Intracellular calcium strongly potentiates agonist-activated TRPC5 channels

Transient receptor potential (TRP) channels are a superfamily of ∼28 different proteins present in mammals (Ramsey et al., 2006; Venkatachalam and Montell, 2007). TRP channels have varied expression patterns and biophysical properties, and they have been implicated in a wide range of physiological responses, including sensory transduction pathways involved in temperature, pressure, light, taste, and painful chemical stimuli.

The seven members of the canonical subfamily of TRP channels (TRPC) are most similar to the Drosophila TRP channel required in fly vision. All of the TRPC subfamily members are nonselective cation channels with modest Ca²⁺ permeability that, at present, appear to be activated by both G protein–coupled receptors (GPCRs) and growth factor receptors coupled to PLC. The TRPC subfamily can be further divided into TRPC1/C4/C5 and TRPC3/C6/C7. These can form homotetramers, or in some cases heterotetramers, between subfamily members (Strubing et al., 2001, 2003; Hofmann et al., 2002). For the TRPC3/C6/C7 subfamily, activation may be mediated, at least in part, by diacylglycerol released by PLC hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP₂) into diacylglycerol and inositol tris (1,4,5) phosphate (IP₃). For the TRPC1/C4/C5 subfamily of channels, the intermediates between receptor, PLC activation, and channel activation are not fully understood.

Early work established that in hippocampal neurons, TRPC1/5 heteromers and TRPC5 homomers were expressed in neuronal cell bodies, whereas only TRPC5 homomers were trafficked to neuronal processes (Greka et al., 2003; Bezzerides et al., 2004). We have found no evidence for functional TRPC1 homomeric channels in heterologous expression systems (Strubing et al., 2001, 2003), and the evidence that they exist in native cells is conflicting (Kim et al., 2003; Hartmann et al., 2008). In contrast, TRPC1/4, TRPC1/5, and TRPC4 and TRPC5 homomeric channels clearly exist in both heterologous expression and native tissues.

Modulatory influences on the activity of the TRPC1/C4/C5 subfamily channels have been described, typically for TRPC4 or TRPC5 homomeric channels. These include inhibition of TRPC4 (α isoform) channels by PIP₂ (Otsuguro et al., 2008), inhibition of TRPC5 channels by nucleotides (Dattilo et al., 2008), desensitization of TRPC5 channels by PKC phosphorylation (Zhu et al., 2005), and interaction of both TRPC4 and TRPC5 channels with Ca²⁺ binding proteins. For example, calmodulin has been reported to accelerate TRPC5 agonist-activated current (Ordaz et al., 2005), presumably via a direct interaction with the channel; upstream effects via myosin light chain kinase activation have also been reported (Kim et al., 2006; Shimizu et al., 2006). Furthermore, Ca²⁺ binding protein 1 has been reported to inhibit intracellular Ca²⁺ activation of TRPC5 current (Kinoshita-Kawada et al., 2005), whereas neuronal Ca²⁺ sensor 1 (NCS-1) appears to have permissive or stimulatory actions (Hui et al., 2006).

The multitude of Ca²⁺-dependent effects, causing both stimulation and inhibition, is likely crucial for control of TRPC4 and TRPC5 activity. Thus, a detailed knowledge of the effects of Ca²⁺ on the activity of these channels is important in understanding their potential physiological roles where they are expressed. This includes smooth muscle cells, neuroendocrine cells, and numerous neuronal cell types (Plant and Schaefer, 2005). Here, we focus on TRPC5 homomeric channel currents. We examined the regulation of heterologously expressed TRPC5 activity by intracellular Ca²⁺ using both whole cell and single-channel patch clamp recordings. We find that ∼1 µM intracellular [Ca²⁺] can lead to a 25-fold increase in agonist-activated TRPC5 current at physiological voltages, likely resulting from an increase in channel open probability.

Human embryonic kidney (HEK) cells were cultured in a 1:1 mixture of DMEM/F-12 medium supplemented with 10% fetal bovine serum at 37°C with 5% CO₂ (Invitrogen). Cells were passaged 6–24 h before transfection and transfected in 35-mm dishes using 5 µl Lipofectamine 2000 (Invitrogen) with 1–1.5 µg of mouse TRPC5 (in pCI-neo) or mouse TRPC5-GFP (in pIRES, after excision of internal ribosomal entry site to generate a C-terminal fusion of TRPC5 and GFP [Bezzerides et al., 2004]) and 0.8 µg of human muscarinic type 1 receptor (M1R; in pcDNA3). 0.05–0.1 µg eGFP-C1 or 0.25 µg DsRed-N1 (for fura-2 imaging experiments) was used as a transfection marker (Clontech Laboratories, Inc.). In experiments where mutant calmodulin with all four E-F hands disrupted (CaM1234) was used (Fig. 14), 1.64 µg CaM1234/pcDNA3zeo was included with the above plasmids. After 2–4 h of transfection, fresh media was added. 12–24 h later, the cells were plated onto glass coverslips. Recordings were made up to 8 h after plating. For inside-out single-channel recordings, coverslips were coated with 0.1 mg/ml poly-l-lysine to improve cell adhesion and subsequent patch excision.

Whole cell and single-channel currents were recorded using Axopatch 200A and 200B amplifiers, acquired and controlled using a Digidata 1320A and pClamp 9 (MDS Analytical Technologies). Analysis was typically performed in IGOR Pro 5 (Wavemetrics) and Prism 4 (Graphpad). Single-channel analysis was performed in pClamp 9. Pipettes were pulled from borosilicate glass capillaries (WPI or A-M Systems), had resistances of 2–4 MΩ for whole cell recordings and 4–7 MΩ for single-channel recordings, and were wrapped with parafilm near their tips to reduce capacitance. Whole cell currents were filtered at 5 kHz (4-pole low-pass Bessel) and sampled at 20 kHz. During whole cell recordings the capacity current was minimized by amplifier circuitry, and the series resistance was compensated by 85%. We found that the strongly outwardly rectifying TRPC5 current was sensitive to the series resistance remaining after compensation. Thus, outward currents were analyzed only in cells where the initial uncompensated series resistance was <10 MΩ. To calculate the current density, cell capacitance was either measured from the uncompensated response to −10-mV pulses (average of 8–14 sweeps acquired at 50 at 10 kHz bandwidth) or from the capacitance value on the amplifier after compensation. All voltages were corrected for the −11- or −13-mV measured junction potentials present between the internal solutions and the 2 Ca external solution present before seal formation. Experiments were performed at room temperature (21–24°C).

TRPC5 currents were elicited by voltage ramps consisting of a 40-ms step to −100 mV, followed by a 200-ms ramp to +100 mV, 40 ms at +100 mV, and finally a 20–40-ms step to −60 mV. Voltage ramps were applied at 0.5 Hz from a holding potential of −40 mV. Inward currents were the average of 25 ms during the holding potential preceding each ramp; we chose this potential to minimize the contribution of a contaminating Cl⁻ current (reversal potential ∼−53 mV), which was occasionally recorded in cells using high internal [Ca²⁺]. Due to constitutive TRPC5 activity, currents were not typically leak subtracted; however, constitutive current was absent in recordings made using internal solutions with free <350 nM [Ca²⁺] in nominally Ca-free external solutions without agonist. In these cases (i.e., Figs. 6, 12, and 13), the inward current was leak subtracted with the average of four to eight sweeps, recorded in nominally Ca²⁺-free external solution before carbachol (CCh) addition.

To examine possible changes in instantaneous I-V relation for TRPC5 channels during potentiation by Ca²⁺ (Fig. 9), we used two protocols that gave similar results. In both, voltage steps to a conditioning potential of either +90 (for 50 ms) or +100 mV (for 15 ms) activated TRPC5 channels, and then tail currents were measured from 0.7 to 0.85 ms after repolarization to potentials ranging from −120 through +100 mV. In one set of experiments, repolarization was to −120, −90, −60, −30, and +60 mV for 40 ms. These were applied in a single sweep immediately after one another; sweep to sweep onset was 1 Hz. In another set of experiments, repolarization was to −100 to +100 mV (in 5-mV increments) for 50 ms; sweep to sweep onset was 2 Hz.

The standard Cs-Asp internal solution contained (in mM): 150 Cs-aspartate, 2 MgCl₂, 0.36 CaCl₂, 1 EGTA, 4 MgATP, 0.3 NaGTP, and 10 HEPES, pH 7.20 with CsOH (calculated free [Ca²⁺] = 100 nM). In several experiments, we varied the amount of CaCl₂ and buffer added to internal solutions to set free [Ca²⁺] to different values; the precise components of each internal solution are noted in the figure legends. The free [Ca²⁺] for all internals was calculated using WebMaxChelator (http://www.stanford.edu/∼cpatton/maxc.html), assuming an ionic strength of 0.16. The standard external solution (“2 Ca external”) contained (in mM): 150 NaCl, 4 KCl, 2 CaCl₂, 1 MgCl₂, 10 HEPES, and 10 d-glucose, pH 7.40, with NaOH. 2 Ba external solution was identical, except 2 mM BaCl₂ replaced the CaCl₂. “Nominally Ca-free” external solution contained (in mM): 150 NaCl, 4 KCl, 3 MgCl₂, 10 HEPES, and 10 d-glucose, pH 7.40, with NaOH. In some experiments (e.g., Fig. 12), the concentrations of divalent cations Ca²⁺ and Mg²⁺ were varied; again, the solution composition is noted in the figure legend.

