The Characterization of cytotoxic metabolites from Fijian marine invertebrates

Marine organisms have been shown to produce a wide variety of biologically active secondary metabolites, often without precedent among known terrestrial natural products. As part of a continuing program to screen Fijian marine invertebrates for novel bioactive compounds, two organisms, the sponge Jaspis johnstoni and the didemnid tunicate Lissoclinum patella, have been investigated. Isolation of the active constituents of these organisms was directed by various bioassays and utilized standard chromatographic techniques. The structures were solved using a variety of spectroscopic methods with the major emphasis placed on high field two-dimensional nuclear magnetic resonance (2D NMR), single crystal x-ray diffraction and tandem mass spectrometry. Several of the structure proofs also relied heavily on analysis of key degradation products. The major active metabolite in the Jaspis sponge was found to be a novel compound of mixed peptide/polyketide biosynthesis called jaspamide. Jaspamide is a potent natural insecticide and antifungal agent with moderate cytotoxicity. The 7-deazapurine nucleosides toyocamycin and 5-(methoxycarbonyl)tubercidin and the corresponding aglycones were also isolated. Previous work has shown L. patella, collected from Palau and Australia, to produce a family of cytotoxic, cyclic peptides all containing a thiazole and usually an oxazoline amino acid. In contrast, the Fijian tunicate contained a series of thiazole-containing macrolides, the patellazoles, and a group of modified, thiazoline-containing cyclic peptides, the patellins. The patellazoles were extremely effective antitumor compounds in the National Cancer Institute's human cell line protocol with mean IC(50)s of 10{-3} to less than 10{-6}mu-g/mL. They also exhibited potent antiviral activity in vitro but were too toxic for in vivo use. The structures were solved primarily using NMR techniques, including a new phase-sensitive 2D INADEQUATE experiment. The three-dimensional structure of patellin 2 was solved by x-ray crystallography and the two-dimensional structures of patellins 1 and 3-5 were assigned primarily by fast atom bombardment tandem mass spectrometry. These peptides are unique in having two modified threonine or serine dimethylallyl ethers and in the tendency for most members of the family to exist in two conformations, the ratio of which is a function of solvent.

THE CHARACTERIZATION OF CYTOTOXIC MET ABOLITES FROM FIJIAN MARl1\TE INVERTEBRATES by T. Mark Zabriskie A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Medicinal Chemistry The University of Utah December 1989 Copyright © T. Mark Zabriskie 1989 All Rights Reserved THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a dissertation submitted by T. Mark Zabriskie This dissenation has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. 25 October, 1989 Chair: Chris M. Ireland 25 October, 1989 Arthur D. Broom 25 October, 1989 C. Dale Poulter // . 2S!October, 1989 1 'i' I \'J~, l~: A 1\ ,t.<!\..., L\" :~:: .' (} t .. c"--___ _ William W. Epstcln THE UNIVERSITY OF UTAH GRADUATE SCHOOL FIN AL READING APPROV AL To the Graduate Council of the University of Utah: I have read the dissertation of T. Mark Zabriskie in its final fonn and have found that (1) its format, citations and bibliographic style are consistent and acceptable; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is ready for submission to The Graduate School. Date Chris M. Ireland Chair, Supervisory Committee Approved for the Major Department ad~~z(:.d~ Art ur D. Broom Chair/Dean Approved for the Graduate Council B. Gale Dick Dean of The Graduate School ABSTRACT Marine organisms have been shown to produce a wide variety of biologically active secondary metabolites, often without precedent among known terrestrial natural products. As part of a continuing program to screen Fijian marine invertebrates for novel bioactive compounds, two organisms, the sponge ]aspis johnston; and the didemnid tunicate Lissoclinum patella, have been investigated. Isolation of the active constituents of these organisms was directed by various bioassays and utilized standard chromatographic techniques. The structures were solved using a variety of spectroscopic methods with the major emphasis placed on high field two-dimensional nuclear magnetic resonance (2D NMR), single crystal x-ray diffraction and tandem mass spectrometry. Several of the structure proofs also relied heavily on analysis of key degradation products. The major active metabolite in the ]aspis sponge was found to be a novel compound of mixed peptide/polyketide biosynthesis called jaspamide. Jaspamide is a potent natural insecticide and antifungal agent with moderate cytotoxicity. The 7-deazapurine nucleosides toyocamycin and 5-(methoxy­carbonyl) tubercidin and the corresponding aglycones were also isolated. Previous work has shown L. patella, collected from Palau and Australia, to produce a family of cytotoxic, cyclic peptides all containing a thiazole and usually an oxazoline amino acid. In contrast, the Fijian tunicate contained a series of thiazole-containing macrolides, the patellazoles, and a group of modified, thiazoline-containing cyclic peptides, the patellins. The patellazoles w~re extremely effective antitumor compounds in the National Cancer Institute's human cell line protocol with mean ICSOs of 10-3 to less than 10-6 J.1g/mL. They also exhibited potent antiviral activity in vitro but were too toxic for in vivo use. The structures were solved primarily using NMR techniques, including a new phase-sensitive 20 INADEQUATE experimen t. The three-dimensional structure of patellin 2 was solved by x-ray crystallography and the two-dimensional structures of patellins 1 and 3-5 were assigned primarily by fast atom bombardment tandem mass spectrometry. These pep tides are unique in having two modified threonine or serine dimethylallyl ethers and in the tendency for most members of the family to exist in two conformations, the ratio of which is a function of solvent. v For Tracy, Matthew and Alex TABLE OF CONTENTS ABSTRACT ....................................................................................................... iv LIST OF FIGURES ............................................................................................ ix LIST OF TABLES .............................................................................................. xii ACKN'OWLEDGMEN'I'S ................................................................................ xiii I. INTRODUCTION ................................................................................... 1 Background ........................................................................................... 1 Li tera ture Review ................................................................................ 5 ll. THE CHEMISTRY OF JASPIS JOHNSTONI ..................................... 33 Introduction .......................................................................................... 33 Sponge Collection; Extraction and Purification of Jaspamide ....................................................... : .................................. 33 Structure Determination of Jaspamide ........................................... 35 Biological Activity Profile of Jaspamide ......................................... 45 Isolation of Modified Nucleosides From J. johnstoni ................. 47 Characterization of the Jaspis Nucleosides .................................... 48 Biological Activity of the Jaspis Nucleosides ................................ .52 m. THE CHEMISTRY OF USSOCLINUM PATELLA ........................... 53 In trod uction .......................................................................................... 53 Tunicate Collection; Extraction and Purification of the Patellazoles ........................................................ .53 Structure Determination of the Patellazoles ................................. 54 Minor Patellazoles ............................................................................... 75 Biological Activity of the Patellazoles ............................................. 77 Isolation and Purification of the Patellins ...................................... 79 General Characteristics of the Patellins ........................................... 79 Structure Determination of the Patellins ....................................... 83 Biological Activity of the Patellins ................................................... 102 IV. EXPERIMENTAL .................................................................................... 103 General Experimental Procedures .................................................... 103 The Chemistry of ]aspis johnstoni ................................................... 104 The Chemistry of Lissoclinum patella ............................................ 109 Appendices A. NMR SPECTRA OF ]ASPIS COMPOUNDS ................................... 124 B. NMR SPECTRA OF TIlE PATELLAZOLES ................................... 131 C. NMR SPECTRA OF TIlE PATELLINS ............................................ 145 D. CID MASS SPECTRA OF TIlE PATELLINS ................................... 155 REFERENCES ................................................................................................... 162 viii UST OF FIGURES Figure 1. Phyletic distribution of marine natural products reported between 1977 and 1985 .............................................................................. 2 2. Distribution of nitrogenous marine natural products ...................... 4 3. Structures of the patellamides and ascidiacyclamide ......................... 18 4. Structures of the lissoclinamides and ulicyclamide ........................... 18 5. Structures of the prelissoclinum peptides ............................................ 20 6. Structures of the didemnins .................................................................... 25 7. Structures of tunichromes An 1-3 and Mm 1 and 2 ........................... 27 8. Expansion of the 200 MHz COSY spectrum of jaspamide ................. 40 9. lH-1H NMR couplings observed in the polypropionate unit of jaspamide ....................................................................................... 40 10. Major fragment ions seen in the FAB mass spectrum of 79 ............ 42 11. Computer-generated perspective drawing of jaspamide acetate ........................................................................................................... 45 12. ElMS fragmentation pathways of 5-(C02Me)tubercidin .................... 50 13. Partial structures of patellazole C constructed from a double quantum filtered, phase-sensitive COSY experiment ....................... 59 14. Structural elements of patellazole C unaccounted for by ~OSY ....................................................................................................... 60 15. INAPT correlations seen in patellazole C ............................................ 68 16. COLCX: correlations seen for patellazole C ........................................... 71 17. Selected traces from the 2D phase-sensitive INADEQUATE spectrum of patellazole C ......................................................................... 73 18. Carbon-carbon connectivities in patellazole C established by 2D phase-sensitive INADEQUATE ........................................................ 74 19. 50 MHz 13C NMR spectra of the olefinic region of patellin 2 ......... 82 20. Computer-generated perspective drawing of patellin 2 .................... 86 21. Partial structures of patellin 1 constructed from lH-1H COSY data ........................................................................................ 90 22. Extended partial structures of patellin 1 ............................................... 91 23. FAB/MS/MS fragmentation scheme for cyclic peptides ................... 94 24. 300 MHz 1 H NMR spectrum of jaspamide ........................................... 125 25. 50 MHz 13C NMR spectrum of jaspamide ............................................ l26 26. 200 MHz 1 H NMR spectrum of toyocamycin ...................................... 127 27. 50 MHz 13C NMR spectrum of toyocamycin ....................................... 128 28. 200 MHz lH NMR spectrum of 5-(methoxycarbonyl) tubercidin ..................................................................................................... 129 29. 50 MHz 13C NMR spectrum of 5-(methoxycarbonyl) tubercidin ..................................................................................................... 130 30. 500 MHz 1 H NMR spectrum of patellazole A ..................................... 132 31. 125 MHz 13C NMR spectrum of patellazole A .................................... 133 32. 500 MHz lH NMR spectrum of patellazole B ...................................... l34 33. 125 MHz 13C NMR spectrum of patellazole B ..................................... 135 34. 500 MHz lH NMR spectrum of patellazole C ...................................... 136 35. 125 MHz 13C NMR spectrum of patellazole C ..................................... 137 36. 500 MHz DQCOSY spectrum of patellazole C ...................................... 138 x 37. 400 MHz lH NMR spectrum of the patellazole D and E mixture ......................................................................................... 139 38. 100 MHz 13C NMR spectrum of the patellazole D and E mixture ......................................................................................... 140 39. 200 MHz lH NMR spectrum of patellazole F ...................................... 141 40. 100 MHz 13C NMR spectrum of patellazole F ..................................... 142 41. 200 MHz 1 H NMR spectrum of patellazole G ...................................... 143 42. 100 MHz 13~ NMR spectrum of patellazole G ..................................... l44 43. 500 MHz lH NMR spectrum of patellin 1 ............................................ 146 44. 125 MHz 13C NMR spectrum of patellin 1 ........................................... 147 45. 500 MHz DQCOSY spectrum of patellin 1 ............................................ 148 46. 500 MHz 1 H NMR spectrum of patellin 2 ............................................ 149 47. 50 MHz 13C NMR spectrum of patellin 2 ............................................. 150 48. 400 MHz lH NMR spectrum of patellin 3 ............................................ 151 49. 100 MHz 13C NMR spectrum of patellin 3 ........................................... 152 50. 400 MHz lH NMR spectrum of patellin 5 ............................................ 153 51. 100 MHz 13C NMR spectrum of patellin 5 ........................................... 154 52. CID spectrum of the m/z 597 ion of patellin 1 .................................... 156 53. CID spectrum of the m/z 597 ion of patellin 2 .................................... 157 54. CID spectrum of the m/z 807 ion of patellin 3 .................................... 158 55. CID spectrum of the m/z 793 ion of patellin 4 .................................... 159 56. CID spectrum of the m/z 813 ion of patellin 5 .................................... 160 57. CID spectrum of the 2N HCI hydrolysis product from patellin 3 ............................................................................................ 161 xi LIST OF TABLES 1. IH and 13C NMR Data (CDCI3) for Jaspamide ..................................... 37 2. Comparison of the 13C NMR Data of L-Abrine to the 2-Bromoabrine of Jaspamide ................................................................... 38 3. IH and 13C NMR Data for Toyocamycin and 5-(C02Me)tubercidin ................................................................................. 49 4. 1 H and 13C NMR Assignments of Patellazole C ................................. 