Aza-boronic acids as non-β-lactam inhibitors of AmpC-β-lactamase

Article abstract of DOI:10.1016/j.bmcl.2004.05.054

With the aim of improving the ability of non-β-lactam inhibitors to inhibit AmpC-β-lactamase, a series of 3-aza-phenyl-boronic acid derivatives was obtained using in parallel synthesis. The molecules were tested against Escherichia coli AmpC-β-lactamase.

Full text of DOI:10.1016/j.bmcl.2004.05.054

Bioorganic & Medicinal Chemistry Letters 14 (2004) 3979–3983
Aza-boronic acids as non-b-lactam inhibitors of
AmpC-b-lactamaseq
a
b
a
ꢀ a
Valentina Buzzoni, Jesus Blazquez, Stefania Ferrari, Samuele Calo,
Alberto Venturellia and M. Paola Costia,*
aDipartimento di Scienze Farmaceutiche, Universitꢀa degli Studi di Modena e Reggio Emilia, Via Campi 183, 41100 Modena, Italy
b
ꢁ
Departamento de Biotecnologıa Microbiana, Centro Nacional de Biotecnologia (CSIC),
Campus UAM. Cantoblanco, 28049 Madrid, Spain
Received 26 December 2003; revised 15 May 2004; accepted 21 May 2004
Available online 19 June 2004
Abstract—With the aim of improving the ability of non-b-lactam inhibitors to inhibit AmpC-b-lactamase, a series of 3-aza-phenyl-
boronic acid derivatives was obtained using in parallel synthesis. The molecules were tested against Escherichia coli AmpC-b-
lactamase. The best inhibitors, 3-(2-hydroxy-naphthalen-1-ylazo)-phenyl-boronic acid (12) and 3-(2,4-dihydroxy-naphthalen-1-
ylazo)-phenyl-boronic acid (14), showed apparent inhibition constant values (Ki) of 0.3 and 0.45 lM and increased the potency of
the semi-synthetic cephalosporin antibiotic, ceftazidime, lowering its minimum inhibitory concentration (MIC) value of 50%,
against Gram-negative bacteria strains, producing high levels of AmpC-b-lactamase.
Ó 2004 Elsevier Ltd. All rights reserved.
1. Introduction
sulbactam) (Fig. 1). Among the b-lactamase inhibitors,
clavulanic acid and sulbactam (Fig. 1) can inhibit class
A b-lactamases, but are mostly ineffective against class
C b-lactamases. Class C b-lactamases are present
in Gram-negative microorganisms, such as Enterobac-
ter cloacae and Pseudomonas aeruginosa, which cause
serious health problems. The urgency to develop new
b-Lactam antibiotics inhibit bacterial cell wall synthesis
and therefore the proliferation of microorganisms. The
over-expression of b-lactamases is one of the most
common and well-studied mechanisms of b-lactam
antibiotic resistance.1–3 b-Lactamases compete with
penicillin binding proteins (PBP) in binding b-lactam
antibiotics. b-Lactamases deactivate the b-lactam mole-
cules by hydrolyzing the b-lactam ring, thus preventing
the interaction of the drug with the PBPs. Different
classes of b-lactamases are known. The most clinically
important are class A b-lactamases, which include the
plasmid-based TEM penicillinase, and class C b-lacta-
mases, represented by cephalosporinases, such as
AmpC-b-lactamase.4–6 To overcome the action of these
enzymes, medicinal chemists have introduced ‘b-lacta-
mase resistant’ b-lactams (e.g. aztreonam) or b-lactam-
based b-lactamase inhibitors (e.g. clavulanic acid and
H2N
O
H
SO3H
O
O
S
O
N
S
OH
O
N
N
N
N
O
H
O
N
COOH
COOH
O
HOOC
clavulanic acid
sulbactam
aztreonam
Br
S
SO2
SO2
NH
N
N
NH2
Keywords: AmpC-b-lactamase; Enzyme inhibitors; Phenyl-boronic
acids; In parallel chemistry.
q Supplementary data associated with this article can be found, in the
online version, at doi:10.1016/j.bmcl.2004.05.054
* Corresponding author. Tel.: +39-059-205-5134; fax: +39-059-205-
5131; e-mail: costimp@unimore.it
HO
B(OH)2
B(OH)2
B(OH)2
1
3
2
Figure 1. b-Lactamases inhibitors.
