Download The

Mechanistic Studies on 4-Dimethylallyltryptophan Synthase and the N-Prenyltransferase CymD

QI QIAN

B.Sc., Peking University, 2009

A THESIS SUMBITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Chemistry)

The University of British Columbia (Vancouver)

August, 2015

Prenylated Indole alkaloids comprise a large group of biologically active molecules that include the ergot alkaloids. Prenylation is often important for the activity of these compounds and is catalyzed by an emerging new class of enzyme, the indole prenyltransferases. These enzymes are metal independent and share a unique αββα fold. 4-Dimethylallyltryptophan synthase

(DMATS) is an indole prenyltransferase that transfers the dimethylallyl group onto the C-4 position of L-tryptophan, in the first committed step of ergot alkaloid biosynthesis. It was previously shown to employ a dissociative mechanism, and two important catalytic residues, E89 and K174, have been identified from crystallographic studies. In this work, four mutants were prepared by mutating

E89 and K174 to either glutamine or alanine. The results from kinetic studies and positional isotope exchange (PIX) experiments on all four mutants were consistent with the roles proposed for these two residues. Upon examination of the products in the mutant-catalyzed reactions, one unusual product was identified from the mutant K174A. A hexahydropyrroloindole structure was first proposed and later confirmed by obtaining an authentic sample through chemical synthesis. After examining the positioning of the substrates in the active site, a new mechanism involving a Cope rearrangement was proposed for DMATS.

Another indole prenyltransferase CymD catalyzes a ‘reverse’ prenylation on the N-1 position of L-tryptophan. In this work, a series of mechanistic studies were carried out to probe its mechanism. Fluorinated tryptophan analogs demonstrated a modest effect on the rate of catalysis, suggesting no positive charge accumulation on the indole ring. A krel value of 1.0 × 10-2 was

ii determined with E-F-DMAPP, indicating that significant positive charge accumulates on the allylic moiety during the transition state of catalysis. PIX experiments with L-tryptophan did not show any isotopic scrambling, however, isotopic scrambling was observed with fluorinated tryptophans.

This indicates that a discrete allylic carbocation intermediate is generated. Lastly, solvent kinetic studies presented a primary KIE of 2.3, indicating that the deprotonation of the N-H is a rate- determining step. A hybrid mechanism was proposed for CymD, in which dissociation first forms an allylic cation and then deprotonation direct the indole for nucleophilic attack.

iii

Preface

A version of Chapter 2 has been published and some of the experimental results are reproduced with permission from: Qian, Q.*; Luk, L. Y. P.*; Tanner, M. E. J. Am. Chem. Soc. 2011,

32, 12342-12345 (© 2011 American Chemical Society). Luk carried out the site-directed mutagenesis and prepared all the plasmids for the mutants. Luk and the author of this thesis carried out the protein overexpression and most of the PIX experiments together. The kinetic studies and the characterization of the unusual product from mutant K174A reported in this publication and

Chapter 2 were performed by the author of this thesis under the supervision of Professor Tanner.

A version of Chapter 3 has been published and some of the experimental results are reproduced with permission from: Qian, Q.; Schultz, A. W.; Moore, B. S.; Tanner, M. E.

University of California, San Diego. However, all the mechanistic studies reported in this publication and Chapter 3 were performed by the author of this thesis under the supervision of

Professor Tanner.

* co-first authorship

Table of Contents

Abstract ...... ii

Preface ...... iv

Table of Contents ...... v

List of Tables ...... ix

List of Figures ...... x

List of Symbols and Abbreviations ...... xiv

Acknowledgements ...... xx Dedication ...... xxi

Chapter 1. Alkaloids and Prenyltransferases ...... 1

1.2.1 Terpenoid Indole Alkaloids ...... 6

1.2.2 Prenylated Indole Alkaloids ...... 8

1.3.1 Isoprenyl Pyrophosphate Synthases ...... 13

1.3.2 Protein Prenyltransferases ...... 21

1.3.3 Aromatic Prenyltranferases ...... 25

Chapter 2. Mechanistic Studies on 4-Dimethylallyltryptophan Synthase ...... 45

2.1.1 Preparation of Mutant Genes by Site-Directed Mutagenesis ...... 49

2.1.2 Overexpression of 4-Dimethylallyltryptophan Synthase Mutants ...... 49

2.3.1 Introduction to Positional Isotope Exchange ...... 54

2.3.2 Synthesis of [1-18O]-DMAPP for PIX experiments ...... 55

2.3.3 PIX Experiments on DMAT Synthase Mutants ...... 56

2.3.3.1 PIX Reactions Catalyzed by DMAT Synthase Mutants E89A and E89Q .... 59

2.3.3.2 PIX Reactions Catalyzed by the DMATs Mutants K174A and K174Q ...... 63

2.4.1 Examination of the Enzymatic Reaction Products Using 1H NMR Spectroscopy . 67

2.4.2 CE Analysis of the Products from the K174A Catalyzed Reaction ...... 70

2.4.3 Confirmation of the Structure of Product X by Synthesis ...... 72

2.7.1 Attempts to Convert the Reverse-Prenylated Compound X to DMAT ...... 86

2.7.2 Substrate Analog Studies Aimed at Trapping the Iminium Intermediate ...... 87

2.9.1 Materials and General Methods ...... 94

2.9.2 Overexpression and Purification of the DMATs Mutants ...... 95

2.9.3 Steady-State Kinetic Characterization of DMAT synthase Mutants ...... 96

2.9.4 Synthesis [1-18O]-DMAPP (20) ...... 96

2.9.5 PIX Experiments on DMAT Synthase Mutants ...... 97

2.9.6 Examination of the Enzyme Reaction Products Using 1H NMR Spectroscopy ..... 98 vi

2.9.7 CE Analysis of the Products from the K174A Reaction ...... 98

2.9.8 Synthesis of Compound X ...... 99

2.9.9 Synthesis of 4-Aza-Tryptophan (47) ...... 100

2.9.10 Synthesis of (S)-2-Amino-3-(1H-pyrrol-3-yl)propanoic acid (49) ...... 101

Chapter 3. Mechanistic Studies on Cyclomarin/Cyclomarazine N-prenyltransferase CymD ……………………………………………………………...………………………..102

3.3.1 Characterization of Activity Using 1H NMR Spectroscopy ...... 107

3.3.2 Steady-State Kinetic Characterization of CymD ...... 108

3.4.1 Characterization of CymD Activity with Fluorinated Substrate Anaglogs ...... 110

3.4.2 Kinetic Studies Using Fluorinated Substrates ...... 112

3.4.3 Mechanistic Studies Using Fluorinated DMAPP Analogs ...... 115

3.5.1 PIX Experiment with L-Tryptophan ...... 118

3.5.2 PIX Experiments on Fluorinated Tryptophan ...... 121

3.6.1 Introduction ...... 124

3.6.2 Application of A Solvent Kinetic Isotope Effect Study on CymD ...... 126

3.9.1 Materials and General Methods ...... 134

3.9.2 Protein Purification ...... 135

3.9.3 Activity Assays of CymD on Tryptophan and Its Fluorinated Analogs ...... 136 vii

3.9.4 Kinetics Studies on CymD ...... 136

3.9.5 Positional Isotope Exchange Studies on CymD ...... 137 3.9.6 Solvent Kinetic Isotope Effect Measurement ...... 138

Bibliography ...... 140

viii

List of Tables

Table 1.1 The comparison of the krel values of fluorinated GPP analogs found in FPPs reactions and solvolysis……………………….………………………….…………………..……..……20

Table 2.1 Kinetic constants of wild type and mutant enzymes……………………...…………....53

Table 2.2 The measured partitioning ratios for the wild-type DMATs and mutant-catalyzed PIX reactions………………………………...……………………………..………….…..……66

Table 3.1 Kinetic constants obtained with the fluorinated tryptophans….………………..…….112

Table 3.2 Calculated proton affinities for substituted indoles (kcal/mol)…...…………………..113

rel Table 3.3 The k values of monofluorinated analogs from different SN1 or SN2 reactions…….117

List of Figures

Figure 1.1 Examples of important alkaloids...... 3 Figure 1.2 Some simple indole alkaloids and terpenoid indole alkaloids...... 6 Figure 1.3 Biosynthesis of the precursor strictosidine in terpenoid indole alkaloid formation. .... 7 Figure 1.4 Prenylation of indole core leads to a variety of prenylated indole alkaloids...... 9 Figure 1.5 Some important ergot alkaloids...... 11 Figure 1.6 A general scheme for a prenyltransferase reaction...... 12 Figure 1.7 Isoprene biosynthesis catalyzed by an IPP synthase...... 13

Figure 1.8 Two mechanisms, SN1 and SN2, proposed for FPP synthase...... 15 Figure 1.9 Stereochemical investigations of the FPP synthase-catalyzed reaction using isotopically labeled substrates...... 16

Figure 1.10 Trifluoromethyl substituted dimethylallyl pyrophosphates E-CF3-DMAPP (E-4) and Z-CF3-DMAPP (Z-4), dimethylallyl methanesulfonate (DMA-Ms, 5) and its fluorinated analogs E-6 and Z-6...... 17 Figure 1.11 Fluorinated geranyl pyrophosphate (2-F-GPP, 7) and fluorinated geranyl methanesulfonate (8)...... 19 Figure 1.12 Fluorinated geranyl pyrophosphate (9-11) and the corresponding methanesulfonate derivatives (12-14)...... 19 Figure 1.13 Crystal structure of FPP synthase and graphic representation of the active site with FPP and IPP bound...... 21 Figure 1.14 Protein prenylation catalyzed by protein prenyltransferases...... 22

Figure 1.15 Proposed SN1 and SN2 mechanisms for protein farnesyltransferase...... 23

Figure 1.16 Fluorinated FPP analogs 3-CH2F-FPP (15) and 3-CF3-FPP (16)...... 25 Figure 1.17 A General reaction catalyzed by an aromatic prenyltransferase...... 25 Figure 1.18 PT-barrel fold found in the structure of the ABBA prenyltransferase NphB of Streptomyces...... 27 Figure 1.19 Aromatic prenyltransferases NphB, CloQ and NovQ in the biosynthesis of natural products...... 28 Figure 1.20 Indole prenyltransferases can prenylate every position of the indole core...... 30 Figure 1.21 Prenylation catalyzed by 4-DMAT synthase...... 32 x

Figure 1.22 Investigation of stereochemistry in the reaction catalyzed by DMAT synthase using isotopically labeled substrates 17, 18 and 19...... 33

Figure 1.23 SN1 and SN2 mechanisms proposed for 4-DMAT synthase...... 34 Figure 1.24 Positional isotope exchange observed when using [1-18O]-DMAPP (20) together with L-Trp in DMAT synthase-catalyzed reaction...... 38

Figure 1.25 Deuterium-labeled substrate analogs [1, 1-2H]-DMAPP (21) and [4-2H]-L-tryptophan (22) used in KIE experiments and D,L-6-fluorotryptophan (23)...... 39 Figure 1.26 The dissociative mechanism for DMAT reaction and the hypothesized reaction coordinate...... 39 Figure 1.27 The 3D architecture of DMAT synthase (top left), the active center of DMAT synthase (top right), and DMAPP analog DMSPP (bottom)...... 40 Figure 1.28 Proposed involvement of two residues Lys174 and Glu89 in the DMATS reaction...... 41 Figure 1.29 Cyclomarin/cyclomarazine N-prenyltransferase CymD catalyze the first step in the biosynthesis of cylcomarin A and cyclomarazine A and produces N-DMAT...... 42 Figure 2.1 Mutation of E89 and K174 increases the energy barriers of the nucleophilic attack and rearomatization steps...... 47 Figure 2.2 Illustration of the continuous coupled pyrophosphate assay...... 51 Figure 2.3 Kinetics plots for the wild-type and mutant DMAT synthases...... 53 Figure 2.4 A schematic representation for the isotopic scrambling in PIX experiments with a phosphate ester...... 55 Figure 2.5 Synthesis of [1-18O]-DMAPP (20)...... 56 Figure 2.6 31P NMR spectra showing the α-phosphorus signals of the DMAPP mixture in the PIX reactions catalyzed by the E89 mutants...... 60 Figure 2.7 31P NMR spectra showing the β-phosphorus signals of the DMAPP in the E89A- catalyzed PIX reaction...... 62 Figure 2.8 31P NMR spectra showing the α-phosphorus signals of the DMAPP mixture in the PIX reactions catalyzed by the K174 mutants...... 64

1 Figure 2.9 H NMR spectrum (400 MHz, D2O) of the K174A reaction mixture...... 69 Figure 2.10 Possible explanations for the production of reverse-prenylated products in the K174A catalyzed reaction...... 70 Figure 2.11 UV spectra of DMAT and the unusual product (X) and the structure of chimonanthine...... 72 xi

Figure 2.12 Synthesis of compound X using Danishefsky’s procedure...... 74

1 Figure 2.13 A comparison of H NMR spectra (D2O) of compound X obtained from the K174A reaction, chemical synthesis and the mixture of both...... 75 Figure 2.14 Two possible pathways leading to the formation of the ‘reverse’ prenylated compound X...... 76

Figure 2.15 Active site architecture of DMAT synthase in complex with L-tryptophan and DMSPP...... 78 Figure 2.16 A Cope mechanism for the reactions catalyzed by the wild-type DMAT synthase and the K174A mutant...... 79 Figure 2.17 Hypothetical Cope rearrangement in ergot alkaloids biosynthesis proposed by Wenkert and Sliwa...... 80 Figure 2.18 Studies of the Cope rearrangement with model compounds carried out by Wenkert.[16] ...... 81 Figure 2.19 Model compound study from the Arigoni group...... 82 Figure 2.20 Non-enzymatic reaction involving a Cope rearrangement onto the C-4 position of an indole...... 83 Figure 2.21 Cope rearrangement reaction under ambient temperature...... 84 Figure 2.22 Studies on 4-DMATS using C-4 substituted Trp analogs...... 85 Figure 2.23 Cope rearrangement under biomimetic conditions...... 86 Figure 2.24 Proposed reactions with substrate analogs...... 88 Figure 2.25 Enzymatic synthesis of 4-aza-Trp (47) using tryptophan synthase...... 89 Figure 2.26 Synthetic scheme for analog 49...... 90 Figure 3.1 The prenylation reaction catalyzed by CymD...... 103 Figure 3.2 Two possible dissociative mechanisms proposed for CymD...... 104 Figure 3.3 A possible associative mechanism for CymD...... 105

1 Figure 3.4 H NMR spectrum (400 MHz, D2O) of N-DMAT in the enzyme reaction mixture. 108

Figure 3.5 Kinetic plot for CymD with varying concentrations of L-tryptophan and 20 μM DMAPP...... 110 Figure 3.6 Two fluorinated substrate analogs studied in the enzyme reaction...... 111 Figure 3.7 Kinetics plots of the CymD reaction with fluorinated tryptophans 55 and 23...... 112 Figure 3.8 E-F-DMAPP (56) and the kinetic rates measured for DMAPP and E-F-DMAPP.. . 116 xii

Figure 3.9 Proposed PIX if a dissociative mechanism were operative for CymD...... 119

Figure 3.10 31P NMR spectra showing the α–phosphorus signals for the PIX experiment with L- Trp...... 120 Figure 3.11 31P NMR spectra showing the α–phosphorus signals for the PIX experiment with 4- F-Trp (55)...... 122 Figure 3.12 31P NMR spectra showing the α–phosphorus signals for the PIX experiment with 6- F-Trp (23)...... 123 Figure 3.13 Kinetic plots of rate/[E] vs. substrate concentration demonstrating a solvent kinetic isotope effect...... 127 Figure 3.14 Proposed hybrid mechanism for the reaction catalyzed by CymD...... 130 Figure 3.15 Reactions catalyzed by N-prenyltransferases FtmPT2 and cTrpPT...... 132 Figure 3.16 Cyclopiazonic acid (CPA) biosynthesis pathway in Aspergillus sp...... 133 Figure 3.17 Proposed mechanism for CpaD and its possible product from mutant K177A...... 134

xiii

List of Symbols and Abbreviations

ε molar absorptivity

ε360 extinction coefficient at 360 nm

λmax absorption maximum

ABBA a unique protein fold found in certain aromatic prenyltransferases,

such as NphB and DMATS

Ac acetyl

AcOH acetic acid

Ado adenosine

ADP adenosine diphosphate

ATP adenosine triphosphate

Boc tert-butyloxycarbonyl

Bn benzyl

BSA bovine serum albumin

Cbz carboxybenzyl cymd gene encoding for N-prenyltransferase CymD

D deuterium (2H)

Da Dalton

DFT density function theory

Dk deuterium kinetic isotope effect

xiv

DMAPP dimethylallyl pyrophosphate

DMAT dimethylallyltryptophan

DMATS dimethylallyltryptophan synthase dmaW gene of dimethylallyltryptophan synthase from Claviceps sp.

DNA deoxyribonucleic acid

E. coli Escherichia coli

EDTA ethylenediaminetetraacetate, disodium salt

ESI-MS electrospray ionization mass spectrometry

Et3N triethylamine

EtNH2 ethylamine

Et2O diethyl ether

EtOAc ethyl acetate fgaPT2 gene of dimethylallyltryptophan synthase from Aspergillus

fumigatus

FPP farnesyl pyrophosphate

FPPs farnesyl pyrophosphate synthase

GPP geranyl pyrophosphate

HPLC high performance liquid chromatography iPrOH isopropyl alcohol

IPP isopentenyl diphosphate

IPTG isopropyl 1-thio-β-D-galactopyranoside xv

J coupling constant (NMR); subscripts indicate coupling partners kcat catalytic rate constant kH/kD deuterium kinetic isotope effect kH, kD rate of reaction involving protiated and deuterated substrates kDa kilodalton

KIE kinetic isotope effect

KM Michaelis constant

LB Luria Broth medium

Me methyl

MeOH methanol

MeOTf methyl trifluoromethanesulfonate

MESG 2-amino-6-mercapto-7-methylpurine ribonucleoside

MWCO molecular weight cut-off

NMR nuclear magnetic resonance

NphB bacterial aromatic prenyltransferase isolated from Streptomyces sp.

strain CL190

N-PSP N-phenylselenophthalimide

OD600 optical density at 600 nm

PPTS pyridinium p-toluenesulfonate p-TsOH p-toluenesulfonic acid

PCR polymerase chain reaction xvi pet. ether petroleum ether

PFTase protein farnesyltransferase

Ph phenyl pH acidity of a aqueous solution

Pi inorganic phosphate

PPi pyrophosphate

PIX positional isotope exchange

Re The substituents on a prochiral carbon decrease in priority

clockwise

Si The substituents on a prochiral carbon decrease in priority

counterclockwise rpm revolutions per minute

RT room temperature

SDM site-directed mutagenesis

SDS-PAGE sodium dodecylsulfate polyacrylamide gel electrophoresis

Sp. species

TEAP triethylammonium phosphate

TFA trifluoroacetic acid

TB Terrific Broth

THF tetrahydrofuran

TMSI trimethylsilyl iodide xvii

Tris 2-amino-2-(hydroxymethyl)-1,3-propanediol

UV ultraviolet

υ initial reaction velocity (rate)

υrxn estimated rate of DMAPP converting to product

υPIX estimated isotopic scrambling rate

υPIX/υrxn estimated partitioning ratio measured in PIX experiment

WT wild type

A Ala alanine

C Cys cysteine

D Asp aspartate

E Glu glutamate

F Phe phenylalanine

G Gly glycine

H His histidine

I Ile isoleucine

K Lys lysine

L Leu leucine

M Met methionine

N Asn asparagine

P Pro proline xviii

Q Gln glutamine

R Arg arginine

S Ser serine

T Thr threonine

V Val valine

W Trp tryptophan

Y Tyr tyrosine

xix

Acknowledgements

I would like to take this chance to thank my supervisor Prof. Martin Tanner for his guidance, patience and endless support through my Ph.D. studies. I doubt I will ever get a better boss. I also owe my thanks to all the present and past Tanner group members. I would like to give my special thanks to past group member Dr. Louis Luk for starting a wonderful project and teaching me all the important laboratory techniques when I joined the Tanner group six years ago. I owe my gratitude to the staff members working in Chemistry Department including Dr. Elena Polishchuk and Dr. Jessi Chen from Biological Services Laboratory, Dr. Paul Xia and Dr. Maria Ezhova from

NMR facility and Dr. Yun Ling and David Wong from mass spectrometry facility. Lastly, I am forever grateful for all the support from my family and friends.

Dedication

people who never stop wondering

xxi

Chapter 1. Alkaloids and Prenyltransferases

Alkaloids

Alkaloids are a large group of nitrogen-containing molecules usually produced in plants, fungi, and bacteria as secondary metabolites. They possess a large variety of biological properties and thus have been widely involved in human history and our daily livies as pharmaceuticals, stimulants and narcotics.[1], [2], [3], [4] Morphine (Figure 1.1), one of many alkaloids found in opium, is a potent drug used to relieve severe pain. It was first isolated from opium poppy by Friedrich

Serturner in 1806, and is believed to be the first isolated plant alkaloid in history.[1-2] However, centuries before the isolation of morphine, opium poppies were cultivated and served as important pain killers. Ultimately, their use as a recreational drug led to the two Opium Wars in

1839 and 1858.[5] Another well-known alkaloid is nicotine (Figure 1.1), which is produced in various plant leaves to serve as an antiherbivoric and antipathogenic compound. As the native component of tobacco, nicotine serves as a stimulant, however, it also causes addiction to tobacco smoking and accounts for a large number of deaths around the globe every year.

Recently more and more alkaloids have been discovered and granted important use in the pharmacological field. For example, vinblastine and taxol are used as anticancer drugs;[1] ergotamine is used for treatment of migraine headaches;[4] and tryprostatin A and tryprostatin B were found to exhibit potent cytotoxity against different kinds of cancer cells.[1],[6] We are also familiar with caffeine in coffee, atropine as an essential medicine, as well as cocaine and the notorious acetylated morphine derivative, heroin (Figure 1.1).[1]

R OH H N N O N O H N N N H NCH3 N N N MeO2C R O H3CO Nicotine OH Caffeine R = OH Morphine N OAc H R = OAc Heroin Vinblastine CO CH H 2 3 HO AcO O OH H O N O N N O H N H O OCH O 3 O O H N O OH AcO OH OOCH2Ph H O Ph NHCOPh O HN Taxol Ergotamine Cocaine

Figure 1.1 Examples of important alkaloids.

Due to their various biological properties, alkaloids have been the subject of extensive research ever since the isolation of morphine. To date, there have been over 20,000 alkaloids identified and characterized, yet there is still not a clear definition of the term ‘alkaloid’ because of the blurry boundary between alkaloids and complex amines.[5] Most researchers think of alkaloids as amino acid-derived, nitrogen-containing alkaline compounds. Amino acids, peptides, proteins, nucleic acids, and nucleotides are usually not considered as alkaloids. Furthermore, many different classification systems are possible for alkaloids due to the great variety in their biochemical origin, chemical structure and pharmacological response. Two broad divisions of alkaloids are non-heterocyclic and heterocyclic. Non-heterocyclic alkaloids, or atypical alkaloids, do not contain nitrogen atoms in a cyclic ring system. Heterocyclic alkaloids, or typical alkaloids,

3 possess nitrogen atoms in a cyclic ring and they can be further divided into 14 subcategories according to their core ring structure.[5] Some of the commonly seen ones include indole alkaloids such as ergotamine and vinblastine, pyridine and piperidine alkaloids such as nicotine and piperine, purine alkaloids such as caffeine, isoquinoline alkaloids such as morphine and heroin and lastly tropane alkaloids such as cocaine.

Early research focused on the structural elucidation of important alkaloids and often involved the chemical synthesis of these compounds. For instance, nicotine’s structure was not characterized until it was synthesized in 1904 despite the fact that it only contains one chiral center. Morphine has five chiral centers and its characterization took almost 150 years after its isolation in 1806.[1] Given their often very complex structures, it was no surprise that scientists raised one question: how does Nature synthesize them in a biological setting? Starting from the late 1950s, researchers began studies of alkaloid biosynthetic pathways by feeding radiolabeled precursors to plants and analyzing the resultant radioactive alkaloids after chemical degradation.[7] Later on, with the development of new analytical technologies, isotopically labeled alkaloids were analyzed by mass spectrometry and nuclear magnetic resonance (NMR) spectroscopy. During the 1970s, researchers started utilizing plant cell culture as source for purifying biosynthetic enzymes involved in alkaloid biosynthesis.[1] And finally in the 1990s, with the advances in techniques of molecular biology, researchers were able to clone the genes involved in alkaloid biosynthesis and overexpress them heterologously, therefore enabling experimental work on individual enzymes.[2, 8], [9] These technical advances greatly facilitated the understanding of alkaloid biosynthesis. 4

Indole Alkaloids

Containing over 4,000 compounds, the indole alkaloids represent one of the largest classes of alkaloids. The unifying feature of these compounds is that they all contain an indole moiety.[5],

[7], [10] Even though the chemical complexity of indole alkaloids can vary greatly, they are mainly derived from tryptophan and its decarboxylation product, tryptamine, with only a few exceptions.[5], [7] Based on their chemical structures, indole alkaloids can be further distinguished into two subgroups, isoprenoids and non-isoprenoids. The latter subgroup consists of simple derivatives of indole (e.g. serotonin), pyrolo-indoles (e.g. physostigmine) and β-carbolines (e.g. harmine, Figure 1.2).[11] Isoprenoid indole alkaloids can be further categorized into terpenoid indole alkaloids (TIAs) and prenylated indole alkaloids.[7], [11], [12]

HO H N O NH2 H3CO N N O N H N H H

Serotonin Physostigmine Harmine

O N N H HO N H O O MeO N H H H OH MeO C O N 2

Camptothecin Ajmalicine Quinine

Figure 1.2 Some simple indole alkaloids and terpenoid indole alkaloids.

