Identification of matrix metalloproteinase inhibitors by chemical arrays

Abstract

Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases that degrade many extracellular matrix components and that have been implicated in the pathogenesis of various human diseases including cancer metastasis. Here, we screened MMP-9 inhibitors using photo-cross-linked chemical arrays, which can detect small-molecule ligand–protein interactions on a chip in a high-throughput manner. The array slides were probed sequentially with His-MMP-9, anti-His antibody, and a Cy5-labeled secondary antibody and then scanned with a microarray scanner. We obtained 27 hits among 24,275 compounds from the NPDepo library; 2 of the identified compounds (isoxazole compound 1 and naphthofluorescein) inhibited MMP-9 enzyme activity in vitro. We further explored 17 analogs of 1 and found that compound 18 had the strongest inhibitory activity. Compound 18 also inhibited other MMPs, including MMP-2, MMP-12, and MMP-13 and significantly inhibited cell migration in human fibrosarcoma HT1080 cells. These results suggest that 18 is a broad-spectrum MMP inhibitor.

Graphical abstract

We identify novel MMP inhibitors using chemical arrays, which detect small molecule–protein interactions on a chip in a high-throughput manner.

Key words:

• matrix metalloproteinase
• inhibitor
• chemical arrays

Matrix metalloproteinases (MMPs) are a family of zinc-dependent endopeptidases that degrade multiple components of the extracellular matrix (ECM), including collagens, fibronectin, laminin, proteoglycans, and elastin.^1,2) MMPs play a role in many physiological processes, such as ECM homeostasis, tissue remodeling, and organogenesis. However, MMPs can also participate in a wide range of pathological processes, such as cancer metastasis, rheumatoid arthritis, and chronic obstructive pulmonary disease.³⁻⁸⁾

In humans, 23 MMP proteins have been identified, which can be roughly classified based on substrate specificity into collagenases, gelatinases, stromelysins, matrilysins, membrane-type MMPs, and others. MMPs all possess an N-terminal signal peptide that targets them to the secretory pathway, a propeptide domain for enzyme latency, and a catalytic domain with a highly conserved zinc-binding site. The propeptide domain contains the conserved “cysteine switch” motif PRCGXPD at the N-terminus, which binds to the catalytic Zn²⁺ ion to block its activity. Most MMPs also possess a C-terminal hemopexin domain, which is coupled to the catalytic domain by a flexible hinge region and which is an important mediator of protein–protein interactions. MMPs are secreted as inactive zymogens, and they are activated via removal of the propeptide domain from the active site, for example, by proteolysis of other MMPs or furin-like serine proteases.^1,2)

Many small-molecule MMP inhibitors (MMPIs) have been designed to target MMPs in cancer.⁹⁻¹¹⁾ The first generation of MMPIs were peptidomimetics, which mimic the structure of collagen. They act as competitive inhibitors and chelate the zinc ion present at the active site. To improve specificity and oral bioavailability, non-peptidomimetics were synthesized, based on the 3D conformation of the MMP active site. Other MMPIs include tetracycline derivatives and novel mechanism-based MMPIs such as SB-3CT (2-((4-phenoxyphenylsulfonyl)methyl)thiirane). After the failure of early broad-spectrum MMPIs in the treatment of cancer in clinical trials, drug development strategies focused on more specific MMPIs. In addition, recent studies have shown that several MMPs have non-proteolytic functions, for example, through the hemopexin domain of MMPs.¹²⁻¹⁵⁾ Furthermore, several MMPs have been shown to provide tumor-suppressing functions in different stages of cancer progression.¹⁶⁾ Thus, much work remains for the elucidation of each MMP function and the development of new MMPIs.

Small-molecule microarrays represent a powerful new platform for facilitating target-based screens and reverse chemical genetics studies.¹⁷⁾ Thousands of small molecules are arrayed on a glass slide and used to screen for protein–small molecule interactions. This small-molecule microarray format has great advantages of miniaturization, throughput, and generality with regard to screening protocols. Typically, solid surfaces are derivatized with certain functional groups, and small molecules with a reactive functional group readily attach to the surface upon printing via a selective-coupling reaction. For example, the first small-molecule microarrays were prepared using the Michael addition reaction, involving small molecules containing free thiols printed on slides coated with vinyl sulfone or maleimide groups.¹⁸⁾ However, such selective-coupling approaches may limit the number of analyzable compounds and may mask the potential protein binding site.

