Potent inhibitors of LpxC for the treatment of gram-negative infections

Article abstract of DOI:10.1021/jm2014748

In this paper, we present the synthesis and SAR as well as selectivity, pharmacokinetic, and infection model data for representative analogues of a novel series of potent antibacterial LpxC inhibitors represented by hydroxamic acid 1a.

Full text of DOI:10.1021/jm2014748

Article
pubs.acs.org/jmc
Potent Inhibitors of LpxC for the Treatment of Gram-Negative
Infections
Matthew F. Brown,*,† Usa Reilly,† Joseph A. Abramite,† Joel T. Arcari,† Robert Oliver,† Rose A. Barham,†
Ye Che,‡ Jinshan Michael Chen,† Elizabeth M. Collantes,‡ Seung Won Chung,† Charlene Desbonnet,§
Jonathan Doty,† Matthew Doroski,† Juntyma J. Engtrakul,∥ Thomas M. Harris,† Michael Huband,§
John D. Knafels,⊥ Karen L. Leach,∥ Shenping Liu,⊥ Anthony Marfat,† Andrea Marra,§ Eric McElroy,†
Michael Melnick,† Carol A. Menard,# Justin I. Montgomery,† Lisa Mullins,§ Mark. C. Noe,†
John O’Donnell,§ Joseph Penzien,§ Mark S. Plummer,† Loren M. Price,† Veerabahu Shanmugasundaram,‡
Christy Thoma,§ Daniel P. Uccello,† Joseph S. Warmus,† and Donn G. Wishka†
†Worldwide Medicinal Chemistry, ‡Computational Chemistry, §Antibacterials Research Unit, ∥Drug Metabolism, ⊥Structural Biology,
#Primary Pharmacology, Pfizer Global Research and Development, Eastern Point Road, Groton, Connecticut 06340, United States
S
* Supporting Information
ABSTRACT: In this paper, we present the synthesis and SAR as well as
selectivity, pharmacokinetic, and infection model data for representative
analogues of a novel series of potent antibacterial LpxC inhibitors
represented by hydroxamic acid 1a.
and an outer membrane composed of peptidoglycan. In addi-
tion to these membranes, Gram-negative bacteria possess an
additional outer membrane decorated with lipopolysaccharide
(LPS). LPS is composed of three distinct units. The outer-
most unit is the O-antigen, which is a glycan polymer. This is
linked to a sugar-containing core domain which is appended
to the membrane anchoring group, lipid A (endotoxin). This
outer membrane provides a substantial protective barrier, and
Gram-negative bacteria lacking lipid A are either not viable or
are highly susceptible to a range of anti-infective drugs, suggest-
ing that enzymes involved in lipid A production could represent
new drug targets.6 Nine unique enzymes catalyze the synthesis
of lipid A and UDP-3-O-(R-3-hydroxymyristol)-N-acetylglucos-
amine deacetylase (LpxC) is a cytosolic zinc−metalloamidase
responsible for carrying out the second biosynthetic step.7 The
essential role LpxC plays in the biosynthesis of lipid A, coupled
with a lack of homology with mammalian proteins, has cata-
lyzed efforts to identify small molecule LpxC inhibitors for the
treatment of serious Gram-negative infections.
INTRODUCTION
■
Drug resistant Gram-negative infections are an increasing threat
to public health, especially for seriously ill, hospitalized patients.
Infections arising from multidrug resistant Pseudomonas aeruginosa
and Acinetobacter species as well as carbapenem resistant Klebsiella
pneumoniae present formidable challenges for the medical
community as few treatment options remain.1 Unfortunately,
while resistance to current therapies continues to spread, new
clinical agents to treat these infections are few in number.2 In
addition, the costs and regulatory challenges associated with the
development of new antibacterial agents, combined with the per-
ception of modest future profitability, has unfortunately resulted in
most major pharmaceutical companies exiting anti-infective drug
research at a time when it is most needed.3
While it has been regularly stated that there is a great need to
identify novel lead series with new antibacterial modes of action,
in reality this goal remains highly elusive. A 2007 paper from
GlaxoSmithKline nicely illustrates this challenge as a meager five
lead series were identified from 67 target-based high-throughput
screens conducted over a 7 year period.4 Inhibition of cell wall
biosynthesis is a well-precedented anti-infective strategy (i.e.,
β-lactams), and efforts to identify novel targets essential to cell
wall biosynthesis in Gram-negative bacteria have been reported.5
Gram-positive bacteria contain both a cytoplasmic inner membrane
Inhibitors were first disclosed by researchers at Merck in the
1990s (e.g., L-161,240, Figure 1),8 and this was followed by a
Received: November 1, 2011
Published: December 18, 2011
© 2011 American Chemical Society
914
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
Figure 1. LpxC inhibitors.
a
Scheme 1. General Preparation of Biaryl Analogues 1
a
Reagents and conditions: (a) NaSO2Me, EtOH, reflux; (b) 1-bromo-4-(2-bromoethyl)benzene, Cs2CO3, DMF; (c) (i) LiOH, THF, MeOH, H2O,
(ii) classical resolution using (−)-ephedrine, IPA, H2O; (d) CDMT, NMM, NH2OTHP, 2-MeTHF; (e) ArB(OH)2 or ArBr, Pd(II) EnCat, K2CO3,
EtOH, H2O, 80 °C; (f) HCl 1,4-dioxane, H2O; (g) bis(pinacolato)diborane, Pd(dppf)Cl2, KOAc, 1,4-dioxane, 100 °C; (h) HOBt, Et3N,
NH2OTHP, EDCI, CH2Cl2.
