A minimalistic approach to identify substrate binding features in B1 Metallo-β-lactamases

Article abstract of DOI:10.1016/j.bmcl.2007.06.089

The 2-oxoazetidinylacetate sodium salt was synthesized as a model of a minimal β-lactam drug. This compound and the monobactam aztreonam were assayed as substrates of the Metallo-β-lactamase BcII. None of them was hydrolyzed by the enzyme. While the azeti

Full text of DOI:10.1016/j.bmcl.2007.06.089

Bioorganic & Medicinal Chemistry Letters 17 (2007) 5171–5174
A minimalistic approach to identify substrate binding features
in B1 Metallo-b-lactamases
a
b
c
´
Andres A. Poeylaut-Palena, Pablo E. Tomatis, Andreas I. Karsisiotis,
Christian Damblon,d Ernesto G. Mataa,* and Alejandro J. Vilab,*
aInstituto de Quı´mica Orga´nica de Sı´ntesis (UNR-CONICET), Suipacha 570, S2002LRK Rosario, Argentina
bInstituto de Biologı´a Molecular y Celular de Rosario (UNR-CONICET), Suipacha 570, S2002LRK Rosario, Argentina
cHenry Wellcome Laboratories of Structural Biology, Department of Biochemistry, University of Leicester, UK
dCentre de Biophysique Mole´culaire-CNRS, Orleans, France
Received 30 May 2007; revised 26 June 2007; accepted 28 June 2007
Available online 5 July 2007
Abstract—The 2-oxoazetidinylacetate sodium salt was synthesized as a model of a minimal b-lactam drug. This compound and the
monobactam aztreonam were assayed as substrates of the Metallo-b-lactamase BcII. None of them was hydrolyzed by the enzyme.
While the azetidinone was not able to bind BcII, aztreonam was shown to bind in a nonproductive mode. These results provide an
explanation for the unability of Metallo-b-lactamases to inactive monobactams and give some clues for inhibitor design.
Ó 2007 Elsevier Ltd. All rights reserved.
Metallo-b-lactamases (MbLs) are zinc enzymes that
hydrolyze b-lactam antibiotics, representing one of the
main resistance mechanisms developed by bacteria
toward antibiotics.1 These enzymes differ from the
well-known serine-b-lactamases in that they display a
broad substrate range, being capable of inactivating
penicillins (1), carbapenems (2), and cefalosporins (3)
(Fig. 1).2 Despite sharing a conserved fold, MbLs are
quite diverse in active site structure, and (possibly)
catalytic mechanisms.3,4 MbLs can bind up to two zinc
ions, which are essential for substrate binding and
hydrolysis.5 The broad structural variation among
substrates and enzymes has impeded the identification
of common structural determinants of substrate
recognition, thus thwarting inhibitor design.6
MbL substrates share the b-lactam ring and a carboxyl-
ate moiety a to the b-lactam nitrogen, which we tenta-
tively define as the ‘minimal b-lactam substrate’. This
carboxylate group is supposed to interact with Zn2
and a positively charged residue in the active site, while
Figure 1. Representative examples for the most common b-lactam
antibiotic families. Penicillin (1), carbapenem (2), cefalosporin (3), and
monobactam (4), and 2-oxo-azetidin-1-yl-acetate sodium salt (5), the
Zn1 delivers the attacking nucleophile.7,8 On the other
proposed ‘minimal b-lactam substrate’ structure.
Keywords: Metallo-b-lactamases; Inhibitor design; Ligand binding;
Monocyclic b-lactams.
hand, monobactams, such as aztreonam (4), are the only
family of b-lactam antibiotics that are not efficiently
hydrolyzed by MbLs. So far, the reasons that make
*
Corresponding
authors.
Tel.:
+543414370477
(E.G.M.);
+543414350661x108 (A.J.V.); e-mail addresses: mata@iquios.gov.ar;
vila@ibr.gov.ar
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.06.089

