Calcium chloride, cadmium chloride, and glucagon were purchased from Sigma-Aldrich. 4-Hydroxyn tamoxifen, ICI-182,780, 2-APB, DY131, and GSK4716 were purchased from Tocris.

HEK293T cells (RRID: CVCL_0063) were a gift from Dr. Anna Riegel, Georgetown University. HepG2 cells (RRID: CVCL_0027) and MCF-7 cells (RRID: CVCL_0031) were acquired from Tissue Culture and Biobanking Shared Resource (Georgetown University). For cell passaging, HEK293T cells were maintained in improved minimal essential medium (IMEM; Corning) containing 10% fetal bovine serum (FBS; Sigma-Aldrich). HepG2 and MCF-7 cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM; Corning) containing 10% FBS. For the assays, 5 × 10⁵) cells were plated in a six-well plate with 3 mL of lipoic acid-free IMEM (Crystalgen or VitaScientific) containing 5% charcoal-stripped calf serum (CCS; Sigma-Aldrich) for 48 h. The serum was diluted to 1% CCS. To increase the concentration of intracellular calcium, cells were treated with 1–10 mM calcium to activate calcium-sensing receptors. To obtain an intracellular concentration of cadmium similar to calcium, cells were treated with 5–10 µM cadmium. Cells were also treated with 100 nM glucagon, 1 µM DY131, or 1 µM GSK4716. All treatments were done in the absence or presence of 5 µM 4-hydroxy tamoxifen, 50 µM 2-APB, or 1 µM ICI-182,780 for 3 to 48 h.

Following treatment, RNA was isolated using Trizol (Life Technologies). The reverse transcription reaction was performed using Maxima™ H Minus cDNA Synthesis Master Mix with dsDNase (ThermoFisher Scientific). qPCR assays were performed using Taqman probes (ThermoFisher Scientific), SsoAdvanced Universal Probes supermix (Bio-Rad), or PowerUp™ SYBR™ Green Master Mix (ThermoFisher Scientific). qPCR results were presented as fold change compared to control by using the 2^−ΔΔCt method and normalizing to the ribosomal protein P0 (RPLP0) mRNA (Taqman assays) or β-galactosidase mRNA (SyBr Green assay) ( Table 1 ).

HepG2 or HEK293T cells were plated in T75 flasks in IMEM with 5% CCS for 2 days and then cells were treated with 100 nM glucagon or 1–10 mM calcium in the absence or presence of 50 µM 2-APB for 3 to 24 h in IMEM with 1% CCS. After treatment, cells were trypsinized and washed with Hanks’ balanced salt solution (HBSS; Gibco) twice. Fluo-4 AM (15 µL; Invitrogen) was dissolved in 3 mL of HBSS and mixed with cells at 37°C for 30 min. Then, cells were washed with HBSS twice. Cells (5 × 10⁵) were used for fluorescence measurement by using the fluorimeter (Photon Technology International). The excitation wavelength was 494 nm and the emission wavelength was 516 nm. The first reading was labeled as F. Cells were then treated with 20 µL of Triton X-100 (Boston Bioscience) and the reading was labeled as F _max. Cells were then treated with 40 µL of EGTA (Promega) and the reading was labeled as F _min. The calcium concentration was calculated as follows: C a = K d   ×   ( F − F min ) ÷ ( F max − F ) , where K _d is the affinity of the dye for calcium and its value is 335.

