Sodium Deficiency Regulates Rat Adrenal Zona Glomerulosa Gene Expression

Abstract

Aldosterone is the primary adrenocortical hormone regulating sodium retention, and its production is under the control of the renin-angiotensin-aldosterone system (RAAS). In vitro, angiotensin II can induce aldosterone production in adrenocortical cells without causing cell proliferation. In vivo, a low-sodium diet activates the RAAS and aldosterone production, at least in part, through an expansion of the adrenal zona glomerulosa (zG) layer. Although these mechanisms have been investigated, RAAS effects on zG gene expression have not been fully elucidated. In this study, we took an unbiased approach to define the complete list of zG transcripts involved in RAAS activation. Adrenal glands were collected from 11-week old Sprague-Dawley rats fed either sodium-deficient (SDef), normal sodium (NS), or high-sodium (HS) diet for 72 hours, and laser-captured zG RNA was analyzed on microarrays containing 27 342 probe sets. When the SDef transcriptome was compared with NS transcriptome (SDef/NS comparison), only 79 and 10 probe sets were found to be up- and down-regulated more than two-fold in SDef, respectively. In SDef/HS comparison, 201 and 68 probe sets were up- and down-regulated in SDef, respectively. Upon gene ontology (GO) analysis of these gene sets, we identified three groups of functionally related GO terms: cell proliferation-associated (group 1), response to stimulus-associated (group 2), and cholesterol/steroid metabolism-associated (group 3) GO terms. Although genes in group 1 may play a critical role in zG layer expansion, those in groups 2 and 3 may have important functions in aldosterone production, and further investigations on these genes are warranted.

Aldosterone, which is synthesized in the adrenal zona glomerulosa (zG) under the control of the renin-angiotensin-aldosterone system (RAAS), is the most potent mineralocorticoid involved in maintenance of water and sodium homeostasis in rodents and humans (1, 2). Activation of the RAAS is controlled to a certain extent by a drop in distal tubule sodium levels that is sensed by renal macula densa cells, which subsequently release renin, leading to an elevation in circulating downstream hormones: angiotensin II (Ang II) and aldosterone (2). Therefore, as a result of a low-sodium diet, increased aldosterone leads to sodium reabsorption and hence retention of sodium in the body (2). Moreover, the low-sodium diet invokes the increase in serum aldosterone through both expansion of zG cell layer and up-regulation of aldosterone synthase (CYP11B2) expression in zG cells (3). However, molecular mechanisms of in vivo aldosterone production have not been fully elucidated.

Adrenocortical genes up-regulated in response to RAAS activation have been extensively investigated, and a number of early and late response genes have been identified. For the analyses of early response genes, adrenocortical primary culture cells, mouse Y1 adrenocortical tumor cell lines, and human adrenocortical cancer cell lines, such as H295R cells, were primarily used (4–6). Ang II activates diverse signaling pathways that result in rapid induction of numerous transcription factors, including nuclear receptor subfamily 4, group A, member 1–3 (Nr4a1-3). However, the levels of these transcription factors start decreasing after 6–12 hours and return to their baseline values within 24 hours (5, 6). Along with these transcription factors, Ang II stimulation also increases aldosterone synthase (CYP11B2) mRNA levels, which reach a maximum value within 12–24 hours and subsequently decrease (6, 7). In murine Y1 cells and H295R cells, NR4A1 and NR4A2 can increase transcription of 3 enzymes responsible for aldosterone synthesis: 3-β-hydroxysteroid dehydrogenase type 2 (HSD3B2), 21-hydroxylase (Cyp21), and CYP11B2 (8–10). Therefore, these transcription factors are considered to be early response genes in aldosterone production (7).

Interestingly, chronic in vivo administration (7–14 d) of Ang II increases expression of its own receptors, Ang II receptor type 1A and 1B (Agtr1a and Agtr1b) in rat zG (11, 12), and this effect is blocked by Ang II receptor antagonist, losartan (12), suggesting that Ang II receptor induction may have a role in aldosterone regulation. In addition to Ang II receptor induction, chronic Ang II treatment also increases adrenal steroidogenesis capacity by increasing cholesterol intake in the zG cells via lipoproteins, such as high-density lipoprotein (HDL), low-density lipoprotein (LDL), and very LDL (13). In H295R cells, HDL receptor, scavenger receptor class B, member 1 (SCARB1) is up-regulated by Ang II stimulation (14, 15). In addition, HDL stimulates aldosterone production mainly by activating CYP11B2 protein expression (16). LDL receptor (LDLR) expression is stimulated by phorbol-12-myristate-13 acetate, a potent activator of aldosterone-regulating protein kinase C (14). Very LDL also increases CYP11B2 mRNA expression in a concentration-dependent manner (17). In addition to the chronic effects on aldosterone synthesis, Ang II also stimulates proliferation of zG cells in vivo and in vitro (18, 19). Overall, chronic Ang II administration may increase the capacity for steroidogenesis by increasing Ang II receptors and cholesterol uptake as well as by zG proliferation.

