Recent Progress of the Synthetic Studies of Biologically Active Marine Cyclic Peptides and Depsipeptides

Yasumasa Hamada and Takayuki Shioiri

Graduate School of Pharmaceutical Sciences, Chiba University, Yayoi-cho, Inage-ku, Chiba 263-8522, Japan, and Graduate School of Environmental and Human Sciences, Meijo University, Shiogamaguchi, Tempaku, Nagoya 468-8502, Japan

--------Chem. Rev., 105 (12), 4441 -4482, 2005

Contents

• 1.Introduction
• 2.Cyclic Peptides from Cyanobacteria
• 2.1.Antillatoxin
• 2.1.1.Total Synthesis of (4S,5R)-Antillatoxin Having the Proposed Structure
• 2.1.2.Total Synthesis of (4R,5R)-Antillatoxin (Natural Form) Having the Revised Structure
• 2.1.3.Biological Activities of Antillatoxins
• 2.2.Somamide A
• 2.2.1.Total Synthesis
• 2.2.2.Stereospecific Dehydration of -Hydroxy- -amino Acids Using Martin's Sulfurane
• 2.3.Kahalalide F
• 2.4.Apratoxin A
• 2.5.Lyngbyabellins A and B
• 2.5.1.Total Synthesis of Lyngbyabellin A
• 2.5.2.Total Synthesis of Lyngbyabellin B
• 2.6.Dendroamide A
• 3.Cyclic Peptides from Sponges
• 3.1.Microsclerodermin E
• 3.2.Phakellistatins
• 3.3.Cyclotheonamides E2 and E3
• 3.4.Papuamides A and B
• 3.5.Halipeptins A and B
• 4.Cyclic Peptides from Red Alga
• 4.1.Ceratospongamides
• 5.Cyclic Peptides from Ascidians
• 5.1.Tamandarins A and B>
• 5.2.Mollamide
• 5.3.Trunkamide A
• 6.Cyclic Peptides from Sea Hare
• 6.1.Aurilide
• 7.Cyclic Peptides from Bacteria
• 7.1.Cyclomarins
• 8.Conclusion
• 9.Acknowledgments
• 10.Abbreviations
• 11.References

1. Introduction

Cyclic peptides including cyclic depsipeptides are still a growing research area even in the 21st century. ^[1] They are mainly produced by marine organisms and terrestrial microorganisms. Marine organisms are a well-known rich source of biologically active cyclic peptides having unique structures. They contain unusual amino acids and building blocks. Most of them will offer a new frontier in both synthetic organic chemistry and biological activities. In particular, unique unusual structures will offer the challenge of exploitation of novel synthetic methods, reactions, reagents, catalysts, etc. Thanks to the progress of isolation procedures such as HPLC and methods for structure determination such as NMR and mass spectral means, numbers of natural products of marine origin having both unique structures and intriguing biological activities are increasing. ^[2] However, generally speaking, it is rather difficult to isolate a larger amount of marine natural products because of minute constituents in organisms, difficulty of collection of organisms, and resistance of laboratory culturing. Thus, the limits of the quantity preclude the precise structure determination as well as clarification of detailed biological activities. In fact, the proposed structures of many cyclic peptides have been revised by synthetic works. Thus, the total synthesis is still playing a final means for the structure determination of marine natural products just like several decades ago. Furthermore, the efficient large-scale production of marine natural products by synthesis will offer an opportunity to investigate their biological activities in detail. One of the characteristic features of cyclic peptides will be their conformational rigidity and stability in vivo, in contrast to their linear counterparts. In addition, unusual amino acid and non-amino acid moieties of marine cyclic peptides will offer the lead structures of new biologically useful compounds.

This review focuses on examples of broad interest in the recent progress of the synthetic studies of marine cyclic peptides and depsipeptides having unique structures and intriguing biological activities. There are some reviews on the synthesis of cyclic peptides which generally describe the progress until the end of the 20th century.^[3] This review is not intended to be comprehensive, but the more recent results will be presented. Some microorganisms of freshwater origin also produce interesting cyclic peptides,3a but most of them are out of the scope of this review. Diazonamides isolated from ascidian also belong to a kind of cyclic peptides, but it is also out of the scope because its peptide part is so small in the whole structure and the recent progress is too enormous to review.^[4]4

2. Cyclic Peptides from Cyanobacteria
Cyanobacteria (blue-green algae) of aquatic origin are known to produce a number of interesting biologically active cyclic peptides, ^[5] which are sometimes called cyanopeptides.^[5c] They are now known as a rich source of the lead compounds for pharmaceuticals and biologically useful compounds. In particular, recent findings that the biologically active metabolites of various marine organisms are actually produced by cyanobacteria as symbionts or feeding materials have prompted studies on the metabolites of cyanobacteria.

2.1. Antillatoxin
Antillatoxin is a cyclic lipodepsipeptide isolated in low yield by Gerwick and co-workers^[6] from the marine cyanobacterium Lyngbya majuscula collected in Curaçao. It shows a strong ichthyotoxicity and neurotoxicity which is derived by activation of the mammalian voltage-gated sodium channel at a pharmacological site.^[7] The unique structure 1a involving many methyl and one tert-butyl functions was determined by extensive NMR spectral studies, and the absolute configurations of the N-methylvaline and alanine parts were deduced to be (S) by comparison of chiral-phase TLC results of their acid hydrolysis products. The stereochemistry at the C4 and C5 positions was assigned to be (4S,5R) by combination of NMR spectra (NOESY and J value), CD spectroscopy, and molecular modeling (the lowest energy conformation). As described later, the (4S,5R) configurations of 1a was revised to the (4R,5R) ones as 1b by synthetic works (Figure 1).^[8]

--------------------------------------------------------------------------------
Figure 1

--------------------------------------------------------------------------------

2.1.1. Total Synthesis of (4S,5R)-Antillatoxin Having the Proposed Structure
The first total synthesis of antillatoxin having the proposed structure 1a was achieved by Yokokawa and Shioiri.^[8a] They constructed the whole molecule by coupling of the tripeptide unit with the conjugated diene unit. The lactamization instead of lactonization was employed to construct the macrocyclic depsipeptide because of the higher nucleophilicity of the amino group compared to the hydroxyl one.^[9]

Preparation of the tripeptide unit 2 was rather straightforward utilizing diethyl phosphorocyanidate (DEPC, (EtO)₂P(O)CN) ^[10] and bis(2-oxo-3-oxazolidinyl)phosphinic chloride (BopCl) ^[11] as the coupling reagents, as shown in Scheme 1.

--------------------------------------------------------------------------------
Scheme 1

--------------------------------------------------------------------------------

Synthesis of the conjugated diene unit was carried out in two ways. The first route employed Horner-Emmons reaction twice to construct the diene, as shown in Scheme 2. The Horner-Emmons reaction of pivalaldehyde (3) with the phosphonate 4 mainly afforded the E-isomer 5, which was sequentially treated with diisobutylaluminum hydride (DIBAL, i-Bu₂AlH), chemical manganese dioxide (CMD),^[12] and the phosphonate 6, giving the conjugated diene 7. Reduction of 7 with DIBAL afforded the required alcohol 8. The improved route to the diene 8 employed Suzuki-Miyaura coupling^[13] (Scheme 2). The pentyne 9 underwent the hydroboration with catecholborane followed by hydrolysis to give the boronic acid 10. The Suzuki-Miyaura coupling of 10 with the iodide 11 afforded the dienyl alcohol 8.

--------------------------------------------------------------------------------
Scheme 2

--------------------------------------------------------------------------------

The Evans syn-selective asymmetric aldol reaction^[14] was employed to construct the C4 and C5 chirality in antillatoxin (1a), shown in Scheme 3. Oxidation of the dienyl alcohol 8 with CMD followed by the Evans aldol reaction with the boron enolate from the carboximide 12 afforded the aldol adduct 13. After removal of the chiral auxiliary, methyl esterification afforded the ester 14. Protection of the secondary alcohol with triethylsilyl (TES, Et3Si) chloride and then treatment with DIBAL afforded the alcohol 15, which was oxidized with tetrapropylammonium perruthenate (TPAP, Pr4NRuO4).^[15] The resulting aldehyde underwent the Still-Horner cis-selective olefination^[16] with the phosphonate 16 to give the (Z)-ester 17. Acidic treatment of 17 afforded the ,-unsaturated lactone 18, which was converted to the selenolactone 19 using phenylselenomethyllithium.^[17] The stereochemistry of 19 was confirmed by NOE experiments.

--------------------------------------------------------------------------------
Scheme 3

--------------------------------------------------------------------------------

Ring opening of the selenolactone 19 by saponification, allyl esterification, followed by coupling with the tripeptide 2 using 1-[3-(dimethylaminopropyl)]-3-ethylcarbodiimide hydrochloride (EDCI·HCl, Me₂N(CH₂)₃-N=C=N-Et·HCl) afforded the linear ester 20, from which the phenylselenyl group was oxidatively removed with sodium periodate. The resulting exo-methylene compound 21 underwent deprotection of both N- and C-terminals at the same time, and final macrocyclization was achieved with diphenyl phosphorazidate (DPPA, (PhO)₂P(O)N₃)^[10,18] in the presence of sodium hydrogen carbonate to produce (4S,5R)-antillatoxin (1a) (Scheme 4). However, the NMR spectrum of the synthetic antillatoxin (1a) was revealed to be different from those of natural antillatoxin. In addition, the optical rotation of the synthetic antillatoxin (1a) ([]D= -55 (c 0.24, MeOH)) was different from that of the natural one ([]D = -140 (c 0.13, MeOH)). This discrepancy clearly indicated that structure 1a did not show the stereostructure of natural antillatoxin.

--------------------------------------------------------------------------------
Scheme 4

--------------------------------------------------------------------------------

In this synthetic work^[8] choice of the substrates and sites for macrolactamization was very important because the other substrates 22-25 shown in Figure 2 could not afford the cyclized products by attempted macrocyclization.

--------------------------------------------------------------------------------
Figure 2

--------------------------------------------------------------------------------

A little bit later White and co-workers^[19] also achieved the total synthesis of (4S,5R)-antillatoxin and reached the same conclusion.

The ester 26 was first converted to the allysilane 28 by addition of an excess amount of Grignard reagent 27 in the presence of cerium chloride, followed by the Peterson elimination with silica gel. Reaction of 28 with the dithienium salt 29 and then treatment with boron trifluoride etherate afforded the alcohol 30, which was converted to the aldehyde 31 with o-iodoxybenzoic acid (IBX), as shown in Scheme 5.

