INFORMATION TO USERS

The most advanced technology has been used to photograph and reproduce this manuscript from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

University Microfilms International A Bell & Howell Information Company 300 North Zeeb Road. Ann Arbor. Ml 48106-1346 USA 313/761-4700 800 521-0600

Order Number 9111825

Studies on the biosynthesis of the modified-peptide antibiotic, thiostrepton

Zeng, Zhaopie, Ph.D.

The Ohio State University, 1990

UMI 300 N. Zeeb Rd. Ann Arbor, MI 48106

NOTE TO USERS

THE ORIGINAL DOCUMENT RECEIVED BY U.M.I. CONTAINED PAGES WITH SLANTED AND POOR PRINT. PAGES WERE FILMED AS RECEIVED.

THIS REPRODUCTION IS THE BEST AVAILABLE COPY.

STUDIES ON THE BIOSYNTHESIS OF THE MODIFIED-PEPTIDE ANTIBIOTIC, THIOSTREPTON

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Zhaopie Zeng, B.S.

**************

The Ohio State University

1990

Dissertation Committee: Approved by Heinz G. Floss Leo A. Paquette Viresh Rawal ^Adviser Department of Chemistry To my parents ACKNOWLEDGEMENTS

I would like to express my most sincere gratitude to my research advisor, Professor Heinz G. Floss, for his guidence, encouragement, and patience throughout my graduate study.

I want to give special thank you to Professor John M. Beale and Dr.

Ulla Mocek for their help with acquisition of NMR data, valuable suggestions and friendship. I also would like to express my appreciation to Kay Kampsen for her help during my years in

Professor Floss' group.

Finally, I would like to express my gratitude to my friends and family, expeciallly my parents and my brothers for their supports and encouragements. VITA

March 28,1962 ...... Born - Shanghai, China. July 1985...... B.S., Department of Chemistry, East China Normal University, Shanghai, China. Sept. 1985-June 1987 ...... Teaching Assistant, Department of Chemistry, the Ohio State University, Columbus, Ohio. June 1987-present ...... Graduate Research Associate, Department of Chemistry, the Ohio State University, Ohio.

PUBLICATIONS

P. Zhou, D. O'Hagan, U. Mocek, Z. Zeng, L.-D. Yuen, T. Frenzel, C.J. Unkefer, J.M. Beale and H.G. Floss, "Biosynthesis of the Antibiotic Thiostrepton. Methylation of Tryptophan in the Formation of the Quinaldic Acid Moiety by Transfer of the Methionine Methyl Group with Net Retention of Configuration", J. Am. Chem. Soc., 111, 7274 (1989)

FIELDS OF STUDY

Major Field: Chemistry

Studies in Organic Chemistry TABLE OF CONTENTS

PAGE ACKNOWLEDGEMENTS...... in VITA...... iv LIST OF TABLES ...... vii LIST OF FIGURES...... viii LIST OF ABBREVIATIONS...... xii INTRODUCTION...... 1 RESULTS...... 16 Fermentation ...... 16 Isolation of Thiostrepton ...... 17 Thiostrepton Assay by HPLC and Production Curve 18 NMR Spectroscopy...... 20 D,L-[1-13C]lsoleucine Synthesis and Feeding 33 D,L-[1-13C,3-2H ]Threonine Synthesis and Feeding ...... 37 Formation of Piperideine Ring ...... 44 Mode of Incorporation of into Other Structural Elements ...... 51 Formation of Quinaldic Acid Moiety ...... 62 Stereochemistry of Hydrogen Loss from the p-Carbon of Tryptophan ...... 64 Role of 4-(1-Hydroxyethyl)quinoline-2-carboxylic Acid ...... 65 Attempted 180 2 Feeding Experiment ...... 67 15N Precursor Feedings ...... 70 DISCUSSION...... 87 EXPERIMENTAL SECTION...... 106 Materials and General Methods ...... 106 1. Materials...... 106 2. General Methods ...... 107 Fermentation...... 108 Isolation of Thiostrepton ...... 110

v Thiostrepton Assay by HPLC ...... 110 Feeding Experiments with Labeled Precursors 111 Synthesis of D,L-[1 -13C]lsoleucine ...... 113 Derivatization of the Mixture of Isoleucine and Alloisoleucine ...... 114 Synthesis of Threonine ...... 115 Synthesis of (ZS.S’RJ-P'^HjTryptophan ...... 117 Synthesis of (ZS^'SHS'^Hp-ryptophan ...... 118 Feeding Experiment with Mixture of (2'S,3'S)- [3'-3H]-Tryptophan and D,L-[3'-14C]Tryptophan 119 Feeding Experiment with Mixture of (2'S,3'R)- [3'-3H]-Tryptophan and D,L-[3'-14C]Tryptophan ...... 119 Feeding Experiment with Mixture of (2'S,3'S)- [3'-3H]-Tryptophan and L-[3'-14C]Tryptophan ...... 119 Feeding Experiment with Mixture of (2'S,3'R)- [3'-3H]Tryptophan and L-[3'-^4C]Tryptophan ...... 120 Co-crystallization of a Mixture of Tryptophan ...... 120 4-(1 -Hydroxyethyl)quinaldic Acid Synthesis ...... 121 4-(1-Hydroxy-[1 -3H]ethyl)quinaldie Acid Synthesis...... 123 Feeding Experiment with the 4-(1-Hydroxy- [1 -3H]ethyl)quinaldic Acid ...... 124 180 2 Experiment...... 125 Fermentation with Different Amounts of (NH4)2S 0 4 in Production Medium ...... 125 15NH4CI Feeding ...... 126 15N-Glycine Feeding ...... 126 Fermentation with Different Amounts of Serine 127 Feeding Experiment with 15NH4CI and a 3-Fold Excess of Unlabeled Serine ...... 127 REFERENCES...... 132

vi LIST OF TABLES

TABLE PAGE

1. 13C and 1H NMR spectral data of thiostrepton ...... 31 2. 13C-13C-Coupling patterns in thiostrepton derived from L-[1,2-13C2]- and L-[2,3-13C2]serine ...... 32 3 3 H/14c Ratios in feeding experiments with radiolabeled tryptophan ...... 66 4. 2D-NMR COSY shift correlations of thiostrepton...... 82/83 5. 1H-NMR assignm ents of thiostrepton ...... 84/85 6. 2D-NMR 1H/15N shift correlations of thiostrepton 86 7. Required number of biosynthetic genes ...... 105 8. Feeding experiments with labeled precursor ...... 112 9. HPLC system for tracing tryptophan synthesis ...... 129 10. HPLC system for purification of tryptophan ...... 130 11. HPLC system for purification of thiostrepton ...... 131 LIST O F FIGURES

RGURE PAGE

1. The structure of thiostrepton and a shorthand designation for its structural components ...... 2 2. The structure of nosiheptide and a shorthand designation for its structural components ...... 4 3. Berninamycin ...... 8 4. Labeling pattern of nosiheptide from various 13C-labeled amino acids ...... 9 5. Labeling pattern of thiostrepton from various 13C-labeled amino acids ...... 11 6. Incorporation of chiral methionine into quinaldic acid moiety of thiostrepton ...... 14 7. Thiostrepton production curve in S.laurentii fermentation ...... 19 8. 1H-NMR spectrum of thiostrepton in CDCI3:CD3OD 4:1... 21 9. 1H-1 H COSY spectrum of thiostrepton in CDCI3:CD3OD 4:1...... 22 10. 1H-13C COLOC spectrum of thiostrepton in CDCI3:CD3OD 4:1 ...... 23 11. 13C-NMR spectrum of thiostrepton in CDCI3:CD3OD 4:1. 24 12 1H-13C correlation spectrum of thiostrepton in CDCI3:CD3OD 4:1 ...... 26 13. 2D-INADEQUATE spectrum of thiostrepton from [1,2-13C2]serine feeding ...... 27/28 14. 2D-INADEQUATE spectrum of thiostrepton from [2,3-13C2]serine feeding ...... 29/30 15. Synthesis of D,L-[1 -13C]isoleucine ...... 34 16. The 13C-NMR spectrum of thiostrepton biosynthesized from D,L-[1 -13C]isoleucine ...... 35 17. The incorporation of D,L-[1-13C]isoleucine into thiostrepton ...... 36

viii 18. Synthesis of D,L-[1“13C,3-2H]threonine ...... 38 19. The 1H-NMR spectrum of standard L-threonine and synthesized D.L-threonine ...... 39 20. The 1H-NMR spectrum of D,L-[1-13C,3-2H]threonine.... 40 21. The 13C-NMR spectrum of thiostrepton biosynthesized from D,L-[1-13C,3-2H]threonine ...... 41 22. The incorporation of D,L-[1 -13C,3-2H]threonine into thiostrepton...... 42 23. The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 2S,3S-[3-13C,2H1]serine (left) and 2S-[3-13C 2H2]serine (right) at PipC3 and PipC4 region...... 47 24. The 2D Hetcor NMR of thiostrepton biosynthesized from 2S-[3-13C,2H2]serine ...... 48 25. The 2D Hetcor NMR of thiostrepton biosynthesized from,2S,3S-[3-13C ^H^serine ...... 49/50 26. The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 25.35-[3-13C 2H.,]serine (left) and 2S-[3-13C 2H2]serine (right) at the DealaCS region ...... 52 27. The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 25.35-[3-13C,2H1]serine (left) and 2S-[3-13C,2H2]serine (right) at the AlaMe region...... 54

ix 28. The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 2S-[3-13C,2H2]serine at the ThzCS and QC3 region...... 56 29. The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 25.35-[3-13C^H^serine at the ThzC5 and QC3 region...... 57 30. The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 25.35-[3-13C,2H1]serine (left) and 2S-[3-13C 2H2]sGrine (right) at the CysC5 region ...... 58 31. The intact incorporation of L-[1,2-13C2]serine and L[2,3-13C2]serine into thiostrepton. Heavy lines denote coupling between carbon atoms ...... 59 32. The incorporation of 2S- [3-13C,2H2]serine into thiostrepton...... 60 33. The incorporation of 2S.3S- [3-13C^H ^serine into thiostrepton...... 61 34. The 13C-NMR spectrum of thiostrepton biosynthesized from L-[indole-15N, 1',2'-13C2]- tryptophan ...... 63 35. The synthesis of tritiated 4-(1-hydroxyethyl)- quinoline-2-carboxylic acid ...... 68 36. The 3H-NMR of 4-(1 -hydroxy-[1 -3H]ethyl)- quinaldic acid ...... 69

x 37. The 1H-NMR spectrum of thiostrepton in DMSO-d6 76 38. The 1H-NMR spectrum of thiostrepton in DMSO-d6 with D20 exchange...... 77 39. The 1H-1H COSY spectrum of thiostrepton in DMSO-d6... 78 40. The 15N DEPT spectrum of thiostrepton ...... 79 41. The 2D-inverse heteronuclear multiple quantum correlation with BIRD pulse spectrum of thiostrepton ...... 80 42. The 15N-NMR spectra of thiostrepton biosynthesized from 15N-labeled precursors ...... 81 43. Formation of quinaldic acid moiety ...... 90 44. Incorporation of tritiated 4-(1 -hydroxyethyl)- quinolinic acid into thiostrepton ...... 91 45. Biosynthesis of gliotoxin ...... 92 46. Formation of piperideine ring ...... 98 47. Primary peptide precursors of nosiheptide and thiostrepton ...... 101 48. Diagram of the system for fermentation in 180-containing atmosphere ...... 126

xi LIST OF ABBREVIATIONS

COLOC correlation spectroscopy via long couplings COSY correlation spectroscopy DMF N,N-dimethylformamide DMSO dimethyl sulfoxide 2D two-dimensional GC-MS gas chromatography-mass spectrometry H-C HETCOR proton, carbon heteronuclear correlation spectroscopy HPLC high performance liquid chromatography INADEQUATE incredible natural abundance double quantum transfer experiment FVM frozen-vegetative mycelia J coupling constant NMR nuclear magnetic resonance THF tetrahydrofuran TLC thin layer chromatography UV ultraviolet uCi micro-Curie INTRODUCTION

Thiostrepton (Figure 1) was first isolated from fermentation broths inoculated with Streptomyces azureus.1-2-3 The antibiotic bryamycin which was isolated independently from S. hawaiiensis4 appears to be identical to thiostrepton. S. laurentii 5 was also found to produce the same antibiotic. The compound inhibits protein synthesis in gram-positive bacteria. Most gram-negative bacteria are impermeable to thiostrepton and are therefore resistant to its action.

Thiostrepton is insoluble in water, low-boiling alcohols and nonpolar organic solvents such as hexane or benzene. Because of the low water solubility, at least 95% of the antibiotic remains with the mycelium when the latter is separated from the fermentation broth. The antibiotic can be extracted from the moist filter cake by extraction into chloroform and further purified by repeated crystallization and column chromatography over alumina. Thiostrepton melts with decomposition over the range of 246 to 256°C. It does not dissolve in dilute aqueous acid or base. The ultraviolet spectrum shows no maxima, but displays characteristic shoulders at 225, 250 and 280 nm with E1cm equal to 520, 380 and

1 2

I Deolo (2 ) H Deolo (3)j

iT ht (4)1

Alo (2) H Deolo (I)] fDeolo (2) j 1 Deolo(3) I G p D [ f e ® ] p iTMOl . Srhr(^)t-m-fTir|

Hfl i / * , | | I H | l P .e-Elol A Il

s‘ " 4 C J* ^ 2 5^S z NH H 4 5 \ ■ F 6 vCH,CH3 « « " ' 1 Alo (2) H ^ M J CH?< I jj 5 «CH, 0 HO H0 8X \\ ,_____ ,\ // | xvH „ 0 fThflPl S - ^ JC

Thr (2) j

H H

CH. I '"'H

Thsln = Thiostreptine residue Q » Quinaldic acid precursor Cys s Thiazoline Thz * Thlazole ring Pip * Piperidine ring

Figure 1 The structure of thiostrepton and a shorthand designation for its structural components. 3 225, respectively.

The complete structure of thiostrepton has been determined by degradation studies71’74 and X-ray crystallographic analysis.6 The molecule is composed of two loops of highly modified amino acid residues including three thiazole rings, one thiazoline ring in the larger loop, one quinolinic acid moiety in the shorter loop, and several oxidized or dehydrogenated amino acids. The two peptide loops connect at the piperideine ring of thiostrepton. The side chain contains another thiazole ring and two dehydroalanine residues.

A number of antibiotics, the siomycins,7 thiopeptins,8 micrococcins,9-10 sporangiomycin,11 Sch18640,12 the thiocillins13 and nosiheptide14 (Figure 2) are all grouped in a thiostrepton antibiotic family since they share a common architecture characterized by a macrocycle incorporating four thiazole rings and one pyridine or reduced pyridine ring, a side chain consisting of another thiazole and one or two dehydroalanine moieties, and an extra loop of more variable structure.

Consistent with this common architecture, these antibiotics share a common mode of action. These highly modified, sulfur-rich polypeptide antibiotics inhibit protein synthesis in gram-positive bacteria. It was first observed that 35S-thiostrepton binds to the

SOS but not the 30S ribosomal subunits of E. coli ribosomes. Later Iml ° H H i

Thz = Thiazole ring Thr = Threonine residue Cys = residue Pyr = Pyridine ring Ind = Indolic acid precursor Deala = Dehydroalanine residue Glu = y-hydroxyl glutamic acid But =. Dehydrobutyrine residue Figure 2 The structure of nosiheptide and a shorthand

designation for its structural components. on, it was found that the binding is specifically to the 23S rRNA; ribosomal protein L11 is also prominently involved in the binding of thiostrepton; it is altered or absent in various thiostrepton- resistant mutants. However, these drugs can also bind to certain particles which lack L11 or proteins L7 and L12. These ribosomal proteins are required to interact normally with various supernatant factors, such as elongation factor Tu and elongation factor G and with GTP. Therefore, these antibiotics inhibit aminoacyl-tRNA binding and translocation reactions. Resistance to these antibiotics is due to methylation of a distinct site on the 23S rRNA by a specific methylase, resulting in the inhibition of antibiotic binding.15'17

Although nosiheptide is used commercially as a growth promotant for poultry18 and thiostrepton is used in veterinary medicine in the treatment of bovine mastitis, none of the thiopeptide antibiotics have found applications in human medicine. This is due essentially to their low solubility, which prevents parenteral use, and poor uptake from the gastrointestinal tract, which results in the development of resistance before high enough serum concentrations to kill pathogens can build up. Clearly, structural modification of these antibiotics would be highly desirable, but in view of their structural complexity, the prospects for achieving this by total synthesis or chemical structure modification are not very good. Therefore, biological structure modification is a worthwhile alternative 6 approach. Such an approach requires an understanding of the biosynthetic pathway, its genetic control and of the molecular mode of action of these compounds.

There are two types of biosynthetic mechanisms involved in peptide antibiotic synthesis. One is a protein template mechanism and the other is the ribosomal protein synthesis mechanism. Work on the biosynthesis of peptides and proteins has been carried out for more than 40 years. As the determination of protein structures progressed, the involvement of nucleic acid templates, a general polymerization mechanism and ribosomal machinery in their formation became evident. In the sixties, some attention was focused on alternative processes, independent of nucleic acids, for the formation of certain smaller peptides.20 During the seventies, several multienzyme systems were isolated and studied,20 some of which contain multifunctional polypeptides and an essential thiol group. It is now generally accepted that most microbial peptide antibiotics are synthesized by a protein thiotemplate mechanism. However, peptides containing non-protein constituents may also be derived from precursors of ribosomal origin, as has been demonstrated for ,21 subtilin22 and epidermin23. Therefore, it is necessary to clarify the biosynthetic mechanism of each individual antibiotic peptide. The mechanism of biosynthesis can be determined by inhibitors of protein synthesis, such as chloramphenicol, borrelidin, or puromycin. synthesis should not be inhibited by these agents. Nosiheptide formation by S. actuosus was tested for inhibition by chloramphenicol and found to be insensitive to this agent, suggesting that it involves nonribosomal synthesis. This was recently further probed by R. Woodman of our group, who confirmed that concentrations of chloramphenicol which

inhibit protein synthesis in S. actuosus and S. azureus, respectively, did not inhibit the synthesis of nosiheptide and thiostrepton. Furthermore, oligonucleotide probes coding for the peptide sequences corresponding to nosiheptide and thiostrepton, respectively, were synthesized and used to hybridize against the DNA of S. actuosus and S. azureus; no homologous sequences were detected. All this evidence strongly points to a template-directed process for these thiopeptide antibiotics.