HepG2 cells were cultured in 100-mm dishes in IMEM with 5% CCS for 2 days and then treated with 100 nM glucagon, 10 mM calcium, or 5 µM cadmium for 3 h in IMEM with 1% CCS. Cells were then washed with phosphate-buffered saline (PBS) and lysed by radioimmunoprecipitation assay (RIPA) buffer. After removing cell debris by centrifuge, the protein concentration was determined by the Bio-Rad Protein Assay Dye Reagent (Bio-Rad Laboratories). Cell lysates were denatured by boiling and ran on 10% SDS-PAGE gel under reducing conditions, transferred to a 0.45-µm nitrocellulose membrane, blocked by 5% non-fat dry milk in PBS-Tween, and incubated with anti-PEPCK1 antibody (A2036, ABclonal, RRID: AB_2764060, 1:500), anti-G6PC antibody (ab93857, Abcam, RRID: AB_10903775, 1:200), anti-solute carrier family 2 member 2 (SLC2A2) antibody (ab192599, Abcam, 1:1,000), or anti-actin antibody (MAB1501, Millipore Sigma, RRID: AB_2223041, 1:2,000) at 4°C overnight and followed by anti-rabbit IgG (SA00001-2, Proteintech, RRID: AB_2722564, 1:5,000) or anti-mouse IgG (5220-0341, SeraCare, RRID: AB_2891080, 1:10,000) for 1 h at room temperature. The target protein signals were visualized using the Western Lightening^® Chemiluminescence Reagent Plus (Perkin-Elmer). The protein amount was measured by ImageJ and presented as fold change (± SEM) compared to control and normalized to the level of actin.

HEK293T cells (3 × 10⁶) cells were plated in 150-mm culture dishes in IMEM with 5% CCS for 3 days. The HEK293T cells were then transfected with 25 µg of pSG5-HA-ERRγ or pCMX-hERRγ for 24 h. Cells were treated with 10 mM calcium or 10 µM cadmium in IMEM with 1% CCS for 1 h. HepG2 cells were grown in the same condition for 2 days, then treated with 100 nM glucagon, 10 mM calcium, or 10 µM cadmium in IMEM with 1% CCS for 1 h. Cells were fixed with 1% formaldehyde (ThermoFisher Scientific) for 5 min at room temperature, quenched by 125 mM glycine (Sigma-Aldrich) for 5 min, and washed with ice-cold PBS twice. Cells were collected by scraping and centrifugation, and then treated with nuclei isolation buffer (50 mM Tris-Cl, 60 mM KCl, 0.5% NP40, and 10 mM DTT) for 15 min on ice followed by centrifugation at 430 rcf for 10 min to pellet the nuclei. Isolated nuclei were then lysed with nuclei lysis buffer (EMD Millipore) on ice for 10 min, followed by flash-freezing in liquid nitrogen and thawing at room temperature to facilitate lysis. Chromatin was then sheared with the Bioruptor^® Plus sonicator (Diagenode) until the size was reduced to 300–500 bp. The soluble chromatin was then subjected to immunoprecipitation using anti-HA tag antibody (ab1424, Abcam, RRID: AB_301017) for PSG5-HA-ERRγ, and anti-ERRγ antibodies (pp-h6812, R&D, RRID: AB_604965) for pCMX-hERRγ or IgG (ab18413, Abcam, RRID: AB_2631983). Antibody/chromatin complexes were captured by Protein A/G magnetic beads (Thermofisher) on a magnetic rack, washed sequentially with low-salt buffer (EMD Millipore), high-salt buffer (EMD Millipore), LiCl buffer (EMD Millipore), and Tris-EDTA (Sigma-Aldrich), and eluted in chromatin immunoprecipitation assay (ChIP) elution buffer (1% SDS, 100 mM NaHCO₃). Cross-link was reversed in 0.54 M NaCl overnight at 65°C. The samples were then treated with RNase (Fisher Scientific) and purified by the PCR purification kit (QIAGEN). DNA regions of interest were amplified by RT-PCR. To determine the potential ERRE sites of target genes, the UCSC Genome Browser was used. TNAAGGTCA as ERREs sequence template was used to find the potential ERRE of PEPCK1, G6PC, PDK4, and SLC2A2. The PDK4 ChIP primer sequence was previously published, and its ERRE location sequence was also verified in the UCSC Genome Browser (41).

