Structure-based design of a novel, potent, and selective inhibitor for MMP-13 utilizing NMR spectroscopy and computer-aided molecular design

Article abstract of DOI:10.1021/ja001547g

The high-resolution NMR solution structure of the catalytic fragment of human collagenase-3 (MMP-13) was used as a starting point for structure-based design of selective inhibitors for MMP-13. The major structural difference observed between the MMP struc

Full text of DOI:10.1021/ja001547g

9648
J. Am. Chem. Soc. 2000, 122, 9648-9654
Structure-Based Design of a Novel, Potent, and Selective Inhibitor for
MMP-13 Utilizing NMR Spectroscopy and Computer-Aided
Molecular Design
James M. Chen,*,‡ Frances C. Nelson,† Jeremy I. Levin,† Dominick Mobilio,
Franklin J. Moy, Ramaswamy Nilakantan, Arie Zask,† and Robert Powers*
Contribution from the Department of Biological Chemistry, Wyeth Research, 85 Bolton St., Cambridge,
Massachusetts 02140 and the Department of Chemical Sciences, Wyeth Research, 401 N. Middletown Rd.,
Pearl RiVer, NY 10965
ReceiVed May 4, 2000. ReVised Manuscript ReceiVed July 12, 2000
Abstract: The high-resolution NMR solution structure of the catalytic fragment of human collagenase-3 (MMP-
13) was used as a starting point for structure-based design of selective inhibitors for MMP-13. The major
structural difference observed between the MMP structures is the relative size and shape of the S1′ pocket
where this pocket is significantly longer for MMP-13, nearly reaching the surface of the protein. On the basis
of the extended nature of the MMP-13 S1′ pocket an inhibitor potent and selective for MMP-13 was designed
from an initial high throughput screening (HTS) lead. CL-82198 was identified as a weak (10 µM) inhibitor
against MMP-13 while demonstrating no activity against MMP-1, MMP-9, or the related enzyme TACE. The
drug-like properties of CL-82198 made it an ideal candidate for optimization of enzyme potency and selectivity.
On the basis of NMR binding studies, it was shown that inhibitor CL-82198 bound within the entire S1′
pocket of MMP-13 which is the basis of its selectivity against MMP-1, MMP-9, and TACE. A strategy utilizing
this information was devised for designing new inhibitors that showed enhanced selectivity toward MMP-13.
Our design strategy combined the critical selectivity features of CL-82198 with the known potency features of
a nonspecific MMP inhibitor (WAY-152177) to generate a potent and selective MMP-13 inhibitor (WAY-
170523). WAY-170523 has an IC50 of 17 nM for MMP-13 and showed >5800-, 56-, and >500-fold selectivity
against MMP-1, MMP-9, and TACE, respectively.
A structure-based approach to designing potent and selective
the basis of differential expression in normal breast tissues and
in breast carcinoma. Recently, both an NMR and X-ray structure
of inhibited MMP-13 have been reported.8,9
inhibitors has established itself as an important component of
the drug development process (for reviews see refs 1 and 2).
This is evident by the extensive structural data available for
the matrix metalloproteinase (MMP) family of enzymes and the
emergence of unique inhibitors based on this structural informa-
tion (for reviews see refs 3-7). The MMPs are involved in the
degradation of the extracellular matrix that is associated with
normal tissue remodeling processes such as pregnancy, wound
healing, and angiogenesis. MMP expression and activity is
highly controlled because of the degradative nature of these
enzymes where the apparent loss in this regulation results in
the pathological destruction of connective tissue and the ensuing
disease state. Thus, the MMPs are a highly active set of targets
for the design of therapeutic agents for the disease areas of
arthritis and oncology. The MMP family is composed of a
number of enzymes where MMP-13 was recently identified on
The MMPs are generally categorized based on their substrate
specificity, where the collagenase subfamily of MMP-1, MMP-
8, and MMP-13 selectively cleaves native interstitial collagens
(types I, II, and III). It is likely that only a subset of MMP
enzymes will be involved in a particular disease as is evident
by the overexpression of MMP-13 in breast carcinoma and
MMP-1 in papillary carcinomas. Therefore, the current paradigm
in the development of MMP inhibitors is to design specificity
into the structures of the small molecules based on the
postulation that a broad-spectrum MMP inhibitor would provide
a higher exposure to toxic side effects.
Nevertheless, there is a close similarity in the overall three-
dimensional fold for the MMP family of proteins, which is
consistent with the relatively high sequence homology (>40%),
making the design of a selective MMP inhibitor a nontrivial
task. The concept of designing selective MMP inhibitors has
been facilitated by the extensive structural data available for
the MMPs where a significant difference in the size and shape
of the S1′ pocket has been observed.8-26 This structural
* Address correspondence to these authors at the Department of
Biological Chemistry.
† Department of Chemical Sciences.
‡ Current address: Tularik Inc., Two Corporate Dr., South San Francisco,
CA 94010. Phone: (650)-825-7458. E-mail: jchen@tularik.com.
(1) Gubernator, K.; Boehm, H. J. Methods Princ. Med. Chem 1998, 6,
15-36.
(8) 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, 217-221.
(2) Kubinyi, H. Curr. Opin. Drug DiscoVery DeV. 1998, 1, 4-15.
(3) Woessner, J. F., Jr. FASEB J. 1991, 5, 2145-54.
(4) Ries, C.; Petrides, E. Biol. Chem. Hoppe-Seyler 1995, 376, 345-55.
(5) Browner, M. F. Perspect. Drug DiscoVery Des. 1995, 2, 343-51.
(6) Morphy, J. R.; Millican, T. A.; Porter, J. R. Curr. Med. Chem. 1995,
2, 743-62.
