Novel Small-Molecule PGC-1α Transcriptional Regulator With Beneficial Effects on Diabetic db/db Mice

Forskolin, luciferase, and luciferase antibody were obtained from Sigma-Aldrich. The 2600-base pair PGC-1α promoter, delta MEF and delta CRE binding element promoter-driven luciferase reporter plasmids, were obtained from Addgene. Radiolabeled [9,10-³H(N)]-palmitic acid was purchased from PerkinElmer. The transfection reagent Lipofectamine 2000 was obtained from Invitrogen, and the siLentFect lipid was obtained from Bio-Rad. Luciferase substrate was purchased from Promega. Antibodies against cAMP response element- binding protein (CREBP), phospho-CREBP (Ser133), p38 mitogen-activated protein kinase (MAPK), and phospho-p38 MAPK (Thr180/Tyr182) were purchased from Cell Signaling Technology; the antibody against muscle actin was from Santa Cruz. Other materials were obtained as previously described (26).

A pGL3-basic luciferase reporter plasmid containing the 2600-base pair PGC-1α promoter (27) and pcDNA3.1, in a ratio of 10:1 (0.8 μg of total DNA per well), were cotransfected into HEK293 cells in a 24-well plate by Lipofectamine 2000. Stable cell lines (PGC-1α-luc) were selected based on their resistance to 1 mg/mL G418 and a strong luciferase enzyme signal.

Stable HEK293 PGC-1α-luc cells were plated into 384-well plates at approximately 3000 cells per well. After overnight culture, compounds at a final concentration of 2 μg/mL were added to the culture medium. At the end of 24 h, luciferase substrate was added to each well and the released luciferin signal then was detected using an EnVision microplate reader.

Luciferase was diluted to 20 units and incubated with compounds in a 384-well plate. Then, luciferase substrate was added, and the released luciferin signal was detected using an EnVision microplate reader.

The assay was initiated by adding [9,10–³H(N)]-palmitic acid to a final concentration of 250 μmol/L and 1.5 μCi per well in Dulbecco’s modified Eagle’s medium (DMEM). After incubation with the compound for 4 h in differentiated L6 myotubes, a sample from each well was added to charcoal slurry for centrifugation and then the radioactivity was measured.

For plasmid transfection, L6 myoblasts were transfected with Lipofectamine 2000 containing the pGL-3 basic vector, PGC-1α promoter luciferase, delta MEF, or delta CRE promoter luciferase. Six hours after transfection, the medium was changed for differentiation. After being cultured overnight, the cells were incubated with compound for 24 h. For transfection of small interfering RNA (siRNA), 20 nmol/L siRNA (sense: GAUUCAAACUCAGACGAUU; antisense: AAUCGUCUGAGUUUGAAUC) targeting PGC-1α was transfected on day 4 of L6 myotube differentiation using siLentFect lipid in serum-free DMEM. Six hours after transfection, the medium was replaced with normal medium. The next day, the cells were treated with compound for 24 h for studies.

The animal experiments were approved by the Animal Ethics Committee of the Shanghai Institute of Materia Medica. Male C57BKS db/db and lean littermate C57BKS mice were housed in a temperature-controlled room (22 ± 2°C) with a 12-h light/dark cycle. For the pharmacokinetic studies, a single dose of ZLN005 was administered to db/db mice. The plasma and tissue samples were collected at different time points, and the compound concentrations were determined by liquid chromatography–mass spectrometry/mass spectrometry. For chronic treatment, six to eight 8-week-old mice were assigned randomly to the various treatment groups by body weight and glucose levels. Lean and db/db mice received oral administration of either vehicle (0.5% methylcellulose), ZLN005, or metformin for 6 weeks. Body weight and food intake were recorded daily. Fasting blood glucose levels were measured after 6 h of fasting. At the 3rd week of treatment, the effect of chronic ZLN005 administration on the respiratory exchange ratio (RER) of db/db mice was measured using an eight-chamber indirect calorimeter. The glucose tolerance test (1.5 g · kg⁻¹ glucose i.p.) and insulin tolerance test (1 unit · kg⁻¹ insulin i.p.) were performed after 6 h of starvation. The pyruvate tolerance test (PTT; 1.5 g · kg⁻¹ sodium pyruvate i.p.) was performed in mice that had been fasted overnight. Plasma concentrations of nonesterified fatty acids (NEFAs), triglycerides, and insulin were measured as previously described (26). Plasma β-hydroxybutyrate was measured using a β-hydroxybutyrate assay kit (Abcam). At the end of treatment, the animals were killed, and their tissues were freeze-clamped for further experiments. For preliminary safety evaluation, Sprague-Dawley rats were orally administered either vehicle or ZLN005 for 14 days. The rats were checked every day for death and clinical symptoms. At day 15, the animals were anatomized and blood was collected for further investigation.

