Stereospecific Total Synthesis of Somocystinamide A

Abstract

The first total synthesis of somocystinamide A, a disulfide dimer with extremely labile enamide functional groups, was accomplished in a concise and stereospecific manner. Somocystinamide A is reported to possess exceptionally potent anti-angiogenic and tumoricidal activities. The current work should enable further pharmacological investigation of this important natural product.

Marine cyanobacteria are a rich source of biomedically relevant secondary metabolites that are of unique molecular architecture.¹ In line with this theme, somocystinamide A (1) was isolated from a mixed assemblage of Schizothrix and Lyngbya and shown to possess remarkable biological properties.^2,3 Initially, 1 only showed moderate activity against mouse neuroblastoma cells (Neuro-2a).³ In subsequent studies, however, its IC₅₀ against human umbilical vein endothelial cells (HUVECs) was found to be 500 fM.² This astonishing in vitro finding was verified in zebra fish wherein antiangiogenetic effects were observed at 80 nM media concentration. Despite this potency, 1 was shown to have no observable adverse effects on zebra fish even at 30 μM. In addition, 1 was shown to trigger apoptosis in tumor cells via caspase-8 activation.² This activity profile supports the development of 1, or analogs thereof, for potential use in cancer treatment.

Biosynthetically, somocystinamide A appears to be assembled through alternating NRPS-PKS elements with a unique termination of a PKS unit via decarboxylation and dehydration to furnish the terminal olefin as seen in curacin A.⁴ Methylation of the enamide using S-adenosyl methionine, a signature decoration in marine cyano-bacterial natural products,⁵ produces the tertiary enamide. Secondary enamides have been observed in many natural products and their preparation has been studied extensively in recent years.⁶ Tertiary enamides, however, are encountered very rarely in natural products and as such, the strategies for their preparation are relatively undeveloped and scarce.^6,7 Furthermore, the presence of the disulfide group in 1 requires great care and consideration during the course of synthesis.⁸ For example, synthetic investigation of epidithiapiperazinedione natural products (such as 2) has met with much difficulty in the installation of the disulfide.^8,9 To date, only one complete total synthesis of a compound of this class has been reported.¹⁰ Another case in point is psammaplin A (3); in all three of the published syntheses of 3, the sulfur atoms were introduced as a disulfide in the final step so as to avoid side reactions.¹¹

It was envisioned in the synthesis of 1 that the key carbon-carbon connection at the internal olefin would be made by olefin cross metathesis using a ruthenium catalyst.¹² Accordingly, terminal olefin 6 was prepared from the known aldehyde 5¹³ via a Wittig reaction.¹⁴ Thiazolidine was chosen as the protecting group for the thiol of 4 because of its relative stability and its tandem protection of the carbamate proton.

Screening of commercially available ruthenium catalysts revealed that the second generation Hoveyda-Grubbs catalyst (11) was optimal (table 1).^12c Furthermore, this reaction was optimized for multi-gram scale by adjusting the concentration and number of equivalents of 7. Good stereoselectivity was observed in all cases (e.g. trans : cis = 18:1, entry 8). Minimizing the amount of 7 facilitated the purification process because the major side product of the reaction, dimer 12, closely eluted with the desired product 8 during chromatography. In line with recent reports that some ruthenium catalysts are more functional group-tolerant than initially suspected,¹⁵ alkyl sulfides are apparently very well tolerated by 11, but not by 9, suggesting that there may be competition between tri-cyclohexylphosphine and the sulfide 6 for binding as a ligand on ruthenium.¹⁶

Table 1.

Olefin cross metathesis with various conditions to produce 8

	catalyst	conc.a	esterb	yield of 8c
1	9, 20 mol %	0.04	7, 10 equiv	0%
2	10, 20 mol %	0.04	7, 10 equiv	26% (na)
3	10, 4 mol %	0.04	7, 3 equiv	23% (57%)
4	11, 5 mol %	0.03	7, 3 equiv	81% (94%)
5	11, 5 mol %	0.03	12, 1.5 equiv	53% (55%)
6	11, 2.5 mol %	0.03	7, 3 equiv	44% (na)
7d	11, 5 mol %	0.04	7, 3 equiv	73% (83%)
8e	11, 5 mol %	0.2	7, 2.2 equiv	82% (82%)

^aConcentration of 6 (M);

^bEquivalents of 7 or 12 with respect to 6;

^cIsolated yields. Yields in parentheses are based on recovered starting material;

^d3.2 g scale (ca. 10 times more than entries 1~6);

^e5.5 g scale.

