Discovery of the first potent and selective Mycobacterium tuberculosis Zmp1 inhibitor

Article abstract of DOI:10.1016/j.bmcl.2014.04.004

The Mycobacterium tuberculosis extracellular zinc metalloprotease 1 (Zmp1) has been proposed to play a key role in phagosome maturation and to enhance the survival of Mycobacterium tuberculosis in the host. Consequently, small molecule inhibitors of Zmp1

Full text of DOI:10.1016/j.bmcl.2014.04.004

Bioorganic & Medicinal Chemistry Letters 24 (2014) 2508–2511
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of the first potent and selective Mycobacterium
tuberculosis Zmp1 inhibitor
Mattia Mori a,b, Francesca Moraca a, Davide Deodato a, Davide M. Ferraris c, Petra Selchow d,
Peter Sander d,e, Menico Rizzi c, , Maurizio Botta a,f,
⇑
⇑
a Dipartimento di Biotecnologie, Chimica e Farmacia, University of Siena, Via Aldo Moro 2, 53100 Siena, Italy
b CLNS Istituto Italiano di Tecnologia@Sapienza, Viale Regina Elena 291, 00161 Roma, Italy
c Dipartimento di Scienze del Farmaco, University of Piemonte Orientale ‘A. Avogadro’, Viale Largo Donegani 2, 28100 Novara, Italy
d Institute of Medical Microbiology, University of Zurich, Gloriastrasse 32, CH-8006 Zurich, Switzerland
e National Reference Laboratory for Mycobacteria (NRLM), Gloriastrasse 32, CH-8006 Zurich, Switzerland
f Sbarro Institute for Cancer Research and Molecular Medicine, Center for Biotechnology, College of Science and Technology, Temple University,
BioLife Science Bldg., Suite 333, 1900 N 12th Street, Philadelphia, PA 19122, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The Mycobacterium tuberculosis extracellular zinc metalloprotease 1 (Zmp1) has been proposed to play a
key role in phagosome maturation and to enhance the survival of Mycobacterium tuberculosis in the host.
Consequently, small molecule inhibitors of Zmp1 are of pivotal importance as a tool to better understand
the pathogenicity of Zmp1 and as lead candidates for pharmacological intervention. Here we combined in
silico structure-based inhibitor design with biochemical studies to discover and characterize the first
potent competitive Zmp1 inhibitor showing a Ki of 94 nM and a high selectivity for Zmp1 with respect
to human Neprilysin.
Received 18 March 2014
Revised 31 March 2014
Accepted 2 April 2014
Available online 13 April 2014
Keywords:
Tuberculosis
Zmp1
Ó 2014 Elsevier Ltd. All rights reserved.
Virtual screening
Enzyme inhibitors
Metalloproteases
It is widely appreciated that novel pharmacological approaches
against tuberculosis (TB) are urgently needed and may be stimu-
lated by the characterization and validation of novel protein targets
that sustain tuberculosis pathogenicity. In this respect, the Mycobac-
terium tuberculosis (Mtb) zinc-dependent metalloprotease 1 (Zmp1),
a member of the M13 endopeptidase family,1,2 has been proposed to
play a key role during phagosome maturation and, therefore, can be
proposed as a potential therapeutic target against TB.
