Covalent inhibition of serine β-lactamases by novel hydroxamic acid derivatives

Article abstract of DOI:10.1021/bi4003887

The effectiveness of β-lactam antibiotics is greatly limited by the ability of bacteria to produce β-lactamases. These enzymes catalyze the hydrolysis of β-lactams and thus loss of their antibiotic activity. The search for inhibitors of β-lactamases began soon after β-lactams were introduced into medical practice and continues today. Some time ago, we introduced a new class of covalent serine β-lactamase inhibitors, the O-aryloxycarbonyl hydroxamates, that inactivated these enzymes by a unique mechanism in which the active site became cross-linked. We describe in this paper some new variants of this class of inhibitor. First, we investigated compounds in which more polar hydroxamates were incorporated. These were generally not more active than the original compounds against representative class A and class C β-lactamases, but one of them, 1-(benzoyl)-O- (phenoxycarbonyl)-3-hydroxyurea, was significantly more stable in solution, thus revealing a useful platform for further design. Second, we describe a series of O-(arylphosphoryl) hydroxamates that are also irreversible inactivators of class A and class C β-lactamases, by phosphorylation of the enzyme, as revealed by mass spectra. These compounds did not, however, cross-link the enzyme active site. A striking feature of their structure-activity profile was that hydroxamate remained the leaving group on enzyme phosphorylation rather than aryloxide, even though the aryloxide was intrinsically the better leaving group, as indicated by pK_a values and demonstrated by the products of hydrolysis in free solution. Model building suggested that this phenomenon arises from the relative affinity of the enzyme active site components for the two leaving groups. The results obtained for both groups of inhibitors are important for further optimization of these inhibitors.

Full text of DOI:10.1021/bi4003887

Article
pubs.acs.org/biochemistry
Covalent Inhibition of Serine β‑Lactamases by Novel Hydroxamic
Acid Derivatives
Ronak Tilvawala and R. F. Pratt*
Department of Chemistry, Wesleyan University, Lawn Avenue, Middletown, Connecticut 06459, United States
S
* Supporting Information
ABSTRACT: The effectiveness of β-lactam antibiotics is greatly limited by the ability of bacteria to produce
β-lactamases. These enzymes catalyze the hydrolysis of β-lactams and thus loss of their antibiotic activity. The
search for inhibitors of β-lactamases began soon after β-lactams were introduced into medical practice and
continues today. Some time ago, we introduced a new class of covalent serine β-lactamase inhibitors, the
O-aryloxycarbonyl hydroxamates, that inactivated these enzymes by a unique mechanism in which the active
site became cross-linked. We describe in this paper some new variants of this class of inhibitor. First, we
investigated compounds in which more polar hydroxamates were incorporated. These were generally not more
active than the original compounds against representative class A and class C β-lactamases, but one of them, 1-(benzoyl)-O-
(phenoxycarbonyl)-3-hydroxyurea, was significantly more stable in solution, thus revealing a useful platform for further design.
Second, we describe a series of O-(arylphosphoryl) hydroxamates that are also irreversible inactivators of class A and class C β-
lactamases, by phosphorylation of the enzyme, as revealed by mass spectra. These compounds did not, however, cross-link the
enzyme active site. A striking feature of their structure−activity profile was that hydroxamate remained the leaving group on
enzyme phosphorylation rather than aryloxide, even though the aryloxide was intrinsically the better leaving group, as indicated
by pKa values and demonstrated by the products of hydrolysis in free solution. Model building suggested that this phenomenon
arises from the relative affinity of the enzyme active site components for the two leaving groups. The results obtained for both
groups of inhibitors are important for further optimization of these inhibitors.
acterial resistance to β-lactam antibiotics continues to
increase through evolution of β-lactamases.1 These
acyclic substrates 2 but have the new feature of two possible
leaving groups, as indicated in structure 1.
B
enzymes have become able to catalyze hydrolytic destruction of
essentially all β-lactam antibiotics.2 Class C β-lactamases are of
concern in Gram-negative bacteria, particularly enterobacteria.3−5
The efficacy of β-lactams can be restored, however, by
administration of a specific β-lactamase inhibitor together with a
threatened β-lactam. There has been, and continues to be, much
research directed toward new β-lactamase inhibitors.6 Most of the
effective β-lactamase inhibitors known to date, including those that
are currently used in clinical practice, are themselves β-lactams and
act as mechanism-based inhibitors.7 These, too, eventually succumb
to β-lactamase evolution.8
A study of 1 suggested that inhibition of the class C
Enterobacter cloacae P99 β-lactamase was initiated by
nucleophilic attack on the reactive carbonyl group by the
active site serine hydroxyl, followed by loss of the aryloxide
leaving group to yield acyl-enzyme 3 (Scheme 1). This acyl-
Scheme 1
The broad specificity and genetic plasticity of β-lactamases,
necessary to their biological function, dictate the situation in
which potent broad spectrum noncovalent inhibitors are
probably not possible; tellingly, none have been discovered
from exhaustive natural product screens.9 Thus, the only useful
β-lactamase inhibitors are likely to be of the covalent type.
Although covalent inhibitors have had a spotty reputation in
pharmaceutical circles,10 many clinically employed drugs are
enzyme inhibitors of this kind, and there are recent calls for
more attention to be paid to them.11
A number of non-β-lactam, covalent inhibitors of β-lactamases
that, in principle, should not be susceptible to β-lactamases have
been described over the years, including various acylating
agents,12−15 phosphylating agents, sulfonating agents, and boronic
acids, for example.12 Recently, we described a new class of covalent
β-lactamase inhibitors of general structure 1.15,16 These
compounds are acylating agents and direct analogues of the
enzyme was then attacked by the amine of Lys 315, displacing
the second leaving group, the hydroxamate, to yield an inert
cross-linked active site.15
A similar rationale has now led to phosphates 4 as direct
analogues of known phosphonate inhibitors 5.
Received: March 26, 2013
Revised: April 29, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Wiltshire, U.K.). Compounds 25 and 26 were synthesized by R.
