Solution-phase parallel synthesis of a pharmacophore library of HUN-7293 analogues: A general chemical mutagenesis approach to defining structure-function properties of naturally occurring cyclic (depsi)peptides

Article abstract of DOI:10.1021/ja020166v

HUN-7293 (1), a naturally occurring cyclic heptadepsipeptide, is a potent inhibitor of cell adhesion molecule expression (VCAM-1, ICAM-1, E-selectin), the overexpression of which is characteristic of chronic inflammatory diseases. Representative of a gene

Full text of DOI:10.1021/ja020166v

Published on Web 04/19/2002
Solution-Phase Parallel Synthesis of a Pharmacophore
Library of HUN-7293 Analogues: A General Chemical
Mutagenesis Approach To Defining Structure-Function
Properties of Naturally Occurring Cyclic (Depsi)peptides
Yan Chen,† Melitta Bilban,† Carolyn A. Foster,‡ and Dale L. Boger*,†
Contribution from the Department of Chemistry and The Skaggs Institute for Chemical Biology,
The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, and
NoVartis Forschungsinstitut, Brunner Strasse 59, A-1235, Vienna, Austria
Received February 1, 2002
Abstract: HUN-7293 (1), a naturally occurring cyclic heptadepsipeptide, is a potent inhibitor of cell adhesion
molecule expression (VCAM-1, ICAM-1, E-selectin), the overexpression of which is characteristic of chronic
inflammatory diseases. Representative of a general approach to defining structure-function relationships
of such cyclic (depsi)peptides, the parallel synthesis and evaluation of a complete library of key HUN-7293
analogues are detailed enlisting solution-phase techniques and simple acid-base liquid-liquid extractions
for isolation and purification of intermediates and final products. Significant to the design of the studies
and unique to solution-phase techniques, the library was assembled superimposing a divergent synthetic
strategy onto a convergent total synthesis. An alanine scan and N-methyl deletion of each residue of the
cyclic heptadepsipeptide identified key sites responsible for or contributing to the biological properties.
The simultaneous preparation of a complete set of individual residue analogues further simplifying the
structure allowed an assessment of each structural feature of 1, providing a detailed account of the structure-
function relationships in a single study. Within this pharmacophore library prepared by systematic chemical
mutagenesis of the natural product structure, simplified analogues possessing comparable potency and,
in some instances, improved selectivity were identified. One potent member of this library proved to be an
additional natural product in its own right, which we have come to refer to as HUN-7293B (8), being isolated
from the microbial strain F/94-499709.
The naturally occurring cyclic heptadepsipeptide HUN-7293
(1, Figure 1) was first isolated in 1992 from a fungal broth in
a screen for inhibitors of inducible cell adhesion molecule
autoimmune diseases characterized by cell adhesion molecule
overexpression or upregulation. As detailed in recent studies,4
diverse signals act on endothelial cells to activate members of
the nuclear factor of Kappa B (NF-κB) transcription factor
family. NF-κB family members are cytoplasmic homo- or
heterodimer proteins inactivated when complexed with members
of a family of inhibitor proteins, IκB. Phosphorylation of IκB,
releasing the active NF-κB, results in translocation to the nucleus
and initiation of transcription of its targeted genes including
the cell surface adhesion molecules. Aside from antisense5 and
antibody6 approaches, only a limited number of other groups
have detailed work on small molecule inhibitors7 of the
expression of cell surface adhesion molecules and, unlike HUN-
7293 (1),1 each target the NF-κB signal transduction pathway.
1
expression.1 Its structure was subsequently established by H
NMR spectroscopic methods and confirmed by X-ray analysis.2
Independently, the same cyclic depsipeptide was isolated by a
Japanese group from a different fungal species based on a screen
for anti-HIV compounds.3 Endothelial cell associated molecules
including intercellular adhesion molecule 1 (ICAM-1), vascular
cell adhesion molecule 1 (VCAM-1), and E-selectin play
important roles in the immune response by recruiting and
regulating leukocyte migration and cell-to-cell interactions at a
site of inflammation. Modulation of these interactions through
the inhibition of such cell adhesion molecule expression should
prove useful for the treatment of chronic inflammatory and
We recently disclosed the first total synthesis of HUN-7293
(1) by a convergent approach which served to define several
* Corresponding author. E-mail: boger@scripps.edu.
† The Scripps Research Institute.
‡ Novartis Forschungsinstitut.
(4) Collins, T.; Read, M. A.; Neish, A. S.; Whitley, M. Z.; Thanos, D.; Maniatis,
T. FASEB J. 1995, 9, 899-909. Thanos, D.; Maniatis, T. Cell 1995, 80,
529-532. May, M. J.; Ghosh, S. Immunol. Today 1998, 19, 80-88.
(5) Glover, J. M.; Leeds, J. M.; Mant, T. G. K.; Amin, D.; Kisner, D. L.;
Zuckerman, J. E.; Geary, R. S.; Levin, A. A.; Shanahan, W. R., Jr. J.
Pharmacol. Exp. Ther. 1997, 282, 1173-1180. Yacyshyn, B. R.; Bowen-
Yacyshyn, M. B.; Jewell, L.; Tami, J. A.; Bennett, C. F.; Kisner, D. L.;
Shanahan, W. R., Jr. Gastroenterology 1998, 114, 1133-1142.
(6) Kavanaugh, A. F.; Schulze-Koops, H.; Davis, L. S.; Lipsky, P. E. Arthritis
Rheum. 1997, 40, 849-853.
(1) Foster, C. A.; Dreyfuss, M.; Mandak, B.; Meingassner, J. G.; Naegeli, H.
U.; Nussbaumer, A.; Oberer, L.; Scheel, G.; Swoboda, E.-M. J. Dermatol.
1994, 21, 847-854. Foster, C. A. J. Aller. Clin. Immunol. 1996, 98, S270.
Wang, S.; Winiski, A. P.; Scheel, G.; Geyl, D.; Oberhauser, B.; Traber,
R.; Foster, C. A. Mol. Biol. Cell 2000, 11, 467A.
(2) Hommel, U.; Weber, H.-P.; Oberer, L.; Naegeli, H. U.; Oberhauser, B.;
Foster, C. A. FEBS Lett. 1996, 379, 69-73.
(3) Itazaki, H.; Fujiwara, T.; Sato, A.; Kawamura, Y.; Matsumoto, K. Japanese
Kokai Tokkyo Koho 1995, 11 pp.
