Hormonal control of the renal immune response and antibacterial host defense by arginine vasopressin

The identification of collecting duct cells as a preferential adhesion site for UPECs in an in vivo model of ascending UTI (6) led us to analyze the mechanisms of cell activation in a highly differentiated collecting duct cell line, mpkCCD_cl4, derived from collecting ducts microdissected from the kidney of a transgenic SVPK-Tag mouse harboring the large T antigen under the control of a truncated pyruvate kinase promoter fused to an SV40 enhancer (21). These cells formed confluent layers of epithelial-shaped cells, which in turn formed domes, a characteristic feature of layers of ion-transporting epithelial cells (Fig. S1). mpkCCD_cl4 cells expressed TLR4 as well as the accessory protein MD-2 and myeloid differentiation factor 88 (MyD88; Fig. S1). They also expressed the mRNA of CD14 (Fig. S1), indicating that mpkCCD_cl4 cells might autonomously produce CD14, rendering them independent of serum-derived soluble CD14. Western blotting using a specific polyclonal rabbit anti-TLR4 antiserum revealed a major protein band of the predicted size of ∼96 kD, which was not detected in the presence of an excess of the peptide used for immunization (Fig. 1 A). FACS analysis revealed predominantly intracellular staining of TLR4 (Fig. 1 B), and immunofluorescence studies showed that TLR4 was mainly concentrated in the perinuclear region of the cytoplasm (Fig. 1 C, top), where it colocalized with the Golgi marker CTR433 (Fig. 1 C, bottom) (24). These findings indicate that mpkCCD_cl4 cells express the LPS receptor complex TLR4–MD-2, which is mainly detected in the intracellular compartment (i.e., in the Golgi apparatus), as previously reported in pulmonary epithelial cells (25), endothelial cells (26), and intestinal epithelial m-IC_cl2 cells (27).

The recognition of LPS by TLR4 induces a cascade of events leading to activation of the nuclear transcription factor NF-κB signaling pathway and the subsequent production of chemokines and cytokines (28). IκB-α undergoes ubiquitination and proteasome-mediated degradation, leading to the release and nuclear translocation of the NF-κB subunit p65/RelA (29, 30). In mpkCCD_cl4 cells, LPS induced the nuclear translocation of NF-κB (Fig. 1 D, top). This nuclear redistribution was completely abolished after treatment with the specific NF-κB inhibitor SN50. LPS also induced a dose-dependent increase in luciferase activity in cells transiently transfected with a NF-κB luciferase reporter gene construct (Fig. 1 D, bottom). 50 ng/ml TNF-α, used as positive control, similarly induced a significant increase in luciferase activity. Finally, LPS induced dose- and time-dependent stimulation of the secretion of the chemokine macrophage inflammatory protein 2 (MIP-2; Fig. 1 E), which was very significantly reduced in the presence of SN50 or polymyxin B, a competitive inhibitor of LPS (not depicted). These results indicate that mpkCCD_cl4 cells provide a suitable in vitro cell system for analyzing regulatory mechanisms controlling TLR4-mediated cellular activation in renal collecting duct cells.

Previous studies had demonstrated that AVP induces a rapid rise in cellular cAMP content and stimulates the reabsorption of NaCl and water by renal collecting duct cells (17–20). We investigated whether dDAVP could affect innate immune recognition of LPS in mpkCCD_cl4 epithelial cells. Incubating cells with 10⁻⁸ M dDAVP alone or with LPS for 6 h induced a significant rise in cAMP as compared with untreated cells (Fig. 2 A). LPS induced rapid time-dependent degradation of IκB-α and stimulated the phosphorylation of the NF-κB subunit p65/RelA (phosphorylated p65; Fig. 2 B). The time-dependent degradation of IκB-α and phosphorylation of the NF-κB subunit p65/RelA (phosphorylated p65) caused by LPS was no longer observed in the presence of dDAVP (Fig. 2 B). Experiments using mpkCCD_cl4 cells confirmed that dDAVP induced a dose-dependent decrease of the LPS-induced activation of NF-κB (Fig. 2 C). Consistent with these findings, dDAVP caused dose-dependent inhibition of LPS-stimulated secretion of MIP-2 and TNF-α (Fig. 2 D). A constant concentration of 10⁻⁸ M dDAVP also transactivated NF-κB and stimulated the secretion of MIP-2, and TNF-α inhibited to the same extent at all concentrations of LPS tested (Fig. 2, E and F), indicating that dDAVP does not bind to LPS in solution. To confirm that the inhibitory action of dDAVP on LPS-mediated cell activation involved the binding of dDAVP to V2R, the production of MIP-2 was measured in mpkCCD_cl4 cells preincubated with increasing concentrations of SR121463B (see Materials and methods), a selective, nonpeptide V2R antagonist (31). Stimulations were subsequently performed with 10 ng/ml LPS plus 10⁻⁸ M dDAVP. SR121463B produced concentration-dependent antagonism of the inhibitory action of dDAVP on LPS-induced MIP-2 secretion: at 10⁻⁷ M, SR121463B completely antagonized the inhibitory action of dDAVP and restored the stimulatory activity of LPS (Fig. 2 G). The V2R-mediated inhibitory action of dDAVP appeared to be specific because neither dDAVP alone nor dDAVP plus SR121463B affected PMA-induced cell stimulation (Fig. 2 H). We then checked that dDAVP was acting specifically on the TLR4-mediated cellular activation. Extinction of TLR4 mRNA expression by a specific TLR4 small-interfering RNA (siRNA) completely inhibited the secretion of MIP-2 stimulated by LPS in mpkCCD_cl4 cells, whereas a negative control siRNA had no effect (Fig. S2). As a control, TLR4 extinction induced by the siRNA used had no effect on PMA stimulation. Involvement of PKA in the inhibitory action of dDAVP on LPS activation was demonstrated by preincubating the cells for 30 min with 5 × 10⁻⁵ M of the selective PKA inhibitor H89. H89 significantly reduced the inhibitory effect of dDAVP on the LPS-mediated secretion of MIP-2 (LPS + dDAVP = 429 ± 31 pg/ml; LPS + dDAVP + H89 = 616 ± 11 pg/ml; n = 4; P < 0.05). In contrast, preincubating cells with 10⁻⁶ M GF109203X, a protein kinase C inhibitor, did not affect the inhibitory action of dDAVP on the LPS-mediated secretion of MIP-2 (LPS + dDAVP = 495 ± 17 pg/ml; LPS + dDAVP + GF109203X = 503 ± 18 pg/ml; n = 4). Collectively, these findings indicate that dDAVP acts as a potent modulator of the TLR4-dependent activation of renal CCD cells.