Following host infection, Mtb is phagocytized by macrophages
into the phagosome that, in turn, should undergo to a critical mat-
uration step which culminates with the formation of the phago-
lysosome that leads to Mtb clearance.3 However, Mtb has been
found to inhibit the phagosome maturation step, precluding mac-
rophage killing and remaining indefinitely in a latent non-virulent
form in infected persons. Mtb is then re-activated as soon as an
immunosuppressive condition occurs, thus preventing a definitive
eradication of the pathogen from the host, giving rise to TB
recurrences.4
Zmp1 has been proposed to impact on the phagosome matura-
tion by inhibiting caspase-1 activation and the processing of pro-
IL-1b into IL-1b.3 Indeed, the suppression of the zmp1 gene has
been found to reactivate caspase-1 as well as the production of
IL-1b, thus leading to phagosome maturation and the clearance
of the pathogen.5 A recent report by Muttucumaru et al. claims that
deletion of the zmp1 gene causes bacterial hypervirulence in a
murine model,6 contrarily to what was previously observed by
Master and colleagues, where zmp1 deletion led to virulence atten-
uation.3 Therefore, the precise biological role of Zmp1 in either Mtb
virulence or host immune response escape is still under debate and
has been not fully deciphered yet, although it is clear that the
enzyme plays a major role in the host–pathogen interaction.
Indeed, a BCG mutant in which the zmp1 gene has been deleted
(zmp1 rBCG) was developed7 and preclinical studies of the rBCG
D
zmp1 vaccine are currently in progress.8 Therefore, the discovery
of Zmp1 inhibitors is of high relevance to better assessing the role
played by the enzyme in the pathogenicity of tuberculosis. More-
over, the fact that Zmp1 localizes outside the Mtb cell wall, which
is difficult to penetrate, reinforces Zmp1 as attractive pharmaco-
logical target for the development of small molecule inhibitors.
Recently, the first crystal structure of Zmp1 in complex with the
⇑
Corresponding authors. Tel.: +39 0577 234306; fax: +39 0577 234333 (M.B.);
Tel.: +39 0321 375712; fax: +39 0321 375821 (M.R.).
E-mail address: botta.maurizio@gmail.com (M. Botta).
http://dx.doi.org/10.1016/j.bmcl.2014.04.004
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

M. Mori et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2508–2511
2509
generic zinc metalloprotease inhibitor phosphoramidon has been
solved at 2.60 Å resolution (PDB ID: 3ZUK),9 elucidating key
Zmp1/inhibitor interactions and representing therefore a valuable
tool to design specific small molecule Zmp1 inhibitors by means of
computer-aided drug design.
Openeye Scientific Software); (ii) docking of the most promising
hits identified by ROCS towards Zmp1; (iii) rescoring by means
of the Molecular Mechanics Generalized Born Surface Area method
(MM-GBSA) and (iv) visual inspection. First, we generated a ROCS
query based on the energy minimized binding mode of phosphora-
midon deprived of the rhamnose moiety, because it is mostly sol-
vent exposed and does not interact directly with Zmp1 residues.9
The query was then used to screen the whole Asinex database
(about 600,000 small molecules) and the 10% top ranking mole-
cules sorted by the Tanimoto Combo score were docked towards
the catalytic site of Zmp1 in its energy minimized structure. It is
widely appreciated that rescoring docking poses with an external
potential energy function generally improve the quality of a virtual
screening protocol. Based on our experience, the ligand delta
energy of binding of the 1000 top ranking small molecules was
then calculated by means of the MM-GBSA method.14 Most of
these molecules bear a well-known zinc binding group, mostly car-
boxylic acid or sulfonate groups. Finally, top ranking 500 small
molecules endowed with the lowest MM-GBSA delta energy of
binding to Zmp1 were visually inspected. Among them, 18 most
interesting molecules were selected (ZTB1–ZTB18), based on a
combination of binding mode, rescoring energy and chemical
diversity each other as well as with respect to phosphoramidon
or other known Zn-proteases inhibitors, and submitted to bio-
chemical investigations. Chemical structures of selected com-
pounds, the detailed experimental procedure adopted for
carrying out the enzymatic assays on recombinant Mtb Zmp1
and human Neprilysin, and for the calculation of the Ki values for
the different inhibitors are reported as Supplementary information.
