All electrical recordings were made from whole primary cilia of mIMCD-3 cells as described previously (Kleene and Kleene 2012). In short, mIMCD-3 cells (murine epithelial cells from the renal inner medullary collecting duct, CRL-2123, American Type Culture Collection (Rauchman et al., 1993)) were cultured on beads that were not attached to the recording chamber. By applying suction to a recording pipette, a single primary cilium could be made to enter the pipette. If the resistance between the membrane and the pipette was sufficient (≥1 GΩ, or 0.5 GΩ with the low-Ca²⁺ external solution), the cilium was excised from the cell. This left the cilium inside the recording pipette in the inside-out configuration. The pipette containing the cilium was then transferred among different solutions that bathed the cytoplasmic face of the membrane. A schematic diagram appears elsewhere (Kleene and Kleene 2017). Recordings were done under voltage clamp using equipment and software as previously described (Kleene and Kleene 2012). The temperature of the recording chamber was not controlled other than by the room thermostat (24°C). During acquisition, currents were low-pass filtered at 2 kHz and digitized at 5 kHz. For figures, and for analysis of single TRPV4 channels, recordings were again low-pass filtered at 500 Hz. With the exception of the TRPM4 experiments, a correction of 2.7 mV was applied for a liquid junction potential between each internal bath (which contained mostly KCl) and its salt bridge (which contained mostly NaCl). All potentials are given as membrane potential (i.e. internal relative to external).

The beads coated with cells were stored in a standard external (extracellular) solution containing (in mM) NaCl 140, KCl 5, CaCl₂ 2, MgCl₂ 2, sodium pyruvate 2, HEPES 5, and D-glucose 9.4, adjusted to pH 7.4 with NaOH. The recording pipettes were filled with this same external solution with two exceptions. In testing the effect of low external Ca²⁺, total CaCl₂ was reduced to 0.8 mM, and 2 mM BAPTA was added, giving 0.1 µM free Ca²⁺. When studying TRPM4 current, the standard external solution was modified to include KCl 145 and NaCl 0 to maximize the TRPM4 current (Flannery et al., 2015). All internal (cytoplasmic) solutions included KCl 140, NaCl 5, MgCl₂ 2, HEPES 5, and D-glucose 5, adjusted to pH 7.4 with KOH. Each internal solution also included CaCl₂ and a Ca²⁺ buffer as described below. U-73122, an inhibitor of phospholipase C (5 µM), and GSK1016790A, a TRPV4 agonist (50 nM) were added to some internal solutions as noted.

For some studies, I selected cilia with no detectable PC2 channels. To check for the presence of TRPP2 channels, the cilium was exposed to one of the solutions with reduced Ca²⁺ buffering. Voltage was maintained at +60 mV. If no large-conductance channels were seen to open within 2 min, active TRPP2 channels were judged to be absent.

Apparent association constants K' _Ca between Ca²⁺ and each of the three Ca²⁺ buffers used were determined by Scatchard analysis (Bers 1982) with a Ca²⁺-selective electrode (Orion 932000). The K' _Ca values were 6.7 × 10⁶ M⁻¹ for BAPTA, 2.0 × 10⁶ M⁻¹ for EGTA, and 0.78 × 10⁶ M⁻¹ for dibromoBAPTA. The x-intercept of the Scatchard plot also yielded the purity of the chelator, which was 84–88%. The impurity is reportedly water (Harrison and Bers 1987; Marks and Maxfield 1991). Calcium chloride solutions were prepared by dilution from a standard 0.1 M solution (Orion 922006) traceable to the National Institute of Standards and Technology. Table 1 lists the buffered Ca²⁺ solutions prepared. Our double-distilled water contained 9 ± 1 µM Ca²⁺ (n = 6 determinations), and this is included in the values shown for [Ca²⁺]_total. The low-Ca²⁺ solutions buffered with BAPTA or EGTA were not sensitive to small errors in the concentrations of total Ca²⁺ or buffer. In all cases, a 2% error (high or low) in either quantity gave a concentration of free Ca²⁺ between 13 and 21 nM.

BAPTA, EGTA, GSK1016790A, and U-73122 were purchased from MilliporeSigma and dibromoBAPTA from Santa Cruz Biotechnology. GSK1016790A was diluted from a 100 µM stock solution and U-73122 from a 5 mM stock solution, both in DMSO.

For voltage jumps, the cilium was clamped at a prepulse potential for 10 s, which was then stepped to a more positive (postpulse) potential for 10 s. Prepulse potentials tested ranged from −140 mV to +20 mV and postpulse potentials from +20 mV to +80 mV. For analysis, the mean current in the absence of detectable channel openings (leak current) was subtracted in QuB (State University of New York, Buffalo, NY). Transient currents are defined as the mean current during the first 3 s of the postpulse minus the mean current during the final 3 s of the postpulse. For steady-state currents, mean currents over a 20 s recording, minus the leak current, are reported.

