Mechanical Force and Cytoplasmic Ca²⁺ Activate Yeast TRPY1 in Parallel

Abstract

The ability to sense mechanical and osmotic stimuli is vital to all organisms from mammals to bacteria. Members of the transient receptor potential (TRP) ion-channel family have attracted intense attention for their involvement in mechanosensation. The yeast homologue TRPY1 can clearly be activated by hypertonic shock in vivo and by stretch force under patch clamp. Like its animal counterparts, TRPY1 is polymodal, being gated by membrane stretch force and by cytoplasmic Ca²⁺. Here, we investigated how these two gating principles interact. We found that stretch force can induce some channel activation without cytoplasmic Ca²⁺. Tens of micromolar Ca²⁺ greatly enhance the observed force-induced activities, with open probabilities following well the Boltzmann distribution, in which the two gating energies are summed as exponents. To map this formalism to structures, we found Ca²⁺-binding proteins such as calmodulin or calcineurin to be unnecessary. However, removing a dense cluster of negative charges in the C-terminal cytoplasmic domain of TRPY1 greatly diminishes the Ca²⁺ activation as well as its influence on force activation. We also found a strategic point upstream of this charge cluster, at which insertion of amino acids weakens Ca²⁺ activation considerably but leaves the mechanosensitivity nearly intact. These results led to a structure–function model in which Ca²⁺ binding to the cytoplasmic domain and stretching of the membrane-embedded domain both generate gating force, reaching the gate in parallel.

Introduction

Mechanical senses are common to all creatures, from hearing, touch and balance in animals to gravitropism in plants and osmotic regulation in fungi and bacteria (Kung 2005; Martinac et al. 2008). These processes involve converting different forms of mechanical stimuli into electrical signals by force-sensitive molecular transducers. Such transducers in prokaryotes are exemplified by the extensively characterized bacterial mechanosensitive ion channels of large conductance (MscL) and small conductance (MscS), which protect bacteria from osmotic down-shock (Martinac et al. 2008). However, in eukaryotes, the molecular identities of such force-transducing channels are often elusive. Members of the transient receptor potential (TRP) ion channel family are apparently involved in mechanosensation. For example, human TRPP1 and TRPP2 are proposed to sense fluid flow in kidney, rat TRPV4 responds to hypotonic swelling, fly TRPVs Nanchung and Inactive function in hearing and the worm TRPV (Osm-9) is required for nose touch and hypertonic stress (Christensen and Corey 2007). Yet, whether these animal channels are the first-line force sensors or downstream amplifiers or serve other roles in mechanosensation remains unclear. In contrast, the yeast TRP channel, TRPY1, expressed in the vacuolar membrane, is clearly a mechanically sensitive molecule. It is activated not only by osmotic up-shock in vivo (Denis and Cyert 2002) but also by membrane stretch under patch clamp (Zhou et al. 2003), making it a useful model for eukaryotic primary mechanotransducers. In addition, TRPY1 requires no heterologous expression, exhibits a large 300-pS signal over noise (Palmer et al. 2001; Zhou et al. 2003) and can easily be manipulated by genetics (Su et al. 2007; Zhou et al. 2007).

Most TRP channels are polymodal. They respond to a variety of stimuli including voltage, temperature, pH and chemicals, besides mechanical force (Venkatachalam and Montell 2007). One of the most physiologically important second messengers, intracellular free Ca²⁺, has been reported to modulate many TRP channels. For instance, TRPM2 requires Ca²⁺ for activation (McHugh et al. 2003). TRPM4 and TRPM5 both exhibit Ca²⁺ activation and subsequent inactivation, with calmodulin being required for TRPM4 but not TRPM5 (Hofmann et al. 2003; Launay et al. 2002; Nilius et al. 2005). TRPA1 is proposed to be directly activated by Ca²⁺ through an N-terminal EF-hand structure (Doerner et al. 2007; Zurborg et al. 2007), although a recent study has now challenged EF-hand’s being the Ca²⁺ sensor (Wang et al. 2008). TRPY1 has been known to be activated by cytoplasmic Ca²⁺ since the registry of its channel activity (Bertl and Slayman 1990; Wada et al. 1987), long before its molecular identification (Palmer et al. 2001). Since the discovery of its mechanosensitivity under patch clamp, the action of Ca²⁺ is now thought to be an integral component in a Ca²⁺-induced Ca²⁺ release (CICR) positive feedback that amplifies its response to mechanical force (Zhou et al. 2003).

