Indazole-based ligands for estrogen-related receptor α as potential anti-diabetic agents

Article abstract of DOI:10.1016/j.ejmech.2017.07.015

Estrogen-related receptor α (ERRα) is an orphan nuclear receptor that has been functionally implicated in the regulation of energy homeostasis. Herein is described the development of indazole-based N-alkylthiazolidenediones, which function in biochemical

Full text of DOI:10.1016/j.ejmech.2017.07.015

Accepted Manuscript
Indazole-based ligands for estrogen-related receptor α as potential anti-diabetic
agents
Raymond J. Patch, Hui Huang, Sharmila Patel, Wing Cheung, Guozhang Xu, Bao-
Ping Zhao, Derek A. Beauchamp, Dionisios Rentzeperis, John G. Geisler, Hossein
B. Askari, Jianying Liu, Jyotsna Kasturi, Meghan Towers, Micheal D. Gaul, Mark R.
Player
PII:
S0223-5234(17)30537-8
10.1016/j.ejmech.2017.07.015
EJMECH 9578
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 5 August 2016
Revised Date: 23 June 2017
Accepted Date: 11 July 2017
Please cite this article as: R.J. Patch, H. Huang, S. Patel, W. Cheung, G. Xu, B.-P. Zhao, D.A.
Beauchamp, D. Rentzeperis, J.G. Geisler, H.B. Askari, J. Liu, J. Kasturi, M. Towers, M.D. Gaul,
M.R. Player, Indazole-based ligands for estrogen-related receptor α as potential anti-diabetic agents,
European Journal of Medicinal Chemistry (2017), doi: 10.1016/j.ejmech.2017.07.015.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Indazole-based Ligands for Estrogen-Related Receptor
α
†
as Potential Anti-Diabetic Agents
Raymond J. Patch, Hui Huang, Sharmila Patel, Wing Cheung, Guozhang Xu, Bao-Ping Zhao, Derek A.
Beauchamp, Dionisios Rentzeperis, John G. Geisler, Hossein B. Askari, Jianying Liu, Jyotsna Kasturi,
Meghan Towers, Micheal D. Gaul and Mark R. Player*
Janssen Research & Development, LLC, Welsh and McKean Roads, Spring House, PA 19477-0776,
USA
RECEIVED DATE
TITLE RUNNING HEAD: Estrogen-Related Receptor α Ligands
†
X-ray crystallographic information files can be obtained free of charge via
www.ccdc.cam.ac.uk/consts/retrieving.html (or from CCDC, 12 Union Road, Cambridge CB2 1EZ,
U.K.; fax +44 1223 336033; email deposit@ccdc.cam.ac.uk) CCDC 899222 for 20A and CCDC 898578
for 20B.
*
To whom correspondence should be addressed. Phone: 215-628-7860. Fax: 215-540-4616. E-mail:
mplayer@its.jnj.com.
a
Abbreviations: ERR, estrogen-related receptor; ER, estrogen receptor; ERE, estrogen response
element; ERRE, estrogen-related receptor response element; LBD, ligand binding domain; PPAR,
peroxisome proliferator-activated receptor; PGC-1, PPARγ coactivator 1; OXPHOS, oxidative
1

ACCEPTED MANUSCRIPT
phosphorylation; RIP-140, receptor-interacting protein 140; PEPCK, phosphoenolpyruvate
carboxykinase; MCAD, medium-chain acyl coenzyme A dehydrogenase; PDK, pyruvate dehydrogenase
-/-
kinase; apoA-IV, apolipoprotein A-IV; ERRα , ERRα null; FRET, fluorescence resonance energy
transfer; TR-FRET time-resolved FRET; hSRC, human steroid receptor coactivator; TZD, thiazolidine-
2
,4-dione; CYP, cytochrome P450; LBD, ligand binding domain; LXR, liver X receptor; RAR, retinoic
acid receptor; E2, 3,17β-estradiol; AF-2, activation function 2; TG, triglyceride; FFA, free fatty acids;
OGTT, oral glucose-tolerance test.
ABSTRACT: Estrogen-related receptor α (ERRα) is an orphan nuclear receptor that has been
functionally implicated in the regulation of energy homeostasis. Herein is described the development of
indazole-based N-alkylthiazolidenediones, which function in biochemical assays as selective inverse
agonists against this receptor. Series optimization provided several potent analogues that inhibited the
recruitment of a co-activator peptide fragment in vitro (IC50s < 50 nM) and reduced fasted circulating
insulin and triglyceride levels in a sub-chronic pre-diabetic rat model when administered orally (10
mg/kg). A multi-parametric optimization strategy led to the identification of 50 as an advanced lead,
which was more extensively evaluated in additional diabetic models. Chronic oral administration of 50
in two murine models of obesity and insulin resistance improved glucose control and reduced
circulating triglycerides with efficacies similar to that of rosiglitazone. Importantly, these effects were
attained without the concomitant weight gain that is typically observed with the latter agent. Thus, these
studies provide additional support for the development of such molecules for the potential treatment of
metabolic diseases.
KEYWORDS: Estrogen-related receptor α, ERRα, nuclear receptor, anti-diabetic, metabolic
syndrome, indazole, thiazolidenedione, TZD, insulin resistance
Introduction
2

ACCEPTED MANUSCRIPT
a
Estrogen-related receptors (ERRs) comprise a subfamily of orphan nuclear receptors, whose
members were identified on the basis of their DNA binding domain sequence similarities relative to
1
,2
those of estrogen receptors (ERα and ERβ).
Unlike these ERs, the ERRs are considered to be
constitutively active as they do not utilize estrogenic ligands or require such ligand-induced
conformational changes in order to regulate nuclear target transcription. Rather, they rely on co-
activator (or co-repressor) binding protein interactions for their functional activities. However, the
ERRs and ERs do share sequence similarities, particularly in their DNA-binding domains (69%
3
homology between ERRα and ERα). Thus, while preferentially binding to DNA sites termed ERR
response elements (ERREs), ERRs additionally interact with the classical DNA estrogen response
3
,4
elements (EREs) in vitro. As a result, the ERRs and ERs share several target genes and co-regulatory
5
proteins, leading to cross-talk between these two systems. ERRs may therefore provide compensatory
functional responses or may directly modulate estrogenic signals in various tissues, such as reproductive
organs, mammary glands and bone.
Much accumulated evidence additionally supports a primary role of ERRα in the regulation of
metabolism and energy homeostasis. ERRα is predominately expressed in tissues that derive energy
from fatty acid metabolism; in the adult mouse, for example, highest expression levels are found in
6
brown adipose, heart and kidney. Fatty acid β-oxidation in these tissues is under the control of the rate-
limiting enzyme, medium-chain acyl coenzyme A dehydrogenase (MCAD), whose transcribing gene,
6
ACADM, was shown to be regulated by ERRα in vitro.
ERRα-mediated transcriptions are regulated by interactions with co-activator and co-repressor
proteins, whose functions in various metabolic processes are well established. The expression profile of
this receptor correlates with those of PPARγ co-activator-1 proteins (isoforms PGC-1α and PGC-1β),
principal metabolic modulators for several transcription factors and strong inducers of ERRα-mediated
7
-10
transcriptional activity.
While PGC-1α coactivates numerous nuclear receptors by way of common
leucine-rich interaction motifs, it additionally harbors a unique binding motif (LLxYL) that is
3

ACCEPTED MANUSCRIPT
9
specifically required for engagement with ERRα. PGC-1β is a selective co-activator for the ERR
10
family, being most potent for ERRγ and of reduced but similar potencies for ERRα and ERRβ. The
energy regulating co-repressor ligand, receptor-interacting protein 140 (RIP-140), which suppresses the
1
1,12
expression of several genes involved in glucose and lipid metabolism
has also been shown to
1
3
interact with ERRα and affect its target gene transactivations.
In turn, ERRα up-regulates RIP-140
1
4
gene transcription during adipogenesis, implicating a role for the former in maintaining energy
homeostasis in adipocytes.
Several other genes with established roles in metabolic processes are influenced by ERRα in vitro.
Transcription of the phosphoenolpyruvate carboxykinase (PEPCK) gene, which encodes a rate-
determining gluconeogenic enzyme and whose expression is enhanced by PGC-1α, is repressed by
1
5
ERRα in hepatocytes. Pyruvate dehydrogenase kinase 4 (PDK4), a key negative regulator of muscle
and liver glucose oxidation, has also been shown to be up-regulated in response to ERRα/PGC-1α-
1
6-18
driven transcription in these tissues.
Basal levels of apolipoprotein A-IV (apoA-IV), a post-prandial
satiety signal that is synthesized in the intestine following lipid ingestion, have also been reported to be
under the transcriptional control of ERRα in mice, and an ERRα regulatory function in intestinal lipid
1
9
transport and energy homeostasis has been suggested. ERRα has been shown to act in concert with
PGC-1α to induce the expression of multiple genes encoding mitochondrial proteins, thereby
influencing various aspects of mitochondrial biogenesis and function, including mitochondrial DNA
2
0
replication and transcription, the TCA cycle, fatty acid oxidation, and oxidative phosphorylation.
The initial characterization of the physiological role of ERRα and its implication in the regulation of
-/-
21
metabolism in vivo was derived from ERRα-deficient (ERRα , null) mice.
The null mice were
viable and fertile and exhibited no observable anatomical abnormalities upon histological examination.
In comparison with control (wild-type) littermates, these mice were noticeably lean, displaying reduced
(~17%) body weight and a reduction (50-60% decrease) in inguinal, epididymal and peritoneal fat pads
at 10 weeks of age. Histological analysis of fat tissue revealed a normal number of fat cells, but a
4

