Structural aspects for evolution β-lactamases from penicillin-binding proteins

Article abstract of DOI:10.1021/ja034861u

Penicillin-binding proteins (PBPs), biosynthetic enzymes of bacterial cell wall assembly, and β-lactamases, resistance enzymes to β-lactam antibiotics, are related to each other from an evolutionary point of view. Massova and Mobashery (Antimicrob. Agents

Full text of DOI:10.1021/ja034861u

Published on Web 07/19/2003
Structural Aspects for Evolution of â-Lactamases from
Penicillin-Binding Proteins
Samy O. Meroueh,†,| George Minasov,‡,| Wenlin Lee,† Brian K. Shoichet,§ and
Shahriar Mobashery*,†
Contribution from the Department of Chemistry and Biochemistry, UniVersity of Notre Dame,
Notre Dame, Indiana 46556, Department of Molecular Pharmacology and Biological Chemistry,
Northwestern UniVersity, 303 East Chicago AVenue, Chicago, Illinois 60611, and Department of
Pharmaceutical Chemistry, UniVersity of California, San Francisco, Genentech Hall,
San Francisco, California 94143
Received February 25, 2003; E-mail: mobashery@nd.edu
Abstract: Penicillin-binding proteins (PBPs), biosynthetic enzymes of bacterial cell wall assembly, and
â-lactamases, resistance enzymes to â-lactam antibiotics, are related to each other from an evolutionary
point of view. Massova and Mobashery (Antimicrob. Agents Chemother. 1998, 42, 1-17) have proposed
that for â-lactamases to have become effective at their function as antibiotic resistance enzymes, they
would have had to undergo structure alterations such that they would not interact with the peptidoglycan,
which is the substrate for PBPs. A cephalosporin analogue, 7â-[N-Acetyl-L-alanyl-γ-D-glutamyl-L-lysine]-
3-acetoxymethyl-3-cephem-carboxylic acid (compound 6), was conceived and synthesized to test this notion.
The X-ray structure of the complex of this cephalosporin bound to the active site of the deacylation-deficient
Q120L/Y150E variant of the class C AmpC â-lactamase from Escherichia coli was solved at 1.71 Å
resolution. This complex revealed that the surface for interaction with the strand of peptidoglycan that
acylates the active site, which is present in PBPs, is absent in the â-lactamase active site. Furthermore,
insertion of a peptide in the â-lactamase active site at a location where the second strand of peptidoglycan
in some PBPs binds has effectively abolished the possibility for such interaction with the â-lactamase. A
2.6 ns dynamics simulation was carried out for the complex, which revealed that the peptidoglycan surrogate
(i.e., the active-site-bound ligand) undergoes substantial motion and is not stabilized for binding within the
active site. These factors taken together disclose the set of structure modifications in the antibiotic resistance
enzyme that prevent it from interacting with the peptidoglycan, en route to achieving catalytic proficiency
for their intended function.
Sequence similarity, a shared protein fold, conservation of
structural motifs, and mechanistic features such as active site
acylation by their respective substrates at specific serine residues
have been proposed as consistent with kinship of â-lactamases
and penicillin-binding proteins (PBPs).1-4 â-Lactamases are
bacterial resistance enzymes to â-lactam antibiotics, and PBPs
are biosynthetic enzymes involved in assembly and processing
of the cell wall. It is clear that, despite the divergence of the
sequences, the general protein fold for these families of enzymes
is preserved. The divergence of the sequences has allowed for
impressive diversification of function among these proteins.
It is widely thought that the genes for the more ancient PBPs
gave rise to those of â-lactamases.2,4-6 PBPs process the
building blocks for the cell wall, namely the peptidoglycan (1).
Two of the important PBP activities are the DD-transpeptidase
and DD-peptidase reactions. The former activity is responsible
for the final step of cell wall assembly by cross-linking two
strands of peptidoglycan (species 4). The latter activity removes
the C-terminal D-alanine from the peptidoglycan (species 5), a
process that moderates the degree of cross-linking. The enzymes
that carry out these activities catalyze their reactions in two steps
that involve an intermediary acyl-enzyme species (2).
â-Lactamases of classes A, C, and D also go through an
acyl-enzyme species in their turnover of â-lactam antibiotics.
These antibiotics mimic the structure of the acyl-D-Ala-D-Ala
portion of the peptidoglycan.7 However, the catalytic machiner-
† University of Notre Dame.
‡ Northwestern University.
§ University of California, San Francisco.
| These authors contributed equally.
(1) Massova, I.; Mobashery, S. Curr. Pharm. Des. 1999, 5, 929-937.
(2) Massova, I.; Mobashery, S. Antimicrob. Agents Chemother. 1998, 42, 1-17.
(3) Kelly, J. A.; Dideberg, O.; Charlier, P.; Wery, J. P.; Libert, M.; Moews, P.
C.; Knox, J. R.; Duez, C.; Fraipont, C.; Joris, B.; Dusart, J.; Frere, J. M.;
Ghuysen, J. M. Science 1986, 231, 1429-1431.
(5) Ghuysen, J. M.; Charlier, P.; Coyette, J.; Duez, C.; Fonze, E.; Fraipont,
C.; Goffin, C.; Joris, B.; Nguyen-Disteche, M. Microb. Drug Resist. 1996,
2, 163-175.
(6) Ghuysen, J. M. Annu. ReV. Microbiol. 1991, 45, 37-67.
