain movements in AK are much more complex than the traditional two domain view. It is shown that there are substantial conformational changes induced in four separate domains of the enzyme. These movements allow intimate interactions to form between the enzyme and both the nucleotide and phosphagen substrates and are thought to be primarily induced upon nucleotide binding (Forstner et al., 1998). These conformational changes appear largely responsible for the stabilization of two flexible loops and the subsequent coordination of critical reactive residues in the active site. One of these loops had been suggested to be a “substrate specificity” loop. It is within the small domain (residues 62-68), is of a length that is inversely related to the size of the phosphagen substrate and thought to participate in a lock-and-key mediation of specificity (Suzuki et al., 2000). This function was tested by the kinetic analysis of several chimeric constructs designed with CK loop elements in an AK backbone (Azzi et al., 2004). These constructs had measurable arginine kinase activity, no new creatine kinase activity, and no significant effect on the relative binding of phosphoarginine. These data suggest that this loop alone could not account for the high degree of discrimination of this enzyme.

8 1.5 Mechanistic Studies Over a half a century of kinetic, spectroscopic and crystallographic investigations have provided a significant amount of mechanistic data for phosphagen kinase reactions (Kenyon and Reed, 1983). These data suggest that phosphagen kinases share a common catalytic mechanism (Murali et al., 1993; Murali et al., 1994). Steady state kinetics and product inhibition studies indicate a rapid equilibrium, random-order, bimolecular-bimolecular reaction (Blethen, 1972; Hansen and Knowles, 1981) with synergy of binding. Stereo-specific isotope investigations suggest a direct, in-line γ-phosphoryl transfer (Hansen and Knowles, 1991). The pattern of chemical shifts and line shapes of the nucleotide complexes of CK parallel those for the corresponding complexes of AK, indicating structural and/or conformational similarity of the phosphate chains of nucleotides bound to the two enzymes (Rao, 1981). Despite their overall similarities a difference in the active site is indicated by the pH independence (pH 6.0-9.0) of the chemical shift of the β-phosphate of MgADP bound to creatine kinase, whereas with AK this resonance shows a pKa approximately 7.5 (Murali, 1993; 1994). The NMR data further suggest that the nucleotide constituent of the transition state analogue complex retains its overall structure regardless of the phosphagen moiety (Ray et al., 1996). The putative transition state of both CK and AK can be mimicked in a stable transition state analogue complex (TSAC) with Mg-ADP, creatine or arginine phosphagen, and a nitrate molecule that mimics the terminal phosphate of ATP during in-line transfer (Milner-White and Watts, 1971). Arginine kinase specifically catalyzes the magnesium-dependent reversible transphosphorylation between N-phospho-L-arginine (PArg) and ATP.

MgADP + PArg + H+ ⇔ MgATP + Arginine (Equation 2)

In this scheme the forward reaction is acid catalyzed and the backward reaction is base catalyzed and most likely requires that the active site contain at least one residue that can act as both. Abstraction of a proton from the guanidinium (Cook et al., 1981) is thought to be an early catalytic step. Inspection of the high resolution AK TSAC structure showed that Glu225 and Glu314 were the only potential catalytic residues close to the guanidinium (Yousef et al., 2003). However, a kinetic evaluation of several site-directed mutants of these residues ruled them out as catalytic bases (Pruett et al., 2003). Nevertheless, it was demonstrated that these residues

9 enhance the reaction rate and may mediate their effect by restricting substrate motion or in substrate alignment. This hypothesis is supported by the structure of the mismatched complex of creatine and MgADP in AK (Azzi et al., 2004). This structure shows that the guanidinium of the creatine ligand occupies a near equivalent position to the native arginine guanidinium. However creatine fails to be a substrate for AK. This is most likely due to the lack of enzyme mediated restraints on the substrate position and loss of substrate alignment. A conserved cysteine residue has been implicated in catalysis. The catalytic role of this residue was investigated by mutating it to serine, alanine, asparagine, and aspartate (Gattis et al., 2004). Steady-state binding constants were slightly increased, more so for phospho-L-arginine than ADP. Substrate binding synergy (defined as an increase in the binding affinity of a ligand in the presence of a second ligand, and expressed as the Km/Kia) observed in many phosphagen kinases, was reduced or eliminated in these mutant enzymes. Together, these results contradict the prevailing point of view that the cysteine mediates a substrate-induced conformational change. Furthermore, they confirm that it is the thiolate form that is relevant to catalysis, and suggest that one of its roles is to help to enhance the catalytic rate through electrostatic stabilization of the transition state. The importance of metal ions to the catalytic process of many enzymes has been well documented (Lee et al., 2000; Hanna and Doudna, 2000; Allen et al., 1994; Adams and Taylor, 1993). The role of the magnesium ion in both AK and CK has been probed by substitution with other divalent metal ions (Mn+2, Co+2, Zn+2, Cd+2) by Infra-Red Spectroscopy (IR), Nuclear Magnetic Resonance (NMR), Proton Paramagnetic Resonance (PPR) and Electron Paramagnetic Resonance (EPR) methods (Morrison, 1973; Ray et al., 1998; Jaori et al., 1998; Buttlaire and Cohn, 1974; Kenyon and Reed, 1983). An analysis of the geometry of metal-ligand interactions relevant to proteins shows that the metal-oxygen distances have a mean range of 2.07Å (Harding, 1999). In the high-resolution TSAC structure of AK the magnesium has two coordination shells. The first shell is at a radius of 2Å and consists of O1α, O1β of ADP, O1 of nitrate and three water molecules. The second coordination shell is at 4Å and consists of residues Glu224, Glu225, and Glu314. These negatively charged residues interact indirectly with the magnesium via hydrogen bonds with three magnesium water ligands. The kinetic analysis of several site directed mutants of residues Glu225 and Glu314 showed a decrease in catalytic rate but very little effect on the binding affinity for the cognate arginine ligand. These data suggest that

10 residues 225 and 314 play an accessory role in catalysis and most likely mediate their effect by positioning the guanidinium (Pruett et al., 2003). In AK the magnesium likely plays four major roles in catalysis: (1) It allows close approach of the gamma and beta phosphate groups of the nucleotide; (2) It facilitates large conformational changes in going from the substrate-free to substrate bound forms (Forstner et al., 1998); (3) It helps configure the active site by coordinating residue Glu314; (4) It most likely functions by withdrawing electrons from the Pγ-

OPβ bond or in positioning the phosphate groups.

1.6 Substrate Alignment Although infrequently appreciated, proximity and alignment are key components of the reactive process in many bimolecular reactions (Bruice and Benkovic, 2000; Bruice, 2002). Because of entropic contributions the complexity involved in coordinating the reactive moieties of two separate substrate molecules is increased. Hence, it is likely that alignment could play a larger role in bimolecular reactions than unimolecular reactions. However, there is no consensus on the importance of alignment relative to other catalytic mechanisms such as: (1) acid-base catalysis; (2) substrate strain; (3) covalent catalysis; (4) transition-state stabilization; and (5) ground-state destabilization. As a bidirectional catalyst, arginine kinase is an excellent paradigm for the study of questions fundamental to the catalysis of multiple substrate reactions. In the 1970's, Jencks and Koshland argued over the relative importance of substrate alignment in multi-substrate enzymes, though they agree that these proteins differ from unimolecular enzymes. Koshland proposed that alignment of substrates could be an important component of catalysis due to special “orbital steering” effects (Koshland., 1952; Bruice TC., 1972; Dafforn et al., 1971; Hoare DG., 1972; Moriarty RM and Adams T., 1973; Mesecar et al., 1997) wereas Jenks suggests that proximity effects maybe dominated by translational entropy and are insufficient to contribute significantly to catalysis. Some of the contention has arisen from a narrow focus on entropic terms, but calculations on unimolecular and bimolecular systems indicate that the adoption of near-attack conformers is accompanied by larger enthalpic changes than entropic (Lightstone and Bruice, 1996,1998). Thus, there are compelling reasons to understand substrate pre-alignment whenever there is a possibility to observe it directly. Evidence to support Koshland’s orbital steering

11 suggestion was obtained by structural modification studies with isocitrate dehydrogenase (Wolfenden et al., 1999). Small modifications were made to the structure of the cofactors for this enzyme, which led to a slight misalignment of the bound cofactors. Because the substrate must react with one of the cofactors during catalysis, misalignment of the cofactor would translate into a perturbed reaction trajectory that should affect the catalytic power of the enzyme. In fact the reaction with the modified cofactors resulted in large decreases in the reaction rate with only small changes in the orientation of the substrates. In contrast evidence to support Jenks’ point of view stemmed from measurements of TΔS‡ that showed small entropic contributions to unimolecular enzyme catalysis. Specifically these results suggest that (i) many of the motions that are free in the reactant state of the reference solution are also free at the transition state. (ii) binding to the enzyme does not completely freeze the motion of the reacting fragments so that the entropy at the transition state is not zero. These data agree with other previous theoretical studies (Villa et al., 2000; Sham et al., 2000; Warshel A., 2003). Thus the role of substrate alignment in bimolecular catalysis remains an important topic due to the larger entropic effects in multi-substrate reactions and because pre-organization involves both entropy and enthalpy. The high-resolution, wild type AK structure (1.2Å) in the TSAC (Yousef et al., 2003) was solved as a non-covalent complex with no perturbation of the enzyme structure. Thus, in this model visualization of substrate alignment is free of many of the biases reported in prior TSA structure analyses where substrates were covalently modified to achieve a stable complex. The high-resolution AK structure shows very precise alignment of substrates with angles of approach within a few degrees of ideal for in-line transfer of the gamma phosphate of ATP. The donor and acceptor atoms (ADP-O3β and guanidyl Nη2 of substrate arginine) are positioned to form bonds almost orthogonal to the nitrate plane. Regarding the nucleophilic attack by the guanidyl Nη2, the optimal angle for the attacking lone pair is ≈ 110° to the Nη2-Cγ bond in a plane perpendicular to the guanidine plane (Zhou et al., 1998). The transition state structure of AK shows the approach angle between arginine and nitrate to be 112°, 2° from ideal. For the reverse reaction, extended Hückel calculation of phosphate valence electron density suggests that the optimal angle for nucleophilic attack by the ADP-O3β is 100° (Zhou et al., 1998) with respect to the Pβ-O3β bond. In the high-resolution structure the observed angle is ≈ 96°, 4° from optimal. This precision of alignment is similar to that envisioned in Koshland’s orbital steering hypothesis

12 (Koshland., 1952; Bruice TC., 1972; Dafforn et al., 1971; Hoare DG., 1972; Moriarty RM and Adams T., 1973; Mesecar et al., 1997). Even though these calculations were called into question by Jenks, the alignment is more precise than required by his original estimate of the impact of entropy on reaction rate. Nevertheless, it is still not possible to discriminate between several possibilities: (i) is the precise alignment purely accidental? (ii) has the enzyme evolved to exploit marginal improvements in rate on the basis of precise alignment? or (iii) are the catalytic effects of precise alignment underestimated by the approximations of current theory (Yousef et al., 2002). The ability to study a bimolecular reactive mechanism trapped at the transition state provides a unique opportunity to uncover how subtle structural changes can lead to large differences in specificity of metabolic enzymes.

1.7 Alternate Substrates and Enzyme Investigations For many years enzyme-substrate interactions have been studied by both classical kinetics and biophysical techniques. A significant portion of our current understanding of enzyme-ligand interactions has been derived from the investigation of enzyme mutants and alternate ligand complexes. To date structural data on unfavorable protein-ligand interactions are sparse, partly because structures of weakly bound ligands are more difficult to obtain. However, weakly binding ligands provide important information on catalysis and the role of substrate restraints in relation to conformational stability. This is especially true when minor changes in substrate chemistry have a significant impact on catalysis. These systems can be used to investigate the minimal bonding restraints necessary for ligand coordination. Several potential phosphagen substrates including, L-citrulline (CIT), L-ornithine (ORN) and D-arginine (DARG) and more recently imino-ethyl-ornithine (ILO) have been used to investigate kinetic and structural parameters of enzyme catalysis in different systems. These ligands are present in several structures in the Protein Data Bank (PDB) including those of: lysine-arginine-ornithine binding protein (1LAH), ornithine-transcarbamoylase (1FB5, 1A1S, 1C9Y, 1EP9), agininosuccinate synthetase (1J21), dimethylarginine dimethylaminohydrolase (2C6Z), and arginine deiminase (1LXY) to name a few. These same substrates have been used to investigate AK by kinetic, NMR, EPR and PPR studies (Pereira et al., 1991; Buttlaire and

13 Cohn 1974; Smith, 1978; Urschel et al., 2006). Previously measured inhibition constants, for L- arginine derivatives indicate that in AK substrate affinity depends on the length of the carbon chain connecting the guanidine and amino acid moieties of the phosphagen substrate (L- isoleucine>L-valine>L-α-aminobutyrate>L-alanine) as well as the bonding interactions of the α- carbon atom (L-arginine>agmatine and L-ornithine>putrescine) (Watts et al., 1980). It has also been shown that in AK arginine analogues lacking the α-amine competitively inhibit L-arginine binding and also protect against inactivation by 5,5′-dithiobis-(2-nitobenzoic acid), which binds to solvent exposed cysteins (Feng et al., 2005; Pan et al, 2004). Interestingly, in AK analogues lacking the carboxylate group do not protect (Watts et al., 1980). The structure of the mismatched complex of creatine and MgADP in AK shows the guanidinium occupying a near equivalent position to the native arginine guanidinium (Azzi et al., 2004). Nevertheless there is a 40-fold reduction in the turnover rate, which the authors hypothesize is most likely due to the lack of enzyme mediated restraints on the substrate position and loss of favorable entropic contributions to the reactive process. These preliminary data prompted the use of these ligands (L-imino-ethyl-ornithine, L-ornithine, L-citrulline, and D- arginine) in a crystallographic investigation to determine the structural basis for the observed kinetic parameters. Of particular interest is the investigation of the effects of substrate alignment in catalysis and to determine the parameters governing substrate binding and coordination of the ligand enzyme complex. These ligands retain a varied representation of guanidinium configurations with DARG retaining the cognate guanidinium, ILO containing an sp3 nitrogen, ORN containing an sp2 nitrogen and CIT being neutral.

CHAPTER 2 MATERIALS AND METHODS 2.1 Materials All materials were purchased from Sigma-Aldrich unless otherwise specified. A detailed list of all consumable materials, chemicals, reagents, solutions, buffers, instrumentation and software used in these experiments are listed in the appendices.

2.2 Methods The protein samples used in the crystallographic investigations presented here were prepared from a pre-existing bacterial expression construct, Strong and Ellington (1996). The protein was purified from inclusion bodies and refolded. Outlined below are the methods used for expression and purification of an AK protein sample.

2.2.1 Bacterial Cultures All bacterial cultures were processed in a sterile working environment. Gloves were worn at all times and sterile pipet tips, bottles, and culture flasks were used. For overnight cultures to be used for inoculation of expression cultures, 10ml of L.B.-Amp media in a Falcon® (15ml) tube were inoculated from a frozen bacterial stock (maintained at -20°C) and incubated overnight at 37°C in an incubator shaking at 200rpm (rotation radius of 5cm) overnight. To maintain a constant plasmid stock, larger cultures were used for plasmid preparation (Midiprep): 200ml of L.B. media in an Erlenmeyer flask were inoculated with E.coli DH5α cells containing the original pET vector constructed by Strong and Ellington (1996).

2.2.2 Protein Overexpression in Bacteria Arginine kinase (Limulus polyphemous) was cloned and expressed in E. coli BL21-DE3 (Strong & Ellington, 1996). Cultures were grown in L.B. media under ampicillin selection and induced with 1mM IPTG at mid log phase (O.D.600= 0.5-0.6), as measured in a plastic cuvette (using L.B.-Amp media as a blank). Cells were then incubated for 12hr at 37° with constant rotation (mixing) at 250rpm.

2.2.3 Protein Purification 2.2.3.1 Preparation of Lysate One liter of induced E. coli cells was centrifuged at 6,000 x g for 15min at 4°C using a Beckman JA10 rotor. The pelleted cells were resuspended in 40ml of lysis buffer (400µl of fresh PMSF [100mM] was added just prior to lysis) and lysed with a French press at 1250 psi. The sample was passed through the pressure cell twice to increase the efficiency of lysis. The lysate was cleared by centrifugation at 10,000 x g in a JA14 rotor for 30 min at 4°C.

