Renal medullary gene expression in aquaporin-1 null mice
http://www.100md.com
《美国生理学杂志》
Department of Physiology, College of Medicine, and Arizona Research Laboratories, Genomics Research Laboratory, University of Arizona, Tucson, Arizona
ABSTRACT
Mice that lack the aquaporin-1 gene (AQP1) lack a functional countercurrent multiplier mechanism, fail to concentrate the inner medullary (IM) interstitium, and present with a urinary concentrating defect. In this study, we use DNA microarrays to identify the gene expression profile of the IM of AQP1 null mice and corresponding changes in gene expression resulting from a loss of a hypertonic medullary interstitium. An ANOVA analysis model, CARMA, was used to isolate the knockout effect while taking into account experimental variability associated with microarray studies. In this study 5,701 genes of the possible 12,000 genes on the array were included in the ANOVA; 531 genes were identified as demonstrating a >1.5-fold up- or downregulation between the wild-type and knockout groups. We randomly selected 35 genes for confirmation by real-time PCR, and 29 of the 35 genes were confirmed using this method. The overall pattern of gene expression in the AQP1 null mice was one of downregulation compared with gene expression in the renal medullas of the wild-type mice. Heat shock proteins 105 and 94, aldose reductase, adenylate kinase 2, aldolase B, aldehyde reductase 6, and p8 were decreased in the AQP1 null mice. Carboxylesterase 3, matrilin 2, lipocalin 2, and transforming growth factor- were increased in IM of AQP1 null mice. In addition, we observed a loss of vasopressin type 2 receptor mRNA expression in renal medullas of the AQP1 null mice. Thus the loss of the hyperosmotic renal interstitium, due to a loss of the concentrating mechanism, drastically altered not only the phenotype of these animals but also their renal medullary gene expression profile.
osmolarity; aquaporin; microarray; real-time PCR; vasopressin
IN THE KIDNEY, AQUAPORIN-1 water channels (AQP1) have been shown to be expressed in the apical and basolateral epithelial cell membranes of proximal tubules, thin descending limbs of Henle's loop, and in endothelial cells of descending vasa recta (25–27). Countercurrent multiplication relies on active solute transport in the thick ascending limb of Henle (TAL) and the rapid osmotic equilibration along the thin descending limb of Henle (20); therefore, the presence of AQP1 in the thin descending limbs plays a vital role in this mechanism. A study of AQP1 null mice demonstrated their failure to produce concentrated urine on water deprivation (8). Further studies demonstrated that fluid reabsorption along the proximal tubule of the AQP1 null mice was reduced, due to the impairment of near-isosmolar fluid reabsorption in this segment (36). The reduction in proximal tubule fluid reabsorption in these mice did not translate to an increase in the distal delivery of fluid (32). Moreover, the loss of an effect of a V2 receptor-specific agonist, 1-deamino-8-D-arginine vasopressin (dDAVP), which should equalize urine and medullary interstitial osmolality, also indicated that the medullary interstitium of AQP1 null mice is not appropriately hypertonic. Therefore, the diuresis seen in AQP1 null mice resulted primarily from reduced fluid absorption in the collecting ducts (32). Taken together, these results suggest that the primary renal defect in AQP1 null mice is the inability to generate a hypertonic medullary interstitium by countercurrent multiplication (22). Thus these animals can be considered to be "countercurrent multiplier knockouts," providing a tool to investigate how the loss of high local osmolality affects inner medullary cell gene expression in vivo, and the present study uses DNA microarrays to identify this effect. These animals could also provide us with a tool to investigate the independent effects of vasopressin and high local osmolality on gene expression in renal cells in vivo.
METHODS
Animals. Mice lacking aquaporin-1 (AQP1) were generated by homologous recombination in embryonic stem cells as previously reported (22). Animals were bred and maintained in the animal facility of the University of Arizona Health Sciences Center under National Institutes of Health guidelines. AQP1 genotypes designated (+) for the wild-type allele and (–) for the targeted allele were determined by PCR analysis of genomic DNA isolated from tail biopsies. F2 generation AQP1 +/+ and –/– male animals derived from crosses of AQP1 +/– were used in our studies. We used mice that were 10 wk old. Mice received regular food and water ad libitum.
RNA isolation, amplification, and cDNA purification. Full methodology for the RNA purification, amplification, and cDNA production can be found in supplemental data C. RNA was isolated from mouse inner medullas using a Qiagen RNeasy Mini Kit (cat. no. 74104) according to the manufacturer's protocol for isolation from tissue. The RNA was quantified using a spectrophotometer and run on a 1% agarose gel to determine both its quality and purity. Throughout the following procedures, all samples were kept as individual samples (i.e., from a single mouse) and termed C1C3 for wild-type mice and E1E3 for AQP1 null mice. RNA was amplified using a MessageAmp kit (Ambion cat. no. 1750) according to the manufacturer's protocols. Three micrograms of total RNA from each individual mouse were used as a template for each amplification reaction, and this gave a yield of 50 μg of amplified RNA (aRNA). This amplified RNA was then reverse transcribed to cDNA using an EndoFree RT kit (Ambion cat. no. 1740) according to the manufacturer's protocol. Amino allyl-modified cDNA was purified using PCR purification columns (Qiagen cat. no. 28104) according to the manufacturer's protocol. The modified cDNA was labeled with Alexa dyes via free amine modification (Molecular Probes, Eugene, OR, A20002 [GenBank] Alexa Fluor 546 and A20006 [GenBank] Alexa Fluor 647).
Microarray slide preparation and hybridization. Microarrays for our study were prepared within the Genomic Research Laboratory (GRL) at the University of Arizona using the NIA mouse 15K clone set http://lgsun.grc.nia.nih.gov/cDNA/15k.html. Full methodology for the production of the microarrays and the hybridization protocols can be found in supplemental data C. We randomly selected 15 clones from our array results for sequence confirmation, and all 15 were identical to that published for the clone set on http://lgsun.grc.nia.nih.gov/, confirming the accuracy of the print and the clone set used to make the arrays. Labeled modified cDNA in hybridization buffer was loaded onto a slide and set to hybridize at 47°C for a minimum of 16 h. After completion, a short wash was run in the hybridization station after which the slide was removed and dipped in 0.05x SSC, to remove any residual nonhybridized cDNA. The slide was dried and analyzed using the arrayWORx eCCD-based microarray scanner from Applied Precision, capable of multichannel fluorescence scanning.
Microarray analysis. A multivariate experimental approach was used in the analysis of microarray data. This approach enables us to analyze a variety of variables in a microarray study (i.e., time course, treatment, condition, genotype) as well as identify and eliminate sources of experimental variance inherent to microarray data (i.e., array variation, dye performance). This approach is based on an ANOVA statistical model using a custom software package, CARMA (1). CARMA permits a robust characterization and classification of the data as well as provides outputs of residuals and other statistical parameters as references for users. Data were reduced and evaluated by CARMA and the results were submitted to the Gene Expression Omnibus (GEO) and can be found under GEO reference number GSE1298 [NCBI GEO] .
Real-time quantitative PCR. Real-time quantitative PCR was carried out using the RotorGene RG3000 (Corbett Research) sequence detection system and SYBR Green reagents from Qiagen (Quantitect Sybr Green PCR Kit, cat. no. 20414). Primers were designed using Primer3 software (29) and are listed in supplemental data B along with the gene accession number for the target gene. Three micrograms of total or amplified RNA were reverse transcribed with the Endofree RT kit, according to the manufacturer's protocol (Ambion). The cDNA was diluted to 8 ng/μl and the PCR reaction mixture contained 5 μl of Sybr master mix, 0.4 μl 25 mM MgCl2, 0.6 μl RNAse-free water, 100 pmol of forward and reverse primers, and 16 ng cDNA, in a volume of 10 μl. Each reaction was performed in triplicate at 95°C, 15 min; then 95°C, 15 s, and 58°C, 15 s, and 20 s at 72°C for 40 cycles. This was followed by a melt cycle that consisted of stepwise increase in temperature from 72 to 99°C.
A single predominant peak was observed in the dissociation curve of each gene, supporting the specificity of the PCR product. Ct numbers (threshold values) were set within the exponential phase of the PCR and were used to calculate the expression levels of the genes of interest and were normalized to endogenous cellular dynactin RNA. The level of dynactin RNA was measured in parallel samples using dynactin-specific primers.
