Granulatimide and Isogranulatimide, Aromatic Alkaloids with G2 Checkpoint Inhibition Activity Isolated from the Brazilian Ascidian Didemnum granulatum: Structure Elucidation and Synthesis

Article abstract of DOI:10.1021/jo981607p

Crude methanol extracts of the ascidian Didemnum granulatum collected in Brazil showed activity in a new screen for G2 cell cycle checkpoint inhibitors. Bioassay-guided fractionation of the extract yielded the known alkaloids didemnimides A (1) and D (2), the new alkaloid didemnimide E (3), and a new G2 checkpoint inhibitor. Two candidate structures for the inhibitor, named granulatimide (4) and isogranulatimide (5), have been prepared via a short and efficient biomimetric synthesis involving the photolysis of didemnimide A (1). The synthesis revealed that the correct structure for the naturally occurring G2 checkpoint inhibitor is isogranulatimide (5). Granulatimide (4), the other candidate structure, was also found to be a G2 checkpoint inhibitor, and it was subsequently detected in chromatographic fractions associated with purification of D. granulatum alkaloids. Granulatimide (4) and isogranulatimide (5) represent the first examples of a new class of G2 specific cell cycle checkpoint inhibitors and the first ones identified through a rational screening program.

Full text of DOI:10.1021/jo981607p

9850
J . Org. Chem. 1998, 63, 9850-9856
Gr a n u la tim id e a n d Isogr a n u la tim id e, Ar om a tic Alk a loid s w ith G2
Ch eck p oin t In h ibition Activity Isola ted fr om th e Br a zilia n
Ascid ia n Did em n u m gr a n u la tu m : Str u ctu r e Elu cid a tion a n d
Syn th esis†
Roberto G. S. Berlinck,*,‡,§ Robert Britton,§ Edward Piers,*,| Lynette Lim, Michel Roberge,
Rosana Moreira da Rocha,X and Raymond J . Andersen*,§
Departments of Chemistry and Earth and Ocean Sciences, University of British Columbia, Vancouver,
BC, Canada V6T 1Z1, Department of Biochemistry and Molecular Biology, University of British
Columbia, Vancouver, BC, Canada V6T 1Z3, and Departamento de Zoologia, Setor de Ciencias
Biologicas, Universidade Federal do Parana, Curitiba, Parana, Brazil
Received August 6, 1998
Crude methanol extracts of the ascidian Didemnum granulatum collected in Brazil showed activity
in a new screen for G2 cell cycle checkpoint inhibitors. Bioassay-guided fractionation of the extract
yielded the known alkaloids didemnimides A (1) and D (2), the new alkaloid didemnimide E (3),
and a new G2 checkpoint inhibitor. Two candidate structures for the inhibitor, named granulatimide
(4) and isogranulatimide (5), have been prepared via a short and efficient biomimetic synthesis
involving the photolysis of didemnimide A (1). The synthesis revealed that the correct structure
for the naturally occurring G2 checkpoint inhibitor is isogranulatimide (5). Granulatimide (4), the
other candidate structure, was also found to be a G2 checkpoint inhibitor, and it was subsequently
detected in chromatographic fractions associated with purification of D. granulatum alkaloids.
Granulatimide (4) and isogranulatimide (5) represent the first examples of a new class of G2 specific
cell cycle checkpoint inhibitors and the first ones identified through a rational screening program.
In tr od u ction
tion of genetic abnormalities. Cells with mutated p53
tumor suppressor genes (≈50% of all solid tumor cancer
cells) are unable to activate the G1 checkpoint in re-
sponse to DNA damage. However, the G2 checkpoint,
although usually weaker than in normal cells, still
provides them with an opportunity to repair their DNA
after damage and to continue successfully dividing.
Inhibitors of the G2 checkpoint when used alone should
have no effect on normal or cancer cells. When used in
combination with a DNA-damaging agent, they should
have little effect on the ability of normal cells to survive,
but should dramatically increase the killing of p53-
cancer cells. This is because normal cells should have
time to repair DNA damage as they can still activate
their G1 checkpoint, and their G2 checkpoint is typically
strong and would require a higher dose of checkpoint
inhibitor to be blocked. However, p53- cancer cells,
which lack a G1 checkpoint and have a weak G2
checkpoint, should be forced into a premature and lethal
mitosis.
The ultimate goal in the chemotherapeutic treatment
of cancer is to develop drugs that are highly effective
against tumor cells and have little or no effect on healthy
cells. Recent discoveries of clear genetic differences
between healthy and cancer cells have presented a
number of interesting opportunities for developing strat-
egies to selectively target cancer cells by taking advan-
tage of their genetic abnormalities. One such approach
would be to use a DNA damaging agent in combination
with a G2 specific cell cycle checkpoint inhibitor.1
Normal cells respond to DNA damage by activating
checkpoints that temporarily halt growth and division
to allow time for DNA repair.2 The G1 checkpoint allows
DNA repair before it is replicated in S phase, and the
G2 checkpoint permits DNA repair before chromosomes
are segregated in mitosis, thus preventing the propaga-
* Address correspondence to Raymond Andersen, Department of
Chemistry, University of British Columbia, Vancouver, B.C., V6T 1Z1.
FAX 604 822 6091. Email: randersn@unixg.ubc.ca.
† Dedicated to Professor Sergio de Almeida Rodrigues on the occasion
of his 60th birthday.
Very few G2 checkpoint inhibitors are known. Two
groups have been found serendipitously: the purine
analogues (caffeine, pentoxifylline, etc.), and staurospo-
rine and UCN-01 (7-hydroxystaurosporine).3 In a variety
of paired cell lines differing only in their p53 status, G2
checkpoint inhibitors induced greater sensitivity to DNA
damage in p53- cells than in p53+ cells.4 For example,
‡ On leave from the Instituto de Quimica de Sao Carlos, Univer-
sidade de Sao Paulo, CP. 780, CEP 13560-970, Sao Carlos, SP, Brazil.
Tel. 00-55-16-2739954. Fax: 00-55-16-2739987. Email: rberlink@
iqsc.sc.usp.br.
