Total Synthesis of the Antitumor Marine Sponge Alkaloid Agelastatin A

Article abstract of DOI:10.1021/ja992487l

The first total synthesis of the cytotoxic marine metabolite agelastatin A (1) has been achieved in about 14 steps performed in 12 operations in approximately 7 percent overall yield starting from cyclopentadiene. Hetero Diels-Alder cycloaddition of cyclo

Full text of DOI:10.1021/ja992487l

9574
J. Am. Chem. Soc. 1999, 121, 9574-9579
Total Synthesis of the Antitumor Marine Sponge Alkaloid
Agelastatin A
Didier Stien, Glen T. Anderson, Charles E. Chase, Yung-hyo Koh, and
Steven M. Weinreb*
Contribution from the Department of Chemistry, The PennsylVania State UniVersity,
UniVersity Park, PennsylVania 16802
ReceiVed July 14, 1999
Abstract: The first total synthesis of the cytotoxic marine metabolite agelastatin A (1) has been achieved in
about 14 steps performed in 12 operations in approximately 7% overall yield starting from cyclopentadiene.
Hetero Diels-Alder cycloaddition of cyclopentadiene with N-sulfinyl methyl carbamate (7) afforded cycloadduct
8, which without purification was converted to allylic sulfoxide 9 and then by a [2,3]-sigmatropic rearrangement
into bicyclic oxazolidinone 11. The C-5a nitrogen was introduced into the oxazolidinone Boc derivative 16 by
a Sharpless/Kresze allylic amination with SES sulfodiimide 12c. Palladium-promoted cyclization of 2-acyl
pyrroles 20 and 21 via a π-allylpalladium intermediate 22 led to ABC-tricycles 23 and 24, respectively. A
hydroxyl urea D-ring model system was constructed by hydroborating 24, leading eventually to keto amide 31
and then to tetracycle 33. A modified strategy was developed for synthesis of the pivotal tricyclic ketone 58,
involving as key steps a chemoselective hydrolysis of N-Boc oxazolidinone 54 and an internal conjugate addition
of pyrrolo cyclopentenone 57. A TMS group was used as a convenient substitute for the C-1 bromine substituent
of agelastatin A, and thus silylpyrrole 58 could be converted to bromopyrrole 59. Finally, the D-ring could be
annulated onto an R-amino ketone derived from 59 using methyl isocycanate, providing racemic agelastatin A
(1).
In 1993 Pietro and co-workers described the isolation of
agelastatin A from the deep water marine sponge Agelas
1a-c
dendromorpha collected in the Coral Sea near New Caledonia.
The structure and relative stereochemistry of this unique
tetracyclic metabolite were established as shown in 1 by a series
of degradative and spectroscopic experiments. The absolute
configuration of agelastatin A at the four stereogenic centers
around the carbocyclic C-ring was determined to be 5aS,5bS,
8
aS,9aR by a combination of molecular modeling of the
conformation of the compound and application of a CD exciton-
1
b
are needed for optimal activity. Alkylation or acylation of these
functional groups, as well as removal of the C-1 pyrrole
bromine, leads to a significant loss of potency. In addition,
agelastatin A is highly toxic in a brine shrimp bioassay and
also is insecticidal against beet army worm and corn root worm.2
The highly unusual heterocyclic array contained in this small
family of marine alkaloids, coupled with the interesting biologi-
cal activity of agelastatin A, makes them attractive targets for
coupling method. Agelastatin A is accompanied by a small
amount of agelastatin B, the corresponding C-1,2 dibromopy-
rrole 2. Metabolites 1 and 2 are inseparable, but a derivative of
the latter could be obtained in pure form for characterization
purposes. More recently, Molinski et al. reported the isolation
from the West Australian sponge Cymbastela sp. of two new
closely related minor metabolites, agelastatin C (3) and D (4),
2
along with agelastatin A. Agelastatin C is identical to 1, except
3
,4
total synthesis. We conceived of an approach to the agelast-
atins via the strategy shown in Scheme 1 starting from
cyclopentadiene as the source of the carbocyclic C-ring. The
basic plan was to use the two double bonds of cyclopentadiene
as handles to introduce the four nitrogens and stereogenic centers
of the natural product. The expectation was that the configu-
ration at the equilibratable hydroxy urea center at C-8a would
be self-controlled by the fact that the cis-fusion of the C and
D-rings is calculated to be ∼20 kcal/mol more stable than the
that it is hydroxylated at C-5b. Interestingly, this compound
does not seem prone to spontaneous loss of N-methylurea.