RESULTS
Microarray analysis. We isolated total RNA from the medullas of wild-type (samples C1, C2, C3) and AQP1 null mice (samples E1, E2, E3) for microarray analysis of gene expression. Total RNA from each sample was amplified before being labeled in the cDNA reaction. The hybridization scheme we used for microarray analysis is based on the interwoven loop design as previously described (19). Each sample from an individual mouse was hybridized to an array four times, generating four independent replicate measures of that particular RNA sample (see MIAME document, supplemental data). An ANOVA model was used to isolate the knockout effect while taking into account experimental effects (array, dye, spot) and thus determined the effect of AQP1 expression on gene expression patterns in the renal medulla. The two channels for each array were normalized using intensity- and location-dependent Lowess regression. Data were first transformed using a Loglin function, which performs a log transformation for higher intensities, and a linear transformation for lower intensities (16). The ANOVA was performed on a gene by gene basis and was limited to genes that were measured confidently on a minimum of three of the four hybridizations, for at least one sample. In this study 5,701 genes of thepossible 12,000 genes on the array were included in the ANOVA. We considered genes to be significantly differentially expressed if the ANOVA P value was <0.05. Genes that exhibited small changes in gene expression but were identified as significant because of unusually small variance were excluded based on a cutoff of >1.5-fold up- or downregulation between the wild-type and AQP1 null mice; 531 genes were in this selected group. An output file showing individual measurements, gene name (if known) and links to the NIA and Genbank databases was generated (see supplemental data A). A large proportion of the genes selected as significantly differentially expressed were genes showing a downregulation in expression in the renal medullas of the AQP1 null mice compared with wild-type controls; of 531 genes in the microarray output, 465 genes were downregulated and 66 genes were upregulated in the AQP1 null mice compared with wild-type mice.
Multiple genes identified as encoding members of the mitochondrial electron transport chain were downregulated in the renal medullas of the AQP1 null mice (see supplemental data A). Among those with the largest decrease in expression in AQP1 null mice, compared with wild-type mice, were cytochrome c oxidase (Cox) subunits (Cox I, 5-fold decrease; Cox IVa, 3-fold decrease), mitochondrial H+-ATP synthase F1 and F0 complex subunits (ATP5C1, 2.5-fold decrease; ATP5L, 2.4-fold decrease), NADH dehydrogenase (ubiquinone) (NADH, 2.4-fold decrease), and malate dehydrogenase (MDH; 2-fold decrease). Several heat shock and stress genes were identified as being significantly decreased in expression in the AQP1 null mice; heat shock protein 105 kDa (HSP105; 5-fold decrease), osmotic stress protein 94 kDa (OSP94; 2.7-fold decrease), heat shock protein 30 kDa (HSP30; 1.7-fold decrease), heat-responsive protein 12 kDa (HRP12;, 1.7-fold decrease), aldose reductase (AR; 4-fold decrease), and the reactive oxygen species scavenger, superoxide dismutase 1 (SOD1; 2.4-fold decrease).
Validation of array results via real-time quantitative PCR. Quantitative real-time PCR was used to confirm the validity of our array analysis. We confirmed our results for a randomly selected number of genes that were either up- or downregulated in the AQP1 null mice, compared with wild-type mice, using the same amplified RNA samples that were used for the microarrays. Additionally, we confirmed the up- and downregulation of specific genes in a second set of RNA samples from wild-type and AQP1 null mice; these RNA samples were not amplified before use in the real-time PCR in order to eliminate any potential error associated with amplification. The amplified RNA is considered a measure of the validity of the microarray analysis, and the total RNA a confirmation of the mRNA expression levels in a biological replicate. We used dynactin as an internal standard for the real-time analysis as it was unchanged on the microarray data. Initial observations for -actin and GAPDH in our array output demonstrated that both of these commonly used housekeeping genes were differentially expressed, with a greater than 1.5-fold decrease in expression in the AQP1 null animals compared with wild-type (see supplemental data A).
The data collected from the real-time PCR on amplified RNA samples were compared with that obtained from the microarray experiments and can be found in supplemental data D. The real-time PCR results confirmed the microarray data in 29 of 35 randomly selected genes. The data collected from the real-time PCR using the unamplified RNA, from a separate set of mice, were compared with that obtained from the microarray experiments and are presented graphically in Figs. 1, 2, 3, and 4. The real-time PCR results from the unamplified RNA confirmed the microarray data in 25 of 28 genes selected for confirmation. Figures 1–3 show the real-time PCR results for selected genes that decreased in the renal medullas of AQP1 null mice compared with wild-type mice. Figure 4 shows the results for the randomly selected genes that increased in the renal medullas of AQP1 null mice.
As previously mentioned, multiple genes encoding members of the mitochondrial electron transport chain were found to decrease in the microarray data and we selected several genes in order to reexamine the mRNA levels by real-time PCR in a separate set of mice. Figure 1 presents the data from the real-time PCR assay (3 wild-type vs. 3 AQP1 null mice) plotted against the corresponding microarray values. NADH dehydrogenase (0.63 ± 0.13 compared with 1.00 ± 0.1 in wild-type medullas, P < 0.05), the mitochondrial proton pump, H+-ATP synthase subunits ATP 5C1 (0.24 ± 0.01 compared with 1.00 ± 0.20 in wild-type medullas, P < 0.05), and ATP 5L (0.12 ± 0.03 compared with 1.00 ± 0.23 in wild-type medullas, P < 0.05) all demonstrated a significant decrease in mRNA abundance in AQP1 null compared with wild-type mice. We also demonstrated a significant decrease in the abundance of the Na+-K+-ATPase -subunit in the AQP1 null mice (0.53 ± 0.13 compared with 1.00 ± 0.16 in wild-type medullas, P < 0.05) and confirmed that adenylate kinase 2 (AK2), a metabolic enzyme often linked to mitochondrial metabolic rate, was significantly decreased in the AQP1 null mice (0.24 ± 0.06 compared with 1.00 ± 0.03 in wild-type medullas, P < 0.05).
Renal medullary cells are usually exposed to hyperosmotic conditions and many stress genes and heat shock proteins have been studied in this region of the kidney. Figure 2 presents the results for the real-time PCR assay (3 wild-type vs. 3 AQP1 null mice) plotted against the corresponding microarray values for several genes that fell into this functional category of heat and stress related genes. Heat shock protein 105 (HSP105) and osmotic stress protein 94 (OSP94) were confirmed as decreasing in abundance in the renal medullary cells of AQP1 null mice (HSP105, 0.18 ± 0.01 compared with 1.00 ± 0.10, P < 0.05; OSP94, 0.11 ± 0.01 compared with 1.00 ± 0.16, P < 0.05 in wild-type medullas). Aldose reductase was significantly decreased in the renal medullas of the AQP1 null mice (0.06 ± 0.01 compared with 1.00 ± 0.58 in wild-type medullas, P < 0.05). We also demonstrated that mRNA abundance for the serum and glucocorticoid regulated kinase 1 (SGK1), a signaling molecule known to mediate the effects of aldosterone on epithelial sodium transport, via the epithelial sodium channel (ENaC) (28) was decreased in abundance in the AQP1 null mice (0.12 ± 0.04 compared with 1.00 ± 0.35 in wild-type medullas, P < 0.05). Also shown in Fig. 2 are data for the mRNA expression of both P8, a gene whose function is unknown, and tissue plasminogen activator (tPA). Both were both significantly decreased in the AQP1 null mice (P8, 0.15 ± 0.03 compared with 1.00 ± 0.24, P < 0.05; tPA, 0.23 ± 0.03 compared with 1.00 ± 0.21, P < 0.05 in wild-type medullas).
Several genes, whose function in the kidney and more specifically function in the renal medulla are unknown, were reexamined by real-time PCR assay and these data are presented in Fig. 3. For example, mRNA levels for both cell growth-regulating nucleolar protein (LYAR) (0.31 ± 0.02 compared with 1.00 ± 0.06 in wild-type medullas, P < 0.05) and lymphocyte antigen 6 complex (Ly6) (0.4 ± 0.06 compared with 1.00 ± 0.11 in wild-type medullas, P < 0.05) were decreased in the AQP1 null mice.
Unlike the genes observed to decrease in the AQP1 null mice, there was no obvious pattern to be seen in the types of genes observed to increase in the AQP1 null mice. Of the 66 genes that showed a significant increase in the microarray results, we reexamined seven by real-time PCR and the results are presented in Fig. 4. Several of the genes we studied by real-time were classed in the array annotation as structural or matrix protein-encoding genes; carboxylesterase (CES3), matrilin 2 (MATN2) and mouse fat tumor suppressor homolog (mFAT1). The mRNA levels of all three were confirmed as significantly increased in the AQP1 null mice (CES3, 3.75 ± 0.57 compared with 1.00 ± 0.17, P < 0.05; MATN2, 7.77 ± 2.45 compared with 1.00 ± 0.41, P < 0.05; mFAT1, 1.97 ± 0.35 compared with 1.00 ± 0.09 in wild-type medullas, P < 0.05).
