Anthranilate sulfonamide hydroxamate TACE inhibitors. Part 1: Structure-based design of novel acetylenic P1′ groups

Article abstract of DOI:10.1016/S0960-894X(02)00135-X

The structure-based design of potent sulfonamide hydroxamate TACE inhibitors bearing novel acetylenic P1′ groups has led to compounds with excellent in vitro potency against TACE and selectivity over MMP-1.

Full text of DOI:10.1016/S0960-894X(02)00135-X

Bioorganic & Medicinal Chemistry Letters 12 (2002) 1195–1198
Anthranilate Sulfonamide Hydroxamate TACE Inhibitors.
Part 1: Structure-Based Design of Novel Acetylenic P10 Groups
James M. Chen,a,* Guixian Jin,a Amy Sungb and Jeremy I. Levina,*
aWyeth-Ayerst Research, 401N. Middletown Road, Pearl River, NY 10965, USA
bWyeth-Ayerst Research, PO Box CN-8000, Princeton, NJ 08543, USA
Received 10 December 2001; accepted 4 February 2002
Abstract—The structure-based design of potent sulfonamide hydroxamate TACE inhibitors bearing novel acetylenic P10 groups has led to
compounds with excellent in vitro potency against TACE and selectivity over MMP-1. # 2002 Elsevier Science Ltd. All rights reserved.
Rheumatoid arthritis (RA) is a chronic debilitating
autoimmune disease that results in inflammation of the
joints and ultimately irreversible joint erosion.1 It
afflicts 1% of the population. The recent success of
Enbrel1, a soluble tumor necrosis factor-a (TNF-a)
receptor, has been a major advance in the treatment of
RA patients.2 However, the search for improved therapies
with orally active agents that modulate levels of the pro-
inflammatory cytokine TNF-a for RA, as well as other
inflammatory diseases such as Crohn’s disease, is ongo-
ing. One avenue of exploration has been the identification
of small molecule inhibitors of TNF-a converting enzyme
(TACE), the enzyme responsible for the cleavage of 26
kDa membrane-bound TNF into its soluble form, a 17
kDa non-covalently bound homotrimer.3
MMP/TACE inhibitors exemplified by 3a and 3b, and
the more selective TACE inhibitor 3c, has been dis-
cussed (Fig. 2). We now wish to report on the structure-
based design that has led to the discovery of novel pro-
pargylic ether P10 groups, providing potent inhibitors of
TACE with selectivity over MMP-1.
The experimental structure of TACE (ADAM-17), a
member of the adamalysin family of enzymes, was not
known during the initial stages of our discovery effort.
Since the linear sequence of TACE shares 15–28%
This process has been aided by the fact that many of the
now ubiquitous inhibitors of matrix metalloproteinases
(MMPs) have also been found to inhibit TACE.4
Examples of known TACE inhibitors include succinate
hydroxamate 1,5 and macrocylic hydroxamic acids 2a
and 2b (Fig. 1).6
Figure 1. Literature TACE inhibitors.
In an effort to find more effective therapies for the
treatment of diseases such as atherosclerosis, osteo-
arthritis, and cancer, as well as RA, we have recently
disclosed a series of sulfonamide hydroxamic acid inhi-
bitors of MMP-1, MMP-9, MMP-13, and TACE, based
on an anthranilic acid scaffold.7 The SAR of the
anthranilic acid 3- and 5-positions (3, R1 and Br) as well
as the P10 moiety (3, R2), leading to broad spectrum
*Corresponding authors. Tel.: +1-845-602-3053; fax: +1-845-602-
5561; e-mail: levinji@war.wyeth.com (J.I.M.); tel.: +1-650-522-5269;
fax: +1-650-522-5899; e-mail: jchen@gilead.com (J.M.C.)
Figure 2. Sulfonylated anthranilate hydroxamic acids.