As an initial step to elucidate molecular mechanisms of aldosterone production, we previously defined differentially expressed transcripts between the zG and zona fasciculata (zF), the middle layer of adrenal cortex, which produces glucocorticoid (20, 21). As described in our previous report (20), laser-capture microdissection (LCM) was employed to isolate tissues from the two steroid hormone-producing adrenal cortex layers in Sprague-Dawley (SD) rats fed normal sodium (NS) diet. The microarray analysis of these samples revealed the unique gene expression profile of the zG, with 234 transcripts that show at least two-fold greater expression in the zG than zF. These genes may have important roles in zG maintenance and/or aldosterone production. However, more in vivo studies are needed to define the molecular mechanisms of aldosterone regulation. Therefore, in this study, we carried out an unbiased approach to compare rat zG transcriptomes between rats fed with NS, high-sodium (HS) (RAAS-inhibited model), and sodium-deficient (SDef) (RAAS-activated model) diet using LCM and microarray. We determined that only a few hundred genes were differentially regulated by more than two-fold in SDef compared with NS or HS. Interestingly, a number of the genes were associated with cell proliferation and/or cholesterol/steroid metabolism processes. These genes may play critical roles in physiological aldosterone production in zG cells caused by RAAS activation, including both zG layer expansion and aldosterone biosynthesis.

Materials and Methods

Rats

All rat procedures carried out were reviewed and approved by the Georgia Reagents University Institutional Animal Care and Use Committee. Male SD rats (Harlan) were housed in a temperature-controlled environment maintained on a 12-hour light, 12-hour dark cycle and given access to food/water ad libitum. After at least 1 week of acclimation by daily handling and weighing, the rats were randomly divided into 3 groups and provided with diets containing different levels of sodium: SDef (0.01%–0.02% background sodium, TD.90228; Harlan Laboratories, Inc), NS (0.49% NaCl added to the SDef diet, TD.96208), and HS (4% NaCl added to the SDef diet, TD.92034). After 13 or 72 hours of diet treatment, rats were killed via rapid decapitation between 8 and 10 am followed by tissue and blood collection. Adrenal glands were cleaned of adherent tissues and frozen in embedding media (O.C.T. compound; Sakura Finetek U.S.A., Inc) for LCM or fixed in 4% paraformaldehyde for histology.

Aldosterone measurements

Sera were collected from the blood samples, and serum aldosterone levels were determined using RIA kit for aldosterone (Coat-A-Count Aldosterone; Siemens) (16). An aldosterone concentration for any sample below the detection limit (9.06 pg/mL) was adjusted to 9.06 pg/mL to aid statistical analysis. Because natural logarithm transformed values rather than absolute values followed a normal (Gaussian) distribution, natural logarithm transformed serum aldosterone values were used in statistical analysis.

LCM and RNA isolation

Frozen adrenal glands were cut in 7-μm sections onto Superfrost Plus Microscope Slides (Fisher Scientific) or MembraneSlides PEN-Membrane (Leica). These sections were stained with cresyl violet (Sigma) by following the manufacture's ribonuclease-free protocol in the HistoGene LCM Frozen Section Staining kit (Molecular Devices) with an exception of removing the xylene step in staining MembraneSlides PEN-Membrane. To collect enriched populations of aldosterone-producing zG cells, the 3- to 5-cell layers immediately beneath the capsule were carefully captured, and total RNA was isolated using the PicoPure RNA isolation kit (Molecular Devices) as previously described (20). The tissue samples were captured from at least 5 rats per group. However, bioanalysis of the RNA samples revealed severe degradation in some of the samples potentially due to the small starting sample sizes and subsequent complex procedures. Only the RNAs with an acceptable quality were subjected for the following analyses.

