This item has the following downloads:
1 UNCOVERING STEREOCOMPLEMENTARITY IN TWO ALKENE REDUCTASES OF THE O LD Y ELLOW E NZYME FAMILY AND T HE DIFFERENT FATE S OF ELECTRONS IN THE SECOND HALF REACTION OF DIHYDROOROTATE DEHYDROGENASE 1B By YURI ALEXEY ANDREIW POMPEU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVER SI TY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVER SI TY OF FLORIDA 2013
2 2013 Yuri Alexey Andreiw Pompeu
3 To my mom, Elosavete Andreiv, who moved from our native Brazil just to be closer to me and has always supported my decisions no matter how absurd they may seem. I also dedicate this to my Father, Marcus Pompeu, whom I have gotten closer to every year of my life despite the 4,337 miles separating us.
4 ACKNOWLEDGMENTS First, I express my sincere gratitude to my advisor Professor Jon Stewart for his wrong paths. His approachability and guidance were invaluable to me and my career. I acknowledge the extreme patience and willingness to help of Professor Steve Brun er. He has always provided me with valuable insight and discussions from the moment I appeared in his office stating that I wished to learn protein crystallography. I must acknowledge Professor Rob McKenna who has always found the time not only to accommod ate my experiments in his facilities but also to answer questions, irrespective of their quality. I thank Professor David Ostrov, for bringing me into a very exciting project and trusting me with valuable data; I am truly honored. Finally, I thank Profess ors Valerie de Crecy Lagard and Nicole Horenstein for being part of my advisory committee and for their helpful criticism throughout this process. I must also acknowledge members of the Stewart group for many helpful discussions, in particular Dr. Bradford Sullivan, who participated directly in the studies of carvone reductions and Dr. Adam Walton who performed substrate screenings using OYE2.6 and the alkene reductase library of homologs.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 14 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 OYE1 AND BIOCATALYSIS ................................ ................................ ................... 17 The Isolation and Discovery of OYE1 ................................ ................................ ..... 17 History of Purification and Spectral Characterization ................................ .............. 18 The Structure of OYE1 ................................ ................................ ............................ 21 Overall S tructure ................................ ................................ .............................. 22 The FMN B inding E nvironment ................................ ................................ ........ 24 The Active S ite of OYE1 ................................ ................................ ................... 26 The Catalytic Cycle and Mechanism of OYE1 ................................ ........................ 28 Aromatization of Cyclic Enones: A Dismutation Reaction in OYE1 ......................... 31 Biocatalysis and the Potential of OYE1 as a Suitable Catalyst ............................... 33 Expectations and Limitations ................................ ................................ ............ 33 OYEs as Biocatalysts ................................ ................................ ....................... 34 ............ 37 2 UNCOVERING STEREOCOMPLEMENTARITY IN S. pastorianus OYE1 ............. 42 ........................... 42 ..................... 42 Discriminating Between (R) and ( S) ..................... 43 Addressing the Limitations: Experimental Design ................................ ................... 44 The Problem ................................ ................................ ................................ ..... 44 Experimental App roach and Techniques ................................ .......................... 45 Results and Discussion ................................ ................................ ........................... 46 Catalytic Studies of OYE1 M utants ................................ ................................ .. 46 (S) Carvone reductions ................................ ................................ .............. 46 (R) Ca rvone reductions ................................ ................................ .............. 47 Probing the Active Site through Charge Transfer Complex Formation Studies ................................ ................................ ................................ .......... 49 Crystallography S tudies ................................ ................................ .................... 52 Structural overview ................................ ................................ .................... 52 Wild typ e ................................ ................................ ................................ .... 53
6 Trp116Leu mutant ................................ ................................ ...................... 54 Trp116Ile mutant ................................ ................................ ........................ 60 Trp116Ala mutant ................................ ................................ ...................... 64 Trp116Val mutant ................................ ................................ ...................... 68 Trp116 Thr mutant ................................ ................................ ..................... 70 Trp116 Gln mutant ................................ ................................ ..................... 73 Conclusions and Future Work ................................ ................................ ................. 76 Experimental Procedures ................................ ................................ ........................ 81 Enzymatic Reductions Screening (Catalytic Studies) ................................ ....... 81 Charge Transfer Formation Assays ................................ ................................ 82 GC MS Analysis ................................ ................................ ............................... 82 Site Directed Mutagenesis ................................ ................................ ............... 83 Protein Expression and purification ................................ ................................ .. 84 Crystallographic Studies ................................ ................................ ................... 86 Structure Solution ................................ ................................ ............................. 87 Enzyme substrate Complex Modeling ................................ .............................. 88 3 STRUCTURAL AND CATALYTIC CHARACTERIZATION OF P. stipitis OYE2.6, A USEFUL CATALYST FOR ALKENE REDUCTIONS ................................ ........... 89 Background and Motivation ................................ ................................ .................... 89 Results and Discussion ................................ ................................ ........................... 92 Overal l Fold ................................ ................................ ................................ ...... 93 Quaternary Structure ................................ ................................ ........................ 94 Flavin Environment and Phenol Binding ................................ ........................... 97 The Active Site and Ligand Binding ................................ ................................ 100 Catalytic Characterization through Mutagenesis ................................ ............ 107 Conclusions and Future Direction ................................ ................................ ......... 113 Experimental Procedures ................................ ................................ ...................... 116 Protein Purification ................................ ................................ ......................... 116 Native Gel Filtration ................................ ................................ ........................ 117 Crystallogenesis ................................ ................................ ............................. 118 Structure Solution ................................ ................................ ........................... 120 Phenol Binding Studies ................................ ................................ .................. 121 4 CH ARACTERIZATION OF Lactococcus lactis CV56 DIHYDROOROTATE DEHYDROGENASES IA AND IB ................................ ................................ ......... 122 Introduction and Rationale ................................ ................................ .................... 122 Background and Experimental Design ................................ ................................ .. 124 DHOD ase Function ................................ ................................ ........................ 124 Catalytic Cycle and Electron Acceptors ................................ .......................... 125 Structural Similarities and Potential as an Alkene Reductase ........................ 127 Results and Discussion ................................ ................................ ......................... 129 Lactococcus lactis CV56 DHOD 1A Cloning, Expression and Purification ..... 129 Lactococcus lactis CV56 DHOD 1B ................................ ............................... 133 Introduction ................................ ................................ .............................. 133
7 Cloning, e xpression and p urification ................................ ........................ 134 Crystallization and Structure of L. lactis CV56 DHOD 1B ............................... 139 Substrate Specificity and Mutagenesis Targets ................................ .............. 145 Characterization of Ile74Glu and Ile74Asp ................................ ..................... 147 Characterization of the G75N and I74E/G75N Mutants ................................ .. 148 Steady state Kinetics and Catalytic Efficiency ................................ ................ 149 Wild type ................................ ................................ ................................ .. 150 Ile74Asp m utant ................................ ................................ ....................... 150 Ile74Glu m utant ................................ ................................ ........................ 151 Gly75Asn m ut ant ................................ ................................ ..................... 151 Ile74Glu/ Gly75Asn m utant ................................ ................................ ....... 151 Residues 74 and 75 can Disrupt Electron Transfer to the subunit .............. 152 Kinetic studies of oxidase r eactions ................................ ......................... 152 Choosing between NAD + and oxygen: studies of uncoupling .................. 156 Conclusion and Future Directions ................................ ................................ ......... 161 Experimental Procedures ................................ ................................ ...................... 167 Site Directed Mutagenesis ................................ ................................ ............. 167 Protein Expression and P urification ................................ ................................ 168 Crystallographic Studies ................................ ................................ ................. 170 St ructure Solution ................................ ................................ ........................... 171 Steady State Kinetics ................................ ................................ ..................... 171 APPENDIX A SELECTED GC MS DATA ................................ ................................ ................... 173 B PLASMID CONSTRUCTS USED ................................ ................................ ......... 176 C PURIFIED OYE1 SDS PAGE ANALYSIS ................................ ............................. 179 D GEL FILTRATION OF PROTEIN LADDER AND LINEAR REGRESSION ........... 180 LIST OF REFERENCES ................................ ................................ ............................. 181 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 190
8 LIST OF TABLES Table page 1 1 Reductions of 2 and 3 alkyl substituted 2 cyclohexenones ................................ 38 2 1 X ray Crystallographic Data Collection and Refinement Statistics ...................... 59 2 2 X ray Crystallographic Data Collection and Refinement Statistics ...................... 61 2 3 X ray Crystall ographic Data Collection and Refinement Statistics ...................... 67 2 4 X ray Crystallographic Data Collection and Refinement Statistics ...................... 69 2 5 X ray Crystallographic Data Collection and Refinement Statistics ...................... 71 2 6 X ray Crystallographic Data Collection and Refinement Statistics ...................... 75 2 7 Mutagenic primers utilized for constructing the enzymes used for X ray crystallography. ................................ ................................ ................................ .. 83 3 1 X ray Crystallographic Data Collection and Refinement Statistics .................... 106 4 1 X ray Crystallographic Data Collection and Refinement Statistics .................... 144 4 2 Steady sta te parameters of wild type and mutant DHOD 1B enzymes ............ 149 4 3 Steady state parameters of wild type and mutant DHOD 1B enzymes ............ 154 4 4 Catalytic Properties of WT and DHOD 1B mutants. ................................ ......... 162 4 5 Mutagenic primers utilized for constructing the mutant enzymes. .................... 167
9 LIST OF FIGURES Figure page 1 1 Respiratory chain described by Warburg and Christian. ................................ ..... 18 1 2 Spectra of free oxidized and complexed OYE1.. ................................ ................ 20 1 3 The overall structure of OYE1. ................................ ................................ ........... 23 1 4 The FMN environment and direct hydrogen bonding network. ........................... 25 1 5 The active site of OYE1 and interactions with the ligand p HBA.. ...................... 27 1 6 The catalytic cycle of OYE1 shows the reductive and oxidative half reactions.. ................................ ................................ ................................ ........... 28 1 7 The kinetic mechanism of OYE1. S ox is oxidized substrate and S red reduced. ... 29 1 8 The simplified chemical mechanism of OYE1.. ................................ ................... 30 1 9 Observed dismutation reaction products of ODE and 2 cyclohexenone. ............ 31 1 10 Dismutation reaction ................................ ................................ ........................... 32 1 11 2 and 3 alkyl substituted 2 cyclohexenones tested in OYE1 reducti ons. ........... 37 1 12 Stereo view of the modeled pseudo Michaelis complex of 3 ehtyl 2 cyclohexenone and OYE1. ................................ ................................ ................. 39 1 13 The unexpected setereochemical outcomes of OYE1 mediated reductions of (R) and (S) carvone. ................................ ................................ ........................... 41 2 1 The alternative binding mode hypothesis. ................................ .......................... 42 2 2 Stereocomplementarity in OYE1 variants.. ................................ ......................... 45 2 3 The results of (S) carvone reductions.. ................................ ............................... 47 2 4 The results of (R) carvone reductions.. ................................ .............................. 48 2 5 This scheme shows the geometric constraints for (S) and (R) carvone oxidation in the reductive half reaction ................................ ............................... 50 2 6 Charge transfer complex formation activities of Trp116 with (S) carvone (light bars) and (R) carvone (dark bars). ................................ ................................ ..... 51 2 7 Panels A and B show results from OYE1 and ligands co crystallization. ............ 53
10 2 8 Yellow and green crystal forms. ................................ ................................ .......... 55 2 9 The normal binding mode. ................................ ................................ .................. 56 2 10 OYE1 Trp116Leu ................................ ................................ ............................... 57 2 11 OYE1 Trp116Ile ................................ ................................ ................................ .. 62 2 12 De tailed view of Trp116Ile. ................................ ................................ ................ 63 2 13 OYE1 Trp116Ala. ................................ ................................ ............................... 65 2 14 Schematic representation of (R) carvone formation from (+) dihydrocarvone, via an enzymatic oxidation (dehydrogenation). ................................ .................. 66 2 15 OYE1 Trp116Val. ................................ ................................ ............................... 68 2 16 Crystal Structure of the Trp116Thr mutant. ................................ ........................ 70 2 17 OYE1 Trp116Gln. ................................ ................................ ............................... 74 2 18 Alignment of all four binding modes.. ................................ ................................ .. 77 2 19 116 side chain distance and th e fraction of flipped product ................................ ................................ .................. 78 2 20 Hydrophobic pockets created in OYE1. ................................ .............................. 80 2 21 Temperature program for GC MS analysis on DB 17 cloumn. ........................... 82 3 1 Reductions of Baylis Hillman Adducts by Alkene Reductases ........................... 90 3 2 Partial sequence conservation in s even OYE family representatives. ................ 91 3 3 Comparison of OYE1 and OYE2.6 stereoselectivity. ................................ .......... 91 3 4 Superposition of the structures of OYE2.6 (cyan) and OYE1 (orange). .............. 93 3 5 The proposed OYE2.6 dimer. The 2 fold molecular axis is indicated by a dashed line.. ................................ ................................ ................................ ....... 95 3 6 OYE2.6 crystal lattice. ................................ ................................ ........................ 96 3 7 OYE1 crystal lattice. ................................ ................................ ........................... 97 3 8 The FMN biding environment and direct interactions. Hydrogen bonds are shown as black dashes. ................................ ................................ ..................... 98 3 9 Complex formation between OYE2.6 and p CP ................................ ................. 99
11 3 10 Complex formation with p HBA. ................................ ................................ ........ 100 3 11 The active site comparison of OYE2.6 and OYE1. ................................ ........... 101 3 12 OYE2.6 malonate complex. ................................ ................................ ............. 102 3 13 OYE2.6 nicotinamide complex. ................................ ................................ ........ 103 3 14 Crystal structure of OYE2.6 in complex with p chlorophenol. ........................... 105 3 15 Structures of the three Baylis Hillman adducts tested. ................................ ..... 107 3 16 Mutations of positio ns 188 and 191 ................................ ................................ 109 3 17 Effects of mutations of amino acid Thr35. ................................ ........................ 110 3 18 Mutations of positions 113 and 116. ................................ ................................ 111 3 19 Comparison of OYE 2.6 and S. pastorianus OYE1 complexes. ....................... 114 3 20 Elution profile of OYE2.6.. ................................ ................................ ................ 118 3 21 X ray diffraction images of crystals. ................................ ................................ .. 119 4 1 The pyrimidine de novo biosynthesis. ................................ ............................... 125 4 2 The catalytic cycle of DHOD.. ................................ ................................ ........... 126 4 3 Active site alignment of OYE1 and DHOD 1A.. ................................ ................ 129 4 4 The primers utilized to clone the pydA gene from L. lactis CV56. ..................... 130 4 5 SDS PAGE analysis of E. coli cells carrying pYAP 3.. ................................ ..... 131 4 6 Region of L. lactis CV56 genome containing the pyr Z and pyr D genes.. .......... 135 4 7 Absorption spectrum of DHOD 1A.. ................................ ................................ .. 136 4 8 Elution profile of the purified DHOD 1B. The enzyme elutes at 11.4 mL primarily. ................................ ................................ ................................ ........... 137 4 9 The absorption spectrum of the purified DHOD 1B. ................................ ......... 138 4 10 Wild type DHOD 1B crystal optimization ................................ .......................... 140 4 11 X ray diffraction pattern at different temperatures. ................................ ........... 141 4 12 The biological heterotetramer of DHOD 1B.. ................................ .................... 1 42 4 13 The conformational changes in the active site loop. ................................ ......... 143
12 4 14 Potential substrates for DHOD 1B that would complement the substrate range in our existing library of OYE like alkene reductases. ............................ 145 4 15 Active site comparison of DPD and DHOD 1B ................................ ................ 147 4 16 Conc entration dependence of measured initial velocities V o for w.t. ................ 150 4 17 Conc entration dependence of measured initial velociti es V o for the I74D mutant ................................ ................................ ................................ .............. 150 4 18 Conc entration dependence of measured initial velocities V o for the G75N mutant. ................................ ................................ ................................ ............. 151 4 19 Conc entration dependence of measured initial velocities V o for the I74E/G75N double mutant. ................................ ................................ ............... 152 4 20 The possible fates of electrons in DHOD 1B.. ................................ .................. 153 4 21 Conc entration dependence of measured initial velocities V o for wild type DHOD1B. ................................ ................................ ................................ ......... 154 4 22 Conc entration dependence of measured initial velocities V o for the I74D mutant. ................................ ................................ ................................ ............. 155 4 23 Conc entration dependence of measured initial velociti es V o for the I74E mutant. ................................ ................................ ................................ ............. 155 4 24 Conc entration dependence of measured initial velocities V o for the G75N mutant. ................................ ................................ ................................ ............. 156 4 25 Uncoupling by wild type DHOD 1B. ................................ ................................ .. 157 4 26 Uncoupling by I74E DHOD 1B.. ................................ ................................ ........ 158 4 27 Uncoupling by I74D DHOD 1B.. ................................ ................................ ....... 159 4 28 Uncoupling by G75N DHOD 1B.. ................................ ................................ ...... 160 4 29 Uncoupling by I74E/G75N DHOD 1B.. ................................ ............................. 160 4 30 Absorption spectra of the five DHOD 1B enzymes studied.. ............................ 163 4 3 2 SDS PAGE analysis of all DHOD 1B enzymes purified in this study. ............... 170 A 1 Authentic standard (Acros) of (+) dihydrocarvone containing a mixture of isomers.. ................................ ................................ ................................ ........... 173 A 2 Chromatogram of authentic sample of (R) carvone (Acros Organics). ............. 173
13 A 3 GC MS analysis of OY E1 catalyzed (R) carvone reduction.. ........................... 174 A 4 GC MS analysis of W116A OYE1 catalyzed (R) carvone reduction.. ............... 174 A 5 GC MS analysis of OYE1 catalyzed (S) carvone reduction.. ............................ 175 A 6 GC MS analysis of W116I OYE1 catalyzed (S) carvone reduction.. ............... 175 B 1 Plasmid map of all pET OYE W116X used for structural studies. .................... 176 B 2 Plasmid construct of pYAP 7. DHOD 1A with C terminal His tag. .................... 176 B 3 pYAP 5 encodes pyrD with C terminal His tag. ................................ ................ 177 B 4 pYAP 6 encodes pyrZ with N terminal His tag. ................................ ................ 177 B 5 Basic plasmid construct for pYAP 8, pYAP 9, pYAP 1 2 and pYAP 13. Each mutant was created using primers listed in Table 4 4. ................................ ...... 178 C 1 Typical purity of OYE1 enzymes used for X ray crystallography.. .................... 179 D 1 Elution profile of the protein standards used for calibration. ............................. 180 D 2 The linear fit of the molecular weight (log M.W.) versus the relative elution volumes plot. ................................ ................................ ................................ .... 180
14 LIST OF ABBREVIATIONS Abs Absorbance DHO Dihydroorotate DHOD Dihydroorotate dehydrogenase DHOD Dihydroorotate dehydrogenase DPD Dihydropyrimidine dehydrogenase FAD Oxidized flavin adenine dinucleotide FADH 2 Reduced flavin adenine dinucleotide FMN Oxidized flavin mononucleotide FMNH 2 Reduced flavin mononucleotide GC MS Gas chromatography mass spectrometry GDH Glucose dehydrogenase M Mole/liter NAD(P) + Oxidized nicotinamide adenine dinucleotide NAD(P)H Reduced nicotinamide a denine dinucleotide OYE Old yellow enzyme p HBA para Hydroxybenzaldehyde UV Vis Ultra violet and visible light
15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy UNCOVERING STEREOCOMPLEMENTARITY IN TWO ALKENE REDUCTASES OF THE OLD YELLOW ENZYME FAMILY AND THE DIFFERENT FATES OF ELECTRONS IN THE SECOND HALF-REACTION OF DIHYDROOROTATE DEHYDROGENASE 1B By Yuri Alexey Andreiw Pompeu August 2013 Chair: Jon D. Stewart Major: Chemistry The first portion of this work reports the efforts towards uncovering the factors controlling stereoselectivity in two alkene reductases belonging to the Old Yellow Enzyme family. We demonstrate, through site-saturation mutagenesis the impact that each of the 20 amino acids has on the stereochemical outcome of OYE1 mediated carvone reductions. Furthermore, in order to explain the stereochemical diversity observed we investigated, at an atomic level, the most interesting OYE1 variants using high resolution single crystal X-ray crystallography. The structure elucidation of OYE2.6, an OYE1 homolog that displays desirable physical and catalytic properties, is also presented. In addition to the determination of its three-dimensional and quaternary structures, catalytic characterization was done by mutating key active-site residues. Taken together, these results provide significant insight for future protein engineering projects as well as some guidance towards understanding how, in some cases, minute changes can lead to big differences in the stereoselectivity of these alkene reductases. The second part of this dissertation describes how, in an attempt to modify substrate specifity in the enzyme DHOD 1B, we uncover ed the importance of two amino
16 acids Ile74 and Gly75, to the el e ctron transfer mechanism of the enzyme. The design expression, kinetic characterization as well as the structural elucidation of the wild type enzyme is described in detail. In essence, our results suggest that modifications of subunit of L lactis CV56 DHOD1B can impact the example, mutation of Ile 74 to Glu completely abolished catalytic activity towards NAD + while retaining oxidase activity that was comparable to t hat of the wild type enzyme.
17 CHAPTER 1 OYE1 AND BIOCATALY SI S The Isolation and Discovery of OYE 1 Old yellow enzyme (OYE, EC 126.96.36.199), was the first enzyme found to contain flavin as a prosthetic group required for cataly si s and has served as a model for many flavin dependent oxidoreductases. In recent years the catalytic activity of this enzyme and an entire subset of related homologs have appeared in the literature as valuable catalysts with industrial applicat ions 1 bottom yeast Saccharomyces carl sbergen si s (so named after the Danish brewery, Carlsberg, where it was first isolated and later reclas si fied as Saccharomyces pastorianus ) b y Warburg and Christian in 1932 2 Their original aim was to elucidate the nature of biological oxidations in yeast. In the course of their work a yellow protein was gelbe ferment llow ferment) This yellow protein was present in appreciable amounts in the lysate and it was noted that addition of the yellow lysate would facilitate the oxidation of glucose 6 phosphate in the Christian ultimately described a respiratory chain con si sting of the oxidation of glucose 6 phosphate in the presence of NADP + /NADPH and glu cose 6 phosphate dehydrogenase ( Figure 1.1). In 1938, Haas independen tly reported the discovery of a new yellow protein homolog in yeast 3 which then led to the de si gnation of Warbur g and si sted and the original enzyme in the gelbe ferment is now called Old yellow enzyme, or si mply OYE. Remarkably, despite the overabundance of kinetic and structural data about OYE1 and many other homologs
18 that has been accumulated in the last 80 years, the natural substrate and the exact function of OYEs remains a topic for debates among researchers. Figure 1 1 Respiratory chain described by Warburg and Christian. Glucose 6 phosphate is oxidized by glucose 6 phosphate dehydrogenase (zwischenferment) in the presence of NADP + (coferment). Old Yellow Enzyme (gelbe ferment) serves to regenerate NADP + by facilitating a 2 electron transfer from NADPH to molecular oxygen. History of Purification and Spectral Characterization Following the work of Warburg and Christian, Theorell and keson, developed a more robust isolation and purification method which allowed for more accurate and therefore conclu si ve studies of Old Yellow Enzyme 4 Their procedure con si sted of the ly si s of dry yeast cells, followed by several precipitation steps u si ng lead a cetate, cold acetone, ethanol and ammonium sulfate and adsorption onto a 4% calcium phosphate gel. The eluted enzyme could be crystallized through dialy si s u si ng a buffer with gradually increa si ng amounts of ammonium sulfate. Through analy si s of the crysta lline OYE it was determined that the enzyme had a molecular weight of approximately
19 100,000 Da and 2 FMN molecules at a ratio of 1/50,000 Da, suggesting its existence as a functional dimer. The purity of this preparation was assessed u si spe ctroscopic characteristics in the ultraviolet and vi si ble range. It was reported that the enzyme had absorption maxima at 280, 383 and 464 nm. The ratio of A 383 : A 464 was 0.98 and the ratio of A 280 : A 464 was 11.5. The molar extinction coefficient 464 for th e bound flavin mononucleotide had a reported value of 10,600 M 1 cm 1 In 1969, expanding on the work of Theorell, Matthews and Massey were able to improve on the purification of the O ld Yellow Enzyme and carried out an even more detailed characterization of the enzyme 5 They made the observation that the enzyme preparation appeared green and that upon reduction u si ng sodium dithionite the enzyme solution became colorless. Additionally, when the reduced enzyme was dialyzed exten si vely it would return to its native yellow color upon reoxidation Remarkably, when t he oxidized yellow enzyme was mixed with the dialysate it regained its green color oxidized enzyme and that it had a diminished affinity for the reduced enzyme and could therefore be removed by dialy si s. Several attempts were made to identi i t was concluded that the molecule was aromatic with an ionization behavior si milar to that of a phenol hydroxyl group shown to be a substrate for xanthine oxidase, albei UV Vis spectrum as a broad absorption maximum in the 600 680 nm range and it was hypothe si zed that it was the result of the formation of a charge transfer complex ed flavin mononucleotide and the phenolic ligand as it had been observed previously in other flavoproteins 6 Figure1 .2 shows a typical spectrum of
20 oxidized Old Yellow Enzyme in its free form and in its charge form. Figure 1 2 Spectra of free oxidized and complexed OYE1. The solid lined spectrum ( ) is that of free oxidized OYE1. The dashed lined spectrum ( ) is that of OYE1 in 80 M 4 c h l orophenol, a charge transfer complex forming ligand. In a subsequent study, Matthews and Massey were able to isolate and para hydroxybenzaldehyde. This led to the conclu si on that the green appearance of the enzyme was indeed a result of the formation of a charge transfer complex 7 In such complex there are direct interactions between the systems of the electron rich ligand and the electron deficient oxidized flavin. It was also shown that this and other related phenols were capable of I n 1975 Abramovitz and Massey undertook the identification and characterization of many ligands capable of forming charge transfer complexes with Old Yel low Enzyme. This work showed that charge transfer complexes had ch aracteristic
21 absorption maxima in the range of 550 680 nm that varied depending on both the identity of the ligand and the redox potential of the bound flavin. Another important observation was the fact that the observed pH dependence of ligand dissociation strongly suggested that the phenolate form, rather than the conjugate acid form, was respon si ble for complexation 8 Shortly aft er, Abramovitz and Massey exploited phenol binding to develop an affinity purification technique u si ng a phenol functionalized re si n for the sole chromatographic step 9 This further facilitated a s i mpler and more effective isolation of the enzyme from its native host in greater yield. The method con si sted of allowing the oxidized enzyme to bind to the phenol re si n followed by elution through the use of a reducing buffer. The reducing agent used was sodium dithionite which quickly reduces si n is greatly diminished cau si ng the enzyme to elute off the chromatographic column. In 1991 the gene for Old Yellow Enzyme (now termed OYE1 ) was finally cloned and overexpressed in E scherichia coli 10 Comparison of the overexpressed protein with non recombinant en zyme preparations led to the discovery that several OYE isozymes 11 The cloned gene also led to a homogenous source which allowed the enzy unambiguous characterization and later proved to be suitable for structural determination. The Structure of OYE1 Even though some structural information on OYE1 had been inferred based on exten si ve studies of the environment around the bound FMN 12 the complete three dimen si onal st ructure of the protein was highly anticipated. Previous attempts at obtaining OYE1 crystals suitable for structural investigation through X ray
22 crystallography, had largely failed. The only crystal forms that could be obtained were of partially proteolytic ally cleaved enzyme. This was due to the fact that enzyme isolated which formed mixtures of homo and heterodimers 13 With the cloning of the gene encoding OYE1 by Saito et al 10 larger amounts of recombinant OYE1 that appeared electrophoretically ho mogenous could be purified from E. coli cells. Homogenous OYE1 three dimen si onal structure 14 Overall S tructure The protein contained a si ngle domain con si sting of a parallel, eight stranded barrel si milar to that of triosephosphate isomerase (TIM) 15 The largely hydrophobic active si te is acces si ble to solvent via a deep cleft. This cleft is, in part, obscured by a flexible loop that has implications in the dynam ic binding of the NADPH substrate FMN binds to OYE1 at the carboxy terminal end of the barrel in an extended conformation. The isoalloxazine ring lies perpendicular to the barrel axis with the phosphoryl group held between strand no terminus of helix D cau si ng the flavin moiety being placed near the center of the barrel (Fig. 1 3 ). The FMN is mostly buried, surrounded by the carboxy terminal loops which cover the end of the barrel and inhibit the access of solvent to the interior of the barrel. Fox and Karplus also solved the crystal structure of OYE1 with bound para hydroxybenza ldehyde and this structure shed light onto the phy si cal nature of previously discovered charge transfer interactions 7 The 2 resolution was sufficient to confidently define the orientation of the ligand, and its electron den si ty was quite clear after refinement.
23 Figure 1 3 The overall structur e of OYE1 A si ngle protomer of OYE1 is depicted in cartoon representation and the flavin mononucleotide cofactor is shown in ball and stick representation. The inhibitor p hydroxybenzaldehyde is shown in stick representation. This Figure was generated u s i ng the atomic coordinates of PDB code1OYB The phenol ring displaces a chloride ion bound in the active si te of free OYE1 and is stacked above the si face of the flavin ring w ith the planes of the two rings approximately parallel The orientation of the molecule places the phenolate oxygen above the C2 atom of the flavin, interacting with H is1 91 and Asn194 and the aldehyde carbonyl hydrogen bonded to the hydroxyl of Tyr375. The close phy si cal interaction between ligand and FMN in addition to being geomet rically po si tioned to participate in helped explain the nature of charge transfer complex through the soaking of OYE1 crystals in a solution cont aining an NADPH analog. It was
24 concluded that the binding mode of NADH and NADPH cofactors is largely dictated by hydrogen bonding. These interaction s placed carbon C4 o f the nicotinamide directly above N5 of the flavin, in a favorable po si tion for a hydride transfer. The remaining portion of the cofactor appeared to be of less importance and this observation helped explain the fact that OYE1 can be reduced by both NADH a nd NADPH, although a preference for the phosphorylated cofactor is observed. T he FMN Binding E nvironment The tightly bound flavin mononucleotide interacts with amino acids that line the interior of the parallel 8 stranded barrel, many of which are highly conserved amongst member s of the OYE family 14 16 The isoalloxazine moiety of the flavin is involved in di rect hydrogen bonds with the si de chains of Thr37, Gln114, Arg243 and the main chain nitrogen of Gly72. All of these interactions are suspected to have a direct impact on reactivity and or spectral properties of the FMN cofactor in OYE1 and other related flavoproteins 17 Figure 1 4 illustrates the flavin environment in OYE1. Xu et al 17c carried out a more detailed study on the role of Thr37 in flavin reactivity in OYE1. In the crysta l structure, Thr 37 is hydrogen bonded with the C4 carbonyl o xygen of FMN through i ts si de chain hydroxyl group Such hydrogen bonding is likely to play a role in controlling the properties of the enzyme flavin. To investigate the pos si ble role of the hydrogen bond in cataly si s, a Thr to Ala (T37A) mutation of OYE1 was constructed and expressed. The mutation had different effects on the two separate oxidative and reductive half reactions of the enzyme catalytic cycle (this concept will be further exp lored in the following sections).
