Salt-sensitive transcriptome of isolated kidney distal tubule cells

Abstract

In the distal kidney tubule, the steroid hormone aldosterone regulates sodium reabsorption via the epithelial sodium channel (ENaC). Most studies seeking to identify ENaC-regulating aldosterone-induced proteins have used transcriptional profiling of cultured cells. To identify salt-sensitive transcripts in an in vivo model, we used low-NaCl or high-NaCl diet to stimulate or suppress endogenous aldosterone, in combination with magnetic- and fluorescence-activated cell sorting to isolate distal tubule cells from mouse kidney for transcriptional profiling. Of the differentially expressed transcripts, 162 were more abundant in distal tubule cells isolated from mice fed low-NaCl diet, and 161 were more abundant in distal tubule cells isolated from mice fed high-NaCl diet. Enrichment analysis of Gene Ontology biological process terms identified multiple statistically overrepresented pathways among the differentially expressed transcripts that were more abundant in distal tubule cells isolated from mice fed low-NaCl diet, including ion transmembrane transport, regulation of growth, and negative regulation of apoptosis. Analysis of Gene Ontology molecular function terms identified differentially expressed transcription factors, transmembrane transporters, kinases, and G protein-coupled receptors. Finally, comparison with a recently published study of gene expression changes in distal tubule cells in response to administration of aldosterone identified 18 differentially expressed genes in common between the two experiments. When expression of these genes was measured in cortical collecting ducts microdissected from mice fed low-NaCl or high-NaCl diet, eight were differentially expressed. These genes are likely to be regulated directly by aldosterone and may provide insight into aldosterone signaling to ENaC in the distal tubule.

INTRODUCTION

Aldosterone is a steroid hormone secreted by the adrenal zona glomerulosa in response to angiotensin II or hyperkalemia (12, 13, 17). In the kidney, aldosterone acts on the aldosterone-sensitive distal tubule, consisting of the late distal convoluted tubule (DCT2), connecting tubule (CNT), and collecting duct (CD), to enhance sodium reabsorption via the epithelial sodium channel (ENaC). Aldosterone-induced proteins, which are transcriptionally regulated when aldosterone binds to its cytoplasmic receptor, the mineralocorticoid receptor (MR), mediate the action of aldosterone on ENaC. Although many aldosterone-induced proteins have been identified, genetic deletion of these proteins does not produce the renal salt wasting, hyperkalemia, and hypotension observed following genetic deletion of the MR (7, 42). This indicates that physiologically relevant aldosterone-induced proteins remain to be identified (46).

Poulsen and colleagues (33) recently identified aldosterone-induced genes in kidney cells sorted from mice treated with aldosterone. Cells expressing enhanced green fluorescent protein (eGFP) under the control of the promoter for TRPV5, a calcium channel expressed predominately in the CNT, were isolated by fluorescence-activated cell sorting (FACS) and used for RNA-Seq (23). They identified 290 aldosterone-induced genes, including Sgk1 and Tsc22d3 (which encodes GILZ), as well as 257 aldosterone-repressed genes. In an advance over previous studies that used in vitro cell culture models, they used cells rapidly isolated from the kidney for transcriptional profiling. However, administration of aldosterone has diverse metabolic effects, including hypokalemic metabolic alkalosis. We have shown that these metabolic changes have their own effects on transporters in the kidney that are not the direct result of MR activation (42).

To minimize the metabolic changes associated with the administration of aldosterone, we used low-NaCl or high-NaCl diets to chronically stimulate or suppress endogenous aldosterone, respectively. CNT/CD cells, rapidly isolated from mouse kidneys by a combination of magnetic- and fluorescence-activated cell sorting, were then used for RNA-Seq. Using this approach, we identified 323 differentially expressed transcripts. Of the differentially expressed transcripts, 162 were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet. These transcripts were compared with the aldosterone-induced genes identified by Poulsen and colleagues, leading to the identification of 18 differentially expressed genes in common between the two experiments. In cortical collecting ducts (CCDs) microdissected from mice fed low-NaCl or high-NaCl diet, eight of the 18 genes were found to be differentially expressed.

MATERIALS AND METHODS

Ethical statement.

All studies were approved by the Oregon Health & Science University Animal Care and Usage Committee (protocol #IP00000286) and followed the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Experimental animals.

Male mice (species: Mus musculus, strain: C57BL/6J) were used for all experiments. For measurement of physiological parameters and isolation of CNT/CD cells, the mice were 18 wk old with a mean weight of 28.3 g (range: 24.5–32.2 g). Eight mice (4 mice per group) were used for measurement of physiological parameters; six mice (3 mice per group) were used for isolation of CNT/CD cells for RNA-Seq. For immunohistochemistry and microdissection of CCDs, the mice were 15 to 16 wk old with a mean weight of 24.0 g (range: 18.5–27.7 g).

Housing and husbandry.

Mice were housed in a specific pathogen-free facility at Oregon Health & Science University and maintained on a 12 h:12 h light-dark cycle with free access to food and water. For immunohistochemistry, mice were fed a laboratory rodent diet (0.39% Na⁺, PicoLab 5L0D). For measurement of physiological parameters and isolation of CNT/CD cells, mice were fed either high-NaCl (2.4% Na⁺, TD.90230 Envigo) or low-NaCl (0.01–0.02% Na⁺, TD.90228 Envigo) diet for 5 days. For microdissection of CCDs, mice were fed high-NaCl diet for 7 days. Mice were then switched to low-NaCl diet or continued on the high-NaCl diet for 5 more days. Blood electrolyte measurements showed that plasma concentrations of sodium, potassium, chloride, and ionized calcium were similar between the protocol used for isolation of CNT/CD cells and the protocol used for microdissection of CCDs. Throughout each experiment, animal health was monitored by daily assessment of appearance and body weight.