Pelto as part of a previous study in this laboratory.16 Elemental
analyses were conducted by Atlantic Microlab, Inc. (Norcross,
GA). Electrospray mass spectra of enzyme complexes were
obtained by the Mass Spectrometry Laboratory, School of
Chemical Sciences, University of Illinois (Urbana, IL).
We have previously shown that electron-withdrawing groups
on the aryloxy ring of 1 lead to more effective inhibitors.16
Structure−activity studies of 1 suggested that these compounds
may react with the β-lactamase active site with the amido side chain
in an orientation different from that of normal substrates.16 To
probe for unusual polar interactions of the side chain, compounds
6−12 were prepared and evaluated as described below.
Synthesis. (O-Phenoxycarbonyl)-1-benzyl-3-hydroxyurea
(6). A solution of imidazole (0.20 g, 3.01 mmol) in DCM (3 mL)
was added dropwise to 1-benzyl-3-hydroxyurea (0.5 g, 3.01 mmol)
(preparation described in the Supporting Information) dissolved
in dimethoxyethane (10 mL). The resulting solution was stirred
at room temperature for 15 min and the temperature then
reduced to −20 °C under a dry nitrogen atmosphere. Phenyl
chloroformate (0.47 g, 3.01 mmol) in DCM (2 mL) was added
dropwise over 1 min and the mixture stirred for approximately
90 min at 0 °C. The imidazole hydrochloride precipitate was
removed by filtration and the filtrate reduced to dryness by rotary
evaporation. The residue was dissolved in ethyl acetate,
precipitating more imidazole hydrochloride, which was removed
by filtration. The solvent was then removed by evaporation and
the residue further dried overnight under an oil pump vacuum.
The crude solid product was recrystallized from a mixture of
benzene and cyclohexane (1/1) in 6.6% yield: mp 108−110 °C;
1H NMR (d6-DMSO) δ 4.25 (d, J = 7.20 Hz, 2H), 7.22 (m, 8H),
7.45 (t, J = 7.1 Hz, 2H), 7.59 (t, J = 6.9 Hz, 1H), 10.5 (s, 1H);
FTIR (KBr) 1813 cm−1; HRMS (ES+) 287.1031 (M + H+), calcd
for C15H15N2O4 287.1032.
N-(Benzyloxycarbonyl)-O-(phenylphosphoryl)-
hydroxylamine (13). Triethylamine (0.48 g, 4.76 mmol) in dry
DCM (3 mL) was added very slowly dropwise to a precooled
solution of phenyl dichlorophosphate (1.0 g, 4.76 mmol) and
benzyl alcohol (0.51 g, 4.76 mmol) in dry DCM (12 mL) over
a period of 1 h at −20 °C under nitrogen. The reaction mixture
was then stirred for an additional 1 h at that temperature.
Benzyl N-hydroxycarbamate (0.79 g, 4.76 mmol) in dry DCM
(12 mL) was added, followed by dropwise addition of
triethylamine (0.48 g, 4.76 mmol) in dry DCM (3 mL). This
reaction mixture was stirred for an additional 2 h until it reached
room temperature. The solvent was evaporated to dryness, the
residual oil dissolved in ethyl acetate, and the solution washed
several times with 5% aqueous HCl. The organic layer was
separated, dried over Na2SO4, and evaporated to dryness to yield
the benzyl ester of the required product as a colorless oil. This
material was dried under an oil pump vacuum overnight. Then,
without purification, the ester was subjected to debenzylation as
described by Saady et al.17 Thus, DABCO (0.05 g, 0.48 mmol)
was added to a stirred solution of the ester (0.20 g, 0.48 mmol)
in anhydrous toluene (2 mL), under an argon atmosphere. The
reaction mixture was heated under reflux for 2 h before the
solvent was evaporated and the residue dissolved in aqueous
HCl (5%, 10 mL). This solution was extracted with ethyl acetate
and the organic layer dried over Na2SO4. The solvent was then
evaporated under reduced pressure to yield 13. The crude
product was purified by recrystallization from an ethyl acetate/
hexane mixture (1/1) in 60% yield: mp 108−110 °C; 1H NMR
(d6-DMSO) δ 5.12 (s, 2H), 7.17 (m, 3H), 7.36 (m, 7H), 10.93
(s, 1H); 31P NMR (d6-DMSO) δ −4.14; HRMS (ES+) 324.0645
(M + H+), calcd for C14H15NO6P 324.0637.
The dual leaving group potential of phosphyl inhibitors 4 was
explored by means of compounds 13−24.
Kinetics Methods. Absorption spectra and spectrophoto-
metric reaction rates were measured with Hewlett-Packard
8452A and 8452E spectrophotometers. Enzyme concentrations
were determined spectrophotometrically. Kinetics experiments were
conducted at 25 °C in a 20 mM 3-morpholinopropanesulfonic acid
EXPERIMENTAL PROCEDURES
■
The class C P99 β-lactamase from E. cloacae P99 and the TEM-
2 β-lactamase from Escherichia coli were purchased from the
Centre for Applied Microbiology and Research (Porton Down,
B
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
(MOPS) buffer solution at pH 7.5. Stock solutions of the inhibitors
were prepared in DMF and diluted to 1−5% DMF in reaction
mixtures.
Spontaneous Hydrolysis. Spontaneous hydrolysis rates of
6−12 were measured spectrophotometrically in replicate at
appropriate concentrations and wavelengths (Table S1 of the
Supporting Information). The full progress curves were then fit
to a first-order rate equation and the rate constants thus
obtained averaged.
P99 β-Lactamase Inactivation. The P99 β-lactamase (0.5 μM)
was incubated with inhibitor (10−500 μM) in 100 μL of buffer
containing 0.1% BSA. At appropriate times, aliquots (5 μL) of the
reaction mixture were diluted to 1 mL in MOPS buffer containing
cephalothin (300 μM), and hydrolysis of the latter was monitored
spectrophotometrically at 280 nm over a period of time. The mea-
sured initial rates, proportional to the remaining free enzyme con-
centration, were plotted versus incubation time. These data were
fit to Scheme 2 with Dynafit.18
Molecular Modeling. The crystal structure of the P99 β-
lactamase (Protein Data Bank entry 1XX2), including crystallo-
graphic water molecules, was used as the starting point for
computations using Discovery studio version 2.5 (Accelrys).