1.2.1 Terpenoid Indole Alkaloids

Terpenoid indole alkaloids (TIAs) consist of an indole moiety provided by tryptamine and a terpenoid component derived from the iridoid glucoside secologanin (Figure 1.3) and comprise

~3000 compounds that include anticancer drugs vinblastine (Figure 1.1) and camptothecin

(Figure 1.2), the antimalarial drug quinine (Figure 1.2), and the antihypertensive drug ajmalicine

(Figure 1.2). Some TIAs are proposed to play important roles in the defence mechanism of plants against herbivores and pathogens.[2, 7]

The biosynthesis of terpenoid indole alkaloids has been extensively investigated and it is generally accepted that all terpene indole alkaloids are derived from tryptophan and the iridoid

6 terpene secologanin (Figure 1.3). Tryptophan is converted into tryptamine through a decarboxylation reaction catalyzed by tryptophan decarboxylase.[2, 7] Secologanin is itself a natural product and is believed to be ultimately derived from isoprenyl pyrophosphate (IPP), dimethylallyl pyrophosphate (DMAPP) and glucose. Tryptamine and secologanin then participate in a stereoselective Pictet-Spengler condensation catalyzed by strictosidine synthase to produce strictosidine (S configuration at C-5), which is the common precursor to all TIAs.[2,

7] Strictosidine can then undergo various enzyme-catalyzed rearrangements to yield dramatically different chemical structures, the mechanisms of which remain some of the most interesting problems in secondary metabolism.

COOH OH

NH2 NH2 OH tryptophan O N decarboxylase N H H strictosidine tryptophan O tryptamine synthase OH OH NH O H O H OH 5 OPP O O NH DMAPP CO2Me OPP O MeO2C HO OH OH strictosidine at -5 IPP + glucose secologanin S configuration C

Figure 1.3 Biosynthesis of the precursor strictosidine in terpenoid indole alkaloid formation.

1.2.2 Prenylated Indole Alkaloids

Prenylation refers to the attachment of an isoprenyl group onto receptor molecules and it can occur in a ‘normal’ fashion when the C-1 of the prenyl group connects to the receptor molecule or in a ‘reverse’ fashion when the C-3 of the prenyl group connects to the receptor molecule (Figure 1.4, boxed structure). Prenylation on indole rings leads to a large number of natural products, namely prenylated indole alkaloids. This group of natural products comprises another sub-family of isoprenoid indole alkaloids. Unlike terpenoid indole alkaloids, the isoprenoid moiety in prenylated indole alkaloids mainly derives from the direct attachment of prenyl group from a prenyl donor such as dimethylallyl pyrophosphate (DMAPP) onto the indole core of tryptophan derivatives. These alkaloids are mainly distributed in the fungi of the genera

Claviceps, Penicillium and Aspergillus and often display biological activities that are clearly distinct from their non-prenylated precursors.[12], [13], [6]

3 4 R 3 dimethylallyl 5 tryptophan 2 + 2 1 derivatives R' pyrophosphate 6 N R' 1 (DMAPP) OPP 7 H 'normal' 'reverse'

OH O O MeO OH R O N N HN NH N N N H H H HN O N O H H O

R = OMe A fumitremorgin B tryprostatin echinulin R= H tryprostatin B

H HOOC O H N H N H OH N O N N N H H H O O H N O H amauromine lysergic acid shearinine D Figure 1.4 Prenylation of indole core leads to a variety of prenylated indole alkaloids. Two different prenylation modes are shown in the box and the prenyl moieties in the prenylated indole alkaloids are shown as red.

Due to the different modes of prenylation as well as the broad range of indole precursors, prenylated indole alkaloids possess a wide range of chemical structures (Figure 1.4). It is also worth mentioning that the prenyl group can be installed onto each position of the indole or indoline cores. For instance, the important mycotoxin fumitremorgin B contains prenyl groups at both the N-1 and C-2 positions.[13] Tryptostatin A and B, presumably the biosynthetic precursors of fumitremorgin B, are cyclic dipeptides derived from L-tryptophan and L-proline

9 and are prenylated at the C-2 position. Echinulin not only bears two ‘normal’ prenylations at positions C-5 and C-7 but also carries a ‘reverse’ prenylation at the C-2 position. Reverse prenylation is commonly seen at the C-3 position of the indoline ring as in amauromine. Normal prenylation at the C-4 position of the indole ring gives rise to a large class of important indole alkaloids, namely ergot alkaloids, that include lysergic acid and its derivatives. Prenylation on

C-5 and C-6 is also seen in indole diterpenes such as shearinine D.

Ergot alkaloids comprise the most medicinally important family of prenylated indole alkaloids, as they can interact with various receptors in the nervous system. They share a characteristic tetracyclic ergoline ring system (Figure 1.5) and are mainly produced from fungi of the genera Claviceps, Penicillium and Aspergillus. They have long played important roles in human history.[14] In Europe in the Middle Ages, consumption of rye contaminated with ergot alkaloid producing fungi led to widespread epidemics, known as St Antony’s Fire.[14] On the other hand, ergot alkaloids such as ergometrine and ergotamine have been widely utilized as important pharmaceuticals to prevent maternal bleeding after childbirth and to treat migraine headaches, respectively.

O N OH H HO Ph N O H NEt2 NH NH HN O O N O N N H H H O N AcO H H

NH NH N NH H ergoline lysergic acid N H diethylamide C (LSD) ergometrine ergotamine fumigaclavine Figure 1.5 Some important ergot alkaloids. The characteristic ergoline ring system is shown in the box. The indole core is shown in blue and the C5 units derived from DMAPP are shown in red.

Due to their various potent biological activities and their potential as important pharmaceuticals, the biosynthesis of ergot alkaloids has attracted a great deal of attention in recent years. The Groger and Floss groups were pioneers in this field and carried out many feeding experiments between the 1960s and 1990s.[15], [16], [17], [18] The first gene cluster of ergot alkaloid biosynthesis was identified from Claviceps purpurea in 1995.[19] This opened a new era of direct biochemical studies on the biosynthetic enzymes. Since then, over 80 enzymes responsible for ergot alkaloid biosynthesis have been identified. The enzyme responsible for the first committed step of ergot alkaloid biosynthesis, a tryptophan 4-prenyltransferase, will be studied within this thesis. Prenyltransferases are a class of enzyme that catalyze prenylation reactions and will be discussed in the following section.

Prenyltransferases

Prenyltransferases are widely distributed in Nature and are key enzymes in the biosynthesis of many natural products in plants, fungi, bacteria and eukaryotes.[13, 20], [21] They catalyze prenylation reactions that involve the covalent attachment of a prenyl group onto a receptor molecule. The prenyl group usually derives from isoprenyl diphosphates that have different hydrocarbon chain lengths, including C5 (prenyl), C10 (geranyl), C15 (farnesyl) and C20

(geranylgeranyl). In a prenyltransferase-catalyzed reaction, the receptor molecule serves as a nucleophile and forms a bond with the allylic carbon of the prenyl moiety, releasing pyrophosphate as a side product (Figure 1.6). Based on the type of receptor molecules involved in the reaction, prenyltransferases can be grouped into three main classes: isoprenyl pyrophosphate (IPP) synthases, protein prenyltransferases and aromatic prenyltransferases.[13, 20-

21] The structures, mechanisms and functions of some of these prenyltransferases have been extensively studied and will be discussed in the following sections.

R R

prenyltransferase + Acceptor + PPi OPP Acceptor pyrophosphate isoprenyl prenylated product diphosphate R=(C5H8)nH

Figure 1.6 A general scheme for a prenyltransferase reaction. 12

1.3.1 Isoprenyl Pyrophosphate Synthases

Isoprenyl pyrophosphate synthases (IPPSs) are a class of prenyltransferases that catalyze chain elongation reactions starting from isoprenyl pyrophosphate (IPP) to produce a variety of isoprenes with different chain lengths (Figure 1.7).

OPP

IPP IPP synthase IPP synthase + OPP OPP OPP H+ PPi , GPP n

DMAPP isoprenyl pyrophosphate

Figure 1.7 Isoprene biosynthesis catalyzed by an IPP synthase. ‘n’ can range from 1 to over 1000.

Over the last few decades, many IPPSs have been isolated and their genes have been characterized. Alignment of their amino-acid sequences revealed two conserved aspartate-rich

DDxxD domains, which were proposed to be the binding sites for substrates and enzyme cofactor

Mg2+.[20] Multiple sequence alignments performed by the Poulter group confirmed that IPPSs were derived from a common ancestor.[22] Based on the chain length of the product from the enzyme-catalyzed reaction, IPPSs can be further divided into short-chain (n=2-3 in Figure 1.7), medium-chain (n=4-5), long-chain (n=6-10), and polyprenyl pyrophosphate synthases (n>15).

The short-chain IPPSs are probably the most extensively studied group.[23] 13

Short-chain IPPSs catalyze the synthesis of short-chain isoprenyl pyrophosphates including geranyl pyrophosphate (n=1, GPP in Figure 1.7) and farnesyl pyrophosphate (n=2, both formed by farnesyl pyrophosphate synthase, or FPP synthase) as well as geranylgeranyl pyrophosphate

(n=3, formed by geranylgeranyl pyrophosphate synthase, or GGPP synthase). FPP synthases are commonly found in plants, bacteria, and mammals, where they are involved in the biosynthesis of a variety of important terpenes including sterols (structural components of eukaryotic membranes), carotenes (photoreceptors in visual and photosynthetic systems) and polyprenyl pyrophosphates (membrane-soluble carriers of sugars).[24] FPP synthase was first purified and studied by the Porter and Popják groups in the 1960s.[25], [26] The Porter group purified the enzyme from pig liver and proposed that FPP synthase catalyzed two consecutive transfers of allylic pyrophosphate onto isopentenyl pyrophosphate (IPP). They also showed that for the enzyme to be active, either Mg2+ or Mn2+ was required.[25]

The elucidation of the mechanism employed by FPP synthase has long been the focus of research on this enzyme. A dissociative (SN1) mechanism, also referred to as an ionization- condensation-elimination mechanism, was first proposed in the 1950s when the prenyltransferase was implicated as playing a role in sterol biosynthesis (Figure 1.8, top path).[27],[28] The reaction starts with an ionization step that dissociates DMAPP or GPP into an allylic carbocation and pyrophosphate. The double bond of IPP then adds to the allylic carbocation forming a tertiary carbocation intermediate in the condensation step. A final removal of a C-2 proton in the elimination step produces the new double bond in GPP or FPP.

OPP

condensation S 1 N R OPP H OPP R PPi 3o carbocation elimination ionization IPP intermediate

OPP R OPP attack R enzyme-assisted Enz Enz X OPP R=CH GPP R=CH3, DMAPP X 3, R=C H GPP R=C5H9, FPP 5 9, R OPP 2 H SN displacement HR OPP enzyme-linked intermediate R

Figure 1.8 Two mechanisms, SN1 and SN2, proposed for FPP synthase.

In the 1960s, Cornforth and Popják examined the stereochemistry of the FPP synthase- catalyzed reaction during their studies of squalene biosynthesis.[29], [30], [31] When deuterium- labeled substrate (R)-[1-D]-DMAPP (1) was used, an inversion of configuration at C-1 of

DMAPP was observed (Figure 1.9 top reaction). The configuration of C-1 at isopentenyl pyrophosphate, however, did not undergo any change when isotopically labeled IPP (2) was used.

Additionally, the pro-R proton HR from C-2 of isopentenyl pyrophosphate was always eliminated during the generation of the new trans double bond. Finally, the new bond was formed on the Si face of the double bond in IPP as an (S)-stereoisomer was always produced from trans-[4-T] IPP

(3, Figure 1.9 bottom reaction).

H D H D D H + FPP synthase H D OPP OPP OPP H T T (R)-[1-D]-DMAPP (1R, 2S)-[1-D-2-T]-IPP 1 2

FPP synthase + T OPP OPP H OPP DMAPP trans-[4-T]-IPP T H 3

Figure 1.9 Stereochemical investigations of the FPP synthase-catalyzed reaction using isotopically labeled substrates.

With these observations, Cornforth and Popják proposed a SN2 displacement mechanism for the FPP synthase-catalyzed reaction (Figure 1.8 lower path). In this mechanism, the double bond in IPP is attacked by an enzymatic nucleophile X, which in turn initiates an attack from C-4 of

IPP onto the C-1 of DMAPP or GPP displacing pyrophosphate. This presumably generates an enzyme-linked intermediate. A subsequent trans elimination of nucleophile X and HR at the C-2 position of the enzyme-linked intermediate produces the trans product.[29]

The Poulter group, however, argued that stereochemical studies were not sufficient to distinguish between the SN1 and SN2 mechanisms as both have the potential to invert the configuration at the C-1 position of DMAPP if the substrates were held at specific positions in the enzyme active site.[24] In order to further investigate the two proposed mechanisms, the

Poulter group synthesized a series of fluorinated substrate analogs and tested them in mechanistic studies. Trifluoromethyl (CF3) analogs of DMAPP, (E)- and (Z)-trifluoromethyl-2-butenyl

[24, 32] pyrophosphate (E-4 and Z-4 in Figure 1.10), were first examined. The CF3 moiety has a very similar size to that of methyl group yet its strongly electron-withdrawing nature (σ+ = 0.612) 16 should cause it to drastically reduce the rate of allylic carbocation formation.[24, 32] In contrast, the CF3 moiety was known to slightly accelerate the rate of the SN2 displacement of chloride by iodide in the (E)-2-butenyl system.[33] Thus Poulter and his coworkers predicted that the rate of the enzyme reaction should be severely slowed if the SN1 mechanism was employed while it should be slightly enhanced in the SN2 mechanism. This prediction was somewhat questionable as fluorination could also cause rate depression in an ‘exploded’-SN2 mechanism. Besides, it is also possible that the enzyme might simply employ a different mechanism with the fluorinated substrates. Yet, the authors were the first to employ fluorinated substrate analogs in the enzyme mechanistic studies and they did observe some interesting results.

CF 3 CF3 OMs F3C OPP OPP F3C OMs OMs

Z-CF -DMA-Ms E-CF3-DMAPP Z-CF3-DMAPP DMA-Ms E-CF3-DMA-Ms 3 E-4 Z-4 5 E-6 Z-6

Figure 1.10 Trifluoromethyl substituted dimethylallyl pyrophosphates E-CF3-DMAPP (E- 4) and Z-CF3-DMAPP (Z-4), dimethylallyl methanesulfonate (DMA-Ms, 5) and its fluorinated analogs E-6 and Z-6.

The authors examined the non-enzymatic cationic solvolysis of methanesulfonate analog

(DMA-Ms, 5) and its fluorinated analogs E-6 and Z-6 and they calculated the relative reaction rate (krel) by dividing the rate of catalysis obtained with the fluorinated analog by the one obtained with 5 (Figure 1.10). The krel values of 6 in the cationic solvolysis were found to be

-7 1×10 . When the authors incubated E-CF3-DMAPP (E-4) and Z-CF3-DMAPP (Z-4) with IPP

17 and FPP synthase, the rate of condensation was found to decease by a factor of 3×10-7 (also krel), indicating a great depression of the reaction rate.[24], [34] The similarity between the krel values of solvolysis and ones found with the enzyme reaction greatly supports the SN1 mechanism, in which an allylic carbocation intermediate was involved. Yet the inhibition constants for E-4 (Ki

= 51 µM with GPP and Ki = 23 µM with DMAPP) and Z-4 (Ki = 62 µM with GPP and Ki = 29

µM with DMAPP) were considerably larger than the Michaelis constants for the normal substrates, suggesting a much weaker binding of these two analogs in the enzyme active site.

More surprisingly, mixed-linear patterns were obtained in a Lineweaver-Burk plot, indicating interactions between fluorinated DMAPP and both the DMAPP and IPP binding sites. Thus, the authors felt that the observed rate depression could merely result from non-productive binding modes of the fluorinated substrate analogs.

In order to address the problems that arose from mixed inhibition binding in the first system, the Poulter group analyzed another fluorinated analog, 2-fluorogeranyl pyrophosphate

(2-F-GPP, 7, Figure 1.11).[24, 35] The fluorination at C-2 can still slow down the formation of an allylic carbocation, albeit to a lesser extent than a trifluoromethyl group, while the increased size should prevent it from entering the IPP binding site. These predictions were confirmed by experimental results. The reaction rate was found to be slowed by a factor of 8.4 ×10-4 when 2-

F-GPP was used in the enzyme reaction. This value was similar to the depression factor (4.4×

-3 10 ) observed in the SN1 solvolysis of fluorinated geranyl methanesulfonate (8, Figure 1.11). In addition, the KM value of 2-F-GPP was determined to be 1.1 µM, very close to that of GPP (KM

[35] = 0.8 µM). These results supported the involvement of an SN1 dissociation mechanism. 18

OPP OMs F F 2-F-GPP 2-fluorogeranyl methanesulfonate 7 8

Figure 1.11 Fluorinated geranyl pyrophosphate (2-F-GPP, 7) and fluorinated geranyl methanesulfonate (8).

To further investigate the mechanism, the Poulter group tested another set of fluorinated

GPP analogs that carried either a CH2F (9), a CHF2 (10) or a CF3 group (11) at the C-3 position as alternative substrates (Figure 1.12).[36] The author also measured the krel values of the first- order solvolysis of fluorinated geranyl methanesulfonate analogs 12-14 (Figure 1.12) in acetone- water. Similar to the previous experiments, the krel values of the fluorinated GPP analogs 9-11 found in the enzymatic reactions were comparable to the krel values of the corresponding methanesulfonate derivatives 12-14 in the non-enzymatic solvolysis (Table 1.1).[36] These results agreed with an SN1 mechanism.

R R

OPP OMs

R=CH F 9 R=CH F 12 2 3-CH2F-GPP 2 3-CH2F-G-Ms 10 13 R=CHF 2 3-CHF2-GPP R=CHF 2 3-CHF2-G-Ms R=CF 3-CF -GPP 11 R=CF 3-CF -G-Ms 14 3 3 3 3 Figure 1.12 Fluorinated geranyl pyrophosphate (9-11) and the corresponding methanesulfonate derivatives (12-14).

Table 1.1 The comparison of the krel values of fluorinated GPP analogs found in FPPs reactions and solvolysis. Analogs krel krel

(FPP synthase) (Solvolysis)

Compound 9 and 12 3.7×10-4 7.7×10-4

Compound 10 and 13 4.0×10-8 2.2×10-6

Compound 11 and 14 7.7×10-9 4.0×10-7

The crystal structures of FPP synthases have also been extensively investigated.[20] The

Poulter group reported the first crystal structure of the avian enzyme with a resolution of 2.6 Å

(Figure 1.13).[20, 37] This enzyme is composed of 13 α helices, with 10 of them forming a large central cavity around the active site (Figure 1.13). Two conserved DDxxD domains are located inside the cavity, 12 Å apart from each other (Figure 1.13 B). These DDxxD motifs bind to the diphosphate moieties of the substrate GPP and IPP through divalent cation cofactors such as

Mg2+. Together they form the substrate binding sites for IPP and DMAPP/GPP. Site-directed mutagenesis studies have confirmed that mutations of the Asp in these DDxxD motifs to Ala can

[37] significantly decrease the kcat values by 5-7 orders of magnitude. Additionally, a Phe residue that is 12 Å away from the first DDxxD motif has been proposed to be a key residue in determining the hydrocarbon chain length of the product. Mutation of the Phe112 residue to an

Ala transformed the wild-type avian farnesyl diphosphate synthase into synthases capable of

[38] producing geranylgeranyl diphosphate (C20). 20

Phe

Figure 1.13 Crystal structure of FPP synthase and graphic representation of the active site with FPP and IPP bound. A) The model of avian FPPS is shown using a ribbon diagram and the two active sites located in the dimer are shown with orange arrows. B) Graphic representation of the enzyme active site is shown with the black circle representing Phe112. This figure was obtained from the crystallographic report on the enzyme.[20]

1.3.2 Protein Prenyltransferases

Protein prenylation is catalyzed by three protein prenyltransferases: protein farnesyltransferase (PFTase), protein geranylgeranyltransferase type I (GGTase-I) and type II

(GGTase-II). PFTase catalyzes the transfer of a C-15 farnesyl group (from FPP) while GGTase type I and II catalyze the transfer of a C-20 geranylgeranyl group (from GGPP) onto cysteine residues located near the carboxy-terminus of a protein by forming a thioether bond (Figure 1.14).

[39],[40] While a cysteine residue is common in the Caax motif of the proteins prenylated, the aax residues vary greatly in proteins prenylated by different prenyltransferases.[39-40] In addition, all three protein prenyltransferases require Zn2+ and FTase and GGTase-II require Mg2+ for activity. 21

Zn2+ can enhance the binding of protein substrates and also interacts with the Caax thiol and

2+ [40] lowers its pKa, whereas Mg is important for pyrophosphate displacement.

SH + CaaxCOOH n S OPP CaaxCOOH n Protein protein n = 2, FPP prenyltransferase Protein n = 3, GGPP

Figure 1.14 Protein prenylation catalyzed by protein prenyltransferases.

Prenylation constitutes one of the most important post-translational modifications of a protein. In fact, it is crucial for the proper function of more than 100 human proteins including

Ras, Rab, nuclear lamins and G protein. These proteins are required in cell signal pathways, the construction of the cellular framework, and visual signal transduction.[41] More specifically, prenylation of these proteins is necessary to enable them to be anchored to the cell membrane where they fulfill their functions.[20] As an example, the oncogenic forms of the Ras family found in nearly 30 % of human cancers require post-translational farnesylation to be functional. This makes FTase an important potential target for inhibition in anticancer therapy.[42], [43]

As in the case for FPP synthase, two mechanisms, dissociative (SN1) and associative (SN2),

[44], [45], [46] have been proposed for PFTase (Figure 1.15). For the SN1 mechanism, a distinct carbocation will form upon the dissociation of pyrophosphate from FPP. This is then attacked by

[45], [46] the nucleophilic thiol of the cysteine residue. While in the SN2 mechanism, the thiol of the

22 cysteine residue is first deprotonated and coordinates with the Zn2+ cofactor upon substrate binding. This greatly increases the nucleophilicity of the cysteine side chain. The thiolate then attacks the C-1 of FPP and displaces the pyrophosphate in a concerted step. Different approaches have been employed to elucidate the mechanism employed by protein farnesyltransferase, and a lot of evidence has been uncovered supporting an SN2 mechanism with an ‘exploded’ transition state where the C-1 of FPP bears a partial positive charge while metal-bound sulfur in the protein substrate and the bridging-O of the leaving pyrophosphate bear partial negative charges.[45], [46]

Mg2+ PPi RH2C S RH2C Caax Zn COOH SN1 S Protein CaaxCOOH

Protein prenylated protein RH2C OPP FPP SH Mg2+ 2+ CaaxCOOH Mg - RH2C δ Protein RH2C O PPO SN2 PP protein substrate Zn δ + - S δ S Caax COOH Zn CaaxCOOH R=geranyl Protein Protein

transition state 'exploded'

Figure 1.15 Proposed SN1 and SN2 mechanisms for protein farnesyltransferase.

Mechanistic studies on protein farnesyltransferase isolated from yeast were first performed in the Poulter group.[44] Later on, the Fierke group carried out some mechanistic studies on a

23 mammalian protein farnesyltransferase (from rat).[46] Both groups examined the relative reactivity of fluorinated FPP analogs CH2F- and CF3-FPP (compound 15 and 16, Figure 1.16) in the protein farnesyltransferase reaction, as it was expected that the fluorine substitution would destabilize the corresponding allyl carbocations and therefore reduce the reaction rates. Very similar results were obtained from yeast and mammalian PFTase, the fluorination on FPP decreased the reaction rates significantly and the relative catalytic rates (krel) of the analogs 15 and 16 were determined to be 1.5 x 10-1 and 1.3 x 10-3 for the yeast PFTase and 1.8 x 10-2 and

2.6 x 10-4 for the mammalian PFTase, respectively.[44], [46] These results suggest that significant carbocationic character is generated in the transition state for both PFTases. However, the magnitude of the rate decreases were less dramatic than those observed for the corresponding

GPP analogs, 9 and 11, in the trans-FPP synthase reaction and those observed for the methanesulfonate derivatives, 12 and 14, in the non-enzymatic sovolysis (Table 1.1). Both of these latter reactions are believed to proceed through an SN1 mechanism. It was discovered that the effects of the fluorine substitution on PFTase reactivity more closely resemble the effects on solvolysis reactions in the presence of a potent nucleophile such as azide, which proceed via an associative mechanism with an “exploded” transition state.[44] Therefore, it was proposed that

PFTase employs an SN2 pathway through an ‘exploded’ transition state, which carries a considerable carbocationic character at the allylic position.

OPP

R=CH F F-FPP 15 2 3-CH2 R=CF 3-CF -FPP 16 3 3

Figure 1.16 Fluorinated FPP analogs 3-CH2F-FPP (15) and 3-CF3-FPP (16).

1.3.3 Aromatic Prenyltranferases

Aromatic prenyltransferases catalyze the transfer of a prenyl group onto electron-rich aromatic receptors and are responsible for the biosynthesis of a large variety of secondary metabolites in plants, fungi, and bacteria (Figure 1.17).[47], [48], [49] Aromatic prenyltransferases can be subdivided into the membrane-associated and the functionally soluble prenyltransferases.[47] Membrane-associated aromatic prenyltransferases contain the characteristic DDxxD motif responsible for Mg2+-isoprenyl pyrophosphate binding that is also present in FPP synthases. They are involved in the biosynthesis of ubiquinones and membrane lipids in prokaryotes and archaea.[50], [51]

R R aromatic + prenyltransferase OPP n n

isoprenyl aromatic = R electron donating pyrophosphate substrate group

Figure 1.17 A General reaction catalyzed by an aromatic prenyltransferase.

The functionally soluble aromatic prenyltransferases have been identified from different bacterial and fungal species and do not possess a DDxxD motif. Therefore, they lack the dependency on Mg2+ or other divalent cations for activity, suggesting different substrate binding and catalytic mechanisms from FPP synthases. Recent crystallographic studies on enzymes in this category revealed a novel protein fold with an anti-parallel β/α barrel, termed the PT-barrel

(Figure 1.18).[52] This fold consists of five repetitive αββα elements with the β-strands forming the active site in the middle of the barrel and the α-helices forming a solvent-exposed ring around the barrel.[47, 53], [54], [55], [56] For proteins with this characteristic PT-barrel, the name ABBA prenyltransferases has been suggested.[47] Ever since the establishment of this new prenyltransferases family, more and more new members have been identified and characterized, yet mechanistic studies are lacking.

Figure 1.18 PT-barrel fold found in the structure of the ABBA prenyltransferase NphB of Streptomyces. The unique ABBA fold is characterized by a central barrel consisting of ten antiparallel β strands surrounded by ten α helixes.