(9) Moy, F. J.; Chanda, P., K.; Chen, J., M.; Cosmi, S.; Edris, W.; Levin,
J. I.; Wilhelm, J.; Powers, R. J. Mol. Biol. 2000, 302, 673-691.
(10) Bode, W.; Reinemer, P.; Huber, R.; Kleine, T.; Schnierer, S.;
Tschesche, H. EMBO J. 1994, 13, 1263-9.
(7) Zask, A.; Levin, J. I.; Killar, L. M.; Skotnicki, J. S. Curr. Pharm.
Des. 1996, 2, 624-661.
(11) Gooley, P. R.; O’Connell, J. F.; Marcy, A. I. Nat. Struct. Biol. 1994,
1, 111-18.
10.1021/ja001547g CCC: $19.00 © 2000 American Chemical Society
Published on Web 09/21/2000

Structure Based Design of SelectiVe MMP-13 Inhibitors
J. Am. Chem. Soc., Vol. 122, No. 40, 2000 9649
Patrick method) into related sets of structures.31 The active structures
were analyzed by their calculated physical properties (total number of
non-hydrogen atoms, number of heteroatoms, number of hydrogen-
bond donors, number of hydrogen-bond acceptors, calculated log P,
molecular weight) as well as known cytotoxicity as measured by the
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide, when
available.
difference across the MMP family provides an obvious approach
for designing specificity into potent MMP inhibitors by design-
ing compounds that appropriately fill the available space in the
S1′ pocket while taking advantage of sequence differences. A
number of examples have been previously reported using this
approach where some selectivity between MMPs has been
achieved by incorporating a biphenyl into the S1′ pocket.8,27,28
The recent NMR and X-ray structures for MMP-13 indicated
an unusually large S1′ pocket for MMP-13 suggesting that it
may be feasible to take advantage of this feature to design a
selective MMP-13 inhibitor.8,9 Here we report a novel, potent
inhibitor highly selective for MMP-13 that has been identified
by optimizing an initial lead from high throughput screening
based on an NMR structure of its complex with MMP-13 in
conjunction with computer aided molecular design.
The properties of each compound were normalized against a set of
known orally bioavailable drugs.33 Specifically, the mean and standard
deviation of the six calculated properties were determined for the set
of orally bioavailable drugs. Then, for each assayed compound, the
normalized property value was obtained by subtracting the mean value
and dividing by the standard deviation for the known drugs. The
normalized property profiles for each structure in a cluster were plotted
side by side as a column graph to obtain the cluster’s profile, clusters
containing a high percentage of cytotoxic compounds (as measured by
the MTT assay described above) would be eliminated from further
consideration. Similarly, any significant deviation from the profiles of
the orally bioavailable drugs would also eliminate the cluster from
further analysis.
Experimental Methods
High Throughput Screening Analysis. A total of 58079 compounds
were analyzed for MMP-13 inhibitor activity in a high throughput screen
(HTS). Enzyme activity was assessed using nonlabeled recombinant
MMP-13 and a peptide substrate.29,30 MMP-13 activity was determined
using 50 mM HEPES buffer with 5 mM CaCl2, 0.02% Brij 35
(polyoxyethylene 23 lauryl ether), and 0.5% Cysteine at pH 7.0. The
HTS was a kinetic assay where compounds were screened at a
concentration of 10 µg/mL with final MMP-13 and DMSO concentra-
tions of 5 nM and <1.0%, respectively. A total of 385 compounds
were identified that inhibited g40% of MMP-13 activity. From the
385 actives, 162 structures were eliminated by visual inspection using
considerations such as synthetic accessibility and reactivity. The
remaining 223 active compounds were clustered (using the Jarvis-
Synthesis of WAY-170523. General: Melting points were obtained
on a Mel-Temp apparatus and are uncorrected. 1H spectra were recorded
at 75, 100, 300, or 400 MHz as indicated and are expressed in δ (ppm)
with TMS as an internal standard. Infrared spectra were recorded as a
KBr press on a Niclete 710. Low-resolution mass spectra were
determined on a Micromass platform electrospray ionization quadrapole
mass spectrometer. High-resolution mass spectra were determined on
a Bruker 9.4 T FTMS. Elemental analyses were performed on a Perkin-
Elmer Series II CHNS/O Analyzer and are within 0.4% of theory. Thin-
layer chromatography was performed on silica gel 60 F-254 (EM
Reagents). Flash column chromatography was carried out using silica
gel 60 (230-400 mesh). Solvents and reagents were obtained from
commercial sources and used without additional purification. All
chemical yields are not optimized and generally represent the result of
a single experiment.
(12) Lovejoy, B.; Cleasby, A.; Hassell, A. M. Ann. N.Y. Acad. Sci. 1994,
732, 375-8.
(13) Lovejoy, B.; Hassell, A. M.; Luther, M. A. Biochemistry 1994, 33,
8207-17.
Benzofuran-2-carboxylic acid (2-hydroxyethyl)amide (11): To
a solution of 2-benzofurancarboxylic acid (1.0 g, 6.16 mmol), 1-(3-
dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (1.0 g, 6.16
mmol), and 1-hydroxybenzotriazole (1.08 g, 8.0 mmol) in DMF (12
mL) at 0 °C was added ethanolamine (0.37 mL, 6.16 mmol) and
4-methylmorpholine (2.04 mL, 9.24 mmol). The reaction was stirred
at 0 °C for an additional 10 min and then warmed to room temperature
and stirred for 4 h. The reaction mixture was then diluted with ethyl
acetate, washed three times with H2O, once with NaHCO3 (saturated),
and once with brine, dried over MgSO4, and concentrated in vacuo to
provide 0.55 g (43% yield) of the desired product as a pale yellow
solid. Mp 90-91 °C. IR (KBr) 3349, 3256, 1614, 1597, 1539, 1302,
(14) Lovejoy, B.; Cleasby, A.; Hassell, A. M. Science 1994, 263, 375-
7.