The CHARMm force field was employed for all computations.
The partial charges of the enzyme were assigned by Discovery
studio. Structures and partial charges of the PO5 nucleus of the
pentacoordinated phosphorus species were derived from the
results of high-level calculations.19 The positions of the
phosphorus atom and the oxygen atoms bound to it were fixed
because molecular mechanics parameters for this type of structure
were not available. The phosphorane dianion structure (with
neutral Tyr 150) was chosen because it represents a close analogue
of the intermediate immediately formed on general base (Tyr 150
anion)-catalyzed attack of the serine hydroxyl group on a phosphate
monoanion. The pH was set to 7.5, and the total charge on the
complex was −2.0. The side chains of Lys 67 and Lys 315 were
cationic. The enzyme−substrate complex was hydrated with a 20 Å
sphere of water molecules centered at the active site serine. The
hydrated structure was energy-minimized using the steepest descent
algorithm (250 steps) followed by the conjugate gradient method
(750 steps) for 1000 steps. The energy-minimized structures were
heated for 10000 steps (10 ps) to 300 K. The output of the heating
run was then subjected to a 1 ns production run. Several of the 100
saved structures were selected from the production output and
energy-minimized using the protocol mentioned above.
Scheme 2
With compounds such as 8 in which the spontaneous
reaction was too rapid for the procedure described above to be
employed, various concentrations of the inhibitor were
incubated with the enzyme until the reaction was complete.
The reaction mixture was then assayed for β-lactamase activity
against cephalothin (300 μM). These data, initial rates versus
concentration, were also fit to Scheme 2 with Dynafit.18
Active Site Titration with 13. The P99 β-lactamase
(1.0 μM) was incubated with a series of concentrations (0.2,
0.4, 0.6, 0.8, 1.0, and 1.2 μM) of 13 in buffer containing
0.1% BSA. After 24 h, the reaction mixtures were assayed for
β-lactamase activity against 300 μM cephalothin and the initial
rates thus determined plotted against inhibitor concentration to
obtain the partition ratio (k3/k2) from the abscissa intercept.
The data were fit in Kaleidograph.
Mass Spectra of EI Complexes. The P99 β-lactamase (20 μM,
100 μL) was incubated with 6 (1.0 mM; 5 μL of a 20 mM
stock solution in DMF was added) for 15 min. A small aliquot
was taken and assayed for enzyme activity to confirm inactivation.
The reaction was then quenched with 50% aqueous trichloroacetic
acid (final concentration of 5%). The reaction mixture was allowed
to stand on ice for 30 min and centrifuged at 2000 rpm for 8 min
and the supernatant decanted. The precipitate was washed with 5%
trichloroacetic acid (2 × 50 μL) followed by acetone (2 × 50 μL).
The solid residue was dried under an oil pump vacuum and
subjected to electrospray mass spectrometry. A control sample
without inhibitor was prepared in the same way.
RESULTS AND DISCUSSION
■
Acyl hydroxamates 6−12 were prepared by acylation of the
corresponding hydroxamic acids, as previously described.16
Phosph(on)ate diesters 13−23 were prepared by reaction of
appropriate phosphyl chlorides with the required alcohols and
hydroxamic acids in the presence of base. Where appropriate,
the benzyl group was used to protect one phosphyl oxygen and
subsequently removed by DABCO aminolysis.17 Phosphate
monoester 24 was prepared by condensation of phosphoric acid
with benzyl N-hydroxycarbamate.20 Details of the synthesis of 6
and 13 are presented in Experimental Procedures and those of the
other compounds in the Supporting Information.
Acyl derivatives 6−12 hydrolyzed spontaneously in aqueous
buffer. Observed first-order rate constants for this reaction of
6−12, under the conditions employed at pH 7.5, are listed in
Table 1. For comparison, data for parent compound 25 and
methoxycarbonyl analogue 26 are included. Immediately
noticeable from Table 1 is the high spontaneous reactivity of
many of the new compounds, most of them more reactive than
25 itself. The differences in reactivity among 25, 6, 7, and 8,
and between 9 and 26, probably reflect the relative effectiveness
of the hydroxamate carbonyl as an intramolecular nucleophile
in catalyzing the breakdown of these compounds (Scheme 3).
The P99 β-lactamase (20 μM) was incubated with phosphates
13 (500 μM) and 19 (1.2 mM) (total volumes of 100 μL) for 15
and 30 min, respectively. Small aliquots were taken and assayed for
enzyme activity to confirm inactivation. The reaction mixtures
were then prepared for mass spectroscopy as described above.
NMR Study of the Hydrolysis of 13. A solution of 13 in
2H2O containing 20 mM NaHCO3 at p2H 7.2 was heated at
70 °C for 24 h. 1H NMR spectra (300 MHz) were recorded at
suitable intervals as the hydrolysis reaction proceeded.
Nucleophilic amido-carbonyl participation in systems analo-
gous to these is well-documented.21,22 The carbonyl of 1, as a
carbamate, is a poor nucleophile23 but would be enhanced by
the replacement of the carbamate oxygen with nitrogen as in 6
and 8 (and in 9 vs 26). Acylation of this nitrogen, as in 7,
would weaken the nucleophilicity of the hydroxamate carbonyl,
leading to the slow, spontaneous reaction of this compound.
