Optimization and Mechanistic Characterization of Pyridopyrimidine Inhibitors of Bacterial Biotin Carboxylase

Article abstract of DOI:10.1021/acs.jmedchem.9b00625

A major challenge for new antibiotic discovery is predicting the physicochemical properties that enable small molecules to permeate Gram-negative bacterial membranes. We have applied physicochemical lessons from previous work to redesign and improve the antibacterial potency of pyridopyrimidine inhibitors of biotin carboxylase (BC) by up to 64-fold and 16-fold against Escherichia coli and Pseudomonas aeruginosa, respectively. Antibacterial and enzyme potency assessments in the presence of an outer membrane-permeabilizing agent or in efflux-compromised strains indicate that penetration and efflux properties of many redesigned BC inhibitors could be improved to various extents. Spontaneous resistance to the improved pyridopyrimidine inhibitors in P. aeruginosa occurs at very low frequencies between 10^-8 and 10^-9. However, resistant isolates had alarmingly high minimum inhibitory concentration shifts (16- to >128-fold) compared to the parent strain. Whole-genome sequencing of resistant isolates revealed that either BC target mutations or efflux pump overexpression can lead to the development of high-level resistance.

Full text of DOI:10.1021/acs.jmedchem.9b00625

Subscriber access provided by UNIV OF SOUTHERN INDIANA
Article
Optimization and Mechanistic Characterization of
Pyridopyrimidine Inhibitors of Bacterial Biotin Carboxylase
Logan Dane Andrews, Timothy R. Kane, Paola Dozzo, Cat M. Haglund, Darin J. Hilderbrandt, Martin
S. Linsell, Timothy Machajewski, Glen McEnroe, Alisa W. Serio, Kenneth B. Wlasichuk, David B Neau,
Svetlana Pakhomova, Grover L Waldrop, Marc Sharp, Joe Pogliano, Ryan Cirz, and Frederick Cohen
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00625 • Publication Date (Web): 15 Jul 2019
Downloaded from pubs.acs.org on July 16, 2019
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 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
TITLE: Optimization and Mechanistic Characterization of Pyridopyrimidine Inhibitors of
Bacterial Biotin Carboxylase
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
*a
a
a
a
AUTHORS: Logan D. Andrews , Timothy R. Kane , Paola Dozzo , Cat M. Haglund , Darin J.
a
a
a
a
a
Hilderbrandt , Martin S. Linsell , Timothy Machajewski , Glen McEnroe , Alisa W. Serio ,
a
b
b
b
Kenneth B. Wlasichuk , David B. Neau , Svetlana Pakhomova , Grover L. Waldrop , Marc
c
c,d
*a
a
Sharp , Joe Pogliano , Ryan T. Cirz , Frederick Cohen
aFormer employee of Achaogen Inc., 1 Tower Place, Suite 400, South San Francisco, CA 94080,
United States
bDepartment of Biological Sciences, Louisiana State University, Baton Rouge, LA 70803,
United States
cLinnaeus Bioscience Inc. 3210 Merryfield Row, San Diego, CA 92121, United States
dUniversity of California, San Diego 9500 Gilman Drive, La Jolla, 92093, United States
ABSTRACT
A major challenge for new antibiotic discovery is predicting the physicochemical properties that
enable small molecules to permeate Gram-negative bacterial membranes. We have applied
physicochemical lessons from previous work to redesign and improve the antibacterial potency of
pyridopyrimidine inhibitors of biotin carboxylase (BC) by up to 64-fold and 16-fold against E. coli
and P. aeruginosa, respectively. Antibacterial and enzyme potency assessments in the presence of
an outer membrane permeabilizing agent or in efflux-compromised strains indicate that penetration
and efflux properties of many redesigned BC inhibitors could be improved to various extents.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 61
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
Spontaneous resistance to the improved pyridopyrimidine inhibitors in P. aeruginosa occurs at
-8
-9
very low frequencies between 10 to 10 . However, resistant isolates had alarmingly high MIC
shifts (16 to >128-fold) compared to the parent strain. Whole genome sequencing of resistant
isolates revealed that either BC target mutations or efflux pump overexpression can lead to the
development of high-level resistance.
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
ACS Paragon Plus Environment