Several genes, known to be present in collecting duct cells, were not represented on our microarrays; therefore we used real-time PCR to investigate their relative expression levels in the AQP1 null mice. Figure 5 presents the mean relative expression of these selected genes in the AQP1 null mice relative to the wild-type mice plotted on a log scale (± SE). We observed a significant increase in mRNA abundance in the AQP1 null mice for all three of the epithelial sodium transporter subunits, ENaC- (4.92 ± 1.04 compared with 1.00 ± 0.56 in wild-type medullas, P < 0.05), ENaC- (3.98 ± 0.48 compared with 1.00 ± 0.31 in wild-type medullas, P < 0.05), and ENaC- (4.77 ± 0.77 compared with 1.00 ± 0.37 in wild-type medullas, P < 0.05). There was a significant increase in the abundance in the mRNA of AQP3 in the renal medulla of the AQP1 null mice (2.04 ± 0.38 compared with 1.00 ± 0.04 in wild-type medullas, P < 0.05). We also demonstrated that there was no significant difference between the abundance of AQP2 mRNA in the renal medulla of the two sets of animals (0.88 ± 0.49 compared with 1.00 ± 0.25 in wild-type, P = 0.26, not significant). AQP2 expression is known to be regulated in collecting duct principal cells following activation of type 2, vasopressin receptors (V2R) located on the basolateral membrane of collecting duct cells. We examined the levels of V2R messenger RNA in the medulla of the AQP1 null animals and observed a significant decrease in V2R abundance compared with wild-type animals (0.61 ± 0.06 compared with 1.00 ± 0.14 in wild-type medullas, P < 0.05). In contrast, vasopressin 1a receptor (V1aR) mRNA abundance was significantly increased in AQP1 null animals (4.77 ± 0.47 compared with 1.00 ± 0.11 in wild-type medullas, P < 0.05).
DISCUSSION
AQP1 null mice manifest a severe defect in urinary concentrating ability (22). When deprived of water, the mice are unable to concentrate their urine and conserve fluid. The increased urinary flow rates observed with these mice result from reduced fluid absorption in the collecting duct, due to the low medullary interstitial osmolality, and not as a result of an increase in distal delivery of fluid following the decrease in proximal tubule fluid reabsorption (32). The reduction of interstitial osmolality in the AQP1 null mice suggests that the cells of the renal medulla are not exposed to the high NaCl and urea concentrations normally associated with the environment of the renal medulla. The aim of this study was to investigate the gene expression profile of the renal medullary cells of AQP1 null mice compared with those of wild-type mice. We proceeded to analyze the gene expression profile via microarray analysis and confirmed, by real-time quantitative PCR analysis, many of the genes identified as differentially expressed. A dramatic downregulation of gene expression was observed in the renal medulla of the AQP1 null mice, compared with wild-type medulla, and included genes known to be localized to renal medullary cells and previously suggested to function in the protection of cells against osmotic stress. We also identified many genes whose localization and function in the renal medulla are not yet known. We will discuss a few of these observations in the context of renal function.
Heat shock/stress genes. The cells of the renal medulla are normally exposed to high concentrations of NaCl and urea that vary depending on whether the urine is dilute or concentrated. When exposed to hyperosmolar NaCl, renal epithelial cells are known to increase the expression of stress proteins such as HSP70 (31), along with proteins such as aldose reductase and the betaine transporter that allow the cells to accumulate organic osmolytes (14). Aldose reductase is an enzyme that catalyses the synthesis of sorbitol from glucose. In renal cells, hypertonicity is known to increase the transcription of the aldose reductase gene (33), which results in corresponding increases in both mRNA abundance and enzyme activity, thus allowing an increase in the accumulation of renal cell sorbitol (6), suggesting that these proteins help to stabilize cellular biochemical processes during osmotic stress (5). In our study, we demonstrate a decrease in the abundance of several transcripts known to encode heat- and stress-induced proteins, and we confirmed the decrease in mRNA levels for HSP 105, OSP 94 and aldose reductase in the AQP1 null animals, perhaps reflecting the loss of high interstitial osmolality in the kidneys of these mice. OSP94 is a member of the HSP110/SSE stress protein subfamily and likely acts as a molecular chaperone (21). In mouse kidney, previous studies showed that OSP94 mRNA expression paralleled the known corticomedullary osmolality gradient, with highest expression in the inner medulla. Moreover, inner medullary OSP94 expression was increased during water restriction when osmolality was known to increase (21). In concordance with our observations, studies in mIMCD3 cells demonstrated a decrease in the expression of OSP94 and aldose reductase mRNA when osmolarity was decreased (7).
It was reported recently that HSP105, which is known to be induced by heat stress and able to bind to p53 in a temperature-sensitive manner, is located in the renal medulla and has a similar pattern of distribution as OSP94 (23). In fact, the authors suggested that HSP105 and OSP94 are the same protein. Heat-responsive protein 12 (HRP12) is a 12-kDa protein of unknown function with significant sequence similarity to HSP70 and DnaK. It is endogenously expressed at high levels in the liver and the kidney, and its expression is known to be upregulated in response to heat shock (30). A recent proteomic study identified it as having a mitochondrial localization (24).
Our findings confirm and extend previous studies indicating a role for stress protein in renal medullary function. A detailed study into all known heat shock protein family members in the AQP1 null mice would be an interesting future study, as only two members of the family were selected in this study (as a measure to confirm our microarray results). Moreover, the specific roles for this family of proteins in the maintenance of cell integrity in the face of osmotic stress need to be elucidated.
Mitochondrial pathway genes. The kidney is an organ known to have high basal ATP turnover rates. In the kidney and other organs with high energy demands, the postnatal development of ATP-requiring functions is linked to the biogenesis of mitochondria (10, 35). In the thick ascending limb of Henle's loop, the main oxidative cell type of the renal medulla, this mitochondrial biogenesis occurs between 16 days after birth and adulthood, a time frame that corresponds to when the concentrating mechanisms develop in the renal medulla (11). Mitochondrial biogenesis in the kidney is characterized by a doubling of mitochondrial density as well as an increase in the activity of several mitochondrial oxidative enzymes. The rise in postnatal circulating glucocortocoids was shown to be responsible for changes in the numbers of mitochondria but not for the increase in enzyme activity (12).
We observed a decrease in expression of a range of transcripts that encode mitochondrial enzymes and proteins in our microarray study, confirming this general trend by real-time PCR. The F1/F0 ATPase subunits were significantly decreased in the renal medullas of the AQP1 null mice, suggesting that the respiratory capacity per mitochondrium may be decreased in these mice compared with wild-type mice. Malate dehydrogenase, the enzyme that catalyzes the final step of the citric acid cycle, was decreased. In addition, in the AQP1 null mice there was a significant decrease in genes that encode the cytochrome c oxidase complex of enzymes, along with an observed decrease in NADH dehydrogenase (ubiquinone), all proteins associated with the oxidative phosphorylation pathway of the inner mitochondrial membrane. Overall, our results suggest that either the total number of mitochondria in the renal medulla of the AQP1 null mice are reduced or that the activity of individual mitochondria is significantly altered. Interestingly, we also observed a decrease in the mRNA abundance of the Na+-K+-ATPase -subunit in the AQP1 null mice, which also suggests that active sodium transport across the basolateral membranes of the renal medullary cells in these mice is decreased. Because Na+-K+-ATPase activity is the primary ATP-consuming process in all renal cells, a decrease in oxidative capacity could be in response to the decrease in energetic load. Further studies, such as mitochondrial activity assays, and calculation of the number of mitochondria in the renal medullary cells of the AQP1 null mice are required to differentiate between these possibilities.
Vasopressin receptors. Vasopressin receptors are expressed in several cell types along the nephron. V2R are expressed on the basolateral membrane of collecting duct cells, and via activation of adenylyl cyclase mediate an increase in water, urea and sodium permeabilities of these cells. V1aR have been found on the apical membrane of cortical collecting duct cells (2, 18). V1aR activation in collecting duct cells of rabbits has been shown to increase prostaglandin synthesis, leading to the stimulation of phosphodiesterases (4) that would reduce cAMP levels in the cells and blunt the V2R-dependent stimulation of adenylyl cyclase (3). Receptors for V1a are also expressed in the vasa recta of the kidney and stimulation of the V1aR reduces renal blood flow in the medulla during antidiuresis (9).
We observed a significant decrease in the mRNA abundance for the V2R in the inner medulla of the AQP1 null mice. In contrast, the mRNA abundance of the V1aR was significantly increased in these mice compared with wild-type controls. The decrease in V2R message in the renal medulla suggests that vasopressin activation of the cAMP pathway in the renal medulla of these mice may be reduced. Our general observation from the microarray data was that the pattern of gene expression in the renal inner medulla of the AQP1 null mice was one of downregulation compared with genes expressed in the wild-type mice. This raises several questions about the role of AQP1 in establishing the hyperosmotic environment of the renal medulla. Are the genes downregulated in response to the loss of the hyperosmolarity in the renal medulla or is the presence of AQP1 required as an upstream regulator of several gene pathways P8, a transcription factor whose function is currently unknown, was shown to be induced in rat diabetic kidneys (15). P8 was significantly decreased in the cells of the AQP1 null mice. mRNA abundance for all three ENaC subunits was significantly increased in the AQP1 null mice. The abundance of this protein is known to be increased in diabetic kidneys (13, 34). Future studies are planned to compare the gene expression profiles of renal medullary cells in animal models of diseases that have been associated with defects in the handling of salt and water, such as diabetes. Additionally, the effect of water restriction on the gene expression profile of the medullary cells of the AQP1 null mice will be dissected by microarray studies. These studies will allow us to identify the effect of an increase in vasopressin on renal medullary gene expression in the absence of an increase in osmolarity.