§ Departments of Chemistry and Earth and Ocean Science, Univer-
sity of British Columbia.
| Department of Chemistry, University of British Columbia.
Department of Biochemistry and Molecular Biology, University
of British Columbia.
(3) (a) Busse, P. M.; Bose, S. K.; J ones, R. W.; Tomlach, L. J . Radiat.
Res. 1978, 76, 292. (b) Schlegel, R.; Pardee, A. B. Science 1986, 232,
1264. (c) Downes, C. S.; Musk, S. R. R.; Watson, J . V.; J ohnston, R. T.
J . Cell Biol. 1990, 110, 1855. (d) Steinmann, K. E.; Belinski, G. S.;
Lee, D.; Schlegel, R. Proc. Natl. Acad. Sci. USA 1991, 88, 6843. (e)
Andreassen, P. R.; Margolis, R. L. Proc. Natl. Acad. Sci. U.S.A. 1992,
89, 2272. (f) Tam, S. W.; Schlegel, R. Cell Growth Differ. 1992, 3, 811.
(g) Wang, Q.; Fan, S.; Eastman, A.; Worland, P. J .; Sausville, E. A.;
O’Connor, P. M. J . Natl. Cancer Inst. 1996, 88, 956.
X Universidade Federal do Parana.
(1) (a) Holloway, S. L.; Glotzer, M.; King, R. W.; Murray, A. W. Cell
1993, 73, 1393. (b) Weinert, T.; Lydall, D. Seminars Cancer Biol. 1993,
4, 129. (c) Nurse, P. Cell 1997, 91, 865. (d) Kao, G. D.; McKenna, W.
G.; Maity, A.; Blank, K.; Muschel, R. J . Cancer Res. 1997, 57, 753.
(2) (a) Hartwell, L.; Weinert, T.; Kadyk, L.; Garvick, B. Cold Spring
Harbor Symp. Quantum Biol. 1994, 59, 259. (b) Kaufmann, W. K.;
Paules, R. S. FASEB J . 1996, 10, 238.
10.1021/jo981607p CCC: $15.00 © 1998 American Chemical Society
Published on Web 11/25/1998

Aromatic Alkaloids Isolated from a Brazilian Ascidian
J . Org. Chem., Vol. 63, No. 26, 1998 9851
pentoxifylline enhanced cisplatin-induced killing of p53-
MCF-7 cells 30-fold. The known G2 checkpoint inhibitors
are of limited clinical use because they also interact with
a broad range of other cellular processes. However, the
demonstration that even imperfect G2 checkpoint inhibi-
tors can potentiate the cytotoxicity of DNA-damaging
agents in a manner specific for p53- cells suggests that
a rational search for novel inhibitors could yield agents
with greater specificity and potency. It is therefore of
considerable importance to find new and specific check-
point inhibitors that may form the basis of a different
approach to cancer treatment.
The first high-throughput bioassay for G2 cell cycle
checkpoint inhibitors has recently been developed in one
of our laboratories.5 This ELISA-based screen has been
used to evaluate ca. 1300 extracts of marine invertebrates
and marine bacterial cultures for G2 checkpoint inhibi-
tors. Crude methanol extracts of the ascidian Didemnum
granulatum collected in Brazil showed promising activity
in the assay. Fractionation of D. granulatum extracts led
to the isolation of the known alkaloids didemnimides A
(1) and D (2)6 and a new alkaloid didemnimide E (3),
which are all inactive in the checkpoint assay. A small
amount of a novel G2-specific cell cycle checkpoint
inhibitor was also isolated. Two candidate structures for
the inhibitor, named granulatimide (4) and isogranula-
timide (5), have been prepared via a short and efficient
biomimetic synthesis involving the photolysis of didem-
nimide A (1). The synthesis revealed that the correct
structure for the naturally occurring G2 checkpoint
inhibitor was isogranulatimide (5). Granulatimide (4),7
the other candidate structure, was also found to be an
in vitro G2 checkpoint inhibitor, and it was subsequently
detected in chromatographic fractions associated with
purification of D. granulatum alkaloids, possibly as an
artifact. Herein we report the structures of the novel
alkaloids didemnimide E (3), granulatimide (4), and
isogranulatimide (5) and the total synthesis of didem-
nimide A (1), granulatimide (4), and isogranulatimide (5).
multiple times with fresh methanol. The combined crude
ethanol/methanol extract showed promising activity in
a novel assay designed to detect G2 checkpoint inhibi-
tors.5 Initial fractionation of the crude methanol extract
from the Araca Beach, Sao Sebastiao collection of D.
granulatum via Sephadex LH20 (MeOH) chromatogra-
phy gave a number of deep red and orange colored
fractions. One of these fractions showed G2 cell cycle
checkpoint inhibition activity, and further purification
of that fraction by C18 RPHPLC yielded the pure G2
checkpoint inhibitor isogranulatimide (5) as a red amor-
phous solid. RPHPLC fractionation of colored but inactive
fractions gave pure didemnimides A (1) and D (2).
Fractionation of the crude methanol extracts of a second
collection of D. granulatum made at Arquipelago do
Arvoredo and in the Sao Sebastiao Channel using the
same chromatography procedures gave the new alkaloid
didemnimide E (3) as the major constituent, in addition
to 1, 2, 4, and 5. The known alkaloids didemnimides A
(1) and D (2) were identified by comparison of their
spectroscopic data with the literature values.6
Resu lts a n d Discu ssion
The ascidian D. granulatum occurs in rocky, shallow
water marine habitats along the coastline of southern
Brazil. Specimens of D. granulatum were collected by
hand using SCUBA at three different sites, and the
freshly collected ascidians were immediately immersed
in ethanol and stored at -20 °C. Further workup involved
decanting the ethanol followed by blending the ascidian
tissue with methanol and extracting the solid residue
Didemnimide E (3) was obtained as an orange amor-
phous solid that gave a [M + H]+ ion at m/z 293.1048 in
the HRFABMS appropriate for the molecular formula
C16H12N4O2, identical with the molecular formula of
1
didemnimide C (6). Comparison of the H and 13C NMR
data obtained for didemnimide E (3) with the literature
data for didemnimide C (6)6 indicated that the molecules
were closely related (Tables 1 and 2). The 1H and 13C
NMR spectra of didemnimide E (3), like the NMR spectra
of didemnimide C (6), showed well-resolved resonances
for all the protons and carbons in the molecule, indicating
that the tautomerism which caused signal broadening in
the NMR spectra of didemnimide A (1) was no long
(4) (a) Russell, K. J .; Wiens, L. W.; Demers, G. W.; Galloway, D. A.;
Plon, S. E.; Groudine, M. Cancer Res. 1995, 55, 1639. (b) Powell, S.
