Virtual screening identification and chemical optimization of substituted 2-arylbenzimidazoles as new non-zinc-binding MMP-2 inhibitors

Article abstract of DOI:10.1016/j.bmc.2019.115257

Matrix metalloproteinases (MMPs) are a large family of zinc-dependent endoproteases known to exert multiple regulatory roles in tumor progression and invasiveness. This encouraged over the years the approach of MMP, and particularly MMP-2, targeting for anticancer treatment. Early generations of MMP inhibitors, based on aspecific zinc binding groups (ZBGs) assembled on (pseudo)peptide scaffolds, have been discontinued due to the clinical emergence of toxicity and further drawbacks, giving the way to inhibitors with alternative zinc-chelator moieties or not binding the catalytic zinc ion. In the present paper, we continue the search for new non-zinc binding MMP-2 inhibitors: exploiting previously identified compounds, a virtual screening (VS) campaign was carried out and led to the identification of a new class of ligands. The structure-activity relationship (SAR) of the benzimidazole scaffold was explored by synthesis of several analogues whose inhibition activity was tested with enzyme inhibition assays. By performing the molecular simplification approach, we disclosed different sets of single-digit micromolar inhibitors of MMP-2, with up to a ten-fold increase in inhibitory activity and ameliorated selectivity towards off-target MMP-8, compared to selected lead compound. Molecular dynamics calculations conducted on complexes of MMP-2 with docked privileged structures confirmed that analyzed inhibitors avoid targeting the zinc ion and dip inside the S1′ pocket. Present results provide a further enrichment of our insights for the design of novel MMP-2 selective inhibitors.

Full text of DOI:10.1016/j.bmc.2019.115257

Journal Pre-proofs
Virtual screening identification and chemical optimization of substituted 2-ar-
ylbenzimidazoles as new non-zinc-binding MMP-2 inhibitors
Antonio Laghezza, Grazia Luisi, Alessia Caradonna, Antonella Di Pizio, Luca
Piemontese, Fulvio Loiodice, Mariangela Agamennone, Paolo Tortorella
PII:
S0968-0896(19)31474-9
https://doi.org/10.1016/j.bmc.2019.115257
BMC 115257
DOI:
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
3 September 2019
25 November 2019
6 December 2019
Please cite this article as: A. Laghezza, G. Luisi, A. Caradonna, A. Di Pizio, L. Piemontese, F. Loiodice, M.
Agamennone, P. Tortorella, Virtual screening identification and chemical optimization of substituted 2-
arylbenzimidazoles as new non-zinc-binding MMP-2 inhibitors, Bioorganic & Medicinal Chemistry (2019), doi:
https://doi.org/10.1016/j.bmc.2019.115257
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
©
2019 Published by Elsevier Ltd.

VIRTUAL SCREENING IDENTIFICATION AND CHEMICAL OPTIMIZATION
OF SUBSTITUTED 2-ARYLBENZIMIDAZOLES AS NEW
NON-ZINC-BINDING MMP-2 INHIBITORS
b§
a§
b
a1
b
Antonio Laghezza , Grazia Luisi , Alessia Caradonna , Antonella Di Pizio , Luca Piemontese , Fulvio
b
a*
b*
Loiodice , Mariangela Agamennone , Paolo Tortorella
aDepartment of Pharmacy, "G. d'Annunzio" University of Chieti-Pescara, Chieti, Italy
bDepartment of Pharmacy and Pharmaceutical Sciences, "A. Moro" University of Bari, Bari, Italy
1Present address: Section In Silico Biology & Machine Learning, Leibniz-Institute for Food Systems
Biology at the Technical University of Munich, Freising, Germany
*
Corresponding Authors
E-mail address: m.agamennone@unich.it (M. Agamennone), paolo.tortorella@uniba.it (P. Tortorella)
Abstract
Matrix metalloproteinases (MMPs) are a large family of zinc-dependent endoproteases known to exert
multiple regulatory roles in tumor progression and invasiveness. This encouraged over the years the
approach of MMP, and particularly MMP-2, targeting for anticancer treatment. Early generations of MMP
inhibitors, based on aspecific zinc binding groups (ZBGs) assembled on (pseudo)peptide scaffolds, have
been discontinued due to the clinical emergence of toxicity and further drawbacks, giving the way to
inhibitors with alternative zinc-chelator moieties or not binding the catalytic zinc ion.
In the present paper, we continue the search for new non-zinc binding MMP-2 inhibitors: exploiting
previously identified compounds, a virtual screening (VS) campaign was carried out and led to the
identification of a new class of ligands. The structure-activity relationship (SAR) of the benzimidazole
scaffold was explored by synthesis of several analogues whose inhibition activity was tested with enzyme
inhibition assays. By performing the molecular simplification approach, we disclosed different sets of
single-digit micromolar inhibitors of MMP-2, with up to a ten-fold increase in inhibitory activity and
ameliorated selectivity towards off-target MMP-8, compared to selected lead compound. Molecular
1

dynamics calculations conducted on complexes of MMP-2 with docked privileged structures confirmed
that analyzed inhibitors avoid targeting the zinc ion and dip inside the S1’ pocket. Present results provide
a further enrichment of our insights for the design of novel MMP-2 selective inhibitors.
Keywords: cancer; matrix metalloproteases; MMP-2; selectivity; non-zinc-binding inhibitors;
arylbenzimidazoles.
2

1
. Introduction
Cancer remains a leading cause of morbidity and mortality worldwide, despite the outstanding progress
in diagnosis and clinical protocols for treatment. Metastatic tumors, in particular, are highly incurable
and account for about 90% of fatalities in oncologic patients [1].
Tumor cell dissemination at distant sites is a complex, multistep process, concisely characterized by
tissue invasion, pervasive angiogenesis and avoidance of immune and/or apoptotic destruction. The
ability of malignant cells to intravasate into blood and lymphatic vessels, and then to extravasate into the
target tissues, implies the demolition of the main physiological barrier to metastasis, namely the
extracellular matrix (ECM), a dynamic macromolecular structure which, providing both mechanical and
functional sustain to the intracellular spaces, exerts a fine control on cell growth, migration and
differentiation, and tissue plasticity [2].
It is not surprising, then, that an ongoing attention is focused on matrix metalloproteinases (MMPs),
since members of this multifunctional family of calcium- and zinc-dependent ECM disrupting
endopeptidases, have been long recognized as major players in tumor genesis and aggressiveness [3].
As deputed executors of the proteolytic degradation of ECM components (collagen, gelatin, elastin,
fibronectin, and proteoglycans) and non-ECM substrates (growth factors, chemokines and cytokines),
MMPs are involved in several physiological processes. To avoid excessive tissue degradation, MMP
expression and proteolytic activity are finely regulated at the gene transcriptional level and through
zymogen activation and inhibition by endogenous modulators (TIMPs, 2-macroglobulin, and
endostatin).
Although the mechanisms mediating dysregulation of these proteases remain unclear, it is widely
accepted that abnormal MMP expression and/or activity is strongly implied in the development and
progression of metastatic cancer, [4] as well as autoimmune, chronic- [5] and neuro-inflammatory
syndromes [6], cardiovascular pathologies [7], viral and bacterial infections, and neurological disorders
[8], among all Alzheimer’s disease [9].
As a result, the recognition of MMPs as highly attractive targets for anti-cancer intervention has
produced an outstanding advance in the comprehension of their structural, mechanistic, and functional
roles, in an effort to assist the design of synthetic inhibitors tailored for these peptidases [10].
Currently, 28 isoforms of MMPs have been identified in mammals as mainly secreted and
membrane-anchored enzymes, and they have been classified into sub-families according to their substrate
specificity, primary structures and cellular localization.
3