The objective of the research presented in this thesis was to study the biosynthetic pathway of thiostrepton. There are several related compounds, such as nosiheptide and berninamycin, which have been studied by feeding isotope-labeled precursors to the producing organisms followed by analysis of the resulting compounds for isotope incorporation. Berninamycin (Figure 3) is a composed of ten residues, including two 5- methyloxazole rings, five dehydroalanines and a dehydrobutyrine residue.24-25 The feeding experiments with 14C-labeled amino acids suggested that the dehydroalanine residues arise from serine and the thiazole moiety is derived from cysteine. Feeding experiments with 8

Berninomycyl

i n °®a,e Dtoto ^ I '^NH / \Thr OtolQ CH*tfJ Nj© °H i n nh

C-NH

h / i o j J H' "CO-C.A*N h.

Hyvot- D®oto«— 000,0

Figure 3: Berninamycin H za p HOCH2CH-C02H I HN NHz * NHz HSCH z-CHCO eH I A NHz HOCHz—CHCOzH I

Noslheptide

- y T * 's J iv K h / > = o „ r ^ HN

HaC

Figure 4: Labeling pattern of nosiheptide from various 130-labeled amino acids. 10

13C-labeled amino acids in the case of nosiheptide26 (Figure 4) further proved the origins of these moieties. L-[1-13Cj- And L-[3- 13C]serine specifically labeled C1 and C3 of the dehydroalanines, repectively. [3-13C]Cysteine labeled C5 of the thiazole rings. The labeled also labeled the pyridine ring of nosiheptide. The indole moiety was shown by feeding L-[2, 1'-13C]tryptophan to be the product of an intramolecular rearrangement of tryptophan.

Some preliminary feeding experiments have also been carried oiit with the producer of thiostrepton by D. O'Hagan and Pei Zhou in our group. To analyze the isotope incorporations into the molecule, complete and unambiguous assignments of all the signals in the 13C- NMR and 1H-NMR spectra needed to be established. Thiostrepton was the subject of earlier NMR investigations by Tori et al.27 and subsequently by Hensens and Albers-Schonberg28-29. A reexamination of the NMR data by employing a variety of 2D NMR techniques on unlabeled and biosynthetically multiple labeled samples of thiostrepton was performed in our group (see Results Section). The feeding experiments with 13C-labeled precursors in S. laurentii (Figure 5) gave the following results.

The three dehydroalanine moieties are labeled specifically by serine but not cysteine, indicating that they arise by 2

Figure 5: Labeling pattern of thiostrepton from various 13C-labeled amino acids. 12

The thiazole and thiazoline rings each originate from a molecule of cysteine and the carboxyl group of an adjacent amino acid, which was clearly proven by the specific incorporation of cysteine and serine. The mechanism of thiazole ring formation still needs to be examined. The thiazoline ring has R configuration at C4, corresponding in configuration to a D-cysteine. It will be important to know whether this stereochemistry reflects that of the original precursor amino acid in the peptide chain or whether it is generated subsequently in the process of post-assembly modification of this cysteine moiety.

The piperideine ring of thiostrepton as well as the pyridine ring of nosiheptide appear to be formed from two molecules of serine. This could involve the direct tail-to-tail joining of two intact molecules of serine or derivatives thereof, e. g., dehydroalanine as proposed by Bycroft.30 Alternatively, serine may be broken down metabolically, e. g., to glycine and CH2-H4folate, and the pyridine/piperideine ring assembled specifically from these pieces.

The quinaldic acid moiety is also the product of a transformation of tryptophan. The latter accounts for all the carbon atoms except C12, which is contributed by methionine. A methylation of tryptophan at C2 of the indole ring is the first step in this transformation, as demonstrated by efficient and specific incorporation of D,L-2-methyl-[3'-13C]tryptophan. Trapping experiments with D,L-2-methyl-[3'-13C]tryptophan (200 mg/L) showed 5-10% dilution of the isotope in the reisolated material. Butanol extraction of the mycelia followed by derivatization and GC- MS revealed the presense of 2-methyltryptophan, with concentration highest just prior to the appearence of thiostrepton. Cell-free extracts of 36 hr old mycelia of S. laurentii catalyzed the formation of frifiated 2-methyltryptophan from tryptophan and [methyl- 3H]AdoMet (Figure 6). Surprisingly, this reaction proceeds with net retention of configuration of the methyl group. In a series of experiments (mefhy/-R)-[methyl-2H 1,3H]methionine gave 2- methyltryptophan and thiostrepton carrying an R methyl group; and the S isomer gave 2-methyltryptophan and thiostrepton carrying an S methyl group. This methylation reaction does not involve an intermediate methylene group, as in sterol side-chain methylation, because L-[methyl-13C,2H3]methionine is incorporated with complete retention of all three deuterium atoms.

The objective of this thesis research was to continue to study the biosynthesis of thiostrepton. The origins of some of the other moieties, such as the butyrine, threonine, thiostreptin, isoleucine and alanine moieties, still needed to be determined. The mode of formation of some of the highly modified building blocks, such as the thiazole and thiazoline rings, the quinolinic acid moiety, piperideine ring, and their stereochemistry all needed to be studied COOH c h 2 COOH c CH< CH I T NH2 80 % e.e. R 75 % e.e. S

D NH

COOH

COOR HN* 9 R OH

Fs69; 65% 8.8. R F=72; 76% e.e. R Fs33; 59% 9 .8 . S Fs31; 66% 8.8. S

Figure 6: Incorporation of chiral methionine into quinaldic acid moiety of thiostrepton. further. The source of several hydroxy groups which seem not to arise from the original amino acids and the origin of the terminal amide function will also be discussed in this thesis. RESULTS

FERMENTATION

Streptomyces laurentii was used for the feeding experiments producing thiostrepton. Preliminary work was also carried out with

S. azureus as well as S. hawaiiensis. In both cases, however, the yield of thiostrepton was low and the CHCI3 extract contained more impurities compared to that obtained with S. laurentii. At the enrichments achieved, the 13C-NMR analysis required at least 30 mg of pure material in order to obtain interpretable spectra. The fermentation of S. laurentii normally yielded about 70 mg/l.

The storage condition of the culture proved very important for good production. S. laurentii on agar slant kept in a refrigerator degenerated within three monthes. During several transfers from slant to slant, the culture also degenerated. To circumvent this problem, a well producing cultures was grown up in seed medium for 48 hrs and then stored at -78°C in 2 ml aliquots with 15% glycerol (frozen vegetative mycelia, or FVM). The FVM was used to inoculate a first set of slants. The slants were grown for 3 days and transferred

16 17 to new slants. The second generation of slants produced higher thiostrepton levels and was used to inoculate the first stage seed jnedium for all experiments.

Five ml of seed culture were used to inoculate each production culture of 50 ml production medium in a 250 ml Erlenmeyer flask. During the fermentation in the production medium, the temperature must be kept between 26 and 27°C; if it exceeds 28°C, the yield of thiostrepton will be much lower and the product contains more impurities.

ISOLATION OF THIOSTREPTON

Since at least 95% of the thiostrepton is contained in the mycelium, the antibiotic was recovered by repeatedly homogenizing the culture with an equal volume of chloroform in a blender. After separating the chloroform layer and evaporation of the CHCI3, the residue was dissolved in a minimum amount of CHCI3 and the thiostrepton precipitated out with hexane. In most cases, after several reprecipitations, the antibiotic was sufficiently pure for NMR analysis. If necessary, further purification by HPLC was carried out. In the feeding experiments with radioactive precursors, the thiostrepton was purified to constant specific radioactivity by redissolving the precipitate in CH2CI2/EtOH 4:1 and precipitating it 18

with diethyl ether, in addition to the chloroform/hexane precipitation.

THIOSTREPTON ASSAY BY HPLC AND PRODUCTION CURVE

Thiostrepton appears on TLC plates at Rf=0.6 after development

with dioxane/hexane 6:4. Although TLC can be used to check the purity of the antibiotic quickly, HPLC provides a more accurate and rapid quantitative assay. The HPLC system used here was based on the one employed for nosiheptide purification, which was developed by D. Houck26. The system utilizes a Hamilton PRP-1 column and a mobile phase consisting of 60% CH3CN in water. Injections into acetonitrile/water resulted in partial precipitation of the antibiotic on the column, leading to peak tailing. A trace of acetic acid was injected together with thiostrepton to increase the solubility of thiostrepton in organic solvents.

The time course of thiostrepton production was analysed in production medium (50 ml in each 250 ml flask) inoculated with 5 ml of 48 hours old seed culture. Aliquots of the production culture were extracted with CHCI3 and the concentration of thiostrepton was determined by analytical HPLC with UV detection (254 nm filter), using a calibration curve obtained with standard reference thiostrepton. The results are illustrated in Figure 7. Thiostrepton production started at 30 hrs after inoculation and peaked at 72~80 19 mg

20

0 12 24 36 48 60 72 84 96 hr

Figure 7: Thiostrepton production curve in S. laurentii fermentation 2 0 hours. Therefore, in the feeding experiments, the precursors were added to the production cultures in two portions, at a time shortly following the initiation of antibiotic synthesis (30~32 hrs) and at 52 hrs, the midpoint between start and maximum production (72~80 hrs).

NMR SPECTROSCOPY

For the purpose of establishing the biosynthetic pathway to thiostrepton, we needed to confirm the earlier 13C-NMR and 1H-NMR assignments which were based entirely on 1D NMR spectroscopy27’29. We used CDCI3-CD3OD (4 : 1) as the solvent for both 13C-NMR (Figure

11) and 1H-NMR spectroscopy (Figure 8).

All proton signals were assigned by 2D NMR techniques, such as 1H-1H correlation spectroscopy (COSY) (Figure 9) and 1H-13C long range correlation (COLOC) (Figure 10). Except for thiazole Thz(2)H5 (8.13 ppm) which showed a cross correlation to ThstnH2 (5.62 ppm), the other 3 ThzH5 and quinaldic acid QH3 protons did not display any correlations in the COSY spectrum. Their assignment was achieved by the COLOC spectra, in which the ThzH5's displayed two-bond couplings to the corresponding ThzC4 as well as three-bond coupling with ThzC2. The QH3 proton also showed a three-bond coupling to the carbonyl carbon of the quinaldic acid moiety. The assignments are given in Table 1. 9.50 9.00 8.50 8.00 7.50 7.00 6.50 6 00 5.50 5.00 4.50 4.00 3.50 3.00 3.50 3.00 1.50 PPM

Figure 8 :1H-NMR spectrum of thiostrepton in CDCIg : CDgOD 4:1 2 2

THIOSTREPTON IN CDCL3: MEOH 4: 1. 303K C0SY-90

j UL JJL si •**Ai ^ ^ ^ M j w A _ jl_

1.00

* •» 2 .0 0 B-- It— X—

3.00

1

4.00 J

5.00

6 .0 0

*■ 1 7.00 ■=}

5 8.00

9.00

. 10.00

wH.ti|m»iwi|iiniitW|Wwmiprm m ijiiiiiiiii|iiim iii|im ittm itiwnii| wm ^ ^iii|iiiiiinninim iil|wiwiwt;wwwt^niwiiin|wmiiw[miww^iiii jiiiini PPM 10.00 9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 PPM

Figure 9 :1H-1H COSY spectrum of thiostrepton in CDCL : CD3OD 4:1 THIOSTREPTON IN CDCL3: MEOH 4; 1. 313K 6 HZ-TUNED COLOC

Figure 10:1H-13C COLOC spectrum of thiostrepton in CDCI3 : CD3OD 4:1 170 160 140 130 120 100 90 60 70ISO 60 40 PPM

Figure 11:13C-NMR spectrum of thiostrepton in CDCIg : CDgOD 4:1 25

The 13C-NMR spectrum (Figure 11) revealed signals for all 72 carbons. Multiplicities (10 quartets, 7 triplets, 23 doublets and 32 singlets) were determined from an edited distortionless enhancement by polarization transfer (DEPT) analysis.

Proton signals were assigned to their respective carbons through their one-bond couplings based on the 1H-13C COSY spectrum (Figure 12). In the 1H-13C COLOC experiments, the assignments of Deala(2)CO (162.00 ppm) and Thz(2)C2 (166.38 ppm) were confirmed by three- bond coupling to both Deala(2)H3 protons (6.54 and 5.48 ppm) and to Thz(2)H5 (8.13 ppm), respectively. Thz(3)C2 (169.97 ppm) displayed a two bond coupling to Thr(2)H2 (5.62 ppm). QC2, Thz(2)CO and Deala(3)CO were assigned based on 13C-13C correlations from the 2D-INADEQUATE spectrum of thiostrepton obtained in a feeding experiment with L-[1,2-13C2]serine (Figure 13), which will be described later.

Multiple labeling with stable isotopes combined with NMR analysis is an important tool for the elucidation of biosynthetic pathways leading to secondary metabolites.31-32 w hen two adjacent carbons are simultaneously enriched in carbon-13, the coupling of the NMR signals can serve as an assignment criterium, and at the same time solve a biosynthetic problem if the retained coupling is the result of a biochemical process. For both purposes, the feeding experiments with L-[1,2-13C2J- and [2,3-13C2]serine were performed. THIOSTREPTON IN COCL3: ME OH 4:1 C/H CORRELATION SPECTRUM J.M.B.

JL i 1 LU JU l. i i jlJLI i I-jul.

. 4.00

. 5.00

. 6 .0 0

. 7.00

6.00

PPM 130 ISO 110 90 40 SO 100 PPM

Figure 1 2 :1H-13C correlation spectrum of thiostrepton in CDCI3 : CD3OD 4:1 ro O) 27

Ala(1)C2 1000

2000

3000 CysCO CysC4

4000

NEHTZ 100 170 160 150140 130 120 100 80

Figure 13:2D-INADEQUATE spectrum of thiostrepton from [1,2-13C2]serine feeding 28

6000

6500

7000 y OCO OC2

7500

f j— BOOO Thz(2)CO j 8500

9000

9500

HERTZ 170 160 '« 150 140 130 120 PPM

Figure 13(Continued): 2D-INADEQUATE spectrum of thiostrepton from [1,2-13C2]serine feeding 29

. -siooo

Ala{l)C3

. -10000

PipCS . -9000

. -eooo

CysC4 . -7000

T i - r "T“ T* T i i HEflTZ 90 60 70 60 SO 40 30 20 10

Figure 14:2D-INADEQUATE spectrum of thiostrepton from [2,3-13C2]serine feeding 30

1000

sooo

3000

4000

CC2 0C3 sooo

eooo

170 160 190 140 130 120 110 too 80

Figure 14(Continued): 2D-INADEQUATE spectrum of thiostrepton from [2,3-13C2]serine feeding 31

