A variant of ASIC2 mediates sodium retention in nephrotic syndrome

Idiopathic nephrotic syndrome (INS) is characterized by proteinuria and renal sodium retention leading to edema. This sodium retention is usually attributed to epithelial sodium channel (ENaC) activation after plasma aldosterone increase. However, most nephrotic patients show normal aldosterone levels. Using a corticosteroid-clamped (CC) rat model of INS (CC-PAN), we showed that the observed electrogenic and amiloride-sensitive Na retention could not be attributed to ENaC. We then identified a truncated variant of acid-sensing ion channel 2b (ASIC2b) that induced sustained acid-stimulated sodium currents when coexpressed with ASIC2a. Interestingly, CC-PAN nephrotic ASIC2b-null rats did not develop sodium retention. We finally showed that the expression of the truncated ASIC2b in the kidney was dependent on the presence of albumin in the tubule lumen and activation of ERK in renal cells. Finally, the presence of ASIC2 mRNA was also detected in kidney biopsies from patients with INS but not in any of the patients with other renal diseases. We have therefore identified a variant of ASIC2b responsible for the renal Na retention in the pathological context of INS.

in association with ASIC1a and/or ASIC2a may participate in JNa+ in CCD and in sodium retention in CC-PAN rats. This hypothesis was confirmed by the finding that genetic deletion of ASIC2b (see Supplementary Data) abolished the stimulation of JNa+ in CCD and sodium retention in CC-PAN rats, and reduced by over half the volume of ascites without altering proteinuria ( Figure 2B-E). The residual volume of ascites observed in CC-PAN ASIC2b -/rats is likely accounted for by the increased permeability of peritoneal capillaries described in PAN rats (13).

Molecular characterization of a new variant of ASIC2b in CC-PAN rats.
ASICs are mainly expressed in the nervous system where their activation by extracellular acidification induces very brief cation currents that depolarize the cell membrane, allowing activation of nearby voltage-dependent channels or release of neuro-mediators (11, 14,15). Given the transiency of ASIC-driven currents, ASICs cannot sustain epithelial sodium reabsorption. Therefore, we searched for a variant of ASIC2b that could convert ASIC1a or ASIC2a into a channel carrying sustained sodium current. Starting from RNAs extracted from a CC-PAN rat kidney, we generated by 5'-RACE a cDNA the sequence of which was 100% identical to the deposited rat antibody specific for the t-ASIC2b protein, we used a pan anti-ASIC2 antibody directed against an epitope common to ASIC2a and ASIC2b (Abcam, Anti-ACCN1, ab77384). This antibody revealed a faint band ~80kDa (where ASIC2a was detected with the specific antibody) and a main band ~55kDa which was absent in the kidney of ASIC2b -/rats indicating the presence of a t-ASIC2b protein in CC-PAN rat kidneys. The intensity of this band was higher in CCDs from CC-PAN rats as compared with CC-controls ( Figure 4B).
Immuno-histological labeling of isolated CCD with the pan-ASIC2 antibody demonstrated luminal expression of ASIC2a/b in CC-PAN rats but neither in CC-control rats nor in CC-control ASIC2b -/or CC-PAN ASIC2b -/rats ( Figure 4C). ASIC2 labeling was detected in both AE1 positive cells (type A intercalated cells) and in AE1 negative cells (principal cells) as shown in Figure 4D.

Functional characterization of t-ASIC2b/ASIC2a channels in X. laevis oocyte.
We next evaluated whether t-ASIC2b is functional and may alter ASIC2a properties by expression in X.
laevis oocytes. Two-electrode voltage clamp experiments showed that expressing t-ASIC2b alone did not induce any acid-sensitive ion current in oocytes (not shown) whereas its coexpression with ASIC2a modified the acid-induced current mediated by the latter. Figure 5A shows representative traces obtained at a holding potential of -70mV in oocytes expressing ASIC2a alone or combinations of ASIC2a with either ASIC2b or t-ASIC2b. As previously described (16), decreasing the extracellular pH from 7.4 to 4.0 rapidly induced a transient current that spontaneously inactivated almost completely within seconds in ASIC2aexpressing oocytes. Co-expression of ASIC2a with the full length ASIC2b increased the initial (peak) current and decreased the level of inactivation, leading to an enlarged residual (plateau) current. These changes were markedly amplified by the co-expression of t-ASIC2b with ASIC2a. The ratio of the plateau over the peak currents was significantly higher in oocytes coexpressing ASIC2a and t-ASIC2b as compared with those expressing ASIC2a alone or in combination with ASIC2b ( Figure 5B). In oocytes co-expressing ASIC2a and t-ASIC2b, the residual desensitized current remained stable for >5 min (not shown). These findings indicate that co-expression of ASIC2a with a N-ter truncated form of ASIC2b induces the formation of functional channels that allow sustained transport of sodium in response to an acid stimulus.
In the following experiments, we focused our analysis on the residual current carried by the desensitized channel, the one which can account for sodium reabsorption in CCDs.
Within the range of pH tested (7.4 to 4.0), the pH sensitivity of the plateau current was similar in oocytes injected with ASIC2a alone or in combination with ASIC2b or t-ASIC2b ( Figure 5C). Truncated ASIC2b did not alter significantly the apparent low sensitivity of ASIC2a to amiloride ( Figure 5D). Co-expression of t-ASIC2b changed neither the sodium affinity nor the cation selectivity of ASIC2a (not shown).