57 5. IH and 13C NMR Assignments of Patellazole A ................................. 61 6. IH and 13C NMR Assignments of Patellazole B ................................. 63 7. Comparison of the Thiazole in Patellazole C with 2-t-Butyl-4-Methylthiazole ............................................................. 67 8. 1 H and 13C NMR Assignments of Patellin 2 ........................................ B4 9. Collision-Induced Decomposition Analysis of the m/z 597 ion from Patellin 2 ......................................................... 96 10. Collision-Induced Decomposition Analysis of the m/z 597 ion from Patellin 1 ......................................................... 97 11. Collision-Induced Decomposition Analysis of the m/z 807 ion from Patellin 3 ......................................................... 98 12. Comparison of the Collision-Induced Decomposition Spectra of Patellins 3 and 4 ....................................................................... 101 13. Collision-Induced Decomposition Analysis of the m/z 813 ion from Patellin 5 ......................................................... 102 ACKNOWLEDGMENTS I would like to express my deepest gratitude to Professor Chris M. Ireland for his guidance and endless support throughout my graduate career. His trust in my decisions and abilities and his candor in all subjects are greatly responsible for my development as a scientist. Many thanks go to Drs. Tadeusz F. Molinski and Darrell R. Davis for their helpful advice, comments and, most importantly, their friendship. I would also like to thank all my past and present colleagues in the Ireland group and at the University of Utah for making my stay memorable. In particular, I would like to thank Dr. Chad Nelson for acquiring most of my mass spectral data and Mark Foster for per­forming the molecular modeling calculations and proofreading this disserta­tion. My sincerest thanks go to my parents, for their love and support and the values they instilled in me. I also thank my in-laws and my grandparents for their help and encouragement during some very trying times. Most of all I thank Tracy for nine years of love, encouragement and friendship, and for two beautiful sons. Even though she endured a tremendous amount of pain and suffering in the last year, her unselfish support of my work never diminished. Finally, I would like to thank the American Foundation for Pharma­ceutical Education and the University of Utah Research Committee for their generous fellowships, and the National Institutes of Health for funding this research. I. INTRODUCTION Background The study of marine natural products, when compared to terrestrial plant and microorganism studies, is a relatively new area and one that has generated a great deal of interest in the last 15 years. As the search for new pharmaceuticals intensifies and the occurrence of novel classes of biologically active compounds from traditional terrestrial sources becomes less frequent, the Earth's oceans are a logical source of new organisms to be investigated. Many of the secondary metabolites isolated from marine organisms are without precedent in terrestrial natural product studies. Considering that the majority of marine metabolites has come from invertebrate animals (e.g., sponges, tunicates and coelenterates), as opposed to higher plants, fungi and prokaryotic microbes, this is not surprising. The early work in the field focused primarily on the obvious targets; the large and abundant organisms such as coelenterates (sea whips, soft corals), sponges and macroalgae (seaweed) collected from easily accessible waters. These three phyla alone (the various algae phyla are grouped here into one division) were responsible for 90% of the more than 1700 marine natural products reported between 1977 and 1985 (Figure 1).1 However, as the ocean's fluora and fauna began to receive increased pressure from the natural products chemist, the resourceful investigator sought or created less obvious niches to explore. These "niches" included unexplored collection sites, including cold waters2,3; less obvious or difficult to obtain organisms such as 46 30 18 • 'Algae' II Sponges II Coelenterates 383 ~ Echinoderms 0 Tunicates • Bryozoans Ei 'Microbes' Figure 1. Phyletic distribution of marine natural products reported between 1977 and 1985. deep water species,4 tunicates,5,6 bryozoans? and cyanobacteria8; and imple­menting screens for different types of biological activity, including insectici­dal, 9 anti-inflammatory,10 immunoregulatory,11 and cardioregulatory actions.12 Rationale 2 The initial focus of Professor Ireland's research group involved study­ing the secondary metabolism of Pacific tunicates; the basis for this was two­fold. First, a high incidence of biological activity had been demonstrated; 5.2% of 425 tunicate extracts tested before 1979 were active in the National Cancer Institute's P388 murine leukemia screen, third only to the Porifera (5.3%, 1702 tested) and Ctenophora (8.3%, 12 tested).13 Second, although many species had been collected and tested, very little work had been done to address the nature of the biologically active principles. 3 As this group and others began uncovering the identities of the active compounds, a clear trend emerged; nearly 90% of the 46 reported tunicate metabolites in Figure 1 contained nitrogen. This was in definite contrast to the highly studied coelenterates and algae which specialize in acetate-derived secondary metabolites and show only a 7.3% occurrence of nitrogenous com­pounds (Figure 2).1 The ideal biological activity-directed isolation scheme involves the definition of a specific type of activity in a crude extract and tracing it with an appropriate assay through the course of numerous extractions and chromato­graphic steps to the pure, active compound(s). Unfortunately, the reality in most academic labs is that the tum-around time for all but the most basic assays is often too long or the assays too costly to strictly follow this approach. Therefore, when a crude extract displays promising activity in an assay not easily or routinely performed in the isolation lab, it becomes necessary to pursue the active principles using criteria other than the assay. These criteria must increase the likelihood of isolating the active constituents during the interim while the necessary testing is performed. The correlation of nitrogen content with a high incidence of biological activity has prompted the search for nitrogenous metabolites to become a gen­eral criteria in this laboratory, including investigations of organsims other than tunicates. While this is strictly an empirical observation and by no means absolute, the evidence from this group suggests that the odds of isolat­ing novel, biologically active compounds are increased when this marker is pursued. An effective illustration of this point is that every compound isolat­ed by the Ireland group which has generated patentable interest from industry 4 'Microbes' • Non- N Bryozoans • N compounds Tunicates Echinoderms Coelenterates Sponges 'Algae' 646 o 200 400 600 800 Number of Compounds 'Microbes' Bryozoans Tunicates 89.1 Echinoderms Coelenterates Sponges 'Algae' o 20 40 60 80 100 % Nitrogenous Compounds Figure 2. Distribution of nitrogenous marine natural products. 5 has contained nitrogen. Other guidelines are also followed to heighten the likelihood of isolat­ing active compounds. For example, crude extracts often exhibit antimicro­bial activity in conjunction with cytotoxic, antiviral or insecticidal activity. By tracking the antimicrobial fractions using simple overnight agar / disk assays a rapid method is available to follow compounds which are often responsible for the less easily detected activity as well. Of course, identifying antimicro­bial activity is desirable in itself. As part of the continuing effort of Professor Ireland's group to charac­terize bioactive metabolites from Fijian marine invertebrates, two organisms have been investigated; the sponge ]aspis johnstoni and the tunicate Lisso­clinum patella. Both studies began with a defined biological activity (cytotoxi­city vs. murine L1210 leukemia) and then focused on the isolation of the nitrogenous metabolites. Of the four classes of compounds found (two from each organism), three possess marked biological activity. These data serve to further validate pursuing nitrogenous metabolites when initial biological activity is identified. Literature Review The contributions of this thesis to the study of sponge secondary meta­bolites include the characterization of the first compound of mixed peptide/ polypropionate biosynthesis from a sponge and the iaentification of two 7- deazapurine nucleosides along with the corresponding aglycones. For the purposes of background and continuity, brief reviews of nucleosides and peptides isolated from sponges are included in this introduction. In the area of tunicate secondary metabolism, the work outlined in this thesis includes the first reported polypropionate macrolides from an ascidian and a new family of modified cyclic peptides. The investigation of tunicates as sources of biologically active compounds has been greatly intensified re­cently and the number of reported compounds has outpaced available reviews. Therefore, a comprehensive handling of both peptides and poly­ketide metabolites of tunicate origin is presented. Peptide Natural Products from Marine Sponges 6 Very recently, a review covering natural product peptides from marine animals and plants was prepared by members of the Ireland research group.14 Since I am a coauthor of that paper, only a brief introduction to the types and activities of peptides isolated from sponges is presented here. Included are representative examples that illustrate the diverse array of amino acid­derived sponge metabolites and the often novel amino acids found in these peptides from the Porifera. A comprehensive and continuing review of marine natural products, including pep tides, is prepared yearly by Faulkner.lS The first antimicrobial peptide to be identified from a sponge was discodermin A (1), isolated from Discodermia kiiensis, it is a tetradecapeptide containing several D-amino acids and two rare t-butylleucine residues, one of L and one of D configuration.16,17 HCQ-O-Ala-L -Phe-L -Pro-X-D-Trp-L-Arg-O-CYS(OaH)-L -Tt-L -MeGln-O-Leu-L -Asn-L -Thr-Sar J (1): -X-= -D-t-Leu-L-t-Leu­( 2): -X- = -o-Val-L-t-Leu­( 3): -X- = -D-t-Leu-L-Val­( 4): -X- = -D-Val-L-Val- Discodermin A inhibited the growth of Bacillus subtiIus, Staphylococ­cus au reus, Psuedomonas aeruginosa and Escherichia coli at levels as low as 7 1 Jlg/mL. There are now four members of the class (2-4) and, in addition to the antimicrobial activity, the discodermins have been reported to inhibit the development of starfish embryos at 5 Jlg/mL.t8 The rare t-Ieu residues had previously only been found occurring naturally in Actinomycete metabolites; since some Discodermia are known to harbor symbiotic algae and bacteria it is possible that the discodermins are microbial products.16 An Okinawan Theonella sp. sponge has produced five novel cyclo­depsipeptides which inhibit the division of fertilized starfish eggs.19 The structure of the major component, theonellapeptolide Id (5), and that of the N-methylated derivative theonellapeptolide Ie (6) have been solved using tradi tional peptide sequencing and degradation methods.20 They are trideca­depsipeptides containing several N-methyl amino acids, three J3-alanines and an N-terminal valine capped with a methoxyacetyl group. Mea H~ l..oN Me UNO a a\.1/ ,~ a Y ~H R H MeN ".(yN"lrN~HN~O a a a a HN X 0 H H \ 0 Me )--~ ) . JlyN~N~:X~ .~~ g g ~e 0 (5): R=H (6): R= Me The same major metabolite has been independently isolated from a Theonella sp. collected at Okinawa Island and named theonellamine B.21 This structure proof relied mainly on 2D NMR techniques and the activity traced in this isolation was the inhibition of Na+ /K+-transporting ATPase (ICso, pig brain Na+ /K+ ATPase, 7 x 10-6 M). Br HO HN ~oO H:X; o~ ~ -.9HrN~NH OH o ~'-\ I N::=:J o NH HN 0 HOO~ HO C H2N 2 ...... OC~ I , NH CONH2 HN~~ HN 0 o (7) Br Theonellamide F (7) is a quite recent example of another unusual pep­tide with potent antifungal and cytotoxic properties isolated from a Japanese Theonella sp. sponge and appears to be unrelated to the above Theonella peptides.22 The complete- structure, including absolute configuration, was 8 9 derived from spectroscopic and degradation studies and found to contain five unusual amino acids, the most striking of which is the 't-L-histidino-D-ala­nine forming a bicyclic bridge. A similar bridge has been found in moroidin, a toxic peptide from the Australian stinging bush LapoTtea mOToides, between C-2 of a tryptophan and N-l of a histidine imidazole.23 The novel 3-amino-4- hydroxy-6-methyl-8-(p-bromophenyl}-5,7-octadienoic acid (Aboa) is very similar to 3-amino-phenyldecanoic acids that have been reported as constitu­ents of cyclic pep tides isolated from blue-green algae and may indicate a sym­biotic origin of theonellamide F.24,2S Bengamides A-F (8-13) and isobengamide E (14) are metabolites of an unidentified Fijian sponge in the ]aspidae family that exhibit cytotoxic, anthelmintic and antimicrobial activity.26 The structures include a novel caprolactam formed by the cyclization of lysine or &-hydroxylysine and were based primarily on analysis of 2D NMR spectra. I I OMe OH (14) The Australian sponge Dysidea herbacea, collected along the Great Barrier Reef near Townsville, contained the tetramic acid derivative dysidin (15) which contains the first example of a naturally occurring trichloromethyl 10 group.27 Another collection of this sponge near Gladstone, Australia con­tained the diketopiperazine 16 which is composed of two N-methyl trichloro­methylleucine residues.28 (15) (16) The presence of halogenated amino acids such as that in 16 is a fre­quent occurrence in marine peptides although most are halogenated aromatic amino acids. For example, every sponge investigated in the order Verongida produces secondary metabolites derived from a brominated tyrosine. An example is bastadin-1 (17) produced by the phenolic oxidative coupling of two oxime-containing bromotyrosine dimers.29 Bastadin-1 was isolated from the Australian sponge lanthella basta and possessed potent in vitro and some in vivo antimicrobial activity against Gram-positive organisms.30 HO H Br~N HO~ Br N H (17) I N Br OH .... OH 11 Other halogenated aromatic amino acids are found in the closely related cyclodepsipeptides jaspamide (18) and geodiamolides A and B (19 and 20). The 19-membered macrocycle jaspamide is the first example of mixed peptide/polyketide biosynthesis reported from a marine sponge and was inde­pendently isolated from collections of Jaspis johnstoni collected from a marine lake in Palau and Suva Harbor, Fiji.9 Jaspamide contains a tripeptide composed of L-alanine, the new amino acid D-2-bromo-a-N-methyltrypto­phan (2-bromoabrine) and a rare ~-tyrosine, plus a 12-carbon polypropionate unit. The carbon skeleton was established by 2D NMR and FABMS and the relative stereochemistry provided by x-ray diffraction analysis of a crystalline tyrosine-acetate derivative. The absolute configuration was provided by acid hydrolysis followed by derivatization and analysis of the alanine by chiral HPLC. Jaspamide exhibited potent insecticidal activity against Heliothis virescens (LCso = 4 ppm) and was extremely effective against C. albicans at 1 Jlg/ disk. Interestingly, 18 was completely inactive against a variety of Gram­positive and Gram-negative bacteria. OH HN o "~O Br H 3 C .... NyO ~NH (18) HO R () ~o I. 0:xN:J)lN ..... Y O o H 0 NH (19): R = I (20): R = Br 12 Jaspamide has subsequently been reported under the name jasplakino-lide but no stereochemistry was assigned.31 This isolation was also from a Fijian sponge and the name was based on an incorrect taxonomic assignment. In this report jaspamide was credited with in vitro anthelmintic activity against the nematode Nippostrongylus braziliensis at less than 1 Ilg/mL and in vitro cytotoxicity of 0.32 Ilg/mL against larynx epithelial carcinoma and 0.01 Ilg/ mL against human embryonic lung cells. Jaspamide (18) displayed in vivo activity comparable to miconazole nitrate when administered topically as a 2% solution to a Candida vaginal infection in mice, but was inactive against a systemic murine C. albicans infection when administered subcuta­neously. 