Total Synthesis of (+/-)-Cylindricines A and B

Article abstract of DOI:10.1021/jo991020q

Cylindricine A (1) and cylindricine B (2) have been aynthesized in 11 steps and 19 percent overall yield. The key reaction involves the addition of an organocopper species to a bicyclic vinylogous amide, which provides complete stereocontrol over the installation of the hexyl side chain.

Full text of DOI:10.1021/jo991020q

J . Org. Chem. 1999, 64, 8263-8266
8263
Tota l Syn th esis of (()-Cylin d r icin es A a n d B
J ulie Fields Liu and Clayton H. Heathcock*
Department of Chemistry, University of California, Berkeley, California, 94720
Received J une 25, 1999
Cylindricine A (1) and cylindricine B (2) have been synthesized in 11 steps and 19% overall yield.
The key reaction involves the addition of an organocopper species to a bicyclic vinylogous amide,
which provides complete stereocontrol over the installation of the hexyl side chain.
In tr od u ction
Marine ascidians, invertebrates of the phylum Chor-
data, subphylum Tunicata, have been investigated heavily
in recent years as sources of interesting secondary
metabolites. Many of the novel compounds isolated from
these organisms are alkaloids belonging to families such
as the quinolizidines,1,2 indolizidines,3 decahydroquino-
lines,4-7 pyrroloiminoquinones,8 and poly-heteroaromat-
ics.9,10 These molecules exhibit a broad range of biological
activity including inhibition of topoisomerase II8 and
toxicity against murine leukemia,1,9 human colon tumor
cell lines,8 bacteria such as E. coli, S. aureus, and B.
subtilis,3,8-10 C. albicans,3 herpes simplex virus types I
and II,10 and a DNA-repair deficient yeast strain.4
Blackman and co-workers examined several extracts of
the Australian ascidian Clavelina cylindrica and reported
in a series of papers the isolation of a new family of
alkaloids. The tricyclic compounds are known as the
cylindricines A-K.11 Structurally related alkaloids are
the clavepictines,12a fasicularin,12b and lepadiformine,12c
which appears to be very similar, although the accuracy
of the published structure has been called into question.13
Cylindricines A (1) and B (2), the major components
of the initial extract, were the first members of the family
to be identified. Additionally, their skeletal structures are
the first known natural examples of the pyrrolo[2,1-j]-
quinoline and the pyrido[2,1-j]quinoline ring systems,
respectively.11 Cylindricine A and cylindricine B are
isomeric, and a pure solution of either alkaloid as the
free base will isomerize to the 3:2 equilibrium mixture
in 6 days at room temperature. The two molecules are
believed to interconvert via an aziridinium ion.11
(1) Raub, M. F.; Cardellina, J . H.; Choudhary, M. I.; Ni, C.-Z.;
Clardy, J .; Alley, M. C. J . Am. Chem. Soc. 1991, 113, 3178.
(2) Kong, F.; Faulkner, D. J . Tetrahedron Lett. 1991, 32, 3667.
(3) Raub, M. F.; Cardellina, J . H., II; Spande, T. F. Tetrahedron Lett.
1992, 33, 2257.
(4) Patil, A. D.; Freyer, A. J .; Reichwein, R.; Carte, B.; Killmer, L.
B.; Faucette, L.; J ohnson, R. K.; Faulkner, D. J . Tetrahedron Lett. 1997,
38, 363.
(5) Steffan, B. Tetrahedron 1991, 47, 8729.
(6) Kubanek, J .; Williams, D. E.; de Silva, E. D.; Allen, T.; Andersen,
R. J . Tetrahedron Lett. 1995, 36, 6189.
(7) Biard, J . F.; Guyot, S.; Roussakis, C.; Verbist, J . F.; Vercauteren,
J .; Weber, J . F.; Boukef, K. Tetrahedron Lett. 1994, 35, 2691.
