Alkylidene oxapenem β-lactamase inhibitors revisited: Potent broad spectrum activity but new stability challenges

Article abstract of DOI:10.1021/ml5001855

We present a comprehensive study of C6-alkylidene containing oxapenems. We show that this class of β-lactamase inhibitors possesses an unprecedented spectrum with activity against class A, C, and D enzymes. Surprisingly, this class of compounds displayed significant photolytic instability in addition to the known hydrolytic instability. Quantum mechanical calculations were used to develop models to predict the stability of new analogues.

Full text of DOI:10.1021/ml5001855

Letter
pubs.acs.org/acsmedchemlett
Alkylidene Oxapenem β‑Lactamase Inhibitors Revisited: Potent
Broad Spectrum Activity but New Stability Challenges
Matthew D. Miller,†,§ Manoj Kale,‡ Kishore Reddy,‡ Sharon Tentarelli,† Mark Zambrowski,†
Minli Zhang,† Tiffany Palmer,† John Breen,† Sushmita Lahiri,† Pravin S. Shirude,*,‡,⊥
and Jeroen C. Verheijen*,†
†Infection iMed, AstraZeneca, 35 Gatehouse Drive, Waltham, Massachusetts 02451, United States
‡Department of Medicinal Chemistry, AstraZeneca India Pvt. Ltd, Avishkar, Bellary Road, Bangalore 560024, India
S
* Supporting Information
ABSTRACT: We present a comprehensive study of C6-
alkylidene containing oxapenems. We show that this class of β-
lactamase inhibitors possesses an unprecedented spectrum
with activity against class A, C, and D enzymes. Surprisingly,
this class of compounds displayed significant photolytic
instability in addition to the known hydrolytic instability.
Quantum mechanical calculations were used to develop models to predict the stability of new analogues.
KEYWORDS: Oxapenems, β-lactamase inhibitors, Gram-negative infections, quantum mechanical calculations, stability
he β-lactam antibiotics constitute one of the oldest and
most widely used classes of antibacterial agents.
oxapenems to hydrolysis. Pfaendler10 and Wild11 established
some limited structure−activity relationships (SAR) to enhance
stability of the compounds. Wild also described the synthesis of
C6-alkylidene oxapenems, containing a double bond attached
to the nonbridge carbon on the lactam ring, and noted
significant effects of the alkylidene substituent on the stability of
the oxapenem core.12 However, no additional SAR around the
alkylidene oxapenems have been reported.
The structural similarity of the C6-alkylidene oxapenems to
the alkylidene penems, exemplified by the clinical candidate
BLI-48913 (Figure 1), caught our attention. Given AstraZe-
neca’s long-standing interest in developing new β-lactamase
inhibitors,14,15 we decided to revisit this scaffold and explore its
spectrum against clinically relevant β-lactamases, particularly
from class C and D. Herein, we report the extraordinary β-
lactamase inhibitory activity of oxapenems with a C6-alkylidene
substituent. In addition, we describe predictive models based
on quantum mechanical calculations to predict the stability of
oxapenem compounds.
Previous reports have shown that a tert-butyl substituent at
the C2 position led to the highest stability for oxapenems.11,12
Therefore, we focused our efforts on modifications of the C6
position while keeping the C2 tert-butyl substituent in place.
The synthesis of the alkylidene oxapenems was carried out
according to previously described methods (Scheme 1).12
Thus, acetoxyazetidinone 1 was converted into methyl
thioether 2, followed by protection of the β-lactam nitrogen
with a silyl group. An aldol condensation between racemic 3
and various aldehydes gave access to 4 as a mixture of eight
T
Unfortunately, resistance to the class is widespread. The
primary resistance mechanism is through expression of a
collection of enzymes known as β-lactamases that proficiently
hydrolyze the β-lactam to an inactive metabolite. One
successful strategy to circumvent this resistance combines the
β-lactam antibiotic with an agent capable of inhibiting β-
lactamases.1−3 There are currently only three marketed β-
lactamase inhibitors, which only inhibit Ambler class A β-
lactamases.4,5 The clinical utility of these established agents is
limited by the spread of class C and D enzymes. Hence, there is
a significant unmet medical need for broad-spectrum β-
lactamase inhibitors with activity against all three classes of
serine proteases (class A, C, and D).