For macroscopic TRPV4 current-voltage relations, current was measured as voltage was continuously varied from −100 or −140 mV to +100 mV. Current at −90 mV was figured as the mean of the values from −100 mV to −80 mV, and current at +90 mV as the mean of the values from +80 mV to +100 mV. For the single-channel study of TRPV4, 20 s recordings were made at each voltage. Single-channel currents were determined from amplitude histograms.

Results are expressed as mean ± standard error. Data sets were compared with a nonparametric test, either the Wilcoxon signed rank test (for paired measures) or the Mann-Whitney rank sum test (for unpaired measures). Our criterion for significance was p < 0.05.

In the primary cilium of mIMCD-3 cells, in the absence of agonists, PC2 channels are the only large-conductance channel seen, and the only channel activated by micromolar concentrations of intraciliary Ca²⁺ (Kleene and Kleene 2017; Liu et al., 2018). Under certain conditions, it was possible to transiently activate the channels to a level above that seen in the steady state. When the cilium was held at a more negative potential for 10 s and then jumped to a more positive potential, such a transient activation could be observed given reduced internal Ca²⁺ buffering (Figure 1A, left, 0.1 mM BAPTA; Supplementary Figure S1). The transient activity was greater when the prepulse potential was more negative. The internal solution for these tests was prepared to have a free Ca²⁺ concentration of 15 nM, and the external solution included 2 mM Ca²⁺. Transferring the cilium to an internal solution with a greater buffer concentration (2 mM), while maintaining the low free Ca²⁺ concentration, prevented the transient current (Figure 1A, right, 2 mM BAPTA). This indicates that the transient activity results from an increase in intraciliary free Ca²⁺ that can be overcome with sufficient Ca²⁺ buffering. Although there is a driving force for Ca²⁺ to enter the cilium at all of the voltages tested, the prepulse temporarily increases that driving force. Under favorable conditions, even a single PC2 channel might admit enough Ca²⁺ to further activate itself. Cilia with a single active PC2 channel are encountered on occasion. In each of 4 such cases tested, the channel showed a transient activation following a prepulse to a sufficiently negative potential if Ca²⁺ buffering was low (Figure 1B, left). Again, the effect was not seen with increased Ca²⁺ buffering (Figure 1B, right). The reduction of activity with higher Ca²⁺ buffering was reversible (Figure 1C). In each of 17 cilia tested in 0.1 mM BAPTA, a prepulse protocol could be found that led to a transient activation given reduced Ca²⁺ buffering. In 15 cilia tested in both 0.1 and 2 mM BAPTA, the maximum mean transient current was 34 ± 15 pA with 0.1 mM BAPTA and 1.9 ± 1.3 pA with 2 mM BAPTA (Figure 1D). The difference is significant (p < 0.001, Wilcoxon signed rank test for paired measures). Prepulse voltages sufficient to cause transient activation varied from one cilium to the next. Figure 1E shows the most positive prepulse potential able to cause transient activation in 0.1 mM BAPTA.

The voltage prepulses I tested temporarily increased the driving force for Ca²⁺ to enter the cilium. Even at a steady potential, though, PC2 channels might conduct enough Ca²⁺ into the cilium to increase their own activities. This was also confirmed experimentally. In 12 of 13 cilia tested, channel activity over some range of holding potentials was greater when Ca²⁺ buffering was reduced (Figure 2A, left) than with higher Ca²⁺ buffering (Figure 2A, right). The effect was observed even in each of 3 cilia with just one active PC2 channel (Figure 2B) and was reversible (Figure 2C). In the 13 cilia tested, the mean steady-state current was 6.7 ± 1.5 pA with 0.1 mM BAPTA and 0.8 ± 0.5 pA with 2 mM BAPTA (Figure 2D). The difference is significant (p < 0.001, Wilcoxon signed rank test for paired measures).

In all but one cilium, the activity of the channel showed a voltage dependence that has been well described, increasing as the steady-state potential was made more positive. In a single cilium, however, channel activity decreased with low Ca²⁺ buffering as the potential was changed from −20 to +20 mV (Figure 2E). This sequence was repeated 3 times in this cilium with the same result.