N.; DeFrank, J . S.; Connell, P.; Eogan, M.; Preffer, F.; Dombkowski,
D.; Tang, W.; Friend, S. Cancer Res. 1995, 55, 1643. (c) Russell, K. J .;
Wiens, L. W.; Demers, G. W.; Galloway, D. A.; Le, T.; Rice, G. C.;
Bianco, J . A.; Singer, J . W.; Groudine, M. Int. J . Radiat. Oncol. Biol.
Phys. 1996, 36, 1099. (d) Yao, S. L.; Akhtar, A. J .; McKenna, K. A.;
Bedi, G. C.; Sirdransky, D.; Mabry, M.; Ravi, S. J .; Collector, M. I.;
J ones, R. J .; Sharkis, S. J .; Fuchs, E. J .; Bedi, A. Nature Med. 1996, 2,
1140. (e) Bracey, T. S.; Williams, A. C.; Paraskeva, C. Clin. Cancer
Res. 1997, 3, 1371.
(5) Roberge, M.; Berlinck, R. G. S.; Xu, L.; Anderson, H.; Lim, L.;
Curman, D.; Stringer, C. M.; Friend, S. H.; Davies, P.; Vincent, I.;
Haggarty, S.; Kelly, M. T.; Britton, R.; Piers, E. Andersen, R. J . Cancer
Res. 1998, in press.
(6) Vervoort, H. C.; Richards-Gross, S. E.; Fenical, W.; Lee, A. Y.;
Clardy, J . J . Org. Chem. 1997, 62, 1486-1490.
(7) The structure and synthesis of granulatimide was reported at
the 9th International Symposium on Marine Natural Products, Towns-
ville, Australia, J uly 6-10, 1998.
1
present. Significant differences in the H NMR data for
the two compounds was only observed in the chemical
shifts of the N-methyl and imidazole proton resonances
(Table 1), indicating that didemnimide E (3) and didem-
nimide C (6) simply differed in the position of imidazole
N-methylation. The HMBC data obtained for didemnim-
ide E (3) showed correlations between the N-methyl
proton resonance at δ 3.45 (Me-18) and two proton-
bearing imidazole carbon resonances at δ 124.3 (C-14)
and 137.2 (C-16), confirming the structure of didemnim-
ide E (3). All of the additional COSY, HMQC, and HMBC

9852 J . Org. Chem., Vol. 63, No. 26, 1998
Ta ble 1. 1H NMR Da ta for Did em n im id es a n d Gr a n u la tim id es. Recor d ed in DMSO-d 6
Berlinck et al.
atom no.
didemnimide A (1)6
didemnimide C (6)6
didemnimide E (3)
granualtimide (4)
isogranulatimide (5)
1
2
4
5
6
12.45, bs
8.05, bs
12.02, bs
8.07, bs
12.31, bs
8.20, bs
12.58, bs
13.48, bs
-
8.51, d, J ) 7 Hz
7.35, dd, J ) 7, 7 Hz
7.43, dd, J ) 7, 7 Hz
7.67, d, J ) 7 Hz
11.11, bs
8.10, b
9.12, b
-
-
-
7.07, br m
6.87, br m
7.07, br m
7.39, d, J ) 7.8 Hz
11.66, bs
7.71, bs
7.68, bs
10.87, bs
-
6.38, d, 8 Hz
6.77, dd, 8, 8 Hz
7.10, dd, 8, 8 Hz
7.44, d, 8, Hz
11.13, bs
6.43, d, J ) 7 Hz
6.90, dd, J ) 7, 7 Hz
7.16, dd, J ) 7, 7 Hz
7.50, dd, J ) 7, 7 Hz
11.37, bs
8.89, d, J ) 7 Hz
7.30, dd, J ) 7, 7 Hz
7.48, dd, J ) 7, 7 Hz
7
7.61, d, J ) 7 Hz
10
14
16
17
18
10.96, s
-
8.50, s
13.57, s
-
7.07, s
7.70, s
7.75, bs
9.07, bs
3.15, s
3.45, s
Ta ble 2. 13C NMR Da ta for Did em n im id es a n d Gr a n u la tim id es. Recor d ed in DMSO-d 6
C no.
didemnimide A (1)6
didemnimide C (3)6
didemnimide Ed (3)
granulatimided (4)
isogranulatimided (5)
2
3
3a
4
5
6
7
7a
8
9
11
12
13
14
16
NMe
130.7
104.7
125.6
121.7
119.6
121.7
112.0
136.5
n.o.
131.8
105.2
125.0
119.9
120.8
122.6
112.5
136.6
134.1
172.0a
171.8a
134.1
122.6
131.8
140.4
32.3
133.4
104.4
124.3
119.1
121.3
122.9
112.9
136.7
138.2a
170.9b
171.2b
113.5a
122.3
124.3
137.2
34.3
135.4c
113.0
121.4
123.8
120.0
126.1
111.6
140.4
122.7a
169.8b
171.0b
109.5a
125.7c
133.4c
144.5
n.o.
97.9
121.0
122.4
121.9
124.8
112.4
135.5
126.4a
169.0b
170.0b
114.8a
n.o.e
172.9a
172.8a
n.o.
125.7
119.6
136.5
n.o.
n.o.
d
a-c May be interchanged. Assignments are based on HMQC and HMBC data. e Not observed.
data obtained for didemnimide E (3) were consistent with
the proposed structure.
