Pyrazole-Based Lactate Dehydrogenase Inhibitors with Optimized Cell Activity and Pharmacokinetic Properties

Article abstract of DOI:10.1021/acs.jmedchem.0c00916

Lactate dehydrogenase (LDH) catalyzes the conversion of pyruvate to lactate, with concomitant oxidation of reduced nicotinamide adenine dinucleotide as the final step in the glycolytic pathway. Glycolysis plays an important role in the metabolic plasticity of cancer cells and has long been recognized as a potential therapeutic target. Thus, potent, selective inhibitors of LDH represent an attractive therapeutic approach. However, to date, pharmacological agents have failed to achieve significant target engagement in vivo, possibly because the protein is present in cells at very high concentrations. We report herein a lead optimization campaign focused on a pyrazole-based series of compounds, using structure-based design concepts, coupled with optimization of cellular potency, in vitro drug-target residence times, and in vivo PK properties, to identify first-in-class inhibitors that demonstrate LDH inhibition in vivo. The lead compounds, named NCATS-SM1440 (43) and NCATS-SM1441 (52), possess desirable attributes for further studying the effect of in vivo LDH inhibition.

Full text of DOI:10.1021/acs.jmedchem.0c00916

Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE
Article
Pyrazole-Based Lactate Dehydrogenase (LDH) Inhibitors
with Optimized Cell Activity and Pharmacokinetic Properties
Ganesha Rai, Daniel J Urban, Bryan T Mott, Xin Hu, Shyh-Ming Yang, Gloria A Benavides,
Michelle S Johnson, Giuseppe L Squadrito, Kyle R Brimacombe, Tobie D Lee, Dorian M Cheff,
Hu Zhu, Mark J. Henderson, Katherine Pohida, Gary A. Sulikowski, David M. Dranow, Md Kabir,
Pranav Shah, Elias Padilha, Dingyin Tao, Yuhong Fang, Plamen P. Christov, Kwangho Kim,
Somnath Jana, Pavan Muttil, Tamara Anderson, Nitesh K Kunda, Helen J. Hathaway, Donna F
Kusewitt, Nobu Oshima, murali Cherukuri, Douglas R Davies, Jeffrey P Norenberg, Larry A. Sklar,
William J. Moore, Chi V Dang, Gordon M Stott, Leonard M Neckers, Andrew J Flint, Victor Darley
Usmar, Anton Simeonov, Alex G Waterson, Ajit Jadhav, Matthew D. Hall, and David J. Maloney
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c00916 • Publication Date (Web): 09 Sep 2020
Downloaded from pubs.acs.org on September 9, 2020
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 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Dang, Chi; Leonard and Madlyn Abramson Family Cancer Research
Institute
Stott, Gordon; Frederick National Laboratory for Cancer Research
Neckers, Leonard; National Cancer Institute, Urologic Oncology Branch
Flint, Andrew; Frederick National Laboratory for Cancer Research
Usmar, Victor Darley; The University of Alabama at Birmingham
Simeonov, Anton; National Institutes of Health, NIH Chemical Genomic
Center
Waterson, Alex; Vanderbilt Institute of Chemical Biology
Jadhav, Ajit; NIH, NCATS
Hall, Matthew; National Institutes of Health,
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
Maloney, David; Nexus Discovery Advisors ,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 95
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
Pyrazole-Based Lactate Dehydrogenase (LDH) Inhibitors with
Optimized Cell Activity and Pharmacokinetic Properties
a†
a†
†
†
†
Ganesha Rai* , Daniel J. Urban , Bryan T. Mott , Xin Hu , Shyh-Ming Yang , Gloria 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
‡
‡
‡
†
Benavides , Michelle S. Johnson , Giuseppe L. Squadrito Kyle R. Brimacombe , Tobie D.
Lee , Dorian M. Cheff , Hu Zhu , Mark J. Henderson , Katherine Pohida , Gary A.
†
†
†
†
†
⊥
§
†
†
†
†
Sulikowski , David M. Dranow , Md Kabir , Pranav Shah , Elias Padilha , Dingyin Tao ,
†
⊥
⊥
⊥
€
Yuhong Fang , Plamen Christov , Kwangho Kim , Somnath Jana , Pavan Muttil , Tamara
€
€
€
¥
∥
Anderson , Nitesh K. Kunda , Helen J. Hathaway , Donna F. Kusewitt , Nobu Oshima ,
Murali Cherukuri , Douglas R. Davies , Jeffrey P. Norenberg , Larry A. Sklar , William J.
∥
§
€
¥
#
∇
#
∥
#
Moore , Chi V. Dang , Gordon M. Stott , Leonard Neckers , Andrew J. Flint , Victor M.
‡
†
⊥
†
†
Darley-Usmar , Anton Simeonov , Alex G. Waterson* , Ajit Jadhav , Matthew D. Hall ,
David J. Maloney*†
†
National Center for Advancing Translational Sciences, National Institutes of Health, 9800
Medical Center Drive, Rockville, Maryland 20850, United States
‡
Mitochondrial Medicine Laboratory, Department of Pathology, University of Alabama at
Birmingham, Birmingham, Alabama 35294, United States
∥
Urologic Oncology Branch, Center for Cancer Research, National Cancer Institute, 9000
Rockville Pike, Bethesda, Maryland 20892, United States
⊥
3
Vanderbilt Institute of Chemical Biology, Vanderbilt University, Nashville, Tennessee
7232, United States
#
NExT Program Support, Applied/Developmental Research Directorate, Leidos Biomedical
Research, Inc., Frederick National Laboratory for Cancer Research, Frederick, Maryland
1702, United States
2
∇
Abramson Cancer Center, Abramson Family Cancer Research Institute, Perelman School
of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States.
Ludwig Institute for Cancer Research, New York, NY 10017, United States.
§
Beryllium Discovery Corp., 7869 Day Road West, Bainbridge Island, Washington 98110,
United States
€
College of Pharmacy, University of New Mexico Health Sciences Center, Albuquerque,
New Mexico 87131, United States.
¥
Dept of Pathology, University of New Mexico Cancer Center, Albuquerque, New
Mexico 87131, United States.
1
| P a g e
ACS Paragon Plus Environment