Page 3 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
The spread of multi-drug resistant (MDR) bacteria presents a threat to our healthcare
system. An analysis from the Centers for Disease Control indicates that approximately 23,000
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
people die annually in the US from these pathogens, and more recent reports suggest this estimate
2
,3
is conservative. In the future, routine life-saving procedures such as the use of coronary stents,
anti-cancer chemotherapy, and organ transplants could carry the risk of life-threating bacterial
infections. The situation is particularly alarming for MDR infections caused by the so-called
Gram-negative “ESKAPE” pathogens - E. coli, K. pneumoniae, A. baumannii, and P. aeruginosa.
This point is illustrated by the recent discovery of a strain of K. pneumoniae from a US ICU patient
4
that is resistant to 26 antibiotics. To address this threat from MDR pathogens, a variety of new
antibiotic discovery approaches are warranted including optimization of previously discovered
compounds with existing, although sometimes low levels, of antibacterial potency.
One such previously discovered compound that serves as an enticing starting point for
further optimization is the well-characterized biotin carboxylase (BC) inhibitor 1 that was
identified in a high-throughput screen against a membrane-compromised, efflux pump-deficient
5
strain of E. coli (tolC, imp). BC is a member of the fatty acid biosynthesis pathway (Figure S1)
and has attracted attention over the past decade as a promising target for the development of novel,
broad-spectrum antibiotics for Gram-negative bacterial infections.6-11 The first committed step in
this pathway is catalyzed by an essential cytoplasmic acetyl-CoA carboxylase multienzyme
1
2
complex (see also Supporting Information Text S1), and inhibitors of the BC active site of this
complex have shown antibacterial activity against several Gram-negative species.5, 13-17 The BC
active site is highly conserved across a number of bacterial pathogens, including P. aeruginosa, E.
1
8
coli, and H. influenzae. While 1 lacks the potency to become a therapeutic agent, it is a promising
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 61
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
starting point for medicinal chemistry efforts because of its selective on-target activity, low
molecular weight, and available co-crystal structures with BC that enable structure-based
molecular design.
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
Inhibitors of BC must pass through both the outer and inner membranes of the Gram-
negative bacteria to reach this cytoplasmic-residing enzyme. The outer membrane bars the passage
of many exogenous small molecules. However, porins in the membrane allow some hydrophilic
molecules with a size and shape compatible with the diameter of the porin channels to diffuse into
1
9
the periplasm of the cell. Once in the periplasm, molecules must then pass through the
phospholipid inner membrane to reach cytoplasmic targets, a process favored by lower polarity.
Because the molecular properties of the outer and inner membranes differ, the two barriers act
orthogonally to greatly hinder the access of most potential inhibitors to their cytoplasmic target
sites. Even when a molecule successfully penetrates both membranes, efflux pumps that actively
remove foreign compounds from the cell further reduce the concentration of inhibitor reaching a
cytoplasmic target.20,21 These hurdles are major challenges to the development of small molecule
inhibitors designed to kill bacteria by acting within the cytoplasm.
To address the challenge of discovering inhibitors that can readily access cytoplasmic
targets, retrospective chemoinformatic approaches have been used to evaluate the physicochemical
properties of molecules that accumulate effectively in Gram-negative cells.22-24 A pioneering
analysis by O’Shea and Moser, showed that, in general, antibiotics with Gram-negative activity
are significantly more polar (as measured by relative polar surface area and cLogD values) and
have more H-bond acceptor and donor atoms compared to a reference set of drugs from other
2
5
therapeutic areas. For the subset of antibiotics that act against targets in the cytoplasm, the
physicochemical differences compared to the reference set were quite diverse, and less pronounced
4
ACS Paragon Plus Environment

Page 5 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
for some classes of antibiotics, but still trended toward having higher overall polarity.24,25
However, a more recent assessment of Gram-negative cell compound accumulation that examined
charge and polarity (cLogD) properties separately suggests that polarity alone is not predictive of
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
cell accumulation.26
To illustrate the physicochemical properties of cytoplasm-acting inhibitors that are
sufficient for antibacterial potency, we compared a set of exemplar Gram-negative antibiotics
(Table 1). These antibiotics have cLogD values ranging from -3.6 to 2.0 and all contain groups
that ionize near physiological pH (7.4).
Table 1. Examples of cytoplasm-targeted Gram-negative agents with
physicochemical properties of interest
MW
Charge
class
e
Compound
StructuOrec
(g/mol) cLogDd
O
F
OH
Levofloxacin
361
-2.0
Zwitterionic
N
N
N
O
NH2
O
N
Trimethoprim
290
253
323
1.1
-1.1
1.0
Basic
Acidic
Acidic
H2N
N
O
O
O
N
O
O
S
N
Sulfamethoxazole
Chloramphenicol
H
H2N
OH OH
HN
Cl
O2N
OH
Cl
O
O
O
HO
O
HO
NH2
OH
Tetracycline
Nitrofurantoin
ACHN-975a
444
238
369
-3.6
-0.27
2.0
Zwitterionic
Acidic
OH
N
O2N
O
N
N
O
O
N
H
HO
O
NH2
NH
HN
Zwitterionic
O
OH
H
N
N
N
S
N
N
N
Cl
Triusb
420
1.3
Basic
NH2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 61
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
Br
1
N
395
3.3
Neutral
Br
H2N
N
N
NH2
a
b
c
Ref. 27. Ref. 28. Acidic groups colored red; basic groups colored blue.
dcLogD at pH 7.4. Calculated using ACD/Percepta GALAS algorithm,
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
e
Advanced Chemistry Development, Inc. At physiological pH = 7.4.
Comparing the property values in Table 1 reveals that BC inhibitor 1 with a cLogD value
of 3.3 is significantly less polar than the proven antibacterials. In addition and perhaps most
importantly, 1 is the only compound in Table 1 that does not contain an ionizable group. These
comparisons suggest that the potency of 1 could be improved by increasing its polarity and
installing ionizable groups with pK values in the physiological pH range. Further support for this
a
approach was reported recently by the Hergenrother group, which showed that the addition of a
primary amine to a molecule with a pK in the physiological pH range can significantly enhance
a
cellular accumulation.26,29
Herein, we report a structure-guided approach that was used to modulate the
physicochemical properties of 1 to favor membrane penetration and efflux avoidance while
simultaneously maintaining or improving BC target affinity, resulting in a significant overall
improvement in antibacterial potency.
CHEMISTRY
Design Strategy. The previously determined structure of 1 bound to E. coli BC suggested two
vectors allowing for physicochemical optimization while also permitting the retention, or even
improvement of, BC target inhibition. The first vector points from the amine at the 7-position of
2
+
the pyridopyrimidine ring toward two conserved glutamic acid residues that coordinate Mg when
6
ACS Paragon Plus Environment