Sch em e 1
The G2 checkpoint inhibitor gave a [M + H]+ ion in
the HRFABMS at m/z 277.0730 appropriate for a molec-
ular formula of C15H8N4O2. Its UV spectrum (MeOH)
showed bands at 210, 231, 280, and 470 nm consistent
with a polycyclic aromatic chromophore, and its IR
spectrum contained bands at 3294 and 1678 cm-1 indi-
cating NH and amide carbonyl groups. The molecular
formula of the inhibitor differed from that of didemnimide
A (1) only by the loss of two hydrogen atoms, and a
comparison of their NMR data (Tables 1 and 2) showed
that the two compounds were closely related. A NH
proton resonance at δ 11.11 gave HMBC correlations to
carbonyl resonances at δ 169.0 and 170.0 and sp2 carbon
resonances at δ 126.4 and 112.2, confirming the presence
of a maleimide substructure in the inhibitor. The COSY
spectrum identified a four-proton spin system (δ 8.51, d,
J ) 7 Hz (H-4)); 7.43, dd, J ) 7, 7 Hz (H-6); 7.35, dd, J
) 7, 7 Hz (H-5); 7.67, d, J ) 7 Hz (H-7)) that could be
imidazole fragment in the putative biosynthetic precursor
didemnimide A (1) (Scheme 1).
assigned to the H-4 to H-7 protons on an indole residue.
Irradiation of a broad proton resonance at δ 13.48
induced a NOE only in the aromatic doublet at δ 7.68,
which assigned the doublet to H-7 and the broad proton
resonance to the indole NH. The absence of resonances
Evidence in support of a planar polycyclic aromatic
structure for the checkpoint inhibitor could be found in
1
the H NMR data. The large chemical shift observed for
1
the H-4 resonance (δ 8.51) in the inhibitor relative to the
chemical shift observed for the H-4 resonance in didem-
nimide A (1) (δ 7.07) can be attributed to a neighboring
group effect from the C-9 maleimide carbonyl in either
candidate structure 4 or 5. A similar diamagnetic aniso-
in the H NMR spectrum of the inhibitor that could be
assigned to the indole H-2 or H-3 protons indicated the
indole nucleus had substituents at both C-2 and C-3.
Subtraction of the atoms present in the maleimide and
indole fragments (C12H6N2O2) from the molecular formula
of the inhibitor left C3H2N2. These atoms could be
accounted for by a disubstituted imidazole moiety, lead-
ing to the two polycyclic aromatic candidate structures
47 and 5 that represent alternative modes of forming a
bond between the C-2 carbon of the indole fragment and
either a carbon (C-14) or nitrogen atom (N-17) of the
tropic effect deshields the H-4 resonance (DMSO-d6
δ
9.30) relative to the H-8 resonance (δ 7.96) in staurospo-
rine (7).8 The presence of the neighboring group deshield-
(8) Takahashi, I.; Saitoh, Y.; Yoshida, M.; Sano, H.; Nakano, H.;
Morimoto, M.; Tamaoki, T. J . Antibiot. 1989, 42, 571-576.

Aromatic Alkaloids Isolated from a Brazilian Ascidian
J . Org. Chem., Vol. 63, No. 26, 1998 9853
Sch em e 2
ing effect of the maleimide carbonyl in the checkpoint
inhibitor and not in didemnimide A (1) is consistent with
the presence of a rigid planar heterocycle in the inhibitor,
in contrast to the uncyclized didemnimide A (1), where
the X-ray data shows that the indole and maleimide rings
are twisted relative to each other along the C-3 to C-8
bond.6
In principle, a variety of different NMR experiments
should readily distinguish between the candidate struc-
tures 4 and 5. However, the remaining two resonances
1
at δ 8.10 and 9.12 in the H NMR spectrum, that could
be assigned to the imidazole fragment of the inhibitor,
gave very broad signals that failed to show any HMBC/
HMQC correlations or NOEs precluding a clear spectro-
scopic resolution to the structural problem. The broad-
ness of the imidazole resonances in the NMR spectra of
the checkpoint inhibitor was attributed to a tautomeric
equilibrium involving the NH proton. Similar broadening
of the imidazole CH and NH proton resonances was
1
observed in the H NMR spectrum of didemnimide A (1)
and related analogues which do not have a methyl
substituent on one of the imidazole nitrogens.6 Didem-
nimide C (6) and E (3), which do have methyl substitution
on one of the imidazole ring nitrogens, give sharp, well-
resolved NMR spectra. By analogy with the didemnim-
ides, only one of the candidate structures for the check-
point inhibitor, i.e., 4, would be expected to undergo
tautomeric equilibrium involving an imidazole NH. There-
fore, the observed broadening of the 1H NMR signals
assigned to the imidazole fragment of the inhibitor
appeared to be most consistent with candidate structure
4. From a biogenetic perspective, it seemed reasonable
that the co-occurring alkaloid didemnimide A (1) repre-
sented a logical precursor to either candidate structure
4 or 5 (Scheme 1). Structure 4 is related to staurosporine
(7),9 arcyriaflavins,10 and other polycyclic naturally oc-
curring bisindole maleimides, whereas structure 5 is
apparently without precedent among natural products.
Therefore, from a biogenetic perspective, 4 was also
deemed to be the most probable structure.
Since it was not possible to distinguish between
candidate structures 4 and 5 using the available spec-
troscopic data, it was decided to resolve the issue by total
synthesis. The synthesis was also required to generate
sufficient material for complete biological evaluation.
Based on the combination of chemical and biogenetic
arguments presented above, candidate structure 4, which
we named granulatimide, was felt to be the most probable
structure for the checkpoint inhibitor, and it was chosen
as the initial synthetic target.
Sch em e 3
The total syntheses of didemnimide A (1) and granu-
latimide (4) are summarized in Scheme 3. Treatment of
the substituted imidazole 1114 with n-BuLi in THF at
-78 °C, followed by reaction of the resultant 5-lithio
imidazole intermediate with dimethyl oxalate, provided
the required R-keto ester 8 in 66% yield. The condensa-
tion of 8 with indole-3-acetamide (9) was effected ef-
ficiently by treatment of a solution of the two reactants
with t-BuOK in warm DMF in the presence of molecular
sieves. The use of molecular sieves was crucial to the
success of this reaction, since, in the absence of this
material, a second major product resulting from displace-
ment of the phenylthio group by a methoxyl function (i.e.,
10, OMe in place of SPh) was formed.