Single-channel currents were filtered with a 4-pole low-pass Bessel filter at 2 kHz and sampled at 50 kHz. Recordings were converted to idealized traces using pClamp 9 software, including only events that were >0.5 times the mean amplitude. We assumed a dead time of 0.25 ms. TRPC5 channel openings had single-channel conductances of ∼40–50 pS; in some excised patches we observed a smaller amplitude opening of 17.6 ± 1.1 pS (n = 5). We omitted these lower conductance channel-containing patches from analyses.

DsRed-positive cells were loaded with 100 µM fura-2 (pentapotassium salt) via the patch pipette. Cells were held at −40 mV, and the typical voltage ramp protocol was applied at 0.2 Hz. Immediately before each voltage ramp, cells were alternatively illuminated with 340 and 380 nm light (Lambda DG-4; Sutter Instrument Co.) for 100–200 ms, and the emission light >510 nm was captured using a CCD camera (Orca-ER; Hamamatsu Photonics) and analyzed with Slidebook (Intelligent Imaging Innovations) or Metamorph (MDS Analytical Technologies). After background subtraction, the ratio of the 340- and 380-nm excited images was calculated and converted to free [Ca²⁺] using: [Ca²⁺] = K*(R_max−R)/(R−R_min), where K* = 2.7536 µM, R_max = 12.4, and R_min = 0.3. These values were calculated from average values measured using internal solutions with varying [Ca²⁺] in untransfected HEK cells as described by Neher (2005).

BAPTA tetracesium salt and fura-2 pentapotassium salt were purchased from Invitrogen. Calmodulin inhibitory peptide and W-7 were from EMD. Stock solutions were prepared in water (calmodulin inhibitory peptide) or DMSO (W-7) and diluted fresh daily into the internal solution at the indicated concentrations. All other reagents were from Sigma-Aldrich.

Averaged results are presented as mean ± SEM. Population means were tested for significance with Student's two-tailed t test, with P < 0.05 considered significant.

Fig. 1 shows the typical currents evoked by voltage ramps in a HEK cell transiently transfected with mouse TRPC5 and the human M1R, recorded using internal solutions with modest Ca²⁺ buffering (1 mM EGTA). The initial constitutive currents were quite small, with an average current density at −40 mV of −1.4 ± 0.2 pA/pF (n = 30), but outwardly rectifying currents at +100 mV (20.0 ± 2.6 pA/pF) indicated the presence of TRPC5 channels. Upon the addition of 100 µM CCh to activate M1Rs, a current with the characteristic TRPC5 I-V relationship immediately began to develop; the inward current was roughly constant from −100 to −50 mV and declined as the reversal potential (∼−5 mV in our solutions) was approached. At potentials positive to the reversal potential, the current was small and flat until +40 mV, whereupon the current became large and relatively linear (Okada et al., 1998; Schaefer et al., 2000; Strubing et al., 2001).

The TRPC5 current amplitudes steadily increased over ∼60 s after CCh application, until the rate of current change dramatically accelerated, reaching a plateau at +100 mV (8.7-fold increase in current). The inward current at −40 mV increased to an even larger degree (17.9-fold), reaching its new peak 10 s later than the +100-mV peak. Despite the large current increase, the reversal potential did not change and the flat portion of the current voltage from 0 to +40 mV remained.

The different kinetics of current increase at +100 and −40 mV are more obvious when the absolute value of the ratio between these currents is calculated (Fig. 1 B). This ratio initially increased sharply soon after CCh addition, as the outward current developed before the inward. After ∼30 s it reached a new value where the current at +100 mV was ∼16-fold larger than that at −40 mV. However, as the CCh application continued, the rectification ratio steadily increased until a rapid increase led to a peak where the outward current was ∼67-fold larger than the inward. This large asymmetry in the inward and outward currents was maintained for only a short time, as over the next 14 s the ratio declined to ∼7.3, lower than that achieved before the rapid increase in current level.

This biphasic activation, with a delayed, sudden increase in TRPC5 current, was observed in 18 of 30 cells recorded with an internal solution containing 100 nM of calculated free [Ca²⁺] and 1 mM EGTA. These cells had peak current densities at −40 mV ranging from −39 to −512 pA/pF (mean −153.0 ± 27.2 pA/pF). The latency from CCh addition to the peak current in these 18 cells varied from 10 to 88 s, with some cells showing a nearly immediate current increase, whereas in others it developed more slowly (average 38.3 ± 5.5 s; Fig. 1 C). The current at +100 mV consistently reached its new peak 2–10 s before the peak current measured at −40 mV. In the remaining 12 cells, the TRPC5 current activated with an approximately single-exponential time course before reaching steady state at −5.1 ± 0.8 pA/pF at −40 mV (−10.6 pA/pF maximum) for the duration of CCh application (typically 60–90 s). A similar delayed increase in TRPC5 currents activated by the histamine type 1 receptor or internal GTP-γ-S has been reported by Obukhov and Nowycky (2004, 2008), suggesting that the response is not unique to the M1R.

TRPC5 channels are Ca²⁺ permeable, with reported relative permeabilities of Ca²⁺ to Na⁺ (P_Ca/P_Na) ranging from 1 to 14 (Plant and Schaefer, 2005). Thus, the large amplitude currents we recorded suggest that significant Ca²⁺ entry occurred. This Ca²⁺ entry might escape the buffering of the low concentrations of EGTA we used in our internal solution for Fig. 1. Thus, the delayed increase in TRPC5 current could reflect a Ca²⁺ dependence of TRPC5 channels themselves or other components of the agonist activation pathway. If this increase is Ca²⁺ dependent, increasing the buffering capacity of the internal solution might slow or prevent the TRPC5 current potentiation. Fig. 2 A shows the current densities of eight cells measured at −40 mV, recorded using Cs-Asp internal solution with calculated free [Ca²⁺] maintained at 100 nM with 10 mM EGTA. Four of these cells had peak current densities at −40 mV of −14.2 pA/pF or less (−10.9 ± 2.1 pA/pF), similar to cells in Fig. 1 C that did not undergo potentiation. The activation of the current in these four cells followed a similar single-exponential time course. In contrast, the currents at −40 mV from the remaining four cells reached peak densities ranging from −46.8 to −498.6 pA/pF (−244.3 ± 95.1 pA/pF), similar to the cells that underwent potentiation in Fig. 1 C. The latency to peak in these cells was increased with 10 mM EGTA in the internal solution and ranged from 56 to 168 s after CCh application. This increase presumably resulted from the requirement for more Ca²⁺ entry before escape from Ca²⁺ buffer control.

BAPTA and EGTA have similar K_Ds for Ca²⁺ binding, but BAPTA buffers ∼200-fold more rapidly (Naraghi, 1997). Thus, BAPTA can more effectively reduce [Ca²⁺] at small time and spatial scales near Ca²⁺ entry points (Neher, 1998). When TRPC5/M1R-expressing cells were recorded using a Cs-Asp internal solution (100 nM of calculated free [Ca²⁺] now buffered with 10 mM BAPTA), we found that the current amplitudes were small and no delayed current potentiation occurred. Fig. 2 B shows the current densities at −40 mV for nine cells recorded with this internal solution: currents activated with a single-exponential time course that reached an average peak amplitude of −6.3 ± 1.4 pA/pF. This result suggests that tight control of [Ca²⁺] close to Ca²⁺ entry sources by BAPTA can effectively prevent TRPC5 potentiation.

BAPTA itself might limit agonist-dependent signaling by inhibiting PLC (Hardie, 2005). To determine whether the reduction of TRPC5 current by high internal BAPTA results from this inhibition or from its fast Ca²⁺ buffering, we compared the responses in Fig. 2 B to recordings of TRPC5/M1R-expressing cells (1 mM EGTA in the internal solution) when the CCh concentration was reduced to 10 µM (not depicted). Previous work demonstrated that heterologously expressed M1Rs activated with 10 µM CCh should result in ∼40–50% less PIP₂ hydrolysis than with 100 µM CCh (Biddlecome et al., 1996; Willars et al., 1998). Out of 20 cells recorded in this way, 10 displayed biphasic activation with large current amplitudes (average current density at −40 mV = −140.4 ± 28.6 pA/pF), whereas the remaining cells had small current amplitudes (average of −5.7 ± 0.6 pA/pF; n = 10). Thus, despite a lower level of agonist stimulation at M1R receptors, a large fraction of TRPC5/M1R-expressing cells can still undergo potentiation with low internal Ca²⁺ buffering. This suggests that the suppression of potentiation by 10 mM BAPTA is likely the result of its ability to rapidly buffer Ca²⁺ entering the cell.