Agelastatin D is simply an N-demethylated analogue of 1.
Spectroscopic data indicated that 3 and 4 have the same relative
and absolute configuration as 1.
Agelastatin A has been reported to have significant in vitro
1c
activity against L1210 and KB tumor cells. Structure-activity
studies indicate that the C-8a hydroxyl group and both NH’s
2
(1) (a) D’Ambrosio, M.; Guerriero, A.; Debitus, C.; Ribes, O.; Pusset,
trans arrangement. Thus, the intent was to convert cyclopen-
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
(3) For a preliminary account of some agelastatin model studies, see:
Anderson, G. T.; Chase, C. E.; Koh, Y.-H.; Stien, D.; Weinreb, S. M.; Shang,
M. J. Org. Chem. 1998, 63, 7594.
(4) Taken in part from: Anderson, G. T. Ph.D. Thesis, The Pennsylvania
State University, 1995.
1
994, 77, 1895. (c) D’Ambrosio, M.; Guerriero, A.; Ripamonti, M.; Debitus,
C.; Waikedre, J.; Pietra, F. HelV. Chim. Acta 1996, 79, 727.
2) Hong, T. W.; Jimenez, D. R.; Molinski, T. F. J. Nat. Prod. 1998,
1, 158.
(
6
1
0.1021/ja992487l CCC: $18.00 © 1999 American Chemical Society
Published on Web 09/29/1999

Antitumor Alkaloid Agelastatin A Synthesis
J. Am. Chem. Soc., Vol. 121, No. 41, 1999 9575
Scheme 1
Scheme 3a
Scheme 2
a
(
a) X ) Ts; (b) X ) CO Me; (c) X ) SES.
2
utilized ditosyl sulfodiimide 12a, which was heated in refluxing
toluene with 11, followed by treatment with trimethyl phosphite
in methanol, to produce allylic sulfonamide 16′a as a single
regio- and stereoisomer (50%), presumably via an ene reaction
to initally generate 14 and subsequent [2,3]-sigmatropic rear-
tadiene to cyclic carbamate 6 followed by introduction of the
C-5a nitrogen and a 2-acylpyrrole group to give 5. We had some
initial concerns about the possible sensitivity of the bromine in
the pyrrole A-ring as well as its compatibility with some of our
planned transformations, and thus also considered possible
synthetic equivalents of this substituent (vide infra). It might
be noted that the C-1 bromine in agelastatin A is reductively
removed simply by treatment with sodium hydride. In this
paper we report the successful implementation of this strategy
to the first total synthesis of agelastatin A (1).
4
rangement to product 15 (Scheme 3). Our supposition that the
bis-sulfodiimide ene reaction would occur as shown from the
convex face of 11 was later proven by X-ray crystallography
(
vide infra). Similarly, heating the bis-carbamate 12b10 with
olefin 11, followed by phosphite treatment, afforded a single
carbamate 16′b in 73% yield. Unfortunately, despite some effort
we were unable to remove the N-protecting group from either
sulfonamide 16′a or methyl carbamate 16′b to produce the
corresponding allylic primary amine 17.
