Discovery and Optimization of Phosphopantetheine Adenylyltransferase Inhibitors with Gram-Negative Antibacterial Activity

Article abstract of DOI:10.1021/acs.jmedchem.7b01861

In the preceding manuscript [Moreau et al. 2018, 10.1021/acs.jmedchem.7b01691] we described a successful fragment-based lead discovery (FBLD) strategy for discovery of bacterial phosphopantetheine adenylyltransferase inhibitors (PPAT, CoaD). Following several rounds of optimization two promising lead compounds were identified: triazolopyrimidinone 3 and 4-azabenzimidazole 4. Here we disclose our efforts to further optimize these two leads for on-target potency and Gram-negative cellular activity. Enabled by a robust X-ray crystallography system, our structure-based inhibitor design approach delivered compounds with biochemical potencies 4-5 orders of magnitude greater than their respective fragment starting points. Additional optimization was guided by observations on bacterial permeability and physicochemical properties, which ultimately led to the identification of PPAT inhibitors with cellular activity against wild-type E. coli.

Full text of DOI:10.1021/acs.jmedchem.7b01861

Subscriber access provided by UNIV OF SCIENCES PHILADELPHIA
Discovery and Optimization of Phosphopantetheine
Adenylyltransferase Inhibitors with Gram-Negative Antibacterial Activity
Colin K Skepper, Robert J Moreau, Brent A Appleton, Bret M Benton, Joseph E Drumm,
Brian Y Feng, Mei Geng, Andreas Lingel, Yipin Lu, Mulugeta Mamo, Wosenu Mergo, Mina
Mostafavi, Christopher M Rath, Micah Steffek, Kenneth T Takeoka, Kyoko Uehara, Lisha
Wang, Jun-Rong Wei, Lili Xie, Wenjian Xu, Qiong Zhang, Cindy Li, and Javier de Vicente
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01861 • Publication Date (Web): 18 Mar 2018
Downloaded from http://pubs.acs.org on March 18, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery Chemistry
Zhang, Qiong; Novartis Institutes for BioMedical Research Inc, Global
Discovery Chemistry
Li, Cindy; Novartis Institutes for BioMedical Research Inc, Global Discovery
Chemistry
de Vicente, Javier; Novartis Institutes for BioMedical Research Inc, Global
Discovery Chemistry
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Discovery and Optimization of Phosphopantetheine
Adenylyltransferase Inhibitors with GramꢀNegative
Antibacterial Activity
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Colin K. Skepper,* Robert J. Moreau, Brent A. Appleton, Bret M. Benton, Joseph E. Drumm III,
Brian Y. Feng, Mei Geng, Andreas Lingel, Yipin Lu, Mulugeta Mamo, Wosenu Mergo, Mina
Mostafavi, Christopher M. Rath, Micah Steffek, Kenneth T. Takeoka, Kyoko Uehara, Lisha
Wang, JunꢀRong Wei, Lili Xie, Wenjian Xu, Qiong Zhang, Cindy Li, Javier de Vicente
Novartis Institutes for Biomedical Research, 5300 Chiron Way, Emeryville, California 94608,
United States
KEYWORDS. FBS, FBDD, FBLD, PPAT, phosphopantetheine, adenylyltransferase, CoaD,
Gramꢀnegative, antibacterial, antibiotic.
ABSTRACT In the preceding manuscript we described a successful fragmentꢀbased lead
discovery (FBLD) strategy for discovery of bacterial phosphopantetheine adenylyltransferase
inhibitors (PPAT, CoaD). Following several rounds of optimization two promising lead
compounds were identified: triazolopyrimidinone 3 and 4ꢀazabenzimidazole 4. Here we disclose
our efforts to further optimize these two leads for onꢀtarget potency and Gramꢀnegative cellular
activity. Enabled by a robust Xꢀray crystallography system, our structureꢀbased inhibitor design
approach delivered compounds with biochemical potencies 4ꢀ5 orders of magnitude greater than
ACS Paragon Plus Environment
1