RNA amplification, labeling, and microarray

Total RNAs (1–5 ng) from zG samples collected from rats fed diet for 72 hours were submitted to the University of Michigan DNA Sequencing Core MicroArray Core Facility for the subsequent procedures. RNA amplification and reverse transcription were performed using the Ovation Pico WTA System V2 (no. 3300; NuGEN Technologies, Inc). The cDNA was purified using QIAquick PCR Purification kit (no. 28104; QIAGEN) and was fragmented and biotin-labeled using Encore Biotin Module (no. 4200; NuGEN Technologies, Inc), followed by hybridization to Rat Gene 1.1 ST Array Strip (no. 901627; Affymetrix). This array was designed to interrogate almost all genes on the rat genome with 27 342 probe sets comprised of 722 254 unique probes (median 26 probes per probe set). Note that more than 1 probe set existed for some genes, such as the two independent probe sets for ribonucleotide reductase M2 (Rrm2) gene. Also, some probe sets were designed to bind transcripts from multiple genes in a gene family: for example, a probe set for Transcript Cluster ID no. 10904543 detected a cumulative transcript level of 3 CYP11B family member genes (Cyp11b1–Cyp11b3). The details of array design, including probe sequences, are available online at the manufacture's website (http://www.affymetrix.com/estore/browse/products.jsp?productId=prod350004#1_3).

Quantitative RT-PCR (qPCR)

Total RNA obtained via LCM was subjected to linear amplification and conversion to cDNA using the Ovation PicoSL WTA System V2 (no. 3312–48; NuGEN Technologies, Inc). For qPCR, 1 ng of prepared cDNA was mixed with Fast Universal PCR Master Mix (2×, no. 4367848; Applied Biosystems) and TaqMan Gene Expression Assays (Applied Biosystems) specific for the transcripts: Cyp11b2, Rn02396730_g1; cyclin-dependent kinase inhibitor 3 (Cdkn3), Rn01414656_m1; Rrm2, Rn01768870_g1; Nr4a1, Rn00666995_m1; and Scarb1, Rn00580588_m1. Peptidylprolyl isomerase A (Ppia, cyclophilin A) transcript was used for normalization of sample loading (Rn00690933_m1). The delta delta Ct method was used to calculate fold changes (22). Delta Ct values were used for statistical analysis. SEM of fold change was calculated based on delta delta Ct values.

Immunohistochemical (IHC) staining

Localizations of zG-specific CYP11B2 and zF-specific steroid 11-β-hydroxylase (CYP11B1) proteins in rat adrenal glands were visualized using IHC staining as previously described (20, 23).

Statistics for physiological data and qPCR

SigmaPlot 12 software (Systat Software, Inc) was used for statistical analysis of food intake (FI), body weight, serum aldosterone value, and qPCR delta Ct values. Statistical differences between HS, NS, and SDef at different time points were analyzed by either one-way ANOVA, two-way ANOVA, or two-way repeated measures ANOVA. ANOVA test was followed by an appropriate post hoc test to determine the groups with significant differences. P values below 0.05 were considered statistically significant.

Statistics for microarray data

Statistical analysis of microarray data was performed by University of Michigan DNA Sequencing Core MicroArray Core Facility. Microarray data were analyzed using “affy,” “oligo,” and “limma” packages of the Bioconductor implemented in the R statistical environment as follows. The raw expression values for each gene probe sets were normalized using a robust multiarray average method (24). Robust multiarray average normalized values were fit weighted to linear models that are designed specifically for microarray analysis (25). For comparison between groups, t test was performed, and P values were adjusted for multiple comparisons using Benjamini and Hochberg false discovery rate (q value) (26). Gene ontology (GO) analyses were calculated using a conditional hypergeometric test and were pursued using GOstats in the Bioconductor (27). A GO term tree was generated using visualization functionality of AmiGO 1.8 at the GO website (http://amigo.geneontology.org/cgi-bin/amigo/amigo) (28), and the output file was further annotated using GraphViz software (29).

Results

In order to invoke changes in physiological adrenal aldosterone production, male SD rats at 11 weeks of age were divided into 3 groups (n = 12 each) after 1-week acclimation and subjected to different levels of oral sodium loading with HS, NS, and SDef diets. The levels of daily FI in HS and SDef rats were comparable with that of NS rats throughout the 3-day treatment period except for the first 24 hours (d 1) in HS rats, where they consumed about 20% less diet than NS rats (mean ± SE, 17.67 ± 0.71 g vs 22.17 ± 0.89 g; two-way ANOVA, P < .01) (Figure 1A). Consistent with the comparable FI, body weight gain on the 3 diets was similar throughout the 3-day period (two-way repeated measures ANOVA, P > .80), and there was no difference between body weights at the end of the treatment (one-way ANOVA, P = .699) (Figure 1B). Similarly, a shorter diet treatment (13 h) caused a trend of lower FIs in HS rats (HS, 15.23 ± 1.07 g; NS, 17.96 ± 0.45 g; and SDef, 17.93 ± 0.95 g; n = 4 each, one-way ANOVA, P = .085), and there was no difference in their body weight (data not shown). Overall, regardless of salt content, the rats consumed similar amounts and gained comparable levels of body weight.

