New total synthesis of the marine antitumor alkaloid (-)-agelastat in A

Article abstract of DOI:10.1021/ol0490476

(Matrix Presented) A new total synthesis of (-)-agelastatin A (1) has been achieved from the chiral oxazolidinone (-)-3. Although enone transposition was problematic when the Michael ring closure of 2 was attempted with strong base, the desired cyclizatio

Full text of DOI:10.1021/ol0490476

ORGANIC
LETTERS
2004
Vol. 6, No. 15
2615-2618
New Total Synthesis of the Marine
Antitumor Alkaloid (−)-Agelastatin A
,†
Mathias M. Domostoj,† Ed Irving,‡ Feodor Scheinmann,‡ and Karl J. Hale*
The Christopher Ingold Laboratories, The Chemistry Department,
UniVersity College London, 20 Gordon Street, London WC1H 0AJ,
United Kingdom, and Ultrafine, Synergy House, Guildhall Close,
Manchester Science Park, Manchester M15 6SY, United Kingdom
k.j.hale@ucl.ac.uk
Received May 24, 2004
ABSTRACT
A new total synthesis of (−)-agelastatin A (1) has been achieved from the chiral oxazolidinone (−)-3. Although enone transposition was problematic
when the Michael ring closure of 2 was attempted with strong base, the desired cyclization could be effected with Hunig’s base after the
pyrrole nucleus was brominated. Subsequent reduction and monobromination afforded synthetic (−)-agelastatin A (1).
(-)-Agelastatin A (1) is an architecturally unusual antineo-
plastic alkaloid isolated from the axinellid sponge Agelas
dendromorpha by Pietra and co-workers.1 Although (-)-
agelastatin A has been shown to potently inhibit the growth
of L1210 leukemia in mice, its antitumor profile against solid
human tumors or human tumor cell lines has not been
thoroughly investigated.1c In fact, (-)-agelastatin A has only
been screened against a single human KB nasopharyngeal
cancer cell line; its IC50 was 0.075 µg/mL.1c
Given the current scarcity of natural (-)-agelastatin A and
our interest in evaluating its growth-inhibitory effects against
solid human tumors in mice, we recently embarked on its
total synthesis for the purpose of increasing supply.2 In this
Letter, we now report on the successful conclusion of this
venture with our recent conversion of (-)-3 into (-)-1.
In mid-2003, we published a fully stereocontrolled enan-
tiospecific total synthesis of the chiral oxazolidinone (-)-7
(Scheme 1) from D-glucosamine.2 Racemic 7 had previously
been converted into racemic agelastatin A by Weinreb and
co-workers in 1999.3 Although our route constituted an
enantiospecific formal total synthesis of (-)-agelastatin A,
we sought an alternative endgame to that used by Weinreb
et al., as their route had employed the highly dangerous and
expensive methyl isocyanate for introducing the cyclic
hemiaminal subunit.3-5
(3) (a) Stein, D.; Anderson, G. T.; Chase, C. E.; Koh, Y-h.; Weinreb,
S. M. J. Am. Chem. Soc. 1999, 121, 9574. (b) Anderson, G. T.; Chase,
C. E.; Koh, Y.-h.; Stein, D.; Weinreb, S. M. J. Org. Chem. 1998, 63,
7594.
(4) For the first asymmetric synthesis of (-)-agelastatin A, see: (a)
Feldman, K. S.; Saunders: J. C. J. Am. Chem. Soc. 2002, 124, 9060. (b)
Feldman, K. S.; Saunders: J. C.; Laci Wrobleski, M. J. Org. Chem. 2002,
67, 7096.
† University College London.
‡ Ultrafine.
(1) (a) D’Ambrosio, M.; Guerriero, A.; Debitus, C.; Ribes, O.; Pusset,
J.; Leroy, S.; Pietra, F. J. Chem. Soc., Chem. Commun. 1993, 1305. (b)
D’Ambrosio, M.; Guerriero, A.; Chiasera, G.; Pietra, F. HelV. Chim. Acta
1994, 77, 1895. (c) D’Ambrosio, M.; Guerriero, A.; Ripamonti, M.; Debitus,
C.; Waikedre, J.; Pietra, F. HelV. Chim. Acta 1996, 79, 727.
(2) Hale, K. J.; Domostoj, M. M.; Tocher, D. A.; Irving, E.; Scheinmann,
F. Org. Lett. 2003, 5, 2927.
(5) For a synthetic approach to (-)-agelastatin A, see: Baron, E.;
O’Brien, P.; Towers, T. D. Tetrahedron Lett. 2002, 43, 723.
10.1021/ol0490476 CCC: $27.50 © 2004 American Chemical Society
Published on Web 06/19/2004