5172
A. A. Poeylaut-Palena et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5171–5174
aztreonam resistant to MbL-mediated hydrolysis are not
known.3 Here we decided to test whether aztreonam is
able to bind MbLs, and in so doing, to compare its
behavior with that of the model azetidinone compound
(5), representing the ‘minimal b-lactam substrate’.
Substrate hydrolysis. Hydrolysis of aztreonam (4) and
the azetidinone (5) by the MbL BcII was studied by
UV spectroscopy.9 None of these compounds showed
a noticeable hydrolysis over a long time (16 h), even
when adding high enzyme concentrations (10–20 lM).
The absence of hydrolysis was also verified by 1H
NMR spectroscopy. These assays reveal that neither
aztreonam nor azetidinone (5) are suitable substrates
of BcII. This may be attributed to either lack of binding
of these compounds at all to the enzyme, or to a nonpro-
ductive binding mode. Since the finding of a binding,
nonhydrolyzable compound could be exploited to de-
sign a possible inhibitor, we explored binding by differ-
ent biophysical methods.
Figure 2. Corresponding 15N,1H HSQC spectra for free BcII (red) and
BcII plus 40 mM aztreonam (green). The signal labeled as N233 is
assigned to the NH sidechain of the corresponding residue.The NMR
data processing and graphic rendering was done with Sparky (T. D.
Goddard and D. G. Kneller, SPARKY 3, University of California, San
Francisco).
Binding to Co(II)-BcII monitored by UV–vis and fluores-
cence spectroscopy. Co(II)-substitution has been largely
exploited in zinc enzymes as a probe of the metal site
structure, and to follow changes upon exogenous
ligand binding. The spectral features of Co(II)-BcII
have been already characterized in detail.10 The
spectrum is composed of ligand-field bands in the
visible region attributed to the Co1 site and stronger
absorption bands from the Co2 site in the UV region
due to ligand ! metal charge transfer. Addition of aze-
tidinone (5) did not induce spectral changes in the
absorption spectrum of Co(II)-BcII. A steady decrease
in the intensity of all bands was observed at high con-
centrations of added azetidinone (5), suggesting metal
ion dissociation. When aztreonam was added, no
changes could be observed in the visible features of the
spectrum. The charge transfer band of Co(II) BcII at
343 nm was obscured by an intense band of aztreonam
in this range, and could not be followed during this
titration.
final concentration, revealing no binding at all, in agree-
ment with the UV–vis and fluorescence studies.
Addition of aztreonam (4) induced spectral changes at
very high concentrations (>40 mM) (Fig. 2), that are
summarized in Table 1. These changes are small but sig-
nificant. The nature of the resulting shifts indicates a fast
exchange regime, typical of weak ligand binding. The
perturbed residues include the three Zn2 ligands
(D120, C221, H263), K224 (that usually interacts with
the carboxylate moiety of bicyclic b-lactams), S69 (a sec-
ond-shell ligand of the Zn2 site, which is involved in a
hydrogen bond with D120), and residues belonging to
loops 3 and 12, that flank the active site.7 The behavior
of the loops resembles that observed upon inhibitor
binding for the enzyme CcrA,14 which is another B1 sub-
class partner of BcII. In contrast to the situation of
CcrA, there is a neat discrimination between the two
zinc binding sites, since Zn1 ligands appear largely
nonperturbed while Zn2 is involved on aztreonam
binding.
Ligand binding to BcII can also be followed by confor-
mational changes in the enzyme that induce variations
in the intrinsic Trp fluorescence.11 No changes in the
fluorescence intensity could be monitored upon addition
of azetidinone (5) in a stopped flow instrument. More-
over, the addition of azetidinone (5) was unable to mod-
ify nitrocefin binding parameters. In the case of
aztreonam, the absorption of this compound limited
the detection of this phenomenon, suggesting that it
binds very weakly or does not bind at all.
We docked aztreonam in the active site of BcII based
on the CSP map and positioning the sulfonate moiety
on the Zn2 coordination sphere lying away from Zn1
site.15 As a result, the b-lactam moiety is at ꢀ5.5 A
˚
from the attacking nucleophile at the Zn1 site and
in an improper orientation, in a nonproductive bind-
ing mode. The NMR data fully agree with the Co(II)
substitution experiments, that show no interaction of
the ligand with the Co1 site (as results from the ab-
sence of changes in the absorption features in the vis-
ible range). Thus, we can conclude that aztreonam
binds poorly and in a nonproductive way to BcII.
The same result can be extrapolated to all B1 MbLs.
Since aztreonam was designed with 4-methyl substitu-
ent to escape the hydrolysis by serine-b-lactamases, we
cannot completely discard that this moiety is also
effective in this case.
Binding to BcII monitored by NMR spectroscopy.12
Finally, we employed NMR spectroscopy to follow
ligand binding by chemical shift perturbation (CSP) of
the backbone amide residues of BcII.13 Consequently,
15N-BcII was expressed in Escherichia coli BL21(DE3)
pLysS’ cells transformed with the pET-Term-BcII
1
plasmid.10 The corresponding 15N, H HSQC spectra
were recorded upon additions of increasing concentra-
tions of 4 and 5. The HSQC spectrum displayed no
changes when azetidinone (5) was added up to 100 mM