Humans express 23 MMPs, with no fewer than 14 found in the vasculature. The most commonly
dysregulated isoforms in cancer, highly correlated to metastatic potential, are the 72 kDa MMP-2 and
the 92 kDa MMP-9 (type IV collagenases or gelatinases A and B, respectively), which therefore represent
validated goals for therapeutic purposes as well as prospective biomarkers for specific tumors. Because
of its involvement in the breast-to-bone metastatic cascade [11], MMP-2 keeps on being the most
attractive target for inhibition.
Despite wide substrate variability, MMPs display a high degree of structural homology, in particular
for the catalytic domain, containing one functional zinc. The main structural difference between the
family members regards the hydrophobic S1’ cavity and its surrounding -loop that, showing variability
in shape, amino acid sequence, and dynamic behavior, represents the major determinant for MMP
substrate recognition and selective binding to potential inhibitors [12, 13].
The tight requirement of the zinc ion for catalysis and the presence of diverse substrate binding
subsites in the protein have been exploited for the design of the first generations of MMP inhibitors
(MMPIs), which are invariantly built up by coupling a substrate- or structure-based peptide-to-
peptidomimetic backbone to a metal-chelating moiety, the so-called zinc-binding group (ZBG).
Although most of the ZBG-featuring MMPIs result extremely potent in vitro, their application in the
oncologic field has been disappointing to date, essentially due to an intrinsic lack of selectivity towards
those individual MMPs promoting cancer progression, leading to a loss of the protective effects of others
MMPs in tumor stages, as well as to the beneficial role of MMPs and generic metalloproteases in
connective tissue homeostasis [14-17].
As a consequence of their poor selectivity profile, most MMPIs produce dose-limiting toxicity,
which manifests mainly as the musculoskeletal syndrome (MSS) [18]. Further reasons are implied in the
marketing failure of these broad-spectrum inhibitors, including the fact that the clinical trials are
performed on patients with terminal-phase cancer, when inhibitors efficacy is reduced owing to several
overlapping pathways [19] the scarce tolerability and, particularly for MMPIs of the hydroxamate-type,
poor oral bioavailability [20]and loss of efficacy due to metabolic rearrangement of the NH-OH chelating
moiety.
To address these shortcomings, two main approaches have been followed in the search for more
selective MMPIs, characterized in addition by greater oral bioavailability and improved pharmacokinetic
profiles [21]: 1) the introduction of alternative ZBGs [22-26]; 2) the identification of a new class of
ligands that, by lacking the ZBG, are not able to bind the metal site. These latter were designed to take
4

advantage of the steric limitations of the S1’ specificity pocket to exploit new productive interactions for
selectivity. These inhibitors, identified for MMP-13 [27-32], MMP-8 [33] and MMP-12 [34], which
share high S1’ conformational flexibility [35], were shown to be equally potent to broad-spectrum
MMPIs, but much more selective.
The challenge of tuning the activity of disease-related MMPs by exploiting the poorly investigated
interactions in the S1’ site has led our group to carry on a virtual screening-based project to identify novel
non-zinc binding inhibitors (NZBIs) targeting gelatinase A, i.e., MMP-2 [36-39].
Our studies moved from the unprecedented observation that the S1’ binding pattern to a non-zinc
chelating inhibitor by MMP-8 is mostly conserved in the corresponding MMP-2/ligand complex. This
despite the fact that the shorter MMP-2 S1’ loop is unable to give the same conformational changes
induced, upon binding, at the bottom of the S1’ pockets in the MMP-8 and -13 isoforms, which are
thought to be crucial for efficient and selective inhibition [36, 38].
Starting from this probe, and guided by an integrated pharmacophore-docking based screening, we
previously identified micromolar inhibitors of MMP-2 belonging to the hydroxynaphtyridine and
hydroxyquinoline classes. Notably the docked derivatives were not found to bind the catalytic zinc [36].
Hence, the identified structures were submitted to a simplification study in order to disclose the structural
determinants for MMP-2 inhibition [38].
The relevance of the subject and the great potential connected to the identification on NZB MMP-2
inhibitors prompted us to go on with the project. Therefore, we decided to exploit previously obtained
results to setup a new virtual screening workflow aimed to optimize the search for new and selective
non-zinc binding MMP-2 inhibitors. Analogue testing and following synthesis confirmed the prospective
role of the benzimidazole scaffold to develop NZBIs of MMP-2.
2
. Results and discussion
2
.1 Virtual screening
As already stated, the design of NZBIs represents an effective strategy to identify useful MMPIs.
However, just a few MMP-2 inhibitors not binding the zinc ion have been disclosed so far, therefore, a
hierarchical virtual screening (VS) workflow has been setup exploiting previously obtained data about
NZBIs, in particular, considering the active compounds obtained in the first VS campaign and the
synthesized analogues.
5

6
10,000 molecules from Asinex were filtered to eliminate unsuitable compounds by applying the
REOS filter [40]; the 3D structures, stereoisomers, tautomers and protomers of the filtered library
490,000 compounds) were generated obtaining almost 1,300,000 resulting structures that were aligned
(
to the pharmacophore hypothesis obtained as described thereafter.
A ligand-based pharmacophore model was generated taking into account the common structural
features of the fifteen active compounds previously identified (Table S1 of Supporting Information) [36,
3
8]. The best pharmacophore hypothesis was composed of five features ADDRR mapping the two
aromatic functions (R), the acceptor (A) on the ureic CO group and the two hydrogen-bond donor NH of
the urea (D) (Figure 1). The obtained hypothesis showed a very good ability to discriminate among
active and inactive compounds (BEDROC - Boltzmann-Enhanced Discrimination of Receiver Operating
Characteristic = 1, PhaseScore = 1.6). This hypothesis was used to screen the previously mentioned
library, saving compounds that map at least 3 out of 5 features.
Figure 1. Pharmacophore hypothesis obtained by aligning the already identified NZBIs of MMP-2. Structure of the
compounds aligned to the pharmacophore hypothesis is represented as stick (N, O and H are colored on the basis of the atom
type, while carbon atoms are colored on the basis of the entry). The hypothesis is composed by five features: 1 hydrogen bond
(HB) acceptor (red sphere with arrows), 2 HB donors (cyan sphere with arrow), 2 aromatic rings (orange donuts).
The top 2,000 compounds with highest fitness values were submitted to docking simulations, following
a protocol already applied for the rationalization of previously studied inhibitors [38]. Visual inspection
of best ranked compounds brought to the selection of 20 virtual hits (Table S2 of Supporting
6