9
10.1021/ja020166v CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 5431-5440
5431

A R T I C L E S
Chen et al.
use even in the multistep synthesis of the natural product 1 and
its structurally challenging analogues. The library described (4-
34) consists of nearly 40 natural product analogues,15-21 each
prepared using a convergent,12-14 yet divergent22 total synthesis,
and provides an alanine scan of the seven residues, an N-methyl
deletion of the three N-methyl residues, and fundamental
simplifications in each of the nonstandard amino acid side chains
allowing the identification of key residues and structural features
contributing to the biological properties. The simultaneous
inclusion of key analogues of the nonstandard amino acid side
chains permitted the identification of not only essential structural
features, but also nonobvious and subtle structural features. It
is this combination of the alanine scan and N-methyl deletion
library with the systematic single site deletion modifications of
structural features found in each of the individual side chains
that we have come to refer to as a pharmacophore library.
Significant to the design of such studies, the library was
assembled superimposing a divergent synthetic strategy22 onto
a convergent total synthesis. This combination of the beneficial
attributes of a conventional convergent synthetic strategy with
the divergent preparation of a series of structurally related
analogues (e.g., use of one prep of 2 to assemble all residue
1-3 analogues) can only be conveniently conducted using
solution-phase, not solid-phase, techniques and avoids the more
repetitive independent linear synthesis of each analogue.
Solution-Phase Synthesis of HUN-7293 and Aza-HUN-
7293. In preliminary studies designed to optimize the solution-
phase synthesis protocols for implementation in the library
Figure 1.
subtle elements essential to its preparation including backbone
ester formation by a Mitsunobu displacement and the identifica-
tion of MLEU3-LEU4 as an effective macrocyclization site.8 In
addition to identifying the tetrapeptide 2 and tripeptide 3 as key
intermediates in the convergent synthesis, the use of a Mitsunobu
esterification with inversion of the DGCN R-center avoided
racemization of the sensitive NMe-Ala (residue 7) observed
upon carboxylate activation required of a conventional esteri-
fication, and permitted the utilization of a readily available L-
versus D-amino acid precursor to the D-R-hydroxy carboxylic
acid residue. Herein, we detail the implementation of this
approach in the parallel synthesis of a library of key analogues
of 1 used to define in detail the structure-function relationships
of the natural product.
A solution-phase approach to the simultaneous (parallel)
preparation of the library was utilized enlisting simple acid-
base liquid-liquid extractions for isolation and purification of
the synthetic intermediates and final products. This approach,
which we introduced with the multistep preparation of simpler
amide-based small molecules libraries,9-14 provided each in-
termediate and final product sufficiently pure for subsequent
(10) Cheng, S.; Tarby, C. M.; Comer, D. D.; Williams, J. P.; Caporale, L. H.;
Boger, D. L. Bioorg. Med. Chem. 1996, 4, 727-737.
(11) Boger, D. L.; Tarby, C. M.; Caporale, L. H. J. Am. Chem. Soc. 1996, 118,
2109-2110.
(12) Boger, D. L.; Chai, W.; Jin, Q. J. Am. Chem. Soc. 1998, 120, 7220-7225.
(13) Boger, D. L.; Chai, W. Tetrahedron 1998, 54, 3955-3970.
(14) Boger, D. L.; Goldberg, J.; Jiang, W.; Chai, W.; Ducray, P.; Lee, J. K.;
Ozer, R. S.; Andersson, C.-M. Bioorg. Med. Chem. 1998, 6, 1347-1378.
(15) Review: Hall, D. G.; Manku, S.; Wang, F. J. Comb. Chem. 2001, 3, 125-
150.
(16) For representative natural product libraries see: Nicolaou, K. C.; Winssinger,
N.; Pastor, J.; Ninkovic, S.; Sarabia, F.; He, Y.; Vourloumis, D.; Yang, Z.;
Li, T.; Giannakakou, P.; Hamel, E. Nature 1997, 387, 268-272. Nicolaou,
K. C.; Vourloumis, D.; Li, T.; Pastor, J.; Winssinger, N.; He, Y.; Ninkovis,
S.; Sarabia, F.; Vallberg, H.; Roschanger, F.; King, N. P.; Finlay, R. V.;
Giannakakou, P.; Verdier-Pinard, P.; Hamel, E. Angew. Chem., Int. Ed.
Engl. 1997, 36, 2097-2103. Nicolaou, K. C.; Winssinger, N.; Vourloumis,
D.; Ohshima, T.; Kim, S.; Pfefferkorn, J.; Xu, J.-Y.; Li, T. J. Am. Chem.
Soc. 1998, 120, 10814-10826. Dragoli, D. R.; Thompson, L. A.; O’Brien,
J.; Ellman, J. A. J. Comb. Chem. 1999, 1, 534-539. Lee, K. J.; Angulo,
A.; Ghazal, P.; Janda, K. D. Org. Lett. 1999, 1, 1859-1862. Lopez-Pelegrin,
J. A.; Janda, K. D. Chem. Eur. J. 2000, 6, 1917-1922. Xu, R.;
Greiveldinger, G.; Marenus, L. E.; Cooper, A.; Ellman, J. A. J. Am. Chem.
Soc. 1999, 121, 4898-4899. Bhat, L.; Liu, Y.; Victory, S. F.; Himes, R.
H.; Georg, G. I. Bioorg. Med. Chem. Lett. 1998, 8, 3181-3184. Fecik, R.
A.; Frank, K. E.; Gentry, E. J.; Mitscher, L. A.; Shibata, M. Pure Appl.
Chem. 1999, 71, 559-564. Nicolaou, K. C.; Pfefferkorn, J. A.; Barluenga,
S.; Mitchell, H. J.; Roecker, A. J.; Cao, G. Q. J. Am. Chem. Soc. 2000,
122, 9968-9976 and accompanying articles. Wipf, P.; Reeves, J. T.;
Balachandran, R.; Giuliano, K. A.; Hamel, E.; Day, B. W. J. Am. Chem.
Soc. 2000, 122, 9391-9395. de Frutos, O.; Curran, D. P. J. Comb. Chem.
2000, 2, 639-649. Andrus, M. B.; Turner, T. M.; Sauna, Z. E.; Ambudkar,
S. V. J. Org. Chem. 2000, 65, 4973-4983. Pelish, H. E.; Westwood, N.
J.; Feng, Y.; Kirchhausen, T.; Shair, M. D. J. Am. Chem. Soc. 2001, 123,
6740-6741. Chang, Y.-T.; Choi, J.; Ding, S.; Prieschl, E. E.; Baumruker,
T.; Lee, J.-M.; Chung, S.-K.; Shultz, P. G. J. Am. Chem. Soc. 2002, 124,
1856-1857. For natural product libraries prepared using the solution-phase
techniques detailed herein, see refs 17-20.
(7) Boschelli, D. H.; Connor, D. T.; Lesch, M. E.; Schrier, D. J. Bioorg. Med.