Figure 1.

FI and body weight changes. Eleven-week-old male SD rats were maintained on a 12-hour light, 12-hour dark cycle and given access to food/water ad libitum. After at least 1 week of acclimation by daily handling and weighing, the rats were randomly divided into 3 groups at day 0 and provided with diets containing different levels of sodium: SDef, NS, and HS. A, Amount of FI before and after diet change. FI was measured daily from the day before diet change (d −1) to the day of killing (d 3). Error bar, SEM. Statistical analysis was performed by two-way ANOVA with Holm-Sidak post hoc method. ***, P < .001. B, Body weight change. Body weights were measured daily throughout the experiment. Error bar, SEM; n.s., not significant.

These changes in oral sodium intake successfully altered serum aldosterone levels as well as both CYP11B2 mRNA and protein expression in adrenal zG cells. The 13-hour treatment (n = 4 per diet) caused an elevation of serum aldosterone level in SDef rats compared with that of NS rats (median [25 percentile–75 percentile], 201.7 [90.5–311.8] pg/mL vs 26.4 [13.2–48.6] pg/mL; two-way ANOVA, P < .01), whereas that of HS rats showed a trend of decline (9.06 [9.06–9.63] pg/mL; P = .085) (Figure 2A). Correspondingly, the 72-hour treatment (n = 12 per diet) caused an increase in aldosterone production in SDef rats compared with that of NS rats (763.1 [515.8–1060] pg/mL vs 34.0 [19.9–69.5] pg/mL; P < .01), and that of HS rats showed a trend of decline (20.8 [12.8–29.2] pg/mL; P = .081) (Figure 2A). Consistently, Cyp11b2 transcript levels in laser-captured zG cells were significantly different between HS, NS, and SDef rats at both 13- and 72-hour time points (n = 3–5 per group, two-way ANOVA, P < .01) (Figure 2B). As for the protein expression in adrenal cortex, IHC staining (n = 4 each) revealed no apparent difference in CYP11B2 staining between diets at the 13-hour time point (Figure 2C, top row), indicating that little or no change occurred at the protein level in the first 13 hours. In contrast, at the 72-hour time point, the differences between diets were evident, where almost no CYP11B2 staining was detectable in HS adrenals, and prominent CYP11B2 staining was observed in SDef adrenals (Figure. 2C, mid row). Furthermore, double CYP11B1/CY11B2 IHC staining showed that distances between capsule and CYP11B1-positive zF layers were comparable among all diet groups, suggesting that the 72-hour diet treatment had little effect on the numbers of cells in the histological zG layer (Figure 2C, bottom row). These results indicated that the diet treatments caused changes in adrenal aldosterone production and associated mRNA/protein expression, and the changes were more prominent at the 72-hour time point, with little effect on the zG cell numbers. Based on these observations, we decided to determine zG transcriptome differences at the 72-hour time point with the LCM and microarray technologies.

Figure 2.

Serum aldosterone and CYP11B2 expression in the zG. After 13 or 72 hours of diet treatment of SDef, NS, and HS, rats were killed via rapid decapitation followed by blood and adrenal glands collection. A, Serum aldosterone among rat groups. Serum aldosterone levels were determined by RIA. Natural logarithm transformed serum aldosterone values were used in statistical analysis (two-way ANOVA with Holm-Sidak post hoc method). The boundary of the box indicates the 25th and 75th percentiles, and a line within the box marks the median. Whiskers below and above the box indicate the 10th and 90th percentiles, respectively. **, P < .01; ***, P < .001. B, qPCR for Cyp11b2. Laser-captured enriched populations of aldosterone-producing zG cells were used for Cyp11b2 qPCR. Fold changes were calculated by comparing with 72-hour-treated NS. Statistical analysis was performed by two-way ANOVA with Student-Newman-Keuls post hoc method using delta Ct value from qPCR. Error bar, SEM. **, P < .01; ***, P < .001. C, IHC, single-immunostaining for CYP11B2 with diaminobenzidine (brown) on 13-hour (top row)- and 72-hour (middle row)-treated adrenals as well as double-immunostaining for CYP11B2 with 5-bromo-4-chloro-3′-indolyphosphate (blue) and CYP11B1 with diaminobenzidine (brown) on 72-hour-treated adrenals (bottom row).