C
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Table 1. Rate Constants for Inhibition of the P99
β-Lactamase by Acyl Hydroxamates
a
b
inhibitor
k0 (s−1
2.5 × 10−4
)
k1 (M−1 s−1
)
k3/k2
cd
,
25
6
6100 200
1640 50
2.0 0.1
f
0.0191 0.0001
ND
e
e
e
e
[(2.41 0.01) × 10−4
(8.35 0.04) × 10−5
[(7.27 0.13) × 10−5
0.0643 0.0005
]
]
]
[400 9]
7.7
c
7
320 160
580 15
21
e
[800 69]
7480 250
[1700 350]
0.87 0.01
0.19 0.01
f
8
ND
e
[(2.54 0.02) × 10−4
(2.28 0.01) × 10−3
very slow
(3.66 0.01) × 10−3
(7.2 0.1) × 10−3
(2.93 0.03) × 10−5
2.05
f
9
ND
d
f
26
10
11
12
ND
f
201
4320 420
64
1
ND
f
ND
f
7
ND
a
b
Spontaneous hydrolysis rate constant. k3/k2 is the turnover/
c
inactivation ratio (Scheme 1). Inhibition data fit to Scheme 1.
d
e
f
From ref 16. pH 5.0. Not determined. Small values of k3/k2 (≤20)
for these compounds could not be precisely determined because of the
high spontaneous hydrolysis rates and/or the low rates of inhibition.
The combination of a poor intramolecular nucleophile and a
poor leaving group in 26 greatly impedes the spontaneous
breakdown of 26.24
The lability of both 6 and 8 is significantly lower at pH 5
(Table 1). This suggests that the reactive entity in Scheme 3
may actually be anion 27. Compound 25 has an acidic NH with
a pKa of 6.8,16 and the current compounds would be expected
to be comparably acidic, with some variation dictated by the
details of the structure. Compounds 10 and 11 hydrolyze
rapidly in solution, probably from direct water and hydroxide
attack; the hydroxamate leaving group would be much better in
these cases (pKa values of 6.0 and 7.1, respectively25) because
of the two activating acyl groups on the hydroxamate nitrogen.
Figure 1. (A) Activity of the P99 β-lactamase (0.15 μM) as a function
of time in the presence of 6 (10.0 μM). (B) Activity of the P99
β-lactamase (0.15 μM) after complete reaction with 8 at various
concentrations.
Compounds 6−12, in general, were time-dependent,
irreversible inactivators of the class C E. cloacae P99 β-lactamase
(see Figure 1, for example) and of the class A TEM-2 enzyme.
Second-order rate constants of inactivation of the P99 enzyme
were obtained as described in Experimental Procedures, and the
results are listed in Table 1. It is unlikely that the (transiently
stable) dioxazolinone product from the reaction of 27 is a
significant inhibitor.26 The results show that, at pH 7.5, only
compound 8 is (slightly) more effective than parent compound
25; 6 and 7 are markedly less effective inactivators. The same
positive effect of hydrazine incorporation is seen upon
comparison of methyl esters 9 and 26. This may be due
to the more effective hydrogen bond acceptor (hydrazide
carbonyl) present in 8 and 9. The demethylated analogue
of 8, compound 28, may also be an effective inhibitor, but
attempts to synthesize it failed, probably because of its instability.
The value of an aromatic ring on the hydroxamic acid moiety
is apparent in the comparison of 10 and 11. The imide moiety
of 10 and 11 does not, however, provide them any advantage
as inhibitors over the original amide of 25.
It is also interesting to note at this stage the activity of
N,N-diacyl-hydroxylamine 12, in comparison with that of the
N,O-diacyl isomer 25. As recorded in Table 1, 12 inhibited the
P99 enzyme less effectively than did 25. This order of activity
is in accord with the order of spontaneous hydrolytic lability
(Table 1) and the carbonyl stretch frequencies in the infrared
spectra of these molecules (the highest-frequency carbonyl is
1764 cm−1 for 12 and 1808 cm−1 for 2516).
As described in the introductory section, inactivation of the
P99 β-lactamase by 25 is accompanied by irreversible cross-linking
Scheme 3
D
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
of the active site by insertion of an inhibitor-derived carbonyl
moiety between the active site Ser 67 and Lys 315. It is likely that
the inhibitors described in this paper also follow this mechanism. A
mass spectrum of the inactive complex of 6 with the enzyme
showed, as with 25, addition of a small fragment with a mass of
25 5, consistent with Scheme 1.
The effectiveness of 25 as an inhibitor at pH 7.5 is limited by
its existence as an unreactive anion at that pH.16 It is more
effective in the pH range of 5−7 where it gains a proton to
form the reactive neutral species (Scheme 1). Values of k1 for
6−8 at pH 5 are also listed in Table 1. It can be seen that 7 also
becomes more effective at pH 5, but 6 and 8 become less
effective. These changes probably reflect the resultant of the
effects of the NH pKa’s of 6 and 8, which are likely to be higher
than that of 25, and the pKa of the enzyme active site whose
protonation reduces its reactivity.16
The hydrolysis of 13 at 70 °C in 20 mM NaHCO3 was
monitored for 4 days by 1H NMR, which showed approximately
two-thirds of it hydrolyzed to phenol and 24 at that time. It is
interesting that the first leaving group appeared to be phenoxide
(phenol, pKa of 10.027) rather than the hydroxamate (benzyl
N-hydroxycarbamate, pKa of 9.228) (see further discussion below).
The better 4-nitrophenoxide leaving group of 16 also left first in
the spontaneous hydrolysis of this compound.
Phosphylates 13−22 are stoichiometric 1/1 irreversible
inactivators of the P99 β-lactamase (see Figure 2, for example).
The practical effectiveness of the new inhibitors is greatly
restricted by their instability in solution [k0 (Table 1)], where
the exception is 7, which is much more stable, as discussed above.
Although N,N-diacyl compound 12 is, like 7, more stable, it is also
a poorer inhibitor than its N,O-diacyl analogue 25 (Table 1). As
also described above, pH is important, because the stability of 6
and 8 does increase markedly at lower pH values.