variety of reports describing inhibitors with improved enzyme
potency and antibacterial spectrum, with CHIR-090 being one of
the most extensively studied inhibitors.9 Crystallography and NMR
studies have provided a wealth of information regarding LpxC
protein conformation as well as the binding mode of various small
molecule inhibitors at the active site.10 L-161,240, CHIR-090, and
other inhibitors described in the literature are structurally similar in
that they contain a zinc-binding group (e.g., hydroxamic acid) and
a hydrophobic tail which mimics the hydroxy-myristate fatty acid
group found in the natural enzyme substrate. These lipophilic
moieties make important van der Waals interactions with the
enzyme in a hydrophobic tunnel, which are critical for potent
enzyme inhibition. Here, we provide initial details regarding a
novel series of alkyl hydroxamic acid LpxC inhibitors, exemplified
by compound 1a (Figure 1). The potent activity exemplified by
this series can, in part, be attributed to an important hydrogen-
bonding interaction between the alkylsulfone headgroup and a
lysine residue located near the catalytic site (Lys238).
each enantiomer (2R)-5 and (2S)-5 were obtained to assign the
absolute stereochemistry. The preferred (2R)-5 enantiomer, which
provides the more active hydroxamic acid analogues 1, was con-
verted to the phenyl bromide template 6 under standard amide
coupling conditions with O-(tetrahydro-2H-pyran-2-yl)hydroxyl-
amine (NH2OTHP). The boronate ester template 7 was prepared
in a two-step sequence from 5 involving a palladium-catalyzed cross-
coupling reaction with bis(pinacolato)diborane followed by an amide
formation with NH2OTHP. Templates 6 and 7 were subjected to
Suzuki−Miyaura cross coupling reactions with aryl boronic acids
and aryl bromides, respectively, and subsequent acid-mediated
deprotections to provide the desired hydroxamic acids 1a−x.
The biphenyl analogues 10, 14, and 17, and generally all
analogues described herein, were prepared in a similar fashion
to that of 1a−x whereby incorporation of the hydrophobic tail
is achieved through alkylation of a sulfone ester moiety. For
example, treatment of methylsulfone 8 with sodium hydride
and 4-(2-bromoethyl)biphenyl provided ester 9, which was
converted to the hydroxamic acid 10 under direct amidation
conditions with hydroxylamine hydrochloride (Scheme 2). The
phenylsulfone 11, prepared in a similar fashion to that of
methylsulfone 3, was treated with cesium carbonate and 4-(2-
iodoethyl)biphenyl to provide the biphenyl intermediate 12.
After basic hydrolysis, the corresponding free acid (13) was
converted to the hydroxamic acid 14. Likewise, the sulfolane
derivative 15 was alkylated under basic conditions with 4-(2-
iodoethyl)biphenyl to provide ester 16, which was transformed
using standard hydrolysis/amidation/deprotection procedures
to the desired hydroxamic acid 17.
CHEMISTRY
■
The majority of the LpxC inhibitors that are described within
are biaryl derivatives (1a−x), and these were prepared in short
order from templates 6 and 7, as illustrated in Scheme 1.
Treatment of chloride 2 with sodium methanesulfinate in
refluxing ethanol provided the methylsulfone 3, which was
alkylated with 1-bromo-4-(2-bromoethyl)benzene under basic
conditions to provide ester 4. Saponification of 4 provided the
acid 5, as a racemic mixture, that was subjected to a classical
resolution with (−)-ephedrine. Single crystal X-ray structures of
915
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
a
Scheme 2. Synthesis of Biphenyl Analogues 10, 14, and 17
a
Reagents and conditions: (a) (i) NaH, DMF, 0 °C; (ii) 4-(2-bromoethyl)biphenyl, 0 °C to rt; (b) NH2OH·HCl, NaOMe, MeOH, THF, 0 °C to rt;
(c) NaSO2Ph, EtOH, reflux; (d) 4-(2-iodoethyl)biphenyl, Cs2CO3, DMF; (e) LiOH, THF, MeOH, H2O; (f) CDMT, NMM, NH2OTHP,
2-MeTHF; (g) HCl, 1,4-dioxane, H2O; (h) NH2OTHP, PyBOP, DIPEA, CH2Cl2.
a
The preparation of analogues 19, 20, and 22, containing hetero-
atoms linking the biaryl tail group to the hydroxamate headgroup,
is shown in Scheme 3. Deprotonation of sulfone 3 with lithium
hexamethyldisilazide (LiHMDS), followed by treatment with
(4-bromophenyl)-(chloromethyl)sulfane, furnished an arylbromide
intermediate that was coupled with phenyl boronic acid under
typical Suzuki−Miyaura conditions to provide ester 18. Hydro-
lysis, amide coupling with NH2OTHP, and acid-promoted depro-
tection of 18 led to the hydroxamic acid 19, which was oxidized to
the sulfone 20 with oxone. Compound 22 was produced in a
similar fashion, whereby 3 was first alkylated with 4-(chloro-
methoxy)biphenyl to provide ether 21, which was then converted
by the same sequence of reactions to the hydroxamic acid 22.
A variety of methods were utilized to prepare the phenyl
ether analogues 24, 26, 27, 29 and 30, as illustrated in Scheme 4.
The alkylation of sulfone 3 with 1-(benzyloxy)-4-(2-iodoethyl)-
benzene under basic conditions provided ester 23, which upon
saponification was converted to the hydroxamic acid 24, as
previously described. Hydrogenolysis of 23 with Pearlman’s
catalyst and cyclohexene provided the common phenol inter-
mediate 25. Methylation of 25, followed by conversion of the
ester moiety to the hydroxamic acid in the usual manner, pro-
duced 26. Mitsunobu reactions were employed to install the
cyclopentyl and (4-pyridyl)methyl groups of 27 and 29,
respectively, and an mCPBA oxidation of the (4-pyridyl)methyl
group of 28 ultimately led to the (4-pyridinyl)methyl N-oxide
30. In general, the hydroxamic acid moieties of these analogues
were introduced in a similar fashion (ester hydrolysis, amide
bond formation, and an acidic deprotection).