32 J aspamide has also been isolated from aNew Guinea collection of J. johnstoni.33 The geodiamolides were isolated from a Geodia sp. sponge gathered in the West Indies.34 These 18-membered cyclodepsipeptid~s are composed of a tripeptide, consisting of two L-alanines and a D-3-halo-N-methyltyrosine, and the same 12-carbon polypropionate hydroxy acid found in jaspamide. Geodi­amolide A (19) is the 3-iodotyrosine analogue and geodiamolide B (20) con­tains a 3-bromotyrosine. Both of these structures, including the absolute con .. figuration, were solved by x-ray analysis. Like jaspamide, these compounds were active antifungal agents but were devoid of antibacterial activity. An interesting point arises from the uncanny structural similarities of jaspamide and the geodiamolides, in parti­cular the identical L-alanine-tetrapropionate unit: is this an example of a product from similar or identical symbiotic microorganisms or is this a chemical clue of taxonomic importance? While this question is far from having a simple answer, it is worth noting that both Geodia and ]aspis sponges belong to the order Choristida in the class Demospongia; the genus 13 Geodia belongs to the family Geodiidae and the genus Jaspis is a member of the family Jaspidae.35 However, the recent report that the Fijian J. johnstoni also contained the known Streptomyces metabolite toyocamycin would seem to indicate the presence of a microorganism which could also be responsible for the production of the cyclodepsipeptide.36 Because of the impressive biological activity of jaspamide and the structural similarity of the geodiamolides, there has been considerable inter­est in the synthesis of these compounds. Several different groups have now reported syntheses of jaspamide and the geodiamolides.37,38,39,40,41 N ucleosides from Sponges The initial interest in exploring sponge secondary metabolites as a source of new, biologically active compounds is generally attributed to Bergmann and coworkers. Their pioneering work in the early 1950s on the Caribbean sponge Cryptotethya crypta led to the discovery of the arabinonuc­leosides spongouridine (21), spongothymidine (22) and spongosine (23).42,43,44,45 These compounds later served as templates for the design of Ara-A (24) and Ara-C (25), synthetic analogs used clinically as antiviral and antitumor agents.46,47,48,49 Interestingly, in 1984 Ara-A (9-~-D-arabinofuran­osyladenine) was found as a natural product in the Mediterranean gorgonian Eunicella cavolini along with the 3t-O-acetyl derivative.50 There are now a number of novel nucleosides and bases isolated from marine sponges, most of which possess notable biological activity. There have also been several reports of nucleosides containing the rare pyrrolo[2,3-d]pyrimidine base. 14 HN0 :yR :):N :):N N:) N I ~ II ~ oAN I ~ H3CO~N N HO ~ HO~ HOW HOW HO HO HO HO OH HO HO (21): R= H (23) (24) (25) (22): R=CH3 The Australian sponge Tedania digitata was found to contain the nucleoside 1-methylisoguanosine (26) which has a wide array of pharmacol­ogical properties.51,52 This compound elicits cardiovascular, hypothermic and skeletal muscle relaxant effects when administered orally to mice and rats.53 Compound 26 interacts directly with adenosine receptors and appar­ently owes its prolonged activity to deamination resistance, thereby acting as a long lasting adenosine analog.54 The California dorid nudibranch Anisodoris nobilis has also been found to contain 1-methylisoguanosine and presumably accumulates it from a dietary source.55,56 The sponge Hymeniacidon sanguinea, collected in the English Channel, contains the novel purines 1,9-dimethyl-6-imino-8-oxopurine (27) along with the known compound 1-methyl-6-iminopurine (28) (spongo­purine). 57,58 Another new purine, 1,3,7-trimethylguanine (29) was reported from the New Zealand sponge Latrunculia brevis.59 This compound was inactive against the P388 murine leukemia cell line and all antimicrobial and antiviral assays. 15 NH2 NH NH 0 MeN:.):N) O)..N6 N MeN:J:~>=O MeN:J:N) MeN :)cMI Ne) HO~ ~N N ~N N HNAN N Me H Me HO OH (26) (27) (28) (29) Examples of the novel pyrrolo[2,3-d]pyrimidine base (7-deazapurine) include 4-amino-5-bromo-pyrrolo[2,3-d]pyrimidine (30), isolated from a new species of Echinodictyum.60 Pharmacological properties include eNS and cardiovascular activity and potent bronchodilation. Mycalisines A and B (31 and 32) were found in a Mycale sp. sponge collected from the Gulf of Sagami,Japan.61 These highly modified nucleosides were reported to inhibit the division of fertilized starfish eggs but were completely inactive against a wide variety of bacteria. N~ ~NJl/ H (30) (31) (32) A collection of ]aspis johnstoni from the Fiji Islands contained the known compounds toyocamycin (33) and 5-(methoxycarbonyl)tubercidin (34).36 Toyocamycin is a common Streptomyces metabolite and 5-(methoxy- carbonyl)tuberddin was first reported as an intermediate in the synthesis of sangi vam ycin. 62,63 (33) (34) Pep tides from Marine Tunicates The area and number of tunicate peptides has advanced at such a rate in the two years since the preparation of our previously mentioned review that a comprehensive treatment of the subject is now required.14 16 In 1980, Ireland and Scheuer reported that the tunicate Lissoclinum patella, collected near Palau in the Western Caroline Islands, contained ulithiacyclamide (35) and ulicyclamide (36), the first peptides isolated from a tunicate.5 Since then, L. patella has proven to be a rich source of novel marine natural products. There are now 16 reported examples in this family of lipophilic, cyclic pep tides; all have been isolated from tunicates collected either in Palau or Australia and all share the common feature of a thiazole and usually an oxazoline amino acid. These metabolites may be further cat­egorized into one of three structural classes: the cyc100ctapeptide patella­mides and asddiacyclamide (Figure 3); the cycloheptapeptide lissoclinamides and ulicyclamide (Figure 4); and the disulfide-containing ulithiacylamide (35). 17 )- (35) (36) Patellamides A, Band C (37,38 and 39) and lissodinamides 1,2 and 3 (42, 43 and 44) were reported by Ireland and co-workers to be constituents of the Palauan L. patella.64,65 During the course of this work two useful methods in the structure determination of peptides were developed. First, Biskupiak and Ireland reported a general method for determining the abso­lute configuration of the characteristic thiazole amino acids found in all the lissoclinum peptides. This method is based on the reaction of thiazoles with singlet oxygen to form a cydo-adduct, which upon hydrolysis gives an a­amino acid.66 The second method involved using the COSY-45 2D NMR experiment to aid in sequencing small peptides. This experiment utilizes a 450 mixing pulse in place of the standard 900 second pulse in the normal1H-1H COSY experiment. The effect of the 450 pulse is the reduction of passive coupling, thus the transferred magnetization is distributed over fewer spins and can lead to stronger long-range cross peaks because of the increased sensitivi- Compound Rl R2 R3 Ri Patellamide A (37): H D-val D-val L-ile Patellamide B (38): Me D-ala D-phe L-Ieu Patellamide C (39): Me D-ala D-phe L-val Patellamide 0 (40): Me D-ala D-phe L-ile Ascidiacyclamide (41): Me D-val D-val L-ile Figure 3. Structures of the patellamides and ascidiacyclamide. Compound Lissoclinamide 1 (42): Lissoclinamide 2 (43): Lissoclinamide 3 (44): Lissoclinamide 4 (45): Lissoclinamide 5 (46): Lissoclinamide 6 (47): Ulicyclamide (36): x thiazole thiazoline thiazoline thiazoline thiazole thiazoline thiazole L-val D-ile D-ile D-val D-val D-val L-ile Figure 4. Structures of the lissoclinamides and ulicyclamide. D-ile D-ala D-ala L-phe L-phe D-phe D-ala 18 19 ty.67,68 An added advantage of this pulse sequence is reduced strong coupling within the multiplets along the diagonal. Using this technique, Sesin et al. were able to detect five-bond couplings between a ... protons of adjacent amino acids.69 This coupling was determined to be as small as 0.2 Hz and was not detectable in the one dimen­sional spectrum. Detection of these five-bond couplings helped revise the initially proposed structures of patellamides A-C (the original assignments had the thiazoles attached to C-2 of the oxazolines). The revised structures have also been confirmed by synthesis.70,71,72 Three minor components of the Palauan tunicate have also been iso­lated and the structures determined to be prelissoclinamide 2 (48), preuli­cyclamide (49) and prepatellamide B formate (50) (Figure 5).69 These com­pounds lack an oxazoline present in the namesake metabolites and appear to be biosynthetic precursors rather than degradation products. The conditions required to convert lissoclinamide 2 (43) to 48 (5% H2S04/MeOH reflux) are more harsh than extraction conditions and no degradation of 43,36 or 38 was observed upon extended storage.69 This conversion also allowed the absolute stereochemistry of prelissociinamide 2 to be assigned. The absolute configura­tion of the other two minor components was not determined. In 1983, a group from the Suntory Institute in Japan isolated ascidia­cyclamide (41) from an unidentified ascidian collected from Rodda Reef, Queensland, Australia.73 The structure of 41 differs from that of patellamide A only in the substitution of threonine for serine in one of the oxazoline residues and results in a symmetrical molecule. Ulithiacyclamide was also found in this tunicate; therefore it would appear extremely likely that the organism the Japanese group investigated was L. patella. The structure of ascidiacyclamide has been confirmed by synthesis and also by x-ray ana- 20 Prelissodinamide 2 (48) Preulicyclamide (49) Prepatellamide B Formate (SO) Figure 5. Structures of the prelissoclinum peptides. 21 lysis. 74,75 A conformational study of 41 has also appeared comparing the solid-state and solution structures?6 The study found close agreement between the two structures and showed that 41 adopts a saddle-shaped confor­mation that orients all the amide protons to the interior of the ring. Very recently, there have been two published isolations of new and known lissoc1inum peptides from L. patella collected along the Great Barrier Reef, Queensland, Australia. Schmitz and coworkers reported isolating four new peptides, patellamide D (40) and lissoclinamides 4-6 (45, 46 and 47), in addition to the known compounds ascidiacyc1amide (41) and ulithiacyclamide (35).77 The new structures were solved primarily by NMR methods along with acid hydrolysis and chiral GC and HPLC analyses, but the investigation also involved x-ray crystallographic analysis of patellamide D (40). The second investigation into the Australian L. patella was conducted by Hawkins and coworkers and reported finding the same compounds as Schmitz et al. with the exception of lissoclinamide 6.78 The stereochemistry of several amino acids in these pep tides was not determined but is undoubt­edly the same as that reported by Schmitz. Hawkins and his colleagues fur­ther reported that the peptides could also be found in the isolated Prochloron algae cells in similar weight-to-weight ratios as with whole animal extracts, but no speculation was made on the true source of the metabolites. In contrast to the Palauan and Australian L. patella, the same tunicate collected from the Fiji Islands contained a family of extremely cytotoxic, thiazole-containing polypropionates, the patellazoles, in addition to a new class of modified, cyclic pep tides named the patellins. The structure of the major component of the family, patellin 2 (51), has been determined by x-ray crystallography and degradation studies.79 Patellin 2 lacks the characteristic thiazole and oxazoline amino acids of the previous lissoclinum peptides and 22 instead has a thiazoline and two threonines uniquely modified as dimethyl-allyl ethers. Interestingly, 51 wa.s seen by NMR to exist in two conformations, the ratio of which was a function of the dielectric constant of the NMR solvent. It was also seen in the x-ray analysis that the molecule was able to adopt two distinct conformations and that the difference arose from the thiazoline ring residing in either a flat or puckered form. Molecular dynamics calculations further confirmed that two conformations having distinct energy minima were available to patellin 2 (51) but resulted from a cis-trans inversion ·of the proline rather than the inversion of the thiazoline seen in the x-ray structure. (51) Ulithiacyclamide is the most potent cytotoxin of the lissoclinum pep­tides, exhibiting in vitro anticancer activity against the L1210 murine leuke­mia cell line (ICSO at 0.35 Ilg/mL), the CEM cell line (0.01 Ilg/mL) and HeLa cells (0.1 Ilg/mL).64 In vivo testing towards P1534J murine leukemia gave a therapuetic index (T /C) of 188 with 1 mg/kg repetitive doses.80 The patella­mides are approximately an order of magnitude less toxic than ulithiacycla- 23 mide and the lissoclinamides show only borderline activity against L1210 cells (IC50'S > 10 Jlg/mL).64,65 Patellin 2 exhibited no cytotoxicity. The Great Barrier Reef tunicate Lissoclinum bistratum has recently been shown by Hawkins and coworkers to contain two cyclic hexapeptides, bistratamides A and B (52 and 53).81 The structures were solved using 2D NMR methods and have the characteristic thiazole and oxazoline moieties found in the patellamides (Figure 3) and lissoclinamides (Figure 4). (52): X = thiazoline (53): X = thiazole In similar fashion to their previously cited work on L. patella, the authors separated the symbiotic Prochloron from the ascidian and demon­strated that the pep tides were concentrated in the algae. Perhaps most intriguing was their finding that the host tunicate, depleted of its symbiotic Prochloron, contained two unique nitrogenous, polyketide-derived macro­cycles which exhibited potent cytotoxicity. These compounds will be discussed in greater depth in the following section on polyketide-derived metabolites from tunicates. Bistratamides A and B were found to have mild 24 cytotoxic activity against the T24 and :MRC5CVl human cell1ines (ICsos of 50 and 100 Jlg/mL, respectively). The Caribbean didemnid tunicate, Trididemnum solidum, contains an important new class of cyclic depsipeptides, the didemnins. The structures of didemnins A-C (Figure 6, 54-56) were disclosed first and shown to contain the unique depsipeptide link hydroxyisovalerylpropionate (HIP) and were report­ed to have a new allo stereoisomer of statine.6,82 It was shown later, by syn­thesis and x-ray crystallography, that the correct structure involves replacing statine with isostatine.83,84,85,86,87 More impressive than the unique structures of 54-56 is the potent in vitro and in vivo biological activity. Didemnins A and B inhibited Herpes simplex viruses I and II at 1.0 JiM and 0.05 JlM concentrations, Rift Valley fever virus at 1.37 and 0.04 Jlg/mL, Venezuelan equine encephalomyelitis at 0.43 and O.os Jlg/mL and yellow fever virus at 0.4 and O.qs Jlg/mL, respective­ly. Mice infected with Rift Valley fever showed 90% survival when treated with didemnin B at 0.25 mg/kg but some drug related deaths were observed at this dose.88 Didemnin B (55) has also demonstrated in vivo anticancer activi­ty against P388 