ERRγ Regulates Cardiac, Gastric, and Renal Potassium Homeostasis

Abstract

Energy production by oxidative metabolism in kidney, stomach, and heart, is primarily expended in establishing ion gradients to drive renal electrolyte homeostasis, gastric acid secretion, and cardiac muscle contraction, respectively. In addition to orchestrating transcriptional control of oxidative metabolism, the orphan nuclear receptor, estrogen-related receptor γ (ERRγ), coordinates expression of genes central to ion homeostasis in oxidative tissues. Renal, gastric, and cardiac tissues subjected to genomic analysis of expression in perinatal ERRγ null mice revealed a characteristic dysregulation of genes involved in transport processes, exemplified by the voltage-gated potassium channel, Kcne2. Consistently, ERRγ null animals die during the first 72 h of life with elevated serum potassium, reductions in key gastric acid production markers, and cardiac arrhythmia with prolonged QT intervals. In addition, we find altered expression of several genes associated with hypertension in ERRγ null mice. These findings suggest a potential role for genetic polymorphisms at the ERRγ locus and ERRγ modulators in the etiology and treatment of renal, gastric, and cardiac dysfunction.

In addition to regulation of oxidative metabolism, ERRγ regulates ion homeostasis, especially potassium, in renal, gastric and cardiac tissues and appears to influence hypertension in mouse and human.

In the highly oxidative renal, gastric, and cardiac tissues, ATP is largely consumed in the powering of the sodium-potassium adenosine triphosphatase (ATPase) to establish sodium and potassium ion gradients. The resulting Na⁺ and K⁺ gradients are responsible, in turn, for symporting and antiporting additional ions and metabolites such as protons, calcium, and glucose central to producing urine, gastric acid, and cardiac muscle contraction. Although dysregulated ion homeostasis by membrane proteins has been demonstrated in several diseases and syndromes such as Timothy syndrome, Brugada syndrome, Liddle disease, and Dent disease (1,2,3,4,5), the transcriptional mechanisms that link the processes of energy production and ion homeostasis are not well understood. Several nuclear receptors and their ligands, such as thyroid hormone receptor, glucocorticoid receptor, mineralocorticoid receptor, vitamin D receptor, and the peroxisome proliferator-activated receptor (PPAR) are known to exert profound effects on ion homeostasis and/or cellular energetics (6,7,8,9,10). Therefore, nuclear receptors are likely candidates to coordinate cellular ion homeostasis and energy production at a transcriptional level.

The estrogen related-receptor γ (ERRγ), along with ERRα and ERRβ, are orphan nuclear receptors, identified by sequence similarity to the estrogen receptor ERα, that are active in the absence of estrogenic ligands (11,12). Despite their constitutive activity, mammalian ERRs exhibit limited affinities for a variety of synthetic ligands, including diethylstilbestrol and 4-hydroxytamoxifen (4-OHT), which act as inverse agonists (13,14). The widespread chemotherapeutic use of tamoxifen and its metabolite, 4-OHT, with untoward side effects, and the recent identification of environmental pollutants as endocrine disruptors, such as bisphenol A and organochlorine pesticides that oppose the action of 4-OHT on ERRγ, suggest that ERRs may be novel contributors to the actions of these compounds (15,16,17).

Relatively little is known about the ERRs compared with classic endocrine nuclear receptors, such as glucocorticoid receptor, thyroid hormone receptor, or the more recently recognized PPARs (18). ERRα has been demonstrated to have a critical role in oxidative metabolism energy production and has been implicated as a risk factor in Type 2 diabetes (19,20,21,22,23). ERRα null mice have a reduced capacity for mitochondrial energy production and nonshivering thermogenesis, and antisense experiments suggest a role in osteogenesis (22,24,25).

The study of ERRβ has been complicated by the midgestational lethality of knockout (KO) mice resulting from placental defects resembling the administration of the ERRβ inverse agonist diethylstilbestrol (14,26). Tetraploid rescue or conditional KO strategies have demonstrated a role for ERRβ in primordial germ cells, the regulation of central nervous system development, and the production of the specific ionic constituents of inner ear endolymph (27,28).