Scheme 3. Synthesis of biphenyl analogues 19, 20, and 22
a
Reagents and conditions: (a) (i) LiHMDS, THF, NMP, 0 °C, (ii)
(4-bromophenyl)-(chloromethyl)sulfane, NaI, 0 °C to rt; (b) PhB-
(OH)2, Pd(PPh3)4, K2CO3, 1,4-dioxane, H2O, 80 °C; (c) NaOH, 1,4-
dioxane, H2O; (d) (i) CDMT, NMM, 2-MeTHF, (ii) NH2OTHP;
(e) PPTS, EtOH; (f) oxone, MeOH, H2O; (g) (i) NaH, DMP, 0 °C,
(ii) 4-(chloromethoxy)biphenyl, NaI, 0 °C to rt.
The thioether analogue 33 was prepared from the
enantiopure acid 5, beginning with the conversion to the
methyl ester via alkylation with methyl iodide (Scheme 5).
A palladium-catalyzed cross-coupling with bis(pinacolato)-
diborane provided the pinacol boronate ester, which was then
916
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
Scheme 4. Synthesis of Phenyl Ether Analogues 24, 26, 27,
29, and 30
RESULTS AND DISCUSSION
■
a
A novel structural feature of the series reported here is the
alkyl-linked sulfone-containing headgroup, which differs from
the amide-linked threonine-containing headgroup found in
CHIR-090 and related inhibitors. The sulfone moiety is engaged in
a hydrogen-bonding interaction with Lys238, an interaction
maintained by the threonine hydroxyl group common to the
CHIR-090-like inhibitors (Figure 2). The methyl group attached
Figure 2. X-ray crystal structure of 1t with Pseudomonas aeruginosa
LpxC enzyme.
to the sulfone resides in a fairly small hydrophobic space defined
by a cluster of phenylalanine residues, the closest being Phe191;
a similar van der Waals interaction is observed for the threonine
methyl group of the CHIR-090 analogues. This space will not
accommodate larger substituents. For example, the correspond-
ing analogue wherein the sulfone methyl group of 1a has been
replaced by the larger phenyl group (Table 1, compound 14) is
significantly less potent than 1a in the Pseudomonas aeruginosa
LpxC enzyme assay (PaeLpxC). While a simple bond rotation
would place the phenyl group in solvent exposed space and still
enable interaction of a sulfone oxygen with Lys238, the loss in
potency suggests that this conformation is not optimal, perhaps
due to the placement of the polar sulfone oxygens in the region
of the hydrophobic phenylalanine cluster, as well as the loss of
the van der Waals interaction present with the methyl-con-
taining analogue 1a. Interestingly, locking the sulfone moiety
into a ring as shown in compound 17 also leads to a significant
loss of enzyme activity, which may again be explained by an
unfavorable placement of the sulfone oxygens near the
phenylalanine cluster. Enzyme inhibitory activity in this
series resides with analogues with (R) absolute stereochemistry
(compare (R) enantiomer 1a (1.37 nM) to (S) enantiomer
(S)-1a (>95.2 nM)). While the binding site appears capable of
accommodating both configurations, compounds in the (S)
configuration place the methylsulfone group into solvent-
exposed space, thus losing the interaction with Lys238. This
result demonstrates that maintaining this H-bonding inter-
action is critical for achieving significant potency in this series.
The methyl group appended to the quaternary center in com-
pound 1a is located in a generally solvent exposed region,
perhaps making a positive van der Waals contact with a nearby
Met62. Racemic compound 10, the des-methyl analogue of
compound 1a, shows an approximate 10-fold loss of enzyme
potency relative to 1a. The absence of substitution at this
a
Reagents and conditions: (a) 1-(benzyloxy)-4-(2-iodoethyl)benzene,
Cs2CO3, DMF, 50 °C; (b) LiOH, THF, MeOH, H2O; (c) (i) oxalyl
chloride, DMF, DCM, (ii) NH2OTMS; (d) Pd(OH)2/C, cyclohexene,
EtOH, reflux; (e) MeI, Cs2CO3, DMF; (f) cyclopentanol, PPh3,
DIAD, THF, 0 °C to rt; (g) NaOH, 1,4-dioxane, H2O; (h) (i) CDMT,
NMM, 2-MeTHF, (ii) NH2OTHP; (i) PPTS, EtOH; (j) pyridin-4-
ylmethanol, PPh3, DIAD, THF, 0 °C to rt; (k) KOH, 2-MeTHF, H2O,
60 °C; (l) mCPBA, HOAc, 55 °C; (m) propylphosphonic anhydride,
Et3N, NH2OTHP, 2-MeTHF.
a
Scheme 5. Synthesis of Thioether Analogue 33
a
Reagents and conditions: (a) MeI, NaHCO3, THF, H2O; (b)
bis(pinacolato)diborane, Pd(dppf)Cl2, KOAc, 1,4-dioxane, 100 °C; (c)
NaIO4, NH4OAc, acetone, H2O; (d) phenyl disulfide, CuI, 2,2′-
bipyridine, DMSO, H2O, 100 °C; (e) NaOH, 1,4-dioxane, H2O; (f)
HOBt, Et3N, NH2OTHP, EDCI, CH2Cl2; (g) HCl, MeOH, 1,4-
dioxane, H2O.
subjected to oxidative conditions to furnish the boronic acid 31.
