Design, synthesis and biological evaluation of novel TRβ selective agonists sustained by ADME-toxicity analysis

Article abstract of DOI:10.1016/j.ejmech.2019.112006

Although triiodothyronine (T3) induces several beneficial effects on lipid metabolism, its use is hampered by toxic side-effects, such as tachycardia, arrhythmia, heart failure, bone and muscle catabolism and mood disturbances. Since the α isoform of thyroid hormone receptors (TRs) is the main cause of T3-related harmful effects, several efforts have been made to develop selective agonists of the β isoform that could induce some beneficial effects (i.e. lowering triglyceride and cholesterol levels reducing obesity and improving metabolic syndrome), while overcoming most of the adverse T3-dependent side effects. Herein, we describe the drug discovery process sustained by ADME-Toxicity analysis that led us to identify novel agonists with selectivity for the isoform TRβ and an acceptable off-target and absorption, distribution metabolism, excretion and toxicity (ADME-Tox) profile. Within the small series of compounds synthesized, derivatives 1 and 3, emerge from this analysis as “potentially safe” to be engaged in preclinical studies. In in vitro investigation proved that both compounds were able to reduce lipid accumulation in HepG2 and promote lipolysis with comparable effects to those elicited by T3, used as reference drug. Moreover, a preliminary in vivo study confirmed the apparent lack of toxicity, thus suggesting compounds 1 and 3 as new potential TRβ-selective thyromimetics.

Full text of DOI:10.1016/j.ejmech.2019.112006

Journal Pre-proof
Design, synthesis and biological evaluation of novel TRβ selective agonists sustained
by ADME-toxicity analysis
Massimiliano Runfola, Simona Sestito, Lorenza Bellusci, Valeria La Pietra, Vincenzo
Maria D'Amore, Marta Anna Kowalik, Grazia Chiellini, Sheraz Gul, Andrea Perra,
Amedeo Columbano, Luciana Marinelli, Ettore Novellino, Simona Rapposelli
PII:
S0223-5234(19)31163-8
DOI:
https://doi.org/10.1016/j.ejmech.2019.112006
EJMECH 112006
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 24 September 2019
Revised Date: 11 December 2019
Accepted Date: 20 December 2019
Please cite this article as: M. Runfola, S. Sestito, L. Bellusci, V. La Pietra, V.M. D'Amore, M.A. Kowalik,
G. Chiellini, S. Gul, A. Perra, A. Columbano, L. Marinelli, E. Novellino, S. Rapposelli, Design, synthesis
and biological evaluation of novel TRβ selective agonists sustained by ADME-toxicity analysis,
European Journal of Medicinal Chemistry (2020), doi: https://doi.org/10.1016/j.ejmech.2019.112006.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
© 2019 Published by Elsevier Masson SAS.

Design, synthesis and biological evaluation of novel TRβ selective agonists
sustained by ADME-Toxicity analysis
Massimiliano Runfola1#, Simona Sestito1#, Lorenza Bellusci2, Valeria La Pietra3, Vincenzo
Maria D’Amore3, Marta Anna Kowalik4, Grazia Chiellini2*, Sheraz Gul5, Andrea Perra4,
Amedeo Columbano4, Luciana Marinelli3, Ettore Novellino3, Simona Rapposelli1,6*
1Department of Pharmacy, University of Pisa, Pisa, 56126, Italy;
2Department of Pathology, University of Pisa, Pisa, 56126, Italy;
3Department of Pharmacy, University of Naples Federico II, Italy;
4Department of Biomedical Sciences, Unit of Oncology and Molecular Pathology, University
of Cagliari, Cagliari, Italy;
5Fraunhofer Institute for Molecular Biology & Applied Ecology – ScreeningPort, Hamburg,
Germany;
6Interdepartmental Research Centre for Biology and Pathology of Aging, University of Pisa,
Pisa, Italy.
# M.R and S.S equally contributed;
*For S.R.: phone, +39 050 2219582; fax, +39 050 2219577; Email:
simona.rapposelli@unipi.it
*For G.C.: phone, +39 050 2218662; E-mail, grazia.chiellini@unipi.it

HIGHLIGHTS
• A novel series of thyromimetics was developed and their ADME-Toxicity profile was
assessed.
• To simulate the interaction of selected compounds with the putative receptor-binding site
in silico docking-studies were performed
• Compounds 1 and 3 proved to be TRβ selective agonists and promote lipolysis through the
activation of AMPK/ACC pathway
• Preliminary in vivo studies of thyromimetics 1 and 3 confirm a safe-profile of both
compounds, lacking both cardio- and hepatotoxicity.
ABSTRACT
Although triiodothyronine (T3) induces several beneficial effects on lipid metabolism, its use
is hampered by toxic side-effects, such as tachycardia, arrhythmia, heart failure, bone and
muscle catabolism and mood disturbances. Since the α isoform of thyroid hormone receptors
(TRs) is the main cause of T3-related harmful effects, several efforts have been made to
develop selective agonists of the β isoform that could induce some beneficial effects (i.e.
lowering triglyceride and cholesterol levels reducing obesity and improving metabolic
syndrome), while overcoming most of the adverse T3-dependent side effects. Herein, we
describe the drug discovery process sustained by ADME-Toxicity analysis that led us to
identify novel agonists with selectivity for the isoform TRβ and an acceptable off-target and
absorption, distribution metabolism, excretion and toxicity (ADME-Tox) profile. Within the
small series of compounds synthesized, derivatives 1 and 3, emerge from this analysis as
“potentially safe” to be engaged in preclinical studies. In in vitro investigation proved that
both compounds were able to reduce lipid accumulation in HepG2 and promote lipolysis with
comparable effects to those elicited by T3, used as reference drug. Moreover, a preliminary in
vivo study confirmed the apparent lack of toxicity, thus suggesting compounds 1 and 3 as
new potential TRβ-selective thyromimetics.

KEYWORDS: Triiodothyronine, thyronamine, TRβ selective agonist, fatty-liver disorder,
liver regeneration
INTRODUCTION
Thyroid hormones (THs) are essential regulatory molecules for normal growth and
development and for maintaining metabolic homeostasis [1]. Most of the activities elicited by
THs are mediated by nuclear thyroid hormones receptors (TRs), which are members of the
nuclear hormone receptor family and act as ligand-activated transcription factors. There are
two main TR isoforms encoded on separate genes, namely TRα and TRβ. Both TR isoforms
bind 3,5,3'-triiodothyronine (T3) and mediate TH-regulated gene expression through
interactions with DNA response elements and with a complex array of transcriptional
cofactors, including corepressors (CoRs), coactivators (CoAs), integrators like CREB-
binding protein (CBP), and general transcription factors (GTFs) [2]. TH/TR binding exerts
profound effects in several physiological processes. In the central nervous system (CNS) THs
signaling is involved in development and maintenance of brain function, influencing various
activities such as neuronal and glial cell differentiation, myelination, and neurogenesis [3, 4].
In the liver, THs influence hepatic lipid metabolism through multiple pathways, with
insightful effects on energy expenditure [5], fat oxidation [6] and cholesterol metabolism [7].
Conversely, alteration of cellular TH signaling in the liver plays a key role in the onset or
progression of several liver-associated diseases, such as non-alcoholic fatty liver disease
(NAFLD), and hepatocellular carcinoma (HCC) [8]. Therefore, modulating the function of
THs offers the potential to treat a wide variety of liver diseases and cancer, as well.
Unfortunately, the use of THs as therapeutic agents is hampered by the lack of selectivity and
the consequent adverse side effects (such as increased heart rate, cardiac hypertrophy, muscle
wasting, and reduced bone density) that are mainly due to their binding to TRα receptors. In
order to overcome these drawbacks, selective activation of the β thyroid hormone receptor
(TRβ) is an appropriate method to develop new treatments for several chronic diseases with a
reduced side effect profile.
During the past two decades, a number of selective TRβ agonists have been developed. These
include the analogues Sobetirome (GC-1), KB-141 and the Hep-Direct prodrug VK2809 that
exhibits most of the beneficial properties of T3 in the absence of deleterious effects [9-11].

