Loss of primary cilia increases polycystin-2 and TRPV4 and the appearance of a nonselective cation channel in the mouse cortical collecting duct

Abstract

Flow-related bending of cilia results in Ca²⁺ influx through a polycystin-1 (Pkd1) and polycystin-2 (Pkd2) complex, both of which are members of the transient receptor potential (TRP) family (TRPP1 and TRPP2, respectively). Deletion of this complex as well as cilia result in polycystic kidney disease. The Ca²⁺ influx pathway has been previously characterized in immortalized collecting duct cells without cilia and found to be a 23-pS channel that was a multimere of TRPP2 and TRPV4. The purpose of the present study was to determine if this TRPP2 and TRPV4 multimere exists in vivo. Apical channel activity was measured using the patch-clamp technique from isolated split-open cortical collecting ducts from adult conditional knockout mice with (Ift88^flox/flox) or without (Ift88^−/−) cilia. Single tubules were isolated for measurements of mRNA for Pkd1, Pkd2, Trpv4, and epithelial Na⁺ channel subunits. The predominant channel activity from Ift88^flox/flox mice was from epithelial Na⁺ channel [5-pS Na⁺-selective channels with long mean open times (475.7 ± 83.26 ms) and open probability > 0.2]. With the loss of cilia, the predominant conductance was a 23-pS nonselective cation channel (reversal potential near 0) with a short mean open time (72 ± 17 ms), open probability < 0.08, and a characteristic flickery opening. Loss of cilia increased mRNA levels for Pkd2 and Trpv4 from single isolated cortical collecting ducts. In conclusion, 23-pS channels exist in vivo, and activity of this channel is elevated with loss of cilia, consistent with previous finding of an elevated-unregulated Ca²⁺-permeable pathway at the apical membrane of collecting duct cells that lack cilia.

INTRODUCTION

Autosomal dominant polycystic kidney disease (PKD) involves mutations in PKD1 and PKD2 genes, which encode for polycystin-1 and polycystin-2, respectively, either of which results in renal cystogenesis (4). The majority of humans with autosomal dominant polycystic kidney disease have mutations in polycystin-1 (PC1), a poorly understood protein that is thought to serve as a mechanosensor (19). Polycystin-2 (PC2) is a Ca²⁺-permeable channel and has been shown to partner with PC1 (26, 36). Both proteins are members of the transient receptor potential (TRP) family, most of which appear to function as nonselective cation channels (13). The other nomenclature for these proteins is TRPP1(polycystin-1) and TRPP2 (polycystin-2). The TRPP1/TRPP2 complex has been extensively studied in cilia, where they normally serve to transduce flow-related ciliary movement into Ca²⁺ influx and mobilization (3, 19, 40). In fact, PKD is known as a ciliopathy since deletion of cilia also results in cystogenesis (17), which is a model that resembles autosomal recessive PKD. In a previous study (31), we used an immortalized cortical collecting duct (CCD) cell line derived from Oak Ridge polycystic kidney (Orpk) mouse (hypomorph of Ift88) and rescued cells as a control. The loss of cilia inhibited flow-mediated Ca²⁺ entry; however, there was an increase in unregulated apical Ca²⁺ entry compared with control cells with intact cilia (31). Thus, TRPP2 was still present and functional but was insensitive to variations in apical flow. A further study (46) was performed using patch clamp to identify this Ca²⁺-permeable pathway, and we found a 23-pS Ca²⁺-permeable cation channel, the expression of which was elevated in the absence of cilia. One characteristic of the TRP family is the extensive interactions within and across subfamily members to form multimeres with distinct biophysical properties. Previous studies using heterologous expression systems have shown that TRPP2 can partner with either TRPV4 or TRPC1 (18, 26). Whether these multimeres occur naturally in renal epithelial cells is unknown, which prompted us to determine if the 23-pS channel at the apical membrane might be a TRP multimere. Using siRNA knockdown experiments, it was determined that this 23-pS channel was a multimere of TRPP2 and TRPV4 (46).