Putative post-translational modifications of t-ASIC2b.
The comparison of the apparent molecular weight of native t-ASIC2b (around 55-60 kDa, see Figure 4B) and that of the transfected one in OKP or HEK cells (around 65 kDa Figure 3B) suggests a putative, specific post-translational modification in the kidney. To try to explain this discrepancy, we showed that the native t-ASIC2b was not sensitive to PNGase whereas the t-ASIC2b expressed in oocyte is glycosylated ( Figure 6A). The absence of glycosylation, obtained after directed mutagenesis of the two glycosylation sites (N416 and N443) did not modify the function of t-ASIC2b when expressed in oocytes, giving a similar amiloride-sensitive current as the WT t-ASIC2b ( Figure   6B-C).
Expression of ASIC2a/b in nephrotic patients. We next evaluated whether ASIC2a/b is also expressed in the kidney of patients with minimal change disease. RT-qPCR revealed the presence of ASIC2a/b mRNA in kidney biopsies from 5/8 patients with idiopathic nephrotic syndrome whereas it was undetectable in the remaining 3 (figure 7A). All patients with non glomerular renal diseases (11/11) displayed no or very low expression of ASIC2a/b mRNA.
Immunohistochemistry showed ASIC2a/b labeling in collecting ducts of some, but not all, nephrotic patients whereas labeling was never observed in non-nephrotic patients (figure 7B).

Signaling of ASIC2b induction during nephrotic syndrome.
It has been reported that albumin abnormally present in the kidney ultrafiltrate of nephrotic rats is endocytosed in CCDs, and that this process triggers several intracellular signaling cascades (17, 18). We therefore investigated the role of albumin in the induction of ASIC2b in CC-PAN rats. For this purpose, we used Nagase analbuminemic rats (NAR), a strain spontaneously lacking the albumin gene.
Despite analbuminemia, CC-NARs developed massive but slightly lower proteinuria than control rats in response to PAN ( Figure 8A). This proteinuria mainly consisted of proteins of higher molecular weight than albumin ( Figure 8B). In CC-NARs, PAN increased neither Asic2b mRNA expression in CCD nor sodium retention ( Figure 8C & 8D). The volume of ascites was reduced by half in NARs as compared to WT rats (in ml/100g body wt ± SE; WT: 7.3 ± 0.5, n=6; NAR: 3.8 ± 0.5, n=6: p<0.001). Interestingly, JNa+ in CCDs from nephrotic NARs was not increased compared with non-nephrotic NARs ( Figure 8E), indicating that albumin participates in the stimulation of ASIC2-dependent Na + reabsorption.
We next evaluated the possible involvement of ERK pathway in the induction of ASIC2b because we previously reported that this pathway is activated by albuminuria (18). By semiquantitative immunofluorescence, we confirmed that phospho-ERK labelling was increased in CCDs from CC-PAN rats as compared with CC-rats, and that this effect was abolished in NARs ( Figure 9A), indicating that ERK phosphorylation is induced by albumin, probably through its endocytosis. ERK phosphorylation was also prevented by in vivo treatment of CC-PAN rats with the mitogen-activated protein kinase kinase inhibitor U0126 ( Figure 9B). U0126 treatment also abolished the induction of Asic2b mRNA expression and the positivation of sodium balance ( Figure 10A-B), and reduced by over half the volume of ascites (in ml/100g body wt ± SE; Control: 7.3 ± 0.5, n=6; U0126: 3.0 ± 0.4, n=5: p<0.001). We also found that U0126 prevented the over-expression of the two subunits of Na,K-ATPase, the motor for sodium reabsorption in CCD ( Figure 10C-D). Unfortunately, we were not able to dissect native CCDs from U0126-treated rats to measure JNa+ by in vitro microperfusion in these rats.
Altogether, these results indicate that nephrotic albuminuria activates the ERK pathway and subsequently induces the expression of ASIC2b and sodium retention in CCDs. This mechanism is specific for albumin. This conclusion raises a question regarding the reversal of sodium retention in PAN rats. As a matter of fact, it has been shown than within 12 days following the administration of PAN, sodium balance and Na,K-ATPase activity in CCD return to basal levels and that ascites disappears despite the maintenance of massive proteinuria.
We confirmed that within 12 days sodium balance was restored to basal level and ascites was reduced while proteinuria remained high ( Figure 11A). We observed that at that time albumin was no longer accumulated in CCDs ( Figure 11B) and that ERK phosphorylation and Asic2b mRNAs had turned back to basal levels ( Figure 11C-D).