--------------------------------------------------------------------------------
Scheme 5

--------------------------------------------------------------------------------

In a parallel procedure the alkyne 9 underwent hydrozirconation and then iodination with N-iodosuccinimide (NIS) to produce the iodide 32, from which the eneyne 34 was obtained by coupling with the Grignard reagent 33, shown in Scheme 6. Conversion of the eneyne 34 to the dienylstannane 35 followed by iodination afforded the diene iodide 36. After the halogen-lithium exchange reaction with tert-butyllithium, addition to the aldehyde 31 afforded the syn-adduct 37 as a major product. The ratio of syn and anti isomers was 8:1, which is easily explained by the Felkin-Anh model. Coupling of 37 with the tripeptide 38 using EDCI afforded the linear peptide 39, from which the dithiane part was converted to the carboxylic acid 40 via the corresponding aldehyde. Removal of the trichloroethoxycarbonyl (Troc) group followed by macrolactamization with O-(7-azabenzotriazol-1-yl)-N,N,N',N'-tetramethyluronium hexafluorophosphate (HATU)-diisopropylethylamine (DIPEA)^[20] afforded (4S,5R)-antillatoxin (1a), which was also not identical to natural antillatoxin. Thus, the White group also concluded that the proposed structure of natural antillatoxin should be revised.

--------------------------------------------------------------------------------
Scheme 6

--------------------------------------------------------------------------------

Loh and co-workers also reported their synthetic studies of antillatoxin.^[21] They developed indium-mediated allylation reactions of carbonyl compounds with allylic bromides. As shown in Scheme 7, the indium-mediated allylation of 42, prepared from pivalaldehyde (3) using aldol reactions twice, with the allylic bromide 41 smoothly proceeded in saturated ammonium chloride catalyzed by lanthanide triflate to give the adduct 43 as a mixture of syn and anti isomers (93:7). Reduction of the ester 43 afforded the alcohol 44, which was converted to the carbonate 45. Although the carbonate 45 was first reported to be transformed into the ,-unsaturated amide 46 by an insertion reaction of carbon monoxide followed by the coupling reaction with alanine methyl ester, the real product later proved to be the non-carbonylated one 47.21c

--------------------------------------------------------------------------------
Scheme 7

--------------------------------------------------------------------------------

Furthermore, the indium-mediated allylation reaction of the aldehyde 42 with the bromide 48 proceeded under analogous conditions to give a mixture of the homoallylic alcohols 49 in which the syn isomer 49a was the major one (syn:anti = 72:28).^[21d]

2.1.2. Total Synthesis of (4R,5R)-Antillatoxin (Natural Form) Having the Revised Structure
When unavailability of natural products precludes further structural studies by spectral and other methods, synthetic studies of the compounds having analogous structures will be the best choice for structure determination. This was the case for antillatoxin also. Thus, (4R,5R)-antillatoxin (1b) was selected as the next synthetic target,^[8b-d] which was proposed as the second possible configuration by the Gerwick group.^[6] Since the relationship between the C4 and C5 positions is anti, the anti-selective aldol reaction developed by Abiko and Masamune^[22] was employed to construct the required stereochemistry, as shown in Scheme 8. The aldehyde obtained from the dienyl alcohol 8 by its CMD oxidation was subjected to the anti-aldol reaction with the (E)-enolate generated from the propionate ester 50 of mesitylenesulfonyl (MesSO₂) norephedrine derivative using dicyclohexylboron triflate and triethylamine. After protection of the secondary alcohol with TESOTf, the resulting anti-aldol adduct 51 was converted to the primary alcohol 52 with DIBAL. The alcohol 52 was transformed into (4R,5R)-antillatoxin (1b) in the same way as developed in the synthesis of (4S,5R)-antillatoxin (1a), as summarized in Scheme 8. The synthetic (4R,5R)-antillatoxin (1b) was identical to the natural antillatoxin by comparison of the spectral details and optical rotation. Thus, the structure of natural antillatoxin was revised to be 1b by total synthesis.^[8b-d]

--------------------------------------------------------------------------------
Scheme 8

--------------------------------------------------------------------------------

In addition, (4R,5S)- and (4S,5S)-antillatoxins were analogously synthesized using the antipodal syn- or anti-selective aldol reactions as key steps.

2.1.3. Biological Activities of Antillatoxins
The natural antillatoxin has been shown to produce neuronal death, which is prevented by co-application of an NMDA receptor antagonist.^[7d] In addition, antillatoxin was revealed to be a novel activator of voltage-gated sodium channels (VGSC).^[7c]

Using four synthesized stereoisomers, (4R,5R)-, (4S,5R), (4S,5S)-, and (4R,5S)-antillatoxins, detailed biological evaluation was carried out in different biological assay systems: ichthyotoxicity to gold fish, microphysiometry using cerebeller granule cells (CGCs), lactose dehydrogenase (LDH) efflux from CGCs, monitoring of intracellular Ca²⁺ concentrations in CGCs, and cytotoxicity to Neuro 2a cells.^[7c] The natural antillatoxin, the (4R,5R)-isomer, was revealed to be greater than 25-fold more potent than any of the other stereoisomers.

2.2. Somamide A2.2.1. Total Synthesis
Somamide A was isolated by Gerwick and co-workers^[23] from assemblages of the marine cyanobacteria Lyngbya majuscula and Schizothrix sp. from the Fijian Island. Its structure was determined to be the 19-membered macrocyclic depsipeptide 58 having a 3-amino-6-hydroxy-2-piperidone (Ahp) unit, a (Z)-2-amino-2-butenoic acid (Abu) unit, and a sulfoxide function, as shown in Figure 3. The Ahp unit has been recently characterized as a constituent in more than 60 19-membered cyclic depsipeptides derived from cyanobacteria.^[5] They generally exhibit an interesting and significant inhibiting action against peptide proteases. The Ahp moiety may be biosynthetically derived from glutamate and probably plays an important role for protease inhibition because it may participate in converting the cyclic depsipeptides into a bioactive conformation due to the conformationally restricted structure and hydrogen bonding with the free hydroxyl group.

--------------------------------------------------------------------------------
Figure 3

--------------------------------------------------------------------------------

The first total synthesis of a cyclic depsipeptide possessing the Ahp unit was accomplished by Yokokawa, Inaizumi, and Shioiri ^[24] for micropeptin T-20 (59). This Ahp-depsipeptide was isolated from the cyanobacterium Microcyctis aeruginosa of freshwater origin.^[25] Its total synthesis revealed that the proposed structure 59 should be reexamined.

The total synthesis of somamide A (58) by Yokokawa and Shioiri ^[26] is the second example of the total synthesis of the Ahp-depsipeptides. The retrosynthetic plan is shown in Scheme 9, in which the cyclic precursor 60 was disconnected into three fragments 61-63. The precursor of the Ahp unit in 60 was the 2-amino-5-hydroxypentanoic acid unit, which was cyclized to the Ahp moiety at the last stage of the synthesis.

--------------------------------------------------------------------------------
Scheme 9

--------------------------------------------------------------------------------

As for the synthesis of the Abu-containing depsipeptide fragment 61, Martin's sulfurane (diphenyl bis(1,1,1,3,3,3-hexafluoro-2-phenyl-2-propyl)sulfurane)^[27] was employed to dehydrate the corresponding threonine derivative 67, which was straightforwardly prepared from HCl·H-Met-OMe (64) via the alcohol 65 and the depsipeptide 66, shown in Scheme 10. Dehydration of 67 with Martin's sulfurane afforded the (Z)-Abu derivative 61 in excellent yield. Application of Martin's sulfurane to the stereospecific dehydration of the amino acid derivatives was further developed by Yokokawa and Shioiri,^[28] which will be described later in section 2.2.2.

--------------------------------------------------------------------------------
Scheme 10

--------------------------------------------------------------------------------

The dipeptide 62 was prepared from Boc-N-MeTyr(TBDPS)-OTce (68) by acidic deprotection and then coupling with Boc-Phe-OH utilizing BopCl, ^[11] and the 2-amino-5-hydroxypentanoic acid derivative 63 was smoothly prepared from Aloc-Glu-OBzl (69) via 70 and 71, as shown in Scheme 11.

--------------------------------------------------------------------------------
Scheme 11

--------------------------------------------------------------------------------

Respective deprotection of 61 and 62 followed by coupling with DEPC afforded the depsipeptide 72, which, after acidic deprotection, was condensed with 63 and treated with TBSCl to give the linear peptide 73 having the full carbon skeleton of somamide A (58), as shown in Scheme 12. Simultaneous deprotection of both N- and C-terminal protective groups followed by macrolactamization with pentafluorophenyl diphenylphosphinate (FDPP, Ph₂P(O)OC₆F₅)^[29] afforded the cyclic depsipeptide 74.

--------------------------------------------------------------------------------
Scheme 12

--------------------------------------------------------------------------------

Construction of the Ahp unit in 58 was achieved by the method established in the total synthesis of micropeptin T-20.^[24] Thus, after removal of the TBS function, oxidation of the resulting alcohol 60 with IBX and then treatment with tetrabutylammonium fluoride (TBAF, Bu₄N⁺F^-) afforded somamide A (58) as a major product and the sulfide 75 as a minor one, shown in Scheme 13. The latter was transformed into 58 on standing in air or with hydrogen peroxide, which would suggest that the sulfide 75 is a true natural product and somamide A is an artifact.

--------------------------------------------------------------------------------
Scheme 13

--------------------------------------------------------------------------------

2.2.2. Stereospecific Dehydration of -Hydroxy--amino Acids Using Martin's Sulfurane
As described in section 2.2.1, Martin's sulfurane proved to be very effective for construction of (Z)-Abu. Further investigations^[28] revealed that threo-N-acyl--hydroxy--amino acid derivatives 76 afford (Z)-,-dehydroamino acids 77 using Martin's sulfurane, while erythro-N-acyl--hydroxy--amino acid amides 78 are converted to 4,5-trans-oxazolines 79 under analogous reaction conditions, as shown in Scheme 14. One noteworthy stereospecific feature of the method is demonstrated in the dehydration of the tripeptide 80 having both threo and erythro configurations to the oxazoline-Abu product 81.

--------------------------------------------------------------------------------
Scheme 14

--------------------------------------------------------------------------------

2.3. Kahalalide F
Kahalalides are a family of peptides bearing a variety of cyclic or acyclic skeletons isolated from the sacoglossan mollusc Elysia rufescens and the green alga Bryopsis sp.^[30] Among these peptides kahalalide F is the largest 19-membered cyclic depsipeptide, which is known to exhibit very interesting antitumor activity^[31] and is in phase II clinical trials for treatment of lung and prostate cancers and melanoma. The first assigned primary structure is consisted of mostly common amino acids except (Z)-Abu and 5-methylhexanoic acid. Further investigation of the absolute stereochemistry of kahalalide F led to two structures with different stereochemistry in Val (3) and Val (4) parts proposed independently by Scheuer et al.^[32] and Rinehardt et al.^[33] As described later, the discrepancy was unambiguously solved by the synthetic work of Giralt, Albericio, and co-workers.^[34] Although kahalalide B, having only common amino acids and 5-methylhexanoic acid, was also synthesized by the same workers,^[35] the following discussion will be limited to kahalalide F (82) due to the subject of this review (Figure 4).

--------------------------------------------------------------------------------
Figure 4

--------------------------------------------------------------------------------

Giralt, Albericio, and co-workers synthesized the two proposed structures because of the uncertainty regarding the stereochemistry at Val (3) and Val (4) and accomplished the synthesis and structural determination of kahalalide F. From the simplicity of the structure, their synthesis includes solid-phase synthesis of the linear precursors and subsequent macrocyclization of the off-resin peptides at the D-Val/Phe site. Construction of (Z)-Abu was carried out by stereoselective dehydration of the threonine precursor at a late stage of the synthesis (Scheme 15).