murine leukemia (T /C 199 at 1.0 mg/kg).82 Didemnin B has also been evaluated against human tumors in a stem cell assay.89 Tumor cells from 8 of 17 patients showed sensitivity to 55 with the median ICSO being 4.2 x 10-3 Jlg/mL. Didemnin B is now in phase II human clinical trials as an anticancer agent. Added to the list of impressive activities associated with didemnin B is its effect as an immunosuppressive agent. In a Simonsen parental-to-Ft graft-versus-host assay, didemnin B showed 71 % inhibition of splenomegally with repetitive doses of 0.3 mg/kg.ll R) OH 0 "~O o o 0yNH ##~O N I H I HN 0000-(. ~~ I OMe O•H ~ • N X= H3C"'Y ~ o 0 OH I y= H3C~'1:; o Z= H N o OJyC\~ Didemnin A (54): Didemnin B (55): Didemnin C (56): Didemnin D (57): Didemnin E (58): Nordidemnin B (59): Figure 6. Structures of the didemnins. Rt = Me; Rt = Me; Rt = Me; Rt = Me; Rt = Me; Rt=H; o 0 R2=H R2=X R2=Y R2 = Z (n = 3) R2 = Z (n = 2) R2=X 25 26 There have been several other reports of new didemnins as well as confirmation of structures proposed primarily on the basis of mass spectral data. The structures of two minor compounds, didemnins D (57) and E (58), have been reported and share the same core structure as didemnin B, with sidechain extensions of three and two L-glutamines, respectively, capped with an L-pyroglutamate residue.84 Nordidemnin B (59) was originally reported by Gloer and the structure based on GC/MS analYSis of hydrolysate products.90 The structure of 59 has been confirmed by synthesis and the NMR spectra completely assigned.91,92 The tunichromes are blood pigments found in at least two species of heavy metal-sequestering tunicates and were originally thought to stabilize these metals at the low pH of a sci dian blood. It now seems that the tunichro­mes and the metal are found in different types of blood cells.93 However, of major interest to the natural products chemist are the modified tripeptide structures of the pigments themselves and the complex isolation required to obtain pure samples of these extremely unstable molecules. Recently three bright yellow pigments, tunichromes B 1-3 (Figure 7, 60- 62; since renamed tunichromes An 1-3) have been isolated from the blood of the vanadium-sequestering, black, Floridian tunicate Ascidia nigra.93,94 The extreme air and water sensitivity of these molecules required an isolation scheme conducted entirely under deoxygenated, dry argon and implemented a prototype centrifugal counter-current chromatograph (CCCC, now called a centrifugal partition chromatograph or CPC).95 Over a 5 year period approx­imately 6000 tunicates were required to work out an isolation scheme and eventually supply 0.5 mg of pure tunichrome An-1 (60). A more recent investigation into the blood pigments of the iron-accu­mulating tunicate Molguia manhattensis has produced tunichromes Mm-l OH o OH An-1 (60): X = OH; Y = OH An-2 (61): X = H; Y = OH An-3 (62): X = H; Y = H ~N:L o NH H HO~N HO~ 0 Mm-1 (63) o H HO~~N I .dfI$ 0 HO Mm-2 (64) OH OH OH OH OH Figure 7. Structures of tunichromes An 1-3 and Mm 1 and 2. 27 28 and Mm-2 (Figure 7, 63 and 64).93 Like tunichromes An 1-3, these are modi-fied tripeptides, but rather than having N-terminal, L-phenylalanine-derived residues, Mm-1 has a glycine and Mm-2 has a L-Ieucine. Polyketide-Derived Macrocycles from Tunicates As discussed earlier in the introduction, tunicates seem to specialize in producing secondary metabolites derived at least in part from amino acids; nearly 90% of the tu~cate metabolites described contain nitrogen. It is inter­esting though, if not surprising, that the only examples of polyketide-derived macrocycles reported from tunicates are also nitrogenous. The first nonpeptidyl macrocycles isolated from a tunicate were the patellazoles, 24-membered, thiazole-containing macrolides that are extremely potent cytotoxins.96,97 Patellazoles A, B and C (65,66 and 67) were reported by Zabriskie and coworkers as constituents in the Fijian tunicate Lissoclinum patella, while Corley et al. isolated patellazole B from L. patella collected near Guam. Both independent structure proofs focused primarily on 2D NMR techniques. Crucial data in solving the structure of patellazole C (67) were obtained from a novel phase-sensitive 20 INADEQUATE experiment and isolation of the thiazole moiety liberated by ozonolysis. Because of limited sample, the NMR studies on patellazole B (66) focused on modern inverse­detection experiments, which nicely complemented many of the carbon­observed NMR experiments in the first report. Patellazoles A-C were reported as having mean ICsos against human tumor cell lines ranging from 10-3 to 10-6 ~g/mL and activity against the fungus Candida albicans.96 The paper reporting patellazole B cited a similar ICSO (3 x 10-6 ~g/mL) versus the KB cellline.97 These impressive values make the patellazoles the most potent cytotoxins yet isolated from a tunicate 29 (as a comparison, didemnin B is reported as having an 1Cso versus the L1210 cell line of 1.1 x 10-3 Jlg/ml).82 (65): Rl = H; R2 = H (66): Rl = H; R2 = OH (67): Rl = OH; R2 = OH Two other 24-membered macrolides, named iejimalides A (68) and B (69), have been found in the Okinawan tunicate Eudistoma cf. rigida.98 These compounds exhibit marked activity in vitro towards the L1210 (ICso = 0.062 and 0.032 Jlg/mL) and L5178Y (ICso = 0.022 and 0.001 Jlg/mL) murine leukemia cell lines. The structures were solved with 2D NMR methods; particularly important were HMBC and phase-sensitive NOESY experiments used to connect partial structures. Amino acid analysis of the hydrolysis pro­duct (6 N HCn produced one equivalent of L-serine, as determined by chiral HPLC. In 1988, Gouiffes et ale disclosed the incomplete structure of a com- 30 pound, referred to as bistramide A, isolated from the ascidian Lissoclinum bistratum collected in New Caledonia.99 They reported a molecular formula of C4oH6SN 20S in addition to a linear structure accounting for all the atoms in the molecule. They were unable, however, to assign three ring-forming ether linkages needed to fulfill the unsaturation requirements. A short time later, Hawkins and coworkers reported the structures of bistratenes A and B (70 and 71), cytotoxic compounds isolated from L. bistratum gathered on the Great Barrier Reef, Australia.81 Bistratene A is undoubtedly identical to bistramide A; the same molecular formula and carbon skeleton are described for both compounds. MeO OMe (68): R = H (69): R = Me N:rO ~yH H 0 . HO Using the ordinary HETCOR experiment with delays optimized for long-range couplings, Hawkins and coworkers reported seeing correlations establishing the required ring connections that the French group could not detect using the COLOC experiment. Bistratene A (70) and B (71) are reported as being cytotoxic towards the 1'24 and MRC5CVl cell lines at 0.07 and 0.09 Ilg/mL, respectively. (70): R = H (71): R = Ac Discussion of Tunicate-Algal Symbiosis 31 The preceding subsections on tunicate metabolites leads to a final point of discussion; that is the origin of the increased percentage of nitrogenous metabolites observed for the tunicates. Tunicates are well-known for existing in obligate and nonobligate syrrlbiosis with unicellular prokaryotic algae of the genus Prochloron.100,lOl The exact nature of the symbiosis has not been delineated but algal-tunicate symbioses are more common in tropical waters where the level of dissolved nutrients is low.102 This tends to suggest that the algae playa role in the host's nutrition. There is a convincing body of evidence showing that the Prochloron photosynthates are transferred to the host where they may help meet the host's carbon requirements.103,l04,lOS In the studies by Kremer et al., Prochloron cells were isolated from Lissoclinum 32 patella and incubated with NaH14C03 for 1 hour. Sonication and solvent partition of the cells resulted in 50% of the water soluble label being found in amino acids and 20% resided in the intermediates of the tricarboxylic acid cycle. 104 A recent report by Parry has shown that the L. patella algal symbiont can assimilate nitrogen from 15N labeled ammonium sulfate and incorporate it into glutamine.106 The ocean is a poor source of ammonia, with an estima­ted concentration of only 0.2 J.1M,107 but as much as 95% of the nonprotein nitrogen excreted by tunicates is in the form of ammonia, which could create an ammonia-rich environment for the Prochloron.108 Although Parry saw no evidence for 15N incorporation in the host tissues, this may have been due to short incubation times, a large host pool of the translocated products and the relative insensitivity of 15N NMR. However, the implications are that the Prochloron not only supplies the tunicate with energy-rich photosynthetic products but may also provide usable nitrogenous compounds produced from the excreted ammonia of the tunicate. The role of the symbiont in the production of secondary metabolites is an issue that is still unclear. The occurrence of such different metabolites as the lissoclinamides / patellamides and the patellazoles from the same organ­ism points to a different origin for each class of compounds. The report by Hawkins et al. showing that in L. bistratum the bistratamides were concentrated in the algal cells and the bistratenes were contained in the tunicate tissue further indicates different origins for the metabolites.81 Unfortunately, our studies on L. patella did not involve separation of the algae and tunicate, so whether or not the patellazoles and patellins are predominantly found in the host or symbiont, respectively, is still a mystery and one that deserves attention. ll. TIIE CHEMISTRY OF JASPIS JOHNSTONI Introduction Sponges of the genus Jaspis have received little attention from marine natural products chemists; prior to this study only two investigations into the secondary metabolites of Jaspis sponges have appeared in the literature. Both investigations were carried out on J. stellifera collected from Fiji and the southern Great Barrier Reef. The Fijian sponge was shown to contain three bright yellow metabolites (72-74) having the rare malabaricane triterpene skeleton.109 The sponge collected off the Australian coast also contained metabolites with this skeleton and were found to be malabaracan-28-oic acids (75-78).110 Our decision to investigate the bright orange, encrusting sponge Jaspis johnstoni was prompted by preliminary cytotoxicity data of the crude meth­anolic extract (L1210 cells; ICSO of 0.28 J.1g/ml). In addition, the crude extract showed good insecticidal activity against the tobacco budworm Heliothis virescens and was very active against the fungus Candida albicans. Sponge Collection: Extraction and Purification of Jaspamide The sponge J. johnstoni was first collected at Suva Harbor, the Fiji Islands in November 1984 and kept frozen until lyopholization. A methanol extract, obtained by soaking 73 g of freeze-dried, homogenized sponge, was successively extracted with hexanes, carbon tetrachloride, chloroform and 1- R (72): R = 0 (73): oR = H, OAe (~) (75): R= H (76): R = Ae (78) 34 butanol. Thin layer chromatography of these fractions (silica gel, EtOAc) showed a single UV active, nitrogenous compound (Rf 0.40) in the CC4 soluble material. Aside from pigments, this compound also comprised the bulk of the CHCl3 fraction with several minor, very polar, UV active and nitrogenous spots evident near the origin. 35 Following the insecticidal and antifungal activity led us to the CC4 (178 mg) and CHCl3 (620 mg) soluble material, which was pooled and filtered through a silica gel 60 column (2.4 x 10 cm, EtOAc) to remove the pigments and minor polar metabolites. Final purification by HPLC (silica gel, EtOAc: Hexanes, 8:2) gave 80 mg of jaspamide (18) as a colorless oil (0.11 % yield). Br OH 31 HN 34 •• ~O .... NyO H3 C ~ 47 NH 1 o (18) Structure Determination of Jaspamide Initial NMR analyses of jaspamide (18) are summarized in Table 1 and provided the first clues to the peptidyl nature of the compound. Seen in the 1 H NMR'spectrum (Appendix A) were characteristic broad, exchangeable sig­nals at 6.63 and 7.65 ppm assigned as amide protons and a singlet at 2.98 ppm 36 assigned as an N-methyl. Highly coupled signals in the 04.5 to 5.9 region were consistent with a-protons, and the doublet methyl signals in the upfield region further pointed to a peptide structure with several hydrophobic amino acids. In the 13C spectrum (Appendix A) four carbonyls were clearly seen in addition to signals in the 45 to 60 ppm region accounting for the a-carbons. Also notable in the NMR spectra of 18 were 16 signals in the aromatic region of the carbon spectrum including a resonance at 155.7 ppm, characteristic of the phenol carbon of a tyrosine unit, and an olefinic methyl singlet at 1.56 ppm and an exchangeable singlet at 08.70 in the 1H spectrum. A 2D 1H-13C heteronuclear chemical shift correlation (HETCOR) experiment at 400 MHz allowed the assignment of all the one-bond correlations given in Table 1.111 Low resolution positive ion FAB mass spectrometry produced a spec­trum with two ions of near equal intensity at m/z 709 and 711 indicating the presence of a single bromine atom in the molecule. A l1igh resolution peak match mass measurement on the m/z 709 ion established a molecular form­ula of CJ6H4sN406Br (found 709.2596 (MH+); C3t#4iN406Br requires 709.2602). The first indication that 18 was a depsipeptide was seen in the IR and 13C NMR spectra. Infrared absorptions at 1715, 1684, 1674 and 1655 cm-1, in conjunction with carbon resonances at 175.1, 174.4, 170.5, 168.9 and 70.8 ppm, indicated an ester and three amide subunits. The first and simplest substructure was found to be alanine. In a 2D homonuclear correlation (COSY) experiment,112 the broadened multiplet at 4.75 ppm showed strong couplings to the exchangeable, broad singlet at 6.63 ppm as well as to the methyl group at 0.70 ppm. The second unit to be solved was that of 2-bromoabrine (2-Bromo-a-N­methyltryptophan). The proton NMR contained characteristic signals and coupling patterns for a 1,2-disubstituted aromatic ring: 0 7.24 (d, J = 7.3 Hz), Table 1 37 IH and 13C NMR Data (CDCh) for Jaspamide C# 13C 8 (EEm)a,b 1 H 8 (EEm) (mult., J (Hz»C 1 H_l H Connectivitiesd 2 40.1 2.50 (m) H-3A, Me-21 3 40.7 A- 2.38 (dd, 15.7, 10.8) H-3B, H-2, H-5 B- 1.89 (d, 15.7) H-3A, H-5 4 131.1 5 127.8 4.75 (d, 7.1) H-6, H-3(A & B) 6 29.2 2.23 (m) H-5, H-7, Me-23 7 43.3 1.32 (m) H-6, H-8 8 70.8 4.62 (m) H-7, Me-24 11 39.7 A- 2.65 (dd, 15.0,4.7) H-IIB, H-12 B - 2.65 (dd, 15.0, 5.5) H-IIA, H-12 12 49.0 5.26 (dd, 8.4,4.7) H-ll (A & B), N-13H H-27, H-31 N-13H 7.65 (d, 8.4) H-12 15 55.5 5.85 (dd, 10.2, 6.5) H-34 (A & B) 18 45.8 4.75 (m) H-19, Me-47 N-19H 6.63 (bs) H-18 21 20.3 1.12 (d, 6.8) H-2 22 18.5 1.56 (s) 23 21.9 0.81 (d, 6.5) H-6 24 19.0 1.05 (d,6.3) H-8 26 133.6 27 127.1 6.94 (d, 8.3) H-28, H-12 28 115.6 6.66 (d, 8.3) H-27 29 155.7 30 115.6 6.66 (d, 8.3) H-27 31 127.1 6.94 (d, 8.3) H-28, H-12 34 23.2 A- 3.38 (dd, 15.2,6.3) H-34B, H-15 B - 3.24 (dd, 15.2, 10.5) H-34A, H-15 N-35H 8.70 (s) 36 109.0 37 111.1 38 131.3 39 118.1 7.24 (d, 7.3) H-40 40 122.3 7.13 (dd, 7.7, 7.3) H-39, H-41 41 120.9 7.10 (dd, 7.7, 7.3) H-40, H-42 42 110.6 7.56 (d, 7.3) H-41, H-35 43 136.1 45 30.8 2.98 (s) 47 17.8 0.70 (d, 6.9) H-18, H-19 a Measured at 100 MHz and referenced to the center line of CDC13 (77.0 ppm). b Unassigned carbonyl resonances: 8175.1, 174.4, 170.5, 168.9. c Measured at 300 MHz and referenced to residual CHC13 (7.24 ppm). d Observed in 2D COSY experiment. 