The two questions asked in the present study were as follows: 1) does the 23-pS channel exist in vivo and 2) are protein expression and channel activity increased in the absence of cilia. This work was performed by assessing mRNA levels and channel activity with the patch-clamp technique in isolated split-open collecting duct (CD) principal cells derived from adult conditional knockout mice with (Ift88^flox/flox) or without (Ift88^−/−) cilia. These adult Ift88^−/− mice, which lack cilia, have almost no cystic structures at the time of study, so that it was not possible to patch cystic cells. However, cilia in this conditional mouse model have been deleted globally and, therefore, noncystic CD cells should exhibit the basic phenotypic characteristic of PKD, although we cannot exclude that cystic cells may have undergone additional transformations.

MATERIAL AND METHODS

Mice.

Development of the Ift88 floxed allele mice has been previously reported (8). Ift88 conditional knockout mice were generated by crossbreeding Ift88^flox/flox female mice with male mice that express tamoxifen-inducible Cre (CAGG-creER) (12). Four-week-old Ift88 conditional floxed allele mice with and without Cre were injected with tamoxifen (6 mg/40 g body wt) every other day for three doses for global knockout of the Ift88 gene. Genotyping for Ift88^flox/flox mice was performed by PCR using primer sequences as previously described (2). Both male and female mice were included in the study. IFT88 protein was deleted by 14 days after tamoxifen, as previously reported (8). Mice were maintained in accordance with the Institutional Animal Care and Use Committee regulations at the Medical University of South Carolina, Emory University, and University of Alabama at Birmingham. The animal study was performed in accordance with Animal Research: Reporting of In Vivo Experiments guidelines (16).

Single channel patch-clamp methods.

Four weeks after tamoxifen, mice were euthanized, renal tubules were manually dissected, and the cortical CD (CCD) was identified by the branching morphology of the tubules (N = 4–6 mice, both sexes included). Tubules were placed in physiological saline solution [containing (in mM) 140 NaCl, 5 KCl, 1 CaCl₂, and 10 HEPES adjusted to pH 7.4 with NaOH] in a plastic dish before being split open to reveal the apical surface of the cells before single channel patch clamp as previously described for patch clamp of cells in culture (32, 45). Briefly, a microelectrode (recording pipette resistance: 7–8 MΩ) was filled with physiological buffer solution in which Li⁺ was substituted for Na⁺ [containing (in mM) 140 LiCl, 1 CaCl₂, and 10 HEPES adjusted to pH 7.4 with NaOH] and lowered to a single cell before application of a small amount of suction to achieve a >1-GΩ seal. Principal cells were identified by their charactereristic size and flattened appearance under interference contrast. Epithelial Na⁺ channels (ENaCs) were identified by characteristic channel kinetics (long mean open and closed times of >0.5 s) and the current-voltage relationship of the channel (unit conductance close to 5 pS, between −20 and +20mV, a very positive, >40-mV reversal potential, and inward rectification). TRPV4 channels were identified as 23-pS channels with a reversal potential near zero that could be inhibited by the selective TRPV4 inhibitor GSK-2193874 (obtained from Tocris, Minneapolis, MN) (5, 6, 41).

Single tubule quantitative RT-PCR.

Four weeks after tamoxifen treatement, mice were euthanized and CDs were hand dissected using fine forceps under a microscope. The CCD was identified by morphology and selected only if the mRNA level for aquaporin-2 was higher than GAPDH. Each single tubule was then transferred to an Eppendorf tube containing lysis buffer (RNeasy Micro kit, Qiagen, Valencia, CA) and lysed by tip sonication for 20 s. Total RNA was isolated from lysates using the above RNeasy Micro kit (Qiagen). About 20 ng of total RNA were used for cDNA synthesis using the RT² first-strand kit and reverse transcribed using RT² SYBR Green MasterMix (Qiagen) and the following primers: Pkd1, 5′-GTTGTCCTCTCCATACAGCA-3′ and 5′-GATGTTAGCCGTGTCCGT-3′; Pkd2, 5′-ACAGTGTCCGAGTGTAGTAGT-3′ and 5′-CCGAGAGAAGTACCTGAAAAGTG-3′; Trpv4, 5′-GACCACGTTGATGAAGGAC-3′ and 5′-CTGGAGATCCTGGTGTACAAC-3′; Enac-α, 5′-ACCCCGTGAGTCTCAACATC-3′ and 5′-CCTGGCGAGTGTAGGAAGAG-3′; Enac-β, 5′-CATCCAGGCTGTCTTCATT-3′ and 5′-ACATGCTGAGGCAGGTCTCT-3′; and Enac-γ, 5′-ACACCACCCTCCACTGAGAC-3′ and 5′-CTGTGAGCTGGGAGGAAAAG-3′ (Integrated DNA Technologies, Caraville, IA). GAPDH was used as a reference gene for normalization. Relative message levels were calculated using ΔΔC_T (where C_T is threshold cycle), normalized to GAPDH levels. Data are presented fold changes relative to the Ift88^+/+ group from at least four separate determinations (N = 4–5 mice, 3–4 tubules from each mouse).