Page 3 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
their respective fragment starting points. Additional optimization was guided by observations on
bacterial permeability and physicochemical properties, which ultimately led to the identification
of PPAT inhibitors with cellular activity against wildꢀtype E. coli.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
INTRODUCTION
Antibiotics have been a foundational component of medical practice since the 1940s. The
steady spread of resistance, however, threatens to undermine the continued use of many
traditional antibiotic classes and may eventually erode the medical advances that these drugs
1
,2
have made possible.
This issue is compounded by the fact that the rate of discovery of new
3
antibiotics has been declining for decades. Even as the World Health Organization (WHO),
Centers for Disease Control and Prevention (CDC) and others sound the alarm over an
impending “postꢀantibiotic era”, the majority of drugs in clinical use today still belong to
4
structural classes that were discovered during the middle of the last century.
There is an especially urgent need for new antibiotics that target multidrugꢀresistant (MDR)
and pandrugꢀresistant (PDR) Gramꢀnegative bacteria, including those identified in the ESKAPE
family (i.e. Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and
Enterobacter spp.). Diverse resistance mechanisms have been described that negate the efficacy
5
of all classes of drugs used to treat infections caused by these pathogens, including colistin,
6
which is widely regarded as the antibiotic of last resort. Bacteria that possess multiple
7
resistance determinants can cause infections that are virtually untreatable.
Despite high unmet medical need and decades of research in the field, the discovery of new
agents to treat Gramꢀnegative infections is challenging. This is due, in part, to the fact that Gramꢀ
negative bacteria present significant barriers to entry for small molecule drugs. These barriers
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
include two cell membranes with opposing requirements for permeation and an array of efflux
8
pumps with broad substrate recognition.
It has been recognized that compounds with
physicochemical properties suitable for Gramꢀnegative cell entry are not well represented in
corporate screening decks. Thus, it has historically been challenging to translate hits from highꢀ
throughput screens against Gramꢀnegative bacterial targets into leads with wholeꢀcell activity
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
9
suitable for further development.
One successful strategy to mitigate the challenges associated with antibacterial drug discovery
is the modification of existing antibiotic scaffolds. While this approach has provided many of the
3
drugs in use today, addressing the longꢀterm need for effective antibiotics will require that it be
complemented by development of compounds that inhibit novel targets. One such target that has
received attention recently is phosphopantetheine adenylyltransferase (PPAT, also known as
CoaD). PPAT is a hexameric enzyme that catalyzes the penultimate step in coenzyme A (CoA)
biosynthesis. Coenzyme A is the universal acyl chain carrier required for many metabolic
1
0
processes including the citric acid cycle and fatty acid biosynthesis.
In bacteria, CoA is a
critical coꢀfactor in the biogenesis of membrane lipids, peptidoglycan, teichoic acids (in Gramꢀ
1
1,12
positive organisms) and Lipid A (in Gramꢀnegative organisms).
CoA is synthesized in five
1
3
steps starting from pantothenic acid (vitamin B5, Figure 1). PPAT catalyzes the fourth step in
the sequence in which an adenosine monophosphate unit derived from ATP is appended to 4’ꢀ
phosphopantetheine via a diphosphate ester linkage with the concomitant expulsion of inorganic
1
4
pyrophosphate. PPAT is essential for bacterial growth and exhibits high sequence homology
between species, but it shares relatively little sequence homology with the human ortholog,
1
5,16
making it an attractive target for antibacterial drug development.
ACS Paragon Plus Environment
3

Page 5 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Schematic represenation of coenzyme A biosynthesis.
1
7
Several small molecule inhibitors of bacterial PPAT have been described previously,
including two compounds from AstraZeneca (1, 2) that demonstrated in vitro and in vivo
1
8
inhibition of Gramꢀpositive bacterial growth. In the previous manuscript we detailed our own
19
efforts to identify inhibitors of Gramꢀnegative bacterial PPAT using an FBLD approach. All
fragment hits were found to bind at the 4’ꢀphosphopantetheine site of E.coli PPAT. After several
rounds of hitꢀtoꢀlead optimization we identified lead compounds 3 and 4, which exhibited
nanomolar IC values in our biochemical assay and displayed modest cellular activity against E.
5
0
coli ꢁtolC. Below we describe further optimization of this series to deliver picomolar inhibitors
of PPAT from Gramꢀnegative bacteria that demonstrate growth inhibitory activity against WT E.
coli.
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Structures of known bacterial PPAT inhibitors from AstraZeneca (1, 2), lead
compounds (3, 4) and fragment hit 5. Biochemical potency, ligand efficiency (LE) and
(c)logD7.4 shown for compounds 3ꢀ5.
RESULTS AND DISCUSSION
1
9
As described in the preceding manuscript triazolopyrimidinone 3 efficiently inhibited E. coli
PPAT and displayed modest cellular potency against the effluxꢀdeficient E. coli ꢁtolC mutant
strain (Figure 2). Compound 3 also exhibited physicochemical properties in line with
expectations for a potential Gramꢀnegative antibiotic. An Xꢀray coꢀcrystal structure with E. coli
PPAT revealed that 3 is anchored in the pantetheine pocket of the enzyme by five direct or
waterꢀmediated hydrogen bonds (Figure 3). Specifically, the triazolopyrimidinone core interacts
directly with the NH of M74 and the backbone carbonyl of D72, and through a water molecule
with the NH of A75 and the sidechain carbonyl of N106. The benzylic cyanomethyl forms an
additional backbone interaction with the NH of S39. In an effort to further improve potency, we
ACS Paragon Plus Environment
5