In this study, we have identified a critical role for ERRγ in the regulation of genes involved in transport across three highly metabolic tissues: heart, stomach, and kidney. Genomic and physiological analyses indicate that ERRγ regulates the cardiac, gastric, and renal expression of key ion homeostatic genes, including Kcne2, resulting in hyperkalemia, gastric parietal cell abnormalities, and prolonged electrocardiographic QT intervals of ERRγ null mice. Furthermore, in addition to altered expression of several genes associated with hypertension in ERRγ null mice, we observed a correlation between specific ERRγ single-nucleotide polymorphism (SNP) genotypes and altered blood pressure in humans.

Results

Transport genes are dysregulated in oxidative tissues of ERRγ null mice

Two lines of ERRγ null mice (termed S, for Salk, and G for GlaxoSmithKline) were generated with LacZ-neomycin resistance cassettes targeted to the second exon of ERRγ encoding the DNA-binding domain followed by a premature stop codon [Deltagen and (29)]. The S strain has been described previously, and the G strain is new to this study. Both KO lines eliminated mRNA and protein for ERRγ and resulted in animals that exhibited perinatal lethality by the second day of life [postnatal d2 (P2)], as evidenced by the recovery of homozygotes in Mendelian ratios at P0, and subsequent death of the homozygotes which appeared runted and failed to obtain milk (29). Because this lethality was independent of strain or housing conditions, the mice were not backcrossed and the experiments were thus performed on mice of mixed SV129/C57BL/6 background. A complete histological examination of the visceral organs did not uncover any substantial defects in the development of the pups (29).

To explore the genetic basis of the perinatal lethality in ERRγ null mice, gene expression analysis (MOE430A; Affymetrix, Santa Clara, CA) of brain, liver, salivary gland, stomach, and kidney from perinatal ERRγ null and wild-type (WT) mice was performed. These findings were then compared with our previous perinatal expression analysis (Affymetrix, GeneChip MG430 2.0) to uncover processes that are regulated by ERRγ across tissues (20,29). Expression analysis was performed using an online data analysis suite (Vampire) (30).

Collectively, expression analysis detected altered expression of 1788 genes in the heart, kidney, and stomach tissues of ERRγ null mice (Fig. 1A). Functionally, 7–29% of these genes are associated with the gene ontology (GO) classifications of Transport (GO:0006810) and Localization (GO:0051179), in addition to the 8–15% associated with the previously reported role of ERRγ in metabolic processes (GO:0008152) (Fig. 1B) (20,29). At the intersection of heart, stomach, and kidney expression changes, 150 genes associated with metabolic (e.g. Ndufa4, Hadha, Mdh1) and ion handing (e.g. Atp6v0b, Slc15a2) processes were detected, notably in the context of prolonged electrocardiograph (ECG) QT intervals and hyperkalemia in ERRγ null mice, as discussed below (Fig. 1C) (31,32,33,34,35). Furthermore, within this intersection approximately 17% of these genes have an association with hypertension (e.g. Vasn, Vegfb, Kcna5) (36,37,38). The known role of ionic fluxes in the highly metabolic tissues of heart, stomach, and kidney, led us to explore the transport gene set in more detail.

Figure 1.

Genomic analysis of ERRγ function. Stomach (S) and kidney (K) genome-wide analyses of perinatal gene expression were compared with our previous heart (H) perinatal expression analysis to uncover processes that are regulated by ERRγ across tissues. A, Expression analysis detected altered expression of 1788 genes in the heart, kidney, and stomach tissues of ERRγ null mice. B, Functionally, 7–29% of these genes are associated with the GO classifications of Transport (GO:0006810) and Localization (GO:0051179), in addition to the 8–15% associated with the previously reported role of ERRγ in metabolic processes (GO:0008152). C, At the intersection of heart, stomach, and kidney expression changes, 150 genes associated with metabolic, and ion-handing processes were detected, as well as hypertension. Heart is Salk strain; stomach and kidney are GlaxoSmithKline strain (Materials and Methods).

