Transcriptomic analysis reveals inflammatory and metabolic pathways that are regulated by renal perfusion pressure in the outer medulla of Dahl-S rats

Abstract

Studies exploring the development of hypertension have traditionally been unable to distinguish which of the observed changes are underlying causes from those that are a consequence of elevated blood pressure. In this study, a custom-designed servo-control system was utilized to precisely control renal perfusion pressure to the left kidney continuously during the development of hypertension in Dahl salt-sensitive rats. In this way, we maintained the left kidney at control blood pressure while the right kidney was exposed to hypertensive pressures. As each kidney was exposed to the same circulating factors, differences between them represent changes induced by pressure alone. RNA sequencing analysis identified 1,613 differently expressed genes affected by renal perfusion pressure. Three pathway analysis methods were applied, one a novel approach incorporating arterial pressure as an input variable allowing a more direct connection between the expression of genes and pressure. The statistical analysis proposed several novel pathways by which pressure affects renal physiology. We confirmed the effects of pressure on p-Jnk regulation, in which the hypertensive medullas show increased p-Jnk/Jnk ratios relative to the left (0.79 ± 0.11 vs. 0.53 ± 0.10, P < 0.01, n = 8). We also confirmed pathway predictions of mitochondrial function, in which the respiratory control ratio of hypertensive vs. control mitochondria are significantly reduced (7.9 ± 1.2 vs. 10.4 ± 1.8, P < 0.01, n = 6) and metabolomic profile, in which 14 metabolites differed significantly between hypertensive and control medullas (P < 0.05, n = 5). These findings demonstrate that subtle differences in the transcriptome can be used to predict functional changes of the kidney as a consequence of pressure elevation.

INTRODUCTION

In Dahl salt-sensitive (SS) rats, as in the clinical condition, hypertension is accelerated by the consumption of a high-salt diet and disease progression is associated with the development of chronic kidney disease (20). Following several weeks of a high-salt diet SS rats show substantial renal damage, characterized by tubular protein casts, renal immune-cell infiltration and reductions in glomerular filtration rate (5, 7).

Studies in our laboratory and others have applied transcriptomic methods such as RNA-Seq and microarray to reveal pathways responsible for renal injury that occurs with high-salt feeding (22–24, 40). However, these studies were unable to separate the effects of salt and pressure. Therefore, it is impossible to determine whether observed changes are the consequence of high pressure, secondary effects caused by salt, or both.

To determine the effects of pressure per se, absent the potential confounding factors of extrarenal effects or salt alone, we developed a system in which renal perfusion pressure (RPP) to a single kidney in the SS rat is mechanically maintained at control pressure. Previous studies have shown that differences in pressure in kidneys in the same rat have profound physiological effects (10, 25, 28–30). Specifically, this approach has demonstrated that in SS rats fed a 4.0% NaCl diet (high salt, HS), renal injury is exacerbated in the right-hypertensive kidney relative to the left servo-controlled normotensive kidney (25). Affymetrix microarrays also revealed differentially expressed genes within the outer medullary tissue between the left and right kidneys. This provided the first insights related to the molecular pathways that mediate the deleterious effects of pressure (25).

The present study expands upon this earlier experiment. First, we have analyzed the transcriptome at an earlier time point (7 days rather than 14 days of HS diet) to detect pathways that are more likely to be causes of the pathophysiology of renal injury, rather than consequences. Second, we have utilized RNA-Seq, which is not limited to previously selected gene sequences and offers a greater dynamic range than earlier microarray studies as well as increased precision in signal quantification. Third, advancements in statistical methodologies have enabled considerably more robust interpretations of transcriptomic data. As with our previous microarray studies (25), the experiments detailed here focus on the effects of RPP on transcriptomic changes in the renal outer medulla. Previous studies from our laboratory have highlighted the key role of the outer medulla in the development of salt-sensitive hypertension in SS rats (4). Enhanced oxidative stress in the renal medulla of SS rats causes reduced medullary blood flow and hypoxia, which may stimulate renal injury, inflammation, and fibrotic injury. Using the servo-control technique we were able to specifically isolate the effects of perfusion pressure on these phenomena.