EGTA should be less effective than BAPTA at restricting the accumulation of intraciliary Ca²⁺. Although the two buffers have similar equilibrium association constants with Ca²⁺ at cellular pH, the rate of Ca²⁺ binding by BAPTA is roughly 100 times faster than the rate for EGTA (Dargan and Parker 2003). Thus a sufficiently rapid influx of Ca²⁺ may overcome buffering by EGTA but not by BAPTA. I have confirmed this prediction by comparing the effects of the two Ca²⁺ buffers. With 0.25 mM EGTA as the buffer in the internal solution, it was possible in each of 11 cilia tested to identify a prepulse potential that allowed a transient activation of PC2 channels after jumping to a more positive potential (Figure 3A, left, EGTA; Supplementary Figure S2). The transient activation was absent when the cilia were transferred to an internal solution with 0.25 mM BAPTA instead of EGTA (Figure 3A, right, BAPTA). The internal solutions with both EGTA and BAPTA were prepared to have free Ca²⁺ concentrations of about 20 nM. With EGTA, the transient activation was seen in each of 4 cilia tested that had a single active PC2 channel (Figure 3B), and the effect of BAPTA was reversible (Figure 3C). In 11 cilia tested in both EGTA and BAPTA (each at 0.25 mM), the maximum mean transient current was reduced from 22 ± 8 pA with EGTA to 0.3 ± 0.3 pA with BAPTA (Figure 3D). The effect of buffer choice is significant (p < 0.001, Wilcoxon signed rank test for paired measures). Figure 3E shows the most positive prepulse potential able to cause transient activation in each of 16 cilia tested in EGTA.

In most cilia tested, steady-state channel activity was greater at some holding potentials with EGTA as the Ca²⁺ buffer instead of BAPTA (each at 0.25 mM). This was seen in cilia with several PC2 channels (Figure 4A) and in 3 cilia each with a single channel (Figure 4B). The effect of exposure to BAPTA was reversible (Figure 4C). In the 13 cilia tested, the mean steady-state current was 7.4 ± 1.8 pA with 0.25 mM EGTA and 3.0 ± 1.6 pA with 0.25 mM BAPTA (Figure 4D). The difference is significant (p = 0.010, Wilcoxon signed rank test for paired measures).

Each experiment shown so far compares two internal solutions that were prepared to have identical concentrations of free Ca²⁺, in all cases well below the ∼1 µM or more needed to activate PC2 in previous reports (Kleene and Kleene 2017; Liu et al., 2018). Voltage protocols were found that allowed activation of PC2 channels to levels expected for much higher intraciliary Ca²⁺ concentrations than the concentrations initially provided. The dependence of this activation on the concentrations and rate constants of the Ca²⁺ buffers shows that the activation is Ca²⁺-dependent. Apparently there is a source of intraciliary Ca²⁺ beyond that present in the internal solutions provided at the start. To test whether the source is external Ca²⁺ that crosses the membrane to the intraciliary compartment, the experiments were repeated with external Ca²⁺ reduced from 2 mM to 0.1 µM. With reduced external Ca²⁺, voltage prepulses even to −140 mV caused no transient activation of PC2 (Figure 5A). With 2 mM external Ca²⁺, the maximum mean transient current following a prepulse averaged 34 ± 15 pA (n = 15 cilia); with 0.1 µM external Ca²⁺ the value was 0.8 ± 0.4 pA (n = 9 cilia) (Figure 5B). The values are significantly different (p < 0.001, Mann-Whitney rank sum test for unpaired measures). This was true (p = 0.0024) even if the three very high values from the population tested in 2 mM Ca²⁺ were excluded. With reduced external Ca²⁺, steady-state PC2 current at +40 or +60 mV did not depend on the concentration of the Ca²⁺ buffer (Figure 5C). The mean currents averaged 1.9 ± 0.7 pA with 0.1 mM BAPTA and 1.2 ± 0.6 pA with 2 mM BAPTA. The difference is not significant (n = 10 cilia, p = 0.084, Wilcoxon signed rank test for paired measures). With reduced cytoplasmic Ca²⁺ buffering, mean steady-state current was greater with 2 mM external Ca²⁺ (6.7 ± 1.5 pA, n = 13 cilia) than with 0.1 µM external Ca²⁺ (1.9 ± 0.7 pA, n = 13 cilia) (Figure 5D). The difference is significant (p = 0.0099, Mann-Whitney rank sum test for unpaired measures).

The additional Ca²⁺ that activates PC2 channels when Ca²⁺ buffering is reduced may enter through the PC2 channels themselves, but the experiments thus far do not allow that conclusion. An ideal control would be a PC2 channel that has the usual sensitivity to intraciliary Ca²⁺ but does not conduct Ca²⁺. If transient activation were to be seen with such a channel, it would indicate that there is a Ca²⁺ source other than flow through PC2 itself. No such PC2 variant has been described, but the renal primary cilium has another channel that can serve this purpose: TRPM4. These channels are present in the primary cilia of most mIMCD-3 cells and, like other TRPM4 channels reported, are activated by internal Ca²⁺ but do not conduct Ca²⁺ (Mathar et al., 2014; Flannery et al., 2015). In typical experimental conditions, including those used above, they are much less sensitive than PC2 to internal Ca²⁺. When phospholipase C is inhibited, though, TRPM4, like PC2, is activated by micromolar levels of internal Ca²⁺ (Zhang et al., 2005; Nilius et al., 2006; Flannery et al., 2015). Many of the renal primary cilia lack active PC2 channels (Kleene and Kleene 2017; Ha et al., 2020), and such cilia were chosen to study TRPM4 currents in isolation. In these cilia, I inhibited phospholipase C by adding 5 µM U-73122 to the internal solutions and looked for a source of Ca²⁺ influx that might activate TRPM4 channels.