The copper-mediated cross coupling of 31 with phenyl disulfide
provided the thioether ester 32, which was transformed to the final
target 33, as previously described.
917
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
a
Table 1. LpxC Inhibitors
918
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
Table 1. continued
919
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
Table 1. continued
a
b
c
Analogues are (R)-enantiomers unless otherwise indicated. clogD calculated at pH 7.4 with ACD pchbat version 9.3. Ciprofloxacin.
(S)-enantiomer. racemic. n = 1 data. n = 2 data.
d
e
f
g
center may allow racemization to occur with isolated
enantiomers due to the acidic nature of the methine proton.
Therefore, in addition to van der Waals interactions, the methyl
group found in 1a could play an important role of locking the
molecule in the more potent (R) stereochemical configuration.
The composition of the tether group linking the sulfone
headgroup to the biphenyl moiety was briefly explored. The
two-carbon linker found in 1a is preferred while analogues with
the thioether (19), sulfone (20), and ether linkages (22) retain
reasonable potency. Of this set, the ether linked analogue 22
is least potent. The preferred C−CH2−O−Car torsion angle
for ether analogue 22 would be expected to be close to 0°,11
quite different from the angle observed in the crystal structures
for the ethylene-linked analogues (Figure 2). As observed with
literature LpxC inhibitors, the hydroxamic acid in our series
interacts with the active-site zinc in a chelating fashion and the
aryl tailgroup occupies the enzymatic hydrophobic tunnel. The
hydroxamic acid is an efficient zinc-binding group, and replace-
ment of this moiety with a carboxylic acid also capable of binding
zinc leads to significantly diminished potency (compare hydro-
xamic acid 1v (3.68 nM) with carboxylic acid 1w (>92.5 nM). The
morpholino group appended to the 4-position of the terminal aryl
ring in compound 1t exits the end of the hydrophobic tunnel and
resides in solvent-exposed space (Figure 2).
It has been reported in the literature that the moiety linked
to the zinc-binding headgroup needs to be a hydrophobic group
of adequate length to efficiently interact with the hydrophobic
tunnel and maintain reasonable potency.8,9c Therefore, removal
of the terminal aryl ring of compound 1a leads to significantly
diminished enzymatic potency (compound 34, >100 nM).
Adding a 4-methoxy group to the aryl ring offers no improve-
ment (26), however, a variety of other analogues with various
oxygen or sulfur-linked lipophilic groups provide a significant
boost in potency relative to 34 and 26 (e.g., 24, 27, 29 and 33).
Generally speaking, the biphenyl tailgroup present in
compound 1a provides the highest level of potency and
antibacterial spectrum, therefore, the effect of terminal ring
substitution of the biaryl group was extensively explored.
Direct attachment of polar functionality at the 2-position of
the terminal ring as shown with amide 1e leads to significantly
reduced enzymatic and cellular potency due to the desolvation
cost associated with placement of polar functionality in the
LpxC hydrophobic tunnel. In addition, the 2-position is
tolerant to small, lipophilic functional groups. However, activity
is reduced as the size of this group is increased. For example,
the 2-ethyl analogue 1d is 19-fold less potent than the des-ethyl
analogue 1a. A similar trend is observed at the 3-position, how-
ever, this position is more tolerant to varying size and polarity
as compared to the 2-position. With its placement near the
mouth of the hydrophobic tunnel, the 4-position of the terminal
aryl ring is tolerant to a wider range of functionality, and attach-
ment of a variety of groups at this position results in single-digit
nM enzyme inhibitors with impressive anti-Pseudomonal whole-
cell activity (1j−m, 1u, 1v, 1x). For example, while placement
of a polar primary amide at the 2- or 3-positions leads to
diminished potency relative to 1a, the 4-amide analogue 1m
demonstrates equivalent enzymatic activity. A variety of analo-
gues with basic functionality at the 4-position show decent
enzyme inhibition (1q−1t), however, the guanidine-containing
analogue 1s is substantially less active than the other amine-
containing analogues with regard to whole-cell activity. The
benzoic acid analogue 1o shows rather weak enzymatic activity,
however, addition of an alkylether linkage to place the polar acidic
functionality in solvent exposed space outside of the hydrophobic
tunnel provides a moderate improvement in potency (1p).
Replacement of the terminal aryl ring of compound 1a with a
variety of heteroaryl groups typically provides reduced potency.
For example, the pyridyl analogue 1n shows an approximate 20-
fold loss of enzymatic potency as compared to 1a. A number of
literature inhibitors, including CHIR-090, link a biphenyl
acetylene moiety to the zinc-binding headgroup to achieve
optimal potency. Given the safety concerns associated with
mechanism-based inactivation of cytochrome P450 enzymes
mediated by alkyne-containing drugs,12 it is noteworthy that
high-level enzyme potency and whole-cell activity is achieved in
this series in the absence of an arylacetylene.