25 Figure 1 4. The FMN environment and direct hydrogen bonding network. Figure was generate d u si ng the crystallographic model of OYE1 (PDB code 4GBU). Hydrogen bonds are indicated as dashed lines. The properties of the muta nt enzyme were examined in detail through direct measurement of redox potential, binding of phenolic ligands, rapid reaction studies of both the reductive and oxidative half reactions, and catalytic turnover The mutant enzyme has enhanced activity in the oxidative half reaction (loss of electrons by FMNH 2 ) but the reductive (gain of electrons by FMN) half reaction is slowed down by more than one order of magnitude. The peaks of the absorption spectra for enzyme bound with phenolic compounds are shifted toward shorter wavelengths than those of wild type OYE1, con si stent with its lower redox potential. It is suggested that Thr37 in the wild type OYE1 increases the redox potential of the enzyme by stabilizing the negative charge of the reduced flavin through hydrogen bonding with it.
26 The active si te of OYE1 The active si te of OYE1 is si tuated directly ab ove the si face of the FMN. It is relatively small with a volume of approximately 460 3 (calculated u si ng the CASTp server 18 ). It is connected to bulk solvent by a small channel that is roughly 8 wide and 10 deep. It is also a predominantly hydrophobic cavity as it is defined by amino acids Thr37, Tyr82, Trp116, Leu118, Phe123, Tyr196, Phe250, Pro295, Phe296 and Tyr375. Two re si dues posses si ng more polar properties are found in the back wall of the active si te namely His 191 and Asn194 Both si de chains of His191 and Asn194 are po si tioned directly above the pyrimidine portion of the flavin ring, and are hydrogen bond donors ( with nearly ideal geometry ) for carbonyl and or phenol(ate) groups of ligands. In fact, as observed in the crystal structure, the hydroxyl group of p HBA is in excellent po si tion to form hydrogen bonds with of His191 and of Asn194 ( Figure 1 5) These interactions help place the ligand s meta carbon directly above the N5 of flavin in a spatial arrangement favora ble for a hydride transfer. These observations, along with the NADPH analog binding orientation (discussed above), led to the belief that this binding mode is also shared by other substrates of OYE1. Also evident in the crys tal structure t he aldehyde group of p HBA forms a hydrogen bond with the si de chain hydroxyl of Tyr375, which moved si gnificantly in comparison to the unliganded OYE1 structure. Given their po si tion in the active si te and the evidence for direct interacti ons with ligands, His191 and Asn194 were thought to be of great relevance in the mechanism of OYE1. In 1998, Brown et al. carried out studies specific OYE1 mutants in order to evaluate the roles played by these two amino acids
27 Figure 1 5. The active si te of OYE1 and interactions with the ligand p HBA. Hydrogen bonds are indicated by dashed lines. The FMN cofactor is shown in ball and stick, the amino acids and p HBA as sticks. The Figure was generated u si ng coordinates from PDB code 1OYB. Mutations of H is 191 to Asn, Asn194 to His, and a double mutation, H191N/N194H, were made within OYE1. The N194H protein was expressed at very low levels and could not be purified. The H191N mutant and the H191N/N194H double mutant were successfully expressed and purifie d and their properties were evaluated. T he two mutant forms had the expected effect on phenolic ligand binding, i.e decreased binding affinity and decreased charge transfer absorbance. In the case of para chlorophenol, the K D value increased from 1 M in the wild type to 2400 M in the H191N mutant It had been previously established that the formation of charge transfer complexes occur via the phenolate form of ligands 8 thus i t was concluded th at the mutations caused a decreas ed stabilization of the phenolate anion Reduction of the H191N mutant enzyme by NADPH was si milar to that of OYE1, but the reduction rate constant for NADH was greatly decreased. The double mutant enzyme had an increased rate constant for
28 reduction by NADPH, but the reduction rate constant with NADH was lower by a factor of 15. The reactivity of OYE1 and the mutant enzymes with oxygen was si milar, but the reactivity of 2 cyclohexenone was greatly decreased by all of the mutations. The crystal structures of the two mutant forms showed only minor changes from that of the wild type enzyme. The Catalytic Cycle and Mechanism of OYE1 Si nce its early discovery in 1932, OYE1 has been known to catalyze the oxidation of NADPH in the presence of molecular oxygen (O 2 ) as an electron acceptor Th e discovery of other substrates that could serve as electron acceptors in addition to vast spectral knowledge allowed for the conception of a clear model for the catalytic cycle of the enzyme ( Figure 1 6) 11 19 Figure 1 6. The catalytic cycle of OYE1 shows the reductive and oxidative half reactions. Oxygen and the model substrate 2 cyclohexenone are shown as pos si ble electrons acceptors. The accepted reaction cycle suggests a ping pong mechanism. In fact, data for kinetic assays with different substrates have always been reported to yield sets of
29 parallel lines in double reciprocal plots. Additionally, the initial observation that the nicotinamide binds to the same si te as substrate analogs do also serve d as substantiating evidence for a clas si c p ing p ong mechanism ( Figure 1 7) Figure 1 7. The kinetic mechanism of OYE1. S ox is ox idized substrate and S red reduced. By the mid about the roles of several of the re si dues forming the active si te of OYE1. A surplus of kinetic and spectroscopic data derived not only from the wild type enzyme but also from several m utants had been reported as well In a 1995 study Vaz et al. showed through deuterium labeling experiments that OYE1 catalyzed the net trans addition of H 2 unsaturated aldehyde, namely cinnamaldehyde 19 These findings and the existence of an atomic model of the enzyme enabled the propo si tion of detailed chemical mechanism of the enzyme. In the reductive half reaction, reduced nicotinamide cofactors bind OYE1. In the active si te two hydrogen bonds formed by His191, Asn194 and the nicotinamide amide carbonyl place its carbon C4 directly above N5 of the flavin favor ing a hydride transfer. After completion of the reductive half reaction, oxidized cofactor dissociates and oxidized substrates can enter the ac tive si te to react. Si milarly, t he substrate is properly oriented primarily through the hydrogen bonds formed by its key role in substrate orientation, the polar nature of these hydrogen bonds
30 accentuated by ability of carrying a po si tive charged is thought to aid in stabi liz ation of the enolate form of the substrate. This stabilization would activate the substrate by increa si ng the carbon electron deficien cy making it more susceptible to a hydride attack from the flavin. carbon would build el ectron den si carbon and its protonation would complete the net trans addition of H 2 Tyr196 is p o si tioned carbon at a distance of 3.41 , making it an obvious candidate for the protonation step. Results show that mutation of Tyr196 to Phe severely impairs the rate of the oxidative half reaction in OYE1 20 with k cat being decreased by nearly 6 orders of magnitude with 2 cyclohexenone Figure 1 8 shows the chemical mechanism of the oxidative half reaction for OYE1. Figure 1 8 The si mplified chemical mechanism of OYE1. The illustration shows the oxidative half reaction, wh erein fully reduced flavin (FMNH 2 ) transfers a carbon of 2 cyclohex enone and protonation of the carbon occurs concomitantly.
31 Aromatization of Cyclic Enones: A Dismutation Reaction in OYE1 unsaturated ketones could be reduced by OYE1, the se arch for novel and more diverse substrates began. Curiously it was observed that when OYE1 was allowed to react with certain cyclic 2 ene 1 ones in the absence of an electron donor such as NADPH a long wavelength absorption band develop ed in the UV Vis spectrum of the enzyme. This band had absorption maxima values in the 550 680 nm range and it was a rather broad peak strongly resembling those previously seen in charge transfer complex spectra. In 1995 Vaz et al carried out a more detailed investigation of this phenomenon and reported a new reaction pathway in OYE1 19 Figure 1 9 shows two of the reactions observed in th e study. Figure 1 9. Observed dismutatio n reaction products of ODE and 2 cyclohexenone. In this dismutation pathway, the reductive half reaction is accomplished by abstracting electrons from the enone to generate FMNH 2 the reduced form of the enzyme. This results in a two electron oxid ation of the enone creating a second C C double bond. In the case of si x membered ring enones, the oxidized molecule is a 2,5
32 diene 1 one that can undergo a thermodynamically favorable tautomerization to gain aromaticity and form a phenol ( Figure 1 10 C ) In the oxidative half reaction, t he enzyme can reduce a second molecule of substrate to return to its original oxidized state thus completing the catalytic cycle. One consequence of this proposed mechanism is that in each half reaction the substrate must bind in different orientations in the active si te, more specifically in different orientations relative to N5 of the flavin. Figure s 1 10 A,B show a rationalization of the different binding modes in the reductive and oxidative half reactions. Figure 1 10 Dismutation reactions. Panels A and B illustrate the proposed binding modes of ODE in the reductive and oxidative half reactions, respectively. Panel C shows the proposed si mplified mechanism for the formation of the phenolic inhibitor from a cyclic enone
33 Biocataly si s and the Potential of OYE1 as a Suitable Catalyst Expectations and Limitations Over the past few years the use of enzymes as biocatalysts for the prepara tion of enantiopure active compounds has become an established manufacturing process in the speciality (applied fine chemicals) and pharmaceutical industry. The employment of biocatalysts is attractive for synthetic organic chemists as a consequence of the controlled insertion of chirality and selectivity with regard to regio and stereochem istry in molecules and the si mple usage in technical production units. Thus many researchers are discovering the advantages in utiliz ing these natural catalysts for producing optically active, higher value molecules. In particular as multiple process steps can be circumvented or combined the natural cat alysts are contributing to more e conomic industrial processes Natural evolution has provided a number of enzymes with high enantio and po si tional (regio) selectivitie s suitable for the synthe si s of e nantiom erically pure compounds with few byproducts 21 making b iocataly si s an attractive option in synthe si s In fact, e nzymes have been exploited in organic chemistry for more than o ne hundred years with a number of notable industrial examples 22 and references therein Therefore, the use of enzymes for the production of high value pharmaceuticals and fine chemicals is at the fo refront of the quest for mild and ecologically sustainable processes 23 B iocataly si s has, however, traditionally suffered in many cases from limitations associated with narrow substrate scope, incomplete enantioselectivities and i nsufficient robustness. I n reality the range of chemical reactions open to biocataly si s is limited, due to the disadvantageous reaction conditions in an aqueous environment near phy si ological conditions and enzyme usual limited reactivity towards more complex
34 larger chemical compounds Nature has evolved a pool of catalysts which operat e most effectively under phy si ological conditions and on a narrow range of natural substrates. While these characteristics are optimal for sustaining life, naturally occurring enzymes are often not suitable for biocatalytic processes without further tailor ing or rede si gn The pioneering work of Frances Arnold and Pim Stemmer on directed evolution set the ground for implementing evolution of enzymes for the synthe si s of technological useful compounds 24 O YEs as Biocatalyst s Asymmetric cataly si s plays a pivotal role in modern synthetic organic chemistry and the increa si ng demand for optically pure building blocks has fueled interest in identifying and creating new stereoselective catalysts In particular, the use of alkene reductases is regarded as an attractive option si nce the reduction of a C C double bond generates two adjacent sp 3 centers and opens the pos si bility for the creation of two distinct chiral centers in a si ngle step. Existing synthetic methods of asymmetric hydrogenations currently involv e the use of precious metals in conjunction with chiral phosphines In addition, the o rganocatalytic hydrogenation of cyclic enones to allow the synthe si s of a wider range of enantiomerically pure organic compounds has been developed as well 25 However, enzymes continue to be competitive catalysts in asymme tric syntheses due to their low costs, ability to manipulate the substrate specificity through si te directed mutagene si s) and the need for typically mild operating conditions such as atmospheric pressure, mild temperature s and pH The O ld Y ellow E nzyme family is a clas si c example of enzyme s able to catalyze the reduction of certain alkenes to produce many commer cially useful substrates and synthons. Si nce the discovery of its first family member nearly 80 years ago, these
35 flavin dependent enzymes have been shown to catalyze the reduction of the C C double unsaturated ketones, aldehydes, nitro alkenes carboxylic acids and esters 26 27 28 29 The potential applicability of OYEs has also been tested in improving 2 amino acids 30 A s mentioned before, there are some limitations on the applicability of enzymes for exten si ve organic synthe si s and OYEs s hare many of these inadequacies. Firstly, most of the family members, including OYE1, display optimal pH values for activity near phy si ological values and the same trend is followed in their optimal temperature pr ofile, which is around 37 C. Two enzymes in this class namely Thermus scotoducts SA 10 chromate reductase (CrS) 31 and Thermoanaerobacter pseudoethanolicus E39 TOYE 32 have been reported to display higher tolerance s towards elevated temperature s Both enzymes maintained si gnificant activities in temperatures near 60 65 C Perhaps not surpri si ngly these values are close to the opti mal growth temperature of the ir respective host organisms Another asp ect that can keep biocatalysts from being suitable for a synthetic route is the lack of denaturation re si stance in the presence of organic solvents. It is often the case that solvent tol erance is related to temperature tolerance and researchers may turn to thermophilic organisms in search of a suitable biocatalyst. While it may be pos si ble to find an enzyme that catalyzes the reaction of interest in a thermophilic organism which tolerates harsh solvents and gives the expected optically pure enantiomer, the reality is that, frequently, a biocatalyst has several inherent limitations that keep it from being ideal.
36 In the case of OYEs, and the majority of naturally occurring enzymes, the subs trate range is rather narrow. Enzymes generally evolve to catalyze a specific reaction, it is therefore expected that a great preference towards a si ngle substrate be observed. Even though OYEs have been shown to work on a variety of substrates a more detailed analy si s indicates that most of the substrates share some common features and have limited diver si ty as a whole activated and are usually conjugated to carbonyls of ketones, aldehydes, esters and to a lesser extent acids. In addition, s ome nitro alkenes can serve as substrates 26 33 and t he reduction of nitrile containing alkenes rep orted although it is uncertain whether this enzyme is a true member of the OYE family 34 Substrate si ze is another limitation in this class of enzymes. As describe d in previous sections, the active si te of OYE1 is relatively small and se cluded by a narrow op ening which limits the nature and the number of substituents that can be found in substrates This observation generally holds true for other members that have been structurally characterized to date. Another challenge encountered with OYEs is related to their stereoselectivity. Owing to the generally conserved active si te architecture that is seen across several members of the family, OYEs tend to display little stereochemical diver si ty. The work of Bougioukou (PhD the si s and references therein) has show n that many OYE1 display nearly identical stereoselectivities towards a number of enones, enals and nitro alkenes. The lack of stereochemical diver si ty in OYE mediated alkene reductions is a natural consequence of the multipoint binding of the alkene subst rate and FMN along with the fixed location of the highly conserved proton donor Tyr196.
37 Earlier Studies of OYE1 Si ze and Stereocomplementarity Substrate si ze restriction in OYE1 is particularly apparent in the 3 al kyl substituted 2 cyclohexeno ne class of substrates ( Figure 1 11) In 2006, Swiderska and Stewart 28 conducted a systematic investigation that looked to shed some light on the enzyme active si te restrictions. A series of 2 and 3 alkyl substituted cyclohexenones were screened in OYE1 mediated reductions ( Figure 1 11). It was evident that that as the si ze of the alkyl substituents increased, their conver si ons efficiencies decreased. The general substrate preferences are: Me>Et> n Pr> i Pr>> n Bu for enones substituted at po si tion 3 and a si milar trend was seen for substit utions at po si tion 2 (Me>Et). Under the experimental conditions, 100% of 3 methyl 2 cyclohexenone was reduced while no reaction was detected with the substrate 3 butyl 2 cyclohexenone ( Table 1 1). Figure 1 11. 2 and 3 alkyl substituted 2 cyclohexenones tested in OYE1 reductions
38 Si nce the generation of chiral centers from the starting alkene requires substitutions at po si tions 2 and 3, improvement upon si ze limitations could po si tively applicability for a wider range of synthetic routes. Table 1 1. Reductions of 2 and 3 alkyl substituted 2 cyclohexenones 1 ND. Not detected. The results reported in this study provided some guidance for new investigations on the active si te of OYE1. They also helped draw attention to a particular location in the active si te architecture and its potential as a target for subsequent protein engin eering studies. In 2009, d uring their efforts to address the problem of low rates for larger substrates, Padhi et al unexpectedly uncovered a solution to a different limitation that affects OYEs, namely the lack of stereochemical diver si ty in these enzymes 3 5 Specifically, it was found that conservative changes to a si ngle re si due (Trp116) inverted the stereoselectivity of S. pastorianus OYE 1 for one alkene substrate (( S ) carvone) This open ed many pos si bilities for u si ng this enzyme and related biocataly sts in asymmetric synthe si s Ketone R group Conver si on (%) e.e. (%) Configuration 1a Me 100 94 S 1b Et 76 94 S 1c n Pr 25 89 S 1d i Pr 18 90 S 1e n Bu ND 1 3a Me 100 96 R 3b Et 16 94 R
39 It was suspected that diminished rates for OYE 1 mediated reductions of substrates such as 3 ethyl 2 cyclohexenone were a consequence of unfavorable steric interactions with active si te re si dues. The authors hypothe si zed that i ncrea si ng the active si te volume would alleviate these adverse contacts. To guide these efforts, they modeled a pos si ble Michaelis complex of this substrate ( Figure 1 1 2) and focused attention on re si dues within 5 of the bound substrate as candidates for replacement. Trp116 was selected si nce its si de chain is closest to the 3 substituents of 2 cyclohexenones and changes at this po si tion would be expected to have the largest effect on increa si ng the active si te volume Si te saturation mutagene si s utilizi ng homologous recombination in yeast cells was performed and 200 random colonies had their activities screened for 3 methyl 2 cyclohexenone reduction. Only 19 colonies encoded mutants that sustained at least 50% of the activity seen in wild type. DNA seque ncing revealed that si x encoded Trp at po si tion 116, si x encoded Phe, three encoded Ile ( on had an extra mutation), and there were si ngle examples of Leu, Met, and Tyr. Figure 1 12. Stereo view of the m odeled pseudo Michaelis complex of 3 ehtyl 2 cyclohexenone and OYE1. Reprinted from J. Am. Chem. Soc 2009, 131, 3271 3280 Copyright 2009 American Chemical Society.
40 Because Phe and Ile were found most commonly, these variants were chosen for further characterization and were pu rified as G ST fu si on protein s It was expected that OYE variants with smaller re si dues at po si tion 116 would have higher reduction rates for 3 substituted 2 cyclohexenones; however, this was not observed. In fact, both the Phe and Ile OYE mutants had lower sp ecific a ctivities for both 3 me t h yl and 3 ethyl 2 cyclohexenone than the wild type values, although the enantioselectivities were identical (>98% e.e. favoring the (S) products) Si nce the initial hypothe si s relating improved reaction rates with the identity of th e re si due at po si tion 116 proved incorrect, purified wild type, W116F, and W116I OYE 1 proteins were tested against a panel of representative enones and enals to see whether these mutations had other effects on OYE properties. The enantiomers of carvone provided a surpri si ng result: while (R) carvone was reduced by all three proteins to the same trans product with very high diastereoselectivity, the fate of the (S) a ntipode depended on the identity of the re si due at po si tion 116 ( Figure 1 13 ). The wild ty pe and W116F reduced (S) carvone to cis ( 2 R, 5 S) dihydrocarvone By contrast, the W116I mutant afforded trans (2 S ,5 S ) dihydrocarvone The origin of the altered stereoselectivity displayed by the W116I OYE variant was probed by u si ng (4 R ) NADPD as the cofactor for reducing (S) carvone MS analy si s of the trans product revealed incorporation of a si ngle deuterium atom, as expected (76% of the total product). The 1 H NMR spectrum showed that the C 3 axial proton ( H 3 ax ) of trans (2 S ,5 S ) dihydrocarvone had be en deuterated This as si gnment was supported by careful analy si s of the coupling patterns
41 Figure 1 13. The unexpected setereochemical outcomes of OYE1 mediated reductions of (R) and (S) carvone. The Trp116Ile variant reduced (S) carvone to yield a product that was different than that produced by the wild type enzyme. Taken together, these data indicate that the W116I mutant carries out net trans addition of H 2 across the double bond of (S) carvone however, the orienta tion compared to that of (R) carvone T his was the first report that an alternative substrate binding mode has been created by protein engineering for a member of the OYE family of alkene reductases
42 CHAPTER 2 UNCOVERING STEREOCOMPLEMENTARITY IN S. pasto rianus OYE1 Stere ocomplementarity in OYE1: Acces si Product the Different Substrate Binding Modes The work of Padhi et al provided clear evidence that it was pos si ble to achieve stereocomplementary products by making changes in the active si te of OYE1. As indicated by the 2 H NMR studies, the Trp116Ile enzyme like the wild type enzyme catalyzed the net trans addition of H 2 to (S) carvone This result is not entirel y unexpected con si dering the geometric constraints imposed by the spatial arrangement of Tyr196 and the FMN cofactor. This observation then suggested that the reversed stereoselectivity was a result of a binding orientation that placed the oppo si te face of the alkene above the si face of the flavin ring. Alternative binding modes, si milar to those proposed by Vaz et al., would offer a plau si ble explanation for the observed stereochemical outcome. Furthermore, previous in si lico studies had also suggested th e fea si bility of alternative Michaelis complexes 35 T he two hypothe si zed binding modes for (S) carvone are illustrate d in Figure 2 1 Figure 2 1 The alternative binding mode hypothe si s. A. The normal binding mode. B. The flipped binding mode
43 Discriminating Between (R) and (S) While the alternative binding mode hypothe si s provides a clear explanation for the observed stereochemical diver si ty displayed b y Trp116Ile, it does not offer much guidance in understanding t he precise mechanism for a flipped binding mode. Neither does it explain how a particular enzyme variant can discriminate between flipping (R) and (S) carvone. At a first glance, one obvious explanation can be linked to the fact that tryptophan is much lar ger than isoleucine. A Trp si de chain has a n approximate volume of 169 3 while Ile is si gnificantly smaller at 108 3 (calculated from the values reported by Zamyatnin 36 ) A Trp to Ile mutation would, therefore, si gnificantly increase the active si te volume pos si bly allowing for an alternative substrate orientation. However, more careful deliberation raises an interesting question tha t cannot be answered on the base s of molecular volumes alone. How is it pos si ble that the Trp116Ile mutant can lead to a reduction product of (S) carvone that is oppo si te to that of wild type while yielding an (R) carvone reduction product that is the same as that of wild type OYE1? A nother question worthy of asking is: What effect does each amino acid substitution at po si tion 116 Given the way Padhi and co workers carried out their or i ginal studies, it was unclear whether all 19 pos si ble Trp116 variants were generated. Secondly, the screening procedure was de si gned to identify OYE1 variants that had high 3 methyl 2 cyclohexenone reductase activity which could have led to the omis si on of many variants with other interesting properties. If an understanding of the precise factors that control stereoselectivity in OYE1 was to be achieved, a library of position 116 variants that is fu lly saturated was in order.
44 Addres si ng the Limitations : Experimental De si gn The Problem particularly in the area of asymmetric reductions One challenge in developing biocatal ytic methods for organic synthe si s is that gaining access to both product enantiomers can be difficult, but also essential in achieving acceptance of these catalysts in preparative organic chemistry. OYEs are particular ly challenging in this regard. Becaus e of active site spatial constraints as well as the geometric constraints of the FMN mediated hydride transfer, these enzymes often show very high stereoselectivities and readily provide one product enantiomer in appreciable excess H owever, obtaining the prove much more difficult making protein engineering an attractive strategy. With one exception ( in (S) carvone) we were unable to discover an enantiocomplementary OYE 1 for any of the substrates we investigated and were therefore able to fulfill only few of our methodological goals. While our earlier studies had yielded one stereocomplementary OYE 1 varia nt for (S) carvone reduction, this was an isolated example, and an analogous solution for (R) carvone was not found in the limited number of Trp116 replacements examined ( Figure 2 2 ) To address this problem, a more systematic study of mutations in position 116 was undertaken. Moreover, while we had deduced that ( S ) carvone should have bound in explain the observed stereochemistry, we had no direct experimental evid ence to support this contention We therefore gathered extensive structural data in order to not only validate our previous hypotheses but also to provide valuable guidance for protein engineering efforts.
45 Figure 2 2 Stereocompleme ntarity in OYE1 variants. While both product diastereomers could be obtained for (S) carvone, no variant could mimic this result for (R) carvone. Experimental Approach and Techniques Our goal was not onl y to solve this particular synthetic problem and gain access to all pos si ble asymmetric reduction products but also to discover general principles that might guide future efforts to adapt these versatile biocatalysts for even wider synthetic applications. Another aspect that we sought to fully explain was the precise mechanism for the observed stereochemical inver si on and also we looked to corroborate our flipped binding mode hypothesis with experimental evidence. In order to address the aforementioned con cerns, a full library that contained a complete set of Trp 116 amino acid replacements for OYE 1 was constructed through
46 si te directed mut a gene si s Several Trp116 variants had been previously generated by Dr. W. Colin Conerly, and with the help of Dr. Bradford Sullivan, a library containing all 19 possible amino acid replacements was constructed. Each mutant had its properties evaluated with respect to carvone reductions and ability to carry out carvone dismutation. Additionall y in an attempt to gain an understanding of the mechanism for stereocomplementarity in this system at an atomic level we turned to si ngle crystal X ray diffraction studies Trp116 that exhibited the most interesting properties were overexpressed and puri fied to homogeneity for crystallization and structure determination Results a nd Discus si on Catalytic Studies of OYE1 M utants (S) Carvone reductions Replacements for Trp 116 had si gnificant impacts on both the efficiency and the stereoselectivity of (S) carvone reductions While most polar replacements particularly those with charged si de chains were highly detrimental, the majority of non polar amino substitutions yielded functional OYE1 variants. The main exceptions were Asn and Gln, both of which gave good catalytic efficiencies and mainly the product expected ng observation was that the Trp116 Leu mutant afforded mostly the product derived from the normal substrate binding mode whereas the closely related amino acids Ile and Val gave almost exclu si which is nearly isosteric with Val, gave much lower catalytic efficiency. These were the first indications that su bstrate binding and catalytic activity could be altered by very subtle changes to the OYE1 active si te structure. Figure 2 3 summarizes the results
47 obtained. The x axis represents the conver si on percentage while the y axis lists the diastereomeric excess (d.e. %). The values for diastereomeric excesses were calculated as follows: Where d 1 is the amount of diastereomer 1 (major product) and d 2 the amount of diastereomer 2 (minor prod uct) observed The peaks were as si gned through GC MS analyses of a uthentic standards (Appendix A 1 ). Figure 2 3 The results of (S) carvone reductions Each square ( ) represents a si ngle Trp116 replacement and has an associated conver si on and d. e. percentage value. Wild type (W) is indicated by a star. (R) Carvone reductions Trp 116 substitution also had si gnificant impacts on the outcomes of (R) carvone reductions ( Figure 2 4). As before, polar amino acids (except for Asn and Gln) dramatically decreased alkene reduction efficiency. Surpri si ngly, all variants except two
48 (Ala and Val) gave almost exclu si vely the product derived from the normal substrate binding mode. The behavior of the Val mutant is even more surpri si ng si nce the closely relate d Leu and Ile variants productively bound (R) carvone only in the normal orientation. From a practical point of view, the Ala and Val variants are useful in completing the collection of diastereoselective alkene reductases that allow selective access to a ll four stereoisomeric products from (S) and (R) carvone ( Figure 2 2 ). Figure 2 4 The results of (R) si ngle Trp116 replacement and has an associated conver si on and d.e. percentage value. Wild type (W) is indicated by a star. The results of carvone red uction by wild type and Trp 116 variants of OYE1 are summarized in Figure 2 3 and 2 4 A si de from those mutants with little or no catalytic activity, successful replacements for Trp 116 can be divided into thre e categories: 1) those that productively bound both (S) and (R) carvone mainly in the normal orientation; 2) those that bound (S) (R) carvone normally; 3) those that productively bound both (S) and (R) carvone in
49 Probing the Active Si te through Charge Transfer Complex Formation Studies In the absence of an external reductant such as NADPH, OYE1 catalyzes a d ismutation reaction in which H 2 is abstracted from one substrate molecule and tra nsferred to a second 19 In the case of 2 cyclohexenones, this process is essentially irrever si ble si nce the oxidation product is a cyclohexadienone that rapidly tautomerizes to the corresp onding phenol. These phenols subsequently form high affinity charge transfer complexes with OYE1 that feature absorbance maxima at ca 650 nm 8 that can be recognized visually by the change in en zyme color from bright yellow (oxidized FMN form) to green. Carvone dismutation requires that the reacting hydrogens be oriented in a trans diaxial arrangement ( Figure 2 5 ). Because the location of the FMN is fixed within the protein environment, the subs trate must be po si tioned so that its C hydrogen lies within N5 37 In addition, the carbonyl oxygen must form hydrogen bonds to the si de chains of His191 and Asn 194 to facilitate the polar elimination of H 2 These requirements, in addition to avoiding steric conflict between the isopropenyl si de chain and th e FMN cofactor, define the active si te locations of (S) and (R) carvone to be those shown in Figure 2 5 While the ring carbons of (S) carvone lie in the same locations as those of a phenol inhibitor, those of (R) carvone must be shifted somewhat in order to maintain the proper locations of C and the carbonyl oxygen. This has the effect of moving the (R) substrate into closer proximity to the si de chain of the amino acid at po si tion 116 as compared to its (S) counterpart. We used this approach to gain so me information on the effective active volume created by Trp116 mutations. We would later use this information in an attempt to draw parallels between charge transfer complex formation
50 Figure 2 5 This scheme shows the geometric constraints for (S) and (R) carvone oxida tion in the reductive half reaction In order to react, trans diaxial hydrogens must be properly po si tioned. This implicates different orientations for (R) and (S) carvone. The ability to form phenol 5 from both (S) and (R) carvone were determined spectrophotometrically for all catalytically active OYE1 Trp116 mutants (this criterion excluded the Glu, Arg, Cys, Lys, Thr and Gly vari ants si nce they showed less than 30% activity i n NADPH driven reductions). Because the oxidation product, phenol 5 is a potent inhibitor of OYE1 progress curves for these reactions are inherently non linear. We therefore chose a fixed reaction time (24 hrs.) and assessed the fraction of enzyme with bound phenol 5 at that time This complexation generates a charge transfer band centered at 648 nm, and we determined the fraction of OYE1 with bound 5 by measuring the A 462 / A 648 ratios both at the start of the reaction and after 24 hr. Values for A 462 / A 648 of completely saturated enzymes were determined by adding an excess of p chlorophenol in separate experiments to determine the final A 462 / A 648 ratio of a completely saturated protein phenol complex. Preliminary studies had shown that
51 the charge t ransfer bands from p chlorophenol were a good approximation for those formed by phenol 5 ; the former was used here because it was commercially available These data allowed the fractional saturation for each dismutation reaction to be calculated after 24 hr ( Figure 2 6 ) Interestingly both Gln and Asp replacements failed to show spectra typical of charge transfer complexes and the results obtained were inconclu si ve. Figure 2 6 Charge transfer complex formation activities of Trp116 with (S) carvone (li ght bars) and (R) carvone (dark bars). The horizontal bars represent the mean saturation percentage for each substrate. In general, (S) carvone was a better substrate than (R) carvone at forming phenol 5 The mean value for ( S ) carvone binding saturation is 73% in contrast to 65% for the R enantiomer. In the case of the wild type enzyme the largest discrepancy in charge transfer complex formation was observed (S) carvone reached nearly 40% saturation contrasting (R) is general trend is in fair agreement with the proposed model ( vide supra ).