(15) Spurlino, J. C.; Smallwood, A. M.; Carlton, D. D.; Banks, T. M.;
Vavra, K. J.; Johnson, J. S.; Cook, E. R.; Falvo, J.; Wahl, R. C.; et al.
Proteins: Struct., Funct., Genet. 1994, 19, 98-109.
(16) Stams, T.; Spurlino, J. C.; Smith, D. L.; Wahl, R. C.; Ho, T. F.;
Qoronfleh, M. W.; Banks, T. M.; Rubin, B. Nat. Struct. Biol. 1994, 1, 119-
123.
(17) Becker, J. W.; Marcy, A. I.; Rokosz, L. L. Protein Sci. 1995, 4,
1966-76.
(18) Gonnella, N. C.; Bohacek, R.; Zhang, X.; Kolossvary, I.; Paris, C.
G.; Melton, R.; Winter, C.; Hu, S.-I.; Ganu, V. Proc. Natl. Acad. Sci. U.S.A.
1995, 92, 462-6.
(19) Van Doren, S. R.; Kurochkin, A. V.; Hu, W.; Ye, Q.-Z.; Johnson,
L. L.; Hupe, D. J.; Zuiderweg, E. R. P. Protein Sci. 1995, 4, 2487-2498.
(20) Botos, I.; Scapozza, L.; Zhang, D.; Liotta, L. A.; Meyer, E. F. Proc.
Natl. Acad. Sci. U.S.A. 1996, 93, 2749-2754.
1
1182, 1069, 743 cm-1. H NMR (300 MHz, DMSO-d6) δ 3.43 (dq, J
) 5.85, 52.0 Hz, 5 H), 4.78 (t, J ) 5.60 Hz, 1 H), 7.33 (m, 1 H), 7.46
(m, 1 H), 7.53 (d, J ) 0.84 Hz, 1 H), 7.65 (dd, J ) 0.66, 8.56 Hz, 1
H), 7.73 (d, J ) 18.69 Hz, 1 H), 8.63, (t, J ) 5.70 Hz). 13C NMR (75
MHz, DMSO-d6) δ 41.93, 59.89, 109.57, 112.08, 123.06, 124.0, 127.07,
127.51, 149.58, 154.49, 158.53. Electrospray mass spec: m/z 205.8
[(M + H)+ C11H12NO3 requires 206.07]. Anal. Calcd for C11H11NO3:
C, 64.38; H, 5.40; N, 6.83. Found: C, 64.20; H, 5.30; N, 6.84.
(21) Broutin, I.; Arnoux, B.; Riche, C.; Lecroisey, A.; Keil, B.; Pascard,
C.; Ducruix, A. Acta Crystallogr., Sect. D: Biol. Crystallogr. 1996, D52,
380-92.
(22) Gooley, P. R.; O’Connell, J. F.; Marcy, A. I.; Cuca, G. C.; Axel,
M. G.; Caldwell, C. G.; Hagmann, W. K.; Becker, J. W. J. Biomol. NMR
1996, 7, 8-28.
(23) Betz, M.; Huxley, P.; Davies, S. J.; Mushtaq, Y.; Pieper, M.;
Tschesche, H.; Bode, W.; Gomis-Rueth, F. X. Eur. J. Biochem. 1997, 247,
356-363.
2-[[(4-{2-[(1-Benzofuran-2-ylcarbonyl)amino]ethoxy}phenyl)-
sulfonyl](benzyl)amino]-3,5-dimethylbenzoic acid (13): Benzofuran-
2-carboxylic acid (2-hydroxyethyl)amide (157 mg, 0.76 mmol) was
dissolved in DMF (1.5 mL) and cooled to 0 °C. Sodium hydride (76
mg, 1.9 mmol) was added and the reaction was stirred for 30 min. To
this solution was added acid 12 (289 mg, 0.73 mmol) in DMF (2 mL).
The reaction was held at 0 °C for an additional 30 min and then allowed
to warm to room temperature and stirred overnight. The reaction was
then quenched with 1 M HCl and extracted three times with CH2Cl2.
(24) Gonnella, N. C.; Li, Y.-C.; Zhang, X.; Paris, C. G. Bioorg. Med.
Chem. 1997, 5, 2193-2201.
(25) Moy, F. J.; Chanda, P. K.; Cosmi, S.; Pisano, M. R.; Urbano, C.;
Wilhelm, J.; Powers, R. Biochemistry 1998, 37, 1495-1504.
(26) Moy, F. J.; Chanda, P., K.; Chen, J. M.; Cosmi, S.; Edris, W.;
Skotnicki, J. S.; Wilhelm, J.; Powers, R. Biochemistry 1999, 38, 7085-
7096.
(27) Hajduk, P. J.; Sheppard, G.; Nettesheim, D. G.; Olejniczak, E. J.
Am. Chem. Soc. 1997, 119, 5818-5827.
(28) Olejniczak, E. T.; Hajduk, P. J.; Marcotte, P. A.; Nettesheim J. Am.
Chem. Soc. 1997, 119, 5828-5832.
(29) Weingarten, H.; Feder, J. Anal. Biochem. 1985, 24, 437-440.
(30) Bickett, D. M.; Green, M. D.; Berman, J.; Dezube, M.; Howe, A.
S.; Brown, P. J.; Roth, J. T.; McGeehan, G. M. Anal. Biochem. 1993, 212,
58-64.