The top trace of Figure 6A shows the steady-state TRPM4 current activated at +60 mV when internal free Ca²⁺ was buffered to 10 µM with 2 mM dibromoBAPTA. Under this condition a macroscopic TRPM4 current was activated that averaged 5.8 ± 1.0 pA at +60 mV (n = 18 cilia). When a cilium was moved to an internal solution with just 15 nM free Ca²⁺ but also reduced Ca²⁺ buffering (0.1 mM BAPTA), almost no TRPM4 current was seen (Figure 6A, middle recording). In this case the mean TRPM4 current averaged 0.14 ± 0.04 pA (n = 18 cilia). The reduction of TRPM4 current is significant (Figure 6B, n = 18 cilia, p < 0.001, Wilcoxon signed rank test for paired measures). On return to the internal solution with 10 µM free Ca²⁺, the TRPM4 current was restored (Figure 6A, bottom trace). On average, the ratio of the TRPM4 current seen with 15 nM Ca²⁺ and reduced buffering to that activated in 10 µM free Ca²⁺ was 0.03 ± 0.01 (n = 18 cilia).

To verify that the Ca²⁺-activated current was conducted through TRPM4 channels, the experiment was repeated in cells in which TRPM4 was knocked down with shRNA (Flannery et al., 2015). In cilia from such cells, the steady-state current activated by 10 µM free Ca²⁺ was 0.21 ± 0.05 pA (n = 8 cilia), which is significantly less that the current in wild-type cells (Figure 6C, p < 0.001, Mann-Whitney rank sum test for unpaired measures). With reduced Ca²⁺ buffering, a prepulse to a negative voltage did not give rise to a detectable transient TRPM4 current in cilia from wild-type cells when the voltage was jumped from −120 or −140 mV to +60 mV (Figure 6D). Following a prepulse, the mean transient current was 0.15 ± 0.12 pA (n = 10 cilia). In the absence of PC2 channels, I found no evidence of a Ca²⁺ influx that could activate TRPM4 even with reduced Ca²⁺ buffering. This supports the hypothesis that the activation of PC2 described above occurs following an influx of Ca²⁺ through the PC2 channels themselves.

The primary cilia of mIMCD-3 cells also express TRPV4 channels. To study TRPV4 currents in isolation, cilia lacking active PC2 channels were again selected. With internal Ca²⁺ buffered with 0.25 mM BAPTA, the ciliary macroscopic current-voltage relation showed a small leak conductance (Figure 7A, black). Addition of the TRPV4 agonist GSK1016790A (50 nM; Willette et al., 2008) activated an outwardly rectifying macroscopic current (Figure 7A, red). The activation was reversible on removal of the agonist (Figure 7A, orange). With 0.25 mM BAPTA, the additional conductance seen in the presence of GSK1016790A averaged 1,648 ± 449 pS measured between +20 mV and +80 mV (n = 18 cilia). The mean reversal potential of the additional current (intersection of the red and black recordings) was −9.0 ± 0.6 mV (n = 14 cilia).

In a few cilia, the number of active channels was low enough to allow resolution of single-channel events (Figure 7B). The average single-channel conductance was 39 pS when measured between −140 and −60 mV and 90 pS between +20 and +80 mV (Figure 7C, blue lines). The current-voltage relation of the macroscopic current (Figure 7C, gray) was similar to that of the single channels, suggesting that the channel open probability does not strongly depend on voltage. The relief of outward rectification at very negative potentials is characteristic of TRPV4 in the presence of extracellular Ca²⁺ (Voets et al., 2002).