Statistical analysis.

Results are shown as means ± SE. The significance of the results was determined by a Fisher exact test and one-way ANOVA followed by a Tukey test for post hoc comparison or a Student’s t-test (Prism 6, GraphPad, La Jolla, CA). P values of <0.05 indicated statistical significance.

RESULTS

Loss of cilia results in the appearance of a 23-pS nonselective cation channel activity.

Figure 1A shows a representative image of an isolated CCD before (left) and after (right) one end of the tubule was split open. Single channel records from the apical membrane showed two distinct currents (Fig. 1B). The current predominantly seen in CCDs derived from Ift88^−/− mice that lack cilia (Fig. 1B, top channel) are currents that are characteristic of a nonselective channel (at least one channel in 16 of 20 patches: mean probability = 0.192 ± 0.135). The current-voltage (I-V) relationship from the nonselective channels was linear with a conductance of 22.6 pS and a reversal potential at zero (Fig. 1C). We know that the channel permeability is cationic nonselective since the sum of cations, Na⁺ and K⁺, inside the cell and outside must be the same, thus producing a reversal potential of zero. Cl⁻, the major anion in these cells, is lower inside the cells than outside, so the reversal potential would be nonzero if the channel were anion permeable.

Fig. 1.

A: Image of an isolated collecting duct before (left) and after (right) one end of a tubule from a mouse kidney was split open. A patch electrode on a principal cell is visible in the bottom left of the image (right). The diameter of the principal cell in the image was ~15 μm. B: single channel records from isolated split-open tubules from Ift88^−/− and control Ift88^flox/flox mice. In control mice, the currents were inward with long mean open times (O) and mean closed times (C) characteristic of epithelial Na⁺ channels (ENaCs). Principal cells from Ift88^−/− mice had two types of channels. Most cells had 23-pS nonselective (NSC) cation channels (reversal potential near zero) with short mean open times (72 ± 17 ms), an open probability = 0.192 ± 0.135, and characteristic flickery openings. C: current-voltage relationship for ENaCs from control mice (bottom) and NSC channels from Ift88^−/− mice (top). Different symbols represent different patches. In the top graph (NSC channel), the current-voltage plot was linear with a conductance of 22.6 pS and an intercept of 1.8 mV (r² = 0.968). In the bottom graph (ENaCs), the inward rectification and very positive reversal potential were characteristic of ENaCs. The line through the ENaC current-voltage relationship is the best nonlinear least-squares fit to the Goldman-Hodgkin-Katz equation to the control data. The fit predicted that intracellular Na⁺ was 11.6 ± 3.56 mM and that the apical membrane potential was −15.9 ± 9.70 with a principal cell Na⁺ permeability of 2.22 ± 0.507 × 10⁻⁷cm·s. The zero current potential was about +37 mV. This implies that the Na⁺ reversal potential was approximately +53 mV. Principal cell Na⁺ permeability was 2.22 ± 0.507 × 10⁻⁷cm·s. The conductance of the channel between −60 and 0 mV was ~5.2 pS, similar to that previously reported in rat connecting tubules (1, 24, 27, 44, 47). D: frequency of NSC channels versus ENaCs in principal cells from Ift88^−/− and control mice. The frequency of ENaCs was significantly reduced in Ift88^−/− compared with control mice (P < 0.001); the frequency of 23-pS NSC channels was significantly increased in Ift88^−/− compared with control mice (P < 0.001). E: box plot for open probability in Ift88^−/− compared with control mice. The boxes indicate 25−75% confidence levels, the line is the median value, the dashed line is the mean value, the whiskers indicate 5% and 95% confidence levels, and the dots show original data. Data were from five patches containing one NSC channel on five tubules from five separate Ift88^−/− mice and from six patches containing one ENaC on six tubules from six separate control mice.