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.05.054

3980
V. Buzzoni et al. / Bioorg. Med. Chem. Lett. 14(2004) 3979–3983
compounds that are able to inhibit class C b-lacta-
on the E. coli AmpC-bL crystal structure, a library of 42
compounds was designed and synthesized.16 The best
compound, 3-(4-benzensulfonyl-thiophene-e-sulfonyl-
amino)-phenylboronic acid (2, Fig. 1), showed a Ki
of 0.08 lM, which is about one hundred-fold less than
the Ki of the starting hit, 1.16 Further studies showed
that some phenyl-aza derivatives, such as 5-(4-bromo-
phenylazo)-2-hydroxy-phenyl-boronic acid (3, Fig. 1),
inhibit E. coli AmpC-bL, with a Ki value of 3 lM
(Shoichet B. K. personal communication).
mases7;8 is clear.
In previous papers, the discovery of phenyl-boronic acid
derivatives as new hits in the development of class C b-
lactamases inhibitors, using a cycle of molecular mod-
elling, synthetic chemistry and enzymatic assays9;10 was
described. These molecules act as transition state ana-
logues competing with the b-lactam antibiotic and
deactivating the enzyme.11–15 Virtual screening of the
commercially available boronic acid derivatives, using
the DOCK 3.5 program, led to the identification of 3-
amino-phenyl-boronic acid (1, Fig. 1), which shows a Ki
of 7.3 lM against Escherichia coli AmpC-b-lactamase
(bL).10 This molecule was recognized to be a potential
scaffold for new inhibitors: starting from this scaffold
and using structure-based drug design approaches based
With the aim of combining the 3-amino-phenyl-boronic
acid feature with the phenyl-aza moiety, compound 1
was selected as the starting scaffold to design and syn-
thesize a small library of 3-aza-phenyl-boronic acid
derivatives. Visual inspection of the crystal structure of
the complex E. coli AmpC-bL-1 clearly suggested that a
Table 1. Apparent inhibition constants (Ki lM) of compounds 1, 4–14, 15a, 16a against E. coli AmpC-b-lactamase
3
R
2
1
4
B(OH)2
Compound
R
Ki (lM)
Compound
R
Ki (lM)
1
NH2
7.3
10
2.0
5
7
N
N
N
N
SO2
9
10
8
5
6
SO3H
S
SO3H
OH
OH
6
9
7
8
N
N
N
N
4
5
3.5
0.7
11
12
5.0
0.3
5
6
8
7
6
7
8
5
5
7
6
9
10
8
HO
OH
N
N
OH
HO
7
6
5
8
NO2
N
N
N
N
6
1.5
13
0.9
9
10
7
5
6
N
N
OH
5
Cl
6
9
7
8
5
6
7
8
3.0
1.0
14
0.45
1.8
OH
HO
Cl
O
N
N
HO
8
6
N
N
CH3
N
5
9
N
15a
7
N
N
N
CH3
13
10
OH
12 11
HO
NH2
8
6
N
5
N
N
OH
N
5
9
2.7
16a
0.7
7
N
N
7
12
9
6
B(OH)2
OH
11 10

V. Buzzoni et al. / Bioorg. Med. Chem. Lett. 14(2004) 3979–3983
3981
substituent in position 3 can fill the ‘R1’ pocket of the
enzyme, which is an essentially unexplored binding site
in the substrate binding domain, typically occupied
by groups in position 6 of b-lactams.11;14;16 Thirteen new
compounds were obtained and tested against E. coli
AmpC-bL. Some selected compounds were also tested
against Gram-negative resistant bacteria: E. coli
and 16 (Scheme 2) were prepared separately, starting
from 3-amino-phenol via diazotation of the amino
group in position 3, and further reaction with substi-
tuted aromatic rings. Subsequently, they were reacted
with compound
1 to give compounds 15a, 16a
(Scheme 2). The purity of all the synthesized compounds
was determined by both TLC and elementary analy-
ses. For all the compounds, structural characteriza-
tion was achieved by nuclear magnetic resonance
(1H NMR). For a few compounds, UV spectra were also
recorded (all the details may be found in supplementary
materials).
K12AmpC-Ent,
E. coli derepressed, Enterobacter
cloacae derepressed 12991 ED, Citrobacter freundii de-
repressed 91/98-2, and Pseudomonas aeruginosa dere-
pressed 88/98.
