Introduction

The worldwide prevalence of obesity has doubled from 1980 to 2015, with nearly 10% of the world’s population now classified as obese or overweight1. Obesity adversely affects the physiological functions of the body and increases the risk of developing multiple diseases, including diabetes2, cardiovascular disease2,3, and cancer4. Statins and Ezetimibe, which inhibit the biosynthesis of cholesterol in the liver and absorption of cholesterol from the small intestine, respectively, have been used as therapeutic agents for obesity but are not effective in all patients5.

In recent years, thyroid hormones (TH) have gained attention as a new treatment for obesity6,7. The effects of thyroid hormones are mediated by their receptors (THRs), which belong to the nuclear receptor superfamily. Thyroxine (T4), a prohormone produced in the thyroid gland, is converted to biologically active triiodothyronine (T3) by iodothyronine deiodinases 1 and 2 (DIO1 and DIO2)8. THR forms a complex with the retinoid X receptor (RXR) and a corepressor, such as NcoR or SMRT, on the thyroid hormone-responsive elements (TRE) of target genes. The binding of T3 to THR facilitates the dissociation of the corepressor and the recruitment of a coactivator, such as steroid receptor coactivator-1 (SRC-1), leading to the transcriptional activation of target genes9. There are two subtypes of THRs, α, and β; the α subtype (THRα) is highly expressed in the brain, heart, and muscle, while the β subtype (THRβ) is expressed mostly in the liver and in the pituitary10.

T3 has the ability to activate the metabolism of lipids in the liver and adipose tissues via THRβ, while an excess amount of T3 can lead to tachycardia, as well as, bone loss and muscle wasting through THRα11. The biological functions induced by THRβ binding are, therefore, beneficial to address abnormal lipid metabolism and obesity, but T3 shows almost no selectivity between receptors, making it difficult to avoid severe side effects derived from binding to THRα12.

To overcome this issue, considerable efforts have been devoted to the development of THRβ-selective TH analogs by derivatizing T313,14. GC-1, developed by Scanlan et al., is the most representative of THRβ-selective TH analogs, showing a 10-fold lower affinity for THRα than T3 without significant loss of affinity to THRβ15. Although the effects on lipid metabolism of GC-1 have been demonstrated in numerous in vitro and animal models, clinical trials were terminated after phase 113. GC-24 is another THRβ selective agonist showing ten-fold greater selectivity for THRβ against THRα compared with GC-1. These two molecules have high structural similarity, differing only in the presence of an isopropyl group at the 3’ position in the left part of GC-1 compared to a benzyl group in the same position for GC-2416.

This study uses cross-coupling chemistry17 to develop more potent, more selective, and less toxic GC-1 derivatives based on the structure-receptor selectivity relationship between GC-1 and GC-24. Compound ZTA-261 has the highest affinity for THRβ after screening multiple compounds with a highly congested diarylmethane structure. This study explores the potential of ZTA-261 as a THRβ-selective synthetic thyroid hormone mimic. THRβ-selectivity is assessed using a competitive ligand binding assay using full-length THRs. ZTA-261 is administered to high-fat diet-induced obesity model mice to evaluate its effect on lipid metabolism by quantifying body weight, epididymal adipose weight, and levels of total cholesterol and triglycerides in the serum and liver. The expression of lipid metabolism-related genes in the liver is analyzed to gain insight into the mechanism of action of ZTA-261. The toxicity of ZTA-261 to the liver, heart, and bone is evaluated by measuring serum alanine aminotransferase (ALT) levels and by histological analyses of the heart and bone.

Methods

Construction of expression vectors of THRα and THRβ

The coding sequences of full-length human THRα and THRβ were amplified by PCR. SC307938 (OriGene) and pF1KB3732 (Kazusa DNA Res. Inst.) were used as templates. The primer sequences used were as follows:

THRα sense: 5’-CGGAATTCATGGAACAGAAGCCAAGCAAGG-3’

THRα antisense: 5’-TGCGGTACCTTAGACTTCCTGATCCTC-3’

THRβ sense: 5’-CGGAATTCATGACTCCCAACAGTATGACAG-3’

THRβ antisense: 5’-GGCTCTAGATTAATCCTCGAACACTTCCAAGAAC-3’

The PCR products of THRα and β were cloned into EcoRI–KpnI and EcoRI–XbaI sites of pTNT (Promega), respectively. The resultant expression vectors, pTNT-THRα and pTNT-THRβ, were used for the synthesis of THRs by in vitro translation.

In vitro translation of full-length THRs

THRα and THRβ were synthesized by in vitro translation using TNT T7 Quick Coupled Transcription and Translation kit (Promega) according to the manufacturer’s protocol. After incubation at 30 °C for 90 min, the reaction mixtures were stored at −80 °C until use.

3,5,3’-[125I]-Triiodothyronine ([125I]-T3)-displacement assay

The [125I]-T3-displacement assay was performed as described by Chapo et al.18 with slight modifications. 2 μL of in vitro translation mixture of either THRα or THRβ were mixed with 0.5 nM [125I]-T3 (7.4 MBq/mL, 81.4 TBq/mmol, PerkinElmer) and serially diluted non-radioactive T3 or TH analogs from 10−5 to 10−11 M in 100 μL of E400 buffer (20 mM potassium phosphate buffer, pH 8.0, 400 mM NaCl, 0.5 mM EDTA, 1 mM MgCl2, 1 mM monothioglycerol 500 μg/mL calf thymus histone). The reaction mixture was incubated overnight at 4 °C. A nitrocellulose membrane (Optitran, 0.45 μm in pore size, Whatman) pre-blocked by E-400 buffer was placed on a dot-blot apparatus (Minifold I, GE Healthcare), and the reaction mixtures were filtered through the membrane followed by three washes with 200 μL E-400 buffer. The membranes were removed from the manifold, allowed to air-dry for 5 min, wrapped in plastic film, and exposed to an imaging plate (BAS IP MS 2040 E, FUJI FILM) for 4 h. Radioactive signals were scanned using Typhoon FLA9000 (GE Healthcare) and quantified by Image Quant TL software (GE Healthcare). The IC50 values for each competitor were calculated by fitting the dose–response data to the log (inhibitor) vs. response (three parameters) model using GraphPad Prism (version 8.0; GraphPad Software).