--------------------------------------------------------------------------------
Scheme 15

--------------------------------------------------------------------------------

The tetrapeptide resin 84 was synthesized from the Fmoc-D-Val-resin, prepared from the commercially available chlorotrityl chloride resin (83), by a sequential attachment of D-alloisoleucine, D-allothreonine, and D-alloisoleucine derivatives using Fmoc/t-Bu strategy and HATU/DIPEA as the coupling reagent. Ester linkage between 84 and Aloc-Val-OH was formed using diisopropylcarbodiimide (DIC) in the presence of DMAP. For chain elongation to the decapeptide 86 from 85, six amino acids were sequentially attached and capped with 5-methylhexanoic acid at the N-terminal. Construction of (Z)-Abu was carried out by two different methods: (1) after side-chain elongation from 86 to the tridecapeptide 87, stereoselective formation of the (Z)-Abu residue on the resin^[36] (method A) by Fukase's method^[37] using EDCI and cuprous chloride and (2) direct introduction of the dipeptide, Aloc-Phe-(Z)-Abu-OH, to 86 with HATU/DIPEA (method B). The Aloc group in 88 was deprotected with Pd(PPh₃)₄ and phenylsilane,^[38] and cleavage from the resin with TFA/CH₂Cl₂ (1:99) afforded the linear depsipeptide, which was subjected to macrocyclization using 1H-benzotriazol-1-yloxytripyrrolidinophosphonium hexafluorophosphate (PyBOP)-DIPEA^[39] to furnish, after removal of side-chain protection with TFA/H₂O (95:5), kahalalide F (82) in 10-14% yields. The synthetic kahalalide F and the diastereomer were compared by coeluting on HPLC with an authentic sample of kahalalide F, and the stereochemistry at the Val (3) and Val (4) residues was unambiguously determined as R and S, respectively. Interestingly, the biological activity of the diastereoisomer with (S)-Val (3) and (R)-Val (4) was 10 times less active than that of natural kahalalide F with (R)-Val (3) and (S)-Val (4).

2.4. Apratoxin A
Apratoxins A-C, isolated from cyanobacterial Lyngbya sp. collected at Guam^[40a] and Palau^[40b] in 2001, are highly functionalized cyclic depsipeptides that are known to exhibit potent cytotoxicity against KB cell and LoVo cancer cell (IC₅₀ = 0.52 and 0.36 nM). The pharmacological profiles of the unique peptides are unknown, and their elucidation is limited by their scarcity from marine origin. The structural feature of apratoxins shown in Figure 5 includes a hybrid structure from polypeptide and polyketide, which contains the 4-vinylthiazoline fused with novel 3,7-dihydroxy-2,5,8,8-tetramethylnonanoic acid.

--------------------------------------------------------------------------------
Figure 5

--------------------------------------------------------------------------------

The first total synthesis of apratoxin A (89) was accomplished by Chen and Forsyth.^[41] Their synthetic feature includes final macrocyclization at the Ile/Pro site, coupling of two segments, the ester 90 and the triamide 91, by thiol ester formation as the latent thiazoline, and thiazoline formation by a one-pot Staudinger reduction-intramolecular aza-Wittig (S-aW) process^[42] developed by them. It has been reported that the thiazoline part of 89 is susceptible to acid conditions and undergoes dehydration to give the less active apratoxin analogue, (E)-34,35-dehydroapratoxin A. ^[40b] It was therefore necessary to introduce a 4-vinylthiazoline moiety at a late stage of the synthesis.

Synthesis of the ester fragment 90, 3,7-dihydroxy-2,5,8,8-tetramethylnonanoic acid derivative with L-proline, was achieved in 12 steps, as shown in Scheme 16. The known alcohol 92, readily available by Brown's asymmetric allylation,^[43] was acylated with acrylic acid and N-methyl-2-chloropyridinium iodide,^[44] and the resulting ester 93 was cyclized by ring-closing metathesis (RCM)^[45] using the Grubbs catalyst. Introduction of the C37 methyl group to the ,-unsaturated lactone was performed by conjugate addition with a higher order methyl cuprate.^[46] Reductive cleavage of the lactone 94 with LiAlH₄ and TBS protection of the primary alcohol provided the secondary alcohol 95, which was subjected to esterification with Boc-Pro-OH by the Yamaguchi method,^[47] subsequent deprotection of the TBS group, and TPAP oxidation of the primary alcohol to afford the aldehyde 96. The anti-aldol part in 90 was constructed by Paterson's anti-selective aldol condensation^[48] from 96. The resulting aldol with the skeleton of the segment 90 was protected with TBSOTf, and Paterson's chiral auxiliary was cleaved in two steps to furnish the ester 90.

--------------------------------------------------------------------------------
Scheme 16

--------------------------------------------------------------------------------

The triamide segment 91 was prepared from Boc-MeIle-OMe by a sequential coupling of Boc-MeAla-OH, Boc-Tyr(Me)-OH, and the ,-unsaturated carboxylic acid 100 using (7-azabenzotriazole-1-yloxy)tripyrrolidinophosphonium hexafluorophosphate (PyAOP)/DIPEA^[49] and subsequent exchange of the protected primary hydroxyl group in 101 to the thiol one using the Mitsunobu method^[50] in three steps. The thiol-containing ,-unsaturated acid residue in 91 served as a precursor to the vinylogous cysteine residue at a later stage (Scheme 17).

--------------------------------------------------------------------------------
Scheme 17

--------------------------------------------------------------------------------

The thiol ester 102 was efficiently formed from the two segments 90 and 91 using DPPA and triethylamine according to the method developed by Yokoyama, Shioiri, and Yamada.^[51] For introduction of the nitrogen function, the PMB ether in 102 was deprotected with DDQ and subsequent one-step azidation^[52] of the allylic alcohol was effective by employing DPPA again together with triphenylphosphine and diisopropyl azadicarboxylate (DIAD) to afford the azide 103 in 90% yield, Scheme 18.

--------------------------------------------------------------------------------
Scheme 18

--------------------------------------------------------------------------------

The labile thiazoline was efficiently constructed using the S-aW process developed^[42] using triphenylphosphine after exchange of the TBS protection to the TES one for the solution of final difficult deprotection. The Boc group of 104 with acid-labile 2-hydroxyethyl-4-vinylthiazoline moiety was carefully deprotected in two steps by conversion to the TBS urethane using TBSOTf and its cleavage with TBAF according to the Ohfune's procedure.^[53] After saponification with lithium hydroxide, macrocyclization of the free peptide with PyAOP/DIPEA smoothly proceeded at room temperature for 2 h to give, after careful TES deprotection with HF-acetonitrile, apratoxin A (89) in 47% yield (three steps). Synthetic apratoxin A was identical in all respects with spectroscopic data provided for the natural substance, Scheme 19.

--------------------------------------------------------------------------------
Scheme 19

--------------------------------------------------------------------------------

The oxazoline-containing analogue of apratoxin A was synthesized by Ma and co-workers.^[54] In the synthesis, preparation of the polyketide fragment was accomplished by a very similar approach to that described above (Forsyth's method). Very recently the polyketide fragment was synthesized by Xu and co-workers.^[55]

2.5. Lyngbyabellins A and B
As described in sections 2.1, 2.2, and 2.4, Lyngbya majuscula is a rich source of biologically active secondary metabolites, and lyngbyabellins A and B were also isolated from this cyanobacterium.^[56] These cyclic depsipeptides exhibit attractive cytotoxic properties against the human cancer cell lines. Their structures 105 and 106 are closely related each other, and lyngbyabellin A (105) has two thiazole rings, while one of the thiazole rings is replaced by a thiazoline ring in lyngbyabellin B (106), shown in Figure 6. These show a resemblance to the structure of dolabellin (107), a metabolite isolated from the sea hare Dolabella auricularia.^[57] This striking structural relationship will support a cyanobacterial origin for dolabellin.

--------------------------------------------------------------------------------
Figure 6

--------------------------------------------------------------------------------

Lynbyabellins A and B were synthesized by Yokokawa, Shioiri, and co-workers.^[58] In their strategy the molecules were disconnected at the peptide and ester bonds, and macrolactamization was accomplished by activation of the carboxyl group at the C-terminal peptide where no epimerization occurred. Construction of the thiazoline ring of 106 was postponed to the final stage of the synthesis because of its facile racemization.

The required thiazole amino acids 111 were prepared using the method established by Hamada and Shioiri^[59] from the corresponding amino acids 108 through (1) transformation into the Weinreb amides 109, (2) reduction with LiAlH4, (3) condensation with cysteine methyl ester, and (4) dehydrogenation with CMD, as shown in Scheme 20.

--------------------------------------------------------------------------------
Scheme 20

--------------------------------------------------------------------------------

The ,-dihydroxy thiazole 118 was prepared from the thiazole 113, which was obtained by condensation of the aldehyde 112 with cysteine methyl ester followed by the CMD oxidation. Although the thiazole 113 was prepared in only two steps, the yield was miserable because of the instability of the intermediate thiazolidine, shown in Scheme 21. Alternatively, preparation of the thiazole 113 was accomplished from Fmoc-S-trityl cysteine (114) through (1) methyl esterification using trimethylsilyldiazomethane (TMSCHN₂, Me₃SiCHN₂),^[60] (2) deprotection of the Fmoc (9-Fluorenylmethoxycarbonyl) group, and (3) coupling with 3-methylcrotonic acid. The resulting cysteine amide 115 underwent titanium(IV)-mediated tandem deprotection-dehydrocyclization^[61] to give the thiazoline 116, which was dehydrogenated with 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU)/BrCCl₃^[62] to produce the thiazole 113. After replacement of the methyl ester of 113 with the 2-trimethylsilylethyl (TMSE) one, the Sharpless asymmetric dihydroxylation^[63] using AD-mix- afforded the ,-dihydroxy thiazole 118 with 91% enantiomeric excess (ee).

--------------------------------------------------------------------------------
Scheme 21

--------------------------------------------------------------------------------

The dichlorinated -hydroxy acid fragment 123 was constructed with excellent stereoselectivity (97% ee) by enantioselective aldol reaction of the ketene acetal 119 with the aldehyde 120 using the chiral oxazaborolidine 121 developed by Kiyooka,^[64] followed by replacement of the ester group of 122, shown in Scheme 22.

--------------------------------------------------------------------------------
Scheme 22

--------------------------------------------------------------------------------

2.5.1. Total Synthesis of Lyngbyabellin A
The total synthesis of lyngbyabellin A (105) was started by dicyclohexylcarbodiimide (DCC)-mediated coupling of the dichloro ester 123 with the dipeptide 124, prepared from the thiazole 111a and Boc-Gly-OH, shown in Scheme 23. After deprotection of the allyl group from the ester 125, coupling with the diol 118 using DCC afforded the linear precursor 126. After deprotection, the macrolactamization was achieved under high dilution conditions using DPPA in the presence of sodium hydrogen carbonate to give lyngbyabellin A (105).