β-Lactamase inhibitors based on the oxapenem scaffold have
been known for several decades, beginning with a paper
describing the synthesis in the 1970s.6 Later research showed
greater potency of C6-alkoxy substituted oxapenems (Figure 1)
against several enzymes relative to the marketed drug clavulanic
acid.7−9 These reports mention the susceptibility of the
Received: May 7, 2014
Accepted: June 20, 2014
Published: June 20, 2014
Figure 1. Structures of C6-alkoxy and C6-alkylidene oxapenems, and
penem BLI-489.
© 2014 American Chemical Society
915
dx.doi.org/10.1021/ml5001855 | ACS Med. Chem. Lett. 2014, 5, 915−920

ACS Medicinal Chemistry Letters
Letter
a
Scheme 1. Synthesis of Oxapenems
a
Reagents: (a) EtOH, NaSMe, 0−25 °C, 60−70%; (b) DMF, TBDMSCl, DABCO, 80−90%; (c) RCHO, LDA, THF, −78 °C, 40−50%; (d) Ac2O,
TEA, DMAP, 80−90%; (e) TBAF/AcOH, THF, 60−80%; (f) Cs2CO3/ACN, 50−70%; (g) LHMDS/THF, 60−70%; (h) NCS, silica, DCM, 30−
t
40%; (i) BuOK/tBuOH, THF, 20−30%; (j) TPP, Pd(PPh3)4/DCM, sodium-2-ethyl hexanoate, 40−50%.
diastereomers. Protection of the hydroxyl group (→ 5) was
followed by desilylation (→ 6) and alkylation to give 7.
Acylation of the acetate in 7 with pivaloyl chloride gave access
to intermediate 8. Conversion of the methyl thioether into a
chloride (9) set the stage for a base-mediated cyclization with
concomitant elimination of the acetoxy group to form the C6-
alkylidene (10). It should be noted that the last step removed
two chiral centers, resulting in a significant reduction of
complexity as the mixture no longer consisted of eight
diastereomers but was instead a single racemate. The synthesis
was completed by deallylation to give target compound 11 as
the sodium salt. The yields for each step were generally
moderate to good. A slightly lower yield was observed for the
key aldol condensation (3 → 4) as well as for the last two steps.
The lower yields for the last two steps presumably result from
the susceptibility of alkylidene penems to hydrolysis.
We next evaluated the inhibitory activities of the oxapenems
against representative serine β-lactamases from class A, C, and
D and compared them with those of the marketed β-lactamase
inhibitor clavulanic acid (12, Table 1). β-Lactamase inhibitory
activities were evaluated following 5 min preincubation in an
enzymatic turnover assay utilizing the colorimetric substrate
nitrocefin. The racemic 2-furan E-alkylidene 13, first reported
by Wild et al.,12 was significantly more potent than
enantiomerically pure clavulanic acid, particularly against class
C and D enzymes. In fact, 13 showed single digit to double
digit nanomolar IC50s against all class A, C, and D β-lactamases
tested. The observed inhibitory activity for 13 is among the best
reported to date for any β-lactamase inhibitor. To assess the
role of the alkylidene substituent, compound 14, lacking the
alkylidene, was synthesized. It was evident by comparing 13
and 14 that the conjugated double bond at the C6 position of
the oxapenem scaffold is essential to obtain broad-spectrum
activity, particularly against class A enzymes. However, the
desalkylidene compound 14 remained surprisingly potent
against class C enzymes (and to a slightly lesser extent against
class D enzymes).
compounds 15E and 16E were significantly more potent
inhibitors than their corresponding Z isomers. Table 1 also
showed that replacement of the 2-furan (13) with a phenyl
group (15) resulted in increased potency against class C
enzymes, but reduced activity against the class A carbapene-
mase KPC-2. While compound 16E, containing a 3-furan
group, was more potent against KPC-2 than the compound
with the phenyl substituent (15E), its activity was less than that
of the 2-furan (13), particularly against class C enzymes.
Finally, we examined the effect of introduction of
substituents on the aromatic ring by making the o-, m-, and
p-methoxyphenyl alkylidenes (17−19). As can be seen from
Table 1, substitution at each of these 3 positions was tolerated,
with the resulting compounds displaying similar or slightly
improved potency compared to unsubstituted phenyl 15E.