ACCEPTED MANUSCRIPT
reduction in adipocyte size. Consistent with these findings, lipogenesis was reduced by 30-55% in
adipose tissues of these mice. Food consumption, energy expenditure and fasting levels of glucose, free
fatty acids and triglycerides were unchanged in the null mice maintained on a standard chow diet.
-/-
However, the ERRα mice were resistant to weight gain induced by a high-fat diet. Gene expression
profiling of adipose tissues of these animals revealed alterations in the expression levels of several
genes involved in fatty acid metabolism (including ACADM), lipogenesis and energy metabolism.
Intestinal gene expression analysis of the null mice revealed down regulation of several proteins
1
9
involved in oxidative phosphorylation and lipid handling (including APOA4).
-
/-
The lean, diet-induced obesity resistant phenotype presented by the ERRα mouse would appear to be
incongruous with many of the functions ascribed to this orphan receptor as determined from mechanistic
studies carried out in vitro. Rather, a phenotype more reflective of a general decrease in oxidative
metabolism and mitochondrial biogenesis, coupled with an increase in gluconeogenesis might have been
anticipated in animals lacking ERRα transcriptional input. It is possible that in the absence of signaling
from ERRα, compensatory mechanisms arising from other nuclear receptors (such as ERRγ, for
example) or transcriptional pathways become operational. Under such circumstances, certain
components of ERRα-dependent activities relating to energy homeostasis may be functionally
maintained, while the influence of others may become more apparent in the whole animal. This
possibility complicates the inferential interpretation of biochemical results obtained in isolated systems
and has made it difficult to predict a priori the therapeutic outcome of selective ERRα modulation in
disease states.
We have previously described the diaryl ether-based thiazolidenedione 1 (Figure 1), an ERRα-
selective ligand, which inhibited the recruitment of a polypeptide fragment of the hSRC2 co-activator to
the ERRα ligand binding domain in biochemical assays (IC50s < 150 nM) and cellular 2-hybrid reporter
2
2
assays (IC s 1-5 µM). In diet-induced murine models of obesity and in an overt diabetic rat model,
5
0
oral administration of 1 normalized insulin and circulating triglyceride levels, improved insulin
5

ACCEPTED MANUSCRIPT
sensitivity and was body weight neutral. Extension of our initial SAR studies led to the identification of
the N-1-benzylindazolyl system as a suitable replacement for the phenoxyphenyl moiety of this lead
series. Furthermore, derivatization of the TZD moiety was well tolerated in the benzylindazole series
and was a primary focus of SAR development studies. Herein we summarize the optimization efforts,
which culminated in the identification of 50, a novel ERRα modulator with improved efficacy in
diabetic animal models.
Chemistry
A general synthetic route for the preparation of series analogues is outlined in Scheme 1. Benzylation
of commercially available indazole-5-carboxaldehyde 2 proceeds in non-regioselective fashion,
providing a readily separable mixture (~1:1) of N-1- and N-2-benzyl derivatives (3a-c and 4a-c,
2
2
respectively). Knoevenagel condensation of 3a-c with 2,4-thiazolidinedione as previously described
afforded the Z-olefinic isomers 5a-c, which could be derivatized with a variety of electrophilic reagents
under nucleophilic displacement conditions or by Mitsunobu methodology to provide alkylated
analogues, A.
3-Methylindazolyl analogues were prepared from 3-methyl-1H-indazole-5-carbaldehyde 9, whose
synthesis is outlined in Scheme 2. In this case, benzylation of 5-bromo-3-methylindazole 6 proceeded
2
3
regioselectively (~ 4:1) in favor of the N-1-derivative, 7. Palladium-catalyzed cyanation provided the
corresponding 5-cyanoindazole (8), which upon treatment with aluminum-nickel alloy in aqueous
formic acid was reduced to indazolecarboxaldehyde 9.
Piperidinylthiazolidinedione analogues were accessed through ring-opening reactions with Boc-
piperidine-3,4-epoxide 10. As shown in Scheme 3, reaction of thiazolidine-2,4-dione with epoxide 10
under conditions of microwave heating afforded a regioisomeric mixture of 4- and 3-substituted
piperidinyl alcohols (11 and 12, respectively), which could be partially separated by silica gel
2
4
®
chromatography. Treatment of these alcohols with Deoxo-Fluor afforded the corresponding trans-
fluoropiperidinyl-TZDs 13 and 14, whose chromatographic purification was more conveniently
performed than that of the preceding alcohol mixture (11 and 12). The fluoromethinyl proton provided
6

ACCEPTED MANUSCRIPT
1
the diagnostic signal in the H NMR spectrum that enabled the assignment of trans configuration for 13.
In this case, this signal appears as a geminally coupled (JHF = 50 Hz) pair of triplets of doublets (J =
1
0.2, 5.9 Hz) centered at 5.15 ppm in CDCl ; the triplet arising from two trans diaxial couplings to
3
adjacent protons. Thus, this fluorination occurred with net retention of configuration, presumably the
2
5
result of anchiomeric assistance from the neighboring thiazolidinedione ring system. Knoevenagel
condensation of 13 with indazol-5-carbaldehyde 3b provided the corresponding piperidinyl derivative,
4
8.
A modified synthetic strategy was utilized for the targeted preparation of 49 on a larger scale (ca., 50
g). In this method (Scheme 4), reaction of indazolidenylthiazolidinedione 5b with Boc-piperidine-3,4-
epoxide 10 provided direct access to trans-hydroxypiperidinyl-thiazolidinediones 15 and 16. Under
2
6
conditions of magnesium perchlorate catalysis, this conversion occurred somewhat regioselectively,
2
7
favoring nucleophilic attack at the 4-position of the piperidineoxide (15 : 16 ~ 3.5 : 1). Fluorination as
before afforded the corresponding mixture of trans-fluoropiperidinyl-thiazolidinediones 17 and 18,
whose purification was again more conveniently performed at this fluoropiperidine stage. Acidic
deprotection of 17 provided 48, which underwent reductive methylation to afford methylpiperidine 49.
Process development aspects of this strategy have been described elsewhere in greater detail along with
2
8
a refined procedure that proved to be suitable for larger scale synthesis (500 g scale).
Resolution of the isomeric trans-fluoropiperidines comprising 49 was accomplished by chiral
®
chromatographic separation of the Boc-protected precursor, 17, on a Chiralcel -OD column (Scheme 5).
Deprotection of resolved antipodes 17A and 17B provided chiral trans-piperidines 19A and 19B, which
were methylated as above under reductive amination conditions to provide chiral methylpiperidines, 50
and 51, respectively. In order to determine the absolute configurations of these chiral analogues, the
corresponding 4-nitrobenzylated derivatives were prepared by reductive amination reactions of 19A and
1
9B with nitrobenzaldehyde. X-ray diffraction-quality crystals of 20A were obtained by crystallization
from nitromethane, whereas crystallization of 20B from acetonitrile afforded suitable crystals for
structure determination. The absolute configurations of nitrobenzylpiperidines 20A and 20B were
7

ACCEPTED MANUSCRIPT
determined to be (3R,4R) and (3S,4S), respectively. Thus, elaboration of precursor Boc-piperidine 17A
as outlined in Scheme 5 led to 50, whose stereochemical configuration must be (3R,4R) at these centers.
Similar treatment of Boc-piperidine 17B provided 51, whose stereochemical configuration must
therefore be (3S,4S). Figure 2 depicts the ball-and-stick renderings of crystallized nitrobenzyl
derivatives 20A and 20B.
Series SAR
Compounds were evaluated in vitro for their abilities to inhibit the binding of a coactivator
polypeptide derived from hSRC2 using a time-resolved fluorescence resonance energy transfer assay
2
2
(
TR-FRET), as previously described. With respect to potency in this assay, a wide range of polar
substituents stemming from the TZD ring nitrogen position were tolerated. This was not an unexpected
finding in light of the anticipated binding mode of these molecules, which was based upon the co-crystal
2
2
structure of ERRα with 1. Substituents at this position were not expected to make productive contact
with the ERRα protein, since these groups would be oriented to project outwards from the protein
towards solvent. However, these substituents could likely influence ancillary pharmacology as well as
pharmacokinetic properties of this series. Thus, acidic and basic functionalities were investigated and
found to be equally well accommodated. Small, neutral polar groups also afforded potent molecules,
however these derivatives were generally not sufficiently soluble for formulation and further
characterization in vivo.
N-1-Benzyl substituent SAR indicated a preference for 2,4-disubstituted systems, particularly those
bearing a 2-trifluoromethyl group. Requirements for the 4-substituent were somewhat less strict, with
MeO, Cl and CF groups being preferred. In a probe set of analogues (Table 1), MeO- and Cl-
3
substituted derivatives were found to be equipotent at inhibiting coactivator binding, whereas the
corresponding CF -substituted derivatives were generally less active (ex., 30). Furthermore, the poor
3
3
solubilities of the CF -substituted analogues precluded their pharmacological evaluations in vivo, thus
they were not pursued beyond these initial studies. Pharmacological evaluation of some of the more
potent 4-methoxy analogues in a sub-chronic fa/fa rat model (described below) revealed only modest
8