Measurement of physiological parameters.

Electrolytes, urea nitrogen, total carbon dioxide, hematocrit, hemoglobin, and anion gap were measured in heparinized whole blood obtained by cardiac puncture, using an i-STAT Handheld Blood Analyzer with a CHEM8+ cartridge (Abbott Point of Care). The anion gap was calculated by the i-STAT Handheld Blood Analyzer using the equation: anion gap = Na⁺ − (Cl⁻+ total CO₂). Plasma aldosterone was measured by ELISA assay (IBL-America).

Immunohistochemistry.

Mice were anesthetized by intraperitoneal injection of ketamine cocktail (ketamine 50 mg/kg, xylazine 5 mg/kg, acepromazine 0.5 mg/kg). The kidneys were fixed by retrograde perfusion via the abdominal aorta with 50 ml 3% paraformaldehyde/PBS followed by 10 ml 300 mosmol/kgH₂O sucrose/PBS. After the kidneys were removed, they were stored overnight at 4°C in 800 mosmol/kgH₂O sucrose/PBS, cut transversely into 2 to 3 mm thick slices, and embedded in Tissue-Tek OCT compound (VWR). For immunohistochemistry, 5 µm thick sections were cut and mounted on Superfrost Plus glass slides (ThermoFisher Scientific).

For antibody labeling, sections were rehydrated for 3 × 5 min in PBS, washed for 30 min in 0.5% Triton X-100/PBS, then rinsed for 3 × 5 min in PBS. Following a 30 min incubation in 5% milk/PBS to reduce nonspecific staining, sections were incubated overnight at 4°C with the primary antibodies diluted in 5% milk/PBS: monoclonal L1-CAM rat antibody (MAB5674, R&D Systems) diluted 1:500, polyclonal NCC rabbit antibody [Bostanjoglo et al. (5) of the Ellison laboratory] diluted 1:1,000, and polyclonal AQP2 goat antibody (Santa Cruz Biotechnology) diluted 1:500. Sections were washed 3 × 5 min in PBS, then incubated with the secondary antibodies anti-goat Alexa Fluor 488, anti-rat Alexa Fluor 594, and anti-rabbit Alexa Fluor 647 for 45 min at room temperature. Following incubation with the secondary antibodies, sections were again washed 3 × 5 min in PBS. Sections were then mounted with ProLong Diamond Antifade Mountant (ThermoFisher Scientific) and imaged under a Nikon A1R laser-scanning confocal microscope.

Determination of kidney tubule segment-specific expression of L1-CAM.

Expression of L1-CAM, also known as neural cell adhesion molecule L1, was previously shown to be restricted to the CD within the human kidney (21). To evaluate the expression of L1-CAM along the tubule in mouse kidney, we used immunofluorescence staining. CNTs were identified by cytoplasmic staining for calbindin-D_28K (CALB1), a calcium binding protein expressed predominantly in the CNT. CNT/CDs were identified by apical staining for aquaporin 2 (AQP2), a water channel expressed in CD principal cells. Basolateral staining for L1-CAM was observed in both CNTs and CDs, demonstrating L1-CAM also has kidney tubule segment-specific expression in the distal tubule of mouse kidney (Fig. 1, A and B).

Fig. 1.

The cell adhesion protein L1-CAM was used to isolate connecting tubule (CNT)/collecting duct (CD) cells from mouse kidney. A: localization of L1-CAM (red), aquaporin (AQP)2 (green), and NCC (magenta) in the kidney cortex of wild-type mice. *Tubules that are L1-CAM and AQP2 positive. >Tubules that are NCC positive. B: localization of L1-CAM (red), NCC (green), and calbindin-D_28K (CALB1, magenta) in the kidney cortex of wild-type mice. #Tubules that are L1-CAM and CALB1 positive. Arrowhead indicates tubules that are NCC positive. C: procedure for isolating L1-CAM+ kidney cells. D–F: example analysis of L1-CAM⁺ kidney cells. The antibodies for this experiment were CD31-APC/CD45-APC/label check reagent-PE; the live/dead discriminator was propidium iodide. Acquisition and analysis were performed on a Becton Dickinson Influx Cell Sorter. Gating was as follows: Forward scatter/side scatter to exclude cellular debris (D), Trigger pulse width/forward scatter to exclude non-single cells (E), live gate (PI negative, to select viable cells) (F), and APC/PE to exclude L1-CAM⁺CD31⁺ and L1-CAM⁺CD45⁺ cells (G). FACS, fluorescence-activated cell sorting; MACS, magnetic-activated cell sorting.

Isolation of CNT/CD cells.

Although cell type-specific markers for 43 kidney cell types can be detected in whole-kidney RNA-Seq data, transcriptional changes observed with whole-kidney RNA-Seq cannot be attributed to a particular cell type (11). Proximal tubule cells represent the dominant cell type in the kidney and contribute the majority of whole-kidney mRNA (11). By comparison, CNT/CD cells represent a small percentage of the total kidney cell population (31). Thus, to detect transcriptional changes specifically in CNT/CD cells, we developed a protocol to isolate CNT/CD cells from mouse kidney by magnetic-activated cell sorting (MACS) and FACS following mechanical and enzymatic separation of kidney tissue (Fig. 1C).

