Investigation of the Biosynthesis of the Nucleoside Antibiotic Sphaerimicin

University of Kentucky UKnowledge

Theses and Dissertations--Pharmacy College of Pharmacy

2021

INVESTIGATION OF THE BIOSYNTHESIS OF THE NUCLEOSIDE ANTIBIOTIC SPHAERIMICIN

Jonathan Overbay University of Kentucky, [email protected] Digital Object Identifier: https://doi.org/10.13023/etd.2021.151

Right click to open a feedback form in a new tab to let us know how this document benefits ou.y

Recommended Citation Overbay, Jonathan, "INVESTIGATION OF THE BIOSYNTHESIS OF THE NUCLEOSIDE ANTIBIOTIC SPHAERIMICIN" (2021). Theses and Dissertations--Pharmacy. 126. https://uknowledge.uky.edu/pharmacy_etds/126

This Doctoral Dissertation is brought to you for free and open access by the College of Pharmacy at UKnowledge. It has been accepted for inclusion in Theses and Dissertations--Pharmacy by an authorized administrator of UKnowledge. For more information, please contact [email protected]. STUDENT AGREEMENT:

I represent that my thesis or dissertation and abstract are my original work. Proper attribution has been given to all outside sources. I understand that I am solely responsible for obtaining any needed copyright permissions. I have obtained needed written permission statement(s) from the owner(s) of each third-party copyrighted matter to be included in my work, allowing electronic distribution (if such use is not permitted by the fair use doctrine) which will be submitted to UKnowledge as Additional File.

I hereby grant to The University of Kentucky and its agents the irrevocable, non-exclusive, and royalty-free license to archive and make accessible my work in whole or in part in all forms of media, now or hereafter known. I agree that the document mentioned above may be made available immediately for worldwide access unless an embargo applies.

I retain all other ownership rights to the copyright of my work. I also retain the right to use in future works (such as articles or books) all or part of my work. I understand that I am free to register the copyright to my work.

REVIEW, APPROVAL AND ACCEPTANCE

The document mentioned above has been reviewed and accepted by the student’s advisor, on behalf of the advisory committee, and by the Director of Graduate Studies (DGS), on behalf of the program; we verify that this is the final, approved version of the student’s thesis including all changes required by the advisory committee. The undersigned agree to abide by the statements above.

Jonathan Overbay, Student

Dr. Steven G. Van Lanen, Major Professor

Dr. David Feola, Director of Graduate Studies INVESTIGATION OF THE BIOSYNTHESIS OF THE NUCLEOSIDE ANTIBIOTIC SPHAERIMICIN

______

DISSERTATION ______

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the College of Pharmacy at the University of Kentucky

By Jonathan W. Overbay Lexington, Kentucky Director: Dr. Steven G. Van Lanen, Professor of Pharmaceutical Sciences Lexington, Kentucky 2021

INVESTIGATION OF THE BIOSYNTHESIS OF THE NUCLEOSIDE ANTIBIOTIC SPHAERIMICIN

Antibiotic-resistance has become a widespread problem in the United States and across the globe. Meanwhile, new antibiotics are entering the clinic at an alarmingly low rate. Highly-modified nucleosides, a class of natural products often produced by actinobacteria, target MraY bacterial translocase I. MraY is a clinically unexploited enzyme target that is ubiquitous and essential to peptidoglycan cell wall biosynthesis. The nucleoside antibiotics known vary in efficacy and the functionalities contributing to improved activity is poorly understood. Sphaerimicin, a newly discovered modified nucleoside, has potent inhibitory activity with an IC50 of 13.65 nM against MraY. In general, sphaerimicin is primarily effective against gram-positive bacteria (MIC ranges from 2-16 μg/mL against Enterococcus faecium and Staphylococcus aureus), but little is known about the biosynthesis and mechanism of action. Sphaerimicin has highly unusual structural features, including a heavily modified ribosylated glycl-uridine disaccharide core that is appended to a dihydroxylated piperidine ring.

The novel biosynthesis of these features was investigated, leading to the functional characterization of six enzymes critical for the disaccharide core formation from uridine monophosphate. Recently, a unique S-adenosylmethionine- and PLP-dependent alpha- aminobutyric acid transferase and a nonheme Fe(II)- and alpha-ketogluterate-dependent hydroxylase from the sphaerimicin biosynthetic pathway were characterized. This part of the pathway extends the carbon scaffold, which will be crucial for formation of the piperidine ring. Not only does this study provide new chemical entities to help better understand MraY as a target, it could potentially reveal new enzymatic chemistries that will power innovative chemoenzymatic synthesis and genome mining to uncover new natural products.

KEYWORDS: Enzyme, Mechanism, Biosynthesis, Nucleoside, Antibiotic, PLP

Jonathan Overbay (Name of Student)

01/31/2021 Date

INVESTIGATION OF THE BIOSYNTHESIS OF THE NUCLEOSIDE ANTIBIOTIC SPHAERIMICIN

By Jonathan W. Overbay

Steven G. Van Lanen Director of Dissertation

David Feola Director of Graduate Studies

01/31/2021 Date

DEDICATION

To my family – my wife, mother, father, and everyone else who has supported me over the years. ACKNOWLEDGMENTS

I would like to thank my mentor, Dr. Steven G. Van Lanen, for his support. I would also like to thank my committee members: Drs. Jürgen Rohr, Jon Thorson, and Luke Moe, for their invaluable advice over the years.

A major thanks to my fellow lab members Drs. Zheng Cui, Minakshi Bhardwaj, Xiaodong Liu, Han Nguyen, Matt McErlean, and Erfu Yan, as well as Ashely Biecker, Jasmine Woods, and Hoda Saghaeiannejad-Esfahani.

I would also like to acknowledge my family and friends for their support and encouragement.

iii TABLE OF CONTENTS

ACKNOWLEDGMENTS ...... iii

LIST OF TABLES ...... vii

LIST OF FIGURES ...... viii

LIST OF ABBREVIATIONS ...... xiv

Relevancy of Studying Biosynthesis of Highly-Modified Nucleoside Antibiotics ...... 1 1.1 Antibiotic resistance ...... 1 1.2 Natural product biosynthesis ...... 2 1.3 MraY bacterial translocase I ...... 3 1.4 GlyU-ADR nucleoside antibiotics ...... 5 1.5 Sphaerimicin: the most recently discovered GlyU-ADR-containing natural product ...... 8 1.6 Contents of the dissertation ...... 10

Current Understanding of PLP Chemistries ...... 11 2.1 Introduction - pyridoxal 5'-phosphate function & structure ...... 11 2.2 Alanine racemase ...... 15 2.3 γ-addition and elimination catalysis ...... 18 2.3.1 Cystathionine gamma-synthases / lyases and LolC ...... 18 2.3.2 Alkyl transfer in nucleoside antibiotic biosynthesis ...... 20 2.3.3 Formation of alkyl-substituted pipecolate in citrinadin and fusaric acid biosynthesis ...... 21 2.3.4 The γ-elimination, α-addition (Claisen Condensation) catalysis of CqsA .... 24 2.3.5 The γ-elimination of 1-Aminocyclopropane-1-carboxylic Acid Synthases .. 25 2.3.6 PLP-dependent gamma-elimination in alkyne functional group biosynthesis 27 2.4 Beta-addition and -elimination catalysis ...... 28 2.4.1 Beta-replacement in lanthionine biosynthesis ...... 28 2.4.2 Beta-replacement in viomycin and staphyloferrin biosynthesis ...... 29 2.4.3 The β-elminination catalysis of serine dehydratase ...... 30 2.4.4 Beta-decarboxylation ...... 30 2.4.5 PLP-dependent Thiol Formation ...... 31

iv 2.5 The α-replacement catalysis of transaldolases in production of beta-hydroxy- alpha-amino acids ...... 32 2.6 PLP-dependent oxidation catalysis...... 32 2.7 Conclusion ...... 36

. Evidence of an Unusual Mechanism for an Aminobutyric Acid Transferase ...... 37 3.1 Introduction ...... 37 3.2 Materials and Methods ...... 39 3.2.1 General experimental procedures ...... 39 3.2.2 Cloning, overexpression, and purification of proteins ...... 40 3.2.3 Assay of SphL, LipJ, and Mur24 half-reactions ...... 42 3.2.4 Mur24 deuterium exchange assay ...... 43 3.2.5 Determining the regiochemistry of isotope exchange ...... 43 3.2.6 MTA product-inhibition assays ...... 44 3.2.7 Kinetics of Mur24 ...... 44 3.2.8 FPLC analysis of SphL and LipJ ...... 44 3.2.9 Protein modeling of SphL and LipJ ...... 45 3.3 Results and Discussion ...... 45 3.3.1 Tracking proton-exchanges throughout catalysis ...... 45 3.3.2 Rate constant determination and kinetic isotope effects ...... 48 3.4 Conclusion ...... 52 3.5 Supporting Information ...... 58

Downstream Biosynthesis of Sphaerimicin: Formation of the Dihydroxylated Piperidine Ring ...... 65 4.1 Introduction ...... 65 4.2 Materials and Methods ...... 65 4.2.1 General experimental procedures ...... 65 4.2.2 Cloning, overexpression, and purification of proteins ...... 66 4.2.3 Activity assay of SphK and SphN ...... 66 4.2.4 FPLC analysis of SphK and SphN ...... 66 4.3 Results and Discussion ...... 66 4.3.1 Functional assignment of SphK as a β-hydroxylase ...... 66 4.3.2 Functional assignment of SphN as a putative transketolase ...... 69 4.4 Conclusion ...... 70 4.5 Supporting Information ...... 71

References ...... 73

Vita ...... 81

LIST OF TABLES

Table 1.1 MIC values of sphaerimicin A against a range of bacterial strains...... 9 Table 1.2 IC50 values of several highly-modified nucleoside antibiotics against MraY. ...9 Table 3.1 Kinetic parameters and kinetic isotope effects determined with respect to five isotopologues. k* = kcat / Km **DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L-Met, concentrations were doubled in reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L-Met isotopologue)...... 51 Table 4.1 Binding motifs predicted in SphK based on a homologous protein (proline 3- hydroxylase; PDBe 1E5R)...... 68

SI Table 3.1 Primers used for cloning sphL...... 58 SI Table 3.2 LCMS gradient method used to detect 2-oxobutyrate-MBTH...... 58 SI Table 4.1 Primers used for cloning wildtype and mutant SphK expression, as well as SphN expression...... 71

vii

LIST OF FIGURES

Figure 1.1 Adapted representation of WHO data on number of new classes of antibiotics entering the clinic: the golden age of drug discovery at the peak – and a discovery void as of recent...... 2 Figure 1.2 Partial illustration of cell wall biosynthesis noting how MraY fits into the overall pathway. Also noted are a selection of current antibiotics used to clinically to treat bacterial infections by inhibiting downstream peptidoglycan biosynthesis on the cell surface...... 4 Figure 1.3 Select GlyU-ADR nucleoside antibiotics. Components include: GlyU-ADR core (blue), modified N-alkylamine linker (red), and additional tailoring (black)...... 6 Figure 1.4 Dioxygenase LipL activity and the phosphonate UMP analog used to probe the mechanism. Hydroxylation first occurs, followed by an oxidative dephosphorylation...... 7 Figure 1.5 Biosynthesis of GlyU-ADR shared among this subset of highly-modified nucleosides. Two canonical UMP molecules can be transformed into the GlyU-ADR core of these large nucleoside antibiotics using 6 enzymatic steps...... 8 Figure 2.1 Wild-type HDC activity compared to a critical mutation capable of modulating the PLP-depdendent chemistry that occurs...... 13 Figure 2.2 General aminotransferase mechanism. First, the alpha-position is deprotonated; this drives electrons into PLP, affording the quinonoid intermediate; next, water is utilized from the aqueous environment and the electron sink is again utilized to release the product...... 15 Figure 2.3 General decarboxylase mechanism. In this case, the electron sink stabilizes the carbonanion at the alpha-position; because of this stabilization, carbon dioxide can be eliminated from the reactant...... 15 Figure 2.4 PLP-depedent mechanism of alanine racemase depicted. Residues on opposing sides of the substrate allow interconversion between stereoisomers...... 18 Figure 2.5 CGS reaction, wherein homocysteine of OSHS is eliminated and used in the addition of cysteine...... 20 Figure 2.6 Proposed LolC activity. An analogous mechanism to CGS; however, a

nitrogen heteroatom is used to form a new Cγ-N bond...... 20

viii

Figure 2.7 CndF activity using PLP-dependent gamma-replacement chemistry to break a

Cγ-O bond and create a new Cγ-C bond...... 23

Figure 2.8 Fub7 uses PLP-dependent gamma-replacement to break a Cγ-O bond and create a Cγ-C bond; unlike CndF, catalysis uses an aldyhyde substituent via enolate reactivity...... 24 Figure 2.9 CqsA mechanism wherein AdoMet is utilized in a gamma-elimination, breaking a Cγ-S bond; this is followed by alpha-addition acyltranfer-like activity...... 25 Figure 2.10 ACC synthase mechanism wherein the alpha-carbon is deprotonated by a neighboring lysine residue, forming a ketimine intermediate. The electron movement is reversed, leading to the ketimine pi electrons to attack the gamma-position (thereby breaking the Cγ-S bond and forming the cyclopropane ring...... 26 Figure 2.11 BesB in the context of beta-ethynylserine biosynthesis in which the PLP- dependent gamma-elimination of chlorine leads to alkyne formation...... 27 Figure 2.12 VioB PLP-dependent activity, where beta-elimination is followed by beta- addition at using free ammonia nearby during catalysis...... 29

Figure 2.13 SbnA catalyzes beta-replacement chemistry, wherein a Cβ-O bond is broken

and a Cβ-N bond is created...... 30 Figure 2.14 PLP-dependent beta-replacement, wherein carbon dioxide is first released from the substrate...... 31 Figure 2.15 Transaldolase (LipK) mechanism constituting an alpha-elimination of acetaldehyde and addition of an amino acid, leading for the formation of GlyU...... 32 Figure 2.16 Proposed mechanism of RohP by Hedges et al. ACS Chem Biol 2018...... 34 Figure 2.17 Proposed mechanism of CuaB...... 35 Figure 2.18 Proposed mechanism of Cap15...... 35 Figure 3.1 Functional characterization of Mur24 led to the discover of a PLP-dependent alkyltransferase that performs γ-elimination and γ-addition in a single active-site...... 38 Figure 3.2 Mass spectra of varying L-Met isotopologues enzymatically converted to AdoMet via hMAT2A, subsequently converted to 2-oxobutyrate by Mur24, and lastly derivatized with MBTH...... 47

Figure 3.3 MS spectra of Mur24 half-reaction performed in D2O. Correcting for natural isotopic abundancies: 0D species, 25%; 1D species, 20%; 2D species, 27%; 3D species, 10%; 4D species, 6%; 5D species, 9%...... 48 Figure 3.4 Single turnover kinetics of ABTase half-reaction (n = 2). Reactions varying [Mur24] (●) included 50 mM KH2PO4, pH 8; 50 mM KCl; 10 mM MgCl2; 40 µM PLP; 500 µM AdoMet; and 240 nM, 480 nM, or 960 nM Mur24. Reactions varying [AdoMet] (■) included identical conditions, apart from 960 nM Mur24 and 125 nM, 250 nM, or 500 nM AdoMet. Slopes of the linear regressions were 1.2 and 0.96, respectively...... 50 Figure 3.5 Calculation of rate constant with respect to both Mur24 (α) and AdoMet (β) using the rate equation shown. The half-reaction was determined to be essentially 1st- order with respect to both α and β (the slope of each respective linear regression from Figure 3.3, indicating an overall 2nd-order reaction). The 2nd-order rate constant was calculated using this data...... 50 Figure 3.6 Michaelis-Menten kinetics of Mur24 using methionine isotopologues. Reactions included 50 mM KH2PO4, pH 8; 50 mM KCl; 10 mM MgCl2; 40 µM PLP; 2 mM ATP; 1 µM hMAT2A; 275 nM Mur24. [L-Met isotopologues] used to calculate Michaelis-Menten curve included 0 µM, 2.0 µM, 3.9 µM, 7.8 µM, 15.6 µM, 62.5 µM, and 125 µM. *DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L-Met, concentrations were doubled in reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L-Met isotopologue)...... 51 Figure 3.7 KIE under varying pH and temperature conditions. *DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L-Met, concentrations were doubled in reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L- Met isotopologue...... 52 Figure 3.8 Full catalytic cascade with multiple pathways proposed for Mur24: Since the enzyme does not copurify with PLP, we propose AdoMet covalently attaches to PLP forming the aldimine externally (I); as the AdoMet-PLP complex enters the active site, the oxyanion of PLP deprotonates Lys234 as it enters a partially hydrophobic pocket (II);

once AdoMet-PLP complex is in position, Cα is deprotonated (III) and the quinonoid state of PLP is short-lived as C4' picks up a proton from Lys234 (IV); in the presence of the acceptor substrate (9), a conformational change in the active site occurs positioning

the hydroxy-group of the pyridine away from the hydrophobic pocket and toward one of two arginine residues (V); this leads to the protonation of the ketimine, generating an iminium and oxyanion – consequently lowering the pKa of Hβ proton and allowing for the

deprotonation of Cβ (VI); this leads to the E configuration until MTA is eliminated and the s-trans configuration is generated (VII); when the acceptor substrate is present (9) an aza-Michael-like addition occurs (path a of VIII) and electron movement is reversed affording 10; when no acceptor substrate is present to generate the kinetically favorable product, the thermodynamically-favored s-cis product is formed (path b of VIII); the new orientation facilitates a 1,5-sigmatropic rearrangement to occur (IX) leading to the PLP- bound aminocrotonate (X); alternatively, a 1,5-prototropic rearrangement could occur in path b whereby C4' is deprotonated by a protein residue and a proton is picked up at Cγ from the aqueous environment or an active site residue...... 55 Figure 4.1 Substrates tested with SphK in the presence of 2-oxobutyrate, iron(II), and oxygen...... 67 Figure 4.2 HPLC trace of SphK activity: the sole substrate found active with SphK was nucleoside-intermediate 1A...... 68 Figure 4.3 Relative activity of alanine-mutant SphK proteins compared to wildtype...... 68 Figure 4.4 Canonical mechanism of a transketolase double-displacement reaction, wherein a 2-carbon donor covalently binds the thiamine pyrophosphate coenzyme (I); this 2-carbon unit will be added to the acceptor substrate (II). In this assay, the half- reaction measured using a real-time enzyme-coupled assay, whereby glyceraldehyde 3- phosphate generated as the product after xylulose 5-phosphate modification; this assay utilized catalytically-efficient, commercially-available triosephosphate isomerase and glycerophosphate dehydrogenase in order to detect the oxidation of NADH using UV-vis measurements at 340 nm...... 69 Figure 4.5 Relative activity of SphN based on UV-vis-based real-time enzyme-coupled assay: Initial velocity of SphN compared to a commercially-available positive-control E. coli transketolase (Left) and initial velocity of SphN using varying enzyme concentrations...... 70