52 Crystallography S tudies Structural overview To help understand what molecular interactions controlled the orientation of substrate binding in wild type OYE 1 and its Trp 116 replacements, we determined the three dimen si onal structures for several key variants in the presence of substrates and / or other ligands. To avoid potential interference during crystallization, the GST fu si on tag was deleted and proteins were purified by affinity chroma tography on a phenol containing re si n followed by final polishing by gel filtration chromatography (see experimental section for full details) All variants crystallized in a hanging drop vapor diffu si on setup under very si milar conditions and in the same space group as reported by Fox and Karplus for the wild type enzyme (P 4 3 2 1 2) 14 As expected, the overall structures for all mutants w ere nearly identical (RMSD 0.10 0.27 ). The only exception was the extra barrel helix B (re si dues 145 153), which exhibited greater deviation. This region usually exhibited higher B factors and makes contacts with adjacent protein m olecules in the cr ystal lattice These variations usually seemed to correlate with small differences in unit cell dimensions and are likely to be caused by crystal packing restraints. All structures were solved by molecular replacement u si ng the wild type OYE1 structure ( P DB accession code 1OYA) or one of the first mutants determined during the course of this study (3RND) All of our structures showed a slightly different protein structure near the C terminus than reported previously as well as an ordered Mg 2+ ion A si de fr om these minor differences, the overall structures reported here ar e highly si milar to that of 1OYA The po si tions of the mutant si de chains at po si tion 116 were readily apparent from the high resolution omit electron den si ty difference maps m F o
53 D F c To oaking experiments were carried out for all key variants by adding neat (R) and (S) carvone to pre grown OYE1 crystals in the mother liquor prior to mounting Attempts to obtain co crystallize the usu ally resulted in no crystal growth or poorly diffracting crystals (figure 2 7A ,B ). We also carried out some soaking experiments with (+) dihydrocarvone, which was added to crystals of the Th r, Ala and Val OYE1 variants. Figure 2 7 Panels A and B show res ults from OYE1 and ligands co crystallization. These are not suitable for X ray diffraction studies All crystals initially exhibited the characteristic bright yellow color of oxidized flavin; however, after several days of soaking in the presence of carvon e, some mutants gradually adopted a green color. This was particularly apparent in crystals of the Val and Ala replacements. By contrast, crystals of the Ile and Gln variants remained bright yellow, even after prolonged ligand soaking times. When ligands were observed in the active si well def ined by the electron den si ty. Ligand atoms were added manually only after the protein and most water molecules had been refined to avoid m odel bias Wild type Crystals of wild type OYE1 were soaked individually with both (S) and (R) carvone and complete, high resolution data sets were collected for both experiments.
54 No electron den si ty that could be modeled as carvone was observed; instead, electron den si ty that could be best interpreted as a chloride ion was found at the location occupied by a phenolate oxygen atom in charg e transfer complex structures OYE1 is known to bind monovalent anions including chloride with a K D of about 8 mM 38 and c hloride is present in the crystallization buffer at approximately 400 mM. For these reasons, c si te ligand found in the absence of additional ligand soaking. That it remained bound to the wild type enzyme even in the presence of saturating carvone implies that the thermodynamic stability of these complexes was insufficient to displace even this weakly bound active si te chloride Trp116 Leu mutant Based on the nature of the carvone reduction products obtained from this protein, we surmised that the Trp116 Leu OYE 1 variant must have bound both (R) and (S) carvo Numerous soaking experiments were carried out in an effort to observe the latter within the active si te; unfortunately, only a spherical region of electron den si ty best modeled as a chloride ion was found when (S) carvone was used. By contrast, we were successful with (R) carvone, although the nature of the active si te ligand depended on the time between subs trate addition and crystal cool ing in liq uid nitrogen Initially yellow crystals soaked with (R) carvone fo r two days at 4 C gradually beca me green ( Figure 2 8 A, right panel ). When the soaking time was decreased to 2.5 hr before coolin g in liquid nitrogen, the crystals remained yellow ( Figure 2 8A left panel) We suspect ed that the green color was a result of a charge transfer complex formation between the enzyme and a phenol resulting from a dismutation reaction during the soaking period ( Figure 2 5 ). In order to validate this hypotheis we turned to in crystallo UV Vis spectroscopy. In these experiments,
55 Trp 1 16Leu si ngle crystals that had been soaked in (R) carvone for different time lengths were mounted in typical manner, for X ray data collection at 100 K After determining the optimal crystal orientation, UV Vis (350 800 nm) absorbance spectra were collect ed u si ng a microspectrophotometer that is focused on the crystal rotation axis. Complete data sets were collected for both (yellow and green) crystal forms, which exhibited si gnificantly different absorba nce spectra in the 550 800 nm ( Figure 2 8 B ). Figure 2 8 Yellow and green crystal forms. Panel A photographs of the mounted Trp 116L eu crystals. Crystals were soaked for 2.5 hrs. (left) or for 48 hrs. (right). Panel B shows the spectra collected for both crystal forms. C Experimental electron den si ty maps (omit m Fo D Fc) for the ligands in the active si te. The spectra for the green crystal form w ere con si stent with those of charge transfer complexes between OYE and a phenolic inhibitor (see previous sections for discus si on) max value of ca 675 nm observed for the green crystals during X ray data collection, we suspected that (R) carvone had in fact, undergone
56 dismutation and that the active si te ligand was actually phenol 5 Si nce no phenolic molecules were present during the purification procedure or in the crystallization buffer, we rationalized that phenolic ligand formation must have occurred post crystallization. This notion was supported by the ligand electron den si ty maps which was fit much better by phenol 5 than by the starting substrate (R) carvone ( Figure 2 8 C). As expected, the orientation of phenol 5 in the active si te matched our expectation s for the normal binding mode wherein the ligand is directly superimposed onto the p hydroxybenzaldehyde inhibito r as initially reported by Fox and Karplus 14 ( Figure 2 9 ) Figure 2 9 The normal binding mode. Superposition of phenol 5 and p HBA. Tr p116Leu atoms are shown as thick lines in cyan color. OYE1 wild type is shown as thinner ball and stick representation. The phenolic oxygen of phenol 5 is ideally positioned to hydrogen bond to His191 and Asn194 and its aromatic ring is directly above the si face of the flavin mononucleotide ring As highlighted in Figure 2 8 the phenolic ligand ring displays flat topology and is oriented parallel to the isoalloxazine rings of the flavin. This
57 conform si de chain of Phe 296 (closest approach of 3.20 ); nevertheless, the phenyl si de chain predominantly occupied one wel l defined conformation Tyr375 is also in close proximity and has some evid ence for multiple conformations although a major position could be deduced from its electron density. In a ddition, t he Leu 116 si de chain w as also well defined with no unduly close contacts between the methyl group of 5 and this protein re si due ( Figure 2 10 A ) Figure 2 10 OYE1 Trp116Leu. A Crystal structure of phenol 5 bound to Trp116Leu OYE1. 2mFo DFc Electron den si ty maps for relevant re si dues are shown at si dues are colored in cyan, FMN in yellow and the ligand in green. The red mesh is the omit difference (mF o DF c ) map of B Crystal structure of (R) carvone 4 bound to Trp116Leu OYE1. Electron den si ty maps are shown at si dues. Protein re s i dues are colored in cyan, FMN in yellow and ligand in green. The red mesh is the omit difference (mF o DF c B oth the ligand electron den si ty from the yellow crystals of (R) carvone soaked Trp 116Leu OYE1 and the absence of an absorption band in the 500 800 nm range were con si stent with unmodified substrate in the active si te ( Figure 2 10 B) In fact, the
58 experimenta l 2mFo DFc and mFo DFc electron den si ty maps made (R) Carvone ea si ly identifiable. As in the previous structure, the ( R ) carvone ligand was found in the occupying a po si tion si milar to that of phenol 5 seen in the structure of the green crystal. The carbonyl oxygen of 4 was ideally po si tioned to form hydrogen bonds wit h both N 191 and N Its isopropenyl group pointed away from the si de chain of Leu 116 and made conta cts with the si de chains of Phe250, Pro 295 and Phe296. Minor adjustments were also observed in the si de chain po si tions of Phe 250 and Ty r 375, which appeared to p opulate multiple conformations. The C 2 methyl group of 4 make s hydrophobic contacts with C C 116. The distance and angle from the FMN N5 atom (3.75 and 105, respectively) suggests that this complex mimics the catalytically productive Michaelis complex and orientation by the Leu variant. One important difference between t his intact (R) carvone complex and that with dismutation product 5 was that the non planar carvone provoked two key changes in the protein structure. First, the electron de n si ty for the si de chain of Phe 296 was weak, and it was clear that several conformat ions were present in the ensemble of protein molecu le s within the crystal ( Figure 2 10 B ). Likewise, the si de chain of Tyr375 was also found in at least two conformations. These observations suggest that additional changes at po si tions 296 and 375 that cre ate additional space for substrate binding might yield even more catalytica lly efficiency OYE1 variants. The close approach of the isopropenyl group to the abovementioned protein re si dues may also explain why we were unable to observe a complex with (S) ca rvone desp ite numerous soaking attempts.
59 Table 2 1 X ray Crystallographic Data Collection and Refinement Statistics PDB Acces si on code 4GXM 4GWE Ligand Soaked (R) carvone (R) carvone Observed active si te ligand Phenol 5 (R) carvone X ray source NSLS X26C NSLS X26C Space group P 4 3 2 1 2 P 4 3 2 1 2 Unit cell dimen si ons a = b, c () 141.12, 42.92 140.78, 42.76 Resolution () 34.2 1.36 (1.41 1.36) 28.2 1.45 (1.49 1.44) Unique reflections 87,060 (5,990) 70,039 (4,324) Completeness (%) 93.78 (65.40) 90.83 (56.74) Multiplicity 16.5 (7.5) 12.8 (11.7) R sym [b] 0.062 (0.38) 0.044 (0.18) 39.89 (4.49) 48.82 (12.97) R work [c] R free [d] 0.108, 0.137 0.115, 0.145 Ramachandran statistics [e] Favored (%) Allowed (%) Outliers (%) 98 2 0 98 2 0 Number of protein, solvent and ligand atoms 3,287, 633, 100 3,255, 590, 65 Average B factors ( 2 ) Protein Solvent FMN Ligand 11.5 25.4 7.5 18.1 13 26.7 10.0 19.1 [a] Values in parentheses denote data for the highest resolution bin. [b] Rmerge = hkl i |Ii(hkl) [I(hkl)]|/ hkl i Ii(hkl), where Ii(hkl) is the inten si ty of the ith observation of unique reflection hkl. [c] R work |F o (hkl)| |F c (hkl)| o (hkl)|. [d] R free is calculated in the same manner as R work u si ng 10% of the reflection data not included during the refinement. [e] Statistics generated u si ng MOLPROBITY 39
60 While the po si tions of the isopropenyl si de chains occupy si milar locations in the two carvone enantiomers, their exact locations differ slightly and in cases where they are nearly in van der Waals contact with protein re si dues, v ery minor changes in the location of this moiety may have si gnificant impacts Both X ray data sets of the Trp116Leu enzyme yielded high quality atomic models. The complete crystallographic parameters, data collection and refinement statistics can be found in Table 2 1 Trp116 Ile mutant This OYE1 mutant was of particular interest to us since it exhibited a unique behavior. Trp116Ile catalyzed the reduction of ( S ) carvone with a nearly complete reversal of stereochemistry compared to the wild type enzyme. Curiously, it showed the same stereochemistry as the wild type enzyme in the redu ction of ( R ) carvone. We therefore attempted to obtain crystal structures of complexes formed between the enzyme and both ( R ) and ( S ) carvone. Crystals soaked with (R) carvone yield ed active si te electron den si ty maps that were not consistent with the ( R ) carvone substrate. Rather, the best interpretation was consistent with a chloride ion which is present in the crystallization buffer and known to bind to OYE1. One important aspect is that, in the high resolution crystal structure (1.55 ) the Ile11 6 si de chain showed very well defined electron den si ty for all of its atoms ( Figure 2 12 A ). The si de chains of Phe296 and Tyr375 showed indications of disorder having weak electron den si ty and high B factors As anticipated, this substitution si de of the OYE1 active si te just below Tyr196
61 Table 2 2 X ray Crystallographic Data Collection and Refinement Statistics PDB Acces si on code 4H6K 4GE8 Ligand Soaked (R) carvone (S) carvone Observed active si te ligand Chloride (S) carvone X ray source NSLS X25 NSLS X25 Space group P 4 3 2 1 2 P 4 3 2 1 2 Unit cell dimen si ons a = b, c () 140.93, 42.46 140,93, 42.48 Resolution () 31.5 1.55 (1.60 1.55) 24.2 1.5 (1.55 1.49) Unique reflections 62,517 (6,122) 68,916 (6,757) Completeness (%) 99.94 (99.69) 99.75 (99.22) Multiplicity 9.1 (8.7) 11.1 (10.4) R sym [b] 0.080 (0.90) 0.048 (0.63) 14.22 (1.89) 29.40 (2.83) R work [c] R free [d] 0.182, 0.204 0.137, 0.170 Ramachandran statistics [e] Favored (%) Allowed (%) Outliers (%) 97 3 0 98 2 0 Number of protein, solvent and ligand atoms 3,216, 313, 56 3,256, 465, 91 Average B factors ( 2 ) Protein Solvent FMN Ligand 29.6 38 19.5 21.2 38.0 13.7 23.9 [a ] Values in parentheses denote data for the highest resolution bin. [b] Rmerge = hkl i |Ii(hkl) [I(hkl)]|/ hkl i Ii(hkl), where Ii(hkl) is the inten si ty of the ith observation of unique reflection hkl. [c] R work |F o (hkl)| |F c (hkl)| o (hkl )|. [d] R free is calculated in the same manner as R work u si ng 10% of the reflection data not included during the refinement. [e] Statistics generated u si ng MOLPROBITY 39
62 A ddition of (S) carvone to these crystals yielded a pseudo Michaelis complex that demonstrated directly for the first time that carvone could be accommo dated in the active si te of S. pastorianus OYE1 ( Figure 2 11 B ). The si de chain of Tyr196 moves si gnificantly to bind slight backbone shift are observed in the cr ystal structure of the complex. (S) Carvone was stacked above the FMN co factor with its carbonyl oxygen atom po si tioned to form hydrogen bonds with both N His191 and N iso propenyl group lies in a hydrop hobic pocket directly above Thr 37 and makes hydrophobic contacts with the si de chain of the I le introduced at po si tion 116. The si de chains of Tyr375 and Phe296 showed well defined electron den si ty, contrasting what was seen in the crystal structure of the uncomplexed enzyme. The dis tance and angle from N5 of the FMN (3.70 and 89, respectively) were within the range of values expected 37 for catalytically relevant complex. Figure 2 11 OYE1 Trp116Ile. A The free enzyme structure. B Complex of enzyme and (S) carvone. The gray mesh represents 2mFo mesh in B represents the omit mFo codes 4H6K (A), 4GE8 (B). Protein re si dues are colored in cyan, FMN in yellow and the ligand in green
63 Interestingly, the si de chain of Ile11 6 was no longer well defined at the methyl ter minus in this pseudo Michaelis complex; rather, disorder was observed at the point of closest approach to the bound substrate ( Figure 2 12 B ) Moreove r, the main conformer of Ile had the iso propenyl group of ca rvone. This suggests that there may be some steric interference, even in a variant that allows (S) The se steric clashes between the si de chain of Ile at po si tion 116 and the iso propenyl moiety of the substrate may also explain why the same mutant was unable to bind (R) carvone in an analogous s ubstituent s in these substrate e nantiomers occupy si milar, but not identical, spatial po si tions. The complete c rystallographic parameters, data collection and refinement stat istics can be found in Table 2 2 Figure 2 12 Detailed view of Trp116Ile. A Electron den si ty observed in the chloride structure. B Electron den si ty observed in the (S) carvone complex structure. The gray mesh represents 2mFo represents the omit mFo The main conformations of Ile116 have rotated 180 apart. Protein re si dues are colored in cyan, FMN in yellow and the ligand in green
64 Trp116 Ala mutant Based on the alkene reduction products ob tained, we deduced that the Trp 116 Ala l mutant must have bound both (S) and (R) carvone primar orientations. These predictions were borne out by the results of soak ing experiments, which yielded electron den si ty maps that were only con si stent with the expected liga nd binding orientation ( Figure 2 13 B ). Unfortunately, the high dismutation activity of this OYE1 variant toward both carvone enantiomers meant that only p henol 5 could be observed in the crystals ) even after relatively short soaking times The po si tion of phenol 5 in both complexes was very si milar to that observed for the (S) Trp116Ile variant. The iso propenyl group of 5 rests above the si de chain of Thr37 a nd its phenolic oxygen makes short (2.71 and 2.77 ) and well aligned hydrogen bonds wit h of His191 and of Asn 194 respectively As mentioned above, crystals soaked with (S) and (R) carvone solutions yielded electron den si ty maps that were best fitted by phenol 5 ( Figure 2 13 D). The final models were essentially indistinguishable independently of which enantiomer of carvone they had been exposed to Curio u sly, s oaking crystals of the Trp 116Ala mutant with (+) dihydrocarvone 6 overnight at 4 C yielded active si te electron den si ty best fit by (R) carvone 4 in the flipped orientation ( Figure 2 13 A C ). One pos si ble route to the crystallographically observed co mplex is summarized in Figure 2 14 Trans elimination of H 2 from 6 wou ld most logically occur from a binding arrangement in which the cofactor and (R) carvone within the active si complex, the FMNH 2 can either transfer a pair of electrons to the alkene (back reaction to re form 6 ) or reduce a different acceptor (likely O 2 ) in an essentially irrever si ble step
65 Figure 2 13 OYE1 Trp116Ala. A Crystal structure of Trp116Ala in complex with (R) carvone. B Trp116Ala with phenol 5 bound in the active si te. C Omit mFo DFc map of (R) carvone at 3 Omit mFo DFc map of phenol 5 at 4.5 Protein re si dues are colored in cyan, FMN in yellow and the ligand in green This would leave an oxidized su bstrate in the active si te with an oxidized cofactor. In the absence of an external reductant as in the crystallization drop, this dead end complex crystallographically observed comp lex. Whether this change in substrate binding orientation involves dissociation and re binding of (R) carvone or occurs intramolecularly is unknown; however, the high external concentration of (+) dihydrocarvone would diminish the chances of re binding any (R) carvone released to the bulk solution, suggesting that the latter pathway may be more likely It is also pos si ble that H 2 is eliminated from the minor, cis isomer of (+) dihydrocarvone (present at 15% in the commercial reagent used). This would place (R) carvone in the active si te
66 in the same orientation as observed experimentally One puzzling observation is that little or no further dismutation of (R) carvone to phenol 5 was observed in the crystal structure, despite the demonstrated ability of this mutant to catalyze this reaction when (R) carvo ne was the starting substrate. It is pos si ble that the enzyme had insufficient time to carry out the second dismutation step prior to flash cool ing in liquid nitrogen Figure 2 14 Schematic representation o f (R) carvone for mation from (+) dihydrocarvone, via an enzymatic oxidation (dehydrogenation). (R) carvone complex with the Trp 116 Ala mutant was formed, this structure provide d valuable information binding mode, clearly demonstrating that this mutant is capable of interacting with substr ate in this manner ( Figure 2 14 ). That the final (R) carvone complex involved the preferred even when the alternative is acces si bl e. This led us to hypothe si ze that OYE1 intrin si arrangement is sterically precluded by an amino acid si de chain at po si tion 116 is the binding orientation utilized. This notion was further supported by studies of the Trp 116 Val mutant.
67 Table 2 3 X ray Crystallographic Data Collection and Refinement Statistics PDB Acces si on code 4GBU 4K7V Ligand Soaked (S) carvone (+) dihydrocarvone Observed active si te ligand Phenol 5 (R) carvone X ray source NSLS X6A NSLS X6A Space group P 4 3 2 1 2 P 4 3 2 1 2 Unit cell dimen si ons a = b, c () 141.12, 42.81 140.85, 42.84 Resolution () 21.0 1.18 (1.22 1.17) 35.42 1.52 (1.57 1.516) Unique reflections 141,261 (13,929) 66,315 (6,492) Completeness (%) 99.71 (99.59) 98.52 (97.79) Multiplicity 7.1 (6.6) 7.8 (7.2) R sym [b] 0.076 (0.55) 0.093 (0.524) 15.10 (2.05) 22.67 (3.44) R work [c] R free [d] 0.100, 0.121 0.145, 0.183 Ramachandran statistics [e] Favored (%) Allowed (%) Outliers (%) 98 2 0 97 3 0 Number of protein, solvent and ligand atoms 3,281, 658, 107 3,205, 500, 62 Average B factors ( 2 ) Protein Solvent FMN Ligand 11.5 26.1 6.9 16.3 18.5 30.90 16.6 25.3 [a] Values in parentheses denote data for the highest resolution bin. [b] Rmerge = hkl i |Ii( hkl) [I(hkl)]|/ hkl i Ii(hkl), where Ii(hkl) is the inten si ty of the ith observation of unique reflection hkl. [c] R work |F o (hkl)| |F c (hkl)| o (hkl)|. [d] R free is calculated in the same manner as R work u si ng 10% of the reflection data not included during the refinement. [e] Statistics generated u si ng MOLPROBITY 39
68 Trp116 Val mutant As observed with the A l a variant, soaking crystals of the Trp116Val OYE1 mutant with either (S) or (R) carvone overnight at 4 C gave complexes with dismutation product 5 in the ( Figure 2 15 B) The ligand location was si milar to that observed for the Ala structures, although C8 of the ligand was shifted by 0.64 in order to accommodate contact s with and of the Val at po si tion 116. The phenolic oxygen remained at essentially the same po si tion as seen earlier in the Trp116Ala complexes. Figure 2 15 OYE1 Trp116Val. A Crystal structure of Trp116 Val in complex with (R) carvone. B Trp116 Val with phenol 5 bound in the active si te. Grey mesh is 2mFo DFc electron den si mFo Protein re si dues are colored in cyan, FMN in yellow and the ligand in green Soaking the Trp116Val mutant with (+) dihydrocarvone 6 yielded complex of (R) carvone within the active si ding orientation ( Figure 2 15 A ). This complex likely arises by the same mechanism as described above for the Ala case ( Figure 2 14 ) and further underscores the notion that even when both ligand binding orientations are acces si ble, OYE1
69 Table 2 4 X ray Crystallographic Data Collection and Refinement Statistics PDB Acces si on code 4H4I 4K8E 4K8H Ligand Soaked (S) carvone (R) carvone (+) dihydrocarvone Observed active si te ligand Phenol 5 Phenol 5 (R) carvone X ray source NSLS beamline X6A NSLS X6A NSLS beamline X6A Space group P 4 3 2 1 2 P 4 3 2 1 2 P 4 3 2 1 2 Unit cell dimen si ons 141.35, 42.85 141.443, 42.851 a = b, c () 141.00, 42.82 Resolution () 23.6 1.25 (1.29 1.25) 24.2 1.27 (1.315 1.269) 35.48 1.55 (1.605 1.55) Unique reflections 119,164 (11.776) 113,485 (11,067) 65,589 (6,247) Completeness (%) 99.98 (99.95) 99.44 (98.44) 99.95 (99.57) Multiplicity 10.2 (9.9) 9.9 (9.0) 12.2 (12.2) R sym [b] 0.10 (0.55) 0.069 (0.547) 0.086 (0.676) 19.58 (3.07) 20.33 (2.46) 24.60 (3.79) R work [c] R free [d] 0.124, 0.146 0.107, 0.127 0.124, 0.162 Ramachandran statistics [e] Favored (%) Allowed (%) Outliers (%) 98 2 0 98 2 0 97 3 0 Number of protein, solvent and ligand atoms 3,286, 660, 108 3,325, 642, 97 3,320, 542, 72 Average B factors ( 2 ) Protein Solvent FMN Ligands 9.8 26.5 5.8 14.7 11.7 27.90 8.6 16.6 15.00 27.60 12.1 19.3
70 Trp116 Thr mutant The Thr mutant was chosen for additional study because this re si due althoug h very different in polarity, is near ly isosteric with Val Our hypothe si s was that if substrate binding orientation was controlled primarily by steric factors, the Thr and Val variants should show si milar properties. In other words, the same invers ion of stereoselectivity for both ( R ) and ( S ) carvone displayed by the Ala mutant was expected in the Thr mutant. However, t his was not observed. Substituting Thr for Trp116 was highly detrimental for the catalytic efficiency of OYE1 and this mutant showe d essentially no conver si on for (R) carvone and relatively poor conver si on of (S) carvone with r eversed stereoselectivity.( Figure s 2 3 2 4 ). Figure 2 16 Crystal Structure of the Trp116Thr mutant. The g rey mesh shows 2mFo DFc electron den si Protein re si dues are colored in cyan, FMN in yellow and the waters are shown as red spheres.
71 Table 2 5 X ray Crystallographic Data Collection and Refinement Statistics PDB Acces si on code 4K7Y Ligand Soaked (R) carvone Observed active si te ligand Chloride X ray source NSLS X6A Space group P 4 3 2 1 2 Unit cell dimen si ons a = b, c () 141.37, 42.83 Resolution () 23.59 1.2 (1.242 1.199) Unique reflections 135,136 (13,135) Completeness (%) 99.80 (98.28) Multiplicity 6.0 (5.6) R sym [b] 0.067 (0.570) 14.34 (1.87) R work [c] R free [d] 0.1150, 0.1350 Ramachandran statistics [e] Favored (%) Allowed (%) Outliers (%) 97 3 0 Number of protein, solvent and ligand atoms 3,358, 666, 58 Average B factors ( 2 ) Protein Solvent FMN Ligands 11.4 26.40 8.3 a] Values in parentheses denote data for the highest resolution bin. [b] Rmerge = hkl i |Ii(hkl) [I(hkl)]|/ hkl i Ii(hkl), where Ii(hkl) is the inten si ty of the ith observation of unique reflection hkl. [c] R work |F o (hkl)| |F c (hkl)| o (hkl)|. [d] R free is calculated in the same manner as R work u si ng 10% of the reflection data not included during the refinement. [e] Statistics generated u si ng MOLPROBITY 39
72 We soaked crystals of the Thr variant individually with (S) and (R) carvone as well as (+) dihydrocarvone and collected complete, high resolution data sets for each. In all cases, no active si te electron den si ty beyond that of chloride (buffer component) was observed. The h ighest resolution structure (1.20 ) was derive d from soaking with (R) carvone (4K7Y). Crystal soaked with (R) carvone and (+) dihydrocarvone were indistinguishable and were fully refined. Surp ri si ngly, the si de chain of Asn 194 appeared to occupy at least two different conformations in this structure w ith an approximate occupan c y of 65 : 35 ratio. To the best of our knowledge, this phenomenon has not been observed in any other OYE1 structure ( Figure 2 16 ). While the major conformer of Asn 194 was identical to that observed normally, the second placed the si de chain carbonyl above the FMN, occupying the same space normally filled by a chloride ion in an unliganded OYE 1 structure Modeling only the two conformations of Asn 194 led to a po si tive peak in the Fo Fc difference map, which could be eliminated by adding a chloride ion whose occupancy mirrored that of the Asn 194 si de ch ain in its normal conformation. It is pos si ble to model a hydrogen bond network between the si de chain of Th r116, water 723 and the si de chain of Asn 194 (in the second conformation) a nd these favorable interactions may partially explain t he observed structural changes. It should be noted that locating the si de chain of Asn 194 directly above the FMN would preclude substrate binding and this may be the reason that this Trp116 variant sho wed poor catalytic efficiency. The si de of Thr116 also showed evidence for multiple conformations and participates directly in a hydrogen b ond network with solvent molecules not previously observed in other structures The change in hydrophobicity of the binding pocket coupled to the unusual movement of
73 Asn194 (re si due known to be critical for substrate binding) may explain the lack of success in obtaining structures of the enzyme substrate complexes. The complete cryst allographic parameters, data collection and refinement statistics can be found in Table 2 5 Trp116 Gln mutant The final mutant selected for structural st udies was the Trp 116Gln variant si nce this protein showed good catalytic activity toward both (R) and (S) carvone with altered stereosele ctivity for the latter ( Figure 2 3 2 4 ). In this regard, its properties closely resemble that of the Ile mutant. We were unable to determine a dismutation activity for the Gln mutant, either because it was unable to c atalyze the reaction or it was unable to form a stable charge transfer complex carvone dismutation product 5 under the experimental conditions Attempts to observe (S) carvone within the active si te of the Trp116 Gln mutant by soaking with neat ligand were unsuccessful Soaking experiments with (R) carvone did afford a complex; however, there were several unusual features Based on the si milar to that seen for the Val a nd Ala variants. Instead, the substrate was observed in oppo si te stereoisomer ( Figure 2 17 A ). catalyt ically productive arrangement however First, the distances between the 2 of His191 and 2 of Asn 194 we re 3.7 and 4.5 , respectively. Such distances preclude hydrogen bonds, which are known to be essential for catalytic turnover by OYE1 40 An analogous binding mode for phenols, lacking these hydrogen bonds, would also be con si ste nt with our inability to observe a charge transfer
74 complex in dismutation studies of the Gln mutant In wild type OYE1 a hydrogen bond between the phenol oxygen and the si de chain of His191 stabilizes the former in its electron rich anionic form, allowing a charge transfer complex with the electron deficient FMN Figure 2 17 OYE1 Trp116Gln. A The complex of Trp116Gln and (R) carvone. The g rey mesh shows 2mFo DFc electron den si red mesh is the omit mFo B Superpo si tion of the all binding modes observed. The normal and flipped binding modes are shown in thin magenta lines, the non canonical binding observed in the Gln116 variant is shown in thick green lines. By mutating His191 Massey demonstrated that loss of this hydrogen bond eliminated charge transfer complex formation; in our case, it is ligand re po si tioning that eliminates the interaction. The geometric relationship between bound (R) carvone and the FMN cofactor also argues against the catalytic r elevance of the observed complex. In particular, the angle formed between C of carvone and N5 and N10 of the FMN is 66.5 and the C to N5 distance is 3.9 . Both of these values fall well out si de the ranges observed by Fraaije in his survey of flavoprot eins bound to their respective substrates and / or products, strongly suggesting that hydride transfer would be
75 Table 2 6 X ray Crystallographic Data Collection and Refinement Statistics PDB Acces si on code 3TXZ Ligand Soaked (R) carvone Observed active si te ligand (R) carvone X ray source X25 Space group P 4 3 2 1 2 Unit cell dimen si ons a = b, c () 141.36, 42.69 Resolution () 45.0 1.7 (1.76 1.7) Unique reflections 48,165 (4,726) Completeness (%) 99.99 (99.96) Multiplicity 9.5 (6.2) R sym [b] 0.052 (0.58) 17.10 (2.22) R work [c] R free [d] 0.161, 0.186 Ramachandran statistics [e] Favored (%) Allowed (%) Outliers (%) 98 2 0 Number of protein, solvent and ligand atoms 3,244, 368, 43 Average B factors ( 2 ) Protein Solvent FMN Ligands 23.5 36.1 16.8 25.6 a] Values in parentheses denote data for the highest resolution bin. [b] Rmerge = hkl i |Ii(hkl) [I(hkl)]|/ hkl i Ii(hkl), where Ii(hkl) is the inten si ty of the ith observation of unique reflection hkl. [c] R work |F o (hkl)| |F c (hkl)| o (hkl)|. [d] R free is calculated in the same manner as R work u si ng 10% of the reflection data not included during the refinement. [e] Statistics generated u si ng MOLPROBITY 39
76 impos si ble from this geometry. The isopropenyl group of (R) carvone has moved away from the si de chain of Gln116 and makes contacts primarily with Thr37 and Tyr375. The po si tion of (R) carvone is unlike those observed in either flipped or normal binding modes ( Figure 2 17 B). The complete crystallographic parameters, data collection and refinement stat istics can be f ound in Table 2 6 Conclu si ons and Future Work The main goals of this work were to: a) engineer stereocomplementary OYE1 variants, therefore gaining access to all four pos si ble diastereomers of dihydrocarvone; b) uncover the general principles by which th e orientation of substrate binding is controlled in OYE1. We therefore combined data from catalytic studies (alkene reduction and dismutation activity) with X ray crystallography. Given the hydrophobicity si te, wh ich is also predominantly hydrophobic, we initially hypothe si zed that substrate binding is largely dictated by steric constraints. The crystallographic studies provided experimental data for four pos si ble types of OYE1 ligand complexes, namely: a) carvone bound in a normal orientation; b) carvone 5 ) in normal orientation and d) superimposed the crystal structures of each of the cases described. Structures were overlaid with respect to their FMN cofactor only. All four overlaid structures show that the oxygen atoms of the ligands either a ketone carbonyl or a phenol hydroxyl occupy essentially identical po si tion s. Their location is ideal for forming hydrogen bonds with both His191 and Asn194, reinforcing the notion that this interaction is of high importance to ligand binding ( Figure 2 18 ).
