Amino Acid Derivatives as New Zinc Binding Groups for the Design of Selective Matrix Metalloproteinase Inhibitors

A number of matrix metalloproteinases (MMPs) are important medicinal targets for conditions ranging from rheumatoid arthritis to cardiomyopathy, periodontal disease, liver cirrhosis, multiple sclerosis, and cancer invasion and metastasis, where they showed to have a dual role, inhibiting or promoting important processes involved in the pathology. MMPs contain a zinc (II) ion in the protein active site. Small-molecule inhibitors of these metalloproteins are designed to bind directly to the active site metal ions. In an effort to devise new approaches to selective inhibitors, in this paper, we describe the synthesis and preliminary biological evaluation of amino acid derivatives as new zinc binding groups (ZBGs). The incorporation of selected metal-binding functions in more complex biphenyl sulfonamide moieties allowed the identification of one compound able to interact selectively with different MMP enzymatic isoforms.


Introduction
Matrix metalloproteinases (MMPs) are 23-member zincdependent endopeptidases family involved in the extracellular matrix turnover [1]. Their aberrant regulation has been implicated in tumoral process, where they showed to have a dual role inhibiting or promoting cell growth and survival, angiogenesis and metastasis [2,3] differentiation [4], and inflammation and immune surveillance [5]. Moreover, MMPs are overexpressed in a variety of tumor types, and their overexpression is associated with tumor aggressiveness and poor prognosis [6]. The specific alteration of the MMPs in malignant tissues and their participation in some of the major oncogenic mechanisms have both fuelled interest in the design and evaluation of MMP inhibitors (MMPIs) as anticancer agents [7,8]. Generally, the MMPIs design involves peptide or peptidomimetic backbones containing a zinc-binding group (ZBG) able to interact with both the subpockets surrounding the active site (S 1 and S 1 , S 2 , and S 3 ) and the zinc (II) ion present in the catalytic site, respectively [9,10]. The greater part of MMPIs research has focused on developing the peptide or peptidomimetic containing a hydroxamic acid as chelating group. Although this design has produced potent inhibitors such as Batimastat [11,12] and Marimastat [13] (Figure 1), none of these MMPIs has successfully completed clinical trials.
The inability of hydroxamates to produce clinically viable compounds has been attributed to low oral availability, poor in vivo stability, and undesirable side effects associated with these compounds [14]. This has prompted the investigation of a limited number of nonhydroxamate-based MMPIs [15][16][17][18][19]. We present herein the results obtained with a small library of compounds synthesized and tested as potential ZBGs. The compounds were selected on the basis of some similarities to hydroxamates, such as the possibility to form five-member chelates ( Figure 2), but with potentially enhanced pharmacokinetic properties such as a better hydrolytic stability and/or proposed increased affinity for the MMP zinc (II). The designed ligands have a general 2-aminopropane-1,3-disubstituted structure which might be visualized as an amino acid derivative with the carbon atom connected through two carbons to heteroatoms with lone pairs or simply electron availability (R and R 1 ). These functional groups are sulfhydryl (SH), alcohol (OH), imidazole, cyano (CN), and azide (N 3 ) which are able to interact as Lewis-base in the coordination of the catalytic zinc ion. Their symmetric and asymmetric combination gave rise to a small ZBGs library ( Table 1). The two carbons rotational freedom could allow the chelating groups R and R 1 to orient themselves as better as possible in direction of the zinc ion.
According to the preliminary results of enzymatic inhibition activities, we further synthesized, from the most interesting ligands, a small series of sulfonamide derivatives containing a phenoxyphenyl group. This moiety has been widely used in the design of MMPs inhibitors as side chain of choice able to interact with the enzymatic S 1 subsite which plays a pivotal role in the determination of inhibition selectivity [20,21]. The aims of the current study were to screen a range of nonhydroxamate structures as new ZBGs and to evaluate the enzymatic activity of small molecules designed to interact with the subpocket S 1 and with the zinc (II) ion present in the catalytic site of MMPs.

