sign in sign up
<<
Home , Baccatin III, Biosynthesis, Semisynthesis, Taxadiene synthase, Tesetaxel

DEVELOPMENT OF A FOUR-STEP SEMI- OF THE ANTICANCER DRUG AND ITS ANALOGUES

By

Chelsea Thornburg

A DISSERTATION

Submitted to Michigan State University in partial fulfillment of the requirements for the degree of

Biochemistry and Molecular ‒ Doctor of Philosophy

2015 ABSTRACT

DEVELOPMENT OF A FOUR-STEP SEMI-BIOSYNTHESIS OF THE ANTICANCER DRUG PACLITAXEL AND ITS ANALOGUES

By

Chelsea Thornburg

Paclitaxel (Taxol®) is a widely used chemotherapeutic drug with additional medical applications in drug-eluting stents as an anti-restenosis treatment. Paclitaxel is a structurally complex natural with an excellent scaffold for designing analogs with pharmacological properties. To date, clinically approved analogs include and for the treatment of additional cancers. Currently, fermentation methods produce paclitaxel and large quantities of the precursors 10-deacetylbaccatin III

(10-DAB) and baccatin III. The complexity of the semi-characterized ~19-step paclitaxel biosynthetic pathway limits bioengineering attempts. However, the availability of 10-DAB and baccatin III suggests a semi-biosynthetic pathway to paclitaxel starting with these precursors is feasible. We have designed a short, simple biosynthetic pathway, capable of making paclitaxel, analogs, and/or valuable precursors for the semi-synthesis of additional analogs of biological interest.

The paclitaxel biosynthesis baccatin III: 3-amino-13-O-phenylpropanoyl

CoA (BAPT) and the bacterial (2R,3S)-phenylisoserinyl CoA (PheAT) produce N-debenzoylpaclitaxel, N-debenzoyldocetaxel, or analogs. The addition of the paclitaxel biosynthetic N-debenzoyltaxol-N-benzoyltransferase (NDTNBT) and the bacterial benzoate CoA ligase (BadA) produce paclitaxel or other N-acylated analogs. In this dissertation, BAPT and BadA are kinetically characterized. The specificity of BadA was systematically investigated with a series of 24 substrates. Six crystal structures of BadA in complex with different substrates, including benzoyl AMP, are used to explain BadA reactivity and propose rational mutations (A227A, H333A, and I334A) that expand substrate specificity and provide insight into the BadA mechanism and connect with established regulatory mechanisms in .

Major hurdles including solubility and substrate availability, were overcome in order to characterize BAPT activity in the proposed semi-biosynthetic pathway. BAPT was purified as a fusion with maltose binding protein and its (2R,3S)-phenylisoserinyl

CoA substrate was biosynthesized. To our knowledge this is the first time (2R,3S)- phenylisoserinyl CoA has been isolated in quantitative yields high enough to allow for characterization of the Michaelis-Menten kinetic constants (kcat and KM) for BAPT. This dissertation also describes the combination of BAPT and a bacterial ligase (PheAT) to produce N-debenzoylpaclitaxel and N-debenzoyl-10-deacetylpaclitaxel, precursors of paclitaxel and docetaxel, respectively. Biosynthesis of a biologically active paclitaxel analog, N-2-furanyl-N-debenzoylpaclitaxel, using the aforementioned , is also demonstrated as proof-of-principle that this semi-biosynthetic pathway may shorten the number of steps required to make certain paclitaxel (and docetaxel) analogs of interest.

Copyright by CHELSEA THORNBURG 2015

ACKNOWLEDGEMENTS

I would like to acknowledge my advisor, Dr. Kevin Walker, for his support during my time here at Michigan State University. I also would like to acknowledge Dr. Dan Jones for all his advice and assistance in learning . Dr. Jim Geiger kindly trusted me with his equipment and was a great collaborator for the BadA ligase work.

The and (BMB) department has been a wonderful academic home. Faculty members were always willing to discuss any research problems I encountered along the way. The following professors sat down with me at some point and said helpful things: Bill Henry, Tom Sharkey, Honggao Yan, Kaillathe “Pappan”

Padmanabhan, Charlie Hoogstraten, and Kristin Parent “KP”.

I have to thank my family for all their love and support. My mom- Kristen

Thornburg, my Papi- Mark Santas, my sisters- Caitlin Thornburg and Rhoda Brew-Appiah, my bro-in-law Matt Seidel, and my niece Madison are always there for me even though they have no idea what I do all day. My GREAT aunt Frankie and uncle Jim welcomed me into their home and are two of my favorite people. I am also grateful to the lovely Janelle and James Sabo (and the girls: Claire, Katherine, and Sophia) for welcoming me into their home for Thanksgiving these past few years.

I also need to thank all the people I have lost during my doctoral program. My grandfather- Newton Thornburg, my grandmother- Cloteel Atkins, my dear friend Pam

Movalson and her daughter, Christine. I miss you all.

v TABLE OF CONTENTS

LIST OF TABLES ...... x

LIST OF FIGURES ...... xi

KEY TO ABBREVIATIONS ...... xvii

Chapter 1. Clinical use and production of paclitaxel and analogs of clinical interest ...... 1 1.1 Introduction ...... 1 1.1.1 Clinical uses of paclitaxel ...... 1 1.1.2 Clinical uses of paclitaxel analogs ...... 3 1.1.3 A brief history of paclitaxel ...... 3 1.1.4 Paclitaxel mode of action ...... 5 1.1.5 Paclitaxel biosynthesis ...... 7 1.1.6 Paclitaxel production ...... 11 1.1.7 Semi-biosynthesis of paclitaxel, precursors, and analogs ...... 14 REFERENCES ...... 19

Chapter 2. Kinetically- and crystallographically-guided mutations of a benzoate CoA ligase (BadA) elucidate mechanism and expand substrate permissivity ...... 33 2.1 Introduction ...... 33 2.2 Experimental ...... 38 2.2.1 Materials ...... 38 2.2.2 Plasmids ...... 39 2.2.3 BadA protein expression and purification ...... 39 2.2.4 BadA kinetic assays ...... 40 2.2.5 BadA assay analysis by liquid chromatography mass spectrometry ...... 41 2.2.6 BadA mutations ...... 42 2.2.7 Kinetic analysis ...... 43 2.2.8 BadA crystal structures ...... 43 2.2.8.1 of R. palustris benzoate: coenzyme A ligase (BadA) ...... 43 2.2.8.2 Co-crystallization to obtain the ligand bound structure ...... 44 2.2.8.3 Data processing and refinement of BadA ...... 44 2.2.9 Calculation of covalent van der Waals volumes and lengths ...... 45 2.3 Results ...... 46 2.3.1 Solving the BadA structure ...... 46 2.3.1.1 Domain orientation ...... 46 2.3.1.2 Features of the BadA ...... 47 2.3.2 Kinetic properties of BadA ...... 50 2.3.3 Substrate turnover by BadA ...... 50 2.3.3.1 Halogenated benzoates ...... 50 2.3.3.2 Benzoates with strongly electron-withdrawing substituents ...... 52 2.3.3.3 Benzoates with strongly electron-donating substituents ...... 52

vi 2.3.3.4 Turnover of heteroaromatic carboxylates ...... 53 2.3.3.5 Turnover of non-aromatic carbocycle carboxylates ...... 54 2.3.4 Rational Mutation of the BadA Active Site ...... 54 2.3.4.1 Ala227Gly-BadA mutant ...... 56 2.3.4.2 Ile334Ala-BadA mutant ...... 56 2.3.4.3 His333Ala-BadA mutant ...... 57 2.3.4.4 Leu332Ala-BadA mutant ...... 57 2.4 Discussion ...... 58 2.4.1 BadA structure and homology ...... 58 2.4.2 Catalytically important residues in BadA ...... 62 2.4.3 Structural rationale for substrate specificity of BadA...... 64 2.4.3.1 Non-aromatic carbocycle carboxylates ...... 67 2.4.4 Analysis of point mutants of BadA ...... 68 2.5 Conclusions ...... 70 APPENDIX ...... 72 REFERENCES ...... 93

Chapter 3. Expression and purification of the Taxus cuspidata baccatin III: 3-amino-13-O- phenylpropanoyl transferase (BAPT) ...... 99 3.1 Introduction ...... 99 3.2 Experimental ...... 107 3.2.1 Materials ...... 107 3.2.2 BAPT purification from Pichia pastoris...... 107 3.2.2.1 Cloning of pHisBAPT and pMycBAPT ...... 107 3.2.2.2 BAPT expression in P. pastoris ...... 109 3.2.2.3 BAPT Western blot ...... 109 3.2.3 BAPT purification from Escherichia coli ...... 110 3.2.3.1 Cloning of pNterBAPT ...... 110 3.2.3.2 Cloning of pCterBAPT...... 110 3.2.3.3 Cloning of pOptBapt ...... 111 3.2.3.4 Cloning of pMBP-CterBAPT and pMBP-NterBAPT ...... 111 3.2.3.5 E. coli strains ...... 112 3.2.3.6 BAPT expression in E. coli ...... 112 3.2.3.7 Nickel-affinity chromatography ...... 113 3.2.3.8 Ammonium sulfate fractionation ...... 114 3.2.3.9 exchange chromatography ...... 114 3.2.4 Optimized NterBAPT expression and purification...... 115 3.2.5 MBP-BAPT expression and purification ...... 115 3.2.6 Band densitometry ...... 116 3.2.7 Proteomics...... 117 3.2.8 BAPT activity assays ...... 117 3.3 Results and discussion ...... 119 3.3.1 BAPT expression in Pichia pastoris ...... 119 3.3.2 BAPT expression and purification in E. coli ...... 122 3.3.2.1 Expression optimization in E. coli ...... 122 3.3.2.2 Purification by nickel-affinity chromatography ...... 125

vii 3.3.2.3 Ammonium sulfate precipitation ...... 127 3.3.2.4 Ion exchange chromatography ...... 127 3.3.3 Optimized NterBAPT purification ...... 128 3.3.4 BAPT activity assays ...... 131 3.3.5 MBP-BAPT expression and purification ...... 135 3.4 Conclusions ...... 138 3.5 Future Research ...... 140 APPENDIX ...... 141 REFERENCES ...... 147

Chapter 4. Paclitaxel analog biosynthesis from baccatin III with a four-enzyme in vitro system and characterization of baccatin III-3-amino-13-O-phenylpropanoyl CoA transferase (BAPT) ...... 153 4.1 Introduction ...... 153 4.1.1 Engineering the paclitaxel biosynthetic pathway ...... 153 4.1.2 Proposed paclitaxel semi-biosynthesis ...... 160 4.1.3 Applications in paclitaxel analog production ...... 162 4.1.4 BAPT characterization and proof-of-principle for paclitaxel/analog biosynthesis ...... 163 4.2 Experimental ...... 164 4.2.1 Materials ...... 164 4.2.2 of (3R)-β-phenylalanyl CoA ...... 164 4.2.3 Method I. Synthesis of (2R,3S)-phenylisoserinyl CoA ...... 166 4.2.4 Method II. Synthesis of (2R,3S)-phenylisoserinyl CoA ...... 167 4.2.5 Method III. Synthesis of (2R,3S)-phenylisoserinyl CoA ...... 167 4.2.6 Method IV. Synthesis of (2R,3S)-phenylisoserinyl CoA ...... 168 4.2.7 PheAT purification...... 169 4.2.8 Biosynthesis of (2R,3S)-phenylisoserinyl CoA ...... 169 4.2.9 Purification of (2R,3S)-phenylisoserinyl CoA ...... 170 4.2.10 HPLC Analysis of acyl CoA thioesters ...... 171 4.2.11 Acyl CoA purity analysis (Ellman assay) ...... 172 4.2.12 BAPT, BadA, and NDTNBT purification ...... 173 4.2.13 BAPT kinetic assays ...... 173 4.2.14 Kinetic analysis ...... 174 4.2.15 Liquid chromatography mass spectrometry: BAPT assay analysis ...... 174 4.2.16 PheAT and BAPT coupled reactions ...... 175 4.2.17 Production of paclitaxel and its analogs in a coupled enzyme assay ...... 175 4.3 Results and discussion ...... 177 4.3.1 Synthesis of acyl CoA substrates ...... 177 4.3.1.1 Synthesis of (3R)-β-phenylalanyl CoA ...... 178 4.3.1.2 Synthesis of (2R,3S)-Phenylisoserinyl CoA ...... 179 4.3.2 Biosynthesis of (2R,3S)-PhIS CoA...... 182 4.3.3 Michaelis-Menten kinetics of BAPT ...... 184 4.3.3.1 (3R)-β-phenylalanyl CoA and baccatin III substrates ...... 184 4.3.3.2 (2R,3S)-Phenylisoserinyl CoA and baccatin III substrates ...... 186

viii 4.3.3.3 (2R,3S)-Phenylisoserinyl CoA and 10-deacetylbaccatin III as substrates ...... 186 4.3.4 BAPT activity and substrate assisted ...... 187 4.3.5 Coupled biosynthesis of N-debenzoylpaclitaxel: PheAT and BAPT...... 193 4.3.6 Biosynthesis of N-2-furanoyl-N-debenzoylpaclitaxel: a paclitaxel analog………...... 194 4.4 Conclusions ...... 197 4.5 Future Research ...... 199 APPENDIX ...... 201 REFERENCES ...... 223

ix LIST OF TABLES

Table 1.1. Clinically-approved , their current uses, and open clinical trials...... 3

Table 2.1. Kinetic parameters of BadA for various substrates...... 51

Table 2.2. Relative apparent maximum rates of BadA and point mutants for various substrates...... 55

Table 3.1. Biologically active analogs as active as paclitaxel with modifications in the C13- phenylisoserinyl sidechain...... 106

Table 3.2. Proteomics analysis of purified NterBAPT...... 130

Table 3.3. Proteomics analysis of purified MBP-BAPT...... 137

Table 4.1. Simple analogs of paclitaxel with biological activity...... 162

Table 4.2. Kinetic constants of enzymes required for proposed paclitaxel biosynthesis from baccatin III...... 187

Table 4.3. Michaelis-Menten kinetic constants from BAHD acyltransferase family members...... 191

x LIST OF FIGURES

Figure 1.1. Clinically-approved taxanes ...... 2

Figure 1.2. Paclitaxel mode of action ...... 5

Figure 1.3. Taxoid derivatives of taxa-4(5),11(12)-...... 7

Figure 1.4. Paclitaxel biosynthesis: Part 1 ...... 8

Figure 1.5. Paclitaxel biosynthesis: Part 2 ...... 10

Figure 1.6. Examples of off-pathway metabolites that divert away from paclitaxel production ...... 11

Figure 1.7. Semi-synthesis of paclitaxel from 10-DAB ...... 13

Figure 1.8. Taxanes clinically approved and in clinical trials ...... 14

Figure 1.9. Coupled enzyme biosynthesis of paclitaxel...... 15

Figure 2.1. Representative natural products derived from acyl CoA thioesters ...... 34

Figure 2.2. ATP-dependent, two-step mechanism of a benzoate: CoA ligase in the presence of magnesium ...... 36

Figure 2.3. Acyl CoA detection by tandem mass spectrometry ...... 41

Figure 2.4. Crystal structure of BadA in complex with benzoate ...... 46

Figure 2.5. The active site of the BadA structure ...... 49

Figure 2.6. BadA structural alignments and C-terminal domain orientation ...... 61

Figure 2.7. Active site of BadA showing possible polar contacts between Lys427 and Bz- AMP...... 63

Figure A.1.1. Michaelis-Menten plot of biosynthetic benzoyl CoA catalyzed by BadA. .73

Figure A.1.2. Michaelis-Menten plot of biosynthetic thiophene-2-carbonyl CoA catalyzed by BadA ...... 73

Figure A.1.3. Michaelis-Menten plot of biosynthetic 3-furoyl CoA catalyzed by BadA ..74

Figure A.1.4. Michaelis-Menten plot of biosynthetic cyclohexanoyl CoA catalyzed by BadA ...... 74

xi Figure A.1.5. Michaelis-Menten plot of biosynthetic 1-cyclohexen-1-oyl CoA catalyzed by BadA...... 75

Figure A.1.6. Michaelis-Menten plot of biosynthetic 3-cyclohexen-1-oyl CoA catalyzed by BadA...... 75

Figure A.1.7. Michaelis-Menten plot of biosynthetic 2-fluorobenzoyl CoA catalyzed by BadA...... 76

Figure A.1.8. Michaelis-Menten plot of biosynthetic 3-fluorobenzoyl CoA catalyzed by BadA ...... 76

Figure A.1.9. Michaelis-Menten plot of biosynthetic 4-fluorobenzoyl CoA catalyzed by BadA...... 77

Figure A.1.10. Michaelis-Menten plot of biosynthetic 2-chlorobenzoyl CoA catalyzed by BadA...... 77

Figure A.1.11. Michaelis-Menten plot of biosynthetic 3-chlorobenzoyl CoA catalyzed by BadA ...... 78

Figure A.1.12. Michaelis-Menten plot of biosynthetic 4-chlorobenzoyl CoA catalyzed by BadA ...... 78

Figure A.1.13. Michaelis-Menten plot of biosynthetic 2-aminobenzoyl CoA catalyzed by BadA ...... 79

Figure A.1.14. Michaelis-Menten plot of biosynthetic 3-aminobenzoate CoA catalyzed by BadA ...... 79

Figure A.1.15. Michaelis-Menten plot of biosynthetic 4-aminobenzoyl CoA catalyzed by BadA ...... 80

Figure A.1.16. Michaelis-Menten plot of biosynthetic 2-hydroxybenzoyl CoA catalyzed by BadA ...... 80

Figure A.1.17. Michaelis-Menten plot of biosynthetic 3-hydroxybenzoyl CoA catalyzed by BadA ...... 81

Figure A.1.18. Michaelis-Menten plot of biosynthetic 4-hydroxybenzoyl CoA catalyzed by BadA ...... 81

Figure A.1.19. Michaelis-Menten plot of biosynthetic 2-methylbenzoyl CoA catalyzed by BadA ...... 82

Figure A.1.20. Michaelis-Menten plot of biosynthetic 3-methylbenzoyl CoA catalyzed by BadA ...... 82

xii Figure A.1.21. Michaelis-Menten plot of biosynthetic 4-methylbenzoate CoA catalyzed by BadA ...... 83

Figure A.1.22. Michaelis-Menten plot of biosynthetic 2-cyanobenzoyl CoA catalyzed by BadA ...... 83

Figure A.1.23. Michaelis-Menten plot of biosynthetic 2-methoxybenzoyl CoA catalyzed by BadA ...... 84

Figure A.1.24. Michaelis-Menten plot of biosynthetic 2-nitrobenzoyl CoA catalyzed by BadA ...... 84

Figure A.1.25. Crystals of Rhodopseudomonas palustris benzoate CoA ligase ...... 85

Figure A.1.26. Overlay of BadA structures in complex with bound ligands...... 85

Figure A.1.27. Michaelis-Menten plot of benzoyl CoA catalyzed by the Lys427Ala-BadA mutant...... 86

Figure A.1.28. Benzoyl AMP (Bz-AMP) produced by BadA and the Lys427Ala-BadA mutant ...... 87

Figure A.1.29. Shifted position of Tyr432 in BadA bound with bzAMP ...... 88

Figure A.1.30. Partial sequence alignment of selected benzoate CoA ...... 89

Figure A.1.31. Comparison of the position of Phe226 of BadA (in the thiolation conformation) and that of Phe236 of BCLM (in the adenylation conformation) bound to benzoate ...... 89

Figure A.1.32. Partial amino acid sequence alignment of a broad selection of ATP- dependent adenylases and coenzyme A ligases...... 90

Figure A.1.33. Relative position of benzoate and benzoyl AMP in the BadA active site. 90

Figure 3.1. Proposed mechanism of BAHD acyltransferases ...... 100

Figure 3.2. Examples of BAHD enzyme products...... 101

Figure 3.3. BAHD structure of the Sorghum hydroxycinnamoyltransferase (HCT) ...... 102

Figure 3.4. Condensed paclitaxel biosynthesis from geranylgeranyl diphosphate showing the BAHD acyltransferases in the pathway...... 103

Figure 3.5. SDS-PAGE of whole cell lysate from expressions of BAPT clones from P. pastoris ...... 120

Figure 3.6. NterBAPT expression in crude cell lysates...... 121

xiii Figure 3.7. Soluble expression of NterBAPT with varied IPTG concentrations ...... 122

Figure 3.8. Expression and induction of OptBAPT in E. coli ...... 123

Figure 3.9. Relative expression levels of NterBAPT and the Taxus N-benzoyltransferase (NDTNBT) by SDS-PAGE...... 124

Figure 3.10. SDS-PAGE of NterBAPT purification by nickel-affinity chromatography125

Figure 3.11. NterBAPT activity during nickel-affinity chromatography ...... 126

Figure 3.12. Ammonium sulfate fractionation of NterBAPT clarified lysate ...... 127

Figure 3.13. Purification of NterBAPT by multiple chromatography steps ...... 129

Figure 3.14. Coupled assay schematic with PheAT and BAPT ...... 131

Figure 3.15. Biosynthesis of N-debenzoyl-2′-deoxypaclitaxel by BAPT ...... 132

Figure 3.16. Representative LC-ESI/MS/MS of a BAPT assay...... 134

Figure 3.17. Domain structures of MBP-BAPT fusion ...... 135

Figure 3.18. Representative purification of MBP-BAPT...... 136

Figure A.2.1. Multiple sequence alignment of Taxus cuspidata acyltransferases ...... 142

Figure A.2.2. HisBAPT amino acid sequence for expression in P. pastoris ...... 143

Figure A.2.3. cMycHisBAPT amino acid sequence for expression in P. pastoris...... 143

Figure A.2.4. NterBAPT amino acid sequence for expression in E. coli...... 143

Figure A.2.5. CterBAPT amino acid sequence for expression in E. coli ...... 143

Figure A.2.6. Codon optimized bapt gene sequence ...... 144

Figure A.2.7. MBP-CterBAPT (95.2 kDa) amino acid sequence for expression in E. coli...... 145

Figure A.2.8. Quantification of BAPT by band densitometry...... 145

Figure A.2.9. SDS-PAGE gel of purified NterBAPT and MBP-CterBAPT for proteomics analysis...... 146

Figure 4.1. Condensed scheme of paclitaxel biosynthesis ...... 155

Figure 4.2. Products of an engineered chimeric cytochrome P450 -5α- hydroxylase expressed in E. coli ...... 157

xiv Figure 4.3. Coupled enzyme biosynthesis of paclitaxel ...... 159

Figure 4.4. Taxanes clinically approved and in clinical trials ...... 161

Figure 4.5. Structures of (3R)-β-phenylalanyl coenzyme A (R = H) and (2R,3S)- phenylisoserinyl coenzyme A (R = OH)...... 177

Figure 4.6. of (3R)-β-Phenylalanyl CoA ...... 179

Figure 4.7. Conformation of (2R,3S)-phenylisoserine ...... 181

Figure 4.8. Biosynthesis of (2R,3S)-phenylisoserinyl CoA with PheAT, a truncated form of tyrocidine synthetase A (TycA) ...... 182

Figure 4.9. Biosynthetic reaction progress curve for (2R,3S)-PhIS CoA ...... 183

Figure 4.10. BAPT-catalyzed enzyme reaction ...... 184

Figure 4.11. Multiple sequence alignment of bapt genes from Taxus sp...... 188

Figure 4.12. Proposed mechanisms for BAHD acyltransferases and BAPT ...... 189

Figure 4.13. Homology model of BAPT based on the hydroxycinnamoyl transferase (PDB: 4G0B) from Coffea canephora ...... 192

Figure 4.14. N-debenzoylpaclitaxel biosynthesized from BAPT and PheAT coupled reactions ...... 193

Figure 4.15. Biosynthesis of N-2-furanoyl-N-debenzoylpaclitaxel with PheAT, BAPT, BadA, and NDTNBT ...... 194

Figure 4.16. Overlapping substrate specificities of PheAT, BAPT, BadA, and NDTNBT ...... 196

Figure A.3.1. 1H-NMR of R-N-Boc-3-amino-3-phenylpropanoic acid...... 202

Figure A.3.2. 1H-NMR of chemically synthesized (3R)-β-phenylalanyl CoA...... 203

Figure A.3.3. HPLC chromatogram of purified (3R)-β-phenylalanyl CoA...... 204

Figure A.3.4. Mass spectra of purified (3R)-β-phenylalanyl CoA ...... 205

Figure A.3.5. LC-ESI/MSMS chromatogram showing the production of N-debenzoyl-2′- deoxypaclitaxel...... 206

Figure A.3.6. (Scheme I) Synthesis of (2R,3S)-phenylisoserinyl CoA...... 206

Figure A.3.7. (Scheme II) (2R,3S)-PhIS CoA synthesis with 2-OH protection...... 207

xv Figure A.3.8. (Scheme III) (2R,3S)-PhIS CoA synthesis by DCC/ HOBt coupling...... 207

Figure A.3.9. (Scheme IV) (2R,3S)-PhIS CoA synthesis by DCC/NHS coupling...... 208

Figure A.3.10. 1H-NMR of (2R,3S)-N-Boc-3-amino-3-phenylisoserine...... 209

Figure A.3.11. 1H-NMR of biosynthetic (2R,3S)-phenylisoserinyl CoA...... 210

Figure A.3.12. 13C-NMR of biosynthetic (2R,3S)-phenylisoserinyl CoA...... 211

Figure A.3.13. HPLC analysis of purified (2R,3S)-phenylisoserinyl CoA...... 212

Figure A.3.14. Mass spectra of purified (2R,3S)-phenylalanyl CoA ...... 213

Figure A.3.15. Michaelis-Menten kinetic plot of biosynthetic N-debenzoylpaclitaxel produced by MBP-BAPT ...... 214

Figure A.3.16. Michaelis-Menten kinetic plot of biosynthetic N-debenzoyl-2′- deoxypaclitaxel produced by MBP-BAPT ...... 214

Figure A.3.17. Michaelis-Menten kinetic plot of biosynthetic N-debenzoyl-2′- deoxypaclitaxel produced by MBP-BAPT...... 215

Figure A.3.18. BAPT activity with 10-DAB and (2R,3S)-PhIS CoA...... 216

Figure A.3.19. Purified recombinant PheAT ...... 217

Figure A.3.20. Production of N-debenzoyl-2’-deoxypaclitaxel by BAPT...... 218

Figure A.3.21. Production of N-debenzoylpaclitaxel by BAPT...... 220

Figure A.3.22. BAPT and PheAT coupled assay production of N-debenzoylpaclitaxel. 221

Figure A.3.23. Multi-enzyme biosynthesis of N-furanoyl-N-debenzoylpaclitaxel...... 222

xvi KEY TO ABBREVIATIONS

(2R,3S)-PhIS CoA (2R,3S)-phenylisoserinyl CoA

(3R)-β-Phe CoA (3R)-β-phenylalanyl CoA

4CL 4-chlorobenzoate CoA ligase (Alcaligenes sp.)

10-DAB 10-deacetylbaccatin III

Å angstrom

A absorbance

Ac acetyl

Aq aqueous

AIDS acquired immune deficiency syndrome

AMP monophosphate

ATP

BadA benzoate CoA ligase (Rhodopseudomonas

palustris)

BAPT baccatin III: 3-amino-13-O-phenylpropanoyl

CoA transferase

BCLM benzoate CoA ligase (Burkholderia

xenovorans)

Bz benzoyl

BSA bovine serum albumin

BzO benzoate

CAN cerium(IV) ammonium nitrate

xvii CH2Cl2 methylene chloride

CoA coenzyme A

Cyc cyclohexane

1-Cyc 1-cyclohexen-1-carboxylic acid

3-Cyc 3-cyclohexen-1-carboxylic acid

DBAT 10-deacetylbaccatin III 10-O-acetyltransferase

DTT dithiothreitol

DMADP dimethylallyldiphosphate

EDTA ethylenediaminetetraacetic acid

EtOAc ethyl h hour

GTP triphosphate

H2O water

HCl hydrochloric acid

HEPES 4-(2-hydroxyethyl)-1-

piperazineethanesulfonic acid

HPLC high performance liquid chromatography

IDP isopentenyl diphosphate

IPTG isopropyl β-D-1-thiogalactopyranoside kDa kilodalton

LB Luria Bertani

NHS N-hydroxysuccinimide ester

OD optical density

xviii MBP maltose binding protein

MeOH methanol min minute

MS mass spectrometry

MOPS 3-(N-morpholino)propanesulfonic acid

NDTNBT N-debenzoyltaxol-N-benzoyltransferase

NMR nuclear magnetic resonance PCF plant cell fermentation PCR polymerase chain reaction Ph phenyl PheAT phenylisoserine CoA ligase PMSF phenylmethanesulfonylfluoride

PPi inorganic RPM revolutions per minute r.t. room temperature s second

SDS-PAGE sodium docecylsulfate polyacrylamide gel

electrophoresis

TES triethylsilyl

THF tetrahydrofuran

TLC thin layer chromatography

TS taxadiene synthase

UV ultraviolet

xix Chapter 1. Clinical use and production of paclitaxel and analogs of clinical interest

1.1 Introduction

Cancer is a leading cause of disease and death worldwide (1). In the United States, the ten most common forms include cancers of the breast, lung, prostate, colon and rectum, bladder, thyroid, kidney, endometrium, and pancreas as well as melanoma and leukemia

(1). The most lethal of these cancers are certain pancreatic, lung, leukemia, and colorectal cancers with mortality rates between 30% and 80% (2). Paclitaxel and the analogs docetaxel and cabazitaxel are approved drugs for use with many types of cancer. Novel preparations of these compounds and other analogs show promise for treating previously chemoresistant cancers (3).

1.1.1 Clinical uses of paclitaxel

Since paclitaxel (Taxol®) was originally approved as a chemotherapeutic for - refractory ovarian cancer in 1992, its usage has expanded to include treatment for a wide variety of cancers including metastatic , non-small cell lung cancer, and

AIDS-related Kaposi’s sarcoma (1). More recently, paclitaxel was approved as nano- particle albumin-bound (nab)-paclitaxel (under the brand name Abraxane) in combination with (a nucleoside analog) for the first-line treatment of pancreatic metastatic adenocarcinoma (3, 4). Conjugated nab-paclitaxel was designed to facilitate greater drug solubility and uptake by tumor cells (33%) over sb-paclitaxel (paclitaxel solubilized in

Cremophor® EL formulation) (5).

1 Paclitaxel is also used off-label for the treatment of endometrial (6), cervical (7), and gastroesophageal cancers (8-10). There are currently 345 open clinical trials for paclitaxel and nab-paclitaxel in the United States (Table 1.1) (11). Many of these clinical trials are for paclitaxel in combination therapies with other drugs for treating different breast, ovarian, cervical, pancreatic, melanoma, non-small cell lung, cutaneous squamous cell carcinoma, and bile duct (cholangiocarcinoma) cancers among others (11).

Non-cancer applications with paclitaxel include its use as a coating in drug-eluting stents to prevent in-stent restenosis in heart disease patients (12-14). Usage at lower doses shows activity with other diseases (15). In particular, paclitaxel shows preliminary activity preventing pulmonary fibrosis (16), inflammation (17, 18), and axon regeneration (19, 20).

The broad chemotherapeutic activity of paclitaxel and its growing usage in other diseases stands to increase its demand (21).

Figure 1.1. Clinically-approved taxanes. Brand names are shown in parentheses.

2 1.1.2 Clinical uses of paclitaxel analogs

Docetaxel is a semi-synthetic analog of paclitaxel with modifications at the C10 and the

3′N sidechain positions (Figure 1.1). Docetaxel was originally approved for the treatment of non-small cell lung cancer in 1999 (1). It was subsequently approved for the treatment of prostate, breast, gastric, and head and neck cancers (1). In 2010, cabazitaxel, a second analog of paclitaxel with modifications at the C7, C10, and 3’N positions was approved as a second-line treatment for castration-resistant hormone refractory prostate cancer (Figure

1.1) (22). There are currently 143 open clinical trials for docetaxel and 11 for cabazitaxel in the United States (Table 1.1) (11). Many of these clinical trials are for first or second- line combination therapies in different types of cancer such as prostate, head and neck, urothelial cell, breast, solid tumor, and stomach cancers.

Table 1.1. Clinically-approved taxanes, their current uses, and open clinical trials.

Clinical Trials1 Drug Currently Approved Uses United States Worldwide Ovarian, breast, non-small cell lung cancers Paclitaxel 240 783 and AIDS-related Kaposi’s sarcoma Breast, non-small cell lung, and pancreatic Nab-paclitaxel 105 152 cancer Breast, prostate, non-small cell lung, Docetaxel 143 461 stomach, and head and neck cancers

Cabazitaxel prostate cancer 11 36

1Numbers represent clinical trials that are open and/or currently accepting patients. Clinical trial information was obtained from the National Institutes of Health (www.clinicaltrials.gov)

1.1.3 A brief history of paclitaxel

A cytotoxic extract from the Pacific yew, Taxus brevifolia was discovered in 1964 and years later the active ingredient was identified as Taxol® (paclitaxel) (23, 24). Early studies showed that paclitaxel blocked cell division in the G2 and/or M phase of mitosis by promoting polymerization unlike other mitotic inhibitors (at the time) which

3 promoted microtubule disassembly (25-27). By the late 1980s, clinical trials were underway and paclitaxel was approved for hormone-refractory ovarian cancer in 1992 and breast cancer in 1994 (28).

Successful clinical use of paclitaxel hinged on a steady drug supply. Paclitaxel is a highly complex diterpenoid with 11 chiral centers and 5 (Figure 1.1). Because of this complexity there was no simple synthetic route available to produce commercial quantities of the drug. Instead, paclitaxel was isolated from its natural source, the bark of the slow-growing Pacific yew tree. Isolating commercial quantities of paclitaxel from the

Pacific yew proved unsustainable and fraught with environmental concerns including habitat destruction of the threatened northern spotted owl (23). Eventually, a total organic synthesis for paclitaxel was developed, requiring approximately 40 reaction steps (29-35).

Because of the large number of steps required, the was not cost-effective and a semi-synthetic method was sought (23, 36). The clinical development of paclitaxel was aided by a Cooperative Research and Development Agreement (CRADA) between the

National Institutes of Health (NIH) and Bristol Myers Squibb (BMS). In return for licensing rights, BMS supported the production of paclitaxel from yew, the development of a semi-synthetic commercial production method, developed paclitaxel formulation for clinical use, and organized large-scale clinical trials (23). The use and importance of paclitaxel are expected to increase as cancer treatments are more available worldwide (37).

4

Figure 1.2. Paclitaxel mode of action. Structure of α/β tubulin, microtubule, and paclitaxel-bound microtubule. Reprinted with permission from Macmillan Publishers Ltd: Nat. Rev. Drug Disc. 9, 790-803, 2010.

1.1.4 Paclitaxel mode of action

Paclitaxel is active against a broad range of cancers. The accepted pharmacological mode of action suggests paclitaxel stabilizes causing arrest and subsequent cell death (26, 27, 38). Microtubules are composed of polymeric chains of α and β tubulin subunits, which polymerize head to tail as α/β heterodimers (Figure 1.2) (39,

40). As cells grow and divide, microtubules dynamically polymerize and depolymerize, facilitating normal chromosomal segregation and intracellular transport (40). This dynamic instability of microtubules requires GTP- to function (39-41). Structural studies with tubulin and paclitaxel suggest that paclitaxel binding to β-tubulin allosterically induces a “GTP-bound” state, preventing conformational changes in the tubulin and thus depolymerization (41). Although prior studies report paclitaxel arrests cancer cell cycle leading to cell death, more recent experiments demonstrated that paclitaxel activity in

5 tumors generated multiple mitotic spindles, resulting in polar body formation, impaired cytokinesis, and ultimately apoptosis (42, 43).

While paclitaxel ultimately causes cell death in many cancer cell types, paclitaxel resistance is well-documented and can be an acquired or innate characteristic of a particular cancer type (44-46). Mechanisms of drug resistance include P-glycoprotein efflux pump expression (47), altered expression of regulatory proteins (45, 48), as well as microtubule specific effects with mutated β-tubulin, varied levels of β-tubulin isotypes, and chemical modification of tubulin (reviewed in (44)). In an effort to overcome paclitaxel resistance, numerous analogs were synthesized based upon structure and function studies (49-53).

Modifications to the paclitaxel structure alter its reactivity and drug resistance in cancer cells. For example, the analog docetaxel (54) is more toxic than paclitaxel and is approved for use in additional cancer types such as prostate cancer. Yet another paclitaxel analog, cabazitaxel, was approved as a second-line treatment for hormone-refractory prostate cancer in patients already treated with docetaxel (22).

6

Figure 1.3. Taxoid derivatives of taxa-4(5),11(12)-diene. Taxadiene synthase (TS) cyclizes GGDP to make taxa-4(5),11(12)-diene, which is a precursor for many different enzymes including those on the paclitaxel pathway. Several downstream biosynthetic products are shown to illustrate the diversity and complexity of the taxoids.

1.1.5 Paclitaxel biosynthesis

Interest in paclitaxel biosynthesis is largely driven by efforts to optimize plant cell fermentation (PCF) technologies to produce biosynthetic paclitaxel (21, 55-58). While the downstream, second half of the pathway is well-characterized, the first half is only partially characterized (59, 60). The pathway is highly complex with hundreds of taxoids derived from the precursor taxa-4(5),11(12)-diene (Figure 1.3) (59, 61, 62). Many of the enzymes involved in paclitaxel biosynthesis are identified, but reaction order and substrate specificities of individual enzymes remain relatively unknown, particularly in early

7

Figure 1.4. Paclitaxel biosynthesis: Part 1. Enzyme abbreviations are as follows. IPPI, isopentenyl diphosphate , GGDPS, geranylgeranyl diphosphate synthase, TS, taxadiene synthase, T5αOH, taxadiene-5-α-ol hydroxylase, T13αOH, -13α-hydroxylase, TAT, taxadiene-5α-ol-O-acetyltransferase T10βOH, taxane-10β-hydroxylase. Enzymes not shown: T1βOH, taxane-1β-hydroxylase, T2αOH, taxane- 2α-hydroxylase, T7βOH, taxane-7β-hydroxylase, T9αOH, taxane-9α-hydroxylase, TBT, taxane-2α-O- benzoyltransferase, epoxidase*, oxomutase*, C9-oxidase*, Enzymes marked with (*) are unidentified. Compound abbreviations: isopentenyl diphosphate (IDP), dimethylallyl diphosphate (DMADP), geranylgeranyl diphosphate (GGDP), 10-deacetylbaccatin III (10-DAB).

pathway enzymes (63-65). Here, a general outline of the paclitaxel biosynthetic pathway is presented.

The primary metabolite precursors, isopentenyl diphosphate (IDP) and dimethylallyl diphosphate (DMADP), are derived from the methylerythritol pathway in the as opposed to the mevalonate pathway in the (66). These isoprenoids are sequentially coupled to produce geranylgeranyl diphosphate (GGDP), the committed

8 diterpenoid precursor (Figure 1.4). The GGDP precursor is cyclized by taxadiene synthase to produce taxa-4(5),11(12)-diene, the committed precursor for taxoid biosynthesis (67,

68). Paclitaxel is produced from taxa-4(5),11(12)-diene by a series of ~19-enzymatic steps

(Figure 1.4 and Figure 1.5) (60, 69). The exact order of these enzymatic reactions is unclear

(Figure 1.4). Numerous taxanes have been identified that could be intermediates in the first half of the pathway (prior to the formation of 10-deacetylbaccatin III (10-DAB)), demonstrating the difficulty in defining a linear pathway to paclitaxel (59, 60, 62). Taxa-

4(5),11(12)-diene is modified by a series of at least 7 cytochrome P450-dependent hydroxylases, a C4-C20-epoxidase, an oxomutase, a C9 oxidase, and 5 acyltransferases

(Figure 1.4, Figure 1.5) (64, 70-81). The taxadiene-5-α-ol hydroxylase (T5αOH) hydroxylates taxa-4(5),11(12)-diene precursor at the C5 position, and also migrates the double bond to form taxadiene-5α-ol (Figure 1.4) (77, 78). Despite great progress identifying taxane structures and screening cDNA libraries from induced Taxus cell cultures, the order of enzyme reactions, and which side-products are produced during the first half of the pathway remain unresolved.

In contrast with the first half of the pathway, the biosynthesis of paclitaxel from 10-

DAB is well-described (Figure 1.5). The 10-deacetylbaccatin III 10-O-acetyltransferase

(DBAT) acetylates 10-DAB to form baccatin III (63, 82, 83). The baccatin III: 3-amino-

13-O-phenylpropanoyl CoA transferase (BAPT) acylates baccatin III to add the (3R)-β- phenylalanyl sidechain (71). The β-phenylalanyl CoA ligase responsible for producing the activated sidechain intermediate was recently identified, but its substrate specificity was not characterized in depth (58). Although the gene for the proposed C2′-hydroxylase is unidentified, the C2′ position on the sidechain is most likely hydroxylated after the

9

Figure 1.5. Paclitaxel biosynthesis: Part 2. Enzyme abbreviations are as follows: DBAT, 10- deacetylbaccatin III 10-O-acetyltransferase, BAPT, baccatin III: 3-amino-13-O-phenylpropanoyl CoA transferase, PAM, aminomutase, β-PhL, β-phenylalanine CoA ligase, NDTNBT, N- debenzoyltaxol-N-benzoyltransferase. Enzymes marked with (*) are unidentified. Compound abbreviations: 10-deacetylbaccatin III, (10-DAB), N-debenzoyl-2′-deoxytaxol (NDB2T).

sidechain is attached to baccatin III (80). The last enzyme on the pathway, N- debenzoyltaxol-N-benzoyltransferase (NDTNBT) benzoylates the 3′N on N- debenzoylpaclitaxel to form paclitaxel (64, 73).

As mentioned, there are hundreds of taxanes in plant cells derived from taxa-

4(5),11(12)-diene, many of which are either enzyme byproducts or a result of pathway branches (21, 60, 84) (Figure 1.6). Efforts to increase flux toward paclitaxel production in plant cells have centered on profiling gene expression in methyl jasmonate-induced cell cultures to identify genes involved in the regulation of paclitaxel biosynthesis (84-86).

10

Figure 1.6. Examples of off-pathway metabolites that divert flux away from paclitaxel production.

Genetic transformation techniques with Taxus cells are in development and not yet routine, limiting the engineering capabilities of PCF (87-89). Using these techniques, several repressive factors of paclitaxel biosynthetic genes were recently identified and may prove useful as targets to increase paclitaxel production in PCF (90).

1.1.6 Paclitaxel production

Paclitaxel purification from the bark of the Pacific yew is low yielding (<0.05 %), environmentally unsustainable, and economically unfeasible (37). Even so, paclitaxel was

11 harvested from the yew tree for early studies and clinical trials (23). Growing clinical success of paclitaxel prompted enormous efforts into developing synthetic production methods. However, due to its structural complexity, the complete organic synthesis of paclitaxel is also untenable with more than 40 synthetic steps (29-34). The discovery of high levels of the precursors 10-DAB and baccatin III in the needles of Taxus made the development of semi-synthetic methods possible as yew needles are a more renewable resource than yew bark (23, 91, 92). The semi-synthetic production of paclitaxel from 10-

DAB was developed by the Holton research group in collaboration with BMS (36, 93-95)

(Figure 1.7). This method substantially improved paclitaxel production compared with isolation from the natural source (36). However, multiple protection and deprotection steps were required, making this method costly in the long-term. The development of suspended

Taxus plant cell fermentation technology to produce biosynthetic paclitaxel is replacing semi-synthesis for commercial production (37, 96). Phyton Biotech developed this technology and currently supplies a majority of paclitaxel worldwide (57, 97). Taxus cells biosynthesize paclitaxel in response to a number of elicitors both organic and inorganic, the strongest and most well-known being methyl jasmonate (98-100). In the presence of methyl jasmonate, T. baccata PCF produces paclitaxel and baccatin III at levels of ~50 mg/L compared with ~0.4 mg/L in unelicited cells (100). However, yields produced from

Taxus cell cultures are highly variable depending on the Taxus sp., the elicitors present, and growth conditions. In addition to paclitaxel production, PCF produces other taxanes as byproducts. Large quantities of 10-DAB and baccatin III are also produced by PCF and used as semi-synthetic precursors for the production of clinically-approved docetaxel and

12

Figure 1.7. Semi-synthesis of paclitaxel from 10-DAB. (i) TESCl, , DMF, r.t., 96% yield; (ii) LiHMDS, CH3COCl, THF, 82% yield, (iii) LDA, THF, (iv) PMP aldimine, 96% yield for two steps, (v) CAN, CH3CN, H2O, 88%, (vi) BzCOCl, TEA, DMAP, CH2Cl2, r.t., 97%, (vii) LiHMDS, THF, -40 °C, 30 min, (viii) HF/pyridine, pyridine/MeCN, 0 °C to r.t., 18 h, 80% for two steps. The overall yield of paclitaxel is approximately ~50 % without considering the synthesis of the C13 sidechain components. This synthesis is adapted from (95, 101).

13

Figure 1.8. Taxanes clinically approved and in clinical trials. A. The paclitaxel. B. Docetaxel. C. Cabazitaxel. D. DJ-927 (Tesetaxel) completed phase II clinical trials for several cancers. E. BMS-275183 (Phase I – terminated). F. TL-00139 (Milataxel) completed phase II clinical trials for malignant mesothelioma. G. MST-997 (Phase I- terminated). H. TPI-287 ongoing clinical trials for several cancers and Alzheimer’s disease. Brand names are listed in parentheses (clinicaltrials.gov).

cabazitaxel, as well as other analogs in clinical trials including TPI-287 for Alzheimer’s disease (Figure 4.4) (51).

1.1.7 Semi-biosynthesis of paclitaxel, precursors, and analogs

Although PCF methodologies were adapted to produce paclitaxel in commercial quantities, docetaxel, cabazitaxel and other analogs must be semi-synthesized from 10-DAB or baccatin III, requiring multiple protection/deprotection steps, and large quantities of and organic solvents (51, 102-109). Paclitaxel production transitioned to PCF because it was more cost-effective and environmentally friendly. Lack of a biosynthetic

14

Figure 1.9. Coupled enzyme biosynthesis of paclitaxel. I. (2R,3S)-phenylisoserinyl CoA production by PheAT. II. Sidechain attachment to baccatin III by BAPT to produce the docetaxel precursor (R1 = H) (III). IV. BadA catalyzes the production of benzoyl CoA. V. NDTNBT transfers a benzoyl group to the 3′-N of N- debenzoylpaclitaxel to produce paclitaxel. Baccatin III: R1 = . 10-Deacetylbaccatin III (10- DAB): R1 = H.

pathway for certain analogs makes in vivo biosynthetic engineering impossible, but development of a heterologous pathway in could make precursor biosynthesis possible by exploiting the broad substrate specificity of biosynthetic enzymes and feeding precursors into the growth media.

15 Recently, bioengineering efforts with the paclitaxel pathway have increased in heterologous such as E. coli, S. cerevisiae, and A. thaliana (110-120). Thus far, progress is limited to the earliest pathway steps due to incomplete characterization of the pathway (111, 119). The eukaryotic, membrane-bound enzymes, such as the cytochrome

P450 hydroxylases, must be expressed in yeast or engineered for expression in bacteria

(81, 111, 121). Byproducts produced by pathway enzymes with uncharacterized substrate specificities (78, 122), and unidentified pathway bottlenecks (116), complicate bioengineering attempts. The complexity of the non-linear paclitaxel pathway and its incomplete characterization suggest bioengineering paclitaxel from primary and minimizing side-product formation will be extremely difficult (60, 96).

Given these concerns and the availability of the 10-DAB and baccatin III precursors from PCF, we proposed the construction of a simple, biocatalytic pathway that bypasses many of the problems associated with engineering the entire paclitaxel pathway. The approach is versatile, ultimately enabling the production of paclitaxel, analogs, and precursors of interest in E. coli. The cytoplasmic BAPT and NDTNBT Taxus acyltransferases, and two corresponding cognate, bacterial acyl CoA ligases, PheAT and

BadA, were combined in a 4-enzyme biosynthetic reaction (Figure 1.9) (123, 124). The use of all 4 enzymes with the starting material baccatin III produces paclitaxel. By omitting the NDTNBT acyltransferase and the BadA ligase, N-debenzoylpaclitaxel precursors are made. Analogs of pharmaceutical interest are subsequently made by N-.