dDAVP-induced down-regulation of LPS-mediated NF-κB activation suggests that this polypeptide hormone may alter the activation (i.e., phosphorylation) of signaling molecules. The phosphorylation of signaling molecules is regulated not only by protein kinases, but also by Ser/Thr protein phosphatases (32). PP1 and PP2A, as well as a variety of protein kinases, are involved in the regulation of signaling pathways by a phosphorylation/dephosphorylation mechanism (32). Interestingly, Ser/Thr PP2A is highly expressed in the distal nephron (33) and is activated by dDAVP in mouse renal collecting duct cells (34). We therefore investigated whether dDAVP activated Ser/Thr protein phosphatase activity, and whether calyculin A and okadaic acid, two potent inhibitors of Ser/Thr PP1 and PP2A (35), impaired the inhibitory effect of dDAVP on LPS-mediated cell stimulation. Incubating mpkCCD_cl4 cells with 10⁻⁸ M dDAVP for 6 h significantly stimulated PP2A activity (Fig. 3 A). 10 ng/ml LPS for 6 h also slightly increased the PP2A activity stimulated by dDAVP (Fig. 3 A). In the absence of dDAVP, preincubating the cells with 10 nM calyculin A or okadaic acid had no effect on LPS-stimulated secretion of MIP-2 or TNF-α (Fig. 3 B). In sharp contrast, the two protein phosphatase inhibitors prevented the inhibitory effect of dDAVP on the LPS-stimulated secretion of MIP-2 and TNF-α (Fig. 3 B). Although calyculin A and okadaic acid both inhibit PP1 and PP2A activities, the latter is a more selective inhibitor of PP2A at the 10-nM dose used (36, 37). 10 nM calyculin A and okadaic acid also impaired the inhibitory effect of 10⁻⁸ M dDAVP on the stimulated secretion of MIP-2 and TNF-α at all concentrations of LPS (1–1,000 ng/ml) tested (Fig. S3). These findings strongly suggest that the increase in PP2A activity is directly responsible for the inhibitory effect of dDAVP on the LPS-stimulated secretion of both MIP-2 and TNF-α by renal collecting duct cells.

CFTR, a Cl⁻ channel activated by cAMP-dependent PKA, can be dephosphorylated and inactivated by several Ser/Thr protein phosphatases (38–41). CFTR mediates the electrogenic apical secretion of Cl⁻ stimulated by dDAVP in renal collecting duct cells (12, 13). mpkCCD_cl4 cells express both ENaC and the CFTR Cl⁻ channel (Fig. 4 A). Two recent studies have demonstrated that the CFTR Cl⁻ channel is directly linked to PP2A (42, 43). Inhibitors of PP2A were also shown to prolong the deactivation of cAMP-activated CFTR Cl⁻ channels in colonic carcinoma Caco-2 cells (42) and to increase CFTR channel activity in excised patches of airway and intestinal epithelium (43). Although protein phosphatase inhibitors had no substantial effects on dDAVP-stimulated Cl⁻ fluxes in mpkCCD_cl4 cells (unpublished data), we wanted to find out whether CFTR was involved in the observed PP2A-dependent inhibitory effect of dDAVP on LPS-mediated cell activation. We therefore tested the effects of various Cl⁻ channel inhibitors on mpkCCD_cl4 cells exposed to LPS alone or to LPS plus dDAVP for 6 h (Fig. 4 B). 5-nitro-2(3phenylpropyl-amino)benzoate (NPPB) and glybenclamide (Gly) both inhibit mouse CFTR at high concentrations and also affect other Cl⁻ transporters, as well as K⁺-ATP–sensitive channels (13, 44–46). The more potent and specific CFTR inhibitor CFTR_inh-172 (see Materials and methods) was therefore also included in the analysis (47). 10⁻⁴ M NPPB and Gly significantly impaired the inhibitory activity of dDAVP on LPS-mediated secretions of MIP-2 and TNF-α, whereas 10⁻⁵ M amiloride, a potent inhibitor of ENaC, did not affect cytokine secretion (Fig. 4 B). Importantly, 10⁻⁷ M CFTR_inh-172 also significantly diminished the inhibitory effect of dDAVP on the LPS-mediated secretion of MIP-2 and TNF-α (Fig. 4 B). Similar to what has been reported for calyculin A and okadaic acid, 10⁻⁷ M CFTR_inh-172 completely impaired the inhibitory effect of 10⁻⁸ M dDAVP on the stimulation of the secretion of MIP-2 and TNF-α at all concentrations of LPS (1–1,000 ng/ml) tested (Fig. S3).