38 7.13 (dd, ] = 7.7, 7.3 Hz), 7.10 (dd, ] = 7.7,7.3 Hz) and 7.56 (bd, ] = 7.3 Hz). Because the broad doublet at 7.56 ppm showed coupling in the COSY experi­ment to the exchangeable singlet at 8.70 ppm and the 13C NMR spectrum con­tained shifts at a 111.1, 118.1 and 136.1 consistent with C-3, C-3a and C-7a, respectively, of an indole, a 2-substituted tryptophan was strongly supported. The N-methyl in the molecule was assigned to this residue based on the lack of a-proton to amide proton coupling. The 13C spectrum of commercially available L-a-N-methyltryptophan is compared to that of jaspamide in Table 2. All of the signals correlate closely with the exception of C-2. The chemical shift of a 109.0 in 18 is 10 ppm upfield from that of abrine and is consistent with a strongly shielding substituent such as bromine in that position. Fur­ther support for the placement of the bromine at this position came from the FABMS data. An ion at m/z 631 was the highest mass fragment seen in the spectrum that did not have the isotope pattern for bromine. This accounts for Table 2 Comparison of the 13C NMR Data of L-Abrine to the 2-Bromoabrine of Jaspamide Carbon # L-Abrine 2-Bromoabrine C-2 119.0 109.0 C-3 111.3 110.1 C-3a 127.6 131.3 C-4 119.2 118.1 C-5 124.3 122.3 C-6 121.8 120.8 C-7 112.0 110.6 C-7a 136.4 136.1 39 a loss of 78 and 80 mass units, respectively, from the parent ions; one mass unit less than required for bromine and was initially very perplexing. How­ever, a report in the literature was brought to our attention showing that halogenated nuc1eosides undergo extensive dehalogenation when ionized by FAB in a glycerol matrix.113 The halogen presumably undergoes a radical exchange with a hydrogen from the glycerol matrix. This reaction is not specific for nucleosides; thyroxine, a polyiodinated tyrosine derivative, was also observed to undergo halogen/hydrogen exchange of one and two iodine atoms. This knowledge gave additional support for the attachment of the bromine in jaspamide to an aromatic site. The remaining amino acid, ~-tyrosine, was assigned over the more common and isomeric tyrosine based on an observed coupling in the COSY experiment between the ortho protons of the aromatic ring at 6.94 ppm and the me thine proton at 5.26 ppm (Figure 8). This correlation is most consistent with the four-bond benzylic relationship in ~-tyrosine rather than a five-bond homobenzylic relationship required in tyrosine. The rare ~-tyrosine residue has been reported before to occur in the edeine peptides.114 The final partial structure of jaspamide is a 12-carbon hydroxy acid con­taining four methyl groups on alternating carbons, characteristic of propion­ate biosynthesis. A combination of single frequency irradiation and 2D COSY NMR experiments enabled the construction of this unit as shown in Figure 9. The diastereotopic protons on C-3 both showed allylic coupling to the olefinic proton of C-5 allowing connections to be made across the double bond. Inter­estingly, only the downfield C-3 proton showed coupling to H-2, indicating a certain amount of rigidity exists in the system. The above four subunits of jaspamide account for 15 of the 16 degrees of unsaturation defined by the molecular formula. A cyclic structure was i 8.0 i' ,t i '.0 6.11 . 5.2 4. .. p" Figure 8. Expansion of the 200 MHz COSY spectrum of jaspamide. R o Figure 9. IH-IH NMR couplings observed in the polypropionate uni t of jaspamide. 40 41 proposed to account for the additional unsaturation required by the molecu-lar formula as well as the hydrophobicity of the compound. This was further supported by the failure of 18 to react with ninhydrin. At this point it came to our attention that Dr. Tadeusz F. Molinski and Mr. Andrew Marcus, working in the laboratory of Professor D. J. Faulkner at the Scripps Institution of Oceanography in La Jolla, California had also isolated a small amount of a compound from a sponge collected at a marine lake in Palau which was identical to jaspamide. Because we had a much larger sample it was decided that we would continue to pursue the spectral characterization of the compound while the Scripps collaborators attempted to obtain a crystalline sample. The next task was to assemble the proper sequence of the four partial structures. The method of choice for determining the sequence of modified cyclic and linear peptides is fast atom bombardment (FAB) mass spectromet­ry. 1 15,l16,l17 When dealing with small amounts of medium sized peptides, NMR sequencing techniques suffer from poor sensitivity and complications from weak nuclear Overhauser effects (NOE) in this molecular weight range. The classical peptide sequencing method of Edman degradation is often of limited utility in marine peptide studies since many of these compounds contain highly modified or unusual amino acids for which chromatographic standards are not available. A more in depth discussion on the use of FABMS in the structure determination of cyclic peptides is presented later in the discussion on the patellins. Unless an ester linkage (depsipeptide) occurs or one amino acid a­nitrogen is significantly more basic than the others, cyclic peptides can present a complex FABMS sequencing problem. This is because protonation to form the molecular ion and subsequent ring opening can occur at each amino acid. ES LIBRAR\ 42 Thus, the sequence determination of jaspamide was greatly facilitated by the depsipeptide linkage. Saponification of 18 with methanolic potassium hyd­roxide, followed by treatment with diazomethane, resulted in the linear methyl ester 79 which was analyzed by FABMS for sequence information. Key fragment ions were peak matched to further confirm ion composition. Figure 10 shows the linear derivative 79 and assignments of the key fragment ions. The cleavages are denoted using the widely accepted peptide ion nomenclature proposed by Roepstorff and Fohlman.118 The cleavage of the tertiary amide bond between the alanine and 2-bromoabrine resulted in the most intense fragments seen in the mass spectrum. High resolution mea­surements on the B2 and Y2" ions corresponded to fragments with molecular formulas of C1sH2~03 and C22H2SN304Br and account for the N-terminal loss of the hydroxy acid and alanine and the loss of 2-bromoabrine and methyl ~-tyrosine from the C-terminus, respectively. Another intense ion at m/z 546, B3, was also peak matched and corresponded to C27H37N304Br HO 54S* Yt " 474 196 O Y 2" B Y It H I 1 ,NJ. ~ .N~OCH3 • N n H ~ n o : I 0 I 0 B2 268 (79) B3 0,1 546 OH Figure 10. Major fragment ions seen in the FAB mass spectrum of 79. resulting from the loss of methyl (3-tyrosine, placing it as the C-terminal residue and serving to finalize the two-dimensional structure of jaspamide. Stereochemical Analysis of Jaspamide by Chiral HPLC 43 With the carbon skeleton of jaspamide defined, the task remained to define the stereochemistry of the six chiral centers in the molecule. Since our attempts at crystallization had failed to date, efforts were directed to the acid hydrolysis, derivatization and analysis of the three amino acids. Several com­plications made this less than a straightforward task. First, tryptophan does not survive conventional 6 N HCI hydrolysis due to oxidation, so a method using the nonoxidizing 4 N methanesulfonic acid and 0.2% 3-(2-aminoeth­yl) indole as a catalyst had to be employed.119 Second, the past method of choice in these laboratories for determining the absolute configuration of amino acids had been chiral gas chromatography. However, tryptophan and (3-tyrosine, even when derivatized as trifluoroacetamide methyl esters, are not volatile at temperatures compatible with the chiral liquid stationary phase of these columns. Therefore, chiral HPLC methods were investigated to circumvent this problem. A number of systems using various chiral mobile phases with a CIS stationary phase and different types of derivatives were initially tried but all failed to perform to our satisfaction. The method ultimately chosen involved analysis of the amino acids on a Pirkle-type column employing (R)-N-(3,5-dinitrobenzoyl)phenylglycine as the covalently bound chiral stationary phase (CSP). Pirkle columns separate optical isomers on the basis that three inter­actions with a chiral surface will serve to differentiate enantiomers if one of these interactions involves the chiral center.120,12I The interactions involved in forming and differentiating the CSP-solute complex with the Pirkle 44 columns are x-x stacking, dipole-dipole stacking, hydrogen bonding and steric hinderance. Once CSP-solute recognition and complexation occurs, chiral discrimination between enantiomers is usually caused by a steric interaction which weakens, or prevents, formation of one of the enantiomer-CSP com­plexes, thus causing that isomer to elute first.122 Initial trial separation studies on commercially available Land D,L­tryptophan, Land D,L tyrosine, and Land D,L-alanine derivatized as their N- 3,5-dinitrobenzoyl amides resulted in little to no enantiomeric resolution. Similar efforts using the same amino acids as their N-dansyl derivatives further treated with diazomethane to form the methyl esters, produced acceptable results for the separation of D,L alanine but not for D,L tyrosine or tryptophan. Although not all of the amino acid configurations could be unambiguously assigned, the decision was made to hydrolyze a sample of jaspamide to set the absolute configuration of the alanine and to try and isolate the other pieces. The hydrolysate of 18 from 4 N MeS04H, 0.2% 3-(2-aminoethyl)indole as catalyst, was derivatized with dansyl chloride and diazomethane and ana­lyzed by chiral HPLC on the Pirkle column (75:25 Hexanes:EtOAc). Analysis showed jaspamide contained L-alanine but failed to yield significant amounts of the other expected products. X-ray Crystal Diffraction Studies on Jaspamide Acetate At this point the Scripps group was successful in growing a crystal of a ~-tyrosine acetate derivative suitable for x-ray diffraction studies. The x-ray solution was provided by Professor Jon Clardy and Mr. Chang-fu Xu at Cornell University, Ithaca; New York and served to confirm the proposed sequence and define the relative stereochemistry including the trans-4,5 45 double bond. Since the alanine had been found to have the 5 configuration by chiral HPLC, the absolute configuration of the entire molecule is 25, 6R, 85, 12R, 15R, 185 as shown in 18. A computer-generated perspective drawing of the x-ray model of jaspamide acetate is shown in Figure 11. Biological Activity Profile of Jaspamide Biological testing of pure jaspamide revealed it to be an extremely effec­tive antifungal and insecticidal metabolite. A 7.6 mm sterile disk impreg­nated with 1 Jlg of jaspamide resulted in an 11 mm zone of inhibition against Figure 11. Computer-generated perspective drawing of jaspamide acetate. Hydrogens are omitted for clarity. The acetate at 032 is extensively disordered and not shown. 46 Candida albicans. This makes jaspamide one of the most potent anti-Candida substances encountered in this program, yet interestingly, it was completely inactive against a variety of Gram-positive and Gram-negative bacteria including Bacillus subtilus, Escherichia coli, Psuedomonas aeruginosa and Staphylococcus aureus. Jaspamide is the first compound from our investigation into Fijian marine invertebrate chemistry which has shown good insecticidal activity. Since the H. virescens' assay is a recent addition to our testing regime a brief description is warranted. The cotton or tobacco budworm, Heliothis virescens, is an important commercial pest, feeding on cotton, tobacco, lettuce and most garden vege­tables. The assay tests for feeding inhibition by mixing a known amount of a compound into an agar and cellulose based food pellet which is then offered to the insect as the only food source. The amount of added compound is varied to find the effective concentration for 50% growth inhibition (ECso) and the lethal concentration to 50% of the organisms (LC50)' Jaspamide (18) was found to have an ECSO of 1.5 ppm and an LCso of 4 ppm, which is com­parable to a number of commercial pesticides. Azadirachtin, considered to be the prototypical natural product insecticide, exhibits an LCSO of 1 ppm in this assay. 123 The insecticidal and anti-Candida properties of jaspamide were impres­sive enough to warrant further testing. Preliminary in vitro tests were very encouraging and recollection of 1.4 kg of J. johnstoni in November, 1986 provided nearly one gram of jaspamide for in vivo analysis. Unfortunately, the animal studies showed the toxicity of 18 to be too high to justify further interest as an insecticide or antifungal agent. The linear methyl ester of jaspamide (79), prepared for the mass spec- 47 tral sequencing studies, was evaluated for insecticidal activity but was comp-letely inactive. Debromojaspamide was prepared by treatment of jaspamide with Raney nickel and found to be equipotent against H. virescens. Isolation of Modified Nucleosides From 1. johnstoni With the new, larger 1986 collection of J. johnstoni the minor, UV active, polar compounds detected in the 1984 collection could now be isolated in sufficient quantity to allow their structures to be solved. The methanolic extract from 300 g of lyophilized and homogenized sponges was successively extracted with hexanes, carbon tetrachloride and chloroform, in a modified Kupchan solvent partitioning scheme, to yield 1.23 g of the CHCl3 soluble material containing the polar Jaspis metabolites. The sought-after material was separated from jaspamide by passing the CHCl3 fraction over a C 18 reverse phase scrub column in 80% aqueous methanol. This crude mixture showed good antifungal activity against C. albicans and justified further pursuit of the responsible compound(s). Final isolation of the individual components was achieved using reverse phase HPLC (1:1 MeOH:H20) and produced 24.7 mg of toyocamycin (33), 43.0 mg of 5-(methoxycarbonyl)­tubercidin (34),3.5 mg of toyocamycin aglycone (80) and 8.6 mg of 5-(methoxy­carbonyl) tubercidin aglycone (81) all as white solids. NH2 eN N~' ~)ll R (33): R = Ribose (80): R = H (34): R = Ribose (81): R = H 48 Characterization of the Jaspis Nucleosides The first indication of the nucleoside nature of 33 and 34 was found in the 1 H NMR spectra, which are summarized in Table 3; the spectra can be found in Appendix A. The two spectra were remarkably similar with the single exception of an O-methyl signal seen at 3.66 ppm in the spectrum of 34. Otherwise, both spectra contained two aromatic singlets and six signals with the proper coupling patterns and chemical shifts of a ribose moiety. Because the IH NMR spectra of compounds 80 and 81 lacked the sugar resonances of the two major polar metabolites, this suggested that they were the aglycones of 33 and 34. Also, the presence of the O-methyl singlet in the spectrum of aglycone 81 was evidence that this modification occurred on the nucleobase of 34 and not on the ribose. The electron impact mass spectrum (ElMS) of 5-(methoxycarbonyl)­tubercidin showed a parent ion at m/z 324 which was futther substantiated by an MH+ ion (m/z 325) seen by liquid chromatography thermospray mass spectrometry (LCMS). Characteristic fragment ions in the E1 mass spectrum at m/z 192 (base + I, 100%), 193 (base + 2), 205 (base + 14), 221 (base + 30) and 235 (base + 44) unequivocally confirmed that 34 was a ribosyl nucleoside. These informative ions also established a molecular weight of 191 for the base and provided further evidence for placing the O-methyl group on the base portion of the molecule rather than the sugar.124 A diagram of the decompositions leading to these characteristic fragment ions is shown in Figure 12. Although the IH NMR and mass spectral data of 34 were very inform­ative, the crucial pieces of data used in solving this structure were found in the 13C NMR and IR spectra. The 13 signals in the 13C NMR spectrum of 5- (methoxycarbonyl)tubercidin (Table 3 and Appendix A), together with the mass spectral data, allowed for a molecular formula of C13Hl~406 and Table 3 IH and 13C NMR Data for Toyocamycin and 5-(C02Me)tubercidin 50 MHz 13C NMR (DMSO-D6)a #C Toyocamycin (33) 5-(C02Me)tubercidin (34) 2 4 4a 5 6 7a 11 21 31 41 51 CN CO OCH3 153.6 157.0 101.3 83.0 132.4 150.2 87.8 74.3 70.2 85.5 61.2 115.4 153.2 157.7 100.7 106.2 130.0 151.2 87.6 74.2 70.4 85.5 61.3 165.4 52.0 200 MHz lH NMR (m, ] (Hz» (Pyridine-Ds)b #H Toyocamycin (33) 5-(C02Me)tubercidin (34) 2 8.71 (s, IH) 6 8.60 (s, IH) l' 6.81 (d,4.8, IH) 21 5.18 (dd, 4.8, 4.8,lH) 3' 5.02 (dd, 4.8, 4.4, IH) 4' 4.72 (m, IH) 51 4.31 (dd, 12.1, 2.4, IH) 5 I 4.20 (bd, 12.1, IH) 0CH3 8.62 (s, IH) 8.60 (s, IH) 6.94 (d, 5.3, IH) 5.31 (dd, 5.3, 5.1, IH) 5.28 (dd, 6.2, 5.1, IH) 4.74 (m, IH) 4.30 (dd, 12.2,2.6, IH) 4.18 (dd, 12.2,2.9, IH) 3.66 (s,3H) aReferenced to internal DMSO-D6 (39.5 ppm). bReferenced to center signal of residual pyridine-D4H (7.55 ppm). 