The highly convergent total synthesis that was envis-
aged, as outlined in general terms in Scheme 2, was to
include two key steps. The first of these was the conden-
sation of the R-keto ester 8 with indole-3-acetamide (9),
which was expected11 to furnish the maleimide 10. The
second key transformation was envisioned to be the
photochemically induced bond formation12 between the
indole C-2 carbon and the imidazole C-4 carbon (C-14)
of the natural product didemnimide A (1)6,13 to afford
granulatimide (4). Presumably, 1 would be readily de-
rived by removal of the two protecting groups from the
imidazole moiety of 10.
(11) Related condensations of indole-3-acetamide and other mono-
substituted acetamides with dialkyl oxalates have been reported. See,
for example, (a) Rooney, C. S.; Randall, W. C.; Streeter, K. B.; Zeigler,
C.; Cragoe, E. J .; Schwam, H.; Michelson, S. R.; Williams, H. W. R.;
Eichler, E.; Duggan, D. E.; Ulm, E. H.; Noll, R. M. J . Med. Chem. 1983,
26, 6, 700. (b) Neel, D. A.; J irousek, M. R.; McDonald, J . H. Bioorg.
Med. Chem. Lett. 1998, 8, 47. After the work reported herein had been
completed and accepted for publication, a paper describing condensa-
tions similar to 8 + 9 f 10 was published: Faul, M. M.; Winneroski,
L. L.; Krumrich, C. A. J . Org. Chem. 1998, 63, 6053.
(12) For related cyclizations, see (a) Rawal, V. H.; J ones, R. J .; Cava,
M. P. Tetrahedron Lett. 1985, 26, 2423. (b) Rawal, V. H.; J ones, R. J .;
Cava, M. P. J . Org. Chem. 1987, 52, 2, 19. (c) Terpin, A.; Winklhofer,
C.; Schumann, S.; Steglich, W. Tetrahedron 1998, 54, 1745.
(13) For the first total synthesis of a didemnimide (didemnimide
C) via a route less convergent than that reported herein for didem-
nimide A, see ref 12c.
(9) Tamaoki, T.; Nomoto, H.; Takahashi, I.; Kato, Y.; Morimoto, M.;
Tomita, F. Biochem. Biophys. Res. Commun. 1986, 135, 397.
(10) Steglich, W.; Steffan, B.; Kopanski, L.; Eckhardt, G. Angew.
Chem., Int. Ed. Engl. 1980, 19, 459.
(14) Achab, S.; Guyot, M.; Potier, P. J . Am. Chem. Soc. 1978, 100,
3918, and references cited therein.

9854 J . Org. Chem., Vol. 63, No. 26, 1998
Berlinck et al.
Sch em e 4
Treatment of the tetracycle 10 with W-2 Raney nickel
in refluxing EtOH15 furnished an excellent yield of the
desulfurized product 12. Removal of the methoxymethyl
group was accomplished by reaction of the latter sub-
stance with 10 equiv of boron tribromide in refluxing
CH2Cl2. Reaction at room temperature and (or) use of
lesser amounts of the reagent resulted in the recovery of
considerable amounts of starting material. The product
of this transformation, didemnimide A (1), exhibited
spectral data identical with those reported6 for the
natural product.
Photolysis of a solution of didemnimide A (1) in
acetonitrile containing a small amount of palladium on
carbon12a produced, in 91% yield, granulatimide (4), and
in 8% yield, isogranulatimide (5). The spectroscopic data
obtained for granulatimide (4) were in full accord with
the assigned structure. A series of difference NOE
experiments were particularly instrumental in assigning
the 1H NMR spectra and confirming the structure.
Irradiation of a broad singlet at δ 12.58 induced a NOE
in a doublet at δ 7.61, thereby assigning the broad singlet
to the indole NH and the doublet to H-7. Similarly,
irradiation of the broad singlet at δ 13.57 induced a NOE
in a sharp singlet at δ 8.5, which led to the assignment
of the downfield singlet to the imidazole NH and the
upfield singlet to H-16. It is interesting to note, that
contrary to our expectations, granulatimide (4) gives
sharp, well-resolved 1H NMR resonances for all of the
hydrogen atoms in the molecule. The imidazole NH in
granulatimide (4) is strongly deshielded (i.e., δ 13.57),
and it does not show a NOE to the indole NH, suggesting
that it participates in a strong intramolecular hydrogen
bond to the maleimide C-11 carbonyl. Semiempirical
calculations carried out on the HN-17 and HN-15 tau-
tomers of granulatimide (4) showed that the HN-17
tautomer is approximately 7 kcal lower in energy.16
Calculations also show that the HN-15 tautomer has a
severe steric interaction between the indole NH (N-1) and
imidazole NH (N-15) that leads to a distortion of the
planar aromatic heterocycle. The combination of an
intramolecular hydrogen bond stabilizing the HN-17
tautomer and a severe steric interaction destabilizing the
HN-15 tautomer may be responsible for suppressing the
imidazole NH tautomerization in granulatimide (4).
Comparison of the NMR data for synthetic granula-
timide (4) with the data for the checkpoint inhibitor
isolated from the extract of the first D. granulatum
collection (Tables 1 and 2) clearly revealed that the two
molecules were different. Therefore, we concluded that
the G2 checkpoint inhibitor must be isogranulatimide (5),
the only alternate candidate structure consistent with
the spectroscopic data and a biogenesis from didemnim-
ide A (1) (Scheme 1). A minor product from the photo-
chemical conversion of didemnimide A (1) to granula-
timide (4) had TLC properties identical with the natural
isogranulatimide (5). Their 1H NMR spectra were also
nearly identical in all respects, except for the two
resonances that could be assigned to the imidazole H-14
and H-16 protons. The synthetic product gave sharp,
well-resolved singlets at δ 7.88 and 8.93, while the
natural sample gave broad resonances at δ 8.10 and 9.12.