Animals and treatment

Male 7-week-old C57BL/6J mice were obtained from a local supplier (Nihon SLC, Shizuoka, Japan). The mice were individually housed in cages (Innocage; ORIENTAL GIKEN) at 23 °C with a 12-h-light–dark cycle starting at 08:00 a.m. After one week of acclimation, the animals were randomly divided into each experimental group and fed either a normal diet (ND) (10 kcal% fat) (D12450B; RESEARCH DIETS) or a high-fat diet (HFD) (60 kcal% fat) (D12492; RESEARCH DIETS). After 8 weeks of rearing on an ND or HFD, the animals were intraperitoneally injected with either saline, T3 at 0.1 or 1 µmol/kg day (65 or 650 μg/kg day), GC-1 at 0.1 or 1 µmol/kg day (33 or 330 μg/kg day), or ZTA-261 at 0.1 or 1 µmol/kg day (43 or 430 μg/kg day) for three weeks. The injection was prepared by dissolving the compounds in 0.5 N NaOH at a concentration of 100 mM, which was then further diluted with saline to a concentration of either 0.05 or 0.5 mM. The volume of the solution injected was determined based on the body weight of the animal, measured once a week, with 0.05 and 0.5 mM solutions injected at 0.1 and 1 µmol/kg day, respectively. Each group consisted of 10 animals, except for the ND (n = 8) and HFD with saline (vehicle, n = 9) groups. In the ND-treated group, one animal died during the experimental period, and one animal was excluded from the analysis after finding tumor-like lesions in the liver. In the HFD-treated group, one animal was excluded from the analysis because its body weight was less than two standard deviations below the mean.

The injection was given 6 h after light on. Food consumption and body weights were measured weekly. Food and water were provided to the animals ad libitum. The animals were treated according to the guidelines of Nagoya University, and all experimental procedures were approved by the Animal Experiment Committee of Nagoya University (A210715-003).

Tissue collection

After three weeks of daily injections, the body weights of the mice were measured. Whole blood was collected through an abdominal aorta puncture under isoflurane anesthesia. Subsequently, the animals were euthanized using deep isoflurane anesthesia, and the liver, heart, and epididymal adipose tissues were removed from the body. Liver tissues for lipid analysis were weighed and stored at −80 °C, and for the histological experiment were frozen on dry ice embedded in Surgipath FSC 22 (Leica Biosystems) and stored at −80 °C. Hearts were weighed and fixed in 4% paraformaldehyde for 48 h in a cassette (Sakura Finetek Japan, Tokyo, Japan). Femurs were isolated, wrapped in KimWipes (Kimberly Clark), soaked in saline, and stored at −80 °C. Epididymal adipose tissue was weighed.

Hepatic lipid analysis

Frozen livers were homogenized with chloroform-methanol (2:1), and the chloroform layer into which the liver lipids were extracted was collected. This extract was used to measure hepatic total lipid, triglyceride (TG), and total cholesterol (TC) contents. The TG and TC levels were measured using LabAssay Cholesterol and LabAssay Triglyceride (FUJIFILM Wako Pure Chemicals), respectively, according to the manufacturer’s protocol. The total liver lipid content was measured according to the method described by Forch et al.19.

Oil Red O staining of the liver section

20 µm sections were prepared from frozen liver tissues using Cryostat (CM3050 S, Leica). Sections were affixed to silane-coated glass slides, dried at room temperature, and stored at −80 °C until use. Before staining, sections were fixed with 1% PFA/1 × PBS, pH 7.4 (Gibco) for 5 min, immersed in 1 × PBS for 5 min, distilled water for 5 min, and 60% 2-propanol solution for 1 min. Oil Red O staining solution was prepared by mixing 0.36 g of Oil Red O (Sigma) with 120 mL of 2-propanol and 80 mL of distilled water, allowing the solution to stand at room temperature for 1 h, then filtering the solution and leaving it overnight at 37 °C. Sections were immersed in the Oil Red O staining solution at 37 °C for 15 min, rinsed in 60% 2-propanol for 1 min, and in distilled water for 1 min. Sections were counterstained with Meyer’s hematoxylin. Sections were observed under a microscope (BX43; Olympus) at ×100 magnification, and digital images were captured using CellSens software (Olympus). The area of red-stained fat droplets was quantified using Image J (National Institutes of Health), and the percentage of the stained area in the total image area was calculated. Nine different parts in three consecutive sections per mouse were analyzed.

Gene expression analysis by real-time quantitative PCR (RT-qPCR)

Total RNA was extracted from frozen liver tissue using the QIAzol reagent (Qiagen) according to the manufacturer’s protocol. Extracted RNA was stored at −80 °C until use. One microgram of total RNA was subjected to cDNA synthesis using the ReverTra Ace qPCR RT Kit (TOYOBO). qPCR reaction was performed using QuantStudio 3 (Applied Biosystems) in 20 µL of reaction mixture containing 2 µL of cDNA, 10 µL of TB Green Premix Ex Taq™ II (TAKARA), 0.4 µL of ROX DyeII (TAKARA) and 0.4 µM each of gene-specific primers. Rps18 was selected as an internal control gene from the Mouse Housekeeping Gene primer set (TAKARA) using RefFinder software20,21 (Supplementary Fig. 1). Relative expression was calculated using Pfaffl’s method22. The sequences of the PCR primers used were as follows:

