Synthetic ERRα/β/γ Agonist Induces an ERRα-Dependent Acute Aerobic Exercise Response and Enhances Exercise Capacity

ABSTRACT: Repetitive physical exercise induces physiological adaptations in skeletal muscle that improves exercise performance and is effective for the prevention and treatment of several diseases. Genetic evidence indicates that the orphan nuclear receptors estrogen receptor-related receptors (ERRs) play an important role in skeletal muscle exercise capacity. Three ERR subtypes exist (ERR α , β , and γ ), and although ERR β / γ agonists have been designed, there have been significant difficulties in designing compounds with ERR α agonist activity. Additionally, there are limited synthetic agonists that can be used to target ERRs in vivo . Here, we report the identification of a synthetic ERR pan agonist, SLU-PP-332, that targets all three ERRs but has the highest potency for ERR α . Additionally, SLU-PP-332 has sufficient pharmacokinetic properties to be used as an in vivo chemical tool. SLU-PP-332 increases mitochondrial function and cellular respiration in a skeletal muscle cell line. When administered to mice, SLU-PP-332 increased the type IIa o x idative skeletal muscle fibers and enhanced exercise endurance. We also observed that SLU-PP-332 induced an ERR α -specific acute aerobic exercise genetic program, and the ERR α activation was critical for enhancing exercise endurance in mice. These data indicate the feasibility of targeting ERR α for the development of compounds that act as exercise mimetics that may be effective in the treatment of numerous metabolic disorders and to improve muscle function in the aging.

Characterization of SLU-PP-332 as a Novel Pan ERR Agonist with Potent ERR α Activity. The identification of the compound C29 as an ERR α inverse agonist 21 , 22 suggested that ERR α may not be an intractable drug target. However, as an inverse agonist, C29 acts to block the constitutive transcriptional activation activity of ERR α , and it is uncertain if one could design a small molecule that could act as an ERR α agonist enhancing the strong constitutive activity of the receptor. Using a rational drug design approach, we optimized the ERR β / γ selective agonist scaffold, GSK4716, for ERR α activity and gained 50-fold ERR α potency. We utilized the X-ray crystal structure of the ligand binding domain (LBD) of ERR γ bound to GSK4716 ( Figures 1A , B and S1 ) (PDB ID: 2GPP) 23 and subsequently modeled GSK4716 bound to the LBD ERR α in order to assess how we might optimize such binding to design an ERR α agonist ( Figure 1C ). In contrast to the first report of GSK4716 displaying no ERR α activity, we observed very weak agonist activity in an ERR cotransfection assay ( Figure 1F ) and believed that the GSK4716 scaffold may be useful as an initiation point to develop high-affinity ERR α agonists. The X-ray structure of ERR γ LBD bound to GSK4716 is the only X-ray crystal structure for any ERR bound with an agonist ligand ( Figure 1B ). 23 In this structure, the agonist GSK4716 binds in a previously unidentified pocket dubbed the “agonist” pocket near the solvent-exposed surface of the receptor ( Figure S1 ). 23 The phenolic hydroxyl group of GSK4716 interacts with Asp328 ( Figure 1B ) while the carbonyl of the acyl hydrazone bridges to two water molecules, one of which interacts with Arg316 and the other water molecule interacts with Leu309 ( Figure S1 ). Molecular modeling of GSK4716 in the LBD of ERR α followed by energy minimization to refine the protein–ligand complexes reveals similar interactions to that observed with ERR γ X-ray crystal structure ( Figure 1C ). Our strategy to design high-affinity ERR α agonists was based on optimizing ligand interactions with Phe328 that is specifically in ERR α . We employed a strategy to optimize GSK4716 based on converting the isopropyl phenyl group of GSK4716 to a more hydrophobic moiety that could potentially gain affinity by interacting with the Phe328 in ERR α . Molecular modeling of a compound with a naphthalene substituent in place of the isopropyl phenyl group (SLU-PP-332; Figure 1A bottom and Figures 1C , S1 , and S2 ) in the LBD of ERR γ and ERR α predicted the newly added phenyl group to make π – π interactions with Phe435 (ERR γ ) or Phe495 and Phe328 (ERR α ). We hypothesized that this modification would improve the affinity of the ligand in both receptors, but particularly toward ERR α due to a potential π – π stacking interaction between the ligand naphthalene group and Phe328 (the corresponding alanine residue in ERR β and ERR γ is unable to make such interactions). As predicted, SLU-PP-332 gained substantial ERR α potency vs GSK4716 in addition to a moderate increase in potency for ERR γ in a full-length ERR cell-based cotransfection/reporter assay ( Figure 1F vs Figure 1G ) (SLU-PP-332: ERR α EC 50 = 98 nM, ERR β = 230 nM, ERR γ = 430 nM vs GSK4716: ERR α EC 50 > 5000 nM, ERR β > 5000 nM, ERR γ = 1200 nM). SLU-PP-332 displayed a degree of ERR α selectivity with 4.4-fold selectivity for ERR α over ERR γ and 2.3-fold ERR α over ERR β . SLU-PP-332 also displayed activity at all ERRs in a Gal4-ERR LBD chimeric cotransfection assay, and similar to the full-length ERR cotransfection assays, SLU-PP-332 was more potent at ERR α than ERR β and ERR γ ( Figure S3A ). SLU-PP-332 was selective for the ERRs as it did not alter the activity of either ER α or ER β , or other nuclear receptors in cotransfection assays ( Figure S3B ). Direct binding of SLU-PP-332 to ERR α was confirmed by limited proteolysis, where the LBD is subjected to chymotrypsin in the presence and absence of SLU-PP-332, and the ability of the drug to “protect” fragments of the LBD from digestion due to a conformational change in the LBD is detected. 24 – 26 As shown in Figure S3C , SLU-PP-332 dosedependently protects a fragment of the ERR α from protease digestion, consistent with direct binding of the drug to ERR α . Direct binding of SLU-PP-332 to ERR γ was also confirmed using differential scanning fluorimetry, where the compound dose-dependently increased the thermal stability of the purified ERR γ LBD ( Figure S3D ). Unfortunately, ERR α did not function well in the differential scanning fluorimetry assay, and ERR γ did not function in the limited proteolysis assay, and thus we were unable to obtain biochemical data from these two receptors in the identical biochemical assay. Although it is not clear why both receptors did not function in the same biochemical assay, we have observed for a number of nuclear receptors that they do necessarily function well in these biochemical assay

