Elucidation of mechanism of inhibition and X-ray structure of the TEM-1 β-lactamase from Escherichia coli inhibited by a N-sulfonyloxy-β-lactam

Article abstract of DOI:10.1021/ja990400q

Class A β-lactamase are inactivated by a novel type of monocyclic ?- lactams described recently (J. Am. Chem. Soc. 1995, 117, 5938). A compound of this class, (±)-trans-1-N-(tosyloxy)-3-(1-hydroxyethyl)-4-phenyl-2- azetidinone, is synthesized, is shown to acylate the active site of the TEM- 1 β-lactamase from Escherichia coli rapidly, and resists deacylation for several days. The crystal structure of the enzyme-inhibitor complex was determined at 1.95 A? resolution. The features of the three-dimensional structure of this acyl-enzyme species and mechanistic studies revealed that a fragmentation of the inactivation ensued on acylation of the active-site serine and that the ester carbonyl oxygen was outside the oxyanion hole. This ester carbonyl makes a strong hydrogen bond to the protonated form of the side chain of Glu-166, the general base for deacylation of the typical acyl- enzyme intermediates in the normal catalytic process. Furthermore, interactions within the active site mandated the existence of the former β- lactam amine as an imine or a ketone, and not as an enamine or an enol, and shed light on the unique mechanism of action of these enzyme inactivators. This type of inactivator holds the promise of application for inhibition of other enzymes.

Full text of DOI:10.1021/ja990400q

VOLUME 121, NUMBER 23
J UNE 16, 1999
© Copyright 1999 by the
American Chemical Society
Elucidation of Mechanism of Inhibition and X-ray Structure of the
TEM-1 â-Lactamase from Escherichia coli Inhibited by a
N-Sulfonyloxy-â-lactam
Peter Sware´n,† Irina Massova,‡ John R. Bellettini,§ Alexey Bulychev,‡ Laurent Maveyraud,†
Lakshmi P. Kotra,‡ Marvin J. Miller,§ Shahriar Mobashery,*,‡ and Jean-Pierre Samama*,†
Contribution from the Groupe de Cristallographie Biologique, Institut de Pharmacologie et de Biologie
Structurale du CNRS, 205 route de Narbonne, 31077-Toulouse Cedex, France, Department of Chemistry,
Wayne State UniVersity, Detroit, Michigan 48202, and Department of Chemistry and Biochemistry,
UniVersity of Notre Dame, Notre Dame, Indiana 46556
ReceiVed February 8, 1999
Abstract: Class A â-lactamases are inactivated by a novel type of monocyclic â-lactams described recently
(J. Am. Chem. Soc. 1995, 117, 5938). A compound of this class, (()-trans-1-N-(tosyloxy)-3-(1-hydroxyethyl)-
4-phenyl-2-azetidinone, is synthesized, is shown to acylate the active site of the TEM-1 â-lactamase from
Escherichia coli rapidly, and resists deacylation for several days. The crystal structure of the enzyme-inhibitor
complex was determined at 1.95 Å resolution. The features of the three-dimensional structure of this acyl-
enzyme species and mechanistic studies revealed that a fragmentation of the inactivator ensued on acylation
of the active-site serine and that the ester carbonyl oxygen was outside the oxyanion hole. This ester carbonyl
makes a strong hydrogen bond to the protonated form of the side chain of Glu-166, the general base for
deacylation of the typical acyl-enzyme intermediates in the normal catalytic process. Furthermore, interactions
within the active site mandated the existence of the former â-lactam amine as an imine or a ketone, and not
as an enamine or an enol, and shed light on the unique mechanism of action of these enzyme inactivators.
This type of inactivator holds the promise of application for inhibition of other enzymes.
The hydrolytic action of â-lactamases is the primary cause
of bacterial resistance to â-lactam antibiotics.1-4 These enzymes
have been selected to be present in many pathogens, and the
use of a wide range of these antibiotics in the clinic has
facilitated evolution of the enzymes for expression of many
phenotypic resistance profiles.1-4 There are over 250 known
â-lactamases, of which one group, the class A enzymes, are
the most common.3 In light of their prevalence, inhibitors of
class A â-lactamases are highly sought, and three such inhibitors,
clavulanic acid, sulbactam, and tazobactam,3 have been intro-
duced to the clinic. However, an increasing number of variants
of â-lactamases resistant to the inhibition process have been
identified since 1992.5 It has been demonstrated that the
mechanism of inactivation of these enzymes by all three
† Institut de Pharmacologie et de Biologie Structurale du CNRS.
‡ Wayne State University.
§ University of Notre Dame.
(1) Massova, I.; Mobashery, S. Antimicrob. Agents Chemother. 1998,
42, 1.
(2) Medeiros, A. A. Clin. Infect. Dis. 1997, 24 (Suppl 1), S19.
(3) Bush, K.; Mobashery, S. In ResolVing the Antibiotic Paradox:
Progress in Understanding Drug Resistance and DeVelopment of New
Antibiotics; Rosen, B. P., Mobashery, S., Eds.; Plenum Press: New York,
1998; pp 71-98.
(4) Bush, K.; Jacoby, G. A.; Medeiros, A. A. Antimicrob. Agents
Chemother. 1995 39, 1211. Matagne, A.; Lamotte-Brasseur, J.; Fre´re, J.
M. Biochem. J. 1998, 330, 581.
10.1021/ja990400q CCC: $18.00 © 1999 American Chemical Society
Published on Web 05/27/1999

5354 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Sware´n et al.
compounds is similar,6 and variants of these enzymes may thus
evolve, which would, in principle, give cross-resistance to all
three inactivators.5 Therefore, new â-lactamase inhibitors that
function by distinct mechanisms are highly desirable.
Scheme 1
We have reported our findings on a novel class of mono-
bactam inhibitors for class A â-lactamases.7 These molecules
have the general structure shown below (1). This class of
compounds distinguishes itself in its ability to inactivate the
enzyme exceedingly rapidly. Compound 2, a prototype inhibitor
Results and Discussion
of this kind, acylates the active site of the enzyme in seconds.