The methyl ester 8 was hydrolyzed to obtain the carboxylic acid 13 in order to avoid undesired reduction to the primary alcohol. Afterward, the thiol and the carbamate of 13 were reductively deprotected by sodium in liquid ammonia (scheme 2).^13a Re-protection of the carboxylic acid as a methyl ester, deprotection of the amine, and acetylation yielded 14 in a good yield. Simultaneous basic hydrolysis of the methyl ester and the thioacetate in the presence of O₂ cleanly caused dimerization to the disulfide in one pot to give 15. However, attempts to effect the dimerization by conventional means, such as treatment with I₂, did not give good yields.¹⁷

Scheme 2.

Synthesis of disulfide 15 (somocystinoic acid)

With 15 (named “somocystinoic acid”) in hand, various conditions were investigated to couple the in-situ generated imine 16 to the corresponding di-acyl chloride 20, the formation of which was verified by the reaction with methylamine to produce 17. In most cases, the starting material decomposed while in some cases a trace amount of 1 was observed. This result was curious because there are reports of synthesis of simple enamides via acylation of the corresponding acid chloride with imine.^7,18 It is possible that the putative acyl iminium ion intermediate 21 is intercepted via an intramolecular reaction due to its dimeric nature (scheme 3).¹⁸ In support of this hypothesis, only tautomer 16 and not 19 was observed by ¹H and ¹³C NMR in CD₂Cl₂.¹⁹

Scheme 3.

Possible mechanisms for enamide formation

We then turned our attention to the recently developed Cu-mediated vinylation reaction.⁶ It has been reported, although with little experimental evidence, that this approach is inapplicable to acylic secondary amides.^6b,d,f

Therefore, coupling between a simple amide 23 and a commercially available vinyl bromide 24 was investigated, but found ineffective, giving support to these earlier reports.

Observation that the hydrolytic decomposition of 1 to 17 occurs with relatively low activation energy³ inspired us to carry out the opposite reaction,²⁰ specifically condensation of the aldehyde 18²¹ with 17. A Soxhlet extraction apparatus was found to be a convenient vessel for small scale reactions and allowed the use of solvents heavier than water.²² Observing that the putative intermediate 21 did not yield 1, we hypothesized that the E1 pathway was not viable (scheme 3). In support of this hypothesis, use of a more polar solvent, THF, decreased the yield in comparison to 1,2-dichloroethane, a less polar solvent.²³ The best result was obtained when TsOH was used as the catalyst, which gave a 41% yield. Further investigation of this reaction is underway on model systems.

The analytical data (¹H and ¹³C NMR, MS, UV, IR and optical rotation) for synthetic 1 were essentially identical to those for natural somocystinamide A,²⁴ thus confirming the originally assigned structure.³ The bioactivity of the synthetic product 1 was evaluated in the murine Neuro-2a cancer cell line, but its activity was highly variable,²⁵ possibly due to the unusual solubility properties of 1, or to factors which we do not currently understand. We also tested synthetic 1 in the brine shrimp toxicity model²⁶ and observed significantly impaired motility in treated (at 1, 10 and 100 μg/mL) versus control shrimps (DMSO). We also noted a much decreased quanitity of intestinal contents in the treated group. These observations underscore the value of a synthetic supply of somocystinamide A for it is clear that it possesses biological properties not yet understood.

In conclusion, the first total synthesis of somocystinamide A was achieved despite the presence of two quite challenging functional groups in the molecule. The current synthesis is fairly robust for all steps but the last two such that >1 g of 15 has been prepared smoothly, and scale-up necessary for in vivo testing of this biomedically exciting compound should now be possible. However, our work demonstrates the need for a more robust reaction to prepare tertiary enamides, a development which would open the door for synthesis of other natural products possessing this functional group, such as the laingolides.²⁷

Figure 1.

Examples of disulfide-containing natural products.

Scheme 1.

Cross metathesis to 8

Scheme 4.

Synthesis of somocystinamide A (1)