Chem. 1996, 4, 557-562. Boschelli, D. H.; Kramer, J. B.; Khatana, S. S.;
Sorenson, R. J.; Connor, D. T.; Ferin, M. A.; Wright, C. D.; Lesch, M. E.;
Imre, K.; Okonkwo, G. C.; Schrier, D. J.; Conroy, M. C.; Ferguson, E.;
Woelle, J.; Saxena, U. J. Med. Chem. 1995, 38, 4597-4614. Sullivan, R.
W.; Bigam, C. G.; Erdman, P. E.; Palanki, M. S. S.; Anderson, D. W.;
Goldman, M. E.; Ransone, L. J.; Suto, M. J. J. Med. Chem. 1998, 41, 413-
419. Stewart, A. O.; Bhatia, P. A.; McCarty, C. M.; Patel, M. V.; Staeger,
M. A.; Arendsen, D. L.; Gunawardana, I. W.; Melcher, L. M.; Zhu, G.-D.;
Boyd, S. A.; Fry, D. G.; Cool, B. L.; Kifle, L.; Lartey, K.; Marsh, K. C.;
Kempf-Grote, A. J.; Kilgannon, P.; Wisdom, W.; Meyer, J.; Gallatin, W.
M.; Okasinski, G. F. J. Med. Chem. 2001, 44, 988-1002. Meng, C. Q.;
Zheng, X. S.; Holt, L. A.; Hoong, L. K.; Somers, P. K.; Hill, R. R.; Saxena,
U. Bioorg. Med. Chem. Lett. 2001, 11, 1823.
(17) Boger, D. L.; Fink, B. E.; Hedrick, M. P. J. Am. Chem. Soc. 2000, 122,
6382-6394 (distamycin A).
(18) Boger, D. L.; Dechantsreiter, M. A.; Ishii, T.; Fink, B. E.; Hedrick, M. P.
Bioorg. Med. Chem. 2000, 8, 2049-2057 (distamycin A).
(19) Boger, D. L.; Lee, J. K. J. Org. Chem. 2000, 65, 5996-6000 (triostin A).
(20) Boger, D. L.; Schmitt, H.; Fink, B. E.; Hedrick, M. P. J. Org. Chem. 2001,
66, 6654-6661 (CC-1065).
(8) Boger, D. L.; Keim, H.; Oberhauser, B.; Schreiner, E. P.; Foster, C. A. J.
Am. Chem. Soc. 1999, 121, 6197-6205.
(9) Cheng, S.; Comer, D. D.; Williams, J. P.; Boger, D. L. J. Am. Chem. Soc.
1996, 118, 2567-2573.
(21) Combinatorial natural products biosynthesis: Hutchinson, C. R. Curr. Opin.
Microbiol. 1998, 1, 319-329.
(22) Divergent total synthesis: Boger, D. L.; Brotherton, C. E. J. Org. Chem.
1984, 49, 4050-4055.
9
5432 J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002

Synthesis of a Pharmacophore Library of HUN-7293 Analogues
A R T I C L E S
Scheme 1
tion isolations with minimal formation of undesired diastereo-
mers, the peptide coupling conditions were optimized with the
amines used in excess to ensure the complete consumption of
the intermediate activated esters derived from the carboxylic
acids. The three dipeptides 37, 42, and 45 were obtained cleanly
in good yields after simple acid-base liquid-liquid extractions
of couplings promoted by EDCI-HOAt (Scheme 1). The excess
starting amine, any residual unreacted carboxylic acid, reagents
(EDCI-HOAt), and the reagent byproducts (from EDCI) were
removed in the extractions. Selective deprotection of the N-Boc
group in the presence of the t-Bu ester in 37 with HCl-EtOAc25
provided the amine HCl salt 38, which was used directly in the
next coupling reaction. Tripeptide 3 was obtained in good yield
and purity from 38 and the R-hydroxycarboxylic acid 39 using
EDCI-HOAt (2,6-lutidine) and the acid-base liquid-liquid
extraction for purification and isolation. Similarly, tetrapeptide
2 was prepared employing EDCI-HOAt (2,6-lutidine) for the
coupling of dipeptides 43 and 46 following N-Boc deprotection
of 45 (HCl-EtOAc) and methyl ester hydrolysis of 42 (LiOH).
Under these conditions, the diastereomeric ratio favoring the
desired tetrapeptide 47 was increased to 32:1 compared to 13:1
using HATU-NaHCO3.8 Notably, this coupling entails activation
of a racemization prone N-acyl N-methyl amino acid and was
carefully monitored in the synthesis of analogues containing
modifications in residues 4-6. Mitsunobu esterification of 2,
derived from LiOH hydrolysis of 47, with the tripeptide 3
provided the key linear heptadepsipeptide 48 (47%). Whereas
the product of the amide coupling reaction utilized to assemble
the linear heptapeptide precursor to aza-HUN-7293 (4) was
isolated and purified through the acid-base liquid-liquid
extraction protocol,25 the Mitsunobu esterification product 48
and those leading to 5-32 were further purified by chroma-
tography. In principle this may not have been necessary, but it
did simplify the isolation and improved the purity of the final
products. Linear heptadepsipeptide 48 was deprotected upon
treatment with TFA, and converted to the amine HCl salt. Final
macrolactamization to provide 1 (64%) was achieved by the
treatment with DPPA-iPr2NEt in 1 mM DMF. These cyclization
conditions were first developed for the preparation of aza-HUN-
7293 (4),23 which was established to close much less effectively
than 1 itself, and subsequently established to be optimal for
synthesis of the HUN-7293 analogues. In contrast, treatment
of the deprotected linear heptadepsipeptide derived from 48 with
3
HOAt-EDCI in the absence of base afforded C2 -epi-1 exclu-
sively (32-44%) without the detection of 1. Analogous
observations were made with the cyclization of the linear
heptapeptide leading to 4 (aza-HUN-7293) although this sub-
strate failed to close as effectively as 1. Cyclization conducted
with DPPA (0 °C, 5 mM DMF) in the presence of iPr2NEt (16:
1, 27%) or NaHCO3 (21:1, 10%) provided 4 with little
synthesis, the preparations of HUN-7293 (1) and its backbone
aza analogue 4 were conducted enlisting acid-base liquid-
liquid extractions for the purification of the intermediates. The
results obtained with the synthesis of 1 are detailed in Scheme
1 and details of the synthesis of 4 are provided in the Supporting
Information.23,24 The initial preparation of 4 along with 1
permitted an early assessment of the importance of the backbone
depsipeptide ester resulting in its maintenance in the library.
The tetrapeptide 2 and tripeptide 3 were prepared starting
with the five amino acids and R-hydroxycarboxylic acid
previously disclosed.8 To obtain pure products from the extrac-
3
competitive C2 epimerization, whereas closure promoted by
BOP (25 °C, 5 mM CH3CN) in the presence of DMAP (5:1,
22%), iPr2NEt (3:1, 25%), or NaHCO3 (1:2.5, 5%) suffered
3
substantial C2 epimerization, and those promoted by EDCI-
HOAt (0-25 °C, 5 mM DMF) with or without NaHCO3
suffered near complete C23 epimerization (1:18-32, 40-85%).