Page 3 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Abstract: Lactate dehydrogenase (LDH) catalyzes the conversion of pyruvate to lactate, with
concomitant oxidation of NADH as the final step in the glycolytic pathway. Glycolysis plays
an important role in the metabolic plasticity of cancer cells and has long been recognized as a
potential therapeutic target. Thus, potent, selective inhibitors of LDH represent an attractive
therapeutic approach. However, to date, pharmacological agents have failed to achieve
significant target engagement in vivo, possibly because the protein is present in cells at very
high concentrations. We report herein a lead optimization campaign focused on a pyrazole-
based series of compounds, using structure-based design concepts, coupled with optimization
of cellular potency, in vitro drug-target residence times, and in vivo PK properties, to identify
first-in-class inhibitors that demonstrate LDH inhibition in vivo. The lead compounds, named
NCATS-SM1440 (43) and NCATS-SM1441 (52) possess desirable attributes for further
studying the effect of in vivo LDH inhibition.ꢀ
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
| P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 95
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
INTRODUCTION
Cancer cells exhibit deregulated metabolic characteristics that are remarkably different
from normal cells, demonstrating an avidity for glucose uptake and catabolism, ultimately
producing lactate and generating ATP through the cytosolic, non-oxidative glycolytic
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
+
pathway. Glycolysis can be the preferred NAD regeneration and energy production process
for cancer cells, even in the presence of oxygen, rather than the more energy-efficient
mitochondrial oxidative phosphorylation. This is referred to as aerobic glycolysis, and the
1
preference for aerobic glycolysis is also known as the Warburg effect. While aerobic
glycolysis is an inefficient way to generate ATP (compared with oxidative phosphorylation),
the rate of ATP generation is rapid. In addition, it is hypothesized that rapidly proliferating
+
cancer cells have adapted this approach to regenerate NAD and to support the production of
essential cellular building blocks such as amino acids, lipids, and nucleotides needed to
2
support rapid cell growth. Indeed, many non-cancer cells use a combination of oxidative
phosphorylation and glycolysis to achieve the needed metabolic plasticity to serve their
biological function. While the molecular basis of aberrant cancer metabolism and its role in
3
cancer development and progression have yet to be fully elucidated, the Warburg effect, and
the enzymes in glycolysis, have long been recognized as potential targets for the selective
killing of cancer cells.
Many glycolytic enzymes are overexpressed in cancer, including the lactate
dehydrogenase (LDH) enzymes A (LDHA) and B (LDHB).4,5 LDH is a tetrameric protein
composed of the products of the LDHA (subunit M) and LDHB (subunit H) genes. The
tetrameric combination of these gene products generates 5 LDH isoforms with different
combinations of subunits depending on the cell type. The LDH5 (4M) isoform is the primary
6
form expressed in cancer cells, although other isoforms have also been reported. All LDH
isoforms catalyze the last step in the glycolytic pathway converting pyruvate to lactate while
3
| P a g e
ACS Paragon Plus Environment

Page 5 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
+
regenerating NAD from NADH. The lactate produced is then secreted from cells via the
monocarboxylate transporter (MCT) proteins.
Significant evidence exists to support the development of LDH inhibitors as a
therapeutic option for cancer treatment. Genetic knockdown of LDHA has been shown to elicit
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
7
cell death or delayed cell growth in cell lines from colorectal carcinoma (siRNA), Burkitt
8
9
10
lymphoma (siRNA), hepatocellular carcinoma (siRNA), pancreatic cancer (shRNA), and
1
1
mouse mammary tumor cells (shRNA). For example, mouse mammary tumor cells lacking
1
1
LDHA implanted as xenografts demonstrated dramatically reduced tumor growth. And, in a
genetically engineered mouse model (GEMM) of non-small cell lung cancer, induced
knockout of mouse LDHA led to the regression of established tumors without serious systemic
12
toxicity. In contrast, LDHB knockdown has been reported not to significantly impact tumor
cell survival.7
While the therapeutic potential of LDHA inhibition appears to be substantial, the
discovery and development of LDH inhibitors has proven to be challenging. For example,
because of the μM concentrations of LDH enzyme in cancer cells, an effective inhibitor will
likely need to bind with exceptionally high affinity and also achieve high intracellular
concentrations to enable a therapeutic level of target engagement. To date, no inhibitors of
LDH which meet these criteria have been reported. The first reported LDH inhibitors came
from academic groups (e.g. FX-1113a and NHI-2 ) with efforts from biotech and
13b
14
1
5
pharmaceutical companies being disclosed more recently. Molecules from GlaxoSmithKline
16
17
(
GSK) and Genentech (GNE-140) have demonstrated in vitro inhibition of cellular lactate
production and cytotoxicity in cancer cells, but the relatively poor pharmacokinetics of these
compounds limited their usefulness for testing of the therapeutic hypothesis in vivo. More
recently, LDH inhibitors (e.g. 'Compound 24c' in Figure 1) that show modest inhibition of
4
| P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 95
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
MiaPaCa-2 xenograft growth have been described; but target engagement in vivo was not
1
8
demonstrated .
We recently reported the identification of a pyrazole-based hit from quantitative high-
throughput screening (qHTS) and used structure-based design to develop nM inhibitors of
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
9
LDH enzyme activity, exemplified by 1. Compound 1 inhibited LDH in the highly glycolytic
MiaPaCa-2 (human pancreatic cancer) and A673 (human Ewing’s sarcoma) cell lines,
demonstrating sub-µM suppression of cellular lactate output and inhibition of cell growth.
While 1 showed favorable in vitro ADME properties (e.g., microsomal stability, solubility),
its PK profile was not suitable for in vivo use. Herein, we report the optimization of the
pyrazole-based chemotype using binding constants (particularly off-rates or drug-target
residence time) and cell-based activity to guide improvements in potency, while also
optimizing pharmacokinetic properties. These efforts ultimately resulted in lead compounds
43 (NCATS-SM1440) and 52 (NCATS-SM1441) that have enabled studies of the effects of
pharmacological LDH inhibition in animal models of Ewing’s sarcoma20 and human
pancreatic cancers21 as well as interrogation of a synergistic role in T cell-based
immunotherapy.22
5
| P a g e
ACS Paragon Plus Environment

Page 7 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Cl
S
HO
O
O
HN
S
CN
Cl
F
Cl
HO
O
N
F
O
NH O
N
H
NH
N
S
HN
O
O
N
S
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
O
N
O
H2NO2S
O
GSK 2837808A
GNE140
'Compound 22'
Ph
SO NH
2
2
F
O
N
O
O
O
N
N
S
N
O
N
CO H
2
'
Compound 24c'
1
F
F
SO NH2
2
SO NH2
2
F
S
N
N
S
N
N
S
N
S
N
CO H
2
CO H
2
4
3 (NCATS-SM1440)
52 (NCATS-SM1441)
Biochemical
Potency

<800 nM Cellular
Optimal PK
profile
X
In vivo LDHA
inhibition
X
Potency
GSK 2837808A
X
'
Compound 22'






