Synthetic Studies Towards the Core Structure of Nakadomarin A by a Thioamide-Based Strategy

The tricyclic BCD substructure of the marine natural product nakadomarin A has been synthesised. The strategy utilised a key carbon–carbon bond-forming reaction between a furan and an N-acyliminium ion derived from a secondary thiolactam. In addition, a novel three-component coupling reaction between a thioamide, an allylic bromide and an isocyanate, leading to the establishment of two new stereogenic centres, is reported. Two key steps in a projected total synthesis of nakadomarin A have been realised by using the unique chemistry of thioamides. Formation of the carbocyclic B ring can be effected by nucleophilic attack of a furan on a thiolactam-derived iminium ion, and the key quaternary centre can be established by a novel three-component coupling reaction.


Introduction
Nakadomarin A (1), a hexacyclic alkaloid of the manzamine family, was isolated in 1997 by Kobayashi et al. from the marine sponge Amphimedon sp., collected off the Kerama Islands in Japan. [1] It displays cytotoxicity against L1210 murine lymphoma cells, is an inhibitor of cyclin-dependent kinase 4 and shows antifungal and antibacterial activity.
The structure of nakadomarin A incorporates a tetracyclic furan-containing core (the ABCD ring system) fused to an eight-membered E ring and bridged by a fifteen-membered F ring. The combination of intriguing structural features and promising biological activity has made the compound a popular target for synthesis. Several total syntheses have been published, [2] as have numerous reports on the synthesis of various substructures. [3][4][5][6][7][8][9] Our initial retrosynthetic analysis of nakadomarin A is depicted in Scheme 1. Late-stage construction of the E and F rings by ring-closing metathesis would lead back to the tetracyclic substructure 2. We planned to install the piper-subsequent reductive removal of the sulfur. Padwa and coworkers have shown that mono-N-substituted thioamides, on treatment with α-bromoacyl chlorides, are transformed through N-acylation and S-alkylation into potent N-acyliminium ion electrophiles, which react intramolecularly with nucleophilic alkenes, indoles and benzenes. [10] We planned to establish the vicinal stereocentres of compound 4 through thia-Claisen rearrangement of a diene such as 5; reaction on the exo face of the bicyclic system, through the chair conformation shown, should lead to the desired stereoisomer 4. We expected that 5 would be accessible from the pyroglutamate-derived thiolactam 6 and (Z)allylic bromide 7. [11] In this paper, we describe our model studies in two areas of this retrosynthesis: the cyclisation of a thiolactam-derived iminium ion to generate the BCD ring system of nakadomarin, and the development of a thia-Claisen rearrangement strategy to establish the stereogenic centres in a compound related to 4.