GC-1 emerged in 1995 [9] as the first halogen free thyromimetics. It was shown to bind
preferentially to TRβ vs TRα and was initially studied as an hypocholesterolemic compound
able to stimulate hepatic pathways without harmful side effects [12]. Due to the beneficial
effects of GC-1 and KB2115, the latter known as Eprotirome, entered human clinical trials
for dyslipidaemia, and yielded encouraging results in the absence of harmful effects typically
associated with THs high levels (for reviews, see Tancevski et al. 2011 [13], Meruvu et al.
2013 [14]). The progression of GC-1 was discontinued despite a promising phase I trial for
the treatment of lipid metabolism disorders and obesity, with no phase II trials being planned.
Additional applications of GC-1 have been proposed in orphan indications and this drug is
now in phase I/II trial for a rare disease (X-Linked Adrenoleukodystrophy) (Clinicaltrial.gov
Identifier: NCT03196765). Eprotirome, progressed to phase III clinical trial but was
terminated due to unexpected side effects in animal studies [15]. Another selective
thyromimetic, namely Resmetirom (MGL-3196) was designed as a liver-specific TRβ agonist
able to reduce hepatic lipid synthesis and has progressed to phase II trials (ClinicalTrials.gov
Identifier: NCT02912260). In addition, VK2809 is a novel liver-directed thyroid receptor
beta agonist pro-drug [16] that possesses selectivity for liver tissue, and after undergoing
cleavage by a cytochrome P450 enzyme, is able to release the selective thyromimetic
MB07344 which binds to TRβ receptor (Figure 1). Currently, VK2809 has progressed to
Phase 2 clinical trial in patients with primary hypercholesterolemia and non-alcoholic fatty
liver disease (NAFLD). (ClinicalTrials.gov identifier: NCT02927184). Although many
thyromimetics are currently in clinical trials, none have been approved by FDA (Figure 1).

Figure 1. Structures of TRβ agonists which have been progressed in drug discovery.
Despite the challenges with progressing TRβ selective agonists due to their side effect
profiles, their use to treat other human pathologies could be explored as part of drug
repositioning strategies. Accordingly, the identification of novel agonists with selectivity for
both the TRβ and the liver could represent a valid strategy to regulate metabolic disorders.
This paper reports the results of the design and synthesis of a novel series of TRβ agonists.
The thyromimetic-analogues were evaluated for their off-target and ADME-Tox properties
and the most optimal compounds progressed to in vitro profiling in a nuclear receptor
coregulator assay (Thermofisher®) and for their ability to promote lipolysis in human
hepatoma cells (HepG2). Further investigation to determine the cellular pathways involved in
their lipid-lowering ability was also carried-out. Additionally, to rationalize the
pharmacological results, in silico docking studies on selected compounds were also
performed in order to simulate their interaction with the putative receptor-binding site.
Finally, we performed preliminary in vivo studies to assess the effect of two thyromimetics

(compound 3 and its pro-drug 1), which were shown to be associated with the most optimal
off-target and ADME-Tox profile, to determine their ability to modulate cholesterol serum
levels, triglicerides, glucose, bilirubin and transaminases in rats.
RESULTS & DISCUSSION
Rational design of new thyromimetic molecules
Our search for therapeutic agents focused on the design and synthesis of agonists endowed
with high selectivity for TRβ and enhanced hepato-specificity. To this end, we designed and
synthesized a new class of thyromimetics based on a diphenylmethane scaffold. This moiety
has been widely recognized as an effective replacement for the biaryl-ether core of THs [16-
20]. Notably, in previously developed diphenylmethane-based thyromimetics, the oxyacetic
acid side chain was shown to have a critical role in conferring TRβ selectivity [19]. On the
basis of this finding, we designed compounds 1-4 and 9-11 (Figure 2). In compounds 3, 4, 9
and 10, the hydroxyl group at the 4'-position in T3 was replaced by an amino group. This
type of modification generates the zwitterion that could be detrimental for the bioavailability
of the molecule. Hence, we synthesized the corresponding acetamide-analogues of 3 and 4,
namely 1 and 2, which could represent potential pro-drugs with improved drug-like
properties. Oral administration of these pro-drugs, by first pass metabolism through the liver,
should lead to an efficient generation of the corresponding parent compounds. Based on our
previous experience gained during the development of SG-compounds as thyronamine
analogues of T1AM, it is known that the oxyethylamine sidechain represents a valid
precursor of the oxyacetic group that is formed following the metabolic activation by MAOs
[18]. Therefore, we synthesized the derivatives 5-8 as possible precursors of the
corresponding thyromimetics with an oxyacetic structure. The two iodine atoms present in
the T3 inner ring have been replaced with hydrogen atoms (2 and 4), with two methyl groups
(1, 3, 5, 7, 10 and 11) or with a hydrogen and CF3 (or H and F) as in compounds 6, 8 and 9.
As to the iodine atom in the outer ring of T3, it has been replaced by an isopropyl group (1-6)
or by a hydrogen atom (7-11).

Figure 2. Structures of novel thyromimetics 1-11 with a biphenylmethane scaffold.
Chemistry
A crucial component in our synthetic procedures is the synthesis of the biphenylmethane
scaffold which was obtained with the palladium (0)-catalyzed Suzuki-Miyaura cross coupling
reaction between the substituted benzylbromide and selected phenyl boronic acid derivatives.
The Suzuki-Miyaura reaction was performed under standard or microwave-assisted
conditions, with yields ranging from 22% and 60% as reported in Table 1. In particular,
benzylbromide 15, which was not commercially available, was obtained as shown in Scheme
1. A one-pot reaction between 2-isopropylaniline, acetic anhydride, and bromine was carried
out in acetic acid providing the bromobenzene derivative 12 with high yields. Then,
formylation of 12 with n-BuLi and DMF at –78 °C led to benzaldehyde-derivative 13, which
was reduced with NaBH4 to provide the benzylic alcohol 14. Finally, 15 was obtained by
bromination of 14, using CBr4 and PPh3 (Appel reaction).

Scheme 1. Synthesis of benzylbromide 15. Reagents and conditions. i: Ac2O, AcOH, 117-
118 °C, 1 h; ii: Br2, AcOH, 65 °C, 30' ; iii: n-BuLi, DMF, THF, –78 °C → rt, 3 h; iv: NaBH4,
MeOH, 0 °C → rt, 12 h; v: CBr4, PPh3, DCM, 0 °C → rt, 12 h.
R1
R1
Catalyst (5mol%)
Base (1.5 eq)
B(OH)2
R2
R
X
R
X
Br
+
MeO
R2
OMe
Substrate
Product
A: R=H, X=NO2
B: R=iPr, X=NHAc
Substratea
Entry
Bb
R1
R2
Product Base
Catalyst
Solvent
Temp. Time Yieldd
Me
Me
16
17
18
24
K2CO3
K2CO3
K2CO3
PdCl2
acetone/H2O 40 °C
6h
60%
30%
22%
39%
100
Acetone/H2O
°C
Bb
Bb
Aa
H
H
H
H
PdCl2
10'c
12h
20'c
CF3
F
Pd(PPh3)4 1,4 dioxane
1,2-
90 °C
110
°C
Ba(OH)2*8H2O Pd(PPh3)4
DME/H2O
Aa
Aa
CF3
Me
H
25
26
K2CO3
K2CO3
Pd(PPh3)4 1,4 dioxane
90 °C
6h
52%
32%
Me
PdCl2
acetone/H2O rt
72h
a
b
Table 1. Cross-Coupling Reactions. commercially available starting material; compound
c
d
15; reaction was performed using a microwave reactor; yields refer to products after
compound purification which was performed by flash chromatography (for further
information on eluent mixture, consult supplementary material).