The ability to append a range of functional groups at the
4-position of the terminal aryl ring provides the potential to fine-
tune molecular properties. That said, analogues in this series
with relatively low clogD values generally demonstrate reduced
enzymatic and cellular potency. For example, the highly polar
N-oxide found in compound 30 (clogD −1.4) is located in
solvent exposed space, yet this molecule is 4-fold less potent
than the corresponding pyridine 29 (clogD 0.5) in the enzyme
assay and ≥16-fold less active in the cell-based assay. This trend
has been generally observed in this series as molecules with low
clogD values (<0) regularly show reduced enzymatic and
whole-cell potency. Conversely, compounds from this series
920
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
Table 2. MIC90 Values for Select Analogues
MIC90 (μg/mL)
strains
n
cipro
Azt
CHIR-090
1a
1k
1l
1t
1u
1v
1x
Pseudomonas aeruginosa
Klebsiella pneumoniae
Escherichia coli
22
22
11
11
11
11
64
64
64
>64
>64
32
4
2
4
16
2
1
8
1
2
4
1
4
4
1
4
2
1
8
2
16
4
>64
1
0.25
0.5
0.5
>64
0.5
1
0.5
0.5
2
Enterobacter aerogenes
Citrobacter freundii
4
1
1
1
2
16
8
4
32
>64
32
8
2
4
Acinetobacter baumannii
64
32
32
16
>64
>64
>64
>64
with relatively high clogD values (>3) occasionally demonstrate
Gram-positive antibacterial whole-cell activity (Staphylococcus
aureus strain SA21 (ATCC 29213), Table 1), suggesting the
potential for off-target activity because Gram-positive bacteria
lack LpxC enzyme. Therefore, analogue design efforts were
focused within polarity space (clogD ∼ 0−3) to provide
optimal enzymatic and Gram-negative whole-cell activity while
minimizing Gram-positive activity. While oxidative metabolism
is not the major contributor to drug clearance in this series
(vide infra), focusing design at the lower end of this clogD scale
also reduced the extent of oxidative metabolism observed in
liver microsome incubations (unpublished results).
Micromolar whole-cell activity is common for Gram-negative
drug research programs and highlights the formidable cellular
barriers and resistance mechanisms present in these pathogens.
Given that drugs with cytosolic targets must pass through the
rather impenetrable outer membrane as well as the inner
membrane, while avoiding a number of multidrug resistant
efflux pumps to reach their biological targets,13 it is remarkable
that effective bacterial killing is ever achieved. The relative
impact of efflux on whole-cell activity can be illustrated by
determining the MIC shift that occurs for compound 1a when
comparing data from the wild-type Pseudomonas aeruginosa
strain PA01 (MIC = 0.125 μg/mL = 0.37 μM) to a PA01-
modified strain (PAO280) which lacks major efflux pumps
(Δ(mexAB−OprM) Δ(mexXY)Δ(MexZ)) (MIC = 0.008
μg/mL = 23 nM).14 The magnitude of the efflux-mediated
MIC shift illustrated by this example (∼ 16-fold) is consistently
observed with analogues in this series.
Having identified potent PaeLpxC enzyme inhibitors with
whole-cell activity, MIC90 values were obtained for select
compounds vs a panel of important Gram-negative pathogens,
including Pseudomonas aeruginosa, Klebsiella pneumoniae,
Escherichia coli, Enterobacter aerogenes, Citrobacter freundii, and
Acinetobacter baumannii (Table 2). The LpxC inhibitors
(CHIR-090 included) significantly outperform the comparator
agents ciprofloxacin and aztreonam with the exception that
ciprofloxacin demonstrates similar activity vs Enterobacter
aerogenes. None of the compounds evaluated in this study
exhibit significant activity vs Acinetobacter baumannii. In
general, the LpxC inhibitors described here exhibit diminished
Acinetobacter LpxC enzyme activity as compared to PaeLpxC,
leading to relatively weak whole-cell Acinetobacter activity.
Analogues 1a and 1v were further profiled against a larger
panel of 91 clinically relevant strains of Pseudomonas aeruginosa.
Phenol 1v was selected for this advanced evaluation based on
solid activity observed in the small panel study described above,
combined with the ability to utilize the phenol as a prodrug
handle to maximize aqueous solubility in future in vivo studies.
Both compounds demonstrate impressive MIC90 values of 2
and 4 μg/mL (Table 3). The LpxC inhibitors show superior
activity to all the comparator drugs in this study with the
Table 3. MIC90 Data vs 91 Strains of Pseudomonas aeruginosa
compd
range (μg/mL)
MIC50 (μg/mL)
MIC90 (μg/mL)
1a
1v
0.008−4
0.015−8
0.5
1
2
4
ciprofloxacin
amikacin
0.008 to >64
0.5 to >64
0.25−2
2
32
32
1
4
polymyxin B
cefepime
0.5
16
16
0.25 to >64
64
>64
ceftazidime
1 to >64
exception of polymyxin B, which is considered to be a drug of
last resort for the treatment of multidrug resistant Pseudomonas
aeruginosa infections due to its considerable toxicity.
Given the structural similarity of this series of LpxC in-
hibitors to known hydroxamic acid MMP inhibitors, compound
1a was also evaluated against a panel of metalloproteases. Good
selectivity was observed except for two targets showing Ki
values <10 μM (MMP12 and MMP13, Table 4). The selectivity
Table 4. MMP Activity for 1a
hMMPs
Ki (μM)
hMMPs
Ki (μM)
MMP1
MMP2
MMP3
MMP7
MMP8
MMP9
MMP10
MMP12
MMP13
>10
>10
>10
>10
>10
>10
>10
1.35
7.73
MMP14
MMP15
MMP16
MMP20
MMP24
MMP25
MMP26
TACE
>10
>10
>10
>10
>10
>10
>10
>10
of compound 1a was also determined against a panel of 120
targets at a 10 μM drug concentration.15 Minimal off-target
activity was observed as >50% inhibition was only obtained in
two instances; human norepinephrine (NE) transporter (IC50 =
340 nM, Ki = 250 nM) and PDE11 (IC50 = 18 μM).
