Synthetic route to the GE3 cyclodepsipeptide.

Article abstract of DOI:10.1021/ol025895u

[reaction: see text] A reasonably efficient [2 + 2 + 2] fragment condensation strategy has been developed for assembling the cyclodepsipeptide sector of GE3 that involves 5-7. A Carpino HATU-mediated macrolactamization was used to close the 19-membered cy

Full text of DOI:10.1021/ol025895u

ORGANIC
LETTERS
2002
Vol. 4, No. 11
1903-1906
Synthetic Route to the GE3
Cyclodepsipeptide
Karl J. Hale* and Linos Lazarides
The Christopher Ingold Laboratories, Department of Chemistry, UniVersity College
London, 20 Gordon Street, London WC1H 0AJ, United Kingdom
k.j.hale@ucl.ac.uk
Received March 19, 2002
ABSTRACT
A reasonably efficient [2 + 2 + 2] fragment condensation strategy has been developed for assembling the cyclodepsipeptide sector of GE3
that involves 5−7. A Carpino HATU-mediated macrolactamization was used to close the 19-membered cyclodepsipeptide ring.
Medical interest in the antitumor agent GE3 has risen steadily
ever since it was found to display potent growth inhibitory
effects against solid human tumors in mice and it was
suggested to have a novel mechanism of action.1
GE32 and its sister molecule A83586C major prizes for total
synthesis.3 In this paper, we now report a reasonably efficient
synthetic route to the cyclodepsipeptide sector of GE3. We
believe that it will not only expedite a future chemical
synthesis of the natural product but also allow for the creation
of a diverse collection of analogues. The work reported
herein constitutes a further development and augmentation
of our previous synthetic studies on this class.3
Specifically, a single 2 mg/kg dosage of GE3 brings about
a 47% reduction in tumor size in immunocompromised mice
transplanted with PSN-1 human pancreatic carcinoma after
only 11 days,1 and importantly, at this low dosage, GE3 is
neither lethal nor highly toxic. GE3 is thought to act by
preventing deregulated E2F/DP transcription factors from
binding to target genes that encode for proteins critically
involved in cancer cell division, a mechanism that has yet
to be clarified and confirmed by other biology groups. In
the case of A431 human lung cancer cells, this mode of
action is correlated with an ability to halt cell cycle
progression from the G1 to S phase. GE3 is also able to
selectively inhibit cyclin A gene expression in Saos-2 cells
without repressing â-actin gene expression. These combined
biological effects and the current dearth of supply have made
The main points of difference between GE3 and A83586C
lie in the N-methyl D-Leu residue, which replaces the
N-methyl-D-Ala unit in the cyclodepsipeptide core, and in
(2) For a synthesis of the GE3 acyl side chain, see: Makino, K.; Henmi,
Y.; Hamada, Y. Synlett 2002, 613.
(3) For the first total synthesis of A83586C, see: (a) Hale, K. J.; Cai, J.
J. Chem. Soc. Chem. Comm. 1997, 2319. (b) Hale, K. J.; Cai, J.; Delisser,
V. M. Tetrahedron Lett. 1996, 37, 9345-9348. (c) Hale, K. J.; Cai, J.
Tetrahedron Lett. 1996, 37, 4233-4236. (d) Hale, K. J.; Cai, J.; Manaviazar,
S.; Peak, S. A. Tetrahedron Lett. 1995, 36, 6965-6968. (e) Hale, K. J.;
Delisser, V. M.; Yeh, L.-K.; Peak, S. A.; Manaviazar, S.; Bhatia, G. S.
Tetrahedron Lett. 1994, 35, 7685-7688. (f) Hale, K. J.; Manaviazar, S.;
Delisser, V. M. Tetrahedron 1994, 50, 9181-9188. (g) Hale, K. J.; Bhatia,
G. S.; Peak, S. A.; Manaviazar, S. Tetrahedron Lett. 1993, 34, 5343-5346.
(h) For a detailed account of our A83586C synthesis and the synthesis of
other complex cyclodepsipeptides, see: Hale, K. J.; Bhatia, G. S.; Frigerio,
M. The Chemical Synthesis of Natural Products; Hale, K. J., Ed.; Sheffield
Academic Press: Sheffield, 2000; Chapter 12, p 349.
(1) Sakai, Y.; Yoshida, T.; Tsujita, T.; Ochiai, K.; Agatsuma, T.; Saitoh,
Y.; Tanaka, F.; Akiyama, T.; Akinaga, S.; Mizukami, T. J. Antibiot. 1997,
50, 659.
10.1021/ol025895u CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/02/2002