Information), taking into account the structural diversity and the presence of an aromatic function
interacting with the His201 imidazole ring (MMP-2 numbering). These compounds have been purchased
and submitted to the following biological evaluation.
2
.2.
Enzyme inhibition assays
The inhibitory activity of selected hits was evaluated in vitro by fluorometric assays using the
commercially available catalytic domain of MMP-2, -8, -9 and -13. A preliminary screening was
conducted on MMP-2, calculating the percentage of inhibition at 100 M (Figure 2). IC values were
5
0
then measured only for compounds showing the highest inhibition.
1
00
80
60
40
20
0
Figure 2. Percentage of inhibition of MMP-2 catalytic domain measured at 100 M of inhibitor concentration. Tested
compounds are reported with their supplier code (Table 2 of Supporting Information).
The new virtual screening run afforded one hit SYN 23017509 (1), structurally different from the
previously identified compounds, and with a micromolar activity against MMP-2 and MMP-13 (IC =72
5
0
±
2 and 93± 1 µM, respectively), while it resulted inactive against MMP-8. This compound represents
the first example of benzimidazole-based non-zinc binding MMP-2 inhibitor.
7

Figure 3. Structure of 1 analogues selected for enzyme inhibition assays.
2
.3.
Compound 1 series confirmation
To verify the validity of the identified scaffold, a substructure search in the Asinex library was carried
out considering the benzimidazole fragment as the scaffold and 11 analogues (2-12) were selected and
purchased for further testing (Figure 3).
The inhibitory activity of 1 analogues was evaluated measuring the percentage of enzyme inhibition at
1
00 M toward MMP-2, -8, -9, and -13. The IC values have been measured only for compounds 2 and
5
0
8
that showed the highest percentage of inhibition (all the other compounds presented a negligible
inhibition at 100 M). Obtained data have confirmed the role of benzimidazole scaffold for bioactivity
as 2 and 8 have shown an even better inhibition in comparison to 1; in particular, 2 displayed good
activity against MMP-2, and MMP-8, and -9, while it resulted inactive against MMP-13, while 8
confirms the interesting inhibitory profile of 1 sparing the off-target isoform MMP-8 and slightly
increasing the potency toward MMP-2 and MMP-13; however, 8 was found devoid of activity towards
MMP-9 (Table 1). Based on these preliminary activity results, and taking into account the major
8

chemical versatility for SAR elucidation of compound 2, and particularly its C ring side chain (see
below), compared to 8, our family of congeners was built up starting from inhibitor 2.
IC50 µM
Compound
MMP-2
MMP-8
MMP-9
MMP-13
1
2
72 ± 2
>100
>100
93 ± 1
31 ± 5.3
48 ± 2
32 ± 4
>100
26.6 ± 2.3
>100
> 100
8
52 ± 7
Table 1. MMP enzyme inhibition data measured for most active benzimidazole analogues.
2
.4.
Hit optimization
On the basis of these evidences, to have a detailed SAR for this class of inhibitors, we planned to
synthesize derivatives of 2 exploring: i) the substitution on the benzimidazole ring A; ii) the nature and
position of the substituent on the central phenyl nucleus (C); iii) the substitution of the benzimidazole
nucleus with benzoxazole and benzothiazole; iv) the replacement of the phenyl ring (C) with a furan
(Figure 4).
Figure 4. Structural modifications realized on 2 moieties.
9

Chemistry
Synthesis of 2-arylbenzimidazole, 2-arylbenzoxazole and 2-arylbenzothiazole derivatives
Synthesis of N-(benzothiazol-2-yl)-4-nitrobenzamides
Commercially available 1,2-diaminobenzenes were acylated with opportune acyl chlorides and then
easily cyclized under acid conditions to the corresponding 2-arylbenzimidazoles 33, 34, 37-42, 55 and
5
6 (Scheme 1).
Scheme 1. Reagents and conditions: a) i. pyridine, reflux, overnight. ii. p-TsA, xylene, reflux, overnight; b) DMSO, reflux,
h; c) 1N NaOH, THF, rt, 4h; d) EDC, NEt , 2-phenylethylamine, dry DMF, N , rt, 22h; e) CH I, K CO , dry DMF, overnight;
f) Fe, 6N HCl, EtOH, reflux, 3-5h; g) NEt , RCOCl or Ph-SO Cl, dry THF, rt, N , 3-12h; h) Cs CO , (t-Boc) O, MeCN, rt, 4-
1
3
2
3
2
3
3
2
2
2
3
2
1
2h; i) H
2
, 5 or 10% Pd/C, EtOH/EtOAc, 3 bar, rt, 12-24h; l) TFA, dry DCM, rt, 2-12h;. i)
2
Depending on R substitution, the 2-arylbenzimidazole series was split into not-nitrated (33-35, 80,
and 13) and nitrated compounds (37-42, 55, and 56).
2
The small group of 2-arylbenzimidazoles with R ≠ NO included the esters methyl 4’-benzoate (34)
2
and its methyl 3’-benzoate analogue, which were subjected to mild alkaline hydrolysis to afford the
1
0

corresponding benzoic acid derivatives 35, as last product, and 80, respectively. Final condensation of
8
0 with 2-phenylethylamine in presence of EDC and NEt yielded inverted amide 13.
3
On the other side, 2-nitroarylbenzimidazoles 37-42, 55, and 56 were the precursors, via reduction
and subsequent acylation, of the most substantial set of candidates. 42 was also converted into its N1-
methylated 2-nitroarylbenzimidazole analog (43) by efficient alkylation with CH I/K CO in dry DMF.
3
2
3
Due to the poor stability of bromine atom to catalytic hydrogenation, the reduction of nitro aromatic
group for 6-bromo derivatives 37 and 38, as well as for 41 and 42, to give the corresponding aniline-
derivatives 36, 78, 79, and 44, was carried out in excellent to acceptable yields by treatment with iron
powder in 6N HCl (Scheme 1, route 1). Subsequent acylation using the opportune acyl chlorides and
NEt in anhydrous THF produced 17, 22, 23, 30-32. Unfortunately, this procedure gave rise to numerous
3
by-products and the desired compounds were isolated in low yields (16% for 32). Thus, for the synthesis
of remaining derivatives we decided to protect the imidazole nitrogen to avoid its involvement (Scheme
1
, route 2): Boc-protection was quantitatively introduced (57-60) and kept all through the sequential
steps, i.e. catalytic hydrogenation of nitro group to give 61-64, subsequent acylation with the required
acyl chlorides (65-77), till the final acidolytic removal, affording the desired N-deprotected products 14-
1
6, 18-21, 24-29, with good overall yields.
Benzene substitution of the benzimidazole scaffold gave rise to the 1,3-tautomers which could be
1
seen in some H NMR spectra of the 2-arylbenzimidazole derivatives. Equivalence and fast exchange of
1
the 4,7-protons and 5,6-protons accounted for the broad signals of the aromatic protons in the H NMR
spectra [41] On the basis of this evidence, being the position of 4/7 and 5/6 equivalent, the 2-
arylbenzimidazole derivatives are shown with the benzene substituents in position 6 and 7. This is not
valid for those compounds substituted on the benzimidazole nitrogen, as N-methylated 43, and all N-Boc
intermediates envisioned in route 2.
2
-(5-Nitrofuran-2-yl-)benzimidazole (45) was assembled as described in satisfactory yields and was
not further processed; it represents the only benzimidazole to be prepared featuring a furanyl heterocycle
as bioisosteric replacement of the 2-phenyl substituent (Scheme 2).
Scheme 2. Reagents and conditions: a) i. pyridine, reflux, overnight. ii. p-TsA, xylene, reflux, overnight.
1
1

