Nanomolar inhibitors of AmpC β-lactamase

Article abstract of DOI:10.1021/ja0288338

β-lactamases are the most widespread resistance mechanism to β-lactam antibiotics, such as the penicillins and the cephalosporins. In an effort to combat these enzymes, a combination of stereoselective organic synthesis, enzymology, microbiology, and X-ra

Full text of DOI:10.1021/ja0288338

Published on Web 12/24/2002
Nanomolar Inhibitors of AmpC â-Lactamase
Federica Morandi,† Emilia Caselli,‡ Stefania Morandi,‡ Pamela J. Focia,§
Jesu´s Bla´zquez,| Brian K. Shoichet,*,† and Fabio Prati*,‡
Contribution from the Department of Pharmaceutical Chemistry, UniVersity of California,
San Francisco, Mission Bay Genentech Hall, 600 16th Street, Mail Box 2240,
San Francisco, California 94143, Dipartimento di Chimica, UniVersita´ degli studi di Modena e
Reggio Emilia, Via Campi 183, Modena, Italy, Department of Molecular Pharmacology and
Biological Chemistry, Northwestern UniVersity, 303 East Chicago AVenue,
Chicago, Illinois 60611, and SerVicio de Microbiologia, Hospital Ramo´n y Cajal,
National Institute of Health (INSALUD), Madrid, Spain
Received October 5, 2002 ; E-mail: shoichet@cgl.ucsf.edu; prati.fabio@unimore.it
Abstract: â-lactamases are the most widespread resistance mechanism to â-lactam antibiotics, such as
the penicillins and the cephalosporins. In an effort to combat these enzymes, a combination of stereoselective
organic synthesis, enzymology, microbiology, and X-ray crystallography was used to design and evaluate
new carboxyphenyl-glycylboronic acid transition-state analogue inhibitors of the class C â-lactamase AmpC.
The new compounds improve inhibition by over 2 orders of magnitude compared to analogous glycylboronic
acids, with Ki values as low as 1 nM. On the basis of the differential binding of different analogues, the
introduced carboxylate alone contributes about 2.1 kcal/mol in affinity. This carboxylate corresponds to the
ubiquitous C3(4)′ carboxylate of â-lactams, and this energy represents the first thermodynamic measurement
of the importance of this group in molecular recognition by class C â-lactamases. The structures of AmpC
in complex with two of these inhibitors were determined by X-ray crystallography at 1.72 and 1.83 Å
resolution. These structures suggest a structural basis for the high affinity of the new compounds and
provide templates for further design. The highest affinity inhibitor was 5 orders of magnitude more selective
for AmpC than for characteristic serine proteases, such as chymotrypsin. This inhibitor reversed the
resistance of clinical pathogens to the third generation cephalosporin ceftazidime; it may serve as a lead
compound for drug discovery to combat bacterial resistance to â-lactam antibiotics.
Introduction
Novel compounds are actively pursued as leads for drug
discovery. Chemically, novel leads provide insight about the
recognition determinants of the target receptor. Biologically,
they can have specificities that substrate analogues lack and
elude barriers or defenses to which substrate analogues fall
victim.
The need for such new biological effects is keenly felt in the
search for inhibitors of â-lactamases. These enzymes are the
major resistance determinants to â-lactam antibiotics, including
the penicillins and the cephalosporins, and threaten public
health.1,2 To combat these enzymes, â-lactam inhibitors such
as clavulanic acid, or “â-lactamase resistant” â-lactams such
as ceftazidime, have been introduced (Figure 1). The similarity
of these â-lactams to the original substrates has allowed
resistance to develop further. Broad-spectrum â-lactamases, such
as the class C â-lactamase AmpC, have spread among bacteria.
Point substitutions have resulted in mutants of once narrow-
Figure 1. Chemical structures of several â-lactam ligands; the R1 side
chains are marked. (a) The substrate cephalothin. (b) The “â-lactamase
resistant” ceftazidime. (c) The inhibitor ATMO-carbacephem.4 (d) The
inhibitor clavulanic acid.
† University of California, San Francisco.
‡ Universita´ degli studi di Modena e Reggio Emilia.
§ Northwestern University.
spectrum class A â-lactamases, leading to enzymes such as
TEM-30 and TEM-64 that are either less inhibited by, or can
simply hydrolyze, the “â-lactamase resistant” compounds.
Recently, new substrate analogues have been described that can
| National Institute of Health (INSALUD).
(1) Matagne, A.; Dubus, A.; Galleni, M.; Frere, J. M. Nat. Prod. Rep. 1999,
16, 1-19.
(2) Rice, L. B.; Bonomo, R. A. Drug Resist. Updates 2000, 3, 178-189.
9
10.1021/ja0288338 CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 685-695
685

A R T I C L E S
Morandi et al.
such as boronic acids. These inhibitors replace the â-lactam
recognition motif with a boronic acid, which makes a reversible,
dative covalent bond with the active site serine residue forming
a tetrahedral adduct (Figure 2a). Replacing the lactam group
with a boronic acid makes these inhibitors novel enough to
evade many of the resistance mechanisms that now jeopardize
â-lactams.11 By the deployment of side chains normally found
in substrates, it has been possible to improve the potency of
these compounds, down to 5.9 nM for TEM-1.9,12 We previously
found that glycylboronic acids (Figure 2b) inhibit AmpC
competitively, with Ki values as low as 20 nM.10
The glycylboronic acids resemble half of the â-lactam
molecule, bearing the R1 side chain of substrates but lacking
recognition elements corresponding to the thiazolidine or
dihydrothiazine rings of penicillins or cephalosporins, respec-
tively (Figure 1a). The absence of a negatively charged group
in a position corresponding to the C4′ position of the dihy-
drothiazine ring seems particularly noteworthy. All â-lactams
bear a carboxylic or sulfonic acid at this position. In class A
â-lactamases, this group is a key recognition element.9,12,13 In
class C â-lactamases, the role of this group is less understood.
Mutant and substrate analyses suggest that such a charged group
is not key for recognition.13 Conversely, structural analyses of
the binding determinants (“hot spots”) of AmpC suggest that
there is a carboxylate binding site on the enzyme.14 The
contribution to binding energy that such a carboxylate might
make, if any, remains in doubt, owing to the irreversible binding
of â-lactams to â-lactamases, preventing equilibrium-based
thermodynamic analyses.
Here, we describe a structure-based approach to the design
and testing of m.carboxyphenylglycylboronic acids (Figure 2c).
The m.carboxylate group is meant to correspond to the C4′
carboxylate of cephalosporins in their tetrahedral high-energy
intermediate form, mimicking both the distance to the tetrahedral
center and the absolute stereochemistry of the chiral carbon.
These new compounds improve inhibition by over 2 orders of
magnitude compared to analogous glycylboronic acids (Figure
2b) that lack this group. Since these inhibitors bind reversibly
to the enzyme, the different Ki values allow for a thermodynamic
analysis of affinity. By making small substitutions to the
inhibitors, we may understand the contribution of the carboxylic
acid to binding (Table 1). The structures of AmpC in complex
with two of these inhibitors, determined by X-ray crystal-
lography, allow us to investigate the structural bases for binding.
The selectivity of these inhibitors for AmpC versus characteristic
serine proteases, which boronic acids are known to inhibit, is
considered, as is their efficacy against pathogenic bacteria
expressing class C â-lactamases.
Figure 2. Comparison between the deacylation high energy intermediate
of a cephalosporin in a serine â-lactamase (a), a transition-state analogue
glycylboronic acid (b), and a transition-state analogue m.carboxyphenyl-
glycylboronic acid (c).
inhibit these mutant3 and broad-spectrum â-lactamases4 with
IC50 values as low as 100 nM (Figure 1). Given their similarity
to substrates, one worries that resistance will rapidly develop
against these new agents as well.
A more ambitious strategy abandons substrate information
altogether, focusing instead on the structure of the receptor as
the sole template for design. Structure-based screening ap-
proaches have discovered inhibitors dissimilar to both substrates
and substrate analogues.5,6 These novel inhibitors may evade
traditional, pre-evolved resistance mechanisms. Conversely,
novel inhibitors of AmpC â-lactamase are relatively weak, with
Ki values in the 25 µM range.6
Between the extremes of substrate analogues and structure-
based discovery lie transition-state analogues (Figure 2),7-10
We arrive at compounds that inhibit AmpC with Ki values
as low as 1 nM. This affinity is realized by adding more and
more functional groups stolen from the â-lactam substrates. If
novelty really is a virtue for inhibitor discovery, it may be asked
whether, in doing so, we have not entered a Faustian bargain.
This is a point to which we will return.