We sought to determine if SLU-PP-332 could potentially be used as a chemical probe to evaluate the activation of ERR function in vivo ; thus, we first assessed in vivo exposure after intraperitoneal (i.p.) administration in mice ( Figure 2A ). Mice were administered SLU-PP-332 (30 mg/kg, i.p.), and plasma and muscle were collected 2- and 6 h post administration and analyzed by mass spectrometry. Two hours after administration, levels of SLU-PP-332 were highest in skeletal muscle (∼0.6 μ M), while levels in the plasma were lower (∼0.2 μ M) ( Figure 2A ). We observed no overt toxicity in mice administered SLU-PP-332 (50 mg/kg b.i.d., i.p.) for 10 days which is consistent with the normal complete blood count and electrolyte levels 33 ( Table S1 ). We also observed no significant alterations in serum creatine kinase, suggesting a lack of skeletal muscle toxicity ( Figure S5 ). We next examined the effects of chronic SLU-PP-332 treatment on muscle physiology and function. Three-month-old C57BL/6J mice were administered SLU-PP-332 for 15 days (50 mg/kg, b.i.d., i.p.), followed by an examination of quadricep muscle histology. To increase drug exposure in the efficacy studies experiment, we increased the dose to 50 mg/kg. To avoid the effect of ERR on facultative thermogenesis and cold tolerance, 34 we maintained the mice at thermoneutrality (30 °C). Histology was performed on unfixed muscle and stained for hematoxylin and eosin ( Figure S6A ) and succinate dehydrogenase (SDH) activity ( Figure 2B ). Mice treated with SLU-PP-332 displayed a more oxidative muscle phenotype (greater SDH staining) ( Figure 2B ). We also assessed expression of key proteins within the oxidative phosphorylation complexes and observed an increase in complex I (NDUFB8) and complex V (ATP5A) expression in the gastrocnemius muscle in response to SLU-PP-332 treatment ( Figure 2C ). Consistent with these observations, we also observed an increase in cytochrome c protein expression in response to SLU-PP-332 treatment in the gastrocnemius muscle ( Figure 2D ). Sections were also stained for laminin and consistent with an increased oxidative phenotype as the myofibers were smaller in diameter 35 in SLU-PP-332-treated mice ( Figure 2E ). Electron micrographs of the quadriceps display an increase in mitochondria content in muscles from drug-treated mice compared to vehicle-treated ( Figure 2F ). Mitochondrial DNA concentrations in the skeletal muscle were also increased (relative to nuclear DNA) consistent with an increase in mitochondrial number ( Figure 2G ). We also noted that SLU-PP-332 treatment resulted in increased oxidative type IIa muscle fibers ( Figure 2H , I ) consistent with the increased expression of Myosin IIA protein by western blot ( Figure 2J ). This observation aligned with an increase in expression of Myh6 , which encodes a myosin heavy chain subtype that is associated with type IIa muscle fibers 36 ( Figure 2K ). These data indicate that pharmacological activation of ERRs leads to an increased oxidative capacity of skeletal muscle and an increase in type IIa muscle fibers and suggested that treatment of mice with SLU-PP-332 may lead to an increase in exercise endurance. To assess this, we treated sedentary mice with SLU-PP-332 or vehicle for 7 days (b.i.d., i.p. 50 mg/kg) and subjected them to exercise until exhaustion on a rodent treadmill. Plasma glucose levels were monitored following the exercise to confirm exhaustion ( Figure S6B ). As shown in Figure 2L , mice treated with the ERR agonist were able to run ∼70% longer and ∼45% further than vehicle-treated mice. In a separate study, where sedentary mice were treated in an identical manner except longer duration (2 weeks), we noted an increase in grip strength as well ( Figure 2M ). In order to assess alterations in gene expression due to drug treatment, we prepared RNA/cDNA from gastrocnemius and quadricep muscles from mice treated with SLU-PP-332 or vehicle (50 mg/kg, b.i.d., i.p., 10 days). Muscles were obtained 3 h post the final administration, and global changes in gene expression were assessed by RNA-Seq. We observed 442 differentially expressed genes (DEGs) in the quadricep muscle vs 238 DEGs in the gastrocnemius muscle (FDR < 0.05, |fold change (FC)| > 1.5) ( Figure 3A , B and Supporting Information Part 2 ). There was a significant overlap between the quadricep and gastrocnemius muscle DEGs, as shown in Figure 3C ( p < 2.2 × 10 −16 by Fisher’s exact test). The top 10 pathways (KEGG and Wikipathway) upregulated and downregulated by SLU-PP-332 treatment in either muscle type are shown in Table S2 . We observed substantial overlap in those pathways between muscle types that were significantly ( p < 0.05) modulated. Pathways that were significantly SLU-PP-332 upregulated in both muscle types are shown in Figure 3D , E . Interestingly, these pathways are quite distinct from those identified as downregulated in the skeletal muscle ERR α / γ double KO mice, w