(8) Copp, B. R.; Ireland, C. M.; Barrows, L. R. J . Org. Chem. 1991,
56, 4596.
(9) Charyulu, G. A.; McKee, T. C.; Ireland, C. M. Tetrahedron Lett.
1989, 30, 4201.
(10) Rinehart J r., K. L.; Kobayashi, J .; Harbour, G. C.; Gilmore, J .;
Mascal, M.; Holt, T. G.; Shield, L. S.; Lafargue, F. J . Am. Chem. Soc.
1987, 109, 3378.
(11) (a) Blackman, A. J .; Li, C.; Hockless, D. C. R.; Skelton, B. W.;
White, A. H. Tetrahedron 1993, 49, 8645. (b) Li, C.; Blackman, A. J .
Aust. J . Chem. 1994, 47, 1355. (c) Li, C.; Blackman, A. J . Aust. J . Chem.
1995, 48, 955.
(12) (a) Raub, M. F.; Cardelina, J . H., II; Choudhary, M. I.; Ni, C.-
Z.; Clardy, J .; Alley, M. C. J . Am. Chem. Soc. 1991, 113, 3178. (b) Patil.
A. D.; Freyer, A. J .; Reichwein, R.; Carte, B.; Killmer, L. B.; Faucette,
L.; J ohnson, R. K.; Faulkner, D. J . Tetrahedron Lett. 1997, 38, 363. (c)
Briard, J . F.; Guyot, S.; Roussakis, C.; Verbist, J . F.; Vercauteren, J .;
Weber, J . F.; Boukef, K. Tetrahedron Lett. 1994, 35, 2691.
(13) (a) Werner, K. M.; de los Santos, J . M.; Weinreb, S. M.; Shang,
M. J . Org. Chem. 1999, 64, 686. (b) Pearson, W. H.; Barta, N. S.;
Kampf, J . W. Tetrahedron Lett. 1997, 38, 3369. (c) Pearson, W. H.;
Ren, Y. J . Org. Chem. 1999, 64, 688.
The novel skeletal structures of cylindricines A and B,
as well as their interesting ability to interconvert, made
these two natural products intriguing targets for total
synthesis. At the outset of this project, no published
syntheses for any of the cylindricine alkaloids existed.
However, during the course of our work on cylindricines
A and B a report was published by Snider et al. detailing
the syntheses of (()-cylindricines A, D, and E, which
10.1021/jo991020q CCC: $18.00 © 1999 American Chemical Society
Published on Web 10/14/1999

8264 J . Org. Chem., Vol. 64, No. 22, 1999
Liu and Heathcock
Sch em e 1
monium hydroxide as cosolvents. After 16 h at 100 °C
the bicyclic ketone was isolated in 78% yield as a
separable mixture of three isomers. These isomers were
identified by comparison to model compounds lacking the
butenyl chain; the identities of the model compounds
were proven using X-ray crystallography.16 Compounds
20-22 have since been published by Snider and Liu, and
1
their H NMR data matches ours.14
The desired isomer 20 was the major product, while
the undesired isomers 21 and 22 were present in lesser
amounts. However, the approximately 2:1 desired/
undesired isomer ratio was not very impressive, so we
attempted to recycle the minor isomers by resubjecting
them to the double-Michael reaction conditions. Unfor-
tunately, after extended heating only a trace amount of
20 was seen; apparently the retro-Michael reaction is not
as facile as might be assumed. While we were able to
perform the retro-Michael reaction using model com-
pounds that lacked the butenyl chain by first generating
the quaternary amine salt with excess iodomethane,16
attempts to generate the quaternary amine salts of 21
and 22 were unsuccessful due to the steric bulk provided
by the quaternary carbon substituents R to the nitrogen.
This disappointing result encouraged us to explore more
selective options for installation of the hexyl side chain.