When the BAPTA concentration in the internal solution was reduced to ∼1 mM and Ca²⁺ was added to give 100 nM of calculated free [Ca²⁺], the resulting TRPC5 current amplitudes were similar to those recorded with 1 mM EGTA (Fig. 2 C). In nine cells recorded with ∼1 mM BAPTA in the internal (three with 1 mM BAPTA and six with 0.99 mM BAPTA + 100 µM fura-2), seven cells had peak current densities at −40 mV from −31.9 to −417 pA/pF (−116 ± 52.3 pA/pF), with latencies from 20 to 270 s. The two other cells had small CCh-elicited currents after reaching steady state (−2.0 and −2.3 pA/pF). The long latencies in two cells (220 and 270 s) possibly resulted from the slower application of ramp voltages used in BAPTA/fura-2 experiments (0.2 Hz vs. the 0.5 Hz used in all other experiments). Ca²⁺ entry might be enhanced during the hyperpolarizing portion of the voltage ramp, and slower application of ramp stimuli likely resulted in a slower intracellular [Ca²⁺] elevation.

The fact that increasing the Ca²⁺ buffering capacity of the internal solution slows or prevents the sudden potentiation of the TRPC5 current suggests that potentiation requires an increase in intracellular [Ca²⁺]. In these HEK cells, Ca²⁺ presumably enters through TRPC5 channels, but it might also be released from endoplasmic reticular stores in response to IP₃. To determine if Ca²⁺ entry is required for the potentiation of TRPC5 channels, we tested whether the removal of external Ca²⁺ can prevent or reverse the delayed increase in CCh-activated TRPC5 current (Fig. 3). For these experiments, we used a Cs-Asp internal solution with 100 nM of calculated free [Ca²⁺] using 1 mM BAPTA. As shown for the cell in Fig. 3 A, voltage ramps applied before the addition of CCh elicited a small constitutive TRPC5 current that was strongly reduced after the external solution was switched to nominally Ca-free external solution with 3 mM MgCl₂ (51% reduction in current at +100 mV; 86% reduction in current at −40 mV). This reduction was complete within the 2 s between applied voltage ramps and likely resulted from a decrease in the potentiation of TRPC5 channels due to extracellular Ca²⁺ binding at a site in the presumed S5-pore linker region of the channel (Jung et al., 2003). This site is formed by a series of acidic residues and binds not only Ca²⁺ but also lanthanides (lanthanum and gadolinium) to increase channel open probability in a voltage-dependent manner. Constitutive TRPC5 current in the absence of agonist might result from constitutive activity of the overexpressed M1Rs (e.g., recruitment of PLC by Gβγ despite low Ca²⁺ concentrations [Drin and Scarlata, 2007]) or of PLC itself.

The application of 100 µM CCh in nominally Ca-free external solution to the cell shown in Fig. 3 (A–C) activated TRPC5 current with a single-exponential time course, reaching a steady-state amplitude of 2,100 pA at +100 mV and −22 pA at −40 mV. Over the course of the 60-s CCh application, the TRPC5 current remained constant and never exhibited spontaneous potentiation. When the extracellular solution was switched to 2 Ca external solution (still with 100 µM CCh), both the inward and outward current immediately increased, again, likely due to potentiation at the extracellular Ca²⁺/lanthanide site (Fig. 3 B). Over the course of the next 28 s, the current further increased, reaching a plateau that was largely maintained while extracellular Ca²⁺ was present. At this point, the peak current at +100 mV was approximately eightfold larger relative to the current recorded immediately after Ca²⁺ readdition; the inward current at −40 mV was ∼18-fold larger. Removal of external Ca²⁺ reversed the current potentiation, with the current immediately falling after Ca²⁺ removal, and continuing to decline over the next ∼90 s. The TRPC5 current finally reached a new steady-state amplitude that was smaller than the initial amplitude of the CCh-activated current in nominally Ca-free external solution. This reduction might result from a desensitization of the overall response, arising either from reductions in the activity of TRPC5 channels themselves or other components of the signaling pathway (e.g., receptors, heterotrimeric G proteins, or PLC). A subsequent readdition of extracellular Ca²⁺ (still with CCh) in this cell generated another similar response, with an immediate increase in TRPC5 current amplitude followed by a delayed peak. Although the peak currents after the first external Ca²⁺ addition were 8–18-fold larger than currents recorded immediately after the return to 2 Ca external solution, they retained the reversal potential and flat portion of the TRPC5 I-V relation from 0 to +40 mV (Fig. 3 C). Also, removal of external Ca²⁺ after CCh activation of TRPC5 did not cause a change in the reversal potential, strongly suggesting that the effects occur by altering the activity of TRPC5 channels and not unrelated conductances.

We quantified the extent of Ca²⁺ potentiation of TRPC5 current as the peak amplitude in 2 Ca external solution relative to the amplitude immediately after Ca²⁺ readdition. From eight TRPC5/M1R-expressing cells recorded in this way (i.e., as in Fig. 3 A), the average potentiation of the current at +100 mV was 4.75 ± 0.62-fold. In contrast, the inward current at −40 mV was potentiated to a much greater extent, with the peak current 24.0 ± 8.5-fold larger. Replacement of external Ca²⁺ with Ba²⁺ also led to a reversal of TRPC5 current potentiation (e.g., Fig. 9 B), similar to Mg²⁺ replacement. The greater potentiation of inward current relative to outward current resulted in a more linear I-V relation (the flat portion positive to the reversal potential was maintained). A similar finding was recently reported by Obukhov and Nowycky (2008).

Although 18 of 30 TRPC5/M1R-expressing cells recorded in weakly buffered internal Ca²⁺ underwent potentiation, the 12 cells that did not undergo potentiation might have had lower expression levels that resulted in insufficient Ca²⁺ entry. Alternatively, modulatory processes might have prevented the ability of TRPC5 channels in this subset of cells from responding to any increase in intracellular [Ca²⁺]. To distinguish between these alternatives, we increased the calcium in the external solution to 10 mM for cells that failed to undergo potentiation after initial activation by 100 µM CCh in 2 Ca external solution. As shown in Fig. 4 (A and B), application of 10 Ca external solution rapidly, but not immediately, induced potentiation. The current increased over 8 s to a level approximately sevenfold greater than in 2 Ca external solution, and this level was maintained. Interestingly, returning cells to 2 Ca external did not abolish the potentiation, suggesting that the increased TRPC5 current after potentiation might have been sufficient to maintain high intracellular [Ca²⁺] and potentiation.

In 13 of 15 TRPC5/M1R-expressing cells treated in this manner, we observed potentiation of the current after extracellular perfusion of 10 Ca external solution and CCh (Fig. 4 C). The average current density at −40 mV in 2 Ca external solution was −7.3 ± 1.2 pA/pF, which increased to a peak of −70.1 ± 10.6 pA/pF (n = 13) after external [Ca²⁺] was increased to 10 mM. The average potentiation of the current at −40 mV was 14.6 ± 2.7-fold (n = 13; Fig. 4 D). As shown in Fig. 4 C, the latency to peak current after the addition of 10 Ca external solution was bimodal: in 7 of the 13 cells the peak occurred 8.3 ± 1.0 s (n = 7) after 10 mM Ca²⁺ application, whereas in the remaining 6 cells the peak occurred with a significantly longer latency of 44.0 ± 4.7 s. Both values are longer than the solution exchange time and presumably reflect the time course of the intracellular [Ca²⁺] rise.

We tested whether a different Gαq/11-coupled receptor can lead to potentiation of TRPC5 currents. Coexpression of the mGluR1 receptor with TRPC5 channels and activation by 100 µM (RS)-3,5-dihydroxyphenylglycine alone did not result in potentiation in five tested cells, likely due to the resulting small currents (−1.35 ± 0.28 pA/pF at −40 mV, not depicted). However, subsequent application of 10 Ca external solution potentiated TRPC5 currents in four of the five TRPC5/mGluR1-expressing HEK cells, on average 22.6 ± 4.3-fold (the remaining cell current did not change). The latency to peak was much longer in these cells (101.8 ± 31.7 s; n = 4) than in the M1R-expressing cells after the addition of 10 Ca external. The reasons for such large differences in current activation by the two receptors are unclear, but they might reflect differing expression levels or receptor-dependent modulatory effects. However, Ca²⁺ potentiation can occur in response to the activation of multiple Gαq/11-coupled receptors.