These assays were performed as previously described (26,28–31).

These assays were performed as previously described (26). In addition, HEK293 cells were cultured in high-glucose DMEM supplemented with 10% FBS. For real-time PCR, all samples were run in duplex and normalized to actin or tubulin expression. For details of primer sequences, see Supplementary Table 1.

The results are presented as the mean ± SE. Differences between the groups were analyzed with the Student’s t test. P < 0.05 was regarded as statistically significant.

Studies have shown an inverse correlation between levels of PCG-1α in muscle and mitochondrial activity, insulin resistance, and T2DM (32,33). Therefore, we created an HEK293 stable cell line containing luciferase expression driven by the human PGC-1α promoter (PGC-1α-luc); then the luciferase reporter assay was optimized for application in automated HTS. After we randomly screened a library containing 48,000 pure synthetic compounds with diverse structures, the novel compound ZLN027 (Supplementary Fig. 1A) was found to increase the PGC-1α promoter reporter by l.7-fold (Supplementary Fig. 1B). Forskolin and dexamethasone, which were reported to stimulate expression of PGC-1α, had stimulatory effects at 1.4- and 2.0-fold, respectively. By checking the cell survival rate, we found obvious cytotoxicity for ZLN027 in HEK293 and L6 myotubes. Extensive data mining identified another compound, ZLN005 (Fig. 1A), which had the same chemical backbone but did not have obvious cytotoxicity (Supplementary Fig. 1D and E). ZLN005 was observed to potently inhibit luciferases, but ZLN027 was not (Supplementary Fig. 1C). These results might explain the discrepancy between the two compounds in the PGC-1α promoter reporter assay. The transcriptional modulatory effect of ZLN005 in L6 myotubes was also investigated. As shown in Fig. 1B, ZLN005 increased PGC-1α mRNA levels in a dose-dependent manner; 20 μmol/L ZLN005 caused a threefold increase over the control after 24 h. At 10 μmol/L, the PGC-1α mRNA levels were increased to almost the same extent at 16 to 48 h (Fig. 1C).

PGC-1α is a powerful transcriptional coregulator of GLUT4 and mitochondrial genes, including components of the electron transport system. As seen in Fig. 1D, the mRNA levels of GLUT4, NRF1, estrogen-related receptor α (ERRα), cytochrome c oxidase 5b (cox5b), and acyl-CoA oxidase (24,34) were increased by ZLN005 (10 μmol/L).

Because GLUT4 plays a crucial role in glucose uptake in skeletal muscle, we investigated this effect of ZLN005 further. As shown in Fig. 1E, ZLN005 stimulated glucose uptake dose dependently after 24 h of treatment, with 20 μmol/L of ZLN005 resulting in a 1.8-fold improvement. Because the induction of the expression of genes involved in mitochondrial biogenesis and fatty acid oxidation (FAO) (35) was observed. As shown in Fig. 1F, ZLN005 increased oxidation of palmitic acid dose dependently, with 20 μmol/L ZLN005 resulting in a 1.28-fold increase after 24 h compared with control. These results demonstrated that ZLN005 stimulated the expression of PGC-1α and downstream genes in skeletal muscle cells and improved glucose utilization and FAO.

