Design, synthesis, and evaluation of α-ketoheterocycles as class C β-lactamase inhibitors

Article abstract of DOI:10.1016/S0968-0896(01)00107-9

A series of specific α-ketoheterocycles (benzoxazole, thiazole, imidazole, tetrazole, and thiazole-4-carboxylate) has been synthesized in order to assess their potential as β-lactamase inhibitors. The syntheses were achieved either by construction of the heterocycle (benzoxazole) from an appropriate α-hydroxyimidate, followed by oxidation of the alcohol, or by direct reaction of methyl phenaceturate with a lithiated heterocycle. The properties of these compounds in aqueous solution are described and their inhibitory activity against β-lactamases assessed. They did inhibit the class C β-lactamase of Enterobacter cloacae P99 but not the TEM β-lactamase. The most effective inhibitor of the former enzyme (K_i = 0.11 mM) was 5-(phenylacetylglycyl) tetrazole, probably because it is an anion at neutral pH. Interpretation of the results was aided by computational models of the tetrahedral adducts. Most of the compounds also inhibited α-chymotrypsin but not porcine pancreatic elastase.

Full text of DOI:10.1016/S0968-0896(01)00107-9

Bioorganic & Medicinal Chemistry 9 (2001) 2035–2044
Design, Synthesis, and Evaluation of ꢀ-Ketoheterocycles as
Class C ꢁ-Lactamase Inhibitors
Sanjai Kumar, Andre L. Pearson and R. F. Pratt*
Department of Chemistry, Wesleyan University, Middletown, CT 06459, USA
Received 29 January 2001; accepted 16 March 2001
Abstract—A series of specific a-ketoheterocycles (benzoxazole, thiazole, imidazole, tetrazole, and thiazole-4-carboxylate) has been
synthesized in order to assess their potential as b-lactamase inhibitors. The syntheses were achieved either by construction of the
heterocycle (benzoxazole) from an appropriate a-hydroxyimidate, followed by oxidation of the alcohol, or by direct reaction of
methyl phenaceturate with a lithiated heterocycle. The properties of these compounds in aqueous solution are described and their
inhibitory activity against b-lactamases assessed. They did inhibit the class C b-lactamase of Enterobacter cloacae P99 but not the
TEM b-lactamase. The most effective inhibitor of the former enzyme (Ki=0.11 mM) was 5-(phenylacetylglycyl) tetrazole, probably
because it is an anion at neutral pH. Interpretation of the results was aided by computational models of the tetrahedral adducts.
Most of the compounds also inhibited a-chymotrypsin but not porcine pancreatic elastase. # 2001 Elsevier Science Ltd. All rights
reserved.
Introduction
antibiotics against resistant organisms.11 In the absence
of any suitable commercial inhibitor of class C b-lacta-
mases, there has been a pressing need in the research
community to produce specific inhibitors against these
enzymes. This work reports the synthesis and kinetics
studies, in conjunction with molecular modeling, of
specific a-ketoheterocycles as a new class of potential
inhibitors of class C b-lactamases.
In recent years, resistance to antibiotics has been
increasing due to the use, overuse and misuse of broad
spectrum antibiotics and the ability of microbes to
exchange resistance genes.1,2 One of the most common
forms of bacterial resistance to the b-lactam antibiotics
comes from a groupof enzymes called b-lactamases that
catalyzes the hydrolysis of b-lactam antibiotics (Scheme 1)
rendering them ineffective.3
A wide variety of inhibitors of serine proteases have
been designed and synthesized over the last 50 years.12
These include covalent modifying reagents, including,
recently, b-lactams,13 and transition state analogue pre-
cursors such as peptidyl trifluoromethyl ketones and
peptidylboronic acids.14,15 Elastase inhibitors of this
type, for example, have been sought for the treatment of
emphysema.16,17 Recently Edwards et al. have described
peptidyl a-ketoheterocycle analogues as inhibitors of
human neutrophil elastase (HNE).18ꢀ20 These com-
pounds are believed to form anionic tetrahedral adducts
with the active site serine hydroxyl group(Scheme 2)
Over the last 10 years, the prevalence of bacterial resis-
tance to third generation cephalosporins has increased
rapidly because of high levels of expression of class C b-
lactamases.4,5 Although originally found chromoso-
mally encoded, these enzymes have recently been found
on plasmids,6 thus raising the chance that the resistance
gene may spread to other species. The class C b-lacta-
mases (also known as cephalosporinases) have become a
major concern in the medical community7ꢀ10 owing to
their broad specificity for third generation cephalospor-
ins. The b-lactamase inhibitors used clinically at pre-
sent, clavulanate, sulbactam and tazobactam, are
ineffective against class C b-lactamases and thus have
not been used in combination therapy with b-lactam
*Corresponding author. Tel.: +1-860-685-2629; fax: +1-860-685-
2211; e-mail: rpratt@wesleyan.edu
Scheme 1.
0968-0896/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(01)00107-9

2036
S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
and thus generate transition state analogue structures.
Furthermore, the crystal structure of an elastase inhi-
bitor complex showed that His 57, the active site general
acid/base catalyst, is involved as a general acid in stabi-
lizing the tetrahedral adduct by forming a bifurcated
hydrogen bond to both the active site Ser 195 Og and
the heterocyclic nitrogen atom (Scheme 2).18 We there-
fore decided to investigate the potential of this class of
inhibitor against b-lactamases. We thought it possible
that such inhibitors might trapthe general acid/base
catalysts of b-lactamase active sites. The identity of such
catalysts has long been controversial.21 The following
compounds 1–5 (Scheme 3) were synthesized and
evaluated as inhibitors, primarily of a class C b-lacta-
mase. They contain a combination of an electrophilic a-
ketoheterocycle and the amido side chain favored by
b-lactamases.