Hydroxamic acid-containing drugs often suffer relatively high
plasma drug clearance due to metabolism of the hydroxamate
moiety, with glucuronide conjugate formation and aldehyde
oxidase-mediated metabolism being common issues.16
A
number of analogues in this series were evaluated in rats, and
in all cases, moderate to high clearance was observed following
iv dosing (Table 5). Glucuronide formation and aldehyde
oxidase-mediated reduction of the hydroxamate moiety to the
primary amide were determined to be major routes of meta-
bolism (unpublished results). The metabolic liability of the
hydroxamate moiety can be illustrated by comparing the
rapid clearance observed with hydroxamate analogue 1v
(65 mL/min/kg) following iv dosing in rats to the relatively
921
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
computer-controlled fluidic system combined with a triple quadrupole
mass spectrometer (BioTrove RapidFire).18 The enzyme assay was
conducted in 384 well polypropylene plates (Thermo 4312) in a total
volume of 50 μL containing 100 pM purified Pseudomonas aeruginosa
LpxC enzyme. The substrate buffer contains 100 mM Na2PO4, 1 mg/mL
bovine serum albumin, pH 7.0. The enzyme buffer contains 5 mM
Na2PO4 and 1.5 mg/mL bovine serum albumin, pH 7.5. Compounds
(5 μL in 100% DMSO: compound final concentration range 100 μM
to 1.0 pM) were preincubated with 20 μL of LpxC enzyme for 30 min
at room temperature. The substrate UDP-3-O-(R-hydroxydecanoyl)-
N-acetylglucosamine was added to initiate the assay, 25 μL at 1.0 μM
(final concentration 0.5 μM). Assay plates were sealed, incubated at
room temperature for 60 min, and the reaction was stopped by the
addition of 10 μL of 1 N hydrochloric acid. The plates were read on
the BioTrove RapidFire. Percent substrate conversion was calculated
and plotted to obtain IC50 values (at least n = 3 unless specified
otherwise).
Antimicrobial Susceptibility Testing. The minimum inhibitory
concentration (MIC) values were determined using the broth
microdilution protocol according to the methods of the Clinical and
Laboratory Standards Insitute (CLSI). Inoculum Preparation and Direct
Colony Suspension: Bacterial strains were grown on Tryptic soy agar
with 5% sheep blood for 24 h prior to testing. Colonies of strains were
suspended into sterile saline until a turbidity standard of 0.5
McFarland was achieved (approximately 1 × 108 CFU/mL). This
suspension was diluted 1:400 in Mueller−Hinton broth (2.5 × 105
CFU/mL). Drug Dilution Tray Preparation: Compounds are diluted in
appropriate solvent to 2200 μg/mL. The microbroth dilution stock
plates were prepared in 2-fold dilution series, 64−0.06 μg drug/mL
(high dilution series) and 0.25−0.00025 μg drug/mL (low dilution
series). For tube 1 of the high dilution series, 200 μL of the 2220 μg/
mL stock was added to duplicate wells of a 96-well microtiter plate.
For tube 1 of the low concentration series, 200 μL of an 8.33 μg/mL
stock was added to duplicate rows of a 96-well microtiter plate. Serial
2-fold decremental dilutions were made using a BioMek FX robot
(Beckman Coulter Inc., Fullerton, California) with 10 of the remaining
11 wells, each of which contained 100 μL of the appropriate solvent/
diluent. Row 12 contained solvent/diluent only and served as the
control. Daughter plates were spotted (3.0 μL/well) from the stock
plates listed above using the BioMek FX robot, yielding the final
concentration of drug as described above. Tray Inoculation: Organisms
were inoculated (100 μL volumes) using the BioMek FX robot. The
final number of bacterial cells per well was approximately 2.5 × 104
CFU/mL. The inoculated trays were placed in stacks of no more than
4 and covered with an empty tray. The trays were incubated for 16−
20 h at 35 °C in an ambient air incubator. MIC Test Results: After
inoculation and incubation, the degree of bacterial growth was
estimated visually with the aid of a Test Reading Mirror (Dynex
Technologies 220−16) in a darkened room with a single light shining
directly through the top of the microbroth tray. The MIC was the
lowest concentration of drug which prevented macroscopically visible
growth under the conditions of the test. Determining the MIC50 and the
MIC90: After all the MICs for each subset of strains were read, the
values were sorted from highest to lowest and the MICrange, MIC50,
and MIC90 were calculated.
Table 5. Rat Pharmacokinetic Data
dose
(mg/kg)
Cl
Vdss
(L/kg)
AUC
T1/2
compd
(mL/min/kg)
(μg·h/mL)
(h)
1a
1l
10
10
10
10
58
53
65
15
2.7
1.4
2.6
0.4
2.8
3.1
3.2
0.6
1.4
0.7
1v
1w
2.5
10.9
low clearance observed for the corresponding carboxylic acid
1w (15 mL/min/kg).
In general, compounds in this series exhibit significant
binding to human plasma protein (e.g., compound 1a, HFu =
2%). A number of literature references suggest that a drug
discovery strategy to lower human projected dose based on
reducing plasma protein binding may be flawed, with the
possible exception being programs focused on high clearance
series with a need for iv drug delivery.17 The compounds
described herein certainly exhibit moderately high clearance in
rat, and iv delivery is a must-have for therapies targeting serious
Gram-negative infections. Therefore, reducing protein binding
became a goal for this medicinal chemistry program and design
strategies to address this issue will be discussed in a subsequent
publication.