Gluconeogenesis plays an important role in regulating glucose levels, and PGC-1α is known to stimulate this process (36). ZLN005 had no effect on levels of PGC-1α mRNA in rat primary hepatocytes after 24 h of treatment (Fig. 2A). The expression of the key gluconeogenic enzyme coregulated by PGC-1α, phosphoenolpyruvate carboxykinase (PEPCK), was not affected by ZLN005 (Fig. 2B), nor did ZLN005 increase glucose production (Fig. 2C) in primary hepatocytes. The contrary effects of ZLN005 in L6 myotubes and primary hepatocytes suggested that expression of PGC-1α was regulated in a cell type–specific manner.

To evaluate the dependence of PGC-1α in the effects of ZLN005, L6 myotubes were transfected with PGC-1α siRNA. ZLN005-stimulated expression of the PGC-1α gene and oxidation of palmitic acid was blocked upon PGC-1α silencing (Fig. 3A and B). This indicated that ZLN005-stimulated FAO was mediated by PGC-1α.

To clarify the mechanism of ZLN005 acting on PGC-1α, transfection of the PGC-1α promoter harboring truncated mutations in the MEF2 and CREBP binding sites were analyzed (27). Because ZLN005 inhibits luciferase activity, luciferase protein levels were investigated as the direct readout for transcriptional activity. The PGC-1α promoter drives luciferase induction upon treatment with ZLN005 (10 μmol/L), but the promoter constructs with the truncated CREBP binding site showed little difference compared with the PGC-1α promoter. However, a truncated MEF2 binding site resulted in an ablation of the increase in luciferase protein in response to treatment with ZLN005 (Fig. 3C). This suggested that ZLN005 stimulation of PGC-1α expression was dependent on the MEF2 binding site.

Studies have reported that p38 MAPK, the energy sensor protein AMP-activated protein kinase (AMPK), and cAMP-responsive CREBP signaling pathways control expression of PGC-1α (37,38). Therefore, we determined whether treatment of L6 myotubes with ZLN005 (10 μmol/L) would lead to activation of AMPK, p38 MAPK, or CREBP. Phosphorylation levels of p38, which phosphorylates and activates MEF2 (39) and phosphorylates PGC-1α protein (40), slightly decreased, suggesting the p38 pathway was not the cause of the increased expression of PGC-1α. Phosphorylation of CREBP remained unchanged, which was consistent with the previously observed lack of effect on luciferase expression for the CREBP binding site. Meanwhile, the phosphorylation of AMPK and its downstream acetyl-CoA carboxylase (ACC) were increased significantly (Fig. 3D).

The dose-dependent activation of AMPK by ZLN005 was confirmed at 24 h (Fig. 3E). Thus we examined whether the effects of ZLN005 on glucose uptake were dependent on AMPK using the AMPK inhibitor compound C. Pretreatment with compound C attenuated the ability of ZLN005 to induce phosphorylation of AMPK and expression of PGC-1α (Fig. 3F and G). These results suggested that PGC-1α expression stimulated by ZLN005 in L6 myotubes was dependent on the AMPK pathway. To investigate the mechanism of AMPK activation by ZLN005, we measured the direct effect of ZLN005 on AMPKα2β1γ1 and AMPKα2β2γ1 on a molecular level, and no catalytic kinase activation was observed (data not shown). We found it could increase the ADP-to-ATP ratio in L6 myotubes (Fig. 3H) and mildly increase oxygen consumption in isolated muscle mitochondria (Fig. 3I). This suggests that ZLN005 increases the ADP-to-ATP ratio to activate AMPK by uncoupling mitochondria respiration in L6 myotubes.

Pharmacokinetic studies of db/db mice showed that ZLN005 was absorbed into the plasma quickly, reaching a concentration of 3.7 μmol/L within 15 min. The concentration declined to 0.44 μmol/L within 4 h (Fig. 4A) after a single oral dose of 15 mg/kg. ZLN005 reached a fairly high concentration of 60.9 μmol/L within 15 min in liver tissue, which then declined to 10.9 μmol/L within 4 h (Fig. 4B).The concentration of ZLN005 in muscle tissue was stable at approximately 3–4 μmol/L over 4 h.