Substituting baccatin III for 10-DAB leads to the production of docetaxel analogs or the docetaxel precursor, N-debenzoyl-10-deacetylpaclitaxel. Benefits of using the late stage paclitaxel biosynthetic enzymes, BAPT and NDTNBT, include bypassing uncharacterized

16 biosynthetic pathway steps and an opportunity to exploit the acyl CoA substrate permissivity and regioselectivity of these enzymes to make variously acylated analogs of interest.

In order to develop this pathway and demonstrate proof-of-principle for paclitaxel and analog biosynthesis, the activities of BadA, a benzoate CoA ligase, and BAPT, a baccatin III: β-phenylpropanoyl CoA transferase, were studied (Figure 1.9). The substrate specificity of BadA was systematically probed via Michaelis-Menten kinetic characterization with an array of substrates, and activity was also interpreted in the context of six crystal structures (125). Targeted mutation of the BadA active site and activity analysis led to new insights regarding enzyme mechanism and expanded substrate specificity compared with wild-type BadA (Chapter 2).

The characterization of BAPT activity in vitro first required protein expression optimization and purification. Initial discovery and characterization of BAPT enzyme activity was limited to estimations of KM with (3R)-phenylalanyl CoA and (2R,3S)- phenylisoserinyl CoA, performed with assays containing impure BAPT lysate (71). In this work, BAPT was purified substantially compared with initial experiments, overcoming solubility hurdles (Chapter 3). Determination of the Michaelis-Menten kinetic parameters of BAPT required the synthetic and biosynthetic production of commercially unavailable acyl CoA substrates including (3R)-phenylalanyl CoA and (2R,3S)-phenylisoserinyl CoA.

BAPT was combined with PheAT to demonstrate the biosynthesis of paclitaxel or docetaxel precursors from (2R,3S)-phenylisoserine and baccatin III or 10-DAB, respectively (Chapter 4).

17 Ultimately, BAPT was combined with PheAT, NDTBT, and BadA to produce N-

2-furanyl-N-debenzoylpaclitaxel, a biologically active analog of paclitaxel (Chapter 4).

This demonstrates the potential for these enzymes to be combined and developed in bacteria to produce paclitaxel/docetaxel/cabazitaxel precursors and other analogs of clinical interest. Semi-biosynthesis of docetaxel (or precursors) by this method has the ability to shorten the number of synthetic steps, potentially increasing production and lowering the cost to patients. By exploiting the substrate specificities of enzymes in this pathway the number of steps required to produce a given paclitaxel analog may be reduced.

The combination of bacterial CoA ligases and plant acyltransferases to produce acylated natural products may be extended to additional analogs by pairing an acyltransferase with acyl CoA ligases with different substrate specificities.

Organic synthesis of biologically relevant, structurally complex natural products is often difficult and costly, requiring large volumes of environmentally harmful organic solvents. Development of semi-biosynthetic enzymatic processes can bypass synthetic hurdles and potentially make production of novel compounds possible. Combinatory biosynthetic enzymes from different organisms can aid in this process.

18

REFERENCES

19 REFERENCES

1. National Cancer Institute: Common Cancer Types, (2015), Bethesda, MD: National Cancer Institute; [accessed 2015, Oct 02]. http://www.cancer.gov/types/common- cancers.

2. Cancer Facts & Figures 2015, (2015), Atlanta, GA: American Cancer Society; [accessed 2015, Oct 02]. http://www.cancer.org/research/cancerfactsstatistics/ cancerfactsfigures2015/.

3. Von Hoff, D. D., Ervin, T., Arena, F. P., Chiorean, E. G., Infante, J., Moore, M., Seay, T., Tjulandin, S. A., Ma, W. W., Saleh, M. N., Harris, M., Reni, M., Dowden, S., Laheru, D., Bahary, N., Ramanathan, R. K., Tabernero, J., Hidalgo, M., Goldstein, D., Van Cutsem, E., Wei, X. Y., Iglesias, J., and Renschler, M. F. (2013) Increased survival in pancreatic cancer with nab-paclitaxel plus gemcitabine, New Engl. J. Med. 369, 1691-1703.

4. Von Hoff, D. D., Ramanathan, R. K., Borad, M. J., Laheru, D. A., Smith, L. S., Wood, T. E., Korn, R. L., Desai, N., Trieu, V., Iglesias, J. L., Zhang, H., Soon- Shiong, P., Shi, T., Rajeshkumar, N. V., Maitra, A., and Hidalgo, M. (2011) Gemcitabine plus nab-paclitaxel Is an active regimen in patients with advanced pancreatic cancer: A Phase I/II trial, J. Clin. Oncol. 29, 4548-4554.

5. Yardley, D. A. (2013) nab-Paclitaxel mechanisms of action and delivery, J. Control. Release 170, 365-372.

6. Sorbe, B., Andersson, H., Boman, K., Rosenberg, P., and Kalling, M. (2008) Treatment of primary advanced and recurrent endometrial carcinoma with a combination of and paclitaxel—long‐term follow‐up, Int. J. Gynecol. Cancer 18, 803-808.

7. McGuire, W. P., Blessing, J. A., Moore, D., Lentz, S. S., and Photopulos, G. (1996) Paclitaxel has moderate activity in squamous cervix cancer: A gynecologic oncology group study, J. Clin. Oncol. 14, 792-795.

8. Lee, H. H., Ye, S., Li, X. J., Lee, K. B., Park, M. H., and Kim, S. M. (2014) Combination treatment with paclitaxel and inhibits growth of human esophageal squamous cancer cells by inactivation of Akt, Oncol. Rep. 31, 183-188.

9. Ruppert, B. N., Watkins, J. M., Shirai, K., Wahlquist, A. E., Garrett-Mayer, E., Aguero, E. G., Sherman, C. A., Reed, C. E., and Sharma, A. K. (2010) / versus carboplatin/paclitaxel as definitive chemoradiotherapy for locoregionally advanced esophageal cancer, Am. J. Clin. Oncol-Canc. 33, 346- 352.

20 10. Wang, H., Ryu, J., Gandara, D., Bold, R. J., Urayama, S., Tanaka, M., Goldberg, Z., Follette, D., Narayan, S., and Lau, D. (2007) A phase II study of paclitaxel, carboplatin, and radiation with or without surgery for esophageal cancer, J. Thorac. Oncol. 2, 153-157.

11. ClinicalTrials.gov. (2015). Bethesda, MD: U.S. National Institutes of Health [accessed 2015, Oct 6]. http://www.clinicaltrials.gov.

12. Bajaj, N. R., and Garratt, K. N. (2010) Clinical studies with paclitaxel - Eluting stent systems, Curr. Pharm. Des. 16, 4025-4036.

13. Farb, A., Heller, P. F., Shroff, S., Cheng, L., Kolodgie, F. D., Carter, A. J., Scott, D. S., Froehlich, J., and Virmani, R. (2001) Pathological analysis of local delivery of paclitaxel via a polymer-coated stent, Circulation 104, 473-479.

14. Park, S. J., Shim, W. H., Ho, D. S., Raizner, A. E., Park, S. W., Hong, M. K., Lee, C. W., Choi, D. H., Jang, Y. S., Lam, R., Weissman, N. J., and Mintz, G. S. (2003) A paclitaxel-eluting stent for the prevention of coronary restenosis, New Engl. J. Med. 348, 1537-1545.

15. Zhang, D. S., Yang, R. H., Wang, S. X., and Dong, Z. (2014) Paclitaxel: New uses for an old drug, Drug Des. Dev. Ther. 8, 279-284.

16. Wang, C., Song, X., Li, Y., Han, F., Gao, S., Wang, X., Xie, S., and Lv, C. (2013) Low-dose paclitaxel ameliorates pulmonary fibrosis by suppressing TGF- β1/Smad3 pathway via miR-140 upregulation, PLoS One 8, e70725.

17. Mirzapoiazova, T., Kolosova, I. A., Moreno, L., Sammani, S., Garcia, J. G., and Verin, A. D. (2007) Suppression of endotoxin-induced inflammation by Taxol, Eur. Resp. J. 30, 429-435.

18. Petrache, I., Birukova, A., Ramirez, S. I., Garcia, J. G., and Verin, A. D. (2003) The role of the microtubules in tumor necrosis factor-α–induced endothelial cell permeability, Am. J. Resp. Cell Mol. 28, 574-581.

19. Sengottuvel, V., Leibinger, M., Pfreimer, M., Andreadaki, A., and Fischer, D. (2011) Taxol facilitates axon regeneration in the mature CNS, J. Neurosci. 31, 2688-2699.

20. Hellal, F., Hurtado, A., Ruschel, J., Flynn, K. C., Laskowski, C. J., Umlauf, M., Kapitein, L. C., Strikis, D., Lemmon, V., and Bixby, J. (2011) Microtubule stabilization reduces scarring and causes axon regeneration after spinal cord injury, Science 331, 928-931.

21. Onrubia, M., Cusido, R. M., Ramirez, K., Hernandez-Vazquez, L., Moyano, E., Bonfill, M., and Palazon, J. (2013) Bioprocessing of plant in vitro systems for the mass production of pharmaceutically important metabolites: paclitaxel and its derivatives, Curr. Med. Chem. 20, 880-891.

21 22. de Bono, J. S., Oudard, S., Ozguroglu, M., Hansen, S., Machiels, J.-P., Kocak, I., Gravis, G., Bodrogi, I., Mackenzie, M. J., and Shen, L. (2010) Prednisone plus cabazitaxel or for metastatic castration-resistant prostate cancer progressing after docetaxel treatment: a randomised open-label trial, Lancet 376, 1147-1154.

23. Goodman, J., and Walsh, V. (2001) The story of Taxol: Nature and politics in the pursuit of an anti-cancer drug, Cambridge University Press, Cambridge, UK.

24. Wani, M. C., Taylor, H. L., Wall, M. E., Coggon, P., and McPhail, A. T. (1971) Plant antitumor agents. VI. The isolation and structure of Taxol, a novel antileukemic and antitumor agent from Taxus brevifolia, J. Am. Chem. Soc. 93, 2325-2327.

25. Jordan, M. A., and Wilson, L. (2004) Microtubules as a target for anticancer drugs, Nat. Rev. Cancer 4, 253-265.

26. Schiff, P. B., Fant, J., and Horwitz, S. B. (1979) Promotion of microtubule assembly in vitro by Taxol, Nature 277, 665-667.

27. Schiff, P. B., and Horwitz, S. B. (1980) Taxol stabilizes microtubules in mouse fibroblast cells, P. Natl. Acad. Sci. U.S.A. 77, 1561-1565.

28. Holmes, F. A., Walters, R. S., Theriault, R. L., Forman, A. D., Newton, L. K., Raber, M. N., Buzdar, A. U., Frye, D. K., and Hortobagyi, G. N. (1991) Phase II trial of Taxol, an active drug in the treatment of metastatic breast cancer, J. Natl. Cancer Inst. 83, 1797-1805.

29. Holton, R. A., Kim, H. B., Somoza, C., Liang, F., Biediger, R. J., Boatman, P. D., Shindo, M., Smith, C. C., and Kim, S. (1994) First total synthesis of Taxol. 2. Completion of the C and D rings, J. Am. Chem. Soc. 116, 1599-1600.

30. Holton, R. A., Somoza, C., Kim, H. B., Liang, F., Biediger, R. J., Boatman, P. D., Shindo, M., Smith, C. C., and Kim, S. (1994) First total synthesis of Taxol. 1. Functionalization of the B ring, J. Am. Chem. Soc. 116, 1597-1598.

31. Nicolaou, K. C., Liu, J. J., Yang, Z., Ueno, H., Sorensen, E. J., Claiborne, C. F., Guy, R. K., Hwang, C. K., Nakada, M., and Nantermet, P. G. (1995) Total synthesis of Taxol 2. Construction of A-ring and C-ring intermediates and initial attempts to construct the ABC ring-system, J. Am. Chem. Soc. 117, 634-644.

32. Nicolaou, K. C., Nantermet, P. G., Ueno, H., Guy, R. K., Couladouros, E. A., and Sorensen, E. J. (1995) Total synthesis of Taxol 1. Retrosynthesis, degradation and reconstitution, J. Am. Chem. Soc. 117, 624-633.

33. Nicolaou, K. C., Ueno, H., Liu, J. J., Nantermet, P. G., Yang, Z., Renaud, J., Paulvannan, K., and Chadha, R. (1995) Total synthesis of Taxol 4. The final stages and completion of the synthesis, J. Am. Chem. Soc. 117, 653-659.

22 34. Nicolaou, K. C., Yang, Z., Liu, J. J., Nantermet, P. G., Claiborne, C. F., Renaud, J., Guy, R. K., and Shibayama, K. (1995) Total synthesis of Taxol 3. Formation of Taxol's ABC ring skeleton, J. Am. Chem. Soc. 117, 645-652.

35. Nicolaou, K. C., Yang, Z., Liu, J. J., Ueno, H., Nantermet, P. G., Guy, R. K., Claiborne, C. F., Renaud, J., Couladouros, E. A., Paulvannan, K., and et al. (1994) Total synthesis of Taxol, Nature 367, 630-634.

36. Holton, R. A., Biediger, R. J., Boatman, P. D. (1995) Taxol: Science and Applications. , CRC Press, Boca Raton, FL.

37. Malik, S., Cusido, R. M., Mirjalili, M. H., Moyano, E., Palazon, J., and Bonfill, M. (2011) Production of the anticancer drug Taxol in Taxus baccata suspension cultures: A review, Process Biochem. 46, 23-34.

38. Parness, J., and Horwitz, S. B. (1981) Taxol binds to polymerized tubulin J. Cell Biol. 91, 479-487.

39. Nogales, E. (2000) Structural insights into microtubule function, Annu. Rev. Biochem. 69, 277-302.

40. Subramanian, R., and Kapoor, Tarun M. (2012) Building complexity: Insights into self-organized assembly of mcrotubule-based architectures, Dev. Cell 23, 874-885.

41. Alushin, Gregory M., Lander, Gabriel C., Kellogg, Elizabeth H., Zhang, R., Baker, D., and Nogales, E. (2014) High-resolution microtubule structures reveal the structural transitions in αβ-tubulin upon GTP hydrolysis, Cell 157, 1117-1129.

42. Weaver, B. A. (2014) How Taxol/paclitaxel kills cancer cells, Mol. Biol. Cell 25, 2677-2681.

43. Zasadil, L. M., Andersen, K. A., Yeum, D., Rocque, G. B., Wilke, L. G., Tevaarwerk, A. J., Raines, R. T., Burkard, M. E., and Weaver, B. A. (2014) Cytotoxicity of paclitaxel in breast cancer is due to chromosome missegregation on multipolar spindles, Sci. Transl. Med. 6, 229ra243.

44. Orr, G. A., Verdier-Pinard, P., McDaid, H., and Horwitz, S. B. (2003) Mechanisms of Taxol resistance related to microtubules, Oncogene 22, 7280-7295.

45. Richardson, A., and Kaye, S. B. (2005) Drug resistance in ovarian cancer: The emerging importance of gene transcription and spatio-temporal regulation of resistance, Drug Resist. Update 8, 311-321.

46. Mozzetti, S., Ferlini, C., Concolino, P., Filippetti, F., Raspaglio, G., Prislei, S., Gallo, D., Martinelli, E., Ranelletti, F. O., and Ferrandina, G. (2005) Class III β- tubulin overexpression is a prominent mechanism of paclitaxel resistance in ovarian cancer patients, Clin. Cancer Res. 11, 298-305.

23 47. Ramachandra, M., Ambudkar, S. V., Chen, D., Hrycyna, C. A., Dey, S., Gottesman, M. M., and Pastan, I. (1998) Human P-glycoprotein exhibits reduced affinity for substrates during a catalytic transition state, Biochemistry-US 37, 5010-5019.

48. Lee, D.-k., Kim, Y. H., Kim, J.-S., and Seol, W. (2004) Induction and characterization of Taxol-resistance phenotypes with a transiently expressed artificial transcriptional activator library, Nucleic Acids Res. 32, e116-e116.

49. Ganesh, T., Guza, R. C., Bane, S., Ravindra, R., Shanker, N., Lakdawala, A. S., Snyder, J. P., and Kingston, D. G. (2004) The bioactive Taxol conformation on - tubulin: experimental evidence from highly active constrained analogs, P. Natl. Acad. Sci. U.S.A. 101, 10006-10011.

50. Geney, R., Sun, L., Pera, P., Bernacki, R. J., Xia, S., Horwitz, S. B., Simmerling, C. L., and Ojima, I. (2005) Use of the tubulin bound paclitaxel conformation for structure-based rational drug design, Chem. Biol. 12, 339-348.

51. Dubois, J. (2006) Recent progress in the development of docetaxel and paclitaxel analogues. Exp. Opin. Ther. Pat. 16, 1481-1496.

52. Fang, W.-S., and Wang, S.-R. (2008) Structural studies of Taxol analogues for . Exp. Opin. Drug Disc. 3, 1109-1122.

53. Yared, J. A., and Tkaczuk, K. H. (2012) Update on taxane development: New analogs and new formulations, Drug Des. Dev. Ther. 6, 371-384.

54. Bissery, M. C., Guenard, D., Guerittevoegelein, F., and Lavelle, F. (1991) Experimental antitumor-activity of Taxotere (RP-56976, NSC-628503), a Taxol analog, Cancer Res. 51, 4845-4852.

55. Yu, W. B., Liang, X., and Zhu, P. (2013) High-cell-density fermentation and pilot- scale biocatalytic studies of an engineered yeast expressing the heterologous glycoside of 7--xylosyltaxanes, J. Ind. Microbiol. Biotechnol. 40, 133- 140.

56. Hampel, D., Mau, C. J. D., and Croteau, R. B. (2009) Taxol biosynthesis: Identification and characterization of two acetyl CoA: taxoid-O-acetyl that divert pathway flux away from Taxol production, Arch. Biochem. Biophys. 487, 91-97.

57. Cusido, R. M., Onrubia, M., Sabater-Jara, A. B., Moyano, E., Bonfill, M., Goossens, A., Pedreno, M. A., and Palazon, J. (2014) A rational approach to improving the biotechnological production of taxanes in plant cell cultures of Taxus spp, Biotechnol. Adv. 32, 1157-1167.

58. Ramirez-Estrada, K., Altabella, T., Onrubia, M., Moyano, E., Notredame, C., Osuna, L., Vanden Bossche, R., Goossens, A., Cusido, R. M., and Palazon, J.

24 (2015) Transcript profiling of jasmonate-elicited Taxus cells reveals a - phenylalanine-CoA ligase, Plant Biotechnol. J.

59. Guerra-Bubb, J., Croteau, R., and Williams, R. M. (2012) The early stages of Taxol biosynthesis: An interim report on the synthesis and identification of early pathway metabolites, Nat. Prod. Rep. 29, 683-696.

60. Ketchum, R. E. B., and Croteau, R. B. (2006) The Taxus metabolome and the elucidation of the Taxol® biosynthetic pathway in cell suspension cultures, In Plant Metabolomics (Saito, K., Dixon, R., and Willmitzer, L., Eds.), pp 291-309, Springer Berlin Heidelberg.

61. Kingston, D. G. (2000) Recent advances in the chemistry of Taxol, J. Nat. Prod. 63, 726-734.

62. Baloglu, E., and Kingston, D. G. I. (1999) The taxane diterpenoids, J. Nat. Prod. 62, 1448-1472.

63. Loncaric, C., Merriweather, E., and Walker, K. D. (2006) Profiling a Taxol pathway 10--acetyltransferase: Assessment of the specificity and the production of baccatin III by in vivo acetylation in E. coli, Chem. Biol. 13, 309-317.

64. Long, R. M., Lagisetti, C., Coates, R. M., and Croteau, R. B. (2008) Specificity of the N-benzoyl transferase responsible for the last step of Taxol biosynthesis, Arch. Biochem. Biophys. 477, 384-389.

65. Nevarez, D. M., Mengistu, Y. A., Nawarathne, I. N., and Walker, K. D. (2009) An N-aroyltransferase of the BAHD superfamily has broad aroyl CoA specificity in vitro with analogues of N-dearoylpaclitaxel, J. Am. Chem. Soc. 131, 5994-6002.

66. Eisenreich, W., Menhard, B., Hylands, P. J., Zenk, M. H., and Bacher, A. (1996) Studies on the biosynthesis of taxol: The taxane carbon skeleton is not of mevalonoid origin, P. Natl. Acad. Sci. U.S.A. 93, 6431-6436.

67. Hezari, M., Lewis, N. G., and Croteau, R. (1995) Purification and characterization of taxa-4(5),11(12)-diene synthase from pacific yew (Taxus brevifolia) that catalyzes the first committed step of Taxol biosynthesis, Arch. Biochem. Biophys. 322, 437-444.

68. Koksal, M., Jin, Y., Coates, R. M., Croteau, R., and Christianson, D. W. (2011) Taxadiene synthase structure and evolution of modular architecture in terpene biosynthesis, Nature 469, 116-120.

69. Heinig, U., and Jennewein, S. (2009) Taxol: A complex diterpenoid natural product with an evolutionarily obscure origin, Afr. J. Biotechnol. 8, 1370-1385.

25 70. Walker, K., and Croteau, R. (2000) Taxol biosynthesis: Molecular cloning of a benzoyl-CoA : taxane 2--O-benzoyltransferase cDNA from Taxus and functional expression in Escherichia coli, P. Natl. Acad. Sci. U.S.A. 97, 13591-13596.

71. Walker, K., Fujisaki, S., Long, R., and Croteau, R. (2002) Molecular cloning and heterologous expression of the C-13 phenylpropanoid side chain-CoA acyltransferase that functions in Taxol biosynthesis, P. Natl. Acad. Sci. U.S.A. 99, 12715-12720.

72. Walker, K., Ketchum, R. E. B., Hezari, M., Gatfield, D., Goleniowski, M., Barthol, A., and Croteau, R. (1999) Partial purification and characterization of acetyl coenzyme A: Taxa-4(20),11(12)-dien-5--ol O-acetyl transferase that catalyzes the first acylation step of Taxol biosynthesis, Arch. Biochem. Biophys. 364, 273-279.

73. Walker, K., Long, R., and Croteau, R. (2002) The final acylation step in Taxol biosynthesis: Cloning of the taxoid C13-side-chain N-benzoyltransferase from Taxus, P. Natl. Acad. Sci. U.S.A. 99, 9166-9171.

74. Walker, K. D., Klettke, K., Akiyama, T., and Croteau, R. (2004) Cloning, heterologous expression, and characterization of a phenylalanine aminomutase involved in Taxol biosynthesis, J. Biol. Chem. 279, 53947-53954.

75. Chau, M., and Croteau, R. (2004) Molecular cloning and characterization of a cytochrome P450 taxoid 2--hydroxylase involved in Taxol biosynthesis, Arch. Biochem. Biophys. 427, 48-57.

76. Chau, M., Jennewein, S., Walker, K., and Croteau, R. (2004) Taxol biosynthesis: Molecular cloning of a cytochrome P450 characterization taxoid 7--hydroxylase, Chem. Biol. 11, 663-672.

77. Hefner, J., Rubenstein, S. M., Ketchum, R. E. B., Gibson, D. M., Williams, R. M., and Croteau, R. (1996) Cytochrome P450-catalyzed hydroxylation of taxa- 4(5),11(12)-diene to taxa-4(20),11(12)-dien-5-ol: The first oxygenation step in Taxol biosynthesis, Chem. Biol. 3, 479-489.

78. Jennewein, S., Long, R. M., Williams, R. M., and Croteau, R. (2004) Cytochrome P450 taxadiene 5-hydroxylase, a mechanistically unusual monooxygenase catalyzing the first oxygenation step of Taxol biosynthesis, Chem. Biol. 11, 379- 387.

79. Jennewein, S., Rithner, C. D., Williams, R. M., and Croteau, R. B. (2001) Taxol biosynthesis: taxane 13α-hydroxylase is a cytochrome P450-dependent monooxygenase, P. Natl. Acad. Sci. U.S.A. 98, 13595-13600.

80. Long, R. M., and Croteau, R. (2005) Preliminary assessment of the C13-side chain 2 '-hydroxylase involved in Taxol biosynthesis, Biochem. Bioph. Res. Co. 338, 410- 417.

26 81. Schoendorf, A., Rithner, C. D., Williams, R. M., and Croteau, R. B. (2001) Molecular cloning of a cytochrome P450 taxane 10-hydroxylase cDNA from Taxus and functional expression in yeast, P. Natl. Acad. Sci. U.S.A. 98, 1501-1506.

82. Ondari, M. E., and Walker, K. D. (2008) The Taxol pathway 10-O-acetyltransferase shows regioselective promiscuity with the oxetane hydroxyl of 4-deacetyltaxanes, J. Am. Chem. Soc. 130, 17187-17194.

83. Fang, J., and Ewald, D. (2004) Expression cloned cDNA for 10-deacetylbaccatin III-10-O-acetyltransferase in Escherichia coli: A comparative study of three fusion systems, Protein Expres. Purif. 35, 17-24.

84. Song, G. H., Zhao, C. F., Zhang, M., Fu, C. H., Zhang, H., and Yu, L. J. (2014) Correlation analysis of the taxane core modification, enzyme expression, and metabolite accumulation profiles under methyl jasmonate treatment, Biotechnol. Progr. 30, 269-280.

85. Nims, E., Dubois, C. P., Roberts, S. C., and Walker, E. L. (2006) Expression profiling of genes involved in paclitaxel biosynthesis for targeted metabolic engineering, Metab. Eng. 8, 385-394.

86. Lenka, S. K., Boutaoui, N., Paulose, B., Vongpaseuth, K., Normanly, J., Roberts, S. C., and Walker, E. L. (2012) Identification and expression analysis of methyl jasmonate responsive ESTs in paclitaxel producing Taxus cuspidata suspension culture cells, BMC Genomics 13, 148.

87. Ketchum, R. E. B., Wherland, L., and Croteau, R. B. (2007) Stable transformation and long-term maintenance of transgenic Taxus cell suspension cultures, Plant Cell Rep. 26, 1025-1033.

88. Li, S. T., Fu, C. H., Zhang, M., Zhang, Y., Xie, S., and Yu, L. J. (2012) Enhancing Taxol biosynthesis by overexpressing a 9-cis-epoxycarotenoid dioxygenase gene in transgenic cell lines of Taxus chinensis, Plant Mol. Biol. Rep. 30, 1125-1130.

89. Vongpaseuth, K., Nims, E., Amand, M. S., Walker, E. L., and Roberts, S. C. (2007) Development of a particle bombardment-mediated transient transformation system for Taxus spp. cells in culture, Biotechnol. Progr. 23, 1180-1185.

90. Lenka, S. K., Nims, N. E., Vongpaseuth, K., Boshar, R. A., Roberts, S. C., and Walker, E. L. (2015) Jasmonate-responsive expression of paclitaxel biosynthesis genes in Taxus cuspidata cultured cells is negatively regulated by the bHLH transcription factors TcJAMYC1, TcJAMYC2, and TcJAMYC4, Front. Plant Sci. 6.

91. Appendino, G. (1995) The of the yew tree, Nat. Prod. Rep. 12, 349- 360.

27 92. Denis, J. N., Greene, A. E., Guenard, D., Guerittevoegelein, F., Mangatal, L., and Potier, P. (1988) A highly efficient, practical approach to natural Taxol., J. Am. Chem. Soc. 110, 5917-5919.

93. Ojima, I., Slater, J. C., Kuduk, S. D., Takeuchi, C. S., Gimi, R. H., Sun, C. M., Park, Y. H., Pera, P., Veith, J. M., and Bernacki, R. J. (1997) Syntheses and structure- activity relationships of taxoids derived from 14-hydroxy-10-deacetylbaccatin III, J. Med. Chem. 40, 267-278.

94. Ojima, I., Chen, J., Sun, L., Borella, C. P., Wang, T., Miller, M. L., Lin, S., Geng, X., Kuznetsova, L., and Qu, C. (2008) Design, synthesis, and biological evaluation of new-generation taxoids, J. Med. Chem. 51, 3203-3221.

95. Ojima, I., Fumero-Oderda, C. L., Kuduk, S. D., Ma, Z., Kirikae, F., and Kirikae, T. (2003) Structure–activity relationship study of taxoids for their ability to activate murine macrophages as well as inhibit the growth of macrophage-like cells, Bioorg. Med. Chem. 11, 2867-2888.

96. Expósito, O., Bonfill, M., Moyano, E., Onrubia, M., Mirjalili, M., Cusido, R., and Palazon, J. (2009) Biotechnological production of Taxol and related taxoids: Current state and prospects, Anticancer Agents Med. Chem. 9, 109-121.

97. Roberts, S. C., and Shuler, M. L. (1997) Large-scale plant , Curr. Opin. Biotechnol. 8, 154-159.

98. Ketchum, R. E. B., Gibson, D. M., Croteau, R. B., and Shuler, M. L. (1999) The kinetics of taxoid accumulation in cell suspension cultures of Taxus following elicitation with methyl jasmonate, Biotechnol. Bioeng. 62, 97-105.

99. Yuan, Y. J., Wei, Z. J., Miao, Z. Q., and Wu, J. C. (2002) Acting paths of elicitors on Taxol biosynthesis pathway and their synergistic effect, Biochem. Eng. J. 10, 77-83.

100. Yukimune, Y., Tabata, H., Higashi, Y., and Hara, Y. (1996) Methyl jasmonate- induced overproduction of paclitaxel and baccatin III in Taxus cell suspension cultures, Nat. Biotechnol. 14, 1129-1132.

101. Nawarathne, I. N. (2011) Modification of biologically important natural product scaffolds through biocatalysis, using Taxus acyltransferases.

102. Ali, S. M., Hoemann, M. Z., Aube, J., Mitscher, L. A., Georg, G. I., McCall, R., and Jayasinghe, L. R. (1995) Novel cytotoxic 3'-(tert-Butyl) 3'-diphenyl analogs of paclitaxel and docetaxel, J. Med. Chem. 38, 3821-3828.

103. Chang, J., Hao, X. D., Hao, Y. P., Lu, H. F., Yu, J. M., and Sun, X. (2013) Design, synthesis and cytotoxicity of novel 3'-N-alkoxycarbonyl docetaxel analogs, Bioorg. Med. Chem. Lett. 23, 6834-6837.

28 104. Commerçon, A., Bourzat, J., Didier, E., and Lavelle, F. (1995) Practical and antimitotic activity of docetaxel and side-chain analogs, Taxane Anticancer Agents 583, 233-246.

105. Georg, G. I., Harriman, G. C. B., Hepperle, M., Clowers, J. S., VanderVelde, D. G., and Himes, R. H. (1996) Synthesis, conformational analysis, and biological evaluation of heteroaromatic taxanes, J. Org. Chem. 61, 2664-2676.

106. Li, C. H., Qiu, Y. T., Li, X., Liu, N. J., and Yao, Z. Y. (2014) Biological evaluation of new antitumor taxoids: Alteration of substitution at the C-7 and C-10 of docetaxel, Bioorg. Med. Chem. Lett. 24, 855-859.

107. Ojima, I., Park, Y. H., Fenoglio, I., Duclos, O., Sun, C. M., Kuduk, S. D., Zucco, M., Appendino, G., Pera, P., Veith, J. M., Bernacki, R. J., Bissery, M. C., Combeau, C., Vrignaud, P., Riou, J. F., and Lavelle, F. (1995) Synthesis and structure-activity relationships of new taxoids, In Taxane Anticancer Agents: Basic Science and Current Status (Georg, G. I., Chem, T. T., Ojima, I., and Vyas, D. M., Eds.), pp 262-275, Amer. Chemical Soc., Washington.

108. Ojima, I., Slater, J. C., Michaud, E., Kuduk, S. D., Bounaud, P. Y., Vrignaud, P., Bissery, M. C., Veith, J. M., Pera, P., and Bernacki, R. J. (1996) Syntheses and structure-activity relationships of the second-generation antitumor taxoids: Exceptional activity against drug-resistant cancer cells, J. Med. Chem. 39, 3889- 3896.

109. Shionoya, M., Jimbo, T., Kitagawa, M., Soga, T., and Tohgo, A. (2003) DJ-927, a novel oral taxane, overcomes P-glycoprotein-mediated multidrug resistance in vitro and in vivo, Cancer Sci. 94, 459-466.

110. Ajikumar, P. K., Tyo, K., Carlsen, S., Mucha, O., Phon, T. H., and Stephanopoulos, G. (2008) Terpenoids: opportunities for biosynthesis of natural product drugs using engineered microorganisms, Mol. Pharm. 5, 167-190.

111. Ajikumar, P. K., Xiao, W. H., Tyo, K. E., Wang, Y., Simeon, F., Leonard, E., Mucha, O., Phon, T. H., Pfeifer, B., and Stephanopoulos, G. (2010) Isoprenoid pathway optimization for Taxol precursor overproduction in Escherichia coli, Science 330, 70-74.

112. Boghigian, B. A., Salas, D., Ajikumar, P. K., Stephanopoulos, G., and Pfeifer, B. A. (2012) Analysis of heterologous taxadiene production in K- and B-derived Escherichia coli, Appl. Microbiol. Biotechnol. 93, 1651-1661.

113. Jiang, M., Stephanopoulos, G., and Pfeifer, B. A. (2012) Toward biosynthetic design and implementation of Escherichia coli-derived paclitaxel and other heterologous polyisoprene compounds, Appl. Environ. Microbiol. 78, 2497-2504.

114. Besumbes, O., Sauret-Gueto, S., Phillips, M. A., Imperial, S., Rodriguez- Concepcion, M., and Boronat, A. (2004) Metabolic engineering of isoprenoid

29 biosynthesis in Arabidopsis for the production of taxadiene, the first committed precursor of Taxol, Biotechnol. Bioeng. 88, 168-175.

115. Chang, M. C. Y., Eachus, R. A., Trieu, W., Ro, D. K., and Keasling, J. D. (2007) Engineering Escherichia coli for production of functionalized terpenoids using plant P450s, Nat. Chem. Biol. 3, 274-277.

116. Dejong, J. M., Liu, Y., Bollon, A. P., Long, R. M., Jennewein, S., Williams, D., and Croteau, R. B. (2006) of Taxol biosynthetic genes in Saccharomyces cerevisiae, Biotechnol. Bioeng. 93, 212-224.

117. Engels, B., Dahm, P., and Jennewein, S. (2008) Metabolic engineering of taxadiene biosynthesis in yeast as a first step towards Taxol (Paclitaxel) production, Metab. Eng. 10, 201-206.

118. Huang, Q. L., Roessner, C. A., Croteau, R., and Scott, A. I. (2001) Engineering Escherichia coli for the synthesis of taxadiene, a key intermediate in the biosynthesis of taxol, Bioorg. Med. Chem. 9, 2237-2242.

119. Zhou, K., Qiao, K., Edgar, S., and Stephanopoulos, G. (2015) Distributing a among a microbial consortium enhances production of natural products, Nat. Biotechnol. 33, 377-383.

120. Yadav, V. G. (2014) Combining metabolic pathway design and retrosynthetic planning for the design of a novel semisynthetic manufacturing scheme for paclitaxel, Org. Process Res. Dev. 18, 816-826.

121. Jennewein, S., Park, H., DeJong, J. M., Long, R. M., Bollon, A. P., and Croteau, R. B. (2005) Coexpression in yeast of Taxus cytochrome P450 reductase with cytochrome P450 oxygenases involved in Taxol biosynthesis, Biotechnol. Bioeng. 89, 588-598.

122. Rontein, D., Onillon, S., Herbette, G., Lesot, A., Werck-Reichhart, D., Sallaud, C., and Tissier, A. (2008) CYP725A4 from yew catalyzes complex structural rearrangement of taxa-4(5),11(12)-diene into the cyclic ether 5(12)-oxa-3(11)- cyclotaxane, J. Biol. Chem. 283, 6067-6075.

123. Geissler, J. F., Harwood, C. S., and Gibson, J. (1988) Purification and properties of benzoate-coenzyme A ligase, a Rhodopseudomonas palustris enzyme involved in the anaerobic degradation of benzoate, J. Bacteriol. 170, 1709-1714.

124. Muchiri, R., and Walker, K. D. (2012) Taxol biosynthesis: Tyrocidine synthetase A catalyzes the production of phenylisoserinyl CoA and other amino phenylpropanoyl thioesters, Chem. Biol. 19, 679-685.

125. Thornburg, C. K., Wortas-Strom, S., Nosrati, M., Geiger, J. H., and Walker, K. D. (2015) Kinetically-and crystallographically-guided mutations of a benzoate CoA

30 ligase (BadA) elucidate mechanism and expand substrate permissivity, Biochemistry-US 54, 6230-6242.

31

Chapter 2 is adapted from the following publication:

Kinetically and crystallographically guided mutations of a benzoate CoA ligase (BadA) elucidate mechanism and expand substrate permissivity. Biochemistry-US, 54, 6230-6242, 2015.

Chelsea K. Thornburg, Susan Wortas-Strom, Meisam Nosrati, James H. Geiger and Kevin D. Walker

BadA crystal screening and structure determination was performed by Susan W. Strom and Meisam Nosrati. All other work was performed by Chelsea K. Thornburg.

32 Chapter 2. Kinetically- and crystallographically-guided mutations of a benzoate CoA ligase (BadA) elucidate mechanism and expand substrate permissivity

2.1 Introduction

Acyl-group adenylation by ATP, in both primary and specialized metabolism, is common to all living organisms. The superfamily of ATP-dependent adenylation enzymes

(PFAM00501) is organized into three classes (1). Class I includes the non-ribosomal synthetase (NRPS) adenylation domains, the acyl CoA ligases, luciferase (2). Classes II and III comprise the aminoacyl-tRNA synthetases and

NRPS-independent siderophore (NIS) enzymes, respectively (3). The focus of the current study centers on an acyl CoA ligase from Class I. Plants and bacteria employ aroyl CoA thioesters for the biosynthesis of specialized metabolites. For example, anthraniloyl CoA is a precursor of a quorum sensing in Pseudomonas aeruginosa (4); benzoyl CoA lies on the pathway to enterocin (a bacteriocin) in the marine actinomycete Salinispora (5), to the chemotherapeutic Taxol in Taxus plants (6, 7), and to the volatile fragrant benzylbenzoate in plants (8). Biosynthesis of the anti-bacterial and anti-fungal aureothin proceeds through 4-nitrobenzoyl CoA in soil bacterium Streptomyces thioluteus (Figure

2.1) (9, 10).

Accounts of aerobic and anaerobic degradation of environmental pollutants such as polycyclic aromatic (PAHs) and polychlorinated biphenyls (PCBs) to benzoate derivatives in bacteria include the characterization of several different aroyl CoA ligases from Rhodopseudomonas palustris, Burkholderia xenoverans, Pseudomonas strain

33

Benzoyl CoA

Benzylbenzoate

Enterocin Taxol Anthraniloyl CoA 4-Nitrobenzoyl CoA

PQS Aureothin (Pseudomonas Quorum Signal) Figure 2.1. Representative natural products derived from acyl CoA thioesters.

34 CBS3, Thauera aromatica, and Pseudomonas aeruginosa (11-16). enzymes in these bacteria target the aroyl CoA thioesters for dearomatization and ring cleavage along catabolic pathways (17, 18). These organisms are thus candidates for bioremediation of man-made PAHs and PCBs to reduce their environmental impact (17, 19).

ATP-dependent CoA ligases use two half-reactions to catalyze thioesterification

(Figure 2.2) (20). The enzyme binds an alkyl- or aryl carboxylate and a Mg2+-ATP complex for subsequent coupling during the first half-reaction. The oxyanion of the carboxylate nucleophilically attacks ATP, releases diphosphate, and forms an acyl-AMP mixed anhydride intermediate. In the second half-reaction, the nucleophilic of CoA attacks the carbonyl of the acyl-AMP intermediate to make the thioester. Other crystal structures from this ligase family support the hypothesis that the enzymes adopt adenylation- and thiolation-conformations to catalyze their respective half-reactions (15, 21, 22).

Several active aroyl CoA ligase structures have been determined in complex with acyl carboxylates, cofactors, acyl-adenylates, AMP, ATP, and CoA (14, 15, 21, 23-29). R. palustris, chosen as a potential bacterial candidate for bioremediation, has at least three distinct aroyl CoA ligases with different substrate specificities and regulatory characteristics (13, 30). Of these, a benzoate CoA ligase (designated BadA) was identified in an earlier study (12) that reported on the relative Vmax (i.e., Vrel) values of BadA for different substrates compared to benzoate (12). Here, we explicitly calculated the apparent

app app kinetic parameters (kcat and KM ) of BadA for 25 different carboxylate substrates to further understand the steric, mesomeric, and inductive effects of the substituents on the mechanism. We intend to repurpose BadA for use in coupled enzyme assays to

35

Figure 2.2. ATP-dependent, two-step mechanism of a benzoate: CoA ligase in the presence of magnesium. BzO: Benzoate; Bz-AMP: Benzoyl AMP; and BzCoA: Benzoyl Coenzyme A. The ortho-, meta- and para-carbons of the aryl ring of BzO are highlighted with carbon numbering 2-, 3-, and 4-, respectively.

biosynthesize non-natural acyl CoAs. These thioesters will serve as substrates for CoA- dependent N- and O-acyltransferases that acylate many bioactive specialized metabolites from plants (31, 32). Calculation of the intrinsic catalytic constants of BadA for non-natural carboxylates is an important step towards this goal.

Also described herein are five X-ray crystal structures of recombinant BadA at ~1.80

Å resolution complexed with various aryl carboxylates and one with benzoyl-AMP (Bz-

AMP). BadA is structurally similar to other CoA ligases in its class, comprising separate

N-terminal and C-terminal domains joined by a flexible hinge. Based upon the structures of other aroyl CoA ligases, we expected the C-terminal domain of BadA-carboxylate complexes to remain in an "un-rotated" adenylate-forming conformation as seen in other benzoate derivatives in complex with ligases (21, 23, 27). Surprisingly, however, each complex showed BadA in the conformation primed for thioesterification. The homologous human medium chain fatty acyl CoA synthetase (ACSM2A) (24) provides precedent for the binding of benzoate (BzO) to BadA in the thiolation conformation. By analogy,

ACSM2A co-crystallized with the surrogate substrate ibuprofen in the thiolation conformation (24). In addition, each of the aryl carboxylates binds the BadA active site in a unique configuration rotated ~60° about the phenyl ring relative to BzO bound in other

36 structures and the Bz-AMP bound in BadA. Here, mutagenesis of the active site supports the importance of both orientations of the substrate in the active site. Together, the data support a mechanism where benzoate tightly binds BadA in the thiolation conformation, followed by rotation in the active site upon ATP binding for the acyl adenylation step. The structures further provide a basis for rational mutagenesis to remove sterically challenging residues from the active site and expand the substrate scope of BadA. The results of a series of mutations are also described here. Using these structures, we also determined that the non-conserved Lys427, present in the BadA active site in the thiolation conformation, is not required for acyl adenylate formation but is necessary for thioesterification.

37 2.2 Experimental

2.2.1 Materials

Benzoic acid (98%) was obtained from Mallinkrodt Pharmaceuticals (St. Louis, MO). 2-

Aminobenzoic acid (98%), 3-aminobenzoic acid (98%), 2-cyanobenzoic acid (96%), cyclohexane carboxylic acid (98%), 1-cyclohexene-1-carboxylic acid (97%), 3- cyclohexene-1-carboxylic acid (98%), 4-ethylbenzoic acid (99%), 2-hydroxybenzoic acid

(99+%), 3-hydroxybenzoic acid (99%), 2-methoxybenzoic acid (98%), 2-nitrobenzoic acid

(95%), 3-nitrobenzoic acid (99%), 2-picolinic acid (99%), and thiophene-2-carboxylic acid

(99%) were purchased from Alfa Aesar (Ward Hill, MA); acetyl CoA (98%). 4-

Aminobenzoic acid (99%), ATP (98%), benzoyl CoA (98%), 2,6-dichlorobenzoic acid

(98%), 3-chlorobenzoic acid (99%), cinnamic acid (99%), 3-cyanobenzoic acid (98%), 4- cyanobenzoic acid (98%), 3,4-dimethylbenzoic acid (98%), 2-fluorobenzoic acid (97%),

3-fluorobenzoic acid (97%), 4-fluorobenzoic acid (98%), L- (99%), 4- hydroxybenzoic acid (99%), 2-methylbenzoic acid (99%), 3-methoxybenzoic acid (99%),

4-methoxybenzoic acid (99%), 3-methylbenzoic acid (99%), 4-methylbenzoic acid (98%), phenylacetic acid (99%), L-phenylalanine, and L-tyrosine (98%) were obtained from

Sigma-Aldrich (St. Louis, MO). 2-Chlorobenzoic acid (>98.0%) and 4-chlorobenzoic acid

(99%) were from TCI America (Portland, OR). 4-Nitrobenzoic acid (99%) was purchased from Acros Organics (New Jersey). Coenzyme A (~95%) was purchased from Lee

Biosolutions (St. Louis, MO).

38 2.2.2 Plasmids

The badA cDNA from Rhodopseudomonas palustris was generously provided by Caroline

Harwood (University of Washington, Seattle, WA). The gene was amplified by PCR and subcloned into the pET28a (Novagen) expression vector with the following forward primers: 5'-TAT GAA TGC AGC CGC GGT C-3', 5'-TGA ATG CAG CCG CGG TCA

C-3' and reverse primers: 5'-GTC AGC CCA ACA CAC CC-5', and 5'-TCG AGT CAG

CCC AAC ACA CC-3'. Directional cloning of badA was confirmed by DNA sequencing

(Michigan State University Genomics Core). The resulting plasmid was named pBadA.

2.2.3 BadA protein expression and purification

The pBadA expression vector (pET28a) was used to transform BL21(DE3) E. coli

(Invitrogen, Carlsbad, CA). A single colony was selected and used to inoculate a 10-mL culture of Luria-Bertani (LB) media (Accumedia) containing 50 µg/mL kanamycin

(Roche). The culture was grown overnight at 37 °C. This seed culture (5 mL) was used to inoculate fresh LB media (1 L) containing 50 µg/mL kanamycin. This culture was grown at 37 °C until A600 = 0.8. Gene expression was induced by 0.5 mM isopropyl β-D-1- thiogalactopyranoside (IPTG), and the cultures were grown for 5 h at 18 °C and harvested by centrifugation at 6,000g. The bacterial pellet was resuspended in Buffer A (50 mM

Na2PO4, 300 mM NaCl, 15 mM imidazole, 5% glycerol, pH 8.0) containing EDTA-free

Protease Inhibitor Cocktail tablets (Roche Life Sciences, Indianapolis, IN). Cells were lysed with a Misonix XL 2020 sonicator and centrifuged at 18,000g for 30 min. The supernatant was passed through a 0.2 µm filter (Millipore, Billerica, MA) and loaded onto a Ni2+–NTA Qiagen column pre-equilibrated with Buffer A. The column was washed with

5 column volumes (CV) of Buffer A and eluted with 3 CV of Buffer A containing 250 mM

39 imidazole. Each fraction was analyzed by SDS-PAGE. Fractions containing BadA were combined, loaded into a 10 kDa MW cutoff Dialysis Cassette (Thermo Scientific Pierce,

Grand Island, NY), and dialyzed overnight against Buffer B (50 mM NaPO4 containing

5% glycerol, pH 8.0) for kinetic analysis or Buffer C (20 mM Tris, pH 8.0) for protein crystallization experiments. Dialyzed protein was concentrated in a Millipore Amicon

Ultra 30 kDa cutoff concentrator to ~10 mg/mL (estimated by the Coomassie (Bradford)

Protein Assays (Thermo Scientific Pierce, Grand Island, NY). The molecular weight of

BadA (501 amino acids) was verified by liquid chromatography/electrospray ionization/ mass spectrometry (LC-ESI/MS) on a Q-ToF Ultima Global mass spectrometer (Waters,

Milford, MA). Protein aliquots were frozen in liquid nitrogen and stored at –80 °C.

2.2.4 BadA kinetic assays

Stocks of ATP (10 mM) and CoA (10 mM) were dissolved in Buffer B, MgCl2 (at 100 mM) was stored in water, and stocks of the aromatic carboxylic acids (each at 100 mM) were dissolved in methanol. Aromatic carboxylic acids were then diluted in water. To establish steady-state kinetic rates of BadA with respect to protein concentration and time for each carboxylate (1 µM to 4 mM) and other reactants: ATP (250 µM), CoA (250 µM),

MgCl2 (750 µM) were combined in Buffer B and pre-incubated at 31 °C for 10 min before the addition of 0.1 to 20 µg/mL BadA (90 µL total volume). Assays were acid-quenched

(pH 3) with 8.8% formic acid in water. Acetyl CoA was added as an internal standard at a final concentration of 1 µM. Methanol concentrations were held constant at 1% (v/v) among assays with varied concentrations of carboxylic acids.