Table 1. I3C and ’H NMR spectral data of thiostrepton (tf in ppm, J in Hz). lie CO 173.71s Deala(3) 3 104.271 6.37 £(d. 1.17). 5.52 Z Ala(2) CO 173.31s (d. 1.17) C ysC O 171.97s Deala(2) 3 103.281 6.54 £ (d. 1.93). 5.48 Z Cys 2 170.16s (d. 1.93) Thz(3) 2 169.97 s D eala(l) 3 102.791 5.66 £ ( d . 1.92), 5.19 Z T hz(l) 2 169.73 s (s. br) A la(l) CO 168.84 s Cys 4 78.98 d 4.83 (dd, 9.60. 12.60) Thz(4) 2 168.35s Thsln 3 77.19s Thz(2) 2 166.38 s Thr(2) 3 72.00 d 6.19 (m) Deala(3) CO 165.99 s Thstn 4 67.73 d 3.67 (d. 6.48) T hr(l) CO 165.47 s Q 8 67.34 d 4.32 (d. 1.75) Deala(l) CO 162.92s T h r(l) y 66.46 d 1.35 (m) Thz(2) CO 162.10s lie 2 65.60 d 2.81 (d. 4.58) Deala(2) CO 162.00 s Q 11 64.35 d 5.16 (d. 6.43) Pip 2 161.94 s Pip 6 64.21 d 5.16 (s. br) Thz( 1) CO 161.67 s Q 7 59.02 d 3.46 (dd, 1.70, 5.50) Q CO 160.80 s Pip 5 57.57 s Thz(4) CO 159.56 s Thr(2) 2 55.83 d 5.62 (d. 7.62) Thz(3) 4 157.17s T h r(l) 2 55.63 d 4.27 (dd, 3.25. 7.60) Q 9 154.55 s Thstn 2 53.06 d 5.62 (d, 9.90) Q 4 153.41s Ala(2) 2 51.95 d 4.59 (pd, 6.48, 7.77) Thz(2) 4 150.09 s A la(l) 2 49.35 d 3.67 (dq. 5.40, 6.81) Thz(4) 4 149.92 s lie 3 38.49 d 1.69 (m) Thz(l) 4 146.40 s Cys 5 34.781 3.51 (dd. 9.00. 11.50) Q 2 143.56 s Pip 4b 29.16t 3.92 (mot-H) 2.16 (m p-H) Deala(2) 2 134.20 s lie 4 24.571 1.21 (m), 0.95 Im' Deala(3) 2 133.02 s Pip 3 24.571 3.32 (m p-H), 2.76 (m oc-H) But 3 132.53d 6.07 (q, 7.10) Q C H 3 22.53 q 1.21 (d. 6.61) D eala(l) 2 132.28 s T h r(l) C H 3 18.92 q 0.70 (d. 6.50) Q 6 129.99 d 6.23 (dd, 5.62. 9.90) A la(l) C H 3 18.88 q 1.03 (d. 6.80) But 2 128.45 s Ala(2) C H 3 18.81 q 1.28 (d, 6.62) Thz(4) 5 127.55 d 8.14 (s) Thr(2) CH3 18.68 q 1.56 (d, 6.47) Q 10 127.20 s Thstn 3-CH3 18.34 q 0.99 (s) Thz(2) 5 125.36 d 8.13 (s) Thstn 5 15.88 q 1.15 (d, 6.63) T hz(l) 5 124.88 d 8.01 (s) He 3-CH 3 15.52 q 0.82 (d, 6.91) Q 5 123.15 d 6.73 (d, 10.08) But C H 3 15.20q 1.46 (d, 7.05) Q 3 122.26 d 7.13 (s) H e5 11.25 q 0.74 (t, 7.13) Thz(3) 5 118.16 d 7.40 (s) Deala(2) NH 9.82 (s, br); piperidine (Pip) 5-NH 9.72 (s, br); Deala(3) NH 8.96 (s, br); Thr(2) NH 8.62 (d 8 80)- But NH 8.47 (s, br); Dea)a(l) NH 7.82 (s, br); Ala(l) NH 7.62 (d, 5.33); Thsln NH 7.43 (d, 9.92) Ala(2) NH 6 99 (d, 7.78); Thr(l) NH 6.91 (d. 7.66). • The unusually low frequency chemical shift of Thr(l) 3-H (1.35ppm) results from a strong shielding interaction due to the proximity of the proton to the ring currents of the dihydroquinoline ring81. k The well-known deshielding of an equatorial proton in a six-mcmbered ring explains the resonance of Pip 4-H, at 3.92 ppm. Even greater downfield shifts for the corresponding protons were found for related compounds like the thiopeptins81. The deshielding is probably enhanced due to the orientation of the proton to the thiazole and an acylamino substituents at C-58'. Table 2. ’’C-^C-Coupling patterns In 1 derived from L-{1,2-13CJ- and L-p.S-^CJserine

L-[2,3-1*CJserine experiment: L-(1,2-,sCJserine experiment:

Carbon signal JE Carbon signal Carbon signal (8, ppm) (Hz) (Hz) (8, ppm) (Hz) (Hz)(8, ppm)

Ala(2)C3 (18.81) 37 36 Ala(2)C2 (51.95) 53 53 Ala(2)CO (173.31) CysC5 (34.78) 31 31 CysC4 (78.98) 57 56 CysCO (171.97) PipC4 (29.16) 35 34 PipC5 (57.57) 59 61 Thz(1)C2 (169.73) Ala(1)C3 (18.88) 34 34 Ala(1)C2 (49.35) 53 53 A!a(1)CO (168.84) PipC3 (24.57) 47 49 PipC2 (161.94) 49 45 Thz(4)C2 (168.35) Deala(3)C3 (104.27) 77 77 Deala(3)C2 (133.02) 65 65 Deala(3)CO (165.99) Deaia(1)C3 (102.79) 80 77 Deala(l)C2 (132.28) 67 67 Deala(1)CO (162.92) Thz(2)C5 (125.36) 67 66 Thz(2)C4 (150.09) 78 78 Thz(2)CO (162.10) Deala(2)C3 (103.28) 77 76 Deala(2)C2 (134.20) 68 68 Deala(2)CO (162.00) Thz(1)C5 (124.88) 68 66 Thz(1)C4 (146.40) 78 77 Thz(1)CO (161.67) QC3 (122.26) 63 63 QC2 (143.56) 85 85 QCO (160.80) Thz(4)C5 (127.55) 65 65 Thz(4)C4 (149.92) 78 78 Thz(4)CO (159.56) Thz(3)C5 (118.16) 67 67 Thz(3)C4 (157.17) 59 59 PipC6 (64.21) 33 The 2D-INADEQUATE technique can detect the signals at all the double quantum frequencies by incrementing the evolution time following the generation of double quantum coherence. Carbons sharing the same double quantum frequency are necessarily connected and the coupled carbons will give signals at a double quantum frequency equal to the sum of their chemical shifts, relative to the transmitter (Figure 13 + 14). Since the INADEQUATE technique33 is inherently insensitive, only 13C-enriched pairs will be detected for thiostrepton (Table 2). Most correlations were shown in the 2D INADEQUATE experiment by coupling, except for PipC3-PipC2 which was determined in 1D 13C-NMR.

Cultures of S. laurentii were fed a number of 13C-labeled amino acids to obtain positional information on the origin of the residues that make up thiostrepton. Relative 13C abundances were calculated by normalizing the peak integrals within specific regions of the spectra of natural abundance and labeled samples.

D,L-[1-13C]ISOLEUCINE SYNTHESIS AND FEEDING

Labeled isoleucine was synthesized from K13CN by a Strecker reaction (Figure 15). Since the precursor, 2-methylbutanal, is a mixture of enantiomers, the product contains two sets of diastereomers, D,L-[1-13C]isoleucine and D,L-[1-13C]alloisoleucine. after derivatizing the product mixture to the N-pentafluoropropronyl CHjCHoCHCHO + K13CN 1-n h 3.C2H ^H ^H 4CI CH,CH./CHCHCOOH I 2 . H C lfco n c.-) | I CH3 V 1 C H /IH 2

Figure 15: Synthesis of D,L-[1-13C]isoleucine lleCO

I

Thz(2)C2

T~ —T" t - T* 174 172 170 166 164 162 PPM

Figure 16: The 13C-NMR spectrum of thiostrepton biosynthesized from D,L-[1-13C]isoleucine CHaCHjjCHCHCOOH CHjNH*

0ch 2ch 3

HO. H H h3c4 n

HNS O HtC ho ' = x h 3 1 H

h 3C OH

Figure 17: The incorporation of D,L-[1-13C]isoleucine into thiostrepton 37

isopropyl esters, we checked the derivatives by GC-MS on a chiral column. The chromatograms showed four peaks with similar mass spectral fragmentation patterns as expected. The ratio of the four peaks is 3.2 : 1 : 3.2 : 1.2, therefore the ratio of D,L-[1-13C]- isoleucine to alloisoleucine is either 74% to 26% or 26% to 74%. Co­ injection with the derivative of L-isoleucine established that the minor component is isoleucine and the major one is alloisoleucine.

A sample of the mixture of isoleucine and alloisoleucine was fed

to cultures of S. laurentii. The resulting thiostrepton showed 13C- enrichment in the carboxyl group of the isoleucine moiety (173.71 ppm, 4.1% enrichment) and also in C-2 of the thiazole(2) ring (166.38 ppm, 6.4% enrichment) (Figure 16, 17). The latter observation indicates that the thiostreptine residue arises from isoleucine, presumably by some oxidative modifications.

D,L-[1-13C]THREONINE SYNTHESIS AND FEEDING

The Strecker reaction was also used to synthesize threonine carrying a 13C label in the carboxyl group (Figure 18). The synthesis started with a reduction of pyruvic aldehyde dimethyl acetal by NaBH4. The hydroxy group was then protected by benzylation

followed by treatment with TFA (trifluoroacetic acid) to afford D,L- 2-benzyloxypropionaldehyde which was used in the Strecker reaction. During the reaction, the benzyl protecting group was CHjCCH(OCH3)2 NaBH4 ► CH3CHCH(0CH3)2 O OH

1.NH4CI, k 13c n . c h 3c h c h ( o c h 3)2 — -ch3chcho ------3— ----- ► 2. HCS (eonc.J OBz OBz

CHjCHCHCOOH I I o h n h 2

Figure 18: Synthesis of D,L-[1-13C, 3-2H]threonine 39

j ' X j L

I ' 1 ■ ' "1"’ I "T* * ' |" -T—I ■ I "*"* ‘ V '-* ' r , r’' I 1 ■'’ 1 ■ 1 1 I 1 ' ’ I ' ■ \ ■ I ■ I 1 ■ ■ I ■ ■ ■ 1 . I ■ I '» «1 1 «'■ ' I ■ , ■ I ■ | ■ fi.00 4.90 4.M 4.40 4.80 4.00 S.«0 8.80 8.40 8.80 8.80 8.80 8.80 8.40 8.80 8.00 1.80 1.80)L 1.40 1.80 1.00 .80 .80

Figure 19: The 1H-NMR spectrum of standard L-threonine (top) and synthesized D,L-threonine /

L JU l J l

' ’ I 1 ' ■ 1 I '■ r - Y 8 .8 0 8.00 4.80 4.00 S .80 8.00 mi 2.80 2.00 S. 80 1.00 .80 O.t

Figure 20: The 1H-NMR spectrum of D,L-[1 -13C,3-2H]threonine IJeCO

Thz(3)C2 Thz(2)C2 CysC2

Thr(1)CO

tn in IMIR IM

Figure 21: The 13C-NMR spectrum of thiostrepton biosynthesized from D,L-[1 -13C,3-2H]threonine CHjCHCHCOOH I I o h n h 2

z H H3c vv CH2CH3

HO. H H

HN^O

HO = CH3 H

0 /CH3 HO E H3C OH

Figure 22: The incorporation of D,L-[1-13C,3-2H]threonine into thiostrepton 43

removed by hydrolysis to produce the free hydroxy amino acid directly. Four isomers (two diastereomeric racemates) were obtained. By comparison with the 1H-NMR of standard L-threonine, it was deduced that L-threonine and its enantiomer made up 84.2% of the mixture (Figure 19). Therefore 42.1% of the mixture was L- threonine. L-[1-13C, 3-2H]Threonine was synthesized by the same procedure, using NaB2H4 in the reduction step and K13CN in the

Strecker reaction step (Figure 20). This material contained 37% L- threonine.

In analogy to the origin of the dehydroalanine moieties, the butyrine moiety is formed by dehydration of threonine, as evidenced by approximately equal enrichments in C1 of threonine(l) (165.47 ppm, 1.8% enrichment), C2 of thiazole(3) (169.97 ppm, 1.7% enrichment ) and C2 of the thiazoline residue ( 170.16 ppm, 1.2% enrichment ) after feeding the mixture of D,L-[1-13C]-threonine and -allothreonine(Figure 21, 22). Interestingly, substantial incorporation of 13C from C1 of threonine is seen into the two carbons derived from the carboxyl group of isoleucine ( Thz(2)C2 : 166.38 ppm, 2.3% enrichment; lleCO: 173.71 ppm, 2.5% enrichment ), reflecting the role of threonine as a precursor in isoleucine biosynthesis34. The presence of deuterium from C3 of the threonine precursor is not detected in thiostrepton by 13C-NMR analysis due to the intrinsic line broadening and the low intensity of the natural abundance lines of the attached carbons. 4 4

FORMATION OF PIPERIDEINE RING

The biosynthesis of the piperideine ring is highly unusual. The incorporation of D,L-[1-13C]- and D,L-[3-13C]serine into thiostrepton suggested that the piperideine ring may be formed by a "tail-to-tail" joining of two serine units, the nitrogen and carbons-2,3,4,5 being derived from serine, while carbon-6 is donated by a molecule of cysteine. But there is an alternative route for the biosynthesis of the piperideine ring that can lead to the labeling patterns observed in the [13C]serine experiments: Glycine, derived from carbons 1 and 2 of serine could be the precursor of the Thz(1)C2-PipC5 and Thz(4)C2-PipC2 pieces while methylenetetrahydrofolate derived from C-3 of serine provides PipC3 and PipC4. These possibilities, and possible other alternatives, can be distinguished by probing with "bond-labeled" precursors, i.e., precursors carrying multiple intramolecular stable isotope labels at the two termini of specific bonds, which of the bonds in serine are incorporated intact into the product.

The general rationale for the use of such "bond-labeled" precursors is as follows: In the added precursor the two isotopes at the two ends of a bond are present strictly in the same molecules, as evidenced by 13C-13C coupling or a 2H isotope shift on a 13C-NMR signal, as the case may be. When the labeled precursor is fed to a culture it will be diluted extensively by unlabeled endogenous 45 material. Hence, if the bond between the two isotopes is broken, the two pieces will each be diluted with large amounts of unlabeled material. In any subsequent bond formation between these two positions the statistical probability of recombining two isotopic nuclei in the same molecule is very small, resulting in loss of the 13Q..13Q coupling or 2H isotope shift. Conversely, observation of retention of the coupling or isotope shift indicates that the two isotopes have been incorporated specifically from the same precursor molecule, suggesting that the bond between them has been transferred intact from precursor to product.

In the present case, the two precursors used to probe for intact incorporation of serine were L-[1,2-13C2]- and L-[2,3-13C2]serine

(Figure 13, 14). The results found in Table 2 demonstrate clearly that the segments PipC3-PipC2-Thz(4)C2 and PipC4-PipC5-Thz(1 )C2 are each derived intact from serine. In the L-[2,3-13C2]serine feeding experiment, the 2D-INADEQUATE spectrum of the resulting thiostrepton showed the connectivity PipC3-PipC2 and PipC4-PipC5 with C-C coupling constants: 47 Hz for PipC3 , 49 Hz for PipC2, 35 Hz for PipC4 and 34 Hz for PipC5. The experiment with L-[1,2- i3C2]serine demonstrated the biochemical connectivities PipC2

(J=49 Hz)-Thz(4)C2 (J=45 Hz) and PipC5(J=59 Hz)-Thz(1)C2(J=61 Hz) (Figure 31). The mechanism for connecting the hydroxymethyl carbons of the two serine residues is intruiging and merits further study. Feeding experiments with L-[3-13C,2H2]- and 2S,3S-[3-13C,2H1]serine may provide some information about the origin of the hydrogens at piperideine C3 and C4. If a 13C with its attached deuterium is incorporated into the metabolite, the deuterium, proton decoupled 13C-NMR spectrum will show separate peaks for deuterated and non- deuterated 13C nuclei since the a-deuterium isotope effect results in a slight upfield shift of the deuterated 13C resonance (~0.2-0.3 ppm per D atom). Upon removal of the deuterium decoupling the deuterated 13C signals disappear into the background due to 2H coupling and quadrupolar line broadening. In the thiostrepton from the L-[3-13C,2H2]serine feeding experiment, the 13C-NMR (Figure 23) showed that one deuterium was incorporated at PipC3, resulting in a 0.25 ppm upfield shift of the signal at 24.33 ppm, but did not give a clear result for PipC4 because of an overlap with other peaks. The 1D {1H,2H} 13C-NMR spectra of thiostrepton from the 2S,3S-[3- 13C,2H-|]serine feeding (Figure 23) did not give a clear result for

PipC3 and PipC4 either due to the same problem. Fortunately, an alternative experiment, 2D selective 1H,13C heteronuclear shift correlation NMR with 2H decoupling35 (2D Hetcor NMR), resolved the overlapping resonances. The selective heteronuclear correlation pulse sequence with deuterium decoupling and with the variable pulse angle 0 set to 90° gave simpler CH plots, which display only carbons bearing a single proton. In the spectrum of the thiostrepton 47

PipC-4 PipC-3 PipC-4

T" _T_ 2 8 2 8 2 6 2 4 PPM PPM

Figure 23: The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 2S,3S- [S-^C^H^serine (left) and 2S-[3-13C,2H2]- serine (right) at PipC3 and PipC4 region. uu J i l -X ■>. -.A -I

0.0

. 9 .0 0 PipH-3a

PPM 60 40 PPM

Figure 24: The 2D Hetcor NMR of thiostrepton biosynthesized from 2S-[3-13C,2H2]serine

■t* 00 -jlJL L

S .0 0

. 9 .0 0

PipH-4a

. 6.00 3

. 7 .0 0

. 1.00

60 40 9 0

Figure 25: The 2D Hetcor NMR of thiostrepton biosynthesized from 2S,3S-[3-13C,2H1]serine 1 L _ j l I I. — . I L — ______K,Ji

• Deala(1)H-3E

! 1 . > Deala(3)H-3E 1 t Deala(2)H-3E , 1

t QH-3

4 Thz(3)H-5

Thz(4)H.5 | |' Ttal1)H-5 Thz(2)H-5

~ l >------r — I------1------1------r— i----- ■------1 | I r — i------1 : r— —t— I i [— >40 IM HO >10 >00

Figure 25(Continued): The 2D Hetcor NMR of thiostrepton biosynthesized from 2S,3S-[3-13C,2H1]serine cn o from the L-[3-13C,2H2]serine feeding (Figure 24) there exists one correlation between PipC3 (24.57 ppm) and PipH3a (2.76 ppm), which confirms that there is one deuterium incorporated at PipH3, and demonstrates that it is located in the p position. For PipC4, the spectrum does not show any correlation; thus it could be either labeled by two deuteriums or none. However, in the spectrum of thiostrepton from the 2S,3S-[3-13C,2H1]serine feeding (Figure 25), a correlation signal for PipC4-PipH4a (3.92 ppm) was evident, which indicates that one deuterium is incorporated at PipH4p. This also implies that there must be two deuterium atoms incorporated at PipC4 from L-[3-13C,2H2]serine. Furthermore, no deuterium is incorporated at PipC3 from 2S,3S-[3-13C,2H1]serine, which shows that the pro-S hydrogen from C3 of serine is the one which is removed during the formation of the piperideine ring (Figure 32,33).