Our results indicate that a rise in intracellular [Ca²⁺] is required to potentiate TRPC5 current, but its detailed concentration dependence is unclear. Overexpression of TRPC5 channels might lead to non-physiological [Ca²⁺] after the activation of M1Rs. To determine the range of intracellular [Ca²⁺] required for TRPC5 potentiation, we recorded TRPC5 currents and monitored intracellular [Ca²⁺] using 100 µM fura-2 added to the internal solution (∼100 nM of calculated free [Ca²⁺] with 0.99 mM BAPTA). Fig. 5 shows the currents and the calculated [Ca²⁺] recorded from one such TRPC5/M1R-expressing cell after the application of 100 µM CCh. Both the TRPC5 current at −40 mV and intracellular [Ca²⁺] initially increased slowly, but after 150 s in CCh the rise in the current and intracellular [Ca²⁺] accelerated. At the point where the time derivative (dI/dt) of the current at −40 mV reached 50% of its maximum, the calculated free [Ca²⁺] was 945 nM (in this cell). As the CCh application continued, TRPC5 current and [Ca²⁺] increased further until both reached a plateau. The fura-2 signal indicated that [Ca²⁺] was ∼4 µM at this point, although with its high affinity (K_D ∼250 nM) the fura-2 was very likely saturated, and real intracellular [Ca²⁺] might have been even higher. It is clear that the 0.99 mM BAPTA present in the internal solution is unable to adequately buffer Ca²⁺ entering through TRPC5 channels.

From six cells where TRPC5 currents and intracellular [Ca²⁺] were simultaneously measured, five cells underwent potentiation of the current (Fig. 5 C). In those cells, the intracellular [Ca²⁺] at 50% maximal dI/dt ranged from 611 to 1,802 nM (mean 1,117 ± 222 nM). Cells in which the 50% maximal TRPC5 current dI/dt occurred at lower [Ca²⁺] tended to reach higher absolute dI/dt values and have shorter latencies compared with cells where the current and intracellular [Ca²⁺] developed slowly. The variation in calcium sensitivity is visible when the current densities at −40 mV, from the initial application of CCh until the peak, are normalized to their own maximum and plotted against the recorded intracellular [Ca²⁺] (Fig. 5 D). Despite latencies varying from 48 to 216 s in these cells, the TRPC5 current increased in a graded manner with intracellular [Ca²⁺]. One of the five cells had a very high initial [Ca²⁺] of ∼1,500 nM and significant basal current; the 50% maximal dI/dt point occurred at 1,802 nM.

In the one cell that did not undergo potentiation, TRPC5 current activated with a single-exponential time course, and the intracellular [Ca²⁺] reached a transient peak of 615 nM, before declining to ∼360 nM over 125 s. The current in this cell also declined, displaying an unusually fast desensitization that was rarely seen in other cells.

The fura-2 imaging experiments above indicate that intracellular [Ca²⁺] over ∼600 nM enables the potentiation of TRPC5 channels; however, [Ca²⁺] was constantly changing in those cells. Any delays between the [Ca²⁺] elevation and channel potentiation might lead to an underestimate of the Ca²⁺ sensitivity for potentiation. To confirm that ∼1 µM intracellular [Ca²⁺] causes potentiation of TRPC5 channels, we recorded CCh-activated currents from TRPC5/M1R-expressing HEK cells using internal solutions where calculated free [Ca²⁺] was buffered to levels varying from 100 nM to 17 µM. These experiments were performed in nominally Ca-free external solution to prevent escape from control of the internal Ca²⁺ buffer, as was evident in Fig. 5.

Fig. 6 A shows the CCh-activated TRPC5 current recorded from cells using a Cs-Asp internal solution with 350 nM of calculated free [Ca²⁺] buffered with 10 mM BAPTA, or with 1.8 µM of calculated free [Ca²⁺] buffered with 5 mM HEDTA. The application of 100 µM CCh in both cases activated a current during voltage ramps that had the typical TRPC5 shape: outwardly rectifying, with reversal near 0 mV and a flat region between 0 and +40 mV. The activation of the current followed a single-exponential time course and was complete within ∼30 s; after this point, the current maintained its amplitude or declined slightly as the agonist application continued. However, the amplitude of the current density at −40 mV recorded with 350 nM of calculated free Ca²⁺ was −1.23 pA/pF, whereas with 1.8 µM of internal [Ca²⁺] it was >100-fold larger, at −131.8 pA/pF.

Fig. 6 B summarizes the average CCh-activated current densities at −40 mV for TRPC5/M1R-expressing cells recorded with a range of internal [Ca²⁺]. The TRPC5 current amplitude has a sharp dependence on calculated free [Ca²⁺] in the internal solution and was essentially maximal when internal [Ca²⁺] was >1 µM, similar to the result from the fura-2 imaging experiments. When the pipette solution calculated free [Ca²⁺] was buffered to 100, 200, or 350 nM (all with 10 mM BAPTA), the currents at −40 mV were small, similar to the cell shown in Fig. 6 A. The average current density at −40 mV from cells recorded with internal solutions up to 350 nM of calculated free [Ca²⁺] was less than ∼−2.5 pA/pF. These values are similar to the TRPC5 current amplitudes recorded using internal solutions with 100 nM of calculated free [Ca²⁺] buffered with 1 mM EGTA or BAPTA in external solutions with no added Ca (e.g., Fig. 3, trace b; average current density at −40 mV was −0.94 ± 0.29 pA/pF, n = 7, and −4.78 ± 1.04, n = 6, respectively). The similarity in current density amplitudes again suggests that regardless of possible direct inhibition of PLC by BAPTA, TRPC5 currents are small when calculated free [Ca²⁺] was in the range of 100 to 350 nM.

In contrast, CCh-activated TRPC5 currents recorded with internal solutions >1 µM were much larger. The average current densities at −40 mV from cells recorded using internal solutions with 1.4, 1.8, or 17 µM of calculated free [Ca²⁺] (buffered with 10 or 5 and 5 mM HEDTA, respectively) ranged from ∼−74 to −102 pA/pF. These values are only slightly larger than the average current density at −40 mV reached in cells recorded with 100 nM of calculated free [Ca²⁺]/1 mM BAPTA internal solutions immediately after the return to nominally Ca-free external solution from the potentiated amplitude (e.g., Fig. 3 C, trace e), which was −69.04 ± 13.69 pA/pF (n = 8). This suggests that TRPC5 channels recorded using high [Ca²⁺] internal solutions in nominally Ca-free external solution might be in the same potentiated state as that reached by most cells recorded using 1 mM EGTA- or BAPTA-buffered internal solutions and normal 2 Ca external solution.

The biphasic agonist activation resulting in large amplitude and more linear inward TRPC5 currents bears some resemblance to the large enhancement of TRPV1- and TRPV3-mediated currents during strong agonist stimulation (Chung et al., 2005; Chung et al., 2008). In these channels, prolonged activation with high concentrations of agonist reportedly can lead to an increase in the permeability to large cations such as NMDG. To directly assess whether TRPC5 channels change their permeability properties during potentiation by intracellular Ca²⁺, we monitored the reversal potential of the CCh-activated current when extracellular Na⁺ was completely replaced by NMDG (Fig. 7). 100 µM CCh was initially applied to TRPC5/M1R-expressing cells in a nominally Ca-free external solution with NMDG; the absence of external Ca²⁺ prevented the potentiation of TRPC5 current. Then, the external solution was changed to 2 Ca external solution, with 150 mM NaCl and 2 mM CaCl₂. This allowed Ca²⁺ to enter and caused subsequent potentiation of TRPC5 channels. For the cell shown in Fig. 7 A, the very large peak current at −100 mV of −11.6 nA indicates that the channels were very likely potentiated. Next, the external solution was replaced with one containing NMDG, still with 2 mM CaCl₂ present to maintain high intracellular [Ca²⁺] and current potentiation. Finally, we removed Ca²⁺ from the external NMDG solution and monitored the reversal potential soon after external Ca²⁺ removal, when TRPC5 channels were still potentiated, and ∼60 s later, when potentiation was almost fully reversed. In both cases the external solution was identical, so that any change in the reversal potential would result from alterations in active conductances or their properties.