Scheme 2 illustrates the synthetic procedure to obtain compounds 1-6. Briefly, phenol
derivatives 19-21 were obtained by demethylation of purified cross coupling products 16-18
by using BBr3. The following O-alkylation reaction of 20 and 21 with bromoacetic acid led to
compound 1 and 2, which have undergone deacetylation in presence of hydrochloric acid at
concentrated grade to provide desired compound 3 or 4. Compounds 5 and 6 were obtained
by reaction of phenolic derivative 19 or 21 with bromoacetonitrile, followed by reduction of
intermediates 22 and 23 in presence of LiAlH4 and AlCl3.
Scheme 2. Synthesis of final compounds 1-6. Reagents and conditions. (i) BBr3, DCM, –10
°C, 1 h; (ii) BrCH2COOH, DMF, Cs2CO3, rt, 1 h; (iii) BrCH2CN, DMF, Cs2CO3, rt, 30'; (iv)
HCl 37%, H2O, 120 °C, 12 h; (v) LiAlH4, AlCl3, THF, rt → 66 °C, 12 h.
Compounds 7-11 were obtained according to the synthetic procedure reported in Scheme 3.
Briefly, purified cross-coupling products 24-26 were treated with BBr3 to afford the phenol
derivatives 27-29 which have undergone O-alkylation reaction with bromoacetonitrile or
ethyl bromoacetate leading to intermediates 30-32 and 33, respectively. Compound 9 was
obtained by selective nitro-group reduction (35), followed by acid hydrolysis with HCl at

concentrated grade. Treatment of 33 with hydrazine and FeCl3 led to ethyl ester 36 which was
converted to compound 10 by mild saponification. Finally, the conversion of amine moiety of
10 to a phenolic group using diazonium salt led to compound 11.
Scheme 3 Synthesis of final compounds 7-11. Reagents and conditions: A. (i) BBr3, DCM, –
10 °C, 1 h; (ii) BrCH2CN, DMF, Cs2CO3, rt, 20', 1 h; (iii) LiAlH4, AlCl3, THF, rt → 66 °C,
12 h; (iv) H2, Pd/C, MeOH, 12 h; (v) H2SO4, NaNO2, H2O, 100 °C, 1 h; (vi) HCl 37%, H2O,
100 °C, 4 h; B. (i) BrCH2COOEt, DMF, Cs2CO3, rt, 1 h; (ii) H2, Pd/C, EtOH, 150', rt (iii)
NaOH 10%, MeOH, 66 °C, 1 h; (iv) H2SO4, NaNO2, H2O, 100 °C, 1 h.
Off-target and ADME-Tox profiling

In order to identify the most promising compound for progression in the drug discovery
process, the novel thyromimetics 1-11 were screened in vitro at 10 ꢀM against a panel of
early ADME-Tox assays comprising cytotoxicity (U2OS, HEK, hTERT and MCF-7),
cardiotoxicity (hERG), cytochrome P450 inhibition (CYP1A2, CYP2C9, CYP2C19,
CYP2D6 and CYP3A4), and off-targets such as Aurora B kinase and phosphodiesterase
(PDE4C1), and epigenetic enzymes (SIRT7, HDAC4, HDAC6, HDAC8 and HDAC9), see
Figure 3. At first, the cytotoxicity of 1-11 were determined after 24 h or 48 h of treatment in
the four cell-lines. Reassuringly, only compound 6 was associated with a low degree of
cytotoxicity at 10 µM (65% at 24 h and 53% at 48 h) towards MCF-7 cell-line (see “paper
data in brief” for details). [Ref]
h
h
s
h
h
d
h
h
4
8
h
h
n
2
4
4
8
4
4
8
4
9
B
4
8
4
u
2
2
3
3
6
1
C
2
9
1
4
A
2
o
4
6
8
9
C
A
a
r
9
9
2
7
7
-
F
T
T
A
C
C
D
p
-
S
S
C
C
C
7
2
4
R
R
1
2
2
2
3
P
G
o
r
u
F
T
A
A
A
O
O
2
K
K
E
m
E
E
P
P
P
P
R
C
C
R
I
2
E
E
D
D
D
D
o
T
T
Y
Y
Y
Y
Y
D
P
E
C
1
2
3
4
5
6
7
8
9
U
U
H
H
h
h
M
M
C
C
C
C
C
S
H
H
H
H
A
h
2 -6.21 -8
-8 -0.08 -5
16 -3.53 -7
-7
2
1
1
3
6
0
48
1
13
33
14
14
8
36
33
-15
7
24
28
86
30
96
6
37
21
29
84
7
5
-15 -13
-12 -8
-2 -21
11 -14
7 5.927
-3
10
9
11
-1
32
13
53
4
23 2.656
31 4.38
15 13.9
26 6.617
23 3.507
45 9.552
17 8.486
17 17.78
32 8.047
23 4.31
-5 -2
11 14
-6
7
19
10
6
6
-9 -20
14 -19
-10 -15
-11 -32
-7 -21
-2 -16
15 -26
-8 -17
13 -33
1
-9
4
0 13.76
7
1
39
17
-1 -13
5
17
33 3.158
1
6
5
62 100
99
83
89
51
56
47
6
-2 -10
-15
-2 -10
34 13.05 17
8 3.447 -3
15 55
16 65
-1 38
97 100 100
82 100 100
97 129
5
1
-6
4
3
3
70
67
83
74
5
9 8.035
2
0
7
36
22
15
28
4
91
18
6
97
56
98
40
-9
10
6
2
12
3
-2
11
7 3.819 10
-2 3.649 -4
2 3.234 -6
9
13
4
47 103
10
11
-3
-5
-1 37
-2 24
-3 -26 -30
43 -15
-2
9
3
-11
21
2
3
14
9
-19
9
5
Cell growth inibition
ADME-Tox & Off Target Liability Panel
Classification
% Inhibition at 10 µM [compound]
Acceptable profile
<50
51-90
>91
A flag that requires remedial action
Major issue that requires significant attention
Off-target
Tox profile
Figure
and ADME-
3.
of compounds 1-11. All compounds were screened at 10 µM in triplicate. A traffic-light
system was used to characterize the effect of the new thyromimetics in the in vitro off-target
and ADME-Tox assays.
The ADME-Tox profile of a compound is a major factor when selecting the best candidate(s)
to progress in the drug discovery value chain. Among several enzymes responsible for
xenobiotic transformation, the cytochrome P450 (CYP450) enzyme plays a key role in phase
I oxidative metabolism and most of drug biotransformation are essentially achieved by five
isoforms: CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 [21]. Interestingly, while
the carboxylic compounds were associated with minimal adverse effect on CYP450 enzyme

activity, compounds 5-8 yielded considerably more inhibition against all CYP450 enzymes,
probably as a consequence of the oxyethylamine moiety. In order to assess the off-target
liability of newly synthesized molecules, all compounds were tested against several well-
known off-targets. For this study, enzymes belonging to different classes were selected.
SIRT7, HDAC4, HDAC6, HDAC8 and HDAC9 were chosen as representatives of an
undesirable epigenetic modulation [22, 23]. No compounds were associated with adverse
inhibition of these targets. Moreover, the new synthesized molecules were also shown to be
inactive against Aurora B kinase and PDE4C1 enzymes which are known to be essential for
cell growth and whose inhibition has been involved in tumorigenesis. Finally, the ability of
the newly synthesized molecules to modulate hERG ion-channel activity was assessed using
the Predictor hERG FP assay (Thermofisher®). The hERG ion-channel plays a key role in
cardiac action potential and represents one of the main reasons for failure of clinical
candidates or, even worse, a cause for withdrawal of drugs from the market. For this reason,
it is considered important to assess the effect of compounds on hERG ion-channel activity at
an early stage of drug discovery process. Compounds 1-4, 10 and 11 did not significantly
alter the hERG ion-channel activity, such that the percentage of inhibition was <40 % at 10
µM. More importantly, compound 1, 10, and 11 were shown to inhibit hERG ion-channel
activity <25 % at 10 µM (Figure 4). Overall, the off-target and ADME-Tox profiling led to
the identification of 6 compounds (1-4, 10, 11) with the most optimal profiles for
progression. Next, we proceeded with biological evaluation to assess their agonist activity
against TRα and TRβ.