ECG and potassium channel expression abnormalities in ERRγ null mice

Having previously described a molecular explanation for the prolonged electrocardiogram QRS complex in ERRγ null animals, we sought a molecular basis for the prolonged electrocardiogram QT interval (Fig. 2A) (29). ERRγ, robustly expressed in the heart, has been shown previously to play a role in regulating mitochondrial and electrical properties despite a lack of overt histological abnormalities of the ERRγ null hearts at postnatal day 0 (Fig. 2, B and C) (20,29). Quantitative RT-PCR analysis revealed down-regulated gene expression for channels associated with long QT syndromes in ERRγ null heart, relative to WT controls (39,40). Kcne1, which encodes the MinK channel that forms a delayed inward rectifier channel essential for cardiac rhythm and rate, was significantly down-regulated (Fig. 2D). Furthermore, expression of the MinK-related peptide, Kcne2, and its pore-forming partner, HERG, encoded by Kcnh2, were reduced by approximately 30–50% in ERRγ null mice relative to WT littermate controls (Fig. 1D). Furthermore stable expression of short hairpin RNA directed against ERRγ in HL1 cells produced an approximately 70% reduction in Kcne1 expression, relative to control (Fig. 2E). Additionally, Kcnh2 expression was reduced by about 30% in the HL1-shERRγ cells, demonstrating that ERRγ regulates expression of Kcne1 and Kchn2 both in vivo and in vitro (Fig. 2E).

Figure 2.

ERRγ loss results in prolonged QT interval and down-regulated potassium channels. A, Representative averaged ECG demonstrates fractionated QRS complex in ERRγ heterozygous (HET) and KO and prolonged QT intervals in ERRγ KO mice at E18.5 (29). B, Whole-mount X-gal staining reveals the ubiquitous expression of ERRγ throughout the heart at E18.5. C, P0 ERRγ null mice have normal cardiomyocyte organization by hematoxylin and eosin (H&E) histology. D, Loss of ERRγ results in the altered expression of potassium channels of the heart. E, HL-1 cardiomyocyte cell line stably expressing small interfering RNA against ERRγ (siERRγ) has down-regulated expression of potassium channels, as seen in vivo. Values ± sd; *, P < 0.05; **, P < 0.01; ***, P < 0.001 by two-tailed Student's t test. All figures from Salk mice. GFP, Green fluorescent protein.

Gastric regulation of ion channel expression by ERRγ

Robust ERRγ expression was observed in the fundus of the stomach by both immunohistochemistry and whole-mount X-gal staining (Fig. 3, A and B), and ERRγ colocalizes with the parietal cell marker, Atp4b (Fig. 3E). Gastric expression of H⁺/K⁺ ATPase is markedly reduced, however, in ERRγ null animals (Fig. 3C), suggesting a potential role for ERRγ in regulating gastric parietal cell function and/or development. We observed down-regulation of several genes involved in parietal function (below), with reduced gastric parietal cell Dolichos biflorus agglutinin (DBA)-lectin staining, suggesting a global reduction in parietal cell markers in ERRγ null animals at embryonic d 18.5 (E18.5) (Fig. 3D) (41,42). Importantly, whereas the one potassium channel essential for gastric acid secretion, Kcnq, was not down-regulated, Kcne2 is reduced 4- to 5-fold in ERRγ null stomach relative to controls (Fig. 3H). Furthermore, expression of parietal anion exchanger, Slc4a2, was down-regulated by approximately 4-fold in the stomachs of ERRγ null mice at P2 (data not shown). Because physiological or histamine-induced acidification of the stomach does not begin until several days after birth, gastric pH measurements at P1were not acidic and not different than controls (data not shown) (43).

Figure 3.

Loss of ERRγ alters parietal cell markers. Panel A, Immunohistochemistry against β-galactosidase reveals parietal cells express ERRγ. Panel B, Whole-mount X-gal staining demonstrates ERRγ expression is limited to the secretory portion of the gastric fundus. Panel C, A striking reduction in the expression of the H⁺/K⁺ATPase is seen in ERRγ null mice. Panel D, Although DBA-lectin staining is reduced in ERRγ null mice, parietal cells are present. Panels E–G, Anti-ERRγ staining colocalizes with Atp4b expression. Panel H, Loss of ERRγ in stomach results in profound down-regulation of parietal cell markers by triplicate quantitative RT-PCR (n = 3 WT; 9 HET; 6 KO). Values ± sd; *, P < 0.05; **, P < 0.01; ***, P < 0.001 by two-tailed Student's t test. S, Salk mice; G, GlaxoSmithKline mice; HET, heterozygous.