2.2.3.2 Recovery of Crude Inclusion Bodies The inclusion bodies were washed to remove lipid, DNA, and contaminating protein. Washing proceeded as follows: the inclusion bodies were resuspended in 50ml of wash buffer, incubated for 30 min at room temperature, then spun at 5000 x g in a Beckman JA20 rotor. This process was repeated four times. In the final step the buffer was removed and the inclusion bodies were resuspended in a storage buffer (10mM Tris-HCl, 1mM EDTA, 1mM DTT, 10mM KCl, 0.02%

NaN3, pH=8.0)

2.2.3.3 Unfolding of Inclusion Bodies Although 1M urea is typically adequate to denature a protein, this is not true for AK as determined by difference absorption urea titration (Copeland, 1993), which showed a sigmoidal unfolding transition between 2M and 4.5M. These data were confirmed by others (France & Grossman, 1996). Therefore the washed AK inclusion bodies were unfolded for one-hour incubation in 8M urea at room temperature. The unfolded protein solution was clarified by centrifugation at 8,000 x g for one hour in a Beckman JA20 rotor and then filtered through a 0.22µm membrane filter.

2.2.3.4 Chromatography and Protein Refolding The unfolded protein was diluted from 8M to 6M urea and partially purified by size exclusion chromatography in the unfolded state. The sample was passed over a Sephacryl® S300 column [driven by FPLC (AKTA®)] pre-equilibrated in a buffer containing 6M urea (S300 buffer). The sample was run at a flow rate of 0.4ml/min and 0.5ml fractions were collected starting 2.5h after

16 the sample was loaded. The peak fractions were pooled and the protein concentration was determined by measurement of the optical density using the following equation, (assuming an extinction coefficient of 0.6 for the unfolded AK protein)

Concentration (mg/ml) = (A280) • (Dilution) / 0.6 ml/cm/mg (Equation 3)

The sample was diluted to 0.25mg/ml and slowly refolded by sequential dialysis against refolding buffers containing 4M, 2M, 0.5M, and 0M urea all at 2 liter volumes. The sample was dialyzed in the sequential buffers for four-hours at 4°C and the dialysis tubes were inverted during transfer from one buffer to the next to assure proper mixing. In addition to assure removal of residual urea the dialysis step at 0M urea was performed in duplicate. The dialyzed protein was concentrated to 5ml using a stirred cell Amicon® concentrator with a 10,000 Dalton (NMWCO) membrane at 20psi. The protein was then further concentrated to 2ml using a Centricon® concentrator with a 10,000 Dalton (NMWCO). The concentration of the refolded protein sample was calculated after optical density measurements using the equation described above. As the extinction coefficient changes in going from and unfolded to a folded state an extinction coefficient of 0.76 ml/cm/mg was used for the refolded AK enzyme (Zhou et al., 1998). The protein sample was diluted in water and measured at 280nm in a quartz microcuvette (100µl) against a blank of water (Warburg and Christian, 1941). The refolded protein was further purified by a second round of size exclusion chromatography using a Sephacryl® S-100 HR (HiPrep® 16/60, Amersham) column to separate the improperly folded enzyme from the properly folded protein. For this step the column was equilibrated in S100 buffer and the protein sample was concentrated to between 10-20mg/ml in a total volume of 2ml. The sample was injected into the column using a 2ml loop and separated using a flow rate of 0.5ml/min. Fractions of 0.5ml were collected starting 20 min after the sample was injected. The peak fractions of the properly folded protein, which eluted with a retention volume near 50ml, were pooled and dialyzed overnight against 2 liters of “Regular” buffer (recipe given in materials and methods) to remove any salt prior to the final purification step on a DEAE matrix. In the final step the sample was purified with an anion exchange column, HiPrep® 16/10 DEAE, (Amersham). The S100 sample was diluted to 5ml with “Regular” buffer and loaded onto

17 a DEAE column using a “Super-loop” (5ml injection volume) with a flow rate of 0.5ml/min. The bound protein was eluted with a linear salt gradient (KCl) from 10mM to 100mM. The AK protein eluted at near 60mM KCl and was baseline resolved from all other contaminating protein. The peak fractions of the final purified AK (original template, inclusion body) sample were pooled and dialyzed against a buffer containing 10mM Tris, 10mM KCl, 1mM DTT, 4mM azide, pH=7.5 at 25°C and concentrated to 32 mg/ml and stored at 4°C.

2.2.4 SDS-PAGE (reducing conditions) Proteins were separated according to their molecular weight by SDS-PAGE under reducing conditions (Laemmli, 1970).

2.2.4.1 Polyacrylamide Gels Mini-gels (thickness 0.75mm) were cast using the Mini Protean 3 system (Bio-Rad). The stacking and separating gels were prepared without ammonium persulfate (APS). APS was then added to the separating gel to start polymerization. The gels were cast into the assembled glass plates and overlayed with isopropanol. After 20min of polymerization, the isopropanol was decanted, the polymerized gel was washed with distilled water (dH2O) and APS was added to the stacking gel. The stacking gel was poured into the casting mold, a 15 well comb was inserted into the top of the gel and allowed to polymerize for 20min. Unused gels were wrapped in a wet paper towel, wrapped in plastic wrap and stored at 4°C until needed.

2.2.4.2 Sample Preparation and Electrophoresis

Protein samples were diluted in distilled H2O and 5x sample buffer was added. After 5min of boiling at 95°C, samples were centrifuged at 10,000g for 2min and pipetted into the wells of the polyacrylamide gel. Electrophoresis was carried out at 100V for 45min at room temperature. After the dye front ran off the bottom of the resolving gel the assembly was taken apart and the gel was removed and washed in distilled water to eliminate residual SDS.

2.2.4.3 Coomassie Staining Gels were fixed and stained by soaking in Coomassie staining solution (Neuhoff et al., 1985) at room temperature for 60 min on a rotating shaker. The background Coomassie stain was

18 removed using a destaining solution till the desired contrast of background stain was achieved. At this point the gels were transferred to distilled water and soaked overnight to remove residual methanol and acetic acid. The gels were then dried by dessication in Gel-Wrap® (Bio Systems, NJ) and clamped into a 10″ x 10″ acrylic casting base and left at room temperature for two days (or till dry).

2.2.4.4 Silver Staining Protein samples from each step of the purification procedure were examined by SDS-PAGE and visualized using a silver staining technique. Silver staining (Nesterenko et al., 1994) can detect protein bands corresponding to less than 100ng and was used to detect any impurities in the protein samples. The following solutions were freshly prepared for staining of one gel: 1. Fixing solution: 30ml 50% acetone + 750µl 50% TCA + 12.5µl 37% formaldehyde

2. Dithionite solution: 25mg NA2S2O3 + 100ml dH2O

3. Silver staining solution: 80mg AgNO3 in 30ml dH2O + 300µl 37% formaldehyde 4. Developer: 30ml Bicarbonate solution + 12.5µl thiosulfate solution + 12.5µl 37% formaldehyde During staining the gels were placed on a shaker for incubation times longer than 5min, and solutions were removed by pipetting. Gloves were worn when handling gels to avoid contamination of the gel surface with proteins or oils from hands, which would be detected by the silver stain. The stained gels were scanned to obtain a permanent record.

2.3 Protein Crystallization Biocrystallization is a process involving nucleation and growth, where molecules have to be brought into a supersaturated state. Described in this section is a systematic approach for improving protein crystals by growing them in the metastable zone using the vapor diffusion technique. Screening around known conditions was performed to establish a working schematic saturation diagram for the crystallization of the AK in the TSAC at room temperature. The diagram assisted in the refinement of crystallization conditions by allowing for the definition of the nucleation and growth phases.

2.3.1 Crystal Growth Arginine kinase had previously been crystallized over a narrow window of conditions. Due to the large substrate induced conformational changes in this enzyme, which could have a negative affect on the crystal lattice, co-crystallization of the enzyme-substrate complex was pursued in place of soaking. Co-crystallization with non-cognate ligands had the potential to disturb the requirements for crystallization, and some effort was first made to find a broad range of conditions that would provide more robust crystallization. The first parameter screened was the HEPES concentration. The HEPES concentration was varied from 25mM to 100mM in 25mM increments. The second parameter investigated was pH. The pH was screened from 6.0 to 9.0 in 0.5 unit increments. The third factor optimized was the magnesium concentration which was tested from an initial concentration of 50mM up to 150mM in 10mM increments. This method of optimization proceeded in an iterative manner through four rounds of optimization with each trial being compared with the optimized conditions from the previous test. As a final step in the optimization of room temperature crystallization conditions a schematic saturation diagram was determined. The schematic saturation diagram is a plot of the states of crystallization as a function of the protein versus the precipitation concentration. The progression from the soluble to the supersaturated state is defined by three zones: 1. soluble, 2. metastable, and 3. precipitation. The transition between states are delineated by two lines the nucleation-line and the precipitation-line. The optimum conditions for crystal growth are thought to be in a metastable zone of conditions, which allows crystal growth without the

20 production of further nuclei (Ataka, 1993; McPherson, 1999; Ducruix and Giegé, 1992; Stura and Wilson, 1992). The area of conditions called the “metastable zone” lies between the nucleation and precipitation curves on the saturation diagram. In order to define the saturation diagram the protein concentration was varied from 10mg/ml to 60mg/ml, in 10mg/ml increments and the precipitating reagent polyethylene glycol (PEG) was varied from 10% to 40% in 10% increments. Nucleation is the first step to crystal growth (Feher and Kam, 1985; Chayen, et al., 1997). In order to define the nucleation-line each of the six protein concentrations were setup in a vapor diffusion trial against each of the PEG concentrations. A dog whisker (Shetland Sheep Dog, (Sheltie) “Bailey”) was cleaned with ethanol and used to transfer a micro-nucleate from a preformed crystal to each of the vapor diffusion trials. Each condition was monitored for the presence or absence of a crystal. The protein concentrations where crystals did not form were noted for each precipitation concentration and graphed using an excel spreadsheet. Each data point was repeated 5 times. The precipitation line was determined, to a reasonable approximation, by setting up a linear screen of each protein concentrations versus each PEG concentration and monitored for precipitation. The conditions where no precipitation formed were plotted against the conditions where precipitation appeared using an excel spreadsheet. For quality control the refractive index of each precipitation solutions was monitored. The refractive index is an indication of the amount of solute in a solution. Monitoring this parameter assured that a linear precipitation screen was produced and significantly decreased trial-to-trial variation due to inaccuracies in pipetting or measuring reagents. Once the optimized room temperature conditions for the wild type AK TSAC conditions were defined, it was necessary to reapply these conditions to the crystallization of the non- cognate complexes with L-imino-ethyl-ornithine (ILO), L-ornithine (ORN), L-citrulline (CIT), and D-arginine (DARG) (figure 2.3.1).

Fig. 2.3.1. Chemical structure of arginine and arginine analogues.

Because of the expense of some alternate substrates, the previously described method of crystallization (Zhou et al., 1998) wherein the enzyme was dialyzed against a large volume (50ml) of the substrate mixture was altered. In the modified vapor diffusion method trials were performed by mixing protein stock (30 mg/ml), 5x alternate analogue complex (AAC) buffer, and precipitating reagent in a ratio of 2µl: 0.5µl: 2µl, respectively. The AAC buffer consisted of

5mM MgCl2, 4mM K-ADP, 50mM NaNO3 and 5mM NaN3 at pH= 7.5. Because of the altered binding affinities for the non-cognate ligands each was screened in concentrations from 20mM to 60mM in 10mM increments. To compensate for the dilution of the substrate stocks upon addition to the crystallization trials the AAC buffer solution was made as a 5x concentrate. Initial (de novo) crystals for each complex grew within two weeks. The crystallization conditions for each complex was further optimized by changing the relative amount of protein/ precipitant/ and AAC buffer.

2.3.2 Crystal Freezing and Cryoprotection To minimize radiation damage, all diffraction studies were carried out at 100K. To prevent ice damage cryoprotectant solutions were used. Crystals for each of the alternate complexes were harvested with nylon cryoloops that ranged in size from 0.2mm-1.0mm in size depending on the size of the crystal. Each crystal was transferred to a Pyrex™ depression well containing a cryoprotectant of 20% glycerol + 1X AAC buffer and soaked for 30 seconds. The crystals were then scooped up quickly and moved directly to a dry stream of liquid nitrogen at the goniometer head (Oxford Cryosystems).

2.4 Diffraction Measurements 2.4.1 Data Collection Before diffraction measurements were started, the cryostream was given 1.5hr to cool to 100K and the rotating anode was powered up to 40kV, 90mA, and 1.2kW. The crystals were mounted on the goniometer head and centered so that they were in the middle of the X-ray beam for ϕ settings of 0° and 90°. Initially two diffraction images were taken, one with ϕ=0° and the other with ϕ=90°, with an oscillation of 1.0°, a detector distance of 200mm and an exposure of 2min. Once diffraction quality crystals were identified, data were collected using X-rays from the copper Kα edge (1.5418Å) produced on a rotating anode x-ray generator with a 0.3mm collimator. Data for each of the complexes were collected using a 1.0° oscillation (starting at ϕ=0°) and a crystal to detector distance of 100mm. For each data set exposure times were varied between 5 to 10min per frame, depending on diffraction intensities, and collected using a Mar- CCD detector.

2.5 X-Ray Crystallography Processing 2.5.1 Indexing, Integration and Scaling Indexing, integration, and scaling were done using DENZO and SCALEPACK ® as implemented in the HKL suite (Otowinowski and Minor, 1997). The details are given in the Results section.

2.5.2 Conversion to Structure Factor Amplitudes After data reduction the SCALEPACK output file was converted to the format (.mtz) that is used by the CCP4 programs with SCALEPACK2MTZ (CCP4, 1994). The intensities were converted to structure factor amplitudes using TRUNCATE (CCP4, 1994). This program takes into account that for weak reflections a negative intensity value can be observed due to the statistical distribution of background noise (experimental error). Failure to consider these reflections would bias the data, but structure factor amplitudes cannot be derived from them by calculating the square root of the intensities. As the true intensities of these reflections are positive, TRUNCATE assigns small positive values to their structure factor amplitudes, inferring them according to Bayesian statistics (French and Wilson, 1978). The program also displays a Wilson plot (Wilson, 1949), which is essentially a plot of the intensities of reflections versus the resolution. For datasets of protein crystals, it has a characteristic maximum at 6 Å and then ideally decreases in a straight line towards higher resolution. From the gradient of the line, the overall B-factor (temperature factor), of the dataset can be estimated. At this point the reflections to be used for cross-validation of the model during refinement were separated from the ones to be used for refinement calculations. If a model is validated with the same structure factors to which it was fit, any model can be fit to given data by increasing the parameter to data ratio. This is called over-fitting. In order to prevent over-fitting, a small subset of the reflections, the so-called free set, is excluded from all calculations. As each spot in the diffraction image contains information about all the atoms of the model, a good model should predict the intensities of the reflections in the free set. This is assessed by the free R free factor (R ) (Brünger, 1992), an R factor between Fobs and Fcalc, the observed and calculated structure factors, for the free set of reflections:

(Equation 4)

In order to generate the Rfree set, a list of all observable hkl values for the experiment was generated with UNIQUE (CCP4, 1994). Numbers from 0-9 were generated and randomly

24 distributed over the hkl values as free R flags with FREERFLAG (CCP4, 1994). 10% (approximately; 8,000-10,000) of the reflections (those carrying the free R flag 0) were from then on excluded from calculations and refinement in order to monitor the refinement process. After that, the list containing HKL’s and free R flags were merged with the list containing the structure factor amplitudes with the program CAD (CCP4, 1994).