Discussion
Sodium reabsorption in the CCD proceeds via two pathways: the classical electrogenic pathway mediated by amiloride-sensitive ENaC on the apical side and basolateral Na,K-ATPase, and an electroneutral, thiazide-sensitive pathway energized by the basolateral H-ATPase and requiring the concerted activity of apical sodium dependent-and -independent chloride/bicarbonate exchangers and of a basolateral sodium-bicarbonate cotransporter (10, 19). These two pathways originate from principal cells and B-type intercalated cells respectively. In contrast, A-type intercalated can secrete sodium via basolateral Na/K/Cl cotransporter and apical H(Na),K-ATPase (20, 21). Here we show that sodium reabsorption in CCD from CC-PAN rats is electrogenic and amiloride-sensitive but independent of any ENaC subunit, since none of them is expressed at the apical cell border (6), revealing the existence of a third pathway for sodium reabsorption. This pathway involves a newly characterized truncated variant of ASIC2b which, in association with ASIC2a, can carry sustained apical sodium entry. This channel is expressed at the apical border of all cell types constituting CCDs suggesting that they might all participate in sodium reabsorption. However, this appears unlikely for A-type intercalated cells because they do not express any known sodium pump at their basolateral side.
Based on co-immunoprecipitation and functional studies, Ugawa and coworkers have shown that rat ASIC2a and ASIC2b biochemically assemble to constitute functional (16) channels. The variant of ASIC2b identified in this study differs from ASIC2b by the deletion of most of its intracellular N-terminal domain (71/88 amino acid residues). The N-terminal domain of ASICs is not necessary for subunit association since the trimeric structure of chicken ASIC1 was deduced from crystal structures obtained from proteins with truncated intracellular domains (22, 23). This indicates that ASIC2a and truncated ASIC2b can assemble to constitute functional channels likely made of three subunits like all functional ASICs. However, the stoichiometry of assembly remains unknown and may be variable in vivo as well as in our coexpression studies. Consequently, the macroscopic currents measured in X. laevis oocyes may stem from a mix of three types of trimeric channels containing 0, 1 or 2 subunits of t-ASIC2b.
Current kinetics models propose that ASICs exist under three functional states: a closed, an open and a desensitized state in which the channel is partially open but cannot be activated by acid stimulus. Following acid stimulus, the open and desensitized states translate into peak and plateau currents respectively. It was previously reported that the main effect of ASIC2b is to increase the current carried by desensitized ASIC2a (16). We confirmed this effect of ASIC2b and found that it is amplified by t-ASIC2b. Thus, the kinetics of the inward current mediated by ASIC2a/t-ASIC2b resembles that of ENaC-mediated current, except for the presence of a transient peak of small amplitude as regards to the remaining plateau.
Another important issue concerns the intensity of the current carried by ASIC2a/t-ASIC2b compared with ENaC-mediated current. Formally, this question cannot be answered by electrophysiological analysis in X. laevis oocyte since the current intensity depends on the expression level of the respective channels and possibly on the stoichiometry of t-ASIC2b/ASIC2a assembly. Nonetheless, it is worth mentioning that the macroscopic Na + current measured in ASIC2a and t-ASIC2b-expressing oocytes under the desensitized state (~3µA at a holding potential of -70mV) is of the same order of magnitude as that initially reported in ENaC-expressing oocytes (~1µA at -100mV) (24). Altogether our findings indicate that, given their conductive properties under desensitized state, hetero-trimers made of ASIC2a and t-ASIC2b may substitute for ENaC and allow sustained sodium reabsorption in CCD principal cells.
Conversely to ENaC which opens stochastically in absence of stimulus, ASIC2 requires an acid stimulus to open and to desensitize. Thus, the increased abundance of ASIC2 subunits in CCDs from CC-PAN rats is not sufficient to account for sodium reabsorption; this also requires the presence of an ASIC2 activating factor. ASIC2a requires low pH for maximal activation (pH50 ~4.0, maximal current, pH~2.0) (16), and association with t-ASIC2b did not modify this pH dependency. The pH of the luminal fluid prevailing in CCDs in vivo is estimated in the 6.0-6.5 range (25, 26) but, because type A intercalated cells of the CCD secrete protons and due to the presence of unstirred layers, the pH at the external surface of the apical membrane might be 0.5-1.0 units lower, i.e. in the 5.0-6.0 range. Based on X. laevis expression experiments shown in Fig 5C, these pH would induce 20-40% of the current carried by desensitized ASIC2a/t-ASIC2b (Iplateau) at pH 4.0, which may be sufficient to account for the rate of sodium reabsorption determined by in vitro microperfusion. In addition, one cannot exclude that factors abnormally present in the urine of nephrotic rats or produced by CCD cells may also increase the pH sensitivity of ASIC2a/t-ASIC2b or activate it independently of pH.
Such mechanisms have been reported with other ASICs, e.g. NO and arachidonic acid potentiate ASIC-mediated proton-gated currents (27, 28) and the arginine metabolites agmatine and arcaine activate ASIC in a proton-independent manner (29, 30). Moderate proteolysis of ASIC2 may also activate it, as previously reported for ENaC and plasmin (31) , the excretion of which is increased during nephrotic syndrome (32).
The apparent low amiloride-sensitivity of t-ASIC2b/ASIC2a observed in X. laevis oocytes (IC50≈50 µM) contrasts with the high sensitivity of sodium transport observed in vitro in microperfused CCDs (full inhibition with 10 µM amiloride). These differences in amiloride sensitivity likely stem from the fact that acid pH required to activate ASIC2 in X. laevis oocytes strongly decreases its sensitivity to amiloride (16). This further support the notion that acid pH is probably not the unique activating factor of ASIC2 in ASDN during nephrotic syndrome.
The mechanism of sodium retention in nephrotic rats varies according to their aldosterone status: when animals display high plasma aldosterone level, increased sodium reabsorption is mediated by the classical ENaC-dependent pathway and there is no evidence for ASIC2 expression, whereas blunting of hyper aldosteronemia switches off the ENaC pathway and triggers the ASIC2-dependent one. This suggests the existence of a balance between factors that reciprocally trigger and repress the renal expression of ENaC and ASIC2.
Aldosterone is the major inducer of ENaC expression in the ASDN; whether or not it represses the expression of ASIC2 remains to be established. Here we report that the phosphorylation of ERK brought about by the endocytosis of albumin mediates the over expression of ASIC2 in the ASDN and the stimulation of sodium transport. In contrast, ERK phosphorylation reduces ENaC activity by different mechanisms including the decrease of its open probability, increasing its membrane retrieval (33) and decreasing expression of its mRNA expression (34). Thus, endocytosis of albumin and subsequent activation of ERK pathway appears as a major factor that switches on and off ASIC2 and ENaC pathways respectively. This raises the question of the mechanism responsible for the escape of sodium retention to albuminuria observed in the long term. Our data show that it proceeds at the step of albumin endocytosis, possibly via the down regulation of the albumin receptor 24p3R (17).
Our preliminary results in nephrotic patients indicate that expression of ASIC2 in the ASDN is not restricted to the PAN rat model. The fact that only part of the patients studied displayed ASIC2 expression in their ASDN might be related to the fact that, in nephrotic rats, ASIC2 is only found in corticosteroid clamped animals and that only a fraction of nephrotic patients displays normal aldosterone status (7). A further study will be necessary to evaluate whether expression of Asic2 in nephrotic patients is correlated with low plasma aldosterone level. More generally, it is noteworthy that the alternate ATG initiating codon found in the rat Asic2b sequence is conserved not only in the human sequence (NM_183377.1) but also in mice (NM_007384.3).
It has been reported previously that channels made of aENaC and ASIC1a constitute functional cation reabsorbing channels that participate in fluid clearance by lungs (35). Here we describe for the first time the functional expression and role of channels exclusively made of ASIC subunits out of excitable cells. It also shows that deletion of the intracellular N-terminal domain of ASIC2b modifies its properties and allows converting the transient ASIC2a into a long lasting epithelial channel. Thus, ASICs and ENaC share more than sequence similarities since both can perform epithelial sodium transport. Whether variants of ENaC might work as transient sodium channels is a stimulating hypothesis.