1
b
Our approach to bicyclic carbamate 11 was based upon
N-sulfinyl dienophile Diels-Alder methodology previously
To solve this deprotection problem, we turned to the SES
1
1
(
â-trimethylsilylethanesulfonyl) derivative. Thus, new sulfo-
5
developed in these laboratories (Scheme 2). Therefore, con-
diimide 12c was easily prepared by standard methodology from
â-trimethylsilylethanesulfonamide (SES-NH2) (see Supporting
Information). The reaction of this diimide with alkene carbamate
densation of cyclopentadiene and N-sulfinyl methyl carbamate
6
(
7) could be effected at 0 °C in benzene to afford cycloadduct
8
in high yield. This compound was prone to retro Diels-Alder
1
1, followed by phosphite treatment, did indeed provide the
desired SES-protected allylic amination product 16′c in 50-
0% yields. However, the reaction tended to be unreliable,
reaction at room temperature upon attempted chromatographic
purification and thus was immediately treated with phenyl-
7
6
magnesium bromide to produce allylic sulfoxide 9 (86% yield
based upon dienophile 7). Upon heating with HMPT in ethanol,
sulfoxide 9 underwent a [2,3]-sigmatropic rearrangement via
the sulfenate ester 10 to afford approximately a 1:1 mixture of
the desired olefinic oxazolidinone 11 and the uncyclized hydroxy
ethyl carbamate 12. However, compound 12 could be cyclized
to 11 with potassium tert-butoxide in high yield.8
particularly on large-scale runs. Thus, it was found best that
olefin carbamate 11 first be converted to the Boc derivative
1
3, which reproducibly underwent the Sharpless/Kresze ami-
nation on heating with 12c, followed by sodium borohydride,
to produce the allylic sulfonamide 16c. This compound proved
difficult to purify, but cleavage of the Boc group with TFA
afforded compound 16′c which could be isolated in pure form
in high overall yield. Cleavage of the SES group with TBAF
subsequently yielded the requisite primary amine 17. To firmly
establish the structure and stereochemistry of these Sharpless/
Kresze amination products, sulfonamide 16c was converted to
its bis-N-Boc derivative, whose structure was determined by
We next moved to introduction of the C-5a nitrogen into
oxazolidinone 11, and investigated applying the Sharpless/
9
Kresze allylic amination procedure to this compound.
A
particular concern here was the level of regio- and stereoselec-
tivity we might achieve in this process. Initial experiments
1
2
(
5) For reviews and lead references to N-sulfinyl dienophile [4 +
]-cycloadditions, see: (a) Weinreb, S. M. Acc. Chem. Res. 1988, 21, 313.
b) Boger, D. L.; Weinreb, S. M. Hetero Diels-Alder Methodology in
X-ray analysis.
2
Since amine 17 and some of the later compounds derived
from this intermediate tended to be rather polar and water
soluble, a second series of compounds was prepared in which
the amine and carbamate were N-protected with hydrophobic
(
Organic Synthesis; Academic Press: San Diego, 1987; Chapter 1. (c)
Weinreb, S. M. Heterodienophile Additions to Dienes. In ComprehensiVe
Organic Synthesis; Trost, B. M., Fleming, I., Eds.; Pergamon: Oxford, 1991;
Vol. 5, p 401.
(
6) Garigipati, R. S.; Freyer, A. J.; Whittle, R. R.; Weinreb, S. M. J.
(10) Kresze, G.; Munsterer, H. J. Org. Chem. 1983, 48, 3561.
(11) (a) Weinreb, S. M.; Demko, D. M.; Lessen, T. A.; Demers, J. P.
Tetrahedron Lett. 1986, 27, 2099. (b) For an improved preparation of SES-
Cl, see: Weinreb, S. M.; Chase, C. E.; Wipf, P.; Venkatraman, S. Org.
Synth. 1997, 75, 161.
(12) We thank Dr. M. Shang (University of Notre Dame) for this analysis.
X-ray data for this compound can be found in the Supporting Information
for ref 3.
Am. Chem. Soc. 1984, 106, 7861.
(
7) Cf. Macaluso, A.; Hamer, J. J. Org. Chem. 1966, 31, 3049.
(8) For some other syntheses of oxazolidinone 11, see: Mulvihill, M.
J.; Gage, J. L.; Miller, M. J. J. Org. Chem. 1998, 63, 3357. Muxworthy, J.
P.; Wilkinson, J. A.; Procter, G. Tetrahedron Lett. 1995, 36, 7539.
(9) (a) Sharpless, K. B.; Hori, T. J. Org. Chem. 1976, 41, 176. (b) Bussas,
R.; Kresze, G. Liebigs Ann. Chem. 1980, 629.

9576 J. Am. Chem. Soc., Vol. 121, No. 41, 1999
Stien et al.