ERRγ binds and activates Atp4a and Kcne2 promoter fragments

In silico predictions of (ERREs, Fig. 4A) in both the Atp4a and Kcne2 proximal promoters were confirmed by in vitro plate-based immobilized DNA-binding assays using nuclear extracts prepared from 293A cells overexpressing an N-terminal epitope-tagged ERRγ (V5-ERRγ, Fig. 4, B–D; Materials and Methods) (44). Briefly, 2 pmol of biotinylated ERRE (consensus, Atp4a or Kcne2) were immobilized to streptavidin-coated wells in a 96-well plate (supplemental Fig. 1 published as supplemental data on The Endocrine Society's Journals Online web site at http://mend. endojournals.org). Nuclear extracts containing V5-tagged ERRγ were added to the wells in the presence of increasing amounts (fold excess) of either a WT or a mutated ERRE, as indicated. Unbound proteins were washed away, and the bound ERRγ was detected using a peroxidase-conjugated anti-V5 antibody and colorimetric reaction. In vitro analysis detected specific binding of ERRγ to WT ERREs (consensus, Atp4a or Kcne2) that could be competed by excess corresponding WT ERRE, but not by a corresponding mutant ERRE (Fig. 4, C and D). In cell culture, transcriptional activity of either Atp4a or Kcne2 promoter regions fused to luciferase reporters was up-regulated nearly 2-fold by coexpression of ERRγ and the coactivator PGC1-α (PPAR γ coactivator 1), but not by either molecule alone (Fig. 4E). Additionally, this transcriptional activity was ERRE dependent and sensitive to ERRE mutations in either promoter. In vivo, ERRγ bound the ERREs in Atp4a or Kcne2 promoters as indicated by chromatin immunoprecipitation of chromatin isolated from adult mouse stomach by nonquantitative PCR (supplemental Fig. 2C).

Figure 4.

ERRγ binds and directly regulates Kcne2 and Atp4a promoters. A,) The mouse Atp4a and Kcne2 promoters contain putative ERREs (Materials and Methods and supplemental Fig. 1). B–D, ERRγ binds the putative ERREs found in Atp4a and Kcne2 genes. Two picomoles of biotinylated gene-specific or a consensus ERRE were preimmobilized on a strepavidin-coated 96-well plate. Nuclear extracts containing V5-tagged ERRγ were added to the wells in the presence of increasing amounts of either a WT or a mutated ERRE as indicated. Unbound proteins were washed away, and the bound ERRγ was detected using an anti-V5 antibody. E and F, Luciferase induction was observed in CV-1 cell transfection of reporter constructs containing WT Atp4a and Kcne2 promoter fragments, by nonmutant variants, when ERRγ and PGC-1α are cotransfected. *, P < 0.05; **, P < 0.01, by two-tailed Student's t test.

Renal alterations in potassium channels and pressor hormones in ERRγ null mice

The kidney is central in regulating the ionic composition of the blood and the control of vascular tone via pressor hormones that raise blood pressure, such as bradykinin and kallidin of the kallikrein-kinin system (KKS) (45). ERRγ is highly expressed in renal structures relevant to serum ion and blood pressure homeostasis, such as the renal cortex and collecting ducts, and notably within cells surrounding the distal tubules that mediate aldosterone-stimulated sodium and water reuptake (Fig. 4, A–D). Gene expression analysis identified altered expression of genes that regulate serum potassium and blood pressure, e.g. the renal KKS. Specifically, the potassium channel, Kcnj1 (Romk1) was down-regulated in the E18.5 null kidney by approximately 80% by quantitative PCR, relative to WT levels (Fig. 4E). Similarly, renal Kcne1 was reduced by approximately 50% in ERRγ null kidney (Fig. 4E). Consistently, Kcne2 was also down-regulated (Fig. 4E). The most dramatically reduced renal genes were kallikrein 1 (Klk1) and kallikrein 6 (Klk6) of the KKS (Fig. 4E). Whereas the role of Klk6 in the kidney is still being explored, Klk1 cleaves kininogen to lysl-bradykinin, which generates bradykinin to regulate renin secretion, ion homeostasis, and NO formation to lower blood pressure. These changes in renal gene expression did not alter aldosterone (data not shown) or serum sodium levels (Fig. 5B); however, serum potassium was significantly elevated in E18.5 ERRγ null pups, relative to WT controls (Fig. 5C).

Figure 5.