Confluent mpkCCD_cl4 cells exhibit a highly differentiated, tight epithelial phenotype when grown on semipermeable filters and develop high transepithelial electrical resistance (>1,500 Ω.cm²) (21). The polarized (apical and/or basal) secretion of MIP-2 was measured in the apical and basal medium bathing confluent mpkCCD_cl4 cells grown on filters after being exposed to 10 ng/ml LPS for 6 h. Apical addition of LPS resulted in an increase in predominantly apically directed secretion of MIP-2 (Fig. 4 C). In contrast, basal addition of LPS induced only a moderate increase in the secretion of MIP-2, which appeared to be predominantly basally directed (Fig. 4 C). These findings demonstrate the presence of a predominantly apically oriented mechanism of LPS recognition and chemokine secretion in mpkCCD_cl4 cells. Similarly, apical addition of 10⁻⁷ M CFTR_inh-172 restored the LPS-stimulated apical and basal secretion of MIP-2 and TNF-α after the apical or basal addition of LPS to confluent cultures of dDAVP-treated mpkCCD_cl4 cells grown on filters (Fig. 4 C). These findings indicate that the cAMP-regulated CFTR Cl⁻ channel, together with PP1 and PP2A, is involved in regulating the bipolarized cytokine secretion induced by LPS in renal collecting duct cells.

To assess further the role of CFTR in the down-regulation of the LPS-induced inflammatory responses elicited by dDAVP, experiments were performed using primary cultures of CCDs dissected from cftr^m1unc and wild-type counterpart mice (hereafter referred to as cftr^+/+ and cftr^−/− CCDs, respectively). Cultured cftr^+/+ and cftr^−/− CCDs formed confluent layers of cuboid-shaped cells expressing K₈–K₁₈ cytokeratins and the tight junction–associated protein ZO-1 (Fig. 5 A). These cell layers were devoid of hematopoietic cells and did not express the bone marrow–derived cell marker CD45 (Fig. 5 B). cftr^+/+ and cftr^−/− CCDs both expressed TLR4, whereas the expected size of CFTR transcripts was only observed in cftr^+/+ and not in cftr^−/− CCDs (Fig. 5 C). dDAVP impaired the LPS-induced activation of the phosphorylated NF-κB subunit p65/RelA in cftr^+/+ CCDs (Fig. 5, D and E). As in mpkCCD_cl4 cells, the decrease in phosphorylated p65 caused by dDAVP was no longer detected in LPS-treated wild-type cftr^+/+ CCDs incubated with 10⁻⁷ M CFTR_inh-172, 10 nM okadaic acid, or calyculin A (Fig. 5 D). In sharp contrast, cftr^−/− CCDs were completely resistant to the effect of dDAVP (Fig. 5, D and E). In line with these findings, dDAVP significantly reduced the LPS-induced secretion of MIP-2 and TNF-α in cftr^+/+ but not in cftr^−/− CCDs (Fig. 5 F). Moreover, calyculin A, okadaic acid, and CFTR_inh-172 had no effect on LPS-treated cftr^−/− CCDs (Fig. 5 F). These findings indicate that the inhibitory activity of dDAVP requires functional CFTR to control the activity of the Ser/Thr protein phosphatases in renal collecting duct cells.

To study the dDAVP-mediated effect on LPS-induced cell activation in a biologically relevant context, confluent cultures of CCDs dissected from TLR4-defective C3H/HeJ mice and from TLR4-expressing C3H/HeN mice (hereafter referred to as Lps^d and Lpsⁿ CCDs, respectively) were analyzed after exposure to UPECs (6). dDAVP inhibited the secretion of MIP-2 and TNF-α in Lpsⁿ CCDs after the addition of LPS or HT7 UPEC isolates (Fig. 6, A and B). In contrast, no detectable LPS or dDAVP-induced effect was noted in Lps^d CCDs (Fig. 6 A). In line with a previous study, which had demonstrated that UPEC isolates also activate a TLR4-independent signaling pathway (6), we found that incubating Lps^d CCDs with UPECs resulted in residual cell stimulation. However, the levels of MIP-2 and TNF-α secreted by Lps^d CCDs reached only ∼50% of those for Lpsⁿ CCDs. Importantly, TLR4-independent cell activation was not affected by preincubating with dDAVP (Fig. 6 B). These findings demonstrate that the suppressor effect of dDAVP is limited to TLR4-mediated cell activation and suggest that it may well be of biological relevance during UPEC infection.