Scheme 4
Scheme 6
′
Scheme 5
We decided to also explore the possibility of utilizing
palladium-based technology for construction of the agelastatin
D-ring. Since the dibenzyl compound 24 was better behaved
and easier to handle than primary amine 23, it was used for
these studies. Therefore, amine 24 was first converted to
N-methyl urea 26, which was then treated with a catalytic
amount of Pd(OAc)2 with Cu(OAc)2 as the stoichiometric
reoxidant to afford the tetracyclic ene urea 28, whose structure
was confirmed by 2D NMR (HMBC) (Scheme 6).1 It seems
reasonable that this transformation occurs via a syn addition of
7,18
benzyl groups. Therefore, allylic sulfonamide 16′c was dialkyl-
ated with benzyl bromide to afford 18, which was then
deprotected to provide benzylamine 19 (Scheme 4).
2
+
Pd and the urea nitrogen to the olefinic double bond of 26 to
produce 27. Subsequent syn â-hydride elimination from 27
would then yield the observed ene urea product 28. Interestingly,
we have been unable to find a close literature analogy for the
syn stereochemical outcome of the first step in this process,
With substrates 17 and 19 in hand, we next turned to
construction of an intact ABC-ring tricycle. For the initial
feasibility studies, it was decided to simply use pyrrole
1
9
-carbonyl chloride13 rather than a 5-brominated derivative.
although related anti additions are known. It is possible that
the R-face of olefin 26 is simply too hindered to allow the
desired anti addition to occur in this particular case. Since
hydration of cyclization product 28 would not afford the correct
C/D-ring functionality, this approach was abandoned.20
2
Alkylation of amines 17 and 19 with this acid chloride afforded
the desired amides 20 and 21, respectively, in good yields
(
Scheme 5). We were pleased to find that exposure of amide
0
carbamate 20 to Pd aforded the desired tricyclic amine 23 as
a single stereoisomer. This compound proved difficult to isolate
and purify and was therefore characterized as the benzyl
We next turned to an alternative strategy to selectively
functionalize the olefinic double bond of tricycle 24. Thus,
amine 24 was converted to the Boc derivative 29, which could
be hydroborated with diborane (Scheme 7). After treatment of
the boranes with hydrogen peroxide, the resulting mixture of
alcohols was directly oxidized with PDC to provide a mixture
of the desired ketone 31 (37% yield from olefin 29) along with
an alcohol tentatively assigned structure 30. Attempts were made
to improve the regioselectivity here, but other hydroborating
reagents were either unreactive toward alkene 29 (e.g., 9-BBN)
or caused extensive decomposition (e.g., thexyl borane, BHBr2).
It was possible, however, to remove the Boc group from 31
with TFA to generate amino ketone 32, which could be
condensed with triphosgene, followed by methylamine to
produce the correctly functionalized agelastatin A tetracycle 33.
Although the conversion of tricyclic alkene 24 to ketone 31
proceeded in only mediocre yield, this sequence proved that an
carbamate derivative 25. In particular, this tricycle showed a
1
5
% H NMR NOE enhancement between protons Ha and Hb,
and neither of these protons showed any enhancement with Hc,
thereby confirming the stereochemistry of the cyclization
4
product 23. Similarly, palladium-promoted cyclization of the
dibenzyl compound 21 led to the tricyclic alkene 24. We believe
these cyclizations proceed via a π-allylpalladium intermediate
2
2 which probably forms with inversion from the allylic
14,15
carbamates 20 and 21.
It is known that soft nucleophiles
3
can add either syn or anti to palladium in such η -complexes
and for stereoelectronic reasons in these cases only syn
cyclization of the pyrrole nitrogen onto the π-allylpalladium
species is possible, giving the desired cis-fused B/C ring
products. It should also be added that, although various nitrogen
heterocycles including indoles have previously been N-alkylated
with π-allylpalladium complexes, apparently this is the first time
a pyrrole has been used in such a reaction.1
(17) van Benthem, R. A. T. M.; Hiemstra, H.; Longarela, G. M.;
Speckamp, W. N. Tetrahedron Lett. 1994, 35, 9281.
