Targeted disruption of the Kcnj5 gene in the female mouse lowers aldosterone levels

Aldosterone plays a central role in blood pressure regulation, and its dysregulation is seen as increasingly important in hypertension. Up to 10% of hypertension is now attributed to primary aldosteronism (PA), characterized by autonomous aldosterone production, independent of the renin–angiotensin pathway [1]. Recent guidance suggests that a third of PA patients have unilateral aldosterone producing adenomas (APAs) in an adrenal gland [1]. Exome sequencing has also shown that over a half of these APAs harbour somatic mutations in the potassium channel encoding gene, inwardly rectifying K channel CNJ type (KCNJ) (KCNJ5) [2–4]. The importance of these KCNJ5 mutations in PA has been reviewed in detail [5,6].

The prevalence of KCNJ5 mutations in APAs varies between cohorts, with up to 70% of APAs in some south Asian cohorts harbouring KCNJ5 mutations [7]. KCNJ5 encodes a G-protein regulated inwardly rectifying potassium channel (GIRK4), with the majority of mutations identified in APAs affecting the cation selectivity of this channel, and resulting in increased Na⁺ permeability [2,8,9]. The current consensus posits that this increased Na⁺ permeability of the mutant GIRK4 allows Na⁺ influx into the normally hyperpolarized aldosterone-producing cells of the zona glomerulosa (ZG) causing them to depolarize [4]. This depolarization opens voltage-gated calcium channels that activate Ca²⁺/calmodulin-dependent protein kinases, increasing transcription of aldosterone synthase (CYP11B2) and eventually increasing aldosterone production.

Despite extensive characterization of the KCNJ5 mutations identified in APAs, little is known about the role or importance of the wild-type (WT) GIRK4 channel in aldosterone regulation. Yet, KCJN5 is expressed at higher levels in the adrenal than the atria (http://www.gtexportal.org/home/gene/KCNJ5) where its role in the muscarinic currents in the heart is well understood. This differential expression of transcript in the adrenal is also seen for other potassium channels such as the two-pore K channel (K2P) TASK1 (KCNK3) channel (http://www.gtexportal.org/home/gene/KCNK3). The ZG cells have been shown to have a resting membrane potential of approximately −80 mV, close to the E_k of potassium (−90 mV) in these cells, and TASK channels are thought to be important contributors to the high resting K permeability of rodent ZG cells [10,11]. WT GIRK4 channels, probably as heterotetrameric channels with KCNJ3, could contribute to the basal hyperpolarization of the ZG cell. Equally they may have a role in ZG repolarization after they are depolarized in response to ATII, since the Gβγ subunits liberated activate GIRK4 channels [12,13]. This would suggest an inhibitory role for the channel, resulting in the ZG cell requiring a larger depolarizing stimulus for aldosterone production.

By utilizing the previously established KCNJ5 (GIRK4) (^−/−) knockout (KO) mouse line [14,15], we have investigated the role of WT GIRK4 in the mouse adrenal and its impact on aldosterone secretion.

They were a generous gift from Dr Kevin Wickman (Department of Pharmacology, University of Minnesota, Minneapolis, MN, U.S.A.) and Dr Matteo Mangoni (Centre National de Recherche Scientifique (CNRS UMR 5203), Department of Physiology, Montpelier, France) and were maintained in Cambridge by outcrossing with WT C57/BLJ6 mice that were also used as the littermate controls, animals were used for experiments aged 13–16 weeks [14,15]. The KCNJ5 genotype of each mouse used was confirmed by PCR: neomycin primers 5′ ATGGATTGCACGCAGGTT 3′, 5′ GATACCGTAAAGCACGAGGAAG 3′; coding exon 1 (exon 3 modern mRNA), 5′ TAGAACCACAGGACACCTAGTGAG 3′, 5′ CATTGCCTACGGACGGG 3′. The animal research was regulated under U.K. law, specifically the Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012 following ethical review by the University of Cambridge Animal Welfare and Ethical Review Body.