49 m/z324(~) BH+ --Ia. ..... CI H II o m/z 221 (B+30) tp HO} BH+ BH+ H O~ H --I....... CI H2 -CHO. . BI H+ .r I ·CHHC~O" CHO HC 2 m/ z 235 (B+44) II o m/z 205 (B+14) Figure 12. EIMS fragmentation pathways of S-(C02Me)tubercidin. so provided the first evidence that this nucleoside contained a modified purine base. The signal in the 13C NMR spectrum at 16S.4 ppm together with strong absorptions in the IR spectrum at 1701 and 1239 cm·1 and the O-methyl group indicated the presence of an aromatic methyl ester. Furthermore, the upfield sp2 signals at 0 100.7 and 106.2 correspond more closely to C-4a and C-S of a pyrrolo[2,3-d]pyrimidine ring system, commonly called a 7-deazapurine, than to a purine.125 With these data in hand the final structure was determined to be the known compound S-(methoxycarbonyl)tubercidin (34) by comparison to reported spectroscopic data.126 This compound had been prepared as an intermediate in syntheses of other pyrrolo[2,3-d]pyrimidines but this is the first report of its occurrence as a natural product.63,126 51 A molecular weight of 291 for toyocamycin (33) was indicated by parent ions in the EI and LC thermos pray mass spectra of m/z 291 and 292, respec­tively. The ElMS analysis of 33 proved to be nearly as informative as that of (34), showing the same series of characteristic fragment ions corresponding to base + 1 (m/z 159, 100%), base + 2 (m/z 160), base + 14 (m/z 172), base + 30 (m/z 188) and base + 44 (m/z 202).124 Similar to 5-(methoxycarbonyl)tubercidin, the crucial elements used in solving the structure of 33 were found in the IR and carbon NMR spectra. The 13C NMR spectrum contained 12 signals (Table 3) and together with a molecular weight of 291 match a molecular formula of C12H13NS04. The seven non-ribosyl carbons in the 13C spectrum again pointed to a substituted 7-deazapurine base. A strong absorption at 2230 cm-1 in the IR spectrum, together with the 115.4 ppm quaternary carbon seen in the 13C NMR spectrum, suggested the nucleoside base was substituted with a nitrile. The 13C spectrum also contained a resonance for a quaternary carbon at 83.0 ppm; the extreme upfield chemical shift indicated it must be substituted with the strongly shielding nitrile. Comparison of spectral data to published values confirmed the structure of 33 as the known Streptomyces metabolite toyocamycin.62,125 Structure assignment of the aglycones 80 and 81 was based on compar­ison of NMR and ElMS data with the respective nudeosides. Isolation condi­tions were never severe enough to hydrolyze the nucleosides and therefore these are naturally occurring compounds. 52 Biological Activity of the Jaspis Nuc1eosides Routine anti-Candida testing of toyocamycin (33) and 5-(methoxycarb­onyl) tubercidin (34) showed 33 to produce a 13 mm zone of partial inhibition at 1.25 Jlg/ disk while 34 exhibited no activity. Testing for anti-leukemia activ­ity versus the L1210 cell line gave ID50'S of 2.6 x 10-3 and 2.7 x 10-2 Jlg/mL for 33 and 34, respectively. In the initial synthesis of 5-(methoxycarbonyl)tuber­cidin, Rao reported that it extended the lives of mice inflicted with L1210 leukemia by as much as 37%.63 Because 33 and 34 are known compounds no further detailed testing was undertaken in our lab. The aglycones were not tested because of insufficient material. m. THE CHE:MlSTRY OF USSOCUNUM PATELLA Introduction The ascidian (a.k.a tunicate, sea squirt) Lissoclinum patella is a colonial invertebrate organism belonging to the Didemnidae family (Phylum Chor­data, Subphylum Tunicata, Class Ascidiacea). The Didemnidae are partially classified by the occurrence of symbiotic, unicellular algae.100,lOl In L. patella this symbiont is easily seen as a dark green mottling below the transparent outer mantle of the tunicate. Algae of the genus Prochloron are reported to be the only symbionts of L. patella127; however, the algae from the collection of animals investigated here were not classified. Our interest in this tunicate collected near Dravuni Island, Fiji was prompted in part by the observation that the peptide secondary metabolite composition and relative abundances in L. patella appear to be highly geo­graphically dependent.128 Additionally, the crude extract displayed impres­sive cytotoxicity (L1210, ICSO of 0.38 Jlg/mL) and had good antiviral activity (ICSO of 2.63 Jlg/mL) in preliminary screens against vesicular stomatitis virus. Tunicate Collection; Extraction and Purification of the Patellazoles The tunicate was collected from depths of -1 to -3 meters on Astrolobe Reef near Dravuni Island, Fiji in November 1984 and kept frozen until used. Two hundred-twenty grams of lyopholized animals were homogenized in methanol and the filtrate successively extracted with hexanes, carbon tetra- 54 chloride and chloroform. Cytotoxicity assays (L1210) showed the initial acti­vity of the crude extract was primarily concentrated in the CC4 fraction (2.42 g) with some activity still in the CHCl3 soluble material (1.24 g). The CC4 fraction was passed over a silica gel gravity column (2.8 x 50 em, in two parts; EtOAc/MeOH gradient). The early eluting fractions contained three nitro­genous, UV active compounds which were separated by successive reverse phase and silica gel HPLC; yielding 96.5 mg of patellazole A (65), 144.3 mg of patellazole B (66) and 312 mg of patellazole C (67), all as colorless glasses. 4 3~3O4 0 4 0 R2 3 (65): Rl = H; R2 = H (66): Rl = H; R2 = OH (67): Rl = OH; R2 = OH 4S Structure Determination of the Patellazoles The majority of the information used in solving the structures of the patellazoles was obtained on the major component, patellazole C (67). This is 55 particularly true for many of the time consuming 2D NMR experiments such as COLOC and INADEQUATE. However, the full NMR assignments and partial structures of the less abundant patellazoles A (65) and B (66) were also deduced independently. The NMR data for 65 and 66 are discussed below only when describing the differences from 67. Tables with complete spectral assignments for patellazoles A and B are included in this section and the lH and 13C spectra can be seen in Appendix B. Similarly, much of the physical data for 65 and 66 are nearly identical to that of 67 and are not described in detail here but can be found in the experimental section. Physical Data of Patellazoles A,B and C The infrared spectrum of patellazole C (67) had a strong absorption centered at 1728 and another at 1708 cm-1 attributed to an ester(s) and a ketone, respectively. Broad absorptions from 3460 to 3474 cm-1 indicated several hydroxyls in the molecule. The UV spectrum of 67 contained a single absorption maximum at 241 nm (MeOH, e 26 000) and the molecule was optically active; [a]D -1000 (c 1.06, CH2CL2). Low resolution, positive ion FAB MS of 67 produced an MH+ ion at m/z 920, indicating an odd number of nitro gens in the compound. A high resolution peak match measurement on the MH+ ion established a molecular formula of C49H77N013S (found 920.5179; requires 920.5197). The molecular formula was further confirmed by elemental analysis and requires 12 degrees of un saturation in the molecule. A deuterium-exchanged FAB MS experi­ment and computer analysis of the resulting isotope pattern established a total of six active hydrogens in the molecule.129,130 This was also supported by deuterium-exchanged lH NMR experiments on 67. 56 FAB MS of patellazoles A and B produced MH+ ions at m/z 888 and 904 and high resolution measurements gave molecular formulas of C49H77N011S and C49H77N012S, respectively. The increasing number of oxygens from 65 to 67 in conjunction with increasingly polar chromatographic properties sug­gested patellazoles A,B and C differed only by the number of hydroxyls they contained. The initial proton NMR analysis of 67 (Appendix B) provided the first clue to the polypropionate nature of the compound; there were 12 methyl signals upfield of 2 ppm, seven of which were doublets. Also notable in the 1H spectrum were an O-methyl at 3.25 ppm, six olefinic resonances, and a crowded region from 3 to 4.5 ppm indicated a high number of protons on carbons bearing oxygen. The 125 MHz carbon NMR spectrum (Appendix B) resolved al149 signals; the downfield region included a ketone carbon at 0 216.4, three sp2 carbons attached to heteroatoms at 0 175.4, 174.8 and 171.4 and 10 signals from 154.1 to 114.2 ppm accounted for five double bonds, one of which was extremely polarized. Furthermore, these sp2 carbon signals account for 9 of the 12 unsaturations in the molecule, leaving three rings to be assigned. Notable upfield signals were 11 carbons from 87.0 to 60.0 ppm, providing further evidence of the high number of oxygens in the molecule. A series of DEPT experiments established the presence of 71 carbon-bound protons (13 methyls, six methylenes and 20 methines) and agrees with the molecular formula and active hydrogen count from the mass spectral data.131 Partial Structures from NMR Correlation Spectroscopy All one-bond 1H-13C correlations were established using the HETCOR experiment and are presented in Table 4.111 The first partial structures were 57 Table 4 IH and 13C NMR Assignments of Patellazole C C# 13(: (ppm) (mult)a,b 1 H (ppm) (mult,! (Hz»C 1 171.36 (s) 2 81.25 (s) 3 32.65 (t) A- 1.90 (bdd, 13.6,4.4) B- 2.62 (dd, 13.6, 4.4) 4 32.14 (t) A- 1.06 (m) B- 1.44 (m) 5 28.17 (d) 1.85 (bm) 6 44.32 (t) A- 1.30 (ddd, 11.3,11.3,2.3) B- 1.56 (dd, 11.3,5.7) 7 72.56 (d) 3.87 (bm) 8 50.05 (d) 3.13 (dq, 9.5, 6.8) 9 216.43 (s) 10 56.28 (d) 4.28 (ddd, 10.8, 7.6, 4.0) 11 124.85 (d) 522 (dd, 10.8, 10.8) 12 134.39 (d) 5.93 (dt, 10.8, 5.7) 13 32.37 (0 A- 153 (m) B- 3.48 (dd, 12.1, 12.1) 14 38.86 (d) 1.95 (m) 15 74.32 (d) 3.68 (bd, 9.7) 16 69.44 (d) 3.90 (bd, 8.6) 17 87.01 (d) 4.09 (d, 8.6) 18 132.15 (s) 19 133.36 (d) 6.38 (d, 10.8) 20 125.31 (d) 6.62 (dd, 15.3, 10.8) 21 136.14 (d) 6.30 (dd, 15.3, 5.6) 22 35.69 (d) 3.25 (m) 23 85.62 (d) 4.87 (d, 2.3) 24 75.38 (s) 25 130.40 (d) 5.47 (s) 26 133.96 (s) 27 34.68 (t) A- 352 (d, 13.4) B- 3.83 (d, 13.4) 28 154.05 (s) 29 114.21 (d) 626 (s) 30 174.76 (s) 31 59.98 (s) 32 65.00 (d) 2.73 (q, 5.4) 33 14.03 (q) 0.92 (d, 5.4) 34 175.39 (s) 35 49.49 (d) 251 (dq, 9.0, 7.1) 36 69.29 (d) 3.85 (m) 37 20.36 (q) 1.11 (d,,6.5) 38 24.20 (q) 1.49 (s) 39 18.13 (q) 0.93 (d, 6.8) 40 13.52 (q) 0.82 (d, 6.8) Table 4, cont. C# 13(: (ppm) (mult)a,b 1H (ppm) (mult,! (Hz»C 41 62.52 (0 A- 3.77 (dd, 10.8,4.0) B- 4.05 (bdd, 10.8, 7.6) 42 16.03 (q) 1.12 (d, 7.1) 43 56.01 (q) 3.26 (5) 44 10.99 (q) 2.00 (5) 45 19.08 (q) 1.60 (d, 7.1) 46 27.49 (q) 1.47 (5) 47 24.79 (q) 1.62 (fd, 1.1) 48 15.58 (q) 1.67 (5) 49 14.85 (q) 1.12 (d, 7.1) a measured at 125 MHz; referenced to C6D6 (128.0 ppm). b multiplicity determined with DEPT experiment. C measured at 500 MHz; referenced to 4DSH (7.15 ppm). 58 59 constructed from a 500 MHz double quantum-filtered, phase-sensitive COSy (DQCOSY) experiment acquired on patellazole C and are outlined in Figure 13.132,133 Apart from the disconnection between H-15 and H-16, the oc..x:OSY allowed the assignment of all the contiguous spin systems in the molecule. The atoms and functionalities not accounted for by the correlation experi­ment are summarized in Figure 14. H 6.30 H6.38 H H 1.53 , " , • 31 ~H2.S1 CH3 1.12 " . ,'$ $ OH , 4.3 H3C H 3.85 I.U· ~ HH 5 . 47 47 C~.k 25 -- 28 ....,... ..... ...' -\ -~ ~CH 2.73 H~ 0.9~ .. , ,. .. ...... ,', , H c4) 1.47 3 2<0H.. , 6.1S ".'. . ......3 4 ," , Figure 13. Partial structures of patellazole C constructed from a double quantum filtered, phase-sensitive COSy experiment. - CH3 1.49 - CH3 1.67 o ./ 'CH3 3.25 o A 216.43 -X\J 154.05 r-\.-114.21 H 6.25 X=NorO x A 171.36 60 x A 175.39 174.76 Figure 14. Structural elements of patellazole C unaccounted for by DQCOSY. Analogous correlation experiments on patellazoles A and B revealed the differences between the three metabolites. Both 65 and 66 have a methyl group on C-IO, in place of the hydroxymethyl in patellazole C. A triplet methyl signal (0.90 ppm) and correlations in the COSY spectrum showed patellazole A also lacked the hydroxyl at C-36, on the butyrate ester side chain. Table 5 lists the full NMR assignments for patellazole A and Table 6 contains the assignments for patellazole B. Before the partial structures assembled from the DQCOSY experiment could be confidently extended, the location and types of functionalities har­boring the nitrogen and sulfur atoms needed to be addressed. Raney Nickel Reduction of Patellazole A An attempt was made to desulfurize patellazole A by hydrogenation with Raney nickel. Patellazole A (65) was treated with excess Raney nickel in EtOH at room temperature. Two reaction products, one major, less polar Table 5 61 IH and 13C N:MR Assignments of Patellazole A C# 13(: (ppm) (mult)a,b 1 H (ppm) (mult,! (Hz»C 1 172.00(s) 2 80.75(s) 3 32.37(t) A- 1.70 (m) B- 2.63 (dt, 13.2, 5.9) 4 32.15(t) A- 1.02 (m) B- 1.25 (m) 5 28.46(d) 1.68 (m) 6 44.25(0 A- 1.20 (m) B- 1.59 (m) 7 73.33(d) 3.88 (bm) 8 48.75(d) 3.25 (dq, 9.5, 6.8) 9 214.11(s) 10 48.62(d) 4.13 (dq, 10.7,6.6) 11 130.61(d) 5.28 (dd, 10.7, 10.7) 12 131.72(d) 5.84 (dt, 10.7, 6.1) 13 33.33(t) A- 1.60 (m) B- 3.53 (dd, 11.9, 11.9) 14 38.90(d) 2.08 (m) 15 74.91(d) 3.69 (bm) 16 69.30(d) 3.85 (dd, 8.3, 2.5) 17 88.19(d) 4.00 (d, 8.3) 18 132.45(s) 19 132.63(d) 6.32 (d, 10.8) 20 125.87(d) 6.66 (dd, 15.3, 10.8) 21 136.51(d) 6.41 (dd, 15.3, 5.7) 22 36.12(d) 3.25 (m) 23 85.85(d) 4.92 (bs) 24 75.82(s) 25 131.93(d) 5.34 (s) 26 133.18(s) 27 34.74(t) A- 3.15 (d, 13.7) B- 4.11 (d, 13.7) 28 154.27(s) 29 114.66(d) 6.20 (s) 30 174.78(s) 31 60.56(s) 32 65.93(d) 2.69 (q, 5.4) 33 14.17(q) 0.93 (d, 5.4) 34 175.64(s) 35 41.73(d) 2.31 (dq, 13.6, 7.0) 36 27.02(0 A- 1.35 (m) B- 1.72 (dq, 13.6, 7.3) 37 12.00(q) 0.90 (t, 7.3) 38 24.01(q) 1.45 (s) 39 18.38(q) 0.71 (d, 6.4) C# 40 41 42 43 44 45 46 47 48 49 Table 5, cont. 13(: (ppm) (mult)a,b 14.09(q) 15.35(q) 16.14(q) 56.28(q) 11.42(q) 19.22(q) 27.30(q) 24.75(q) 16.08(q) 17.22(9) 1 H (ppm) (mult,! (Hz»C 0.89 (d, 6.8) 1.32 (d, 6.6) 1.01 (d, 6.6) 3.17 (s) 1.97 (s) 159 (d, 7.0) 1.44 (s) 154 (s) 1.79 (s) 1.17 (d, 7.0) a measured at 125 MHz; referenced to C6D6 (128.0 ppm). b multiplicity determined with DEPT experiment. C measured at 500 MHz; referenced to C6DSH (7.15 ppm). 62 Table 6 63 IH and 13C NMR Assignments of Patellazole B C# 13(: (ppm) (mult)a,b 1 H (ppm) (mult,] (Hz»C 1 171.95 (5) 2 81.36 (5) 3 32.43 (t) A- 1.83 (dt, 13.4,4.6) B- 2.60 (dt, 13.4, 4.6) 4 32.34 (t) A- 1.00 (dq, 13.4, 3.5) B- 1.42 (m) 5 28.61 (d) 1.73 (bm) 6 44.52 (t) A- 1.28 (ddd, 13.8, 12.0, 23) B- 1.55 (m) 7 73.03 (d) 3.88 (m) 8 48.97 (d) 3.25 (m) 9 214.57 (5) 10 48.54 (d) 4.09 (dq, 10.7,6.7) 11 130.63 (d) 5.29 (dd, 10.7, 10.7) 12 131.67 (d) 5.86 (dt, 10.7, 5.5) 13 32.91 (t) A- 1.52 (m) B- 3.57 (dd, 12.1, 12.1) 15 74.76 (d) 3.67 (bd, 9.3) 16 69.69 (d) 3.91 (bd, 8.8) 17 87.51 (d) 4.09 (d, 8.8) 18 132.57 (5) 19 133.55 (d) 6.38 (d, 10.8) 20 125.59 (d) 6.62 (dd, 15.3, 10.8) 21 136.59 (d) 6.35 (dd, 15.3, 5.6) 22 35.87 (d) 3.30 (m) 23 86.23 (d) 4.82 (d, 2.1) 24 75.70 (5) 25 131.14 (d) 5.37 (5) 26 133.82 (5) 27 34.85 (t) A- 3.37 (d, 13.4) B- 3.94 (d, 13.4) 28 154.29 (5) 29 114.62 (d) 6.22 (5) 30 174.99 (s) 31 60.33 (5) 32 65.55 (d) 2.71 (q, 5.5) 33 14.19 (q) 0.90 (d, 5.5) 34 175.73 (5) 35 49.66 (d) 2.49 (dq, 8.5, 7.1) 36 69.62 (d) 3.85 (m) 37 20.54 (q) 1.10 (d, 6.5) 38 24.38 (q) 1.46 (5) 39 18.33 (q) 0.72 (d, 6.6) 40 13.92 (q) 0.83 (d, 6.7) C# 41 42 43 44 45 46 47 48 49 Table 6, cont. 13(: (ppm) (mult)a,b 15.50 (q) 16.13 (q) 56.24 (q) 11.18(q) 19.10 (q) 27.42 (q) 24.86 (q) 15.98 (q) 14.94 (9) 1 H (ppm) (mult" (Hz»C 1.33 (d, 6.7) 1.11 (d, 7.0) 3.22 (s) 2.00 (s) 155 (d, 7.4) 1.44 (s) 157 (bs) 1.75 (s) 1.09 (d, 7.1) a measured at 125 MHz; referenced to C6D6 (128.0 ppm). b multiplicity determined with DEPT experiment. C measured at 500 MHz; referenced to C6DsH (7.15 ppm). 