(3) Mourey, L.; Kotra, L. P.; Bellettini, J.; Bulychev, A.; O’Brien, M.; Miller,
M. J.; Mobashery, S.; Samama, J. P. J. Biol. Chem. 1999, 274, 25260-
25265.
(4) Trehan, I.; Morandi, F.; Blaszczak, L. C.; Shoichet, B. K. Chem. Biol. 2002,
9, 1-10.
(5) Toney, J. H.; Fitzgerald, P. M.; Grover-Sharma, N.; Olson, S. H.; May,
W. J.; Sundelof, J. G.; Vanderwall, D. E.; Cleary, K. A.; Grant, S. K.; Wu,
J. K.; Kozarich, J. W.; Pompliano, D. L.; Hammond, G. G. Chem. Biol.
1998, 5, 185-196.
(6) Powers, R. A.; Morandi, F.; Shoichet, B. K. Structure 2002, 10, 1013-
(10) Caselli, E.; Powers, R. A.; Blasczcak, L. C.; Wu, C. Y.; Prati, F.; Shoichet,
B. K. Chem. Biol. 2001, 8, 17-31.
1023.
(7) Crompton, I. E.; Cuthbert, B. K.; Lowe, G.; Waley, S. G. Biochem. J. 1988,
251, 453-459.
(11) Powers, R. A.; Blazquez, J.; Weston, G. S.; Morosini, M. I.; Baquero, F.;
Shoichet, B. K. Protein Sci. 1999, 8, 2330-2337.
(8) Chen, C. C.; Rahil, J.; Pratt, R. F.; Herzberg, O. J. Mol. Biol. 1993, 234,
165-178.
(12) Ness, S.; Martin, R.; Kindler, A. M.; Paetzel, M.; Gold, M.; Jensen, S. E.;
Jones, J. B.; Strynadka, N. C. Biochemistry 2000, 39, 5312-5321.
(13) Frere, J. M. Mol. Microbiol. 1995, 16, 385-395.
(9) Strynadka, N. C.; Martin, R.; Jensen, S. E.; Gold, M.; Jones, J. B. Nat.
Struct. Biol. 1996, 3, 688-695.
(14) Powers, R. A.; Shoichet, B. K. J. Med. Chem. 2002, 45, 3222-3234.
9
686 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Nanomolar â-Lactamase Inhibitors
A R T I C L E S
Table 1. Ki Values of the Glycylboronic Acids against AmpC
a These Ki values were determined in 50 mM KPi pH 7.0.10 The values in Tris are typically 2-fold lower. b Differential free energy of binding relative
to compound 1, calculated at 298 K. Values are calculated using ∆∆G ) -RT lnKi,N/Ki,1, where N represents the compound to which compound 1 is being
compared. Positive values indicate improved affinity. c Not obtained.
Results
Synthesis. The synthesis of (R)-[1-acylamino-1-(3-carboxy-
moiety, showed a greater than 98% diastereoselectivity of the
“one pot” reaction.19,20 All attempts to synthesize 13 and 14
failed, probably because of steric hindrance (see Discussion).
Following the same protocol, we obtained compound 20
(Scheme 2) in 56% overall yield, starting from the (+)-
pinanediol phenylboronate 19. The conversion of the pinanediol
esters to the free boronic acids 15-17 and 21 was achieved
through hydrolysis in degassed HCl under reflux for 1 h.12,21
This treatment also led to the deprotection of the carboxy moiety.
Compound 17 was obtained by extraction with EtOAc, and
compounds 15, 16, and 21 were recovered from the aqueous
phenyl)]methylboronic acids 15-17 employed the general
strategy developed by Matteson et al (Scheme 1);15,16 (+)-
pinanediol was chosen as a chiral auxiliary to guide the
stereochemical course of the Matteson homologation.17,18 Pro-
tection of the carboxy moiety of 3-bromobenzoic acid 5 as the
oxazolidine derivative 6,12 followed by boronation of the
corresponding lithium derivative at -78 °C with B(OCH3)3 and
trans-esterification with (+)-pinanediol, afforded the desired
compound 7 (overall yield of 70% from compound 5). Com-
pound 7 was converted in a “one pot” reaction to compounds
10-12 (16-25% overall yield) to avoid the epimerization of
the intermediate R-chloro derivative 8;15,18 occasionally, com-
pounds 8 and 9 were isolated and characterized. 1H NMR
analysis of compounds 10-12, particularly the diagnostic signals
of the amide NH peaks and the Hendo hydrogen of the pinanyl
1
phase. All compounds were fully characterized by H and 13C
NMR, IR, mass spectra, and elemental analyses except for
compounds 15-17 and 21, for which mass spectra were
unobtainable. Nevertheless, the X-ray structures, combined with
the NMR and IR analyses, unambiguously identified these
compounds.
(15) Matteson, D. S.; Ray, R.; Rocks, R. R.; Tsai, D. J. S. Organometallics
1983, 2, 1536-1543.
(16) Matteson, D. S. Acc. Chem. Res. 1988, 21, 294-300.
(17) Matteson, D. S. Chem. ReV. 1989, 89, 1535-1551.
(18) Matteson, D. S. J. Organomet. Chem. 1999, 581, 51-65.
(19) Tsai, D. J. S.; Jesthi, P. K.; Matteson, D. S. Organometallics 1983, 2, 1543-
1545.
(20) Matteson, D. S.; Sadhu, K. M.; Peterson, M. L. J. Am. Chem. Soc. 1986,
108, 810-819.
(21) Martin, R.; Jones, J. B. Tetrahedron Lett. 1995, 36, 8399-8402.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 687

A R T I C L E S
Morandi et al.
Scheme 1
Scheme 2
Enzymology and Binding Affinities. All boronic acids
inhibitors of AmpC â-lactamase that we have previously studied
have been reversible, fast-on fast-off, competitive inhibi-
tors.10,11,22 The phenylglycylboronic acids, especially the 1 nM
inhibitor compound 16, showed a time-dependent inhibition of
AmpC. Notwithstanding this, they were all reversible inhibitors,
displaying a classic time-dependent recovery from inhibition
during a reaction initiated with substrate (i.e., reaction rates
increased after an initial lag phase and then reached a steady-
state plateau). The time dependence in the inhibition thus reflects
a slow off-rate. Consistent with reversibility, the inhibitors could
be competed off by increasing substrate concentration. Perhaps
the simplest model to explain the time-dependent effect with
these inhibitors is a convolution of their high affinities, a
nanomolar Kd value alone would lead to an off-rate on the
second time scale, and the dative-covalent nature of the serine-
boron bond. There was no significant conformational change
in the enzyme site in the complexed structures, suggesting that
enzyme reorganization did not present a significant barrier to
the inhibitor leaving the site. We have accounted for the
incubation effect in the Ki values reported for these inhibitors
(see Methods).
(Table 1). In these compounds, the addition of the m.carboxy-
phenyl side chain improves the binding energy by about 3 kcal/
mol.
We reasoned that if we began with better leads than 1 and 4,
which had inhibited AmpC with Ki values of 18.5 and 0.320
µM, respectively,10 we would achieve still better inhibitors. We
therefore turned to the m.carboxyphenylglycylboronic acid
analogues of compounds 2 and 3, which had inhibited AmpC
with Ki values of 0.150 and 0.020 µM, respectively (Table 1).
Surprisingly, compound 17, bearing the R1 side chain of
cloxacillin, was 40-fold less potent (higher Ki) than the original
analogue lacking the m.carboxyphenyl group (compound 2). For
this compound, the addition of the m.carboxyphenyl group made
the binding energy about 2 kcal/mol worse (Table 1). We were
unable to synthesize the analogue bearing the ceftazidime side
chain (14).
X-ray Crystallographic Structure Determination. To in-
vestigate the structural bases for this dramatic reversal of relative
affinities, and to understand detailed recognition, we determined
the crystal structure of AmpC in complex with 16 to 1.83 Å
resolution (Table 2). Proline and glycine residues excluded,
92.2% of the amino acids were in the most favored regions of
the Ramachandran plot (7.8% in the additionally allowed
regions).23
To investigate the effect on inhibition of adding an
m.carboxyphenyl group, we first measured the potency of
relatively simple derivatives bearing an acetyl R1 side chain
(compound 15) and an R1 side chain resembling that of
cephalothin (compound 16). These compounds were 150-fold
and 300-fold more potent than the lead compounds lacking the
m.carboxyphenyl side chain, compounds 1 and 4, respectively
The position of the inhibitor in the active site was unambigu-
ously identified in the initial |Fo| - |Fc| difference map
contoured at 3σ. Electron density connected the Oγ of the
catalytic Ser64 to the boron atom of the inhibitors (Figure 3a).