the nature of the C(29)-tertiary alcohol alkyl grouping that,
in the case of GE3, is a methyl rather than an ethyl. Although
these two changes might appear rather innocuous to the
unitiated in this field, our most recent synthetic experiences
have shown that they will almost certainly necessitate that a
different synthetic strategy be developed for setting the
C(29)-OH and that modified tactics be adopted for assembly
of the cyclodepsipeptide.
The synthesis of dipeptide 5 commenced from known (3S)-
N(1)-Z-N(2)-Fmoc-piperazic acid 8 (Scheme 2), which is
Scheme 2. Synthesis of Linear Hexapeptide 15a
Our overall retrosynthetic planning for GE3 (Scheme 1)
is currently predicated upon the biogenetically inspired
Scheme 1. Retrosynthetic Analysis of GE3
a Reagents and conditions: (a) DCC (1.1 equiv), BocNHNH2
(1.2 equiv), THF (0.3 M), 0 °C, 1 h, and then rt, 20 h. (b) Et2NH
(40 equiv), MeCN (0.12 M), rt, 15 min. (c) 10 (1 equiv), 11 (1
equiv), AgCN (1.5 equiv), C6H6 (0.3 M), 80 °C, 50 min. (d) Et2NH
(40 equiv), MeCN (0.12 M), rt, 15 min. (e) Premix acid 6 with
BOP-Cl at -20 °C in CH2Cl2 (0.15 M), add in collidine (1.1 equiv)
over 1 min, stir for 20 min, and then add solid 5 (1 equiv) followed
by collidine (1.1 equiv) over 1 min, warm to rt, and stir for 16 h.
(f) Et2NH (40 equiv), MeCN (0.12 M), rt, 15 min. (g) 14 (1 equiv),
7 (1 equiv), AgCN (1.5 equiv), C6H6 (0.12 M), 80 °C, 3-5 min.
coupling of cyclodepsipeptide 3 with the activated benzo-
triazole ester 2; a strategy that will deliberately avoid the
use of protecting groups at the final stages of the synthesis.
Our proposed pathway to the cyclodepsipeptide 3 was
founded upon the successful [2 + 2 + 2] union of 5-7.
readily accessible through our recently introduced tandem
electrophilic hydrazination/nucleophilic cyclization protocol.3e,4
Compound 8 was initially converted to the acyl hydrazide 9
by treatment with BocNHNH2, DCC, and THF over a 20 h
1904
Org. Lett., Vol. 4, No. 11, 2002

