Evolution of a 4-Benzyloxy-benzylamino Chemotype to Provide Efficacious, Potent, and Isoform Selective PPARα Agonists as Leads for Retinal Disorders

Article abstract of DOI:10.1021/acs.jmedchem.9b01189

Peroxisome proliferator-activated receptor alpha (PPARα) is expressed in retinal Müller cells, endothelial cells, and in retinal pigment epithelium; agonism of PPARα with genetic or pharmacological tools ameliorates inflammation, vascular leakage, neurodegeneration, and neovascularization associated with retinal diseases in animal models. As such, PPARα is a promising drug target for diabetic retinopathy and age-related macular degeneration. Herein, we report proof-of-concept in vivo efficacy in an streptozotocin-induced vascular leakage model (rat) and preliminary pharmacokinetic assessment of a first-generation lead 4a (A91). Additionally, we present the design, synthesis, and evaluation of second-generation analogues, which led to the discovery of 4u and related compounds that reach cellular potencies <50 nM and exhibit >2,700-fold selectivity for PPARα over other PPAR isoforms. These studies identify a pipeline of candidates positioned for detailed PK/PD and pre-clinical evaluation.

Full text of DOI:10.1021/acs.jmedchem.9b01189

Subscriber access provided by Karolinska Institutet, University Library
Article
Evolution of a 4-Benzyloxy-benzylamino Chemotype to Provide Efficacious,
Potent, and Isoform Selective PPAR# Agonists as Leads for Retinal Disorders
Xiaozheng Dou, Dinesh Nath, Henry Shin, Elmar Nurmemmedov,
Philip Bourne, Jian-xing Ma, and Adam S Duerfeldt
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b01189 • Publication Date (Web): 25 Feb 2020
Downloaded from pubs.acs.org on February 28, 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 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Evolution of a 4-Benzyloxy-benzylamino Chemotype to Provide Efficacious,
Potent, and Isoform Selective PPARα Agonists as Leads for Retinal Disorders
Xiaozheng Dou,†,§ Dinesh Nath, Henry Shin, Elmar Nurmemmedov, Philip C. Bourne, Jian-Xing Ma,
†,§
‡
‖
§
‡
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
and Adam S. Duerfeldt*†,§
†
§
Institute for Natural Products Applications and Research Technologies and Department of Chemistry &
Biochemistry, University of Oklahoma, 101 Stephenson Parkway, Stephenson Life Sciences Research
Center, Norman, Oklahoma 73019, United States
‡Department of Physiology, University of Oklahoma Health Sciences Center, 941 Stanton L. Young
Boulevard, Oklahoma City, Oklahoma 73104, United States
‖John Wayne Cancer Institute & Pacific Neuroscience Institute at Providence Saint John’s Health Center,
2
200 Santa Monica Boulevard, Santa Monica, California 90404, United States
*
To whom correspondence should be addressed: adam.duerfeldt@ou.edu | 405-325-2232
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 70
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
ABSTRACT
Peroxisome proliferator-activated receptor alpha (PPARα) is expressed in retinal Müller cells, endothelial
cells, and in retinal pigment epithelium, and agonism of PPARα with genetic or pharmacological tools
ameliorates inflammation, vascular leakage, neurodegeneration, and neovascularization associated with
retinal diseases in animal models. As such, PPARα is a promising drug target for diabetic retinopathy and
age-related macular degeneration. Herein, we report proof-of-concept in vivo efficacy in an STZ-induced
vascular leakage model (rat) and preliminary pharmacokinetic assessment of a first-generation lead 4a
(A91). Additionally, we present the design, synthesis, and evaluation of second-generation analogs, which
led to the discovery of 4u and related compounds that reach cellular potencies <50 nM and exhibit >2,700-
fold selectivity for PPARα over other PPAR isoforms. These studies identify a pipeline of candidates
positioned for detailed PK/PD and pre-clinical evaluation.
0
1
2
3
4
5
6
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
Keywords: PPARα, age-related macular degeneration, diabetic retinopathy, PPAR selectivity, structure-
based design
ACS Paragon Plus Environment

Page 3 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Diabetic retinopathy (DR) is a common complication of diabetes (~30% of diabetics) and is
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
the leading cause of blindness in the working population. Currently, >40% of the patient
population fails to respond to the gold-standard treatment, direct intraocular injection of vascular
2
endothelial growth factor (VEGF) antibodies. New therapies that are superior or complementary
to current approaches would be of great value to patients, especially for those that remain
refractory to current options. Ideally, unlike current treatment options, new approaches should be
non-invasive (to the eye), affordable, and not reliant on specialized facilities.
The promise of peroxisome proliferator-activated receptor alpha (PPARα) agonism as a novel
strategy for treating DR has been confirmed in two independent human clinical trials (FIELD and
ACCORD), wherein orally administered Fenofibrate (Feno), a clinically used drug for
hyperlipidemia, demonstrated robust protective effects against DR and retinal neovascularization
(NV) in type 2 diabetic patients.3-4 This protection greatly reduced the need for patients to undergo
disease modifying laser photocoagulation therapy. Since the report of these results, mechanistic
and genetic studies have determined that the protective effects of Feno are unrelated to its lipid-
lowering activity, but rather result from agonism of PPARα by the Feno metabolite fenofibric acid
(
FenoFA).5-6 Feno/FenoFA however, suffer from low ocular distribution, low affinity and selectivity
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 70
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
for PPARα, and dose-limiting toxicities, all of which will likely limit its use as a DR therapy. Proof-
of-concept, however, for targeting retinal PPARα with systemically available small molecules has
been established with Feno and revealed that small molecule agonists with improved PK/PD
profiles have high promise to become first-in-class non-invasive treatment options for DR and
related ocular conditions (e.g., age-related macular degeneration and retinopathy of
prematurity).7-8
0
1
2
3
4
5
6
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
As important members of the nuclear receptor superfamily, PPARs play essential roles in
regulating cellular differentiation, development, metabolism, and tumorigenesis of higher
organisms. Thus, these receptors have been considered attractive targets for treating metabolic
disorders and inflammatory conditions.9-10 PPARs consist of 3 subtypes: PPAR, PPARγ and
PPARδ/. Although the ligand-binding domains share high sequence similarity (60-70%), the
isoforms exhibit distinct tissue distribution, biological functionality, and can be targeted
selectively.11 Within this family, PPARα has been implicated in hyperlipidemia, diabetes and
inflammatory disorders; and agonism of this isoform provides pharmacological benefits for these
conditions.12-13 Only recently, however, have the roles of PPARα in regulating VEGF,
inflammation, mitochondrial function, apoptosis, and NV in diabetic retinae been revealed, thus
elucidating a new avenue for PPARα agonists as therapeutics for oculovascular diseases.14-15
ACS Paragon Plus Environment