(31) Jarvis, R. A.; Patrick, E. A. A. IEEE Trans. Comput. 1973, C-22,
1025-1034.
(32) Denzoit, F.; Lang, R. J. Immunol. Methods 1986, 89, 271-277.
(33) Martin, E. J.; Critchlow, R. E.; Spellmeyer, D. C.; Rosenberg, S.;
Spear, K. L.; Blaney, J. M. Pharmacochem. Libr. 1998, 29, 133-146.

9650 J. Am. Chem. Soc., Vol. 122, No. 40, 2000
Chen et al.
The organics were combined, washed with water and brine, dried over
Na2SO4, and concentrated in vacuo to provide an oil that was purified
via column chromatography using hexane/ethyl acetate (1:1) followed
by CHCl3/MeOH (95:5) as eluant to yield 253 mg (58%) of the desired
product as a white solid. Mp 90-94 °C. 1H NMR (400 MHz, DMSO-
d6) δ 1.46 (s, 3 H), 2.25 (s, 3 H), 3.70 (q, J ) 5.72 Hz, 2 H), 4.23 (t,
J ) 5.74 Hz, 2 H), 4.62 (d, J ) 14.2 Hz, 1 H), 5.02 (d, J ) 14.2 Hz,
1 H), 7.05-7.80 (comp m, 16 H), 9.00 (t, J ) 5.64 Hz, 1 H), 12.70 (s,
1 H). High-resolution mass spectra (ESI) m/z 599.18424 [(M + H)+
C33H31N2O7S requires 599.18465].
points for the indirectly acquired dimensions to increase resolution.
Linear prediction by the means of the mirror image technique was used
only for constant-time experiments.40 In all cases data were processed
with a skewed sine-bell apodization function and one zero-filling was
used in all dimensions.
The general approach for determining the structure of CL-82198
bound to MMP-13 based on a previously solved NMR structure was
described in detail previously.26
The resonance assignments and bound conformation of CL-82198
in the MMP-13:CL-82198 complex were based on the 2D 12C/12C-
filtered NOESY,41,42 2D 12C/12C-filtered TOCSY,41,42 and 12C/12C-filtered
COSY experiments.43
Benzofuran-2-carboxylic acid (2-{4-[benzyl-(2-hydroxycarbam-
oyl-4,6-dimethylphenyl)sulfamoyl]phenoxy}ethyl)amide (WAY-
170523): Acid 13 (170 mg, 0.28 mmol) was dissolved in DMF (5 mL).
To this was added hydroxybenzotriazole (92 mg, 0.68 mmol) followed
by 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (150
mg, 0.78 mmol). The reaction was stirred for 1 h and then hydroxy-
lamine hydrochloride (156 mg, 2.24 mmol) and triethylamine (0.39
mL, 2.8 mmol) were added and the resulting mixture was stirred
overnight. The reaction was then diluted with CH2Cl2 and washed with
water, 1 M HCl, NaHCO3 (saturated), and brine, dried over Na2SO4,
and concentrated in vacuo to provide an oil that was triturated with
ether to provide the desired hydroxamic acid as a beige solid (58%
yield). IR (KBr) 3380, 1653, 1595, 1496, 1257, 1153, 1093, 853, 750,
1
The assignments of the H, 15N, and 13C resonances of MMP-13 in
the MMP-13:CL-82198 complex were based on the previous assign-
ments for the MMP-13:WAY-151693 complex34 in combination with
a minimal set of experiments: 2D 1H-15N HSQC, 3D 15N-edited
NOESY,44,45 CBCA(CO)NH,46 C(CO)NH,47 HNHA,48 and HNCA.49
The MMP13:CL-82198 structure is based on observed NOEs from
the 3D 15N-edited NOESY44,45 and 3D 13C-edited/12C-filtered NOE-
SY.50,48 The 3D 15N-edited NOESY and 3D 13C-edited/12C-filtered
NOESY experiments were collected with 100 and 110 ms mixing times,
respectively. The acquisition parameters for each of the experiments
used in determining the solution structure of the MMP-13:CL-82198
complex were as reported previously.26,51 As a result of the weak binding
affinity of CL-82198 with MMP-13 only a minimal number of
intermolecular NOEs were observed. The number of observed NOEs
was insufficient for a detailed simulated annealing approach for solving
the structure of the MMP-13:CL-82198 complex, but was sufficient
for validating a minimized docked model of the complex.
1
700 cm-1. H NMR (400 MHz, DMSO-d6) δ 1.54 (s, 3 H), 2.25 (s, 3
H), 3.70 (q, J ) 5.7 Hz, 2 H), 4.25 (t, J ) 5.8 Hz, 2 H), 4.73 (d, J )
2 Hz, 2 H), 6.98 (d, J ) 5.72 Hz, 2 H), 7.07-7.24 (comp m, 4 H),
7.41 (dt, J ) 7.4, 52 Hz, 2 H), 7.58 (d, J ) 0.68 Hz, 1 H), 7.66 (d, J
) 8.4 Hz, 1 H), 7.79 (m, 2 H), 8.83 (t, J ) 2.5 Hz, 1 H), 8.97 (t, J )
5.6 Hz, 1 H). High-resolution mass spectra (ESI) m/z 614.19532 [(M
+ H)+ C33H32N3O7S requires 614.19555].
Molecular Analysis and Design. The minimized models of CL-
82198 and WAY-152177 complexed to MMP-13 were prepared as
previously described.52,53 Using molecular dynamics methods (Sybyl
v6.4 from Tripos Inc), protein regions within 5 Å from CL-82198 were
sampled along with the inhibitor, whereas everything else remained
rigid during the simulations. Upon energy convergence, the last 50
frames from the final 100-ps run were averaged and this averaged
structure underwent a final minimization. The final MMP-13:CL-82198
model appeared to have optimized possible polar and van der Waals
interactions (Figure 4B). Validity of the final MMP-13:CL-82198 model
was determined by consistency with the experimental NOEs observed
for the complex. The identical procedure was applied to the complex
of MMP-13 and WAY-152177 (experimental structure not shown).