X


X


X
X
X
X
X


GNE140
X
1
X
'
Compound 24c'
Not disclosed
a43 = NCATS-SM1440
b52 = NCATS-SM1441


Figure 1. Representative previously described LDH inhibitors and comparison with new
leads 43 = NCATS-SM1440 and 52 = NCATS-SM1441. Named as NCI-006 in references
0 and 21. Named as NCI-737 in reference 20.
a
b
2
CHEMISTRY
A robust linear synthetic strategy was used for the optimization of the cyclopropylmethyl and
sulfonamide regions (Scheme 1 and Scheme 2). The commercially available carboxylic acids
a-l were reacted with 1,2,3-benzotriazole in the presence of thionyl chloride to give acyl
benzotriazole intermediates Ia-l. The reaction of these intermediates (Ia-l) with substituted
acetophenones IIa-d, mediated by magnesium bromide ethyl etherate in the presence of
6
| P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 95
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
Hunig’s base, afforded 1,3-diketone derivatives IIIa-p (Table S3) in good yields.
Subsequently, cesium carbonate-mediated alkylation of 1,3-diketone intermediates IIIa-p
with 4- (bromomethyl)benzenesulfonamide derivatives IVa-c in DMSO at room temperature
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
19
generated advanced intermediates Va-r, as described in our previous paper . Initial attempts
with tosic acid-catalyzed cyclocondensation of 1,3-diketone derivatives IIIa-p with ethyl 2-
hydrazinylthiazole-4-carboxylate hydrobromide generated a mixture of both pyrazole
regioisomers. We previously reported that the 5-aryl regioisomer is inactive in LDHA assays,
so formation of this regioisomer is undesired. The product ratio of the desired isomer was poor
(
≤10%), particularly when R is electron-donating alkyl groups such as cyclopropylmethyl or
1
cyclopropylethyl groups which were found to be optimal for activity (e.g. Va-f, Vh-j, and Vn-
r). Therefore, it was neither efficient nor practical to use this cyclization protocol to synthesize
the quantities of the advanced bromide containing intermediates VIIa-f, VIIh-j, and VIIn-r
needed for late-stage optimization efforts.
The ratio of the two regioisomers produced in the reaction depends on reactivity differences
dictated by the electronic environment around the two keto groups. To improve the
regioisomeric ratio, we envisioned utilizing these reactivity differences by blocking the
undesired reactivity of the 1-keto group via initial formation of an enamine intermediate prior
to addition of the ethyl 2-hydrazinylthiazole-4-carboxylate. Extensive optimization identified
improved conditions, in which heating 1,3-diketones Va-r in ethanol in the presence of 0.5
equivalents of both pyrrolidine and tosic acid for 1-2 h at reflux, then adding ethyl 2-
hydrazinylthiazole-4-carboxylate and stirring at reflux overnight allowed us to obtain a ~
5
0:50 mixture of the two regioisomers. The desired isomer was isolated as the second peak
using reversed-phase (C18) chromatography, as the desired regioisomer was slightly less polar
1
9
(
as judged by LC/MS analysis and according to our previous reports ). Further reaction
optimization proved elusive, and the investigation of alternative routes to obtain exclusively
7
| P a g e
ACS Paragon Plus Environment

Page 9 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the desired isomer met with limited success. Thus, for all subsequent scale-up and analog
syntheses, we used the above procedure. Compound 1 and analogs 2-9 were obtained by
hydrolysis of the intermediates VIIa, VIIg-m, and VIIb-d, respectively, using lithium
hydroxide in ethanol and subsequent HPLC purification.
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 1: Syntheses of intermediates VIa-r and analogs 1-9a.
Br
N
N
O
O
O
O
N
3
R3
2
R2
R2
a
b
R1
R1
OH
R1
X
X
O
a-l
SO NH
2
2
IIIa-p
from Ia-l & IIa-d)
R =
Ia-l
c
IIa-d
IVa-c
1
(
a
b
e
F
a: X = H; R = Ph
a
: R = H
F
2
3
b: X = F; R = Ph
b
:
R = 2-F
f
2
3
d
F
c: X = H; R = Br
c
: R3 = 3-F
2
CF3
F
F3C
d
:
X = F; R = Br
2
i
l
g
j
h
k
SO NH2
2
SO NH2
2
R3
R3
R1
O
O
R1
R2
X
N
N
N
CO Et
2
R1
c
d
N
N
S
N
X
S
X
VIIa-r
Va-r
SO NH2
R2
R3
2
CO Et
2
R2
VIa-r
R =
1
e
CF3
R =
1
k: X = H; R = Ph; R = H
2
3
a: X = H; R = Ph; R = H
2
3
R =
1
b: X = H; R = Ph; R = 2-F
F3C
1 from VIIa
2
3
Analogs 2-6 from VIIg-m
Analogs 7-9 from VIIb-d
l: X = H; R = Ph; R = H
2
3
c: X = H; R = Ph; R = 3-F
2
3
F
d: X = F; R = Ph; R = H
R =
1
2
3
F
e: X = H; R = Br; R = 2-F
2
3
m: X = H; R = Ph; R = H
2
3
f: X = F; R = Br; R = 2-F
2
3
R =
1
F
R =
1
F
n: X = H; R = Br; R = 2-F
2
3
R =
1
g: X = H; R = Ph; R = H
2
3
R =
1
q: X = H; R = Br; R = 2-F
2
3
R =
1
o: X = H; R = Br; R = 2-F
2
3
h: X = H; R = Ph; R = H
2
3
R =
1
R =
1
i: X = H; R = Br; R = 2-F
2
3
j: X = F; R = Br; R = 2-F
p: X = H; R = Br; R = 2-F
r: X = H; R = Br; R = 2-F
2 3
2
3
2
3
8
| P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 95
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
a
Reagents and conditions: (a) SOCl , CH Cl , 4 h, 91-100% (b) MgBr ·OEt , iPr NEt,
2
2
2
2
2
2
CH Cl , 12 h, 60-69% (c) Cs CO , DMSO, 1 h, 55-83% (d) i. Pyrrolidine (0.5 eq), TsOH (0.5
2
2
2
3
eq), EtOH, reflux, 1-2 h ii. Ethyl 2-hydrazinylthiazole-4-carboxylate·HBr, reflux overnight
e) LiOH, THF/MeOH/H O, 1 h.
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
The synthesis of analogs 11-89 listed in Tables 2-4 was achieved utilizing either Sonogashira
or Suzuki coupling of the advanced 3-bromo aryl intermediates VIIe-f, VIIi-j, and VIIn-r,
followed by hydrolysis with LiOH (Scheme 2). For the Sonogashira coupling, we developed
a robust room temperature, copper-free protocol catalyzed by commercially available
monoligated precatalysts, [P(t-Bu) ] Pd(crotyl)Cl or [DTBNpP] Pd(crotyl)Cl, in the presence
3
2
3
of DABCO in dioxane . This method allowed us to quickly generate an alkyne library and
explore the SAR around the biphenyl region with access to more complex structures.
Moreover, the protocol requires only a minimal work-up that could further facilitate rapid
analog generation during our library synthesis and is amenable to providing large quantities
of key intermediates (e.g.VIIIa-d) to support advanced studies on top compounds. Using this
method, analogs 14-16, 20-32, 36-43, 48-49, 51-52, 74-75, and 84-89 were synthesized
directly from intermediates VIIe-f, VIIi-j, and VIIn-r with the corresponding commercially
available terminal alkynes IX and the [P(t-Bu) ] Pd(crotyl)Cl catalyst in dioxane. Syntheses
3
of analogs 33-35 is summarized in scheme 3. To rapidly explore the SAR with aryl or
heteroaryl R substitutions, a slightly modified 3-step protocol was used for analogs 44-47, 53-
6
7, and 76-83. The synthesis began with [P(t-Bu) ]Pd(crotyl)Cl-catalyzed Sonogashira
3
coupling of intermediates VIIe-f and VIIi-j with trimethylsilylacetylene, followed by
deprotection of the TMS group with CsF, to give intermediates VIIIa-d. Further Sonogashira
coupling of these alkyne intermediates with the corresponding aryl or heteroaryl bromides X
using the same conditions, followed by hydrolysis with LiOH, furnished analogs 44-45, 53-
61, 66-67, and 76-79. Analog 11 (R = H) was obtained upon hydrolysis of the intermediate
9
| P a g e
ACS Paragon Plus Environment