Page 7 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
sought to leverage the information gained from the 39 fragmentꢀbound Xꢀray coꢀcrystal
structures of E. coli PPAT that were obtained during our FBLD efforts. Of particular interest
was the binding mode of methoxytryptamine derivative 5, which was shown to partially overlap
with 3 but was distinct in that the M74 and L102 sidechains were displaced and a hydrogen bond
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
was established between the amide NH of the N106 sidechain and the methoxy group of the
2
ligand (Figure 3). In fact, ten other fragments were observed to have similar binding modes
2
0
suggesting that this region of the pantetheine pocket represented a binding “hot spot” that could
be exploited to improve the affinity of compound 3.
Figure 3. Left. Xꢀray crystal structure of compound 3 bound to PPAT. PPAT protein is
represented as yellow ribbon with key binding site residues displayed as yellow ballꢀandꢀstick
models. Compound 3 is displayed as an orange ballꢀandꢀstick model. The key binding site water
molecule is shown as a red sphere. Hydrogen bonds are displayed as light blue dotted lines.
Right. Overylay of the crystal structure of PPAT in complex with 3 and the crystal structure of
PPAT in complex with 5. PPAT in the PPATꢀ5 complex is represented as cyan ribbon with key
binding site residues displayed as cyan ballꢀandꢀstick models. Key binding site residues in the
PPATꢀ3 complex are shown as yellow ballꢀandꢀstick models. Compounds 3 and 5 are shown as
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ballꢀandꢀstick models with 3 in orange and 5 in white. The key binding site water molecule is
shown as a red sphere. Hydrogen bonds are displayed as light blue dotted lines.
Given the spatial relationship between triazolopyrimidinone 3 and tryptamine 5 in the binding
pocket, a series of compounds was designed and synthesized that merged the structural features
of the two inhibitors. We found that phenyl ether 7 inhibited enzymatic activity at a level
comparable to parent analog 6 and displayed weak cellular activity against E. coli ꢁtolC (Table
1). The Xꢀray coꢀcrystal structure of 7 with E. coli PPAT verified that the triazolopyrimidinone
core was able to maintain the network of hydrogen bonds observed for 3, and that the sidechains
of M74 and L102 had enveloped the anisole moiety (Figure 4). However, it was also clear that
the the interaction with N106 could be further optimized. We expected that a carbonylꢀbased
hydrogen bond acceptor would exhibit more favorable alignment relative to the side chain of
N106, and that reinstallation of the benzylic cyanomethyl would further enhance potency. This
was indeed the case as methyl benzoate 8 exhibited a 3ꢀfold improvement in enzymatic
inhibition and a 8ꢀfold improvement in cellular activity against E. coli ꢁtolC relative to parent
compound 3. We subsequently discovered that the distal phenyl ring of 8 could be replaced with
a saturated heterocycle to provide piperidine 9. This modification lowered the measured logD7.4
by nearly 1.4 units and improved the solubility >40ꢀfold while maintaining biochemical potency.
The potency of compounds 8 and 9 approached the lower limit of our enzymatic assay ([PPAT]
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
=
24 nM), and for such compounds we frequently turned to surface plasmon resonance (SPR) to
generate binding affinity data. In the case of 8 and 9 we did not observe a significant
discrepancy between biochemical IC50 and SPR K , however this was not true of all potent
D
compounds (vide infra).
ACS Paragon Plus Environment
7

Page 9 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. Xꢀray coꢀcrystal structures of 7 and 9 bound to PPAT from E. coli. PPAT is
represented as yellow ribbon with key binding site residues displayed as yellow ballꢀandꢀstick
models. Compounds are displayed as orange ballꢀandꢀstick models. The key binding site water
molecule is shown as a red sphere. Hydrogen bonds are displayed as light blue dotted lines.
Table 1. Biochemical, SPR and MIC data of triazolopyrimidinones 4, 6-9.
H
N
N
H
R1
N
N
N
O
R2
1
2
Cmpd
R
R
E. coli
E. coli
P.
E. coli
logD7.4
PPAT IC50 PPAT SPR aeruginosa
ꢁtolC MIC
(µg/mL)
(
nM)
K (nM)
D
PPAT IC50
nM)
(
3
6
7
CH
CH
CH
2
3
3
CN Cl
Cl
37
9.0
174
NT
49
16
1.2
1.1
NT
251
167
969
567
>128
64
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
8
9
CH
2
CN
CN
12
13
5.6
11
16
11
2
8
2.0
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
CH
0.64
2
NT = not tested.
Despite improvements in logD7.4 and biochemical potency, compounds bearing the
triazolopyrimidinone core continued to exhibit only modest MIC values against E. coli ꢁtolC.
2
1
The weakly acidic nature of the core heterocycle and high polarity of derivatives such as 9
tPSA = 137) suggested that poor inner membrane permeability may account for the lack of
(
cellular activity. We had previously observed that compound 4, which features a 4ꢀ
azabenzimidazole core, was capable of making interactions with the enzyme identical to those
19
made by triazolopyrimidinone 3. Furthermore, 4 exhibited improved cellular activity against E.
coli ꢁtolC relative to 3, possibly as a result of superior permeability. To probe the generality of
this observation, we synthesized and tested piperidine ether 10 (Table 2). We were pleased to
find that 4 and 10 compared favorably to 3 and 9, respectively; despite similar biochemical
potencies, the 4ꢀazabenzimidazoles displayed improved activity against E. coli ꢁtolC, albeit with
somewhat higher logD7.4 values. With these results in hand, we set out to optimize the moiety
between the terminal hydrogen bond acceptor (i.e. the carbamate) and the phenyl ring. This
effort led to the identification of several potent analogs, including methyleneꢀlinked piperidine
1
1. Compound 11 represents an example of a compound with potency that apparently exceeded
the lower limit of detection in our biochemical assay, as indicated by SPR (K = 75 pM,
D
ACS Paragon Plus Environment
9