Cyp7a1 Forward: 5’-AGCAACTAAACAACCTGCCAGTACTA-3’

Cyp7a1 Reverse: 5’-GTCCGGATATTCAAGGATGCA-3’

Srebp1-c Forward: 5’-GGAGCCATGGATTGCACATT-3’

Srebp1-c Reverse: 5’-AGGAAGGCTTCCAGAGAGGA-3’

Ldlr Forward: 5’-TGACGGGCTGGCGGTAGACTG-3’

Ldlr Reverse: 5’-AGTGTGATGCCATTTGGCCAC-3’

Pnpla2 Forward: 5’-CAACCTTCGCAATCTCTAC-3’

Pnpla2 Reverse: 5’-TTCAGTAGGCCATTCCTC-3’

Dio1 Forward: 5’-CCCCTGGTGTTGAACTTTG A-3’

Dio1 Reverse: 5’-CTGTGGCGTGAGCTTCTTC-3’

Me1 Forward: 5’-AGTATCCATGACAAAGGGCAC-3’

Me1 Reverse: 5’-ATCCCATTACAGCCAAGGTC-3’

Thrsp Forward: 5’-AAACCAGCGAGGCTGAGAACGA-3’

Thrsp Reverse: 5’-CAGGTGGGTAAGGATGTGATGG-3’

Serum lipid and ALT analysis

The blood samples were left to stand for 1 h at room temperature, then centrifuged at 4 °C for 15 min at 800 × g, and the supernatant was collected. Serum TC and TG levels were measured using LabAssay Cholesterol and LabAssay Triglyceride, respectively (FUJIFILM Wako Pure Chemical Industries), according to the manufacturer’s protocol. Serum ALT levels were measured using Alanine Aminotransferase (ALT or SGPT) Activity Colorimetric/Fluorometric Assay Kit (BioVision) according to the manufacturer’s protocol.

Hematoxylin–eosin staining of the heart section

Heart tissues were fixed, dehydrated, and permeated with paraffin using an automatic fixation and encapsulation device (Sakura Finetech, Japan). Paraffin blocks were prepared using Tissue Tech TEC 6 (Sakura Finetech, Japan), and 5 µm-thick transverse transmural sections were prepared with a microtome (Leica). The sections were deparaffinized, rehydrated, and immersed in Meyer’s hematoxylin solution (FUJIFILM Wako Pure Chemical Industries), followed by eosin staining. Sections were dehydrated and mounted using Entellan New (Merck). Sections were observed under a microscope (BX43; Olympus) at × 400 magnification and digital images were captured using cellSens (Olympus). 100 cells were randomly selected from the left ventricle of an individual mouse, and their trans-nuclear widths were measured according to Oláh et al.23 using Image J (National Institute of Health).

X-ray micro-computed tomography (μCT) of isolated mouse femur

μCT scanning of the mouse distal femur was performed using TDM 1000 (Yamato Scientific) with the following conditions: X-ray tube potential, 60 kV; tube current, 60 μA; voxel size, 9.2 μm; integration time, 125 ms; projections, 1200. A brass filter of 0.1 mm thickness was placed in the beam path to reduce the beam hardening effect. Image analysis was performed using TRI/3D-BON-FCS64 (RATOC System Engineering). Trabecular bone parameters, including bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), trabecular number (Tb.N), and connectivity density (Conn. D) were calculated from the region 0.3–2.3 mm proximal to the epiphyseal growth plate. Five samples were randomly selected for analysis.

Measurement of serum T3 concentration

To eliminate proteins from the serum samples, 300 μL of acetonitrile containing 1 μg/mL of gabapentin as an internal standard was added to the serum. The mixture was then placed on ice for 15 min and centrifuged for 15 min at 20,000 × g. MonoSpin C18 cartridges (GL Science, Tokyo, Japan) were pre-conditioned with 300 μL of acetonitrile, followed by 300 μL of purified water, loaded with 200 μL of supernatants, and then washed with 300 μL of purified water. Elution was performed with 100 μL of acetonitrile. The eluate was dried, reconstituted with 30 μL of purified water, and injected into the LC–MS system.

A Cadenza CD-C18 column (3 × 150 mm, 3 μm; Imtakt, Kyoto, Japan) was used for chromatographic separation. All the experiments were performed using an Ultimate 3000 UHPLC system connected to a Q Exactive Plus Orbitrap mass spectrometer (Thermo Fisher Scientific). The injection volume for all the samples was 10 μL. The mobile phase used was 100% purified water (A) and 100% acetonitrile (B). The following gradient elution program was used: 0–2 min (5% B), 2–6 min (5–95% B), 6–8 min (95% B), 8–8.1 min (95-5% B), and 8.1–12 min (5% B) at a flow rate of 0.4 mL/min. The temperature of the autosampler and the column were maintained at 10 and 37 °C, respectively.

The MS parameters were set as follows: sheath gas flow rate, 40 psi; Aux gas flow rate, 10; spray voltage, 4 kV; capillary temperature, 350 °C; resolution, 70,000; AGC target, 3 × 106; maximum IT, 200 ms. Detection was performed in positive ion mode for gabapentin (m/z = 172.1322) and T3 (m/z = 651.80). Data were acquired and processed by the Xcalibur data system (Version 4.4, Thermo Fisher Scientific). The peak areas of the analytes were normalized to those of the internal standard.