In summary, gene array analysis, using a new approach for statistical identification of expression changes, CARMA, suggested that several families of proteins are downregulated in the renal medullas of AQP1 null mice. These global gene profile findings were verified by real-time quantitative PCR demonstrating the validity of our approach. Limitations due to the limited number of genes and gene families confirmed remain an issue, but the combination of gene array studies with subsequent verification clearly provides an approach for evaluating general changes in gene expression associated with models of renal dysfunction.
GRANTS
This work was funded by a grant from the University of Arizona Foundation (H. L. Brooks) and by National Institutes of Health Grant DK-064706 (H. L. Brooks).
ACKNOWLEDGMENTS
We are grateful to A. Verkman (UCSF) for providing the AQP1 null mice and to C. Weber (UA) for technical assistance.
Supplemental data sets A–D can be found at http://ajprenal.physiology.org/cgi/content/full/00207.2004/DC1.
FOOTNOTES
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
REFERENCES
Adamson C, Maitra N, Bahl J, Greer KA, Klewer S, Hoying JB, and Morkin E. Regulation of gene expression in cardiomyocytes by thyroid hormone and thyroid hormone analogs, 3,5 diiodothyropropionic acid and CGS 23425. JPET 311: 18, 2004.
Amorim JB and Malnic G. V1 receptors in luminal action of vasopressin on distal K+ secretion. Am J Physiol Renal Physiol 278: F809–F816, 2000.
Bankir L. Antidiuretic action of vasopressin: quantitative aspects and interaction between V1a and V2 receptormediated effects. Cardiovasc Res 51: 372–390, 2001.
Breyer MD and Ando Y. Hormonal signaling and regulation of salt and water transport in the collecting duct. Annu Rev Physiol 56: 711–739, 1994.
Burg MB. Response of renal inner medullary epithelial cells to osmotic stress. Comp Biochem Physiol A 133: 661–666, 2002.
Burg MB, Kwon ED, and Kultz D. Regulation of gene expression by hypertonicity. Annu Rev Physiol 59: 437–455, 1997.
Cai Q, Ferraris JD, and Burg MB. Greater tolerance of renal medullary cells for a slow increase in osmolality is associated with enhanced expression of HSP70 and other osmoprotective genes. Am J Physiol Renal Physiol 286: F58–F67, 2004.
Chou CL, Knepper MA, Hoek AN, Brown D, Yang B, Ma T, and Verkman AS. Reduced water permeability and altered ultrastructure in thin descending limb of Henle in aquaporin1 null mice. J Clin Invest 103: 491–496, 1999.
Cowley AW Jr. Control of the renal medullary circulation by vasopressin V1 and V2 receptors in the rat. Exp Physiol 85: 223S–231S, 2000.
Djouadi F, Bastin J, Gilbert T, Rotig A, Rustin P, and MerletBenichou C. Mitochondrial biogenesis and development of respiratory chain enzymes in kidney cells: role of glucocorticoids. Am J Physiol Cell Physiol 267: C245–C254, 1994.
Djouadi F, Bastin J, Kelly DP, and Merlet Benichou C. Transcriptional regulation by glucocorticoids of mitochondrial oxidative enzyme genes in the developing rat kidney. Biochem J 315: 555–562, 1996.
Djouadi F, Wijkhuisen A, and Bastin J. Coordinate development of oxidative enzymes and NaKATPase in thick ascending limb: role of corticosteroids. Am J Physiol Renal Physiol 263: F237–F242, 1992.
Ecelbarger CA, Knepper MA, and Verbalis JG. Increased abundance of distal sodium transporters in rat kidney during vasopressin escape. J Am Soc Nephrol 12: 207–217, 2001.
GarciaPerez A and Burg MB. Role of organic osmolytes in adaptation of renal cells to high osmolality. J Membr Biol 119: 113, 1991.
Goruppi S, Bonventre JV, and Kyriakis JM. Signaling pathways and lateonset gene induction associated with renal mesangial cell hypertrophy. EMBO J 21: 5427–5436, 2002.
Greer KA and Hoying JB. Implementation of a statistical linear model for microarray data analysis. FASEB J 17: A78, 2003.
Ikeda M, Yoshitomi K, Imai M, and Kurokawa K. Cell Ca2+ response to luminal vasopressin in cortical collecting tubule principal cells. Kidney Int 45: 811–816, 1994.
Kerr MK, Martin M, and Churchill GA. Analysis of variance for gene expression microarray data. J Comput Biol 7: 819–837, 2000.
Knepper MA and Rector FC. Urinary concentration and dilution. In: The Kidney, edited by Brenner BM and Rector FC. Philadelphia, PA: Elsevier, 1995, p. 532–570.
Kojima R, Randall J, Brenner BM, and Gullans SR. Osmotic stress protein 94 (Osp94). A new member of the Hsp110/SSE gene subfamily. J Biol Chem 271: 12327–12332, 1996.
Ma T, Yang B, Gillespie A, Carlson EJ, Epstein CJ, and Verkman AS. Severely impaired urinary concentrating ability in transgenic mice lacking aquaporin1 water channels. J Biol Chem 273: 4296–4299, 1998.
Matsumori M, Itoh H, Toyoshima I, Komatsuda A, Sawada K, Fukuda J, Tanaka T, Okubo A, Kinouchi H, Mizoi K, Hama T, Suzuki A, Hamada F, Otaka M, Shoji Y, and Takada G. Characterization of the 105kDa molecular chaperone. Identification, biochemical properties, and localization. Eur J Biochem 269: 5632–5641, 2002.
Mootha VK, Bunkenborg J, Olsen JV, Hjerrild M, Wisniewski JR, Stahl E, Bolouri MS, Ray HN, Sihag S, Kamal M, Patterson N, Lander ES, and Mann M. Integrated analysis of protein composition, tissue diversity, and gene regulation in mouse mitochondria. Cell 115: 629–640, 2003.
Nielsen S, Pallone T, Smith BL, Christensen EI, Agre P, and Maunsbach AB. Aquaporin1 water channels in short and long loop descending thin limbs and in descending vasa recta in rat kidney. Am J Physiol Renal Physiol 268: F1023–F1037, 1995.
Nielsen S, Smith BL, Christensen EI, Knepper MA, and Agre P. CHIP28 water channels are localized in constitutively waterpermeable segments of the nephron. J Cell Biol 120: 371–383, 1993.
Pallone TL, Edwards A, Ma T, Silldorff EP, and Verkman AS. Requirement of aquaporin1 for NaCl-driven water transport across descending vasa recta. J Clin Invest 105: 215–222, 2000.
Pearce D. SGK1 regulation of epithelial sodium transport. Cell Physiol Biochem 13: 13–20, 2003.
Rozen S and Skaletsky HJ. Primer3 on the WWW for general users and for biologist programmers. In: Bioinformatics Methods and Protocols: Methods in Molecular Biology, edited by Krawetz S and Misener S. Totowa, NJ: Humana, 2000, p. 365–386.
Samuel SJ, Tzung SP, and Cohen SA. Hrp12, a novel heatresponsive, tissuespecific, phosphorylated protein isolated from mouse liver. Hepatology 25: 1213–1222, 1997.
Santos BC, Chevaile A, Kojima R, and Gullans SR. Characterization of the Hsp110/SSE gene family response to hyperosmolality and other stresses. Am J Physiol Renal Physiol 274: F1054–F1061, 1998.
Schnermann J, Chou CL, Ma T, Traynor T, Knepper MA, and Verkman AS. Defective proximal tubular fluid reabsorption in transgenic aquaporin1 null mice. Proc Natl Acad Sci USA 95: 9660–9664, 1998.
Smardo F, Burg M, and Garcia Perez A. Kidney aldose reductase gene transcriptionis osmotically regulated. Am J Physiol Cell Physiol 262: C776–C782, 1992.
Song J, Knepper MA, Verbalis JG, and Ecelbarger CA. Increased renal ENaC subunit and sodium transporter abundances in streptozotocininduced type 1 diabetes. Am J Physiol Renal Physiol 285: F1125–F1137, 2003.
Valcarce C, Navarrete RM, Encabo P, Loeches E, Satrustegui J, and Cuezva JM. Postnatal development of rat liver mitochondrial functions. The roles of protein synthesis and of adenine nucleotides. J Biol Chem 263: 77677775, 1988.