HepG2 cells or MCF-7 cells were cultured on 18-mm circular cover slides in a 12-well plate in IMEM with 5% CCS for 2 days. For HepG2 cells, the serum was diluted to 1% on day 3 and cells were treated with 100 nM glucagon or 10 mM calcium for 3 h. For MCF-7, cells were grown in IMEM with 1% CCS for 24 h. Cells were then washed with PBS, fixed in 10% buffered formalin (Fisher Scientific) for 20 min and permeabilized using 0.5% Triton X-100 in PBS buffer for 5 min. Cells were then incubated in PBS containing 3% bovine serum albumin at room temperature for 40 min. Cells were treated with 1 µg/mL anti-ERRγ antibodies at 4°C overnight and then with anti-mouse polyclonal antibody (Alexa Fluor 594, Invitrogen, RRID: AB_2313921, 1:500) at room temperature for 60 min. F-actin was stained with Phalloidin-iFluor 488 Conjugate (AAT Bioquest, 1:500) for 60 min at room temperature. The nuclei were stained by 1 μg/mL 4,6-diamidino-2-phenylindole (DAPI, Boehringer Mannheim) for 5 min. The images were captured by Leica SP8.

HepG2 cells were cultured in IMEM with 5% CCS for 2 days and then treated with 7.5 mM calcium, 10 mM calcium, 5 µM cadmium, or 10 µM cadmium in IMEM with 1% CCS for 24 h. Glucose concentration in the medium was measured by Glucose Assay Kit-WST (G264-05, dojindo).

PSG5-HA-ERRγ (ERRγ splice variant 1) was a gift from Dr. Rebecca Riggins, Georgetown University (42). pCMX-hERRγ (ERRγ splice variant 3) was a gift from Dr. Hongwu Chen, University of California, Davis (43). pCMV-GAL4 plasmid containing GAL4, pEQ176 plasmid containing β-galactosidase, and pG6 (5’Pro) plasmid containing chloramphenicol acetyltransferase (GAL-4-CAT) were purchased from Addgene. The amino acid sequences of the LBDs of ERRγ-1, -2, and -3 are identical. Because the sequences of the LBDs are the same, the LBD was subcloned from the ERRγ-1 construct. To create the GAL-4-ERRγ-LBD construct, the hinge region and LBD (L187-V458) of ERRγ from pSG5-HA-ERRγ was fused with the GAL-4 DNA binding domain from pCMV-GAL4 to form GAL-4-ERRγ-LBD. Restriction enzyme cut sites SacI were created in the 5′ terminus and XbaI was created in the 3′ terminus. Then, corresponding enzymes were used to fuse GAL-4-ERRγ-LBD with the empty pCMV vector. To create GAL-4-ERRγ-LBD mutants, mutation primers were designed using the QuikChange Primer Design of Agilent and synthesized by Integrated DNA Technologies. The GAL-4-ERRγ-LBD plasmid was mutated using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent), and results were verified by sequencing. For the transfection assay, cells were transfected with 2.5 μg of pSG5-HA-ERRγ or pCMX-hERRγ using Trans-IT LT-1 (MirusBio) or SuperFect (QIAGEN) for 24 h. For the CAT reporter assay, cells were transfected with 2 μg of wild-type GAL-4-ERRγ-LBD or GAL-4-ERRγ-LBD mutant, 2 μg of pG6 (5’Pro), and 1 μg of pEQ176 using Trans-IT LT-1 for 24 h.

Raw microarray data (accession no. GSE136595) were downloaded from Gene Expression Omnibus. Analysis of microarray data was performed by gene set enrichment analysis (GSEA) v4.1.0 for Windows. The hallmark gene sets and ontology gene sets from the Molecular Signatures Database were used in our analysis.

The structure of ERRγ was obtained from Protein Data Bank (ID: 2GP7). The A chain of 2GP7 was then submitted to the Metal Ion-Binding Site Prediction and Docking Server (MIB; http://bioinfo.cmu.edu.tw/MIB/) or Autodock Vina to predict the corresponding calcium docking site (44, 45). All molecular dynamics simulations were performed by using QwikMD in VMD and the CHARMM36 force field. During the molecular dynamics simulations, the temperature of 37°C was set over a period of 1 ns. Resolution was set as Max, then solvent was set as Explicit, and other parameters were set based on the instruction of QwikMD. All analyses were performed by using the VMD plug-in (46). Graphs were created by PyMOL.

All statistical analyses were performed in Prism. Data are presented as the mean ± standard error of the mean (SEM). Statistical differences were evaluated by one-way analysis of variance (ANOVA) followed by Fisher’s LSD test or t-test. Statistical significance is defined as a p-value of ≤0.05. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001.