Loss of ERRγ alters renal markers. A, Hematoxylin and eosin staining of kidney does not reveal significant differences in ERRγ null mice. B, Anti-β-galactosidase staining indicates ERRγ expression in the distal tubules. C, X-gal staining of frozen sections shows that ERRγ is expressed in the distal tubules and collecting ducts of the kidney. D, Whole-mount X-gal staining reveals the high expression of ERRγ throughout the cortex of the kidney. E, In kidney, loss of ERRγ produces a down-regulation of potassium channels and components of the KSS as detected by triplicate quantitative PCR (n = 3 WT; 9 HET; 6 KO). F, Serum sodium levels did not increase to statistically significant degree in ERRγ null mice at E18.5. G, Serum potassium levels were increased by more than 70% in ERRγ null animals at E18.5. Values ± sd; *, P < 0.05; **, P < 0.01; ***, P < 0.001 by two-tailed Student's t test. S, Salk mice; G, GlaxoSmithKline mice; H&E, hematoxylin and eosin; HET, heterozygous.

Consistent with this finding we observed that specific ERRγ genotypes in humans are associated with changes in blood pressure. This association was determined by examining 20 SNPs in 947 lean Caucasians (Materials and Methods). Specifically, two SNPs were found to be associated with hypertension and/or elevated blood pressure (supplemental Table 1). For SNP RS2789734, in ESRRG intron 3, the minor GG genotype was associated with a decreased risk of developing hypertension (OR = 0.54, P = 0.04) and lower systolic blood pressure (P = 0.005), independent of age, gender, or BMI. A second SNP, RS2789718, was also identified in intron 3 with the minor TT genotype that is associated with increased systolic (P = 0.01) and diastolic blood pressure (P = 0.03), suggesting, that although this result does not meet genome-wide significance, ERRγ may represent a critical locus that both positively and negatively influences blood pressure.

Discussion

There are partially intersecting arms of transcriptional regulation by PPARs and ERRs that are dependent on more general coactivating molecules, such as PGC1a. KO mouse models for each gene have been studied with phenotypes spanning midgestational lethality (ERRβ, PPARγ) to neonatal lethality (ERRγ) to relatively mild metabolic changes (ERRα), highlighting the varied compensatory capacities of these transcriptional networks. Consistently, loss of ERRγ alters gene expression in a complex manner that is of relatively broad scope and low magnitude and yet incompatible with life. Analysis of these gene expression changes in the light of phenotypic ECG changes, hyperkalemia, and the consideration of functions of the tissues with high ERRγ expression led us to examine the expression of genes known to contribute to the function of heart, stomach, and kidney.

In this study we found that loss of ERRγ results in markedly reduced gene expression of key potassium channel subunits and profound dysregulation of potassium homeostasis. The centrality of potassium homeostasis in the function of cardiac, gastric, and renal tissues is reflected by prolonged electrocardiogram QT intervals, down-regulation of parietal cell genes, and raised serum potassium levels, respectively. The significance of these findings is underscored by the identification of human polymorphisms in the ESRRG locus that are associated with altered blood pressure. These findings demonstrate a role for ERRγ in the genetic regulation of potassium homeostasis at molecular, tissue and whole animal levels.

ERRγ and prolonged QT intervals

In the heart, birth marks a transition toward the preferred use of lipids over carbohydrates and increased work by the right ventricle of the heart to perfuse the pulmonary vasculature. Loss of ERRγ blunts this transition to oxidative metabolism and alters the expression of key potassium channels and subunits, Kcne1, Kcne2, and Kcnh2, that along with increased serum potassium (below), contribute to an increased electrocardiographic QT interval (46). This prolonged QT interval is consistent with down-regulation of the Kcnh2 gene (HERG), a voltage-gated potassium channel, which when mutated causes long QT syndrome type 2 (LQT2) (47). Similarly, Kcne1 and Kcne2 encode small integral membrane proteins that assemble with Kcnh2 to modulate activity and consistently, mutations in these genes cause LQT5 and LQT6, respectively (48,49). Furthermore, high-dose tamoxifen has been associated with ECG disturbances in humans undergoing chemotherapy and stresses the importance of considering the role of ERRγ in these arrhythmias (50).