Figure 4.6 Proposed biosynthetic pathway of sphaerimicin piperidine ring formation based on bioinformatics and preliminary biochemical/functional protein characterization...... 71

SI Figure 3.1 hMAT2A coupled reaction to make AdoMet in situ does not affect the initial velocity of ABTases when compared to using a commercial AdoMet stock...... 58 SI Figure 3.2 MBTH-derivatized product from HCl-catalyzed tautomerization of 2- oxobutyrate isotopologue formed from AdoMet-3,3,4,4-d4. After biocatalysis, one deuterium was lost to proton exchange in H2O. After HCl-catalyzed tautomerization (26% conversion after isotope distribution correction, assuming no β, γ-hydride shift), one additional deuterium was lost. This indicates that biocatalysis and HCl-catalyzed tautomerization each led to one proton being exchanged at the β-position throughout the alkyl-group modifications. (Note: L-Met isotopologue stocks used were 99% pure, accounting for the minor amounts of natural isotopic abundancies.) ...... 59 SI Figure 3.3 SDS-PAGE gel of IMAC-purified SphL and LipJ, both His-tagged recombinant proteins heterologously expressed in E. coli BL21 cells...... 59 SI Figure 3.4 Commercially available lysozyme, bovine serum albumin, and catalase were used to construct a standard curve (R2=0.9981, n = 2). The linear regression was used to approximate the native sizes of both SphL and LipJ (176 kDa and 210 kDa, respectively). Thi This roughly corresponds to a tetrameric complex for the quaternary structure of each enzyme...... 60 SI Figure 3.5 Following MBTH derivatization of 2-oxobutyrate, two stereoisomers are formed with distinct retention times (extracted ion chromatogram shown). For kinetic studies, total peak area was used in all calculations...... 60 SI Figure 3.6 Half-reaction of Mur24 in D2O significantly lowers the kcat (kcat(H) / kcat(D) = 2.4), suggesting that the active site is exposed to the aqueous environment throughout half-reaction catalysis and steps following MTA-elimination can perturb kinetics. Km could not be accurately determined from the data points collected...... 61 SI Figure 3.7 Michaelis-Menten kinetics comparing pure L-Met-3,3,4,4-d4 to a DL-Met- 3,3,4,4-d4 mix. *DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L-Met, concentrations were doubled in

xii

reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L-Met isotopologue). This demonstrates that a DL-Met-2-d1 mix as opposed to pure L-Met-2-d1 in kinetic studies has no unintended impact on KIE studies...... 61 SI Figure 3.8 Michaelis-Menten kinetics comparing L-Met to L-Se-Met (no statistically significant difference in any kinetic parameters present, suggesting elementary step of MTA leaving is not rate-limiting)...... 62 SI Figure 3.9 End-point assay of a 24-hour reaction with varying [Mur24]. This indicates that the enzyme is acting as a pseudo-substrate, whereby more enzyme will lead to more product, even once the reaction has reached equilibrium...... 62 SI Figure 3.10 Lineweaver-Burke plot of MTA product-inhibition. Considering standard error, the mode of inhibition looks most likely to be noncompetitive, indicative of MTA being the first product released during catalysis...... 63 SI Figure 3.11 Predicted SphL structure modelled after an ACC synthase (PDB: 1YNU) with PLP. Circles black is an alpha-helix in 1YNU that is not present in the model of SphL ...... 63 SI Figure 3.12 Binding pocket of SphL modelled using ACC synthase (PDB: 1YNU) with PLP. Most notable difference is the greater hydrophobicity in a region of SphL active-site compared to 1YNU...... 64 SI Figure 4.1 SDS-PAGE gel of IMAC-purified SphK, SphO, and SphN, both His-tagged recombinant proteins heterologously expressed in E. coli BL21 cells...... 72 SI Figure 4.2 Agilent 6120 Quadrupole and Agilent 6230 TOF MS spectra...... 72

xiii

LIST OF ABBREVIATIONS

α-KB α-ketobutyrate α-KG α-ketoglutarate ABTase Aminobutyric acid transferase ACC 1-aminocyclopropanecarboxylic acid ACN Acetonitrile AdoMet S-adenosyl-l-methionine Ado-Se-Met S-adenosyl-l-Se-methionine ADR 5'-amino-5'-deoxyribosyl ADR-GlylU (5''-amino-5''-deoxyribosyl)-5'-C-glycyluridine AQC 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate CBL Cystathionine β-lyase CBS Cystathionine β-synthase CDC Center for Disease Control CGL Cystathionine γ-lyase CGS Cystathionine γ-synthases FPLC Fast protein liquid chromatography GABA gamma-Aminobutyric acid GlyU 5'-C-glycyluridine HDC Histidine decarboxylase HRMS High-resolution mass spectrometry IC Inhibitory concentration IMAC Immobilized metal affinity chromatography IPTG Isopropyl β-D-1-thiogalactopyranoside KIE Kinetic isotope effect LC Liquid chromatography LIC Ligation-independent cloning

xiv

MBTH 3-methyl-2-benzothiazolinone hydrazone hydrochloride MGL Methionine gamma-lyase MIC Minimum inhibitory concentrations MS Mass spectrometry MTA Methylthioadenosine MWCO Molecular weight cut off NP Natural product NRPS nonribosomal peptide synthase OASS O-acetylserine sulfhydrylase OD Optical density OSHS O-succinyl-l-homoserine PCR Polymerase chain reaction PLP Pyridoxal 5'-phosphate SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel electrophoresis SHMT Serine Hydroxymethyltransferase THF N5, N10-methylene tetrahydrofolate TOF Time of flight UDP Uridine diphosphate UMP Uridine monosphosphate UV Ultraviolet

VG VINYLGLYCINE WHO World Health Organization

RELEVANCY OF STUDYING BIOSYNTHESIS OF HIGHLY-MODIFIED NUCLEOSIDE ANTIBIOTICS

1.1 Antibiotic resistance

Antibiotic resistance has become a widespread problem in the United States and across the globe. Out of billions of bacterial cells, some evolve through random mutation to have drug-resistance. Antimicrobial drugs have been susceptible to antibiotic resistance since they were first used to treat infections; penicillin resistance was identified the same decade it was introduced. This antibiotic resistance threat has long been known and was predicted by Alexander Fleming in the 1940’s before resistance had even developed [1-3].

Selective pressure of antibiotics on microorganisms accelerates this environmental resistance, as the bacteria that survive the antibiotic treatment are now more likely to have and share a form of drug-resistance with other bacteria [4]. This can have ripple effects, impacting both the individual and the community. Allergic reactions, disruption of gut microbiota, and increased incidence of drug-resistance are all adverse effects of this selective pressure [2]. Just a few of the most serious resistance stains, according to the

CDC, includes carbapenem-resistant Enterobacteriacae, extended spectrum β-lactamase producing Enterobacteriacae, vacomycin-resistant Enterobacteriacae, and methicillin- resistant Staphylococcus aureus [2].

This is a multifaceted problem and an interdisciplinary team is needed to answer it from multiple angles. Treating and preventing a drug-resistance crisis involves four major areas of focus. This includes improved prescribing/stewardship from physicians, tracking big data by biostatisticians, preventing infections through public health awareness, and

developing new drugs from the benchtop and translating that into successful clinical trials

[2]. Vancomycin, which took three decades to develop resistance, proved to be an excellent

example of how carefully prescribing an effective drug only when necessary could keep

remaining “last-resort” drugs effective for decades.

1.2 Natural product biosynthesis

New antibiotics are desperately needed. Penicillin sparked interest in natural

product drug discover, leading to a “discovery golden age” of biosynthetically-derived

therapeutics from the 1940-50’s. Although organizations like the CDC and WHO promote

the advance of new antibiotics, little clinical progress has been made in the last few decades. Apart from daptomycin, a lipopeptide discovered in the late 1980’s, there have been little-to-no new classes of antibiotics entering the clinic since 1981, creating a

discovery void (Error! Reference source not found.) [3]. CDC data suggests this has had

a detrimental impact: an increase in death and illness originating from antibiotic-resistance

(estimates of over two million illnesses and 23 thousand deaths) [2].

Figure 1.1 Adapted representation of WHO data on number of new classes of antibiotics entering the clinic: the golden age of drug discovery at the peak – and a discovery void as of recent.

The development of new antimicrobial drugs using natural products (i.e. compounds derived from secondary metabolism in a variety of organisms) is a promising route [5]. Natural products (NPs) or semi-synthetic derivatives of NPs are known to vary greatly in molecular weight, number of rings, ring systems, heteroatoms, rotatable bonds, stereocenters, and hydrophilicity [6, 7]. Approximately half of all new drug approvals have originated from NPs, likely due to their rich chemical diversity [5, 6]. This large and unique region of underexplored chemical space compared to synthetic origins has fueled NP-based drug discovery [6]. Additionally, the machinery orchestrating these biosynthetic transformations often offer new chemistries not visible by traditional synthetic means [8].

1.3 MraY bacterial translocase I

Antibacterial activity can occur through several mechanisms such as the inhibition of nucleic acid metabolism and repair, protein synthesis, folate synthesis, and enzymes involved in cell wall biosynthesis [9, 10]. With little overlap between cell wall biosynthesis in prokaryotic cells and any cellular activity in eukaryotic cells, cytotoxicity is less likely to occur in mammalian cells when inhibiting this pathway. Further, inhibition of peptidoglycan cell wall biosynthesis is only partially explored as just six of twelve minimal enzymatic steps have known targets [11-13]. MraY bacterial translocase I is currently an unexploited target in the clinic. MraY catalyzes the second stage of cell wall biosynthesis, transferring a phosphor-N-acetylmuramyl pentapeptide from uridine monophosphate to a

C55 undecaprenyl phosphate (C55-P) held in the lipid membrane to create undecaprenyl-

pyrophosphoryl-MurNAc-pentapeptide (lipid I, see Error! Reference source not found.)

[9, 14-18]. This will go on to be an integral part of the peptidoglycan cell wall.

Additionally, eukaryotic cytotoxicity is less of a concern, as similar proteins in eukaryotic

cells share very little sequence similarity to MraY. This is a proven pathway to target:

transglycosylase inhibitors (e.g. vancomycin) and transpeptidase inhibitors (e.g. β-lactams) are widely used in the clinic. Tunicamycin, a well-known nucleoside antibiotic, can inhibit this step of biosynthesis; however, tunicamycin displays prominent cytotoxicity with inhibitory activity against dolichyl phosphate:N-acetylglucosamine-1-phosphate transferase [9, 19]. This enzyme catalyzes a similar reaction to MraY in eukaryotic cells.

Interestingly, MraY is ubiquitous and essential across gram-positive and gram-negative bacteria. Considering this, MraY is a promising target in combating antibiotic resistance.

Figure 1.2 Partial illustration of cell wall biosynthesis noting how MraY fits into the overall pathway. Also noted are a selection of current antibiotics used to clinically to treat bacterial infections by inhibiting downstream peptidoglycan biosynthesis on the cell surface.

Only a few crystal structures of MraY have been resolved since its discovery in the

1990’s, so current structural insight into MraY is minimal. There are significant differences between the MraY native substrates and known inhibitors. Many inhibitors lack the diphosphate found in the natural substrate (UDP-MurNAc-pentapeptide), instead baring an extra modified ribosyl unit. One recent study of muraymycin D2 binding MraY supported that a large conformational change may accommodate these major differences.

Muraymycin D2 does not interact with three acidic residues, nor the magnesium cofactor,

essential to native substrate binding [20-22]. The functionalities of these inhibitors that

strengthen or weaken enzyme-inhibitor interaction are poorly understood. The work described throughout this thesis should provide tools to further inform structure-activity relationships of this interaction, getting closer to the answer of what makes an MraY inhibitor efficacious.

1.4 GlyU-ADR nucleoside antibiotics

A subset of MraY inhibitors include nucleoside antibiotics with a GlyU-ADR core

derived from UMP [14-16, 23]. Examples of this subset includes FR-900493,

muraminomicins, muramycins, liposidomycins, caprazamycins, sphaerimicins, and A-

90289 (Figure 1.3) [14-18, 23-25]. New members of this subset are still being discovered,

for example Cui et al discovered three new muraymycins most recently [26]. The GlyU-

ADR-containing nucleosides have thus far displayed no cytotoxicity to eukaryotic cells, in

contrast to tunicamycin. Many of these compounds have potent antimicrobial activity with

IC50 values ranging from 10-70 nM and MICs as low as 1 µg/mL on gram-positive bacteria such as methicillin-resistant S. aureus and vancomycin-resistant E. faecium [24].

Generally, highly-modified nucleosides are effective against gram-positive strains. This preclinical data, though limited, makes these promising new chemical entities for possible clinical use.

O O O O O OH O H3C O O OH O 5 N H NH H NH H2N N O O N O O O H N O H N O O O O O O O H N O 2 H H2N O H HO H HO OH O H HO HO OH HO FR-900493 Muraminomicin F

12 NH Liposidomycin O O O O O OH N NH2 O O H C O O O OH 3 OH O 6 OH O H N H H H H NH H NH N N N N O O N O N O O H H N O H N O O O O O O O HN O O H2N O H H N O H 2 Caprazamycin HN N H HO OH H HO H O HO O HO OH O S O OH Muraymycin A A-90289 A

Figure 1.3 Select GlyU-ADR nucleoside antibiotics. Components include: GlyU-ADR core (blue), modified N-alkylamine linker (red), and additional tailoring (black).

The biosynthetic pathways of these highly-modified nucleosides involve an array of unusual enzyme chemistry, all starting with canonical UMP. From UMP, large scaffolds with multiple stereocenters are installed. These compounds are often appended to other biosynthetically derived components, such as nonribosomal peptides and polyketides.

Biochemical characterization of this catalysis is essential, as bioinformatics commonly yields inaccurate data with such small training sets.

Oxidative dephosphorylation is the first step in these nucleoside biosynthetic pathways. This is carried out by a nonheme, Fe(II)-, α-KG-, oxygen-dependent dioxygenase, transforming UMP into a uridine 5'-aldehyde intermediate. This catalysis was probed by Goswami et al (Error! Reference source not found.); it was found that a single enzyme (LipL) is exhibiting UMP hydroxylase and phosphorylase activity [18, 27, 28].

Figure 1.4 Dioxygenase LipL activity and the phosphonate UMP analog used to probe the mechanism. Hydroxylation first occurs, followed by an oxidative dephosphorylation.

At this point, the pathway splits. Uridine 5'-aldehyde is utilized in two ways. For

one, it can be converted into GlyU by a transaldolase (LipK) [29]. The other way is via a

typical PLP-dependent transamination, forming 5'-amino-5'-deoxy-uridine. A phosphorylase and nucleotidyltransferase will activate the modified ribose sugar (ADR).

Then, a glycosyltransferase can join GlyU and the activated ADR to make the GlyU-ADR

disaccharide core (Figure 1.5) [15, 30, 31].

O O OH NH OH Glycosyl- H2N NH O H2N transferase O O N O N O O P O O O O , PL HO H2N H - hr ase L T ol H H NH NH sald HO OH -KG ran Fe(II), α H T HO OH HO OH N O N O O O O PO Dioxygenase L - H Me Tra t , O HO OH ns P HO OH a LP m NH ina se N O Phosphorylase O O H2N H2N Nucleotidyl- ONDP transferase HO OH HO OH

Figure 1.5 Biosynthesis of GlyU-ADR shared among this subset of highly-modified nucleosides. Two canonical UMP molecules can be transformed into the GlyU-ADR core of these large nucleoside antibiotics using 6 enzymatic steps.

Before further scaffold extension, there is evidence that all homologous

biosynthetic pathways require phosphorylation at the 3''-position by a kinase (e.g. the

recently characterized Mur28 of muraymycin biosynthesis) [32]. This is hypothesized to

be part of a self-resistance mechanism, as antibacterial activity of the phosphorylated

intermediates is significantly hindered [32]. Protecting the producing strain during

biosynthesis this way is has been observed before, for example during caprazamycin

production [33].

1.5 Sphaerimicin: the most recently discovered GlyU-ADR-containing natural

product

The most recently discovered nucleoside antibiotic containing a GlyU-ADR core is the sphaerimicins. Sphaerimicins A-D have been isolated from Sphaerisporangium sp.

SANK 60911 using a structure-based gene targeting screen [24]. Sphaerimicin A has been structurally elucidated and found to have a unique 3'-sulfonation, as well as a

dihydroxylated piperidine bridged to the 5''-amine of a 5-amino-5-deoxyribose moiety.

This compound exhibits effective MICs against a variety of bacterial strains and possess potent IC50 values against MraY [24].

Table 1.1 MIC values of sphaerimicin A against a range of bacterial strains. Organism MIC (µg/mL) Streptococcus pneumoniae ATCC 49619 (THB) 1 + Streptococcus pneumoniae ATCC 49619 (G ) 16 + Streptococcus pyogenes ATCC 12344 (G ) 16 + Staphylococcus aureus ATCC 6538P (G ) 4 + Staphylococcus aureus 10925 (G ) 8 + Staphylococcus epidermidis ATCC 14990 (G ) 8 + Enterococcus faecalis ATCC 29212 (G ) 2 + Enterococcus faecium ATCC 19434 (G ) 2 - Moraxella catarrhalis ATCC 25238 (G ) 4 - Haemophilus influenzae ATCC 49247 (G ) 8 - Haemophilus influenzae Rd (G ) 32

Table 1.2 IC50 values of several highly-modified nucleoside antibiotics against MraY.

MraY inhibitor IC50 (nM) Muraminomicin F 11.47 Sphaerimicin A 13.65 Sphaerimicin B 44.49 Sphaerimicin C 65.82 Sphaerimicin D 12.23

1.6 Contents of the dissertation

Described throughout this work is how the GlyU-ADR intermediate is initially

utilized in a similar (but unusual) way in each of these homologous natural product

biosynthetic pathways. Established herein is a unique PLP-dependent gamma-replacement, wherein a Cγ-S bond is broken and a Cγ-N bond is created. This forms the critical N-

alkylamine linker where further tailoring will occur in downstream biosynthesis. First, a

current understanding of PLP chemistries is reviewed (Chapter 2). Next, studies used to

probe the mechanism of the PLP-dependent alkyltransfer are described (Chapter 3).

Finally, preliminary evidence for a few ways the biosynthetic pathways diverge,

specifically concerning the formation of the dihydroxylated piperidine ring in the

sphaerimicins, is discussed (Chapter 4).