The immediate acylated species (3) undergoes rearrangement
to give the inhibited species (4), following the release of the
good leaving group. Recovery of the enzyme activity occurred
within 2-3 h, and we suggested that this species resists
deacylation in light of the fact that it is an R,â-unsaturated ester.
In this report, we have explored the details of the mechanism
of action of this type of enzyme inhibitors by kinetics measure-
ments and X-ray structure determination of the TEM-1 â-lac-
tamase-inhibitor complex. The study was aided by the design
and synthesis of inactivator 5 that resists deacylation from the
One of the best inhibitors of this class that we had studied
previously was compound 2. This compound modified the active
site Ser-70 of the TEM-1 â-lactamase rapidly [kinact/Ki ) (1.12
( 0.08) × 104 M-1 s-1].7 To stabilize the acyl-enzyme
intermediate and thus improve the inhibitory properties, we
incorporated the hydroxyethyl group in the structure of inacti-
vator 2 to arrive at compound 5. Indeed, earlier work by others8
and by us9,10 had indicated that the presence of the 1(R)-R-
hydroxyethyl group adjacent to the lactam carbonyl of other
classes of â-lactams imparted substantial stability to the acyl-
enzyme species. The magnitude of this effect on turnover kinetic
parameters could be as much as 50 000-fold,9 and the effect
stems from the stereoelectronic interactions of 1(R)-R-hydroxy-
ethyl group within the active site. Compound 5 was therefore
expected to resist strongly the deacylation process.
Scheme 1 shows the synthesis for compound 5. The synthesis
commenced with the reduction of methyl acetoacetate (6) by
sodium borohydride in methanol to afford methyl 3-hydroxy-
butanoate (7) in 71% yield. Treatment of 7 with 2 equiv of
LDA, to form the dianion, followed by the addition of
benzaldehyde afforded a mixture of diastereomeric aldol
products 8 (four diastereomers as a 39:33:24:4 mixture) in 74%
yield. Saponification of the methyl esters of the mixture of
diastereomers using 1.0 M NaOH afforded the corresponding
carboxylic acids, which were coupled with O-allylhydroxyl-
amine using the water-soluble carbodiimide EDC-HCl to
affordhydroxamates 9 in 48% yield. The mixture was used
directly in the next step. The â-hydroxy hydroxamates were
cyclized to the corresponding â-lactams using Mitsunobu
conditions11 in the presence of di-tert-butyl azodicarboxylate
(DBAD) and triphenylphosphine (PPh3) to afford a diastereo-
meric mixture of four â-lactams, which were separated by
column chromatography to obtain the pure isomers 10-13. Only
the relative stereochemistry at the hydroxyl bearing carbon of
active site of the enzyme for several days. This inactivator was
used to modify the TEM-1 â-lactamase from Escherichia coli,
the parent enzyme of a group of 69 class A â-lactamases (as of
May 1999). The crystal structure of the complex is described
herein, which helped elucidate the mechanism of action of the
inactivator.
(5) Vedel, G.; Belaaouaj, A.; Gilly, L.; Labia, R.; Philippon, A.; Ne´vot,
P.; Paul. G. J. Antimicrob. Chemother. 1992, 30, 449. Blazquez, J.; Baquero,
M. R.; Canton, R.; Alos, I.; Baquero, F. Antimicrob. Agents Chemother.
1993, 37, 2059. Belaaouaj, A.; Lapoumeroulie, C.; Canica, M. M.; Vedel,
G.; Ne´vot, P.; Krishnamoorthy, R.; Paul. G. FEMS Microbiol. Lett. 1994,
120, 75. Farzaneh, S.; Chaibi, E. B.; Peduzzi, J.; Barthelemy, M.; Labia,
R.; Blazquez, J.; Baquero, F. Antimicrob. Agents Chemother. 1996, 40, 2434.
Bret, L.; Chaibi, E. B.; Chanal-Claris, C.; Sirot, D.; Labia, R.; Sirot, J.
Antimicrob. Agents Chemother. 1997, 41, 2547. Bermudes, H.; Jude, F.;
Arpin, C.; Quentin, C. Antimicrob. Agents Chemother. 1997, 41, 222.
Canica, M. M.; Lu, C. Y.; Krishnamoorthy, R.; Paul, G. C. J. Mol. EVol.
1997, 44, 57. Sirot, D.; Recule, C.; Chaibi, E. B.; Bret, L.; Croize, J.; Chanal-
Claris, C.; Labia, R.; Sirot, J. Antimicrob. Agents Chemother. 1997, 41,
1322. Canica, M. M.; Caroff, N.; Barthelemy, M.; Labia, R.; Krishnamoor-
thy, R.; Paul, G.; Dupret, J. M. Antimicrob. Agents Chemother. 1998, 42,
1323. Vakulenko, S. B.; Geryk, B.; Kotra, L. P.; Mobashery, S.; Lerner, S.
A. Antimicrob. Agents Chemother. 1998, 43, 1542. A Web site gives updated
information on â-lactamases and their clinical variants: http://www.lahey-
.org/studies/webt.htm.
(8) Cama, L. D.; Wildonger, K. J.; Guthikonda, R.; Ratcliffe, R. W.;
Christensen, B. G. Tetrahedron 1983, 39, 2513. Cama, L.; Christensen, B.
G. J. Am. Chem. Soc. 1978, 100, 8006. Shih, D. H.; Hannah, J.; Christensen,
B. G. J. Am. Chem. Soc. 1978, 100, 8004. Iwata, H.; Tanaka, R.; Ishiguro,
M. J. Antibiot. 1990, 43, 901. Murakami, M.; Aoki, T.; Matsuura, M.;
Nagata, W. J. Antibiot. 1990, 43, 1441.
(9) Taibi, P.; Mobashery, S. J. Am. Chem. Soc. 1995, 117, 7600.