40 2.2.5 BadA assay analysis by liquid chromatography mass spectrometry

Liquid chromatography–electrospray tandem mass spectrometry (LC–ESI/MS/MS) in negative ion mode was used to quantify the biosynthetic acyl CoA products. An autosampler (at 10 °C) connected to a UPLC system (Waters Corp., Milford, MA) injected a 10-µL aliquot of each assay onto an Ascentis Express C18 HPLC column (2.7 µm, 5 cm

× 2.1 mm, at 30 °C, Sigma-Aldrich). The column was eluted at 0.4 mL/min with 2.5% solvent B (100% acetonitrile) and 97.5% solvent A (0.05% triethylamine in water (pH 5.5)) with a 0.5 min hold, followed by a linear gradient to 20% solvent B over 4 min, then increased to 100% solvent B over 2 min, and finally lowered to 2.5% solvent B over 0.5 min. The needle was washed with 2 mL each of 100% isopropanol and then with 10% acetonitrile in water prior to each injection. The HPLC effluent was directed to an electrospray ionization mass spectrometer (Quattro Premier XL, Waters Corp, Milford,

MA) in negative ion mode, with a cone voltage of 60 V and collision energy of 44 eV.

Each aryl CoA was quantified by multiple reaction monitoring of the [M – H]– → m/z 408 transition, common to each aryl CoA tested (Figure 2.3). Peak areas (calculated using the

MassLynx data analysis software, Waters Corp., Milford, MA) were converted to product concentrations using a standard curve for a series of benzoyl CoA concentrations (61 nM to 15.6 µM, n = 3) that were normalized to an internal standard acetyl CoA (1 µM).

Figure 2.3. Acyl CoA detection by tandem mass spectrometry. Multiple-reaction monitoring (MRM) LC- ESI/MS/MS [M - H]– → m/z 408 transition ion in negative ion mode for aryl CoA thioesters.

41 2.2.6 BadA mutations

Point mutations of the badA gene were generated by site-directed mutagenesis, using the following forward and reverse primer pairs (mutated bases are underlined): A227G: 5'-

CCA AAC TGT TTT TCG GCT ACG GCC TCG GCA ACG-3' and 5'-CGT TGC CGA

GGC CGT AGC CGA AAA ACA GTT TGG-3'; L333A: 5'-CGG CTC GAC CGA GAT

GGC GCA CAT CTT TCT GTC GAA C-3' and 5'-GTT CGA CAG AAA GAT GTG CGC

CAT CTC GGT CGA GCC G-3'; H334A: 5'-CGG CTC GAC CGA GAT GCT GGC GAT

CTT CCT GTC GAA TTT G-3' and 5'-CAA ATT CGA CAG GAA GAT CGC CAG CAT

CTC GGT CGA GCC G-3'; I335A: 5'-CCG AGA TGC TGC ACGCG TTT CTG TCG

AAC CTG C-3' and 5'-GCA GGT TCG ACA GAA ACG CGT GCA GCA TCT CGG-3';

K427A: 5'-CGA CAT GCT GGC GGT CAG CGG CAT CTA TGT CAG CCC GTT CGA

GAT CG-3' and 5'- GCC GCT GAC CGC CAG CAT GTC GTC GGT GCG GCC CGC

ATA GGT GTA G-3'. The pBadA plasmid was PCR amplified using Phusion HF polymerase (New England Biolabs, Ipswich, MA) under the following protocol: 95 °C for

3 min, followed by 20 cycles of 95 °C for 1 min, 58 °C for 1 min, 72 °C for 4.5 min, with a final elongation step of 72 °C for 7 min. PCR product was digested with DpnI (New

England Biolabs) at 37 °C for 2 h, then used to transform DH5α E. coli (Invitrogen, Thermo

Life Sciences, Grand Island, NY). The resulting colonies were inoculated into starter cultures and grown for DNA purification (PureYield Plasmid MiniPrep System, Promega,

Fitchburg, WI) and DNA sequencing for the badA point mutations at the Michigan State

University Research Technology Support Facility. Mutant plasmids were used to transform

BL21(DE3) E. coli (Invitrogen). The expression, isolation and purification of mutant enzyme were identical to that described for wild-type BadA.

42 2.2.7 Kinetic analysis

To calculate the kinetic constants, each substrate was varied (1 – 4000 μM) in separate assays. The steady-state concentrations of assay reagents were 250 µM ATP, 250 µM CoA,

2.5 mM MgCl2, 0.1% methanol in 50 mM NaHPO4 buffer (pH 8) with varying concentration of carboxylate substrate. BadA concentrations are noted for each substrate.

Resultant CoA thioester products were quantified after terminating the reaction as

app app described previously. The apparent kinetic parameters (KM and kcat ) were calculated by non-linear regression with Origin Pro 9.0 software (Northampton, MA) (Figures A.1.1–

app app A.1.24), using the Michaelis-Menten equation: vo = kcat [Eo][S]/(KM + [S]). Relative steady-state rates for mutant enzymes Ala227Gly-BadA, Leu332Ala-BadA, His333Ala-

BadA, and Ile334Ala-BadA were calculated for 15 aryl carboxylates. Assays with purified mutant protein contained 0.02 mg/mL mutant enzyme, 250 μM ATP, 1 mM carboxylate,

750 μM Mg2+, and 250 μM CoA. Relative velocities at apparent saturation are reported as percentages relative to BadA with benzoate. For the Lys427A-BadA mutant, the biosynthetic benzoyl-AMP and benzoyl CoA products catalyzed by the mutant were analyzed separately by HPLC (with A254 detection) and LC-ESI-MS/MS. Control assays contained the necessary cofactors and reagents but no mutant enzyme.

2.2.8 BadA crystal structures

2.2.8.1 Crystallization of R. palustris benzoate: coenzyme A ligase (BadA)

Purified BadA protein at (17.5 mg/mL) was used for de novo protein crystallization trials using six different 96-condition Crystal Screen (Hampton Research) matrices. A Gryphon

Protein Crystallization robot (Art Robbins Instruments) was used for setting up sitting-drop

43 vapor diffusion crystallization trials. The trials were kept at 22 °C throughout crystallization. About 45 different conditions produced crystals of varying quality; the best was from the Wizard I/II Screen A10 condition (20% PEG-2000 MME (w/v), 0.1 M Tris-

HCl, pH 7). This condition was then optimized in crystal boxes employing the hanging- drop vapor diffusion crystallization method. 2 µL hanging drops were used consisting of an equal measure of protein and reservoir solutions. The best crystals, used for data collection, were grown from a reservoir solution containing 0.1 M Tris-HCl, pH 7.0 and

15% PEG 3350 (w/v).

2.2.8.2 Co-crystallization to obtain the ligand bound structure

Co-crystallization was employed by adding separately each carboxylic acid ligand (4 equiv) to the BadA protein solubilized in 0.1 M Tris buffer, pH 8.0. For the multiple- component cocrystallization, ATP, CoA and benzoic acid (each at 1 eq) were added to the protein solution. Each co-crystallization mixture was incubated for 10 min at 4 °C followed by crystallization using a reservoir of 20 mM Tris buffer at 15% PEG 3350 at variable pH.

The best crystals grew in a range of pH from 6.5 to 7.5.

2.2.8.3 Data processing and refinement of BadA

Crystals were soaked in cryoprotectant (0.2 M Tris-HCl, pH 6.5 – 7.5, containing 20%

PEG-3350 and 20% glycerol), mounted in CryoLoops (Hampton Research), and then flash- frozen in liquid nitrogen. The native X-ray diffraction intensity data were collected under a continuous stream of nitrogen at 100 K on beamline 21-ID-G, LS-CAT (Argonne

National Laboratory, Advanced Photon Source, Chicago, IL) at a wavelength of 0.97872

Å. Raw diffraction data were indexed using the HKL2000 software package (33). A search of the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank

44 (PDB) revealed one known structure for a benzoate CoA ligase (BCLM) from Burkholderia xenovorans LB400 (PDB 2V7B) (14). Query of the Swiss-Model server (34) produced a threaded protein structure homology-model of BadA based on the BCLM structure. The

BadA structure was solved by molecular replacement using this model and the MOLREP program in the CCP4 suite (35-37). Though the N-terminal domain was well-positioned within the density, the C-terminal domain was not. The model was corrected using the

Buccaneer program in the CCP4 suite to thread the amino acid sequence into the density.

Further corrections were made manually using the COOT program (38). Jligand version

1.0.9 (39) of the CCP4 suite was used to add the template restraints for the ligands prior to refinement. The structures were refined using REFMAC in the CCP4 program suite. Water were added using REFMAC and COOT near the end of the refinement.

2.2.9 Calculation of covalent van der Waals volumes and lengths

The molecular volumes of the substituents on the aryl carboxylate substrates were estimated as Connolly solvent-excluded volume (40) using ChemBio3D Ultra software

(ver. 13.0, Perkin Elmer) and a probe radius of 0.1 Å. The volume of a phenyl anion was subtracted from the estimated volume of a phenyl attached to a substituent. The geometries of polyatomic substituents such as CH3, OH, MeO, CN, NH2, and NO2, attached to phenyl were MM2 energy minimized using the default parameters of the ChemBio3D Ultra software. The covalent van der Waals lengths of CN and Cl substituents attached to a phenyl ring were calculated using the distance between two and atomic radius of the terminal .

45 2.3 Results

2.3.1 Solving the BadA structure

BadA crystallized as a complex with benzoate (BzO) in the absence of added benzoate.

BadA also co-crystallized with 2-fluorobenzoate (2-F-BzO), 2-methylbenzoate (2-Me-

BzO), 3-furoate (3-Fur), and thiophene-2-carboxylate (Thio-2-C) (Figure 2.2 and Figures

A1 – A24 for structures). BadA was also co-crystallized with both benzoate and ATP resulting in a bound Bz-AMP intermediate, the product of the first half-reaction (see Figure

2.2), bound in the active site. All structures were refined to resolutions of at least 1.8 Å

(Table A1) and helped guide the interpretation of the kinetics data.

Figure 2.4. Crystal structure of BadA in complex with benzoate. The substrate is shown in yellow and enclosed in electron density map (2Fo - Fc). The N-terminal domain (green) and the C-terminal domain (blue) in the "thiolation conformation." BadA (PDB: 4EAT)

2.3.1.1 Domain orientation

The BadA structures in complex with BzO (PDB: 4EAT) (Figure 2.4 and Figure A.1.26),

2-Me-BzO (PDB: 4RLQ), Thio-2-C (PDB: 4RMN), 3-Fur (PDB: 4RM3), 2-F-BzO (PDB:

4RM2) and Bz-AMP (PDB: 4ZJZ) (Figure A.1.26) had the characteristic N-terminal and

46 C-terminal domains of other ATP-dependent adenylases (2, 28). The N-terminal (residues

1 – 434) and C-terminal (residues 435 – 522) domains of BadA fold analogously to the benzoate CoA ligase BCLM (PDB: 2V7B) (14) with an RMSD between the peptide backbones of ~0.5 Å and 0.6 Å, respectively (41). The N-terminal domain contains the benzoate , and the C-terminal domain contacts the outer edge of the benzoate binding pocket, positioning benzoate in the active site through a charged interaction between the carboxylate of benzoate and Lys427 (Figure 2.4).

2.3.1.2 Features of the BadA active site

The BadA active site is largely hydrophobic with many binding contacts between the substrate and the polypeptide backbone. In all BadA structures bound to the aryl carboxylate, the active site coordinates the carboxylate substrate with Lys427. In this position, the para-carbon of BzO neighbors the carbonyl of Leu332 along the peptide backbone, and the two meta-carbons of BzO point towards His333 and Ala227 on either side (Figure 2.5, A and D). The si- and re-faces of the BzO are positioned between backbone bonds comprising Gly327, Ser328, and Thr329 on one face, and Tyr228 on the other face.

Co-crystallization of BadA with BzO and ATP yielded a Bz-AMP intermediate in the active site with the C-terminal domain also poised in the thiolation conformation (21,

26). The active sites of structures bound to BzO or Bz-AMP showed almost no structural differences. However, when Bz-AMP forms, Lys427 moves from an interaction with the carboxylate of the substrate to several polar contacts with the Bz-AMP and the peptide backbone (Figure 2.5, B and D). We evaluated the importance of Lys427 of BadA for catalysis by making a Lys427Ala-BadA mutant and compared its activity against BadA.

47 app -5 –1 The mutant made benzoyl CoA at a rate (kcat = (83 ± 1.8) × 10 s ), more than 33000-

app –1 app fold slower than wild type BadA (kcat = 28 s ), yet the KM (1.5 ± 0.31 µM) for this mutant was nearly the same for the wild-type BadA (4.4 ± 0.65 µM) (Figure A.1.27). The mutant and BadA catalyzed approximately equal amounts of benzoyl-AMP in the absence of CoA (Figure A.1.28). These data indicated that Lys427 is needed for the thiolation reaction.

When the Bz-AMP anhydride forms, other structural differences include a decrease in the dihedral angle between the carboxylate and the aryl ring from 37° for BzO to 12° for the anhydride. Further, Tyr432 shifted toward the phosphate oxygen of the AMP anhydride and away from Arg421 to accommodate the Bz-AMP (Figure A.1.29). The benzoyl moiety also rotates ~60° about the C6-axis of the aromatic ring and moves the ortho-carbon from an unhindered region of the active site towards residue Ala227, which previously resided near the meta-carbon (Figure 2.5, C and D). This rotation positions the benzoyl moiety of

Bz-AMP similar to those of benzoates bound in other CoA ligase structures, regardless of the C-terminal domain orientation (14, 21, 23).

48

A B

C D

Figure 2.5. The active site of the BadA structure. A. Three-dimensional rendering of BadA in complex with benzoate (BzO) (cyan) (PDB: 4EAT) and B. benzoyl-AMP (Bz-AMP) (yellow) (PDB: 4ZJZ); key, steric and substrate binding residues (green) are highlighted. C. Overlay of the active site orientations of BzO (cyan) and Bz-AMP (yellow) in complex with BadA. Structure alignments were made with UCSF Chimera 1.10 (42). D. Two-dimensional rendering of the active site showing distances (Å) between BzO (bold black structure) and Bz group of Bz-AMP (bold blue structure) and key, steric active site residues. (Insets) Two views of the dihedral angle between the carboxylate oxygens and the phenyl ring of BzO and Bz-AMP (partial structure) bound in BadA. Heteroatoms are colored as oxygen (red), phosphorus (orange), and nitrogen (blue).

49 2.3.2 Kinetic properties of BadA

The R. palustris BadA was recombinantly overexpressed in E. coli and incubated separately with various heteroaromatic, alkenoyl, alkyl, and aryl carboxylates to calculate the kinetic constants for each substrate (Table 2.1). Benzoate substrates with substituents of differing steric and electronic properties were chosen to systematically probe the active site of BadA. We quantified the biosynthetic CoA thioester products by LC-ESI-MS/MS.

2.3.3 Substrate turnover by BadA

app –1 app The kcat (28 ± 0.9 s ) and KM (4.4 ± 0.65 µM) of BadA for the natural substrate BzO are

app –1 app consistent with those reported earlier (kcat = 27 s ; KM = 0.6 – 2 µM) (12). These values were compared to the Michaelis parameters of BadA for 21 mono-substituted benzoates, three cyclohexane derived carboxylates, and two heteroaromatic carboxylates to their corresponding CoA thioesters (Table 2.1). Although KM is not a true dissociation constant, in this study it will serve as a means of comparing enzyme interactions with different carboxylate substrates.

2.3.3.1 Halogenated benzoates

app BadA turned over substrates 2-F-, 3-F-, and 4-F-BzO as well as BzO, and the KM values of BadA for 2-F-BzO and 4-F-BzO were similar to that for BzO, but 3-F-BzO was an

app outlier (Table 2.1). Interestingly, the KM for 3-F-BzO was 17-fold higher than for BzO, reducing its relative catalytic efficiency 13-fold and suggesting that an interaction between the meta-position and active site residues affected binding but not turnover.

50

Table 2.1. Kinetic parameters of BadA for various substrates.

a app -1 app app app -1 -1 Substrate kcat (s ) KM (µM) kcat /KM (s µM ) BzO 28 ± 0.9 4.4 ± 0.65 6.4 Thio-2-C 4.8 ± 0.2 37 ± 5.7 0.13 3-Fur 6.7 ± 0.5 71 ± 12 0.094 PhAc ND - - Cinn ND - - Non-aromatic carboxylates 3-Cyc 16 ± 1.03 56 ± 9.1 0.29 1-Cyc 5.6 ± 0.24 52 ± 7.1 0.11 Cyc 13.5 ± 1.07 1470 ± 213 0.0092 Ortho-substituted benzoates 2-F 34 ± 0.7 8.1 ± 0.91 4.2 2-OH 0.63 ± 0.01 1.6 ± 0.2 0.39 2-NH2 3.9 ± 0.12 21 ± 2.3 0.19 2-CN 0.11 ± 0.0034 213 ± 22 0.00052 2-Cl 2.2 ± 0.07 130 ± 12 0.017 2-Me 1.2 ± 0.042 44 ± 6.1 0.027 2-MeO 0.011 ± 0.0005 789 ± 82 0.000014 2-NO2 0.27 ± 0.011 2015 ± 167 0.00013 Meta-substituted benzoates 3-F 35 ± 1.4 73 ± 11 0.48 3-OH 1.8 ± 0.04 229 ± 18 0.0079 3-NH2 1.8 ± 0.11 174 ± 32 0.010 3-CN Slow - - 3-Cl 0.33 ± 0.01 290 ± 23 0.0011 3-Me 0.54 ± 0.018 215 ± 26 0.0025 3-MeO ND - - 3-NO2 ND - - Para-substituted benzoates 4-F 22 ± 0.66 6.6 ± 0.9 3.3 4-OH 0.93 ± 0.037 158 ± 17 0.0059 4-NH2 0.18 ± 0.004 25 ± 3 0.0072 4-CN Slow - - 4-Cl 0.29 ± 0.012 83 ± 11 0.0035 4-Me 0.7 ± 0.035 590 ± 73 0.0012 4-MeO ND - - 4-NO2 ND - - ND: Not Detectable, below . Slow: The CoA thioester product was detected at 0.4 pmol s-1 at apparent saturation, even at BadA concentrations up to 100 ug/mL. Initial steady-state rates at low substrate concentrations were too slow to calculate kinetic parameters. Standard error in parentheses. aThio-2-C: thiophene-2-carboxylate, 3-Fur: 3-furoate, PhAc: phenylacetate, Cinn: cinnamate, 3-Cyc: 3-cyclohexene- carboxlyate, 1-Cyc: 1-cyclohexene-carboxlyate, Cyc: cyclohexane carboxylate. Aromatic ring positions are designated as o-, m-, and p- (ortho-, meta-, and para-, respectively). Standard errors in parentheses were calculated for the non-linear regression fit by the Origin Pro 9.0 software.

51 BadA turned over 2-chlorobenzoate (2-Cl-BzO) better than 3-Cl-BzO or 4-Cl-BzO, in a trend similar to the fluorobenzoates. However, compared to BzO, turnover rates were

app ~10-fold slower for 2-Cl-BzO and 100-fold slower for both 3-Cl- and 4-Cl-BzO. The KM values of BadA for the Cl-BzO series (between 83 and 290 μM) were much higher than

app app the corresponding F-BzO substrates (<10 μM). Accordingly, the kcat /KM values of BadA for the Cl-BzO substrates were considerably reduced (~380-fold for 2-Cl-BzO and up to

~6000-fold for 3-Cl-BzO) compared to that for BzO. These data suggest the larger chlorine

(43) likely affects the binding interaction of all regioisomers of the chlorinated substrates and prevents them from regularly adopting a catalytic conformation, causing slower turnover (Table 2.1).

2.3.3.2 Benzoates with strongly electron-withdrawing substituents

The 3-NO2- and 4-NO2-BzO substrates were not productive, while 2-NO2-BzO was turned over (Table 2.1). Similarly, 2-cyanobenzoate (2-CN-BzO) was the only productive substrate among the isomeric CN-BzO substrates, but ~2.5-fold less active than 2-NO2-

app BzO (Table 2.1). The KM of BadA for 2-CN-BzO was ~10-fold lower than for 2-NO2-

BzO (2000 µM). These data are consistent with the observation that substituent regiochemistry affects substrate binding and subsequent turnover.

2.3.3.3 Benzoates with strongly electron-donating substituents

The catalytic efficiency of 2-hydroxybenzoate (2-OH-BzO) was 16-fold lower than BzO,

app app due primarily to its lower kcat (Table 2.1). By comparison, the KM values of BadA for 3-

OH- and 4-OH-BzO) were substantially higher than for 2-OH-BzO, yet were turned over similarly.

52 app BadA did not bind 2-aminobenzoate (2-NH2-BzO) (KM = 21 µM) as well as 2-OH-

BzO (1.6 µM), and was turned over 6-fold faster than 2-OH-Bz (Table 2.1). The catalytic

–1 -1 efficiencies of BadA for 3-NH2-BzO and 3-OH-BzO were similar (~0.1 s ·µM ) because

app app of their comparable kcat and KM values. 4-NH2-BzO binds BadA ~6-fold with more

app app apparent affinity (KM = 25 μM) than the 4-OH-BzO (KM = 160 μM), which is reciprocal to the ~6-fold difference in product release (0.18 and 0.93 s–1, respectively) of the corresponding CoA thioesters, thus equalizing the catalytic efficiencies of BadA for these substrates.

BadA turned over each Me-BzO regioisomer in almost the same order of reactivity

app as OH-, NH2-, and Cl-substituted substrates (ortho > meta > para) (Table 2.1). The kcat

app values for BadA with Me-BzO substrates compared to BzO were slower, and the KM values were higher. The larger Michaelis constants were the dominant factors that affected the catalytic efficiency of BadA for the Me-substituted substrates. For the MeO-BzO substrates, BadA turned over the 2-MeO-BzO 2,000-fold slower than BzO, with a 180-fold

app higher KM than for BzO. 3-Methoxybenzoyl (3-MeO-Bz) CoA and 4-MeO-Bz CoA were not detectable under standard assay conditions, and were below the detection limit (at 0.5 pmol/mL) of the mass spectrometer.

2.3.3.4 Turnover of heteroaromatic carboxylates

BadA turned over the two heteroaromatics Thio-2-C and 3-Fur at similar rates (Table 2.1).

app app The kcat and the KM of BadA for Thio-2-C are ~6-fold slower and ~8-fold higher, respectively, than the values for BzO; whereas, the values for 3-Fur were ~4-fold slower and ~14-fold higher, respectively. These data showed that 3-Fur was turned over (1.4-fold)

53 faster by BadA, but had less apparent affinity (1.9-fold) for the enzyme than Thio-2-C, thus equalizing their catalytic efficiencies at ~0.1 s-1·µM-1 (Table 2.1).

2.3.3.5 Turnover of non-aromatic carbocycle carboxylates

app Intrinsic kcat values of BadA for non-aromatic carboxylates are reported here for the first time. The turnover rates of cyclohexane carboxylate, 1-cyclohexene-1-carboxylate (1-

Cyc), and 3-cyclohexene-1-carboxylate (3-Cyc) were similar (only 5 to 1.8-fold lower) to

app that of BzO. The KM of cyclohexane carboxylate (Cyc) was ~330-fold lower than BzO among these carbocycles. This trend is also reflected in the lower catalytic efficiency of

app BadA for Cyc than for 1-Cyc and 3-Cyc. However, the kcat of BadA for Cyc was surprisingly 48% compared with the turnover of BzO. This result improves upon an earlier

app BadA study where the relative apparent maximum rate (Vrel ) was at 1% for Cyc compared with that for BzO, likely due to a low Cyc concentration (250 µM) (12), below the

app calculated KM = 1470 µM in this study.

2.3.4 Rational Mutation of the BadA Active Site

BadA structures in complex with carboxylate substrates and a benzoyl-AMP intermediate guided site-directed mutagenesis on the wild-type BadA to remove steric bulk from the active site and expand the substrate preference for benzoates with meta- and para- substituents. Four residues (Ala227, Leu332, His333, and Ile334) surround the phenyl ring of BzO in the active site (Figure 2.5). The residue positioned near a given carbon of BzO depends on whether the BzO is in the carboxylate-bound orientation (Figure 2.5A), or the rotated Bz-AMP-bound orientation (Figure 2.5B). For example, Ala227 is next to the meta- carbon of BzO (Figure 2.5A) and the ortho-carbon of Bz-AMP (Figure 2.5B). Similarly, the peptide backbone of Leu332 is near the para- and meta-carbons of BzO and Bz-AMP,

54

Table 2.2. Relative apparent maximum rates of BadA and point mutants for various substrates.

app a app Substituent (Vrel ) Vrel of BadA Point Mutations on BzO BadA A227G I334A H333A L332A H (BzO) 100b 86 93 89 89 (8.8)c (4) (3) (10) (8) 2-CH3 10.6 42 1.8 11 1.7 (0.5) (1) (<0.1) (1) (0.2) 2-NO2 1.6 1.2 0.03 1.4 0.12 (0.10) (0.012) (0.002) (<0.1) (0.01) 2-OCH3 0.09 10.4 0.003 0.058 0.004 (0.008) (0.33) (<0.001) (0.005) (<0.001) 2-CN 0.02 12.0 0.03 0.08 0.05 (0.002) (0.82) (<0.01) (0.01) (0.002) 2,6-DiCl ND 0.69 ND ND ND (0.02) 3-CH3 2.04 7.0 49 44 1.5 (0.4) (0.3) (1.4) (3.6) (0.12) 3-OCH3 ≤0.001 2.9 70 10.7 0.086 (0.32) (6.2) (0.24) (0.009) 3-CN ND 0.038 8.5 2.5 0.017 (1.3) (0.15) (0.002) 3-NO2 ND 0.015 0.23 1.2 ND (0.0023) (0.036) (0.034) 4-CH3 5.02 0.29 14 4.5 0.063 (0.304) (0.014) (0.35) (0.202) (0.004) 4-CN ≤0.001 ≤ 0.001 0.010 2.9 ND (0.001) (0.3) 4-OCH3 ND < 0.001 39 0.93 ND (2.9) (0.019) 4-NO2 ND ND ≤ 0.001 0.11 ND (0.010) 4-Ethyl ND ND 9.4 0.86 ND (0.83) (0.058) 3,4-DiCH3 <0.001 0.016 ND 0.69 ND (0.0032) (0.11) (PhAc) ND ND 0.012 < 0.001 ND (0.0022) a app Vrel are normalized (as percentage (%)) to the apparent relative Vmax of BadA for benzoate. Highlighted (bold) are inactive substrates with BadA that are active with a mutant, or the mutant is 10-fold more active than BadA. b app c Vrel for BadA with benzoate is 15.1 nmol/min. Standard deviations are in parenthesis (n = 3). ND = Not Detected.

55 respectively. The His333 side chain points toward the meta- and para-positions of BzO and Bz-AMP, respectively, while the Ile334 side chain is near the ortho-carbon of BzO and the meta-carbon of Bz-AMP. Therefore, targeted mutations of Ala227Gly, Leu332Ala,

His333Ala, and Ile334Ala should show the relative importance of the two orientations for reactivity. The latter three were mutated to Ala instead of a smaller, flexible Gly to maintain conformational rigidity in the peptide backbone. Each mutant turned over the natural substrate benzoate at 86% to 93% the rate of BadA (Table 2.2).

2.3.4.1 Ala227Gly-BadA mutant

app The maximum relative velocities (Vrel ) of the Ala227Gly-BadA mutant increased for ortho-substituted BzO substrates (2-Me (~4-fold), 2-CN (600-fold), and 2-MeO (116- fold)) compared with those of BadA. Ala227Gly-BadA also uniquely converted 2,6- dichlorobenzoate (2,6-diCl-BzO) to its CoA thioester while BadA and the other mutant enzymes did not. Increases in activity with meta-substituted BzO substrates were also observed. A ~3-fold increase in activity with 3-Me-BzO was observed in addition to a

~2900-fold (or greater) increase in activity with 3-MeO-BzO. Novel, but low level activities with 3-CN-BzO and 3-NO2-BzO were observed. Compared with the activities of

BadA for substrates with para-substituents, the mutant activities were roughly the same, except the mutation markedly decreased the turnover of 4-Me-BzO by 17-fold (Table 2.2).

2.3.4.2 Ile334Ala-BadA mutant

app The Ile334Ala-BadA mutant had a higher Vrel for meta- and para-substituted benzoates

app but lower for all ortho-substituted benzoates tested. The Vrel values of Ile334Ala-BadA compared with those of BadA increased 70,000-fold for 3-MeO-BzO, 24-fold for 3-Me-

BzO, 2.8-fold for 4-Me-BzO, and 7-fold for 4-CN-BzO (Table 2.2). The BadA Ile334Ala

56 mutant had expanded substrate specificity for previously inactive substrates: 4-MeO

(39%), 4-ethylbenzoate (4-Ethyl-BzO) (9.4%), 3-CN-BzO (8.5%), 3-NO2-BzO (0.2%), a benzoate homolog, phenylacetate (PhAc) (0.01%) and 4-NO2-BzO (0.001%) (Table 2.2).

The Ile334Ala mutation, designed to help binding and turnover of BadA for ortho- substituted substrates, dramatically increased the catalysis of larger meta- and para- substituents

2.3.4.3 His333Ala-BadA mutant

His333Ala-BadA mutant, designed to help binding and turnover of BzO substrates with substituents at the meta-position resulted in increases in activity with both meta- and para-

app substituted BzO substrates, but did not significantly change the Vrel for ortho-substituted

app BzO substrates compared to BadA. Compared with BadA, the Vrel values of His333Ala-

BadA increased 11,000-fold for 3-MeO-BzO, 22-fold for 3-Me-BzO, 3,400-fold for the

3,4-diMe-BzO, and 2,100-fold for 4-CN-BzO (Table 2.2). New activities (relative to BzO

(at 100%)) for this mutation include 3-CN- (2.5%), 3-NO2- (1.2%), 4-MeO- (0.9%), 4-

NO2- (0.1%), and 4-Ethyl- (0.9%) BzO substrates (Table 2.2).

2.3.4.4 Leu332Ala-BadA mutant

The Leu332Ala mutant was designed to increase binding and turnover of para-substituted benzoic acids. Upon testing this mutant, we observed a significant loss of activity (~90%) for nearly all substrates except BzO. Small increases in activity with 3-MeO-BzO and 3-

CN-BzO were observed. The Leu332Ala mutant turned over BzO at 89% the rate of BadA; other mutants tested turned over the natural substrate at a similar rate.

57 2.4 Discussion

In this study, to provide insight on active site residues that define substrate selectivity, we solved the crystal structure of BadA and used this information to design point mutants with greater substrate permissivity. In addition, we measured the Michaelis parameters of BadA for various substrates to evaluate how steric effects and electronics, through interactions with a heteroatom or substituent, affect binding and turnover.

2.4.1 BadA structure and homology

Homologous adenylase structures of CoA ligases and AMP ligases (20-30% similarity) in the protein databank (including BadA) show that enzymes in this family typically fold into a larger N-terminal domain (400-550 residues) and a smaller C-terminal domain (~110 residues), where the active site lies between the domains. The N-terminal domain contains nearly all the residues that bind the carboxylate substrate. The N-terminal domain also binds the adenosyl group of ATP while the C-terminal domain residues coordinate the and phosphate groups. Several related CoA ligase crystal structures support a domain-alternation mechanism where the C-terminal domain rotates ~140° from its adenylation conformation to a thiolation conformation after the acyl-AMP forms and CoA binds (21). The domain rotation occurs on a flexible hinge loop that contains a conserved

Asp residue (26), and this conformational change configures the ligase for the thiolation step in the second half-reaction (44). The adenylation conformation supposedly forms when the carboxylate alone (or with ATP) binds the active site in the N-terminal domain

(24, 26). After the formation of the acyl anhydride intermediate and binding of CoA, the

C-terminal domain rotates into the thiolation conformation. This rotation places key

58 catalytic C-terminal residues in the active site and primes the enzyme for thioesterification

(21, 26, 45).

To our knowledge, all known structures of ATP-dependent aroyl CoA ligases in complex with only their natural carboxylate substrate adopt the adenylation conformation

(14, 23, 25, 28). The structure of a 4-chlorobenzoate CoA ligase (CBAL) from Alcaligenes sp. AL3007 in complex with 4-chlorobenzoate (4-Cl-BzO) (PDB: 1T5D) or 4-Cl-Bz-AMP

(PDB: 3CW8) adopts the adenylation conformation (Figure 2.6A). Subsequently, CBAL adopts the CoA-bound thiolation conformation when complexed with a phenacyl analog of 4-chlorobenzoyl CoA and AMP (PDB: 3CW9) (Figure 2.6B). The crystal structure of a benzoate CoA ligase (BCLM) (PDB: 2V7B) from Burkholderia xenoverans (14) (61% sequence identity with BadA), in complex with only benzoate, is in the adenylation conformation (Figure 2.6B). In contrast, all six BadA structures in complex with either aryl carboxylates or Bz-AMP are found surprisingly in the thiolation conformation, even in the absence of CoA. This observation deviates from the predicted sequential domain- alternation mechanism proposed for ATP-dependent CoA ligases (21). The conserved

Asp424 hinge residue in BadA and the flanking residues of the flexible loop are nearly identical with those in BCLM. Reasons for why the BadA and BCLM structures adopt different C-domain conformations when bound to aryl carboxylate substrates remain obscure, particularly, since a BadA structure in the adenylation conformation is unavailable. However, one possible explanation is that the two enzymes have different resting states; BadA assumes the thiolation conformation, while BCLM adopts the adenylation conformation when benzoate is bound.

59 Further, sequence and structural alignments show that the active site residues (Figures A30 and A31) and architectures of the N-terminal domains of BadA and BCLM in complex with

app benzoate are identical. It is therefore interesting that the relative kcat values for BadA with fluorobenzoates ranged between 75% and 117% relative to BzO (100%); whereas, BCLM turned over 2-F-BzO at 2% the rate for BzO (14). Active site residue Phe236 in the adenylation conformation of BCLM is located <4 Å above the carboxylate group of the bound BzO and blocks the CoA channel. The analogous Phe226 residue in BadA is offset by a ~72° rotation away from the BzO, opening the CoA-binding channel (Figure A.1.31).

Because the BadA and BCLM structures have different C-domain conformations, it is difficult to provide a structural rationale for differences in substrate specificity between

BadA and BCLM.

60

A B

C D

Figure 2.6. BadA structural alignments and C-terminal domain orientation. A. C-Terminal domains (ligands omitted for clarity) of CBAL in complex with 4-Cl-BzO (PDB: 1T5D) (green) and 4-Cl-Bz-AMP (PDB: 3CW8) (yellow), in the adenylation conformation, are overlaid with those of BadA in complex with BzO (PDB: 4EAT) (red) and Bz-AMP (PDB: 4ZJZ) (cyan) in the thiolation conformation. B. C-Terminal domains (ligands omitted for clarity) of CBAL in complex with both 4-Chlorophenacyl CoA and AMP (PDB: 3CW9) (green), and BCLM in complex with BzO (PDB: 2V7B) (purple) are overlaid with that of BadA in complex with BzO (PDB: 4EAT) (red) in the thiolation conformation. C. Relative positions of catalytically important lysine residues in BadA and BCLM. C-terminal domain (blue) is in the thiolation conformation: Lys427 (green) of BadA (PDB: 4ZJZ) is in the active site and Lys512 (red) is solvent exposed. D. C-terminal domain (blue) is in the adenylation conformation: Lys520 (red) of BCLM (PDB: 2V7B) is in the active site and Lys433 (green) is solvent exposed. N-terminal domains (gray).

61 2.4.2 Catalytically important lysine residues in BadA

Despite the different conformations of the C-terminal domain for BadA and BCLM, each uses a distinct C-terminal domain lysine (BadA: Lys427/512 and BCLM: Lys433/520) to orient the BzO substrate in either conformational state. In the adenylation conformation,

BCLM uses Lys520 (aligns with Lys512 of BadA) to coordinate BzO. An earlier study showed that mutation to alanine or modification by native enzymatic Nε-acetylation of

Lys512 of BadA reduced the benzoyl CoA product formation by 99% (30, 46). In all BadA structures in the thiolation conformation presented here, Lys427 coordinates BzO in the active site and Lys512 is far from the active site and solvent-exposed in the C-domain

(Figure 2.6, C and D). The apo structure of the ACSM2A crystallized in the thiolation conformation, supporting the idea that this conformation is available for binding by the carboxylate ligand (24). A thiolation conformation resting state would make the conserved lysine (Lys512 of BadA), present in all ATP-dependent CoA ligases, solvent accessible for post-translational regulation by Nε-acetyltransferases/deacetylases (30, 46, 47).

The surface-exposed, conserved lysine residue is accessible to regulatory acetyltransferases/deacetylases when these ligases adopt the thiolation conformation. By contrast, the adenylation conformation returns this lysine to the active site for adenylation catalysis. Lys427 of BadA makes four polar contacts with Bz-AMP, one with the benzoyl oxygen, three with the AMP moiety, and two to Gly303 and Gly430 (Figure 2.7). These contacts likely anchor Bz-AMP, and the positive charge of Lys427 primes the AMP- leaving group for CoA transesterification. Based on this information, we predicted a

Lys427Ala-BadA mutant would slow the second thioesterification step of the BadA reaction, and found indeed that the mutant was 99.9% less active compared to BadA.

62 However, the rate of acyl-AMP formation appears unaffected. In the absence of structural data, the importance of Lys427 of BadA on catalysis was unexpected due to the semi- conserved nature of the residue.

Together, the earlier Lys512 modification study and the Lys427 mutagenesis described here suggest that when BadA assumes the adenylation conformation (likely stabilized when ATP binds), Lys512 enters the active site. It is imagined that the latter residue then coordinates and primes BzO for nucleophilic attack on ATP based on homology with BCLM in the adenylation conformation. This mechanistic sequence is analogous to that described in an earlier study for ACSM2A (24). The latter proceeds through a proposed series of conformations [thiolation- (carboxylate-binding step), adenylation- (ATP-binding step), and returns to thiolation (acyl-AMP forming, CoA binding, and thioesterification steps)] to catalyze its thioesterification reaction.

Figure 2.7. Active site of BadA showing possible polar contacts between Lys427 and Bz-AMP. The Bz- AMP ligand (yellow), Gly304, and Gly431 (green) and distances (Å) are shown. Heteroatoms are colored as oxygen (red), phosphorus (orange), and nitrogen (blue).

63 Lys427 is not highly conserved among ligases, however BCLM contains a presumed functionally-similar lysine residue (Lys433) at the same position as the active site Lys427 of BadA, as does the phenylalanine adenylation domain (PheA) of gramicidin synthetase I from Brevibacillus brevis, and the D-alanyl carrier protein ligase (DltA) from Bacillus cereus (28, 48). In other ligases, such as the dihydroxybenzoate CoA ligase (DhbE) from

Bacillus subtilis (Asn434) or 4-chlorobenzoate CoA ligase (CBAL) (Ile406), other residues

(23, 25) align with Lys427 of BadA (Figure A.1.32). These residues are predicted to orient similarly to Lys427 of BadA in the thiolation conformation based upon sequence and structural alignments. The importance of Lys427 for BadA catalysis of the thioesterification reaction led us to postulate that lysine surrogates of DhbE and CBAL perform a similar function in stabilizing the acyl-AMP intermediate. Likely, a combination of Asn434 and Arg428 of DhbE and Asn411 and Arg400 of CBAL (similar to the highly- conserved Arg421 of BadA) form polar contacts with the acyl-AMP intermediates. We anticipate that lysine residues or nearby polar residues positioned similarly to that of

Lys427 and Arg421 of BadA are important for thioesterification in other ATP-dependent

CoA ligases.

2.4.3 Structural rationale for substrate specificity of BadA

Through structural characterization of BadA, the different trajectories of BzO and the pathway intermediate Bz-AMP suggest that the substrate must rotate ~60° counterclockwise (relative to re-face, see Figure 2.5D) about the C6-axis of symmetry. This rotation causes the ring carbons of BzO to move past multiple active site residues to access the Bz-AMP trajectory (Figure 2.5D). Upon binding BzO, the residues in BadA closest (≤

3.8 Å) to one ortho-carbon are Ile334 and Ala302. The other ortho-carbon points towards

64 a void within the active site where the phosphate group of Bz-AMP will ultimately reside

(Figure 2.5 and Figure A.1.33), supporting the favored binding of ortho-substituted benzoates in BadA. The structures of ortho-substituted benzoates bound to BadA all show them oriented such that the substituents occupy this void region, and are not pointed toward

Ile334 and Ala302. The meta-carbons of the benzoates align with the Ala227 and His333, and the para-carbon is nearest the carbonyl of Leu332 on the peptide backbone. The effects

app of the steric residues were evidenced generally by the higher KM values of BadA for meta- and para-substituents with increasing covalent van der Waals volumes F < OH < NH2 <

CN < Cl < Me < NO2 < MeO of benzoates (Table A2) over the ortho-regioisomers (Table

2.1).

When the substrate rotates toward the Bz-AMP conformation, the steric active site residues presumably reduced access to the catalytically proper orientation and generally

app also slowed the kcat of BadA for substrates with larger substituents (Table 2.1). BadA typically turned over 2-substituted substrates faster than the 3- and 4-substituted isomers.

The observation is consistent with the kinetics data for substrates other than the F-BzO series. Fluorine is isosteric to hydrogen (43, 49), and thus the steric demand of the F-BzO

app substrates is similar to BzO, as reflected in the similar KM values of BadA for 2- and 4-F-

BzO and BzO. But overall, the enzyme turnover and binding affinity are principally affected by the regiochemistry and steric effects of the substituents (Table 2.1). For instance, BadA only turned over 2-MeO-BzO presumably by placing the bulky towards the open channel in the active site. We imagine the bulky methoxy group clashed with Ala227 before and after rotation of the benzoyl moiety of the substrate. As a

65 result, the binding affinity and turnover of BadA were reduced for 2-MeO-BzO compared to BzO, and 3- and 4-MeO-BzO were not productive.

Attempts to co-crystallize BadA with 4-Me-BzO (98% purity) intriguingly resulted in a structure containing the 2-methyl isomer (not shown). This isomeric selection strongly supported the idea that reduced steric interactions in the active site around one ortho-carbon of the substrate enabled highly regiospecific binding of ortho-substituted substrates. It is

app worth noting, that 2-OH-BzO had the lowest KM of all substrates tested with BadA.

Evaluation of the BadA-BzO complex suggests that the 2-OH group may interact with the nearby hydroxyl of Thr329, creating a second binding contact (Figure 2.5D). In addition to interacting with the carboxylate of the bound BzO, Thr329 would provide an additional binding contact that may also affect release of the 2-OH-Bz CoA as shown by the faster release of the 3- and 4-OH-Bz CoA products.

Similarly, the regiochemistries of the splayed, planar NO2 group and the extended, linear CN substituent likely prevented 3-NO2-, 3-CN-, the 4-NO2-, and 4-CN-BzO from binding and being turned over by BadA. As with the bulky 2-MeO-BzO substrate, the 2-

NO2- and -CN-BzO were productive and converted slowly to their CoA thioesters by BadA

(Table 2.1). The linear 2-CN substituent extends into the open channel of the active site

app (Figure 2.5, A and D) (43). The kcat for 2-CN-BzO, however, is lower than nearly all of the 2-substituted BzO substrates with smaller substituents, and this observation is consistent with the steric effects of Ala227 affecting catalysis as the substrate pivots during

app adenylation. To illustrate further, ortho-substituted substrates 2-CN-BzO (KM = 213 µM)

app and 2-Cl-BzO (KM = 130 µM) bind BadA similarly. However, the estimated covalent van der Waals length of cyano (CAr-CN, dw = 3.1 Å) extends further than that of chloro (CAr-

66 app Cl, dw = 2.5 Å) (ChemBioDraw, ver. 13.0), and this difference may have reduced the kcat of BadA for 2-CN-BzO by 200-fold compared with that for 2-Cl-BzO.

In summary, the size and position of the substituent on a BzO analog affect the relative rates of reactivity significantly more than the electronics of the substituent. BadA is more forgiving of ortho-substituents than para-substituents, consistent with the relatively open channel in the BadA structures next to the ortho-carbon of the BzO substrate. In contrast, substituents at the meta- and para-positions significantly impact activity, except for fluorinated substituents, which are isosteric with hydrogen and thus have no additional steric requirements.

2.4.3.1 Non-aromatic carbocycle carboxylates

The Cyc substrate bound BadA the least among the non-aromatic carbocycles likely because the staggered ring conformation and additional hydrogens clashed with active site

app residues. Estimated by KM , substrates 1-Cyc and 3-Cyc have ~12 to 13-fold more apparent affinity for BadA than cyclohexane (Table 2.1). The planarity of the double bond in these mono-unsaturated carbocycles likely reduced steric interfaces between the ring and active site residues within the BadA active site. The turnover of 1-Cyc was lower than that for

Cyc (Table 2.1) while the placement of the double bond in 3-Cyc strongly increased catalytic turnover. The double bond positioned between the meta- and para-carbons of the ring in the latter removes two hydrogen atoms from the substrate that binds in a sterically crowded area near Ala227. Decreasing this steric hindrance likely explains the higher

app turnover and lower KM of 3-Cyc relative to Cyc and 1-Cyc (Table 2.1).

67 2.4.4 Analysis of point mutants of BadA

Mutational analyses also supported the substrate rotation mechanism in the BadA active site during catalysis. The four mutants Ala227Gly-BadA, Ile334Ala-BadA, His333Ala-

BadA, and Leu332Ala-BadA in this study were designed to reduce the steric-clash encountered by the substituents during substrate binding and rotation to form Bz-AMP within BadA. For example, the proposed 60° rotation causes the ortho- and meta-carbons of the initially bound BzO to move to the former locations of the meta-and para-carbons, respectively. Support for substrate rotation arises from a representative Ala227Gly-BadA mutation that reduced the steric interactions near the ortho- and meta-carbons of BzO. This mutant showed increased activity over BadA for ortho-substituted substrates that were imagined to bind and rotate within a sterically more favorable active site. This same mutant showed low but novel activities with meta-substituted BzO substrates, suggesting that the

Ala227Gly mutation enabled meta-substituted substrates to now bind and rotate to a catalytically competent adenylation orientation (Table 2.2).

Collectively, the Ala227Gly-BadA, Ile334Ala-BadA, and His333Ala-BadA mutants turned over new substrates (3-MeO, 3-CN, 3-NO2, 4-MeO, 4-NO2, 4-Ethyl, 3,4-

DiCH3, and, surprisingly, phenylacetate (PhAc), a precursor of penicillin G) that BadA could not (Table 2.2). Ile334Ala-BadA and His333Ala-BadA were active with nearly all of the 16 substrates tested, including PhAc, making these mutants broadly active 2-, 3-, or

4-substituted benzoate CoA ligases. Ala227Gly-BadA uniquely turned over 2,6-DiCl-BzO.

The results of the point mutations of BadA again supported a steric argument for BadA reactivity as opposed to one for substituent electronic effects. It is unclear why the

Leu332Ala-BadA mutant, designed to reduce the steric interactions around the para-

68 carbon of the substrate, unexpectedly did not turn over the para-substituted analogs, and slowed the turnover of several non-natural substrates tested (Table 2.2). The peptide backbone carbonyl of Leu332 contributes to the active site architecture, while the Leu side chain is oriented away from the active site. The Leu side chain most likely engages in structural, hydrophobic interactions (Figure 2.5, A and B). Therefore, we suspect the

Leu332Ala mutation may have malformed the active site of BadA.

69 2.5 Conclusions

Structures of BadA in complex with aryl carboxylate or aryl carbonyl-AMP show a persistent thiolation conformation before adenylation or CoA binding. This pre-CoA- bound conformation of BadA deviates from available structures of bacterial aryl carboxylate CoA ligases in the protein data bank. This suggests that the enzyme dynamics of BadA may be unique among benzoate CoA ligases during the substrate binding, adenylation, and CoA-thioesterification steps. This suggests two possible subclasses of benzoate CoA ligases, i) those whose thiolation conformation is the resting state and therefore bind benzoate in this conformation, and ii) those whose resting state is the adenylation conformation, and bind benzoate in this conformation. Additional biochemical and structural analyses are required to further support this hypothesis. Further, similarity in the protein architecture and ligands bound, yet differences in the overall tertiary protein structure, aid in understanding the mechanisms of action. With this understanding, it was possible to design mutant constructs that were dramatically more permissive than the native protein. The Ala227Gly-BadA mutant improved catalysis primarily with ortho-substituted benzoates, while the Ile334Ala-BadA and His333Ala-BadA improved turned over with meta- and para-substituted benzoates over native BadA catalysis. These findings are important for making non-natural acyl CoA thioesters that can be used in a chassis engineered to express natural product pathways that make next-generation bioactive compounds. Expanding the BadA substrate specificity may also help engineer pathways for bioremediation where substrate specificity is often a bottleneck for strain development (50) against pollutants like polycyclic aromatic hydrocarbons (51) and polychlorinated biphenyls (52, 53). In addition, the BadA kinetic constants provide data

70 needed for modeling flux control analysis in synthetic biology and bioengineering (54-57), as well as metabolizing carboxylate-containing xenobiotic drugs in (24).