The boron geometry was tetrahedral, as expected, and key
hydrogen bond interactions in the active site closely resemble
(22) Weston, G. S.; Blazquez, J.; Baquero, F.; Shoichet, B. K. J. Med. Chem.
1998, 41, 4577-4586.
(23) Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton, J. M. J.
Appl. Crystallogr. 1993, 26, 283-291.
9
688 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Nanomolar â-Lactamase Inhibitors
A R T I C L E S
Table 2. X-ray Data Collection and Refinement Statistics
AmpC/16 complex
AmpC/21 complex
cell constants (Å; deg)
a ) 117.96; b ) 77.49;
a ) 118.20; b ) 76.70;
c ) 96.91; â ) 115.78
C2
c ) 97.66; â ) 116.31
C2
space group
resolution (Å)
unique reflections
total observations
1.83
66 946
1.72
82 138
257 829
350 811
R
merge (%)
6.0 (42.5)a
4.0 (12.0)a
completeness (%)
I / σ(I)
resolution range for refinement (Å)
number of protein residues
number of water molecules
RMSD bond lengths (Å)
RMSD bond angles (deg)
96.5 (91.0)a
99.2 (96.0)a
19.5 (2.7)a
30.0 (11.2)a
20.0-1.83 (1.87-1.83)a
20.0-1.72 (1.76-1.72)a
716
468
716
587
0.010
1.57
18.7
21.4
34.8b
44.5b
39.1
0.014
1.76
16.7
18.9
22.7b
36.1b
34.1
R
R
cryst (%)
free (%)
average B-factor, protein atoms (Å2)
average B-factor, inhibitor atoms (Å2)
average B-factor, water molecules (Å2)
a Values in parentheses are for the highest resolution shell. b Values cited were calculated for both molecules in the asymmetric unit.
Figure 3. Stereoviews of 2|Fo| - |Fc| electron density maps (blue) of the refined model of AmpC in complex with (a) compound 16 and (b) compound 21,
contoured at 1σ. In green are the simulated-annealing omit electron density maps for the inhibitors, contoured at 3σ. Carbon atoms are colored gray; oxygen
atoms, red; nitrogen atoms, blue; sulfur atoms, yellow; and boron atoms, purple. The putative deacylating water Wat402 is shown as a red sphere. The
figures were generated using SETOR.39
those typically observed in â-lactamase structures with transition
state analogues and with â-lactams (Figure 4a; Table 3).9,10,12,24,25
The O12 of the boronic acid is placed in the “oxyanion”26 or
“electrophilic”25 hole formed by the backbone amide groups of
Ser64 and Ala318 (Table 3). The O13 of the boronic acid
interacts with the putative catalytic base Tyr150.24,27 Two well-
ordered and highly conserved water molecules are also observed.
(24) Lobkovsky, E.; Billings, E. M.; Moews, P. C.; Rahil, J.; Pratt, R. F.; Knox,
J. R. Biochemistry 1994, 33, 6762-6772.
(25) Usher, K. C.; Blaszczak, L. C.; Weston, G. S.; Shoichet, B. K.; Remington,
S. J. Biochemistry 1998, 37, 16082-16092.
(26) Murphy, B. P.; Pratt, R. F. Biochem. J. 1988, 256, 669-672.
(27) Dubus, A.; Ledent, P.; Lamotte-Brasseur, J.; Frere, J. M. Proteins: Struct.,
Funct., Genet. 1996, 25, 473-485.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 689

A R T I C L E S
Morandi et al.
Table 3. Interactions in Inhibitor Bound and Native AmpC
â-Lactamase
distance (Å)
interaction
AmpC/16a
AmpC/21b
apob,c
S64N-O12
3.1
2.7
3.3
2.7
2.8
3.0
2.9
3.0
3.3
2.8
2.6
3.4
2.6
2.7
3.0
3.1
2.8
6.5
2.6
7.1
3.0
3.1
2.9
3.2
2.8
3.3
2.7
3.0
2.9
2.9
3.0
3.2
2.9
2.7
3.2
2.7
2.8
3.1
3.2
2.9
2.9
2.6
2.6
N.P.
N.P.
N.P.
N.P.d
N.P.
N.P.
N.P.
N.P.
N.P.
2.5
3.2
3.1
3.5
3.5
3.8
2.9
2.7
2.9
N.P.
N.P.
N.P.
2.7
A318N-O12
A318O-O12
Y150OH-O13
Wat402-O12
Wat402-O13
Y150OH-K315Nê
Y150OH-S64Oγ
Y150OH-K67Nê
K67Nê-A220O
K67Nê-S64Oγ
Wat402-T316Oγ1
Wat402-Wat403
Wat403-N346Oδ1
Wat403-R349Nη1
A318O-N9
N152Nδ2-O8
Q120Nꢀ2-O8
N152Oδ1-K67Nê
N152Nδ2-Q120Oꢀ1
Wat181-O22
3.0
N.P.
N.P.
N.P.
Wat469-O23
N189Nδ2-O22
a Distances are for monomer 1 of the asymmetric unit. b Distances are
for monomer 2 of the asymmetric unit. c The apo structure used as reference
is PDB code 1KE4.14 d Not present.
binding residues, such as Thr316 and Asn346, are nearby, no
direct interaction is observed to these residues.
Figure 4. Active site of AmpC in complex with (a) compound 16 and (b)
compound 21. In part c, the complex of AmpC with 16 is overlaid with
AmpC in complex with cephalothin (PDB entry 1KVL). Dashed yellow
lines represent key hydrogen bonds. Atoms are colored as in Figure 1, except
for the inhibitors (16 and 21), whose carbon atoms are colored green, and
for the substrate cephalothin, whose carbon atoms are colored magenta.
Red spheres represent water molecules. Interaction distances are listed in
Table 3. Figures 4 and 5 were generated using MidasPlus.40
To investigate the role of this carboxylate, the derivative 21
was made, lacking the carboxylate but maintaining the phenyl
ring. This compound was 35-fold less active than compound
16, which has the carboxylate (Table 1). Comparing the affinity
of 21 (Ki value of 35 nM) to the affinity of 16 (Ki value of 1
nM) suggests that the carboxylate by itself contributes about
2.1 kcal/mol to the interaction energy with AmpC.
Wat402, which appears to be the deacylating water,11,28-30
interacts with both O12 and O13 of the boronic acid (Figure
4a). The second water molecule, Wat403, interacts with Wat402,
the Oδ1 atom of Asn346, and the Nη1 atom of Arg349 (not
shown). The amide group of the inhibitor is placed in the amide
recognition region defined by Asn152 and Ala318.14 The
nitrogen (N9) of the amide group interacts with the backbone
oxygen of Ala318, and the carbonyl oxygen (O8) interacts with
Nδ2 of Asn152. The benzene ring is in van der Waals contact
with Leu119 and Leu293 (distances range from 3.6 to 4.5 Å
for each leucine, respectively), which form a hydrophobic patch
on AmpC.14 Unexpectedly, the carboxylic acid group is observed
to interact with Nδ2 of Asn289 and two ordered water
molecules, Wat181 and Wat469. Although canonical carboxylate
To understand this result, we determined the crystal structure
of AmpC in complex with 21 to 1.72 Å resolution. Proline and
glycine residues excluded, 92.5% of the amino acids were in
the most favored regions of the Ramachandran plot (7.5% in
the additionally allowed regions),23 with other crystallographic
statistics consistent with a well-determined structure (Table 2).
The position of the inhibitor in the active site was unambigu-
ously identified in the initial |Fo| - |Fc| difference map
contoured at 3σ. Electron density connected the Oγ of the
catalytic Ser64 to the boron atom of the inhibitor; the boron
geometry was tetrahedral, as expected (Figure 3b). The AmpC/
21 complex resembles the AmpC/16 complex, making the most
of the same interactions, save for those involving the deleted
carboxylate group (Figure 4b, Table 3).
Microbiology. To investigate the potential of these com-
pounds to reverse antibiotic resistance, we undertook preliminary
antimicrobial activity studies in bacterial cell culture. The
minimum inhibitory concentration (MIC) of ceftazidime against
eight clinically isolated bacterial pathogens producing class C
(28) Patera, A.; Blaszczak, L. C.; Shoichet, B. K. J. Am. Chem. Soc. 2000, 122,
10504-10512.
(29) Crichlow, G. V.; Nukaga, M.; Doppalapudi, V. R.; Buynak, J. D.; Knox,
J. R. Biochemistry 2001, 40, 6233-6239.
(30) Beadle, B. M.; Trehan, I.; Focia, P. J.; Shoichet, B. K. Structure 2002, 10,
413-442.