Results and Discussion
We first focused on generating the carbocyclic B ring of nakadomarin A in a simple model of the BCD ring system (Scheme 2). N-Benzylpyrrolidin-2-one (8) was alkylated sequentially with iodomethane and 3-(bromomethyl)furan (9), prepared as a solution in diethyl ether, [12] to afford lactam 10. Removal of the benzyl group was effected with lithium in liquid ammonia, and thionation with Lawesson's reagent [13] then gave thiolactam 11. Treatment of this compound with bromoacetyl chloride [10] in toluene at 100°C led to the desired tetracycle 12 in 53 % yield. The same transformation could be effected in higher yield (73 %) by treatment of thiolactam 11 with methyl bromoacetate in toluene at reflux. Scheme 2. Synthesis of the BCD ring system (12).
Our next aim was to extend this study through the preparation of a compound with functionalities that would allow the construction of the eight-membered E ring of nakadomarin A. To this end, l-pyroglutaminol [14] was protected as its benzylidene N,O-acetal 13 (Scheme 3). [15] Attempts to convert 13 into its thiolactam analogue using Lawesson's reagent were unsuccessful, with the only product isolated being the 1,3,5,2-oxathiazaphosphepane 14, in which a P-S unit has been inserted into the benzylic C-O bond. Clivio and co-workers [16] previously reported the insertion of Lawesson's reagent into the C-O bond of a dihydrouracil nucleoside and found that the unwanted ArPS 2 unit was expelled upon extended heating. In our case, however, further heating of 14 in pyridine led only to recovery of the starting material and so a revised strategy was adopted in which thionation took place at a later stage. C-Ethoxycarbonylation of 13 was carried out according to a literature procedure. [17] Subsequent alkylation with 3-(bromomethyl)furan (9) afforded 15 as a 5:1 diastereomeric mixture; the major diastereomer depicted was isolated in 41 % yield. Hydrolysis of the benzylidene group of 15 with p-toluenesulfonic acid was followed by acetylation of the resulting alcohol and thionation to give secondary thiolactam 16.
Two α-bromocarbonyl electrophiles 17 and 18, each containing a further carbon chain for elaboration of the E ring, were prepared for reaction with 16 (Scheme 4). Lactone 17 was synthesised by conversion of δ-valerolactone into the corresponding TMS ketene acetal and treatment with bromine, and acyl chloride 18 was synthesised by double deprotonation of hex-5-enoic acid with LDA, bromination with carbon tetrabromide and then conversion of the resulting α-bromo acid into an acid chloride using oxalyl chloride.
Initial attempts to effect direct reaction of thiolactam 16 with lactone 17 resulted in recovery of the starting materials, and so the nucleophilicity of 16 was increased through deprotonation. Treatment of 16 with sodium hydride followed by reaction with α-bromo-δ-valerolactone (17) afforded thioimidate 19 as a mixture of diastereoisomers (Scheme 4). On heating in toluene in a sealed tube, no acylation of the nitrogen atom was observed; rather the substrate underwent an Eschenmoser sulfide contraction to give vinylogous carbamate 20 as a single unassigned geometric isomer. Given the lack of base or thiophile in the mixture of diastereomers. Attempts to thionate this compound to give 26 by using either Lawesson's reagent or phosphorus pentasulfide were unsuccessful. A reversal of the step order was attempted. Acetonide 27 could be thionated, albeit in low yield, with Lawesson's reagent, whereas the use of other thionating reagents failed; decomposition was observed when 2,4-bis(phenylthio)-1,3dithia-2,4-diphosphetane-2,4-disulfide (Yokoyama's reagent) [19] was employed, and the use of Bergman's P 4 S 10pyridine complex [20] led to recovery of the starting material. A slightly more efficient synthesis of 29 could be achieved by starting from ethyl l-pyroglutamate (30) [21] by thionation with Lawesson's reagent in 82 % yield, [22] followed by the selective reduction of the ester with lithium borohydride and protection using 2,2-dimethoxypropane. However, we were unable to conduct a successful α-acylation of thiolactam 29.
Having failed in attempts to prepare the desired thia-Claisen substrate 26, we instead conceived a three-component coupling approach that would allow the stereospecific formation of two new C-C bonds at the α position of thiolactam 29 in a single laboratory operation. This strategy is illustrated in Scheme 7.  The reaction of thiolactam 29 with an allylic bromide should give N,S-ketene acetal 31. Such compounds are known to readily undergo substitution reactions with electrophiles such as acid chlorides [23] or isocyanates [24] at the nucleophilic carbon atom; our hope was that we could intercept the N,S-ketene acetal 31 with such an electrophile to generate 32, which would then undergo the desired [3,3]sigmatropic rearrangement to give the product 33. An obvious potential pitfall of this chemistry is that the 1,5-diene moiety of 31 may undergo the sigmatropic rearrangement more rapidly than it reacts with the electrophile; if this were the case, the simple thia-Claisen product 34 would be obtained.
Initial studies on the three-component coupling reaction were carried out by using (E)-cinnamyl bromide as the allylic bromide component. Deprotonation of 29 with either n-butyllithium or lithium hexamethyldisilazide (LHMDS) at low temperature was followed by treatment with (E)-cinnamyl bromide. Once alkylation of the sulfur was complete, as judged by TLC, the second electrophile was added and the reaction mixture warmed to room temperature (Scheme 8). A wide range of electrophiles were tested: ethyl chloroformate, acetic formic anhydride, phenyl isocyanate, chlorosulfonyl isocyanate, trichloroacetyl isocyanate and phosgene. However, under none of the conditions tried could any of the putative three-component coupling products 35 be obtained; rather, the only compound isolated was the product 36 from rearrangement of the first-formed N,Sketene acetal. This clearly indicated that the rate of acyl-ation of the N,S-ketene acetal was not competitive with the rate of sigmatropic rearrangement.
A control experiment carried out in the absence of the second electrophile showed that the [3,3]-sigmatropic rearrangement to form 36 was indeed rapid, proceeding in less than 5 min at room temperature. Under these conditions, 36 could be isolated in 73 % yield as a 25:1 mixture of diastereoisomers. The stereochemistry of the major isomer was confirmed by X-ray crystallography (Figure 1). [25,26] Rawal and co-workers previously reported that N,Sketene acetals derived from (Z)-allylic bromides rearrange much more slowly than those derived from the corresponding E isomers (3 h at reflux in THF vs. warming from -78°C to room temperature). [27] In light of this observation, and given the requirement for a (Z)-allylic bromide in the projected total synthesis (Scheme 1), we next investigated the use of (Z)-cinnamyl bromide.
Further investigations into this three-component coupling reaction and its application to the total synthesis of nakadomarin A are ongoing.

Conclusions
We have demonstrated the feasibility of two key steps in a projected total synthesis of nakadomarin A. First, we have shown that the carbocyclic B ring in a tricyclic ABC substructure can be constructed by electrophilic substitution of a furan by using a thiolactam-derived acyliminium electrophile. Secondly, we have developed a novel threecomponent coupling reaction involving a thioamide, allylic bromide and isocyanate to stereoselectively install two vicinal stereogenic centres, one of them quaternary.