MODE OF INCORPORATION OF SERINE INTO OTHER STRUCTURAL ELEMENTS

The 13C-13C coupling patterns seen in the three dehydroalanine moieties of thiostrepton after feeding L-[1,2-13C 2]- and L-[2,3- 13C2]serine (Table 2, Figure 31) confirmed the intact incorporation of serine into these moieties. Two deuterium atoms each were retained in the dehydroalanine moieties from L-[3-13C,2H2Jserine (Figure 26) since {1H, 2H} 13C-NMR showed peaks shifted about 0.50 ppm upfield from the corresponding natural abundance 13C signals. 52

Deafa(2)C-3 Deala(3)C-3

|Deala(1)C-3 Deala(2)C-3

Deala(3)C-3 | Deala(1)C-3

106104 106 104 102

Figure 26: The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 2S>3S-[3-13C,2H1]serine (left) and 2S-[3-13C,2H2]- serine (right) at the DealaC3 region. The stars denote deuterated 13C nuclei signals. 53 One deuterium atom per dehydroalanine unit was retained from 2S,3S-[3-13C,2H1]serine (Figure 26). 2D Hetcor NMR (Figure 25) of the latter thiostrepton sample showed correlations between C3 and the Z hydrogen of the dehydroalanine units (Figure 33) which was therefore derived from the pro-R hydrogen of C3 of serine.

13C-13C couplings were also observed in the two alanine moieties of the thiostrepton samples derived from L-[1,2-13C2]- and L-[2,3- 13C2]serine as precursors (Table 2, Figure 31), albeit less intense, indicating some direct conversion of serine into alanine, probably via pyruvate. As expected, two deuterium atoms and one deuterium, respectively, from L-[3-13C2,2H2]- and 2S,3S-[3-13C2,2Hi]serine were incorporated into each alanine moiety (Figure 27).

The four thiazole rings as well as the thiazoline ring and their attached carboxy groups (or equivalent carbon, in the case of thiazole(4)) each arise from a molecule of cysteine and the carboxyl group of an adjacent amino acid, which provides C2 of the rings. The intact incorporation of serine into these moieties was confirmed by observation of the expected 13C-13C couplings in the thiostrepton samples from the [1,2-13C2]- and [2,3-13C2]serine feedings (Table 2,

31). One atom of deuterium was retained at C5 of each thiazole ring after feeding L-[3-13C,2H2]serine (Figure 28), whereas no deuterium was incorporated from 2S,3S-[3-13C,2H^serine (Figure 29, 25).

Therefore, the hydrogens at C5 of the thiazole rings are derived from 54

I

I

t b W f I Ala(1)Me I Ala(2)Me Vi*. I*" Ala(1)Me

Ala(2)Me

"T “ 7“ T- T” 20 18 20 IB

Figure 27: The 13C-NMR(bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 2S,3S-[3-13C,2H1]serine (left) and 2S-[3-13C ,2H2]- serine (right) at the AlaMe region. The stars denote deuterated 13C nuclei signals. 5 5 the pro-R hydrogen at C3 of serine (Figure 32, 33). The results from the Hetcor NMR analyses of these samples regarding the alanine, dehydroalanine and thiazole ring moieties are all consistent with the 1D {1H,2H} 13C-NMR experiments. 1D 13C-NMR spectra of thiostrepton from the L-[3-13C,2H2]serine feeding clearly showed a

0.51 ppm upfield shifted peak for the signal at 33.90 ppm, which indicates two deuterium atoms were incorporated at C5 of the thiazoline ring. However, the feeding experiment with (2S,3S)-[3- 13C,2H-iJserine did not give a clear indication of deuterium incorporation at that position in either the broadband or single­ frequency 2H decoupled 1D 13C-NMR or the 2D {1H,2H}1 H,13C-Hetcor NMR experiment.

FORMATION OF QUINALDIC ACID MOIETY

MODE OF REARRANGEMENT

That the methylation of tryptophan is the first step in the formation of the quinaldic acid moiety of thiostrepton, preceding the rearrangement, has been proven by feeding and trapping experiments with D,L-2-methyl-[3'-13C]tryptophan and by enzymatic studies. Further conversion of 2-methyltryptophan may involve either (a) cleavage of the N1/C7a bond and connection of the side chain nitrogen to C7a or (b) cleavage of the N1/C2 bond and connection of C2' to N1. These two possibilities were distinguished by feeding a Thz(1)C-5

QC-3

T "T T" I— 128 126 124 128 118 PPM

Figure 28: The 13C-NMR(bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 2 S-[3 -13C,2H2]serine at the ThzC5 and QC3 region. The stars denote deuterated 13C nuclei signals. 57

Thz(4)C-5 Thz(2)C-5 Thz(1)C-5 4 ll Thz(3)C-5 s/

I —r~ I I 128 126 124 122 120 118 PPM

Figure 29: The 13C-NMR (bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from 2S,3S-[3-13C,2H1]serine at the ThzC5 and QC3 region. The star denote deuterated 13C nuclei signals. 58

CysC-5 CysC-5

36 34 36 34

Figure 30: The 13C-NMR(bottom) and 13C-NMR with deuterium decoupling (top) spectra of thiostrepton biosynthesized from aS.SS-p-^C^HJserine (left) and 2S-[3-13C,2H2]- serine (right) at CysC5 region. The stars denote deuterated 13C nuclei signals. o

HN = o NH

NH I

NH

O — HN NH HO NH

,OH OH HO

Figure 31: The intact incorporation of L-[1,2-13C2]serine and L-[2,3-13C2]serine into thiostrepton. Heavy lines denote coupling between carbon atoms. 0 . DD. D

D D

= H H3C x h 2ch 3

H* ° " l M - C D HO. HH HO^lCH 0 s

HN 0 H*C HO = CH3 H

H,C3 OH

Figure 32: The incorporation of 2S-[3-13C,2H2]serine into thiostrepton.«=13C ■0

UN

NH ,ch 2ch 3

ch 3

HO NH

HN,

Figure 33: The incorporation of 2S,3S-[3-13C,2H1]serine into thiostrepton. • =13C 6 2 sample of L-tryptophan labeled with 15N in the indole nitrogen and with 13C in both CT and C2' of the side chain. The thiostrepton from this experiment showed 13C-enrichment in the carboxyl group and C2 of the quinaldic acid moiety and one bond coupling of these two signals to each other. In addition, the QC2 signal at 143.56 ppm displayed a 3.02 Hz one-bond coupling to 15N and the QCO at 160.8 ppm showed a two-bond coupling of 8.08 Hz to the 15N (Figure 34). Hence, option (b) must be followed in the ring expansion to the quinaldic acid system.

STEREOCHEMISTRY OF HYDROGEN LOSS FROM THE B-CARBON OF

IB Y PIQ P HAN

During the rearrangement of 2-methyltryptophan, one of the methylene hydrogens originating from C3' of tryptophan is eliminated. The stereochemistry of this process can be determined by feeding mixtures of S-[3'-14C]tryptophan with either (2'S,3'S)- or (2'S,3'R)-[3,-3H]tryptophan. If the process of hydrogen elimination is completely stereospecific, one of the double labeled precursors will ideally give thiostrepton with an unchanged ratio of 3H/14C and the other will produce thiostrepton containing 14C but no tritium, giving a 3H /14C ratio of zero. In other words, the tritium retention51 will be 100% in one case and 0 in the other. It is possible to determine, in an extension of this experiment, whether only L-tryptophan or also the D isomer serves as a precursor36. This simply requires repeating v V W^fy

162 160 144

h

y^VvV^yVyv»Y V “1— “I— _ i — “1------1— —1— “I— “1— —I— 162 ise 156 154 tag 150 140160 146 144 PPM

Figure 34: The 13C-NMR spectrum of thiostrepton biosynthesized from L-[indole-15N, 1\2'-13C2jtryptophan. the experiment with mixtures of the chirally tritiated L-tryptophan samples and D,L-[3'-14C]tryptophan. One mixture will still give 0% tritium retention. The other will give 200% tritium retention if only L-tryptophan is used, because half the 14C will be in a biologically inactive form. If D and L form are used equally well, the tritium retention will be 100%, and if the D form is used more efficiently than L-tryptophan the tritium retention will be < 100%. Of course, these numbers describe an idealized situation; in reality one can expect significant deviations from this ideal case due to a variety of partial racemization processes taking place in the biological system .

The (2'S,3'R)-[3'-3H]tryptophan was synthesized from indole and (2S,3S)-[3-3H]serine with tryptophan synthase37, whereas (2'S,3'S)- [3'-3H]tryptophan was prepared in the same way from (2S,3R)-[3'- 3H]serine. The results obtained in the feeding experiments with mixtures of these precursors with 14C-tryptophan are summarized in Table 3. (^S.S'RHS'^HJtryptophan mixed with S-[3'-14C] or R,S-[3'- 14C] tryptophan lost more tritium during the conversion to thiostrepton than the mixtures of (2,S,3'S)-[3,-3H]tryptophan with the same [14C]tryptophan samples. This result indicates that the process involves at least preferential, if not stereospecific elimination of the pro-R hydrogen from C-3' of tryptophan. This was later confirmed by the feeding experiment with 2S,3S-[3-13C , 2H]serine which was discussed earlier. However, quantitatively the 65 figures are far from those predicted for the ideal case, indicating that the added precursor must be subject to extensive, ancillary metabolic reactions which lead to considerable racemization at the tritiated center.

Table 3 also shows that the feeding of the chirally tritiated substrates with L-[3'-14C]tryptophan gave higher tritium retentions in thiostrepton compared to the mixtures with D,L-[3'- 14C]tryptophan. This would suggest that D-tryptophan is incorporated more efficiently than the L isomer. While this may mean that D-tryptophan is a more immediate precursor of thiostrepton than the L isomer, it is also possible that the L-form is the more immediate precursor, but it is used extensively for protein synthesis before antibiotic production reaches maximum rates whereas more of the D isomer, which enters the cells more slowly, is still available during the period of most rapid antibiotic synthesis. The latter situation has precedent in the biosynthesis of another antibiotic, pyrrolnitrin, from tryptophan38.

ROLE of 4-f 1 -H YDROXYETH YLVQUI NOUN E-2-CAR B_QXYLIC_ACI D

A logical advanced intermediate in the formation of the quinaldic acid moiety of thiostrepton would be 4-(1- hydroxyethyl)quinaldic acid. To test this notion, the compound was synthesized in labeled form from quinaldic acid (Figure 35). The Table 3 : 3H/14C Ratios in feeding experiments with radiolabeled tryptophan

Precursor 3H/14C in 3H/14C in thiostreDton tryptophan tryptophan3 thiostrepton6 tryptophan

(2,S,3,S)-[3,-3H]/D,L-[3,-14C] 5.7 3.0 0.53 (ZS.S’RJ-P'^Hl/D.L-P’-^C] 2.0 0.80 0.40 (2,S)3,S)-[3,-3H]/L-[3,-14C] 2.47 2.0 0.81 (2,S,3,R)-I3,-3H]/L-[3,-14C] 3.60 1 -6(5) 0.46 a 3H/14C ratio of aliquots cocrystallized with unlabeled thiostrepton to constant 3H/14C. b Average of measurements. 67 acetyl group was introduced at C4 of quinoline by a homolytic acylation reaction with acetyl radicals39. Tritium was introduced by reduction of methyl 4-acetylquinaldate with [3H]NaBH4. After hydrolysis of the tritiated compound, 4-(1 -hydroxy- [1- 3H]ethyl)quinaldic acid was obtained. The proper position of the label was confirmed by 3H-NMR which showed a single peak at 5.90 ppm (Figure 36). The final compound was fed to the thiostrepton producer. The radioactivity of the thiostrepton obtained (4mg) corresponded to 10% incorporation of the precursor. Specific incorporation into thiostrepton was confirmed by precipitation of the material from different solvents to give constant specific radioactivity. The supernatant from each precipitation was checked by HPLC and it was found that the peak of thiostrepton in each case matched the peak of radioactivity.

ATTEMPTED ™ 02 FEEDING EXPERIMENT

A number of oxygen functions in thiostrepton are most likely not carried over from the precursor amino acids. These are the oxygens at C3 and C4 of the thiostreptine residue, the oxygen at C8 of the quinaldic acid and that at C11. In an attempt to obtain information on their origin, fermentations were carried out in a closed system containing 180 2 gas (50 atom% 180). It was hoped that isotopic substitution with oxygen-18 could be detected by natural abundance 13C-NMR, based on the upfield shift in the 13C-NMR signals of 68

CHjCHO FeS04»BuOOH

COOH COOCH3

NaOH

COOCH3

HO i.CH

HCI COOH

Figure 35: The synthesis of tritiated 4-(1 -hydroxyethyl)quinoline-2-carboxylic acid 320.13 MHz TRITIUM NMR ANALYSIS P I/3 SHIFTED SINE-BELL PROCESSING COMPOSITE PULSE PROTON DECOUPLED

i... I... 9.00 e.oo 7.00 6 .0 0 S .00 4.00 3.00 S .00 1.00 PPM

Figure 36: The 3H-NMR spectrum of 4-(1-hydroxy-[1-3H]ethyl)quinaldic acid 70 directly attached carbons according to previous studies.40 The magnitudes of these shifts range from about 0.01 ppm to 0.05 ppm. Unfortunately, no 180-shifted peaks were discernible in the 13C-NMR spectrum, presumably because of the intrinsic line broadening due to the size of the molecule. A similar fermentation with nosiheptide under a 98% 180 2 atmosphere was carried out by another member of our group, and an equally negative result was obtained in the 13C- NMR spectrum, although FAB mass spectrometry clearly showed the presence of 180 in the molecule.

15N PRECURSOR FEEDINGS

A curious feature of the thiopeptide antibiotics, including thiostrepton, is the presence of an unsubstituted amide function at the carboxy terminus of the peptide chain. This amide function may arise by amidation of a carboxy group via an activated ammonia derivative, e.g. carbamylphosphate, as the amide function of glutamine,34 or it may be the remnant of an amide bond to another amino acid, as, for example, the terminal amide function of a number of mammalian neuropeptides, like TRH, LHRH, etc.. In the latter cases, the precursor peptide chain contains an additional carboxy terminal glycine residue which is removed by an oxidative process, leaving its nitrogen behind as the terminal amide function.41*42 By a similar process, any other amino acid might, of course, also serve as the donor of the terminal amide nitrogen in the thiopeptide antibiotics. To shed light on this issue it was desirable to establish the origin of the various nitrogens, and particularly of the terminal amide nitrogen, in these thiopeptide antibiotics, by feeding

experiments with 15N-labeled precursors. Such experiments with S.

actuosus have shown that the terminal amide nitrogen of nosiheptide is derived specifically from serine (R. Tsuchiya, unpublished results), i.e., the precursor peptide giving rise to nosiheptide must carry at least one extra carboxy-terminal serine in addition to the amino acids which are conserved in the final structure, and possibly additional ones.

In an attempt to find out where the terminal amide nitrogen in thiostrepton comes from, feeding experiments with 15NH4CI, [15N]glycine and 15NH4CI plus a 3 fold excess of nonlabeled serine were performed followed by 15N-NMR analysis of the products. Since the production medium contains large amounts of cornsteep liquor, soybean flour and (NH4)2S 0 4, all very good nitrogen sources, the two feeding experiments with 200 mg 15NH4CI and 200mg 15N-glycine in 700 ml production medium did not give very good enrichments due to the large dilution of the 15N. Therefore, the unlabeled nitrogen sources in the medium needed to be reduced. Fermentations with reduced amounts of (NH4)2S 0 4 in the production medium and with equal amounts of NH4CI instead of (NH4)2S 0 4 were carried out.

Comparing the yields from these different fermentations did not show much difference. The thiostrepton samples from feeding 72 experiments with 500 mg of 15NH4CI and 500mg 15N-glycine without (NH4)2S 0 4 in the production medium gave significant signals in the

15N-NMR spectra.

For the further analysis of these spectra it was necessary to work out assignments of the signals in the 15N-NMR spectra of thiostrepton. Since the 15N-NMR assignments are based on the N-H correlation in 2D-NMR, the solvent of the NMR sample had to be changed from the usual CDCI3-CD3OD 4:1 to DMSO-D6 to avoid the exchange of the N-H protons to N-D. As a consequence, the 1H-NMR signal assignments needed to be repeated in this solvent. This was done by low temperature 1H-NMR at atemperature of 320°K (Figure 37), 1H-NMR with D20 exchange (Figure 38) and 1H-1H correlation

spectroscopy (Figure 39).

The quartet of dehydrobutyrine ButH3 at 6.38 ppm (J=6.96Hz), the triplet of lleH5 at 0.82 ppm and the singlet of Thstn3(CH3) at 0.91 ppm were assigned directly from the 1D 1H-NMR spectrum. Four sharp singlets at 8.07, 8.11, 8.32 and 8.48 ppm should represent ThzH5 of the four thiazole rings. One singlet at 7.64 ppm was assigned to the quinaldic acid QH3 proton.

Exchangable protons can often be identified by addition of a drop of D20 to the sample resulting in the disappearance of peaks. In the spectrum of thiostrepton, 9 peaks were no longer observed after D20 73 exchange (Figure 38). These were at 9.89, 9.37, 9.12, 9.07, 9.39, 8.33, 8.11, 8.05 and 7.70 ppm. Their assignments were determined by 1H-1H COSY spectroscopy.