During the course of this preliminary SAR evaluation, we
realized that further optimization of activity would be a fruitless
endeavor if we did not first address the severe stability issues
observed for this initial set of compounds. The stability of the
compounds in Table 1 was too low to allow evaluation in more
advanced models, such as cellular activity assays (inhibition of
bacterial growth) or determination of physicochemical proper-
ties. Surprisingly, we found that even under strictly anhydrous
conditions, the compounds in Table 1 suffered from rapid
degradation. Analysis of the reaction mixtures revealed that the
degradation pathways for anhydrous and aqueous degradation
differed, raising the interesting possibility that this class of
compounds suffered from photolytic instability in addition to
hydrolytic instability.
Indeed, under carefully controlled conditions, we were able
to separate the two distinct degradation pathways (Scheme 2).
Thus, HPLC−MS−MS analysis of the degradation mixture in
aqueous buffer with exclusion of light suggested that hydrolytic
degradation proceeded through opening of the β-lactam with
concomitant decarboxylation (D1). An additional minor peak
with an identical mass and mass-spec fragmentation pattern but
shorter retention time was also detected, suggesting formation
of an isomer of D1 (D2). However, analysis of an anhydrous
DMSO solution of 13 subjected to ambient lighting suggested
formation of three major degradants from two separate
Having confirmed the essentiality of the alkylidene
substituent for broad spectrum coverage, we examined the
role of the alkylidene geometry. As shown in Table 1,
916
dx.doi.org/10.1021/ml5001855 | ACS Med. Chem. Lett. 2014, 5, 915−920

ACS Medicinal Chemistry Letters
Letter
Table 1. Inhibitory Activity of Oxapenems against Representative β-Lactamases
a
b
E. cloacae ARC3525. P. aeuruginosa PAO1, PDC-1.
also detected during the photolysis experiment, but at
significantly lower levels than during aqueous hydrolysis
experiments. Comparison of the hydrolytic and photolytic
experiments revealed that the typical β-lactam hydrolysis
product D3 was not detected when degradation proceeded in
the dark. This counterintuitive observation may indicate rapid
decarboxylation of the initial hydrolysis product D3. It is
conceivable that detection of D3 during photolysis reflects the
formation of a reactive intermediate under anhydrous
conditions, which is converted into D3 in the aqueous mobile
phase during analysis, thereby limiting decarboxylation prior to
detection (and explaining the detection of only minor
quantities of D1 and D2 in the photolysis mixture). Finally, it
should be noted that all structural assignments are tentative,
based on HPLC−MS−MS only, and that formation of isomers
with identical molecular weight cannot be ruled out.
Scheme 2. Putative Decomposition Pathways for Compound
13 during Photolytic (Red Arrow) and Hydrolytic (Blue
Dotted Arrow) Degradation
pathways: ring opening of the β-lactam to give D3 along with
cleavage of the β-lactam ring to give a furan acrylic acid (D4)
and an oxazole (D5). The detection of the latter two products
suggested that this degradation proceeded through a light-
induced retro 2 + 2 addition. A small peak for oxazole D6 was
detected as well during the photolysis experiment, presumably
the result of oxidation of D3. Small peaks for D1 and D2 were
Having established the two different degradation mechanisms
that plagued the C6-alkylidene oxapenems, we set out to
develop robust methods to measure the photolytic and
hydrolytic stability. In order to measure the photolytic stability
of the oxapenems, data was collected in triplicate on anhydrous
917
dx.doi.org/10.1021/ml5001855 | ACS Med. Chem. Lett. 2014, 5, 915−920

ACS Medicinal Chemistry Letters
Letter
solutions in DMSO using an autosampler rack that had been
modified to allow controlled exposure of the samples to light.
Since traditional sun-chamber conditions were found to
degrade the compounds very rapidly, a fluorescent light was
used to slightly accelerate photodegradation over general
laboratory lighting for optimal read-out. Samples were tested
at 6 min intervals in triplicate with an autosampler using
UPLC-MS. Decomposition was determined based on the
quantitative peak area remaining for each measurement.
Quantitation was done using detection of the DAD-UV signal
between 200 and 400 nm.The standard deviation between the
triplicate measurement was reasonable for each of the
compounds tested (<10% of the reported value). The results
of the photolytic stability assays are shown in Table 2 (data
sorted by measured half-life).