64 compound and a minor, more polar compound, were observed on silica gel TLC. The major degradation product was isolated by silica gel HPLC but the minor product decomposed on the column. 65 Low resolution FABMS of the isolated reaction product gave a mole­cular ion at m/z 872 (MH+), a net reduction of 16 mass units from 65. The 13C spectra (1 H decoupled and DEP'D revealed that the quaternary carbon at 60.56 ppm and the methine carbon at 65.93 ppm had shifted downfield to 129.54 and 128.78 ppm, respectively. Another major change was seen in an upfield shift of the sp2 carbon at 174.78 ppm to 171.83 ppm. Similarly, the 1H NMR spectrum had a new quartet resonance in the olefinic region (B 6.52), and two new olefinic methyl signals (B 2.07 (s) and 1.80 (d». The mass spectral and NMR data can be explained only by the reduc­tion of an epoxide in patellazole A followed by the loss of water. Additional support for an epoxide came from a 1H coupled DEPT experiment (DEPT++) performed on patellazole C.134 The methine carbon at B 65.00 in 67, equiva­lent to the 65.93 ppm signal in 65, had a 1JCH of 172.6 Hz, consistent with the increased s character in the C-H bonds of an epoxide.135 Although treatment with Raney nickel failed to disclose the location of the sulfur atom as expected, several key pieces of information were obtained. First, resistance of the sulfur atom in the patellazoles to Raney nickel reduc­tion ruled out a thiol, thioether, thioamide or other types of functionality which are easily desulfurized.136 Second, the confirmation and location of the epoxide were an important piece of information. A third piece of important information came as an indirect result of the Raney nickel epoxide elimination. In analyzing the IH coupled DEPT spectrum of 67, the methine carbon at 114.21 ppm was noted to have a 1JCH of 186.7 Hz. This large coupling is best explained by incorporating this carbon into a small heteroaromatic ring.137 With the nitrogen and sulfur atoms in 67 still unaccounted for, the presence of a thiazole was strongly suggested. Ozonolysis of Patellazole C 66 Unequivocal evidence for the novel thiazole-epoxide moiety came from treating patellazole C with a saturated solution of ozone in CH2CI2. Following a reductive work up with dimethyl sulfide, the sole UV active product was isolated by silica gel HPLC and shown to be the thiazole 82. The 200 MHz 1H NMR was consistent with the structure and a high resolution ElMS measurement confirmed the composition (measured 211.0675; C1(#lJNOS requires 211.0667). The physical data of 82 compared very favor­ably to the reported values for 2-t-butyl-4-methylthiazole and are summarized in. Table 7.138,139 Particularly noteworthy are the close correlations of the chemical shifts and coupling constants of C-5. The syn orientation of the epoxide methyl groups was established by difference nuclear Overhauser effect (NOE) spectroscopy. Irradiation of the methyl doublet at 0.88 ppm produced an NOE to the 1.73 ppm methyl singlet. With the structure of 82 confidently solved, only one more ring in patellazole C remained to be assigned. With the placement of the nitrogen and sulfur into the thiazole, fur­ther assembly of the pieces in Figures 13 and 14 into extended partial struc­tures was accomplished through a combination of the long-range 1H-13C cor­relation experiments COLOC and INAPT (selective INEPT), and a new 2D phase-sensitive INADEQUATE experiment. Table 7 Comparison of the Thiazole in Patellazole C with 2-t-Butyl-4-Methylthiazole Data Thiazole 82 2 -t-Buty 1-4-Methyl thiazole UV (MeOH): IH NMR (H-5): 13C NMR:a o Amax 251 nm (e 4200) 6.46 ppm (C6D6> C6D6(ppm) C-2: 174.76 C-4: 154.05 C-5: 114.21 (JCH = 186.7 Hz) a 13C chemical shifts reported are for patellazole C. Long-Range lH-13C Correlation NMR Experiments on Patellazole C 1,: s-..!( C(CH3b Amax 246 nm (e 4390) 6.52 ppm (CCLs) CCLs (ppm) 178.9 151.5 111.4 (JCH = 185.3 Hz) 67 The INAPT, or selective INEPT, pulse sequence is a one-dimensional experiment that permits selective detection of long-range IH-13C correlations. By selectively exciting a single IH frequency, only those carbons having a scalar coupling to the irradiated proton appear in the 13C spectrum.140,141 A series of INAPT spectra was acquired on patellazole C by irradiating all clearly defined IH signals. The resulting correlations are illustrated in Figure 15 and 68 key connections assigned from the data are discussed below. Irradiation at both H-3B (82.62) and Me-38 (8 1.49) gave strong two and three bond correlations to ~arbons at 881.25 and 171.36, placing the quaternary a-ester carbon at C-2 and the carbonyl at C-1, respectively. An alternative arrangement creating an ester linkage between C-2 and C-l can be proposed but would require four bond correlations through a heteroatom and are less likely to be observed. Subsequent data also eliminate this possibility. irradia­tion of H-8 (83.13), H:-I0 (84.28) and Me-40 (8 0.82) all gave rise to INAPT sig­nals at 216.43 ppm, placing the ketone at C-9 and extending the carbon chain to C-1S. Although H-1S failed to exhibit a vicinal coupling to H-16 in the DQCOSY experiment, H-17 (84.09) showed a correlation to C-1S (874.32) in Figure 15. INAPT correlations seen in patellazole C. 69 the INAPT spectrum. Additional three bond correlations from H-17 to the methoxyl carbon (C-43, 0 56.01), the olefinic methyl group at 10.99 ppm (C-44) and C-19 in the diene (0 133.36) extends the carbon chain to C-23. Upon irradi­ating the narrow doublet of H-23 (04.87) a strong signal occurred in the carbon spectrum at 171.36 ppm, the ester carbonyl previously assigned as C-l. This correlation creates a 24 membered macrolide and satisfies the remaining un­saturation requirement. Further correlations of H-23 to C-24 (0 75.38), Me-46 (0 27.49) and C-25 (0 130.40) lengthen the carbon skeleton to C-26. There were no INAPT data allowing extension of the carbon chain beyond C-26. However, with the isolation of the thiazole from the ozonolysis of patellazole C the assignment of the side chain was complete. Figure 15 also contains INAPT correlations about the thiazole unit that provided additional support for its structure. The remaining correlation in Figure 15 requiring comment occurs be­tween H-35 (02.51) and the carbonyl at 175.39 ppm. Together with the IH-IH correlation data this forms a 2-methyl-3-hydroxybutyrate side chain. Unfort­unately, no evidence was found in the INAPT placing the ester on the main structure. In order to corroborate some weak INAPT correlations and try to place the butyrate side chain, two COLCX: experiments were acquired on patellazole C. The COLOC pulse sequence is a two-dimensional experiment that selects for long-range IH-13C correlations through the small n]CH (0-15 Hz for n > 1) couplings constants.142 It differs from the conventional HETCOR experiment optimized for small] values by including the proton evolution time, tt, into the polarization transfer delay, At. This effectively reduces the time that proton magnetization is lost to transverse relaxation (T2) from t1 + A1 to only At- The result is better sensitivity. 70 COLOC experiments on patellazole C were acquired using n]CH values of 9 and 15 Hz. A summary of relevant correlations extending the partial structures from Figures 13 and 14 is illustrated in Figure 16. Of major importance were correlations placing the isolated C-27 methylene and those further confirming the C-15 to C-16 connection. 2D Phase-Sensitive INADEQUATE Spectrum of Patellazole C In addition to the long-range correlation techniques discussed above, extended partial structures in patellazole C were assigned using a new 2D phase-sensitive INADEQUATE experiment developed by Dr. Charles L. Mayne in the Chemistry Department at the University of Utah)43 Patellazole C was the first natural product investigated with this unique new pulse sequence and a brief description is in order. The ID INADEQUATE experiment was first described by Bax as a method for measuring carbon-carbon coupling constants)44 Since then, it has evolved into a powerful tool for structure elucidation because it provides a means of directly mapping out the carbon skeleton of a molecule. However, the INADEQUATE is one of the least sensitive NMR experiments available. Because the INADEQUATE experiment detects two adjacent 13C nuclei in a molecule, and since the natural abundance of 13C at a given position is 1 %, the occurrence of such a molecule in Nature is only 1 in 10,000. Nevertheless, modern high field NMR spectrometers can perform the experiment on as little as 0.1 mmol of sample. The original one-dimensional experiment was soon extended to dis­perse the scalar couplings in the second dimension according to their double quantum frequencies.145 The 2D INADEQUATE spectrum consists of pairs of doublets with identical double quantum frequencies and with each doublet 71 H3C , ]; 0 H HO CH3 H Figure 16. COLOC correlations seen for patellazole C. split by lIce and centered (except for a small isotope shift) at the chemical shift of the carbon in the molecule. The double quantum frequencies serve to uniquely identify pairs of carbons sharing a scalar coupling, i.e., carbon-carbon bonds. Dr. Mayne's modification of the standard 2D INADEQUATE experi­ment was to implement the method of States et al. for obtaining phase-sensi­tive spectra.146 This requires the aquisition of two separate data sets repre­senting the real and imaginary parts of a complex spectrum. Use of this method permits phase-sensitive display of two-dimensional data in much the same manner as one-dimensional spectra are treated. When used in con­junction with the two-dimensional INADEQUATE experiment, phase- 72 sensitive aquisition is particularly beneficial because the resulting doublets occur in antiphase and are much easier to recognize in low signal-to-noise spectra than are absolute value doublets. Additionally, pure absorption spectra have inherently better signal-to-noise ratios than do spectra acquired in the dispersive mode. One more important aspect of the 2D INADEQUATE experiment arises from the symmetry requirement about the double quantum frequency. Even if one of the antiphase doublets is weak or absent, its chemical shift can be inferred from the position of the visible partner. This is possible because the double quantum frequency is simply the algebraic sum (in hertz, relative to the carrier frequency) of the chemical shifts of the two coupled carbons. The position of the absent partner is located the same distance from the diagonal as the the visible doublet, but on the opposite side. Another modification of the 2D INADEQUATE pulse sequence was required due to the large range of chemical shifts in patellazole C. At high field, the width of a carbon spectrum can exceed 25 kHz; it then becomes diffi­cult to generate rf pulses powerful enough to accurately perturb all of the spins. A pulse sequence replacing the 1800 pulses in the ID INADEQUATE with composite pulses has been published and was adapted for use in the two­dimensional experiment.147 Figure 17 shows representative examples of carbon-carbon correlations seen in the 2D phase-sensitive INADEQUATE spectrum of patellazole C. The data are best displayed as slices through F2, the double quantum frequency, since the antiphase doublets clearly stand out. Figure 18 illustrates all the carbon-carbon connectivities seen for patellazole C (67) as bold lines. The experiment was optimized for sp3 to sp3 couplings using a lIce value of 35 Hz which results in the lack of any observed sp2 to sp2 connections. 0'1 - ,..... U) c.o ex:> ex:> I"") If) ~~ 1""),..... ~~ ex:> If) I"") I"") C 14 to C42 and C22 to C45 C5 to C39 C15 to C14 ... .I".".) -0'1 - ~~ -0'1 ,..... ,..... U) .... ~~ ex:> If) .... U) o U) ex:> I"") Figure 17. Selected traces from the 2D phase-sensitive INADEQUATE spectrum of patellazole c. 73 Figure 18. Carbon-carbon connectivities in patellazole C established by 2D phase-sensitive INADEQUATE. Placement of the 2-Methyl-3-Hydroxybutyrate Ester in the Patellazoles 74 The remaining structural feature to be addressed in patellazole C is the positioning of the 2-methyl-3-hydroxybutyrate on C-2 of the macrolide. The DQCOSY experiment assigned three signals in the proton spectrum at 2.65, 6.15 and 4.36 ppm to hydroxyls on C-16, C-24 and C-36, respectively, and elim­inated placing the ester at these sites. A deuterium exchanged IH spectrum showed sharpening of the signals for H-7, H-15 and H-16, indicating these protons were on carbons bearing a hydroxyl. This left C-2 and C-41 as possible sites of attachment for the butyrate. 75 Direct evidence showing the C-2 oxygen was substituted with the ester came from a deuterium exchanged 13C NMR experiment on patellazole A (which lacks the C-36 and C-41 hydroxyls). Carbons 7, 15, 16 and 24 exhibited isotope-induced up field shifts from 0.1 to 0.17 ppm, whereas carbons 2, 17 and 23 showed negligible changes. With the positioning of the 2-methyl-3-hydroxybutyrate, the two­dimensional structures of patellazoles A, B and C were complete. The final task remained to assign the relative and absolute stereochemistry of the 16 chiral centers. Due to the number of stereocenters involved and the difficulty in relating the configurations of the many remote chiral centers, the stereo­chemistry could be best solved by an x-ray crystal study. Numerous attempts to crystallize the patellazoles failed; nearly every set of conditions resulted in the compounds coming out of solution as an oil. To increase the likelihood of crystallization, a number of derivatives designed to make the molecules more rigid and/or more or less polar were prepared including acetates, p-bromobenzoates, p-bromophenylcarbamates and opened epoxide derivatives; unfortunately, none of the derivatives afforded a crystal­line sample. Minor Patellazoles In addition to the three major patellazoles (65 - 67) described above, four minor metabolites have been isolated but the structures not assigned. Two of these compounds, patellazoles D and E were isolated as 2.5 mg of an inseparable mixture. Patellazole F was found in the largest amount, 6.8 mg and is an isomer of patellazole C. The fourth minor constituent, patellazole G appears to be identical to 31,32 deoxypatellazole A based on comparison of NMR and F ABMS spectral data. 76 Pa tellazoles D and E During the purification of patellazole C, 2.5 mg of a minor, more polar HPLC fraction was collected and shown by NMR to contain two compounds in about a 5:3 ratio. Chromatographic studies failed to reveal a system which could separate the compounds and preliminary data were collected on the mixture. FAB mass spectrometry showed a clear parent ion at m/z 938 (patel­lazole D, MH+) and a (M+Na)+ ion at m/z 960. A less intense ion at m/z 906 was assigned to the parent ion of the other component (patellazole E). The 13C NMR spectrum revealed only 87 resolved signals, preventing much con­jecture as to qualitative differences from the characterized patellazoles. The 1 H NMR spectrum was very complex and provided little useful information. As with the structurally defined members of this family of compounds, patellazole D differs by sixteen mass