Results and Discussion
The heterocycles 1–5 were synthesized by variations of
established methods (Schemes 4–8). Thus, the benzox-
azole 1 was prepared from phenylacetylaminoacetalde-
hyde by construction of the oxazole ring and oxidation
of the resulting alcohol 12, essentially as described by
Edwards et al.18,20 The remaining compounds 2–5 were
prepared by reaction of the appropriate lithiated het-
erocycle with methyl phenaceturate.19,22 Thus, for the
synthesis of 2, 2-lithiothiazole was generated from n-
butyl lithium and 2-bromothiazole, and for that of 4,
the 2-lithio derivative was prepared directly from thia-
zole-4-carboxylic acid. Protected heterocycles were
employed in the remaining two cases, 1-SEM-imidazole
for 323,24 and 1-BOM-tetrazole for 5.25 Reaction of the
lithio derivatives of these protected heterocycles with
methyl phenaceturate yielded the required products in
N-protected form. After some experimentation with
alternative methods,23ꢀ25 it was found that both pro-
tecting groups could be removed by treatment with tri-
fluoroacetic acid under the appropriate conditions, as
described in Experimental.
Scheme 2.
Scheme 3.
Scheme 4.

S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
2037
The properties of the heterocycles believed to be
important to the inhibitory power of 1–5 are their elec-
tron-withdrawing ability (sI) and the basicity of the b-N
atom.26 High electron-withdrawing ability (large posi-
tive sI values) would promote nucleophilic attack by the
active site nucleophile and high basicity of the b-N
would attract adjacent general acids (Scheme 2); toge-
ther these properties should both contribute to a strong
inhibitor. Consideration of the values of these para-
meters for the heterocycles chosen (Table 1) suggests
that the benzoxazole 1 and the tetrazole 5 should be the
best inhibitors. The behavior of these compounds in
solution however can be complicated, with the possibi-
lities of keto, enol, enolate and hydrated keto forms to
be considered. For example, the benzoxazole 1 exists in
the solid state mainly in the keto form (Scheme 9; an
infra-red spectrum shows an intense nC=0 at 1720
two proton doublet at d 4.80 corresponding to the
methylene groupadjacent to the keto grou;p uv
absorption spectra show that this species absorbs maxi-
mally at 291 nm (e=13,250 Mꢀ1 cmꢀ1). In DMSO,
however, the keto form converts essentially completely
over an h into the enol 6, characterized by a vinylic
proton doublet at d 7.75 and a lmax of 326 nm
(e=16,300 Mꢀ1 cmꢀ1) in the NMR and UV absorption
spectra, respectively. A spectrophotometric pH titration
of the enol indicated that its pKa in water is 10.2; the
maximal absorption of the enolate was observed at 344
nm (e=17,600 Mꢀ1 cmꢀ1).
The solubility of 1 in water was insufficient to obtain a
1H NMR spectrum in that solvent alone. Spectra in
acetonitrile/water mixtures show the presence of two
compounds, the keto form and a second form, most
likely the hydrate 8, the latter characterized by a keto-
methylene resonance at d 3.43, that is 1.37 ppm upfield
1
cmꢀ1). In acetonitrile solution, H NMR spectroscopy
indicates that the keto form is also dominant, with a
Scheme 5.
Scheme 6.

2038
S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
from its position in the keto compound, and UV
absorptions peaks at 271, 278, and 301 nm.
spectra in pure water indicate a hydrate content of
around 50%. The presence of this hydrate attests to the
electron-withdrawing power (sI) of the benzoxazole
substituent (Table 1). This is also indicated by the
exchange of the ketomethylene protons with solvent
deuterons in 2H2O; in 50:50 acetonitrile/2H2O, this
In 50:50 acetonitrile/water, the hydrate comprises ca.
15% of the total. This proportion increased as the ace-
tonitrile content of the solvent decreased. Absorption
Scheme 7.
Scheme 8.
Table 1. Inhibition of the P99 b-lactamase and a-chymotrypsin by
the a-ketoheterocycles 1–5; electronic properties of the heterocycles
Inhibitor
Ki (mM)
sIa b-N chargea
P99 b-Lactamase a-Chymotrypsin
b
b
1
2
3
4
5
0.41
0.34
0.27
—
ꢀ0.62
ꢀ0.43
ꢀ0.53
—
ꢀ0.40d
0.55ꢁ0.06
0.34c
0.77c
1.1c
1.5ꢁ0.1
b
0.11ꢁ0.03
0.35c
0.49
aTaken from ref 26.
bNo inhibition observed at 1 mM inhibitor.
cSingle point estimates.
dRefers to neutral tetrazole.
Scheme 9.

S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
2039
process was complete in 2 h. The presence of the
hydrate, however, would decrease the concentration of
the presumed inhibitory form (keto) in solution.
reversible inhibitors of modest potency; their inhibitory
ability did not further increase with time. The thiazole
derivative 2 was more effective than the imidazole, in
accord with the higher sI of the latter compound rather
than its smaller b-N charge. The effectiveness of 3
decreased at lower pH, demonstrating that the inhibitory
form was not the cation.
2
1H NMR spectra of 2–4 in H2O show the presence of
essentially non-hydrated keto compounds. The smaller
amount of hydrate in these cases presumably reflects the
lower electron-withdrawing power of the heterocycles
concerned (Table 1). The NMR spectrum of 3 in 1 M
2HCl however shows the presence of 85% hydrate. The
electron withdrawing power of the imidazole of 3 has
presumably been significantly increased by protonation;
the pKa of 2-acetylimidazolium is reported to be 3.12.27
Although the sI value of the 5-tetrazolyl substituent is
the largest of the heterocycles in Table 1, the heterocycle
is also acidic to the extent that 5 probably exists essen-
tially completely as a tetrazole anion at pH 7; the pKa of
tetrazole is 4.928 and that of a 5-acyl-tetrazole must be
lower. An anionic tetrazole moiety in 5 should decrease
the electrophilicity of the keto-carbonyl with respect to
that of the neutral form, but increase the basicity of the
b-N.