Our final retrosynthesis involved preparing the re-
quired bicyclic ketone with a stereoselective organocopper
addition to a vinylogous amide. The vinylogous amide
could be prepared from a saturated ketone, which could
arise from the double-Michael addition of an ammonia
equivalent to a dienone. The dienone could be formed in
turn via a Horner-Emmons reaction with a â-ketophos-
phonate and paraformaldehyde.
utilizes a similar double-Michael reaction approach to the
bicyclic core of the cylindricines.14 In addition, a synthesis
of (-)-cylindricine C by Molander et al., which also
involves a double-Michael reaction, has recently appeared
in print.15
Since cylindricines A and B interconvert, one could
arrive at the natural mixture of the two compounds by
the direct synthesis of either isomer, followed by equili-
bration to the 3:2 mixture. We envisioned that the third
ring of the tricycle could be formed by attack of an amine
moiety on an electrophilically activated terminal alkene
or by radical cyclization. The bicyclic ketone could result
from a Mannich-Michael reaction sequence beginning
with an appropriately substituted enone. That enone in
turn could potentially be synthesized by a conjugate
addition-elimination reaction with a â-chloro enone.
Resu lts a n d Discu ssion
In an attempt to investigate this possible synthetic
route, model silyl enol ether 12 and N-acylimmonium
precursor 13 were prepared.16 Upon treatment with BF3‚
OEt2 followed by deprotection of the nitrogen, Mannich
product 14 was isolated in high yield (Scheme 1).
However, efforts to synthesize the desired bicyclic product
proved fruitless, as the â-amino moiety of ketone 14
displayed a marked tendency to eliminate from the
molecule under a wide variety of Michael reaction condi-
tions. This result led us to redesign the synthesis, using
instead a double-Michael reaction to deliver the desired
bicycle in one step from a dienone.
A conjugate addition-elimination reaction was suc-
cessfully performed using known â-trifluoromethane-
sulfonyloxy ester 1617 and the Gilman cuprate formed
from CuI and 2 equiv of 4-lithio-1-butene to provide
desired â-substituted ester 17 (Scheme 2). In later larger-
scale reactions, we switched to the lower-order mixed
cuprate reagent derived from CuCN and 1 equiv of
4-lithio-1-butene in an effort to conserve the alkyllithium
reagent, which must be synthesized from 3-buten-1-ol via
the iodide. In both cases, the yield for the reaction was
excellent. Addition of the lithium anion of dimethyl
methylphosphonate provided the desired â-ketophos-
phonate 18 in 96% yield. A Horner-Emmons reaction
with heptanal then afforded the dienone in 88% yield.
The double-Michael addition of ammonia to the dienone
was carried out in a thick-walled resealable Pyrex tube,
using ammonia-saturated ethanol and concentrated am-
The desired Horner-Emmons reaction was successful,
providing dienone 23 in 86% yield (Scheme 3). The
double-Michael reaction was conducted as before, fol-
lowed by the protection of the nitrogen to afford bicycles
25 as approximately 1:1 mixtures of ring-juncture iso-
mers in excellent yield. The next step involved the
conversion of the saturated ketone to the vinylogous
amide substrate needed for the stereoselective organo-
copper addition. We investigated several methods for
effecting this conversion, including the attempted elimi-
nation of various leaving groups R to the carbonyl such
as bromo, phenylselenyl, and tosyloxy moieties. Unfor-
tunately, these reactions largely resulted in the eventual
decomposition of the starting materials rather than
generation of the desired unsaturated compounds. Even-
tually, we determined the most effective method for the
(14) Snider, B. B.; Liu, T. J . Org. Chem. 1997, 62, 5630.
(15) Molander, G. A., Ro¨nn, M. J . Org. Chem. 1999, 64, 5183.
(16) Fields, J . E. Doctoral Thesis, University of California, Berkeley,
1999.
(17) Lin, H.-S.; Rampersaud, A. A.; Zimmerman, K.; Steinberg, M.
I.; Boyd, D. B. J . Chin. Chem. Soc. 1993, 40, 273.