To evaluate in vivo efficacy with compound 1a, a murine
acute systemic infection model was utilized. Mice were admini-
stered 1a (subcutaneous dose; vehicle: 40% β-cyclodextrin
in water) 0.5 and 4 h after an intraperitoneal challenge of
Pseudomonas aeruginosa strain UC12120, a penicillin-resistant,
quinolone-sensitive clinical isolate (1a MIC = 0.25 μg/mL). At
24 h postinfection, the animals were sacrificed and the bacterial
burden in the spleen was determined. The drug dose providing
50% bacterial burden reduction relative to untreated animals
(ED50) was determined to be 35 mg/kg. Impressively, at a
higher dose of 100 mg/kg, the spleen bacterial burden was
reduced below the limits of detection (6.2 log burden reduction
relative to control animals). An additional study was conducted
wherein animal survival at 96 h post infection was utilized as
the end point, thus providing a PD50 = 63.4 mg/kg. In vivo
dose fractionation studies conducted in a murine neutropenic
thigh Pseudomonas aeruginosa infection model suggest that
unbound AUC/MIC correlates best with efficacy for this drug
class. While pharmacokinetic data was not gathered during the
efficacy study described above for 1a, extrapolation from a previous
mouse pharmacokinetic study (50 mg/kg, subcutaneous) provides
an unbound AUC/MIC ∼ 8 for the PD50 study.
CONCLUSION
■
Here we introduce a novel series of LpxC inhibitors which
exhibit potent and selective enzyme inhibitory activity, broad
spectrum whole-cell activity, and in vivo efficacy in a murine
infection model. A key structural feature of this series is the
methylsulfone-containing headgroup which engages in a key
hydrogen-bonding interaction with the active site residue
Lys238. Subsequent studies focused on optimizing the physical
properties of this series to enable the identification of analogues
worthy of advanced development will be discussed in a future
publication.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental details for compounds 1a−1x, 3−7, 9−14, and
16−34. This material is available free of charge via the Internet
at http://pubs.acs.org. are available free of charge via the
Internet at http://pubs.acs.org. Compound 1a will be available
to purchase from Sigma-Aldrich, Tocris, as well as Toronto
Research Chemicals.
EXPERIMENTAL SECTION
LpxC Enzyme Assay. The assay measures the enzymatic
deacetylation of the synthetic lipid substrate UDP-3-O-(R-hydroxy-
decanoyl)-N-acetylglucosamine by mass spectrometry, using a
■
Accession Codes
The PDB code for compound 1t is 3U1Y.
922
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923

Journal of Medicinal Chemistry
Article
Lightle, S. Crystal structure of LpxC from Pseudomonas aeruginosa
complexed with the potent BB-78485 inhibitor. Protein Sci. 2008,
17 (3), 450−457. (d) Coggins, B. E.; McClerren, A. L.; Jiang, L.; Li,
X.; Rudolph, J.; Hindsgaul, O.; Raetz, C. R. H.; Zhou, P. Refined
Solution Structure of the LpxC-TU-514 Complex and pKa Analysis of
an Active Site Histidine: Insights into the Mechanism and Inhibitor
Design. Biochemistry (Moscow) 2005, 44 (4), 1114−1126.
(11) Anderson, G. M. III; Kollman, P. A.; Domelsmith, L. N.; Houk,
K. N. Methoxy group nonplanarity in o-dimethoxybenzenes. Simple
predictive models for conformations and rotational barriers in
alkoxyaromatics. J. Am. Chem. Soc. 1979, 101 (9), 2344−2352.
(12) Kalgutkar, A. S.; Obach, R. S.; Maurer, T. S. Mechanism-based
inactivation of cytochrome P450 enzymes: chemical mechanisms,
structure−activity relationships and relationship to clinical drug−drug
interactions and idiosyncratic adverse drug reactions. Curr. Drug
Metab. 2007, 8 (5), 407−447.
AUTHOR INFORMATION
Corresponding Author
*Phone: (860) 441-3522. E-mail: matthew.f.brown@pfizer.com.
■
ACKNOWLEDGMENTS
We thank David George, Lucinda Lamb, Deborah Gordon, and
Judith Hamel for conducting infection model studies.
■
REFERENCES
■
(1) Boucher Helen, W.; Talbot George, H.; Bradley John, S.;
Edwards John, E.; Gilbert, D.; Rice Louis, B.; Scheld, M.; Spellberg, B.;
Bartlett, J. Bad bugs, no drugs: no ESKAPE! An update from the
Infectious Diseases Society of America. Clin. Infect. Dis. 2009, 48 (1),
1−12.
(2) Devasahayam, G.; Scheld, W. M.; Hoffman, P. S. Newer
antibacterial drugs for a new century. Expert Opin. Invest. Drugs 2010,
19 (2), 215−234.
(3) Fischbach, M. A.; Walsh, C. T. Antibiotics for Emerging
Pathogens. Science 2009, 325 (5944), 1089−1093.
(4) Payne, D. J.; Gwynn, M. N.; Holmes, D. J.; Pompliano, D. L.
Drugs for bad bugs: confronting the challenges of antibacterial
discovery. Nature Rev. Drug Discovery 2007, 6 (1), 29−40.
(5) Cipolla, L.; Polissi, A.; Airoldi, C.; Gabrielli, L.; Merlo, S.;
Nicotra, F. New targets for antibacterial design: Kdo biosynthesis and
LPS machinery transport to the cell surface. Curr. Med. Chem. 2011, 18
(6), 830−852.
(6) (a) Vaara, M. Outer membrane permeability barrier to
azithromycin, clarithromycin, and roxithromycin in gram-negative
enteric bacteria. Antimicrob. Agents Chemother. 1993, 37 (2), 354−356.
(b) Vuorio, R.; Vaara, M. The lipid A biosynthesis mutation lpxA2 of
Escherichia coli results in drastic antibiotic supersusceptibility.
Antimicrob. Agents Chemother. 1992, 36 (4), 826−829.