0960-894X/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(02)00135-X

1196
J. M. Chen et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1195–1198
identity with certain MMP and adamalysin sequences,
and includes the consensus zinc binding motif
HEXGHXLGXXH, a homology model of TACE was
constructed based on its alignment with adamalysin II,
with which it shares 24% identity. Construction of this
model also assumed that small molecules that inhibit
both MMPs and TACE would have a similar binding
motif in both enzyme classes. It was hoped that this
model would provide us with accurate information on
the enzyme–inhibitor interaction sites, particularly in
the zinc binding and S10 regions.
structure of TACE had several interesting structural
characteristics. First, the catalytic zinc motif was highly
similar to known matrix metalloproteinases. Second, in
contrast to the MMPs the model contained only the
single catalytic zinc metal and no other structural
metals. Finally, there were three disulfide bridges within
the catalytic domain serving to stabilize the protein in
the absence of structural metals.
Concurrent with this analysis, the X-ray structure of
TACE with a peptide based inhibitor bound in the
active site was solved at 2.0 A.9 Figure 4 depicts the
ribbon diagrams from the TACE X-ray structure and
the TACE homology model. While there are clear dif-
ferences between the two structures in the outer loop
regions, the areas around the active site zinc (in green)
are quite similar. Figure 5 compares the S10 binding site
surface of the homology model to the X-ray structure.
The three histidine residues chelating the active site zinc
(red) lie to the left of the S10 pocket. It is evident that
the size and shape of the extended S10–S30 channel
shown in the X-ray structure is highly similar to the one
predicted by the homology model. Both TACE struc-
tures indicate that the S10 pocket consists of an initial
region resembling the shallow S10 pocket of MMP-1 in
size and depth. Importantly, the rest of the pocket is no
longer extended linearly, as in the MMPs. Rather the
extended channel is almost perpendicular to the S10
pocket, directed toward the S30 surface. This is clearly
visible in Figure 6, in which compound 3a is docked in the
S10 pocket described by the TACE homology model, with
the zinc-chelating histidine residues on the left. As we had
seen previously for analogues of 3a bound to MMP-13,
the hydroxamate of the inhibitor can chelate to the cata-
lytic zinc in a bidentate fashion, the sulfonamide group
can form a hydrogen bond with the protein backbone and
the methoxyphenyl group is positioned at the entrance of
the hydrophobic S10 pocket.10 The CH3–O–phenyl bonds
are in the eclipsed conformation pointing the methoxy
group directly down the S10–S30 channel of TACE.
Figure 3 is a partial alignment of TACE and adamalysin
II showing the conserved zinc binding sequences. The
three conserved histidines coordinated to the catalytic
zinc metal are shown in red, as is the conserved
methionine found in the Met turn. The residues in pur-
ple are the positions that define the length and depth of
the S10 pocket within the zinc metalloproteinase family.
Based on the conserved nature of the amino acids in this
region of the enzyme, an initial TACE homology model
was constructed with a shallow S10 pocket similar to
that of MMP-1. However, this was not in accord with
the potency and selectivity for TACE displayed by some
of our inhibitors bearing lengthy P10 groups.7b While
extending the depth of the TACE S10 pocket was unlikely,
based on sequence homology, the S10 and S30 pockets
could be connected through a channel, based on specific
side-chain conformations of the leucine (blue) and
valine (magenta) residues for adamalysin II in Figure 3.
These correspond to the glutamate (blue) and leucine
(magenta) residues of TACE in Figure 3. This channel
results in an S10 pocket that is unique in size and shape
when compared to MMP-1, MMP-9, and MMP-13.
Subsequently the full catalytic domain of TACE was
constructed using the COMPOSER (SYBYL) method.
An iterative refinement procedure was utilized for opti-
mizing the initial homology model.8 The final minimized
Given this information, an initial design strategy for
enhancing the TACE potency and selectivity of 3a was
proposed, focusing on utilizing a carbon–carbon triple
Figure 3. Zinc binding sequences of adamalysin II and TACE.
Figure 4. Comparison of ribbon diagrams for the TACE X-ray
structure (left) and TACE homology model (right).
Figure 5. Comparison of TACE active site for homology model (left)
and X-ray (right), showing S10-S30 channel.