Chemistry
The symmetric ligands were prepared starting from serinol (1a) according to synthetic route shown in Scheme 1. After N-Boc-protection, the alcohol groups of 2 were activated as ditosylate derivatives in order to undergo nucleophilic substitution with azide and nitrile salts. Thus, displacement of the OTs group with tetraethylammonium cyanide (TEACN) or sodium azide (NaN 3 ) in DMF using TEA as base led to 4 and 5, respectively, with 80%-82% yields. The final symmetric derivatives 1b and 1c have been obtained after deprotection of 2-amino group using a solution of 25% TFA in dichloromethane.
The ditosylation reaction was the limiting step in this synthetic strategy, described in the literature using pyridine (py) as solvent [22]. In our case, the treatment of 2 with 4toluenesulfonyl chloride in pyridine led to ditosylate derivative 3 in only 2% yield. A preliminary study of the influence of solvents, reaction time, and reactive/base concentration ratio on this reaction was performed in order to (a) improve yields and mono/ditosylate adduct ratio; (b) facilitate workup procedures; (c) use a less toxic solvent.  As shown in Table 2, treatment of 2 with Tos-Cl and TEA in 2.4 : 3 ratio gave the highest yields (85%) and better selectivity (1 : 19) in the formation of ditosylate derivative 3 using dry dichloromethane as solvent (entry 8). Pyridine or pyridine with dimethylaminopyridine as base catalyst gave low yields with a little percentile of dialkylation product (entries 1, 2, and 3), while DCM as solvent was more effective without base catalyst (entries 6, 7, and 8 versus entries 4 and 5).
The symmetric and asymmetric ligands, 1f and 1d, 1e, and 1g-1l, respectively, were prepared according to the synthetic route shown in Scheme 2.

Enzymatic Inhibition Assays
The synthesized ZBGs, compounds 1a-1l, were tested against the catalytic domain of MMP-2 in order to evaluate their chelating capability with respect to acetohydroxamic acid (AHA) which was considered a representative of the standard hydroxamate chelator. All the examined compounds exhibited a higher inhibitory activity compared to AHA ( Table 3). The enzymatic assays revealed that the most interesting compounds are the serinol 1a and the asymmetric ligands 1e and 1g, containing both an imidazole group and an alcohol or a thiol group, respectively. Surprisingly, the cysteinol 1d, despite the well-known zinc thiophilicity, showed a lower enzymatic activity with IC 50 value of 674 M. On the basis of these data, we selected the most active ligands 1a and 1e to be incorporated as ZBG in a more complex structure. The ligands were linked, through a sulfonamide bind, with a phenoxyphenyl group, described in the literature for its well-validated affinity for the S 1 enzymatic subpocket [10,24]. A third ligand 1d, less active, was also chosen in order to evaluate the real influence of ZBG group alone in the enzymatic activity.
These preliminary results showed a different behaviour of ZBGs when they are introduced into a more complex structure indicating that, in this case, the modulation of selectivity does not depend only on ZBGs [29].

Molecular Modeling
In order to rationalize the observed activity data, docking calculations of the ZBGs and compounds 25a, 25d, and 25e were performed on the MMP-2 catalytic domain. Subsequently, they were submitted to a refinement step, thorough minimization of best poses. The applied protocol allowed to correlate predicted and experimental binding energies. It is well known that docking scores hardly correlate with activity data, and to this aim, more accurate calculations are required such as Free Energy Perturbation or Thermodynamic Integration. Among available approaches, Linear Interaction Energy (LIE) represents a good compromise between accuracy and speed of calculations [30,31]. In this approach, the binding process is represented as the replacement of water molecules solvating a ligand by the protein, using an implicit water model. LIE generates a custom scoring function calculating the values of alpha, beta, and gamma coefficients of the following equation: Journal of Amino Acids  where Delta is the calculated binding energy; xxx is the van der Waals, Coulombic, and Cavity energy terms from the bound state; xxx is the van der Waals, Coulombic, and Cavity energy terms from the free state. LIE method applied to our ligands provided a statistically significant correlation between calculated and experimental data, underpinning the validity of predicted docking poses (Table 5). It is worth noting that chiral compounds under study were synthesized and tested as racemic mixture. Consequently, all calculations were carried out for all enantiomers, and quantitative models were generated for both R (Rmodel) and S forms (S-model) separately. Obtained Δ values indicate that the R-model works slightly better than the S-model in predicting activity, as demonstrated by statistical correlation values (Table 6); however, the S-model is able to predict the binding energy with acceptable approximation indicating that experimental activity can be due to the contribution of both enantiomers.
This result is confirmed from the analysis of fragments docking poses in fact that no relevant differences can be observed in the binding of enantiomeric forms of chiral compounds, in the MMP-2 active site.
Moreover, differently than expected, just ligand 1a is able to chelate the zinc ion, providing an explanation of the higher activity observed for this compound. Other fragments give a monodentate binding of the catalytic zinc, and the other electron donating group is usually involved in H-bond interactions with surrounding residues, such as the Pro221 carbonyl oxygen (e.g., 1d), except for compounds containing the imidazole ring (e.g., 1e), involved in a -stacking with the His201 side chain, which represents one of the main interactions formed by MMPIs in the S 1 pocket (Figure 3). This behavior can be attributed to the strict geometrical requirements, which must be fulfilled by chelating group around the zinc ion in MMPs active site.
The binding of sulfonamide derivatives 25a, 25d, and 25e was studied as well through docking calculation and subsequent refinement as previously described on MMP-1, -8, and -9 (Table 7). No statistical correlations are provided in these cases because of the few available data. Docking results show all ligands occupying the S 1 site, except for MMP-1. This isoform, in fact, is known for having a short S 1 pocket, unable to accommodate the large biphenylether portion of  these ligands. The imidazole ring of compounds 25e, the more active towards MMP-1, occupy the hydrophobic pocket of this protein.
Binding mode of sulfonamide derivatives to the other MMPs is well conserved, regardless of chirality: MMP-2, -8, and -9 have a deep S 1 site able to locate the hydrophobic biphenyl ether, whose proximal aromatic ring interacts with the imidazole ring of His201, and the distal ring provides hydrophobic interactions in binding pocket. The sulfonamide moiety provides two H-bonds between a sulfone oxygen and Ala165 and Leu164 NH (MMP-2 numbering) and the sulfonamide NH and the Pro221 CO or alternatively Ala165 CO ( Figure 4). Main differences are observed for the binding of the ZBG; in MMP-2, the ZBG of 25a maintains the ability to chelate the zinc ion.
This chelation, not observed in MMP-8 and -9, can explain the higher activity observed for this ligand in MMP-2.
MMP-8 and -9 zinc ions coordinate all ligands in a monodentate fashion with a similar geometry, similarly to what observed for the ZBG in MMP-2. Therefore, as no chelation is provided by the ZBG in MMP-8 and MMP-9, the zinc thiophilicity seems to play a relevant role in determining activity toward these isoforms.