In order to elucidate the effects of diet treatments on zG cell transcriptome, we used laser-captured zG samples from 72-hour-treated HS (n = 3), NS (n = 4), and SDef (n = 5) rats for microarray analyses. Out of the 27 342 probe sets on the microarray covering 17 061 RefSeq coding transcripts (as of February 2012), only 79 and 10 probe sets were found up-regulated and down-regulated more than two-fold (P < .05), respectively, in SDef compared with NS (SDef/NS comparison, Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). Between SDef and HS (SDef/HS comparison), 201 and 68 probe sets were up- and down-regulated in SDef, respectively (Supplemental Table 2). On the other hand, only 3 probe sets were found up-regulated, and no probe sets were down-regulated in NS compared with HS (NS/HS comparison, Supplemental Table 3), and hence, the analyses hereafter focused on the SDef/HS and SDef/NS comparisons. Intriguingly, most of the probe sets identified in SDef/NS comparison, 71 up-regulated and 6 down-regulated, overlapped with SDef/HS (Figure 3 and gray-shaded probe sets in Supplemental Tables 1 and 2). Overall, we identified a number of differentially expressed zG transcripts between diet treatments with the highest number in the SDef/HS comparison.

Figure 3.

Venn diagram showing the overlaps of differentially regulated genes in the three comparisons. Differentially regulated genes with more than two-fold difference and P < .05 were counted. Each circle of SDef group vs NS group (SDef/NS), SDef vs HS group (SDef/HS), and NS/HS shows the count of up-regulated genes and down-regulated genes in upper and lower side of circles, respectively.

In order to visualize transcript expression levels in individual rats, a heat map was generated for 27 probe sets with more than five-fold differences in the SDef/HS and SDef/NS comparisons (Figure 4), which included 25 up-regulated and 1 down-regulated genes in SDef/HS in addition to 8 up-regulated genes in SDef/NS. Note that all up-regulated genes in SDef/NS were also up-regulated in SDef/HS. The signal intensities in rats in each diet group were relatively uniform throughout the probe sets, suggesting consistent diet effects and LCM samplings.

Figure 4.

Heat map of genes with more than five-fold difference in expression level. The heat map was created for genes with greater than 5-fold changes with P < .05 in the comparison of SDef group vs HS group (SDef/HS) and SDef vs NS group (SDef/NS). Log2 signal intensities were shown in individual adrenal zG cells of 3 HS, 5 SDef, and 4 NS rats. Fold changes (FCs) in SDef/HS and SDef/NS comparisons were shown in the left and right side of the heatmap, respectively.

The microarray results were validated by qPCR for two genes (Cdkn3 and Rrm2) from transcripts that appeared in both SDef/NS and SDef/HS comparisons as well as two genes (Nr4a1 and Scarb1) from those only in SDef/HS (Figure 5). When transcript levels were normalized to NS rats, Cdkn3 showed significant up-regulation in SDef (mean [mean ± SE range], 8.16 [6.44–10.34]-fold) compared with both NS (1.00 [0.74–1.36]-fold; one-way ANOVA, P = .002) and HS (0.15 [0.09–0.27]-fold; P < .001). Similarly, Rrm2 showed significant up-regulation in SDef (16.27 [13.37–19.80]-fold) compared with NS (1.00 [0.56–1.80]-fold; P = .026) and HS (0.03 [0.006–0.16]-fold; P = .001). Nr4a1 and Scarb1 transcripts were up-regulated in SDef compared with HS: 3.02 [2.27–4.03]- vs 0.08 [0.01–0.48]-fold (P = .033) and 1.68 [1.40–2.02]- vs 0.023 [0.17–0.31]-fold (P < .001), respectively. On the whole, qPCR results for 4 genes were consistent with the microarray results and confirmed the validity of microarray findings.

Figure 5.

Confirmation of microarray data using qPCR. Expression levels of 4 genes (A, Cdkn3; B, Rrm2; C, Nr4a1; and D, Scarb1) were determined using qPCR to confirm microarray data. Fold changes of HS and SDef were calculated by referring the value of NS. Error bar, SEM. *, P < .05; **, P < .01; ***, P < .001.