Increased AVP plasma levels are found in a variety of pathophysiological situations associated with dehydration, such as severe renal infection or septic shock. The question arises of the extent that dDAVP influences renal inflammatory responses and bacterial clearance. To answer this question, either 1 ng/μl/h dDAVP or 1 μl/h isotonic saline was administered continuously to adult female Lpsⁿ and Lps^d mice via an implanted osmotic pump while receiving normal water intake (see Materials and methods) (48). Untreated Lpsⁿ and Lps^d mice were kept in metabolic cages before and after the implantation of the osmotic pumps filled with dDAVP to find out whether dDAVP had similar antidiuretic effects in both mouse strains. dDAVP administration induced similar reduction in urine volume in both mouse strains (Table S1). In accordance with these results, urine osmolality increased and plasma osmolality decreased in all HT7-infected mice treated with dDAVP but not in those given isotonic saline (Table S2). Moreover, the magnitude of changes in plasma and urinary osmolalities reflecting the hydroosmotic activity of dDAVP were very similar in the Lpsⁿ and Lps^d mice (Table S2). 24 h after the transurethral inoculation of the HT7 isolates, the expression of proinflammatory mediators and the renal bacterial burden were analyzed. The levels of IL-1β, MIP-2, regulated on activation, normal T cell expressed and secreted (RANTES), MCP-1, and TNF-α expression were lower in the dDAVP-treated Lpsⁿ mice than in mice infused with isotonic saline after bacterial infection (Fig. 7 A). UPEC infection induced only a moderate increase in proinflammatory mediator expression in the kidneys of Lps^d mice, which was not altered by the dDAVP treatment (Fig. 7 A). dDAVP also dampened the production of MIP-2 and TNF-α measured in whole-cell homogenates from HT7-infected Lpsⁿ mice but had no effect on the production of MIP-2 and TNF-α by HT7-infected kidneys of Lps^d mice (Fig. 7 B). These data are in accordance with the in vitro observations and further confirm that the suppressor effect of dDAVP is restricted to TLR4-dependent immune activation. The importance of TLR4 activation during UPEC infection was demonstrated by the finding that the renal bacterial burden was markedly greater in Lps^d than in Lpsⁿ wild-type mice. Importantly, Lpsⁿ mice treated with dDAVP were significantly more susceptible to UPEC infection than their Lpsⁿ counterparts receiving isotonic saline, with a 20-fold greater bacterial CFU 1d after infection (Fig. 7 C). In contrast, dDAVP had no effect on the renal bacterial counts in Lps^d mice (Fig. 7 C).

Chemoattractant chemokines, such as MIP-2, play a key role in the migration of PMNs to mucosal sites of inflammation to produce efficient bacterial clearance (49, 50). In accordance with the lower expression of chemokines in infected Lpsⁿ mouse kidneys treated with dDAVP, the number of Ly6-G–positive PMNs found on kidney sections from infected mice 1d after bacterial challenge was significantly lower in dDAVP-treated Lpsⁿ mouse kidneys than in the saline-infused Lpsⁿ control mice (Fig. 7, D and E). In sharp contrast, the number of PMNs in UPEC-infected Lps^d kidneys, which was significantly lower than that in infected Lpsⁿ mouse kidneys, was not affected by the dDAVP treatment (Fig. 7, D and E). The distribution of neutrophils in the pyelic and inner medullary regions of infected kidneys appears to be rather variable (Fig. 7 D). To confirm the immunohistological findings, neutrophil migration was also assessed by measuring myeloperoxydase (MPO) activity in renal homogenates (51). MPO assay revealed that neutrophil migration was lower in Lps^d than in Lpsⁿ kidneys colonized by HT7 (Fig. 7 F). dDAVP infusion induced a marked decrease in MPO activity in the kidneys from HT7-infected Lpsⁿ mice as compared with that of saline-infused infected Lpsⁿ mice but had no effect on MPO activity in HT7-infected Lps^d kidneys (Fig. 7 F). These findings demonstrate that dDAVP also influences inflammatory responses in vivo and suggest that dDAVP may be involved during bacterial kidney infection. They also raised the question as to whether in vivo blockade of the endogenous V2R, which is highly expressed in collecting duct cells, could prevent the deleterious effect of dDAVP on the inflammatory response and bacterial burden. To answer this question, C3H/HeN Lpsⁿ mice subjected to water restriction were given intraperitoneal injections of 0.03 or 0.3 mg/kg SR121463B 6 h before the transurethral inoculation of HT7. The mice received a second injection of SR121463B 14 h after bacterial infection. Administration of a low or high concentration of SR121463B alone had no effect on the levels of mRNA expression of proinflammatory mediators (Fig. 8 A). As a control, we checked that a high concentration (10⁻⁶ M) of SR121463B did not affect the growth or viability of HT7 E. coli isolates (unpublished data). Also, the administration of low doses of SR121463B did not significantly affect plasma and urine osmolality (Table S3). Consistent with the decrease in urine osmolality caused by aquaresis (31), a high dose of the V2R antagonist (0.3 mg/kg) significantly increased plasma osmolality (Table S3). The administration of both low and high doses of SR121463B to infected mice resulted in significant dose-dependent increases in the expression of proinflammatory mediators and in the production of MIP-2 and TNF-α measured in whole kidney homogenates (Fig. 8, A and B). Compared with untreated infected mice, SR121463B treated mice also displayed a significant dose-dependent decrease in the renal bacterial burden 24 h after challenge (Fig. 8 C). Consistent with the stimulatory effect of the V2R antagonist on proinflammatory mediators, the number of CFU detected in kidneys proportionally decrease as a function of the dose of SR121463B administered (Fig. 8 C), whereas the number of Ly6-G–positive neutrophils infiltrating the kidneys and the levels of MPO activity remained almost the same in infected mice receiving a low dose (0.03 mg/kg) of SR121463B, and then fell significantly in infected mice treated with the high dose (0.3 mg/kg) of SR121463B (Fig. 8, D–F). Only a few UPECs (∼3,000 CFU per kidney) and infiltrating neutrophils were detected in kidneys from 0.3 mg/kg SR121463B–injected mice (Fig. 8, E and F). Furthermore, the small number of infiltrating neutrophils present in the infected kidneys from the 0.3 mg/kg SR121463B–injected mice was very similar to that found in dDAVP-treated Lpsⁿ mice, which exhibited ∼110-fold more bacteria in their kidneys (Fig. 7 D). These results suggest that SR121463B administration promoted rapid renal clearance of bacteria by stimulating proinflammatory mediators. The beneficial effect of the V2R antagonist appeared to be restricted to the kidney, because the number of CFUs detected in the bladder did not considerably differ in untreated and SR121463B-treated mice (unpublished data). These findings support the idea that this potent V2R blocker agent, which may impair the retrograde ascent of UPECs via its aquaretic effects (31), effectively antagonizes the action of endogenous dDAVP, leading to a significant (P < 0.05) increase in the expression of proinflammatory mediators and the subsequent activation of PMN influx, thereby promoting the recruitment of neutrophils to kill bacteria invading kidneys.