Since the two complexes used identical MMP-13 structures, the proteins
were overlapped to depict the positions of the two inhibitors within
the active site. Graphics analysis of the inhibitors showed that the
methylene carbon of CL-82198 containing the 2HB1/2 protons (Figure
2) overlapped identically with the methoxy carbon from WAY-152177.
This analysis indicated the optimal or minimal linkage length required
to connect the benzofuran moiety to the methoxy region of WAY-
152177. The final design scheme is shown in Figure 6 for the hybrid
NMR Sample Preparation. Uniformly (>95%) 15N- and 15N/13C-
labeled human recombinant MMP-13 was expressed in E. coli and
purified as described previously.9,34 1 mM 13C/15N- and 15N-MMP-13
NMR samples were prepared by concentration and buffer exchange
using Millipore Ultrafree-10 centrifugal filters into a buffer containing
10 mM deuterated Tris-base, 100 mM NaCl, 5 mM CaCl2, 0.1 mM
ZnCl2, 2 mM NaN3, and 10 mM deuterated DTT in 90% H2O/10%
D2O or 100% D2O. The 10:1 CL-82198:MMP-13 samples were
prepared by addition of CL-82198 into either a 1 mM 13C/15N- or 15N-
MMP-13 sample followed by pH readjustment. The sample to explore
the potential of competitive inhibition between CL-82198 and WAY-
151693 was prepared by first adding 1 mM WAY-151693 to a 1 mM
15N-MMP-13 sample followed by the addition of 10 mM CL-82198.
The initial MMP-13:WAY-151693 sample was made by buffer
exchange of 15N-MMP-13 into the buffer containing 0.1 mM WAY-
151693 followed by additional buffer exchanges to remove excess
WAY-151693. Finally, 10 mM of CL-82198 was added to the 1 mM
15N-MMP-13:WAY-151693 sample followed by pH readjustment.
NMR Data Collection. All spectra were recorded at 35 °C on a
Bruker AMX-2 600 spectrometer using a gradient enhanced triple-
resonance 1H/13C/15N probe. For spectra recorded in H2O, water
suppression was achieved with the WATERGATE sequence and water-
flip back pulses.35,36 Quadrature detection in the indirectly detected
dimensions was recorded with a States-TPPI hypercomplex phase
increment.37 Spectra were collected with appropriate refocusing delays
to allow for 0,0 or -90,180 phase correction.
(40) Zhu, G.; Bax, A. J. Magn. Reson. 1992, 100, 202-7.
(41) Petros, A. M.; Kawai, M.; Luly, J. R.; Fesik, S. W. FEBS Lett.
1992, 308, 309-14.
(42) Gemmecker, G.; Olejniczak, E. T.; Fesik, S. W. J. Magn. Reson.
1992, 96, 199-204.
(43) Ikura, M.; Bax, A. J. Am. Chem. Soc. 1992, 114, 2433-40.
(44) Marion, D.; Driscoll, P. C.; Kay, L. E.; Wingfield, P. T.; Bax, A.;
Gronenborn, A. M.; Clore, G. M. Biochemistry 1989, 28, 6150-6.
(45) Zuiderweg, E. R. P.; Fesik, S. W. Biochemistry 1989, 28, 2387-
91.
Spectra were processed using the NMRPipe software package38 and
analyzed with PIPP39 on a Sun Ultra10 Workstation. When appropriate,
data processing included a solvent filter, zero-padding data to a power
of 2, linear prediction back one data point of indirectly acquired data
to obtain zero phase corrections, and linear prediction of additional
(46) Grzesiek, S.; Bax, A. J. Am. Chem. Soc. 1992, 114, 6291-3.
(47) Grzesiek, S.; Anglister, J.; Bax, A. J. Magn. Reson., Ser. B 1993,
101, 114-19.
(34) Moy, F. J.; Chanda, P., K.; Cosmi, S.; Edris, W.; Levin, J. I.; Powers,
R. J. Biomol. NMR 2000, in press.
(48) Vuister, G. W.; Bax, A. J. Am. Chem. Soc. 1993, 115, 7772-7.
(49) Kay, L. E.; Ikura, M.; Tschudin, R.; Bax, A. J. Magn. Reson. 1990,
89, 496-514.
(35) Piotto, M.; Saudek, V.; Sklenar, V. J. Biomol. NMR 1992, 2, 661-
5.
(50) Lee, W.; Revington, M. J.; Arrowsmith, C.; Kay, L. E. FEBS Lett.
1994, 350, 87-90.
(36) Grzesiek, S.; Bax, A. J. Am. Chem. Soc. 1993, 115, 12593-4.
(37) Marion, D.; Ikura, M.; Tschudin, R.; Bax, A. J. Magn. Reson. 1989,
85, 393-9.
(51) Moy, F. J.; Seddon, A. P.; Boehlen, P.; Powers, R. Biochemistry
1996, 35, 13552-13561.
(38) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax,
A. J. Biomol. NMR 1995, 6, 277-293.
(52) Chen, J. M.; Sheldon, A.; Pincus, M. R. J. Biomol. Struct. Dyn.
1995, 12, 1129-1159.
(39) Garrett, D. S.; Powers, R.; Gronenborn, A. M.; Clore, G. M. J. Magn.
Reson. 1991, 95, 214-20.
(53) Chen, J. M.; Xu, S. L.; Wawrzak, Z.; Basarab, G. S.; Jordan, D. B.