We developed a rapid and universal coupling method that enables the introduction of a variety of small molecules to a solid support using a photoaffinity reaction.¹⁹⁻²¹⁾ In contrast to conventional methods, this method immobilizes a small molecule on a solid support in a functional group-independent (or non-selective) manner. Therefore, a wide variety of small molecules can be immobilized without chemical modification because chemical immobilization is achieved by UV irradiation. This method also allows small molecules to bind to all potential binding proteins, because the immobilized small molecules display virtually all of their surfaces. The utility of this photo-cross-linked small-molecule microarray has been proven in the discovery of ligands of many type of targets, including prokaryotic proteins,²²⁾ viral proteins,^23,24) and disease-related mammalian proteins.²⁵⁻²⁹⁾

In this study, we describe the identification of MMPIs using chemical arrays. The chemical array-based screen developed here, followed by an enzyme assay, identified isoxazole compound 1 as an MMP-9 inhibitor. By exploring analogs of 1, we identified compound 18, which was approximately 60-fold stronger than 1. Compound 18 inhibited various MMPs, including MMP-2, MMP-9, MMP-12, and MMP-13 and inhibited cell migration in human fibrosarcoma HT1080 cells, suggesting that 18 is a broad-spectrum MMPI.

Materials and methods

Materials

Compound 1 and naphthofluorescein were obtained from the RIKEN Natural Products Depository (NPDepo, Saitama, Japan).³⁰⁻³²⁾ Analogs of 1 were purchased from Namiki Shoji (Tokyo, Japan). All compounds were dissolved in dimethylsulfoxide (DMSO) as stock solutions, which were stored at −20 °C. His-Myc-tagged human recombinant MMP-9 was purified as previously described.³³⁾

Chemical array screening

The chemical arrays were prepared according to our previous reports.^20,25) Solutions of the 24,275 compounds (2.5 mg mL⁻¹ in DMSO) from NPDepo were arrayed onto 9 separate photoaffinity linker-coated glass slides using a chemical arrayer developed at RIKEN. The slides were incubated with TBS-T [10 mM Tris–HCl (pH 8.0), 150 mM NaCl, 0.05% Tween-20] containing 1% skim milk for 1 h at room temperature. Then, the slides were probed at 30 °C for 1 h with 1 μM purified His-Myc-tagged human recombinant MMP-9 in TBS-T containing 1% skim milk. After washing with TBS-T, the probed slides were incubated with anti-His antibody (mouse IgG, 1/100 diluted, GE Healthcare) in TBS-T containing 1% skim milk at 30 °C for 1 h. This incubation was followed by another washing step and incubation with a second antibody (goat anti-mouse IgG, Cy5 conjugate, 50 μg mL⁻¹, Millipore) at 30 °C for 1 h. After the final wash step, slides were scanned with a GenePix microarray scanner (Molecular Devices) using the Cy5 channel (an excitation wavelength of 635 nm and an emission wavelength of 675 nm). The fluorescence signals were quantified using the GenePixPro 6.1 software (Molecular Devices) with local background subtraction. We used data from slides treated with antibodies alone as a reference. The images from 2 slides treated with and without His-Myc-tagged MMP-9 were tinted red and green, respectively, using Photoshop 5.5 software (Adobe Photoshop) and were merged into a composite image.

MMP enzyme assays

MMP enzyme assays were performed according to the manufacturer’s instructions (R&D Systems). Activated human recombinant MMP-1 (R&D Systems), MMP-2 (R&D Systems), MMP-7 (R&D Systems), MMP-9 (R&D Systems), MMP-9 with a truncated C-terminal hemopexin domain (MMP-9/ΔHemopexin, Calbiochem), MMP-12 (R&D Systems), or MMP-13 (R&D Systems) was preincubated with the compounds for 30 min at 37 °C, and then, the fluorogenic peptide substrate was added (Mca-PLGL-Dpa-AR-NH₂, R&D Systems). Fluorescence emission at 405 nm with excitation at 320 nm was measured in kinetic mode for 30 min at 37 °C, using a fluorescent microplate reader (SpectraMax M2^e, Molecular Devices).

Cell culture and cell growth assays

The human fibrosarcoma cell line HT1080 (RIKEN Cell Bank) was cultured in D-MEM (Invitrogen), supplemented with 10% fetal calf serum (Nichirei Biosciences Inc., Tokyo, Japan), 50 U mL⁻¹ penicillin (Invitrogen), and 50 μg mL⁻¹ streptomycin (Invitrogen), at 37 °C in a humidified atmosphere containing 5% CO₂. Cell growth was measured using Cell Count Reagent SF (WST-8 solution, Nacalai Tesque, Kyoto, Japan) as described previously.³⁴⁾

Cell migration assay

Cell migration was determined using a wound-healing scratch assay. HT1080 cells were seeded at high densities (1 × 10⁵ cells/well) into 24-well culture plates and grown overnight. After replacement of the culture medium without serum for 2 h, cells were scratched with a pipette tip to generate a scratch wound and then incubated with or without compound 18 for 24 h. The healing rate was quantified using image analysis of post-culture measurements of the gap size (Adobe Photoshop). Statistical analysis was performed using one-way ANOVA followed by Bonferroni’s post hoc test. A p-value <0.05 was considered statistically significant.