Page 11 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
VIIIb using LiOH in a methanol-water mixture. However, the use of the [P(t-
Bu) ]Pd(crotyl)Cl catalyst under similar conditions failed to give the isolable product in the
3
case of more electron-deficient heteroaryl bromides, such as 1,3-thiazolyl or 1,3-oxazolyl
bromides. Fortunately, switching to a more active catalyst, [DTBNpP]Pd(crotyl)Cl, with
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
o
slightly elevated reaction temperature (60 C in dioxane) facilitated efficient product
formation. Thus, Sonogashira reaction of the intermediates VIIIa-d with corresponding 1,3-
thiazolyl or 1,3-oxazolyl bromides under those conditions in the presence of DABCO,
followed by hydrolysis, furnished analogs 46-47, 62-65 and 80-82. Due to the gaseous nature
and commercially unavailability of alkyne precursors, we used SPhos precatalyst-catalyzed
Suzuki coupling of the corresponding alkynyl trifluorobornates or pinacol esters, followed by
hydrolysis protocol for alkyne analogs 12-13, 17-19 and 50. Finally, Suzuki coupling of
corresponding alkenyl boronates or boronic acids with VIIe-f, using SPhos Pd(crotyl)Cl
o
precatalyst in the presence of potassium phosphate base at 100 C and subsequent hydrolysis
of the ester group, provided analogs 68-73.
Scheme 2: Syntheses of analogs 10-32 and 36-89a.
1
0 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 95
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
X
X
F
F
Br
R
2
SO NH2
SO NH
2
2
a, d
+
R
N
S
N
S
R1
R1
N
N
IX
N
N
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
CO Et
CO2H
4-16, 20-32, 36-43, 48-49: X = F
(Table 2 for R groups)
2
1
R =
VIIe-f, VIIi-j, VIIn-r
1
5
7
7
1-52: X = H ( Table 3 for R groups)
4: X = F; R = 5-methylthiophen-2-yl
5: X = H; R = 5-methylthiophen-2-yl
i. a,
ii. e
TMS
R =
1
84-89: X = H (Table 5 for R and R1 groups)
X
F
X
F
SO NH
2
2
R
SO NH2
2
N
R1
a or b & d
N
+
Br
R
N
S
R1
N
N
X
S
N
CO Et
2
CO H
VIIIa-d
2
4
4-47: X = F (Table 2 for R groups)
3-67: X = H (Table 3 for R groups)
R =
1
R =
1
R =
1
5
a: X = H
b: X =F
c: X = H
11 d: X =F
74: X = F (Table 5 for R group)
R =
d
1
7
6-83: X = H (Table 5 for R groups)
X
X
F
F
Br
R
SO NH2
2
SO NH
2
2
c, d
+
R B(OH)2
N
S
N
N
N
XI
S
N
N
CO Et
CO2H
2
1
0, 12-13, 17-19, 50, 68, 73: X = H (Tables 1-4 for R groups)
VIIe-f
6
9-72: X = F (see Table 4 for R groups)
a
Reagents and conditions: (a) [P(t-Bu) ]Pd(crotyl)Cl, DABCO, dioxane, RT, 1-12 h (b)
3
o
[DTBNpP]Pd(crotyl)Cl, DABCO, dioxane, 60 C, 12 h (c) SPhosPd(crotyl)Cl, K PO ,
3
4
o
dioxane/H O, 100 C, 0.5 h (d) LiOH, THF/MeOH/H O, 1 h. e) CsF, THF-EtOH, RT, 2 h
2
2
Scheme 3: Syntheses of analogs 33-35a.
1
1 | P a g e
ACS Paragon Plus Environment

Page 13 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
F
F
F
F
R
NC
SO NH2
SO NH2
2
2
a, b, c
N
S
N
S
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
CO Et
CO H
2
2
33
1
8a: R = CH OH
2
a
d,f
1
8b: R = CHO
35
F
F
F
F
Br
F3C
SO NH2
SO NH2
2
2
e, f
+
(Bu) Sn
4
CF3
N
N
N
N
S
N
S
N
CO Et
CO H
2
2
VIIf
34
a
Reagents and conditions: (a) Dess-Martin periodinane, DCM, RT, 2 h (b) NaN , CF SO H,
3
3
3
o
ACN, RT, 12 h (c) (CH ) Sn-OH, dichloroethane, MW, 110 C, 1 h (d) Deoxo-Fluor, DCM,
3
3
t-
o
RT, 12 h (e) P( Bu) HBF ([PdCl(allyl)] dioxan, 80 C, 4 h (f) LiOH, THF/MeOH/H O, 1
3.
4,
2,
2
h.
RESULTS AND DISCUSSION
In our previous work, we described structure-guided optimization of a weakly active
quantitative high-throughput screening hit into 1, a potent, selective LDH inhibitor suitable
1
9
for probing LDH function in cells . Although 1 showed nanomolar inhibition of LDHA in
the biochemical activity assay and was able to achieve reasonable cellular potency, additional
optimization was required to identify highly potent molecules with PK properties suitable for
in vivo studies. Here, we report a focused lead optimization effort that has identified
compounds with good activity in the highly glycolytic MiaPaCa-2 and A673 cell lines and
with improved pharmacokinetic properties.
We previously showed that the thiazole carboxylic acid moiety forms a critical interaction
with the active site R168 and could not be replaced by other groups. The 4-benzyl sulfonamide
1
2 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 95
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
group likewise forms critical hydrogen bond interactions with Asp140, Glu191, and Ile141
that contribute significantly to the potency. Further, the cyclopropylmethyl group forms an
important π stacking interaction with Tyr238 and was found to be a key substituent for
conferring cellular potency. However, previous modifications of the benzyl sulfonamide or
replacements for the cyclopropylmethyl group had been limited, and thus the potential to
improve potency and residence time had not been fully explored. Additionally, in our previous
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
9
co-crystal structure , the binding pose shows that the biphenyl does not fully occupy the
hydrophobic pocket in this region, suggesting an opportunity to introduce other lipophilic
groups in a region that appeared to be the most accommodating for further SAR optimization.
Therefore, the lead optimization efforts described here have focused on these three regions of
the compounds in this series.
Based on our published crystal structures of compounds in this series bound to LDHA, the
SAR campaign began with more extensive investigations of the cyclopropylmethyl group.
Initially, a difluoro group was introduced into the cyclopropyl ring (2), anticipating additional
hydrogen bonding interaction with the phenol of residue Tyr238 (See Figure 2). Although
analog 2 maintained biochemical potency (24 nM vs. 25 nM), it showed reduced cellular
potency and cytotoxicity compared to 1 (A673 lactate = 1423 vs. 450 nM, A673 cytotox =
4
642 vs. 2450 nM; MiaPaCa-2 lactate = 1721 vs. 636 nM, cytotox = 10,660 vs. 8448 nM). To
extend deeper into this binding site the cyclopropylethyl (3) analog was made. This group
resulted in reduced biochemical LDHA activity and offered only marginal improvement in
cellular potency. This trend appeared to be general, as it replicated across numerous matched
pairs of analogs (see 43 vs. 74, 52 vs. 75, 54 vs. 77, 56 vs. 78, 57 vs. 79, 63 vs. 80, 64 vs. 81,
65 vs. 82, and 67 vs. 83 in Tables 2-4). Replacing the cyclopropylmethyl group with other
fluorinated alkyl groups (4-6) diminished potency in the biochemical assay and drastically
reduced the cellular activity compared to 1. Adding a branched substitution to the methylene
1
3 | P a g e
ACS Paragon Plus Environment