A central issue in the study of polymodal ion channels is to clarify how different stimuli reach the same channel gate. Understanding how different stimuli act and interact in gating can lead to clear insights. For example, distinguishing the action of menthol from coolness on TRPM8 (Bandell et al. 2006) and that of warmth from camphor on TRPV3 (Grandl et al. 2008) have shed light on chemo- and thermosensitivity. To use TRPY1 to understand the mechanosensitivity in eukaryotes, we need to sort out the relation between stretch force and Ca²⁺ in gating this channel, as this study sets out to do.

Materials and Methods

Yeast Strains and Cultures

Except for the calmodulin mutant and its parent, the wild-type parental strains were BY4742 (MATα his3Δ leu2Δ lys2Δ ura3Δ) and yvc1Δ mutant (a chromosomal knockout strain, YOR088 W; yvc1::km). For the calmodulin experiments, calmodulin wild-type strain CMD1 ⁺ and calmodulin mutant strain cmd1-6 were in the CYR1 background and were kindly provided by Dr. T. N. Davis (Geiser et al. 1991). For electrophysiology, standard yeast medium with yeast extract, peptone and dextrose (YPD) was used. For luminometric experiments, complete minimal medium with dextrose (CMD) or a modified version with ammonium chloride replacing ammonium sulfate (DCD) was used (Denis and Cyert 2002; Su et al. 2007; Zhou et al. 2007).

Mutagenesis

All site-directed mutagenesis experiments were performed using the standard overlap extension PCR method (Ho et al. 1989). Different PCR fragments carrying the corresponding mutations were cloned into HindIII/XhoI sites of YVC1 (the gene that encodes TRPY1 protein) in a prYVC plasmid carrying a GPD promoter and a URA3 marker. All mutants were verified by sequencing. Sequenced plasmids were linearized by EagI/MluI before transforming into yvc1Δ for gene integration. Transformed yeast mutant strains with correct markers were sequenced again from the genome to verify the yvc1-mutant integrants. All experiments were performed on the wild type or these integrants.

Luminometry

Luminometric measurements were conducted as previously described (Batiza et al. 1996; Denis and Cyert 2002). Briefly, yeast cells were transformed with pEVP11/AEQ plasmid bearing the apoaequorin gene and a LEU2 marker. Transformants were selected on CMD-LEU plates. Colonies were picked and precultured in DCD-LEU medium at 30°C for 24 h and then subcultured in fresh medium supplemented with 2 μM coelenterazine for 24 h in the dark, before the up-shock-induced luminescence was measured. Cells (20 μl) were shocked with 200 μl hypertonic solutions (0.75–2.5 M NaCl, 50 mM MES [pH 7], 25 mM EGTA), and the luminescence was recorded with a Berthold LB 9507 luminometer (Berthold Detection Systems, Oak Ridge, TN).

Solutions

For patch-clamp recordings, the pipette solution contained the following (in mM): 150 KCl, 5 MgCl₂, 100 sorbitol, 10 HEPES-KOH, pH 7.2. The bath solution contained 150 KCl, 5 MgCl₂, 100 sorbitol, 10 HEPES-KOH (pH 7.2) and different Ca²⁺ concentrations adjusted by CaCl₂ and EGTA. For [Ca²⁺] above 10⁻⁵ M, CaCl₂ was directly added to EGTA-free bath solution. For [Ca²⁺] below 10⁻⁵ M, appropriate amounts of CaCl₂ recommended by MaxChelator (www.stanford.edu/~cpatton/maxc.html/) were added to the bath solution containing 2 mM EGTA.