Conclusion
Herein, we described the design, synthesis, inhibitory activity, and molecular modeling studies of new non-hydroxamatebased MMPIs. The adopted synthetic strategy enabled the setting-up of a small ZBGs library through a simple and easily accessible pool of reactions. The biological screening of this library led to the identification of two ZBGs that were incorporated in a more complex structure able to interact with the S 1 enzymatic site. The biological data for compounds 25a and 25e confirmed the inhibition trend of the respective ZBGs against MMP-2. Compound 25d, containing a less potent chelating group (1d versus 1a and 1e), was equipotent to 25a against MMP-2 and more potent than 25a against MMP-8 and MMP-9 (10-and 13-fold, resp.). Molecular modeling studies provided a rationalization of the experimental data, suggesting a putative binding mode of studied ligands in MMPs active site. These preliminary results indicate the importance of testing and selecting firstly compounds containing the minimums structural requirements necessary for a specific biological activity. Furthermore, taking in consideration the complex role of MMPs in the cellular and tumoral homeostasis, the development of selective inhibitors is desirable in order to shed further light on the protein function, signalling pathways, and role in disease of different MMPs [32][33][34]. Thus, compound 25d identified in this preliminary study as MMP-8 and MMP-9 inhibitors could be submitted to a rational process of hit optimization with the aim to improve its potency and selectivity of action. The introduction of these new fragments into different peptide structures with the aim to synthesize selective MMPs inhibitors and to explore their structure-activity relationships is currently under study in our laboratory.
Proenzymes were activated immediately prior to use withaminophenylmercuric acetate (APMA 2 mM for 1 h at 37 ∘ C for MMP-2 and MMP-8, APMA 2 mM for 2 h at 37 ∘ C for MMP-1, and APMA 1 mM for 1 h at 37 ∘ C for MMP-9). For assay measurements, the inhibitor stock solutions ( is the initial velocity in the absence of the inhibitor. Results were analyzed using SoftMax Pro software and Origin software.

General.
Reagents, starting materials, and solvents were purchased from commercial suppliers and used as received. Analytical TLC was performed on plates coated with a 0.25 mm layer of silica gel 60 F254 Merck and preparative TLC on 20 cm × 20 cm glass plates coated with a 0.5 mm layer of silica gel PF254Merck. Silica gel 60 (300-400 mesh, Merck) was used for flash chromatography. Melting points were determined by a Kofler apparatus and are uncorrected. 1 H NMR and 13 C NMR spectra were recorded with a Varian-400 spectrometer, operating at 400 and 100 MHz, respectively. Chemical shifts are reported in values (ppm) relative to internal Me 4 Si, and values are reported in hertz (Hz). ESIMS experiments were performed on an Applied Biosystems API 2000 triple-quadrupole spectrometer.