Vallon V, Verkman AS, and Schnermann J. Luminal hypotonicity in proximal tubules of aquaporin1knockout mice. Am J Physiol Renal Physiol 278: F1030–F1033, 2000.(Matthew R. McReynolds, Ka)
ABSTRACT
Mice that lack the aquaporin-1 gene (AQP1) lack a functional countercurrent multiplier mechanism, fail to concentrate the inner medullary (IM) interstitium, and present with a urinary concentrating defect. In this study, we use DNA microarrays to identify the gene expression profile of the IM of AQP1 null mice and corresponding changes in gene expression resulting from a loss of a hypertonic medullary interstitium. An ANOVA analysis model, CARMA, was used to isolate the knockout effect while taking into account experimental variability associated with microarray studies. In this study 5,701 genes of the possible 12,000 genes on the array were included in the ANOVA; 531 genes were identified as demonstrating a >1.5-fold up- or downregulation between the wild-type and knockout groups. We randomly selected 35 genes for confirmation by real-time PCR, and 29 of the 35 genes were confirmed using this method. The overall pattern of gene expression in the AQP1 null mice was one of downregulation compared with gene expression in the renal medullas of the wild-type mice. Heat shock proteins 105 and 94, aldose reductase, adenylate kinase 2, aldolase B, aldehyde reductase 6, and p8 were decreased in the AQP1 null mice. Carboxylesterase 3, matrilin 2, lipocalin 2, and transforming growth factor- were increased in IM of AQP1 null mice. In addition, we observed a loss of vasopressin type 2 receptor mRNA expression in renal medullas of the AQP1 null mice. Thus the loss of the hyperosmotic renal interstitium, due to a loss of the concentrating mechanism, drastically altered not only the phenotype of these animals but also their renal medullary gene expression profile.
osmolarity; aquaporin; microarray; real-time PCR; vasopressin
IN THE KIDNEY, AQUAPORIN-1 water channels (AQP1) have been shown to be expressed in the apical and basolateral epithelial cell membranes of proximal tubules, thin descending limbs of Henle's loop, and in endothelial cells of descending vasa recta (25–27). Countercurrent multiplication relies on active solute transport in the thick ascending limb of Henle (TAL) and the rapid osmotic equilibration along the thin descending limb of Henle (20); therefore, the presence of AQP1 in the thin descending limbs plays a vital role in this mechanism. A study of AQP1 null mice demonstrated their failure to produce concentrated urine on water deprivation (8). Further studies demonstrated that fluid reabsorption along the proximal tubule of the AQP1 null mice was reduced, due to the impairment of near-isosmolar fluid reabsorption in this segment (36). The reduction in proximal tubule fluid reabsorption in these mice did not translate to an increase in the distal delivery of fluid (32). Moreover, the loss of an effect of a V2 receptor-specific agonist, 1-deamino-8-D-arginine vasopressin (dDAVP), which should equalize urine and medullary interstitial osmolality, also indicated that the medullary interstitium of AQP1 null mice is not appropriately hypertonic. Therefore, the diuresis seen in AQP1 null mice resulted primarily from reduced fluid absorption in the collecting ducts (32). Taken together, these results suggest that the primary renal defect in AQP1 null mice is the inability to generate a hypertonic medullary interstitium by countercurrent multiplication (22). Thus these animals can be considered to be "countercurrent multiplier knockouts," providing a tool to investigate how the loss of high local osmolality affects inner medullary cell gene expression in vivo, and the present study uses DNA microarrays to identify this effect. These animals could also provide us with a tool to investigate the independent effects of vasopressin and high local osmolality on gene expression in renal cells in vivo.
METHODS
Animals. Mice lacking aquaporin-1 (AQP1) were generated by homologous recombination in embryonic stem cells as previously reported (22). Animals were bred and maintained in the animal facility of the University of Arizona Health Sciences Center under National Institutes of Health guidelines. AQP1 genotypes designated (+) for the wild-type allele and (–) for the targeted allele were determined by PCR analysis of genomic DNA isolated from tail biopsies. F2 generation AQP1 +/+ and –/– male animals derived from crosses of AQP1 +/– were used in our studies. We used mice that were 10 wk old. Mice received regular food and water ad libitum.
RNA isolation, amplification, and cDNA purification. Full methodology for the RNA purification, amplification, and cDNA production can be found in supplemental data C. RNA was isolated from mouse inner medullas using a Qiagen RNeasy Mini Kit (cat. no. 74104) according to the manufacturer's protocol for isolation from tissue. The RNA was quantified using a spectrophotometer and run on a 1% agarose gel to determine both its quality and purity. Throughout the following procedures, all samples were kept as individual samples (i.e., from a single mouse) and termed C1C3 for wild-type mice and E1E3 for AQP1 null mice. RNA was amplified using a MessageAmp kit (Ambion cat. no. 1750) according to the manufacturer's protocols. Three micrograms of total RNA from each individual mouse were used as a template for each amplification reaction, and this gave a yield of 50 μg of amplified RNA (aRNA). This amplified RNA was then reverse transcribed to cDNA using an EndoFree RT kit (Ambion cat. no. 1740) according to the manufacturer's protocol. Amino allyl-modified cDNA was purified using PCR purification columns (Qiagen cat. no. 28104) according to the manufacturer's protocol. The modified cDNA was labeled with Alexa dyes via free amine modification (Molecular Probes, Eugene, OR, A20002 [GenBank] Alexa Fluor 546 and A20006 [GenBank] Alexa Fluor 647).
Microarray slide preparation and hybridization. Microarrays for our study were prepared within the Genomic Research Laboratory (GRL) at the University of Arizona using the NIA mouse 15K clone set http://lgsun.grc.nia.nih.gov/cDNA/15k.html. Full methodology for the production of the microarrays and the hybridization protocols can be found in supplemental data C. We randomly selected 15 clones from our array results for sequence confirmation, and all 15 were identical to that published for the clone set on http://lgsun.grc.nia.nih.gov/, confirming the accuracy of the print and the clone set used to make the arrays. Labeled modified cDNA in hybridization buffer was loaded onto a slide and set to hybridize at 47°C for a minimum of 16 h. After completion, a short wash was run in the hybridization station after which the slide was removed and dipped in 0.05x SSC, to remove any residual nonhybridized cDNA. The slide was dried and analyzed using the arrayWORx eCCD-based microarray scanner from Applied Precision, capable of multichannel fluorescence scanning.
Microarray analysis. A multivariate experimental approach was used in the analysis of microarray data. This approach enables us to analyze a variety of variables in a microarray study (i.e., time course, treatment, condition, genotype) as well as identify and eliminate sources of experimental variance inherent to microarray data (i.e., array variation, dye performance). This approach is based on an ANOVA statistical model using a custom software package, CARMA (1). CARMA permits a robust characterization and classification of the data as well as provides outputs of residuals and other statistical parameters as references for users. Data were reduced and evaluated by CARMA and the results were submitted to the Gene Expression Omnibus (GEO) and can be found under GEO reference number GSE1298 [NCBI GEO] .
Real-time quantitative PCR. Real-time quantitative PCR was carried out using the RotorGene RG3000 (Corbett Research) sequence detection system and SYBR Green reagents from Qiagen (Quantitect Sybr Green PCR Kit, cat. no. 20414). Primers were designed using Primer3 software (29) and are listed in supplemental data B along with the gene accession number for the target gene. Three micrograms of total or amplified RNA were reverse transcribed with the Endofree RT kit, according to the manufacturer's protocol (Ambion). The cDNA was diluted to 8 ng/μl and the PCR reaction mixture contained 5 μl of Sybr master mix, 0.4 μl 25 mM MgCl2, 0.6 μl RNAse-free water, 100 pmol of forward and reverse primers, and 16 ng cDNA, in a volume of 10 μl. Each reaction was performed in triplicate at 95°C, 15 min; then 95°C, 15 s, and 58°C, 15 s, and 20 s at 72°C for 40 cycles. This was followed by a melt cycle that consisted of stepwise increase in temperature from 72 to 99°C.
A single predominant peak was observed in the dissociation curve of each gene, supporting the specificity of the PCR product. Ct numbers (threshold values) were set within the exponential phase of the PCR and were used to calculate the expression levels of the genes of interest and were normalized to endogenous cellular dynactin RNA. The level of dynactin RNA was measured in parallel samples using dynactin-specific primers.