3
Aza-HUN-7293 (4) and its C2 epimer were compared with
3
1 and its C2 epimer in the ability to inhibit inducible cell
(23) Boger, D. L.; Chen, Y.; Foster, C. A. Bioorg. Med. Chem. Lett. 2000, 10,
1741-1744.
(24) Full experimental details are provided in the Supporting Information.
(25) Gibson, F. S.; Bergmeier, S. C.; Rapoport, H. J. Org. Chem. 1994, 59,
3216-3218.
9
J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002 5433

A R T I C L E S
Chen et al.
Figure 2.
adhesion molecule expression (VCAM-1, ICAM-1, E-selectin)
in ELISA assays previously disclosed (Figure 2).1,2 In each case,
4 proved to be 20- to 60-fold less potent than 1 at inhibiting
VCAM-1, ICAM-1, or E-selectin expression.23 The natural
product exhibits a 24-fold selectivity for VCAM-1 versus
ICAM-1 expression and an even greater 44-fold selectivity
versus E-selectin. This same selectivity, albeit reduced, was
observed with 423 and the subsequent screening of our library
was conducted comparing only the inhibition of inducible
VCAM-1 and ICAM-1 expression. In addition, the C23 epimer
of aza-HUN-7293 exhibited a further 10-fold loss in potency
relative to 4 suggesting that the C23 stereochemistry is important
to the properties of 1. In contrast to this behavior of 4 versus
Figure 3.
each amino acid side chain with a methyl group while retaining
the stereochemistry of the peptide backbone, the significance
of each individual side chain can be identified.28 Although this
approach has its limitations, it has been used to provide insight
into structure-function relationships of a protein, especially
when the functional mechanism is unknown or its 3D structure
is unavailable.29Similarly, in the absence of a defined target,
the importance of each residue of the cyclic heptadepsipeptide
structure of HUN-7293 could not be anticipated. Along with
additional modifications and simplifications in each of the
individual side chains, an alanine scan of the structure was
conducted. The synthesis of each analogue constituting the
alanine scan is presented in the following treatments of each
residue. However, the results of their examination are sum-
marized in Figure 3 since they provide an initial, albeit
surprisingly incomplete, overview of the relative importance of
each residue. With the exception of residue 7, which already
contains a methyl group in the natural product (R7 ) Me), the
replacement of each side chain with a methyl group led to a
reduction in potency. However, the magnitude of this loss in
activity varied over a 1000-fold range with the R1 replacement
being only ca. 2-fold lower, the R2-R4 replacements leading
to 10- to 100-fold reductions, and the R5 and R6 replacements
leading to even larger >100- to 1000-fold reductions with R5
being by far the most significant residue. Just as interesting and
despite the potency reductions, the methyl substitutions for the
R1-R4 side chains enhanced the VCAM-1 versus ICAM-1
selectivity, indicating that the changes in this half of the
3
3
C2 -epi-4, C2 -epi-1 was only 2-fold less potent than 1 against
VCAM-1 and only 5-fold less potent against ICAM-1.26
Although this result was not expected, it is consistent with
subsequent observations made with a series of such C23 epimers
that we incorporated into the following studies. Consequently,
the C23 stereochemistry was critically assessed following library
macrocyclization in light of the potential epimerization. Having
established that the depsipeptide ester is important to the
biological properties of 1 and facilitated macrocyclization,23 the
library was assembled incorporating the backbone ester. In
principle, this did not alter the reliance on extraction purification.
However, we did find that chromatographic purification fol-
lowing the Mitsunobu esterification facilitated the isolation of
the pure final products by removing minor contaminant byprod-
ucts and diastereomers accumulating throughout the syntheses.
In addition, and in order to ensure confidence in the testing
results, the final analogues 4-34 were purified by HPLC
3
chromatography prior to assay ensuring removal of any C2 -
epimer and other potential diastereomer contaminants.
Alanine Scan. Alanine-scanning mutagenesis is an informa-
tive method for defining epitopes in proteins.27 By replacing
3
(26) The C2 epimer of 1 has also been prepared by degradation of 1 through
selective formation of the residue 3 thioamide, thioimidate formation,
(28) Humphrey, J. M.; Chamberlin, A. R. Chem. ReV. 1997, 97, 2243-2266.
(29) For examples of such an N-methyl deletion scan, see: Boger, D. L.;
Yohannes, D.; Myers, J. B. J. Org. Chem. 1992, 57, 1319-1321. Boger,
D. L.; Yohannes, D.; Zhou, J.; Patane, M. A. J. Am. Chem. Soc. 1993,
115, 3420-3430. Boger, D. L.; Patane, M. A.; Zhou, J. J. Am. Chem. Soc.
1994, 116, 8544-8556. Boger, D. L.; Patane, M. A.; Zhou, J. J. Am. Chem.
Soc. 1995, 117, 7357-7363. Boger, D. L.; Zhou, J. J. Am. Chem. Soc.
1995, 117, 7364-7378.
3
hydrolysis, and subsequent macrolactamization with partial C2 epimer-
ization. Its activity was found to be identical with that of our synthetic
sample of C23-epi-1, IC50 ) 1.5 nM (VCAM-1), 84 nM (ICAM-1). We
thank Berndt Oberhauser (Novartis) for sharing these observations in
advance of publication.
(27) Di Cera, E. AdV. Protein Chem. 1998, 51, 59-119 and the references
therein.
9
5434 J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002

Synthesis of a Pharmacophore Library of HUN-7293 Analogues
A R T I C L E S
molecule diminished inhibition of ICAM-1 expression more
significantly, whereas this selectivity was lost along with the
activity with methyl substitutions at R5 and R6 with the change
at R5 again being especially detrimental. From these results
alone, one could anticipate that substantial changes in the R1
side chain could be tolerated, including efforts to enhance
selectivity or physiochemical, distribution, stability, and toxicity
profiles, whereas changes to R2-R4 might prove more chal-
lenging, and those to R5 and R6 are likely to be especially
detrimental. Consequently, the results of the more subtle side
chain modifications which are detailed in the following sections
proved especially valuable at addressing inaccuracies in inter-
preting such an incomplete data set.