2.5.3 Molecular Replacement One way to solve the phase problem is to possess an atomic model, from which estimates of the phases can be computed. Such a model can be obtained when you know the structure of a related protein, or even the same protein from a different crystal form. But to build up an atomic model of the new crystal form, it is necessary to work out how the model should be oriented and positioned in the new cell. Molecular replacement is the technique used to solve that problem. Although the structure factor amplitudes were derived from the diffraction data the initial structure factor phases were obtained by molecular replacement. If the content of the unit cell, i.e. the structure, position and orientation of the molecules, are known, structure factor phases and amplitudes can be calculated by Fourier transformation. This is used to obtain initial phases. First, a guess at the structure to be solved is made by using model coordinates of an existing structure, usually a homologous protein. This model, termed the “phasing” model, has to be placed into the unit cell in the same orientation and position as the molecule in the crystal in order to calculate initial phases. If the positioning is correct, the phases should be similar enough to the phases of the target molecule to calculate a first electron density map. The crucial step in the process is the search for the correct placement of the search model within the unit cell. Here, this was done using two separate programs: AMORE (CCP4, 1994; Navaza, 1994) and CNS (Brünger et al., 1998).

In the first method the program AMORE was used to optimize the orientation of the search model to correlate to the diffraction data represented as structure factors (ROTING program part). Then the best results for the rotation search were used as starting values for the translation search (TRAING program part). Instead of calculating coordinates for each rotation and translation and then calculating structure factors, AMORE computes the structure factors directly from the initial model, the rotation matrix, and the translation vector. This allows for a quick search of all configurations. Finally, AMORE performs a rigid body refinement on all

25 solutions for which the translation function had a peak height greater than 50% of the maximum peak height observed. In the second method the program CNS was used. The initial scale files (.sca) from SCALEPACK® were converted to a reflection file (.hkl) using the CNS routine. Using this reflection file a random set of reflections (5%) was selected for use as a cross-validation file [free R] (.cv) using the make_cv.inp input file. Next a molecular topology file (.mtf) was produced using generate.inp. All three files (.pdb, .mtf, and .hkl) were used in a cross-rotation analysis (cross_rotation.inp) with the high-resolution AK TSAC structure (PDB: 1M15) as the probing model. The top five solutions from the rotation function analysis were tested separately in a translation function search (translation.inp). The top five solutions of each translation analysis were rigid body refined and the model with the best statistics was considered the starting model for further refinement. After the initial molecular replacement it was decided to use only the output files from the CNS protocol to proceed.

2.6 Graphical Rendering and Structural Analysis The most important step in all crystallographic investigations is the structural interpretation and accurate graphical representation of the final models. Upon completion of all structural refinement processes each of the non-cognate complexes were investigated using several software programs. The first set of programs used were analytical tools to measure statistical variations in the data, distances, bond angles, and thermal factors. These methods included the following programs:

PROCHECK: Used to generate a Ramachandran Plot of the residue bond angles to assure there are no disallowed angles Swiss-PDB viewer: Used to measure inter-atomic distances COOT: Used to measure bond angles and bond distances O: Used as a visualization tool and to allow for the manipulation of protein structure during model building.

26 The second series of programs were used as tools to render 3-Dimensional molecular models from the atomic coordinates of each structure. These representations allowed for a coherent visualization of larger structures such as: secondary structure, domain arrangement, and the nature of domain motions upon substrate binding. These methods included the following programs: COOT, Web-Lab viewer PRO, MOL-MOL, Raster, DYNO, RasMol, and DynDom. All graphics were rendered, enhanced, and text edited using: POV-Ray, Maya (MAC), PowerPoint, Microsoft-Paint, and PhotoShop.

CHAPTER 3 RESULTS 3.1 Expression and Purification Arginine kinase (Limulus polyphemus) was over-expressed in E. coli BL21 (DE3) pLysS (a gift Dr. Ross Ellington). Overall, this purification protocol produced 20-40mg of pure protein per liter of culture. The purity of the protein sample was qualified at each step by running a representative sample on an SDS-PAGE gel and silver staining it (figure 3.1.1).

Fig. 3.1.1. Silver stained SDS-PAGE of purification steps for arginine kinase. Lanes: (1) Marker; (2) Washed inclusion bodies; (3) After S300 denaturing conditions; (4) After S100 refolded protein; (5) After DEAE final purification.

3.2 Optimization of Crystallization

Because of the extensive amount of preparation time expected for the optimization of crystallization conditions for each of the non-cognate complexes, and the limited ability to work for long periods of time at 4°, it was decided to pursue an optimized room temperature crystallization protocol. The AK TSAC complex was chosen as the trial system. The previously

28 reported crystallization conditions for AK (25mM HEPES and 50mM MgCl2 at pH 7.5, 4˚C; Zhou et al., 1998) produced poor crystals at room temperature (figure 3.2.1).

Fig. 3.2.1. Crystals of AK in 25mM HEPES, 50mM MgCl2, pH=7.5. Grown at room temperature.

Observation with a polarized microscope showed that these de novo crystals had unacceptable qualities (cracks, poor lattice formation, ill defined edges, and low birefringence). These crystals diffracted very weakly and to only low resolution (3.4Å). The first round of optimization consisted of raising the HEPES concentration to 100mM and screening the Mg+2 concentrations from 50mM to 150mM in 10mM increments. This allowed for the optimization of the magnesium concentration. Using a final, optimized concentration of 100mM MgCl2 resulted in the crystals seen in figure 3.2.2. These crystals are much smaller than the crystals in figure 3.2.1 and were never tested for diffraction because of their small size.

Fig. 3.2.2. Crystals of AK in 100mM HEPES, 100mM MgCl2, pH= 7.5. Grown at room temperature.

Fig. 3.2.3. Crystals of AK in 100mM HEPES, 100mM MgCl2, pH=8.0. Grown at room temperature.

The last parameter to be optimized was the pH which was screened from 7.0 to 8.0 in 0.5 pH unit increments. The crystals at pH 7.0 were thin needle clusters of poor quality but increasing the pH to 8.0 gave well-formed 3-D crystals with well defined edges, good refraction of polarized light, and large size (figure 3.2.3). As a final step for describing the optimal conditions for crystallization it was necessary to refine the relative protein concentration versus precipitating reagent (PEG) concentration. In essence this is a phase diagram of the state transitions for the protein ligand complex in going from a solution state to a crystallized state. These data allow for the control of the rate of super- saturation, hence controlling nucleation and crystal growth. Here, a linear screen of the precipitating reagent (PEG 6000) from 0%-35% in 5% increments was made. The precipitation stocks were standardized by measuring the refractive indices for all solutions and plotting it versus its calculated concentration (figure 3.2.4). The refractive index is an indication of the total solute in a given solution and since only the PEG 6000 concentration is being varied it is primarily representative of the total PEG present in each solution.

Refractive Index of a linear screen: PEG 6000 100mM Hepes/ 100mM MgCl (room temp.) 1.39 1.387

1.38 1.38 1.377 1.374 1.371 1.37 1.3692 1.366 1.3622 1.36 1.3601 refractive index 1.358 1.3551 1.35 Refractive Index

1.34 1.339

1.33 0 5 10 15 20 25 30 35 40 % PEG 6000

Fig. 3.2.4. Calibration curve plotting the refractive index of several precipitation solutions in a linear screen. This method is used to monitor the production of a linear precipitation screen.

This method is a quantitative monitor for the production of a linear precipitation screen. This is a convenient and effective way to monitor mechanical errors such as pipetting or weighing of reagents and decreases trial-to-trial variations. A large amount of protein was purified and concentrated to 60mg/ml as determined by UV spectroscopy. The protein sample was then diluted to 50mg/ml, 40mg/ml, 30mg/ml, 20mg/ml, and 10mg/ml. For each protein concentration, crystallization screens were set up at each PEG concentration (0%-35%; 5% increments) in a grid screening method. By observing each vapor diffusion trial and noting where precipitation formed, it was possible to define the precipitation zone of a saturation diagram. To define the nucleation line of the saturation diagram seeding techniques were employed. A linear screen of precipitation reagent and protein was set up and a micro-nucleate was transferred to each condition from a de novo crystal and monitored for crystal growth. The concentration of precipitation reagent at the interface between no crystal growth and crystal growth was considered to be the nucleation point for each protein concentration. Each trial was repeated 5 times and the data points representing the line of precipitation are the average value of all trials. Combining the nucleation and precipitation data allowed for the generation of a saturation diagram (figure 3.2.5) for the crystallization of arginine kinase in the TSAC form at room temperature. This was accomplished per previously defined methods for other proteins (Durbin and Feher., 1996; Saridakis et al., 2003; Asherie., 2004). The final crystallization conditions for AK TSAC were determined to be: 26% PEG 6000, 100mM HEPES, 100mM

MgCl2 at pH 8.0. This solution was mixed in a 1:1 ratio with AK TSAC at 30mg/ml. The final crystals show excellent visual properties under a polarizing microscope and diffract to better than 1.5Å on a rotating anode x-ray generator with 15 minute exposures and a 0.5 degree sweep- width. These optimized room temperature conditions were used for the crystallization of each of the alternate substrate complexes.

32 Saturation Diagram PEG 6000 100mM Hepes/100mM MgCl room temp. Arginine kinase TSAC (ori/inclusion body)

Protein Concentration (mg/ml) 10

0 10 16 20 27 32 % PEG 6000 w /v

Fig. 3.2.5. A saturation diagram showing the solubility of AK as a function of the concentration of precipitant present. Here three zones are defined: (1) soluble zone ; (2) metastable zone ; (3) precipitation zone

Because of the expensive nature of some of the alternate substrates the previously defined methods of crystallization (Zhou et al., 1998) where the enzyme was dialyzed against the substrate mixture had to be altered. In this modified method crystals were obtained by the hanging drop vapor diffusion method by mixing protein stock (30 mg/ml), 5x alternate analogue complex buffer (AAC), and precipitating reagent in a ratio of 2µl: 0.5µl: 2µl, respectively. The concentration of the non-cognate ligand was optimized by screening from 10mM to 50mM and keeping all other transition state components the same concentration as the WT TSAC. To compensate for the dilution of the substrate stocks upon addition to the crystallization trials the AAC buffer solution was made as a 5x concentrate and added appropriately. Diffraction quality crystals for each of the alternate complexes were obtained by micro-seeding denovo crystals into a fresh drop of protein, AAC buffer, and precipitating reagent. Each alternate substrate complex produced multiple crystal forms, very similar to the crystal forms of the wild type TSAC (figure 3.2.3), with average dimensions of 0.6mm x 0.4mm x 0.3mm. All crystal forms tested gave the

33 same space-group and unit cell parameters but some crystal forms were noted to give significantly higher resolution diffraction and better overall data statistics.

3.3 Data Collection The data set for each of the alternate analogue structures were collected from a single crystal at cryogenic temperature (100K). The diffraction images were displayed with AUTOMAR (mar research) for visual analysis (figure 3.3.1). The first step in extracting the data from the diffraction images is to assign the diffraction spots to their point in the reciprocal lattice,

“indexing”. Coordinates of selected spots in one image, displayed by XDISP (HKL, Otwinowski and Minor, 1997), were entered into DENZO by using the peak search routine. DENZO then calculated for each possible lattice type the orientation of the crystal in the beam and the unit cell dimensions best fit to generate the diffraction image. The lattice type of the crystal, determined as the one with most restraints that still fit to the dataset reasonably well. After indexing, DENZO had all the information necessary to predict the position of all spots on the diffraction images. The intensity of these spots was then determined by integration. A square box was defined that contained a spot in the middle surrounded by a guard region and the background.

Fig. 3.3.1. A diffraction pattern from an arginine kinase (TSAC) crystal grown using the optimized room temperature crystallization conditions. This image was taken with a 10 minute exposure on a rotating anode generator and the detector edge is at 1.7Å.

This box was overlaid on each reflection; the intensity of the spot pixels was integrated and corrected for noise contributions derived from the background level. The guard region, with contributions from both signal and noise, was excluded from the calculations. The mosaicities of each crystal were moderately-high so a large portion of the reflections recorded were partial. For each diffraction image, integration yielded a list of reflections, defined by their h, k, and l values, and their intensities. Those lists still contained partial reflections. In order to generate one list of unique reflections, data were merged by estimating all partials, dividing by the partiality, and scaling with SCALEPACK to correct empirically for fluctuations in beam intensity and structure of the crystal. The completeness and the accuracy of the datasets were assessed during scaling.

35 Table 3.3.1: Crystallographic data for molecular replacement and final models.

Data Processing ILO ORN CIT DARG

Space group P212121 P212121 P212121 P212121 Unit cell a=62.63 a=64.89 a=65.1 a= 64.10 b=68.11 b=70.15 b=70.7 b=65.41 parameters Å c=91.88 c=79.64 c=79.8 c=85.90 Resolution Å 50 - 2.45 50 - 2.0 50 - 2.0 50 - 2.8 (outer shell) (2.5-2.45) (2.5-2.0) (2.5-2.0) (3.5-2.8) No. of observations 966,345 1,309,639 1,564,521 950,345 No. of unique 170,175 201,208 322,133 122,278 reflections % Completeness 100% (94.4) 99.8% (96.2) 99.7% (97.8) 98.3% (92.1)

Rsym (outer shell) 5.6 (21.4) 7.8 (30.2) 8.4 (27.5) 4.9 (16.1) I/σ 16.6 (9.3) 19.3 (6.4) 21.1 (2.2) 14.3 (6.8) Redundancy 4 6 8 4 Structural ILO ORN CIT DARG Refinement Rwork / Rfree 21.2 / 23.1 19.4 / 22.1 21.7 / 23.5 24.3 / 25.2 R.m.s deviations from ideal values Bond length (Å) 0.014 0.008 0.011 0.016 Bond angles (°) 1.6 1.4 1.4 1.7 Agreement of (Φ,Ψ) with Ramachandran plot Favorable region 90.7% (0%) 92.3% (0)% 91.2% (0%) 88% (0%) (Disallowed) Mean B value (Å2) Phosphagen ILO : 23.1 ORN: 19.7 CIT: 22.8 DARG: 28.9 mimic Protein (2,848atoms) (2,816atoms) (2,832atoms) (2,821atoms) 15.6 14.7 15.3 20.3 Main-chain atoms 14.2 13.3 13.5 18.9 Side-chain 17.4 15.9 17.2 23.2 Water molecules 32.7 (309) 28.9 (245) 33.5 (291) 35.7 (215)

The data sets showed normal statistical profiles in that the completeness decreased from low to high resolution whereas the R factors, indicating inaccuracy, increased. The results for data processing are listed in table 3.3.1. All complexes induce the substrate bound form of the

36 enzyme and are nearly isomorphous with the wild type AK TSAC, type 2 (Zhou et al., 1998). In all cases phasing was solved by molecular replacement using the high-resolution AK TSAC structure as a model (PDB ID: 1M15). Cross-rotation and translation searches using CNS (Read et al., 1998) yielded a unique solution for each structure. The solution that provided the best initial refinement statistics was used to calculate a Fo-Fc and 2Fo-Fc simulated annealing, σ-A weighted, composite omit-map for use in manual refinement with the program “O” (Jones et al., 1991).

3.4 Molecular Replacement and Refinement

The structure factor phases necessary to calculate the electron density map for each structure were determined by molecular replacement (Rossmann, 1972). The structures of all the alternate complexes were refined using the same series of steps. The steps used to solve the ILO structure are outlined below. The high-resolution AK TSAC structure (PDB: 1M15) was used as a search model and positioned into the crystal unit cell using the rotation and translation function searches of CNS (Brünger, 1992). The diffraction data were used in the range of 10-2.0Å. Good correlation of the rotation and translation of the search molecule with the diffraction data was indicated by peaks in the rotation and translation functions. Use of the high-resolution AK TSAC as a phasing model brought each structure within the convergence radius of a good model (Rfree < 34%). A reference set of 5% of the reflections were set aside for calculation of the Rfree (Brünger, 1992). A working set of the remaining 95% of the reflections was used throughout the automated CNS refinement. Rigid body refinement with CNS improved Rfree from 34 to 32%, simulated annealing dropping Rfree from 32 to 31%, and thermal factor refinement dropped Rfree to 27%. After several rounds of refinement the electron density was checked for definitive substrate density and if density was present the substrates were modeled in order of ADP/ Mg+2/ Nitrate and phosphagen analogue.

Fig. 3.4.1. Progression of model refinement for the ILO model. Rfree calculations are based on 2σ cut-offs for CNS.