RESULTS
Microarray analysis. We isolated total RNA from the medullas of wild-type (samples C1, C2, C3) and AQP1 null mice (samples E1, E2, E3) for microarray analysis of gene expression. Total RNA from each sample was amplified before being labeled in the cDNA reaction. The hybridization scheme we used for microarray analysis is based on the interwoven loop design as previously described (19). Each sample from an individual mouse was hybridized to an array four times, generating four independent replicate measures of that particular RNA sample (see MIAME document, supplemental data). An ANOVA model was used to isolate the knockout effect while taking into account experimental effects (array, dye, spot) and thus determined the effect of AQP1 expression on gene expression patterns in the renal medulla. The two channels for each array were normalized using intensity- and location-dependent Lowess regression. Data were first transformed using a Loglin function, which performs a log transformation for higher intensities, and a linear transformation for lower intensities (16). The ANOVA was performed on a gene by gene basis and was limited to genes that were measured confidently on a minimum of three of the four hybridizations, for at least one sample. In this study 5,701 genes of thepossible 12,000 genes on the array were included in the ANOVA. We considered genes to be significantly differentially expressed if the ANOVA P value was <0.05. Genes that exhibited small changes in gene expression but were identified as significant because of unusually small variance were excluded based on a cutoff of >1.5-fold up- or downregulation between the wild-type and AQP1 null mice; 531 genes were in this selected group. An output file showing individual measurements, gene name (if known) and links to the NIA and Genbank databases was generated (see supplemental data A). A large proportion of the genes selected as significantly differentially expressed were genes showing a downregulation in expression in the renal medullas of the AQP1 null mice compared with wild-type controls; of 531 genes in the microarray output, 465 genes were downregulated and 66 genes were upregulated in the AQP1 null mice compared with wild-type mice.
Multiple genes identified as encoding members of the mitochondrial electron transport chain were downregulated in the renal medullas of the AQP1 null mice (see supplemental data A). Among those with the largest decrease in expression in AQP1 null mice, compared with wild-type mice, were cytochrome c oxidase (Cox) subunits (Cox I, 5-fold decrease; Cox IVa, 3-fold decrease), mitochondrial H+-ATP synthase F1 and F0 complex subunits (ATP5C1, 2.5-fold decrease; ATP5L, 2.4-fold decrease), NADH dehydrogenase (ubiquinone) (NADH, 2.4-fold decrease), and malate dehydrogenase (MDH; 2-fold decrease). Several heat shock and stress genes were identified as being significantly decreased in expression in the AQP1 null mice; heat shock protein 105 kDa (HSP105; 5-fold decrease), osmotic stress protein 94 kDa (OSP94; 2.7-fold decrease), heat shock protein 30 kDa (HSP30; 1.7-fold decrease), heat-responsive protein 12 kDa (HRP12;, 1.7-fold decrease), aldose reductase (AR; 4-fold decrease), and the reactive oxygen species scavenger, superoxide dismutase 1 (SOD1; 2.4-fold decrease).
Validation of array results via real-time quantitative PCR. Quantitative real-time PCR was used to confirm the validity of our array analysis. We confirmed our results for a randomly selected number of genes that were either up- or downregulated in the AQP1 null mice, compared with wild-type mice, using the same amplified RNA samples that were used for the microarrays. Additionally, we confirmed the up- and downregulation of specific genes in a second set of RNA samples from wild-type and AQP1 null mice; these RNA samples were not amplified before use in the real-time PCR in order to eliminate any potential error associated with amplification. The amplified RNA is considered a measure of the validity of the microarray analysis, and the total RNA a confirmation of the mRNA expression levels in a biological replicate. We used dynactin as an internal standard for the real-time analysis as it was unchanged on the microarray data. Initial observations for -actin and GAPDH in our array output demonstrated that both of these commonly used housekeeping genes were differentially expressed, with a greater than 1.5-fold decrease in expression in the AQP1 null animals compared with wild-type (see supplemental data A).
The data collected from the real-time PCR on amplified RNA samples were compared with that obtained from the microarray experiments and can be found in supplemental data D. The real-time PCR results confirmed the microarray data in 29 of 35 randomly selected genes. The data collected from the real-time PCR using the unamplified RNA, from a separate set of mice, were compared with that obtained from the microarray experiments and are presented graphically in Figs. 1, 2, 3, and 4. The real-time PCR results from the unamplified RNA confirmed the microarray data in 25 of 28 genes selected for confirmation. Figures 1–3 show the real-time PCR results for selected genes that decreased in the renal medullas of AQP1 null mice compared with wild-type mice. Figure 4 shows the results for the randomly selected genes that increased in the renal medullas of AQP1 null mice.
As previously mentioned, multiple genes encoding members of the mitochondrial electron transport chain were found to decrease in the microarray data and we selected several genes in order to reexamine the mRNA levels by real-time PCR in a separate set of mice. Figure 1 presents the data from the real-time PCR assay (3 wild-type vs. 3 AQP1 null mice) plotted against the corresponding microarray values. NADH dehydrogenase (0.63 ± 0.13 compared with 1.00 ± 0.1 in wild-type medullas, P < 0.05), the mitochondrial proton pump, H+-ATP synthase subunits ATP 5C1 (0.24 ± 0.01 compared with 1.00 ± 0.20 in wild-type medullas, P < 0.05), and ATP 5L (0.12 ± 0.03 compared with 1.00 ± 0.23 in wild-type medullas, P < 0.05) all demonstrated a significant decrease in mRNA abundance in AQP1 null compared with wild-type mice. We also demonstrated a significant decrease in the abundance of the Na+-K+-ATPase -subunit in the AQP1 null mice (0.53 ± 0.13 compared with 1.00 ± 0.16 in wild-type medullas, P < 0.05) and confirmed that adenylate kinase 2 (AK2), a metabolic enzyme often linked to mitochondrial metabolic rate, was significantly decreased in the AQP1 null mice (0.24 ± 0.06 compared with 1.00 ± 0.03 in wild-type medullas, P < 0.05).
Renal medullary cells are usually exposed to hyperosmotic conditions and many stress genes and heat shock proteins have been studied in this region of the kidney. Figure 2 presents the results for the real-time PCR assay (3 wild-type vs. 3 AQP1 null mice) plotted against the corresponding microarray values for several genes that fell into this functional category of heat and stress related genes. Heat shock protein 105 (HSP105) and osmotic stress protein 94 (OSP94) were confirmed as decreasing in abundance in the renal medullary cells of AQP1 null mice (HSP105, 0.18 ± 0.01 compared with 1.00 ± 0.10, P < 0.05; OSP94, 0.11 ± 0.01 compared with 1.00 ± 0.16, P < 0.05 in wild-type medullas). Aldose reductase was significantly decreased in the renal medullas of the AQP1 null mice (0.06 ± 0.01 compared with 1.00 ± 0.58 in wild-type medullas, P < 0.05). We also demonstrated that mRNA abundance for the serum and glucocorticoid regulated kinase 1 (SGK1), a signaling molecule known to mediate the effects of aldosterone on epithelial sodium transport, via the epithelial sodium channel (ENaC) (28) was decreased in abundance in the AQP1 null mice (0.12 ± 0.04 compared with 1.00 ± 0.35 in wild-type medullas, P < 0.05). Also shown in Fig. 2 are data for the mRNA expression of both P8, a gene whose function is unknown, and tissue plasminogen activator (tPA). Both were both significantly decreased in the AQP1 null mice (P8, 0.15 ± 0.03 compared with 1.00 ± 0.24, P < 0.05; tPA, 0.23 ± 0.03 compared with 1.00 ± 0.21, P < 0.05 in wild-type medullas).
Several genes, whose function in the kidney and more specifically function in the renal medulla are unknown, were reexamined by real-time PCR assay and these data are presented in Fig. 3. For example, mRNA levels for both cell growth-regulating nucleolar protein (LYAR) (0.31 ± 0.02 compared with 1.00 ± 0.06 in wild-type medullas, P < 0.05) and lymphocyte antigen 6 complex (Ly6) (0.4 ± 0.06 compared with 1.00 ± 0.11 in wild-type medullas, P < 0.05) were decreased in the AQP1 null mice.
Unlike the genes observed to decrease in the AQP1 null mice, there was no obvious pattern to be seen in the types of genes observed to increase in the AQP1 null mice. Of the 66 genes that showed a significant increase in the microarray results, we reexamined seven by real-time PCR and the results are presented in Fig. 4. Several of the genes we studied by real-time were classed in the array annotation as structural or matrix protein-encoding genes; carboxylesterase (CES3), matrilin 2 (MATN2) and mouse fat tumor suppressor homolog (mFAT1). The mRNA levels of all three were confirmed as significantly increased in the AQP1 null mice (CES3, 3.75 ± 0.57 compared with 1.00 ± 0.17, P < 0.05; MATN2, 7.77 ± 2.45 compared with 1.00 ± 0.41, P < 0.05; mFAT1, 1.97 ± 0.35 compared with 1.00 ± 0.09 in wild-type medullas, P < 0.05).