71

APPENDIX

72 APPENDIX

0.3

0.2

0.1

(nmol/min)

V

0.0

0 50 100 150 Benzoate (M)

Figure A.1.1. Michaelis-Menten plot of biosynthetic benzoyl CoA catalyzed by BadA (0.1 µg/mL) -1 from benzoate (BzO). Vmax = 0.25 ± 0.008 nmol/min, KM = 4.40 ± 0.65 µM, kcat = 27.5 ± 0.87 s , (n = 3).

0.15

0.10

0.05

(nmol/min)

V

0.00

0 100 200 300 Thio-2-C (M)

Figure A.1.2. Michaelis-Menten plot of biosynthetic thiophene-2-carbonyl CoA catalyzed by BadA (0.3 µg/mL) from thiophene-2-carboxylate (Thio-2-C). Vmax = 0.13 ± 0.07 nmol/min, KM = 37 ± 5.7 µM, -1 kcat = 4.8 ± 0.23 s , (n = 3).

73

0.05

0.04

0.03

0.02

(nmol/min)

V

0.01

0.00 0 50 100 150 200 250 3-Fur (M)

Figure A.1.3. Michaelis-Menten plot of biosynthetic 3-furoyl CoA catalyzed by BadA (0.1 µg/mL) -1 from 3-Furoate (3-Fur). Vmax = 0.058 ± 0.0043 nmol/min, KM = 70.8 ± 12.3µM, kcat = 6.74 ± 0.48 s , (n = 3).

0.40

0.35

0.30

0.25

0.20

0.15

(nmol/min)

V 0.10

0.05

0.00

0 500 1000 1500 2000 Cyc (M)

Figure A.1.4. Michaelis-Menten plot of biosynthetic cyclohexanoyl CoA catalyzed by BadA (0.5 µg/mL) from cyclohexane carboxylate (Cyc). Vmax = 0.62 ± 0.05 nmol/min, KM = 1466 ± 213 µM, kcat = 13.5 ± 1.07 s-1 (n = 3).

74

0.06

0.04

0.02

(nmol/min)

V

0.00

0 200 400 600 1-Cyc (M)

Figure A.1.5. Michaelis-Menten plot of biosynthetic 1-cyclohexen-1-oyl CoA catalyzed by BadA (0.1 µg/mL) from 1-cyclohexen-1-carboxylate (1-Cyc). Vmax = 0.051 ± 0.0022 nmol/min, KM = 52 ± 7.1 µM, -1 kcat = 5.6 ± 0.24 s . (n = 3).

0.6

0.4

0.2

(nmol/min)

V

0.0

0 100 200 3-Cyc (M)

Figure A.1.6. Michaelis-Menten plot of biosynthetic 3-cyclohexen-1-oyl CoA catalyzed by BadA (0.4 µg/mL) from 3-cyclohexen-1-carboxylate (3-Cyc). Vmax = 0.58 ± 0.04 nmol/min, KM = 56.2 ± 9.1 µM, kcat = 16 ± 1.03 s-1 (n = 3 for [S] = 10, 20, 35, 200, otherwise n = 2).

75

0.4

0.3

0.2

(nmol/min) 0.1

V

0.0

0 100 200 300 2-F-BzO (M)

Figure A.1.7. Michaelis-Menten plot of biosynthetic 2-fluorobenzoyl CoA catalyzed by BadA (0.1 µg/mL) from 2-fluorobenzoate (2-F-BzO). Vmax = 0.31 ± 0.0064 nmol/min, KM = 8.1 ± 0.91 µM, kcat = 34 ± 0.7 s-1, (n = 3 for [S] = 5, 15, 50, 75, 100, 200, otherwise n = 2).

0.4

0.3

0.2

(nmol/min) 0.1

V

0.0

0 200 400 600 800 1000 3-F-BzO (M)

Figure A.1.8. Michaelis-Menten plot of biosynthetic 3-fluorobenzoyl CoA catalyzed by BadA (0.1 µg/mL) from 3-fluorobenzoate (3-F-BzO). Vmax = 0.32 ± 0.013 nmol/min, KM = 73 ± 11 µM, kcat = 35 ± 1.45 s-1, (n = 3).

76

0.2

0.1

(nmol/min)

V

0.0

0 100 200 4-F-BzO (M)

Figure A.1.9. Michaelis-Menten plot of biosynthetic 4-fluorobenzoyl CoA catalyzed by BadA (0.1 µg/mL) from 4-fluorobenzoate (4-F-BzO). Vmax = 0.2 ± 0.006 nmol/min, KM = 6.6 ± 0.9 µM, kcat = 22 ± 0.66 s-1, (n = 3 for [S] = 20, 100, 150, otherwise n = 2).

0.04

0.03

0.02

(nmol/min)

V 0.01

0.00 0 250 500 750 1000 1250 2-Cl-BzO (M)

Figure A.1.10. Michaelis-Menten plot of biosynthetic 2-chlorobenzoyl CoA catalyzed by BadA (0.2 µg/mL) from 2-chlorobenzoate (2-Cl-BzO). Vmax = 0.04 ± 0.001 nmol/min, KM = 126 ± 12.3 µM, kcat = 2.21 ± 0.068 s-1, (n = 3).

77

0.006

0.004

(nmol/min) 0.002

V

0.000 0 200 400 600 800 1000 3-Cl-BzO (M)

Figure A.1.11. Michaelis-Menten plot of biosynthetic 3-chlorobenzoyl CoA catalyzed by BadA (0.2 µg/mL) from 3-chlorobenzoate (3-Cl-BzO). Vmax = 0.0061 ± 0.0002 nmol/min, KM = 289 ± 23 µM, kcat = 0.33 ± 0.011 s-1, (n = 3).

0.006

0.004

0.002

(nmol/min)

V

0.000

-250 0 250 500 750 1000 1250 4-Cl-BzO (M)

Figure A.1.12. Michaelis-Menten plot of biosynthetic 4-chlorobenzoyl CoA catalyzed by BadA (0.2 µg/mL) from 4-chlorobenzoate (4-Cl-BzO). Vmax = 0.0054 ± 0.00022 nmol/min, KM = 83.2 ± 11.1 µM, kcat = 0.29 ± 0.012 s-1, (n = 3).

78

0.4

0.3

0.2

(nmol/min)

V

0.1

0.0 0 50 100 150 200 250 2NH -BzO (M) 2 Figure A.1.13. Michaelis-Menten plot of biosynthetic 2-aminobenzoyl CoA catalyzed by BadA (1 µg/mL) from 2-aminobenzoate (2-NH2-BzO). Vmax = 0.36 ± 0.01 nmol/min, KM = 21 ± 2.3 µM, kcat = 3.9 ± 0.1 s-1, (n = 3).

1.0

0.8

0.6

0.4

(nmol/min)

V 0.2

0.0 0 250 500 750 1000 1250 3-NH -BzO (M) 2 Figure A.1.14. Michaelis-Menten plot of biosynthetic 3-aminobenzoate CoA catalyzed by BadA (5 µg/mL) from 3-aminobenzoate (3-NH2-BzO). Vmax = 0.82 ± 0.05 nmol/min, KM = 174 ± 32 µM, kcat = 1.8 ± 0.1 s-1, (n = 3).

79

0.125

0.100

0.075

0.050

(nmol/min) 0.025

V

0.000

-0.025 -500 0 500 1000 1500 2000 2500 4-NH -BzO (M) 2 Figure A.1.15. Michaelis-Menten plot of biosynthetic 4-aminobenzoyl CoA catalyzed by BadA (5 µg/mL) from 4-aminobenzoate (4-NH2-BzO). Vmax = 0.081 ± 0.00195 nmol/min, KM = 25.3 ± 3.01 µM, -1 kcat = 0.18 ± 0.0043 s , (n = 3).

0.08

0.06

0.04

(nmol/min) 0.02

V

0.00

0 100 200 300 400 500 2-OH-BzO (M)

Figure A.1.16. Michaelis-Menten plot of biosynthetic 2-hydroxybenzoyl CoA catalyzed by BadA (0.1 µg/mL) from 2-hydroxybenzoate (2-OH-BzO). Vmax = 0.058 ± 0.0009 nmol/min, KM = 1.6 ± 0.2 µM, kcat = 0.633 ± 0.01, (n = 3).

80

1.00

0.75

0.50

(nmol/min)

V 0.25

0.00 0 500 1000 1500 2000 2500 3-OH-BzO (M) Figure A.1.17. Michaelis-Menten plot of biosynthetic 3-hydroxybenzoyl CoA catalyzed by BadA (5 µg/mL) from 3-hydroxybenzoate (3-OH-BzO). Vmax = 0.81 ± 0.02 nmol/min, KM = 229 ± 18 µM, kcat = 1.8 ± 0.04 s-1, (n = 3).

0.5

0.4

0.3

0.2

(nmol/min)

V 0.1

0.0 0 250 500 750 1000 1250 4-OH-BzO (M) Figure A.1.18. Michaelis-Menten plot of biosynthetic 4-hydroxybenzoyl CoA catalyzed by BadA (5 µg/mL) from 4-hydroxybenzoate (4-OH-BzO). Vmax = 0.43 ± 0.02 nmol/min, KM = 160 ± 17 µM, kcat = 0.9 ± 0.04 s-1, (n = 3).

81

0.4

0.3

0.2

(nmol/min)

V 0.1

0.0

0 500 1000 1500 2-Me-BzO (M)

Figure A.1.19. Michaelis-Menten plot of biosynthetic 2-methylbenzoyl CoA catalyzed by BadA (3 µg/mL) from 2-methylbenzoate (2-Me-BzO). Vmax = 0.33 ± 0.012 nmol/min, KM = 44 ± 6 µM, kcat = 1.2 ± 0.042 s-1, (n = 3).

0.15

0.10

0.05

(nmol/min)

V

0.00

-500 0 500 1000 1500 2000 2500 3-Me-BzO (M) Figure A.1.20. Michaelis-Menten plot of biosynthetic 3-methylbenzoyl CoA catalyzed by BadA (3 µg/mL) from 3-methylbenzoate (3-Me-BzO). Vmax = 0.15 ± 0.0050 nmol/min, KM = 215 ± 25.6 µM, kcat = 0.54 ± 0.018 s-1, (n = 3).

82

0.3

0.2

0.1

(nmol/min)

V

0.0

-500 0 500 1000 1500 2000 2500 4-Me-BzO (M)

Figure A.1.21. Michaelis-Menten plot of biosynthetic 4-methylbenzoate CoA catalyzed by BadA (5 µg/mL) from 4-methylbenzoate (4-Me-BzO). Vmax = 0.32 ± 0.016 nmol/min, KM = 590 ± 73 µM, kcat = 0.7 ± 0.035 s-1, (n = 3).

0.04

0.02

(nmol/min)

V

0.00

-500 0 500 1000 1500 2000 2500 2-CN-BzO (M)

Figure A.1.22. Michaelis-Menten plot of biosynthetic 2-cyanobenzoyl CoA catalyzed by BadA (5 µg/mL) from 2-cyanobenzoate (2-CN-BzO). Vmax = 0.050 ± 0.0016 nmol/min, KM = 213 ± 22 µM, kcat = 0.11 ± 0.0034 s-1, (n = 3).

83

0.020

0.015

0.010

(nmol/min) 0.005

V

0.000

-500 0 500 1000 1500 2000 2500 2-MeO-BzO (M)

Figure A.1.23. Michaelis-Menten plot of biosynthetic 2-methoxybenzoyl CoA catalyzed by BadA (20 µg/mL) from 2-methoxybenzoate (2-MeO-BzO). Vmax = 0.021 ± 0.00096 nmol/min, KM = 789 ± 82 µM, -1 kcat = 0.011 ± 0.00052 s , (n = 3).

0.4

0.3

0.2

(nmol/min) 0.1

V

0.0

-1000 0 1000 2000 3000 4000 5000 2-NO BzO (M) 2 Figure A.1.24. Michaelis-Menten plot of biosynthetic 2-nitrobenzoyl CoA catalyzed by BadA (20 µg/mL) from 2-nitrobenzoate (2-NO2-BzO). Vmax = 0.5 ± 0.02 nmol/min, KM = 2015 ± 167 µM, kcat = 0.27 ± 0.011 s-1, (n = 3).

84

Figure A.1.25. Crystals of Rhodopseudomonas palustris benzoate CoA ligase (BadA in complex with BzO).

Figure A.1.26. Overlay of BadA structures in complex with bound ligands. Substrate bound enzymes are as follows BzO (tan), 2-F-BzO (purple), 2-Me-BzO (salmon), 3-Fur (green), Bz-AMP (blue), and Thio- 2-C (pink) (Left panel). Active site magnified to show overlay of bound substrates (Right panel).

85

0.0015

0.0010

0.0005

(nmol/min)

V

0.0000

0 100 200 300 Benzoate (M)

Figure A.1.27. Michaelis-Menten plot of benzoyl CoA catalyzed by the Lys427Ala-BadA mutant. Enzyme was (15 µg/mL) with the benzoate (BzO) substrate. Vmax = 0.0011 ± 0.000025 nmol/min, KM = -1 1.5 ± 0.31 µM, kcat = 0.05 ± 0.0011 s , (n = 4).

86 50 A. AMP ATP 40 Bz-AMP 30 O

20 AMP

(Arbitrary Units) (Arbitrary

258 10 A 5.8 min

0 0 2 4 6 8 Time (min) 50 B. AMP 40 ATP

30 O

20 AMP

(Arbitrary Units) (Arbitrary

258 10 5.8 min

A 0 0 2 4 6 8 Time (min) 50 C. ATP 40

30

20

(Arbitrary Units) (Arbitrary

258 10

A

0 0 2 4 6 8 Time (min) D. 50 ATP 40

30

20

(Arbitrary Units) Units) (Arbitrary

258 10

A

0 0 2 4 6 8 Time (min) E.

Figure A.1.28. Benzoyl AMP (Bz-AMP) produced by BadA and the Lys427Ala-BadA mutant. A. Lys427Ala-BadA mutant and separately B. wild-type BadA with 1 mM benzoate (BzO) and 1 mM ATP at 31 °C for 45 min. Negative control assays C. incubated with 1 mM BzO, 1 mM ATP and no enzyme, D. BadA and 1 mM ATP, and E. Lys427Ala-BadA mutant and 1 mM ATP. Assays contained 250 µg/mL ligase, and the product mixtures were analyzed by reverse-phase HPLC with A258-monitoring of the effluent. The Bz-AMP eluted at Rt = 5.8 min. F. Tandem mass spectrometry (MS/MS, negative ion mode) fragmentation of the biosynthetic Bz-AMP ([M - H]– = m/z 450) fraction isolated from the reverse-phase HPLC after incubation of the Lys427Ala-BadA mutant with 1 mM BzO and 1 mM ATP (see panel A). The MS/MS fragmentation was identical for the biosynthetic Bz-AMP fraction isolated from the reverse- phase HPLC after incubation of BadA with 1 mM BzO and 1 mM ATP (see panel B).

87 Figure A.1.28 (cont’d).

328 F. 100

80

60

40 134

Rel. Ion Abundance Rel. Ion 20 257 80 192 0 100 200 300 400 500 m/z

Figure A.1.29. Shifted position of Tyr432 in BadA bound with bzAMP. The Tyr432 residue is shifted toward the phosphate oxygen (phosphorus: orange; phosphate oxygen: red) of the AMP anhydride and away from Arg421 to accommodate the Bz-AMP.

88

BCLM ...//...176 KAAATGCDDIAFWLYSSGSTGKPKGTVHTHANLYWTAELYAKPILGIAE--NDVVFSAAK 233 BCLC ...//...177 KAVASGCDDIAFWLYSSGSTGKPKGTVHTHANLYWTAELYAKAILGIVE--QDVVFSAAK 234 BadA ...//...166 KPAATQADDPAFWLYSSGSTGRPKGVVHTHANPYWTSELYGRNTLHLRE--DDVCFSAAK 223 4CBL ...//...146 EDPQREPAQPAFIFYTSGTTGLPKAAIIPQR-AAESRVLFMSTQVGLRHGRHNVVLGLMP 204

BCLM LFFAYGLGNGLTFPLSVGATAILMAERPTADAIFARLVE----HRPTVFYGVPTLYANML 289 BCLC LFFAYGLGNGLTFPLSVGATAVLMAERPTPDAIFARLTR----HKPTVFYGVPTLYASML 290 BadA LFFAYGLGNALTFPMTVGATTLLMGERPTPDAVFKRWLGGVGGVKPTVFYGAPTGYAGML 283 4CBL LYHVVGFFAVLVAALALDGTYVVIEEFRPVDALQLVQQE-----QVTSLFATPTHLDALA 259

BCLM VSPNLP--ARADVAIRICTSAGEALPREIGERFTAHFGCEILDGIGSTEMLHIFLSNRAG 347 BCLC ASPNVP--AREDLALRICTSAGEALPREIGERFTARFGAEILDGIGSTEMLHIFLSNRAG 348 BadA AAPNLP--SRDQVALRLASSAGEALPAEIGQRFQRHFGLDIVDGIGSTEMLHIFLSNLPD 341 4CBL AAAAHAGSSLKLDSLRHVTFAGATMPDAVLETVHQHLPGEKVNIYGTTEAMNSLYMRQPK 319

BCLM ...//...466 GLVKTRAFVVLKREFAPSEILAEELKAFVKDR-LAPHKYPRDIVFVDDLPKTATGKIQRF 524 BCLC ...//...467 GLVKTRAFVALKGEFVASDALADELKAFVKGR-LAPHKYPRDIVFVDDLPKTATGKIQRF 525 BadA ...//...461 GLTKPKAYVVPR---PGQTLSETELKTFIKDR-LAPYKYPRSTVFVAELPKTATGKIQRF 516 4CBL ...//...440 WGQSVTACVVPR---LGETLSADALDTFCRSSELADFKRPKRYFILDQLPKNALNKVLRR 496

BCLM KLREQ--- 529 BCLC KLREQL-- 531 BadA KLREGVLG 524 4CBL QLVQQVSS 504

Figure A.1.30. Partial amino acid sequence alignment of selected benzoate CoA ligases. BCLM and BCLC from B. xenovorans and BadA from R. palustris (active site residues are highlighted in black), and a 4-chlorobenzoate coenzyme A ligase (4-CBAL, listed as 4CBL), from Alcaligenes sp. Strain AL3007 (active site residues are highlighted in gray) are shown. The alignment was calculated with ClustalW2.

Figure A.1.31. Comparison of the position of Phe226 of BadA (in the thiolation conformation) and that of Phe236 of BCLM (in the adenylation conformation) bound to benzoate. The phenylalanine residue rotates into the CoA binding site acting as a gate. In the adenylation conformation, the phenylalanine residue (BCLM) blocks the CoA binding site. In the thiolation conformation, the phenylalanine residue (BadA) rotates out of the active site, opening the channel.

89

BadA YVRNDDGSYTYAGRTDDMLKVSGIYVSPFEIEATLVQHPGVLEAAVVGVADEHGLTKPKA 467 BCLm YCRLPNGCYVYAGRSDDMLKVSGQYVSPVEVEMVLVQHDAVLEAAVVGVD-HGGLVKTRA 472 bACS ARRDEDGYYWITGRVDDVLNVSGHRLGTAEIESALVAHPKIAEAAVVGIPHAIKGQAIYA 561 yACS AAKDKDGYIWILGRVDDVVNVSGHRLSTAEIEAAIIEDPIVAECAVVGFNDDLTGQAVAA 570 4CBL AVWTPEGTVRILGRVDDMIISGGENIHPSEIERVLGTAPGVTEVVVIGLADQRWGQSVTA 446 DhbE VRLTRDGYIVVEGRAKDQINRGGEKVAAEEVENHLLAHPAVHDAAMVSMPDQFLGERSCV 474 PheA ARWLSDGNIEYLGRIDNQVKIRGHRVELEEVESILLKHMYISETAVSVHKDHQEQPYLCA 474 DltA -GYVENGLLFYNGRLDFQIKLHGYRMELEEIEHHLRACSYVEGAVIVPIKKGEKYDYLLA 443 :* ** . : * : *:* : : .: .

BadA YVVPRPG---QTLSE------TELKTFIKDRLAPYKYPRSTVFVAELPKTATGKIQRF 517 BCLm FVVLKREFAPSEILA------EELKAFVKDRLAPHKYPRDIVFVDDLPKTATGKIQRFK 525 bACS YVTLNHGEEPSP------ELYAEVRNWVRKEIGPLATPDVLHWTDSLPKTRSGKIMRRI 614 yACS FVVLKNKSSWSTATDDELQDIKKHLVFTVRKDIGPFAAPKLIILVDDLPKTRSGKIMRRI 630 4CBL CVVPRLGETLSADAL------DTFCRSSELADFKRPKRYFILDQLPKNALNKVLRRQ 497 DhbE FIIPRDEAPKAA------ELKAFLRERGLAAYKIPDRVEFVESFPQTGVGKVSKKA 524 PheA YFVSEKHIPL------EQLRQFSSEELPTYMIPSYFIQLDKMPLTSNGKIDRKQ 522 DltA VVVPGEHSFEKEF------KLTSAIKKELNERLPNYMIPRKFMYQSSIPMTPNGKVDRKK 497 . : * .:* . *: :

Figure A.1.32. Partial amino acid sequence alignment of a broad selection of ATP-dependent adenylases and coenzyme A ligases. A portion of the C-terminal domain is shown to illustrate the complete conservation of Lys512 in BadA (magenta) and the partial conservation of Lys427 in BadA (cyan). BadA: benzoate coenzyme A ligase from Rhodopseudomonas palustris, BCLM: benzoate coenzyme A ligase from Burkholderia xenovorans LB400, bACS: Acetyl coenzyme A synthetase from Salmonella enterica, yACS: Acetyl coenzyme A synthetase from Saccharomyces cerevisieae, 4CBL: 4- chlorobenzoate coenzyme A ligase from Alcaligenes sp. AL3007, DhbE: 2,3-Dihydroxybenzoate adenylase from Bacillus subtilis, PheA: phenylalanine activating domain from gramicidin synthetase 1 from Brevibacillus brevis, DltA: D-alanyl carrier protein ligase from Bacillus cereus. Structures were aligned by Clustal Omega (EMBL-EBI).

Figure A.1.33. Relative position of benzoate and benzoyl AMP in the BadA active site. Black dot on BzO (cyan) highlights the ortho-carbon position near the void in the BadA active site where the phosphoryl group of Bz-AMP (yellow) resides.

90

Table A.1.1. X-ray Crystallography Data and Refinement Statistics. 2-Thiophene 2-Methyl 2-Fluoro- Benzoyl- Substrate Carboxylic 2-Furoic Acid Benzoic acid Benzoic acid Benzoic acid AMP Acid

Space group P21 P21 P21 P21 P21 P21

a (Å) 58.659 58.644 58.637 58.885 58.650 58.615

b (Å) 94.808 94.609 95.110 95.749 96.02 95.611

c (Å) 95.346 95.701 95.330 98.605 95.38 95.706

 (°) 90 90 90 90 90 90

 (°) 104.92 104.92 104.88 110.43 104.65 104.53

γ (°) 90 90 90 90 90 90

Molecules per 2 2 2 2 2 2 Asymmetric Unit

Total reflection 442714 541417 344579 389465 705943 430613

Unique 115478 99162 93457 92681 93581 112317 Reflection 96.9 97.5 98.54 97.82 98.6 96.34 Completeness (%) (74.3) (80) (89.89) (80.81) (96.58) (81.37) 17.1 21.84 16.91 12.68 21.7 17.9 Average I/ (1.35) (1.97) (1.61) (1.55) (2.73) (2.0) 7.7 9.70 11.1 10.5 8.9 8.8 R (%) merge (44.3) (47.2) (57.0) (47.5) (63.7) (43.4)

Resolution (Å) 50-1.63 50-1.72 50-1.76 50-1.77 50-1.80 50-1.73 (Last Shell) (1.66-1.63) (1.77-1.72) (1.79-1.76) (1.81-1.77) (1.84-1.80) (1.73-1.70)

Rcryst/Rfree (%) 15.81/19.31 15.65 /19.37 18.58/22.93 15.91/20.17 15.5/19.1 15.37/19.2

RMSD From Ideal Values

Bond Length (Å) 0.0192 0.0181 0.022 0.0196 0.022 0.022

Bond Angle (°) 1.9244 1.9736 1.966 1.961 1.969 1.981

Average B factor 21.264 25.67 21.754 19.3 19.69 18.812

Number of water 577 603 632 738 655 647 molecules

PDB IDs 4RLQ 4RMN 4RM3 4RM2 4EAT 4ZJZ aValues in the parenthesis refer to the last resolution shell.

91

Table A.1.2. Calculated van der Waals Volume (Vw) 3 Substituent Vw (Å ) F 8.0 HO 8.9 NH2 11.8 CN 15.5 Cl 17.4 Me 19.9 NO2 23.0 MeO 25.1

92

REFERENCES

93 REFERENCES

1. Schmelz, S., and Naismith, J. H. (2009) Adenylate-forming enzymes, Curr. Opin. Struct. Biol. 19, 666-671. 2. Conti, E., Franks, N. P., and Brick, P. (1996) Crystal structure of firefly luciferase throws light on a superfamily of adenylate-forming enzymes, Structure 4, 287-298. 3. Schmelz, S., Kadi, N., McMahon, S. A., Song, L., Oves-Costales, D., Oke, M., Liu, H., Johnson, K. A., Carter, L. G., Botting, C. H., White, M. F., Challis, G. L., and Naismith, J. H. (2009) AcsD catalyzes enantioselective citrate desymmetrization in siderophore biosynthesis, Nat. Chem. Biol. 5, 174-182. 4. Storz, M. P., Brengel, C., Weidel, E., Hoffmann, M., Hollemeyer, K., Steinbach, A., Muller, R., Empting, M., and Hartmann, R. W. (2013) Biochemical and biophysical analysis of a chiral PqsD inhibitor revealing tight-binding behavior and enantiomers with contrary thermodynamic signatures, ACS Chem. Biol. 8, 2794- 2801. 5. Bonet, B., Teufel, R., Crüsemann, M., Ziemert, N., and Moore, B. S. (2015) Direct capture and heterologous expression of Salinispora natural product genes for the biosynthesis of enterocin, J. Nat. Prod. 78, 539-542. 6. Walker, K., Long, R., and Croteau, R. (2002) The final acylation step in Taxol biosynthesis: Cloning of the taxoid C13-side-chain N-benzoyltransferase from Taxus, P. Natl. Acad. Sci. U.S.A. 99, 9166-9171. 7. Walker, K., and Croteau, R. (2000) Taxol biosynthesis: Molecular cloning of a benzoyl-CoA : taxane 2α-O-benzoyltransferase cDNA from Taxus and functional expression in Escherichia coli, Proc. Natl. Acad. Sci. U.S.A. 97, 13591-13596. 8. Dudareva, N., Klempien, A., Muhlemann, J. K., and Kaplan, I. (2013) Biosynthesis, function and metabolic engineering of plant volatile organic compounds, New Phytol. 198, 16-32. 9. Ziehl, M., He, J., Dahse, H.-M., and Hertweck, C. (2005) Mutasynthesis of aureonitrile: An aureothin derivative with significantly improved cytostatic effect, Angew. Chem. Int. Ed. 44, 1202-1205. 10. He, J., and Hertweck, C. (2003) Iteration as programmed event during polyketide assembly; molecular analysis of the aureothin biosynthesis gene cluster, Chemistry & biology 10, 1225-1232. 11. Coleman, J. P., Hudson, L. L., McKnight, S. L., Farrow, J. M., 3rd, Calfee, M. W., Lindsey, C. A., and Pesci, E. C. (2008) Pseudomonas aeruginosa PqsA is an anthranilate-coenzyme A ligase, J. Bacteriol. 190, 1247-1255.

94 12. Geissler, J. F., Harwood, C. S., and Gibson, J. (1988) Purification and properties of benzoate-coenzyme A ligase, a Rhodopseudomonas palustris enzyme involved in the anaerobic degradation of benzoate, J. Bacteriol. 170, 1709-1714. 13. Egland, P. G., Gibson, J., and Harwood, C. S. (1995) Benzoate-coenzyme A ligase, encoded by BadA, is one of 3 ligases able to catalyze benzoyl-coenzyme A formation during anaerobic growth of Rhodopseudomonas palustris on benzoate, J. Bacteriol. 177, 6545-6551. 14. Bains, J., and Boulanger, M. J. (2007) Biochemical and structural characterization of the paralogous benzoate CoA Ligases from Burkholderia xenovorans LB400: Defining the entry point into the novel benzoate oxidation (box) pathway, J. Mol. Biol. 373, 965-977. 15. Chang, K. H., Xiang, H., and Dunaway-Mariano, D. (1997) Acyl-adenylate motif of the acyl-adenylate/thioester-forming enzyme superfamily: a site-directed mutagenesis study with the Pseudomonas sp. strain CBS3 4- chlorobenzoate:coenzyme A ligase, Biochemistry-US 36, 15650-15659. 16. Schuhle, K., Gescher, J., Feil, U., Paul, M., Jahn, M., Schagger, H., and Fuchs, G. (2003) Benzoate-coenzyme A ligase from Thauera aromatica: an enzyme acting in anaerobic and aerobic pathways, J. Bacteriol. 185, 4920-4929. 17. Carmona, M., Zamarro, M. T., Blazquez, B., Durante-Rodriguez, G., Juarez, J. F., Valderrama, J. A., Barragan, M. J., Garcia, J. L., and Diaz, E. (2009) Anaerobic catabolism of aromatic compounds: A genetic and genomic view, Microbiol. Mol. Biol. Rev. 73, 71-133. 18. Fuchs, G. (2008) Anaerobic metabolism of aromatic compounds, Ann. N. Y. Acad. Sci. 1125, 82-99. 19. Seo, J. S., Keum, Y. S., and Li, Q. X. (2009) Bacterial degradation of aromatic compounds, Int. J. Env. Res. Public Health 6, 278-309. 20. Berg, P. (1956) Acyl adenylates; an enzymatic mechanism of acetate activation, J. Biol. Chem. 222, 991-1013. 21. Reger, A. S., Wu, R., Dunaway-Mariano, D., and Gulick, A. M. (2008) Structural characterization of a 140 degrees domain movement in the two-step reaction catalyzed by 4-chlorobenzoate:CoA ligase, Biochemistry-US 47, 8016-8025. 22. Muchiri, R., and Walker, K. D. (2012) Taxol biosynthesis: Tyrocidine synthetase A catalyzes the production of phenylisoserinyl CoA and other amino phenylpropanoyl thioesters, Chem. Biol. 19, 679-685. 23. May, J. J., Kessler, N., Marahiel, M. A., and Stubbs, M. T. (2002) Crystal structure of DhbE, an archetype for aryl acid activating domains of modular synthetases, P. Natl. Acad. Sci. U.S.A. 99, 12120-12125.

95 24. Kochan, G., Pilka, E. S., von Delft, F., Oppermann, U., and Yue, W. W. (2009) Structural snapshots for the conformation-dependent catalysis by human medium- chain acyl-coenzyme A synthetase ACSM2A, J. Mol. Biol. 388, 997-1008. 25. Gulick, A. M., Lu, X., and Dunaway-Mariano, D. (2004) Crystal structure of 4- chlorobenzoate:CoA ligase/synthetase in the unliganded and aryl substrate-bound states, Biochemistry-US 43, 8670-8679. 26. Reger, A. S., Carney, J. M., and Gulick, A. M. (2007) Biochemical and crystallographic analysis of substrate binding and conformational changes in acetyl-CoA synthetase, Biochemistry-US 46, 6536-6546. 27. Hu, Y., Gai, Y., Yin, L., Wang, X., Feng, C., Feng, L., Li, D., Jiang, X. N., and Wang, D. C. (2010) Crystal structures of a Populus tomentosa 4-coumarate:CoA ligase shed light on its enzymatic mechanisms, The Plant cell 22, 3093-3104. 28. Conti, E., Stachelhaus, T., Marahiel, M. A., and Brick, P. (1997) Structural basis for the activation of phenylalanine in the non-ribosomal biosynthesis of gramicidin S, EMBO Journal 16, 4174-4183. 29. Berg, P. (1956) Acyl adenylates; the synthesis and properties of adenyl acetate, J. Biol. Chem. 222, 1015-1023. 30. Crosby, H. A., Heiniger, E. K., Harwood, C. S., and Escalante-Semerena, J. C. (2010) Reversible N-lysine acetylation regulates the activity of acyl-CoA synthetases involved in anaerobic benzoate catabolism in Rhodopseudomonas palustris, Mol. Microbiol. 76, 874-888. 31. Beuerle, T., and Pichersky, E. (2002) Purification and characterization of benzoate:coenzyme A ligase from Clarkia breweri, Arch. Biochem. Biophys. 400, 258-264. 32. Gaid, M. M., Scharnhop, H., Ramadan, H., Beuerle, T., and Beerhues, L. (2011) 4- Coumarate:CoA ligase family members from elicitor-treated Sorbus aucuparia cell cultures, J. Plant Physiol. 168, 944-951. 33. Otwinowski, Z., and Minor, W. (1997) Processing of X-ray diffraction data collected in oscillation mode, Method. Enzymol. 276, 307-326. 34. Arnold, K., Bordoli, L., Kopp, J., and Schwede, T. (2006) The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling, Bioinformatics 22, 195-201. 35. Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., Powell, H. R., Read, R. J., Vagin, A., and Wilson, K. S. (2011) Overview of the CCP4 suite and current developments, Acta Crystallogr. Sect. D. Biol. Crystallogr. 67, 235-242.

96 36. Vagin, A., and Teplyakov, A. (1997) MOLREP: an automated program for molecular replacement, J. Appl. Crystallogr. 30, 1022-1025. 37. Vagin, A. A., Steiner, R. A., Lebedev, A. A., Potterton, L., McNicholas, S., Long, F., and Murshudov, G. N. (2004) REFMAC5 dictionary: organization of prior chemical knowledge and guidelines for its use, Acta Crystallogr. Sect. D. Biol. Crystallogr. 60, 2184-2195. 38. Emsley, P., Lohkamp, B., Scott, W. G., and Cowtan, K. (2010) Features and development of COOT, Acta Crystallogr. Sect. D. Biol. Crystallogr. 66, 486-501. 39. Lebedev, A. A., Young, P., Isupov, M. N., Moroz, O. V., Vagin, A. A., and Murshudov, G. N. (2012) JLigand: a graphical tool for the CCP4 template-restraint library, Acta Crystallogr. Sect. D. Biol. Crystallogr. 68, 431-440. 40. Connolly, M. L. (1993) The molecular surface package, J. Mol. Graphics 11, 139- 141. 41. Schrodinger, LLC. (2010) The PyMOL Molecular Graphics System, Version 1.3r1. 42. Pettersen, E., Goddard, T., Huang, C., Couch, G., Greenblatt, D., Meng, E., and Ferrin, T. (2004 ) UCSF Chimera--a visualization system for exploratory research and analysis J. Comput. Chem. 25, 1605-1612. 43. Batsanov, S. S. (2001) Van der Waals Radii of Elements, Inorganic Materials 37, 871-885. 44. Gulick, A. M., Starai, V. J., Horswill, A. R., Homick, K. M., and Escalante- Semerena, J. C. (2003) The 1.75 A crystal structure of acetyl-CoA synthetase bound to adenosine-5 '-propylphosphate and coenzyme A, Biochemistry-US 42, 2866- 2873. 45. Wu, R., Reger, A. S., Lu, X., Gulick, A. M., and Dunaway-Mariano, D. (2009) The mechanism of domain alternation in the acyl-adenylate forming ligase superfamily member 4-chlorobenzoate: coenzyme A ligase, Biochemistry-US 48, 4115-4125. 46. Crosby, H. A., Pelletier, D. A., Hurst, G. B., and Escalante-Semerena, J. C. (2012) System-wide studies of N-lysine acetylation in Rhodopseudomonas palustris reveal substrate specificity of protein acetyltransferases, J. Biol. Chem. 287, 15590-15601. 47. Gardner, J. G., Grundy, F. J., Henkin, T. M., and Escalante-Semerena, J. C. (2006) Control of acetyl-coenzyme A synthetase (AcsA) activity by acetylation/deacetylation without NAD(+) involvement in Bacillus subtilis, J. Bacteriol. 188, 5460-5468. 48. Du, L., He, Y., and Luo, Y. (2008) Crystal structure and enantiomer selection by D-alanyl carrier protein ligase DltA from Bacillus cereus, Biochemistry 47, 11473- 11480.

97 49. Rowland, R. S., and Taylor, R. (1996) Intermolecular nonbonded contact distances in organic crystal structures: Comparison with distances expected from van der Waals radii, J. Phys. Chem. 100, 7384-7391. 50. Xun, L. (2012) Microbial Degradation of Polychlorophenols, In Environmental Science and Engineering (Singh, S. N., Ed.), pp 1-30, Springer, New York. 51. Flanagan, P. V., Kelleher, B. P., and Allen, C. C. R. (2013) Assessment of anaerobic biodegradation of aromatic hydrocarbons: The impact of molecular biology approaches, Geomicrobiol. J. 31, 276-284. 52. Lauby-Secretan, B., Loomis, D., Grosse, Y., Ghissassi, F. E., Bouvard, V., Benbrahim-Tallaa, L., Guha, N., Baan, R., Mattock, H., and Straif, K. Carcinogenicity of polychlorinated biphenyls and polybrominated biphenyls, Lancet Oncol. 14, 287-288. 53. Passatore, L., Rossetti, S., Juwarkar, A. A., and Massacci, A. (2014) Phytoremediation and bioremediation of polychlorinated biphenyls (PCBs): State of knowledge and research perspectives, J. Hazard. Mater. 278, 189-202. 54. Nawarathne, I. N., and Walker, K. D. (2010) Point mutations (Q19P and N23K) increase the operational solubility of a 2--O-benzoyltransferase that conveys various acyl groups from CoA to a taxane acceptor, J. Nat. Prod. 73, 151-159. 55. Ajikumar, P. K., Xiao, W. H., Tyo, K. E. J., Wang, Y., Simeon, F., Leonard, E., Mucha, O., Phon, T. H., Pfeifer, B., and Stephanopoulos, G. (2010) Isoprenoid pathway optimization for Taxol precursor overproduction in Escherichia coli, Science 330, 70-74. 56. Jeya, M., Kim, T. S., Tiwari, M. K., Li, J., Zhao, H., and Lee, J. K. (2012) The Botrytis cinerea type III shows unprecedented high catalytic efficiency toward long chain acyl-CoAs, Mol. Biosyst. 8, 2864-2867. 57. Yu, Z.-L., Liu, J., Wang, F.-Q., Dai, M., Zhao, B.-H., He, J.-G., and Zhang, H. (2011) Cloning and characterization of a novel CoA-ligase gene from Penicillium chrysogenum, Folia Microbiol. 56, 246-252.

98 Chapter 3. Expression and purification of the Taxus cuspidata baccatin III: 3- amino-13-O-phenylpropanoyl transferase (BAPT)

3.1 Introduction

Biosynthesis of the anticancer drug paclitaxel requires multiple acyltransferases in the

BAHD enzyme superfamily, named for the first four discovered enzymes: a benzylalcohol

O-acetyltransferase (BEAT), an anthocyanin O-hydroxycinnamoyl-transferase (AHCT), an anthranilate N-hydroxycinnamoyl/benzoyltransferase (HCBT), and deacetylvindoline

4-O-acetyltransferase (DAT) (1). BAHD enzymes catalyze aliphatic or aromatic acylation of N- or O-acyl acceptor molecules in plant and fungal secondary metabolism pathways

(Figure 3.2). These pathways include the modification of phenolic compounds for biosynthesis (2), anthocyanin (3-5), volatile compounds (6, 7), and phytoalexin (8) biosynthesis among others (1, 9, 10). BAHD enzymes range in size from 48 to 55 kDa, have low sequence identity (~25-34 %) between members, with diverse substrate specificities (1, 10) (Figure 3.2). Despite vastly different substrate specificities, BAHD enzymes contain a number of conserved motifs including the catalytic HXXXD motif in the active site and the structural DFGWG motif in the C-terminal half of the enzyme

(Figure 3.3). Elucidation of the first BAHD crystal structure of vinorine synthase (and subsequent crystal structure) clarified the role of these highly conserved motifs (5, 11-14).

The conserved histidine of the HXXXD motif is proposed to act as a general base to deprotonate the acyl acceptor for nucleophilic attack on the acyl donor and the conserved aspartate orients away from the active site performing a structural role (11) (Figure 3.1).

99

Figure 3.1. Proposed mechanism of BAHD acyltransferases. I. The conserved histidine in the HXXXD motif acts as a general base upon the acyl acceptor, which nucleophilically attacks the carbonyl of the acyl donor thioester. II. A tetrahedral intermediate is formed which subsequently collapses. III. The catalytic histidine is restored to a deprotonated state and the acyl acceptor is acylated. Coenzyme A (CoA-SH) is released.

The conserved DFGWG motif is located in the C-terminal domain, away from the active site and appears to perform a structural role (11, 15). The determination of additional crystal structures showed conservation of the global structure of these enzymes, the functional conservation of the motifs HXXXD and DFGWD, and the variability between active site residues to accommodate vastly different acyl donor and acceptor groups (5, 12-

14). For example, while vinorine synthase crystallized in the apo form, the sorghum

(Sorghum bicolor) hydroxycinnamoyl transferase (HCT) crystallized in complex with the reaction products, p-coumaroyl shikimate and coenzyme A (CoA-SH) (Figure 3.3) (13). A large channel through the middle of the BAHD acyltransferase forms the active site. Both the acyl acceptor (shikimate) and the acyl donor (p-coumaroyl CoA) enter the channel and face each other in close proximity to the conserved histidine residue (His162 in HCT)

(Figure 3.3). The peptide backbones of all five crystallized BAHD enzymes align well with each other. Although these acyltransferases may bind dramatically different substrates, they appear to utilize the same binding sites and the conserved catalytic histidine residue.

100

Figure 3.2. Examples of BAHD enzyme products. The final acylation on each product is blue.

101 A.

B.

Figure 3.3. BAHD structure of the Sorghum hydroxycinnamoyltransferase (HCT). A. HCT structure (blue) is shown with conserved HXXXD motif (cyan) and DFGWG motif (yellow). The active site channel is delineated by a black circle. B. HCT structure bound with the p-coumaroyl shikimate and CoA-SH (magenta) products of the acyltransferase reaction. PDB: 4KEC

102 There are five BAHD acyltransferase steps in paclitaxel biosynthesis (Figure 3.4).

All five of these acyltransferases share 57-65 % sequence identity with each other (Figure

A.2.1). Several have been purified and biochemically characterized including taxadiene-

5α-ol-O-acetyltransferase (TAT) (16), taxane-2α-O-benzoyltransferase (TBT) (17, 18), 10- deacetylbaccatin III 10-O-acetyltransferase (DBAT) (19), baccatin III: 3-amino-13-O- phenylpropanoyl transferase (BAPT) (20), N-debenzoylpaclitaxol-N-benzoyltransferase

(NDTNBT) (21-23).

Figure 3.4. Condensed paclitaxel biosynthesis from geranylgeranyl diphosphate showing the BAHD acyltransferases in the pathway. TAT, taxadiene-5α-ol-O-acetyltransferase, TBT, taxane-2α-O- benzoyltransferase, DBAT, 10-deacetylbaccatin III 10-O-acetyltransferase, BAPT, baccatin III: 3-amino-13- O-phenylpropanoyl CoA transferase, NDTNBT, N-debenzoylpaclitaxol-N-benzoyltransferase. Acylations are shown in blue and the final 2′-hydroxylation performed by an unidentified 2′-hydroxylase (T2′OH) is shown in red. The larger arrow represents multiple reaction steps.

103 BAPT is more active with the pathway precursor, (3R)-β-phenylalanyl CoA ((3R)-β-Phe

CoA) acyl donor than the (2R,3S)-phenylisoserinyl CoA ((2R,3S)-Phe CoA) acyl donor.

Characterization of BAPT activity with (2R,3S)-phenylisoserinyl CoA is important for developing a condensed paclitaxel biosynthesis that bypasses the need to identify, characterize, and express the C2′-hydroxylase (T2′OH) (Figure 3.4). The substrate specificity of BAPT, particularly with aryl-substituted phenylisoserines, has never been investigated. A number of active analogs of paclitaxel have been chemically synthesized with modifications at this position (Table 3.1). Investigations into BAPT substrate specificity would show the potential to produce paclitaxel and docetaxel analogs of interest with modified C13-sidechains (20). Previous studies with BAPT used crude lysate and estimated BAPT concentrations at 1% of total protein. Estimates of the KM for BAPT with baccatin III and (3R)-β-phenylalanyl CoA ((3R)-β-Phe CoA) were determined at 2.4 µM and 4.9 µM respectively (20). The catalytic turnover of BAPT has never been determined.

Previous research with BAHD acyltransferases purified protein by standard overexpression and chromatography techniques. Vinorine synthase was expressed as a polyhistidine-tagged enzyme and purified by nickel-nitriloacetic acid (Ni2+-NTA) affinity chromatography, anion-exchange chromatography, and gel- chromatography (11,

24). The HCT enzyme was also expressed as a polyhistidine tagged protein and purified by nickel-affinity and cation-exchange chromatography at yields of 10.4 mg/L (13). Other

BAHD acyltransferases were also purified from bacteria, but in cases where BAHD expression in E. coli resulted in insoluble protein, the yeast Pichia pastoris was successfully used to express the enzyme (4, 25). In order to purify recombinant BAPT enzyme, both P. pastoris and E. coli were investigated for the recombinant expression of

104 soluble, active BAPT enzyme. This chapter describes the optimization of the expression and purification of BAPT. Expression and induction conditions are discussed, as well as various column chromatography purification methods. A coupled activity assay was also developed in order to follow BAPT activity during purification optimization due to the commercial unavailability of the substrate (3R)-β-Phe CoA.

105

Table 3.1. Biologically active analogs as active as paclitaxel with modifications in the C13- phenylisoserinyl sidechain.

+ R1 R2 R3 Biological Activity Assay Reference LCC61, LCC6-MDR2, MCF73, 2-methyl-prop-1-enyl cyclohexyl propanoyl NCI/ADR4 LCC6, LCC6-MDR, MCF7, 2-methylpropyl cyclohexyl propanoyl (26) NCI/ADR 2-methyl-prop-1-enyl t-butoxy propanoyl MCF7, NCI/ADR 2-methyl-prop-1-enyl t-butoxy c-PrCO MCF7, NCI/ADR cyclohexyl t-butoxy H P3885, M6 (27) trimethyl t-butoxy H B167, M (28) 4-chlorophenyl phenyl acetyl M, B16 (29) 4-methoxyphenyl phenyl acetyl M, B16 (30) 4-fluorophenyl phenyl acetyl M, B16 cyclohexyl phenyl acetyl M, B16 (31) 2-pyridyl phenyl acetyl M, B16 4-pyridyl phenyl acetyl M, B16 (32) 2-furyl phenyl acetyl M, B16 2-methyl-prop-1-enyl phenyl acetyl J774.18 4-fluorobenzyl t-butoxy acetyl J774.1 (33) 4-fluorobenzoyl t-butoxy H J774.1 +Activities with the same or better activity than paclitaxel 1LCC6 breast cancer cell line 2LCC6-MDR multidrug resistant breast cancer cell line (Pgp+) 3MCF7 breast cancer cell line 4NCI/ADR multidrug resistant human ovarian cancer cell line 5P388 leukemia cell line 6In vitro microtubule assembly assay 7B16 melanoma cell line 8J774.1 murine macrophage cell line.

106 3.2 Experimental

3.2.1 Materials

Chloramphenicol (99%) and HEPES (>99%) were obtained from Fluka/Sigma Aldrich (St.