ACCEPTED MANUSCRIPT
pharmacodynamic effects (reductions of plasma insulin and triglyceride levels), whereas 4-chloro-
substituted analogues provided more robust responses in vivo. Thus, the majority of our efforts were
focused on developing series analogues bearing the N-1-(4-chloro-2-trifluoromethylbenzyl) group.
A sub-chronic pre-diabetic model was established for assessing basic pharmacological efficacy of
29
select analogues. In this assay, 9.6-week old fa/fa rats, maintained on a high fat diet, were dosed
orally with test compound (10 mg/kg) for 4 days. On day 5, fed plasma insulin and triglyceride levels
were measured; those compounds providing substantial decreases in both parameters were candidates
for further characterization in vivo. Because of its modest throughput, compound selection for this
assay was based on prior ancillary pharmacological profiling, in which selectivity (Cerep receptor/ion
3
channel panel), CYP P450 interactions, microsomal stability and hERG activity (inhibition of H-
astemizole binding) were evaluated. During the optimization process, selected examples were
additionally evaluated in patch-clamp assays, the results of which confirmed a strong correlation
between inhibitory activity in the hERG binding assay and functional channel activity for this series.
In general, analogues bearing an acidic functionality from the TZD exhibited favorable ancillary
pharmacological properties: high selectivity, low propensity for CYP interactions and minimal hERG
liability (IC > 50 µM). As shown in Table 2, analogue potencies in vitro were of similar magnitude
5
0
(within 3-fold); however, effects on insulin and triglyceride levels were generally moderate and
variable. In the fa/fa model, treatment with carboxylate derivatives 25 and 26 produced significant
reductions in plasma insulin levels (-50 % and -30 %, respectively) but only moderately reduced
triglycerides (-14% and -19%, respectively). Similarly, sulfamide analogues 35 and 37 had more
pronounced effects on insulin levels than on triglyceride levels, whereas the opposite trend was
observed with dimethylsulfamide analogue, 36, which more effectively reduced triglyceride levels (-
4
6%). Notable exceptions were tetrazolylmethyl (32) and acylsulfonamido (34) derivatives, each of
which provided robust effects on both parameters in vivo. However, in subsequent pharmacokinetic
studies, both analogues exhibited extremely poor oral bioavailabilities (plasma concentrations below the
9

ACCEPTED MANUSCRIPT
level of detection). Additionally, the tetrazole moiety introduced a CYP 2C9 liability and thus was not
investigated further.
Introduction of a methyl group at the 3-position of the indazole ring was well tolerated with respect to
potency in vitro. In some instances, greatly improved bioavailability was observed, though this did not
necessarily translate to increased efficacy in vivo. For example, reduction of plasma insulin levels were
much less pronounced following treatment with 38 than with 37, in spite of a 40-fold increase in rat
plasma levels of the former analogue (data not shown).
Increased efficacy was generally observed from analogues derivatized with basic functionalities;
however, these improvements were frequently accompanied by off-target activities. When assayed
against a standard Cerep panel, compounds that exhibited significant activities (> 50% inhibition at 10
µM) against multiple receptors/channels (10 or more) were regarded as being too non-selective for
advancement. Similarly excluded were those analogues exhibiting significant activities (IC50 < 10 µM)
in either the hERG binding assay or in microsomal CYP binding assays. Table 3 exemplifies a limited
set of some particularly efficacious molecules that were studied during the course of lead optimization
efforts, along with key liabilities that were identified for each, thus precluding their further progression.
In addition to the off-target activities identified above, poor oral bioavailabilities also impacted
several members of this series. For example, five particularly efficacious molecules, 28, 39, 42, 44 and
4
5 had very high volumes of distribution in SD rats and did not achieve plasma levels greater than 100
ng/mL, following oral dosing (10 mg/kg). Efforts to attenuate the amine basicity in this region of the
molecule were undertaken as a means of potentially mitigating the above-mentioned off-target effects
and improving bioavailability. Since fluorination of a piperidine ring is able to reduce its pKa by as
3
0
much as 2 log units, this modification was introduced in the present series. Indeed, the 3-fluoro
analogue of 40, namely 48, was devoid of the CYP liabilities of the unfluorinated compound, though a
3
1
potential for hERG interaction was detected.
Furthermore, the methylated homologue, 49 (the 3-
fluoro derivative of 41), was found to be highly efficacious in vivo and free from both hERG channel
and CYP liabilities.
1
0

ACCEPTED MANUSCRIPT
In the 4-day fa/fa model, treatment with 49 not only afforded considerable reductions on fed insulin
and triglyceride levels (-73% and -57%, respectively), but additionally reduced fed glucose and free
fatty acid levels significantly (-43% and -35%, respectively). Importantly, these effects were realized
without a concomitant increase in body weight gain. Resolution of racemic 49 was therefore
undertaken; its isomeric constituents, 50 and 51, were prepared and found to be equipotent in vitro. The
former however, was somewhat more efficacious in vivo, as assessed in an extended (10-day) fa/fa
study, in which slightly greater reductions in fed plasma insulin (-67% and -51%, respectively) and fed
plasma triglyceride (-68% and -51%, respectively) levels were realized at day 5. Fed plasma glucose
levels were also reduced by both isomers (-32% and -27%, respectively; the former reaching
significance). On day 10 of this study, both isomers provided significant improvements in glucose
handling during an OGTT, (-46% and -37% glucose AUC0-120, respectively). Most notably, 50, in
contrast to 51, provided these effects without inducing body weight gain, increasing food consumption
or adversely affecting liver enzymes. Thus, 50 was selected as an advanced lead for more extensive
characterization in vivo.
Selectivity Studies
The potential of 50 to interact with other nuclear receptors was investigated under a variety of assay
conditions (Table 4). TR-FRET assays similar to that described above for ERRα were used to evaluate
32
33
functional activities against ERRβ and ERRγ, whereas commercial fluorescence polarization assays
(
Invitrogen) provided the means for evaluating activities against ERα, ERβ, PR and PPARγ ligand
3
4
binding. Compared with its potent ERRα inhibitory activity (IC50 = 0.023 µM), 50 proved much less
potent (> 400-fold selectivity) at inhibiting co-activator peptide binding with ERRβ or ERRγ (IC s > 10
5
0
µM) and did not compete with ERα, ERβ, PPAR-γ or PR ligand binding at the highest test
concentrations studied. Other nuclear receptors (PPARα, PPARδ, LXRα, LXRβ, RXRα) were
similarly unaffected by 50, either as agonists or antagonists, as determined by corresponding co-
3
5
activator recruitment (TR-FRET) assays (IC s > 10 µM; Table 4).
5
0
1
1

ACCEPTED MANUSCRIPT
Compound 50 was also evaluated for selectivity in a general binding screen (Cerep) against a standard
panel of 51 receptors and ion channels. At a test concentration of 10 µM, 50 did not inhibit paradigm
compound binding greater than 40%, with the exceptions of the chloride channel, against which
picrotoxin binding was partially inhibited (50%) and the Na channel (site 2), against which veratridine
binding was partially inhibited (69%). Follow-up testing in a patch clamp hNav1.2 functional assay
showed no significant effects on current at concentrations up to 10 µM. Similarly, against a standard
kinase panel of 209 enzymes (Invitrogen), 50 showed no significant inhibitory activities (> 30%) at a
test concentration of 10 µM and an ATP concentration of 100 µM.
Pharmacokinetic Studies
Pharmacokinetic assessment of 50 was initially examined in the Sprague-Dawley rat at doses of 2
mg/kg (iv) and 10 mg/kg (po) (Table 5). Following iv administration, plasma concentrations declined in
a biphasic manner with plasma concentrations measured out to 24 h. Compared to hepatic blood flow,
5
0 had a moderate clearance and had a very high volume of distribution relative to total body water.
Absorption of 50 following oral administration was rapid with maximal concentrations reaching 683
ng/mL between 1-2 h. The oral half-life of 4.7 h was similar to that determined following intravenous
administration. The calculated bioavailability of 83.5% was based on AUC24h values for both iv and
oral administrations. In the CD-1 mouse (5 mg/kg, iv; 20 mg/kg, po), 50 had a low clearance and again
exhibited a large volume of distribution. The oral half-life in this species (10.6 h) was significantly
longer than that calculated in the SD rat.
Oral pharmacokinetic assessment of 50 was also carried out in two efficacy species. In the fa/fa rat,
absorption of 50 after oral administration (15 mg/kg) was rapid and extensive, with maximal plasma
concentrations reaching 4023 ng/mL at 2.5 h. Exposure of 50 was higher, and the terminal elimination
half-life (6.7 h) was longer than had been observed in the SD rat. In ob/ob mice, absorption of 50 after
oral administration was also rapid, with maximal plasma levels (829 ng/mL) being measured at 2 h.
The terminal elimination half-life (8.1 h) was similar to that found in the CD-1 mouse. The Cmax and
1
2