Based on the type of aromatic rings in their substrates, the ABBA prenyltransferase superfamily can be divided into two families, the phenol/phenazine prenyltransferases and the indole prenyltransferases.[53] The former comprises the well-studied enzymes NphB (previously known as Orf2), CloQ and NovQ that are found in bacteria genera Streptomyces and they are involved in the biosynthesis of the anti-oxidant naphterpin and the antibiotics clorobiocin and novobiocin, respectively (Figure 1.19).[54, 57], [58]

O H HO OH HO OH HO GPP H H NphB O OH OH OH OH 2 OH O naphterpin THN possible proposed precusor substrate intermediate OH OH COOH COOH H N O O O O DMAPP MeO O O O O R CloQ/NovQ OH 1 H O N CH3 OH OH R2 = clorobiocin, R1 = Cl R2

4-hydroxyphenylpyruvate novobiocin R = Me R = NH 4-HPP , 1 2 2

Figure 1.19 Aromatic prenyltransferases NphB, CloQ and NovQ in the biosynthesis of natural products. Red bonds represent the carbons derived from the prenyl donor.

The first ABBA prenyltransferase studied was CloQ and its biochemical characterization has been carried out.[57],[55] CloQ was purified from Streptomyces roseochromogenes as a soluble monomeric 35 kDa protein. The natural substrates for CloQ were identified to be 4- hydroxylphenylpyruvate and DMAPP (Figure 1.19). The true physiological substrate of NphB remains unknown, however, it has been proposed that derivatives of 1,3,6,8- tetrahydroxynaphthalene (THN) could be possible candidates. In vitro activity had been observed with a variety of dihydroxy-containing THN analogues.[54]

Indole Prenyltransferases

Indole prenyltransferases utilize indole-containing compounds as their aromatic substrates. 28

They have been studied since the 1950s yet were only recently recognized as another subfamily of the ABBA prenyltransferase superfamily.[53, 59] They account for the wide variety of prenylated indole alkaloids previously discussed in section 1.2.2 and are known to prenylate on every possible position of the indole core. These enzymes not only give rise to an astounding diversity of indole alkaloids but also enhance the bioactivities and bioavailabilities of these compounds. Therefore, investigations of indole prenyltransferases would help to understand their biosynthetic relationships and the evolution of these enzymes. As a matter of fact, there has been a significant increase of the number of the identified indole prenyltransferases in the past decade.[59],[60],[61]

Recently, indole prenyltransferases that can install a prenyl moiety onto each of the seven positions of the indole ring have been biochemically characterized (Figure 1.20). For instance,

CymD from the marine actinobacterium Salinispora arenicola and FtmPT2 from Aspergillus fumigatus can catalyze prenylation at the N-1 position on the indole ring of tryptophan or its derivative;[62],[63] FtmPT1 and FgaPT1 from Aspergillus fumigatus catalyze C-2 prenylation during the biosynthesis of fumitremorgin B and fumigaclavine C, respectively;[63],[64] CdpNPT from Aspergillus fumigatus and AnaPT from Neosartorya fischeri can catalyze C-3 prenylation;[65],[66] FgaPT2 from Aspergillus fumigatus, MaPT from Malbranchea aurentiaca and CpaD from Aspergillus sp. all catalyze C-4 prenylation;[9],[67],[68] and lastly, 5-DMATS from

Aspergillus clavatus, IptA from Streptomyces sp. SN-593 and 7-DMATS from Aspergillus fumigatus, catalyze the prenylation at the C-5, C-6 and C-7 positions, respectively.[69],[70],[71]

R H3C 4 'normal' 'reverse' 5 3 H3C H3C prenylation 2 indole substrate CH3 prenylation 6 N 7 H 1 Prenyltransfeases, DMAPP

O N1 C2 C3 COOH N O NH NH2 HN H N O N N N H H H O

CymD FtmPT1 CdpNPT

C4 C6 C7 C5 COOH COOH COOH NH COOH 2 NH2 NH NH 2 N N 2 H H N N H H 4-DMATs IptA 7-DMATs 5-DMATs

Figure 1.20 Indole prenyltransferases can prenylate every position of the indole core.

The indole prenyltransferases identified so far have been shown to share sequence similarities ranging from 25 % to 70 %.[13],[59] The proteins were all proven to be soluble and share the characteristic anti-parallel β/α barrel. More importantly, unlike other membrane-bound aromatic prenyltransferases, they do not require divalent metal ions for catalytic activity.

Most indole prenyltransferases utilize DMAPP as their prenyl donor while their indole substrates are typically tryptophan or tryptophan-containing cyclic dipeptides. The Li group has carried out a lot of substrate specificity studies on different indole prenyltransferases.[72],[73],[59],[74],[75] They have demonstrated that indole prenyltransferases have very broad substrate specificity towards their aromatic substrates. Therefore, they have proposed 30 that indole prenyltransferases could be used in the chemo-enzymatic synthesis of novel indole alkaloids.

Two particular indole prenyltransferases, 4-dimethylallyltryptophan synthase and the indole

N-prenyltransferase CymD, will be studied in this thesis, details of which will be discussed in the following sections.

4-Dimethylallyltryptophan Synthase

4-Dimethylallyltryptophan synthase (DMATS) catalyzes an electrophilic aromatic substitution in which the prenyl group from dimethylallyl pyrophosphate (DMAPP) is transferred onto the C-4 position of tryptophan in a normal fashion (Figure 1.21). DMAT synthase has been identified in different fungi including Claviceps purpurea, Aspergillus fumigatus and Malbranchea aurantiaca. It is the enzyme involved in the first committed step in the biosynthesis of the ergot alkaloids such as lysergic acid derivatives and fumigaclavine C

(Figure 1.5). Due to the important biological and pharmacological properties of the ergot alkaloids and the unusual regioselectivity displayed in the reaction, this enzyme has attracted a great deal of interest from researchers in the past decades.

COO COO

NH3 NH3 DMATS O O O P O P O N N O O PPi H H L-tryptophan DMAPP 4-DMAT

Figure 1.21 Prenylation catalyzed by 4-DMAT synthase.

In the early stages of research on this enzyme, the Floss group isolated 4-DMAT synthase from the mycelia of Claviceps sp., SD 58, and purified it to apparent homogeneity. Subsequently, they examined the stereochemistry of the reaction catalyzed by this enzyme with the use of isotopically labeled substrates.[17],[18] An inversion of stereochemistry at C-1 of DMAPP was observed when tritium labeled substrates (R)-[1-3H]-DMAPP (17) and (S)-[1-3H]-DMAPP (18) were utilized in the reaction (Figure 1.22). Additionally, the geometry of the allylic double bond

2 was maintained when deuterium labeled substrate (Z)-[methyl- H3]-DMAPP (19) was used in the reaction.

COOH H COOH DMAT T NH2 T + synthase NH H 2 OPP N H 3 N (R)-[1- H]-DMAPP H 17 COOH DMAT NH2 T COOH H + synthase H NH T N 2 OPP H 3 (S)-[1- H]-DMAPP N 18 H COOH CD3 CD3 DMAT NH COOH + 2 synthase

OPP N NH2 H Z 2H ( )-[methyl- 3]-DMAPP N 19 H

Figure 1.22 Investigation of stereochemistry in the reaction catalyzed by DMAT synthase using isotopically labeled substrates 17, 18 and 19.

Even though an inversion of the stereochemistry at C-1 of DMAPP was observed, the

Floss group felt that a dissociative mechanism involving a carbocation intermediate could not be precluded. Therefore, they proposed both dissociative and associative mechanisms for the

DMATS-catalyzed reaction, namely SN1 and SN2 mechanisms (Figure 1.23). In the SN1 mechanism, the reaction starts with the dissociation of the C-1/OPP bond of DMAPP into an ion pair involving an allylic carbocation stabilized by the enzyme-bound pyrophosphate. Then a direct attack from the C-4 of tryptophan onto the face opposite the pyrophosphate ensures the inversion at C-1 of DMAPP as well as the retention of double bond geometry. Lastly, a deprotonation at C-4 of the resulting arenium intermediate yields 4-DMAT. In the SN2 mechanism, the same arenium intermediate forms through an associative ‘exploded’ transition 33 state where both the leaving of the pyrophosphate group and the attack from C-4 of tryptophan happen in a synchronized manner. In this case, the inversion of stereochemistry at C-1 is expected as a result of the SN2 transition state.

COOH PPi NH2

S 1 N mechanism N COOH H BH: COOH COOH NH2 H NH NH2 2 N OPP H N N O H H DMAPP OH L-Tryptophan P arenium intermediate 4-DMAT O O P O δ SN2 mechanism O O COOH C δ NH H 2

N δ H SN2 'exploded' transition state

Figure 1.23 SN1 and SN2 mechanisms proposed for 4-DMAT synthase.

It was not obvious whether an SN1 or SN2 mechanism was involved in the reaction catalyzed by DMAT synthase as both mechanisms have precedence with different types of prenyltranferases. For instance, isoprenyl pyrophosphate synthases such as farnesyl pyrophosphate synthase (FPPs) are generally thought to employ SN1 mechanisms due to the weak nucleophilicity of the alkene nucleophile. On the other hand, for protein farnesyltransferase catalyzed reactions, the prenyl acceptor is a thiolate which has significantly stronger 34 nucleophilicity, and they are believed to utilize a SN2 mechanism. In the case of DMAT synthase, the nucleophilicity of the indole lies between that of an alkene and a thiolate, therefore both SN1 and SN2 mechanisms might be operational. Further investigations were needed to address this interesting problem in enzymology.

The Floss group first established the method of isolating this enzyme from a Claviceps sp. strain.[17] Later on, this method was further modified by the Rilling and Poulter groups.[8],[76]

Unfortunately, all these methods required multiple chromatographic and precipitation steps resulting in a low yield of the enzyme. In order to further probe the catalytic mechanism of

DMAT synthase, it was necessary to establish an efficient method for the production of a large quantity of pure recombinant enzyme.

Despite these constraints, a large number of mechanistic studies have been conducted on the reaction catalyzed by DMAT synthase. The Poulter group has investigated the mechanism by studying the substituent effects of electron-withdrawing and electron-donating groups on both tryptophan and DMAPP.[77] It was observed that upon fluorination of DMAPP, the reaction rate decreased by over 100-fold. This is consistent with the notion that electron-withdrawing fluorination can greatly destabilize the allylic carbocation formed in the course of catalysis. In addition, they measured the reaction rates using tryptophan and its 7-substituted analogs and constructed a Hammett plot which gave a good linear correlation with a negative slope. This is consistent with an electrophilic substitution yet suggests that there is less development of positive charge at the transition state than that observed in non-enzymatic electrophilic substitution reactions. Ultimately, this study did not clearly differentiate between the two 35 mechanisms since one might expect similar results with either one.

In 1995, by using the partial amino acid sequence of the purified DMAT synthase, Schardl and coworkers cloned the first DMAT synthase gene, dmaW, and its cDNA from Claviceps purpurea, and overexpressed it in the yeast Saccharomyces cerevisiae.[19] No effort was made to purify the enzyme to homogeneity; instead the crude yeast extract was directly used in enzyme assays. Due to the low enzyme activity obtained in this work, product identification relied solely on HPLC retention time, UV spectra and mass spectrometry, and no NMR data of the product was acquired.

In 2005, the Li group was able to identify another putative DMAT synthase gene, fgaPT2, from the available genome of Aspergillus fumigatus, then clone and express it in Saccharomyces cerevisiae.[9],[72] More importantly, they successfully purified the enzyme to near homogeneity for the first time and characterized the enzyme biochemically. FgaPT2 was found to be a soluble, dimeric protein with a subunit size of 52 kDa that lacked the putative prenylpyrophosphate binding site (N/D)DXXD motif commonly found in other prenyltransferases. This suggested it might operate with a different mechanism. Tryptophan and several methylated derivatives were treated with FgaPT2 and products from these reactions were isolated by HPLC and then studied using NMR spectroscopy and MS. With these studies, the enzyme product was confirmed to be

4-DMAT and methylated DMATs were also observed when methylated Trp was used, at a lower yield comparied to that of native substrate tryptophan.[72]

Furthermore, in 2008, a third DMAT synthase gene, MaPT, was identified from the genome of Malbranchea aurentiaca.[67] The gene was successfully overexpressed in E. coli, which soon 36 became a standard procedure due to the high yield of enzyme production.

These newly discovered DMAT synthases were all found to lack the presumed prenylpyrophosphate binding DD(xx)D motif similar to other bacterial prenyltransferases such as NphB, CloQ and NovQ. Secondly, these prenyltransferases do not require a divalent metal cation like Mg2+, Mn2+, or Zn2+ in order to function. Hence, they were believed to belong to the same emerging new class of aromatic prenyltransferases. Therefore, solving the mechanism of one enzyme could possibly shine insights onto other members of the family.

More recently, research studies carried out by Luk and Tanner aimed at discerning between the dissociative and associative mechanism were reported.[78] In order to probe the transient formation of the allylic carbocation, the authors employed positional isotope exchange (PIX) experiments. An 18O isotopically labeled DMAPP analog [1-18O]-DMAPP (20) was synthesized, which bears an 18O label at the bridging position between allyl group and pyrophosphate group.

When compound 20 was treated with FgaPT2, it was found that the 18O label partially scrambled onto the non-bridging position of the recovered DMAPP (see red atoms in Figure 1.24) while control experiments showed that no scrambling occurred in the absence of tryptophan.[78] This suggests that the enzyme catalyzes a reversible cleavage of the C-O bond in DMAPP. However, the observed PIX could be explained by either the SN1 or SN2 mechanism. In an SN1 mechanism, an allylic carbocation /pyrophosphate ion pair will be generated and its lifetime should be long enough for the rotation of the O-P bond in pyrophosphate to occur. In an SN2 mechanism, PIX could be possible if a subsequent step, such as deprotonation of the arenium intermediate, were rate-determining, and the SN2 step was reversible. 37

18 allylic cation/PPi [1- O]-DMAPP allylic cation/PPi DMAPP ion pair 20 ion pair 18O-non-bridging O O O O PO P O PO P O rotation PO P O PO P O O O + O O + + + L-Trp L-Trp L-Trp L-Trp

Figure 1.24 Positional isotope exchange observed when using [1-18O]-DMAPP (20) together with L-Trp in DMAT synthase-catalyzed reaction. Red circles denote for 18O label.

To address these possibilities, the authors investigated the nature of the rate-determining step using kinetic isotope effect (KIE) experiments.[78] Two deuterium-labeled substrate analogs

[1, 1-2H]-DMAPP and [4-2H]-L-tryptophan were prepared (compound 21 and 22 in Figure 1.25).

With [1, 1-2H]-DMAPP (21), a normal secondary KIE was observed, indicating cleavage of the

C-O bond of DMAPP is a partially rate-determining step of catalysis and that the transition state has considerable carbocation character. With [4-2H]-L-tryptophan (22), an inverse secondary

KIE was observed, which suggests C-C bond formation is also partially rate-determining. Since both the C-O bond cleavage and C-C bond formation are rate-determining steps, the occurrence of PIX could only be explained by a reversible formation of an allylic carbocation/pyrophosphate

2 ion pair in an SN1 mechanism. The absence of a primary KIE with [4- H]-L-tryptophan (22) also indicates that re-aromatization of the arenium intermediate is not the rate-determining step of catalysis. In addition, when a PIX experiment was carried out using an unreactive substrate 6- fluorotryptophan (23 in Figure 1.25), complete scrambling of the isotopic label was observed indicating the reversible formation of an allylic carbocation.

COOH COOH D NH D D 2 NH2 OPP N N H F H 2 2 [1,1- H]-DMAPP [4- H]-L-tryptophan D,L-6-fluorotryptophan 21 22 23

Figure 1.25 Deuterium-labeled substrate analogs [1, 1-2H]-DMAPP (21) and [4-2H]-L- tryptophan (22) used in KIE experiments and D,L-6-fluorotryptophan (23).

The authors then proposed that their results supported a dissociative (SN1) mechanism

(Figure 1.26).[78] The normal secondary KIE observed with [1, 1-2H]-DMAPP (21) and the inverse secondary KIE observed with [4-2H]-L-tryptophan (22) imply that both the dissociation step and the C-C formation step should be partially rate-determining (see the reaction coordinate in Figure 1.26). The last deprotonation step, however, is not rate-determining based on the absence of a primary KIE with [4-2H]-L-tryptophan (22).

B DMAPP PPi COOH COOH COOH PPO COOH NH2 H NH NH 2 2 NH2

+ N N - H DMAT N N H H H H L-Trp arenium intermediate E

Figure 1.26 The dissociative mechanism for DMAT reaction and the hypothesized reaction coordinate.

The structure of DMAT synthase in a complex with tryptophan and the unreactive substrate analog, dimethylallyl-S-thiolopyrophosphate (DMSPP, Figure 1.27), has also been recently reported by Stehle et. al. (Figure 1.27).[56] It was confirmed that the structure of FgaPT2 shares the same topology as other cation-independent prenyltransferases, such as NphB, and they all share the unique ααββ fold. It was also noticed that the diphosphate-leaving group of DMAPP was tightly bound with a large number of amino acid residues, possibly leading to the decrease of the energy barrier for carbocation formation from DMAPP. In addition, the authors proposed that the formed allylic carbocation could be stabilized by cation-π interactions with Y345 on one face and the indole of the substrate on the other (Figure 1.27, top right).[56]

O O O O P P S O O DMSPP

Figure 1.27 The 3D architecture of DMAT synthase (top left), the active center of DMAT 40 synthase (top right), and DMAPP analog DMSPP (bottom). Crystal structures were taken from PDB 3I4X.[56]

Furthermore, this structure of the substrate-bound enzyme implicates two active site residues as playing key roles in catalysis. The carboxylate group of Glu89 was hydrogen-bonded to the indole N-H, thus it could enhance the nucleophilicity of the aromatic ring and promote its attack onto the dimethylallyl carbocation (Figure 1.28). This could occur either via deprotonation or by formation of a charged hydrogen bond. On the other hand, the amine group of Lys174 was proposed to deprotonate the arenium intermediate leading to the re-aromatization of DMAT.

These hypotheses still await experimental confirmation.

Lys174-NH2 Lys174-NH2 Lys174-NH3 DMAPP NH PPi 2 NH2 NH2 NH2 PPO COOH H COOH COOH COOH

+ N N - H DMAT N N H H arenium H H L-Trp O O O intermediate O O O Glu89 Glu89 Glu89 Figure 1.28 Proposed involvement of two residues Lys174 and Glu89 in the DMATS reaction.

Cyclomarin/Cyclomarazine N-Prenyltransferase CymD.

Cyclomarin/cyclomarazine prenyltransferase CymD is a prenylating enzyme found in the marine actinobacterium Salinispora arenicola CNS-205 by Moore et. al..[62],[79] CymD catalyzes 41 a reverse prenylation of the N-1 position of tryptophan to yield N-(1,1-dimethyl-1- allyl)tryptophan (N-DMAT, Figure 1.29). This compound then serves as a substrate for a heptamodular nonribosomal peptide synthetase CymA in S. arenicola leading to the antibacterial cyclomarazine A and the anti-inflammatory cyclomarin A.

O O HO N O Me H HN O COOH COOH O PPO NH O NH O NH NH NH HN O 2 CymD 2 CymA O O HN N N N Me H OH N O N-DMAT N H

cyclomarazine A O A cyclomarin Figure 1.29 Cyclomarin/cyclomarazine N-prenyltransferase CymD catalyze the first step in the biosynthesis of cylcomarin A and cyclomarazine A and produces N-DMAT.

In 2008, the Moore group first proprosed that the cymD gene from S. arenicola encodes for a prenyltranferase as the corresponding protein sequence showed 24% identity to the prenyltransferase LtxC.[79],[80] When the authors inactivated the cymD gene in a cymD- knockout mutant, the production of the natural cyclomarins and cyclomarazines was eliminated in S. arenicola. Taken together with the finding of a surprising low production of desprenyl cyclomarin, they proposed that the prenylation should occur in the early stages of cylcomarin biosynthesis.[79] 42

Later in 2010, the Moore group established the biological function of CymD as an indole

N-prenyltransferase.[62] The authors synthesized N-DMAT chemically and confirmed that its structure is identical to that of the pathway intermediate produced by the enzyme. Additionally, the authors treated the cymD- knockout mutant with chemically synthesized N-DMAT and successfully restored the production of cyclomarin and cylcomarazine to wild-type levels. These observations confirmed CymD as a bacterial N-dimethylallyltryptophan synthase.

The CymD reaction is of interest from a mechanistic point of view because it involves the alkylation of an indole nitrogen that is very non-nucleophilic due to the participation of its lone pair in the aromaticity of the indole ring. Also this enzyme shares sequence similarity with the

DMAT synthase FgaPT2. Both enzymes are thought to possess a α/β barrel fold and catalyze cation-independent prenyl transfers. Yet no mechanistic studies had been performed on this enzyme. We therefore questioned whether this enzyme utilizes a similar mechanism as the

DMAT synthase FgaPT2 does.

Thesis Goals

The first goal of this thesis is to further probe the role of some key residues found in the enzyme active site of the DMAT synthase FgaPT2 and hopefully help to understand the mechanism of other prenyltransferases from the same family. Mutants of FgaPT2 were prepared and examined with a series of mechanistic experiments. One of the FgaPT2 mutants produced an unusual product, which eventually led us to propose a new mechanism for FgaPT2 involving 43 a Cope rearrangement. A more detailed discussion of the mechanistic studies on FgaPT2 will be covered in Chapter 2.

In the light of the new rearrangement mechanism we proposed for FgaPT2, we considered the possibility of other indole prenyltransferases sharing similar mechanisms. Mechanistic studies were carried out on another indole prenyltransferase CymD using a series of techniques including fluorinated substrate studies, PIX experiments and solvent kinetic isotopic studies. All of these experiments eventually argued against the possibility of a rearrangement mechanism and led us to propose a novel mechanism for the prenyltransferase CymD. Detailed discussion of the mechanistic studies on CymD will be included in Chapter 3.

Chapter 2. Mechanistic Studies on 4-Dimethylallyltryptophan Synthase

A version of this chapter has been published and some of the experimental results are reproduced with permission from: Qian, Q.*; Luk, L. Y. P.*; Tanner, M. E. J. Am. Chem. Soc. 2011, 32, 12342-12345 (© 2011 American Chemical Society). * co-first authorship

As stated in section 1.5, 4-dimethylallyltryptophan (DMAT) synthase belongs to the family of indole prenyltransferases. Due to the interesting reaction catalyzed by this enzyme, a substantial amount of research has been devoted to understanding its mechanism. In the past decade, from the two postulated mechanisms, evidence favored the SN1 mechanism largely due to the observation of reversible prenyl-phosphate bond breakage by positional isotope exchange

(PIX) and kinetic isotope effect (KIE) experiments.

At the onset of this study, we aimed to investigate the catalytic roles of the active site residues using site-directed mutagenesis and PIX experiments. In the crystallographic structure published by Metzger et al., two key active site residues Glu89 and Lys174 were proposed to play important roles in the catalytic mechanism and mutation of these two residues had significant impact on the catalytic activity of the enzyme.[56] In light of previous work in our group, we envisioned that combining the PIX experiments and KIE measurements with site- directed mutagenesis could help us to further probe the catalytic mechanism of this enzyme.

In the structure solved by Metzger et al., the carboxylate group of Glu89 was observed to form a hydrogen-bond to the N-H of the indole ring.[56] Thus, it is likely that it assists catalysis either by deprotonation of the indole as it attacks the carbocation intermediate or by forming a charged hydrogen bond with the resulting iminium ion.[56] If Glu89 is mutated to a non-ionizable hydrogen-bond acceptor (E89Q mutation), or a non-polar residue (E89A mutation), an increase of the energy barrier for the nucleophilic attack step would be expected. This should make the nucleophilic attack step cleanly rate-determining, which, in turn, can increase the lifetime of the carbocation intermediate (Figure 2.1). For the mutation of the glutamate to a non-polar alanine, 46 the enzyme might lose activity completely as was reported in the crystallographic studies, while the mutation to a weaker hydrogen-bond acceptor glutamine might still retain some activity.

Additionally, if PIX experiments are carried out with these two mutants, the extent of PIX, (or the partitioning ratio (υPIX/υrxn), to be precise), should increase substantially as compared to that of the wild-type enzyme-catalyzed reaction. This would be expected because the ionization step that forms the carbocation would be more freely reversible in the mutant reactions due to the increased energy barrier in the subsequent step. Furthermore, the fate of the dimethylallyl carbocation might change due to the extended lifetime in the mutant reactions; for example, the carbocation could be attacked by water to form either a primary or tertiary alcohol, instead of the expected product. As a result, we will also examine the products from the mutant reactions by 1H-NMR spectroscopy.

K174Q PPi K174A DMAPP NH2 NH2 NH2 PPO COOH H COOH COOH

NH2 + 4 COOH N N - H N 3 H arenium H DMAT H 2 intermediate N 1 E89Q H L-Trp E89A

Figure 2.1 Mutation of E89 and K174 increases the energy barriers of the nucleophilic attack and rearomatization steps. Red dotted line represents the effect of E89 mutation and the blue dotted line represents the effect of K174 mutation.

From the position of Lys174 in the crystal structure, this residue was believed to serve as a base to deprotonate the arenium intermediate and reconstitute the aromaticity. If Lys174 is mutated to the less basic residue glutamine (K174Q mutation) or the truncated residue alanine

(K174A mutation), the energy barrier for the deprotonation step should increase substantially and likely this will promote it to be a cleanly rate-determining step. Thus the probability that the previous steps would be reversible should also increase (Figure 2.1). In previous studies, Metzger et al. reported that the mutant K174Q maintained about 40% activity. Considering the side chain of glutamine is 1 Å shorter than that of lysine, they proposed the activity might be attributed to a polarized water molecule that can deprotonate the arenium intermediate.[56] Metzger et al. also prepared the mutant K174E, and found that it lost about 98% of the wild type activity. This is reasonable because the introduction of a negative charge perturbs the active site more dramatically. A less perturbing K174A mutation has not yet been examined and its activity remains to be studied. As in the case for the E89 mutations, the potential for reversible carbocation formation should increase due to the increased energy barrier for the rearomatization step, thus the partitioning ratio (vPIX/vrxn) should increase accordingly for both mutants K174Q and K174A. 1H NMR spectroscopy will also be used to monitor the mutant reactions and determine the product compositions.

