C O M M U N I C A T I O N S
in the case of a larger substituent (Table 1, fragment 1), broad-spectrum
activity is observed. This is likely due to the strong interaction of the
biphenyl substituent with the deep S1′ pocket of these MMPs. For BISL-
1, a greater preference for MMP-2 (Table 1, fragments 42 and 43) is
obtained. This is best illustrated by the p-trifluoromethylphenyl derivatives
(Table 1, fragments 3 and 43), where both chelating sulfonamides inhibit
MMP-2 over MMP-3 and -9, but the benzimidazole compound (Table 1,
fragment 43) shows reduced activity against MMP-8. This fragment is
uniquely potent against MMP-2 when compared with the other MMPs
tested and is particularly notable for discriminating between the gelatinases
(MMP-2 and -9).11 This finding clearly demonstrates that the core scaffold
has a significant effect on the selectivity of these lead compounds.
mol higher in energy than the conformation observed with MMP-
2. Again, these calculations are consistent with the experimental
data showing that fragment 43 is selective for MMP-2. The best
poses of fragments 3 and 43 in the active site of MMP-2 are shown
in Figure 2. Interestingly, both fragments have the sulfonamide
moiety pointed toward the unprimed side of the active site.11 In
spite of the intrinsic limitations of classical approaches to accurately
describe systems involving metal ions, the docking analysis was
otherwise wholly consistent with the experimental findings sum-
marized (Table 1) and provides some initial insight into the binding
of these fragments. Because the electrostatic component is the most
important contribution to the calculated ∆Eele+Vdw, our findings also
suggest that, despite the homology among MMPs, differences in
the electrostatic environment around the Zn(II) ion (Figure S5) may
have a significant effect on inhibitor-receptor recognition.
Table 1. IC50 Values (µM) of Select Fragments against Four MMPs
Figure 2. Model structure of the complex formed between MMP-2 (gray
surface, zinc shown as orange sphere) and fragments 3 (left) and 43 (right).
In summary, we have used known scaffolds from fluorescent
Zn(II) sensors as ZBGs in the design of two focused fragment
libraries. Most of the fragments exhibited a preference for MMP-
2, which was generally consistent with computational analysis. One
fragment (Table 1, fragment 43) shows low micromolar activity
against MMP-2 and no significant activity against MMP-3, -8, and
-9. Ongoing efforts to elaborate these new scaffolds are expected
to produce more potent, semiselective MMPi.
Evidence that these chelating sulfonamide fragments were binding
to the active site Zn(II) ion was provided by three control compounds
with structural similarities to fragment 3. Each of these control
compounds (Scheme 1, 78-80) has a reduced metal-binding capacity
due to either (i) lacking the two nitrogen donor atoms required to
achieve chelation (79) or (ii) having an insufficiently acidic N-H group
(78, 80) such that metal binding cannot compete with ligand proton-
ation. At a concentration of 500 µM all three control compounds
showed <30% inhibition of MMP-2 (>200-fold loss of activity vs 3),
indicating that the QSL-1 and BISL-1 libraries inhibit MMPs by Zn(II)
ion coordination. In addition, extended X-ray adsorption fine structure
(EXAFS) spectroscopy of fragment 3 with an MMP indicate binding
to the active site Zn(II) ion (Figure S6), providing direct evidence that
these fragments bind to the MMP metal ion.
Acknowledgment. This work was supported by the National
Science Foundation (instrumentation grants CHE-9709183, CHE-
0116662, and CHE-0741968), the American Heart Association
(0970028N), and the NIH (R01 HL080049; R21 HL094571). J.A.M.
is supported in part by NSF, NIH, HHMI, CTBP, and NBCR.
Supporting Information Available: Experimental details, Figures
S1-S6, and Tables S1-S2. This material is available free of charge
References
To better understand the activity of these fragments, computational
docking studies were performed on fragments 3 and 43 (Table 1, fragments
3 and 43) in MMP-2, -3, -8, and -9 using the Glide software package
(Schro¨dinger, Supporting Information). The 100 best poses against each
MMP were filtered based on distance/geometry between the ZBG and
the active site Zn(II) ion (Supporting Information).3 This filtering procedure
ensured that only structures with geometric parameters that produce
reasonable metal chelation are evaluated.
(1) Congreve, M.; Chessari, G.; Tisi, D.; Woodhead, A. J. J. Med. Chem. 2008,
51, 3661–3680.
(2) McRae, R.; Bagchi, P.; Sumalekshmy, S.; Fahrni, C. J. Chem. ReV. 2009,
109, 4780–4827.
(3) Fahrni, C. J.; O’Halloran, T. V. J. Am. Chem. Soc. 1999, 121, 11448–
11458.
(4) Fahrni, C. J.; Henary, M. M.; VanDerveer, D. G. J. Phys. Chem. A 2002,
106, 7655–7663.
(5) Henary, M. M.; Wu, Y. G.; Fahrni, C. J. Chem.sEur. J. 2004, 10, 3015–3025.
(6) Hendrickson, K. M.; Geue, J. P.; Wyness, O.; Lincoln, S. F.; Ward, A. D.
J. Am. Chem. Soc. 2003, 125, 3889–3895.
(7) Chai, S. C.; Ye, Q.-Z. Bioorg. Med. Chem. Lett. 2009, 19, 6862–6864.
(8) Coussens, L. M.; Fingleton, B.; Matrisian, L. M. Science 2002, 295, 2387–
2392.
(9) Jacobsen, F. E.; Lewis, J. A.; Cohen, S. M. ChemMedChem 2007, 2, 152–171.
(10) Agrawal, A.; Romero-Perez, D.; Jacobsen, J. A.; Villarreal, F. J.; Cohen,
S. M. ChemMedChem 2008, 3, 812–820.
(11) Whittaker, M.; Floyd, C. D.; Brown, P.; Gearing, A. J. H. Chem. ReV.
1999, 99, 2735–2776, and references therein.
Fragment 3 generated acceptable poses for all four MMPs;
however, the lowest energy poses for MMP-3 and -9 were >3.5
kcal/mol higher (∆Eele+Vdw, Table S1) than those for MMP-2 and
-8, consistent with the selectivity for this fragment for the latter
MMPs (Table 1). Acceptable poses for fragment 43 were readily
obtained with MMP-2, but not with the other MMPs examined.
Structures of fragment 43 in MMP-3, -8, and -9 were all >3.4 kcal/
JA101088J
9
J. AM. CHEM. SOC. VOL. 132, NO. 24, 2010 8233