In accord with the COSY shift correlations and coupling constants (Table 4), the doublet at 1.73 ppm is assigned to the proton of ButCH3 and the one at 9.12 ppm correlating to the proton of ButCH3 is assigned to ButNH. A set of signals at 0.82 (t), 1.18 (m), 1.42 (m), 1.78 (m), 0.89 (d) and 3.10 (overlap) are assigned to lleH5, lleH4, lleH4, lleH3,lle3(CH3) and lleH2, respectively. The methylene groups

of three dehydroalanine residues are observed as three sets of two singlets, each set being correlated to one NH. Their assignments are: 9.89 (DealaNH), 6.41 (DealaH3E), 5.72 (DealaH3z); 9.37 (DealaNH), 6.04 (Deala H3E), 5.70 (DealaH3z); 9.12 (DealaNH), 5.60 (DealaH3E) and 5.31 (DealaH3z). Another two sets of signals are assigned to the two alanine moieties: 8.33 (AlaNH), 4.32 (AlaH2), 1.25 (AlaCH3); 8.05 (AlaNH), 4.25 (AlaH2) and 1.15 (AlaCH3). Significant signals at 6.84

(d), 6.25 (dd), 3.42 (m), and 3.61 (d) are assigned to QH5, QH6, QH7, and QH8. The methyl group at 1.01 ppm is assigned to QCH3 and one proton at 5.60 ppm correlated to QCH3 is assigned to QH11. The doublet at 5.38 ppm with correlation to 8.39 ppm is assigned to ThstnH2; it becomes a singlet after D20 exchange. One methyl group at 1.06 ppm as well as a proton at 3.65 ppm are assigned to ThstnCH3 and ThstnH4, respectively (Table 5). The signal correlated to ButCH3 is assigned to CysH4, the signals at 3.42 and 3.62 ppm are 74 then assigned to CysH5. Two more sets of signals could be assigned to the two threonine moieties: 0.93 (ThrCH3), 4.08 (ThrH3), 4.60 (ThrH2), 7.70 (ThrNH); 1.25 (Thr’CH3), 5.08 (Thr'H3), 4.60 (Thr'H2) and 8.05 ppm (Thr'NH). The assignments of these two threonine moieties as well as the piperideine and cysteine moieties, based on these three NMR experiments, are only tentative.

The 1H-signal assignments (Table 5) served as the basis for assigning the 15N resonances. Multiplicities of the 15N signals were determined by a 15N DEPT experiment (Figure 40), which cleanly distinguished the terminal amide nitrogen from other resonances. A 2D-inverse heteronuclear multiple quantum correlation experiment (HMQC) with BIRD pulse, which only detects protons coupled to 15N, was employed in the assignment of the protonated nitrogens (Table 6, Figure 41). Since the non-protonated nitrogens were of no concern in the present experiments, no effort was made to assign them.

Figure 42 shows the 15N-NMR spectra of thiostrepton obtained after feeding 15NH4CI, [15N]glycine, and 15NH4CI plus a 3-fold excess of nonlabeled serine. As can be seen, no preferential labeling of the dehydroalanine moieties by glycine was observed. This is probably due to rapid metabolic interconversion of different nitrogen sourses in this organism. To determine if serine (which is biosynthesized from glycine) is the source of the amide nitrogen, a 3-fold excess of nonlabeled serine was fed along with 15NH4CI. The NMR analysis 75 showed no significant relative intensity differences compared to the previous experiment with 15NH4CI alone although all enrichments were diminished. If serine is the immediate source of the amide nitrogen, we would expect the relative intensities of the signals for the terminal amide nitrogen and the dehydroalanine nitrogen to remain the same relative to each other but to decrease relative to other signals. If the terminal amide nitrogen is furnished by glycine, we would expect that only the dehydroalanine nitrogen signals but not the terminal amide nitrogen signal should decrease relative to other signals. The data seem to be consistent with the notion that serine is the source of the amide nitrogen, but they do not prove it. A more definite resolution of this issue, as in the case of nosiheptide, would require a strain in which different nitrogen sources are metabolized more selectively, i.e., more slowly. Attempts to use S. azureus or S. hawaiiensis instead of S. laurentii, in hopes that they might meet this requirement better, had to be abandoned because both organisms produced much less thiostrepton than is needed for 15N-NMR experiments.

In the 15N-NMR and 15N DEPT experiment, there always appeared signals of a second similar compound in the sample. This second compound might be a closely related metabolite or a decomposition product of thiostrepton. It was not possible to separate this second compound from thiostrepton. THIOSTAEPKM IN *MS0 fH20-SUPPflESSI0N)

T -~T~ TT M lr" "T ■"T~ I ■~r~ ” I' ■ ■"T- ”7^- i’ ■

Figure 37: The 1H-NMR spectrum of thiostrepton in DMSO-d6 THIOSTREPTON IN OMSO (4020)

6.50 8.00 7.50 7.00 6.50 6.00 5.00 4.00 3.50 3.002.50 2.00 S.50 1.00 .50 PPM Figure 38: The 1H-NMR spectrum of thiostrepton in DMSO-d6 with D20 exchange. 78

THIOSTREPTON IN 0NS0-08 COSY-BO 4K

jlJLl j i J j L

a J 1 .00

3.00

3 .0 0

4.00 i

< 0.00

6 .0 0

« 7.0 0

6 .0 0

8.0 0

10.00 PPM 8.00 7 .0 06.00 6.00 8.00 4.00 3.00 2.00 1,00

Figure 39: The 1H-1H COSY spectrum of thiostrepton in DMSO-d, ZPZ THIOSTREPTON M~1S OEPT 11

T T T T T T T ▼ T T T T T T T T T T “|- 120 110 too 140

Figure 40: The 15N DEPT spectrum of thiostrepton THIOSTREPTON IN OMSO INVERSE 1H/15N HNQC ANALYSIS J.N. K A L E 80

1 K U kL J '

100

. 110

. 120

. 130

. 140

PPM 10.00 8.60 8.00 8 .8 0 8.00 7 .8 0 7 .0 0 PPM

Figure 41: The 2D-inverse heteronuciear multiple quantum correlation with BIRD pulse spectrum of thiostrepton 81

[15N]glcine

15NH4CI

15NH4CI with 3 fold excess serine

b&l " tit '* Mf ’ " im BM W W ** sW ’J gig 1 »}|,JB»l IW w Figure 42: The 1SN-NMR spectra of thiostrepton biosynthesized from 15N-labeled precursors. 82

Table 4 : 2D-NMR COSY shift correlations of thiostrepton

Assegnment Chemical shifts(ppm)

lleH5 0.82 / \ / \ lleH4 lleH4 1.18 1.42 \ / \ / lleH3 1.78 / \ / \ lleH3(CH3) lleH2 0.89 3.10

ButH-3 ButCH3 ButNH 6.38 «— > 1.73 <—* 9.12 j / CysH4 5.23 / \ / \ CysH5 CysH5 3.42 3.62

DealaNH 9.89 / Y / \ DealaH3 « DealaH3 5.72 « —> 6.41 83

Table 4 (Continued)

DealaNH 9.37 / \ / \ DealaH3 *-* DealaH3 5.70 * —> 6.04

DealaNH 9.12 / \ / \ DealaH3 DealaH3 5.60 < -----> 5.31

QH5 6.84 / \ / \ QH8«*QH7 » QH6 3.61 3.42 < > 6.25

QH11 * QCH3 5.60 «— » 1.06

AlaNH AlaH2 «—► AlaCH3 8.33^4.32^1.25

AlaNH AlaH2 ► AlaCH3 8.05 <*4.32 **1.15

ThstinH2 « * ThstinNH 5.38 ^— * 8.39

ThstinH4 « ► ThstinCH3 3.60 «— » 1.06 84 Table 5 :1H-NMR assignments of thiostrepton

Assignment Chemical shifts(ppm) Coupling Constants(Hz)

lleH5 0.82(t) 7.26 lle3(CH3) 0.89(d) 6.90 Thstn3(CHg) 0.92(s) ThrCHg 0.93(d) QCH3 1.01 (br) ThstnHS 1.06(d) 6.36 AlaCHg 1.15(d) 7.08 lleH4 1.18(m) Ala'CHg 1.25 Thr'CHg 1.27 lleH4 1.42(m) BUtCHg 1.73(d) 6.90 lleH2 3.10 QH7 3.42 CysH5 3.42 ThstinH4 3.60 QH8 3.61 CysH5 3.62 ThrH3 4.08 AlaH2 4.32 Ala'H2 4.32 ThrH2 4.60 Thr'H2 4.60 Thr'H3 5.08 85

Table 5 (Continued)

CysH4 5.23 DealaH3(c) 5.31 (s) ThstinH2 5.38(d) 8.94 QH11 5.60 DealaH3(t) 5.60 Deala'H3(c) 5.70 Deala"H3(c) 5.72 Deala'H3(t) 6.04 QH6 6.25(dd) 10.1+3.31 ButH3 6.38(q) 6.96 Deala"H3(t) 6.41(d) 0.98 QH5 6.84(dd) 10.1+1.31 QH3 7.64(s) ThrNH 7.70 Thr'NH 8.05(br) AlaNH 8.05(br) ThzH5 8.07(s) Thz'H5 8.11(s) DealaNH 8.11(br) Thz"H5 8.32 Ala'NH 8.33 ThstinNH 8.39 Thz"'H5 8.48 ButNH 912 Deala'NH 9.37 86

Table 6 :2D-NMR 1H/1SN shift correlations of thiostrepton

Assignment Chemical shifts(ppm) isN 1H

DealaNH 129.12 9.89 AlaNH 128.66 8.33 ThstinNH 128.28 8.39 DealaNH +ButNH 124.42 9.12 DealaNH 123.53 9.37 AlaNH 116.09 8.05 DISCUSSION

The thiostrepton molecule consists of several distinct structural moieties, such as the thiazole and thiazoline rings, the piperideine ring, the quinolinic acid moiety and some moieties, which all present different biosynthetic problems.

The quinolinic acid moiety is formed from L-tryptophan, as had already been determined earlier by specific incorporation of L-[1\

13C]tryptophan. The latter accounts for all the carbon atoms except

C12, which is contributed by methionine. Since 2 -methyltryptophan was found to be a precursor of thiostrepton, the quinolinic acid

moiety is the product that results from the rearrangement of 2 - methyltryptophan. The quinoline nucleus is found in a variety of naturally occurring compounds. Studies of the biosynthesis of a number of these compounds indicate the existence of several pathways leading to this ring system. The biosynthesis of kynurenic acid and its relatives, which is well understood, is one of these pathw ays .43 Tryptophan serves as the precursor of the quinoline ring. The transformation involves cleavage of the C2/C3 bond of the

87 88 indole ring followed by excision of C2 and reconnection of the indole nitrogen to the a-carbon. The quinoline ring of quinine, a plant alkaloid, is also a product of tryptophan metabolism .44-47 This mode of conversion involves cleavage between C2 and the indole nitrogen and connection of the indole nitrogen to the a-carbon of the side chain; C2 of the indole nucleus is preserved in the product. Anthranilic acid, which is derived from tryptophan in some cases43, has also been implicated in the biosynthesis of several quinoline derivatives of bacterial, fungal and plant origin .2 8 -66 The transformation of 2 -methyltryptophan into the quinaldic acid moiety in thiostrepton seems to involve a ring expansion similar to that leading to the formation of the quinine type alkaloids. Unequivocal evidence that the rearrangement involves cleavage of the N1/C2 bond of tryptophan and reconnection of N1 to the a-carbon of the side-chain comes from a feeding experiment with tryptophan labeled intramolecularly with 15N in the indole nitrogen and with 13C at C1' and C2’ of the side chain. The 13C-NMR spectrum of the resulting thiostrepton showed enrichment in the carboxyl group and C2 of the quinaldic acid, with the signal for C 2 displaying one-bond coupling to 15N and that for the carboxyl group showing two-bond 13C -15N coupling. A mechanistically reasonable pathway for the transformation of the indole to the quinoline system, which is consistent with the experimental data, is portrayed in Figure 43. This process has chemical precedent in the hypochlorite-catalyzed conversion of 2-methyltryptophan to 4-acetylquinoline .4 8 89

Incorporation of tritiated 4-(1-hydroxyethyl)quinolinic acid into thiostrepton suggests that the methylation and ring expansion of tryptophan occur prior to the attachment of this moiety to the remainder of the molecule (Figure 44).

The dihydrobenzene structure in the quinaldic acid moiety of thiostrepton might be produced by epoxidation at C7/C8 of the aromatic ring followed by attack of the amino acid nitrogen of isoleucine on the epoxide. This proposal has precedent in the biosynthesis of gliotoxin and aranotin49-50(Figure 45). The amino- alcohol systems were suggested to arise by the interaction of an epoxide with the phenylalanine-derived nitrogen atom. This notion is

supported by the fact that [ar- 2 H5]phenylalanine was incorporated _ into gliotoxin without loss of any of the aromatic hydrogen atoms, and by the non-incorporation of labeled m-tyrosine. The stereochemistry of the dihydrobenzene structure in thiostrepton is consistent with this proposed origin, since opening of the epoxide by

attack of a nucleophile would be an anti process.

The biochemical origin of the dehydroamino acid residues of peptide antibiotics has been the subject of some speculation through

the years .24>25-52*53 It has been shown that the dehydroalanine

moieties in berninamycin24,25 and nosiheptide 37 are formed by

elimination of water from serine. This is also the case for the three dehydroalanine residues in thiostrepton. Intact incorporation of NH NH

COOH COOH

CH

H H NH CH COOH

CH O COOH NH

CH

HN' OH

o<0 Figure 43: Formation of quinaldic acid moiety. Figure 44: incorporation of tritiated 4-(1-hydroxyethyl)quinolinic acid into thiostrepton. r 0H phenylalanine .... v' II I* 0 H"'JL/OH Cyclo-L-phenylalanyl H°2C -L-seryl serine

iT-s gliotoxin NH

OH

Figure 45: Biosynthesis of gliotoxin. 93 serine into these moieties is confirmed by the coupling patterns observed in thiostrepton after feeding [1,2- 13C]- and [2,3- 13C]serine.

The incorporation of D,L-[1- 13C]threonine into the butyrine moiety supports the general hypothesis on conversion of (3-hydroxyamino acids to a.p-unsaturated residues. However, the stereochemistry of the dehydration had not been studied. In most enzymatic eliminations of water where the stereochemistry is known, such as in fumarase and enolase the elimination is anti. Fumarase first breaks the C-OH bond to generate a carbocation intermediate, whereas enolase cleaves the C-H bond to produce a carbanion intermediate. Some enzymatic eliminations of water are known that proceed with overall syn geometry, such as those catalyzed by 3- methylglutaconyl-CoA hydratase and 5-dehydroquinate dehydrase; In our case, the conversion of (2S,3S)-[3- 13C, 2H ^ se rin e into thiostrepton showed incorporation of the deuterium atom into the E position of dehydroalanine. If we assume that the serine residues undergoing dehydration have L configuration, this reaction therefore proceeds via an anti elimination route. The same steric course is evident for the threonine to butyrine conversion from the configurations of precursor and product.

The formation of the four thiazole rings, as well as the thiazoline ring, from cysteine has analogy in the generation of the thiazole rings of nosiheptide from cysteine ,37 the oxazole rings of berninamycin from threonine ,2 4 -25 and the oxazole ring of 94 virginiamycin from serine .53 The intact incorporation of the precursor into these moieties was again confirmed by feeding [ 1 ,2 - i 8C2]- and [2,3- 13C 2]serine. From the experiment with (2S,3S)-[3-

13C,2H ^serine, it follows that the pro-S hydrogen is removed.

Assuming that the conversion of serine to cysteine proceeds in S. laurentii with the same stereochemistry as in E. coli, namely retention of configuration at C354, and that the cysteine residues in the peptide which are converted to thiazoles have L configuration, the dehydrogenation reaction must proceed with a n ti stereochemistry. This contrasts with a variety of dehydrogenation reactions which all occur with syn stereochemistry, e.g., in cryptoechinulin55, mycelianamide56, elaiomycin 57 and N- carbobenzoxy-L-tryptophan 58 but conforms to the stereochemistry of formation of the oxazole ring in virginiamycin59. In the latter case it was suggested59 that the reaction does not proceed by dehydrogenation of a dihydrothiazole, but rather by dehydrogenation of the serine unit to the C3 aldehyde followed by enolization and ring closure of the enol. However, our results argue against an analogous thioaldehyde mechanism for thiostrepton biosynthesis, on the following grounds.

The thiazoline ring may either be an intermediate stage in the formation of the aromatic thiazole ring system, or it may arise by subsequent reduction of a thiazole ring. The feeding experiment with

L-[3-13C,2 H2]serine which revealed the presence of two atoms of deuterium at C-5 of the thiazoline ring rules out the second possibility. Surprisingly, however, we did not detect the required presence of one atom of deuterium at this position in the thiostrepton from the feeding experiment with (2S,3S)-[3- ^C^H-Jserine. Neither broadband nor single-frequency decoupling experiments nor an inverse 2 D{2 H}1H,13C-HETCOR experiments provided clear evidence for the presence of deuterium. Since the evidence for the presence of two atoms of deuterium at that position in the sample from L-[3- 13C,2 H2]serine was unequivocal, the reason for this failure must be technical, e.g., the lower enrichment of the sample. In any case, the retention of both deuterium atoms from C3 of serine in the thiazoline ring shows that the latter is not formed by reduction of a thiazole. Therefore the configuration at C4 of that ring, which corresponds to a D amino acid, signals the presence of a D-cysteine residue at that position in the peptide chain. One may speculate that the other cysteine residues which give rise to thiazoles have L configuration, and that the D configuration at this site is the reason that this ring is not fully oxidized to the aromatic thiazole level. In addition, this result argues against ring formation via a thioaldehyde enol, since such a mechanism would not proceed via a thiazoline intermediate. Of course, the possibility cannot be ruled out that the thiazole and thiazoline rings are formed by mechanistically different paths. 96 The dihydroxyisoleucine moiety (thiostreptine) was found to arise from isoleucine, possibly by 3,4-dehydrogenation, epoxidation and hydration of the epoxide or alternatively, by two consecutive hydroxylations. Determination of the incorporation of 180 2 into this moiety would distinguish the two mechanisms. Since attempts to locate 180 by direct 13C-NMR of thiostrepton were unsuccessful, a different method for the determination of the incorporation of 180 from atmospheric oxygen is needed. One possibility to solve the problem is to degrade the resulting thiostrepton by acid h y d ro ly sis 60’61 followed by analysis of the dihydroxyisoleucine fragment employing 13C-NMR.

