J Biol Inorg Chem (2006) 11: 131–138 DOI 10.1007/s00775-005-0053-x

O R I GI N A L A R T IC L E

David T. Puerta Æ Michael O. Griffin Æ Jana A. Lewis Diego Romero-Perez Æ Ricardo Garcia Francisco J. Villarreal Æ Seth M. Cohen

Heterocyclic zinc-binding groups for use in next-generation matrix metalloproteinase inhibitors: potency, toxicity, and reactivity Received: 7 September 2005 / Accepted: 22 October 2005 / Published online: 3 December 2005
SBIC 2005

Abstract In an effort to improve the zinc-chelating portion of matrix metalloproteinase (MMP) inhibitors, we have developed a family of heterocyclic zinc-binding groups (ZBGs) as alternatives to the widely used hydroxamic acid moiety. Elaborating on findings from an earlier report, we performed in vitro inhibition assays with recombinant MMP-1, MMP-2, and in a cell culture assay using neonatal rat cardiac fibroblast cells. In both recombinant and cell culture assays, the new ZBGs were found to be effective inhibitors, typically 10–100-fold more potent than acetohydroxamic acid. The toxicity of these chelators was examined by using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide and 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt cytotoxicity assays, which demonstrate that most of these compounds are nontoxic at concentrations of almost 100 lM. To address the possible interaction of sulfurcontaining ZBGs with biological reductants, the reactivity of these chelators with 5,5¢-dithiobis(2-nitrobenzoic acid) was examined. Finally, thione ZBGs were shown to be effective inhibitors of cell invasion through an extracellular matrix membrane. The data presented herein suggest these heterocyclic ZBGs are potent, nontoxic, and biocompatible compounds that show promise for incorporation into a new family of MMP inhibitors. Keywords Hydroxypyridinones Æ Metalloproteinases Æ Pyrones Æ Sulfur ligands Æ Zinc D. T. Puerta Æ J. A. Lewis Æ S. M. Cohen (&) Department of Chemistry and Biochemistry, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0358, USA E-mail: [email protected] Tel.: +1-858-8225596 Fax: +1-858-8225598 M. O. Griffin Æ D. Romero-Perez Æ R. Garcia Æ F. J. Villarreal Department of Medicine, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0358, USA

Introduction Matrix metalloproteinases (MMPs) are zinc(II)-dependent enzymes responsible for the hydrolytic cleavage of the components of the extracellular matrix (ECM), such as collagen, elastin, and gelatin [1]. The tissue breakdown mediated by MMPs facilitates numerous physiological processes, such as wound healing, angiogenesis, and cell division [2–4]. In contrast, the activity of MMPs has also been implicated in pathologies such as cardiovascular disease, cancer, and arthritis [5, 6]. As a result, MMPs have been a highly sought after therapeutic target for more than 2 decades. MMP inhibitors (MPIs) generally follow a two-component strategy: a peptidomimetic backbone is designed to interact in a noncovalent fashion with the MMP active site (in specific substrate interaction sites termed ‘‘subsites’’), while an appended zinc(II)-chelating moiety (often referred to as a zinc-binding group, ZBG) binds via coordinate– covalent bonds to the hydrolytic zinc(II) ion, rendering the enzyme inactive. Thousands of MPIs have been synthesized, and some have been tested in clinical trials. We estimate that more than 90% of MPIs contain a hydroxamic acid as the ZBG [1, 7]. The majority of research effort has been placed on the development of the MPI backbone to obtain specificity toward MMPs, while relatively little attention has been paid to improving the zinc-chelating portion of the inhibitor. Interest in non-hydroxamate MPIs has increased recently in light of the troubled history of hydroxamatebased MPIs in the clinic; however, hydroxamate-based MPIs continue to dominate this field [4, 6, 8, 9]. Hydroxamic acid MPIs are potent compounds in vitro, but their clinical efficacy has been limited by poor oral availability and biocompatibility. Hydroxamic acids are rapidly hydrolyzed in vivo generating a carboxylic acid, which is generally much less effective as a ZBG [10]. This hydrolysis also produces potentially harmful and unwanted metabolites such as nitric oxide [11]. In light of these limitations, we have concerned ourselves