Page 5 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Specifically, previous studies demonstrate that PPARα knockout alone results in retinal
mitochondrial dysfunction, a declined electroretinogram (ERG) response, and retinal cell
apoptosis.16 Furthermore, PPARα levels in the retina are decreased in diabetic animal models
and in diabetic humans and that agonism of PPARα restores balance.14-15 Although PPARα
agonism was intensely investigated in the 1990s as a therapeutic approach for dyslipidemia, and
numerous potent agonists exist in literature, none of the known chemotypes have been evaluated
in animal models for efficacy against retinal conditions. Rather than attempting to repurpose
known compounds, often lacking full disclosure of physicochemical and PK/PD profiles, we
decided to embark on a discovery process to identify and develop novel PPARα agonistic
chemotypes as new compositions for DR and AMD leads.
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
Recently, we published first-generation structure activity relationship (SAR) studies on a new
PPARα agonistic chemotype, 4-carboxy-2-phenylquinolines, represented by Y-0452 (Figure 1).17
Guided by in-silico studies, we proposed that deconstruction of the rigid quinoline core would
provide a more synthetically tractable and flexible scaffold, expected to exhibit improved
complementarity to the U-shaped ligand binding pocket. Based on this design strategy, we
developed the 4-benzyloxy-benzylamino chemotype to assess our hypothesis. In general, this
small molecule family exhibits improved potency and selectivity over both Y-0452 (initial hit) and
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 70
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
FenoFA, thus validating our rationale for further development of this chemotype and pursuing
proof of concept studies. Compound A91 (Figure 1), a representative analogue from this series
was advanced through rigorous biochemical evaluation to demonstrate typical downstream
responses of PPARα agonism including PPARα upregulation, induction of target genes (e.g.,
0
1
2
3
4
5
6
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
17
Acadm, Cpt1a, Fabp3, and Slc25a20) and attenuation of cell migration. Having identified a new
class of PPARα agonists and demonstrating useful SAR, the purpose of the work presented
herein was to assess the preliminary PK and in vivo protection against vascular leakage of A91
to establish proof-of-concept and develop second-generation derivatives aimed at improving
potency and selectivity of this chemotype.
OMe
functionalize
C
O
first-
generation
H
B
second-
generation
Cl
N
N
Goals:
- initial PK studies
- POC in vivo efficacy
- improve potency
- enhance selectivity
A
remove
rigidity
transpose
carboxylic acid
A91
This work
HO
Y-0452
PPAR EC50 = 25-50 M
O
HO
O
PPAR EC50 = 6 M
>20-fold selective
for PPAR
Figure 1. Overall strategy for hit optimization and the objective of the studies presented herein.
RESULTS AND DISCUSSION
PK/PD Profile of First-Generation Lead. To gain insight about the metabolic stability and rate of
clearance, both of which influence bioavailability, A91 was assessed for intrinsic clearance in
ACS Paragon Plus Environment

Page 7 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
human and rat (Sprague-Dawley) liver microsomes. As shown in Figure 2A, compound A91
proved to be rather stable, with the mean half-life (t ) exceeding 60 minutes in both human and
1/2
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
rat microsomes. These results indicate that this chemotype is likely to exhibit low clearance and
that phase-I hepatic metabolism is unlikely to limit bioavailability.
A
C
B
OMe
O
O
H
N
HO
A91
Figure 2. Pharmacokinetic profile of A91. (A) Stability in liver microsomes. A91 was pre-incubated with pooled liver
microsomes in phosphate buffer (pH 7.4) for 5 min at 37 °C. The system was activated by the addition of an NADPH-regenerating
system and incubated for 0, 15, 30, 45, and 60 min before being quenched with an acetonitrile/methanol mixture. Samples were
processed and analyzed by HPLC-MS/MS to determine the peak area remaining. Experiments completed in duplicate and mean
values are depicted. (B) Assessment of A91 (10 μM) for irreversible inhibition of the major CYP450 drug metabolizing isoforms.
Peak areas corresponding to the metabolite of known substrates for each isoform are recorded. The percent of control activity was
calculated by comparing the peak area obtained in the presence of A91 to that obtained in the absence of A91. Percent inhibition
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 70
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
was calculated by subtracting the percent control activity from 100. Time-dependent CYP inhibition is demonstrated if the percent
inhibition from samples pre-incubated with the NADPH-regenerating system is larger than those without NADPH-regeneration.
Negative values are a reflection of the variability in the background noise of the experiment. Experiments completed in duplicate
and mean values are depicted. (C) Assessment of A91 for inhibition of the hERG channel. The degree of inhibition (%) was
obtained by measuring the tail current amplitude, which is induced by a one-second test pulse to -40 mV after a two-second pulse
to +20 mV, before and after drug incubation. Experiments completed in duplicate and mean values are depicted.
0
1
2
3
4
5
6
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
In parallel, A91 was assessed for evidence of time-dependent inhibition against the major
drug metabolizing cytochrome P450 isoforms (i.e., 1A, 2C9, 2C19, 2D6, and 3A). In this assay,
peak areas corresponding to the metabolite of control substrates were recorded in the presence
or absence of A91. This analysis was completed in the presence and absence of NADPH, to
reveal time-dependence of inhibition. Test compounds which exhibit higher inhibition of metabolite
formation in the absence of NADPH are deemed time-dependent inhibitors. As shown in Figure
2B, A91 exhibits no evidence of time-dependent inhibition for any of the CYP450s evaluated. This
observation is especially important for CYP3A, as this family is responsible for the metabolism of
statins, a drug class required for a majority of diabetics. These results indicate that CYP drug-
drug interactions due to irreversible or tight binding are unlikely with this chemotype.
Lastly, A91 was evaluated for inhibition of the human ether-á-go-go (hERG) potassium
channel. Inhibition of this channel is a flagship sign of cardiac toxicity. For this experiment, the
automated whole cell patch-clamp (Qpatch 48) technique was used to record outward potassium
ACS Paragon Plus Environment