Results

Chemical array-based screening for MMP-9 inhibitors

To identify novel MMP-9 inhibitors, we first performed chemical array-based screening to obtain small molecules that interacted with MMP-9. The chemical arrays of immobilized NPDepo library compounds were probed sequentially with or without purified His-Myc-tagged human recombinant MMP-9, anti-His antibody, and a Cy5-labeled secondary antibody and then scanned with a microarray scanner (Fig. 1(A)). The fluorescence signals from two slides treated with and without His-Myc-tagged MMP-9 were quantified. In this first screening, 27 hits were obtained among the 24,275 compounds (Fig. 1(B)).

Fig. 1. Chemical array-based screening for MMP-9 inhibitors.

Notes: (A) Representative fluorescent images of the chemical arrays. The chemical arrays were probed sequentially with purified His-Myc-tagged human recombinant MMP-9, anti-His antibody, and a Cy5-labeled secondary antibody and then scanned with a microarray scanner. The fluorescent images of 2 slides, with and without His-Myc-tagged MMP-9, were tinted red and green, respectively, and merged into a composite image. Candidate compounds are indicated with white arrows. (B) Overview of the screening results. The first screening using chemical arrays identified 27 hits among 24,275 compounds from the NPDepo library. In the second screening using an in vitro enzyme assay, 2 compounds (1 and naphthofluorescein) were found to inhibit MMP-9 enzyme activity. (C) Chemical structures of 1 and naphthofluorescein.

We next performed an in vitro enzyme assay to examine the inhibitory activity of these 27 compounds on MMP-9. In this second screening, two of the compounds (isoxazole compound 1 and naphthofluorescein) inhibited MMP-9 enzyme activity (Fig. 1(B) and (C)). The IC₅₀ values of 1 and naphthofluorescein against MMP-9 were 233 (Table 1) and 15 μM, respectively.

Table 1. Inhibition of MMP-9 by analogs of 1.

Compound	Structure	Compound	Structure
IC50 (μM)		IC50 (μM)
1		10
233 ± 38		>300
2		11
>30		>30
3		12
164 ± 39		11.0 ± 10.4
4		13
267 ± 31		22.6 ± 3.4
5		14
>30		9.1 ± 5.1
6		15
>300		>30
7		16
>300		25.6 ± 0.1
8		17
>30		101 ± 27
9		18
239 ± 21		3.8 ± 1.2

Notes:

Human recombinant MMP-9 was preincubated with the compounds for 30 min at 37 °C, and then, the fluorogenic peptide substrate was added. Fluorescent intensity was measured in kinetic mode for 30 min at 37 °C. IC₅₀ values are means ± SD (n = 3).

Inhibition of MMP-9 by analogs of 1

To obtain more potent inhibitors of MMP-9, we further explored analogs of 1. An in vitro MMP-9 enzyme assay showed that compound 18 had the strongest inhibitory activity, with an IC₅₀ value of 3.8 μM; it was approximately 60-fold stronger than 1 (Table 1). Overall, analogs with an amide (12, 16, and 18), a sulfonamide (13 and 17), or a carbamate (14) showed increased potency of enzyme inhibition.

Compound 18 is a broad-spectrum MMPI

We next examined the selectivity and potency of 18 on various MMPs. As shown in Table 2, 18 inhibited the enzyme activities of MMP-2, MMP-12, and MMP-13, as well as MMP-9 and MMP-9/ΔHemopexin, with almost identical potency. Compound 18 did not inhibit the enzyme activities of MMP-1 or MMP-7 up to 100 μM (Table 2). These results suggest that 18 is a broad-spectrum MMPI.

Table 2. Inhibition of various MMPs by 18.

Enzyme	IC50 (μM)
MMP-1	>100
MMP-2	2.1 ± 0.4
MMP-7	>100
MMP-9	3.8 ± 1.2
MMP-9/ΔHemopexin	3.6 ± 1.4
MMP-12	3.2 ± 0.9
MMP-13	4.3 ± 1.0

Notes:

Each human recombinant MMP was preincubated with 18 for 30 min at 37 °C, and then, the fluorogenic peptide substrate was added. Fluorescent intensity was measured in kinetic mode for 30 min at 37 °C. IC₅₀ values are means ± SD (n = 3 or 4).