--------------------------------------------------------------------------------
Scheme 23

--------------------------------------------------------------------------------

2.5.2. Total Synthesis of Lyngbyabellin B
Instead of the thiazole ring in lyngbyabellin A (105), lyngbyabellin B (106) has a thiazoline ring which will readily undergo racemization during synthesis and handling. Hence, construction of the thiazoline ring was postponed to the later stage of synthesis and Wipf's oxazoline-thiazoline interconversion protocol^[65] was employed.

The remaining two building blocks for the preparation of lyngbyabellin B (106) were the diol 129 and the dipeptide 131, which were, respectively, prepared using the Sharpless asymmetric dihydroxylation and the DEPC coupling, as shown in Scheme 24.

--------------------------------------------------------------------------------
Scheme 24

--------------------------------------------------------------------------------

The thiazole 111b was hydrolyzed and condensed with the dichlorohydroxy ester 123 using DCC to give the ester 132. After removal of the allyl group from 132, the resulting carboxylic acid sluggishly underwent coupling with the dihydroxy ester 129, and it was necessary to activate the carboxyl group with 2,4,6-trichlorobenzoyl chloride^[47] before coupling. The coupling product 133 was connected with the dipeptide 131 using the DEPC method after deprotection to give the linear peptide 134. Removal of the protective groups at both N- and C- terminals and then macrolactamization with FDPP afforded the macrocycle 135 (Scheme 25).

--------------------------------------------------------------------------------
Scheme 25

--------------------------------------------------------------------------------

Final formation of the thiazoline ring was accomplished by Wipf's oxazoline-thiazoline interconversion protocol,^[65] as shown in Scheme 26. Thus, treatment of 135 with TBAF and then (diethylamino)sulfur trifluoride (DAST, (Et₂N)S⁺F₃^-) afforded the oxazoline 137. After thiolysis of the oxazoline 137, the resulting thioamide 138 was treated with DAST to give lyngbyabellin B (106) in excellent yield.

--------------------------------------------------------------------------------
Scheme 26

--------------------------------------------------------------------------------

2.6. Dendroamide A
As already shown there are many marine cyclic peptides having oxazole, oxazoline, thiazole, and thiazoline amino acids as their constituents.^3e Dendroamide A (139) is also a modified cyclic peptide having the oxazole and thiazole rings and was isolated from the terrestrial cyanobacterium Stigonema dendroideum fremy on the basis of its ability to reverse drug resistance in tumor cells that overexpress either of the transport proteins, P-glycoprotein or MRP 1.^[66] Because of this multidrug resistance reversing activity, dendroamide A may serve as the starting point for the synthesis of a variety of analogues for structure-biological activity analyses (Figure 7).

--------------------------------------------------------------------------------
Figure 7

--------------------------------------------------------------------------------

The first synthesis of dendroamide A (139) was achieved by Bertram and Pattenden.^[67] Further total syntheses were reported by the groups of Smith,^[68] Kelly,^[69] and Shin.^[70]

Bertram and Pattenden^[67] first used a linear approach for the total synthesis of 139. The required oxazole amino acid 142 was prepared from Z-D-Ala-DL-Thr-OMe (140) by oxidation with Dess-Martin periodinane,^[71] cyclization with triphenylphosphine-iodine in the presence of triethylamine,^[72] and then deprotection of the Z group, as shown in Scheme 27. The thiazole derivative 144 was prepared through a modified Hantzsch method^[73] by condensation of ethyl bromopyruvate with the thioamide 143 of alanine. Utilizing these oxazole and thiazole derivatives, sequential couplings of 142 and 144, 145 and 146 with EDCI-N-hydroxybenzotriazole (HOBt) in the presence of N-methylmorpholine (NMM) afforded the linear precursor 147. Deprotection followed by macrolactamization with FDPP/DIPEA afforded dendroamide A (139) in 91% yield.

--------------------------------------------------------------------------------
Scheme 27

--------------------------------------------------------------------------------

Next, Bertram and Pattenden applied an interesting method of metal-templated assemblies developed by them to the one-pot synthesis of 139.^[67] A 1:1:1 mixture of the heterocyclic amino acids 149, 150, and 151, respectively, derived from 142, 144, and 146, underwent the metal-templating reactions by treatment with various metal tetrafluoroborate salts in the presence of DIPEA and then FDPP, giving a mixture of dendroamide A (139) and its analogues such as 152. When no metal salts were used the yield was 75% and the percentage composition of cyclic products was 23:29 for 139 and 152. Although use of AgBF₄ afforded 152 only, Ca(BF₄)₂ was found to act as a better template in the assembly of trimeric cyclic peptides to raise the proportion of 139 to 52% with only 23% of analogue 152 (Scheme 28).

--------------------------------------------------------------------------------
Scheme 28

--------------------------------------------------------------------------------

As already pointed out earlier, one of the most important strategies for the preparation of macrocyclic peptides is selection of the most suitable point for the cyclization of linear precursors. Prior to the synthesis of dendroamide A Xia and Smith first determined the lowest energy conformation of each of the three possible acyl-azide linear precursors (see the disconnection A-C in Figure 7) using both MM2 and MOPAC modeling.^[68] Disconnection B was selected as the most suitable point because the distance between the N-terminal nitrogen and the C-terminal carbon was revealed to be the closest by calculation.

Preparation of the required oxazole amino acid 155 was carried out from the dipeptide, Boc-D-Ala-L-Thr-OMe (153), by dehydration with Burgess reagent (Et₃N⁺SO₂N-CO₂Me)^[74] and then oxidation of the epimeric oxazoline mixture 154 with bromotrichloromethane and DBU,⁶² as shown in Scheme 29. The chiral thiazole amino acids 156 and 157 were prepared using a modified Hantzsch reaction^[73] just like Scheme 27 (from 143 to 144). Construction of the full carbon skeleton of 139 started from deprotection of the thiazoles 156 and 157. The dipeptide 158 obtained using DIC and HOBt was deprotected and coupled with the oxazole carboxylic acid 155 to give the linear precursor 159. Deprotection at the C- and N-terminals followed by macrocyclization with DPPA afforded dendromide A (139) in 56% yield together with its conformational isomer in 18% yield.

--------------------------------------------------------------------------------
Scheme 29

--------------------------------------------------------------------------------

Highlights of the synthesis of dendroamide A (139) by You and Kelly^[69] will be application of bis(triphenyl)oxodiphosphonium trifluoromethanesulfonate^[75] to the preparation of the oxazole structure from a -ketodipeptide and the thiazolines from fully protected cysteine-containing dipeptides.

Synthesis of the oxazole amino acid 162 is outlined in Scheme 30. The -ketodipeptide 161 prepared by the Dess-Martin oxidation of 160 smoothly afforded the oxazole 162 by treatment with bis(triphenyl)oxodiphosphonium trifluoromethanesulfonate generated from triphenylphosphine oxide and triflic anhydride. The triphenylphosphine-iodine reagent might be used in this case^[67]67 like the oxobisphosphonium salt but requires use of a base, increasing the risk of racemization.^[72]

--------------------------------------------------------------------------------
Scheme 30

--------------------------------------------------------------------------------

Synthesis of the thiazole amino acids 165a and 165b, respectively, started from 163a and 163b, which were easily transformed into the thiazolines 164a and 164b, respectively, utilizing bis(triphenyl)oxodiphosphonium trifluoromethanesulfonate. Oxidation of 164a and 164b with active manganese dioxide afforded the required thiazoles 165a and 165b, respectively.

Synthesis of the linear precursor 167 proceeded from three heterocyclic amino acids 162, 165a, and 165b. Removal of the Fmoc group of 165a was carried out with diethylamine, while deprotection of the allyl group of 165b was accomplished utilizing a palladium catalyst, generated from palladium acetate and polystyrene-supported triphenylphosphine (PS-Ph-PPh2) in the presence of phenylsilane.^[76] Coupling with O-benzotriazol-1-yl-N,N,N',N'-tetramethyluronium hexafluorophosphate (HBTU)/HOBt/DIPEA, deprotection from the product 166 with diethylamine, and then coupling with the carboxylic acid derived from oxazole 162 yielded the linear precursor 167. After deprotection, final cyclization with PyBOP/DMAP was achieved by slowly adding the deprotected linear peptide to a solution of PyBOP/DMAP in CH₂Cl₂/DMF (2/1) with a syringe pump over 8 h to give dendroamide A (139) in good yield. These reaction conditions are analogous to high-dilution conditions essential for macrocyclization.

The key step for the preparation of dendroamide A (139) by Yonezawa, Tani, and Shin is formation of the oxazole and thiazole rings from dehydropeptides.^[70] The starting dehydrodipeptide 169, prepared from Boc-L-Ser-D-Ala-OMe (168), was converted to the bromide 170 with N-bromosuccinimide (NBS), and then treatment with TFA afforded the bromo pyruvate derivative 171. The Hantzsch condensation of 171 with the thioamide 172 produced the thiazole peptide 173, which was converted to the corresponding thioamide 174 according to the usual procedure, shown in Scheme 31.

--------------------------------------------------------------------------------
Scheme 31

--------------------------------------------------------------------------------

Preparation of the required oxazole fragment 182 was started by deprotection of the Boc group of 175a and then coupling with the serine derivative 176 using benzotriazol-1-yloxytris(dimethylamino)phosphonium hexafluorophosphate (BOP), as shown in Scheme 32. The oxazole ring was formed by treatment of 177a with NBS and then cesium carbonate. Removal of the isopropylidene group from the resulting oxazole peptide 178 followed by dehydration afforded the dehydropeptide 180, which was treated with NBS in methanol and then TFA to give the bromide 182 via 181.

--------------------------------------------------------------------------------
Scheme 32

--------------------------------------------------------------------------------

Condensation of the thioamide 174 with the bromide 182 using potassium hydrogen carbonate afforded the N,O-diprotected linear precursor 183, which after deprotection was cyclized with BOP and DIPEA to give dendroamide A (139) in 57% yield (Scheme 33).

--------------------------------------------------------------------------------
Scheme 33

--------------------------------------------------------------------------------

3. Cyclic Peptides from Sponges
3.1. Microsclerodermin E
Microsclerodermins A-I, isolated from both Theonella sp. and Microscleroderma sp., are a growing family of 23-membered cyclic peptides that are known to exhibit antifungal activity as well as cytotoxicity.^[77] The significant features common to all members of these molecules include six amino acid residues, four of which ((3R)-4-amino-3-hydroxybutyric acid (GABOB), a modified tryptophan, an unusual 3-aminopyrrolidone-4-acetic acid (PyrAA), and aromatic 3-amino-2,4,5-trihydroxy acids (AETD, AMMTD, etc.)) are uncommon components. There are slight variations on the tryptophan, the 3-aminopyrrolidinone-4-acetic acid (PyrAA), and aromatic 3-amino-2,4,5-trihydroxy acids. To date, synthetic studies on microsclerodermin A have been reported by Sasaki, Hamada, and Shioiri,^[78] and recently, the first total synthesis of microsclerodermin E (184) was achieved by Zhu and Ma.^[79] In this section the discussion will be limited to Ma's synthesis because the works of Hamada and Shioiri were complied in their review3a (Figure 8).