132

with improving inhibitor–metal ion complexation. As a result, we have developed a facile strategy to evaluate the structural properties of this interaction [12–14] and have identified a small library of novel ZBGs for use in MPIs [15]. An initial evaluation of these ZBGs was performed using an in vitro assay with purified MMP-3. In these experiments four different groups of chelators were studied (Fig. 1): hydroxypyridinones (1–3, 6), pyrones (4, 5), hydroxypyridinethiones (7–9), and thiopyrones (10, 11). These compounds represent only the zincchelating portion of a new MPI and do not contain the critical peptidomimetic backbone. Lacking the critical backbone, the ZBGs were compared with the inhibitory activity of acetohydroxamic acid (AHA) as a representative of the standard hydroxamate chelator found in most MPIs. All of the new compounds were found to be more potent inhibitors of MMP-3 than AHA [15]. Furthermore, the ZBGs were found to have a greater affinity than AHA for the zinc(II) ion in a model complex of the MMP active site [15]. The cyclic structures of these compounds were anticipated to confer enhanced bioavailability and biostability when incorporated into MPIs. The improved potency and stability of these ZBGs were proposed as important advancements in the development of next-generation metalloprotein inhibitors. The present report expands upon our initial study of heterocyclic ZBGs 1–11, by presenting important data on the biocompatibility and efficacy of these compounds. The inhibitory activity of these new compounds is attributed to enhanced zinc binding, not from other nonspecific protein interactions. To evaluate this hypothesis, the inhibitory activities of these ZBGs against MMP-1 and MMP-2 are reported here. The three MMPs examined (MMP-1, MMP-2, and MMP-3) have distinct subsite structures, but all have a highly conserved active center containing a zinc(II) ion bound by three histidine residues. In addition, to evaluate the potential therapeutic application of these compounds, assays using neonatal rat cardiac fibroblast (CF) and fibrosarcoma HT1080 cells were performed. The ability of these ZBGs to inhibit invasion of HT1080 cells through a matrix membrane was examined. The cytotoxicity of these compounds was examined using a standard cell viability assay. The reaction of some

Fig. 1 Heterocyclic zincbinding group (ZBGs) proposed for use in matrix metalloproteinase (MMP) inhibitors. Acetohydroxamic acid (AHA) was utilized as a benchmark and N-(methyl)mercaptoacetamide (N-MMAA) as a representative thiol chelator in the experiments described herein

ligands with 5,5¢-dithiobis(2-nitrobenzoic acid) (DTNB) was studied to identify possible nonspecific and undesired reactivity with cysteine residues and other biological reductants that the compounds might encounter in vivo. The results show that these ligands are effective chelators that will be potent ZBGs when incorporated into novel MPIs.

Materials and methods Preparation of ZBGs Compounds 1–11 were purchased from a commercial supplier or prepared as previously described [16, 17]. 1H NMR spectra of the O,S ligands (7–11) were used to show that only monomeric species of each ligand were present in solution. For all compounds, the hydroxyl proton was present in the 1H NMR spectra and the spectral data for 9 were also consistent with published values for the monomeric O,S ligand [18]. No dimerassociated resonances were present in the NMR spectra and no dimer peaks were found in the mass spectrometry analyses, indicating that the compounds are pure as synthesized. Recombinant MMP assays Activities of Escherichia coli recombinant human MMP1 catalytic domain (amino acids 81–249, 19.9 kDa) and MMP-2 catalytic domain (amino acids 81–423, 40 kDa) were measured utilizing a 96-well microplate fluorescent assay kit purchased from BIOMOL International, following the procedure provided with the kit. Experiments were performed using a Bio-Tek Flx 800 fluorescence plate reader and Nunc white 96-well plates. The compounds (Fig. 1) were dissolved in dimethyl sulfoxide (DMSO) and further diluted (500 times) in assay buffer [50 mM N-(2-hydroxyethyl)piperazine-N¢-ethanesulfonic acid, 10 mM CaCl2, 0.05% Brij-35, pH 7.5]. MMP-1 and MMP-2 were incubated individually with varying concentrations of different inhibitors for 1 h at 37 C, followed by addition of substrate to initiate the assay. Reactions were agitated by shaking for 1 s after each

