Structure-based design and synthesis of novel non-zinc chelating MMP-12 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2005.05.079

A new class of MMP-12 inhibitors was discovered and optimized using structure-based drug design methods. Modeling studies using a known MMP-12 crystal structure identified a new interaction mode for these new MMP-12 inhibitors. Further optimization resulted in the discovery of a compound displaying nanomolar activity against MMP-12 and which was co-crystallized with MMP-12.

Full text of DOI:10.1016/j.bmcl.2005.05.079

Bioorganic & Medicinal Chemistry Letters 15 (2005) 3787–3790
Structure-based design and synthesis of novel non-zinc
chelating MMP-12 inhibitors
Anne-Claude Dublanchet,* Pierre Ducrot, Charles Andrianjara, Margaret OÕGara,
`
´
Renaud Morales, Delphine Compere, Alexis Denis, Stephane Blais, Philippe Cluzeau,
´
Karine Courte, Jacques Hamon, Franc¸ois Moreau, Marie-Laure Prunet and Anita Tertre
Departments of Chemistry and Research Technologies, Pfizer Global Research and Development, Fresnes Laboratories,
3-9 rue de la Loge, BP 100, F-94265 Fresnes Cedex, France
Received 24 March 2004; revised 17 November 2004; accepted 12 May 2005
Available online 5 July 2005
Abstract—A new class of MMP-12 inhibitors was discovered and optimized using structure-based drug design methods. Modeling
studies using a known MMP-12 crystal structure identified a new interaction mode for these new MMP-12 inhibitors. Further opti-
mization resulted in the discovery of a compound displaying nanomolar activity against MMP-12 and which was co-crystallized
with MMP-12.
Ó 2005 Elsevier Ltd. All rights reserved.
Matrix metalloproteinases (MMPs) are a family of zinc
dependent endoproteinases, involved in the turnover
and remodeling of an extracellular matrix and therefore
are therapeutic targets for the treatment of many inflam-
matory, malignant, and degenerative diseases.1 For
example, inhibitors of MMP-12 are sought for the pre-
vention of chronic obstructive pulmonary disease
(COPD) development.2 The vast majority of reported
MMP inhibitors can be viewed as zinc chelating
compounds either via a hydroxamate moiety or, less
effectively, via other groups, including reverse hydroxa-
mate, thiol, carboxylate, phosphonic acid, and phosphi-
nate functions. Indeed, X-ray crystal structures of
MMPs with bound inhibitors reveal that inhibitor bind-
ing interactions typically include coordination of the ac-
tive site zinc by a ligand (e.g., hydroxamic acid) and
occupancy of the S10 pocket with a hydrophobic group.
However, in the development of new therapeutic agents
targeting MMPs, there remains considerable room for
new structural classes not coordinating the active site
zinc atom.3 Moreover, exploiting the structural differ-
ences between the MMP subtypes4 (Table 1), currently
an important issue in this area, should lead to more spe-
Table 1. Comparison between various MMP amino acid sequences
close to the inhibitor putative binding site
MMP-12 I180 L214 T215 V235 T239 K241 V243
MMP-1
MMP-2
MMP-3
MMP-7
MMP-8
MMP-9
MMP-13
MMP-14
N
L
V
T
I
R
L
L
Y
L
L
L
L
V
V
V
A
V
V
V
V
L
L
L
V
L
L
L
I
S
T
S
I
T
T
L
G
R
T
T
M
L
T
N
M
I
H
G
A
R
T
L
L
F
F
Q
cific inhibitors, postulated to avoid the toxic side effects
supposedly linked to broad spectrum inhibitors. We
now describe the discovery of a non-zinc chelating com-
pound class optimized for its MMP-12 activity using a
structure-based design process.
To identify non-competitive or uncompetitive inhibitors
from those that are competitive, the MMP-12 high-
throughput screen was conducted in the presence of
5 mM acetohydroxamic acid (AH).5 Compound 1
(Table 2) was thereby identified as an initial lead.
Compound 1 alone exhibited weak inhibition of
MMP-12 (IC50 = 13 lM) but proved to be much more
potent (IC50 = 0.23 lM) in the presence of AH and
showed a complete lack of activity against MMP-1, -3,
and -9. The data measured in the presence of AH
Keywords: Matrix metalloprotease 12; Human metalloelastase; Non-
zinc chelating inhibitors.