Indicative of the increased precision of RNA-Seq, 1,613 genes were found to be significantly different between the right-hypertensive and left servo-controlled outer medullas compared with 57 genes in our earlier microarray analysis. Because of this greatly increased breadth we were able to take advantage of pathway-level analyses, which enhanced our ability to detect physiological patterns. Specifically, we applied three distinct pathway analytical approaches to identify pathways that were significantly overrepresented in the hypertensive renal outer medulla. Importantly, this enabled an unbiased view of the data set, avoiding the possibility of choosing genes based on prior expectation. In validation of this approach, the kidneys exposed to high pressures showed increases in immune cell infiltration and defects in mitochondrial function and increases in Jnk signaling in the outer medulla.

METHODS

Experimental animals.

Male SS/JrHsdMcwi (SS) rats were obtained at weaning from a colony developed and maintained at the Medical College of Wisconsin. All experimental protocols were approved by the Medical College of Wisconsin Institutional Animal Care and Use Committee.

Surgical preparation.

Prior to surgery rats were trained to a bidirectional turntable cage (Rodent Workstation with Raturn system; Bioanalytical Systems, West Lafayette, IN) for ≥7 days. Rats were surgically prepared, under isoflurane anesthetic, at 9–12 wk as described (10, 30). In brief, an inflatable silastic occluder was placed around the aorta between the renal arteries. Catheters were placed in the carotid and femoral arteries for the measurement of mean arterial pressure above and below the cuff, respectively. A catheter was implanted in the femoral vein for the administration of analgesics and anesthetics. Catheters were tunneled subcutaneously and exited between the shoulder blades. After ≥7 days recovery blood pressure measurements were initiated.

Servo-control protocol.

Saline was continuously infused into the venous catheter at 6.9 μl/min, and mean arterial pressure was recorded 24 h/day from both carotid and femoral catheters. Baseline measurements on 0.4% NaCl diet were made over 3 days to obtain stable control measurements. Subsequently, rats were swapped to HS and measurements continued for 7 days. During the HS diet RPP to the left kidney was continuously maintained within ~5 mmHg of the average control value. Significant differences in RPP were assessed by two-way repeated -measures ANOVA.

Tissue collection.

Rats were anesthetized with pentobarbital, and both kidneys were flushed with saline. The bottom pole of the kidney was isolated, and the outer medulla separated from cortex and snap-frozen for RNA isolation or metabolomics measurement.

RNA isolation and library preparation.

Eight servo-controlled left outer medullas and their corresponding hypertensive-right outer medullas were isolated for RNA analysis. Total RNA was isolated with Trizol reagent and assessed on Agilent 2100 Bioanalyzer. Samples with RNA integrity number ≥ 8 were considered high quality and used for sequencing. Libraries were prepared with Illumina’s TruSeq RNA Sample Prep Kit (RS-122-2001) and sequenced on an Illumina HiSeq2500 sequencer as described (24). Sixteen libraries were multiplexed on two lanes of a flow-cell. Sequence reads were aligned to rat genome version Rn5, and expression levels were quantified as fragments per kilobase exon per million reads (FPKM). The FPKM values were quantile normalized, and log2FPKM values were used for statistical analysis.

Detection of significantly different genes and pathways.

Significant difference in gene expression was measured by paired t-test on the log2FPKM data set. P values were false discovery rate (FDR)-corrected by the Benjamini-Hochberg method as described (1).

The list of significantly different genes was analyzed with DAVID 6.7 to find overrepresented pathways. To enable comparison to other analytical methods, DAVID analysis used only the set of pathways from the Kyoto Encyclopedia of Genes and Genomes (KEGG) (19). The same list of genes, and their expression levels, was analyzed with Ingenuity Pathway Analysis.