Page 7 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
substrate ATP is bound (Figure 1A). The nearest oxygen atoms of these glutamic acid residue side
chains are 4.9 and 6.1 Å away from the 7-position nitrogen atom of 1. We hypothesized that this
vector would provide an opportunity to place an ionizable amine group that would reduce the
cLogD value of the compound while also engaging the nearby glutamic acid residues via H-
bonding and/or electrostatic interactions. The second vector points off the dihalo-aromatic ring
through a small slot in the active site surface toward solvent (Figure 1B). We begin by focusing
our attention on the first vector at the 7-position of the aromatic ring system and subsequently
return to the second vector for further optimization. In the final phase of the campaign compounds
consisting of derivatizations at both vectors are presented.
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. Compound 1 (carbon, green; nitrogen, blue; bromine, dark red; all rendered as sticks)
co-crystallized with E. coli BC at 2.1 Å as reported in Ref. 5 (PDB code 2V58). (A) View of
compound 1 in the BC active site showing the 7-position vector that points toward two proximal
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 61
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
glutamate residues as illustrated by the two arrows. The distance from the 7-position nitrogen atom
and nearest oxygen atoms of Glu288 and Glu276 are 4.9 and 6.1 Å, respectively. (B) View of the
solvent accessible surface representation, in transparent grey, of E. coli BC co-crystallized with 1.
The second vector points off the dihalo-aromatic ring of 1 through a small slot in the active site
surface toward solvent. Images prepared using PyMOL.
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
Synthesis. Simple 2,7-diaminopyridopyrimidines such as 1 and 4a-e are readily synthesized in
one step from aryl acetonitriles as shown in Scheme 1.
Scheme 1. Synthesis of 6-aryl pyridopyrimidines
O
Y
X
a
CN
H
N
+
7
Y
H2N
N
3
NH2
X
2
H N
2
N
N
NH2
4a-e
2a-e
a: X = Y = Cl
b: X = H, Y = Ph
c: X = H, Y = 3-pyridyl
d: X =Cl, Y = 3-pyridyl
e: X = Cl, Y = 4-methoxy-3-pyridyl
Reagents and conditions: (a) Base, solvent see Experimental Section.
To install an amine at the 7-position, we developed a novel synthetic route that allowed for
differentiation of the 2- and 7- positions as illustrated in Scheme 2. The route begins with a
condensation between aryl acetonitrile 2a and pyrimidine 5, wherein the eventual amine at the 2-
position is masked as a thiomethyl group. Following condensation, the 7-position is converted to
pyridone 7 by nitration and hydrolysis. Chlorination with POCl provided 8 with a handle for
3
installation of various amines by simple S Ar displacement. The sulfide was converted to an amine
n
8
ACS Paragon Plus Environment

Page 9 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
by oxidation to the sulfoxide and displacement with ammonia to provide analogs 12a-b. If
required, final Boc deprotection was affected with TFA, to afford inhibitors 13c-h. This route
could also be modified to allow for late-stage diversification via Suzuki reaction, as illustrated in
Scheme 3, for the synthesis of compounds 14a–e.
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
Scheme 2. Synthesis of Compounds 12a, b, and 13c-h
X
O
b
For 6a
a
N
H
N
2a, 2f +
Cl
H N
N
N
Z
H N
N
5
SMe
2
2
6
6
a Y = Cl, Z = SMe
f Y = OiBu; Z = SMe
e
f
6
g Y = OiBu Z = S(O)Me
f Y = OiBu Z = NH2
4
Cl
Cl
d
c
N
N
Cl
Cl
HO
N
N
SMe
Cl
N
N
SMe
NH
9
a-h
7a
8b
See Table 2
Cl
g
N
Cl
13c-h
For 12c-h
N
N
N
X
10a-h X = SMe
1a-h X = S(O)Me
12a-h X = NH2
e
f
1
a
Reagents and conditions: (a) NaH, DMF 45%; (b) NaNO , trifluoroacetic acid, 90%; (c) POCl ,
2
3
DMF (cat), 95 °C, 98%; (d) 9a-h, MeCN, 1,4-dioxane, 80 °C; (e) mCPBA or Davies oxazirdine,
DCM; (f) NH , MeOH, THF, 65 °C; (g) Trifluoroacetic acid, DCM.
3
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 61
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
Scheme 3. Synthesis of Compounds 14a–d
Y
Y
c
b
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
a
N
N
2d, 2g
+
5
Cl
Cl
H N
HO
N
N
SMe
N
N
SMe
2
7d Y = 3-pyridyl
7
7
6
d Y = 3-pyridyl
6f Y = OiBu
g Y = Br
f Y = OiBu
g Y = Br
6
Y
Y
d
N
e
N
Cl
Cl
N
N
N
SMe
Cl
N
N
SMe
BocHN
NH
9h
BocHN
8d Y = 3-pyridyl
8f Y = OiBu
8g Y = Br
10i Y = 3-pyridyl
1
1
0j Y = OiBu
0k Y = Br
Y
Y
f
N
N
Cl
Cl
N
N
N
S(O)Me
N
N
N
NH2
BocHN
BocHN
11i Y = 3-pyridyl
1
1
1j Y = OiBu
1k Y = Br
1
2i Y = 3-pyridyl
12j Y = OiBu
12k Y = Br
h, g
g
14b Y = 4-pyridyl
4c Y = 5-fluoro-6-methoxy-3-pyridyl
4d Y = 2-methylthiazol-5-yl
1
1
4a Y = 3-pyridyl
4e Y = OiBu
1
1
Reagents and conditions: (a) NaH, DMF; (b) NaNO , trifluoroacetic acid; (c) POCl , DMF (cat),
2
3
9
5 °C; (d) 9h, MeCN, 1,4-dioxane, 80 °C; (e) mCPBA or Davies oxazirdine, DCM; (f) NH3,
MeOH, THF, 65 °C; (g) Trifluoroacetic acid, DCM; (h) R(BOH) , Pd(dppf)Cl , Na CO , dioxane,
2
2
2
3
water, 80 °C.
1
0
ACS Paragon Plus Environment

