Given the potential of the cryptophycins as lead structures
for new antitumor agents, interest in their synthesis and that
of analogues has grown significantly over the past 10 years.
The first reports on the total synthesis appeared in the mid-
1990s from the research groups of Kitagawa4a and Moore
and Titus.4b These reports were soon followed by numerous
approaches comprising both formal and total syntheses.5
Many of these studies have focused on the synthesis of (+)-
cryptophycin-3 (3, Figure 1), the deoxy counterpart of (+)-
cryptophycin-1 (1), displaying diminished cytotoxicity by
100-fold.2a Efforts to discover and develop cryptophycin
analogues possessing improved pharmacokinetic properties
led to (+)-cryptophycin-52 (2, Figure 1),6 a semisynthetic,
designed to enhance hydrolytic stability relative to (+)-1 by
incorporation of a second methyl substituent at C(6) (Figure
1). Cryptophycin-52 (2), currently in phase II clinical trials,
is the most potent suppresser of microtubule dynamics
discovered to date.7 Moreover, in contrast to other antimitotic
agents, such as Taxol, vinblastine, and vincristine, (+)-
cryptophycin-52 (2) was shown to be minimally affected by
multidrug resistance.7a
Figure 2. (a) Overlay of X-ray structure of cryptophycin-3 (3)
and a simplified model of 4 and (b) overlay of X-ray structure of
cryptophycin-3 (3) and a simplified model of 5.
In view of the significant potential of the cryptophycins
for cancer therapy, we initiated a synthetic program with
the specific aim to develop a new class of analogues,
exploiting the concept of nonpeptide peptidomimetics.8 This
design strategy entails replacement of the 16-membered
macrolide ring of the cryptophycins with a suitable nonpep-
tide scaffold and attachment of the appropriate cryptophycin
side chains with the required spatial orientation to mimic
the conformation of the natural product. A premise of this
strategy is that modification of the macrolide ring, which
cause conformational change of side chains, significantly
diminishes cytotoxicity.2a
At the outset of this program, little was known about the
bioactive conformation of the cryptophycins bound to tubulin.
We therefore took as a working hypothesis that the available
X-ray structure of cryptophycin-3 (3, Figure 2) would
comprise a reasonable representation of the solution con-
formation for scaffold design. Importantly, NOE studies by
Moore et al. revealed that the preferred conformation of the
side chains in DMSO solution appears to be nearly identical
with that observed in the crystal structure.2a Extensive
computer modeling studies suggested that a seven-membered
ring could serve as a viable macrolide ring surrogate. The
efficacy of the seven-membered ring in a variety of drugs
(e.g., benzodiazapenes) is believed to be related to the modest
degree of flexibility that permits receptor-induced fit. Also
important from the design perspective, the overall size of
the seven-membered ring is smaller than that of the cryp-
tophycin macrocycle (Figure 2). Thus, potential cryptophycin
mimics having the putative biologically important substitu-
ents will not significantly occupy a volume larger than that
of the cryptophycin and thereby avoid deleterious steric
interactions during receptor binding.
Reasoning that an aryl substituent attached to the azepine
ring nitrogen would nicely overlay the substituted tyrosine
moiety in unit B (Figure 2a), and to alleviate the problem of
possible hydrophobic collapse induced by van der Waals
interactions between the two phenyl groups (as seen in our
initial modeling studies), we incorporated a urea functionality
in the side chain. An additional advantage of the urea moiety
is the potential to mimic the corresponding carbonyl and NH
groups in the BC peptide linkage of the cryptophycins. Upon
incorporation of the remaining cryptophycin side chains on
the azepine scaffold, the Monte Carlo conformational analysis
revealed that the model structures (Figure 2) reproduced well
the geometry of the side chains of cryptophycin-3 (3) at
comparatively small energy costs.9 As further illustrated in
Figure 2a and 2b, we selected two different linkages for the
side chain of unit A. A tertiary amine permitted excellent
overlap with cryptophycin-3 (3), when unit A possessed an
olefin (Figure 2a). Unfortunately, incorporation of a similar
side chain possessing the â-epoxide (4R,5R) led to decom-
position during the synthesis. On the other hand, use of an
ether linkage (Figure 2b) did enable construction of the
epoxide. However, to obtain best congruency in the side
chain conformation, one additional methylene moiety was
required. With these structural constraints in mind, we turned
to the synthesis of 4 and 5 from epoxide 6 as outlined
retrosynthetically in Scheme 1.
(3) (a) Mooberry, S. L.; Busquets, L.; Tien, G. Int. J. Cancer 1997, 73,
440 and references therein. (b) Bai, R.; Durso, N. A.; Sackett, D. L.; Hamel,
E. Biochemistry 1999, 38, 14302.
(4) (a) Kobayashi, M.; Kurosu, M.; Wang, W.; Kitagawa, I. Chem.
Pharm. Bull. 1994, 42, 2394. (b) Barrow, R. A.; Hemscheidt, T.; Liang, J.;
Paik, S.; Moore, R. E.; Tius, M. A. J. Am. Chem. Soc. 1995, 117, 2479.
(5) Eggen, M.; Nair, S. K.; Georg, G. I. Org. Lett. 2001, 3, 1813 and
references therein.
(6) (a) Norman, B. H.; Hemscheidt, T.; Schultz, R. M.; Andis, S. L. J.
Org. Chem. 1998, 63, 5288. (b) Liang, J.; Moher, E. D.; Moore, R. E.;
Hoard, D. W. J. Org. Chem. 2000, 65, 3143.
(7) (a) Wagner, M. M.; Paul, D. C.; Shih, C.; Jordan, M. A.; Wilson, L.;
Williams, D. C. Cancer Chemother. Pharmacol. 1999, 43, 115 and
references therein. (b) Panda, D.; Ananthnarayan, V.; Larson, G.; Shih, C.;
Jordan, M. A.; Wilson, L. Biochemistry 2000, 39, 14121 (c) Kessel, D.;
Luo, Y. Cancer Lett. 2000, 151, 25.
(8) Hirschmann, R.; Ducry, L.; Smith, A. B., III J. Org. Chem. 2000,
65, 8307 and references therein.
(9) Specific details of these studies are outlined in Supporting Informa-
tion.
4064
Org. Lett., Vol. 3, No. 25, 2001