The piperideine ring of thiostrepton, in analogy to the pyridine ring of nosiheptide ,37 arises from two molecules of serine which are connected tail to tail through their carbon atoms 3, and from the carboxyl group of a cysteine which provides C 6 . Although labeling of PipC3 and PipC4 by C3 of serine was not clearly proven by the single labeling experiment, due to signal overlap, the double labeling experiments clearly demonstrated the incorporation of two intact serine molecules. This also rules out the possibility that serine gives rise to the piperideine ring indirectly, via metabolism to C ^ and C 2 fragments.26*37 The L-[3-13C,2H2]serine experiment showed retention of one atom of deuterium at C3p and two deuterium atoms at C4, while the (2S,3S)-[3-13C,2 H^serine feeding showed no deuterium at C3 and one deuterium at C4p. A plausible mechanism 97 for this process, adopted from an earlier proposal by Bycroft and

G ow land ,30 seem s to be the cyclo-addition of two dehydroalanine residues followed by dehydration (Figure 46). The stereochemistry of the two dehydroalanine residues undergoing this reaction is presumably the same as that of the other three dehydroalanine moieties which result from anti elimination of water from serine. According to the configuration at C5 of thiostrepton and that at C4 deduced from the (2S,3S)-[3- 13C,2H-Jserine feeding experiment, the ene component must undergo a syn addition on the re-re face. The diene component is supposed to be formed by enolization of 1. Two stereoisomers could be formed, but 2 is the more likely intermediate since the conformation of 2 is closer to that of product, which does not require large steric rearrangement. The 1,4- addition of diene can occur in a syn or anti mode, which could result in either one of four isomers: 4 (3si, 6 si), 5 (3re, 6 re), 6 (3re, 6 si) and 7 (3s/, 6 re). These four isomers will require different stereochemistries for the 1,4-elimination of water: syn from 4, 5 or anti from 6 , 7 . If the elimination is concerted, it should be syn. Therefore, 4 or 5 will be the more likely intermediate. Between these two, again by the argument that the steric arrangement at C 6 in 4 is closer to that of the product, it seem s that it is the more likely intermediate. The final step, 1,4-hydrogenation, must be anti, as demonstrated by the fact that L-3- 13 C,2 H2]serine gave a piperideine ring with the unlabeled hydrogen at C3 in the a-position and by the p orientation of the hydrogen at C 6 . r»r c + HO^ Y X NHR T OH 3 /

,NHR \ H O " > " Ho > "

4 5 6 7

-H20

.NHR [2H‘]

Figure 46: Formation of piperideine ring. In feeding experiment with 2S,3S-[3-13C,2H1]serine, HA=HD= 2H, HB=HC=1H; In feeding experiment with S-[3-13C 2H2]serine, HA=HB=HC=HD= 2H. 99 Since all the modified residues of thiostrepton, as well as nosiheptide, can be traced to natural amino acids, it is logical to assume that the structure is originally synthesized as a linear peptide containing all the amino acids except the tryptophan giving rise to the quinaldic acid in the extra loop. The timing and mode of attachment of this tryptophan residue is still unclear. In the case of nosiheptide, formation of the indolic acid moiety presumably requires the free carboxyl group of tryptophan, and the amino nitrogen of tryptophan has been shown not to be incorporated specifically into any nitrogen of the antibiotic ;37 the tryptophan therefore cannot be involved in a peptide bond. Thiostrepton most likely is formed by a similar pathway in which the tryptophan is not part of the original peptide chain (Figure 47). The results define the minimum sequence of a peptide precursor of thiostrepton as H 2N-L- ller -L-Ala2 -S e r 3 '-L-Ala4'-Ser*-C ys 2-L-Thr3-Thr4-D -C ys 5-L-lle3-

C y s 7 -L-Thrfl-Cys 3-Ser*3-Cys * 7 -S er* 3-S e r* 3-OH. The carboxy- terminal amide nitrogen may be derived specifically from another serine. This was demonstrated clearly in the case of nosiheptide by feeding experiments with 15NH4CI, [15N]glycine, and [ 15N]glycine in the presence of a threefold excess of unlabeled serine. This suggests that the peptide precursor carries at least one more serine, and possibly additional amino acids, at its carboxy terminus. Similar experiments on thiostrepton gave less clear-cut results, but are nevertheless consistent with the idea that the peptide precursor of 100 thiostrepton may also contain at least one more carboxy-terminal serine.

Thiostrepton and related antibiotics share the same overall architecture but are made up of somewhat different building blocks. For instance, the results obtained from studies of nosiheptide biosynthesis suggest that a primary peptide precursor should have the following minimum sequence: H 2 N -S e r 1-C y s 2~L-Thr3-T hr4-

C y s 5-L -G lu 6 - C y s 7 -L -C y s 6 -C y s 9- S e r 10- C y s 11- S e r 12-OH. If one removes the residues lie1' through Ala4' from the thiostrepton precursor peptide, the rest of the sequence would be very similar to that of the nosiheptide precursor except for substitution of lie 6 instead of Glu, Thr 6 instead of Cys and one more serine at the C- terminus. With those modifications of the constituent amino acids, similar peptides give rise to a similar major loop and side chain.

S erin e 1 and serine 10 must be joined to form the pyridine/piperideine ring, cysteine residues numbers 2 ,5,7,9, and 11 form thiazole or thiazoline rings. From these comparisons, it seems very likely that the biosynthetic machineries for thiostrepton and nosiheptide formation may be very similar, particularly with respect to peptide synthesis.

Inhibition experiments with chloramphenicol as well as the negative outcome of hybridization experiments probing for the presence of base sequences coding for the precursor peptides of Nosiheptide Precursor

H2N-serr-cys2-thr3-thr4-cys5-glu6-cys7- cys3-cys9-serro-cys 11-ser12-ser13OH

trp pyruvate

nosiheptide

Thiostrepton Precursor

H2N-iler-ala2'-ser3'-ala4'-ser?-cys 2 thr3-thr4-cys5- ile6-cys 7-th r3-cys 9-ser 10-cys11-ser 12-ser 13-(ser14)-OH

(pyruvate) hp

thiostrepton Figure 47: Primary peptide precursors of nosiheptide and thiostrepton. 1 0 2 nosiheptide or thiostrepton in the DNA of the producing organisms point to a non-ribosomal process for the assembly of these precursor peptides. With 12-18 amino acids in a non-repetitive sequence, the compounds fall between the simple peptide antibiotics, like gramicidin S, bacitracin or thyrocidin ,20 which are all synthesized on an enzyme template, and the larger - containing antibiotics, like nisin 21 or subtilin22, which are obtained from larger pro-peptides that are assembled ribosomally. This places thiostrepton among the largest peptide antibiotics produced by a non-ribosomal process. The largest peptide which has been shown to be synthesized by a multienzyme thiotemplate mechanism is alamethicin, a nonadecapeptide, as reported by Mohr and Kleinkauf63.

The peptide synthase assembling the precursor could be one multienzyme complex carrying out the entire process from activated individual amino acids to the complete peptide on one carrier without release of any intermediates. Alternatively, each amino acid may be added to a growing peptide by a separate enzyme. Based on the precedent of other peptide antibiotics, the involvement of a multienzyme complex seems more likely. By analogy to the well characterized multienzyme systems for the assembly of peptide antibiotics, like gramicidin S or bacitracin ,20 such an enzyme complex should be quite large, probably having a molecular weight between 1 and 2 million dalton. 103 Quite a number of enzymes must be involved in the elaboration of these highly modified thiopeptides in addition to the peptide synthase (Table 7). In the case of thiostrepton, at least fifteen additional enzymes are predicted to be required, and possibly far more, depending on whether repetitive steps occurring in different parts of the molecule require one or multiple sets of enzymes. Formation of dehydroalanine and dehydrobutyrine from serine or threonine must use at least one dehydratase, more likely two for the two different amino acids, and possibly as many as four if each dehydroalanine is made by a separate enzyme. Thiazole ring formation probably involves cyclization and dehydration, which would require two enzymes. Thiostreptine arises from isoleucine, possibly via dehydrogenation, epoxidation and hydration of the epoxide, which needs three enzymes, or via two consecutive hydroxylations, which would require two enzymes. The piperideine ring was proposed to arise from the cyclization of two dehydroalanines and subsequent dehydration and reduction of the six-membered ring, which requires at least two enzymes in addition to the formation of the dehydroalanine. Rearrangement of 2- methyltryptophan which involves cleavage of the N1/C2 bond of the indole ring, reconnection of the a-carbon of the side chain to the indole nitrogen and reduction of the carbonyl group, requires at least three enzymes. The methylation of tryptophan needs one more enzyme. Thus the most likely number of enzymes required is 20, counting the peptide synthase as one enzyme. Therefore, even by the most conservative estimates at least 20-30 kbp of DNA will be required to code for thiostrepton biosynthesis, and more likely the number will be at least twice that. The cloning and characterization of this DNA, now underway in collaboration with the laboratory of Professor W. R. Strohl64, is a major challenge. 105

Table 7: Required Number of Biosynthetic Genes

Nosiheptide

Min. Likelv Max.

Peptide synthase 1 1 1 Racemase 1 Peptide hydrolase 1 Amide synthesis 1 1 1 Serine/threonine dehydratase 1 2 2 Thiazole formation 2 2 10 Hydroxyglutamate formation 1 1 1 Pyridine formation 3 4 6 Indolic acid formation 4 4 5 Indolic acid attachment 1 2 2 14 17 30

Thiostrepton

Min. Likelv Max.

Peptide synthase 1 1 1 Racemase 1 1 Peptide hydrolase 1 Amide synthesis 1 1 1 Serine/threonine dehydratase 1 2 4 Thiazole formation 2 2 9 Thiostreptine formation 2 3 3 Piperideine formation 2 3 5 Quinaldic acid formation 4 4 6 Quinaldic acid attachment 3 3 3 16 20 EXPERIMENTAL SECTION

MATERIALS AND GENERAL METHODS

All chemicals and solvents obtained from commercials sources were of the highest quality available and most were used without further purification. Reaction solvents were purified by distillation from appropriate drying agents: THF (Na/benzophenone), CH 2CI2

(P20 5), ether (Na/benzophenone), DMF (CaH2). Other solvents were purified as described .70 Ingredients for fermentation media were purchased from Difco, except soybean flour and corn steep liquor which were purchased from Sigma. Oxygen-18 gas was purchased from Cambridge Isotopes. Streptomyces laurentii ATCC31255,

Streptomyces azureus ATCC14921, Streptomyces hawaiiensis ATCC12236 were obtained from the American Type Culture Collection. Authentic thiostrepton was obtained as a gift from Squibb Corporation or was purchased from Calbiochem.

L-[indole- 15N,1',2'-13C2]Tryptophan was synthesized by Dr. Ursula

Mocek from [15N ]indole 66 (90.4%) and L-[1,2- 13C 2]serine with tryptophan synthase and (2S,3S)-[3- 3 H]- and (2S,3R)-[3- 3H ]serine were synthesized by Nobue Fujii in this laboratory54. L-[3- 13C, 2 H2]-, 106 107 (2S,3S)-[3-13C, s h ^ - , L-[1 ,2-13C 2]- and L-[2,3- 13C2]serine were gifts from the Los Alamos Stable Isotope Resource. Tritiated NaBH4 was purchased from ICN Biomedicals, L-[3'- 14C]trytophan from NEN

Research Products and D,L-[3'- 14C]tryptophan from the Amersham Corporation. Tryptophan synthase (15 U/mg, 1 mg protein/ml) was a gift from Dr. Edith W. Miles, NIH, Bethesda. 15NH4CI and [ 15N]glycine were purchased from Aldrich.

2. Ssi2smLM sM s

Melting points were determined on a Mel-Temp Laboratory Device and are uncorrected. Preparative separations were performed by flash chromatography18 on 230-400 mesh silica gel from Aldrich. Routine GC/MS identification of synthetic intermediates was performed on a Hewlett-Packard 5970A gas chromatograph with 5790 mass selective detector. Radioactive samples were counted in a Packard Minaxi-b Tricarb 4000 liquid scintillation counter or a Beckman LS 1801 liquid scintillation counter using Aquasol-2 (New England Nuclear) scintillation cocktail. Counting efficiencies were determined using [ 14C]- and [ 3H]toluene as internal standards. A New Brunswick rotary shaker G25-R was used for fermentations.

The 1H-, 2H-, 3H-, 13C- and 15N-NMR spectra were recorded on an IBM AF-300 spectrometer operating at 7.1 Tesla field strength and equipped with process controller, inverse accessory, inverse probe, 108 and Aspect 3000 data system with array processor. Broadband proton decoupling was employed for 13C-NMR. Chemical shifts are given in parts per million (ppm) relative to (CH3)4Si (TMS) as internal standard or adjusted to the TMS scale by reference to the solvent signal. Coupling constants (J) are given in Hertz (Hz). Splitting patterns are designated: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet. 15N-NMR spectroscopy was performed at a frequency of 30.412 MHz with internal deuterium lock. 2D-lnverse Heteronuclear Multiple Quantum Correlation experiments with BIRD pulse were conducted with the decoupler as the observe channel (90° 1H pulse: 10.5 usee) and the broadband transmitter as the X-nucleus channel (90° 15N pulse: 18.0 usee). To avoid signal modulation by spin rate fluctuations, samples were measured in a non-spinning mode and shimmed accordingly.

FERMENTATION

Trial fermentations were carried out with Streptomyces azure us

(ATCC 14921), Streptomyces laurentii (ATCC 31255), and

Streptomyces hawaiiensis (ATCC 12236), all reported, to produce thiostrepton. Among these three, S. laurentii produced the highest yields of thiostrepton, and was selected for further work. The S. laurentii culture obtained from ATCC was transferred from a lyophile tube and incubated on yeast malt extract agar (YME agar) for | 109 three days at 28°C. Small portions of the slant were used to inoculate 50 ml seed cultures in 250-ml Erlenmeyer flasks in the following seed medium(per liter): soybean flour (15 g), dextrose (30 g), C aC 03 (2.5 g), NaCI (1.0 g). These were incubated at 28°C in a rotary shaker at 250 rpm for two days. After the growth was complete, aliquots (2 ml) were stored aseptically with glycerol (15%) at -78°C (frozen vegetative mycelia or FVM). The FVMs could be stored for at least one year without any decrease of the thiostrepton yield and were used to start a new generation on YME agar. Following incubation for three days at 28°C, the slants were transfered to another new set of slants which were used to inoculate the seed medium of a fermentation.

Of a seed culture, 5 ml was used to inoculate 50 ml of production medium in 250 ml Erlenmeyer flasks, which were grown at 26.5°C for 72 hrs to 80 hrs with rotary shaking at 250 rpm. The production medium contained (per liter): glucose (1 OOg), soybean flour (20g), yeast extract (1.0g), corn steep liquor (35 g), KH2P 0 4 (1.0 g) Na2H P 0 4

(1.4 g), (NH4)2 S 0 4 (1.0 g), C aS 04 (2.0 g), M gS04 7H 20 (0.2 g),

F eS 0 4.7H20 (0.2 g), MnS0 4.2H20 (0.5 g), Na2Mo04-2H20 (5 mg), H3B 0 3

(5 mg), C uS04-2H20 (10 mg) ZnS04-7H20 (10 mg) and CoCI 2.6H20 (50 mg). The pH value of the production medium was adjusted to 7.0 with NaOH before sterilization for 15 min. at 121°C. 110 ISOLATION OF THIOSTREPTON

Following the completion of the fermentation, an equal volume of

CHCI3 was added to the culture and the mixture homogenized in a blender. The aqueous layer with the broken mycelia was then separated from the CHCI 3 layer by centrifugation at 7000 rpm for 10 min. The organic phase was evaporated to dryness under vacuum below 35°C. The extraction of the aqueous phase containing the mycelia was repeated. The residue of the organic phase was dissolved in a minimum amount of CHCI 3 and the thiostrepton precipitated by addition of hexane. The precipitate was collected by centrifugation . The precipitation procedure was repeated twice. If necessary, the precipitate was redissolved in a minimum amount of

CH 2CI2/EtOH, 4:1. and thiostrepton was precipitated by the addition of ethyl ether and collected by centrifugation.The purification of thiostrepton by HPLC was also used sometimes for higher purity. The HPLC system is shown in Table 10.

THIOSTREPTON ASSAY BY HPLC

Rapid assays of thiostrepton in the fermentation broths were carried out using a HPLC methodology as described below. Aliquots

(1 ml) of the culture were taken and agitated with the same amount of CHCI3. The cells were centrifuged down and the organic phase was combined with an equal volume of glacial acetic acid. An aliquot of 111 the resulting solution (15 ui) was injected directly into the HPLC column.

The HPLC system consisted of a Beckman model 116A pump, a Beckman 21OA injector, a Hamilton PRP-1 C18 column (10 urn, 250 x 4.1 mm), a Waters UV absorbance detector, and a Fisher Recordall recorder. The detector was equipped with a 254 nm filter. The solvent system consisted of 60% acetonitrile in water at a flow rate of 0.5 ml/min. The retention time of thiostrepton under these conditions was approximately 9 min.

This method was used to establish the production curve and to monitor fermentations. A set of standard thiostrepton solutions of different concentrations was used for calibration.