The other parameter that is important in defining the
inhibitory potency of these compounds is the partition ratio
k3/k2 (Scheme 1), favoring hydrolysis of the intermediate acyl-
enzyme versus inactivation. This was small, ∼2, for 25. It is also
small (≤20) for 6 and 8−12, although it is difficult to accurately
determine because of the large background hydrolysis rates at
pH 7.5. The interesting exception is 7, for which the turnover
rate greatly exceeds the inactivation rate. As discussed
previously,16 partitioning of the acyl-enzyme is probably
determined by its relative occupancy of two conformations of
the inhibitor in the active site (Scheme 4), one (29a) in which
the acyl-enzyme carbonyl is held in the oxyanion hole in the
manner of a normal substrate, leading to hydrolysis. In the
other (29b), the carbonyl is hydrogen-bonded to the Tyr 150
hydroxyl; this conformation leads to attack by Lys 315 and
inactivation by cross-linking. It seems that the acyl-enzyme
from compound 7, unlike the others, and possibly due to an
interaction of the benzoyl carbonyl oxygen with a component
of the active site, e.g., Asn 152 or Gln 120, prefers conformation
29a and hydrolysis. More structural information is needed to
design inhibitors with a preference for 29b, but 7 would be a
useful place to begin.
Scheme 4
Figure 2. (A) Activity of the P99 β-lactamase (0.5 μM) as a function of
time in the presence of 13 (10.0 μM). (B) Activity of the P99 β-lactamase
(1.0 μM) after complete reaction with 13 at various concentrations.
Second-order inactivation constants are listed in Table 2. The
k1 value of 49 M−1 s−1 for 13 is smaller than the value of
200 M−1 s−1 for the close phosphonate analogue 30 (X = H),
a difference that is typical for nucleophilic attack at the
phosphorus of a phosphate versus a phosphonate.29,30 Although
the values of k1 listed in Table 2 generally appear to be low, it
has been shown that structural optimization of 30 can lead to
k1 values of at least 105 M−1 s−1 for both class A and class C
β-lactamases.31 One important improvement in 30 lay in
adding a better aryloxide leaving group. For example, the k1
value for 30 (X = NO2) is 5.6 × 104 M−1 s−1.32
Compound 6 was also, like 25, an inactivator of the class A
TEM-2 β-lactamase, with a k1 value of 240 15 M−1 s−1, cf.
13.4 M−1 s−1 for 25.16 The significantly higher value for 6 may
signal the formation of an additional hydrogen bond between 6,
containing an additional NH hydrogen bond donor in the side
chain, and the polar active site of the TEM-2 enzyme, which
contains, for example, Glu 166 and Asn 171.
In contrast to the acyl derivatives, the phosphates are stable to
spontaneous hydrolysis at neutral pH and room temperature.
E
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
been the first to leave from the phosphoryl reaction center.
This must reflect the differences in the transition state
geometry and electronic structure between the acylation and
phosphylation transition states. It is possible, however, to
achieve a hydroxamoyl phosphoryl enzyme 32 by means of
symmetrical phosphate 19. This compound was, however, a
weaker inhibitor than 13, implying that neither hydroxamate
moiety interacts strongly with the enzyme. The enzyme
inactivated by 19 yielded a mass spectrum (Figure S1B of the
Supporting Information) with an increase of 234 amu versus
the free enzyme mass. This correlates well with addition of
(PhCH2OCONHOPO2 minus H) (230 amu) to the enzyme.
The reactivity of phosphyl hydroxamates with β-lactamases
does, however, seem to be a predictable function of the
phosphyl group. As would be expected,28,29,35 phosphonate 20
is a more effective inhibitor than phosphate 18. The ability of
hydroxamate as a leaving group is also seen in the results for 20,
which is a considerably better inhibitor of the P99 enzyme than
4-nitrophenyl phenylphosphonate (k1 = 2.9 M−1 s−1).36
Additional hydrophobicity in the hydroxamate moiety, as in
21, produces a decrease in inhibitory activity as compared with
that of 13, while such an addition to the aryloxy moiety, as in
22, leads to a small increase. This difference in response
presumably reflects the nature of the respective binding sites
(see below). The inactivity of 23, like that of the analogous acyl
derivative, is probably due to the dissociation of the NH to
form an unreactive dianion.16 Finally, the inactivity of 24 is
likely due to the poor electrophilicity of phosphorus in this
molecule, even as a monoanion, and the absence of general acid
catalysis32,37 to facilitate a dissociative mechanism involving
quasi-free metaphosphate.
We employed molecular modeling to understand the results
described above with the phosphates in terms of active site
structure. We began with the premise that the mechanism of
the reaction between the β-lactamase and a phosphate diester
involved bimolecular displacement at phosphorus with a
trigonal bipyramidal pentacoordinated phosphorus intermedi-
ate.33 In the reactive complex, one would expect the incoming
serine oxygen nucleophile to take one apical position and a
leaving group, either hydroxamate or phenoxide, to take the
other. We also considered the possibility of pseudorotation of
the first-formed pentacoodinated intermediate, but this did not
seem to lead to more strongly interactive structures. Although
pseudorotation at an enzyme active site is unlikely and, to the
best of our knowledge, not unambiguously proven to occur for
any enzyme,38,39 it may be more likely to arise when unnatural
active site ligands are investigated or promiscuous or moonlighting
enzymes.40 The participation of a sixth phosphorus ligand is also a
possibility but not one yet demonstrated in any particular case.40
The experimental evidence (see above) suggests that the
hydroxamate must prefer the second apical position even when
the phenoxide is apparently a better leaving group, such as the
4-nitrophenoxide in 16 (4-nitrophenol, pKa of 7.1)27 and the
pentafluorophenoxide in 17 (pentafluorophenol, pKa of 5.5).41
A combination of model building, molecular dynamics, and
molecular mechanics energy minimizations led to the structures
of Figure 3 as best representing the likely pentacoordinated
intermediates.
Table 2. Rate Constants for Inhibition of the P99 β-
Lactamase by Phosphylates
phosphate
k1 (M−1 s−1
)
phosphate
k1 (M−1 s−1
)
13
14
15
16
17
18
49
31
3
1
19
20
21
22
23
24
6.7 0.2
1100
6
3.1 0.1
12.0 1.4
0.89 0.17
3.5 0.4
93
NI
NI
1
a
a
130
1
a
No inhibition observed under the conditions employed.