Page 9 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
currents from a single cell in the presence and absence of the compound of interest. As shown
in Figure 2C, compound A91 exhibits low potency for the hERG channel (compounds that exhibit
≥50% inhibition at 10 μM are flagged as toxic), indicating that this chemotype is unlikely to exhibit
hERG-related cardiotoxicity, especially if potency for PPARα is improved. It is also worth noting
that A91 exhibits no measurable cytotoxicity in the retinal pigment epithelium photoreceptor cell-
line 661W at concentrations as high as 200 µM as measured in a colorimetric enzyme coupled
LDH detection cytotoxicity assay (supplementary).
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
In vivo Efficacy of First-Generation Lead. With a promising initial PK profile, and cellular
potency and selectivity surpassing FenoFA, we advanced A91 to proof-of-concept in vivo efficacy
studies in a well-established streptozotocin (STZ)-induced rat model of DR, which assesses the
ability of a compound of interest to attentuate retinal vascular leakage—a major culprit behind
diabetic macular edema and consequential vision loss. As shown in Figure 3A, systemic
administration (i.p.) of compound A91 reduces retinal vascular leakage in diabetic rats to the
range of the non-diabetic cohort at relatively the same dose as FenoFA. Of interesting note, A91
seems to lack signs of hepatomegaly (Figure 3B), a common side-effect of Feno/FenoFA
observed in rodent models but not in humans. These in vivo results paired with the initial PK
assessment, demonstrate that A91 1) exhibits in vivo efficacy in a relevant DR model following
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 70
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
systemic administration, 2) crosses the blood-retinal barrier and reaches the retina, 3) is
bioavailable, 4) survives first-pass metabolism and clearance mechanisms well enough to
maintain efficacy, and 5) demonstrates a relatively safe profile (no observable toxicity) after daily
i.p. injection for one-month.
0
1
2
3
4
5
6
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. (A) Effects on retinal permeability and (B) liver phenotype for compound A91 in comparison to FenoFA. 7-8 week-old
male Brown Norway rats were injected with streptozotocin (STZ, 55 mg/kg) to induce diabetes. Two weeks after STZ injections,
daily treatment (i.p. injection) with compound A91 or FenoFA commenced and lasted for 28 days. Student’s t-test was used to
compare groups for statistical significance and values are mean ± SD, # P<0.05 (vs. FenoFA), Retinal permeability was measured
using Evans blue as tracer and normalized by total retinal protein. *P<0.05 (vs. STZ-DMSO).
Inspired by promising PK and in vivo results for A91, we commenced second-generation SAR
studies with an aim of enhancing potency and isoform selectivity of this chemotype for PPARα. Since no
co-crystal structure of A91 existed at the onset of this work, we developed an in-silico model using the
Schrodinger Drug Discovery Suite to gain insight into the putative binding mode and guide our evolution
of this chemotype. For these studies we selected PDB 2P54, a co-crystal structure of a known PPARα
selective agonist GW590735 bound to human PPARα (hPPARα). To validate our docking approach,
ACS Paragon Plus Environment