Page 11 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
RESULTS AND DISCUSSION
Structure-activity relationships for derivatives off the pyridopyrimidine 7-position.
Compound 4a was chosen as the parent molecule for this first campaign because it is identical to
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
except for the substitution of the dibromo ring with a dichloro ring, which results in a slightly
lower cLogD. In addition, the IC value measured for 4a is within 2-fold of the IC value
5
0
50
measured here for 1, indicating that the ring substitution does not have a large impact on BC target-
inhibition. As expected based on the potency reported for 1, compound 4a has limited
microbiological activity against wild type (WT) ESKAPE pathogens A. baumannii, E. coli, and K.
pneumoniae, and no activity against P. aeruginosa was detected (MIC >256g/mL) (Table 2).
To test the ability of 4a to penetrate the Gram-negative outer membrane, we conducted
MIC assays in the presence and absence of the permeabilizing agent polymyxin B nonpeptide
(PMBN). At the concentration used (8 g/mL), PMBN does not have an antibacterial effect but is
expected to render the outer membrane more permeable by interacting with and disrupting the
negatively charged lipopolysaccharide (LPS) molecules that otherwise maintain the outer
3
0
membrane order. When PMBN is included, the MIC value of compound 4a against E. coli is
reduced 32-fold from 32 to 1 g/mL. This significant enhancement of potency in the presence of
PMBN suggests that when the outer membrane is compromised, a greater proportion of 4a is able
to reach the BC active site in the cytoplasm. By contrast, ACHN-975 (Table 1), which also targets
a cytoplasmic enzyme but has physicochemical properties that may be more conducive to
membrane penetration, is only 2- to 4-fold more potent in the presence of PMBN suggesting that
a greater proportion of ACHN-975 readily crosses the outer membrane.
To test the contribution of efflux to the limited potency of 4a, an E. coli (tolC) mutant
was tested in the MIC assay. With this key efflux component deleted, the potency of 4a is
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 61
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
strengthened by 64-fold (32 to 0.5 g/mL), suggesting that WT E. coli uses a TolC-containing
efflux pump to dramatically reduce the intracellular concentration of 4a. Similar results were
observed when comparing the MIC values of 4a against WT and efflux-compromised strains of P.
aeruginosa (Table 2).
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
When PMBN was included with compound 4a the MIC for E. coli tolC strain is 0.25
g/mL, the lowest value observed for all of the strains and conditions tested. Because the efflux-
compromised and WT strains are expected to have the same levels of LPS in their outer
membranes, we initially reasoned that PMBN should reduce the MIC by the same proportion for
both strains. Instead, the presence of PMBN induced only a 2-fold reduction in MIC against the
efflux-compromised strain (0.5 to 0.25 g/mL), compared to a 32-fold reduction against the WT
strain (32 to 1 g/mL). This disconnect suggests that when the outer membrane and efflux
mechanisms are both compromised, other properties of the bacterial cell and inhibitor, such as the
BC enzyme affinity and/or the ability of the inner membrane to exclude the inhibitor, become
limiting to the observed MIC potency.
Structural elaboration of 4a at the 7-position of the pyridopyrimidine ring yielded
compounds 12a-b and 13c-h shown in Table 2. To test our structural expectation that groups at
this position could be readily accommodated by the BC enzyme, we measured the IC values
5
0
against purified P. aeruginosa BC. The results in Table 2 show that the IC values measured are
5
0
all similar or better compared to 4a, demonstrating that the 7-position modification does not disrupt
BC inhibition and in one case (13h) improves inhibition by nearly 10-fold.
1
2
ACS Paragon Plus Environment

Page 13 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2. MIC values, IC values, and physicochemical properties of interest for compounds elaborated at the 7-position
5
0
Cl
N
Cl
R
N
F
N
N
NH2
H
O
H2N
HO
HO
N
N
N
N
N
N
N
-R
H N
2
H2
N
H2N
H2
N
HO
H2
N
H2
N
H2N
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
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
H
Organisma
4a
256
32
12a
12b
>64
32
13c
13d
13e
13f
13g
13h
128
2
A. baumannii
E. coli
>256
64
>256
16
>256
32
>256
64
>256
16
>128
32
K. pneumoniae
P. aeruginosa
E. coli +PMBN
E. coli (ΔtolC)
E. coli (ΔtolC) +PMBN
64
256
>256
4
0.5
0.12
256
>64
>64
2
0.25
0.12
>64
256
>256
8
1
0.25
256
256
>256
4
2
1
>256
>256
8
1
0.25
>64
64
128
>128
4
1
0.5
64
8
>256
1
0.5
0.25
16
>256
2
0.5
0.12
64
>256
0.25
0.12
0.06
8
b
c
P. aeruginosa (ΔmexAB, etc.)
256
E. coli +/-PMBN
E. coli +/-efflux
32
64
16
128
4
16
128
2
2
8
8
64
4
8
32
8
32
8
16
16
16
E. coli (ΔtolC) +/-PMBN
P. aeruginosa +/-efflux
IC50 P. aeruginosa BC (nM)
2
4
2
4
2
2
>16
320
307
4.8
2.8
>1
-
>1
190
405
7.6
2.3
>1
440
407
7.9
1.6
-
>4
64
>2
>32
33
170
403
(3.2)
2.9
140
390
5.0
3.5
83
405
8.7
0.6
320
403
10.1
0.8
MW (g/mol)
389
9.6
-0.1
387
8.0
2.7
d
pK
a
clogD (pH 7.4)
aThe strains of each bacterial species were as follows: A. baumannii (wild type) ATCC 19606; E. coli (wild type) DM4100; K.
b
pneumoniae (wild type) ATCC 43816; P. aeruginosa (wild type) PAO1. The E. coli ΔtolC strain contains a targeted knockout of the
c
efflux pump component TolC. The P. aeruginosa ΔmexAB, etc. strain contains targeted knockouts of efflux pumps MexAB-OprM,
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 61
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
MexCD-OprJ, and MexEF-OprN, and the efflux pump components MexXY are expected to be compromised by the absence of OprM.
d
Calculated pK values are in parentheses.
a
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
1
4
ACS Paragon Plus Environment