Statistics

Data in Fig. 2 are expressed as mean ± standard deviation (SD), and subjected to non-linear regression analysis. Data in Figs. 3, 4, 7, and 8 are expressed as mean ± standard error (SEM). Data in Figs. 3, 4, 7, and 8 were evaluated using one-way ANOVA followed by Tukey’s post-hoc test. Data in Figs. 5b and 6 are expressed as median, maximum, minimum, and interquartile values and were evaluated using the Kruskal–Wallis test, followed by Dunn’s multiple comparisons test. All statistical tests were performed using GraphPad Prism (version 7.0, version 8.0, or version 10.0; GraphPad Software).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Results

Design and synthesis of THRβ-selective TH analogs

We began our investigation by designing a new THRβ-selective TH analog based on key structural features of T3, GC-1, and GC-24. The hydrophobic isopropyl and benzyl groups at the 3’ position of the phenol in GC-1 and GC-24 have been shown to be critical for high THRβ-selectivity (Fig. 1a)15,16. Notably, X-ray crystallographic analysis of human THRβ with TH analogs has shown that these substituents are located in the hydrophobic region of the hormone-binding pocket, resulting in improved fit and increased affinity for THRβ16. The 3,5-dimethyl substituents mimic iodide substituents in the natural hormone from a space-filling point of view and, thus, are important structural elements that fix the two aromatic rings in a perpendicular arrangement. This places the molecule in the active conformation seen for effective TH analogs24,25. Therefore, we introduced a naphthol group to enhance the hydrophobic interaction with THRβ and increase structural rigidity. We also employed trifluoromethyl groups26 to enhance the rigid conformation while mimicking the electronic properties of natural iodide substituents. New thyroid hormone analogs were synthesized, three of which are shown in Fig. 1b. The synthetic route to ZTA-245 and ZTA-261 is shown in Scheme (Fig. 1c). Methyl 4-bromo-2,6-bis(trifluoromethyl)benzoate 1, which was prepared from 1,3-bis(trifluoromethyl)benzene27, was subjected to Cu-catalyzed cross-coupling with benzyl alcohol to afford the ester 2. The reduction of 2 followed by chlorination gave benzyl chloride 3 as an activated coupling partner. In the presence of a simple Pd catalyst, Suzuki–Miyaura cross-coupling of 3 with (4-methoxynaphthalen-1-yl) boronic acid proceeded smoothly, giving the corresponding diarylmethane 4. After removal of the benzyl group, the resulting phenol 5 was alkylated with t-butyl bromoacetate to afford ester 6. Basic hydrolysis and demethylation provided ZTA-245. For the synthesis of ZTA-261, common intermediate 5 was converted into the triflate 7, which underwent Pd-catalyzed Migita–Kosugi–Stille cross-coupling with allyltributylstannane to form alkene derivative 8. Finally, oxidative cleavage of the alkene unit followed by demethylation furnished ZTA-261.

Fig. 1: Design and synthesis of THRβ-selective TH analogs.
figure 1

To enhance the hydrophobic interaction with THRβ, the 3’-substituted phenol unit was replaced with naphthol. Ortho methyl groups were replaced with bulky trifluoromethyl groups, which can provide a more rigid conformation that will be effective for binding with THRs and are more electronically similar to the iodide groups in natural T3 (a). The representative analogs (b), and synthetic scheme for ZTA-245 and 261 (c) were shown. Bn benzyl, DIBAL-H diisobutylaluminium hydride, DCM dichloromethane, TFA trifluoroacetic acid, DMF N,N-dimethylformamide, Tf trifluoromethanesulfonyl, EtOAc ethyl acetate.

The affinity and selectivity of these new hormone derivatives for THRα and β were tentatively evaluated by a cell-based assay. In this assay, the binding of TH analogs to the ligand binding domain of either THRα or β is converted to luminescence via a luciferase reporter (Supplementary Fig. 2). From these studies, ZTA-261 became the focus as it was found to have the highest affinity for THRβ among the compounds studied (Supplementary Table 1).

THRβ selectivity assessment of TH analogs using full-length THR

The THRβ-selectivity of T3, GC-1, and ZTA-261 were assessed using a radioligand-displacement assay. Full-length human THRα and THRβ were synthesized by in vitro transcription-translation and were incubated with a fixed concentration of [125I] T3 in the presence of non-labeled TH or TH-analogs at the concentrations indicated in Fig. 2, followed by filtration with a nitrocellulose membrane to separate the THR-bound and THR-free compounds.

Fig. 2: [125I]-T3-displacement assay.
figure 2

In vitro translation mixture of either THRα or THRβ was incubated with 0.5 nM of [125I]-T3 and serially diluted non-radioactive T3 (a), GC-1 (b), or ZTA-261 (c). [125I]-T3 bound to THRs was captured on a nitrocellulose membrane, and signals were quantified and plotted; the signals of [125I]-T3 bound to THRs in the absence of non-radioactive compounds were set to 100%. Data are shown as mean ± SD (n = 3). Data were fitted to the log (inhibitor) vs. response (three parameters) model using GraphPad Prism (version 8.0; GraphPad Software).

Using this assay, T3, GC-1, and ZTA-261 were shown to bind with similar affinity to THRβ, with IC50 values of 3.6, 3.4, and 6.3 nM, respectively (Fig. 2 and Table 1). The key difference between these compounds is seen in their binding to THRα. While T3 binds to THRα with approximately the same affinity as it binds THRβ (IC50 of 3.4 nM), GC-1 and ZTA-261 show a much lower affinity for THRα (IC50 of 73 and 660 nM, respectively). Thus, the affinities of GC-1 and ZTA-261 to THRα are approximately 20- and 100-fold lower than that to THRβ, respectively (Fig. 2 and Table 1). These results suggest that ZTA-261 is a highly THRβ-selective analog.

Table 1 IC50 values of T3, GC-1, and ZTA-261

Effects of TH analogs on body weight and accumulation of visceral fat

To assess the effect of TH analogs in vivo, T3, GC-1, and ZTA-261, either at low (0.1 µmol/kg day) or high (1 µmol/kg day) doses, were injected intraperitoneally into the high-fat diet (HFD)-induced obesity model mice according to the previous report28. During the experimental period, food consumption did not significantly differ between the vehicle and experimental groups (Supplementary Fig. 3).