ERRγ and gastric parietal cell development

We observed the presence of parietal cell markers in the ERRγ null stomach before the age when gastric acid secretion begins, but the reduction of ERRγ target genes, Atpa4, Atpb4, and Kcne2, suggests ERRγ null animals might develop achlorhydria if they are to survive (Fig. 3). For instance, the genetic deletion or pharmacological inactivation of the H⁺/K⁺ ATPase or Kcne2 disrupts gastric acid production (51,52,53,54). We predict that ERRγ activity may be a target for the development of therapies to address gastric disorders such as gastroduodenal ulcers, reflux esophagitis, and stress-related mucosal disease. In addition, Kcne2 has been identified as a growth regulator of stomach tumors, indicating that pharmacological modulation of ERRγ activity may favorably impact gastric tumor progression by modifying the expression of Kcne2 (55).

ERRγ, hyperkalemia, and hypertension

The kidney is a complex, highly metabolic organ that establishes the ionic composition of the blood and the hormonal control of vascular tone. The striking down-regulation of the KKS gene, Klk1, implicates ERRγ as a key regulator of salt homeostasis and blood pressure. Klk1 expression is regulated by salt intake, and polymorphisms in its promoter have been associated with hypertension in humans. In addition, urinary levels of Klk1 are inversely related to blood pressure in humans (56). ERRγ may be a critical regulator of vascular health and kidney function via the KSS system and bradykinin to oppose the hypertensive actions of the renin-angiotensin system.

The decreased renal expression of the potassium channel genes, Kcnj1, Kcne1, and Kcne2, and elevated serum potassium levels in ERRγ null animals indicate that ERRγ transcriptionally regulates renal potassium homeostasis. This regulation is highlighted by Kcnj1, a component of the renal K_ATP channel that excretes excess potassium into the urine. Our results and the previously described regulation of Knce1 and Kcnq1 by ERRβ suggests that ERRβ/γ may exert combinatorial control over a subset of genes as has been demonstrated for ERRα/γ (20,27). Dysregulated endocytosis of KCNJ1 has been implicated in elevated serum potassium levels (hyperkalemia) in a subset of patients with hypertension (57). It follows that the down-regulation of Kcnj in E18.5 ERRγ null animals is consistent with the observed hyperkalemia (Fig. 5G).

The clinical relevance of ERRγ dysfunction is revealed in the association of SNPs in the human ESRRG gene that correlate with both increases and decreases in blood pressure; providing the first evidence of a human genetic disorder associated with ERRγ (supplemental Table 1). Although the molecular mechanism by which ERRγ function is altered by the intronic SNPs is not known in humans, it is suggestive of a possible influence on blood pressure by haploinsufficiency or a hypomorphic allele.

Genomic analyses

The role of ERRγ in transcriptional regulation has been examined by studying expression changes in perinatal ERRγ KO and WT stomach, kidney, and heart. The resulting analysis of the current studies, and their interpretation in the context of previous findings, provides a clearer picture of the role of ERRγ across tissues, e.g. how ERRγ may relate to mechanisms by which cells coregulate metabolism and transport functions to achieve their differentiated roles in mature tissues.

In conclusion, we report an unanticipated role for ERRγ in the control of ion homeostasis in the highly oxidative cardiac, gastric, and renal tissues. The widespread use of the selective ER modulator, 4-OHT, and increasing abundance of environmental pollutants that may alter the activity of ERRγ focus attention toward side effects and toxicities that alter cardiac, gastric, and renal function (15,17,18,50,58). Furthermore, development of agents that preserve ERR activity during selective ER modulator administration may prevent toxicities that limit the use of high-dose therapies to treat ER-dependent disease (59,60).

Materials and Methods

Experimental animals

All procedures performed were in compliance with the Animal Welfare Act and United States Department of Agriculture regulations and approved by the GlaxoSmithKline Institutional Animal Care and Use Committee or the Salk Institute IACUC. ERRγ-null mice termed “G strain” were generated by Deltagen (Redwood City, CA) by homologous recombination using a targeting vector that deletes nucleotides 586–610 in exon 2 of the ERRγ open reading frame, or as previously described (S strain) (29). This targeting event removes the first zinc finger of the DNA-binding domain and results in a frameshift in all three reading frames that introduces numerous stop codons. No ERRγ protein or message was detected (data not shown). Mice were maintained on standard laboratory chow and allowed food and water ad libitum. Mice were of mixed SV129/C57BL/6 background.