Page 11 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
biochemical IC = 9300 pM). Thioether 12 also displayed subꢀnanomolar affinity for PPAT
5
0
(
SPR K = 280 pM). The Xꢀray coꢀcrystal structures of compounds 9 and 11 provide a possible
D
rationale for the remarkable binding affinity of the latter analogue relative to 10 (SPR K = 75
D
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
pM for 11 vs. 1900 pM for 10; no crystal structure was available for 10). As seen in Figure 5,
the core heterocycles of 9 and 11 occupy nearly identical positions and orientations. The
piperidine moiety of compound 11, however, adopts a bound conformation that is likely closer to
the ground state conformation than the corresponding moiety in 9. Specifically, the piperidine
aryloxy substituent of 9 adopts an axial orientation whereas the corresponding methylene
substituent in 11 enjoys an equatorial disposition. In addition, the ~90° torsion angle between
the phenyl ring, the methylene linker and the piperidine tertiary carbon of 11 is in line with the
preferred conformation for primary alkyl substituents on aryl rings. In contrast, primary and
secondary alkoxy substituents prefer to adopt a planar conformation (torsion angle ~0/180°) as
2
2
determined from an anlaysis of CSD and PDB structures. Finally, the methylene linker group
in 11 makes lipophilic interactions with nearby residues, in particular A37, F70 and G9. In
contrast, compound 10 projects an ether oxygen toward the same region of the binding pocket.
ACS Paragon Plus Environment
10

Journal of Medicinal Chemistry
Page 12 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Overlay of Xꢀray coꢀcrystal structures of 9 and 11 bound to PPAT from E. coli. PPAT
is represented as yellow ribbon with key binding site residues displayed as yellow ballꢀandꢀstick
models. Compounds are shown as ballꢀandꢀstick models with 11 in orange and 9 in cyan. The
key binding site water molecule is shown as a red sphere. Hydrogen bonds are displayed as light
blue dotted lines.
Significantly, both 11 and 12 displayed weak but measurable MIC values against WT E. coli
(128 and 64 µg/mL, respectively). This demonstration of WT Gramꢀnegative activity was a
major milestone in the program and encouraged further exploration. More polar analogs of 12,
such as sulfone 13 and sulfonylpiperizine 14, had comparable biochemical potency but reduced
cellular activity; piperazinone 15 was markedly less active. Truncation of the piperidine ring of
1
0 to a pyrrolidine provided diastereomers 16 and 17, with the preferred diastereomer (17)
having activity similar to 10, except for 16ꢀfold weaker cellular activity against E. coli ꢁtolC.
Replacing the linking oxygen atom with carbon provided pyrrolidines 18 and 19, whose
improved biochemical potencies were in line with previous observations in the piperidine series,
with 19 displaying a measurable MIC against WT E. coli (64 µg/mL).
ACS Paragon Plus Environment
11

Page 13 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2. Biochemical, SPR and MIC data for azabenzimidazoles 10-19.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
Cmpd
R
E. coli
E. coli
P.
E. coli
E. coli
logD7.4
PPAT
PPAT
aeruginosa
PPAT IC50
(nM)
ꢁtolC MIC WT MIC
IC (nM) SPR KD
(µg/mL)
4
(µg/mL)
>128
5
0
(
nM)
4
Cl
56
10
64
3.3
3.0
1
0
11
1.9
9.2
0.25
0.125
0.06
2
>128
11
12
9.3
8.2
9.3
0.075
0.28
NT
2.5
2.7
2.4
128
2.8
3.7
1.9
64
O
1
3
>128
O
S
N
O
OMe
ACS Paragon Plus Environment
12