These NMR discrepancies were attributed to the use of
different chromatographic procedures to purify the two
samples resulting in different protonation states for the
natural and synthetic isogranulatimides. Therefore, both
the natural and synthetic samples were repurified using
identical reversed phase HPLC conditions (1:1 CH3CN-
0.05% TFA) resulting in them having identical HPLC
1
retention times and H NMR spectra. The formation of
isogranulatimide (5) via the photolysis of didemnimide
A (1), while unanticipated, confirms the structure of the
natural product. Scheme 4 proposes a possible radical
mechanism for this interesting transformation.17
During the investigation of the photochemical conver-
sion of didemnimide A (1) to granulatimide (4), it was
discovered that the reaction proceeded rapidly and in
good yield in DMSO upon irradiation with natural
sunlight and oxidation with air. D. granulatum grows in
very shallow water where it is exposed to intense sunlight
and didemnimide A (1) is a major constituent of D.
granulatum extracts, so it seemed reasonable to expect
that granulatimide (4) would be a co-occurring metabo-
lite. Granulatimide (4) is extremely insoluble in most
common organic solvents including methanol and etha-
nol, and it can only be effectively solubilized with DMF
or DMSO. Therefore, in hindsight it was apparent that
our standard extraction procedures using ethanol/
methanol would be expected to yield only small amounts
of 4. Nevertheless, with synthetic granulatimide (4) as a
TLC reference it was possible to identify 4 in chroma-
tography fractions generated during purification of D.
granulatum alkaloids. Unfortunately, the extracted as-
cidian tissue had been thrown away by the time the
solubility and chromatographic properties of synthetic
granulatimide (4) were discovered. Therefore, the true
concentration of granulatimide in D. granulatum and the
issue of whether the granulatimide found in the chro-
matography fractions is an artifact formed by photo-
chemical transformation of didemnimide A (1) during
fractionation or a bona fide natural product remain to
be resolved by a future collection of D. granulatum.
(15) Ohta, S.; Yamamoto, T.; Kawasaki, I.; Yamashita, M.; Katsuma,
H.; Nasako, R.; Kobayashi, K.; Ogawa, K. Chem. Pharm. Bull. 1992,
40, 2681.
(16) Calculations were performed using HyperChem release 4 for
Windows, Polack-Ribiere Alogrithm, in vacuo.
(17) See: CRC Handbook of Organic Photochemistry and Photobi-
ology; Harspool, W. M., Ed., CRC Press: New York, 1995; p 513.

Aromatic Alkaloids Isolated from a Brazilian Ascidian
J . Org. Chem., Vol. 63, No. 26, 1998 9855
pure didemnimide A (1), didemnimide D (2), didemnimide E
(3) (0.012% wet wt.), and isogranulatimide (5). Once synthetic
granulatimide (4) was in hand as an authentic TLC reference
(see below), small amounts of 4 were also identified in the
pooled chromatography fractions.
Did em n im id e E (3). Obtained as an orange amorphous
solid; for 1H NMR and 13C NMR, see Tables 1 and 2;
HRFABMS m/z 293.1048 [M + H]+, calcd for C16H13N4O2,
293.1039.
Granulatimide (4) and isogranulatimide (5) are novel
alkaloids that both possess heterocyclic aromatic skel-
etons without precedent in natural products. Biogeneti-
cally they represent alternative modes of cyclization of
the putative precursor didemnimde A (1) (Scheme 1). A
short and highly efficient biomimetic synthesis that gives
granulatimide (4) in high yield and isogranulatimide (5)
in relatively low yield has been completed (Scheme 3).
The facile photochemical transformation of didemnimide
A (1) into both granulatimide (4) and isogranulatimide
(5) in the laboratory suggests that a similar photochemi-
cal reaction may be responsible for their formation in D.
granulatum. Granulatimide (4) and isogranulatimide (5)
are both G2 specific cell cycle checkpoint inhibitors with
in vitro IC50 values in the range of 1 to 1.8 µM.5 These
novel alkaloids represent a new structural class of G2
checkpoint inhibitors and the first one discovered as part
of a rational screening program. The availability of
synthetic granulatimide (4) and isogranulatimide (5)
should facilitate their use as tools to test, for the first
time in vivo, the hypothesis that a G2 checkpoint inhibi-
tor combined with a DNA damaging agent will lead to
selective killing of p53- cancer cells.
Isogr a n u la tim id e (5). Obtained as a red amorphous solid;
UV (MeOH) λmax (ꢀ) 210 (10234), 231 (10648), 280 (6552), 470
(1876); IR (film) 3294, 1678, 1586, 1568, 1421, 1367, 1205,
1
1135, 802, 745, 724 cm-1; for H NMR and 13C NMR, see Tables
1 and 2; HRFABMS m/z 277.0730 [M + H]+, calcd for
C15H9N4O2, 277.0714.
Syn t h esis. Met h yl 2-(1-(m et h oxym et h yl)-2-(p h en yl-
th io)im id a zol-5-yl)glyoxyla te (8). To a cold (-78 °C), stirred
solution of 1-(methoxymethyl)-2-(phenylthio)imidazole15 (11)
(340 mg, 1.55 mmol) in dry THF (8.0 mL) was added a solution
of n-BuLi (1.47 M in hexanes, 1.26 mL, 1.85 mmol) and the
mixture was stirred for 1 h. A solution of dimethyl oxalate (540
mg, 4.58 mmol) in dry THF (2.0 mL) was added, and the
mixture was stirred at -78 °C for an additional 1.25 h. The
reaction mixture was treated with saturated aqueous NH4Cl
(5.0 mL) and Et2O (20 mL). The phases were separated, and
the aqueous layer was extracted with Et2O (25 mL). The
combined organic extracts were washed (brine, 10 mL), dried
(MgSO4), and concentrated. Flash chromatography (35 g of
silica gel, 1:1 petroleum ether-Et2O) of the crude material
afforded 305 mg (66%) of 8 as a beige solid that exhibited mp
59-60 °C; IR (KBr) 1729, 1657, 1305, 1273, 1167, 1114, 748
Exp er im en ta l Section
Gen er a l Exp er im en ta l P r oced u r es. TLC was carried out
by using commercial aluminum-backed silica gel 60 plates.