133

fluorescence measurement. Upon cleavage of the fluorescent substrate, Mca-Pro-Leu-Gly-Leu-Dpa-Ala-ArgNH2 [0.4 mM in assay; Mca is 7-(methoxycoumarin-4yl)-acetyl; Dpa is N-3-(2,4-dinitrophenyl)-L-a-b-diaminopropionyl] at the Gly-Leu bond, Mca fluorescence (kex=335 nm, kem=405 nm) was measured at 60-s intervals for 30 min. The IC50 values obtained were corrected for competitive absorption as previously described [15]. CF preparation Neonatal rat CFs were prepared as previously described [19]. Briefly, CFs were prepared from hearts of 1–2-dayold Sprague–Dawley rats. Following collagenase digestions (4 times), non-myocyte cells (mostly fibroblasts) were isolated by Percoll density gradient. The cell suspension was plated onto uncoated tissue culture dishes for 30 min to allow preferential attachment of CFs to the bottom of the dish. The nonadherent cells were removed and fresh medium (Dulbecco’s modified Eagle’s medium plus 10% fetal bovine serum) was added. CFs were allowed to proliferate to confluence and then trypsinized and frozen at 70 C. CF stocks were freshly plated for each experiment and used from first or second passages. When reaching 90% confluence, the cells were serum deprived for 24 h and then treated according to the experimental design (vide infra). Cell culture assays To activate CF MMPs, cells were treated with plasminogen (60 lg/ml, Sigma) for 16 h. MMP activity was determined by adding 10 mM of substrate (Mca-ProLeu-Gly-Leu-Dpa-Ala-Arg-NH2) (BIOMOL International) in assay buffer [50 mM tris(hydroxymethyl)aminomethane, pH 7.5, 150 mM NaCl, 5 mM CaCl2] with 8 lg/ml aprotinin to neutralize residual plasmin activity, to the culture medium. Fluorescence measurements were collected in a similar manner as described for the purified MMP assays (vide supra). Data were fit to the sigmoidal Hill equation y=[ZBG]/([ZBG]+k) using Prism 3.0 (GraphPad Software, San Diego, CA, USA). In this equation, y is the rate of substrate hydrolysis as a fraction of maximal substrate hydrolysis, and k is the ZBG concentration at which activity is half-maximal (IC50 value). Cell viability assays To assess ZBG cell toxicity, 3-(4,5-dimethylthiazol-2-yl)2,5-diphenyltetrazolium bromide (MTT, Sigma) and 3(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)2-(4-sulfophenyl)-2H-tetrazolium salt (MTS, Promega) cell viability assays were performed according to the manufacturer’s instructions. MTT and MTS are

substrates that provide a colorimetric signal in response to viable mitochondria. CF cells were incubated with varying concentrations of inhibitors (0.001–10 mM) for 24 h. After incubation, the cell medium was removed and the cells were rinsed three times with fresh medium to remove the excess ZBGs, which could interfere with the MTT or MTS reaction. The cells were then lysed and combined with MTT or MTS in medium for 1–2 h. Following solubilization (required only for MTT), absorbance at 570 nm was measured using a microplate reader (Bio-Tek l-Quant Instrument, Winooski, VT, USA). To qualitatively assess ZBG cytotoxicity, photomicrographic images of untreated and inhibitor-treated CFs were collected 24 h after treatment at several does of each ZBG. Thiol reactivity To evaluate inhibitor reactivity with free thiols, assays with DTNB were performed. Inhibitors were dissolved in DMSO and further diluted (250 times) into assay buffer (50 mM 2-morpholinoethanesulfonate, 10 mM CaCl2, pH 6.0). The assay buffer was chosen to mimic the buffer used previously for colorimetric MMP-3 assays [15]. Inhibitor solutions (1 ml of 200 lM) were quickly mixed with a solution of DTNB (1 ml of 1 mM) in assay buffer (giving a 5:1 DTNB-to-ZBG ratio) in a 10-mm quartz cuvette and sealed using a Teflon stopper (VWR). For compounds 7 and 9, additional reactions with other ratios of DTNB to ZBG (10:1, 1:1, and 1:10) were also studied. The spectra of the solutions were monitored with one reading per minute for 225 min at 405 nm (wavelength used for colorimetric MMP assays) and at 412 nm (kmax for TNB2 , e=13,600 M 1 cm 1 at pH 6.0) [20]. Spectra were collected with a PerkinElmer Lambda 25 spectrophotometer. Absorbance versus time was plotted in order to provide information about the rate and extent of reaction with DTNB. Cell invasion assays Cell invasion assays were obtained from Chemicon International and used according to the manufacturer’s instructions with minor modifications. Fibrosarcoma HT1080 cells were grown in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum and were serumstarved for 24 h prior to initiating the assay. The ZBGs were dissolved in DMSO and diluted 500-fold in serumfree Dulbecco’s modified Eagle’s medium. All wells contained the same concentration of DMSO during the assay. As a chemoattractant, 10% fetal bovine serum was used. The final solutions, containing digested cells, CyQuant GR Dye, and lysis buffer, were transferred to white 96-well Nunic flat-bottom plates. The plates were read using a Bio-Tek Flx 800 fluorescence plate reader using kex=460 nm and kem=516 nm. The ZBGs were monitored for inhibition of invasion at 1 mM (AHA,

134

1–6) and 100 lM (8–10). Owing to problems with toxicity at 100 lM, 7 was examined at 50 lM. All experiments were repeated in quadruplicate.