Preparation of Dimethylallyltryptophan Synthase and Mutants

2.1.1 Preparation of Mutant Genes by Site-Directed Mutagenesis

The mutant genes encoding for E89A, E89Q, K174A and K174Q were prepared by Louis

Luk using the QuikChange® Site-Directed Mutagenesis kit. The DNA sequences of the DMATS mutant plasmids were confirmed by sequencing the entire gene.

2.1.2 Overexpression of 4-Dimethylallyltryptophan Synthase Mutants

When I joined the project, I started with the production of the mutants together with Louis

Luk. Production of the mutants (E89A, E89Q, K174A and K174Q) was achieved using a modification of the protocol previously described.[72] After using the plasmid fgaPT2/pET28a to transform E. coli Rosetta(DE3) pLysS cells, cell cultures were grown in Terrific Broth (TB) medium until an OD600 of 0.6 was reached. Cells were then induced for overexpression by the addition of isopropyl-1-thio-β-D-galactopyranoside (IPTG). After an additional 16 h growth, cells were harvested and lysed in a French pressure cell, and the crude lysate was clarified by centrifugation before application onto a column of immobilized Ni2+ ion affinity resin. The enzyme was eluted from the column using a step gradient of imidazole buffer. Initially, the pooled enzyme fractions were passed through a size exclusion column in order to remove imidazole, yet it was discovered later that imidazole was still present in the enzyme solution 49 when the corresponding signals appeared in 1H NMR spectra. Significant loss of the enzyme activity was observed when attempts were made to remove imidazole by centrifugal ultrafiltration, therefore, dialysis was chosen instead. In the end, typically ~10 mg of enzyme was purified from 1 L of culture, which was sufficient for further characterization.

Steady-State Kinetic Characterization of DMATS Mutants

In past studies, different approaches had been utilized for analyzing the kinetics of DMAT synthase.[8-9, 77] The Poulter group used [1-3H]-DMAPP and measured the incorporation of radioactivity into the product DMAT.[8, 77] The Li group employed HPLC to determine the reaction progress of stopped enzyme reactions by using an authentic DMAT sample as the standard.[9] In our group, Luk used a commercially available continuous coupled pyrophosphate assay to determine the kinetic activity of DMAT synthase as pyrophosphate was one of the products in the enzyme reaction.[78] This coupled assay proved to greatly facilitate the kinetic studies on this enzyme because no separation of the products, or quenching of the reaction at different time points, was required. Furthermore, this assay exempts the use of radio-labeled substrates or authentic samples of the product if substrate analogs were to be examined.

In the continuous coupled assay, the pyrophosphate (PPi) generated in the reactions was measured indirectly by first hydrolyzing it into two equivalents of phosphate (Pi) using inorganic pyrophosphatase as one of the coupling enzymes (Figure 2.2). Subsequently, phosphate is then consumed by another coupling enzyme purine nucleoside phosphorylase (PNPase), that can 50 convert 2-amino-6-mercapto-7-methylpurine ribonucleoside (MESG) to ribose 1-phosphate and

2-amino-6-mercapto-7-methylpurine. The resulting 2-amino-6-mercapto-7-methylpurine has a distinctive absorption maximum at 360 nm, thus, the accompanying absorption change allows quantitation of inorganic pyrophosphate produced in the enzyme reaction.

COOH DMAT COOH NH2 synthase + NH2 + PPi N OPP H N L-tryptophan DMAPP H inorganic DMAT pyrophosphatase

2 Pi S S + HO PNPase N N O N NH + O N N O P O N NH2 N NH2 OH OH HO O O 2-amino-6-mercapto-7-methylpurine ribose-1-phosphate Abs = 360 nm OH OH MESG max

Figure 2.2 Illustration of the continuous coupled pyrophosphate assay.

Using this coupled pyrophosphate assay, Luk was able to determine the kinetic constant kcat

-1 to be 0.27 ± 0.05 s at saturating levels of both DMAPP and L-tryptophan. The KM value for

DMAPP was determined to be 11±4 μM by varying the DMAPP concentration and holding the

L-tryptophan concentration constant at a saturating level. Similarly, KM for L-tryptophan was determined to be 10±4 μM by varying the L-tryptophan concentration and holding the DMAPP

51 concentration constant at a saturating level.[78] The same assay was used in the kinetic studies of all four DMAT synthase mutants as well as wild-type DMAT synthase, for the purpose of comparison. The hyperbolic curves with L-Trp as the variable substrate for WT, E89Q, K174A and K174AQ are shown in Figure 2.3. Kinetic parameters were determined from the fit of the initial velocities to the Michaelis-Menten equation (Table 2.1), and The reported error ranges were calculated from the non-linear least square fitting on results of kinetic trials carried out on the same day, with enzymes and substrates from the same batch. However, systematic error can occur due to factors such as the use of different batches of enzymes and substrates, errors in concentration measurements, as well as temperature fluctuation. These typically lead to ~20% error when measurements were performed on different days. The kcat and KM value for E89A could not be determined due to the low activity of the mutant and an upper value of kcat was estimated. Since all the mutations would likely only change the substrate binding environment for L-tryptophan, KM values for DMAPP were not measured.

0 10 20 30

) A 2 2 -1 M s -8 10 (

1 1 Rate Rate

0 10 20 30 [L-tryptophan] (µM)

C D

Figure 2.3 Kinetics plots for the wild-type and mutant DMAT synthases. Traces A, B, C, and D are for WT, E89Q, K174Q and K174A, respectively.

Table 2.1 Kinetic constants of wild-type and mutant enzymes

Enzyme WT E89Q E89A K174Q K174A

-1 a kcat (s ) 0.47 ± 0.02 0.015 ± 0.001 <0.001 0.39 ± 0.01 0.023±0.001

KM, Trp (μM) 6.7 ± 0.8 6.8 ± 0.5 N/A 8.6 ± 0.5 8.8 ± 1.0 a. The errors were calculated from the fitted curves.

The measured kinetic constants for wild-type DMAT synthase were consistent with values 53 reported by Luk.[78] The mutation of Glu89 to either Gln or Ala resulted a 19-fold or > 400-fold reduction in the value of kcat, respectively, which we interpret as being consistent with the notion that this residue plays a key role in increasing the nucleophilicity of the indole ring. On the other hand, the mutation of Lys174 to Gln resulted in only a 17% reduction of kcat (consistent with

[56] previous studies) while mutation of this residue to Ala resulted in a 20-fold reduction of kcat.

This is somewhat surprising since the structural analysis clearly indicated that Lys174 is in a position to act as a Brønsted base that deprotonates the arenium ion intermediate. We postulate that the Gln residue at this position may still be able to facilitate deprotonation since the proton on the arenium ion intermediate is quite acidic. However, an Ala residue would lack any potential to act as a base thus the rate is more considerably reduced. In addition, the KM values of L- tryptophan for all three active mutants did not change significantly, which suggests that the introduction of these mutations did not dramatically alter the substrate binding environment for

L-tryptophan. Thus the decreases of the kcat likely resulted from the poorer transition state stabilization as opposed to ground state destabilization.

Positional Isotope Exchange Experiments on DMAT Synthase Mutants

2.3.1 Introduction to Positional Isotope Exchange

The positional isotope exchange (PIX) experiment is a commonly used technique that monitors the isotopic scrambling to provide evidence for a reversible bond breakage event in a 54 mechanism.[81] This technique often involves a substrate containing a phosphate group with an

18O label at a position bridging to a leaving group. In cases where PIX is used to monitor

18 cleavage of carbon-phosphate ester bonds, when the R1- O bond breaks, the three P-O bonds may become torsionally equivalent via rotation around the P-O-R2 bond (Figure 2.4). If the breakage of this bond is reversible and the rotation occurs at a higher rate than the bond reformation, the 18O label will be randomly scrambled among the bridging and non-bridging positions in the recovered starting material. The observation of this isotopic scrambling can provide important insights into the behavior of enzyme-bound intermediates, which makes PIX very useful in mechanistic studies.

18 18 bridging O non-bridging O O O O O P OR P OR P OR P R 2 R 2 2 OR 1 O 1 O R1 O R1 O 2 O O O O

R = allyl 1 R = 2 phosphate Figure 2.4 A schematic representation for the isotopic scrambling in PIX experiments with 18 a phosphate ester. The solid red circles denote for the O label and R1 forms a carbocation after the C-O bond breakage.

2.3.2 Synthesis of [1-18O]-DMAPP for PIX experiments

The synthesis of [1-18O]-DMAPP (20) was completed using a published procedure (Figure

2.5).[78] Triethylorthoacetate was refluxed with 95% enriched 18O water and p-toluenesulfonic

18 acid, and the resulting ester was then hydrolyzed to produce the sodium [1,1- O2]-acetate (24).

Then the 18O-labeled acetate was heated with dimethylallyl chloride and a small amount of

18 triethylbenzylammonium chloride (Bn(NEt3)Cl) to produce the dimethylallyl [1, 1- O2]-acetate

(25). Deacetylation of the ester in anhydrous methoxide gave the [1-18O]-dimethylallyl alcohol

(26). Finally, 26 was incubated with triethylammonium phosphate (TEAP) in the presence of

18 trichloroacetonitrile (CCl3CN) to yield [1- O]-DMAPP (20). This method gave a reasonable yield (overall ~8%) and the 18O-incorporation rate was determined to be around 90% by negative

ESI-MS spectrometry (m/z = 247 and 245 for labeled and non-labeled DMAPP, respectively).

O OEt 1. H2O / pTosOH BnN(Et3)Cl + OEt 2. H O /NaOCH ONa OEt 2 3 Cl triethylorthoacetate 24 dimethylallyl chloride

NaOCH3 TEAP

MeOH CCl3CN O OH OPP

O 25 26 20

Figure 2.5 Synthesis of [1-18O]-DMAPP (20). Solid red circle represents an 18O atom.

2.3.3 PIX Experiments on DMAT Synthase Mutants

Following a procedure previously described by Luk et al.,[78] we examined the effect of the mutations on the extent of isotope scrambling. In the PIX experiments, a mixed sample of 40% 56

DMAPP and 60% [1-18O]-DMAPP was incubated with 0.8 equivalents L-tryptophan in the presence of DMAT synthase for extended periods of time in buffer containing EDTA. The reaction was monitored using 31P NMR spectroscopy, which reports on both the extent of reaction (amount of DMAPP vs. pyrophosphate) as well as the extent of PIX (appearance of a non-bridging α-phosphorus signal at upper field). When the reaction had stopped, usually after

24 h of incubation, the reaction was quenched by removing the enzyme by centrifugation through a membrane filter (10 kDa MWCO). The crude reaction mixture was then lyophilized before acquiring the final 31P NMR spectrum.

Since the changes in phosphorus chemical shifts that resulted from isotopic scrambling are small in magnitude (~ 0.01 ppm), some precautions were necessary in order to obtain the high resolution 31P NMR spectra needed to resolve the signals. The enzyme reactions were carried out in the presence of EDTA. This was done to sequester divalent cations Mg2+ and Ca2+, which were known to have broadening effects on the 31P NMR spectra. A small amount of specialized chelating resin (Chelex-100) was also added to the crude reaction mixture to remove residual amounts of divalent cations prior to the acquisition. This resin is particularly convenient since it could be present in the NMR tube and did not cause any noticeable interference during data acquisition. The acquisition parameters of the spectrometer were also carefully adjusted in order to increase the resolution of the 31P NMR spectra. The sweep width was set to 20 ppm (centered at -5 ppm) and the acquisition time was increased to 27 s. In addition, mathematical processing of the acquired FID using apodization with exponential and Gaussian functions was used to optimize the resolution of the 31P NMR spectra. 57

In the DMAT synthase catalyzed reaction, once the dimethylallyl cation/ pyrophosphate ion pair is formed in the enzyme active site, it can either partition forward to form the products

DMAT and pyrophosphate, or backward to regenerate DMAPP during which isotopic scrambling can occur. Therefore, the partitioning ratio (υPIX/υrxn), which is the ratio of the extent of scrambling to the extent of reaction (%PIX/%rxn under initial velocity conditions), is a useful indicator of any changes in energy barrier of the steps that occur subsequent to the dimethylallyl carbocation formation step.

Both the scrambling extent (%PIX) and the reaction extent (%rxn) can be determined by the peak integrations in the 31P NMR spectra. When the isotopic scrambling proceeds to completion,

66.7% of the recovered DMAPP should bear a non-bridging 18O label. Thus, the value of %PIX was determined by dividing the integration of the 31P signal due to the α-phosphorus of DMAPP bearing a non-bridging 18O label by the combined integration of signals due to the α-phosphorus of DMAPP bearing either a bridging or a non-bridging 18O label and then dividing this number by 66.7%. This value was found to be 0.22 in the wild-type PIX experiment.[78]

The value of %rxn was determined by dividing the integration of the total pyrophosphate product signals by the integration of the total pyrophosphate and DMAPP signals. This value was found to be 0.57 in the wild-type PIX experiment.[78] Hence, the partitioning ratio for the wild-type enzyme (υPIX/υrxn) = (%PIX/%rxn) = 0.22/0.57 = 0.39.

2.3.3.1 PIX Reactions Catalyzed by DMAT Synthase Mutants E89A and E89Q

The PIX experiments were first carried out on DMAT synthase mutants E89A and E89Q.

31P NMR spectra were acquired both before and after the enzyme reactions. For the DMAPP and

[1-18O]-DMAPP mixture, the corresponding α-phosphorus signals are two sets of doublets centered at δ -9.28 ppm and δ -9.30 ppm, respectively (Figure 2.6, A). After 24 h incubation, a set of new signals corresponding to unlabeled and 18O-labeled pyrophosphate appeared at δ -

5.94 ppm and δ -5.96 ppm, respectively. The integrations of the pyrophosphate signals were used to calculate the percentage of DMAPP that had been converted to pyrophosphate (%rxn), this provides an estimate for the reaction conversion rate (υrxn). In the reaction catalyzed by E89Q, about 19% of the DMAPP had been consumed, while in the E89A catalyzed reaction about 8% of the DMAPP had been consumed. The observation of pyrophosphate formation in the E89A catalyzed reaction was quite surprising as the mutant E89A was deemed essentially inactive from the kinetic studies. It seems that this mutant still retains a small amount of activity and that the reactions can still occur upon extended incubations. Alternatively, it is possible that a water molecule could occasionally slip into the enlarged active site of the mutant enzyme to promote a hydrolysis reaction of DMAPP.

O O O O O P O P O O P O P O O O O O DMAPP 18 A. Before enzyme addition [1- O]-DMAPP 20

α-18 [ O]-DMAPP B. E89A PIX reaction O O O P O P O O O

C. E89Q PIX reaction

-5.8 -5.9 -6.0 -6.1 -9.1 -9.2 -9.3 -9.4 ppm Figure 2.6 31P NMR spectra showing the α-phosphorus signals of the DMAPP mixture in the PIX reactions catalyzed by the E89 mutants. A), spectrum before enzyme addition; B), spectrum for the E89A catalyzed reaction; C), spectrum for the E89Q catalyzed reaction.

Isotopic scrambling (PIX) was observed in both the E89A and E89Q catalyzed reactions

(Figure 2.6). After the reaction was carried out by the Glu89 mutants, a new doublet appeared at

δ -9.31 ppm which is in the region of the α-phosphorus signals. The larger upfield shift of the signals is consistent with a higher P-18O bond order and indicates the formation of non-bridging

[α-18O]-DMAPP. The percentage of non-bridging [α-18O]-DMAPP was calculated to be 21% and

18% based on the peak integrations. Thus the percentage of PIX (%PIX) was determined to be

31% and 27% in the E89A and E89Q reactions, respectively. Using this data, the partitioning ratios (υPIX/υrxn) of the E89A and E89Q reactions were determined to be 3.9 and 1.5, signifying a 10x and 3.8x increase over the partitioning ratio observed in wild type reaction, respectively

(Table 2.2).

Surprisingly, in the PIX reaction catalyzed by E89A, in addition to the new signals from [α-

18O]-DMAPP, another doublet appeared at δ -5.41 ppm (Figure 2.7). This new set of signals likely corresponds to β-phosphorus with 18O label attached at a position with a bond order greater than one (Figure 2.7). It is possible that a movement of the pyrophosphate moiety relative to that of the dimethylallyl cation occasionally allows the pyrophosphate to flip inside the active site.

This motion would interconvert the positions of the phosphates and move the 18O label from the

α- to the β- position of the recovered DMAPP to give the ‘18O-flipping’ product. The extent of

‘18O-flipping’ was evaluated by dividing the integral of the [β-18O]-DMAPP signal by the combined integrals of the β-phosphorus signals. It turned out that a significant amount of recovered starting material (9% after 8% DMAPP consumption) showed DMAPP bearing 18O isotope at the terminal β-position. In another words, in the recovered DMAPP, 9% of the pyrophosphate had flipped inside the active site before reattaching to dimethylallyl carbocation.

It is not completely clear why pyrophosphate flipping is only observed with E89A. We postulate that the extensive shortening of the side chain in this mutant possibly increased the volume of the active site and provides more room for such a dramatic motion to occur. Additionally, the severely reduced activity of the mutant might also increase the probability of the reversible 61 formation of the allylic carbocation/pyrophosphate ion pair and result in a greater chance for the pyrophosphate flipping. It should be noted that we did not include the amount of ‘18O-flipping’ during the calculation of the PIX extent for E89A due to the complexity of the calculations. As a result, the partitioning ratio (vPIX/vrxn) for E89A is probably underestimated.

A. Before PIX experiment

O O α-18 O P O P O [ O]-DMAPP O O + unlabeled DMAPP

B. After PIX experiment

O O O P O P O O O - [β 18O]-DMAPP

ppm -5.20 -5.30 -5.40 -5.50 -5.60

Figure 2.7 31P NMR spectra showing the β-phosphorus signals of the DMAPP in the E89A- catalyzed PIX reaction. (A) spectrum before PIX experiment. (B) sepectrum after PIX experiment. Red circle represents a mixture of 18O and 16O in labeled and unlabeled DMAPP, respectively. Solid red circle represents the 18O label scrambled onto the β-phosphorus.

The residue Glu89 was proposed to promote the nucleophilic attack of the indole ring by hydrogen-bonding to the indole N-H, as previous described in the crystallographic studies.[56]

This is consistent with our observation that the E89A mutation resulted a 10-fold increase of the partitioning ratio (υPIX/υrxn). This could be attributed to two factors. Firstly, the nucleophilicity 62 of the indole ring will be greatly reduced as the alanine residue in the mutant enzyme lacks the polar interaction with the indole N-H. This will greatly increase the energy barrier of the nucleophilic attack step and decrease the commitment to catalysis. On the other hand, the formation of the allylic carbocation/pyrophosphate ion pair should become more reversible, which provides additional chance for the isotope scrambling (and pyrophosphate flipping in the case of E89A). The partitioning ratio of the E89Q mutation was only increased by 3.8-fold, implying that the barrier of the nucleophilic attack step increased to a smaller extent. Since glutamine is a weak hydrogen-bond acceptor, we suspect that it might still retain the capability to catalyze the nucleophilic attack, albeit much less efficiently. Additionally, it should be noted that the increase in magnitude of the υPIX/υrxn with the two mutants roughly correlates to the decrease of the kcat values. This is consistent with the notion that reduced enzyme activity might contribute to a larger extent of PIX.

2.3.3.2 PIX Reactions Catalyzed by the DMATs Mutants K174A and K174Q

We also inspected the PIX reactions catalyzed by the DMATs mutants K174A and K174Q.

Before the enzyme reaction, the corresponding α-phosphorus signals of DMAPP and [1-18O]-

DMAPP appear as two sets of doublets centered at δ -9.28 ppm and δ -9.30 ppm, respectively

(Figure 2.8). After extensive incubation with the mutants at 37 oC, a new set of signals corresponding to the unlabeled and 18O-labeled pyrophosphate appeared at δ -6.14 ppm and δ -

6.15 ppm, respectively. Using the same method described previously, the extents of the K174A 63 and K174Q reactions were determined to be 6% and 33%, respectively.

O O O O O P O P O O P O P O DMAPP A. Before enzyme addition O O O O

[1-18O]-DMAPP 20

α-18 [ O]-DMAPP B. K174A PIX reaction O O O P O P O

O O

C. K174Q PIX reaction

ppm -6.0 -6.1 -6.2 -6.3 -9.1 -9.2 -9.3 -9.4 -9.5

Figure 2.8 31P NMR spectra showing the α-phosphorus signals of the DMAPP mixture in the PIX reactions catalyzed by the K174 mutants. A), spectrum before enzyme addition; B),

spectrum for the K174A catalyzed reation; C), spectrum for the K174Q catalyzed reaction.

Isotope scrambling was also observed in both the K174A- and K174Q-catalyzed PIX reactions (Figure 2.8). A new set of doublets appeared at δ -9.31 ppm in the region of the α- phosphorus signals, indicating the formation of [α-18O]-DMAPP. The amount of [α-18O]-

DMAPP in the recovered DMAPP was calculated to be 17% and 16% and thus the percentage of PIX was estimated to be 26% and 24% in the K174A and K174Q reactions, respectively. With 64 this information, the partitioning ratios (υPIX/υrxn) of the K174A and K174Q reactions were estimated to be 4.3 and 0.74, showing an 11x and 1.9x increase, respectively, when compared to the partitioning ratio observed in the wild-type-catalyzed reaction (Table 2.2). In addition, the

31P NMR spectra of the Lys174 mutant reactions did not show any signs of the ‘18O-flipping’, as was observed in the E89A reaction.

The residue Lys174 was proposed to deprotonate the arenium intermediate and reconstitute the aromaticity.[56] Our observation that the K174A mutation resulted in an 11-fold increase of the partitioning ratio (υPIX/υrxn) could be attributed to two factors. Firstly, the non-basic alanine mutation lacks any potential to deprotonate the arenium intermediate, which will significantly increase the energy barrier of the rearomatization step thus decreasing the commitment of the arenium intermediate towards product formation. In addition, the increased energy barrier of the rearomatization step should make the previous steps including the formation of the allylic carbocation/pyrophosphate ion pair more reversible, thereby increasing the amount of isotopic scrambling. It was also observed that the partitioning ratio of the K174Q mutant only underwent a 1.9-fold increase, implying a much smaller change to the barrier of the rearomatization step.

Similar to the results with the E89 mutants, it was noticed that the increase in the magnitude of the υPIX/υrxn roughly correlated to the decrease in the kcat values.

It was previously reported in the crystallographic studies that the K174Q mutant was partially active and it was proposed that the carboxylate group of the L-Trp could polarize a water molecule to deprotonate the arenium intermediate.[56] Our observations of the rather different activities with the alanine and glutamine mutants argue against this hypothesis. The alanine 65 mutant would provide more room for the water molecule to get involved in the catalysis than glutamine mutant does, therefore a higher activity would be expected for K174A based on their hypothesis. Hence, we postulate that the glutamine residue is still basic enough to deprotonate the arenium intermediate as the arenium intermediate is quite acidic, thus both the kcat and partitioning ratio (υPIX/υrxn) of K174Q were very similar to those observed with wild-type enzyme. However, a non-basic alanine residue would lack the ability to deprotonate the arenium and therefore a much more significant change in both the kcat value and partitioning ratio

(υPIX/υrxn) is observed.

Table 2.2 The measured partitioning ratios for the wild-type DMATs and mutant-catalyzed PIX reactions

DMATS Partitioning Ratio (υPIX/υrxn) Relative change in υPIX/υrxn wild-typea 0.39 1x E89A 3.9 10x E89Q 1.5 3.8x K174A 4.3 11x K174Q 0.74 1.9x a. The results with wild-type DMATS are from the literature.[78]

Analysis of the Products in the Mutant-Catalyzed Enzyme Reactions

Although it was encouraging to find out that the results from PIX experiments agreed with our hypotheses about the catalytic roles of the residues Glu89 and Lys174, we still felt the

66 necessity to examine the products of the mutant-catalyzed enzyme reactions. As mentioned in the beginning of this chapter, it was possible that the mutations could change the fate of dimethylallyl carbocation. For instance, if a water molecule could access the mutated enzyme active site and attack the dimethylallyl carbocation, either a primary or tertiary allylic alcohol could be formed. Thus the observed formation of pyrophosphate and isotope scrambling do not necessarily mean that DMAT was produced. Therefore, a series of experiments were carried out to examine the enzymatic reaction products in the mutant catalyzed reactions.

2.4.1 Examination of the Enzymatic Reaction Products Using 1H NMR Spectroscopy

Since our primary concern was that the dimethylallyl carbocation might be attacked by a water molecule to form an allylic alcohol, we first examined the mutant-catalyzed reaction using

1H NMR spectroscopy. For the PIX experiments, the mutants were purified in Tris·HCl buffer, which does not interfere with the acquisition of 31P NMR spectrum. However, the presence of

Tris buffer in the enzyme solution will dramatically complicate the 1H NMR spectra, thus all the mutant enzymes were exchanged into phosphate buffer before running the enzyme reactions.

Buffer exchange was accomplished by dialysis against phosphate buffer. A solution containing

L-tryptophan, DMAPP and mutant enzyme in phosphate buffer was incubated for 10 h at 37 oC before the enzyme was removed by ultrafiltration. The filtrate was lyophilized and a 1H NMR spectrum was obtained in D2O. In most cases, the HOD peak at δ 4.7 ppm was predominant and made it hard to observe the signals from allylic protons, thus an NMR protocol with solvent 67 suppression was utilized to effectively remove the water peaks.

In the wild-type enzyme catalyzed reaction, the vinylic proton on the DMAT prenyl moiety appears as a characteristic triplet at δ 5.25 ppm due to its coupling with the adjacent methylene group. The 1H NMR spectra of the reactions catalyzed by the mutants E89A, E89Q and K174Q were identical to that of the reaction catalyzed by the wild-type enzyme, implying that DMAT was the only product originating from DMAPP for all three mutants. With K174A, however, only a small amount of DMAT was detected while a new set of signals containing a doublet of doublets (1H) and two doublets (2H) appeared around δ 5-6 ppm, the characteristic region for vinylic protons (Figure 2.9). In addition, the two singlets (6H) corresponding to the two methyl groups on the prenyl moiety shifted to δ 0.9-1.0 ppm, compared to those of DMAT that appear at δ 1.6 ppm. Based on the integrations of the vinylic protons, about 10% of the product was

DMAT while the majority was an unexpected compound X. The coupling pattern and integration of the vinylic signals, together with the shifted methyl signals, suggested that the unknown product was ‘reverse’ prenylated. The simplicity of the spectrum suggests the ‘reverse’ prenylated product was formed as a single stereoisomer.