The tetrazole 5 is the best b-lactamase inhibitor of the
five compounds. Its Ki only increased slightly to 0.28
mM at pH 5, suggesting that the inhibitory species at
pH 7.5 was in fact the tetrazole anion 19. This would
probably have the least electrophilic carbonyl group,
but would have the most basic nitrogen and also a
negative charge. The latter point may well be the domi-
nant effect here. Essentially all good substrates and
inhbitors of the P99 b-lactamase (and, indeed, of most
b-lactamases) are negatively charged. The active site has
significant complementary positive charge, provided by
the conserved lysine residues, 67 and 315, and the a-2
helix dipole.29,30
A computational model of 5 (as the anion 19) bound as
a tetrahedral carbonyl adduct at the active site was con-
structed, based on the crystal structure of an analogous
phosphonate inhibitor as described in Experimental. The
The results of experiments to determine the effectiveness
of 1–5 as class C b-lactamase inhibitors are given in
Table 1. These compounds were found to be fast
Table 2. Structural parameters from molecular dynamics of enzyme-inhibitor complexes
Distance (A) or Angle (deg)a
Structural parameter
1
2
3
4
5
Tyr 150 Oz-ligand Nb distance
Tyr 150 Oz-Ser 64 Og distance
Tyr 150 CE–Oz-ligand Nb angle
Ser 64 Og–Ctet-ligand Ca–Nb dihedral
ligand Oꢀ-Ser 64 N distanceb
4.3ꢁ0.5
4.9
83
3.4ꢁ0.5
2.9
93
3.7ꢁ0.5
2.8
79
3.6ꢁ0.8
3.1
83
3.1ꢁ0.2c
3.1
107c
32ꢁ17
27ꢁ35
80ꢁ23
5ꢁ21
ꢀ70ꢁ21c
3.0
2.9
2.9
3.0
2.9
aAverage and standard deviation of 100 snapshots from 10-20 psec.
bMajor oxyanion hole interaction.
cThe Nb closer to Tyr 150 Oz.
Figure 1. Stereo view of the optimized tetrahedral adduct of the anion of the tetrazole 5 with the P99 b-lactamase.

2040
S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
conformation of the bound inhibitor, and in particular
the conformation of the heterocycle at the active site,
was investigated by means of a 20 ps molecular dynam-
ics run. Some important structural information derived
from these runs, both for the tetrazole and also for the
other heterocycles, as discussed below, is given in Table 2.
It was found, in most snapshots of the tetrazole com-
plex, that the hydroxyl oxygen of Tyr 150 was within
hydrogen-bonding distance of the b-N of the tetrazole
ring and both of these moieties were close to the
ammonium groupof Lys 315. (A water molecule was
also generally found associated with the b-N of the tet-
razole.) Although, the Tyr 150 hydroxyl was generally
within hydrogen-bonding distance, it was directed, as
indicated by the orientation of the tetrazole ring (Ser 64
Og–Ctet-ligand Ca–Nb dihedral angle; Table 2), not so
much at a tetrazole nitrogen lone pair but rather
directly to the negatively charged ring of the hetero-
cycle. It was also noticed that the hydroxyl groupof Thr
316 was within hydrogen-bonding distance of one of the
g-N atoms of the tetrazole ring in many of these snap-
shots. Figure 1 shows an energy-minimized structure
derived from a typical snapshot. As well as the features
mentioned above, this shows the tetrahedral oxyanion
in its hole and the usual disposition of the amido side
chain with respect to active site functional groups.29ꢀ31
One distinct difference however, both in the minimized
structure and in most of the dynamics snapshots, is that
Lys 67 seems to have pulled away from the hydroxyl of
Tyr 150, the side-chain carbonyl of Asn 152, and Ser 64
Og, and taken a position apparently hydrogen-bonded
to the side chain carbonyl of the ligand.
complex of 5 with the b-lactamase suggested that
although the carboxylate could indeed interact strongly
with Lys 315 and Thr 316, bringing about a significant
degree of tethering of the heterocyclic ring into a con-
formation appropriate for interaction with Tyr 150, the
ligand did not remain as firmly bound as was the tetrazole
(e.g., see the Y150-b-N distance in Table 2).
A final surprise was the ineffectiveness of the benzox-
azole 1 as a b-lactamase inhibitor. The 2-benzoxazolyl
substituent appears to represent an attractive combina-
tion of electron-withdrawing power and b-N basicity
(Table 1). Indeed, benzoxazole analogues were among
the most powerful elastase inhibitors prepared by
Edwards et al.19 The likely reason for the failure of 1 as
a P99 b-lactamase inhibitor was demonstrated by mole-
cular modeling. A dynamics run suggested that it may
not be possible to accomodate the benzene ring of the
heterocycle at the active site and also maintain the
important hydrogen-bonding contacts mentioned above
between the active site and the ligand. In fact, the side
chains of Lys 315 and Tyr 150 moved away from the
ligand as the simulation proceeded (see the Tyr 150-b-N
distance in Table 2). The dynamics showed that the
benzoxazole was rather less mobile at the active site
than the monocyclic heterocycles, probably because of
its greater size which led to more steric interactions with
the protein surrounding the active site. Unfortunately, it
was not able, even in this less mobile state, to interact
productively with Tyr 150 and Lys 315. The rigidity of
the tetrahedral adduct also suggests that the benzox-
azole might have greater difficulty in binding as the
ketone and reacting with the nucleophilic serine than
would the smaller heterocycles. These factors probably
lead to the inhibitory impotence of 1. The a-ketohe-
terocycles 1–5 had some inhibitory activity against a-
chymotrypsin (Table 1), but none against elastase. They
were also ineffective against the class A TEM b-lacta-
mase and the Streptomyces R61 DD-peptidase. The les-
ser effectiveness of 1–5 against the latter two enzymes
than against the P99 b-lactamase mimics the relative
activity of phosphonate monoesters against these
enzymes.32
Similar dynamics runs with the thiazole and imidazole
inhibitors, 2 and 3, found the heterocyclic rings in simi-
lar positions to that shown for the tetrazole in Figure 1
but with Tyr 150 and Lys 67 more closely attending Ser
64 Og than the heterocycle (Table 2). The average Tyr
150 Oz-ligand b-N distance is also less favorable than in
the tetrazole case (Table 2). This apparently weaker
interaction between heterocycle and Tyr 150, although
probably offset to some degree by the greater electro-
philicity of the carbonyl group in these cases, may
explain the poorer performance of 2 and 3 as inhibitors
with respect to 5. In all of these cases, a barrier to the
anticipated optimal orientation of the heterocycle, as
shown in Scheme 2, was the eclipsing of the Ser 64 Og–
Ctet bond by the C¼N bond of the heterocycle. This is
overcome, apparently, in elastase, but not so in the P99
b-lactamase, as indicated by the average Ser 64 Og–Ctet-
ligand Ca–Nb dihedral angles (Table 2).