Several genes, known to be present in collecting duct cells, were not represented on our microarrays; therefore we used real-time PCR to investigate their relative expression levels in the AQP1 null mice. Figure 5 presents the mean relative expression of these selected genes in the AQP1 null mice relative to the wild-type mice plotted on a log scale (± SE). We observed a significant increase in mRNA abundance in the AQP1 null mice for all three of the epithelial sodium transporter subunits, ENaC- (4.92 ± 1.04 compared with 1.00 ± 0.56 in wild-type medullas, P < 0.05), ENaC- (3.98 ± 0.48 compared with 1.00 ± 0.31 in wild-type medullas, P < 0.05), and ENaC- (4.77 ± 0.77 compared with 1.00 ± 0.37 in wild-type medullas, P < 0.05). There was a significant increase in the abundance in the mRNA of AQP3 in the renal medulla of the AQP1 null mice (2.04 ± 0.38 compared with 1.00 ± 0.04 in wild-type medullas, P < 0.05). We also demonstrated that there was no significant difference between the abundance of AQP2 mRNA in the renal medulla of the two sets of animals (0.88 ± 0.49 compared with 1.00 ± 0.25 in wild-type, P = 0.26, not significant). AQP2 expression is known to be regulated in collecting duct principal cells following activation of type 2, vasopressin receptors (V2R) located on the basolateral membrane of collecting duct cells. We examined the levels of V2R messenger RNA in the medulla of the AQP1 null animals and observed a significant decrease in V2R abundance compared with wild-type animals (0.61 ± 0.06 compared with 1.00 ± 0.14 in wild-type medullas, P < 0.05). In contrast, vasopressin 1a receptor (V1aR) mRNA abundance was significantly increased in AQP1 null animals (4.77 ± 0.47 compared with 1.00 ± 0.11 in wild-type medullas, P < 0.05).
DISCUSSION
AQP1 null mice manifest a severe defect in urinary concentrating ability (22). When deprived of water, the mice are unable to concentrate their urine and conserve fluid. The increased urinary flow rates observed with these mice result from reduced fluid absorption in the collecting duct, due to the low medullary interstitial osmolality, and not as a result of an increase in distal delivery of fluid following the decrease in proximal tubule fluid reabsorption (32). The reduction of interstitial osmolality in the AQP1 null mice suggests that the cells of the renal medulla are not exposed to the high NaCl and urea concentrations normally associated with the environment of the renal medulla. The aim of this study was to investigate the gene expression profile of the renal medullary cells of AQP1 null mice compared with those of wild-type mice. We proceeded to analyze the gene expression profile via microarray analysis and confirmed, by real-time quantitative PCR analysis, many of the genes identified as differentially expressed. A dramatic downregulation of gene expression was observed in the renal medulla of the AQP1 null mice, compared with wild-type medulla, and included genes known to be localized to renal medullary cells and previously suggested to function in the protection of cells against osmotic stress. We also identified many genes whose localization and function in the renal medulla are not yet known. We will discuss a few of these observations in the context of renal function.
Heat shock/stress genes. The cells of the renal medulla are normally exposed to high concentrations of NaCl and urea that vary depending on whether the urine is dilute or concentrated. When exposed to hyperosmolar NaCl, renal epithelial cells are known to increase the expression of stress proteins such as HSP70 (31), along with proteins such as aldose reductase and the betaine transporter that allow the cells to accumulate organic osmolytes (14). Aldose reductase is an enzyme that catalyses the synthesis of sorbitol from glucose. In renal cells, hypertonicity is known to increase the transcription of the aldose reductase gene (33), which results in corresponding increases in both mRNA abundance and enzyme activity, thus allowing an increase in the accumulation of renal cell sorbitol (6), suggesting that these proteins help to stabilize cellular biochemical processes during osmotic stress (5). In our study, we demonstrate a decrease in the abundance of several transcripts known to encode heat- and stress-induced proteins, and we confirmed the decrease in mRNA levels for HSP 105, OSP 94 and aldose reductase in the AQP1 null animals, perhaps reflecting the loss of high interstitial osmolality in the kidneys of these mice. OSP94 is a member of the HSP110/SSE stress protein subfamily and likely acts as a molecular chaperone (21). In mouse kidney, previous studies showed that OSP94 mRNA expression paralleled the known corticomedullary osmolality gradient, with highest expression in the inner medulla. Moreover, inner medullary OSP94 expression was increased during water restriction when osmolality was known to increase (21). In concordance with our observations, studies in mIMCD3 cells demonstrated a decrease in the expression of OSP94 and aldose reductase mRNA when osmolarity was decreased (7).
It was reported recently that HSP105, which is known to be induced by heat stress and able to bind to p53 in a temperature-sensitive manner, is located in the renal medulla and has a similar pattern of distribution as OSP94 (23). In fact, the authors suggested that HSP105 and OSP94 are the same protein. Heat-responsive protein 12 (HRP12) is a 12-kDa protein of unknown function with significant sequence similarity to HSP70 and DnaK. It is endogenously expressed at high levels in the liver and the kidney, and its expression is known to be upregulated in response to heat shock (30). A recent proteomic study identified it as having a mitochondrial localization (24).
Our findings confirm and extend previous studies indicating a role for stress protein in renal medullary function. A detailed study into all known heat shock protein family members in the AQP1 null mice would be an interesting future study, as only two members of the family were selected in this study (as a measure to confirm our microarray results). Moreover, the specific roles for this family of proteins in the maintenance of cell integrity in the face of osmotic stress need to be elucidated.
Mitochondrial pathway genes. The kidney is an organ known to have high basal ATP turnover rates. In the kidney and other organs with high energy demands, the postnatal development of ATP-requiring functions is linked to the biogenesis of mitochondria (10, 35). In the thick ascending limb of Henle's loop, the main oxidative cell type of the renal medulla, this mitochondrial biogenesis occurs between 16 days after birth and adulthood, a time frame that corresponds to when the concentrating mechanisms develop in the renal medulla (11). Mitochondrial biogenesis in the kidney is characterized by a doubling of mitochondrial density as well as an increase in the activity of several mitochondrial oxidative enzymes. The rise in postnatal circulating glucocortocoids was shown to be responsible for changes in the numbers of mitochondria but not for the increase in enzyme activity (12).
We observed a decrease in expression of a range of transcripts that encode mitochondrial enzymes and proteins in our microarray study, confirming this general trend by real-time PCR. The F1/F0 ATPase subunits were significantly decreased in the renal medullas of the AQP1 null mice, suggesting that the respiratory capacity per mitochondrium may be decreased in these mice compared with wild-type mice. Malate dehydrogenase, the enzyme that catalyzes the final step of the citric acid cycle, was decreased. In addition, in the AQP1 null mice there was a significant decrease in genes that encode the cytochrome c oxidase complex of enzymes, along with an observed decrease in NADH dehydrogenase (ubiquinone), all proteins associated with the oxidative phosphorylation pathway of the inner mitochondrial membrane. Overall, our results suggest that either the total number of mitochondria in the renal medulla of the AQP1 null mice are reduced or that the activity of individual mitochondria is significantly altered. Interestingly, we also observed a decrease in the mRNA abundance of the Na+-K+-ATPase -subunit in the AQP1 null mice, which also suggests that active sodium transport across the basolateral membranes of the renal medullary cells in these mice is decreased. Because Na+-K+-ATPase activity is the primary ATP-consuming process in all renal cells, a decrease in oxidative capacity could be in response to the decrease in energetic load. Further studies, such as mitochondrial activity assays, and calculation of the number of mitochondria in the renal medullary cells of the AQP1 null mice are required to differentiate between these possibilities.
Vasopressin receptors. Vasopressin receptors are expressed in several cell types along the nephron. V2R are expressed on the basolateral membrane of collecting duct cells, and via activation of adenylyl cyclase mediate an increase in water, urea and sodium permeabilities of these cells. V1aR have been found on the apical membrane of cortical collecting duct cells (2, 18). V1aR activation in collecting duct cells of rabbits has been shown to increase prostaglandin synthesis, leading to the stimulation of phosphodiesterases (4) that would reduce cAMP levels in the cells and blunt the V2R-dependent stimulation of adenylyl cyclase (3). Receptors for V1a are also expressed in the vasa recta of the kidney and stimulation of the V1aR reduces renal blood flow in the medulla during antidiuresis (9).
We observed a significant decrease in the mRNA abundance for the V2R in the inner medulla of the AQP1 null mice. In contrast, the mRNA abundance of the V1aR was significantly increased in these mice compared with wild-type controls. The decrease in V2R message in the renal medulla suggests that vasopressin activation of the cAMP pathway in the renal medulla of these mice may be reduced. Our general observation from the microarray data was that the pattern of gene expression in the renal inner medulla of the AQP1 null mice was one of downregulation compared with genes expressed in the wild-type mice. This raises several questions about the role of AQP1 in establishing the hyperosmotic environment of the renal medulla. Are the genes downregulated in response to the loss of the hyperosmolarity in the renal medulla or is the presence of AQP1 required as an upstream regulator of several gene pathways P8, a transcription factor whose function is currently unknown, was shown to be induced in rat diabetic kidneys (15). P8 was significantly decreased in the cells of the AQP1 null mice. mRNA abundance for all three ENaC subunits was significantly increased in the AQP1 null mice. The abundance of this protein is known to be increased in diabetic kidneys (13, 34). Future studies are planned to compare the gene expression profiles of renal medullary cells in animal models of diseases that have been associated with defects in the handling of salt and water, such as diabetes. Additionally, the effect of water restriction on the gene expression profile of the medullary cells of the AQP1 null mice will be dissected by microarray studies. These studies will allow us to identify the effect of an increase in vasopressin on renal medullary gene expression in the absence of an increase in osmolarity.