Figure 4. hERG ion-channel inhibition profile of compounds 1-11. All compounds were
tested at 10 µM in triplicate, and raw data were normalized to positive control (Ctrl =
Compound E-4051) and negative control (DMSO) and the % Inhibition of the hERG ion-
channel activity calculated. Results were compared to control using the One-Way ANOVA
test. *** p<0.005.
Nuclear receptor activation
The activity of compounds 1-4 and 10-11 on TRα and TRβ was assessed using the
LanthaScreenTM TR-FRET Nuclear Receptor Coregulator Assay (service provided by
Invitrogen Corporation, USA) in 10-point dose-response experiments for each compound.
Among the tested compounds, only 3 and 11 were active against TRα and TRβ. Interestingly,
compound 11 activated both TR isoforms with EC50 towards TRα and TRβ of respectively
54 and 32 nM. Apart from ability to activate the receptors, the selectivity towards the TRβ
isoform is pivotal in order to avoid side effects related to TRα activation in vivo. Above all,
compound 3 was shown to be a TRβ agonist with EC50 value 458 nM, whereas it only
activated TRα to 30 % (at 1µM) (see Supporting Information for details). Based upon these
results, compound 3 was selected for further biological investigations. In addition, compound

1 was also selected for further study, as it was originally designed as pro-drug of compound
3, and more importantly, it was also found to possess an acceptable off-target and ADME-
Tox profile.
Computational studies
In order to elucidate the activity and the preferential binding of compound 3 for TRβ with
respect to TRα, docking calculations were carried out with the aid of Glide 6.7 (Maestro
packaging) into the ligand-binding-cavity (LBC) of the two receptors, using the structures co-
crystallized with GC-1, PDB: 3ILZ and 3IMY, respectively (See Experimental Section for X-
ray selection). Bearing in mind the importance of water-mediated interactions for GC-1,
which is a potent thyromimetic as well as a very close analogue of 3, our calculations were
performed retaining crystal water molecules within 3 Å of the ligand. As for TRβ, the best
ranked docking pose predicted for 3 (Figure 5) is highly superimposable with the crystal
structure of GC-1 into the thyroid receptor. Specifically, the ligand carboxylate forms a H-
bond with the backbone NH group of the N331 and engages two charge-reinforced H-bonds
with the guanidinium group of R282. Moreover, water-mediated contacts are found with the
R316 and R320 side chains. As for the dimethyl phenyl moiety, it establishes many van der
Waals interactions with the LBC lipophilic side chains of L330, I275, I353, M313 and M310
and, thanks to the 3,5 di-methyl substitutions, it is locked in the active perpendicular
conformation with respect to the terminal phenyl ring [24]. Other profitable lipophilic
interactions are detected between the latter and the L341, F272, I276, L346 side chains and
between the iso-propyl substituent with the residues F269, F272, M442, L346 and F451.
Finally, the 3 amino group H-bonds the imidazoline nitrogen atom of the H435. Of note,
when the zwitterionic form of 3 is docked into the receptor, numerous bad contacts are found
between the cationic amino group and the surrounding lipophilic residues of the LBC, which
might be responsible of the lower activity of 3 with respect to its hydroxyl analogue GC-1
(data from the literature). Moreover, the bulkiness of the residues F269, F455 and F439 does
not allow a proper accommodation of the 4'-NH-acetyl group of compound 1, generating
intramolecular steric clashes and hampering the formation of an H-bond with H435,
justifying the loss of activity of this prodrug (see Figure SI2 together with comments on the
other derivatives of the series that can be found in Supporting Information).

Figure 5. Binding mode of compound 3 in the TRβ LBC. The receptor is displayed in grey
cartoon, while the ligand and the main protein residues are depicted in magenta and gray
sticks respectively. H3 helix and the N331 loop are in transparency and non-polar hydrogens
are omitted for sake of clarity. H-bonds are shown as black dashed lines.
As for the selectivity profile of 3, in TRα a pose very similar to the co-crystallized structure
of GC-1 with the receptor is found. This outcome is not surprising since it is known that only
a single residue mutation (S277 in TRα versus N331 in TRβ) differentiates within the TRα
and TRβ binding pockets, and this mutation does not alter the overall shape and
characteristics of the two clefts. However, a deeper analysis shows that our ligand is not able
to interact with R262 (R316 in TRβ) and just one water-mediated contact is found with the
R228 side chain (R282 in TRβ). In this regards, crystallographic analysis and MD studies
[25, 26], have demonstrated that this single mutation is able to affect the orientation of one of
the interacting arginines (R228 in TRα and R282 in TRβ). Specifically, in TRβ, the N331
side chain interacts with the R282 guanidinium, which in turn is permanently and optimally
oriented toward the ligand carboxylate allowing GC-1 to stabilize a rich H-bonds network
with the three surrounding arginines, justifying its preferential binding to TRβ. As described

above, analogously to GC-1, thyromimetic compound 3 is able to establish a broad and stable
interaction pattern with TRβ arginines, including strong and direct contacts with R282, while
part of these interactions are lost in TRα. Reasonably, this might in part justify the
preferential binding of 3 to TRβ with respect to TRα. Also, it is known that in TRβ LBC, the
residues lining the outer ring of thyromimetics (especially the M442 residue) [27], are more
flexible with respect to their corresponding TRα residues, so that large 3'-substituents and
certain 4'-substituents [28], increment selectivity toward this isoform. Thus, it is possible that
3, possessing the novel combination of 3'-iPr e 4'-NH2 substituents on the outer ring, it is
better accommodated within the TRβ pocket.
Effects on lipid metabolism
Abnormal accumulation of lipids in hepatocytes is a prominent aspect in several liver
diseases such as NAFLD (non-alcoholic fatty liver spectrum disease) and alcoholic steato-
hepatitis (ASH) [29]. In order to investigate whether the new thyromimetic molecules 1 and 3
could promote lipolysis, and whether derivative 1 is capable of efficiently delivering the
corresponding parent compound 3, we evaluated their effect on lipid metabolism in the
human hepatoma cells HepG2. Initially HepG2 cells were incubated for 24 h with 1 and 3 at
1 µM and 10 ꢀM, with T3 as reference drug (Figure 6). Oil-red O staining was used to
monitor intracellular lipid accumulation (Figure 7). Cloroquine (CLO) 25 ꢀM was used as
positive controls.
150
100
*
**
*
**
#
***
**
***
50
0

Figure 6. Effects of 1 and 3 on total lipid accumulation in HepG2 cells. Cells were treated
for 24 h with 1, 3 or T3 at 1 and 10 ꢀM. 25 ꢀM Cloroquine (CLO) was used as positive
control. Oil red O stained intercellular oil droplets were eluted with isopropanol and
quantified by spectrophotometry analysis at 510 nm. Values represent the mean ± SEM of 4
to 8 experiments. The groups were compared using the One-Way ANOVA followed by
Tukey’s range test. *p<0.05 vs. Ctrl; **p<0.01 vs. Ctrl; ***p<0.005 vs. Ctrl; #p<0.05 vs. 3
1µM.
10
&
***
#
***
8
6
4
2
0
$
***
**
*
Figure 7. Thyromimetic-derivatives 1 and 3 induce lipolysis in HepG2 cells. Glycerol
released in the culture medium (0.5 ml) of HepG2 cells after 24 h treatment with 1, 3 and T3
at 1 and 10 ꢀM. 25 ꢀM Cloroquine (CLO) was used as positive control. Values represent the
mean ± SEM of 4 to 8 experiments. The groups were compared using the One-Way ANOVA
followed by Tukey’s range test. *p<0.05 vs. Ctrl; **p<0.01 vs. Ctrl; ***p<0.005 vs. Ctrl;
#p<0.005 vs. compound 1 1 ꢀM; $p<0.01 vs. compound 3 1 ꢀM; &p<0.01 vs. T3 1 ꢀM.
Oil Red O staining revealed that the newly designed compound 3 reduced total lipid
accumulation into lipid droplets with a potency comparable or even higher than equimolar
doses of T3 (Figure 6). This result also shows that compound 1 exerts the same effect of 3 in
reducing lipid accumulation, whereas the increase in glycerol released levels seems to be
more efficient when compared with 3. This result let us to speculate that thyromimetic
derivative 1 is a pro-drug which possesses improved drug-like properties when compared to 3
(zwitterionic molecule) and this contributes to increase the bioavailability of the molecule
inside the hepatocytes. Further pharmacokinetic investigations need to be carried out in the
near future to confirm our hypothesis.