Page 15 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
group (84, 85, and 89) or adding substitution on the cyclopropyl ring (86-88) further reduced
both biochemical and cellular potency, relative to 75. The primary focus was to gain cellular
potency, yet moieties other than cyclopropylmethyl or cyclopropylethyl failed to improve
activity in both lactate and cytotoxicity cellular assays. Therefore, the preferred
cyclopropylmethyl group was utilized in subsequent analogs. We next explored fluorine
substitution at the 2- (7) or 3- positions (8) of the benzenesulfonamide ring. Analog 7, with a
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
-fluorobenzene sulfonamide substitution, showed similar biochemical potency to 1, and
slightly decreased activity in the lactate assay, but exhibited an improvement in cytotoxicity.
The 3-fluoro analog, 8, exhibited a significant decrease in activity across all assays. Initially,
we were unable to rationalize these observations, as neither fluorine group appeared
positioned to contribute an additional specific interaction with LDH, based on the previous
crystal structure. However, crystal structures with 2-fluoro-containing analogs (23, 52 Figure
2) clearly show a preferred orientation of the ring, with the fluorine occupying a position
pointed into the protein and away from the space occupied by the biphenyl or alkyne
substitutions. Based on the previously reported correlation between the binding kinetics and
the cellular potency, we subsequently found that the 2-fluoro substitution improves the off-
rate of the molecule (Table 6), which might explain the enhanced cytotoxic effects in
MiaPaCa-2 and A673 cellular assays. Therefore, the 2-fluoro group was retained as an optimal
substitution to evaluate further modifications of the lead molecule.
Table 1. Biochemical and Cellular LDH Inhibition of 2−10 with Comparator 1.
SO NH
SO NH
SO NH
2
2
2
2
2
2
R
2
R
3
R
CO2H
N
CO2H
N
CO H
2
N
N
N
N
N
N
N
Ph
F
Ph
Ph
S
S
S
1-6
7-8
9-10
1
4 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 95
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
HEK293
Cells
A673 cells
MiaPaCa-2 cells
Biochemical
Compd
No.
R
LDHA
lactate
inh
lactate
Cytotox.
inh.
Cytotox.
IC50
(nM)
CETSA
a
IC50
(nM)
IC (nM)
a
5
0
IC50
IC50
(nM)
IC (nM)
a
a
50
a
a
(
nM)
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
2
3
4
5
6
7
8
9
24
25
36
450
2454
4642
636
1721
804
8448
1066
-
501
727
15849
303
2529
150
-
F
1423
509
F
1959
6195
274
9719
20186
1861
613
148b
19400
12791
4760
867
3168b
>30000
17706
3376
CF3
CF3
4104
36
>30000
24440
1144
F
F
2-F
25
3-F
H
229
23
4465
835
11471
2647
6465
1224
585
5747
8448
401
141
1
0
F
23
465
1344
3416
a
IC values represent the half maximal (50%) inhibitory concentration as determined in the
HTS assay using a dose response in 1536-well format. (n = 2 for lactate and n = 3 for
cytotoxicity, CETSA and biochemical LDHA). Efficacy < 20 % with curve class 4, therefore
50
b
it should be considered as inactive.
1
9
Previous SAR suggested that the 3-biphenyl moiety provided significant levels of enzyme
and cellular potency. However, this functional group failed to provide the pharmacokinetic
profile needed for evaluation in in vivo studies. An initial assessment showed no clear SAR
trend regarding the presence of a 4-fluoro group (7 and 10, Table 1). Consequently, we
employed both the 4-fluoro and 4-H for further modifications of the 3-phenyl group in this
region. In addition to extensive SAR in the biphenyl series, to be published in due course, we
explored alkynes and alkenes as phenyl isosteres. We rationalized that alkynes, with a linear
1
5 | P a g e
ACS Paragon Plus Environment