9
690 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Nanomolar â-Lactamase Inhibitors
A R T I C L E S
Table 4. Synergy of Compounds 16 and 21 with Ceftazidime
against Bacteria Producing â-Lactamase
most species of AmpC a polar residue is found at this position).
This led us to wonder if the carboxylate was really driving
affinity and not some other group. For instance, the phenyl ring
of the m.carboxyphenyl group formed van der Waals interactions
with the hydrophobic patch on AmpC made up of Leu119 and
Leu293, previously found to be a hot spot for ligand interactions
in the enzyme.14 Perhaps it was the phenyl ring, not the
carboxylate, that was responsible for the nanomolar inhibition
of compound 16.
To investigate this possibility, we synthesized compound 21,
the analogue of 16 that retains the phenyl ring but replaces the
m.carboxylate with a hydrogen (Table 3). If the phenyl ring
was responsible for most of the 3.4 kcal/mol improvement in
interaction energy, we would expect the affinity of this analogue
to resemble that of 16. Instead, it lost 35-fold activity compared
to 16, suggesting that the carboxylate is responsible for most
of the affinity gain (Table 1). The structure of the AmpC/21
complex, determined by X-ray crystallography to 1.72 Å, shows
that the two compounds bind similarly in the AmpC site and
that no significant enzyme rearrangement has occurred between
the two complexes.
a
MIC (µg/mL)
b
strain
CAZ
CAZ + 16b,c
CAZ + 21b,c
C. freundii
256
32
256
256
32
256
256
64
8
1
16
8
8
4
16
8
E. coli 1d
E. coli 2d
E. cloacae 1d
E. cloacae 2d
E. cloacae 3d
P. aeruginosa 1d
P. aeruginosa 2d
4
8
16
16
8
16
32
8
a Minimum inhibitory concentration. b CAZ ) ceftazidime. c The ratio
of ceftazidime to inhibitor was 1:1. d Strains defined in Materials and
Methods.
Table 5. Selectivity of 16 and 21 for AmpC versus Serine
Proteases
enzyme
IC50 (µM) for 16
IC50 (µM) for 21
AmpC
0.0026
150
0.090
6.0
R-chymotrypsin
â-trypsin
elastase
.1000
.1000
500
128
This returns us to the original premise that the carboxylate
is a key recognition feature for AmpC. Since the phenylgly-
cylboronic acids are reversible inhibitors, we can for the first
time assign an energetic value to having this functional group:
2.1 kcal/mol. This is a large value, considering the need to
desolvate the carboxylate, though not unprecedented for a polar-
ionic interaction.31 Whether this energy owes directly to the
hydrogen bond between the carboxylate and Asn289, or to more
general interactions with the residues in the carboxylate binding
site (Figure 4c),24,28-30,32 must await protein mutagenesis
studies.31 What is clear now is that this carboxylate contributes
significantly to the binding of this inhibitor, and by extension
to the recognition of â-lactam substrates by class C â-lacta-
mases.
â-lactamases ranged from 256 µg/mL to 32 µg/mL. Both 16
and 21 showed synergy with ceftazidime against all strains.
Compound 16 was slightly more potent than compound 21,
improving the MIC values of ceftazidime by between 8- and
32-fold (Table 4).
Selectivity. To investigate the selectivity of these compounds,
16 and 21 were tested against the serine proteases R-chymo-
trypsin, â-trypsin, and elastase (Table 5). Compound 16 was
57 000-fold more selective for AmpC over R-chymotrypsin and
190 000-fold more selective for AmpC over elastase. Compound
21 was 60-fold and 1400-fold more selective for AmpC over
R-chymotrypsin and elastase, respectively. Neither compound
had any measurable activity against â-trypsin below 1 mM.
It is appropriate to ask then why, if the carboxylate contributes
so much to binding affinity, is the m.carboxyphenylglycyl-
boronic acid 17 such a poor inhibitor, nearly 40-fold worse than
its analogous glycylboronic acid 2, the latter of which lacks
the carboxylate? An overlay of the crystal structures of the
AmpC complex with the glycylboronic acid 210 and that of
AmpC/16 explains this unanticipated drop in affinity (Figure
5). The bulky R1 side chains of 2 and 17, which bend back
toward the boronic acid group, will sterically clash with the
introduced m.carboxyphenyl moiety in 17, disrupting its interac-
tions with the enzyme; this would also be true of 14.4,33 The
inability to even synthesize the m.carboxyphenylglycylboronic
acid 14, an analogue of the 20 nM glycylboronic acid 3,10
presumably also reflects this crowding. This was disappointing,
as we had (naively) hoped that 14 would see the same
improvement as 16, reaching a Ki better than 100 pM. When
transition-state analogues are being designed for â-lactamases,
care must be taken that the two side chains, each of which
individually can contribute significantly to binding, are sterically
consistent with one another.
Discussion
The aspect of this study that first attracted our attention was
the high affinity of compound 16, which inhibits AmpC
â-lactamase with a Ki value of 1 nM (Table 1). This is a 300-
fold improvement over the parent compound 4 (Table 1), which
lacks the m.carboxyphenyl group.10 This result supports the view
that a C3(4)′ acidic group, which is ubiquitous among â-lactams,
is a key recognition feature in AmpC. Although this is well
accepted for the class A â-lactamases, the contribution of the
â-lactam carboxylate to recognition by the class C enzymes has
been uncertain.13 It is appropriate to consider what enzyme
groups are responsible for complementing this functionality.
To answer this question, we determined the structure of the
complex between AmpC and compound 16 by X-ray crystal-
lography (Figure 4a, Table 2). We expected to see the carbox-
ylate of 16 interact with Asn346 or Thr316, which both
crystallography and mutagenesis studies suggest are responsible
for interacting with the C3(4)′ carboxylates of penicillins and
cephalosporins.24,28-30 Instead, we found that the carboxylate
hydrogen bonded with the nearby Asn289 and two ordered water
molecules new to this structure (Figure 4a and c; Table 3).
Asn289, though in the active site region, has not previously
been implicated as a functional residue in AmpC and is only
modestly conserved among class C â-lactamases (although in
(31) Fersht, A. R.; Shi, J.; Knill-Jones, J.; Lowe, D. M.; Wlkinson, A. J.; Blow,
D. M.; Brick, P.; Carter, P.; Waye, M. M. Y.; Winter, G. Nature 1985,
314, 235-238.
(32) Oefner, C.; D’Arcy, A.; Daly, J. J.; Gubernator, K.; Charnas, R. L.; Heinze,
I.; Hubschwerlen, C.; Winkler, F. K. Nature 1990, 343, 284-288.
(33) Powers, R. A.; Caselli, E.; Focia, P. J.; Prati, F.; Shoichet, B. K.
Biochemistry 2001, 40, 9207-9214.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 691

A R T I C L E S
Morandi et al.
tamases, providing energetic values for interactions observed
in the structures of their complexes and by extension those of
â-lactams. These crystal structures are templates for yet further
improvement in the potency of this family of inhibitors. Whether
such potency can be converted into therapeutic use as antire-
sistance agents will depend at least partly on whether their
similarity to substrates makes them more or less sensitive to
existing resistance mechanisms. This consideration may bear
upon the design of many antiresistance therapeutics; studies to
address it are only now beginning.
Materials and Methods
Synthesis and Analysis. All reactions were performed under argon
using oven-dried glassware. Solvents were dried according to classical
procedures. A cold bath at -100 °C was prepared by addition of liquid
nitrogen to a precooled (-80 °C) mixture of 1:1 EtOH/MeOH.
Chromatographic purification of the compounds was performed on silica
gel (0.05-0.20 mm). Melting points were obtained on a Bu¨chi 510
apparatus and are uncorrected. Optical rotations were recorded at 20
°C on a Perkin-Elmer 241 polarimeter and are in 10-1 deg cm2 g-1. IR
spectra were determined in KBr pellets (for solids) and films (for
Figure 5. Overlay of the AmpC/16 and AmpC/2 complexes (PDB entry
1FSY for AmpC/2),10 each determined by X-ray crystallography. Carbon
atoms of 16 are colored gray, and carbon atoms of 2 are colored green.
Even with this caveat, compounds of this family have several
attractive features. Compound 16 is highly selective for AmpC,
having little affinity for serine proteases such as chymotrypsin
(Table 5). This high level of selectivity may reflect the
hydrophilicity of compound 16 and its chiral display of relatively
dense functionality. Whereas the activity of these compounds
in cell culture (Table 4) is several orders of magnitude worse
than their activity as enzyme inhibitors, they are nevertheless
relatively potent at reversing the resistance of clinical pathogens
such as E. cloacae, which are currently such a problem in
hospitals.
liquids) on a Perkin-Elmer 1600 Series spectrophotometer. 1H and 13
C
NMR spectra were recorded on a Bruker DPX-200 (at 200 and 50 MHz,
respectively) spectrometer: chemical shifts are reported in δ values
from TMS as the internal standard. Mass spectra were determined on
a Finnigan MAT SSQ A mass spectrometer (EI, 70 eV). Elemental
analyses were performed on a Carlo Erba Elemental Analyzer 1110.