Ha 1' Hb 2' 4' COO 5' NH 2 2 CH3-4'/5' N H 1 H compound X

H-2

HOD Ha Hb H-2'

.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0. ppm

1 Figure 2.9 H NMR spectrum (400 MHz, D2O) of the K174A reaction mixture. Assignments of the characteristic proton signals from the reverse prenylation are shown in the spectrum. Excess tryptophan and natural product DMAT is also present in the spectrum. The NMR spectrum is referenced to the HOD peak in D2O (δ 4.7 ppm).

Initially, we postulated two possibilities that could account for the appearance of the reverse prenylation signals. It is possible that a water molecule could access the active site of the mutant enzyme and quench the dimethylallyl carbocation to produce a tertiary allylic alcohol.

Alternatively, the mutation of Lys174 to alanine might provide some extra room and allow the dimethylallyl carbocation to be mispositioned within the active site. This might result in a nucleophilic attack from the indole ring onto the tertiary carbon of the dimethylallyl carbocation and produce a reverse-prenylated product (Figure 2.10). In order to distinguish between these

69 possibilities, 2-methyl-3-buten-2-ol was added into the reaction mixture and the resulting 1H

NMR spectrum showed two distinct sets of vinylic proton signals. Therefore, it was concluded that the unknown product in the K174A reaction did not result from the hydrolysis of DMAPP.

As a first step in determining the structure of the unknown product, the products of the K174A reaction were analyzed by capillary electrophoresis (CE) coupled with photodiode array (PDA) detection. This can provide a good separation as well as a UV spectrum of the unusual product.

H H O

K174A-CH3 OH PPO COO COO DMAPP 2-methyl-3-buten-2-ol NH NH3 3 PPi COO N N NH H H 3 L-Trp O N O H Glu89 reverse prenylated product?

Figure 2.10 Possible explanations for the production of reverse-prenylated products in the K174A catalyzed reaction.

2.4.2 CE Analysis of the Products from the K174A Catalyzed Reaction

Reactions catalyzed by the wild-type enzyme as well as the K174A mutant were incubated at 37 ºC for 16 h to ensure completion. The enzymes were then removed by ultracentrifugation before the reaction mixtures were lyophilized. The residues were dissolved in 1 mL deionized

70 water before the capillary electrophoresis analysis. The capillary electrophoresis analyses were carried out in the laboratory of Prof. David Chen at the University of British Columbia. With an authentic sample of DMAT, we were able to confirm its elution time to be 10.2 min and that of the unknown product to be 9.4 min. Also, the UV spectra of both DMAT and the unusual product

X produced by K174A were acquired (Figure 2.11). It was clear that the indole chromophore

(λmax = 278 nm) had been disrupted and two new absorption maxima at 237 nm and 292 nm had appeared. We later discovered that the two new absorption maxima were very similar to those observed with the hexahydropyrroloindole alkaloids such as chimonanthine (λmax = 240 and 294 nm, acidic conditions, Figure 2.11).[82] This discovery led us to suspect that the unusual product

(X) also bore the characteristic tricyclic ring system of the hexahydropyrroloindole alkaloids

(Figure 2.11). From the relative position of the prenyl and indole group observed in the structural studies (Figure 1.27), one could expect that the prenyl group would add onto the Re face at the

C-3 position of the indole ring (Figure 2.10). Furthermore, it is known that cis-fused 5-membered ring systems are more stable. Therefore, the expected stereochemistry of compound X is shown in Figure 2.11.

COO COO NH3 R NH2 N H N H H X DMAT compound

278 nm 237 nm 292 nm

H H C H 3 N N chimonanthine

λmax = 240 and 294 nm N N (acidic conditions) H CH3 H Figure 2.11 UV spectra of DMAT and the unusual product (X) and the structure of chimonanthine.

2.4.3 Confirmation of the Structure of Product X by Synthesis

The proposed chemical structure of compound X can account for the observed signals in the

1H NMR spectrum acquired from the K174A reaction (Figure 2.9). The doublet of doublets at δ

5.99 ppm is attributed to the vinylic H-2’. The doublet at δ 5.18 ppm has a coupling constant of

J = 10.9 Hz and therefore is assigned to the terminal vinylic proton Ha that is cis to H-2’, whereas the doublet at δ 5.13 ppm with a coupling constant of J = 17.4 Hz is assigned to the terminal vinylic proton Hb that is trans to H-2’. Additionally, the singlet that appeared at δ 5.31 ppm is associated with the proton at the C-2 position. The shifting of methyl signals to δ 1.10 and 0.99 ppm is attributed to the fact that they are no longer directly attached to the double bond as in the

72 case of DMAT. Nevertheless, it would be more convincing if an authentic sample of compound

X could be synthesized and its 1H NMR spectra compared. Fortunately, the Danishefsky group had prepared the methyl ester of our suspected compound X during their synthesis of amauromine, which provided us with a convenient way to obtain a sample of this material.[83]

Following the previously described procedure, the L-tryptophan methyl ester was first protected using Boc anhydride to give N,N’-Boc L-tryptophan methyl ester (27, Figure 2.12).

Compound 27 was then treated with N-phenylselenophthalimide (N-PSP) in the presence of equal molecular amount of pyridinium p-toluenesulfonate (PPTS) to construct the tricyclic hexahydropyrroloindole ring system and give compound 28 as a mixture of stereoisomers. It has been shown that the resulted hexahydropyrroloindole structure was formed as an 18:1 mixture of stereoisomers resulting from addition to both the Re and Si face at the C-3 position. Due to the difficulties in separation, the mixture was carried on and the selenophenyl group was replaced by a reverse-oriented prenyl group using prenyltri(n-butyl)tin to yield cis-fused hexahydropyrroloindole 29. The Boc protecting groups were removed from 29 with trimethylsilyl iodide (TMSI) to render the methyl ester of the desired product (30). Since the unprotected hexahydropyrroloindole ring is not stable in acidic conditions, the methyl ester was hydrolyzed under carefully controlled basic conditions and the resulting compound X was carefully neutralized to pH 7 using an acidic resin. No further purification of compound X was attempted. The 1H NMR spectrum of compound X was then acquired.

N-PSP O COOMe COOMe N SePh NaOH PhSe COOMe NH2 NHBoc (n-Bu)4NHSO4 O NBoc PPTS N (Boc)2O N N H H Boc Boc 28 L-tryptophan methyl ester 27 MeOTf

t-Bu N t-Bu Sn(Bu)3

COOH LiOH COOMe TMSl COOMe o THF/H O MeCN, 0 C NH 2 NH NBoc N H N H N H H H Boc X 30 29

Figure 2.12 Synthesis of compound X using Danishefsky’s procedure. Synthesis of compound 30 was published by Danishefsky et al..[83]

The spectra of the enzymatically-produced and synthetic compound indeed turned out to be very similar (Figure 2.13). Nonetheless, small differences in certain chemical shifts and coupling constants were observed, possibly due to minor differences in the pH of the NMR samples.

Attempts were then made in order to carefully adjust the NMR samples to the same pH in the hope of getting the identical 1H NMR spectra, however, it was nearly impossible to get all the chemical shifts to be identical. Finally, a sample of synthetic compound X (roughly one equivalent) was added into the enzyme reaction mixture and a 1H NMR spectrum was acquired, which unambiguously showed that the vinylic proton signals and the upfield methyl signals overlapped and the intensities increased accordingly (Figure 2.13). This clearly demonstrates that the structure of compound X was as depicted in Figure 2.7. More importantly, it confirmed

74 that the unknown product bears the R configuration at C-3.

.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0. ppm

1 Figure 2.13 A comparison of H NMR spectra (D2O) of compound X obtained from the K174A reaction, chemical synthesis and the mixture of both. The top spectrum corresponds to the K174A catalyzed reaction, the middle spectrum is for the chemically synthesized X and the bottom spectrum shows the reaction mixture spiked with approximately one equivalent of compound X.

Re-examination of the DMATs Enzyme Mechanism

With the chemical structure and stereochemistry of the K174A product compound X confirmed, we began to question how this ‘reverse’ prenylated compound was produced. Given

75 the relative positioning of the two substrates DMAPP and L-Trp inside the enzyme active site, there are several possible explanations that could account for its production (Figure 2.14). One possible explanation for the formation of this reverse-prenylated compound involves an initial attack from the C-4 position of Trp to render the normal arenium intermediate, and in the absence of a base to deprotonate the C-4 position, a Cope rearrangement takes place and generates an iminium intermediate that can cyclize to give compound X (Figure 2.14, pathway A).

Alternatively, it is possible that a direct attack from the C-3 position of tryptophan onto the tertiary center of the dimethylallyl carbocation can provide the iminium intermediate, which can quickly cyclize to give the ‘reverse’ prenylated product (Figure 2.14, pathway B).

COOH COOH H NH NH 2 2 arenium DMAPP Pathway A intermediate N N PPO COOH H H Cope rearrangement COOH NH2

NH2 N N H COOH COOH H L-Trp NH Pathway B 2 NH2 'reverse' prenylated compound X N N H H iminium intermediate Figure 2.14 Two possible pathways leading to the formation of the ‘reverse’ prenylated compound X.

In order to probe the likelihood of these two possibilities, we examined the X-ray crystal structure of DMAT synthase in complex with tryptophan and DM-S-PP as this best resembles 76 the Michaelis complex. In the active site of DMAT synthase, it was noticed that the plane of the dimethylallyl group lies parallel to that of the indole ring and that, surprisingly, the distance between C-4 of tryptophan and C-1 of DMSPP is 3.8 Å, which is very similar to the 3.5 Å distance between C-3 of tryptophan and C-3 of DMSPP (Figure 2.15). Assuming that the dimethylallyl group does not undergo a significant movement in the active site upon dissociation of the pyrophosphate, the two possible pathways leading to the ‘reverse’ prenylated product will be equally likely from a geometric perspective. Ample evidence has shown that the C-3 position on indole rings is the strongly favored position for electrophilic addition.[84],[85] A relevant example involves reaction between indole and prenylbromide, which selectively produces 3- prenylindole as the major product.[86] Thus from the perspective of intrinsic chemical reactivity, an initial nucleophilic attack from C-3 of Trp onto the dimethylallyl carbocation will be favored once the carbocation is formed. Therefore, it is not clear how the wild-type enzyme can prevent such a C-3 attack from occurring.

Figure 2.15 Active site architecture of DMAT synthase in complex with L-tryptophan and DMSPP. [56] Key active site residues K174 and E89 are shown. Distances were measured based on the crystal structure taken from PDB 3I4X.[56]

Given the observations that the K174A mutant forms a compound that is ‘reverse’ prenylated at C-3 and that nucleophilic attack from C-3 of Trp is actually more favorable based on both the intrinsic reactivity and the crystallographic results, we re-evaluated the previously proposed mechanism for the wild-type DMAT synthase reaction. We felt it is reasonable to propose that wild-type DMAT synthase mechanism actually involves an initial nucleophilic attack from C-3 of tryptophan onto the tertiary center of the dimethylallyl carbocation to give a ‘reverse’ prenylated iminium intermediate (Figure 2.16). Subsequently, a reversible Cope rearrangement interconverts this species with the ‘normal’ C-4 prenylated arenium intermediate, which gives the normal product DMAT upon deprotonation. However, in the case of the K174A mutant, due to the lack of a basic residue required to deprotonate the arenium intermediate, the more 78 thermodynamically stable iminium intermediate is released from the active site and readily cyclizes to produce compound X (Figure 2.16).

Lys174 PPi NH2 COO COO COO H Cope NH3 NH3 NH3

N N N H H H iminium arenium intermediate intermediate K174A WT DMATs mutant

COO COO NH3 NH2 N N H H 'reverse' prenylated DMAT X compound Figure 2.16 A Cope mechanism for the reactions catalyzed by the wild-type DMAT synthase and the K174A mutant.

The Feasibility of the Cope Mechanism for DMAT Synthase

After the fact, we found that the Wenkert group had proposed such a ‘Cope mechanism’ back in 1970s in order to address how substitution takes place at the relatively unactivated C-4 position of the indole ring (Figure 2.17).[16],[87] In their proposed mechanism, the initial step involves an attack from the Re face of the indole C-3 onto the tertiary position of the carbocation

79 to give a ‘reverse’ prenylated iminium intermediate. Ring closure readily occurs to produce the hexahydropyrroloindole compound. A reversible Cope rearrangement then interconverts this species with the C-4 ‘normal’ prenylated arenium intermediate, and a final elimination renders product.

O O O O H

HN N N HN N N N N H H H Figure 2.17 Hypothetical Cope rearrangement in ergot alkaloids biosynthesis proposed by Wenkert and Sliwa.[87]

When the Wenkert group proposed the Cope mechanism for ergot biosynthesis, they synthesized the model indoline compounds 31 and 32 to test the feasibility of such a non- enzymatic rearrangement (Figure 2.18).[16],[87] They heated the model compounds up to 250 oC, yet no Cope rearrangement products were observed in either case, and they concluded that it is unlikely that a Cope rearrangement is involved in ergot alkaloid biosynthesis.

250 oC X N N H 31 H

250 oC X N N H H 32

Figure 2.18 Studies of the Cope rearrangement with model compounds carried out by Wenkert.[16]

We would argue, however, that the order of occurrence of the Cope rearrangement and the cyclization would greatly affect the kinetics of the rearrangement. In our mechanism, the Cope rearrangement happens prior to the cyclization, and the important imine functionality

(unsaturation between N1 and C2 of the indole ring) is present during the rearrangement (Figure

2.16). While the compounds 31 and 32 were designed to model a potential Cope rearrangement in a ring-closed hexahydropyrolloindole, they are imperfect mimics of the iminium intermediate in our mechanism (Figure 2.18). The absence of the unsaturation would certainly influence the preferred conformation of the molecule and could preclude the the Cope rearrangement from occurring. Additionally, the imine (or iminium) in our proposed mechanism becomes extensively conjugated to the rest of the π-system in the transition state of the Cope rearrangement. This type of stabilization could lead to a lowering of the reaction barrier that would not be present in the

Wenkert model compounds. 81

A more analogous model reaction was carried out in Arigoni group. It was never published in the literature but has been reviewed in an article by Floss.[16] They prepared compound 33 that contains the imine functionality but lacks the dimethyl substituents, which would be expected to accelerate the Cope reaction via a gem-dimethyl-like effect.[88] Upon heating the compound 33 to 250 oC, the N-alkylated compound 34 was produced (Figure 2.19). Therefore, the compound did undergo an apparent aza-Cope rearrangement, albeit in the opposite direction around the indole ring. However, from an analysis of the DMATs active site, it is clear that the prenyl group is held in an orientation that favors the Cope rearrangement towards the C-4 position.

250 oC 3 h N N

33 34

Figure 2.19 Model compound study from the Arigoni group.

Better supporting evidence for the feasibility of the pericyclic rearrangement on the indole ring can be found in the Claisen rearrangement studies on indoloquinoline derivative 35 (Figure

2.20).[89] Vo û t e et al. showed that upon reflux of 35 in toluene for 5 h, a Claisen rearrangement first resulted in compound 36. When compound 36 was allowed to further reflux for 5 days, a mixture of compounds 37 (72%) and 38 (12%) were produced. Compounds 37 and 38 are presumably generated by Cope rearrangements onto either the indole N-1 or C-4 positions, respectively. This observation clearly indicates that such a rearrangement is feasible under non-

82 enzymatic conditions even though an elevated temperature might be required.

O O PhMe O PhMe O reflux reflux 5 h, 89% 5 days + N N N N N N Me N N Me Me H Me

35 36 37 (72%) 38 (12%) Figure 2.20 Non-enzymatic reaction involving a Cope rearrangement onto the C-4 position of an indole.

Our hypothesis of a DMATS mechanism involving a Cope rearrangement has triggered model studies on this rearrangement as well.[90],[91] Schwartzer et al. showed that compound 39 will spontaneously rearrange at room temperature to give compound 40 (Figure 2.21).[90] This presumably occurs via a Cope rearrangement onto the indole C-4 position followed by tautomerization. A fused-three membered ring is present in compound 40, which certainly contributes to the driving force for the rearrangement upon releasing its ring strain. However, it is conceivable that the gem-dimethyl-like effect in the iminium intermediate can accelerate the rearrangement as well.[88]

THF H 3 h tautomerize O O R. T. O N N N H H H 39 40

Figure 2.21 Cope rearrangement reaction under ambient temperature.

Rudolf et al. probed the 4-DMATs reaction with tryptophan analogs containing substituents at the 4-position of the indole ring.[92] When 4-methyl-L-tryptophan (41) was employed as a substrate, there was a very low level of catalytic activity (0.3 %). However, the major product was identified to be a reverse C-3 prenylated hexahydropyrroloindole 42, which is consistent with an initial nucleophilic attack from C-3 followed by the release of the reverse C-3 prenylated intermediate as was suggested in the Cope mechanism (Figure 2.22). A minor product 43 was normal N-1 prenylated, which could result from an aza-Cope rearrangement of the reverse C-3 prenylated intermediate. Another minor product 44 was normal C-3 prenylated, indicating that some motion of the dimethylallyl cation was possible when a C-4 substituent is introduced.

Surprisingly, when 4-methoxy- or 4-amino-L-tryptophan were tested, products that were isolated were normal prenylated at the C-5 and C-7 positions and no C-3 prenylation was observed. The authors forwarded this as evidence against the Cope mechanism since these products could not have arrived from a C-3 reverse prenylated intermediate. However, we felt that a methoxy, or especially an amine, substitution at C-4 of Trp would dramatically activate the C-5 and C-7 positions for electrophilic aromatic substitution and that the use of the activated Trp analogs may

84 cause the change of prenylation positions.[85] Therefore, a Cope mechanism remains feasible for the normal 4-DMATS mechanism.

COO COO COO NH2 R= Me, 41 N NH2 + + NH2 COO N N R H H NH3 42 43 44 DMATs COO N DMAPP COO COO NH OMe 2 + H NH2 NH NH 2 3 NH3 + + R=NH , OMe N 2 N N H H H

Figure 2.22 Studies on 4-DMATS using C-4 substituted Trp analogs.

More recently, the most compelling evidence for a Cope rearrangement from the C-3 to C-4 positions on the indole has been reported by Viswanathan et al. in their attempt to mimic the

DMATS reaction.[93] They prepared a series of C-3 ‘reverse’ prenylated compounds and heated these compounds in phosphate buffer with microwave irradiation. Products bearing the prenyl moiety on the C-4 position were formed, possibly through a two-step mechanism that is similar to the one we proposed for DMATS (Figure 2.23). A Cope rearrangement first generated nonaromatic intermediates, which are similar to the arenium intermediate generated in DMATS reaction. Then a tautomerization reconstituted the aromaticity, corresponding to the deprotonation by Lys174 in the DMATS reaction. The authors also investigated the pH 85 dependence of the rearrangement on protected tryptamine derivative 45. Under acidic conditions

(pH =6.0 - 6.5), no rearrangement product 46 was detected, however, the yield of product 46 improved at higher pH and was maximized at pH 8.8. This can be explained by the need for a base to deprotonate at C-4, similar to the role of Lys174 in DMATS.

R H Cope H R R O O O N phosphate buffer pH H µ N N W, 150 W H dependent H 150 oC R= CH CH NPhth 45 intermediate R= CH CH NPhth 46 2 2 , 2 2 , Figure 2.23 Cope rearrangement under biomimetic conditions.

Further Experimental Attempts to Gain Support for the Cope Mechanism

2.7.1 Attempts to Convert the Reverse-Prenylated Compound X to DMAT

Since the interconversion of the reverse-prenylated compound X and the iminium intermediate (i.e., ring opening) is expected to be reversible in solution, we proposed to test whether wild-type DMAT synthase is capable of converting compound X into DMAT. In a first attempt, the K174A catalyzed reaction mixture was incubated with wild-type DMAT synthase for extended periods of time before the sample was analyzed. Pyrophosphate was also added in order to ensure it was available for binding, yet no significant conversion of compound X to 86

DMAT was observed. Subsequently, after the chemical synthesis had been completed, we incubated the synthetic compound X with wild-type DMAT synthase under similar conditions, but again failed to detect any noticeable conversion. We argue that the failure of this attempt might be due to the thermodynamic stability of the ring-closed form of compound X in solution.

Since DMAT synthase would not be able to catalyze the ring-opening of compound X, there might not be sufficient amounts of the ring-opened iminium intermediate available for catalysis.

In the wild-type enzyme-catalyzed reaction, such a ring closure could be prevented by orienting the α-amino group away from C-2 or by keeping it in a protonated form throughout the lifetime of the enzyme-bound intermediate. Therefore, the failure of this attempt does not preclude the possible Cope mechanism for DMAT synthase.

2.7.2 Substrate Analog Studies Aimed at Trapping the Iminium Intermediate

We envisioned that solid support for the Cope rearrangement mechanism could be obtained if the iminium intermediate in the wild-type reaction, or a derivative thereof, could be trapped and isolated. To start with, we proposed that the iminium intermediate might be reduced in the presence of reducing agent (NaBH4) either within the active site or free in solution. Wild-type reactions were therefore carried out in the presence of NaBH4 in phosphate buffer, yet no noticeable amount of a reduced reverse-prenylated compound could be detected even after the reaction had gone to completion. We interpreted this as evidence that the wild-type enzyme binds the iminium intermediate tightly in the active site and that the active site is inaccessible to 87 reagents such as borohydride.

Alternatively, we proposed to use tryptophan analogs that are unable to undergo the Cope rearrangement yet are still capable of attacking from the C-3 position. If they are still able to undergo the initial C-3 nucleophilic attack, the iminium intermediate might be released and cyclize to render products similar to compound X. We devised two substrate analogs in the hope of identifying reverse-prenylated compounds (Figure 2.23). For compound 4-aza-tryptophan

(47), the C-4 position was replaced by nitrogen, therefore we suspect that, after the initial attack on C-3 position, the resulting iminium intermediate would be released and cyclize to produce a tricyclic product 48. For another substrate analog (S)-2-amino-3-(1H-pyrrol-3-yl)propanoic acid

(49), there is no C-4 position for the Cope rearrangement and again we anticipated that the iminium intermediate might be released and cyclize to render a ring-closed product 50.

COO COO COO NH N NH3 N N 3 DMAPP NH2 N N N H H H 47 iminium intermediate 48

COO COO

NH3 DMAPP NH3 COO N N N N H H H H 2 49 iminium intermediate 50

Figure 2.24 Proposed reactions with substrate analogs.

4-Aza-Trp (47) was synthesized enzymatically with tryptophan synthase following a previously reported procedure (Figure 2.24).[94] Several synthetic routes were available at the time of our proposal. However, we chose to use tryptophan synthase with the substrate 4- azaindole because of its simplicity and the reported high yield. A commercially available sample of E. coli pre-transformed with a plasmid expressing tryptophan synthase was ordered from

ATCC. A crude cell lysate was used to carry out the biotransformation without any further purification. After the enzymatic conversion, the product 4-aza-Trp was purified using C18 reversed-phase chromatography. Subsequently, 4-aza-Trp was treated with DMAPP in the presence of wild-type DMATs and the reaction mixture was carefully examined by 1H NMR spectroscopy. Unfortunately, no production of tricyclic compound with reverse-prenylation could be detected and starting materials were recovered.

COO N HO N Trp synthase NH3 + o 37 C, 3 N H N COO days. N H 3 H

4-azaindole L-serine 47

Figure 2.25 Enzymatic synthesis of 4-aza-Trp (47) using tryptophan synthase.

The synthesis of substrate analog 49 was achieved using a known procedure (Figure 2.25).[95]

A Boc protection of pyrrole-3-carboaldehyde gave compound 51. The side chain was extended

89 though a Wittig reaction to give 52. The double bond was then reduced to render a racemic mixture of compound 53. Deprotection of the mixture was first attempted using the method described in the synthesis of reverse-prenylated compound X, yet no reaction was observed.

Subsequently, the mixture was subjected to basic deprotection, which effectively removed the

Boc and methyl ester group to give 54. The Cbz protecting group was then removed by hydrogenolysis to yield the desired compound 49 as a racemic mixture. This compound was then treated with DMAPP in the presence of wild-type DMAT synthase, however, no reaction could be observed and only starting materials were recovered.

O H H H NHCbz P NHCbz O (OMe)2 O NiCl2 O COOMe (Boc)2 COOMe NaBH4

N N TMG N H Boc - 30 oC Boc pyrrole-3-carboaldyhyde 51 52

NHCbz NHCbz NH2 1. MeONa/MeOH COOH Pd/C COOH COOMe 2. H2O H , MeOH N N 2 N Boc H H 53 54 49

Figure 2.26 Synthetic scheme for analog 49. Synthesis of 53 has been reported in the literature.[95]

Even though these attempts failed to provide further evidence for the Cope rearrangement mechanism, it is possible that the modifications made on these substrate analogs resulted in poor substrate binding that dramatically impacted their ability to participate in catalysis.

Summary and Conclusion

In summary, crystallographic studies implicated two important residues, E89 and K174, in the enzyme active site of 4-DMAT synthase as playing important roles in the enzyme mechanism.

The E89 residue forms a hydrogen-bond to the N-H of the indole ring and likely enhances the nucleophilicity of the indole, while the K174 residue is thought to act as the base that deprotonates the arenium intermediate and reconstitutes the aromaticity. These hypotheses have been examined through studies on the mutants generated by site-directed mutagenesis.

When the E89 residue is mutated to glutamine or alanine (E89Q and E89A, respectively), kinetic studies have shown a significant decrease of the value of kcat (E89A is nearly inactive).

This is consistent with the notion that E89 plays a key role in increasing the nucleophilicity of the indole ring either by deprotonation at N-1 or through hydrogen-bonding. The mutation of

K174 to glutamine resulted in only a 17% reduction in the value of kcat while mutation of this residue to alanine resulted in a 20-fold reduction. This surprising observation was interpreted as being due to a Gln residue at this position maintaining the ability to deprotonate the rather acidic arenium intermediate. However, an Ala residue would lack any potential to act as a base and result in the more profound rate decrease. In addition, the KM values of L-tryptophan for all three active mutants did not undergo significant change, which suggests that the decreased activity resulted from transition state destabilization rather than poor substrate binding.