A. A. Poeylaut-Palena et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5171–5174
5173
Table 1. Most significant chemical shift perturbations on BcII residues
upon addition of aztreonam (4)
Training Network (MEBEL, contract HPRN-CT-
´
2002-00264). Dr. Fernando Pasteran from Malbran
Institute is acknowledged for kindly providing us
Aztreonam.
Residuea
CSP (ppm)
Locationb
N233c
S69
F61
0.64
0.61
0.53
0.51
0.51
0.51
0.44
0.39
0.33
0.33
0.31
0.29
0.29
0.29
0.26
0.20
0.10
0.04
0.00
Loop 12
S4, Zn2 second-shell ligand
Loop 3
Loop 12
G232
L58
N51
Supplementary data
S3
S3
Supporting information is available with detailed proce-
dures for the synthesis of azetidinone (5) and the spec-
troscopic data of azetidinone (5) and the synthetic
precursors. Also, UV–vis spectroscopy data of Co(II)
substituted BcII titration with aztreonam and azetidi-
none (5) are included. Supplementary data associated
with this article can be found, in the online version, at
doi:10.1016/j.bmcl.2007.06.089.
L231
H263
Y238
H55
Loop 12
S12, Zn2 ligand
Loop 12
S3
D120
A235
D230
K224
G63
Zn2 ligand
Loop 12
Loop 12
Loop 12
Loop 3
C221
H116
H118
H196
Zn2 ligand
Zn1 ligand
Zn1 ligand
Zn1 ligand
References and notes
1. Crowder, M. W.; Spencer, J.; Vila, A. J. Acc. Chem. Res.
2006, 39, 721.
2. Felici, A.; Amicosante, G.; Oratore, A.; Strom, R.;
Ledent, P.; Joris, B.; Fanuel, L.; Frere, J. M. Biochem.
J. 1993, 291, 151.
The three His ligands of the Zn1 site are included for comparison.
a BBL numbering.
b S, b-strand; Zn1, zinc 1 coordination site; Zn2, zinc 2 coordination
site, for loops’ numbering, see Dal Peraro et al.7
c The perturbed signal corresponds to the Nd2–Hd21 of residue N233.
3. Fisher, J. F.; Meroueh, S. O.; Mobashery, S. Chem. Rev.
2005, 105, 395.
4. Galleni, M.; Lamotte-Brasseur, J.; Rossolini, G. M.;
Spencer, J.; Dideberg, O.; Frere, J.-M. Antimicrob. Agents
Chemother. 2001, 45, 660.
5. Rasia, R. M.; Vila, A. J. J. Biol. Chem. 2004, 279, 26046.
6. Toney, J. H.; Moloughney, J. G. Curr. Opin. Investig.
Drugs 2004, 5, 823.
7. Dal Peraro, M.; Vila, A. J.; Carloni, P. Proteins 2004, 54, 412.
8. Spencer, J.; Read, J.; Sessions, R. B.; Howell, S.; Black-
burn, G. M.; Gamblin, S. J. J. Am. Chem. Soc. 2005, 127,
14439.
Regarding azetidinone (5), the ‘minimal b-lactam sub-
strate’, it is striking that this compound is not able to
bind BcII. This may be attributed to the mobility of
the carboxylate moiety when the second ring is not pres-
ent, or to the absence of other substituents that (not
being conserved) may provide additional nonspecific
anchoring points in the enzyme structure. Analyzing
KM from several substrates it is clear that substrates
without a substituent chain in position 6 of the penicil-
lanic nucleus, or position 7 of the cephalosporanic nu-
cleus, have the highest KM values.2,16 Overall, this
suggests that the disparate substituents in this position
help in providing nonspecific, hydrophobic anchoring
points to b-lactam substrates. Given that azetidinone
(5) and aztreonam (4) are both susceptible to alkaline
hydrolysis, the lack of an efficient binding mode is the
principal factor for the absence of enzymatic hydrolysis
for both compounds.
9. Bacillus cereus MbL BcII was obtained according to already
published methods by Orellano et al. For UV spectroscopy,
hydrolysis assays were carried out in 10 mM KH2PO4,
200 mM NaCl, 20 lM ZnSO4, 50 lg/mL BSA, pH 7.5, at
303 K. For aztreonam, hydrolysis was also tested in 10 mM
Hepes, 200 mM NaCl, 20 lM ZnSO4, and BSA 50 lg/mL,
pH 7.5. Spectral parameters are: aztreonam kmax = 318 nm,
De = 660 Mꢁ1 cmꢁ1; azetidinone (5) kmax = 195 nm. The
spectral features of 5 were determined by following its
alkaline hydrolysis. For 1H NMR spectroscopy, hydrolysis
assays were carried out in a D2O buffer containing KH2PO4
10 mM, NaCl 200 mM, BSA 50 lg/mL, pH 7.5, at 303 K.
The samples were incubated overnight with 10 lM BcII.
The NMR experiments were recorded in a Bruker Avance-
300 spectrometer.
10. Orellano, E. G.; Girardini, J. E.; Cricco, J. A.; Ceccarelli,
E. A.; Vila, A. J. Biochemistry 1998, 37, 10173.
11. Fluorescence quenching experiments were carried on an
Applied Photophysics SX.18-MVR stopped-flow appara-
tus. The method was the same as already reported by
Rasia et al. Final enzyme concentration was 20 lM in
100 mM Hepes, 200 mM NaCl, pH 7.5, at 288 K. Aztre-
onam concentrations were in the lM range. The highest
employed azetidinone concentration was 150 mM.
12. For NMR experiments, labeled BcII was obtained as
follows: an 8 h LB (150 lg/mL ampicillin and 35 lg/mL
cloramphenicol) culture of the transformed cells was
pelleted and inoculated into a 100 mL M9 medium
supplemented with 4 g/L glucose, 1.2 g/L (NH4)2SO4,
1 mM MgSO4, 10 lM CaCl2, 150 lg/mL ampicillin, and
Here we show that the minimal b-lactam substrate is lar-
ger than expected, suggesting that some seemingly ancil-
lary, hydrophobic, substituents may play a role in
assisting substrate binding. This should be taken into
account for future inhibitor design efforts.
Acknowledgments
This work was supported by grants from ANPCyT to
E.G.M. and A.J.V., and a grant from HHMI to
A.J.V. E.G.M. and A.J.V. are staff members of CONI-
CET. A.A.P. was former fellow of ANPCyT and is reci-
pient of a doctoral fellowship from CONICET. P.E.T. is
recipient of a doctoral fellowship from CONICET.
A.I.K. was supported by a European Union Research