Biochemistry 1998, in press.

Structure Based Design of SelectiVe MMP-13 Inhibitors
J. Am. Chem. Soc., Vol. 122, No. 40, 2000 9651
Figure 2. Illustration of the CL-82198 inhibitor with the corresponding
proton labels.
inhibitors but only 223 of the 385 hits were identified as viable
synthetic targets. The 223 hits were clustered based on structural
similarities and the properties of these compounds were
compared against the properties of the set of orally available
drugs. This profile analysis provides an initial means to predict
the likelihood that an HTS hit may have the characteristics of
an orally bioavailable drug. Using these criteria, the cluster
containing CL-82198 was identified for further analysis. The
normalized property profile for this cluster is shown in Figure
1A and the corresponding structures for the cluster are shown
in Figure 1B. It is clear from Figure 1 that none of the five
compounds in the cluster exhibit significant deviations in
physical properties relative to the set of orally available drugs.
Also, it is apparent that members of the cluster are close
structural analogues suggesting a structure-activity relationship.
These results identified this cluster as a possible source of a
potentially new series of MMP-13 inhibitors. CL-82198 (com-
pound 1) was shown to exhibit weak inhibition of MMP-13
(89% at the 10 µg/mL), but more intriguing was the observation
of a complete lack of activity against other MMPs (MMP-1,
MMP-9, and TACE). The profile of CL-82198 illustrated in
Figure 1A indicates that the compound has properties similar
to orally available drugs suggesting that it would be an ideal
candidate for optimization of its enzyme potency and selectivity.
The primary structure of CL-82198 along with the proton
naming convention is shown in Figure 2. A common feature of
known MMP inhibitor structures is the presence of a Zn-chelator
that plays a fundamental role in its activity. In most cases Zn
chelation occurs from the presence of a hydroxamic acid in the
structure of the small molecule. As apparent from the structure
of CL-82198 (Figure 2), the compound does not contain an
obvious substituent that would chelate Zn. Thus, the unique
structure of CL-82198 suggested a potential novel mechanism
for inhibition of MMP-13 further strengthening the choice of
CL-82198 as an initial lead candidate. Therefore, the identifica-
tion of CL-82198 as a candidate to optimize its activity and
selectivity was based on three unique observations: its intrinsic
MMP-13 selectivity, its structural profile similar to known
bioavailable drugs, and finally its apparent novel structure.
NMR Structure of the MMP-13:CL-82198 Complex. The
NMR binding studies provided critical information pertaining
to the mechanism of CL-82198 inhibition of MMP-13 and the
method for designing increased potency. The major question
presented when CL-82198 was identified from HTS was its
unknown MMP-13 binding site and its method for inducing
MMP-13 inhibition. Our previous work on the NMR structure
of MMP-13 complexed with WAY-151693 and MMP-1 com-
plexed with CGS-27023A provided the framework and meth-
odology to analyze CL-82198 bound to MMP-13.9,26
Figure 1. (A) A column graph depicting the normalized property values
for each structure in the cluster. For each compound the normalized
values of the six properties are shown by different colored columns:
total number of non-hydrogen atoms (blue); number of heteroatoms
(green); number of hydrogen-bond donors (red); number of hydrogen-
bond acceptors (yellow); calculated log P (purple); and molecular
weight (orange). The profiles for the five different compounds are
arrayed next to each other. Note that every compound in the cluster
deviates less than 1 standard deviation from the mean value for the
marketed drugs in respect to each of the six properties. The mean values
for the marketed drugs are as follows: total number of non-hydrogen
atoms (25); number of heteroatoms (6.9); number of hydrogen-bond
donors (1.8); number of hydrogen-bond acceptors (4.7); calculated log
P (2.2); and molecular weight (335). (B) The structures of the five
compounds comprising the cluster. The values in parentheses are the
percent inhibition values measured in the MMP-13 inhibition assay.
CL-82198 is labeled as compound 1 in the cluster.
inhibitor. The homology model of MMP-9 was constructed using the
COMPOSER program (Tripos INC, Sybyl v.6.4).
Results and Discussion
High Throughput Screening Analysis. CL-82198 was
identified as an initial lead from the analysis of the MMP-13
high throughput screen (HTS). A total of 385 compounds from
the 58079 compounds screened were identified as MMP-13
The CL-82198 MMP-13 binding site was initially identified
from chemical shift perturbation in the 1H-15N HSQC spectra.
The observed perturbations were mapped onto a GRASP surface
as depicted in Figure 3. It is apparent that the major effect of

9652 J. Am. Chem. Soc., Vol. 122, No. 40, 2000
Chen et al.
Figure 3. Two GRASP surface views for MMP-13 colored by the observed chemical shift perturbation induced by the binding of CL-82198: (A)
looking down the S1′ binding pocket and (B) the base of the S1′ binding pocket. Red corresponds to a negative chemical shift change and blue
corresponds to a positive chemical shift change.