CURRENT UNDERSTANDING OF PLP CHEMISTRIES

2.1 Introduction - pyridoxal 5'-phosphate function & structure

Pyridoxine (vitamin B6) is a precursor of pyridoxal 5'-phosphate (PLP). PLP-

dependent enzyme catalysis is ubiquitous and incredibly diverse, considering nature has

used the same coenzyme to fulfill so many different physiological needs. Recently, a study

estimated that 4% of total enzyme activity requires PLP [34-36]. Additionally, PLP-

dependent enzymes encompass 5 of the 6 enzyme classes categorized by the Enzyme

Commission of the International Union of Biochemistry (including oxidoreductases,

transferases, hydrolases, lyases, and isomerases) [34, 35]. More than 230 different enzyme

activities have been reported for catalysis utilizing PLP, which is done so by electron

movements enabled by the highly conjugated electron sink of the PLP pyridine ring [37-

42]. This allows for stabilization of carbanions through delocalization of the negative

charge throughout the conjugated system [43-45]. PLP is used in primary metabolism –

interconverting alpha-keto acids and amino acids to allow exchange of carbon sources

between the Kreb’s cycle and amino acid metabolism, as well as cell signaling. For

example, in the regulation of neurotransmitters, GABA aminotransferase (GABA-AT) can

catalyze 2-oxobutyrate to L-glutamate. GABA-AT can also interchange between GABA

and succinic semialdehyde. Another PLP-dependent enzyme, glutamate decarboxylase links these two reactions, converting L-glutamate to GABA [46]. This makes PLP central

to shuttling carbon sources between neural transmission and aerobic metabolism.

Additionally, PLP-dependent catalysis has been utilized in numerous ways to create a vast

number of nonproteinogenic amino acids [47, 48]. Clearly, PLP is important to vital

cellular functions and has brought about incredibly diverse chemistry achieved through

acidification and exchange of protons otherwise too unreactive to modify.

Let’s consider how enzymes typically utilize PLP. First, an internal aldimine will

form in the enzyme active site typically from a lysine residue in the active site. This is

followed by an external aldimine forming with the substrate, often an amino acid. From

here, a number of bonds can be cleaved at the alpha position, wherein the electrons will be

stored and dispensed via the electron sink of the PLP pyridine ring. Typically, the enzyme

will bind substrates in a way that the Cα-H bond is held in parallel to the p orbitals of the π system [49]. This weakens the bond, lowering vibrational force constants, allowing for effective deprotonation [49]. Formation of new linkages can occur with incoming protons or a second substrate [40, 42]. Experimentally, different conjugations of PLP bound to substrate can be tracked using UV-visible spectroscopy throughout the catalytic cycle, such as the formation of the internal and external aldimine. Additionally, a quinonoid forms when PLP captures electrons from the substrate and stabilizes this, like after alpha-carbon deprotonation, though this is not always possible to observe in the lab setting [50].

Proteins direct the diverse chemistry that can take place; active site mutations have been demonstrated to modulate activity, such as an aminotransferase acquiring decarboxylase-like activity [51]. In another case, a single tyrosine mutant in human and

Morganella morganii histidine decarboxylases (hHDC and mHDC, respectively) altered the directed chemistry, causing a decarboxylation dependent oxidative deamination to occur, rather than the wild-type activity of a simple decarboxylation to form histamine

(Figure 2.1, [52]). The tyrosine is thought to be responsible for the protonation of the

substrate alpha-carbon, allowing for histamine formation once CO2 leaves and the

quinonoid intermediate is formed. With a phenylalanine mutation, this protonation is

prevented. Kaminaga et al propose that, in the presence of oxygen, this leads to a peroxy-

aldimine intermediate. Once released from PLP, this peroxy-imidazole ethanamine would

release hydrogen peroxide, forming imidazole ethanimine. Exchanging with water affords

the imidazole acetaldehyde product. Thus, a minor loss of a hydroxy group on an active

site residue converts a decarboxylase into an aldehyde synthase. Nature has evolved to give

PLP-dependent enzymes a multitude of functionality, capable of acting on amino acid, oxoacid, and amine substrates [48]. Through PLP-enabled protonation and deprotonation

events at alpha-, beta-, and gamma-position carbons of substrates, a variety of eliminations

and replacements can occur. This greatly adds to the chemical complexity that Nature

creates and provides crucial synthesis, degradation, and interconversion of amino acids –

linking carbon and nitrogen metabolism [42].

N N - N + O CO2 NH3 HN Y334F/hHDC HN + hHDC HN Y260F/mHDC H N mHDC 3

Figure 2.1 Wild-type HDC activity compared to a critical mutation capable of modulating the PLP-depdendent chemistry that occurs.

PLP-dependent enzymes can be divided into five independent classes based on fold-type [35]. Alpha family PLP-dependent enzymes (Fold Type I) encompass a combination of parallel beta sheets and alpha-helices resembling the aspartate aminotransferase homodimer. Often, this fold-type will catalyze a change in covalency at

the α-position, such as aminotransferases (Figure 2.2) [53], serine hydroxy

methyltransferases, and transaldolases (e.g. LipK [29]). Fold Type II structures resemble

the beta-subunit of tryptophan synthase. These Beta family PLP-dependent enzymes (Fold

Type II) typically change covalency at the β-position, including tryptophan synthase β, serine dehydratases [53], and SbnA [54]. The alpha family is the largest, followed by the beta family [35]. Two smaller families – the D-alanine aminotransferase and alanine racemase family – each consist of only a few enzymes (Fold Type III and IV, respectively).

Lastly, Fold Type V is based on glycogen phosphorylase. This is a unique fold-type, as it

is the only class of PLP-dependent enzymes that does not go through a geminal diamine intermediate between the deprotonated substrate amine and the internal aldimine Schiff base [35]. Many PLP-dependent enzymes do not fit into any of these families functionally, such as various amino acid racemases and decarboxylases (Figure 2.3), as well as ACC synthases involved in ethylene biosynthesis [53, 55]. PLP-dependent enzymes probably diverged into distinct classes based on the chemistries they perform – later diverging further based on substrate specificity [42]. Aminotranferases and decarboxylases can be found in multiple lineages, suggesting convergent evolution [42]. PLP likely emerged early in biological evolution, before divergence of eukaryotes and archebacteria 1500 million years ago [42]. PLP can form an “external” aldimine in the absence of the enzyme-PLP internal aldimine. For example, PLP reversibly binds L-Met to form an aldimine with a Kd

of 3.6 mM at pH 7.5 [56]. The first biological catalysts may have been metal ions and

organic cofactors that coordinated the useful electron-sink molecule [42]. Described below

are some of the known α, β, and/or γ elimination and addition catalytic strategies employed by PLP known to date.

Figure 2.2 General aminotransferase mechanism. First, the alpha-position is deprotonated; this drives electrons into PLP, affording the quinonoid intermediate; next, water is utilized from the aqueous environment and the electron sink is again utilized to release the product.

O H+ R - R O CO NH 2 N

HO -2 HO -2 OPO3 OPO3 N N H H Figure 2.3 General decarboxylase mechanism. In this case, the electron sink stabilizes the carbonanion at the alpha-position; because of this stabilization, carbon dioxide can be eliminated from the reactant.

2.2 Alanine racemase

Though the alanine racemase family constitutes a small portion of PLP-dependent

enzymes, mechanistic studies into such enzymes have given significant insights into PLP-

dependent catalysis. Alanine racemase is the enzyme responsible for the interconversion

between D- and L-isomers of alanine (Figure 2.4). Michael Toney et al have explored alanine racemases from Bacillus strains, finding that during L- to D-isomerization an active site Tyr265 residue is responsible for proton abstraction from L-Ala and a Lys39 residue is used for reprotonation. During D- to L-isomerization, the Lys39 residue deprotonates and

the Tyr265 residue reprotonates [41, 57-60].

Contrary to the PLP resonance previously described, the full extent of the PLP

electron sink is not utilized during this alanine racemase activity. In many PLP-dependent

enzymes, a conserved aspartate residue in the active site nearby the pyridine nitrogen helps

maintain a formal positive charge. This creates an electronic environment that causes an

intramolecular interaction where the hydroxyl of PLP hydrogen bonds with the aldimine nitrogen, causing it to mostly exist in a protonated state [61]. In alanine racemases studied

here, an Arg219 residue interacts with the pyridine nitrogen. The basicity of guanidine

group of arginine prevents the pyridine from being formally protonated [62]. This is

reflected in structural studies [40, 44, 63] and the inability to detect the diagnostic UV

spectra of a quinonoid intermediate during catalysis [41]. Though this destabilizes the

carbanionic intermediate, the mechanism still proceeds through a stepwise model of

protonation and reprotonation, rather than a concerted double proton transfer, which would

bypass the need for a carbanionic intermediate to exist. This seems counterintuitive –

destabilizing the conjugation system of PLP, even though the alanine racemase mechanism

proceeds through the typical carbanionic intermediate. Toney et al hypothesize that this

may serve to avoid side reactions such as transamination, which is seen in other PLP-

dependent racemases that lack an arginine residue in proximity to the nitrogen pyridine and

possess the typical charged state [62, 64, 65]. The aldimine nitrogen must still be

protonated to activate it for nucleophilic attack of the substrate; this is called

transamination, which often creates a geminal diamine intermediate. Based on 15N solid-

state NMR studies, it is thought that this activation is instead achieved through protonation

(or strong OHO hydrogen-bonding) of the pyridine hydroxyl [61]. This means there are

two modes of internal aldimine activation – through either N- or O-protonation. Toney et

al has also explored the importance of small, femtosecond time scale active site protein

motions in alanine racemase catalysis.

The Born-Oppenheimer approximation states that isotopic substitution only alters

vibrational frequency, leaving electrostatic properties unaltered. Vern Schramm pioneered

using this principle to disrupt femtosecond motion of HIV-1 protease and purine nucleoside

phosphorylase. While loop and domain motions at micro- to milli-second time scale are

central to providing catalytically productive environments (e.g. opening and closing active

site as needed), smaller protein motions surrounding the active site in the femto- to pico-

second time scale are involved in transition state barrier crossing. Using “heavy enzymes”

with major incorporation of heavier isotopes into the peptide sequence (i.e. kinetic isotope

effect, KIE), Schramm et al demonstrated that bond vibrations near the catalytic site in this

time scale play a role in transition-state formation [66-69]. This disruption is caused by

local heavy atoms in the catalytic site leading to a reduced probability that the enzyme

finds proper enzymatic transition state conformation.

Michael Toney et al have applied a similar approach to alanine racemase. This is a

novel addition to this area of research as this was the first time heavy-enzyme KIEs were used to observe an enzyme-catalyzed reaction where proton transfer was the rate-limiting step (a common motif in PLP-dependent reactions [40, 45, 59, 70, 71]). Toney et al studies showed that isotope effects from enzyme and substrate deuteration are not independent of each other, indicating that these two species are coupled in the rate-limiting event [72].

This means that femto-second protein motion during proton transfer is critical to transition state barrier crossing. Additionally, they found that intrinsic KIEs are relatively small in

either direction of catalysis (1.66 and 1.57 for the L to D and D to L conversion,

respectively). This is likely owed to proton tunneling between the alpha-carbon and either residue responsible for proton transfer (Tyr265 or Lys39 acting as the acid-base catalysts in their studies with Geobacillus stearothermophilus, like their previous studies in Bacillus

alanine racemases) [65, 72]. Together, this mechanistic work shows that protein residues

and small femtosecond time scale conformational changes in the active site aid in directing

specific chemistry take place.

+ Lys39-NH3 + Lys39-NH3 Lys39-NH2 Tyr265-O Tyr265-OH Tyr265-OH Hα - - H - CO2 CO2 CO2

NH NH NH L-Ala + E-PLP O O O D-Ala + E-PLP O P OH O P OH O P OH O O HO O O O N N N

Figure 2.4 PLP-depedent mechanism of alanine racemase depicted. Residues on opposing sides of the substrate allow interconversion between stereoisomers.

2.3 γ-addition and elimination catalysis

2.3.1 Cystathionine gamma-synthases / lyases and LolC

Transsulfuration and reverse transsulfuration are critical de novo pathways for L-

cysteine and L-homocysteine interconversion, respectively [73, 74]. Transsulfuration, critical in plant and bacterial metabolism, uses cystathionine gamma-synthase (CGS) and cystathionine beta-lyase (CBL) to interconvert between L-cysteine / O-succinyl-L-

homoserine and cystathionine [75]. In fungi, cystathionine-gamma-synthase catalyzes C-

C bond breaking and a C-S bond formation [47]. Overall, the gamma-elimination of O-

acetyl- L-homoserine gives a vinylglycine ketimine, which is subject to the nucleophilic

attack of the thiol group of cysteine. In contrast, reverse transsulfuration of mammalian

and fungal metabolism, uses cystathionine beta-synthase (CBS) and cystathionine gamma-

lyase (CGL) to interconvert between L-homocysteine / L-serine and cystathionine [74].

Mammalian and fungal CBS utilizes homocysteine and serine for form cystathionine through beta-replacement. CGL performs an alpha, gamma-elimination on cystathionine, yielding cysteine, 2-oxobuytrate, and ammonia. Plant and bacterial CGSs catalyze the replacement of substituents at the γ-position of amino acids through a common α-proton exchange as well as a rapid β-proton exchange [76].

CGSs can utilize OSHS or L-VG as a C4N donor to form cystathionine in the presence of the C4N acceptor cysteine (γ-replacement product, Figure 2.5) [77]. Without a

C4N acceptor present, both OSHS and L-VG can form 2-oxobutyrate through an

aminocrotonate-PLP intermediate (γ-elimination product) [78]. There is evidence that the

CGS reaction yields the Z-conformation in the L-VG-PLP intermediate [79]. Conversely,

cystathionine gamma-lyase catalyzes the alpha, gamma-hydrolysis of L-cystathionine to form L-cysteine and 2-oxobutyrate. Remarkably, N360S mutation in the parasite

Toxoplasma gondii cystathionine gamma-lyase (TgCGL) changes the specifity of

elimination. This modification directs catalysis to perform an alpha,beta-elimination, rather

than the alpha, gamma-elimination found in wild type TgCGL [73]. LolC, a critical enzyme in loline alkaloid biosynthesis, is proposed to have similar activity to CGS, creating a new

Cγ-N bond rather than a new Cγ-S bond (Figure 2.6) [80].

O O O HO OH O NH2 NH2 O CSG HO S OH PLP O NH2 NH2 Succ. HO SH O Figure 2.5 CGS reaction, wherein homocysteine of OSHS is eliminated and used in the addition of cysteine.

O XO OH CO2H NH2 O LolC N OH CO2H PLP NH2 NH XOH

Figure 2.6 Proposed LolC activity. An analogous mechanism to CGS; however, a nitrogen heteroatom is used to form a new Cγ-N bond.

2.3.2 Alkyl transfer in nucleoside antibiotic biosynthesis

Another PLP-dependent γ -replacement catalytic strategy for transferring a C4N

group was recently reported. An aminobutyrate transferase employ Cγ-S bond breaking and

Cγ-N bond formation – a greater feat in regards to nucleophilicity compared to the CGS

Cγ-O bond is breaking and Cγ-S bond formation [81, 82]. Evidence shows that these

ABTases, such as Mur24, cannot utilize L-VG as an alternative donor, an interesting distinction from CGS that perform similar chemistry [78]. Analogous to CGS, α-KB/2- oxobutyrate is formed in the absence of the C4N acceptor, making it a reasonable

hypothesis that γ-replacement and -elimination reactions diverge from a common

intermediate (proposed to be L-VG, like CGS). The sequence similarity suggests nucleoside ABTases to be like 1-aminocyclpropane-1-carboxylate synthases, but they have

more in common with CGS mechanistically as indicated by proton exchange experiments

with substrate isotopologues and reactions conducted in D2O [82]. First, the alpha-proton is abstracted (as in many PLP-dependent enzymes). This is likely followed by immediate

C4' protonation, generating ketimine II. This is indicated by difficulty in measuring any quinonoid formation during Mur24 catalysis. Now more acidic due to the vicinal iminium, the beta-proton is abstracted forming enamine III. Subsequently, the beta, gamma-

unsaturated ketimine IV is formed as MTA is eliminated. The amine nucleophile (i.e.

nucleoside intermediate) is deprotonated and attacks the gamma-carbon in an aza-Michael-

like addition (described in Chapter 3). It is possible that Lys234 of Mur24 is responsible for internal aldimine formation, alpha- and beta-proton transfers [82]. The natural alkyl-

donor substrate of both ACC synthases and nucleoside ABTases is AdoMet [83]. Similar

to loline alkaloid biosynthesis, a simple methyltransferase-like SN2-type substitution looks possible (such as that seen in polyamine biosynthesis like spermidine [82, 84]), but the chemical strategy to transfer this group is much more intricate. This mechanism is further explored in the contents of this dissertation in Chapter 3.

2.3.3 Formation of alkyl-substituted pipecolate in citrinadin and fusaric acid

biosynthesis

Pipecolate and derivatives are intriguing building blocks that are frequently used in natural products such as rapamycin, where the moiety plays a critical role in target potency.

Though pipecolate is not an unusual natural compound – found in plant, fungal, and microbial biosynthesis – alkyl-substituted pipecolates are very rare. Recently, a two-step enzymatic synthesis of (2S,6S)-6-methyl pipecolate was characterized in citrinadin

biosynthesis by the Tang lab. First, a gamma-elimination of O-acetyl-L-homoserine occurs.

This generates vinylglycine ketimine, wherein a nucleophilic attack by acetoacetate will

create a new carbon-carbon bond. This full gamma-replacement is catalyzed by a single,

PLP-dependent enzyme, CdnF (Figure 2.7), which has sequence similarity to cystathionine-gamma-synthase. This is in equilibrium with a cyclic Schiff base, which is the substrate for the stereoselective imine reductase, CndE [85]. CndF was shown to form

an internal aldimine with a lysine residue, typical of many PLP-dependent enzymes. The mechanism of this route is distinct from typical pipecolate biosynthesis, which leads to an aliphatic carbon at the position where alkyl-substitution would need to occur [86].

CndF shows a flexible substrate specificity toward the nucleophile. Though the natural substrate is acetoacetate, CndF can accept 6 beta-keto carboxylate and ester substituted substrates [85]. This shows that decarboxylation of the acetoacetate is not required for enol formation, meaning that a general base in the active site must be responsible for abstracting the alpha-proton of the beta-keto substrates. With respect to medicinal chemistry, this is interesting because the R-group (naturally a methyl group) of the CndE product is in the mature natural product following NRPS-modification and tailoring reactions. This means that CndF could potentially be used with various substrates for compound optimization (although this would currently be limited due to downstream specificity in the biosynthetic pathway). With respect to enzymology, this is interesting because (though it is mechanistically analogous to cystathionine-gamma-synthase) this was the first example of a PLP-dependent gamma-substitution that catalyzes a C-C bond formation.

More recently, the Tang lab has characterized another Cγ-O bond breaking and Cγ-

C bond forming PLP-dependent enzyme Fub7 in fusaric acid biosynthesis.

Biosynthetically, Fub7 catalyzes a gamma-replacement reaction, converting O-acetyl-L-

homoserine and n-hexanal to acetate and (2S)-2-amino-5-formylnonanoic acid (Figure 2.8,

[87]). This is a chemically intriguing reaction. PLP-dependent enzymes using aldehyde substrates often use them as electrophiles, however, in this cause the enolate tautomer of n-hexanal is used as a nucleophile. This product is in equilibrium with the cyclic unsaturated pipecolate, a precursor to fusaric acid [87]. Using Fub7, the group chemoenzymatically synthesized a series of pipecolate analogues including 5-alkyl-, 5,5- dialkyl-, and 5,5,6-trialkyl-l-pipecolic acids [88].