Total Synthesis of (()-Cylindricines A and B
J . Org. Chem., Vol. 64, No. 22, 1999 8265
Sch em e 2a
F igu r e 1. Global minimization structures of cis and trans
vinylogous amide 27b.
Sch em e 4a
a
Reagents: (a) CH2CHCH2CH2Cu(CN)Li, Et2O, -50 °C; (b)
LiCH2P(O)(OMe)2, THF, -78 °C to rt; (c) C6H13CHO, LiCl,
i-Pr2NEt, CH3CN; (d). NH3-satd EtOH, concd NH4OH, 100 °C.
Sch em e 3a
a
Reagent: (a) C6H13MgBr, CuBr‚Me2S, BF3‚OEt2, THF, -78
°C; (b) TBAF, THF (for 28c and 29c only).
alkyl or aryl group. The question, therefore, was whether
vinylogous amides 27 would occupy conformations in
which axial attack by the incoming organocopper species
would furnish the desired stereoisomers.
Molecular modeling22 supported our expectation that
both of the vinylogous amide isomers would indeed
occupy appropriate conformations (Figure 1) with the
butenyl chains in the axial orientation. However, should
the selectivity of the addition prove acceptable in only
one isomer, we could simply epimerize bicyclic ketone 24
iteratively to a single isomer earlier in the synthesis.
The organocopper addition was successfully performed
using hexylmagnesium bromide in the presence of BF3‚
OEt2 and CuBr‚Me2S. We were pleased to find that the
reaction provided complete stereoselectivity in the case
of trans isomers 27 and excellent stereoselectivity in the
case of the cis isomers, as only a trace of the undesired
product isomer was seen (Scheme 4). This positive result
eliminated any need for the epimerization of bicyclic
ketone 24. The next step proved rather difficult, as the
exceedingly hindered nitrogen atom was surprisingly
challenging to deprotect. We initially examined the use
of a methyl carbamate protecting group. Carbamates 28a
and 29a were treated exhaustively with basic hydrolysis
conditions, acidic removal conditions, and nucleophilic
cleavage conditions such as the use of LiI/collidine or
mercaptan anion/HMPA. All attempts to remove the
protecting group failed, resulting in recovery of the
starting material. We switched to the TFA amide pro-
a
Reagents: (a) NaH, (CH2O)n, benzene; (b) NH3-satd EtOH,
concd NH4OH, 105 °C; (c) ClCO2Me, K2CO3, CH3CN (or) TFAA,
Et3N, CH2Cl2 (or) TEOC-OSu, K2CO3, CH3CN; (d) LDA, TIPSOTf,
THF, -78 °C to rt; (e) CAN, DMF, 0 °C.
formation of the vinylogous amides to be the ceric
ammonium nitrate oxidation of triisopropylsilyl enol
ethers 26,18,19 which were easily formed from ketones 25.
This mild procedure provided excellent yields of the cis
and trans vinylogous amide products, which were sepa-
rated via HPLC.
The next challenge was the stereoselective addition of
an organocopper species to the vinylogous amide sub-
strates. Comins and co-workers have developed an excel-
lent method for achieving this type of addition to N-acyl
vinylogous amides that involves the use of a Grignard
reagent in the presence of BF3‚OEt2 and CuBr‚Me2S.20,21
The method has been demonstrated to provide excellent
stereoselectivity for the axial installation of the incoming
(18) Evans, P. A.; Longmire, J . M.; Modi, D. P. Tetrahedron Lett.
1995, 36, 3985.
(19) Evans, P. A.; Nelson, J . D. J . Org. Chem. 1996, 61, 7600.
(20) Brown, J . D.; Foley, M. A.; Comins, D. L. J . Am. Chem. Soc.
1988, 110, 7445.
(22) Molecular modeling was performed using Macromodel with the
MM3 force field. Global minimizations were performed using the Monte
Carlo method.
(21) Comins, D. L.; Dehghani, A. J . Org. Chem. 1995, 60, 794.