J. M. Chen et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1195–1198
1197
bond to occupy the S10–S30 channel (Fig. 7). Extending
an acetylenic moiety from the methoxy carbon of 3a,
resulting in propargyl ether 6a, appeared to be ideal
because the shape of the acetylene matches the shape of
the channel cavity, increasing favorable van der Waal’s
contacts with the extended channel region. It was also
expected that by varying the length of the propargylic
P10 group increased selectivity for TACE could be
effected. For example, the length and preferred con-
formation of inhibitor 6b should provide a better fit for
the S10 pocket of TACE than for that of MMP-1,
MMP-9, or MMP-13. Examination of the binding site
features of the TACE model compared to the active
sites of MMPs was accomplished using the analysis
capabilities within SYBYL. Depicted in Figure 8 is the
binding motif of designed inhibitor 6b in the TACE
homology model (gold) compared to models of MMP-1
(green) and MMP-13 (brown). In each enzyme the
catalytic zinc atom (blue) is situated adjacent to the S10
binding pocket. Among the MMPs, the structural dif-
ferences between the S10 pockets are based largely on
the linear length or depth of each pocket.11 The rigid
4-carbon chain of butynyl ether 6b, in bent or extended
conformations, cannot be accommodated in the S10
pocket of MMP-1, MMP-9 (not shown), or MMP-13.
However, the shape of TACE S10 pocket differs sig-
nificantly from that of the MMPs, allowing the butynyl
ether of 6b to fit.
A series of related alkyl and propargylic ether deriva-
tives was therefore desired to examine the effect of
length and rigidity of the P10 group on TACE potency
and selectivity in vitro. The desired sulfonamide
hydroxamic acids were prepared as described in Scheme
1. Thus, the desired anthranilic acid ester, 4, was reacted
with 4-fluorobenzenesulfonyl chloride and then alky-
lated with iodomethane to provide the 4-fluorobenzene
sulfonamide, 5a. SNAr displacement of the fluoride of 5a
with excess 2-butyn-1-ol in the presence of sodium
hydride, followed by in situ isomerization to the allenyl
ether and subsequent aqueous HCl quench gave phenol
5b.12 Mitsunobu alkylation of phenol 5b with the
appropriate alkyl or propargylic alcohol then gave ethers
5c. The fully functionalized anthranilate ester sulfona-
mides were then hydrolyzed to the carboxylic acids and
converted into the hydroxamic acids, 6, via the acid
chlorides.
The in vitro activities for a series of anthranilate
hydroxamates bearing alkyl and propargylic ether P10
groups are shown in Table 1.13 Thus, our original design
target, propargyl ether 6a, is slightly more potent
against TACE than methoxy derivative 3a, but has
virtually the same TACE/MMP selectivity profile.
Attaching a methyl group to the terminal acetylene of
6a provides butynyl ether 6b, which is essentially equi-
potent to 6a against TACE enzyme. In addition, as
predicted, 6b has substantially increased selectivity for
Figure 6. Inhibitor 3a in the TACE active site with the S10 pocket
shown.
Figure 8. Designed inhibitor 6b in the S10 pockets of MMP-1 (left),
MMP-13 (center), and TACE (right).
Scheme 1. (i) 4-F-PhSO2Cl, TEA; (ii) CH3I, K2CO3; (iii) (a)
CH3CCCH2OH, NaH, DMF, 80 ꢀC; (b) HCl, H2O; (iv) R1OH, PPh3,
DEAD; (v) NaOH; (vi) (a) (COCl)2, DMF; (b) NH2OH.
Figure 7. Designed propargylic ether inhibitor 6b bound to TACE.

1198
J. M. Chen et al. / Bioorg. Med. Chem. Lett. 12 (2002) 1195–1198
Table 1. In vitro activity of alkyl and propargylic ether P10 groups
References and Notes
1. Choy, E. H. S.; Panayi, G. S. N. Engl. J. Med. 2001, 344, 907.
2. Garrison, L.; McDonnell, N. D. Ann. Rheum. Dis. 1999,
58 (Suppl. I), 165.