To determine whether calcium can mediate the cross-talk between glucagon and ERRγ, HepG2 cells were treated with glucagon (100 nM) or calcium (7.5 or 10 mM) in the absence or presence of the ERRγ inhibitor 4-hydroxy tamoxifen (5 μM). Treatment with glucagon for 3 h resulted in an approximately 2.1-, 2.4-, and 1.7-fold increase in the expression of PEPCK1, G6PC, and SLC2A2, respectively, which was significant and blocked by the ERRγ antagonist 4-hydroxy tamoxifen ( Figures 1A–C ). There was no significant effect on the expression of PDK4, which is not part of the gluconeogenesis pathway (data not shown). Similar to treatment with glucagon, treatment with calcium for 3 h significantly increased expression of PEPCK1, G6PC, and SLC2A2 by 2.6-, 2.3-, and 1.9-fold, respectively, which was blocked by 4-hydroxy tamoxifen. Treatment with calcium for 24 h resulted in a significant increase in the expression of PEPCK1, G6PC, and PDK4 by approximately 1.8-, 4.1-, and 5.6-fold, respectively ( Figures 1E–H ). 4-Hydroxy tamoxifen blocked the effects of calcium on PEPCK1 and PDK4 but failed to block the effects of calcium on G6PC and increased its expression. There is some evidence suggesting that 4-hydroxy tamoxifen can influence metabolism, which may be due to its effect on the expression of G6PC (47, 48).

Glucagon increases intracellular calcium through the inositol 1,4,5 trisphosphate receptor (IP3R)-mediated release of calcium from the endoplasmic reticulum, whereas high levels of extracellular calcium can increase intracellular calcium through the activation of calcium receptors and channels (49, 50). To determine whether intracellular calcium mediates the effects of glucagon and extracellular calcium on the activity of ERRγ, the concentration of intracellular calcium was measured following treatment with glucagon (100 nM) in the absence or presence of the IP3R inhibitor 2-APB (50 μM). The concentration of intracellular calcium was also measured following treatment with extracellular calcium (7.5 or 10 mM) ( Figure 1D ). As expected, treatment of HepG2 cells with glucagon for 3 h increased the concentration of intracellular calcium from approximately 300 (± 18) nM to 456 (± 51) nM that was blocked by 2-APB (196 ± 24 nM). Treatment with increasing concentrations of calcium for 3 h also increased the concentration of intracellular calcium to approximately 481 (± 50) to 602 (± 99) nM and treatment with increasing concentrations of calcium for 24 h further increased intracellular calcium concentration from 315 (± 53) nM to approximately 642 (± 85) to 797 (± 78) nM.

To determine whether cadmium mimics the effects of calcium on the activation of ERRγ and disrupts the glucagon signaling pathway, HepG2 cells were treated with cadmium (5 μM) for 24 h. Treatment with cadmium resulted in a significant increase of approximately 3.0-, 3.5-, 1.9-, and 8.1-fold in the expression of PEPCK1, G6PC, SLC2A2, and PDK4, respectively ( Figures 1I–L ). Similar to glucagon and calcium, 4-hydroxy tamoxifen blocked the effects of cadmium on PEPCK1, SLC2A2, and PDK4 but not the effect on G6PC.

To determine whether treatment with glucagon, calcium, or cadmium increased protein expression, the amount of PEPCK1, G6PC, and SLC2A2 was measured by Western blot analysis. Similar to the increase in mRNA, treatment with glucagon resulted in a significant 2.5- and 2.1-fold increase in PEPCK1 and SLC2A2, respectively. Treatment with calcium resulted in a 2.8-, 2.3-, and 3.0-fold increase in PEPCK1, G6PC, and SLC2A2, respectively, and treatment with cadmium resulted in a 3.5- and 3.6-fold increase in G6PC and SLC2A2, respectively, which were significant ( Figures 1M–P ).