In order to better understand the biological meaning of the observed changes in gene expression, GO analysis was performed using the lists of differentially expressed genes (>two-fold difference, P < .1) in the SDef/NS and SDef/HS comparisons. GO is a major bioinformatics initiative with the aim of standardizing the representation of gene and gene product attributes using a systematic language (ontology) (30). The analysis took advantage of the hierarchical structure of the GO database by testing child terms (more specific gene sets) first, and if they were significant, it removed their genes from the parent term (more general gene sets) before the parent term was tested. Although we performed GO analysis on all 3 key biological domains, including biological processes (BPs), cellular components, and molecular functions, we described only the results of the BP domain in this report, because the other domains provided little insight on the functional attributes of our gene lists.

The GO analysis identified 35 and 42 significant GO terms in SDef/NS and SDef/HS comparisons (Supplemental Tables 4 and 5), respectively, and 18 terms were shared between these 2 GO term lists (gray shaded in Supplemental Tables 4 and 5). In order to illustrate the relations of these GO terms, we generated a hierarchical GO term tree, where significant GO terms from both lists and their nonsignificant parent terms were plotted (Supplemental Figure 1). With the visualization of GO term tree, we identified at least three independent groups of functionally related GO terms in the tree. The terms in the first of these 3 arbitrary groups (group 1) are broadly associated with cell proliferation, where many of the GO terms contained words such as mitosis, cell cycle, chromosome segregation, and so on. A number of genes in the heat map (Figure 4) were associated with these terms, including Cdkn3 (Figure 5A), Rrm2 (Figure 5B), and Top2a (6.15-fold up in SDef/NS comparison and 9.25-fold up in SDef/HS comparison, both P < .001). Intriguingly, about 80% of the SDef/NS terms (27 out of 35) and 40% of SDef/HS terms (16 out of 42) belonged to this group, indicating that SDef diet, when compared with other diets, significantly altered expression levels of the genes associated with cell proliferation process.

The second GO term group (group 2) consisted of 9 SDef/HS terms that were broadly associated with cellular responses to stimuli. The types of stimuli in these terms were broad, including chemical, steroid hormone, growth factors, and more. This suggested that, rather than a specific set of genes for a particular stimulus, SDef diet induced changes in expression levels of a wide array of genes commonly affected by various stimuli. For example, Ccl5 gene (3.70-fold down in SDef/HS comparison; P = .039) was involved in all 9 GO terms, Thbs1 (7.14-fold down in SDef/HS comparison; P < .001) in 8 terms, and Fabp4 (10.6-fold up in SDef/HS comparison; P = .095) in 7 terms. Notably, some of the genes vital for aldosterone production were also registered in these terms, including Cyp11b2 (in 4 terms, 12.55-fold up in SDef/HS comparison; P < .001), Nr4a1 (4 terms, 8.82-fold up in SDef/HS comparison; P = .018), Ldlr (3 terms, 5.21-fold up in SDef/HS comparison; P = .002), and Scarb1 (1 term, 6.11-fold up in SDef/HS comparison; P < .001).

The third GO term group (group 3) consisted of only two SDef/HS terms. However, it contained GO terms associated with cholesterol metabolism and steroid biosynthesis. These two BPs play a critical role in aldosterone synthesis and involved genes such as Scarb1, Cyp11b2, and Ldlr in the lists. Detection of these terms was consistent with our experimental design and proved the reliability of our approach. Overall, the GO analysis indicated that many genes that overlapped in both SDef/NS and SDef/HS comparisons were involved in cell proliferation, and the genes specific for SDef/HS may be associated with a wide range of BPs, including responses to stimuli and steroid biosynthesis.

Discussion

In this study, we defined for the first time changes in the zG transcriptome induced by different dietary sodium levels using an unbiased method. Microarray analysis revealed hundreds of genes differentially regulated in response to changes in dietary sodium intake, including known aldosterone-regulating genes, such as Cyp11b2, Nr4a1-3, and Scarb1. In addition, GO analysis of the differentially expressed genes visualized an overview of the changes in transcriptomes. These results provided a complete picture of in vivo zG transcriptome changes after 72 hours of RAAS system activation, which is invaluable for elucidating the molecular mechanisms of aldosterone regulation.