77 Interestingly, whether the ligand was aromatic or non aromatic, the po si tions of the branch point carbon C8 on the isopropenyl substituent were in very close agreement While the po si tions of atom C8 are somewhat fixed, the atom of the methyl and methylene groups ( C9 and C10 ) are able to move due to the facile rot ation of the C5 C8 si ngle bond. Figure 2 18 Alignment of all four binding modes. (R) carvone in the normal orientation (pink), phenol 5 in the normal orientation (purple), (R) carvone in the flipped orientation (magenta), phenol 5 in the flipped orientation (gree n) We attempted to correlate the results of the catalytic assays with a variety of molecular distances and angles based on the crystal structures. The most informative geometric data turned out to be the distances between the ligand C8 po si tion and the closest atom of the amino acid at po si tion 116 For Trp116 variants that bound the ligand in a flipped orientation, the distances were measured directly from the X ray crystal structures. In cases where no ligand in the flipped orientation was observed but the structures were known (w.t., Gln and Ile), phenol 5 was modeled in the active si te based on alignment
78 as shown in Figure 2 18 Finally, for variants lacking crystal structures (Met, Asn, Phe, and Tyr ) both the ligand and the si de chains were modeled. We found a good correlation between the interatomic distances from C8 atom to the si de chain of amino acid 116 of and the fraction of product derived from ( Figure 2 19 ) In the case of (S) carvone, the relat ionship was surpri si ngly con si stent. If the ligand to protein distance was less than ca. 3.7 , the enzyme was essentially unable to productively bind the substrate in a flipped orientation ( Figure 2 19 ). Given the C C bond distances of C8 C9 and C10 in a ddition to the pos si ble C5 C8 bond rotation this crit ical distance likely reflects complexes where the protein substrate distance would be smaller than the ir van der Waals contact limits. steric factors. Figure 2 19 116 si de chain distance and the fraction of flipped product. The data (S) carvone and as squares ( ) for (R) carvone.
79 Three Trp116 variants showed si gnificant deviations from the predicted behavior for (R) carvone, namely Asn, Ile and Met; in each of these cases additional factors, not accounted for in the modeling are likely to contribute. The crystallographic studies of Trp116Gln showed that this substitution provoked (R) carvone to bind in a n unusual man ner It is pos si ble that the Asn si de caused si milar perturbations. In the case of the Trp116Ile enzyme, the crystal structure revealed that the isobutyryl si de chain had undergone conformational rearrangements to accommodate the bound ligand. This observa tion suggested that some steric interference occurred and in this case very minute differences in substrate structure may dictate whether or not flipping would be observed. This hypothe si s seems to offer a plau si ble explanation to the fact that Ile allowed for (R) carvone flipping but not (S) carvone. M inor steric interference coupled to several conform ers, also offer a rationalization for t he last amino acid substitution t hat escape s the overall trend namely Met Perhaps the most striking conclu si on of this study is the fact that in OYE1, the For example the wild type enzyme (Trp116) reduces (S) carvone to cis (2 R ,5 S ) dihydrocarvone with a dia stereomeric excess of 90%. In contrast, the Trp116Ile variant nearly completely invert e d the stereochemical outcome yielding a 90% d.e. of trans (2 S ,5 S ) dihydrocarvone. Why would an increase active si te volume on one si when clearly the wild type enzyme was able to carry out cataly si and only when this substrate orientation becomes precluded by a large amino acid si de chain will the redu ction product derive from a normal binding mode.
80 The obvious question will ask why, then, is flipped the preferred orientation ? To address this we decided to analyze the active si te more carefully. We calculated the pockets and cavities present in the wil d type enzyme as well as in the Trp116Val mutant si nce this variant was able to invert the stereochemistry of both (R) and (S) carvon e reductions. It was evident that this substitution had created an extra pocket that very nicely accommodated the isopropenyl group of carvone ( Figure 2 20 A). This pocket absent in the wild type ( Figure 2 20 B) had a shape nearly complementary to si de chain and was predominantly hydrophobic in nature. Figure 2 20 Hydrophobic pockets in OYE1. A The a ctive si te pocket of Trp116Val. (R) carvone is shown in green sticks. B The active si te pocket in wild type OYE1. The isopropenyl group of (R) carvone si orientation. The pockets are shown as semi transparent surfaces. A si de chain makes close contacts with Phe250, Pro295, Phe296 and Tyr375. In fact, both Tyr375 and Phe296 appear to move si gnificantly in an attempt to accommoda te this binding orientation with the latter being essentially disordered. In
81 T rp116Val, by si mply replacing the large Trp re si due with a smaller si de chain access to a new bi nding pocket is created. As noted previously this pocket has the ideal characteristics for accommodating carvone. Its shape is nearly ideal to fit the isopropenyl group. Additionally, it is formed by the Thr37, Met39, Gly72, Phe74, Tyr82, Ala85, Leu118 an d Val116 making it suited to interact with a hydrophobic substrate such as carvone. One final important observation is that upon substrate binding no appreciable movement of any of the si de chains that delineate this pocket is observed, further validating the idea of it being a more natural fit for this substrate. This work has described the discovery of OYE1 variants that are fully stereocomplementary, giving access to all four diastereomers of dihydrocarvone. A rationalization based on the wealth of stru ctural and catalytic data is also presented. The principles that govern substrate binding, and therefore stereoselectivity in OYE1, are likely to be shared by other members of this family. Hopefully the studies described herein can provide guidance and mot ivation for many protein engineering projects to come making this class of enzymes even more suitable for biocataly si s. Experimental Procedures Enzymatic Reductions Screening (Catalytic Studies) Full library screen reactions were carried out aerobically in a 96 deepwell plate containing 10 mM substrate in 0.1 M KP i pH 7.0 in a total volume of 300 L 0.3 mM NADP H was kept at a constant concentration via a regeneration system con si sting of 200 mM g lucose and 25 units/mL of GDH (glucose dehydrogenase). The catalyst loading was approximately 100 grams. Reactions were incubated at 22C for 16 hours. The reactions were extracted by addition of 500 L of ethyl acetate followed by vigorous
82 mixing. The pl ate was then centrifuged at 3,000 g for 10 minutes; finally 300 L of the organic layer was collected and used for GC MS analy si s. Charge Transfer Formation Assays Each enzyme was added to final concentration ca. 15 M in 0.1 M KP i pH 8.0. Substrate (e ither (R) or (S) carvone) was added to a final concentration of 2 mM from a 100 mM ethanol stock. The reactions were incubated for 24 hrs at room temperature (near 22 C in a rotatory apparatus for mixing The total reaction volume was 250 L. All enzyme stocks had their concentrations estimated spectrophotometrically u si ng an 462 = 11,700 M 1 cm 1 In separate experiments, each variant was incubated with 3 mM p to form charge transfer complexes as well as evaluate their spectrum at full binding saturation. This information was used to calculate the fractional charge transfer complex formation with (R) and (S) carvone. GC MS Analy si s Figure 2 21 Temperature program for GC MS analysis on DB 17 cloumn. All samples were analyzed u si ng DB 17 column (0.25 mm x 30 m) with the following temperature program: initial temperature of 90C ; followed by an i ncrease of
83 10 C/min up to 130 C; increase rate of 2 C/min up to 150 C; increase rate of 20 C/min up to 250 C for 5 minutes ( Figure 1 4 ) Typical retention times were 7.23 min for the trans (2 R ,5 R ) dihydrocarvone and trans (2 S ,5 S ) dihydrocarvone products, 7.62 min for cis (2 S ,5 R ) dihydrocarvone and cis (2 R ,5 S ) dihydrocarvone (Appendix A, Figure A 1) and 8.8 4 min for the substrates (R) and (S) carvone. (Appendix A Figure A 2 ) Appendix A shows results for typical analyses of the enzymatic reduc tions. Si te Directed Mutagene si s All mutants used for were generated by PCR u si ng primers with the de si red mutation at po si tion 116 in a modification of the method published by Zheng et al 41 Primers were de si GG GTT CAG TTA XXX GTT TTG GGT TGG GCT GCT TTC CCA GA CA ACC CAA AAC XXX TAA C TG AAC CCC AAA CGA ACG AT T TCT mutation. ) Table 2 7 lists primers used in this study. Table 2 7 Mutagenic primers utilized for constructing the enzymes used for X ray crystallography. Mutation Prim er Sequence W116A fwd W116A rev GG GTT CAG TTA GCT GTT TTG GGT TGG GCT GCT TTC CCA GA CA ACC CAA AAC AGC TAA C TG AAC CCC AAA CGA ACG AT T TCT W116L fwd W116L rev GG GTT CAG TTA TTG GTT TTG GGT TGG GCT GCT TTC CCA GA CA ACC CAA AAC CAA TAA C TG AAC CCC AAA CGA ACG AT T TCT W116Q fwd W116Q rev GG GTT CAG TTA CAA GTT TTG GGT TGG GCT GCT TTC CCA GA CA ACC CAA AAC TTG TAA C TG AAC CCC AAA CGA ACG AT T TCT W116T fwd W116T rev GG GTT CAG TTA ACC GTT TTG GGT TGG GCT GCT TTC CCA GA CA ACC CAA AAC GGT TAA C TG AAC CCC AAA CGA ACG AT T TCT W116V fwd W116V rev GG GTT CAG TTA GTT GTT TTG GGT TGG GCT GCT TTC CCA GA CA ACC CAA AAC AAC TAA C TG AAC CCC AAA CGA ACG AT T TCT Each PCR reaction (total volume 100 L) containing 5 Phu si on HF Buffer (20 L), p ET3b OYE (Appendix B) (20 ng), forward and reverse primers (0.5 M each), dNTPs (200 M each), and Phu si on Hot Start II High Fidelity DNA Polymerase (1 U)
84 was subjected to an initial denaturation step of 98 C (30 s) followed by 25 cycles of 98 C (10 s) and 72 C (4 min) followed by a final incubation at 72C (7 min). Amplicons were purified by DNA spin columns, digested with DpnI at 37C (10 U for 4 h followed by an additi onal 10 U for 4 h) then purified by an additional DNA spin column. Aliquots (5 L) were used to transform E. coli JM109 (75 L) by electroporation. SOC medium was added (600 L), then the samples were incubated for 1 h at 37 C prior to selection on LB med ium supplemented with ampicillin Plasmid DNA purified (spin columns) from randomly chosen colonies was analyzed by DNA sequencing to identify the de si red OYE1 variants. The Trp116Ile mutant was constructed by digestion of pET3b OYE and pSKP3 B30 (plasmid described in 35 ) with ApaI and KpnI. The resulting small fragment of pSKP3 B30 resulting from the digestion and contained the OYE1 W116I coding region was purified u si ng a low melt agarose gel Ligation was performed u si ng this small fragment and the large fragment of the pET3b OYE ApaI/KpnI double digest which had also been purified by a low melt agarose gel. The ligation reaction used 1.75 g of the pET3b OYE fragment and 0.243 g of pSKP3 B30 small fragment. 1 L of the resulting ligation mixture was used to transform electro competent E. coli JM109 cells (39 L) by electroporation. SOC medium was added (600 L), then the samples were incubated for 1 h at 37 C prior to selection on LB medium supplemented with ampicillin. Plasmid DNA purified (spin columns ) from randomly chosen colonies was analyzed by DNA sequencing to id entify the de si red OYE1 W116I variant Protein Expres si on and purification Plasmids were transformed into E. coli BL21 G old ( DE3 ) cells and grown in LB broth supplemented with 0.1% glucose w/w, and 200 g/mL ampicillin Cultures were
85 induced by addition of 0.4 mM isopropyl thiogalacto si de (IPTG) when the optical den si ty at 600 nm reached 0.6. The temperature was decreased to 30 C during protein expres si on. Cells were harvested after 16 h ours at a den si ty of approximately 8 g/L. Cells were then pelleted by centrifugation at 5350 x g for 10 minutes and the pellet was resuspended to 1g/mL in 100mM Tris HCl pH 8.0containing 10 M phenylmethanesulfonyl fluoride (PMSF). All of the subsequent steps were performed at 4 C. The resuspended cells were lysed at 15,000 p si in a French pressure cell u si ng a Carver laboratory press. Lysed cells had debris removed by centrifugation at 20,000 x g for 1 hour. To remove nucleic acids, protamine sulfate (1 mg/mL) was added. Nucleic acids were centrifuged out at 10,000 x g for 15 min. The cell lysate had its pH adjusted to 8.5 by the addition of ammonium hydroxide. Solid ammonium sulfate was added to 78% saturation level and the lysate was centrifuged at 20, 000 x g for 1 hour. The precipitate was resuspended in minimal amount of 100mM Tris HCl, 100 mM ammonium sulfate, 10 M PMSF at pH 8.0, and dialyzed against the same buffer for 6 hours (buffer was exchanged after 3 hours). The lysate was then dialyzed agai nst a new buffer of the same recipe containing an additional 10 mM sodium dithionite. This step served the purpose of removing any small phenolic type molecules that ma y be complexed with the enzyme. This step is also referred to as de greening in allu si on to the long wavelength ch arge transfer complexes of OYE1 One last dialy si s was performed to ensure removal of all of the sodium dithionite, and that the enzyme would return to its oxidized state before chromatography. The enzyme was purified by affinity chromatography u si ng an N ( 4 hydroxybenzoyl) aminohexyl agarose gel matrix. The column was packed with about 3 mL of the matrix and connected to a Pharmacia FPLC
86 system. The column was equilibrated with buffer containing 100mM Tris HCl, 100 mM ammonium su lfate pH 8.0. Cell extract was loaded onto the column and it was washed with buffer until the absorbance at 280 nm became stable at approximately 0.2 A.U. For elution, the column was washed with 3 volumes of buffer supplemented with 6 mM sodium dithionite. The elution buffer was deoxygenated and flushed with nitrogen gas before addition of sodium dithionite The enzyme eluted upon reduction by dithionite. Enzyme was collected in four 3 mL fractions. Fractions were pooled and concentrated by centrifugation u si ng a 10,000 MWCO membrane. For regeneration of the affinity matrix, the column was washed with 3 volumes of buffer containing 6 M guanidine HCl, 0.2 M sodium acetate at pH 5.0. Purified protein was then loaded onto a Superdex 200 HR 10 30 (Pharmacia) fo r further purification. The c olumn was equilibrated with low ionic strength buffer, 50 mM Tris HCl, 50 mM NaCl, 10 M PMSF. Only the major peak at 280 nm was collected. A typical run would con si st of a 10 mg of protein loading and after this step the prote in samples were homogenous as indicated by SDS PAGE ( Appendix C ) Crystallog raphic Studies Crystals were obtained u si ng by a hanging drop vapor diffu si on method at 4 C. The protein solution had approximately 40 mg/mL protein in 50mM Tris HCl, 50mM NaCl, 10 M PMSF pH 7.5. The reservoir solutio n contained 35% PEG 400, 200 mM MgCl 2 and 100 mM HEPES pH 8.3. Crystal usually formed in about 5 to 7 days and would grow for another 7 to 14 days. Crystals grew as yellow tetragonal rods typically diffracted to 2.0 1.2 and measured on average 180 m x 180 m x 5 00 m. For substrate soaking experiments, the crystals were soaked in mother liquo r with the
87 addition of saturati n g amounts of ligand for varying amount of time depending on the experiment. Data collection w as performed under cryogenic conditions at 100 K in Brookhaven National Laboratory beamlines X25, X6A and X26C. Crystals were mounted on nylon loops and plunged into liquid nitrogen prior to data collection. The best results were obtained when no further c ryo protection protocol was employed. The mother proved suitable for cryogenic data collection. These crysta s l were also si sts of allowing a cryo cooled crystal to warm up to room temperature followed by flash cool ing. In many cases, this was an efficient way of minimizing ice rings seen on the diffraction pattern. Structure Solution X ray data were processed u si ng HKL2000 or iMOSFLM 42 43 .The structures were solved by molecular replacement u si ng AUTOMR in the PHENIX suite 44 The search used a high resolution model of OYE1 (PDB code 3RND) The initial search model was devoid of water molecules and any ligands. Additionally, re si due 116 was initially modeled as alanine. Inspection of the electron den si ty maps was followed by manual rebuilding, addition of ordered waters and refinement of ea coordinates and temperature factors against the X ray data. During refinement, inspection of the calculated 2 m Fo D Fc and m Fo D Fc difference maps showed clear po si tive electron den si ty for the si de chain of re si due 116 and above the isoalloxazine moiety of the bound FMN cofactor. This allowed for building of the correct re si due 116 si de chains and unambiguous placement of the carvone ligands. This process was performed iteratively until re si duals errors R work and R free converged. The complete crys tallographic parameters, data collection and refinement statistics can be found in Results and Discus si on
88 Enzyme substrate Complex Modeling The complexes generated were W116F/M/N/P/Q/Y and OYE1 w.t. bound to inhibitor (phenol 5 vide supra ) The enzyme inhibitor complexes were generated by manually building each si de chain in COOT 45 and clashes were calculated u si ng the built in MOLPROBITY 39 utility The initial po si tion of the si de chains were selected on the ba si s of: a) the likelihood of the rotamer; b) the si milarity to an amino acid in an existing crystal structure model, e.g. phenylanlanine was placed si milarly to tryptophan. Because of the constraints regarding po si tion of the ligand such as, hydrogen bonding to N194 and H191 distance to the FMN is oalloxazine ring an d distance to the proton donor Y 196, the ligand was manually placed in a po si tion si milar to the observed in the crystal structures. The resulting enzyme inhibitor models were then subjected to geometry optimization/energy minimization u si ng REFMAC5 46 The resulting re fined models were then inspected for validity. If shifts greater than 0.20 in the inhibitor final coordinates were observed, the model was not con si dered for the measurements as it would likely represent an unproductive complex. This was the case for the W116Y and w.t. enzymes and si mple least squares superpo si tion was used to estimate steric clashes. The model of W116P was not used for measurements at all as energy minimization could not be performed successfully. This likely comes from the fact that mut ation to a P116 would cause major backbone rearrangement in the protein and energy minimization procedures failed to find suitable conformations.
89 CHAPTER 3 STRUCTURAL AND CATALYTIC CHARACTERIZATION OF P stipitis OYE2.6 A USEFUL CATALYST FOR ALKENE REDU CTIONS Background and Motivation O ur group was highly interested in discovering OYE family enzymes that woul d offer enantiocomplementarity as a chieving such goal can provide a solution to one of the key factors limits the applicability of enzymes in synthetic processes. The previous chapter discusse s how s ome of our earliest protein engineering studies revealed that replacing an active si te Trp re si due (po si tion116) with Ile radically changed the stereoselectivity of S pastorianus OYE1 Initially this observation seemed to be unique to the reduction of carvones, however further investigation showed it was not an isolated occurrence 47 Our group tested the stereoselectivity of OYE1 Trp116 variants in the reduction of three substrates that are useful intermediates in a number of synthetic routes 48 49 50 Once again, po si tion 116 was able to influence the enantioselectivity and the outcome depended on the nature of the Trp116 replacement ( Figure 3 1). Interestingly, Trp116 is a well conserved re si du e in the OYE1 family ( Figure 3 2 ) W e therefore pursued old yellow enzyme homologs posses si ng re si dues other than Trp at the po si tion corresponding to 116 of S. pastorianus OYE1. This l ed t o the cloning and overexpres si on of OYE 2.6 from the xylose fermenting yeast Pichia stipitis CBS 6054 (Bougioukou PhD the si s). Based on sequence alignment and computer modeling from the S. pastorianus OYE1 structure, Ile 113 in OYE 2.6 was expected to occupy a si milar active si te location as Trp 116.
90 Figure 3 1. Reductions of Baylis Hillman Adducts by Alkene Reductases On this ba si s, we expected th at the stereoselectivity of OYE 2.6 would be analogous to that of the S. pastorianus OYE1 Trp 116Ile mutant however, the reality proved more complex. In general, OYE2.6 has proven to be a superior alkene reductase with regard to stereoselectivity, conver si on rate and stability and was successfully employed in the reductions of citral 51 In some cases, its stereoselectivity was oppo si te that of S. pastorianus OYE1, but this was not univer sally true. For example OYE1 reduces 2 hydroxymethylcyclopent 2 enone to the ( R ) product mainly In contrast OYE2.6 afford ed the ( S ) enantiomer and th is was in accord with our initial hypothe si s. However,
91 curiously, OYE2.6 fails to display a si milar inv er si on of stereochemistry in the reduction of (S) carvone ( Figure 3 3 ). Figure 3 2 Partial sequence conservation in seven OYE family representatives. OYE2.6, OYE1, estrogen binding protein (EBP1), oxophytodienoate reductase (OPR1), pentaerythritol tetranitrate reductase (PETNR), morphinone reductase (MR) and N ethylmaleimide redu ctase (NemA). Po si tion 113 in OYE 2.6 (116 in OYE1) is indicated by an arro w. Figure 3 3 Comparison of OYE1 and OYE2.6 stereoselectivity. Despite exhibiting superior and sometimes complementary catalytic properties t here are currently two impediments to the wider use of OYE2.6 in chemical synthe si s: a relatively limited kno wledge of substrate and stereoselectivity c haracteristics and an unknown three dimen si onal structure. Because the latter is particularly critical for protein engineering studies of this enzyme, we undertook the elucidation of the X ray
92 crystal structure of OYE 2.6. In addition, the results of four si te saturation mutagene si s libraries that allow for meaningful comparisons with S. pastorianus OYE1 are presented. These will point the way toward future a pproaches to obtaining even more useful variants of OYE 2.6. Results and Discus si on Native OYE2.6 was overexpressed in E. coli BL21 Gold (DE3) carrying the plasmid pBS3. The enzyme could be purified by ammonium sulfate fractionation followed by affinity and gel filtration chromatography. This afforded a prepara tion of the enzyme of >9 5 % purity After screening approximately 300 conditions, crystals were observed when organic salts were the main precipitants at near neutral pH values. The most promi si ng, obtained in 2.4 M sodium malonate, pH 7.0, diffracted to a maximum resolution of 2.0 u si ng a Rigaku Cu rotating anode X ray source. Unfortunately, these proved unsuitable for crystallographic studies as they were not si ngular and yielded diffraction patterns that could not be indexed accurately. Because attempts at optimizing pH and salt concentrations proved unsuccessful, a further screening of 96 additives coupled with the original conditions (2.4 M sodium malonate, pH 7.0) was carried out. The best crystals were seen in the original conditions suppleme nted with 2% 2 propanol and formed from a ca. 40 mg/mL protein solution. After adequate cryoprotection the crystals diffracted to 1. 4 u si ng synchrotron source radiation at 100 K T he crystals belonged to space group P 6 3 2 2 (number 182) and t he unit cel l measured 127.1 127.1 , 123.4 The asymmetric unit con si sted of one OYE2.6 molecule with a solvent content of 63% with a Matthews coefficient of 3.23 3 /Dal 52 ( See Experimental for more details )
93 Overall Fold The initial OYE2.6 structure was solved by molecular replacement u si ng a modified ver si on of S. pastorianus OYE1 as the search model (1OYA). The X ray data yielded a high quality model that was refined to 1.5 resolution with final Rwork and Rfree values of 0.143 and 0.168, respectively ( Table 3 1 ). The enzyme belongs to the TIM phosphate binding family and exhibits the eight stranded / barrel fold characteristic of all OYEs first seen in triosephosphate isomerase (TIM) 53 and glycolate oxidase (GOX). The overall structure of OYE2.6 is si milar to those of other old yellow enzyme homologs, particularly to th at of S. pastorianus OYE1 ( Figure 3 3 ) These share 43% sequence identity and also have closely related backbone po si tioning (aligning 325 carbons with a root mean square deviation of 0.92 ) Figure 3 4 Superpo si tion of the structures of OYE2.6 (cyan) and OYE1 (orange) The regions of highest dis si milarity are shown in cartoon representation and are labeled
94 The most important structural difference between P. stipitis OYE2.6 and S. pastorianus OYE1 is found in loop L6 (Glu287 Ala302) ( Figure 3 4 ). Finally, the crystal structure also reveals that in OYE2.6, Ile113 is spatially equivalent to Trp116 in OYE1 as had been surmised initially. we carried out structural comparison s to othe r members. After S. pastorianus OYE1, t he n ext most si milar structures are Pseudomonas putida morphinone reductase (MR, PDB code 1GWJ) 16b (291 aligned, 1.26 r.m.s.d.), Enterobacter cloacae pentaerythritol tetranitrate reductase (PETNR, 1H50) 16c (232 C aligned, 1.38 r.m.s.d.), followed by Bacillus subtilis YqjM (1Z41) 16d (191 C aligned, 2.57 r.m.s.d.) (Appendix D) These results suggest that OYE2.6 and OYE1 can be grouped into the same class of OYE homologs Quaternary Structure Be si des variations in active si te re si dues and architecture, differences in quaternary structure are also an aspect that can help in the clas si fication of OYE family enzymes. For instance, S. pastorianus OYE1 is dimeric while its plant homolog OPR1 is monomeri c 54 I n contrast the bacterial enzyme TOYE exists as an octamer and even higher oligomer ic states 32 In an attempt to unambiguously clas si fy OYE2.6 we attempted to characterize its quaternary structure through native gel filtration and analy si s of the molecular contacts in its crystalline form Under our experimental conditions, homogenous recombinant OYE2.6 had an elution volume con si stent with that of a n 89 kDa assembly (Experimental Procedures). Given that OYE2.6 is a 407 amino acid protein wi th a theoretical molecular weight of 46.1 kDa this result strongly
95 oligomeric state in its crystal form we had to study the crystal contacts of each protomer si nce the crystal lographic asymmetric unit con si sts of a si ngle molecule. One particular interaction buries the most surface area ( total of 1520 2 ) and is composed of two protomers related by a two fold crystallographic symmetry operator suggesting this could be the dime ric interface. This interface involves heli x 4 (re si dues 212 216) helix 5 (252 260) and helix 6 (263 268, 271) in both molecules ( Figure 3 5 ). As these helices are nearly perpendicular to the two fold axis, each helix interacts with its symmetry mate and are the same helices as those seen in S. pastorianus OYE1 dimer 14 Figure 3 5 The proposed OYE2.6 dimer. The 2 fold molecular a xis is indicated by a dashed line. Helices 4, 5 and 6 are label. The orientation of the crystallographic axes are shown for reference. As previously noted by Fox and Karplus, t he dimer interface area r epresents about 5% of the total surface area of an OYE monomer and is near the lower end of valu es
96 seen for dimeric proteins 55 This may be one of the r easons why software prediction by PISA 56 results in a different dimeric interface although no si ngle solution is heavily favored. OYE2.6 crystallizes in space group P 6 3 2 2 having 12 symmetry operators and each monomer makes contacts with five other symmetry mates ( Figure 3 6 A ). Dis si milarly, OYE1 crystallizes in P 4 3 2 1 2 and makes con tacts with seven symmetry mates ( Figure 3 7 A ). Figure 3 6. OYE2.6 crystal lattice. A Crystal packing of OYE2.6 in P 6 3 2 2. viewed along the crystallographic c axis. The unit cell is dr awn as a black polygon and the dimer is shown in purple. B Magnified view of the two molecules forming the homodimer. D espite crystallizing in different space group s, the relative po si tions of the two monomers forming the dimers in OYE2.6 and OYE1 are nea rly identical (figures 3 6 B and 3 7 B). The same helices, namely, helix 4, 5 an d 6, interact with their symmetry mate equivalent in the crystal structure of both enzymes. This obs ervation not only supports the structure of the dimeric assembly in OYE2.6 but also suggests that OYE2.6, like OYE1, has evolved these specific contacts t o exist as a dimer Taken together these results support the notion that OYE2.6 and OYE1 are indeed in the same class of fungal OYEs.