Page 11 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
constraints, and parameters, GW590735 was extracted, exposed to MM2 energy-minimization, and re-
docked into the hPPARα ligand binding domain to ensure that the results reproduced the bound
conformation of the ligand. The overlay of co-crystallized and docked GW590735 showed excellent
congruence (RMSD = 0.155 Å). Maintaining the same constraints and parameters, A91 was docked into
hPPARα and the results were visually inspected for strategies to improve or introduce key interactions.
Since our gatekeeper assay is a cell-based assessment and thus not based on direct binding affinity, scores
were not used to prioritize ligands at this stage. Instead, visual inspection and structural modification
hypotheses dictated the prioritization of derivatives selected for synthesis and assessment.
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
Our approach focused on three key principles. First, maintain a suitably positioned
carboxylate (or isostere) to capture key hydrogen bonding interactions. Previous studies
demonstrate the significance of hydrogen bond interactions between hPPARα Ser280, Tyr314,
His440, and Tyr464 on the AF-2 helix and the carboxylate of known PPARα ligands.18-25 The
ability to interact with all four of these residues is believed to be responsible for triggering full
19
agonism of hPPARα. Poorer agonists tend to only interact with some of these hydrogen-bonding
partners. It is worth noting that while historical PPARα literature suggests a carboxylic acid is
required for activity (a pharmacophore), examples of PPARα agonists lacking this functionality
have been reported,13, 26 demonstrating the -COOH motif can be altered. Secondly, we wanted to
maintain the N-benzyl linkage to provide a complementary geometry for the U-shaped
architecture of the ligand-binding domain. Third, our in-silico studies consistently predicted the
aryl tail (C-ring) to bind a hydrophobic pocket comprised of helices 6 and 7 and defined by Ile241,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 70
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
Leu247, Ala250, Leu254, and Ala333. Literature precedence demonstrates that improvement of
this interaction with other PPARα chemotypes has proven beneficial to enhancing ligand potency
and selectivity.23-25
0
1
2
3
4
5
6
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
Stage I SAR: Linker Extension, C-Ring Modification, Carboxylate Location. The first stage
of our SAR (Figure 4) focused on assessing various hydrophobic substituents on the C-ring, probing the
effects of linker extension, and determining optimal location for the carboxylate motif (or isostere).
Stage I
O
R
n
O
H
Previous leads
N
m
O
H
*
His440
X
N
*
-
COOH at
3- or 4-position
4
5
a (A91), X = -COOH, EC50 = 4.42 M
a, X = -C(CH ) COOH, EC = 25.3 M
F
Tyr464 Stage I selection
3
2
50
Ser280
O
O
H
C-ring modification
Linker extension
Tyr314
N
HO
Carboxylate location
4b, EC50 of PPAR = 0.77 M
Stage II
HO
F
Methylation
F
Metabolism blocking
Isosterism
F
O
O
O
R3
O
O
H
N
R4 R5
H
N
R
N
HO
R1
R2
R
Stage III
F
R1 Best performer
F
F
Stage II selection
O
R4
O
O
O
R5
O
H
N
H
N
R2
N
HO
HO
R3
4u, EC50 of PPAR = 37 nM
24, EC50 of PPAR = 18 nM
Figure 4. Multi-stage approach to structural improvement of A91. Model depicts 4a (A91) docked to PDBID 2P54.
To generate stage I analogues, 4-hydroxybenzaldehyde (1) was coupled with various benzyl
bromides (2a-l) to yield the corresponding 4-benzyloxy-benzaldehydes 3a-l (Scheme 1A). These
precursors were then coupled with 3-aminobenzoic acid (or 4-aminobenzoic acid in the case of 4b*) via
ACS Paragon Plus Environment

Page 13 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
reductive amination to obtain 4a-4l and 4b*. Following a similar synthetic approach, 5a, 5b, 5b*, and 5f
that include the typical fibrate head-group were synthesized (Scheme 1A). The only difference in the
generation of 5a, 5b, 5b* and 5f was that the carboxylate remained protected as the ethyl ester until being
revealed in the final step following saponification. Our previous studies indicated that the 4-benzyloxy-
benzylamino chemotype deviates from the fibrates in that incorporation of the classical α-gem-dimethyl
carboxylate “head-group” leads to pan-agonism and decreases potency. While these affects are obviously
opposite to our goals, these analogues were generated to determine if this trend continued to propagate
through this series.
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 generate derivatives with extended linkers the synthesis began with the transformation of
primary alcohols 6a-b into the corresponding alkyl bromides 7a-b (Scheme 1B). These precursors were
then utilized to O-alkylate 4-hydroxybenzaldehyde to generate 8a-b, which were coupled with 3-
aminobenzoic acid through reductive amination to provide 4m and 4n that included an extra –CH – between
2
the B and C-rings (Scheme 1B). Analogue 4o with an additional –CH – between the A and B rings and the
2
B and C rings was also generated. Chemoselective alkylation of tyrosol with 7b afforded 9, which was
converted to the corresponding alkyl bromide 10 that was used to N-alkylate 3-aminobenzoic acid (Scheme
1
B).
Scheme 1. Synthesis of stage I analogues a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
A
4
2
1
R1
1
1
a, R = 4-OMe
g, R = 4-CN
O
1
1
b, R = 4-F
h, R = 2-F
H
1
1
N
c, R = 4-Cl
i, R = 3-F
1
1
b
d, R = 4-Br
j, R = 2,4-difluoro
1
1
e, R = 4-I
k, R = 3,4-difluoro
O
4a-l and 4b*
1
1
OH
f, R = 4-CF
l, R = 3,5-difluoro
3
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
OH
R1
2
R1
R1
a
O
+
1
O
O
Br
O
H
N
1
2a-l
3a-l
c, d
O
5a, 5b, 5b*, 5f
O
OH
B
O
Y
O
O
H
b
c
N
R1
O
HO
HO
R1
R1
Y = OH (6a-b)
Y = Br (7a-b)
8a-b
1
4
m, R = H
a
1
a
,
R = H
1
4
n, R = OMe
1
b, R = OMe
O
O
H
N
R1
d
e
Y
R1
O
Y = OH (9)
Y = Br (10)
1
4
o, R = OMe
a
aReagents and conditions: A) (a) K
CO , DMF; (b) i) 3-aminobenzoic acid or 4-amino benzoic acid, toluene, 130 °C, 2-3 h; ii)
2 3
NaBH(OAc)
3
, acetic acid, THF, 12 h, 32-97%; (c) i) ethyl 2-(3-aminophenoxy)-2-methylpropanoate or ethyl 2-(4-aminophenoxy)-
.
2
6
-methylpropanoate, toluene, 130 °C, 2-3 h; ii) NaBH(OAc)
0%. B) (a) PBr , diethyl ether; (b) 4-hydroxybenzaldehyde, K
, acetic acid, THF, 12 h, 96%; (d) tyrosol, K
3
, acetic acid, THF, 12 h; (d) LiOH H
2
O, THF/CH
3
OH/H
2
O, 12 h, 14-
3
2
CO
3
, DMF, 40-65%; (c) 3-aminobenzoic acid, toluene, 130 °C, 2-
3 h; ii) NaBH(OAc)
3
2
3 3
CO , DMF, 26%; (e) 3-aminobenzoic acid, Et N, DMF, 25%.
All yields reported are unoptimized. Asterisks indicate the -COOH motif is attached para instead of meta.
Stage I derivatives were evaluated for in vitro human PPARα (hPPARα) agonism in a commercially
available hPPARα luciferase cell reporter assay (Indigo Biosciences). This assay utilizes non-human
mammalian cells engineered to express hPPARα with the luciferase reporter gene functionally linked to a
PPARα-response promoter. Thus, quantifying changes in luciferase expression in compound treated
reporter cells provides a sensitive method to quantify changes in PPARα activity. As an initial first-pass
assessment, all compounds were evaluated at two concentrations, 5 µM and 50 μM. This two-concentration
approach is a cost-effective means to provide insight into 1) rough estimates of potency, 2) therapeutic
ACS Paragon Plus Environment