Electrophysiology

Yeast vacuoles were generated as previously described (Su et al. 2007; Zhou et al. 2007). All patch-clamp recordings were performed with Clampex 10.0 (Axon Instruments, Union City, CA). Data were sampled at 5 kHz and filtered at 1 kHz. Glass pipettes (100 μl; Drummond Scientific, Broomall, PA) were pulled with a P-97 Micropipette Puller (Sutter Instrument, Novato, CA) to 3–5 MΩ tip resistance. With negative pressure in the pipette, 2–3 GΩ seals were formed before being broken by transient voltage pulses to reach the whole-vacuole mode. Whole-vacuole mode patches were subsequently excised into the cytoplasmic side–out mode. Data were acquired at −30 mV voltage, with the vacuolar side being positive. Bath perfusion was used to change [Ca²⁺]. Mechanical stimulation was applied as pressure exerted with a 60-ml syringe and gauged with an Omega (Stamford, CT) PX140 pressure sensor. Experiments were performed at room temperature (21–23°C). Data were analyzed using Clampfit 10.0 (Axon Instruments) and Microsoft (Redmond, WA) Excel. Data are presented as the mean ± SD. Significant differences were tested using Student’s unpaired t-test (P < 0.05).

Curve Fitting for P _o vs. Pressure at Different [Ca²⁺]

We first generated the predictions from the parallel model and then assessed how well they fit the data. The gating energy for force activation is ΔG _Force = (γ−γ₀) ΔA, where γ is the membrane tension, γ₀ the tension for half maximum activation and ΔA the change in area between the closed and the open states. The gating energy for Ca²⁺ activation is ΔG _Ca = k _B T ln (K _m [Ca²⁺]ⁿ), where k _B T has its usual meaning, K _m is the equilibrium constant of Ca²⁺ binding and n is the number of Ca²⁺ ions bound per channel. The parallel model dictates that P _o /(1−P _o ) = exp [(ΔG _Force + ΔG _Ca)/k _B T] = exp {[(γ−γ₀) ΔA + k _B T ln (K _m [Ca²⁺]ⁿ)]/k _B T}. The tension (γ) is directly proportional to the applied pressure (P) according to the Young–Laplace equation as the geometry of the patch is constant within one patch. Thus, further transformation converted the Boltzmann equation into P _o /(1−P _o ) = exp (a P + b). For each [Ca²⁺], the rises of P _o over the range of P in Fig. 1a were fitted to this equation with a and b being the fitting parameters (Fig. 1b). In this equation, the a values are dictated by ΔA and by the geometry of the patch, both of which remain constant within the same patch. Therefore, constancy of the a values at different [Ca²⁺] is supportive of the parallel model. On the other hand, the b values should correlate linearly with log [Ca²⁺], if the parallel model is correct. The a values in the fits from different [Ca²⁺] were evaluated for constancy and the b values were evaluated for linearity against log [Ca²⁺] (Fig. 1d). The curve fitting was carried out using SigmaPlot 10 (Systat, Richmond, CA).

Fig. 1

Relationship between TRPY1’s mechanosensitivity and Ca²⁺ sensitivity. a One representative recording from repeated studies displays channel activities in response to four different pressures (60, 120, 180 and 240 mmHg, left to right) at four different [Ca²⁺] (10⁻⁹, 10⁻⁷, 10⁻⁵ and 10⁻³, bottom to top). All patch-clamp recordings were performed on isolated vacuoles in excised cytoplasmic side–out configuration at −30 mV (vacuolar side positive). In the absence of Ca²⁺ (10⁻⁹ M), channels are weakly activated by strong pressure (240 mmHg) (bottom trace). Above 10⁻⁷ M Ca²⁺, force-induced responses become evident. At a high [Ca²⁺] (10⁻³ M), robust basal Ca²⁺-induced activities mask most of the pressure responses. b P _o vs. pressure at different [Ca²⁺], from the data shown in a fitted with Boltzmann curves. c Comparison of P _o with or without the application of 240 mmHg pressure at different [Ca²⁺] shows a synergistic effect. Note, the pressure-induced P _o increase at 10⁻⁵ M Ca²⁺ is much larger than that at 10⁻⁷ M Ca²⁺. d Plot of the b value vs. log [Ca²⁺]. The b value, a curve-fitting parameter, shows a linear correlation with log [Ca²⁺] predicted by the parallel model. See “Materials and Methods”