(7) Tipper, D. J.; Strominger, J. L. Proc. Natl. Acad. Sci. U.S.A. 1965, 54,
1133-1141.
(4) Kotra, L. P.; Samama, J. P.; Mobashery, S. In Bacterial Resistance to
Antimicrobials, Mechanisms, Genetics, Medical Pratice and Public Health;
Lewis, A., Salyers, A. A., Haber, H., Wax, R. G., Eds.; Marcel Dekker:
2002; p 123.
9
9612
J. AM. CHEM. SOC. 2003, 125, 9612-9618
10.1021/ja034861u CCC: $25.00 © 2003 American Chemical Society

â-Lactamases from Penicillin-Binding Proteins
A R T I C L E S
ies of these three classes of â-lactamases are distinct and have
evolved independently from different ancestral PBP progeni-
tors.2,4
designed cephalosporin molecule that provides structural evi-
dence for the proposal of Massova and Mobashery. Molecular
dynamics simulations of this complex provide additional insights
into these issues, suggesting that a portion of the ligand
undergoes substantial motion in the active site of this enzyme
in agreement with X-ray diffraction data, which revealed the
existence of two possible conformers of the bound ligand. The
2.6 ns molecular dynamics simulation provided a detailed
account of the motion of the ligand within the active site, which
incidentally samples both conformers of the ligand seen within
the X-ray structure of the complex.
Experimental Procedures
Compound 14 was prepared by a literature method.10 The Q120L/
Y150E mutant of E. coli AmpC â-lactamase was made by oligonucle-
otide-directed mutagenesis, expressed, and purified as previously
described.11
N-Fluorenylmethoxycarbonyl-L-alanyl-D-glutamic Acid, R-tert-
Butyl, γ-Benzyl Diester (10). To the mixture of N-Fmoc-L-alanine (8)
(748 mg, 2.40 mmol), D-glutamic acid R-tert-butyl-γ-benzyl diester
(9) (587 mg, 2.00 mmol), and HOBt (324 mg, 2.40 mmol) in CH2Cl2
(20 mL) was added DCC (480 mg, 2.33 mmol). The resultant mixture
was stirred at room temperature overnight. The precipitated DCU was
filtered, and the filtrate was concentrated on a rotary evaporator. Ethyl
acetate (30 mL) was added to the residue, and the suspension of
additional DCU was filtered. The solvent was removed in vacuo, and
the residue was redissolved in EtOAc. This solution was washed with
1 N HCl, water, sat. NaHCO3, water, and brine and then was dried
over MgSO4. The solvent was removed in vacuo, and the product was
purified by column chromatography (silica gel, 5:3 then 1:1 hexane/
1
EtOAc) to give a white solid (761 mg, 65%). H NMR (CDCl3): δ
1.39-1.43 (overlapping d and s, 12H), 1.91-2.05 (m, 1H), 2.18-2.27
(m, 1H), 2.35-2.51 (m, 2H), 4.20-4.23 (m, 1H), 4.32-4.39 (m, 3H),
4.50-4.55 (m, 1H), 5.10 (s, 2H), 5.61 (bd, 1H, J ) 7.6 Hz), 6.94 (bd,
1H, J ) 7.2 Hz), 7.28-7.40 (m, 9H), 7.58 (bd, 2H, J ) 6.8 Hz), 7.75
(d, 2H, J ) 8 Hz). 13C NMR (CDCl3): δ 19.20, 27.67, 28.19, 30.45,
47.31, 50.72, 52.37, 66.77, 67.34, 82.79, 120.23, 125.37, 127.33, 127.95,
128.50, 128.54, 128.81, 141.52, 144.00, 156.20, 170.93, 172.51, 172.88.
EI HRMS 529.1973 (M+ - C4H9, calcd for C30H29N2O7 529.1975).
N-Acetyl-L-alanyl-D-glutamic Acid, R-tert-Butyl, γ-Benzyl Diester
(11). A round-bottomed flask was charged with compound 10 (453
mg, 0.772 mmol) and 10% piperidine/CH2Cl2 (22 mL), and the resultant
solution was stirred at room temperature for 1 h. The solution was
washed with phosphate buffer (pH 5.5, 1 M), water, and brine and
then was dried over MgSO4. The solvent was removed in vacuo, and
the crude residue was redissolved in CH2Cl2 (20 mL). To this solution
was added Et3N (115 µL, 0.849 mmol) and acetic anhydride (80 µL,
0.849 mmol), and the resultant mixture was stirred at room temperature
overnight. The solvent was removed on a rotary evaporator, and the
product was purified by column chromatography (silica gel, 1:4 hexane/
EtOAc) to give a colorless oil (314 mg, 100%). 1H NMR (CDCl3): δ
1.39 (d, 3H, J ) 7.2 Hz), 1.48 (s, 9H), 1.98-2.54 (m, 7H), 4.46-4.59
(m, 2H), 5.15 (s, 2H), 6.28 (bd, 1H, J ) 7.2 Hz), 6.94 (bd, 1H, J )
7.6 Hz), 7.34-7.41 (m, 5H). 13C NMR (CDCl3): δ 19.7, 24.3, 28.3,
29.1, 31.4, 49.8, 53.4, 67.7, 83.6, 129.3, 129.4, 129.7, 136.8, 171.1,
171.6, 173.3, 173.8; EI HRMS 350.1476 (M+ - C4H8, calcd for
C17H22N2O6 350.1478).
