Voltage is a partial activator of rat thermosensitive TRP channels

Abstract

TRPV1 and TRPM8 are sensory nerve ion channels activated by heating and cooling, respectively. A variety of physical and chemical stimuli activate these receptors in a synergistic manner but the underlying mechanisms are unclear. Both channels are voltage sensitive, and temperature and ligands modulate this voltage dependence. Thus, a voltage-sensing mechanism has become an attractive model to explain the generalized gating of these and other thermo-sensitive TRP channels. We show here using whole-cell and single channel measurements that voltage produces only a partial activation of TRPV1 and TRPM8. At room temperature (20–25°C) membrane depolarization evokes responses that saturate at ∼50–60% of the maximum open probability. Furthermore, high concentrations of capsaicin (10 μm), resiniferatoxin (5 μm) and menthol (6 mm) reveal voltage-independent gating. Similarly, other modes of TRPV1 regulation including heat, protein kinase C-dependent phosphorylation, and protons enhance both the efficacy and sensitivity of voltage activation. In contrast, the TRPV1 antagonist capsazepine produces the opposite effects. These data can be explained by an allosteric model in which voltage, temperature, agonists and inverse agonists are independently coupled, either positively or negatively, to channel gating. Thus, voltage acts separately but in concert with other stimuli to regulate channel activation, and, therefore, a voltage-sensitive mechanism is unlikely to represent a final, gating mechanism for these channels.

Transient receptor potential (TRP) channels are gated by a diverse range of physical and chemical stimuli. Products of phospholipase C signalling including diacylgylcerol, phosphatidylinositol 4,5-bisphosphate, and protein kinase C represent critical, conserved pathways for TRP regulation (Clapham, 2003). In addition, several TRP channels are gated by temperature including TRPV1-4, TRPM5 and TRPM8 (Dhaka et al. 2006). The molecular basis for thermal sensitivity is not clear. Many temperature-sensitive TRPs are voltage sensitive and therefore it has been proposed that a voltage-sensitive mechanism may contribute to thermal sensing (Voets et al. 2004; Nilius et al. 2005; Talavera et al. 2005). This interrelationship is exemplified by the capsaicin receptor TRPV1, arguably the best studied TRP channel. TRPV1 is gated directly by heat, capsaicin (Caterina et al. 1997), protons (Tominaga et al. 1998), cations (Ahern et al. 2005, 2006), spider toxins (Siemens et al. 2006) and a number of fatty acids (Zygmunt et al. 1999; Hwang et al. 2000; Smart et al. 2000; Huang et al. 2002; Ahern, 2003; Movahed et al. 2005; Matta et al. 2007). Several studies also indicate that a voltage-sensitive process plays a fundamental role in TRPV1 regulation (Piper et al. 1999; Gunthorpe et al. 2000; Ahern & Premkumar, 2002; Vlachova et al. 2003; Voets et al. 2004). Notably, Voets et al. (2004) showed that depolarization alone activates the channel at room temperature (21°C), albeit with a high threshold; the voltage for half-maximal activation (V_1/2) is ∼+150 mV. However, heat or capsaicin can shift this relationship to less depolarized potentials such that the channel opens under physiological voltages. Further, depolarization reduces the threshold for activation by heat and capsaicin. Interestingly, the cold- and menthol-activated channel TRPM8 exhibits a similar voltage sensitivity to TRPV1, albeit with the opposite temperature dependence (Voets et al. 2004). Therefore, it has been proposed that voltage sensitivity may represent a fundamental mechanism by which thermal and other stimuli regulate channel gating.

If gating of TRPV1 and TRPM8 occurs through a final, voltage-dependent mechanism, then one would expect that the maximal open probability would be attained at some finite potential, independently of other stimuli. To examine this, we looked for changes in the sensitivity and efficacy of voltage activation at varying temperatures and in the presence of channel agonists. Our data confirm that temperature and agonists alter the voltage sensitivity of TRPV1 and TRPM8. However, we show that these stimuli also enhance the maximal activation produced by depolarization. Further, activation with ‘saturating’ concentrations of agonists reveals a voltage-independent component of gating. Thus, it appears that membrane depolarization acts as a partial activator of these channels that synergizes with other stimuli, and therefore does not represent the underlying mechanism for gating. We present an allosteric model to account for voltage, temperature, agonist and antagonist modulation of these TRP channels