2. Chemistry
3. Biological results and discussion
A small library of 13 aza-derivatives was synthesized
using in parallel liquid phase chemistry (4–14, 15a,
16a)17;18 (Table 1). The first library (4–14) was prepared
starting from compound 1, via diazotation of the amino
group and further reaction with substituted aromatic
rings (Scheme 1). Filtration under vacuum of the reac-
tion mixtures provided compounds 12–14, as highly
pure material. Although the reaction was simple and
quick, a column chromatography purification step was
required to provide compounds 4–8; for all the other
compounds, liquid/liquid extraction was performed to
provide highly pure compounds. The aza compounds 15
All the compounds were tested against E. coli AmpC-bL
and apparent inhibition constant (Ki) values, in com-
petition with the substrate cephalothin (Table 1), were
obtained using spectrophotometric assay.10 All the
experiments were repeated at least three times and
the standard error was under 20%. The Ki range of the
newly synthesized compound was 0.3–5 lM and 6 out of
13 compounds were active in the sub-micromolar range
(0.3–1 lM). The span for the Ki values is at least ten
times. The naphthol derivatives, compounds 12 and 14
show the best inhibitory activity, with Ki values of 0.3
and 0.45 lM, respectively. Compound 13 shows an
affinity that is three times lower, in comparison with the
ortho-substituted compound 12 (Ki ¼ 0.9 and 0.3 lM,
respectively). Among the phenol derivatives, compound
5 is five times more potent, in comparison with the
ortho-substituted compound 4 (0.7 lM, compared to
3.5 lM). A nitro group added in ortho to the hydroxyl of
compound 4 (compound 6), does not greatly modify the
affinity of the molecule; on the contrary, two chlorine
substituents on compound 5 (compound 7), decrease the
affinity fourfold. Among the naphthalene derivatives,
the introduction of an acidic group, such as a sulfonylic
group, was attempted in order to improve solubility in
water, but only compound 11 was obtained. Unfortu-
nately, the activity of compound 11 was about ten times
lower than that of the other naphthalene derivatives
(12–14). By combining the phenol and the naphthol
moieties that proved to be the best (compounds 12 and
14), compounds 15a and 16a were obtained. They
showed Ki values lower than 2 lM, thus not seriously
improving the affinity in comparison with the starting
molecule 4, (Ki ¼ 3.5 lM). Compound 10 shows the
same affinity for the enzyme as the mono-phenol
derivatives. These results demonstrate that the active site
is accessible to bulky groups, so a three-ring system can
be allocated to the enzyme pocket, but the compounds
are not as active as expected, based on simple synergistic
considerations. It seems that the rigidity of the aza-
bridge does not allow good adaptation to the enzyme
binding site, as observed in a previous paper, where a
new class of 3-amino-phenyl-boronic acid, with a sul-
fonamide and carbamide bridge, was synthesized.16 Also
in that case, the more rigid carbamide bridge inferred
lower affinity in comparison with the highly flexible
sulfonamide bridge. It seems that the flexible fragments
R
N
N
NH2
i, ii
B(OH)2
B(OH)2
1
4-14
Scheme 1. Reagents and conditions: (i) NaNO2, HCl 5 N, 0 °C; (ii) R,
NaOH 10%.
R1
NH2
R1
R2
N
iii, iv
HO
N
+
R2
OH
15 R1: OH, R2: H
16 R1: OH, R2: OH
NH2
v, vi
B(OH)2
N
B(OH)2
N
R1
N
HO
N
R2
15a R1: OH, R2: H
16a R1: OH, R2: OH
Scheme 2. Reagents and conditions: (iii) NaNO2, HCl 5 N, 0 °C; (iv)
R, NaOH 10%; (v) NaOH 10%; (vi) NaNO2, HCl 5 N, 0 °C

3982
V. Buzzoni et al. / Bioorg. Med. Chem. Lett. 14(2004) 3979–3983
linked to the core skeleton are necessary for better
molecular recognition, which requires the adaptation of
the ligand to the enzyme.
protein and form some structure–activity relationships.
The crystallographic structure of E. coli AmpC-bL in
complex with compound 216 was considered (Fig. 3).
The ligand adopts a jack-knife structure and forms most
of the interactions through the 3-sulfonyl-amino-phenyl-
boronic acid fragment of the molecule. In particular, the
borine forms a dative covalent bond with Ser64, the two
hydroxyls of the boronic acid group are involved in
hydrogen bonds with the main chain of Ser64, Ala318
and the side chain of Tyr150. The phenyl ring forms a
dipole–quadrupole interaction with Asn152, while the
sulfonamide nitrogen forms a dipole–quadrupole inter-
action with Tyr221 and the sulfonamide oxygen inter-
acts with Tyr221 and Ser212. Moreover, four water
molecules, one of which is highly conserved, contributes
to the stabilization of this part of the complex. The
distal part of the molecule forms a hydrogen bond with
a water molecule through one sulfone oxygen molecule,
whereas the phenyl ring interacts with Leu119 and
Leu293.16 This molecule shows a Ki value of 0.08 lM,16
about four times lower than the best compound studied
in this work, which is compound 12.