A few benzoxazole and benzothiazole derivatives were considered as isosteric counterparts of the
benzimidazole scaffold. Starting from commercially available 2-aminophenols and 4-nitrobenzoyl
chloride, 2-(4’-nitroaryl)benzoxazole 46, 52 and 53 were prepared in the same conditions described for
benzimidazoles; 2-(4’-nitroaryl)benzothiazole 47 was instead obtained by refluxing 2-aminothiophenol
and 4-nitrobenzaldehyde in DMSO (Scheme 1). Owing to the lack of the reactive nitrogen, 2-
nitroarylbenzoxazole 53 and 2-nitroarylbenzothiazole 47 were processed to 54 and 49 through aniline
intermediates 81 and 48 by following route 1 conditions, i.e. iron-mediated reduction in acidic conditions
followed by smooth acylation.
In order to evaluate whether the phenyl appendage needed to be directly linked, via a C-C bond, to
the 2-position of the heteroaromatic nucleus for the inhibitory activity, or some fragment was tolerated
between the two rings, compounds 50 and 51, containing an interposed NHCO unit, were synthesized in
acceptable yields through acylation of corresponding 2-amminobenzothiazoles with p-nitrobenzoyl
chloride in the presence of NEt in dry THF (Scheme 3).
3
Scheme 3. Reagents and conditions: a) NEt , dry THF, rt, 3-12h.
3
2
.5.
Biochemical assays and SAR analysis
All synthesized compounds were tested in enzyme inhibition assays against MMP-2, -8, -9 and -13
Tables 2-5) and compared with lead compound 2. Inhibition on MMP-9 has been evaluated because of
(
its similarity to MMP-2.
Our first SAR investigation regarded those analogues most closely related to 2 in structure and size.
Thus, intermediate 2-aminoarylbenzimidazoles 36, 44, 78, 79, and N-(Boc) 2-aminoarylbenzimidazoles
6
1-64 opened the way to amidated derivatives 14-32 (Scheme 1). In order to better explore the
importance of nature and position of the substituent on the benzimidazole ring, the methyl group present
1
2

in the 6-position of 2 was replaced with a hydrogen atom (14) or moved to the position 7 (15). Obtained
derivatives were overall less active than 2 against tested MMPs, except for the conserved activity of 15
on MMP-8. A slight improvement in activity was observed after replacement of the methyl residue with
a bromine atom (17 vs 2) (Table 2).
It came to our attention that the 3'-phenylacetamide group in 6 (Figure 3) could be elongated of one
1
methylene unit to give inhibitor 14 (R = H), with little gain in potency against tested MMPs with respect
to 6. We were then intrigued in investigating the homologation approach for the phenylalkanoylamide
chain linked to the C ring, with the following results compared to parent 2 (Table 2):

the addition of a further CH group in the alkanoyl amide appendage positioned in 3’ did not alter
2
significantly the inhibitory activity against MMP-2, as shown for the 7-methyl substituted analog 16
in comparison to 15, although promoted a negligible increase in selectivity towards MMP-2 with
1
regard to MMP-8 (s = 0.7 vs. 0.5, Table 2); compound 20 (R = H) resulted, however, a more
efficient inhibitor than inferior homologue 14 against all MMPs. The opposite was observed by
moving the PhCH CH CH CONH chain from the 3’- to the 4’-position of the C ring, exemplified
2
2
2
1
by the sensible drop in activity for inhibitor 21 compared to 23 (R = H), as well for the 6-bromine
derivative 22 versus 32; despite this, 21 exhibited an acceptable selectivity towards the target
isoform;

more significantly, the truncation of the phenylpropionamide to the simpler benzamide chain led to
a noticeable gain in activity in comparison to 2 and its congeners towards MMP-2 and MMP-13.
1
The effect, which is already evident in the 3’-benzamide derivative 18 (R = H, see 14), is observed
1
1
also for the 4’-substituted inhibitors 24 (R = H, vs 23) and 27 (R = 6-CH , see 26), with the
3
1
exception of 31 (R = 6-Br) which is practically equipotent to 32 on MMP-2. Compounds 24, 27,
and 31 are de facto the first undersized derivatives exhibiting a one-digit micromolar inhibition on
both targets, and emerging selectivities, particularly 31 (s = 7.8). Compound 18 well exemplifies the
tolerance to bioisosteric replacement in different parts of the amide chain linked to the C ring; in
fact, it represents the efficient carba-analog of the pyridine-containing inhibitor 4 (Figure 3), which
was found practically devoid of activity against the whole MMP spectrum; this could be in relation
to the higher electron density of the pyridine nucleus compared to the 18 phenyl ring. Secondly, 18
amide bond was replaced with a sulfonamide junction, known to introduce electronic and
geometrical properties which may be peculiar to recognition and binding: the resulting surrogate 19
was found a weaker inhibitor towards MMP-2 and MMP-13 with respect to its amide counterpart,
1
3

thus evidencing the critical queries for binding affinity. Once the aryl-spacer length required for
optimal activity was settled, we explored the effect of the substitution on the distal aromatic ring:
curiously, the addition of a bromine atom in the 4-position resulted in a slightly less efficient
inhibitor (25 compared to 24), while the introduction of the likewise electron-withdrawing nitro
group in 4- (28 and 30) or 3-position (29) afforded an approximate two-fold increase in activity
compared to unsubstituted 27 and 31, and resulted in remarkably potent inhibition against MMP-2
and MMP-13 (IC50 values of 3.3±0.4/1.8±0.4, 3.1±0.3/3.9±1.6, and 3.9±0.9/2.2±0.2 M,
respectively), with 28 and 29 featuring good selectivity. It is interesting to note that benzimidazole
derivatives 28-30 exhibit equal ligand efficiency (LE) values (0.27).
1
4