At sacrifice, the body weight of HFD-fed mice injected with the vehicle was ~1.5 times higher than that in ND-treated mice (Fig. 3a). Mice injected with GC-1 at both low and high doses and high doses of T3 and ZTA-261 had decreased body weight by ~20% compared to those treated with vehicle (Fig. 3a). Treatment with low doses of T3 and ZTA-261 did not significantly reduce body weight (Fig. 3a), suggesting that ZTA-261 can reduce body weight in mice to the same extent as the natural thyroid hormone T3, but less effectively than GC-1. The weight of the epididymal adipose tissue decreased significantly following T3, GC-1, and a high dose of ZTA-261 treatment, whereas a low dose of ZTA-261 did not show any obvious effect (Fig. 3b).

Fig. 3: Effects of intraperitoneal injection of TH analogs on body weight and deposition of visceral fat.
figure 3

Male 8-weeks-old C57BL/6J mice were fed either a normal diet (ND; 10 kcal% fat) or a high-fat diet (HFD; 60 kcal% fat) and maintained for 8 weeks. Then, animals were intraperitoneally injected with either saline (Vehicle), T3 (0.1 or 1 µmol/kg day), GC-1 (0.1 or 1 µmol/kg day), or ZTA-261 (0.1 or 1 µmol/kg day). After three weeks of injection, body weight (a), and weight of the epididymal adipose tissue (b) were measured. Data are shown as the mean ± SEM (n = 8–10). Significant differences among groups are indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, one-way ANOVA with Tukey’s post-hoc test).

Effects of TH analogs on lipid metabolism

To investigate the effects of the TH analogs on lipid metabolism, we measured the levels of total cholesterol and triglycerides in plasma and liver, and total lipids in liver. Serum cholesterol levels decreased in all experimental groups compared to the vehicle control group (Fig. 4a). Serum triglyceride levels were also decreased by injection of GC-1 at a low dose and ZTA-261 at a high dose, whereas there was no significant difference between the T3- and vehicle-injected groups (Fig. 4b). Although liver cholesterol levels showed no significant differences in any of the experimental groups (Fig. 4c), the levels of total lipids and triglycerides in the liver were significantly decreased by GC-1, ZTA-261, and high-dose T3 treatment (Fig. 4d, e). These results suggest that ZTA-261 can improve lipid metabolism in mice as effectively as GC-1.

Fig. 4: Effects of TH analogs on the accumulation of lipids in mice.
figure 4

Mice fed either a normal diet (ND; 10 kcal% fat) or a high-fat diet (HFD; 60 kcal% fat) were intraperitoneally injected with either saline, T3 (0.1 or 1 µmol/ kg·day), GC-1 (0.1 or 1 µmol/kg·day) or ZTA-261 (0.1 or 1 µmol/kg·day). After 3 weeks of injection, serum and liver tissues were collected, and the amount of serum cholesterol (a), serum triglycerides (b), liver cholesterol (c), liver triglycerides (d), and liver total lipids (e) were measured. Data are shown as the mean ± SEM (n = 8–10). Significant differences among groups are indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, one-way ANOVA with Tukey’s post-hoc test).

Histological evaluation of lipid accumulation in the liver

Excessive accumulation of lipid droplets in hepatocytes is characteristic of non-alcoholic fatty liver disease (NAFLD)29. To evaluate the effect of the TH analogs on liver lipids, liver sections were stained with Oil Red O to visualize the accumulation of lipid droplets in hepatocytes (Fig. 5a and Supplementary Fig. 4). Treatment with ZTA-261 at a dose of 1 µmol/kg·day significantly reduced the Oil Red O-positive area compared to the vehicle to a level as low as the ND-treated group (Fig. 5a, b).

Fig. 5: Oil Red O staining of the liver section of mice treated with TH analogs.
figure 5

Mice fed either a normal diet (ND; 10 kcal% fat) or a high-fat diet (HFD; 60 kcal% fat) were intraperitoneally injected with either saline, T3 (0.1 or 1 µmol/kg day), GC-1 (0.1 or 1 µmol/kg day) or ZTA-261 (0.1 or 1 µmol/kg day). After three weeks of injection, liver tissues were collected, frozen sections were prepared, and subjected to Oil Red O staining (a). Stained areas were quantified and normalized to the total image area. Data are shown as box plots, with median, 25th and 75th percentiles and maximum and minimum values (n = 8–10). Significant differences among groups are indicated by asterisks (***P < 0.001 and ****P < 0.0001, Kruskal–Wallis test with Dunn’s multiple comparisons test) (b).

Gene expression analysis in the liver

To explore the mechanism by which ZTA-261 promotes lipid metabolism, we examined the expression of THR-regulated genes and genes involved in lipid metabolism in the liver30 by reverse transcription-quantitative PCR (RT-qPCR) analysis upon administration of the compound at 1 μmol/kg day (Fig. 6). We first examined the expression of THR-regulated genes, including Iodothyronine deiodinase 1 (Dio1), Malic enzyme 1 (Me1), and Thyroid hormone responsive (Thrsp). Dio1 catalyzes the conversion of T4 to T3. Me1 encodes an enzyme that generates NADPH for fatty acid synthesis. Thrsp is involved in triglyceride biosynthesis. The expression of Dio1 and Me1 were significantly upregulated by T3, GC-1, and ZTA-261 (Fig. 6a, b). GC-1 and ZTA-261 increased the expression of Thrsp, but T3 showed no significant difference compared with the vehicle (Fig. 6c).