TRPV4 channels can be activated by internal Ca²⁺ and also conduct Ca²⁺ (Strotmann et al., 2003). In other systems, this influx potentiates their activity by positive feedback, producing a regenerative response (Strotmann et al., 2003; White et al., 2016). I was able to demonstrate this effect in renal primary cilia. With reduced buffering of internal Ca²⁺ (0.25 mM BAPTA), 50 nM internal GSK1016790A activated an outwardly rectifying current (Figure 8A, compare black and red). When the internal solution was replaced with one still containing 50 nM GSK1016790A but with BAPTA increased to 2 mM (Figure 8A, orange), the agonist-activated current was greatly reduced. With 0.25 mM BAPTA, the additional current seen in the presence of GSK1016790A averaged −16 ± 6 pA at −90 mV and 86 ± 20 pA at +90 mV (Figure 8B, n = 7 cilia). With 2 mM BAPTA, the additional current averaged 3.8 ± 1.3 pA at −90 mV and 3.7 ± 3.6 pA at +90 mV (Figure 8B, n = 7 cilia). The effect of buffer concentration is significant at both potentials (p = 0.0016 at either potential, Wilcoxon signed rank test for paired measures).

Both PC2 and TRPV4 conduct Ca²⁺ and can be activated by intraciliary Ca²⁺. Thus a Ca²⁺ influx through one of these channel types might activate the other. I tested whether ciliary PC2 channels could show a secondary activation following direct activation of TRPV4 in the presence of 0.25 mM BAPTA. Figure 9A shows the outcome in a cilium with just one active PC2 channel. With reduced internal Ca²⁺ and no agonist, the PC2 channel opened only at strongly depolarizing potentials (Figure 9A, left, black), as expected (Kleene and Kleene 2017; Liu et al., 2018). Addition of the TRPV4 agonist GSK1016790A (50 nM) increased the current at most potentials (Figure 9A, left, red). In general, it is not straightforward to assess the proportions of the PC2 and TRPV4 currents after addition of the agonist. GSK1016790A activates a TRPV4 current with a reversal potential near −9.0 mV (Figure 7A); the reversal potential of PC2 under these conditions is −61 mV (Kleene and Kleene 2017). The mixed current activated by GSK1016790A reversed at −13 mV (Figure 9A, left, intersection of the red and black recordings), suggesting that it was mostly carried by TRPV4 channels. Near the reversal potential for TRPV4 (−9.0 mV), there is no net TRPV4 current, so one can measure the PC2 current in isolation. (Note that even at this potential there is still an inward driving force for Ca²⁺. The influx of Ca²⁺ and Na⁺ is cancelled by an efflux of K⁺.) On addition of 50 nM GSK1016790A, the open probability of the single PC2 channel at −5 mV increased from almost 0 to a value approaching 1 (Figure 9A, right). On return to a solution lacking GSK1016790A, the current at all potentials returned to the lower level (Figure 9A, left, orange), and the single PC2 channel returned to an open probability near 0 (Figure 9A, right, bottom).

In a second cilium, the agonist activated a mean PC2 current at −5 mV of 177 pA (Figure 9B, right), which equals the current expected from 32 PC2 channels, assuming an open probability of 1 (Kleene and Kleene 2017). The current activated by GSK1016790A in this cilium reversed at −52 mV (Figure 9B, left), suggesting that most of the activated current was carried by PC2 channels. Of 8 cilia tested, 7 showed increased PC2 activity at −5 mV in the presence of 50 nM GSK1016790A (Figure 9C). PC2 channels were almost never seen to open at −5 mV in the absence of GSK1016790A; the mean channel current never exceeded 0.17 pA and averaged 0.025 ± 0.021 pA (n = 8 cilia). In the presence of GSK1016790A, the mean current in the same cilia at −5 mV averaged 37 ± 21 pA (n = 8 cilia). The effect of GSK1016790A on PC2 activity is significant (p = 0.0078, Wilcoxon signed rank test for paired measures).

The work presented here identifies conditions that allow regenerative Ca²⁺ signaling in the primary cilia of renal epithelial cells. Figure 10 summarizes that system. The cilia contain at least two types of ion channels, PC2 and TRPV4, that conduct Ca²⁺ from the external compartment and also are activated by intraciliary Ca²⁺. This can result in positive feedback: As Ca²⁺ enters the cilium, it increases the activity of PC2 and TRPV4 channels, causing further Ca²⁺ entry, which causes further activation, and so on. Intraciliary Ca²⁺ also increases the open probability of TRPM4 channels, but those do not conduct Ca²⁺. Opening any of these channels should depolarize the cilium. That may further contribute to a regenerative response, since the PC2 channels are also activated by depolarization (Kleene and Kleene 2017; Liu et al., 2018). PC2 is also part of a regenerative system in the endoplasmic reticulum of renal cells (Sammels et al., 2010). There PC2 closely interacts with another Ca²⁺-activated, Ca²⁺-conducting channel, the inositol 1,4,5-trisphosphate receptor (IP₃R). The interaction potentiates Ca²⁺-induced Ca²⁺ release unless the channel function of PC2 is eliminated by mutation. The potentiation is sensitive to the speed and concentrations of the Ca²⁺ buffers chosen. A similar system was described in oocytes overexpressing PC2 and IP₃R (Li et al., 2005). By contrast, in artificial bilayers, PC2 inhibits the activity of single channels of another Ca²⁺-activated, Ca²⁺-conducting channel, the ryanodine receptor (Anyatonwu et al., 2007).