Methods

HEK cell electrophysiology

HEK 293F cells (Invitrogen) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 1% non-essential amino acids and 10% fetal calf serum. Cell cultures were maintained at 37°C with 5% CO₂ on poly d-lysine-coated glass coverslips. Cells were transfected with rat TRPV1 or rat TRPM8 (a gift from David Julius, University of California, San Francisco), S502A/S800A TRPV1 (a gift from Makoto Tominaga, National Institutes of Natural Sciences (Japan)) and GFP cDNA using Lipofectamine™ Transfection Reagent (Invitrogen) according to the manufacturer's instructions and used 24–48 h after transfection. Whole-cell and cell attached patch clamp recordings were performed using an EPC8 amplifier (HEKA). For whole-cell recordings series resistance compensation was routinely set at ∼40–50%. The current signal was low-pass filtered at 1–3 kHz and sampled at 4 kHz. Currents were further filtered for display purposes. The bath solution contained (mm): 140 NaCl, 4 KCl, 1 MgCl₂, 1.2 CaCl₂, 10 Hepes, 10 glucose, pH 7.4 (290 mosmol l⁻¹). In cell-attached experiments, NaCl was replaced with 144 mm KCl. The pipette solution contained (mm) 140 CsCl, 10 NaCl, 10 Hepes, 5 EGTA, 2 MgATP and 0.03 GTP, pH 7.3, and in cell-attached experiments 140 NaCl, 10 Hepes, 5 EGTA, pH 7.3. Solutions were applied via a gravity fed system with an outlet ∼100 μm in diameter and positioned ∼100–200 μm from the cell of interest. Separate outlets were used to apply capsaicin and RTX solutions to avoid contamination. For heat experiments, temperature was controlled using an in-line solution heater and verified with a probe positioned adjacent to the cell of interest (Warner Instruments). For experiments with protons (pH 6), in most cases, endogenous ASIC channels were desensitized by pretreatment with a pH 6.8 solution for 45–60 s as described by Gunthorpe et al. (2000). Furthermore, ∼3 s after the application of protons (pH 6) there were no detectable current in non-transfected HEK cells during voltage steps indicating that ASIC currents did not contaminate TRPV1 current traces.

Data analysis

A family of test potentials for 100 ms duration was used to study voltage-dependent activation. The steady-state currents at the end of the test potentials were used to plot the conductance as a function of the test potential using the following Boltzmann function:

where G is the conductance, G_min and G_max are the minimum and maximum conductance, V_1/2 is the potential that elicits half maximal conductance, z is the gating charge, F is the Faraday constant and R is the gas constant. In cell-attached experiments, open probability was calculated as the current integral divided by the product of the current amplitude, the number of channels, and the time duration of the voltage steps (50 ms). Channel amplitudes were measured by fitting all-points histograms to Gaussian functions. All fitting and modelling were performed using Origin software (OriginLab Corp., Northampton, MA, USA).

Chemicals

Capsazepine, phorbol 12,13-dibutyrate (PDBu), menthol and resiniferatoxin (RTX) were obtained from Sigma (St Louis, MO, USA). Capsaicin was obtained from Tocris Cookson (Ellisville, MO, USA). Drugs were prepared as stock solutions in ethanol and diluted into physiological solution prior to experiments.

Statistical analysis

Data are given as means ± s.e.m. and statistical significance (P < 0.05) was evaluated using Student's t test for paired or unpaired data as indicated.

Results

Depolarization evokes submaximal TRPV1 currents

We studied voltage-dependent properties of TRPV1 by performing whole-cell patch-clamp experiments in TRPV1-expressing HEK293 cells. Figure 1A shows representative current traces resulting from a family of voltage steps either under control conditions (25°C) or in the presence of capsaicin (50 nm) or protons (pH 6). Figure 1B shows plots of peak conductance versus voltage. The smooth curves are best fits to a Boltzmann function used to calculate the maximal conductance (G_max) and voltage for half-maximal activation (V_1/2). Under control conditions voltage alone activated the channel with a V_1/2 of 121 ± 6 mV. Capsaicin and protons shifted the V_1/2 to 69 ± 8 mV and 48 ± 10 mV, respectively. Figure 1C summarizes these data from five experiments; protons reduced the V_1/2 by ∼74 mV, while capsaicin (50 nm and 10 μm) reduced the V_1/2 by 53 and 215 mV, respectively. These shifts in V_1/2 values are similar to those previously reported for TRPV1 ligands (Voets et al. 2004; Matta et al. 2007). In addition to enhanced voltage sensitivity, it is very apparent that the TRPV1 agonists increased the G_max. Under control conditions the G_max was about 60% of that observed with a saturating concentration of capsaicin (10 μm, at +200 mV, Fig. 1D). This value increased with 50 nm capsaicin (0.82 ± 0.02) and protons (pH 6.0, 0.98 ± 0.04).