8266 J . Org. Chem., Vol. 64, No. 22, 1999
Liu and Heathcock
Sch em e 5a
success with the metal-coordinated radical procedures of
Stella.24,25 Use of TiCl3 in aqueous acetic acid returned a
mixture of the simple amine and the starting N-chlor-
amine, but use of copper(I) and copper(II) salts in aqueous
acetic acid/THF provided the desired tricyclic product 31
in 85% yield as a 1.14:1.00 mixture of cylindricine A and
epi-cylindricine A. These compounds were separated, and
cylindricine A was allowed to convert to the natural
mixture of cylindricines A and B as a solution in C6D6.26
a
Reagents: (a) NCS, CH2Cl2; (b) AgNO3, MeOH;(c) CuCl, CuCl2,
THF, H2O, AcOH, 0 °C.
Con clu sion
tecting group on the theory that it would be easier to
remove, but this acyl group proved equally difficult to
hydrolyze. In the case of amides 28b and 29b, after
extended reaction times all conditions resulted in the
complete elimination of the TFA amide moiety from the
molecule, thus providing dienone 19. Presumably, this
occurs with the TFA amide and not with the methyl
carbamate due to the former protecting group’s increased
electron-withdrawing ability. We ultimately moved to the
use of the TEOC protecting group in the hope that its
TMS cleavage site would extend well outside the hin-
dered amine pocket, leaving the TMS moiety readily
susceptible to fluoride ion attack. Carbamates 28c and
29c were duly prepared, and we were pleased to find that
upon treatment with TBAF the desired free amine was
liberated cleanly in 89% yield.
With the desired amine in hand, we examined a
number of methods for closing the third ring. Treating
amine 20 with electrophilic reagents such as NBS, NCS,
or I2 failed to produce any of the desired tricyclic products.
However, the NCS reaction resulted in the isolation of a
surprisingly stable N-chloramine in 97% yield (Scheme
5), and we decided to investigate the possibility of
converting this compound to the tricyclic cylindricine ring
system via nitrenium ion closure onto the terminal olefin.
Unfortunately upon treatment of 30 with a series of
silver(I) salts in MeOH the nitrenium ion appears to
undergo spin-inversion,23 abstracting hydrogen from the
solvent to provide simple amine 20. We eventually found
In conclusion, tricyclic alkaloids cylindricine A and
cylindricine B have been synthesized in 11 steps and in
19% overall yield from known â-trifluoromethane-
sulfonyloxy ester 16. The key reaction involved the addi-
tion of an organocopper species to a bicyclic vinylogous
amide, which provided complete stereocontrol over the
installation of the hexyl side chain.
Ack n ow led gm en t. This work was supported by a
research grant from the National Institutes of Health
(GM 46057). The authors thank Professor Adrian J .
Blackman of the University of Tasmania, Hobart, for
1
kindly providing H NMR spectra of natural 1 and 2.
Su p p or tin g In for m a tion Ava ila ble: Full experimental
procedures and characterization data for all compounds pre-
pared, as well as selected 1H NMR spectra. This material is
available free of charge via the Internet at http://pubs.acs.org.
J O991020Q
(23) Gassman, P. G. Acc. Chem. Res. 1970, 3, 26.
(24) Bougeois, J .-L.; Stella, L.; Surzur, J .-M. Tetrahedron Lett. 1981,
22, 61.
(25) Stella, L. Angew. Chem., Int. Ed. Engl. 1983, 22, 337.
(26) The 1H NMR spectra of synthetic cylindricine A and of the
mixture of synthetic cylindricines A and B were compared to 1H NMR
spectra kindly provided by Professor Adrian J . Blackman of the
University of Tasmania, Hobart. The spectra of the natural compounds
were plotted with reference to the solvent peak of C6D6 at 7.40 ppm;
the spectra of the synthetic compounds were plotted with reference to
the solvent peak of C6D6 at 7.16 ppm. With allowance made for this
difference, the data were identical.