The results described in this paper appear to elaborate
earlier indications29 that specific electrophilic carbonyl
compounds are, at best, only modest inhibitors of
b-lactamases. These compounds have also, to date, not
demonstrated any activity against DD-peptidases.
Computational studies suggest that, in general, tetra-
hedral adducts at the active site of the P99 b-lactamase
without an electronegative heteroatom a to the tetra-
hedral atom do not interact strongly with the enzyme.29
There is, in fact, no direct evidence that 1–5 do actually
form tetrahedral adducts at the b-lactamase active site.
Inhibition, where it occurs, may just involve non-cova-
lent interaction with the enzyme. The latter would pre-
sumably be enhanced by the negative charge on the
tetrazole anion of 5. Although the incorporation of a
carboxylate, such as is found in b-lactamase substrates,
into b-lactamase inhibitors seems an appropriate strat-
egy, and one which has succeeded with boronates,33
even a rationally placed carboxylate, as in 4, was unable
Model-building suggested that a carboxylate group
added to the 4-position of an imidazole or thiazole
might interact favorably with the Lys 315 ammonium
ion and the Thr 316 hydroxyl groupin a similar way to
the carboxylate of a substrate;31 thus, compound 5 was
prepared. Unfortunately, it was not an effective inhi-
bitor (Table 1). A major factor leading to this result is
probably the high dehydration energy of the carbox-
ylate group. 3-(Phenylacetamido)-pyruvic acid is also a
poor inhibitor of this enzyme and probably for the same
reason.29 A molecular dynamics study (20 ps) of the

S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
2041
to enhance the activity of the heterocyclic ketones stud-
ied in this work. It seems unlikely at this time that
carbonyl compounds will prove to be useful against
b-lactam recognizing enzymes.
Atomic charges on the protein at pH 7.0 were assigned
by the software. The charges on the inhibitor molecule
were generated by MNDO calculations (MOPAC 6.0
module) on a peptide-bound inhibitor. The enzyme-
inhibitor complex was hydrated with a 15 A sphere with
water molecules centered at the active site Ser 64 Og. A
temperature of 300 K and a dielectric constant of 1 were
employed in all the simulations. The molecular dynam-
ics simulations in a CV force field were run for 20 ps
with coordinates saved every 100 steps. The trajectories
obtained in the simulation were analyzed in the Analysis
module of INSIGHT II. Energy minimization was also
performed in the CV force field as described.29
Experimental
Enzymes and substrates
The class C b-lactamase of Enterobacter cloacae P99
and the class A TEM-2-b-lactamase from Escherichia
coli W3310 were purchased from the center for Applied
Microbiology and Research (Porton Down, Wiltshire,
UK) and used as received. The Streptomyces R61 DD-
peptidase was the generous gift of Dr. J.-M. Frere of the
University of Liege, Liege, Belgium. Bovine a-chymo-
trypsin and its substrate N-succinyl-alanyl-alanyl-prolyl-
phenylalanyl-p-nitroanilide, porcine pancreatic elastase
(PPE) and its substrate N-succinyl-alanyl-alanyl-prolyl-
leucyl-p-nitroanilide were purchased from Sigma Chemical
Co.
Synthesis
1H and 13C NMR spectra were obtained from a Varian
300 MHz spectrometer and IR spectra were recorded on
a Perkin–Elmer BX FTIR V2.0 spectrophotometer.
Thin-layer chromatography was performed on 250 m
silica gel plates (Analtech, Newark, DE, USA). Column
chromatography were carried out on silica gel (grade
9385, 230–400; Merck), Sephadex LH-20 (bead size 20–
100 m, Sigma) and Sephadex G-10 (40–120 m, Sigma).
Mass spectra were obtained from the Mass Spectro-
metry Laboratory, School of Chemical Sciences, Uni-
versity of Illinois, IL, USA. Elemental analysis were
performed by Desert Analytics, Tucson, AZ, USA.
Analytical and kinetic methods
Absorption spectra and spectrophotometric reaction
rates, all at 25 ꢂC, were measured with a Hewlett-Pack-
ard HP8452A spectrophotometer. Kinetics experiments
with the P99 b-lactamase were performed in 20 mM
MOPS buffer (pH 7) and 20 mM pyridine buffer (pH 5)
using nitrocefin (Unipath, Ogdensburg, NY, USA) as
the substrate. The activities of 1–5 as competitive inhi-
bitors of the P99 b-lactamase (Ki values) were deter-
mined by measuring initial rates of product appearance
at 482 nm as a function of inhibitor concentration, and
fitting the data to a simple competitive inhibition equa-
tion. Similar experiments with bovine a-chymotrypsin
were performed spectrophotometrically at 410 nm
using N-succinyl-alanyl-alanyl-prolyl-phenylalanyl-p-
nitroanilide as the substrate (Km=0.043 mM) in 0.1 M
Tris–HCl, 0.01 mM CaCl2 buffer at pH 7.8.34 The
porcine pancreatic elastase assays were performed using
N-succinyl-alanyl-alanyl-prolyl-leucyl-p-nitroanilide
substrate in 0.1 M Tris, 0.01 M CaCl2 buffer at pH 7.8
as described by Largman et al.35 The absence of time
dependence of activity loss of the b-lactamase in the
presence of 1–5 was demonstrated by measuring the
enzyme activity in incubation mixtures of enzyme and
inhibitor as a function of time: small aliquots were
removed at appropriate intervals after mixing and
assayed against nitrocefin.