Fig. 6: Effect of TH analogs on the expression of THR-regulated genes and lipid metabolism-related genes in the liver.
figure 6

Total RNA was extracted from frozen liver tissue from mice fed with a high-fat diet administered either saline, T3, GC-1, or ZTA-261 at 1 µmol/kg day, and subjected to RT-qPCR analysis. Relative expression levels of Dio1 (a), Me1 (b), Thrsp (c), Cyp7a1 (d), Ldlr (e), Srebp1c (f), and Pnpla2 (g) were determined by Pfaffl’s method using Rsp18 as a reference gene. Data are shown as box plots, with median, 25th and 75th percentiles and maximum and minimum values (n = 9–10). Significant differences among the groups are indicated by different letters Significant differences among groups are indicated by asterisks (*P < 0.05, **P < 0.01, and ***P < 0.001, Kruskal–Wallis test with Dunn’s multiple comparisons test).

We checked the expression levels of four lipid metabolism-related genes, Cytochrome P450, family 7, subfamily a, polypeptide 1 (Cyp7a1), Low-density lipoprotein receptor (Ldlr), Sterol regulatory element binding protein-1c (Srebp1c), and Patatin-like phoshodomain containing 2 (Pnpla2). CYP7A1 is a rate-limiting enzyme involved in cholesterol catabolism. LDLR promotes the cellular uptake of LDL and facilitates cholesterol degradation. The expression of Cyp7a1 (Fig. 6d) and Ldlr (Fig. 6e) in the T3-, GC-1-, and ZTA-261-treated groups was not significantly different from that in the vehicle-treated group. The expression level of Srebp1c, which encodes a transcription factor that activates the synthesis of fatty acids and triglycerides, was significantly lower in the T3 and GC-1 groups than in the vehicle group. Although not statistically significant, the ZTA-261-treated group showed decreased Srebp1c expression (Fig. 6f). Levels of patatin-like phosphodomain-containing 2 (Pnpla2), also known as adipose triglyceride lipase (ATGL), which encodes an enzyme that catalyzes the hydrolysis of triglycerides31, were significantly higher in the GC-1 and ZTA-261-treated group than in the vehicle group (Fig. 6g). These results suggest that GC-1 and ZTA-261 promote lipid metabolism by activating similar pathways.

Toxicity assessment of TH analogs in the liver and heart

To evaluate the hepatotoxicity of ZTA-261 in vivo, we measured the serum levels of alanine aminotransferase (ALT), a marker of hepatocyte necrosis. A significant increase in serum ALT level was observed when GC-1 was administered at 1 μmol/kg day compared to vehicles (Fig. 7a). On the other hand, ZTA-261-administered mice did not show a significant increase in serum ALT levels compared to vehicles at either 0.1 or 1 μmol/kg day (Fig. 7a), suggesting that the ZTA-261 is less hepatotoxic than GC-1.

Fig. 7: Toxicity assessment of TH analogs on the liver and heart.
figure 7

Serum was collected from mice fed with either a normal diet or a high-fat diet injected with either saline, T3 (0.1 or 1 µmol/kg day), GC-1 (0.1 or 1 µmol/kg day), or ZTA-261 (0.1 or 1 µmol/kg day) and ALT levels were measured to assess hepatic toxicity of compounds. Data are shown as the mean ± SEM (n = 8–10). Significant differences among groups are indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, one-way ANOVA with Tukey’s post-hoc test) (a). Heart weight was measured to assess heart toxicity. Data are shown as the mean ± SEM (n = 8–10). Significant differences among groups are indicated by asterisks (*P < 0.05, and ****P < 0.0001, one-way ANOVA with Tukey’s post-hoc test) (b). Heart sections were prepared and subjected to hematoxylin–eosin staining. 100 cells were randomly selected from the left ventricle of an individual mouse, their trans-nuclear widths were measured, and average values were calculated. Data are shown as the mean ± SEM (n = 8–10). Significant differences among groups are indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, one-way ANOVA with Tukey’s post-hoc test) (c).

Left ventricular hypertrophy is often associated with hyperthyroidism caused by increased heart rate mediated by human THRα32. Therefore, THRβ-selective analogs are expected to circumvent this adverse effect on the heart. We measured the heart weight of mice treated with the T3 or TH analogs to examine whether cardiac hypertrophy was induced. Administration of both high (1 μmol/kg day) and low (0.1 μmol/kg day) doses of T3 or a high dose of GC-1 caused a significant increase in heart weight compared to vehicles (Fig. 7b). No such effect was found in the heart of animals treated at either high or low doses of ZTA-261 (Fig. 7b). To test whether the increase in heart weight resulted from an increase in the size of the cardiomyocytes in the left ventricle, we stained the heart sections with hematoxylin-eosin and measured the transnuclear width of the left ventricular cardiomyocytes (Supplementary Fig. 5 and Fig. 7c). Both high and low doses of T3, a high dose of GC-1, and a high dose of ZTA-261 administration significantly increased the width of the myocytes. However, the effect sizes were larger in animals treated with T3 compared to those treated with GC-1 or ZTA-261 (Fig. 7c).