Bayesian regression analysis.

Using a method described recently (34, 40), we analyzed the RNA-Seq data set to find pathways that discriminate on the basis of pressure. Inputs into the method include both the set of gene expression data (log2FPKM) and the set of pathways to be analyzed (biochemical pathways from KEGG). In addition, a regression analysis was performed with the measured RPP. ΔFPKM and Δ blood pressure (BP) were calculated as the difference in expression level for each gene and BP, respectively, between each pair of kidneys. The paired regression analysis was performed on this new set of variables/end points.

Seahorse analysis.

Mitochondria were isolated from the right-hypertensive and left servo-controlled outer medulla of additional servo-control rats and assessed with an XF96 Seahorse Plate Reader as described (32). Respiratory control ratio (RCR) was calculated as the quotient of post-ADP-stimulated respiration and postoligomycin-stimulated respiration. Significant differences were determined by paired t-test.

Metabolomic analysis.

Outer medullas were isolated, snap-frozen, and analyzed by gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) at the Michigan Regional Comprehensive Metabolomics Resource Core. Data were normalized per rat, and metabolite levels expressed as the ratio of each measurement to the rat mean. Six left/right outer medulla pairs were analyzed. Using the Sign method by Filzmozer et al. (12) for the detection of multivariate outliers, a single rat was found to greatly differ from the other five rats. Additionally, this outlier rat showed an opposite signal in 15 of 22 metabolites that showed a consistent signal in the remaining five rats. Consequently, as this rat was discordant in both a holistic and an individual metabolite analysis, this rat was removed as an outlier. Holistic analysis was performed by principal component analysis (PCA) in R (Fig. 5). Individual metabolite differences were detected by paired t-test.

Fig. 5.

Metabolites were extracted from snap frozen left servo-controlled and right-hypertensive medullas and measured by mass spectrometry. A: principal component analysis reveals separate metabolic state in left servo-controlled and right hypertensive medullas. Ellipses represent 2 SD of a bivariate normal distribution. One pair exhibited a substantial difference from all other pairs and was removed as an outlier. B: 14 metabolites were found to be significantly different (paired t-test P < 0.05, n = 5) between left and right medullas.

Western blot.

Western blot analysis was performed as previously described (11). Briefly, protein samples were prepared from the outer medullas used for RNA-Seq. After transfer and blocking with 5% milk/Tris-buffered saline-Tween 20, PVDF membranes were probed with a Jnk1 pan antibody (Biosource, 44-690) and phosphor-SAPK/Jnk (Thr183/Tyr185) (Cell Signaling Technology, #9255). Images were captured by Chemidoc, and band intensity analyzed with ImageLab.

Statistical analysis.

Statistical tests were performed with R 3.0.1 and Matlab R2016b. Statistical tests for each method are described in their respective sections. Statistical significance is defined as P ≤ 0.05, FDR ≤ 0.05, or Pr ≥ 0.5, as appropriate.

RESULTS

Servo control maintains control pressure during 7 days of a 4.0% NaCl challenge.

Figure 1 illustrates the RPP in the left servo-controlled and right-hypertensive kidneys of conscious SS rats. These data correspond to rats that form a subset of the experimental group reported by Evans et al. (10). Here, we used banked tissue from eight rats that showed a wide range in the hypertensive response to an HS diet. These rats take full advantage of the continuous Bayesian regression analysis (BRA) described below, which depends on a large spread in BP responses within the group. After the switch to an HS diet, the pressures in the right-hypertensive kidneys increased from 132 ± 2 to 158 ± 5 mmHg (P < 0.05), while the corresponding left servo-controlled kidneys remained unchanged (127 ± 2 vs. 126 ± 1 mmHg) over the 7-day period.

Fig. 1.