The electron density for each structure showed clear density for all residues as well as each ligand, except for the nitrate in the CIT complex, which is not retained. After the substrates were modeled automated water picking was performed and the positions and thermal factors were refined. This dropped the Rfree from 27.3% to 24.2%. Several rounds of manual rebuilding into

2Fo-Fc and Fo-Fc maps using the program O alternated with automatic refinement resulted in a final Rfree of 23.1% (R= 21.2%). The final model statistics for each of the structures are listed in table 3.3.1 and a graphical presentation of the progression through refinement is presented in figure 3.4.1. Analysis of the Ramachandran plots (Ramakrishnan and Ramachandran, 1965) for all structures showed that the conformational angles of 356 residues are in the most favored regions, two residues (Asp126, Thr273) in the generously allowed region, and none in the disallowed region, as defined by PROCHECK (Laskowski et al., 1993), figure 3.4.2.

Fig. 3.4.2 Ramachandran Plot for the final ILO model showing two residues in the generously allowed region (Asp161, Thr273) and none in the disallowed region. Image generated with PROCHECK.

3.5 Induced Fit Many proteins have been analyzed for substrate induced structural changes (Gerstein et al. 1994). An analysis of arginine kinase showed that this enzymes undergoes more complicated conformational changes than had been previously suggested (Yousef et al., 2002) with three domains moving relative to one fixed domain. These domains are not linear segments of residues but are rather non-continuous segments of residues that can be spatially distant form one another. An analysis of the overall induced fit domain movements of AK bound to the alternate complexes was performed using the program Dynamic Domain (DynDom) (Collaborative Computational Project 1994; Hayward and Berendsen, 1998). The DynDom program defines the fixed domain and the moving domains with an assessment of the movements and the rigid displacement of a domain. Hence, displacement vectors can be replaced by a rotation axis whose length is equal to the rotation amplitude. In this measure of domain movement displacement vectors are calculated for short backbone segments by a least squares analysis of the two conformations (in most cases the substrate-free and substrate bound forms). From this data rotation axes are generated and a clustering algorithm detects residues with spatially close axis of equal length and defines them as domains. Dynamic domain analysis of open and closed arginine kinase shows that three domains move relative to one fixed domain about specific rotational axes (Yousef et al., 2002). These axes designate movement with the color of the arrowhead representing the domain that is moving and the color of the shaft designating the domain toward which movement occurs. These non- intuitive movements are much more complex than the standard view of four linear domains moving in a mutual fashion to close the binding pocket and embrace the substrates.

40 Table 3.5.1. Residue content of each dynamic domain for each structure.

Fixed Domain Domain #1 Wild Type ILO ORN CIT DAR Wild Type ILO ORN CIT DAR 98-111 96-99 100-112 96-100 96-100 7-97 7-95 7-95 7-95 7-95 124-127 126-149 125-126 126-148 125-151 129-136 268-273 268-271 268-271 268-271 156-163 168-178 156-162 161-163 156-163 147-155 220-222 184-229 231-254 168-235 168-267 164-168 227-261 256-267 277-285 237-252 272-279 223-226 276-285 274-276 329-352 255-267 331-335 262-275 329-352 272-278 341-347

Domain #2 Domain #3 Wild Type ILO ORN CIT DAR Wild Type ILO ORN CIT DAR 169-197 100-112 96-99 N.A. N.A. 112-123 112-123 112-123 101 112-124 205-219 125 127-149 286-328 286-328 286-328 103-105 280-330 156-163 163-164 110-125 230-255 168-174 236 277-285 185-215 279-352 329-352 217-230 255-267 272-276

DynDom analysis shows that residues in non-sequential areas of the structure move in a concerted fashion to achieve the substrate bound configuration. The residue content of each domain in each of the analogue complexes is listed in table 3.5.1. This table defines the nonlinear segments that make up each dynamic domain. These data show domain one making a hinge rotation of 17.5º relative to the fixed domain. This domain carries many important elements such as the phosphagen “specificity” loop (residues 61-68), Cys 271, and a highly conserved NEEDH segment (residues 223-227).

Fig. 3.5.1. Ribbon diagram representing the dynamic domains for each non-cognate complex. The fixed domains are colored: fixed domain=yellow, domain 1=blue, domain 2=orange, domain 3=purple. Domain movements are represented by a rotational axis with the color of the head of the arrow designating which doamin is moving and the shaft representing the domain to which the movement occurs. Image generated with DynDom.

42 Domain two shows a hinge rotation of 17.6º about an axis near orthogonal to the domain one axis. This domain carries residues His185, Ser122, and His284 all of which interact with the ADP substrate and are conserved in phosphogen kinases (Pineda & Ellington 1999). Domain three undergoes a hinge rotation of 13.5º relative to the fixed domain. This domain carries the putative catalytic residue Glu314. Each of the dynamic domains defined by DynDom analysis showed movements relative to the fixed domain. However, there is one non-intuitive movement of domain one relative to domain two (not the fixed domain). In this motion domain one rotates by 13.5º along an axis near orthogonal to all other rotational axes. This movement is unique to AK (Suzuki et al. 1997) and may help to stabilize the conformation of the “specificity” loop. In the structural analysis presented here many of the residues that are assigned to each moving domain remain near the same as in the wild type model, represented in schematic form in figure 3.5.1. There are however differences seen in the angle and magnitude of the rotational axes and the designation of specific nonlinear residue segments from one domain to another. The structures of ILO and ORN show four distinct dynamic domains, much like that described for wild type (TSAC), but CIT and DARG show only three dynamic domains with domain two being designated as part of the fixed domain. The ILO structure shows four dynamic domains (figure 3.5.1a), much like the wild type, but differs in residue content and placement of the rotational axes. Domain one (blue) in ILO is quite different from the designation in wild type and shows a root mean square deviation of 0.21Å when compared to the wild type structure. Domain one exhibits a rotation of 13.1º relative to the fixed domain (yellow) whereas in the wild type structure it rotates 17.6º. The angle between the rotational axis and a line joining the centers of mass of domain one and the fixed domain is 75.2º in ILO and 70.7º for wild type. Domain two (orange) shows an R.M.S.D. of 0.17Å and rotates 14º whereas it only rotates 13.1º in wild type. The angle between the rotational axis and a line joining the centers of mass of domain two and the fixed domain is 80.7º in ILO and 53º for wild type. Domain three (purple) in ILO contains the same residues as in wild type with an R.M.S.D. of 0.53Å and rotates 13.2º relative to the fixed domain whereas it only rotates 17.7º in wild type. The angle between the rotational axis and a line joining the centers of mass of domain three and the fixed domain is 67.9º in ILO and 90.9º for wild type. Although DynDom fails to detect a rotational vector for the movement of domain one relative to the fixed domain, it does detect an extra non-typical displacement of domain three relative to domain two. Here, domain three rotates 11.9º relative to domain two

43 and the angle between the rotational axis and a line joining the centers of mass of domain three and domain two is 55.3º. The ORN structure shows four dynamic domains (figure 3.5.1b) much like ILO and wild type and differs only slightly in residue content from the ILO structure. Domain one (blue) in ORN shows a root mean square deviation of 0.28Å when compared to the wild type structure and rotates 13.8º relative to the fixed domain. The angle between the rotational axis and a line joining the centers of mass of domain one and the fixed domain is 84.2º in ORN. Domain two (orange) in ORN has an R.M.S.D. of 0.29Å and rotates 13.3º. The angle between the rotational axis and a line joining the centers of mass of domain two and the fixed domain is 55.6º in ORN. Domain three (purple) in ORN contains the same residues as in wild type with an R.M.S.D. of 0.60Å and rotates 12.38º relative to the fixed domain. The angle between the rotational axis and a line joining the centers of mass of domain three and the fixed domain is 76.3º in ORN. Like the ILO structure DynDom fails to detect a rotational vector for the movement of domain one relative to the fixed domain. Although CIT retains much of the same molecular character as arginine the enzyme shows quite different dynamic domain movements in DynDom analysis (figure 3.5.1d). Domain one (blue) in CIT has a root mean square deviation between all 131 atoms of 0.22Å. But this domain rotates only 15º relative to the fixed domain and the angle between the rotational axis and a line joining the centers of mass of domain one and the fixed domain is 77.8º in CIT. Domain three (purple) has an R.M.S.D. of 0.61Å and rotates 18.9º. The angle between the rotational axis and a line joining the centers of mass of domain three and the fixed domain is 78.0º in CIT. Due to a displacement vector ratio of greater than 1:1 for intra-domain vs. inter- domain movements the clustering algorithm of DynDom fails to detect domain two. As a consequence, two rotational vectors between, domain two and the fixed domain and between domain one and domain two are absent in the citrulline system. Despite this absence of rotational axes and dissimilar magnitude of rotation about the remaining axes there is still a very similar coordination of the dynamic domains. Other than the obvious chiral differences the structure of DARG is a chemical mimic of L-arginine. Even so the enzyme shows dynamic domain movements that are more like CIT than wild type. Domain one (blue) in DARG (figure 3.5.1e) has a root mean square deviation of 0.33Å and rotates 15.7º relative to the fixed domain. The angle between the rotational axis and a

44 line joining the centers of mass of domain one and the fixed domain is 77.3º. Domain three (purple) has an R.M.S.D. of 0.68Å and rotates 22.6º relative to the fixed domain. The angle between the rotational axis and a line joining the centers of mass of domain three and the fixed domain is 63.0º. As in the CIT analysis DynDom fails to detect domain two in DARG. As a consequence two rotational vectors (domain two vs. fixed domain, domain one vs. domain two) seen in the wild type system are absent in the DARG system.

3.6 Active site Coordination Although the chemical composition of the non-cognate substrates is very similar to the native L- arginine substrate they impact the position of active site residues and the coordination of

+2 - substrates (ADP, Mg and NO3 ). The cross-validated Luzzati Error for each structure is: ILO= 0.15, ORN=0.19, CIT=0.27, and DARG=0.24. The ILO molecule is very similar to arginine but has a carbon atom replacing a terminal nitrogen. The ILO ligand is aligned in a nearly identical position as the cognate arginine substrate but the guanidine-like structure is displaced by 0.55Å (figure 3.6.1a). In this model there is no significant deviation in many of the active site residues (Tyr68, Glu225, Cys271, Thr273). However, there is a displacement in the side chain atoms of Glu314 which are displaced by 0.61Å and the imidizole ring of His315 which is displaced 0.92Å. The displacement of the His315 side chain is indicated by a torsion about the Cα/Cβ bond changing from –48.2º in the wild type TSAC to –58.3º in the ILO structure. The ADP and nitrate ligands retain their native coordination but the Mg+2 ion shifts nearly 0.40Å away from the ADP O1α and O1β atoms. The ORN molecule is the shortest of all the non-cognate substrates in the test group. It terminates at Nε and does not possess a guanidine moiety. The α-amine is in nearly the same position as in the wild type complex but the carboxylate group is displaced by 0.61Å (figure 3.6.1b.). The Nε atom of ORN is positioned in nearly the same position as the non-reactive nitrogen (i.e. closer to Cys271 than Glu314) of the cognate arginine guanidinium. In this model there is no significant deviation in the side chain position of residues Glu225 and Cys271. However, the side chains of residues Tyr68, Thr273, Glu314 and His315 have moved 0.35Å, 0.51Å, 0.30Å and 1.6Å. The torsion about the Cα/Cβ in His315 is –64.2º. In this structure the Mg+2 has moved 0.35Å away from the ADP O1α and O1β atoms. The distance between the

45 +2 - Mg and the NO3 has increased 2.0Å and allows for the coordination of a water molecule between the magnesium and the nitrate. The bonding distances are 1.9Å between the Mg+2 and the water, and 2.2Å between the water and nitrate. The dislocation of the nitrate away from O3β of ADP imparts a small shift of 0.30Å in the O2α position. This in-turn imparts strain in the flexible loop that carries Glu314 to maintain the water-mediated bond between the phosphate O2α and the backbone nitrogen of Glu314 and may be responsible for the displacement of the His315 side chain. Citrulline retains many of the chemical moieties of arginine but has an oxygen replacing a terminal nitrogen of the guanidinium. A rotation in the substrate about the Cδ/ Nε bond from +74.3º in arginine to –116.1º in CIT “flips” the guanidine-like configuration of CIT out of the native binding site and proximal to the side chain of Glu314 making new bonds with the backbone of Glu314. The entire substrate is translated into the binding pocket by nearly 0.7Å but retains near native bonding interactions with the specificity loop (residues 62-68). In spite of major deviations in substrate conformation and position there are relatively small changes in the positions of residues Tyr68, Glu225, Cys271, and Thr273 (figure 3.6.1c.). However residue Glu314 has been displaced 0.55Å and there is a torsion about the Cα/Cβ in His315 of -59.0º. This torsion has induced a 0.73Å shift in the ring of His315. The most significant difference is

- the lack of a NO3 molecule and the positioning of O3β on ADP, which has moved 0.73Å. The D-arginine structure displays the most prolific changes in the coordination of active site residues of all models (figure 3.6.1d). The reactive nitrogen of the guanidinium is displaced nearly 1.0Å and the carboxylate is rotated near 90º about the Cα/Cβ bond and translated by 1.6Å to a parallel plane with the ring of Tyr68. In the cognate arginine complex the substrate carboxylate makes two bonds with the backbone of Gly64 (2.9Å) and Val65 (2.8Å). In the DARG complex the carboxylate forms two distinctly new bonding interactions with the backbone of Gly66 and through a water molecule with Cys271. There is a disruption in the coordination of nearly all active residues and this structure is the most changed complex with an R.M.S.D of 0.34Å. The positions of Thr273, Glu225 and Cys271 have moved 0.37Å, 0.44Å and 0.63Å respectively. The backbone Cα’s of Glu314 and His315 have been moved 0.53Å. This imparts a shift of 0.58Å in the position of the Glu314 side-chain and a 0.35Å shift in the position of the His315 ring. The movement of His315 is not the same torsion like movement seen in

Fig. 3.6.1. Overlay of wild type and analogue structures. Models were aligned by least squares fit of the fixed domain residues. Only critical active site residues and substrates are presented. Image generated with Web-lab Viewer Pro.

ILO, ORN, and CIT but rather a pure translation away from the substrate binding pocket. Residues Glu225, Cys271 and Thr273 have moved approximately 0.4Å, 0.6Å and 0.4Å, respectively. In this structure there is no significant displacement of the nitrate ligand. The Mg+2 ion is displaced 1.0Å with an increase in the distance to O1α and O1β of ADP by almost 0.3Å and 0.9Å, respectively. Dislocation of the Mg+2 may be responsible for a shift in the position of the ADP O2α by nearly 0.4Å and the loss of a water ligand between the Mg+2 and the side chain of Glu225. The distances between active site residue side chains can be used as a method to distinguish a change in the active site volume. In ILO the distance between the two delta carbons of Glu225 and Glu314 has increased 0.43Å (figure 3.6.2a) and the distance from

47 - Tyr68 to His315 has increased 0.33Å. The distance from NO3 to Glu314 has increased 0.31Å +2 - and the distance from Mg to the NO3 has increased 0.35Å. In the ORN structure the inter- residue distances from Tyr68 to Glu314 and Glu225 to Glu314 shorten by 0.45Å and 0.21Å - respectively (figure 3.6.2b). The distance from the Cδ of Glu314 to the nitrogen of NO3 - increases 0.5Å, while the distance from Cδ of Glu225 to NO3 increases 0.2Å. This movement is parallel to the plane of the two side chains for Glu225 and Glu314. The distances from both Glu225 and Glu314 to the Mg+2 have not been drastically changed. In the CIT structure there is a reduction in the distance from residue Glu314 to Tyr68 by 0.4Å, effectively “squeezing” the binding pocket together. Although the nitrate is not retained there is very little deviation in the rest of the critical active site distances. In the DARG structure the overall inter side-chain distance between several active site residues has increased. An increase in distance is seen between Glu225 / Glu314 = 0.37Å, Tyr68 / Cys271= 0.52Å, Glu225 / Mg+2=0.35Å, Glu314 / - +2 - NO3 =0.43Å, and Mg / NO3 = 0.81Å.

Fig. 3.6.2. A schematic representation of the interactive site distances noted in non- cognate AK complex. Image generated with Web-lab Viewer Pro.

CHAPTER 4 DISCUSSION 4.1 Protein Purification Arginine kinase was expressed in E coli and purified using gel filtration and ion exchange chromatography. Although a significant portion of the expressed protein was sequestered in inclusion bodies, unfolding and refolding of the protein resulted in a soluble sample that retained its native binding ability and overall fold. This protein was soluble enough to use at high concentration and behaved well in crystallization trials. The protein bound all arginine analogues in a mimic of the TSA complex.