Kidneys were cut into small pieces with a scalpel, then incubated with collagenase type II (1.0 mg/ml, GIBCO) and DNase type I (Thermo Scientific) in Hanks’ balanced salt solution with Ca²⁺ and Mg²⁺ (37°C for 30 min). After repetitive pipetting with a serological pipette, the suspension was filtered through a 70 µm nylon mesh cell strainer. The cells were centrifuged at 300 g for 10 min and resuspended in flow cytometry butter (1.86 mg/ml EDTA and 5.0 mg/ml BSA in 1× PBS). The cells were incubated with anti-L1-CAM (CD171) microbeads (130-101-548, Miltenyi Biotec) for 15 min at 4°C, then centrifuged at 300 g for 10 min and resuspended in flow cytometry buffer. The cells were then incubated with PE-conjugated label check reagent (130-098-866, Miltenyi Biotec), APC-conjugated anti-CD31 antibody (130-102-571, Miltenyi Biotec), APC-conjugated anti-CD45 antibody (130-102-544, Miltenyi Biotec), and propidium iodide (130-093-233, Miltenyi Biotec) for 10 min at 4°C. The PE-conjugated label check reagent, which binds the anti-L1-CAM microbeads, allowed the selection of L1-CAM⁺ cells by FACS in addition to MACS. The APC-conjugated anti-CD31 and anti-CD45 antibodies allowed the exclusion of endothelial and immune cells, which can express L1-CAM, by FACS (14, 27). Propidium iodide, a fluorescent intercalating agent, allowed the exclusion of nonviable cells by FACs. Following cell labeling, the cells were centrifuged at 300 g for 10 min and resuspended in flow cytometry buffer.

Following selection of L1-CAM⁺ cells from the cell suspension by the Posseld-positive selection program on an AutoMACS Pro Separator (Miltenyi), CNT/CD cells (L1-CAM⁺CD31^-CD45^- cells; PE channel) were separated from cellular debris (Fig. 1D), nonsingle cells (Fig. 1E), and L1-CAM⁺ endothelial and immune cells (Fig. 1G) on a Becton Dickinson Influx cell sorter. Viable cells were selected by excluding cells stained with the live/dead discriminator propidium iodide (Fig. 1F). L1-CAM⁺CD31^-CD45^- cells were sorted directly into QIAzol lysis reagent (79306; Qiagen). To confirm that the L1-CAM⁺CD31^-CD45^- cells isolated by this method were enriched for CNT/CD cells, quantitative PCR (qPCR) was used to compare expression of Slc34a1 (a proximal tubule marker) and Aqp2 in unsorted versus sorted kidney cells. Compared with unsorted kidney cells, sorted kidney cells had sixfold higher expression of Aqp2 and 151-fold lower expression of Slc34a1, confirming that the sorted cells were enriched for CNT/CD cells.

RNA extraction, library construction, and RNA sequencing.

Total RNA was extracted from the CNT/CD cells with an RNeasy Plus Universal Mini Kit (73404, Qiagen), according to the manufacturer’s protocol. The RNA quality was assessed with an Agilent 2100 Bioanalyzer. Library preparation and RNA sequencing were performed using Illumina’s standard TruSeq RNA protocol with polyA selection. PolyA⁺ RNA was isolated using oligo-DT magnetic beads and chemically fragmented to 200–300 base pairs. The fragmented polyA⁺ RNA was converted to double-stranded cDNA by random hexamer priming. The double-stranded cDNA was treated to blunt the ends, and a single “A” base was added to the blunt ends. Proprietary Illumina adaptors were ligated to the cDNA fragments. Libraries were amplified using PCR and cleaned using AMPure XP beads (Agencourt). Library quality was assessed using an Agilent 2100 Bioanalyzer. After quantification of individual libraries by real-time PCR, equimolar library pools were prepared. The final concentration of the multiplexed pools was verified by real-time PCR. Libraries were sequenced with a 100-cycle single read protocol on the Illumina HiSeq 2500 platform.

Data processing and differential expression analysis.

Base call files were converted to fastq files using Bcl2Fastq (Illumina), and the data were assessed for quality using fastQC. The reads were aligned to the mouse genome (mm10 assembly) using Subread (24); the aligned reads were mapped to transcripts using featureCounts (25). Transcripts with fewer than five reads across all samples were excluded from the differential expression analysis. Differentially expressed transcripts were identified using the edgeR package (36). A negative binomial model for count data was applied using two estimates of dispersion: common dispersion, which uses a single estimate of dispersion for all transcripts and detects transcripts with more outlier values, and tagwise dispersion, which estimates dispersion for each transcript and detects transcripts with more consistent results. P values were adjusted for multiple comparisons by the Benjamini Hochberg false discovery rate (FDR) procedure (3). The raw and analyzed files were uploaded to Gene Expression Omnibus under the accession GSE122995.

Gene Ontology enrichment analysis.

Gene Ontology (GO) enrichment analysis was performed using BiNGO (28). Differentially expressed transcripts with greater abundance in low-NaCl diet versus high-NaCl diet were used as the test set; the whole annotation was used as the reference set. Overrepresentation of GO Biological Process terms in the differentially expressed transcripts was determined by the hypergeometric test; P values were adjusted for multiple comparisons by the Benjamini Hochberg FDR procedure (3).

Microdissection of cortical collecting ducts.

CCDs were dissected from mouse kidney by the technique developed by the Knepper laboratory (45). The microdissected CCDs were immediately transferred to ice-cold QIAzol lysis reagent (79306, Qiagen) and then stored at −80°C until RNA extraction.