Consequently, we investigated the effect of changing the
aryloxide leaving group of 13 (although the hydroxamate is also
a potential leaving group). In 14 and 15, the phenoxide of 13 is
replaced with poorer alkoxide leaving groups, and in 16 and 17,
it is replaced with much better leaving groups. Counter-
intuitively, the alkoxy leaving groups produced only a small
decrease in rate, and even more surprisingly, the p-nitro and
pentafluoro substitutions of the phenoxide led to decreases in
rate (Table 2). These observations are incompatible with rate-
determining nucleophilic displacement of phenoxide in 13 by
the active site serine hydroxyl. The observations are, however,
in accord with nucleophilic attack on phosphorus with
displacement of the hydroxamate to yield phosphoryl-enzyme
31. A concerted associative mechanism would be expected for
nucleophilic displacement at a phosphate diester monoanion,33
which, in this case, as indicated by the substituent effects, would
involve a transition state with more bond breaking (to
hydroxamate) than formation (to Ser OH). Alternatively, general
acid catalysis for departure of the hydroxamate could be
indicated, but this seems unlikely for such a good leaving
group and is thought to be unavailable for phosphyl transfer
reactions at the P99 β-lactamase active site.32 Rate-significant
bond breaking to hydroxamate is also suggested by the activity of
18 being higher than that of 13 (the anion of PhOCONHOH,
pKa of 8.4,28 should be a better leaving group than that of
PhCH2OCONHOH, pKa of 9.228).
Compounds 13−24 had low activity against the class A TEM-2
or the class D OXA-1 β-lactamases, which resembles the situation
with 30 and these enzymes.34 Nonetheless, as with 30,31 structural
optimization might well lead to useful inhibitors.
A mass spectrum (Figure S1A of the Supporting
Information) of the inert complex derived from the reaction
of the P99 β-lactamase with 13 showed a mass enhancement of
155 amu compared with that of the original enzyme. This
corresponds well to addition of (PhPO2 minus H) (155 amu),
as in 31, and thus loss of the hydroxamate from phosphorus.
Notably, the phenoxide group was also not lost, and thus, cross-
linking did not occur.
In Figure 3A, phenoxide is in the (apical) leaving group
position and the hydroxamate in the equatorial. Neither group
seems to enjoy any interaction with the enzyme that would confer
specificity. In Figure 3B, the hydroxamate is certainly in a good
position to leave. It has no direct interaction with the enzyme, and
Thus, contrary to observations in the carbonyl system and to
observations in free solution (see above), the hydroxamate has
F
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
Figure 3. Stereoviews of energy-minimized pentacoordinated phosphorus intermediate structures formed on reaction of 13 with the P99 β-
lactamase. Only heavy atoms are shown. (A) Phenoxide in the (apical) leaving group position. (B) Hydroxamate in the (apical) leaving group
position.
thus, its rate of departure is most directly influenced by its intrinsic
leaving group ability, as expressed to a first approximation by its
conjugate acid pKa. The presence of additional hydrophobic
character may therefore not greatly affect its departure as seen in
the results with 21. It is interesting to note here that the
phenylacetamido moiety of 30 does appear in models42 to interact
more strongly, by hydrogen bonding, with the enzyme (occupying
the classical side chain amide binding site comprising the side
chain NH2 group of Asn 152 and the backbone carbonyl oxygen
of Ser 318), than the carbamate group of 13. In the case of 30, the
aryloxy leaving group, in an apical position, appears to interact
with the aromatic ring of Tyr 150, in a manner similar to that of
the aryloxide of 13 in Figure 3B (see below).
The equatorial disposition of the other potential leaving
group, aryloxide or alkoxide, however, is also favorable to the
structure of Figure 3B (cf. Figure 3A). The phenoxide ring of
13 is able to interact favorably with both Tyr 150 (hydro-
phobic/quadrupole) and Asn 152 (dipole/quadrupole). The
quadrupolar interactions would be enhanced if the electron
density on the phenoxide ring were decreased, as in 16 and 17.
It is therefore possible that the aryloxy group would prefer to
remain in the equatorial (nonleaving) position in the complexes
derived from 16 and 17, as well as from 13, rather than take up
the apical (leaving group) position. The ability of enzyme active
sites to select the leaving group of a phosphoryl transfer
reaction to yield a kinetic rather than a thermodynamic product
has been previously observed in a substrate analogue43,44 but
not, to the best of our knowledge, in an inhibitor. It should be
noted that an extended hydrophobic aryloxide might reorient to
allow hydrophobic interaction with Tyr 221 rather than Tyr 150,
as seen in a variety of boronic acid and noncovalent inhibitors.45,46
The aryl group of arylphosphonyl inhibitors,36 also in an equatorial
position, appears to interact with Tyr 150 in a manner similar to
that of the aryloxy leaving group of 13 shown in Figure 3B. The
poorer alkoxide phosphorus ligands of 14 and 15 would naturally
take up equatorial positions as the hydroxamate prepared to depart
from the apical position. The structure of Figure 3B, therefore,
seems to better accommodate the experimental results.
G
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
(3) Georgopapadakou, N. H. (2008) Antibiotic resistance in
enterobacteria. In Bacterial resistance to antimicrobials (Wax, R. G.,
Ed.) 2nd ed., pp 343−362, CRC Press, Boca Raton, FL.
CONCLUSIONS
■
O-Aryloxycarbonyl hydroxamates, 1, inactivate serine β-lactamases
by cross-linking the active site (Scheme 1).16 The new derivatives,
6−12, noting particularly those containing side chains elaborated
with polar hydrogen bonding functionality (6−9), appear to inhibit
by the same mechanism used by 1 and generally have similar
activity as inhibitors. The new compounds, however, are generally
less stable in aqueous solution than 25, with the interesting
exception of hydroxyurea derivative 7. Further structure−activity
studies based on 7 may lead to a useful new class of inhibitors.