Formaldehyde fixed paraffin embedded (FFPE) samples were cut using a microtone to 5-μM sections. Sections were deparaffinized in histoclear II (National Diagnostics, Atlanta, GA) and dehydrated in graded ethanol ending in ddH₂O. Antigen retrieval was performed using standard procedure in the 2100-Retriever (http://www.aptum-bio.com) using commercial universal antigen retrieval solution (http://www.aptum-bio.com).

Mounted tissue sections were stained using the Envision DAB enhancer kit from Dako following manufacturer’s protocol with anti-DAB2 (disabled 2) (http://www.bdbiosciences.com). The following commercial antibodies were used:

An optimized immunohistochemical protocol using the Envision Plus IHC kit from DAKO was used to stain DAB2. In brief, FFPE 5 μM sections mounted on to slides were rapidly rehydrated through Histoclear II and reducing ethanol solutions, each for 20 s. Antigen retrieval was performed at 60°C overnight with universal antigen retrieval solution (Aptum Biologicals, https://www.proteogenix-products.com), peroxidases blocked for 10 min, and both primary (anti-DAB2) and secondary antibodies incubated for 10 min each. Between each step, slides were washed twice for 1 min in PBS. Following staining, slides were rapidly dehydrated in ascending ethanol solutions for 20 s each, ending with 100% ethanol, allowed to dry and used immediately for laser capture. Laser capture was performed on a Leica LMD6, with samples falling directly into lysis buffer before immediate extraction using the Qiagen RNeasy FFPE kit (https://www.qiagen.com) according to manufacturer’s instructions.

Animals were anaesthetized under 3% isoflourane in O₂ and a terminal blood sample collected by cardiac puncture; blood was immediately transferred to lithium heparin coated tubes for plasma collection. This fresh blood was also analysed on the iSTAT using EC8+ cartridges (www.pointofcare.abbott). Plasma was removed, snap frozen and stored at −80°C.

Aldosterone was measured from 10 μl frozen plasma using the Cisbio aldosterone assay kit according to manufacturer’s instructions (www.cisbio.com), in which charcoal-stripped serum was used for dilution of the assay standards.

Total RNA, including miRNA was extracted using Qiagen miRNeasy kit from 16 whole mouse adrenals taken from animals between 13 and 16 weeks of age, four from each group; male WT (^+/+), male KCNJ5 (^−/−), female WT and female KCNJ5 (^−/−). Quality of extracted RNA was analysed using an Agilent TapeStation (www.genomics.agilent.com), all RNA samples reached RIN numbers >9. The RNAseq analysis was performed by Cambridge Genome Services (www.genomicservices.path.cam.ac.uk) using Ilumina Mouse WG6 beadchips (approximately 45000 markers mapped to NCBI RefSeq build 36.2, release 22) and analysed using the EdgeR Bioconductor package (https://bioconductor.org/packages/release/bioc/html/edgeR.html). Pathway analysis of the differentially expressed genes was performed with the Ingenuity Package Analysis (IPA®) software package (https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis/). The RNAseq data used in this manuscript have been publically deposited on GEO.

H295R cells were maintained as described previously [16]. Cells were treated with the described drugs or DMSO as vehicle control for 24 h after plating in 96-well plates. The following drugs were used: rosiglitazone (www.sigmaaldrich.com R2408), fenofibrate (www.sigmaaldrich.com F6020) or GW 6471 (www.sigmaaldrich.com G5045). For the last 24 h of treatment, cells were also treated with 10 nM angiotensin II (Ang-II) (www.sigmaaldrich.com, A9525) ± 10 nM losartan (www.sigmaaldrich.com 61188). At the final time point, medium was removed for aldosterone quantification and cell viability measured by MTT assay as described previously [16]. Aldosterone production was normalized by viability within each well. For gene expression experiments, medium was discarded and cells lysed for RNA extraction using Purelink RNA extraction kit (www.thermofisher.com).

Reverse transcription was carried out using Superscript IV (www.thermofisher.com) as per manufacturer’s protocol. Gene expression analysis was carried out using the TaqMan® Fast Advanced Master Mix and validated TaqMan® Gene Expression Assays according to manufacturer’s protocol (www.thermofisher.com) on an ABI 7500 platform. Probes used: mouse CYP11B2, FAM-MGB, Mm01204955_g1, mouse KCNJ5, FAM-MGB, Mm01175829_m1 and eukaryotic 18S, VIC-MGB, 4319413E. Relative gene expression was calculated by the ΔC_T method against 18S as the housekeeper and expressed as a multiple of the control value (set to 1) [17].