Servo control maintains control pressure in the left kidney. Over the 7-day time course, the right kidneys (gray) experience steady increase in pressure while the left kidneys (black) remain at control pressure (n = 8). HS, high salt; RPP, renal perfusion pressure.

1,613 genes significantly different between left servo-controlled and right-hypertensive outer medullas.

Eight pairs of left servo-controlled and right-hypertensive outer medullas were collected for RNA-Seq analysis. A heat map of the transcriptomic data did not reveal a consistent separation between left and right outer medullas (Fig. 2A). However, when paired on a per-rat basis a strong signal differentiating left from right outer medullas became apparent (Fig. 2B). We found 1,613 genes to be significantly different between right-hypertensive and left servo-controlled outer medullas (FDR ≤ 0.05).

Fig. 2.

Heat maps reveal a paired signal distinguishing the left servo-controlled medullas from the right hypertensive medullas (n = 8). RNA expression data for the servo-controlled and hypertensive medullas was analyzed with heat maps. A: the unpaired data set did not show a consistent signal distinguishing between left and right medullas, as indicated by the lack of separation between the left and right samples in the heat map. B: in contrast, when the data were expressed in paired format, the left and right samples show distinct separation. C: DAVID analysis finds KEGG pathways overrepresented in the list of 1,613 significantly different genes. D: IPA analysis finds similar pathways overrepresented in the list of 1,613 genes. IPA, Ingenuity Pathway Analysis; FDR, false discovery rate; L, left; R, right.

DAVID and IPA discover biological pathways that are overrepresented in the list of 1,613 genes.

Figure 2, C and D, presents the pathways significantly overrepresented in the list of 1,613 genes determined by two widely used analytical packages: DAVID and IPA. These methods use the identities of the significantly different genes to find pathways that are “overrepresented” in the data set, i.e., pathways that have more genes in the data set than would be expected if genes were selected randomly. As shown, considerable overlap was found in the pathways identified by these two analyses although different names are given to similar pathways (e.g., oxidative phosphorylation in the DAVID analysis and mitochondrial dysfunction in IPA).

BRA discovers pathways that discriminate on the basis of pressure.

DAVID and IPA designate pathways of interest based upon the identities of the differentially regulated genes but do not use the expression data directly. In contrast, the BRA (34) directly uses the measured RNA expression data to discover which pathways display consistent patterns of expression that discriminate between the left servo-controlled outer medulla and the right-hypertensive outer medulla. For example, the probability of 0.88 assigned to the “chemokine signaling pathway” in the discrete BRA indicates that there is an 88% probability that the genes in chemokine signaling pathway” display a consistent pattern that distinguishes the right-hypertensive outer medullas from the left servo-controlled outer medullas. In addition to the analysis that distinguishes right from left outer medullas (discrete end point), the BRA also incorporated the measured RPPs associated with each RNA-Seq library (continuous end point). This enabled discovery of pathways that vary continuously in their expression patterns as pressure increases. That is, unlike the discrete analysis, which detects pathways that have binary expression patterns between left and right medullas, analogous to flipping a switch and turning the genes in the pathway on or off, the continuous analysis finds pathways that show a smoother transition over the range of pressures observed in the HS day 7 medullas, analogous to turning a dial that regulates gene expression levels. Table 1 presents the pathways with posterior probability (Pr) ≥0.5 for the BRA with either a discrete or continuous response. In most cases, both analyses agreed in their assessment of the pathways. However, in the case of chemokine signaling pathway, the discrete BRA found a Pr of 0.88, while the continuous BRA found a Pr of 0.21, a difference of 0.67. In contrast, the discrete analysis found a Pr for “cytokine-cytokine receptor interaction” of 0.53, while the continuous analysis found a Pr of 0.72, a difference of 0.19.

Table 1.