Renal tubule collecting duct cells play an active role in initiating the inflammatory host defense during renal bacterial infection (6). In this paper, we show that the antidiuretic peptide AVP, which acts on collecting duct cells, is also a potent immunomodulator of the innate response elicited by UPECs. Previous studies have demonstrated that the stimulation of proinflammatory responses in a variety of cell types is associated with down-regulation of cAMP and increased expression of intracellular adhesion molecule (ICAM) 1 (CD54) and vascular cell adhesion molecule 1, which are involved in the cell-to-cell contact-mediated host response (17, 52). In contrast, activation of cAMP-mediated signaling may suppress the immune response elicited by LPS and TNF-α. Agents that increase cellular cAMP have also been shown to inhibit the TNF-induced expression of ICAM-1 and vascular cell adhesion molecule 1 in human lung epithelial A549 cells (52), airway smooth muscle cells (17), and activated macrophages (18, 19). In this study, dDAVP also inhibited the LPS-induced expression of ICAM in mpkCCD_cl4 cells (unpublished data). A rise in cellular cAMP content was also shown to induce selective suppression of the activation of NF-κB in splenic B lymphocytes by blocking the phosphorylation of NF-κB/RelA and the degradation of IκB-α (53). Similar to these previous findings, we show here that dDAVP inhibited both the LPS-induced degradation of IκB-α and its stimulation of phosphorylated p65/RelA, as well as LPS-mediated NF-κB activation and cytokine secretion in mpkCCD_cl4 cells. Ser/Thr protein phosphatases play key roles in the regulation of phosphorylation of the NF-κB transcription factors. The SV40 small antigen, which associates with PP2A and inhibits its activity, enhances the activity of NF-κB (54). PP2A has been shown to dephosphorylate RelA directly in melanoma cell lines (55). In addition, the inhibition of PP1 and PP2A by okadaic acid induces the nuclear translocation and activation of NF-κB, as well as its activation in Jurkat cells and human neutrophils (56, 57). We show that after specific binding to V2R, dDAVP stimulates PP2A activity in both untreated and LPS-treated mpkCCD_cl4 cells. These findings are consistent with those of a previous immunohistochemical study demonstrating that the expression of the PP2A protein is restricted to the epithelial cells of the distal nephron of adult rat and mouse kidneys (33, 34). We show that inhibition of PP2A activity by exposure to low concentrations of okadaic acid and calyculin A blunted the inhibitory effect of dDAVP on the LPS-stimulated secretion of MIP-2 and TNF. This suggests that the cAMP-dependent PKA activation caused by dDAVP leads to increased PP2A activity, which in turn inhibits IκB-α degradation and p65/RelA phosphorylation in renal collecting duct epithelial cells.

We also demonstrate that the inhibitory action of dDAVP on cellular activation requires an intact CFTR. The regulation of CFTR by PKA has been extensively studied. The regulatory domain (R domain) of CFTR harbors multiple consensus sites for phosphorylation by PKA. The phosphorylation of the R domain regulates the opening of the CFTR channel (58). Multiple phosphatases may deactivate CFTR (38–41). PP2A has been shown to be one of the most potent protein phosphatases involved in the dephosphorylation of purified CFTR. Vastiau et al. (42) reported a direct and functional interaction between CFTR and PP2A. Using mass spectrometry, Thelin et al. (43) demonstrated that the PP2A catalytic A regulatory (Aα) and B′ (B′ε) regulatory subunits associate with CFTR, and that the latter binds directly to the C terminus of the CFTR molecule. PP2A inhibitors have also been shown to increase cAMP-stimulated CFTR currents in excised patches of airway epithelia and in intact mouse jejunum, and to delay their deactivation in colonic carcinoma Caco-2 cells (43, 44), which is consistent with physical interaction between CFTR and PP2A. Although neither calyculin A nor okadaic acid had any effect on dDAVP-stimulated CFTR-mediated Cl⁻ secretion, they both impaired the inhibitory effect of dDAVP on p65/RelA phosphorylation and on the secretion of MIP-2 and TNF-α elicited by LPS in mpkCCD_cl4 cells. CFTR blocking agents also impaired the inhibitory action of dDAVP on LPS-mediated activation. Collectively, these findings suggest that the closely associated CFTR and PP2A proteins are key signaling molecules activated by dDAVP, and that they down-regulate LPS-stimulated, TLR4-mediated inflammatory responses in renal collecting duct cells.