Page 15 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
window, and 3) relative levels of agonism. A 5:50 ratio <<1.0 indicates the compound is much more active
at 50 μM than 5 μM, meaning the compound has poor potency. A 5:50 ratio ~1.0 indicates equal activity at
both concentrations, meaning the EC50 is <5 μM and is non-cytotoxic to the cells at a 10-fold higher
concentration, as cell death results in a decrease in luminescence production. As one would presume, a 5:50
ratio of >1.0 indicates a potency <5 μM, but also demonstrates toxicity at 50 μM. Therefore, 5:50 ratios 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
~1.0 are desirable. GW590735, a highly selective (≥500-fold over PPARγ and PPARδ) PPARα partial
agonist was employed as a positive control.24
Table 1. In vitro Activity of 4a-la
R
O
H
N
O
OH
EC50
EC50
hPPARγ
(µM)
EC50
hPPARδ
(µM)
Ratio
Fold-
hPPARα
(5/50)b
Signalc
(µM)d
SIe
Compound
a (A91)
R
4
4-OMe
4-F
0.5
0.8
0.3
0.6
0.8
0.7
0.9
0.6
0.8
0.9
1.2
1.1
1.1
1.0
1.4
1.3
0.4
1.5
1.3
1.1
0.9
1.0
1.0
1.2
1.2
1.2
1.1
1.0
4.43±0.01
0.77±0.03
>100
>100
>100
>100
>20
4
4
b
>125
b*
4-F
4
c
4-Cl
4-Br
4-I
0.83±0.04
0.97±0.14
1.45±0.02
0.96±0.11
4d
4
e
4
f
4-CF
3
4
g
4-CN
2-F
4h
4
i
3-F
1.18±0.05
0.54±0.09
0.36±0.01
4
j
2,4-difluoro
3,4-difluoro
3,5-difluoro
>100
>100
>175
4k
4
l
GW590735
0.015±0.002
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 70
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
Rosiglitazone
GW0742
0.28±0.3
0.0019±0.002
aAll analogues contain the carboxylate head-group at the meta-position of the A-ring unless otherwise indicated. * indicates a para-
b
c
0
1
2
3
4
5
6
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
carboxylate. Ratio of relative light unit (RLU) signal at 5 μM and 50 μM compound concentrations. Ratio of maximal signal
d
(
RLU) strength observed for the compound of interest to that obtained with GW590735. EC50 values represent the mean ± SEM
e
of atleast two separate experiments performed in triplicate. SI = EC50 (PPARγ or PPARδ) / EC50 (PPARα). Blank cells indicate
compound was not selected for testing in the corresponding assay.
In line with our hypothesis, the incorporation of lipophilic or electron withdrawing substituents on
the C-ring of the benzoic acid chemotype (4b-l) provides an improvement in cellular activity in comparison
to A91 (4a). Most analogues in the 4b-l series exhibit 5:50 ratios >0.5, indicating EC values <5 µM (Table
5
0
1
). Comparison of 4b and 4b* demonstrates that the 3-carboxylate is favored over the 4-carboxylate. Within
this class, we advanced 8 derivatives to multi-point dose-response analysis in the same luciferase assay.
Compounds were chosen based on a combination of interest, 5:50 ratio, and level of agonism. As shown in
Table 1, the 4-F (4b), 4-Br (4d), 2,4-difluoro (4j), and 3,4-difluoro (4k) analogues all exhibit
submicromolar activity and are ~2-3 fold more active than the 4-I (4e) and 4-CF (4f). This result was
3
somewhat forecasted by our in-silico results, which predicted that substituents larger than a bromine at the
4
-position would be problematic. Even though the 3-F (4i) is less active than the 4-F (4b), it was still more
active than A91. Thus, it is not surprising that inclusion of fluorine at both the 3- and 4-positions (4k) of
this scaffold results in additive properties and is the most potent compound of this series. Inclusion of the
typical fibrate head-group (5a, 5b, 5b*, and 5f) once again leads to decreased potency (Table 2) but often
higher levels of agonism than the benzoic acid relatives (5a vs A91, 5b vs 4b, and 5f vs 4f). As shown in
Table 3, analogs containing extended linkers 4m–4o exhibited poor potency (4m and 4n) or poor activation
(
4o) in comparison to related non-homologated derivatives.
Table 2. In vitro Activity of 5a, 5b, 5b*, 5fa
ACS Paragon Plus Environment