CL-82198 on the chemical shifts of MMP-13 occurs in the
proximity of the S1′ pocket suggesting that CL-82198 sits in
this pocket. From the NMR and X-ray structures of MMP-13,
it was determined that the S1′ pocket for MMP-13 is very deep
and linear in shape while nearly reaching the surface of the
protein. In fact, a number of residues at the surface of MMP-
13 near the base of the S1′ pocket show significant chemical
shift perturbation in the presence of CL-82198. Since CL-82198
is a linear molecule, docking studies would place the inhibitor
stretched throughout the linear S1′ pocket of MMP-13. The only
question remaining was whether to place the morpholine or the
benzofuran moiety of CL-82198 at one end of the pocket,
adjacent to the catalytic zinc or at the opposite end, distant from
the zinc atom. Property analysis of the enzyme’s S1′ pocket
depicts that the end adjacent to the zinc is relatively polar
whereas the opposite end is hydrophobic. This analysis led us
to dock CL-82198 with the morpholine ring adjacent to the
catalytic zinc atom with the benzofuran moiety siting in a
hydrophobic pocket formed by L115, L136, F149, and P152 at
the base of the S1′ pocket. To further verify the proposed
binding of CL-82198 in the S1′ pocket of MMP-13, a simple
competition experiment with WAY-151693 was conducted. The
1H-15N HSQC experiment for the MMP-13:CL-82198 complex
was collected in the presence of WAY-151693. The presence
of WAY-151693 displaced all of CL-82198 as evident by the
distinct differences in the 1H-15N HSQC spectra, which further
suggests that both compounds bind in the S1′ pocket (data not
shown).
The relative orientation and binding of CL-82198 with MMP-
13 was further confirmed by the observation of intermolecular
NOEs between CL-82198 and MMP-13 from the 3D 13C-edited/
12C-filtered NOESY experiment. The NOESY spectra were
collected in the presence of a 10-fold excess of CL-82198
because of the weak affinity of CL-82198 with MMP-13.
Nevertheless, a total of 16 NOEs were observed between CL-
82198 and L81, L115, V116, Y141, T142, and Y143, which
support the initial positioning of CL-82198 in the MMP-13 S1′
Figure 4. (A) Expanded 2D plane from the 3D 13C-edited/12C-filtered
NOESY experiment corresponding to NOEs from L82 δ and L115 δ
to the labeled resonances from CL-82198. (B) Expanded view of the
NMR MMP-13:CL-82198 complex where the MMP-13 active site is
shown as a transparent surface with CL-82198 shown as liquorice
bonds. View is looking at the S1′ pocket.
pocket. An expanded 2D plane from the 3D 13C-edited/12C-
filtered NOESY experiment is shown in Figure 4A, which
demonstrates examples of some key intermolecular NOEs
between CL-82198 benzofuran group resonances and L115 δ
and CL-82198 resonances proximal to the morpholine ring and
L82 δ. The complex of CL-82198 with MMP-13 was subjected
to energy refinement using the NMR results as constraints.26,52
An expanded view of the refined NMR structure of MMP-13
complexed with CL-82198 is shown in Figure 4B. The modeling
results depict the morpholine oxygen forming a hydrogen bond

Structure Based Design of SelectiVe MMP-13 Inhibitors
Table 1. IC50 and Selectivity Data
J. Am. Chem. Soc., Vol. 122, No. 40, 2000 9653
compd
MMP-1
MMP -9
MMP -13
TACE
S-1a
S-9a
S-TACEa
WAY-152177
WAY-159062
WAY-159063
WAY-170523
82 nM
21 nM
46 nM
71 nM
945 nM
15 nM
75 nM
301 nM
17 nM
240 nM
5.5×
1.4×
0.6×
0.2×
56×
16×
750 nM
470 nM
10.0×
3.4×
6.3×
2.2×
>500×
1025 nM
NA (10 µM)
664 nM
19% (1 µM)
>5800×
a Selectivity data presented as a ratio of the MMP or TACE IC50 with MMP-13. NA, no activity observed at the indicated concentration.
Table 2. Sequence Comparison
positiona
MMP-1
MMP-9
MMP-13
115
144
ARG
VAL
LEU
ARG
LEU
VAL
a The residue numbering corresponds to the sequence for the MMP-
13 structure.
with the backbone amide group of Leu-82 and the benzofuran
group packs deep in the S1′ pocket with the peptide bond linker
portion forming hydrogen bonds with protein backbone groups.
The complex shows no apparent interactions between the
inhibitor and the catalytic zinc, justifying the ligands micromolar
potency.
Structures of MMP-1, MMP-9, and MMP-13. The recent
NMR solution structures of MMP-1 and MMP-13 were used
as starting points for molecular modeling and analysis.9,25,26 At
the start of this study, the experimental structure of MMP-9
was not available, therefore, a model was developed based on
its strong homology to MMP-1 (54% identity around the
catalytic domain). Overall, the catalytic site of MMP-9 is similar
to the corresponding sites in MMP-1 and MMP-13. All three
structures were used as starting points for analysis and synthetic
design.
Figure 5. Design scheme dividing WAY-159062 into two components
corresponding to its potency component and its selectivity component.
Basis for the design of a hybrid with CL-82198.
other potent inhibitors. WAY-159062 is an example of a potent
and selective inhibitor for MMP-9 and MMP-13 (see Table 1).
Structurally similar inhibitors were positioned into the active
site of MMP-13 based on the NMR solution structure of MMP-
13 complexed with WAY-1516939 and MMP-1 complexed with
CGS-27023A.26
Figure 5 shows the critical regions of WAY-159062, which
can be broken down into two components, WAY-152177, which
represents the zinc chelating portion of the compound that
contributes to the binding potency, and the toluene group (1A),
which contributes to enhanced ligand selectivity against MMP-
1. The strategy was to design a new inhibitor based on replacing
the toluene group (1A) with a component of CL-82198 critical
for binding within the extended S1′ pocket of MMP-13. The
overlay of the NMR solution structure for CL-82198 with the
model for WAY-152177 is shown in Figure 6B. The close
similarity between the positioning of the two structures made
it readily apparent that it would be possible to generate a hybrid
of the two structures combining the potent WAY-152177 with
the selective component of CL-82198 (Figure 6). These results
were then used to design the proposed hybrid inhibitor WAY-
170523 (see Figures 5 and 6). The synthesis of the inhibitor
was accomplished in a straightforward manner beginning with
the known54 acid (12) (Scheme 1). Fluoride displacement with
the sodium salt of benzofuran alcohol (11) (prepared via
condensation of benzofuran carboxylic acid with ethanolamine)
provided the elaborated acid (13). Condensation of the acid with
hydroxylamine hydrochloride mediated by hydroxybenzotriazole
and 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochlo-
ride then produced the desired hydroxamic acid. When tested
against several members of the MMP family, the data in Table
1 clearly show that the new inhibitor, WAY-170523, has better
potency compared to WAY-159062 in addition to improved
selectivity toward MMP-13. With a proof of concept now
demonstrated for establishing selectivity versus MMP-9, this
hybrid inhibitor now provides a starting point for the design of
potent, selective, orally bioavailable inhibitors of MMP-13.