Animals.
Experiments were carried out on male rats (150-170 g at the onset of the experimentation) fed a standard laboratory chow (A04, Safe, Augy, France) with free access to water. Sprague-Dawley rats were from Charles Rivers (L'Abresles, France) and Nagase analbuminemic rats (NARs) were from Japan SLC (Shizuoka, Japan). For surgery, animals were anaesthetized by intraperitoneal injection of a mix including Domitor (Pfizer, 0.5 μg/g body wt), Climasol (Graeub, 2 μg/g body wt) and Fentanyl Jansen (Janssen Cilag Lab, 5 ng /g body wt). Animals were awakened by a subcutaneous injection of a mix containing Antisedan (Pfizer, 750 ng/g body wt), Sarmasol (Graeub, 200 ng/g body wt) and Narcan (Aguettant, 133 ng/g body wt). Before killing, animals were anaesthetized with pentobarbital (Sanofi, France, 50 mg/kg body wt, I.P.). Corticosteroid clamp was achieved by bilateral adrenalectomy and supplementation with aldosterone (10 µg/kg/day) and dexamethasone (14 µg/kg/day) through subcutaneous osmotic pump (ALZET, Charles River) (6) Nephrotic syndrome was induced the day after surgery for corticosteroid clamp by a single intra-jugular injection of aminonucleoside puromycine (PAN) (Sigma-Aldrich, 150 mg/kg body wt). Control rats received a single injection of isotonic NaCl (1ml /100 g body wt). A group of rats was treated with the ERK kinase inhibitor U0126 by daily subcutaneous injection (3mg/10 g body wt/day in a mixture of DMSO/sesame oil, 16%/84% vol/vol). Treatment started the day of corticosteroid clamp. Animals were studied 6 days after vehicle or PAN injection, at the time of maximum of sodium retention and proteinuria, or after 12 days when sodium balance was restored (36). To induce ENaC expression, rats were fed a Na + -depleted diet (Safe, synthetic diet containing 0.11g Na + /kg instead of 2.5g/kg). Rats were studied 14 days after the onset of the low Na + diet.