In summary, gene array analysis, using a new approach for statistical identification of expression changes, CARMA, suggested that several families of proteins are downregulated in the renal medullas of AQP1 null mice. These global gene profile findings were verified by real-time quantitative PCR demonstrating the validity of our approach. Limitations due to the limited number of genes and gene families confirmed remain an issue, but the combination of gene array studies with subsequent verification clearly provides an approach for evaluating general changes in gene expression associated with models of renal dysfunction.
GRANTS
This work was funded by a grant from the University of Arizona Foundation (H. L. Brooks) and by National Institutes of Health Grant DK-064706 (H. L. Brooks).
ACKNOWLEDGMENTS
We are grateful to A. Verkman (UCSF) for providing the AQP1 null mice and to C. Weber (UA) for technical assistance.
Supplemental data sets A–D can be found at http://ajprenal.physiology.org/cgi/content/full/00207.2004/DC1.
FOOTNOTES
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
REFERENCES
Adamson C, Maitra N, Bahl J, Greer KA, Klewer S, Hoying JB, and Morkin E. Regulation of gene expression in cardiomyocytes by thyroid hormone and thyroid hormone analogs, 3,5 diiodothyropropionic acid and CGS 23425. JPET 311: 18, 2004.
Amorim JB and Malnic G. V1 receptors in luminal action of vasopressin on distal K+ secretion. Am J Physiol Renal Physiol 278: F809–F816, 2000.
Bankir L. Antidiuretic action of vasopressin: quantitative aspects and interaction between V1a and V2 receptormediated effects. Cardiovasc Res 51: 372–390, 2001.
Breyer MD and Ando Y. Hormonal signaling and regulation of salt and water transport in the collecting duct. Annu Rev Physiol 56: 711–739, 1994.
Burg MB. Response of renal inner medullary epithelial cells to osmotic stress. Comp Biochem Physiol A 133: 661–666, 2002.
Burg MB, Kwon ED, and Kultz D. Regulation of gene expression by hypertonicity. Annu Rev Physiol 59: 437–455, 1997.
Cai Q, Ferraris JD, and Burg MB. Greater tolerance of renal medullary cells for a slow increase in osmolality is associated with enhanced expression of HSP70 and other osmoprotective genes. Am J Physiol Renal Physiol 286: F58–F67, 2004.
Chou CL, Knepper MA, Hoek AN, Brown D, Yang B, Ma T, and Verkman AS. Reduced water permeability and altered ultrastructure in thin descending limb of Henle in aquaporin1 null mice. J Clin Invest 103: 491–496, 1999.
Cowley AW Jr. Control of the renal medullary circulation by vasopressin V1 and V2 receptors in the rat. Exp Physiol 85: 223S–231S, 2000.
Djouadi F, Bastin J, Gilbert T, Rotig A, Rustin P, and MerletBenichou C. Mitochondrial biogenesis and development of respiratory chain enzymes in kidney cells: role of glucocorticoids. Am J Physiol Cell Physiol 267: C245–C254, 1994.
Djouadi F, Bastin J, Kelly DP, and Merlet Benichou C. Transcriptional regulation by glucocorticoids of mitochondrial oxidative enzyme genes in the developing rat kidney. Biochem J 315: 555–562, 1996.
Djouadi F, Wijkhuisen A, and Bastin J. Coordinate development of oxidative enzymes and NaKATPase in thick ascending limb: role of corticosteroids. Am J Physiol Renal Physiol 263: F237–F242, 1992.
Ecelbarger CA, Knepper MA, and Verbalis JG. Increased abundance of distal sodium transporters in rat kidney during vasopressin escape. J Am Soc Nephrol 12: 207–217, 2001.
GarciaPerez A and Burg MB. Role of organic osmolytes in adaptation of renal cells to high osmolality. J Membr Biol 119: 113, 1991.
Goruppi S, Bonventre JV, and Kyriakis JM. Signaling pathways and lateonset gene induction associated with renal mesangial cell hypertrophy. EMBO J 21: 5427–5436, 2002.
Greer KA and Hoying JB. Implementation of a statistical linear model for microarray data analysis. FASEB J 17: A78, 2003.
Ikeda M, Yoshitomi K, Imai M, and Kurokawa K. Cell Ca2+ response to luminal vasopressin in cortical collecting tubule principal cells. Kidney Int 45: 811–816, 1994.
Kerr MK, Martin M, and Churchill GA. Analysis of variance for gene expression microarray data. J Comput Biol 7: 819–837, 2000.
Knepper MA and Rector FC. Urinary concentration and dilution. In: The Kidney, edited by Brenner BM and Rector FC. Philadelphia, PA: Elsevier, 1995, p. 532–570.
Kojima R, Randall J, Brenner BM, and Gullans SR. Osmotic stress protein 94 (Osp94). A new member of the Hsp110/SSE gene subfamily. J Biol Chem 271: 12327–12332, 1996.
Ma T, Yang B, Gillespie A, Carlson EJ, Epstein CJ, and Verkman AS. Severely impaired urinary concentrating ability in transgenic mice lacking aquaporin1 water channels. J Biol Chem 273: 4296–4299, 1998.
Matsumori M, Itoh H, Toyoshima I, Komatsuda A, Sawada K, Fukuda J, Tanaka T, Okubo A, Kinouchi H, Mizoi K, Hama T, Suzuki A, Hamada F, Otaka M, Shoji Y, and Takada G. Characterization of the 105kDa molecular chaperone. Identification, biochemical properties, and localization. Eur J Biochem 269: 5632–5641, 2002.
Mootha VK, Bunkenborg J, Olsen JV, Hjerrild M, Wisniewski JR, Stahl E, Bolouri MS, Ray HN, Sihag S, Kamal M, Patterson N, Lander ES, and Mann M. Integrated analysis of protein composition, tissue diversity, and gene regulation in mouse mitochondria. Cell 115: 629–640, 2003.
Nielsen S, Pallone T, Smith BL, Christensen EI, Agre P, and Maunsbach AB. Aquaporin1 water channels in short and long loop descending thin limbs and in descending vasa recta in rat kidney. Am J Physiol Renal Physiol 268: F1023–F1037, 1995.
Nielsen S, Smith BL, Christensen EI, Knepper MA, and Agre P. CHIP28 water channels are localized in constitutively waterpermeable segments of the nephron. J Cell Biol 120: 371–383, 1993.
Pallone TL, Edwards A, Ma T, Silldorff EP, and Verkman AS. Requirement of aquaporin1 for NaCl-driven water transport across descending vasa recta. J Clin Invest 105: 215–222, 2000.
Pearce D. SGK1 regulation of epithelial sodium transport. Cell Physiol Biochem 13: 13–20, 2003.
Rozen S and Skaletsky HJ. Primer3 on the WWW for general users and for biologist programmers. In: Bioinformatics Methods and Protocols: Methods in Molecular Biology, edited by Krawetz S and Misener S. Totowa, NJ: Humana, 2000, p. 365–386.
Samuel SJ, Tzung SP, and Cohen SA. Hrp12, a novel heatresponsive, tissuespecific, phosphorylated protein isolated from mouse liver. Hepatology 25: 1213–1222, 1997.
Santos BC, Chevaile A, Kojima R, and Gullans SR. Characterization of the Hsp110/SSE gene family response to hyperosmolality and other stresses. Am J Physiol Renal Physiol 274: F1054–F1061, 1998.
Schnermann J, Chou CL, Ma T, Traynor T, Knepper MA, and Verkman AS. Defective proximal tubular fluid reabsorption in transgenic aquaporin1 null mice. Proc Natl Acad Sci USA 95: 9660–9664, 1998.
Smardo F, Burg M, and Garcia Perez A. Kidney aldose reductase gene transcriptionis osmotically regulated. Am J Physiol Cell Physiol 262: C776–C782, 1992.
Song J, Knepper MA, Verbalis JG, and Ecelbarger CA. Increased renal ENaC subunit and sodium transporter abundances in streptozotocininduced type 1 diabetes. Am J Physiol Renal Physiol 285: F1125–F1137, 2003.
Valcarce C, Navarrete RM, Encabo P, Loeches E, Satrustegui J, and Cuezva JM. Postnatal development of rat liver mitochondrial functions. The roles of protein synthesis and of adenine nucleotides. J Biol Chem 263: 77677775, 1988.
Vallon V, Verkman AS, and Schnermann J. Luminal hypotonicity in proximal tubules of aquaporin1knockout mice. Am J Physiol Renal Physiol 278: F1030–F1033, 2000.(Matthew R. McReynolds, Ka)