For further studies, we considered compounds 12 and
14, which showed the lowest Ki values. The minimum
inhibitory concentration (MIC) of ceftazidime (Ce-
ftAZ), in the presence and absence of compounds 12 or
14, was determined,19 against several Gram-negative
resistant clinical isolates that show an AmpC-dere-
pressed phenotype, E. coli derepressed, E. cloacae de-
repressed 12991 ED, C. freundii derepressed 91/98-2,
and P. aeruginosa derepressed 88/98, and the laboratory
engineered E. coli K12AmpC-Ent, which expresses
AmpC from E. cloacae cloned in a multi-copy plasmid
(Table 2, Fig. 2). The compounds were only weakly
active against the Gram-negative bacteria. Compounds
12 and 14 lowered the CeftAZ MIC values by 50%
against E. coli der. and P. aureuginosa der. 88/98.
Although a reduction of this level is generally not con-
sidered significant, this result was obtained consistently
in different experiments. Only compound 12 decreases
the CeftAZ MIC value fourfold against C. freundii der.
91/98-2, thus the association of the antibiotic CeftAZ
with these boronic compounds in vivo seems to
strengthen the activity of the antibiotic alone.
Visual inspection20 of the three-dimensional active site
of E. coli AmpC-bL allowed us to understand how
compounds 4–14, 15a and 16a can interact with the
Table 2. Minimum inhibitory concentration (MIC) (lg/mL) of ceft-
azidime (CeftAZ) for compounds 12 and 14
CeftAZ
CeftAZ/12
(1/1)
CeftAZ/14
(1/1)
E. coli K12AmpC-Ent
E. coli derepressed
E. cloacae derepressed
12991 ED
64
64
64
64
64
128
256
256
256
C. freundii derepressed
91/98-2
P. aeruginosa
64
16
64
64
64
128
Figure 3. Surface of the active site of E. coli AmpC-bL with compound
2 bound.
derepressed 88/98
300
250
200
150
100
50
CeftAZ
CeftAZ/12 (1/1)
CeftAZ/14 (1/1)
0
E. coli
K12 AmpC-Ent
E. coli
derepressed
E. cloacae
C. freundii
P. aeruginosa
derepressed 12991 ED derepressed 91/98-2 derepressed 88/98
Figure 2. Minimum inhibitory concentration (MIC) (lg/mL) of ceftazidime (CeftAZ) for compounds 12 and 14. The compounds were dissolved in
50% DMSO and dilutions were performed using growth medium. An adequate final concentration in which to determine the minimum inhibitory
concentration (MIC) was obtained where the concentration of DMSO was maintained below 5%.

V. Buzzoni et al. / Bioorg. Med. Chem. Lett. 14(2004) 3979–3983
3983
Compounds 4–14, 15a and 16a were drawn three-
dimensionally using InsightII:20 their phenyl boronic
acid moieties were rigidly superimposed on the same
moiety as compound 2, while the aza-substituents of the
molecules were manually fitted inside the binding site of
E. coli AmpC-bL. The aza bridge of these molecules
imposes a rigid planar conformation, which is more
difficult to allocate to the active site of the enzyme.
Moreover, these compounds, lacking the sulfonamide
bridge, do not interact with Tyr221 and Ser212. How-
ever, it is reasonable to consider that these molecules
conserve the interaction using the boronic acid group,
but in order to accommodate the aza moiety, the C–B
bond of compounds 4–9, 11–14 has to rotate by about
31° so that the second ring interacts with His210,
Val211, Glu61 and Gly320. In this binding mode, the
distance between Asn152and the first phenyl ring in-
creases. As a consequence, the dipole–quadrupole
interaction is weaker. In this binding mode, the enzyme
has to slightly change its conformation in the region of
Val211 and Glu61.
bad contacts, making the interactions with L119 and
L293 weaker.
Acknowledgements
We would like to thank Dr. Tondi and Dr. Brian K.
Shoichet (UCSF, San Francisco, USA) for reading the
manuscript and Ely Lilly, Indianapolis for their financial
support and NIH GM63815 B.K.S. grant. We also
thank Dr. Stefano Ghelli for his help with the NMR
experiments, R. Gallesi of the Microanalyses Labora-
tory, Centro Interdipartimentale Grandi Strumenti
(CIGS) of Modena University and Centro Inter-
dipartimentale di Calcolo Automatico ed Informatica
Applicata (CICAIA) of Modena University.