Page 15 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In addition to having similar BC affinity compared to 4a, the compounds 12a and 12b, also
have similar cLogD and pK values. Thus, it was expected that these compounds would perform
a
similarly to 4a in the MIC panel. The MIC data for 12a and 12b in Table 2 reflect these
expectations, with similarly low potency against WT strains and similarly strong MIC shifts
observed in the presence of PMBN and against the efflux mutants. These observations are
consistent with the premise that compounds with high cLogD values and no ionizable groups at
physiological pH values have difficulty entering cells and avoiding efflux.
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
To test the physicochemical hypothesis further, compound 12b was modified with the
addition of a nitrogen atom with a pK value near physiological pH, resulting in amine-containing
a
compound 13c. In addition to becoming ionizable, 13c also has a modest decrease in cLogD.
Importantly, the BC affinity of 13c remains relatively constant compared to the parent compound
so that any changes in potency can more likely be attributed to changes in the two important
physicochemical properties. Table 2 shows that 13c does not have a large increase in antibacterial
potency against WT strains. However, the presence of PMBN improves the potency of 13c by only
2
-fold, suggesting that it crosses the outer membrane more easily than compounds 4a and 12a-b,
which have MIC shifts of 16- to 32-fold in the presence of PMBN. Likewise, the 16-fold shift
observed with 13c against the E. coli efflux-compromised strain compared to WT strain is much
lower than the 64- to 128-fold shifts seen with compounds 4a and 12a-b. The lower fold-change
suggests that compound 13c also avoids efflux more readily than the neutral compounds with
higher cLogD values. Despite these apparent cellular penetration improvements and maintenance
of BC target affinity, the lack of corresponding improvement against WT strains suggests that 13c
has lost potency due to some other difference(s) compared to 4a and 12a-b. While currently
untested, one hypothesis is that 13c has a reduced ability to cross the Gram-negative inner
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 61
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
membrane relative to 4a and 12a-b. The direct measurement of the accumulation of compound
inside the cell could help test this hypothesis.
Further derivatives of 13c were designed to increase the pK and lower the cLogD, again
a
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
with the expectation that such changes would improve outer membrane penetration and efflux
avoidance. Contrary to this expectation, these derivatives (13d-g in Table 2) all have higher MIC
shifts in the presence of PMBN compared to 13c (8-fold versus 2-fold, respectively). In addition,
all had equal or larger MIC shifts against the efflux compromised strain compared to 13c. These
observations suggest that an unknown counteracting property of these derivatives reduce their
ability to penetrate the outer membrane and avoid efflux relative to 13c. One possible explanation
is that the basicity of the amines of 13d-g is too high since they all have pK values greater than
a
1
3c. If true, this would suggest there is a narrow pK range for this scaffold to achieve optimal
a
cellular penetration. When the majority of the compound population is positively charged at
physiological pH, as is the case when the pK is >8, the membrane penetration and efflux avoidance
a
properties of the compounds may get worse despite relatively low cLogD values. While this model
also remains untested, we focused our further medicinal chemistry efforts in this series on making
compounds with amine pK values less than 8.
a
After further rounds of elaboration at the 7-position, the most potent compound against the
WT strains identified 13h, which contained an amino azabicyclohexane group with a pK value of
a
8
. This compound had similar physicochemical properties as compounds 13d-g and yielded similar
MIC shifts in the presence of PMBN and against the efflux compromised strain. However, 13h
achieved the lowest MIC values against the WT strains of interest (except against P. aeruginosa,
which remained above the measurable range), presumably due to its approximately 10-fold
improvement in BC target affinity compared to the parent 4a. The resiliency of the WT P.
1
6
ACS Paragon Plus Environment

Page 17 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
aeruginosa strain was not unexpected as this species is known to have a less permeable membrane
and a more extensive efflux system compared to the other ESKAPE pathogens tested here.31-35
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
Structure-activity relationships of slot position derivatives. In an effort to further improve the
potency of parent compound 4a, we turned our attention to the second vector of interest pointing
off the dihalo-aromatic ring (Figure 1B). Structural elaboration at this position occurred in two
phases. The first phase involved the addition of various aryl groups off of the dibromo aromatic
ring that were designed to fit within the small surface slot of the BC active site (Figure 1B). These
modifications did not contain groups capable of ionizing near physiological pH nor did they result
in large reductions to the cLogD so it was not expected that these compounds would have enhanced
membrane penetration or efflux avoidance.
These new compounds were tested against our MIC panel to assess cellular entry properties
and overall potency (Table 3). In accordance with expectations, this set of compounds had large
MIC shifts when PMBN was present, similar to parent compounds 1 and 4a. These compounds
were also much more potent against the E. coli efflux-compromised strain relative to the WT strain,
with MIC shifts ranging from 64- to >512-fold. These observations support the prediction that this
set of compounds would retain relatively poor outer membrane permeability and high efflux
potential because of unoptimized physicochemical properties. However, we note that compound
4
c has a cLogD value a full unit lower than parent 1, yet the MIC fold-shift for this compound
suggests no improvement in either outer membrane penetration or efflux avoidance. While this
reduction in cLogD value may not be large enough to yield predictable penetration and efflux
avoidance improvements, it may also be that introduction of polarity at this particular position on
the molecule is ineffective for reasons not captured by the simple expectation based solely on bulk
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 61
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
physicochemical properties. Given that 4c has the same cLogD as 13c, which did not show large
MIC shifts in the presence of PMBN or when efflux was compromised, it may be that the presence
of an ionizable amine, as on 13c and not 4c, is the more important property in determining cellular
accumulation compared to the cLogD parameter.
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
8
ACS Paragon Plus Environment