Toxicity assessment of TH analogs in bone

Bone morphometry was performed using X-ray micro-computed tomography (μCT) to evaluate the side effects of the TH analogs on bone. The metaphysis of the distal femur was analyzed (Fig. 8a), and the following parameters describing the microstructure of the trabecular bone were measured: BV/TV (%), Tb. N (mm−1), Tb. Sp (μm), Tb. N (μm) and Conn. D (mm−3) (Fig. 8b–f). BV/TV indicates the fraction of bone volume (BV) in the total volume of interest (TV). The BV/TV value in the 1 μmol T3-treated group was significantly lower than those in the vehicle- and 1 μmol ZTA-261-treated groups (Fig. 8b). A significant decrease in BV/TV was also observed in the 1 μmol GC-1-treated group compared to the vehicle group, but not between the vehicle and ZTA-261-treated groups (Fig. 8b). Tb. N is the number of trabeculae per unit length. Tb. N was the lowest in the 1 μmol T3-treated group, followed by the 1 μmol GC-1-treated group. The Tb. N values were significantly lower in these two groups than those in the vehicle- and ZTA-261-treated groups (Fig. 8c). Tb. Sp indicates the mean distance between trabeculae. It was significantly higher in the 1 μmol T3-treated group than that in the other groups. The ZTA-261-treated group did not show a significant difference compared with the vehicle-treated group (Fig. 8d). There was no significant difference in Tb. Th, trabecular thickness, in the four groups (Fig. 8e). Conn. D. characterizes the redundancy of trabecular connections normalized by TV. It was significantly decreased in the T3 and GC-1-treated groups compared to that in the vehicle groups (Fig. 8f). A slight decrease was also observed in the ZTA-261 treated group compared to that in the vehicle group, but the decrease was not significant (Fig. 8f). These parameters except Tb. Th were severely affected in the T3-treated group, which is consistent with the fact that T3 has no selectivity for THRα and β. From these results, the harmful effects on the bone were reduced in the THRβ-selective analogs; however, when comparing GC-1 and ZTA-261, the toxicity was further decreased in ZTA-261.

Fig. 8: Toxicity assessment of TH analogs on the bone by μCT analysis.
figure 8

Femur was collected from mice fed with a high-fat diet (60 kcal% fat) injected with either saline, T3 (1 µmol/kg day), GC-1 (1 µmol/kg day), or ZTA-261 (1 µmol/kg day). The trabecular region of the distal femur of each mouse was subjected to μCT scanning. Representative μCT images of the distal femur (a). The analyzed region, indicated by a white rectangle, spanned 0.3–2.3 mm proximal to the red line marking the epiphyseal growth plate. Trabecular bone parameters including bone volume fraction (BV/TV) (b), trabecular number (Tb.N) (c), trabecular separation (Tb.Sp) (d), trabecular thickness (Tb.Th) (e), and connectivity density (Conn.D) (f) were calculated. Data are shown as the mean ± SEM (n = 5). Significant differences among groups are indicated by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, one-way ANOVA with Tukey’s post-hoc test).

Blood–brain barrier permeability of TH analogs

TH stimulates the differentiation of oligodendrocyte progenitors and myelin production during central nervous system (CNS) development. In addition to its developmental functions, TH facilitates remyelination after pathological demyelination in the adult CNS33. Thus, TH and its analogs are potential treatments for CNS disorders such as multiple sclerosis.

Previously, the CNS distribution of GC-1 was assessed in mice and was found to be distributed to the CNS with efficiency at the lowest limit for compounds approved as CNS drugs34. Permeability of the blood-brain barrier (BBB) is a key property of drugs that determines their distribution to the CNS. Therefore, we tested BBB transparency in vitro using the BBB kit RBT-24H (PharmaCo-Cell)35. We applied 10 μM of either GC-1 or ZTA-261 to the wells, mimicking the luminal side of the BBB, and allowed them to penetrate into the wells mimicking the abluminal side for 30 min. Caffeine and digoxin were used as the positive and negative controls, respectively. The apparent permeabilities (106 cm/s) of these compounds were calculated using the following equation:

$${P}_{{{\rm {app}}}}=\,\frac{{V}_{{\rm {A}}}}{A{\cdot }{\left[C\right]}_{{{{luminal}}}}}{{{{\rm{\cdot }}}}}\frac{\Delta {\left[C\right]}_{{{{Abluminal}}}}}{\Delta T}$$
(1)

where \({V}_{\rm {{A}}}\) is the volume on the abluminal side, \(A\) is the surface area, \({\left[C\right]}_{{{\rm {luminal}}}}\) is the initial concentration of drugs on the luminal side, \(\Delta {\left[C\right]}_{{{\rm {Abluminal}}}}\) is the change in the concentration of drugs on the abluminal side within the assay period \(\Delta T\) (30 min). Papp values of GC-1 and ZTA-261 were 14.6 ± 3.54 and 7.28 ± 2.55, respectively, while caffeine, known as a CNS stimulant, showed Papp values of 76.4 ± 8.35 (Supplementary Table 2). The Papp value of digoxin could not be determined because the concentration of digoxin on the abluminal side was below the detection limit (0.1 μM). These results indicate that GC-1 had limited permeability to the CNS, which is consistent with a previous report34, and that the permeability of ZTA-261 was even smaller than that of GC−1. These results suggest that ZTA-261 is particularly useful for treating liver hyperlipidemia without affecting the CNS. Future studies are required to identify transporters for GC-1 and ZTA-261.

Effects of TH analogs on serum T3 levels

To investigate the impact of the TH analogs on endogenous TH levels, we measured the concentration of T3 in serum samples collected from animals injected with either T3, GC-1 or ZTA-261 at 1 μmol/kg day (Table 2). In the T3-injected animals, the serum T3 concentration was 743.5 ± 78.0 ng/dL, which was about 10 times higher than the control group, while in the GC-1 and ZTA-261-injected groups, the amount of T3 is below the quantification limit. The increase in T3 levels in the T3-injected group may reflect exogenous T3. It is possible that GC-1 and ZTA-261 inhibit endogenous T3 synthesis and secretion by acting on the pituitary gland, which does not have a BBB, but on the hypothalamus.

Table 2 Serum T3 concentration

Discussion

Several THRβ-selective agonists have been developed recently36,37,38, and one such molecule, MGL-3196, is currently under phase 3 trials for the treatment of non-alcoholic steatohepatitis (NASH) and NAFLD39. Considering this, THRβ is a promising drug target for improving lipid metabolism and NASH/NAFLD.