â-Lactamases have specialized as antibiotic resistance en-
zymes. Indeed, at least â-lactamases of classes A8 and C9 have
fully evolved for their function by having become diffusion-
limited for their preferred substrates. Massova and Mobashery
have argued that for this catalytic proficiency to have reached
its full potential, â-lactamases had to undergo alteration of their
structures such that they would not bind to the peptidoglycan.2
It was proposed based on the study of the existing X-ray
structures for â-lactamases and PBPs that the corresponding
surface of the PBP that would recognize the peptidoglycan
strand that acylates the active site (present in both DD-
transpeptidases and DD-peptidases) has been retracted in
â-lactamases, such that molecular recognition of the peptidogly-
can at this subsite of the active site would be impaired in
â-lactamases. Furthermore, it was proposed that independent
insertions of several amino acids in each of the cases of classes
A and C of â-lactamases in the middle of the binding site for
the second peptidoglycan strand (in the case of DD-transpep-
tidases) would abolish binding to the PBP substrate at this
subsite. In the course of evolution of â-lactamases, the advent
of these two sets of alterations in the structure of a given parental
PBP would liberate the nascent â-lactamase from the interactions
with the PBP substrate, facilitating its evolutionary trajectory
toward the emergence of bona fide resistance enzymes.
N-Acetyl-L-alanyl-γ-D-glutamyl(R-OtBu)-L-lysine(ꢀ-N-Boc), Ben-
zyl Ester (13). A solution of compound 11 (270 mg, 0.664 mmol) in
MeOH (20 mL) was charged with Pd/C (5% Degussa type, 32 mg),
and the mixture was stirred vigorously under an ambient pressure of
In this report, we present results of X-ray crystallography of
the class C â-lactamase from Escherichia coli bound to a
(10) Micetich, R. G.; Singh, R.; Singh, M. P.; Shaw, C. C. Synthetic Commun.
1986, 16, 453-460.
(11) Patera, A.; Blaszczak, L. C.; Shoichet, B. K. J. Am. Chem. Soc. 2000, 122,
10504-10512.
(8) Hardy, L. W.; Kirsch, J. F. Biochemistry 1984, 23, 1275-1282.
(9) Bulychev, A.; Mobashery, S. Antimicrob. Agents Chemother. 1999, 43,
1743-1746.
9
J. AM. CHEM. SOC. VOL. 125, NO. 32, 2003 9613

A R T I C L E S
Meroueh et al.
Table 1. Crystallographic Statistics and Refinement Results
hydrogen at room temperature for 3 h. The mixture was then filtered,
and the filtrate was concentrated in vacuo to give the crude acid as a
white solid (200 mg). 1H NMR (CDCl3): δ 1.37 (d, 3H, J ) 6.9 Hz),
1.45 (s, 9H), 1.93-2.48 (m, 7H), 4.39-4.66 (m, 2H), 6.81 (bd, 1H, J
) 7.8 Hz), 7.30 (bd, 1H, J ) 7.8 Hz). 13C NMR (CDCl3): δ 18.5,
23.0, 26.8, 27.9, 30.1, 48.9, 52.5, 82.5, 170.6, 170.8, 172.7, 176.1. This
crude product was redissolved in CH2Cl2/DMF (12/2 mL), and to this
solution was added 12 (256 mg, 0.762 mmol), HOBt (86 mg, 0.635
mmol), and DCC (118 mg, 0.571 mmol). The resultant mixture was
stirred at room temperature overnight. Some CHCl3 (∼50 mL) was
added to dilute the reaction mixture, and the organic solution was
washed with 5% NaHCO3, 1 N HCl, water, and brine and then was
dried over MgSO4. The solvent was removed in vacuo, and the product
was purified by column chromatography (silica gel, 50:1 CHCl3/MeOH)
space group
C2
cell constants: (Å), (deg)
a ) 118.81, b ) 76.30, c ) 98.13,
â ) 116.21
resolutiona (Å)
total observations
unique reflections
25.0-1.71 (1.77-1.71)
440 555
83 273
4.1 (12.2)
98.1 (97.0)
32.7 (10.7)
713
825
0.017
2.0
12.5
42.8
33.0
15.4
R
merge (%)
completeness (%)
I / σI
number of protein residues
number of water molecules
RMSD bond lengths (Å)
RMSD bond angles (deg)
average B-factor for protein
average B-factor for compound 6
average B-factor for solvent
R-factor (%)
1
to give the title compound as a white solid (200 mg, 47%). H NMR
(CD3OD): δ 1.38 (d, 3H, J ) 7 Hz), 1.45 (s, 9H), 1.47 (s, 9H), 1.64-
2.32 (overlapping s and m, 13H), 3.02-3.06 (m, 2H), 4.20-4.28 (m,
3H), 5.15 (s, 2H), 7.36-7.44 (m, 5H). 13C NMR (CD3OD): δ 16.90,
19.73, 21.63, 24.35, 25.77, 28.30, 28.78, 29.11, 31.43, 39.34, 49.84,
52.27, 53.40, 67.74, 81.03, 83.62, 129.31, 129.44, 129.75, 136.83,
159.01, 171.10, 171.63, 172.89, 173.31, 173.82. Mp 48-50 °C. ESI
MS 634.36 (M+, calcd for C32H50N4O9 634.36).