Results and discussion Purified MMP assays Our previous report of the compounds depicted in Fig. 1 as novel ZBGs involved an assay using recombinant MMP-3 [15]. AHA was used as a representative benchmark ZBG found in most inhibitors. All the compounds tested showed greater inhibitory activity than AHA; compounds 1–6 demonstrated inhibitory activities 3–16fold greater than AHA, while the most marked improvement in activity was observed with compounds 7–11, which were 70–700-fold more potent. It is important to note that the modest IC50 values of these compounds are due to the lack of a peptidomimetic backbone; this is best exemplified by AHA, which has an IC50 value of approximately 25 mM alone (against MMP-3), but when attached to the appropriate backbone has generated many nanomolar potency inhibitors [1, 7, 9]. The affinity for the individual components of an MPI (ZBG vs. peptidomimetic backbone) has been revealed by the structure activity relationship by NMR method, which showed a greater-than-additive effect by linking the two components together [21, 22]. To demonstrate that compounds 1–11 inhibit MMPs by zinc(II) chelation and not through interactions specific to the protein backbone or side chains of MMP-3, two different MMPs (MMP-1 and MMP-2) were assayed in this study. The zinc(II)-tris(histidine) hydrolytic active site is conserved throughout the MMP family, but the surrounding subsites in MMPs are not. MMP-1 (collagenase) and MMP-2 (gelatinase A) have distinct active sites and both differ from that of MMP-3 (stromelysin) [1]. The shallow S1¢ pocket in MMP-1 differs distinctly from the large hydrophobic channel found in MMP-2 and MMP-3. Most synthetic inhibitors

of MMPs have exploited the S1¢ pocket to obtain selectivity over MMP-1 [1]. MMP-2 and MMP-3 are more closely related; however, there are disparities in the S3¢ subsite of the two proteinases. Inhibitors have been designed that are selective for MMP-3 over MMP-2 by exploiting the differences in the S3¢ subsite [23]. In addition to structural differences among these MMPs, the catalytic activities of these enzymes show divergent dependences on pH. MMP-3 has optimal catalytic activity at pH 6, while MMP-1 and MMP-2 are most efficient at pH 7.5 [24]. A subset (1, 2, 5, 7–9, 11) of the compounds from our initial study was tested against MMP-1 and MMP-2. This subset represents all of the ligand motifs examined in the initial study: AHA (control), hydroxypyridinones (1, 2), pyrones (5), hydroxypyridinethiones (7–9), and thiopyrones (11) (Fig. 1). Consistent with inhibitory data on MMP-3, the compounds tested (1, 2, 5, 7–9, 11) were more potent inhibitors of MMP-1 and MMP-2 than AHA (Table 1). The O,O donor ligands (1, 2, 5) were fourfold to tenfold more potent inhibitors of MMP-1 and threefold to sixfold more potent inhibitors of MMP-2 than AHA (Fig. 2). The O,S ligands (7–9, 11) showed notable improvements in inhibition, where they demonstrated a 60–280-fold increase in potency over AHA for MMP-1 and a 40–260-fold increase for MMP2. The apparent decrease in efficacy of the compounds against MMP-1 and MMP-2 compared with MMP-3 may be due to the differences in the assay conditions. The assays were run at the optimal pH values for each enzyme; MMP-3 is most active at pH 6, while MMP-1 and MMP-2 are most active at pH 7.5. The change in pH from 6 to 7.5 may have an effect on the protonation state of either the inhibitor or the amino acid side chains in the active site of the MMPs. It has been shown that the protonation state of a catalytic Glu residue in the active site strongly affects the potency of inhibitors. A more acidic inhibitor (e.g., carboxylate) will be less potent at a higher pH, where the Glu residue is deprotonated, while a basic inhibitor (e.g., hydroxamate) will be less potent when the pH of the system is lower than the

Table 1 IC50 values (lM) for zinc-binding groups (ZBGs) against MMP-1, MMP-2, and MMP-3, and in cardiac fibroblast (CF) cell culture ZBG

MMP-1a

MMP-2a

MMP-3b

CF culturea

AHA 1 2 4 5 6 7 8 9 10 11

41,600 (±400) 5,960 (±40) 4,200 (±300) Not determined 4,200 (±300) Not determined 490 (±10) 680 (±20) 150 (±10) Not determined 400 (±10)