ACCEPTED MANUSCRIPT
AUC24h values were determined to be approximately half of those obtained in the CD-1 mouse at a
comparable dose.
Pharmacological Studies
The pharmacodynamic effects of 50 in the fa/fa rat model were examined in a chronic efficacy study
to confirm and extend the preliminary findings obtained during our initial subchronic evaluation. In this
3
2-day dose range study, animals were administered daily oral doses of either 50 (1, 3, 10 or 50 mg/kg)
or rosiglitazone (10 mg/kg), the latter of which was used for an active comparator group. After 30 days
of treatment with 50, statistically significant reductions in the fasting glucose levels at all test doses
were obtained (Figure 3). Dose-dependent reductions in fasting plasma insulin levels were also effected
throughout the entire dose range (Table 6). Significant dose-dependent reductions in the glucose AUC0-
2h
were also realized during an OGTT, indicating improved glucose homeostasis (Figures 3 and 4).
Glucose excursion was reduced by 23% in the 1 mg/kg group, whereas at 10 mg/kg, maximal efficacy
was attained (55% reduction in glucose AUC). At 3 mg/kg, 50 improved glucose excursion to an
equivalent extent as rosiglitazone at its maximally efficacious dose in this model (10 mg/kg).
Significant reductions in insulin AUC were also measured during the OGTT, indicating a dose-related
improvement in insulin sensitivity.
Plasma concentrations of 50 were measured at early and late time points during the study to correlate
drug exposure with the observed pharmacology (Table 6). Increases in drug level approached dose
proportionality at the lower dose range, but this proportionality was not maintained through 50 mg/kg
dosing. After 26 days of repeated dosing, systemic compound accumulation appears to have resulted at
the lower doses, as indicated by the increased AUCs observed at day 26, compared to those at day 1.
However, this phenomenon was not observed at the 50 mg/kg dose, possibly indicating a saturation of
the compound accumulation mechanism(s) at repeated high dosages. With respect to fasting glucose
levels and glucose excursion, most significant effects of 50 were obtained at plasma exposures > 10,000
ng•hr/mL, whereas higher exposures (> 23,000 ng•hr/mL) provided the more robust effects on insulin
levels.
1
3

ACCEPTED MANUSCRIPT
No significant changes in body weight or body weight gain were induced by 50 relative to vehicle
after 28 days of dosing (Table 7). Modest increases in food intake were observed at 3 and 10 mg/kg
doses, but at the 50 mg/kg dose, the increase did not reach significance. In contrast, treatment with
rosiglitazone produced a 52% increase in body weight gain relative to vehicle over the course of this
study period and a 29% increase in food intake. At necropsy (day 32), the epididymal and brown
adipose fat pads, liver and heart were excised and weighed. A dose-related increase in the brown
adipose tissue was observed in the animals treated with 50; a similar significant increase in this tissue
was observed in the rosiglitazone treatment group. An unexpected increase in weight of the epididymal
fat pad was observed, which had not been detected in a related 15-day, single dose study (data not
shown), suggesting a possible redistribution or partitioning of lipids into this fat pad over chronic
dosing. Consistent with this notion, dose-related decreases in liver weight and concomitant decreases in
hepatic and systemic triglycerides were observed, indicating a resolution of hepatic steatosis. This was
also reflected in an improvement in AST levels. Pancreatic insulin content was increased in all groups
where improved glucose control (during the OGTT) had been observed, including the rosiglitazone
cohort. No significant effects on cholesterol levels were afforded by either compound.
In a separate tissue distribution study, following 11 days of oral administration of 50 (12.5 mg/kg)
compound concentrations were measured in several tissues (fat >> liver > muscle ~ kidney > heart),
confirming an accumulation of compound in these tissues upon repeated administration (Figure 5).
Thus, fat and liver appear to be the predominant target tissues for compound uptake in this species.
Slow redistribution from these tissues would be consistent with the observed prolonged systemic
exposure to 50 following chronic administration.
The robust effects of 50 on triglyceride levels, coupled with its high level of accumulation in adipose
tissue prompted an investigation for the production of adipose-derived factors such as adiponectin,
whose induction would be expected to influence glucose tolerance and insulin sensitivity. In fact, dose-
dependent increases in steady-state levels of adiponectin were observed following treatment with 50,
3
9
which reached significance at 3 mg/kg by day 4 (Figure 6). At the 10 and 50 mg/kg doses, plasma
1
4

ACCEPTED MANUSCRIPT
adiponectin levels were increased by 1.6- and 1.9-fold, respectively. By day 32, the increases were
more pronounced as adiponectin levels were 2.1- and 3.3-fold above vehicle at these same respective
doses. Similar increases in adiponectin levels were obtained by rosiglitazone treatment. These results
are consistent with studies correlating increased adiponectin levels with improved peripheral insulin
4
0
sensitivity.
To further delineate the mechanism(s) by which 50 improved the dyslipidemic phenotype of the
Zucker fa/fa rats, expression levels of specific ERRα- regulated genes were measured from the adipose
tissue of the treated animals. Particular genes investigated were those involved in the tricarboxylic acid
cycle, fatty acid beta oxidation, and gluconeogenesis, as well as PGC-1α (energy metabolism and
mitochondrial biogenesis). Given the increased plasma levels of adiponectin that occurred following
treatment with 50, ADIPOQ expression was also investigated. In addition, the observed attenuation of
plasma triglycerides prompted an examination of ANGPTL4 expression levels. Thus, RNA was
extracted from the epididymal fats pads of the treated animals and the expression of a subset of genes
was confirmed by quantitative RT-PCR (Figure 7). Significant increases in expressions of known
ERRα-regulated genes were observed, including ACADM (fatty acid beta oxidation), PCK1
(
gluconeogenesis) and IDH3A (TCA cycle). ERRα gene expression (ESRRA) itself was also enhanced,
4
1
which may reflect the stimulation of a positive feed-forward mechanism. Significant increases were
also detected in the expressions of ADIPOQ and ANGPTL4. Although the trend towards elevated
PGC1
α
expression did not reach significance, a subsequent whole genome microarray analysis revealed
(+ 2.25-fold, p = 0.007; see Supporting Information, Table 1).
an increased expression of PGC1
β
Functional and pathway analyses of differentially regulated transcripts from this broad microarray
®
(
Ingenuity Systems, IPA) revealed prominent effects on canonical pathways associated with the TCA
cycle, mitochondrial dysfunction and fatty acid processing. Notably affected were genes associated
with fatty acid transport (FABP3, + 6.5-fold), fatty acid oxidation (EHHADH, + 3.2-fold; ALDH1A1,
1
5

ACCEPTED MANUSCRIPT
2
.2-fold), mitochondrial electron transport (UCP1, + 3.2-fold; UCP3, + 2.3-fold) and mitochondrial
respiration (COX8B, + 2.2-fold).
A second monogenic rodent, the ob/ob mouse, which serves as a model of obesity, hyperlipidemia,
insulin resistance and type-2 diabetes, was employed to confirm the ability of 50 to attenuate
hyperglycemia and improve insulin sensitivity. A 32-day chronic efficacy study in this model was
conducted to determine the maximally efficacious dose of 50 on glycemic control and body weight, as
compared to rosiglitazone. With animals treated with 50, dose-dependent reductions in fed glucose and
HbA1c levels were observed. During an OGTT, treatment with 50 at doses of 10 and 50 mg/kg resulted
in statistically significant reductions in the glucose AUC0-2h (41%, and 57%, respectively), indicating
improved glucose tolerance. As shown in Figure 8 and Table 8, at these doses, the effectiveness of 50 at
improving glucose control (determined by the OGTT as well as by fasting plasma glucose levels) was
similar to that of rosiglitazone at its maximally effective dose.
Similar to observations in the chronic fa/fa rat study, assessment of plasma levels of 50 on the first
and last days of the study (Table 8) revealed increases in drug level that approached dose
proportionality. The extent of compound absorption (Cmax) was less on day 28 than on day 1. However,
extended terminal elimination phases were observed at all doses, which were additionally reflected by
increased AUCs at day 28, compared to those at day 1. Thus, compound accumulation appears to have
resulted from 28 days of repeated dosing in this species as well. However, in contrast to results from
the above fa/fa rat study, saturating levels of 50 were not attained upon repeated dosing in the ob/ob
mouse, as day 28 AUCs exceeded those at day 1 through the entire dose range (3 - 50 mg/kg). In this
study, significant effects on fasted glucose, glucose excursion during the OGTT (day 28) and % HbA1c
(day 32) were realized at drug exposures ranging between ~ 6,000 and 30,000 ng•h/mL (Table 8).
No significant changes in body weight, body weight gain or food intake were observed at any dose of
5
0 or with rosiglitazone (Table 9). Dose-dependent decreases in epididymal white adipose tissue
(EWAT) mass were noted, consistent with a redistribution of this fat pad. Also, significant increases in
1
6