(10) Miyashita, K.; Massova, I.; Mobashery, S. Bioorg. Med. Chem. Lett.
1996, 6, 319.
(6) Massova, I.; Mobashery, S. Acc. Chem. Res. 1997, 30, 162.
(7) Bulychev, A.; O’Brien, M. E.; Massova, I.; Teng, M.; Gibson, T.
A.; Miller, M. J.; Mobashery, S. J. Am. Chem. Soc. 1995, 117, 5938.
(11) Miller, M. J. Acc. Chem. Res. 1986, 19, 49.

Inhibition and X-ray Structure of TEM-1 â-Lactamase
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5355
the hydroxyethyl side chain of each compound could not be
assigned unambiguously. The stereochemistry of compound 5
at that site was assigned from the interactions observed in the
X-ray structure of compound 5 in the TEM-1 â-lactamase
complex. We thought that complexation of phosphorus with diol
9 would possibly allow for some selectivity for cyclization at
the benzylic position over the secondary position; however, this
selectivity was not observed. The regiochemistry of the cy-
clization products was determined using 2-D COSY NMR
experiments. In compounds 10 and 11, the proton that is coupled
to the hydroxyl proton showed coupling with the C-3 hydrogen
of the â-lactam ring. In contrast, in compounds 12 and 13, the
proton coupled to the hydroxyl proton showed coupling with
the methyl group on the C-3 substituent in addition to the C-3
hydrogen of â-lactam ring. The stereochemistry on the â-lactam
ring was determined from the coupling constants of the C-3
and C-4 protons on the ring. The stereochemistry of the site of
attachment of the hydroxyl group could not be determined at
this stage. â-Lactam 12 was allowed to react with palladium
diacetate [Pd(OAc)2] and triphenylphosphine in the presence
of pyrrolidine (as the allyl cation scavenger) to afford the
N-hydroxy â-lactam, which was treated with p-toluenesulfonyl
chloride in the presence of triethylamine to afford N-tosyloxy
â-lactam 5 in 37% yield. We noticed that, after approximately
3 days of storage of â-lactam 5 at room temperature, it was
contaminated with a rearrangement product, which we attribute
to structure 14, although the structure assignment is equivocal.
If our notion for the structure of this rearrangement product is
correct, the tosylate group may first transfer to the free hydroxyl
group, and then the resulting N-hydroxy â-lactam may undergo
a known rearrangement12 to afford 14.
Compound 5 inactivated the TEM-1 â-lactamase rapidly. The
rate of this process could not be attenuated effectively in
competition with substrates to permit evaluation of individual
parameters for enzyme inhibition. However, we have used three
parameters in evaluation of the kinetic properties of this
compound with this enzyme. The second-order rate constant
for inactivation (kinact/Ki) was assessed for the inactivation
process; the larger the number, the faster the process. This
parameter was evaluated at (3.20 ( 0.15) × 103 M-1 s-1. The
partition ratio (kcat/kinact) was evaluated to be 7, which is an
expression of the efficiency of the inactivation process; that is
to say, seven molecules were hydrolyzed before the onset of
enzyme inactivation. We evaluated the ease for recovery of
activity (krec), as a measure of the longevity for the inactive
enzyme species; the smaller this rate constant, the better the
inactivator. As expected, we observed exceedingly slow deacyl-
ation for the inactivated enzyme complex, and the recovery of
activity was biphasic. A relatively rapid phase accounting for
only 8% of the recovery of activity could be discerned only in
the presence of high concentrations of substrates (not seen in
Figure 1) and proceeded with a rate constant comparable to those
for the other monobactams of this type [krec ) (1.6 ( 0.4) ×
10-3 s-1]. However, the slower phase for deacylation took place
over a remarkably long period of several days (Figure 1), which
incidentally allowed for recovery of no more than 15% of
activity in 6 days [krec ) (1.3 ( 0.2) × 10-5 s-1]. The rate
differential for the two phases of the recovery of activity is over
120-fold. As will be described subsequently in this manuscript,
there are two acyl-enzyme species for the inhibited complex.
One is an immediate acyl-enzyme species with the ester
carbonyl housed in the active-site oxyanion hole, and the other
Figure 1. Activity profile of the TEM-1 â-lactamase as a function of
time in the presence of compound 5.
(seen in the X-ray structure) has the ester carbonyl displaced
from the oxyanion hole. We suggest that the first, the immediate
acyl-enzyme species, is the one that gives the more rapid rate
constant for deacylation, and the other, the one with the carbonyl
out of the oxyanion hole, accounts for the slower step for
deacylation.
The monobactam inactivators are recognized by the TEM-1
â-lactamase, whereby they form a noncovalent preacylation
complex. Enzyme acylation takes place rapidly to give the
immediate acylated species (such as 3), and a species such as
4 would account for enzyme inactivation.7 We noted that, on
inactivation of the enzyme by compounds 2 and 5, the tosylate
was eliminated quantitatively but that no chromophore formation
was seen with compound 5, in contrast to compound 2.
Structures 15-18 show the possibilities for the nature of the
acyl-enzyme species of the TEM-1 â-lactamase with com-
pound5. The atoms of structures 15 and 17 are numbered
arbitrarily (atoms 1-13). In light of the fact that the chro-
mophore at 276 nm for the R,â-unsaturated ester7 was not seen
with the use of compound 5, we concluded that structures 15
and 17 may be unlikely. Hydrolysis of the iminium group of
16, the immediate product of the loss of the leaving group,
would generate 18, and either 16 or 18 may account for the
collective data for experiments in solution. These species are
simply tautomers of 15 and 17, respectively.
Crystals of the inactivated enzyme, produced by soaking the
native TEM-1 â-lactamase crystals with inhibitor 5, were
sensitive to X-ray exposure. Crystals diffracted weakly when
data collection was performed at -10 °C, but the diffraction
pattern was much improved when crystals were flash-cooled at
100 K. A 1.95 Å data set was collected under these conditions
(Table 1; Experimental Section). Crystals belonged to the space
(12) Zercher, C. K.; Miller, M. J. Tetrahedron Lett. 1989, 30, 7009.