Because PGC-1α is a key regulator of oxidative metabolism, we studied the effects of ZLN005 on whole-body fat oxidation using indirect calorimetry. The db/db mice were dosed with ZLN005 (15 mg · kg⁻¹ · day⁻¹ ) or vehicle for 2 weeks. The last administration was given 4 h before the experiment. The animals were monitored for oxygen consumption and CO₂ production for 21 h (between 1700 and 1400 h). There were no significant changes in Vo₂ and Vco₂ (Fig. 4C and D), but a decrease in the RER was observed, indicating a shift to fatty acid use (Fig. 4E). The heat was almost unchanged during the 21-h measurement, except for a slight increase during cycles 11 to 20 (Fig. 4F).

To assess the antidiabetic efficacy of ZLN005 in vivo, we investigated the effect of chronic ZLN005 administration in db/db mice and lean mice. ZLN005 (15 mg · kg⁻¹ · day⁻¹) was administered orally for 6 weeks, and metformin (250 mg · kg⁻¹ · day⁻¹) was used as a positive control. During the treatment, ZLN005 did not affect body weight gain or food intake in either db/db mice or lean mice (Fig. 5A and B). In lean mice, plasma glucose generally was unchanged by ZLN005 treatment. In db/db mice, however, random blood glucose and fasting blood glucose levels decreased significantly over 4 weeks by ZLN005 and metformin treatment (Fig. 5C and D). ZLN005 did not alter glucose tolerance in lean mice, but in db/db mice it improved glucose clearance, as evidenced by the approximately 14% decrease in the area under the curve (AUC) (Fig. 5E). An insulin tolerance test revealed that treatment with ZLN005 significantly decreased insulin resistance in db/db mice, as evidenced by the approximately 18% decrease in the AUC (Fig. 5F). A PTT also was performed in db/db mice, and ZLN005 improved pyruvate tolerance, as evidenced by the 16% decrease in the AUC (Fig. 5G).

In db/db mice, plasma NEFA and triglyceride levels were decreased by 20% and 37%, respectively, and cholesterol was decreased by 10% (Table 1) with ZLN005 treatment. Plasma insulin and β-hydroxybutyrate content, liver/bodyweight index and adipose composition, and muscle and liver triglyceride levels, however, were not ameliorated by treatment with ZLN005 or metformin.

Together, these results indicated that ZLN005 had antihyperglycemia and antihyperlipidemia effects and increased insulin sensitivity in db/db mice but not lean mice. These results suggested that ZLN005 might act selectively in the diabetic mouse model and have no adverse effects on normal lean mice.

To clarify the tissue-specific efficacy, we found that chronic treatment with ZLN005 led to significant induction of PGC-1α mRNA expression and PGC-1α target genes in the gastrocnemius, including GLUT4 and mitochondria OXPHOS genes such as ERRα, cytochrome-c, cox5b, ATPase-F1α, and uncoupling protein 3. Genes related to FAO, such as medium-chain acyl-CoA dehydrogenase (MCAD) and long-chain acyl-CoA dehydrogenase, also were induced. However, NRF1 and carnitine palmitoyl transferase 1 remained unchanged (Fig. 6A).

We also investigated gene expression in liver tissue. Unlike in skeletal muscle, PGC-1α mRNA was down-regulated by 34% compared with the vehicle group. Its downstream gluconeogenic genes, including glucose-6-phosphatase (G6Pase) and PEPCK, were reduced by 31% and 27%, respectively (Fig. 6B), in response to ZLN005 treatment. PGC-1α is known to coordinate the mitochondria OXPHOS and FAO in liver, so the expression of mitochondria OXPHOS genes, such as cytochrome-c, ERRα, cox5b, and ATPase-F1α, also were investigated. However, no reduction was observed in the mRNA level of these genes. Similarly, no changes in carnitine palmitoyl transferase 1, MCAD, and acyl-CoA oxidase were observed.

PGC-1α has emerged as a master regulator of mitochondria, and mitochondrial biogenesis resulting from an increased number of mitochondrial DNA copies. Indeed, mitochondrial DNA was increased (31%) by ZLN005 in the gastrocnemius of db/db mice, an effect that was not observed in liver (Fig. 6C). We also measured AMPK and ACC phosphorylation in abdominal muscle and liver tissue of db/db mice. Consistent with the results found at the cellular level, AMPK and ACC phosphorylation were increased in abdominal muscle but not in liver tissue (Fig. 6D and E). These differing effects suggested that ZLN005 increased PGC-1α expression in a tissue-specific manner in vivo.