ACCEPTED MANUSCRIPT
brown adipose tissue (BAT) were effected by both agents. However, in contrast to rosiglitazone-treated
animals, whose liver masses were increased relative to vehicle control animals, no significant changes in
liver weight were noted in the group treated with 50.
Similar to findings in the fa/fa rat, triglyceride levels were also significantly reduced in this animal
model. In addition, significant reductions in cholesterol were achieved at the 10 and 50 mg/kg doses; an
effect that was not obtained in the rat model. Dose-dependent increases in steady-state levels of
adiponectin were also detected in this model following treatment with 50 (Figure 9). When assessed at
the end of the 32-day treatment regimen, plasma adiponectin levels were significantly elevated (1.4-,
2
.2- and 2.8-fold; 3, 10 and 50 mg/kg dose groups, respectively). A similar increase in adiponectin level
was observed in the rosiglitazone group (2.6-fold). No changes in ALT or AST levels were noted at any
dose of 50, whereas significant increases in both parameters were observed in rosiglitazone-treated
animals. Moderate decreases in creatinine levels were observed in animals treated with either agent.
BUN levels however, though not affected at any dose of 50, were significantly increased in the
rosiglitazone-treated cohort.
Discussion
Given the extensive range of metabolically-relevant genes whose transcriptions can be modulated by
ERRα, it is evident that this nuclear receptor plays a prominent role in the regulation of mammalian
4
2
metabolic function and energy homeostasis. The extent, however, to which this receptor exerts such
regulatory control in vivo has proven difficult to delineate, due to the complex network of co-regulatory
proteins and overlapping pathways that may become differentially activated in response to variable
energy demands. In the face of deficient (or excessive) ERRα-mediated regulatory input, compensatory
signaling mechanisms may become operational in the whole animal such that some of the anticipated
effects of receptor dysregulation are not realized. The ERRα-null mouse model provided initial insight
regarding the phenotypic consequences of global repression of ERRα-mediated transcriptions and
provided the impetus for investigating the potential of selective modulators of this receptor as
1
7

ACCEPTED MANUSCRIPT
therapeutic agents for metabolic diseases. Indeed, in studies using murine models of obesity and
diabetes, we demonstrated that multiple aspects of this lean phenotype (normalized serum triglyceride
and insulin levels; improved glucose tolerance) could also be realized through pharmacological
2
2
intervention with the ERRα modulator, 1.
In the present work, the development of a second generation series of compounds is described, which
differs structurally from the first-generation series in that 1) a benzylindazole core has replaced the
phenoxyphenyl ether, and 2) an alkyl group projects from the thiazolidinedione nitrogen. Exploration of
a large set of N-alkyl substituents spanning a range of polarities indicated broad tolerance at this
position with respect to effects on SRC2 coactivator peptide/receptor binding. This finding is consistent
with the anticipated binding mode of these compounds (inferred from the X-ray structure of the receptor
2
2
complex with 1), in which substituents at this position would not be expected to engage in productive
receptor interactions, but rather would project away from the receptor core and remain extensively
solvated. Analogues bearing basic functionalities tended to provide more robust responses in vivo as
compared with acidic derivatives, though such compounds generally exhibited off-target activities that
precluded their advancement. However, by attenuating the basicity of this functionality, off-target
activities could be minimized. This strategy ultimately provided lead compound, 50, a highly selective
modulator of ERRα with no demonstrable activity against other metabolically-relevant nuclear
receptors and negligible off-target activities, as assessed in general receptor/kinase selectivity screening.
It should be highlighted that embedded within the structure of compound 50 is an olefinic
thiazolidinedione moiety, a functionality that has been described as having the potential to render such
4
3
molecules as pan assay interference compounds (PAINS) in various assay systems. However, the high
degree of selectivity exhibited by 50 when evaluated against a broad range of targets provides clear
2
2
indication of a lack of such promiscuous activity for this molecule, as was similarly observed with 1.
The Zucker fa/fa rat serves as a monogenic model of prediabetes arising from a mutated fa gene,
29
which produces a truncated leptin receptor that cannot interact with its natural peptide hormone.
When maintained on a high-fat diet, these animals develop an obese, insulin resistant phenotype and
1
8

ACCEPTED MANUSCRIPT
exhibit elevated serum lipid, triglyceride and cholesterol levels. In this model, chronic administration of
5
0 provided dose-dependent improvements in insulin sensitivity, glucose handling, serum and liver
triglyceride levels and provided an increase in circulating levels of adiponectin. Notably, in contrast to
the effects afforded by rosiglitazone treatment, the improvements in metabolic parameters provided by
5
0 were realized without concomitant increases in food intake or body weight gain. Pharmacokinetic
analysis revealed an increase in compound exposure following repeated dosing that would be consistent
with a mechanism(s) of compound accumulation and re-distribution. In support of this, tissue
distribution studies showed particularly high levels of 50 in adipose tissue after 11 days of dosing (12.5
mg/kg).
The ability of 50 to attenuate hyperglycemia and improve insulin sensitivity in a second rodent model
of obesity-induced diabetes, the ob/ob mouse, was also investigated. This animal presents a phenotype
that includes insulin resistance, hyperinsulinemia, hyperlipidemia and glucose intolerance. In this
model, chronic treatment with 50 provided dose-dependent improvements in glucose tolerance, as
evidenced by reductions in circulating levels of glucose (in response to an oral glucose challenge as well
as fasted levels) and HbA1c. Serum triglyceride levels were also significantly reduced, as were serum
cholesterol levels; in fa/fa rats, the cholesterol reductions did not reach statistical significance. In
contrast, rosiglitazone treatment at a maximally efficacious dose did not provide an improvement in
cholesterol levels in ob/ob mice, though it did lead to decreased circulating triglyceride levels. Both
agents effected increases in adiponectin levels, the magnitudes of which were similar to those obtained
in the fa/fa rat. Again, the improvements in glycemic control realized upon treatment with 50 were
provided without effecting changes in body weight or body weight gain. BAT mass was increased by
5
0 and rosiglitazone in both animal models; however, EWAT was uniquely decreased by both agents in
the ob/ob mouse. A pharmacokinetic profile somewhat similar to that determined in the fa/fa rat was
also noted in the ob/ob mouse following repeated administration of 50. Thus, in both species, extended
terminal elimination phases were observed, and compound accumulation was indicated.
1
9

ACCEPTED MANUSCRIPT
Previously, the differential gene analyses of liver tissue from C57BL/6 mice maintained on a high fat
diet and treated with 1 revealed enrichment in a number of ontology terms associated with fatty acid
2
2
metabolism, synthesis and oxidation, lipid transport and mitochondrial function. In those animals, the
liver had proven to be the most responsive tissue with respect to such differential analysis. However, in
the present study, given the high adipose levels of 50 that were detected in the fa/fa rat in response to
chronic dosing, microarray gene analyses were performed on RNA extracted from this tissue. Several
of the genes up-regulated in liver following treatment with 1 (mice) were found to be similarly affected
in adipose tissue by 50 (fa/fa rats), including ERRα-regulated TCA cycle genes (IDH3A, IDH3B), and
fatty acid oxidation genes (ACADM, HADHB, ACOX1). Interestingly, whereas adipose levels of
2
2
ACADM had not been affected by 1, in the present study, significant up-regulation of this gene was
found to be induced by 50, possibly reflecting tissue-selective pharmacology.
Increased ADIPOQ expression was effected by both ERRα modulators, and likely accounts for the
increased serum adiponectin levels afforded by treatment with 50 that were observed in the present
study. However, the mechanism(s) by which a selective ERRα modulator such as 50 would effect such
an increase is not clear. It has been shown that transactivation of ADIPOQ can be mediated through a
direct interaction of a PPARγ/RXRα heterodimer on the adiponectin promoter, and further augmented
4
4
by PPARγ agonists.
Accordingly, rosiglitazone treatment of fa/fa rats provided the expected
significant increase in adiponectin levels. However, the increase in adiponectin levels resulting from
treatment with 50 would appear to result from a distinct mechanism independent of direct PPARγ or
RXRα activation. No evidence of binding of 50 to either receptor was detected in vitro, nor were gene
expression levels of either PPAR
regimen. However, although RXR
significant increase in expression of RXR
in a compensatory capacity for this pathway. A slight increase in ADIPOR2 gene expression was also
γ
or RXR
gene expression was unchanged, treatment with 50 did lead to a
in adipose (+ 2.6-fold; p < 0.001), which possibly may serve
α in adipose tissue affected following a 15 d treatment
α
γ
2
0