units, probably by addition of an extra hydroxyl. Patellazole E, at a molecular weight of 905 may possibly be a dihyd­ropatellazole B, a lower homolog of patellazole C, or a respective isomer. The best clues to the modifications contained in these compounds are the lack of any signals in the 13C spectrum for epoxide carbons. Correspondingly, there are at least 18 carbon signals between 69 and 90 ppm in the mixture whereas the same region in patellazole C only has eight resonances. Patellazole F Patellazole F eluted immediately before patellazole C on reverse phase HPLC (15% aqueous MeOH) and with 6.8 mg of material, was the most abun­dant of the minor compounds. Low resolution FAB mass spectrometry estab­lished a MH+ of 920 mass units, the same molecular weight as patellazole C. The 13C NMR data revealed several shifted signals in the olefinic region and the carbon at 72.56 ppm in patellazole C (C-7) had become one of three signals 77 at 69 ppm in patellazole F. With the data available, the most plausible structure for patellazole F is that of the C-7 epimer of patellazole C. Patellazole G The hexane soluble material (1.53 g) from the solvent partition showed several nonpolar, UV active compounds by TLC and was treated similarly to the CCLt fraction. Final purification by silica gel HPLC produced small amounts of patellazoles A and Band 3.5 mg of the minor metabolite patellazole G. FABMS of patellazole G showed a strong MH+ ion at m/z 872, match­ing a molecular formula with one oxygen less than patellazole A. The 13C NMR data contained 12 olefinic signals and prompted spectral comparison to 31,32 deoxypatellazole A. The IH spectra of the two compounds, obtained in benzene-D6, were superimposable and the 13C signals of patellazole G were within 0.2 ppm of the corresponding carbons in the DEPT spectrum of 31,32 deoxypatellazole A. While obtaining a carbon spectrum of patellazole G in CDCl3, the only solvent that a full spectrum of 31,32 deoxypatellazole A was recorded in, the sample decomposed; therefore, a comparison of the full 13C spectra is impossible because the 31,32-deoxypatellazole A sample was submit­ted for biological evaluation. Without additional evidence from data such as UV and IR spectra, an unequivocal assignment of the structure of patellazole G cannot be made. Biological Activity of the Patellazoles The patellazoles are the most potent cytotoxins of any tunicate meta­bolite and have generated considerable interest frolll: the pharmaceutical 78 industry, leading to the filing of a patent application. The most impressive activity was identified in the National Cancer Institute's human cell line protocol. Patellazole B possessed the highest activity with a mean in vitro ICso of less than 10-6 Jlg/mL against more than 30 tumor cell lines. In the same assay, patellazoles A and C demonstrated ICso's of 3 x 10-4 Jlg/mL and 3 x 10-3 J.1g/mL, respectively. These values are essentially the same as those found for the patellazoles against the L1210 murine leukemia cell line. The crude extract of L. patella was found to have good antiviral activity (ICso = 2.63 Jlg/mL, TI = 38 against vesicular stomatitis virus) and the pure patellazoles Band C were submitted for further testing.148 Both compounds demonstrated activity against Herpes simplex viruses 1 and 2 (HSV-l and -2) at concentrations which were not toxic to the Vero host cells. Patellazole B had the highest antiviral activity with ICSO's of 5 x 10-4 and 6 x 10-2 J.1g/mL, respectively, and a cytotoxic ICso of 0.75 J.1g/mL. This equates to an in vitro therapeutic index of 1500 for HSV-1. In the same assay, patellazole Chad antiviral ICso's of 0.3 and 0.8 J.1g/mL and a cytotoxic ICSO of 4 J.1g/mL. The antiviral activity of patellazoles Band C was also explored against influenza A, human rhinovirus (HRV-14) and human cytomegalovirus (HCMV) but proved to be too toxic towards the MRC-5 and MDCK host cells. Patellazole B has been investigated for in vivo antiviral activity. Mice infected with HSV-l were treated with patellazole B and acyclovir (ACV) at several different doses and schedules. Untreated control animals experienced a 90% mortality rate with a median survival time (MST) of 7.5 days. Animals treated with patellazole B at doses of 0.5 mg/kg/day and 1.0 mg/kg/day given every other day had an MST of 8.0 days, while animals treated with higher and lower doses had an MST of 7.5 days. Acyclovir (40 or 80 mg/kg/day) ex­hibited significant protection having an MST >12 days at both dosage levels. 79 These data indicate patellazole B has no significant antiviral activity in mur-ine infections at levels below the LDSO, determined to be 1-2 mg/kg/day given every other day. Biochemical studies have shown that the patellazoles stimulate incorporation of labelled thymidine into DNA. This observation indicates that the compounds cause damage to DNA which is then repaired. The rel­ative resistance to cytotoxicity observed in Vero cells appears to be the result of an increased capacity to repair DNA damage. Isolation and Purification of the Patellins In addition to the patellazoles, the CC4 fraction from the solvent parti­tion scheme was also found to contain a second family of compounds. In­spection of the later effluent from the silica gel column used to isolate the patellazoles showed several weakly UV active, nitrogenous compounds. Sep­aration of these compounds by successive reverse phase and silica gel HPLC yielded 8.5 mg of patellazole C (67),4.8 mg of patellin 1 (83), 134 mg of patellin 2 (51), 83.2 mg of mixed patellins 3 (84) and 4 (85), and 5.4 mg of patel lin 5 (86). General Characteristics of the Patellins Preliminary NMR analyses of the chromatographic fractions contain­ing the patellins suggested that, except for patellin 1 (83), each consisted of a mixture of closely related peptides. This was indicated by nearly every signal in the spectra having a smaller sister signal. After repeated attempts to separ­ate the mixtures failed, it was observed that the ratio of doubled signals in the NMR spectra could be varied with solvent and confirmed that the purported mixtures represented single peptides existing in two conformations. The conformer ratios vary with the dielectric constant, Er, of the solvent: solvents 80 (83) 14 S'F 16 • 30~29'#~N8 11 ~ 18. 0 N~NH K,: 31 28 0 H 7 0 20 21 0 33 5 :t4 4 N NH 3 2 26 25 27 (51) -<~~J~ o ~ HN,l.,UO- ~ o (84) :1y ~t ~HNJ,l.,u~O- ~ o (85) --=: ~,# s~o o NH ~N HNto~ o (86) 81 82 with high values of Er, such as acetone or acetonitrile, give spectra with the greatest difference in conformer distribution while those solvents with low dielectric constants, like methylene chloride or chloroform, result in spectra displaying nearly equal conformer populations. This is most easily seen in the uncrowded olefinic region of the 13C NMR spectrum of patellin 2 (51) illustrated in Figure 19. While the doubled NMR spectra were simplified when acquired in acetone-D6 and acetonitrile-D31 the patellins were only slightly soluble in I 14(J I 133 PPM i 12~ Figure 19. 50 MHz 13C NMR spectra of the olefinic region of patellin 2 (0.14 M sample, 0.5 Hz line broadening). Top, spectrum acquired in acetone-D6. Bottom, spectrum acquired in CDCl3. 83 these solvents and the clearly visible minor conformers served to complicate all two-dimensional correlation experiments. For this reason alternative methods to N:MR were required to solve the patellin structures. Structure Determination of the Pa tellins The Structure of Pa tellin 2 The first of these cyclic peptides studied was patellin 2 (51), the major component of the family. The infrared spectrum of 51 showed a broad NH stretching absorption at 3327 cm-I and strong amide carbonyl absorptions cen­tered at 1670 cm-I . The UV spectrum exhibited an absorption at 210 nm (E 8400) with a weak shoulder at 248 nm (E 2450). FAB mass spectrometry estab­lished a molecular weight of 732. Because the molecule gave a weak parent ion, the high resolution peak match measurements were made on an m/z 597 fragment ion arising from the loss of two CsHs groups (found 597.3069; C27IttsN607S requires 597.3071). Although considerable effort was invested in N:MR analysis of 51 (1 H and 13C assignments are listed in Table 8 and spectra are included in Appen­dix C), the complex spectra made it obvious that the most direct method of solving the structure would be x-ray crystallography. Fortunately a crystalline sample of 51, obtained as flat plates from aqueous MeOH, was found suitable for x-ray analysis. The x-ray solution was provided by Prof. Jon Clardy and Mr. Thomas J. Stout at Cornell University, Ithaca, New York and served to define the complete relative stereostructure of patellin 2 (51): a computer­generated perspective drawing is given in Figure 20. The absolute configuration of patellin 2 was not independently defined by the x-ray analysis and was determined using Marfey's procedure for deter­mining the chirality of amino acids as their 1-fluoro-2,4-dinitrophenyl-5-L- Table 8 84 1H and 13C NMR Assignments of Patellin 2 C# 13C (ppm) (mult)a,b lH (ppm) (mult,] (Hz»C 1 174.46 (5)- 2 48.94 (t) A- 3.97 (m) B- 3.74 (m) 3 25.85 (t) 2.05 (m) 4 30.76 (t) A- 2.38 (m) B- 1.93 (m) 5 64.99 (d) 4.19 (bd, 10.8) 6 173.87 (5)- N2W 6.28 (d, 9.3) 7 59.45 (d)+ 4.29 (dd, 9.3, 1.7) 8 68.10 (d)+ 4.46 (dq, 6.3, 1.7) 9 21.49 (q)+ 1.10 (d, 6.3) 10 171.42 (5)- N3H 7.72 (d, 8.2) 11 50.45 (d) 4.75 (m) 12 44.06 (t) A- 2.08 (m) B- 1.44 (m) 13 25.38 (d) 1.79 (5pt, 6.6) 14 23.27 (q) 0.95 (d, 6.6) 15 22.69 (q) 0.90 (d, 6.6) 16 170.96 (5)- 17 34.20 (t) A- 3.77 (dd, 11.1,8.1) B- 3.54 (dd, 11.1, 10.2) 18 79.35 (d) 5.42 (ddd, 10.2, 8.1,2.0) 19 170.54 (s)- N5H§ 7.48 (d, 10.0) 20 60.94 (d)§ 4.18 (dd, 10.0, 1.4) 21 68.45 (d)§ 4.37 (dq, 6.2, 1.4) 22 21.29 (q)§ 1.21 (d, 6.2) 23 170.21 (5)- N6H 7.29 (d, 7.2) 24 56.80 (d) 4.54 (dd, 7.2, 1.8) 25 31.52 (d) 2.38 (m) 26 21.10 (q) 1.09 (d, 6.8) 27 18.02 (q) 0.88 (d, 6.8) 28 76.29 (5)- 29 28.01 (q)- 1.28 (s) 30 28.01 (q)- 1.18 (5) 31 145.93 (d)· 5.93 (dd, 17.62, 10.81) 32 113.36 (t)· A- 5.17 (dd, 17.62, 1.13) B- 5.05 (dd, 10.81, 1.13) 33 76.11 (s)# 34 26.77 (q)# 1.28 (5) C# 35 36 37 Table 8, cont. 13C (ppm) (mult)a,b 26.32 (q)# 145.84 (d)# 113.64 (t)# IH (ppm) (mult,] (Hz»C 1.18 (s) 5.87 (dd, 17.64, 10.83) A- 5.10 (dd, 17.64, 1.22) B- 5.02 (dd, 10.83, 1.22) a Signals marked with It are interchangeable. The spin system marked with t is interchangeable with that marked § and the system denoted by • is not distinguished from the corresponding system marked with #. b Measured at 100 MHz and referenced to internal acetone-D6. Multiplicity determined by DEPT experiment. C Measured at 400 MHz and referenced to residual acetone-DsH. 1 H_13C correlations assigned using HETCOR. 85 86 C17b Figure 20. Computer-generated perspective drawing of patellin 2. Hydrogens have been omitted for clarity. Both conformations of the thiazoline are superimposed. alanine amide (FDAA) diastereomers.149 Thin layer chromatography (What­man precoated C1S F254; 1:1 H20:MeOH) of the FDAA derivatized total acid hydrolysate of 51 against like derivatized standards, unequivocally showed that the valine and leucine residues possessed the L configuration. These assignments were further confirmed by reverse phase HPLC which also served to assign the L stereochemistry to the threonines and proline. The absolute configuration of patellin 2 is then defined as 55, 75, 8R, 115, 18R, 205, 21R, 245. Conformational Analysis of Patellin 2 In addition to defining the three-dimensional structure, the x-ray ana­lysis, similar to the solution NMR spectra, also revealed two conformations were available to patellin 2. The thiazoline ring was disordered in the solid 87 state and had to be modeled with two distinct conformations. The simplest analysis, illustrated in Figure 20, involved disordering two atoms, Sl and C17, as shown. Basically two different C16-S1-C17-C18 torsional angles were found: 21.10 and -31.30 • In an attempt to provide additional insight into the disorder, the results of the crystal structure analysis were investigated by Prof. Clardy with molecular mechanics (MacroModel),150 and two distinct minima with torsional angles essentially the same as those in the solid state were found. The molecular mechanics analysis disclosed that the two conforma­tions differed by only 0.15 kcal/mol while the presumed intermediate inter­converting the two conformations was only 0.20 kcal/mol above the higher energy. This energy barrier is too low to account for the observation of two discrete species on the NMR timescale; therefore, another type of conforma­tional interconversion must be responsible for the solvent dependent doubling of signals in the NMR spectra. More rigorous molecular modeling studies were undertaken in our laboratory by Mr. Mark P. Foster using the software package of CHARMm and QUANTA.151 By subjecting a crude set of three-dimensional coordinates to a regime of molecular dynamics calculations, over 900 possible structure con­formations were generated. A random sample of these was selected and then energy minimized using a modified Newton-Raphson algorithm until the energy change between steps was less than 0.001 kcal/mol. The result was good convergence to two structures of distinct energy minima. However, instead of differing in the conformation of the thiazoline like the crystal structures, the computer-generated structures reflected cis and trans isomers of the proline amide bond. The two conformers differed in energy by 1.1 kcal/mol and the barrier to inversion was determined to be 21 kcal/mol. 88 Investigation of the literature revealed that cis -trans isomerism of prolines was not uncommon for small cyclic peptides and that cyclic hexapep­tides were observed to have greater freedom than cyclic octapeptides.152 Furthermore, it has been reported that cis -trans isomers could be distinguish­ed in solution by the difference in chemical shifts of the ~ and 'Y carbons; in cis-proline these signals are further separated than in trans-proline.153 The chemical shifts of the ~ and 'Y proline carbons in the major conformer of S1 (C-3 and C-4 in Table 8) differ by 4.91 ppm. Reinvestigation of the 13C NMR spectrum of patellin 2 showed that the corresponding carbons of the minor conformer differed by 7.65 ppm (22.45 and 30.10 ppm, respectively), indicating that the predominant signals arise from the trans conformer. Presumably, the thiazoline inversion observed in the solid state is also occurring in solution. However, as previously stated, the 0.2 kcal/mol inver­sion barrier is low enough that detection of two discrete. thiazoline species would not be possible on the NMR timescale. Conversely, NMR spectra of patellin 2 acquired at low temperature showed a predominance of the trans conformer and a clear increase in the population of the cis conformer could be seen as the temperature was raised. Further investigation of the conformations of the patellins is being carried out in our lab by Mark Foster. These studies involve the analysis of scalar couplings, quantitative NOE spectroscopy and molecular mechanics to arrive at global structures modeled in solvents of different dielectric constants. The Partial Structures of Patellin 1 The minor metabolite patellin 1 (83) was the single member of this family of compounds that was not seen to exist in multiple conformations by 89 NMR. Characteristic amide absorptions were observed in the IR at 3349, 1677, 1673, 1655, 1650 cm-1 and the UV spectrum contained maxima at 220 (e 3800) and 247 (sh) run. Positive and negative ion FABMS established a molecular weight of 732 and a molecular formula of C37H6oN607S was confirmed by HRFABMS (observed 733.4247 (MH+), C37~lN607S requires 733.4324) and defines 11 degrees of unsaturation in the molecule. Patellin 1 failed to react with ninhydrin, thereby indicating a cyclic structure or a modified N-termi-nus. T