Generation of ASIC2b null rats.
Invalidation of Accn1 gene encoding the ASIC2 proteins was performed using Crispr-Cas9 technology (TRIP, INSERM UMR1064, Nantes, France). Zygotes from Sprague-Dawley rats were microinjected with a single guide RNA (sgRNA, 10ng/ µl) designed to target exon 1 of the Accn1 gene and Cas9 mRNA (50ng/µl) as previously described (37).This resulted in a deletion of 20 nt and a frameshift of the coding region leading to the early appearance of a premature stop codon. Embryos were then implanted in pseudopregnant females and grown until full term. To genotype the animals, a 953 pb genomic DNA fragment around the targeted region was PCR to detect the deleted fragment in KO animals by gel electrophoresis.
Urine proteinogram. Urine samples were thawed and centrifuged 10 min at 3300g. 9 µl of supernatant (pure for control rats, 1/10 diluted in H2O for PAN rats) were half diluted in Glycerol/blue solution and separated by SDS-PAGE. After migration gels were rinsed twice with water and stained overnight in Coomassie blue solution (Pageblue protein staining, Life Technologies) at room temperature. The gels were then rinsed for at least 48h in water.  (7), with bath and perfusate containing (in mM): 138 NaCl, 1.2 MgSO4, 2 K2HPO4, 2 calcium lactate, 1 sodium citrate, 5.5 glucose, 5 alanine, 12 creatinine (bath continuously gassed with 95% O2/5% CO2). pH was adjusted either to 7.4 with Hepes 10 mM and Tris 5 mM or to 6.0 with MES 5 mM. Transepithelial voltage (PDte) was measured via microelectrodes connected through an Ag/AgCl half-cell to an electrometer. Tubules were perfused at low rate (~2 nl/min). Concentrations of Na + , and creatinine were determined by HPLC, and ion flux (JNa + ) was calculated as:

Microdissection of cortical collecting ducts (CCDs
where [Na + ]p and Na + ]c are the concentration of Na + in the perfusate and collection, respectively, Vp and Vc are the perfusion and collection rates, respectively, L is the tubule length, and t is the collection time. For each tubule, Na + flux was calculated as the mean of four successive 10-15min collection periods. Vp was calculated as: where [creat]c and [creat]p are the concentrations of creatinine in the collection and perfusate, respectively.