Journal of Medicinal Chemistry
Page 14 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
O
1
1
4
5
26
1.1
NT
13
32
>128
NT
2.0
1.4
O
S
N
N
O
OMe
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
231
488
NT
1
1
1
1
6
7
8
9
536
12
NT
7.7
NT
NT
502
9.3
8.3
3.8
>128
>128
>128
>128
64
2.8
2.7
3.1
NT
4
46
1
20
0.125
NT = not tested.
Additional insight into the SAR of this series was provided by the use of a novel, mass
2
3
spectrometry (MS)ꢀguided metabolomics assay. This assay, described in detail elsewhere, was
designed to measure the concentration of either free CoAꢀSH or acetyl CoA in bacterial cells in
ACS Paragon Plus Environment
13

Page 15 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the presence or absence of a PPAT inhibitor. The reported IC values indicate the concentration
5
0
of compound required to reduce the level of a given CoA metabolite by 50% relative to untreated
control cells. Importantly, this assay allowed us to assess target engagement in effluxꢀ
competent, WT Gramꢀnegative cells at concentrations well below those required to observe an
MIC. Table 3 shows CoAꢀSH metabolite data for compounds 3, 4, 9 and 10.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Compound 4 proved to be slightly more potent than 3 in the E. coli ꢁtolC metabolite assay.
However, in WT E. coli 4 was at least 40ꢀfold more potent than 3 (IC50 = 4.76 µM for 4 vs.
>200 µM for 3). These data are consistent with the hypothesis that the triazolopyrimidinone core
suffers from generally poor permeability as a consequence of slow permeation, efflux or (more
likely) both. This trend continued for compounds 9 and 10, where the azabenzimidazole analog
(10) was approximately 20ꢀfold more potent than 9 as measured by the metabolite assay in both
WT and effluxꢀdeficient E. coli.
Comparing the data for compounds 4 and 10, however, illustrates one of the challenges of
Gramꢀnegative antibacterial drug discovery. As discussed above, compound 10 was designed to
take advantage of a binding “hot spot” in the PPAT active site by engaging in a hydrogenꢀ
bonding interaction with the side chain of N106. The success of this design principle was
manifested by a 5ꢀfold improvement in biochemical potency for 10 compared to 4. This
translated to a 15ꢀfold improvement in metabolite inhibition and a 16ꢀfold improvement in MIC
against E. coli ꢁtolC. However, 10 was 2–3ꢀfold less potent than 4 in the metabolite assay
against WT E. coli. Put another way, the E. coli WT/ꢁtolC metabolite IC50 ratio is ~13 for
compound 4, compared to ~500 for compound 10. Thus, while the intrinsic potency of
compound 4 was improved by elaboration to 10, this came at the cost of significantly increased
efflux in WT E. coli cells. Although further optimization of 10 provided compounds with onꢀ
ACS Paragon Plus Environment
14

Journal of Medicinal Chemistry
Page 16 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
scale MICs against WT E. coli, such as 11, 12 and 19, a very large efflux shift was still evident.
The identification of additional structural features that would improve both biochemical potency
and Gramꢀnegative cellular permeability and/or efflux therefore became a priority for the
program.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 3. Comparison of biochemical IC , metabolite IC and MIC data for 3, 4, 9 and 10.
5
0
50
E.
PPAT IC50 CoAꢀSH
nM) metabolite
coli E. coli ꢁtolC E. coli WT E. coli
E. coli WT
MIC
(µg/mL)
CoAꢀSH
ꢁtolC MIC
(
metabolite
(µg/mL)
IC (µM)
IC (µM)
50
50
3
7
0.59
>200
16
>128
3
4
5
1
6
3
0.37
0.45
4.76
4
8
>128
>128
>200
9
ACS Paragon Plus Environment
15

Page 17 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
0.024
11.9
0.25
>128
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
0
Inspection of the Xꢀray coꢀcrystal structure of 11 overlaid with that of dephosphoꢀCoA
revealed that the tertiary carbon atom of the piperidine ring of 11 offered a vector for growing
toward the polar portion of the active site that accommodates the phosphate ester linkage of the
enzymatic reaction product (Figure 6). We hoped that growing toward this pocket would result
in formation of additional polar interactions with the protein, or even ATP, thereby providing a
boost in potency without increasing (or perhaps even reducing) lipophilicity. A number of
compounds were prepared to test this hypothesis (Table 4). Small, polar substituents branching
from the tertiary piperidine carbon of 11 were well tolerated (20-23) and provided PPAT
inhibitors with potency comparable to the parent compound. In addition, larger substituents
including aliphatic alcohols, acids and nitriles (24-27) also provided potent compounds (many
with pM K values), but failed to deliver a compound with improved activity over 11 in cellular
D
assays. Further efforts were met with a similar lack of success and this strategy was
deprioritized.
ACS Paragon Plus Environment
16