Hirose, T.; Chiba, K.; Mishio, S.; Nakano, J.; Uno, H. Heterocycles 1982,
19, 1019.

5356 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Table 1. Data Processing and Refinement Statistics
Sware´n et al.
15 and 17 cannot be accommodated by the information from
the X-ray structure, consistent with the findings from the
biochemical studies (discussed earlier). On the other hand, the
electron density corresponding to the C4-C5 bond, to the C7,
C9, and C13 atoms, and to the hydroxyl and methyl groups of
the 6R substituent agree with the binding of species 16 and/or
18 to the protein. Based on hydrogen bond and van der Waals
patterns, the stereochemistry at carbon 6 is S. The hydroxyl
group is found at 2.7 Å from Ser-130 Oγ and the methyl group
at 3.5 Å from the phenyl ring of Tyr-105.
The absence of a well-defined electron density corresponding
to the iminium/ketone group and to the phenyl ring, the value
of the refined occupancy of the bound inhibitor (0.74), the
increasing mobility according to the temperature factors from
atom 1 to 5 (20.2 to 36.9 Å2), and the unusual sensitivity of the
TEM crystals upon complex formation with compound 5
collectively suggest that the structure of the inhibited species
in the crystal is not uniform. These observations would be in
line with the coexistence of species 18 along with 16, in the
modified enzyme. The iminium moiety of 16 would be located
at 2.4 Å to the main-chain oxygen atom of Ala-237 and at 2.9
overall resolution
range 35.0-1.95
highest resolution
shell 2.00-1.95
no. of observns
no. of ind reflns
completeness (%)
Rmerge
I/σI
48398
14329
88.0
0.043
24.9
3636
1122
96.4
0.082
16.1
a
R
factor: 0.206
Rfree: 0.258
2014 protein atoms
349 solvent atoms
( B ) 19.5 Å2)
( B ) 32.1 Å2)
rms deviation
bond length: 0.015 Å
dihedral angles: 21.8°
bond angles: 1.3°
improper angles: 0.75°
a Rmerge ) ∑∑| I - Ii|/∑∑Ii.
group P212121, with cell parameters a ) 40.94 Å, b ) 59.31
Å, and c ) 87.58 Å.
The first cycles of refinement, which included a rigid body
optimization as the initial step, were performed using reflections
from 20 to 1.95 Å and the native TEM-1 â-lactamase structure
as the initial model. The Rfactor and the Rfree values dropped from
0.55 to 0.23 and from 0.56 to 0.29, respectively. The electron
density maps computed at this stage showed a continuous
electron density in the active site extending from Ser-70 Oγ,
which indicated binding of the inactivator as an acyl-enzyme
complex. In the last rounds of refinement, the significant electron
densities found in the oxyanion hole and in the vicinity of Arg-
244 were assigned as a water molecule and a sulfate ion,
respectively. Atoms of the inhibitor were introduced at this
stage. All protein atoms, except a few solvent-exposed side
chains, were visible in the final electron density map, which
allowed inclusion of 349 water molecules and one sulfate ion.
The final Rfactor was 0.206 (Rfree ) 0.258 ) for all reflections
between 20 and 1.95 Å (Table 1).
Electron density beyond atom 5 of the inactivator bound in
the active site was present from the first step of refinement.
However, it did not significantly improve in the next steps in
order to allow an unambiguous model building of the atom at
position 8 and of the phenyl group. Such an observation is
usually interpreted to be the result of the motion of the chemical
groups. This clearly applies to our case, as will be elaborated
below.
1
Å to the Oδ atom of the Asn-170 side chain (Scheme 2). These
interactions would not be compatible with the same binding of
species 18, which possesses an oxygen atom at position 8. A
rotation about the C4-C5 bond, which is only possible for
species 16 and 18, likely explains the weak electron density
for atoms beyond atom 5.
Despite acylation by the inhibitor, the TEM-1 enzyme still
binds a water molecule in the oxyanion hole, at 2.9 and 2.4 Å
to the main-chain nitrogen atoms of residues 70 and 237,
respectively. In contrast to the inhibitor, this water molecule
has the same occupancy (1.0) and temperature factor (15.2 Å2)
as the neighboring protein atoms. It occupies a position similar
to that found in the unmodified enzyme. This observation
indicates that, after acylation of Ser-70 by the inhibitor and the
motion of the ester carbonyl out of the oxyanion hole, a water
molecule moved in to reoccupy the void in the oxyanion hole.
An interesting comparison could be made between the binding
modes of the two acyl-enzyme intermediates of imipenem (a
carbapenem antibiotic) and of compound 5.13 In the X-ray
structure of the TEM-imipenem complex, the carbonyl ester
was also found outside the oxyanion hole, but oriented toward
Ser-130 Oγ. However, no water molecule was found in the
oxyanion hole because of the dynamic motion of the oxygen
atom of the carbonyl ester in and out of this site, as demonstrated
by molecular dynamics simulations.13 Such a dynamic situation
does not appear to be the case for the inhibitory species of 5
seen in the X-ray structure. The tetrahedral electron density in
the vicinity of Arg-244 was attributed to a sulfate ion supplied
by the crystallization medium. It was located at 2.5-3.1 Å to
The protein structure in the complex is very similar to the
structure of the native enzyme. A global superimposition of both
structures based on the main-chain atoms led to a rms deviation
of 0.4 Å. Atoms O1 (Ser-70 Oγ), C2, O3, C4, and C6 of the
inhibitor and the hydroxyl and methyl substituents on C6 are
common to the four plausible species that may account for
inhibition of the enzyme (structures 15-18). The electron
density showed that Ser-70 Oγ is acylated and that the ester
oxygen atom O3 is not in the oxyanion hole. The ester carbonyl
has displaced the hydrolytic water molecule and is located at
2
the Nú atom of Lys-234, the Oγ atom of Ser-235, and the Nη
of Arg-244. This region of the protein is characterized by a
significant positive electrostatic potential and may provide, in
solution, a binding site for the sulfonate moiety of 5 on its
departure from the immediate acyl-enzyme intermediate. We
hasten to add that the presence of the sulfate ion is not required
for enzyme inactivation, since all of the kinetics experiments
were performed in its absence.