15,000 (±3,000) 5,600 (±100) 2,600 (±400) Not determined 2,600 (±100) Not determined 100 (±40) 380 (±10) 60 (±10) Not determined 140 (±10)

25,000 (±4,000) 1,600 (±100) 5,100 (±200) 7,200 (±1,200) 5,700 (±100) 5,700 (±200) 35 (±3) 362 (±3) 140 (±20) 120 (±40) 210 (±20)

8,700 (±400) 3,240 (±140) 2,430 (±130) 850 (±40) 273 (±6) 2,870 (±90) 790 (±30) 135 (±2) 51 (±2) 125 (±2) 86 (±2)

Compound 3 was not evaluated owing to poor aqueous solubility a Values based on at least three independent experiments b From Ref. [17]

135 Fig. 2 IC50 values of the compounds (AHA, 1–11, excluding 3) against MMP-1 (black), MMP-2 (crossed), and MMP-3 (gray). The inset highlights the most potent compounds (7–11)

pKa (approximately 5.6) of the Glu residue [24]. The two most acidic ZBGs, 1 and 7 (pKa 5.8 and 4.6, respectively) [25–27], display the most significant decrease in potency, which is consistent with the aforementioned trends. The other ZBGs examined (including AHA) have pKa values that are more basic (approximately 8.1–9.5) [25, 26] [28], and are less affected by the increase in pH. Nevertheless, the overall inhibitory trend for different ZBGs is consistent for all three MMPs. On average the ZBGs were twofold more potent against MMP-2 than MMP-1 (Fig. 2). Control experiments without inhibitor showed that MMP-1 is less active than MMP-2 (data not shown). Assuming that the ZBGs act as competitive inhibitors, which is consistent with a hypothesis of active-site metal binding, we would expect that inhibition of the less-active MMP-1 will exhibit higher IC50 values relative to MMP-2. The fact that AHA exhibits the same difference in potency between MMP-1 and MMP-2 supports the hypothesis that the differences in IC50 values between MMPs are due to differences in enzyme activity and not because of nonspecific protein interactions of the ZBGs. All of the compounds presented here are small, simple molecules with little functionality and are unlikely to have any strong, specific interactions with the protein subsites. Overall, compared with AHA, the compounds tested here proved to be superior inhibitors of MMP-1, MMP-2, and MMP-3. The inhibition of three different classes of MMPs by these compounds with similar IC50 values strongly suggests that these ligands inhibit the enzyme through enhanced zinc(II) binding and not through noncovalent or other unanticipated protein interactions. This conclusion is further supported by the correlation of IC50 values with relative binding affinities of these chelators toward an MMP model complex [15]. Spectroscopic studies with cobalt(II)- or cadmium(II)-substituted MMPs will be used to test this hypothesis and determine whether these chelators inhibit MMPs by activesite metal binding.

Cell culture assays The promising inhibitory potential of these compounds against recombinant MMPs led us to examine their activities in a cellular assay. CF cultures were selected for evaluating new ZBGs on the basis of two primary reasons. First, CF cells have a well-documented capacity to produce and secrete a variety of MMPs into culture media in zymogen form, including MMP-2 and MMP-9 [29]. The CF-secreted MMPs are amenable to activation by several mechanisms, including protease-mediated cleavage by enzymes such as plasminogen [30]. Addition of plasminogen activates MMPs by first itself being cleaved by urokinase plasminogen activator (uPA) present on the cell surface and then cleaving the prodomain of the MMP [31]. A second reason for using CF cultures is an increasing body of experimental work indicating that the use of MPIs may have a therapeutic benefit following myocardial infarction [32]. The compounds in Fig. 1 (not including N(methyl)mercaptoacetamide, N-MMAA) were analyzed for the ability to inhibit plasmin-activated MMPs produced by CFs in culture. All of the new ZBGs showed greater potency than AHA, with 9 demonstrating the strongest inhibition (Table 1). The order of potency largely paralleled that seen in vitro using purified MMPs. Of notable exception, 5 demonstrated significantly greater potency, while 7 showed markedly reduced potency in CF culture. These discrepancies may be due to interactions of these compounds with other components of the cellular media. At the concentrations of ZBG used to inhibit these enzymes these compounds will complex loosely bound metal ions in the culture medium and may inhibit metalloenzymes other than MMPs, which might perturb the observed efficacy in this assay. Therefore, assessing the MMP inhibitory capacity of new compounds in culture provides additional information regarding their efficacy toward a ‘‘physiological’’ mix of MMPs that might remain concealed by in vitro assessment alone.