Journal of Medicinal Chemistry
Page 18 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. Xꢀray coꢀcrystal structure of compound 11 bound to PPAT overlaid with the bound
conformation of dephosphoꢀCoA in it’s complex crystal structure with PPAT (pdb code: 1b6t).
PPAT in the complex with 11 is represented as yellow ribbon with key binding site residues
displayed as yellow ballꢀandꢀstick models. Ligands are shown as ballꢀandꢀstick models with 11
in orange and dephosphoꢀCoA in cyan. The key binding site water molecule is shown as a red
sphere. Hydrogen bonds are displayed as light blue dotted lines.
ACS Paragon Plus Environment
17

Page 19 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 4. Biochemical, SPR and MIC data of imidazolopyridines 20-27.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
Cmpd
R
E. coli
PPAT
E. coli
PPAT
P.
E. coli
ꢁtolC
MIC
E. coli
WT
MIC
logD7.4
aeruginosa
PPAT IC50
IC50
SPR KD
(
nM)
(nM)
(nM)
(µg/mL)
(µg/mL)
2
2
2
0
1
2
10
25
13
NT
8.2
9.7
8.0
2
>128
>128
>128
2.1
NT
1
NT
0.02
0.25
16
2
3
4.9
8.8
3.2
0.49
NT
8.1
10
0.25
0.25
0.5
>128
>128
>128
3.5
NT
NT
24
2
5
0.85
8.2
2
2
6
7
5.1
7.5
0.54
2.0
13
24
0.5
32
>128
>128
NT
NT
NT = not tested.
In parallel with our survey of the phosphate ester binding site, we also explored a distinct
hydrophobic pocket defined by the sidechains of Y98, Q101, L102 and L131 that was first
identified from an Xꢀray coꢀcrystal structure with HTS hit 28 (Figure 7).
The
ACS Paragon Plus Environment
18

Journal of Medicinal Chemistry
Page 20 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
triazolopyrimidinone core of compound 3 provided a convenient vector for growing toward this
pocket with more advanced inhibitors, and this led to the series of triazolopyrimidines
summarized in Table 5.
Even without the key nitrile substituent common to the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
triazolopyrimidinone and azabenzimidazole series, initial compounds such as diarylamines 29
and 30 displayed strong inhibition of PPAT and potent inhibition of the growth of E. coli ꢁtolC.
In the context of Nꢀacetyl indoline 30, introduction of the aforementioned nitrile (31) provided at
24
least a 2ꢀfold improvement in biochemical and cellular potency while reducing logD . Aside
7
.4
from the expected array of hydrogen bonds between the core scaffold and protein, much of the
binding affinity for these inhibitors is derived from hydrophobic interactions between the
aromatic heterocycle appended to the triazolopyrimidine core and the sidechains of L102, L131,
25
Q101, as well as πꢀπ stacking with Y98. Expansion of the SAR revealed a preference for a
twoꢀcarbon linker between the C7 amine and a pendant aromatic group; phenethylamine 32
displayed good activity that could be further enhanced by rigidifying the linker as a transꢀ
substituted cyclopropane (33). Replacement of the phenyl ring of 32 with aromatic heterocycles
was well tolerated and led to the identification of more polar analogues such as pyrazole 34,
which was shown by Xꢀray crystallography to bind in the Y98 pocket and engage in a hydrogen
bond with the sidechain carbonyl of Q101. While installation of the benzylic cyanomethyl again
led to a marked improvement in activity (35), activity against WT E. coli remained elusive.
ACS Paragon Plus Environment
19

Page 21 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
N
N
N
N
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Cl
NH
28
E. coli PPAT IC50 = 4.4 ꢀM
LE = 0.28, LogD7.4 >5
Figure 7. Xꢀray coꢀcrystal structures of 28 bound to PPAT from E. coli. PPAT Protein is
represented as yellow ribbon with key binding site residues displayed as grey ballꢀandꢀstick
models. Ligand is shown as an organge ballꢀandꢀstick model. The key binding site water
molecule is shown as a red sphere. Hydrogen bonds are displayed as light blue dotted lines.
Table 5. Biochemical, SPR and MIC data of triazolopyrimidines 29-39.
N
N
H
R1
N
N
N
R4
Cl
4
1
Cmpd
R
R
E. coli E. coli P.
PPAT
IC50
E. coli E. coli logD7.4
PPAT
SPR KD osa
(nM)
aerugin ꢁtolC
WT
MIC
MIC
(
nM)
PPAT
IC50
(µg/mL) (µg/mL)
(nM)
29
CH
3
14
8.8
29
1
>128
4.4
ACS Paragon Plus Environment
20