1
2.8 Å from the Oꢀ oxygen atom of Glu-166, which indicates
that the side-chain function of this amino acid should be
protonated in the complex (Figure 2A and Scheme 2; the
distances shown in the scheme for hydrogen bonds are for those
between the two heteroatoms).
The structural constraints brought about by the planar
character of the potential inactivated species 15 and 17 would
bring the hydroxyl or the amino group, respectively, at atom
position 8, to an unfavorable 2.0 Å distance from the side chain
of Asn-170. The electron density of Asn-170 is well-defined,
and according to the structure refinement, no alternate confor-
mation for its side chain may be postulated. Hence, structures
A number of features about this inhibitor are novel. First,
acylation of the active-site serine by our inactivator is driven
by the highly favorable steric and electrostatic interactions of
the inhibitor with the active-site functions. A key interaction
(13) Maveyraud, L.; Mourey, L.; Kotra, L. P.; Pedelacq, J.-D.; Guillet,
V.; Mobashery, S. and Samama, J. P. J. Am. Chem. Soc. 1998, 120, 9748.

Inhibition and X-ray Structure of TEM-1 â-Lactamase
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5357
Figure 2. (A) Stereoview of the crystal structure of the TEM-1 â-lactamase from E. coli modified by 5. The final [3Fobs - 2Fcalc, Φcalc] electron
density map has been contoured at 1 σ (20-1.95 Å). (B) Stereoview of the energy-minimized structure for the immediate acyl-enzyme intermediate
of compound 5 with the TEM-1 â-lactamase from E. coli. The arrow shows the carbonyl moiety of the acyl-enzyme intermediate. In the immediate
acyl-enzyme intermediate (image B), the ester carbonyl occupies the oxyanion hole, and in the X-ray structure (image A), this moiety moved out
of the oxyanion hole and a water molecule (shown as spheres at 3 o’clock) occupies this site.
Scheme 2
inhibitor in the active site. The immediate acyl-enzyme
intermediate (Figure 2B) forms on acylation of Ser-70 and loses
this good leaving groupsthe sulfonatesby either an R- or
â-elimination process as discussed previously,7 to give rise to
the final inactivated species. The loss of the sulfonate moiety
gives additional flexibility to the acyl-enzyme species, which
rearranges into a new low-energy entity. This is the inactivated
species that one sees in the crystal structure of the TEM-1
â-lactamase modified by 5, which shows structural elements
unique to this type of enzyme inactivation, and indeed is quite
distinct compared to all known structures for inhibited â-lac-
tamases.
As described earlier, the hydroxyethyl group in inactivator 5
was meant to prevent the early deacylation of the acyl-enzyme
intermediate. Indeed, this moiety would play this role at the
stage of the formation of the immediate acyl-enzyme species
(Figure 2B). However, after the departure of the sulfonate group,
the strong interactions it forms with the side chain of Ser-130
(Scheme 2 and Figure 2A) direct the oxygen of the ester moiety
into the space that the hydrolytic water normally occupies, such
that the water molecule is displaced from the active site. This
here is with the sulfonate of the inhibitor, which serves as a
surrogate for the invariant carboxylates for typical substrates
found in this position.7 These interactions readily anchor the

5358 J. Am. Chem. Soc., Vol. 121, No. 23, 1999
Sware´n et al.
cannula. The mixture was stirred for 3.5 h at -78 °C, at which time
the reaction was quenched by the addition of saturated NH4Cl, and
was subsequently extracted with ether. The ether layer was washed
with 1.0 M HCl, saturated NaHCO3, and brine, dried (MgSO4), and
filtered, and the solvent was evaporated off to afford a yellow oil that
was chromatographed on silica gel eluting with 30% EtOAc in hexanes
to afford 7.0 g (74%) of 8 (four diastereomers) as a light yellow oil:
is the structural reason that imparts hydrolytic stability to this
inhibited enzyme species. Compound 5 was synthesized as a
racemic mixture. The S-stereogenicity at the atom position 6
was deduced from the polar and van der Waals interactions
observed in the crystal. Modeling experiments were performed
with both enantiomers at this position, as we were not certain
of the absolute configuration before the crystal structure became
available.
The type of monobactam inhibitor for â-lactamases reported
herein operates by a mechanism entirely distinct from those of
the clinically used inactivators. Whereas the chemistry described
in this report was envisioned for â-lactamases, the principles
disclosed should be of general application for any enzyme that
operates by covalent catalysis.
1
Rf 0.30 (2:3, EtOAc:hexanes); IR (neat) 3445, 1725 cm-1; H NMR
(300 MHz, CDCl3) δ 1.21 (d, J ) 6.5 Hz), 1.25 (d, J ) 6.6 Hz), 1.29
(d, J ) 6.4 Hz), 2.70-2.78 (m), 2.81-2.91 (m), 3.03 (d, J ) 5.0 Hz),
3.28 (d, J ) 8.6 Hz), 3.57 (s), 3.63 (s), 3.72 (s), 3.73-3.90 (m), 4.08-
4.30 (m), 5.06-5.23 (m), 5.26 (t, J ) 5.5 Hz), 7.26-7.43 (m); 13C
NMR (75 MHz, CDCl3) δ 21.07, 21.61, 21.67, 22.13, 22.36, 42.50,
51.38, 51.66, 51.78, 56.98, 57.61, 58.94, 59.06, 60.65, 64.19, 65.71,
66.44, 67.03, 70.61, 71.49, 73.18, 73.97, 125.46, 125.93, 126.15, 126.54,
127.54, 127.74, 127.98, 128.21, 128.34, 128.38, 128.46, 141.62, 141.71,
171.90, 173.22, 173.58, 174.24; HRMS (FAB) calcd for C12H17O4
(MH+) 225.1127, found 225.1109.