6
(18) For reviews of heteroatom/olefin cyclizations promoted by electro-
(
(
13) Boatman, R. J.; Whitlock, H. W. J. Org. Chem. 1976, 41, 3050.
14) For formation of π-allylpalladium complexes from oxazolidinones,
philes, see: (a) Harding, K. E.; Tiner, T. H. Electrophilic Heteroatom
Cyclizations. In ComprehensiVe Organic Synthesis; Trost, B. M., Fleming,
I., Eds.; Pergamon: Oxford, 1991; Vol 4, p 363. (b) Frederickson, M.; Grigg,
R. Org. Prep. Proced. Int. 1997, 29, 33, 63.
see: Cook, G. R.; Shanker, P. S. Tetrahedron Lett. 1998, 39, 3405. Cook,
G. R.; Shanker, P. S. Tetrahedron Lett. 1998, 39, 4991.
(
15) For reviews of the stereochemistry of allylic palladations, see: (a)
(19) See, for example: Backvall, J.-E.; Andersson, P. G. J. Am. Chem.
Soc. 1990, 112, 3683. Andersson, P. G.; Aranyos, A. Tetrahedron Lett.
1994, 35, 4441.
Frost, C. G.; Howarth, J.; Williams, J. M. J. Tetrahedron: Asymmetry 1992,
3
, 1089. (b) B a¨ ckvall, J. E. New J. Chem. 1990, 14, 447.
16) Cf. Billups, W. E.; Erkes, R. S.; Reed, L. E. Synth. Commun. 1980,
0, 147.
(
(20) Several unsuccessful attempts were also made to effect selenocy-
1
clization of alkene urea 26 using PhSeX.18

Antitumor Alkaloid Agelastatin A Synthesis
J. Am. Chem. Soc., Vol. 121, No. 41, 1999 9577
Scheme 7
Scheme 9
Scheme 8
1
6′c could be selectively alkylated on the sulfonamide nitrogen
with p-methoxybenzyl chloride to give 40, thereby introducing
a PMB-protecting group (Scheme 9). This substituent was used
for purposes of improving organic solubility of intermediates
and also since we believed it would have a beneficial effect in
some of the subsequent operations (vide infra). The alkylation
product 40 was then converted to Boc derivative 41, and the
SES group could be removed with TBAF to yield the secondary
amine 42. It was then possible to acylate amine 42 in good
yield with the 5-TMS-pyrrole-2-carbonyl chloride, prepared
from lithio carboxylate 39 and oxalyl chloride, affording silyl
pyrroloamide 43. The corresponding bromo pyrroloamide 44
could be prepared similarly from lithio carboxylate 36, but only
in poor yield. We were pleased to find that with LiOH
chemoselective hydrolysis of the oxazolidinone moiety of 43
occurred in preference to that of the Boc group to give a 4/1
mixture of 46/45. In addition, no hydrolytic cleavage of the
pyrroloamide functionality was observed in this step. We
initially believed that the hydrolysis preference for the cyclic
carbamate was due to a blocking of the Boc group by the PMB
substituent, but this supposition later proved to be incorrect (vide
infra).
Since compounds 45 and 46 were not easily separable, the
mixture was exposed to PDC in DMF to yield a chromato-
graphically separable mixture of the desired enone 47 (61%
overall yield from oxazolidinone 43) along with recovered 45
(Scheme 10). The latter compound 45 could then be recycled
by conversion to 43 in good yield. Attempts also were made to
repeat this same sequence with bromopyrrole substrate 44, but
only extensive decomposition occurred on basic hydrolysis, and
this series of compounds was therefore abandoned.
intermediate like 31 is, in fact, useful for annulation of the
agelastatin D-ring and thus prompted a change in our synthetic
strategy.
2
4
At this stage of the synthesis, procedures were developed for
preparation of pyrrole-2-carboxylic acid derivatives bearing the
bromine, or a synthetic equivalent. Unfortunately 5-bromopy-
rrole-2-carboxylic acid cannot be made efficiently by bromi-
nation of pyrrole-2-carboxylic acid since a complex mixture of
21
regioisomers is produced. Thus, we have utilized methodology
reported by Cava, et al. to synthesize some potentially useful
22a,23
pyrrole A-ring substrates for the agelastatins.