Genomics

We performed Affymetrix expression array analysis of tissues (brain, salivary, liver, stomach, and kidney) collected from perinatal mice. RNA was extracted by Trizol, per manufacturer's instructions. RNA was labeled and hybridized to Affymetrix, MOE430A chips per manufacturer protocol. Heart gene expression was derived from previously published work and referenced to these findings (29). Changes in gene expression and gene ontology classifications were detected using an online analysis platform (30).

Cardiac electrophysiology

Standard lead II ECG recordings were performed on neonatal (E18.5, ∼1.25 g) mice as previously described (29). Chlorided 0.004-inch sterling silver wires (Surepure Chemetals, Florham Park, NJ) were sutured in a lead II arrangement. Amplified ECGs (100 sec) were digitized at 6 kHz (National Instruments, Austin, TX), recorded with a custom MATLAB software application (QRS), and stored for later analysis. The acquired ECG records were analyzed with computer-aided semiautomatic wave analysis and quantification of the ECG waveform. The analysis program was used to group 100 individual beats based on identification of the time points (at which the QRS complex crossed the isoelectric line of the ECG recording) from which quantitative measurements were derived by the software. All ECG recordings were made while mice were left recumbent on a 37 C water blanket.

High-salt diet

The high-salt diet consisted of 0.9% NaCl drinking water with NaCl increases of 0.1%/d for 9 d. Mice were then fed high-salt diet (Laboratory Rodent diet 5001 supplemented with 8% NaCl from LabDiet; Purina Mills, LLC, Gray Summit, MO) for 8 wk. Multiple urinary chemistry parameters were determined. Nonnormal log-transformed values were analyzed by ANOVA. The ANOVA modeled the genotype (+/+, null/+), diet (regular, high salt), and genotype-by-diet interaction effects (using SAS PROC MIXED). Some markers showed unequal variances across treatment groups requiring fitting of an unequal variance model. P values < 0.05 were considered statistically significant. Mean fold changes were reported.

Gene expression analysis

RNA was prepared using Trizol reagent. Affymetrix analysis was performed according to manufacturers instructions. Real-time quantitative PCR analysis was performed using an ABI PRISM 7700 Sequence Detection System instrument and software (PE Applied Biosystems, Inc., Foster City, CA). Primers and probes were designed using Primer Express Version 2.0.0 (Applied Biosystems) and synthesized by Keystone Laboratories (Camarillo, CA). All primers and probes were cross-referenced with the National Center for Biotechnology Information Blast program. Fold induction values were calculated by subtracting the mean threshold cycle number (Ct) for each treatment group from the mean Ct for the vehicle group and raising 2 to the power of this difference. For animal studies, the average of each treatment group (three to five animals/group) was used.

Quantitative RT-PCR

Assays were performed as previously described (29). Briefly, for each biological sample, quantitative PCR reactions were performed in triplicate, and expression was normalized to Gapdh or U36b4 expression. Bar graphs represent the averaged relative expression of the biological samples and the sem, assigning WT a relative expression of 1 for each indicated transcript.

Immunohistochemistry

Rabbit anti-ERRγ serum was generated in s using a peptide (404-AGQHMEDPRRAGKMLM-419) from mouse ERRγ helix 9 (Research Genetics/Invitrogen, Carlsbad, CA). Antibodies were affinity purified using the same peptide and tested by Western blotting. No cross-reactivity to ERRα or ERRβ was noted (data not shown).

Staining was performed on formalin-fixed paraffin-embedded tissue samples sectioned at 4 μm. Epitope unmasking was performed as necessary by immersing slides in citrate buffer (Vector Antigen Unmasking solution at 1:100 dilution; Vector Laboratories, Inc., Burlingame, CA) according to manufacturer's instructions. Antibodies were used at the following concentrations: rabbit anti-ERRγ : 2.1 μg/ml; H+/K+ ATPase β-subunit, 1:10,000 (MA3–923, Affinity BioReagents, Golden, CO); β-galactosidase, 2- (200–4136, Rockland Immunochemicals; Biotinylated DBA Lectin, 3 μg/ml (B-1035Vector Laboratories). Secondary antibodies were goat antirabbit AlexaFluor 488 and goat antimouse AlexaFluor 568 (Invitrogen). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole. Colorimetric immunohistochemical staining was performed on a Discovery staining system (Ventana Medical Systems Inc., Tucson, AZ). All procedures were carried out according to the manufacturer's directions with approximately 3-fold higher antibody concentrations.