136

Fig. 3 3-(4,5-Dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)2-(4-sulfophenyl)-2H-tetrazolium salt cell viability assay results. The relative toxicities of the compounds (AHA, 1–11, excluding 3) in neonatal rat cardiac fibroblasts at concentrations of 0.001 mM

(black), 0.01 mM (white), 0.1 mM (crossed), and 1.0 mM (gray) of the ZBG listed. Colorimetric assay values were normalized to those of untreated cells

Cell toxicity

compounds (including AHA) showed lower cell viability, particularly at high concentrations (data not shown). The photomicrographs shown in Fig. 4 are representative of cell images obtained from control (i.e., untreated) CFs and 0.1- and 1.0-mM doses of compound 9. At the 0.1-mM dose of 9, CFs showed little evidence of disruption in cell morphology and are comparable to those of untreated cells. In contrast, at the 1.0-mM dose of 9, CFs demonstrated disruption of normal cell morphology and evidence of cell death. For all ZBGs, visual evidence for disruption of cell morphology was obtained only at doses of 1.0 mM or higher (data not shown). In summary, none of the compounds examined are acutely toxic, which indicates they are viable moieties for incorporation into full MPIs.

Compounds 1–11 have been shown both in vitro and in cell culture to be potent, broad-spectrum inhibitors of MMPs (compound 3 has not been completely evaluated owing to poor aqueous solubility). We sought to address the question of cytotoxicity and biocompatibility using a cell viability assay. In this assay, cellular viability is monitored by the reduction of MTT or MTS by mitochondrial enzymes, producing a characteristic blue color. Several of the O,S donor atom ZBGs (7–11) were found to react with MTT and MTS, producing false positives for cell viability in our initial assay attempts. A control experiment between MTT and compounds 7–11 in the absence of CF cells confirmed this reactivity (data not shown). To eliminate this undesired side reaction, the assay protocol was slightly modified by introducing an additional wash step prior to MTT/MTS addition. After incubating the CF cells with inhibitors for 24 h, the cells were washed with fresh medium to remove any excess inhibitors from the wells. Subsequently, the MTT or the MTS was added as per the standard protocol to evaluate the viability of the CF cells. This simple rinse step gave reproducible MTT and MTS assay results; the assays along with photomicrographs of the CF cells were used for evaluation of inhibitor cytotoxicity. Figure 3 shows the results of the MTS experiments with CF cells. The new ZBGs showed varying degrees of cytotoxicity, with the sulfur-containing compounds (7–11) generally being more toxic. All of the compounds showed low toxicity at concentrations up to 100 lM (with the exception of 7), and compounds 1–6 were comparable to AHA. The MTT assay gave similar results to those obtained with MTS, although all of the

Thiol reactivity In our earlier study, a colorimetric assay with purified MMP-3 was performed to confirm the potency of some ZBGs [15]. Activity was monitored upon cleavage of a thioester substrate, where the free thiol produced upon cleavage by MMPs reacts with DTNB to form TNB2 , which can be observed spectrophotometrically. In certain cases, we observed a chemical reaction between the O,S donor ligands and the DTNB reagent used in the assay. In addition, some O,S chelators were found to undergo a reaction with MTT in cell viability assays (vide supra). These observations prompted us to examine the reactivity of the ZBGs in Fig. 1 with DTNB, as a simple model of reactivity with biological species containing thiol moieties. The reactivity toward thiol-containing biological species

137

Fig. 4 Photomicrographs of cardiac fibroblast cells in the absence (a) and presence of two concentrations of compound 9: 0.1 mM (b) and 1.0 mM (c). The image in c is representative of those observed in cells exposed to high concentrations of several of the ZBGs