ACCEPTED MANUSCRIPT
detected (+ 1.25-fold; p = 0.024), which may be an additional contributing factor for the improved
insulin sensitivity afforded by 50.
The means by which 50 afforded an increased expression of ANGPTL4 in fa/fa rat adipose is
presently unclear. ANGPTL4 is generally regarded as a PPAR target gene that encodes angiopoietin-
like protein 4, an adipokine whose expression is responsive to nutritional state, being elevated during
4
5, 46
times of fasting and reduced under the fed state, particularly with a high fat diet.
Overexpression of
ANGPTL4 has been shown to improve glucose tolerance in mice, while at the same time inducing
4
7
hyperlipidemia and hepatic steatosis. Microarray analysis however, revealed no significant changes in
expression levels of PPARA, PPARD or PPARG in the adipose of animals treated with 50. Nonetheless,
a modest but significant increase in ANGPTL4 expression resulted from treatment of fa/fa rats with 50,
which may contribute to the improved glucose tolerance that was observed following chronic treatment.
However, increased ANGPTL4 expression would appear to be inconsistent with the normalization of
circulating and hepatic triglycerides provided by 50 or rosiglitazone. Thus, these results may indicate
that alternative operational mechanisms (upregulation of fatty acid oxidation pathways, for example)
provide overriding influences on lipidemia in these animals. It is also possible that in spite of increased
gene expression, sustained protein levels of ANGPTL4 were insufficiently elevated by either treatment
4
7
to effect dyslipidemia, as has previously been postulated.
The results presented herein expand upon our earlier studies on the metabolic consequences of
2
2
modulating ERRα activity in vivo.
In two animal models of obesity and diabetes, significant
improvements in several parameters associated with glucose intolerance resulted from treatment with
the novel, selective ERRα modulator, 50. While the underlying mechanisms responsible for effecting
these metabolic improvements may involve some overlapping PPAR-responsive pathways, ERRα
modulators such as 50 also appear to provide access to unique, PPAR-independent mechanisms by
which metabolic function can be improved in vivo. These findings provide further evidence in support
2
1

ACCEPTED MANUSCRIPT
of the proposal that appropriate pharmacological modulation of ERRα function may constitute a viable
therapeutic approach for the treatment of metabolic diseases such as diabetes.
Experimental Section
Reagents and solvents were obtained from commercial suppliers and used without further purification.
Flash chromatography was performed with pre-packed silica gel cartridges (Redisep or Analogix) using
a Teledyne-Isco CombiFlash Companion chromatography system. Radial chromatography was
performed on a Cyclograph™ centrifugal chromatography system (Analtech, Inc.) with Silica Gel GF
rotor plates (Analtech, Inc.; 1000 µm, 2000 µm, or 4000 µm). Preparative HPLC purifications were
carried out on a Gilson 333/334 system configured with a 215 liquid handler and 155 UV/Vis detector
using a YMC C18 ODS-A column (5 µm, 30 mm × 100 mm). The HPLC methods incorporated UV
detection at 214 and 254 nm, ambient column temperature, a flow rate of 32.0 mL/min, and a binary
1
solvent system of 0.01% TFA in water (solvent A) and 0.01% TFA in CH CN (solvent B). H-NMR
3
spectra were recorded on a Bruker B-ACS-120 (400 MHz) spectrometer, a Varian Mercury (400 MHz)
spectrometer, or a Varian VXR-300 spectrometer (300 MHz) at room temperature. 2-D NMR spectra
were recorded on a Bruker DRX 500 (500 MHz) spectrometer. Chemical shifts are given in ppm (δ),
coupling constants (J) are in hertz (Hz), and signals are designated as follows: (s) singlet, (d) doublet, (t)
triplet, (q) quartet, (m) multiplet, (br s) broad singlet, (br d) broad doublet, (dd) doublet of doublet,
(ddd) doublet of doublet of doublet, (dtd) doublet of triplet of doublet. The purities of test compounds
were determined by LC-MS analysis, which was performed on a system consisting of an APCI
ionization source on a Finnigan LCQ ion trap mass spectrometer, a Hewlett Packard G1315A diode
array detector, a Hewlett Packard G1315A quaternary gradient pumping system, a Gilson 215
configured as an autosampler, and an Agilent Zorbax SB-C18 HPLC column (5 µm, 2.1 mm × 50 mm).
The HPLC method incorporated UV detection at 214 nm, ambient column temperature, a flow rate of
1
.0 mL/min, a 30 µL injection volume and a binary solvent system of 0.01% TFA in water (solvent A)
and 0.01% TFA in MeCN (solvent B). The following gradient (8 min) was used: 0-6 min, 5-100% B; 6-
2
2

ACCEPTED MANUSCRIPT
7
min, 100% B; 7-7.1 min, 5% B; 7.1-8 min, 5% B. All final compounds had a minimal purity of 98%.
Elemental analysis of compound 50 was performed by QTI, Inc., Whitehouse, NJ, and was within
2
8
±0.4% of the theoretical value.
Syntheses of N-1-Benzylindazoles (General Procedure A): A solution of indazole-5-carbaldehyde
1
(52.0 mmol) and an appropriately substituted benzyl bromide (62.1 mmol) in DMF (120 mL) was
treated with Cs CO (17 g, 52.2 mmol), and the mixture was heated at 90 °C for 16 h. The reaction was
2
3
cooled to RT and partitioned between EtOAc and H
brine, dried (Na SO ) and concentrated under reduced pressure. Purification of the regioisomeric
product mixture by silica gel chromatography (EtOAc/hexane or DCM/hexane) afforded the desired N-
-benzyl-1H-indazole-5-carbaldehyde (3a-c). (The corresponding N-2 benzylated regioisomers 4a-c,
which eluted later, was generally not isolated.)
-[4-Methoxy-2-(trifluoromethyl)benzyl]-1H-indazole-5-carbaldehyde (3a). H NMR (400 MHz,
CDCl 10.05 (s, 1H), 8.30 (s, 1H), 8.26 (d, J = 1.17 Hz, 1H), 7.92 (dd, J = 9.00, 1.57 Hz, 1H), 7.38
d, J = 8.61 Hz, 1H), 7.24 (d, J = 2.74 Hz, 1H), 6.88 (dd, J = 8.61, 2.35 Hz, 1H), 6.71 (d, J = 8.61 Hz,
2
O. The organic phase was washed with water (3x),
2
4
1
1
1
3
) δ
(
1
3
1
H), 5.78 (s, 2H), 3.80 (s, 3H). LCMS (APCI+) m/z: mass calcd. for C H F N O : 334.09, found:
1
7
13
3
2
2
+
35.1 [(M+H) ].
1
-[4-Chloro-2-(trifluoromethyl)benzyl]-1H-indazole-5-carbaldehyde (3b). H NMR (400 MHz,
CDCl
3
)
δ
10.06 (s, 1H), 8.50 (s, 1H), 8.46 (d, J = 0.73 Hz, 1H), 7.92 (dd, J = 8.80, 1.22 Hz, 1H), 7.89
(d, J = 2.20 Hz, 1H), 7.83 (d, J = 8.80 Hz, 1H), 7.66 (dd, J = 8.31, 1.96 Hz, 1H), 6.78 (d, J = 8.31 Hz,
1
H), 5.90 (s, 2H). LCMS (APCI+) m/z: mass calcd. for C H ClF N O: 338.04, found: 339.1
1
6
10
3
2
+
[
(M+H) ].
1
1
1
-[2,4-Bis(trifluoromethyl)benzyl]-1H-indazole-5-carbaldehyde (3c). H NMR (400 MHz, CDCl
3
)
δ
0.08 (s, 1H), 8.34 (s, 1H), 8.31 (s, 1H), 7.99 (s, 1H), 7.96 (dd, J = 8.84, 1.26 Hz, 1H), 7.64 (d, J = 8.08
Hz, 1H), 7.37 (d, J = 8.84 Hz, 1H), 6.83 (d, J = 8.08 Hz, 1H), 5.91 (s, 2H). LCMS (APCI+) m/z: mass
+
calcd. for C H F N O: 372.07, found: 373.1 [(M+H) ].
1
7
10
6
2
2
3