4.2 Crystallization Crystallization trials were performed to optimize a room temperature crystallization condition that would allow for the formation of high quality crystals in a very short time. Extensive screening near previously reported crystallization conditions culminated in the formulation of an optimum crystallization buffer that could be used for the high-through put production of diffraction quality crystals of AK TSAC at room temperature. These conditions were extended to crystallize all four of the non-cognate complexes in transition state analog form with the Mg- nucleotide, phosphagen mimic, and a nitrate molecule. The ILO, ORN, and CIT crystals were grown using a final ligand concentration of 20mM, but the DARG ligand had to be increased to 50mM before quality crystals could be obtained. This would suggest that DARG has a lower binding affinity than the other ligands. All TSAC ligands are bound to the enzyme with the exception of the nitrate in the CIT structure, which is not retained. All complexes displayed several different crystal morphologies but each had excellent canonical properties (defined edges, good birefringence, and large size). Diffraction quality crystals for each of the alternate complexes were obtained by micro-seeding de novo crystals from the initial vapor diffusion trials into a fresh drop of protein, substrate buffer, and precipitation reagent. These “second generation” crystals gave many different crystal forms. Some forms diffracted better than others but all gave the same space-group with only minor deviations in unit cell parameters.

4.3 Induced Fit and Dynamic Domains Many proteins have been analyzed for substrate induced structural changes (Gerstein et al. 1994). In comparison, AK undergoes more complicated conformational changes than has been seen in many of these other systems. Dynamic Domain (DynDom) (Collaborative Computational Project 1994; Hayward and Berendsen, 1998) analysis of the substrate-free and substrate-bound states of AK show that three domains move relative to one fixed domain about specific rotational axes (Yousef et al., 2002). The use of arginine homologues has allowed for the identification of integral domain movements and lent some insight into the relevant function of specific active site moieties. Single atom substitutions in the chemical structure of the phosphagen impact complex movements of various “rigid” domains and even influence the arrangement of specific intra- domain secondary structures. A constant theme in all domain analysis is the presence of three fundamental groups the fixed domain, dynamic domain #1 and dynamic domain #3. These three “primary” domains carry all the necessary residues (Cys271, Glu225, “specificity loop”, and Glu314) to construct an integral but flawed active site. The noted change in the angle and magnitude of the axis of rotation suggest that there is an altered movement in the dynamic domains but the ligand bound structures of the non-cognate complexes are very similar to the wild type TSAC. Deviations in the induced fit movements are seen most clearly by the absence of a rotational axis between domain #1 and the fixed domain in ILO and ORN. The absence of this rotational movement is compensated for by a shift in the angle of the rotational axis between domain #1 and domain #2. In the ILO and ORN structures the alpha helix that consists of residues Lys175 and Asp184, is not included as part of dynamic domain #2 but two loops: 1. Asn223 – His227, 2. Cys139 – Pro135, are included. This suggests that the elemental moieties of the guanidinium affect the relative displacement of these secondary structures. In the CIT and DAR models the rotational axis between domain one and the fixed domain is present but domain #2 has become part of the fixed domain. The axis of rotation for domain #1 relative to the fixed domain is more like the axis seen for the rotation of domain #1 relative to domain #2 in the other structures including wild type TSAC. Also, in CIT and DAR

51 there is an increase in the rotational magnitude of domain #3 (CIT=6º and DAR=10º). This is most likely a result of compensating for the non-ideal movement of domain #1. Of further interest is the lack of domain #2 in the CIT and DARG structures. These ligands bind to the active site in a very different configuration than ARG, ILO and ORN. In the CIT model the guanidine-like structure “flips” out of the binding pocket and the DARG ligand adopts a significantly different configuration at the amino acid end of the molecule. This lack of native induced fit movements may effect catalysis in two ways: 1. The CIT and DARG models undergo relatively simple induced fit movements which may decrease the overall enthalpic contribution to the activation energy 2. The lack of induced fit movement may compromise the active site configuration and reduce the ability of the enzyme to restrain the substrate increasing the entropic contribution to catalysis. This is indicated by the fact that the DARG ligand has a nearly 3x higher B-factor than the cognate ARG substrate. The Root Mean Square Difference (RMSD) between the Cα backbone of the wild type TSAC complex and the alternate substrate complexes is: ILO=0.25Å, CIT= 0.31Å, ORN=0.27Å, and DARG=0.34Å. Taken together these data suggest that the final position of secondary structural elements is very similar for all structures (figure 4.3.1) but the domain movements that are used to achieve this configuration have different axes and magnitude of rotation. The only notable difference between the ligands of the analogue complexes and the wild type complex is the presence of the phosphagen homologue. Thus the variations in domain designations are likely a result of altered axes and magnitudes of rotation in the dynamic domains and indicate that the substrate induced conformational changes can be altered to compensate for non-ideality within active site moieties.

Fig. 4.3.1. Ribbon diagram showing a structural comparison of the wild type AK enzyme with all non-cognate structures. All structures are least square fit according to the fixed domain, as defined by dynamic domain analysis. Each structure is represented by a different color: WT=yellow, ILO=purple, CIT=green, ORN= light blue, DARG= orange. Image generated with Web-lab Viewer Pro.

4.4 Enzyme Stabilization

Although structural flexibility is a key component in enzyme catalysis it is sometimes necessary to restrict the movement of reactive residues that are located on highly flexible loops. In AK there are four loops that carry many of the residues believed to mediate catalytic processes (Tyr68, Glu225, Cys271, and Glu314). Each of these loops undergoes substantial movements in the transition from the substrate-free to the ligand bound state. An inspection of the AK TSAC structure shows that these flexible regions are flanked by alpha-helices, were many of the stabilizing bonding interactions are located. The “specificity” loop (residues 62-68) is stabilized by multiple bonding interactions. The first bond is between Asn60 and Thr197 (located on a transient alpha-helix) and the bonding distance changes only 0.3Å in going from the substrate- free to the substrate bound form. The other side of this flexible loop is stabilized by a bonding 53 interaction between the backbone nitrogen of Ala69 and the backbone oxygen of Phe270. This bonding distance is exquisitely retained at 2.8Å ± 0.2Å in the transition from the open to bound state. This salt bridge also maintains the overall positional relationship between Tyr68 and Cys271. Also helping to stabilize the active site are two hydrogen bonds found between Leu27 and Val65 (2.8Å) and Leu26 and Gly64 (2.9Å). These residues are located on two separate alpha-helices both of which are designated as part of dynamic domain one. The distance between Leu27 and Val65 is very well maintained in the substrate-free, TSAC, ILO, ORN and CIT structures (2.9Å ± 0.2). The distance from Leu26 to Gly64 is well maintained at 3.0Å in most complexes but increases to more than 3.6Å in the DARG model. The DARG structure is likely due to a rearrangement of the bonding restraints between the carboxylate and both residues Gly64 and Val65, which shifts the backbone position of Gly64 and increases the distance between Leu26 and Gly64. The rotational movement of domain one relative to domain two places residues Asp62 and Arg193 in close proximity and allows for the formation of a bond. The distance between these two residues in the substrate-free form is nearly 6.6Å and closes to 2.8Å in the substrate bound structure. This distance shows only minor fluctuations in most non-cognate structures. This is the only interaction, near the specificity loop, that is a direct consequence of the rotation of domain one relative to domain two and is conserved and unique among AK’s. Investigation of the substrate-free and high-resolution TSAC structures of AK suggests that coordination of a second flexible loop (309-320) most likely results from “flipping” of residue Glu314 into the binding pocket (Yousef et al., 2003). The distance between these two residues decreases from nearly 30Å in the substrate-free enzyme to 2.8Å in the TSAC structure. Examination of the non-cognate complexes shows that coordination of this loop is produced by a bonding interaction between Asp183 (domain 2) and Arg312 (domain 3) at a distance of 2.81Å. This bonding restraint serves to stabilize two separate flexible regions (175-204, 309-320) and contributes to the overall stability of the enzyme. In the formation of this bonding interaction two alpha helices (a surface exposed and a transient alpha helix) are pulled closer to the active site where both helix dipoles could influence the reactive complex. Notably it is likely that only three bonding interactions (Glu225, α-amine and carboxylate) would be sufficient to stabilize the overall configuration of the phosphagen substrate. This observation is supported by the ORN structure, which has only these three bonding restraints and

54 makes no bonding interaction with Glu314, and the ILO structure which has a significantly weakened interaction with Glu314. This suggests that the Glu314 bonding restraint may be superfluous for phosphagen binding but likely plays a role in substrate alignment. Furthermore, coordination of Glu314 most likely occurs subsequent to phosphagen binding, consistent with its positional dependence on the presence or absence of the guanidinium in the ILO, ORN, and DARG structures.

4.5 Active Site Coordination 4.5.1 Active site pre-organization Previous investigations of AK have suggested that the Mg-nucleotide complex binds to the fixed domain, induces large conformational changes, and is responsible for mediating long-range bonding interactions that serve to stabilize the active site (Forstner et al., Yousef et al., 2003). The investigation of the high-resolution structure shows two water molecules bound near the active site and mediating bonding interactions between the Mg+2 and Glu225 (water #1) and between Glu225 and Tyr68 (water #2) (figure 4.5.1.1a). The investigation of the substrate-free and TSAC bound structure of AK suggest that water #1 may play a role in the reactive mechanism (Yousef et al., 2003). An analysis of the analogue complexes shows that disruption of phosphagen coordination has a direct impact on the coordination of these water molecules. The demonstrated displacement of these water molecules suggest that they may play a role in active site processes. Here we propose that these water molecules are responsible for the stabilization of the active site configuration. The notion of a stable conformation for residues Tyr68 and Glu225 is supported by both the CIT and DARG structures. In the CIT structure there is a loss of the nitrate and a total displacement of the guanidino group. Despite the loss of this multitude of bonding restraints the coordination of Glu225 and the position of these two water molecules are not significantly affected. In the DARG structure the bonding restraints between the α-amine and Tyr68 are disrupted with no impact on the coordination of Tyr68. However, in the DARG complex water #1 is displaced from the active site and the bonds of water #2 are increased significantly (figure 4.5.1.1e). These data suggest that coordination of residue Try68 and Glu225 are not dependent on local bonding restraints with the phosphagen ligand but may be mediated by the water-

55 bonding network. These data suggest that a pre-organized active site “scaffold” (consisting of residues Tyr68 and Glu225) could be stabilized by local water-bonding interactions. Although these bonding interactions may help to stabilize the relative position of Tyr68 and Glu225 they are not absolutely rigid constraints. The inherent plasticity of these bonding restraints allows for subtle fluctuations in the overall distance from Tyr68 to Glu225. This distance harmonizes to the distance between the guanidinium and the α-amine of the cognate phosphagen substrate. This would allow for substrate discrimination by phosphagen length and through bonding interactions with both the guanidinium and amino acid moieties. Mediation of bonding restraints at both the nucleotide and phosphagen binding sites through an inter-residue water-bonding network (Tyr68 to Mg+2) may explain the 5 fold reduction in binding of both phospho-arginine and ADP in the Y68S mutant (Azzi et al., 2004). Disruption of this water-bonding network could impact the bonding restraints at both sites. The implication of these water-bonding restraints may also explain the retention of activity in this mutant as the water molecules may compensate for deviations in local binding restraints lost by the mutation.

Fig. 4.5.1.1. Graphical representation of the water-bonding network in the active site of each alternate analogue complex: a= WT, b=ILO, c=ORN, d=CIT, e=DARG. Image generated with Web-lab Viewer Pro.

Fig. 4.5.1.1. Continued.

4.5.2 Overlay of wild type and analogue complexes Non-cognate substrates have been used to probe the catalytic parameters of many enzymes. A structural investigation of the impact of inhibitory substrates that bind weaker than the cognate ligand can provide insight into the importance of specific bonding restraints and their contribution to catalysis. The investigation of the analogue complexes presented here show that residues Tyr68, Glu225, and Cys271 retain their relative coordination very well. However, Glu314, His315, the nitrate, and the magnesium are all very sensitive to deviations in the substrate structure. This may suggest that the coordination of these moieties may play more of a role in catalysis than in substrate binding or that they less ordered/ more mobile.

4.5.3 Inter-residue distances in the active site One factor that can affect enzyme catalysis is a decrease in enzyme mediated substrate restraints. A decrease in substrate restraints may be indicated by a change in the active site volume or an increase in the B-factor for a given substrate. An analysis of the surface of the binding pocket shows that there is little difference in the overall structure of the active site of each complex (figure 4.5.3.1).

Fig. 4.5.3.1. Surface view of the binding pocket for each phosphagen analogue complex. This view demonstrates the overall similarity of each

active site conformation. Image generated with PyMOL (MAC).

However, a closer analysis shows that some of the non-cognate substrates have a significant impact on the distances between the side chains of specific active site residues. These data represent measurements diagonally and perpendicularly across the active site and are indicative of changes in the overall active site volume. The ILO molecule binds to AK in a configuration that is very similar to the native phosphagen and has little effect on the relative distances between many of the residue side chains. This suggests that there is little impact on the active

60 site volume and most likely no collateral effect on the ability of the enzyme to restrain the substrate position. However, ILO fails to act as a substrate in AK because the reactive (sp3) nitrogen resides in the non-reactive position as designated in the cognate AK TSAC complex (i.e. closer to Cys271). In the ORN and CIT structures there is a decrease in the bonding distance from residue 68 to 314 and from 225 to 314. This suggests that the active site volume is decreased and may indicate an increase in the restraint of the substrate position. Although this is likely not relevant for CIT activity because of the loss of nitrate binding it may provide part of the explanation for the inactivity of the ORN complex were the ligands are held too rigid to allow catalysis. In the DARG structure there is an increase in the distance from 68 to 314 and from 225 to 314. This is the only model where an apparent increase in the binding pocket volume was noted. This observation combined with the decreased binding affinity of DARG (as indicated by the necessity to increase the DARG concentration during crystallization) suggests that the inactivity of DARG could be due to a weakening of bonding restraints between the enzyme and the substrate. This is further indicated by an increase in the B-factor of the DARG ligand (the only ligand with an increased thermal factor) compared to the cognate ARG ligand. It should be noted that a deviation in active site volume is not just dependent on the size of the ligand being bound, as indicated by the comparison of the large ILO and small ORN ligand active sites.

4.6 Ligand Binding 4.6.1 Phosphagen bonding restraints The kinetic investigation of AK mutants shows that disruption of the position of the guanidinium has an effect on catalysis but very little effect on substrate binding (Azzi et al., 2004; Pruett et al., 2003). Here we have shown that ILO and ORN bind to AK but are not reactive. The ILO molecule contains an sp3 nitrogen and the ORN molecule contains an sp2 nitrogen, both of which could be catalytically active but each fails to be a substrate because of their coordination. Furthermore, we used ORN and DARG to show that coordination of the guanidinium is also important for optimal catalysis. These data suggest that the chemical redundancy of the guanidinium is important for proper alignment and that proper alignment is important to catalysis. The fact that these substrates bind to AK even in the absence of (ORN) or total

61 displacement of (CIT) the guanidinium suggests that the bonding restraints mediated by the guanidinium contribute primarily to substrate alignment not to substrate binding. This contention is supported by kinetic investigations which demonstrate that ORN binds nearly as well as the cognate arginine and is a better competitive inhibitor than CIT (Watts et al., 1980). This confirms the previous suggestion (Pruett et al., 2003) that residues Glu225 and Glu314 most likely mediate their effect by altering substrate alignment. An examination of the CIT and DARG structures shows that despite the weakening of (CIT) or total disruption of (DARG) the α-amine bonding restraints these ligands still bind to the active site. Therefore like the guanidinium, the α-amine may not be critical for phosphagen binding but may play a critical role in reactivity. This also confirms previous kinetic investigations that show that substrates lacking the α-amine still act as inhibitors in the AK reaction (Watts et al., 1980). The carboxylate moiety of the substrate mediates bonding restraints with the backbone of residues in the “specificity” loop (residues 62-68). This flexible loop facilitates competent bonding interactions even in the presence of major translational (CIT) and rotational (DARG) deviations of the carboxylate. These new bonding interactions are sufficient to allow substrate binding. Previous investigations have shown that DARG displays the weakest binding affinity of all these non-cognate substrates (Watts et al., 1980). Major structural deviations in the DARG structure are seen in the guanidinium and carboxylate coordination. Taken together these data indicate that the bonding restraints mediated by the carboxylate are more important to substrate binding than the restraints mediated by the guanidinium. This explains previous results that show that substrates lacking the carboxylate fail to protect against inactivation with 5,5′-dithiobis-(2-nitobenzoic acid) (Feng et al., 2005; Pan et al., 2004) and that mutation of Phe63 and Leu65 (which make bonding interactions with the carboxylate) to glycine in the Stichopus AK enzyme demonstrates a remarkable increase in the

Kmarg (Uda and Suzuki., 2004). Interaction of the carboxylate group directly with the backbone of the flexible loop rather than the side chains of local residues and the tolerance of deviations in substrate structure may provide an explanation for the retention of activity in some site-directed mutants of loop elements that may not have fully disrupted interactions with the native backbone interactions (Azzi et al., 2004).