The ERRs play important roles in the regulation of energy metabolism and fuel selection. Loss of ERR α or ERR γ function leads to reduced muscle oxidative function and reduced functional endurance; 16 , 37 , 56 thus, pharmacological activation of these receptors may lead to beneficial metabolic effects associated with increased skeletal muscle activity for the treatment of metabolic diseases. In this study, we characterize the ability of a pan ERR α / β / γ synthetic agonist with ∼4-fold ERR α selectivity over ERR γ (SLU-PP-332) to function as an exercise mimetic and improve muscle and metabolic function both in vitro and in vivo . The key novelty of this compound is the ability to activate ERR α as well as the pharmacokinetic properties sufficient for in vivo studies. However, it is important to note that our studies with this compound were performed under conditions where all three ERRs would be activated, and this compound does not have sufficient specificity to probe the individual functions of ERR paralogs. Given the unique pharmacological activity of this compound to target activation of ERR α , we employed a genetic model combined with SLUPP-332 to identify an ERR α -dependent pathway that is linked to a previously identified acute aerobic exercise genetic response. We found that SLU-PP-332 treatment induces the expression of DDIT4 via specific activation of ERR α . DDIT4 is a key protein that is induced after short bouts of aerobic exercise that is responsible for inducing an acute aerobic exercise genetic program that leads to a range of physiological adaptations to exercise. 48 We found that Ddit4 is a direct ERR α target gene, and previous data indicating that Ddit4 null mice display reduced exercise endurance 49 is consistent with our results, indicating that SLU-PP-332 treatment, which induces Ddit4 expression, enhances exercise endurance. The array of genes that are induced by both SLU-PP-332 and acute aerobic exercise have been linked mechanistically to improved exercise endurance, increased fatty acid oxidation, and/or improved metabolic efficiency, which are all physiological components of the adaptive response to exercise. Of course, the key gene we examined, Ddit4 , is associated with mitochondrial function and improved exercise endurance, 49 and another gene within this program we examined, Slc25a25 , is also similarly associated with these endpoints. 51 Most importantly, the effects of SLU-PP-332 on exercise endurance are dependent on ERR α as mice with muscle-specific deletion of this receptor are refractory to improved performance. It must be reinforced that the pharmacodynamic properties of SLU-PP-332 are such that at the dose we used to assess ERR activation of all three ERR paralogs would be targeted. Previously, it had been reported that skeletal muscle overexpression of ERR γ in mice led to improved endurance; 17 however, we noted substantial differences in the array of genes regulated by overexpression of ERRs compared to pharmacological activation of ERRs. The acute aerobic exercise gene program was not identified when either ERR α or ERR γ was overexpressed or knocked out. 9 , 12 We believe that transient activation of ERRs may be quite distinct than either chronic overexpression or complete loss of the receptor. Furthermore, overexpression of ERR(s) likely provides for an unremitting level of elevated transcriptional activity that cannot be mimicked by pharmacological activation of this receptor class that already displays strong constitutive transactivation activity. Multiple bouts of aerobic exercise (2 h/day for 8 days on a rodent treadmill) have been shown to induce ERR α (∼1.5-fold) and ERR γ (∼2.1-fold) expression within the gastrocnemius muscle in mice. 18 Short-term aerobic exercise (cycling) in humans has been shown to induce ERR α (mRNA and protein expression) in skeletal muscle (m. vastus lateralis) to a similar extent (1.5–2-fold). 6 These data suggest that the ∼2-fold increase in the ERR transcriptional activity that we observe with SLU-PP-332 treatment is likely more similar to physiological changes in ERR activity induced by exercise than experimental models of skeletal muscle overexpression of ERR(s) or VP-16 ERR fusion proteins. Thus, pharmacological activation of ERR may be more closely aligned with driving physiological changes that are similar to normal exercise adaptation such as induction of the acute aerobic exercise response rather than chronic overexpression of ERR or similar key regulator proteins. In summary, activation of ERR α by SLU-PP-332 as an exercise mimetic induces an acute aerobic exercise program that leads to an array of physiological adaptations that are associated with exercise, including increased o x idative fibers in a muscle, increased fatty acid o x idation, and enhanced exercise endurance. Several nuclear receptors, such as LXR, FXR, PPAR α , PPAR δ , PPAR γ , and REV-ERB, among others, have been evaluated or utilized as targets f