Table 2. In vitro inhibitory activity, selectivity and ligand efficiencies for benzimidazole derivatives 13-32 compared to lead molecule 2, evaluated by fluorometric
assay using commercially available catalytic domain of MMP-2, -8, -9 and -13. IC50 µM values are reported as the mean ± SEM of at least 2 independent experiments
performed in triplicate.
MMP-8/
MMP-2 MMP-2* MMP-2*
LE
LLE
Compd
2
R1
R2
MMP-2
MMP-8
MMP-9 MMP-13
6-CH
6-CH
H
3
3
3'-NHCOCH
3'-CONHCH
3'-NHCOCH
2
2
2
CH
CH
CH
2
2
2
Ph
Ph
Ph
Ph
31±5
45±5
32±4
40±2
26.6±2.3
37.7±1.5
68±27
>100
>100
1.0
0.9
0.5
0.5
0.7
0.8
3.9
0.7
0.7
>1.0
2.2
3.4
4.3
5.9
3.0
5.3
7.7
5.4
0.23
0.22
0.21
0.21
0.21
0.24
0.28
0.23
0.22
0.20
0.22
0.25
0.30
0.28
0.23
0.28
0.27
0.27
-0.80
-0.54
-0.74
-1.12
-1.54
-0.77
0.54
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
87±26
64±18
60±2.5
16 ±1
41.9±1.4
29±6
>100
7-CH
7-CH
6-Br
H
3
3
3'-NHCOCH
3'-NHCOCH CH
3'-NHCOCH CH
3'-NHCOPh
3'-NHSO Ph
2
CH
2
60±29
>100
2
2
CH
2
Ph
40.9±0.4
38.6±2.8
91±56
2
2
Ph
12.5 ±1.5 14.5±0.5
73±5
11.9±2.7
65.4±3.0
47±7
46±16
43.2±1.4
34.6±2.6
>100
57±6
36±9
35±15
H
2
94±13
0.41
-0.92
-1.23
-1.50
-0.08
0.81
H
3'-NHCOCH
4'-NHCOCH
4'-NHCOCH
2
2
2
CH
CH
CH
2
2
2
CH
CH
CH
2
2
2
Ph
Ph
Ph
43±12
>100
81.2±1.3
50±3
H
96±36
31±1
6-Br
H
67±1.8
64±6
58.0±2.5
29±9
22.2±2.0
15.2±0.3
7.5±2.5
10.9±1.2
17.9±2.3
6.5±0.6
1.8±0.4
2.2±0.2
4'-NHCOCH
4'-NHCOPh
4'-NHCO-Ph-4-Br
4'-NHCOCH CH Ph
4'-NHCO-Ph
2
CH
2
Ph
19±3
H
6.5±2.5
9.4± 1.5
25.3±1.0
8.6±0.7
3.3± 0.4
3.9±0.9
28±17
16±10
41±16
60±9
H
55± 1
-0.11
-0.71
0.18
6-CH
6-CH
6-CH
6-CH
3
3
3
3
2
2
77±7
45.3±0.7
25.5±1.0
20.9±2.9
28.4±2.5
10.7±0.5
14.5±1.4
4'-NHCOPh-4-NO
4'-NHCOPh-3-NO
2
2
0.65
0.58
1
5

3
3
3
0
1
2
6-Br
6-Br
6-Br
4'-NHCOPh-4-NO
4'-NHCOPh
2
3.1±0.3
6.42±0.29
7.0±0.1
7.9±2.8
50±2
8±1
6.4 ±1.1
32 ±1.9
11.5±1.5
3.9±1.6
3.9±1.6
6.5±1.5
2.6
7.8
1.1
0.27
0.28
0.26
0.43
0.05
-0.42
2 2
4'-NHCOCH CH Ph
*
LE: the ligand efficiency of compounds was calculated by converting in G the IC50 value toward MMP-2 and then dividing the obtained value by the number
of ligand heavy atoms [42].
*LLE: lipophilic ligand efficiency was calculated by subtracting the LE to the calculated logP value [43].
*
1
6

Based on these initial results, the effective weight of the phenylalkanoylamide chain in improving
the interactions network inside the enzymatic cleft remained to be established; thus, the molecular
simplification strategy was carried to the utmost extremes, with the investigation of simpler
arylbenzimidazoles lacking this appendage; this further set comprised most of intermediates in the
chemical routes to our bulkier inhibitors, with some newly synthesized compound (Scheme 1); here the
2
importance of small R substituents directly linked to the C aromatic ring was ascertained (inhibitors 33-
4
4, Table 3).
1
7

Table 3. In vitro inhibitory activity, selectivity and ligand efficiencies for benzimidazole derivatives 33-45 evaluated by fluorometric assay using commercially
available catalytic domain of MMP-2, -8, -9 and -13. IC50 µM values are reported as the mean ± SEM of at least 2 independent experiments performed in triplicate.
MMP-8/
MMP-2
-
LE
MMP-2
-
LLE
MMP-2
-
Compd
R1
R2
X
MMP-2
MMP-8
MMP-9
MMP-13
3
3
3
3
3
3
3
4
4
4
4
4
4
3
4
5
6
7
8
9
0
1
2
3
4
5
6-Br
H
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NCH
NH
>100
9.0±3
>100
26.3±1.4
35±7
>100
20.7±1.2
27±4
>100
19±8.6
43±5
6-Br 4'-COOCH
3
2.9
1.1
2.6
1.1
-
0.35
0.33
0.40
0.37
-
0.99
1.95
1.78
1.07
-
6-Br
6-Br
6-Br
6-Br
6-CH
6-CH
H
4'-COOH
31.5 ±1.5
10.0±1.5
8.8±0.8
>100
4'-NH
4'-NO
3'-NO
4'-NO
3'-NO
4'-NO
4'-NO
4'-NO
4'-NH
2
2
2
2
2
2
2
2
2
26.0±2.5
10±4
21.7±2.0
10±0.1
>100
21±4
10.5±2.5
47±4
>100
3.5
1.9
0.8
6.0
>5.6
-
0.37
0.35
0.31
0.35
0.30
-
1.45
1.11
0.87
2.22
1.35
-
3
3
6.5±0.7
14±3.9
82±5
23±12
27±14.4
64±4
22±6
6.1±0.1
15.5±2.4
>100
36.6±0.1
56.5±0.5
23.3±10.5
>100
6-NO
5/6-NO
6-NH
2
4.2±0.2
18±12
>100
25±10
>100
12±7
2
3
14±9
2
>100
>100
>100
31±7
27±5
32±4
33±4
0.9
0.36
2.14
1
8

2
Although the unsubstituted derivative 33 (R = H) was found inactive against all the MMPs under
2
scrutiny, R replacement with a 4’-positioned acid group (35) led to no perturbing effects compared to 2,
and rather introduced inhibitory activity against MMP-13. The substitution with an ester (34) or amine
group (36) in 4’ even led to a meaningful increase of the inhibitory activity towards the target isozyme
MMP-2, and to an appreciable MMP-13 inhibition, compared to the lead compound. Out of the context
2
of R simplification but following this outcome, we considered 80, a 3’-carboxy analogue of 35, for
ammidation with 2-phenylethylamine, just to explore the amide inversion approach referred to 2
(Scheme 1). Unfortunately, the inverso-isomer at the NHCO junction 13 (Table 2) resulted a weaker
inhibitor towards the whole panel of MMPs, compared to the prototype.
The introduction of a nitro group in 4'-position provided the extremely efficient inhibitors 37 and
3
9, each approximately equipotent towards MMP-2 and MMP-13 (IC = 8.8±0.8/10.5±2.5 M, and
5
0
6
.5±0.7/6.1±0.1 M, respectively), whereas the positioning of the same group in 3' determined a loss of
the inhibitory potency against all tested MMPs, more relevant for the 6-bromine derivative 38 (see 37)
with respect to the 6-CH -substituted inhibitor 40 (vs. 39). The presence of two nitro groups, in both 6-
3
and 4'-positions (42), further increased the inhibitory activity against MMP-2 (IC = 4.2±0.2 M). It is
5
0
interesting to note that efficient inhibitors 36, 37, and 39 exhibit the highest values of LE (LE values of
.40, 0.37, and 0.37, respectively); they all share a small 4-NO -phenyl (4-NH -phenyl for inhibitor 36)
0
2
2
appendage directly linked to the benzimidazole nucleus, while in the three most efficient MMP-2
inhibitors (derivatives 28-30) a further carbamidophenyl moiety is interposed between the NO -phenyl
2
group and the benzimidazole; accordingly they share a far lower LE value (0.27) as to the aforementioned
inhibitors.
The reduction of both nitro groups in 6- and 4'-position (44 versus 42) plainly truncated compound
activity. Finally, the methylation of the 42 benzimidazole nitrogen to give product 43 as isomeric mixture
led to poorer inhibitory results towards MMP-2, associated with a full loss of activity toward MMP-8
and MMP-9. Hence the importance of having the free NH of benzimidazole moiety, which seems to be
involved as hydrogen bond donor (HD) effector in interactions important for the inhibitory activity of
related compounds.
Our further synthetic efforts concerned the replacement of the 4’-activated phenyl ring with the 2-
(5-nitrofuran-2-yl) bioisosteric nucleus: the change afforded the good inhibitor 45, equipotent against the
whole MMP panel, which is however superior only to the 4’-NO -substituted arylbenzimidazole 41 (R1
2
=
H) towards MMP-2 and MMP-13 (Scheme 1, Table 3).
1
9