Page 19 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 3. MIC values, IC values, and physicochemical properties of interest for compounds
5
0
elaborated at the slot-position
R
N
H2N
N
N
NH2
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
OMe
Br
OiBu
N
N
N
-R
Br
Cl
Cl
Cl
Organisma
1
4b
>256
>256
>256
>256
2
0.5
0.25
>256
4c
>256
256
256
>256
4
2
0.5
16
4d
4e
4f
A. baumannii
E. coli
128
32
>256
8
>256
8
>128
32
K. pneumoniae
P. aeruginosa
E. coli +PMBN
E. coli (ΔtolC)
E. coli (ΔtolC) +PMBN
64
32
>16
>256
0.25
0.06
0.03
8
>128
>128
1
0.25
≤0.06
32
>256
0.5
0.5
0.12
16
>256
0.5
0.12
0.06
4
b
c
P. aeruginosa (ΔmexAB, etc.)
E. coli +/-PMBN
E. coli +/-efflux
64
64
>128
>512
2
64
128
4
16
64
32
128
2
32
128
>4
E. coli (ΔtolC) +/-PMBN
P. aeruginosa +/-efflux
IC50 P. aeruginosa BC (nM)
MW (g/mol)
4
2
>16
180
395
4.8
3.3
-
>16
130
314
(4.9)
2.3
>64
36
>32
32
>4
79
93
313
(4.7)
3.2
349
(4.4)
2.9
379
(3.8)
2.6
344
(3.9)
2.9
d
pK
a
clogD (pH 7.4)
aThe strains of each bacterial species were as follows: A. baumannii (wild type) ATCC 19606; E. coli
b
(wild type) DM4100; K. pneumoniae (wild type) ATCC 43816; P. aeruginosa (wild type) PAO1. The
c
E. coli ΔtolC strain contains a targeted knockout of the efflux pump component TolC. The P.
aeruginosa ΔmexAB, etc. strain contains targeted knockouts of efflux pumps MexAB-OprM, MexCD-
OprJ, and MexEF-OprN, and the efflux pump components MexXY are expected to be compromised by
d
the absence of OprM. Calculated pK values are in parentheses.
a
To assess the effect of the structural changes off the dibromo aromatic ring on target BC
inhibition, IC values were measured using purified P. aeruginosa BC. As these derivatives were
5
0
designed to sterically fit into a small slot extending from the BC active site but not make new
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 61
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
specific interactions, it was expected that the IC values of the elaborated compounds would be
5
0
quite similar to parent compound 1. As expected, the IC50 values reported in Table 3 are
approximately equal to or lower by up to 6-fold of 1, indicating that the introduction of additional
ring systems at this position do not cause major disruptions to target inhibition. Wary of the
potential for the additional aromatic ring to reduce solubility, it was discovered that certain
aliphatic groups, such as isobutyl 4f, could also fit the slot position and afford activity against the
enzyme.
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
For the final phase of the structural elaboration off the dihalo-aromatic ring we chose a
starting scaffold with the best amine-containing group identified during the derivatization
campaign at the 7-position of the pyridopyrimidine ring, 13h. The first two related compounds in
the series, 14a and 14b (Table 4), maintained the physicochemical properties of the parent
compound 13h and showed an improvement in BC inhibition, with IC values recorded at the
5
0
lower limit of detection of 15 nM – an improvement of 2-fold relative to the parent 13h. This
BC target inhibition improvement could be responsible for the small MIC improvements observed
against the WT strains (with the exception of A. baumannii). No substantially different MIC shifts
were observed in the presence of PMBN or when efflux was compromised, as expected given that
1
4a and 14b have pK and cLogD values similar to the parent comparator 13h and all compounds
a
contain an ionizable amine.
2
0
ACS Paragon Plus Environment

Page 21 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 4. MIC values, IC values, and physicochemical properties of interest for compounds
5
0
elaborated in the final phase of the medicinal chemistry campaign
R
N
Cl
H
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
N
N
N
NH2
H2N
H
F
N
OMe
S
-R
O
N
N
N
Organisma
14a
14b
14c
32
14d
128
0.5
2
14e
A. baumannii
E. coli
>256
1
128
1
32
2
1
K. pneumoniae
P. aeruginosa
E. coli +PMBN
E. coli (ΔtolC)
E. coli (ΔtolC) +PMBN
8
4
4
8
256
0.12
0.03
0.008
1
128
0.12
0.03
0.015
2
128
0.06
0.015
0.008
2
128
>64
0.06
0.03
0.015
8
0.03
0.008
0.004
1
b
c
P. aeruginosa (ΔmexAB, etc.)
E. coli +/-PMBN
E. coli +/-efflux
8
8
16
64
16
64
32
64
32
32
P. aeruginosa +/-efflux
IC50 P. aeruginosa BC (nM)
256
≤15
430
8.4
2.5
64
64
128
≤15
450
(8.1)
2.6
>8
≤15
430
(8.1)
2.5
≤15
478
(8.1)
2.4
≤15
425
8.1
2.8
MW (g/mol)
d
pK
a
cLogD (pH 7.4)
aThe strains of each bacterial species were as follows: A. baumannii (wild type) ATCC
1
9606; E. coli (wild type) DM4100; K. pneumoniae (wild type) ATCC 43816; P.
b
aeruginosa (wild type) PAO1; S. aureus (wild type) ATCC 29213. The E. coli ΔtolC strain
3
contains a targeted knockout of the efflux pump component TolC. The P. aeruginosa
ΔmexAB, etc. strain contains targeted knockouts of efflux pumps MexAB-OprM, MexCD-
OprJ, and MexEF-OprN, and the efflux pump components MexXY are expected to be
4
compromised by the absence of OprM. Calculated pK values are in parentheses.
a
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 61
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
The next compounds in the series, 14c-14e, all maintained BC IC values below 15 nM
5
0
and achieved the most potent microbiological activity observed in the series. As these compounds
also maintained the general physicochemical properties of the starting scaffold, the MIC shifts
observed in the presence of PMBN or against the efflux compromised strain were similar to the
shifts observed with the parent compound.
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
Structural Biology. To assess the structural consequences of the 7-position modifications, the X-
ray co-crystal structure of 13f with the WT BC enzyme from E. coli was solved at 2.06 Å (see
Table S1 for crystallographic statistics). Figure 2A shows that the pyridopyrimidine ring scaffold
5
of 13f occupies the same position as the previously reported structure of 1. Thus, neither the
dibromo- to dichloro-substitution nor the modification off the 7-position altered the scaffold
binding position in the BC active site pocket.
Figure 2. BC co-crystal structures of new pyridopyrimidine inhibitors overlaid with previously
identified parent compound 1 co-crystallized with E. coli BC (shown in green; PDB 2V58). (A)
Compound 13f (carbon, magenta; nitrogen, blue; chlorine, light green; all rendered as sticks) co-
crystallized with E. coli BC at 2.06 Å (magenta ribbon cartoon; PDB code 6OI9). The core
pyridopyrimidine scaffold of 13f is found in the identical position as that of parent compound 1.
The 7-position moiety extending off 13f points toward the conserved glutamate residues placing
2
2
ACS Paragon Plus Environment