Transcription cofactor PGC-1α is precisely regulated in a number of metabolic tissues and physiological contexts, such as in response to fasting in the liver and exercise in muscle. It is dysregulated in pathological states, such as mitochondrial dysfunction in skeletal muscle and hypergluconeogenesis. It may, therefore, be a potential target for drug discovery in the treatment of metabolic syndromes. Arany et al. (34) recently set up an HTS method for the identification of small molecules that induced PGC-1α in primary skeletal muscle cells, demonstrating that PGC-1α could be regulated. Here, we used a luciferase assay to identify small molecules and demonstrate positive effects on PGC-1α transcriptional regulation. During this process the shortcomings of a luciferase assay were uncovered: small molecules inhibiting luciferase activity affected the outcome of the HTS.

Exercise induces PGC-1α in skeletal muscle, where it induces GLUT4 and mitochondrial OXPHOS. Increasing PGC-1α activity in skeletal muscle should, therefore, be beneficial in metabolic disorders. First, we chose L6 myotubes to validate ZLN005. We found that it increased PGC-1α and downstream gene expression. Consistent with increased gene expression, FAO and glucose uptake also were up-regulated in L6 myotubes. An siRNA study then demonstrated that ZLN005-stimulated oxidation of palmitic acid was dependent on PGC-1α.

In the chronic treatment of db/db mice, the GLUT4 and mitochondrial OXPHOS genes were up-regulated in the gastrocnemius. Some of the regulators of FAO, such as long-chain acyl-CoA dehydrogenase and MCAD, also were up-regulated by ZLN005, consistent with ZLN005 promoting palmitic acid oxidation in L6 myotubes. The number of mitochondria was increased simultaneously, demonstrating that ZLN005 might promote mitochondria biogenesis.

Liver gluconeogenesis is known to be highly regulated by PGC-1α (19). Excessive gluconeogenesis is one of the symptoms of T2DM. We observed down-regulation of hepatic expression of PGC-1α, PEPCK, and G6Pase and improvement of PTT results in db/db mice. Several studies have shown that liver-specific down-regulation of PGC-1α leads to defects in lipid metabolism and hepatic insulin resistance (41), suggesting PGC-1α regulation must be tightly connected to lipid homeostasis and glucose metabolism.

Because decreased expression of PGC-1α was observed in liver tissue, the influence of PGC-1α on FAO and mitochondria biogenesis genes was explored next. It is noteworthy that these expression levels remained unchanged. It has been reported that mice with a germ line-–targeted disruption of PGC-1α exhibited normal mitochondrial abundance and morphology in the liver (42). Thus, compensatory pathways to maintain the function of liver mitochondria might exist. It also was reported that increased insulin sensitivity might elicit the phosphorylation and inhibition of PGC-1α through kinase Akt2/protein kinase B-β (43). In contrast to the effects observed in db/db mice, ZLN005 had no effect on the expression of PGC-1α and PEPCK in primary hepatocytes. Therefore, we speculated that alterations in PGC-1α in the liver tissue of db/db mice might be a secondary phenomenon.

The effect of ZLN005 incorporation in vivo in lean mice and db/db mice was tested. After a 2-week treatment, a reduction in the RER was observed, which reflected the preferential use of lipids over carbohydrates as a source of energy. After a 6-week treatment, ZLN005 improved glucose homeostasis and insulin sensitivity, possibly from muscle glucose uptake because an increase in GLUT4 expression was observed. This effect also was possible through diminished production of hepatic glucose, a principal constituent of whole-body glucose homeostasis. The improvement in PTT values may have resulted from down-regulation of PEPCK and G6Pase gene expression. Lipid profiles, such as triglycerides, NEFAs, and cholesterol levels, were reduced, potentially as a consequence of enhanced use of whole-body fat.