2-(3-Bromophenyl)-4,4-dimethyl-4,5-dihydro-oxazole (6) was synthe-
sized as described.12
(+)-Pinanediol 3-(4,4-Dimethyl-4,5-dihydro-oxazol-2-yl)phenyl-
boronate (7). n-BuLi (2.5 mL of a 2.5 M solution in hexane, 6.23
mmol) was added dropwise with stirring to a solution of 6 (1.51 g,
5.93 mmol) in THF (9.5 mL) at - 78 °C under argon. After 30 min,
a solution of trimethylborate (0.7 mL, 5.93 mmol) in THF (2 mL) was
added, and the mixture was stirred for 1.5 h; thereafter, the resulting
yellow solution was quenched with TMSCl (0.75 mL, 5.93 mmol) and
allowed to reach rt. After 1 h, (+)-pinanediol (1.01 g, 5.93 mmol)
dissolved in a minimum amount of anhydrous Et2O was added as one
portion to the solution and then stirred overnight. The reaction mixture
was partitioned in Et2O (16 mL) and H2O (10 mL), and the aqueous
phase was extracted with Et2O (2 × 10 mL). The combined organic
phases were dried on MgSO4, filtered, and concentrated to give an
orange oil, which was purified by chromatography (7:3 EtPet/EtOAc)
and crystallization (EtPet), affording 7 (1.14 g, 54%) as a white
crystalline solid, mp 98-101 °C, [R]D ) +11.2 (c 1.5, CHCl3). IR
We return, then, to the question raised in the Introduction:
how novel is this series, and what are its prospects as
antiresistance agents; will resistance easily arise against it? After
all, the increased affinity of the transition-state analogues has
been achieved through a series of raids on the â-lactam arsenal,
resulting in compounds that, except for their boronic acid group,
are by now looking rather like â-lactams.
Two views may be considered. The first suggests that
resistance will be hardest to develop against analogues that most
closely resemble substrates. To achieve resistance, an enzyme
must distinguish between the inhibitor and the substrate, since
it must still act on the latter, and the more the inhibitor resembles
the substrate, the more difficult this will be. In this model,
transition-state analogues such as compound 16 increasingly
resemble the substrate, while remaining impervious to hydroly-
sis, and should be difficult to develop resistance against.
1
(KBr): 1646 cm-1. H NMR (CDCl3): δ 0.81 (3H, s, pinanyl CH3),
1.22 (1H, d, J ) 10.5 Hz, pinanyl Hendo), 1.33 (3H, s, pinanyl CH3),
1.40 (6H, s, 2CH3), 1.50 (3H, s, pinanyl CH3), 1.5-2.5 (5H, m, pinanyl
protons), 4.22 (2H, s, CH2O), 4.47 (1H, dd, J ) 8.8, 2.0 Hz, pinanyl
CHOB), 7.43 (1H, t, J ) 7.7 Hz, H5 arom.), 7.92 (1H, dt, J ) 7.7, 1.5
Hz, H6 arom.), 8.05 (1H, dt, J ) 7.7, 1.5 Hz, H4 arom.), 8.43 (1H, br
s, H2 arom.). 13C NMR (CDCl3): δ 24.4, 26.9, 27.5, 28.8 (2C), 29.1,
35.9, 38.6, 40.0, 51.9, 68.0, 78.8, 79.5, 86.8, 128.1, 131.2, 135.0, 137.8,
162.5 (CB and aromatic quaternary C not seen). EIMS: m/z 353 (M+),
338 (base peak), 323, 282, 242, 202, 186, 130, 103, 67, 55. Anal. Calcd
for C21H28NO3B: C, 71.39; H, 7.99; N, 3.96. Found: C, 71.01; H,
8.65; N, 3.89.
A second view recalls that most bacterial resistance mech-
anisms are ancient in the biosphere and preadapted to recognize
an antibiotic class. In the case of â-lactams, these mechanisms
include receptors that bind â-lactams and up-regulate â-lac-
tamases, the deletion of porin channels through which â-lactams
diffuse, and point mutants, such as those in TEM and SHV
â-lactamases, that reduce affinity for â-lactam inhibitors such
as clavulanate. In this model, the more a â-lactamase inhibitor
resembles a â-lactam, the more subject it will be to these pre-
evolved resistance mechanisms.
One Pot General Procedure for the Synthesis of (+)-Pinanediol
(1R)-1-Acylamino-1-[3-(4,4-dimethyl-4,5-dihydro-oxazol-2-yl)phen-
yl]methylboronate (10-12). Dichloromethyllithium was generated by
adding n-BuLi (1.24 mL of a 2.5 M solution in hexane, 3.11 mmol)
dropwise to a solution of CH2Cl2 (0.29 mL, 4.53 mmol) in THF (10
mL) with stirring at -100 °C under argon: toward the end of the BuLi
addition, precipitation of the white microcrystalline LiCHCl2 became
In this work, we have used stereoselective organic synthesis,
enzymology, microbiology, and X-ray crystallography to design
m.carboxyphenylglycylboronic acids that inhibit class C â-lac-
tamases with Ki values as low as 1 nM. As reversible inhibitors,
these molecules probe the recognition determinants of â-lac-
9
692 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Nanomolar â-Lactamase Inhibitors
A R T I C L E S
evident. After 30 min, the mixture was treated with the above pinanediol
boronate 7 (1.00 g, 2.83 mmol) and allowed to reach rt with stirring.
The tetrahedral boronate adduct precipitated as an abundant white solid
at -80 °C and redissolved upon warming. After 1 h at 0 °C, the solution
was cooled to -78 °C; LiN(TMS)2 (3.11 mL of a 1 M solution in
THF, 3.11 mmol) was added, and the resulting solution was allowed
to warm gradually to 20 °C and stirred overnight. The desilylation,
acylation, and purification of the product were carried out as described
for compounds 10-12.
2-thiophenacetylchloride (383 µL, 3.11 mmol) in THF (1 mL) was
slowly added, and the resulting mixture was allowed to warm to rt
overnight. EtOAc (65 mL) and H2O (15 mL) were added, and the
aqueous phase was extracted with EtOAc (2 × 15 mL). The combined
organic phases were dried (MgSO4) and concentrated to give an orange
oil which was purified by gradient chromatography (9:1 EtOAc/EtPet,
95:5 EtOAc/MeOH), affording 11 (360 mg, 25% overall yield from 7)
as a pale orange solid, mp 120 °C, [R]D ) -24.2 (c 2.1, CDCl3), de >
1
98%. IR (KBr): 1646, 1600 cm-1. H NMR (CDCl3): δ 0.84 (3H, s,
Analytical samples of compounds 8 and 9 were prepared and
characterized.
pinanyl CH3), 1.23 (1H, d, J ) 10.5 Hz, pinanyl Hendo), 1.27 (1H, s,
pinanyl CH3), 1.37 (3H, s, pinanyl CH3), 1.38 (3H, s, CH3), 1.39 (3H,
s, CH3), 1.5-2.5 (5H, m, pinanyl protons), 4.13 (1H, br d, J ) 2.0 Hz,
CHB), 4.01 (2H, br s, CH2CONH), 4.11 (2H, s, CH2O), 4.25 (1H, dd,
J ) 8.8, 2.0 Hz, pinanyl CHOB), 6.86 (1H, br, NHCO), 7.01 (2H, m,
CHCHS-CHCS), 7.33 (3H, m, CHCHS-H6-H5 arom.), 7.77 (2H, m,
H2-H4 arom.). 13C NMR (CDCl3): δ 24.8, 27.2, 28.0, 28.58, 28.64,
29.5, 34.3, 36.9, 38.8, 40.6, 48.4 (br, CB), 52.9, 67.2, 77.7, 80.7, 84.8,
126.5, 126.7, 126.9, 128.0, 128.7, 129.1, 132.0, 134.6, 142.4, 175.5
(two quaternary C not seen). EIMS: m/z 506 (M+, base peak), 473,
327, 270, 203, 135, 97, 93, 67, 55. Anal. Calcd for C28H35BN2O4S: C,
66.40; H, 6.97; N, 5.53; S, 6.33. Found: C, 66.26; H, 7.23; N, 5.29; S,
6.58.