Rfreeb (%)
19.1
b
a Values in parentheses are for the highest-resolution shell. Rfree was
calculated with 5% of the reflections set aside randomly.
164.44, 166.34, 172.36, 172.85, 174.14, 174.20, 174.26, 174.32. ESI
MS 643.25 (M+, calcd for C26H39N6O11S 643.24).
Benzhydryl 7â-[N-Acetyl-L-alanyl-γ-D-glutamyl(R-OtBu)-L-lysine-
(ꢀ-N-Boc)]-3-acetoxymethyl-3-cephem-carboxylate (15). To the solu-
tion of 13 (200 mg, 0.315 mmol) in MeOH was added Pd/C (5%
Degussa, 20 mg), and the mixture was stirred under an ambient pressure
of hydrogen at room temperature overnight. The mixture was filtered,
and the filtrate was concentrated in vacuo to give the crude acid as a
Crystallization and Data Collection. Crystals of Q120L/Y150E
mutant were grown by seeding techniques. Droplets of 10 µL containing
3.9 mg/mL of the protein in 1.0 M potassium phosphate buffer (pH
8.7) were seeded with microcrystals of WT AmpC and placed over a
1.7 M potassium phosphate buffer well solution, pH 8.7. Crystals
appeared within 5-7 days after equilibration at 22 °C and grew to a
maximum size of 0.2 × 0.2 × 0.2 mm3 in a week. Crystals were soaked
in a 50 mM solution of compound 6 (1.7 M KPi, pH 8.7) for 3 h and
then dipped in cryo-protectant solution (50 mM of compound 6 and
20% sucrose in 1.7 M KPi, pH 8.7) for 1 min. The crystals were flash-
cooled in liquid nitrogen. Diffraction data were collected from a single
crystal on the 5ID beamline of DND-CAT at Advance Photon Source
(Argonne, IL). Two sets of data, at high and low resolution, were
collected at a wavelength of 1.0000 Å using an MARCCD detector. A
total of 450 frames were integrated, and 440 555 reflections were scaled
and merged using the HKL package.12 The data are 98.1% complete
to 1.71 Å resolution (Table 1). Crystals belong to the space group C2,
with two AmpC molecules in the asymmetric unit.
1
white solid (177 mg). H NMR (CD3OD): δ 1.35-1.54 (overlapping
s, d, and m, 25H), 1.72-1.23 (overlapping s and m, 6H), 3.05-3.11
(m, 2H), 4.27-4.47 (overlapping bs and m, 4H), 4.95 (bs, 1H), 6.00
(bs, 1H), 7.09 (bs, 1H). This crude product was redissolved in CH2Cl2
(10 mL), and to this solution was added 14 (143 mg, 0.325 mmol),
HOBt (44 mg, 0.325 mmol), and DCC (60 mg, 0.292 mmol). The
resultant mixture was stirred at room temperature overnight. Some CH2-
Cl2 (∼50 mL) was added to dilute the reaction mixture, and the organic
solution was washed with 5% NaHCO3, 10% citric acid, water, and
brine and then was dried over MgSO4. The solvent was removed in
vacuo, and the product was purified by column chromatography (silica
gel, 2:1 EtOAc/hexane, then 15:1 CH2Cl2/2-propanol) to give an off
white powder (86 mg, 28%). Mp 168-170 °C. 1H NMR (CD3OD): δ
1.36 (d, 3H, J ) 7 Hz), 1.44 and 1.46 (2 overlapping s, 18H), 1.66-
2.33 (overlapping s and m, 16H), 3.02-3.05 (m, 2H), 3.51 (d, 1H, J
) 18.5 Hz), 3.64 (d, 1H, J ) 18.5 Hz), 4.23-4.37 (m, 3H), 4.74 (d,
1H, J ) 13 Hz), 4.96 (d, 1H, J ) 13 Hz), 5.11 (d, 1H, J ) 5 Hz), 5.73
(d, 1H, J ) 5 Hz), 6.91 (s, 1H), 7.25-7.44 (m, 10H). 13C NMR (CD3-
OD): δ 16.57, 19.34, 20.57, 22.90, 26.59, 27.08, 27.99, 28.27, 28.41,
31.10, 31.91, 40.06, 49.12, 52.43, 52.91, 57.76, 59.80, 64.25, 78.85,
81.24, 83.62, 126.26, 127.41, 127.85, 128.38, 128.51, 128,79, 128.86,
140.18, 140.43, 157.82, 161.61, 171.45, 171.52, 172.59, 173.59, 174.29,
174.34, 174.38, 174.49. FAB MS 965 (M+ + H, calcd for C48H65N6O13S
965).
Structure Solution and Refinement. The phases were determined
by molecular replacement using AmoRe.13 The best result from the
single body solution was fixed in order to find a second monomer in
the asymmetric unit. The solution of this run was used as a starting
model for rigid body refinement in CNS.14 After torsion angle simulated
annealing and several rounds of Cartesian and B-factor refinement, the
model was manually corrected using 2Fo-Fc and Fo-Fc sigmaA-
weighted electron density maps displayed with TURBO.15 The com-
pound was then fit in the site using Fo-Fc difference electron density.
Further refinement in CNS led to Rcryst and Rfree values of 17.4 and
19.7%, respectively, for all data to 1.71 Å. At this stage, alternative
conformations were added and further refinement was conducted in
REFMAC.16 After several rounds of refinement using TLS parameters17
and manual corrections, the refinement converged to Rcryst and Rfree
values of 15.4 and 19.1%, respectively (Table 1).