Normally distributed data were analysed by ANOVA with post-hoc testing or Student’s t tests as appropriate using Prism 6 software (www.graphpad.com). Significance was taken as P<0.05.

There were no obvious macroscopic differences between the adrenal glands recovered from KCNJ5 KO (^−/−) compared with WT (^+/+) mice. Sections of the glands also showed that zonation between the cortex and medulla (M) was maintained. Using the specific disabled 2 (DAB2) marker [18], the ZG also had a similar depth in both KO (^−/−) and WT (^+/+) glands (Figure 1 below).

We next determined the expression and localization of Kcnj5 in the adrenal gland of WT (^+/+) C57BL/6 mice. Due to the lack of a commercially suitable antibody with specificity for the channel in the mouse (Supporting Data), we carried out Kcnj5 gene expression analysis by qPCR of WT (^+/+) mouse adrenal tissue. The specificity of the KCNJ5 gene expression assay was confirmed by gene expression analysis in WT (^+/+) and KCNJ5 KO (^−/−) brain and adrenal cDNA (Supporting Data).

To confirm that Kcnj5 was specifically expressed in the outer ZG, as in the human adrenal cortex, laser capture microdissection was used to recover tissue from the ZG, zona fasciculata (ZF) and M. We confirmed successful microdissection by examining Cyp11B2 gene expression as a specific marker for the ZG (Figure 2 below). This showed that only cDNA extracted from the ZG expressed both Cyp11B2 and Kcnj5 genes, confirming that in the mouse Kcnj5 is expressed specifically by the aldosterone-producing cells of the ZG.

We measured plasma aldosterone levels in male and female adult KCNJ5 KO (^−/−), heterozygous (^+/−) and WT mice (^+/+), both at baseline and 30 min after an Ang-II challenge (200 mg/kg IP). Female WT mice displayed significantly higher plasma aldosterone levels than age-matched WT males (Figure 3). Basal plasma aldosterone levels were also unchanged across genotypes in male mice (Figure 3A below). In contrast, female Kcnj5 KO animals had significantly lower basal plasma aldosterone levels than female WT littermates. The levels in heterozygous females were intermediate. These differences were not explained by differences in Cyp11B2 gene expression (Figure 3B below). Aldosterone levels after Ang-II challenge were significantly higher in female compared with male KO mice, despite having similar basal levels of aldosterone (Figure 3C below).

Blood electrolytes in male and female Kcnj5-deficient mice were measured in fresh whole blood using an iSTAT device. The blood levels of K showed a downward trend from WT to homozygous KO and were consistently lower in females compared with males. However, neither of these trends were statistically significant. No differences were observed between sexes or genotypes in blood levels of sodium, chloride, glucose, urea, HCO₃ or percentage haematocrit. There was also no significant difference in body weight between genotypes, although males were significantly heavier than females throughout (Figure 4).

To explore the molecular basis for the sex differences in aldosterone secretion, we carried out RNAseq analysis of WT and Kcnj5 (^−/−) KO adrenals from both male and female mice. The RNAseq results showed that many more genes were differentially expressed in females compared with males (Table 1 below).

The five differentially expressed genes in the male did not include any genes plausibly linked to aldosterone production (Supplementary Data online). Of note, Kcnj5 was itself overexpressed in both male and female KO adrenals (Supplementary Data online). However, the Sashimi plot (Supporting Data) of the RNAseq data showed that exon 3 of Kcnj5 was not expressed in keeping with its deletion and replacement with a neomycin resistance gene (Neo^r) in the deletion cassette. This explains our failure to detect a full-length Kcnj5 cDNA on qPCR of the Kcnj5 KO (^−/−) adrenals.