was of concern, as this can prove disadvantageous for in vivo applications [22, 33]. To examine the stability of the compounds, their reaction with DTNB was monitored spectrophotometrically [20]. Reactivity with DTNB to produce TNB2 was monitored at 412 and 405 nm, which represents the kmax of TNB2 and the wavelength at which the colorimetric MMP-3 assay was performed, respectively [15]. For comparison, N-MMAA was used as a representative free thiol ZBG. As expected, O,O donor ligands such as AHA and 5 showed no reactivity with DTNB after more than 3 h of incubation. In contrast, the free thiol N-MMAA was a very reactive compound toward DTNB; within 8 min, 91% of N-MMAA in solution had reacted with DTNB (data not shown). The different hydroxypyridinethione and thiopyrone isomers showed varying reactivity with DTNB, which could be categorized into three groups. Hydroxypyridinethione 8 was the sole member of the first group of compounds which showed no reactivity with DTNB, similar to the O,O donor ligands. In the second group, hydroxypyridinethiones 7 and 9 showed a rapid initial reaction with DTNB followed by a steady plateau in the kinetic trace. The degree of conversion for 7 and 9 was significantly less than that of N-MMAA and was dependent on the initial concentration of the ZBG and DTNB, consistent with the establishment of a concentration-dependent equilibrium. The third group of compounds was thiopyrones 10 and 11, which showed a gradual conversion over time. The rate of reaction of thiopyrones 10 and 11 with DTNB is different, and varied substantially depending on the thiopyrone ring substituents (data not shown). Using this simple, qualitative analysis, both the O,O and O,S donor ligands in Fig. 1 were found to be much less reactive toward DTNB when compared with a free thiol. In addition, the reactivity of thione O,S ligands varied depending on the ring structure (e.g., position of endocyclic heteroatoms, ring substituents). Therefore, the design of complete MPIs based on these chelators can be directed such that the most unreactive ZBGs are employed. Cell invasion assays Several of the ZBGs in Fig. 1 were tested for their ability to inhibit a biological MMP-dependent process, namely

the invasion of fibrosarcoma HT1080 cells through a membrane comprised of ECM proteins. In the cell invasion assay, cells move from an upper to a lower chamber as a function of their ability to degrade a synthetic membrane composed of ECM proteins. Thus, this cell invasion process is dependent on MMP activity. A standard cell invasion assay kit was used to examine this phenomenon and the results are summarized in Fig. 5. Most of the O,O donor ligands (AHA, 2, 5) had little or no effect on invasion at a concentration of 1 mM; however, the O,O ligand compound 1 did inhibit invasion by 70% at this same concentration (data not shown). The O,S donor ZBGs (8, 9, 11) inhibited invasion by 52–62% at a concentration of 100 lM, while compound 7 inhibited invasion by 53% at a concentration of 50 lM. The ability of these compounds to inhibit cell invasion correlates well with their in vitro ability to inhibit MMPs, where the O,S donor compounds are far

Fig. 5 Cell invasion assay results. Relative fluorescence units (RFU, 516 nm) for several ZBGs; an increase in fluorescence indicates more invasion of fibrosarcoma HT1080 cells through an extracellular matrix membrane. The control contains no inhibitor, for all others AHA 1 mM, 7 50 lM, 8 100 lM, 9 100 lM, and 10 100 lM

138

more potent MMP inhibitors. These experiments demonstrate the ability of these compounds to inhibit an MMP-dependent process in an in vitro model of cell invasion.

Concluding remarks MPI design and synthesis have yielded a vast number of compounds with a variety of backbone substituents that are designed to interact with specific MMP subsites. In comparison, little progress has been made on improving the ZBG of the inhibitor that binds directly to the ubiquitous active-site zinc(II) ion. With the limitations of hydroxamic acids as a ZBG clearly established, it is increasingly apparent that improved ligands must be developed for second-generation MPIs [22, 34]. As part of an ongoing study of new chelators for use in MPIs, the inhibitory properties of compounds 1–11 were evaluated against MMPs under a variety of conditions. Recombinant MMP and cell culture inhibition assays confirm that these compounds are effective inhibitors of MMPs that likely bind to the active-site zinc(II) ion. Cytotoxicity assays performed on primary cultures of CFs show these chelators are essentially nontoxic at high micromolar concentrations. Disulfide-exchange studies with DTNB found that the O,S donor ligands react with DTNB to different degrees, but all of the heterocyclic compounds are far less reactive than simple thiol-based ligands. The O,S donor ligands can inhibit cellular invasion through an ECM membrane in vitro. In summary, the experiments presented here support our initial report that potent, nontoxic, and biocompatible alternatives to the hydroxamic acid chelator are available for use in MPIs. Synthetic efforts to prepare complete MPIs based on these new compounds will exploit the vast array of peptidomimetic backbones already studied for hydroxamate-based inhibitors. Acknowledgements This work was supported by the University of California, San Diego, a Chris and Warren Hellman Faculty Scholar award (S.M.C.), an award from the American Heart Association (S.M.C.), a Pilot Project Grant from the Rebecca and John Moores U.C.S.D. Cancer Center, a Cottrell Scholar Award from the Research Corporation (S.M.C.), and N.I.H. grants HL43617 (F.J.V.) and HL-07444 (R.G.). J.A.L. was supported in part by a GAANN fellowship (GM-602020-03), an ARCS award, and a U.C. TSR&TP Fellowship.