Experimental Section
(()-O-Allyl 3-Hydroxy-2-(hydroxyphenyl)methylbutanohydrox-
amate (9). A stirred solution of 1.5 g (6.5 mmol) of the methyl ester
8 in 8.0 mL of THF was mixed with 7.2 mL of a 1.0 M NaOH solution
at room temperature. The turbid solution was stirred for 2 h and was
subsequently diluted with ether (10 mL). The layers were separated,
and 0.92 g (8.4 mmol) of O-allyl hydroxylamine hydrochloride was
added to the aqueous portion. The pH of the aqueous layer was adjusted
to 4.5 using 1.0 M HCl, and 1.9 g (9.7 mmol) of EDC‚HCl was added
in four portions over a period of 1.5 h, while maintaining the pH of
the solution at 4.5. The solution was extracted with EtOAc (3×). The
combined organic layer was washed with 1.0 M HCl, 5% NaHCO3
solution, and brine, dried (Na2SO4), and filtered, and the solvent was
evaporated to afford 0.83 g (48%) of 9 as a white foam. The 1H NMR
was very complex due to the mixture of diastereomers, as well as the
presence of hydroxamate and hydroximate forms. The mixture was used
directly in the next step: IR (KBr) 3300 (broad), 1645 cm-1; HRMS
(FAB) calcd for C14H20NO4 (MH+) 266.1392, found 266.1392; MS
(FAB) m/z 288 (M + Na+).
Melting points (mp’s) were determined on a Thomas-Hoover
capillary melting point apparatus in open capillaries and are uncorrected.
Infrared (IR) spectra were obtained on a Perkin-Elmer 1420 IR
spectrophotometer and were calibrated with the 1601 cm-1 band of
polystyrene. Nuclear magnetic resonance (NMR) spectra were obtained
on a General Electric GN-300, a Varian Unity Plus 300, or a Varian
VXR500S spectrometer. 1H NMR chemical shifts are reported in parts
per million relative to tetramethylsilane (0.00 ppm). 13C NMR spectra
were referenced relative to the center peak of CDCl3 (77.00 ppm). Fast-
atom bombardment (FAB, xenon, 3-nitrobenzyl alcohol matrix) mass
spectra were obtained on a JEOL JMS-AX505HA mass spectrometer.
Thin-layer chromatography (TLC) was conducted on silica gel 60 F254
(0.2 mm thickness, aluminum support), and the chromatograms were
visualized with ultraviolet light and by dipping in 10% phosphomo-
lybdic acid (PMA) in ethanol, followed by heating. Flash column
chromatography was performed using silica gel 60 (EM Science, 230-
400 mesh ASTM). Elemental analysis was performed by M-H-W
Laboratories (Phoenix, AZ). Anhydrous acetonitrile and triethylamine
were freshly distilled from calcium hydride under an atmosphere of
nitrogen and transferred via syringe or cannula. Bulk grade ethyl acetate
(EtOAc) and Skellysolve B (referred to simply as “hexanes”) were
distilled before use. All purchased reagents were of reagent grade quality
and were used without further purification. Penicillin G was purchased
from Sigma. The wild-type class A TEM-1 â-lactamase from E. coli
was purified according to literature methods.14 All kinetic and spectral
measurements were made on a Perkin-Elmer Lambda 3B or Hewlett-
Packard 452 diode-array spectrophotometer. The enzyme assay and
methods for determination of kinetic parameters were according to
published procedures.7
General Procedure for the Preparation of â-Lactams 10-13. A
stirred solution of 662 mg (2.49 mmol) of hydroxamate 9 in 20.0 mL
of anhydrous THF was charged with 654 mg (2.49 mmol) of
triphenylphosphine, followed by 574 mg (2.49 mmol) of di-tert-butyl
azodicarboxylate (DBAD) at ice-water temperature under an atmo-
sphere of nitrogen. The resulting solution was allowed to warm to room
temperature and was allowed to stir for an additional 20 h. The solvent
was evaporated, and the residue was chromatographed on silica gel
eluting with 30% EtOAc in hexanes to afford 142 mg (23%) of 10 as
a white solid, 77 mg (12%) of 11 as a colorless oil, 185 mg (30%) of
12 as a colorless oil, and 71 mg (12%) of 13 as a colorless oil.
(()-cis-1-(Allyloxy)-3-[(hydroxyphenyl)methyl]-4-methyl-2-aze-
tidinone (10): Rf 0.43 (1:1, EtOAc:hexanes); mp 86-87 °C (hexanes);
(()-Methyl 3-Hydroxybutanoate (7). A solution of 9.0 g (78 mmol)
of methyl acetoacetate (6) in 300 mL of methanol was stirred at ice-
water temperature under an atmosphere of nitrogen. Sodium borohy-
dride (0.98 g, 26 mmol) was added to this solution, and the mixture
was stirred for 30 min. Subsequently, the reaction was quenched by
the addition of 100 mL of brine, the mixture was extracted with ether
(3×), dried (MgSO4), and filtered, and the solvent was evaporated in
vacuo to afford 6.5 g (71%) of 7 as a colorless oil: 1H NMR (300
MHz, CDCl3) δ 1.23 (d, J ) 6.3 Hz, 3H), 2.39-2.55 (m, 2H), 3.07 (br
s, 1H), 3.71 (s, 3H), 4.14-4.27 (m, 1H); 13C NMR (75 MHz, CDCl3)
δ 22.31, 42.64, 51.39, 63.97, 172.85.