2-(Phenylacetylglycyl) benzoxazole (1)
The synthesis of 2-(phenylacetylglycyl) benzoxazole is
outlined in Scheme 4.
Phenylacetylaminoacetaldehyde (10). This was prepared
by the procedure of Brutshy et al.36 from aminoace-
taldehyde diethyl acetal and phenylacetyl chloride. The
product was crystallized from benzene/petroleum ether,
collected by filtration, and dried under vacuum. It was
characterized by its mp, 108–110 ꢂC (uncorr) [lit.36
113.5–115.5 ꢂC (corr)] and 1H NMR spectrum
[(300 MHz, DMSO-d6), d 3.65 (s, 2H), 4.2 (t, 2H), 6.1
(br s, 1H), 7.25–7.45 (m, 5H)].
Phenylacetylaminoacetaldehyde cyanohydrin (11).
A
solution of phenylacetylaminoacetaldehyde (10 g, 56
mmol) in methylene chloride (177.3 mL) under nitrogen
was treated with acetone cyanohydrin (15.54 mL) and
triethylamine (4.71 mL). The resulting mixture was stir-
red for four h. After evaporation of solvent, the residue
was taken upin ethyl acetate, washed with brine and
dried over MgSO4. The crude product was purified
using flash column chromatography on silica gel with a
hexane/ethyl acetate (1:1) mixture as solvent. The pro-
duct was isolated as a pale, white brittle solid in a yield of
Molecular modeling
A previously described general procedure for molecular
modeling was adopted.29 The starting point for these
studies was the crystal structure of the class C b-lacta-
mase from E. cloacae P99 with a phosphonate inhibitor
bound covalently to the active site serine residue; the
structure with the arylacetamido side chain closer to the
b-3 strand was chosen.30 The simulations were per-
formed on a IBM 3CT computer with INSIGHT II 97.0
(MSI Technologies, Inc., San Diego, CA, USA).
ꢂ
57% (6.56 g); mp57–58 C, 1H NMR (300 MHz, CDCl3)
d 3.45 (dt, J=14.2, 6.2 Hz, 1H), 3.63 (s, 2H), 4.6 (br s,
1H), 5.4 (br s, 1H), 6.4 (br s, 1H), 7.2–7.4 (m, 5H).
1-(2-benzoxazolyl)-2-(phenylacetylamino) ethanol (12). A
mixture of chloroform (10.4 mL) and absolute ethanol
(9.5 mL) under nitrogen at 0 ꢂC was treated with acetyl
chloride (10.5 mL) dropwise over 15 min. To this was

2042
S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
added phenylacetylaminoacetaldehyde cyanohydrin
(11), 1.0 g (4.9 mmol), in chloroform (10.4 mL) and the
reaction mixture was stirred for an hour. The solvent
was evaporated while keeping the temperature at 25 ꢂC
or below. The crude imidate, dissolved in anhydrous
ethanol (24.6 mL), was refluxed with o-aminophenol
(0.59 g, 5.4 mmol) for 6 h. After the solvent had been
removed by evaporation, the residue was dissolved in
ethyl acetate, washed successively with 1N NaOH, 1 N
HCl, saturated NaHCO3 and brine, dried over MgSO4,
and evaporated to dryness in vacuo. The brownish
crude material thus obtained was recrystallized from hot
benzene/petroleum ether (20–40 ꢂC). The product was
isolated as a colorless granular solid in a yield of 85%
g) yield. 1H NMR (300 MHz, CDCl3) d 3.5 (s, 2H), 3.63
(s, 3H), 3.86 (d, J=6.2 Hz, 2H), 7.2–7.4 (m, 5H), 8.5 (br
t, 1H).
2-(Phenylacetylglycyl) thiazole (2). Under an atmo-
sphere of nitrogen, to a cold (ꢀ78 ꢂC), stirred solution
of n-butyl lithium [15.2 mmol; 9.5 mL of a 1.6 M solu-
tion in hexane (Aldrich)] in THF (18.3 mL) was added
dropwise a solution of 2-bromothiazole [1.37 mL, 15.2
mmol (Aldrich)] in the same solvent (18.3 mL). After
the yellow solution had been stirred at ꢀ78 ꢂC for 30
min, a solution of phenylacetylglycine methyl ester (13)
(1.44 g, 6.94 mmol) in 18.5 mL THF was added slowly.
The mixture was stirred at ꢀ78 ꢂC for 1 h. The reaction
was quenched by the addition of 30 mL satd. NaHCO3
and 75 mL water. Solvent THF was removed by rotary
evaporation and the residual mixture was treated with
125 mL of ethyl acetate. The organic layer was washed
with water, brine, dried over Na2SO4, and evaporated
to dryness. The crude product was purified by silica gel
flash column chromatography and eluted with 2%
methanol in chloroform. Recrystallization from diiso-
propyl ether gave a white crystalline solid in a yield of
4% (60 mg).The product was characterized by mp 103–
104 ꢂC; 1H NMR (300 MHz, CDCl3) d 3.72 (s, 2H), 4.92
(d, J=5.01 Hz, 2H), 7.22–7.44 (m, 5H), 7.74 (d, J=2.9
Hz, 1H), 8.06 (d, J=3.0 Hz, 1H), 6.04 (br t, 1H, NH);
nmax (KBr) 1705 cmꢀ1 (C¼O). Anal. calcd for
C13H12N2O2S: C, 59.98; H, 4.65; N, 10.65; S, 12.32.