62 4.6.2 Mg+2 and nitrate coordination Previous investigations of AK by PPR, EPR and NMR (Bulttier and Midred, 1974; Ray et al., 2002) suggest that the phosphoryl moiety moves about 1.0Å during in-line transfer. In the non- cognate complexes presented here, the position of the Mg+2 shifts by as much as 1.0Å. These data suggest that the increase in bonding distance noted in the earlier studies may have been due to a shift in the position of the metal ion not an overall displacement of the phosphoryl group during transfer. In the DARG complex disruption of the magnesium position has little impact on the coordination of the nitrate and in the ORN complex the displacement of the nitrate has little effect on magnesium coordination. Taken together these observations suggest that at the transition state the nitrate is not restrained by a bonding interaction with the Mg+2. Displacement of the Nε atom in the ORN complex has a dramatic affect on the position of the nitrate with a displacement of 1.6Å. This observation along with the loss of nitrate binding after complete displacement of the guanidinium in the CIT structure indicate that a positive charge at Nε may be important for the precise alignment of the nitrate in the mimic high-energy transition state.

4.7 Reactive Alignment It has been suggested that in AK catalysis may be significantly impacted by the precise alignment of the gamma phosphate with the reactive nitrogen of the guanidinium (Zhou et al., 1998; Pruett et al., 2003; Yousef et al., 2002). In the high-resolution TSAC structure of AK it was noted that the substrates were aligned within 3.0˚ of optimal for in-line transfer of the phosphoryl group. These structures confirm previous results that indicate a reactive mechanism that involves the precise pre-alignment of two substrate moieties so that the orbitals are aligned for optimal reaction trajectories (Zhou et al., 1998; Yousef et al., 2003; Pruett et al,. 2003).

Fig. 4.7.1. Details of substrate alignment in the wild type verse the analogue complexes. Emphasizing the near-linearity of phosphate transfer. Figure: a= WT, b= ILO, c=ORN, and d=DARG. The CIT structure was not analyzed because the nitrate was not retained. Image generated with Web-lab Viewer Pro.

The structures of the analogue complexes show varying degrees of misalignment in the angles of nucleophilic attack at the guanidinium (Nη2) and ADP (O3). The ILO structure (although non-

64 reactive due to position of the CH2) shows the smallest deviation in the angle of attack (2.0˚) were as DARG and ORN (which could act as substrates) are much higher at 5.0˚ and 12.0˚ respectively (figure 4.7.1a,b,c,d). The structure of these “subtly” distorted complexes provides experimental evidence that a deviation of a few tenths of an angstrom may make the difference between a fully active and completely inactive enzyme. The minor guanidine deviations in ILO and DARG most likely have a minimal impact on catalytic parameters, such as pK perturbation, lending weight to the importance of substrate alignment. Disruption of the guanidinium position has a major impact on catalysis. In the ILO, ORN and DARG structures a displacement of the guanidinium is mirrored with a displacement in the coordination of the Mg+2. The DARG structure shows the most significant displacement of the magnesium (1.0Å) with a weakening of the bonding interactions with both O1α and O1β of ADP by 0.4Å and 0.9Å, respectively. This decrease in local bonding restraints for the terminal phosphates alters transition state stabilization and would negatively impact catalysis. This describes one mechanism through which an enzyme can mediate substrate specificity with only the cognate substrates binding to the active site and being held sufficiently rigid for efficient catalysis. In the DARG complex, disruption of the magnesium position has little impact on the coordination of the nitrate and in the ORN complex the displacement of the nitrate has little effect on magnesium coordination. Taken together these observations suggest that at the transition state the nitrate is not restrained by a bonding interaction with the Mg+2. However, displacement of the Nε atom in the ORN complex has a dramatic affect on the position of the nitrate with a displacement of 1.6Å. This observation combined with the loss of nitrate binding after complete displacement of the guanidinium in the CIT structure indicate that a positive charge at Nε may be important for the precise alignment of the nitrate in the mimic high-energy transition state.

CHAPTER 5 CONCLUSION

An enzyme has many ways to facilitate catalysis. The conundrum that the enzyme must solve is to bind tightly only the unstable transition state structure (lifetime equal to one bond vibration), but not substrate or product. In many enzyme catalyzed reactions the movement of dynamic domains is responsible for the close proximity of many secondary structural elements and allows for interactions between specific enzyme structures. These interactions play a role in restricting the movement of highly flexible loops or secondary structures within a protein (reducing the unfavorable contributions to the activation barrier) and are responsible for forming a competent active site that can align the reactive moieties of the ligands. The formation of a competent active site provides for the rigid restraint of reactive moieties and plays a role in the alignment of critical ligand elements. In AK the Mg-nucleotide complex makes many bonding interactions with the fixed domain (Yousef et al., 2003). Furthermore, the work presented here suggests that bonding of the phosphagen ligand is primarily through the carboxylate raising the possibility that the phosphagen may not require the closed state for binding. These two observations suggest that the induced fit movements noted in AK may serve only to bring the two substrates in close juxtaposition (as well as optimizing the electrostatic environment of the active site) in order to facilitate efficient catalysis but are not responsible for substrate binding. These data also imply that the motion of domain #1 (Tyr68, Cys271) and domain #3 (Glu314) are most likely subsequent to nucleotide binding. Two forms of structural evidence support this contention. First, the “specificity” loop (residues 62-68) on domain #1 is tethered by intra-domain bonds that don’t change distances from the substrate-free to ligand bound state. This imparts a rather rigid nature to this area and the loop cannot undergo large positional changes to compensate for deviations in substrate coordination. Second, in DynDom analysis there is retention of the angle of the rotation axis for domain #3 whereas the axis of rotation for domain #1 deviates significantly from structure to structure. This suggests that deviations in ligand structure are compensated for by changes in the angle and magnitude of the rotation axis for domain #1 more

66 than any other domain. The notion that the phosphagen can bind through its interactions at the carboxylate would further suggest that the interactions between the guanidinium and the enzyme serve only to stabilize the reactive configuration at the transition state and do not play a role in substrate binding. The relative contribution of substrate alignment to catalysis has been the subject of significant debate for many years. The structure of a reactive complex trapped in a mimic of the high-energy transition state and in the absence of any enzyme mutation or covalent manipulation of the substrate complex is the quintessential model for addressing this question. The observation of the DARG structure demonstrates that minor deviations in substrate alignment (a few tenths of an angstrom or a few degrees) can render a ligand completely inactive. Retention of the native guanidinio group in this complex (which should be reactive) provides direct evidence for the importance of substrate alignment in catalysis. Furthermore, the observation that organization of the active site plays a role in catalysis but not substrate binding raises an interesting catalytic question. Of critical importance is whether the interactions between the organized active-site and the substrate strain the ligand (such that the optimal position in the reaction is enhanced) or whether these contacts simply interact with a substrate conformer that approaches a local geometric minimum. The observation that arginine kinase can bind many alternate phosphagen conformations (CIT, DARG) and the notion that the bonding interaction between the guanidinium and the enzyme serve primarily to stabilize the reactive configuration suggest that the substrate(s) is most likely strained to achieve catalytic coordination. This may also explain why some of the ligands are contorted into chemically unfavorable geometric configurations. These findings could be validated by a quantum mechanical and molecular mechanical analysis of the reactive complex of each of the analogue structures as well as a series of binary complexes in which activity could be correlated with the distortion from ideal geometry. Moreover, the complexity of the induced-fit changes in AK enhances the necessity to investigate the timing of conformational changes, a parameter that is well suited for investigation by NMR methods.

APPENDIX A: SUPPLIES Pipetes Baffled flasks Centrifugal bottles Falcon tubes (15ml & 50ml) Coverslips, siliconized Crystallization trays Filter apparatuses (large vessel) Memebrane filters (0.45µm & 0.22µm ) Large volume concentrators Concentrators Microcon YM10 Pipet tips (1µl – 1ml) Capillaries (0.1, 0.2, 0.5, 1.0mm I.D.) Mounting wax Plastic cuvettes Syringes (3ml, 5ml, 20ml, 60ml) Syringe filters (0.22µm. 0.45µm) Eppendorf tubes (0.5, 1.5, 2.0ml) Cryo-tubes (NUNC) Nylon loops (0.1, 0.2, 0.5, 1.0mm)

APPENDIX B: CHEMICALS

Acrylamide-Bisacrylamide Ampicillin, sodium salt Ammonium persulfate (APS) Bradford reagent Bio-Rad Protein Assay Coomassie Brilliant Blue R-250 Ethylenediaminetetraacetic acid (EDTA) > 99% Ethanol Isopropyl-beta-D-thiogalactopyranoside (IPTG) Methanol TEMED 1-Piperazineethane sulfonic acid, 4-(2-hydroxyethyl) (HEPES) Sodium chloride (NaCl) Ethidium bromide Glacial acetic acid Sodium dodecyl sulfate (SDS) Tris-HCL (Trizma)™ Tris-Base β-Mercaptoethanol phenylmethylsulfonyl fluoride (PMSF) Glycerol L-Arginine L-Imino-ethyl-ornithine L-Citrulline L-Ornithine D-Arginine Sodium nitrate (NaNO3) Magnesium chloride (MgCl2) Sodium azide (NaN3) Phosphate buffered saline (PBS) Phosphate (mono & di-basic)

APPENDIX C: SOLUTIONS

Ampicillin stock……………………..50 mg/ml ampicillin in dH2O

APS solution………………………...30mg APS in 300µl H2O

Coomassie destaining solution……...50ml methanol 75ml glacial acetic acid H2O up to 1 liter

Coomassie staining solution………...0.5g Coomassie Brilliant Blue 250ml methanol 30ml glacial acetic acid H2O up to 500ml

IPTG stock (1M)……………………2.38g IPTG in 10ml H2O, stored at -20°C

LB media……………………………10g Trypton, 5g Yeast extract, 10g NaCl

LB-Amp media……………………...LB with 50µg/ml ampicillin

3X SDS sample buffer………………750mg SDS 1.5ml β-Mercaptoethanol 1.2ml of 0.25% (w/v) Bromophenol Blue stock 1.0ml stacking gel buffer 4.0ml of 50% (v/v) glycerol H2O up to 10ml

(10x) Running Buffer (1 liter)………Tris-base 30.3g Glycine 144g SDS 10g

Separating gel buffer………………..1.5M Tris pH 8.8, titrated with HCl at 25°C

Stacking gel buffer………………….1.5M Tris pH 8.8, titrated with HCl at 25°C

Soda solution………………………..520mg Na2CO3 in 60ml H2O

Stop solution………………………...7% (v/v) glacial acetic acid in H2O

Thiosulfate solution…………………10% (w/v) Na2S2O3 in H2O, stored at 4°

APPENDIX D: BUFFERS

Lysis buffer: 50mM Tris-HCl Unfolding Buffer: 7mM DTT 50mM Tris-HCl 1µM PMSF 8M Urea pH = 8.0 1mM DTT 1mM EDTA Regular Buffer: pH = 8.0 10mM Tris-HCl 1mM EDTA Refolding Buffers: 10mM KCl 50mM Tris-HCl 1mM DTT 1mM EDTA 0.02% NaN3 100mM β-mercaptoethanol pH = 8.0 Urea: 4M, 2M, 0.5M pH = 8.0 Inclusion Body Wash Buffer: 50mM Tris-HCL Buffer: gel filtration-denatured (S300) 2% Triton-X 100 50mM Tris-HCl 10mM EDTA 6M Urea 1M Urea 1mM DTT pH = 8.0 1mM EDTA pH = 8.

Buffer: gel filtration (S100) AAC Buffer: 10mM Tris-HCl 5mM MgCl2 1mM EDTA 4mM K-ADP 100mM KCl 50mM NaNO3 0.02% NaN3 20mM ILO/ ORN/ CIT pH = 8.0 5mM NaN3 1mM DTT Buffer: anion exchange (DEAE) pH = 7.5 10mM Tris-HCl 1mM EDTA AAC Buffer: D-arginine 10mM KCl 5mM MgCl2 1mM DTT 4mM K-ADP 0.02% NaN3 50mM NaNO3 pH = 8.0 50mM D-arginine 5mM NaN3 1mM DTT pH = 7.5

APPENDIX E: INSTRUMENTS

AKTA (FPLC) GE Healthcare (Piscataway, NJ) Anion exchange column (DEAE) GE Healthcare (Piscataway, NJ) Anion exchange column (mono-Q) GE Healthcare (Piscataway, NJ) Analytical balance Sartorius (Gottingen) Autoclave Systec(Wettenberg) Centrifuge J2-21 Beckman Coulter (Fullerton, CA) Centrifuge (tabletop) Sorvall RC-6 Thermo Electron Corp. (Asheville, NC) Cryo-systems Oxford Cyosystems (Devens, MA) French pressure cell STANDSTED Inc. Superdex S300 HiLoad (16/60) GE Healthcare (Piscataway, NJ) Superdex S100 HiLoad (10/30) GE Healthcare (Piscataway, NJ) Leica Microscope (with polarizer) Leica Microsystems (Bannockburn, IL) Mar-CCD detector MARUSA (Evanston, IL) Microfluidizer Microfluidics (Newton,MA) Mini Protean 3 cell BIO-RAD Laboratories (Hercules,CA) Orbital shaker Cole-Parmer (Vernon Hills, IL) pH electrode Fisher Scientific (Pittsburgh, PA) Photometer Smart-Spec™ BIO-RAD Laboratories (Hercules,CA) Pipetman pipettes Gilson (Middleton, USA) PowerPac BIO-RAD Laboratories (Hercules,CA) Quarts microcuvette 100µl, 1cm path length Hellma (Jena) Rigaku R-axis II Rigaku (The Woodlands, TX) Shaker incubator New Brunswick Scientif (Edison, NJ) Ultra-pure water Millipore (Billerica, MA)

APPENDIX F: COMPUTATIONAL SOFTWARE

Automar Mar research GmbH (Hamburg) CCP4 5.0.2 Collaborative Computational Project, number 4 (1994) CNS Collaborative CNS effort (Yale University) COOT Emsley and Cowtan (2004) DynDom Theoretical and Comp. Biophysics Group (Univ. of Illinois) HKL HKL research, Inc. (Charlottesville, USA) O Uppsala Software Factory Pov-Ray Persistence of Vision Ray Tracer Pty. Ltd. Swiss PDB Viewer GlaxoSmithKline R&D and Swiss Institute of Bioinformatics Web-Lab viewer Pro Accelrys Inc. PyMOL DeLano Scientific L.L.C PROCHECK European Bioinformatics Institute

REFERENCES

Alonso GD, Pereira CA, Remedi MS, Paveto MC, Cochella L, Ivaldi MS, Gerez de Burgos NM, Torres HN, Flawia MM. 2001. Arginine kinase of the flagellate protozoa Trypanosoma cruzi. Regulation of its expression and catalytic activity. FEBS Lett. 498(1): 22-5.

Alber, T. C., R. C. Davenport, D. A. Giammona, E. Lolis, G. A. Petsko, and D. Ringe. 1987. Crystallography and site-directed mutagenesis of yeast triosephosphate isomerase: what can we learn and catalysis from a "simple" enzyme. Cold Spring Harb. Symp. Quant. Biol. 52: 603-613.

Amadeu TP, Costa AM. 2006. Nitric oxide synthesis inhibition alters rat cutaneous wound healing. J Cutan Pathol. 33(7): 465-73.