Figure 1. Capsaicin and protons enhance the sensitivity and efficacy of voltage activation for TRPV1.

A, representative current traces in response to a family of voltage steps (voltage protocols included above current traces, capsaicin is abbreviated cap). B, the representative Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. C, the mean shift in the V_1/2 from control conditions at 25°C (n = 5). D, the mean fractional G_max normalized to 10 μm capsaicin at +200 mV (n = 5, *P = 0.01, **P = 0.007).

Next, to confirm that these changes in G_max reflected an increase in channel gating and not changes in single channel conductance, we performed single channel measurements from cell-attached patches. We used the cell-attached configuration rather than cell-free patches because we found this aided the patch stability during large depolarizations. In addition, we used a high [K⁺] bathing solution to eliminate membrane potential (see Methods). Figure 2A shows representative current traces activated by brief (50 ms) voltage steps from 0 to +200 mV before and after addition of 1 μm capsaicin. Figure 2B shows the ensemble of 13 consecutive sweeps. Similar to the whole-cell data, capsaicin increased the peak current at the end of the pulse by ∼2 fold and the overall open probability (P_o) from 0.18 to 0.48. Figure 2C shows all-points histograms used to derive the single-channel conductance. Under control conditions, the conductance was ∼115 pS, whereas, in the presence of 1 μm capsaicin the conductance was modestly decreased to ∼90 pS. Thus, the larger ensemble current in the presence of capsaicin (Fig. 2B) reflects an increase in P_o. Furthermore, we measured single channel P_o over a large voltage range (−230 to +250 mV) before and after capsaicin (50 nm, note we used a lower capsaicin concentration here to improve patch stability during extended depolarizations). Figure 2D shows that capsaicin increased the maximal P_o from 0.19 ± 0.03 to 0.41 ± 0.02 (n = 5, P < 0.01) and shifted the V_1/2 from 150 ± 2 mV to 98 ± 4 mV. Taken together, these results indicate that membrane depolarization alone is not sufficient to activate TRPV1 fully. Rather, voltage acts in synergy with other agonists – capsaicin and protons – to maximally activate the channel.

Figure 2. Capsaicin increases the maximal, voltage-evoked activity of TRPV1 in cell-attached patches.

A, representative traces of single channel activity in cell-attached patches. Activity was evoked by a 50 ms voltage pulse from 0 mV to 200 mV in the absence (left) or presence (right) of 1 μm capsaicin. Note that the bath contained a high [K⁺] concentration in order to nullify cell membrane potential. B, the ensemble of 13 consecutive traces under control conditions (left) or 1 μm capsaicin (right). C, all-points histograms from selected data segments at +200 mV. The smooth lines are the best fits to Gaussian functions. D, single channel P_o versus voltage before and after 50 nm capsaicin. Activity was evoked by 40 mV voltage steps between −230 mV and +250 mV. The smooth line is the best fit to a Boltzmann function (parameters are given in text). Data are means of 5 cells and 5–7 trials per cell.

High concentrations of capsaicin and RTX reveal voltage-independent gating of TRPV1

Although TRPV1 shuts at negative potentials in the presence of low concentrations of agonists (see Fig. 1B) we found that higher concentrations of capsaicin revealed voltage-independent activation. Figure 3A and B shows that in the presence of 10 μm capsaicin, a significant fraction of current remained even at −220 mV. Indeed, Boltzmann fits revealed that channels did not fully shut; approximately 40% of the current was voltage independent. There was no difference in the effective gating charge (z) between control and capsaicin (0.69 and 0.70, respectively). The percentage of voltage-independent gating in the presence of 10 μm capsaicin was variable (range 20–100%), but on average it was 45% (Fig. 3F). To explore this further we tested effects of the ultrapotent agonist, resiniferatoxin (RTX, Fig. 3C and D). Strikingly, in the presence of 5 μm RTX, TRPV1 currents exhibited voltage-independent gating, with a near ohmic conductance over a 400 mV voltage range (from −220 to 200 mV) in all cells tested (n = 5). We also observed a near ohmic conductance in ∼20% of recordings with 10 μm capsaicin. These data strongly indicate that chemical and voltage activation of TRPV1 are separable.

Figure 3. Saturating concentrations of capsaicin and RTX uncover voltage-independent gating.