The cAMP-regulated CFTR is associated with inflammatory processes in other cell types. For example, defective CFTR function is responsible for cystic fibrosis, a chronic inflammatory lung disease characterized by bacterial colonization of the respiratory mucosa and an exaggerated and destructive chronic inflammatory response (59). CFTR has been shown to regulate the expression of RANTES, IL-8, IL-10, and inducible nitric oxide synthase (60). Interestingly, Estell et al. (60) also demonstrated that the CFTR-mediated NF-κB activation and RANTES expression require the insertion of CFTR into the plasma membrane. In this study, we show that the activation of the cAMP–PKA pathway by dDAVP, which down-regulates LPS-induced inflammatory responses in renal collecting duct cells, also requires an intact CFTR, which is inserted into the apical membrane of collecting duct cells in response to cAMP stimulation (12, 13). The inhibitory action of dDAVP seems to be restricted to the TLR4-mediated signaling pathway, because administration of the peptide hormone to mice infected by UPECs only altered inflammatory responses mediated by the TLR4 signaling cascade and not pathways independent of TLR4 (6). Our findings also suggest that activation of the PP2A–CFTR complex by dDAVP involves a dephosphorylation process on the TLR4-mediated, NF-κB signaling pathway. This scenario is consistent with a previous report showing that NF-κB activation and increased IL-8 production in cystic fibrosis tracheal epithelial cells is abrogated when dominant-negative signaling molecules, such as MyD88, are expressed (61).

Under conditions of dehydration, a rise in the circulatory level of vasopressin will stimulate NaCl and water reabsorption from the terminal parts of the nephron. We show that AVP, via its stimulatory effect on CFTR and PP2A, simultaneously reduces host inflammatory responses and favors renal bacterial invasion. This was illustrated by the fact that chronic administration of dDAVP reduces the renal expression of proinflammatory mediators and the recruitment of PMNs, and increased the susceptibility of Lpsⁿ (but not Lps^d) mice to UPEC infection. The reduction in urine flow resulting from increased NaCl and water reabsorption may also contribute to the increased susceptibility to ascending bacterial infection. However, there is no reason why this effect should be restricted to Lpsⁿ mice.

It is generally agreed that adequate hydration helps to improve the resolution of UTIs and even to prevent them, although there is no direct clinical evidence that dehydration promotes UTIs in humans (62). However, previous experimental studies have demonstrated that water deprivation does considerably increase the risk of E. coli–induced pyelonephritis in rats (63) and that enterococcal-induced pyelonephritis in rats could be cured by sustained water diuresis (64). These studies are in accordance with the present findings showing that vasopressin impairs immune response both in vivo and in vitro. This means that it is conceivable that a sustained increased in the concentration of vasopressin in kidneys would tend to inhibit the local immune response and favor the bacterial colonization by ascending uropathogens. Such situations could account, at least in part, for the high frequency of UTI in elderly patients, who are particularly susceptible to dehydration.

The action of dDAVP in the down-regulation of inflammatory response observed in UPEC-infected kidneys has been confirmed by the blocking experiments using a V2R antagonist. Previous binding studies have demonstrated that nonpeptide V2R antagonists, including salts of SR121463 and Tolvaptan, display high competitive affinity for the renal V2R from different species, including mice and humans (31, 65, 66). These very selective compounds also display only very low affinity for other AVP receptor subtypes and do not bind to many of the receptors that are unrelated to AVP (65, 66). In in vivo studies, these V2R antagonists induce marked aquaresis in healthy and diseased animals, and clinical trials have shown that they improve hyponatremia, congestive heart failure, and various other diseases associated with volume overload (31, 66–68). Oral or intravenous administration of SR121463 salts induce dose-dependent aquaresis lasting 2–6 h (65). Unlike conventional diuretics, such as furosemide, the action of the SR121463 salts is purely aquaretic, with no major changes in urinary Na⁺ or K⁺ excretion (31). We show that blockade of V2Rs by SR121463B stimulated the expression of proinflammatory mediators in the kidneys of infected mice. Furthermore, the combination of the stimulated expression of proinflammatory mediators and aquaresis caused by high doses of SR121463B (0.3 mg/kg) resulted in the virtually complete clearance of UPECs colonizing the kidneys. These findings strongly suggest that in vivo, as in cultured collecting duct cells, dDAVP has a potent inhibitory effect on the inflammatory response in UPEC-infected kidneys.