*
Corresponding author. Tel.: +33 1 58 31 70 61; fax: +33 1 48 68 96
96; e-mail: anne-claude.dublanchet@wanadoo.fr
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.05.079

3788
A.-C. Dublanchet et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3787–3790
Table 2. Compound 1, initial hit identified from a mass screen of
MMP-12 catalytic domain in the presence of 5 mM acetohydroxamate
(AH)
the docking experiments.7d The neutral state of the
thiophene moiety creates a hydrophobic field in its vicin-
ity. As a result, the thiophene group makes no interac-
tion other than hydrophobic contacts with His218
(lower side) and Leu181 (upper side), and does not
chelate the zinc atom, being at a distance of more than
Enzyme
IC50 (lM)
Enzyme
IC50 (lM)
MMP-12
13
MMP-2
MMP-3
MMP-8
MMP-9
26
MMP-12 + 5 mM AH
MMP-13
MMP-1
0.23
24
>101
25
>101
˚
4 A. In summary, the thiophene ring constitutes a good
linker in optimizing both length and orientation of the
compound. This newly discovered interaction mode of
MMP-12 inhibitors has motivated us to choose com-
pound 1 as an initial lead candidate for optimization.
The docked complex suggests that potency could be im-
proved by optimizing interactions with specific residues
lying in the S10 pocket, that is, through hydrogen bond-
ing either to Lys241 backbone NH or carbonyl oxygen,
as well as hydrogen bonding to Thr239 carbonyl oxygen
or side- chain hydroxyl group. Furthermore, targeting
Lys241 and Thr239 should improve selectivity versus
MMP-2,3,9,13, and 14. Moreover, previous studies4
have shown that MMP selectivity for zinc binding inhib-
itors can be achieved by optimizing the size of the
moiety located in the S10 pocket, which is differently
shaped for each MMP subtype.
>101
suggest that compound 1 achieves a synergistic inhibi-
tion of MMP-12 with AH and thus binds to the protein
in a different way compared to AH. Moreover, com-
pound 1 does not contain an obvious substituent that
would chelate the zinc atom. As part of a structure-
based design program, modeling studies were initiated
using the crystal structure of the inactive E219A mutant
MMP-12 determined by Lang et al.6 We replaced
Ala219 by a Glu having the side-chain conformation ob-
served for Glu202 in MMP-13 (pdb code 830c). Docking
experiments were conducted with AGdock (proprietary
software).7a For each of the suggested binding modes,
compound 1 was involved in hydrophobic interactions
with the S10 pocket through the phenyl moiety, more
precisely with Tyr240 methylene groups. Additionally,
the amide moiety forms two hydrogen bonds with the
backbone NH of Leu181 and the backbone carbonyl
of Pro238. Several binding modes were observed from
the docking experiments for the ethyl-morpholine
moiety. The preferred orientation is stabilized by
hydrophobic contacts between the morpholino methylene
groups and the p-electron of the backbone Gly179-Ile180
peptidic bond (Fig. 1). This type of interaction, that is,
interaction between an aliphatic group and p-electrons,
has already been observed in crystal structure data.7b
On the basis of these results, our main strategy was to
optimize the occupancy of the S10 pocket and hence
maximize hydrophobic interactions. On the basis of
docking studies of compounds extended on the phenyl
ring side with hydrophobic substituents, analogues rep-
resented by structures 2–5 have been designed and syn-
thesized. As illustrated in Table 3, compounds 2–5
were found to be much more potent than the initial hit.
In parallel with the affinity measurements, we also exam-
ined the physico-chemical properties of our inhibitors.
Not surprisingly, compounds 2–5 of Table 3 exhibit
poor solubility and high variability of the microsomal
metabolic stability, probably related to the solubility
issues. A reason accounting for the poor solubility
observed in a majority of compounds synthesized to ex-
plore the SAR lies in their high lipophilicity, due to their
design to fit in the hydrophobic S10 pocket. More gener-
ally, we observed a poor microsomal metabolic stability
on a large number of compounds bearing an aminoeth-
ylmorpholino group, probably related to the hydrolysis
of the amide bond by liver enzymes. To address this is-
sue, a variety of substituents have been investigated for
the replacement of this amide substituent, which is open
to the solvent according to our docking studies. Table 3
provides a couple of examples of the groups selected
with compounds 6 and 7. A common feature is the pres-
ence of a carboxylate group, which probably helps in
maintaining some solubility via solvent interactions for
these molecules, which are mainly hydrophobic.
The overall orientation of the compound clearly match-
es the preferred region for hydrophobic contacts, as cal-
culated by MOE7c contact statistics (green region on
Fig. 1). Noteworthy is that the His218 side chain, which
constitutes the zinc finger structure, is kept neutral for
Since the majority of compounds displaying the best
affinity were biphenyl thienyl derivatives, with a linear,
almost planar, shape, presumably favoring efficient crys-
tal packing, compounds were designed to disrupt the
planarity of the molecule. Compound 9 is one typical
example of this design where we tried to break the
planarity. The affinity and solubility data show that
compound 9 displays a better solubility than the
Figure 1. Overlap of compound 1 docked into the active site of MMP-
12 cat domain and hydrophobic (green) and polar (red) preferred
regions—MOE contact statistics.