Journal of Medicinal Chemistry
Page 22 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
3
3
0
1
2
CH
CH
CH
3
2
3
7.8
CN 4.0
48
1.2
NT
NT
6.8
4.3
151
0.125
0.06
4
>128
>128
>128
4.6
3.9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
>4.8
3
3
3
4
CH
3
3
16
36
6.1
5.9
34
2
2
>128
>128
4.9
2.9
CH
228
3
3
5
6
CH
2
CN 5.4
1.1
6.1
0.5
>128
3.8
CH
CH
CH
307
CN 21
270
NT
3.9
NT
218
9.7
32
>128
>128
>128
3.3
2.4
NT
3
2
3
37
4
3
8
235
>128
ACS Paragon Plus Environment
21

Page 23 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3
9
CH
2
CN 14
21
12
4
>128
2.8
NT = not tested.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The aforementioned lipophilicityꢀdriven binding resulted in inefficient inhibitors (LipE ~1.5ꢀ2)
and it became imperative to discover new groups at C7 that could not only modulate the
physicochemical properties of our inhibitors, but also address our goal of identifying compounds
with reduced efflux and/or improved permeability in WT Gramꢀnegative bacteria. We began by
introducing small, weakly basic amines (36-39) and found that compounds lacking the nitrile
substituent (36 and 38) were relatively weak inhibitors of PPAT; however, upon introduction of
the nitrile (37 and 39) a dramatic boost in potency was observed (Table 5). Xꢀray coꢀcrystal
structures showed that the Y98/Q101/L102 hydrophobic pocket had collapsed, as expected, and
suggested that the terminal nitrogen of piperazine 36 was engaged in a hydrogen bond with the
Y98 phenol. Although these compounds had only modest MIC values against E. coli ꢁtolC,
2
3
consideration of metabolite inhibition data proved instructive and encouraging (Table 6).
While 31 demonstrated relatively potent reduction of CoAꢀSH levels in ꢁtolC and WT E. coli
consistent with its biochemical potency and E. coli ꢁtolC MIC), it suffered from a high logD7.4
(
and negligible solubility. The pyrazole analog 35 was somewhat improved in this regard, but
piperazine 37 displayed an intriguing profile. Despite modest intrinsic potency, 37 demonstrated
robust inhibition of CoAꢀSH formation in the metabolite assay against WT E. coli cells, as well
as a more favorable logD7.4 and much improved solubility. Clearly, this warranted further
exploration.
ACS Paragon Plus Environment
22

Journal of Medicinal Chemistry
Page 24 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 6. Biochemical IC , metabolite IC , MIC, logD and solubility data for 31, 35, and 37.
5
0
50
7.4
E. coli
PPAT
IC50
E. coli
ꢁtolC
CoAꢀSH
E. coli
WT CoAꢀ ꢁtolC
SH MIC
E. coli
E. coli
WT
MIC
logD7.4
Sol.
(PBS)
(µg/mL)
(
nM)
metabolite metabolite (µg/mL) (µg/mL)
IC (µM) IC (µM)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
50
4.0
5.4
21
<0.006
0.012
0.097
2.4
0.06
0.5
4
>128
>128
>128
3.9
<0.6
23.8
>729
3
1
5
74
41
3.8
2.4
3
37
In an attempt to capitalize on these new discoveries, we hypothesized that the combination of a
weakly basic amine with a substituent capable of engaging in a πꢀπ stacking interaction with
Y98 would improve both Gramꢀnegative bacterial permeation and intrinsic potency, respectively.
Guided by the wealth of structural information we had accumulated, we synthesized S
benzylamine 40 and were gratified to find that it prevented the growth of WT E. coli at
concentrations as low as 32 µg/mL (Table 7). Further, the poor inhibitory activity of
diastereomer 41 and alcohol 42, particularly in cellular assays, helped support our hypothesis
that an appropriately substituted amine could increase potency and confer improved permeability
ACS Paragon Plus Environment
23

Page 25 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
against WT E. coli. Closeꢀin SAR showed that substitution of the amine or the linker ablated
both biochemical and cellular potency (data not shown), but thiophenes (43-44) and fluorophenyl
analogs (45-47) were well tolerated and expanded the number of compounds with activity
against WT E. coli. Unfortunately, replacement of the phenyl ring with the more polar pyridine
ring (48-50) proved deleterious to cellular activity. The data for the fluorophenyl and pyridyl
series of compounds were suggestive of improved cellular activity with proximity between the
hydrogen bond acceptor and the amine. Our hypothesis was that this arrangement might allow
for an internal hydrogen bond that could improve innerꢀmembrane permeability. As a result, we
synthesized oꢀmethoxyphenyl compound 51. This analog was the most potent compound of the
series and it bound to the enzyme in a manner consistent with previous observations (Figure 8).
Unfortunately, further optimization of this example was unsuccessful, as were parallel efforts to
reduce the lipophilicity, such as by replacing the phenyl ring with a cyclopropane (52) or 5ꢀ
membered heteroaromatic groups (53-54).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 7. Biochemical, SPR and MIC data of triazolopyrimidines 40-54.
4
Cmpd
R
E. coli E. coli P.
E.
aeruginosa ꢁtolC
SPR KD PPAT IC50 MIC
coli E.
coli logD7.4
PPAT
IC50
PPAT
WT MIC
(µg/mL)
32
(
nM)
(nM)
(nM)
(µg/mL)
4
0
6.6
ND
4.1
0.25
3.4
ACS Paragon Plus Environment
24