High probability (Prob ≥ 0.5) pathways from Bayesian regression analysis

Pathway	Prob(Left vs. Right) Discrete	Prob(BP) Continuous
Chemokine signaling pathway	0.88	0.21
Neuroactive ligand-receptor pathway	0.62	0.62
Pathways in cancer	0.55	0.63
Cytokine-cytokine receptor interaction	0.53	0.72
Cell cycle	0.42	0.52
Purine metabolism	0.45	0.50

The Bayesian regression analysis (BRA) finds pathways that are likely to discriminate between either the left and right kidney (discrete analysis) or on the basis of blood pressure (continuous analysis). The BRA directly uses the expression data values as fragments per kilobase exon per million reads to make inferences.

Paired BRA detects difference in Jnk regulation.

The previous BRA does not take advantage of the paired nature of the transcriptomic data set. Here, we calculated ΔBP and ΔFPKM between the pairs of outer medullas, then the BRA was run as before on these new calculated variables/end points. A pathway with a high probability of activation indicates a high probability of a large difference between the left servo-controlled kidney and the right hypertensive kidney.

Although the paired BRA estimated several pathways in common with the unpaired BRA, MAPK signaling was detected only by paired BRA (Pr = 0.57). Additionally, the paired analysis assessed a 13% increase in probability of the HTLV-1 infection pathway relative to the unpaired analysis. Both the HTLV-1 and MAPK pathways contained Jnk signaling, suggesting that this may be responsible for the increase in probabilities. Consistent with this suggestion, Western blot analysis demonstrated an increase pJnk/Jnk ratio in the right-hypertensive vs. left control outer medullas (0.79 ± 0.11 vs. 0.53 ± 0.10, P < 0.01, n = 8; Fig. 3).

Fig. 3.

Western blot analysis revealed a significantly higher phosphorylated Jnk (pJnk) to Jnk in the right-hypertensive medullas (gray bars) relative to the left servo-controlled medullas (black bars). **P < 0.01, paired t-test, n = 8.

Seahorse analysis of mitochondrial function finds decreased respiratory control ratio in the right-hypertensive renal outer medulla.

Both DAVID and IPA nominated oxidative phosphorylation in the renal outer medulla as a pathway that differed between right-hypertensive outer medullas and left servo-controlled outer medullas. To further explore this pathway, we performed a mitochondrial stress test on isolated mitochondria from the left and right outer medullas of servo-controlled rats (Fig. 4). During the low-salt control period the average right RPP in this group of rats was 129 ± 2 mmHg and the left RPP was 124 ± 2 mmHg. After 7 days of HS the right RPP had increased to 150 ± 3 mmHg, whereas the left RPP remained stable at 124 ± 2 mmHg. The RCR was calculated as the ratio of the ADP-stimulated state 3 respiration (maximally stimulated respiration) and the oligomycin-stimulated state 4 respiration (inhibition of ATP synthase, residual oxygen consumption is due to proton leak across the inner mitochondrial membrane). A high RCR is reflective of a high capacity for ATP turnover and substrate oxidation with low proton leak and therefore indicates healthy mitochondria. In contrast, reductions in RCR reflect mitochondria dysfunction (2). A significant decrease in the RCR of the right-hypertensive outer medulla’s mitochondria was observed compared with mitochondria of the left servo-controlled outer medulla (7.9 ± 1.2 vs. 10.4 ± 1.8, n = 6, P < 0.01). This difference in RCR was driven by a significant increase in the state 4 respiration (31.6 ± 8.0 vs. 24.1 ± 7.5, P < 0.05) of the hypertensive outer medullas in the absence of a significant difference in state 3 respiration (243.4 ± 66.8 vs. 241.5 ± 83.1, P = 0.9).

Fig. 4.