Flash chromatography18 was carried out with 230-400 mesh
silica gel (E. Merck). THF was distilled from sodium/benzophe-
none, and CH2Cl2 was distilled from CaH2. Commercial EtOH
(reagent grade) and MeCN (HPLC grade) were used without
further purification. Molecular sieves were dried under vacuum
with heating for 5 h prior to use. All reactions were carried
out under an atmosphere of dry argon using glassware that
had been thoroughly flame dried.
Collection a n d Extr a ction of D. gr a n u la tu m . The
ascidian D. granulatum (85 g wet wt.) was collected during
the summer of 1995 at depths of 1 m at the rocky shore of
Araca beach, Sao Sebastiao (Sao Paulo state, southeastern
Brazil). A second collection was made at the Arquipelago do
Arvoredo (Santa Catarina state, southern Brazil. 150 g wet
wt.) and in the Sao Sebastiao Channel (Sao Paulo state. 185
g wet wt.) during November 1997. Freshly collected animals
were stored in ethanol at -20 °C. Animals obtained from the
different collections were processed separately as follows: after
decantation of ethanol, the animal was blended and extracted
three times with methanol. The ethanol and methanol extracts
were combined, filtered, and evaporated in vacuo. The organic
material was dissolved in 8:2 MeOH-CH2Cl2 and filtered for
elimination of inorganic salts.
Isola tion a n d Str u ctu r e Deter m in a tion . The organic
extract (1.7 g) obtained from the first collection of D. granu-
latum was subjected to chromatography on Sephadex LH20
(MeOH), affording eight fractions with clearly distinctive
constituents by TLC (CH2Cl2-MeOH 9:1, UV at 254 nm). The
seventh fraction (0.0131 g) which exhibited G2 cell cycle
checkpoint inhibition activity contained almost pure isogranu-
latimide (5). RPHPLC purification of this fraction (1:1 CH3CN-
0.05% TFA) yielded pure isogranulatimide (5) (0.006% wet
wt.). RPHPLC purification (30:70 CH3CN-0.05% TFA) of
individual inactive fractions yielded didemnimides A (1)
(0.004% wet wt.)6 and D (2) (0.004% wet wt.).6
The organic extracts obtained (total weight: 5.8 g) from the
second collection were separately subjected to chromatography
on Sephadex LH20 (MeOH), giving fractions with compositions
similar to those from the first collection as determined by TLC
and RP HPLC analyses. These fractions were pooled according
to TLC properties and further separated by RP HPLC to give
cm-1 1H NMR (CDCl3, 400 MHz) δ 8.23 (s,1 H), 7.52-7.54
;
(m, 2 H), 7.37-7.38 (m, 3 H), 5.81 (s, 2 H), 3.91 (s, 3 H), 3.36
(s, 3 H); 13C NMR (CDCl3, 100 MHz) δ 172.2, 161.6, 154.1,
145.4, 145.3, 133.2, 129.5, 129.2, 128.7, 75.8, 56.5, 53.0;
HREIMS calcd for C14H14N2O4S 306.0674, found 306.0674.
Anal. Calcd C, 54.89; H, 4.61; N, 9.15. Found: C, 55.00; H,
4.66; N, 8.94.
3-[4-(1-(Meth oxym eth yl)-1H-2-(p h en ylth io)im id a zol-5-
yl)-2,5-d ioxo-2,5-d ih yd r o-1H-p yr r ol-3-yl]in d ole (10). To a
stirred solution of t-BuOK (150 mg, 1.34 mmol) in DMF (3.0
mL) at room temperature were added sequentially 4 Å mo-
lecular sieves (∼500 mg) and a solution of 8 (167 mg, 0.54
mmol) and indole-3-acetamide 9 (114 mg, 0.66 mmol) in DMF
(4.0 mL). The reaction mixture was heated to 45 °C and stirred
for 12 h. The dark purple solution was treated with hydro-
chloric acid (1 N, 3.0 mL) and EtOAc (30 mL). The phases were
separated, and the aqueous layer was extracted with EtOAc
(2 × 10 mL). The combined organic extracts were washed
(brine, 4 × 15 mL), dried (MgSO4), and concentrated. Flash
chromatography (25 g of silica gel, 30:1 CH2Cl2-MeOH) of the
crude material afforded 208 mg (90%) of 10 as an orange solid
that exhibited mp 243 °C; IR (KBr) 3400-2600, 1765, 1708,
1537, 1341, 741 cm-1; 1H NMR (400 MHz) δ 12.11 (br s, 1 H),
11.25 (br s, 1 H), 8.14 (s, 1 H), 7.46 (d, 1 H, J ) 7.9 Hz), 7.18-
7.33 (m, 6 H), 7.14 (dd, 1 H, J ) 8, 8 Hz), 6.77 (dd, 1 H, J )
8, 8 Hz), 6.51 (d, 1 H, J ) 8.0), 5.00 (s, 2 H), 2.97 (s, 3 H); 13
C
NMR (100 MHz) δ 171.7, 171.6, 140.3, 136.7, 135.3, 133.8,
133.5, 132.5, 129.5, 128.4, 127.2, 125.7, 124.3, 122.7, 120.8,
120.2, 117.4, 112.6, 104.6, 75.8, 55.9; HREIMS calcd for
C
23H18N4O3S 430.1099, found 430.1099.