Page 17 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
conformation and minimal steric bulk, could present the terminal groups to a hydrophobic
region in the enzyme, and potentially provide tighter binding. In an initial SAR exploration, a
simple terminal alkyne 11 (Table 2) retained biochemical activity relative to 10 (Table 1),
with only modest loss in potency toward lactate output (< 2 fold) and cytotoxicity (< 3 fold)
in both cell lines. We thus explored a variety of substitutions on the alkyne to better understand
the SAR in this hydrophobic space in LDH (Tables 2 and 3). Initially, we explored the terminal
alkyne region with numerous alkyl and cycloalkyl groups, which could occupy the
hydrophobic pocket. Most changes were tolerated, with steadily improved cell-based activities
noted as the size of the alkyl group increases (e.g. 16-17 and 20). Introducing electron-
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
withdrawing polar groups, particularly terminal -CN (33) and -CF groups (34) which are
3
strong hydrogen bond acceptors, markedly reduced cellular activity. A similar trend of lower
cell potency was observed for analogs 18-19 and 35-36, which also possess electron-
withdrawing groups capable of hydrogen bond interactions. Introducing cycloalkyl groups
onto the terminal alkyne region resulted in very potent compounds in the biochemical assay,
but the activity in cellular assays was more variable. Groups such as cyclopropyl (21),
cyclobutyl (22), and cyclopentyl (23 & 51) showed promising potency increases in the cellular
assays. When a methylene linker was introduced (37 vs 21 and 38 vs 23), potency in the lactate
inhibition assay increased or remained similar while less potency was observed in the
cytotoxicity assays. Interestingly, when the cycloalkyl groups contained hydrogen bond
acceptor or donor atoms, biochemical potency was maintained, but cellular activities
significantly dropped (24-32 and 39-41 in Table 2), consistent with our previous observations.
As the SAR indicated a preference for lipophilic groups at the terminal position of the alkyne,
we synthesized analog 42, which replaced the terminal alkyl group with a 2-thienyl group.
Though analog 42 was less potent in the biochemical assay (34 nM), it showed improved
cellular potency (Table 2) in lactate (438 nM in A673; 307 nM in MiaPaca-2), and cytotoxicity
1
6 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 95
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
assays (265 nM in A673; 268 nM in MiaPaca-2) compared to 10. As an unsubstituted thienyl
2
4
moiety could be metabolically labile , we incorporated a 5-methyl thienyl group, which
retained similar potency in the cellular assays (43 and 52, which also are named as NCATS-
SM1440 and NCATS-SM1441 respectively). Replacing the methyl group of the thiophene
with other alkyl groups (45, 55-57) maintained the potency; however, bulkier substituents
such as a t-butyl group decreased potency (45). Moving the 5-methyl group to the 3-position
of the thiophene resulted in similar cellular potency (53), whereas switching to the 5-
methylthien-3-yl substitution pattern (54 and 77) slightly decreased activity. A 3,5- (66) or
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
,5- dimethyl substitution pattern (67 & 83) on the thiophene further improved cellular
potency. Interestingly, introduction of halogens onto the 5-alkylthiophene significantly
reduced the cellular potency (58-60). Replacing the thiophene with similarly sized
heterocycles was well tolerated. For instance, a 5-methylfuran (44 and 61) showed similar
inhibition of lactate production and cytotoxicity in both cell types compared to the thiophene
analogs (43 and 52). Introducing a methylthiazole (46-47, 63-65, and 80-82) or methyloxazole
(
62) also retained biochemical and cellular potencies. The more hydrophilic methyl imidazole
48) or methylpyrazole (49) analogs showed good biochemical potency but less favorable
(
cellular effects in lactate and cytotoxicity assays.
Table 2. Biochemical and Cellular LDH Inhibition of Analogs 11−49.
2
SO NH2
F
R
N
CO2H
N
N
S
F
11-49
A673 cells
HEK293
Cells
MiaPaCa-2 cells
Biochemica
l LDHA
Compd
No.
lactate
lactate
inh.
R
Cytotox.
Cytotox. CETSA
inh.
IC50
nM)
IC50
IC50
(nM)
IC50
(nM)
a
IC (nM)
IC50
5
0
a
a
a
(nM)
a
a
(
(nM)
1
7 | P a g e
ACS Paragon Plus Environment

Page 19 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
2
H
CH3
9
902
505
487
420
405
4346
332
716
833
773
660
450
470
361
298
4025
322
662
746
845
253
119
112
106
106
11
14
24
50
13
Et
n-Pr
n-Bu
1
1
4
5
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
6
21
16
284
141
1011
935
487
137
756
845
2053
75
1
1
7
8
3
8
2475
769
1798
1225
1659
525
1327
1092
577
119
OH
OMe
19
2
2
0
1
54
9
284
695
378
935
132
593
487
554
428
909
323
573
89
100
133
22
23
2
2
2
2
2
2
3
4
5
6
7
8
22
33
7
509
450
100
689
333
454
314
593
112
95
F
1931
1052
1324
12241
1268
1056
5937
2187
1177
2647
3587
973
1479
898
2178
1236
2263
2994
1224
882
150
84
HO
O
O
10
7
1047
4873
831
126
1499
89
F
O
6
HO
29
7
O
O
3
3
0
1
1
1054
2635
867
201
815
2579
F
O
5
3871
1738
HO
32
NH
15
12241
4010
9052
8910
4401
17302
3
3
3
3
4
5
CN
CF3
15
450
15
>30000
>30000
474
14005
163
>30000 >30000 >30000 >30000
CF2H
F
937
1482
741
1327
3
6
14
800
1134
822
1020
127
3
3
7
8
31
202
8
234
487
1016
939
259
424
1030
848
89
178
267
39
1086
973
1007
1102
N
1
8 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 95
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
40
9
933
15349
438
1380
5834
265
937
18852
307
1225
18061
268
178
5045
89
O
O
4
4
1
2
107
32
N
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
S
43
(NCATS-
SM1440)
57
873
119
403
257
106
S
4
4
47
544
528
547
552
79
O
4
5
713
529
743
507
665
10602
S
N
4
4
6
7
33
52
937
569
803
867
488
909
-
S
N
1324
1603
170
S
N
4
4
8
9
8
9
1597
1603
1542
593
1066
1141
1387
619
299
189
N
N
N
a
IC values represent the half maximal (50%) inhibitory concentration as determined in the
5
0
HTS assay using a dose response in 1536-well format. (n = 2 for lactate and n = 3 for
cytotoxicity, CETSA and biochemical LDHA).
Table 3. Biochemical and Cellular LDH Inhibition of Analogs 50−67.
SO NH2
2
F
R
N
CO2H
N
N
S
50-67
HEK29
Cells
Compd
Biochemic
A673 cells
MiaPaCa-2 cells
3
1
9 | P a g e
ACS Paragon Plus Environment