ACCEPTED MANUSCRIPT
Syntheses of [N-1-Benzylindazol-5-ylmethylene]-1,3-thiazolidine-2,4-diones (General Procedure
B): A mixture of thiazolidine-2,4-dione (1.0 mmol), substituted 1-benzyl-1H-indazolecarbaldehyde
from Procedure A (1.0 mmol), NH OAc (2.0 mmol) and acetic acid (2.0 mL) and was heated at 110 °C
4
for 12 h. After cooling to rt, the precipitated product was collected by filtration, washed extensively
with water and triturated with EtOAc/hexane or cold EtOH to afford the desired pure 5-[1-(substituted
benzyl)-1H-indazol-5-ylmethylidene]-1,3-thiazolidine-2,4-dione product.
(5Z)-5-({1-[4-methoxy-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
1
thiazolidine (5a). H NMR (400 MHz, DMSO-d ) δ 12.58 (br s., 1H), 8.32 (d, J = 0.73 Hz, 1H), 8.11
6
(s, 1H), 7.94 (s, 1H), 7.75 (d, J = 8.80 Hz, 1H), 7.63 (dd, J = 1.59, 8.93 Hz, 1H), 7.27 (d, J = 2.69 Hz,
1
H), 7.13 (dd, J = 2.69, 8.80 Hz, 1H), 6.79 (d, J = 8.80 Hz, 1H), 5.78 (s, 2H), 3.80 (s, 3H). LCMS
+
(
APCI+) m/z: mass calcd. for C H F N O S: 433.07, found: 434.3 [(M+H) ].
2
0
14
3
3
3
(5Z)-5-({1-[4-Chloro-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
1
thiazolidine (5b). H NMR (400 MHz, DMSO-d ) δ 12.59 (br s, 1H), 8.36 (d, J = 0.73 Hz, 1H), 8.13
6
(s, 1H), 7.95 (s, 1H), 7.88 (d, J = 2.20 Hz, 1H), 7.79 (d, J = 8.80 Hz, 1H), 7.66 (dd, J = 1.34, 8.93 Hz,
2
4
H), 6.76 (d, J = 8.31 Hz, 1H), 5.86 (s, 2H). LCMS (APCI+) m/z: mass calcd. for C H ClF N O S:
19 11 3 3 2
+
37.02, found: 438.3 [(M+H) ].
(5Z)-5-({1-[2,4-Bis-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
1
thiazolidine (5c). H NMR (400 MHz, DMSO-d ) δ 12.61 (s, 1H), 8.39 (s, 1H), 8.14 (d, J = 12.13 Hz,
6
2
1
4
H), 7.97 (d, J = 11.12 Hz, 1H), 7.96 (s, 1H), 7.83 (d, J = 8.84 Hz, 1H), 7.67 (dd, J = 8.84, 1.52 Hz,
H), 6.91 (d, J = 8.34 Hz, 1H), 5.99 (s, 2H). LCMS (APCI+) m/z: mass calcd. for C H F N O S:
2
0
11
6
3
2
+
70.04, found: 513.2 [(M + H + MeCN) ].
[(5Z)-5-({1-[4-Methoxy-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
thiazolidin-3-yl]acetic acid (21). (A) A mixture of thiazolidinedione 5a (200 mg; 0.46 mmol), tert-
butyl bromoacetate (104 µL; 0.81 mmol) and K CO (164 mg; 1.19 mmol) in DMF (3 mL) was stirred
2
3
at 60 °C for 2 h, then cooled to RT. Water (12 mL) was added to the mixture, producing a milky
2
4

ACCEPTED MANUSCRIPT
solution which was extracted with ether and the combined organics were concentrated. The crude
product thus obtained was purified by silica gel chromatography (50 – 67%) DCM/hexanes to afford
(
5Z)-tert-butyl 2-(5-((1-(4-methoxy-2-(trifluoromethyl)benzyl)-1H-indazol-5-yl)methylene)-2,4-dioxo-
thiazolidin-3-yl)acetate (217 mg; 86%) as a pale yellow solid, which was used directly.
B) A solution of the above ester (0.82 mmol) in DCM (5 mL) and TFA (5 mL) was stirred at RT for 2
h, then concentrated under reduced pressure. The solid thus obtained was boiled in MeCN (10 mL) for
(
5
min, then allowed to cool to RT. The product was filtered and washed with MeCN to afford
1
carboxylic acid 21 as an off-white powder. H NMR (400 MHz, DMSO-d ) δ 13.46 (br s, 1H), 8.35 (s,
6
1
2
3
4
H), 8.20 (s, 1H), 8.15 (s, 1H), 7.78 (d, J = 9.00 Hz, 1H), 7.69 (dd, J = 1.57, 9.00 Hz, 1H), 7.28 (d, J =
.74 Hz, 1H), 7.14 (dd, J = 2.93, 8.80 Hz, 1H), 6.81 (d, J = 9.00 Hz, 1H), 5.79 (s, 2H), 4.39 (s, 2H),
+
.80 (s, 3H). LCMS (APCI+) m/z: mass calcd. for C H F N O S: 491.08, found: 492.3 [(M+H) ].
22
16
3
3
5
-[(5Z)-5-({1-[4-Methoxy-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
thiazolidin-3-yl]cyclohexanecarboxylic acid (22). (A) To a solution of 1,3-thiazolidine-2,4-dione
(
585 mg; 5 mmol), ethyl 4-hydroxycyclohexanecarboxylate (1.03g; 6 mmol) and triphenylphosphine
1.57g; 6 mmol) in THF (10 mL) was added a solution of diisopropyl azidodicarboxylate (1.18 mL; 6
(
mmol) in THF (5 mL) in dropwise fashion. Upon complete addition, the reaction was heated to 70 °C
and allowed to stir overnight. After cooling to room temperature, the reaction was concentrated and the
residual mixture was purified by silica gel chromatography eluting with DCM to afford ethyl 4-(2,4-
dioxothiazolidin-3-yl)cyclohexanecarboxylate (528 mg, 39%) as a colorless syrup that was used
directly.
(B) A mixture of ethyl 4-(2,4-dioxothiazolidin-3-yl)cyclohexanecarboxylate (528 mg; 1.95 mmol),
carboxaldehyde 3a (600mg; 1.8 mmol) and NH OAc (277mg; 3.6 mmol) in HOAc was heated at 110
4
°
C for 12 h. After cooling to rt, the reaction was concentrated and the residual mixture was purified by
silica gel chromatography (0 – 2% MeOH/DCM) to afford [(5Z)-5-({1-[4-methoxy-2-
trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-thiazolidin-3-
(
2
5

ACCEPTED MANUSCRIPT
yl]cyclohexanecarboxylic acid ethyl ester (391 mg; 37 %) as a pale yellow solid, which was used
directly.
(C) A mixture of the above ester (391 mg; 0.67 mmol), 10% HCl (5 mL) and dioxane (10 mL) was
heated at 95 °C for 24 h, then concentrated under reduced pressure. The residue was triturated with
minimal MeCN/ether (1:2), filtered and washed with ether to afford pure carboxylic acid 22 (304 mg;
1
8
2%) as a pale yellow powder. H NMR (400 MHz, DMSO-d
6
) δ 12.25 (br s, 1H), 8.33 (s, 1H), 8.14
(d, J = 2.74 Hz, 1H), 8.03 (d, J = 6.26 Hz, 1H), 7.76(d, J = 9.00 Hz, 1H), 7.65 (ddd, J = 1.57, 3.52, 9.00
Hz, 1H), 7.27 (d, J = 2.74 Hz, 1H), 7.14 (dd, J = 2.54, 8.80 Hz, 1H), 6.80 (d, J = 8.61 Hz, 1H), 5.78 (s,
2
H), 4.05 - 4.27 (m, 1H), 3.80 (s, 3H), 1.94 - 2.30 (m, 5H), 1.69 - 1.85 (m, 1H), 1.31 - 1.68 (m, 3H).
+
LCMS (APCI+) m/z: mass calcd. for C H F N O S: 559.14, found: 560.0 [(M+H) ].
2
7
24
3
3
5
(
5Z)-5-[(1-{[4-(Methyloxy)-2-(trifluoromethyl)phenyl]methyl}-1H-indazol-5-yl)methylidene]-3-(2-
morpholin-4-ylethyl)-1,3-thiazolidine-2,4-dione Hydrochloride (23). A mixture of 4-(2-
chloroethyl)morpholine hydrochloride (140 mg; 0.75 mmol), indazolydinylthiazolidine-2,4-dione 5a
215 mg; 0.5 mmol) and K CO (205 mg; 1.5 mmol) in DMF (2.5 mL) was heated at 90 °C overnight.
(
2
3
After cooling to room temperature, the mixture was diluted with water (8 mL) and extracted with DCM.
The organic phase was dried (K CO ), concentrated and the residue was purified by silica gel
2
3
chromatography (0 – 2% MeOH/DCM) to afford the product as a fluffy, pale yellow solid. The material
was taken up in MeOH (5 mL) and treated with a large excess of HCl/ether. The resulting precipitate
was filtered and washed with ether to afford the corresponding HCl salt (219 mg; 75%) as a white
1
powder (slight yellow tint). H NMR (400MHz, DMSO-d ) δ 10.05 (br s, 1 H), 8.35 (s, 1 H), 8.19 (s, 1
6
H), 8.12 (s, 1 H), 7.74 - 7.86 (m, 1 H), 7.69 (dd, J = 1.57, 9.00 Hz, 1 H), 7.28 (d, J = 2.74 Hz, 1 H), 7.14
(dd, J = 2.35, 8.61 Hz, 1 H), 6.84 (d, J = 8.61 Hz, 1 H), 5.79 (s, 2 H), 3.90 - 4.17 (m, 4 H), 3.80 (s, 3 H),
3
.52 - 3.73 (m, 4 H), 3.4-3.5 (m, 2H), 3.08 – 3.24 (m, 2 H). LCMS (APCI+) m/z: mass calcd. for
+
C
26
H
25
F
3
N
4
O
4
S: 546.15, found: 547.2 [(M+H) ].
2
6