Results

Activation of TPRY1 by Membrane Stretch Force and by Ca²⁺

We assessed the relation between stretch-force activation and Ca²⁺ activation repeatedly using patch clamp in excised cytoplasmic-side-out configuration. Results from a representative experiment that covers a large range of [Ca²⁺] and stretch force are shown in Fig. 1. In the virtual absence of Ca²⁺ (10⁻⁹ M, Fig. 1a, bottom trace), there is nearly no spontaneous channel activity. Strong membrane stretches (e.g., by 240 mmHg pressure), however, can activate the channels, albeit weakly, showing that the mechanosensitivity of TRPY1 does not have an absolute dependence on the presence of cytoplasmic Ca²⁺. This mechanosensitivity is consistent with the function of TRPY1 in vivo, which must start to open upon osmotic stress from a resting steady state with very little free Ca²⁺ in the cytoplasm. Pressures higher than 240 mmHg tend to break the patch seals and cannot practically be examined in our system. Increasing [Ca²⁺] promotes the baseline open probability (P _o ) in the absence of applied pressure, with a sharp rise between 10⁻⁴ and 10⁻³ M Ca²⁺, from P _o  ~ 0.1 to P _o  ~ 0.6 (Fig. 1a). Furthermore, the pressure-induced P _o rises steeply with the increase in [Ca²⁺] (Fig. 1a, top three traces). Channel activation by applied pressure is plotted in Fig. 1b for different [Ca²⁺]. A robust pressure-induced response starts to be evident at [Ca²⁺] above 10⁻⁵ M (Fig. 1a, second trace, b), and when [Ca²⁺] approaches 10⁻³ M, this response is masked by the prominent Ca²⁺-induced basal activities (Fig. 1a, top trace, b). Comparison of P _o with or without the application of pressure (240 mmHg) at different [Ca²⁺] shows an apparent synergistic effect (Fig. 1c). That is, the pressure-induced P _o increase at 10⁻⁵ M Ca²⁺ is much larger than that at 10⁻⁷ M Ca²⁺, an effect that cannot be ascribed to the Ca²⁺-induced P _o increase alone.

In simplified models, the gating force from Ca²⁺ binding and that from membrane stretch can work in series or in parallel. In a serial model, Ca²⁺ binding may be viewed as a prerequisite step for stretch-force sensing. Such a model is not supported by our data because TRPY1 remains mechanically sensitive in the virtual absence of Ca²⁺ (Fig. 1a, bottom trace). In a parallel model, two stimuli are independently sensed and transmitted to the gate. The integration of the two influences can be evaluated by the sum of the two energy exponents. As given in “Materials and Methods,” this model converts the Boltzmann equation into P _o /(1−P _o ) = exp (a P + b). This equation was used to fit the actual data points shown in Fig. 1b, with a and b being the fitting parameters. The a values from the fits were found to be 0.012, 0.010 and 0.008 for 10⁻⁷, 10⁻⁵ and 10⁻³ M Ca²⁺, respectively. For 10⁻⁹ M Ca²⁺ the a value could not be accurately given due to the extreme low activity. Thus, the a values show a reasonable constancy, being 0.01 ± 0.002 (mean ± SD), supporting the parallel model. The b values were plotted with log [Ca²⁺] in Fig. 1d, showing b vs. log [Ca²⁺] being linear, also in favor of a parallel model. In this model, the apparent synergism between force activation and Ca²⁺ activation in TRPY1 can be attributed to the sum of the two exponents of gating energies.