Scheme 1. Our Proposed New Endgame for (-)-Agelastatin
A (1)
Scheme 2. Route to Cyclopentenone 2
The new endgame that we envisaged (Scheme 1) would
transform (-)-3 into the chiral cyclopentenone 2 and would
exploit a base-catalyzed intramolecular Michael addition to
fuse the pyrrole and cyclopentane rings. Ketone 4 would then
be hydrogenolyzed and brominated to complete the total
synthesis of (-)-1.
We commenced our route to (-)-2 with the selective
N-carbamoylation of oxazolidinone (-)-32 with acid chloride
56 (Scheme 2). Using a combination of n-BuLi and DABCO
as the bases, the N-carbamoyl oxazolidinone 8 was readily
formed in 88-94% yield. Compound 8 was then subjected
to a second N-acylation step, on this occasion with 2 equiv
of the pyrrole acid chloride 6, excess Et3N, and DMAP in
CH2Cl2 and THF; the desired pyrrolocarboxamide 9 was
isolated in 89% yield. The SES group7 of 9 was then cleaved
with Bu3SnH and AIBN8 in PhMe at reflux. The free
pyrroloamide 10 was typically obtained in 60% yield.
To the best of our knowledge, this is the first time that a
SES group has been reductively removed from an amide
nitrogen by Bu3SnH under free radical conditions. Impor-
tantly, our new protocol for cleaving amido-SES groups
works reasonably well, as demonstrated here. It certainly out-
performed all of the fluoride-induced cleavage protocols that
we investigated on this and related substrates. A small
amount of the C-desilylated product 11 (up to 7%) was also
sometimes encountered in this Bu3SnH reduction depending
on its overall duration.
Our next objective was to hydrolyze the oxazolidinone
ring of 10 while leaving the urethane unit intact. Although
this reaction could be accomplished cleanly with aqueous
LiOH in THF at room temperature, it did require long
reaction times (128 h) to deliver workable yields of product.
In this regard, the desired allylic alcohol 12 could usually
be isolated in 37-40% yield along with 48% of 10, which
was then recycled. On the basis of the quantity of 10 that
was typically recovered, the yield of 12 was calculated to
be 77%.
Of the various oxidants that were evaluated for converting
12 into 2, pyridinium dichromate in DMF was by far the
most effective. Enone 2 was usually fashioned in 73-76%
yield by this method (Scheme 2). The TPAP/NMO system
could also be used, but the yields were approximately 10%
lower (ca. 65%).
(6) Yoakim, C.; Ogilvie, W. W.; Cameron, D. R.; Chabot, C.; Guse, I.;
Hache, B.; Naud, J.; O’Meara, J. A.; Plante, R.; Deziel, R. J. Med. Chem.
1998, 41, 2882.
(7) (a) Weinreb, S. M.; Demko, D. M.; Lessen, T. A.; Demers, J. P.
Tetrahedron Lett. 1986, 27, 2099. (b) Weinreb, S. M.; Chase, C. E.; Wipf,
P. Venkatraman, Org. Synth. 1997, 75, 161.
(8) (a) For the first report of phenylsulfonyl groups R to a carbonyl being
reductively removed by Bu3SnH and AIBN, see: (a) Smith, A. B., III; Hale,
K. J.; McCauley, J. A., Jr. Tetrahedron Lett. 1989, 30, 5579. (b) For the
first application of this method to arylsulfonylated amides, see: Parsons,
A. F.; Pettifer, R. M. Tetrahedron Lett. 1996, 37, 1667.
We next investigated the base-mediated ring closure of
enone 2 to secure ketone 4 (Schemes 1 and 3). Et3N in
MeOH or 10 mol % Bu3P in THF9 were completely
(9) Stewart, I. C.; Bergman, R. G.; Toste, F. D. J. Am. Chem. Soc. 2003,
125, 8696.
2616
Org. Lett., Vol. 6, No. 15, 2004