Not surprisingly, the concentrations and speeds of the internal Ca²⁺ buffers profoundly influenced the extent of positive feedback in the cilium. With low internal free Ca²⁺ and 2 mM BAPTA, the open probability of PC2 does not exceed ∼0.2, even at favorable potentials (e.g. +80 mV; Kleene and Kleene 2017). When the initial internal solution contained low Ca²⁺ but just 0.1 mM BAPTA, PC2 channels neared a fully open state even at −40 mV (Figure 2B). This reflects an intraciliary free Ca²⁺ concentration >10 μM, as judged by measurements made with stronger Ca²⁺ buffering (2 mM BAPTA; Kleene and Kleene 2017). This is an increase of at least 500-fold over the free Ca²⁺ initially provided (15–21 nM). The additional intraciliary Ca²⁺ arose from Ca²⁺ flowing through PC2 channels into the cilium, where it was able to overcome the dilute Ca²⁺ buffer and accumulate. Consistent with this explanation, positive feedback could not be achieved with low external Ca²⁺, nor with a channel (TRPM4) that is activated by Ca²⁺ but does not conduct it. When BAPTA was replaced with EGTA, a buffer that is slower to bind Ca²⁺, positive feedback was enhanced (Figures 3, 4). Activation of TRPV4 by an agonist was also dependent on Ca²⁺ buffering; it was successful with 0.25 mM BAPTA but not 2 mM BAPTA (Figure 8). Regenerative signaling by TRPV4 has been described in other systems (Strotmann et al., 2003; White et al., 2016).

In a limiting case, a single PC2 channel was able to conduct enough Ca²⁺ into the cilium to further activate itself (Figures 1B, C, 2B,C, 3B, 4B). This presumably requires diffusion of Ca²⁺ over a very short distance, from the mouth of the channel pore to an internal Ca²⁺-activation site on the same channel, which has not been identified (Vien et al., 2020). This self-activation was somewhat more effective with EGTA than with BAPTA (Figures 3, 4). In models of other Ca²⁺-conducting channels, internal Ca²⁺ ranges from 50 µM to over 200 μM at the center of the channel (Naraghi and Neher 1997; Neher 1998; Berkefeld et al., 2006; Foskett et al., 2007). The effects of EGTA and BAPTA are expected to differ only over distances of about 20–200 nm from the Ca²⁺ source (Neher 1998; Berkefeld et al., 2006), suggesting that the site of PC2 self-activation by Ca²⁺ is within this distance from the pore. Whether the Ca²⁺ domain of a single channel extends to other channels should depend on the Ca²⁺ buffers and on the distributions of the channels along the length of the cilium. In dendritic segments, Ca²⁺ signals can spread over distances of 5–500 µm (Augustine et al., 2003). Internal Ca²⁺ can be as high as 5–10 µM even 200 nm from the Ca²⁺ source (Neher 1998). This would be more than enough to activate the ciliary channels, but the size of the domain depends strongly on the speed and concentration of the buffer (Berkefeld et al., 2006). In cilia with multiple PC2 channels, the regenerative effects might have resulted strictly from individual channels locally self-activating. However, activation of TRPV4 channels with an agonist increased intraciliary Ca²⁺ enough for secondary activation of PC2 channels (Figure 9). The relative positions of the TRPV4 and PC2 channels are not known.

It would be instructive to model this system, but one would need to know the identities, concentrations, mobilities, rate constants, and Ca²⁺ affinities of any mobile or immobile intraciliary Ca²⁺ buffers; the locations and Ca²⁺ conductances of the relevant channels; the locations and sensitivities of their Ca²⁺-activation sites; and the diffusion geometry within the cilium. Most of these parameters are not yet available. The cilium does not contain endoplasmic reticulum. There could be membrane transporters that eject Ca²⁺, although none have been described in these cilia. The results presented here suggest some obvious tradeoffs that a model might help to explain. For example, making the voltage more negative increases the driving force for Ca²⁺ entry, which should increase the response. However, it also decreases the open probability of the PC2 channels (Kleene and Kleene 2017; Liu et al., 2018), which should decrease the response. In most cilia, more negative voltage increased the response, suggesting that the driving force for Ca²⁺ dominated. However, a single cilium (Figure 2E) showed the opposite effect of voltage. The voltages needed to optimize the response varied substantially from cilium to cilium (Figures 1E, 3E). A model could reveal how sensitive the response is to variation in each parameter.