Page 21 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
lactat Cytoto lactat Cytoto
No.
al LDHA
R
CETSA
e inh.
x
e inh.
IC50
(nM)
x
a
IC50
IC (nM)
5
0
IC50
nM)
IC50
(nM)
695
IC50
(nM)
619
a
(
nM)
150
89
a
a
a
a
(
5
0
CH3
9
713
525
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
51
47
298
760
419
281
377
5
2
(
NCATS
4
0
105
367
347
119
-
S
SM1441)
5
3
101
87
390
494
419
359
417
657
452
332
119
95
S
54
S
5
5
5
5
6
7
385
201
169
296
403
333
528
332
347
203
434
491
590
342
905
179
183
270
S
S
S
58
167
154
434
36
218
345
485
450
452
528
833
285
408
286
337
432
556
528
803
296
113
134
113
168
Cl
S
5
6
9
0
CF H
2
S
CF3
S
61
O
O
S
N
6
6
6
6
2
3
4
5
12
17
43
13
525
527
713
552
309
532
347
450
487
671
360
641
471
119
90
N
N
720
38
S
S
1174
345
1180
101
N
6
6
6
7
137
235
528
716
218
218
533
810
201
299
S
S
161
2
0 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 95
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
a
IC values represent the half maximal (50%) inhibitory concentration as determined in the
5
0
HTS assay using a dose response in 1536-well format. (n = 2 for lactate and n = 3 for
cytotoxicity, CETSA and biochemical LDHA).
We also examined the replacement of the alkyne linker (68-73, Table 4) with an alkene motif.
This slightly diminished the cellular potency despite low nanomolar biochemical activity. The
alkene analogs 68 and 69 showed marginally inferior cellular potency compared to
corresponding alkyne analogs 52 and 23. A similar trend was also observed for other matched
pairs, including for alkene analog 70 vs. alkyne 21, 71 vs. 12, and 72 vs. 17.
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 4. Biochemical and Cellular LDH Inhibition of Compounds 68−73.
2
SO NH2
2
SO NH2
F
F
Y
N
CO2H
N
CO2H
N
N
N
N
F3C
S
S
68-72
73
X
HEK293
Cells
A673 cells
MiaPaCa-2 cells
Biochemical
Comp
d No.
Y
X
LDHA
lactate
lactate
Cytotox
inh
Cytotox
inh
IC50
nM)
CETSA
a
IC (nM)
IC50
IC50
(nM)a
a
50
IC50
(nM)
IC (nM)
nM)a
50
(
a
a
(
6
6
8
9
H
F
52
74
357
371
452
331
149
471
454
158
108
S
488
70
71
F
F
15
11
320
405
833
614
231
321
837
528
134
126
CH3
7
7
2
3
F
-
59
39
361
713
2263
255
972
1899
112
398
-
16023
17376
2
1 | P a g e
ACS Paragon Plus Environment

Page 23 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
IC values represent the half maximal (50%) inhibitory concentration as determined in the
5
0
HTS assay using a dose response in 1536-well format. (n = 2 for lactate and n = 3 for
cytotoxicity, CETSA and biochemical LDHA).
Recalling the improved cellular potency of earlier analogs with the ethylcyclopropane
pyrazole substitution, we synthesized analogs 74-83 (Table 5) combining this substitution
with optimal groups from the terminal alkyne region. As summarized in Table 5, analogs 74-
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
8
3 showed a noteworthy improvement in cellular potency when compared to their
corresponding cyclopropylmethyl analogs mentioned above.
Table 5. Biochemical and Cellular LDH Inhibition of Analogs 74−89.
2
SO NH2
2
SO NH2
F
F
R
R
N
CO2H
N
CO2H
S
N
N
N
N
S
S
74-83
84-89
X
Bioche
HEK293
Cells
A673 cells
MiaPaCa-2 cells
lactate
m
Compd
No.
R
X
LDHA
lactate
inh
Cytotox
IC50
(nM)
Cytotox CETSA
inh
IC50
(nM)a
IC50
(nM)
IC50
(nM)
IC50
IC50
a
a
a
(nM)a
nM)a
(
7
7
4
5
F
H
H
202
390
174
547
218
187
137
197
172
417
226
171
137
200
152
95
S
117
52
S
76
77
O
H
H
160
306
434
434
197
320
254
387
223
265
142
189
S
7
8
S
2
2 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 95
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
7
8
8
8
9
0
1
2
S
H
H
H
H
636
33
572
364
434
300
571
180
218
219
379
417
334
253
496
173
218
219
299
84
N
S
S
N
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
S
24
17756
90
24
83
H
487
162
402
188
402
299
S
84
85
86
87
88
89
-
-
-
-
-
-
874
4540
307
685
18264
346
968
5899
590
1027
5144
409
9350
7427
1049
469
708
10751
251
122
485
526
459
175
1174
5270
364
4686
26351
515
5258
20931
1027
3183
1864
1219
a
IC values represent the half maximal (50%) inhibitory concentration as determined in the
5
0
HTS assay using a dose response in 1536-well format. (n = 2 for lactate and n = 3 for
cytotoxicity, CETSA and biochemical LDHA).
1
9
In our previous report , we described several crystal structures with similar analogs bound to
LDHA, establishing the binding mode of this series of compounds within the LDHA catalytic
site, and helping to guide our lead optimization campaign toward the identification of
compounds with improved potency. As part of the optimization campaign, we obtained co-
crystal structures of analogs 23 and 52 bound to LDHA in the presence of cofactor NADH
(
Figure 2). Analogs 23 and 52 present similar binding poses as the previously reported
2
3 | P a g e
ACS Paragon Plus Environment

Page 25 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
molecules, with the carboxylate making a clearly defined salt bridge interaction with Arg168
and a hydrogen bond with Thr247. Additionally, the cyclopropylmethyl group and
cyclopropylethyl group both maintain a potential pseudo π−π interaction between the
cyclopropyl group and active-site tyrosine Tyr238; this substitution was found to be vital for
potent cellular activity. As demonstrated previously, the 4-benzyl sulfonamide motif makes
well-defined H-bonding interactions with side chains of Asp140 and Glu191 as well as with
the main chain nitrogens of Asp140, Ile141, and Glu191 to confer potent biochemical and
cellular activities. Further, the cyclopentylalkyne in analog 23 and 5-methyl thienyl alkyne
analog 52 presented similar binding poses compared to the biphenyl ortholog. The linear rigid
confirmation of the alkynes enables the presentation of moieties that offer increased
hydrophobic interactions with LDHA while offering lower lipophilicity compared with a
phenyl ring. It is unclear from the crystal structure exactly how some of these variations
improve cellular potency. For example, the 2-fluoro moiety on the benzenesulfonamide ring
is oriented in the same way in both structures. Although this group does occupy a small sub-
pocket in the enzyme, no specific interactions with the protein are evident. The changed
binding kinetics that this substitution produces (vide infra) may instead result from subtle
electronic modulation of the hydrogen bonding interactions, which ultimately provided better
cellular potency. These efforts demonstrate that a multi-parameter optimization strategy that
utilizes structure-based design combined with targeted improvement of residence time, can
elicit significant improvements in cellular potency, at least in the context of LDHA inhibitors.
All of the synthesized analogs were profiled for their biochemical activity against both LDH
isozymes, LDHA and LDHB, and several compounds were tested against a related ‘off-target’
dehydrogenase, malate dehydrogenase (MDH). All analogs possess nearly identical potency
against LDHB compared to LDHA (Table S1) and thus display a highly significant correlation
coefficient (Figure S1). Recent reports allude to the need for a pan LDH inhibitor in order 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
2
4 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 95
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
25
elicit a cellular phenotype , highlighting a possible advantage of this profile. All the analogs
showed little or no biochemical activity against MDH, with over 2,500-fold selectivity for
LDHA/LDHB (Table S1).
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: Crystal structure of LDHA bound to inhibitor 23 (A, pdb code 6Q0D) and 52 (B,
pdb code 6Q13). LDHA is shown in ribbon (blue) and key residues in the active site are shown
in green. Small molecule inhibitors are shown in sticks with salmon- and magenta-colored
carbons.
While compounds inactive in the biochemical assay elicit no cellular activity, indicating the
on-target nature of the observed effects, we noted only a marginal correlation between the
biochemical LDHA inhibition and inhibition of cellular lactate production throughout this
optimization campaign. This poor correlation may have hampered prior efforts to identify cell-
active LDHA inhibitors. The unusual trend and poor correlation between biochemical and
cellular potency may be due to the disparity in the concentration of LDHA in the biochemical
assay (2 nM) versus the extremely high concentration of intracellular LDH (estimated to be
in the range of 2−17 μM) in glycolytic tumor cells. Furthermore, the nature of the active site
preferentially selects for highly lipophilic compounds with an acidic moiety such as a
carboxylic acid, which can introduce additional permeability and protein binding challenges
that might reduce cellular efficacy and thus influence the correlation. Though structure-guided
optimization was crucial in driving the cellular potency, it was also difficult to explain the
2
5 | P a g e
ACS Paragon Plus Environment