Found: C, 60.23; H, 4.43; N, 10.65; S, 12.30.
(1.0 g), mp110–112 ꢂC, H NMR (300 MHz, CDCl3) d
1
3.6 (s, 2H), 3.7–3.8 (m, 1H), 3.9–4.0 (m, 1H), 5.1 (m,
1H), 7.15 (m, 2H), 7.25 (m, 3H), 7.5 (m, 1H), 7.65 (m,
1H). 13C NMR (300 MHz, CDCl3) d 44 (CH2), 45
(CH2), 68 (COH), 111 (ArCH), 121 (ArCH), 125
(ArCH), 126 (ArCH), 135 (ArCH), 141 (ArCN), 151
(ArCO), 166, 173 (CO, CN).
2-(phenylacetylglycyl) benzoxazole (1). tert-Butyl alco-
hol (0.11 mL) was added to a suspension of 1-(2-ben-
zoxazolyl)-2-(phenylacetylamino)ethanol (12) (0.366 g,
1.25 mmol) and Dess–Martin periodinane (1,1,1-triace-
toxy-1,1,-dihydro-1,2-benziodoxol-3(1H)-one)37 (1.48 g,
3.5 mmol) in dichloromethane (8.6 mL) and the result-
ing slurry stirred at room temperature for 1 h. The
reaction mixture was then partitioned between ethyl
acetate and saturated Na2S2O3 (1:1), washed with satu-
rated NaHCO3 and brine, dried over MgSO4 and eva-
porated to dryness. Recrystallization of the crude
product from benzene afforded a colorless solid (0.34 g;
2-(Phenylacetylglycyl) imidazole (3)
The synthesis of 2-(phenylacetylglycyl) imidazole is
outlined in Scheme 6.
94%). It was characterized as follows: mp150 ꢂC; H
1
NMR (300 MHz, CDCl3) d 3.71 (s, 2H), 4.93 (d, J=5.4
Hz, 2H), 6.25 (br s, 1H), 7.3–7.4 (m, 5H), 7.48 (t, J=7.8
Hz, 1H), 7.57 (t, J=7.8 Hz, 1H), 7.67 (d, J=8.1 Hz,
1H), 7.89 (d, J=8.1 Hz, 1H); 13C NMR (300 MHz,
DMSO-d6) d 43 (CH2), 44 (CH2), 115 (ArCH), 120
ArCH), 125 (ArCH), 126 (ArCH), 127 (ArCH), 128
(ArCH), 129 (ArCH), 136 (ArCH), 146 (ArCH), 171
and 172 (CONH and C¼N), 195 (CO); nmax (KBr) 1720
cmꢀ1 (C¼O). Anal. calcd for C17H14N2O3: C, 69.40; H,
4.75; N, 9.52. Found: C, 69.30; H, 4.82; N, 9.57.
1-[20-(trimethylsilyl)ethoxy]methyl-2-(phenylacetylglycyl)-
imidazole (15). To a cold (ꢀ78 ꢂC), stirred solution of
n-butyl lithium [25.2 mmol; 15.8 mL of a 1.6 M solution
in hexane (Aldrich)] under nitrogen atmosphere in THF
(30.4 mL) was added dropwise a solution of 1-SEM-
imidazole (14)24 (5 g, 25.22 mmol) in the same solvent
(25 mL). After the solution had been stirred at ꢀ78 ꢂC
for 1 h, a solution of phenylacetylglycine methyl ester
(13) (2.39 g, 11.51 mmol) in 30.4 mL THF was added
slowly. The mixture was stirred for 1 h at ꢀ78 ꢂC. The
reaction was quenched with 50 mL of saturated
NaHCO3. The reaction mixture was taken into 125 mL
of water and the THF was removed by rotary evapora-
tion. The residual aqueous layer was extracted with
ethyl acetate. The organic layer was washed with water
(2ꢃ), brine (2ꢃ), dried over Na2SO4 and evaporated to
dryness. The crude product, a dark tan oil, was further
purified by silica gel flash column chromatography
using a ethyl acetate/hexane (4:1) mixture to elute the
colorless product in 42% (1.8 g) yield. 1H NMR
(300 MHz, CDCl3) d 0 (s, 9H), 0.95 (t, J=5.8 Hz, 2H),
3.55 (t, J=6.8 Hz, 2H), 3.70 (s, 2H), 4.82 (br s, 1H),
5.74 (s, 2H), 6.4 (br s, 1H), 7.2–7.44 (m, 7H).
2-(Phenylacetylglycyl) thiazole (2)
The synthesis of 2-(phenylacetylglycyl) thiazole is out-
lined in Scheme 5.
Phenylacetylglycine methyl ester (13). To a stirred ice
cold solution of glycine methyl ester hydrochloride (20
g, 159 mmol) in CHCl3 (440 mL) was added 44.3 mL
(318 mmol) of ice cold triethylamine. To this reaction
mixture 21.0 mL (159 mmol) of phenylacetyl chloride in
100 mL CHCl3 was added dropwise at 0 ꢂC and the
mixture stirred continuously for 2 h. The reaction mix-
ture was then allowed to come to room temperature and
stirred for a further 4 h. The organic layer was washed
with water, dried over MgSO4 and evaporated to dry-
ness. Recrystallization of the crude product from diiso-
propyl ether gave a white crystalline solid in 52% (15.5
2-(Phenylacetylglycyl)imidazole (3). 1-SEM-2-(phenyla-
cetylglycyl)imidazole (15) (50 mg, 0.13 mmol) was stir-
red at room temperature with 3 mL of trifluoroacetic

S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
2043
acid/methylene chloride (2:1) and reaction was followed
by monitoring the disappearance of starting material on
TLC [Rf=0.5 with ethyl acetate/hexane (4:1)]. After 4 h,
the reaction was stopped and the solvents were evapo-
rated to dryness to give the product in protonated form.