(+)-Pinanediol (1S)-1-Chloro-1-[3-(4,4-dimethyl-4,5-dihydro-oxazol-
2-yl)phenyl]methylboronate (8): [R]D ) +3.5 (c 2.3, CHCl3). IR
1
(film): 1650 cm-1. H NMR (CDCl3): δ 0.86 (3H, s, pinanyl CH3),
1.17 (1H, d, J ) 10.5 Hz, pinanyl Hendo), 1.31 (3H, s, pinanyl CH3),
1.41 (6H, s, 2CH3), 1.43 (3H, s, pinanyl CH3), 1.5-2.5 (5H, m, pinanyl
protons), 4.13 (2H, s, CH2O), 4.41 (1H, dd, J ) 8.8 Hz, 2.0, pinanyl
CHOB), 4.58 (1H, br s, CHB), 7.41 (1H, t, J ) 7.7 Hz, H5 arom.),
7.64 (1H, dt, J ) 7.7, 1.5 Hz, H6 arom.), 7.88 (1H, dt, J ) 7.7, 1.5 Hz,
H4 arom.), 8.02 (1H, t, J ) 1.5 Hz, H2 arom.). 13C NMR (CDCl3): δ
24.3, 26.6, 27.4, 28.7 (3C), 35.6, 38.6, 39.7, 45.0 (br, CB), 51.7, 67.7,
79.3, 79.6, 87.5, 128.1, 128.6, 129.0, 131.7, 139.6, 162.2. EIMS: m/z
401-403 (M+), 386-388, 371-373, 250-252, 223-225 (base peak),
189, 135, 93, 67, 55.
(+)-Pinanediol (1R)-1-{[3-(2-Chlorophenyl)-5-methylisoxazole-
4-carbonyl]amino}-1-[3-(4,4-dimethyl-4,5-dihydro-oxazol-2-yl)phen-
yl]methylboronate (12). The boronate 12 was prepared on the same
scale following the procedure described for 11 using 3-(2-chlorophenyl)-
5-methylisoxazolyl-4-carbonylchloride (796 mg, 3.11 mmol) as the
acylating agent. After extraction, the combined organic phases were
dried (MgSO4) and concentrated in vacuo to give an orange oil, which
was purified by gradient chromatography (Et2O, 95:5 Et2O/MeOH) and
triturated with Et2O, affording 12 (276 mg, 16% overall yield from 7)
as a pale yellow solid, mp 140 °C, [R]D ) +182.7 (c 2.0, CHCl3),
(+)-Pinanediol (1R)-1-(N-Bistrimethylsilylamino)-1-[3-(4,4-dimethyl-
4,5-dihydro-oxazol-2-yl)phenyl]methyboronate (9). [R]D ) -2.7 (c 2.0,
1
CHCl3). IR (film): 1650 cm-1. H NMR (CDCl3): δ 0.13 (18H, 2Si-
(CH3)3), 0.89 (3H, s, pinanyl CH3), 1.30 (1H, d, J ) 10.5 Hz, pinanyl
Hendo), 1.35 (3H, s, pinanyl CH3), 1.41 (6H, s, 2CH3), 1.46 (3H, s,
pinanyl CH3), 1.5-2.5 (5H, m, pinanyl protons), 4.12 (2H, s, CH2O),
4.14 (1H, br s, CHB), 4.39 (1H, dd, J ) 8.8, 2.0 Hz, pinanyl CHOB),
7.33 (1H, t, J ) 7.7 Hz, H5 arom.), 7.62 (1H, dm, J ) 7.7 Hz, H4
arom.), 7.79 (1H, dm, J ) 7.7 Hz, H6 arom.), 8.05 (1H, m, H2 arom.).
13C NMR (CDCl3): δ 2.8 (6C), 24.4, 26.9, 27.4, 28.7 (3C), 35.8, 38.6,
39.9, 47.0 (br, CB), 51.9, 67.8, 79.0, 79.4, 86.5, 125.8, 127.1, 127.6,
127.9, 129.7, 145.587, 163.1. EIMS: m/z 526 (M+), 511, 453, 275,
203, 135, 130, 93, 73 (base peak), 67, 55.
1
de > 98%. IR (KBr): 1647, 1601 cm-1. H NMR (CDCl3): δ 0.83
(3H, s, pinanyl CH3), 1.10 (1H, d, J ) 10.5 Hz, pinanyl Hendo), 1.28
(3H, s, pinanyl CH3), 1.33 (3H, s, pinanyl CH3), 1.41 (6H, s, 2CH3),
1.5-2.5 (5H, m, pinanyl protons), 2.84 (3H, s, CH3CON), 4.12 (2H, s,
CH2O), 4.19 (1H, dd, J ) 8.8, 2.0 Hz, pinanyl CHOB), 4.22 (1H, br
d, J ) 2.4 Hz, CHB), 6.10 (1H, br d, J ) 2.4 Hz, NHCO), 7.19 (1H,
dt, J ) 7.5, 1.6 Hz, H4 arom.), 7.30 (1H, t, J ) 7.5 Hz, H5 arom.),
7.51 (4H, m, other aromatic protons), 7.67 (1H, m, J ) 1.6 Hz, H2
arom.), 7.78 (1H, dt, J ) 7.5, 1.6 Hz, H6 arom.). 13C NMR (CDCl3):
δ 13.8, 24.4, 26.7, 27.8, 28.8 (2C), 29.0, 36.2, 38.5, 40.0, 45.2 (br,
CB), 52.2, 67.9, 78.0, 79.5, 85.6, 126.3, 126.7, 128.0, 128.7, 129.6,
130.9, 131.9, 132.4, 134.4,140.7, 163.7, 176.2, (isoxazole quaternary
C not seen). EIMS: m/z 601-603 (M+), 566 (base peak), 449, 423-
425, 338, 316, 203, 178, 131, 93, 67, 55. Anal. Calcd for C33H37BN3O5-
Cl: C, 65.85; H, 6.20; N, 6.98. Found: C, 65.59; H, 6.28; N, 6.83.
(+)-Pinanediol (1R)-1-Acetylamino-1-[3-(4,4-dimethyl-4,5-dihy-
dro-oxazol-2-yl)phenyl]methylboronate (10). After 16 h at rt, the
reaction mixture containing the silylamino derivative 9 was cooled at
-78 °C and treated with a solution of Ac2O (1.13 mL, 12.00 mmol)
and AcOH (194 µL, 3.40 mmol) in THF (2 mL), and then was allowed
to warm to rt and stirred overnight. The solution was partitioned in
EtOAc (60 mL) and H2O (12 mL), and the aqueous phase was extracted
with EtOAc (2 × 15 mL). The combined organic phases were washed
with 5% NaHCO3, H2O (6.5 mL), and saturated NaCl (6.5 mL), dried
(MgSO4), and concentrated in vacuo to give a brownish oil which was
purified by chromatography (95:5 Et2O/MeOH) and crystallization
(EtOAc), affording 10 (291 mg, 24% overall yield from 7) as a white
solid, mp 196 °C, [R]D ) -136.9 (c 2.1, CHCl3), de > 98%. IR
General Procedure for the Synthesis of Free Boronic Acids 15-
17. The pinanediol esters 10-12 (0.30 mmol) were deprotected in
degassed HCl (3 N, 7 mL) for 1 h at 120 °C under argon, and the
resulting mixture was extracted with EtOAc (2 × 15 mL). The
corresponding hydrolysis products 15 and 16 were isolated from the
aqueous phase, while compound 17 was isolated from the organic phase.
1
(KBr): 1648, 1600 cm-1. H NMR (CDCl3): δ 0.80 (3H, s, pinanyl
CH3), 1.22 (9H, br s, 2CH3-pinanyl CH3), 1.32 (3H, s, pinanyl CH3),
1.39 (1H, d, J ) 10.5 Hz, pinanyl Hendo), 1.5-2.5 (8H, m, pinanyl
protons and CH3CONH at 2.16, d, J ) 0.67 Hz), 3.92 (1H, br s, CHB),
4.05 (3H, m, pinanyl CHOB-CH2O), 7.28 (2H, m, H5-H6 arom.),
7.68 (2H, m, H2-H4 arom.), 10.18 (1H, br, NHCO). 13C NMR
(CDCl3): δ 18.1, 24.5, 27.0, 27.8, 28.5, 28.7, 29.5, 37.1, 38.5, 40.5,
51.0 (br, CB), 53.0, 67.7, 76.6, 79.4, 83.4, 125.0, 125.8, 127.7, 128.4,
130.3, 142.7, 163.363, 177.0. EIMS: m/z 424 (M+), 381, 272, 245
(base peak), 228, 203, 131, 93, 67, 55. Anal. Calcd for C24H33BN2O4:
C, 67.90; H, 7.84; N, 6.60. Found: C, 68.32; H, 7.59; N, 6.31.