Four out of the eighteen molecules (Fig. 1) proved to inhibit
In this multidisciplinary work, we combined a two-round struc-
ture-based virtual screening with organic synthesis, enzymatic and
kinetics assays to discover potent Zmp1 inhibitors endowed with a
significant chemical diversity with respect to generic metallopro-
tease inhibitor phosphoramidon. Starting from the crystallographic
structure of Zmp1/phosphoramidon complex,9 the chain A was
retained due to the less number of gaps with respect to the chain
B. Preliminary self-docking trials were performed by using differ-
ent docking programs (GOLD, Glide, AutoDock)10,11 to probe their
capability to reproduce the crystallographic binding mode of
phosphoramidon. Due to the general failure of the self-docking
procedure, also after removal of the rhamnose moiety from the
co-crystalized inhibitor, the chain A of Zmp1/phosphoramidon
was submitted to energy minimization in explicit solvent by
means of Amber11.12 Repeating the self-docking of phosphorami-
don towards such refined Zmp1 structure provided satisfactory
results. Moreover, different protonation schemes for zinc-binding
residues His493, His497 and Glu560, for Glu494 and His622
(Zmp1 residues are numbered as in the abovementioned X-ray
structure) were tested and the best agreement between computa-
tional and experimental pose of phosphoramidon was obtained by
keeping the non-zinc-binding residues Glu494 and His622 in their
protonated form (Supplementary information). The Goldscore
function of GOLD (version 4.0.1) provided the lowest root mean
square deviation (RMSD = 0.315 Å) between the docking-based
and the energy minimized crystallographic binding pose of phos-
phoramidon (Supplementary information). In line with previous
reports, here we obtained significant benefits by performing struc-
ture-based ligand design towards computationally refined protein
structures.13,14
Zmp1 catalytic activity at 100
active with a Ki of 32.9 0.4
l
M, with ZTB12 being the most
lM (Table 1). The docking-based
binding mode of ZTB12 (Supplementary information) shows the
sulfonate group interacting with Glu494 and His622 as well as
with the catalytic Zn(II) in a geometry that resembles zinc coordi-
nation. The aromatic core is stacked with the side chain of Arg616,
whereas the non-zinc-binding sulfonate establishes H-bonds with
the guanidine group of Arg616.
After these preliminary molecular modeling analysis, we per-
formed virtual screening by means of a protocol composed of four
subsequent steps: (i) a fast screening with ROCS15 (version 3.1.2,
Figure 1. Chemical structure of active Zmp1 inhibitors described in this work.

2510
M. Mori et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2508–2511
Table 1
Inhibition of Zmp1 catalytic activity by small molecules selected in the first-round
virtual screening
Compda
% Inhibition at 100
l
M
Ki (lM)
ZTB7
ZTB9
ZTB10
ZTB12
54
16
28
78
n.db
n.d
n.d
32.9 0.4
a
Molecules not showing inhibition at tested conditions are not reported in the
table.
b
Not determined.
Since ZTB12 was the most potent hit identified in the first-
round virtual screening, it was selected as the reference starting
point of the second-round virtual screening, which aimed at
improving the ligand inhibitory potency towards Zmp1. A new
ROCS query was built on the docking-based binding conformation
of ZTB12 and used to screen the ‘Drugs Now’ subset of the ZINC
database, which was composed of over six million compounds
(the docking conformation of ZTB12 and the respective ROCS query
are in Supplementary information). 10% top-ranking molecules fil-
tered by ROCS were selected for docking. However, due to the large
dimension of such test-set, only compounds bearing at least a car-
boxyl or a sulfonate zinc binding groups were docked towards
Zmp1 (about 11,000 molecules). After rescoring and visual inspec-
tion, nine molecules (ZTB19–ZTB27, see Supplementary informa-
tion) were deemed top priority and submitted to in vitro tests
using pure recombinant Mtb Zmp1. Results of enzymatic assays
showed that five out of the nine compounds delivered by the sec-
ond-round virtual screening proved to inhibit Zmp1 in vitro with
higher potency than ZTB12 (Table 2 and Fig. 1).