The current–voltage curves of CCh-activated TRPC5 channels recorded in nominally Ca-free NMDG external solutions are shown in Fig. 7 B. Inward currents at −100 mV were small, only 1.5–2.5% of the level in normal Ca²⁺- and Na⁺-containing external solution, reflecting the poor absolute permeability of NMDG⁺. The reversal potential of these currents was −65 mV. At voltages depolarized from the reversal potential, the outward current carried by Cs⁺ increased, until it began to decline as V_m reached 0 to +25 mV. This reduction is presumably the result of increasing block by internal Mg²⁺, which is responsible for the flat portion of the I-V in normal external sodium (Obukhov and Nowycky, 2005). Positive to ∼+25 mV, the current was strongly outwardly rectifying. As the potentiation of TRPC5 channels declined, there was a minimal shift in the reversal potential even as the current at −100 mV decreased from ∼−250 to ∼−12 pA. This demonstrates that although TRPC5 is somewhat permeable to NMDG, its permeation does not change with intracellular Ca²⁺ potentiation. From two cells, the reversal potential of TRPC5 current in NMDG external solutions immediately after the removal of external Ca²⁺ was ∼3 mV more depolarized than after the potentiation declined. This value is much smaller than the >40-mV depolarizing shift demonstrated in NMDG external solutions for either TRPV1 channels (Chung et al., 2008) or TRPV3 channels (Chung et al., 2005).

TRPC5 and TRPC4 are closely related, sharing 80% sequence similarity. They also share many of the same functional characteristics, including activation by Gαq/11-coupled receptors and potentiation by external Ca²⁺ and lanthanides (Schaefer et al., 2000, 2002; Strubing et al., 2001). We tested whether intracellular Ca²⁺ can potentiate TRPC4 channels as well as TRPC5 channels. In HEK cells expressing the TRPC4β isoform and M1Rs, the application of 100 µM CCh activated a current with a similar I-V relation to TRPC5 (Fig. 8 B). We used this isoform as it tends to result in larger currents compared with TRPC4α. Despite the use of internal solutions with 100 nM of calculated free [Ca²⁺] and 1 mM EGTA, we never observed biphasic activation in any of the 10 cells recorded for up to 60 s in CCh. Instead, the current activated with a single-exponential time course and remained at steady state throughout the CCh application (Fig. 8 A). The average TRPC4β-mediated currents were small, −8.35 ± 2.9 pA/pF at −40 mV (n = 10), similar in amplitude to TRPC5-expressing cells that did not undergo potentiation (e.g., Fig. 1 C). One cell did have a peak current density at −40 mV of −30.0 pA/pF, but its activation was not biphasic.

Because the addition of 10 Ca external solution potentiated TRPC5 current in 13 of 15 cells (Fig. 4), we tested whether high external Ca²⁺ would also potentiate TRPC4 current (Fig. 8 B). TRPC4β current was activated by the application of 100 µM CCh for 20–30 s, until a new steady state was reached. When the external solution was changed to 10 Ca external solution, only one of three cells underwent an increase in current amplitude at −40 mV. The current peaked after 6 s at an amplitude 2.9-fold larger than in the standard 2 Ca external solution. This is much smaller than the ∼14-fold increase we found for TRPC5 channels. After the peak, the current began to decline rapidly, such that after 64 s the current was reduced by 65%.

The small inward current amplitudes recorded from TRPC4β/M1R-expressing cells might have resulted in only modest intracellular [Ca²⁺] increases, despite raising external Ca²⁺ to 10 mM. Thus, we tested whether supplying elevated [Ca²⁺] in the internal solution could lead to large amplitude TRPC4β currents. As in similar experiments with TRPC5 channels (Fig. 6), we recorded these currents in external solutions with nominally Ca-free external solution, preventing Ca²⁺ entry (not depicted). The average current density at −40 mV recorded from three TRPC4β/M1R-expressing cells using the internal solution with 1.8 µM [Ca²⁺] and 5 mM HEDTA was −32.6 ± 7.7 pA/pF. Thus, average currents were enhanced approximately fourfold relative to currents recorded with low Ca²⁺ buffering (despite the absence of Ca²⁺ in the external solution), suggesting that intracellular Ca²⁺ also potentiates TRPC4β channels. The inability of 10 mM of external Ca²⁺ to potentiate currents and the smaller average current densities with elevated [Ca²⁺] in the internal might result from lower expression of TRPC4β channels or from the apparently enhanced desensitization of TRPC4 compared with TRPC5 (Schaefer et al., 2000). We did not test other isoforms of TRPC4; whether they might differ in potentiation by Ca²⁺ is unknown.

In principle, the intracellular Ca²⁺-dependent potentiation of TRPC5 channels could result from increases in the number of active channels, in the single-channel conductance, or in the channel open probability. The different extent and kinetics of inward current potentiation relative to outward current suggest that increases in the number of active channels alone cannot underlie calcium potentiation. We tested whether the conductance of TRPC5 channels changed by measuring the open-channel I-V relation before and after potentiation by intracellular Ca²⁺. CCh-activated TRPC5 currents were recorded using pipette solutions with low calcium buffering in the presence of standard 2 Ca external solution. To measure the open-channel I-V, 15 (or 50)-ms conditioning pulses were applied to activate TRPC5 channels, and the tail currents immediately after hyperpolarization to a range of potentials were recorded. Fig. 9 A shows the voltage-dependent activation and deactivation of TRPC5 channels 15 and 91 s after the application of 100 µM CCh. Currents were initially small in amplitude (+1.19 nA at +100 mV), presumably before intracellular Ca²⁺ potentiation. The current continued increasing, interrupted only by the replacement of external calcium with barium, which prevented the current increase similar to magnesium replacement (Fig. 9 B, c.f., Fig. 3). After external calcium was reapplied, the TRPC5 current rapidly increased as calcium potentiation occurred, reaching a peak amplitude of +17.45 nA. Furthermore, large inward tail currents were visible in response to the hyperpolarizing steps. Note that the current amplitudes at +100 mV were constant throughout the sweep; under these conditions, there is little or no use dependence of TRPC5 channels during repetitive depolarizations.

To compare the open-channel I-V before and after potentiation, when current amplitudes varied up to a factor of 10, we normalized the currents to the amplitude of the current immediately after the repolarization to +60 mV (Fig. 9 C). We measured the relative instantaneous tail current during the hyperpolarizing potentials at the same time point where the TRPC5 channel open probability, reached during the +100-mV conditioning pulse (or +90 mV for some cells), remains constant. We used the smaller current amplitudes at +60 mV for normalization rather than the +100-mV amplitude to minimize possible effects of uncompensated series resistance when currents at +100 mV were very large. From six cells we found that the instantaneous I-V relationship was not affected by intracellular Ca²⁺ potentiation (Fig. 9 D). For example, the relative peak tail current at −60 mV was −0.967 ± 0.076 before potentiation and −1.107 ± 0.058 after potentiation. The larger inward currents after potentiation were instead caused by a slower and less complete deactivation of the tail current. Tail currents at −60 mV before intracellular Ca²⁺ potentiation deactivated with a single exponential with a tau of 2.54 ± 0.44 ms, which increased to 3.91 ± 0.54 ms after potentiation. Thus, the fivefold difference in potentiation between inward and outward TRPC5 currents is not likely to arise from large, asymmetrical changes in channel conductance.

To gain further understanding of the biophysical changes that occur during potentiation of TRPC5 by intracellular Ca²⁺, we performed cell-attached recordings from TRPC5/M1R-expressing HEK cells using pipettes filled with the standard 2 Ca external solution and 100 µM CCh (Fig. 10, A–C). Cells were initially bathed in 2 Ca external solution, and TRPC5 channel activity at −50 mV was recorded for ∼20 s. Despite the presence of CCh in the pipette, TRPC5 activity remained low. At this point, 100 µM CCh in 2 Ca external solution was added to the bath, activating M1R receptors and TRPC5 channels throughout the cell. In two of nine cells, this led to a dramatic increase in NP_o (number of channels multiplied by the open probability) of 14.0 ± 0.39-fold (P = 0.017), with a 14.3–17.1-s delay. Because in these experiments it is unknown whether intracellular [Ca²⁺] reached levels sufficient to cause potentiation, we then added 10 Ca external solution with CCh to the bath to increase Ca²⁺ entry, similar to whole cell experiments shown in Fig. 4. In the two cells that already displayed an elevated NP_o, the addition of 10 Ca external solution with CCh caused no further change. In six of the remaining seven cells tested, perfusion of 10 Ca external solution and CCh caused a dramatic increase in NP_o: the average peak increase in NP_o was ∼10-fold, with a delay <10 s. To test whether this increase in TRPC5 channel NP_o was due to the increase in intracellular [Ca²⁺], we changed the bath solution to nominally Ca free with CCh in four patches (Fig. 10 D). This led to a large decrease in NP_o that reached levels similar to initial 2 Ca external solution alone (0.86 ± 0.32-fold; P = 0.16). The lower peak potentiation of NP_o in these experiments compared with whole cell experiments might result from lower [Ca²⁺] caused by a reduced driving force on Ca²⁺ in the cell-attached configuration. The activation of large TRPC5 currents after bath CCh addition would likely drive cells to ∼0 mV, compared with the −40-mV holding potential used in whole cell recordings.