97 Figure 3 7 OYE1 crystal lattice. A Cry stal packing of OYE1 in P 4 3 2 1 2 viewed along the crystallographic c axis the dimer is shown in purple The unit cell is drawn as a black polygon. B Magnified view of the two molecules forming the homodimer. Flavin Environment and Phenol Binding The flav in mononucleotide cofactor was readily apparent in the initial electron den si ty maps. As anticipated, the FMN was well ordered and displayed well defined electron den si ty and low temperature factors ca. 17 2 ( 22 2 for protein atoms) con si stent with the K D values near 10 10 M usually seen in OYE enzymes The cofactor is bound in the s 8 strand parallel barrel and makes exten si ve contacts with several highly conserved re si dues ( Figure 3 8 ) negatively ch arged phosphate group makes a salt bridge with Arg 347 bond with both water 31 and Arg 240, whose si de chain The dimethyl benzene moiety makes contacts with the aromatic re si dues Phe373 and Tyr 374 and its isoall oxazine ring is po si tioned to form direct hydrogen bonds with Gln 111 Thr35, Arg 240 and the main chain nitrogen of Ala68. Gln 111 is absolutely conserved in the OYE family and this si d e chain, along with that of Thr 35, is implicated in regulating the FMN reactivity in S. pastorianus OYE1 17c 17a
98 Figure 3 8 The FMN biding environment and direct interactions. Hydrogen bonds are shown as black dashes. We used spectral changes that accompany phenol binding to OYE2.6 to characterize its active si te character and cofactor properties. OYE2.6 bound both p chlorophenol and p hydroxybenzaldehyde tightly and these complexes were accompanied by long wavelength charge transfer bands with max at 600 and 555 nm, respectively. Massey determined that the po si tions of these charge transfer bands for S pastorianus OYE1 depended upon both hydrogen bonding interac tions with the phenol(ate) oxygen and the FMN reduction potential 8 In OYE1, the phenolic oxygen interacts with the si de chains of His 191 and Asn 194; by contrast, two histidines occupy these po si tions in OYE2.6 (His 188 and 191). Compared with the corresponding values for S. pastorianus OYE1, max values for both p chlorophenol and p
99 hydroxybenzaldehyde interacting with OYE2.6 are blue shifted by 59 and 27 nm, respectively. Relatively small effects on charge transfer max values were observed from changing phenol(ate) hydrogen bonding partners in OYE1, whereas substituting the enzyme with FMN analogs posses si ng different reduction potentials led to more si gnificant wavelength shifts 57 Given the si milarity in active si te structures between OYE2.6 and OYE1 particularly in re si dues that interact directly with the flavin the differences in charge transfer max values are therefore more likely to reflect small differences in flavin local environment rather than a si gnificant redox potential difference U si ng the nearly linaer relationship between FMN E max values for p chlorophenol and p hydroxybenzaldehyde established for S. pastorianus OYE1, we estimate d that the E value for OYE2.6 is ca. 248 mV. This is si milar to tha t measured for wild type OYE1 ( 230 mV). Figure 3 9 Complex formation between OYE2.6 and p CP. A The spectral changes in OYE2.6 upon ligand binding. The concentration of p CP was increased from 0 to 120 M. B Binding saturation curve and non linear fit.
100 OYE2.6 bound p chlorophenol with and the formation of the charge transfer max of 600 nm. The binding saturation could be fit u si ng a specific b inding non linear regres si on model (GraphPad). The estimated K D is 9.8 M 600 of 5170 M 1 cm 1 ( Figure 3 9 ). In contrast, OYE1 forms a charge transfer complex with p max 645 4400 M 1 cm 1 40 In the case of p hydroxybenzaldehyde, as expected, a charge transfer complex formation max at 555 of approximately 5400 M 1 cm 1 OYE2.6 titration with a solution of p h ydroxybenzaldehyde resulted in binding saturation behavior that was best interpr eted as very tight binding with a K D value being con si derably smaller than the enzyme concentration of 2 4 M ( Figure 3 10 ). Figure 3 10 Complex formation with p HBA. A The spectral changes in OYE2.6 upon ligand binding. The concentration of p HBA was increased from 0 to 120 M. B Binding saturation curve and non linear fit. The poor fit is due to an observed infinitely tight binding saturation. The Active Si te and Ligand Binding As mentioned before, the active si te of OYE2.6 displays some si gnificant differences relative to that of OYE1. First, the pair of hydrogen bond donors in the active si te is composed of two histidine re si dues (188 and 191) instead of a hi si tidine and an
101 asparagine. As had been predicted based on sequence alignment an Ile (113) replace s a tryptophan and is very close to the bound ligands. Loop L6, which usually exhibits large plasticity across the OYE family, is also si gnificantly different The portion of the loop that defines the active si te has an entirely different amino acid sequence. Interestingly, in OYE2.6 a glycine re si due (292) replaces a proline (Pro295 in OYE1). Given their largely different phy si cal characters, this substitution is very likely to be a factor in the substrate specificity differences between t hese two enzymes ( Figure 3 11 ). Figure 3 11 The active si te comparison of OYE2.6 and OYE1. OYE2.6 re si dues are shown as thick lines and cyan carbons, while OYE1 re si dues are in thin lines with orange carbons. The labels indicate the re si dues in OYE2.6. The initial OYE2.6 crystals displaye d strong electron den si ty peaks directly above the si face of the FMN This density region appeared symmetrical and had its strongest peak in optima l position to interact with His 188 and 191 Since no ligands were
102 delib erately added throughout purification or crystallization steps, we turned our attention to buffer components. Malonate, present in the crystallization buffer at 2.4 M, was a near perfect f it for this density ( Figure 3 12 A ) In this complex, one carboxylate oxygen is po si tioned to form hydrogen bonds with the si de chains of histidines 188 and 191, which likely mimics their interactions with the carbonyl oxygens of conjugated alkene substrates as previously reported by us (previous chapter) and others. Figu re 3 1 2 OYE2.6 malonate complex. A Crystal structure of OYE2.6 in complex with malonate. B The conformation of malonate bound in the active si te. C Conformation of malonates seen on the surface of protein. The omit mFo DFc map of malonate contoured at DFc map of the Protein re si dues are colored in cyan, FMN in yellow and the ligand in green The other carboxylate oxygen form s hydrogen bond with an ordered water molecule (WAT301) which in turn is hydrogen bonded to the si de chains of Tyr78 and Th 35. Curiously, w hile several well ordered malonate molecules were observed in the surface
103 of the protein, only this active si te malonate was found in the higher energy planar conformation ( Figur e 3 12 B). The fact that the active site malonate appears in a higher system of the flat flavin ring. A ll solvent exposed malonate molecules were observed in the expected bent form ( Figure 3 12 C) Crystals grown from a second preparation of OYE2.6 also showed additional electron den si ty in the active si te region; however, the shape differed si gnificantly from that observed previously This density suggested the species was a six membered ring that appeared aromatic. Additionally, some extra density suggested th at the molecule carr ied a three atom group with roughly trigonally planar geometry. These observations would be consistent with the shape of molecules such as benzoic acid, nitro benzene, nic otinic acid and nicotinamide among others. Figure 3 1 3 OYE2.6 nicotinamide complex. A Crystal structure of OYE2.6 in complex with nicotinamide. The o mit mFo DFc map as a red mesh. The 2mFo Protein re si dues are colored in cyan, FMN in yellow and the ligand in green B Alignment of structures of OYE homologs containing nicotinamide derivatives.
1 04 Given that the ident ity of the ligand would be constrained by what molecules are likely to be present throughout the protein overexpression, isolation and crystallization processes, we deduced that it was free nicotinamide. This ligand could be successfully fit into the elect ron density and was refined to an occupancy of 95% ( Figure 3 13 and Table 3 1 ). This fortuitous complex likely mirrors that between OYE2.6 and its natural substrate, NADP H. The amide carbonyl oxygen is located within hydrogen bond distances of the imidazol e si de chains of both His188 and His191 and the distance and angle between the nicotinamide C4 and the FMN are con si stent with the geometry of hydride transfer (3.4 and 94 respectively) 37 This ligand conformation is also in agreement with that of a previously reported OYE1 complex with an NADP + analog 58 Further supporting the notion that this a rrangeme nt mimics the productive NADPH/OYE2.6 Michaelis complex, the n icotinamide binding orientation within OYE2.6 is analogous to those observed for tetrahydro nicotinamide within the active si tes of T pseudethanolicus TOYE 32 and morphinone reductase 59 ( Figure 3 13B) Unfortunately, neither UV Vis nor fluorescence s pectroscopy yielded usable spectral changes for nicotinamide binding to OYE 2.6, so its binding constant could not be measured by these methods. To provide a model for how potential substrates might bind to the active si te and for further comparison to its homolog OYE1, we attempted to obtain a crystal structure of the complex between OYE2.6 and p chlorophenol by soaking crystalline enzyme in cryoprotectant buffer containing 2 mM. The OYE2.6 crystals were very sensitive to osmotic pressure and osmolality ch anges which made soaking experiments very difficult. Long soaking times led to increased diffraction mosaicity and crystal cracking.
105 The best crystal used for data collection resulted from a short ( ca. 2 seconds ) soaking time. This crystal diffracted to a resolution of 2 and the electron density maps revealed the position of the p chlorophenol inhibitor ( Figure 3 1 4 and Table 3 1 ). The phenolic oxygen was po si tioned between the si de chains of His188 and 191 and a meta ring carb on was 3.5 above the FMN N5 atom Figure 3 1 4 Crystal structure of OYE2.6 in complex with p chlorophenol The omit mFo DFc map of the active Protein re si dues are co lored in cyan, FMN in yellow and the ligand in green carbons of OYE2.6 substrates would be in an analogous location. The angle between this carbon and N5 and N10 of the FMN was 105 These values are con si stent with those expected for efficient hydride transfer to a cyclohexenone carbon. As the model neared completion, some areas of positive electron density remained and could be best
106 explained by a part ially (n ear 35%) occupied malonate molecule. This is probably a result of the high concentration of malonate in the crystallization buffer and the short soaking times with p chlorophenol. Table 3 1 X ray Crystallographic Data Collection and Refinement Statisti cs Accession code 3TJL 3UPW 4DF2 A ctive site ligand malonate nicotinamide / malonate p chlorophenol X ray source NSLS X25 NSLS X25 Space group P 6 3 2 2 P 6 3 2 2 P 6 3 2 2 Unit cell dimensions a= b, c () 127.16, 122.45 90, 90, 120 127.30, 123.06 90,90, 120 127.73 123.64 90, 90, 120 Resolution () 44.10 1.50 44.24 1.78 24.86 2.02 Unique reflections 92,358 (2,365) 1 56,484 (1,608) 38,149 (1,269) Completeness (%) 99.07 (96.0) 99.76 (96.0) 96.70 (99.0) Redundancy 7.6 (4.7) 10.5 (4.2) 7.2 (6.9) R merge 2 0.056 (0.60) 0.047 (0.55) 0.093 (0.70) 33.4 (2.0) 37.1 (2.1) 14.2 (2.1) R work 3 R free 4 0.143, 0.168 0.157, 0.185 0.160, 0.198 Ramachandran statistics 5 Favored (%) Allowed (%) Outliers (%) 97.60 2.40 0 97.80 2.20 0 97.52 2.48 0 Number of protein, solvent and ligand atoms 3,238, 517, 52 3,211, 445, 65 3,218, 394, 53 Average B factors ( 2 ) Protein Solvent FMN Ligands 22.21 35.20 16.22 33.30, 28.40, 23.10 23.20 34.40 19.17 34.90, 31.80, 38.0 (malonate); 21.70 (nicotinamide) 31.40 38.87 26.21 39.30, 26.20 (malonate); 27.01 ( p chlorophenol) a] Values in parentheses denote data for the highest resolution bin. [b] Rmerge = hkl i |Ii(hkl) [I(hkl)]|/ hkl i Ii(hkl), where Ii(hkl) is the inten si ty of the ith observation of unique reflection hkl.
107 [c] R work |F o (hkl)| |F c (hkl)| o (hkl)|. [d] R free is calculated in the same manner as R work u si ng 10% of the reflection data not included during the refinement. [e] Statistics generated u si ng M OLPROBITY 39 Catalytic Characterization through Mutagenesis After structural characterization of OYE2.6 was completed, we turned our attention to probing its structure/function relationships with respect to alkene reducti ons. We chose three representative Baylis Hillman adducts 1 3 for catalytic studies of OYE2.6 ( Figure 3 15 ) since they possess additional functionality that makes them suitable for further useful chemical transformations following asymmetric alkene reducti ons 48 50 Figure 3 15 Structures of the three Baylis Hillman adducts tested. While a number of catalytic hydrogenation strategies for the asymmetric reduction of 3 has been reported 60 and references therein we are not aware of simi lar catalysts being applied to cyclic enones 1 and 2 In addition, we have previously used the same series for characterizing S. pastorianus OYE1 mutants, thereby facilitating direct c omparison of their substrate conversions and stereoselectivity 47 F our site saturated mutagenesis libraries of OYE2.6 were created using a variation of the method published by Zheng and Reymond employing a reduced genetic alphabet (NNK) that contained 32 codons but still encompassed all twenty amino acids 41 After extensive optimization b y Drs. Adam Walton and Bradford Sullivan in our laboratory libraries
108 routinely contained > 16 amino acid replacements (usually 18 or 19). After verifying their diversity at the pooled plasmid DNA stage, this mixture was used to transform the E. coli expres sion host In the case of the Ile 113 library, simple site directed mutagenesis methods were used to obtain the three missing codons so that a complete set of amino acid replacements could be screened. Enones 1 3 were reduced under whole cell conditions in the presence of glucose for NADPH regeneration by host metabolic pathways. In essence, His188 in OYE 2.6 showed little tolerance toward amino acid substitutions and only the wild type protein was highly active ( Figure 3 16 A) The Ser, Gln and Asn variants reduced substrates 1 and 2 poorly and showed no detectable activity toward s 3 All variants that retain ed some catalytic activity showed the same ( S ) stereoselectivity as seen in the wild type enzyme. In S. pastorianus OYE1, the corresponding position (19 1) is also occupied by His and its function was probed by Brown and Massey In their study, t his residue was changed to Asn and while the variant retained catalytic activity, the K M value for 2 cyclohex enone increased significantly 40 The differential response toward analogous substitutions highlights the subtle differences in active site architecture s even among highly similar alkene reductases as OYE family enzymes Despite the lack of success in identifying functional histidine replacements in this particular case, others have found that site saturation libraries can sometimes reveal un expected results 61 Scrutton mutated the corresponding His residue in E. cloacae PETNR (position 181) and found that this library had very low average catalytic activity when tested against carbonyl conjugated
109 alkenes 62 On the other hand, several useful replacements for this same PETNR variant were identified for certain nitroalk ene substrates 63 Figure 3 1 6 Mutations of positions 188 and 191. A Effects of mutations of amino acid His188. B Effects of m u tations of position His191. The enone substrates 1 2 and 3 are colored in green, red and blue respectively. The wild type enzyme is indicated by an arrow. Grey letters indicate substitutions that were missing in the library used for screening. Position 191 in OYE 2. 6 was slightly more tolerant toward mutation, with both the native His and the Asn mutant proteins showing significant catalytic activity while n o changes in enantioselectivity were observed for the three enones investigated here ( Figure 3 16 B) As observed in the previous case, the catalytic activities of the mutant proteins were generally consistent across all three enone substrates. This underscores the point that the catalytic assays reveal general characteristics of the enzymes and suggests that similar behavior will apply to other potential substrates as well. The analogous position in S. pastorianus OYE1 (194) contains Asn and i nterestingly, att empts to prepare the single Asn 194His mutant of S. pastorianus OYE1 were unsuccessful 40 The analogous His in E. cloacae PETNR (position 184) also showed
110 limited tolerance for amino acid replacem ent, although the Asn mutant retained 20% of the wild type activity for 2 cyclohexenone In both OYE 2.6 and S. pastorianus OYE1, the side chain of a key Thr residue forms a hydrogen bond with O4 of the bound FMN. For OYE1, loss of this interaction slowed the reductive half reaction by an order of magnitude but increased the oxidative half reaction in the Thr37Ala mutant. Other OYE family members contain Thr or Cys at the analogous position. Based on these precedents, we created and examined a systema tic re placement library for Thr 35 in OYE2.6 ( Figure 3 1 7 ). Figure 3 17 Effects of mutations of amino acid Thr35. The enone substrates 1, 2 and 3 are colored in green, red and blue respectively. The wild type enzyme is indicated by an arrow. Grey letters indic ate substitutions that were missing in the library used for screening. For OYE2.6 the Ser replacement had essentially the same properties as the wild type enzyme, consistent with the importance of hydrogen bonding with the FMN cofactor. The Ala variant a lso possessed activity, albeit diminished. The most surprising result was that several other residues (Cys, Leu, Met, Gln and Val) could also functionally replace Thr 35, though with lower efficiencies. No change in
111 stereoselectivity was found for any of t he Thr 35 substitutions. The observed tolerance toward changes at this position bodes well for f uture work in active site remodeling. Our previous studies of S. pastorianus OYE1 highlighted the key influence of Trp116 in controlling reactivity 35 47 and similar observations have been made for both E. cloacae PETNR 62 and B. subtilis YqjM 64 As noted above, the corresponding position in native OYE2.6 (113) is occupied by Ile. A number of amino acids replacements were tolerated by OYE2.6 at this position ( Figure 3 18 A ). Small to medium hydrophobic residues (Ala, Leu, Met, Phe and Val) as well as polar, uncharged amino acids (Asn, Cys, Gln, His, Ser and Thr) could replace the native Ile with good to excellent retention of catalytic activity. Figure 3 1 8 Mutations of p ositions 113 and 116. A Effects of mutations of amino acid Ile113. B Effects of motations of position Trp116 in OYE1. The enone substrates 1, 2 and 3 are colored in green, red and blue respectively. The wild type enzyme is indicated by an arrow On the o ther hand, charged amino acids, Pro and larger hydrophobic residues ( Tyr and Trp) yielded inactive proteins toward the Baylis Hillman adducts examined here. One difference between the position 113 mutants and the libraries described above is that
112 conversio ns differ significantly between the two cyclic enones and the acyclic Roche ester precursor. With regard to substrate conversion, the response of OYE2.6 toward amino acid substitutions at position 113 closely mirrors that of a Trp116 replacement library in S. pastorianus OYE1 ( Figure 3 18 B) 47 Like O YE2.6, OYE1 prefers cyclic enones 1 and 2 over acyclic alkene 3 and similar ranges of residues at position 116 are acceptable substitutes for the native Trp. In some cases, this similarity extends to changes in stereoselectivity. For example, wild type OYE 1 (with Trp at position 116) reduces 2 with predominantly (R) selectivity (60% e e ); by contrast, the Trp11 6 Ile mutant afforded 91% ee (S) product 47 For OYE2.6, wild type (with Ile at position 113) reduced 2 with high (S) selectivity (>90% ee) while the Ile11 3 Trp mutant provided 11% ee favoring the (R) enantiomer. However, o ne important differenc e between the behavior of OYE2.6 and S. pastorianus OYE1 mutants involves the acyclic Roche ester precursor 3 Wild type OYE2.6 and all of the Ile113 mutants that accept this s ubstrate catalyze reductions with essentially complete (S) selectivity, although some with relatively poor conversion. By contrast, mutating Trp116 in S. pastorianus OYE1 provides a range of stereoselectivities for this substrate, ranging from >98% ee (R) (wild type) to >98% ee (S) (Trp116 Gln These differences highlight the distribu tive nature of the substrate binding pocket within alkene reductases. While single amino substitutions can be sufficient to alter stereoselectivity in some favorable cases, multiple substitutions are more often required. Comparing the structures of OYE2.6 and S. pastorianus OYE1 in light of these mutagenesis data provides a much better roadmap for future protein engineering studies.
113 Conclu si ons and Future Direction A major motivation for determining the crystal structure of OYE2.6 was to understand i ts stereoselectivity, particularly why it is generally oppo si te to that of wild type S. pastorianus OYE1 for Baylis Hillman adducts 1 3 but identical to that of the OYE1 Trp 116I le m utant These observation s in addition to the fact that the crystal structu re revealed this enzyme possesses an Ile in place of a Trp for OYE1 would suggest that OYE2.6 is essential ly a mimic of OYE1 Trp116Ile mutant. However in catalyzing the reductions of certain substrates, such as (S) carvone, this simplification proves incorrect and OYE2.6 does not share the stereoselectivity of OYE1 Trp116Ile. We were intrigued by this behavior and hoped that elucidation of its three dimensional structure would enable us to rationalize these results We decided to directly compar e the structures of OYE2.6 with that of OYE1 Trp113Ile with 2 hydroxymethyl 2 cyclopentenone 2 bound in the active site Interestingly, i n two of the three OYE 2.6 structures solved an ordered water molecule which is absent in OYE1 structures hydrogen bo nds to the si de chains of Thr 35 and Tyr78 ( Figure 3 12 A and 3 13 A ). When the flavin cofactors of the OYE2.6 /malonate complex and the S. pastorianus W116I OYE1/enone 2 complex were overlaid, the hydroxy l group of 2 was in the same location as these ordere d water molecules in OYE 2.6 ( Figure 3 1 9 ). Our previous results have argued that this protein substrate interaction in S. pastorianus OYE1 plays a key role in determining which face of the system will accept hydride from reduced FMN. It is therefore reasonable to suggest that an analogous hydrogen bond is also formed between OYE2.6 and the hydroxymethyl groups of 1 3 and this provides a structural rationale for the observed stereoselectivity. We carried out structural alignments of OY E2.6 with wild type and the
114 Trp 116I le mutant S. pastorianus OYE1, P. putida morphinone reductase (MR ), E. cloacae pentaerythritol tetranitrate reductase (PETNR), B. subtil i s YqjM, T pseudethanolicus E39 TOYE and tomato oxophytodienoate reductase OPR1 to determine whether these proteins also contain an analogous ordered water molecule. Figure 3 1 9 Comparison of OYE 2.6 and S. pastorianus OYE1 complexes. The structures of native OYE 2.6 with bound malonate (cyan) and the structure of the S. pastorianus W116I OYE1 mutant with enone 2 bound (orange) were overlaid using all FMN atoms (yellow carbons). Key side chains and an ordered water molecule are shown. Numbers correspond to OYE2.6 sequence. Interestingly, this water was only observed near the FMN in O YE2.6 In the case of OYE1, MR, PETNR and OPR1, a tryptophan re si due fills this volume and eliminates the pos si bility of an ordered water molecule at this location. Both YqjM and TOYE possess Ala in this location that provides sufficient space; however, th e two re si dues that form hydrogen bonds with the ordered water in OYE2.6 (Thr35 and Tyr 78) either cannot form h ydrogen bonds (Ile) or have a suitable re si due (Cys), but in unfavorable orientations. Taken together, this may explain the absence of this critical ordered water
115 molecule in the other proteins. The po si tions of the active si te Thr si de chains (po si tions 35 and 37 in OYE2.6 and S. pastorianus OYE1, respectively) are essentially indistinguishable. The same is true for the Il e si de chain ( Ile11 3 in OYE2.6 and the W116I si de chain in S. pastorianus OYE1), with one important difference. In the OYE1 mutant, the Ile si de chain was observed in two conformations, only one of which allows productive substrate binding. Likewise, two b inding orientations of enone 2 were observed in the complex with W116I S. pastorianus OYE1, one potentially relevant to cataly si s and the other clearly incorrect By contrast, the corresponding Ile si de chain in OYE2.6 ( Ile11 3) was only found in a si ngle c onformation, which matched the productive arrangement in the W116I S. pastorianus OYE1 mutant (Figure 2 17 ). This si ngular c onformation of Ile11 3 may be one reason that OYE2.6 is a more efficient catalyst when compared to the W116I mutant of S. pastorianus OYE1. Despite these si milarities, there are also key differences between OYE 2.6 and S. pastorianus OYE1 even the OYE1 W116I mutant. For example, when reducing (S) carvone, both wild type S. pastorianus OYE1 and OYE2.6 share the same stereoselectivity, w hich is oppo si te tha t of the OYE1 W116I mutant This not only points out the special nature of Baylis Hillman substrate binding to alkene reductase active si tes, bu t also underscores the fact OYE 2.6 is not si mply the same as the W116I mutant of S. pastoria nus OYE1. Other active si te re si dues of OYE2.6 must also influence the orientation of substrate binding. In the previous chapter, the altered stereoselectivity observed by some Trp116 replacements of OYE1 was explained by the creation of a hydrophobic po cket that was capable of accommodating the isopropenyl chain of carvone. This would allow for a substrate flipping and therefore a different stereochemical outcome. In OYE2.6, an
116 analogous pocket seems to exist, however it is precluded by Tyr78 whose side chain occupies a slightly different position compared to its OYE1 counterpart (Tyr82). Our group has used the structural knowledge gained throughout the course of this work and new amino acid positions are being targeted for mutagenesis. Future efforts w ill focus on expanding the substrate range for OYE2.6 as well as the identification of stereocomplementary variants. Another are currently being explored in our group is the increase in thermostability of the enzyme through combining directed evolution and rational design. Experimental Procedures Protein Purif i cation E. coli BL21 (DE3) harboring plasmid pBS3 (a derivative of pET 26b containing t he OYE 2.6 coding region flanked by NdeI and XhoI si tes) was grown at 37 C in LB medium supplemented with 50 g/mL kanamycin. IPTG was added to a final c oncentration of 400 mM when the culture had reached an OD 600 value of 0.6 and shaking was continued at 30 C for an additional 8 10 h. Cells were harvested by centrifugation at 5, 000 x g resuspended in 100 mM Tris Cl 10 mM PMSF, p H 8.0 at a concentration of 1 g / mL and lysed by a French pressure cell at 12,000 p si All purification steps were carried out at 4 C. Debris was removed by centrifugation (18, 000 x g for 45 60 min). Solid ammonium sulfate was added to the s uper natant to reach 78% saturation; insoluble material was then removed by centrifugation (18,000 x g for 30 min). The pellet was resuspended in a minimal vol ume of 100 mM Tris Cl, 50 mM ammonium sulfate pH 8.0, protamine sulfate was then added to a final concentration of 1 mg / mL and the resulting insoluble material was removed by centrifugation (15,000 g for 20 min). The dark yellow supernatant was dialyzed against
117 100 mM Tris Cl, 50 mM ammonium sulfate pH 8.0 for 8 h rs This was replaced by additional dialy si s buffer supplemented with 10 mM sodium dithionite for 6 h. In the final dialy si s step (8 h), sodium dithionite was absent in the buffer, yielding a bright yellow dialysate. The crude protein sample was loaded onto an N (4 hydroxybenzoyl) aminohexyl agarose affinity column previously equilibrated with 100 mM Tris Cl, 50 mM ammonium sulfate pH 8.0. The column matrix became dark brown, indicating successful OYE2.6 binding. After loading, the column was washed with the starting buffer at a flow rat e of 0.5 mL / min until the A bs 280 value of the flow through solution was stabilized at ca 0.2. OYE2.6 was eluted by washing the column with deoxygenated starting buffer supplemented with 4 mM sodium dithionite. The protein fractions with the most intense y ellow color (after subsequent air oxidation) were pooled and concentrated by ultrafiltration to a pproximately 10 mg / mL Final purification was achieved by gel filtration c hromatography on a Superdex 200 column (Pharmacia) equilibrated with 30 mM Tris Cl, 30 mM NaCl pH 7.5. The eluted protein was concentrated by ultrafiltration to ca. 40 mg / mL u si ng a 280 value of 55,810 M 1 cm 1 as calculated by ExPASy  prior to crystallization studies. Native Gel Filtration For gel filtration experiments a Superdex 200 HR 10/30 (Pharmacia) column was equilibrated with five column volumes of 100 mM KP i pH 7.0 buffer at 4 C and a flow rate of 0.5 mL/min For calibration, 250 L of a protein mixture containing thyroglobulin OYE1, BSA, ovalbumin, myoglobin at a concent ration of 2 3 mg/mL each was injected. The elution volumes were recorded by monitoring the absorbance at 280 nm. The void volume was determined by injecting a 1 mg/mL sample of blue dextran Appendix D shows the elution of the protein standards. The elutio n volumes were
118 plotted versus the logarithm of each protein molecular weight and fitted by a linear regression (R 2 = 0.9931) (Appendix D, Figure D 2) The elution volume for OYE2.6 (250 L at 10 mg/mL) was 13.6 mL at 4C and a flow rate of 0.5 mL/min ( Figure 3 20). This elution volume resulted in a calculated molecular of 88 kDa.The theoretical molecular based on amino acid sequence is 45 kDa which suggests OYE2.6 is a dimer. Figure 3 20. Elution profile of OYE2.6. The elution volume is consistent wit h that of a homodimer. Crystallogene si s A p rotein solution of ca. 40 mg/mL OYE2.6 in 30 mM Tris HCl, 30 mM NaCl pH 7.5 was used in screening for crystallization conditions. In 96 well crystallization plates, 1 L of protein solution was mixed with an equa l volume of 96 unique crystallization buffers (Classics II Suite, Qiagen). This procedure was repeated using a different set of crystallization conditions (Index, Hampton) The most promising results appeared in 2.4 M sodium malonate pH 7.0. A crystal was screened using a home anode diffractometer (Rigaku) and diffraction to a maximum resolution of 2 was observed. Unfortunately these data could not be utilized for structure solution as they were highly mosaic and appeared to contain m ultiple poorly define d lattices that could
119 not be indexed. As this condition proved recalcitrant to optimization, a screening kit containing 96 unique additives was used in combination with the original crystallization buffer. Finally, the best c rystals w ere grown u si ng the si tting drop vapor diffu si on method on polystyrene microbridges from 2.4M sodium malonate, pH 7.0 buffer supplemented with 1 3% 2 propanol. These crystallization conditions were not ideal for cryo crystallography and data collected at 1 00 K varied significantly in quality and an ideal cryo protection protocol could not be readily indentified. A more systematic investigation showed that the best diffrac tion data were obtained when crystals were briefly washed in 3.0 M sodium malonate pH 7 .0, 10% glycerol ( Figure 3 21B) Figure 3 21. X ray diffraction images of crystals that had been briefly washed in 3.0 M sodium malonate pH 7.0 and A 5% glycerol; B 10% glycerol; C 15% glycerol and D 20% glycerol v/v.