RNA extraction and quantitative PCR.

Total RNA was isolated from the microdissected CCDs using an RNeasy Plus Universal Mini Kit (73404, Qiagen), according to the manufacturer’s protocol. The RNA was then reverse-transcribed into cDNA using a High-Capacity RNA-to-cDNA Kit (4387406, ThermoFisher Scientific). qPCR was performed using TaqMan reagents (ThermoFisher Scientific); 18S rRNA was used as the reference gene. Gene expression data were calculated by the ΔCt method.

Statistical analyses.

Statistical analyses of transcriptome data are described under “data processing and differential expression analysis” and “Gene Ontology enrichment analysis.” All other statistical tests were performed as indicated in the figure legends using Prism 7 for Mac OS X (GraphPad Software).

RESULTS

L1-CAM expression in the mouse kidney.

We investigated the feasibility of using L1-CAM as a cell surface marker of the distal tubule for sorting cells. We first examined the colocalization of L1-CAM expression with the principal cell CD and CNT marker AQP2; L1-CAM and AQP2 were extensively colocalized in the same cells (Fig. 1A). In addition to colabeling the same tubules, cells within a tubule that were negative for AQP2 were also negative for L1-CAM. This suggests that L1-CAM is expressed in principal cells and not in intercalated cells (22). We colabeled kidney sections with L1-CAM and the CNT marker CALB1 to determine whether L1-CAM is expressed in the CNT (Fig. 1B). We observed that there are tubules that are positive for both L1-CAM and CALB1, suggesting that L1-CAM is also expressed in the CNT. Additionally, we observed L1-CAM-positive tubules in the cortex, outer medulla, and inner medulla (data not shown). We observed no L1-CAM-positive tubules that were also positive for the distal convoluted tubule marker NCC. We concluded that L1-CAM is expressed in both the CNT and the CD and is an appropriate cell-surface marker for sorting distal tubule cells.

Physiological effects of dietary sodium intake.

C57BL/6J mice were fed either low-NaCl or high-NaCl diet for 5 days, causing a significant difference in plasma aldosterone without causing other metabolic changes (Fig. 2A). Specifically, blood electrolytes, total carbon dioxide, blood urea nitrogen, hematocrit, hemoglobin, and anion gap were similar between the two groups (Fig. 2, B–J). Urinary electrolytes were not measured. We have previously published data showing changes in urinary electrolytes in response to changes in diet (42). These data indicate that at 5 days, differences in urinary electrolytes reflect differences in dietary intake rather than differences in renal function.

Fig. 2.

C57BL/6J mice fed low-NaCl (0.01–0.02% Na⁺) diet had markedly elevated plasma aldosterone compared with mice fed high-NaCl (2.4% Na⁺) diet. However, other physiological parameters were similar between the two groups. Aldosterone was measured in plasma; other parameters were measured in whole blood. A: aldosterone, B: sodium, C: potassium, D: chloride, E: total carbon dioxide, F: ionized calcium, G: glucose, H: hematocrit, I: hemoglobin, and J: anion gap. Data in A–J displayed as means ± SE; comparisons by unpaired t-test with two-tailed P value. ***P < 0.0001.

RNA-Seq profiling of CNT/CD cells.

CNT/CD cells, isolated as above from mouse kidneys, were used to prepare cDNA libraries for single-end sequencing. The cDNA libraries were sequenced to a mean depth of 113 million reads (range: 106 million–126 million) per sample with a mean alignment of 76.3% (range: 72.2–79.1%). The aligned reads mapped to 59,121 transcripts. Transcripts with fewer than five reads across all samples or with log₂ counts per million reads ≤1 were excluded from the differential expression analysis. Of the remaining 7,273 transcripts, 323 were differentially expressed (FDR < 0.05) between the low-NaCl and high-NaCl diet groups. Of the differentially expressed transcripts, 162 were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet, and 161 were more abundant in the CNT/CD cells isolated from mice fed high-NaCl diet.

Bioinformatic analysis of the CNT/CD cell transcriptome.

GO enrichment analysis was used to identify biological processes associated with differentially expressed transcripts that were more abundant in CNT/CD cells isolated from mice fed low-NaCl diet. Biological Process terms that were statistically overrepresented among these transcripts were identified by the bioinformatics application BiNGO (28). Ion transmembrane transport (GO:0034220), regulation of growth (GO:0040008), and negative regulation of apoptosis (GO:0043066) were among the overrepresented Biological Process terms. Of the genes are associated with each of these terms; those that were differentially expressed are illustrated by the heat maps shown in Fig. 3. The heat maps show higher expression of these genes in the CNT/CD cells isolated from mice fed low-NaCl diet.

Fig. 3.

Heat maps illustrating transcripts associated with the Gene Ontology (GO) classifications “GO:0034220 ion transmembrane transport” (A), “GO:0040008 regulation of growth” (B), and “GO:0043066 negative regulation of apoptosis” (C) that were differentially expressed in L1-CAM⁺ kidney cells isolated from C57BL/6J mice fed low-NaCl (0.01–0.02% Na⁺) versus high-NaCl (2.4% Na⁺) diet.

GO Molecular Function terms were used to identify differentially expressed transcripts associated with DNA-binding transcription factor activity (GO:0003700), transmembrane transporter activity (GO:0022857), kinase activity (GO:0016301), and G protein-coupled receptor (GPCR) activity (GO:0004930).