Anderson, C. M., F. H. Zucker, and T. A. Steitz. 1979. Space-Filling Models of Kinase Clefts and Conformation Changes. Science. 204: 375-380.

Asherie N., 2004. Protein crystallization and phase diagrams. Methods. 34(3):266-72.

Ataka M. 1993. Protein crystal growth: An approach based on phase diagram determination. Phase Transitions 45: 205-219.

Azzi A., Clark S.A., Ellington W.R., and Chapman M.S. (2004) Protein Sci. 13: 575-585.

Babbitt, P. C., G. L. Kenyon, I. D. Kuntz, F. E. Cohen, J. D. Baxter, P. A. Benfield, J. D. Buskin, W. A. Gilbert, S. D. Hauschka, J. P. Hossle, C. P. Ordahl, M. L. Pearson, J.-C. Perriard, L. A. Pickering, S. D. Putney, B. L. West, and R. A. Ziven. 1986. Comparisons of creatine kinase primary structures. J. Protein Chem. 5: 1-14.

Baylis C. 2006. Arginine, arginine analogs and nitric oxide production in chronic kidney disease. Nat Clin Pract Nephrol. 2(4): 209-20. Review.

Bennett, W. S., Jr., and T. A. Steitz. 1980. Structure of a complex between yeast hexokinase A and glucose II. Detailed comparisons of conformation and active site configuration with the native hexokinase B monomer and dimer. J. Mol. Biol. 140:211- 230.

Bernstein, B.E. and Hol, W.G.J. 1998. Crystal structures of substrates and products bound to the phosphoglycerate kinase active site reveal the catalytic mechanism. Biochemistry 37: 4429–4436.

Blaszczyk, J., Li, Y., Yan, H., and Ji, X. 2001. Crystal structure of unligated guanylate kinase from yeast reveals GMP-induced conformational changes. J. Mol. Biol. 307: 247– 257.

Blethen, S. L. 1972. Kinetic properties of the arginine kinase isoenzymes of Limulus

74 polyphemus. Arch. Biochem. Biophys. 149: 244-251.

Borders CL Jr, Snider MJ, Wolfenden R, Edmiston PL. 2002. Determination of the affinity of each component of a composite quaternary transition-state analogue complex of creatine kinase. Biochemistry. 41(22): 6995-7000.

Boyer, R. 1998. Concepts in Biochemistry. Brooks/Cole, Pacific Grove, CA.

Brown AE, Grossman SH. 2004. The mechanism and modes of inhibition of arginine kinase from the cockroach (Periplaneta americana). Arch Insect Biochem Physiol. 57(4): 166-77.

Bruice TC. 1972. Views on approximation, orbital steering, and enzymatic and model reactions. Cold Spring Harb Symp Quant Biol. 36: 21-7.

Bruice, T.C. 2002. A view at the millennium: The efficiency of enzymatic catalysis. Acc. Chem. Res. 35:139-148.

Brünger, A.T. 1992. Free R value: A novel statistical quantity for assessing the accuracy of crystal structures. Nature 355: 472–475.

Brünger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., et al. 1998. Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr. 54: 905–921.

Brünger, A. T. (1997) The Free R Value: A More Objective Statistic for Crystallography, Methods Enzymol. 277: 366-396

Chapman, M. S., and Blanc, E. 1997. Potential use of Real Space Refinement in Protein Structure Determination. Acta Crystallographica D53: 203-6

Chayen NE, Radcliffe JW, Blow DM. 1993. Control of nucleation in the crystallization of lysozyme. Protein Sci 2: 113-118.

Cheung, C.W. and Mas, M.T. 1996. Substrate-induced conformational changes in yeast 3-phosphoglycerate kinase monitored by fluorescence of single tryptophan probes. Protein Sci. 5: 1144–1149.

Collaborative Computational Project, N. 1994. The CCP4 Suite: Programs for Protein Crystallography. Acta Crystallogr. D50: 760-763.

Connolly, M. L. 1983. Solvent accessible surfaces of proteins and nucleic acids. Science. 221: 709-713.

Cook, P. F., Kenyon, G. L., and Cleland, W. W. (1981) Use of pH Studies to Elucidate the Catalytic Mechanism of Rabbit Muscle Creatine Kinase, Biochemistry 20: 1204-1210.

Cowtan, K.D., and Main, P. 1996. Phase combination and cross validation in iterated

75 density-modification calculations. Acta Crystallographica D52, 43-48.

Curran JN, Winter DC, Bouchier-Hayes D. 2006. Biological fate and clinical implications of arginine metabolism in tissue healing. Wound Repair Regen. 14(4): 376- 86.

Dafforn, A., and D. E. Koshland, Jr. 1971. Theoretical aspects of orbital steering. Proc. Natl. Acad. Sci. USA. 68: 2463-2467.

Dobritzsch, D., Ricagno, S., Schneider, G., Schnackerz, K.D., and Lindqvist, Y. 2002. Crystal structure of the productive ternary complex of dihydropyrimidine dehydrogenase with NADPH and 5-iodouracil: Implications for mechanism of inhibition and electron transfer. J. Biol. Chem. 277: 13155–13166.

Ducruix, A. and Giege, R. 1992. Methods of Crystallization. In Crystallization of Nucleic Acids and Proteins., A. Ducruix and R. Giege, eds. (Oxford: IRL Press).

Dumas, C., and J. Camonis. 1993. Cloning and sequence analysis of the cDNA for arginine kinase of lobster muscle. J. Biol. Chem. 268: 21599-21606.

Dumas, C., and J. Janin. 1983. Conformational changes in arginine kinase upon ligand binding seen by small-angle x-ray scattering. FEBS Lett. 153: 128-130.

Durbin S.D., and Feher. 1996. Protein crystallization. Annu Rev Phys Chem. 47: 171-204. Review.

Eder, M., Schlattner, U., Becker, A., Wallimann, T., Kabsch, W., and Fritz-Wolf, K. 1999. Crystal structure of brain-type creatine kinase at 1.41 Å resolution. Protein Sci. 8: 2258–2269.

Eder, M., Fritz-Wolf, K., Kabsch, W., Wallimann, T., and Schlattner, U. 2000. Crystal structure of human ubiquitous mitochondrial creatine kinase. Proteins 39: 216–225.

Eisenberg, D., and A. D. McLachlan. 1986. Solvation energy in protein folding and binding. Nature. 319: 199-203.

Ellington, W.R. 1989. Phosphocreatine represents a thermodynamic and functional improvement over other muscle phosphagens. J. Exp. Biol. 143: 177-194.

Ellington, W.R. 2001. Evolution and physiological roles of phosphagen systems. Annu. Rev. Physiol. 63: 289–325.

Fayos R, Melacini G, Newlon MG, Burns L, Scott JD, Jennings PA. 2003. Induction of flexibility through protein-protein interactions. J. Biol. Chem. 278(20): 18581-7.

Feher G, Kam Z . 1985. Nucleation and growth of protein crystals: General principles and assays. Methods Enzymol 114: 77-112.

Fersht, A. 1985. Enzyme Structure and Function. Freeman, New York. 76

French GS., and Wilson KS., 1978. On the treatment of negative intensity observations. Acta Cryst. 34: 517-525.

Forstner, M., M. Kriechbaum, M. P. Laggner, and T. Wallimann. 1996. Changes of creatine kinase structure upon ligand binding as seen by small-angle scattering. J. Mol. Struct. 383: 217-227.

Forstner, M., K. Manfred, P. Laggner, and T. Wallimann. 1998. Structural changes of creatine kinase upon substrate binding. Biophys. J. 75: 1016-1023.

Fritz-Wolf, K., T. Schnyder, T. Wallimann, and W. Kabsch. 1996. Structure of mitochondrial creatine kinase. Nature. 381: 341-345.

Furter, R., Furter-Graves, M., and Walliman, T. 1993. Creatine kinase: The reactive cystein is required for synergism but is nonessential for catalysis. Biochemistry 32: 7022– 7029.

Gerstein, M., and C. Chothia. 1991. Analysis of protein loop closure. Two types of hinges produce one motion in lactate dehydrogenase. J. Mol. Biol. 220: 133-149.

Gerstein, M., and M. Krebs. 1998. A database of macromolecular motions. Nucleic Acids Res. 26: 4280-4290.

Gerstein, M., A. M. Lesk, and C. Chothia. 1994. Structural mechanisms for domain movements in proteins. Biochemistry. 33: 6739-49.

Gross, M., E. M. Furter-Graves, T. Wallimann, H. M. Eppenberger, and R. Furter. 1994. The tryptophan residues of mitochondrial creatine kinase: roles of Trp-223, Trp-206, and Trp-264 in active-site and quaternary structure formation. Protein Sci. 3: 1058-1068.

Gross, M., and T. Wallimann. 1995. Dimer-dimer interaction in octameric mitochondrial creatine kinase. Biochemistry. 34: 6660-6667.

Guex, N., and Peitsch, M. C. (1997) SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling, Electrophoresis 18: 2714-2723.

Hanna R, Doudna JA. 2000. Metal ions in ribozyme folding and catalysis. Curr Opin Chem Biol. 4(2): 166-70. Review.

Hansen, D. E., and J. R. Knowles. 1981. The stereochemical course of the reaction catalyzed by creatine kinase. J. Biol. Chem. 256: 5967-5969.

Hayashi T, Tsujino T, Iwata S, Nonaka H, Emoto N, Yano Y, Otani S, Hayashi Y, Itoh H, Yokoyama M. Decreased ornithine decarboxylase activity in the kidneys of Dahl salt- sensitive rats. Hypertens Res. 25(5): 787-95.

Hayward, S. 1999. Structural principles governing domain motions in proteins. Proteins 36: 425-435. 77

Hayward, S. and Berendsen, H.J. 1998. Systematic analysis of domain motions in proteins from conformational change: new results on citrate synthase and T4 lysozyme. Proteins 30: 144-154.

Henikoff, S., and J. G. Henikoff. 1992. Amino acid substitution matrices from protein blocks. Proc. Natl. Acad. Sci. USA. 89: 10915-10919.

Hoare DG. 1972. Significance of molecular alignment and orbital steering in mechanisms for enzymatic catalysis. Nature. 236(5348): 437-40.

Jencks, W. P., and M. I. Mage. 1974. "Orbital steering," entropy and rate accelerations. Biochem. Biophys. Res. Commun. 57: 887-892.

Jones, T. A., and M. Kjeldgaard. 1997. Electron-density map interpretation. Methods Enzymol. 277: 173-208.

Jones, T. A., Zou, J.-Y., Cowan, S. W., and Kjeldgaard, M. 1991. Improved Methods for Building Protein Models in Electron Density Maps and Location of Errors in thee Models. Acta. Crystallographica A47: 110-9

Joseph, D., G. A. Petsko, and M. Karplus. 1990. Anatomy of a conformational change: hinged "lid" motion of the triosephosphate isomerase loop. Science. 249: 1425-1428.

Kabsch, W. 1976. A solution for the best rotation to relate two sets of vectors. Acta Crystallogr. A. 32: 922-923.

Kabsch, W., and K. Fritz-Wolf. 1997. Mitochondrial creatine kinase a square protein. Curr. Opin. Struct. Biol. 7: 811-818.

Kammermeier H. 1993. Meaning of energetic parameters. Basic Res. Cardiol. 88:380–84

Kammermeier H. 1987. Why do cells need phosphocreatine and a phosphocreatine shuttle? J. Mol. Cell. Cardiol. 19: 115–18

Kammermeier H, Schmidt P, Jungling E. 1982. Free energy change of ATP-hydrolysis: a causal factor of early hypoxic failure of the myocardium. J. Mol. Cell. Cardiol. 14: 267– 77

Karlsson, R., Zheng, J., Xuong, N., and Sowadski, J.M. 1993. Structure of mammalian catalytic subunit of cAMP-dependent protein kinase and an inhibitor peptide displays an open conformation. Acta Cryst. D 49: 381–388.

Kenyon, G. L., and G. H. Reed. 1983. Creatine kinase: structure-activity relationships. Adv. Enzymol. 54: 367-426.

Khuchua ZA, Vasiljeva EV, Clark JF, Korchazhkina OV, Branishte T, Kapelko VI, Kuznetsov AV, Ventura-Clapier R, Steinschneider AYa, Lakomkin VL, et al. 1992. The creatine kinase system and cardiomyopathy. Am J Cardiovasc Pathol. 4(3): 223-34. 78

Kleywegt, G. J., and Jones, A. T. (1996) Efficient Rebuilding of Protein Structures, Acta Crystallogr. D52: 829-832.

Knighton, D.R., Zheng, J.H., Ten Eyck, L.F., Ashford, V.A., Xuong, N.H., Taylor, S.S., and Sowadski, J.M. 1991. Crystal structure of the catalytic subunit of cyclic adenosine monophosphate-dependent protein kinase. Science 253: 407–414.

Knowles, J.R. 1991. To build an enzyme. Phil. Trans. R. Soc. Lond. Ser. B. 332: 115– 121.

Koshland, D. E. J. 1958. Application of a theory of enzyme specificity to protein synthesis. Proc. Natl. Acad. Sci. USA. 44: 98-104.

Kraulis, P. 1991. MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24: 946-950.

Lahiri, S. D., Wang, P. F., Babbitt, P. C., McLeish, M. J., Kenyon, G. L., and Allen, K. N. 2002. The 2.1 A Structure of Torpedo californica Creatine Kinase Complexed with 2+ - the ADP-Mg -NO3 -Creatine Transition-State Analogue Complex, Biochemistry 4: 13861-13867.

Laemmli UK. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature. 227(5259): 680-5.

Laskowski, R. A., MacArthur, M. W., Moss, D. S. and Thornton, J. M. 1993. PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26: 283-91

Lee LV, Poyner RR, Vu MV, Cleland WW. 2000. Role of metal ions in the reaction catalyzed by L-ribulose-5-phosphate 4-epimerase. Biochemistry. 39(16): 4821-30.

Lolis, E., and Petsko, G. 1990. Transition-State Analogues in Protein Crystallography: Probes of the Structural Source of Enzyme Catalysis. Annual Review of Biochemistry 59: 597-630

Lui, N. S. T., and L. Cunningham. 1966. Cooperative effects of substrates and substrate analogs on the conformation of creatine phosphokinase. Biochemistry. 5: 144-149.

McDonald, R. C., T. A. Steitz, and D. M. Engelman. 1979. Yeast hexokinase in solution exhibits a large conformational change upon binding glucose or glucose 6-phosphate. Biochemistry. 18: 338-342.

McPherson, A. 1982. Preperation and Analysis of protein Crystals (Malabar, FL: Kreiger).

McPherson, A 1973. A Preliminary Crystallographic Investigation of Rabbit Muscle Creatine Kinase. Journal of Molecular Biology 82: 83-86

79 McPherson, A. 1999. Crystallization of Biological Macromolecules. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Merritt, E. A., and D. J. Bacon. 1997. Raster3D: photorealistic molecular graphics. Methods Enzymol. 277: 505-525.

Mesecar, A.D., Stoddard, B.L., and Koshland Jr., D.E. 1997. Orbital steering in the catalytic power of enzymes: Small structural changes with large catalytic consequences. Science 277: 202-206.

Minor, W. 1993. XDisplayF Program (West Lafayette: Purdue University).

Moradi H, Kwok V, Vaziri ND. 2006 Effect of chronic renal failure on arginase and argininosuccinate synthetase expression. Am J Nephrol. 26(3): 310-8.

Moriarty RM, Adams T. 1973. A criticism of the use of certain bridged bicyclic hydroxycarboxylic acids as model compounds for the concept of orbital steering. J Am Chem Soc. 95(12): 4070-1

Morrison, J. F. 1973. Arginine kinase and other invertebrate guanidino kinases. in The Enzymes (Boyer, P. D., Ed.) pp 457-486, Academic Press, New York.

Mühlebach, S., M. Gross, T. Wirz, T. Wallimann, J. C. Perriard, and M. Wyss. 1994. Sequence homology and structure predictions of the creatine kinase isoenzymes. Mol. Cell Biochem. 133/4: 245-262.

Murali N, Jarori GK, Rao BD. 1994. Two-dimensional transferred nuclear Overhauser effect spectroscopy (TRNOESY) studies of nucleotide conformations in arginine kinase complexes. Biochemistry. 33(47): 14227-36.