ACCEPTED MANUSCRIPT
(5Z)-5-({1-[4-Methoxy-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-3-(2-pyrrolidin-
1
-ylethyl)-1,3-thiazolidine-2,4-dione (24). To a solution of thiazolidinedione 5a (27 mg, 0.06 mmol)
in anhydrous DMF (1.5 ml) was added 1-(2-chloroethyl)pyrrolidine hydrochloride (12.7 mg, 0.07
mmol), followed by K CO (25.8 mg, 0.19 mmol). The resulting mixture was stirred at 65 °C for 2 h,
2
3
then diluted with water and extracted with EtOAc (3X). The combined organic extracts were washed
with brine, dried (Na SO ), filtered and concentrated under reduced pressure. The residue was purified
2
4
by silica gel chromatography eluting with DCM, then EtOAc to afford the desired product (26.1 mg;
1
7
1
8
9%) as a yellow solid. H NMR (400 MHz, CDCl ) δ 8.19 (d, J = 0.78 Hz, 1H), 8.01 (s, 1H), 7.94 (m,
3
H), 7.48 (dd, J = 8.61, 1.57 Hz, 1H), 7.36 (d, J = 8.61 Hz, 1H), 7.23 (d, J = 2.74 Hz, 1H), 6.88 (dd, J =
.61, 2.74 Hz, 1H), 6.71 (d, J = 8.61 Hz, 1H), 5.76 (s, 2H), 3.91 (t, J = 6.65 Hz, 2H), 3.80 (s, 3H), 2.76
(
t, J = 7.04 Hz, 2H), 2.59 (m, 4H), 1.77 (m, 4H). LCMS (APCI+) m/z: mass calcd. for C H F N O S:
26 25 3 4 3
+
5
30.16, found: 531.3 [(M+H) ].
[(5Z)-5-({1-[4-Chloro-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
thiazolidin-3-yl]acetic acid (25). (A) A mixture of thiazolidinedione 5b (438 mg; 1 mmol), tert-butyl
bromoacetate (225 µL; 1.5 mmol) and K CO (415 mg; 3 mmol) in DMF (4 mL) was stirred at 60 °C
2
3
for 1-2 h, then cooled to RT. Water (12 mL) was added to the mixture, producing a milky suspension;
ether (4 mL) was then added affording a triphasic mixture (the solid product was suspended in the ether
phase). After removal of the aqueous phase, hexane (2 mL) was added and the mixture was filtered and
washed with ether/hexane (1:2) to afford pure 5-[1-(substituted benzyl)-1H-indazol-5-ylmethylene]-
(
2,4-dioxo-1,3-thiazolidin-3-yl)acetic acid tert-butyl ester (473 mg; 86%) as a slightly yellow-tinted
white powder, which was used directly.
B) A solution of the above ester (473 mg; 0.86 mmol) in DCM (5 mL) and TFA (5 mL) was stirred at
(
RT for 2 h, then concentrated under reduced pressure. The solid thus obtained was boiled in MeCN (10
mL) for 5 min, then allowed to cool to RT. The product was filtered and washed with MeCN to afford
1
acid 25 as an off-white powder (425 mg, 99%). H NMR (400 MHz, DMSO-d ) δ 13.46 (br s, 1H),
6
2
7

ACCEPTED MANUSCRIPT
8
.38 (s, 1H), 8.21 (s, 1H), 8.16 (s, 1H), 7.89 (d, J = 1.96 Hz, 1H), 7.83 (d, J = 8.61 Hz, 1H), 7.71 (dd, J
=
1.56, 9.00 Hz, 1H), 7.66 (dd, J = 1.96, 8.22 Hz, 1H), 6.78 (d, J = 8.61 Hz, 1H), 5.88 (s, 2H), 4.40 (s,
2
H), 3.34 (s, 2H). LCMS (APCI+) m/z: mass calcd. for C H ClF N O S: 495.03, found: 496.2
2
1
13
3
3
4
+
[
(M+H) ].
4
-[(5Z)-5-({1-[4-Chloro-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
thiazolidin-3-yl]cyclohexanecarboxylic acid (26). (A) To a solution of indazolydinylthiazolidine-2,4-
dione 5b (500 mg; 1.14mmol), ethyl 4-hydroxycyclohexanecarboxylate (368 µL; 2.28 mmol) and
triphenylphosphine (599 mg; 2.28 mmol) in THF (20 mL) was added a solution of diisopropyl
azidodicarboxylate (450 µL; 2.28 mmol) in a dropwise fashion. After stirring at room temperature for 2
h, additional portions (1.14 mmol each) of alcohol, triphenylphosphine and diisopropyl
azidodicarboxylate were added and stirring was continued for an additional 2 h, at which time LCMS
analysis indicated complete consumption of 5b. The reaction was concentrated and the residue was
taken up in DCM and purified by silica gel chromatography eluting with (0 – 40 %) EtOAc/hexanes to
afford [(5Z)-5-({1-[-2,4-bis(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-2,4-dioxo-1,3-
thiazolidin-3-yl]cyclohexanecarboxylic acid ethyl ester (600 mg; 89 %) as a white solid (following
recrystallization from MeCN), which was used directly.
(B) A mixture of the above ester (450 mg; 0.76 mmol), 10% HCl (5 mL) and dioxane (10 mL) was
heated at 100 °C for 12 h, then cooled to 0 °C, upon which the product precipitated as a white solid.
The precipitate was filtered, washed with water and air-dried to afforded pure carboxylic acid 26 (410
mg; 96%) as a white powder, which was directly converted to the corresponding ethanolamine salt.
(C) A portion of the above carboxylic acid (300 mg; 0.53 mmol) was dissolved in hot THF (1.5 mL)
and ethanolamine (48 µL; 0.8 mmol) was added, resulting in the formation of a heavy precipitate. After
cooling to room temperature, the reaction was treated with ether (20 mL) and filtered to furnish the
1
corresponding ethanolamine salt (307 mg, 92%). H NMR (DMSO-d ) δ 8.36 (s, 1H), 8.15 (s, 1H),
6
8
.04 (s, 1H), 7.89 (s, 1H), 7.80 (d, J = 8.61 Hz, 1H), 7.60 - 7.74 (m, 2H), 6.77 (d, J = 8.61 Hz, 1H), 5.86
2
8

ACCEPTED MANUSCRIPT
(s, 2H), 4.13 (t, J = 12.33 Hz, 1H), 3.43 (t, J = 5.48 Hz, 2H, ethanolamine), 2.66 (t, J = 5.28 Hz, 2H,
ethanolamine), 1.92 - 2.22 (m, 5H), 1.73 (d, J = 10.96 Hz, 2H), 1.28 - 1.48 (m, 2H). LCMS (APCI+)
+
•
m/z: mass calcd. for C H ClF N O S: 563.09, found: 564.0 [(M) ].
2
6
21
3
3
4
(5Z)-5-({1-[4-Chloro-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-3-(2-morpholin-4-
ylethyl)-1,3-thiazolidine-2,4-dione (27). To a solution of thiazolidinedione 5b (30 mg, 0.07 mmol) in
anhydrous DMF (0.7 ml) was added 4-(2-chloroethyl)morpholine hydrochloride (15 mg, 0.08 mmol),
followed by K
with water and extracted with EtOAc (3X). The combined organic extracts were washed with brine,
dried (Na SO ), filtered and concentrated under reduced pressure. The residue was purified by silica gel
2 3
CO (30 mg, 0.2 mmol). The resulting mixture was stirred at 70 °C for 3 h, then diluted
2
4
chromatography (67% EtOAc /hexanes) to afford the desired product (32 mg; 83%) as a yellow solid.
1
H NMR (400 MHz, CDCl ) δ 8.22 (s, 1H), 8.01 (s, 1H), 7.97 (s, 1H), 7.72 (d, J = 1.96 Hz, 1H), 7.52
3
(dd, J = 1.76, 8.80 Hz, 1H), 7.30 - 7.39 (m, 2H), 6.67 (d, J = 8.61 Hz, 1H), 5.80 (s, 2H), 3.90 (t, J = 6.46
Hz, 2H), 3.66 (t, J = 4.50 Hz, 4H), 2.64 (t, J = 6.46 Hz, 2H), 2.52 (br s, 4H). LCMS (APCI+) m/z: mass
+
calcd. for C H ClF N O S: 550.11, found: 551.3 [(M+H) ].
2
5
22
3
4
3
(5Z)-5-({1-[4-Chloro-2-(trifluoromethyl)benzyl]-1H-indazol-5-yl}methylidene)-3-(2-pyrrolidin-1-
ylethyl)-1,3-thiazolidine-2,4-dione (28). A mixture of 1-(2-bromoethyl)pyrrolidine hydrobromide (354
mg; 1.38 mmol), indazolydinyl-thiazolidine-2,4-dione 5b (400 mg; 0.92 mmol) and K CO (317 mg;
2
3
2
.3 mmol) in DMSO (5 mL) was heated at 80 °C for 5 h. After cooling to rt, the reaction was diluted
with water and extracted with EtOAc. The organic phase was back-extracted with saturated NaCl, dried
Na SO ) and concentrated under reduced pressure. The residue was then purified by preparative HPLC
using a gradient (10% - 100%) of MeCN/H O (0.01% TFA). After evaporation of MeCN, combined
(
2
4
2
product-containing fractions were neutralized with sat’d NaHCO and extracted with DCM. The
3
organic phase was dried (Na SO ) and concentrated to afford 28 (260 mg, 53%) as an off-white solid.
2
4
1
H NMR (400 MHz, CDCl ) δ 8.21 (s, 1H), 8.01 (s, 1H), 7.96 (s, 1H), 7.71 (d, J = 2.02 Hz, 1H), 7.51
3
(dd, J = 1.52, 8.84 Hz, 1H), 7.29 - 7.39 (m, 2H), 6.67 (d, J = 8.59 Hz, 1H), 5.79 (s, 2H), 3.92 (t, J = 6.82
2
9