O O

R1 OH - - OOC OAc H2 - PLP R1 N COO R1 N COO + NH3 CndF CndE

- - OOC OOC - - OOC OOC + PLP reductase NH3 O aminotransferase + N [H] H2N NH + NH + 3 3 H2O

Figure 2.7 CndF activity using PLP-dependent gamma-replacement chemistry to break a Cγ-O bond and create a new Cγ-C bond.

O O BH BH H H B H - H - H - - CO2 AcO CO2 - AcO CO2 AcO CO2 AcO CO2 NH NH NH NH O BH NH O O O O P OH O P OH O O O P OH O P OH P OH O O O O O O O O O O N N H N N N H H H H

H O H H O H O O HA HA H - - - - CO CO CO 2 CO2 2 2 NH B NH NH NH H O O O O O P OH P O P OH O P OH O O OH O O O O O O O N N N N H H H H

-H2O tautomerization - - O CO2 - H2N N CO2 N CO2 H

Figure 2.8 Fub7 uses PLP-dependent gamma-replacement to break a Cγ-O bond and create a Cγ-C bond; unlike CndF, catalysis uses an aldyhyde substituent via enolate reactivity.

2.3.4 The γ-elimination, α-addition (Claisen Condensation) catalysis of CqsA

PLP-dependent enzymes yielding a γ-replacement first make a stabilized α- carbanion, followed by a deprotonation at the newly acidic Cβ-position. This leads to the

leaving of the γ-substituent, creating a fully conjugated intermediate with electrophilicity

at the Cγ [79]. Catalysis can also include a mix of elimination and addition at varying

positions, such as the loss of γ-elimination and α-addition (acyltransfer activity) of CqsA

involved in Vibrio cholera autoinducer-1 biosynthesis [89].

CqsA is critical to the biosynthesis of the quorum-sensing signal CAI-1, coupling

S-adenosyl-L-methionine (AdoMet) and decanoyl-coenzyme A to form a CAI-1 precursor,

Ea-CAI-1. Catalysis combines the γ-elimination of the MTA moiety of AdoMet, forming a vinylglycine-PLP intermediate. This will then undergo an acyltransferase reaction,

leading to the addition of a decanoyl moiety at the α-position (Figure 2.9). This was a novel addition to the known biosynthetic arsenal, as an enzyme capable of doing both of these catalytic transformations in one active site had previously been unheard of entirely [89].

Undoubtedly, the use of AdoMet as a cosubstrate in enzymatic reactions has led to

interesting chemistries. For example, using AdoMet as an alkyl-donor in spermidine

biosynthesis, as well as the more mechanistically intriguing PLP-dependent alkyl-transfer

of CAI-1 and nucleoside antibiotic biosynthesis. This makes for an excellent example of

how Nature has evolved strategies to leverage such an abundant molecule stereotypically

reserved for amino acid metabolism and methyltransferase reactions. Another unusual use

of AdoMet is the cyclization of the aminobutyrate substituent in colibactin biosynthesis

[90, 91], as well as the PLP-dependent chemistry is found in ethylene biosynthesis.

CoAS R O H O O + H + S - S - - Ad O Ad O MTA O NH N N

HO -2 HO -2 HO -2 OPO3 OPO3 OPO3 N N N H H H Figure 2.9 CqsA mechanism wherein AdoMet is utilized in a gamma-elimination, breaking a Cγ-S bond; this is followed by alpha-addition acyltranfer-like activity.

2.3.5 The γ-elimination of 1-Aminocyclopropane-1-carboxylic Acid Synthases

1-Aminocyclopropane-1-carboxylic acid synthases (ACC synthases) are essential to plants in the formation of ethylene, a plant hormone responsible for fruit-ripening and

other regulative processes. ACC synthases catalyze Cα-Cγ bond formation and Cγ-S bond

cleavage of AdoMet to form ACC and MTA (Figure 2.10), the rate-limiting step in ethylene formation. Evidence suggests that, unlike CGS, Cβ protons do not participate in

the reaction [55]. This is an interesting part of the mechanism distinct from the ABTases described in this dissertation, as they share the most sequence similarity with ACC

synthases but do appear to utilize Cβ protons. Unexpectedly, an ACC synthase has also been functionally characterized from the guangnamycin biosynthetic pathway (a bacterial

producing strain) [92]. Oddly, a gene in humans has been found encoding a putative ACC

synthase, though studies thus far have only shown activity toward deaminating VG [93].

We propose that he lysine residue essential to alpha-carbon deprotonation is quickly pulled

away from the catalytic region via a glutamate residue via electrostatic interactions. This

prevents any other interactions from occurring, leading to the formation of the

cyclopropane ring.

O O O + H + S - S - - Ad O Ad O MTA O NH N N

HO -2 HO -2 HO -2 OPO3 OPO3 OPO3 N N N H H H Figure 2.10 ACC synthase mechanism wherein the alpha-carbon is deprotonated by a neighboring lysine residue, forming a ketimine intermediate. The electron movement is reversed, leading to the ketimine pi electrons to attack the gamma-position (thereby breaking the Cγ-S bond and forming the cyclopropane ring.

2.3.6 PLP-dependent gamma-elimination in alkyne functional group biosynthesis

Six enzymes are utilized during the formation of beta-ethynylserine from L-lysine in S. cattleya. One of these enzymes, BesB is PLP-dependent with sequence similarity to the cystathionine-beta-lyase/cystathionine-gamma-synthase family. Once the halogenase and oxidase, BesD and BesC, respectively, chlorinates and oxidatively cleaves L-lysine,

BesB catalyzes the gamma-elimination of chloride from 4-chloro-allyl-L-glycine to form

an L-propargylclycine intermediate (Figure 2.11). In probing the BesB mechanism, a deuterium exchange was monitored with high-resolution mass spectroscopy. Using 4-

+ chloro-allylglycine as a substrate in D2O, [M+2D] L-propargylclycine was detected, which

would indicate a mechanism analogous to other PLP-dependent gamma-eliminations requiring two deprotonation-protonation events. It is proposed that alpha-carbon deprotonation forms a 4-chloro-allylglycine-ketimine intermediate. A second, beta-carbon deprotonation initiates the chloride elimination, possibly forming an allene intermediate that is isomerized into the final propargyl group [94]. This could be instrumental in the design of bioorgonthonal platforms for click-chemistry.

- + - + - - OOC NH3 BesD OOC NH3 BesC OOC BesB OOC - - + α + α NH + Cl NH + NH3 O2, KG, Cl NH3 Cl O2, KG, Cl 3 PLP 3

Figure 2.11 BesB in the context of beta-ethynylserine biosynthesis in which the PLP- dependent gamma-elimination of chlorine leads to alkyne formation.

2.4 Beta-addition and -elimination catalysis

2.4.1 Beta-replacement in lanthionine biosynthesis

Lanthionine synthase, a homologue of O-acetylserine sulfhydrylase (OASS),

catalyzes the beta-replacement of L-cysteine with another molecule of L-cysteine, releasing

H2S and L-lanthionine [95, 96]. This is achieved first through typical internal aldimine

formation with a lysine residue. Once the external aldimine is formed with L-cysteine,

proton transfer between the alpha-carbon and the same lysine residue leads to the release

of H2S and formation of ketoenamine and enolimine intermediates in resonance. Next, the addition of another L-cysteine molecule leads for the formation of L-lanthionine [96, 97].

This reaction is highly similar to cystathionine beta-synthase (CBS), however lanthionine

synthase prefers formation of lanthionine as a product, while CBS prefers cystathionine

(both are capable of making either product to an extent) [98]. Wolthers et al found that a

S224A lanthionine synthase mutant changes this ratio of product formation, increasing

turnover rate at the cost of catalytic preference for L-lanthionine formation. Product

formation changes from a 2:1 ratio of lanthionine formation over cystathionine, to a 1:1

formation of mixed product. This indicates that the presence of the serine residue in the

active site is critical for binding specificity of the second substrate (creating a preference

of a second L-cysteine molecule over L-homocysteine, like in CBS catalysis). Likely, this

is a case where Nature has evolved to prefer specificity over efficiency.

2.4.2 Beta-replacement in viomycin and staphyloferrin biosynthesis

VioB, a 2,3-diaminopropionate synthase from viomycin biosynthesis, utilizes L-

serine/O-acetyl-L-serine to create L-2,3-diaminopropionate. Sharing homology to cysteine synthases and serine dehydratases, VioB facilitates a PLP-dependent beta-substituent replacement and elimination. VioB has a unique nucleophile; while most PLP-dependent enzymes use a thio- or amino-nucleophile (often on other amino acids such as cysteine),

VioB uses free ammonia in solution (possibly generated by another enzyme in the biosynthetic pathway) as a nucleophile (Figure 2.12) [99].

NH + + 3 NH2 + VioK - H3N - CO2 CO2 NH3

- CO2

R NH + VioB NH O VioB 3 - + O C NH O P OH - 2 3 O + O O2C NH3 N

Figure 2.12 VioB PLP-dependent activity, where beta-elimination is followed by beta- addition at using free ammonia nearby during catalysis.

SbnA is responsible for the first step in the biosynthesis of the siderophore,

staphyloferrin B. Catalysis coverts O-phospho-L-serine and L-glutamate to N-(1-amino-1-

carboxy-2-ethyl)-glutamic acid (Figure 2.13) [54]. As expected, SbnA creates an

absorption spectrum of 412 nm in the presence of PLP (indicative of an internal aldimine

forming with a lysine residue in the active site. Though there is homology to VioB, only

O-phospho-L-serine was a substrate for SbnA, rather than L-serine/O-acetyl-L-serine. Upon

29 addition of O-phospho-L-serine, an external aminoacrylate intermediate is formed, as indicated by a shift in the UV-Vis spectrum to a 324 and 467 nm peak [100].

COOH O COOH SbnA -2 O OPO3 OH PLP NH2 -3 HOOC NH2 PO HOOC N OH 4 H NH2

Figure 2.13 SbnA catalyzes beta-replacement chemistry, wherein a Cβ-O bond is broken and a Cβ-N bond is created.

2.4.3 The β-elminination catalysis of serine dehydratase

Serine dehydratases perform a β-elimination of the hydroxyl group of serine. It was found that a conformational change in PLP was essential for nucleophilic attack of the substrate, as well as concerted α-proton transfer from the substrate to a protein residue in the active site and directing substrate carbanion formation [38].

2.4.4 Beta-decarboxylation

Although much less common, beta-decarboxylation has been studied. Most novel of these is UstD – from the ustiloxin B biosynthetic pathway of the pathogenic fungus,

Ustilaginoidea virens. In this catalysis, the beta-decarboxylation of aspartate leads to formation of an enamine. This enzyme activity is unique, because the enamine formed is reacted with an aldehyde substrate in the same active site [101] (Figure 2.14). Though this bi-bi ordered sequential PLP-dependent replacement chemistry is typical in gamma- position catalysis, it is rather unusual in beta-position chemistries.

O R - - S OOC UstD OOC UstD - OOC CO2H + PLP + NH NH + 3 3 O R NH O S 3 ustiloxin B

Figure 2.14 PLP-dependent beta-replacement, wherein carbon dioxide is first released from the substrate.

2.4.5 PLP-dependent Thiol Formation

Methionine gamma-lyase (MGL) is commonly associated with L-methionine

catabolism, decomposing L-methionine to methanethiol, 2-oxobutyrate, and ammonia.

Interestingly, MGL is also capable of the beta-elimination of S-substituted cysteines,

generating the respective alkylthiol, pyruvate, and ammonia. In the case of Citrobacter

intermedius MGL, a Tyr 113 residue plays a critical role in gamma-elimination, whereas a

Tyr58 residue plays a critical role in beta-elimination [102]. This contrasts with the more recently discovered C-S lyase from Lactobacillus delbrueckii (a lactic acid bacterium).

This C-S lyase has broad substrate specificity (aminoethyl-L-cysteine, L-cystine, L-

cysteine, L-cystathionine) but appears to only be capable of beta-elimination [103]. As seen with lanthionine synthases, single protein residues can play a critical role in substrate specificity. In the case of Pseudomonas putida MGL, a C116H renders the enzyme completely inactive toward the L-cysteine and L-methionine (the wild-type substrates), but opens up activity to L-homocysteine [104].

2.5 The α-replacement catalysis of transaldolases in production of beta-hydroxy-

alpha-amino acids

LipK, a transaldolase from the A-90289 biosynthetic gene cluster, joins L-threonine and uridine 5'-aldehyde in the biosynthesis of a particular subset of nucleoside antibiotics to form a GlyU precursor through alpha-replacement chemistry (Figure 2.15). Initially annotated as a SHMT, LipK has distinct regions in the protein sequence that differentiate its activity from bona fide SHMTs [29]. Classically, SHMTs are used to convert glycine to serine using THF as a C1-unit donor. LipK was the first enzyme found to do this unusual formation of a β-hydroxy, α-amino acid using a uridine precursor, creating two new stereocenters in the process. There is a whole class of enzyme homologues capable of analogous catalytic activity; notably, AbmH of albomycin biosynthesis uses a thioheptose nucleoside intermediate [105].

O OH O O O - R - O O NH NH

HO -2 HO -2 OPO3 OPO3 N N H H Figure 2.15 Transaldolase (LipK) mechanism constituting an alpha-elimination of acetaldehyde and addition of an amino acid, leading for the formation of GlyU.

2.6 PLP-dependent oxidation catalysis

PLP-dependent enzymes can also perform oxidation chemistry, such as oxidative dehydrogenation typically catalyzed by metal- or Flavin-dependent enzymes. For example,

MppP from the mannopeptimycin biosynthetic pathway, which is partially responsible for

the formation of L-enduracididine. There is evidence that MppP converts L-arginine and

dioxygen to (S)-4-hydroxy-2-ketoarginine via unusual PLP-, O2-dependent chemistry.

Interestingly, even though O2 is required- it is not incorporated into the product. Instead,

the hydroxyl and ketone oxygen atoms are derived solely from water, while the dioxygen

molecule is reduced to H2O2 [106]. Mannopeptimycin, teixobactin, and enduracidin are all

nonribosomal peptide antibiotics that bind the lipid II intermediate during peptidoglycan

cell wall biosynthesis (similar to vancomycin mode of action, but different binding

interactions) [107]. Ind4, from the indolmycin biosynthetic pathway, has been shown to

catalyze a similar 4-electron oxidation of L-arginine, forming 4,5-dehydroarginine [108]

An Ind4 homolog, RohP, has recently been discovered to have activity similar to MppP in

which an oxidized and hydroxylated product is formed (Figure 2.16) [109]. A distinctly

different PLP-, O2-dependent mechanism for oxidation has been proposed for CuaB most

recently. Similar to RohP, water is suggested to be the oxygen source, however CuaB

seems to perform an α-addition, decarboxylation, and hydroxylation in one catalytic cycle

(Figure 2.17) [110]. Cap15, a monooxygenase-decarboxylase from the capuramycin biosynthetic pathway, performs a very different oxidation reaction. This reaction converts an α-amino acid into a carboxamide. In the case of Cap15, an oxygen from O2 is

incorporated into the substrate in a PLP-dependent catalysis (Figure 2.18) [37].

Characteristic of most PLP-dependent reactions, PLP is thought to stabilize the conjugated carbanion, allowing a reaction with O2 owed to the oxidation potential of the stabilized

carbanion [108].

Katherine Ryan et al discusses how these PLP-dependent enzymes utilizing molecular oxygen as a cosubstrate are found multiple protein families. They suggest this may indicate that this type of catalysis could be directed in other non-related PLP-

dependent enzymes through directed-evolution experiments [111].

NH2 O NH2 O NH2 O H - - - H2N N O H2N N O H2N N O H H H NH N O2 H NH - -H H O HO -2 HO C -2 2 2 HO -2 OPO3 OPO3 OPO3 N N N H H H

H+

NH2 O NH2 O - NH2 O H N N O - 2 H N N O - H 2 H2N N O NH H O H NH 2 NH H2O -H HO -2 O OPO HO -2 2 2 3 OPO HO -2 3 OPO3 N H N H N H2

NH2 O - H2N N O H NH O NH2 O OH NH 2 H2O - - H N N O HO -2 H2N N O 2 OPO H H OH 3 OH NH2 O N H Figure 2.16 Proposed mechanism of RohP by Hedges et al. ACS Chem Biol 2018.

ACP O O O O H - O S - - O O R O NH NH NH R - H HO -2 HO C -2 HO -2 OPO3 OPO3 OPO3 N N N H H H

H2O CO2 H+ O O OH O O O O R R R

NH NH O2 NH

HO -2 HO -2 HO -2 OPO3 OPO3 OPO3 N N N H H H -H 2O2

O OH R NH

HO -2 OPO3 N H Figure 2.17 Proposed mechanism of CuaB.

OH O OH O OH O H - - - R O R O R O NH NH NH - H HO -2 HO C -2 HO -2 OPO3 OPO3 OPO3 N N N H H H

O2 O HOHO O HO O OH O O O - R + R O R O O H NH NH NH

HO -2 HO -2 HO -2 OPO3 OPO3 OPO3 N N N H H H -CO 2

OH + O H OH R OH O -H R NH O 2 NH HO -2 OPO3 HO -2 OPO3 N H N H Figure 2.18 Proposed mechanism of Cap15.

2.7 Conclusion

The chemical diversity capable solely through utilizing PLP as a coenzyme in

enzymology is impressive. Nature has evolved a multitude of ways to direct this reactivity

into branching mechanistic pathways. Understanding the full extent to which PLP is

utilized in nature is beneficial to both expanding the library of biosynthetic gene clusters

and natural products, as well as furthering the potential of PLP-dependent chemistry in

synthetic biology. Some discoveries in thermophilic PLP-dependent enzymes has opened

the possibility of using high temperatures and significant concentrations of organic solvents

(e.g. ethanol, methanol, DMSO, and acetonitrile) during catalysis, as these enzymes are able to retain catalytic activity. This opens the possibility of using PLP-dependent chemistry with a wider variety of nonpolar substrates [112, 113]. Classifying PLP- dependent enzymes based on sequence can be challenging. To streamline this, Stephanie

Sieber et al have worked on generating cofactor probes (PLP mimics) that cells can uptake and metabolize in situ. These probes can then bind to unknown PLP-dependent enzymes and, with NaBH4, be rendered irreversible. Through these studies, they found that 29% of

known PLP-dependent enzymes are still poorly characterized, with many more still undiscovered entirely [113, 114]. This means that more investigation into PLP-dependent enzymes has great potential.

EVIDENCE OF AN UNUSUAL MECHANISM FOR AN AMINOBUTYRIC ACID TRANSFERASE

3.1 Introduction

Highly-modified nucleosides are a diverse group of natural products. Their

biological activity can vary from antiviral reverse transcriptase and RNA polymerase

inhibitors to antibacterial translocase I inhibitors. One group of highly-modified

nucleosides inhibiting bacterial translocase I contains a disaccharide core that consists of a

5''-amino-5''-deoxyribose (ADR) glycosylated (5'S, 6'S)-5'-C-glycyluridine (GlyU). This

ADR-GlyU core is represented by FR-900493, caprazamycin B, A-90289 B, sphaerimicin

A, and muraymycin D1 and D2. All share a 6'-N-alkylamine linker that can be appended to a variety of other natural products such as glycosylated fatty acids, polyketides, or nonribosomal peptides. The ADR-GlyU biosynthesis has been described previously, wherein six enzymes form the disaccharide core using UMP and L-Thr [18, 27, 29-31].