The 72-hour (3 d) time point that we used in microarray analysis showed clear sodium diet effects on mRNA, protein, and aldosterone production levels with no apparent increase in zG cell number. Mitani et al (3) have elegantly shown the expansion of CYP11B2-positive zG cell layer in rats during a 10-day time course of low-sodium diet. In the report, they showed that most zG cells before low-sodium diet administration have little CYP11B2 expression, whereas CYP11B2-positive zG cells occupy the entire subcapsular area after a 3-day treatment before an increase in cell number. Consistent with their findings, our IHC results also showed that more zG cells became CYP11B2-positive without changing thickness of the zG after 72-hour SDef. In addition, we analyzed a shorter time point, 13 hours. Interestingly, protein expression level was not changed by SDef or HS IHC, but Cyp11b2 transcript level was clearly altered in the zG. This suggests that mRNA level of CYP11B2 in the zG was regulated within 13 hours, but the CYP11B2 protein level follows to CYP11B2 mRNA after 13 hours. This is in agreement with in vitro cell culture studies showing increased CYP11B2 mRNA within 6 hours in treatment with Ang II (6). Increased serum aldosterone level in SDef at 13 hours may be caused by increased expression and phosphorylation of steroidogenic acute regulatory protein, which is known as an “early regulatory step of aldosterone production” (13). Mitani et al (3) also reported that the CYP11B2-positive zG cells begin to proliferate resulting in zone expansion to 10–15 or more cells in depth, after 6–10 days of SDef. Intriguingly, our microarray results clearly showed that changes in proliferative gene expression were apparent after 72-hour SDef treatment (group 1 of Supplemental Figure 1). These genes may be involved in the subsequent zG expansion at 6–10 days of SDef treatment (3).

Among the 26 genes in the heat map (Figure 4), 13 genes (50%) belonged to “cell proliferation associated” GO terms (group 1 in Supplemental Tables 4 and 5): Cdkn3, Rrm2, Top2a, stathmin 1 (Stmn1), Nuf2, Tpx2, Ccnd2, cyclin B1 (Ccnb1), Casc5, Nusap1, Ndc80, Cenpf, and Thbs1. Rrm2 and Top2a were the two highest up-regulated genes belonging to proliferation associated GO term (group 1 in Supplemental Tables 4 and 5, see also Supplemental Figure 1). Rrm2 is an S-phase-specific and rate-limiting protein for the synthesis of chromosomal DNA, and therefore, zG cells in SDef rats may have a higher capacity of chromosomal DNA synthesis (32). Top2a gene produces DNA topoisomerase IIα; an enzyme important for chromosome condensation, chromatid separation, and relief of torsional stress that occurs during DNA replication (33). Hence, Top2a up-regulation in SDef-zG cells may suggest a higher capacity of DNA synthesis as well. In the aspect of cell cycle regulation genes, Cdkn3, cyclin B2 (Ccnb2), and Ccnb1 in group 1 also appeared in the Figure 4 as highly up-regulated genes in SDef. Cdkn3 inhibits cyclin-dependent kinase 2, which is essential for the gap 1/synthesis (G₁/S) phase transition in cell cycle (34), and hence, Cdkn3 may be indirectly involved in G₁ elongation. As for cyclin B2 and cyclin B1, their expression and complex formation with cyclin-dependent kinase 1, followed by relocation into nucleus, causes cells to progress into mitosis (M) phase (35). Overall, SDef-zG showed up-regulation of genes involved in cell proliferation and mitosis. These genes in group 1 may be involved in the subsequent zG expansion.

As mentioned in the Results section, genes vital for aldosterone production were included in “response to stimulus associated GO terms” (group 2 in Supplemental Table 5 and Supplemental Figure 1) and/or “cholesterol/steroid metabolism associated GO terms” (group 3). Actually, a number of genes overlapped in groups 2 and 3. It is known that transcription factors such as Nr4a1, Nr4a2, and Nr4a3, which were included in group 2, are acutely regulated genes related to Ang II-stimulated aldosterone production in vitro (early response genes) (6, 8, 10, 15). Consistently, in the current in vivo study, all Nr4a genes were up-regulated in SDef/HS (Nr4a1, fold change 8.8-fold; Nr4a2, 4.7-fold; and Nr4a3, 2.1-fold) (Figure 4 and Supplemental Table 2). In addition to these early response genes, two chronic response genes appeared in groups 2/3: Scarb1 in both groups and Agtr1a in group 2. Scarb1 is up-regulated in H295R cells after a 72-hour Ang II stimulation (7), which was consistent with the current in vivo study. In fact, HDL, a ligand of Scarb1, increases CYP11B2 expression and aldosterone production in H295R cells (16). Thus, HDL and its receptor Scarb1 seems to have an important role in aldosterone production under activated RAAS. As for the down-regulated genes in groups 2/3, sonic hedgehog (Shh) (0.45-fold in SDef/HS) and wingless-type MMTV integration site family, member 4 (Wnt4) (0.37-fold; P = .004) were listed. Both Shh and Wnt4 have a crucial role for adrenal homeostasis (36). Shh mRNA is expressed in the outer portion of developing mouse adrenal, and conditional inactivation of Shh in the adrenal cortex causes severe hypoplasia and histological disorganization of the adrenal cortex (37). WNT4 is also expressed in the zG (20, 21), and Wnt4-deficient mice lead to a decrease in aldosterone production (38). Thus, groups 2/3 genes, the ones discussed above, as well as those that have not been studied yet in the context of aldosterone regulation, may play a critical roles in RAAS-induced aldosterone production in vivo.