To determine whether treatment with calcium and cadmium increased the concentration of glucose, the amount of glucose in the medium was measured. Treatment of HepG2 with calcium or cadmium for 24 h also significantly increased the concentration of glucose in the medium from 18 (± 1) mM to approximately 22 (± 2) to 25 (± 3) mM ( Figure 1Q ). Taken together, the results suggest that glucagon increases the expression of genes involved in gluconeogenesis through an increase in intracellular calcium and the subsequent activation of ERRγ and metals that mimic calcium increase the activity of ERRγ.

To further show that the change in gene expression is due to the activation of ERRγ by glucagon (100 nM), calcium (7.5 or 10 mM), and cadmium (5 or 10 μM), the recruitment of ERRγ to the enhancer regions of PEPCK1, G6PC, SLC2A2, and PDK4 was measured by a ChIP-qPCR assay. ERRγ is recruited to the ERR response element (TNAAGGTCA) in the enhancer regions of target genes and regulates gene expression. Through database searches, potential ERREs were identified in the enhancer region of the target genes ( Figures 2A–D ). Treatment with glucagon resulted in a significant enrichment of approximately 4.8-, 7.3-, and 6.0-fold of ERRγ on the enhancer regions of PEPCK1, G6PC, and SLC2A2, respectively. Treatment with calcium resulted in an approximately 3.7-, 3.7-, 4.7-, and 3.2-fold enrichment to the enhancer regions of PEPCK1, G6PC, SLC2A2, and PDK4, respectively, and treatment with cadmium resulted in an approximately 3.7-, 6.5-, and 2.6-fold enrichment to the enhancer regions of G6PC, SLC2A2, and PDK4, respectively, which were significant.

To determine the effects on the localization of ERRγ, HepG2 cells were treated with glucagon or calcium and the effects on ERRγ were measured by immunofluorescence. In control cells, ERRγ was localized to the periphery of the cell nucleus. Following treatment with glucagon or calcium, ERRγ re-localized in the nucleus ( Figure 2E ). Taken together, the results suggest that glucagon, calcium, and cadmium activate and recruit ERRγ to the ERRE in the enhancer regions of genes involved in gluconeogenesis.

Mutational analysis and molecular modeling show that calcium interacts with the LBD of ERα to activate the receptor (37). The high homology between the LBDs of ERα and ERRγ suggests that the LBD of ERRγ contains potential calcium and cadmium interaction sites. ERRγ-1 is the canonical isoform of the receptor and shares the same LBD with other ERRγ splice variants. The LBDs of the splice variants ERRγ-1, -2, and -3 are identical. Calmodulin binds to ERRγ through two binding motifs (51). One calmodulin binding motif is located at the boundary between the N-terminus and DNA binding domain, and the other binding motif is located at the boundary of the hinge region and the N-terminus of the LBD. Both motifs are needed for calmodulin binding to ERRγ. To address the mechanism by which calcium and cadmium activate ERRγ, HepG2 cells were transfected with a chimeric receptor that contains the DNA binding domain of GAL4 and the LBD of ERRγ (GAL-4-ERRγ-LBD) and a CAT reporter construct. Following transfection, the cells were treated with glucagon (100 nM), calcium (7.5 or 10 mM), or cadmium (5 μM) in the absence or presence of the IP3 receptor inhibitor 2-APB (50 μM) or 4-hydroxy tamoxifen (5 μM) ( Figures 2F–H ). Treatment with glucagon for 3 h resulted in a significant increase of approximately 9.7-fold in the expression of CAT that was blocked by 2-APB and 4-hydroxy tamoxifen. Treatment with calcium for 3 and 24 h resulted in a significant increase of approximately 1.9- to 5.7-fold in the expression of CAT that was also blocked by 4-hydroxy tamoxifen. Similar to calcium, treatment with cadmium for 24 h resulted in a significant 4.5-fold increase of CAT that was blocked by 4-hydroxy tamoxifen. Taken together, the results suggest that calcium mediates the effects of glucagon through the LBD of ERRγ and that cadmium also mimics the effects of calcium through the LBD of ERRγ.