Bioinformatic analysis suggests that this part of the biosynthetic pathway is shared among

the group [24].

Most recently, the 6'-N-alkylamine linker was found to have a similar biosynthetic

origin in these homologous nucleoside biosynthetic pathways. Utilizing a muraymycin-

producing mutant strain, Streptomyces NRRL30475, feeding experiments were conducted

with isotopically labeled precursors. Analysis of this study revealed L-Met to be the origin

of the C3N unit [82]. Mur24 (from the muraymycin biosynthetic pathway) was functionally assigned as a pyridoxine-5'-phosphate (PLP)-dependent AdoMet:phospho-ADR-GlyU aminobutyric acid transferase (ABTase), responsible for the γ-replacement of AdoMet by

breaking the Cγ-S bond and generating a new Cγ-N bond with the nucleoside intermediate

(Figure 3.1) [82]. Interestingly, the phosphorylated version of ADR-GlyU, a cryptic

intermediate formed by Mur28, was found to be the only adequate acceptor substrate of

Mur24 [32, 82]. Though not apparent from the structure, the same strategy is used to form the piperidine ring of sphaerimicin and A-90289 using SphL and LipJ, respectively [82].

OH H H H H2N COOH O N COOH O H O O Mur24 H N NH H2N O N NH O O O O H N H PLP H 2 O H2N AdoMet MTA O HO OH OH HO OH RO O OH ADR-GlyU (8) R = H HO P O 2- Mur28 3''-O-phospho-ADR-GlyU (9) R = PO3 OH 10

Figure 3.1 Functional characterization of Mur24 led to the discover of a PLP-dependent alkyltransferase that performs γ-elimination and γ-addition in a single active-site.

PLP offers incredible chemical versatility to enzymes, where protein structure can direct an array of reactivity. This is likely why PLP-dependent enzymes encompass 5 of the 6 enzyme classes categorized by the Enzyme Commission of the International Union of Biochemistry (including oxidoreductases, transferases, hydrolases, lyases, and isomerases), accounting for more than 230 distinct enzyme activities [34, 35, 37-42]. This activity is enabled by the highly conjugated electron sink of the PLP pyridine ring, allowing the stabilization of carbanions through delocalization of the negative charge throughout the conjugated system [34, 35, 37-45]. In total, around 4% of total enzyme activity utilizes

PLP [34-36].

The PLP-dependent alanine racemase mechanism has been explored extensively.

Through observing the kinetic isotope effects (KIEs), wherein heavy atoms with lower vibrational frequency are utilized to perturb reactivity, the proton transfer at the alpha

position was found to be the rate-limiting step with marginal KIEs reported between 1.2-

1.7 [40, 45, 59, 70, 71]. With the discovery of the new ABTase, along with knowing that

primary KIEs indicate a rate-limiting or subsequent product-determining step, we looked to understand the mechanism of this unusual gamma-replacement in a similar manner. In doing so, we propose that this new family of ABTases likely proceed through a γ- elimination, γ-addition mechanism, capable of a half-reaction uni-bi ordered sequential mechanism, wherein both paths release methylthioadenosine (MTA) as the first product.

Using AdoMet isotopologues, KIEs across a range of pH and temperatures highlight that the rate-limiting step is Cα deprotonation, though Cβ is also perturbed by a deuterium

isotope effect. Additionally, protein modeling and D2O “wash-in” experiments and D2O

isotope effects of the half-reaction allude to the possibility of two parallel pathways after

γ-elimination of MTA for offloading product in the absence of acceptor substrate (9) – both

a prototropic and sigmatropic rearrangement.

3.2 Materials and Methods

3.2.1 General experimental procedures

UV spectra were recorded on a FLUOstar Omega plate reader (BMG LABTECH

GmbH, Offenburg, Germany). An Agilent 6120 Quadrupole MSD mass spectrometer

(Agilent Technologies, Santa Clara, CA) equipped with an Agilent 1200 Series Quaternary

LC system and Eclipse XDB-C18 column (150 x 4.6 mm, 5 μM) was utilized for LC-MS

analysis. HR-ESI-MS spectra was recorded on an Agilent 6230 TOF equipped with an

Agilent 1260 Infinity II LC system and Phenomenex Gemini C18 column (100 x 3 mm, 3

μM). All solvents used were of LC-MS grade and purchased from Pharmco-AAPER

(Brookfield, CT). MassHunter’s Qualitative Navigator was used for all HRMS analysis.

FPLC was performed on a GE AKTA system with a Superdex 200 Increase 10/300 GL

size exclusion chromatography column.

3.2.2 Cloning, overexpression, and purification of proteins

The gene sphL was synthesized and provided by Zhaoyong Yang. The expression construct for the gene hMat2A was provided by Prof. Chunming Liu. The gene mur24 was

amplified from genomic DNA extracted from Streptomyces sp. NRRL30473, provided by

Dr. Zheng Cui. All genes were amplified by PCR using Phusion DNA polymerase from

Thermo Scientific with the supplied HF buffer (Thermo Scientific, Rockford, IL) and

primers synthesized by Integrated DNA Technologies (Coralville, IA; SI Table 3.1).

Purified PCR products were ligated into pET30 Xa/LIC (Novagen, Darmstadt, Germany), pET28a, or pXY200 using standard digestion and ligation protocol. Plasmids were confirmed by DNA sequencing (ACGT, INC, Wheeling, IL).

Plasmids pET28a-hMAT2a, pET30-lipJ, and pET30-sphL were transformed into chemically competent E. coli BL21(DE3) cells (New England Biolabs, Ipswich, MA).

Transformed strains were grown in LB media with 50 μg/mL kanamycin. Cultures were grown in 1 L LB media with 50 μg/mL kanamycin at 37 oC until cell density reached an

o OD600 of 0.4-0.6, at which point cultures were cooled to 18 C expression was induced with

0.1 mM IPTG. Cells were harvested after 12-18-hour incubation at 18oC, 250 rpm, and

lysed in Buffer A (100 mM KH2PO4, 300 mM NaCl, pH 8.0) using a Qsonica sonicator

(Qsonica LLC, Newtown, CT). Sonication protocol comprised of a 2 min sonicating time at 40% amplitude with 10 s pulses, separated by 50 s rest/cooling periods. Cell debris was pelleted using high speed (14000 rpm) centrifugation. Supernatant was then purified using

IMAC chromatography with HisPurTM Ni-NTA agarose (Thermo Scientific, Rockford,

IL). Undesired proteins were washed from the resin with low (10-20 mM) concentrations of imidazole spiked in Buffer A. The desired His-tagged protein was then eluted with 200 mM imidazole spiked Buffer A. Purified proteins were concentrated and a buffer exchange using Buffer B (25 mM KH2PO4, 100 mM NaCl, pH 8.0) was completed using forced

dialysis via Amicon Ultra 10000 MWCO centrifugal filters (Millipore, Burlington, MA).

This protein solution was drop frozen using liquid nitrogen and stored at -80oC. Extinction

coefficients were calculated using the ProtParam tool from ExPASY. Protein

concentrations were determined using a Nanodrop (Thermo Scientific, Rockford, IL) and

purity was assessed by 12% acrylamide SDS-PAGE.

Plasmid pXY200-mur24 was transformed into S. lividans TK24 using PEG-

mediated protoplast transformation and plated on R2YE media with 50 μg/mL apramycin.

After a 6-day incubation at 28oC, positive tranformants were confirmed by colony PCR

using InstaGene Matrix (Bio-Rad, Hercules, CA). This recombinant strain was then used

to inoculate 50 mL R2YE media with 50 μg/mL apramycin and incubated for 3 days at

28oC, 250 rpm. Protein expression was induced with 50 μg/mL thiostrepton and the culture was incubated in the same conditions for an additional 24 hours. The cells were then

'harvested and enzymatically lysed in Buffer A with 4.0 mg/mL lysozyme. After a 30 min incubation at 30oC, the suspension was physically lysed using a Qsonica sonicator (Qsonica

LLC, Newtown, CT). Sonication protocol comprised of a 4 min sonicating time at 40%

amplitude with 10 s pulses, separated by 50 s rest/cooling periods. Protein purification and

storage was as completed as previously described.

3.2.3 Assay of SphL, LipJ, and Mur24 half-reactions

Half-reactions were prepared with 50 mM KH2PO4, pH 8.0, 50 mM KCl, 10 mM

MgCl2, 2 mM aminobutyric acid donor (AdoMet, Ado-Se-Met, or S-methylmethionine),

40 μM PLP, and 275 nM ABTase (SphL, LipJ, or Mur24). Reactions were incubated at

22oC overnight.

Derivatizing agent was used to facilitate effective UV-vis absorbance and ionization for mass spectrometry. AQC (a derivatizing agent for amines) was used to bind the α-amino group of L-VG. The reaction was terminated with ACN in a 1:1 ratio. Proteins

were pelleted with centrifugation and supernatant was combined with 0.2 M sodium borate

buffer, pH 8.8 and 3.0 mg/mL AQC ACN solution in a 1:3:1 ratio. Mixtures were incubated

at 55oC for 10 minutes, cooled to room temperature, and transferred to autosampler vials.

Reactions were analyzed using LC-MS with the gradient shown (SI Table 3.2).

Assays for analysis of 2-oxobutyrate produced during the reaction utilized MBTH

(a derivatizing agent for α-keto acids) using an adapted method originally described by

Tanaka [115]. Reactions were terminated using ACN with 4% formic acid in a 1:1 ratio.

Proteins were pelleted with centrifugation and supernatant was combined with 8 mM

MBTH aqueous solution in a 2.5:1 ratio. Mixtures were incubated at 50oC for 30 minutes,

cooled to room temperature, and transferred to autosampler vials. Reactions were analyzed using LC-MS with the gradient shown (SI Table 3.2).

3.2.4 Mur24 deuterium exchange assay

For deuterium exchange assays, AdoMet isotopologues were generated in situ using ATP, methionine, and hMAT2A. Reactions were prepared with 50 mM KH2PO4, pH

8.0, 50 mM KCl, 10 mM MgCl2, 2 mM ATP, 1 mM L-Met isotopologue (L-Met, L-Met-2-

d1, L-Met-3,3,4,4-d4, L-Met-2,3,3,4,4-d5-methyl-d3, L-Met-S-methyl-d3, or L-Se- Met), 40

μM PLP, 1 μM hMAT2A and 275 nM ABTase (SphL, LipJ, or Mur24). D2O “wash-in”

o half-reactions were performed in 62% D2O. Reactions were incubated at 22 C overnight

and terminated using ACN with 4% formic acid (1:1 ratio to reaction solution). MBTH derivatizing protocol and LC-MS gradient described previously was used for analysis.

3.2.5 Determining the regiochemistry of isotope exchange

The product of the Mur24 deuterium exchange assay utilizing the L-Met

isotopologue, L-Met-3,3,4,4-d4, led to the loss of a single deuterium. In order to determine the location of this deuterium exchange, reactions were subjected to HCl-catalyzed tautomerization (1.5 M HCl, incubated for 1 hour) in conjunction with HRMS.

Tautomerization in water led to washing out any deuteriums at the beta-position of the 2- oxobutyrate product, leaving only the deuterium(s) at the gamma-position.

3.2.6 MTA product-inhibition assays

Michaelis-Menten kinetics were measured with similar reaction conditions for

Mur24. AdoMet was generated in situ with hMAT2A as described. [L-Met] included 0, 2,

3.9, 7.8, 15.6, 62.5, and 125 µM. This series of kinetics was repeated with 0, 1, and 10 mM

MTA. Obtained kinetics were converted to a Lineweaver-Burke plot for data interpretation.

3.2.7 Kinetics of Mur24

Reactions were prepared as previously described. For kinetics, however, a 40 min

timepoint was used for all reaction incubations to capture the initial velocity as determined

by a time-course assay and generation of AdoMet in situ via hMAT2A was ensured to not

limit the reaction rate (SI Figure 3.1). Additionally, a nine-point standard curve was

generated using commercial 2-oxobutyrate ranging from 20 nM to 10 μM. Standards and

reactions were derivatized with MBTH. Reactions were analyzed using HR-MS. Peak area

from extracted ion chromatograms (using both stereoisomer peaks, see SI Figure 3.5) was

converted to [2-oxobutyrate] (nM) using the standard curve. GraphPad Prism 7 was used for all kinetic analysis. Additionally, kinetic assays were repeated at pH 7.0 and 9.0, as

well as 28 and 37oC temperatures.

3.2.8 FPLC analysis of SphL and LipJ

Proteins were overexpressed and purified as described (Error! Reference source not found.). Rather than being drop frozen, however, proteins were kept on ice while FPLC

was prepared. An isocratic flow of Buffer C (20 mM HEPES, 250 nM NaCl, pH 8) at 0.4-

0.6 mL/min was used to maintain a pressure of no more than 3.0 mPa2 and 280 nm wavelength was monitored. Monomeric protein standards were used to create a standard curve including lysozyme, BSA, and catalase (14.4, 66.5, and 232 kDa, respectively).

Following elution of all protein standards, protein of interest was injected, and 1.0 mL fractions were collected until all eluents were expelled. SDS-PAGE was used to confirm the collection containing the protein of interest. The native size of the protein was calculated using the elution time of appropriate collection and the generated standard curve.

Calculated protein native size was divided by denatured monomeric size and rounded to the nearest whole number to give the number of units in complex to create the quaternary structure.

3.2.9 Protein modeling of SphL and LipJ

Protein modeling was initially processed using the Phyre2 web portal [116]. From

these results, models with the best alignment coverage were analyzed in PyMOL to probe

differences and similarities in the overall tertiary structure and the residues that look to be

critical in the active site.

3.3 Results and Discussion

3.3.1 Tracking proton-exchanges throughout catalysis

Typical of many gamma-replacement enzymes, we found that in the absence of the alkyl-acceptor substrate the enzyme performs an elimination half-reaction. ABTases

Mur24, SphL, and LipJ are all capable of converting AdoMet into methylthioadenosine

(MTA) and an aminocrotonate intermediate that is spontaneously converted to 2- oxobutyrate in the presence of water. We proposed that this encompassed all deprotonation events that could account for rate-limiting steps, with only an aza-Michael-like addition and cascade of reprotonation to follow. Since the alkyl-acceptor nucleoside intermediate is synthetically limited, we opted to explore the mechanism of these novel ABTases using the half-reactions. Additionally, Mur24 was used in all mechanistic studies, as it was found to be the most active homolog.

Using AdoMet isotopologues generated in situ using hMAT2A and L-Met

isotopologues, we found that L-Met (I) and L-Met-2-d1 (II) reactions formed a product

mass of 264 m/z (the mass of MBTH-derivatized 2-oxobutyrate, Figure 3.2). L-Met-

3,3,4,4-d4 (III) and L-Met-2,3,3,4,4-d4-Mt-d3 (IV) gave a product mass of 267 m/z (Figure

3.2, cont’d), suggesting that one proton at either the beta- or gamma-position is exchanged throughout catalysis. To determine the regiochemistry of this proton exchange, HCl- catalyzed tautomerization was utilized following the enzymatic reaction. This led to enolate formation and an exchange of protons between the beta-position of 2-oxobutyrate and water. Subsequence product-derivatization and LCMS analysis revealed that 26% of

2-oxobutyrate [264+3 m/z] was converted to [264+2 m/z], indicating that the enzymatic proton exchange also occurred at the beta-position (SI Figure 3.2).

Figure 3.2 Mass spectra of varying L-Met isotopologues enzymatically converted to AdoMet via hMAT2A, subsequently converted to 2-oxobutyrate by Mur24, and lastly derivatized with MBTH.

Performing the half-reaction in D2O led to an intriguing mass spectrum (Figure

3.3), where species containing anywhere from 0-5 deuterium atoms was detected. The

primary product of [264+2 m/z], in conjunction with the presence of D2O significantly perturbing Mur24 kinetics (see 3.3.2), suggests a possible 1,5-sigmatropic rearrangement

that occurs when the acceptor (nucleoside-intermediate) substrate is not present. Additional protein modelling (SI Figure 3.11-3.12) aides in the discussion of this possibility in the conclusion of this chapter.

3000

2000

Intensity 1000

0 260 262 264 266 268 270 m/z

3.3.2 Rate constant determination and kinetic isotope effects

Next, single-turnover kinetics were observed with respect to both AdoMet and

Mur24. This showed that the half-reaction is 1st-order with respect to each (Figure 3.4, based on the log-based kinetics graphed), with an overall 2nd-order rate constant of (3.9 ±

0.8) x 10-6 nm-1 min-1 (Figure 3.5). Michaelis-Menten kinetics revealed a similar 2nd-order

-6 -1 -1 - rate constant (5.6 x 10 nm min ), with a Km of 3.3 ± 1 µM and a kcat of (19 ± 0.9) x 10

3 min-1 (SI Figure 3.8).

KIEs of various AdoMet isotopologues were measured with respect to 1st-order

nd (kcat) and 2 -order (kcat / Km) kinetics (Figure 3.6, Table 1.1). Protonation tracked via

endpoint assays described previously only allow a snapshot of before and after catalysis;

KIEs allow better understanding of protonation states throughout catalysis. Regarding L-

Met-2-d1, only a DL-mixture (1:1) was readily available. Since hMAT2A is only capable

of converting the L-stereoisomer into AdoMet, any DL-Met-2-d1 concentrations were

doubled. To ensure that DL-mixtures do not interfere with generating in situ AdoMet

isotopologues and subsequently measuring KIEs, L-Met-3,3,4,4-d4 KIE was compared to

2X DL-Met-3,3,4,4-d4, wherein essentially no variation between the two was detected (SI

Figure 3.7). KIEs were measured at 22, 28, and 37oC, as well as pH 7, 8 (optimal enzyme

activity), and 9 (Scheme 1.1). KIEs of Cα- and Cβ-deprotonation were more significant at lower pH and higher temperatures with respect to 2nd-order rate constants. 1st-order KIEs

were much tighter across the varying conditions, though pH still greatly affects 1st-order

KIEs. KIEs were consistently higher when Cα-deprotonation was perturbed, although

perturbation of Cβ-deprotonation also led to a lesser KIE. Based on this varying KIE data

(Scheme 1.1), we propose that proton-abstraction from the donor substrate is rate-limiting

at lower pH environments. As pH is increased, steps after elimination of MTA may

contribute more to kcat instead, as kinetics of the half-reaction in D2O are greatly perturbed

(SI Figure 3.6). This may occur through a combination of 1,5-sigmatropic and 1,5-

prototropic rearrangement (Figure 3.7).