A, representative current traces in response to a family of voltage steps (voltage protocols included above current traces) in the absence and presence of 10 μm capsaicin. B, the representative Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. C, representative current traces in response to a family of voltage steps in the absence and presence of 5 μm RTX. D, boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. E, summary of the V_1/2 values for control and 10 μm capsaicin at 25°C (n = 9, **P = 0.0005). F, summary of the voltage-independent component (G_min, n = 15 for capsaicin and n = 5 for RTX). G, summary of the mean fractional G_max normalized to 10 μm capsaicin or 5 μm RTX (n = 5).

Depolarization does not fully activate TRPM8

The cold and menthol receptor TRPM8 is also sensitive to voltage, and like TRPV1, both thermal and chemical stimuli alter the voltage sensitivity (Brauchi et al. 2004; Voets et al. 2004). We therefore asked whether voltage can fully activate TRPM8. Figure 4A and B shows that menthol enhanced both the efficacy and sensitivity of activation by voltage. At 25°C, 0.8 mm menthol increased the G_max by approximately 2-fold (Fig. 4G), shifted the V_1/2 from 121.6 ± 5.6 mV to −5.3 ± 27.2 mV (Fig. 4E), and reduced z from 0.7 ± 0.02 to 0.56 ± 0.06 (P < 0.02). At a higher concentration of menthol (6 mm), the V_1/2 was further shifted to −110 ± 30.9 mV (Fig. 4C–E), z was reduced to 0.29 ± 0.03 (P < 0.005) and channels did not completely shut at negative potentials; approximately 5% current remained with 0.8 mm menthol and this was enhanced to ∼34% with 6 mm menthol (Fig. 4F). Thus, similar to TRPV1, TRPM8 exhibits voltage-independent gating.

Figure 4. Voltage partially activates TRPM8.

A, representative current traces in response to a family of voltage steps (voltage protocols included above current traces) in the absence and presence of 0.8 mm menthol. B, the representative Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. C, representative current traces in response to a family of voltage steps in the absence and presence of 6 mm menthol. D, Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. E, summary of the V_1/2 values for control, 0.8 mm menthol, and 6 mm menthol at 25°C (n = 5, **P = 0.003, ***P = 0.0002). F, summary of the voltage-independent fractional response (n = 5, **P = 0.002). G, summary of the mean fractional G_max normalized to 6 mm or 0.8 mm menthol (n = 5).

Heat and PKC enhance the sensitivity and efficacy for voltage activation of TRPV1

Our data indicate that depolarization does not fully activate TRPV1 and TRPM8, and moreover both channels exhibit voltage-independent gating. Thus, voltage sensitivity cannot underlie chemical gating of these channels. Nonetheless, it is possible that a voltage-sensitive mechanism subserves thermal sensitivity. Indeed, heat is recognized as a partial TRPV1 agonist with an efficacy less than half of capsaicin (Tominaga et al. 1998). We therefore explored the voltage dependence of TRPV1 at varying temperatures to test whether temperature and voltage act in an additive or non-additive manner. As previously reported (Voets et al. 2004) the sensitivity of voltage-dependent activation was enhanced with increasing temperature (Fig. 5A and B); the V_1/2 values at 20, 25 and 32°C were 149.9 ± 8.7 mV, 121.4 ± 6.1 mV and 67.3 ± 13.4 mV, respectively (Fig. 5C) and z values were not significantly different (0.63 ± 0.02, 0.65 ± 0.02 and 0.60 ± 0.01, respectively). However, we found that temperature, similar to chemical stimuli (capsaicin and protons), also increased the efficacy of voltage activation. At 20°C the fractional G_max (normalized to a maximal response at +200 mV obtained with pH 6.0) was 0.51 ± 0.04, and this increased to 0.61 ± 0.02 and 0.93 ± 0.07 when the temperature was, respectively, increased to 25°C and 32°C (Fig. 5A, B and D). It is important to note that the normalization procedure employed eliminates effects of temperature on single channel conductance and therefore these changes in G_max reflect changes in channel gating. To explore this further we measured single channel activity in cell-attached patches. Increasing the temperature from 25°C to 32°C increased the P_o recorded at +210 mV from 0.21 ± 0.02 to 0.35 ± 0.03 (n = 3, P < 0.01).

Figure 5. Heat enhances the sensitivity and efficacy of voltage activation for TRPV1.

A, representative current traces in response to a family of voltage steps at 25°C and 32°C in the absence or presence of protons (pH 6.0). B, Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. C, summary of the V_1/2 values at the given temperatures (n = 4–5). D, summary of the fractional G_max normalized to pH 6.0 (n = 4–5). *P < 0.05 **P < 0.005. Data were compared to 20°C.