Genes that did not belong to groups 1–3 may also have important roles in aldosterone production. Pyroglutamylated RFamide peptide receptor (Qrfpr), retinol binding protein (Rbp7), transmembrane protein 178A (Tmem178), PDZ binding kinase (Pbk), and Stmn1 were the 5 highest up-regulated genes that did not belong to groups 1–3 (Figure 4). Among these genes, Qrfpr was reported to have a role in aldosterone regulation. Qrfpr is a G protein-coupled receptor, and iv administration of pyroglutamylated RFamide peptide, the ligand of QRFPR, causes a release of aldosterone from adrenal gland (39). Like Qrfpr, other genes may have roles in aldosterone production. However, further investigation is needed to elucidate their functions.

Although this study presented a comprehensive picture of zG transcriptome changes associated with physiological aldosterone production, some limitations may have to be taken into account when interpreting the results. First, physiological aldosterone production can be stimulated not only by a SDef diet but also by an elevated potassium intake. Dierks et al (40) have shown that mice on a 4 week high-potassium diet increase aldosterone production and transcriptome changes in the whole adrenal glands. Intriguingly, however, among the top 30 potassium up-regulated genes, only cyclin A2 (Ccna2) and Ccnb1 overlapped with our SDef/HS and SDef/NS comparisons. Because Dierks et al (40) have elegantly shown that expression of adrenal genes changes in a time-dependent manner, the limited overlap may be due to the difference in treatment duration, ie, 3 days in our rats vs 4 weeks in the mice of Dierks et al (40). Another likely cause of limited overlap may be the differences in tissue sources, where Dierks et al (40) used whole mouse adrenals, whereas we used a laser-captured population of rat zG.

Second, we assume that most of the transcriptome changes in SDef diet were caused by RAAS activation and hence a direct action of elevated circulating Ang II. However, it is possible that other factors, including changes in serum sodium or potassium concentrations and/or neurological responses to the low-sodium diet, may have had an effect on zG transcriptome changes. Further experiments, such as a chronic Ang II infusion model, may need to be employed to further dissociate these effects.

The current study focused on the physiological changes of zG transcriptome, which may or may not have relevance to adrenal pathologies. However, we believe that the current gene list will provide a reference point for the transcriptome analyses of pathological conditions. As a proof of concept, we compared the gene list in SDef/HS comparison with that of human aldosterone-producing adenoma (APA)-associated genes. At least 3 laboratories have published comprehensive lists of APA-associated genes by comparing the APA transcriptome with that of adjacent normal zG (31, 41, 42). Our initial expectation was that there would be a significant overlap between SDef/HS and APA/adjacent-normal comparisons, because both cases compared aldosterone-producing zG (-like) tissue and aldosterone-suppressed normal zG tissue. To our surprise, however, Cyp11b2 was the only common gene between the comparisons. The cause of this limited gene profile overlap may simply be due to the species difference, ie, rat vs human. Alternatively, however, the limited overlap may indicate that aldosterone production in APA is induced by a pathway only present in the pathology of APA. This suggests the feasibility of APA-specific pharmacological inhibition of aldosterone production by targeting this disease-specific pathway.

In summary, we have described for the first time rat zG transcriptome changes caused by dietary sodium intake and identified more than 280 differentially regulated genes. GO analysis identified that some of these genes may play a role in cell proliferation (group 1), response to stimulus (group 2), and cholesterol/steroid metabolism (group 3). Although more functional investigations are needed, genes in group 1 may be involved in zG expansion in response to RAAS activation, and some of the genes in groups 2/3 may be associated with aldosterone overproduction in zG cells. The gene list identified in this study provides adrenal researchers a broad analysis of the zonal effects of sodium diet, thereby providing new opportunities for adrenal and aldosterone research fields.