Page 17 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
R
O
H
N
O
O
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
OH
Ratio
Fold-
(5/50)b
Signalc
Compound
R
5
a
OMe
0.1
0.2
0.6
0.2
1.8
2.0
1.8
1.2
5
5
b
F
F
b*
5
f
CF
3
aAll analogues contain the carboxylate head-group at the meta-position of the A-ring unless otherwise indicated. * indicates a para-
b
c
carboxylate. Ratio of relative light unit (RLU) signal at 5 μM and 50 μM compound concentrations. Ratio of maximal signal
(
RLU) strength observed for the compound of interest to that obtained with GW590735.
Table 3. In vitro Activity of 4m-o
R
O
O
n
H
N
HO
m
Ratio
Fold-
Compound
m
n
R
(5/50)a
Signalb
4
m
1
1
2
2
H
0.1
0.1
0.9
0.9
4
n
OMe
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
o
2
2
OMe
1.1
0.1
a
b
Ratio of relative light unit (RLU) signal at 5 μM and 50 μM compound concentrations. Ratio of maximal signal (RLU) strength
observed for the compound of interest to that obtained with GW590735.
From the stage I cohort, analogues 4b and 4j were selected to move forward to selectivity
assessment, wherein equivalent luciferase assays were conducted on isogenic cell-lines engineered to
overexpress either hPPARγ or hPPARδ. Neither compound produced significant levels of hPPARγ or
hPPARδ agonism up to 100 µM compound concentration, indicating a selectivity of >125-fold (4b) and
0
1
2
3
4
5
6
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
>
175-fold (4j) for hPPARα. Rosiglitazone, a known selective PPARγ agonist, and GW0742, a known
selective PPARδ agonist, were employed as positive controls in these isogenic cell-lines.
Stage II SAR: Strategic Methylation, Metabolism Blocking, Carboxylate Isosteric
Replacement. From the stage I results, compound 4b was selected as the base structure from this family
for advancement. Three strategies were leveraged to advance the SAR in stage II 1) methylation at strategic
positions within the scaffold that were expected to enhance complementarity with the shape and
hydrophobicity of the PPARα ligand binding pocket; 2) blockage of predicted sites of metabolism; and 3)
isosteric replacement of the carboxylate.
In silico prediction of major sites of metabolism for A91 with Schrodinger’s P450 Site of
Metabolism software identified the A-ring as a potential liability for CYP450 oxidation (supplementary).
As such, we synthesized 4p and 4q (Scheme 2) to interrogate the effects of fluorine blockage on activity.
These two analogues were generated by coupling 3b to 5-amino-2-fluorobenzoic acid (11a) or 5-amino-3-
fluorobenzoic acid (11b) to produce 4p and 4q, respectively. The 5-fluoro variant (4q) exhibits a ~3-fold
improvement in activity over the 4-fluoro derivative (4p) and is nearly equipotent to 4b (Table 4).
Interestingly, incorporation of a chlorine at the 5-position of the A-ring (4r) greatly decreases potency (5:50
ratio = 0.11), while incorporation of a bromine at the 5-position (4s) maintains potency in comparison to
4
b but introduces apparent cytotoxicity at elevated concentrations (5:50 ratio = 1.3). These results suggest
ACS Paragon Plus Environment

Page 19 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
that fluorine incorporation should be a viable strategy to block sites on the A-ring if proven to be
problematic as this series progresses through development.
Scheme 2. Synthesis of 4p-4sa
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
F
F
O
O
2
5
H N
O
2
a
O
OH
H
N
1
3
4
2
5
+
HO
6
3
1
6
O
R
4
3b
11a, R = 4-F 11c, R = 5-Cl
1b, R = 5-F 11d, R = 5-Br
R
4p,R = 4-F 4r, R = 5-Cl
4q,R = 5-F 4s,R = 5-Br
1
aReagents and conditions: (a) i) 11, toluene, 130 °C, 2-3 h; ii) NaBH(OAc)
, acetic acid, THF, 25 °C, 12 h, 78-86%.
3
Table 4. In vitro Activity of 4p-s
F
O
O
H
2
N
HO
3
1
6
4
5
R
EC50
hPPARα
(µM)c
EC50
EC50
hPPARδ
(µM)
Ratio
Fold-
hPPARγ
Compound
R
(5/50)a Signalb
(µM)
SId
4
4
p
q
4-F
5-F
0.8
1.1
0.1
1.3
1.2
1.1
1.1
1.4
1.61±0.06
0.58±0.04
>100
>100
>100
>170
>100
4r
5-Cl
5-Br
4
s
1.12±0.08
>100
a
b
Ratio of relative light unit (RLU) signal at 5 μM and 50 μM compound concentrations. Ratio of signal maximal signal (RLU)
c
strength observed for the compound of interest to that obtained with GW590735. EC50 values represent the mean ± SEM of atleast
d
two separate experiments performed in triplicate. EC50 (PPARγ or PPARδ) / EC50 (PPARα). Blank cells indicate compound was
not selected for testing in the corresponding assay.
2
7
Inspired by frequently reported methyl effects in drug discovery, analogues containing one or two
methyl groups at different positions were synthesized (Scheme 3) using variations of chemistry discussed
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 70
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
in Schemes 1 and 2 and evaluated in the luciferase assay (4t-z, Table 5). Specifically, we interrogated the
effects of methylation on the 2- and 3-position of the B-ring, the 4-position of the A-ring, the nitrogen
linker, and all possible variations on the benzylic position of the linker between the B- and C-rings (another
predicted site of metabolic liability). Interestingly, installation of a methyl group at the 3-position of B-ring
0
1
2
3
4
5
6
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
(4u) resulted in a drastic improvement in cellular activity, dropping the EC50 to ~40 nM, a ~21-fold
improvement from the des-methyl relative, 4b, and improving the selectivity for the PPARα isoform to
>2,700-fold (Figure 5). However, methylation at alternate sites (4t and 4v-z) failed to induce similar effects
on the activity, indicating that the improvement observed with 4u is not likely to arise from an increase in
general hydrophobicity but rather from improving aspects of target engagement.
Scheme 3. Synthesis of methylated derivatives 4t-za
ACS Paragon Plus Environment