The increase in NP_o was apparently derived from a substantial increase in the frequency of channel openings (Fig. 10 E): the average increase during Ca²⁺ potentiation was approximately ninefold, whereas 0 Ca²⁺ and CCh perfusion returned frequency to near initial levels. Channel mean open time also increased (Fig. 10 G) on average approximately twofold, which reversed upon perfusion of nominally Ca-free external solution (1.1 ± 0.2; P = 0.34). Concurrently, channel mean closed time decreased substantially: before potentiation closed time was 64.6 ± 4.5 ms (n = 697 closings), but during potentiation it dropped to 15.4 ± 0.2 ms (n = 751 closings; P < 0.0001).

It is unclear whether the changes in NP_o were the result of an increase in the number of active channels in the patch or an increase in single-channel open probability. Three patches contained occasional double openings that could be observed under all conditions, indicating the presence of multiple channels. However, we never observed an increase in the number of open channel levels during Ca²⁺ potentiation of TRPC5, although in some cases NP_o was quite low (ranging from 0.003 to 0.376). The NP_o of patches with one apparent channel is similar to previous reports (Schaefer et al., 2000; Jung et al., 2003; Obukhov and Nowycky, 2008). For those patches in which only one channel was observed even after Ca²⁺ potentiation (average NP_o was 0.0562 ± 0.034, range 0.0033–0.156; n = 5), we used the following equation to estimate the probability that the observed openings during Ca²⁺ potentiation were actually the result of two channels (Colquhoun and Hawkes, 1995): P(r > n_O) = µ⁽ⁿ_o⁻¹⁾, where µ = (1−P_o)/(1−P_o/2), and n_O = observed number of single-channel openings (in our case, n_O = 1,562). By this calculation, the maximum probability of more than one channel being present during Ca²⁺ potentiation in those patches was ∼10⁻²⁰. Therefore, we suggest that the primary effects of Ca²⁺ potentiation are to increase open probability via a small increase in mean open time and a larger decrease in closed time.

We found that the single-channel conductance was unchanged during Ca²⁺ potentiation of TRPC5 (Fig. 10 F). Before Ca²⁺ potentiation, the average single-channel conductance was 50.36 ± 4.1 pS, whereas after potentiation it was 49.65 ± 4.7 pS (after 0 Ca²⁺ and CCh perfusion it was 53.19 ± 5.1 pS). The single-channel conductance of TRPC5 at negative membrane potentials is reportedly 38–48 pS (Schaefer et al., 2000; Yamada et al., 2000; Strubing et al., 2001). The slightly larger conductance that we observed might result from an added contribution of the cell's resting potential. The single-channel current–voltage relation (Fig. 10 B) was outwardly rectifying, closely resembling previous reports (Jung et al., 2003; Obukhov and Nowycky, 2008).

These data support the hypothesis that Ca²⁺ entry through a single TRPC5 channel does not cause maximal potentiation of that channel. 100 µM CCh was included in the pipette so that TRPC5 channels within the patch were likely to be activated and passing Ca²⁺ for ∼30 s before bath perfusion of CCh. If the resulting Ca²⁺ elevation near the activated TRPC5 channels in the patch saturated the potentiation mechanism, increases in bulk Ca²⁺ after the addition of CCh and 2 or 10 mM Ca²⁺ to the bath would not induce further potentiation. Instead, the ∼10-fold potentiation of NP_o we observed after the bath addition of CCh and Ca²⁺ suggests that Ca²⁺ that has flowed into the cell through other TRPC5 channels is also sensed by the TRPC5 recorded in the patch.

To directly determine the effect of varying [Ca²⁺] at the cytosolic face of TRPC5 channels, we performed excised inside-out patch recordings from TRPC5 and M1R-transfected HEK cells, again with 2 Ca external solution and 100 µM CCh in the pipette. The cytosolic face of the patch was exposed to solutions of varying calculated free [Ca²⁺] to determine whether TRPC5 channels were intrinsically sensitive to Ca²⁺. In 5 of 12 patches, we observed a dramatic potentiation of single-channel activity when the intracellular surface of the channel was exposed to 1.8 and 100 µM Ca²⁺, with NP_o increasing ∼23-fold and 45-fold over 100 nM Ca²⁺, respectively (Fig. 11, A–E). One of these responsive patches was dramatically stimulated by Ca²⁺ with NP_o increasing ∼150-fold in 100 µM Ca²⁺. Latency to peak activity ranged from <2 to 20 s. Single-channel conductance ranged from 45 to 49 pS and did not systematically change with bath solution. Similar to the cell-attached recordings, the increase in NP_o was due to a large increase in the number of channel openings and increased mean open time. Compared with 0.1 µM of calculated free Ca²⁺, the frequency of openings in 1.8 µM of calculated free Ca²⁺ increased by ∼15-fold, and in 100 µM Ca²⁺ by ∼34-fold. The stimulation of channel activity was also caused by an increase in mean open time, which was ∼50% for both 1.8 µM of calculated free Ca²⁺ and in 100 µM Ca²⁺. Returning patches to low [Ca²⁺] caused the channel activity to decrease, indicating that the potentiation was reversible, as in whole cell recordings.

The remaining seven patches were not responsive to changes in [Ca²⁺] (Fig. 11 F). Most of these patches initially displayed a relatively high NP_o, even in 0.1 µM of calculated [Ca²⁺], but open times and conductance were similar to the other five patches. These patches might contain channels that are already potentiated and locked in this state. Collectively, these results suggest that in some cases the ability of Ca²⁺ to strongly increase TRPC5 channel activity can survive patch excision, and that this potentiation might reflect the same processes that occur during the potentiation seen in whole cell experiments. However, the absence of potentiation in some patches suggests that an important factor was lost upon patch excision.

The results in Figs. 10 and 11 suggest that intracellular [Ca²⁺] can increase the open probability of TRPC5 at voltages ∼−50 mV, and our whole cell experiments suggest that the extent of this potentiation is voltage dependent, increasing with hyperpolarization. Potentiation of TRPC5 channels by external Ca²⁺ and lanthanides also enhances channel open probability, which increases with hyperpolarization (Jung et al., 2003). If both mechanisms exist, the ability of external Ca²⁺/lanthanides to further increase current might be reduced when channels are already potentiated by elevated intracellular [Ca²⁺]. The experiments in Fig. 3 suggest that this is the case. TRPC5 channels initially activated in nominally Ca-free external solution were not potentiated, and when 2 Ca external solution was added, the immediate increase in current presumably resulted from binding at the external Ca²⁺/lanthanide site. On average, the current at −40 mV increased 7.02 ± 1.40-fold, whereas the current at +100 mV increased 1.61 ± 1.17-fold (n = 8; compare traces b and c in Fig. 3). After the addition of external Ca²⁺ led to potentiation of the TRPC5 current, intracellular [Ca²⁺] presumably reached high levels. At this point the removal of external Ca²⁺ resulted in an immediate decrease in current amplitude, reflecting loss of the potentiation at the Ca²⁺/lanthanide site (compare traces d and e in Fig. 3). The average potentiation by 2 mM Ca²⁺ at this point was 2.22 ± 0.17-fold at −40 mV and 1.11 ± 0.04-fold at +100 mV (n = 8). Thus, after TRPC5 channels were potentiated by intracellular [Ca²⁺], the effect of extracellular Ca²⁺ was reduced by 68% at −40 mV and ∼31% at +100 mV.

We tested the proposed interaction between internal Ca²⁺ and external Ca²⁺/lanthanide potentiation using internal solutions with low (200 nM) or high (1.4–17 µM) calculated [Ca²⁺] to keep TRPC5 channels in non-potentiated and potentiated states, respectively. Using these internal solutions, we measured the response as external Ca²⁺ was increased from 0 to 10 mM (Fig. 12). TRPC5 channels were initially activated by the addition of 100 µM CCh in nominally Ca-free external solutions. After 45–60 s in CCh, the current had reached steady state and external [Ca²⁺] was increased to 0.3, 1, 3, and 10 mM. The effect of changing external Ca²⁺ on TRPC5 current was immediate and remained constant during each 10-s [Ca²⁺] application. In some cells recorded with low internal [Ca²⁺] (Fig. 12 A), the application of 10 mM Ca²⁺ first led to an immediate increase in current that then further developed, presumably as Ca²⁺ entry escaped control of BAPTA in the internal solution. In contrast, for some cells recorded with high internal [Ca²⁺] (Fig. 12 B), the application of 10 Ca external initially increased current (presumably as the potentiation at the external site increased), followed by a decline (as the current desensitized). For all cells, the current was averaged from the sweep immediately after the application of 10 Ca external solution exchange to minimize any contribution of these changes.