FEEDING EXPERIMENTS WITH LABELED PRECURSORS

During feeding experiments, labeled precursors were added to the production cultures shortly after the initiation of antibiotic synthesis, usually 32 hrs after inoculation of the production medium with 5 ml of the seed culture. Each precursor was dissolved in deionized water and sterilized by filtration through a sterile 0.2 urn Millipore filter (Gelman). In the stable isotope labeling experiments, the precursors were fed to the production cultures in two portions Table 8: Feeding experiments with labeled precursor

Precursor Total amount Harvest ml of Yield(mg) of of precursor(mg) time(hr) culture thiostrepton

L-[indole-15N, 1 \2'-13C2]tryptophan 70 72 500 10 L-[1,2-13C2]serine 100 76 700 71 L-[2,3,-13C2]serine 100 72 700 90 L-[3-13C,2H2Jserine 100 76 900 43 L-p-^C^HJserine 100 80 850 45 D,L-[1-13C]isoleucine 100 76 800 80 D,L-[1 -13C]threonine 100 72 750 47 113 at 30-32 and 52 hr in concentrations of 50 to 70 mg per liter each tim e.

The cultures were harvested 72 to 80 hrs after inoculation (Table 8).

SYNTHESIS OF D,L-[1-13C]IS0LEUCINE (1)67,68

Anhydrous ammonia was bubbled into 40 ml of absolute ethanol in a 500 ml round bottom flask for 15 minutes, until the solution was saturated. To this solution was added 35 ml of NH4OH solution (28.8%), 1.36 g (0.02 moles) of K^CN, and 1.26 g (0.023 moles) of dry NH4CI. At room temperature the homogeneous solution was stirred and 2.20 ml (0.021 moles) of 2-methylbutanal was added dropwise. The reaction flask was sealed with a stopcock and allowed to stir at room temperature for 16 hours. Upon completion of the reaction , the ethanol and ammonia were removed from the mixture under reduced pressure on a rotary evaporator to leave a clear oil in the reaction flask. To this was added 20 ml of 12 N HCI and the solution was heated at 100°C for 4 hours. The solution was then cooled, 100 ml of distilled water was added and the mixture evaporated to a small volume to remove HCI vapor. The removal of HCI was repeated three times. Then, 100 ml of distilled water was added and the solution was neutralized with 1N NaOH. The solution was allowed to sit in a refrigerator overnight to give a white solid mixture of isoleucine and alloisoleucine (0.6 g, 23% yield from K13CN 114

99% 13C). 1H-NMR (D20): 5 0.75-1.07 (m,6H), 1.12-1.60 (m,2H), 1.77-2.20 (m,1H), 3.62 (t), 3.67 (t), 3.87 (t) & 3.96 (t) (1H).

DERIVATIZATION OF THE MIXTURE OF ISOLEUCINE AND ALLOISOLEUCINE

To 2 mg of the mixture of amino acids in 0.5 ml of H20 was added 1 ml of 1N HCI (pH<1.0) and the solution was evaporated to dryness. A solution (3 ml) containing 5 ml of isopropyl alcohol and 1 ml of acetyl chloride, prepared at 0°C, was added to the residue. The mixture was transferred to a capped vial and stirred at 100°C in an oil bath for one hour. The solvent was blown off with dry nitrogen at room temperature. Methylene chloride (2 ml) and pentafluoropropanoic anhydride (1 ml) were then added and the solution stirred at 100°C in an oil bath for another hour. After the solution had cooled to room temperature, the solvent was again removed in a stream of dry nitrogen. The product was dissolved in CH2CI2 and analysed by GC-MS on a Chirasil Val III chiral column. Four peaks were observed which had similar or identical fragmentation patterns. Peak 1 (m/z, rel intensity): 264(M+-CH = CH2, 2), 233(17), 232(M+-OCH(CH3)2, 60), 221(13), 204(41), 176(12), 164(10), 119(11), 69(100), 43(59), 41(69), 39(12); Peak 2: 264(1.5), 233(14), 232(48), 221(13), 204(41), 176(11), 164(9), 119(11), 69(100), 57(16), 43(60), 41(75), 39(13); peak 3: 264(2), 233(17), 232(58), 221(14), 204(44), 176(13), 115

164(11), 69(100), 57(13), 43(56), 41(69), 39(12); peak 4: 264(5), 233(24), 232(79), 221(20), 204(59), 176(16), 164(13), 69(100), 57(14), 43(53), 41(59), 39(11). The ratio of the mixture peak 1 : peak 2 : peak 3 : peak 4 = 3.2 : 1 : 3.2 : 1.2.

SYNTHESIS OF THREONINE

D.L-Lactic aldehvde__dimethvl _ acetal (21; Ten ml (10.2 g, 82.7 mmoles) of pyruvic aldehyde dimethyl acetal was dissolved in MeOH (50 ml) and 2.0 g of NaBH4 was added. The solution was stirred for 5 hrs, then 50 ml of water was added and the MeOH was evaporated under reduced pressure. The solution was extracted with ether. The ether layer was dried over MgS04, concentrated and purified by silical gel column filtration with ethyl acetate/hexane 1:4 as eluant to give 5 g of product (48% yield). 1H-NMR (CDCI3): 8 1.62 (d, J=6.0 Hz,3H), 2.75 (broad, 1H), 3.88 (d, J=10.0 Hz, 6H), 4.22 (m, 1H), 4.54 (d, J=5.7 Hz,1H).

1.1-Dimethoxv-2-benzvloxvpropane (3169: To 0.5 g (4.2 mmoles) of 2 in DMF (2 ml) was added 0.28 g (9.3 mmoles) of NaH (80% in mineral oil) at 0°C. After the evolution of hydrogen gas ceased, 0.78 g (4.5 mmoles) of benzyl bromide was added to the solution. The reaction mixture was stirred at room temperature for 1 hr. The solvent was then removed under reduced pressure. The residue was dissolved in 10 ml of water and the solution was extracted with ether to give 116 0.52 g of crude product. Purification by silica gel column chromatography (ethyl acetate/hexane 1:4 as eluant) yielded 0.45 g (50%) of the pure liquid product. 1H-NMR (CDCI3): 5 1.18 (d, J=6.0 Hz, 3H), 3.42 (d, J=3.0 Hz, 6H), 3.56 (m, 1H), 4.20 (d, J=5.5 Hz, 1H), 4.64 (s, 2H), 7.15-7.40 (m, 5H); GC-MS: 179 (M+-OMe, 0.1), 92 (15), 91 (100), 77 (15), 76 (18), 72 (59), 71 (11), 65 (60), 63 (12), 51 (21), 47 (90), 45 (27), 44 (27), 43 (21), 41 (24), 39 (35), 31 (63).

2-fBenzvloxv^propanal (4): To 3 g (14.3 mmoles) of 3 were added 9 ml of TFA and 1 ml of H20 and the mixture w as stirred under nitrogen for 4 days. Most of the solvent was evaporated and the residue was purified by silica gel chromatography (ethyl acetate/hexane 1:4 as eluant) to give 1.90 g (11.6 mmoles) of product (81% yield). 1H-NMR (CDCI3): d 1.31 (d, J=7 Hz, 3H), 3.88 (qd, J=7 , 2 Hz, 1H), 4.58 (d, J=12 Hz, 1H), 4.64 (d, J=12 Hz, 1H), 7.40-7.25 (m, 5H), 9.65 (d, J=2 Hz, 1H); GC-MS: 135 (M+-CHO, 0.9), 134 (9), 92 (10), 91 (100), 64 (19), 51 (8).

Threonine (5^: Fifteen ml of ethanol were saturated with NH3 gas. To the solution were added 10 ml of 14.8 N NH4OH, 0.52 g (7.88 mmoles) of KCN, and 0.50 g (9.3 mmoles) of dry NH4CI. The homogeneous solution was stirred at room temperature and 1.90 g (11.5 mmoles) of 4 was added dropwise over a 10 minute period. The reaction flask was closed with a stopper and the mixture stirred for 24 hrs. Following the reaction, the ethanol and ammonia were removed under 117 reduced pressure on a rotary evaporator. Five ml of conc. HCI was added and the solution heated at 100°C for 3 hrs. After the dark brown solution had cooled, 100 ml of distilled water was added and the solvent was evaporated to remove HCI. The removal of HCI was repeated three times. Fifty ml of water and 200 mg of activated charcoal were then added. The solution was heated to boiling and filtered hot to yield a clear solution which was adjusted to pH 2~3 with 2 N NaOH. The compound was purified on a column (1 cm x 20 cm) of Dowex-50W (H+ form) cation exchange resin (eluant 5% NH4OH) to give 600 mg (4.96 mmoles, 63%) of product consisting of a mixture of threonine and allothreonine containing 42.1% of L-threonine. 1H-NMR (D20): S 1,22 (d) & 1.35 (d) (J=6.6 Hz, 3H), 3.60 (d) & 3.85 (J=5 Hz, 1H), 4.25 (m, 1H).

M-13C. 3-2H1threonine: The procedure was the same as described for the synthesis of unlabeled threonine, except that NaB2H4 (98% 2H) and K13CN (99% 13C, 0.5 g, 7.57 mmoles) were used instead of NaBH4 and KCN. 1H-NMR (D20): 1.20 (s) &1.35 (s) (3H), 3.58 (d) & 3.85 (d) (1H). The yield was 550 mg (4.5 mmoles, 60%). The percentage of L-threonine in the mixture was 37% (99% 13C, 98% 2H).

Synthesis of (2'S,3,RH3'-3H]Tryptophan37

To about 10 uCi of (2S,3S)-[3-3H)serine (100 uCi/umole) were added 10ul of fresh 5 mM pyridoxal phosphate solution (adjusted 118 with NaOH to pH 8), 100 ul of HEPES (N-[2-hydroxyethyl]piperazine-N'-[2-ethanesulfonic acid]) buffer (1M, pH 8), 25 ul of 5 M NaCI and 20 ul of 10 M indole solution. After confirming that the pH of the solution was around 8, 3 ul tryptophan synthase (45 mU) was added and the mixture incubated at 37°C. The percentage conversion of serine to tryptophan was checked by HPLC (see table 9) during the incubation. After 45 min., 67% of the serine had been converted to tryptophan, and after 90 min. and 2 hrs the reaction had tapered off at 75% conversion. A second identical incubation with 20 uCi of (2S,3S)-[3-3H ^serine was carried out and combined with the first. Work up by HPLC (see Table 10) gave 16 uCi of (2S,,3,R)-[3'-3H]tryptophan.

Synthesis of ^'S ^’SMS'^HJTryptophan

The procedure was the same as described for the synthesis of (2S',3R)-[3,-3H]tryptophan. Thirty uCi of 2S,3R-[3-3H]serine (100 uCi/umole) were used as substrate and produced 19 uCi of tryptophan after purification by HPLC.

Feeding Experiment with Mixture of ( 2 'S ,3 'S ) “[3 ,-3H ] - Tryptophan and D,L-[3'-14C]Tryptophan

(2,S,3'S)-[3'-3H]Tryptophan (8.58 uCi) and D,L-[3'-14C]tryptophan (1.51 uCi) were mixed and fed to one 50 ml production culture of 119 Streptomyces laurentii 32 hr after inoculation. The culture was harvested at 76 hrs and extracted. To the labeled thiostrepton was added 50 mg of authentic thiostrepton. Three methods were used for purification of the diluted thiostrepton. The thiostrepton was purified by HPLC (see table 11) to give a 3H/14C ratio of 3.03, whereas recrystallization from CHCI3/MeOH at -20°C gave material 3H/14C=2.97 and preparative TLC (solvent system: dioxane/hexane 6:4, silica gel) gave 3H/14C=2.90.

Feeding Experiment with Mixture of (2'S,3'R)-[3,-3H ] - Tryptophan and D,L-[3'-14C]Tryptophan

(2'S,3'R)-[3'-3H]tryptophan (3.74 uCi) and D,L-[3’-14C]tryptophan

(1.87 uCi) were mixed and fed to Streptomyces laurentii 36 hrs after inoculation of the production culture. The culture was harvested at 72hr. Thiostrepton purified by HPLC gave a 3H/14C ratio of 0.74, that purified by preparative TLC gave 3H/14C=0.87.

Feeding Experiment with Mixture of 2'S,3'S-[3,-3H ] - Tryptophan and L-[3'-14C]Tryptophan

A mixture of (2,S,3'S)-[3'-3H]tryptophan (2.43 uCi) and

L-[3'-14C]tryptophan (0.99 uCi) was fed to a 48 hr old Streptomyces laurentii culture. The culture was harvested at 74 hr. Thiostrepton 120 purified by HPLC gave a 3H/14C ratio of 1.78 wheras purification by preparative TLC gave 3H/14C=2.15.

Feeding Experiment with Mixture of 2'S,3'R-[3,-3H ] - Tryptophan and L-[3'-14C]Tryptophan:

A mixture of (2'S,3'R)-[3'-3H]tryptophan and L-[3'-14C]tryptophan was fed to Streptomyces laurentii 40 hrs' after inoculation. The culture was harvested at 74 hrs. Thiostrepton purified by HPLC gave a 3H/14C ratio of 1.63 and that purified by preparative TLC gave a ratio of 1.67.

Co-crystallization of a Mixture of Tryptophan

To a mixture of (2'Sf3'R)-[3'-3H] tryptophan and D,L-[3'-14C]tryptophan (2% of sample used in feeding experiment) was added 30 mg of L-tryptophan and 200 ul of 0.5 N NaOH. The mixture was heated at 75°C until the L-tryptophan had totally dissolved. EtOH (1 OOul) and one drop of AcOH were then added. White crystals formed after the solution was cooled down to room temperature. The solid was filtered off and washed with EtOH (3x100 ul) and ether (3x1 ml ) to give 18.6 mg of tryptophan. The recystallization was repeated four times and the radioactivity of 3H and 14C was checked each time until the 3H/14C ratio was constant. 121

The same procedure was used for the other three mixtures of tryptophan (see Table 3).

4-(1-Hydroxyethyl)quinaIdic Acid (9) Synthesis

Methvl auinaldate (6): To 5 g (29 mmoles) of quinaldic acid dissolved in methanol (50 ml), diazomethane (freshly made from Diazald, Aldrich) was added until the solution turned yellow. After evaporation of the solvent, the residue was purified by column chromatography on silica gel (ethyl acetate / hexane 4:1 as eluant). The yield was 4 g (72.6%). Melting point: 8Q~82°C. 1H-NMR (CDCI3): d 2.87 (s, 3H), 7.65 (t, J=7.0 Hz, 1H), 7.78 (t, J=7.0 Hz, 1H), 7.87 (d, , J=8.0 Hz, 1H), 8.13 (d, J=8.1 Hz, 1H), 8.20 (d, J=7.9 Hz, 1H), 8.27 (d, J=8.0 Hz, 1H).

Methvl 4-Acetvlauinaldate (7):^9 To 3 g (16 mmoles) of 6 and 10 ml (179 mmoles) of acetaldehyde in a mixture of water (15 ml), acetic acid (15 ml), and sulfuric acid (6 ml), were added dropwise simultaneously and separately t-butyl hydroperoxide (30 mmoles) and a solution of ferrous sulphate (8.61 g, 30 mmoles) in water (20 ml) with stirring at room temperature for 2 hrs. After dilution with water, the product was extracted with ether and purified by column chromatography on silica gel (ethyl acetate/ hexane 4 : 1 as eluant). One g (4 mmoles) of product was obtained (27%). Melting point:

108-116°C. 1H-NMR(CDCI3): 8 2.77 (s, 3H), 4.09 (s, 3H), 7.72(ddd, 122

J=1.3, 6.9, 8.0 Hz, 1H), 7.81 (ddd, J=1.4, 6.9, 8.0 Hz, 1H), 8.32 (dd, J= 1.4, 7.9 Hz, 1H), 8.39 (s, 1H), 8.50 (dd, J=1.2, 7.4 Hz, 1H).

Methvl 4-M-Hvdroxvethvnauinaldate (8): To a solution of 1 g of 7 in 50 ml of MeOH, 0.19 g of NaBH4 was added. The solution was stirred for 2 hrs, and 20 ml of water was then added. The MeOH was evaporated, the aqueous solution extracted with ether, and the extract dried with MgS04. After evaporation of the ether, the residue was purified by chromatography on silica gel (hexane/EtOAc 2:1 as eluant) to give 640 mg of product (64%). Melting point: 110~114°C. 1H-NMR (CDCI3): 5 1.63 (d, J=6.6 Hz, 3H), 4.02 (s, 3H), 5.64 (q, J=6.5 Hz, 1H), 7.60 (ddd, J= 1.3, 6.9, 8.4 Hz, 1H), 7.72 (ddd, J=1.4, 7.0, 8.3 Hz, 1H), 8.03 (d, J=8.3Hz, 1H), 8.27 (d , J=10.6 Hz, 1H), 8.27 (s, 1H).

To 640 mg of 8 was added 10 ml of 1N NaOH. The solution was stirred for 24 hrs and then was washed with ether, followed by acidified with 1 N HCI to pH 3. The solution was extracted with n-BuOH and concentrated. Four hundred mg of white solid product was obtained (66%). Melting point: 205°C (decomposition), 1H-NMR (D20): 8 1.65 (d, J=7.0 Hz, 3H), 5.90 (q, J=6.4 Hz, 1H), 7.95 (t, J=6.9 Hz, 1H), 8.15 (t, J=7.0 Hz, 1H), 8.34 (d, J=8.4 Hz, 1H), 8.45 (d, J=8.4 Hz, 1H), 8.54 (s, 1H). FAB-MS: 218 (M+1). 123 4-(1-Hydroxy-[1-3H]ethyl)quinaldic Acid Synthesis

Methvl 4-M-Hvdroxv-ri-3H1ethvnauirialdate (10): A sample of 0.6 mg of 7 was dissolved in 0.5 ml of MeOH and added to 5 mCi of NaB3H4 (1.4 Ci/mmole). The mixture was kept for 0.5 hr at room temperature while the progress of the reaction was checked by TLC (ethyl acetate/hexane 1:1 as solvent, Rf=0.2) followed by radioactivity analysis with a radiochromatogram scanner. The solution was acidified with 0.01 N HCI and stirred overnight. Water (200 ul) was then added and the mixture kept for 3 hrs. The product was isolated and purified by preparative TLC on aluminium sheets of silica gel (solvent system: ethyl acetate/hexane 1:1). The product region was cut out and eluted with MeOH. The total product contained 0.4 mCi of radioactivity.