--------------------------------------------------------------------------------
Figure 8

--------------------------------------------------------------------------------

Microsclerodermine E (184) has Trp(CO₂H), dehydroPyrAA, and 3-amino-10-(p-ethoxyphenyl)-2,4,5-trihydroxydeca-7,9-dienoic acid (AETD) as unusual components. Synthesis of microsclerodermin E was accomplished by [4 + 2] segment strategy including the segment coupling of the tetrapeptide 185 and the dipeptide 186 at AETD/dehydroPyrAA site and macrocyclization at Gly/GABOB site.

AETD was constructed from commercially available -gluconolactone with four asymmetric centers using Julia coupling as a key step, shown in Scheme 34. -Gluconolactone was first converted to the epoxide 187 by a sequence of reactions including protection with 2,2-dimethoxypropane and p-toluenesulfonic acid (TsOH), reduction of the ester, activation of the primary alcohol with TsCl, and treatment with potassium carbonate. The resulting epoxide 187 underwent an epoxide-opening reaction with the lithium anion generated from methyl phenyl sulfone in the presence of boron trifluoride etherate^[80] to produce the sulfone 188. For introduction of the nitrogen function, the 4,5-6,7-bisacetonide 188 was subjected to deprotection with hydrochloric acid followed by regioselective reprotection with 2,2-dimethoxypropane and TsOH to yield the isomeric 3,4-6,7-bisacetonide 189 in 67% yield. Transformation of the hydroxyl group to the amino one was achieved in four steps by mesylation of 189, azidation of the mesylate with sodium azide, hydrogenation of the azide to the amine, and N-protection with trifluoroacetic anhydride to produce the trifluoroacetamide 190. The chain extension was carried out by a three-step Julia coupling with the anion generated from 190 with p-ethoxycinnamaldehyde. Benzoylation of the resulting alcohol followed by treatment with sodium amalgam in methanol provided the aromatic diene 191 with the skeleton of AETD together with the corresponding (Z)-isomer in a 5:1 ratio. The diene 191 was transformed to the AETD derivative 192 by a sequence of selective deprotection and selective protection followed by reductive deprotection^[81] of the N-trifluoroacetyl group.

--------------------------------------------------------------------------------
Scheme 34

--------------------------------------------------------------------------------

Synthesis of the dipeptide 197 commenced with preparation of the GABOB derivative 195 from the known diol 193 in five steps. The diol 193 underwent selective tosylation of the primary hydroxyl group and subsequent azidation using sodium azide in DMF to produce the GABOB skeleton, which was converted to the active ester 195 by TBS protection of the secondary alcohol followed by replacement of the methyl ester with the OSu ester. Coupling of 195 with the AETD derivative 192 was conducted under refluxing conditions to provide the dipeptide 196, which was converted to the dipeptide active ester 197 for the next segment coupling by Dess-Martin oxidation of the primary hydroxyl group, sodium chlorite oxidation to the carboxylic acid, and subsequent active ester formation using EDCI/HOSu (Scheme 35).

--------------------------------------------------------------------------------
Scheme 35

--------------------------------------------------------------------------------

The dehydroPyrAA 202 was prepared from aspartic acid, as shown in Scheme 36. Boc-Asp(OBzl)-OH was converted to the diBoc derivative 198 by allyl ester formation, N-tert-butoxycarbonylation with di-tert-butyl dicarbonate (Boc₂O) in the presence of DMAP, and subsequent deprotection of the allyl ester using Wilkinson's catalyst. Treatment of 198 with thionyl chloride and pyridine in DMF provided the N-Boc-N-carbonic anhydride 199, which reacted with the lithium enolate generated from trimethylsilylethyl acetate to afford the enantiomerically pure -ketoester 200a without any racemization. In the case of employing known methods using the imidazolide and the pentafluorophenyl active ester, considerable racemization was observed. A different strategy using the N-carbonic anhydride^[82] was therefore required for construction of the aspartic acid -ketoester. For construction of the pyrrolidinone ring, 200a was hydrogenolyzed, and the resulting carboxylic acid 200b was converted to the carboxamide by treatment with ammonia via the mixed anhydride, which directly cyclized to the hydroxypyrrolidinone 201. Mesylation of 201 was spontaneously accompanied by dehydration to form the dehydropyrrolidinone amino acid (dehydroPyrAA) 202a. In the course of conversion to the active ester 202b 202a was found to be susceptible to basic conditions due to the unfavorable racemization. Deprotection of the TMSE ester with TBAF caused considerable racemization. The additive TsOH together with TBAF was a better choice and minimized racemization to a 4:1 R/S ratio.

--------------------------------------------------------------------------------
Scheme 36

--------------------------------------------------------------------------------

For preparation of the tryptophan derivative, introduction of the cyano group at the C2 position was performed according to Danishefsky's procedure^[83] from N,N-dibenzyltryptophan methyl ester (203), shown in Scheme 37. Chlorination of 203 with tert-butyl hypochlorite and subsequent cyanation with trimethylsilyl cyanide in the presence of boron trifluoride etherate gave the 2-cyanotryptophan 204. Hydrolysis of the cyano group in 204, esterification of the resulting carboxylic acid, and subsequent hydrogenolysis of the N,N-dibenzyl group with Pearlman catalyst afforded the Trp(CO₂Me) derivative 205.

--------------------------------------------------------------------------------
Scheme 37

--------------------------------------------------------------------------------

Trp(CO₂Me) 205 was coupled with Boc-Sar-OH, and the resulting dipeptide was saponified and condensed with H-Gly-OTMSE using EDCI/HOBt, as shown in Scheme 38. The tripeptide 207, after treatment with TFA, was then elongated by coupling with the dehydroPyrAA-activated ester 202b to yield the tetrapeptide 208, which after deprotection was again condensed with the dipeptide segment active ester 197 to give the linear precursor 209 for microsclerodermin E.

--------------------------------------------------------------------------------
Scheme 38

--------------------------------------------------------------------------------

For final assembly to microsclerodermin E the ester function in 209 was deprotected with TBAF and the azide was reduced to the amine with trimethylphosphine^[84] to provide the free-linear precursor, which was subjected to macrocyclization with DPPA for 14 days to furnish microsclerodermin E methyl ester in 40% yield together with 9% yield of the separable diastereomer at the dehydroPyrAA residue, as summarized in Scheme 39. Saponification at the Trp(CO₂Me) part and then mild removal of the MOM group from the acid-labile protected microsclerodermin with Amberlyst-15 resin^[85] gave microsclerodermin E (184), which was identical in all respects with spectroscopic data provided for the natural substance. In the reduction of the azide function it is notable that standard reagents, Ph₃P, n-Bu₃P, and SnCl₂, are ineffective for this case and the only sterically less hindered reagent, Me₃P, gives the desired product. In addition, use of EDCI/1-hydroxy-7-azabenzotriazole (HOAt) in the macrocyclization was ineffective and the shorter reaction time in the macrocyclization using even DPPA resulted in a much lower yield.

--------------------------------------------------------------------------------
Scheme 39

--------------------------------------------------------------------------------

3.2. Phakellistatins
Phakellistatin 3 (211) and isophakellistatin 3 (212) were isolated from the Western Indian ocean sponge Phakellia carteri.^[86] Phakellistatin 3 was shown to inhibit P388 leukemia cell growth (ED₅₀ = 0.4M), while the diastereoisomer, isophakellistatin 3, was inactive. Both of these cyclic peptides contain the 3a-hydroxypyrrolidino[2,3-b]indoline (Hpi) moiety, which will be biosynthetically derived from the tryptophan moiety by oxidation. In fact, phakellistatin 13 (210) containing the Trp residue was isolated from the sponge Phakellia fusca Thiele, collected at Yongxing Island in China.^[87] Phakellistatin 13 was also cytotoxic against the human hepatoma BEL-7404 cell line with an ED₅₀< 10^-2 g/mL. As shown in Scheme 40 its structure 210 is identical to those of 211 and 212 except for the Trp residue.

--------------------------------------------------------------------------------
Scheme 40

--------------------------------------------------------------------------------

The unique peptides 211 and 212 having the Hpi residue were synthesized from 210 by Greenman, Hach, and Van Vranken^[88] just like the biosynthetic pathway. The linear precursor heptapeptides 213 and 214 were synthesized on chlorotrityl resin from Fmoc-protected amino acids using HBTU and cleaved from resin with TFA-1,2-ethanedithiol. Macrocyclization of 213 was carried out using HBTU/HOBt/DIPEA to give the cyclic lactam in 60% yield, from which phakellistatin 13 (210) was obtained by acidic treatment. In contrast, the efficiency for macrocyclization of the hexapeptide 214 proved to be poor (16%). This synthesis establishes the undetermined absolute configuration of the Trp moiety in 210 to be L.

Utilizing the synthesized phakellistatin 13 (210), photooxidation with rose bengal as a sensitizer^[89] afforded a 1:1 mixture of phakellistatin 3 (211) and isophakellistatin 3 (212) in a combined yield of 20%. Multiple rounds of HPLC purification produced phakellistatin 3 (211) with over 95% purity, but isophakellistatin 3 (212) was obtained in 85% purity along with 15% phakellistatin 3 (211). Although the efficiency of the oxidation should be much improved, this work demonstrates for the first time that a tryptophan residue can be directly converted to the corresponding 3a-hydroxypyrrolo[2,3-b]indoline in a full-length peptide. Further, spectroscopic and molecular modeling studies revealed a similar backbone conformation for all three cyclic peptides despite their markedly different biological activity against tumor cell lines.

Before the above synthetic work phakellistatins 1, 2, 5, 10, and 11, shown in Figure 9, isolated from the marine sponges of genus Phakellia, were synthesized in a similar manner. ^[90] Although these phakellistatins were reported to show a moderate cytotoxicity, the synthetic specimens were found to be chemically but not biologically (cancer cell lines) identical to the natural products. The reason for this discrepancy might be a conformational difference, especially around the proline residue, or more likely the presence of a trace amount of a highly active antineoplastic agent that noncovalently binds to the natural cyclic peptides. The results clearly indicated the importance of the total syntheses for confirmation of the biological activity of the natural products.

Figure 9

In 1990 Fusetani and co-workers reported isolation of cyclotheonamides A and B ^[91a] from a sponge of the genus Theonella, and then they isolated closely related congeners C-E ^[91b] in 1998. The basic structural motif of the cyclotheonamides consists of a 19-membered cyclic peptide constructed from five amino acid residues, two of which (vinylogous tyrosine and -ketoarginine) are common, unique components in the family. These unique cyclic peptides attracted considerable attention immediately for interesting inhibitory activity to serine proteases. Syntheses of cyclotheonamides A and B have been accomplished by several groups 92 and compiled as a review. 3e The following discussion will be limited to the most recent work, 93 the synthesis of cyclotheonamides E 2 and E 3 ( 215a and 215b) , which differ from counterparts A and B in the presence of D -alloisoleucine residue in place of D -phenylalanine and the side-chain containing benzolylalanine and isovalerylalanine residues (Figure 10).