Page 23 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
its proximal amine group nitrogen atom 3.16 and 2.17 Å in mol A and 2.97 and 2.65 Å in mol B
away from the nearest oxygen atoms of the carboxylate groups of Glu288 and Glu276,
respectively. (B) Compound 14a (carbon, light blue; nitrogen, blue; chlorine, light green; all
rendered as sticks) co-crystallized with the H. influenzae BC at 2.5 Å (light blue-ribbon cartoon;
PDB code 6OI8). The T-loop (Ref. 36) in the H. influenzae crystal structure is presumably
disordered, as observed previously in other BC crystal structures (Ref. 37) and not represented by
the ribbon cartoon. The core pyridopyrimidine scaffold of 14a is found in the identical position as
that of parent compound 1. The 7-position moiety extending off 14a points toward the conserved
glutamate residues placing its proximal amine group nitrogen atom 3.29 and 2.91 Å away from
the nearest oxygen atoms of the carboxylate groups of Glu288 and Glu276, respectively. (C) The
vector of 14a that extends off the dihalo-aromatic ring is positioned, as expected, within the small
slot in the active site surface as shown by the partially transparent solvent accessible surface
representation in grey. Images prepared using PyMOL.
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
Consistent with our hypothesis based on the structure of 1, the 7-position modification of
1
3f extends further into the interior of the BC enzyme placing the distal amine within H-bonding
distance to Glu288 and Glu276. Consistent with this possibility, the IC results show a 5-fold
5
0
improvement in the inhibition of 13f compared to 4a, possibly reflecting the binding contribution
of the new H-bond interactions to the conserved glutamate residues that are formed by 13f;
however, mutation studies involving both glutamate residues would be needed to more definitively
assign the binding affinity consequences of the 7-position modification.
To investigate the structural consequences of modifications off the dihalo ring, the crystal
structure of 14a in complex with WT BC enzyme from H. influenzae was solved at 2.5 Å (Figure
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 61
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
2
B and C; see Table S1 for crystallographic statistics). The active site of the H. influenzae BC
enzyme is nearly identical to the other Gram-negative species of interest.5,18 The structure shows
that the pyridopyrimidine ring scaffold of 14a occupies the same position as 13f and the previously
reported structure of 1 demonstrating that the pyridyl modification off the dihalo ring does not
alter the binding conformation of the core ring system. This modification is positioned, as intended,
within the slot region of the BC active site, potentially making packing interactions to the residues
in the walls of the slot (Figure 2C). The structure also shows that the 7-position moiety of 14a
extends toward the conserved Glu288 and Glu276 residues positioning the primary amine of 14a
within H-bonding distance to the two acid residues (Figure 2B). Thus, both modifications found
on 14a had the intended structural consequences.
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
Bacterial Cytological Profiling. The optimization of antibacterial compounds may be
compromised by off-target activity that would complicate interpretation of structure-activity
38
relationships. To test that the improved antibacterial potency of the novel BC inhibitors was due
to on-target activity, Bacterial Cytological Profiling was used. Briefly, this technique uses high
resolution fluorescence microscopy to compare the effects on cytological structure of novel
antibiotics to a set of reference control antibiotics with known targets.39,40 Compounds that inhibit
different pathways generate reproducible and distinct cytological signatures that can be used to
infer the mechanism of action of unknown compounds.39-44 When E. coli cells were treated with
4
the previously reported BC inhibitor 1 at ~3-fold the MIC for 2 hours they formed short chains of
3
to 8 cells, a phenotype that is characteristic of lipid biogenesis inhibitors (Figure 3). Two novel
inhibitors, 13f and 14e, were tested in the profiling assay. It was found that both showed
2
4
ACS Paragon Plus Environment

Page 25 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
cytological profiles very similar to each other and to parent compound 1 suggesting that they
inhibit the same target in vivo.
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 3. Cells treated with BC inhibitors produce chains of
cells. Cells were treated with either DMSO or compound at the
indicated concentrations for 2 hours. The top panel shows FM4-64
stained cell membranes (red) and the bottom panel shows both cell
membranes and DNA stained with DAPI (blue). Scale bar equals 1
micron.
Selectivity characterization. Compound 1 is a member of a class of compounds that were
4
5
originally identified as inhibitors of VEGFR2 and FGFR1 eukaryotic protein kinases. However,
5
in the original report, 1 was shown to have acceptable selectivity for bacterial BC over these and
2
8 other eukaryotic protein kinases. To test if the potency improvements of the derivative
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 61
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
compounds synthesized in the current work came at the expense of losses in selectivity, two
particularly promising compounds, 14a and 14e, were subjected to a similar panel of kinases.
Compounds were profiled at a single concentration of 25  in the Z’-LYTE kinase activity assay
at Invitrogen. The results are shown in Table 5. Compared to parent compound 1, the novel
compounds retain excellent and even improved selectivity against this panel of kinases.
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 5. Percent inhibition of eukaryotic kinases at
a compound concentration of 25 M
compound
kinase
1
14a
14e
ABL1
FYN
SRC
BTK
97
93
93
92
78
58
52
47
26
24
21
20
20
18
16
13
10
7
38
12
8
16
57
65
21
47
1
1
7
9
25
4
30
12
8
32
60
6
24
40
6
5
15
8
23
4
3
FGFR1
EGFR (ErbB1)
MAP4K4 (HGK)
NTRK1 (TRKA)
MAPK1 (ERK2)
MAPK14 (p38 alpha) Direct
MET (cMet)
AURKA (Aurora A)
PAK4
CDK1/cyclin B
CSNK1D (CK1 delta)
CHEK2 (CHK2)
MAPKAPK2
3
20
26
32
3
67
21
32
3
SGK (SGK1)
INSR
6
GSK3B (GSK3 beta)
IKBKB (IKK beta)
AKT1 (PKB alpha)
CSNK2A1 (CK2 alpha 1)
PDK1 Direct
6
5
4
4
4
5
6
0
7
8
7
1
3
0
0
PRKCZ (PKC zeta)
CHEK1 (CHK1)
IKBKE (IKK epsilon)
0
-1
-7
5
28
3
-8
42
0
Determination of resistance liabilities. One concern with inhibitors targeting a single enzyme is
that single point mutations in the target enzyme may lead to high-level resistance. Due to the high
2
6
ACS Paragon Plus Environment