(E)-4-Hydroxy-N ′ -(naphthalen-2-ylmethylene)benzohydrazide. To a solution of 2-naphthaldehyde (1.0 g, 6.6 mmol) in toluene (100 mL) was added 4-hydro x ybenzohydrazide (1.1 g, 6.6 mmol) portion wise. The mi x ture was allowed to stir for 18 h at reflu x . A solid precipitated, which was recrystallized from a 1:9 mi x ture of methanol and ether to obtain the title compound as a white solid (1.3 g, 68%); 1 H NMR (400 MHz, DMSO- d 6 ) δ 11.81 (s, 1H), 10.18 (s, 1H), 8.63 (s, 1H), 8.12 (s, 1H), 8.06–7.84 (m, 6H), 7.54 (dp, J = 6.5, 3.5 Hz, 2H), 6.93 (dd, J = 8.8, 2.3Hz, 2H). 13 C NMR (101 MHz, DMSO- d 6 ) δ 162.96, 160.81, 146.88, 133.68, 132.93, 132.33, 129.81, 128.49, 128.30, 127.79, 127.03, 126.73, 123.95, 122.74, 115.11. High-resolution mass spectrometry (HRMS) calculated for C 18 H 15 N 2 O 2 (M + H) + : 291.11280, found: 291.11284.

