A biomimetic enantioselective approach to the decahydroquinoline class of dendrobatid alkaloids.

Frogs of the neotropical family Dendrobatidae produce a remarkably diverse array of biologically active alkaloids. One of the major classes of these amphibian alkaloids are the decahydroquinolines, which have been isolated not only from skin extracts of dendrobatid and mantelline frogs but also from bufonid toads, tunicates, marine flatworms, and myrmicine ants. They possess either a cis or trans decahydroquinoline ring fusion, with a side chain substituent at both the C-2 and C-5 positions and, in the lepadin series, an acylated hydroxy group at the C-3 position. The most representative decahydroquinoline alkaloid is cis-195A (formerly called pumiliotoxin C), first isolated in 1969 from a Panamanian population of Dendrobates pumilio. The source of amphibian alkaloids remains an unresolved and challenging question, in particular after the discovery that some of these alkaloids also occur in ants, thus strengthening a dietary hypothesis for their origin in frogs. Although there are no conclusive studies concerning the biosynthesis of these toxins and, consequently, little is known about the biosynthetic pathways, there has been speculation as to possible derivation from the polyketide route by aminocyclization of polycarbonyl intermediates (A), leading to either 2,5-disubstituted decahydroquinolines (C) or spiropiperidines (histrionicotoxins). In accordance with this hypothesis, a plausible biosynthetic pathway to the decahydroquinoline class of dendrobatid alkaloids is depicted in Scheme 1. The structural diversity and pharmacological activity associated with this class of alkaloids, as well as the limited amounts available from natural sources, have stimulated considerable synthetic effort in this area, including some biomimetic approaches. In this context we present herein a biomimetic enantioselective approach to the decahydroquinoline class of dendrobatid alkaloids, which has culminated in the biomimetic synthesis of ()-pumiliotoxin C. Our synthetic approach involves the use of an appropriate 1,5polycarbonyl derivative as a synthetic equivalent of the hypothetical biogenetic polyketide intermediate A and (R)-phenylglycinol as a chiral latent form of ammonia to induce the key enantioselective biomimetic aminocyclization to the target hydroquinoline system. To evaluate the feasibility of our proposal, we initially used 1,5tetracarbonyl compound 2 [A: R = OEt; R = (CH2)3CO2Et], which was easily accessible in excellent yield (82%) by reaction of glutaryl dichloride with 4-ethoxy-4-oxobutylzinc bromide (1) in the presence of Pd(PPh3)4 as the catalyst (Scheme 2). To our delight, heating a benzene solution of 2 and (R)phenylglycinol under Dean-Stark conditions in the presence of a catalytic amount of AcOH, resulted in the straightforward construction of the hydroquinoline ring system, with generation of two stereogenic centers, leading directly to the enantiopure tricyclic lactam 3 in 35% yield. Cyclohexenone 4 (22%) and hydroquinolone 5 (25% yield; nearly equimolecular mixture of stereoisomers) were also isolated. Formation of 3 can be rationalized by considering that, after an initial aldol cyclization from the symmetrical starting diketone 2, the resulting -oxoester 4 undergoes a phenylglycinolpromoted cyclocondensation reaction, in a process that mimics the proposed biosynthetic pathway depicted in Scheme 1. In accordance with this interpretation, 2 was first cyclized to cyclohexenone 4 in excellent yield (90%), by treatment with 1N aqueous LiOH followed by reesterification, and then converted to lactam 3 (60% yield) by reaction with (R)-phenylglycinol in refluxing C6H6-cat AcOH. [12] On the other hand, the formation of 5 in the direct reaction of 2 with (R)phenylglycinol is a consequence of the initial generation of an oxazolidine, which then undergoes two successive cyclizations as depicted in Scheme 3. By choosing the appropriate 1,5-polycarbonyl derivative, the above biomimetic double cyclocondensation can be adapted to the enantioselective synthesis of a variety of 2,5-disubstituted decahydroquinoline derivatives, as exemplified by the synthesis of the decahydroquinoline alkaloid cis-195A outlined in Scheme 4. The required diketoester 6 was prepared in 65% yield by Pd-catalyzed coupling of 5-oxohexanoyl chloride with the functionalized organozinc derivative 1 and stereoselectively converted as in the above series to a tricyclic hydroquinolone lactam (8) in excellent overall yield. Thus, the initial aldol cyclocondensation took place in [ [*] Prof. Dr. M. Amat, Dr. R. Griera, R. Fabregat, Prof. Dr. J. Bosch* Laboratory of Organic Chemistry, Faculty of Pharmacy, and Institute of Biomedicine (IBUB), University of Barcelona Av. Joan XXIII s/n, 08028-Barcelona (Spain) Fax: (+34)93-402-45-39 E-mail: [email protected] [email protected] www.ub.edu/farmaco/grupos/amatbosch/indice.htm