DNA-binding transcription factors interact selectively with promoters and enhancers to alter gene transcription. Transcription factors determine the transcriptional response to physiological stimuli, including changes in plasma aldosterone. Tsc22d3 (which encodes GILZ) is itself a transcription factor and was identified as an aldosterone-induced transcript in mpkCCD_c14 cells (35, 38). The GO molecular function term DNA-binding transcription factor activity (GO:0003700) was used to identify additional transcription factors that were differentially expressed in the CNT/CD cells (Table 1). FoxI1 and Dmrt2, which were previously identified as intercalated cell-specific transcription factors within the CD, were among the differentially expressed transcription factors that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet (8, 22). Bcl6 and Egr1, which were differentially expressed in the kidneys of male rats with developmentally programmed hypertension, were among the differentially expressed transcription factors that were more abundant in the CNT/CD cells isolated from mice fed high-NaCl diet (41).

Table 1.

Differentially expressed genes included in GO:0003700 DNA-binding transcription factor activity

Gene Symbol	Gene Name	logFC	logCPM	P Value	FDR
FoxI1	Forkhead box I1	−1.39	8.06	4.09E-06	3.82E-04
Dmrt2	Doublesex and mab-3 related transcription factor 2	−1.33	4.06	1.28E-05	8.85E-04
Ankrd1	Ankyrin repeat domain 1	−1.34	3.28	1.30E-05	8.93E-04
Tsc22d3	TSC22 domain family member 3	−1.41	1.23	1.74E-05	1.14E-03
Zfp750	Zinc finger protein 750	−1.02	3.02	9.42E-04	2.34E-02
Cited2	Cbp/p300 interacting transactivator with Glu/Asp-rich carboxy-terminal domain 2	1.03	2.45	9.72E-04	2.39E-02
Foxm1	Forkhead box M1	1.04	2.17	9.03E-04	2.27E-02
Deaf1	Deaf1	1.07	2.36	5.80E-04	1.63E-02
Tfap4	Transcription factor AP-4	1.09	3.40	3.51E-04	1.12E-02
Scx	Scleraxis bHLH transcription factor	1.09	4.46	3.08E-04	1.01E-02
Bcl6	B cell CLL/lymphoma 6	1.14	4.11	1.74E-04	6.56E-03
Klf2	Krüppel-like factor 2	1.28	1.23	1.00E-04	4.44E-03
Jun	Jun proto-oncogene	1.46	7.77	1.36E-06	1.69E-04
Id3	Inhibitor of DNA binding 3	1.63	7.02	8.01E-08	1.61E-05
Junb	JunB proto-oncogene	1.63	6.68	7.56E-08	1.55E-05
Nr4a1	Nuclear receptor subfamily 4 group A member 1	1.94	5.34	2.89E-10	1.46E-07
Ier2	Immediate early response 2	2.03	7.34	4.21E-11	2.67E-08
Egr1	Early growth response 1	2.16	7.98	2.62E-12	2.34E-09
Fos	Fos proto-oncogene, AP-1 transcription factor subunit	2.49	8.14	1.46E-15	2.87E-12
Fosb	FosB proto-oncogene, AP-1 transcription factor subunit	2.91	5.50	7.68E-20	3.78E-16

CPM, counts per million reads; FC, fold change; FDR, false discovery rate.

Transmembrane transporters are involved in the transport of solutes across the plasma membrane. In the CD, aldosterone regulates transmembrane transport to determine the final concentrations of urinary Na⁺ and K⁺. The GO molecular function term transmembrane transporter activity (GO:0022857) was used to identify transmembrane transporters that were differentially expressed in the CNT/CD cells (Table 2). Several of the differentially expressed transmembrane transporters that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet, including two subunits of the proton ATPase (Atp6v0d2 and Atp6v1g3), aquaporin 6 (Aqp6), anion exchange protein 1 (Slc4a1), and pendrin (Slc26a4), were previously found to be expressed predominantly in intercalated cells by single-cell RNA profiling of CD cells isolated from mouse kidney (8). In contrast, several of the differentially expressed transmembrane transporters that were more abundant in the CNT/CD cells isolated from mice fed high-NaCl diet, including aquaporins 2 and 3 (Aqp2 and Aqp3) and the renal outer medullary potassium channel ROMK (Kcnj1), were previously found to be expressed predominantly in principal cells in the same study (8).

Table 2.

Differentially expressed genes included in GO:0022857 transmembrane transporter activity

