Synthesis of the tricyclic core of sarain A. Use of formaldehyde in an intramolecular Grigg azomethine ylide cyclization

Full text of DOI:10.1021/jo981801b

9616
J . Org. Chem. 1998, 63, 9616-9617
Syn th esis of th e Tr icyclic Cor e of Sa r a in A.
Use of F or m a ld eh yd e in a n In tr a m olecu la r
Gr igg Azom eth in e Ylid e Cycliza tion
Derek J . Denhart, David A. Griffith, and
Clayton H. Heathcock*
Department of Chemistry, University of California,
Berkeley, California 94720
Received September 4, 1998
Sarain A (1) is an alkaloid first isolated by Cimino and
co-workers from a sponge collected in the Bay of Naples.1
In subsequent work, two congeners having one or two more
carbons and a Z-double bond in the macrocyclic ring (sarains
B and C) were isolated from the same source.2 The bizarre
structure of sarain A has elicited considerable synthetic
attention, and two syntheses of the tricyclic core have been
recorded, a synthesis of (()-2 by Weinreb and co-workers3
and a recently communicated synthesis of 3 by Overman and
co-workers.4 Our own initial efforts in this direction, previ-
ously summarized,5 reached the stage of intermediate 4, but
we were unable to remove the N-tosyl group or cause the
nitrogen atom of the sulfonamide to add to the R,â-unsatur-
ated ester. In this paper, we report a revised route that
solves these problems.
F igu r e 1. ORTEP structure of compound 23.
molecular azomethine ylide cyclization, generating the
azomethine ylide by flash vacuum pyrolysis of an aziridine
activated by ester and amide carbonyl groups.4 However, the
steps leading to the required aziridine were relatively low-
yielding and the flash-vacuum pyrolysis procedure seriously
limited throughput. As a result, the previous approach was
not ideal for an early step in a lengthy total synthesis.
Therefore, we turned to the procedure of Grigg and co-
workers,8 whereby the azomethine ylide is generated by
condensation of an R-amino acid derivative with an alde-
hyde. To this end, amine 12 was treated with paraformal-
dehyde in refluxing toluene to obtain the bicyclic heterocycle
13 in good yield. For the cyclization, yields of up to 97% have
been observed on a small scale, but larger scale seems to
result in lower yields. The current best result is 78% on a
scale of 7.6 g of starting amine. In all cases, the remainder
of the material is a mixture of less polar, olefin-containing
products that is presumed to be polymer.
The pyrrolidine nitrogen was benzylated by reaction with
benzyl bromide and sodium carbonate, and the angular ester
function was then reduced in a two-step process employing
diisobutylaluminum hydride, followed by sodium borohy-
dride. The resulting primary alcohol was protected as the
tert-butyldimethylsilyl ether (14). Treatment of 14 with
sodium and tert-butyl alcohol in liquid ammonia cleanly
removed the benzyl groups from the lactam and primary
alcohol, leaving the N-benzylamine intact. Reaction of this
product with triethylsilyl chloride gave 15 in excellent yield.
To avoid a previously observed rearrangement,9 it was
necessary to change the pyrrolidine nitrogen protecting
group at this point. To this end, the N-benzylamine was then
cleaved by catalytic hydrogenolysis and the resulting amine
was protected as the benzyloxycarbonyl derivative (16).
Treatment of lactam 16 with lithium hexamethyldisila-
zane, followed by p-nitrobenzenesulfonyl chloride (“nosyl”
chloride), gave the N-nosyl derivative. The triethylsilyl group
was removed by reaction with camphorsulfonic acid to
provide primary alcohol 17. Methyl ester 18 was prepared
by a three-stage process consisting of Moffatt-Swern oxida-
tion,10 oxidation of the resulting aldehyde with sodium
chlorite,11 and treatment of the resulting carboxylic acid with
methyl iodide and potassium carbonate. Treatment of this
material with lithium hexamethyldisilazane resulted in
clean isomerization to â-keto ester 19.
As shown in Scheme 1, unsaturated amine 7 was prepared
by Wittig reaction of phosphorane 5 and aldehyde 6.6 This
amine was converted into the mixed anhydride with pivalic
acid, which was coupled with the tert-butylcarbamate de-
rived from monoethyl malonate (10), prepared in a straight-
forward manner from the commercially available diethyl
aminomalonate (8).7 The resulting amide 11 was treated
with trifluoroacetic acid to remove the tert-butoxycarbonyl
group and provide primary amine 12.
In our previous approach to the sarain core, we con-
structed the 3,9-diazabicyclo[4.3.0]nonane by an intra-
(1) (a) Cimino, G.; Puliti, R.; Scognamiglio, G.; Spinella, A.; Trivellone,
E. Pure Appl. Chem. 1989, 61, 535. (b) Cimino, G.; Mattia, C. A.; Mazzarella,
L.; Puliti, R.; Scognamiglio, G.; Spinella, A.; Trivellone, E. Tetrahedron 1989,
45, 3863.
(2) (a) Cimino, G.; Scognamiglio, G.; Spinella, A.; Trivellone, E. J . Nat.
Prod. 1990, 53, 1519. (b) Guo, Y.-W.; Madaio, A.; Scognamiglio, G.;
Trivellone, E.; Cimino, G. Tetrahedron 1996, 52, 8341.
(3) (a) Sisko, J .; Weinreb, S. M. J . Org. Chem. 1991, 56, 3210. (b) Sisko,
J .; Henry, J . R.; Weinreb, S. M. J . Org. Chem. 1993, 58, 4945.
(4) Downham, R.; Ng, F. W.; Overman, L. E. J . Org. Chem. 1998, 63,
8096.
(5) Heathcock, C. H.; Clasby, M.; Griffith, D. A.; Henke, B. R.; Sharp,
M. J . Synlett 1995, 467.
(6) Szczepankiewicz, B. G.; Heathcock, C. H. Tetrahedron 1997, 53, 8853.
(7) (a) Tarbell, D. S.; Yamamoto, Y.; Pope, B. M. Proc. Natl. Acad. Sci.
U.S.A. 1972, 69, 730. (b) Henke, B. R.; Kouklis, A. J .; Heathcock, C. H. J .
Org. Chem. 1992, 57, 7056.
(8) (a) For a comprehensive review, see: Grigg, R. Chem. Soc. Rev. 1987,
16, 89. (b) For a similar reaction of a formaldehyde imine, see: Husinec,
S.; Savic, V.; Porter, A. E. A. Tetrahedron Lett. 1988, 29, 6649.
(9) Griffith, D. A.; Heathcock, C. H. Tetrahedron Lett. 1995, 36, 2381.
(10) Omura, K.; Swern, D. Tetrahedron 1978, 34, 1651.
(11) Lindgren, B. O.; Nilsson, T. Act. Chem. Scand. 1973, 27, 888.
10.1021/jo981801b CCC: $15.00 © 1998 American Chemical Society
Published on Web 12/01/1998