Figure 2. Docking based binding mode of ZTB23(R) within the catalytic site of
Zmp1. Polar contacts and zinc coordination are highlighted as black dashed lines.
ZTB23(R) is showed as violet sticks. Zmp1 is showed as green cartoon and
transparent surface, with key residues showed as sticks.
Ph
Ph
Ph
i
ii
S
S
NaOOC
NaOOC
N
H
N
H
S
HOOC
NH2
30
SNa
NaO
32
31
O
iii
S
S
Ph
Ph
S
COOH
N
S
iv
N
HOOC
HOOC
O
O
ZTB23(R)
33
A significant improvement of the hit enrichment was observed,
as more than 55% of molecules selected in silico were active
in vitro. The most potent Zmp1 inhibitor was ZTB23 with a Ki of
Scheme 1. Synthesis of title compound, reaction and conditions: (i) carbon
disulfide, NaOH, H2O, rt overnight; (ii) sodium chloroacetate solution, rt 2 h; (iii)
6 N HCl, two drops POCl3, 75 °C, 4 h; (iv) p-carboxybenzaldehyde, b-alanine,
CH3COOH reflux, 6 h.
0.39 0.088 lM. Since ZTB23 was purchased as racemic mixture
of –R and –S enantiomers (thereafter referred as ZTB23(R) and
ZTB23(S), respectively), in depth docking studies were performed
to predict their binding mode and affinity to Zmp1 as well as both
enantiomers were synthesized by means of stereoselective synthe-
sis. Molecular modeling suggested that, in both enantiomers, the
carboxyl group of the phenylalanine moiety interacts with the cat-
isolated yield). The final Knoeveneagel condensation between 33
and p-carboxybenzaldehyde in acetic acid furnished ZTB23(R) in
very good isolated yield (83%) and high enantiomeric purity as
demonstrated by HPLC analysis (Supplementary information).
The synthesis of ZTB23(S) follows the same synthetic pipeline
alytic zinc ion in
whereas significant differences were observed for the remaining
portion of the molecules. In particular, the -phenylalanine aro-
a geometry resembling zinc coordination,
L
starting with the L-phenylalanine (Supplementary information).
matic ring of ZTB23(S) is well inserted into the hydrophobic S10
cavity with the rhodanine moiety connecting the zinc-binding por-
tion of the inhibitor with the benzoic acid moiety, which take place
in proximity of the large S20 sub-site and interacts with the non-
conserved Arg616. In the docking-based binding mode of ZTB23(R)
Based on a survey performed on Scifinder and PubChem databases
performed at the time of experiments, we may claim that this is
the first report on the synthesis of active inhibitors ZTB23(R) and
ZTB23(S).17
Enzymatic assays showed a Ki of 1.7 0.3
lM for ZTB23(S) and
(Fig. 2), the
D-phenylalanine aromatic ring is projected towards the
a Ki of 0.094 0.054 M for ZTB23(R) (Table 2), reinforcing the
l
solvent in a region occupied by the sugar moiety of phosphorami-
don in its crystallographic pose,9 whereas the rhodanine performs
a H-bond with Asn452 and the benzoic acid moiety is well inserted
in the large S20 sub-site where it interacts with Thr606. Notably,
ZTB23(R) establishes profitable H-bond interactions with Glu494
and His622, which may reinforce the inhibitor affinity to Zmp1
and may lead to a higher inhibitory potency.