I have identified laboratory conditions that strongly potentiate Ca²⁺ signals in the cilium. Those conditions cannot perfectly mimic an undisturbed cell. I recorded at 24°C. At 37°C, the native Ca²⁺ buffers are probably somewhat more effective. The apparent association constant for BAPTA with Ca²⁺, for example, increases by a factor of 1.28 at 37°C, as estimated by the method of Marks and Maxfield (1991). It is also possible that the recording procedure itself changes the structure of the cilium. Learning how well the natural conditions support regenerative Ca²⁺ signals will be challenging. Experimental evidence is sparse. In neuronal models, the concentrations of mobile and immobile buffers are typically estimated to be below 100 µM (Bartol et al., 2015; Ohadi and Rangamani 2019), which in the present study was low enough to allow regenerative signals. The endogenous buffers in those models (e.g. calbindin) have equilibrium association constants with Ca²⁺ that are similar to that of BAPTA or lower (Bartol et al., 2015). In cells of the retinal pigment epithelium, resting intraciliary Ca²⁺ is 580–742 nM (Delling et al., 2013), which may be enough to weakly activate PC2 channels even with strong Ca²⁺ buffering (Kleene and Kleene 2017; Liu et al., 2018). Resting Ca²⁺ is higher in the cilium than in the cytoplasm (107 nM; Delling et al., 2013), suggesting that ciliary channels support some Ca²⁺ entry even at rest. While it seems possible that a small initial stimulus could generate a powerful Ca²⁺ signal in the cilium, evidence for this in intact cells is not yet available. A PC2 agonist increases intraciliary Ca²⁺ in mIMCD-3 cells, but the concentration reached was not determined (Ha et al., 2020).

For convenience, the channels studied here are described as PC2 and TRPV4. However, the subunit compositions of these channels are uncertain. The large-conductance channel, referred to here as PC2, has long been suggested to be a heteromultimer of PC2 and a related protein, polycystin-1 (PC1; reviewed in Douguet et al., 2019), although PC2-dependent channels similar to the native channel persist in cells lacking PC1 (Liu et al., 2018). Circumstantial evidence suggests that the native ciliary PC2 channel may also include TRPM3. The native channel requires expression of both PC2 and TRPM3; in the absence of TRPM3, PC2 traffics to the cilium but functional channels are not detectable (Kleene et al., 2019). Furthermore, the native ciliary channel is activated by pregnenolone sulfate, which also activates TRPM3, and inhibited by isosakuranetin, a specific inhibitor of TRPM3 (Kleene et al., 2019).

The channel referred to here as TRPV4 is activated by GSK1016790A, a TRPV4 agonist (Willette et al., 2008), and inhibited by the selective TRPV4 antagonist HC-067047 (Willette et al., 2008; Everaerts et al., 2010; Siroky et al., 2017). The presence of TRPV4 in renal primary cilia is established (Köttgen et al., 2008; Du et al., 2012; Zhang et al., 2013; Lee et al., 2015; Siroky et al., 2017; Kohli et al., 2017). In the presence of extracellular Ca²⁺, the complex rectification of the current-voltage relation for the ciliary current activated by GSK1016790A (Figure 7C, gray) is the same as that described for TRPV4 (Voets et al., 2002). Exogenously expressed PC2 and TRPV4 can form a heterotetrameric channel with a 2:2 stoichiometry (Stewart et al., 2010). In the plasma membrane of renal epithelial cells, TRPV4 can interact with PC2 (Köttgen et al., 2008; Du et al., 2012; Zhang et al., 2013) and form channels with a conductance of 23 pS (Zhang et al., 2013; Saigusa et al., 2019). In primary cultures from human kidneys, GSK1016790A activates a channel of ∼20 pS at negative potentials (Tomilin et al., 2018). However, channels with a similar conductance were not seen in this study. Instead, the TRPV4 agonist activated channels with conductances of 39 pS at negative potentials and 90 pS at positive potentials (Figure 7C). These are typical values for TRPV4 channels (30–60 pS at negative potentials, 80–100 pS at positive potentials; Nilius et al., 2004; Garcia-Elias et al., 2014). TRPV4 appears to play several roles in renal function (reviewed in Pochynyuk et al., 2013).