To conclude, this study demonstrates that dDAVP, in addition to its key role in fluid homeostasis, also influences local innate immune recognition in the kidney by inhibiting TLR4-mediated cell activation in collecting duct epithelial cells. This pathway involves both PP2A and intact CFTR Cl⁻ channel activity. Importantly, the negative regulatory action of dDAVP described targets the site to which ascending uropathogenic E. coli preferentially adheres in vivo, and may therefore produce an important downstream effect on host defense activation, bacterial proliferation, and disease progression. Identification of this novel negative regulatory mechanism illustrates the important influence of hormonal control on local innate immune recognition and identifies a previously unrecognized factor of disease susceptibility and clinical outcome.

NPPB was obtained from Research Biochemicals International. The thiazolidinone CFTR inhibitor 3-[(3-trifluoromethyl)phenyl]-5-[(4-carboxyphenyl)methylene]-2-thioxo-4-thiazolidinone (CFTR_inh-172) was provided by A.S. Verkman (University of California, San Francisco, San Francisco, CA). The V2R antagonist SR121463B (1-[4(N-tert-butylcarbamoyl)-2-methoxybenzene sulfonyl]-5-ethoxy-3-spiro-[4-(2-morpholinoethoxy)cyclohexane]indol-2-one, phosphate monohydrate) was provided by C. Serradeil-Le Gal (Sanofi-Aventis, Toulouse, France). SR121463B was dissolved in 10⁻² M DMSO and then in 0.9% NaCl to produce the appropriate final concentration. PKA (H89) and PKC (GF109203X) inhibitors were obtained from Merck Biosciences. All reagents used were tested for the absence of LPS contamination using the Limulus amebocyte lysate assay (BioWhittaker).

Adult female C3H/HeN (Lpsⁿ) and C3H/HeJ (Lps^d) mice were obtained from the Jackson Laboratory. Homozygous mutant cftr^m1unc (cftr^−/−) mice were obtained by a targeted mutation of the cftr gene (23), and their wild-type counterpart (cftr^+/+) mice were obtained from the Centre de Développement des Techniques Avancées pour l'Expérimentation Animale (provided by M.F. Bertrand, Centre National de la Recherche Scientifique, Orléans, France). Mice originally derived from ES129/Sv cells injected into C57BL/6J mice embryos have been further backcrossed on a C57BL/6J background for three generations and were then intercrossed. All mice were housed under specific pathogen-free conditions and used when 8–13 wk of age. Renal retrograde UTI was performed on Lpsⁿ and Lps^d mice, as previously described (6), using UPEC strain HT7. This UPEC strain was isolated from the urine of a woman with acute pyelonephritis and expresses the pap adhesin-encoding genes but lacks the hly α-hemolysin–encoding gene. The experimental procedures for the continuous infusion of dDAVP and administration of the V2R antagonist SR121463B before bacterial retrograde infection are listed in Supplemental materials and methods . The experiments were performed in accordance with the guidelines of the French Agricultural Office and in compliance with the legislation governing animal studies.

Experiments were performed on mpkCCD_cl4 cells (21) seeded on glass coverslips, Petri dishes, or Transwell permeable filters (0.4-μm pore size, 1-cm² insert growth area; Corning Costar Corp.) and on primary cultures of isolated CCDs microdissected from the kidneys of Lpsⁿ and Lps^d mice, or from the kidneys of cftr^−/− and cftr^+/+ mice, as previously described (6, 21). Cells were grown in a modified defined medium (DMEM/Ham's F12; 1:1 vol/vol; Invitrogen) supplemented with hormones and 2% fetal calf serum (21) in a 5% CO₂-95% air atmosphere. Confluent cells were incubated with purified E. coli (0111:B4 LPS Ultra-Pure; InvivoGen) or HT7 (5 × 10⁵ bacteria per well).

15 × 10⁶ mpkCCD_cl4 cells per milliliter were transfected by electroporation with the p(κB)₃ IFN-Luc plasmid (a Luciferase cis-reporter system containing 7× AP-1 and 3× NF-κB enhancer elements), as previously described (69). Luciferase activity was measured with a luminometer using the Luciferase Assay System (Promega) according to the manufacturer's instructions.

The experimental procedure for TLR4 siRNA experiments is described in Supplemental materials and methods.

Total RNA was extracted from confluent whole kidneys or cultured CCD cells using the RNeasy mini kit (QIAGEN), according to the manufacturer's instructions, and reverse transcribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen). cDNA was subjected to quantitative real-time PCR using a detector (Chromo 4; MJ Research). The mouse primers and TaqMan probes for IL-1β, MIP-2, MCP-1, RANTES, TNF-α, and β-actin were previously described (6). PCR data were reported as the relative increase in mRNA transcripts to that found in kidneys of naive mice or untreated cultured isolated CCDs, and were corrected by the respective levels of β-actin mRNA, which was used as the internal standard. cDNA and 400 ng of non–reverse-transcribed RNA were also subjected to RT-PCR using TLR4, MD-2, CD14, MyD88, α-ENaC, CFTR, or CD45 primers, as previously described (6).