(7) Young, K.; Silver, L. L.; Bramhill, D.; Cameron, P.; Eveland, S. S.;
Raetz, C. R. H.; Hyland, S. A.; Anderson, M. S. The envA
permeability/cell division gene of Escherichia coli encodes the second
enzyme of lipid A biosynthesis. UDP-3-O-(R-3-hydroxymyristoyl)-N-
acetylglucosamine deacetylase. J. Biol. Chem. 1995, 270 (51), 30384−
30391.
(8) Onishi, H. R.; Pelak, B. A.; Gerckens, L. S.; Silver, L. L.; Kahan, F.
M.; Chen, M.-H.; Patchett, A. A.; Galloway, S. M.; Hyland, S. A.; et al.
Antibacterial agents that inhibit lipid A biosynthesis. Science 1996, 274
(5289), 980−982.
(9) (a) Barb, A. W.; Zhou, P. Mechanism and inhibition of LpxC: an
essential zinc-dependent deacetylase of bacterial lipid A synthesis.
Curr. Pharm. Biotechnol. 2008, 9 (1), 9−15. (b) Cuny, G. D., A new
class of UDP-3-O-(R-3-hydroxymyristol)-N-acetylglucosamine deace-
tylase (LpxC) inhibitors for the treatment of Gram-negative infections:
PCT application WO 2008027466. Expert Opin. Ther. Pat. 2009,
19(6), 893−899; (c) Faruk Mansoor, U.; Vitharana, D.; Reddy, P. A.;
Daubaras, D. L.; McNicholas, P.; Orth, P.; Black, T.; Arshad Siddiqui,
M. Design and synthesis of potent Gram-negative specific LpxC
inhibitors. Bioorg. Med. Chem. Lett. 2011, 21 (4), 1155−1161.
(d) Liang, X.; Lee, C.-J.; Chen, X.; Chung, H. S.; Zeng, D.; Raetz,
C. R. H.; Li, Y.; Zhou, P.; Toone, E. J. Syntheses, structures and
antibiotic activities of LpxC inhibitors based on the diacetylene
scaffold. Bioorg. Med. Chem. 2011, 19 (2), 852−860. (e) Barb, A. W.;
Leavy, T. M.; Robins, L. I.; Guan, Z.; Six, D. A.; Zhou, P.; Bertozzi,
C. R.; Raetz, C. R. H. Uridine-based inhibitors as new leads for
antibiotics targeting Escherichia coli LpxC. Biochemistry (Moscow)
2009, 48 (14), 3068−3077.
(13) Lomovskaya, O.; Zgurskaya, H. I.; Totrov, M.; Watkins, W. J.
Waltzing transporters and “the dance macabre” between humans and
bacteria. Nature Rev. Drug Discovery 2007, 6 (1), 56−65.
(14) Chuanchuen, R.; Beinlich, K.; Hoang, T. T.; Becher, A.;
Karkhoff-Schweizer, R. R.; Schweizer, H. P. Cross-resistance between
triclosan and antibiotics in Pseudomonas aeruginosa is mediated by
multidrug efflux pumps: exposure of a susceptible mutant strain to
triclosan selects nfxB mutants overexpressing MexCD-OprJ. Anti-
microb. Agents Chemother. 2001, 45 (2), 428−432.
(15) Screen was performed by Cerep Corp. 15318 NE 95th St.
Redmond, WA 98052, USA; www.cerep.com.
(16) Dalvie, D.; Cosker, T.; Boyden, T.; Zhou, S.; Schroeder, C.;
Potchoiba, M. J. Metabolism distribution and excretion of a matrix
metalloproteinase-13 inhibitor, 4-[4-(4-fluorophenoxy)-benzenesul-
fonylamino]tetrahydropyran-4-carboxylic acid hydroxyamide (CP-
544439), in rats and dogs: assessment of the metabolic profile of
CP-544439 in plasma and urine of humans. Drug Metab. Dispos. 2008,
36 (9), 1869−1883.
(17) (a) Benet, L. Z.; Hoener, B.-A. Changes in plasma protein
binding have little clinical relevance. Clin. Pharmacol. Ther. 2002,
71 (3), 115−121. (b) Smith, D. A.; Di, L.; Kerns, E. H. The effect of
plasma protein binding on in vivo efficacy: misconceptions in drug
discovery. Nature Rev. Drug Discovery 2010, 9 (12), 929−939.
(18) Langsdorf, E. F.; Malikzay, A.; Lamarr, W. A.; Daubaras, D.;
Kravec, C.; Zhang, R.; Hart, R.; Monsma, F.; Black, T.; Ozbal, C. C.;
Miesel, L.; Lunn, C. A. Screening for antibacterial inhibitors of the
UDP-3-O-(R-3-hydroxymyristoyl)-N-acetylglucosamine deacetylase
(LpxC) using a high-throughput mass spectrometry assay. J. Biomol.
Screening 2010, 15 (1), 52−61.
(10) (a) Cole, K. E.; Gattis, S. G.; Angell, H. D.; Fierke, C. A.;
Christianson, D. W. Structure of the Metal-Dependent Deacetylase
LpxC from Yersinia enterocolitica Complexed with the Potent Inhi-
bitor CHIR-090. Biochemistry (Moscow) 2011, 50 (2), 258−265.
(b) Whittington, D. A.; Rusche, K. M.; Shin, H.; Fierke, C. A.;
Christianson, D. W. Crystal structure of LpxC, a zinc-dependent
deacetylase essential for endotoxin biosynthesis. Proc. Natl. Acad. Sci.
U.S.A. 2003, 100 (14), 8146−8150. (c) Mochalkin, I.; Knafels, J. D.;
923
dx.doi.org/10.1021/jm2014748 | J. Med. Chem. 2012, 55, 914−923