(+)-Pinanediol (1R)-1-(2-Thienylacetylamino)-1-[3-(4,4-dimethyl-
4,5-dihydro-oxazol-2-yl)phenyl]methylboronate (11). Anhydrous MeOH
(1.24 mL of a 2.5 M solution in THF, 3.11 mmol) was added to the
solution containing the silylamino derivative 9 and was stirred for 1 h
at -10 °C and then for 1 h at rt. The solution was cooled at -78 °C,
(1R)-1-Acetylamino-1-(3-carboxyphenyl)methylboronic Acid (15).
The free boronic acid 15 was isolated from the aqueous phase after
removal of the solvent under reduced pressure. Crystallization of the
crude residue (boiling acetone) afforded 15 together with an equimolar
amount of 2-methyl-2-amino-1-propanol as a light brown solid (58 mg,
1
59%), mp 90-92 °C. IR (KBr): 3425, 1698, 1620 cm-1
. H NMR
(DMSO): δ 1.20 (6H, s, 2CH3), 2.10 (4H, br s, CH3CONH-CHB),
3.35 (2H, s, CH2O), 6.98 (1H, d, J ) 7.8 Hz, H6 arom.), 7.28 (1H, t,
J ) 7.8 Hz, H5 arom.), 7.56 (1H, s, H2 arom.), 7.69 (1H, d, J ) 7.8
Hz, H4 arom.), 8.02 (3H, br, B(OH)2-COOH), 9.42 (1H, br, NHCO).
13C NMR (DMSO): δ 18.5, 23.1 (2C), 31 (br, CB), 55.3, 37.3, 126.5,
126.7, 128.3, 130.9, 143.7, 168.5, 176.0. The EIMS was not obtainable.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 693

A R T I C L E S
Morandi et al.
(1R)-1-(2-Thienylacetylamino)-1-(3-carboxyphenyl)methylboron-
ic Acid (16). The free boronic acid 16 was isolated from the aqueous
phase after removal of the solvent under reduced pressure. Crystal-
lization of the crude residue (H2O) afforded 16 as an ivory solid (51
mg, 55%), mp 228 °C (dec), [R]D ) -65.5 (c 0.5, CH3OH). IR (KBr):
3398, 1704, 1606 cm-1. 1H NMR (CD3OD): δ 3.98 (1H, br s, CHB),
4.27 (2H, s, CH2CONH), 7.05 (1H, dd, J ) 5.1, 3.2 Hz, CHCHS),
7.19 (1H, br d, J ) 3.2 Hz, CHCS), 7.41 (3H, m, CHCHS and H5-H6
arom.), 7.85 (2H, m, H2-H4 arom.). 13C NMR (CD3OD): δ 31.0, 53.0
(br, CB), 55.3, 126.2, 127.0, 127.2, 127.4, 128.2, 128.4, 130.6, 130.8,
133.5, 141.5, 168.9, 178.5. Anal. Calcd for C14H14BNO5S: C, 52.69;
H, 4.42; N, 4.39; S, 10.05. Found: C, 52.71; H, 4.59; N, 4.31; S, 9.87.
The EIMS was not obtainable.
30.9, 53.4 (br, CB), 125.7, 125.9, 126.0, 127.3, 128.0, 128.1, 129.0,
133.6, 178.0. Anal. Calcd for C13H14BNO3S: C, 56.75; H, 5.13; N,
5.09; S, 11.65. Found: C, 56.53; H, 5.07; N, 4.90; S, 11.57. The EIMS
was not obtainable.
Enzymology. The phenylglycylboronic acids were dissolved in
DMSO at a concentration of 50 mM; more dilute stocks (10 mM to
1µM) were subsequently prepared as necessary. Kinetic measurements
were performed using nitrocefin as substrate in 50 mM Tris buffer,
pH 7.0, and monitored in an HP8453 UV-vis spectrophotometer. The
concentration of AmpC was determined spectrophotometrically in stock
solutions made from lyophilized powder; this enzyme had been
previously expressed and purified.25 The concentration of enzyme in
all reactions was 1.75 nM, except when assaying compound 16, where
the concentration used was reduced 2- to 4-fold (the results from the
0.8 nM enzyme concentration reactions are reported here). To determine
Ki values, inhibitor and enzyme were incubated together at their final
concentration in the cuvettes for 5 min before the reaction was initiated
by the addition of 200 µM substrate. Ki values for compounds 15, 16,
17, and 21 were obtained by comparison of progress curves in the
presence and absence of inhibitor, using the method described by
Waley,34 which has been widely used for boronic acid inhibitors of
â-lactamases. In these analyses, sufficient inhibitor was used to give
at least 50% inhibition; the Ki values reported are the averages calculated
from reactions at six different inhibitor concentrations, each of which
was repeated three times. The lowest concentration of the 1 nM inhibitor
16 in these assays was 3 nM. Because this concentration approaches
that of enzyme, inhibition values for these reactions were checked
against reactions run with enzyme at 0.4 nM to ensure that the measured
inhibition values were not significantly perturbed by the enzyme’s effect
on free inhibitor concentration. In all reactions, rates were measured
after reactions had overcome their initial lag phase and had reached a
steady state.
(1R)-1-{[3-(2-Chlorophenyl)-5-methylisoxazole-4-carbonyl]amino}-
1-(3-carboxyphenyl)methylboronic Acid (17). The free boronic acid
17 was obtained from the organic phase after anhydrification (MgSO4).
Removal of the solvent under reduced pressure and crystallization of
the crude residue (acetone/Et2O) afforded an ivory solid (82 mg, 50%),
mp 180-190 °C, [R]D ) -36.3 (c 0.3, CH3OH). IR (KBr): 3409, 1702,
1
1630 cm-1. H NMR (DMSO): δ 2.65 (3H, s, CH3CON), 3.58 (1H,
br s, CHB), 6.90 (1H, d, J ) 7.2 Hz, H6 arom.), 7.11 (1H, t, J ) 7.2
Hz, H5 arom.), 7.50 (8H, m, other aromatic protons-B(OH)2), 8.9 (0.5
H, br, NHCO), 12.6 (1H, br, COOH). 13C NMR (DMSO): δ 12.5,
51.0 (br, CB), 125.7, 126.2, 126.7, 127.3, 127.4, 129.7, 129.8, 130.4,
131.5, 131.6, 131.7, 132.6, 142.6, 159.2, 163.8, 167.5, 172.8. Anal.
Calcd for C19H16BClN2O6: C, 55.04; H, 3.89; N, 6.76. Found: C, 54.91;
H, 4.03; N, 6.51. The EIMS was not obtainable.
(+)-Pinanediolphenylboronate (19). A solution of (+)-pinanediol
(665 mg, 3.9 mmol) and phenylboronic acid (18) (476 mg, 3.9 mmol)
in THF (5 mL) was stirred for 10 min, concentrated, and distilled to
yield 21 (879 mg, 88%), bp 135 °C (1.5 mmHg) as a viscous colorless
oil which solidified on standing, mp 50 °C, [R]D ) +14.9 (c 2.3,
CHCl3). IR (KBr): 2910, 1361, 1094, 701 cm-1. 1H NMR (CDCl3): δ
0.92 (3H, s, pinanyl CH3), 1.25 (1H, d, J ) 10.5 Hz, pinanyl Hendo),
1.34 (3H, s, pinanyl CH3), 1.50 (3H, s, pinanyl CH3), 1.9-2.6 (5H, m,
pinanyl protons), 4.48 (1H, dd, J ) 8.8, 2.0 Hz, pinanyl CHOB), 7.34-
7.54 (3H, m, Hm-Hp), 7.85 (2H, dd, J ) 7.9 Hz, 1.6, Ho). 13C NMR
(CDCl3): δ 24.4, 26.9, 27.5, 29.1, 36.0, 38.6, 40.0, 51.9, 78.7, 86.6,
128.1, 131.5, 135.2. Aromatic CB not seen. EIMS: m/z 256 (M+), 241,
215, 187, 173, 108 (base peak), 93, 77. Anal. Calcd for C16H21BO2:
C, 75.02; H, 8.26. Found: C, 74.77; H, 8.53.