A.-C. Dublanchet et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3787–3790
Table 3. In vitro profile of compounds 2–11
3789
Compound
R1
R2
IC50, (lM)
MMP-12
IC50, (lM)
MMP-13
Sol (ug/ml)a
t1/2 (min)b
2
Phenyl
4-Pyridyl
Cyclohexyl
2-Morpholin-4-yl-ethyl
2-Morpholin-4-yl-ethyl
0.4
0.1
<1
304
112
3
3
<1
4
2-Morpholin-4-yl-ethyl
2-Morpholin-4-yl-ethyl
0.46
0.057
0.4
37
54
5
4-Acetylphenyl
Trifluoromethoxy
Trifluoromethoxy
Phenyl
<1
20
6
trans-4-Methylcyclohexane-carboxylic acid
4-Methylphenylacetic acid
4-Methylphenylacetic acid
248
7
0.4
100
<1
NMc
NCd
NMc
NMc
8
0.07
0.09
0.18
8.3
>1.2
2.5
9
2-Hydroxyphenyl
Trifluoromethoxy
Trifluoromethoxy
4-Pyridyl
4-Methylphenylacetic acid
134
>500
>500
>500
10a
10b
11
2-Carboxy-1-phenyl-ethyl, R enantiomer
2-Carboxy-1-phenyl-ethyl, S enantiomer
2-Carboxy-1-phenyl-ethyl, R enantiomer
2.2
0.014
0.27
99
a Thermodynamic solubility measured at pH 7.4.
b Metabolic stability in human hepatic microsomes.
c Non-metabolized.
d Not characterized because of solubility issues.
non-substituted biphenyl compound 8 and remains as
potent as compound 8.
Lastly, this (2R)-amino(phenyl)propanoic acid was com-
bined with one of the active hydrophobic phenyl-pyridyl
substituents to generate compound 11, which is soluble
and displays good affinity for MMP-12 (IC50 = 14 nM)
and, to a less extent, for MMP-13 (IC50 = 270 nM).
Eventually, the binding mode of compound 11 was
confirmed by an X-ray crystal structure of MMP-12
catalytic domain complexed with compound 11 with a
Motivated by these results, we decided to explore the
amide side chain, supposedly exposed to the solvent.
We planned to carryout interactions with the protein
residues in the vicinity of this moiety, although interac-
tions with this region of the protein were extremily
difficult to predict. Targeted residues were Ile180 for
hydrophobic interactions and Thr239 to form an accep-
tor or donor hydrogen bond. Several molecules were
designed in that manner and compound 10 is an exam-
ple where we have been able to predict which one of
the enantiomers was the most active one.
8
˚
resolution of 2.5 A (Fig. 3).
The data clearly show that the biaryl moiety fits into the
S10 pocket of the enzyme, where binding interactions are
mainly hydrophobic. Indeed, the core structures of thio-
phene and phenyl aromatic rings make hydrophobic
contacts with Thr215, Tyr240 methylene, and His218.
Similarly, lipophilic interactions take place observed be-
tween the 4-pyridyl group and Lys241 c- and e-methyl-
ene groups on the upper side, as well as Val235 on the
lower side. Additionally, key hydrogen bonds exist be-
tween the carbonyl oxygen and the backbone NH of
Leu181 and NH of Ala182, as well as between one
carboxylate oxygen atom and the backbone NH of
Tyr240, whereas the other carboxylate oxygen atom
binds to Gly179 carbonyl oxygen via a water molecule.
Docking studies have shown that the R isomer 10a
would be stabilized by hydrophobic contacts between
its phenyl ring and Ile180, which placed the carboxylate
group in the most hydrophilic region (Fig. 2), whereas
the S enantiomer would have its phenyl ring in the
hydrophilic region and its carboxylate anion in the
vinicity of the hydrophobic Ile180. As expected, on syn-
thesis and testing of both enantiomers, the R compound
proved to be the most active one.
Figure 2. Compound 10a docked into the active site of MMP-12 cat
domain. Hydrophobic regions are colored in brown, while polar
surface areas are colored in blue.
Figure 3. X-ray structure of a MMP-12 catalytic domain complexed
with compound 11.

3790
A.-C. Dublanchet et al. / Bioorg. Med. Chem. Lett. 15 (2005) 3787–3790
Synthesis of inhibitor 1 and related analogues is based
on the versatile palladium-catalyzed coupling chemistry.
Compounds 1–4 were synthesized, as shown in Scheme
1. Briefly, an oxidation and amide formation sequence
yielded the common intermediate A, which was either
directly coupled under Suzuki conditions with boronic
acids to give products 1 and 2 or transformed into a tri-
flate to yield final product 3 or eventually derivatized
into a stannane to undergo a Stille coupling to yield 4.