4-(1-Hvdroxv. f1-3H1auinaldic acid (11): The product from the above reaction was evaporated to dryness and to the residue were added 5 ml of H20 and 200 ul of 0.01 N NaOH. The solution was stirred overnight and then neutralized with 0.01N HCI. After adding an additional 200ul of 0.01 N HCI the solution was stirred overnight. 3H-NMR: 8 5.90 (>98%) 124 Feeding Experiment with 4-(1-Hydroxyl-[3H])ethylquinaldic Acid

The solution of 11 (400 uCi) was fed to five 40 hr old production cultures of S. laurentii. At 78 hrs, the culture were harvested, the thiostrepton was isolated and purified by one precipitation. Four mg of thiostrepton containing 40 uCi of radioactivity was obtained.

Fifty mg of unlabeled thiostrepton was added to 5 uCi of the labeled thiostrepton and the mixture was dissolved in a minimum amount of CHCI3 and precipitated with hexane. This precipitation was repeated once. This precipitate was then redissolved in CH2CI2/EtOH 4:1 and precipitated with ether. This precipitation was repeated once. The supernatants were saved each time for further purification by HPLC (Table 11). A small portion of each sample was injected into the HPLC column and the fractions were collected every two minutes. Thiostrepton eluted from 16 to 18 min.. All fractions were counted and the concentation of thiostrepton was calculated based on a calibration curve obtained with standard reference thiostrepton. The specific radioactivities of thiostrepton measured in the first three supernatants and the last precipitate were: 1.5x10*4 uCi/mg, 1.09x10-4 uCi/mg, 1.16x1(H uCi/mg and 1.6x1(H uCi/mg, respectively. 125

1&02 e x p e r im e n t

Production cultures of S. laurentii were first grown for 30 hrs. Six modified 250-ml Erlenmyer flasks, each containing 50 ml of production culture, were connected to the closed system shown in Figure 48. The oxygen reservoir contained 1 L of 50% 180-enriched oxygen gas. The system utilizes a modified Optima aquarium pump (Rolf Hagen Inc., Montreal, Canada), which had been sealed in glass, to circulate the gas at 4 L/min. The air in the fermentation flasks was kept in contact with the circulating air through a sponge plug. Expired C02 was removed by passing the air stream over 1.5 L of 5 N KOH. Concentrated H2S 04 was employed to remove excess moisture from the air stream. An adequate level of moisture was supplied to the air stream by passing it over water. Slightly positive pressure was kept in the system by a pressure-equalizing buret (2 L) containing 1 N CaS04 solution. The other conditions of the fermentation were unchanged. The average oxygen consumption rate was approximately 20 ml per hour for 300 ml of fermentation medium. The fermentation was stopped at 72 hours and about 10 mg of thiostrepton was isolated.

FERMENTATION WITH DIFFERENT AMOUNTS OF (NH4)2S 0 4 IN PRODUCTION MEDIUM

Four different concentrations of (NH4)2S 0 4 in the production 126

Pump

5NKOH

S tir r e r

/ W ater

1N C uSQ

Sponc a Plug R otary Shaker

Production Culture

Figure 48 . Diagram of the system for fermentation in 180-containing atmosphere 127 medium were tried: 500mg/L, 350mg/L, 225mg/L, Omg/L; All other ingredients and fermentation conditions were unchanged. Each experiment consisted of two 50 ml cultures. To each 50 ml culture was fed 6.25 mg of NH4CI (12.5 mg) in two portions at 34 and 50 hours after inoculation. The fermentation was terminated at 80 hours. Each pair of flasks with the same amount of (NH4)2S 0 4 was combined and 100 ml of CHCI3 was added. The mixture was blended and the CHCI3 phase was separated by centrifugation. After evaporation of the CHCI3, the residue was dissolved in 15 ml of DMSO, and 20 ul of each sample was subjected to analytical HPLC (Table 11) for determination of the thiostrepton concentration.

15NH4CI f e e d i n g

To 18 flasks (900 ml) of production culture was added 500 mg of 15NH4CI (99 atom % 15N) in two portions at 34 and 52 hrs. The fermentation was terminated at 80 hrs. The yield of thiostrepton after precipitation from CHCI3/hexane was 50 mg.

15N-GLYCINE FEEDING

To 18 flasks (900 ml) of production culture was added 500 mg [15N]glycine (99 atom % 15N) in two portions at 34 and 50 hrs. The cultures were harvested at 82 hrs. Thiostrepton was isolated by precipitation from CHCI3/hexane and the yield was 50 mg. 128

FERMENTATION WITH DIFFERENT AMOUNTS OF SERINE

To two 50 ml production cultures was added 0.05, 0.1, 0.15, 0.2, 0.25, 0.3 g of serine, respectively, each in two portions at 32 and 56 hrs, after inoculation. The cultures were harvested at 80 hr. and the two parallel cultures were combined. To each 100 ml culture was added 50 ml of CHCI3 and the mixture was agitated in a blender. The organic phase was collected by centrifugation and 10 ul of it was subjected to analytical HPLC (Table 11) for quantitation of thiostrepton.

FEEDING EXPERIMENT WITH 15NH4CI AND A 3-FOLD EXCESS OF UNLABELED SERINE

Four portions of a mixture of 125 mg of 15NH4CI and 0.375 g of serine (total 500 mg of 15NH4CI and 1.5 g of serine) were added to 25 flasks of production culture at 30, 39, 48, 54 hour of growth. The fermentation was terminated at 90 hours. The yield of thiostrepton w as 20 mg. Table 9. HPLC system for tracing tryptophan synthesis:

Injector: Beckman 21OA Pump: Beckman Model 126 Column: Alltech Econosphere C-18 (10um, 250x4.6mm) Integrator: Spectra-physics SP 4270 Detector: Packard Trace-ll Flow Rate: 1ml/min. Detection: radioactivity Retention Time: serine: 3.5 min. tryptophan: 6.5 min. Gradient Elution Mobile Phase Time A% B% 0 90 10 5 78 10 8 0 100 11 0 100 17 90 10 Eluent: A: 5% AcCN in 10mM NaOAc pH 4.5 B: 80% AcCN in 10 mM NaOAc, pH 4.5 Table 10. HPLC system for purification of tryptophan

Injector: Beckman 210A Pump: Beckman Model 126 Column: Whatman Partisil C-18 (10um, Magnum 500x9.4mm) Integrator: Spectra-physics SP 4270 Detector: Hitachi Anspec Model L 3000 Flow Rate: 4ml/min. Detection: 254nm Retention Time: 11 min. Gradient Elution: Mobile Phase Time A% B% 0 90 10 8 78 22 10 0 100 12 0 100 15 90 10 Eluent: A: 5% MeCN, 0.01% H3P 0 4 B: 90%MeCN, 0.01%H3P 04 Table 11. HPLC system for purification of thiostrepton

Injector: Beckman 210 Pump: Beckman Model 126 Column: Whatman Partisil C-18 (10um, Magnum) Integrator: Spectra-physics SP 4270 Detector: Hitachi Anspec Model L 3000 Flow Rate: 1 ml/min. Detection: 250nm Retention Time: 16.5min Gradient Elution: Mobile Phase Time A% B% 0 60 40 17 0 100 19 0 100 21 60 40 Eluent: A: 5% MeCN, 0.01% H3P 0 4 B: 90% MeCN, 0.01%H3P 0 4 REFERENCES

1. J. F. Pagano, M. J. Weinstein, H. A. Stout and R. Donovick, Antibiotics

Ann. 554 (1955-1956).

2. J. Vandeputte and J. D. Dutcher, Antibiotics Ann. 560 (1955-1956). 3. B. A. Steinberg, W. P. Jambor, and L. O. Suydam, ibid., 562 (1955-1956). 4. M. J. Cron, D. F. Whitehead, I. R. Hooper, B. Heinemann and J. Lein,

Antibiot. Chemother. 6, 63 (1959). 5. W. H. Trejo, L. D. Dean, J. Plussec, E. Meyers and W. E. J. Brown,

J. Antibiotics (Tokyo) 30, 639 (1977). 6. B. Anderson, D. Crowfoot Hodgkin and M. A. Viswamitra,

Nature 225. 233 (1970). 7. K. Tori, K. Tokura, K. Okabe, M. Ebata and H. Otsuka, Tetrahedron

Lett. 3,185 (1976).

8. O. D. Hensens and G. Albers-Schonberg, J. Antibiotics (Tokyo) 36, 814(1983).

9. J. Walker, J. Chem. Soc., Chem. Comm. 706 (1977).

10. B. W. Bycroft and M. S. Gowland. J. Chem. Soc., Chem. Comm. 256 (1978). 11. J. E. Thiemann, C. Coronelli, H. Pagani, G. Bereha, G. Tamoni and V.

Arioli, J. Antibiotics (Tokyo) 21, 525 (1968).

132 133

12. M. S. Puar, A. K. Ganguly, A. Afonso, R. Brambilla, P. Mangiaracina,

O. Sarre and R. D. MacFarlane, J. Am. Chem. Soc. 103, 5231 (1981).

13. J. Shoji, T. Kato, Y. Yoshimura and K. Tori, J. Antibiotics (Tokyo) 34,1126 (1981). 14. C. Pascard, A. Ducruix, J. Lunel and T. Prange, J. Am. Chem. Soc. 99, 6418 (1977).

15. S. Pestka and J. W. Bodley, Antibiotics, Vol. Ill, J. W. Corcoran and F. E. Hahn ed. p. 551, Berlin, Heidelberg, New York: Springer 1975. 16. E. Cundliffe, Mechanism of Action of Antibacterial Agents Vol. I, Springer-Verlag, p. 329,1979.

17. E. Cundliffe and J. Thompson, J. Gen. Microbiol. 126,185 (1981).

18. L Keppens and F. Degrate, Rev. Agric. (Brussels) 32,159. (1979).

19. J. Thompson, F. Schmidt and E. Cundliffe, J. Biol. Chem. 257, 7915(1982).

20. H. Kleinkauf and H. von Dohren, in Peptide Antibiotics.

Biosynthesis and Functions; H. Kleinkauf and H. von Dohren, Eds. Walter de Gruyter; Berlin, 1982, and references therein.

21. C. Kaletta and K.-D. Entian, J. Bacteriol. 171,1597 (1989).

22. S. Banerjee and J. N. Hansen, J. Biol. Chem. 263, 9508 (1988). 23. N. Schnell, K.-D. Entian, U. Schneider, F. Gotz, H. Zahner, R.

Kellner and G. Jung, Nature (London) 333, 276 (1988).

24. C. J. Pearce and K. L. Rinehart, J. Am. Chem. Soc. 101,5069 (1979). 134

25. K. L. Rinehart, D. D. Weller and C. J. Pearce, J. Nat. Prod. 43,1 (1980). 26. D. Houck, Ph.D. Thesis, Ohio State University, 1986. 27. K. Tori, K. Tokura, K. Ckabe, M. Ebata, H. Otsuka and G. Lukacs,

Tetrahedron Lett. 185 (1976).

28. M. Luckner and C. Ritter, Tetrahedron Lett. 741 (1965).

29. O. D. Hensens and G. Albers-Schonberg, J. Abtibiotics 36, 832 (1983). 30. B. W. Bycroft and M. S. Gowland, J. Chem. Soc., Chem. Commun. 456 (1978). 31. M. Tanabe, Spec. Period. Rep., Biosyn. 4, 204 (1975).

32. U. Sequin and A. I. Scott, Science 186,101 (1974).

33. A. Bax, R. Freeman, S. P. Kempsell, J. Am. Chem. Soc. 102,4849 (1980). 34. L. Stryer, Biochemistry (3rd Ed), W. H. Freeman and Company. New York, 1988, PP 588.

35. L. A. Trimble, P. B. Reese and J. C. Vederas, J. Am. Chem. Soc. 107, 2175(1985).

36. E. Leistner, R. N. Gupta and I. D. Spenser, J. Am. Chem. Soc. 95 4040 (1973). 37. D. R. Houck, L. -C. Chen, P. J. Keller, J. M. Beale, H. G. Floss, J.

Am. Chem. Soc. 110, 5800 (1988). 38. C. J. Chang, H. G. Floss, D. J. Hook, J. A. Mabe, P. E. Manni, L. L.

Martin, K. Schroder and T. L. Shieh, J. Antibiotics. 34, 555 (1981). 135

39. T. Caromna, G. Fronza and F. Minisci, J. C. S. Perkin II. 10, 1477 (1972).

40. J. C. Vederas, Can. J. Chem. 60,1637 (1982).

41. C. C. Giembotski, B. A. Eipper, R. E. Mains, J. Biol. Chem. 259,6385 (1984).

42. B. A. Eipper, A. C. Mayers, R. E. Mains, Endocrinology 116, 2497 (1985). 43. A. Meister, Biochemistry of the Amino Acids, Vol. 2,2nd ed., Academic Press, New York, 1965.

44. N. Kowanka and E. Leete, J. Am. Chem. Soc. 84, 4919 (1962).

45. E. Leete and J. N. Wemple, J. Am. Chem. Soc. 91,2698 (1969).

46. A. R. Battersby and E. S. Hall, Chem. Commun. 194 (1970).

47. A. R. Battersby and R. J. Parry, Chem. Commun. 31 (1971). if 48. E. E. van Tamelen and V. B. Haarstad, Tetrahedron Lett. 390 (1961).

49. J. D. Bu' Lock and A. P. Ryles, Chem. Commun. 1404 (1970).

50. G. A. Cordell, Introduction to Alkaloids, Chapter 8.6. Wiley, New York, 1981. 51. H. G. Floss, "Radiotracers in Biosynthetic Studies", Chapter 15 in

E. A. Evans, M. Muramatsu (Ed), Radiotracer Techniques and

Applications, Marcel Dekker, Inc., New York (1977). Vol. 2, p. 689.

52. B. W. Bycroft, Nature (London) 224, 595 (1969). 53. D. G. I. Kingston, M. X. Kolpak, J. W. Lefevre and I.

Borup-Grochtmann, J. Am. Chem. Soc. 105,5106 (1983). 136

54. H. G. Floss, E. Schleicher and R. Potts, J. Biol. Chem. 251, 5478 (1976). 55. R. Cardillo, C. Fuganti, D. Ghiringhelli and P. Grasselli, J. Chem.

Soc., Chem. Commun. 778 (1975).

56. G. W. Kirby and S. Narayanaswami, J. Chem. Soc., Chem. Commun. 322 (1973). 57. R. J. Parry, H. S. P. Rao, and J. Mueller, J. Am. Chem. Soc. 104, 339. (1982). 58. M. E. Gustafson, D. Miller, P. J. Davis, and J. P. Rosazza, C. J. Chang

and H. G. Floss, J. Chem. Soc., Chem Comm. 842 (1977).

59. M. B. Purvis, D. G. I. Kingston, N. Fujii and H. G. Floss, J. Chem. Soc.

Chem. Comm. 302 (1987). 60. M. Bodanszky, J. Fried, J. T. Sheehan, N. J. Williams, J. Alicino,

A. I. Cohen, B. T. Keeler and C. A. Birkhimer, J. Am. Chem. Soc. 86, 2478 (1964). 61. D. F. W. Cross, G. W. Kenner, R. C. Sheppard and C. E. Stehr, J.

Chem. Soc. London 2143 (1963). 62. H. G. Floss; J. M. Beale; D. C. Dosch; T. Frenzel; D. R. Houck; J. J. Lee, Y. Li, T. Kamigauchi, U. Mocek, K. O'Hagan, R. Tsuchiya, R. Woodman,

L.-D. Yuen, Z. Zeng. P. Zhou, W. R. Strohl, Proceedings of the 5th

FAOB Congress, Seoul, (in press).

63. H. Mohr and H. Kleinkauf, Biochim Biophys Acta 526, 375 (1978). 64. Y. Li, D. C. Dosch, R. Woodman, H. G. Floss, and W. R. Strohl,

Developments in Industrial Microbiology, (in press). 137

65. M. Luckner, Abhandl. Deut. Akad. Wiss., Berlin, Kl. Chem. Geol. Biol., No.3, 445 (1966).

66. A. W. Murray and D. L. Williams, Organic Synthesis with isotopes, Interscience; New York, 1958, p. 931,

67. D. G. Ott, Synthesis with Stable Isotopes, Wiley, Interscience.

68. B. D. Andresen, J . Labelled Compounds 14, 589 (1978).

69. G. Sugano and M. Miyoshi, J. Org. Chem. 41, 2342 (1976).

70. D. D. Perrin, W. L. F. Armarego and D. R. Perrin, Purification of

Laboratory Chemicals, 2nd ed. Pergamon Press, Ltd., New York, 1980. 71. M. Bodanszky, J. T. Sheehan, J. Fried, N. J. Williams and C. A. Birkhimer, J. Am. Chem. Soc., 82,4747 (1960). 72. C. N. C. Drey, G. W. Kenner, H. D. Law, R. C. Sheppard, M. Bodenszky,

J. Fried. N. J. Williams and J. T. Sheehan, ibid., 83,3906 (1961).

73. M. Bodanszky, J. Alicino, C. A. Birdhimer, and N. J. Williams, ibid., 84, 2003 (1962). 74. M. Bodanszky, J. Fried, J. T. Sheehan, N. J. Williams, J. Alicino, A.

Cohen, B. T. Keeler and C. A. Birkhimer, ibid., 86, (2478).