Positional isotope exchange experiments confirmed the increase of isotope scrambling with 91 all four mutants, albeit to different extents. However, it was noticed that the extent of increase in

PIX roughly correlated to the corresponding decrease of the kcat values. These observations are consistent with the notion that the mutation of E89 and K174 to less polar residues increased the energy barrier of post-ionization steps, which will increase the reversibility of ionization while decreasing the commitment to catalysis.

When the products formed by the mutants were analyzed, an unusual product was identified from the reaction catalyzed by K174A. The 1H NMR spectrum of this product showed a new set of signals corresponding to reverse prenylated vinylic protons as well as two upfield shifted methyl signals. The structure of the unusual product was first postulated to be a reverse- prenylated hexahydropyrroloindole based on the UV spectrum obtained from capillary electrophoresis analysis. Later on, we successfully synthesized an authentic sample of the compound following a modification of a previously reported procedure. Due to the small discrepancies present in the 1H NMR spectra of the synthesized and enzyme-produced compounds, a spectrum of a 1:1 mixture of the compounds was obtained, which unambiguously confirmed the structure and the stereochemistry of this unusual product.

Two possible explanations were proposed for the production of the unusual compound X. It is possible that a nucleophilic attack first occurs at the C-4 position of indole and a subsequent rearrangement generates the reverse C-3 prenylated product. Alternatively, a direct nucleophilic attack from the C-3 position of indole onto the tertiary center of the allylic carbocation could render the reverse-prenylated product. After examining the X-ray crystal structure of DMAT synthase, we found that the distances required for the two possible attacks were very close, 92 suggesting that nucleophilic attack from either C-4 or C-3 of the indole are equally likely from a geometric perspective. These observations led us to the question of how DMAT synthase can catalyze prenylation regioselectively at the C-4 position while the C-3 position has been shown to be a more nucleophilic site. Therefore, a new mechanism involving a Cope rearrangement was proposed for wild-type DMAT synthase. In this mechanism, the prenyl group would first add to the C-3 position of the indole ring in a ‘reverse’ fashion. A Cope rearrangement would then migrate it to the C-4 position in a ‘normal’ fashion. This mechanism could explain the regio- specificity, the structural analysis, and the intrinsic chemical reactivity.

In conclusion, this work successively proved the catalytic roles of two residues identified from the crystallographic studies. Furthermore, the production of an unusual product compound

X with K174A mutant could be utilized to further diversify the ergot alkaloids family if such mutation could be introduced into in vivo conditions. More importantly, enzyme-catalyzed pericyclic mechanisms are somewhat rare and are only observed in a handful of enzymes including chorismate mutase, precorrin-8x methyl mutase, isochorismate pyruvate lyase and the

Diels-Alderases. Although pericyclic rearrangements have been proposed to play roles in the biosynthesis of the ergot alkaloids since the 1970s,[16] this work provides the first mechanistic evidence using a purified enzyme. While the mechanism still remains to be testified, we envision such a rearrangement mechanism could be a real possibility for this, and perhaps other, indole prenyltransferases. For example, similar findings have recently been reported on indole prenyltransferase FtmPT1 which catalyzes a ‘normal’ C-2 prenylation with brevianamide F. One active site mutant of FtmPT1 was also found to produce an unnatural C-3 ‘reverse’ –prenylated 93 hexahydropyrroloindole compound.[96] And later on, the wild-type enzyme was found to also produce compound X when L-tryptophan was used as an alternate substrate.[97] Given the great similarity of these two enzymes’ active site binding environment, it is reasonable to believe that a ‘reverse’ prenylation at C-3 is the first step of the catalysis as well.

Experimental Procedures

2.9.1 Materials and General Methods

All reagents were purchased from Sigma-Aldrich or Alfa Aesar and used without further

18 purification unless otherwise stated. D2O (99.9%) and H2 O (97%) was purchased from

Cambridge Isotope Laboratories, Inc.

Thin layer chromatography was performed on aluminum-backed sheets of silica gel 60F254

(Merck) of thickness 0.2 mm. Compounds were visualized by UV or by staining with a molybdate solution, containing H2SO4 (31 mL), ammonium molybdate (21 g), and Ce(SO4) (1 g) in water (500 mL). Silica gel (SiliaFlash® F60, 230 – 400 mesh, SiliCycle) was used for column chromatography.

Rosetta (DE3) pLysS E. coli cells were purchased from Novagen. Centrifugal filters (4 mL,

10 000 MWCO) were purchased from Millipore. Chelating SepharoseTM Fast Flow resin was purchased from GE Healthcare. Bradford assay dye was purchased from Bio-Rad. EnzChek®

Pyrophosphate Assay Kit was purchased from Life Technologies, Inc. Acryl-cuvettes for use in 94 enzyme kinetic assays were from Sarstedt.

Protein concentrations were determined using Bradford assay with BSA as standards. The enzyme kinetic assays were conducted on Cary 300 UV-vis spectrophotometer with an attached temperature controller.

1H NMR spectra were obtained on the Bruker AV400inv spectrometer at a field strength of

400 MHz. Proton-decoupled 31P NMR spectra were recorded on the Bruker AV400inv spectrometer at a field strength of 162 MHz.

2.9.2 Overexpression and Purification of the DMATs Mutants

Overexpression of the mutants (E89A, E89Q, K174A and K174Q) was achieved using a modification of the protocol described by Li et al.[9] E. coli. Rosetta(DE3) pLysS cells (Novagen) containing the recombinant mutated fgaPT2/pET28a plasmid were grown at 37 oC in 1 L of

Terrific Broth (TB) medium containing 35 μg/mL chloramphenicol and 30 μg/mL kanamycin until an OD600 of 0.6 was reached. Cells were induced for overexpression by the addition of 119 mg (0.5 mM) of isopropyl-1-thio-β-D-galactopyranoside (IPTG). Cells were harvested and lysed by French Press in buffer A (50 mM Tris-HCl, pH 7.5) containing 20 mM imidazole and 500 mM NaCl). The cell lysate was cleared by centrifugation at 8000 × g for 1 h before the supernatant was loaded onto a column of Chelating SepharoseTM Fast Flow resin (10 mL, loaded with 100 mM NiSO4 and then equilibrated with buffer A containing 500 mM NaCl). The column was washed with wash buffer (first with buffer A containing 500 mM NaCl, then with buffer A 95 containing 125 mM imidazole and 500 mM NaCl) and the enzyme was eluted with elution buffer

(buffer A containing 500 mM imidazole and 500 mM NaCl). Typically ~10 mg of enzyme was purified from 1 L of culture.

2.9.3 Steady-State Kinetic Characterization of DMAT synthase Mutants

Enzyme kinetics experiments were conducted using an EnzChek® Pyrophosphate Assay Kit

[78] (Invitrogen) following a previously described procedure. The KM values for L-tryptophan were measured in the presence of 200 μM DMAPP (saturating). For WT, E89Q, K174A and K174Q,

3 μg, 25 μg, 18 μg and 5 μg of enzyme was used per assay, respectively. Kinetic parameters for these enzymes were determined from the fit of the initial velocities to Michaelis-Menten equation using software program Origin Pro 8. The kcat and KM value for E89A could not be determined due to the low activity of the mutant and an upper value of kcat was estimated.

Enzyme mass of 35 kDa was used to calculate the enzyme concentrations.

2.9.4 Synthesis [1-18O]-DMAPP (20)

The synthesis of [1-18O]-DMAPP (20) was completed using the procedure described by Luk et al.[78] Dimethylallyl chloride (1.55 g, 14.8 mmol) was heated at 100 oC for 4.5 h in the presence

18 of triethylbenzylammonium chloride (TEBA) (0.387g, 1.7 mmol) and sodium [1, 1- O2]-acetate 96

(1.47 g, 17.0 mmol). The resultant ester 25 was then extracted with dichloromethane (3 x 10 mL) and concentrated to dryness under vacuum. The residue was dissolved in 20 mL of anhydrous methanol solution containing sodium methoxide (2.6 M) and refluxed for 4 h. The reaction was quenched by addition of 20 mL saturated NH4Cl solution and extracted three times with 20 mL diethyl ether. The organic layer was separated, dried with Na2SO4 and evaporated to dryness using rotovap. The product [1-18O]-dimethylallyl alcohol (26) was purified by Kugelrohr distillation. Compound 26 was coupled to triethylammonium phosphate (TEAP) in the presence

18 18 of trichloroacetonitrile (CCl3CN) to yield [1- O]-DMAPP (20). The extent of O incorporation was determined to be ~90% by ESI-MS.

2.9.5 PIX Experiments on DMAT Synthase Mutants

The PIX experiments with E89A, E89Q, K174A and K174Q were carried out as described previously with the exception that the amount of enzyme was increased to 5 mg.[78] A solution containing 1-[18O]-DMAPP and unlabeled DMAPP (total concentration of 30 mM in 1.0 mL with 63% 18O incorporation) and L-tryptophan (24 mM) in Tris-HCl buffer (50 mM, pD 7.5,

31 prepared using D2O) was prepared and the P NMR spectrum was collected. After the enzyme was added, the mixture was incubated at 37 ºC for 24 h. A 1.0 mL sample of the solution was subjected to ultrafiltration (Amicon Ultra-4, 10000 MWCO, 5000 rpm for 15 min, 4 °C) in order to remove the enzyme. Chelex-100 resin (50 mg of 100-200 mesh, Na+ form, pre-rinsed with

31 D2O) was added to the filtrate and the mixture was vortexed extensively. A second P NMR 97 spectrum was then acquired. The proton-decoupled 31P NMR spectra were recorded on a Bruker

AV400inv spectrometer operating at a frequency 162 MHz. Acquisition parameters included a

2437 Hz (20 ppm) sweep width centered at -5 ppm with a 27 s acquisition time. Well-resolved spectra were achieved after 200 to 1000 scans. All the spectra were optimized using appodization with exponential and Gaussian functions to achieve higher resolution.

2.9.6 Examination of the Enzyme Reaction Products Using 1H NMR Spectroscopy

For the mutant enzymes that were used in the product examinations, 10 mL enzyme eluents were dialyzed against phosphate buffer (3 × 500 mL, 20 mM, pH 7.5) for 12 h at 4 oC. Then a solution containing 2 mM L-tryptophan, 2 mM DMAPP and 9 mg mutant enzyme in phosphate buffer (20 mM, pH 7.5, final volume 10 mL) was prepared. The mixture was incubated for 10 h at 37 oC before the enzyme was removed by ultrafiltration (Amicon Ultra-4, 10 000 MWCO).

1 The filtrate was lyophilized and dissolved in 1 mL D2O and then a H NMR spectrum was obtained. In most cases, the water peak at δ 4.8 ppm was predominant and makes it hard to observe the signals from vinylic protons, thus a NMR protocol with solvent suppression was utilized to effectively remove the water peaks.

2.9.7 CE Analysis of the Products from the K174A Reaction

Separation of the DMAT (10%) and compound X (90%) produced by the K174A mutant 98 was accomplished using capillary electrophoresis with helps from Prof. David Chen’s group at the University of British Columbia. The analysis was carried out with a 60 cm polyethyleneimine coated capillary (I.D. 50 μm, O.D. 360 μm and effective length 48 cm). Before each run, the capillary was preconditioned by flushing with 1% formic acid water solution for 5 min, which also served as the background electrolyte. A sample of the K174A reaction mixture was loaded by pressure injection for 1.5 psi × 5 s. The DMAT and compound X were eluted at 10.2 min and

9.4 min, respectively, when a -15 kV voltage was applied at 15 oC. Photodiode array detection gives the UV spectra of DMAT and compound X.

2.9.8 Synthesis of Compound X

The methyl ester 30 was synthesized according to the method described by Danishefsky et al.[83] LiOH was added to a solution of the methyl ester 30 (39 mg, 0.14 mmol) in a 1:1 mixture of MeOH and 1, 4-dioxane (8 mL). The mixture was stirred for 30 min before Amberlite 120H resin (washed with MeOH and dH2O) was added. The resin was then removed by filtration when the pH of the mixture reached neutrality and the filtrate was evaporated in vacuo to afford an approximately 80% pure sample of compound X as judged by 1H NMR spectroscopy. A 1H NMR spectrum was obtained in a deuterated phosphate buffer (pD 7.5) with the HOD signal referenced

1 to δ 4.7 ppm. H NMR (400 MHz, D2O) δ 7.38 (d, J = 7.6 Hz, 1H), 7.24 (t, J = 7.6 Hz, 1H), 6.93

(t, J = 7.5 Hz, 1H), 6.79 (d, J = 7.8 Hz, 1H), 6.01 (dd, J = 17.4, 10.9 Hz, 1H), 5.36 (s, 1H), 5.15

(d, J = 10.9 Hz, 1H), 5.10 (d, J = 17.4 Hz, 1H), 3.58 (dd, J = 11.2, 6.2 Hz, 1H), 2.53 (dd, J = 99

13.1, 6.3 Hz, 1H), 2.40 (dd, J = 15.0, 9.3 Hz, 1H), 1.09 (s, 3H), 1.00 (s, 3H). Negative ESI-MS

(M-H)-, 271.4.

2.9.9 Synthesis of 4-Aza-Tryptophan (47)

4-aza-Trp (47) was synthesized enzymatically with tryptophan synthase following a previously reported procedure (Figure 2.20).[94] A commercially available sample of E. coli pre- transformed with a plasmid expressing tryptophan synthase was ordered from ATCC. This sample was used to start a 10 mL Luria Broth (LB) culture containing 100 μg/mL ampicillin.

After grown at 37 oC for 12 hrs, 5 mL of this overnight culture was used to inoculate 1 L LB culture with 100 μg/mL ampicillin, the culture was allowed to grow at 37 oC for 12 h. The cells were pelleted and washed with 30 mL 0.1 M NaCl solution before the cells were resuspended in

Tris buffer (100 mM with 5 mM EDTA, 10 mM mercaptomethanol, 1 mM PMSF and 0.1 mM

PLP, pH 8.0) and lysed by French Press. The cell lysate was centrifuged to remove the cell debris and the supernatant was used as a synthetic agent.

Crude cell lysate (5 mL) was sealed in dialysis tubing (10, 000 MWCO) and transferred into a 100 mL phosphate buffer solution (0.1 M, pH 7.8) containing 4-azaindole (0.143 g, 1.21 mmol), L-serine (0.127 g, 1.21 mmol) and PLP (0.8 mg). The reaction was incubated in an orbital shaker (37 ºC, 180 rpm, 3 days). The reaction mixture was then extracted with ethyl acetate (3

× 30 mL) to remove any unreacted 4-azaindole. The aqueous layer was concentrated to 20 mL under reduced pressure before it was loaded onto a reversed-phase C18 column. Water was used 100 to elute L-serine and methanol was used to elute 4-aza-tryptophan (47). Fractions containing 47 were pooled together and evaporated to dryness. 34 mg of 47 was obtained. 1H NMR (400 MHz,

D2O) δ 8.32 (dd, J = 4.8, 1.2 Hz, 1H), 7.89 (dd, J = 8.2, 1.3 Hz, 1H), 7.49 (s, 1H), 7.23 (dd, J =

8.2, 4.8 Hz, 1H), 3.60 (dd, J = 7.7, 5.6 Hz, 1H), 3.22 (dd, J = 14.4, 5.6 Hz, 1H), 2.99 (dd, J =

14.5, 7.7 Hz, 1H). Negative ESI-MS (M-H)-, 204.2.

2.9.10 Synthesis of (S)-2-Amino-3-(1H-pyrrol-3-yl)propanoic acid (49)

The synthesis of substrate analog 49 was achieved a known procedure (Figure 2.25).[95]

101

Chapter 3. Mechanistic Studies on Cyclomarin/Cyclomarazine N-

prenyltransferase CymD

A version of this Chapter has been published and some of the experimental results are reproduced with permission from: Qian, Q.; Schultz, A. W.; Moore, B. S.; Tanner, M. E. Biochemistry 2012, 51, 7733-7739 (© 2012 American Chemical Society).

102

Introduction

The first-committed step in the biosynthesis of the antibacterial cyclomarazine A and the anti-inflammatory cyclomarin A is catalyzed by the N-prenyltransferase CymD, the first indole prenyltransferase characterized from bacteria.[62], [79] The reaction catalyzed by CymD involves a reverse prenylation of tryptophan at the N-1 position to produce N-dimethylallyltryptophan (N-

DMAT, Figure 3.1).

COOH COOH NH NH2 CymD 2 OPP + N N H DMAPP L-Trp N-DMAT

Figure 3.1 The prenylation reaction catalyzed by CymD.

The possible mechanisms involved in this enzyme reaction attracted our interest as the prenyl group is added onto an indole nitrogen that is very non-nucleophilic due to the participation of its lone pair in the aromaticity of the indole ring. In addition, no mechanistic studies had been reported on this enzyme prior to our work.

One potential mechanism that has been proposed for another N-prenyltransferase cTrpPT involves the dissociation of DMAPP to form a dimethylallyl carbocation, followed by nucleophilic attack from the indole nitrogen, and a final deprotonation to give the product (Figure

3.2, Path A).[98],[99] This mechanism is problematic because the indole nitrogen lone pair is non- nucleophilic and it has been well established that the C-3 and C-2 positions are the most 103 nucleophilic positions on an indole ring.[85-86] For example, reaction of indole and prenylbromide exclusively leads to production of 3-prenylindole and 2-prenylindole.[86] Therefore, it is questionable whether the enzyme can prevent nucleophilic attack from C-3 and C-2 of the indole ring once the dimethylallyl carbocation is generated in the enzyme active site.

DMAPP OPP COOH COOH COOH Path A/B NH NH NH2 Path A 2 2 B NH N N ammonium H H intermediate L-tryptophan A Path B -H+

COOH COOH COOH NH2 NH2 NH2 -H+ aza-Cope N N N H N-DMAT iminium intermediate

Figure 3.2 Two possible dissociative mechanisms proposed for CymD.

Another possible mechanism also involves the dissociation of DMAPP to form the dimethylallyl carbocation. In this case, however, we proposed that the initial nucleophilic attack occurs at the C-3 position of the indole ring to produce a normal prenylated iminium intermediate

(Figure 3.2, Path B). This step is reasonable given that the C-3 of indole is the more nucleophilic site. A subsequent deprotonation then occurs to produce a neutral normal prenylated imine. 104

Lastly, an aza-Cope rearrangement renders the reverse prenylated product N-DMAT. As a Cope rearrangement mechanism has been proposed for DMATS (Figure 2.14), a similar mechanism could be a real possibility for this enzyme as well, except that the rearrangement occurs from C-

3 to C-4 in the case of DMATS as opposed to occurring from C-3 to N-1 for CymD. Additionally, ample precedence has demonstrated that the aza-Cope rearrangement from C-3 to N-1 on indole ring is feasible in non-enzymatic conditions (see section 2.6).[16, 89] Taken together, we felt a rearrangement mechanism would address the question of how CymD can prenylate the non- nucleophilic indole nitrogen in such a regioselective manner.

We also considered a possible associative mechanism for CymD that involves an initial deprotonation at N-1 to produce an anionic nucleophile, which then attacks DMAPP in an SN2’ reaction to render the reverse N-prenylated product N-DMAT (Figure 3.3). The pKa of indole nitrogen is 17 in water and 21 in DMSO, so a deprotonation step is not an unreasonable proposition since there is ample precedence for enzymatic deprotonations of carbon acids with

[100],[101],[102] similar pKa. An associative mechanism has previously been proposed for the mechanism of protein farnesyltransferase in which an excellent nucleophile (a cysteine thiolate) is prenylated (see Figure 1.15).

anionic intermediate DMAPP COO COO COO PPO NH 3 NH3 NH3 N OPP N N -H+ H N-DMAT L-tryptophan

Figure 3.3 A possible associative mechanism for CymD. 105

In our previous studies on DMATS, we acquired evidence in support of a Cope rearrangement mechanism using fluorinated substrate analogs studies, PIX experiments and KIE experiments.[78],[103] We envisioned that we could use similar techniques in mechanistic studies of CymD.

This Chapter will describe our kinetic characterization of CymD and our attempts to elucidate the reaction mechanism.

Overexpression and Purification of CymD

A plasmid pHIS8-cymD encoding for His-tagged CymD was obtained from the Moore group as part of our collaboration. The protein production and purification was carried out using a protocol slightly modified from their published procedure.[62] The pHIS8-cymD plasmid was used to transform E. coli cells, then cell cultures were grown in an autoinduction medium, which does not require the addition of IPTG for the induction of protein overexpression. The E. coli cells were lysed in a French pressure cell, and the crude lysate was loaded onto a column of immobilized metal ion affinity resin that had been charged with NiSO4. The His-tagged CymD was eluted using a stepwise gradient of imidazole buffer. Typically, 10 mg of enzyme was purified from 500 mL of culture. ∼

Attempts were then made to remove the imidazole from the most concentrated enzyme fractions. All techniques used, including ultracentrifugation, dialysis, and size exclusion

106 chromatography, resulted in a significant loss of enzyme activity. Therefore, the enzyme containing imidazole was used directly in activity studies. For the enzyme reactions that were studied with 1H-NMR spectroscopy, the enzyme was dialyzed against phosphate buffer using 10 kD cutoff dialysis tubing to remove imidazole effectively.

Characterization of N-Prenyltransferase CymD Activity

3.3.1 Characterization of Activity Using 1H NMR Spectroscopy

The preliminary studies on the activity of CymD were monitored using 1H NMR spectroscopic analysis. The substrates L-tryptophan and DMAPP (7.5 mM and 5 mM, respectively) were incubated with the enzyme (imidazole removed by dialysis) in phosphate buffer at 37 oC. The reactions were usually completed after 4 h, based on the disappearance of the DMAPP proton signals. The enzyme was then removed by ultracentrifugation, the crude reaction mixture was lyophilized, and an 1H NMR spectrum was obtained after the residue was

1 dissolved in D2O. L-Tryptophan was used in excess to reduce the complexity in the H NMR spectrum of the reaction mixture. No purification of the product N-dimethylallyltryptophan (N-

DMAT) was attempted due to the fact that the reverse prenylation in the product will give rise to a characteristic set of proton signals. It was found that the observed proton signals agreed well with those previously reported for the 1H NMR spectrum of N-DMAT. [62] In particular, the characteristic proton signals of the reverse prenyl group appeared. These include the vinylic H- 107

2’ doublet of a doublet at δ 6.10 ppm, the two vinylic H-1’ doublets at δ 5.20 and 5.10 ppm and the singlet of the methyl groups H-4’ and H-5’ at δ 1.70 ppm (Figure 3.4). In a control reaction without enzyme, incubations of L-tryptophan and DMAPP did not produce any products.

COO β α 3 NH3 2 CH3-4'/5' N 1 Hb 1' 5' 2' Ha 4' N-DMAT HOD

aromatic H

H- α H-β H-2' Ha-1' Hb-1'

1 Figure 3.4 H NMR spectrum (400 MHz, D2O) of N-DMAT in the enzyme reaction mixture. Assignments of the characteristic proton signals from reverse prenylation are shown in the spectrum. Excess tryptophan is also present in the spectrum.

3.3.2 Steady-State Kinetic Characterization of CymD

As mentioned in Chapter 2 (section 2.2), a commercially available continuous coupled assay for pyrophosphate proved to be very useful in the determination of the rate of a prenyltransferase

108 reaction. The technique involves measuring the initial velocities at various substrate concentrations, when the concentration of the other substrate is held constant at a saturating level.

The kinetic parameters are determined from the fit of the initial velocities to Michaelis-Menten equation. The same assay was used in the kinetic studies of CymD.

With DMAPP as the variable substrate, a maximal rate was observed when the DMAPP concentration was at 2 μM. Further increasing the DMAPP concentration only displayed an inhibition effect on the initial velocity. Therefore, it was not possible to measure a KM value for

DMAPP as the assay was not sufficiently sensitive to measure initial velocities at lower concentrations.

L-Tryptophan was then used as the variable substrate when the DMAPP concentration was held constant at 20 μM. A hyperbolic curve was obtained and we were able to determine the

-1 kinetic constants for CymD; kcat = 0.10±0.02 s and KM,Trp = 4.6±0.7 μM (Figure 3.5). The reported error ranges were calculated from the non-linear least square fitting on results of kinetic trials carried out on the same day, with enzymes and substrates from the same batch. However, systematic error can occur due to factors such as use of different batches of enzymes and substrates, errors in concentration measurements, as well as temperature fluctuation. These typically lead to ~20% error when measurements were performed on different days. The kcat value found for CymD was 5 times lower than that of DMATs (Table 2.1). This observation could be attributed to DMAPP substrate inhibition, yet preliminary results only demonstrated a 2-fold decrease of reaction rate when the DMAPP concentration changed from 2 μM to 100 μM.

Alternatively, this slow reaction rate might have resulted from the poor nucleophilicity of the 109 indole nitrogen that gets prenylated in the reaction

0 10 20 30

0.08 0.08 ) -1 s (

0.04 0.04 Rate/[E]

0.00 0.00 0 10 20 30 [L-Tryptophan] (µM)

Figure 3.5 Kinetic plot for CymD with varying concentrations of L-tryptophan and 20 μM DMAPP.

Mechanistic Studies Using Fluorinated Substrate Analogs

3.4.1 Characterization of CymD Activity with Fluorinated Substrate Anaglogs

The use of fluorinated substrates has proven to be an effective method of probing enzyme- catalyzed reactions that involve the formation of cationic intermediates and transition states.[104]

We envisioned that fluorinated substrate analogs could be of help in the studies the CymD- catalyzed reaction as well. More specifically, two commercially available compounds D,L-4- fluorotryptophan (55, Figure 3.6) and D,L-6-fluorotryptophan (23, Figure 3.6) were acquired and studied in the enzyme-catalyzed reaction. The fluoro-substitution on either position of the indole 110 ring significantly decreases the nucleophilicity of the indole ring.[85] As a result, the reaction rate would be greatly hampered if the enzyme employs the two dissociative mechanisms (Figure 3.2) since both require the indole ring to act as a nucleophile and involve a buildup of positive charge on the indole ring in the transition state. If the enzyme employs the associative mechanism

(Figure 3.3), however, the fluoro-substitution can stabilize the anionic intermediate and potentially accelerate the reaction rate.