N-Methyl Deletion. Peptides containing N-methyl groups
often display special properties, including increased hydropho-
bicity, increased stability and bioavailability, along with promot-
ing the adoption of cis amide bonds or disrupting intramolecular
H-bonds which are key to establishing the conformation of the
molecules.28,29 HUN-7293 adopts a single, well-defined con-
formation in solution (>90% in CDCl3/DMSO and MeOH/H2O,
g60-70% in CDCl3) which is very similar to that found in the
X-ray structure.2 Prominent structural features of this major
conformation are two cis amide bonds between residues 2 and
3 and residues 4 and 5 where two of the N-methyl amides are
found. Two strong transannular H-bonds are also observed
(N(2)-H/OdC(6), N(6)-H/OdC(3)). This N-methylation state
and intramolecular H-bonding along with the depsipeptide ester
leaves only one amide NH available to serve as a H-bond donor
for intermolecular H-bonding (N(4)-H). One key set of
analogues is 5-7 which constitute the N-methyl deletion at each
of the three N-methyl amides of 1. Upon reflection on the
conduct of the work, we would now recommend extending such
studies to include an N-methyl scan of each secondary amide
residue of the native cyclic depsipeptide which could reveal
the additional importance of the structural or H-bonding
characteristics of the secondary amides.30,31However, at the time
our work was conducted, the potential benefit of such a
N-methyl scan escaped us.
The synthesis of these three key analogues requires the
incorporation of the parent NH versus N-methyl derivative of
residues 3, 5, and 7. Those of residues 3 and 7 are simply L-Leu
and L-Ala and that of residue 5 constituted an intermediate in
our synthesis of the MTO5 residue of HUN-7293.8 The details
of the synthesis of 5-7 are provided in the Supporting
Information.
The evaluation of 5-7 was revealing (Figure 4). The
N-methyl amide linking residues 4 and 5 was found to be
essential as its removal with 6 resulted in the complete loss of
activity (>10 000-fold). This is consistent with the amide
adopting a cis amide stereochemistry serving as a critical
structural element establishing the conformation of the molecule.
Consistent with this expectation, the strong shielding of one of
the Leu4 â1 protons (δ -0.37), indicative of the stacking
interaction with the MTO5 indole seen in the NMR and X-ray
structures, is no longer observed in the 1H NMR of 6. Although
it adopts a well-defined conformation as judged by its sharp
Figure 4.
1
single resonance H NMR, this conformation differs from that
of 1. This largest loss of activity with a modification to residue
5, like that observed in the alanine scan, indicates a critical role
for this residue.
The removal of the additional two N-methyl amides also
reduced activity, albeit not nearly as substantially, but the
changes also productively enhanced the VCAM-1 selectivity.
The most interesting of these was the modest effect of removing
the residue 7 N-methyl group with 7. This only reduced the
VCAM-1 potency 7-fold (IC50 ) 6.8 nM versus 1 nM),
providing an analogue with a sub 10 nM IC50, with a much
improved VCAM-1 versus ICAM-1 selectivity (180-fold versus
24-fold for 1). Thus, removal of the residue 7 N-methyl group
impacted the inhibition of ICAM-1 expression (50-fold reduc-
tion), but had a much more modest effect on inhibition of
VCAM-1 expression (7-fold). Consistent with this behavior, 7
adopts at least two major and nearly equivalently populated
conformations as judged by H NMR. The H NMR of one of
these conformations exhibits the strongly shielded Leu4 â1
proton (δ -0.50) identical with that found in 1 whereas that of
the second is less significantly, but still substantially, shielded
(δ 0.00).
Removal of the residue 3 N-methyl group providing 5 more
significantly reduced activity (11-fold) and had a significant but
less pronounced impact on the selectivity improving it beyond
that observed with 1 (54- vs 24-fold for 1). Interestingly, 5
adopts a single, well-defined solution conformation as judged
by 1H NMR that does not appear on the surface to differ
substantially from that of 1. For example, the diagnostic Leu4
â1 proton exhibits the same strongly shielded chemical shift (δ
1
1
1
-0.28 for 5 vs -0.37 for 1). However, the H NMR spectra
differs considerably in the δ 4.0-5.5 region (R-protons),
indicating a more subtle conformational change, likely the result
of adoption of a trans amide between residues 2 and 3. Thus,
while the role of the residue 3 N-methyl amide is not yet defined,
it is clearly important.
Residue 1. Residue 1, because of its side chain functionality,
is the one residue that has been extensively explored through
derivatization of the natural product itself.32 Consequently, this
residue was not extensively examined in our work and a number
of obvious modifications that would have ordinarily been
(30) Rajeswaran, W. G.; Hocart, S. J.; Murphy, W. A.; Taylor, J. E.; Coy, D.
H. J. Med. Chem. 2001, 44, 1305-1311 and 1416-1421.
(31) For examples of such an N-methyl scan, see: Boger, D. L.; Teramoto, S.;
Cai, H. Bioorg. Med. Chem. 1996, 4, 179-194. Boger, D. L.; Teramoto,
S.; Cai, H. Bioorg. Med. Chem. 1997, 5, 1573-1590.
9
J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002 5435

A R T I C L E S
Chen et al.
Figure 6.
Figure 5.
concurrent examinations of 11 and 12 proved especially
revealing (Figure 6).
incorporated were not pursued. The two analogues prepared, 8
and 9, constitute the removal of the -CH2CN of the side chain
(R1 ) CH3) and the side chain extension by one carbon (R1 )
CH2CH2CH2CN). These analogues would not be easily prepared
from 1 itself yet are revealing in the importance of the presence
of the side chain or its functionality. Their synthesis required
the incorporation of commercially available (S)-2-hydroxypro-
pionic acid and (S)-5-cyano-2-hydroxypentanoic acid (prepared
from L-homoglutamine), into the modified tripeptide 3, Mit-
sunobu esterification with tetrapeptide 2, deprotection, and
macrocyclization.
The preparation of 10-12 required L-Ala, (S)-2-aminooc-
tanoic acid,33 and L-Leu, respectively, for incorporation into
analogues of 37, and the subsequent modified tripeptide 3.
Mitsunobu esterification with tetrapeptide 2, subsequent depro-
3
tection, and macrocyclization provided 10-12. C2 -epi-12 was
3
also synthesized and used to assess the importance of the C2
center relative to the biological activity of 12. The modified
linear heptadepsipeptide 48 was deprotected and treated with
3
EDCI-HOAt in the absence of added base to give C2 -epi-12
3
in 44% isolated yield (C2 -epi-12:12 ) 5.5:1). Under comparable
The examination of 8 and 9 was especially revealing (Figure
5). Simple alanine substitution with 8 for the more elaborate
side chain had a modest ca. 2-fold effect lowering the inhibition
of VCAM-1 providing a simplified analogue that still exhibited
potent (IC50 ) 2.3 nM) inhibition of VCAM-1 expression. A
greater 7-fold loss in potency against ICAM-1 was observed,
and the VCAM-1/ICAM-1 selectivity increased accordingly. In
contrast, 9 was 5- and 8-fold less potent than 1 in reducing
VCAM-1 and ICAM-1 expression, respectively, indicating that
simple extension of the side chain by one carbon was modestly
detrimental. Of these two analogues, 8 exhibits a potency that
approaches that of 1, and exhibits a better VCAM-1 selectivity.