120 After growth, crystals were soaked in harvesting buffer (3.0 M sodium malonate, 10% v/v glycerol, pH 7.0) and flash cooled in liquid nitrogen for data collection. The crystals belonged to space group P 6 3 2 2. To prepare the p chlorophenol complex, crystals were briefly soaked in har vesting buffer supplemented with 2 mM p chlorophenol. Increa si ng soaking times increased mosaicity and decreased the resolution of the crystals. The best results (diffraction data to 2.02 ) resulted from soaking for short periods of time (2 5 seconds ) Structure Solution Reflection data were processed u si ng the HKL3000 program suite 42 Phases were obtained by molecular replacement u si n g the AutoMR utility of PHENIX. 44 u si ng a modification of S. pastorianus OYE1 (PDB code 1OYA) as the search model. All ligands, water molecules and non identical si de chains were rem oved prior to molecular replacement. One solution was found in space group P 6 3 2 2. Moreover, the initially calculated 2 m F o D F and m F o D F c maps showed electron den si ty patterns that could be ea si ly identified as FMN, further validating the molecular replacement solution. The initial model con si sted of 385 amino acids (most devoid of si de chains) and an R free value of 0.33. After the initial si mulated annealing refinement followed by refinement with individual xyz parameters and B factors, the R free dr opped to 0.24. Inspection of the newly phased electron den si ty maps allowed placement of the mis si ng re si dues and showed that some regions of the protein backbone required rebuilding. Refining the more complete model improved the phases and electron den si t y maps clearly showed both ordered solvent molecules and amino acid si de chains. Iterative cycles of model building and refinement were continued to yield the final protein structures. Model building was carried out with COOT 45 and refinement was performed by PHENIX.
121 After the protein and cofactor modeling was completed, additional electron den si ty lying above the FMN was apparent in both the 2mF o F c and F o F c maps. Either malonate alone (3TJL) or nicotinamide (3UPW) completely accounted for this additional active si te electron den si ty. In the case of the p chlorophenol soaked structure, the final electron den si ty maps suggested only partial occupancy of this ligand (approximately 60%), this is con si stent with the very short ligand soaking time. Additional electron den si ty beyond that of pchlorophenol was modeled as a malonate molecule in the same orientation as observed in the native structure. The depo si ted structure (4DF2) has both ligands in the model. All protein s tructure figures were created u si ng PyMOL Schr dinger, LLC) Phenol Binding Studies Assays were performed at 20 C in 100 mM KP i pH 7.0 in a total volume of 1 mL. Purified OYE2.6 was added to a final concentration of ca 20 M. The actual protein concentr ation present during each experiment was calculated from the initial value of A 460 u si ng 460 =10,600. Aliquots of the appropriate phenol ( p chlorophenol or p hydroxy benzaldehyde; 4.0 mM stocks in EtOH) were added to the protein solution and the mixtures were stirred briefly. Spectra were recorded at ca 1 min intervals until no further 3 min). This process was c ontinued until apparent saturation was reached.
122 CHAPTER 4 CHARACTERIZATION OF Lactococcus lactis CV56 DIHYDROOR OTATE DEHYDROGENASES IA AND IB Introduction and Rationale The previous chapters have discussed the applicability of OYEs as biocatalysts in asymmetric reductions. More specifically, the focus has been on the enantioselectivity of these enzymes and we have described our efforts towards understanding the factors that control substrate stereoselectivity. Since gaining access to stereocomplementary products could open many doors to the application of alkene reductases in synthetic processes, thi s has been an ar ea of great interest in our group. A nother factor that can limit the suitability of OYEs and other enzymes for biocatalysis is the rather narrow substrate range usually displayed Generally speaking, an enzyme evolves to catalyze one particular chemical reaction and it is, therefore, expected that a high preference for a single substrate or substrate class is displayed A direct consequence of this is that many chemists will part ways with success as they attempt to increase the value and complexity of the substrates being pursued. One possible way to address this particular limitation is through the use of protein engineering employing mutagenesis and directed evolution While these approaches can certainly prove successful, it is vir tually impossible to predict the precise impact, if any, that each modification will have on an enzyme. Another point of concern is the fact that abundant effort can be devoted to a particular project without any guarantee th at results will be satisfactory In light of these potential challenges associated with protein engineering, one may consider alternative approaches and rely on natural evolution to provide us with enzymes that have already evolved to accept different substrates while catalyzing equival ent chemical transformations.
123 In the case of alkene reductases in the OYE family the substrate range is generally narrow due to the small active site coupled to the geometric restraints imposed by the bound flavin mononucleotide and the side chain involved in protonation (usually a Tyr residue) Two particular classes of substrates are generally poorly enones As has been extensively discussed in chapters 1 and 2, OYE family membe rs are able to catalyze the net trans hydrogenation of activated alkenes and, t o a fair approximation, the reactivity of these enzymes is dictated by the highly conserved three dimensiona l arrangement of the hydrogen bond donors His and Asn/His, the proton don or Tyr196 and the hydride donor N5 of flavin. Given the importance and lack of plasticity of the spatial arrangement in the active sites of some of the same structural features per haps they could catalyze the same chemical reaction irrespective of any overall sequence similarity or substrate specificity In other words, what other enzymes possess a bound FMN as well as amino acid side chains equivalent to those of His191, Asn 194 and Tyr196 having similar spatial arrangements relative to one another? Searches through the structural data bank utility PDBeFold 65 led us to the enzyme dihydroorotate dehydrogenase (DHODase or simply DHOD ) Like OYE1, DHODase belongs to the TIM phosphate bindin g family and has a tightly non covalently associated FMN in the center of its eight barrel domain. While countless enzymes share these same features, deeper examination of the DHODase active site structure led us to believe it could wor k as an alkene reductase complementing OYEs.
124 Background and Experimental Design DHODase Function D ihydroorotate dehydrogenase s catalyze t he two electron oxidation of L dihydroorotate to orotate This conversion is the only redox step in the universal pyrimidine de novo biosynthesis The formation of orotate is the fourth overall step in the biosynthetic monophosphate (OMP) ( Figure 4 1 ). Although all DHODS are flavoproteins, the properties of enzymes f rom different organisms vary considerably 66 The enzymes from Gram positive bacteria, archaea, and some lower eukaryotes, e.g., Saccharomyces cerevisiae are soluble proteins, which use soluble substances as electron acceptors. The se are members of family 1. The DHODs belong ing to family 2 are found mainly in eukaryotic organism and Gram negative bacteria. Family 2 enzymes such as human DHOD, are attached to membranes and use respira tory quinones as final electron acceptors 67 Soluble and membrane associa ted DHODs usually have less than 20% sequence identity and the membrane bound enzymes usually display an additional 40 amino acid domain which forms the binding site for quinones in their N termini and which have no counterpart in soluble DHODs Soluble DH ODs also diverge significantly from each other and have been further divided into two subgroups, family 1A and 1B which show ca. 30% sequence identity 68 69 69b Family 1A enzymes are found in anaerobic yeasts, some protozoa and in milk fermenting bacteria such as Lactococcus lactis 69b 70 and Enterococcus faecalis 71 72 A notable remark is that these two bacteria possess both family 1A and 1B DHODs. Family 1A DHODs are soluble homodimeric proteins with one FMN molecule bound to each subunit and use fumarate as an electron acceptor and are sometimes termed
125 fumarate reductases. In contrast, DHODs of fam ily 1B, predominant in Gram positive bacteria 2 2 heterotetramers and are able to use NAD + as an electron acceptor 73 Independent of their family or subfamily, all DHODs are closely related mechanistically in particular with respect to their first h alf reaction Figure 4 1 The pyrimidine de novo biosynthesis. Step four is an oxidation catalyzed by the enzyme dihydroorotate dehydrogenase ( DHODase ) Catalytic Cycle and Electron Acceptors Similar to OYEs, the catalytic cycle of DHODs can be divided into two separate half reactions in which the FMN is first reduced to FMNH 2 and subsequent ly reoxidized to FMN ( Figure 4 2 ). In the first half reaction, the substrate L 5,6 dihydroorotate binds to
126 the active site of the FMN containing subunit where a Cys residue is ideally oriented to abstract a proton from C5 of dihydroorotate. This polarization leads to a concomitant transfer of hydride to the FMN resulting in the formation of orotate and FMNH 2 The first half reaction appears to be conserve d in all DHODs 17b 74 75 Conversely, the mechanism for the second half reaction depends on the identity of each DHOD. Figure 4 2. The catalytic cycle of DHOD. In the first half reaction, electrons abstracted from dihydroorotate reduce FMN; in the second half reaction, FMNH 2 is reoxidized by a variety of electron acceptors. In summary, the second half reaction of DHODs regenerates oxidized FMN via different electron acceptors in order to prepare the enzyme for the next catalytic turnover. The identity of the electron acceptor differs in the many DHODs and different classes utilize different classes of molecules. Family 2 enzymes use membrane associated respiratory quinones and their mechanism has been less extensively studied although several inhibitors have been successfully d esigned 76 Family 1 DHODs are
127 soluble proteins and use soluble electron acceptors such as fumarate (DHOD 1A) and NAD + (DHOD 1B) One of the main differences between 1A and 1B enzymes is found in their quaternary structures and this is directly related to their mechanism and substrate specificity. Family 1A enzymes are homodimeric and both firs t and second half reactions take place in the same subunit. After oxidation of dihydroorotate, electron s from the reduced flavin are then transferred to the second substrate, fumarate, leading to the formation of succinate in the second half reaction The overall process constitutes a one site ping pong Bi Bi kinetic mechanism. DHODs of family 1B are hetero tetrameric subunits. As in type 1A enzymes, subunits are the site of the first half reaction and contain an FMN cofactor Type 1B enzymes generally cannot use fumarate as an acceptor 73 and electrons are transferred to a bound FAD via a n iron sulfur ( 2 Fe 2 S ) cluster in the subunit In this subunit, NAD + binds and electrons are transferred from the FADH 2 completing the cycle. Other molecules such as ferricyanide, the dye DCIP, and oxygen can also serve as electron acceptors for all DHODs. However with the exception of oxygen, these are thought to be gratuitous acceptors and likely have no physiological role. Structural Similarities and Potential as an Alkene Reductase As mentioned above, we looked for enzymes that exhibited a scaffold similar to that of OYE1. More specifically we looked for similarities in the three dimensional structure with respect to catalytically relevant amino acid residues hoping the enzyme would be able to catalyze chemical conversions that were equivalent to those of OYE1 while accepting different classes of substrates. The first enzyme we turned our attention to was L. lactis DHOD 1A.
128 Like many OYEs, this enzyme is a cytosolic homodimer and each monomer folds into 8 barrel with a tightly bound FMN cofac tor and it is classified as FMN linked oxidoreductase. Despite these obvious similariti es, sequence alignment of DHOD 1 A and OYE1 suggest s that these enzymes are not related and a mere 11.2% (45/400) sequence identity is observed. However, structural alignment with respect to the bound flavin mononucleotide group reveals some inte resting similarities. In OYE1, the most catalytically important features of the active site are residues 191, 194, 196 and 37. His191 and Asn194 are responsible for properly orienting the substrate as well increasing its reactivity through stabilization of an enol intermediate. Tyr196 is directly involved in protonation of the substrate to complete the net hydrogenatio n of the C C double bond while Thr37 is crucial for regulating the reactivity of the FMN 17c Interestingly, DHOD 1A seems to possess residues in similar locations that are, at least in principle, capable of serving the same functions as those in OYE1 ( Figure 4 3) The crystal structure of DHOD 1A in complex with orotic acid 77 reveals that a carbonyl oxygen of the pyrimidine ring is hydrogen bonded to Asn127 and Asn193 which placing carbon is 3.5 from the N5 of flavin. These values are in close agreeme nt with to those seen in structures of OYE1/substrate complexes and seem to describe a favorable geometry for hydride transfer. Another aspect of DHOD 1A that sparked our interest is related to the substrate of its second half reaction, fumarate. In the se cond half reaction, DHOD 1A catalyzes the reduction of fumarate to succinate and some DHODs have been classified as fumarate reductase s in T. cruzi 78 This observation was of particular importance to us
129 since OYE1 and our library of alkene reductases consistently failed to reduce unsaturated acid s and this could further dive rsify our biocatalytic toolbox. The fact that DHOD 1A is able to convert fumaric acid into succinate indicates that its bound FMN cofactor has a redox potential that is suitable for reduction of alkenes conjugated to carboxylic acids. Since we deduced that a change in substrate selectivity would be more easily achievable than tuning the FMN redox potential, we cloned DHOD. Figure 4 3. Active site alignment of OYE1 and DHOD 1A. The residues are labeled as (OYE1/DHOD) numbers. DHOD 1A residues are shown in thick cyan sticks and OYE1 in thin magenta lines. Results and Discussion L actococcus lactis CV56 DHODase 1A Cloning E xpression and P urification DHOD 1A was cloned from the pur ified genomic DNA of L. lactis CV56 using p rimers designed to amplify the pydA gene using PCR while inserting Nde I and Xho I recognition sites ( Figure 4 4 A) These restric tion sites allowed for direct ligation of the pydA gene into the vector pET 22(b)+ (Novagen ) After ligation, E coli JM109 cells were transf ormed using the construct and four random colonies were picked for plasmid
130 extraction. DNA sequencing results showed that the resulting plasmid (pYAP 3) contained t he pydA gene enco ding DHOD 1A ( Figure 4 4 B) Figure 4 4. A The primers utilized to clone the pydA gene from L. lactis CV56. B The final construct of the plasmid pYAP lactamase and confers ampicillin resistance. The pYAP 3 plasmid was transformed into E. coli BL21 Gold (DE3) and protein overexpression was initiated by addition of 0.5 mM IPTG. Cells were harvested after 8 hrs and t he dihydroorotate dehydrogenase activity was assayed using lysate from cells that had been disrupted via sonication. At this sta ge the lysate exhibited the characteristic yellow color of flavins suggesting that DHOD 1A was in fact being overexpressed. The lysate total activity was 2.25 M s 1 L 1 To determine whether the
131 dehydrogenase activity observed was due to DHOD1A or other contaminant s in the lysate the assay was run using the same cell line lacking plasmid DNA. These cells had a measured dihydroorotate dehydrogenase ac tivity of < 0.2 M s 1 L 1 Analysis of the cells using SDS PAGE further validated the overexpression of D HOD 1A as a strong band consistent with the theoretical molecular weight of 33kDa ( Figure 4 5). Figure 4 5. SDS PAGE analysis of E. coli cells carrying pYAP 3. Lane A, B and C contain different amounts of boiled cells harvested prior to induction. Lanes F and G contain samples of boiled cells 8 hrs after induction. In order to carry out more studies of the enzyme we pursued a purification route that could yield large amounts of high purity enzyme. The parent vector of plasmid pYAP 3, pET 22(b)+, is desi gned to express a C terminal hexa histidyl fragment ideal for metal affinity chromatography. In order to express this portion, a stop codon naturally present in the pydA gene needed to be removed. Using a modified procedure of Zheng et al 41 we designed mutagenic primers that would change the original TAA codon to a GCT codon which encod es an Ala residue. This resulted in a construct pYAP 7, that carries the gene for DHOD 1A containing a C terminal His 6 tag named. Using the plasmid pYAP 7 (appendix E, Figure E 1), DHOD 1A could be successfully
132 overexpress ed and purified by nic kel affinity chromatography which was also catalytically active Our initial motivation for cloning DHOD 1A was the possibility of using this enzyme as a biocatalyst that could complement the substrate panel for alkene reductases in the OYE family. While D HOD 1A can successfully reduce fumarate and possibly other unsaturated carboxylic acids, this enzyme had one particular limitation that makes it non ideal for transformations of preparative scale. In a biocatalytic process, often times, the amount of start ing material to be converted can range from a couple of milligrams to near a gram. In the case of asymmetric reductions using alkene reductases, such large scale operation can become costly as the price of the electron donor NADPH fluctuates around US$600 US$1000 dollars/gram. One way to circumvent the prohibitive prices is to couple a cofactor regeneration system to the biocatalytic process. A widely used method for the regeneration of NADPH is the glucose/glucose dehydrogenase system 79 which is compatible with most OYE enzymes. In the case of DHOD 1A, this system cannot be used and an analogous reducing equivalents regeneration route is not readily accessibl e. This ultimate electron source is 5,6 dihydroorotate and since we could not find a feasible way regenerate this molecule if DHOD 1A were to be employed in a preparative scale process an equimolar amount of L 5,6 dihydroorotate would be needed. Even though the price of US$ 200 dollars/gram for this compound may not seem as overwhelming as th at of nicotinamide cofactors, it is far from insignificant and this prompted us to pursue a different enzyme in DHOD 1B.
133 Lactococcus lactis CV56 DHOD 1B Introduction In l donors, we turned our attention to a different dihydroorotate dehydrogenase pre sent in L. lactis DHOD 1B. This enzyme differs from DHOD 1A in several aspects that make it a more promising target in expanding the substrate panel for OYE like alkene reductases. In contrast to the homodimeric forms of type 1A enzymes, DHOD 1B is a soluble sub subunit s are closely related in both sequence and structure to the family 1A homodimers. They are the site of dihydroorotate dehydrogenation and consist primarily of a 8 barrel (TIM barrel ) domain which contain the FMN cofactor s s ubunit in DHOD 1B is the site of electron transfer and adopts a conformation similar to that seen in proteins belonging to the NADPH ferredoxin reductase superfamily 80 It is subdivide into two domains. The N terminal domain, consisting of the first 100 amino acids serves as the binding site for an FAD molecule The C terminal domain comprises an open / structure. It resembles the NAD binding domains of the ferredoxin reductase family and is the putative NAD + binding site in DHOD 1B This structural feature offers a logical explanation for why ability to use NAD + as an electron acceptor is unique to famil y 1B enzymes In 1996, Nielsen et al. 73 unambiguously identified the genes encoding both the subunit ( pyr subunit ( pyr K) in Lactococcus lactis Furthermore, all of the cofactors were identified and it was demonstrated subunit ( pyr K) was essential for the reduction of NAD + at the expense of dihydroorotate It was also shown that, under acidic conditions, the enzyme could catalyze the reverse of the
134 physiological reaction, namely the reduction of orotate to di hydroorotate in the presence of NADH. Even though the physiological relevance of this result was uncertain at the time it made DHOD 1B an attractive target for our studies. Much like in OYE1 catalyzed alkene reductions, t he conversion of orotate to 5,6 d ihydroorotate involves the net trans hydrogenation of a C C double bond at the expense of NADH electrons. Additionally, since a variety of NADH cofactor regeneration systems have been described with success, this enzyme could be, at least in principle, a u seful biocatalyst at a reasonable cost. Cloning e xpression and p urification Lactococcus lactis is one of the few organisms that carries the genes for both class 1A and 1B DHODs allowing us to clone the 1B enzyme from the same genomic DNA preparation used for DHOD 1A In the 1,360 to 1,370 kilobase region of the L lactis CV56 genome the pyrZ gene is located directly upstream from the pyr D gene. pyr Z gene and a reverse pyr D gene ( Figure 4 6 ). The p rimers inserted restriction sites for the endonucleases Nde I and Xho I which allowed the ligation of the 1796 base pair fragment into pET 22(b)+. The resulting construct was terme d pYAP 5 and contained the pyrZ gene and the pyrD gene expressing a C terminal His 6 tag. ( appendix B, Figure B 2 ) Initially, the protein was overexpressed in E. coli BL21 Gold (DE3) cells that were grown in LB broth supplemented with ampicillin utilizing standard protocols and the protein could be successfully purified using nickel affinity chromatography. However this proved unsuccessful and the purified enzyme had low catalytic activity it had a
135 molecular weight much lower than that of the tetramer (65k Da versus 125kDa) and exhibited a UV Vis spectrum that was not consistent with the homotetrameric form ( Figure 4 7) lack of a broad absorbance band in the range of 520 560 nm which is observed in the heterotetramer. Taken together, these observations suggest that our enzyme preparation failed to produce the correct enzyme form in any appreciable amount and we surmised that either the construct pYAP 5 or the expression protocol, or both needed to be changed in order to achiev e more satisfactory results. Figure 4 6. Region of L. lactis CV56 genome containing the pyr Z and pyr D genes. The black arrows indicate the location of the primers and the primers sequences are shown above and below the gene cluster diagram. In their original characterization of DHOD 1B, Nielsen et al. had described the us ing anion
136 exchange chromatography and it was established that the yellowish green form was a homodimer composed of FMN binding subunits ( pyr D) while the brownish orange enzyme was a heterotetramer consisting of two pyr D and two pyr Based on the published description and the visual appearance of our enzyme solutions we concluded that the preparation we obtained using cells harboring pYAP 5 grown in LB medium overwhelmingly expressed the pyr D protein and nearly undetectable amounts of the subunit, pyr Z To repair the inadequacies in expression of the heterotetrameric form of DHOD 1B, changes were made in both the expression protocol and in the construct of the protein. Figure 4 7. Absorption spectrum of DHOD 1A. The lack of a broad abso rbance band from 520 560 nm explains the yellow green appearance of the enzyme solution. The peaks at 360 and 460 nm are typical of FMN cofactors. Escherichia coli is one of the most widely used expression hosts due to its ability to grow rapidly and to h igh density while generally giving good yields of targ et proteins However, expression has been especially troublesome for proteins containing Fe S clusters 81 and much effort has been devoted toward achieving better results. Ballou and co workers showed that optimizati on of growth conditions can lead to high
137 expression levels of some Fe S cluster containing proteins without the need for co expression of genes in the isc cluster 82 Based on these results, we changed the growth medium accordingly and c ells were grown in TB (terrific broth) medium instead of LB broth and were supplemented with ferric citrate, ferrous sulfate and cysteine a t the time of induction. Another change in the expression protocol was that cells were induced at a late log phase (O .D. 600 3.0). To maximize the amount of protein that could be purified we also changed the construct of the plasmid being used. The plasmid pYAP subunit pyr Z subunit pyr D protein with an additional hexahistidyl portion i n its C terminus. In a scenario where the expression of the Fe subunit is subunit, this construct will lead to the isolation of homodimeric DHOD almost exclusively. To overcome this deficiency, we us ed pET 15b (Novagen ) to construct a plasmid, pYAP 6, which encodes pyr Z with an N terminal His 6 tag and pyr D in its native state Figure 4 8. Elution profile of the purified DHOD 1B. The enzyme elutes at 11.4 mL primarily.
138 Cells harboring pYAP 6 were grown in TB and protein expression was carried out based on the results of Ballou and co workers 82 Under these conditions the cell lysate displayed a light brown color characteristic of iron sulfur containing proteins suggesting th at the pyr Z was being expressed appreciably. The clarified lysate was applied onto a nitrilotriacetic acid agarose resin that had been loaded with Ni 2+ and the resin was brown color indicating that the protein had bound succes sfully. The protein was eluted and concentrated by ultra filtration to a concentration of 1.5 mg/mL, and the final yield was approximately 5 mg/L of culture. Figure 4 9. The absorption spectrum of the purified DHOD 1B The enzyme spectrum is shown as a thick dashed line and the spectrum of free FMN is show n in thin dash dot dash line One interesting observation is that the enzyme was fairly unstable and had the tendency to aggregate in buffer even at low temperatures near 4 C. However, if cleavage of the N terminal His6 tag using thrombin was performed, the enzyme exhibited much bette r stability and could be sotred for months with retention of activity.
139 2 2 heterotetramer we carried out native gel filtration experiments. Under the experimental conditions, the elution vol ume for the protein was 1 1.40 mL ( Figure 4 8) and a calculated molecular weight of 115 kDa (appendix D) which is in good agreement with the theoretical molecular weight for DHOD 1B of 126 kDa. The absorption spectrum of the purified enzyme showed a broad a bsorbance band in the 520 560 nm region also suggesting subunits and Fe subunit s ( Figure 4 9). Crystallization and Structure of L. lactis CV56 DHOD 1B In anticipation to future protein enginee ring pr o jects involving DHOD 1B we decided to crystallize the wild type enzyme. In 2000, Rowla n d et al published the structure of the first family 1B enzyme 83 h owever the source of the protein the y crystallized was different than the one we used in th is study and the differences in sequence were 18 amino acids and 29 subunits respectively F urthermore, if we could find conditions that yield diffraction qua lity crystals they could be use d for gaining structural information a bout different variants we may construct as well as different enzyme substrate complexes. To screen crystallization conditions, enzyme was concentrated to ca. 22 mg/mL in 50 mM NaP i pH 6.0, 10% v/v glycerol solutions. Crystallization drops were set up w ith either free enzyme or with the ligand orotate in addition to enzyme. After screening 192 conditions, t he most promising crystallization results were obtained using PEG of different molecular weights as a precipitant in buffers of acidic pH (3.5 5.0). U nfortunately the crystals originally formed as thin plates with several layers and diffracted X rays poorly yielding no interpretable data ( Figure 4 10 A ) Crystal s could be
140 obtained reproducibly from 0.1 M sodium acetate pH 4.0, PEG 4,000 8% w/v solution s, therefore this buffer was further optimized. In general, the addition of several polyols improved crystal morphology. In particular the addition of 3% glycerol resulted in crystals appeared singular and possessed nicely defined faces (4 10 B). Figure 4 10. Wild type DHOD 1B crystal optimization. A Initial crystals obtained in sodium acetate pH 4.0, 8% w/v PEG 4,000. B Crystals after addition of 3% v/v glycerol. One puzzling observation was the fact that despite the significant improvement in their morphology, these crystals failed to yield satisfactory X ray diffraction data. After several attempts at collecting data under cryogenic temperatures (100 K), we decided to test the diffraction of these crystals at room temperature. One notable dif ference between these two collection protocols is the fact that, in room temperature collection, there is no need to treat the crystals with a cryoprotecting buffer such as 30% glycerol. While cryoprotection is a necessary step prior to cooling the crystal s, it can sometimes be detrimental to X ray diffraction. Many crystals exhibit extreme sensitivity to the composition of their mother liquor and the slightest changes in osmolality and ionic strength of this solution can lead to a decreased diffraction res olution or even complete dissolution of the crystal. In the case of our DHO D 1B the crystals diffracted to a
141 maximum resolution of 2.4 at room temperature while the crystals that had been cryoprotected and frozen in liquid nitrogen usually gave diffrac tion patterns that were hardly usable. One limitation associated with data collection at room temperature, and also one of the main motivations for the development of cryocrystallography, is the high susceptibility to radiation damage causing loss of diffr action. This was unfortunately the case for us and a full data set could not be collected at room temperature even though the crystals diffracted well enough. Figure 4 11. X ray diffraction pattern at different temperatures. A Typical diffraction patter Diffraction pattern of crystal at room temperature. Despite difficulties with finding an appropriate cryoprotection protocol, on e full data set was collected on a single crystal. This crystal exhibited high anisotropy, w hich causes the diffraction pattern to be highly variable depending on the crystal orientation relative to the X ray beam. By utilizing select X ray images from different sections of the crystal, d iffraction data up to 2.7 could be integrated and scaled with an overall completeness of 96.5%. T he structure was solved by molecular replacement using the structure of the closely related protein published by Rowland et al (PDB code 1EP2)
142 The crystals belonged to the orthorhombic space group C 2 2 2 1 with a solvent content of 54.2% and a Matthews coefficient of 2.68 3 /Da. The asymmetric unit contained two with unit cell dimensions of 80.2 x 151.9 x 214.8 . Interestingly, the contacts and the surface area of the interface between the two hete rodimers in the asymmetric unit suggest this is not the heterotetrameric biological assembly Rather a neighboring heterodimer, related by a crystallographic 2 fold symmetry operator appears to form heterotetramer that is stable in solution and not a cryst allographic artifact. This proposed heterotetramer assembly buries 37% of the total protein surface area, which is almost three times as much as any of the other protein interface s and has the same overall arrangement as reported by Rowland et al. ( Figure 4 12). The final model was refined to residuals error values R work and R free of 0.22 and 0.26, respectively. The complete crystallographic parameters, data collection and refinement stat istics can be found in Table 4 1 Figure 4 12. The biological heterotetramer of DHOD 1B. T he protein backbone is shown as cartoon representation, the FMN and FAD cofactors are shown as sticks, the 2Fe 2S clusters are shown as spheres.