7â-[N-Acetyl-L-alanyl-γ-D-glutamyl-L-lysine]-3-acetoxymethyl-3-
cephem-carboxylic Acid TFA Salt (6). A solution of compound 15
(11 mg, 0.0114 mmol) in TFA (1 mL) and anisole (1 drop) was stirred
at ice-water temperature for 10 min, at which time the ice-water bath
was removed and the stirring was continued for another 20 min. The
solvent was removed in vacuo. The product was precipitated from
MeOH/ether to give an off-white powder (9 mg, quant.). Mp 167 °C
(12) Otwinowski, Z.; Minor, W. Methods Enzymol. 1997, 276, 307-326.
(13) Navaza, J. Acta Crystallogr., Sect. A 1994, 50, 157-163.
(14) Bru¨nger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros, P.;
Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, M.; Pannu,
N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren, G. L. Acta
Crystallogr., Sect. D 1998, 54, 905-921.
1
(dec); H NMR (CD3OD): δ 1.32-2.05 (overlapping m and s, 16H),
2.22-2.35 (m, 3H), 2.93 (t, 2H, J ) 7.5 Hz), 3.42 (d, 1H, J ) 18 Hz),
3.62 (d, 1H, J ) 18 Hz), 4.31-4.41 (m, 3H), 4.83 (d, 1H, J ) 13 Hz),
5.04-5.07 (2 overlapping d, 2H), 5.68 (d, 1H, J ) 4.5 Hz). 13C NMR
(CD3OD): δ 16.90, 19.78, 21.63, 22.54, 25.77, 26.75, 27.43, 31.11,
31.66, 39.34, 49.74, 52.27, 53.49, 57.55, 59.33, 63.98, 119.72, 130.28,
(15) Cambillau, C.; Roussel, A. Turbo Frodo, OpenGL Ed. ed.; Universite Aix-
Marseille II: Marseille, France, 1997.
(16) Murshudov, G. N.; Lebedev, A.; Vagin, A. A.; Wilson, K. S.; Dodson, E.
J. Acta Crystallogr., Sect. D 1999, 55, 247-255.
(17) Winn, M. D.; Isupov, M. N.; Murshudov, G. N. Acta Crystallogr., Sect. D
2001, 57, 122-133.
9
9614 J. AM. CHEM. SOC. VOL. 125, NO. 32, 2003

â-Lactamases from Penicillin-Binding Proteins
A R T I C L E S
Scheme 1
Computational Procedures. The X-ray structure of the acyl-
enzyme complex provided the initial coordinates for the molecular
dynamics simulations. Crystallographic waters were retained, and
hydrogen atoms were added to the protein using the “protonate”
program, which is part of the AMBER 718 suite of programs. AMBER
force field parameters and atomic charges were assigned to all atoms
using the “parm99” set of parameters. The Sybyl program (Tripos Inc.,
St. Louis, MO) was used for the manipulation and visualization of all
structures and for the protonation of the bound ligand. The atomic
charges of the bound ligand were determined using the RESP
methodology.19 This consisted of first optimizing the molecules using
the AM1 Hamiltonian, followed by an HF/6-31G* single-point energy
calculation to determine the electrostatic potential around the molecule,
which was subsequently used in the two-stage RESP fitting procedure.
The Gaussian 98 package20 was used to carry out all ab initio
calculations. The acyl-enzyme complex was immersed in a box of
TIP3P21 water molecules such that no atom in the acyl-enzyme
complex was within 12 Å from any side of the box; after solvation of
the 5650-atom acyl-enzyme complex, the system consisted of a total
of 53 830 atoms. All bonds involving hydrogen atoms were constrained
using the SHAKE algorithm, and a 2-fs time step was used. The particle
mesh Ewald22 method was used to treat long-range electrostatics. Water
molecules were first energy minimized and equilibrated by running a
short simulation with the acyl-enzyme species fixed using Cartesian
restraints. This was followed by a series of energy minimizations where
the Cartesian restraints were gradually relaxed from 500 kcal/Å2 to 0
kcal/Å2 and the system was subsequently slowly heated to 300 K via
a 300 ps molecular dynamics run. Another 50 ps simulation was carried
out at 300 K for further equilibration. A 2.6 ns simulation at constant
pressure and temperature (1 atm and 300 K, respectively) was then
carried out on an in-house beowulf cluster. Snapshots were collected
every 0.2 ps.
Results and Discussion
Cephalosporin 6 was conceived as a probe in exploring the
potential surface interactions of peptidoglycan with the class C
â-lactamase. The â-lactam nucleus of the cephalosporin mimics
the two D-Ala residues of the peptidoglycan (the portions traced
in red). The acyl moiety at the C7 amine (shown in blue) traces
most of the remainder of the peptide in the peptidoglycan. The
combined red and blue portions of the structure of 6 represent
the strand of peptidoglycan that would acylate the active-site
serine (i.e., E-OH). Species 7, in essence, depicts the first
peptidoglycan strand that has acylated the active-site serine. On
active-site acylation, the “terminal D-Ala residue” from the
strand of the peptidoglycan would be eliminated (the portions
of 7 shown in red).