We then compared the genes differentially expressed in male compared with female WT mice and those expressed in male compared with female KCNJ5 (^−/−) KO adrenals. This identified 184 genes that were shared between WT and Kcnj5 (^−/−) KO adrenals (Figure 5, below). In both analyses, sex-specific genes encoded on the X and Y chromosomes (Uty, Eif2s3y, Kdm5d and Xist) were ranked among the top ten differentially regulated genes (Supplementary Data). Two aldo-keto reductases Akr1c18 and Akr1d1 were down-regulated in females of both genotypes compared with their male counterparts, and the potassium channel encoding gene Kcnk1 had a consistently lower gene expression in males compared with females of the same genotype. There were also a number of differentially expressed genes unique to each genotype, including the Y chromosome gene Ddx3y, that was only differentially expressed in WT adrenals.

To look for functional connections between the differentially expressed genes in the female adrenals, we undertook pathway analysis using the Ingenuity® software package (Ingenuity Pathway Analysis (IPA)) using the genes differentially expressed (log₂FC >±1.5) between female WT and Kcnj5 KO (^−/−) adrenals. This identified five canonical pathways and upstream regulators that highlighted a major role for the RXR nuclear receptor in controlling adrenal function (Table 2 below). RXR is in turn regulated by forming heterodimers with peroxisome proliferator activated receptor (PPAR) α and PPARγ. This analysis also predicted that the pathways through these nuclear receptors were uniformly down-regulated in the KO female adrenals.

IPA of the adrenal RNAseq data suggested that PPARα- and PPARγ-mediated pathways were involved in regulating adrenal aldosterone production. To test this in vitro, we exposed adrenal H295R cells to PPAR pathway agonists. The PPARγ agonist, rosiglitazone (10 μM), produced no significant change in aldosterone production from H295R cells treated for 48 h with the drug (Figure 6, below). However, rosiglitazone effectively blocked aldosterone production stimulated by 10 nM Ang-II for 24 h, suggesting that PPARγ activation may inhibit Ang-II-mediated aldosterone production. In contrast, cells treated with the PPARα agonist, fenofibrate (10 µM), showed a significant Ang-II independent increase in aldosterone production. This increase in aldosterone production was accompanied by a significant elevation in Cyp11b2 gene expression after 24 h fenofibrate treatment, which could be reversed by addition of the specific PPARα antagonist GW6471.

To see if the in vitro effect of PPARα activation on aldosterone production could occur in vivo, WT female mice were treated with fenofibrate (100 mg/kg PO) or vehicle (olive oil) daily for 2 weeks. Mice that received fenofibrate showed a significant increase in plasma aldosterone levels compared with control treated mice, but the effect was masked by the effect of the olive oil vehicle on aldosterone production (Figure 7, below). There was also a modest increase in Cyp11B2 gene expression in adrenals from fenofibrate treated mice compared with control (Figure 7). Together this supported the in vitro finding in H295R cells, showing that activation of PPARα by fenofibrate leads to increase in aldosterone production.

Interest in the role of the GIRK4 K channel in controlling adrenal aldosterone production has emerged only in the last few years with the discovery that both germline and somatic mutants in the KCNJ5 gene cause hyperaldosteronism [2,5]. However, there are no data on the role played by the WT GIRK4 channel in the normal adrenal. Our studies reported here show that female but not male Kcnj5-deficient mice have lowered plasma aldosterone levels compared with WT counterparts. This finding is mirrored by the observation of transcriptional changes in the female Kcnj5 (^−/−) KO adrenal, which are not present in male Kcnj5 (^−/−) KO adrenals. IPA analysis indicated that the pattern of differentially expressed genes in the female KCNJ5 KO compared with WT adrenals implicated PPARα and PPARγ pathways in the regulation of aldosterone secretion. We have confirmed the role of PPARα by demonstrating that activation of PPARα with fenofibrate, a specific agonist for this nuclear receptor, increases aldosterone production both in adrenal H295R cells and mice.