To further demonstrate that calcium and cadmium activate ERRγ, HEK293T cells, which do not express detectable amounts of ERRγ or calcium sensing receptor and channels, were transfected with splice variants of ERRγ. The transfected cells were treated for 12, 24, or 48 hours with calcium (7.5 mM or 10 mM) or cadmium (5 μM or 10 μM) in the absence or presence of 4-hydroxy tamoxifen (5 μM). Transfection was verified by qPCR ( Supplementary Figure 1 ). Similar to HepG2 cells, calcium or cadmium consistently increased the expression of PEPCK1, G6PC, and PDK4 in HEK293T cells transfected with ERRγ1 or ERRγ3. Treatment with calcium resulted in a maximal 3.2-, 4.4-, and 2.3-fold increase in PEPCK1, G6PC, and PDK4, respectively, in cells transfected with ERRγ1 ( Figures 3A–C ) and a maximal 2.5-, 3.8-, and 2.2-fold increase in PEPCK1, G6PC, and PDK4, respectively, in cells transfected with ERRγ3 ( Figures 3G–I ). Treatment with cadmium also resulted in a maximal 2.9-, 5.6-, and 3.2-fold increase in PEPCK1, G6PC, and PDK4, respectively, in cells transfected with ERRγ1 ( Figures 3D–F ) and a maximal 2.2-, 7.7-, and 2.6-fold increase in PEPCK1, G6PC, and PDK4, respectively, in cells transfected with ERRγ3 ( Figure 3J–L ). Treatment with calcium or cadmium resulted in 2.0-fold increase or 4.8-fold increase in SLC2A2 in cells transfected with ERRγ1 but had no effects on cells transfected with ERRγ3 ( Figures 4C, F ). The effects of calcium and cadmium were blocked by 4-hydroxy tamoxifen. ERRγ1 is the canonical isoform of the receptor. Compared to ERRγ1, ERRγ splice variant 3 (ERRγ3) is missing the first 23 amino acids and Y234 is replaced with LLWSDPAD. A diagram of the ERRγ splice variants 1 and 3 is shown in ( Figure 4L ). To determine whether intracellular calcium alters the activity of the ERRγ splice variants, the concentration of intracellular calcium was measured following treatment with calcium (1, 3, 5, and 10 mM) ( Figure 4I ). As expected, treatment with increasing concentrations of calcium increased the intracellular calcium concentration from 191 (±17) nM to 286 (±10) – 630 (±17) nM. Taken together, the results suggest that calcium and cadmium activate different ERRγ splice variants.

To determine whether calcium and cadmium regulate the expression of genes in other metabolic pathways, genes in the glycolysis and pentose phosphate pathways were also measured in HEK293T cells transfected with ERRγ. Depending on the ERRγ splice variant, treatment with either calcium or cadmium had inconsistent effects on other metabolic genes ( Figure 4 and Supplementary Figure 2 ). For example, treatment with calcium decreased hexokinase 2 (HK2) in the glycolysis pathway by approximately 66% or 48% in cells transfected with ERRγ1 or ERRγ3 ( Figures 4A, G ), whereas treatment with cadmium decreased HK2 by approximately 59% in cells transfected with ERRγ1 but had no effect in cells transfected with ERRγ3 ( Figures 4D, J ). The effects of calcium and cadmium were partially blocked by 4-hydroxy tamoxifen. Treatment with calcium also decreased hexose-6-phosphate dehydrogenase (H6PD) by approximately 47% in cells transfected with ERRγ1 and 53% in cells transfected with ERRγ3 ( Figures 4B, H ). Treatment with cadmium also decreased H6PD by an approximately 53% in cells transfected with ERRγ1 and 67% in cells transfected with ERRγ3 ( Figures 4E, K ). However, the effects of calcium and cadmium were not blocked by 4-hydroxy tamoxifen.

To determine whether the effects of calcium and cadmium are specific to ERRγ, non-transfected HEK293T cells were also treated with calcium or cadmium. Treatment with calcium did not change the expression of PEPCK1 or PDK4. Treatment of cadmium did not change the expression of PEPCK1 but increased PDK4 ( Supplementary Figure 3 ). PDK4 is known to be regulated by both ERRα and ERRγ, which may contribute to the increase of PDK4 (52). Non-transfected HEK293T cells have undetectable levels of G6PC and SLC2A2 (data not shown). Taken together, the results suggest that calcium and cadmium can activate different ERRγ splice variants and regulate ERRγ-responsive genes.