As previously described, 67 – 69 HEK293 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum at 37 °C under 5% CO 2 . Twenty-four hours prior to transfection, HEK293 cells were plated in 96-well plates at a density of 2 × 10 4 cells/well. GAL4-NR-LBD or FLAG-ERR-FL plasmids were used in the luciferase assay.

Bioenergetics profile tests in C2C12 myoblasts were conducted, as described by Nicholls et al. 70 The day before (24 h) the assay, C2C12 cells were seeded (10,000/well) in growth media on the 96-well XF Flu x Analyzer (Seahorse) cell plate.

Fresh cryo-sections (10 μ m) were incubated for 1 h with Mouse on Mouse (M.O.M, Vector Lab) incubation media and then incubated with BA–D5, SC71, or BF-F3 antibodies (Developmental Studies Hybridoma Bank) for 45 min at 37 °C, in PBS–1% bovine serum albumin (BSA). Sections were washed three times for 5 min in PBS and then incubated with secondary antibodies diluted in PBS–1% BSA for 30 min at 37 °C. Sections were washed three times for 5 min in PBS and mounted using ProLong Gold mounting media (Thermo Fisher) under glass coverslips. Fresh cryo-sections (10 μ m) were incubated for 30 min at 37 °C in incubation medium (50 mM phosphate buffer, sodium succinate 13.5 mg/mL, NBT 10 mg/mL in water) placed in a Coplin Jar, and then the section was rinsed in PBS. After staining sections were fi x ed in 10% formalin–PBS solution for 5 min at room temperature (RT) and then rinsed in 15% alcohol for 5 min. Slides were mounted with an aqueous medium and sealed. Cryostat sections (10 μ m) were fi x ed for 20 min in 3% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4. Sections were blocked with 8% BSA in PBS-1 h at RT and then incubated at 4 °C overnight with primary antibody for laminin at a 1:200 dilution. Sections were then washed and incubated with anti-rabbit-FITC (1:1000) for 1 h at room temperature. Sections were mounted with fluorescent mounting medium containing Dapi (Vector lab) under glass coverslips. All quantifications were performed using ImageJ software.

For all experiments, 8–10 male C57BL6/J mice per group (12 weeks of age for chow) were administered a dose of SLU-PP-332 50 mg/kg (i.p., b.i.d.) or vehicle for 28 or 12 days. At termination of the experiment, tissues were collected for gene expression analysis by real-time qPCR using methods previously described. Food intake and body weight were monitored daily in these experiments, and body composition was measured prior to initiation and termination of the experiments by NMR (Bruker BioSpin LF50). Plasma was collected for triglyceride and cholesterol measurements. All b.i.d. dosing was performed, with dosing occurring at CT0 and CT12.

Skeletal muscle-specific KO mice, ERR α fl/fl and ERR α −/− mice (20–24 week old, 30.7 + 0.26 g b.w.) were segregated into vehicle-treated or SLU-PP-332-treated groups ( n = 5/group). Mice were administered (i.p.) vehicle (10% DMSO, 15% Kolliphor EL in sterile saline) or 25 mg/kg SLU-PP-332 for 15 days. Prior to run performance, trial mice were acclimated to the treadmill (Columbus Instruments E x er 3/6 motorized treadmill) for 3 days (10 min at 10 m/min, then 2 min at 15 m/min). To assess aerobic run performance, mice were run at 10, 12.5, and 15 m/min for 3 min at each speed, after which the speed was increased by 1 m/min every 2 min (max speed 28 m/min) until exhaustion. Basal and post-run blood lactate readings were read to confirm exhaustion. Mice were sacrificed, and hindlimb muscles were collected 24 h after the run performance test.