Louis, MO). MOPS (>99%) was obtained from Research Products International, Corp. (Mt

Prospect, IL). ATP, DTT, IPTG, ampicillin, kanamycin, cobalt-affinity chromatography resin, and PMSF were purchased from Gold Bio (St. Louis, MO). The following reagents were purchased from New England Biolabs (Ipswich, MA); dNTPs, Phusion HF DNA polymerase, all restriction enzymes, and T4 DNA ligase. The QIAGEN Plasmid Mini Prep

Kit and Gel Extraction Kit were obtained from QIAGEN (Valencia, CA). The following reagents were purchased from Promega (Madison, WI); PureYield Plasmid Mini

Purification System and Wizard SV Gel and PCR Clean-Up System. Coenzyme A (95%) was obtained from Lee Biosolutions (St. Louis, MO). (3R)-β-phenylalanine (98%) was purchased from Peptech (Burlington, MA). EDTA-free protease inhibitor cocktail tablets were purchased from Roche Life Sciences (Indianapolis, IN). All taxanes including baccatin III (>98%) and docetaxel (>98%) were purchased from Natland International

Corporation (Research Triangle Park, NC). The PfuTurbo DNA Polymerase, zeocin, and

E. coli strain BL21(DE3) were sourced from Invitrogen (Carlsbad, CA), The pET28a expression vector, Rosetta(DE3), and Rosetta(DE3)(pLysS) came from (Novagen, EMD

Millipore, Billerica, MA),

3.2.2 BAPT purification from Pichia pastoris

3.2.2.1 Cloning of pHisBAPT and pMycBAPT

The bapt gene (Genbank: JF338879.1) was amplified with the following primers

HisBAPT-Fwd: 5’-CAC GAA TTC AAA AAT GTC TAT GAA GAA GAC AGG TTC-

107 3’ and HisBAPT-Rev: 5’-CTT CTC GAG TCA ATG ATG ATG ATG ATG ATG TAA

CTT TGA CGG ACA CAC TTT AG-3’. These primers introduced restriction sites (EcoRI and XhoI), a C-terminal 6X-histidine tag, and a onto the ends of the amplicon.

Products were amplified by polymerase chain reaction (PCR) using PfU Turbo DNA polymerase (Invitrogen, Carlsbad, CA) per the manufacturer’s instructions. PCR conditions used an initial 95 °C for 4 min denaturation step, followed by 34 cycles at 95

°C for 1 min, 60 °C for 1 min, 72 °C for 4 min, and a final extension at 72 °C for 10 min.

PCR products were analyzed by agarose gel electrophoresis (1% agarose in 1X TAE), double-digested with EcoRI and XhoI digested (New England Biolabs, Ipswich, MA), according to manufacturer’s instructions, and gel purified using the Qiagen Gel Extraction

Kit. The pPICZc plasmid (Invitrogen, Carlsbad, CA) was also double-digested and gel purified. The gel-purified EcoRI/XhoI cut PCR insert and pPICZc vector were ligated (3:1 insert: vector molar ratio) with T4 DNA ligase (16 hours at 14 °C) according to the manufacturer’s instructions. The ligation was used to transform DH5α E. coli (Invitrogen,

Carlsbad, CA) and grown on LB plates (16 hour, 37 °C) containing 25 µg/mL zeocin.

Positive clones were screened by PCR and analyzed by DNA sequencing (Michigan State

University Research Technology Support Facility (RTSF) Genomics Core). The amino acid sequence of HisBAPT is listed in (Figure A.2.2). The same cloning procedure was used to generate the pMycBAPT expression plasmid containing a C-terminal Myc epitope tag, followed by a polyhistidine tag. The pMycBAPT bapt insert was amplified with primers HisBAPT-Fwd (above) and MycBAPT-Rev: 5’-CTT CCG CGG TAA CTT TGA

CGG ACA CAC TT-3’ to add EcoRI and SacII restriction sites. The MycBapt amplicon was ligated in EcoRI and SacII digested pPICZc (Figure A.2.3).

108 3.2.2.2 BAPT expression in P. pastoris

Expression vectors pHisBAPT and pMycBapt were linearized with the PmeI restriction enzyme and transformed into P. pastoris strain X33 by electroporation according to the

EasySelect Pichia Expression Kit manufacturer’s instructions. Cells were plated on YPDS

(1% yeast extract, 2% peptone, 2% dextrose, 1M sorbitol) plates containing zeocin (100 and 1000 µg/mL) and grown at 30 °C. Colonies were also screened for growth on glycerol or methanol as a carbon source and subsequently tested for BAPT expression according to manufacturer’s instructions. Cells were kept on ice throughout lysis procedures. Lysis methods included vortexing with glass beads (0.5 mm), use of a Beadbeater (BioSpec) with glass beads (0.5 mm), or sonication with a Misonix XL 2020 sonicator. Lysate was centrifuged at 18000g for 10 minutes. The supernatant was then centrifuged at 100000g for

1 hour. The final supernatant (or clarified lysate, “CL”) was analyzed for expression by

SDS-PAGE, Western blot, and activity assay.

3.2.2.3 BAPT Western blot

Western blots were performed to detect BAPT protein containing an N-terminal or C- terminal poly-histidine tag. Samples were transferred from an SDS-PAGE gel onto a PVDF membrane (Immobilon) (12 hours at 4 °C, 30 V) gel in transfer buffer (25 mM Tris (pH

8.3), 192 mM , 10% (v/v) MeOH). The PVDF membrane, was carefully removed, soaked in methanol, dried, rinsed with water, and incubated with blocking solution (1%

(w/v) BSA in 10 mM phosphate buffer (pH7.2), 0.9% (w/v) NaCl) for one hour, incubated with mouse anti-6X His tag antibody conjugated to alkaline (1:5000 dilution)

(AbCam, Cat. No. ab81652, Cambridge, UK) in 1X TBS (50 mM Tris-Cl, (pH 7.6), 150 mM NaCl) buffer for 1 hour, rinsed with 1X TBS three times for 10 minutes each, then

109 color-developed by incubation with 1-Step NBT/BCIP (ThermoFisher Scientific,

Waltham, MA). Color development was stopped by rinsing with 1X TBS, followed by a

10 min incubation with 50 mM EDTA in 1X TBS.

3.2.3 BAPT purification from Escherichia coli

3.2.3.1 Cloning of pNterBAPT

The bapt gene was amplified (Pfu Turbo DNA polymerase) with the primers NterBAPT-

Fwd: 5’-GTT CAT ATG AAG AAG ACA GGT TCG TTT GCA GAG TT-3’ and

NterBAPT-Rev: 5’- GAT GGA TCC TCA TCA TAA CTT TGA CGG ACA CAC TT-3’.

PCR conditions used an initial 95 °C for 4 min denaturation step, followed by 34 cycles at

95 °C for 1 min, 57 °C for 1 min, 72 °C for 4 min, and a final extension at 72 °C for 10 min. PCR products were analyzed by agarose gel electrophoresis (1 % agarose in 1X TAE), and gel purified using the QIAGEN Gel Extraction Kit. The pET28a plasmid and PCR amplified insert were double-digested with NdeI and HindIII, followed by agarose gel electrophoresis and gel purification (as described above). The gel-purified PCR insert and

NdeI/HindIII cut pET28a vector were ligated and transformed as described above, into

BL21(DE3) E. coli (Invitrogen, Carlsbad, CA) on LB plates containing kanamycin (50

µg/mL). Positive clones were screened by PCR as described above. The amino acid sequence of expressed NterBAPT is listed in Figure A.2.4.

3.2.3.2 Cloning of pCterBAPT

Cloning of pCterBAPT was performed similarly to pNterBAPT, except for the following modifications. The gene was amplified by PCR with two primer sets. Both PCR products were then combined, boiled, and re-annealed to make sticky-ended PCR products (34).

Primers were as follows: PCR1: CterBAPT-Fwd1: 5’-AAG AAG ACA GGT TCG TTT

110 GCA GA-3’ and CterBAPT-Rev1: 5’-TTA ACT TTG ACG GAC ACA CTT TAG-3’ and

PCR2: CterBAPT-Fwd2: 5’-CAT GAA GAA GAC AGG TTC GTT TGC-3’ and

CterBAPT-Rev2: 5’-AGC TTT AAC TTT GAC GGA CAC ACT T-3’. PCR products were then combined to generate insert with compatible ends for cloning. PCR1 and 2 were combined and incubated at 95 °C for 5 minutes in a heat block, which was then removed and allowed to slowly cool to 40 °C. The PCR amplicon was then ligated into NcoI and

HindIII digested pET28a. The amino acid sequence of expressed CterBAPT is listed in

(Figure A.2.5).

3.2.3.3 Cloning of pOptBapt

The codon-optimized bapt gene (Figure A.2.6) was ordered from Genscript (Piscataway,

NJ), cloned in the pET28a expression vector for expression with a C-terminal polyhistidine tag. The amino acid sequence of expressed OptBAPT is listed in Figure A.2.6.

3.2.3.4 Cloning of pMBP-CterBAPT and pMBP-NterBAPT

To generate MBP-CterBAPT, the codon-optimized bapt gene (opt-bapt) was amplified by

PCR with two primer sets. Both PCR products were then combined, boiled, and re- annealed to make sticky-ended PCR products as described above (34). Primers were as follows: CterMBP-PCR1 (CterFwd1: 5’-GAT CCA TGG GCA AGA AAA CCG GT-3’,

CterRev1: 5’-TCA GCC GGA TCT CAG TGG T-3’) and CterMBP-PCR2 (CterFwd2: 5’-

CAT GGG CAA GAA AAC CGG T-3’, CterRev2: 5’-AGC TTC AGC CGG ATC TCA

GTG GT-3’). To generate MBP-NterBAPT, the non-codon optimized bapt gene (Nter- bapt) was made with the following primers: NterMBP-PCR1 (NterFwd1: 5’-GAT CCA

TGG GCA GCA GCC ATC AT-3’, NterRev1: 5’-TCC TCA TAA CTT TGA CGG ACA

C-3’) and NterMBP-PCR2 (NterFwd2: 5’-CAT GGG CAG CAG CCA TCA T-3’,

111 NterRev2: 5’- AGG TTC CTC ATA ACT TTG ACG GAC AC-3’). Products were amplified with Phusion HF DNA polymerase (New England Biolabs) per the manufacturer’s instructions. The PCR conditions used an initial 95 °C for 4 min denaturation step, followed by 34 cycles at 95 °C for 1 min, 57 °C for 1 min, 72 °C for 1.5 min, and a final extension at 72 °C for 7 min. PCR products were then combined to generate insert with compatible ends for cloning as described above. Annealed sticky-ended bapt was gel-purified using the Wizard SV Gel and PCR Clean-Up System (Promega, Madison,

WI). The pMAL-c2X plasmid (New England Biolabs, Ipswich, MA) was digested with

BamHI and HindIII and gel-purified (as described above). The gel-purified PCR insert and

BamHI/HindIII cut pMAL-c2X vector were ligated and transformed as described above.

Transformants were selected on 100 µg/mL ampicillin. Positive clones were screened by

PCR and DNA sequenced as described above.

3.2.3.5 E. coli strains

The pNterBAPT plasmid was transformed into BL21(DE3) E. coli for initial purification experiments. pNterBAPT was also transformed into Rosetta (DE3) and Rosetta (pLysS) E. coli to express rare tRNA codons. Other expression plasmids, including pCterBAPT, pOptBAPT, and pMBPCterBAPT were also transformed into these E. coli strains.

3.2.3.6 BAPT expression in E. coli

BAPT was expressed from pNterBAPT, pCterBAPT, pOptBAPT, and pMBP-CterBAPT in E. coli. Growth temperatures were modified for optimal expression during the initial growth phase, and also upon induction of protein expression by IPTG. A starter culture of

E. coli supplemented with the appropriate (50 µg/mL kanamycin or 100 µg/mL ampicillin, and 25 µg/mL chloramphenicol for Rosetta E. coli) was grown overnight at 37

112 °C. Variations of growth and induction were tested. Variation 1: The starter culture was used to inoculate (7 mL) a 1 L culture, which was then grown at 37 °C (or 30 °C) to an

OD600 of 0.6, induced by IPTG addition, and finally the temperature was lowered to 18 °C for 18 hr. Variation 2. The starter culture was used to inoculate (7 mL) a 1 L culture which was then grown for 18 hours at 25 °C, followed by IPTG induction, and continued incubation for 1 to 2 hours. IPTG concentrations (50 µM to 1 mM) were varied in both growth variations and induction of BAPT was qualitatively observed by SDS-PAGE.

3.2.3.7 Nickel-affinity chromatography

E. coli cell pellets were resuspended (3 mL/g(wet weight)) in ice cold Lysis Buffer (50 mM

NaPO4 (pH 8), 300 mM NaCl, 15 mM imidazole, and 5% glycerol). EDTA-free protease inhibitor cocktail tablets were added according to manufacturer’s instructions. The cell resuspension was kept on ice, lysed in batches (15 g pellet = ~four liters culture), and sonicated for 15 minutes in three cycles of 10 sec pulses followed by 20 sec rests with a

Misonix XL 2020 sonicator. Lysed cells were centrifuged at 18,000g for 20 minutes. The supernatant was further centrifuged in a Beckman Coulter Ultra Centrifuge at ~100000g for 1.5 hours. The supernatant was loaded onto a Fast-Flow Ni2+–NTA column (10 mL)

(Qiagen) pre-equilibrated with Lysis Buffer at 4 °C. The column was washed with 5 column volumes (cv) of Lysis Buffer, and eluted with 3 cv of Lysis Buffer containing 250 mM imidazole. The eluent was concentrated in a Millipore Amicon Ultra 30 kDa cutoff concentrator and buffer exchanged with 30 mM MOPS, pH 8.0 with 5% glycerol (Buffer

B). The procedure above was modified with Lysis Buffer containing 1 M NaCl or 10% glycerol.

113 3.2.3.8 Ammonium sulfate fractionation

Lysate was prepared as described above with pNterBAPT in BL21(DE3) E. coli. Lysate was gently stirred on ice for 15 minutes. Ammonium sulfate was then slowly added in 5-

10 % (w/v) increments. After each 10 % increment, the lysate was stirred for 20 minutes and centrifuged at 8000 g for 15 minutes. The supernatant was collected, placed on ice, and gently stirred. The next 10 % ammonium sulfate fraction was added and the process repeated. Fractions were collected at each step for analysis by SDS-PAGE and activity assays.

3.2.3.9 Ion exchange chromatography

NterBAPT was partially purified by nickel-affinity chromatography, desalted, and concentrated prior to anion-exchange chromatography. Fast-Flow Q Sepharose (~25 mL)

(GE Healthcare Bio-Sciences, Pittsburgh, PA) was equilibrated in 30 mM MOPS, (pH 8), with 5% glycerol. Partially purified NterBAPT was added to the ion exchange resin, the flowthrough (FT = ~30 mL) and washes (Q1-Q5 = 4 x 30 mL) were collected. Bound protein was eluted (QE = ~45 mL) from the resin with 30 mM MOPS, (pH 8), with 5% glycerol and 1 M NaCl. Fractions were analyzed for activity and by SDS-PAGE for purity.

Cation-exchange chromatography was performed identically to the anion-exchange chromatography described above. The resin was SP-sepharose (30 mL) (GE Healthcare

Bio-Sciences, Pittsburgh, PA). Fractions were collected and analyzed for BAPT activity and by SDS-PAGE for purity.

114 3.2.4 Optimized NterBAPT expression and purification.

Optimized purification of NterBAPT required three chromatography steps: nickel-affinity chromatography, anion-exchange chromatography, and cobalt-affinity chromatography.

The nickel-affinity purification and anion-exchange purification were performed as described above using E. coli from 10 L (~45 g) of cell culture. However, the desalted

Nter-BAPT eluent was concentrated (<5 mL) prior to loading on the Fast Flow Q sepharose resin. Only the first two flowthrough fractions (Q1 and Q2) contained NterBAPT activity.

These fractions were combined and directly loaded onto a cobalt-affinity column (1.5 mL)

(Gold Bio, St. Louis, MO). Cobalt Wash Buffer was 30 mM MOPS, (pH 8), 300 mM NaCl,

15 mM imidazole, and 5% glycerol. Cobalt Elution Buffer was the same as the Cobalt

Wash Buffer, but with 250 µM imidazole. The flowthrough (Co-FT), wash (5 cv) (Co-W), and elution (3 cv) (Co-E) were collected and analyzed for activity and purity by SDS-

PAGE analysis. The final cobalt elution (Co-E) was desalted, concentrated in 30 mM

MOPS, (pH 8), 5% glycerol, flash-frozen (40 µL aliquots) in liquid nitrogen and stored at

-80 °C. Quantification of the 53 kDa band corresponding to NterBAPT was determined by gel densitometry (see below).

3.2.5 MBP-BAPT expression and purification

BAPT was expressed as a maltose binding protein (MBP) fusion protein in

BL21(DE3) E. coli (Invitrogen, Carlsbad, CA) from both the pMBP-NterBAPT and pMBP-CterBAPT expression plasmids. A single colony was selected and used to inoculate a 100-mL culture of LB media containing 100 µg/mL ampicillin. The culture was grown overnight at 37 °C. Aliquots of this starter culture (7 mL) were used to inoculate a flask of fresh LB media (1 L) containing 100 µg/mL ampicillin. MBP-BAPT was purified from

115 eight liters which were grown at 30 °C until OD600 = 0.6. Gene expression was induced by

0.1 mM IPTG for ~16 hours at 18 °C. Bacteria were harvested by centrifugation at 8,000g for 5 minutes. Pellets were resuspended in Buffer A (30 mM MOPS, (pH 8), 300 mM NaCl,

15 mM imidazole, 5% glycerol) at 3 mL/g pellet (wet weight). PMSF (100 mM stock in isopropanol) was added to the cell suspension at a final concentration of 1 mM. MBP-

BAPT was purified by nickel-affinity chromatography as described above. The eluent was loaded onto amylose-binding resin (1 mL), pre-equilibrated with Buffer B. The amylose column was then washed with 5 cv of Buffer B and eluted with 3 cv of Buffer B containing

10 mM maltose. The eluted MBP-BAPT was further concentrated in a Millipore Amicon

Ultra 30 kDa cutoff concentrator and buffer-exchanged with Buffer B to a volume of ~1-3 mL. MBP-BAPT was aliquoted (40 µL), flash-frozen in liquid nitrogen, and stored at –80

°C.

Final concentrations were ~0.3 mg/mL (estimated by the Coomassie (Bradford)

Protein Assays) (Thermo Scientific Pierce, Grand Island, NY). Protein was analyzed for purity by SDS-PAGE, band densitometry, and proteomics.

3.2.6 Band densitometry

Purified protein was separated by SDS-PAGE (10% running gel) and stained by Coomassie blue or by silver. For band densitometry, purified BadA (see Chapter 2) (99%) was run in serial dilutions to generate a standard curve (Figure A.2.8). Band densities were determined on a BioRad Molecular Imager Gel Doc XR+ using ImageLab software. Linear regression was used to generate a standard curve (Excel, Microsoft, Redmond, WA). BAPT (band density at ~53 kDa) levels were calculated from this curve. Quantification of total protein

116 was determined by Coomassie (Bradford) Protein Assays (Thermo Scientific Pierce, Grand

Island, NY).

3.2.7 Proteomics

Purified BAPT samples were submitted to the Michigan State University Research

Technology Support Facility (RTSF) Proteomics Core for identification of co-purifying proteins with BAPT.

3.2.8 BAPT activity assays

Coupled assays with PheAT, the (3R)-β-Phe CoA ligase, were used to follow BAPT activity throughout purification. PheAT was prepared as previously described (35). Assays were prepared on ice and incubated at 31 °C for 5 minutes prior to enzyme addition. The initial PheAT assay (200 µL) contained baccatin III (1 mM in acetonitrile, 10 % (v/v)), followed by 200 mM MOPS (pH 8), 3.5 mM MgCl2, 1 mM ATP, 1 mM (3R)-β- phenylalanine, 1 mM CoA, and PheAT (0.75 mg/mL). After one hour at 31 °C, the BAPT purification fraction (50 µL) was added and allowed to react for an additional two hours.

The reaction was stopped by acidification (pH 5) with 8.8% formic acid, followed by ethyl acetate (2 mL) addition. The docetaxel internal standard (1 µM) was added to the reaction, and extracted twice with EtOAc (2 × 2 mL). The organic layers were pooled and dried under a nitrogen stream. Coupled assay products were analyzed by LC-ESI-MS/MS with multiple reaction monitoring of the N-debenzoyl-2′-deoxypaclitaxel ion fragmentation [M

+ H]+ = m/z 734 → m/z 509 and of the internal standard (docetaxel) fragmentation (1 µM) ion [M + H]+ = m/z 808 → m/z 509. Product peaks were quantified by calculating the ratio of the N-debenzoyl-2′-deoxypaclitaxel peak area to that of the docetaxel (1 µM) internal

117 standard. The amount of product formed was converted to units of activity (µmol/min).

Total protein (mg) was quantified by Coomassie Bradford assays.

118 3.3 Results and discussion

3.3.1 BAPT expression in Pichia pastoris

Success with other BAHD family members led us to believe that P. pastoris may be a more suitable host for high-level, soluble expression of BAPT than E. coli (4, 25). The bapt gene was cloned into the pPICZc expression vector as an N-terminal His-tagged (pHisBAPT) and a cMyc epitope-tagged (pMycBAPT) protein for methanol-inducible, cytoplasmic expression in the yeast, P. pastoris. The pPICZ vector was designed for stable integration in P. pastoris where the 5’AOX1 (alcohol dehydrogenase) promoter region in pPICZ recombines with the yeast chromosomal 5’AOX1 promoter to integrate the bapt gene such that expression is driven by the AOX1 promoter when grown with methanol as the sole carbon source. Multiple HisBAPT clones were screened for BAPT expression after 16 and

41 hours (Figure 3.5A). Minor increases in band density consistent with HisBAPT (~52 kDa) were observed in induced samples but not in background samples transformed with pPICZ empty vector (Figure 3.5B). Large-scale expression and the addition of nickel- affinity chromatography consistently showed no HisBAPT (~52 kDa) present by SDS-

PAGE (data not shown). Unfortunately, expression of P. pastoris clones containing integrated cMycBAPT was also not detectable, either by SDS-PAGE or western blot

(Figure 3.5C and D). No protein bands consistent with cMycBAPT (54 kDa) were observed. BAPT activity assays were also negative.

119

A. B.

C.

D.

Figure 3.5. SDS-PAGE of whole cell lysate from expressions of BAPT clones from P. pastoris. Cells were induced for 16 hours or as labeled, with 0.5 % methanol and 1000 µg/mL zeocin. A. Expression of five clones of HisBAPT (52 kDa). B. Background expression with five clones containing pPICZ empty vector. C. Expression of five clones of MycBAPT (54 kDa). D. Representative western blot of MycBAPT clones with anti-6XHis tag antibody. Arrow indicates expected band location for expressed BAPT protein. The 6X- His tagged BadA, benzoate CoA ligase, was used as a positive control. Ladder is Kaleidoscope Prestained Protein Ladder (BioRad, Hercules, CA).

120

Despite successes with other BAHD family members, the P. pastoris expression system proved unsuccessful at producing active, soluble BAPT. Although multiple clones (~20 per construct) were tested for expression and activity we were never able to detect any

BAPT by western blot or activity assays with (3R)-β-Phe CoA and baccatin III (data not shown). The persistent lack of activity in P. pastoris, combined with active BAPT expression in E. coli redirected purification efforts from yeast to bacteria.

Figure 3.6. NterBAPT expression in crude cell lysates. Four NterBAPT (52.7 kDa) clones in BL21(DE3) E. coli were screened for expression. Induction conditions included growth at 37 °C to an OD600 of 0.6, addition of 0.25 mM ITPG, and growth at 18 °C for 16 hours. Uninduced (U) and induced (I) whole cell lysates were compared to empty vector, pET28a (1), NterBAPT Clone #5 (2), NterBAPT Clone #10 (3), Clone #15 (4), and Clone #20 (5). NterBAPT expression is observed strongly in Clone #20 and less so in Clone #5 and Clone #10. There is no expression in the empty vector control or in Clone #15. Positive control (+) is BadA, a benzoate CoA ligase. Ladder is PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA).

121 3.3.2 BAPT expression and purification in E. coli

3.3.2.1 Expression optimization in E. coli

BAPT was expressed as an N-terminal polyhistidine-tagged protein from pNterBAPT, hereafter referred to as NterBAPT. Induced whole cell lysate showed a ~53 kDa band consistent with the expected size (52.7 kDa) of NterBAPT (Figure 3.6 and Figure 3.9).

Subsequent purification steps showed NterBAPT to be in inclusion bodies (data not

Figure 3.7. Soluble expression of NterBAPT with varied IPTG concentrations. E. coli were induced with IPTG concentrations from 50 µM to 500 µM. Nickel-affinity column elutions are shown for each IPTG concentration. Samples were either induced for 1 hour or 18 hours at 18 °C. Arrows point to induced band at ~53 kDa. The band density decreases from 50 µM to 500 µM IPTG.

shown). Although most of the enzyme appeared to be lost, some NterBAPT activity was retained in the clarified lysate. Efforts to improve the amount of soluble protein included modifying growth and expression conditions, moving the polyhistidine tag from the N- to

C-terminal (pCterBAPT), and the generation of an expression vector containing a C- terminal His-tagged codon-optimized bapt gene for expression in E. coli (pOptBAPT).

Increasing induction time and varying IPTG concentrations did not substantially affect accumulation of soluble NterBAPT or CterBAPT (Figure 3.7). Additionally, more soluble

NterBAPT was found with low concentrations (50 µM) of IPTG (Figure 3.7). There is no

122 apparent difference between expression induction for 3 hours or 18 hours with pOptBAPT, suggesting the low solubility of BAPT is not related to codon usage in E. coli (Figure 3.8).

Based upon IPTG concentration and temperature induction studies, the optimal NterBAPT expression conditions were growth at 30 °C for 16 hours, and induction with 50 µM IPTG for 1.5 hours at 18 °C.

Expression of BAPT in different E. coli cell types, including Rosetta(DE3) and

Rosetta(DE3)(pLysS) (36), which contain expression plasmids for tRNAs of rare codons in E. coli, were also not effective at increasing soluble BAPT (not shown). These results are consistent with the expression of OptBAPT, where no change in soluble protein was observed, also suggesting that problems with BAPT solubility are not related to codon usage. A comparison of NterBAPT expression with a T. cuspidata N-debenzoylpaclitaxel-

Figure 3.8. Expression and induction of OptBAPT in E. coli. Codon optimized, OptBAPT (52.1 kDa), expression levels in E. coli BL21(DE3) cells induced for 3 hours or 18 hours at 18 °C with 0.25 mM IPTG. Uninduced (U) and induced (I) whole cell lysates, the wash (W) fraction from nickel-affinity chromatography and elution fractions (1-6) were compared between 3 hour and 18 hour induced E. coli. Arrows point to the expected 52 kDa band corresponding to OptBAPT expression. Ladder is PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA)

123 N-benzoyltransferase (NDTNBT) (21), a similar paclitaxel biosynthetic acyltransferase

(10), shows the differences in soluble expression between the two highly similar enzymes

(56.5% identity and 71.4% similarity, (Figure A.2.1), (Figure 3.9). NDTNBT is more soluble than BAPT expressed from any of the constructs tested. The uninduced (U) and induced (I) whole cell lysates of NterBAPT and NDTNBT have similar expression patterns. Upon purification by nickel-affinity chromatography, it is apparent from examining the elution (E) fractions, that larger proportions of NDTNBT than BAPT is soluble and bound to the nickel-affinity resin (Figure 3.9). The low solubility of BAPT is likely due to its conformation in E. coli and interaction with cellular chaperones (37).

Figure 3.9. Relative expression levels of NterBAPT and the Taxus N-benzoyltransferase (NDTNBT) by SDS-PAGE. Uninduced (U), induced (I) for 18 hours at 18 °C with 0.25 mM IPTG, whole cell lysates and the nickel-affinity column elution (E) fraction from each purification are shown. The arrow points to the induced ~53 kDa band of NterBAPT. NDTNBT is ~53 kDa. Both enzymes were prepared identically in the same pET28a plasmid background. Ladder is PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA)

124

Figure 3.10. SDS-PAGE of NterBAPT purification by nickel-affinity chromatography. Fractions are as follows CL: clarified lysate, FT: flowthrough, W1: wash 1, W2: wash 2, Ec: concentrated eluent. Bands consistent with NterBAPT (52.7 kDa) are visible near the labeled 55 kDa ladder band. The ladder is PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA).

3.3.2.2 Purification by nickel-affinity chromatography

NterBAPT was expressed in E. coli cells (either BL21(DE3), Rosetta, or Rosetta (pLysS)), and grown under optimal conditions as described above. The clarified lysate was applied to nickel-affinity resin where the flowthrough (FT), washes (W), and elution (E) fractions were analyzed by SDS-PAGE and for activity (Figure 3.10 and Figure 3.11). The eluent

(E) was highly impure with numerous bands present on the gel (Figure 3.10, Lane Ec).

Most activity was retained on the column, with only 8% lost in the FT and W fractions

(Figure 3.11A). However, 66% of activity was lost in the elution (E) fraction compared with the clarified lysate (CL). This loss of activity could not be accounted for with activity measured in the other fractions. The loss was reproducible and unaffected by increased glycerol concentration (5% to 10%) or increased NaCl concentration (0.3 M to 1 M). It is possible that imidazole affects the stability and decreases activity during elution with high

(>200 mM) imidazole concentrations. Unfortunately, lower imidazole concentrations were insufficient to elute NterBAPT from the nickel-affinity resin (Figure 3.11B).

125

A. B.

100 100

80 80

60 60

40 40

Normalized Relative Relative to CL % Total % Activity 20 20

0 NterBAPT Activity (%) 0 CL FT W1 W2 Ec 10 25 50 100 200 Nickel-Affinity Chromatography Nickel-Affinity Chromatography Fractions Imidazole Elutions (mM)

C.

Protein in Activity Specific Activity Specific Activity Fraction Assay (mg) (µmol/min) (µmol/min mg) (Fold Change relative to CL) CL 0.59 2.77 × 10-5 4.69 × 10-5 1 FT 0.52 1.08 × 10-6 2.08 × 10-6 0.04 -5 -6 W1 1.79 1.47 × 10 8.22 × 10 0.18 -5 -5 W2 0.21 1.40 × 10 6.69 × 10 1.4 -4 -3 Ec 0.24 3.68 × 10 1.53 × 10 33

Figure 3.11. NterBAPT activity during nickel-affinity chromatography. Fractions are as follows CL: clarified lysate, FT: flowthrough, W1: wash 1, W2: wash 2, Ec: concentrated eluent. A. Bar graph showing the loss of total activity over the course of chromatography. B. Bar graph showing that NterBAPT binds nickel-affinity resin. Concentrations above 200 µM are required to elute all enzyme from the column. C. NterBAPT specific activity table of nickel-affinity purification fractions.

126 100 80 60 40 20

Activity Activity Remaining (%) 0 10 20 25 30 35 40

Ammonium Sulfate (% (w/v))

Figure 3.12. Ammonium sulfate fractionation of NterBAPT clarified lysate.

3.3.2.3 Ammonium sulfate precipitation

Although nickel-affinity chromatography increased the specific activity of NterBAPT compared to the lysate (33-fold), the elution was still impure, making it difficult to precisely quantify NterBAPT, an important consideration for determining Michaelis-

Menten kinetic parameters (Figure 3.11C). Ammonium sulfate precipitation was tested as a means of precipitating undesired contaminating proteins early in the purification.

Ammonium sulfate was added to the lysate stepwise in 5% or 10% (w/v) increments and

NterBAPT activity was measured (Figure 3.12). NterBAPT activity decreased dramatically

(~45%) after the addition of 20% (w/v) ammonium sulfate. Most bacterial proteins precipitated between 35% - 60% (w/v) ammonium sulfate (Figure 3.12). Since BAPT activity disappears before precipitation of protein impurities, ammonium sulfate fractionation was abandoned as a means of increasing NterBAPT purity.

3.3.2.4 Ion exchange chromatography

Because nickel-affinity chromatography was insufficient to purify NterBAPT, an additional chromatographic step was sought. The calculated pI for NterBAPT is 8.0,

127 suggesting that the enzyme may bind to either anion or cation-exchange resin in solution.

Both anion (Q Sepharose) and cation (SP Sepharose) exchange resins were tested as a second chromatography method with partially purified NterBAPT (nickel-affinity chromatography elution). Collected fractions were analyzed by SDS-PAGE (Figure 3.13).

NterBAPT appeared to bind the cation-exchange resin, but not the anion-exchange resin

(data not shown). Protein impurities from the nickel-affinity purification step were retained on the anion-exchange resin. The flowthrough (Q) collected from anion-exchange chromatography contained BAPT activity while most impurities were retained on the column resin (Figure 3.13). For this reason anion-exchange chromatography was chosen as the second step in the NterBAPT protein purification.

3.3.3 Optimized NterBAPT purification

The NterBAPT purification was optimized with three chromatography steps: nickel- affinity, anion-exchange (Q resin), and cobalt-affinity. Purification by the nickel-affinity and anion-exchange chromatography was performed as described above. The addition of a final cobalt-affinity chromatography step substantially improved the NterBAPT purification (Figure 3.13). Eluent from this column was 40-80% pure (on average) with several protein impurities observed on a silver-stained SDS-PAGE gel (Figure 3.18).

Analysis of the activity in each fraction showed a definite increase in specific activity

(~14000-fold) despite the loss of some BAPT activity during nickel-affinity

128

A. B.

C. Specific Activity Protein Activity Specific Activity Total Activity Fraction (Fold Change (mg) (µmol/min) (µmol/min mg) (% Relative to CL) relative to CL) CL 6444 4.20 × 10-3 6.52 × 10-7 100 1.0 FT 3536 1.98 × 10-4 5.61 × 10-8 4.7 0.1 -3 -6 Ni-Ec 275 1.98 × 10 7.21 × 10 47 11.0 Q1 21.5 1.06 × 10-3 4.95 × 10-5 25 75.8 Q2 12.6 1.98 × 10-4 1.58 × 10-5 4.7 24.2 Q3-4 31.5 6.67 × 10-5 2.12 × 10-6 1.6 3.2 Q5 82.7 1.16 × 10-5 1.41 × 10-7 0.3 0.2 Q6 56.6 3.05 × 10-5 5.39 × 10-7 0.7 0.8 -3 -3 Co-Ec 0.126 1.22 × 10 9.65 × 10 29 14800

Figure 3.13. Purification of NterBAPT by multiple chromatography steps. A. The nickel-affinity column concentrated eluent (Ec), anion-exchange column flowthrough/wash (Q1-5), and elution (Q6) fractions are shown. B. Fractions Q1 and Q2 were pooled, concentrated (Qc) and loaded onto cobalt-affinity resin. The flowthrough (FT), wash (W) and elution (Co-E) fractions are shown. The Co-E fraction contains a 53 kDa band consistent with NterBAPT. C. Specific activity of the NterBAPT purification. The activity in both Q1 and Q2 were combined prior to loading on a cobalt-affinity column. Fractions are labeled as follows: CL; clarified lysate, FT; flowthrough, W; wash, Ec; concentrated eluent, Q; flowthrough/wash fractions from the anion-exchange column, Qc; concentrated Q1 and Q2, Co-E; eluted protein from the cobalt-affinity column. Ladder (L) is the PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA).

129 chromatography (Figure 3.13).The dramatic increase in specific activity is a function of the low amounts of soluble BAPT and the large amount of E. coli processed during purification. From this three column purification, on average, 0.087 mg (1.65 nmol, MW

= 52.7 kDa) of NterBAPT (accounting for ~40-80% purity based upon gel densitometry) was purified from 40 g (wet weight from ~8 L) of E. coli. Yields from this purification were approximately 0.0022 mg of NterBAPT per gram of E. coli (wet weight) (Figure

3.13). The purified NterBAPT was concentrated and proteins were identified by proteomics with tandem mass spectrometry analysis (Figure A.2.9). Results confirmed the presence of

NterBAPT and identified the major impurities (Table 3.2). The chaperones GroEL and

DnaK were identified, which is not surprising considering the limited solubility of

NterBAPT. Other impurities included a cAMP protein, UDP-D-glucuronate dehydrogenase, and 23S rRNA m1g745 methyltransferase (Table 3.2). These proteins may have binding-affinity for NterBAPT, but why they co-purified remains unclear.

Table 3.2. Proteomics analysis of purified NterBAPT.

NterBAPT Protein Impurity Size (kDa) GroEL 57 cAMP receptor protein 24 UDP-D-glucuronate dehydrogenase 74 DnaK 69 23S rRNA m1g745 methyltransferase 30

130

Figure 3.14. Coupled assay schematic with PheAT and BAPT. PheAT produces (3R)-β-phenylalanyl CoA (IV.) from (3R)-β-phenylalanine (III.) which is a substrate for BAPT. Baccatin III (I.) acts as an acyl acceptor. BAPT transfers the (3R)-β-phenylalanyl group to the 13-O position on baccatin III to form the acylated product N-debenzoyl-2′-deoxypaclitaxel (II.).

3.3.4 BAPT activity assays

A coupled enzyme assay was developed with a modified bacterial acyl CoA ligase to detect

BAPT activity since the substrate (3R)-β-Phe CoA is commercially unavailable. The TycA

(PheATE), α-phenylalanine epimerase/non-ribosomal peptide synthetase (NRPS), from the tycrocidine synthetase A biosynthetic pathway in Bacillus brevis was recently discovered to have broader activity (35, 38, 39). The wild-type PheATE was truncated by removal of the epimerization (“E”) domain and mutated in the thiolation (“T”) domain to prevent transfer of the amino acid to the growing tyrocidine peptide chain (35).

These modifications resulted in an active enzyme, PheAT, capable of biosynthesizing

131

A.

B.

120 100 80 60 40 20

Total Ion Count (%) Count Ion Total 0 0 1 2 3 4 5 Time (min) Figure 3.15. Biosynthesis of N-debenzoyl-2′-deoxypaclitaxel by BAPT. A. Structure of N-debenzoyl-2′- deoxypaclitaxel (m/z 734 → m/z 509). Loss of the (3R)-β-phenylalanyl sidechain and an acetyl group yield a fragment of m/z 509. B. Representative LC-ESI/MSMS chromatogram showing the production of N- debenzoyl-2′-deoxypaclitaxel (3.0 minutes) (m/z 734 → m/z 509 (black), docetaxel internal standard: m/z 808 → m/z 509 (red)) by MBP-BAPT (Amylose column elution fraction) and PheAT coupled assays containing baccatin III, (3R)-β-phenylalanine, ATP, and CoA substrates. Mass spectrometry performed on a Waters Quattro micro API LC/MS/MS (Waters, Milford, MA).

(3R)-β-Phe CoA from (3R)-β-phenylalanine, ATP, and CoA, a required precursor for the

BAPT biosynthetic reaction (Figure 3.14). The coupled assay system produces N- debenzoyl-2′-deoxypaclitaxel in two steps. First, PheAT biosynthesizes (3R)-β-Phe CoA.

Second, BAPT catalyzes the transfer of the (3R)-β-phenylalanyl moiety to baccatin III, a paclitaxel precursor (Figure 3.14) (35, 40). Assays were analyzed by LC-ESI-MS-MS chromatography to detect N-debenzoyl-2′-deoxypaclitaxel ([M + H]+ = m/z 734) (Figure

132 3.15). The major product, N-debenzoyl-2′-deoxypaclitaxel, elutes at ~3 minutes. However, a low-level, broad secondary peak elutes at ~2.4 minutes. This secondary peak was present in all assays containing (3R)-β-Phe CoA and baccatin III, including PheAT coupled assays and assays with chemically synthesized (3R)-β-Phe CoA, suggesting the broad peak is derived from assay products. Tandem mass spectrometry of each peak shows nearly identical fragmentation patterns which are consistent with N-debenzoyl-2′-deoxypaclitaxel

(41, 42) (Figure 3.16). It is possible that the broad 2.4 min peak (Figure 3.16B, C) represents isomers of N-debenzoyl-2′-deoxypaclitaxel ([M+H]+ = m/z 734) (43, 44).

Isomerization is reported under neutral to basic conditions when paclitaxel epimerizes at the C7 position to form 7-epi-paclitaxel and possibly at the C10 position as well (44).

BAPT activity assays are incubated at pH 8, followed by dilute acid quenching (pH 5) with

8.8% formic acid in order to stop the reaction because of residual BAPT activity after ethyl acetate extraction. The assay products were exposed to both basic and acidic conditions in which some low-level degradation products may occur. The formation of multiple low- level isomers of N-debenzoyl-2′-deoxypaclitaxel is consistent with the broad, poorly resolved peak at 2.4 minutes. Activity levels were quantified in each fraction by integrating the area under the major product peak at ~3 min. This assay was used to follow BAPT activity over the course of the purification.

133 A.

B.

C.

D.

E.

Figure 3.16. Representative LC-ESI/MS/MS of a BAPT assay. A. Structures and fragmentation of N- debenzoyl-2′-deoxypaclitaxel. Docetaxel sidechain mass is in parentheses. B. Total ion chromatogram for the fragmentation of N-debenzoyl-2′-deoxypaclitaxel (m/z 734). C. Product ion spectrum of m/z 734 at 2.4 minutes. D. m/z 734 at 2.95 minutes and E. docetaxel (m/z 808) at 4.4 minutes. Total ion chromatogram for the docetaxel product ion spectrum is not shown.

134 3.3.5 MBP-BAPT expression and purification

We intended to pursue kinetic analysis, paclitaxel biosynthetic assays, and crystallography with purified BAPT protein. Amounts of purified Nter-BAPT were too low for protein crystallization screens or for large-scale assays. Consequently, another method for soluble

BAPT expression was sought. Fusion proteins such as the maltose binding protein (MBP)

(45), N utilization substance A (NusA) (46, 47), thioredoxin (TRX) (48), and SUMO (49) are commonly used for enhancing the solubility of proteins (50, 51). We chose to investigate whether the addition of an MBP fusion partner would enhance the expression and solubility of BAPT compared with previous efforts. The bapt gene was cloned in the pMal-c2X expression vector with an N- or C-terminal 6X-His tag (Figure 3.17). We hypothesized that the MBP-CterBAPT construct would be more efficient at binding nickel- affinity resin than MBP-NterBAPT due to accessibility of the polyhistidine tag. Initial expression tests confirmed higher soluble expression of MBP-CterBAPT. Therefore, the fusion protein was used for large-scale expression and purification optimization.

Figure 3.17. Domain structures of MBP-BAPT fusion proteins. MBP-NterBAPT and MBP-CterBAPT are shown above. Maltose binding protein (MBP), polyhistidine tag (6X-His).

135 A. B.

Figure 3.18. Representative purification of MBP-BAPT. A. The enzyme was purified by nickel and amylose affinity chromatography as labeled on an SDS-PAGE gel. Fractions are P2; insoluble protein, CL; clarified lysate; FT; flowthrough, W1; wash 1, W2; wash 2, E; elution, Ec; concentrated elution. B. SDS- PAGE gel (silver-stained) showing purified NterBAPT (1) and MBP-CterBAPT (2) (52.7 and 95.2 kDa, respectively) from BL21(DE3) E. coli. Ladder is PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA).

The fusion protein MBP-CterBAPT was purified by nickel-affinity and amylose- affinity chromatography (Figure 3.18). Active protein was detected in the lysate by activity assay, but inclusion bodies containing MBP-CterBAPT were still present in E. coli expressing the MBP-BAPT fusion protein (Figure 3.18A). Expression optimization included reducing the growth temperature from 37 °C to 30 °C for ~14 h to an OD600 of

~2, followed by a 90 min IPTG (100 µM) induction. Nickel and amylose-affinity chromatography were performed in succession to yield 1.4 mg (14.7 nmol, MW = 95.2 kDa) of purified MBP-CterBAPT (accounting for ~50 % purity based upon densitometry) from 35 g (wet weight) of E. coli or ~0.039 mg/g(wet weight) (Figure 3.18). This was a 10-fold increase (mol BAPT per gram of E. coli) in the amount of MBP-CterBAPT purified compared with the previous NterBAPT methodology.

A proteomics analysis identified the major impurities as GroEL, DnaK, and 2- oxoglutarate dehydrogenase E1 component (Table 3.3). GroEL and DnaK are both

136 Table 3.3. Proteomics analysis of purified MBP-BAPT.

MBP-BAPT Protein Impurity Size (kDa) GroEL 57 DnaK 69 2-oxoglutarate dehydrogenase 105 E1 component

chaperones that assist in protein folding (52) and 2-oxoglutarate dehydrogenase E1 component is an enzyme in the primary metabolic tricarboxylic acid (TCA) cycle. Given the low solubility of BAPT, it is not surprising that chaperones co-purified with the enzyme even after passing through multiple columns. Reasons why these and other proteins were retained after two different affinity column purification steps remain unclear. These proteins likely have affinity for BAPT and possibly the Ni2+-NTA or amylose resin.

Yields of MBP-CterBAPT were an order of magnitude higher than NterBAPT with fewer purification steps. For these reasons, MBP-CterBAPT was used to determine the

Michaelis-Menten kinetic constants with substrates of interest including, (3R)-β-Phe CoA,

(2R,3S)-PhIS CoA, baccatin III and 10-deacetylbaccatin III (Chapter 4).

137 3.4 Conclusions

Other BAHD acyltransferases were successfully expressed in both P. pastoris and E. coli at high yields for protein crystallography and kinetics experiments (5, 11-14). BAPT expression was barely detectable in whole cell lysates of P. pastoris and acyltransferase activity was never detected. Definitive expression and activity of BAPT in E. coli whole cell lysates led us to abandon the P. pastoris expression system. Soluble BAPT levels were consistently low (~0.002 mg/g E. coli), regardless of whether BAPT was codon-optimized or expressed as an N- or C-terminal polyhistidine-tagged protein. Ultimately, BAPT was purified as an N-terminal polyhistidine-tagged protein (NterBAPT) to >50 % purity (by

SDS-PAGE band densitometry) requiring three sequential chromatographic steps, including nickel-affinity, anion-exchange, and cobalt-affinity chromatography. Proteomics analysis identified the major impurities as E. coli chaperones GroEL and DnaK.

The low yields and limited solubility of NterBAPT were not compatible with the long-term goals of BAPT kinetic characterization for paclitaxel biosynthesis and protein crystallography. Maltose binding protein (MBP) is known to increase the solubility of fused protein partners through an unidentified mechanism (45). A C-terminal polyhistidine-tagged, MBP-BAPT fusion protein was hypothesized to increase the solubility of BAPT over previous methods. Fusion protein was expressed in E. coli and purified by two sequential chromatographic steps, including nickel-affinity and amylose- affinity chromatography. MBP-BAPT was purified to >50 % purity (by SDS-PAGE band densitometry) at 1.4 mg (14.7 nmol, MW = 95.2 kDa) or ~0.039 mg/g(wet weight) with greater solubility and 10-fold (mol BAPT per g of E. coli) higher yields than any prior BAPT

138 expression system. Proteomics identified the major impurities as GroEL and DnaK, similar to the previous expression system.

139 3.5 Future Research

Purified MBP-BAPT was used for kinetic characterization (Chapter 4) with the acyl donors, (3R)-Phe CoA and (2R,3S)-PhIS CoA, and the acyl acceptors, baccatin III and 10- deacetylbaccatin III. MBP-BAPT was used to demonstrate the biosynthesis of N- debenzoylpaclitaxel (the penultimate paclitaxel precursor), N-debenzoyl-10- deacetylpaclitaxel (the penultimate docetaxel precursor) and N-debenzoyl-N-2- furanylpaclitaxel (a biologically relevant paclitaxel analog) (53).

Future research with BAPT will include attempts at protein crystallization- alone and as an MBP fusion protein (54-56). BAPT has unique characteristics that make a crystal structure of high interest. The conserved HXXXD motif in all BAHD members is naturally modified to GXXXD in the BAPT enzyme. How the substrates bind the active site for catalysis without the conserved catalytic histidine residue is unknown. Structural information may help explain the significance of the natural GXXXD mutation and implications for BAPT catalysis.