To further verify that the change of gene expression was due to the activation of ERRγ by calcium or cadmium, the recruitment of ERRγ to the enhancer regions of PEPCK1, G6PC, PDK4, and SLC2A2 was measured by a ChIP-qPCR assay in HEK293T cells transfected with ERRγ1 or ERRγ3 ( Figures 5A–D ). For HEK293T cells transfected with ERRγ1, treatment with calcium (10 mM) resulted in an approximately 3.4-, 4.3-, 5.3-, and 3.3-fold enrichment and treatment with cadmium (10 µM) resulted in a 2.8-, 2.3-, 2.5-, and 2.5-fold enrichment to the enhancer regions of PEPCK1, G6PC, PDK4, and SLC2A2, respectively. For HEK293T cells transfected with ERRγ3, treatment with calcium resulted in an approximately 4.3-, 3.5-, and 4.4-fold enrichment and treatment with cadmium resulted in a 7.9-, 7.0-, and 8.3-fold enrichment to the enhancer regions of PEPCK1, G6PC, and PDK4, respectively, and the increase was not significant. In non-transfected cells, treatment with calcium or cadmium had no effect on enrichment. The results suggest that calcium and cadmium recruit ERRγ to ERREs in the enhancer regions of target genes and induce gene expression in HEK293T cells.

To further address the mechanism by which calcium and cadmium activate ERRγ, HEK293T cells were co-transfected with GAL-4-ERRγ-LBD and a CAT reporter construct and treated with ERRγ agonists (DY131 and GSK4716), calcium, or cadmium in the absence or presence of the 4-hydroxy tamoxifen. Treatment with DY131 (1 µM) or GSK4716 (1 µM) resulted in an approximately 1.8- and 2.0-fold increase in CAT expression, respectively. Treatment with calcium (7.5 or 10 mM) for 48 h or cadmium (10 µM) for 24 h resulted in a significant increase of approximately 1.9- to 2.0- and 2.4-fold in CAT expression, respectively. Treatment with 4-hydroxy tamoxifen blocked the effects of calcium and cadmium ( Figures 6A, B ), suggesting that calcium and cadmium modulate ERRγ activity though its LBD.

Our previous study identified four potential calcium interaction sites on the aqueous surface of the LBD of ERα (37). The first site is formed by H377, E380, and C381 on helix H4. The second site is formed by H516, N519, and E523 at the interface between helices H10 and H11. The third site is formed by K529, N532, and V534 on helix H11 and in the loop connecting helices H11 and H12 and the fourth site is formed by D538, E542, and D545 on helix H12. The high homology between the LBDs of ERα and ERRγ suggests that the corresponding sites on ERRγ are potential calcium and cadmium interaction sites. Sequence alignment identified four sites in the LBD of ERRγ (53). The first site is formed by S299, Q302, and S303 on helix H4. The second site is formed by Q426, T429, and Q433 at the interface between helices H10 and H11. The third site is formed by K439, K443, and V444 on helix H11 and in the loop connecting helices H11 and H12 and the fourth site is formed by K448, E452, and E455 on helix H12. The amino acids S303 and T429 of ERRγ have side chain groups similar to the corresponding amino acids C381 and N519 of ERα. The amino acid S303 is also conserved among the ERR family and E452 is conserved between the ERR family and ERα ( Supplementary Figures 5A–C ). To determine whether these sites are potential interaction sites of calcium and cadmium with the ERRγ, S303 on helix H4, T429 on helix H10, and E452 on helix H12 were individually mutated to alanine in the GAL-4-ERRγ-LBD, transiently transfected into HEK293T cells, and treated with calcium, cadmium, DY131, or GSK4716. Similar to wild-type GAL-4-ERRγ-LBD, treatment of GAL-4-ERRγ-LBD mutants S303A, T429A, or E452A with DY131 or GSK4716 resulted in a 1.6- to 1.7-fold or 1.6- to 1.9-fold increase of CAT, respectively, indicating that mutation of these amino acids did not alter the activity of the LBD. However, treatment of the mutants with calcium or cadmium did not significantly increase CAT expression, suggesting that S303, T429, and E452 are potential calcium and cadmium interaction sites on the ERRγ LBD ( Figures 6C–H ). The ability of DY131 and GSK4716 to activate S303A, T429A, and E452A may be due to their mechanism of activation. DY131 and GSK4716 bind in the pocket of the LBD, whereas S303, T429, and E452 are located on the aqueous surface of the LBD. For example, GSK4716 interacts with Asp328 and rearranges Arg316 and Glu275 in the pocket (54).