Subjecting the enzymatic reaction to a dilution curve of MTA reveals product

inhibition that noncompetitively inhibits the half-reaction (SI Figure 3.10). This suggests that MTA is the first product released during catalysis (Figure 3.7). With this proposed mechanism in mind, the protonation state of the active site residues is never fully reestablished, meaning that further catalysis must rely on a thermodynamic equilibrium of proton exchange with the active site residues and water. This is likely why half-reaction

(uni-bi ordered sequential) activity is drastically reduced compared to the full gamma- replacement (bi-bi ordered sequential or double-displacement) activity. Reinforcing this notion, adding more enzyme to endpoint activity assays (24-hour reactions) shows that more enzyme leads to more product even when equilibrium has been obtained (SI Figure

3.9). This suggests that during half-reaction kinetics, Mur24 is acting as a pseudo-substrate.

Another possible explanation of hindered activity with respect to the half-reaction is the need to for the L-vinylglycine-PLP intermediate to rotate from s-trans to s-cis (a less energetically favored orientation) as shown in Figure 3.7 in order for the half-reaction to go to completion.

Figure 3.4 Single turnover kinetics of ABTase half-reaction (n = 2). Reactions varying [Mur24] (●) included 50 mM KH2PO4, pH 8; 50 mM KCl; 10 mM MgCl2; 40 µM PLP; 500 µM AdoMet; and 240 nM, 480 nM, or 960 nM Mur24. Reactions varying [AdoMet] (■) included identical conditions, apart from 960 nM Mur24 and 125 nM, 250 nM, or 500 nM AdoMet. Slopes of the linear regressions were 1.2 and 0.96, respectively.

Figure 3.5 Calculation of rate constant with respect to both Mur24 (α) and AdoMet (β) using the rate equation shown. The half-reaction was determined to be essentially 1st- order with respect to both α and β (the slope of each respective linear regression from Figure 3.3, indicating an overall 2nd-order reaction). The 2nd-order rate constant was calculated using this data.

Figure 3.6 Michaelis-Menten kinetics of Mur24 using methionine isotopologues. Reactions included 50 mM KH2PO4, pH 8; 50 mM KCl; 10 mM MgCl2; 40 µM PLP; 2 mM ATP; 1 µM hMAT2A; 275 nM Mur24. [L-Met isotopologues] used to calculate Michaelis-Menten curve included 0 µM, 2.0 µM, 3.9 µM, 7.8 µM, 15.6 µM, 62.5 µM, and 125 µM. *DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L-Met, concentrations were doubled in reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L-Met isotopologue).

Table 3.1 Kinetic parameters and kinetic isotope effects determined with respect to five isotopologues. k* = kcat / Km **DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L-Met, concentrations were doubled in reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L-Met isotopologue).

-1 kcat (min ) Km (μM) kcat/Km (k*) kcat(H)/ kcat(D) k*(H)/k*(D)

L-Met (9.2 ± 0.5) x 10-2 3.4 ± 0.7 27 x 10-3 ------L-Met-2-d1** (5.2 ± 0.4) x 10-2 3.5 ± 1.1 15 x 10-3 1.8 1.8 L-Met-3,3,4,4-d4 (7.9 ± 0.5) x 10-2 3.9 ± 0.9 20 x 10-3 1.2 1.3 L-Met-2,3,3,4,4-d5 (4.8 ± 0.3) x 10-2 2.1 ± 0.6 23 x 10-3 1.9 1.2 L-Met-3,3,4,4-d4** (7.3 ± 0.4) x 10-2 3.1 ± 0.8 23 x 10-3 1.3 1.1

pH-dep 1st order KIE pH-dep 2nd order KIE 3 6 L-Met-2-d1* L-Met-2-d1* )

L-Met-3,3,4,4-d4 ) L-Met-3,3,4,4-d4 (D) (D) cat 2 L-Met-2,3,3,4,4-d5 4 L-Met-2,3,3,4,4-d5 k k* / / (H) (H) k* cat k 1 2 k* = kcat/Km KIE ( KIE (

0 0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 6.5 7.0 7.5 8.0 8.5 9.0 9.5 pH pH temp-dep 2nd order KIE temp-dep 1st order KIE 15 2.5 L-Met-2-d1* L-Met-2-d1*

) L-Met-3,3,4,4-d4 )

2.0 L-Met-3,3,4,4-d4 (D) (D) 10 L-Met-2,3,3,4,4-d5 k* / cat L-Met-2,3,3,4,4-d5 k /

1.5 (H) (H) k* cat

k 1.0 5 k* = kcat/Km KIE (

KIE ( 0.5

0.0 0 20 25 30 35 40 20 25 30 35 40 o o Temperature ( C) Temperature ( C)

Figure 3.7 KIE under varying pH and temperature conditions. *DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L- Met, concentrations were doubled in reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L- Met isotopologue.

3.4 Conclusion

AdoMet is capable of forming an aldimine with PLP (AdoMet pKa 7.7; optimal enzyme activity and aldimine formation is pH 8). This, along with Mur24 and SphL not copurifying with PLP, leads us to propose that AdoMet and PLP may covalently form an aldimine before entering the enzyme active site (I) shown in Figure 3.7. Based on KIEs

measured, we propose that at low pH the deprotonation of Cα is rate-limiting (III), though

Cβ deprotonation is also perturbed in KIE studies (VI). Since a quinonoid state has not been

detected in the lab, we propose that this state of PLP is short-lived in the catalysis (IV).

Based on structural and biochemical evidence in methionine-γ-lyase and cystathionine-γ-

synthase studies, we believe the enamine is in E-configuration when MTA is eliminated

(VII). Therefore, the 1,4-conjugated L-vinylglycine-PLP intermediate is initially in the s- 52

trans configuration, which is kinetically favored for the full-reaction (path a). At higher pH conditions, we propose that rearrangement following MTA-elimination become rate- limiting in the half-reaction (path b), as evidenced by D2O kinetics (SI Figure 3.6) and

deuterium incorporation in overnight reactions (Figure 3.3). This may occur through a 1,5-

sigmatropic rearrangement if the intermediate is rotated to the thermodynamically-favored

s-cis configuration (IX) or, alternatively, a 1,5-prototropic rearrangement whereby C4' is deprotonated by a protein residue and a proton is picked up at Cγ from the aqueous

environment or an active site residue. Either rearrangement will lead to the PLP-bound aminocrotonate intermediate (X) that can be offloaded and spontaneously transformed into

2-oxobutyrate in aqueous solution.

ABTases SphL and Mur24 share several active site residues with ACC synthases, which they share primary and secondary sequence similarity with and are in closely connected clades in phylogenetic analysis [82]. Modeling based on the crystal structure of

ACC synthase 1YNU shows that an arginine residue appears coordinate the carboxylate of

AdoMet (R411 of SphL and Mur24; R407 of 1YNU). Another arginine appears to

coordinate the phosphate of PLP (R242 of SphL and Mur24; R281 of 1YNU). A lysine is

in the position appropriate for Cα deprotonation characteristic of PLP-dependent enzymes

(K234 of SphL and Mur24; K273 of 1YNU). Based on modeling, no other key residues

appear within a range possible to facilitate the acid-base chemistry required for the alkyl

transfer. Since the lysine responsible for internal aldimine formation has been shown to

participate in α-proton exchange in other enzymes [44, 49, 60, 72], we propose that K234

may be responsible for both α- and β-proton exchanges that occur throughout the catalytic

cycle of ABTases, well as protonation of the oxyanion of PLP as it enters the active-site

(III, VI, and II, respectively in Figure 3.7). Another possibility is that there is cofactor- facilitated proton-exchanges occurring, whereby the phosphate of PLP can participate in catalysis. As determined by phylogenetic analysis, ABTases are in a distinct clade from most ACC synthases. ABTases SphL and Mur24 have a glycine in place of a glutamate in the active-site where substrates would bind to interact with PLP (G44, G41, and E47 for

SphL, Mur24, and 1YNU, respectively). Additionally, SphL and Mur24 have a phenylalanine in place of a tyrosine. This creates a more hydrophobic pocket in a region of the active-site (3D modeling in SI Figure 3.12) that has an impact on protonation-states of

PLP (Figure 3.7). SphL and Mur24 also lack an alpha helix near where active-site chemistry would take place (SI Figure 3.11). Lastly, ACC synthase 1YNU possesses a glutamate residue near the active Lys273 (circled red in Figure 3.7), whereas Mur24 has a valine residue; this likely means that Mur24 Lys234 is not pulled away from catalysis by electrostatic interactions that would typically occur in the ACC synthase. Together, these differences suggest how ABTases may have evolved to have distinct activity from ACC synthases, allowing a bisubstrate reaction to occur. Interestingly, gel-filtration data suggests ABTases are natively homo-tetramers (SI Figure 3.3-3.4), similar to cystathionine-γ-synthases rather than ACC synthases, which are homo-dimers.

Lys Lys Ad Gluacc O Lys H Arg Arg HO S H a H a S S B: H2N N S HB H2N N O R R O R HO O Arg H O NH O 2 NH2 O NH3 R O O NH O N N N O Phe O Phe O O P O Hydrophobic O O P O pocket Gly O P OH Hydrophobic O P OH Hydrophobic O O O pocket O pocket O R Ile O Gly O N N Asp N Ile N H H H O O I II III IV O O Asp Asp

Lys 9 9 Lys Lys + Arg Arg H N b 3 H2N b B: BH Arg H H a :B H S :B H N N S H2N N HB H N N R 2 R S 2 O O R H NH2 O NH2 E NH2 O Conformation change with O Arg O H a H Arg N cosubstrate NH H N N a O 2 NH H2N N binding O O O P OH Hydrophobic O P O NH2 NH O O pocket O O P O 2 O O O N H N N H H O V O VI O VII O O Asp O Asp Asp

Lys Lys 9 Arg b Arg H N 9 b 2 (a) H HB BH H2N N BH H H N HB H N N (a) 2 2 O NH MTA 2 O NH2 (b) O H Arg (a) (a) a O Arg NH H2N N H a 10 O NH H2N N Reverse sequence O O P O NH2 O O O P O NH2 O O N H N VIII H O O O Asp O Asp

(b)

Lys Lys

Arg Arg b b H H B: HB B: HB H2N N H2N N O O NH2 NH2

O Arg O Arg H a H a O O O H NH H2N N NH H2N N H O O O 2 Reverse O O O NH NH O P O 2 O P O 2 sequence NH O O O O 3 NH2 O N N H H O O IX X O O Asp Asp Figure 3.8 Full catalytic cascade with multiple pathways proposed for Mur24: Since the enzyme does not copurify with PLP, we propose AdoMet covalently attaches to PLP forming the aldimine externally (I); as the AdoMet-PLP complex enters the active site, the oxyanion of PLP deprotonates Lys234 as it enters a partially hydrophobic pocket (II); once AdoMet-PLP complex is in position, Cα is deprotonated (III) and the quinonoid state of PLP is short-lived as C4' picks up a proton from Lys234 (IV); in the presence of the acceptor substrate (9), a conformational change in the active site occurs positioning the hydroxy-group of the pyridine away from the hydrophobic pocket and toward one of two arginine residues (V); this leads to the protonation of the ketimine, generating an iminium and oxyanion – consequently lowering the pKa of Hβ proton and allowing for the deprotonation of Cβ (VI); this leads to the E configuration until MTA is eliminated and the s-trans configuration is generated (VII); when the acceptor substrate is present (9) an aza- Michael-like addition occurs (path a of VIII) and electron movement is reversed affording

10; when no acceptor substrate is present to generate the kinetically favorable product, the thermodynamically-favored s-cis product is formed (path b of VIII); the new orientation facilitates a 1,5-sigmatropic rearrangement to occur (IX) leading to the PLP-bound aminocrotonate (X); alternatively, a 1,5-prototropic rearrangement could occur in path b whereby C4' is deprotonated by a protein residue and a proton is picked up at Cγ from the aqueous environment or an active site residue.

Though the enzyme appears to be irreversible, the full reaction product is capable

of proton-exchange at the α- and β-positions. However, no proton-exchange occurred with

2-oxobutyrate nor L-vinylglycine, likely because the true enzymatic product is the unstable

aminocrotonate that readily tautomerizes to the imine and exchanges with water to form 2-

oxobutyrate in aqueous solution.

As earlier studies suggested, ABTase substrate specificity remains strict. Using

AdoMet with various S-nucleophiles as acceptors (e.g. L-Cys and 2-mercaptoethanol) demonstrated no gamma-replacement activity. This means that strengthening the nucleophile from the native nitrogen of the phospho-nucleoside intermediate (9) to a sulfur does not overcome substrate specificity alone.

There is a growing body of interesting PLP-dependent gamma-position chemistry.

Most established of these is the cystathionine gamma-synthase (CGS) and lyases (CGL) involved in transsulfuration and reverse transsulfuration, respectively [73, 74]. CGS catalyzes the gamma-replacement of O-succinylhomoserine, eliminating succinate and replacing it with cysteine for form cystathionine [47, 75]. CGL eliminates cysteine from cystathionine, releasing 2-oxobutyrate and ammonium [74]. Mirroring the KIEs measured with the newly discovered ABTases, plant and bacterial CGSs catalyze the replacement of substituents at the γ-position of amino acids through a common α-proton exchange as well

as a rapid β-proton exchange [76]. LolC, a critical enzyme in loline alkaloid biosynthesis,

is proposed to have similar activity to CGS, breaking a Cγ-O bond, but creating a new Cγ-

N bond rather than a new Cγ-S bond [80]. Herein, we provide further evidence that

ABTases such as Mur24 are employing a similar mechanism to catalyze Cγ-S bond

breaking and Cγ-N bond formation – a greater feat in regards to nucleophilicity compared to the CGS Cγ-O bond is breaking and Cγ-S bond formation [81, 82]. Similar to loline alkaloid biosynthesis, a simple methyltransferase-like SN2-type substitution looks possible

(such as that seen in polyamine biosynthesis like spermidine [82, 84]), but the chemical

strategy to transfer this group is much more intricate. Recently, utilization of PLP-

dependent gamma-replacement catalysis has been discovered in the formation of alkyl-

substituted pipecolate structures in citrinadin and fusaric acid biosynthesis (CndF and

Fub7, respectively). Cγ-O bond breaking and Cγ-C bond formation of CndF and Fub7 is

similarly uses PLP-dependent gamma-replacement catalysis, however this is done with

somewhat more substrate flexibility than the ABTases [85, 87, 88]. Intriguingly, Fub7

utilizes an aldehyde substrate as a nucleophile (in the tautomerized state), rather than using

it as an electrophile [87]. Clearly, this family of enzyme activities has a lot left to be

uncovered. Though the general mechanism appears to be similar thus far – the nucleophiles

and electrophiles used, the substrate specificity, and the elementary rates constants

governing rate-limiting and subsequent steps can all vary greatly. This work adds a better

mechanistic understanding to this growing body of PLP-dependent gamma-replacement

enzymes.

3.5 Supporting Information

SI Table 3.1 Primers used for cloning sphL. Primer Sequence Note sphL_pET30A_FWD 5'-GGTATTGAGGGTCGCATGAGTGTGCCGCTGGATG -3' Used for SphL sphL_pET30A_REV 5'-AGAGGAGAGTTAGAGCCGTTAAGCTGCCAGGTCCGG-3' expression

SI Table 3.2 LCMS gradient method used to detect 2-oxobutyrate-MBTH. H2O, 0.1% ACN, 0.1% Time Flow formic acid formic acid 1 0 min 95% 5% 0.4 mL/min 2 10 min 20% 80% 0.4 mL/min 3 16 min 20% 80% 0.4 mL/min 4 17 min 95% 5% 0.4 mL/min 5 22 min 95% 5% 0.4 mL/min

SI Figure 3.1 hMAT2A coupled reaction to make AdoMet in situ does not affect the initial velocity of ABTases when compared to using a commercial AdoMet stock.

40000 1.0 0 M HCl 1.5 M HCl 30000

* p < 0.000004 0.5 20000 Intensity * 10000 Relative Abundance

0.0 0 260 265 270 264 265 266 267 268 269 270 m/z m/z (rounded to whole number)

NH COOH COOH N 2 D D - MTA D D COO O HO D D N N O D +H N S+ N PLP 3 dil.HCl D D R2 ABTase OR OR HO OH COOH COOH O D HO D D D

SI Figure 3.2 MBTH-derivatized product from HCl-catalyzed tautomerization of 2- oxobutyrate isotopologue formed from AdoMet-3,3,4,4-d4. After biocatalysis, one deuterium was lost to proton exchange in H2O. After HCl-catalyzed tautomerization (26% conversion after isotope distribution correction, assuming no β, γ-hydride shift), one additional deuterium was lost. This indicates that biocatalysis and HCl-catalyzed tautomerization each led to one proton being exchanged at the β-position throughout the alkyl-group modifications. (Note: L-Met isotopologue stocks used were 99% pure, accounting for the minor amounts of natural isotopic abundancies.)

SphL LipJ

75 kDa 75 kDa

50 kDa 50 kDa 37 kDa 37 kDa 25 kDa 25 kDa 15 kDa 15 kDa

SI Figure 3.3 SDS-PAGE gel of IMAC-purified SphL and LipJ, both His-tagged recombinant proteins heterologously expressed in E. coli BL21 cells.

10 Elution (mL) 5

1.0 1.5 2.0 2.5 log(MW), (kDa)

SI Figure 3.4 Commercially available lysozyme, bovine serum albumin, and catalase were used to construct a standard curve (R2=0.9981, n = 2). The linear regression was used to approximate the native sizes of both SphL and LipJ (176 kDa and 210 kDa, respectively). Thi This roughly corresponds to a tetrameric complex for the quaternary structure of each enzyme.

CH O CH3 3 N N OH + N N N C S S O NH2 C OH + O MBTH

CH3 N O N C OH S N C

8×106

6×106

4×106 Intensity 2×106

0 0 5 10 15 20 25 Time (minutes)

SI Figure 3.5 Following MBTH derivatization of 2-oxobutyrate, two stereoisomers are formed with distinct retention times (extracted ion chromatogram shown). For kinetic studies, total peak area was used in all calculations.

0.10 H2O

) 0.08 D2O -1

0.06

0.04 /[Enz] (min o

V 0.02

0.00 0 50 100 [Substrate] (µM)

SI Figure 3.6 Half-reaction of Mur24 in D2O significantly lowers the kcat (kcat(H) / kcat(D) = 2.4), suggesting that the active site is exposed to the aqueous environment throughout half-reaction catalysis and steps following MTA-elimination can perturb kinetics. Km could not be accurately determined from the data points collected.

0.10 L-Met-3,3,4,4-d4

) 0.08 L-Met-3,3,4,4-d4* -1

0.06

0.04 /[Enz] (min o

V 0.02

0.00 0 50 100 [Substrate] (µM)

SI Figure 3.7 Michaelis-Menten kinetics comparing pure L-Met-3,3,4,4-d4 to a DL-Met- 3,3,4,4-d4 mix. *DL-Met was used in place of pure L-Met; enantiomeric purity approximately 1:1; since hMAT2A only utilized L-Met, concentrations were doubled in reactions (i.e. 250 µM DL-mixture used to obtain 125 µM L-Met isotopologue). This demonstrates that a DL-Met-2-d1 mix as opposed to pure L-Met-2-d1 in kinetic studies has no unintended impact on KIE studies.