Kidneys were fixed in Dubosc-Brazil solution, rinsed in PBS, embedded in paraffin, and stained with hematoxylin and eosin or periodic acid Schiff. Infiltrating PMNs were detected by immunohistochemical staining using an anti–Ly6-G (GR-1) antibody (1:1,000; BD Biosciences). The number of Ly6-G–labeled neutrophils per surface area (10⁴ μm²) was counted on five different kidney tissue sections for each of the experimental conditions tested. Indirect immunofluorescence studies were performed as previously described (6) using an affinity-purified rabbit anti-TLR4 antiserum (1:100) (27), the mouse monoclonal CTR433 (provided by M. Bornens, Centre National de la Recherche Scientifique, Institut Curie, Paris, France), or the anti–NF-κB antibody (Santa Cruz Biotechnology Inc.). Cy3- and Alexa Fluor 488–conjugated IgG secondary antibodies were obtained from Jackson ImmunoResearch Laboratories. In some cases, nuclei were stained with Hoechst 33258 (Pierce Chemical Co). Specimens were examined using a confocal laser scanning microscope (510-META; Carl Zeiss MicroImaging, Inc.), and the images generated were photographed.

TLR4/MD-2 intracellular staining was visualized using a rat monoclonal anti-TLR4/MD-2 (MTS510) antibody provided by K. Miyake (University of Tokyo, Tokyo, Japan) and a secondary goat anti–rat Cy5-conjugated antibody (Jackson ImmunoResearch Laboratories) after fixation (Cytofix; BD Biosciences) with or without permeabilization in Ca²⁺, Mg²⁺-free PBS containing 0.5% saponin and 2% FCS. Rat anti-hemagglutinin monoclonal antibody (Boehringer) was used as an isotype control.

Confluent cells were rinsed with PBS, scraped into 50 μl of lysis buffer (62.5 mM Tris-HCl [pH 6.8], 2% SDS, 10% glycerol, 50 mM DTT), and sonicated for 15 s at 4°C. Antibodies against TLR4 (27), IκB-α, total or phosphorylated NF-κB p65 (Cell Signaling Technology, Inc.), and β-actin (1:1,000; Sigma-Aldrich) were used to detect the corresponding antigens, as previously described (6).

Cell supernatants from mpkCCD_cl4 cells or isolated CCDs in primary culture were collected, and the concentrations of MIP-2 and TNF-α were determined using DuoSet mouse MIP-2 and TNF-α ELISA kits (R&D Systems), according to the manufacturer's instructions. Kidneys from dDAVP- or SR121463B-treated Lpsⁿ and Lps^d mice, some of which had been infected with HT7 isolates, were homogenized in 1 ml PBS and kept at −80°C until use. Tissue samples were thawed and assayed to measure the levels of MIP-2 and TNF-α production. Results were standardized to the amount of protein detected for each sample using a protein assay (Bio-Rad Laboratories) with BSA as standard. MPO activity was also measured using an ELISA kit (Hycult biotechnology), according to the manufacturer's instructions.

PP2A activity was determined using a Ser/Thr phosphatase assay kit (Upstate Biotechnology), according to the directions supplied by the manufacturer.

Confluent mpkCCD_cl4 cells were incubated with dDAVP, or with LPS or LPS plus dDAVP for 6 h. The cAMP content was measured using the cAMP Biotrak EIA system (GE Healthcare), according to the manufacturer's instructions.

Results are expressed as means ± SE. Significant differences were analyzed using the unpaired Student's t test and by analysis of variance using the Student-Newman-Keuls test and the Bonferroni t test for multiple comparison procedures. P < 0.05 was considered significant.

Fig. S1 depicts the expression of TLR4 and associated signaling molecules in confluent mpkCCD_cl4 cells. Fig. S2 provides an illustration of mpkCCD_cl4 cells forming domes and the mRNA expression of TLR4 and associated signaling molecules. Fig. S3 illustrates the antagonist effect of Ser/Thr protein phosphatases and CFTR on the inhibitory action of dDAVP on LPS-mediated cytokine secretion. Table S1 summarizes the effects of dDAVP infusion on the volume of urine excreted by Lpsⁿ and Lps^d mice housed in metabolic cages. Tables S2 and S3 show blood and urinary osmolality values 24 h after the inoculation of UPECs to untreated and dDAVP-treated mice or SR121463B-treated mice, respectively. Supplemental materials and methods provides information about the conditions of retrograde infection studies, dDAVP infusion, urine collection of mice acclimatized to metabolic cages, and administration of the V2R antagonist and TLR4 siRNA studies.

We are indebted to F. Cluzeaud, D. Gütle, and E. Pedruzzi for technical assistance. We would like to thank M. Muffat Joly for her skillful assistance with the in vivo studies using metabolic cages. We thank A.S. Verkman for the gift of CFTR_inh-172. We thank C. Serradeil-Le Gal for the generous gift of SR121463B. We also thank M.F. Bertrand, who kindly provided us with the cftr^m1unc mice and wild-type counterparts. We thank E. Ogier-Denis for his critical reading of the manuscript.

This work was funded by INSERM and in part by grants from the Association Vaincre la Mucoviscidose (to A. Vandewalle), the Deutsche Forschungsgemeinschaft (Ho 2236/5-1), the Swedish Research Council (K2003-31P-14792), Cancerfonden, and the University of Freiburg (to M.W. Hornef). C. Chassin was supported by a doctoral student grant from the French Ministère de la Défense (Délégation Générale de l'Armement/Mission pour la Recherche et l'Innovation Scientifique). A. Vandewalle was in receipt of an Interface INSERM-AP-HP fellowship.

The authors have no conflicting financial interests.