General Procedure for Synthesis of Amino Alcohols Derived from Amino Acids (9-11)
Ethyl chloroformate (1.2 eq) and N-methylmorfoline (1.2 eq) at 0 ∘ C were added to a solution of Boc-Cys(Trt)-OH (6) or Boc-Ser(OtBu)-OH (7) or flask Boc-His(Boc)-OH (8) (1 mmol) in THF (4 mL). After 1 h, the reaction was filtered off, and NaBH 4 (3 eq) dissolved in 2 mL of water was added. The reaction was then stirred at room temperature for 3 h, washed with H 3 O + , dried with Na 2 SO 4 , and evaporated 8 Journal of Amino Acids under reduced pressure. Chromatography purification of the corresponding residues using -hexane/AcOEt: 2/1 yielded, in each case, the amino alcohol derivatives.

General Procedure for Removal of the Boc Protecting Group: Synthesis of Final Ligands 1d and 1e
The compounds 9 or 11 were dissolved in a 1 : 1 mixture DCM/TFA (10 mL), adding triethylsilane (0.1 eq) as scavenger. The reaction was stirred for 2 h at room temperature and evaporated under reduced pressure to yield the title derivatives as TFA salt.

General Procedure for Synthesis of Tosilated Derivatives 12-14
To a 25 mL round-bottom flask, 9, 10, or 11 (1.1 mmol) was added and dissolved in dry DCM (10 mL). After reached 0 ∘ C, paratoluensulfonyl chloride (1.2 eq) and TEA (1.5 eq) were added. The reaction is stirred for 10 h, washed with water, dried with Na 2 SO 4 , and evaporated under reduced pressure. The crudes were then purified by chromatographic column using -hexane/AcOEt: 3/1 as elution system.

General Procedure for the Synthesis of Thio Derivatives 15 and 16
To a 25 mL round-bottom flask, 12 or 14 (1.1 mmol) was added, and dissolved in DMF (10 mL). TEA (1.5 eq) and Trt-SH (1.2 eq) were then added and the reaction was stirred for 10 h at room temperature. The reaction mixture was then washed with water, dried with Na 2 SO 4 , and evaporated under reduced pressure. The crudes were then purified by chromatographic column using TLC: -hexane/AcOEt: 4/1 as eluent system.

Synthesis of Final Ligands 1f and 1g
The compounds 15 or 16 were dissolved in a 1 : 1 mixture DCM/TFA (10 mL), adding triethylsilane (0.1 eq) as scavenger. The reaction was stirred for 2 h at room temperature and evaporated under reduced pressure to yield the title derivatives as TFA salt.

General Procedure for the Synthesis of Cyano Derivatives 17 and 18
To a 25 mL round-bottom flask, 12 or 13 (1.1 mmol) was added and dissolved in DMF (10 mL). TEA (1.5 eq) and TEACN (1.2 eq) were added, and the reaction was stirred for 10 h at room temperature. The reaction mixtures were then washed with water, dried with Na 2 SO 4 , and evaporated under reduced pressure. The crudes were purified by chromatographic column using -hexane/AcOEt: 3/1

Synthesis of Final Ligands 1h and 1i
The intermediates 17 and 18 were dissolved in a 1 : 1 mixture DCM/TFA (10 mL), adding triethylsilane (0.1 eq) as scavenger. The reaction was stirred for 2 h at room temperature and evaporated under reduced pressure to afford the title compounds as TFA salt.

General Procedure for the Synthesis of Azido Derivatives 19-21
To a 25 mL round-bottom flask, 12, 13, or 14 (1.1 mmol) were added and dissolved in DMF (10 mL). TEA (3 eq) and NaN 3 (2.4 eq) were added, and the reactions were stirred for 10 h at room temperature. The reaction mixtures were washed with water, dried with Na 2 SO 4 , and evaporated under reduced pressure. The crudes were then purified by chromatographic column using -hexane/AcOEt: 3/1 as eluent system.

Synthesis of Final Derivatives 1j-1l
The intermediates 19, 20, and 21 were dissolved in a 1 : 1 mixture DCM/TFA (10 mL), adding triethylsilane (0.1 eq) as scavenger. The reaction was stirred for 2 h at room temperature and evaporated under reduced pressure to afford the title compounds as TFA salt.