1
IR (KBr) 3415 (OH), 1740 (CO) cm-1; H NMR (300 MHz, CDCl3)
δ 1.30 (d, J ) 6.4 Hz, 3H), 2.86 (brs, 1H), 3.38 (dd, J ) 7.5, 5.5 Hz,
1H), 3.93-4.04 (m, 1H), 4.38-4.52 (m, 2H), 4.90 (d, J ) 7.5 Hz,
1H), 5.30-5.44 (m, 2H), 5.93-6.09 (m, 1H), 7.27-7.48 (m, 5H); 13
C
NMR (75 MHz, CDCl3) δ 13.73, 53.09, 56.54, 70.56, 77.22, 120.78,
126.98, 128.24, 128.59, 132.11, 141.15, 165.19; HRMS (FAB) calcd
for C14H18NO3 (MH+) 248.1287, found 248.1285. Anal. Calcd for
C14H17NO3: C, 68.00; H, 6.93, N, 5.66. Found: C, 67.89; H, 7.12; N,
5.51.
(()-trans-1-(Allyloxy)-3-[(hydroxyphenyl)methyl]-4-methyl-2-
azetidinone (11): Rf 0.39 (1:1, EtOAc:hexanes); IR (neat) 3425 (OH),
1760 (CO) cm-1; 1H NMR (300 MHz, CDCl3) δ 1.20 (d, J ) 6.0 Hz,
3H), 2.88 (dd, J ) 6.9, 1.8 Hz, 1H), 3.26 (d, J ) 2.7 Hz, 1H), 3.68
(qd, J ) 6.0, 2.1 Hz, 1H), 4.18-4.32 (m, 2H), 4.94 (dd, J ) 7.2, 2.1
Hz, 1H), 5.18-5.32 (m, 2H), 5.78-5.95 (m, 1H), 7.25-7.44 (m, 5H);
13C NMR (75 MHz, CDCl3) δ 17.10, 56.47, 58.74, 72.30, 77.09, 120.81,
126.23, 128.18, 128.53, 132.04, 140.62, 164.54; HRMS (FAB) calcd
for C14H18NO3 (MH+) 248.1287, found 248.1284.
(()-Methyl 3-Hydroxy-2-[(1-hydroxy-1-phenyl)methyl]butanoate
(8). n-Butyllithium (34 mL of a 2.5 M solution in hexanes, 85 mmol)
was added to a stirred solution of 11.2 mL (85.4 mmol) of diisopro-
pylamine in 34 mL of anhydrous THF at ice-water temperature under
an atmosphere of argon. After 10 min of stirring, the solution was cooled
to -78 °C and was stirred for an additional 5 min. This solution was
charged with 5.0 g (42 mmol) of methyl 3-hydroxybutanoate (7) in 23
mL of anhydrous THF via cannula, and the resultant solution was stirred
at the same temperature for 1 h. A solution of benzaldehyde (5.0 g, 47
mmol) in 23 mL of anhydrous THF was added to the mixture via
(()-trans-1-(Allyloxy)-3-(1-hydroxyethyl)-4-phenyl-2-azetidino-
ne (12): Rf 0.38 (1:1, EtOAc:hexanes); IR (neat) 3450 (OH), 1765 (CO)
1
cm-1; H NMR (300 MHz, CDCl3) δ 1.35 (d, J ) 6.3 Hz, 3H), 2.36
(14) Zafaralla, G.; Manavathu, E. K.; Lerner, S. A.; Mobashery, S.
Biochemistry 1992, 31, 3847.
(br s, 1H), 2.85 (dd, J ) 6.0, 2.4 Hz, 1H), 4.10-4.24 (m, 1H), 4.28-

Inhibition and X-ray Structure of TEM-1 â-Lactamase
J. Am. Chem. Soc., Vol. 121, No. 23, 1999 5359
4.44 (m, 2H), 4.72 (d, J ) 2.1 Hz, 1H), 5.22-5.39 (m, 2H), 5.83-
6.02 (m, 1H), 7.33-7.46 (m, 5H); 13C NMR (75 MHz, CDCl3) δ 21.35,
61.24, 63.28, 65.84, 77.05, 120.68, 126.61, 128.75, 128.92, 132.20,
136.64, 165.83; HRMS (FAB) calcd for C14H18NO3 (MH+) 248.1287,
found 248.1289.
Diffracted intensities were measured on the synchrotron beamline
DW32 at LURE (Orsay, France). The wavelength was 0.975 Å, and
the data were processed using MOSFLM (v. 5.40).16 An 88% complete
data set to 1.95 Å resolution (Rsym ) 0.043; Table 1) was collected.
Refinement with CNS,17 applying a bulk solvent correction, was
carried out against all data (14192 reflections) except a randomly
extracted set of structure factors used for calculation of the free R
value.18 The coordinates of the native TEM-1 enzyme were used as
the initial model.15 Rigid body refinement was performed in order to
account for the variation in cell parameters between the native and the
enzyme-inhibitor crystals. In each cycle of refinement, manual
corrections using the program O19 were followed by slow cooling using
torsion angle molecular dynamics, energy minimization, and temper-
ature factor refinement with maximum likelihood target function.
Electrostatic energy terms were turned off. Water molecules were
introduced as neutral oxygen atoms when they appeared as positive
(()-cis-1-(Allyloxy)-3-(1-hydroxyethyl)-4-phenyl-2-azetidinone (13):
Rf 0.21 (1:1, EtOAc:hexanes); IR (neat) 3415 (OH), 1765 (CO) cm-1
;
1H NMR (300 MHz, CDCl3) δ 0.84 (d, J ) 6.6 Hz, 3H), 2.41 (br s,
1H), 3.15 (dd, J ) 8.4, 5.7 Hz, 1H), 3.65-3.80 (m, 1H), 4.38-4.55
(m, 2H), 5.05 (d, J ) 5.7 Hz, 1H), 5.28-5.42 (m, 2H), 5.89-6.03 (m,
1H,), 7.32-7.46 (m, 5H); 13C NMR (75 MHz, CDCl3) δ 20.88, 56.60,
63.56, 64.76, 76.82, 120.76, 127.45, 128.85, 128.72, 132.15, 133.59,
166.09; HRMS (FAB) calcd for C14H18NO3 (MH+) 248.1287, found
248.1295.