Common Ca²⁺-Binding Proteins such as Calmodulin and Calcineurin are not Required for Ca²⁺ Activation of TRPY1

The synergistic effect between force sensitivity and Ca²⁺ sensitivity encouraged us to further investigate the Ca²⁺ activation of TRPY1 in structural terms. Since many known Ca²⁺-activated ion channels are regulated through Ca²⁺-binding protein partners, we asked whether this is the case with TRPY1. Osmotic up-shocks activate TRPY1 in vivo to release Ca²⁺ from the vacuole into the cytoplasm, which can be measured by transgenic aequorin (Batiza et al. 1996; Denis and Cyert 2002). We have surveyed a collection of 4,810 yeast single-gene deletion strains for their TRPY1-dependent osmotic up-shock responses in search for possible interacting elements that regulate TRPY1 (Loukin et al. 2008). Deletions of most Ca²⁺-related proteins, such as the well-known CaM kinases, do not significantly affect TRPY1’s hypertonic response. The calcineurin deletant cnb1, which lacks the Ca²⁺-binding calcineurin B subunit, however, has a higher vacuolar [Ca²⁺] and consequently an exaggerated Ca²⁺ response but no clear alteration of TRPY1’s channel behaviors under patch clamp (Loukin et al. 2008). Therefore, it is unlikely that calcineurin regulates TRPY1’s Ca²⁺ gating process. One caveat of the screen is that vital genes are not included in the deletion collection and the calmodulin gene is among them.

Calmodulin is a prominent candidate since it is known to be the Ca²⁺-binding subunit in many channels including several TRPs (Nilius et al. 2004, 2005; Saimi and Kung 2002). The Saccharomyces cerevisiae genome has a single calmodulin gene, CMD1. CMD1 is vital. However, the mutant cmd1-6, with multiple mutations engineered to remove all of CMD1’s ability to bind Ca²⁺, grows normally (Geiser et al. 1991). We therefore used cmd1-6 to test whether Ca²⁺-calmodulin is needed to activate TRPY1. The Ca²⁺ responses upon 1.5 M NaCl osmotic up-shock from cmd1-6 and its parent CMD1 ⁺ are not significantly different (Fig. 2a). The normalized peak responses to up-shocks, ranging 0.75–2 M NaCl, are also not significantly different between the two (Fig. 2b). We also directly examined the activities of TRPY1 channels in cmd1-6 and CMD1 ⁺ vacuoles with patch clamp. Ensemble currents at different [Ca²⁺] appear similar between the two (Fig. 2c, d). Mechanosensitivity of the channels, as assessed in excised cytoplasmic side–out patches under 120 mmHg pressure, also appears to be unaffected by the cmd1-6 mutation (Fig. 2c, d). Although sizable variations may obscure individual differences (note SD Fig. 2d), it seems clear that Ca²⁺-calmodulin is not a direct partner in TRPY1’s Ca²⁺ activation. Thus, our screen and mutant analyses suggest that Ca²⁺ probably directly activates TRPY1.

Fig. 2

Ca²⁺-calmodulin is not required for the Ca²⁺ activation of TRPY1. a Osmotic up-shock-induced Ca²⁺ releases of calmodulin mutant cmd1-6 (gray) and its parent CMD1 ⁺ (black) are not significantly different. Arrow indicates the addition of 1.5 M NaCl. Luminescence was gauged as relative light units (RLUs) with transgenic aequorin. b Normalized peak responses to the shocks of different concentrations of NaCl (0.75–2 M) of cmd1-6 (gray) and CMD1 ⁺ (black) appear similar. c Patch-clamp recordings of CMD1 ⁺ (top traces) and cmd1-6 (bottom traces) at 10⁻⁴ M Ca²⁺ (left traces) and 10⁻³ M Ca²⁺ (right traces) do not differ prominently. Pressure of 120 mmHg was applied as indicated. d Averaged responses show no statistically evident difference between the calmodulin mutant (n = 10) and its parent (n = 6). Mean ± SD