3-[4-(1-(Met h oxym et h yl)-1H -im id a zol-5-yl)-2,5-d ioxo-
2,5-d ih yd r o-1H-p yr r ol-3-yl]in d ole (12). To a refluxing solu-
tion of 10 (52 mg, 0.12 mmol) in EtOH (5.0 mL) was added
Raney Ni (W-2, 50% slurry in water, ∼150 mg), and the
suspension was refluxed for 1 h. An additional amount of
Raney Ni (W-2, 50% slurry in water, ∼100 mg) was added,
and the mixture was refluxed for an additional 3 h. The
reaction mixture was then cooled to room temperature and
filtered through Celite. The Celite was washed with CH2Cl2-
MeOH (1:1, 75 mL), and the combined organic washes were
concentrated. Flash chromatography (14 g of silica gel, 20:1
CH2Cl2-MeOH) of the crude material afforded 33 mg (85%)
of 12 as an orange solid: mp > 235 °C dec; IR (KBr) 3400-
2600, 1765, 1703, 1537, 1440, 1342, 1113 cm-1; 1H NMR (400
(18) Still, W. C.; Kahn, M.; Mitra, A. J . Org. Chem. 1978, 43, 2923.

9856 J . Org. Chem., Vol. 63, No. 26, 1998
Berlinck et al.
MHz) δ 12.03 (br s, 1 H), 11.18 (br s, 1 H), 8.07 (s, 1 H), 7.95
(s, 1 H), 7.44 (d, 1 H, J ) 8.5 Hz), 7.09 (dd, 1 H, J ) 7, 7 Hz),
7.01 (s, 1 H), 6.79 (dd, 1 H, J ) 7, 7 Hz), 6.45 (d, 1 H, J ) 8.0
Hz), 5.02 (s, 2 H), 3.07 (s, 3 H); 13C NMR (100 MHz) δ 171.8,
171.7, 140.5, 136.5, 134.3, 132.9, 131.8, 124.3, 122.4, 121.7,
120.5, 120.3, 118.4, 112.3, 104.7, 76.4, 55.6; HREIMS cacld
for C17H14N4O3 322.1066, found 322.1066.
(30 mL), and the resultant solution was washed (brine, 4 ×
20 mL), dried (MgSO4), and concentrated. Flash chromatog-
raphy (20 g of silica gel, 10:1 CH2Cl2-MeOH) of the crude
material afforded 11.7 mg (91%) of 4 as a yellow solid and 1.0
mg (8%) of 5 as a red solid. Final purification was achieved on
RPHPLC (1:1 CH3CN-0.05% TFA). Gr a n u la tim id e (4): IR
(KBr) 3246, 2925, 1743, 1698, 1328, 1227, 743 cm-1; for 1H
NMR (400 MHz), see Table 1; for 13C NMR, see Table 2;
Did em n im id e A (1). To a stirred suspension of 12 (67.4
mg, 0.209 mmol) in CH2Cl2 (15.0 mL) at room temperature
was added a solution of BBr3 in CH2Cl2 (1.0 M, 2.1 mL, 2.1
mmol), and the deep blue solution was heated under reflux
for 5 h. The mixture was cooled to room temperature and
treated with saturated aqueous NaHCO3 (10 mL) and EtOAc
(20 mL) and then was stirred for 0.25 h. The phases were
separated, and the aqueous layer was extracted with EtOAc
(2 × 15 mL). The combined organic extracts were washed
(brine, 10 mL), dried (MgSO4), and concentrated. Flash chro-
matography (20 g of silica gel, 15:1 CH2Cl2-MeOH) of the
crude material afforded 15.8 mg of recovered 12 as well as
45.0 mg (77%, 100% based on recovered starting material) of
1 as an orange solid: IR (KBr) 3247, 1760, 1708, 1557, 1423,
HREIMS calcd for
C15H8N4O2 276.0647, found 276.0647.
Isogr a n u la tim id e (5): see above and Tables 1 and 2.
Ack n ow led gm en t. The authors thank Professor
J ose Carlos de Freitas, director of the Centro de Biologia
Marinha of the Universidade de Sao Paulo, for providing
many facilities during the ascidian collection, and M.
Netherton and J . Scheffer for help with the photoreac-
tions. R.G.S.B. thanks the Fundacao de Amparo a
Pesquisa do Estado de Sao Paulo (FAPESP) for a
visiting professor fellowship (97/06799-6). Financial
support was provided by grants from FAPESP (to
R.G.S.B.), the National Cancer Institute of Canada (to
R.J .A.), the Natural Sciences and Engineering Research
Council of Canada (a Postgraduate Fellowship to R.B.
and grants to R.J .A. and E.P.), and the Canadian Breast
Cancer Research Initiative (to M.R.).
1
1345, 1240, 747 cm-1; H NMR (400 MHz, major tautomer) δ
12.45 (br s, 1 H), 11.66 (br s, 1 H), 10.87 (br s, 1 H), 8.05 (s, 1
H), 7.71 (br s, 1 H), 7.68 (br s, 1 H), 7.39 (br d, 1 H, J ) 7.8
Hz), 7.07 (br m, 2 H), 6.87 (br m, 1 H); 13C NMR (100 MHz) δ
172.8, 172.6, 136.1, 135.9, 130.9, 130.5, 126.0, 126.0, 121.7,
121.3, 119.7, 119.2, 112.2, 111.5, 105.0; HREIMS calcd for
C
15H10N4O2 278.0804, found 278.0804.
Gr a n u la tim id e (4) a n d Isogr a n u la tim id e (5). To a
Su p p or tin g In for m a tion Ava ila ble: NMR spectra for
didemnimide A (3), synthetic granulatimide (4), natural and
synthetic isogranulatimide (5), synthetic didemnimide A (1),
and the synthetic intermediates 8, 10, and 12 (41 pages). This
material is contained in libraries on microfiche, immediately
follows this article in the microfilm version of the journal, and
can be ordered from the ACS; see any current masthead page
for ordering information.
solution of 1 (12.9 mg, 0.046 mmol) in MeCN (5.0 mL) was
added a catalytic amount of palladium-on-carbon (10% Pd),
and the resulting mixture was sparged with argon for 0.5 h.
The mixture was irradiated (450 W Hanovia medium-pressure
mercury vapor lamp, quartz reaction vessel) for 6.5 h. The
reaction mixture was filtered through Celite, the Celite was
washed with DMF (10 mL), and the combined filtrate was
concentrated. The remaining material was taken up in EtOAc
J O981607P