The selectivity of compounds 16 and 21 was tested by determining
their activity against the serine proteases R-chymotrypsin (bovine
pancreatic), â-trypsin (bovine pancreatic), and elastase (porcine pan-
creatic), all from Sigma (St. Louis, MO). The substrates for R-chy-
motrypsin (succinyl-Ala-Ala-Pro-Phe-p-nitroanilide) and â-trypsin (N-
benzoyl-L-arginine ethyl ester) were also purchased from Sigma. The
elastase substrate (MeOSuc-Ala-Ala-Pro-Val-pNA) was purchased from
Calbiochem (San Diego, CA). Substrates were diluted from 20 mM
DMSO stock solutions, and all reactions were performed in 50 mM
Tris buffer, pH 7.0, 25 °C. For R-chymotrypsin, 0.001 mg/mL enzyme
and the inhibitor at its final concentration were incubated in the cuvette
for 5 min before the reaction was initiated by the addition of 200 µM
substrate. The reaction was monitored at 410 nm. For â-trypsin, 0.004
mg/mL enzyme and the inhibitor at its final concentration were
incubated in the cuvette for 5 min before the reaction was initiated by
the addition of 200 µM substrate. The reaction was monitored at 253
nm. For elastase, 0.006 mg/mL enzyme and the inhibitor at its final
concentration were incubated in the cuvette for 5 min before the reaction
was initiated by the addition of 640 µM substrate. The reaction was
monitored at 385 nm. Initial rate fits to the absorbance data for the
first 100 s were used to determine reaction velocities.
(+)-Pinanediol (1R)-1-(2-Thienylacetylamino)-1-phenylmethyl-
boronate (20). The product was synthesized from 19 (700 mg, 2.73
mmol) following the procedure described for 11. The crude residue
was purified by chromatography (7:3 EtPet/EtOAc), affording 20 as a
white solid (631 mg, 56% overall yield), mp 50-55 °C, [R]D ) +7.1
1
(c 2.4, CHCl3), de > 98%. IR (KBr): 1602 cm-1. H NMR (CDCl3):
δ 0.80 (3H, s, pinanyl CH3), 1.15 (1H, d, J ) 10.5 Hz, pinanyl Hendo),
1.23 (3H, s, pinanyl CH3), 1.35 (3H, s, pinanyl CH3), 1.4-2.5 (5H, m,
pinanyl protons), 3.97 (2H, br s, CH2CONH), 4.06 (2H, br s, CHB),
4.22 (2H, dd, J ) 8.8, 2.0 Hz, pinanyl CHOB), 6.53 (1H, br, NHCO),
7.00 (2H, m, CHCHS-Hp), 7.18 (3H, m Hm-CHS), 7.26 (3H, m
CHCS-Ho). 13C NMR (CDCl3): δ 24.8, 27.0, 27.9, 29.3, 35.2, 36.7,
38.8, 40.5, 47.5 (br, CB), 52.7, 78.0, 85.4, 126.8, 126.9, 127.0, 128.3,
128.8, 129.1, 135.3, 140.9, 174.3. EIMS: m/z 409 (base peak, M+),
394, 376, 340, 311, 284, 257, 230, 173, 117, 97, 91, 69. Anal. Calcd
for C23H28BNO3S: C, 67.48; H, 6.89; N, 3.42; S, 7.83. Found: C, 67.21;
H, 7.01; N, 3.36; S, 7.52.
Crystal Growth and Structure Determination. Cocrystals of
AmpC in complex with compounds 16 and 21 were grown by vapor
diffusion in hanging drops equilibrated over 1.7 M potassium phosphate
buffer (pH 8.7) using microseeding techniques. The initial concentration
of the protein in the drop was 3.8 mg/mL, and the concentrations of
compounds 16 and 21 were 705 µM and 588 µM, respectively. The
compounds were added to the crystallization drops in a 1.2% DMSO,
1 M potassium phosphate buffer (pH 8.7) solution. Crystals appeared
in 5-7 days after equilibration at 23 °C. Before data collection, crystals
were immersed in a cryoprotectant solution of 25% sucrose, 1.7 M
(1R)-1-(2-Thienylacetylamino)-1-phenylmethylboronic Acid (21).
The product was synthesized from 20 (387 mg, 0.95 mmol) following
the protocol described for 15-17. The combined aqueous phases were
concentrated, affording the free boronic acid 21 as an ivory solid (103
mg, 58%.), mp 90-100 °C, [R]D ) -3.1 (c 2.1, CD3OD). IR (KBr):
1
3420, 1628 cm-1. H NMR (CD3OD): δ 3.82 (1H, br s, CHB), 4.16
(2H, br s, CH2CONH), 6.98-7.32 (7H, m, CHCHS-CHCS-Ho-Hm-
Hp), 7.38 (1H, dd, J ) 5.3, 1.0 Hz, CHCHS). 13C NMR (CD3OD): δ
(34) Waley, S. G. Biochem. J. 1982, 205, 631-633.
9
694 J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003

Nanomolar â-Lactamase Inhibitors
A R T I C L E S
potassium phosphate, pH 8.7, for about 30 s, and were flash cooled in
liquid nitrogen. Data were measured on the DND-CAT beam line
(5IDB) of the Advance Photon Source at Argonne National Lab at 100
K using a Mar-CCD detector. Both data sets were measured from single
crystals. Reflections were indexed, integrated, and scaled using the HKL
software package.35 For both structures, the space group was C2, with
two molecules in the asymmetric unit, each containing 358 residues.
The initial phasing model was an AmpC/boronic acid complexed
structure (PDB entry 1FSY), with inhibitor, water molecules, and ions
removed. The model was positioned by rigid body refinement and
refined using the maximum likelihood target in CNS36 including
simulated annealing, positional minimization, and individual B-factor
refinement, with a bulk solvent correction. Sigma A-weighted electron
density maps were calculated using CNS and used in further steps of
manual model rebuilding and placement of water molecules with the
program O.37 The inhibitors were built into the 2|Fo| - |Fc| and |Fo| -
|Fc| electron density maps in each active site of the asymmetric unit.
Subsequent refinement cycles consisted of positional minimization and
B-factor refinement in CNS.
cloacae 12991 ED (Table 4, E. cloacae 1, 2, and 3, respectively),
Pseudomonas aeruginosa 279/88, and P. aeruginosa JMSMA7 (Table
4, P. aeruginosa 1 and 2, respectively). Minimum inhibitor concentra-
tion (MIC) values were determined with Mueller-Hinton Broth II using
the microdilution method according to NCCLS guidelines.38
Data Deposition. The coordinates and structure factors have been
deposited in the Protein Data Bank under accession codes 1MXO and
1MY8 for AmpC in complex with compound 16 and 21, respectively.
Acknowledgment. F.M. and E.C. contributed equally to this
work. Supported by NIH GM63815 (B.K.S., PI). We thank R.
Powers and B. Beadle for preparation of AmpC and are
especially grateful to G. Minasov, I. Trehan, and B. Beadle for
extensive assistance with crystallography. We thank MDL Inc.
(San Leandro, CA) for use of the ACD database. We thank the
Centro Interdipartimentale Grandi Strumenti of Modena for
NMR and mass spectra. We thank R. Powers, I. Trehan, L.
Blasczcak, J. Horn, and T. Roth for reading this manuscript.
Crystallographic data were collected at the DuPont-Northwest-
ern-Dow Collaborative Access Team (DND-CAT) Synchrotron
Research Center at the Advanced Photon Source. DND-CAT
is supported by grants from the NSF, the State of Illinois, and
the U.S. Department of Energy.
Microbiology. Compounds 16 and 21 were tested for synergy with
the â-lactam ceftazidime against pathogenic bacteria from clinical
isolates at the Hospital Ramo´n y Cajal; these bacteria were resistant to
â-lactams because of the expression of a class C â-lactamase. Strains
of bacteria tested were Citrobacter freundii mut756-CAZ, Escherichia
coli 72929 Hip, E. coli 4774 Hip (Table 4, E. coli 1 and 2, respectively),
Enterobacter cloacae 72527 ED, E. cloacae 8411 CAZ-R clon7, E.
JA0288338
(35) Otwinowski, Z.; Minor, W. Methods Enzymol. 1997, 276, 307-326.
(36) Brunger, 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.
(37) Jones, T. A.; Zou, J. Y.; Cowan, S. W.; Kjeldgaard, M. Acta Crystallogr.,
Sect. A 1991, 47, 110-119.
(38) National Committee for Clinical Laboratory Standards, Methods for dilution
antimicrobial susceptibility test for bacteria that grow aerobically. Approved
Standard M7-A4; National Committee for Clinical Laboratory Standards:
Villanova, PA, 1997; Vol. 17.
(39) Evans, S. V. J. Mol. Graphics 1993, 11, 134-138.
(40) Ferrin, T. E.; Huang, C. C.; Jarvis, L. E.; Langridge, R. J. Mol. Graphics
1988, 6, 13-27.
9
J. AM. CHEM. SOC. VOL. 125, NO. 3, 2003 695