It has been suggested that Ca²⁺ fluxes through ciliary ion channels, including PC2, may function to prevent cystogenesis in healthy renal epithelial cells (Jin et al., 2014b; Chebib et al., 2015; Wang et al., 2018). In the cell as a whole, Ca²⁺ and cAMP influence the cellular proliferation and fluid secretion that characterize PKD (reviewed in Calvet 2015). cAMP inhibits cystogenesis in normal cells but promotes it in cystic cells (Hanaoka and Guggino 2000; Yamaguchi et al., 2000). In cultured human renal cells from ADPKD patients, expression of adenylate cyclases AC5 and AC6 is increased, as is the basal cAMP level (Pinto et al., 2012). In mice with PC1 or PC2 knocked out in the kidney, concurrent elimination of AC6 or AC5 (respectively) reduces cystogenesis (Rees et al., 2014; Wang et al., 2018). Cystic cells often have depressed cytoplasmic Ca²⁺ (Yamaguchi et al., 2006; Zaika et al., 2013; Jin et al., 2014b; Tomilin et al., 2018; but see also; Cabrita et al., 2021), and increasing this Ca²⁺ can restore a normal response to cAMP (Yamaguchi et al., 2006; Gradilone et al., 2010; Zaika et al., 2013). Within the cilium, Ca²⁺ may influence the level of cAMP (Wachten and Mick 2021), perhaps by inhibiting AC5 and AC6. (On exogenous expression, each of these shows steadily decreasing activity as internal Ca²⁺ is increased from 0.1 µM to 1 mM (Guillou et al., 1999; Hu et al., 2002).) The primary cilia of renal epithelial cells express AC5 and AC6 (Raychowdhury et al., 2009; Chien et al., 2010; Choi et al., 2011; Mick et al., 2015; Wang et al., 2018), and the cyclases interact in the cilium with PC2 (Choi et al., 2011; Wang et al., 2018). One can thus propose that Ca²⁺ entering through PC2 channels might suppress the cyclases and thus the synthesis of cAMP (Wang et al., 2018). Consistent with this, deletion of PC2 increased forskolin-stimulated intracellular cAMP levels. This was corrected by reexpression of wild-type PC2 but not by a mutant lacking Ca²⁺ channel activity (Choi et al., 2011). Although the relevant proteins form a complex in the cilium, their effects on cystogenesis are not known to reside there. It is known that cAMP in the cilium has functions distinct from those in the cytoplasm (Sherpa et al., 2019; Truong et al., 2021).

It also remains to learn what stimuli might initiate an increase in intraciliary Ca²⁺, perhaps regenerative, in vivo. Such an increase has been reported on deflection of the cilium (Su et al., 2013; Jin et al., 2014a). However, it is debated whether a ciliary Ca²⁺ increase is mediated by ciliary channels or by backflow of Ca²⁺ from the cell body into the cilium (Delling et al., 2016). In mIMCD-3 cells, a soluble N-terminal fragment of PC1 activates PC2 channels and initiates an influx of Ca²⁺ into the primary cilium (Ha et al., 2020). Also in mIMCD-3 cells, pregnenolone sulfate, a natural metabolite, activates PC2 channels but at much higher concentrations than are present in vivo (Kleene et al., 2019).

An unexplored possibility is that TRPM3 and TRPV4 may be osmotically activated in the cilium. TRPM3 and TRPV4 are typically activated by hypoosmolality (Oberwinkler and Philipp 2014; Toft-Bertelsen and MacAulay 2021), although this has not been demonstrated in the cilium. TRPM3, TRPV4, and the primary cilium are all required for a normal osmotic response in renal epithelial cells, and that function of TRPM3 appears to reside in the cilium (Siroky et al., 2017). Activation of TRPM3 and TRPV4 could be beneficial in PKD. In a cell line derived from the cortical collecting duct, TRPV4 mediates a Ca²⁺ influx in response to hypotonicity (Wu et al., 2007). Reducing the osmolality of the renal filtrate that bathes the cilia might osmotically activate TRPM3 and TRPV4 in the cilium. This could in turn increase ciliary Ca²⁺, which could then inhibit the ciliary cyclases, reduce ciliary cAMP, and slow the progression of PKD. The increased ciliary Ca²⁺ could secondarily activate PC2, but such a model does not explain why the channel function of PC2 in particular is needed to prevent cystogenesis. In a model of autosomal recessive PKD, increased water intake, which can reduce the osmolality of the urine, did slow the progression of the cystic disease (Nagao et al., 2006). Increased water intake has long been discussed as a potential therapy for PKD (Nagao et al., 2006; Torres et al., 2009; Wang et al., 2013) and has been the basis of multiple clinical trials. If such trials are effective, it will be useful to evaluate the roles, if any, of ciliary osmosensitive TRP channels. To date, there is no evidence that defects in TRPM3 or TRPV4 impact renal function. Defects in TRPV4 do not lead to cyst formation in mouse or zebrafish (Köttgen et al., 2008), and existing studies of mice lacking TRPM3 (Vriens et al., 2011; Hughes et al., 2012; Brown et al., 2015; Vandewauw et al., 2018) have not commented on renal function. It is conceivable that TRPM3 and TRPV4 comprise a redundant, fail-safe system such that elimination of either channel can be compensated by the remaining channel.