0.025 L-Met

) 0.020 L-Se-Met -1

0.015

0.010 /[Enz] (min o

V 0.005

0.000 0 50 100 250 [Substrate] (µM)

SI Figure 3.8 Michaelis-Menten kinetics comparing L-Met to L-Se-Met (no statistically significant difference in any kinetic parameters present, suggesting elementary step of MTA leaving is not rate-limiting).

15000

10000

5000 [2-oxobutyrate] (nM) 0 200 400 800 [Mur24] (nM)

SI Figure 3.9 End-point assay of a 24-hour reaction with varying [Mur24]. This indicates that the enzyme is acting as a pseudo-substrate, whereby more enzyme will lead to more product, even once the reaction has reached equilibrium.

60 0 mM MTA 50 1 mM MTA 10 mM MTA o 40 1/V 30

-0.2 0.0 0.2 0.4 0.6 1/[Substrate]

SI Figure 3.10 Lineweaver-Burke plot of MTA product-inhibition. Considering standard error, the mode of inhibition looks most likely to be noncompetitive, indicative of MTA being the first product released during catalysis.

SI Figure 3.11 Predicted SphL structure modelled after an ACC synthase (PDB: 1YNU) with PLP. Circles black is an alpha-helix in 1YNU that is not present in the model of SphL

SI Figure 3.12 Binding pocket of SphL modelled using ACC synthase (PDB: 1YNU) with PLP. Most notable difference is the greater hydrophobicity in a region of SphL active-site compared to 1YNU.

DOWNSTREAM BIOSYNTHESIS OF SPHAERIMICIN: FORMATION OF THE

DIHYDROXYLATED PIPERIDINE RING

4.1 Introduction

With the functional and mechanistic characterization of ABTases described by

Zheng et al [82] and in Chapter 3 of this dissertation, it is now known that all ADR-GlyU-

containing highly-modified nucleosides contain a commonly derived N-alkylamine linker

– achieved biosynthetically via PLP-dependent gamma-replacement catalysis using

AdoMet as a cosubstrate. Clearly, from this point the biosynthetic routes vary widely in additional tailoring (apparent from the structures of the mature natural products). This chapter describes some of the preliminary bioinformatics and biochemical characterization of downstream biosynthesis in forming the dihydroxylated piperidine ring of sphaerimicin.

Lastly, a proposed pathway based on this data is discussed.

4.2 Materials and Methods

4.2.1 General experimental procedures

See 3.2.1.

4.2.2 Cloning, overexpression, and purification of proteins

See 3.2.2. Mutants H117A and D119A were made using primers generated by IDT

(SI Table 4.1).

4.2.3 Activity assay of SphK and SphN

Wildtype and mutant SphK assays were set up using 500 µM subsrate, 1 mM 2-

oxobutyrate, 50 µM Fe(II), and 500 nM of recombinant protein. Reactions were incubated

overnight. Proteins were precipitated using 1:1 water:acetonitrile. Reactions were then analyzed on both an Agilent 6120 Quadrupole and Agilent 6230 TOF using the HILIC column.

4.2.4 FPLC analysis of SphK and SphN

See 3.2.8.

4.3 Results and Discussion

4.3.1 Functional assignment of SphK as a β-hydroxylase

A putative β-hydroxylase resembling a L-proline cis-4-hydroxylase [117, 118],

SphK, has shown activity with 1A to form 2 (Figure 4.1). This suggests that an 1A is an intermediate in sphaerimicin biosynthesis and that hydroxylation occurs after N- alkylamine transfer (HPLC traces shown in Figure 4.2) catalyzed by SphL, the ABTase 66

homologue described in Chapter 3. Due to limited substrate, stereochemistry of the

hydroxyl-incorporation was not determined. Hydroxylation likely doesn’t occur prior to

the alkyltransfer, as SphK has no activity with AdoMet (1C). Additionally, SphK has no

activity toward decarboxylated-1A (1B), suggesting that hydroxylation immediately

follows alkyltransfer in the biosynthetic pathway. Based on secondary sequence alignment

predicted by the HHpred online-tool, SphK shares several critical residues with a proline-

3-hydroxylase (Table 4.1). This includes an iron(II)-binding motif (HXDXnH) and a 2- oxobutyrate binding-residue. Successful mutations of two of these residues shows significantly hindered activity in overnight-reactions (Figure 4.3). This is likely due to poorer coordination of the cofactor in the active-site, typical of other Fe(II)-dependent mutant proteins. Quaternary structural analysis using gel-filtration showed that SphK is natively a homo-trimeric complex.

O O OH OH α-KG Succ OH H H NH H H NH HOOC N O CO HOOC N O 2 2 O H N O H N O NH2 O NH2 O O Fe(II) O 1A O O H SphK H H2N H2N HO OH HO OH RO OH RO OH 2 1B H2N

NH2 - N COO N N 1C + + O H3N S N

HO OH

Figure 4.1 Substrates tested with SphK in the presence of 2-oxobutyrate, iron(II), and oxygen.

Figure 4.2 HPLC trace of SphK activity: the sole substrate found active with SphK was nucleoside-intermediate 1A.

Table 4.1 Binding motifs predicted in SphK based on a homologous protein (proline 3- hydroxylase; PDBe 1E5R).

Protein Fe(II) Binding Motif α-KG Binding

HXDXnH Residue P3H H107 D109 H158 R168 SphK H117 D119 H162 R172

150

100

50 Relative activity

H117A D119A Wildtype Figure 4.3 Relative activity of alanine-mutant SphK proteins compared to wildtype.

4.3.2 Functional assignment of SphN as a putative transketolase

Based on primary and predicted-secondary sequence alignment, SphN is a classified as a putative transketolase. In order to test this, a real-time assay of the half-reaction was examined. Due to the unavailability of the acceptor (nucleoside intermediate), full catalytic activity was not demonstrated. Using gel-filtration, SphN quaternary structure was determined to be natively monomeric.

Figure 4.4 Canonical mechanism of a transketolase double-displacement reaction, wherein a 2-carbon donor covalently binds the thiamine pyrophosphate coenzyme (I); this 2-carbon unit will be added to the acceptor substrate (II). In this assay, the half- reaction measured using a real-time enzyme-coupled assay, whereby glyceraldehyde 3- phosphate generated as the product after xylulose 5-phosphate modification; this assay utilized catalytically-efficient, commercially-available triosephosphate isomerase and glycerophosphate dehydrogenase in order to detect the oxidation of NADH using UV-vis measurements at 340 nm.

1.1 1.1 No transketolase No transketolase E . co li TK (250 ng/uL) SphN (0.28 units/L) 1.0 SphN (24.5 ng/uL) 1.0 SphN (2.8 units/L) n=3 SphN (28.0 units/L) 0.9 0.9 n=3 Relative Abs (340 nm) Abs (340 Relative Relative Abs (340 nm) Abs (340 Relative

0 5 10 0 5 10 Time (minutes) Time (minutes)

Figure 4.5 Relative activity of SphN based on UV-vis-based real-time enzyme-coupled assay: Initial velocity of SphN compared to a commercially-available positive-control E. coli transketolase (Left) and initial velocity of SphN using varying enzyme concentrations.

4.4 Conclusion

Hydroxylation of the N-alkylamine linker (product of SphL) has been demonstrated by the now functionally characterized SphK. We propose that this is followed by a PLP- dependent decarboxylation (Figure 4.6), based on bioinformatics and the activity of homologous Mur23 (from the biosynthesis of the similar highly-modified nucleoside antibiotic, muraymycin). Next, we propose that an oxidative deamination of the N- alkylamine linker via SphO. This enzyme is putatively FMN-dependent and has similarity to other enzymes capable of oxidative deamination. This would be an intriguing addition to the biosynthetic pathway, as an oxidative dephosphorylation has been characterized earlier in the pathway, but never an oxidative deamination. This would make a suitable aldehyde that could be used as a 2-carbon acceptor substrate for the putative transketolase,

SphN. Partial activity of SphN has been demonstrated in this chapter. This would account for the 5-carbon scaffold (see Figure 4.6) necessary to form the piperidine ring shown in the mature natural product, sphaerimicin.

Figure 4.6 Proposed biosynthetic pathway of sphaerimicin piperidine ring formation based on bioinformatics and preliminary biochemical/functional protein characterization.

4.5 Supporting Information

SI Table 4.1 Primers used for cloning wildtype and mutant SphK expression, as well as SphN expression.

Primer Sequence (single-point mutation underlined) Note sphK_pET30A_FWD 5'-GGTATTGAGGGTCGCATGATGAACCACGTTCGTGAAGTCAG -3' Used for SphK sphK_pET30A_REV 5'-AGAGGAGAGTTAGAGCCCTTATTCTGCCGCCGCCACG-3' expression

sphK_H117A_FWD 5'-GGTCGCACCTGTTATTCTGTGGCTCGTGATGAAACGGCCCGCTAC-3' Used for SphK sphK_H117A_REV 5'-GTAGCGGGCCGTTTCATCACGAGCCACAGAATAACAGGTGCGACC-3' mutant expression Used for sphK_D119A_FWD 5'-CTGTTATTCTGTGCATCGTGCTGAAACGGCCCGCTACCACG-3' SphK mutant sphK_ D119A_REV 5'-CGTGGTAGCGGGCCGTTTCAGCACGATGCACAGAATAACAG-3' expression sphN_pET30A_FWD 5'-GGTATTGAGGGTCGCATGAACGCGGCGACCACC-3' Used for SphN sphN_pET30A_REV 5'-AGAGGAGAGTTAGAGCCGTCAACGGTCCTCACGGGTC-3' expression

SphK SphO SphN

75 kDa 75 kDa 75 kDa 50 kDa 50 kDa 50 kDa 37 kDa 37 kDa 37 kDa 25 kDa 25 kDa 25 kDa 15 kDa 15 kDa 15 kDa

SI Figure 4.1 SDS-PAGE gel of IMAC-purified SphK, SphO, and SphN, both His-tagged recombinant proteins heterologously expressed in E. coli BL21 cells.

SI Figure 4.2 Agilent 6120 Quadrupole and Agilent 6230 TOF MS spectra.

REFERENCES

1. McKenna, M., The last resort. Nature, 2013. 499(7459): p. 394. 2. Control, C.f.D. and Prevention, Antibiotic resistance threats in the United States, 2013. 2013: Centres for Disease Control and Prevention, US Department of Health and …. 3. Organization, W.H., Antimicrobial resistance: global report on surveillance. 2014: World Health Organization. 4. Maiden, M.C., Horizontal genetic exchange, evolution, and spread of antibiotic resistance in bacteria. Clinical Infectious Diseases, 1998. 27(Supplement_1): p. S12-S20. 5. Harvey, A.L., R. Edrada-Ebel, and R.J. Quinn, The re-emergence of natural products for drug discovery in the genomics era. Nature reviews. Drug discovery, 2015. 14(2): p. 111. 6. Stratton, C.F., D.J. Newman, and D.S. Tan, Cheminformatic comparison of approved drugs from natural product versus synthetic origins. Bioorganic & medicinal chemistry letters, 2015. 25(21): p. 4802-4807. 7. Doak, B.C., et al., Oral druggable space beyond the rule of 5: insights from drugs and clinical candidates. Chemistry & biology, 2014. 21(9): p. 1115-1142. 8. Walsh, C.T., A chemocentric view of the natural product inventory. Nature chemical biology, 2015. 11(9): p. 620-624. 9. Product Specialist, S.A., Antibiotics for Research Applications. 2006, Biofiles. 10. Walsh, C.T. and T.A. Wencewicz, Prospects for new antibiotics: a molecule- centered perspective. The Journal of antibiotics, 2014. 67(1): p. 7-22. 11. Gautam, A., R. Vyas, and R. Tewari, Peptidoglycan biosynthesis machinery: a rich source of drug targets. Critical reviews in biotechnology, 2011. 31(4): p. 295-336. 12. Vollmer, W. and U. Bertsche, Murein (peptidoglycan) structure, architecture and biosynthesis in Escherichia coli. Biochimica et Biophysica Acta (BBA)- Biomembranes, 2008. 1778(9): p. 1714-1734. 13. Bouhss, A., et al., The biosynthesis of peptidoglycan lipid-linked intermediates. FEMS microbiology reviews, 2007. 32(2): p. 208-233. 14. Niu, G. and H. Tan, Nucleoside antibiotics: biosynthesis, regulation, and biotechnology. Trends in microbiology, 2015. 23(2): p. 110-119. 15. Chi, X., et al., The muraminomicin biosynthetic gene cluster and enzymatic formation of the 2-deoxyaminoribosyl appendage. MedChemComm, 2013. 4(1): p. 239-243. 16. Chi, X., et al., Biosynthetic origin and mechanism of formation of the aminoribosyl moiety of peptidyl nucleoside antibiotics. Journal of the American Chemical Society, 2011. 133(36): p. 14452-14459. 17. Cai, W., et al., The Biosynthesis of Capuramycin-type Antibiotics IDENTIFICATION OF THE A-102395 BIOSYNTHETIC GENE CLUSTER, MECHANISM OF SELF-RESISTANCE, AND FORMATION OF URIDINE-5′- CARBOXAMIDE. Journal of Biological Chemistry, 2015. 290(22): p. 13710- 13724.

18. Yang, Z., et al., Characterization of LipL as a Non-heme, Fe (II)-dependent α- ketoglutarate: UMP Dioxygenase that Generates Uridine-5′-aldehyde during A- 90289 Biosynthesis. Journal of Biological Chemistry, 2011. 286(10): p. 7885- 7892. 19. Wu, X., et al., Deficiency of UDP‐GlcNAc: Dolichol Phosphate N‐ Acetylglucosamine‐1 Phosphate Transferase (DPAGT1) Causes a Novel Congenital Disorder of Glycosylation Type Ij. Human mutation, 2003. 22(2): p. 144-150. 20. Chung, B.C., et al., Structural insights into inhibition of lipid I production in bacterial cell wall synthesis. Nature, 2016. 533(7604): p. 557-560. 21. Chung, B.C., et al., Crystal structure of MraY, an essential membrane enzyme for bacterial cell wall synthesis. Science, 2013. 341(6149): p. 1012-1016. 22. Mashalidis, E.H., et al., Chemical logic of MraY inhibition by antibacterial nucleoside natural products. Nat Commun, 2019. 10(1): p. 2917. 23. Kaysser, L., et al., Analysis of the liposidomycin gene cluster leads to the identification of new caprazamycin derivatives. ChemBioChem, 2010. 11(2): p. 191-196. 24. Funabashi, M., et al., Structure‐Based Gene Targeting Discovery of Sphaerimicin, a Bacterial Translocase I Inhibitor. Angewandte Chemie International Edition, 2013. 52(44): p. 11607-11611. 25. Cheng, L., et al., Identification of the gene cluster involved in muraymycin biosynthesis from Streptomyces sp. NRRL 30471. Molecular BioSystems, 2011. 7(3): p. 920-927. 26. Cui, Z., et al., Antibacterial Muraymycins from Mutant Strains of Streptomyces sp. NRRL 30471. J Nat Prod, 2018. 81(4): p. 942-948. 27. Goswami, A., et al., Evidence that Oxidative Dephosphorylation by the Nonheme Fe(II), alpha-ketoglutarate:UMP Oxygenase occurs by Stereospecific Hydroxylation. FEBS Lett, 2017. 591(3): p. 468-478. 28. Yang, Z., et al., Fe(II)-dependent, Uridine-5'-monophosphate alpha-ketoglutarate Dioxygenases in the Synthesis of 5'-modified Nucleosides. Methods Enzymol, 2012. 516: p. 153-68. 29. Barnard-Britson, S., et al., Amalgamation of nucleosides and amino acids in antibiotic biosynthesis: discovery of an L-threonine:uridine-5'-aldehyde transaldolase. J Am Chem Soc, 2012. 134(45): p. 18514-7. 30. Chi, X., et al., Biosynthetic origin and mechanism of formation of the aminoribosyl moiety of peptidyl nucleoside antibiotics. J Am Chem Soc, 2011. 133(36): p. 14452-9. 31. Cui, Z., et al., Enzymatic Synthesis of the Ribosylated Glycyl-Uridine Disaccharide Core of Peptidyl Nucleoside Antibiotics. Journal of Organic Chemistry, 2018. 32. Cui, Z., et al., Self-Resistance during Muraymycin Biosynthesis: a Complementary Nucleotidyltransferase and Phosphotransferase with Identical Modification Sites and Distinct Temporal Order. Antimicrob Agents Chemother, 2018. 62(7). 33. Shiraishi, T., et al., Biosynthesis of the antituberculous agent caprazamycin: Identification of caprazol-3ʺ-phosphate, an unprecedented caprazamycin-related

metabolite. The Journal of general and applied microbiology, 2016. 62(3): p. 164- 166. 34. Percudani, R. and A. Peracchi, A genomic overview of pyridoxal-phosphate- dependent enzymes. EMBO reports, 2003. 4(9): p. 850-854. 35. Mueser, T.C., et al., Chapter Fifteen - Pyridoxal 5′-phosphate dependent reactions: Analyzing the mechanism of aspartate aminotransferase, in Methods in Enzymology, P.C.E. Moody, Editor. 2020, Academic Press. p. 333-359. 36. Hoegl, A., et al., Mining the cellular inventory of pyridoxal phosphate-dependent enzymes with functionalized cofactor mimics. Nature Chemistry, 2018. 10(12): p. 1234-1245. 37. Huang, Y., et al., Pyridoxal-5'-phosphate as an oxygenase cofactor: Discovery of a carboxamide-forming, alpha-amino acid monooxygenase-decarboxylase. Proc Natl Acad Sci U S A, 2018. 115(5): p. 974-979. 38. Ngo, H.P., et al., PLP undergoes conformational changes during the course of an enzymatic reaction. Acta Crystallogr D Biol Crystallogr, 2014. 70(Pt 2): p. 596- 606. 39. di Salvo, M.L., R. Contestabile, and M.K. Safo, Vitamin B6 salvage enzymes: Mechanism, structure and regulation. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics, 2011. 1814(11): p. 1597-1608. 40. Shaw, J.P., G.A. Petsko, and D. Ringe, Determination of the Structure of Alanine Racemase from Bacillus stearothermophilus at 1.9-Å Resolution. Biochemistry, 1997. 36(6): p. 1329-1342. 41. Sun, S. and M.D. Toney, Evidence for a Two-Base Mechanism Involving Tyrosine-265 from Arginine-219 Mutants of Alanine Racemase. Biochemistry, 1999. 38(13): p. 4058-4065. 42. Christen, P. and P.K. Mehta, From cofactor to enzymes. The molecular evolution of pyridoxal-5′-phosphate-dependent enzymes. The Chemical Record, 2001. 1(6): p. 436-447. 43. Eliot, A.C. and J.F. Kirsch, Pyridoxal phosphate enzymes: mechanistic, structural, and evolutionary considerations. Annu Rev Biochem, 2004. 73: p. 383-415. 44. Griswold, W.R. and M.D. Toney, Role of the pyridine nitrogen in pyridoxal 5'- phosphate catalysis: activity of three classes of PLP enzymes reconstituted with deazapyridoxal 5'-phosphate. J Am Chem Soc, 2011. 133(37): p. 14823-30. 45. Lin, Y.-L., et al., Molecular dynamics simulations of the intramolecular proton transfer and carbanion stabilization in the pyridoxal 5′-phosphate dependent enzymes l-dopa decarboxylase and alanine racemase. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics, 2011. 1814(11): p. 1438-1446. 46. Silverman, R.B., Design and Mechanism of GABA Aminotransferase Inactivators. Treatments for Epilepsies and Addictions. Chem Rev, 2018. 118(7): p. 4037- 4070. 47. Matoba, Y., et al., Catalytic specificity of the Lactobacillus plantarum cystathionine γ-lyase presumed by the crystallographic analysis. Scientific Reports, 2020. 10(1): p. 14886. 48. Du, Y.L. and K.S. Ryan, Pyridoxal phosphate-dependent reactions in the biosynthesis of natural products. Nat Prod Rep, 2019. 36(3): p. 430-457.