(()-trans-1-N-(Tosyloxy)-3-(1-hydroxyethyl)-4-phenyl-2-azetidi-
none (5). Triphenylphosphine (35 mg, 0.13 mmol), palladium diacetate
(8 mg, 0.04 mmol), and pyrrolidine (31 µL, 0.37 mmol) were added to
a stirred solution of 83 mg (0.34 mmol) of â-lactam 12 in 2.0 mL of
anhydrous acetonitrile at room temperature under an atmosphere of
argon for 1.5 h. The mixture was diluted with CH2Cl2 and was washed
with 5% Na2CO3 (3×). The combined aqueous portion was washed
with EtOAc, and it was acidified to pH 4.0 using 3.0 M HCl. The
solution was washed with CH2Cl2 (3×), the combined organic layer
was dried (Na2SO4) and filtered, and the solvent was evaporated to
afford a yellow oil. The oil was dissolved in 2.0 mL of anhydrous
CH2Cl2, and 70 mg (0.37 mmol) of p-toluenesulfonyl chloride and 50
µL (0.36 mmol) of triethylamine were added at room temperature. After
45 min of reaction time, the solvent was evaporated in vacuo to afford
a brown oil. The oil was purified on a Chromatotron device (1 mm
silica gel plate) eluting with 40% EtOAc in hexanes to afford 45 mg
(37%) of 5 as a colorless oil: Rf 0.26 (2:3, EtOAc:hexanes); IR (CDCl3)
peaks at four standard deviations above the mean, in [Fobs - Fcalc, Φcalc
]
electron density maps. Hereafter, the slow-cooling scheme was
abandoned for conventional refinement. Starting geometry of the
inhibitor was obtained by optimization using DISCOVER (MSI). The
final model was comprised of 2014 non-hydrogen protein atoms,
including the covalently bound inhibitor, 349 water molecules, and one
sulfate anion. The final R and free R values are 0.206 and 0.258 ,
respectively. The average B factors are 19.5 and 32.1 Å2 for protein
atoms and solvent, respectively (Table 1). Attempts were made to
crystallographically refine the conformations of the inhibitor bound in
the active starting from arbitrary initial conformations. For some of
them, the ester oxygen atom was located in the oxyanion hole. In all
cases, these independent refinements led to very similar final conforma-
tions of the bound inhibitor where the ester oxygen atom was located
at 2.8 Å from Glu-166.
1
3590, 1795 (CO) cm-1; H NMR (300 MHz, CDCl3) δ 1.32 (d, J )
6.4 Hz, 3H), 1.90 (br s, 1H), 2.45 (s, 3H), 2.95 (dd, J ) 5.3, 3.0 Hz,
1H), 4.08-4.21 (m, 1H), 4.86 (d, J ) 3.0 Hz, 1H), 7.22-7.40 (m,
7H), 7.83 (d, J ) 8.4 Hz, 2H); HRMS (FAB) calcd for C18H20NO5S
(MH+) 362.1062, found 362.1039.
Acknowledgment. I.M. was the recipient of the Rumble and
Heller Fellowships. The use of the Lizzadro Magnetic Reso-
nance Center for the NMR studies at the University of Notre
Dame is gratefully acknowledged. The work in France was
funded in part by INSERM (CRE Contract 930612), the
Regional Midi-Pyrenees (Contract 9200843), and CNRS (to
J.P.S.). The work in the U.S. was supported by the National
Institutes of Health and National Science Foundation (to S.M.)
and Eli Lilly and Co. (to M.J.M.).
Properties of the Rearranged Product 14: Rf 0.17 (2:3, EtOAc:
1
hexanes); IR (CDCl3) 3460, 1760 (CO) cm-1; H NMR (500 MHz,
CDCl3) δ 0.99 (d, J ) 6.5 Hz, 3H), 2.37 (s, 3H), 3.75 (dd, J ) 5.5, 1.0
Hz, 1H), 4.23 (pentet, J ) 6.5 Hz, 1H), 5.29 (d, J ) 8.0 Hz, 1H), 5.36
(br s, 1H), 7.11-7.19 (m, 3H), 7.24-7.35 (m, 4H), 7.55 (d, J ) 8.5
Hz, 2H); 13C NMR (125 MHz, CDCl3) δ 20.29, 21.60, 62.98, 74.14,
84.68, 127.27, 127.75, 129.01, 129.68, 129.84, 132.97, 133.27, 145.11,
157.20; HRMS (FAB) calcd for C18H20NO5S (MH+) 362.1062, found
362.1046.
Protein Crystallography. Orthorhombic crystals (380 × 200 × 200
µm3) of the TEM-1 â-lactamase were prepared as previously de-
scribed.15 Soaking was performed at 4 °C during 4 days by adding the
inhibitor as a racemic mixture (stock solution: 100 mM in 50%
ammonium sulfate, 100 mM Na,K phosphate buffer, pH 7.8) to the
mother liquor to a final concentration of 20 mM. The crystal was drawn
up into a loop, flushed for few seconds with Paratone oil, and flash-
cooled in a N2 gas stream at 100 K.
JA990400Q
(16) Leslie, A. G. W. Computational Aspects of Protein Crystal Data
Analysis, Proceedings of the Daresbury Study Weekend; SERC, Daresbury
Laboratory: Warrington, U.K., 1987; pp 39-50.
(17) Bru¨nger, A. T.; Adams, P. A.; Clore, M. G.; DeLano, W. L.; Gros,
P.; Grosse-Kunstieve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, M.; Pannu,
N. S.; Read, R.; Rice, L. M.; Simonson, T.; Warren, G. L. Acta Crystallogr.
1998, D54, 905.
(18) Bru¨nger, A. T. Nature 1992, 355, 472.
(15) Jelsch, C.; Lenfant, F.; Masson, J. M.; Samama, J. P. J. Mol. Biol.
1992, 223, 377.
(19) Jones, T. A.; Zou, J. Y.; Cowan, S. W.; Kjeldgaard, M. Acta
Crystallogr. 1991, A47, 110.