References 1. Whittaker M, Floyd CD, Brown P, Gearing AJH (1999) Chem Rev 99:2735–2776

2. Massova I, Kotra LP, Fridman R, Mobashery S (1998) FASEB J 12:1075–1095 3. Nagase H, Woessner JF Jr (1999) J Biol Chem 274:21491– 21494 4. Skiles JW, Gonnella NC, Jeng AY (2004) Curr Med Chem 11:2911–2977 5. Lindsey ML (2004) Heart Fail Rev 9:7–19 6. Coussens LM, Fingleton B, Matrisian LM (2002) Science 295:2387–2392 7. Puerta DT, Cohen SM (2004) Curr Top Med Chem 4:1551– 1573 8. Overall CM, Lo´pez-Otı´ n C (2002) Nat Rev Cancer 2:657–672 9. Breuer E, Frant J, Reich R (2005) Expert Opin Ther Pat 15:253–269 10. Singh J, Conzentino P, Cundy K, Gainor JA, Gilliam CL, Gordon TD, Johnson JA, Morgan BA, Schneider ED, Wahl RC, Whipple DA (1995) Bioorg Med Chem Lett 5:337–342 11. Marmion CJ, Murphy T, Docherty JR, Nolan KB (2000) Chem Commun 1153–1154 12. Puerta DT, Cohen SM (2002) Inorg Chem 41:5075–5082 13. Puerta DT, Cohen SM (2003) Inorg Chem 42:3423–3430 14. Jacobsen FE, Cohen SM (2004) Inorg Chem 43:3038–3047 15. Puerta DT, Lewis JA, Cohen SM (2004) J Am Chem Soc 126:8388–8389 16. Lewis JA, Cohen SM (2004) Inorg Chem 43:6534–6536 17. Lewis JA, Puerta DT, Cohen SM (2003) Inorg Chem 42:7455– 7459 18. Monga V, Patrick BO, Orvig C (2005) Inorg Chem 44:2666– 2677 19. Villarreal FJ, Kim NN, Ungab GD, Printz MP, Dillmann WH (1993) Circulation 88:2849–2861 20. Yu L, Dennis EA (1991) Methods Enzymol 197:65–75 21. Hajduk PJ, Sheppard G, Nettesheim DG, Olejniczak ET, Shuker SB, Meadows RP, Steinman DH, Carrerea Jr GM, Marcotte PA, Severin J, Walter K, Smith H, Gubbins E, Simmer R, Holzman TF, Morgan DW, Davidsen SK, Summers JB, Fesik SW (1997) J Am Chem Soc 119:5818–5827 22. Hajduk PJ, Shuker SB, Nettesheim DG, Craig R, Augeri DJ, Betebenner D, Albert DH, Guo Y, Meadows RP, Xu L, Michaelides M, Davidsen SK, Fesik SW (2002) J Med Chem 45:5628–5639 23. Fray MJ, Burslem MF, Dickinson RP (2001) Bioorg Med Chem Lett 11:567–570 24. Johnson LL, Pavlovsky AG, Johnson AR, Janowicz JA, Man C-F, Ortwine DF, Purchase CF II, White AD, Hupe DJ (2000) J Biol Chem 275:11026–11033 25. Liu ZD, Hider RC (2002) Coord Chem Rev 232:151–171 26. Gorden AEV, Xu J, Raymond KN, Durbin P (2003) Chem Rev 103:4207–4282 27. Albert A, Rees CW, Tomlinson AJH (1956) Br J Exp Pathol 37:500–511 28. Lewis JA, Tran BL, Puerta DT, Cohen SM (2005) Dalton Trans 15:2588–2596 29. Xie Z, Singh M, Singh K (2004) J Biol Chem 279:39513–39519 30. Tyagi SC, Lewis K, Pikes D, Marcello A, Mujumdar VS, Smiley LM, Moore CK (1998) J Cell Physiol 176:374–382 31. Festuccia C, Dolo V, Guerra F, Violini S, Muzi P, Pavan A, Bologna M (1998) Clin Exp Metastasis 16:513–528 32. Villarreal FJ, Griffin M, Omens J, Dillmann W, Nguyen J, Covell J (2003) Circulation 108:1487–1492 33. Migdalof BH, Antonaccio MJ, McKinstry DN, Singhvi SM, Lan S-J, Egli P, Kripalani KJ (1984) Drug Metab Rev 15:841– 869 34. Farkas E, Katz Y, Bhusare S, Reich R, Ro¨schenthaler G-V, Ko¨nigsmann M, Breuer E (2004) J Biol Inorg Chem 9:307–315