3. (a) Nelson, F. C.; Zask, A. Exp. Opin. Invest. Drugs
1999, 8, 383. (b) Lowe, C. Exp. Opin. Ther. Pat. 1998, 8, 1309.
(c) Newton, R. C.; Decicco, C. P. J. Med. Chem. 1999, 42,
2295.
Compd
R
MMP-1a MMP-9a MMP-13a TACEa
4. Skiles, J. W.; Gonnella, N. C.; Jeng, A. Y. Curr. Med.
Chem. 2001, 8, 425.
5. Barlaam, B.; Bird, T. G.; Lambert-van der Brempt, C.;
Campbell, D.; Foster, S. J.; Maciewicz, R. J. Med. Chem.
1999, 42, 4890.
3a
6a
6b
6c
6d
CH3
114
113
1616
2488
11
15
304
21
21
5211
154
68
32
CH2CCH
CH2CCCH3
(CH2)3CH3
16
67
34
CH2CC(CH2)3CH3 53% (10)
389
701
6. (a) Xue, C.-B.; Voss, M. E.; Nelson, D. J.; Duan, J. J.-W.;
Cherney, R. J.; Jacobson, I. C.; He, X.; Roderick, J.; Chen, L.;
Corbett, R. L.; Wang, L.; Meyer, D. T.; Kennedy, K.;
DeGrado, W. F.; Hardman, K. D.; Teleha, C. A.; Jaffee, B. D.;
Liu, R.-Q.; Copeland, R. A.; Covington, M. B.; Christ, D. D.;
Trzaskos, J. M.; Newton, R. C.; Magolda, R. L.; Wexler,
R. R.; Decicco, C. P. J. Med. Chem. 2001, 44, 2636. (b) Xue,
C.-B.; He, X.; Corbett, R. L.; Roderick, J.; Wasserman, Z. R.;
Liu, R.-Q.; Jaffee, B. D.; Covington, M. B.; Qian, M.; Trzas-
kos, J. M.; Newton, R. C.; Magolda, R. L.; Wexler, R. R.;
Decicco, C. P. J. Med. Chem. 2001, 44, 3351. (c) For another
series of macrocyclic TACE inhibitors, see: Holms, J.; Mast,
K.; Marcotte, P.; Elmore, I.; Li, J.; Pease, L.; Glaser, K.;
Morgan, D.; Michaelides, M.; Davidsen, S. Bioorg. Med.
Chem. Lett. 2001, 11, 2907.
7. (a) Levin, J. I.; Du, M. T.; DiJoseph, J. F.; Killar, L. M.;
Sung, A.; Walter, T.; Sharr, M. A.; Roth, C. E.; Moy, F. J.;
Powers, R.; Jin, G.; Cowling, R.; Skotnicki, J. S. Bioorg. Med.
Chem. Lett. 2001, 11, 235. (b) Levin, J. I.; Chen, J.; Du, M.;
Hogan, M.; Kincaid, S.; Nelson, F.; Venkatesan, A. M.;
Wehr, T.; Zask, A.; DiJoseph, J.; Killar, L. M.; Skala, S.;
Sung, A.; Sharr, M.; Roth, C.; Jin, G.; Cowling, R.; Mohler,
K. M.; Black, R. A.; March, C. J.; Skotnicki, J. S. Bioorg.
Med. Chem. Lett. 2001, 11, 2189. (c) Levin, J. I.; Chen, J. M.;
Du, M. T.; Nelson, F. C.; Wehr, T.; DiJoseph, J. F.; Killar,
L. M.; Skala, S.; Sung, A.; Sharr, M. A.; Roth, C. E.; Jin, G.;
Cowling, R.; Di, L.; Sherman, M.; Xu, Z. B.; March, C. J.;
Mohler, K. M.; Black, R. A.; Skotnicki, J. S. Bioorg. Med.
Chem. Lett. 2001, 11, 2975.
8. Chen, J. M.; Nelson, F. C.; Levin, J. I.; Mobilio, D.; Moy,
F. J.; Nilakantan, R.; Zask, A.; Powers, R. J. J. Am. Chem.
Soc. 2000, 122, 9648.