Furthermore, considering the role that the benzimidazole scaffold plays as a structural determinant
for binding affinity, we were tempted to investigate the effects of introducing alternative heteroaromatic
nuclei by bioisosteric replacement of nitrogen with O/S atoms on inhibitory activity. Thus, we decided
to prepare a small series of 2-arylbenzoxazole and 2-arylbenzothiazole derivatives corresponding to the
most efficient benzimidazole inhibitors (Scheme 1), for comparison of their inhibitory activity against
MMP-2, -8, -9 and -13 (Table 4). It deserves to be mentioned that the replacement of benzimidazole
with benzoxazole and benzothiazole prevents tautomer formation.
In the series of benzamide-linked 2-arylbenzoxazoles, replacing N (28) with O (54) resulted in an
almost five-fold decrease of inhibitory potency towards MMP-2, MMP-8, and MMP-13. For related 2-
arylbenzothiazoles, the substitution (49) afforded no advantages with respect to the benzimidazole
analogue (24).
By adopting the fruitful simplification approach, 2-arylbenzoxazoles 46, 52, and 53, and 2-
arylbenzothiazole 47 and 48, were prepared (Table 4). Noteworthy was the fact that 2-arylbenzoxazole
2
4
6 and 2-arylbenzothiazole 47, featuring a 4'-nitro substituent as R , resulted more potent than the
corresponding benzimidazole analogue 41 towards all MMPs, showing very low micromolar activities
1
against MMP-2 and MMP-13. Compound 52 (R = 6-CH ) was shown to maintain almost unchanged the
3
inhibitory efficacy of its bioisoster 39. Shifting the methyl group from the 6- to the 5-position of the
benzoxazole nucleus (isomer 53) had no consequences on activity. In the 2-arylbenzothiazole set, the
attempted replacement of the 4'-nitro group with an amino counterpart (48) had a detrimental effect on
inhibitory behaviour, interesting all MMPs.
2
0

Table 4. In vitro inhibitory activity, selectivity and ligand efficiencies for benzimidazole isosteres 46-49 and 52-54 evaluated by fluorometric assay using commercially
available catalytic domain of MMP-2, -8, -9 and -13. IC50 µM values are reported as the mean ± SEM of at least 2 independent experiments performed in triplicate.
MMP-8/
MMP-2 MMP-2 MMP-2
LE
LLE
Compd
R1
R2
X
MMP-2 MMP-8 MMP-9 MMP-13
4
4
4
4
5
5
5
6
7
8
9
2
3
4
H
H
4’-NO
4’-NO
2
2
O
S
5.8±0.1
6.5±1.5
>100
46±2
36±4
7.3±1.8
7.5±0.5
>100
7.9
1.6
-
0.40
0.40
-
1.96
1.11
-0.31
-0.44
1.21
1.25
-0.09
10.5±1.5 11.5±0.5
H
4'-NH
2
S
88±0.1
38±4
79 ±0.1
42±1
H
4'-NHCOPh
S
16.4±1.7
9±1
21±4
2.3
3.2
1.7
0.9
0.27
0.36
0.36
0.24
6-CH
5-CH
5-CH
3
3
3
4’-NO
4’-NO
2
2
O
O
O
29 ±16
18±5
9.4±2.4
8.8±2.2
11±3
9.7±2.5
16.4±5.2 14.9±3.5
4'-NHCOPh-4-NO
2
15.9±6.2 14.7±1.7 20±4
2
1

As final investigation, in the benzothiazole series we examined the effects of removing the C phenyl
ring by direct insertion of the 4-nitrobenzamide chain in position 2 of the heteroaromatic scaffold
(Scheme 3 and Table 5): the general loss of activity, as compared to 49, observed for resulting
derivatives 50 and 51 (with this latter conserving only a residual activity against MMP-13) would seem
to disclose a pivotal role for the central aromatic nucleus in the binding process.
Table 5. In vitro inhibitory (expressed as M IC50 values) activity of 50 and 51.
Compd
MMP-2
MMP-8
MMP-9
MMP-13
5
0
>100
>100
>100
>100
5
1
>100
>100
>100
20±0.1
Taken together, these results indicated that the benzoxazole and benzothiazole scaffolds offer
advantage in terms of activity with respect to the benzimidazole moiety only in the case of compounds
4
6 and 47.
Overall, the following considerations on this new series of compounds can be pointed out:


2-Aryl (or heteroaryl)-substituted benzimidazole or related heteroaromatic scaffolds are essential
for inhibitory activity.
1
2
Compounds with R and R substituents in position 6 and 4' respectively are better inhibitors
against all MMPs under scrutiny, and in particular towards MMP-2, in comparison to positional
isomers or unsubstituted analogues.

As a rule, the presence of electron acceptor groups has favourable effects on the inhibitory
potency of derivatives.
1



Nitro and bromine are the best R groups in terms of increase of inhibitory activity.
2
3- or 4-nitrobenzamide chain and the nitro group are the preferred substituents in R position.
Nitro groups in position 6 and 4' result in high activity against MMP-2, which is drastically
lowered upon reduction of both substituents.
2
2


Replacement of the benzamide moiety of the side chain with a benzenesulfonamide mimic leads
to a decrease in inhibition towards MMP-2 and MMP-13.
In our hit optimization process, the control of candidates’ molecular size and lipophilicity was
corroborated by Ligand Efficiency (LE), Lipophilic Ligand Efficiency (LLE), and related parameters
reported in extenso in Tables 2-4 for the different MMP isoforms under study. In this regard, there is a
growing consensus on the reliability of LLE metric for structural and pharmacokinetical drug
optimization processes [43].
Starting from the non-zinc-binding MMP inhibitor 2 and its phenylpropionamide-containing
congeners, we successfully performed a molecular simplification approach, leading to the selection of
different sets of candidates for further progression, with up to a ten-fold increase in MMP-2 inhibitory
activity and selectivity towards off-target MMP-8, compared to the lead compound. This is surprising,
since quite often the strategy for ligand potency implementation requires an increase in molecular weight
to enhance contacts inside the protein; the only oversized 2 analogue displaying single-digit micromolar
inhibition against MMP-2 is 32, but in the case some aspecific hydrophobic effects are thought to play a
role in binding affinity, as suggested by the elevated clogP value.
The most potent inhibitors 28-30 were obtained through reduction of the phenylpropionamide chain
by two methylene units, with an appreciable decrease in lipophilicity compared to 2; due to substituent
optimization, the number of heavy atoms increases in 28-30 but, despite this, ligand efficiencies result
higher (LE = 0.27) than that of the lead compound (LE = 0.23). The three benzimidazole derivatives are
equipotent against MMP-2 and exhibit equal LE values; it is interesting to note that, although 28 and 29
share the same clogP value being 4/3-NO positional isomers, they slightly differ in their lipophilic
2
efficiencies, ranging from LLE = 0.65 for the most selective 28 (s = 7.7) to LLE = 0.58 for 29 (s = 5.4),
thus suggesting for 28 the implication of more specific contacts inside the target cleft. We can tentatively
argue that the MMP-2 S1’ pocket is more favourably involved, through a suitably oriented charged site,
in interactions with a 4- (in spite of 3-) positioned nitro group on the benzamide moiety.
Derivatives 34, 36, 37, 39, and 42 are among the most potent inhibitors obtained after complete
removal of the benzamide appendage linked to the C ring; this change determined a significant drop in
molecular weight and lipophilicity, but more importantly the loss of the NHCO polar unit and the related
hydrogen donor/acceptor system. Since these compounds are equipotent with analogues 28-30, it could
be proposed that truncation of the benzamide fragment has no effects on the factors driving the affinity
2
3