Gene Symbol	Gene Name	logFC	logCPM	P Value	FDR
S100a6	S100 calcium-binding protein A6	−1.71	5.48	2.05E-08	4.92E-06
Slc4a1	Solute carrier family 4 member 1	−1.70	8.02	2.12E-08	5.03E-06
Aqp6	Aquaporin 6	−1.57	4.45	2.79E-07	4.17E-05
Atp6v1 g3	ATPase H+-transporting V1 subunit G3	−1.44	8.66	1.83E-06	2.06E-04
Slc1a4	Solute carrier family 1 member 4	−1.57	1.01	3.05E-06	3.12E-04
Slc12a1	Solute carrier family 12 member 1	−1.40	4.85	3.84E-06	3.65E-04
Atp6v0d2	ATPase H+-transporting V0 subunit D2	−1.35	7.31	7.56E-06	6.08E-04
Lrrc8b	Leucine-rich repeat-containing 8 VRAC subunit B	−1.36	1.00	4.28E-05	2.29E-03
Ank	Ankyrin 1	−1.20	3.34	8.51E-05	3.89E-03
Aqp6	Aquaporin 6	−1.12	7.62	1.90E-04	7.01E-03
Kcnk5	Potassium two pore domain channel subfamily K member 5	−1.12	3.95	2.11E-04	7.69E-03
Slc26a4	Solute carrier family 26 member 4	−1.07	9.21	3.61E-04	1.14E-02
Slc27a2	Solute carrier family 27 member 2	−1.02	4.97	6.76E-04	1.83E-02
Atp6v0a4	ATPase H+-transporting V0 subunit A4	−1.02	3.66	7.54E-04	1.99E-02
Fxyd2	FXYD domain-containing ion transport regulator 2	−1.02	3.02	9.42E-04	2.34E-02
Aqp3	Aquaporin 3	1.28	10.54	2.03E-05	1.28E-03
Aqp2	Aquaporin 2	1.29	12.27	1.70E-05	1.13E-03
Nipal1	NIPA-like domain containing 1	1.40	2.14	9.11E-06	7.01E-04
Kcns1	Potassium voltage-gated channel modifier subfamily S member 1	1.47	2.97	2.30E-06	2.48E-04
Trpv5	Transient receptor potential cation channel subfamily V member 5	1.45	5.54	1.78E-06	2.02E-04
Kcnj1	Potassium voltage-gated channel subfamily J member 1	2.82	1.08	3.81E-15	6.83E-12
Atp12a	ATPase H+/K+-transporting nongastric alpha2 subunit	3.87	2.34	2.68E-27	5.28E-23

Kinases catalyze the transfer of phosphate groups to substrate molecules and have been shown to play a role in genomic as well as rapid nongenomic actions of aldosterone. The GO molecular function term kinase activity (GO:0016301) was used to identify kinases that were differentially expressed in the CNT/CD cells (Table 3). Sgk1, or serum/glucocorticoid regulated kinase 1, has been consistently identified in gene expression profiles of cell culture models and mouse kidney following aldosterone treatment (4, 9, 20, 26, 29, 30, 37). In the present study, Sgk1 was among the differentially expressed kinases that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet. Two other differentially expressed kinases that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet, Map3k6 and Mapk11, are members of the mitogen-activated protein (MAP) kinase family. Among the kinases that were more abundant in the CNT/CD cells isolated from mice fed high-NaCl diet was Grk4, which regulates renal dopamine receptors and has been associated with the development of hypertension (34).

Table 3.

Differentially expressed genes included in GO:0016301 kinase activity

Gene Symbol	Gene Name	logFC	logCPM	P Value	FDR
Sgk1	Serum/glucocorticoid-regulated kinase 1	−3.42	1.84	1.74E-22	1.72E-18
Map3k6	Mitogen-activated protein kinase kinase kinase 6	−2.27	1.05	6.19E-11	3.81E-08
Mapk11	Mitogen-activated protein kinase 11	−1.50	2.54	1.65E-06	1.90E-04
Mmd2	Monocyte to macrophage differentiation associated 2	−1.04	3.90	5.85E-04	1.63E-02
Trib2	Tribbles pseudokinase 2	−1.03	3.44	7.54E-04	1.99E-02
Insrr	Insulin receptor-related receptor	−0.97	8.03	1.12E-03	2.63E-02
Efnb2	Ephrin B2	0.93	4.48	1.91E-03	3.85E-02
Ccnd1	Cyclin D1	0.93	8.05	1.81E-03	3.74E-02
Grk4	G protein-coupled receptor kinase 4	1.04	1.12	1.45E-03	3.18E-02
Ulk4	Unc-51 like kinase 4	1.13	1.88	3.41E-04	1.10E-02
Tk1	Thymidine kinase 1	1.37	1.35	2.86E-05	1.68E-03
Ccna2	Cyclin A2	1.66	1.16	6.48E-07	8.74E-05
Cdk1	Cyclin-dependent kinase 1	1.71	2.05	9.67E-08	1.84E-05
Pbk	PDZ-binding kinase	2.09	1.94	1.80E-10	9.86E-08

The GO molecular function term GPCR activity (GO:0004930) was used to identify GPCRs that were differentially expressed in the CNT/CD cells (Table 4). Oxoglutarate receptor 1 (Oxgr1) was among the differentially expressed GPCRs that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet. This α-ketoglutarate-activated GPCR plays a role in activation of the Na⁺-independent Cl⁻/${\text{HCO}}_3^{-}$ exchanger pendrin (18). Pendrin, in turn, modulates ENaC abundance and activity by altering urinary concentrations of ${\text{HCO}}_3^{-}$ and ATP (19, 32).

Table 4.

Differentially expressed genes included in GO:0004930 G protein-coupled receptor activity

Gene Symbol	Gene Name	logFC	logCPM	P Value	FDR
Tshr	Thyroid-stimulating hormone receptor	−1.95	2.80	7.09E-10	3.17E-07
Bdkrb2	Bradykinin receptor B2	−1.82	2.64	8.50E-09	2.39E-06
Avpr1a	Arginine vasopressin receptor 1A	−1.34	6.29	8.92E-06	6.94E-04
Sfrp1	Secreted frizzled-related protein 1	−1.33	9.82	1.03E-05	7.69E-04
Oxgr1	Oxoglutarate receptor 1	−1.32	7.02	1.10E-05	8.02E-04
F2r	Coagulation factor II thrombin receptor	−1.38	2.29	1.13E-05	8.12E-04
Adora2b	Adenosine A2b receptor	−1.31	2.73	2.39E-05	1.48E-03
Gprc5a	G protein-coupled receptor class C group 5 member A	−0.98	5.88	1.05E-03	2.52E-02
Fzd7	Frizzled class receptor 7	1.04	6.83	4.95E-04	1.44E-02
Gprc5b	G protein-coupled receptor class C group 5 member B	1.11	6.24	2.27E-04	8.10E-03
Hpgd	15-hydroxyprostaglandin dehydrogenase	1.22	2.73	8.46E-05	3.88E-03
Fzd1	Frizzled class receptor 1	1.49	5.50	8.65E-07	1.13E-04

Comparison of salt-sensitive and aldosterone-induced genes.