period.5 The Fmoc group6 was then excised from 9 with
diethylamine in acetonitrile and 10 chemoselectively coupled
with Fmoc-N-methyl-D-leucinoyl chloride 117 under silver
cyanide-assisted coupling conditions.8 The latter reaction
proceeded cleanly, providing the desired dipeptide 12 in 90%
yield. Next, the Fmoc group was detached from 12 in 84%
yield and amine 5 carefully purified by flash chromatography
prior to attempting its coupling with known 6.3e
juncture, we cleaved the two Boc groups from 15 with
trifluoroacetic acid in CH2Cl2 (Scheme 3) and oxidized the
Scheme 3. Completion of the GE3 Cyclodepsipeptide 3a
A wide range of conditions were evaluated for effecting
the desired amidation, but virtually all either performed badly
or failed, including the BOP-Cl/Et3N system9 that had
preViously worked so successfully in our A83586C synthesis.
After much effort, we eventually discovered that the com-
bination of BOP-Cl and collidine could unite these two
fragments; this regime afforded the pure tetrapeptide 13 in
66% yield.
The enormously beneficial effect of collidine upon the
course of this coupling cannot be understated. Presumably,
its much greater steric bulk, compared with that of Et3N,
helped prevent it from initiating base-catalyzed decomposi-
tion of the intermediary mixed anhydride derived from 6 and
thwarted its cleavage of the Fmoc group at temperatures
above 0 °C. The latter was especially problematic when Et3N
was employed as the base, due to the much slower rate of
coupling of 5 and 6. This was the result of the enhanced
steric hindrance around the N-methyl D-leucine nitrogen in
5, compared with that around its A83586C N-methyl-D-Ala
dipeptide counterpart.3a Breakdown of the mixed anhydride
by Et3N could potentially be initiated by R-deprotonation to
give a ketene intermediate or, alternatively, by enolization
of the (3R)-Piz residue and subsequent intramolecular attack
of the enolate upon the mixed anhydride. These two events
(along with Fmoc deprotection) could be expected to
contribute significantly to the observed decomposition when
Et3N is used as the base. The new BOP-Cl/collidine
modification solves this thorny problem very effectively and
was central to our eventual success in this venture.
After excising the Fmoc group from the (3R)-piperazic
acid residue of 13, the final fragment condensation was
attempted with the previously prepared acid chloride 7.3a The
silver cyanide-mediated8 [4 + 2] coupling of 14 with 7 was
much less successful than the analogous coupling in our
A83586C synthesis.3a After some optimization, however, a
workable 65% yield of 15 was ultimately obtained.10 At this
(4) Hale, K. J.; Cai, J.; Delisser, V.; Manaviazar, S.; Peak, S. A.; Bhatia,
G. S.; Collins, T. C.; Jogiya, N. Tetrahedron 1996, 52, 1047.
(5) (a) BocNHNH2: Boissonnas, R. A. ; Guttman, St.; Jaquenoud, P. A.
HelV. Chim. Acta 1960, 43, 1349. (b) For the use of TrocNHNH2/DCC to
make Troc acyl hydrazides, see: Fujii, N.; Yajima, H. J. Chem. Soc., Perkin
Trans. 1, 1981, 804.
(6) Carpino, L. A.; Han, G. Y. J. Org. Chem. 1972, 37, 3404.
(7) Fmoc-N-Methyl-D-leucine was prepared according to the method in:
Freidinger, R. M.; Hinkle, J. S.; Perlow, D. S.; Arison, B. H. J. Org. Chem.
1983, 48, 77. It was converted to Fmoc-N-methyl-D-leucinoyl chloride by
treatment with (COCl)2 and C6H6 at rt for 2 h and used without purification
after removal of the solvent and excess reagent in Vacuo.
(8) AgCN-mediated N-acylation: Durette, P. L.; Baker, F.; Barker, P.
L.; Boger, J.; Bondy, S. S.; Hammond, M. L.; Lanza, T. J.; Pessalano, A.;
Caldwell, C. G. Tetrahedron Lett. 1990, 31, 1237.
a Reagents and conditions: (a) CF3CO2H (200 equiv), CH2Cl2
(0.06 M), 0 °C, 2 h. (b) NBS (2 equiv), THF:H2O (1:1) (0.04 M),
rt, 2 h. (c) Add 4 (1 equiv) in CH2Cl2 (0.00086 M) to HATU (10
equiv) and NEM (13.5 equiv) in CH2Cl2 (0.00086 M) at 0 °C over
6 h, and then stir at 0 °C for 2 h, and at rt for 60 h. (d) Zn dust (85
equiv), AcOH:H2O (10:1) (0.3 M), rt, 25 min. (e) BnOC(O)Cl (3
equiv), 10% aq NaHCO3/CH2Cl2 (0.1 M), rt, 1 h. (f) H2 (1 atm),
MeOH (0.01 M), HCl (1 equiv), 10% Pd on C (0.6 g per g of 17),
rt, 16 h.
(9) Tung, R. D.; Rich, D. H. J. Am. Chem. Soc. 1985, 107, 4342.
(10) It is important not to prolong the heating of this reaction much
beyond the 3-5 min time frame we have recommended; otherwise,
decomposition ensues.
liberated N-acyl hydrazine with N-bromosuccinimide in THF
and water to obtain acid 4.11 A Carpino HATU-mediated
cyclization12 was then effected at a very high dilution in CH2-
Org. Lett., Vol. 4, No. 11, 2002
1905