In the present study, there were gender differences in plasma aldosterone with the female having significantly higher levels than age-matched males. Even larger female over males differences in plasma aldosterone were reported in the C57/BL6J strain by Heitzmann et al. [19]. In parallel with these differences in aldosterone levels, we have identified major differences in the adrenal transcriptome of the female compared with male mice. Sexual dimorphism within the renin–angiotensin system is well documented and gender differences in WT adrenal gene expression was first reported by El Wakil et al. [20]. They found some 269 genes that were differentially expressed in male compared with female WT mouse adrenals. Other studies have also reported sex differences in adrenal phenotype, morphology and gene expression in both inbred strains and genetically modified mice [21–23]. For example, a microarray analysis of KCNK3 (TASK1) KO (^−/−) mice showed differences in gene expression and phenotype between female and male mice, with castrated male mice having a similar gene expression profile to sham-operated female mice [24].

We found that like human ZG mRNA for KCNJ5 was readily detected in the ZG of the mouse, although it has been reported that KCNJ5 mRNA was undetectable in the rat adrenal cortex (ZG or ZF) [25]. Chen et al. [25] also failed to show staining for KCNJ5 protein. However, in our hands, the antibody they used did not give specific staining for KCNJ5 protein in either IHC sections or Western blots from the mouse adrenal cortex (Supplementary Figure S1). Hence, although we could detect KCNJ5 message, the lack of a specific antibody for mouse KCNJ5 meant we could not confirm that protein was also present in the mouse ZG.

A striking feature of the RNAseq experiments was the finding that KCNJ5 gene products were up-regulated in the KO mice of both sexes. However, the mouse KCNJ5 mRNA is composed of four exons, but only two code for the mature 419aa protein and the majority of this coding region lies in exon 3. In the KO mice, exon 3 has been swapped for a Neo^r coding region, so although the qPCR analysis used a commercial Taqman primer-probe set that maps to the exon 3–4 boundary (Mm01175829_m1, Thermo Fisher), it will only amplify mature mRNA with both coding exons present. This explains our failure to detect a qPCR signal from the KO mice. This was an important negative control, since none of the commercially available antibodies could identify an appropriately sized band for mature GIRK4 protein that was absent from the KO adrenal. It is also notable that other inwardly rectifying K channels were not present among the genes differentially expressed in male or female KO adrenals. It might be expected that loss of GIRK4 activity could be compensated by altered expression of other inward-rectifying K channels. However, we saw no evidence of this in the RNAseq data.

We have identified 170 differentially expressed genes between sexes of our WT adrenals (using a threshold of a four-fold change), and 273 in adrenals from Kcnj5 (^−/−) KO mice. Indeed, the top seven genes from both lists actually overlapped between genotypes (Figure 5). Interestingly, of these seven genes, all three genes that were up-regulated in females (Xist, Akr1c18 and Akr1d1) were also up-regulated in the study by El Wakil et al. [20]. This finding confirms the robustness of RNAseq in correctly identifying differentially expressed genes, as we and El Wakil et al. [20] used different platforms yet identified very similar gene lists. Of the down-regulated genes however, only one gene, Uty, was consistent between ours and the El Wakil et al. [20] study. However, a large number of other down-regulated genes were consistent between our WT or Kcnj5 (^−/−) KO sex comparisons: including Srd5a2, Hmox1, Acan and Susd3, all of which featured in the top 20 differentially expressed genes in both comparisons (see Supplementary Data). Srd5a2 encodes type 2 5α-reductase (3-oxo-5-α-steroid 4-dehydrogenase 2), an enzyme responsible for the reduction in keto-steroids, including corticosterone, testosterone and aldosterone to their reduced metabolites. In female adrenals, the aldo-keto reductases, Akr1c18, and Akr1d1 are highly expressed, and Akr1d1 encodes steroid 5β-reductase which like 5α-reductase converts keto-steroids into their reduced metabolites. Similarly, the Akr1c family of aldo-keto reductases are implicated in progesterone metabolism [26]. The sexual dimorphism in mouse adrenal gene expression is intriguing considering the female predominance in human of APAs expressing KCNJ5 somatic mutations. A meta-analysis of published data on KCNJ5 somatic mutations showed 67% of them occurred in females [27–29]. Conceivably, KCNJ5 mutations could have a greater phenotypic impact on female compared with male adrenal cells because of a greater physiological importance of GIRK4 in the female adrenal. Based on the data presented here, this seems to be the case at least for the mouse. A sex difference in the aldosterone phenotype was also reported in the ITask1 KO mouse with only female adults displaying hyperaldosteronism [24]. The males were unaffected and appeared to be protected by androgen-regulated up-regulated expression of Task3 channels. However, it is important to note that H295R cells that are routinely used as an in vitro model of human adrenocortical cells and were used in our work are karyotypically female (https://www.lgcstandards-atcc.org/products/all/CRL-2128.aspx). Therefore, sublines derived from H295R such as HAC15 are also female. Hence, there is not an easy cell-based approach to explore the hypothesis that somatic KCNJ5 mutations have a larger effect on aldosterone production if expressed in female compared with male adrenal cells.