Compound 18 inhibits cell migration in HT1080 cells

Some MMPIs are known to inhibit migration of human fibrosarcoma HT1080 cells.^35,36) To examine the effect of 18 on cells, we monitored migration of HT1080 cells treated with 18. As shown in Fig. 2, 18 significantly inhibited migration in HT1080 cells with almost no cell toxicity.

Fig. 2. Compound 18 inhibits cell migration in HT1080 cells.

Notes: (A and B) Effect of 18 on cell migration in HT1080 cells. HT1080 cells in serum-free medium were scratched and then incubated with 18 for 24 h (A). The healing rate was quantified using post-culture measurement of gap size, using image analysis (B). Cell migration (%) indicates the percentage of control. Data are shown as the mean ± SD (n = 3). *, p < 0.05 vs. vehicle. (C) Effect of 18 on growth of HT1080 cells. HT1080 cells were treated with 18 for 48 h. Cell growth was analyzed by WST-8 assay. Cell number (%) indicates the percentage of control. Data are shown as the mean ± SD (n = 3).

Discussion

Using a chemical array-based screen for MMP-9 inhibitors, we identified a compound (18) that is a broad-spectrum MMPI, as evidenced by its inhibition of various MMPs in vitro (Table 2) and the inhibition of HT1080 cell migration (Fig. 2). The catalytic Zn²⁺ ion at the active site of MMPs is surrounded by subsite pockets designated as S1, S2, S3, S1′, S2′, and S3′. Of these, the S1′ pocket provides the basis of selectivity for many MMPIs.¹¹⁾ MMP-3, -11, -12, and -13 have deep S1′ pockets; MMP-2, -8, and -9 have intermediate S1′ pockets, but MMP-1 and -7 have shallow S1′ pockets, which are occluded by Arg²¹⁴ and Tyr²¹⁴, respectively. Compound 18 inhibited the enzyme activities of MMP-2, -9, -12, and -13, but not MMP-1 and -7, suggesting that this compound might target near the S1′ pocket of the active site.

We also found that naphthofluorescein inhibits the enzyme activity of MMP-9 with an IC₅₀ value of 15 μM (Fig. 1). Naphthofluorescein is known as a fluorescent indicator or a pH indicator and has several biological activities, such as inhibition of furin.³⁷⁾ Therefore, derivatization would be required to use naphthofluorescein as a specific MMPI.

Recently, small-molecule inhibitors targeting the hemopexin domain of MMPs have been reported.^38,39) For example, Dufour et al. identified small-molecule compounds that specifically bind to the hemopexin domain of MMP-9 using an in silico docking approach.³⁸⁾ These compounds abrogated MMP-9 homodimerization without affecting its catalytic activity and inhibited MMP-9-mediated cell migration. Chemical array screening for MMP-9 ligands identified 27 candidate compounds (Fig. 1). Of these 25 compounds (excluding 1 and naphthofluorescein), some scarcely affected the catalytic activity of MMP-9 (data not shown). Although false positives are possible, such compounds might bind a domain other than the catalytic domain, for example, the hemopexin domain, and might affect MMP-9 function. In this regard, further analyses, such as evaluation of inhibition of MMP-9 dimerization, would be needed to characterize the effect of the compounds.

Chemical array platforms are based on interactions between proteins and small molecules and are adaptable to any type of protein, including non-enzymes and artificial proteins. Therefore, it would be possible to identify small-molecule compounds specifically targeting each MMP domain, using various deletion mutant proteins. Further, chemical array screens for a series of MMPs would allow identification of both broad-spectrum and specific MMPIs in a high-throughput manner.

In summary, we have demonstrated that 18 is a broad-spectrum MMPI using a chemical array-based screen followed by a secondary assay and derivatization. Our study has also shown that our photo-cross-linked chemical arrays are useful for high-throughput drug screening and reverse chemical genetics studies.

Authors contribution

M.K., Y.Y., N.T., and H.O. designed the research plan; M.K., Y.F., and Y.Y. performed the biochemical and cellular experiments and analyzed data; Y.K., K.H., and T.S. performed the chemical array screening and analyzed data; M.K., Y.K., and H.O. wrote the manuscript with assistance from all authors; and N.T. and H.O. supervised the research.

Acknowledgments

We wish to thank Y. Oku (RIKEN), M. Uchida (RIKEN), H. Aono (RIKEN), C. Gao (Osaka University), M. M. Khan (EPFL), T. Nogawa (RIKEN), M. Muroi (RIKEN), N. Watanabe (RIKEN), and T. Saito (RIKEN) for their help with the experiments. This work was supported in part by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO) and Grant-in-Aid for Scientific Research (KAKENHI).

Disclosure statement

The authors declare no conflicts of interest.