Enantiomerically pure ZTB23(S) and ZTB23(R) were stereose-
lectively synthesized as outlined in Scheme 1 (the synthesis of
ZTB23(R) is reported). The rhodanine 33 was obtained by following
the protocol described by Kumar et al.,16 in which commercial
optically pure phenylalanine (30) was reacted with carbon disul-
fide in aqueous sodium hydroxide to give 31. The latter was
reacted in situ with sodium chloroacetate giving 32 that was lastly
cyclized under acidic conditions affording compound 33 (66%
molecular modeling evidence that performing profitable H-bonds
within the Zmp1 catalytic site as well as interacting within the
depth S20 pocket might be key determinants for potent inhibition
of Zmp1. Notably,
a significant selectivity of ZTB23(R) was
observed with respect to the human Neprilysin (selectivity index
>250, Table 2), which is the human enzyme most similar to
Zmp1 according to a BLASTP search (http://blast.ncbi.nlm.nih.
gov/).18
The methyl and benzyl ester derivatives of ZTB23(R), namely
ZTB28(R) and ZTB29(R), were also synthesized as described in
Supplementary information and tested for Zmp1 inhibition. Both
molecules proved to inhibit Zmp1 catalytic activity with
a
weaker potency than ZTB23(R) (Table 2), thus highlighting the
key role of the non-zinc binding benzoic acid group for Zmp1
inhibition.

M. Mori et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2508–2511
2511
Table 2
Supplementary data
Inhibition of Mtb Zmp1 and human Neprilysin catalytic activity by small molecules
selected from the second-round virtual screening.
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.bmcl.
2014.04.004. These data include MOL files and InChiKeys of the
most important compounds described in this article.
Compd
Ki (
l
M) Mtb Zmp1
Ki (
l
M) human Neprilysin
SId
ZTB12a
ZTB19
ZTB20
32.90
3.70
2.70
32.20
0.39 0.09
4.80
1.7 0.30
0.094 0.05
5.7 0.3
1.8 0.1
n.d.c
n.d.
n.d.
n.d.
n.d.
n.d.
37
25
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
21.7
265.9
n.d.
n.d.
ZTB21
References and notes
ZTB23b
ZTB24
1. Petrera, A.; Amstutz, B.; Gioia, M.; Hahnlein, J.; Baici, A.; Selchow, P.; Ferraris, D.
M.; Rizzi, M.; Sbardella, D.; Marini, S.; Coletta, M.; Sander, P. Biol. Chem. 2012,
393, 631.
2. Ferraris, D. M.; Rizzi, M. Encyclopedia of Inorganic and Bioinorganic Chemistry
In Scott, R. A., Ed.; John Wiley: Chichester, 2013. http://dx.doi.org/10.1002/
9781119951438.eibc2171.
3. Master, S. S.; Rampini, S. K.; Davis, A. S.; Keller, C.; Ehlers, S.; Springer, B.;
Timmins, G. S.; Sander, P.; Deretic, V. Cell Host Microbe 2008, 3, 224.
4. Chan, J.; Flynn J. Clin. Immunol. 2004, 110, 2.
ZTB23(S)
ZTB23(R)
ZTB28(R)
ZTB29(R)
a
b
c
Compound 12 has been kept for comparison.
Racemic mixture.
Not determined.
d
Selectivity index calculated as Ki(Neprilysin)/Ki(Mtb Zmp1) ratio.
5. Schroder, K.; Tschopp, J. Cell 2010, 140, 821.
6. Muttucumaru, D. G.; Smith, D. A.; McMinn, E. J.; Reese, V.; Coler, R. N.; Parish, T.
Tuberculosis (Edinb.) 2011, 91, 111.
7. Johansen, P.; Fettelschoss, A.; Amstutz, B.; Selchow, P.; Waeckerle-Men, Y.;
Keller, P.; Deretic, V.; Held, L.; Kundig, T. M.; Bottger, E. C.; Sander, P. Clin.
Vaccine Immunol. 2011, 18, 907.
8. Kaufmann, S. H.; Gengenbacher, M. Curr. Opin. Biotechnol. 2012, 23, 900.
9. Ferraris, D. M.; Sbardella, D.; Petrera, A.; Marini, S.; Amstutz, B.; Coletta, M.;
Sander, P.; Rizzi, M. J. Biol. Chem. 2011, 286, 32475.
10. Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor, R. J. Mol. Biol. 1997, 267,
727.