In vitro THR competition assays revealed that ZTA-261 binds to THRβ with a high affinity, comparable to T3 and GC-1, and with higher THRβ selectivity than GC-1 (Fig. 1, Table 1). The effect of ZTA-261 on lipid metabolism was tested in vivo using a mouse model of high-fat diet-induced obesity (Fig. 4). The results showed that administration of both GC-1 and ZTA-261 led to a significant reduction in the levels of liver total cholesterol, triglycerides, and serum triglycerides. This finding is consistent with the fact that the IC50 values of GC-1 and ZTA-261 for THRβ were quite similar.

We examined the expression of genes regulated by THR and those related to lipid metabolism in the liver using RT-qPCR (Fig. 6). Consistent with previous studies40,41, the administration of T3, GC-1, and ZTA-261 significantly increased Dio1 and Me1 expression. GC-1 and ZTA-261 also upregulated Thrsp, whereas T3 tended to increase Thrsp, although not significantly. These results suggest that these compounds act via THR-dependent signaling (Fig. 6a–c).

Previous studies have shown that GC-1 reduces serum cholesterol levels by inducing Cyp7a142,43. However, another study showed that these compounds have dose-dependent effects on expression levels of Cyp7a1 in mice28. The greatest increase in Cyp7a1 was observed at a dose of 20 nmol/kg/day for both T3 and GC-1, with a gradual decrease as the dose increased28. In this study, Cyp7a1 expression was not affected by T3, GC-1, or ZTA-261 (Fig. 6d and e), which may be explained by the high doses (i.e., 0.1 and 1 mmol/kg/day).

TH is believed to reduce serum cholesterol levels by stimulating Ldlr44. However, experiments using Ldlr−/− mice have shown that TH can decrease serum cholesterol levels independently of Ldlr28,42,43,45. Furthermore, GC-1 treatment did not always induce Ldlr expression42. These findings indicate that treatment with TH and its analogs does not necessarily increase Ldlr expression.

It is worth noting that the results of our gene expression analysis were consistent with the observation that ZTA-261, T3, and GC-1 were more effective at reducing triglyceride levels than cholesterol levels in the liver (Fig. 4c, d). Treatment with T3 and GC-1 significantly decreased the expression of Srebp1c, activating the synthesis of triglycerides, and treatment with ZTA-261 significantly upregulated the expression of Pnpla2, regulating the degradation of triglycerides (Fig. 6f, g). On the other hand, expression of Cyp7a1 and Ldlr was not affected by these compounds (Fig. 6d, e). These results explain why the administration of these compounds decreased triglyceride levels more efficiently than cholesterol.

ZTA-261 was more effective at reducing fat droplets in the liver. The administration of ZTA-261 at 1 µmol/kg·day reduced fat droplets to a level comparable to that in animals fed a normal diet. In contrast, the liver TG levels were similar in the GC-1- and ZTA-261-treated groups. This discrepancy may be explained by the fact that the number of lipid droplets is also increased by drug-induced liver injury46. In this study, serum ALT levels were higher in the GC-1-treated groups than in the ZTA-261-treated groups. The droplets observed in the GC-1-treated group could be partly explained by the side effects of GC-1 on the liver.

T3 causes harmful effects on the bone and cardiac system by binding to THRα, which is a major barrier to its use as an obesity treatment47,48. When we examined serum T3 levels, neither ZTA-261 nor GC-1 induced hyperthyroidism (Table S2). We observed an increase in heart weight in the T3- and 1 µmol GC-1-treated groups but not in ZTA-261-treated groups at any concentration (Fig. 7b). We conducted a micro-CT analysis to evaluate the impact of T3, GC-1, and ZTA-261 on bone structure (Fig. 8). The results showed that administration of T3 led to significant damage to the trabecular bone structure of the distal femur as indicated by the BV/TV, Tb.N., Tb.Sp., and Conn.D. values. This harmful effect was mitigated in the GC-1-treated group. The ZTA-261 treatment group did not show any significant differences in these values compared with the control groups. The differences in bone and cardiac toxicity among T3, GC-1, and ZTA-261 can be attributed to the degree of THRβ selectivity. T3, being non-selective, had a more harmful impact on the heart and bone than THRβ-selective GC-1 and ZTA-261. Comparatively, ZTA-261 was less toxic than GC-1, possibly due to its higher selectivity for THRβ than GC-1. Since white fat is considered to have more THRα activity than THRβ activity, the reduction in epidydimal fat by GC-1 may also reflect the degree of THRβ selectivity.

Drug-induced liver injury is known to be one of the major factors in the discontinuation of clinical trials49. The hepatotoxicity of ZTA-261 was evaluated by quantifying alanine aminotransferase (ALT) released in the blood. Administration of T3 and GC-1 at a dose of 1 µmol/kg significantly increased serum ALT levels, while ZTA-261, when administered at the same concentration, did not. This indicated that ZTA-261 did not cause hepatotoxicity at the dosage required to promote lipid metabolism (Fig. 7a). Notably, the variability in the ALT levels in individual animals was lower in the ZTA-261-treated groups than in the GC-1-treated groups (Fig. 7a). From a clinical perspective, small individual differences in the harmful effects of a compound are considered an appropriate property as a lead for further modifications. Further pharmacokinetics and pharmacodynamics studies are required to establish the potential of ZTA-261 as a pharmaceutical treatment. In this study, we administered a higher dose of compounds for a longer duration than usual to investigate potential side effects caused by THRα. In future studies, the dosage and duration should be carefully considered to further evaluate the effects of ZTA-261.

In conclusion, we developed a TH analog ZTA-261 and assessed its effects and toxicity in vitro and in vivo. ZTA-261 has high affinity and selectivity for THRβ, which accelerates liver lipid metabolism with low toxicity in the heart and bone. These results suggested that ZTA-261 could potentially be used as a drug lead for diseases related to lipid metabolism, such as NAFLD.