Journal of Medicinal Chemistry
Page 26 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
4
4
4
1
2
3
4
46
ND
5.5
3.3
39
13
4.5
8
>128
>128
64
3.2
3.6
3.0
7.0
6.0
0.25
1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4.8
4.8
2.0
6.7
5.2
8.0
0.25
0.5
64
64
3.3
NT
45
4
6
3.7
2.9
5.4
0.25
64
3.7
4
4
7
8
3.4
11
0.46
6.7
5.0
14
0.125
64
3.2
2.2
4
>128
ACS Paragon Plus Environment
25

Page 27 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
4
9
15
4.8
22
4
>128
2.3
5
5
0
1
11
6.5
15
2
>128
32
1.9
3.1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3.9
0.83
7.0
0.25
5
5
2
3
43
12
25
9.9
15
8
1
>128
>128
1.3
NT
ND
5
4
7.7
ND
10
4
>128
NT
NT = not tested.
ACS Paragon Plus Environment
26

Journal of Medicinal Chemistry
Page 28 of 74
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Xꢀray coꢀcrystal structure of 51 bound to PPAT from E. coli. PPAT protein is
represented as yellow ribbon with key binding site residues displayed as grey ballꢀandꢀstick
models. Ligand is shown as an organge ballꢀandꢀstick model. The key binding site water
molecule is shown as a red sphere. Hydrogen bonds are displayed as light blue dotted lines.
In the preceding manuscript, we described a positive correlation between onꢀtarget potency
and electrostatic potential at the core heterocyclic nitrogen atoms that make direct and waterꢀ
19
mediated hydrogen bonds with the protein.
In addition, it was apparent that the
azabenzimidazole core was associated with improved Gramꢀnegative permeability. Combining
these observations with information gleaned from the chemical series described above, we
targeted compound 55, which features a piperazine ring appended to the azabenzimidazole core.
We were pleased to find that this relatively compact molecule exhibited onꢀscale activity against
WT E. coli comparable to 51 but with reduced molecular weight and lipophilicity. The in vitro
profile of 55 is presented in Table 8. Remarkably, although 55 is 10ꢀfold less potent than 51 in
the biochemical assay and by SPR, 55 is only 2ꢀfold less potent against WT E. coli. Given these
promising data, we made a final attempt to improve the activity by hybridizing 10 and 55 to
ACS Paragon Plus Environment
27

Page 29 of 74
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
provide analog 56. We were disappointed to find that the combination proved to be nonꢀadditive
and the overall profile was similar to that of 10. While it was recognized that an analog of
compound 51 featuring the azabenzimidazole core would be an attractive target, cumulative
knowledge acquired on the project to date suggested there were limited avenues for further
optimization of such a compound.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Efflux pumps in Gramꢀnegative bacteria present a notorious barrierꢀtoꢀentry for small
8
molecules. In general, our attempts to identify compounds with reduced susceptibility to efflux
found limited success, suggesting that recognition by pumps might be an inherent feature of the
general scaffold. Over the course of the program we examined in excess of 20 different core
structures in attempts to find alternative chemical matter with an improved profile. Only very
close analogues retained activity against PPAT (e.g. 4ꢀazabenzothiazole, compound 40 in the
1
9
preceding manuscript ), and these compounds did not offer an advantage over our lead
scaffolds. Compounds that featured more drastic changes to the core heterocycle usually
suffered from a precipitous loss of onꢀtarget potency.
The challenge of overcoming efflux is due in part to the presence of multiple redundant
pumps in the Gramꢀnegative cell. The problem is exacerbated for cytoplasmic targets such as
PPAT, as small molecule inhibitors must cross both the inner and outer membranes which
impede the passage of hydrophilic and lipophilic molecules respectively. Despite a concerted
effort from many research groups, it has proven difficult to derive a set of broadly applicable
“rules” based on structure or physicochemical properties that could be used to guide optimization
8,9
of efflux and permeability.
The data in Table 8 indicates that efflux is a somewhat more
significant barrier to permeability for our compounds than influx through the outer membrane.
This is manifested by MIC values against E. coli imp4213 (an outerꢀmembraneꢀdeficient strain)
ACS Paragon Plus Environment
28