Consequently, it was remarkable that ad this work was being
conducted, the potent residue 1 alanine analogue 8 was
discovered to be a natural product in its own right, being isolated
from the microbial strain F/94-499709,32 and we have come to
refer to it as HUN-7293B. Our sample of synthetic 8 was
identical in all compared respects with naturally derived 8 (1H
NMR).32
3
conditions, 48 itself closed to provide exclusively C2 -epi-1
without the detection of 1. Because of its availability, an
examination of the macrocyclization reaction was also conducted
with the linear heptadepsipeptide leading to 12 and the results
are representative of other analogues studied. Three cyclization
conditions were tested: EDCI-HOAt (NaHCO3), which gave 1
in 72%,8 BOP-DMAP, which gave 1 in 80% yield,8 and DPPA-
3
iPr2NEt, which also gave 1 without detection of C2 -epi-1 (vide
supra). The DPPA-iPr2NEt conditions proved optimal for this
3
specific substrate and gave 12 in 59% without detection of C2 -
epi-12, while EDCI-HOAt (NaHCO3) afforded 12 in 47% with
3
12:C2 -epi-12 ) 1:2, and BOP-DMAP provided 12 in 56% with
3
12:C2 -epi-12 ) 6:1.
In contrast to the alanine analogue (100-fold loss), the removal
of the side chain methyl group and the chiral center with 11
resulted in only a 6-fold loss of potency and an analogue that
maintained or possessed a slightly enhanced VCAM-1 selectivity
(Figure 6). Similarly, the removal of the propyl extension of
the side chain with the incorporation of L-Leu with 12 (removal
of three methylenes) resulted in an analogous 6-fold reduction
in potency and a similarly maintained or slightly enhanced
VCAM-1 selectivity. Thus, an aliphatic hydrophobic side chain
is required, but the characteristics of this side chain do not appear
to be as specific as its unique structure might imply. In contrast
to conclusions that might be drawn from the alanine scan,
considerable simplification of this residue may be possible
including others not yet explored in our studies. Importantly,
Residue 2. The results of the alanine scan for residue 2, which
indicated a 100-fold reduction in potency with 10, along with
the unique structure of the side chain incorporating a chiral
center suggested this residue might be highly specific and
resistant to productive changes. Consequently, the results of the
(32) Dreyfuss, M.; Fehr, T.; Foster, C. A.; Geyl, D.; Oberhauser, B. PCT
Internation Patent Application WO 9719104, 1997; Chem. Abstr. 1997,
127, 81792. Dreyfuss, M. M.; Foster, C. A.; Naegeli, H.-U.; Oberhauser,
B. PCT Internation Patent Application WO 9603430, 1996; Chem. Abstr.
1996, 125, 34177. We thank Berndt Oberhauser (Novartis) for supplying
the 1H NMR of authentic HUN-7293B (IC50 ) 2.1/220 nM for VCAM-
1/ICAM-1) for comparison.
(33) Chenault, H. K.; Dahmer, J.; Whitesides, G. M. J. Am. Chem. Soc. 1989,
111, 6354-6364.
9
5436 J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002

Synthesis of a Pharmacophore Library of HUN-7293 Analogues
A R T I C L E S
the simplification achieved with 11 and 12 was accomplished
with the maintenance of single digit nM activity versus VCAM-1
expression with a slightly enhanced VCAM-1 selectivity.
3
3
Analogous to the comparison of 1 and C2 -epi-1, C2 -epi-12
was only 3-fold less potent at inhibiting VCAM-1 expression
than 12 itself but 8-fold less effective at inhibiting ICAM-1
3
3
expression. Thus, C2 -epi-12, like C2 -epi-1, is slightly less
potent, but more selective for inhibition of VCAM-1 expression.
Notably, the adopted conformation of 10 (R2 ) CH3), as
judged by near identical 1H NMR’s, is essentially indistinguish-
able from that of 1, 11, or 12, indicating that the 100-fold
reduction in activity with 10 is not derived from a structural
perturbation, but rather that the side chain modifications in the
series impact the extent to which 1 reaches or interacts with its
biological target.
Residues 3 and 4. Because residues 3 and 4 are NMe-Leu
and L-Leu, respectively, the changes examined were limited to
alanine substitutions, 13 and 14 (Figure 3), the N-methyl deletion
3
3
(5, Figure 4), and the examination of the C2 -epimer of 1 (C2 -
epi-1, Figure 2). With the exception of the latter modification,
each of these changes resulted in 10- to 60-fold losses in potency
with that of the NMe-Ala substitution for NMe-Leu3 being the
3
most significant (60-fold). By contrast, the C2 -epimer of 1 was
nearly as active as 1 against VCAM-1 expression.26 Thus, while
the presence of the hydrophobic side chain is important, its
configuration is not. This suggests its impact does not lie with
specific target interactions or conformational effects on the
cyclic depsipeptide, but rather with properties that influence how
well the compound reaches its biological target. In addition,
each change enhanced the VCAM-1 selectivity, indicating that
the modification had a more detrimental effect on ICAM-1
versus VCAM-1 expression.
The preparation of 13 and 14 simply required incorporation
of NMe-Ala (residue 3, tripeptide 3) and Ala (residue 4,
tetrapeptide 2), respectively, into the cyclic heptadepsipeptide
synthesis. In the preparation of the modified tripeptide 3 required
for 13, the use of HCl-EtOAc for the selective deprotection of
the N-Boc group in the modified dipeptide 37 turned out to be
critical. The alternative use of neat formic acid followed by basic
workup (aqueous NaHCO3), which affords the free amine and
was applicable to dipeptides containing the natural i-Bu R3 side
chain, resulted in formation of the cyclic diketopiperazine
byproduct exclusively.