of the ligand for the protein. The privileged structures, featuring the highest LLE and LE values, are
represented by 42 (LLE = 2.22, LE = 0.35, s = 6.0) and 36 (LLE = 1.78, LE = 0.40). For 36 the reduction
in lipophilicity is partly due to the presence of the polar NH substituent, which is most probably solvent-
2
exposed and consequently engaged in interactions unrelated to the identity of the binding site, as
supported by its lack of selectivity (s = 2.6).
A final set of single-digit micromolar inhibitors of MMP-2 was disclosed by following the
bioisosteric replacement approach, which led to benzoxazole (46, 52 and 53) or benzothiazole (47)
congeners. Particularly arylbenzoxazole 46 deserves to be selected for its properties (LLE = 1.96, LE =
0
.40); 46 represents a good inhibitor also in terms of selectivity (s = 7.9).
Unfortunately, efforts to reach selectivity towards MMP-2 in relation to the off-target MMP-8
(Tables 2 and 3) were disappointing for the new inhibitors. In fact, only eight compounds among all
displayed appreciable selectivity, with s values ranging from 7.9 (46) to 5.3 (27).
2
.6.
Molecular Modeling
In order to achieve a better understanding of the binding process involving this class of NZBIs, an
accurate structure-based study has been carried out by docking all studied compounds in the MMP-2, -
8
, -9 and -13 binding site. Docking calculations have been carried out using Glide and applying a
semi-rigid docking protocol. The benzimidazole scaffold has two possible tautomers and, when
substituted, the two forms are equally probable, therefore both have been considered in the docking
calculations.
Generally, all compounds occupy the S1' site, as expected, with the central phenyl ring (Figure 5)
facing the His201 side chain and the benzimidazole occupying the distal portion of the S1' site. A rather
well conserved H-bond is formed between the benzimidazole NH and the Ala220 carbonyl oxygen. The
amide substituent moves toward the S2' site.
The methyl or Br substituent on the benzimidazole fits in the S1' site improving the hydrophobic
interaction. In some case the Br forms a contact with the Thr229 OH, while the methyl has positive
contacts with Leu197 and Phe232.
2
Moving the R substituent from 3' to 4' does not change activity toward MMP-2 and selectivity
substantially; in fact, while 3' amide NH forms a H-bond with Pro221 CO, the amide CO is not able to
reach the Leu164 and Ala165 NH. In the 4' substituted compounds the amide CO can form H-bonds with
residues on the top of the S1' site.
2
4

It is well known that docking calculations are not able to provide a comprehensive picture of the
binding process as it neglects the role of water molecules and the flexibility of the system; therefore,
Molecular Dynamics (MD) calculations have been carried out for the complexes of MMP-2 with docked
poses of 2, 28 and 42.
Obtained results confirm that ligands occupy the S1’ site and the main interaction involves the
His201 imidazole ring. Details about the MD simulations are reported in the SI.
Mostly conserved interactions have been monitored during the simulation and those retrieved in
more than the 30% of saved frames are reported in Figure 5.
For the ligand 2 the docked binding geometry is almost conserved with the - stacking with His201
side chain and the H-bond between the benzimidazole NH and the Ala220 CO as the most stable. The
phenylpropionamide portion is solvent exposed and moves toward the S2’ site but does not form any
specific interaction, as resulting from the RMSF graph.
Ligand 28, that presents a 4’substitution on the central phenyl ring, quickly reaches a stable binding
conformation. In fact, it shifts toward the bottom of the S1’ site forming a cation- interaction with the
Arg233 side chain and allowing the NO substituent to interact with His201 and Glu202 via a water
2
molecule. The benzimidazole ring loses its H-bond interaction but the amide NH is bound to Gly216 CO
through a water molecule.
Ligand 42 maintains its binding mode for a half of the simulation, largely changing its starting
position in the last 9 ns of simulation. The ligand moves slightly toward the distal portion of the S1’ site
with the nitro-substituents providing interactions with Glu202 and Ala165 mediated by water molecules,
the phenyl ring forms two -stacking interactions with His201 and Tyr223 side chains, while the
benzimidazole NH is H-bonded to Ala220 and Ile222. The benzimidazole NO substituent forms
2
productive contact with Arg233 and Thr229 OH at the bottom of the S1’ site.
Summarizing, 42 results the most promising starting point for a new series of ligands as its binding
mode demonstrates the role of key interactions between the arylbenzimidazole scaffold and the S1’ site
of MMP-2.
MD simulations, moreover, demonstrate that these ligands are not able to bind the catalytic zinc ion
suggesting the possibility to exploit the S1’ site to obtain selective binders.
2
5

2
6

Figure 5. On the left the 3D complexes of MMP-2 (grey cartoon and solid surface) in complex with ligand 2 (A, cyan C
atoms), 28 (C, orange C atoms), and 42 (E, green C atoms) as resulting from MD simulation (an average conformation is
represented). Ligands are reported as sticks and H-bonds are represented as yellow dashed lines. For sake of clarity, only
residues and water molecules at 4 Å from the ligand are represented as lines. On the right, the 2D representation of interactions
between MMP-2 and ligand 2 (B), 28 (D) and 42 (F) retrieved in more than 30% of saved MD frames are depicted.
The inhibition mechanism of non-zinc binding inhibitors has been investigated in particular with respect
to MMP-13 and results strongly relate to the possibility to exploit the diverse amino acid composition of
MMP specificity loop, increasing selectivity [28, 31]. In fact, the large variability of the S1’ loop
sequence among MMPs influences both shape and flexibility of this pocket, allowing the opening of an
accessory pocket in the case of MMP-13 and MMP-8. The occupancy of this transient pocket allowed to
obtain very selective inhibitors. On the other hand, the relevance of protein dynamics in the MMP
inhibition is a well-known issue: Fabre and coworkers collected several examples that underline the need
to account for target flexibility in the search for MMP inhibitors [35]. With respect to non-zinc binding
ligands, in particular, the role of dynamic effects seems to be particularly relevant even when no
conformational rearrangement of the specificity loop is possible. As an example, in a previous work we
attempted to rationalize activity data of two MMP-2 non-zinc binding inhibitors combining different
computational approaches [39]. In that study, results from MD calculations suggest that more active
ligand is able to stabilize the protein structure by increasing the number of protein intramolecular H-
bonds, thus reducing the specificity loop fluctuations. It can be envisioned a similar mechanism also for
the currently developed ligands. However, addressing this issue, i.e. designing ligands for a moving target,
still remains a difficult goal.
3
. Conclusions
In conclusion, starting from previously identified non-zinc-binding ligands, we setup a virtual screening
campaign that allowed the identification of new inhibitors characterized by the phenylbenzimidazole
scaffold. Starting from the selected 2 prototype, the synthesis of a large library of compounds brought us
to disclose more effective MMP-2 inhibitors belonging to the non-zinc-binding class through a structure-
based refinement process supported by LLE and LE correlation and computational studies.
2
7