140

APPENDIX

141 APPENDIX

NDTNBT MEKAGS-TDFHVKKFDPVMVAPSLPSPKATVQLSVVDSLTICRG-IFNTLLVFNAP-DN- BAPT MKKTGSFAEFHVNMIERVMVRPCLPSPKTILPLSAIDNMARA---FSNVLLVYAANMDR- DBAT --MAGS-TEFVVRSLERVMVAPSQPSPKAFLQLSTLDNLPGVRENIFNTLLVYNAS-DR- TAT ---MEK-TDLHVNLIEKVMVGPSPPLPKTTLQLSSIDNLPGVRGSIFNALLIYNAS-PSP TBT ---MGR---FNVDMIERVIVAPCLQSPKNILHLSPIDN--KTRG-LTNILSVYNAS-QRV : * :: *:* *. ** : ** :*. : * * :: *

NDTNBT --ISADPVKIIREALSKVLVYYFPLAGRLRSKEIGELEVECTGDGALFVEAMVEDTISVL BAPT --VSADPAKVIREALSKVLVYYYPFAGRLRNKENGELEVECTGQGVLFLEAMADSDLSVL DBAT --VSVDPAKVIRQALSKVLVYYSPFAGRLRKKENGDLEVECTGEGALFVEAMADTDLSVL TAT TMISADPAKPIREALAKILVYYPPFAGRLRETENGDLEVECTGEGAMFLEAMADNELSVL TBT S-VSADPAKTIREALSKVLVYYPPFAGRLRNTENGDLEVECTGEGAVFVEAMADNDLSVL :*.**.* **:**:*:**** *:*****..* *:*******:*.:*:***.: :***

NDTNBT RDLDDLNPSFQQLVFWHPLDTAIEDLHLVIVQVTRFTCGGIAVGVTLPHSVCDGRGAAQF BAPT TDLDNYNPSFQQLIFSLPQDTDIEDLHLLIVQVTRFTCGGFVVGANVYGSACDAKGFGQF DBAT GDLDDYSPSLEQLLFCLPPDTDIEDIHPLVVQVTRFTCGGFVVGVSFCHGICDGLGAGQF TAT GDFDDSNPSFQQLLFSLPLDTNFKDLSLLVVQVTRFTCGGFVVGVSFHHGVCDGRGAAQF TBT QDFNEYDPSFQQLVFNLREDVNIEDLHLLTVQVTRFTCGGFVVGTRFHHSVSDGKGIGQL *::: .**::**:* *. ::*: : **********:.**. . . .*. * .*:

NDTNBT VTALAEMARGEVKPSLEPIWNRELLNPEDPLH-LQLNQFDSICPPPMLEELGQASFVINV BAPT LQSMAEMARGEVKPSIEPIWNRELVKLEHCMP-FRMSHLQIIHAPVIEEKFVQTSLVINF DBAT LIAMGEMARGEIKPSSEPIWKRELLKPEDPLYRFQYYHFQLICPPSTFGKIVQGSLVITS TAT LKGLAEMARGEVKLSLEPIWNRELVKLDDPKY-LQFFHFEFLRAPSIVEKIVQTYFIIDF TBT LKGMGEMARGEFKPSLEPIWNREMVKPEDIMY-LQFDHFDFIHPPLNLEKSIQASMVISF : .:.******.* * ****:**::: : :. ::: : .* : * ::*

NDTNBT DTIEYMKQCVMEECNEFCSSFEVVAALVWIARTKALQIPHTENVKLLFAMDLRKLFNPPL BAPT EIINHIRRRIMEERKESLSSFEIVAALVWLAKIKAFQIPHSENVKLLFAMDLRRSFNPPL DBAT ETINCIKQCLREESKEFCSAFEVVSALAWIARTRALQIPHSENVKLIFAMDMRKLFNPPL TAT ETINYIKQSVMEECKEFCSSFEVASAMTWIARTRAFQIPESEYVKILFGMDMRNSFNPPL TBT ERINYIKRCMMEECKEFFSAFEVVVALIWLARTKSFRIPPNEYVKIIFPIDMRNSFDSPL : *: :.. : ** :* *:**:. *: *:*. .::.** .* **::* :*:*. *:.**

NDTNBT PNGYYGNAIGTAYAMDNVQDLLNGSLLRAIMIIKKAKADLKDNYSRSRVVTNPYSLDVNK BAPT PHGYYGNAFGIACAMDNVHDLLSGSLLRTIMIIKKSKFSLHKEL-NSKTVMSSSVVDVNT DBAT SKGYYGNFVGTVCAMDNVKDLLSGSLLRVVRIIKKAKVSLNEHF-TSTIVTPRSGSDESI TAT PSGYYGNSIGTACAVDNVQDLLSGSLLRAIMIIKKSKVSLNDNF-KSRAVVKPSELDVNM TBT PKGYYGNAIGNACAMDNVKDLLNGSLLYALMLIKKSKFALNENF-KSRILTKPSTLDANM . ***** .* . *:***:***.**** .: :***:* *:. * : * .

NDTNBT KSDNILALSDWRRLGFYEADFGWGGPLNVSSLQR-LENGLPMFSTFLYLLPAKNKSDGIK BAPT KFEDVVSISDWRHSIYYEVDFGWGDAMNVSTMLQQQEHEKSLPTYFSFLQSTKNMPDGIK DBAT NYENIVGFGDRRRLGFDEVDFGWGHADNVSLVQHGLKDVSVVQSYFLFIRPPKNNPDGIK TAT NHENVVAFADWSRLGFDEVDFGWGNAVSVSPVQQ--QSALAMQNYFLFLKPSKNKPDGIK TBT KHENVVGCGDWRNLGFYEADFGWGNAVNVSPMQQQREHELAMQNYFLFLRSAKNMIDGIK : ::::. .*. . : *.***** . .** : . : : . * :: ..** ****

NDTNBT LLLSCMPPTTLKSFKIVMEAMIEKYVSKV----- BAPT MLMF-MPPSKLKKFKIEIEAMIKKYVTKVCPSKL DBAT ILSF-MPPSIVKSFKFEMETMTNKYVTKP----- TAT ILMF-LPLSKMKSFKIEMEAMMKKYVAKV----- TBT ILMF-MPASMVKPFKIEMEVTINKYVAKICNSKL :* :* : :* **: :*. :***:* Figure A.2.1. Multiple sequence alignment of Taxus cuspidata acyltransferases. Sequence identity between enzymes is 57-65%. The conserved HXXXD and DFGWG motifs are underlined. BAPT is the only BAHD acyltransferase without the conserved histidine residue. Alignment was made with the MUSCLE alignment tool (EMBL EBI, Cambridge, UK).

142

1 MSMKKTGSFA EFHVNMIERV MVRPCLPSPK TILPLSAIDN MARAFSNVLL 50 51 VYAANMDRVS ADPAKVIREA LSKVLVYYYP FAGRLRNKEN GELEVECTGQ 100 101 GVLFLEAMAD SDLSVLTDLD NYNPSFQQLI FSLPQDTDIE DLHLLIVQVT 150 151 RFTCGGFVVG ANVYGSACDA KGFGQFLQSM AEMARGEVKP SIEPIWNREL 200 201 VKLEHCMPFR MSHLQIIHAP VIEEKFVQTS LVINFEIINH IRRRIMEERK 250 251 ESLSSFEIVA ALVWLAKIKA FQIPHSENVK LLFAMDLRRS FNPPLPHGYY 300 301 GNAFGIACAM DNVHDLLSGS LLRTIMIIKK SKFSLHKELN SKTVMSSSVV 350 351 DVNTKFEDVV SISDWRHSIY YEVDFGWGDA MNVSTMLQQQ EHEKSLPTYF 400 401 SFLQSTKNMP DGIKMLMFMP PSKLKKFKIE IEAMIKKYVT KVCPSKLPRK 450 451 HHHHHH 456 Figure A.2.2. HisBAPT amino acid sequence for expression in P. pastoris. Size is 51.9 kDa.

1 MSMKKTGSFA EFHVNMIERV MVRPCLPSPK TILPLSAIDN MARAFSNVLL 50 51 VYAANMDRVS ADPAKVIREA LSKVLVYYYP FAGRLRNKEN GELEVECTGQ 100 101 GVLFLEAMAD SDLSVLTDLD NYNPSFQQLI FSLPQDTDIE DLHLLIVQVT 150 151 RFTCGGFVVG ANVYGSACDA KGFGQFLQSM AEMARGEVKP SIEPIWNREL 200 201 VKLEHCMPFR MSHLQIIHAP VIEEKFVQTS LVINFEIINH IRRRIMEERK 250 251 ESLSSFEIVA ALVWLAKIKA FQIPHSENVK LLFAMDLRRS FNPPLPHGYY 300 301 GNAFGIACAM DNVHDLLSGS LLRTIMIIKK SKFSLHKELN SKTVMSSSVV 350 351 DVNTKFEDVV SISDWRHSIY YEVDFGWGDA MNVSTMLQQQ EHEKSLPTYF 400 401 SFLQSTKNMP DGIKMLMFMP PSKLKKFKIE IEAMIKKYVT KVCPSKLPRK 450 451 RPPAYVEQKL ISEEDLNSAV DHHHHHH 477 Figure A.2.3. cMycHisBAPT amino acid sequence for expression in P. pastoris. The Myc epitope tag is underlined. Size of protein is 54.3 kDa.

1 MGSSHHHHHH SSGLVPRGSH MKKTGSFAEF HVNMIERVMV RPCLPSPKTI 50 51 LPLSAIDNMA RAFSNVLLVY AANMDRVSAD PAKVIREALS KVLVYYYPFA 100 101 GRLRNKENGE LEVECTGQGV LFLEAMADSD LSVLTDLDNY NPSFQQLIFS 150 151 LPQDTDIEDL HLLIVQVTRF TCGGFVVGAN VYGSACDAKG FGQFLQSMAE 200 201 MARGEVKPSI EPIWNRELVK LEHCMPFRMS HLQIIHAPVI EEKFVQTSLV 250 251 INFEIINHIR RRIMEERKES LSSFEIVAAL VWLAKIKAFQ IPHSENVKLL 300 301 FAMDLRRSFN PPLPHGYYGN AFGIACAMDN VHDLLSGSLL RTIMIIKKSK 350 351 FSLHKELNSK TVMSSSVVDV NTKFEDVVSI SDWRHSIYYE VDFGWGDAMN 400 401 VSTMLQQQEH EKSLPTYFSF LQSTKNMPDG IKMLMFMPPS KLKKFKIEIE 450 451 AMIKKYVTKV CPSKL 465 Figure A.2.4. NterBAPT amino acid sequence for expression in E. coli. Size is 52.7 kDa.

1 MKKTGSFAEF HVNMIERVMV RPCLPSPKTI LPLSAIDNMA RAFSNVLLVY 50 51 AANMDRVSAD PAKVIREALS KVLVYYYPFA GRLRNKENGE LEVECTGQGV 100 101 LFLEAMADSD LSVLTDLDNY NPSFQQLIFS LPQDTDIEDL HLLIVQVTRF 150 151 TCGGFVVGAN VYGSACDAKG FGQFLQSMAE MARGEVKPSI EPIWNRELVK 200 201 LEHCMPFRMS HLQIIHAPVI EEKFVQTSLV INFEIINHIR RRIMEERKES 250 251 LSSFEIVAAL VWLAKIKAFQ IPHSENVKLL FAMDLRRSFN PPLPHGYYGN 300 301 AFGIACAMDN VHDLLSGSLL RTIMIIKKSK FSLHKELNSK TVMSSSVVDV 350 351 NTKFEDVVSI SDWRHSIYYE VDFGWGDAMN VSTMLQQQEH EKSLPTYFSF 400 401 LQSTKNMPDG IKMLMFMPPS KLKKFKIEIE AMIKKYVTKV CPSKLKLAAA 450 451 LEHHHHHH 458 Figure A.2.5. CterBAPT amino acid sequence for expression in E. coli. Size is 52.1 kDa.

143

atgggcaagaaaaccggtagctttgccgaatttcatgtcaatatgatcgaacgtgtgatg M G K K T G S F A E F H V N M I E R V M gtgcgtccgtgcctgccgtcgccgaaaacgattctgccgctgtcggcgattgataacatg V R P C L P S P K T I L P L S A I D N M gcgcgcgcctttagcaatgttctgctggtctacgcggccaatatggatcgcgtttccgca A R A F S N V L L V Y A A N M D R V S A gacccggctaaagtcatccgtgaagcactgtcaaaagtgctggtttattactatccgttt D P A K V I R E A L S K V L V Y Y Y P F gctggccgcctgcgtaacaaagaaaatggtgaactggaagtggaatgcaccggccagggt A G R L R N K E N G E L E V E C T G Q G gttctgttcctggaagcaatggctgattccgacctgtcagtgctgacggatctggacaac V L F L E A M A D S D L S V L T D L D N tacaatccgtcctttcagcaactgattttctcactgccgcaagataccgacattgaagat Y N P S F Q Q L I F S L P Q D T D I E D ctgcacctgctgatcgtccaggtgacccgctttacgtgcggcggtttcgtggttggcgcg L H L L I V Q V T R F T C G G F V V G A aatgtttatggttctgcgtgtgacgccaaaggctttggtcaattcctgcagtcgatggca N V Y G S A C D A K G F G Q F L Q S M A gaaatggctcgtggcgaagtgaaaccgagcattgaaccgatctggaaccgcgaactggtt E M A R G E V K P S I E P I W N R E L V aaactggaacattgcatgccgtttcgtatgtcccatctgcaaattatccacgccccggtg K L E H C M P F R M S H L Q I I H A P V attgaagaaaaatttgtccagaccagcctggtgatcaacttcgaaattatcaatcacatt I E E K F V Q T S L V I N F E I I N H I cgtcgccgtatcatggaagaacgtaaagaaagcctgagctcttttgaaatcgtggcagct R R R I M E E R K E S L S S F E I V A A ctggtttggctggcgaaaattaaagccttccagatcccgcattctgaaaacgtgaaactg L V W L A K I K A F Q I P H S E N V K L ctgtttgcaatggatctgcgccgtagtttcaacccgccgctgccgcatggctactatggt L F A M D L R R S F N P P L P H G Y Y G aatgcctttggcattgcgtgtgcgatggataacgttcacgacctgctgtcgggtagcctg N A F G I A C A M D N V H D L L S G S L ctgcgcaccatcatgatcatcaaaaaatcgaaattcagcctgcataaagaactgaattca L R T I M I I K K S K F S L H K E L N S aaaaccgtcatgagttcctcagtcgtggatgtgaacacgaaatttgaagatgttgtctct K T V M S S S V V D V N T K F E D V V S attagtgactggcgtcatagtatctactatgaagttgatttcggctggggtgacgcgatg I S D W R H S I Y Y E V D F G W G D A M aatgtctctacgatgctgcagcaacaggaacacgaaaaaagtctgccgacctatttttct N V S T M L Q Q Q E H E K S L P T Y F S ttcctgcagagtacgaaaaacatgccggatggtattaaaatgctgatgtttatgccgccg F L Q S T K N M P D G I K M L M F M P P agcaaactgaaaaaattcaaaattgaaattgaagcgatgattaaaaaatatgtcacgaaa S K L K K F K I E I E A M I K K Y V T K gtctgtccgtctaaactgaagcttgcggccgcactcgagcaccaccaccaccaccactga V C P S K L K L A A A L E H H H H H H -

Figure A.2.6. Codon optimized bapt gene sequence. A C-terminal polyhistidine tag (optBAPT) (52.1 kDa) is attached and the translated protein sequence for expression in E. coli is shown (Genscript, Piscataway, NJ).

144

1 MKIEEGKLVI WINGDKGYNG LAEVGKKFEK DTGIKVTVEH PDKLEEKFPQ 50 51 VAATGDGPDI IFWAHDRFGG YAQSGLLAEI TPDKAFQDKL YPFTWDAVRY 100 101 NGKLIAYPIA VEALSLIYNK DLLPNPPKTW EEIPALDKEL KAKGKSALMF 150 151 NLQEPYFTWP LIAADGGYAF KYENGKYDIK DVGVDNAGAK AGLTFLVDLI 200 201 KNKHMNADTD YSIAEAAFNK GETAMTINGP WAWSNIDTSK VNYGVTVLPT 250 251 FKGQPSKPFV GVLSAGINAA SPNKELAKEF LENYLLTDEG LEAVNKDKPL 300 301 GAVALKSYEE ELAKDPRIAA TMENAQKGEI MPNIPQMSAF WYAVRTAVIN 350 351 AASGRQTVDE ALKDAQTNSS SNNNNNNNNN NLGIEGRISE FGSMGKKTGS 400 401 FAEFHVNMIE RVMVRPCLPS PKTILPLSAI DNMARAFSNV LLVYAANMDR 450 451 VSADPAKVIR EALSKVLVYY YPFAGRLRNK ENGELEVECT GQGVLFLEAM 500 501 ADSDLSVLTD LDNYNPSFQQ LIFSLPQDTD IEDLHLLIVQ VTRFTCGGFV 550 551 VGANVYGSAC DAKGFGQFLQ SMAEMARGEV KPSIEPIWNR ELVKLEHCMP 600 601 FRMSHLQIIH APVIEEKFVQ TSLVINFEII NHIRRRIMEE RKESLSSFEI 650 651 VAALVWLAKI KAFQIPHSEN VKLLFAMDLR RSFNPPLPHG YYGNAFGIAC 700 701 AMDNVHDLLS GSLLRTIMII KKSKFSLHKE LNSKTVMSSS VVDVNTKFED 750 751 VVSISDWRHS IYYEVDFGWG DAMNVSTMLQ QQEHEKSLPT YFSFLQSTKN 800 801 MPDGIKMLMF MPPSKLKKFK IEIEAMIKKY VTKVCPSKLK LAAALEHHHH 850 851 HH 852

Figure A.2.7. MBP-CterBAPT (95.2 kDa) amino acid sequence for expression in E. coli. The MBP sequence is underlined.

A. B.

40

) 5

30

10 × 20

10 Intensity ( Intensity 0 0 0.01 0.02 0.03 0.04 BadA (mg/mL)

Figure A.2.8. Quantification of BAPT by band densitometry. A. Representative band densities on a silver- stained SDS-PAGE gel. Lanes 1-7: Dilutions of the standard BadA (2.4-38 µg/mL), Lane A: NterBAPT (47 % purity). Lane B: NterBAPT (76% purity). Arrows represent visible protein bands. The NterBAPT band is labeled. B. Standard curve generated from BadA band densities.

145

Figure A.2.9. SDS-PAGE gel of purified NterBAPT and MBP-CterBAPT for proteomics analysis. NterBAPT (52.7 kDa) (lane 1) contains several proteins estimated at 70, 60, 53, and 25 kDa. MBP-CterBAPT (95.2 kDa) (lane 2) contains several proteins at 95, 70, 60, and 53 kDa. Ladder is the PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA).

146

REFERENCES

147 REFERENCES

1. St-Pierre, B., and De Luca, V. (2000) Evolution of acyltransferase genes: Origin and diversification of the BAHD superfamily of acyltransferases involved in secondary metabolism, In Evolution of Metabolic Pathways (Romeo, J. T., Ibrahim, R., Varin, L., and DeLuca, V., Eds.), pp 285-315. 2. Molina, I., and Kosma, D. (2015) Role of HXXXD-motif/BAHD acyltransferases in the biosynthesis of extracellular , Plant Cell Rep. 34, 587-601. 3. Luo, J., Nishiyama, Y., Fuell, C., Taguchi, G., Elliott, K., Hill, L., Tanaka, Y., Kitayama, M., Yamazaki, M., Bailey, P., Parr, A., Michael, A. J., Saito, K., and Martin, C. (2007) Convergent evolution in the BAHD family of acyl transferases: Identification and characterization of anthocyanin acyl transferases from Arabidopsis thaliana, Plant J. 50, 678-695. 4. Suzuki, H., Nishino, T., and Nakayama, T. (2007) Anthocyanin acyltransferase engineered for the synthesis of a novel polyacylated anthocyanin, Plant Biotechnol. 24, 495-501. 5. Unno, H., Ichimaida, F., Suzuki, H., Takahashi, S., Tanaka, Y., Saito, A., Nishino, T., Kusunoki, M., and Nakayama, T. (2007) Structural and mutational studies of anthocyanin malonyltransferases establish the features of BAHD , J. Biol. Chem. 282, 15812-15822. 6. El-Sharkawy, I., Manríquez, D., Flores, F., Regad, F., Bouzayen, M., Latché, A., and Pech, J.-C. (2005) Functional characterization of a melon alcohol acyl- transferase gene gamily involved in the biosynthesis of ester volatiles. Identification of the crucial role of a residue for enzyme activity, Plant Mol. Biol. 59, 345-362. 7. D'Auria, J. C., Chen, F., and Pichersky, E. (2002) Characterization of an acyltransferase capable of synthesizing benzylbenzoate and other volatile esters in flowers and damaged leaves of Clarkia breweri, Plant Physiol. 130, 466-476. 8. Muroi, A., Ishihara, A., Tanaka, C., Ishizuka, A., Takabayashi, J., Miyoshi, H., and Nishioka, T. (2009) Accumulation of hydroxycinnamic acid induced by pathogen infection and identification of agmatine coumaroyltransferase in Arabidopsis thaliana, Planta 230, 517-527. 9. Yu, X. H., Gou, J. Y., and Liu, C. J. (2009) BAHD superfamily of acyl-CoA dependent acyltransferases in Populus and Arabidopsis: Bioinformatics and gene expression, Plant Mol. Biol. 70, 421-442. 10. D'Auria, J. C. (2006) Acyltransferases in plants: A good time to be BAHD, Curr. Opin. Plant Biol. 9, 331-340.

148 11. Ma, X., Koepke, J., Panjikar, S., Fritzsch, G., and Stockigt, J. (2005) Crystal structure of vinorine synthase, the first representative of the BAHD superfamily, J. Biol. Chem. 280, 13576-13583. 12. Lallemand, L. A., Zubieta, C., Lee, S. G., Wang, Y., Acajjaoui, S., Timmins, J., McSweeney, S., Jez, J. M., McCarthy, J. G., and McCarthy, A. A. (2012) A structural basis for the biosynthesis of the major chlorogenic acids found in coffee, Plant Physiol. 160, 249-260. 13. Walker, A. M., Hayes, R. P., Youn, B., Vermerris, W., Sattler, S. E., and Kang, C. (2013) Elucidation of the structure and of sorghum hydroxycinnamoyltransferase and its structural relationship to other coenzyme A- dependent transferases and synthases, Plant Physiol. 162, 640-651. 14. Manjasetty, B. A., Yu, X. H., Panjikar, S., Taguchi, G., Chance, M. R., and Liu, C. J. (2012) Structural basis for modification of flavonol and naphthol glucoconjugates by Nicotiana tabacum malonyltransferase (NtMaT1), Planta 236, 781-793. 15. Morales-Quintana, L., Moya-León, M., and Herrera, R. (2015) Computational study enlightens the structural role of the alcohol acyltransferase DFGWG motif, J. Mol. Model. 21, 1-10. 16. Walker, K., Ketchum, R. E. B., Hezari, M., Gatfield, D., Goleniowski, M., Barthol, A., and Croteau, R. (1999) Partial purification and characterization of acetyl coenzyme A: Taxa-4(20),11(12)-dien-5--ol O-acetyl transferase that catalyzes the first acylation step of Taxol biosynthesis, Arch. Biochem. Biophys. 364, 273-279. 17. Nawarathne, I. N., and Walker, K. D. (2010) Point mutations (Q19P and N23K) increase the operational solubility of a 2--O-benzoyltransferase that conveys various acyl groups from CoA to a taxane acceptor, J. Nat. Prod. 73, 151-159. 18. Walker, K., and Croteau, R. (2000) Taxol biosynthesis: Molecular cloning of a benzoyl-CoA : taxane 2--O-benzoyltransferase cDNA from Taxus and functional expression in Escherichia coli, P. Natl. Acad. Sci. U.S.A. 97, 13591-13596. 19. Ondari, M. E., and Walker, K. D. (2008) The Taxol pathway 10-O-acetyltransferase shows regioselective promiscuity with the oxetane hydroxyl of 4-deacetyltaxanes, J. Am. Chem. Soc. 130, 17187-17194. 20. Walker, K., Fujisaki, S., Long, R., and Croteau, R. (2002) Molecular cloning and heterologous expression of the C-13 phenylpropanoid side chain-CoA acyltransferase that functions in Taxol biosynthesis, P. Natl. Acad. Sci. U.S.A. 99, 12715-12720. 21. Nevarez, D. M., Mengistu, Y. A., Nawarathne, I. N., and Walker, K. D. (2009) An N-aroyltransferase of the BAHD superfamily has broad aroyl CoA specificity in vitro with analogues of N-dearoylpaclitaxel, J. Am. Chem. Soc. 131, 5994-6002.

149 22. Long, R. M., Lagisetti, C., Coates, R. M., and Croteau, R. B. (2008) Specificity of the N-benzoyl transferase responsible for the last step of Taxol biosynthesis, Arch. Biochem. Biophys. 477, 384-389. 23. Walker, K., Long, R., and Croteau, R. (2002) The final acylation step in Taxol biosynthesis: Cloning of the taxoid C13-side-chain N-benzoyltransferase from Taxus, P. Natl. Acad. Sci. U.S.A. 99, 9166-9171. 24. Bayer, A., Ma, X. Y., and Stockigt, J. (2004) Acetyltransfer in natural product biosynthesis - functional cloning and molecular analysis of vinorine synthase, Bioorg. Med. Chem. 12, 2787-2795. 25. Schilmiller, A. L., Charbonneau, A. L., and Last, R. L. (2012) Identification of a BAHD acetyltransferase that produces protective acyl sugars in tomato trichomes, P. Natl. Acad. Sci. U.S.A. 109, 16377-16382. 26. Ojima, I., Chen, J., Sun, L., Borella, C. P., Wang, T., Miller, M. L., Lin, S., Geng, X., Kuznetsova, L., and Qu, C. (2008) Design, synthesis, and biological evaluation of new-generation taxoids, J. Med. Chem. 51, 3203-3221. 27. Ojima, I., Park, Y. H., Fenoglio, I., Duclos, O., Sun, C. M., Kuduk, S. D., Zucco, M., Appendino, G., Pera, P., Veith, J. M., Bernacki, R. J., Bissery, M. C., Combeau, C., Vrignaud, P., Riou, J. F., and Lavelle, F. (1995) Synthesis and structure-activity relationships of new taxoids, In Taxane Anticancer Agents: Basic Science and Current Status (Georg, G. I., Chem, T. T., Ojima, I., and Vyas, D. M., Eds.), pp 262-275, Amer. Chemical Soc., Washington. 28. Ali, S. M., Hoemann, M. Z., Aube, J., Mitscher, L. A., Georg, G. I., McCall, R., and Jayasinghe, L. R. (1995) Novel cytotoxic 3'-(tert-Butyl) 3'-diphenyl analogs of paclitaxel and docetaxel, J. Med. Chem. 38, 3821-3828. 29. Georg, G. I., Cheruvallath, Z. S., Himes, R. H., Mejillano, M. R., and Burke, C. T. (1992) Synthesis of biologically active Taxol analogues with modified phenylisoserine side chains, J. Med. Chem. 35, 4230-4237. 30. Georg, G. I., Cheruvallath, Z. S., Harriman, G. C., Hepperle, M., Park, H., and Himes, R. H. (1994) Synthesis and biology of substituted 3′-phenyl Taxol analogues, Bioorg. Med. Chem. Lett. 4, 2331-2336. 31. Boge, T. C., Himes, R. H., Vander Velde, D. G., and Georg, G. I. (1994) The effect of the aromatic rings of Taxol on biological activity and solution conformation: Synthesis and evaluation of saturated Taxol and Taxotere analogues, J. Med. Chem. 37, 3337-3343. 32. Georg, G. I., Harriman, G. C., Hepperle, M., and Himes, R. H. (1994) Heteroaromatic Taxol analogues: The chemistry and biological activities of 3′-furyl and 3′-pyridyl substituted taxanes, Bioorg. Med. Chem. Lett. 4, 1381-1384.

150 33. Kirikae, T., Ojima, I., Kirikae, F., Ma, Z., Kuduk, S. D., Slater, J. C., Takeuchi, C. S., Bounaud, P.-Y., and Nakano, M. (1996) Structural requirements of taxoids for nitric oxide and tumor necrosis factor production by murine macrophages, Biochem. Bioph. Res. Co. 227, 227-235. 34. Pham, K., LaForge, K., and Kreek, M. (1998) Sticky-end PCR: New method for subcloning, BioTechniques 25, 206-208. 35. Muchiri, R., and Walker, K. D. (2012) Taxol biosynthesis: Tyrocidine synthetase A catalyzes the production of phenylisoserinyl CoA and other amino phenylpropanoyl thioesters, Chem. Biol. 19, 679-685. 36. Tegel, H., Tourle, S., Ottosson, J., and Persson, A. (2010) Increased levels of recombinant human proteins with the Escherichia coli strain Rosetta(DE3), Protein Expres. Purif. 69, 159-167. 37. Sorensen, H. P., and Mortensen, K. K. (2005) Soluble expression of recombinant proteins in the of Escherichia coli, Microb. Cell Fact. 4. 38. Mootz, H. D., and Marahiel, M. A. (1997) The tyrocidine biosynthesis operon of Bacillus brevis: Complete sequence and biochemical characterization of functional internal adenylation domains, J. Bacteriol. 179, 6843-6850. 39. Villiers, B. R. M., and Hollfelder, F. (2009) Mapping the limits of substrate specificity of the adenylation domain of TycA, Chembiochem 10, 671-682. 40. Muchiri, R. N. (2015) Repurposing a non-ribosomal peptide synthetase (tyrocidine synthetase A) for amino acyl CoA biosynthesis, Michigan State University. 41. Zeper, A., Fang, Q., Liang, X., and Takayama, M. (2000) Study on fragmentation behavior of taxoids by tandem mass spectrometry, Chinese Sci. Bull. 45, 688-698. 42. McClure, T. D., Schram, K. H., and Reimer, M. L. (1992) The mass spectrometry of Taxol, J. Am. Soc. Mass Spectrom. 3, 672-679. 43. Tian, J., and Stella, V. J. (2008) Degradation of paclitaxel and related compounds in aqueous solutions II: Nonepimerization degradation under neutral to basic pH conditions, J. Pharm. Sci. 97, 3100-3108. 44. Tian, J., and Stella, V. J. (2008) Degradation of paclitaxel and related compounds in aqueous solutions I: Epimerization, J. Pharm. Sci. 97, 1224-1235. 45. Kapust, R. B., and Waugh, D. S. (1999) Escherichia coli maltose-binding protein is uncommonly effective at promoting the solubility of polypeptides to which it is fused, Protein Sci. 8, 1668-1674. 46. Davis, G. D., Elisee, C., Newham, D. M., and Harrison, R. G. (1999) New fusion protein systems designed to give soluble expression in Escherichia coli, Biotechnol. Bioeng. 65, 382-388.

151 47. De Marco, V., Stier, G., Blandin, S., and de Marco, A. (2004) The solubility and stability of recombinant proteins are increased by their fusion to NusA, Biochem. Bioph. Res. Co. 322, 766-771. 48. Lavallie, E. R., Diblasio, E. A., Kovacic, S., Grant, K. L., Schendel, P. F., and McCoy, J. M. (1993) A thioredoxin gene fusion expression system that circumvents inclusion body formation in the Escherichia coli cytoplasm, Bio-Technol. 11, 187- 193. 49. Butt, T. R., Edavettal, S. C., Hall, J. P., and Mattern, M. R. (2005) SUMO fusion technology for difficult-to-express proteins, Protein Expres. Purif. 43, 1-9. 50. Costa, S., Almeida, A., Castro, A., and Domingues, L. (2014) Fusion tags for protein solubility, purification, and immunogenicity in Escherichia coli: The novel Fh8 system, Front. Microbiol. 5. 51. Young, C. L., Britton, Z. T., and Robinson, A. S. (2012) Recombinant protein expression and purification: A comprehensive review of affinity tags and microbial applications, Biotechnol. J. 7, 620-634. 52. Ben-Zvi, A. P., and Goloubinoff, P. (2001) Mechanisms of disaggregation and refolding of stable protein aggregates by molecular chaperones, J. Struct. Biol. 135, 84-93. 53. Georg, G. I., Harriman, G. C. B., Hepperle, M., Clowers, J. S., VanderVelde, D. G., and Himes, R. H. (1996) Synthesis, conformational analysis, and biological evaluation of heteroaromatic taxanes, J. Org. Chem. 61, 2664-2676. 54. Smyth, D. R., Mrozkiewicz, M. K., McGrath, W. J., Listwan, P., and Kobe, B. (2003) Crystal structures of fusion proteins with large-affinity tags, Protein Sci. 12, 1313-1322. 55. Liu, Y. F., Manna, A., Li, R. G., Martin, W. E., Murphy, R. C., Cheung, A. L., and Zhang, G. Y. (2001) Crystal structure of the SarR protein from Staphylococcus aureus, P. Natl. Acad. Sci. U.S.A. 98, 6877-6882. 56. Jung, J., Bashiri, G., Johnston, J. M., Brown, A. S., Ackerley, D. F., and Baker, E. N. (2014) Crystal structure of the essential Mycobacterium tuberculosis phosphopantetheinyl transferase PptT, solved as a fusion protein with maltose binding protein, J. Struct. Biol. 188, 274-278.

152 Chapter 4. Paclitaxel analog biosynthesis from baccatin III with a four-enzyme in vitro system and characterization of baccatin III-3-amino-13-O- phenylpropanoyl CoA transferase (BAPT)

4.1 Introduction

Clinical approval of paclitaxel for refractory ovarian cancer in 1992 increased demand for this structurally complex drug. Its subsequent approval for additional cancer chemotherapy treatments including multiple combination therapies, applications in restenosis treatments, and its growing usage in new markets further increased demand and production requirements (1). As described in Chapter 1, paclitaxel purification from Taxus sp. yew trees is environmentally unsustainable, and complete chemical synthesis is cost prohibitive

(1, 2). Current production of paclitaxel is largely supplied through suspended plant cell culture fermentation (PCF) and semi-synthetic methods (See Chapter 1). Although production by these techniques dramatically improved upon paclitaxel yields isolated from yew tree bark (0.001–0.05% w/w) shortages of paclitaxel and its analogs occur annually and can be found on the Food and Drug Administration’s shortages list (3). These shortages negatively impact patient healthcare (2, 4-7). Causes for these shortages are related to high demand, manufacturing quality problems, shortages of raw materials, stock-piling by large hospitals, and other economic disincentives (8).

4.1.1 Engineering the paclitaxel biosynthetic pathway

Major efforts are underway to engineer the paclitaxel biosynthetic pathway in heterologous organisms including Escherichia coli, Saccharomyces cerevisiae, and Arabidopsis thaliana (9-13). These efforts are limited by multiple factors including incomplete characterization of the paclitaxel pathway (Figure 1.4). For example, the biosynthesis of

153 10-deacetylbaccatin III (10-DAB) from taxadiene-5α-ol requires enzymes with unidentified substrate specificities as well as an ambiguous order of the enzymatic reactions (14-19). Additionally, genes for several pathway enzymes remain unidentified, including a proposed C1 hydroxylase (T1βOH), C9 hydroxylase (T9αOH), C4-C20- epoxidase, an oxomutase, a C9 oxidase, and a C2′-hydroxylase (C2′αOH). Most recently, the Taxus baccata β-phenylalanine CoA ligase (βPhL) was identified in jasmonic acid- elicited Taxus cell cultures using amplified cDNA screening techniques (20). Additional unidentified gene transcripts homologous to cytochrome P450 hydroxylases, epoxidases, oxidases, and oxomutases were found, but biochemical characterization and substrate specificity studies with these putative pathway enzymes remain daunting challenges (20).

Over 400 taxoids are derived from the paclitaxel precursor taxa-4(5),11(12)-diene to form a complicated secondary metabolite network (21). Enzymes within this network are known to divert flux from paclitaxel biosynthesis (16, 22, 23). Defining a linear biosynthetic route to paclitaxel with limited side-products is difficult given these concerns (19).

Engineering the paclitaxel pathway in bacteria or yeast has benefits over current plant cell fermentation methods. Many paclitaxel biosynthetic enzymes accept multiple substrates and produce multiple products. Moving enzymes to a system with fewer “off- pathway” enzymes will likely reduce by-products and increase flux to the desired product.

Purification of the final product would likely be easier without hundreds of intermediates and by-products produced in Taxus cells. Characteristics required of such a system include an in-depth understanding of pathway enzymes in terms of activity, substrate specificity,

154

Figure 4.1. Condensed scheme of paclitaxel biosynthesis. Abbreviations: IPPI, isopentenyl diphosphate isomerase, GGPPS, geranylgeranyl diphosphate synthase, TS, taxadiene synthase, T5αOH, taxadiene-5-α-ol hydroxylase, T13αOH, taxane-13α-hydroxylase, TAT, taxadiene-5α-ol-O-acetyltransferase T10βOH, taxane-10β-hydroxylase, DBAT, 10-deacetylbaccatin III 10-O-acetyltransferase, BAPT, baccatin III: 3- amino-13-O-phenylpropanoyl CoA transferase, PAM, phenylalanine aminomutase, β-PhL, β-phenylalanine CoA ligase, NDTNBT, N-debenzoyltaxol-N-benzoyltransferase. Enzymes not shown: T1βOH, taxane-1β- hydroxylase, T2αOH, taxane-2α-hydroxylase, T7βOH, taxane-7β-hydroxylase, T9αOH, taxane-9α- hydroxylase, TBT, taxane-2α-O-benzoyltransferase, epoxidase*, oxomutase*, C9-oxidase*, C2′- hydroxylase*. Enzymes marked with (*) are unidentified.

155 cellular location and regulatory control (24). Bacteria and yeast are genetically tractable, can be grown large-scale in a controlled environment, generally use an inexpensive carbon source, are scalable, and have simpler extraction and purification procedures than those from plant cells (24, 25). A wealth of knowledge regarding metabolic engineering and synthetic biology already exists for bacteria and yeast. The most successful example of drug production via engineered S. cerevisiae is the antimalarial drug , originally from Artemisia annua (26). Biosynthesis of the intermediate artemisinic acid was achieved at high yields (initially 100 mg/mL (27), later 25 g/L (28)) and then developed by semi- synthesis to artemisinin or artesunate, an analog (28). This bioengineering approach to artemisinin synthesis is currently in production and proposed to reduce drug costs and make anti-malarial drugs more available worldwide (29).

To date, paclitaxel bio-engineering efforts have focused on early pathway steps

(Figure 1.4). Taxadiene synthase (TS) was first engineered in E. coli to successfully produce taxa-(4)5,(11)12-diene (taxadiene) (30). Heterologous expression of TS in plants and yeast was also successful (10, 11). An ambitious early effort to express eight biosynthetic enzymes including GGDPS, TS, 5αOH, 13αOH, 10β-OH, TAT, TBT, and

DBAT in S. cerevisiae demonstrated production of trace levels of taxadiene-5α-ol.

Subsequent enzymatic products were undetectable and the first C5-oxidation step was concluded to be a bottleneck (10). Later efforts produced taxadiene in E. coli at g/L quantities and also demonstrated 5αOH activity (12). To actively express 5αOH in E. coli, the natural gene was fused to a Taxus reductase and modified in the transmembrane domains (12). A 98% conversion of taxadiene to taxadiene-5α-ol (~50%) and a by-product,

5(12)-oxa-3(11)-cyclotaxane (OCT) (~50%) were observed (Figure 4.2). The OCT by-

156

Figure 4.2. Products of an engineered chimeric cytochrome P450 taxadiene-5α-hydroxylase expressed in E. coli. The desired product, taxadiene-5α-ol, and a side product, 5(12)-oxa-3(11)-cyclotaxane (OCT), were produced in approximately equal amounts.

product was also detected during the heterologous expression of TS and native T5αOH in tobacco (31). However, no natural taxadiene-5α-ol was detected in tobacco plants. The by- product likely resulted from the unique properties of the T5αOH enzyme (12). Whether the chimeric enzyme was better or worse than the wild-type T5αOH due to engineering effects, or the environment of a heterologous bacterial system is unclear.

Challenges in engineering Taxus sp. cytochrome P450 monooxygenase enzymes to function in bacteria include removing the requirement for P450 hydroxylases to be membrane-bound on the in the plant cell, while retaining activity

(32-35). There are ten predicted oxidation steps required to biosynthesize paclitaxel. In order to successfully engineer the entire pathway in bacteria, all P450 hydroxylases, the epoxidase, and the oxidase would likely need to be engineered to optimize expression and activity. Partner reductases are generally also required for heterologous engineering with

P450 hydroxylases (33). Recently, a mutualistic co-culture of both E. coli and S. cerevisiae split early biosynthetic genes between the two microorganisms. E. coli produced and

157 secreted taxadiene, the S. cerevisiae oxidized the taxadiene with two fusion P450 enzymes- the 5α-hydroxylase/reductase, the 10-β-hydroxylase-, and the acetyltransferase, TAT (13).

Isotopic labeling and mass spectrometry suggested the observed diol product originated from taxadiene (13). If additional by-products are formed or substrate specificity is altered during the enzyme engineering process, pathway flux and product purification will likely become increasingly complex and costly.

Most current knowledge regarding the regulation of paclitaxel biosynthesis is the result of elicitor studies on Taxus cell cultures. Genetic transformation of Taxus sp. is developing, but not yet routine (36-38). Methyl jasmonate is well-known to have positive effects on paclitaxel accumulation in cell cultures (39, 40). Recent work used these effects to find several transcription factor repressors common to seven paclitaxel biosynthetic genes (41).

158

Figure 4.3. Coupled enzyme biosynthesis of paclitaxel. I. (2R,3S)-phenylisoserinyl CoA production by PheAT. II. Sidechain attachment to baccatin III by BAPT to produce the docetaxel precursor (R1 = H) (III). IV. BadA catalyzes the production of benzoyl CoA. V. NDTNBT transfers a benzoyl group to the 3′-N of N- debenzoylpaclitaxel to produce paclitaxel. Baccatin III: R1 = acetyl group. 10-Deacetylbaccatin III (10- DAB): R1 = H.

159 4.1.2 Proposed paclitaxel semi-biosynthesis

One possible approach to biosynthesize paclitaxel uses readily available precursors, baccatin III and 10-DAB, as starting materials in a 4-step enzyme system to produce paclitaxel (Figure 1.9). In lieu of finding and optimizing the unidentified 2′-hydroxylase

(2αOH), a phenylisoserine CoA ligase activity with (2R,3S)-phenylisoserine ((2R,3S)-

PhIS) was discovered in PheAT (truncated TycA), a module of the non-ribosomal peptide synthase (NRPS), tyrocidine synthase (42). Baccatin III: 3-amino-13-O-phenylpropanoyl

CoA transferase (BAPT) attaches the sidechain to baccatin III and is active with (2R,3S)- phenylisoserinyl CoA ((2R,3S)-PhIS CoA), albeit at a lower rate than (3R)-β-phenylalanyl

CoA ((3R)-β-Phe CoA) (43). N-debenzoyltaxol-N-benzoyltransferase (NDTNBT) performs the final N-benzoylation step with both N-debenzoyl-2’-deoxypaclitaxel and N- debenzoylpaclitaxel. A broad specificity benzoate CoA ligase (BadA) (see Chapter 2) was used to biosynthesize various acyl CoAs to serve as putative substrates for NDTNBT.

Engineering the genes encoding these four enzymes in a heterologous host will enable production of paclitaxel (and its analogs) (Table 4.1). Paclitaxel analogs already approved and those in clinical trials typically contain modified acyl groups. Precursors can potentially be biosynthesized using the four enzymes in this methodology (Figure 4.4). The costs and complexity of engineering the entire paclitaxel pathway are avoided, no new enzymes need to be characterized, and flux to side products is decreased. For example,

BAPT is selective for PhIS- and β-Phe, but not α-phenylalanyl CoA. It is unlikely that side products will form in the presence of amino acids such as α-phenylalanine.

160

Figure 4.4. Taxanes clinically approved and in clinical trials. A. The natural product paclitaxel. B. Docetaxel. C. Cabazitaxel. D. DJ-927 (Tesetaxel) completed phase II clinical trials for several cancers. E. BMS-275183 (Phase I – terminated). F. TL-00139 completed phase II clinical trials for malignant mesothelioma. G. MST-997 (Phase I- terminated). H. TPI-287 ongoing clinical trials for several cancers and Alzheimer’s disease. Brand names are listed in parentheses (clinicaltrials.gov).

161 Table 4.1. Simple analogs of paclitaxel with biological activity.

+ R1 R2 Biological Activity Assay Reference 3-chlorophenyl Acetyl M.1, B162 4-chlorophenyl Acetyl M, B16 (44) 4-methylphenyl Acetyl M, B16 4-methoxyphenyl Acetyl M, B16 2-furyl* Acetyl M, B16 (45) 3-furyl Acetyl M, B16 2-thiophen Acetyl A27803, PC34 (46) 4-methylphenyl H B16, P3885 2-fluorophenyl H B16, P388 4-fluorophenyl H B16, P388 (47) 4-chlorophenyl H B16, P388 4-methoxyphenyl H B16, P388 +Activities with the same or better activity than paclitaxel *Analog biosynthesized in this work 1In vitro microtubule assembly assay 2B16 melanoma cell line 3A2780 ovarian cancer cell line 4PC3 prostate cancer cell line 5P388 leukemia cell line

4.1.3 Applications in paclitaxel analog production

Substrate specificity studies show PheAT is active with a number of arylisoserines, including meta-substituted (2R,3S)-phenylisoserines, precursors for paclitaxel analogs with demonstrated cytotoxicity against multiple drug-resistant (MDR) cancer cell lines or currently in clinical trials (45, 48-54). Substrate specificity studies with a benzoate CoA ligase (BadA), (see Chapter 2) and N-benzoyltransferase (NDTNBT) have also been performed (55, 56). BadA has broad substrate specificity and single point mutations showed novel substrate activities as well. The Taxus N-benzoyltransferase (NDTNBT) is active with the following acyl acceptors: N-debenzoyl-2′-deoxypaclitaxel, N-debenzoyl-2′- deoxy-10-deacetylpaclitaxel, and N-debenzoylpaclitaxel. The acyl donor activity of

162 NDTNBT includes ortho-, meta-, and para-substituted benzoic acids with a preference for meta- or para-substituted benzoic acids (55). In addition to producing paclitaxel, it is likely this pathway can be used to produce analogs, or precursors to analogs of biological interest by simply changing the growth media to include relevant precursors such as baccatin III or

10-DAB, phenylisoserines, and benzoate surrogates. Many synthetic analogs have been tested in vitro but not further developed.

4.1.4 BAPT characterization and proof-of-principle for paclitaxel/analog

biosynthesis

One aim of this work was to biochemically characterize BAPT with (2R,3S)-PhIS CoA.

Although BAPT is more active with (3R)-β-Phe CoA, paclitaxel is biologically inactive without the 2′-hydroxyl group (57). In order to characterize Michaelis-Menten kinetic constants of BAPT, various synthetic and biosynthetic strategies were employed to obtain

(2R,3S)-PhIS CoA in preparative yields for the first time. BAPT kinetics were determined with both (3R)-β-Phe CoA and (2R,3S)-PhIS CoA. Proof-of-principle for the biosynthesis of N-debenzoylpaclitaxel and paclitaxel analogs are demonstrated.

163 4.2 Experimental

4.2.1 Materials

Coenzyme A (95%) was obtained from Lee Biosolutions (St. Louis, MO). (3R)-β- phenylalanine (98%) was purchased from Peptech (Burlington, MA). (2R,3S)- phenylisoserine (98%) was purchased from Waterstonetech, LLC (Carmel, IN). All taxanes (baccatin III (>98%), 10-deacetyl baccatin III (>98%), docetaxel (>98%) and paclitaxel (>98%) were purchased from Natland International Corporation (Research

Triangle Park, NC). Additional reagents were sourced as follows: HEPES (>99%)

(Fluka/Sigma Aldrich, St. Louis, MO), MOPS (>99%) (Research Products International,

Corp., Mt Prospect, IL), TEA (100%) (J. T. Baker, Center Valley, PA), acetic anhydride

(>99.4%) and trifluoroacetic acid (>99.5%) (EMD Chemicals, Billerica, MA), di-t-butyl- dicarbonate (>99%) and ethyl chloroformate (97%) (Sigma Aldrich, St. Louis, MO), C18 silica gel resin (carbon 23%, 40-63 µm) (Silicycle, Quebec City, Quebec, Canada). ATP, isopropyl β-D-1-thiogalactopyranoside (IPTG), kanamycin, phenylmethylsulfonyl fluoride

(PMSF), and tris(2-carboxylethyl)phosphineHCl (TCEP) were purchased from Gold Bio

(St. Louis, MO).