Next, we tested the effects of protein kinase C (PKC) on voltage-dependent activation. PKC plays a prominent role in TRPV1 regulation; PKC-dependent phosphorylation of TRPV1 markedly sensitizes the channel to activation by temperature and ligands (Premkumar & Ahern, 2000; Vellani et al. 2001; Numazaki et al. 2002). PDBu pretreatment (200 nm, 2 min) caused a leftward shift in the voltage-dependent activation curve, decreasing the V_1/2 by ∼60 mV (Fig. 6A–C) without affecting the gating charge (0.68 ± 0.06). In addition to altering voltage sensitivity, PKC increased the fractional G_max (normalized to pH 6.0) from 0.51 ± 0.04 to 0.80 ± 0.06 (Fig. 6A, B and D). In contrast, PDBu failed to modify the voltage relationship in cells transfected with a mutant TRPV1, lacking key PKC phosphorylation sites (S502A/S800A) (Numazaki et al. 2002). Figure 6C shows that at 25°C, the TRPV1 mutant displays a similar voltage dependency both before and after treatment with PDBu (V_1/2 = 121.4 ± 6.13 mV, 117.2 ± 4.6 mV and 114.9 ± 5.5 mV, respectively, for WT TRPV1, mutant TRPV1, and mutant TRPV1 plus PDBU). Thus, direct phosphorylation of TRPV1 by PKC is a requirement for the effects observed with PDBu. These data show that thermal and PKC stimuli, like TRPV1 ligands, increase both the efficacy and sensitivity of voltage-dependent activation.

Figure 6. PKC-dependent phosphorylation enhances sensitivity and efficacy of voltage activation for TRPV1.

A, representative current traces in response to a family of voltage steps on cells pretreated with PDBu (500 nm) in the absence or presence of protons (pH 6.0). B, the representative Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. C, summary of the shifts in the V_1/2 from control conditions (cells not treated with PDBu) at 20°C (n = 5) or at 25°C when comparing the S502A/S800A mutant (n = 4). D, summary of the fractional G_max normalized to pH 6.0 (n = 5). **P < 0.005.

Capsazepine reduces the sensitivity and efficacy of voltage activation for TRPV1

The prototypic TRPV1 antagonist, capsazepine, and several recently identified compounds are competitive inhibitors of capsaicin binding (Bevan et al. 1992). Significantly, these compounds to varying degrees also inhibit activation by other stimuli including heat (Tominaga et al. 1998), protons (Tominaga et al. 1998; Ahern et al. 2005) and cations (Ahern et al. 2005) suggesting that they may act in an allosteric manner to reduce channel gating, that is, as ‘inverse agonists’. If TRPV1 agonists act to enhance voltage-dependent activation then we reasoned that inverse agonists would produce the opposite effect. Indeed, Fig. 7A and B shows that capsazepine inhibited voltage-dependent activation in a dose-dependent and reversible manner. Capsazepine reduced the G_max by ∼45% with an IC₅₀ of 0.46 ± 0.03 μm and increased the V_1/2 by up to 70 mV with an EC₅₀ of 0.29 ± 0.15 μm (Fig. 7C and D). Although values for the G_max and V_1/2 at high concentrations of capsazepine (≥ 300 nm) were obtained from extrapolated Boltzmann fits, we were nonetheless able to obtain consistent results when all parameters were allowed to vary during the fit. We also explored the effects of capsazepine (5 μm) in cells pretreated with PDBu. Under these conditions voltage activation clearly reaches a maximum both before and after treatment with capsazepine. These data show clearly that capsazepine reduces the G_max to 0.65 ± 0.02 of the maximal response observed with PDBu alone (Fig. 8, n = 4). Further, capsazepine increased the V_1/2 by 116.6 ± 10.1 mV without an appreciable effect on gating charge (0.62 ± 0.02). Thus, agonists and inhibitors of TRPV1 regulate the voltage dependence of activation (both sensitivity and efficacy) in a reciprocal manner.

Figure 7. Capsazepine inhibits voltage-dependent activation of TRPV1.

A, representative current traces in response to a family of voltage steps in the absence or presence of various concentrations of capsazepine (CPZ). B, the representative Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse. C, the Hill plot for the reduction of the G_max by various concentrations of capsazepine (maximal inhibition = 0.53 ± 0.02, IC₅₀ = 0.456 ± 0.029 μm, Hill coefficent (h) = 2.38 ± 0.36). D, the Hill plot for the shifts in V_1/2 values by various concentrations of capsazepine (maximal shift = 72.0 ± 2.0 mV, EC₅₀ = 0.292 ± 0.023 μm, h = 1.49 ± 0.15). For C and D the data points are the mean of 6–8 separate experiments.