Effects on AMPK/ACC pathway
Thyroid hormone mediates lipogenesis, fatty acid β-oxidation, cholesterol synthesis and the
reverse cholesterol transport pathway through different mechanisms such as (a) the induction
of genes transcription involved in metabolism as well as (b) the activation of alternative
cellular pathways implicated in the direct catabolism of fatty acids, such as autophagy [30,
31]. Hepatic autophagy regulates lipid metabolism through elimination of triglyceride
accumulation in liver and prevents the development of steatosis [30, 32]. The 5' AMP-
activated protein kinase (AMPK) is a serine/threonine protein kinase which plays a central
role in regulating cellular metabolism and energy balance in mammalian cells [33]. Once
activated, AMPK triggers several pathways such as glycolysis, fatty acid oxidation, and
lipolysis. The overall effect on lipid metabolism is stimulation of fatty acid oxidation (FAO)
and inhibition of cholesterol and triglycerides synthesis [34]. Since AMPK is a regulator of
autophagy [35, 36], we investigated AMPK as a possible target of thyromimetic analogues 1
and 3. In particular, we examined the protein levels of phosphorylated AMPK (p-AMPK),
and its target phosphorylated acetyl-CoA carboxylase (p-ACC), a rate-limiting enzyme in
cholesterol and fatty acid biosynthesis. As shown in Figure 8, both compounds, when tested
at the higher concentration (i.e. 10 µM), induced a significant enhancement of p-
AMPK/AMPK and p-ACC/ACC ratios. Consistently, the effect induced by thyromimetic
analogue 1 in the p-ACC/ACC ratio was more pronounced than that observed with the
thyroid hormone T3.
A
B
p-ACC
ACC
p-AMPK
AMPK
β-actin
β-actin
2.0
1.5
1.0
0.5
0.0
1.5
1.0
0.5
0.0
**
***
**
**
***
*
**
*
*

Figure 8. Derivatives 1 and 3 regulate metabolism in HepG2 cells via AMPK/ACC pathway.
(A) Representative immunoblotting images and quantitative analysis of AMPK
phosphorylation. (B) Representative immunoblotting images and quantitative analysis of
ACC phosphorylation. * p<0.05 vs. Ctrl; ** p<0.01 vs. Ctrl; *** p<0.005 vs. Ctrl.
Lack of hepatic toxicity and cardiac hypertrophy in thyromimetics-treated rats
Our in vitro results showed that compound 3 and its pro-drug 1 efficiently reduced lipid
accumulation, suggesting their potential therapeutic use in pathological conditions, such as
NAFLD. To assess their effect in vivo we treated F344 male rats for 3 days with daily
intragastric (IG) administrations of 1 or 3 (50, 75 and 100 ꢀg/100g b.w.). Two additional
groups treated with the vehicle or with T3 (20 ꢀg/100g b.w.) were included. As shown in
Figure 9A, despite the brief exposure, T3 caused a significant increase of heart weight, as
well as of the heart/body weight ratio. No such effect was found in the heart of animals
treated with 1 or 3 at all doses used (Figure 9A), showing that the preferential binding to -
and activation of - TRβ does not result in cardiac hypertrophy. As to the liver, while T3
treatment caused a significant reduction of the liver weight/body weight ratio, neither 1 nor 3
caused change in liver weight. Similar results were obtained when the two thyromimetics
were administered at the dose of 75 µg/100 g b.w. intraperitoneally for 3 days (Figure 9B).
Notably, no change in the serum levels of markers of liver injury, such as transaminases or
bilirubin was observed following administration of the two thyromimetics. Interestingly, the
two novel drugs did not exhibit any significant effect on serum levels of cholesterol and
glucose, whereas compound 1 induces a significant reduction of triglycerides when compared
to controls and to T3 (Table 2).

Figure 9. Effect of T3, 1 and 3 on liver/body weight ratio. F344 male rats were treated for 3
days with daily intragastric (IG) injections of 1 or 3 (50, 75 and 100 ꢀg/100 g b.w.) or T3 (20
ꢀg/100 g b.w.). A control group received the vehicle (DMSO 5% in corn oil). Additional two
groups received daily injections of either 1 or 3 at the dose of 75 ug/100 g b.w.
intraperitoneally for 3 days. **P<0.01.

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Treatment
AST/GOT (U/L) ALT/GPT (U/L)
Bilirubin (mg/dl) Glucose (mg/dl) Cholesterol (mg/dl) TGs (mg/dl)
Controls
153 ± 11
159 ± 11
141 ± 11
133 ± 8
108 ± 7
134 ± 6
119 ± 6
102 ± 5
134 ± 10
123 ± 10
46 ± 4
1.4 ± 0.2
1.4 ± 0.3
1.1 ±0.3
0.9 ± 0.1
1.1 ± 0.2
1.3 ± 0.3
1.6 ± 0.2
1.0 ± 0.2
1.5 ± 0.4
1.1 ± 0.2
151 ± 13.5
139 ± 4.4
129 ± 14
132 ± 7.6
164 ± 12
67.2 ± 2.5
65.2 ± 7.1
68.0 ± 5.2
71.4 ± 6.6
67.6 ± 8.1
69.5 ± 9.4
68.6 ± 5.6
70.2 ± 7.0
71.4 ± 3.1
69.6 ± 1.2
86.8 ± 10.4
90.2 ± 9.3
70.4 ± 7.6
67.0 ± 12.0
62.8 ± 6.1*
68.3 ± 9.1
74.0 ± 12.8
67.8 ± 12.5
74.2 ± 9.0
79.6 ± 9.7
T3
53 ± 4.1
46 ± 3.1
46 ± 1.7
37 ± 2.5
41 ± 2.1
43 ± 3.3
42 ± 4.6
38 ± 2.9
36 ± 1.7
1 i.g. (50 ug/100 g)
1 i.g. (75 ug/100 g)
1 i.g. (100 ug/100 g)
3 i.g. (50 ug/100 g)
3 i.g. (75 ug/100 g)
3 i.g. (100 ug/100 g)
1 i.p. (75 ug/100 g)
3 i.p. (75 ug/100 g)
136 ±
6
153 ± 18
166 ± 5.1
132 ± 14
129 ±
9
____________________________________________________________________________________________________________________
Values are expressed as means ± S.D. of 3 to 4 animals per group. *Significantly different from Controls for at least; p < 0.05.
Table 2. Effect of thyromimetics 1 and 3 and T3 on rat Serum Levels of AST/GOT,
ALT/GPT, Bilirubin, Cholesterol and Triglycerides (TGs) after the treatment with 1 or 3 (i.g.,
50, 75 and 100 ꢀg/100 g b.w; or i.p. 75 ꢀg/100 g b.w.) or T3 (20 ꢀg/100 g b.w.) for 3 days.
CONCLUSION
Compound 3 and its pro-drug 1 showed encouraging off-target and ADME-Tox profile that
prompted us to explore them as TRβ selective thyromimetics. The biochemical-based nuclear
receptor assay allowed us to consider compound 3 as a potential TRβ selective agonist able to
influence lipid metabolism and fat oxidation in hepatocytes. Indeed, our in vitro assays
showed that both compound 3 and its pro-drug (1) were able to reduce lipid accumulation in
HepG2 and promote lipolysis with comparable effects to those elicited by T3, used as
reference drug. These results were associated with a safe off-target and ADME-Tox profile.
Importantly, in vivo studies confirmed the apparent lack of toxicity of our thyromimetics.
Indeed, no liver injury, as assayed by serum levels of ALT, AST and bilirubin and no cardiac
hypertrophy, a critical side effect of T3, could be observed in thyromimetic-treated rats.
Interestingly, we found a significant decrease in blood triglyceride content following
administration of both TRβ agonists, suggesting their possible effect on lipid metabolism.
Future in vivo studies will be needed to investigate the effect of both compounds in