Pharmacokinetic studies of SLU-PP-332 in mice were performed, as previously described. 72 Three-month-old C57Bl6/J male mice ( n = 3) were injected (i.p.) at ZT 1 with 30 mg/kg of SLU-PP-332 (5% Tween–5% DMSO–90% PBS). The animals were sacrificed by CO 2 asphyxiation, and tissues were collected at 1, 2, or 4 h after administration of the compound ( n = 4 per time point). Plasma and tissues (liver, quadriceps, and brain) were collected, flash-frozen, and stored at −80 °C until analysis. Tissue samples were weighed and placed into Eppendorf tubes. NaÏve tissue was used to prepare standard curves in the muscle tissue matri x . To each sample or standard tube were added three to five stainless steel beads (2–3 mm) and the appropriate volume of cold 3:1 acetonitrile/water (containing 100 ng/mL extraction internal standard SR8278) 73 to achieve a tissue concentration of 200 mg/mL. Tubes were placed in a bead beater for 2–3 min. Samples and standards (100 μ L) were plated in a 96-well plate, 150 μ L of acetonitrile was added to each well and then centrifuged at 3200 rpm for 5 min at 4 °C. The supernatant (100 μ L) was transferred to a 96-well plate, evaporated to dryness under nitrogen, reconstituted with 100 μ L of 0.1% v/v formic acid in 9:1 water/acetonitrile, and vortexed for 5 min. Plasma samples or standards prepared in a plasma matri x (100 μ L) were added to a 96-well plate. To each well, 400 μ L of cold acetonitrile containing 100 ng/mL extraction internal standard SR8278 was added. The plate was vortexed for 5 min at 4 °C. The supernatant (300 μ L) was transferred to a second 96-well plate, evaporated to dryness under nitrogen, reconstituted with 100 μ L of 0.1% v/v formic acid in 9:1 water/acetonitrile, and vortexed for 5 min. Finally, to each reconstituted tissue or plasma sample, 10 μ L of 1000 ng/mL enalapril in acetonitrile was added as an injection internal standard, and the 96-well plate was vortexed, briefly centrifuged, and submitted for LC/MS analysis. SLU-PP-332 concentrations were determined on a Sciex API-4000 LC/MS system in positive electrospray mode. Analytes were eluted from an Amour C18 reverse phase column (2.1 × 30 mm 2 , 5 μ m) using a 0.1% formic acid (aqueous) to 100% acetonitrile gradient mobile phase system at a flow rate of 0.35 mL/min. Peak areas for the mass transition of m / z 291 > 121 for SLU-PP-332, m / z 394 > 189 for the extraction internal standard SR8278, and m / z 376 > 91 for the injection internal standard enalapril were integrated using Analyst 1.5.1 software. Peak area ratios of SLU-PP-332 area/SR8277 area were plotted against concentration with a 1/ x -weighted linear regression. Enalapril was used to monitor proper injection signals throughout the course of LC/MS analysis.

In vitro translated ERR α full-length (TNT kit; Promega) was used. Briefly, after incubating at room temperature for 15 min with ligands (1–5–10 μ M), receptor proteins were digested at room temperature for 10 min with 10 μ g/mL trypsin. The proteolytic fragments were separated on a 4–15% sodium dodecyl sulfate (SDS) polyacrylamide gel (Bio-Rad) and visualized by Coomassie Blue staining.

Mitochondrial DNA was extracted using QiAamp DNA mini kit, following the manufacturer’s instructions (Qiagen). DNA was quantified using Sybr Select Master Mi x (Applied Biosystems). All samples were run in duplicate, and the analysis was completed by determining ΔΔCt values. The reference gene used was NRDUV1, a genomic DNA marker.

OXPHOS protein expression was assessed using the Thermo Fisher O x Phos Rodent WB Antibody Cocktail (#45–8099). In order to normalize for protein loading, we used the Stain-Free Western Workflow suite from Bio-Rad ( https://www.bio-rad.com/webroot/web/pdf/lsr/literature/Bulletin_RP0051.pdf ). Image Lab 4.0 software, a component of ChemiDoc MP (Bio-rad), was employed, and the relative amount of total protein in each lane on the blot was calculated and used for quantitation normalization. We then identified a band for each OXPHOS complex and then normalized their intensity level to the total protein intensity for each lane. Then, the average and standard deviation were calculated for each group (vehicle and SLU-332).

Supplementary Table 1 Supplementary Table 2 Supplementary Figures