Page 27 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
bacterial burden and error-prone replication, it is feasible for bacteria to sample every possible
point mutation during the course of an infection. If a single-point mutation can disrupt inhibitor
binding without drastically impairing enzyme function then the infection will become resistant [for
example see Ref. 46]. Efflux-mediated resistance is more common than target-based resistance,
but upregulation of a single efflux system usually has a moderate impact on potency.
To determine whether high-level resistance to our inhibitors could arise in P. aeruginosa,
we selected for spontaneous resistant mutants in the efflux-compromised P. aeruginosa strain. The
lack of genes encoding the three major antibiotic efflux systems increases the likelihood of
identifying rare mutations in other genes. However, it should be noted that P. aeruginosa encodes
additional efflux systems that can contribute to resistance.
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
We selected for mutants on agar plates containing 4-, 8-, or 16-fold the MIC of compounds
1
, 14a, or 14e. The spontaneous mutation frequency of these compounds in this strain background
-9
-9
was similar, ranging from 210 for 14e to 910 for 14a at 8-fold above the MIC. These
5
frequencies are low and similar to the frequencies observed for compound 1 previously. We note
however that spontaneous mutation frequencies would likely be much higher for a WT P.
aeruginosa strain carrying the full suite of efflux systems as these additional components can also
mutate to more efficiently efflux lethal compounds.
Several resistant colonies of the efflux-compromised P. aeruginosa strain were found to
harbor mutations in the BC coding sequence. These mutations encoded changes to amino acids
Leu278 and Ile437, which are located in the BC active site according to a structure of the P.
1
8
aeruginosa enzyme in complex with a substrate analog (PDB: 2VQD). Both residues are highly
conserved across BC species including E. coli and H. influenzae.5,18 The I437N and I437T
mutations identified here have also been observed previously when efflux-compromised strains of
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 61
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
5
E. coli and H. influenzae were grown in the presence of 1. The L278F mutation observed here in
5
P. aeruginosa was not identified in either the E. coli or H. influenzae enzymes. However, several
other active site mutations were observed in E. coli and H. influenzae that were not observed in
the current study, possibly because not enough resistant colonies were screened to exhaustively
identify all possible mutants. If this were the case, additional screening would eventually reveal
the same set of resistance conferring mutants for all BC species. Alternatively, the different sets
of active site mutations observed in the BC enzymes from different species may reflect the
influence of subtle BC structural differences that result in truly different sets of possible resistance-
conferring mutations.
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
All three P. aeruginosa BC active site mutations isolated in this study have markedly
stronger effects on the potency of 14a compared to 14e (Table 6). The L278F mutation raised the
MIC of 14a by 256-fold above the MIC of the parent strain, while both I437 mutations raised the
MIC 128-fold. In contrast, 14e was less affected by all three mutations, showing 8-fold MIC shifts
above the parent strain.
Table 6. MIC changes conferred by spontaneous P.
aeruginosa mutations
MIC (mg L-1)
fold-change from WT
Strain
1
16
128
>128
>128
>128
>128
>128
>128
14a
1
256
128
128
2
2
1
32
14e
1
(1)
8
>8
>8
>8
>8
>8
>8
14a
(1)
256
128
128
2
2
1
32
14e
(1)
8
8
8
1
1
1
4
a
WT background
BC L278F
BC I437N
8
64
64
64
8
8
8
32
BC I437T
MexT F201I
MexT P202L
MexT V226L
INV(muxA-mexZ)
aP. aeruginosa ΔmexAB-oprM, ΔmexCD-oprJ, ΔmexEF-oprN
[PAM1626 from Ref. 47]
2
8
ACS Paragon Plus Environment

Page 29 of 61
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Examination of the 14a H. influenzae BC X-ray co-crystal structure overlaid with the
previously reported P. aeruginosa BC with substrate analog co-crystal structure18 offers a
speculative model for how the resistance-conferring mutations might disrupt inhibitor binding but
not significantly impact substrate turnover (Figure 4). In the H. influenzae structure, the
pyridopyrimidine ring moiety of the 14a ligand extends further toward Leu278 than the
corresponding adenine group of the substrate analog. Because of this difference, mutation to the
larger phenylalanine amino acid could sterically hinder 14a binding while leaving substrate and
transition state binding unaffected. By contrast, the structural overlap of the inhibitor and substrate
near the 437 position appears quite extensive in the structural alignment shown in Figure 4. As
the resistance-conferring mutations at this position introduce more polarity, it may be that
electrostatic forces play a large role in affording the differential effect on substrate/transition state
versus inhibitor binding. Further catalytic and inhibition studies as well as additional X-ray crystal
structures would be needed to understand the precise consequences of the resistance-conferring
mutations. These future studies may also reveal why 14e MIC potency is not as largely affected
by the resistance-conferring mutations as the other inhibitors.
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
9
ACS Paragon Plus Environment