experimental models of NAFLD in order to validate them as new tools to treat hepatic
diseases and to confirm that 1 is an efficient pro-drug of TRβ selective agonist 3, with higher
affinity toward the liver.
EXPERIMENTAL SECTION
Chemistry
General Material and Methods. Melting points were determined on a Kofler hot-stage
apparatus and are uncorrected. Chemical shifts (δ) are reported in parts per million downfield
from tetramethylsilane and referenced from solvent references; coupling constants J are
1
reported in hertz. H-NMR, 13C-NMR and 19F-NMR spectra were obtained with a Bruker
TopSpin 3.2 400 MHz spectrometer and were recorded at 400, 101 and 376 MHz,
respectively. 19F and 13C NMR spectra are 1H decoupled. 19F NMR spectra are unreferenced,
corrected from Trifluoroacetic Acid (TFA) as external standard (-76.2 ppm). Signal spectra
were fully decoupled. The following abbreviations are used: singlet (s), doublet (d), triplet (t),
double−doublet (dd), and multiplet (m). The elemental compositions of the compounds
agreed to within ±0.4 % of the calculated values. Chromatographic separation was performed
on silica gel columns by flash (Kieselgel 40, 0.040−0.063 mm; Merck) or gravity column
(Kieselgel 60, 0.063−0.200 mm; Merck) chromatography. The ≥95 % purity of the tested
compounds was confirmed by combustion analysis. Reactions were followed by thin-layer
chromatography (TLC) on Merck aluminum silica gel (60 F254) sheets that were visualized
under a UV lamp. The microwave-assisted procedures were carried out with a Biotage
Initiator+ microwave. Evaporation was performed in vacuo (rotating evaporator). Sodium
sulfate was always used as the drying agent. Commercially available chemicals were
purchased from Sigma-Aldrich or Fluorochem.
Synthesis of 2-(4-(4-acetamido-3-isopropylbenzyl)-3,5-dimethylphenoxy)acetic acid (1).
A solution of phenolic derivative 19 (0.29 mmol) in DMF was treated with Cs2CO3 (473 mg;
1.45 mmol) and BrCH2COOH (0.29 mmol). The mixture obtained was stirred for 1 h at rt and
then, quenched with water. After extraction with DCM, aqueous phase was acidified with
HCl 10% and extracted again with DCM. Organic phases were then collected, dried, filtered
and concentrated to obtain a crude which was washed in hexane, providing final compound 1.
Yield: 82 % (white powder). 1H-NMR (CD3OD-d4): δ 1.12 (d, 6H, J = 6.8 Hz, CH3); 2.12 (s,

3H, CH3CO); 2.18 (s, 6H, CH3); 3.05−3.08 (m, 1H, CH); 3.98 (s, 2H, CH2); 4.35 (s, 2H,
13
CH2); 6.67 (s, 2H, Ar); 6.76 (dd, 1H, J = 2.0, 8.0 Hz, Ar); 7.00−7.08 (m, 2H, Ar) ppm. C-
NMR (CD3OD-d4): δ 180.44 (COOH); 172.91 (CH3CO); [158.17, 145.59, 140.71, 139.23,
132.96, 130.42, 128.55, 126.41, 126.32, 115.34] Ar; 68.49 (OCH2); 34.84 (CH2); 29.06 (CH);
24.18 (CHCH3); 23.70 (CH3); 22.81 (CH3CO;); 20.52 (CH3) ppm. Anal. Calcd for
C22H27NO4: C, 71.52 %; H, 7.37 %; N, 3.79 %. Found: C, 71.41 %; H, 7.42 %; N, 3.59 %.
Synthesis of 2-(4-(4-acetamido-3-isopropylbenzyl)phenoxy)acetic acid (2). The crude
product was obtained following the same synthetic procedure reported for compound 1. The
crude was purified by suspension in hexane. Yield: 64 % (grey powder). 1H NMR (CD3OD):
δ 1.64 (d, 6H, J = 6.8 Hz, CH3); 2.13 (s, 3H, CH3CO); 3.08 (m, 1H, J = 6.8 Hz, CH); 3.89 (s,
2H, CH2); 4.61 (s, 2H, CH2COOH); 6.83 (d, 2H, J = 8.4 Hz, Ar); 6.98 (dd, 1H, J = 2.0, 8.0
Hz, Ar); 7.08 (d, 1H, J = 8 Hz, Ar); 7.12 (d, 2H, J = 8.4 Hz, Ar); 7.17 (d, 1H, J = 2 Hz, Ar)
ppm. 13C NMR (CD3OD): δ 172.81 (COOH); 172.71 (COCH3); [157.74, 145.62, 142.06,
135.55, 133.10, 130.74, 128.54, 127.48, 127.16, 115.51] Ar; 65.88 (CH2COOH); 41.56
(CH2); 29.02 (CH); 23.59 (CH3); 22.67 (CH3CO) ppm. Anal. Calcd for C20H23NO4: C, 70.36
%; H, 6.79 %; N, 4.10 %. Found: C, 70.43 %; H, 6.69 %; N, 4.21 %.
Synthesis of 2-(4-(4-amino-3-isopropylbenzyl)-3,5-dimethylphenoxy)acetic acid (3).
Compound 1 (103 mg, 0.279 mmol) and HCl conc. (8 mL) were mixed with 2 mL of water
and refluxed at 120 °C for 4 h. Upon cooling a solid was formed, mainly constituted by
compound 3. Crude obtained was washed in HCl 10 % and hexane under vigorous stirring.
Yield: 52 % (grey powder). 1H-NMR (CD3OD-d4): δ 1.15 (d, J = 6.8 Hz, 6H, CH3); 2.19 (s,
6H, CH3); 2.92–2.99 (m, 1H, CH); 3.87 (s, 2H, CH2); 4.67 (s, 2H, OCH2); 6.52 (dd, 1H, J =
1.8, 8.0 Hz, Ar); 6.61 (d, 1H, J = 8.0 Hz, Ar); 6.62 (s, 2H, Ar); 6.79 (d, 1H, J = 1.8 Hz, Ar)
ppm. 13C NMR (CD3OD-d4): δ171.77, 157.28, 142.51, 139.50, 134.79, 132.24, 131.59,
126.44, 125.77, 117.72, 115.03, 66.02, 52.53, 34.52, 28.32, 23.05, 20.52 ppm. Anal. Calcd
for C20H25NO3: C, 73.37 %; H, 7.70 %; N, 4.28 %. Found: C, 73.52 %; H, 7.68 %; N, 4.19
%.
Synthesis of 2-(4-(4-amino-3-isopropylbenzyl)phenoxy)acetic acid (4). The crude product
was obtained following the same synthetic procedure reported for compound 3 and purified
by resuspension in hexane. Yield: 96 % (grey powder). 1H NMR (CD3OD-d4): δ 1.26 (d, 6H,
J = 6.8 Hz, CH3); 3.04 (m, 1H, J= 6.8 Hz, CH); 3.94 (s, 2H, CH2); 4.62 (s, 2H, CH2COOH);
6.86 (d, 2H, J = 8.4 Hz, Ar); 7.12 (d, 3H, J = 8.8 Hz, Ar); 7.25 (d, 1H, J = 2.0, 8.0 Hz, Ar);