(18) Case, D. A.; Pearlman, D. A.; Caldwell, J. W.; Cheatham, T. E., III; Wang,
J.; Ross, W. S.; Simmerling, C. L.; Darden, T. A.; Merz, K. M.; Stanton,
R. V.; Cheng, A. L.; Vincent, J. J.; Crowley, M.; Tsui, V.; Gohlke, H.;
Radmer, R. J.; Duan, Y.; Pitera, J.; Massova, I.; Seibel, G. L.; Singh, U.
C.; Weiner, P. K.; Kollman, P. A. AMBER, 7th ed.; University of
California: San Francisco, 2002.
(19) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. J. Chem. Phys.
1993, 97, 10269.
(20) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Marti, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head-
Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian Inc.: Pisttsburgh, PA,
1998.
Compound 6 was synthesized according to Scheme 1. The
tripeptide 13 was constructed by standard peptide coupling
procedures. After the removal of the benzyl group from this
peptide, it was coupled to cephalosporin 14 to give 15. A global
deprotection by TFA afforded compound 6.
To explore how compound 6 would be recognized by a class
C â-lactamase, the crystal structure of its complex with the
Q120L/Y150E mutant of AmpC was determined to 1.71 Å. This
mutant is deacylation-deficient,11 and it allowed us to trap the
(21) Jorgensen, W. L.; Chandraseklar, J.; Maduar, J. D.; Impey, R. W.; Klein,
M. L. J. Chem. Phys. 1983, 79, 926-935.
(22) Darden, T. A.; York, D. M.; Pedersen, L. G. J. Chem. Phys. 1993, 98,
10089.
9
J. AM. CHEM. SOC. VOL. 125, NO. 32, 2003 9615

A R T I C L E S
Meroueh et al.
in correlated movement to accommodate repositioning of species
7 in the active site, and they are also modeled with two
alternative conformations. The conformation of the ligand in
the active site of monomer A is very similar to one of the
alternative conformations of the compound in monomer B
(Figure 1A). Species 7 makes several hydrogen bonds with
active site residues (Figure 1B). The carbonyl oxygen of the
ester moiety is in the “oxyanion” hole, interacting with amide
nitrogens of Ser64 and Ala318. The carbonyl oxygen of the
lysine in the C7 side-chain hydrogen bonds with side-chain
nitrogen of Asn152. Finally, one of the carboxylate oxygens of
the C7 side-chain hydrogen bonds with the Oγ and the main-
chain nitrogen of Ser211.
An intriguing feature of the complex between AmpC and
species 7 was the high degree of movement observed in the
ligand, especially in the distal parts of the C7 side chain.
Although movement in the distal regions of ligands has been
observed in other AmpC/ligand complexes,23-25 it was precisely
this region of the ligand that was most informative regarding
substrate recognition and differences from PBPs. We wondered
if the movement observed in the crystal structure was in some
sense an artifact of crystallography, for instance reflecting
disorder in the crystal or problems with trapping in the
deacylation-deficient mutant, or whether it represented a true
aspect of molecular recognition between the AmpC enzyme and
this substrate. To investigate this question, we turned to the
molecular dynamics simulation of the AmpC/7 complex.
Molecular dynamics simulations provided insights into the
conformational states as a function of time. Over the course of
the 2.6 ns of simulation, snapshots were collected and super-
imposed on the initial X-ray structure. The resulting structures
were used to determine the root-mean-square (rms) deviation
and atomic fluctuations, which provide a measure of the
flexibility for various regions of the protein and ligand.
Evolution of the rms deviation with respect to time of the
CR atoms of the protein and of species 7 are shown in Figure
2A. The rms deviation of the protein is found not to exceed 1.5
Å, consistent with the crystallographic result of a stable protein
structure. The rms deviation value of the ligand, on the other
hand, fluctuated around 2 Å for the first 1.4 ns of simulation,
which subsequently gradually increased reaching a maximum
of 4.2 Å at 1.8 ns. This increase indicates that the ligand is
prone to substantial movement in the active site of the protein.
Figure 2B depicts more closely the dynamics of different parts
of the ligand. The portion of the ligand highlighted in red was
found to be the most mobile, with its rms deviation reaching
2.3 Å, which is more pronounced than the 1.3 and 0.9 Å rms
deviations experienced by the lysine (in blue) and the dihy-
drothiazine ring, highlighted in green. To illustrate the range
of movement that the ligand experiences in the active site of
the â-lactamase, 26 structures were collected at 100 ps intervals
and are shown superimposed in Figure 2C. The L-Ala-γ-D-Glu
region of the ligand (white arrows in Figure 3C) shows
significant movement and samples a wide variety of conforma-
tions. At the initial stages of the simulation, the acetyl carbonyl
oxygen of the ligand was pointing toward the guanidinium
moiety of Arg-201, and the carboxylate moiety of D-Glu was
Figure 1. Stereoview of the active site of the complex between AmpC
and compound 6 (species 7). (A) Electron density maps of the active site
of the monomer B with two alternative conformations of the compound.
The 2Fo-Fc sigmaA-weighted electron density map for protein residues is
shown in blue (contoured at 1 σ), and the Fo-Fc omit electron density
map for the compound is shown in green (contoured at 2.5 σ). Selected
residues of the protein are labeled. One of the alternative conformations is
shown colored as follows: carbon in yellow, nitrogen in blue, oxygen in
red, and sulfur in green, whereas all atoms of the second conformation are
shown in cyan. (B) Hydrogen-bond interactions between selected active-
site residues with one of the alternative conformations of the compound
(nitrogen in blue, oxygen in red, ligand carbon atoms in green, protein carbon
atoms in yellow, and sulfur in magenta).