Phosphyl analogues 13−24 also represent a new class of class
C β-lactamase inhibitors, with the advantage of being quite
stable in solution. These compounds phosphylate the active site
but do not cross-link it. The competition between hydroxamate
and aryloxide as the leaving group from phosphorus was found to
favor hydroxamate, which does not appear to significantly interact
with active site residues. Aryloxide does appear to interact with the
active site through hydrophobic and quadrupolar interactions but
only when placed in an equatorial (nonleaving group) position of
trigonal bipyramidal intermediates. Counterintuitively then, and in
contrast to the situation with acylation inhibitors 1, addition of
electron-withdrawing groups to the aryloxide produced poorer rather
than better inhibitors (16 and 17 vs 13). Better inhibitors should
therefore be sought by optimizing the hydroxamate leaving group
and by enhancing the interactions of the aryloxide with the active site
in pentacoordinated phosphorus intermediates (Figure 3B).
(4) Helfand, M. S. (2008) β-Lactams against emerging superbugs:
Progress and pitfalls. Expert Rev. Clin. Pharmacol. 1, 559−571.
(5) Wouters, J., and Bauvois, C. (2012) Structure of class C β-
lactamases and perspectives in drug design. In β-Lactamases (Frer
̀
e, J.
M., Ed.) Chapter 4, Nova Science, New York.
(6) Drawz, S. M., and Bonomo, R. A. (2010) Three decades of β-
lactamase inhibitors. Clin. Microbiol. Rev. 23, 160−201.
(7) Pratt, R. F. (1992) β-Lactamase inhibition. In The Chemistry of β-
Lactams (Page, M. I., Ed.) Chapter 7, Chapman and Hall, London.
(8) Papp-Wallace, K. M., Bethel, C. R., Distler, A. M., Kasuboski, C.,
Taracila, M., and Bonomo, R. A. (2010) Inhibitor resistance in the
KPC-2 β-lactamase, a preeminent property of this β-lactamase.
Antimicrob. Agents Chemother. 54, 890−897.
(9) Omura, S. (2011) Microbial metabolites: 45 years of wandering,
wondering and discovering. Tetrahedron 67, 6420−6459.
(10) Johnson, D. S., Weerapana, E., and Cravatt, B. F. (2010)
Strategies for discovering and derisking covalent, irreversible enzyme
inhibitors. Future Med. Chem. 2, 949−964.
(11) Singh, J., Petter, R. C., Baillie, T. A., and Whitty, A. (2011) The
resurgence of covalent drugs. Nat. Rev. Drug Discovery 10, 307−317.
(12) Pratt, R. F. (2012) β-Lactamase inhibitors: Non-β-lactams. In β-
Lactamases (Frer
(13) Buynak, J. D. (2012) β-Lactams as inhibitors of β-lactamases. In
β-Lactamases (Frere, J.-M., Ed.) Chapter 9, Nova Science, New York.
̀
e, J.-M., Ed.) Chapter 10, Nova Science, New York.
̀
(14) Kluge, A. F., and Petter, R. C. (2010) Acylating drugs:
Redesigning natural covalent inhibitors. Curr. Opin. Chem. Biol. 14,
421−427.
ASSOCIATED CONTENT
* Supporting Information
Mass spectra of complexes of the P99 β-lactamase with 13 and
19, a table of analytical wavelengths, and synthetic details for
the preparation of compounds 6−12 and 14−24. This material
is available free of charge via the Internet at http://pubs.acs.org.
(15) Wyrembak, P. N., Babaoglu, K., Pelto, R. B., Shoichet, B. K., and
Pratt, R. F. (2007) O-Aryloxycarbonyl hydroxamates: New β-
lactamase inhibitors that cross-link the active site. J. Am. Chem. Soc.
129, 9548−9549.
■
S
(16) Pelto, R. B., and Pratt, R. F. (2008) Kinetics and mechanisms of
inhibition of a serine β-lactamase by O-aryloxycarbonyl hydroxamates.
Biochemistry 47, 12037−12046.
(17) Saady, M., Lebeau, L., and Miskowski, C. (1995) Selective
monodeprotection of phosphate, phosphite, phosphonate and
phosphoramide benzyl esters. J. Org. Chem. 60, 2946−2947.
(18) Kuzmic, P. (1996) Program DYNAFIT for the analysis of
enzyme kinetic data: Application to HIV proteinase. Anal. Biochem.
237, 260−273.
AUTHOR INFORMATION
Corresponding Author
*E-mail: rpratt@wesleyan.edu. Telephone: (860) 685-2629.
Fax: (860) 685-2211.
Funding
This research was supported by National Institutes of Health
Grant AI-17986 (R.F.P.) and by Wesleyan University.
Notes
■
́
(19) Borden, J., Crans, D. C., and Florian, J. (2006) Transition state
analogues for nucleotidyl transfer reactions: Structure and stability of
pentavalent vanadate and phosphate ester dianions. J. Phys. Chem. B
110, 14988−14999.
(20) Dueymes, C., Pirat, C., and Pascal, R. (2008) Facile synthesis of
simple mono-alkyl phosphates from phosphoric acid and alcohols.
Tetrahedron Lett. 49, 5300−5301.
(21) Cabaret, D., Garcia Gonzalez, M., Wakselman, M., Adediran, S.
A., and Pratt, R. F. (2001) Synthesis, hydrolysis, and evaluation of 3-
acylamino-3,4-dihydro-2-oxo-2H-1,3-benzoxazinecarboxylic acids and
linear azadepsipeptides as potential substrates/inhibitors of β-lactam-
recognizing enzymes. Eur. J. Org. Chem., 141−149.
(22) De Jersey, J., Willadsen, P., and Zerner, B. (1969) Oxazolinone
intermediates in the hydrolysis of activated N-acylamino acid esters.
The relevance of oxazolinones to the mechanism of action of serine
proteases. Biochemistry 8, 1959−1967.
(23) Rahil, J., and Pratt, R. F. (1991) Intramolecular participation of
the amide group in acid- and base-catalysed phosphonate monoester
hydrolysis. J. Chem. Soc., Perkin Trans. 2, 947−950.
(24) Farrar, C. R., and Williams, A. (1980) The mechanism of
hydrolysis and ethylaminolysis of N-benzoylglycine and N-benzoyl-
sarcosine esters. J. Chem. Soc., Perkin Trans. 2, 181−187.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Ms. Chuquio Dong, who prepared compounds 11 and
12, and Ms. Tsagan Ednyasheva, who conducted preliminary
experiments on the synthesis of the phosphoryl compounds.