COOH COOH R R1 OPP 1 NH CymD NH2 + 2 o 37 C N R N R2 DMAPP 2 H

R1= F, R2= H, D,L-4-fluorotryptophan 55 R = H, R = F, D,L-6-fluorotryptophan 23 fluorinated N-DMAT 1 2

Figure 3.6 Two fluorinated substrate analogs studied in the enzyme reaction.

The enzyme activity with the fluorinated tryptophan analogs was first monitored by NMR spectroscopy using a protocol similar to the one used for the natural substrates. The enzyme reactions with the fluorinated analogs D,L-4-fluorotryptophan (55) and D,L-6-fluorotryptophan

(23) were allowed to incubate at 37 oC overnight with an increased amount of enzyme (Figure

3.6). After the enzyme was removed, the crude reaction mixture was lyophilized, and an 1H NMR spectrum was obtained after the residue was dissolved in D2O. With both compounds, the production of fluorinated N-DMAT was confirmed by the appearance of the proton signals arising from reverse prenylation. Also, a maximal value of 50% reaction was observed for both 111 compounds, suggesting that only one of the enantiomers (presumably the L-enantiomer) can serve as a substrate for CymD (Figure 3.6).

3.4.2 Kinetic Studies Using Fluorinated Substrates

Using the coupled enzyme assay, kinetic constants for both D,L-4-fluorotryptophan (55) and

D,L-6-fluorotryptophan (23) were determined by measuring the initial velocities at varying concentrations of the tryptophan analogs while keeping DMAPP concentration constant at 20

μM. Hyperbolic curves for both compounds are shown in Figure 3.7, and the determined constants are shown in Table 3.1. The KM values for compound 55 and 23 were determined to be

-1 3.8 μM and 2.6 μM, respectively, while the kcat values were determined to be 0.037 s and 0.011 s-1, respectively.

0 20 40 60 0 20 40 60 0.04 0.04 A B 0.010 0.010 ) ) -1 -1 s ( 0.02 0.02

Rate/[E] (s Rate/[E] 0.005 0.005 Rate/[E]

0.00 0.00 0.000 0.000 0 20 40 60 0 20 40 60 [4-F-DL-Tryptophan] (µM) [6-F-DL-Tryptophan] (µM)

Figure 3.7 Kinetics plots of the CymD reaction with fluorinated tryptophans 55 and 23. Figures A and B are for 55 and 23, respectively. DMAPP concentration was 20 μM.

112

Table 3.1 Kinetic constants obtained with the fluorinated tryptophans.

-1 Trp Substrate kcat (s ) KM (µM)

L-Trp 0.10 ± 0.02 a 4.7 ± 0.7

55 0.037 ± 0.002 3.8 ± 0.5

23 0.011 ± 0.001 2.6 ± 0.3 a The reported error ranges were calculated from the non-linear least square fitting on results of kinetic trials carried out on the same day. Results from different days with different batch of enzyme could vary up to 20%.

It has been demonstrated by Otero et al. that the nucleophilicity at each position of the indole ring is reflected by the corresponding proton affinity.[85] Using density function theory (DFT), proton affinities were caculated for each position of the indole ring in substituted indoles.

Therefore, we were able to obtain the nucleophilicities of the indole ring in compound 55 and

23 and a subset of the data presented in the computational study is shown in Table 3.2. Note that the proton affinity at the C-3 position of indoles is markedly higher than at any other position and this is consistent with experimental evidence demonstrating that electrophiles prefer to add to the C-3 position in solution.[86]

113

Table 3.2 Calculated proton affinities for substituted indoles (kcal/mol).a

compound N-1 C-3 C-4

indole 206.8 220.1 211.8

4-F-indole 202.0 215.9 197.5

6-F-indole 202.7 216.0 205.6

a Data was taken from reference.[85]

Furthermore, it was shown that fluoro-substitution at either the C-4 or C-6 position causes a greater than 4 kcal/mol decrease in the proton affinity at either the N-1 or C-3 position of the indole ring. The dissociative mechanisms shown in Paths A and B (Figure 3.2), require the addition of a carbocation to the N-1 and C-3 positions of the indole ring, respectively, and if these were operative the rates of reaction should be dramatically lowered by the fluorinations.

Such an effect is observed in the case of the DMATs reaction where C-6 fluorination abolished the reactivity of 6-fluorotryptophan even though 6-methyltryptophan is accepted as an active substrate. This is consistent with the notion that C-6 fluorination reduces the proton affinity at

C-4 from 211.8 kcal/mol to 205.6 kcal/mol and at C-3 from 220.1 kcal/mol to 216.0 kcal/mol.[78],[85] Furthermore, PIX experiments with DMATS, [1-18O]-DMAPP, and 6- fluorotryptophan showed that isotopic scrambling proceeds to completion, indicating that substrate binding and carbocation formation still takes place, but the reduced nucleophilicity of the indole prevents the forward reaction from occurring. In the case of CymD, the KM values for compound 55 and 23 were comparable to that of the natural substrate, L-Trp, suggesting that 114 fluoro-substitution did not affect substrate binding. Therefore, the modest effects that fluoro- substitution had on the rate of catalysis (less than 10-fold decrease in the values of kcat) could be used as evidence that there is negligible positive charge build-up on the indole ring during catalysis. This argues against the dissociative mechanisms shown in path A and path B as both scenarios require significant positive charge accumulation on the indole ring upon nucleophilic attack.

3.4.3 Mechanistic Studies Using Fluorinated DMAPP Analogs

As discussed in Chapter 1, fluorinated analogs of isoprenyl diphosphate have long been used to distinguish between associative (SN2) mechanisms and dissociative (SN1) mechanisms in prenyltransferase-catalyzed reactions. The magnitude of the observed rate reductions have been evaluated as the relative reaction rate, krel, that describes the rate measured with the fluorinated compound divided by that measured with the non-fluorinated compound. Useful information is obtained by comparing the krel values measured for the enzyme reaction to the ones measured for the non-enyzmatic solvolysis of a corresponding methanesulfonate. To apply this test to the

CymD reaction, a fluorinated DMAPP analog E-F-DMAPP (56, Figure 3.8) was analyzed in this fashion. This compound was synthesized by another member of our group during studies on

DMATS using a slightly modified literature procedure.[105] The fluorination on DMAPP will greatly decrease the reaction rate if any carbocation character develops on the allylic carbon during catalysis. 115

As mentioned in Section 3.3.2, DMAPP caused an inhibition of the reaction rate when the concentration is higher than 2 μM and we were unable to run a full kinetic characterization on

DMAPP, restrained by the sensitivity of the coupled enzyme assay. Therefore, the relative rates were compared at the saturation concentrations of the diphosphates, 10 μM and 20 μM, with the tryptophan concentration held at 100 μM. For the E-F-DMAPP reaction, it was necessary to concentrate the CymD eluent 15-fold by ultracentrifugation through a membrane, and the final enzyme amount used was 30-fold greater than that used with DMAPP in order to obtain a measurable rate. Under otherwise identical conditions, the reaction rate with E-F-DMAPP was deceased by 100-fold when compared to that with DMAPP (Figure 3.8), rendering a krel value of

0.01.

OPP

F E-F-DMAPP 56

Conc.a 10 μM 20 μM

Rate/[E]

-1 VDMAPP/[E] (s ) 0.031 0.031

-1 -4 -4 VE-F-DMAPP/[E] (s ) 2.5×10 3.2×10

krel 0.008 0.010

Figure 3.8 E-F-DMAPP (56) and the kinetic rates measured for DMAPP and E-F-DMAPP. a In both cases [L-tryptophan] = 100 μM.

116

The first example of fluorinated substrate studies was performed by the Poulter group in their mechanistic analysis of farnesylpyrophosate synthase (FPPs). Solid evidence for a dissociative mechanism was gained when the relative reaction rate of fluorinated substrate was compared to that of the solvolysis of corresponding methanesulfonate (known to proceed via an

[36] rel -4 SN1 mechanism). A k of 3.7 × 10 was determined for 3-CH2F-GPP (9) in the enzyme- catalyzed reaction, which agrees well with that for the solvolysis of its corresponding methanesulfonate 12 (krel = 7.7 × 10-4). This was forwarded as evidence for a carbocation intermediate in the FPPs reaction. Later on, both the Poulter and Fierke groups used fluorinated

FPP analogs in mechanistic studies of protein farnesyltransferase (PFTase) and the latter group

rel -2 [44, 46] determined a k value of 1.8 × 10 for 3-CH2F-FPP (15). This value agrees well with the

rel -2 k value of 6.1 × 10 determined for the azide-assisted SN2 reaction of the corresponding methanesulfonate. This was forwarded as evidence for an associative mechanism with PFTase.

The same idea was utilized again in probing the mechanism for DMATS, however, the results were less clear cut. A krel value of 1.1 × 10-2 was measured when comparing E-F-DMAPP to

DMAPP and this was compared to a krel value of 1.5 × 10-3 for the unimolecular solvolysis rates

[77] of the corresponding methanesulfonates. The datas were interpreted as supporting an SN1 mechanism for catalysis. In subsequent studies, strong evidence has been presented for the existence of a dimethylallyl carbocation intermediate in the DMATS reaction, indicating that a

rel -2 [78] k value of 1.1 × 10 is compatible with a dissociative (SN1) mechanism.

117

rel Table 3.3 The k values of monofluorinated analogs from different SN1 or SN2 reactions.

Mechanism Reaction krel (SN1 or SN2)

-4 Solvolysis of geranyl methanesulfonate 7.7 × 10 SN1

-4 3-CH2F-GPP (9) in FPPs reaction 3.7 × 10 SN1

-2 Farnesyl methanesulfonate with azide 6.1 × 10 SN2

-2 3-CH2F-FPP (15) in PFTase reaction 1.8 × 10 SN2

-2 Solvolysis of dimethylallyl methanesulfonate 1.1 × 10 SN1

-3 E-F-DMAPP in DMATS reaction 1.5 × 10 SN1

E-F-DMAPP in CymD reaction 1.0 × 10-2 ?

The observed krel value of 1.0 × 10-2 for E-F-DMAPP in the CymD reaction is consistent with an associative mechanism involving an ‘exploded’ transition state that bears considerable carbocation character as in the case of PFTase. It is also, however, compatible with a dissociative mechanism involving a discrete carbocation intermediate. Therefore, more mechanistic studies needed to be carried out in order to deduce the mechanism for CymD.

Positional Isotope Exchange Experiments on CymD

3.5.1 PIX Experiment with L-Tryptophan

Positional isotope exchange (PIX) experiments have been used to provide evidence for the 118 reversible formation of a discrete dimethylallyl carbocation intermediate using 18O-labeled

DMAPP.[81b] In the event of a reversible C-O bond breakage, the 18O label would have the chance to scramble from a bridging position in the starting material to a non-bridging position in the recovered DMAPP. We envisioned that PIX experiments could be utilized to probe the possibilities of Path A and B proposed in the dissociative mechanisms (Figure 3.2). The reversible formation of the allylic carbocation/pyrophosphate ion pair could lead to PIX if either dissociative mechanism were operative for CymD (Figure 3.9).

COOH COOH L-Trp NH NH2 2 dissociation N N H H O O O P OP O P OP O O

[1-18O]-DMAPP] 18O-bridging PIX reaction

O COOH O P OP NH2 L-Trp + O

recovered DMAPP N 18O-non-bridging N-DMAT

Figure 3.9 Proposed PIX if a dissociative mechanism were operative for CymD. The red circle denotes the 18O label.

In the PIX experiment for CymD, a sample containing both unlabeled DMAPP and [1-18O]-

DMAPP (30 μM with 56% 18O incorporation) was incubated with tryptophan (20 μM) and

CymD and the reaction progress was monitored using 31P NMR spectroscopy. After the reaction was quenched, the remaining DMAPP was analyzed using 31P NMR spectroscopy. The 119 integration of the pyrophosphate signals indicated that 41% of the DMAPP was consumed in the incubation. Similar to the PIX eperiments for DMATS and the DMATS mutants discussed in

Section 2.3, a chelating resin was added to the NMR sample to remove residual amounts of divalent cation prior to the spectral acquisition. Also the acquisition parameters of the spectrometer were adjusted accordingly. These were necessary steps in order to increase the resolution of the 31P NMR spectra.

To our surprise, the two α-phosphorus doublets representing the unlabeled and [1-18O]-

DMAPP (20) (at -9.28 and -9.30 ppm, respectively) remained unchanged and no formation of the non-bridging [α-18O]-DMAPP (doublet at -9.31 ppm) was observed (less than 3%, Figure

3.10). As a result, no significant PIX was detected in the L-tryptophan reaction.

DMAPP [1-18O]-DMAPP O O 20 O P OP O P OP O O Before PIX experiment

L-Trp rxn

Figure 3.10 31P NMR spectra showing the α–phosphorus signals for the PIX experiment with L-Trp. Top spectrum are acquired before addition of enzyme and the bottom spectrum shows remaining DMAPP after 41% reaction. The red circle represents the 18O label.

120

3.5.2 PIX Experiments on Fluorinated Tryptophan

It has been shown that fluorinated tryptophan can be used to induce more extensive isotopic scrambling. In the studies on DMATS, complete isotopic scrambling was observed when the PIX experiment was carried out with D,L-6-fluorotryptophan (23).[78] Therefore, we proposed to carry out PIX experiments on D,L-4-fluorotryptophan (55) and D,L-6-fluorotryptophan (23). The fluorinated tryptophans are worse nucleophiles, thus they provide a greater chance for the isotopic scrambling to occur.

The PIX experiment with 55 and 23 were done in a similar manner as that with L-Trp.

However, some modifications were made to facilitate the use of these two analogs. In the previous experiment, because L-tryptophan readily reacts to give N-DMAT, only 0.7 equivalents of L-tryptophan was used in the reaction. This guarantees that the concentration of the recovered

DMAPP will not be too low to provide a well-resolved 31P NMR spectrum. Since the fluorinated analogs are racemic mixtures and activity studies revealed that only one enantiomer can serve as substrate for CymD, the concentrations of the analogs were increased to 1.4 equivalents. Lastly, the amount of enzyme was increased by 3-fold.

The use of D,L-4-fluorotryptophan (55) in the PIX reaction showed a significant amount of isotope scrambling after 59 % of the DMAPP had been consumed. Prior to enzyme addition, the

α-phosphorus signals of the unlabeled DMAPP and [1-18O]-DMAPP (20) appeared as two separate doublets at δ -9.28 and -9.29 ppm in the 31P NMR spectrum, respectively (Figure 3.11).

After incubation with CymD, the new upfield doublet at δ -9.31 ppm that represents [α-18O]- 121

DMAPP was observed.

DMAPP [1-18O]-DMAPP O O 20 before rxn O P OP O P OP O O

4-F-Trp rxn O PO P O O α-18 [ O]-DMAPP

ppm -9.2 -9.3 -9.4

Figure 3.11 31P NMR spectra showing the α–phosphorus signals for the PIX experiment with 4-F-Trp (55). The top spectrum was acquired before addition of enzyme and the bottom spectrum was acquired after 59 % DMAPP consumption. The red circle represents the 18O label.

The extent of PIX was calculated using the method previously discussed in section 2.3.3.

The amount of 18O-non-bridging DMAPP product was calculated to be 44% based on peak integration of the α–phosphorus signals, and the percentage of PIX was determined to be 66% in the reaction. Therefore, the partitioning ratio (υPIX/υrxn) of the reaction was estimated to be 1.1, suggesting that the probability of the formed allylic carbocation/pyrophosphate ion pair partitioning forward roughly equals that of its collapse backwards to reform DMAPP.

The use of D,L-6-fluorotryptophan (23) in the PIX reaction also demonstrated isotopic scrambling in the reaction, although to a much lower extent. After 72% of the DMAPP had been consumed in the reaction, the new upfield doublet corresponding to [α-18O]-DMAPP was observed at δ -9.31 ppm (Figure 3.12). The amount of 18O-non-bridging DMAPP was only 122 approximated as 10% and could not be accurately integrated. Therefore, the partitioning ratio

(υPIX/υrxn) of the reaction was not determined.

DMAPP [1-18O]-DMAPP O O 20 O P OP O P OP O O Before PIX experiment

α-18 6-F-Trp rxn [ O]-DMAPP O O P OP O

ppm -9.2 -9.3 -9.4 Figure 3.12 31P NMR spectra showing the α–phosphorus signals for the PIX experiment with 6-F-Trp (23). The top spectrum was acquired before addition of enzyme and the bottom spectrum was acquired after 72% DMAPP consumption. The red circle epresents the 18O label.

In past studies, the use of PIX experiments has proven to be informative with a metal- independent prenyltransferase. In the case of prenyltransferase DMATS, the observed isotope scrambling provided strong support for the existence of the dimethylallylcation intermediate.[78]

The observed isotopic scrambling with the fluorinated tryptophans proves that cleavage of the

C-O bond in DMAPP is reversible during CymD catalysis. This suggests that an allylic carbocation/pyrophosphate ion pair is formed that can either proceed forward towards product or collapse to reform DMAPP. Even though such scrambling was not observed with the natural substrate L-tryptophan, it is very unlikely that different mechanisms could be involved, given the similarities in rates between the reactions of the fluorinated and unsubstituted tryptophans.

We therefore conclude that CymD catalysis proceeds through a carbocation intermediate in all 123 three cases. The scrambling was presumably only observed in the case of the fluorinated tryptophans because of the poorer nucleophilicity of the fluorinated indoles.

Solvent Kinetic Isotope Effect Study on CymD

3.6.1 Introduction

A commonly used method to gain mechanistic insights on enzyme mechanism is a kinetic isotope effect (KIE) determination. KIE refers to the change of rate constants in a chemical reaction when a lighter isotope of substrate is replaced by a heavier isotope.[106],[107] Such a change in the reaction rate constants is only observed when bonding changes at the site of isotopic substitution are involved in the rate-determining step. Therefore, KIE analysis can provide important information on the rate-determining step in enzyme catalysis.

Theoretically, a KIE can be observed with any isotopic change in the substrate. For the ease of understanding, only KIEs involving a change from hydrogen (H) to deuterium (D) will be discussed here. The KIE involving hydrogen and deuterium is generally represented as Dk = kH/kD, where kH and kD are the reaction rate constants for the substrates containing the corresponding isotopes. The magnitude of the KIEs depends on the change of activation energy when the isotope is replaced.

Generally, KIEs can be classified as either primary or secondary. Primary KIEs are expected when the C-H/C-D bonds are broken or formed in a chemical reaction. Secondary KIEs are 124 expected when the bond cleavage occurs on the carbon that the isotopic atoms (H/D) are attached to. In an enzymatic reaction, Dk represents the intrinsic isotope effect on the isotope-sensitive step, therefore, it is rarely measurable unless this step is cleanly rate-limiting. Most enzymatic reactions involve multi-step reactions with several partially rate-determining steps and it is difficult to measure intrinsic KIEs. They are generally governed by the Michaelis-Menten equation:

VSmax [ ] υ = Equation (3.1) KSM +[ ]

(υ is reaction rate, Vmax is the maximum velocity, [S] is the substrate concentration and KM is the

Michaelis constant of the substrate). The kinetic constants, Vmax and Vmax/KM, are independent of substrate concentration and therefore could be used to measure the KIE for an enzyme reaction:

VH, KIE = max Equation (3.2) VDmax ,

(V / KH ), or KIE = max M Equation (3.3) (Vmax / KDM ),

To determine the KIE for an enzymatic reaction, several methods could be used. The simplest way is by comparing the rate versus substrate concentration plots with labeled and unlabeled substrates, with any other substrates kept at fixed concentrations. Kinetic constants,

Vmax and Vmax/KM, could then be determined to measure the KIE. Equation 3.2 will be used in our studies.

A solvent kinetic isotope effect describes the effect of an isotopically substituted solvent on

125 rate constants of enzymatic reactions.[106] This effect can occur, for instance, when the solvent

H2O is exchanged for D2O. In many cases, a solvent KIE could arise from solvent molecules stabilizing the transition state in the reaction or from deprotonation of a water molecule.

However, it could also result from the proton exchange between solvent and substrate prior to enzyme reaction, which effectively produces a deuterium-labeled substrate.

3.6.2 Application of A Solvent Kinetic Isotope Effect Study on CymD

In any of the three proposed mechanisms (Figure 3.2 and 3.3), the indole N-H bond must be cleaved at some point during catalysis. This proton has a pKa of 17, very close to that of H2O

(pKa =15.7), which indicates that it will rapidly exchange with solvent protons. Therefore, incubation of tryptophan in D2O will quickly generate the N-1 deuterated tryptophan analog.

Hence, it is possible to measure a solvent kinetic isotope effect, which can be used to probe whether the deprotonation step is a rate-determining step of the catalysis. Such a solvent kinetic isotope effect would only be expected when the deprotonation step is rate-determining.

The same coupled enzyme assay was used in the determination of enzyme reaction rates in

D2O yet some alterations had to be made. A sample of 20× buffer was divided into two portions and one portion was lyophilized to dryness and resuspended in an equal volume of D2O to produce the deuterated buffer. Assay solutions were then prepared with either the deuterated or nondeuterated buffer. Also note that the 2-amino-6-mercapto-7-methylpurine ribonucleoside

(MESG), purine nucleoside phosphorylase (PNPase), inorganic pyrophosphatase (PPase) used 126 in the D2O assays all had to be prepared in D2O. In both assays, the DMAPP concentration was held constant at 20 μM while the tryptophan concentration was variable. The amount of PPase used in both assays was decreased to 0.1 unit, yet it was confirmed that the enzymatic reactions were still effectively coupled.

Hyperbolic curves were plotted by fitting the initial velocities to the Michaelis-Menten

-1 equation (Figure 3.13). From the plots, the kcat values were determined to be 0.062 s and 0.026

-1 s for the reactions in H2O and D2O, respectively. Therefore, a primary solvent kinetic isotope of 2.3 (on the value of kcat) was observed.

0 10 20 30 0.06 0.06

0.04 0.04 ) -1 s (

Rate/[E] 0.02 0.02

0.00 0.00 0 10 20 30 [L-Tryptophan] (µM)

Figure 3.13 Kinetic plots of rate/[E] vs. substrate concentration demonstrating a solvent kinetic isotope effect. Tryptophan was the variable substrate and saturating DMAPP (20 μM) was employed. Red triangles represent data obtained in H2O and blue squares represent data obtained in D2O. Data were fit to the Michaelis-Menten equation (dashed line is for H2O and solid line is for D2O).

The observation of a primary solvent kinetic isotope effect suggests that a proton transfer is a rate-determining step of catalysis. This very likely corresponds to the deprotonation of the 127 indole N-H (pKa = 17), which generates a reasonably strong base and might be expected to constitute a rate-determining step of catalysis. The deprotonation steps in the dissociative mechanisms are all likely to be non rate-determining due to the relatively high acidity of the ammonium intermediate and the iminium intermediate in Paths A and B, respectively (Figure

3.2). Therefore, the primary solvent kinetic isotope effect is consistent with the notion that CymD might employ the associative mechanism proposed in Figure 3.3.

Summary

The recent discovery of a large family of indole prenyltransferases raises interesting questions as to the similarities and differences in the mechanisms employed in installing the prenyl group at the various positions around the indole ring. The reaction catalyzed by CymD attracted our interest as a very non-nucleophilic indole N-1 position must be alkylated during catalysis. In addition, our studies on another indole prenyltransferase DMATs suggested that a

Cope rearrangement mechanism could be involved during prenylation at the C-4 position (Figure

2.12). This prompted the notion that CymD may employ a pericyclic rearrangement for the prenylation of the indole nitrogen (Figure 3.2, Path B). In order to probe the mechanism that is involved in the CymD-catalyzed reaction, we carried out a series of studies with purified His- tagged enzyme.

Steady-state kinetics were carried out on L-tryptophan, D,L-4-fluorotryptophan (55) and

D,L-6-fluorotryptophan (23) in order to distinguish between associative (SN2) mechanism and dissociative (SN1) mechanisms. They led to the discovery that fluoro-substitution only had a 128 moderate effect on the rate of catalysis (less than 10-fold decrease in the values of kcat). This is in contrast to the much larger effect with 23 observed in the studies of DMATS. The moderate effects that 55 and 23 had with CymD were then interpreted as evidence that there is no significant positive charge accumulation on the indole ring during catalysis. Hence, this finding argues against the dissociative mechanisms shown in Path A and B where significant positive charge accumulates on the indole ring.

In kinetics studies on the fluorinated DMAPP analog E-F-DMAPP (56), a krel value of 1.0 ×

10-2 was determined when compared to DMAPP. While this is consistent with an associative mechanism involving an ‘exploded’ transition state bearing considerable carbocation character as in the case of PFTase, it is also compatible with a dissociative mechanism involving a discrete carbocation intermediate as in the case of DMATs.

Positional isotope exchange experiments were also carried out on L-tryptophan, D,L-4- fluorotryptophan (55) and D,L-6-fluorotryptophan (23). While no isotopic scrambling was observed in the reaction with L-tryptophan, a significant amount of isotopic scrambling was observed with 55 and a smaller amount was observed with 23. These observations indicates that a discrete allylic carbocation intermediate was generated during catalysis, arguing against an associative (SN2) mechanism.

Finally, solvent kinetic isotope effect studies on CymD presented a primary KIE of 2.3, indicating that the deprotonation of the indole N-H is a rate-determining step during catalysis.

This observation argues against a dissociative (SN1) mechanism for CymD as the deprotonations in such mechanisms are not likely to be rate-determining. Instead, it agrees well with an 129 associative (SN2) mechanism involving the deprotonation of the indole N-H as the rate- determining step.

With all the mechanistic studies carried out on CymD, enough evidence was obtained to enable us to propose a hybrid mechanism (Figure 3.14). An initial ionization of DMAPP forms an allylic carbocation/pyrophosphate ion pair. This is consistent with the observations of isotopic scrambling in the PIX experiments. A subsequent deprotonation of the indole N-H promotes the attack on the allylic carbocation. This is consistent with the studies on the fluorinated analogs and the solvent KIE result. The deprotonation could be achieved by an active site residue, either prior to the attack to form an indole anionic intermediate, or in concert with the C-N bond formation. In this mechanism, generation of a positively charged indolinium is circumvented in order to explain the insensitivity of the reaction rate upon fluoro-substitution on the indole ring.