9. Maskos, K.; Fernandez-Catalan, C.; Huber, R.; Bour-
enkov, G. P.; Bartunik, H.; Ellestad, G. A.; Reddy, P.; Wolf-
son, M. F.; Rauch, C. T.; Castner, B. J.; Davis, R.; Clarke,
H. R. G.; Petersen, M.; Fitzner, J. N.; Cerreti, D. P.; March,
C. J.; Paxton, R. J.; Black, R. A.; Bode, W. Proc. Natl. Acad.
Sci. U.S.A. 1998, 95, 3408.
10. (a) Moy, F. J.; Chanda, P. K.; Chen, J. M.; Cosmi, S.;
Edris, W.; Levin, J. I.; Powers, R. J. Mol. Biol. 2000, 302, 671.
(b) Moy, F. J.; Chanda, P. K.; Cosmi, S.; Edris, W.; Levin,
J. I.; Powers, R. J. Biomol. NMR 2000, 17, 269.
11. (a) Bode, W.; Fernandez-catalan, C.; Tschesche, H.;
Grams, F.; Nagase, H.; Maskos, K. Cell. Mol. Life Sci. 1999,
55, 639. (b) Lovejoy, B.; Welch, A. R.; Carr, S.; Luong, C.;
Broka, C.; Hendricks, R. T.; Campbell, J. A.; Walker,
K. A. M.; Martin, R.; Van Wart, H.; Browner, M. F. Nat.
Struct. Biol. 1999, 6, 217.
aIC50 (nM) or % inhibition (mM).
TACE over the MMPs, validating our design strategy.
In particular, 6b is now 100-fold selective for TACE
over MMP-1. That the carbon–carbon triple bond
increases both TACE potency and selectivity is shown
by a comparison of butynyl ether 6b to butyl ether 6c,
which is less active against TACE and nonselective ver-
sus MMP-9 and MMP-13. Extending the acetylenic P10
group further into the S10–S30 channel, as in heptynyl
ether 6d, still affords a potent inhibitor of TACE and
further increases selectivity over both MMP-1 and
MMP-13 relative to 6b. Propargylic ethers 6b and 6d
have therefore provided a dramatic improvement in
TACE/MMP selectivity while retaining or improving
TACE potency.
In summary, we have succeeded in utilizing structure-
based computational design strategies to discover a
novel series of potent TACE inhibitors with excellent
selectivity over MMP-1. Acetylenic P10 groups were
suggested by a homology model of TACE that had
revealed a unique S10–S30 channel for the enzyme, later
confirmed by an X-ray crystal structure. This acetylenic
P10 moiety is also amenable to further chemical modifi-
cations, providing a handle for drug property optimiza-
tion. The extension of this work to additional
propargylic ether P10 groups and their evaluation versus
TACE enzyme and in a cellular assay measuring the
inhibition of TNF-a production is described in the fol-
lowing communication.
The future therapeutic use of TACE inhibitors for the
treatment of diseases modulated by TNF-a may depend
on the resolution of several issues. Included among
these are the determination of a desirable selectivity
profile with respect to the MMPs and the ADAMs, as
well as the impact of TACE inhibition on a-secretase
activity. We believe that the potency and selectivity of
this new class of inhibitor represents an excellent lead
for the development of drugs for the treatment of
rheumatoid arthritis.
12. Levin, J. I.; Du, M. T. Synth. Commun. 2002, in press.
13. (a) Weingarten, H.; Feder, J. Anal. Biochem. 1985, 147,
437. (b) Inhibitor concentrations were run in triplicate. MMP
IC50 determinations were calculated from a four-parameter
logistic fit of the data within a single experiment. (c) Jin, G.;
Black, R.; Wolfson, M.; Rauch, C.; Ellestad, G. A.; Cowling,
R., Anal. Biochem. 2002, 302, 269.
Acknowledgements
We thank Drs. Roy A. Black, Rebecca Cowling, and
Jerauld Skotnicki for helpful discussions during the
course of this work.