Figure 7.
address the contribution of each structural feature of this unusual
side chain (Figure 7). With the exception of 6, this required the
preparation of the N-methyl amino acid precursor, which was
typically assembled by N-methylation of the corresponding
N-Boc carboxylic acid (NaH, CH3I, THF) following the
procedure of Coggins and Benoiton.34 For the amino acid
precursor required for 22a, Ag2O-CH3I was employed and
provided the N-methyl amino acid methyl ester directly. In each
case, the extent of racemization was assessed by chiral phase
HPLC and comparison with a racemic standard. The precursors
to 15 (L-Ala), 16b, 19, 20, 21, 23 (L-Phe), and 24-26 (L-Tyr)
were commercially available as N-Boc amino acids or unpro-
tected amino acids. The precursors to 17 and 18 were prepared
by indole N-methylation or N-benzylation enlisting com-
mercially available tryptophan derivatives. The N-Boc precursor
to 22 was prepared by intermolecular Mitsunobu displacement
Residue 5. Residue 5 proved to be one of the most interesting
and surprising sites in the molecule. Its very unusual side chain
structure, incorporating both the backbone N-methyl amide and
the N-methoxy functionalization of the indole side chain,
suggested a very special role for this subunit. Moreover, the
results of the alanine scan with 15 (1800-fold loss in activity)
and the N-methyl deletion with 6 (>10 000-fold loss in activity)
confirmed a critical role for this residue and implied that even
subtle changes might have profound effects. As discussed earlier,
the N-methylation promotes the formation of a backbone cis
amide, which in turn is critical for adoption of the well-defined
solution conformation of 1. Consequently, the role of this
structural feature of residue 5 is clear. However, the role of the
side chain was not clear and the results of the examination of
additional analogues containing simplifying modifications proved
especially revealing. Fortunately, this residue was thoroughly
examined with a series of key analogues 15-26 designed to
(34) Coggins, J. R.; Benoiton, N. L. Can. J. Chem. 1971, 49, 1968-1971.
9
J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002 5437

A R T I C L E S
Chen et al.
of the primary alcohol derived from Boc-L-Glu(OtBu) (ClCO2-
Et, Et3N; NaBH4, 87%)35 with CbzNHOMe36 (DEAD, PPh3,
57%) as a key step. Notably, the coupling of the modified
dipeptide 43 with 46 to afford the modified tetrapeptide 47 was
conducted with EDCI-HOAt (2,6-lutidine, -30 °C, 3-5 h)
minimizing the competitive epimerization of the intermediate
active ester of the N-acyl-N-methyl carboxylic acid. Finally, 16a
(92%), 22b (99%), and 25 (89%) were obtained by hydrogena-
tion (H2, Pd-C, CH3OH or EtOAc, 25 °C, 1-4 h) of 1, 22a,
and 26, respectively.
The important key analogue comparison of 16a with 1
revealed that removal of the indole N-methoxy group reduced
activity 40-fold. Although this might initially suggest an
important interaction with the biological target, it was also found
to substantially impact the stability of the compound. Thus,
whereas 1 is stable to conventional handling and storage, 16a
rapidly discolors under analogous conditions. This observation
along with those discussed below suggest the role of the
N-methoxy group may not involve a productive interaction with
the biological target itself, but rather that it may be as simple
as improving compound stability or masking a free indole NH.
Unexpectedly and consistent with such an interpretation, a
number of simplified substitutions for residue 5 were found to
be potent and selective inhibitors of VCAM-1 expression. Most
notably, the NMe-L-Phe substitution 23 (IC50 ) 2.3 nM) proved
to be only 2-fold less potent than 1. Also surprisingly potent
were the related analogues 20 and 21, and the comparisons in
this series indicate that both the presence and position of the
hydrophobic aromatic side chain substituent are important (23
> 20 > 21). In contrast to 1 versus 16a, the incorporation of
polar substituents with 24-26 including a methoxy group
substantially reduced the activity of 23. It is the unanticipated
potent activity of 23 that highlights how easy it might be to
naively assign a special importance to an unusual residue that
can even be apparently validated by degradation studies (16a)
or an alanine substitution (15). Thus, the unbiased examination
of a reasonable range of alternative substitutions in a first pass
study accessible using this systematic chemical mutagenesis of
the cyclic depsipeptide can reveal important unanticipated
insights into structure-function properties. Not only is residue
5 key to the activity of 1, but a subsequent focus of SAR studies
on this residue is likely to be highly productive.
Figure 8.
testing of shipped samples of 17 and 18 provided IC50’s of 21
and 19 nM, respectively. However, independent testing of 16a,
17, and 18 employing in-house samples (Novartis) prepared by
degradation of the natural product and semisynthesis38 provided
IC50’s of 15, 3.6, and 1.7 nM,38 respectively. One interpretation
of these results is that the N-methoxy group of 1 is not integral
to its activity, but that it does appear to convey a unique stability
to the natural product contributing to its biological potency.
Consistent with this interpretation, the analogue 16b incorporat-
ing a benzothiophene in place of the N-methoxyindole proved
stable and nearly equipotent with the natural product (IC50 ) 2
nM).
Residue 6. Analogous to residue 2, the specialized structure
of its side chain and the results of the alanine scan with 27
(270-fold loss in activity) suggested residue 6 was especially
important to the properties of 1, second only to residue 5.
Consequently, the results of the examination of 27-29 (Figure
8) were especially interesting, revealing simplifications in the
structure that do not impact the properties. Their preparation
required the incorporation of L-Ala, (S)-2-aminooctanoic acid,33
and L-Leu into the modified tetrapeptide 2, Mitsunobu esteri-
fication with tripeptide 3, deprotection, and macrocyclization.
In contrast to the results of the replacement of this side chain
with a methyl group which resulted in a 270-fold reduction in
potency and loss of VCAM-1 selectivity, the removal of the
side chain methyl group and the chiral center resulted in an
analogue 28 with essentially the same potency and selectivity
as the natural product. Shortening the side chain with 29
removing the propyl extension of Leu reduced activity 7-fold.
In this regard, the lack of sensitivity to removal of the side chain
methyl group and chiral center with 28 contrasts the results with
residue 2 where it was more significant, whereas the sensitivity
to shortening the side chain with 29 was identical to the results
observed with residue 2. Thus, despite the more pronounced
reliance on the residue 6 side chain than residue 2 (compare
alanine scan), a significant simplification could be made with
28 which had little impact on potency.
3
Like the observations made with 1 and 12, the C2 epimers
of 24 and 26 are only 3- to 4-fold less potent against VCAM-1
expression. In the case of 24, its C23 epimer was >10-fold less
effective against ICAM-1, resulting in a greater VCAM-1 versus
ICAM-1 selectivity analogous to observations made with 1 and
12.
After the evaluation of the first round of compounds, three
additional residue 5 analogues (16b, 17, and 18) were prepared
for examination. The intent was to define in greater detail the
role of the structural features of the MTO residue indole. Like
the parent indole 16a, 17 and 18 proved to be modest inhibitors
of VCAM-1 expression. Moreover, they also proved less robust
on storage37 than 1, but easier to handle than 16a. Thus, initial
One feature that became apparent upon the final purification
of the analogues was the dramatic impact that the residue 6,
and residue 2, side chain had on the reverse phase HPLC
(35) Lee, B. H.; Gerfen, G. J.; Miller, M. J. J. Org. Chem. 1984, 49, 2418-
2423.