4.2.2 Chemical synthesis of (3R)-β-phenylalanyl CoA

Synthesis of (3R)-β-phenylalanyl CoA was adapted from a mixed anhydride intermediate method (43, 58-60). The group of (3R)-β-phenylalanine was protected as follows;

(3R)-β-phenylalanine (~243 µmol) was dissolved in 3:2 (v/v) 0.5 M NaHCO3: t-butanol (5 mL) and stirred at room temperature for 20 min. The solution was stirred on ice for 10 min prior to the dropwise addition of di-t-butyl dicarbonate (2 equiv). After 30 min, the flask was removed from ice and stirred at 22 °C for 2 to 4 h and followed by TLC (8:1:1 butanol:

164 water: acetic acid). Upon completion, water (5 mL) was added to the flask and adjusted to pH 2 with 1M HCl. The reaction was then extracted three times with ethyl acetate (20 mL), the organic fraction was dried (anhydrous Na2SO4), filtered, and evaporated under vacuum to yield R-N-Boc-3-amino-3-phenylpropanoic acid as a white powder (40.1 mg, 99%

1 yield). H-NMR (500 MHz, CDCl3) δ (ppm): 1.43 (m, H5), 2.88 (m, H1, H2), 5.12 (br d,

H3), 7.30 – 7.36 (aromatic protons). (Figure A.3.1 for proton numbering).

3R-N-Boc-3-amino-3-phenylpropanoic acid was resuspended in 5:2 (v/v) CH2Cl2:

THF (dry, 10 mL) and stirred under nitrogen. Triethylamine (TEA) was added (1.2 equiv) and stirred at 22 °C for 10 min. The carboxylate activating , ethyl chloroformate

(1.2 equiv) was added dropwise, and the reaction was monitored by TLC (1% acetic acid in diethyl ether) for 2 h. Solvents were evaporated to dryness under a nitrogen stream. The mixed anhydride product was resuspended in t-butanol (0.5 mL) and stirred. CoA (1 equiv)

(dissolved in 1 mL of 0.4 M NaHCO3) was added to the mixed anhydride intermediate and the reaction was monitored by TLC (5:3:2 (v/v/v) 1-butanol, water, acetic acid) for up to 4 h. The reaction was stopped by the addition of 1M HCl (pH 4). The identity of the acyl

CoA product was further verified by LC-ESI-MS.

Without further purification, the R-N-Boc-3-amino-3-phenylpropanoyl CoA was stirred on ice at 0 °C. TFA was added dropwise to a final concentration of 30% (v/v) and stirred for 30 min. The reaction was warmed to room temperature and stirred for 1 h, followed by 1 h on ice. The flask alternated between stirring on ice for 1 h and at room temperature for 1 h each for 3 cycles. The reaction progress was followed by TLC (5:3:2

(v/v/v) 1-butanol, water, acetic acid) and LC-ESI-MS. The crude product was back- extracted with ethyl acetate (3 x 15 mL) to remove t-butanol and organic contaminants.

165 The remaining aqueous phase (< pH 4) was dried under vacuum to remove excess TFA and then lyophilized. (3R)-β-phenylalanyl CoA was resuspended in water (pH 5, 0.5 mL) and purified by C18 column chromatography with a mobile phase of 0.05% TEA (aq.) with

2.5% acetonitrile. Fractions containing the product were pooled, desalted by running through Dowex 50 cation exchange resin, flash-frozen, and lyophilized. The (3R)-β- phenylalanyl CoA was characterized by NMR (Figure A.3.2), UV-HPLC (Figure A.3.3),

1 LC-ESI-MS (Figure A.3.4), and the Ellman assay. H-NMR (500 MHz, D2O) δ (ppm):

0.76 (br s, H11′′), 0.9 (br s, H10′′), 2.93 (m, H9′′), 3.22 (m, H2c), 3.29 - 3.46 (m, H5′′,

H8′′), 3.55 (q, H5′), 3.81 (m, H5′), 4.01 (s, H3′′), 4.20 (br s, H1′′), 4.54 (br s., H4′), 4.81 -

4.85 (m, H2′, H3′, H3c), 5.07 (m, NH2), 6.15 (m, 1′), 7.38 (aromatic H), 8.36 (br, s, adenine

H), 8.6 (m, adenine H); (Figure A.3.2 for proton numbering). LC-ESI-MS m/z 913.1428

- [M-H] ; calculated for C30H44N8O17P3S: 913.1736.

4.2.3 Method I. Synthesis of (2R,3S)-phenylisoserinyl CoA

N-Boc protection of (2R,3S)-phenylisoserine was performed in the same manner as for

(3R)-β-phenylisoserine, however this reaction was conducted in small scale (<10 mg, 98% yield) due to limited substrate availability. Mixed anhydride formation with ethyl chloroformate and thioesterification with CoA were performed as described for the synthesis of (3R)-β-phenylalanyl CoA. N-Boc protected intermediate (~9.8 mg, 98 % yield,

1 95% purity) was analyzed by NMR. H-NMR (500 MHz, CDCl3) δ (ppm): 1.43 (m, H5),

2.88 (m, H2), 5.12 (br. d., H3), 7.30 – 7.36 (aromatic protons). (See Figure A.3.10 for proton numbering).

166 4.2.4 Method II. Synthesis of (2R,3S)-phenylisoserinyl CoA

N-Boc protection of (2R,3S)-phenylisoserine (98% yield) at smallscale (<5 mg) was performed as described above. The hydroxyl group of N-Boc-(2R,3S)-phenylisoserine was protected with tert-butyldimethylsilyl chloride (TBDMS-Cl). Briefly, N-Boc-(2R,3S)- phenylisoserine (16 µmol) was dissolved in DMF (0.2 mL) and stirred with molecular sieves (4 Å) to remove residual water. Imidazole (2.5 equiv) was added to the solution and stirred (~30 min). TBDMS-Cl (1.2 equiv) was then added and allowed to stir for 18 h. If remaining starting material was visible by TLC, more TBDMS-Cl (0.1 equiv) was added.

The product was extracted twice with diethyl ether, then washed with brine to remove residual DMF. After removal of diethyl ether by evaporation, some residual DMF remained and converted yields were estimated to be 80-90% based on TLC. The identity of the product was confirmed by LC-ESI-MS. Without further purification, the mixed anhydride was made with ethyl chloroformate as described above and thiolation by CoA (0.8 equiv) was also performed as previously described. This method was not successful and did not yield any detectable product by the standard methods of analysis used earlier to characterize (3R)-β-Phe CoA. Therefore, this method was abandoned in favor of Method

III.

4.2.5 Method III. Synthesis of (2R,3S)-phenylisoserinyl CoA

N-Boc protection of (2R,3S)-phenylisoserine (98% yield) at smallscale (<5 mg) was performed as described above. Without protection of the 2-hydroxyl group, carboxylate activation was attempted by N,N'-dicyclohexylcarbodiimide (DCC)/ hydroxybenzotriazole

(HOBt) coupling in small scale (<5 mg). Briefly, N-Boc-(2R,3S)-phenylisoserine (25.6

µmol) was added to a round bottom flask, dissolved in 500 µL of 5:2 (v/v) dry

167 THF:CH2Cl2, and stirred for 10 min at 22 °C. TEA (1 equiv) was added and stirred for an additional 10 min. This solution was then added dropwise to a stirred solution of DCC (1.1 equiv) and the reaction progress was monitored by TLC. The HOBt (1.1 equiv) was pre- incubated with TEA (1 equiv) and solubilized in DMF before dropwise addition to the

DCC coupling reaction. The reaction was stirred under nitrogen for 18 h at 22 °C and monitored by TLC (1:9 MeOH: CHCl3 (v/v)). The reaction was filtered, dried under nitrogen and redissolved in t-butanol (0.5 mL). CoA (1 equiv) was dissolved in 0.4 M

Na2HCO3 (0.5 mL). Portions of CoA (50 µL) were added dropwise to the activated intermediate every 10 min. After addition of CoA, the reaction continued for an additional

~3 h and was monitored by TLC (5:3:2 n-butanol: water: acetic acid (v/v/v)). This method consistently yielded a maximum of detectable product (< 5%) by LC-ESI-MS. Therefore, this method was abandoned in favor of Method IV.

4.2.6 Method IV. Synthesis of (2R,3S)-phenylisoserinyl CoA

The method of DCC/HOBt coupling was modified to DCC/NHS (N-hydroxysuccinimide) coupling. The reaction was tested in small scale (~3.55 µmol) as described above for

DCC/HOBt coupling except NHS was added in place of HOBt. The thioesterification reaction was modified by dissolving CoA in degassed 0.4 M Na2HCO3 with 1 mM tris-(2- carboxyethyl) phosphine (TCEP). The activated (2R,3S)-phenylisoserinyl intermediate was added dropwise to the stirring solution of CoA over 30 min and allowed to run for 5 h. This method consistently yielded a maximum of detectable product (<5%) by LC-ESI-

MS. Therefore, this method was abandoned in favor of a biosynthetic method.

168 4.2.7 PheAT purification

PheAT was expressed in BL21(DE3) E. coli (Invitrogen, Carlsbad, CA) (48). A single colony was selected and used to inoculate a 100-mL culture of LB media containing 50

µg/mL kanamycin. The culture was grown overnight at 37 °C. Aliquots of this starter culture (7 mL) were used to inoculate a flask of fresh LB media (1 L) containing 50 µg/mL kanamycin. For large preparations of PheAT, 5 to 10 L were grown at once. This culture was grown at 37 °C until A600 = 0.6. Gene expression was induced by 0.2 mM IPTG, grown overnight (~16 h) at 18 °C, and cells harvested by centrifugation at 8,000g for 5 min. Pellets were resuspended in Buffer A (50 mM HEPES, 300 mM NaCl, 15 mM imidazole, 5% glycerol, pH 8.0) at 3 mL/g pellet (wet weight) with PMSF (1 mM). Cells were kept on ice, lysed with a Misonix XL 2020 sonicator and centrifuged at 18,000g for 20 min. The supernatant was further centrifuged in a Beckman Coulter Ultra Centrifuge at ~100000g for 1.5 h. The supernatant was loaded onto a Ni2+–NTA Qiagen column pre-equilibrated with Buffer A at 4 °C. The column was washed with 5 column volumes (CV) of Buffer A, and eluted with 3 CV of Buffer A containing 250 mM imidazole. The eluent was concentrated in a Millipore Amicon Ultra 30 kDa cutoff concentrator and buffer exchanged with 50 mM HEPES, pH 8.0. Final concentrations were ~50 mg/mL (estimated by the

Coomassie (Bradford) Protein Assays (Thermo Scientific Pierce, Grand Island, NY)).

Protein aliquots were flash-frozen in liquid nitrogen and stored at –80 °C.

4.2.8 Biosynthesis of (2R,3S)-phenylisoserinyl CoA

A large-scale preparative PheAT enzymatic assay was performed under the following conditions. A concentrated solution of PheAT (20 mL of 36 mg/mL) in 10 mM HEPES buffer, pH 8 was gently stirred at 23 °C. MgCl2(6H2O) (100 mg) and (2R,3S)-

169 phenylisoserine (0.17 mmol) were dissolved in the PheAT solution. Separately, ATP (0.17 mmol) and CoA (0.13 mmol) were dissolved in 2 mL each of 10 mM HEPES, pH 8.

Solutions of ATP and CoA were adjusted to pH 8 (0.5 M NaOH) before addition to the stirred PheAT solution. Total reaction volume was 20 mL. Reaction progress was monitored by UV-HPLC. Specifically, the peaks corresponding to CoA (4.7 min) and

(2R,3S)-phenylisoserinyl CoA (6.6 min) were monitored. After 7.5 h, 50% of the CoASH was converted to product and the reaction was stopped by addition of 8.8% formic acid to pH 4 to precipitate PheAT. The precipitated reaction was centrifuged at 5000g for 10 min.

The supernatant was collected, the pellet washed with water (pH 4 with formic acid) and re-centrifuged. Supernatants were combined and filtered through a Millipore Amicon Ultra

30 kDa concentration unit to remove trace protein. The flow-through was collected, flash- frozen in liquid nitrogen, and lyophilized. The lyophilized crude product was then dissolved in 2 mL of ultrapure water (pH 4), for preparative HPLC purification.

4.2.9 Purification of (2R,3S)-phenylisoserinyl CoA

(2R,3S)-phenylisoserinyl CoA (PhIS-CoA) residue was resuspended in 0.1% TFA in water and purified by preparative HPLC (Agilent 1100 Series). Approximately 20 aliquots (100

µL) of the filtered, lyophilized, crude mixture were loaded onto a 100 µL loop and purified on a preparative C18 column (Atlantis C18 OBD, 5 µm, 19 mm × 150 mm). The column was eluted at 4 mL/min with 2.5% solvent B (100% acetonitrile) and 97.5% solvent A

(0.1% trifluoroacetic acid in water) with a 4 min hold, a linear gradient to 30% solvent B over 15 min, then increased to 100% solvent B over 2 min, and finally lowered to 2.5% solvent B over 2 min. The effluent was monitored at 258 nm. Peak fractions were collected, flash-frozen, and lyophilized to yield (2R,3S)-phenylisoserinyl CoA (97.5% pure, 45.5 mg

170 (0.05 mmol) at 30% yield relative to (2R,3S)-phenylisoserine. The purified thioester product was analyzed by NMR (Figure A.3.11, Figure A.3.12), UV-HPLC (λ = 258 nm)

(Figure A.3.13), mass spectrometry (Figure A.3.14) and Ellman assay for free thiol. 1H

NMR (500 MHz, D2O, pH 3 with CD3COOD) : 0.59 (s, 11′′), 0.73 (s, 10′′), 2.16 (br s,

6′′), 2.72 (br s, 9′′), 2.95 – 3.30 (m, 5′′, 8′′), 3.37 (m, 5′), 3.64 (m, 5′), 3.85 (s, 3′′), 4.02 (br s, 1′′), 4.38 (d, J = 7.32 Hz, 2c), 4.46 (d, J = 6.35 Hz, 4′), 4.74 (m, 2′, 3′, 3c), 4.89 (br s,

NH2), 5.99 (d, J = 5.86 Hz, 1’), 7.20 (br s, aromatic H), 8.19 (s, adenine H), 8.44 (s, adenine

13 H). (See Figure A.3.11 for proton numbering) C NMR (126 MHz, D2O) : 20.48 (s, 11′),

23.17 (s, 10′), 30.00 (s, 9′′), 37.57 (s, 5′′, 6′′), 40.43 (s, 8′′), 40.66 (d, J = 7.67 Hz, 2′′), 59.17

(s, 3c), 67.30 (br s, 5′), 74.09 (s, 1′′), 76.32 (m, 2′, 3′′, 3′), 79.24 (s, 2c), 89.60 (br s, 1′, 4′),

120.78 (s, 5), 129.87-134.79 (C4 aromatic carbons), 144.64 (s, 8), 146.91 (s, 4), 150.74 (s,

2), 152.15 (s, 6), 176.33 (s, 7′′), 176.98 (s, 4′′), 205.43 (s, 1c). (See Figure A.3.12 for carbon

- numbering). LC-ESI-MS m/z 929.094 [M – H] ; calculated for C30H44N8O18P3S: 929.1713.

4.2.10 HPLC Analysis of acyl CoA thioesters

Acyl CoA synthesis and purity were evaluated by HPLC. Samples were dissolved or diluted in Milli-Q water, pH 4 and injected (10 µL) onto a C18 column using an Agilent

1100 Series HPLC system. HPLC was used to determine the purity of the biosynthetic acyl

CoA products. An autosampler connected to an HPLC system (Agilent 1100) injected a

10-µL sample onto a C18 HPLC column (3.5 µm, 4.6 mm × 100 mm, at 22 °C, Waters).

The column was eluted at 1 mL/min with 2.5% solvent B (100% acetonitrile) and 97.5% solvent A (0.1% trifluoroacetic acid in water) with a 1 min hold, a linear gradient to 30% solvent B over 10 min, then increased to 100% solvent B over 2 min, and finally lowered to 2.5% solvent B over 2 min. The effluent was monitored at A258. Reaction progress was

171 measured by the disappearance of the peak corresponding to CoASH and the growth of the peak corresponding to the acyl CoA of interest. Peak areas were integrated using Agilent

ChemStation (Agilent Technologies, Inc. 2003, Santa Clara, CA) software to determine the extent of the reaction or the purity of the acyl CoA.

4.2.11 Acyl CoA purity analysis (Ellman assay)

Ellman’s reagent (5,5'‐Dithio‐bis‐(2‐nitrobenzoic acid): DTNB) is applied in a well- established method to quantify free thiol concentration in protein solutions or purified acyl

CoA thioesters (61, 62). Here, Ellman’s reagent was used to determine the free thiol content in a sample of purified acyl CoA that contains salt. Free thiol assays with Ellman’s reagent were adapted (60) as follows. Briefly, DTNB was dissolved in 30 mM MOPS, pH 7 to make a 10 mM stock solution. An aliquot of the 10 mM DTNB (50 µL) stock was diluted

(2.5 mL) with buffer, then mixed with acyl CoA (<0.2 mM) (250 µL) dissolved in the same buffer. The reactants were incubated at room temperature for 10 min before measuring the absorbance at 412 nm on a UV/VIS spectrophotometer. Acyl CoA thioesters began to hydrolyze at longer incubation times, resulting in increasing A412. Beer’s law (εDTNB =

-1 14.15 mM cm-1 at 412 nm) was used to determine the amount of free thiol present in the sample. This number was compared to a control sample; mixed with NaOH (0.5 M final concentration), heated 20 min (55 °C), acidified to pH 7 (1 M HCl) and mixed with the diluted DTNB (as described), and incubated for 20 min at 22 °C. This sample represented the sum of acyl CoA and free CoA present in the mixture. Beer’s law was used to calculate the amount of free thiol and acyl CoA from the absorbances of both hydrolyzed and non- hydrolyzed samples. The amount of salt in the purified acyl CoA material was determined by calculating the total mass of CoA and acyl CoA present from moles as determined by

172 Ellman assays. This mass was substracted from the total mass of purified material to give the amount of salt (non-acyl CoA) material present.

4.2.12 BAPT, BadA, and NDTNBT purification

BAPT was purified as MBP-BAPT as described in Chapter 3 of this dissertation. BadA was purified as described in Chapter 2 of this dissertation. NDTNBT was purified as previously described (55).

4.2.13 BAPT kinetic assays

Solutions of each acyl CoA (10 mM) were dissolved in deionized water, pH 3 (8.8% formic acid). Baccatin III (10 mM) and 10-deacetylbaccatin III (10 mM) were dissolved in acetonitrile or methanol, respectively. To establish steady-state kinetic rates of MBP-

BAPT with respect to enzyme concentration and time, the acyl CoA and taxane were combined in 200 mM MOPS,(pH 8) containing 5% glycerol and pre-incubated at 31 °C for

10 min before the addition of ~0.2 µg (0.024 nmol) MBP-BAPT (200 µL total volume).

Assays were acid-quenched with 8.8% formic acid (aq.) to pH 5 at different timepoints to generate a time-course series. Docetaxel (1 µM) was added as an internal standard.

Concentrations of acetonitrile or methanol (depending upon the taxane substrate) were held constant at 10% (v/v). Assays were then extracted twice with 1 mL of EtOAc and dried under nitrogen and/or vacuum. Dried assays were resuspended in 100 µL of acetonitrile

(baccatin III-containing assays) or methanol (10-deacetylbaccatin III containing assays) and transferred to a GC vial (200 µL insert) for analysis by LC-ESI-MS/MS.

173 4.2.14 Kinetic analysis

To calculate kinetic constants, each substrate was varied (1 – 3000 μM) in separate assays under the predetermined steady-state conditions. The products made in the reaction assays were quantified as described above. Kinetic parameters (KM and kcat) were calculated by non-linear regression with Origin 9.0 using the following equation vo = kcat [Eo][S]/(KM +

[S]) (Northampton, MA) (Figures A.3.15 to A.3.17). Although KM is not a true dissociation constant, in this study it will serve as a means of comparing enzyme interactions with both substrates. Assays were prepared as follows. The taxane was added to a 5 mL glass test tube, 200 mM MOPS (pH 8) was added dropwise to facilitate mixing of the acetonitrile or methanol. The reaction tube was incubated at 31 °C for 10 min, before the acyl CoA and

MBP-BAPT (0.023 nmol in 200 µL) were added to start the assay.

4.2.15 Liquid chromatography mass spectrometry: BAPT assay analysis

LC-ESI/MS/MS in positive ion mode was used to quantify the biosynthetic acyl CoA products. An autosampler (at 10 °C) connected to an HPLC system (Waters Corp., Milford,

MA) injected 10-µL of each processed assay onto an Ascentis Express C18 HPLC column

(2.7 µm, 5 cm × 2.1 mm, at 30 °C, Sigma-Aldrich). Assay products dissolved in acetonitrile or methanol were each analyzed with an HPLC method utilizing acetonitrile or methanol as the mobile phase, respectively. The column was eluted at 0.4 mL/min with 2.5% solvent

B (100% methanol) and 97.5% solvent A (0.5% formic acid in water) with a 0.5 min hold, an immediate increase to 30% solvent B, followed by a linear gradient to 90% solvent B over 4 min, then increased to 100% solvent B over 0.5 min, and finally lowered to 2.5% solvent B over 0.5 min. The needle was washed with 0.4 mL each of 100% isopropanol and then with 10% acetonitrile in water prior to each injection. The HPLC effluent was

174 directed to an electrospray ionization mass spectrometer (Quattro Micro, Waters Corp,

Milford, MA) in positive ion mode, with a cone voltage of 20 V and collision energy of 16 eV. Each taxane was quantified by multiple reaction monitoring of the [M + H]+ → m/z

509 transition, common to both baccatin III, 10-DAB, and docetaxel. Peak areas

(calculated using the MassLynx data analysis software, Waters Corp., Milford, MA) were converted to product concentrations using the docetaxel internal standard of 1 µM).

4.2.16 PheAT and BAPT coupled reactions

Coupled assays for the production of N-debenzoyl-2′-deoxypaclitaxel were prepared on ice and incubated at 31 °C for 5 min before the enzyme was added. Assay volume was 200 µL with 10% (v/v) methanol. Baccatin III (0.75 mM to 3 mM dissolved in methanol), was added first to a 5-mL glass test tube, followed by 200 mM MOPS, (pH 8), 3.5 mM MgCl2,

1 mM ATP, 1 mM phenylisoserine, 0.05 to 2 mM CoA, and PheAT (0.75 mg/mL) to a final volume of 200 µL. After 1 h at 31 °C, MBP-BAPT (0.01 mg/mL) was added and allowed to react for an additional two h. Assays were processed for analysis in the same manner as that for the MBP-BAPT kinetics (see above). Coupled assay products were analyzed by LC-ESI-MS/MS with multiple reaction monitoring of the N- debenzoylpaclitaxel ion [M + H]+ m/z 750 → m/z 509 transition and of the internal standard docetaxel (1 µM) ion [M + H]+ m/z 808 → m/z 509 transition.

4.2.17 Production of paclitaxel and its analogs in a coupled enzyme assay

Four enzymes (PheAT, BAPT, BadA, and NDTNBT) were coupled together to produce paclitaxel (or an N-acyl analog) from the precursor baccatin III. The coupled reaction was tested in vitro where the ligase reactions, catalyzed by PheAT and BadA, were done first, followed by addition of the acyltransferases BAPT and NDTNBT. The ligase reactions

175 were prepared on ice and incubated, shaken (100 rpm) in a 31 °C waterbath for 5 min prior to the addition of PheAT and BadA. Baccatin III (0.5 mM dissolved in methanol) was added to the bottom of a glass test tube and mixed with 200 mM MOPS, pH 8. Substrates were then added to final concentrations of 3.5 mM MgCl2, 1 mM ATP, 50 µM to 1 mM

CoA, 0.5 to 2 mM (2R,3S)-phenylisoserine, and 1 mM benzoic acid (or 2-furoic acid).

BadA (1 µg/mL) and PheAT (~1 mg/mL) were then added to start the assay after temperature equilibration. After 1 h, MBP-BAPT (2.3 µg, 0.023 nmol) and NDTNBT (91

µg, 1.7 nmol) were added to a final volume of 250 µL and incubated for 3 h. Docetaxel (1

µM) was added as an internal standard and the solution was extracted with ethyl acetate (2

× 1 mL). The organic fraction was removed, dried under a stream of nitrogen, and resuspended in 100 µL of methanol. Transition for the production of paclitaxel ([M +

H]+ = m/z 856 → m/z 509), N-debenzoylpaclitaxel ([M + H]+ = m/z 750 → m/z 509), and

N-debenzoyl-N-2-furanylpaclitaxel ([M + H]+ = m/z 844 → m/z 509), were analyzed by

LC-ESI-MS/MS using the same methods for determining MBP-BAPT kinetics described before. Product concentrations were normalized relative to the internal standard of docetaxel (1 µM).

176 4.3 Results and discussion

4.3.1 Synthesis of acyl CoA substrates

Characterization of BAPT acyltransferase activity requires acyl CoA substrates that are not commercially available. Therefore, the substrates (3R)-β-phenylalanyl CoA ((3R)-Phe

CoA) and (2R,3S)-phenylisoserinyl CoA ((2R,3S)-PhIS CoA) were synthesized and biosynthesized, respectively (Figure 4.5). Several methods have been developed for acyl

CoA synthesis involving thioester activation to a mixed anhydride or other activated intermediate with ethyl chloroformate, N,N’-carbonyldiimidazole, N-hydroxysuccinimide or others (58, 60, 63-67). While synthesis of the (3R)-β-Phe CoA was relatively straightforward, synthesis of (2R,3S)-PhIS CoA proved more complicated, ultimately requiring a large-scale biosynthetic approach with a repurposed bacterial adenylation domain (PheAT) from tyrocidine synthetase A (TycA) of Bacillus brevis (42, 48, 68, 69).

Figure 4.5. Structures of (3R)-β-phenylalanyl coenzyme A (R = H) and (2R,3S)-phenylisoserinyl coenzyme A (R = OH).

177 4.3.1.1 Synthesis of (3R)-β-phenylalanyl CoA

A method based on the mixed anhydride intermediate was used to synthesize (3R)-β-Phe

CoA, (Figure 4.6) (43, 67). The amino group of (3R)-β-phenylalanine was N-Boc protected, followed by the addition of ethyl chloroformate to produce a mixed anhydride intermediate.

Incubation with CoA and subsequent deprotection of the 3-amino group yielded (3R)-β-

Phe CoA. Characterization of the purified product included analysis with Ellman’s reagent,

HPLC (Figure A.3.3), mass spectrometry (Figure A.3.4), and NMR (Figure A.3.2). Results from HPLC analysis and Ellman’s reagent assays showed the yield at 82% relative to CoA and ~95% purity (compared to free thiol content) with 50% salt by weight. Problems encountered during synthesis of (3R)-β-Phe CoA were degradation during purification, lyophilization, and high salt. These problems were minimized with desalting resin and by shortening lyophilization time with smaller volumes. The source of retained salt was likely the TEA used during column purification. The (3R)-β-Phe CoA is a zwitterion at pH 5, and likely purified as a salt even after desalting. NMR spectrum analysis agrees with prior syntheses of related compounds (Figure A.3.2 and Figure A.3.3) (70-72). The presence of smaller, shifted peaks in NMR spectra is also observed in the stock material of CoA from multiple sources and are possibly from an isomer of CoA (73). Mass spectrometry analysis of the major purified product was consistent with the expected mass of (3R)-β-Phe CoA, m/z 913 (Figure A.3.4). Tandem mass spectrometry analysis of a BAPT assay with purified

(3R)-β-Phe CoA with baccatin III and showed a mass consistent with N-debenzoyl-2′- deoxypaclitaxel (Figure A.3.5).

178

Figure 4.6. Organic synthesis of (3R)-β-Phenylalanyl CoA.

4.3.1.2 Synthesis of (2R,3S)-Phenylisoserinyl CoA

Synthesis of (2R,3S)-PhIS CoA was more complicated than (3R)-β-Phe CoA. Several different synthetic approaches were employed, but resulted in very little (<1-10%) or no yield. The 2-hydroxyl group on phenylisoserine changes the reactivity of the substrate with activating agents such as ethyl chloroformate. In the first synthetic attempt the 2-hydroxyl position of the phenylisoserine was not protected in order to reduce steric hindrance at the carboxylate reaction center (see Method I in Materials and Methods, Figure A.3.6). The carboxylate was proposed to react in sequence with ethyl chloroformate to form the mixed anhydride intermediate and finally with CoA to form the acyl CoA thioester. Using the same synthetic method to make (2R,3S)-PhIS CoA as used with (3R)-β-Phe CoA, resulted in the production of many side products as observed on TLC and by mass spectrometry.

The side products appeared comprised of the desired (2R,3S)-PhIS CoA in trace amounts and numerous derivatives of the N-Boc-(2R,3S)-phenylisoserine mixed anhydride.

Attempts to modulate cross-reactivity by cooling the reaction on ice, or slowly adding cold, dilute ethyl chloroformate were unsuccessful.

179 The persistent production of by-products in the mixed anhydride method described above, suggested the 2′-hydroxyl group needed to be protected. An ideal 2′-hydroxyl protecting group for this synthesis needed to be small and acid labile at room temperature, to prevent cleavage of the acyl CoA thioester product. The 2-hydroxyl group was protected as a t-butyldimethylsilyl ether (TBDMS ether) (74). This caused an increase from two to five h in incubation time to form the mixed anhydride intermediate compared with (3R)-β- phenylalanine (Figure A.3.7). After reacting the mixed anhydride intermediate with CoA, we observed analytes with mass fragments consistent with N-Boc-(2R,3S)-PhIS CoA.

However, a bulk of the product mixture contained unreacted CoA which likely formed dimers as evidenced by an ion at [CoAS-SCoA-2H]-2 = m/z 765.10 and isotopic peaks separated by 0.5 m/z units (not shown). The low conversion of the TBDMS-protected intermediate was likely caused by steric hindrance that prevented CoA from accessing the reactive carbonyl group. For this reason, an alternative method of carboxylate activation was sought.

After the initial ethylchloroformate approach resulted in extensive byproducts and protection of the 2-hydroxyl group resulted in trace product formation, an alternative to carboxylate activation that prevents the formation of byproducts was designed. The dicyclohexylcarbodiimide (DCC) reagent is well-established in as a means of activating carboxylic acids in peptide coupling chemistry (75, 76). The large size of DCC was predicted to reduce the occurrence of side products through steric hindrance with the 2-hydroxyl protecting group. However, the bulkiness of DCC would also likely be a steric hindrance for the final nucleophilic reaction with CoA. Therefore, DCC was coupled to hydroxybenzotriazole (HOBt), a small, planar leaving group (Figure A.3.8)

180 (75). The activation by HOBt was followed by TLC and by observing the precipitation of the insoluble DCC byproduct dicyclohexylurea (DCU). There was no N-Boc-(2R,3S)- phenylisoserinyl CoA detectable above background levels by mass spectrometry ([M-H]-1

= m/z 1029.22), substantial CoA oxidation was observed, and the activated intermediate appeared to be degraded into phenylisoserine and HOBt.

The DCC/HOBT coupling method was modified to include N-hydroxysuccinimide

(NHS) instead of HOBt (Figure A.3.9) (75). The NHS reagent is more polar than HOBt and the resulting mixed anhydride with (2R,3S)-PhIS was hypothesized to be more water soluble and possibly more reactive with CoA. The non- reducing agent tris-(2- carboxyethyl)-phosphine (TCEP) was also added to prevent CoA dimerization. Upon CoA addition, results were similar with no significant N-Boc-(2R,3S)-phenylisoserinyl CoA observed. However the CoA was not oxidized due to inclusion of the reducing agent,

TCEP.

Figure 4.7. Conformation of (2R,3S)-phenylisoserine. A. Linear representation of (2R,3S)- phenylisoserine. The bond highlighted in blue is selected for the Newman projection. B. Newman projection of (2R,3S)-phenylisoserine. Structure was energy minimized in ChemBio 3D Ultra (Perkin Elmer, Waltham, MA).

In general, the failure to obtain appreciable quantities of N-Boc-(2R,3S)- phenylisoserinyl CoA by all synthetic means tested is indicative of a sterically challenged reaction center (Figure 4.7). Substantial CoA dimerization was observed in these reactions,

181 but not in the (3R)-β-Phe CoA synthetic reaction, suggesting CoA reacted with itself because the reactive intermediate was not accessible. These difficulties made it apparent that a different method for synthesizing (2R,3S)-PhIS CoA in large enough quantities (~50-

100 mg) was needed for the kinetic characterization of BAPT.

Figure 4.8. Biosynthesis of (2R,3S)-phenylisoserinyl CoA with PheAT, a truncated form of tyrocidine synthetase A (TycA). The 2′-hydroxyl group is shown in red.

4.3.2 Biosynthesis of (2R,3S)-PhIS CoA.

Considering the numerous challenges encountered with the organic synthesis of (2R,3S)-

PhIS CoA, this approach was abandoned, and a biosynthetic method was considered.

Recently in the Walker group, the TycA (PheATE) monomer from the non-ribosomal peptide synthase (NRPS) tyrocidine synthetase A biosynthetic pathway was tested as a

CoA ligase and showed broader substrate activity with (3R)-β-phenylalanine and (2R,3S)- phenylisoserine beyond its natural substrate, α-phenylalanine (42, 68, 69). Modified TycA activity produces (3R)-β-Phe CoA and (2R,3S)-PhIS CoA in an ATP dependent manner.

To our knowledge it is the only enzyme currently known to produce (2R,3S)-PhIS CoA.

More recently, a truncation of TycA (PheATE), removed the epimerization (“E”) domain and modified the “T” domain (Figure 4.8). The truncation was active and is hereafter referred to as PheAT (48).

-1 The Michaelis-Menten kinetic parameters of PheAT (kcat = 0.015 ± 0.0028 s and

KM = 440 ± 62 µM for (2R,3S)-PhIS, and KM = 208 ± 57 µM for CoA) were used to

182

Figure 4.9. Biosynthetic reaction progress curve for (2R,3S)-PhIS CoA. A large-scale (20 mL) reaction with PheAT (36 mg/mL), CoA (5 mg/mL), ATP (5 mg/mL), (2R,3S)-PhIS (~1.6 mg/mL) and MgCl2 (5 mg/mL) was monitored by UV-HPLC at 258 nm. The peak corresponding to (2R,3S)-PhIS CoA was normalized to % conversion relative to CoA. Black circles: One PheAT large-scale reaction which was supplemented with additional (2R,3S)-PhIS (0.5 mM) at 7 h (red squares). Black triangles: One PheAT large-scale reaction, which was supplemented with additional ATP (5 mM) at 6 h (blue diamond).

estimate ~700 mg of PheAT is needed to produce ~100 mg of (2R,3S)-PhIS CoA in 1 h

(Table 4.2). PheAT was heterologously expressed as a 6X-His tagged fusion protein from

E. coli and purified by nickel-affinity chromatography at 99% purity. PheAT was highly soluble at 70 mg/mL. The reaction progress of a large-scale (20 mL) enzyme reaction was monitored by HPLC over the course of 6 to 19 h (see Methods) (Figure 4.9). Accumulation of (2R,3S)-PhIS CoA never exceeded 60% (Figure 4.9) perhaps because of a gradual loss of PheAT activity, or because the reaction reached an equilibrium between (2R,3S)-PhIS

CoA production and degradation in the buffer at pH 8.0. Upon quenching the reaction by acidification (pH < 5.0 with 8.8% formic acid) and stabilizing (2R,3S)-PhIS CoA, PheAT also precipitated, facilitating subsequent product purification.

183 Separation of (2R,3S)-PhIS CoA from CoA was a concern during biosynthetic production. The two compounds had similar retention factors (Rf values) with all TLC conditions tested including variations on 5:3:2 butanol:water:acetic acid (v/v/v) which was successful at separating (3R)-Phe CoA from CoA. Therefore, preparative, reverse-phase

HPLC was used for purification to obtain 45.5 mg (50 µmol) in 30% yield (relative to

(2R,3S)-phenylisoserine) at 97% purity by UV-HPLC analysis (Figure A.3.13 and Figure

A.3.14). Previous (2R,3S)-PhIS CoA synthesis isolated 2.9 µmol in 10% yield (80% purity)

(43). This preparative purification of (2R,3S)-PhIS CoA is an improvement over synthetic methods makes it possible to characterize the paclitaxel biosynthetic BAPT enzyme for the first time (Figure 4.10).

Figure 4.10. BAPT-catalyzed enzyme reaction. The proposed native BAPT reaction catalyzes the transfer of (3R)-β-phenylalanyl CoA (R2 = H) to the 13-O of baccatin III (R1 = Acetyl group). BAPT is less active with (2R,3S)-phenylisoserinyl CoA (R2 = OH) when incubated with either baccatin III or 10-DAB (R1 = H).

4.3.3 Michaelis-Menten kinetics of BAPT

4.3.3.1 (3R)-β-phenylalanyl CoA and baccatin III substrates

The importance of the phenylisoserinyl side chain for the efficacy of paclitaxel has been well-established in structure and function studies (57, 77). Current literature supports a biosynthetic pathway in which BAPT catalyzes the C13 acylation of baccatin III with (3R)-

184 β-phenylalanyl CoA to form N-debenzoyl-2′-deoxypaclitaxel (ND2DP), which then hydroxylated at the 2′-position on the sidechain, and N-benzoylated to produce paclitaxel

(43, 56, 78). Recently, the gene for a (β)-phenylalanyl CoA ligase (β-PL) was identified in

Taxus baccata cell cultures elicited with jasmonic acid (20). The β-PhL was active with β- phenylalanine and 4-coumarate. Unfortunately, the activity of β-PhL with phenylisoserines was not characterized in the former study. While the discovery of a Taxus β-PhL is long- awaited, the hydroxylase responsible for adding the 2′-hydroxyl group to ND2DP remains elusive. In our efforts to characterize the activity of BAPT and build the foundation for engineering efforts of paclitaxel biosynthesis from baccatin III, both the synthetic (3R)-β-

Phe CoA and the biosynthetic (2R,3S)-PhIS CoA were utilized as substrates.

Michaelis-Menten constants for BAPT were determined after (bio)synthesizing the needed acyl CoA substrates (Table 4.2). More importantly, this is the first report of the kinetic parameters for BAPT with (2R,3S)-PhIS CoA. The kcat and KM of BAPT for (3R)-

β-Phe CoA with baccatin III at apparent saturation were 0.0583 ± 0.0010 s-1 and 5.6 ± 1.0

µM, respectively (Figure A.3.16). Reciprocally, the kcat and KM of BAPT for baccatin III

-1 with (3R)-β-Phe CoA at apparent saturation were kcat = 0.0575 ± 0.0018 s and a 68 ± 9.6

µM, respectively (Figure A.3.17). These values are compared with those of BAPT with

-1 (2R,3S)-PhIS CoA where kcat = 0.0022 ± 0.00013 s (Figure A.3.15) and KM = 22.4 ± 6.9

µM. Previous estimates of the KM of BAPT using crude lysate from a recombinant BAPT expressed in E. coli, were 2.4 ± 0.5 µM and 4.9 ± 0.3 µM for baccatin III and (3R)-β-Phe

CoA, respectively (43). These estimates different than those calculated with purified

BAPT. In this study, the KM of purified BAPT with baccatin III is 12-times higher than that for (3R)-β-Phe CoA, whereas the earlier report determined a KM that is half that for (3R)-

185 β-Phe CoA. These disparities likely result from the use of crude enzyme preparations of recombinant enzyme expressed without an affinity epitope for purification and subsequent quantification in the earlier study (43). In this study, we expressed BAPT with a polyhistidine affinity epitope tag, purified, and quantified the expressed enzyme with greater accuracy compared with the crude estimates used previously (43).

4.3.3.2 (2R,3S)-Phenylisoserinyl CoA and baccatin III substrates

The presence of the 2′-hydroxyl group on paclitaxel is required for its biological activity.

Thus characterization of BAPT enzymatic activity for (2R,3S)-PhIS CoA will help advance bioengineering efforts toward making paclitaxel and likely its analogs. BAPT is 27-times more active with (3R)-β-Phe CoA than with (2R,3S)-PhIS CoA, (with baccatin III at apparent saturation) which suggests that the proposed paclitaxel biosynthetic pathway may indeed proceed through a β-phenylalanyl intermediate. In addition, the presence of the 2′- hydroxyl of (2R,3S)-PhIS CoA increases the KM 4-fold over that of (3R)-β-Phe CoA. This may be due to unfavorable steric or electronic effects of the added hydroxyl group within the active site.

4.3.3.3 (2R,3S)-Phenylisoserinyl CoA and 10-deacetylbaccatin III as substrates

An effort was made to determine Michaelis Menten kinetic parameters for BAPT with 10- deacetylbaccatin III (10-DAB) and (2R,3S)-PhIS CoA as substrates to product N- debenzoyl-10-deacetylpaclitaxel (or N-debenzoyldocetaxel). Under 1 mM substrate

-4 concentrations, a Vrel of 1.92 × 10 nmol/min was determined compared with the Vrel of

0.018 nmol/min for BAPT with baccatin III using the same preparation of enzyme (Figure

A.3.18). Michaelis Menten kinetic constants of BAPT for (2R,3S)-PhIS CoA and 10-DAB

(at saturation) could not be calculated because of a 91.2-fold reduction in activity with 10-

186 DAB compared to that for baccatin III. Although BAPT is active with both baccatin III and

10-DAB, it is possible that active site mutagenesis of BAPT could facilitate more efficient production of the 10-deacetylated docetaxel precursor and its analogs.

Table 4.2. Kinetic constants of enzymes required for proposed paclitaxel biosynthesis from baccatin III.

Substrate 1 Substrate 2 Substrate 3 -1 1 kcat (s ) Ref. KM (µM) KM (µM) KM (µM) Acyl CoA ligases (2R,3S)-PhIS CoA ATP PheAT 0.015 ± 0.0028 (48) (440 ± 62) (208 ± 57) ND benzoate CoA ATP (Ch. 2) BadA 28 ± 0.9 (4.4 ± 0.65) (90-120) (2-3) (79) Acyltransferases (2R,3S)-PhIS CoA 0.0022 ± This - (22.4 ± 6.9) 0.00013 work BAPT2* (3R)-β-Phe CoA Baccatin III 0.0583 ± This (5.6 ± 1.0) (68 ± 9.6) 0.0011 work Benzoyl CoA N-debenzoylpaclitaxel ND (56) (375 ± 67) (78 ± 11) NDTNBT N-debenzoyl-2′- Benzoyl CoA deoxypaclitaxel 1.5 ± 0.3 (80) (410) (450) *BAPT expressed as MBP-BAPT fusion protein. 1 KM was determined for an enzyme for a given substrate with the additional substrates listed at apparent saturation. 2Inactive substrates include α-phenylalanine, 3-hydroxy-phenylpropanoic acid, and 3-phenylpropanoic acid. Abbreviations: PheAT: phenylalanine CoA ligase from Bacillus subtilis, BadA: benzoate CoA ligase from Rhodopseudomonas palustris, BAPT: baccatin III: 3-amino-13-O-phenylpropanoyl CoA transferase from Taxus cuspidata, NDTNBT: N-debenzoyltaxol-N-benzoyltransferase from Taxus cuspidata.

4.3.4 BAPT activity and substrate assisted catalysis

A conservation and homology (ConSurf: consurf.tau.ac.il/) analyses of 150 homologous

BAHD family members show complete conservation of the motif, HXXXD, containing the catalytic histidine residue (81-83). BAPT was unique among the BAHD family members with a GXXXD instead of the conserved HXXXD motif (Figure 4.11). This difference is interesting considering the proposed catalytic mechanism for this family of enzymes uses the conserved histidine of the HXXXD motif as a general base to deprotonate the OH or

+ NH3 functional group of the acceptor molecule (Figure 4.12). The deprotonated acceptor

187 molecule then attacks the electrophilic center of acyl CoA forming a tetrahedral intermediate, which subsequently collapses to form the acylated acceptor molecule and

CoA as a by-product.

T. cuspidata 1 121…NPSFQQLIFSLPQDTDIEDLHLLIVQVTRFTCGGFVVGANVYGSACDAKGFGQFLQSMAE…180 T. cuspidata 2 NPSFQQLIFSLPQDTDIEDLHLLIVQVTRFTCGGFVVGTNVYGSVCDAKGFGQFLQGMAE T. sumatrana KPSFQQLIFSLPQDTDIEDLHLLIVQVTRFTCGGFVVGANVYGSVCDAKGFGQFLQGMAE T. wallichiana* NPSFQQLIFSLPQDTDIEDLHLLIVQVTRFTCGGFVVGTNVYGSVCDAKGFGQFLQGMAE T. x media 1 NPSFQQLIFSLPQDTDIEDLHLLIVQVTRFTCGGFVVGANVYGSTCDAKGFGQFLQGMAE T. x media 2 NPSFQQLIFSLPQDTDIEDLHLLIVQVTRFTCGGFVVGANVYGSTCDAKGFGQFLQGMAE :*************************************:*****.***********.*** *(var. chinensis)

Figure 4.11. Multiple sequence alignment of bapt genes from Taxus sp. The GXXXD motif (gray) is shown. Sequences used in the alignment are as follows: T. cuspidata 1 (AAL92459.1), T. cuspidata 2 (AGT51232.1), T. sumatrana (ACN62085.1), T. wallichiana (var. chinensis) (AGC11862.1), T. x media 1 (AAT73200.1), and T. x media 2 (AFD32416.1). Multiple sequence alignment made with Clustal Omega (EMBL EBI, Cambridge, UK).

Clearly, BAPT retains activity without the conserved histidine residue despite reports demonstrating that histidine in the HXXXD motif is catalytically important (84,

85). For example, activity was decreased significantly in the BAHD enzymes vinorine synthase and malonyl CoA: anthocyanin 5-O-glucoside-6’’-O-malonyltransferase

(Ss5MaT1) upon histidine mutation to an inert non-catalytic residue such as alanine (84,

85) (Table 4.3). Loss of the conserved histidine in Ss5MaT1 resulted in a large reduction

-1 of the kcat from 8.5 to 0.002 s , on the same order of magnitude as wild-type BAPT activity with (2R,3S)-PhIS CoA (Table 4.3). Without a BAPT crystal structure to provide context for the non-conserved glycine residue, a homology model was created with RaptorX, a structure-based homology modeling tool (86-88) (Figure 4.13). Aligning the BAPT homology model to the hydroxy-cinnamoyl CoA transferase (HCT) acyltransferase from

Coffea canephora helps infer a relative position of (3R)-β-Phe CoA or

188

A.

B.

Figure 4.12. Proposed mechanisms for BAHD acyltransferases and BAPT. A. BAHD acyltransferases with the HXXXD motif utilize the histidine as a base to abstract a proton from the acyl acceptor OH or NH2 moiety. The acyl acceptor then nucleophilically attacks the acyl donor to form a tetrahedral intermediate (Step I). Subsequent collapse of the intermediate displaces CoASH and forms the acylated acceptor (Step II.). B. The proposed substrate-assisted catalytic mechanism for BAPT, which has a non-conserved GXXXD motif. The amino group of (3R)-β-Phe CoA (R = H) deprotonates the C-13 hydroxyl group on the acyl acceptor, baccatin III (R1 = acetyl). The C-13 oxygen nucleophilically attacks the acyl donor forming a tetrahedral intermediate (Step III) which subsequently collapses, forming N-debenzoylpaclitaxel and releasing CoA (Step IV).

189 (2R,3S)-PhIS CoA (89). The residues proposed to reside in the active site within 4Å of the glycine residue are Ala40, Ala41, Ala42, Ser44, Val161, Tyr162, Ala165, Cys166,

Asn300, and Phe302. There are no obvious residues that could act as a general base. Polar residues are positioned such that the peptide backbone orients the residue away from the active site, or they are solvent exposed. Although Ser44 is oriented away from the catalytic center of the enzyme, rotation of the sidechain could bring the functional group into the active site (Figure 4.13). Tyr162 is modeled in a solvent exposed orientation and Cys166 neighbors the conserved Asp167, pointing away from the active site. One limitation of homology modeling of BAHD acyltransferases is the high variability in active site architecture between enzymes, even in structures that bind the same substrate. For instance, malonyl CoA transferases from Nicotiana tabacum (tobacco) (NtMaT1) and red chrysanthemum (Dm3MaT3) both bind malonyl CoA and transfer the malonyl group to an anthocyanin acceptor molecule. However, crystal structures of both enzymes show distinctive active site architectures even though global protein folds are conserved (90, 91).