A C-Terminal Negative Charge Cluster is Needed for TRPY1’s Ca²⁺ Activation

We next tried to find structural elements within the TRPY1 protein that may be required for Ca²⁺ activation. Analyses showed no well-recognized Ca²⁺-binding motifs such as EF-hand or RCK domains in TRPY1. A cluster of acidic amino acids, however, can be found in the presumed C-terminal cytoplasmic domain of the predicted TRPY1 subunit structure (Fig. 3a, residues 555–582). The strongest negative density is four consecutive aspartates (⁵⁷³DDDD⁵⁷⁶), and we engineered an 11-residue deletion, Δ570–580, removing the ⁵⁷⁰GYLDDDDGLFS⁵⁸⁰ peptide to examine the possible functional significance of these negative charges. Luminometric measurements showed that Δ570–580 has no evident response above TRPY1 knockout control (Fig. 3b). We further used patch clamp to directly compare the channel activities of Δ570–580 with those of the wild type. In multiple studies using excised cytoplasmic-side–out patches, the mutant activities are found to be drastically diminished. As shown in Fig. 3c (left), when bathed in 10⁻⁴ M Ca²⁺, the steady-state activities as well as the pressure-induced activities of the wild type are evident but the activities of Δ570–580, though present, are almost completely lost. In 10⁻³ M Ca²⁺ (Fig. 3c, right), the wild-type channels are highly active and approach maximum when pressure is applied. The Δ570–580 again showes significantly reduced basal activity, although activation by stretch force is evident. It should be noted that, although weak, residual Ca²⁺ activation clearly remains in Δ570–580. The activities of Δ570–580 at 10⁻⁴ M Ca²⁺ (Fig. 3c, lower left) are similar to those of wild type at 10⁻⁹ M Ca²⁺ (Fig. 1a, bottom trace), whereas the activities of Δ570–580 at 10⁻³ M Ca²⁺ resembles those of the wild type at 10⁻⁷ M Ca²⁺ (Fig. 1a, third trace). This retained Ca²⁺ activation in Δ570–580 accounts for its increased force responses at a higher [Ca²⁺] (10⁻³ M). The statistical loss of P _o in Δ570–580 is shown in Fig. 3d for comparison. This loss may be underestimated since the true number of conducting units in the patch cannot be estimated for the mutant due to its low activities. It is clear that Δ570–580 not only largely diminishes Ca²⁺ activation but also weakens the synergistic effect on force activation. Thus, the lack of in vivo responses in Δ570–580 (Fig. 3b) could result from not enough activities to trigger and maintain a CICR feedback pathway in the presence of Ca²⁺ sequestering machineries in yeast cells. These results show that Ca²⁺ activation likely initiates from the direct binding of Ca²⁺ to TRPY1 itself and the cluster of negative charges here may be necessary for that binding.

Fig. 3

A dense cluster of negative charges in the C-terminal cytoplasmic domain is required for Ca²⁺ activation. a Schematic representation of a single channel subunit with six transmembrane helices, N- and C-terminal cytoplasmic domains. A cluster of negative charges in the C-terminal cytoplasmic domain (residues 555–582) is shown, with negative charges in bold. A deletion of 11 amino acids (Δ570–580) within this cluster encompassing the strongest negative charge density ⁵⁷³DDDD⁵⁷⁶ is indicated by a dashed line. b In vivo osmotic up-shock-induced luminometric responses of wild type and Δ570–580. NaCl, 1.5 M, elicits a robust signal from wild type, whereas in Δ570–580 there is no clear response above TRPY1 knockout control. c Patch-clamp recordings of wild type (WT, top traces) and Δ570–580 (bottom traces) at 10⁻⁴ M Ca²⁺ (left traces) and 10⁻³ M Ca²⁺ (right traces). In contrast to the wild type’s robust response to both applied pressure and Ca²⁺ at the tested conditions, the Δ570–580 mutant shows much reduced Ca²⁺-induced activities even at high [Ca²⁺] and its observed pressure-induced activities are also weakened. d Comparison of the P _o of the wild-type and the Δ570–580 mutant at 10⁻⁴ M Ca²⁺ (left) and 10⁻³ M Ca²⁺ (right) with or without 120 mmHg pressure. Mean ± SD (n = 10 for wild type and n = 5 for Δ570–580) (*P < 0.001)