Page 21 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
F
F
3
2
1
4
O
3
O
4
O
H
2
1
a
2
5
N
HO
3
O
1
6
R2
R1
R1
4
1
4t-v
3
a, R =H
1
1
2a, R =2-Me
1
F
F
1
2b, R =3-Me
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
O
O
O
H
N
b
N
HO
HO
4b
4w
F
O
CF3
OH
O
R3
N
N
c
4
3
1
4, R = Me, R = COCF
3
13
4
d
MeO
O
O
4
3
R O
O
4
x, R = H, R = H
F
CF3
OH
O
R3
N
N
e
4
3
1
5, R = Me, R = COCF
3
13
F
4
d
MeO
HO
O
4
3
R O
O
O
4
y, R = H, R = H
F
F
O
H
N
f
g
O
1
6
17
4z
HO
O
aReagents and conditions: (a) i) 3-aminobenzoic acid or 3-amino-4-methylbenzoic acid, toluene, 130 °C, 2-3 h; ii) NaBH(OAc)
acetic acid, THF, 25 °C, 12 h, 70-92%; (b) paraformaldehyde, NaBH(OAc) , EtOH, reflux 2 h, 25 °C, 12 h, 43%; (c) (R)-1-(4-
fluorophenyl)ethan-1-ol, di-tert-butyl azodicarboxylate, PPh , toluene; (d) LiOH H
(e) (S)-1-(4-fluorophenyl)ethan-1-ol, di-tert-butyl azodicarboxylate, PPh , toluene; f) 4-hydroxybenzaldehyde, TFAA, DBU,
CuCl , MeCN, 8%; (g) i) 3-aminobenzoic acid, toluene, 130 °C, ii) NaBH(OAc) , AcOH, THF, 12 h, 25%. All yields reported are
,
3
3
.
3
2 3 2
O, THF/CH OH/H O, 25 °C, 12 h, 65-75%.;
3
2
3
unoptimized.
Table 5. In vitro Activity of 4t-z
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 70
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
F
3
2
1
4
O
O
R3
N
R4 R5
HO
R1
R2
0
1
2
3
4
5
6
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
Ratio
Fold-
EC50
Compound
R1
R2
R3
R4
R5
(5/50)a Signalb hPPARα (µM)c
SId
4
t
2-OMe
H
H
H
H
H
H
H
H
H
H
0.8
1.4
0.2
0.8
0.1
0.2
0.2
0.8
0.5
1.0
1.2
1.2
0.3
1.7
0.6
1.6
0.8
1.3
1.4
1.0
1.45±0.06
4
u
3-Me
H
0.037±0.01
<2700
4
v
Me
H
H
H
4w
H
Me
H
H
0.90±0.08
25.6±3.5
4
4
x
y
H
H
Me
H
H
H
H
H
Me
Me
4z
H
H
H
Me
8.52±1.83
0.77±0.03
4.43±0.01
4
a (A91)
>20
4
b
>125
GW590735
0.015±0.002
a
b
Ratio of relative light unit (RLU) signal at 5 μM and 50 μM compound concentrations. Ratio of signal maximal signal (RLU)
c
strength observed for the compound of interest to that obtained with GW590735. EC50 values represent the mean ± SEM of atleast
d
two separate experiments performed in triplicate. EC50 (PPARγ or PPARδ) / EC50 (PPARα). Blank cells indicate compound was
not selected for testing in the corresponding assay.
Figure 5. Dose-response curves for 4u (blue squares, solid line) against PPARα, PPARδ, and PPARγ. Respective controls are
provided as black circles/dashed line.
ACS Paragon Plus Environment

Page 23 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
While the reason for the drastic improvement in activity arising from incorporation of the methyl
group at the 3-position of the B-ring is not obvious, we have interrogated this phenomenon briefly by in
silico methods. Results from these studies suggest that methylation at the 3-position (4u) results in a binding
shift that allows for improved hydrogen bonding with the Ser280, Tyr314, His440, and Tyr464 tetrad
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 6A). Additionally, 4u is predicted to bind with the aromatic ring of the benzoic acid perpendicular
to His440 thus allowing for T-shaped edge-to-face pi-interactions (Figure 6A). Methylation at the 2-
position (4t), however, produces no such shift and 4t overlays very well with the des-methyl analogue 4b
(Figure 6B). As such, it is not surprising that the activity of 4t and 4b are relatively similar.
A
B
His440
Tyr464
His440
Tyr464
Ser280
Ser280
Tyr314
Tyr314
Figure 6. Predicted binding modes of 4b, 4t, and 4u, demonstrating the possible effects of methyl installation on the 2- or 3-
position of the B ring. (A) Overlay of the predicted binding poses for 4b (yellow) and 4u (magenta). (B) Overlay of the predicted
binding poses for 4b (yellow) and 4t (cyan). PDBID: 2P54. The pdb file of the docking model is included in the Supporting
Information.
Coupling 3b to a variety of anilines containing carboxylic acid surrogates (18a-e) allowed us to
assess the effect of isosteric replacement on PPARα agonistic activity (19a-e, Scheme 4). As seen in Table
6
, replacement of the carboxylate with any of the isosteres resulted in dramatic decreases in activity, as
noted by either a low 5:50 ratio or low overall signal. The exception being the tetrazole (19d), which
exhibits decreased potency but comparable activation levels to the original carboxylate. This result confirms
that the carboxylic acid is a key contributor to triggering PPARα agonism but can indeed be modified. As
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 70
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
part of this series, the thio-variant of the fibrate head-group was also synthesized and assessed (19f).
Interestingly, this derivative exhibits an attenuated level of agonism compared to the parent analogue, 5b.
Differences in agonism levels now observed between the benzoic acid, fibrate head-group, and the thio-
fibrate head-group strongly suggest that manipulation of this motif may provide a useful means to tune the
level of agonism induced by a chemotype. This could be important as it is still unknown what level of
stimulation (partial or full-agonism) is optimal for PPARα to have a therapeutic effect in the retina.
0
1
2
3
4
5
6
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 4. Synthesis of isostere containing derivatives 19a-fa
F
O
NH2
H
a
N
X
1
8a-e
X
19a-e
O
F
F
O
b
O
O
H
H
S
N
S
N
O
HO
18f
19f
aReagents and conditions: (a) 3b, toluene, 130 °C, 2-3 h; ii) NaBH(OAc)
THF/CH OH/H O, 12 h, 13%.
.
3
2
, acetic acid, THF, 25 °C, 12 h, 8-80%; (b) LiOH H O,
3
2
Table 6. In vitro Activity of Carboxylate Isosteres
Ratio
Fold-
Compound
9a
X
(5/50)a Signalb
1
O
1.0
0.5
0.1
0.6
NH
S
O
1
9b
O
NH
OH
1
9c
1.2
0.1
O
S
NH2
O
ACS Paragon Plus Environment