5174
A. A. Poeylaut-Palena et al. / Bioorg. Med. Chem. Lett. 17 (2007) 5171–5174
35 lg/mL cloramphenicol. Cells were grown at 37 °C
13. Mollard, C.; Moali, C.; Papamicael, C.; Damblon, C.;
Vessilier, S.; Amicosante, G.; Schofield, C. J.; Galleni, M.;
Frere, J.-M.; Roberts, G. C. K. J. Biol. Chem. 2001, 276,
45015.
14. Fitzgerald, P. M. D.; Wu, J. K.; Toney, J. H. Biochemistry
1998, 37, 6791.
15. Geometry optimization of molecule 4 was carried out
with the MM+ method using HyperChemÓ 7.5. BcII
molecule corresponds to the 1BVT structure already
deposited in the Protein Data Bank. Docking was
performed manually on the HyperChemÓ 7.5 graphical
interface. The energy of the system was not
minimized.
overnight, pelleted, and resuspended in 2 L of the same
fresh medium, but 15N (NH4)2SO4 was used. Cells were
grown for 4–5 h at 37 °C until OD600nm = 0.7 was reached.
Expression was induced by adding 1 mM isopropyl-
thiogalactoside and 100 lM ZnSO4. Cells were cultured
for additional 4–5 h until saturation (OD600nm = 1.8) was
reached. Cells were harvested and BcII was purified as
previously reported by Orellano et al. NMR experiments
were carried out at 308 K in Avance II 600 Bruker
spectrometer using buffer MES 100 mM, NaCl 200 mM,
ZnSO4 200 lM, pH 6.4. BcII samples were in the 0.25–
1.1 mM concentration range. Azetidinone (5) and aztreo-
nam (4) stock solutions were 1 M in the same buffer but
without added zinc(II).
16. Tomatis, P. E.; Rasia, R. M.; Segovia, L.; Vila, A. J. Proc.
Natl. Acad. Sci. U.S.A. 2005, 102, 13761.