Seahorse analysis reveals reduced respiratory control ratio (RCR) in mitochondria from the right hypertensive medulla relative to the left servo-controlled. A: mitochondria were isolated, then attached to an XF96 Seahorse plate, and then stimulated with ADP to induce state 3 respiration, oligomycin to induce state 4, FCCP as an uncoupler, and antimycin A, which blocks respiration. Data represent a single rat; error bars are from triplicate technical replicates. B: RCR is calculated as the quotient of ADP stimulated state 3 respiration and oligomycin stimulated state 4 respiration. When replotted such that data are expressed as % of oligo oxygen consumption rate (OCR), the distinction between left and right post-ADP values is representative of RCR and is increased in the left servo-controlled relative to the right-hypertensive. C: significant difference between left servo-controlled and right-hypertensive RCR values is observed. **P < 0.01 Statistical significance was determined by a paired t-test, n = 6.

Metabolomic analysis finds differentially regulated metabolites in multiple metabolic pathways.

The unpaired BRA nominated differences in purine metabolism that prompted us to examine this and related metabolic pathways. GC-MS and LC-MS were performed to quantitatively assess 62 different metabolites in frozen left and right renal outer medullas from newly prepared servo-controlled rats. In the five rats used in the analysis, the average right RPP during the control salt period was 132 ± 2 mmHg and increased to 154 ± 3 mmHg by day 7 HS. The average left RPP was 129 ± 2 mmHg during control period and was maintained at this level throughout the HS challenge, averaging 128 ± 2 mmHg by day 7 HS. After we removed one extreme outlier, a PCA of the normalized metabolite levels showed a clear separation between the left and right outer medullas (Fig. 5A). Paired t-test on the individual metabolites revealed 14 to be significantly different between left and right outer medullas (Fig. 5B). These include 11 amino acids, two carbohydrates, one fatty acid, and one nucleotide. Although the metabolomic analysis was performed to validate the predicted differences in purine metabolism, no differences in purines were detected. This finding may be due to the sensitivity of the follow-up analysis, or the pathway finding may be indicative of other physiological findings such as infiltration of T-lymphocytes, which are obligate users of the de novo purine synthesis pathway.

DISCUSSION

The current study and approach provide an example of the insights that can be revealed when sequencing techniques are combined with specialized experimental approaches that provide high-quality, well-controlled data. This analysis specifically provides an unbiased examination of the transcripts regulated by pressure in the SS rat fed an HS diet. The servo-control system allowed us to precisely examine the effect of pressure on kidneys while eliminating confounding factors such as hormonal milieu, sympathetic drive, and the secondary effects of salt intake. The pairing of the data between the left (servo-controlled) and right kidneys within each rat together with breadth of coverage provided by RNA-Seq analysis enabled us to more precisely determine more subtle changes in gene transcription than possible in previous studies.

Statistical analysis of RNA-Seq data.

Advances in transcriptome sequencing have expanded the breadth of coverage and reduced bias toward previous genes of interest. This is further enhanced by use of three separate statistical analyses. The first of these, DAVID, and the second, IPA, both estimate P values based upon overrepresentation of genes known to be related to biological pathways of interest (8, 21). BRA enables a more direct use of the quantitative expression data provided by RNA-Seq and estimates probabilities based upon the ability of a pathway to discriminate between two groups (34). Unlike the P values from DAVID and IPA, this is not dependent solely on the identities of the genes that are differentially regulated but rather uses the actual expression values of these identified genes to make this determination. This is demonstrated in the unique detection of the Jnk signaling pathway by the paired BRA. Although an insufficient absolute number of transcripts in this pathway were detected for significance as determined by DAVID/IPA, the expression-based signal detected by the BRA was sufficient to suggest Jnk signaling. The elevated pJnk/Jnk ratio observed by Western blot validated this prediction.

Furthermore, since the BRA is a regression based analysis, it can incorporate either discrete outcomes (i.e., right vs. left outer medulla) or continuous outcomes (i.e., RPP). Distinctions between the discrete and continuous results, as seen for example between chemokine and cytokine pathways, suggest that there may be differences in the causative signals underlying these pathway responses. These distinctions may be examined in future studies targeted at immune responses in hypertension.

Validation of nominated pathways.