This was taken into 10 mL ethyl acetate and washed
with 5 mL saturated NaHCO3. The aqueous layer was
further washed twice with 10 mL ethyl acetate. The
organic layers were combined, dried over Na2SO4 and
evaporated to dryness. Final purification was achieved
by silica gel flash column chromatography with an ethyl
acetate/hexane mixture (17:3) as solvent, followed by
recrystallization from CH3CN to give pale yellow pro-
Thiazole-4-carboxylic acid (18). This compound was
obtained by a small modification of the synthetic route
followed by Jagoe et al.38 To a stirred solution of 9.5 g
(21.1 mol) of formamide in 95 mL THF, which was
cooled to 0 ꢂC, was added all at once 9.5 g (21.4 mmol)
of solid P4S10. After this addition, the ice bath was
removed and the reaction was stirred at room tempera-
ture for 6 h. From the resultant mixture, 35 mL of THF
solution was decanted into a clean dry flask. The solu-
tion was cooled to 0 ꢂC and 10.5 g (63.0 mmol) of solid
3-bromopyruvic acid hydrate (Aldrich) was added all at
once. The reaction mixture was then heated at 45–50 ꢂC
for 5.5 h. After the mixture was cooled to room tem-
perature, the product, which precipitated, was collected
by vacuum filtration and washed with acetone to give
7.1 g (53% yield) of thiazole-4-carboxylic acid hydro-
bromide. This was converted to the neutral form by
treatment with one equivalent of NaHCO3 at 0 ꢂC. The
crude product was purified by recrystallization from
pentanol to give a colorless powder in 27% (3.59 g)
duct in 56% yield (18 mg). Mp160–161 ꢂC; H NMR
1
(300 MHz, CDCl3) d 3.7 (s, 2H), 4.83 (d, J=5.2 Hz, 2H),
7.22–7.42 (m, 7H); nmax (KBr) 1697.7 cmꢀ1 (C¼O); MS
(ESI+): calcd for (C13H13N3O2)+ 244.1, found 244.1.
5-(Phenylacetylglycyl)tetrazole sodium salt (4)
1
The synthesis of 5-(phenylacetylglycyl)tetrazole is out-
lined in Scheme 7.
yield. H NMR (300 MHz, DMSO-d6) d 8.53 (s, 1H),
9.17 (s, 1H).
1-benzyloxymethyl-5-(phenylacetylglycyl)tetrazole (17).
To a stirred cold (ꢀ78 ꢂC) solution of 4 g (21.1 mmol)
1-benzyloxymethyltetrazole (16)25 in 100 mL THF
under a nitrogen atmosphere was added n-butyl lithium
(21.1 mmol; 12.9 mL of 1.6 M solution in hexane)
dropwise. After the solution had been stirred for 0.5 h, a
solution of phenylacetylglycine methyl ester (2.18 g,
10.5 mmol) in 50 mL THF was added slowly and the
mixture was stirred for 40 min. The reaction mixture
was allowed to come to room temperature and the
reaction was quenched with 100 mL saturated NH4Cl.
THF was selectively evaporated and the aqueous layer
was extracted with 150 mL (3ꢃ) ethyl acetate. The
organic layers were combined, washed with 200 mL
brine, dried over MgSO4, and evaporated to dryness.
This crude product was partially purified by silica gel
flash column chromatography using acetone/hexane
mixture (4:6) and, although still contaminated with a
little unreacted ester, was directly used in the next step
Sodium 2-(phenylacetylglycyl) thiazole-4-carboxylate
(5). To a cold (ꢀ78 ꢂC) stirred solution of n-butyl
lithium (35.2 mmol; 22 mL of 1.6 M solution in hexane)
in 20 mL THF was added dropwise a solution of 1.5 g
(11.6 mmol) thiazole-4-carboxylic acid (18) in 45 mL
THF. After the solution had been stirred at ꢀ78 ꢂC for
0.5 h, a solution of 1.19 g (5.7 mmol) phenylacetylgly-
cine methyl ester (13) in 20 mL THF was added slowly .
The reaction mixture was stirred for 45 min at ꢀ78 ꢂC
and for a further 30 min after removing it from the
ꢀ78 ꢂC bath. The reaction was quenched with the suc-
cessive addition of saturated NaHCO3 solution (30 mL)
and water (70 mL). THF was preferentially evaporated
and the aqueous layer was washed with 350 mL ethyl
acetate. After the pH of the aqueous layer was reduced
to 1.5 with HCl, the solution was extracted with 100 mL
ethyl acetate (3ꢃ). The organic layer was then washed
with 100 mL water and 100 mL brine, dried over
MgSO4 and evaporated to dryness to give the crude
product. Purification was achieved using (i) Sephadex
LH-20 column chromatography with methanol as sol-
vent and (ii) G-10 aqueous column chromatography of
the sodium salt to give the final product in 1.5% yield
(26 mg). 1H NMR (300 MHz, DMSO-d6) d 3.58 (s,
2H), 4.74 (d, J=4.7 Hz, 2H,), 7.22–7.39 (m, 5H), 8.20
(s, 1H), 8.52 (br t, 1H); MS (ESI+1): calcd for
(C14H11N2O4SNa) 327.05, found 327.2.
1
of the synthesis. H NMR (300 MHz, CDCl3) d 4.70 (s,
2H), 6.01 (s, 2H), 7.3–7.5 (m, 5H), 8.63 (s, 1H).