Both the present study and the recently published study by Poulsen and colleagues (33) used FACS to isolate cells from the kidney distal tubule for gene expression profiling. The cell selection approach, immunolabeling of L1-CAM versus expression of eGFP under the TRPV5 promoter, and the means of manipulating plasma aldosterone, changing dietary NaCl versus administration of aldosterone by osmotic minipump, differed between the two studies. However, the resulting plasma aldosterone concentrations were similar (548 ± 58 pg/ml in the present study versus 506 ± 61 pg/ml in the study by Poulsen et al.). We compared the differentially expressed transcripts that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet with the aldosterone-induced genes identified by Poulsen et al., identifying 18 genes in common (Fig. 4A). A heat map illustrating differential expression of these genes between the low-NaCl versus high-NaCl diet groups is shown in Fig. 4B.

Fig. 4.

A: differentially expressed genes that were more abundant in collecting duct cells isolated from mice fed low-NaCl diet were compared with aldosterone-induced genes [Poulsen et al., 2018 (33)] to identify differentially expressed genes in common between the two conditions. B: heat map illustrating differential expression in L1-CAM+ kidney cells isolated from C57BL/6J mice fed low-NaCl (0.01–0.02% Na⁺) versus high-NaCl (2.4% Na⁺) diet of the 18 differentially expressed genes in common between the two conditions.

Expression of low-NaCl diet-induced transcripts in microdissected cortical collecting ducts.

We identified 18 differentially expressed transcripts that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet and that were induced by aldosterone in the study by Poulsen and colleagues (33). Differential expression of these genes was evaluated in an independent experiment using C57BL/6J mice fed either low-NaCl or high-NaCl diet to manipulate endogenous aldosterone. CCDs were microdissected from the kidneys, allowing additional confirmation that gene expression changes observed in the sorted cells can be attributed to the CCD. RNA was purified from microdissected CCDs and gene expression was assessed by qPCR. Gene expression was higher in the low-NaCl diet group compared with the high-NaCl diet group for eight of the 18 genes: Ank, Cited4, Crip1, Hmx2, Insrr, Sgk1, Slc26a4, and Spink8 (Fig. 5, A–R).

Fig. 5.

Expression of the 18 differentially expressed genes identified in Fig. 4 was measured in an independent experiment by quantitative (q)PCR in cortical collecting ducts microdissected from C57BL/6J mice fed low-NaCl (0.01–0.02% Na⁺) diet or high-NaCl (2.4% Na⁺) diet. RNA was purified from microdissected cortical collecting ducts, and expression of the following genes was measured by qPCR: A: Ank, B: Atp6v0d2, C: Atp6v1g3, D: Cdr2, E: Cited4, F: Crip1, G: Cry1, H: FoxI1, I: Fxyd2, J: Hepacam2, J: Hmx2, K: Insrr, L: Lcn2, M: Sgk1, N: Slc26a4, O: Spink8, P: Tsc22d3, and Q: Wscd2.

DISCUSSION

Aldosterone modulates the activity of the epithelial sodium channel ENaC by regulating a set of target genes encoding aldosterone-induced proteins. Identifying the key aldosterone-induced proteins that are essential for mediating aldosterone activation of ENaC has proved challenging. Most studies seeking to identify aldosterone-induced transcripts have used gene expression profiling of in vitro model systems (9, 20, 30, 35, 39). These studies identified Sgk1 and Tsc22d3 (which encodes GILZ) as aldosterone-induced transcripts. In Xenopus oocytes, co-expression of Sgk1 with ENaC was found to increase localization of the channel at the plasma membrane and increase channel-mediated Na⁺ current many fold (1, 9, 26, 30, 37). Evidence from Sgk1 knockout mice does indicate that SGK1 plays a role in processing and trafficking of ENaC; yet, ENaC activity does not differ between Sgk1 knockout and wild-type mice, indicating that other factors are also involved in regulation of ENaC (15, 16, 46). Co-expression of Tsc22d3 with ENaC in Xenopus oocytes increases channel activity. Yet, Tsc22d3 knockout mice develop hypernatremia and hypokalemia when fed low-NaCl diet, which contrasts with the hyponatremia and hyperkalemia observed in kidney-specific mineralocorticoid receptor knockout mice under the same conditions (38, 40). As genetic deletion of aldosterone signaling molecules that have been identified to date does not recapitulate the phenotype of mice lacking mineralocorticoid receptors in the kidney (7, 42), key components of aldosterone signaling pathways in vivo remain to be identified.

Poulsen and colleagues recently identified transcripts in kidney cells sorted from mice treated with aldosterone (33). Cells expressing eGFP under the control of the TRPV5 promoter were isolated by FACS and used for RNA-Seq. This approach identified 290 aldosterone-induced genes and 257 aldosterone-repressed genes; among the aldosterone-induced genes were those encoding SGK1 and GILZ. As the administration of aldosterone has diverse metabolic effects, including hypokalemic metabolic alkalosis, and several consequences of aldosterone administration were recently shown to be secondary to its metabolic effects, we manipulated dietary NaCl, an approach that alters plasma aldosterone while minimizing metabolic changes. Additionally, to avoid the limitations of cell culture models, we coupled this approach with rapid isolation of renal collecting duct cells for RNA sequencing. Using this approach, we identified 323 differentially expressed transcripts.