143 As expected, the overall structure is very closely related to that of L. lactis DHOD 1B reported by Rowland, however, one striking difference is observed. In our structure of L. lactis subunit is shifted by as much as 12 . This loop carries the catalytic Cys135 residue and its movement is directly linked to substrate binding and reactivity 83 84 It has been suggested that the loo p would close upon substrate binding to adopt a catalytically active conformation. The open conformation seen in our structure of DHOD 1B in complex with orotate contradicts this notion and it is likely not the catalytically active conformation. In the ope n loop form, the catalytic acid/base of Cys135 is located almost 7 away from th e substrate which contrasts a distance of 3.6 in the closed form Figure 4 13. The conformational changes in the active site loop. The structure solved in this work ha
144 Table 4 1. X ray Crystallographic Data Collection and Refinement Statistics Structure name Orotate DHOD 1B Observed active si te ligand orotate X ray source Space group C 2 2 2 1 Unit cell dimen si ons a, b, c () 80.20 151.91 214.79 90, 90, 90 Resolution () 50.39 2.72 (2.817 2.72) Unique reflections 34552 (3497) Completeness (%) 96.82 (99.57) Multiplicity 5.1 (5.2) R sym [b] 0.13 (0.63) 7.34 (1.69) R work [c] R free [d] 0. 22 (0.30) 0. 26 (0.37) Ramachandran statistics [e] Favored (%) Allowed (%) Outliers (%) 9 6 3.9 0 .1 Number of atoms P rotein solvent ligand 8451 307 211 Average B factors ( 2 ) Protein Solvent Ligands 36.6 32.8 32.4 [ a] Values in parentheses denote data for the highest resolution bin. [b] Rmerge = hkl i |Ii(hkl) [I(hkl)]|/ hkl i Ii(hkl), where Ii(hkl) is the inten si ty of the ith observation of unique reflection hkl. [c] R work |F o (hkl)| |F c (hkl)| o (hkl)|. [d] R free is calculated in the same manner as R work u si ng 10% of the reflection data not included during the refinement. [e] Statistics generated u si ng MOLPROBITY 39
145 Substrate Specificity and Mutagenesis Targets One property of DHOD 1B that was particularly attractive to us was its ability to catalyze the two electron reduction of orotic acid to yield dihydroorotic acid. Our wild type enzyme preparation catalyzed this conversion with a specific activity of 45.7 m ole min 1 mg 1 in the presence of 1 mM orotate and 0.18 mM NADH. We wanted, therefore, that could offer a substrate panel that was c omplementary to that of our OYE type e nzymes ( Figure 4 14) More specifically we envisioned a possible route for the amino acids which can be used in many applications. Furthermore, if we could manipulate the substrate enantioselectivity in DHOD 1B, as we have d one with OYE1 and OYE2.6, the synthesis of both D and L amino acids would be a possibility. Figure 4 14. Potential substrates for DHOD 1B that would complement the substrate range in our existing library of OYE like alkene reductases. The natural subst rate for all DHOD s is 5,6 dihydroorotate and the fact that these enzymes do not accept 5,6 dihydrouracil or other pyrimidines suggests that the 6
146 carboxyl substituent plays a major role in substrate specificity. To achieve our goal of using DHOD as an alke ne reductase to convert a variety of non natural substrates we needed to uncover which active site residues could impact substrate promiscuity and tolerance in DHOD 1B. Additionally, since mutations near the substrate binding site had already proven succes sful in uncovering enantiocomplementary OYE1 variants, we optimistically decided to employ the same strategy for DHOD 1B. Given the importance of the carboxyl substituent in the substrate for DHODs, we chose the initial target for protein mutagenesis to be with this particular moiety We rationalized that enzymes capable of reducing uracil and other pyrimidines could provide some guidance for choosing an amino acid target. We compared the active site of DHOD 1B with that of pig dihydropyrimidine dehydrogenase (DPD) an enzyme that catalyzes the two electron reduction of uracil to yield 5,6 dihydrouracil. Like DHOD 1B, DPD uses a n FMN cofactor for catalysis and the ultimate electron source is reduc ed nicotinamide. DPD is a large multi domain protein with two subunits of 1025 amino acids each but, despite these disparities, the architecture of the FMN domain and the pyrimidine binding site is surprisingly close to that of DHOD 1B. In fact structural alignment of the two proteins using the FMN as reference reveals that many of the residues that define the active site are conserved. Among the subtle differences one amino acid position is particularly interesting, namely Ile74. In DPD the equivalent pos ition is occupied by Glu611 and despite the obvious differences in the physical properties of the two residues, some of the interactions in which Glu611 participate s make it a hi ghly interesting residue. In the crystal structure, Glu611 is well positioned to
147 participate in hydrogen bonding and to form a salt bridge with Lys574 ( Figure 4 15 A) The equivalent lysine in DHOD 1B, Lys48 has been shown to undergo large conformational changes upon substrate binding and is important for its orientation as well as controlling the reactivity of the FMN 77 85 In order to probe the effect of these interactions in DHOD 1B, we constructed two mutant enzymes, Ile74Glu and Ile74Asp. Figure 4 15. Active site comparison of DPD and DHOD 1B. A The crystal structure of 5 iodouracil bound to pig DPD. B The crystal structure of orotate bound to L. lactis CV56 DHOD 1B. The dashed lines indicate hyd rogen bonds and salt bridges. Characterization of Ile74Glu and Ile74Asp The Ile74Glu and Ile74 Asp mutants were constructed with mutagenic primers using the plasmid pYAP 6 as template (Appendix B Figure B 5). Both mutants were expressed and purified as described for the wild type enzyme. The final protein yield was 3 mg/L for I74D DHOD 1B and 12 mg/L for the I74E variant. We wanted to test the hypothesis that by inserting a negative charge in the proximi ty of Lys48 we would alter the substrate preference of DHOD 1B with the goal of allowing the enzyme to accept uracil and 5,6 dihydrouracil as substrates. In addition to
148 this direct interaction with Lys48, we hypothesized that the simple repulsion between t he carboxyl of orotate and the side chain of Asp/Glu could be sufficient to cause the enzyme to accept pyrimidines that lack a carboxyl moiety. Unfortunately this was not the case and like the wild type enzyme, both I74E and I74D mutants failed to catalyz e the reduction of uracil (up to 10 mM) in the presence of 0.2 mM NADH in acidic conditions. We also tested the dihydrouracil dehydrogenase activity of both mutants in the presence of up to 0.2 mM NAD + in basic conditions but could not detect any rates. C haracterization of the G75N and I74E/G75N Mutants Since our supposition that introducing a salt bridge interaction between Lys48 and a negatively charged residue would make uracil a substrate failed, we examined other interactions between substrate and pro tein. Besides being stabilized by the side chain of Lys48, the carboxylate group of orotate appears to be hydrogen bonded to two main chain nitrogen atoms in a n N of residues Gly75 and Leu76 are properly ori ented to provide stabilization for a negatively charged oxygen ( Figure 4 15 B). Curiously, this interaction is not possible in dihydropyrimidine dehydrogenase since two main chain carbonyl oxygens point towards the substrate binding site ( Figure 4 15 A). F urther examination of the amino acid conformations suggest that t his major back bone rearrangement is accomplished by a Ile74 and Gly75. To study this hypothesis, we constructed two additional mutants, Gly75Asn and the double mutant Ile74Glu / Gly75Asn. The mutants were expressed using plasmids pYAP 12 and pYAP 13 (appendix B, Figure B 5), and both proteins could be purified as described in the previous sections. Both of these mutants were expressed at much lower levels compared to the wild type enzyme. As before, these mutants failed to catalyze the reduction of uracil (up
149 to 10 mM) in the presence of 0.2 mM NADH in acidic conditions. We also tested the dihydrouracil dehydrogenase activity of both mutants in the presence of up to 0.2 m M NAD + in basic conditions but could not detect any rates Steady state Kinetics and Catalytic Efficiency Given the lack of success in our attempts to modify the active site of DHOD 1B in order to change its substrate preferences, we decided to characteriz e the catalytic efficiency of each of the five proteins we had successfully purified. We chose to test + as the electron acceptor simulating the physiologically relevant reaction The assays wer e carried out in hypoxic conditions by using the oxygen scavenging system GODCAT 86 This system consumes dissolved oxygen via the oxidation of glucose, catalyzed by glucose oxidase (GOD), to yield gluconic acid and hydrogen peroxide. The hydrogen peroxide is further deg raded by catal ase (CAT). The initial concentration of NAD + was 110 M for all measurements and its consumption was monitored by changes in absorption at 340 nm over 120 seconds. All assays were performed at 23 C. Table 4 2 summarizes the results Table 4 2. Steady state parameters of wild type and mutant DHOD 1 B enzymes Enzyme Apparent K m Apparent k cat k cat / K m M s 1 M 1 s 1 Wild type 18.12.3 35014 1.9x10 7 I74D 1561442 5.700.70 3.6x10 3 I74E no reaction a no reaction G75N 42993 8.660.56 2.0X10 4 I74E/G75N ND b [a] No reaction was detected for this enzyme. [b] P arameters could not be determined accurately.
150 Wild type The wild type enzyme was clearly the most efficient catalyst with a k cat /K m value of 1.9x10 7 Its apparent K m value of 18.1 M was the lowest measured in this study ( Figure 4 16). Figure 4 16. Conc entration dependence of measured initial velocities V o for w.t. Ile74Asp m utant Figure 4 17 Conc entration dependence of measured initial velocities V o for the I74D mutant Based on the results obtained, changing position 74 from Ile to Asp was very cat /K m value was decreased by a
151 factor of 5300. Both substrate binding ( K m ) and catalytic turnover ( k cat ) seemed to ha ve been equally affected negatively by factor s of 86 and 61 respectively. Ile74Glu m utant This enzyme variant showed no detectable DHODase activity under the experimental conditions. Gly75Asn m utant This particular mutation was originally intended to cause back bone in order to destabilize the binding of orotate. Accordingly, i t decreased the enzyme s efficien cy by a factor of 1000 which still made it the best catalyst besides the wild type enzyme. Figure 4 1 8 Conc entration dependence of measured initial velocities V o for the G75N mutant. Ile74Glu/Gly75Asn m utant This double mutant displayed interesting behavior in that its velocity decreased with addition of substrate before saturation levels could be reached. The data obtained could not be fit confidently using classic Michaelis Menten kinetics. The best model to explain the data, albeit a relatively poor one was that of substrate inhibition with a K i value of ca 207 M Using a direct fit approach (GraphPad), ac ceptable curve fits
152 could only be found if certain kinetic parameters were constrained and, while a solution was found, it was statistically no better than a mere approximation. Figure 4 1 9 Conc entration dependence of measured initial velocities V o for the I74E/G75N double mutant. Residues 74 and 75 can Disrupt Electron Transfer to the s ubunit Kinetic s tudies of oxidase r eactions From our kinetic studies it was clear that the amino acid substitutions had negative impact s on the catalytic efficien cy of the enzyme. More specifically, they were detrimental to the dihydroorotate dehydrogenase of DHOD1B. One of the limitations of steady state kinetics is the fact that the overall rate that can be measured will most closely r eflect the slowest step in t he mechanism. In the case of DHOD 1B t he complete catalytic cycle involves several individual steps however its activity was measured based solely on the formation of NADH, the last step of the second half reaction ( Figure 4 2) Like many flavin dependent enzymes, DHOD 1B can use molecular oxygen as the electron acceptor and catalyze oxidase reactions In the case of DHOD 1B, this property of the flavin opens the possibility for alternate paths to accomplish the second
153 (or oxidative) half reaction in the c atalytic cycle ( Figure 4 20 ) After completion of the first half reaction wherein electrons are abstracted from L dihydroorotate, electrons in the reduced FMN cofactor have two fates. They can be transferred to the FAD molecule subunit through the Fe S cluster to ultimately reduce NAD + which would constitute its dehydrogenase activity ( Figure 4 20, 2a). A lternatively electrons can be transferred directly to oxygen in an oxidase reaction. Figure 4 20. The possible fates of electrons in DHOD 1B. In the first half reaction (1), electrons are transferred from dihydroorotate to the FMN. In the second half reaction electrons can be accepted by molecular oxygen (2a) or by NAD + via [2Fe 2S] FAD (2b). We decide to assay the oxidase acitivity of all th e DHOD variants we generated in order to gain a better understanding of the exact role played by these amino acids. Each enzyme was allowed to react with dihydroorotate in the presence of oxygen. The reaction buffer contained no NAD + or any other electron acceptor and r eaction progress
154 was monitored by following the formation of orotate 278 nm. The kinetic parameters for each enzyme are summarized in Table 4 3. Table 4 3. Steady state parameters of wild type and mutant DHOD 1 B enzymes Enzyme Apparent K m Apparent k cat k cat / K m M s 1 M 1 s 1 Wild type 5.930.61 35.450.83 5.9x10 6 I74D 981 123 42.80 2.1 4.3x10 4 I74E 1793246 15.71.1 8.7x10 3 G75N 153 17 12.3 0. 28 8.0 X10 4 I74E/G75N > 2000 a [a] This variant failed to display any saturation behavior in substrate concentrations up to 10 mM. The wild type enzyme is able catalyze the oxidation of dihdroorotate less efficiently in the absence of NAD + While the apparent K m value was little lower, th e Figure 4 21). Figure 4 21 Conc entration dependence of measured initial velocities V o for wild type DHOD1B. The Ile74Asp mutant provided us with a surprising result Where as this muta tion had significantly impaired its dehydrogenase activity, this is was not the case for its
155 oxidase reaction. While the K m value was significantly higher than the wild type enzyme, the turnover number is slightly higher at 42.8 s 1 ( Figure 4 22). Figure 4 22. Conc entration dependence of measured initial velocities V o for the I74D mutant. The Ile74Glu yielded and even more interesting result since it had failed to show any dehydrogenase activity but it was successful in catalyzing the oxidation of dihydro orotate in the presence of oxygen ( Figure 4 23). Figure 4 23. Conc entration dependence of measured initial velocities V o for the I74E mutant. Consistent with the previous observations, the G75N variant had its apparent K m value affected the least and was the most efficient enzyme after the wild type. Its k cat /K m value has decreased by a factor of 7.3 which suggest s that this mutation does not affect the
156 enzyme s oxidase activity to the same extent as seen in the dehydrogen ase reaction. ( Figure 4 24) Figure 4 24. Conc entration dependence of measured initial velocities V o for the G75N mutant. Choosing b etween NAD + and o xygen : studies of u ncoupling Our kinetic investigations strongly suggest that mutations to positions 74 a nd 75 affect not only substrate binding but also the preference for different electron transfer pathways. It seems clear that alteration at these two residues cause the transfer of electron to NAD + to be much les s favorable while retaining its oxidase acti vity. In order to consolidate this notion we carried out experiments that present the enzymes with the cho ice of w h ich electron acceptor they will utilize for catalysis. In other words, under strictly anaerobic conditions, for every molecule of dihydroorotate that is oxidized there will be a molecule of NAD + that becomes reduced However, if oxygen is allowed to enter the system this no longer must hold true. If the enzyme utilizes both oxygen and NAD + to oxidize dihydroorotate the mole amounts o f NAD H and orotate will not be equal and an uncoupling effect will be observed.
157 To measure the uncoupling ability of each enzyme, we carried out two sets of experiments. First each enzyme was forced to catalyze the oxidation of dihydroorotate in the pr esence of NAD + and the absence of oxygen and the formation of both products we re monitored. We refer to this set experiment the same enzyme variant is mixed with the substrates NAD + and dihydroorotate in similar f ashion, but in the presence of oxygen, in the aerobic run. The formation of the products orotate and NADH is monitored and the results from the hypoxic and aerobic runs are compared. Figure 4 25. Uncoupling by wild type DHOD 1B. A Product formation for reaction under hypoxic (red dashed lines) and aerobic conditions (black solid lines). The formation of product is shown as circles (orotate) and squares (NADH). B UV Vis spectra taken at the final reaction time under hypoxic (red dashed lines) and aerobic conditions (black solid lines).
158 The first enzyme that we examined was wild type DHOD 1B. As we had previously observed, this enzyme was very efficient in catalyzing the anaerobic dehydrogenation of dihydroorotate in the presence of NAD + and as expected the formation of orotate and NADH were tightly coupled. They differed by only 1.7 nmoles under hypoxic conditions and the enzyme kept the tight coupling even under aerobic conditions where the difference in moles was only slightly higher, 2.6 nmoles ( Figure 4 25 A). T he UV Vis spectra of the reactions taken at the final time point of 360 s further corroborated these results. The relative peak heights for orotate (278 nm) and NADH (340 nm) remained relative unchanged ( Figure 4 25 B) Figure 4 26. Unco upling by I74E DHOD 1B. A Product formation for reaction under hypoxic (red dashed lines) and aerobic conditions (black solid lines). The formation of product is shown as circles (orotate) and squares (NADH). B UV Vis spectra taken at the final reaction time under hypoxic (red dashed lines) and aerobic conditions (black solid lines). As we had seen during our steady state studies, the Ile74Glu mutant showed no dehydrogenase activity but was fairly efficient in catalyzing the aerobic oxidation of dihydroor otate. These results were certainly validated by the nearly complete
159 uncoupling of orotate and NADH formation Virtually no NADH formation could be detected under either the aerobic or hypoxic reaction conditions The formation of approximately 2.5 nmoles of orotate was observed under hypoxic conditions and probably reflected the fact that the reaction mixture was not strictly anaerobic. Conversely the enzyme was able to convert dihydroorotate to orotate much more efficiently in the presence of oxygen ( Fig ure 4 26). The two enzymes discussed so far are very different in behavior and can be thought of as polar opposites. The wild type enzyme showed little to no tendency of utilizing oxygen as the electron acceptor, while the Ile74Glu variant used oxygen as t he sole electron acceptor. The other variants exhibited behaviors that were not so binary in nature. Rather their activities constituted a continuum that ranged from the wild type to the I le 74 Glu enzyme and are summarized in figures 4 27 through 4 29. Figure 4 27. Uncoupling by I74D DHOD 1B. A .Product formation for reaction under hypoxic (red dashed lines) and aerobic conditions (black solid lines). The formation of product is shown as circles (orotate) and squares (NADH). B UV Vis spectra taken at th e final reaction time under hypoxic (red dashed lines) and aerobic conditions (black solid lines).
160 Figure 4 2 8 Uncoupling by G75N DHOD 1B. A .Product formation for reaction under hypoxic (red dashed lines) and aerobic conditions (black solid lines). The formation of product is shown as circles (orotate) and squares (NADH). B UV Vis spectra taken at the final reaction time under hypoxic (red dashed lines) and aerobic conditions (black solid lines). Figure 4 29. Uncoupling by I74E/G75N DHOD 1B. A .Product formation for reaction under hypoxic (red dashed lines) and aerobic conditions (black solid lines). The formation of product is shown as circles (orotate) and squares (NADH). B UV Vis spectra taken at the final reaction time under hypoxic (red dashed lines) and aerobic conditions (black solid lines).
161 Conclusion and Future Directions In our search for enzymes that could function as alkene reductases and expand our collection of biocatalyst s and their substrate panel we cloned and over expressed two dihydroorotate dehydrogenases from Lactococcus lactis CV56. The type 1B enzyme seemed to be the more suitable candidate due to its ability to accept electrons directly from NADH to carry out the reduction of the C C double of orotate. We therefore puri fied it to homogeneity and solved its crystal structure in complex with orotate. In an I74E, G75N and a double mutant I74D/G75N. These variants showed no tolerance towards the substrate uracil which lacks a carboxyl group present in orotate. Next we used steady state kinetics to characterize the catalytic efficiency of each DHOD 1B variant. Since all of the mutations seemed to have largely detrimental effects on the dehydro genase activity of the enzyme, we took on the investigation of the while binding of the substrate was largely affected as well, the mutations did not show the same negative effect on the e magnitude Additionally, we showed that these mutations largely impacted the electron transfer rates to NAD + Rather, under aerobic conditions, most of the electrons are transferred to oxy gen. subunit, or pyr Z protein, contains the [2Fe 2S] cluster and its structure is closely related to other NAD binding enzymes, formation of NADH is most likely to happen through this subunit. Furthermore, Nielsen et al had showed that the abs subunit did not affect electron transfer to oxygen but was necessary for the formation of NADH 73 Based on these observations and the results we obtained
162 in this study, we suggest that residues Ile74 and Gly75 are important not only for subst subunit ( pyr subunit ( pyr Z). In fact, the Ile74Glu mutant seems to completely block electrons from subunit which results in loss of its dehydrogenase activity while re taining most of its oxidase activity. Each of the five DHOD 1B proteins we investigated had slightly different catalytic properties. In addition, each mutation had effects on the activity which were not the same for the oxidase activity. Table 4 4 s ummarize s the catalytic properties of each of the mutants. Table 4 4. Catalytic Properties of WT and DHOD 1B mutants. Enzyme Dehydrogenase k cat Oxidase k cat Coupl ing Efficiency s 1 s 1 Wild type 35014 35.450.83 0.91 I74D 5.700.70 42.802.1 0.05 I74E 15.71.1 0 G75N 8.660.56 12.30.28 0.03 I74E/G75N 0.16 enzyme for NAD + C.E. This term was calculated using the following equation:
163 Where Oro t and NADH are th e number of moles formed, under hypoxic conditions, o f oro tate and NADH respectively. Oro and NADH are th e number of moles formed, under aerobic conditions, of orotate and NADH respectively. Our results show that mutations of positions 74 and 75 can decrease the OD1B from 0.91 in wild type to zero in the Ile74Glu mutant. Put differently, the wild type enzyme strongly prefers to utilize NAD + as the electron acceptor while the Ile74Glu mutant could only use oxygen. This striking result prompted us to rationalize the possible mechanis m by which this change was taking place. Figure 4 30. Absorption spectra of the five DHOD 1B enzymes studied. Solid lines represent the enzymes and the dashed line shows the spectrum of free FMN. The inset shows the ratio of absorbance at 454/540 nm. T he simplest explanation for the poor ability of some mutants to transfer electrons to the NAD binding domain could be that mutants failed to express this
164 domain altogether. However based on the physical properties of the mutants and their size according to gel filtration experiments, we have no reason to believe that this is an important factor. Additionally, because of how the proteins were constructed for subunit had not been expressed, the protein could not have been purified succe ssfully using metal affinity chromatography and this theory is supported by SDS PAGE experiment (see experimental procedures). Another plausible explanation is that the mutant enzymes do not form/bind the iron sulfur cluster which is essential for inter su bunit electron transfer. To test this, we collected UV Vis absorption spectra of all enzymes since each cofactor has characteristic peaks. The FMN and FAD cofactor have strong peaks ca. 450 nm while the Fe S cluster shows a broad band near 540 nm. The spe ctra suggest that all of the enzymes have successfully complexed the iron sulfur cluster, at least to an appreciable extent as shown by the ratio Abs 454 /Abs 540 ( Figure 4 30). Since all of the enzymes appeared to be structurally sound, we focused our attent ion on the structure of the active site of DHOD 1B in order to rationalize the results obtained. In our crystal structure, Ile74 is positioned close to both the dimethyl benzene moiety of the FMN cofactor and the bound orotate ( Figure 4 15 B). Initially we had focused our attention on the possible interactions between this residue, Lys48 and the substrate. One important detail that was overlooked is the fact that this Ile 74 is also in close proximity of the Fe subunit. Severa l factors suggest that the exact position of this residue seem critical in regulating the reactivity of the FMN and the redox center. First Ile74 is positioned right
165 in between the FMN and Fe subunit ( Figure 4 31 A) Figure 4 31. A subunit are subunit are in orange carbons. The red mesh shows the omit mFo around oro tate and the 2Fe 2S cluster. B The preferred region (dashed circle) for electron transfer from the FMN in DPD, DHOD 1B and DHOD 2. Although this residue i s not strictly conserved across DHOD enzymes, it is almost always a hydrophobic residue I t is conserved in the DHOD B subgroup; it is a methionine DHOD A subgroup, and it is generally tyrosine, methionine, or cysteine in family 2 enzymes It had been previously hypothesized that t he nature of this side chain is probably important in defining the en vironment of the dimethylbenzene part of the flavin isoalloxazine ring, rather than in binding the substrate 77 Our results suggest otherwise and the nature of this residue can impact not only substrate binding but also electron tr ansfer. One curious observation is the fact that there seems to be a geometric preference for electron transfer reactions between the FMN and other electron acceptors.
166 In the case of DHOD 1B, the 2Fe subunit directly behin d the dimethylbenzene rin g of the FMN. In DPD, the [4Fe 4S] is held in place by an adjacent monomer, nonetheless, it is located in a very similar position relative to the FMN confactor. In the membrane associated family 2 enzyme, such as humanDHOD, there i s no external redox center. Rather, this enzyme utilizes membrane bound quinone as the electron acceptor. Curiously, in the crystal structure of human DHOD in complex with the immunosuppressant/anticancer drug brequinar, the drug binds in a long cleft form ed by two N helices that are unique to family 2 enzymes. Here again the position of the drug is very close to those observed in the enzymes that contain other electron transfer cofactor (figure 4 arrangement suggests the existence of an optimal location for electron transfer to and from the FMN. We hypothesize that our mutations to positions 74 and 75 could cause disturbances to this spatial arrangement which would lead to poor electron transfer subunits. Our hypothesis seems to offer an explanation for how the mutations could impair electron transfer. The decreased dehydrogenase activity is likely due to mutations making electron subunit less favorable, and since electron s can be subunit oxidase activity was retained. However, one question remains unanswered. How can the wild type so efficiently transfer electrons to NAD+ even in the presence of oxygen? The I74D mutant seems to have sligh tly higher oxidase activity suggesting that not only does the wild type enzyme prefer NAD+ but also curbs oxidase activity. The answer to this question is not trivial, unfortunately. The exact factors that regulate flavin reactivity towards oxygen and othe r reactive oxygen
167 species (ROS) is still not fully understood. It seems that different enzymes can tune the flavin reactivity in different manners and no real consensus is observed. Despite the apparent complexity of the flavin reactivity control, some fac tors seem to be more recurrent. Mattevi suggests that the polarity and solvation of the flavin binding site environment can largely affect its reactivity through stabilization of the superoxide anion 87 It seems therefore possible that a mutation such as Ile74Asp can exert its effect thorugh a change in active site polarity. Future work should focus towards obtaining crystal structures of each mutant in order to evaluate how the substitutions can perturb the environment around the enzymes redox centers. Also more systematic mu tagenesis studies are likely to shed light onto electron transfer mechanism not only in DHOD 1B bu t in a general sense. Experimental Procedures Si te Directed Mutagene si s All mutants used for were generated by PCR u si ng primers with the de si red mutation at po si tion 74 or 75 in a modification of the method published by Zheng et al 41 Table 4 5 lists the primers used in this study. G75N* indicates that the template used was pYAP 9 which already contained the I74E mutation. Table 4 5 Mutagenic primers utilized for co nstructing the mutant enzymes Mutation Primer Sequence I74E fwd I74E rev TGC TTA ATG CA G AA G GAC TAC AAA ATC CAG GAT TAG AAG T TTT TGT AGT CC T TC T GCA TTA AGC ATG CCG CTC GCT VTT T I74D fwd I74D rev TGC TTA ATG CA GAT G GAC TAC AAA ATC CAG GAT TAG AAG T TTT TGT AGT CC A TC T GCA TTA AGC ATG CCG CTC GCT VTT T G75N fwd G75N rev TT AAT GCA ATT AAT CTA CAA AAT CCA GGA TTA GAA GTT AT GG ATT TTG TAG ATT AAT TGC ATT AAG CAT GCC GCT CGC TG G75N*fwd G75N*rev TT AAT GCA GAA AAT CTA CAA AAT CCA GGA TTA GAA GTT AT GG ATT TTG TAG ATT TTC TGC ATT AAG CAT GCC GCT CGC TG The shaded letters indicate the mutagenic codons.
168 Each PCR reaction (total volume 100 L) containing 5 Phu si on HF Buffer (20 L), p YAP 6 (Appendix B Figure B 4) (20 ng), forward and reverse primers (0.5 M each), dNTPs (200 M each), and Phu si on Hot Start II High Fidelity DNA Polymerase (1 U) was subjected to an initial denaturation step of 98 C (30 s) followed by 25 cycles of 98 C (10 s) and 72 C (4 min) followed by a final incubation at 72C (7 min). Am plicons were purified by DNA spin columns, digested with Dpn I at 37C (10 U for 4 h followed by an additional 10 U for 4 h) then purified by an additional DNA spin column. Aliquots (5 L) were used to transform E. coli JM109 (75 L) by electroporation. SOC medium was added (600 L), then the samples were incubated for 1 h at 37 C prior to selection on LB medium supplemented with ampicillin Plasmid DNA purified (spin columns) from randomly chosen colonies was analyzed by DNA sequencing to identify the de si red DHOD 1B variants. Protein Expres si on and P urification Plasmids were transformed into E. coli BL21 G old ( DE3 ) cells and grown in T B broth supplemented with and 200 g/mL ampicillin Cultures were induced by addition of 0.4 mM isopropyl thiogalacto si de (IPTG) when the optical den si ty Abs600 reached 2.0 At this point, the growth was supplemented with 1mM cysteine, 0.1 mg/mL ferrous sulfate and 0.1 mg/mL ferric citrate. The temperature was decreased to 30 C during protein expres si on. Cells were harvest ed after 16 hours at a den si ty of approximately 12 g/L. Cells were then pelleted by centrifugation at 5350 x g for 10 minutes and pellet resuspended to a density of 1g/mL in 50 mM NaP i pH 6 .0 10% glycerol containing 10 M phenylmethanesulfonyl fluoride (PM SF). All of the subsequent steps were performed at 4 C. The resuspended cells were lysed at 15,000 p si u si ng a Carver
169 laboratory press. Lysed cells had debris removed by centrifugation at 20,000 x g for 1 hour. P rotamine sulfate (1 mg/mL) was added and n ucleic acids were removed by centrifug ation at 10,000 x g for 15 min. For the purification step a n NTA resin that had been pre loaded with 100 mM nickel sulfate and equilibrated with 50 mM NaP i pH 6 .0 10% glycerol, 40 mM imidazole using a Pharmacia FPLC sytem. Cell extract was loaded onto the column and it was washed with buffer until the absorbance at 280 nm be came stable at approximately 0.1 A.U. For elution, the column was washed with 3 column volumes of the same buffer supplemented with 500 mM imidazole. Enzyme was collected in approximately 10 mL. Fractions were pooled and concentrated by centrifugation u si ng a 10,000 MWCO membrane. For regeneration of the affinity matrix, the column was washed with 3 volumes of buffer containing 50 mM To remo ve the His tag, protein was incubated with thrombin (1:1000 mass ratio) after all imidazol e had been removed by dialysis. The protein mixture was then centrifuge d again and reapplied onto the n ickel chelating resin and the flow through contained DHOD 1B. F igure 4 30 shows SDS PAGE anaylisis of all of the DHOD proteins purified in this study. For crystallography, t he flow through was concentrated a nd the p urified protein was then loaded onto a Su perdex 200 HR 10 30 (Pharmacia) for further purification. The c olumn was equilibrated with low ionic strength buffer, 50 mM NaP i 10 % glycerol pH 6 .0 Only the majo r peak at 280 nm was collected. Proteins were concentrated to >2 mg/mL and could be stored in 50% glycerol at 20 C. For crystallization trials the wild t ype protein was concentrated to 22 mg/mL and utilized without freezing.