References and notes
In order to better accommodate these compounds, and
also compounds 10, 15a and 16a, which have a bulkier
side chain, a second binding mode should be considered.
In this case, the C–B bond should be rotated by about
106°, in order to locate the side chain in the region of
Leu119, Leu293 and Asn289. In this case, only the
boronic acid group conserves the interactions with
Ser64, Ala318 and Tyr150, whereas the first phenyl ring
interacts with Asn152.
1. Hide, B. ASM News 1994, 60, 5.
2. Neu, H. C. Science 1992, 257, 1064.
3. Gerard, D. Chem. Biol. 2000, 7, 127.
4. Bush, K.; Jacoby, G. A.; Medeiros, A. A. Antimicrob.
Agents Chemother. 1995, 39, 1211.
5. Pares, S.; Mouz, N.; Petillot, Y.; Hakenbeck, R.; Dide-
berg, O. Nat. Struct. Biol. 1996, 3, 284.
6. Anderson, J. W.; Pratt, R. F. Biochemistry 2000, 39,
12200.
7. Amicosante, G.; Felici, A.; Segatore, B.; Di Marzio, L.;
Franceschini, N.; Di Girolomo, M. J. Chemother. 1989, 1,
394.
On the basis of the two binding modes proposed, we can
derive some general structure–activity relationships
(SAR). A hydroxyl group in para on the second aro-
matic ring (compounds 13, 5 and 7) suggests the first
binding mode in which the hydroxyl group could form a
hydrogen bond with the backbone of His210 and
Val211. In the case of compound 7, showing a threefold
lower Ki value in comparison with compound 5, the
hindrance of the two chlorine atoms in meta-positions
cause the ligand to move slightly, resulting in new
interactions of the hydroxyl group with His210 and
Val211. A hydroxyl group in ortho (compounds 12, 14,
4, 6) suggests the second binding mode in which the
hydroxyl group could form a hydrogen bond with
Asn289. The one order difference in the Ki between
compounds 4 and 12 could be due to nonspecific inter-
actions with the enzyme, which are difficult to describe
using the current model. In the case of compound 6, the
presence of the nitro group in meta-position causes the
ligand to move slightly, and so affecting the interaction
of the hydroxyl group with Asn289. A substituent in
meta, such as the sulfonyl acid group of compound 11,
could also interact in the second binding mode with
Asn289, but owing to its larger dimensions, it is rea-
sonable that the ligand moves a little in order to avoid
8. Dryjanski, M.; Pratt, R. F. Biochemistry 1995, 34, 3561.
9. Beesley, T.; Gascoyne, N.; Knott-Hunziker, V.; Petursson,
S.; Waley, S. G.; Jaurin, B.; Grundstrom, T. Biochem. J.
1983, 209, 229.
10. Weston, G. S.; Blazquez, J.; Baquero, F.; Shoichet, B. K.
J. Med. Chem. 1998, 4, 4577.
11. Lindquist, R. N.; Terry, C. Arch. Biochem. Biophys. 1974,
160, 135.
12. Bone, R.; Shenvi, A. B.; Kettner, C. A.; Agard, D. A.
Biochemistry 1987, 26, 7609.
13. Sutton, L. D.; Stout, J. S.; Hosie, L.; Spencer, P. S.;
Quinn, D. Biochem. Biophys. Res. Commun. 1986, 134,
386.
14. Usher, K. C.; Blaszczak, L. C.; Weston, G. S.; Shoichet, B.
K.; Remington, S. J. Biochemistry 1998, 37, 16082.
15. Powers, R. A.; Blazquez, J.; Weston, G. S.; Morosini, M.
I.; Baquero, F.; Shoichet, B. K. Protein Sci. 1999, 8, 2330.
16. Tondi, D.; Powers, R. A.; Caselli, E.; Negri, M. C.;
Blazquez, J.; Costi, M. P.; Shoichet, B. K. Chem. Biol.
2001, 8, 593.
17. Sugimoto, K.; Ohki, S. Chem. Pharm. Bull. 1964, 12, 1375.
18. Hackmann, E.; Carvalho Ferreira, P. Boll. Chim. Farm.
1975, 114, 501.
19. National Committee for Clinical Laboratory Standards,
Villanova, PA, 1997.
20. InsighII. http://www.accelrys.com., 2000.