Figure 10

Wasserman and Zhang achieved the synthesis of cyclotheonamides E₂( 215a ) and E₃( 215b )^[93] by a [3 + 2] segment coupling strategy, which contains efficient construction of the arginine -ketoacid using the cyano ylide activation method ^[94] developed by themselves. The tripeptide 221 was prepared by reverse stepwise elongation from the N -terminal diaminopropanoic acid 217 , which was synthesized according to the Izumiya's method ^[95] from Me₃Si(CH₂)₂OCO (Teoc)-asparagine ( 216 ), as shown in Scheme 41. After formation of the dipeptide with H-Pro-OBzl using EDCI/HOBt and subsequent removal of the benzyl ester, the resulting dipeptide 218 was coupled with the acyl cyano ylide 220 , which was prepared from Boc-Arg(diCbz)-OH ( 219 ) by C -acylation of cyanomethylidenephosphorane using EDCI/DMAP, to yield the tripeptide 221 .

Scheme 41

On the other hand, Boc-Tyr(TIPS)-al ( 222 ) was olefinated with a Wittig reagent to give the vinylogous tyrosine ( 223 ), shown in Scheme 42. After Boc deprotection, coupling with Boc- D - a Ile-OH using EDCI/HOBt followed by acidic deprotection afforded 224 .

Scheme 42

The tripeptide 221 was exposed to ozone for activation of the cyano ylide, and the resulting acyl cyanide was directly coupled with the dipeptide 224 to provide in good yield the linear precursor 225 , shown in Scheme 43. The allyl function of 225 was replaced with the pentafluorophenyl ester.

Scheme 43

The linear precursor 226 thus obtained was selectively deprotected with hydrogen chloride to afford the free pentapeptide, which was subjected to cyclization in the presence of DMAP and sodium hydrogen carbonate to produce the cyclic peptide 227 in 61% yield. After cleavage of the Teoc group, coupling with N -benzolylalanine and N- isovalerylalanine using EDCI/HOBt followed by TIPS and Cbz deprotection afforded cyclotheonamides E₂( 215a ) and E₃( 215b ), respectively (Scheme 44).

Scheme 44

Papuamides A ( 229 ) and B are a new family of novel cyclic depsipeptides isolated from the marine sponge genus Theonella collected at Papua New Guinea by Boyd et al. ^[96] Papuamides are known to strongly inhibit the infection of human T-lymphoblastanoid cells by HIV-1 RF and also exhibit potent cytotoxicity against a number of human cancer cell lines. These cyclic heptadepsipeptides have a unique structure containing (4 Z ,6 E )-2,3-dihydroxy-2,6,8-trimethyldecadienoic acid (Dhtda) and unusual amino acid residues, such as (3 S ,4 R )-3,4-dimethylglutamine (3,4-DiMeGln), (2 R , 3R )-3-hydroxyleucine (3-OHLeu), and -methoxytyrosine ( -OMeTyr), as shown in Figure 11. The stereochemistry of papuamides remains to be determined because of the uncertainty regarding the stereochemistry in the -OMeTyr and the Dhtda parts. Interestingly, two unusual amino acids, -OMeTyr and 3,4-DiMeGln, are known as common components of the cyclic depsipeptide callipeltin A, ^[97] which shows anti-HIV and antifungal activities. Although total synthesis of the papuamides has been not accomplished yet, the unique structure as well as interesting biological activities of these compounds made considerable efforts directed toward total synthesis of papuamides. ^[98] The discussion in this section will be limited to synthetic efforts of Hamada and co-workers. ^[99]

Figure 11

To accomplish total synthesis as well as structural determination of papuamides, determination of two stereo-undefined components, MeOTyr and Dhtda, is necessary. Boyd and co-workers reported the interesting experiment in which the hydrolysate, H-MeThr-MeOTyr-Hpr-OH ( 230 ) (Hpr: pipecolic acid), derived from solvolysis of papuamide A with triethylamine in methanol shows the anomalous high-field shift (0.32 ppm, doublet) at the methyl proton of the MeThr residue in the NMR spectrum (Figure 11). ^[96] The experiment encouraged Hamada and co-workers to synthesize the four tripeptides with different stereochemistry in the MeOTyr residue. ^[99] The required four diastereomers of the methoxytyrosine were prepared from ( S )- and ( R )-Garner aldehydes ^[100] by a concise route, as shown in Scheme 45. ^[98d] Addition of the benzyloxyphenyllithium to the ( S )-Garner aldehyde in the presence of lithium bromide gave the protected hydroxyamino alcohol 231 in moderate yield with a 1:3 ratio of syn and anti. The anti-rich crude product was directly crystallized to provide pure (2 S ,3 R )- 231a , whose stereochemistry was deduced by the Felkin-Ahn model and clearly determined by the NOE experiment. On the other hand, (2 S ,3 S )- 231b was produced by racemization-free oxidation of a mixture of 231a and 231b with Dess-Martin periodinane and subsequent syn-selective reduction with K-selectride. The latter was sluggish and proved problematic for easy racemization, even at -78 ℃ . One recrystallization of the crude product, however, gave enantiomerically pure (2 S ,3 S )- 231b . Conversion of (2 S ,3 R )- 231a to Boc-(2 S ,3 R )-MeOTyr(Bzl)-OH ( 232a ) was performed by standard manipulation containing O -methylation with sodium hydride and iodomethane, selective removal of the acetonide with TsOH, Parikh-Doering oxidation to the aldehyde, and final reoxidation to the carboxylic acid 232a with sodium chlorite. The diastereomer (2 S ,3 S )- 231b was transformed to Boc-(2 S ,3 S )-MeOTyr(Bzl)-OH ( 232b ) in a similar fashion. The remaining diastereomers, 232c and 232d , were synthesized from the ( R )-Garner aldehyde with similar stereoselectivities.

Scheme 45

The four tripeptides 230a - d required for structural determination of MeOTyr were prepared from Boc-( S )-Hpr-OBzl by stepwise coupling with Boc-MeOTyr(Bzl)-OH and Cbz-MeThr(TBS)-OH using the HATU/DIPEA and EDCI/HOBt methods, shown in Scheme 46. Sequential deprotection of the TBS group with HF-acetonitrile and the Cbz group by hydrogenolysis furnished the free tripeptides 230a - d . Comparisons of the ¹H NMR spectrum of the hydrolysate from natural papuamide A with those of the synthetic four tripeptides 230a - d were consistent with that of the (2 R ,3 R )-isomer 230a . The chemical shifts of the other three diastereomers at the methyl proton of the MeThr residue differed largely from the 0.32 ppm of the naturally derived one. Accordingly, the stereochemistry of the MeOTyr residue in the papuamide A was unambiguously established as 2 R ,3 R . ^[99]

Scheme 46

For efficiently stereoselective production of (2 R ,3 R )-MeOTyr, new Ir-catalyzed asymmetric hydrogenation accompanied by dynamic kinetic resolution was developed. ^[101] Readily available 2-amino-3-keto ester hydrochloride 238 , prepared from glycine methyl ester benzophenoneimine ( 236 ) by acylation with p -benzyloxybenzoyl chloride ( 237 ) in the presence of KO t -Bu followed by acid hydrolysis, was hydrogenated with the Ir catalyst, prepared from [IrClcod] 2 and ( R )-MeOBIPHEP, under 100 atm of hydrogen. After Boc protection, the (2 R ,3 R )-3-hydroxytyrosine derivative 239 was obtained in a >99:1 diastereomeric ratio with 95% ee, as shown in Scheme 47. The reaction is noteworthy because this demonstrates the first example of Ir-catalyzed anti-selective hydrogenation of 2-amino-3-keto ester hydrochlorides with high diastereo- and enantioselectivity through dynamic kinetic resolution. O -Methylation of 239 with trimethyl oxonium tetrafluoroborate and proton sponge followed by saponification of the methyl ester provided the Boc-(2 R ,3 R )-MeOTyr(Bzl)-OH 232a .

Scheme 47

With the stereo-defined methoxytryrosine available, the cyclic depsipeptide skeleton in papuamide B was constructed from (2 R ,3 R )-3-hydroxyleucine benzyl ester 241 , ^[97] which was stereoselectively and efficiently synthesized by asymmetric hydrogenation of the 2-amino-3-keto ester hydrochloride 240 through dynamic kinetic resolution using Ru-( S )-BINAP catalyst developed by Hamada and co-workers, ^[102,103] as shown in Scheme 48. The depsipeptide 244 was synthesized by [2 + 2] segment coupling of the ester fragment 243 , prepared from the hydroxyleucine 242 and Boc-( S )-Hpr-OH with DCC/DMAP followed by cleavage of the t -Bu ester, with H-Ser(Me)-Gly-OTce. After deprotection of the Boc group, coupling of the resulting tetrapeptide with Boc-(2 R ,3 R )-MeOTyr(Bzl)-OH proved to be a new difficult sequence to couple. Standard reagents, HATU/DIPEA, Bop-Cl/DIPEA, ^[11] and Brop/DIPEA, ^[104] for assembling hindered sequence gave no coupling product, whereas FDPP/DIPEA ^[29] and BMTB/DIPEA ^[105] gave only 6% and 20%, respectively, of the pentapeptide 245 . Among the several reagents examined, DEPBT/DIPEA (Goodman reagent) 106 was most effective for the difficult sequence to provide 245 in 76% yield. The following segment coupling of 245 with Troc-Ala-Thr-OH again proved to be problematic for the difficult coupling. HATU/DIPEA and even BMTB/DIPEA afforded no desired product. DEPBT/DIPEA again was most superior to the standard coupling procedures, yielding the hexapeptide 246 in 72% yield. However, careful examination of the product showed partial racemization at the Thr residue. Finally, pure 246 was obtained by stepwise couplings of Boc-Thr(TBS)-OH and Troc-Ala-OH using DEPBT/DIPEA in 50% yield and four steps.

Scheme 48

After deprotection of the Troc and Tce groups with zinc dust and phosphate buffer, macrocyclization at 0.01 M using HATU/DIPEA proceeded in good yield to afford the cyclic depsipeptide 247 , whose ¹H NMR spectrum showed good agreement with that of the cyclic core in natural papuamide B (Scheme 49). ^[99]

Scheme 49

(3 S ,4 R )-3,4-Dimethyl-( S )-glutamine, a common component of cyclic depsipeptides, papuamide A ^[96] and callipeltin A, ^[97] was stereoselectively prepared from ( S )-pyroglutamic acid, as shown in Scheme 50. The bicyclic lactam 248 ^[107] prepared from ( S )-pyroglutamic acid according to Thottathil's procedure ^[108] was converted to the unsaturated lactam 249 . Treatment of 249 with lithium dimethylcuprate (Me₂CuLi) in the presence of chlorotrimethylsilane ^[109] preferentially provided the 6-methylated product 250 in a 19:1 ratio. Methylation of 250 at the C7 position through enolate formation with LDA followed by alkylation with iodomethane afforded the trans -dimethyl lactam 251 with a 97:3 diastereomeric ratio. Conversion of the trans -product to the desired cis -dimethyl lactam 252 was achieved by deprotonation with LDA followed by stereoselective protonation. The benzylidene group was cleaved by exposure to excess trifluoroacetic acid, and the hydroxy and amine functions of the lactam were, respectively, protected with TBS and Boc groups. Ring opening of the lactam 253 into the carboxamide was carried out by ammonolysis with 2.4% ammonia-methanol to give the desired product, which was directly oxidized with RuCl₃-NaIO₄ to give N -Boc-3,4-dimethylglutamine 255 without epimerization. The stereostructure of natural dimethylglutamine was unambiguously confirmed to be 2 S ,3 S ,4 R by comparisons of the CD and NMR spectra of the synthetic 3,4-dimethylpyroglutamic acid prepared from the lactam 253 with the hydrolysate of callipeltin A.