Scheme 3. Attempts at CyclizIng the Enone 2 with Various
Scheme 4. Base-Mediated Isomerization of 2
Bases
In light of all of these failures, several Brønsted and Lewis
acids were surveyed for their ability to instigate the desired
cyclization. PPTS (5 equiv) in THF and MeOH was initially
screened in this capacity. Unfortunately, this led to the pyr-
role TMS of 2 being replaced by hydrogen. The latter product
was also formed when TMSOTf (1.1 equiv) was used to
activate enone 2 in THF; it was coproduced with 13.
Given all of these disappointments, a step backward was
taken, and the Swern oxidation and in situ cyclization of 12
was attempted under conditions analogous to those reported
by Feldman and Saunders on a related system.4 With our
substrate, this led to a 16% isolated yield of 13, a 20% yield
of the enone 2, and a 62% recovery of 12.
Because Et3N in MeOH had already been shown to leave
enone 2 unrearranged, we postulated that we might be able
to bring about cyclization under similar conditions, if we
could somehow lower the pKa of the pyrrole nitrogen in 2.
With this in mind, 2 was reacted with 2 equiv of N-bromo-
succinimide (NBS) in THF for 3 h, in the expectation that a
single 2,3-dibromopyrrole 17 would form, whose pKa would
be considerably lower (Scheme 5). To our surprise, a
multicomponent mixture of mono-, di-, and tribromo pyrroles
arose, and none of the starting enone 2 remained.11 Therefore,
rather than attempting to purify the individual products,
Hunig’s base (5 equiv) was added directly to the reaction
mixture, and the reactants were allowed to stir at room
temperature overnight to bring about the requisite Michael
ring closure. The crude mixture was then extractively worked
up and directly dehalogenated by catalytic hydrogenation
over 10% Pd on C (wet, 0.1 equiv) in MeOH for 2 h, in the
presence of NaOAc (3 equiv). After this operation, TLC
analysis became much clearer (see Supporting Information),
with the desired ketone 4 now appearing as a chromato-
ineffective at mediating this cyclization; enone 2 was always
recovered untouched in either case. Surprisingly, when enone
2 was reacted with DBU (2 equiv) in THF at room
temperature, an unusual rearrangement took place and the
cyclopentenone 13 was isolated in 70% yield (Scheme 3)!10
Presumably enone 13 arose from the γ-deprotonation of 2
to give the cyclopentadienol 15 (Scheme 4), which then
underwent facile deprotonation to create the aromatic 6π-
anion 16, which finally reprotonated in the manner shown.
Sodium hydride was also investigated for effecting the
desired ring closure of 2, but yet again this base failed to
deliver the cyclized ketone 4 (Scheme 3). Instead, a complex
mixture of products arose in which the rearranged cyclo-
pentenone 13 predominated. Potassium hexamethyl-disilazide
in THF likewise gave rise to complicated reaction mixtures,
as did K2CO3 in MeOH. Curiously, Cs2CO3 in MeOH3
produced 14 as the only readily isolable reaction product in
64% yield, along with other decomposition products (Scheme
3).
(10) The cyclopentenone isomerization seen here is analogous to the well-
known base-mediated prostaglandin A1 to B1 rearrangement reported by
Corey in the late 1960s. See: Corey, E. J.; Andersen, N. H.; Carlson, R.
M.; Paust, J.; Vedejs, E.; Vlattas, I.; Winter, R. E. K. J. Am. Chem. Soc.
1968, 90, 3245. See also: Newton, R. F.; Roberts, S. M. In Prostaglandins
and Thromboxanes; Roberts, S. M., Newton, R. F., Eds.; Butterworths:
London, Boston, 1982; Chapter 6, p 84.
(11) The only component of this mixture that we could ever obtain in
reasonably pure condition was the tribromopyrrole 18; it is the major product
of this reaction, as far as we can tell. Use of 3 equiv of NBS for the
bromination does not increase the amount of 18 that is formed but, instead,
causes overbromination and significant product decomposition.
Org. Lett., Vol. 6, No. 15, 2004
2617

Although TLC revealed that several faster-moving prod-
ucts were also usually formed in this combined reaction
sequence, none of them could ever be easily purified or
unambiguously characterized.
Scheme 5. Completion of the Total Synthesis of (-)-1
Initially, we attempted the N-debenzylation of 4 through
catalytic hydrogenation in MeOH over Pearlman’s catalyst
(20% Pd(OH)2 on C). Unfortunately, this led to the methyl
acetal 19 being produced as a single product in 36% yield
(not optimized). Importantly, this undesired course for the
reaction could be completely avoided, simply by conducting
the hydrogenation in THF for 24 h.4 Under these conditions,
20 was obtained in 66-74% yield.
The final step of the synthesis was the site-selective
monobromination of 20 with NBS (0.9 equiv) in THF and
MeOH as described by Feldman and Saunders.4 This
provided synthetic (-)-agelastatin A in 84% yield. Signifi-
cantly, our new route to (-)-agelastatin A has so far delivered
223 mg of the natural product in a highly reproducible
fashion.
This material is now being evaluated against xenografted
solid human tumors in mice and for investigations into the
mechanism of antitumor action. The results of these studies
will be published in due course.
Acknowledgment. We thank Ultrafine (Manchester),
Merck, Sharp & Dohme (Harlow), Pfizer (Sandwich), and
Novartis Pharma AG (Basel) for their much appreciated
financial support.
Supporting Information Available: Full experimental
procedures and detailed spectral data of all key compounds,
copies of 500 MHz 1H and 125 MHz 13C spectra, and HRMS
data. This material is available free of charge via the Internet
at http://pubs.acs.org.
graphically resolvable slower-moving spot that could typi-
cally be isolated in 24-35% overall yield for the three steps
from 2.
OL0490476
2618
Org. Lett., Vol. 6, No. 15, 2004