Communications
J . Org. Chem., Vol. 63, No. 26, 1998 9617
Sch em e 1a
a
Key: (a) (i) 10, pivaloyl chloride, (ii) Et3N, (iii) 7; (b) CF3CO2H; (c) (Boc)2O; (d) NaOH, EtOH; (e) paraformaldehyde, toluene, reflux; (f) PhCH2Br,
Na2CO3, CH2Cl2, H2O, EtOH; (g) diisobutylaluminum hydride; (h) NaBH4; (i) TBSCl; (j) Na, t-BuOH, NH3; (k) TESCl; (l) H2/Pd(OH)2, THF; (m)
PhCH2O2CCl; (n) (i) LiHMDS, THF, (ii) p-NO2C6H4SO2Cl; (o) camphorsulfonic acid, H2O, THF; (p) Moffatt-Swern; (q) NaClO2; (r) MeI, K2CO3; (s)
LiHMDS, THF; (t) NaBH4, MeOH, CH2Cl2; (u) TFAA, pyridine; (v) DBU; (w) PhSH, K2CO3, DMF.
Reduction of the ketonic carbonyl was accomplished by
reaction with sodium borohydride in methanol-methylene
chloride, and the resulting axial secondary alcohol was
dehydrated by conversion into the trifluoroacetate, which
was treated with 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU)
to accomplish elimination. This treatment led to an ap-
proximately equal mixture of the R,â-unsaturated ester 21,
the desired cyclization product 22, and alcohol 20. Treatment
of the mixture of 21 and 22 with thiophenol and potassium
carbonate in dimethylformamide removed the nosyl group,
providing the sarain core, 23. The structure of 23 was
established by single-crystal X-ray analysis (Figure 1).
In summary, we have developed a viable route to the
characteristic 2,8-diazatricyclo[5.4.0.04,11]undecane skeleton
of the sarains. Our route is relatively long (22 steps, 1.4%
overall yield) and requires rather more protecting group
manipulation than we would like.12 However, the route does
have some virtues compared to the previous syntheses of
the sarain core. The Weinreb route, similar to ours in
accessing the 3,9-diazabicyclo[4.3.0]nonane by an intra-
molecular azomethine ylide cyclization, provides an inter-
mediate that is not functionalized at C1.13 The Overman
route, which uses a completely different strategy, gives an
enantiomerically pure product, but it is a 3:1 mixture at C6,
with the undesired diastereomer predominating. We are
currently developing a modified route that should deal with
the protecting group problems. Our plan is to introduce the
less complicated macrocycle by alkylation of the ester
enolate, and moledular models suggest that the ester enolate
should alkylate exclusively from one face and yield the
desired configuration at C5.14
Ack n ow led gm en t. This work was supported by a
research grant from the United States Public Health
Service (GM 15067) and postdoctoral fellowships from the
American Cancer Society (D.A.G.) and NSERC Canada
(D.J .D.).
Su p p or tin g In for m a tion Ava ila ble: Experimental proce-
dures and spectral data for all compounds and X-ray crystal-
lographic data for compound 23 (26 pages).
J O981801B
(12) Our route diverges from the Weinreb route in the way that bicyclic
lactam 17 is transformed into the tricyclic structure of 23. In our approach,
the six-membered lactam ring is opened by a sort of Dieckman reaction to
give the â keto ester 19. After modification of the functionality, the original
six-membered ring is reformed by Michael addition of the pendant amino-
ethyl group, giving 23. In the Weinreb approach, construction of the tricyclic
system is accomplished directly by addition of a pendant allylsilane to an
immonium ion, derived by partial reduction of the lactam carbonyl. In early
work we did explore
a similar strategy using an aldehyde and the
corresponding dioxolane in place of the allylsilane. However, as has been
described in ref 5, these attempts failed. It should be noted that the modified
approach reported here, although more complicated, is not significantly less
efficientsrequiring only five steps to convert 18 into ester 23.
(13) However, in recently completed work, the Weinreb group have
succeeded in introducing a one-carbon unit at C1: Irie, O.; Henry, J .;
Samizu, K.; Weinreb, S. M. J . Org. Chem., in press.
(14) This anticipation has now been confirmed by Weinreb and co-
workers; ref 13.