Compound 5 was prepared in the same manner using
4-(4-bromophenyl)-N-(2-morpholin-4-ylethyl)thiophene-
2-carboxamide intermediate that was obtained through
a Sandmeyer sequence. Compounds 6, 7, and 10 were
prepared from a common intermediate C, derived from
the starting aldehyde following a Suzuki sequence,
through amide bond formation with readily available
amines as shown in Scheme 2. Eventually, preparation
of compound 11 is achieved following the chemical
route shown in Scheme 3.
Scheme 3. Preparation of compound 11. Reagents and conditions:
(a) pyridylboronic acid, Pd(PPh3)4, K3PO4 aq 2 M, DME, 80 °C;
(b) MeOH, HCl, 65 °C; (c) lithium hydroxide, EtOH/H2O; (d) (2R)-
amino(phenyl) propanoate, HATU, DIEA, DMF, rt; (e) lithium
hydroxide, EtOH/H2O and then HCl 2 M.
act with the zinc active-site atom and nevertheless dis-
play affinities for MMP-12 in the nanomolar range.
X-ray crystallography data have confirmed the unique
binding mode predicted from docking experiments. Is-
sued from an activity-based and property-based optimi-
zation program, compounds such as compound 11 may
be regarded as novel, non-peptidic, low molecular
weight, non-zinc binding leads for continued develop-
ment of MMP inhibitors as drugs. On the basis of our
crystal structure and molecular modeling model, this
compound could be viewed as the starting point for
selectivity optimization over MMP-13 and physico-
chemical profile enhancement.
In summary, we have discovered a series of potent
MMP-12 inhibitors incorporating a thiophene template,
an unusual central linker compared to established MMP
binding groups. Structure-based drug design methods
led to compounds with an improved potency for
MMP-12, achieving a 1000-fold enhancement compared
to the initial hit. In contrast to the well-established
MMP inhibitors, the compounds described do not inter-
Acknowledgments
We thank Dr. Ajith Manage (Evotec-OAI) for the prep-
aration of synthetic intermediates in large scale; I. Me-
nier, N. Philippon, and M. Millet for solubility
measurements; and Dr. Annah Mancy and Anne Mou-
´
stie for determining metabolic stability values.
References and notes
Scheme 1. Preparation of compounds 1–4. Reagents and conditions:
(a) AgNO3, EtOH/NaOH aq, 40 °C; (b) (COCl)2, CH2Cl2, DMF cat.,
0 °C then rt; (c) 4-(2-aminoethyl)morpholine, Et3N, CH2Cl2, rt; (d)
R1boronic acid, Pd(PPh3)4, K3PO4 aq 2 M, DME, 80 °C; (f) i—
4-hydroxyphenylboronic acid, Suzuki cond.; ii—trifluoromethanesulf-
onic anhydride, CH2Cl2, pyridine, rt; iii—pyridine-4-boronic acid,
Suzuki cond.; (g) i—hexamethylditin, Pd(PPh3)4, DME, 80 °C; ii—
1-bromo-4-cyclohexylbenzene, LiCl, CuBr2, Pd(PPh3)4, dioxane, 80 °C.
1. (a) Whittaker, M., et al. Chem. Rev. 1999, 99, 2735; (b)
Beckett, R. P., et al. Expert Opin. Ther. Patents 1998, 8,
259.
2. (a) Belvisi, M. G., et al. Inflamm. Res. 2003, 52, 95; (b)
Macrophage Metalloelastase Inhibitors. In New Drugs for
Asthma, Allergy and COPD; Hansel, T. T., Barnes, P. J.,
Eds.; Prog. Resp. Res. Karger: Basel, 2001; Vol 31, pp 177–
180; (c) Hautamaki, R. D., et al. Science 1997, 277, 2002.
3. Chen, J. M., et al. J. Am. Chem. Soc. 2000, 122, 9648.
4. Terp, G. E., et al. J. Med. Chem. 2002, 45, 2675.
5. Dyer, R. D., et al. European Patent Application,
EP1291439A2, 2003.
6. Lang, R., et al. J. Mol. Biol. 2001, 312, 731.
7. (a) AGDOCK, QuikVu, PLCALC, CVT, CVT_PRO and
ADJUSTH, Pfizer private software; (b) Matsumoto, A.,
et al. J. Am. Chem. Soc. 2002, 124, 8891; (c) MOE Version
2002.03, software available from Chemical Computing
Group Inc.; (d) El Yazal, J.; Pang, Y.-P. J. Phys. Chem.
B 1999, 103, 8773.
Scheme 2. Preparation of compounds 6, 7, and 10. Reagents and
conditions: (a) (COCl)2, CH2Cl2, DMF cat., 0 °C then rt; (b) R2NH2,
Et3N, CH2Cl2, rt; (c) lithium hydroxide, EtOH/H2O.
8. Morales, R., et al. J. Mol. Biol. 2004, 341, 1063.