Table 2. Experimental Photolytic Stability Data and
Calculated Absorbance Wavelengths for C6 Alkylidene
Oxapenems; Data Are Sorted Based on Longest Measured
t1/2
Figure 2. Predictive hydrololytic half-life versus experimental hydro-
lytic half-life for oxapenem compounds. Data points are colored based
on geometry of the alkylidene.
compound (nm)
photolytic t1/2 (h)
λcalc‑long
14
115
286
421
420
443
445
435
446
444
456
15Z (1.4 h),12 which compared well with our hydrolytic
stability data (3.1 and 0.5 h, respectively). Performing the
stability experiments multiple times suggested our method was
highly reproducible (Supporting Information). Comparison to
known inhibitors showed that the oxapenems are significantly
less stable: both clavulanic acid and BLI-489 had half-lives of
>50 h in this assay.
16Z
16E
18
0.38
0.32
0.21
0.21
0.20
0.20
0.17
0.22
15E
19
17
15Z
13
Having established a reproducible, quantitative method to
measure hydrolytic stability, we attempted to provide a
predictive model for hydrolytic stability. We anticipated a
greater energy barrier from starting material to transition state
would translate to greater hydrolytic stability. However, neither
the calculated transition state energy (ΔGact) nor endother-
micity (ΔGrxn) (calculated using the B3LYP method and the 6-
31++G(d,p) basis set with the IEF-PCM solvation model)
showed an acceptable correlation with the measured hydrolytic
stability. This observation may indicate that different
compounds degrade through different pathways or have
different rate limiting steps and/or transition states.
At this point, having nine data points for experimental
hydrolytic stability of alkylidene oxapenems, we were able to
generate a hydrolytic stability QSAR model. The potential
descriptors used to generate the model were calculated using
the software Semichem CODESSA (Comprehensive Descrip-
tors for Structural and Statistical Analysis; Semichem, Shawnee
Mission, KS) following AM1 energy minimization of each
structure, and the best model was selected based on the
multilinear regression (MLR) method.
The data in Table 2 showed that every compound with an
extended conjugated system suffered from photolytic instability.
The one compound that was stable (14) lacked the extended
conjugated system and was not expected to absorb visible light.
For comparison, BLI-489 (which also possesses an extended
conjugated system) had a half-life of 0.5 h in this assay, whereas
clavulanic acid (lacking the conjugated system) was stable in
this assay (half-life > 100 h). We used the time-dependent
density functional theory (TD-DFT) method to calculate the
wavelengths of the absorption maxima. These data are also
recorded in Table 2. Confirming our hypothesis, the calculated
absorption maxima for all photolytically labile compounds were
in the visible range, whereas the absorbance maximum for the
most stable compound (14) was outside the visible range.
Moreover, the two alkylidene oxapenems with the shortest
calculated absorbance wavelengths (16Z and 16E) were
somewhat more stable than the alkylidene oxapenems with
longer calculated absorbance wavelengths. Within this series,
we felt confident that we could use the calculated absorbance
maxima to design compounds with improved photolytic
stability.
For analyis of hydrolytic stability, we diluted stock solutions
of the test compounds in DMSO to 200 μM in pH 7.4 buffered
aqueous solutions. The solutions were stored in amber vials in
racks with a controlled temperature (37 °C). Samples were
analyzed by HPLC-UV in triplicate using an autosampler over a
period of at least 8 h with a minimum of 10 injections per vial.
The quantitative peak area, determined by DAD-UV detection
between 210 and 400 nm, remaining for each sample was used
to monitor hydrolysis (Figure 2). Wild et al. reported half-lives
of oxapenems under different conditions, for 15E (4.3 h) and
Using the MLR method, we generated a QSAR model with a
statistically valid correlation between predicted and exper-
imental hydrolytic stability (R2 = 0.88, F-value =53, t test =
−7.2): predicted hydrolytic half-life = −1571.2 − 112.8·
ERes,C−N
.
This correlation included a single descriptor, the average
two-center core−electron resonance energy for a C−N bond
(ERes,C−N), which describes the charge transfer ability between
all carbon−nitrogen bonds. While a correlation between a
property and descriptor does not necessarily indicate a
causative relationship, these compounds contain a single
nitrogen atom (the atom adjacent to the carbonyl), which is
918
dx.doi.org/10.1021/ml5001855 | ACS Med. Chem. Lett. 2014, 5, 915−920

ACS Medicinal Chemistry Letters
Letter
adjacent to the site of attack for hydrolysis. Thus, a mechanistic
correlation between the descriptor and the reaction rate seems
plausible. The overall trend was quite reasonable (Figure 2) and
provided confidence that this model would be able to predict
hydrolytic stability of the oxapenems.