Figure 8. Capsazepine inhibits voltage-dependent activation after PKC-dependent phosphorylation.

A, representative current traces in response to a family of voltage steps in cells pretreated with PDBu (500 nm) in the absence or presence of capsazepine (5 μm). B, the representative Boltzmann fits to the conductances obtained from steady state currents at the end of the pulse.

An allosteric model accounts for temperature, agonist and antagonist modulation of TRPV1 and TRPM8

Voets and colleagues proposed a two-state model to explain voltage and temperature gating of TRPM8 and TRPV1 (Voets et al. 2004), and an eight-state Monod–Wyman–Changeux model to explain menthol and voltage interactions of TRPM8 (Voets et al. 2007). In both models voltage and temperature drive transitions between closed and open states. Thus, these models predict that voltage/temperature alone can produce maximal activity and that a ligand-bound channel still opens in a voltage-sensitive manner. That is, voltage is sufficient and necessary to gate these channels. Our data do not support these models since we find that voltage alone is a submaximal stimulus and that ligands produce voltage-independent gating. On the other hand, Brauchi et al. (2004) have proposed an allosteric model to explain activation of TRPM8 by voltage and temperature. A key feature of this model is the existence of independent voltage and temperature sensors that are coupled to channel gating (see Fig. 9A). Voltage and temperature drive transitions between resting and activated states of these sensors with equilibrium constants J and K, respectively, where J = J₀exp(zFV/RT) and J₀ is the equilibrium constant at 0 mV, and K = exp(−(ΔH − TΔS)/RT). In turn, the voltage and temperature sensors are coupled to channel gating by allosteric coupling constants C and D, respectively, and to each other, by E. Assuming that all voltage sensor(s) in a channel operate in unison (and similarly for temperature sensors) then this model generates four open and four closed states. The open probability with E = 1 is then given by:

(1)

which simplifies to:

(2)

We found that this allosteric model could reproduce temperature and voltage dependent activation of TRPV1. The smooth lines in Fig. 9B show simultaneous fits to eqn (2) for data at 20°C, 25°C and 32°C. These fits yielded parameters: L = 4.2 × 10⁻⁴, J₀ = 0.0169, ΔH = 205 kJ mol⁻¹, ΔS = 615 J mol⁻¹ K⁻¹, D = 1100, C = 23367 and z = 0.6.

Figure 9. Allosteric models for gating of thermo TRP channels.

A, an 8-state channel schematic diagram for allosteric modulation between voltage and temperature activation. See text for details. B, average G/G_max versus voltage plots with simultaneous fits to the 8-state model for TRPV1 (n = 4–6). C, a 16-state channel schematic diagram for allosteric modulation between voltage, temperature and ligand activation. See text for details. D, average G/G_max versus voltage plots with simultaneous fits to the 16-state model for TRPV1 in the presence or absence of capsaicin (25°C, n = 4). E, average G/G_max versus voltage plots with simultaneous fits to the 16-state model for TRPM8 in the presence or absence of menthol (25°C, n = 4). F, average G/G_max versus voltage plots with simultaneous fits to the 16-state model for TRPV1 in the presence or absence of capsazepine (25°C, n = 4).

To model the effects of ligand activation we introduced ligand sensitivity into the model (Fig. 9C) similar to that proposed for Ca²⁺-activated K⁺ channels (Horrigan & Aldrich, 2002), with transitions between bound and unbound states of this sensor governed by equilibrium Q, where Q = [ligand]/K_D. This ligand binding domain is allosterically coupled to gating by the coupling constant, P, and to voltage and temperature sensors by constants E₂ and E₃. The number of molecules of capsaicin required to gate a channel is unknown. However, assuming, for simplicity, that all binding sites(s) need to be occupied for activation then this scheme would produce eight open and eight closed states, and assuming E₂ and E₃ are 1, then open probability is given by:

(3)