RNA extraction and RT-qPCR.
RNAs were extracted from 40-50 CCDs using an RNeasy micro kit (Qiagen, Hilden, Germany) as previously described (18). Frozen human kidney biopsies (2-3mg) were dissolved in 5µl of RLT complemented with b-mercaptoethanol, transferred in a tube and RNAs were extracted as above. RNAs were reverse transcribed using first strand cDNA synthesis kit for RT-PCR (Roche Diagnostics, Meylan, France), according to the manufacturers' protocols. Real time PCR was performed with a LightCycler 480 SYBR Green I master quantitative PCR kit (Roche Diagnostics) according to the manufacturer's protocol, except that reaction volume was reduced 2.5-fold. Specific primers (Table 1) were designed using Probe Design (Roche Diagnostics).
In each experiment, a standardization curve was made using serial dilutions of a standard cDNA stock solution made from rat or human kidney, and data (in arbitrary unit) were calculated as function of the standard curve. For CCDs, results were normalized as a function of Rps23 expression. Data are means ± SE from several animals. Given the heterogeneity of human kidney biopsies in term of cell composition, data were normalized using the distal nephron markers aquaporin 2 (AQP2) and FXYD4, as previously described (38). Immunoblotting. Pools of 30-80 CCDs were solubilized in Laemmli buffer, and proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (GE Healthcare). After blocking, blots were successively incubated with specific anti-ASIC2a antibody (Alomone labs, ASC-012, 1/1000, see Table 2) or pan-ASIC2 antibody (Abcam, ab77384, 1/700) and anti-rabbit IgG antibody coupled to horseradish peroxidase (Promega Two serial sections of human kidney biopsies were used for ASIC2 and aquaporin 2 (AQP2, a marker of the distal nephron) labeling respectively. ASIC labeling was realized as for isolated CCDs. For AQP2, after blocking, slides were incubated with and anti-AQP2 antibody (Santa Cruz, sc709882, 1/500) for 1 h at room temperature, rinsed and incubated 1 h at room temperature with TRITC-conjugated anti rabbit IgG (1/500).
Albumin immunostaining. 8µm rat kidney sections fixed with 4% paraformaldehyde and included in OCT were transferred to Superfrost Gold Glass slides, rinsed with PBS, incubated for 20 min at room temperature in 100mM glycine, permeabilized for 30s with 0.1% triton and incubated 5 min in SDS 1% for antigen retrieval. After blocking, slides were incubated with anti-AE1 antibody (gift from C. Wagner, Institute of Physiology, Zurich, Switzerland, 1/500) for 1h at room temperature, rinsed and incubated with TRITC-coupled anti-rabbit IgG (1/500) for 1h at RT. After rinsing, anti-albumin antibody FITC-coupled was added (DAKO F0117, 1/500).

Quantification of phospho-ERK. 5µm cryo-sections of rat kidney fixed with 4%
paraformaldehyde were transferred to Superfrost Gold Glass slides, rinsed with PBS at 4°C and incubated for 10 min at 94°C in citrate buffer pH 6.0. After blocking with donkey serum (10%, for 30 min at room temperature) slides were incubated overnight at 4°C with anti-AQP2 antibody (Santa Cruz, sc-9882, 1/400) and anti-P-p44/42 MAPK (T202/Y204 antibody (Cell Signaling Technology, 9101S) diluted at 1/400 each in PBS containing 5% milk and 0.01%     1 cctcgggctgaatga atg agc cgg agc ggc gga gcc cgg ctg ccc gcg acc gcg ctc 58 agc ggc ccg gga cgc ttc cgt atg gcc cgc gag cag ccg gcg ccc gtg gcg gtg gcg 115 gca gct agg cag ccc gga gga gac cgg agc ggc gat ccg gcg ctg cag ggg cca 169 ggg gtc gcc cgc agg ggg cgg ccg tcc ctg agt cgc act aaa ttg cac ggg ctg 223 cgg cac atg tgc gcg ggg cgc acg gcg gcg gga ggc tctttc cag cga ……..  Comparison between groups was performed by 2-tailed unpaired t test. A p value less than 0.05 was considered significant. Comparison between groups was performed by variance analysis (one-way ANOVA) followed  -