With models to predict both hydrolytic and photolytic
stability in place, we evaluated a large number of virtual
compounds, possessing various aromatic and aliphatic
alkylidene substituents, in order to identify candidates for
synthesis with acceptable expected stabilities. The criteria we
used were predicted hydrolytic half-life > 20 h and predicted
λcalc‑long < 400 nm. Unfortunately, none of the evaluated
compounds met the hydrolytic half-life cutoff. However, several
compounds were predicted to have lower λcalc‑long (illustrative
examples are given in Table 3). Of these, some compounds
photolytic stability, the described oxapenems could be valuable
tools to explore inhibition of a broad range of β-lactamases. We
have also demonstrated that alkylidene oxapenems suffer from
photolytic instability in addition to their known susceptibility to
hydrolysis. This new observation may explain why previous
attempts to develop C6-alkylidene oxapenems into drug
candidates have failed. Finally, the predictive hydrolytic and
photolytic stability models presented in this letter may prove
valuable for other projects that aim to improve these types of
stabilities in β-lactams or other cores.
ASSOCIATED CONTENT
* Supporting Information
■
S
Full experimental details for compounds synthesized and
descriptions of biological assays. This material is available free
of charge via the Internet at http://pubs.acs.org.
Table 3. Predicted Hydrolytic Half-Life and Photolytic
λcalc‑long for Representative Virtual Alkylidene Oxapenems
AUTHOR INFORMATION
Corresponding Authors
■
*(J.C.V.) Tel: +1 781 839 4919. Fax: +1 781 839 4500. E-mail:
jeroen.verheijen@astrazeneca.com.
*(P.S.S.) Tel: +91 080 23621212. Fax: +91 080 23621214. E-
mail: pravin.shirude@astrazeneca.com; psshirude@gmail.com.
Present Addresses
§Semichem, Inc., Shawnee, Kansas 66216, United States.
⊥Sai Life Sciences Ltd, Pune 411057, India.
Author Contributions
M.M., J.V., and P.S.S. drafted the manuscript. M.M., J.V., S.L.,
and P.S.S. participated in the design and execution of this study.
M.K. and K.R. performed the chemical syntheses. S.T., M.Za.,
and M.Zh. performed stability studies. T.P. and J.B. performed
enzyme inhibition experiments.
Notes
The authors declare the following competing financial
interest(s): The authors are current or former employees of
AstraZeneca and may possess AstraZeneca stock.
ACKNOWLEDGMENTS
■
were predicted to be inherently unstable (cf. compounds 20
and 23). Other compounds (e.g., 21 and 22) were predicted to
be at the lower end of the range of hydrolytic stabilities
observed to this point. While we did not expect these
compounds to become candidates for further development,
we were still interested in preparing these two compounds to
test the validity of our predictive models. Unfortunately,
attempts to make compounds 21 and 22 failed, apparently due
to instability of the final products. The latter was consistent
with the low predicted hydrolytic stability (although it should
be noted that several compounds in Table 1 were successfully
prepared despite similar or lower predicted hydrolytic
stabilities).
Since both models showed a reasonable correlation between
predicted and measured stabilities, at this point we felt
confident that the SAR for hydrolytic and photolytic stability
was divergent and that the identification of a compound in the
alkylidene oxapenem series combining excellent activity,
hydrolytic stability, and photolytic stability would pose a
formidable challenge.
We are grateful to Ying Liu and Vladimir Capka for assistance
with the photolytic and hydrolytic degradation studies.
ABBREVIATIONS
■
itor
SAR, structure−activity relationships; BLI, β-lactamase inhib-
REFERENCES
■
(1) Llarrull, L. I.; Testero, S. A.; Fisher, J. F.; Mobashery, S. The
Future of the β-Lactams. Curr. Opin. Microbiol. 2010, 13, 551−557.
(2) Bush, K.; Macielag, M. J. New β-Lactam Antibiotics and β-
Lactamase Inhibitors. Expert Opin. Ther. Patents 2010, 20, 1277−1293.