Notably, several of the pathways nominated by all three analyses were related to inflammation, including “leukocyte transendothelial migration,” “chemokine signaling,” and “cytokine-cytokine receptor interaction.” This aligns with our recent observation that renal immune cell infiltration was significantly higher in the right-hypertensive kidney relative to the left servo-controlled kidney of servo-controlled rats fed an HS diet for 7 days (10). This observation provides support for the analytical approaches used in this study. Similarly, genes associated with fibrosis were also significantly higher in the right-hypertensive renal outer medulla than the left servo-controlled outer medulla, as reported in our previous servo-control study (10) and consistent with our previous microarray analysis (25). However, both of these previous studies selected fibrosis-related genes based upon physiological knowledge of the kidney. Interestingly, the pathway analysis carried out in the present paired analysis study did not find fibrosis to be significantly different. This is likely a consequence of the large number of fibrosis genes that are not specifically relevant to renal fibrosis per se, which emphasizes the need for kidney-specific pathway analysis tools, as was recently proposed (13).

Mitochondrial dysfunction in hypertension has been reported (3, 6, 9, 41). Impaired mitochondrial function has been implicated in age-related hypertension experimentally (31, 37), and genetic alterations in mitochondrial DNA are associated with hypertension in humans (33, 38). Previous studies from our department demonstrated ultrastructural abnormalities in the mitochondria of medullary thick ascending limbs of SS rats fed HS (15). Additionally, reduced O₂ utilization in the medullas of SS rats has been reported (41). Together these results suggest that mitochondrial function is altered in salt-sensitive hypertension. However, the effects of BP per se on mitochondrial function have not been previously explored. Seahorse measurements on isolated mitochondria assessed the RCR, a metric of mitochondrial function (2). RCR was significantly lower in the right-hypertensive outer medulla than in the left servo-controlled outer medulla, suggestive of reduced mitochondrial function and potentially mitochondrial damage.

Similarly, differences in the metabolome have been implicated in SS hypertension. Here, 14 metabolites were significantly different between high and low pressure outer medullary tissue. Eleven of these were amino acids, each of which was expressed at increased levels in the right-hypertensive outer medulla. Studies from our department have shown connections between specific amino acids and renal function through multiple mechanisms, including regulation of NO release (16, 18) and TCA cycle activity (35, 36). In this study we are able for the first time to demonstrate differences in the abundances of these amino acids as a result of pressure. Notably, our study found increased abundances of leucine and isoleucine in the hypertensive kidney, as predicted by the pathway analysis. Differences in isoleucine regulation in SS rats, relative to salt-resistant rats, have been shown (39). Since isoleucine is converted into acetyl-CoA and succinyl-CoA, this may reflect the known differences in TCA function in SS rats. The data presented suggest that these differences are driven at least partially by pressure. Phenylalanine and tyrosine were also elevated, consistent with many studies showing that dopamine signaling is impaired in humans and animal models of salt-sensitive hypertension including SS rats (14, 17, 26, 27), again suggesting that these differences are the result of differences in RPP.

Summary

The approaches described within this manuscript demonstrate the advantage of combining precise experimental manipulation (servo-control) with accurate and broad measurements of function (RNA-Seq). The paired nature of the kidneys offers a powerful resource that can be taken advantage of with careful experimental design. Using this approach, we have shown that differences in transcriptome can be used to accurately predict both subtle and profound differences in physiological outcomes such as immune cell infiltration, mitochondrial function and metabolomic state.

GRANTS

A. W. Cowley Jr. is supported by the National Institutes of Health (NIH) via the National Heart, Lung and Blood Institute (HL-116264 and HL-082798). L.C. Evans is funded by an American Heart Association Scientist Development Grant (AHA17SDG33660574). A. Dayton is funded by NIH Grant 5F30HL127979-02. This work utilized Core Services supported by NIH Grant DK-097153 to the University of Michigan.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

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