Cl2 (0.00043 M with respect to 4). Fortunately, this protocol
closed the 19-membered cyclodepsipeptide ring reasonably
successfully, the desired macrolactam 16 being isolated in
40% overall yield for the three steps from 15. In this
particular instance, the N-methyl D-leucine residue exerted
a beneficial effect on the outcome of the reaction. It was a
result that contrasted sharply with the macrolactamization
in A83586C, where a 25% yield of product was obtained
solely from the cyclization.3a
To complete the synthesis of 3, we deprotected the Troc
group of 16 with Zn dust in aqueous acetic acid13 and
temporarily capped the liberated nitrogen with a Z-group to
facilitate the final purification of 17. The latter was then
cleanly deprotected by catalytic hydrogenation at atmospheric
pressure over a 10% Pd on carbon catalyst in anhydrous
methanol containing 1 equiv of HCl.3a Deprotection in an
acidic medium helps prevent an O- to N-acyl shift from
occurring in the â-hydroxyleucine residue.14 The latter can
be a problem when such systems are hydrogenated for
protracted periods under neutral or mildly basic conditions.
The final product 3 was obtained in a fairly pure condition
as judged by 500 MHz 1H NMR and HRMS analysis. In the
HRMS, an (M + H)+ ion was observed at m/e 669.3962,
which correlated closely with the calculated value of m/e
669.3936 for C30H53N8O9 (see Supplementary Information).
Proof that the lactone linkage was present was provided by
the IR spectrum of 3, which showed a strong ester CdO
stretching absorption at 1741 cm-1. It is our recommendation
that 3 be used for subsequent couplings without any further
purification.15
One item of note in our synthesis is the global O-
debenzylation of 17, which took place under much less
forcing condtions than were needed for the A83586C
synthesis. In that instance, the cyclodepsipeptide precursor
required 24 h of hydrogenation at 75 psi before it was fully
deprotected. The ease of the present deprotection and the
markedly improved yield of the GE3 macrolactamization
reaction clearly bode well for the development of a future
commercial synthesis of this molecule. These observations,
along with our findings on the [2 + 2] coupling of 5 with 6,
highlight how a change of only one amino acid residue can
dramatically affect the success of a route to a linear or cyclic
peptide.
In summary, a reasonably efficient synthesis of the GE3
cyclodepsipeptide has been developed. Of considerable
significance is the fact that our route proceeds in 5.34%
overall yield from known 8, and its longest linear sequence
is only 13 steps. The prospect of a future total synthesis of
GE3 has now been brought much closer.
Acknowledgment. We thank AVERT for a studentship
to L.L. and Novartis (USA) for additional financial support.
We are grateful to the London School of Pharmacy for
HRMS determinations.
Supporting Information Available: 1H NMR spectra
(500 MHz), 13C NMR spectra (125 MHz), and HRMS data
for all new compounds. This material is available free of
charge via the Internet at http://pubs.acs.org.
OL025895U
(11) Cheung, H. T.; Blout, E. R. J. Org. Chem. 1965, 30, 315.
(12) Carpino, L. A. J. Am. Chem. Soc. 1993, 115, 4937.
(13) (a) Windholz, T. B.; Johnston, D. B. R. Tetrahedron Lett. 1967,
2555. (b) Pfeiffer, F. R.; Cohen, S. R.; Weisbach, J. A. J. Org. Chem. 1969,
34, 2795.
(14) For the occurrence of a similar O- to N-acyl shift during removal
of a Z-group from an acylated â-hydroxyleucine residue, see the Merck
L-156,602 synthesis referred to in ref 8.
(15) Past synthetic experience in these laboratories on handling free
cyclodepsipeptide hydrochlorides of the A83586C class has shown that
decomposition can occur after reverse-phase chromatography with H2O or
H2O/MeOH, when these solvents are subsequently removed on the rotary
evaporator. Given that these hydrochloride salts are usually obtained in a
fairly pure condition after hydrogenation, we recommend that they be used
directly for subsequent couplings without any further purification.
1906
Org. Lett., Vol. 4, No. 11, 2002