Previous work with H295R or HAC15 cells reported that incubation with PPARγ agonists reduced aldosterone production, although reported effects on CYP11B2 mRNA expression are inconsistent [30,31]. However, we could not reproduce these findings using rosiglitazone (Figure 6). At the PPARγ agonist concentrations previously used (10 μM), we found that cell viability was markedly impaired after 72 h of culture (results not shown), which may explain some of the apparent reduction in aldosterone levels reported beyond 48-h incubation. Our findings with the PPARα agonist fenofibrate are clearer with substantial increases in both aldosterone secretion and Cyp11B2 mRNA levels. Since the effect is blocked by the selective antagonist GW6471, fenofibrate is selectively acting through PPARα. It has been suggested previously that renin may be under PPARα regulation [32], but stimulation of aldosterone secretion by fenofibrate has not been reported before and was only revealed here from our IPA analysis. The IPA analysis also suggests that the PPARα pathway is deactivated in the KO mouse adrenal, which may explain at least in part why female mice have lower aldosterone levels. Further work is needed to explore the precise mechanism connecting PPARα signalling and aldosterone secretion, but it is clear that this effect is not simply explained by altered RXR or Pparα expression. A more detailed steroid profiling of the female KO mouse or a metabolomics approach may give important clues to the molecular pathways involved.

In summary, we have shown that Kcnj5 is functionally important for aldosterone secretion in the mouse but it is gender specific. The reduced aldosterone secretion in the female KO mouse may be explained by reduced PPARα signalling, which represents a novel control pathway for aldosterone. Hence, blockade of PPARα signalling may have translational potential in regulating human hyperaldosteronism, although this strategy may be sex limited.

• Mutations in the the potassium channel, KCNJ5, have been recently identified as a cause of hyperaldosteronism in human hypertension. To further explore the role of KCNJ5 in the adrenal gland, we looked at the phenotypic impact in the KO mouse.

• Female but not male mice with homozygous knock of KCNJ5 have reduced circulating levels of aldosterone. We used RNAseq to probe the molecular basis for this difference comparing WT and KO adrenals. The female adrenal has a much larger set of differentially expressed genes compared with the male (396 compared with 7) and pathway analysis of these genes suggested that PPARα was an important regulator of aldosterone synthesis.

• Our work suggests that KCNJ5 has a role in basal as well as pathological aldosterone secretion, although in the mouse, this basal effect may be sex limited. We have also identified a novel regulator of aldosterone secretion that has translational potential to the human adrenal.

I.H., L.L., R.A.M., and N.F. performed the experiments. I.H. and K.M.O. wrote the manuscript and analysed the data. K.M.O. conceived and supervised the work.

The authors declare that there are no competing interests associated with the manuscript.

The work was supported in part by the BHF PhD studentship [grant number FS_12_78_29875 (to I.H.)]. Raya Al Maskari was supported by a PhD studentship from the Oman Government.

Ang-II

angiotensin II

APA

aldosterone producing adenoma

CYP11B2

aldosterone synthase

DAB2

disabled 2

GIRK4

G-coupled inwardly rectifying K channel 4

FFPE

formaldehyde fixed paraffin embedded

IP

intraperitoneal

IPA

ingenuity pathway analysis

KCNJ

inwardly rectifying K channel CNJ type

KO

knockout

M

medulla

PA

primary aldosteronism

PO

oral

PPAR

peroxisome proliferator activated receptor

qPCR

quantitative PCR

WT

wild-type

ZF

zona fasciculata

ZG

zona glomerulosa