Inserted Residues Result in Selective Attenuation of Ca²⁺ Activation

A sequence analysis of TRPY1 shows that there are two patches of high hydrophobicity (HP1 and HP2 in Fig. 4a) between the negative-charge cluster and the end of S6. HP1 and HP2 are connected through ⁵⁰⁹RILPPKRAKD⁵¹⁸. We added four residues, converting ⁵¹¹LPP⁵¹³ to LVPRGSP, to create a thrombin-cutting site. This construct, now called “+4a” (Fig. 4a), did not show the phenotype in the intended thrombin-cleavage experiment (data not shown) but produced an inadvertent but illuminating effect. Under patch clamp, the baseline activities (without pressure) of +4a are considerably lower than the wild type, with the difference being more evident at a higher [Ca²⁺] (10⁻³ M) (Fig. 4b, c, e, *P < 0.05). It should be noted that, compared with the negative-charge deletion Δ570–580 (Fig. 3c, d), the loss of Ca²⁺ activation in +4a is much less severe (Fig. 4b, e). Interestingly, the mechanosensitivity of +4a is nearly intact (Fig. 4b, e). It appears that the Ca²⁺ activation retained in +4a confers sufficient synergism to enable the observed force-induced activities to be nearly normal, resembling the activities of wild type at 10⁻⁵ M Ca²⁺ (Fig. 1a, second trace). To test whether the selective reduction of Ca²⁺ activation is due to the specific sequence of the thrombin-cutting site, we replaced its key charged arginine with the small and uncharged alanine, changing the sequence to LVPAGSP to create the “+4b” construct (Fig. 4a). Under patch clamp, +4b behaves similarly to +4a (Fig. 4b, f). Both the +4a and +4b mutants have evidently lessened Ca²⁺-induced activities, but their activations by pressure are nearly normal. Note that, to assess the P _o of +4a and +4b, we used 10⁻³ M Ca²⁺ together with 240 mmHg pressure to estimate the total number of conducting units. The channel activities in this condition may not be saturated, especially for +4b mutant (Fig. 4e, f), and therefore their normalized P _o values may be slightly underestimated. However, this underestimation does not affect our conclusion since it is very evident that at 10⁻³ M Ca²⁺ the basal activities are much lower than the response under 240 mmHg pressure in +4a and +4b (Fig. 4e, f), compared with the robust baseline activities of wild type in the same condition and its masking effect on further force responses (Fig. 1a, top trace, 4c). Finally, to test whether the length of the linker is a determinant, we generated a construct with the insertion of only one residue, “+1”, converting ⁵¹¹LPP⁵¹³ to LPRP (Fig. 4a). Interestingly, the +1 mutant shows clearly milder decrement of basal activities than the two +4 mutants and has little influence on mechanosensitivity (Fig. 4b, d), as is the case with the +4 mutants. This selective loss of Ca²⁺ activation in the insertion mutants is consistent with the idea that the point of insertion is in the transmission path of the Ca²⁺-gating force. Although the true mechanism is yet to be clarified, amino acid insertions may slacken the transmission path. Since mechanical activation is not greatly affected by the inserted residues, this transmission appears separable from that of the membrane stretch force that also reaches the gate, further supporting a parallel model.