Figure 9D show that fits to this equation (using previously obtained parameters for L, J₀, ΔH, ΔS, D, C and z) reproduced effects of capsaicin (50 nm and 10 μm) on voltage activation at 25°C. The additional parameters were: K_D = 1 × 10⁻⁵ m and P = 750. We similarly modelled menthol (0, 0.8 and 6 mm) activation of TRPM8 (Fig. 9E). Best fits yielded the parameters: L = 1.34 × 10⁻⁶, J₀ = 0.011, ΔH = −223 kJ mol⁻¹, ΔS = −738 J mol⁻¹ K⁻¹, D = 1463 C = 1057, K_D = 10 mm, P = 1176, and z = 0.71 (control), 0.55 (0.8 mm) and 0.3 (6 mm). Note that these ‘apparent’ K_D values are considerably higher than the EC₅₀ values reported for capsaicin (∼500 nm) and menthol (∼100 μm). However, they agree with our experimental observations that micromolar and millimolar concentrations, respectively, of capsaicin and menthol are required to fully activate TRPV1 and TRPM8 at very hyperpolarized potentials (−200 mV). These data reflect the model's prediction that gating is the product of voltage, temperature and ligand sensing, and therefore the intrinsic sensitivity of each sensor will be lower than commonly measured under normal experimental conditions. Finally, we modelled the effects of capsazepine (0.1, 0.3, 1 and 10 μm) on voltage activation of TRPV1, by varying the ligand sensor parameters K_D and P, and keeping the parameters L, J₀, ΔH, ΔS, C, D and z as above. Figure 9F shows the best fits to the data with parameters: K_D = 1.3 × 10⁻⁷ m, P = 0.11. Thus, the model predicts that capsazepine inhibits gating by reducing the allosteric coupling factor P < 1, and this is consistent with capsazepine acting as a negative, allosteric regulator or inverse agonist.

It is important to note that this allosteric model depends on a number of simplifying assumptions, including (1) that ligand, voltage and temperature sensors operate virtually independent of each other, and (2) that each channel can be treated as having only one sensor (or individual sensors that act in unison). The independence of ligand, voltage and temperature sensors may not be strictly accurate. For example, we found that menthol produced a reduction in the apparent gating charge. The physical basis for this is unclear, but it is possible that menthol directly interacts with the voltage sensor. Indeed, charge-neutralizing mutations in the S4 and S4–S5 linker region of TRPM8 can modulate both the gating charge and menthol binding (Voets et al. 2007). In any case, this model to a reasonable degree does account for the multimodal regulation of TRP channels by voltage, temperature, agonists and antagonists.

Discussion

Voltage-independent gating of TRP channels

A voltage-sensitive mechanism is proposed to underlie gating of thermo-sensitive TRP channels (Voets et al. 2004; Nilius et al. 2005; Talavera et al. 2005). According to this hypothesis, these channels are intrinsically voltage sensitive and thermal and chemical stimuli act to increase this voltage sensitivity (producing a leftward shift in the voltage–activation relationship). If voltage is an intrinsic mediator of gating, then one would expect that voltage alone (at some finite potential) would elicit maximal channel activity independently of other stimuli. However, our results show that voltage does not fully activate TRPV1 and TRPM8. At room temperature (20–25°C), voltage elicits a partial activation, ∼50–60% of the maximal activity obtained with capsaicin (10 μm) or menthol (0.8 mm). Single channel recordings of TRPV1 confirmed that capsaicin increased the maximal open probability evoked by depolarization. Similarly, a voltage-independent component of TRPV1 and TRPM8 currents was evident at negative potentials in the presence of higher concentrations of agonists. Further, the ultrapotent agonist, RTX, rendered TRPV1 essentially voltage independent over a 400 mV range. Thus, these data demonstrate that TRP channels can gate independently of voltage. Further, voltage sensitivity does not appear to subserve gating by other modalities. We found that other stimuli known to regulate TRPV1 function including heat, PKC, or protons enhanced both the efficacy and sensitivity of voltage-dependent activation. In particular, it is notable that heat and voltage produce an additive response arguing against the existence of a shared molecular mechanism. Conversely, the TRPV1 antagonist, capsazepine, produced the opposite effect, reducing the efficacy and the sensitivity of voltage activation with near equal potency. These data show that both agonists and antagonists exert similar effects (either positive or negative) on TRP voltage dependence and indicate that thermal and chemical stimuli can act independently of voltage. Our results differ from Voets et al. (2004) who showed that temperature and ligands selectively altered the sensitivity of voltage activation for TRPV1 and TRPM8 without affecting the maximal open probability. The reasons for this discrepancy are not clear; however, it is possible that these authors employed a normalization procedure that obscured any changes in efficacy. In addition, Voets et al. (2004) used ligand concentrations that would have been insufficient to generate the voltage-independent gating observed in the current study. On the other hand, our data agree with Brauchi et al. (2004) who showed that cooling enhanced the maximal open probability of TRPM8 produced by depolarization.