(3) Buynak, J. D. Understanding the Longevity of the β-Lactam
Antibiotics and of Antibiotic/β-Lactamase Inhibitor Combinations.
Biochem. Pharmacol. 2006, 71, 930−940.
(4) Fisher, J. F.; Meroueh, S. O.; Mobashery, S. Bacterial Resistance
to β-Lactam Antibiotics: Compelling Opportunism, Compelling
Opportunity. Chem. Rev. 2005, 105, 395−424.
(5) Drawz, S. M.; Bonomo, R. A. Three Decades of β-Lactamase
Inhibitors. Clin. Microbiol. Rev. 2010, 23, 160−201.
While we were ultimately unable to identify a broad
spectrum inhibitor for clinical development, this study clearly
reveals the unprecedented enzyme inhibitory activity of the
oxapenems. Within the constraints of their hydrolytic and
(6) Cherry, P. C.; Newall, C. E.; Watson, N. S. Preparation of the 7-
Oxo-4-oxa-1-azabicyclo[3.2.0]hept-2-ene System and the Reversible
Cleavage of Its Oxazoline Ring. J. Chem. Soc., Chem. Commun. 1978,
11, 469−470.
919
dx.doi.org/10.1021/ml5001855 | ACS Med. Chem. Lett. 2014, 5, 915−920

ACS Medicinal Chemistry Letters
Letter
(7) Simpson, I. N.; Urch, C. J.; Hagen, G.; Albrecht, R.; Sprinkart, B.;
Pfaendler, H. R. Synthesis and Biological Activity of AM-112 and
Related Oxapenem Analogues. J. Antibiot. 2003, 56, 838−847.
(8) Jamieson, C. E.; Lambert, P. A.; Simpson, I. N. In Vitro and in
Vivo Activities of AM-112, a Novel Oxapenem. Antimicrob. Agents
Chemother. 2003, 47, 1652−1657.
(9) Jamieson, C. E.; Lambert, P. A.; Simpson, I. N. In Vitro Activities
of Novel Oxapenems, Alone and in Combination With Ceftazidime
Against Gram-Positive and Gram-Negative Organisms. Antimicrob.
Agents Chemother. 2003, 47, 2615−2618.
(10) Pfaendler, H. R.; Weisner, F.; Metzger, K. Synthesis and
Antibacterial Activity of (1′R,5R,6R)-2-tert-Butyl-6-(1′-hydroxyethyl)
Oxapenem-3-carboxylic Acid. Biorg. Med. Chem. Lett. 1993, 3, 2211−
2218.
(11) Wild, H.; Metzger, K. Substituted 6-Alkyloxapenems: Potent β-
Lactamase Inhibitors; Synthesis and Biological Characterization. Biorg.
Med. Chem. Lett. 1993, 3, 2205−2210.
(12) Wild, H.; Hartwig, W. First Synthetic Access to 6-(Methylene)-
oxapenems: A New Class of β-Lactamase Inhibitors. Synthesis 1992,
11, 1099−1103.
(13) Petersen, P. J.; Jones, C. H.; Venkatesan, A. M.; Mansour, T. S.;
Projan, S. J.; Bradford, P. A. Establishment of in Vitro Susceptibility
Testing Methodologies and Comparative Activities of Piperacillin in
Combination With the Penem-Lactamase Inhibitor BLI-489. Anti-
microb. Agents Chemother. 2009, 53, 370−384.
́
(14) Ehmann, D. E.; Jahic, H.; Ross, P. L.; Gu, R. F.; Hu, J.; Kern, G.;
Walkup, G. K.; Fisher, S. L. Avibactam Is a Covalent, Reversible, Non-
β-lactam β-Lactamase Inhibitor. Proc. Natl. Acad. Sci. U.S.A. 2012, 109,
11663−11668.
(15) Ehmann, D. E.; Jahic,
́
H.; Ross, P. L.; Gu, R. F.; Hu, J.; Durand-
Reville, T. F.; Lahiri, S.; Thresher, J.; Livchak, S.; Gao, N.; Palmer, T.;
́
Walkup, G. K.; Fisher, S. L. Kinetics of Avibactam Inhibition Against
Class A, C, and D β-Lactamases. J. Biol. Chem. 2013, 288, 27960−
27971.
920
dx.doi.org/10.1021/ml5001855 | ACS Med. Chem. Lett. 2014, 5, 915−920