Structure Revision of Poecillastrin C and the Absolute Configuration of the β‑Hydroxyaspartic Acid Residue

Abstract

The planar structure of poecillastrin C (1) was revised through selective reduction of the ester carbon. The absolute configuration of the β-hydroxyaspartic acid (OHAsp) residue was determined to be D-threo by Marfey’s analysis. The acid hydrolysate of the reduction product of 1 liberated (2R,3R)-2-amino-3,4-dihydroxybutanoic acid, demonstrating that the β-carboxyl group in poecillastrin C was esterified. The structures of poecillastrins B−D and 73-deoxychondropsin A were also revised.

Graphical Abstract

Chondropsins and poecillastrins are unique marine natural products. They were isolated only in small amounts from taxonomically diverse deep-sea sponges; chondropsins A, B, and D from Chondropsis sp.,^1,2 73-deoxychondropsin A and chondropsin C from Ircinia sp.,³ chondropsin A and 73-deoxychondropsin A from Psammoclemma sp.,⁴ poecillastrins A−C from Poecillastra sp.,^5,6 oecillastrinp C (1) and D from Jaspis sp.,⁷ and mirabalin from Siliquariaspongia sp.⁸ Chondropsins and poecillastrins were discovered as potent cytotoxic compounds with characteristic mean-graph profiles in the National Cancer Institute’s 60-cell antitumor screen and later found to inhibit fungal and mammalian vacuolar H⁺-ATPases (V-ATPases).⁹ Due to their potent activity and unique structural features, this class of compounds is considered as promising leads for new therapeutic agents, especially as anticancer agents. However, further developmental studies have been hampered by restricted supply of the compounds from their natural sources. Total synthesis was unfeasible due to limited stereochemical information; only the relative stereo-chemistry of the THP ring has been reported, and none of the absolute configurations of 26 stereogenic centers in chondropsin A has been reported.¹⁰

The most difficult problem in the structure elucidation of chondropsin A was to assign which of the two carboxylic acid moieties in the OHAsp residue formed the ester linkage and which was free. The authors of ref 1 converted the free carboxylic acid group to a methyl ester and observed an NOE between the O-methyl signal and the oxymethine proton. From this analysis, they concluded that the carboxylic acid group attached to the nitrogen-bearing methine formed the ester linkage. Although the distance between the carbomethoxy protons and the α-methine proton is shorter than that between the carbomethoxy protons and the β-methine proton, both distances fall within the range that can give observable NOEs. Because both carboxyl groups in the OHAsp residues are located within three bonds from either of the methine protons, it is not possible to assign the two carboxyl carbons by HMBC data, unless a HMBC cross-peak is observed from the amide proton (H-3) to one of the carboxyl carbons, which was not observed in chondropsin A. In the structure elucidation of other chondropsin congeners, the same mode of ester formation was proposed without firm evidence.

In the course of our search for bioactive marine metabolites, which cause morphological changes in rat embryonic fibroblast 3Y1 cells, we found activity in the extract of a marine sponge Poecillastra sp. collected at Oshima-shinsone. The combined EtOH and CHCl₃/MeOH (1:1) extract of the sponge (70 g, wet weight) was partitioned between CHCl₃ and H₂O, and the H₂O layer was further extracted with n-BuOH. The CHCl₃ and n-BuOH fractions were combined and fractionated by ODS flash chromatography and RP-HPLC to give 3.2 mg of poecillastrin C as the main active constituent (Supporting Information, SI). The structure of poecillastrin C had been assigned as 1. We examined the NMR data of poecillastrin C and found that the spectroscopic data could not exclude the possibility of structure 2. Therefore, we set out to obtain secure chemical evidence for the structure around the OHAsp moiety.

We took advantage of the different reactivity of the ester and carboxylic acid moieties toward hydride reduction (Scheme 1). In structure 1, reduction with NaBH₄ would convert the C-1 ester carbonyl carbon to a hydroxymethyl, while the C-34 carboxylic acid would stay intact, thereby affording 3-amino-2,4-dihydroxybutanoic acid following acid hydrolysis.¹¹ On the other hand, if a compound with structure 2 is reduced with NaBH₄ and then hydrolyzed, it would afford 2-amino-3,4-dihydroxybutanoic acid.

Scheme 1.

Differentiation of the Alternate Modes of Lactone Formation by Chemical Modification

As a prelude, we determined the absolute configuration of two chiral centers in the OHAsp residue. The acid hydrolysate of poecillastrin C was subjected to Marfey’s analysis,¹² which showed that the OHAsp residue was D-threo (SI). Therefore, the acid hydrolysate of the reduction product of poecillastrin C would afford (2R,3S)-3-amino-2,4-dihydroxybutanoic acid (3) if the structure of poecillastrin C is 1; the compound with structure 2 would give (2R,3R)-2-amino-3,4-dihydroxybutanoic acid (4).

The protected forms of compounds 3 and 4 were prepared as follows. The carboxyl group of monoethylfumarate (5) was reduced to the allyl alcohol (Scheme 2), which was protected by the TBS group. The resultant TBS ether (6) was subjected to the Sharpless asymmetric aminohydroxylation¹³ to give a mixture of 7, 8, and their enantiomers. The enantiomeric mixtures were obtained after HPLC separation. The relative configurations of all the products were shown to be syn by conversion to the cyclic carbamates (9, 10, and their enantiomers), in which a ³J_H2,H3 value of 4.8 Hz was observed for each.¹⁴

Scheme 2.

Synthesis of 9, 10, and Their Enantiomers

The optical resolution of the enantiomers and assignment of their absolute configurations were achieved by the modified Mosher’s method (Scheme 3).¹⁵ The mixture of 7 and its enantiomer was derivatized with (R)-α-methoxy-α-trifluoromethylphenylacetyl chloride (MTPACl) to give 11 and 12, which were separated by HPLC. Their ¹H NMR data showed that the absolute configuration of 11 was (2R,3S). The mixture of 8 and its enantiomer was examined in the same way to give 13 and 14, and the configuration of 13 was established as (2R,3R). Then, 11, 13, and the reduction product of poecillastrin C were hydrolyzed and subjected to Marfey’s analysis, which demonstrated that the hydrolysate of 13 and the reduction product of poecillastrin C were identical (Figure 1, SI). Therefore, the structure of poecillastrin C was reassigned as 2.

Scheme 3.

Preparation of (S)-MTPA Esters 11–14

Figure 1.

LC−MS chromatograms of Marfey’s derivatives: (a) acid hydrolysate of 11; (b) acid hydrolysate of 13; (c) acid hydrolysate of the reduction product of poecillastrin C.

The same set of experiments run with authentic samples of poecillastrins B, D, and 73-deoxychondropsin A liberated 4, demonstrating that the mode of ester formation in these compounds had to be revised (SI).

A combination of microscale reduction, hydrolysis, and derivatization reactions was performed on four different poecillastrin and chondropsin macrolides, and the resulting products were compared with appropriate synthetic standards. The natural products all provided derivatives of the OHAsp residue that are only consistent with macrolactonization occurring via the side chain carbonyl and not the α-amino acid carbonyl. In this work, we showed that the planar structures of the chondropsin/poecillastrin class of metabolites need to be revised, setting the stage for further stereochemical assignment of this class of bioactive compounds.