Page 25 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
9d
0.4
1.6
N
NH
N
N
19e
0.9
0.2
0.1
1.1
B
HO
OH
1
9f
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
S
O
O
OH
4
b
0.8
1.0
1.3
1.0
OH
GW590735
a
b
Ratio of relative light unit (RLU) signal at 5 μM and 50 μM compound concentrations. Ratio of signal maximal signal (RLU)
strength observed for the compound of interest to that obtained with GW590735.
Additional Assessment of 4u. As the most potent analogue identified in stages I and II, 4u was
selected for additional assessment. To confirm target engagement, evaluate binding potency, and provide
some insight into selectivity in a relevant human retinal cell line (MIO-M1, Müller cells) we subjected 4u
to a CETSA (cellular thermal shift assay).28-29 Unlike the luciferase assay, MIO-M1 cells are not engineered
to specifically report engagement with PPARα and thus provide a more realistic picture of compound
behavior in a disease related cell setting. In an initial heat gradient, non-ligand associated PPARα displayed
a temperature-dependent decay, with T (50) and T (75) at 40.8 °C and 45.5 °C, respectively (Figure
agg
agg
7A). In a subsequent test, dose-dependent potency of the PPARα agonists, A91 and 4u were tested at 45.5
°C, the temperature point at which 75% of free PPARα protein melted. As shown in Figure 7B, A91
improved PPARα stability incrementally with an EC of 0.3 µM. On the other hand, 4u rescued PPARα
5
0
stability by 3-fold with an EC of 93 nM. The results for 4u correlate well with the EC value obtained
5
0
50
from the luciferase assay (EC = 37 nM), demonstrating consistency of this compound within different
5
0
contexts and eluding to excellent intracellular selectivity.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Cellular thermal shift assay results for A91 and 4u. (A) Thermal melting profile of PPARα. (B) Dose-dependent
stabilization of PPARα by A91 and 4u. Experiment was run at 45.5 °C, the Tagg(75).
Compound 4u was assessed in the same PK studies as 4a (A91). As seen in Figure 8, this analogue
exhibits significantly improved stability in human and rat liver microsomes (Figure 8A), shows no signs
of irreversible inhibition of the major drug metabolizing CYP450s (Figure 8B) and is not considered a
hERG liability (Figure 8C). It is important to note that while this compound exhibits 10% inhibition of
hERG at 10 μM (the same as A91), this analogue is ~20-fold more potent than A91.
ACS Paragon Plus Environment

Page 27 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A
C
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
B
F
O
O
H
N
HO
4u
Figure 8. Pharmacokinetic profile of 4u. (A) 4u was pre-incubated with pooled liver microsomes in phosphate buffer (pH 7.4)
for 5 min at 37 °C. The system was activated by the addition of an NADPH-regenerating system and incubated for 0, 15, 30, 45,
and 60 min before being quenched with an acetonitrile/methanol mixture. Samples were processed and analyzed by HPLC-MS/MS
to determine the peak area remaining. Experiments completed in duplicate and mean values are depicted. (B) Assessment of 4u
(
10 μM) for irreversible inhibition of the major CYP450 drug metabolizing isoforms. Peak areas corresponding to the metabolite
of known substrates for each isoform are recorded. The percent of control activity was calculated by comparing the peak area
obtained in the presence of 4u to that obtained in the absence of 4u. Percent inhibition was calculated by subtracting the percent
control activity from 100. Time-dependent CYP inhibition is demonstrated if the percent inhibition from samples pre-incubated
with the NADPH-regenerating system is larger than those without NADPH-regeneration. Negative values are a reflection of the
variability in the background noise of the experiment. Experiments completed in duplicate and mean values are depicted. (C)
Assessment of 4u for inhibition of the hERG channel. The degree of inhibition (%) was obtained by measuring the tail current
amplitude, which is induced by a one-second test pulse to -40 mV after a two-second pulse to +20 mV, before and after drug
incubation. Experiments completed in duplicate and mean values are depicted.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 70
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
Stage III SAR: Combining Lessons Learned. Due to information gained from the stage I
and II cohorts we designed and synthesized a series that combined beneficial features from each
stage. Utilizing established synthetic routes, compounds 20-33 were synthesized and evaluated
in similar fashion to stage I and II derivatives (Scheme 5). As shown in Table 4, all stage III
analogues exhibit submicromolar activity in the cell luciferase assay, with many derivatives
revealing EC50 values under 50 nM. Comparison of 24-29 to the corresponding des-methyl
relatives demonstrates that the methyl effect is ~10-20-fold more potent for each derivative and
thus exhibits exquisite consistency. Most notably, 22, which contains an iodine at the 4-position
on the C-ring, exhibits a 50-fold improvement in activity in comparison to the des-methyl B-ring
relative. Based on the predicted shift in the binding pocket (Figure 6) we suspect that room in the
C-ring pocket is subsequently freed, allowing space for the larger iodine substituent that is not as
easily accommodated with the des-methyl compound, 4e. Consistent with the effect observed
between 4w and 4b, N-methylation of the linker results in ~2-fold decrease in potency (4u vs. 30).
Furthermore, comparison of 4u and 31 reveals that incorporation of the 3-F on the A-ring results
in a maintenance in potency, a phenomenon first observed between 4b and 4q. Overall, the
results from stage III analogues confirm trends seen in stages I and II, corroborate in silico docking
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 29 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
predictions, and provide compelling evidence that the SAR for this chemotype is consistent, a
desirable yet often elusive property in lead optimization.
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 5. Synthesis of isostere containing derivatives 20-33a
ACS Paragon Plus Environment