To further explore the interaction between calcium and the ERRγ LBD, molecular dynamics simulations were performed. To predict calcium docking sites, the crystal structure of ERRγ LBD (ID: 2GP7) was selected and its A chain representing ERRγ LBD was submitted to both metal ion-binding site for prediction: modeling server (MIB) and Autodock Vina. In agreement with the mutational analysis, MIB predicted that calcium binds to E452 and Autodock Vina predicted that calcium binds to S303 and T429. Apo ERRγ and calcium-docked ERRγ were then analyzed by molecular dynamics simulations (videos 1–3). Root mean square fluctuation (RMSF) analysis suggested that calcium binding stabilizes the LBD structure ( Figure 6J ). This effect is especially significant in the loop connecting helix H9 and helix H10 and in the N-terminus of helix H10 (amino acids 405–420) as calcium-bound ERRγ LBD has lower RMSF compared with apo ERRγ LBD. Previously published molecular dynamics simulations show that the ERRγ agonist BPA also lowers the RMSF in the same region of the LBD, suggesting that calcium mimics another ERRγ agonist to stabilize and activate ERRγ (19). The calcium-bound ERRγ molecular models are shown in Figures 6J–L .

To better understand the role of calcium and cadmium in influencing metabolic genes, we re-analyzed our previously published microarray results using GSEA. In that study, MCF-7 cells were treated with cadmium and microarrays were used to measure gene expression (11). As expected, GSEA showed a significant increase of hallmark estrogen response early and late pathways (data not shown), which is consistent with our previously published studies showing that cadmium binds to and activates ERα. GSEA also identified an increase in several biosynthetic pathways, including an increase in the carbohydrate biosynthetic pathway ( Figure 7A ), suggesting that cadmium influences metabolism in MCF-7 cells. To verify the microarray results, MCF-7 cells were treated with calcium (1, 3, 5, and 10 mM) or cadmium (5 or 10 µM) in the absence or presence of 4-hydroxy tamoxifen (5 μM) or the ERα antagonist ICI-182,780 (fluvestrant, 1 μM). Similar to HepG2 cells and ERRγ-transfected HEK293T cells, treatment with calcium resulted in a maximal 2.8-, 2.2-, and 3.2-fold increase in PEPCK1, G6PC, and SLC2A2, respectively ( Figures 7B–D ) and treatment with cadmium resulted in a maximal 3.8-, 4.0-, and 9.0-fold increase in PEPCK1, G6PC, and SLC2A2, respectively ( Figures 7E–G ). The effects of both calcium and cadmium were blocked by 4-hydroxy tamoxifen. To verify that the effects of calcium and cadmium were not due to the activation of ERα, MCF-7 cells were also treated with ICI-182,780. The antiestrogen did not significantly block the effects of calcium or cadmium, suggesting that metal activation of ERα was not responsible for the increase in PEPCK1, G6PC, and SLC2A2. Lastly, immunofluorescence verified the expression of ERRγ in MCF-7 cells ( Supplementary Figure 4 ). Taken together, the results suggest that calcium and cadmium increase the expression of genes involved in gluconeogenesis through the activation of ERRγ.

This study has revealed that calcium is a potential ligand of ERRγ that mediates the effects of glucagon and that cadmium, which mimics the effects of calcium, disrupts metabolism through ERRγ. While this study identified three potential interaction sites (S303, T429, and E452), other interaction sites and mechanisms of activation remain undetermined. Moreover, this study raised new questions about the role of calcium and cadmium in metabolism. In pursuit of answering these questions, future research should involve in vivo studies and structural crystallography studies.