The current measured predominantly in CCDs derived from Ift88^+/+ control mice, which have cilia (Fig. 1B, bottom channel), were inwardly rectifying with long mean open and closed times of >0.5 s, characteristic of ENaCs. Figure 1C shows a nonlinear I-V curve with a conductance of 5.2 pS (between −20 and +20mV), consistent with ENaCs. It should be noted that ENaC currents were also detected in Ift88^−/− mice but a much lower frequency than the nonselective channel (in 20 patches in the control group, we observed 18 ENaCs; in 20 patches in the Ift88^−/− group, we observed 4 ENaCs). The frequency of channels was significantly different: P < 0.001 by a z-test (Fig. 1D). Figure 1E shows the open probability (P_o) from five patches containing one nonselective channel on five tubules from five separate Ift88^−/− mice and from six patches containing one ENaC channel on six tubules from six separate control mice.

To determine the biophysical properties of the nonselective channel, freshly dissected split-open CCDs from Ift88^−/− mice were pretreated with the TRPV4 inhibitor GSK-2193874 (20 nM) before single channel measurements were performed with patch pipettes containing the inhibitor at the same concentration as the bath. However, it did not affect channel activity, indicating the nonselective channel is not sensitive to TRPV4 inhibitor (not shown).

Loss of cilia increases TRPP2 and TRPV4 in the CCD.

We next measured mRNA by quantitative RT-PCR from single tubules manually dissected from Ift88^−/− and control mouse kidneys (Fig. 2). mRNA levels in single CCDs demonstrated that loss of cilia resulted in an over twofold increase in Pkd2 and Trpv4 levels compared with CCDs derived from mice with intact cilia. There were no differences in mRNA levels for Pkd1, ENaC-α, and ENaC-γ (ENaC-β was below detection) between the presence and absence of cilia in the CCDs.

Fig. 2.

Quantitative RT-PCR from isolated single cortical collecting ducts (CCDs) from Ift88^−/− and control mice. mRNA levels for polycystin-2 (Pkd2) and transient receptor potential V4 (Trpv4) were higher in Ift88^−/− compared with control CCDs (**P < 0.01). In contrast, mRNA for polycystin-1 (Pkd1) and all epithelial Na⁺ channel (ENaC) subunits were similar among CCDs from Ift88^−/− and control mice. Each dot represents a single tubular mRNA value taken from N = 4–5 mice.

DISCUSSION

In a previous study (46), we studied apical membrane channel activity in immortalized CD cells that either expressed cilia or in which cilia had been largely deleted. We observed a predominant 23-pS cation channel that had a significant permeability to Ca²⁺. Expression of this channel activity was extremely low in CCD cells from both ciliated and nonciliated cells, although it was more abundant in cells that lacked cilia. However, apically applied EGF led to a 64-fold increase in channel activity in cells without cilia compared with cells with cilia (46). One explanation for this result is that the EGF receptor is mislocalized to the apical membrane, in the absence of cilia (7), and EGF was added only to the apical surface. Our further work was based on previous heterologous studies that identified potential TRP partners with TRPP2, including TRPC1 and TRPV4 (18, 19). Using siRNA knockout, we found that the 23-pS channel was a multimer of TRPP2 and TRPV4. As shown in other work, the combination of two different TRP channels can produce a channel that has biophysical properties that are distinct from the parent channels. This certainly applies to the 23-pS channel, since the conductance of TRPP2 is 120 pS (9) and that of TRPV4 is an outwardly rectifying channel with a conductance of 99 pS for outward current and 61 pS for inward current (39). The heteromeric nature of the channel likely explains why a TRPV4 inhibitor did not alter channel P_o. This finding in cultured cells confirmed the previous heterologous expression studies and suggested that this channel maybe an important Ca²⁺-permeable pathway in cilia or at the apical membrane. Furthermore, this channel could be responsible for flow-mediated ciliary Ca²⁺ entry and be involved in the pathogenesis of PKD (11, 35).