Readily
available dibromopyrrole 3422 can be monometalated with
n-butyllithium and then carboxylated to provide lithium car-
boxylate 35 in high yield (Scheme 8). It was found most
convenient to remove the N-Boc group from pyrrole 35 simply
by heating the neat salt at 150 °C, affording 36 in quantitative
yield.2 Similarly, treatment of the mono lithio derivative of
2b
34 with TMSCl cleanly gave silylpyrrole 37. This compound
could be lithiated again, and then carboxylated to provide lithium
salt 38. Thermolysis of 38 led to clean removal of the Boc-
protecting group, yielding salt 39.
Our newly conceived strategy for preparation of a tricyclic
ketone substrate like 31 was based upon the possibility of
effecting an intramolecular conjugate addition of a pyrrole
nitrogen to a cyclopentenone to generate the B-ring (vide infra).
To test this approach, we explored the synthesis of an appropri-
ate cyclization substrate by a sequence of reactions which
required some critical chemoselective steps. Thus, oxazolidinone
In the key step of our strategy, we found that, on exposure
to cesium carbonate in methanol, pyrrolo cyclopentenone 47
cyclizes in quantitative yield to tricyclic ketone 48. We were
originally concerned about the viability of this step since the
internal Michael cyclization requires what appears to be the
unfavorable amide rotamer shown in the structure 47. One of
the reasons for installing the PMB group on the amide nitrogen
was to try to populate this rotamer, and subsequent results
indicate that indeed this substituent does appear to have a
significant affect (vide infra).
(
21) Anderson, H. J.; Lee, S.-F. Can. J. Chem. 1965, 43, 409.
(22) (a) Chen, W.; Stephenson, E. K.; Cava, M. P.; Jackson, Y. A. Org.
Synth. 1991, 70, 151 and references therein. (b) Rawal, V. H.; Jones, R. J.;
Cava, M. P. J. Org. Chem. 1987, 52, 19.
(
23) Cf. Denat, F.; Gaspard-Iloughmane, H.; Dubac, J. J. Organomet.
(24) 5-Bromopyrrole-2-carboxylic acid and its acid chloride are known
to be unstable, and this is probably the reason for the poor yield of 44.
2
1
Chem. 1992, 423, 173.

9578 J. Am. Chem. Soc., Vol. 121, No. 41, 1999
Stien et al.
Scheme 10
Scheme 11
Scheme 12
The next important operation in the synthesis was to replace
the TMS group of 48 with a bromine. Since ipso substitutions
of silylpyrroles are not well documented,23 we were gratified
to find that 48 could in fact be cleanly converted to bromopy-
rrole 49 with NBS. To annulate the D-ring onto tricyclic ketone
4
9, the Boc group was removed with TFA as was previously
done with 31 (cf Scheme 7) to give keto ammonium salt 50. In
this case, however, concentration of salt 50 led to formation of
a dimeric pyrazine, whereas with intermediate 32 having a
benzyl group on the amine nitrogen was not prone to such a
dimerization. Thus, it was necessary to treat 50 in situ under
dilute conditions with methyl isocyanate and Hunig’s base,
affording a mixture of 51 and 52. Acid hydrolysis of this mixture
then yielded 5-PMB-agelastatin A (52). Unfortunately, despite
extensive effort we have been unable to remove the PMB group
from 52 using a number of different reagent combinations
recovered oxazolidinone 55 (Scheme 12). It was also possible
to selectively replace the Boc group on 55 to give 54, thus
allowing recycling of the undesired minor hydrolysis product.
With the key pyrrolo cyclopentenone 57 in hand, we next
investigated the pivotal internal conjugate addition step. Indeed,
treatment of 57 with cesium carbonate in methanol led to the
desired tricycle 58. However, the yield of 58 here, although
acceptable (61%), was not as high as in the case of 47. Similar
results were obtained by doing the cyclization with potassium
carbonate in THF/water. It is therefore reasonable to assume
that the N-5 substituent does actually play a role in inducing a
significant population of the required amide rotamer. The
stereochemistry of tricycle 58 was firmly secured by its eventual
conversion into agelastatin A. It was then possible to replace
the silyl group of 58 with NBS as done previously to afford
bromopyrrole 59.