An allosteric model explains voltage, temperature, agonist and antagonist modulation of thermo-TRPs

Our data suggest that voltage acts via a mechanism that is separate from, but tightly coupled to, activation by other stimuli. In support of this hypothesis we present an allosteric model similar to that proposed by Brauchi et al. (2004). This model accounts for the effects of temperature (20–32°C) on voltage-dependent activation of TRPV1. Furthermore, we have expanded this model to explain ligand modulation of both TRPV1 and TRPM8. This adapted model explains capsaicin- and menthol-induced shifts in voltage sensitivity and voltage-independent gating (increase in G_max and G_min). In addition, it explains how inverse agonists, such as capsazepine, modulate voltage dependence in a reciprocal manner to agonists, reducing both the efficacy and sensitivity of voltage activation. These findings provide further support for the idea that voltage-, temperature- and chemical-sensing mechanisms are separate but coupled. Indeed, gating of TRPV1 is known to be highly co-operative. Protons, cations, vanilloids and temperature regulate TRPV1 in a super-additive or synergistic manner. These stimuli are believed to act at distinct domains. Extracellular protons and cations interact with acidic residues near the mouth of the pore (Jordt et al. 2000; Ahern et al. 2005). Vanilloids may interact at several intracellular domains – including Tyr-550 (Gavva et al. 2004), and residues between the 2nd and 3rd transmembrane loop (Jordt & Julius, 2002). Thermal activation appears to require the C-terminus. Exchanging the C-termini of TRPV1 and TRPM8 reverses their temperature sensitivity (Brauchi et al. 2006). Furthermore, distal C-terminal truncations (31 and 42 amino acid residues) in TRPV1 can reduce the temperature threshold without affecting capsaicin or proton responsiveness (Vlachova et al. 2003), although another study finds no change in temperature sensitivity with the deletion of 88 amino acids (Liu et al. 2004). Charged residues in the 4th transmembrane region (S4) and the S4–S5 linker of TRPM8 that are important for gating charge may participate in voltage sensing (Voets et al. 2007). Significantly, two recent studies provide strong evidence for there being separate voltage and temperature sensors. Deletions in the C-terminus of TRPM8 (Brauchi et al. 2007) or single amino acid mutations in the S6 region of TRPV1 (Susankova et al. 2007) can greatly reduce temperature sensitivity while leaving voltage activation largely intact.

Physiological relevance of allosteric gating

An allosteric model agrees with observations that TRPV1 and TRPM8 can convert multiple physical, chemical and thermal stimuli into channel gating and that this occurs in a synergistic manner. Of physiological importance we describe the effects of PKC – a key inflammatory signalling pathway in nociceptors – on TRPV1 voltage sensitivity. PKC-dependent phosphorylation increases the maximal TRPV1 current and also reduces the V_1/2 by 60 mV, confirming our earlier observations that PKC can serve to directly gate the channel during depolarization (Premkumar & Ahern, 2000). Interestingly, these effects of PKC are equivalent to a ∼7°C increase in temperature. However, PKC and temperature clearly operate via separate mechanisms since the PKC effect is abolished in mutant (S502A/S800A) receptors while temperature sensing is intact. At 25°C, PKC reduces the threshold for voltage activation to about −20 mV to 0 mV, and thus significant activation of TRPV1 could only come into play briefly during cell spiking. However, this value will be greatly reduced at physiological temperature (37°C), such that the threshold for voltage activation would be reduced below the normal resting membrane potential of −60 mV. Thus, these results are consistent with observations that direct activation of TRPV1 occurs at ∼33°C following PKC stimulation (Sugiura et al. 2002). In molecular terms, we suggest this arises from the coupled activation of discrete voltage, temperature, and PKC sensors (with potential contribution from proton/cation sensors as well).

In contrast, we show that capsazepine acts as an inverse agonist at TRPV1, producing opposite effects to agonists and PKC on voltage sensitivity. Notably, the inhibitory effects of capsazepine are cancelled by PKC. Therefore, these observations may explain the variable efficacy of this compound. For example, several studies of rat TRPV1 show that capsazepine can effectively inhibit proton- and cation-activated currents (Tominaga et al. 1998; Ahern et al. 2005), whereas other studies show that capsazepine does not block proton-evoked responses (McIntyre et al. 2001; Gavva et al. 2005). It is possible that differential phosphorylation of TRPV1 could partly account for this discrepancy. Thus, in the physiological sense these results predict that antagonists may be less effective under inflammatory conditions that promote channel phosphorylation.