The present study was focused on answering the question of whether this same 23-pS channel was also present in native tissue. For this work, we performed patch clamp in split-open CD tubules taken from Ift88^flox/flox and Ift88^−/− mice with and without cilia. The important finding was that an essentially identical conductance was found in apical membranes of renal principal cells that do not have cilia, as was found in Orpk cells. In line with our cell culture experiments, we found that this channel activity was far greater in cells that lacked cilia compared with cells that had cilia. The reason for this enhanced activity of this channel, in the absence of cilia, is not known, but one speculation is that since these tubules are dissected and directly studied, latent EGF receptor activation could be involved. An alternative possibility is that in ciliated cells, 23-pS channels reside in the membrane very close to the base of the cilia in an area difficult to form patch electrode seals. In nonciliated cells, the region with channels is more accessible to patch electrode so the frequence of observing channels is increased.

Although the single isolated CCD does not reflect a pure principal cell population, our previous imaging indicated that the cell population in the CCD is predominantly principal cells (29). In the present study, we found that mRNA for TRPP2 and TRPV4 was elevated in single tubules from Ift88^−/− compared with control mice. The reason for increased polycystin-2 was not clear, but previous studies in human PKD kidneys have shown both polycystin-1 and polycystin-2 to be increased in the cystic renal epithelium and that it could be a result of a functional cleaved COOH-terminal of the protein (21, 38). This suggests that another explanation for the enhanced channel activity was due to greater expression of TRPP2 and TRPV4 protein, which could explain the elevated channel activity with the loss of cilia. These results are also consistent with our previous study (31), which suggested an enhanced and unregulated Ca²⁺ entry pathway with the loss of cilia.

In other work, it would appear that ENaC channel activity was higher in control compared with Ift88^−/− CCD cells. However, there was no difference in mRNA levels at least for ENaC-α and ENaC-γ. There are probably a number of explanations for this difference between message and channel activity. The simplest is that increases in intracellular Ca²⁺ near the apical membrane (as would be expected with the increased number of 23-pS channels) is known to strongly inhibit ENaC (1, 34). There has been conflicting data on whether ENaC expression and regulation of Na⁺ reabsorption is enhanced or suppressed in PKD (15, 20, 23, 28, 30, 37).

There have been conflicting results regarding cytosolic Ca²⁺ levels in PKD (22, 31, 35, 42, 43). Clearly, there is work that suggests that the overall cell cytosolic Ca²⁺ concentration is lower in PKD (35, 42, 43). Previously, we measured subapical membrane cytosolic Ca²⁺ levels and found them to be elevated with loss of cilia (31). This was accomplished using a membrane-bound form of the Ca²⁺ probe fura-2 and a membrane folding technique. Others have also shown a subapical region with elevated Ca²⁺ associated with Ca²⁺ buffering by large numbers of apical mitochondria (47). If the 23-pS channel in native CDs is in fact the same channel found in vitro and is a Ca²⁺-permeability pathway, this would be consistent with elevated subapical membrane Ca²⁺ levels with loss of cilia, regardless of the overall cell cytoplasmic Ca²⁺ concentration. Although a number of studies have shown flow-dependent increases in cell Ca²⁺ that are dependent on the presence of a functioning cilia (14, 19, 25, 33), a note of caution, however, concerns a recent study (10) that suggested that primary cilia are not a Ca²⁺-responsive mechanosensor. Cilia-mediated cell signaling in the CCD, whether it is Ca²⁺ dependent or not, is complex, and, importantly, how cilia are involved in cystogenesis remains to be defined.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants K08-DK-106465 (to T. Saigusa), R03-DK-119717(to T. Saigusa), P30-DK-074038 (to P. D. Bell), and DK-110409 (to D. C. Eaton).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

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