AmpC/6 complex (species 7). Evaluation of the structure by
Procheck23 showed that all residues are in the most favored and
additionally allowed regions of the Ramachandran plot. The
refined model is well fit in 2Fo-Fc electron density, except
three residues at positions 284-286 of monomer A, which have
disconnected density and were removed from the final model.
The two monomers in the structure are very similar, except for
this loop region; the overall RMSD is 0.86 Å for all CR atoms
and is 0.35 Å with the 280-290 loop omitted.
The electron density in the active sites of both monomers
indicates the presence of species 7 covalently bound to the Oγ
of the Ser64. Admittedly, this density is “patchy,” which might
indicate low occupancy, multiple ligand conformations, or both.
Despite this “patchiness,” there are clear indications that species
7 is present. There is strong electron density for atoms all along
the length of species 7, from the dihydrothiazine sulfur to the
ester carbonyl, to the C7 side-chain amide group and large
portions of the distal C7 peptide, including the lysine moiety.
The relatively high resolution of the structure (1.71 Å) allowed
us to adopt a combined occupancy and conformation strategy
in fitting the ligand to the observed electron density. In monomer
A, species 7 is modeled with half occupancy and it shares the
same space with several water molecules, also modeled with
partial occupancy. In monomer B, the ligand is modeled in two
alternative conformations with an occupancy of 0.5 each (Figure
1A). Several side chains of the active site appear to be involved
(24) Tondi, D.; Powers, R. A.; Caselli, E.; Negri, M. C.; Blazquez, J.; Costi,
M. P.; Shoichet, B. K. Chem. Biol. 2001, 8, 593-611.
(25) Powers, R. A.; Caselli, E.; Focia, P. J.; Prati, F.; Shoichet, B. K.
Biochemistry 2001, 40, 9207-9214.
(23) Lobkovsky, E.; Billings, E. M.; Moews, P. C.; Rahil, J.; Pratt, R. F.; Knox,
J. R. Biochemistry 1994, 33, 6762-6772.
9
9616 J. AM. CHEM. SOC. VOL. 125, NO. 32, 2003

â-Lactamases from Penicillin-Binding Proteins
A R T I C L E S
Figure 3. (A) Stereoview of the Streptomyces R61 DD-carboxypeptidase/
transpeptidase acylated by 16. A water-accessible surface area (shown in
green) depicts the active site and species 17 is represented according to
atom types (white, blue, red, and yellow correspond to C, N, O, and S,
respectively). The blue and red arrows point to the portions of the species
17 colored in blue and red, respectively. (B) Stereoview of AmpC acylated
by compound 6 with the water-accessible surface within the active site
depicted in green. The ligand is depicted in the capped-stick representation
and is colored according to atom types. The white arrow is pointing to the
exocyclic methylene of species 7. A translucent orange/red water-accessible
surface was constructed for the inserted peptide spanning residues 285 to
296.
Figure 2. (A) Root-mean-square (rms) deviation from the X-ray structure
of the CR carbon atoms of the protein (red) and non-hydrogen atoms of
species 7 (blue). (B) Root-mean-square deviation of different portions of
the ligand from the X-ray structure. The inset shows a schematic of species
7 with colors corresponding to the root-mean-square deviations from the
plot. (C) A stereoview of the superimposition of 26 snapshots collected
from the 2.6 ns simulation at 100 ps intervals. Species 7 is shown in capped-
sticks representation and color coded according to atom types (C, N, O,
and S are in white, blue, red, and yellow, respectively). The protein backbone
is shown as an orange tube. The white arrow points to the L-Ala-γ-D-Glu
portion of species 7, while the red and yellow arrows point to lysine and
the six-membered ring, respectively. The blue arrow points to the inserted
loop specific to class C â-lactamases that prevents binding of the second
strand of the peptidoglycan to the active site.
demonstrates within the active site, which includes conformers
1 and 2, both seen by the X-ray diffraction data. The fact that
this portion of the ligand was found to be highly mobile in the
molecular dynamics simulation is also consistent with the X-ray
diffraction data, which indicated that electron density in that
region was such that the structure could not be unambiguously
resolved into one conformer.
Movements in the other regions of species 7 were more
attenuated, as shown by the rms deviations in Figure 2B and as
depicted in Figure 2C. The lysine residue (red arrow, Figure
2C) does not undergo any major conformational change over
the course of the trajectory. The ligand ring (yellow arrow in
Figure 2C) also does not show any large conformational change,
although it is of interest to note that most of the fluctuations
occurs near the carboxylate moiety of the ring.