5-(phenylacetylglycyl) tetrazole sodium salt (4). The pro-
duct (17) from the previous step (100 mg) was stirred
with 5 mL trifluoroacetic acid (TFA) at room tempera-
ture for 5 h. After evaporation of the TFA, the crude
product was dissolved in water and the pH was brought
to 7.5 with the helpof NaHCO 3. The aqueous layer was
first extracted with 20 mL (3ꢃ) ethyl acetate and then
freeze-dried. Final purification was achieved by aqueous
Sephadex G-10 column chromatography, yielding a
white hygroscopic solid in 23% yield. 1H NMR
(300 MHz, DMSO-d6) d 3.6 (s, 2H), 4.6 (d, J=5.01 Hz,
2H), 7.22–7.36 (m, 5H), 8.34 (br s, 1H); MS (ESI-):
calcd for (C11H10O2N5)ꢀ 244.2, Found 244.1
Acknowledgements
This research was supported by the National Institutes
of Health.
References and Notes
1. Levy, S. B. Sci. Am. 1998, 278, 46.
2. Witte, W. Science 1998, 279, 996.
3. Matagne, A.; Dubus, A.; Galleni, M.; Frere, J.-M. Nat.
Prod. Rep. 1999, 16, 1.
Sodium 2-(phenylacetylglycyl)thiazole-4-carboxylate (5)
The synthesis of sodium 2-(phenylacetylglycyl)thiazole-
4-carboxylate is outlined in Scheme 8.

2044
S. Kumar et al. / Bioorg. Med. Chem. 9 (2001) 2035–2044
4. Sanders, C. C.; Sanders, W. E. Clin. Infect. Dis. 1992, 15, 824.
5. Davies, J. Science 1994, 264, 375.
22. Tao, M.; Bihovsky, R.; Kauer, J. C. Bioorg. Med. Chem.
Lett. 1996, 6, 3009.
6. Pornull, K. J.; Rodrigo, G.; Dornbusch, K. J. Antimicrob.
Chemother. 1994, 34, 943.
23. Lipshutz, B. H.; Vaccaro, W.; Huff, B. Tetrahedron Lett.
1986, 27, 4095.
7. Gonzalez, L. M.; Perez-Diaz, J. C.; Ayala, J.; Casellas,
J. M.; Martinez-Beltran, J.; Bush, K.; Baquero, F. Antimicrob.
Ag. Chemother. 1994, 38, 2150.
8. Jacoby, G. A.; Archer, G. L. New England J. Med. 1991,
324, 601.
24. Whitten, J. P.; Matthews, D. P.; McCarthy, J. R. J. Org.
Chem. 1986, 51, 1891.
25. Satoh, Y.; Moliterni, J. Synlett 1998, 528.
26. Chan, A. W. E.; Golec, J. M. C. Bioorg. Med. Chem. 1996,
4, 1673.
9. Bush, K. Curr. Pharm. Des. 1999, 5, 839.
10. Rice, L. B.; Bonomo, R. A. Drug. Resist. Updates 2000, 3,
178.
11. Livermore, D. M. J. Antimicrob. Chemother. 1993,
31 (Suppl. A), 9.
27. De Maria, P.; Fontana, A.; Spinelli, D.; Macaluso, G.
Gazz. Chim. Ital. 1996, 126, 45.
28. Albert, A. J. Chem. Soc. B 1966, 427.
29. Curley, K.; Pratt, R. F. J. Am. Chem. Soc. 1997, 119,
1529.
12. Leung, D.; Abbenante, G.; Fairlie, D. P. J. Med. Chem.
2000, 43, 305.
30. Lobkovsky, E.; Billings, E. M.; Moews, P. C.; Rahil, J.;
Pratt, R. F.; Knox, J. R. Biochemistry 1994, 33, 6762.
31. Bernstein, N. J.; Pratt, R. F. Biochemistry 1999, 38, 10499.
32. Rahil, J.; Pratt, R. F. Biochem. J. 1991, 275, 793.
33. Martin, R.; Jones, J. B. Tetrahedron Lett. 1995, 36,
8399.
13. Mascaretti, O. A.; Boschetti, C. E.; Danelon, G. O.; Mata,
E. G.; Roveri, O. A. Curr. Med. Chem. 1995, 1, 441.
14. Singh, R.; Micetich, R. G. Invest. Drugs J. 2000, 3, 512.
15. Mehdi, S. Bioorg. Chem. 1993, 21, 249.
16. Trainor, D. A. Trends Pharmacol. Sci. 1987, 8, 303.
17. Metz, W. A.; Peet, N. P. Expert Opin. Ther. Pat. 1999, 9, 851.
18. Edwards, P. D.; Meyer, E. F., Jr.; Vijayalakshmi, J.; Tut-
hill, P. A.; Andisik, D. A.; Gomes, B.; Strimpler, A. J. Am.
Chem. Soc. 1992, 114, 1854.
19. Edwards, P. D.; Wolanin, D. J.; Andisik, D. W.; Davis,
M. W. J. Med. Chem. 1995, 38, 76.
20. Edwards, P. D.; Zottola, M. A.; Davis, M.; Williams, J.;
Tuthill, P. A. J. Med. Chem. 1995, 38, 3972.
21. Matagne, A.; Dubus, A.; Galleni, M.; Frere, J.-M. Nat.
Prod. Rep. 1999, 16, 1.
34. Del Mar, E. G.; Largman, C.; Broderick, W. J.; Geokas,
M. C. Analyt. Biochem. 1979, 99, 316.
35. Largman, C. Biochemistry 1983, 22, 3763.
36. Brutshy, F.; Georgian, V.; Kirk, M.-J.; Kornfeld, E. C.;
Ross, S. D.; Wasserman, H. H.; Wenddler, N.; Woodward, R.
B. In The Chemistry of Penicillin; Clarke, H. T., Johnson, J.
R., Robinson, R., Eds.; Princeton University Press: Princeton,
NJ, 1949; p482.
37. Ireland, R. E.; Liu, L. J. Org. Chem. 1993, 58, 2899.
38. Jagoe, C. T. Ph.D. Thesis, Boston College, 1991; pp 150–
288.