7.34 (s, 1H, Ar); 7.91 (s, 1H, COOH) ppm. 13C NMR (CD3OD-d4): δ 171.44 (COOH);
(157.82, 144.57, 143.37, 135.01, 130.83, 128.63, 128.60, 124.20, 115.67) Ar; 66.03
(CH2COOH); 41.29 (CH2); 28.64 (CH); 23.79 (CH3) ppm. Anal. Calcd for C18H21NO3: C,
72.22 %; H, 7.07 %; N, 4.68 %. Found: C, 72.27 %; H, 7.12 %; N, 4.39 %.
General procedure for synthesis of compounds 5, 6 and 8. Under nitrogen atmosphere,
AlCl3 (1.35 mmol) was added to a solution of LiAlH4 (1.35 mmol) in THF. The mixture was
stirred at rt for 5 min; then, a solution of the proper cyanoderivative 22, 23 or 31 (0.15 mmol)
in THF was added dropwise, and the mixture refluxed for 12 h. After cooling at 0 °C, the
mixture was diluted with water, acidified with HCl 10 % and washed with Et2O. Aqueous
phase was alkalinized with NaOH 1N and extracted with chloroform; then, organic phase was
filtered through celite, washed with brine, dried over Na2SO4 and concentrated.
Synthesis of 4-(4-(2-aminoethoxy)-2,6-dimethylbenzyl)-N-ethyl-2-isopropylaniline (5).
The crude product was transformed in the corresponding hydrochloride salt. Yield: 15 %
1
(grey powder) H-NMR (CD3OD-d4): δ 1.23 (d, 6H, J = 6.8 Hz, CH3); 1.36 (t, 3H, J = 7.4
Hz, CH2); 2.20 (s, 6H, CH3); 3.02-3.06 (m, 1H, CH); 3.36-3.40 (m, 4H, CH2NH2, CH2); 4.07
(s, 2H, CH2); 4.22 (t, J = 4.8 Hz, 2H, OCH2); 6.77 (s, 2H, Ar); 6.97 (dd, 1H, J = 1.4,8.0 Hz,
Ar); 7.21 (d, 1H, J = 1.4 Hz, Ar); 7.27 (d, 1H, J = 8.0 Hz, Ar) ppm. 13C NMR (CD3OD-d4): δ
157.89, 144.36, 143.49, 139.71, 130.74, 130.50, 128.61, 127.99, 124.54, 115.19, 65.06,
40.42, 34.67, 28.66, 24.46, 20.43, 20.36, 11.36 ppm. Anal. Calcd for C22H33ClN2O: C, 70.10
%; H, 8.82 %; N, 7.43 %; Found: C, 70.24 %; H, 8.75 %; N, 7.56 %.
Synthesis of 4-(4-(2-aminoethoxy)-2-(trifluoromethyl)benzyl)-N-ethyl-2-isopropylaniline
(6). The crude was purified by transformation in hydrochloric salt and subsequent re-
crystallization in isopropanol/diisopropyl ether. Yield: 17 % (yellow oil). 1H-NMR (CD3OD-
d4): δ 1.27 (d, J = 6.4 Hz, 6H, CH3); 1.36 (t, J =7.4 Hz, 3H, CH3); 3.03−3.04 (m, 1H, CH);
3.38−3.41 (m, 4H, CH2NH2, CH2); 4.19 (s, 2H, CH2); 4.29 (t, 2H, J = 8.0 Hz, CH2O); 7.11
(dd, J =2.0, 8.4 Hz, 1H, Ar); 7.23 (dd, 2H, J = 2.4, 8.4 Hz, 1H, Ar); 7.28-7.37 (m, 4H, Ar)
ppm. 13C-NMR (CD3OD-d4): δ 158.10, 143.52, 135.03, 134.77, 132.28, 131.02, 130.72,
129.55, 128.94, 124.40, 124.19, 119.09, 113.96, 65.76, 40.20, 37.67, 28.74, 24.41, 17.28,
11.42 ppm. Anal. Calcd for C21H28ClF3N2O: C, 60.50 %; H, 6.77 %; N, 6.72 %; Found: C,
60.51 %; H, 6.72 %; N, 6.71 %.
Synthesis of 4-(4-(2-aminoethoxy)-2,6-dimethylbenzyl)phenol (7). Aniline derivative 34
(0.17 mmol, 48 mg) was solubilized in water and treated with H2SO4 (0.05 mL). The mixture

obtained was stirred for 20 min at 0 °C. Then, a solution of NaNO2 (0.17 mmol; 12 mg) in
H2O was added and the reaction refluxed for 1 h. After this time, reaction mixture was cooled
at RT and diluted with AcOEt. Organic phase was then washed with a saturated solution of
NaCl, dried, filtered and concentrated. The crude obtained was purified by precipitation in
1
MeOH/Et2O. Yield 42% (dark yellow powder). H-NMR (CD3OD-d6): δ 2.19 (s, 6H, CH3);
3.09 (t, J = 5.0 Hz; 2H, CH2NH2); 3.88 (s, 2H, CH2); 3.99 (t, J = 5.0 Hz; 2H, OCH2); 6.61 (s,
13
2H, Ar); 6.69 (d, J = 8.6, 2H, Ar); 6.83 (d, J = 8.6, 2H, Ar) ppm. C-NMR (CD3OD-d6): δ
158.24, 156.33, 139.34, 132.20, 131.26, 129.62, 116.09, 115.11, 69.76, 41.84, 34.21, 20.54
ppm. Anal. Calcd for C17H21NO2: C, 75.25 %; H, 7.80 %; N, 5.16 %; Found: C, 75.19 %; H,
7.66 %; N, 4.88 %.
Synthesis of 4-(4-(2-aminoethoxy)-2-(trifluoromethyl)benzyl)aniline (8). The crude
product was transformed in the corresponding hydrochloride salt. Yield: 10 % (yellow oil).
1H-NMR (CD3OD-d4): δ 3.39 (t, J=4.8 Hz, 2H, CH2NH2); 4.19 (s, 2H, CH2); 4.29 (t, J = 4.8
Hz, 2H, CH2); 7.23 (dd, J = 2.4, 8.4 Hz, 1H Ar); 7.27−7.35 (m, 6H Ar) ppm. 13C-NMR
(CD3OD-d4): δ 158.09, 143.25, 135.05, 132.29, 131.44, 130.89, 130.10, 126.97, 124.13,
119.10, 113.98, 65.78, 40.21, 37.53 ppm. Anal. Calcd for C16H17F3N2O: C, 61.93 %; H, 5.52
%; N, 9.03 %; Found: C, 62.07 %; H, 5.42 %; N, 9.26 %.
Synthesis of 2-(4-(4-aminobenzyl)-3-fluorophenoxy)acetic acid (9). A solution of
derivative 35 (28 mg, 0.109 mmol) in H2O was acidified with HCl 37 % and refluxed at 100
°C for 4 h. Then the mixture was concentrated affording the crude 9, which was purified by
1
formation of the corresponding hydrochloride salt. Yield: 77 % (grey powder). H NMR
(CD3OD-d4): δ 3.98 (s, 2H, CH2); 4.66 (s, 2H, CH2COOH); 6.69-6.74 (m, 2H, Ar); 7.15 (d,
13
1H, J=8.4 Hz, Ar); 7.31-7.37 (m, 4H, Ar) ppm. C NMR (CD3OD-d4): δ 172.21(COOH);
[161.28, 159.50, 143.21, 132.57, 131,21, 129.81, 124.10, 121.30, 111.53, 103.49] Ar; 66.03
(CH2COOH); 34.51 (CH2) ppm. Anal. Calcd for C15H14FNO3: C, 65.45 %; H, 5.13 %; N,
5.09 %; Found: C, 65.27 %; H, 5.18 %; N, 5.21 %.
Synthesis of 2-(4-(4-aminobenzyl)-3,5-dimethylphenoxy)acetic acid (10). Ester 36 (0.12
mmol; 39 mg) solubilized in MeOH was treated with an aqueous solution of NaOH 10%
(0.02 mL). Mixture was then refluxed for 1h and then, concentrated in vacuo. The crude
obtained was purified by precipitation in MeOH/Et2O providing compound 10. Yield: 42 %
1
(pale yellow). H-NMR (CD3OD-d4): δ 2.17 (s, 6H, CH3); 3.85 (s, 2H, CH2); 4.34 (s, 2H,
CH2COOH); 6.61 (d, 2H, J = 8.4 Hz, Ar); 6.65 (s, 2H, Ar); 6.72 (d, 2H, J = 8.4 Hz, Ar) ppm.