Navaza, J. 1994. AMORE: An Automated Package for Molecular Replacement. Acta Crystallographica A50: 157-163.

Navaza, J., and Salundjian, P. 1997. AMORE: An Automated Molecular replacement Program Package. Methods in Enzymol. 277: 581-94

Nicholls, A. 1992. GRASP: Graphical Representation and Analysis of Surface Properties. Columbia University, New York.

Otwinowski, Z. and Minor, W. 1997. Processing of X-ray diffraction data collected in oscillation mode. Methods in Enzymol. 276: 307–326.

Otwinowski, Z., and Minor, W. (2001) in International Tables for Crystallography. Crystallography of Biological Molecules (Rossmann, M. G., and Arnold, E., Eds.) pp 226-235, International Union of Crystallography, Dortrecht.

Page, M. I., and W. P. Jencks. 1971. Entropic contributions to rate accelerations in enzymic and intramolecular reactions and the chelate effect. Proc. Natl. Acad. Sci. USA. 68: 1678-1683. 80

Pai, E. F., W. Sachsenheimer, R. H. Schirmer, and G. E. Schulz. 1977. Substrate positions and induced-fit in crystalline adenylate kinase. J. Mol. Biol. 114: 37-45.

Pasternak, R.C. and Braunwald, E. 1991. Acute Myocardial Infarction. In Harrison’s Principles of Internal Medicine, J. Wilson, E. Braunwald, K.J. Isselbacher, R. G. Petersdorf, J.B. Martin, A.S. Fauci and R. K. Root, eds. (New York, NY: McGraw-Hill, Inc.)

Pereira CA, Alonso GD, Ivaldi S, Silber AM, Alves MJ, Torres HN, Flawia MM. 2003. Arginine kinase overexpression improves Trypanosoma cruzi survival capability. FEBS Lett. 554(1-2): 201-5.

Pickover, C.A., McCay, D.B., Enelman, D.M., and Steitz, T.A. 1979. Substrate binding closes the active cleft between the domains in yeast phosphoglycerate kinase. J. Mol. Chem. 254: 11323–11329.

Pineda, A.O. and Ellington, W.R. 1999. Structural and functional implications of the amino acid sequence of dimeric, cytoplasmic and octameric mitochondrial creatine kinase from protostome invertebrate. Eur. J. Biochem. 264: 67–73.

Pruett, P.S., Azzi, A., Clark, S.A., Yousef, M., Gattis, J.L., Somasundaram, T., Ellington, W.R., and Chapman, M.S. (2003) J. Biol. Chem. 29: 26952-26957.

Ramachandran, G.N. and Srinivasan, R. 1970. Fourier methods in crystallography, M J. Buerger, ed. (New York: Wiley).

Ramelli GP, Joncourt F, Luetschg J, Weis J, Tolnay M, Burgunder JM. 2006. Becker muscular dystrophy with marked divergence between clinical and molecular genetic findings: case series. Swiss Med Wkly. 136(11-12): 189-93.

Rao, J. K., G. Bujacz, and A. Wlodawer. 1998. Crystal structure of rabbit muscle creatine kinase. FEBS Lett. 439: 133-137.

Rao, B. D. N., D. H. Buttlaire, and M. Cohn. 1976. 31P NMR studies of the arginine kinase reaction. J. Biol. Chem. 251: 6981-6986.

Read, R. J. 1985. Improved Fourier Coefficients for Maps Using Phases from Partial Structures with Errors. Acta Crystallographica A42: 140-9

Read, R.J. (1994) Model bias and phase combination. In From first map to final model: Proceedings of the CCP4 Study Weekend (eds. S. Sailey et al.), pp. 1–14. Daresbury Laboratory, Warrington, UK.

Reed, G. H., and M. Cohn. 1972. Structural changes induced by substrates and anions at the active site of creatine kinase. J. Biol. Chem. 247: 3073-3081.

Rhee, S., K. D. Parris, C. C. Hyde, S. A. Ahmed, E. W. Miles, and D. R. Davies. 1997. Crystal structures of a mutant (betaK87T) tryptophan synthase alpha2beta2 complex with 81 ligands bound to the active sites of the alpha- and beta-subunits reveal ligand-induced conformational changes. Biochemistry. 36: 7664-7680.

Rojo, M., R. Hovius, R. A. Demel, K. Nicolay, and T. Wallimann. 1991. Mitochondrial creatine kinase mediates contact formation between mitochondrial membranes. J. Biol. Chem. 266: 20290-20295.

Rossmann, M.G. 1972. Molecular replacement method. Gordon and Breach, New York, NY.

Saks VA, Belikova YO, Kuznetsov AV, Khuchua ZA, Branishte TH, Semenovsky ML, Naumov VG. 1991. Phosphocreatine pathway for energy transport: ADP diffusion and cardiomyopathy. Am J Physiol. 261(4 Suppl): 30-8.

Saridakis E, Chayen NE. 2003. Systematic improvement of protein crystals by determining the supersolubility curves of phase diagrams. Biophys J. 84: 1218-22.

Schramm L, Weierich T, Heldbreder E, Zimmermann J, Netzer KO, Wanner C. 2005. Endotoxin-induced acute renal failure in rats: effects of L-arginine and nitric oxide synthase inhibition on renal function. J Nephrol. 18(4): 374-81.

Schreuder, H. A., S. Knight, P. Curmi, I. Andersson, D. Cascio, C.-I. Brändén, and D. S. Eisenberg. 1993. Formation of the active site of ribulose-1,5-bisphosphate carboxylase/oxygenase by a disorder-order transition from the unactivated to the activated form. Proc. Natl. Acad. Sci. USA. 90: 9968-9972.

Schubert, H. L., E. B. Fauman, J. A. Stuckey, J. E. Dixon, and M. A. Saper. 1995. A ligand-induced conformational change in the Yersinia protein tyrosine phosphatase. Protein Sci. 4: 1904-1913.

Schulze-Lohoff E, Brand K, Fees H, Netzker R, Sterzel RB. 1991. Role of ornithine decarboxylase for proliferation of mesangial cells in culture. Kidney Int. 40(4): 684-90.

Schumacher, M. A., D. Carter, D. S. Roos, B. Ullman, and R. G. Brennan. 1996. Crystal structures of Toxoplasma gondii HGXPRTase reveal the catalytic role of a long flexible loop. Nature Struct. Biol. 3: 881-887.

Shen, Y., Tang, L., Zhou, H., and Lin, Z. 2001. Structure of human muscle creatine kinase. Acta Cryst. D57: 1196–1200.

Skulachev VP. 1992. The laws of cell energetics. Eur. J. Biochem. 208: 203–9

Stein, H. D., D. A. Harn, and J. R. David. 1990. A cloned ATP:guanidino kinase in the trematode Schistosoma mansoni has a novel duplicated structure. J. Biol. Chem. 265: 6582-6588.

Strong, S. 1995. Molecular studies of arginine kinase isoforms. Ph.D. dissertation. Florida State University, Tallahassee.

82 Strong, S. J., and W. R. Ellington. 1993. Horseshoe crab sperm contain a unique isoform of arginine kinase that is present in midpiece and flagellum. J. Exp. Zool. 267: 563-571.

Strong, S. J., and W. R. Ellington. 1995. Isolation and sequence analysis of the gene for arginine kinase from the chelicerate arthropod, Limulus polyphemus: insights into catalytically important residues. Biochim. Biophys. Acta. 1246: 197-200.

Stroud, R. M. 1996. Balancing ATP in the cell. Nature Struct. Biol. 3: 567-569.

Stura, E.A. and Wilson, I.A. 1990. Analytical and production seeding techniques. In Methods: A companion to methods in enzymology (ed. C.W. Carter, Jr.) 1: 38-49.

Stura, E. A. and Wilson, I. A. 1992. Seeding techniques. In Crystallisation of Nucleic Acids and Proteins. A. Ducruix, and R. Giege, editors. Oxford University Press, New York. 99–126.

Suzuki, T., and T. Furukohri. 1994. Evolution of phosphagen kinase: primary structure of glycocyamine kinase and arginine kinase from invertebrates. J. Mol. Biol. 237: 353-35.

Suzuki, T., Y. Kawasaki, and T. Furukohri. 1997a. Evolution of phosphagen kinase. Isolation, characterization and cDNA-derived amino acid sequence of two-domain arginine kinase from the sea anemone Anthopleura japonicus. Biochem. J. 328: 301-306.

Suzuki, T., Y. Kawasaki, T. Furukohri, and W. R. Ellington. 1997b. Evolution of phosphagen kinase. VI. Isolation, characterization and cDNA-derived amino acid sequence of lombricine kinase and the earthworm Eisenia foetida, and identification of a possible candidate for the guanidine substrate recognition site. Biochim. Biophys. Acta. 1343: 152-159.

Tisi, D.D., Bax, B.B., and Loew, A.A. 2001. The three-dimensional structure of cytosolic bovine retinal creatine kinase. Acta Crystallogr. D. Biol. Crystallogr. 57: 187–193.

Vinas JL, Sola A, Genesca M, Alfaro V, Pi F, Hotter G. 2006. NO and NOS isoforms in the development of apoptosis in renal ischemia/reperfusion. Free Radic Biol Med. 40(6): 992-1003.

Virden, R., and Watts, D. C. (1966) The Role of Thiol Groups in the Structure and Mechanism of Action of Arginine Kinase, Biochem. J. 99: 162-172.

Volkman B. F., Lipson D., Wemmer D. E., Kern D. 2001. Two-state allosteric behavior in a single-domain signaling protein.Science. 291:2429–2433.

Wallimann, T., M. Wyss, D. Brdiczka, K. Nicolay, and H. M. Eppenberger. 1992. Intracellular compartmentation, structure and function of creatine kinase isoenzymes: the "phospho-creatine circuit" for cellular energy homeostasis. Biochem. J. 281: 21-40.

Warshel. A. 2003. Computer simulations of enzyme catalysis: methods, progress, and insights. Annu Rev Biophys Biomol Struct. 32: 425-43.

83 Watts, D. C. 1973. in The Enzymes (Boyer, P. D., Ed.) pp 383-455, Academic Press, New York.

Watts D.C., Anosike E.0., Pollitt R.J., Lee C. R., Moreland B. 1980 The use of arginine analogues for investigating the functional organization of the arginine-binding site in lobster muscle arginine kinase. Role of the `essential' thiol group. J. Bioche. 185: 593- 599

Wilson AJC. 1949. The probability distribution of X-ray intensities. Acta Cryst. 2: 318- 321.

Wolfenden, R. 1999. Conformational aspects of inhibitor design: enzyme-substrate interactions in the transition state. Bioorg Med Chem. 7(5): 647-52.

Wothe, D. D., H. Charbonneau, and B. M. Shapiro. 1990. The phosphocreatine shuttle of sea urchin sperm: flagellar creatine kinase resulted from a gene triplication. Proc. Natl. Acad. Sci. USA. 87: 5203-5207.

Wriggers, W. and Schulten, K. 1997. Protein domain movements: Detection of rigid domains and visualization of hinges in comparisons of atomic coordinates. Proteins 29: 1-14.

Wyss, M., J. Smeitink, R. A. Wevers, and T. Wallimann. 1992. Mitochondrial creatine kinase: localization, structure and function, and clinical aspects. Biochim. Biophys. Acta. 288: 771-775.

Xie, Q., and M. S. Chapman. 1996. Canine parvovirus capsid structure, analyzed at 2.9 Å resolution. J. Mol. Biol. 264: 497-520.

Yan, H. and Tsai, M.D. 1999. Nucleoside monophosphate kinases: Structure, mechanism, and substrate specificity. Adv. Enzymol. Relat. Areas Mol. Biol. 73: 103-134.

Yousef, M. S., Clark, S. A., Pruett, P. K., Somasundaram, T., Ellington, W. R., and Chapman, M. S. (2003) Induced fit in guanidino kinases-comparison of substrate-free and transition state analogue structures of arginine kinase, Protein Sci. 12: 103-111.

Yousef, M. S., Fabiola, F., Gattis, J. L., Somasundaram, T., and Chapman, M. S. 2002 Refinement of the arginine kinase transition-state analogue complex at 1.2 A resolution: mechanistic insights, Acta Crystallogr., Sect. D: Biol. Crystallogr. 58: 2009-2017.

Zarrine-Afsar A, Mittermaier A, Kay LE, Davidson AR. 2006. Protein stabilization by specific binding of guanidinium to a functional arginine-binding surface on an SH3 domain. Protein Sci. 2006 15(1): 162-70.

Zhou, G., Parthasarathy, G., Somasundaram, T., Ables, A., Roy, L., Strong, S.J., Ellington, W.R., and Chapman, M.S. 1997. Expression, purification from inclusion bodies, and crystal characterization of a transition state analog complex of arginine kinase: a model for studying phosphagen kinases. Protein Sci. 6: 444-449

84 Zhou, G., T. Somasundaram, E. Blanc, Z. Chen, and M. S. Chapman. 1999. Critical initial real-space refinement in the structure determination of arginine kinase. Acta Crystallogr. D Biol. Crystallogr. 55: 835-84.

Zhou, G., T. Somasundaram, E. Blanc, G. Parthsarathy, W. R. Ellington, and M. S. Chapman. 1998. Transition state structure of arginine kinase: implications for catalysis of bimolecular reactions. Proc. Natl. Acad. Sci. USA. 95: 8449-8454.

Zhou, G., Ellington, W.R., and Chapman, M.S. 2000. Induced fit in arginine kinase. Biophys. J. 78: 1541–1550.

85 BIOGRAPHICAL SKETCH

Shawn A. Clark

Education Canton College of Technology A.A.S Medical Laboratory Technology (1994-1996) Canton, NY 13653

Clarkson University B.S. Biology (1996-1998) Potsdam, NY 13676 Academic Advisor: Dr Michel Roberts

Florida State University Ph.D Biochemistry (1998-2006) Tallahassee, FL 32306 Graduate Advisor: Dr. Michael S. Chapman Dissertation Title: “Arginine Kinase; A Crystallographic Investigation of Essential Substrate Structure”

Research Interests Structural Biology, primarily x-ray crystallography of protein-substrate complexes to probe active-site geometry and interactions involving the structural and functional relationships of enzymes.

Steady-State enzyme kinetics and bioinformatics to assist with rational structure-based drug design.

Protein engineering via site-directed mutagenesis and in-silico calculations.

Research Proficiencies RNA extraction and RTPCR Cloning Recombinant protein expression Protein purification by HPLC, FPLC, and low pressure systems Affinity, ion-exchange, size-exclusion, hydrophobic interaction chromatographies. Native and denaturing protein gel electrophoresis; DNA electrophoresis; Isoelectric focusing; Western Blotting Protein and DNA quantification by UV/Vis spectroscopy Protein crystallization Single crystal x-ray diffraction data collection (synchrotron and home source); Crystallographic data processing Molecular replacement and crystallographic refinement using automated reciprocal-space refinement, electron density map calculation, and real-space refinement

86 Protein homology model building.

Professional Experience 2003-2006: Research Fellow Harvard School of Medicine 2000-2003: Graduate Research Assistant; PI- Dr. Michael S. Chapman Florida State University 1998-2000: Teaching Assistant in General and Organic Chemistry Florida State University

Publications 1. Azzi A., Clark S.A., Ellington W.R., and Chapman M.S. (2004) Protein Sci. 13: 575- 585.

2. Pruett, P.S., Azzi, A., Clark, S.A., Yousef, M., Gattis, J.L., Somasundaram, T., Ellington, W.R., and Chapman, M.S. (2003) J. Biol. Chem. 29: 26952-26957.

3. Yousef, M. S., Clark, S.A., Pruett, P. S., Somasundaram, T., Ellington, W. R., and Chapman, M. S. (2003) Protein Sci. 12: 103–111.

4. Davulcu O, Clark SA, Chapman MS, Skalicky JJ. Main chain 1H, 13C, and 15N resonance assignments of the 42-kDa enzyme arginine kinase. (2005) J Biomol NMR. 32(2):178.

Professional Affiliations 2006 Research Scholar Tubingen Institute fur Biochemie (Tubingen, Germany) 2003-present: Harvard School of Medicine 2003-present: Beth Israel Deaconess Medical Center (Div. Infectious Disease) 2004-2006: Aid for Cancer Research Foundation (Research Fellow) 2003-2004: NIH Research Fellow