ABBREVIATIONS
■
BSA, bovine serum albumin; DABCO, 1,4-diazabicyclo[2.2.2]-
octane; DCM, dichloromethane; DMF, dimethylformamide;
DMSO, dimethyl sulfoxide; ESMS, electrospray mass spectrometry;
FTIR, Fourier transform infrared; MOPS, 3-(N-morpholino)-
propanesulfonic acid; NMR, nuclear magnetic resonance.
REFERENCES
■
(1) Sykes, R. (2010) The evaluation of microbial resistance: A
Darwinian perspective. J. Antimicrob. Chemother. 65, 1842−1852.
(2) Oelschlaeger, P., DuPrez, K. J., Welsh, W. J., and Toney, J. H.
(2010) Evolving carbapenemases: Can medicinal chemists advance
one step ahead of the coming storm? J. Med. Chem. 53, 3013−3027.
(25) Pfeiffer, M. J., and Hanna, S. B. (1993) Aminolysis of activated
esters of indole-3-acetic acid in acetonitrile. J. Org. Chem. 58, 735−740.
(26) Adediran, S. A., Lin, G., Pelto, R. B., and Pratt, R. F. (2012)
Crossover inhibition as an indicator of convergent evolution of
H
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX

Biochemistry
Article
enzyme mechanisms: A β-lactamase and a N-terminal nucleophile
hydrolase. FEBS Lett. 586, 4186−4189.
(27) Jencks, W. P., and Regenstein, J. (1975) in Handbook of
Biochemistry and Molecular Biology (Fasman, G. D., Ed.) 3rd ed.,
Physical and Chemical Data, Vol. 1, pp 305−351, CRC Press,
Cleveland, OH.
(28) Calculated using Advanced Chemistry Development (ACD/
Labs) version 11.02 (1994−2013).
(29) Hudson, R. F., and Keay, L. (1960) The mechanism of
hydrolysis of phosphonochloridates and related compounds. I. The
effect of substituents. J. Chem. Soc., 1859−1864.
(30) Hays, H. R. (1971) Reactions of phosphorus esters with
phenylmagnesium bromide. J. Org. Chem. 36, 98−101.
(31) Di Ninno, F., Hammond, M. L., Dykstra, K., Kim, S., Tan, Q.,
Young, K., Hermes, J. D., Chen, H., Raeppel, S., Mannion, M., et al.
(2008) (Hetero)arylsulfonamidomethylphosphonate derivatives as β-
lactamase inhibitors and their preparation, pharmaceutical composi-
tions and use in the treatment of bacterial infection. PCT Int. Appl.
WO 2008073142.
(32) Rahil, J., and Pratt, R. F. (1992) Mechanism of inhibition of the
class C β-lactamase of Enterobacter cloacae P99 by phosphonate
monoesters. Biochemistry 31, 5869−5878.
(33) Cleland, W. W., and Hengge, A. C. (2006) Enzymatic
mechanisms of phosphate and sulfate transfer. Chem. Rev. 106,
3252−3278.
(34) Rahil, J., and Pratt, R. F. (1991) Phosphonate monoester
inhibitors of class A β-lactamases. Biochem. J. 275, 793−795.
(35) Rahil, J., and Pratt, R. F. (1993) Structure-activity relationships
in the inhibition of serine β-lactamases by phosphonic acid derivatives.
Biochem. J. 296, 389−393.
(36) Nagarajan, R., and Pratt, R. F. (2004) Thermodynamic
evaluation of a covalently bonded transition state analogue inhibitor:
Inhibition of β-lactamases by phosphonates. Biochemistry 43, 9664−
9673.
(37) Rahil, J., and Pratt, R. F. (1993) Kinetics and mechanism of β-
lactamase inhibition by phosphonamidates: The quest for a proton.
Biochemistry 32, 10763−10772.
(38) Thatcher, G. R. J., and Kluger, R. (1987) Mechanism and
catalysis of nucleophilic substitution in phosphate esters. Adv. Phys.
Org. Chem. 55, 99−265.
(39) Gerlt, J. A. (1992) Phosphate ester hydrolysis. In The Enzymes
(Sigman, D. S., Ed.) 3rd ed., Vol. XX, Chapter 3, Academic Press, San
Diego.
(40) Holmes, R. R. (2004) Phosphoryl transfer enzymes and
hypervalent phosphorus chemistry. Acc. Chem. Res. 37, 746−753.
(41) Birchall, J. M., and Haszeldine, R. N. (1956) Polyfluoroarenes.
II. Ionization constant of pentafluorophenol. J. Chem. Soc., 3653.
(42) Pratt, R. F. (2002) Functional evolution of the β-lactamase
active site. J. Chem. Soc., Perkin Trans. 2, 851−861.
(43) Dunn, B. M., DiBello, C., and Anfinsen, C. B. (1973) The pH
dependence of the steady state kinetic parameters for staphylococcal
nuclease-catalyzed hydrolysis of deoxythymidine-3′-phosphate in H2O
and D2O. J. Biol. Chem. 248, 4769−4774.
(44) Deiters, J. A., Galluci, J. C., and Holmes, R. R. (1982)
Computer simulation of staphylococcal nuclease action on thymidine-
3′,5′-bis(phosphate) (pdTp). J. Am. Chem. Soc. 104, 5457−5465.
(45) Powers, R. A., Caselli, E., Focia, P. J., Prati, F., and Shoichet, B.
K. (2001) Structures of ceftazidime and its transition-state analogue in
complex with AmpC β-lactamase: Implications for resistance
mutations and inhibitor design. Biochemistry 40, 9207−9214.
(46) Powers, R. A., Morandi, F., and Shoichet, B. K. (2002)
Structure-based discovery of a novel, noncovalent inhibitor of AmpC
β-lactamase. Structure 10, 1013−1023.
I
dx.doi.org/10.1021/bi4003887 | Biochemistry XXXX, XXX, XXX−XXX