13C-NMR (CD3OD-d4): δ 177.01, 157.94, 146.00, 139.15, 131.39, 131.25, 129.36, 117.00,
115.25, 68.47, 34.27, 20.54 ppm. Anal. Calcd for C17H19NO3: C, 71.56 %; H, 6.71 %; N, 4.91
%; Found: C, 71.38 %; H, 6.92 %; N, 5.07 %.
Synthesis of 2-(4-(4-hydroxybenzyl)-3,5-dimethylphenoxy)acetic acid (11). Aniline 10
(0.20 mmol; 64 mg) was dissolved in H2O, added to a solution of H2SO4 at concentrated
grade (0.05 mL), and left under stirring at 0 °C for 20 min. Then, a solution of NaNO2 (0.20
mmol; 14 mg) in H2O was added, and the mixture was refluxed for 1 h. After this time, the
reaction was allowed to cold down to rt and then diluted with AcOEt. Organic phase was
washed with a saturated solution of NaCl, then dried, filtered and evaporated. The crude
1
product was purified by precipitation in AcOEt/Hexane. Yield: 35 % (brown powder). H-
NMR (CD3OD-d4): δ 2.17 (s, 6H, CH3); 3.87 (s, 2H, CH2); 4.59 (s, 2H, CH2COOH);
13
6.60−6.67 (m, 4H, Ar); 6.77 (d, 2H, J = 8.4 Hz, Ar) ppm. C-NMR (CD3OD-d4): δ 173.18,
157.42, 156.32, 139.45, 132.10, 131.86, 129.64, 116.09, 115.12, 65.94, 34.21, 20.53 ppm.
Anal. Calcd for C17H18O4: C, 71.31 %; H, 6.34 %. Found: C, 71.46 %; H, 6.47 %.
X-ray selection. To date, many X-ray structures of TRα and TRβ in complex with T3, GC-1,
KB131084 and other thyromimetics, are available in the Protein Data Bank (PDB). A
superposition on the α-carbon atoms of all these structures, showed that the general protein
folding is highly preserved, with a good conservation of both the secondary structure and the
side chain orientation, especially within the LBC. Thus, among the crystals with highest
resolution, those co-crystallized with our ligands progenitor GC-1 were chosen for our
docking calculations: PDB codes 3ILZ and 3IMY, for TRα and TRβ, respectively.
Docking calculations The ligands tridimensional structure were generated with the Maestro
Build Panel and prepared using Ligprep predicting all tautomeric and protonation states at
physiological pH (7.4 ± 1.5). The receptor structures were prepared using the Protein
Preparation Wizard implemented in Maestro Suite (Schrödinger Release 2019-2: Schrödinger
Suite 2019-1), deleting all water molecules with the exception of those within 3 Å from the
ligand, adding missing hydrogen atoms and minimizing the complexes. The docking grid was
calculated through the grid generation tool in Glide 6.7 [37-39] and centered around the
crystallized ligand using default settings. The OPLS3E force field was employed for docking.
The results of calculations were evaluated and ranked based on the Glide SP scoring function.
In vitro studies

Activity against thyroid hormone receptors. Assays on selected compounds were
performed by SelectScreen™ Biochemical Nuclear Receptor Profiling Service (Thermofisher
Scientific, US) using the LanthaScreenTM TR-FRET Nuclear Receptor Coregulator Assay
(service provided by Invitrogen Corporation, USA), as reported on the company’s website.
(https://www.thermofisher.com/it/en/home/products-and-services/services/custom-
services/screening-and-profiling-services/selectscreen-profiling-service/selectscreen-cell-
based-nuclear-receptor-profiling-services.html).
Human hepatocellular carcinoma (HepG2) cell culture and treatment with compounds
1 and 3. The human hepatocellular carcinoma HepG2 cells line was purchased from the
American Type Culture Collection (ATCC HB-8065, Rockville, MD, USA) and cultured in
low glucose (LG) DMEM supplemented with 10 % FBS, 100 IU/mL penicillin G sodium and
100 ꢀg/mL streptomycin sulfate (Invitrogen, Paisley, UK) incubating in a humidified
atmosphere with 5 % CO2 at 37 °C. These cells were then treated with 1 and 10 µM
concentrations of test compounds for 24 h. After treatments, cells were lysed in a buffer
containing 20 mM Tris–HCl (pH 7.5), 0.9 % NaCl, 0.2 % Triton X-100, and 1 % of the
protease inhibitor cocktail (Sigma-Aldrich, Milan, Italy) and then stored at −80 °C for further
western blot analysis. All the analyses were conducted on cells between the third and the
sixth passage.
Oil red O (ORO) staining of lipid accumulation in HepG2 cells. Total lipid accumulation
was evaluated according to the method previously described by Liu et al. [40] Cells were
seeded at a density of 3.5 × 104 cells/well in 1 mL of lipogenic growth medium (HG-
DMEM) and treated for 24 h with compound 1 or 3, at 1 and 10 µM. Cloroquine (25 µM)
was used as positive control. Subsequently, after collecting the growth media to be used to
perform glycerol level measurements (as detailed below), cells were rinsed twice with PBS
and fixed in 4 % paraformaldehyde in PBS at 4 °C for 30 min. After 3 washes with cold PBS,
cells were stained with ORO working solution, prepared as described above, for 30 min at
room temperature and subsequently rinsed again with PBS.
Determination of glycerol release from HepG2 cells. After treatment with test compounds,
cell culture supernatants were collected from each well and placed in glycerol-free
containers. A 125µg/mL Glycerol Standard Solution (Abcam, Milan, Italy) was used to make

a #1 through #8 standard curve. 25 µL of each supernatant and standard were then transferred
into a 96-well plate and 100µL of either Free Glycerol Assay Reagent (Abcam, Milan, Italy)
or MilliQ water added to each well. After incubation at rt for 15 min, glycerol levels were
measured by reading absorbance at 540 nm (Bio-Rad 680, Milan, Italy).
Western Blotting Analysis. Proteins (20–30 ꢀg) were separated on CriterionTGX TM gel
(4-20 %) and transferred on Immuno-PVDF membrane (Biorad, Milan, Italy) for 30 min. The
membranes were then incubated overnight at 4 °C with one of the following the specific
primary antibodies: anti-p-AMPKα1-thr172, anti-AMPK, anti-p-ACC-ser79, anti-ACC (Cell
Signaling Technology, Danvers, MA, USA). Then, blots were washed 3 times for 10 min
with 1X TBS, 0.1 % Tween®20 and incubated for 2 h with secondary antibody (peroxidase-
coupled anti rabbit in 1X TBS, 0.1 % Tween®20). After washing 3 times for 10 min, the
reactive signals were revealed by an enhanced ECL Western Blotting analysis system
(Amersham, Milan, Italy). Band densitometric analysis was performed using Image Lab
Software (Biorad, Milan, Italy).
In vivo Studies. Five-week-old male F-344 rats purchased from Charles River (Milano, Italy)
were maintained on a standard laboratory diet (Ditta Mucedola, Milano, Italy). The animals
were given food and water ad libitum with a 12 h light/dark daily cycle and were acclimated
for 1 week before the start of the experiment. All procedures were performed in accordance
with the Guidelines for the Care and Use of Laboratory Animals and were approved by the
Italian Ministry of Health. Animals were treated with daily intragastric (IG) injections for 3
days of 1 or 3 (50, 75 and 100 µg/100 g b.w) or T3 (20 ꢀg/100 g b. wt). A control group
received the vehicle (DMSO 5 % in corn oil). Additional two groups received daily injections
of either 1 or 3 at the dose of 75 µg/100 g b. wt. intraperitoneally for 3 days
Analysis of Serum Aspartate aminotransferase (AST), Alanine aminotransferase (ALT),
Triglycerides (TGs) and Cholesterol (CH). Immediately after sacrifice, blood samples were
collected from the abdominal aorta, serum was separated by centrifugation (2000 g for 20
minutes) and tested for triglycerides, cholesterol, aspartate aminotransferase and alanine
aminotransferase using a commercially available kit from Boehringer (Mannheim, Germany).