170 Figure 4 32 SDS PAGE analysis of all DHOD 1B enzymes purified in this study. Lane A, wild type; lane B, I74E; lane C, I74D; lane D, G75N; lane E, I74E/G75N mutant. The theoretical subunit and subunit. Crystallog raphic Studies A protein solution of ca. 22 mg/mL DHOD 1B in 50 mM NaP i pH 6.0, 10% v/v glycerol was used in screening for crystallization conditions. In 96 well crystallization plates, 1 L of protein solution was mixed with an equal volume of 96 unique crystallization buffers (Classics II Suite, Qiagen). This procedure was repeated using a different set of crystallization conditions (PEGRx, Hampton). The most pro mising results appeared in low pH buffers with low to medium molecular weight PEGs as precipitant with the most promising condition being 0.1 M sodium acetate pH 4.0, PEG 4,000 12% w/v. C rystals appeared very quickly ( <1 hr ) and se emed to be made of variou s thin layers. Unfortunately these crystals could not be utilized for structure solution as they were highly mosaic and appeared to contain multiple poorly define lattices that could not be indexed. As this condition proved recalcitrant to optimization, a screening kit containing 96 unique additives was used in combination with the original crystallization buffer Finally, the best c rystals were grown u si ng the hanging drop vapor diffu si on method on silanized glass covers from 0.1 M sodium acetate pH 4.0, P EG 4,000 12% w/v supplemented with 3 % v/v glycerol These crystallization conditions were not ideal
171 for cryo crystallography and data collected at 100 K varied significantly in quality and an ideal cryo protection protocol could not be readily indentified. A more systematic investigation showed that the best diffraction data were obtained when crystals were mounted for data collection in the mother liquor at room temperature (figures 4 10 and 4 11). Structure Solution X ray data were processed u si ng and iMO SFLM 42 43 42 43 .The structure of L. lac tis CV56 DHOD 1B w as solved by molecular replacement u si ng AUTOMR in the PHENIX suite 44 The search used a high resolution model of L. lactis DHOD 1B (PDB code 1EP3 ). The initial search model was devoid of water molecules and any ligands. Inspection of the electron den si ty maps was followed by manual rebuilding, additio n of against the X ray data. During refinement, inspection of the calculated 2 m Fo D Fc and m Fo D Fc difference maps showed clear po si tive electron den si ty for the ligand orotat e as well as the cofactors FMN, FAD and 2Fe 2S This allowed for building of the missing side chains and unambiguous placement of all ligands. This process was performed iteratively until re si duals errors R work and R free converged. The complete crystallogr aphic parameters, data collection and refinement statistics can be found in Results and Discus si on. Steady State Kinetics Enzyme solutions were quantified using a specific absorption coefficient A 452 of 0.6 mg 1 mL 1 73 Enzyme concentrations were 1 nM for wild type, 20 nM for G75N, 65 nM for I74E, 68 nM for I74D and 125 nm for the I74E/G75N variant. Reactions were assayed using a 1 cm quartz cuvette at 23 C in 100 mM KP i pH 7.5, 2 mM L cysteine,
172 10 mM glucose, glucose o xidase 5 U/mL, catalase 2 U/mL and the total reaction volume was 1 mL. Reactions were pre incubated for 2 min and initiated by addition of dihydroorotate. The initial concentration of NAD + was110 M and initial velocities were measured for 60 120 seconds b y monitoring the formation of NADH at 340 nm The data were fitted using the software GraphPad . For oxidase activity measurements the buffer was vigorously stirred prior to data collection to ensure oxygen concentration reached the atmospheric concentr ation of 260 M. The total reaction volume was 200 L and initial velocities were measured for 60 120 seconds by monitoring the formation of orotate at 278 nm.
173 APPENDIX A SELECTED GC MS DATA Figure A 1. Authentic standard (Acros) of (+) dihydrocarvone containing a mixture of isomers. The p eak at 7.2 min is trans (2 R ,5 R ) dihydrocarvone present at 80%. The peak at 7.6 min is cis (2 S ,5 R ) dihydrocarvone present at 20%. Figure A 2. Chromatogram of authentic sample of (R) carvone (Acr os Organics). The peak has retention time of 8.8 min. Because of the non chiral nature of the stationary phase, (S) carvone had the same retention time.
174 Figure A 3. GC MS analy si s of OYE1 catalyzed (R) carvone reduction. Peak identities: 6.0 min, methyl benzoate (standard); 7.2 min, trans (2 R ,5 R ) dihydrocarvone; 7.6 min cis (2 S ,5 R ) dihydrocarvone; 8.8 min, (R) carvone. Figure A 4. GC MS analy si s of W116A OYE1 catalyzed (R) carvone reduction. Peak identity: 6.0 min, methyl benzoate (standard); 7.2 min t rans (2 R ,5 R ) dihydrocarvone; 7.6 min cis (2 S ,5 R ) dihydrocarvone; 8.8 min, (R) carvone.
175 Figure A 5. GC MS analy si s of OYE1 catalyzed (S) carvone reduction. Peak identity: 6.0 min, methyl benzoate (standard); 7.2 min, trans (2 S ,5 S ) dihydrocarvone ; 7.6 min cis (2 R ,5 S )dihydrocarvone ; 8.8 min, (S) carvone. Figure A 6. GC MS analy si s of W116I OYE1 catalyzed (S) carvone reduction. Peak identity: 6.0 min, methyl benzoate (standard); 7.2 min, trans (2 S ,5 S ) dihydrocarvone ; 7.6 min cis (2 R ,5 S )dihydrocarvone ; 8.8 min, (S) carvone.
176 APPENDIX B PLASMID CONSTRUCTS USED Figure B 1. Plasmid map of all pET OYE W116X used for structural studies. Small arrows indicate location of the mutagenic primers used. Figure B 2. Plasmid construct of pYAP 7. DHOD 1A with C terminal His tag.
177 Figure B 3. pYAP 5 encodes pyrD with C terminal His tag. Figure B 4 pYAP 6 encodes pyrZ with N terminal His tag.
178 Figure B 5. Basic plasmid construct for pYAP 8, pYAP 9, pYAP 12 and pYAP 13. Each mutant was created using primers listed in Table 4 4.
179 APPENDIX C PURIFIED OYE1 SDS PAGE ANALY SI S Figure C 1. Typical purity of OYE1 enzymes used for X ray crystallography. Lane s A,B Samples after affinity chromatography. Lanes C,D Samples after following si ze exclu si on chromatography with Superdex 200.
180 APPENDIX D GEL FILTRATION OF PROTEIN LADDER AND LINEAR REGRESSION Figure D 1. Elution profi le of the protein standards used for calibration. Figure D 2. The linear fit of the molecular weight ( l og M.W.) versus the relative elution volumes plot.
181 LIST OF REFERENCES 1. Toogood, H. S.; Gardiner, J. M.; Scrutton, N. S., Biocatalytic Reductions and Chemical Versatility of the Old Yellow Enzyme Family of Flavoprotein Oxidoreductases. Chemcatchem 2010, 2 (8), 892 914. 2. Warburg, O.; Christian, W., On a new oxidation enzyme. Naturwissenschaften 1932, 20 980 981. 3. Haas, E., Isolation of a new yellow enzyme. Biochem Z 1938, 298 (5/6), 378 390. 4. Theorell, H.; Akeson, A., Molecular Weight and Fmn Content of Crystalline Old Yellow Enzyme. Arch Biochem Biophys 1956, 65 (1), 439 448. 5. Matthews, R. G.; Massey, V., Isolation of Old Yellow Enzyme in Free and Complexed Forms. J Biol Chem 1969, 244 (7), 1779 &. 6. Massey, V.; Palmer, G., Charge Transfer Complexes of Lipoyl Dehydrogenase. Biochem J 1961, 81 (1), P22 &. 7. Matthews, R G.; Massey, V.; Sweeley, C. C., Identification of Para Hydroxybenzaldehyde as Ligand in Green Form of Old Yellow Enzyme. J Biol Chem 1975, 250 (24), 9294 9298. 8. Abramovitz, A. S.; Massey, V., Interaction of Phenols with Old Yellow Enzyme Physical Evi dence for Charge Transfer Complexes. J Biol Chem 1976, 251 (17), 5327 5336. 9. Abramovitz, A. S.; Massey, V., Purification of Intact Old Yellow Enzyme Using an Affinity Matrix Fo Sole Chromatographic Step. J Biol Chem 1976, 251 (17), 5321 5326. 10. Saito, K.; Thiele, D. J.; Davio, M.; Lockridge, O.; Massey, V., The Cloning and Expression of a Gene Encoding Old Yellow Enzyme from Saccharomyces Carlsbergensis. J Biol Chem 1991, 266 (31), 20720 20724. 11. Stott, K.; Saito, K.; Thiele, D. J.; Massey, V., Old Ye llow Enzyme the Discovery of Multiple Isozymes and a Family of Related Proteins. J Biol Chem 1993, 268 (9), 6097 6106. 12. Macheroux, P.; Massey, V., 8 Thiocyanatoflavins as Active Site Probes for Flavoproteins. Biochemistry Us 1991, 30 (2), 456 464. 13. Fox, K. M.; Karplus, P. A., Crystallization of Old Yellow Enzyme Illustrates an Effective Strategy for Increasing Protein Crystal Size. J Mol Biol 1993, 234 (2), 502 507.
182 14. Fox, K. M.; Karplus, P. A., Old Yellow Enzyme at 2 Angstrom Resolution Overall Structure, Ligand Binding, and Comparison with Related Flavoproteins. Structure 1994, 2 (11), 1089 1105. 15. Banner, D. W.; Bloomer, A. C.; Petsko, G. A.; Phillips, D. C.; Pog son, C. I.; Wilson, I. A.; Corran, P. H.; Furth, A. J.; Milman, J. D.; Offord, R. E.; Priddle, J. D.; Waley, S. G., Structure of Chicken Muscle Triose Phosphate Isomerase Determined Crystallographically at 2.5a Resolution Using Amino Acid Sequence Data. Na ture 1975, 255 (5510), 609 614. 16. (a) Abendroth, J.; Niefind, K.; May, O.; Siemann, M.; Syldatk, C.; Schomburg, D., The structure of L hydantoinase from Arthobacter aurescens leads to an understanding of dihydropyrimidinase substrate and enantio specific ity. Biochemistry Us 2002, 41 (27), 8589 8597; (b) Barna, T.; Messiha, H. L.; Petosa, C.; Bruce, N. C.; Scrutton, N. S.; Moody, P. C. E., Crystal structure of bacterial morphinone reductase and properties of the C191A mutant enzyme. J Biol Chem 2002, 277 ( 34), 30976 30983; (c) Barna, T. M.; Khan, H.; Bruce, N. C.; Barsukov, I.; Scrutton, N. S.; Moody, P. C. E., Crystal structure of pentaerythritol tetranitrate reductase: "Flipped" binding geometries for steroid substrates in different redox states of the en zyme. J Mol Biol 2001, 310 (2), 433 447; (d) Kitzing, K.; Fitzpatrick, T. B.; Wilken, C.; Sawa, J.; Bourenkov, G. P.; Macheroux, P.; Clausen, T., The 1.3 A crystal structure of the flavoprotein YqjM reveals a novel class of Old Yellow Enzymes. J Biol Chem 2005, 280 (30), 27904 27913. 17. (a) Brown, B. J.; Hyun, J. W.; Duvvuri, S.; Karplus, P. A.; Massey, V., The role of glutamine 114 in Old Yellow Enzyme. J Biol Chem 2002, 277 (3), 2138 2145; (b) Inaoka, D. K.; Sakamoto, K.; Shimizu, H.; Shiba, T.; Kurisu, G.; Nara, T.; Aoki, T.; Kita, K.; Harada, S., Structures of Trypanosoma cruzi dihydroorotate dehydrogenase complexed with substrates and products: Atomic resolution insights into mechanisms of dihydroorotate oxidation and fumarate reduction. Biochemistry U s 2008, 47 (41), 10881 10891; (c) Xu, D.; Kohli, R. M.; Massey, V., The role of threonine 37 in flavin reactivity of the old yellow enzyme. P Natl Acad Sci USA 1999, 96 (7), 3556 3561. 18. Dundas, J.; Ouyang, Z.; Tseng, J.; Binkowski, A.; Turpaz, Y.; Liang J., CASTp: computed atlas of surface topography of proteins with structural and topographical mapping of functionally annotated residues. Nucleic Acids Res 2006, 34 W116 W118. 19. Vaz, A. D. N.; Chakraborty, S.; Massey, V., Old Yellow Enzyme Aromatiza tion of Cyclic Enones and the Mechanism of a Novel Dismutation Reaction. Biochemistry Us 1995, 34 (13), 4246 4256. 20. Kohli, R. M.; Massey, V., The oxidative half reaction of old yellow enzyme The role of tyrosine 196. J Biol Chem 1998, 273 (49), 32763 32770.
183 21. Toscano, M. D.; Woycechowsky, K. J.; Hilvert, D., Minimalist active site redesign: Teaching old enzymes new tricks. Angew Chem Int Edit 2007, 46 (18), 3212 3236. 22. Schmid, A.; Dordick, J. S.; Hauer, B.; Kiener, A.; Wubbolts, M.; Witholt, B., I ndustrial biocatalysis today and tomorrow. Nature 2001, 409 (6817), 258 268. 23. Panke, S.; Held, M.; Wubbolts, M., Trends and innovations in industrial biocatalysis for the production of fine chemicals. Curr Opin Biotech 2004, 15 (4), 272 279. 24. (a) Dougherty, M. J.; Arnold, F. H., Directed evolution: new parts and optimized function. Curr Opin Biotech 2009, 20 (4), 486 491; (b) Romero, P. A.; Arnold, F. H., Exploring protein fitness landscapes by directed evolution. Nat Rev Mol Cell Bio 2009, 10 (12), 866 876. 25. (a) Tuttle, J. B.; Ouellet, S. G.; MacMillan, D. W. C., Organocatalytic transfer hydrogenation of cyclic enones. J Am Chem Soc 2006, 128 (39), 12662 12663; (b) Ouellet, S. G.; Tuttle, J. B.; MacMillan, D. W. C., Enantioselective organoca talytic hydride reduction. J Am Chem Soc 2005, 127 (1), 32 33. 26. Toogood, H. S.; Fryszkowska, A.; Hare, V.; Fisher, K.; Roujeinikova, A.; Leys, D.; Gardiner, J. M.; Stephens, G. M.; Scrutton, N. S., Structure Based Insight into the Asymmetric Bioreductio n of the C=C Double Bond of alpha,beta Unsaturated Nitroalkenes by Pentaerythritol Tetranitrate Reductase. Adv Synth Catal 2008, 350 (17), 2789 2803. 27. Chaparro Riggers, J. F.; Rogers, T. A.; Vazquez Figueroa, E.; Polizzi, K. M.; Bommarius, A. S., Compar ison of three enoate reductases and their potential use for biotransformations. Adv Synth Catal 2007, 349 (8 9), 1521 1531. 28. Swiderska, M. A.; Stewart, J. D., Stereoselective enone reductions by Saccharomyces carlsbergensis old yellow enzyme. J Mol Cata l B Enzym 2006, 42 (1 2), 52 54. 29. Hall, M.; Stueckler, C.; Hauer, B.; Stuermer, R.; Friedrich, T.; Breuer, M.; Kroutil, W.; Faber, K., Asymmetric bioreduction of activated C = C bonds using Zymomonas mobilis NCR enoate reductase and old yellow enzymes O YE 1 3 from yeasts. Eur J Org Chem 2008, (9), 1511 1516. 30. Swiderska, M. A.; Stewart, J. D., Asymmetric bioreductions of beta nitro acrylates as a route to chiral beta(2) Amino acids. Org Lett 2006, 8 (26), 6131 6133. 31. Opperman, D. J.; Piater, L. A.; van Heerden, E., A novel chromate reductase from Thermus scotoductus SA 01 related to old yellow enzyme. J Bacteriol 2008, 190 (8), 3076 3082.
184 32. Adalbjornsson, B. V.; Toogood, H. S.; Fryszkowska, A.; Pudney, C. R.; Jowitt, T. A.; Leys, D.; Scrutton, N. S., Biocatalysis with Thermostable Enzymes: Structure and Properties of a Thermophilic 'ene' Reductase related to Old Yellow Enzyme. Chembiochem 2010, 11 (2), 197 207. 33. Meah, Y.; Massey, V., Old Yellow Enzyme: Stepwise reduction of nitro olefins and cat alysis of aci nitro tautomerization. P Natl Acad Sci USA 2000, 97 (20), 10733 10738. 34. Kosjek, B.; Fleitz, F. J.; Dormer, P. G.; Kuethe, J. T.; Devine, P. N., Asymmetric bioreduction of alpha,beta unsaturated nitriles and ketones. Tetrahedron Asymmetr 2008, 19 (12), 1403 1406. 35. Padhi, S. K.; Bougioukou, D. J.; Stewart, J. D., Site Saturation Mutagenesis of Tryptophan 116 of Saccharomyces pastorianus Old Yellow Enzyme Uncovers Stereocomplementary Variants. J Am Chem Soc 2009, 131 (9), 3271 3280. 36. Z amyatnin, A. A., Amino Acid, Peptide, and Protein Volume in Solution. Annu Rev Biophys Bio 1984, 13 145 165. 37. Fraaije, M. W.; Mattevi, A., Flavoenzymes: diverse catalysts with recurrent features. Trends Biochem Sci 2000, 25 (3), 126 132. 38. Massey, V. ; Schopfer, L. M., Reactivity of Old Yellow Enzyme with Alpha Nadph and Other Pyridine Nucleotide Derivatives. J Biol Chem 1986, 261 (3), 1215 1222. 39. Chen, V. B.; Arendall, W. B.; Headd, J. J.; Keedy, D. A.; Immormino, R. M.; Kapral, G. J.; Murray, L. W .; Richardson, J. S.; Richardson, D. C., MolProbity: all atom structure validation for macromolecular crystallography. Acta Crystallogr D 2010, 66 12 21. 40. Brown, B. J.; Deng, Z.; Karplus, P. A.; Massey, V., On the active site of old yellow enzyme Rol e of histidine 191 and asparagine 194. J Biol Chem 1998, 273 (49), 32753 32762. 41. Zheng, L.; Baumann, U.; Reymond, J. L., An efficient one step site directed and site saturation mutagenesis protocol. Nucleic Acids Res 2004, 32 (14). 42. Otwinowski, Z.; Minor, W., Processing of X ray diffraction data collected in oscillation mode. Method Enzymol 1997, 276 307 326. 43. Battye, T. G. G.; Kontogiannis, L.; Johnson, O.; Powell, H. R.; Leslie, A. G. W., iMOSFLM: a new graphical interface for d iffraction image processing with MOSFLM. Acta Crystallogr D 2011, 67 271 281.
185 44. Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis, I. W.; Echols, N.; Headd, J. J.; Hung, L. W.; Kapral, G. J.; Grosse Kunstleve, R. W.; McCoy, A. J.; Moriarty N. W.; Oeffner, R.; Read, R. J.; Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H., PHENIX: a comprehensive Python based system for macromolecular structure solution. Acta Crystallogr D 2010, 66 213 221. 45. Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K., Features and development of Coot. Acta Crystallogr D 2010, 66 486 501. 46. Murshudov, G. N.; Vagin, A. A.; Dodson, E. J., Refinement of macromolecular structures by the maximum likelihood method. Acta Crystallogr D 1997, 53 240 255. 47. Walton, A. Z.; Conerly, W. C.; Pompeu, Y.; Sullivan, B.; Stewart, J. D., Biocatalytic Reductions of Baylis Hillman Adducts. Acs Catal 2011, 1 (9), 989 993. 48. Stueckler, C.; Winkler, C. K.; Bonnekessel, M.; Faber, K., Asymmetric Synthesis of (R) 3 Hydroxy 2 methylpropanoate ('Roche Ester') and Derivatives via Biocatalytic C=C Bond Reduction. Adv Synth Catal 2010, 352 (14 15), 2663 2666. 49. Chen, A. Q.; Xu, J.; Chiang, W.; Chai, C. L. L., L Threonine catalysed asymmetric alpha hydroxymethylati on of cyclohexanone: application to the synthesis of pharmaceutical compounds and natural products. Tetrahedron 2010, 66 (7), 1489 1495. 50. Mase, N.; Inoue, A.; Nishio, M.; Takabe, K., Organocatalytic alpha hydroxymethylation of cyclopentanone with aqueou s formaldehyde: Easy access to chiral delta lactones. Bioorg Med Chem Lett 2009, 19 (14), 3955 3958. 51. Bougioukou, D. J.; Walton, A. Z.; Stewart, J. D., Towards preparative scale, biocatalytic alkene reductions. Chem Commun 2010, 46 (45), 8558 8560. 52. Matthews, B. W., Solvent Content of Protein Crystals. J Mol Biol 1968, 33 (2), 491 &. 53. Nagano, N.; Orengo, C. A.; Thornton, J. M., One fold with many functions: The evolutionary relationships between TIM barrel families based on their sequences, structu res and functions. J Mol Biol 2002, 321 (5), 741 765. 54. Schaller, F.; Weiler, E. W., Molecular cloning and characterization of 12 oxophytodienoate reductase, an enzyme of the octadecanoid signaling pathway from Arabidopsis thaliana Structural and funct ional relationship to yeast old yellow enzyme. J Biol Chem 1997, 272 (44), 28066 28072. 55. Miller, S.; Lesk, A. M.; Janin, J.; Chothia, C., The Accessible Surface Area and Stability of Oligomeric Proteins. Nature 1987, 328 (6133), 834 836.
186 56. Krissinel, E.; Henrick, K., Inference of macromolecular assemblies from crystalline state. J Mol Biol 2007, 372 (3), 774 797. 57. Stewart, R. C.; Massey, V., Potentiometric Studies of Native and Flavin Substituted Old Yellow Enzyme. J Biol Chem 1985, 260 (25), 3639 3 647. 58. Karplus, P. A.; Fox, K. M.; Massey, V., Flavoprotein structure and mechanism .8. Structure function relations for old yellow enzyme. Faseb J 1995, 9 (15), 1518 1526. 59. Pudney, C. R.; Hay, S.; Pang, J. Y.; Costello, C.; Leys, D.; Sutcliffe, M. J.; Scrutton, N. S., Mutagenesis of morphinone reductase induces multiple reactive configurations and identifies potential ambiguity in kinetic analysis of enzyme tunneling mechanis ms. J Am Chem Soc 2007, 129 (45), 13949 13956. 60. (a) Etayo, P.; Nunez Rico, J. L.; Vidal Ferran, A., Chiral Rhodium Complexes Derived From Electron Rich Phosphine Phosphites as Asymmetric Hydrogenation Catalysts. Organometallics 2011, 30 (24), 6718 6725; (b) Pautigny, C.; Jeulin, S.; Ayad, T.; Zhang, Z. G.; Genet, J. P.; Ratovelomanana Vidal, V., Convenient General Asymmetric Synthesis of Roche Ester Derivatives through Catalytic Asymmetric Hydrogenation: Steric and Electronic Effects of Ligands. Adv Synt h Catal 2008, 350 (16), 2525 2532; (c) Wassenaar, J.; Kuil, M.; Lutz, M.; Spek, A. L.; Reek, J. N. H., Asymmetric Hydrogenation with Highly Active IndolPhos Rh Catalysts: Kinetics and Reaction Mechanism. Chem Eur J 2010, 16 (22), 6509 6517. 61. Yep, A.; Ke nyon, G. L.; McLeish, M. J., Saturation mutagenesis of putative catalytic residues of benzoylformate decarboxylase provides a challenge to the accepted mechanism. P Natl Acad Sci USA 2008, 105 (15), 5733 5738. 62. Hulley, M. E.; Toogood, H. S.; Fryszkowska A.; Mansell, D.; Stephens, G. M.; Gardiner, J. M.; Scrutton, N. S., Focused Directed Evolution of Pentaerythritol Tetranitrate Reductase by Using Automated Anaerobic Kinetic Screening of Site Saturated Libraries. Chembiochem 2010, 11 (17), 2433 2447. 63. Toogood, H. S.; Fryszkowska, A.; Hulley, M.; Sakuma, M.; Mansell, D.; Stephens, G. M.; Gardiner, J. M.; Scrutton, N. S., A Site Saturated Mutagenesis Study of Pentaerythritol Tetranitrate Reductase Reveals that Residues 181 and 184 Influence Ligand Bindin g, Stereochemistry and Reactivity. Chembiochem 2011, 12 (5), 738 749. 64. Bougioukou, D. J.; Kille, S.; Taglieber, A.; Reetz, M. T., Directed Evolution of an Enantioselective Enoate Reductase: Testing the Utility of Iterative Saturation Mutagenesis. Adv Sy nth Catal 2009, 351 (18), 3287 3305. 65. Krissinel, E.; Henrick, K., Multiple alignment of protein structures in three dimensions. Computational Life Sciences, Proceedings 2005, 3695 67 78.
187 66. Bjornberg, O.; Gruner, A. C.; Roepstorff, P.; Jensen, K. F., The activity of Escherichia coli dihydroorotate dehydrogenase is dependent on a conserved loop identified by sequence homology, mutagenesis, and limited proteolysis. Biochemistry Us 1999, 38 (10), 2899 2908. 67. Rawls, J.; Knecht, W.; Diekert, K.; Lill, R. ; Loffler, M., Requirements for the mitochondrial import and localization of dihydroorotate dehydrogenase. Eur J Biochem 2000, 267 (7), 2079 2087. 68. Andersen, P. S.; Jansen, P. J. G.; Hammer, K., 2 Different Dihydroorotate Dehydrogenases in Lactococcus L actis. J Bacteriol 1994, 176 (13), 3975 3982. 69. (a) Rowland, P.; Nielsen, F. S.; Jensen, K. F.; Larsen, S., The crystal structure of the flavin containing enzyme dihydroorotate dehydrogenase A from Lactococcus lactis. Structure 1997, 5 (2), 239 252; (b) Andersen, P. S.; Martinussen, J.; Hammer, K., Sequence analysis and identification of the pyrKDbF operon from Lactococcus lactis including a novel gene, pyrK, involved in pyrimidine biosynthesis. J Bacteriol 1996, 178 (16), 5005 5012. 70. Nielsen, F. S .; Rowland, P.; Larsen, S.; Jensen, K. F., Purification and characterization of dihydroorotate dehydrogenase A from Lactococcus lactis, crystallization and preliminary X ray diffraction studies of the enzyme. Protein Sci 1996, 5 (5), 852 856. 71. Li, X. T. ; Weinstock, G. M.; Murray, B. E., Generation of Auxotrophic Mutants of Enterococcus Faecalis. J Bacteriol 1995, 177 (23), 6866 6873. 72. Marcinkeviciene, J.; Jiang, W. J.; Locke, G.; Kopcho, L. M.; Rogers, M. J.; Copeland, R. A., A second dihydroorotate d ehydrogenase (type A) of the human pathogen Enterococcus faecalis: Expression, purification, and steady state kinetic mechanism. Arch Biochem Biophys 2000, 377 (1), 178 186. 73. Nielsen, F. S.; Andersen, P. S.; Jensen, K. F., The B form of dihydroorotate d ehydrogenase from Lactococcus lactis consists of two different subunits, encoded by the pyrDb and pyrK genes, and contains FMN, FAD, and [FeS] redox centers. J Biol Chem 1996, 271 (46), 29359 29365. 74. Bjornberg, O.; Rowland, P.; Larsen, S.; Jensen, K. F. Active site of dihydroorotate dehydrogenase A from Lactococcus lactis investigated by chemical modification and mutagenesis. Biochemistry Us 1997, 36 (51), 16197 16205. 75. Fagan, R. L.; Nelson, M. N.; Pagano, P. M.; Palfey, B. A., Mechanism of flavin re duction in class 2 dihydroorotate dehydrogenases. Biochemistry Us 2006, 45 (50), 14926 14932.
188 76. Hurt, D. E.; Sutton, A. E.; Clardy, J., Brequinar derivatives and species specific drug design for dihydroorotate dehydrogenase. Bioorg Med Chem Lett 2006, 16 (6), 1610 1615. 77. Rowland, P.; Bjornberg, O.; Nielsen, F. S.; Jensen, K. F.; Larsen, S., The crystal structure of Lactococcus lactis dihydroorotate dehydrogenase A complexed with the enzyme reaction product throws light on its enzymatic function. Protei n Sci 1998, 7 (6), 1269 1279. 78. Takashima, E.; Inaoka, D. K.; Osanai, A.; Nara, T.; Odaka, M.; Aoki, T.; Inaka, K.; Harada, S.; Kita, K., Characterization of the dihydroorotate dehydrogenase as a soluble fumarate reductase in Trypanosoma cruzi. Mol Bioch em Parasit 2002, 122 (2), 189 200. 79. Weckbecker, A.; Groger, H.; Hummel, W., Regeneration of Nicotinamide Coenzymes: Principles and Applications for the Synthesis of Chiral Compounds. Adv Biochem Eng Biot 2010, 120 195 242. 80. Ingelman, M.; Bianchi, V. ; Eklund, H., The three dimensional structure of flavodoxin reductase from Escherichia coli at 1.7 angstrom resolution. J Mol Biol 1997, 268 (1), 147 157. 81. Beinert, H.; Holm, R. H.; Munck, E., Iron sulfur clusters: Nature's modular, multipurpose structures. Science 1997, 277 (5326), 653 659. 82. Jaganaman, S.; Pinto, A.; Tarasev, M.; Ballou, D. P., High levels of expression of the iron sulfur proteins phthala te dioxygenase and phthalate dioxygenase reductase in Escherichia coli. Protein Expres Purif 2007, 52 (2), 273 279. 83. Rowland, P.; Norager, S.; Jensen, K. F.; Larsen, S., Structure of dihydroorotate dehydrogenase B: Electron transfer between two flavin g roups bridged by an iron sulphur cluster. Structure 2000, 8 (12), 1227 1238. 84. Norager, S.; Arent, S.; Bjornberg, O.; Ottosen, M.; Lo Leggio, L.; Jensen, K. F.; Larsen, S., Lactococcus lactis dihydroorotate dehydrogenase A mutants reveal important facets of the enzymatic function. J Biol Chem 2003, 278 (31), 28812 28822. 85. Combe, J. P.; Basran, J.; Hothi, P.; Leys, D.; Rigby, S. E. J.; Munro, A. W.; Scrutton, N. S., Lys D48 is required for charge stabilization, rapid flavin reduction, and internal elect ron transfer in the catalytic cycle of dihydroorotate dehydrogenase B of Lactococcus lactis. J Biol Chem 2006, 281 (26), 17977 17988. 86. Aitken, C. E.; Marshall, R. A.; Puglisi, J. D., An oxygen scavenging system for improvement of dye stability in single molecule fluorescence experiments. Biophys J 2008, 94 (5), 1826 1835.
18 9 87. Mattevi, A., To be or not to be an oxidase: challenging the oxygen reactivity of flavoenzymes. Trends Biochem Sci 2006, 31 (5), 276 283.
190 BIOGRAPHICAL SKETCH Yuri A. A. Pompeu w as born in Tubar o, Santa Catarina, Brazil. He was raised in Sorocaba, So Paulo and after graduating from high school in 2003 he moved to the US to accept a scholarship as a student athlete at Murray State University. Yuri played varsity tennis for all fo ur years and g raduated with a B.S. degree in c hemistry. He performed undergraduate research in the laboratory of Professor Kevin Revell. In 2008, Yuri started his graduate studies at the University of Florida under the supervision of Professor Jon Stewart. His initial research focused on electrochemical reductions mediated by OYE1 but by his second year the focused had completely shifted towards X ray crystallography and structure guided protein engineering. Largely influenced by Professor Steve Bruner, X r ay crystallography became his true passion. Towards the end o f his graduate career, Yuri beca me involved in a project that studied the genetic association between the protein HLA B57:01 and abacavir hypersensitivity syndrome (AHS). Along with Professor Dav id Ostrov, Yuri was able to solve the crystal ligand. This structure provided the structural basis for (AHS), and has been of significant im pact in the field of immunology and has led to the search for other diseases that are now believed to occur through a similar mechanism. In the future Yuri will focus his studies on human immunology, with a particular interest in disease related structural immunology.