These experiments demonstrate that when TRPC5 channels were recorded with low internal [Ca²⁺], the ability of extracellular Ca²⁺ to potentiate the current was much greater than when internal solutions with high [Ca²⁺] were used. For internal solutions with 200 nM of calculated free internal [Ca²⁺], the TRPC5 current densities at −40 mV increased by ∼10-fold in 10 Ca external solution (relative to the current in nominally Ca-free external). For internal solutions with 1.4, 1.8, or 17 µM of free [Ca²⁺], we averaged the results from all cells because the potentiation by external Ca²⁺ was equal for each. Current densities at −40 mV were potentiated by only approximately threefold in the 10 Ca external solution. Overall, potentiation of TRPC5 channels was equivalent for the two sets of internal solutions at 0.3 and 1 Ca external solution. However, potentiation with high calculated free [Ca²⁺] internals was 35% smaller for 3 Ca external solution and 67% smaller for 10 Ca external solution.

We also tested whether potentiation by La³⁺ was reduced when TRPC5 channels were potentiated by elevated intracellular [Ca²⁺] (Fig. 13). We applied 100 µM La³⁺ during whole cell recordings of CCh-activated TRPC5 currents using internal solutions containing 200 nM, 1.8 µM, and 100 µM Ca²⁺ (to avoid competition between Ca²⁺ and La³⁺ for the external binding site, these experiments were performed in nominally Ca-free external solution). After TRPC5 currents reached steady state after CCh application, the addition of 100 µM La³⁺ potentiated TRPC5 current by ∼14.5-fold at −40 mV and ∼5.5-fold at +100 mV (200 nM of calculated free [Ca²⁺] internal). This potentiation is larger than that caused by 3 or 10 Ca external solution (Fig. 12), as expected given the fact that La³⁺ potentiates TRPC5 current in the presence of 2 mM of external Ca²⁺ (Jung et al., 2003).

At higher internal [Ca²⁺], we observed a voltage-dependent reduction in La³⁺ potentiation. With 1.8 µM of calculated internal [Ca²⁺], 100 µM LaCl₃ potentiated TRPC5 currents on average 12.04-fold at −40 mV and 2.06-fold at +100 mV. Increasing internal [Ca²⁺] to 100 µM further reduced La³⁺ potentiation of TRPC5 current to 6.07-fold at −40 mV and 1.55-fold at +100 mV. Thus, the reduction in La³⁺ potentiation was voltage dependent, being more pronounced at +100 mV (approximately threefold reduction) than at −40 mV (∼20% reduction) when the internal solution contained 1.8 µM of calculated [Ca²⁺]. Jung et al. (2003) have demonstrated that the effect of La³⁺ is to increase channel open probability, overcoming direct channel block by La³⁺. Our single-channel data suggest that a primary feature of internal Ca²⁺ potentiation is also to increase channel open probability, which may occlude the action of La³⁺.

The primary sequence of TRPC5 channels contains three putative calmodulin-binding domains; interactions have been demonstrated at the two carboxy-terminal domains (Zhu, 2005). In one case, calmodulin binding appeared to speed TRPC5 channel activation kinetics (Ordaz et al., 2005). We tested Ca²⁺/calmodulin's participation in potentiation by including calmodulin inhibitors in the internal solution and by coexpressing dominant-negative calmodulin (Fig. 14).

In the first of two experiments, we included 1 µM of calmodulin inhibitory peptide in a Cs-Asp–based internal solution with 1 mM EGTA and recorded TRPC5 currents after activation by 100 µM CCh. This concentration of calmodulin inhibitory peptide is well above the 7 nM IC₅₀ determined in vitro (Torok et al., 1998). From five cells recorded in this way, we found that each had biphasic activation, and the peak current density at −40 mV was −113.3 ± 20.0 pA/pF, similar to control cells (e.g., Fig. 1 C). TRPC5 currents recorded with calmodulin inhibitory peptide activated more slowly than control cells, with an ∼2.5-fold increase in the latency to peak (68.7 ± 7.2 s for control, 160 ± 30.63 s for calmodulin inhibitory peptide; P = 0.0087). We also tested whether the calmodulin inhibitor W-7, applied at 10 µM in the 1-mM EGTA internal solution, was able to inhibit the development of large, biphasic TRPC5 currents activated by 100 µM CCh. From each of four cells recorded with W-7 in the pipette, we observed typical biphasic TRPC5 currents, with the average peak current density at −40 mV of −109.5 ± 32.0 pA/pF. In contrast to the calmodulin inhibitory peptide, W-7 treatment did not significantly affect the latency to peak (control latency 68.7 ± 7.2 s vs. W-7–treated cells 123.0 ± 36.4 s; P = 0.17). In these experiments, we used bath perfusion to exchange external solutions, likely causing the increased latencies for activation in control cells (compare Fig. 1).

Finally, we coexpressed a dominant-negative calmodulin in which all four Ca²⁺ binding E-F hands were mutant (crucial aspartate residues mutated to alanine) with TRPC5 and M1Rs. Similar to the results obtained with calmodulin inhibitors, CCh-activated currents recorded from these cells exhibited the typical biphasic response (Fig. 14 C), with an average peak density at −40 mV of −189.2 ± 37.2 pA/pF (n = 7). As in control cells (e.g., Fig. 3), removal of external Ca²⁺ in these cells led to an immediate decline in the current, followed by a slower reduction in the current over the next 90–120 s. There was no significant effect on the latency to peak current in these cells (control cells: 38.3 ± 5.5 s, n = 18; dominant negative cells: 31.4 ± 8.7 s, n = 7).

Our results suggest that calmodulin is not involved in the intracellular Ca²⁺ potentiation of TRPC5 channels. Although previous studies have shown that calmodulin inhibitors do not affect TRPC5 current amplitudes (Ordaz et al., 2005), some report that calmodulin inhibitors reduce agonist-activated TRPC5 currents (Kim et al., 2006; Shimizu et al., 2006). These reports suggest that calmodulin acts via myosin light chain kinase to either affect upstream receptor signaling or the plasma membrane delivery of TRPC5 channels. The presence of large, CCh-activated TRPC5 currents in our experiments might result from the overexpression of M1Rs or the use of low [Ca²⁺] internal buffering.

Our results suggest that intracellular [Ca²⁺] elevations will play an important role in controlling the amplitude of agonist-activated TRPC5 currents. Potentiation of the TRPC5 channel appears to be maximal at [Ca²⁺] > 1 µM, leading to an ∼25-fold increase in inward currents at physiological membrane potentials. These large current amplitudes could generate a strong membrane depolarization and also further enhance Ca²⁺ entry directly through TRPC5 channels themselves. Furthermore, the intracellular [Ca²⁺] dependence and the demonstrated ability of heterologously expressed TRPC5 channels to overcome low internal solution buffering conditions must be considered when evaluating mechanisms proposed to modulate TRPC5 activity. This is especially important in Ca²⁺ imaging experiments, when endogenous buffer capacity is unknown. Maintaining constant [Ca²⁺] levels with high concentrations of internal solution buffers, or removal of external Ca²⁺, will mitigate the confounding effects of uncontrolled intracellular [Ca²⁺].

Our results show that changes in the single-channel conductance do not occur during potentiation, suggesting that some combination of increased channel number (N) or channel open probability is likely to be the cause. Although some fraction of TRPC5 channels are located in intracellular vesicles and can be inserted into the plasma membrane in response to activation of growth factor receptors (Bezzerides et al., 2004), the kinetics of that process (∼2.5 min to peak) are much slower than the typical onset of TRPC5 channel potentiation, which had latencies of ∼30 s. Additionally, our single-channel data suggest that the likelihood of an increase in channel number during potentiation in those experiments is very low. Lastly, the voltage dependence of this potentiation (∼5-fold at +100 mV and 25-fold at −40 mV), and its occlusion of external Ca²⁺/lanthanide potentiation, suggests that changes in channel open probability likely contribute to the effect. TRPC5 channels are voltage dependent, with depolarization-enhancing activation (Obukhov and Nowycky, 2008). One possibility is that the binding of intracellular Ca²⁺ shifts the position of the voltage dependence of activation toward more hyperpolarized potentials, analogous to the effect of Ca²⁺ on large-conductance Ca²⁺-activated potassium (i.e., BK) channels (Cui et al., 1997). Such a time-dependent shift in the activation curve, as calcium increases, might also explain the different kinetics of potentiation for outward and inward current (Fig. 1 B). For example, if the current at +100 mV begins with a nonzero extent of activation, whereas the current at −40 mV begins at the foot of the activation curve, a hyperpolarizing shift in the activation curve will immediately result in increased current at +100 mV, whereas the current at −40 mV will activate with a delay. Such an explanation has been suggested for TRPV1 activation by capsaicin and TRPM8 activation by menthol (Voets et al., 2004). Lastly, the binding of extracellular Ca²⁺/lanthanides might result in a similar shift in the voltage dependence of activation, causing the observed voltage-dependent increase in TRPC5 channel open probability, as suggested by Obukhov and Nowycky (2008).