Using MD simulations, we explored the binding process of most interesting ligands eliciting relevant
interactions between the identified scaffold, providing useful suggestions for next optimization.
4
4
. Experimental section
.1. Chemical methods
All reagents were purchased from Sigma Aldrich Chemicals (Milan, Italy) and were used without
purification. The reactions were monitored by TLC (silica gel, UV254) with UV light (short wave
ultraviolet 254 nm and long wave ultraviolet 365 nm). All anhydrous reactions were performed under
argon or nitrogen atmosphere. The column chromatography was performed using Fluka silica gel 60 Å
(63-200 μm) or silica gel Si 60 (40-63 μm). Mass spectra were recorded on a HP MS 6890-5973 GC/MSD
spectrometer, electron impact 70 eV, equipped with a HP ChemStation or with an Agilent 6530 Series
1
13
Accurate-Mass Quadrupole Time-of-FLIGHT (Q-TOF) LC/MS; H NMR and C NMR were recorded using
the suitable deuterated solvent on a Varian Mercury 300 or 500 NMR Spectrometer. Chemical shifts (δ) are
expressed as parts per million (ppm) and coupling constants (J) in Hertz (Hz). Melting points were determined
in open capillaries on a Gallenkamp electrothermal apparatus and are uncorrected. Microanalyses of solid
compounds were carried out with an Eurovector Euro EA 300 model analyser and were within ±0.4% of the
theoretical values.
General procedure for the preparation of 2-arylbenzimidazoles and 2-arylbenzoxazoles.
Acyl chloride (2 mmol) was added to a solution of 1,2-diaminobenzene or 2-aminophenol (1 mmol) in
anhydrous pyridine (20 ml) and the resulted mixture was heated to reflux overnight. The mixture was
poured into an ice bath and the precipitate obtained was collected, washed with water and used for the
next step without purification. The crude product was suspended in a mixture of p-toluene sulfonic acid
monohydrate (1.6-2.0 mmol) and xylene (50 ml) and refluxed overnight.
For the synthesis of benzimidazoles derivatives, the reaction mixture was partitioned between
EtOAc (100 ml) and 2 M HCl solution (100 ml) and the organic layer extracted twice with 2M HCl
solution (100 ml). The combined aqueous layers were alkalized with 1 M NaOH solution until basic pH
and extracted with EtOAc (3 X 100 ml). The combined organic layers were washed with brine, dried
over anhydrous Na SO , filtrated and evaporated in vacuo.
2
4
2
8

For the synthesis of benzoxazole derivatives and for ester 34, the reaction mixture was partitioned
between EtOAc (100 ml) and NaHCO solution (100 ml). The organic layer was washed with brine, dried
3
over anhydrous Na SO , filtered and evaporated in vacuo.
2
4
The products were opportunely purified by crystallization or column chromatography.
6
-bromo-2-phenyl-1H-benzo[d]imidazole (33): Brown solid, 34% yield; mp: 206-209 °C
1
(
CHCl /Hexane). H NMR (500 MHz, [D ] DMSO): δ = 3.00-3.60 (br, 1H, NH), 7.32 (dd, J = 8.79, J =
3
6
1
.76, 1H, aromatic), 7.49-7.56 (m, 4H, aromatics), 7.76 (s, 1H, aromatic), 8.15 (d, J = 6.44, 2H,
aromatics). GC-MS: m/z (%): 274 [M+ 2] (99), 272 [M] (100).
Methyl 4-(6-bromo-1H-benzo[d]imidazol-2-yl)benzoate (34): Brown solid, 52% yield; mp: 226-229
1
°
C (CHCl ). H NMR (500 MHz, [D ] DMSO): δ = 3.89 (s, 3H, CH ), 7.36-7.38 (dd, J = 8.32, J = 1.96,
3
6
3
1
8
3
H, aromatic), 7.59 (d, J = 8.32, 1H, aromatic), 7.82 (bs, 1H, aromatic), 8.12 (d, J = 8.56, 2H, aromatics),
.30 (d, J = 8.56, 2H, aromatic), 13.00 (bs, 1H, NH). GC-MS: m/z (%): 332 [M+2] (96), 330 [M] (100),
01 (37), 299 (33), 273 (14), 271 (15).
6
-bromo-2-(4-nitrophenyl)-1H-benzo[d]imidazole (37): Brown solid, 84% yield; mp: 254-249 °C
1
(
EtOAc/hexane). H NMR (500 MHz, [D ] DMSO): δ = 7.37-7.39 (dd, J = 8.81, J = 1.96, 1H, aromatic),
6
7
.61 (d, J = 8.81, 1H, aromatic), 7.84 (d, J = 1.96, 1H, aromatic), 8.39-8.43 (m, 4H, aromatics), 13.25
-
-
2
(
bs, 1H, NH). MS(ESI): m/z: 318 [M+2-H] , 316 [M-H] ; MS : m/z (%): 288 (100), 286 (96), 272 (26),
2
70 (26).
6
-bromo-2-(3-nitrophenyl)-1H-benzo[d]imidazole (38): Brown solid, 85% yield; mp: 220-222 °C
1
(
Et O). H NMR (500 MHz, [D ] DMSO): δ = 7.38 (dd, J = 8.32, J = 1.96, 1H, aromatic), 7.60 (d, J =
2
6
8
.81, 1H, aromatic), 7.83-7.87 (m, 2H, aromatics), 8.33-8.35 (m, 1H, aromatic), 8.60 (d, J = 8.81, 1H,
-
aromatic), 8.99 (t, J = 1.96, 1H, aromatic), 13.00 (bs, 1H, NH). MS(ESI): m/z: 318 [M+2-H] , 316 [M-
-
2
H] ; MS : m/z (%): 288 (80), 286 (67), 272 (72), 270 (38), 260 (100), 258 (62).
6
-methyl-2-(4-nitrophenyl)-1H-benzo[d]imidazole (39): yellow solid, 61% yield; mp: 205-207 °C
1
(
EtOAc). H NMR (300 MHz, CDCl ): δ = 2.49 (s, 3H, CH ), 7.15-7.46 (d, J = 8.2, 1H, aromatic), 7.42
3
3
(bs, 1H, aromatic), 7.57 (bs, 1H, aromatic), 8.20 (d, J = 8.79, 2H, aromatics), 8.30 (d, J = 8.79, 2H,
-
2
aromatics), 9.76-10.70 (br, 1H, NH). MS(ESI): m/z: 252 [M-H] ; MS : m/z (%): 222 (100), 206 (23).
76 [M + Na]+.
2
2
9