(36) Kawase, M.; Kitamura, T.; Kikugawa, Y. J. Org. Chem. 1989, 54, 3394-
3403.
(37) Compound 17 was found to be only 20% (Novartis) to 40% (Scripps) pure
after several months of storage whereas 16a discolors more rapidly.
(38) We thank Berndt Oberhauser (Novartis) for sharing these results and
observations in advance of publication.
9
5438 J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002

Synthesis of a Pharmacophore Library of HUN-7293 Analogues
A R T I C L E S
Figure 9.
retention times (82-100% MeOH-H2O gradient) even though
they did not appear to alter the conformational properties (1H
NMR). Retention times of 1 (21 min), 28 (19 min), 29 (14 min),
and 27 (10 min) were observed mirroring the order in decreased
potency. An analogous trend was observed with 1 versus the
residue 2 analogues 11 (18 min), 12 (13 min), and 10 (9.6 min).
Although the correlation between the two residues is not perfect,
the dramatic impact of the two side chains on the hydrophobic
character of the resulting analogue suggests their impact may
go beyond their interaction with a biological target. They may
additionally influence access to, transport to, or the ability to
reach a biological target by virtue of substantially altering the
hydrophobic character of the compound.
Figure 10.
Residue 7. One of the most interesting and unexpected series
of residue modifications centered on residue 7. The N-methyl
deletion series revealed that removal of the residue 7 N-methyl
group with 7 had a modest effect of lowering potency 7-fold,
but substantially improved the VCAM-1 versus ICAM-1
selectivity (180-fold selective). The remaining simplification
possible is removal of the side chain substituting Sar (NMe-
Gly) for NMe-Ala and providing 30. Consistent with the near
identical 1H NMR’s of 1 and 30 indicating little conformational
perturbation accompanying this change, 30 was found to be only
2- to 3-fold less potent than 1, and it also exhibited an enhanced
VCAM-1 selectivity (Figure 9).
Cumulative Simplifications. Following the evaluation of the
initial library, four additional analogues were prepared that
incorporate the acceptable side chain simplifications or changes
which enhanced the VCAM-1 selectivity. The first set (31 and
32) incorporate the residue 1 simplified methyl substitution (2-
fold reduction), the residue 2 and 6 simplified (S)-2-aminooc-
tanoic acid (6-fold and <2-fold reduction, respectively), and
the remarkable residue 5 NMe-Phe substitution (2-fold reduc-
tion). One of these contained the authentic residue 7 NMe-Ala
residue (31) while the second (32) incorporated the additional
simplified Sar substitution (2-fold reduction).
changes predicted a 22-fold loss in activity and the potency of
31 is only 4- to 5-fold different from this na¨ıve expectation. In
contrast, incorporating the additional residue 7 change embodied
in 32, which now contained five simplifications in the structure,
provided only a weakly active analogue that was 1400-fold less
active than 1 and 25-fold less active than the cumulative changes
would predict.
The second set (33 and 34) incorporated only two of the more
significant, yet productive, substitutions. Both incorporate the
remarkable residue 5 NMe-Phe simplification (2-fold reduction),
and 33 incorporates the residue 7 Sar substitution (2-fold
reduction) whereas 34 lacks the residue 7 N-methyl group (Ala
substitution, 7-fold reduction). Thus, 33 contains two simplifica-
tions that minimize an impact on potency, while 34 incorporates
one simplification that enhances VCAM-1/ICAM-1 selectivity
albeit with a more significant impact on potency. The first of
these analogues, 33, exhibited potent VCAM-1 expression
inhibition (IC50 ) 6 nM) that was exactly in line with
expectations based on the individual substitutions (2.6 × 2.3 )
5.98 vs 6) and it maintained the intrinsic VCAM-1/ICAM-1
selectivity. The second analogue, 34 (IC50 ) 43 nM), was
similarly in line with expectations based on the individual
substitutions (6.8 × 2.3 ) 16 vs 43) where it was only 2- to
3-fold less potent than the cumulative changes would predict.
Moreover, the VCAM-1/ICAM-1 selectivity improved with 34,
albeit not as dramatically as with 7 (Ala substitution).
The first of these analogues, 31, which contained four side
chain simplifications, proved to be especially interesting (Figure
10). It exhibited potent VCAM-1 expression inhibitory activity
(IC50 ) 104 nM) without impacting the intrinsic VCAM-1/
ICAM-1 selectivity, albeit at a level that is approximately 100-
fold less potent than 1. The cumulative effect of the individual
9
J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002 5439

A R T I C L E S
Chen et al.
Conclusions
Although similar approaches have been disclosed to optimize
or explore a single residue within a naturally occurring cyclic
peptide, we are not aware of an instance where all residues were
examined simultaneously. Moreover, a more common approach
to initial SAR studies of such natural products has been
systematic derivatization or degradation studies to establish the
structural features important for activity. These efforts, like an
alanine scan, provide an incomplete data set on which to base
future studies.
An assessment of the impact of each residue, each side chain,
and each structural feature of 1 was established enlisting a
complete library of HUN-7293 analogues composed of system-
atic single point changes in the structure prepared by parallel
solution-phase synthesis. This generalizable library approach
of chemical mutagenesis of the cyclic depsipeptide provides a
detailed first level structure-activity study that subsequent
optimization efforts can be confidently based on. One of the
most striking conclusions of the study is that a simple alanine
scan of the naturally occurring depsipeptide may provide a sense
of the importance of each residue, but it provides an incomplete
data set that is problematic to draw substantive conclusions from.
A second surprising observation that was unanticipated given
that 6 of the 7 residues of 1 contain nonstandard side chains or
are modified by N-methylation is that each nonstandard residue
or side chain could be simplified without substantially impacting
Acknowledgment. We gratefully acknowledge the financial
support of the National Institute of Health (CA41101) and award
of a Erwin Schro¨dinger postdoctoral fellowship for M.B. (2000-
2001). We thank Berndt Oberhauser (Novartis) for copies of
1
the H NMR of naturally occurring 8 for comparison and for
3
discussions relating to C2 -epi-1 (ref 28) and 16-18 (ref 38).
Supporting Information Available: Experimental data for the
synthesis of 1 and 4-34, full characterization data for 1, and
4-34, and all synthetic intermediates and 1H NMR spectra for
1, 3-34, 37, 42, 45, 47, and 48 (PDF) are provided. This
material is available free of charge via the Internet at http://
pubs.acs.org.
3
potency (e.g. 8, 11/12, C2 -epi-1, 16b/23, 28/29, 7/30) and, in
some instances, with improvements in the VCAM-1/ICAM-1
selectivity. Many, if not most, of these latter active analogues
would not have been easily anticipated based on the results of
the alanine scan.
JA020166V
9
5440 J. AM. CHEM. SOC. VOL. 124, NO. 19, 2002