Sulfonamide MMP inhibitors similar to this inhibitor have
Comparative analysis of the MMP structures shows that
residue positions 115 and 144 (based on the MMP-13 sequence
numbering), in addition to the length of the specificity loop,
determine the size and shape of the S1′ pockets. Alignment of
the NMR structures for MMP-1 and MMP-13 shows that MMP-
13 contains two additional insertions in the specificity loop. The
homology model of MMP-9 indicates no additional insertions
so its length is identical to that of MMP-1.
Residue positions 115 and 144 are important in establishing
the relative length of the S1′ pockets for the MMPs where the
larger side chain at these positions results in a smaller S1′
pocket. Since residue 115 is spatially closer to the catalytic zinc
than residue 144, a larger side chain for residue 115 will have
a greater impact on defining a smaller S1′ pocket compared to
residue 144. As can be seen in Table 2, MMP-1 has the largest
side chain at position 115, thus its S1′ pocket is the smallest.
MMP-9 has an Arg at position 144 resulting in its S1′ pocket
being longer compared to MMP-1. Conversely, MMP-13 has
short side chains at both positions 115 and 144. The short side
chains combined with an increased length of its specificity loop
result in MMP-13 having the largest S1′ pocket. To summarize,
the sizes of the MMP S1′ pockets are as follows: MMP-13 >
MMP-9 > MMP-1 where this structural feature plays a critical
role in the design strategy for developing a potent and specific
MMP-13 inhibitor.
Design Strategy. A strategy utilizing NMR and molecular
modeling was applied toward the design and synthesis of an
MMP-13 selective inhibitor lead. The basic approach behind
the design strategy is to optimize the affinity of the chemical
lead CL-82198 while maintaining its inherent MMP-13 selectiv-
ity. This can be achieved by taking advantage of the distinct
structural feature of MMP-13, its deep linear S1′ pocket, while
combining overlapping structural features of CL-82198 with
(54) Levin, J. I.; Du, M. T.; Venkatesan, A. M.; Nelson, F. C.; Zask,
A.; Gu, Y.: United States patent number 5,929,097, 1999.

9654 J. Am. Chem. Soc., Vol. 122, No. 40, 2000
Chen et al.
Figure 6. (A) Design scheme showing the flow from CL-82198 and WAY-159062 to WAY-170523. (B) Expanded view of the NMR MMP-13:
CL-82198 complex overlayed with the MMP-13:WAY-152177 model demonstrating the approach to forming the hybrid inhibitor WAY-170523
where the MMP-13 active site is shown as a grid surface with CL-82198 and WAY-152177 is shown as liquorice bonds. View is looking at the
S1′ pocket.
Scheme 1
a potent and selective inhibitor for MMP-13. Specifically, these
data identified the potential utility of the size, shape and nature
of the S1′ pocket in designing specificity into MMP inhibitors.
Additionally, the abundance of information available on the
nature of MMP inhibitors indicated general characteristics that
were necessary for inhibitor potency. The NMR structures for
MMP-13 and MMP-1 were the initial starting points for a
structure-based approach for developing a selective MMP-13
inhibitor. In addition, the structural information for the WAY-
151693 and CGS-27023A complexes was critical for adapting
design elements into the initial CL-82198 structure, since these
complex structures reliably predicted the bound structures of
new analogues. Thus, the combination of NMR spectroscopy
with molecular modeling techniques and HTS data resulted in
the design of a novel, potent, and selective MMP-13 inhibitor
(WAY-170523) which has an IC50 of 17 nM for MMP-13 and
showed >5800-, 56-, and >500-fold selectivity against MMP-
1, MMP-9, and TACE, respectively. To the best of our
knowledge, this represents the first example of a potent MMP-
13 inhibitor that has been shown to be selective against MMP-
9.
shown enhanced oral potency when basic amines have been
incorporated into the molecule.55 Further studies are underway
to reach the combined target of selectivity with oral activity.
Acknowledgment. The authors would like to thank Pranab
K. Chanda, Scott Cosmi, Wade Edris, and Jim Wilhelm for
providing the MMP samples, Mei-Li Sung, Rebecca Cowling,
and Loran M. Killar for providing the IC50 data, Linda Heydt
and Fernando Ramirez for providing the high throughput
screening results, and Kathryn Bracken and Erin Mattingly for
chemical synthesis.
Conclusion
The extensive structural information available for the MMP
family of proteins provided the essential elements for designing
(55) MacPherson, L. J.; Bayburt, E. K.; Capparelli, M. P.; Carroll, B.
J.; Goldstein, R.; Justice, M. R.; Zhu, L.; Hu, S.-I.; Melton, R. A.; Fryer,
L.; Goldberg, R. L.; Doughty, J. R.; Spirito, S.; Blancuzzi, V.; Wilson, D.;
O’Byrne, E. M.; Ganu, V.; Parker, D. T. J. Med. Chem. 1997, 40, 2525-
2532.
JA001547G