An alternative explanation for BAPT activity in the absence of the HXXXD motif is that the free amine of the (3R)-β-Phe CoA may act as a base in an example of substrate- assisted catalysis (SAC) (Figure 4.12) (97, 98). In this hypothesis, the (3R)-β-Phe CoA binds the active site in close proximity to the 3-amino group. Examples of SAC include subtilisin BPN’ (99, 100), lysozyme (101), type II restriction endonucleases (102), and poly-ADP-ribose polymerases (PARPs) (103). If the amino group of the (3R)-β-Phe CoA is catalytically important, substrates lacking an amino group should be inactive. Previous work showed BAPT to be inactive with 3-hydroxy-phenylpropanoyl CoA, phenylpropanoyl CoA and α-phenylalanyl CoA (43, 70). Inactivity with these substrates

190 Table 4.3. Michaelis-Menten kinetic constants from BAHD acyltransferase family members.

BAHD 1 1 -1 Organism Acyl donor Acyl acceptor kcat (s ) Reference acyltransferase (2R,3S)-PhIS CoA 0.0022 ± Baccatin III This work Taxus (22.4 ± 6.9) 0.00013 BAPT cuspidata (3R)-β-Phe CoA Baccatin III 0.0583 ± This work (5.6 ± 1.0) (68 ± 9.6) 0.0011 p-coumaroyl CoA p-coumaryl alcohol OsPMT Oryza sativa 2.6 (92) (281) (141) acetyl CoA Coniferyl alcohol PhCFAT Petunia2 2.05 (93) (30.6) (56.5) Arabidopsis malonyl CoA Cyanin At5MAT 0.02 (94) thaliana (2-15) (167)3 Arabidopsis feruloyl CoA 16-OH-palmitic acid At5g41040 0.0013 (95) thaliana (9.7) (5.1) Erythoxylum benzoyl CoA Methylecgonine EcBAHD7 9.7 (96) coca (93) (369) malonyl CoA Shisonin Ss5MaT1 8.7 Salvia (22) (35) (84) Ss5MaT1 splendens malonyl CoA Shisonin 0.002 H167A (6.4) (5.4) 1 If available, KM is listed after each substrate (µM). 2Petunia x hybrid ‘Mitchell Diploid’ was source of transgenic petunias. 3Cyanin is a non-native substrate.

suggests the presence and position of the 3-amino group is important for activity. Reduced activity of (2R,3S)-PhIS CoA suggests the 2-hydroxyl position may hinder reactivity of the neighboring amine group in the case of SAC, through steric and/or electronic interactions or places the substrate in a catalytically nonproductive conformation. Previous work in the

Walker group (unpublished) showed that a BAPT-Gly163His mutant resulted in no detectable acyl transfer with (3R)-Phe CoA, 3-hydroxyphenylpropanoyl CoA, or phenylpropanoyl CoA to the baccatin III acyl acceptor. These data further support that

BAPT evolved to lack the conserved histidine in the HXXXD motif. Without structural support, this hypothesis is difficult to prove, but clearly an alternative base is required in the BAPT active site to facilitate the base-catalyzed mechanism used by this family of enzymes.

191

A.

B.

Figure 4.13. Homology model of BAPT based on the hydroxycinnamoyl transferase (PDB: 4G0B) from Coffea canephora. α-helices (cyan), β-sheets (red) and loops (purple) are shown. A. An arrow points to the GXXXD (yellow) motif. The active site channel is outlined (circle). Residues proposed to reside within 4 Å of the active site glycine residue include Ala40, Ala41, Ala42, Ser44, Val161, Tyr162, Ala165, Cys166, Asn300, and Phe302 (not shown). B. Homology model of the BAPT active site aligned with the ligand bound Sorghum hydroxycinnamoyltransferase (HCT) (PDB: 4KEC). HCT products, p- coumarylshikimate and CoA are shown as yellow sticks and the conserved His162 is shown in magenta. The distance between His162 and the acceptor oxygen of shikimate is 3.3 Å. BAPT residues within 4 Å of the shikimate oxygen are shown. The deviant Gly163 residue and the nearby Ser44 are labeled as well as Arg41, Cys166, Asn300, and Phe302.

192 9

8

7

6

5

4

3

2

debenzoylpaclitaxel debenzoylpaclitaxel (nmol) -

N 1

0 0.05 0.1 0.25 0.5 1 2 CoA (mM)

Figure 4.14. N-debenzoylpaclitaxel biosynthesized from BAPT and PheAT coupled reactions. Production of N-debenzoylpaclitaxel at two different concentrations of baccatin III at 0.75 mM (gray), 3 mM (dark gray). (n = 2).

4.3.5 Coupled biosynthesis of N-debenzoylpaclitaxel: PheAT and BAPT.

As part of developing proof-of-principle for paclitaxel biosynthesis, coupling assays were performed with PheAT, the enzyme capable of (2R,3S)-phenylisoserine CoA ligase activity. Baccatin III and (2R,3S)-phenylisoserine were incubated with varied amounts of

CoA to determine conditions for maximal production of N-debenzoylpaclitaxel. This was especially important considering the low turnover of PheAT for (2R,3S)-PhIS CoA and its

KM of 208 ± 57 µM for CoA (Table 4.2). CoA concentrations were varied from 0.05 to 2 mM and biosynthetic N-debenzoylpaclitaxel was quantified by LC-ESI-MS/MS methods.

There appears to be no difference in N-debenzoylpaclitaxel production with low (0.1 mM) or high (2 mM) CoA during PheAT/BAPT coupling with baccatin III (at 0.75 mM or 3 mM) (Figure 4.14). However, below 0.1 mM CoA, the production of N- debenzoylpaclitaxel drops significantly, likely because the KM of PheAT for CoA is 208

µM and at concentrations lower than KM, PheAT is not operating at a maximal rate (Vmax).

193 4.3.6 Biosynthesis of N-2-furanoyl-N-debenzoylpaclitaxel: a paclitaxel analog

The four-enzyme (PheAT, BAPT, BadA, and NDTNBT) biosynthesis of paclitaxel was modified due to the presence of 0.1% paclitaxel in our commercial supply of baccatin III.

The low 0.1% paclitaxel contamination is easily observed by mass spectrometric analysis, making quantification of biosynthetic paclitaxel difficult. Therefore, to conclusively

100

100 80

80 60

60 40

Ion Count (%) Ion Count

40 20

20 -NDTBT +NDTBT 0 1 2 3 4 5 Time (min)

Figure 4.15. Biosynthesis of N-2-furanoyl-N-debenzoylpaclitaxel with PheAT, BAPT, BadA, and NDTNBT. Normalized LC-ESI-MS/MS trace in positive mode for the m/z 844 → m/z 509 ion fragmentation. Product peak is at 4.2 min. Background ions are shown for the coupled assay in the absence of NDTNBT. Mass spectrometry performed on a Quattro Micro API LC/MS/MS (Waters, Milford, MA).

194 establish proof of principle for paclitaxel biosynthesis, a biologically active N-2-furoyl-N- debenzoylpaclitaxel analog was biosynthesized (Figure 4.15) (45). The mass of the analog distinguishes it from the paclitaxel background during LC-ESI/MSMS and 2-furoyl CoA is 200% as active as benzoyl CoA with NDTNBT and baccatin III (55). A clear peak corresponding to m/z 844 → m/z 509 appeared at 4.2 min compared to a background control lacking NDTNBT. N-2-furoyl-N-debenzoylpaclitaxel was quantified at 1.06 ± 0.08 µM.

Levels of N-debenzoylpaclitaxel were 25 ± 9 µm (+NDTNBT) and 39 ± 11 µM (-

NDTNBT). Fragmentation of the [M + H]+ = m/z 844 ion was also consistent with taxanes

(Figure A.3.23) (104, 105). N-2-furoyl-N-debenzoylpaclitaxel is one example of an analog that can be made with this system. The combination of PheAT and BAPT yields N- debenzoylpaclitaxel and N-debenzoyldocetaxel precursors (Figure 4.16). Substrate specificities of each enzyme define the range of analogs that can be biosynthesized without enzyme engineering for enhanced efficiency or broader substrate specificity.

195

Figure 4.16. Overlapping substrate specificities of PheAT, BAPT, BadA, and NDTNBT. Substrate specificities of each enzyme are represented as circles. Overlapping circles indicate shared substrate specificity and the range of compounds that can be biosynthesized by coupling those enzymes together.

196 4.4 Conclusions

(2R,3S)-PhIS CoA was biosynthesized and isolated on a preparative scale (45.5 mg, 50

µmol) for the first time from a large-scale PheAT CoA ligase-catalyzed reaction.

Michaelis-Menten constants of BAPT were calculated for (3R)-β-Phe CoA, (2R,3S)-PhIS

CoA, and baccatin III. The catalytic efficiency (kcat/KM) of BAPT for (2R,3S)-PhIS CoA

-5 -1 -1 (kcat/KM = 9.82 × 10 s µM ) is 106-fold lower than for (3R)-β-Phe CoA (kcat/KM = 0.0104

-1 -1 s µM ) as a result of a decrease in kcat and increases in KM for the former. These results agree with previous work with β-PhL and NDTNBT, suggesting that enzymes on the natural biosynthetic pathway may prefer to acylate baccatin III with (3R)-β-Phe CoA over

(2R,3S)-PhIS CoA (56, 78).

The biologically active, anticancer analog N-2-furoyl-N-debenzoylpaclitaxel and the penultimate precursor, N-debenzoylpaclitaxel were produced in coupled enzyme- catalyzed reactions containing bacterial ATP-dependent CoA ligases PheAT and BadA, and Taxus acyltransferases BAPT and NDTNBT. N-Debenzoylpaclitaxel biosynthesized by PheAT and BAPT catalysis is a branchpoint precursor that can be N-acylated, either synthetically or biochemically (44-47, 50, 52, 106). PheAT and BAPT coupling also produced the docetaxel precursor N-debenzoyldocetaxel. Development of this system will bypass hurdles met in current efforts to engineer the paclitaxel biosynthetic pathway in heterologous organisms. Baccatin III and 10-DAB precursors are readily available at 4- fold the amount of paclitaxel from yew needles, or Taxus plant cell fermentation used to make paclitaxel (107, 108). Employing an adapted “downstream” paclitaxel biosynthesis system reduces the need to reverse engineer eukaryotic membrane bound P450 hydroxylases to function in bacterial microorganisms. There is minimal flux

197 away from the production of the desired product and no side reactions of the acyltransferases with other taxanes. No more than four enzymes would need to be optimized for in vivo expression and activity compared with attempting to engineer the entire paclitaxel pathway which requires minimally 19 enzymatic steps.

198 4.5 Future Research

Future research with BAPT will include the development of in vivo assays for the production of paclitaxel. Important factors will include modulation of protein expression, substrate concentrations, and growth effects. Feeding studies in E. coli showed that 10-

DAB can be successfully taken up and acylated by 10-deacetylbaccatin III 10-O- acetyltransferase (DBAT) in vivo (109). Development of a bioengineered in vivo system to produce paclitaxel, analogs, or precursors is likely feasible with feeding of precursors such as baccatin III, 10-DAB, phenylisoserine, and acyl acids.

Expansion of substrate specificity within the four-enzyme coupled system can be facilitated by rational mutagenesis of enzyme active sites. This has already been demonstrated for BadA with the development of expanded activities and novel activities from single point mutants. Construction of a PheAT homology model will inform active site mutagenesis to expand the substrate specificity to substituted phenylisoserines in order to make analogs in the C13-sidechain. Characterization of BAPT substrate specificity with different phenylisoserines will also aid in potential analog production. The synthesis of a series of phenylisoserine derivatives active with PheAT may be coupled to BAPT to test substrate specificities (48).

Methods to alter the substrate specificity of BAPT may prove more challenging.

The natural mutation of the HXXXD motif to GXXXD in BAPT may affect the orientation of the substrate in the active site, limiting the utility of homology modeling. Another residue may have a catalytic role, or SAC may occur. Based upon homology modeling, a number of residues are predicted to be in the active site. Mutagenesis of these residues and

199 activity assays will help determine what limits substrate activity. Although BAPT is difficult to purify in appreciable yields, crystal screens for the BAPT acyltransferase will be an important future endeavor.

Beyond paclitaxel analog biosynthesis, the combination of plant acyltransferases and microbial acyl CoA ligases may be useful in other bioengineering applications. Acyl

CoA ligases are widespread with different substrate specificities. Combinations of acyl

CoA ligases and acyltransferases hold potential for generating specific, structurally complex molecular products, as opposed to organic acylation, which may require redundant protection reactions, lack regiospecificity, and use toxic organic solvents.

200

APPENDIX

201

APPENDIX

Figure A.3.1. 1H-NMR of R-N-Boc-3-amino-3-phenylpropanoic acid.

202

Figure A.3.2. 1H-NMR of chemically synthesized (3R)-β-phenylalanyl CoA.

203

300

(3R)--Phe CoA

200

258

A

100

0 0 5 10 Time (min)

Figure A.3.3. HPLC chromatogram of purified (3R)-β-phenylalanyl CoA.

204 A.

B.

Figure A.3.4. Mass spectra of purified (3R)-β-phenylalanyl CoA. A. LC-ESI-MS of purified (3R)-β- phenylalanyl CoA (Exact mass, M = 914.1836). Main peaks correspond to [M-H]-1 = m/z 913.098, [M- 2H]-2 = m/z 456.047. B. LC-ESI/MSMS of m/z 913. Fragments are labeled. Mass spectrometry was performed on a Xevo G2-S QTof UPLC/MS/MS (Waters, Milford, MA).

205

100 80 60 40

Ion Count (%) Count Ion 20 0 0 1 2 3 4 5 Time (min)

Figure A.3.5. LC-ESI/MSMS chromatogram showing the production of N-debenzoyl-2′- deoxypaclitaxel. Product was measured by multiple reaction monitoring: m/z 734 → m/z 509 by BAPT. Mass spectrometry performed on a Waters Quattro micro API LC/MS/MS (Waters, Milford, MA).

Figure A.3.6. (Scheme I) Synthesis of (2R,3S)-phenylisoserinyl CoA.

206

Figure A.3.7. (Scheme II) (2R,3S)-PhIS CoA synthesis with 2-OH protection with TBDMS Cl.

Figure A.3.8 (Scheme III) (2R,3S)-PhIS CoA synthesis by DCC/ HOBt coupling.

207

Figure A.3.9 (Scheme IV) (2R,3S)-PhIS CoA synthesis by DCC/NHS coupling.

208

Figure A.3.10. 1H-NMR of (2R,3S)-N-Boc-3-amino-3-phenylisoserine. This compound is an intermediate in the attempted synthesis of (2R,3S)- phenylisoserinyl CoA.

209

Figure A.3.11. 1H-NMR of biosynthetic (2R,3S)-phenylisoserinyl CoA.

210

Figure A.3.12. 13C-NMR of biosynthetic (2R,3S)-phenylisoserinyl CoA.

211

0.6

0.4

258

A

0.2

0.0 0 2 4 6 8 10 12 14 Time (min)

Figure A.3.13. HPLC analysis of purified (2R,3S)-phenylisoserinyl CoA.

212 A.

B.

Figure A.3.14. Mass spectra of purified (2R,3S)-phenylalanyl CoA. A. LC-ESI-MS of purified (2R,3S)-Phenylalanyl CoA (Exact mass, M = 930.1785). Main peaks correspond to [M-H]-1 = m/z 929.094, [M-2H]-2 = m/z 464.044. B. LC-ESI/MSMS of m/z 929. Fragments are labeled. Mass spectrometry was performed on a Xevo G2-S QTof UPLC/MS/MS (Waters, Milford, MA).

213

0.004

0.003

0.002

v (nmol/min) 0.001

0.000

0 1000 2000 3000 (2R,3S)-Phenylisoserinyl CoA (M)

Figure A.3.15. Michaelis-Menten kinetic plot of biosynthetic N-debenzoylpaclitaxel produced by MBP-BAPT. Substrates were (2R,3S)-phenylisoserinyl CoA (15 µM to 3 mM) and baccatin III (1 mM). -4 -1 Vmax = 0.003 ± 1.7 x 10 nmol/min, KM = 22.35 ± 6.97 µM, kcat = 0.13 ± 0.0076 min . (n = 3, n = 2 for 2000 and 3000 µM (2R,3S)-phenylisoserinyl CoA).

0.10

0.08

0.06

0.04

v (nmol/min)

0.02

0.00

-250 0 250 500 750 1000 1250 1500 1750 2000 2250 (3R)-Phenylalanyl CoA (mM)

Figure A.3.16. Michaelis-Menten kinetic plot of biosynthetic N-debenzoyl-2′-deoxypaclitaxel produced by MBP-BAPT. Substrates were (3R)-β-phenylalanyl CoA (12.5 µM to 1 mM) and baccatin -1 III (1 mM). Vmax = 0.086 ± 0.0015 nmol/min, KM = 5.6 ± 1.02 µM, kcat = 3.59 ± 0.063 min . (n = 3).

214

0.10

0.08

0.06

0.04

v (nmol/min)

0.02

0.00

0 1000 2000 3000 Baccatin III (M)

Figure A.3.17. Michaelis-Menten kinetic plot of biosynthetic N-debenzoyl-2′-deoxypaclitaxel produced by MBP-BAPT. Substrates were baccatin III (12.5 µM to 1 mM) and (3R)-β-phenylalanyl CoA -1 (1 mM). Vmax = 0.082 ± 0.0026 nmol/min, KM = 68 ± 9.6 µM, kcat = 3.45 ± 0.11 min . (n = 3).

215

A.

100

80

60

40

20

Total Ion Count (%) Count Ion Total 0 0 1 2 3 4 5 Time (min)

B.

0.040 0.035 0.030 0.025 0.020 0.015 0.010

-debenzoyl- 0.005

N (nmol) 10-deacetylpaclitaxel 0.000 0 20 40 60 80 100 120 140 160 180 200 Time (min)

Figure A.3.18. BAPT activity with 10-DAB and (2R,3S)-PhIS CoA. A. N-debenzoyl-10- deacetylpaclitaxel (m/z 708 → m/z 509) and the internal standard docetaxel (1 µM, m/z 808 → m/z 509) -4 are shown in black and red respectively. B. Timecourse of BAPT activity. Vrel is calculated at 1.9 × 10 nmol/min with ~57 µg/mL MBP-BAPT (200 µL assay). Mass spectrometry performed on a Waters Quattro micro API LC/MS/MS (Waters, Milford, MA).

216

Figure A.3.19. Purified recombinant PheAT. Purified PheAT is 69.6 kDa. Ladder is PageRuler Prestained Ladder (Thermo Fisher, Waltham, MA)

217 A.

B.

C.

Figure A.3.20. Production of N-debenzoyl-2’-deoxypaclitaxel by BAPT. A. Structure of N-debenzoyl-2’-deoxypaclitaxel. B. Total ion chromatogram of BAPT assay. Enzyme products are at 2.3 and 3.0 min respectively. C. Fragment ions of minor peak (2.3 min) of the N-debenzoyl-2’-deoxypaclitaxel ion [M +H]+ = m/z 734. D. Fragment ions of major peak (3.0 min) of the N-debenzoyl-2’-deoxypaclitaxel ion [M +H]+ = m/z 734. Mass spectrometry performed on a Waters Quattro micro API LC/MS/MS (Waters, Milford, MA).

218 Figure A.3.20 (cont’d).

D.

219 A.

B.

C.

Figure A.3.21. Production of N-debenzoylpaclitaxel by BAPT. A. Structure of N-debenzoylpaclitaxel. B. Chromatogram of N-debenzoylpaclitaxel ion [M +H]+ = m/z 750. C. Fragment ions of m/z 750. Mass spectrometry performed on a Waters Quattro micro API LC/MS/MS (Waters, Milford, MA).

220

100

80

60

40

Ion Count (%)

20

0 0 1 2 3 4 5 6 Time (min)

Figure A.3.22. BAPT and PheAT coupled assay production of N-debenzoylpaclitaxel. Chromatogram of N-debenzoylpaclitaxel ([M+H]+ = m/z 750) (black) and the internal standard docetaxel ([M + H]+ = m/z 808) (red). Mass spectrometry performed on a Waters Quattro micro API LC/MS/MS (Waters, Milford, MA).

221 A.

B.

C.

Figure A.3.23. Multi-enzyme biosynthesis of N-furanoyl-N-debenzoylpaclitaxel. A. Structure of N-furanoyl-N-debenzoylpaclitaxel. B. Total ion chromatogram of m/z 844 fragment ions. C. Fragment ions of m/z 844 at 4.1 min. Mass spectrometry performed on a Waters Quattro micro API LC/MS/MS (Waters, Milford, MA).

222

REFERENCES

223 REFERENCES

1. Onrubia, M., Cusido, R. M., Ramirez, K., Hernandez-Vazquez, L., Moyano, E., Bonfill, M., and Palazon, J. (2013) Bioprocessing of plant in vitro systems for the mass production of pharmaceutically important metabolites: paclitaxel and its derivatives, Curr. Med. Chem. 20, 880-891. 2. Malik, S., Cusido, R. M., Mirjalili, M. H., Moyano, E., Palazon, J., and Bonfill, M. (2011) Production of the anticancer drug Taxol in Taxus baccata suspension cultures: A review, Process Biochem. 46, 23-34. 3. Drug Shortages, U.S. Food and Drug Administration. (2015) Silver Spring, MD [accessed 2015, Sep. 27]. http://www.fda.gov/Drugs/DrugSafety/DrugShortages/. 4. Poi, M. J., Berger, M., Lustberg, M., Layman, R., Shapiro, C. L., Ramaswamy, B., Mrozek, E., Olson, E., and Wesolowski, R. (2013) Docetaxel-induced toxicities in breast cancer patients subsequent to paclitaxel shortage: a case series and literature review, Support Care Cancer 21, 2679-2686. 5. Havrilesky, L. J., Garfield, C. F., Barnett, J. C., and Cohn, D. E. (2012) Economic impact of paclitaxel shortage in patients with newly diagnosed ovarian cancer, Gynecol. Oncol. 125, 631-634. 6. Drug Shortages. American Society of Health-Science Pharmacists. (2014). Bethesda, MD. [accessed 2014, Sep.14], http://www.ashp.org. 7. Becker, D. J., Talwar, S., Levy, B. P., Thorn, M., Roitman, J., Blum, R. H., Harrison, L. B., and Grossbard, M. L. (2013) Impact of oncology drug shortages on patient therapy: Unplanned treatment changes, J. Oncol. Pract. 9, e122-128. 8. Link, M. P., Hagerty, K., and Kantarjian, H. M. (2012) Chemotherapy drug shortages in the United States: Genesis and potential solutions, J.Clin. Oncol. 30, 692-694. 9. Ajikumar, P. K., Xiao, W. H., Tyo, K. E. J., Wang, Y., Simeon, F., Leonard, E., Mucha, O., Phon, T. H., Pfeifer, B., and Stephanopoulos, G. (2010) Isoprenoid Pathway Optimization for Taxol Precursor Overproduction in Escherichia coli, Science 330, 70-74. 10. Dejong, J. M., Liu, Y., Bollon, A. P., Long, R. M., Jennewein, S., Williams, D., and Croteau, R. B. (2006) Genetic engineering of Taxol biosynthetic genes in Saccharomyces cerevisiae, Biotechnol. Bioeng. 93, 212-224. 11. Besumbes, O., Sauret-Gueto, S., Phillips, M. A., Imperial, S., Rodriguez- Concepcion, M., and Boronat, A. (2004) Metabolic engineering of isoprenoid biosynthesis in Arabidopsis for the production of taxadiene, the first committed precursor of Taxol, Biotechnol. Bioeng. 88, 168-175.

224 12. Ajikumar, P. K., Xiao, W. H., Tyo, K. E., Wang, Y., Simeon, F., Leonard, E., Mucha, O., Phon, T. H., Pfeifer, B., and Stephanopoulos, G. (2010) Isoprenoid pathway optimization for Taxol precursor overproduction in Escherichia coli, Science 330, 70-74. 13. Zhou, K., Qiao, K., Edgar, S., and Stephanopoulos, G. (2015) Distributing a metabolic pathway among a microbial consortium enhances production of natural products, Nat. Biotechnol. 33, 377-383. 14. Guerra-Bubb, J., Croteau, R., and Williams, R. M. (2012) The early stages of Taxol biosynthesis: An interim report on the synthesis and identification of early pathway metabolites, Nat. Prod. Rep. 29, 683-696. 15. Song, G. H., Zhao, C. F., Zhang, M., Fu, C. H., Zhang, H., and Yu, L. J. (2014) Correlation analysis of the taxane core functional group modification, enzyme expression, and metabolite accumulation profiles under methyl jasmonate treatment, Biotechnol. Progr. 30, 269-280. 16. Ketchum, R. E. B., and Croteau, R. B. (2006) The Taxus metabolome and the elucidation of the Taxol® biosynthetic pathway in cell suspension cultures, In Plant Metabolomics (Saito, K., Dixon, R., and Willmitzer, L., Eds.), pp 291-309, Springer Berlin Heidelberg. 17. Chau, M., Jennewein, S., Walker, K., and Croteau, R. (2004) Taxol biosynthesis: Molecular cloning of a cytochrome P450 characterization taxoid 7--hydroxylase, Chem. Biol. 11, 663-672. 18. Chau, M., Walker, K., Long, R., and Croteau, R. (2004) Regioselectivity of taxoid- O-acetyltransferases: Heterologous expression and characterization of a new taxadien-5-ol-O-acetyltransferase, Arch. Biochem. Biophys. 430, 237-246. 19. Jennewein, S., Rithner, C. D., Williams, R. M., and Croteau, R. B. (2001) Taxol biosynthesis: taxane 13α-hydroxylase is a cytochrome P450-dependent monooxygenase, P. Natl. Acad. Sci. U.S.A. 98, 13595-13600. 20. Ramirez-Estrada, K., Altabella, T., Onrubia, M., Moyano, E., Notredame, C., Osuna, L., Vanden Bossche, R., Goossens, A., Cusido, R. M., and Palazon, J. (2015) Transcript profiling of jasmonate-elicited Taxus cells reveals a - phenylalanine-CoA ligase, Plant Biotechnol. J. 21. Baloglu, E., and Kingston, D. G. I. (1999) The taxane diterpenoids, J. Nat. Prod. 62, 1448-1472. 22. Hampel, D., Mau, C. J. D., and Croteau, R. B. (2009) Taxol biosynthesis: Identification and characterization of two acetyl CoA: taxoid-O-acetyl transferases that divert pathway flux away from Taxol production, Arch. Biochem. Biophys. 487, 91-97.

225 23. Yu, W. B., Liang, X., and Zhu, P. (2013) High-cell-density fermentation and pilot- scale biocatalytic studies of an engineered yeast expressing the heterologous of 7--xylosyltaxanes, J. Ind. Microbiol. Biotechnol. 40, 133- 140. 24. Miralpeix, B., Rischer, H., Hakkinen, S. T., Ritala, A., Seppanen-Laakso, T., Oksman-Caldentey, K. M., Capell, T., and Christou, P. (2013) Metabolic engineering of plant secondary products: Which way forward?, Curr. Pharm. Des. 19, 5622-5639. 25. Chang, M. C., and Keasling, J. D. (2006) Production of isoprenoid pharmaceuticals by engineered microbes, Nat. Chem. Biol. 2, 674-681. 26. Maude, R. J., Woodrow, C. J., and White, L. J. (2010) Artemisinin Antimalarials: Preserving the "Magic Bullet", Drug Develop. Res. 71, 12-19. 27. Ro, D. K., Paradise, E. M., Ouellet, M., Fisher, K. J., Newman, K. L., Ndungu, J. M., Ho, K. A., Eachus, R. A., Ham, T. S., Kirby, J., Chang, M. C. Y., Withers, S. T., Shiba, Y., Sarpong, R., and Keasling, J. D. (2006) Production of the antimalarial drug precursor artemisinic acid in engineered yeast, Nature 440, 940-943. 28. Paddon, C. J., Westfall, P. J., Pitera, D. J., Benjamin, K., Fisher, K., McPhee, D., Leavell, M. D., Tai, A., Main, A., Eng, D., Polichuk, D. R., Teoh, K. H., Reed, D. W., Treynor, T., Lenihan, J., Fleck, M., Bajad, S., Dang, G., Dengrove, D., Diola, D., Dorin, G., Ellens, K. W., Fickes, S., Galazzo, J., Gaucher, S. P., Geistlinger, T., Henry, R., Hepp, M., Horning, T., Iqbal, T., Jiang, H., Kizer, L., Lieu, B., Melis, D., Moss, N., Regentin, R., Secrest, S., Tsuruta, H., Vazquez, R., Westblade, L. F., Xu, L., Yu, M., Zhang, Y., Zhao, L., Lievense, J., Covello, P. S., Keasling, J. D., Reiling, K. K., Renninger, N. S., and Newman, J. D. (2013) High-level semi- synthetic production of the potent antimalarial artemisinin, Nature 496, 528-532. 29. Turconi, J., Griolet, F., Guevel, R., Oddon, G., Villa, R., Geatti, A., Hvala, M., Rossen, K., Göller, R., and Burgard, A. (2014) Semisynthetic Artemisinin, the Chemical Path to Industrial Production, Org. Process Res. Dev. 18, 417-422. 30. Huang, Q. L., Roessner, C. A., Croteau, R., and Scott, A. I. (2001) Engineering Escherichia coli for the synthesis of taxadiene, a key intermediate in the biosynthesis of taxol, Bioorg. Med. Chem. 9, 2237-2242. 31. Rontein, D., Onillon, S., Herbette, G., Lesot, A., Werck-Reichhart, D., Sallaud, C., and Tissier, A. (2008) CYP725A4 from yew catalyzes complex structural rearrangement of taxa-4(5),11(12)-diene into the cyclic ether 5(12)-oxa-3(11)- cyclotaxane, J. Biol. Chem. 283, 6067-6075. 32. Chang, M. C. Y., Eachus, R. A., Trieu, W., Ro, D. K., and Keasling, J. D. (2007) Engineering Escherichia coli for production of functionalized terpenoids using plant P450s, Nat. Chem. Biol. 3, 274-277.

226 33. Jennewein, S., Park, H., DeJong, J. M., Long, R. M., Bollon, A. P., and Croteau, R. B. (2005) Coexpression in yeast of Taxus cytochrome P450 reductase with cytochrome P450 oxygenases involved in Taxol biosynthesis, Biotechnol. Bioeng. 89, 588-598. 34. Leonard, E., and Koffas, M. A. (2007) Engineering of artificial plant cytochrome P450 enzymes for synthesis of isoflavones by Escherichia coli, Appl. Environ. Microbiol. 73, 7246-7251. 35. Lundemo, M. T., and Woodley, J. M. (2015) Guidelines for development and implementation of biocatalytic P450 processes, Appl. Microbiol. Biotechnol. 99, 2465-2483. 36. Vongpaseuth, K., Nims, E., Amand, M. S., Walker, E. L., and Roberts, S. C. (2007) Development of a particle bombardment-mediated transient transformation system for Taxus spp. cells in culture, Biotechnol. Progr. 23, 1180-1185. 37. Ketchum, R. E. B., Wherland, L., and Croteau, R. B. (2007) Stable transformation and long-term maintenance of transgenic Taxus cell suspension cultures, Plant Cell Rep. 26, 1025-1033. 38. Li, S. T., Fu, C. H., Zhang, M., Zhang, Y., Xie, S., and Yu, L. J. (2012) Enhancing Taxol biosynthesis by overexpressing a 9-cis-epoxycarotenoid dioxygenase gene in transgenic cell lines of Taxus chinensis, Plant Mol. Biol. Rep. 30, 1125-1130. 39. Mirjalili, N., and Linden, J. C. (1996) Methyl jasmonate induced production of Taxol in suspension cultures of Taxus cuspidata: Ethylene interaction and induction models, Biotechnol. Progr. 12, 110-118. 40. Furmanowa, M., Glowniak, K., SyklowskaBaranek, K., Zgorka, G., and Jozefczyk, A. (1997) Effect of picloram and methyl jasmonate on growth and taxane accumulation in callus culture of Taxus x media var. Hatfieldii, Plant Cell Tiss. Org. 49, 75-79. 41. Lenka, S. K., Nims, N. E., Vongpaseuth, K., Boshar, R. A., Roberts, S. C., and Walker, E. L. (2015) Jasmonate-responsive expression of paclitaxel biosynthesis genes in Taxus cuspidata cultured cells is negatively regulated by the bHLH transcription factors TcJAMYC1, TcJAMYC2, and TcJAMYC4, Front. Plant Sci. 6. 42. Muchiri, R., and Walker, K. D. (2012) Taxol biosynthesis: Tyrocidine synthetase A catalyzes the production of phenylisoserinyl CoA and other amino phenylpropanoyl thioesters, Chem. Biol. 19, 679-685. 43. Walker, K., Fujisaki, S., Long, R., and Croteau, R. (2002) Molecular cloning and heterologous expression of the C-13 phenylpropanoid side chain-CoA acyltransferase that functions in Taxol biosynthesis, P. Natl. Acad. Sci. U.S.A. 99, 12715-12720.

227 44. Georg, G. I., Boge, T. C., Cheruvallath, Z. S., Harriman, G. C., Hepperle, M., Park, H., and Himes, R. H. (1994) Topliss approach to the synthesis of biologically active substituted N-benzoyl Taxol analogues, Bioorg. Med. Chem. Lett. 4, 1825-1830. 45. Georg, G. I., Harriman, G. C. B., Hepperle, M., Clowers, J. S., VanderVelde, D. G., and Himes, R. H. (1996) Synthesis, conformational analysis, and biological evaluation of heteroaromatic taxanes, J. Org. Chem. 61, 2664-2676. 46. Karliga, B., Schilling, J. K., Kingston, D. G., Bane, S., Ravindra, R., and Talinli, N. (2006) Synthesis and Biological Evaluation of N‐(Arylsulfanyl) carbonyl Analogues of Paclitaxel (Taxol), Chemistry & biodiversity 3, 396-404. 47. Commerçon, A., Bourzat, J., Didier, E., and Lavelle, F. (1995) Practical semisynthesis and antimitotic activity of docetaxel and side-chain analogs, Taxane Anticancer Agents 583, 233-246. 48. Muchiri, R. N. (2015) Repurposing a non-ribosomal peptide synthetase (tyrocidine synthetase A) for amino acyl CoA biosynthesis, Michigan State University. 49. Ojima, I., Park, Y. H., Fenoglio, I., Duclos, O., Sun, C. M., Kuduk, S. D., Zucco, M., Appendino, G., Pera, P., Veith, J. M., Bernacki, R. J., Bissery, M. C., Combeau, C., Vrignaud, P., Riou, J. F., and Lavelle, F. (1995) Synthesis and structure-activity relationships of new taxoids, In Taxane Anticancer Agents: Basic Science and Current Status (Georg, G. I., Chem, T. T., Ojima, I., and Vyas, D. M., Eds.), pp 262-275, Amer. Chemical Soc., Washington. 50. Baloglu, E., Hoch, J. M., Chatterjee, S. K., Ravindra, R., Bane, S., and Kingston, D. G. (2003) Synthesis and biological evaluation of C-3'NH/C-10 and C-2/C-10 modified paclitaxel analogues, Bioorg. Med. Chem. 11, 1557-1568. 51. Georg, G. I., Cheruvallath, Z. S., Harriman, G. C. B., Hepperle, M., Park, H., and Himes, R. H. (1994) Synthesis and biology of substituted 3'-phenyl Taxol analogs, Bioorg. Med. Chem. Lett. 4, 2331-2336. 52. Boge, T. C., Himes, R. H., Vander Velde, D. G., and Georg, G. I. (1994) The effect of the aromatic rings of Taxol on biological activity and solution conformation: Synthesis and evaluation of saturated Taxol and Taxotere analogues, J. Med. Chem. 37, 3337-3343. 53. Shionoya, M., Jimbo, T., Kitagawa, M., Soga, T., and Tohgo, A. (2003) DJ-927, a novel oral taxane, overcomes P-glycoprotein-mediated multidrug resistance in vitro and in vivo, Cancer Sci. 94, 459-466. 54. Sampath, D., Greenberger, L. M., Beyer, C., Hari, M., Liu, H., Baxter, M., Yang, S., Rios, C., and Discafani, C. (2006) Preclinical pharmacologic evaluation of MST-997, an orally active taxane with superior in vitro and in vivo efficacy in paclitaxel- and docetaxel-resistant tumor models, Clin. Cancer Res. 12, 3459-3469.

228 55. Nevarez, D. M., Mengistu, Y. A., Nawarathne, I. N., and Walker, K. D. (2009) An N-aroyltransferase of the BAHD superfamily has broad aroyl CoA specificity in vitro with analogues of N-dearoylpaclitaxel, J. Am. Chem. Soc. 131, 5994-6002. 56. Long, R. M., Lagisetti, C., Coates, R. M., and Croteau, R. B. (2008) Specificity of the N-benzoyl transferase responsible for the last step of Taxol biosynthesis, Arch. Biochem. Biophys. 477, 384-389. 57. Swindell, C. S., Krauss, N. E., Horwitz, S. B., and Ringel, I. (1991) Biologically active Taxol analogs with deleted A-ring sidechain substituents and variable C-2' configurations, J. Med. Chem. 34, 1176-1184. 58. Simon, E. J., and Shemin, D. (1953) The preparation of S-succinyl coenzyme A, J. Am. Chem. Soc. 75, 2520-2520. 59. Rasmussen, J. T., Borchers, T., and Knudsen, J. (1990) Comparison of the binding affinities of acyl-CoA-binding protein and fatty-acid-binding protein for long-chain acyl-CoA esters, Biochem. J. 265, 849-855. 60. Silva, M. F. B., Ruiter, J. P. N., Ijlst, L., Allers, P., ten Brink, H. J., Jakobs, C., Duran, M., de Almedia, I. T., and Wander, R. J. A. (2001) Synthesis and intramitochondrial levels of valproyl-coenzyme A metabolites, Anal. Biochem. 290, 60-67. 61. Ellman, G. L. (1959) Tissue sulfhydryl groups, Arch Biochem Biophys 82, 70-77. 62. Eyer, P., Worek, F., Kiderlen, D., Sinko, G., Stuglin, A., Simeon-Rudolf, V., and Reiner, E. (2003) Molar absorption coefficients for the reduced Ellman reagent: Reassessment, Anal. Biochem. 312, 224-227. 63. Pabsch, K., Petersen, M., Rao, N. N., Alfermann, A. W., and Wandrey, C. (1991) Chemo-enzymatic synthesis of rosmarinic acid, Recl. Trav. Chim. Pays-Bas-J. Roy. Neth. Chem. Soc. 110, 199-205. 64. Mangroo, D., and Gerber, G. E. (1990) Synthesis of acyl-CoA thioesters, Biochem. Cell Biol. 68, 308-312. 65. Kawaguchi, A., Yoshimura, T., and Okuda, S. (1981) A new method for the preparation of acyl CoA thioesters, J. Biochem. 89, 337-339. 66. Al'arif, A., and Blecher, M. (1969) Synthesis of fatty acyl CoA and other thiol esters using N-hydroxysuccinimide esters of fatty acids., J. Res. 10, 344-345. 67. Goldman, P., and Vagelos, P. R. (1961) Specificity of triglyceride synthesis from diglycerides in chicken adipose tissue, J. Biol. Chem. 236, 2620-&. 68. Villiers, B. R. M., and Hollfelder, F. (2009) Mapping the limits of substrate specificity of the adenylation domain of TycA, Chembiochem 10, 671-682.

229 69. Mootz, H. D., and Marahiel, M. A. (1997) The tyrocidine biosynthesis operon of Bacillus brevis: Complete nucleotide sequence and biochemical characterization of functional internal adenylation domains, J. Bacteriol. 179, 6843-6850. 70. Nawarathne, I. N. (2011) Modification of biologically important natural product scaffolds through biocatalysis, using Taxus acyltransferases. 71. D'Ordine, R. L., Paneth, P., and Anderson, V. E. (1995) C13 NMR and H1-H1 NOEs of coenzyme A- Conformation of the pantoic acid moiety, Bioorg. Chem. 23, 169- 181. 72. Wu, W. J., Tonge, P. J., and Raleigh, D. P. (1998) Stereospecific H-1 and C-13 NMR assignments of crotonyl CoA and hexadienoyl CoA: Conformational analysis and comparison with protein-CoA complexes, J. Am. Chem. Soc. 120, 9988-9994. 73. Minkler, P. E., Anderson, V. E., Maiti, N. C., Kerner, J., and Hoppel, C. L. (2004) Isolation and identification of two isomeric forms of malonyl-coenzyme A in commercial malonyl-coenzyme A, Anal. Biochem. 328, 203-209. 74. Wuts, P. G., and Greene, T. W. (2006) Greene's protective groups in organic synthesis, John Wiley & Sons, Hoboken, New Jersey. 75. Han, S. Y., and Kim, Y. A. (2004) Recent development of peptide coupling reagents in organic synthesis, Tetrahedron 60, 2447-2467. 76. Humphrey, J. M., and Chamberlin, A. R. (1997) Chemical synthesis of natural product : Coupling methods for the incorporation of noncoded amino acids into peptides, Chem. Rev. 97, 2243-2266. 77. Kingston, D. G. (2007) The shape of things to come: Structural and synthetic studies of Taxol and related compounds, Phytochemistry 68, 1844-1854. 78. Long, R. M., and Croteau, R. (2005) Preliminary assessment of the C13-side chain 2 '-hydroxylase involved in Taxol biosynthesis, Biochem. Bioph. Res. Co. 338, 410- 417. 79. Geissler, J. F., Harwood, C. S., and Gibson, J. (1988) Purification and properties of benzoate-coenzyme A ligase, a Rhodopseudomonas palustris enzyme involved in the anaerobic degradation of benzoate, J. Bacteriol. 170, 1709-1714. 80. Walker, K., Long, R., and Croteau, R. (2002) The final acylation step in Taxol biosynthesis: Cloning of the taxoid C13-side-chain N-benzoyltransferase from Taxus, P. Natl. Acad. Sci. U.S.A. 99, 9166-9171. 81. Ashkenazy, H., Erez, E., Martz, E., Pupko, T., and Ben-Tal, N. (2010) ConSurf 2010: Calculating evolutionary conservation in sequence and structure of proteins and nucleic acids, Nucleic Acids Res. 38, W529-533.

230 82. Celniker, G., Nimrod, G., Ashkenazy, H., Glaser, F., Martz, E., Mayrose, I., Pupko, T., and Ben-Tal, N. (2013) ConSurf: Using evolutionary data to raise testable hypotheses about protein function, Isr. J. Chem. 53, 199-206. 83. Berezin, C., Glaser, F., Rosenberg, J., Paz, I., Pupko, T., Fariselli, P., Casadio, R., and Ben-Tal, N. (2004) ConSeq: The identification of functionally and structurally important residues in protein sequences, Bioinformatics 20, 1322-1324. 84. Suzuki, H., Nakayama, T., and Nishino, T. (2003) Proposed mechanism and functional amino acid residues of malonyl-CoA: anthocyanin 5-O-glucoside-6 '''- O-malonyltransferase from flowers of Salvia splendens, a member of the versatile plant acyltransferase family, Biochemistry-US 42, 1764-1771. 85. Bayer, A., Ma, X. Y., and Stockigt, J. (2004) Acetyltransfer in natural product biosynthesis - functional cloning and molecular analysis of vinorine synthase, Bioorg. Med. Chem. 12, 2787-2795. 86. Kallberg, M., Margaryan, G., Wang, S., Ma, J., and Xu, J. (2014) RaptorX server: a resource for template-based protein structure modeling, Methods in Molecular Biology 1137, 17-27. 87. Kallberg, M., Wang, H., Wang, S., Peng, J., Wang, Z., Lu, H., and Xu, J. (2012) Template-based protein structure modeling using the RaptorX web server, Nat. Protoc. 7, 1511-1522. 88. Peng, J., and Xu, J. (2011) RaptorX: Exploiting structure information for protein alignment by statistical inference, Proteins 79 Suppl. 10, 161-171. 89. Lallemand, L. A., Zubieta, C., Lee, S. G., Wang, Y., Acajjaoui, S., Timmins, J., McSweeney, S., Jez, J. M., McCarthy, J. G., and McCarthy, A. A. (2012) A structural basis for the biosynthesis of the major chlorogenic acids found in coffee, Plant Physiol. 160, 249-260. 90. Manjasetty, B. A., Yu, X. H., Panjikar, S., Taguchi, G., Chance, M. R., and Liu, C. J. (2012) Structural basis for modification of flavonol and naphthol glucoconjugates by Nicotiana tabacum malonyltransferase (NtMaT1), Planta 236, 781-793. 91. Unno, H., Ichimaida, F., Suzuki, H., Takahashi, S., Tanaka, Y., Saito, A., Nishino, T., Kusunoki, M., and Nakayama, T. (2007) Structural and mutational studies of anthocyanin malonyltransferases establish the features of BAHD enzyme catalysis, J. Biol. Chem. 282, 15812-15822. 92. Withers, S., Lu, F., Kim, H., Zhu, Y., Ralph, J., and Wilkerson, C. G. (2012) Identification of grass-specific enzyme that acylates monolignols with p- coumarate, J. Biol. Chem. 287, 8347-8355. 93. Dexter, R., Qualley, A., Kish, C. M., Ma, C. J., Koeduka, T., Nagegowda, D. A., Dudareva, N., Pichersky, E., and Clark, D. (2007) Characterization of a petunia

231 acetyltransferase involved in the biosynthesis of the floral volatile isoeugenol, Plant J. 49, 265-275. 94. D'Auria, J. C., Reichelt, M., Luck, K., Svatos, A., and Gershenzon, J. (2007) Identification and characterization of the BAHD acyltransferase malonyl CoA: anthocyanidin 5-O-glucoside-6''-O-malonyltransferase (At5MAT) in Arabidopsis thaliana, FEBS Lett. 581, 872-878. 95. Gou, J. Y., Yu, X. H., and Liu, C. J. (2009) A hydroxycinnamoyltransferase responsible for synthesizing suberin aromatics in Arabidopsis, P. Natl. Acad. Sci. U.S.A. 106, 18855-18860. 96. Schmidt, G. W., Jirschitzka, J., Porta, T., Reichelt, M., Luck, K., Torre, J. C., Dolke, F., Varesio, E., Hopfgartner, G., Gershenzon, J., and D'Auria, J. C. (2015) The last step in biosynthesis is catalyzed by a BAHD acyltransferase, Plant Physiol. 167, 89-101. 97. Dall'Acqua, W., and Carter, P. (2000) Substrate-assisted catalysis: Molecular basis and biological significance, Protein Sci. 9, 1-9. 98. Kosloff, M., Zor, T., and Selinger, Z. (2000) Substrate-assisted catalysis: Implications for and drug design, Drug Develop. Res. 50, 250-257. 99. Carter, P., Abrahmsen, L., and Wells, J. A. (1991) Probing the mechanism and improving the rate of substrate-assisted catalysis in subtilisin BPN', Biochemistry- US 30, 6142-6148. 100. Carter, P., and Wells, J. A. (1988) Dissecting the of a protease, Nature 332, 564-568. 101. Matsumura, I., and Kirsch, J. F. (1996) Is aspartate 52 essential for catalysis by chicken egg white lysozyme? The role of natural substrate-assisted hydrolysis, Biochemistry-US 35, 1881-1889. 102. Jeltsch, A., Pleckaityte, M., Selent, U., Wolfes, H., Siksnys, V., and Pingoud, A. (1995) Evidence for substrate-assisted catalysis in the DNA cleavage of several restriction endonucleases, Gene 157, 157-162. 103. Kleine, H., Poreba, E., Lesniewicz, K., Hassa, P. O., Hottiger, M. O., Litchfield, D. W., Shilton, B. H., and Luscher, B. (2008) Substrate-assisted catalysis by PARP10 limits its activity to mono-ADP-ribosylation, Mol. Cell 32, 57-69. 104. Zeper, A., Fang, Q., Liang, X., and Takayama, M. (2000) Study on fragmentation behavior of taxoids by tandem mass spectrometry, Chinese Sci. Bull. 45, 688-698. 105. McClure, T. D., Schram, K. H., and Reimer, M. L. (1992) The mass spectrometry of Taxol, J. Am. Soc. Mass Spectrom. 3, 672-679.

232 106. Georg, G. I., Cheruvallath, Z. S., Himes, R. H., Mejillano, M. R., and Burke, C. T. (1992) Synthesis of biologically active Taxol analogues with modified phenylisoserine side chains, J. Med. Chem. 35, 4230-4237. 107. Tabata, H. (2004) Paclitaxel production by plant-cell-culture technology, In Biomanufacturing, pp 1-23, Springer. 108. Bringi, B., Kadkade, P.G., Prince, C. L., Roach, B.L. (2013) Enhanced production of Taxol and taxanes by cell cultures of Taxus species, (Office, U. S. P. a. T., Ed.), Phyton Holdings, LLC, United States of America. 109. Loncaric, C., Merriweather, E., and Walker, K. D. (2006) Profiling a Taxol pathway 10--acetyltransferase: Assessment of the specificity and the production of baccatin III by in vivo acetylation in E. coli, Chem. Biol. 13, 309-317.

233