In addition, the deprotonation of a relatively non-acidic proton explains the observation of a solvent kinetic isotope effect.

L COO COO -tryptophan COO NH3 NH3 NH3 N N H N DMAPP H OPP B PPi N-DMAT

Figure 3.14 Proposed hybrid mechanism for the reaction catalyzed by CymD.

This mechanism helps to explain how the poorly nucleophilic nitrogen gets prenylated in 130 such a regioselective manner. While it is not possible to rule out a similar mechanism involving a deprotonation-assisted prenylation at C-3 followed by an aza-Cope rearrangement (not shown), we feel that the barrier to the rearrangement would likely be high enough to be rate-determining and mask the solvent KIE.

Conclusion

It remains to be testified whether other indole prenyltransferases that prenylate at N-1 utilize the similar mechanism or involve a proposed aza-Cope rearrangement. To further probe our proposed hybrid mechanism for the N-1 prenylation, it will be interesting to mutate the basic residue that presumably deprotonates the indole N-1 position to a less basic residue or a non- polar residue. Such mutations will likely increase the energy barrier of the deprotonation step and isotopic scrambling may be detected with L-Trp in the mutant-catalyzed reactions. Since a conserved Glu that might perform this deprotonation is not readily apparent from sequence alignments, this would require solving the crystal structure of CymD.

It is also possible that other enzymes from the indole prenyltransferase family might employ similar mechanisms when a prenylation on N-1 occurs. One such prenyltransferase, FtmPT2 from Aspergillus fumigatus, catalyzes the N-1 prenylation of 12,13-dihydroxyfumitremorgin C during the last step in the biosynthesis of fumitremorgin B (Figure 3.15).[63] Another recently identified indole prenyltransferase, cTrpPT from Aspergillus oryzae, is capable of catalyzing prenylation on both the N-1 and the C-7 positions of cyclo-L-Trp-L-Trp to produce a N-1 131

‘reverse’ prenylated compound 57 and C-7 ‘normal’ prenylated compound 58 (Figure 3.15)[99]

It would be interesting to test our hypothesized mechanism on these two enzymes, using similar techniques.

OH O OH O OH MeO OH MeO N N FtmPT2 N N N N H H H O O

12,13-dihydroxyfumitremorgin C fumitremorgin B O H N NH O HN H N N NH cTrpPT O 57 HN + N O H H O N NH L-Trp-L-Trp cyclo- N HN H O 58

Figure 3.15 Reactions catalyzed by N-prenyltransferases FtmPT2 and cTrpPT.

Even though our studies on CymD argues against a hypothesized rearrangement mechanism,

I feel that the proposition of the rearrangement as a common step in the prenylation of indole ring is still viable. Further studies on other prenyltransferase might provide additional evidence.

Prenyltransferase CpaD from Aspergillus sp, characterized by Walsh group, is involved in the biosynthesis of potent neurotoxin cyclopiazonic acid (CPA, Figure 3.16).[68] This enzyme

132 catalyzes a C-4 prenylation of tryptophan derivative cyclo-acetoacetyl-L-tryptophan (cAATrp) to generate β-CPA, a precursor for α-CPA.

O O O OH HN OH HN OH N H O CpaD H O O H N DMAPP N H H cAATrp β-CPA N α-CPA H Figure 3.16 Cyclopiazonic acid (CPA) biosynthesis pathway in Aspergillus sp.

This enzyme catalyzes a similar reaction as DMATS does, and it shares high sequence similarity with other prenyltransferases from the ABBA prenyltransferase family. Alignment studies showed that a lysine residue K177, presumably involved in the final deprotonation of the arenium intermediate, is reserved in CpaD. Therefore, it is reasonable to postulate that the mechanism involved in the CpaD-catalyzed reaction could be similar to that utilized by DMATS

(Figure 3.17). After the dissociation of DMAPP, an initial ‘reverse’ prenylation occurs on the C-

3 position of the indole ring produces an iminium intermediate. A subsequent Cope rearrangement interconverts this with an arenium intermediate. And a final deprotonation of the arenium intermediate by K177 producesβ-CPA. Hence, mechanistic studies on CpaD might provide further evidence for the rearrangement mechanism. In our mutagenesis studies of

DMATS, mutant K174A produced a C-3 ‘reverse’ prenylated tricyclic product. Mutation of the conserved lysine residue K177 in CpaD to alanine might provide a similar ‘reverse’ prenylated 133 compound (Figure 3.17).

H N K177 PPi 2 O O HN OH O HN OH H Cope O HN OH O N N O N H H cAATrp H

? K177A CpaD CpaD

O β-CPA

N OH N O H

Figure 3.17 Proposed mechanism for CpaD and its possible product from mutant K177A.

Experimental Procedures

3.9.1 Materials and General Methods

All reagents were purchased from Sigma-Aldrich or Alfa Aesar and used without further purification. D2O (99.9%) was purchased from Cambridge Isotope Laboratories, Inc. D,L-4- fluorotryptophan (55) and D,L-6-fluorotryptophan (23) were purchased from Acros Organics. [1-

18O]-DMAPP (20) was synthesized as described in Section 4.10.2. E-F-DMAPP (56) was obtained from previous studies in our group.[78]

134

Rosetta (DE3) pLysS E. coli cells were purchased from Novagen. Centrifugal filters (4 mL,

10 000 MWCO) were purchased from Millipore. Chelating SepharoseTM Fast Flow resin was purchased from GE Healthcare. Bradford assay dye was purchased from Bio-Rad. EnzChek®

Pyrophosphate Assay Kit was purchased from Life Technologies, Inc. Acryl-cuvettes for use in enzyme kinetic assays were from Sarstedt.

Protein concentrations were determined by Bradford assay using BSA as standard. The enzyme kinetic assays were conducted on Cary 300 UV-vis spectrophotometer with an attached temperature controller.

1H NMR spectra were obtained on the Bruker AV400inv spectrometer at a field strength of

400 MHz. Proton-decoupled 31P NMR spectra were recorded on the Bruker AV400 spectrometer at a field strength of 162 MHz.

3.9.2 Protein Purification

The plasmid pHIS8-cymD was obtained from the Moore group.[62] After transformation of the plasmid into Rosetta (DE3) pLysS E. coli cells, culture was grown at 37 oC in 500 mL of ZYP-

5052 autoinduction medium containing 50 μg/mL kanamycin for 16 h. Cells were harvested and lysed by French Press in buffer A (50 mM Tris-HCl, pH 7.5) containing 20 mM imidazole and

500 mM NaCl). The cell lysate was cleared by centrifugation at 8000 × g for 1 h before the supernatant was loaded onto a column of Chelating SepharoseTM Fast Flow resin (10 mL, loaded with 100 mM NiSO4 and then equilibrated with buffer A containing 500 mM NaCl). The column 135 was washed with wash buffer (first with buffer A containing 500 mM NaCl, then with buffer A containing 125 mM imidazole and 500 mM NaCl) and the enzyme was eluted with elution buffer

(buffer A containing 500 mM imidazole and 500 mM NaCl). Typically about 10 mg of enzyme was purified from 500 mL of culture. For the enzyme used in the activity studies, 10 mL enzyme eluents were dialyzed against 1 L buffer B (50 mM phosphate, pH 7.5) for three times.

3.9.3 Activity Assays of CymD on Tryptophan and Its Fluorinated Analogs

The substrates L-tryptophan and DMAPP (7.5 mM and 5 mM, respectively) were incubated with CymD in 10 mL buffer B at 37 oC for 4 hrs. The enzyme was then removed by ultracentrifugation using Amicon tubes (4 mL, 10 000 MWCO), and the crude reaction was

1 lyophilized and a H NMR spectrum was obtained after the residue was dissolved in D2O.

Typically, ~1 mg of enzyme was used for each assay.

For the activity assays with fluorinated tryptophan 55 and 23, the enzyme reactions were allowed for overnight incubation at 37 oC before the enzyme was removed. Typically, ~ 5mg of enzyme was used for each assay.

3.9.4 Kinetics Studies on CymD

Enzyme kinetics were measured using an EnzChek® Pyrophosphate Assay Kit modified

136 from a previously described procedure.[78] Solutions (final volume 990 μL in buffer A) containing 20 μM DMAPP, L-tryptophan (variable), 100 μM 2-amino-6-mercapto-7- methylpurine ribonucleoside (MESG), 1 U purine nucleoside phosphorylase (PNPase), 0.1 U inorganic pyrophosphatase (PPase) were prepared in quartz cuvettes and equilibrated for 5 min at 35 oC. The enzymatic reaction was then initiated by addition of 10 μL CymD eluent (4.8 μg) and the initial velocity was calculated from the observed increase of absorption at 360 nm (using

ε= 11,000 M-1 cm-1).[108] Kinetic parameters were determined from the fit of the initial velocities to the Michaelis-Menten equation using the software OriginPro 8.

The kinetics for fluorinated tryptophan 55 and 23 were determined in the same manner with the exception that the concentration of the racemic mixture was doubled and the enzyme used per assay for 55 and 23 were 20 μL containing 6.8 μg and 9.6 μg enzyme, respectively.

The kinetics for fluorinated DMAPP E-F-DMAPP (56) were measured with DMAPP and E-

F-DMAPP concentrations at 10, 20 and 100 μM while the tryptophan concentration was held constant at 100 μM. The enzyme concentrations used for the DMAPP and E-F-DMAPP assay were 7.7 μg and 230 μg, respectively.

3.9.5 Positional Isotope Exchange Studies on CymD

PIX experiments were carried out based on a previously reported procedure.[78] A solution containing 1-[18O]-DMAPP and unlabeled DMAPP (total concentration of 30 mM in 1.0 mL with 56% 18O incorporation) and L-tryptophan (20 mM) in Tris-HCl buffer (50 mM, pD 7.5, 137

31 prepared using D2O) was prepared and the P NMR spectrum was collected. A solution of CymD

(1.5 mg/mL in 0.5 mL of column eluent) was added and the mixture was incubated at 37 ºC for

4 h. A 1.0 mL sample of the solution was subjected to ultrafiltration (Amicon Ultra-4, 10000

MWCO, 5000 rpm for 15 min, 4 °C) in order to remove the enzyme. Chelex-100 resin (50 mg

+ of 100-200 mesh, Na -form, pre-rinsed with D2O) was added to the filtrate and the mixture was vortexed extensively. A second 31P NMR spectrum was then acquired. The proton-decoupled 31P

NMR spectra were recorded on a Bruker AV400inv spectrometer operating at a frequency 162

MHz. Acquisition parameters included a 2437 Hz (20 ppm) sweep width centered at -5 ppm with a 27 s acquisition time. Well-resolved spectra were achieved after 200 to 1000 scans. All the spectra were optimized using appodization with exponential and Gaussian functions to achieve higher resolution.

PIX experiments with fluorinated substrates were performed under identical conditions with the exception that the fluorinates tryptophan concentration was increased by 2-fold and three times the amount of CymD was used in each reaction. The value of vPIX/vrxn for 4- fluorotryptophan was calculated as described previously.

3.9.6 Solvent Kinetic Isotope Effect Measurement

Kinetics assays were run using a modification of the coupled enzyme assay. A sample of

20X buffer A (8.0 mL) was divided into two equal portions. One portion was lyophilized to dryness and resuspended in an equal volume of D2O to produce the deuterated buffer. Assay 138 solutions were prepared with either the deuterated or non-deuterated buffer A to reach a final volume of 990 μL that contain L-tryptophan (variable), DMAPP (20μM), 2-amino-6- mercapto-7-methylpurine ribonucleoside (100 μM), purine nucleoside phosphorylase (PNPase)

(0.1 units), inorganic pyrophosphatase (PPase) (0.1 units), all of which had been prepared in either the deuterated or non-deuterated buffers, respectively. These solutions were incubated at

35 oC for 5 min prior to the addition of CymD (4.8 ug in 10 μL non-deuterated column eluent).

The the initial velocity were determined by measuring the rate of UV absorption change at 360 nm using the Cary 300 UV-vis spectrometer.

139

Bibliography

[1] T. M. Kutchan, Plant Cell 1995, 7, 1059-1070. [2] P. J. Facchini, Annu. Rev. Plant Phys. 2001, 52, 29-66. [3] J. Ziegler, P. J. Facchini, Annu. Rev. Plant Biol. 2008, 59, 735-769. [4] C. L. Schardl, D. G. Panaccione, P. Tudzynski, Alkaloids Chem. Biol. 2006, 63, 45-86. [5] W. C. Evans, Trease and Evans pharmacognosy, 16th ed. 2009, 353-413. [6] T. Usui, M. Kondoh, C. B. Cui, T. Mayumi, H. Osada, Biochem. J. 1998, 333, 543-548. [7] S. E. O'Connor, J. J. Maresh, Nat. Prod. Rep. 2006, 23, 532-547. [8] J. C. Gebler, C. D. Poulter, Arch. Biochem. Biophys. 1992, 296, 308-313. [9] I. A. Unsold, S. M. Li, Microbiology 2005, 151, 1499-1505. [10] T. Kawasaki, K. Higuchi, Nat. Prod. Rep. 2005, 22, 761-793. [11] M. Somei, F. Yamada, Nat. Prod. Rep. 2005, 22, 73-103. [12] R. M. Williams, E. M. Stocking, J. F. Sanz-Cervera, Top Curr. Chem. 2000, 209, 97-173. [13] S. M. Li, Phytochemistry 2009, 70, 1746-1757. [14] T. Haarmann, Y. Rolke, S. Giesbert, P. Tudzynski, Mol. Plant Pathol. 2009, 10, 563-577. [15] J. E. Robbers, H. G. Floss, Arch. Biochem. Biophys. 1968, 126, 967-969. [16] H. G. Floss, Tetrahedron 1976, 32, 873-912. [17] S. L. Lee, H. G. Floss, P. Heinstein, Arch. Biochem. Biophys. 1976, 177, 84-94. [18] M. Shibuya, H. M. Chou, M. Fountoulakis, S. Hassam, S. U. Kim, K. Kobayashi, H. Otsuka, E. Rogalska, J. M. Cassady, H. G. Floss, J. Am. Chem. Soc. 1990, 112, 297-304. [19] H. F. Tsai, H. Wang, J. C. Gebler, C. D. Poulter, C. L. Schardl, Biochem. Biophys. Res. Commun. 1995, 216, 119-125. [20] P.-H. Liang, T.-P. Ko, A. H. J. Wang, Eur. J. Biochem. 2002, 269, 3339-3354. [21] P. H. Liang, Biochemistry 2009, 48, 6562-6570. [22] A. J. Chen, P. A. Kroon, C. D. Poulter, Prot. Sci. 1994, 3, 600-607. [23] K. Ogura, T. Koyama, Chem. Rev. 1998, 98, 1263-1276.

140

[24] C. D. Poulter, H. C. Rilling, Acc. Chem. Res. 1978, 11, 307-313. [25] J. K. Dorsey, J. A. Dorsey, J. W. Porter, J. Biol. Chem. 1966, 241, 5353-&. [26] P. W. Holloway, G. Popjak, Biochem. J. 1967, 104, 57-&. [27] F. Lynen, H. Eggerer, U. Henning, I. Kessel, Angew. Chem. Int. Ed. Engl. 1958, 70, 738- 742. [28] H. C. Rilling, K. Bloch, J. Biol. Chem. 1959, 234, 1424-1432. [29] Cornfort.Jw, Cornfort.Rh, G. Popjak, L. Yengoyan, J. Biol. Chem. 1966, 241, 3970-&. [30] G. Popjak, Cornfort.Jw, Biochem. J. 1966, 101, 553-&. [31] Cornfort.Jw, Angew. Chem. Int. Ed. Engl. 1968, 7, 903-&. [32] C. D. Poulter, D. M. Satterwhite, H. C. Rilling, J. Am. Chem. Soc. 1976, 98, 3376-3377. [33] E. T. Mcbee, O. R. Pierce, D. D. Smith, J. Am. Chem. Soc. 1954, 76, 3725-3728. [34] C. D. Poulter, E. A. Mash, J. C. Argyle, O. J. Muscio, H. C. Rilling, J. Am. Chem. Soc. 1979, 101, 6761-6763. [35] C. D. Poulter, J. C. Argyle, E. A. Mash, J. Am. Chem. Soc. 1977, 99, 957-959. [36] C. D. Poulter, P. L. Wiggins, A. T. Le, J. Am. Chem. Soc. 1981, 103, 3926-3927. [37] L. C. Tarshis, M. J. Yan, C. D. Poulter, J. C. Sacchettini, Biochemistry 1994, 33, 10871- 10877. [38] L. C. Tarshis, P. J. Proteau, B. A. Kellogg, J. C. Sacchettini, C. D. Poulter, Proc. Natl. Acad. Sci. USA 1996, 93, 15018-15023. [39] W. R. Schafer, J. Rine, Annu. Rev. Genet. 1992, 26, 209-237. [40] R. Roskoski, Biochem. Biophys. Res. Commun. 2003, 303, 1-7. [41] K. T. Lane, L. S. Beese, J. Lipid Res. 2006, 47, 681-699. [42] S. M. Sebti, A. D. Hamilton, Oncogene 2000, 19, 6584-6593. [43] F. Caponigro, Anti-cancer drugs 2002, 13, 891-897. [44] J. M. Dolence, C. D. Poulter, Proc. Natl. Acad. Sci. USA 1995, 92, 5008-5011. [45] C. M. Harris, C. D. Poulter, Nat. Prod. Rep. 2000, 17, 137-144. [46] C. Huang, K. E. Hightower, C. A. Fierke, Biochemistry 2000, 39, 2593-2602. [47] M. Tello, T. Kuzuyama, L. Heide, J. P. Noel, S. B. Richard, Cell Mol. Life Sci. 2008, 65, 1459-1463. [48] L. Heide, Curr. Opin. Chem. Biol. 2009, 13, 171-179. 141

[49] K. Yazaki, K. Sasaki, Y. Tsurumaru, Phytochemistry 2009, 70, 1739-1745. [50] R. Meganathan, Vitam. Horm. 2001, 61, 173-218. [51] H. Hemmi, K. Shibuya, Y. Takahashi, T. Nakayama, T. Nishino, J. Biol. Chem. 2004, 279, 50197-50203. [52] O. Saleh, Y. Haagen, K. Seeger, L. Heide, Phytochemistry 2009, 70, 1728-1738. [53] T. Bonitz, V. Alva, O. Saleh, A. N. Lupas, L. Heide, PloS One 2011, 6, e27336. [54] T. Kuzuyama, J. P. Noel, S. B. Richard, Nature 2005, 435, 983-987. [55] U. Metzger, S. Keller, C. E. Stevenson, L. Heide, D. M. Lawson, J. Mol. Biol. 2010, 404, 611-626. [56] U. Metzger, C. Schall, G. Zocher, I. Unsold, E. Stec, S. M. Li, L. Heide, T. Stehle, Proc. Natl. Acad. Sci. USA 2009, 106, 14309-14314. [57] F. Pojer, E. Wemakor, B. Kammerer, H. W. Chen, C. T. Walsh, S. M. Li, L. Heide, Proc. Natl. Acad. Sci. USA 2003, 100, 2316-2321. [58] T. Ozaki, S. Mishima, M. Nishiyama, T. Kuzuyama, J. Antibiot. 2009, 62, 385-392. [59] N. Steffan, A. Grundmann, W. B. Yin, A. Kremer, S. M. Li, Curr. Med. Chem. 2009, 16, 218-231. [60] X. Yu, S.-M. Li, Method Enzymol. 2012, 516, 259-278. [61] M. E. Tanner, Nat. Prod. Rep. 2015, 32, 88-101. [62] A. W. Schultz, C. A. Lewis, M. R. Luzung, P. S. Baran, B. S. Moore, J. Nat. Prod. 2010, 73, 373-377. [63] A. Grundmann, T. Kuznetsova, S. S. Afiyatullov, S.-M. Li, ChemBioChem 2008, 9, 2059- 2063. [64] I. A. Unsold, S. M. Li, ChemBioChem 2006, 7, 158-164. [65] J. M. Schuller, G. Zocher, M. Liebhold, X. Xie, M. Stahl, S.-M. Li, T. Stehle, J. Mol. Biol. 2012, 422, 87-99. [66] W.-B. Yin, A. Grundmann, J. Cheng, S.-M. Li, J. Biol. Chem. 2009, 284, 100-109. [67] Y. Ding, R. M. Williams, D. H. Sherman, J. Biol. Chem. 2008, 283, 16068-16076. [68] X. Liu, C. T. Walsh, Biochemistry 2009, 48, 11032-11044. [69] X. Yu, Y. Liu, X. Xie, X. D. Zheng, S. M. Li, J. Biol. Chem. 2012, 287, 1371-1380. [70] S. Takahashi, H. Takagi, A. Toyoda, M. Uramoto, T. Nogawa, M. Ueki, Y. Sakaki, H. Osada, J. Bacteriol. 2010, 192, 2839-2851. 142

[71] A. Kremer, L. Westrich, S. M. Li, Microbiology 2007, 153, 3409-3416. [72] N. Steffan, I. A. Unsold, S. M. Li, ChemBioChem 2007, 8, 1298-1307. [73] A. Kremer, S. M. Li, Appl Microbiol Biot 2008, 79, 951-961. [74] N. Steffan, S. M. Li, Arch. Microbiol. 2009, 191, 461-466. [75] X. Yu, X. L. Xie, S. M. Li, Appl Microbiol Biot 2011, 92, 737-748. [76] W. A. Cress, L. T. Chayet, H. C. Rilling, J. Biol. Chem. 1981, 256, 917-923. [77] J. C. Gebler, A. B. Woodside, C. D. Poulter, J. Am. Chem. Soc. 1992, 114, 7354-7360. [78] L. Y. Luk, M. E. Tanner, J. Am. Chem. Soc. 2009, 131, 13932-13933. [79] A. W. Schultz, D.-C. Oh, J. R. Carney, R. T. Williamson, D. W. Udwary, P. R. Jensen, S. J. Gould, W. Fenical, B. S. Moore, J. Am. Chem. Soc. 2008, 130, 4507-4516. [80] D. J. Edwards, W. H. Gerwick, J. Am. Chem. Soc. 2004, 126, 11432-11433. [81] aI. A. Rose, Adv. Enzymol. Relat. Areas Mol. Biol. 1979, 50, 361-395; bF. M. Raushel, J. J. Villafranca, CRC Crit. Rev. Biochem. 1988, 23, 1-26; cJ. J. Villafranca, Method Enzymol. 1989, 177, 390-403. [82] A. W. Sangster, K. L. Stuart, Chem. Rev. 1965, 65, 69-130. [83] aS. P. Marsden, K. M. Depew, S. J. Danishefsky, J. Am. Chem. Soc. 1994, 116, 11143- 11144; bK. M. Depew, S. P. Marsden, D. Zatorska, A. Zatorski, W. G. Bornmann, S. J. Danishefsky, J. Am. Chem. Soc. 1999, 121, 11953-11963. [84] S. Lakhdar, M. Westermaier, F. Terrier, R. Goumont, T. Boubaker, A. R. Ofial, H. Mayr, J. Org. Chem. 2006, 71, 9088-9095. [85] N. Otero, M. Mandado, R. A. Mosquera, J. Phys. Chem. A 2007, 111, 5557-5562. [86] M. Westermaier, H. Mayr, Org. Lett. 2006, 8, 4791-4794. [87] E. Wenkert, H. Sliwa, Bioorg. Chem. 1977, 6, 443-452. [88] R. M. I. Beesley, C. K.; Thorpe, J. F. , J. Chem. Soc. Trans. 1915, 1080-1106. [89] N. Voute, D. Philp, A. M. Z. Slawin, N. J. Westwood, Org. Biomol. Chem. 2010, 8, 442- 450. [90] D. D. Schwarzer, P. J. Gritsch, T. Gaich, Angew. Chem. Int. Ed. Engl. 2012, 51, 11514- 11516. [91] N. Mahmoodi, Q. Qian, L. Y. P. Luk, M. E. Tanner, Pure Appl. Chem. 2013, 85, 1935- 1948. [92] J. D. Rudolf, H. Wang, C. D. Poulter, J. Am. Chem. Soc. 2013, 135, 1895-1902. 143

[93] K. Thandavamurthy, D. Sharma, S. K. Porwal, D. Ray, R. Viswanathan, J. Org. Chem. 2014, 79, 10049-10067. [94] R. J. M. Goss, P. L. A. Newill, Chem. Commun. 2006, 4924-4925. [95] J. E. Beecher, D. A. Tirrell, Tetrahedron Lett. 1998, 39, 3927-3930. [96] M. Jost, G. Zocher, S. Tarcz, M. Matuschek, X. Xie, S.-M. Li, T. Stehle, J. Am. Chem. Soc. 2010, 132, 17849-17858. [97] N. Mahmoodi, M. E. Tanner, ChemBioChem 2013, 14, 2029-2037. [98] Q. Qian, A. W. Schultz, B. S. Moore, M. E. Tanner, Biochemistry 2012, 51, 7733-7739. [99] H.-X. Zou, X.-L. Xie, U. Linne, X.-D. Zheng, S.-M. Li, Org. Biomol. Chem. 2010, 8, 3037-3044. [100] M. Balón, M. C. Carmona, M. A. Muñoz, J. Hidalgo, Tetrahedron 1989, 45, 7501-7504. [101] F. G. Bordwell, G. E. Drucker, H. E. Fried, J. Org. Chem. 1981, 46, 632-635. [102] J. P. Richard, T. L. Amyes, Curr. Opin. Chem. Biol. 2001, 5, 626-633. [103] L. Y. Luk, Q. Qian, M. E. Tanner, J. Am. Chem. Soc. 2011, 133, 12342-12345. [104] R. Pongdee, H.-w. Liu, Bioorg. Chem. 2004, 32, 393-437. [105] V. J. Davisson, A. B. Woodside, T. R. Neal, K. E. Stremler, M. Muehlbacher, C. D. Poulter, J. Org. Chem. 1986, 51, 4768-4779. [106] D. B. Northrop, Annu. Rev. Biochem. 1981, 50, 103-131. [107] W. W. Cleland, Arch. Biochem. Biophys. 2005, 433, 2-12. [108] M. R. Webb, Proc. Natl. Acad. Sci. USA 1992, 89, 4884-4887.

144