Because PGC-1α is activated in pancreatic β-cells in animal models of obesity and T2DM, poor transcriptional regulation might lead to a deficiency in insulin secretion (44). The unchanged plasma insulin levels suggested that ZLN005 might not show adverse effects on pancreatic β-cells. In addition, the antidiabetic action of ZLN005 was not a direct consequence of food intake because body weight gain and food intake differences between groups were minimal. Furthermore, ZLN005 had no major side effects in lean mice, indicating that it worked specifically in diabetic mice. For preliminary safety evaluation, Sprague-Dawley rats were administered either vehicle or ZLN005 (75 mg · kg⁻¹ · day⁻¹) orally for 14 days. All the animals survived, and no obvious change in the body weight gain or metabolic parameters was observed (Supplementary Tables 2–4).

At the PGC-1α promoter, there is a MEF binding site for the transcription factor MEF2, an insulin receptor substrate binding site for forkhead box class-O and a CRE binding site for activating transcription factor 2 and CREBP, all of which enhance PGC-1α transcription (32). In this case, we investigated the effect of ZLN005 on CRE and MEF binding sites and found that MEF was necessary for the up-regulation of PGC-1α in L6 myocytes.

A number of intracellular pathways are known to impinge on PGC-1α, including signaling by cAMP, AMPK, Ca²⁺, and p38 MAPK (40,45,46). To investigate the mechanism by which ZLN005 enhanced PGC-1α expression in L6 myotubes, we sought to identify signaling pathways and found that phosphorylation of Thr-180/Tyr-182 in p38 MAPK, which phosphorylates and activates MEF2 and activating transcription factor 2, was not increased after treatment with ZLN005. In addition, phosphorylation of Ser-133 of CREBP, which can be stimulated by Ca²⁺ signaling through calmodulin-dependent protein kinase IV and calcineurin A, also did not change (32). However, AMPK was activated by ZLN005 in L6 myotubes and the muscle tissue of db/db mice, and PGC-1α expression and glucose uptake are dependent on AMPK activation at the cellular level. To further address this issue, we found ZLN005 activated AMPK by mildly uncoupling mitochondria to increase the cellular ADP-to-ATP ratio. The weak uncoupling of mitochondria leading to a beneficial effect on metabolism was reported previously (47). AMPK is an energy sensor that is activated upon depletion of energy (48) and phosphorylates PGC-1α on Thr-177 and Ser-538 in skeletal muscle (49). It is plausible that AMPK phosphorylated PGC-1α and that PGC-1α then acted through positive feedback, binding with MEF2C to increase its own gene expression in myotubes (27). It is interesting that the AMPK pathway did not seem to be affected in rat primary hepatocytes (Supplementary Fig. 2) or the liver tissue of db/db mice, possibly because of the different energy states of the two types of cells. Considering the discrepancy in L6 myotubes and primary hepatocytes, and because MEF2 proteins are expressed predominantly in muscle and the brain (50), we speculated that ZLN005 regulates PGC-1α through muscle cell–specific transcription factors such as MEF2. Thus, the down-regulation of PGC-1α mRNA in the liver tissue of db/db mice might occur as a result.

Overall, our results demonstrated that small molecule–mediated tissue-specific activation of PGC-1α in skeletal muscle is feasible and that the novel compound ZLN005 exerts beneficial effects in an animal model of T2DM. These results further support the potential of PGC-1α as a drug target for the treatment of T2DM and metabolic syndromes.

This work was supported by the National Natural Science Foundation of China (Grants 81001463 and 81125023), the National Science and Technology Major Projects for Major New Drugs Innovation and Development (Grant 2012ZX09304011), the National Program on Key Basic Research Project (Grants 2012CB524906 and G1998051104), and the Shanghai Commission of Science and Technology (Grant 11DZ2292200).

No potential conflicts of interest relevant to this article were reported.

L.-N.Z. contributed to research data, discussion, and preparation of the manuscript. F.W., C.-M.X., and T.-C.D. contributed to the animal experiments. H.-Y.Z. contributed to the synthesis of compounds. Y.-Y.F., Y.-Y.L., M.G., and L.-Y.W. contributed to the discussion and research data. J.-Y.L., J.-K.S., and J.L. contributed to the study design, discussion, and editing of the manuscript. J.Y.-L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of data analysis.