We recently reported the synthesis of compound 16 and its
evaluation by X-ray in a complex with a bifunctional DD-
transpeptidase/DD-peptidase.26 The portion in blue of cepha-
losporin 16 mimics the first strand of peptidoglycan that acylates
the active-site serine of the PBP. On enzyme acylation, the
portions in blue and red of species 17 represent the two strands
pointing toward Ser-209 (3.9 Å). The orientation of these
moieties corresponded to that adopted by conformer 1 (see
compound colored according to the atom types in Figure 1A)
in its acyl-enzyme complex with AmpC (conformer 1/AmpC
complex X-ray structure was used as the initial structure for
the molecular dynamics simulation). As the trajectory pro-
gressed, the acetyl moiety gradually flipped over to eventually
occupy a large cavity in the active site (12 o’clock in Figure
2C), while the D-Glu carboxylate occupied the initial position
of the acetyl moiety to form hydrogen bonding interactions with
Arg-201. In the process of flipping, the L-Ala-γ-D-Glu moiety
adopted similar conformations to those found in conformer 2
(see the species colored in cyan in Figure 1A) of the X-ray
structure. The molecular dynamics simulation thus provided a
detailed account of the conformational flexibility that the ligand
(26) Lee, W.; McDonough, M. A.; Kotra, L.; Li, Z. H.; Silvaggi, N. R.; Takeda,
Y.; Kelly, J. A.; Mobashery, S. Proc. Natl. Acad. Sci. U.S.A. 2001, 98,
1427-1431.
9
J. AM. CHEM. SOC. VOL. 125, NO. 32, 2003 9617

A R T I C L E S
Meroueh et al.
of peptidoglycan poised for the formation of the “cross-linked”
cell wall. The 1.2 Å X-ray structure for this complex from the
Kelly laboratory provided the first glimpse of how DD-
transpeptidases may organize two strands of the bacterial
peptidoglycan in their active sites en route to cross-linking.
Figure 3A shows an image of the high-resolution structure of
this complex, with the active site depicted as a Connolly surface
(in green). To the right of the active site, it is open to
accommodate the second strand of the peptidoglycan and the
“wall” at 11 o’clock is the surface for interactions of the first
peptidoglycan. The same perspective is shown for the complex
of species 7 within the active site of AmpC â-lactamase in
Figure 3B, as reported in this manuscript. It is noteworthy that
the “wall” for interactions with the first strand of the pepti-
doglycan is entirely absent in the â-lactamase complex. This
gives a simple structural perspective for why the crystallographic
data and dynamics simulations could not fix the “first strand of
peptidoglycan” in species 7 in an effective binding geometry,
as the enzyme surface here has evolved away from such binding
ability. It is also revealing that the insertion of the peptide within
the active site at 3 o’clock (shown as a translucent orange/red
water-accessible surface in Figure 3B; pointed to by the blue
arrow in Figure 2C) clearly has abolished the ability of a second
strand of the peptidoglycan to bind to this subsite. This inserted
peptide is not known to play any catalytic role in the reaction
of the â-lactamase. The exocyclic methylene group of the six-
membered ring in species 7 is flush against the side of the
insertion peptide (white arrow in Figure 3B), and there is no
room for a larger ligand such as species 17, seen in the active
site of the PBP. These two crystallographic images provide
evidence for the structural reasons behind the lack of recognition
of the peptidoglycan by at least class C â-lactamases, which
must have paved the way for the emergence of a highly
competent catalyst in the destruction of â-lactam antibiotics in
resistant bacteria.
Concluding Remarks. Because of the kinship of â-lacta-
mases and PBPs, much effort has been dedicated to demonstrat-
ing that they could potentially carry out each other’s reaction(s).
This has not been an easy task, although there exist evidence
that some DD-peptidases can hydrolyze â-lactam antibiotics27-29
and some â-lactamases have marginal peptidase activities.30
From the present report, it would appear that at least class C
â-lactamases haVe fully diVested from the ability to recognize
the peptidoglycan. In essence, the utility of cephalosporin 6 was
in the fact that it acylates the active-site serine, forcing
sequestration of the “peptidoglycan strand” (i.e., the C7 sub-
stituent) into the active site. Despite this forced coexistence
within the complex, binding of the surrogate for the peptidogly-
can is not stabilized by the protein. The requisite surface for
the interaction with the peptidoglycan that acylates the active
site serine in PBPs no longer exists in the â-lactamase active
site. Similarly, the peptide insertion within the active site of
â-lactamase abolishes the potential binding site for the second
peptidoglycan in the active site, such as it is in DD-transpep-
tidases. The X-ray structure for the cephalosporin ligand
indicates at least two binding modes. Molecular dynamics
simulations showed that the ligand samples a substantial
conformational space, which incidentally included the two
species seen in the X-ray structure. The evidence presented here
provides structural insight into how this antibiotic resistance
enzyme has divested itself from interaction with the substrate
for the parental enzymes, en route to achieving full catalytic
effectiveness in its intended function in living bacteria.
Acknowledgment. This work was supported by Grants
AI33170 and GM61629 (to S.M.) and by GM63815 (to B.K.S.)
from the National Institutes of Health. We thank Alexandera
Patera for assistance with initial crystallography and for protein
expression. The X-ray crystal structure of compound 6 in
complex with the mutant AmpC has been deposited with the
PDB, accession number 1O07.
JA034861U
(27) Davies, C.; White, S. W.; Nicholas, R. A. J. Biol. Chem. 2001, 276, 616-
623.
(28) Livermore, D. M. J. Antimicrob. Chemother. 1987, 19, 733-742.
(29) Deka, R. K.; Machius, M.; Norgard, M. V.; Tomchick, D. R. J. Biol. Chem.
2002, 277, 41857-41864.
(30) Rhazi, N.; Galleni, M.; Page, M. I.; Frere, J. M. Biochem. J. 1999, 341 (Pt
2), 409-413.
9
9618 J. AM. CHEM. SOC. VOL. 125, NO. 32, 2003