49. Griswold, W.R., et al., Ground-State Electronic Destabilization via Hyperconjugation in Aspartate Aminotransferase. Journal of the American Chemical Society, 2012. 134(20): p. 8436-8438. 50. Cambron, G., et al., Fluorescence of the Schiff bases of pyridoxal and pyridoxal 5'-phosphate withL-isoleucine in aqueous solutions. J Fluoresc, 1996. 6(1): p. 1-6. 51. Pan, P., et al., Shift in pH-rate profile and enhanced discrimination between dicarboxylic and aromatic substrates in mitochondrial aspartate aminotransferase Y70H. Biochemistry, 1994. 33(10): p. 2757-60. 52. Takeshima, D., et al., A single amino acid substitution converts a histidine decarboxylase to an imidazole acetaldehyde synthase. Arch Biochem Biophys, 2020. 693: p. 108551. 53. Alexander, F.W., et al., Evolutionary relationships among pyridoxal‐5′‐ phosphate‐dependent enzymes: Regio‐specific α, β and γ families. European journal of biochemistry, 1994. 219(3): p. 953-960. 54. Kobylarz, M.J., et al., Deciphering the Substrate Specificity of SbnA, the Enzyme Catalyzing the First Step in Staphyloferrin B Biosynthesis. Biochemistry, 2016. 55(6): p. 927-39. 55. Ramalingam, K., et al., Stereochemical course of the reaction catalyzed by the pyridoxal phosphate-dependent enzyme 1-aminocyclopropane-1-carboxylate synthase. Proc Natl Acad Sci U S A, 1985. 82(23): p. 7820-4. 56. Foo, T.C., A.C. Terentis, and K.V. Venkatachalam, A continuous spectrophotometric assay and nonlinear kinetic analysis of methionine gamma- lyase catalysis. Anal Biochem, 2016. 507: p. 21-6. 57. Watanabe, A., et al., Tyrosine 265 of Alanine Racemase Serves as a Base Abstracting α-Hydrogen from L-Alanine: The Counterpart Residue to Lysine 39 Specific to D-Alanine. The Journal of Biochemistry, 1999. 126(4): p. 781-786. 58. Watanabe, A., et al., Role of Tyrosine 265 of Alanine Racemase from Bacillus stearothermophilus. The Journal of Biochemistry, 1999. 125(6): p. 987-990. 59. Watanabe, A., et al., Role of lysine 39 of alanine racemase from Bacillus stearothermophilus that binds pyridoxal 5'-phosphate. Chemical rescue studies of Lys39 --> Ala mutant. J Biol Chem, 1999. 274(7): p. 4189-94. 60. Toney, M.D., Controlling reaction specificity in pyridoxal phosphate enzymes. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics, 2011. 1814(11): p. 1407-1418. 61. Chan-Huot, M., et al., NMR studies of protonation and hydrogen bond states of internal aldimines of pyridoxal 5'-phosphate acid-base in alanine racemase, aspartate aminotransferase, and poly-L-lysine. J Am Chem Soc, 2013. 135(48): p. 18160-75. 62. Limbach, H.-H., et al., Critical hydrogen bonds and protonation states of pyridoxal 5′-phosphate revealed by NMR. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics, 2011. 1814(11): p. 1426-1437. 63. Spies, M.A. and M.D. Toney, Intrinsic Primary and Secondary Hydrogen Kinetic Isotope Effects for Alanine Racemase from Global Analysis of Progress Curves. Journal of the American Chemical Society, 2007. 129(35): p. 10678-10685.

64. Kurokawa, Y., et al., Transamination as a Side-Reaction Catalyzed by Alanine Racemase of Bacillus stearothermophilus1. The Journal of Biochemistry, 1998. 124(6): p. 1163-1169. 65. Spies, M.A. and M.D. Toney, Multiple Hydrogen Kinetic Isotope Effects for Enzymes Catalyzing Exchange with Solvent: Application to Alanine Racemase. Biochemistry, 2003. 42(17): p. 5099-5107. 66. Kipp, D.R., R.G. Silva, and V.L. Schramm, Mass-Dependent Bond Vibrational Dynamics Influence Catalysis by HIV-1 Protease. Journal of the American Chemical Society, 2011. 133(48): p. 19358-19361. 67. Silva, R.G., A.S. Murkin, and V.L. Schramm, Femtosecond dynamics coupled to chemical barrier crossing in a Born-Oppenheimer enzyme. Proceedings of the National Academy of Sciences, 2011. 108(46): p. 18661. 68. Harijan, R.K., et al., Inverse enzyme isotope effects in human purine nucleoside phosphorylase with heavy asparagine labels. Proc Natl Acad Sci U S A, 2018. 115(27): p. E6209-E6216. 69. Schramm, V.L. and S.D. Schwartz, Promoting Vibrations and the Function of Enzymes. Emerging Theoretical and Experimental Convergence. Biochemistry, 2018. 57(24): p. 3299-3308. 70. Toney, M.D. and J.F. Kirsch, Broønsted analysis of aspartate aminotransferase via exogenous catalysis of reactions of an inactive mutant. Protein Science, 1992. 1(1): p. 107-119. 71. Wright, S.K., M.A. Rishavy, and W.W. Cleland, 2H, 13C, and 15N Kinetic Isotope Effects on the Reaction of the Ammonia-Rescued K258A Mutant of Aspartate Aminotransferase. Biochemistry, 2003. 42(27): p. 8369-8376. 72. Toney, M.D., J.N. Castro, and T.A. Addington, Heavy-enzyme kinetic isotope effects on proton transfer in alanine racemase. J Am Chem Soc, 2013. 135(7): p. 2509-11. 73. Maresi, E., et al., Functional Characterization and Structure-Guided Mutational Analysis of the Transsulfuration Enzyme Cystathionine gamma-Lyase from Toxoplasma gondii. Int J Mol Sci, 2018. 19(7). 74. Aitken, S.M., P.H. Lodha, and D.J.K. Morneau, The enzymes of the transsulfuration pathways: Active-site characterizations. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics, 2011. 1814(11): p. 1511-1517. 75. Conter, C., et al., Cystathionine β-synthase is involved in cysteine biosynthesis and H2S generation in Toxoplasma gondii. Scientific Reports, 2020. 10(1): p. 14657. 76. Guggenheim, S. and M. Flavin, Cystathionine gamma-synthase from Salmonella. Spectral changes in the presence of substrates. J Biol Chem, 1971. 246(11): p. 3562-8. 77. Clausen, T., et al., Crystal structure of Escherichia coli cystathionine gamma- synthase at 1.5 A resolution. Embo j, 1998. 17(23): p. 6827-38. 78. Johnston, M., et al., Mechanistic studies with vinylglycine and beta- haloaminobutyrates as substrates for cystathionine gamma-synthetase from Salmonella typhimurium. Biochemistry, 1979. 18(9): p. 1729-38.

79. Chang, M.N. and C.T. Walsh, Stereochemical analysis of. gamma.-replacement and. gamma.-elimination processes catalyzed by a pyridoxal phosphate dependent enzyme. Journal of the American Chemical Society, 1981. 103(16): p. 4921-4927. 80. Spiering, M.J., et al., Gene clusters for insecticidal loline alkaloids in the grass- endophytic fungus Neotyphodium uncinatum. Genetics, 2005. 169(3): p. 1403-14. 81. Brotzel, F. and H. Mayr, Nucleophilicities of amino acids and peptides. Org Biomol Chem, 2007. 5(23): p. 3814-20. 82. Cui, Z., et al., Pyridoxal-5'-phosphate-dependent alkyl transfer in nucleoside antibiotic biosynthesis. Nat Chem Biol, 2020. 83. Brzovic, P., et al., Reaction mechanism of Escherichia coli cystathionine gamma- synthase: direct evidence for a pyridoxamine derivative of vinylglyoxylate as a key intermediate in pyridoxal phosphate dependent gamma-elimination and gamma-replacement reactions. Biochemistry, 1990. 29(2): p. 442-51. 84. Faulkner, J.R., et al., On the sequence of bond formation in loline alkaloid biosynthesis. Chembiochem, 2006. 7(7): p. 1078-88. 85. Chen, M., C.T. Liu, and Y. Tang, Discovery and Biocatalytic Application of a PLP-Dependent Amino Acid gamma-Substitution Enzyme That Catalyzes C-C Bond Formation. J Am Chem Soc, 2020. 142(23): p. 10506-10515. 86. Hartmann, M., et al., Biochemical Principles and Functional Aspects of Pipecolic Acid Biosynthesis in Plant Immunity. Plant Physiol, 2017. 174(1): p. 124-153. 87. Hai, Y., et al., Biosynthesis of Mycotoxin Fusaric Acid and Application of a PLP- Dependent Enzyme for Chemoenzymatic Synthesis of Substituted l-Pipecolic Acids. Journal of the American Chemical Society, 2020. 88. Hedges, J.B. and K.S. Ryan, Biosynthetic Pathways to Nonproteinogenic α-Amino Acids. Chemical Reviews, 2020. 120(6): p. 3161-3209. 89. Wei, Y., et al., Mechanism of Vibrio cholerae autoinducer-1 biosynthesis. ACS Chem Biol, 2011. 6(4): p. 356-65. 90. Van Lanen, S.G., Biosynthesis: SAM cycles up for colibactin. Nat Chem Biol, 2017. 13(10): p. 1059-1061. 91. Zha, L., et al., Colibactin assembly line enzymes use S-adenosylmethionine to build a cyclopropane ring. Nature Chemical Biology, 2017. 13(10): p. 1063-1065. 92. Xu, Z., et al., Discovery and Characterization of 1-Aminocyclopropane-1- carboxylic Acid Synthase of Bacterial Origin. J Am Chem Soc, 2018. 140(49): p. 16957-16961. 93. Koch, K.A., et al., The human cDNA for a homologue of the plant enzyme 1- aminocyclopropane-1-carboxylate synthase encodes a protein lacking that activity. Gene, 2001. 272(1-2): p. 75-84. 94. Marchand, J.A., et al., Discovery of a pathway for terminal-alkyne amino acid biosynthesis. Nature, 2019. 567(7748): p. 420-424. 95. Kezuka, Y., et al., Structural insights into the catalytic mechanism of cysteine (hydroxyl) lyase from the hydrogen sulfide-producing oral pathogen, Fusobacterium nucleatum. Biochemical Journal, 2018. 475(4): p. 733-748. 96. Mothersole, R.G., et al., S224 Presents a Catalytic Trade-off in PLP-Dependent l- Lanthionine Synthase from Fusobacterium nucleatum. Biochemistry, 2020. 59(44): p. 4250-4261.

97. Mothersole, R.G. and K.R. Wolthers, Structural and Kinetic Insight into the Biosynthesis of H2S and l-Lanthionine from l-Cysteine by a Pyridoxal l- Phosphate-Dependent Enzyme from Fusobacterium nucleatum. Biochemistry, 2019. 58(34): p. 3592-3603. 98. Singh, S., et al., Relative contributions of cystathionine beta-synthase and gamma-cystathionase to H2S biogenesis via alternative trans-sulfuration reactions. J Biol Chem, 2009. 284(33): p. 22457-66. 99. Thomas, M.G., Y.A. Chan, and S.G. Ozanick, Deciphering Tuberactinomycin Biosynthesis: Isolation, Sequencing, and Annotation of the Viomycin Biosynthetic Gene Cluster. Antimicrobial Agents and Chemotherapy, 2003. 47(9): p. 2823. 100. Kobylarz, Marek J., et al., Synthesis of L-2,3-Diaminopropionic Acid, a Siderophore and Antibiotic Precursor. Chemistry & Biology, 2014. 21(3): p. 379- 388. 101. Ye, Y., et al., Unveiling the Biosynthetic Pathway of the Ribosomally Synthesized and Post-translationally Modified Peptide Ustiloxin B in Filamentous Fungi. Angew Chem Int Ed Engl, 2016. 55(28): p. 8072-5. 102. Faleev, N.G., et al., Methionine gamma-lyase: mechanistic deductions from the kinetic pH-effects. The role of the ionic state of a substrate in the enzymatic activity. Biochim Biophys Acta, 2009. 1794(10): p. 1414-20. 103. Allegrini, A., et al., Characterization of C-S lyase from Lactobacillus delbrueckii subsp. bulgaricus ATCC BAA-365 and its potential role in food flavour applications. J Biochem, 2017. 161(4): p. 349-360. 104. Sato, D., et al., Structural and mechanistic insights into homocysteine degradation by a mutant of methionine γ-lyase based on substrate-assisted catalysis. Protein Science, 2017. 26(6): p. 1224-1230. 105. Ushimaru, R. and H.W. Liu, Biosynthetic Origin of the Atypical Stereochemistry in the Thioheptose Core of Albomycin Nucleoside Antibiotics. J Am Chem Soc, 2019. 141(6): p. 2211-2214. 106. Han, L., et al., Streptomyces wadayamensis MppP is a PLP-Dependent Oxidase, Not an Oxygenase. Biochemistry, 2018. 57(23): p. 3252-3264. 107. Han, L., et al., Streptomyces wadayamensis MppP Is a Pyridoxal 5'-Phosphate- Dependent L-Arginine alpha-Deaminase, gamma-Hydroxylase in the Enduracididine Biosynthetic Pathway. Biochemistry, 2015. 54(47): p. 7029-40. 108. Du, Y.L., et al., A pyridoxal phosphate-dependent enzyme that oxidizes an unactivated carbon-carbon bond. Nat Chem Biol, 2016. 12(3): p. 194-9. 109. Hedges, J.B., et al., Snapshots of the Catalytic Cycle of an O2, Pyridoxal Phosphate-Dependent Hydroxylase. ACS Chem Biol, 2018. 13(4): p. 965-974. 110. Dai, G.Z., et al., Pyridoxal-5'-phosphate-dependent bifunctional enzyme catalyzed biosynthesis of indolizidine alkaloids in fungi. Proc Natl Acad Sci U S A, 2019. 111. Hoffarth, E.R., K.W. Rothchild, and K.S. Ryan, Emergence of oxygen- and pyridoxal phosphate-dependent reactions. The FEBS Journal, 2020. 287(7): p. 1403-1428. 112. Isupov, M.N., et al., Thermostable Branched-Chain Amino Acid Transaminases From the Archaea Geoglobus acetivorans and Archaeoglobus fulgidus: Biochemical and Structural Characterization. Frontiers in Bioengineering and Biotechnology, 2019. 7(7).

113. Di Salvo, M.L., et al., Editorial: PLP-Dependent Enzymes: Extraordinary Versatile Catalysts and Ideal Biotechnological Tools for the Production of Unnatural Amino Acids and Related Compounds. Frontiers in Bioengineering and Biotechnology, 2020. 8(52). 114. Richts, B., J. Rosenberg, and F.M. Commichau, A Survey of Pyridoxal 5′- Phosphate-Dependent Proteins in the Gram-Positive Model Bacterium Bacillus subtilis. Frontiers in Molecular Biosciences, 2019. 6(32). 115. Tanaka, H., et al., Simultaneous measurement of D-serine dehydratase and d- amino acid oxidase activities by the detection of 2-oxo-acid formation with reverse-phase high-performance liquid chromatography. Anal Biochem, 2007. 362(1): p. 83-8. 116. Kelley, L.A., et al., The Phyre2 web portal for protein modeling, prediction and analysis. Nat Protoc, 2015. 10(6): p. 845-58. 117. Koketsu, K., et al., Refined regio-and stereoselective hydroxylation of L-pipecolic acid by protein engineering of L-proline cis-4-hydroxylase based on the X-ray crystal structure. ACS synthetic biology, 2014. 4(4): p. 383-392. 118. Clifton, I.J., et al., Structure of proline 3‐hydroxylase. The FEBS Journal, 2001. 268(24): p. 6625-6636.

VITA

Education

Ph.D. candidate, Department of Pharmaceutical Sciences (GPA 3.8) University of Kentucky; Lexington, KY 40536 Expected Graduation: May 2021 Advisor: Dr. Steven G. Van Lanen

B.S. in Chemistry (Biochemistry), Class of 2014 Eastern Kentucky University; Richmond, KY Magna cum Laude, Honors Scholar

Publications

Overbay, J.; Cui, Z.; […]; Van Lanen, S. G. “Catalytic Mechanism of Pyridoxal-5'- Phosphate-Dependent Gamma-Replacement Alkyl Transfer during Nucleoside Antibiotic Biosynthesis.” Manuscript in Preparation.

Overbay, J.; Van Lanen, S. G. “The Versatility of Enzyme-Directed Reactivity in Pyridoxal-5'-Phosphate-Dependent Biosynthesis.” J Ind Microbiol Biotechnol. Manuscript in Preparation. Cui, Z.; Overbay, J.; Wang, X.; Liu, X.; Zhang, Y.; Bhardwaj, M.; Lemke, A.; Wiegmann, D.; Niro, G.; Thorson, J. S.; Ducho, C.; Van Lanen, S. G. “Pyridoxal-5'- Phosphate-Dependent Alkyl Transfer in Nucleoside Antibiotic Biosynthesis.” Nat Chem Biol 2020. Cui, Z.; Liu, X; Overbay, J; Cai, W; Wang, X; Lemke, A; Wiegmann, D; Niro, G; Thorson, JS; Ducho, C; Van Lanen, S. G. “Enzymatic Synthesis of the Ribosylated Glycyl-Uridine Disaccharide Core of Peptidyl Nucleoside Antibiotics." J Org Chem. 2018. McErlean, M.; Overbay, J.; Van Lanen, S. G. "Refining and Expanding Nonribosomal Peptide Synthetase Function and Mechanism." J Ind Microbiol Biotechnol. 2019. Sellers, Z. P.; Williams, R.; Overbay, J. W.; Cho, J.; Henderson, M.; Reed, T. T. “Current Therapeutic Modalities, Enzyme Kinetics, and Redox Proteomics in Traumatic Brain Injury.” Traumatic Brain Injury, InTech, Croatia 2014. Employment Research assistant for University of Kentucky: May 2017 – Present Teaching assistant for University of Kentucky: July 2015 – May 2017 Assistant scientist for LGC Science, Inc: May 2014 - June 2015 Research assistant for Eastern Kentucky University: January 2013 – May 2014 Pharmacy Technician for CVS/pharmacy: May 2011 – May 2013