The precise location where calcium acts to potentiate TRPC5 channels is not completely clear from our experiments. A location close to, or perhaps directly on, the channel is supported by the greater effectiveness of 10 mM BAPTA relative to 10 mM EGTA in preventing potentiation (Fig. 2 B). A more distant location is supported by several results, including the dependence of TRPC5 potentiation on global [Ca²⁺], as directly measured by fura-2 (Fig. 5); the ability of 10 mM EGTA to slow, and in some cells prevent, potentiation (Fig. 2 A); and the slow decline in TRPC5 current after the removal of external calcium (Fig. 3). Furthermore, the persistence of potentiation in some excised patches suggests that the mechanism can survive removal from the cell. Lacking detailed information on the contribution of calcium ions to the overall TRPC5 current, as well as uncertainties stemming from buffer saturation and varying channel expression level, it is difficult to model the profile of [Ca²⁺] near open TRPC5 channels. What is clear is that the Ca²⁺ entering through a single TRPC5 channel is insufficient to generate potentiation, or else we would have observed it in every cell. Similarly, we would not have observed potentiation of TRPC5 channels in our cell-attached recordings (where the pipette contained 2 Ca external solution) after the addition of 10 Ca solution to the bath (Fig. 10 A). This dependence on calcium arising from more distant TRPC5 channels might possibly lead to the correlation between global [Ca²⁺] and potentiation, even with a sensor location associated with the channel (yet still relatively distant from the channel mouth).

There are no predicted Ca²⁺ binding sites on TRPC5 channels, but there are three motifs suggested to interact with calmodulin or related Ca²⁺ binding proteins. One is located in the amino terminus, and two are in the carboxy terminus (Zhu, 2005). Ordaz et al. (2005) showed that calmodulin binding to the second carboxy-terminal domain, called CBII (amino acids 828–854 of mouse TRPC5), results in faster agonist activation of TRPC5 currents, but leaves maximal current amplitudes unchanged. This observation is consistent with our result that inhibition of calmodulin does not affect Ca²⁺ potentiation (Fig. 14). Ca²⁺ binding protein 1 has also been reported to bind the carboxy terminus of TRPC5 (over a larger region), although its inhibition of Ca²⁺ activation of TRPC5 channels makes it unlikely to act as the sensor for potentiation (Kinoshita-Kawada et al., 2005).

NCS-1 is another Ca²⁺ binding protein that has effects on Ca_V, K_V, and IP₃ receptor channels (for review see Burgoyne, 2007). NCS-1 has been shown to interact with TRPC5, again in the carboxy-terminal region (Hui et al., 2006). They showed that coexpression of a dominant-negative E-F hand mutant of NCS-1 reduced TRPC5 activity in response to muscarinic receptor activation, extracellular lanthanides, or intracellular Ca²⁺, but not through changes in surface TRPC5 expression. NCS-1 is a possible candidate underlying Ca²⁺ potentiation, though it is unknown whether NCS-1 interacts with TRPC4β, which can also be potentiated.

Ca²⁺ plays a central role in controlling the activity of many TRP channels, either through direct binding to the channel or through the effects of Ca²⁺ binding proteins, such as calmodulin (for review see Zhu, 2005). For example, TRPA1 is strongly potentiated by intracellular Ca²⁺, which may bind at the E-F hand motif located in the amino terminus of the channel (Doerner et al., 2007; Zurborg et al., 2007; Wang et al., 2008). Calmodulin interacts directly with many TRP channels to effect both positive and negative regulation; examples include Drosophila TRPL (Zimmer et al., 2000), TRPC6 and TRPC7 (Shi et al., 2004), TRPV1 (Rosenbaum et al., 2004), TRPV3 (Xiao et al., 2008), TRPV4 (Strotmann et al., 2003), TRPV6 (Lambers et al., 2004), TRPM2 (Tong et al., 2006), and TRPM4 (Nilius et al., 2005).

When using GPCR stimulation to activate TRP channels, it should be noted that numerous steps before channel activation are Ca²⁺ or calmodulin dependent, including receptor desensitization (Mundell et al., 2004; Turner and Raymond, 2005), PLC activation (McCullar et al., 2003; Horowitz et al., 2005), and PKC activation. Calmodulin regulation of components in the upstream activation pathways of some channels can lead to changes in agonist-activated current amplitudes. For example, calmodulin mutant alleles strongly reduce the light-activated currents of TRPL channels but leave quantal bumps unaffected, suggesting that the channels themselves remain functional (Scott et al., 1997). Another example is TRPC5 itself; application of calmodulin inhibitors reduces M1R-activated currents (presumably via myosin light chain kinase) but spares currents activated by intracellular perfusion of GTPγS (bypassing the receptor pathway) (Kim et al., 2006; see also Shimizu et al., 2006).

The properties of TRPC4 and TRPC5 channels share many similarities with a muscarinic receptor–activated cation current recorded in several different gastrointestinal smooth muscle cell types. The muscarinic-activated cation current in these cells has an I-V relation that is inwardly rectifying and flat positive to the reversal potential (Inoue and Isenberg, 1990a), is activated by depolarization (Inoue and Isenberg, 1990b; Zholos and Bolton, 1994), and is potentiated by external La³⁺ (Inoue et al., 1998). Inoue and Isenberg (1990c) have demonstrated that this current, recorded from guinea pig ileum, is also potentiated by intracellular Ca²⁺ with an EC₅₀ of ∼200 nM, and that this potentiation was maximal at >1 µM. In comparison, our results showed that [Ca²⁺] > 350 nM was required to generate potentiation, which was maximal at 1.4 µM [Ca²⁺]. Expression analysis suggests that TRPC4 is more likely to produce the ileal smooth muscle cell muscarinic-activated cation current (Lee et al., 2003; Lee et al., 2005). If true, the higher Ca²⁺ sensitivity of this current might reflect an intrinsic difference between TRPC4 and TRPC5 channels or, alternatively, modulation by endogenous mechanisms.

The muscarinic-activated current depolarizes smooth muscle cells to trigger Ca_V channels, and the resulting Ca²⁺ entry then feeds back to potentiate the cation current, prolonging the depolarization. In contrast, Ca²⁺ entering through TRPC5 channels themselves was the likely source of intracellular [Ca²⁺] elevations in our experiments. This suggests that any [Ca²⁺] elevation, resulting either from Ca²⁺ entry (via Ca_V or Ca²⁺-permeable neurotransmitter-gated channels) or release from internal stores, that reaches a sufficient level can be integrated by TRPC4 and TRPC5 channels and alter their activity.

TRPC5 is expressed in a variety of brain regions, including neurons in the hippocampus, cortex, amygdala, cerebellum, and hypothalamus (Mori et al., 1998; Philipp et al., 1998; Strubing et al., 2001; Fowler et al., 2007). The precise roles of TRPC5 channels in these neuronal cell types remain undefined, although they are excellent candidates for generating nonselective cation currents activated by metabotropic glutamate receptors (Fowler et al., 2007) or neuropeptides (Meis et al., 2007; Kohlmeier et al., 2008). Furthermore, Ca²⁺-dependent nonselective cation currents have been described in neurons from the substantia nigra (Lee and Tepper, 2007) and the entorhinal cortex (Magistretti et al., 2004), but the source of the [Ca²⁺] required to activate these currents is unknown. These currents may be generated by TRPC4 or TRPC5 channels, or by other Ca²⁺-sensitive TRP channels. The typically small amplitude of most neuronal nonselective currents, in the presence of multiple endogenous conductances, makes testing their properties difficult. Such studies might be facilitated by manipulations that increase [Ca²⁺] to maximize TRP channel potentiation.

Resting [Ca²⁺] in most neurons is typically ∼50–150 nM, but repetitive activity and back-propagating action potentials can generate [Ca²⁺] elevations of ∼0.2 to >1 µM in dendrites and dendritic spines via activation of Ca_V channels (e.g., Helmchen et al., 1996; Sabatini et al., 2002). Activation of NMDA receptors can generate even higher levels, >10 µM in spines (Petrozzino et al., 1995). These [Ca²⁺] would be sufficient to potentiate TRPC5 channels when occurring simultaneously with GPCR activation and, again, enhance depolarization and Ca²⁺ entry.

Although the subcellular localization of TRPC5 channels in adult neurons is unknown, TRPC5 is expressed throughout the soma and processes, including growth cones, of embryonic hippocampal neurons (Greka et al., 2003). Ca²⁺ is critically important in controlling growth cone motility (for review see Gomez and Zheng, 2006), and Ca_V channels, internal stores, and receptor-activated channels all contribute to growth cone [Ca²⁺]. The Ca²⁺-dependent potentiation of TRPC5 might make it a crucial integrator of these multiple signals.

We thank the members of the Clapham laboratory for helpful comments and Svetlana Gapon for tissue culture. The dominant-negative calmodulin plasmid was provided by David T. Yue, and the TRPC4β plasmid was provided by Louis H. Philipson.

Angus C. Nairn served as editor.