Annulation of the D-ring onto intermediate 59 required some
experimentation. Removal of the Boc group of 59 with TFA at
room temperature produced a compound which appeared to
rapidly dimerize even in dilute solution. Boc cleavage with triflic
acid at low temperature (-78 °C) followed by treatment with
methyl isocyanate and Hunig’s base afforded only a trace of
agelastatin A, along with what was presumably dimeric material.
Finally, we discovered that we could cleave the Boc group of
(CAN, DDQ, AlCl3, TFA, HBr, etc). In most cases only
decomposition occurred, and no trace of agelastatin A was
observed. In addition, the PMB group could not be cleaved from
earlier intermediates 48 or 49.
In view of our inability to deprotect 52, we decided to explore
the possibility of modifying the chemistry described above to
avoid the use of a protecting group on the N-5 amide nitrogen.
Toward this end, oxazolidinone 16′c was converted to N-Boc
derivative 16c (cf Scheme 3), and this compound was acylated
with the acid chloride derived from silylpyrrole carboxylate 39
to afford N-acylsulfonamide 53 (Scheme 11). Fluoride-induced
cleavage of the SES group from 53 then led to amide 54. We
were pleased to discover that hydrolysis of 54 with LiOH
proceeded in a manner identical to that of the PMB-substituted
compound 43 to afford an inseparable 1/4 mixture of 55 and
5
6, thus indicating that the bulk of the PMB group is not
25
responsible for the selectivity in this step. Oxidation of this
mixture with PDC gave the requisite enone 57 along with
(25) The presence of a substituent at C-5a is important, however, since
5
9 with excess trimethylsilyl iodide at room temperature, and
LiOH hydrolysis of the unsubstituted system 13 affords a 2:1 ratio of Boc
cleavage product 11 to the product of cleavage of the cyclic carbamate.
subsequent addition of methyl isocyanate followed immediately

Antitumor Alkaloid Agelastatin A Synthesis
J. Am. Chem. Soc., Vol. 121, No. 41, 1999 9579
by dilute NaOH led to formation of agelastatin A (1) (61%).
We believe that this annulation process involves initial formation
of a silyl carbamate like 60 (perhaps also N-silylated)2 and upon
addition of aqueous base this intermediate is converted to the
free amine which is rapidly trapped in situ by the methyl
isocyanate. Synthetic agelastatin A had TLC behavior and proton
NMR spectrum identical to those of an authentic sample.
In conclusion, we have successfully achieved the first total
synthesis of the tetracyclic marine sponge metabolite agelastatin
A (1). The synthesis requires approximately about 14 steps
performed in 12 operations (∼7% overall yield) starting from
cyclopentadiene, which is the precursor of the carbocyclic
C-ring. Key steps include an N-sulfinyl dienophile hetero Diels-
Alder reaction, a Sharpless/Kresze allylic amination using a new
SES-substituted sulfodiimide, an internal Michael addition of
a pyrrole nitrogen to a cyclopentenone, and a D-ring annulation
by addition of methyl isocyanate to an R-amino ketone.
6
Acknowledgment. We are grateful to the National Science
Foundation (CHE-97-32038) for generous financial support of
this research and to Dr. Alan Benesi for help in obtaining and
interpreting 2D NMR data. We thank the National Institutes of
Health for a postdoctoral fellowship (1 F32 GM17949) to C.E.C.
and the Ministry of Foreign Affairs (France) for a Lavoisier
Postdoctoral Fellowship to D.S.. We are also grateful to
Professor T. Molinski for a TLC sample and a copy of the proton
NMR spectrum of natural agelastatin A.
Supporting Information Available: Experimental details
and spectral data for all new compounds, along with copies of
proton and carbon NMR spectra (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
(
26) Jung, M. E.; Lyster, M. A. J. Chem. Soc., Chem. Commun. 1978,
15. Lott, R. S.; Chauhan, V. S.; Stammer, C. H. J. Chem. Soc., Chem.
Commun. 1979, 495. Cf. also: Sakaitani, M.; Ohfune, Y. Tetrahedron Lett.
985, 26, 5543.
3
1
JA992487L