Finally, kinetics studies showed that all compounds herein dis-
covered inhibit Zmp1 with a competitive mechanism with respect
to the generic fluorogenic substrate for Zn-metalloproteases
MMP2/MMP7 (Calbiochem Prod. Num 03-32-5032).
In conclusion, we used a multidisciplinary approach mixing
molecular modeling, organic chemistry and biochemistry to dis-
cover potent inhibitors of the Zmp1 of Mtb. The most potent inhib-
itor is ZTB23(R) showing inhibition at nanomolar concentration
in vitro and a significant selectivity with respect to human Nepri-
lysin. ZTB23(R) is therefore a useful tool to investigate the biolog-
ical functions of Zmp1 during infection also at the light of the
increasing interest that the BCG mutant-based vaccine zmp1 rBCG
is receiving as a potential novel vaccines against TB.7 Notably,
ZTB23(R) is also a valuable lead candidate for further optimization
finalized to the development of pharmaceutically relevant anti-TB
candidate agents. To the best of our knowledge, this is the first
medicinal chemistry study focusing on the identification of potent
Zmp1 inhibitors.
11. Morris, G. M.; Huey, R.; Lindstrom, W.; Sanner, M. F.; Belew, R. K.; Goodsell, D.
S.; Olson, A. J. J. Comput. Chem. 2009, 30, 2785.
12. Case, D. A.; Darden, T. A.; Cheatham, T.E.; Simmerling, C. L.; Wang, J.; Duke, R.
E.; Luo, R.; Walker, R. C.; Zhang, W.; Merz, K. M.; Roberts, B.; Hayik, S.; Roitberg,
A.; Seabra, G.; Swails, J.; Goetz, A. W.; Kolossváry, I.; Wong, K. F.; Paesani, F.;
Vanicek, J.; Wolf, R. M.; Liu, J.; Wu, X.; Brozell, S. R.; Steinbrecher, T.; Gohlke, H.;
Cai, Q.; Ye, X.; Wang, J.; Hsieh, M. J.; Cui, G.; Roe, D. R.; Mathews, D. H.; Seetin,
M. G.; Salomon-Ferrer, R.; Sagui, C.; Babin, V.; Luchko, T.; Gusarov, S.;
Kovalenko, A.; Kollman, P. A. AMBER 12, 2012. Available at: http://
ambermd.org/.
13. Mori, M.; Dietrich, U.; Manetti, F.; Botta, M. J. Chem. Inf. Model. 2010, 50, 638.
14. Mori, M.; Manetti, F.; Botta, M. J. Chem. Inf. Model. 2011, 51, 446.
15. Grant, J. A.; Gallardo, M. A.; Pickup, B. T. J. Comput. Chem. 1996, 17, 1653.
16. Kumar, B. R. P.; Basu, P.; Adhikary, L.; Nanjan, M. J. Synth. Commun. 2012, 42,
3089.
17. Survey on Scifinder (https://scifinder.cas.org/) and PubChem (https://
pubchem.ncbi.nlm.nih.gov/) was performed on 28th March 2014. Search of
ZTB23 chemical structure on Scifinder provided only one match corresponding
to compound ZTB23 without assigned stereochemistry (CAS number 306324-
21-0), including 11 commercial sources and a single literature reference (PCT
WO 2009113795) where the synthesis of ZTB23 is not reported. Search on
PubChem provided ZTB23 chemical structure without assigned
stereochemistry and associated to a number of biological tests where the
compound was inactive.
Acknowledgments
This work was partially supported by the European Union FP7
program (Project SysteMTb HEALTH-F4-2010-241587). Research
in the laboratory of P.S. is supported by Swiss National Science
Foundation, EU NewTBVac (No. 241 745) and Tuberculosis Vaccine
Initiative (TBVI).
18. Altschul, S. F.; Gish, W.; Miller, W.; Myers, E. W.; Lipman, D. J. J. Mol. Biol. 1990,
215, 403.