Page 27 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
drastically improved cytotoxic effects for certain analogs as a result of subtle structural
changes during our optimization.
To address such challenges, we reasoned that a highly potent compound with a long drug-
target residence time might more effectively inhibit cellular LDH activity. To better
understand the reason for improved cellular activity emerging from relatively small structural
changes, we complemented the structure-guided SAR with data from an SPR binding assay
to analyze the top cell-active compounds, in search of compounds with longer off-rates. A
summary of SPR data and the corresponding cell data is outlined in Table 6 for compounds
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
3, 38, 43, and 52. All analogs showed more potent binding affinities and longer off-rates
(k ) compared to 1. Analysis of a larger data set across the series (including unpublished
off
data) suggests that the compounds with long off-rates and residence times (τ) [calculated as
−
1
1
/k (s )] offer high cellular potency in both the lactate and cytotoxicity assays. The most
off
-6
-
potent compounds exhibited off-rates that exceeded the detection limit of the assay, 10 sec
1.
Comparing the data presented here with our previously published results suggests that the
2
-fluoro group on the sulfonamide phenyl ring and the alkyne substitution both positively
contribute to longer residence times, leading to improved binding affinity and more potent
cellular activity.
A good correlation was observed for the IC values for our compounds in MiaPaCa-2 and
5
0
A673 cell lines, in both lactate secretion and cytotoxicity assays (Figure S2); however, the
correlation between inhibition of lactate secretion and cytotoxic effect within the same cell
line was poor (Figure S3). As such, a superior lactate inhibition profile did not always translate
into improved cytotoxic effects for some compounds. The reason for such discrepancy is
unclear but may reflect the differences in the two assay conditions. For example, in the lactate
secretion assay, lactate output was measured after 30 min exposure to compound, while
cytotoxicity was assessed after 48 h of incubation with compound. It is possible that the ability
2
6 | P a g e
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 95
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
of compounds to display higher cytotoxic effects depends on a longer residence time along
with higher lactate inhibition.
Given the potential for the inhibitors to alter lactate production or viability due to unidentified
off-target activity, we evaluated our LDHA inhibitors in a recently developed split Nano
Luciferase (NLuc) high-throughput cellular thermal shift assay (SplitLuc-CETSA) to obtain
concentration dependent measurements of intracellular target engagement. We have
previously demonstrated the applicability and performance of our SplitLuc CETSA assay to
detect LDHA target engagement, in both 384 and 1536-well plate formats, for a set of well-
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
6
validated inhibitors . Accordingly, we evaluated the majority of the analogs utilizing the
SplitLuc-CETSA technique in HEK293T cells (Table1-5) where LDHA was appended with a
C-terminal 15-amino acid HiBiT fusion tag. A majority of the active compounds in the lactate
assay showed cellular binding and thermal stabilization of LDHA in the SplitLuc assay. In
general, we observed a good correlation between CETSA IC and lactate IC (Figure S4)
5
0
50
though we used different cell lines in the two assays. Our lead compounds 42 and 53, along
with many other compounds in the lactate assay, demonstrated stabilization of LDHA within
cells. Though biochemically active compounds tended to show good CETSA activity, no clear
correlation emerged between the cellular stabilization of LDHA via CETSA and the
biochemical IC50. As we stated before, a similar disconnect was observed between
biochemical and lactate IC . As such, biochemical LDHA screening appears to be a poor
50
predictor of cellular effects in this series, the SplitLuc-CETSA cell-based screening platform
provides a new avenue for HTS screening in a cellular context, particularly for LDHA.
Table 6. k and Residence time (τ) data correlation to cell activity for selected analogs.
off
A673 cells
MiaPaCa-2 cells
KD
a
Residence Off rate
Cpd. (nM)a
lactate inh cytotox
lactate inh cytotox
4
Time τ (h) k (x10 )
off
a
a
a
IC (nM)
IC (nM)
IC (nM)
IC (nM)
5
0
50
50
50
1
0.11
0.35
8
450
2454
636
8448
2
7 | P a g e
ACS Paragon Plus Environment

Page 29 of 95
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
3
3
8
0.0193
0.1235
277
277
277
277
0.01
509
487
873
760
100
333
424
403
367
314
848
257
347
0.01
0.01
0.01
939
119
105
43 0.0667
0.0634
5
2
a
K and k were determined via SPR on a Mass-1 instrument from Sierra Sensors and
D
off
−
1
residence time τ was calculated as 1/k (s ).
off
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 assess the impact of the LDH inhibitors on the glycolytic pathway, the Glycolysis Stress
Test (GST) was performed in A673 cells (Figure 3). Glycolytic flux is assessed by measuring
the extracellular acidification rate of the media (ECAR) that results from glycolysis-dependent
proton production by the cells. As shown in Figure 3, after a stable baseline is recorded, the
cells were treated with compounds (43 and 52-Figure 3A,B) at increasing concentrations and
the ECAR is measured for approximately 50 min before injection of glucose, which results in
a substantial stimulation of glycolytic flux and proton production. After three further
measurements, oligomycin is injected to inhibit mitochondrial oxidative phosphorylation,
which results in a compensatory increase in ECAR that is then suppressed to basal levels by
the hexokinase inhibitor 2-deoxyglucose. From this experiment, two glycolysis-dependent
parameters can be obtained; the basal glucose-dependent ECAR (glycolysis Figure 3C) and
the maximal glycolytic capacity (shown in Figure 3B). As anticipated, as LDH inhibition
increases with the concentration of compounds 43 or 52, the reserve biochemical capacity of
+
LDH within the glycolytic pathway is exceeded, resulting in a depletion of NAD and,
ultimately, inhibition of the entire pathway, as evidenced by decreased ECAR.
2
8 | P a g e
ACS Paragon Plus Environment