Scheme 50

Halipeptins A ( 256a ) and B ( 256b ) ^[110a] are novel 16-membered cyclic depsipeptides isolated from the marine sponge Haliclona sp. collected in waters off the Vanuatu Islands by Gomez-Paloma and co-workers in 2001. Halipeptin A is known to show strong antiinflammatory activity in vivo, causing 60% reduction of edema in mice at a dose of 0.3 mg/kg. In 2002 Gomez-Paloma et al. reported isolation of halipeptin C ( 256c ), closely related to 256a and 256b, from the same sponge, reexamined the original assignments with a novel oxazetidine ring for halipeptins, and corrected the oxazetidine amino acid to the thiazoline amino acid in halipeptins A and B as shown in Figure 12.^[110b] Snider reported confirmation of the above revision based on synthesis of the oxazetidine amino acid. ^[111]

Figure 12

Halipeptins consist of L -alanine and three unique components, the thiazoline-amino acid (alaThz), N -methyl hydroxyisoleucine ( N -MeOHIle) (or N -MeVal for 256c ), and 3-hydroxy-2,2,4-trimethyl-7-methoxy (or hydroxy for 256b and 256c ) decanoic acid (HTMMD or HTMHD). The structures of HTMMD and HTMHD were determined by extensive NMR studies, and the only relative stereochemistry at C3 and C4 was elucidated to be threo ((3 S ,4 R ) or (3 R ,4 S )) except C7, which was confirmed to be S by Mosher's method using HTMHD. In addition to their potent biological activities, their intriguing structures prompted several groups to initiate efforts directed toward the total synthesis. De Riccardis ^[112] and Hamada ^[113] reported synthesis of the N MeOHIle derivative using a diastereoselective silyl-assisted [3,3]-sigmatropic rearrangement and diastereoselective methylation of the bicyclic lactam derived from pyroglutamic acid, respectively. HTMMD was synthesized by De Riccardis. ^[114] Total synthesis of this unique cyclic depsipeptide, halipeptin A ( 256a ), was accomplished by Ma and co-workers, leading to structural confirmation of the revised halipeptins. ^[115] Ma and co-workers employed [2 + 2] coupling of the ester and amide segments including final macrocyclization at the HTMMD/alaThz site for construction of this molecule.

HTMMD with (3 S ,4 R ,7 S )-stereochemistry was prepared from ( R )-4-methyl-5-valerolactone 257 , an oxidative degradation product of diosgenin, ^[116] in nine steps as shown in Scheme 51. ^[115] Ring cleavage of 257 with sodium methoxide in methanol and TBS protection of the resulting primary alcohol gave the ester 258 . For construction of the stereochemistry at C7 secondary alcohol, 258 was reduced with DIBAL, the aldehyde was alkylated by Brown's asymmetric allylation using D - B -allyldiisopinocampheylborane, ^[117] and the resulting alcohol was O -methylated. Desilylation of 259 with TBAF, hydrogenation of the terminal alkene, and Swern oxidation provided the aldehyde 260 . Asymmetric aldol condensation of 260 with the ketene acetal 119 using chiral oxazaborolidine 121 ^[64] was employed for construction of the C3 and C4 stereochemistry to produce the HTMMD skeleton 261 as a single isomer. After exchange of the ester function to the allyl ester, the alcohol was coupled with Fmoc-Ala-Cl in the presence of DMAP and DIPEA at -15℃ to afford the ester segment 262a in excellent yield. The amount of DMAP and the low temperature were critical for success of the racemization-free esterification in this case. Deprotection of the Fmoc group gave the ester segment 262b .

Scheme 51

N -Methylhydroxyisoleucine was prepared in 13 steps from 2-butyn-1-ol using a slightly modified Tsunoda's diastereoselective aza-Claisen rearrangement ^[118] as the key step, shown in Scheme 52. Mesylation of 2-butyn-1-ol, N -alkylation with ( R )- -methylbenzylamine, followed by N -acylation with Boc-Gly-OH using EDCI/HOBt afforded the alkyne 263 , which was converted to the substrate 264 for the aza-Claisen rearrangement through hydrogenation to the cis -olefin using Lindlar catalyst and then N -deprotection. Exposure of the resulting amide 264 to an excess amount of lithium hexamethyldisilazide (LiHMDS) generated the ( Z )-enolate dianion, which spontaneously rearranged to the 4,5-dehyrdoisoleucine 265 in a 3:1 diastereomeric ratio to provide pure 265 after recrystallization in 52% yield. After hydrolytic cleavage of the chiral auxiliary from 265 , methylation with diazomethane followed by hydroboration of the double bond gave the hydroxyisoleucine 266 . Conversion to the N -MeOHIle building block 267b was performed by protection of 266 with the TIPS group, exchange of the Cbz group to the Boc one, N -methylation with iodomethane and silver oxide, saponification with lithium hydroxide, allyl ester formation, and Boc deprotection with aluminum chloride.

Scheme 52

The thiazoline amino acid was prepared according to Rapoport's method. ^[119] Coupling of -methylserine allyl ester ( 268 ) and the nitrobenzothiotriazolide 269 gave the thiopeptide 270 , which was converted to the thiazoline 271 by a sequence of deprotection with TFA and reprotection with FmocOSu, cyclization to the thiazoline with DAST, ^[120] and cleavage of the allyl group with Pd(PPh₃)₄ and N -methylaniline. The thiazoline 271 was then coupled with the N -MeOHIle 267c using 2-bromo-1-ethylpyridinium tetrafluoroborate (BEP)/DIPEA. ^[121] The C -chirality of the thiazoline part was sensitive to the coupling conditions and found to epimerize to an inseparable 3:1 diastereomeric mixture of the dipeptide 272a . The problem was solved by chromatographic separation of the final product. The resulting dipeptide 272a was then deprotected with Pd(PPh₃)₄ and N -methylaniline to afford the amide segment 272b (Scheme 53).

Scheme 53

Coupling of the ester segment 262b with 272b with BEP/DIPEA furnished the linear protected peptide 273 , as shown in Scheme 54. After sequential deprotection at the C - and N -terminals by palladium chemistry and base-mediated reaction, respectively, macrocyclization with HATU/DIPEA was sluggish but afforded the cyclic depsipeptide, which was deprotected with TBAF to furnish halipeptin A ( 256a ) in 27% yield after chromatographic purification. The C -chirality of the thiazoline moiety again proved to be sensitive to TBAF treatment by changing from a 3:1 to 5:1 ratio, and the epimer at thiazoline was obtained in 5% yield. The synthetic halipeptin A was firmly confirmed by comparisons to the spectral data of natural halipeptin A. The total synthesis of halipeptin A by the Ma group unambiguously established the uncertain stereochemistry in the halipeptins.

Scheme 54

4. Cyclic Peptides from Red Alga

cis,cis - and trans,trans -Ceratospongamides ( 274 and 275 ), conformationally isomeric at the two proline amide bonds, were isolated by Gerwick and co-workers ^[122] from the Indonesian red alga Ceratodictyon spongiosum containing the symbiotic sponge Sigmadocia symbiotica . trans,trans -Ceratospongamide ( 275 ) exhibits potent inhibition of sPLA₂ expression in a cell-based model for antiinflammation (ED₅₀ 32 nM), whereas the cis,cis -isomer is inactive. cis,cis -Ceratospongamide ( 274 ) was claimed^[122] to be converted to the trans,trans -isomer 275 by heating at 175 ℃ in DMSO according to HPLC (Figure 13).

Figure 13

The first total synthesis of cis,cis -ceratospongamide ( 274 ) was achieved by Yokokawa, Sameshima, and Shioiri in two ways utilizing a [5 + 2] convergent strategy. ^[123a-c] Their further investigations of the thermal behavior of the cis,cis -isomer 274 revealed that the proposed structure of the trans,trans -isomer 275 should be revised to be the trans,trans -[ D - allo -Ile]-isomer 276 . ^[123c,d]

In their strategy the oxazoline ring in 274 was to be constructed at the final stage because of its sensitivity to acidic and basic conditions. Activation of the carboxyl group in the macrolactamization was carried out at either thiazole or L -proline to avoid possible racemization.

First, the dipeptide 278 was prepared from Boc-Phe-OH and H-Pro-OMe. The required proline thiazole methyl ester ( 279 , Boc-Pro-Thz-OMe) was prepared from Boc-Pro-OH and H-Cys-OMe by the method just described in the synthesis of lyngbyabellins (see Scheme 20). The pentapeptide derivative 282 was prepared from 279 by sequential coupling of each Boc-amino acid using DEPC-Et₃N for coupling and hydrogen chloride in dioxane for deprotection of the Boc group via 280 and 281 , shown in Scheme 55.

Scheme 55

After deprotection of the N -terminal Boc group of 282 and the C -terminal methyl ester of 278 , coupling with DEPC afforded the linear precursor 283 , as shown in Scheme 56. Analogously, deprotection of the Boc group in 278 and the methyl ester in 282 followed by DEPC coupling afforded another linear precursor 285 . Then comparison of the macrolactamization step at the two sites (Thz/Phe vs Pro/Ile) was carried out after deprotection of 283 and 285 . Cyclization at the Thz/Phe site more smoothly proceeded to give the macrocycle 284 , and FDPP afforded 284 in 63% yield, while the cyclization with FDPP at the Pro/Ile site sluggishly proceeded to give 284 in 27% yield. FDPP proved to be superior to DPPA and HATU in both cases.

Scheme 56

Final dehydrative cyclization of the allo -threonine residue to the oxazoline was performed using [bis(2-methoxyethyl)amino]sulfur trifluoride (Deoxo-Fluor) to give cis,cis -ceratospongamide ( 274 ) in 54% yield, as shown in Scheme 57. ^[123] Later, use of Martin's sulfurane ^[28] proved to be much better (81% yield) for formation of the oxazoline. X-ray analysis of 274 further confirmed its structure. ^[123c,124]

Scheme 57

Attempted thermal isomerization of the cis,cis -isomer 274 to the trans,trans -isomer 275 was carried out according to Gerwick's conditions (175 ℃ for 30 min in DMSO- d 6 ), 122 producing a compound different from 275 . This thermal isomerization much more smoothly proceeded in the presence of PPTS, shown in Scheme 57. The product clearly proved to be trans,trans -[ D - a Ile]-isomer 276 by its synthesis using Boc- D - a Ile-OH, analogous to the synthesis of 274 , and identical to natural trans,trans -ceratospongamide.