Reporter assays

CV1 cells were transfected as previously described (61). Both ERRγ and PGC-1α expression had small effects on the pGL4TKluc plasmid (activating 1.3- to 1.7-fold above pGL4TKluc alone). This activity was subtracted from the results obtained from the reporter gene constructs. The following primer pairs (5′-3′) were used to amplify the genomic fragments in the reporter plasmids. Atp4a mouse chromosome 7:31495785–31496390, 606 bp forward (F): TGGAACAGGAAGTGTGGCTAG; Atp4a reverse (R): GGTCACCTAGGGAAGTAAAG; Kcne2 mouse chromosome 16:92293254–92293671, 418 bp F: TGTCCAGAGTGTCCTTGTCA; Kcne2 R: GTGGGTCTAACTGGTCAG.

Kcne2 F mutant: TGTCCAGAGTGTTTTTGTCA; site-directed mutation on the Atp4a ERRE was performed using the QuikChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA) system and the following primers: Atp4aQC1: CTTCAGGGAGAACTCTCAGAGAATTTGACATTTGCAGGAGA; Atp4aQC2: TCTCCTGCAAATGTCAAATTCTCTGAGTTCTCCCGAAG.

Chromatin immunoprecipitation

Stomach chromatin was prepared and immunoprecipitated as described with the following primers (62). Atp4a TSS F: ACATGCTAGTGGAACCCTGTGGAAT; Atp4a TSS R: CTTGAACAGCTCAG.

Atp4a ERRE F: TGGAACAGGAAGTGTGGCTAG; Atp4a ERRE R: GTAAGTTCTATGCTTTGTGC; Knce2: as above.

DNA-binding assay

Binding of epitope-tagged ERRγ from nuclear extracts to biotinylated double-stranded oligonucleotides immobilized to streptavidin plates was performed as previously described with the following primers (supplemental Fig. 1) (44). Briefly, epitope-tagged ERRγ is first bound to immobilized ERRE. After competition with a putative ERRE, the remaining ERRγ is assayed by enzymatic reaction based on the binding of antibodies targeting the epitope tag of the ERRγ fusion protein. WT consensus ERRE: GTGGACTTAGTTCAAGGTCAGTTAT; mutant consensus ERRE: GTGGACTTAGTTCAAaaaCAGTTAT; Atp4a ERRE1: CTCTCAGAGAAGGTGACATTTGC; Kcne2 ERRE1: TGTCCAGAGTGTCCTTGTCAGGTCACATCA The consensus and putative ERREs are underlined.

Study population and SNP analysis

Inclusion criteria for this study population (n = 947) was a body mass index (BMI) of at least 40^th percentile for their age and sex in the Canadian population, as well as a self-recorded history of no previous BMI/ body weight greater than the 50^th percentile for their age and sex over any 2-yr consecutive period. The average BMI in this population was 21.8 for males and 19.9 for females. Exclusion criteria for this study population were any eating disorders, current use of medication aimed at weight reduction (including herbal supplements), current major depression episode, or a recent significant variation in weight loss due to a medical condition in the past 6 months. Subjects were Caucasian, defined as having three or four grandparents of Caucasian ancestry and were more than 18 yr of age. Subjects were not allowed in the study in case of a known HIV, hepatitis B, or hepatitis C infection. The mean age of the population was 46.6 yr for males and 43.5 yr for females, with 39.7% of the population being male; 15.8% of this population was hypertensive based on defining hypertension as blood pressure (BP) greater than 140/90 or self-reported medical history of high BP or currently taking medications for BP. The study was approved by the Human Research Ethics Committee of the Ottawa Hospital and the Human Research Ethics Committee at the Ottawa Heart Institute. All participants gave their informed consent.

We thank Eren Demirhan for assistance with statistical analysis; the GlaxoSmithKline Transgenics Group for support during many phases of this work; Kristie Powell and Thomas Golding for E18.5 organ harvests; and Kathryn Coulter, J.D., Ph.D., for useful comments on the manuscript. We thank Estelita Ong and Sally Ganley for administrative support. W.A.A. thanks Samuel L. Pfaff for current support.