To assess the processes that underlie adaptation of the collecting duct to low-NaCl diet, we identified GO Biological Process terms that were statistically overrepresented among the differentially expressed transcripts that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet. Transcripts associated with the term “ion transmembrane transport” were statistically overrepresented among these transcripts. This is consistent with evidence from kidney-specific MR knockout mice that the role of the MR in mediating aldosterone effects on ion homeostasis results primarily from its action in the kidney epithelium (7, 42). Transcripts associated with the terms “regulation of growth” and “negative regulation of apoptosis” were also statistically overrepresented among the differentially expressed transcripts that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet. Remodeling of the CD is associated with cell growth and proliferation and has been described in a variety of physiological and pathophysiological states (2, 6, 10, 44). Overrepresentation of transcripts associated with regulation of growth and negative regulation of apoptosis raises the possibility that remodeling plays a role in the adaptation of the CD to low-NaCl diet.

Although our original aim was to isolate principal cells, the transcriptome data clearly demonstrated that the L1-CAM⁺CD31^-CD45^- kidney cell population included both principal and intercalated cells. Using publicly available data from the single cell gene expression profile of the CD published by Chen and colleagues (8), we were able to differentiate intercalated and principal cell-specific transcripts in our CNT/CD cell transcriptome data. The differentially expressed transcripts that were more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet included many intercalated cell-specific transcripts. These include the transcription factor FoxI1 as well as the transmembrane transporters Atp6v0d2, Atp6v1g3, Aqp6, Slc4a1, and Slc26a4. In addition, the α-ketoglutarate-activated GPCR Oxgr1, which plays a role in activation of the Na⁺-independent Cl⁻/${\text{HCO}}_3^{-}$ exchanger pendrin, was more abundant in the CNT/CD cells isolated from mice fed low-NaCl diet (18). Thus, the adaptation of the CD to low-NaCl diet involves transcriptional changes in intercalated cells as well as principal cells.

Several differentially expressed transcripts that were more abundant in CD cells isolated from mice fed high-NaCl diet have been previously associated with the development of hypertension. Two of these transcripts, Bcl6 and Egr1, were identified by gene expression profiling of kidneys from male rats with developmentally programmed hypertension (41). Bcl6 was also differentially expressed in the kidneys of spontaneously hypertensive rats (43). A third differentially expressed transcript that was more abundant in the CNT/CD cells isolated from mice fed high-NaCl diet, Grk4, encodes a kinase that regulates renal dopamine receptors and has also been associated with the development of hypertension (34). Given that several of the differentially expressed transcripts that were more abundant in the CNT/CD cells isolated from mice fed high-NaCl diet have been associated with the development of hypertension, it is possible that one mechanism linking high-sodium diet and hypertension is transcriptional changes in the distal tubule.

The aldosterone-induced transcripts identified by gene expression profiling of in vitro model systems have not fully explained the action of aldosterone on ENaC in vivo, suggesting that key components of aldosterone signaling pathways in vivo differ from what has been observed in cell culture. To identify aldosterone-induced transcripts in vivo, both the present study and the recently published study by Poulsen et al. (33) used FACS to isolate cells from the kidney distal tubule for gene expression profiling. Plasma potassium, which may have its own effects on gene expression in the distal tubule, differed between the two experimental conditions; however, the elevation in plasma aldosterone was similar. Additionally, while both cell sorting protocols isolated both CNT and CD cells, differences in the cell-selection strategies (L1-CAM versus TRPV5) may have resulted in slightly different populations of cells. This may have limited the number of differentially expressed genes in common between the two experiments. However, the genes that are differentially expressed in both conditions are most likely regulated directly by aldosterone. While the differential expression of some genes identified by this approach was not validated by qPCR, the differential expression of many genes was confirmed. Thus, we suggest that comparing physiologically relevant transcriptome data obtained from different models to interrogate aldosterone signaling may identify novel genes and pathways for further study. In future studies evaluating the signaling pathway by which aldosterone activates ENaC in the kidney distal tubule, it will be valuable to use the kidney-specific MR knockout mouse to distinguish between aldosterone-induced transcripts that are regulated by metabolic changes in the setting of high aldosterone versus aldosterone-induced transcripts that are directly regulated by aldosterone via the MR.

GRANTS

This work has been supported by the National Institutes of Health Grant F30 DK-114980 (to E. A. Swanson), the American Heart Association 15POST25710234 (to J. W. Nelson), the U.S. Department of Veterans Affairs Grant I01X002228 (to D. H. Ellison) and Fondation LeDucq Grant 17CVD05 (Transatlantic Network of Excellence).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.A.S., J.W.N., and D.H.E. conceived and designed research; E.A.S., J.W.N., and K.J.E. performed experiments; E.A.S., J.W.N., S.J., and S.M. analyzed data; E.A.S., J.W.N., and D.H.E. interpreted results of experiments; E.A.S. prepared figures; E.A.S. drafted manuscript; E.A.S., J.W.N., and D.H.E. edited and revised manuscript; E.A.S., J.W.N., S.J., K.J.E., C.-L.Y., S.M., and D.H.E. approved final version of manuscript.