The thermodynamic isomerization of 274 occurred by C epimerization of the Ile residue to provide the intermediate cis,cis -[ D - allo -Ile]-isomer 287 via 286 . The epimer 287 was immediately isomerized at the two Phe-Pro peptide bonds to produce the trans,trans -[ D - allo -Ile]-isomer 276 , as shown in Scheme 58. Conformational studies using NMR spectra and molecular mechanics/dynamics calculations support this mechanism. ^[123c]

Scheme 58

Conflicting results about trans,trans -ceratospongamide have been reported by Deng and Taunton, ^[125] who adopted the same [5 + 2] convergent strategy as Yokokawa et al. ^[123] The linear precursor 288 was deprotected and cyclized with BOP in the presence of DMAP to give the macrolactam 284 in excellent yield, shown in Scheme 59. The oxazoline formation was efficiently carried out with Deoxo-Fluor to give cis,cis -ceratospongamide ( 274 ). Then, to prepare trans,trans -ceratospongamide ( 275 ), the Boc group of the linear precursor 288 was replaced with the Fmoc group, and the oxazoline formation was accomplished before macrolactamization to give the Fmoc-oxazoline precursor 289 . Simultaneous deprotection of both N - and C -terminals with LiOH followed by macrocyclization with BOP-DMAP in CH₂Cl₂-DMF afforded a 1:3 ratio of trans,trans - and cis,cis -ceratospongamides. Furthermore, thermal isomerization of 274 in DMSO at 175 ℃ ^[122] afforded a 5:1 mixture favoring the trans,trans -isomer 275 . trans,trans -Ceratospongamide obtained here was identical to the natural one and claimed ^[125] to have the structure 275 . However, it should be corrected as trans,trans -[ D - allo -Ile]-ceratospongamide ( 276 ) according to the experiments by Yokokawa, Shioiri, and co-workers. ^[123,126]

Scheme 59

Kutsumura, Sata, and Nishiyama ^[127] also succeeded in the synthesis of cis,cis -ceratospongamide ( 274 ), as shown in Scheme 60. The linear heptapeptide 290 was prepared by a stepwise connection of (Phe-Pro)-thiazole residue, Phe-Pro-OMe, and Ile- a Thr-OMe utilizing BOP-Et₃N as a coupling reagent. After deprotection at the C - and N -terminals, macrolactamization with DPPA-DIPEA afforded the cyclized peptide as a mixture of two diastereomers. Deoxo-Fluor provided cis,cis -ceratospongamide ( 274 ) in 12% yield together with the cis -oxazoline isomer 292 , isomeric at the oxazoline part, in 46% yield, the latter of which could be converted to 274 with NaOMe.

Scheme 60

Synthesis of cis,cis -ceratospongamide ( 274 ) was also achieved by Chen, Deng, and Ye, 128 who adopted [4 + 3] fragment condensation, macrolactamization, and subsequent cyclodehydration, shown in Scheme 61. Both the linear peptides 283 and 292 were prepared from Phe-Pro-thiazole and Phe-Pro-Ile- a Thr derivatives. Interestingly, macrolactamization of 283 with HATU proceeded much more smoothly than that of 292 , and both produced a mixture of two conformational isomers, cis,cis -isomer 284 and cis,trans -isomer 293 , in a ratio of 1:1.3 in both cases. Finally, the oxazoline ring closure of the mixture afforded cis,cis -ceratospongamide ( 274 ) only, which suggested that conformer 293 would equilibrate to the corresponding isomer 284 prior to formation of the oxazoline ring.

Scheme 61

5. Cyclic Peptides from Ascidians

Tamandarins A ( 294a ) and B ( 294b ) were isolated by Fenical and co-workers from the unidentified species of didemnid ascidian and determined to be cyclic depsipeptides closely related to didemnins, as shown in Figure 14. ^[129] Although tamandarins A and B contain isostatine and norstatine, respectively, similar to the didemnins, the hydroxyisovalerylpropionic acid (HIP) residue in the didemnins is replaced by more simple hydroxyisovaleric acid. Interestingly, the antitumor activities of tamandarins A and B are superior to those of didemnins, and their ED₅₀ values are 1.2-2.0 ng/mL.

Figure 14

Joullié and co-workers ^[130] achieved synthesis of tamandarins A and B using the methods developed for total synthesis of didemnins. ^[3e,131] They employed a [4 + 2] segment coupling strategy. Although alloisoleucine, a starting material for synthesis of the isostatine, is commercially available, Joullié et al. reported a practical route from ( S )-2-methyl-1-butanol ( 295 ) in four steps, ^[132] which includes novel imine formation and diastereoselective Strecker synthesis, as shown in Scheme 62. The alcohol 295 was oxidized with 2,2,6,6-tetramethyl-1-piperidinyloxy free radical (TEMPO) and sodium chlorite, ^[133] and the resulting aldehyde 296 was condensed with the sulfinamide anion generated in situ from ( R )-(+)-menthyl p -toluenesulfinate and (Me₃Si)₂ NLi. Diastereoselective Strecker synthesis of the sulfinylimine 297 was achieved using Et( i -PrO)AlCN to provide the -amino nitrile 298 in 90% de. After purification, the stereochemistry of 298 was established by X-ray crystallography. Acid hydrolysis of 298 afforded the alloisoleucine in 52% overall yield.

>

Scheme 62

Synthesis of the statins, summarized in Scheme 63, was carried out in six steps from alloisoleucine and valine in which N -Cbz-protection, conversion to the activated pentafluorophenyl ester using the carbodiimide method, Claisen condensation with the lithium enolate of methyl acetate, and diastereoselective reduction of the -keto esters 299 with potassium borohydride in methanol gave after diastereomeric purification by crystallization the desired isomers 300 in the pure state. TIPS protection followed by saponification afforded the statine building blocks 301 .

Scheme 63

( S )-2-Hydroxyisovaleric acid ( 302 ) was synthesized by the standard method, deamination of L -valine with retention of stereochemistry, and converted to the allyl ester 303 . The dipeptide segments 305 were prepared by condensation of 303 with the statines 301 by the carbodiimide method in the presence of DMAP and then deprotection of the allyl ester 304 with Pd(PPh₃)₄ and morpholine, followed by conversion to the active ester using pentafluorophenol and EDCI/DMAP (Scheme 64).

Scheme 64

The resulting esters 305 were directly coupled with the tetrapeptide TMSE ester 306 used previously for synthesis of the didemnins, shown in Scheme 65. The linear precursors 307 thus obtained were macrocyclized in three steps using deprotection of the TMSE group with magnesium bromide, hydrogenolysis of the Cbz group, and cyclization with the HATU/DIPEA method to afford the cyclic depsipeptides in 63% yield for the precursor of 294a and in 23% yield for the precursor of 294b . After simultaneous deprotection of Boc and TIPS groups with hydrogen chloride-ethyl acetate, introduction of the side chain was accomplished using BOP for 294a and DEPBT for 294b to furnish tamandarins A ( 294a ) and B ( 294b ), respectively, which were identical to the natural products.

Scheme 65

Mollamide ( 308 ) is a member of a more rare family of cyclic peptides having a reverse-prenyl unit associated as a serine ether residue together with a thiazoline ring. This cyclic peptide was isolated from the ascidian Didemnum molle ^[134] and displays moderate cytotoxicity against a range of cell lines with IC₅₀ values of 1 g/mL against P388 (murine leukaemia) and 2.5 g/mL against A549 (human lung carcinoma), HT29 (human colon carcinoma), and CV1 (monkey kidney fibroblast) cells (Figure 15).

Figure 15

Total synthesis of mollamide ( 308 ) was accomplished by McKeever and Pattenden, ^[135] who began with construction of the pentapeptide 313 from proline methyl ester in seven steps (Scheme 66). DCC-HOBT was used for coupling and acetyl chloride-methanol for the Boc deprotection. The dipeptide 309 was converted to the tripeptide 310 , which reacted with the thioacylating reagent 311 ^[119] to give the tetrapeptide 312 . After removal of the Boc function, addition of Boc-Phe-OH afforded the pentapeptide 313 .

Scheme 66

The required reverse prenylated amino acids 318 were synthesized using the Lewis-acid-assisted ring opening of activated aziridines with alcohols according to the method developed by Okawa and co-workers. ^[136] The chiral aziridine 315 was first prepared from H-Ser-OMe in two steps, as shown in Scheme 67. Replacement of the trityl (Tr) group with more electron-withdrawing groups was necessary to facilitate the aziridine ring opening. After investigations of a number of alternatives, the 4-nitrobenzyloxycarbonyl (PNZ) and the 2,2,2-trichloro- tert -butyloxycarbonyl (TcBoc) carbamate groups were selected. Replacement of the trityl group with the PNZ and TcBoc groups afforded the aziridines 316a and 316b , which underwent ring opening with 2-methyl-3-buten-2-ol in the presence of boron trifluoride etherate to give the O -prenylated amino acids 317a and 317b , respectively. Saponification of these compounds produced the required carboxylic acids 318a and 318b .

Scheme 67

After deprotection of the Boc group from 313 , coupling with 318a and 318b in the presence of EDCI-HOBT proceeded to give 319a and 319b , respectively. Although the PNZ group in 319a was smoothly removed, subsequent purification proved to be problematic. In contrast, removal of the TcBoc group from 319b according to Ciufolini's protocol ^[137] was clean and efficient to give the free amine, which was coupled with Troc-Ile-OH to give the heptapeptide 320 . Removal of the Troc group, ^[137] saponification of the methyl ester group, and then macrolactamization using DPPA afforded the macrocycle 321 . Finally, cyclodehydration of the -hydroxy thioamide unit to the corresponding thiazoline was accomplished using DAST to give mollamide 308 . Formation of the sterically problematic thiazoline ring was carried out at the last stage of the synthesis in this work too (Scheme 68).

Scheme 68

Ascidians of the genus Lissoclinum have proven to be a rich source of novel cyclic peptides. ^[3e] Most of them have interesting cytotoxic properties.

Trunkamide A also belongs to the Lissoclinum peptides and was isolated from the colonial ascidian Lissoclinum sp. collected at the Great Barrier Reef, Australia, by Bowden, Ireland, and co-workers in 1996. ^[138] It is reported to have promising antitumor activity and is under preclinical trials. The structure of trunkamide A was initially assigned as 322a , but extensive synthetic works by Wipf and Uto ^[139] clearly showed that the real structure of trunkamide A is 322b , epimeric at C45, as shown in Figure 16. Trunkamide A is a reverse-prenylated member of cyclic peptides having a thiazoline ring, similar to mollamide ( 308 ).

Figure 16

Total synthesis of natural trunkamide A ( 322b ) was achieved by Wipf and Uto, 139b McKeever and Pattenden, ^[140] and Giralt ^[141] and co-workers. The Wipf and Pattenden groups employed the BF₃-assisted aziridine ring opening for construction of the reverse-prenylated (rPr) serine and threonine side chains as in the mollamide case (see Scheme 67). ^[132,134]

Wipf and Uto ^[139] started their synthesis of trunkamide A having the proposed structure 322a from the reverse-prenylated serine derivative ( 323 ), which was hydrolyzed and coupled with the isoleucine derivative using DEPC to yield the dipeptide 324 , as shown in Scheme 69. After chemoselective removal of the Cbz group with Et₃SiH-Pd(OAc)₂, ^[142] the amine 324 underwent the DEPC coupling with the carboxyli