Fig. 4

Inserted residues result in selective decrement of Ca²⁺ activation. a Schematic representation shows structural motifs between S6 and the negative charge cluster. Two hydrophobic patches (HP1 and HP2) are predicted by hydrophilicity plot. They are connected through residues 509–518. Three different insertion mutations were constructed at this site: +1 mutant adds R after P⁵¹², +4a mutant adds V after L⁵¹¹ as well as RGS after P⁵¹², +4b mutant adds V after L⁵¹¹ and AGS after P⁵¹². b Comparison of the P _o of the wild-type (WT) and the three insertion mutants at 10⁻⁴ M Ca²⁺ (left) and 10⁻³ M Ca²⁺ (right) with or without 240 mmHg pressure. The insertion mutants display selective reduction of Ca²⁺ activation and nearly normal pressure activation. Mean ± SD (WT, n = 10; +1, n = 5; +4a, n = 10; +4b, n = 4) (*P < 0.05) c–f Patch-clamp recordings of channel activities from wild type, +1, +4a and +4b at 10⁻⁴ M Ca²⁺ (left) and 10⁻³ M Ca²⁺ (right). Different pressures were exerted as indicated. Wild type exhibits pronounced baseline activities at 10⁻³ M Ca²⁺ without added pressure, and this robust activation masks most pressure-induced responses. Two +4 mutants significantly reduce the basal activity at 10⁻³ M Ca²⁺ and therefore unveil the observed mechanosensitivity. The +1 mutant displays a milder decrement of Ca²⁺ activation at 10⁻³ M. (The apparent difference between d and e is due to the numbers of channels in the patches)

Discussion

In this study, we investigated the relationship between stretch-force activation and Ca²⁺ activation in the polymodal ion channel TRPY1. Polymodal gating is a common feature in many ion channels. For instance, TRP channels are regulated by voltage, temperature, force, pH and chemicals (Pedersen et al. 2005; Ramsey et al. 2006; Venkatachalam and Montell 2007); tandem pore-domain potassium channels are modulated by force, polyunsaturated fatty acids and pH (Honore 2007); MscS is gated by force, voltage and protons (Hurst et al. 2008); large-conductance Ca²⁺-activated K⁺ (BK) channels are activated by voltage and Ca²⁺ (Cox et al. 1997; Horrigan and Aldrich 2002; Latorre and Brauchi 2006; Magleby 2003). A fundamental question in polymodal channels is how different sorts of physical and chemical stimuli act and interact. Extensive efforts, including mutations and chimeras, have been devoted to solve the relationships among the gating principles (Bandell et al. 2006; Grandl et al. 2008; Honore 2007; Latorre and Brauchi 2006; Magleby 2003; Voets et al. 2004).

Our results shed light on TRPY1’s Ca²⁺ sensitivity, which is different from that of known Ca²⁺-activated TRP channels. Many Ca²⁺-activated TRP channels are reported to require calmodulin as a regulatory subunit (Nilius et al. 2004, 2005; Saimi and Kung 2002). Thus far, only one TRP channel, TRPA1, has been definitively proposed to be directly activated by Ca²⁺, although the location of its Ca²⁺ sensor is still under debate (Doerner et al. 2007; Wang et al. 2008; Zurborg et al. 2007). By screening the yeast deletion collection and by examining the yeast calmodulin mutant (Fig. 2), we excluded common Ca²⁺-binding proteins being involved in regulating TRPY1’s Ca²⁺ sensitivity. Our work indicates that TRPY1 is probably directly activated by Ca²⁺. Intriguingly, there are no well-known Ca²⁺-binding motifs such as EF-hand, RCK domain or BK-like Ca²⁺ bowl in TRPY1. Thus, a novel Ca²⁺-binding structural element may be at work. We found a dense cluster of negative charges to be necessary for Ca²⁺ activation (Fig. 3). They likely contribute to a part of Ca²⁺ binding, although the nature of the binding site awaits structural information in the future.

With regard to the mechanosensitivity in TRP channels, the yeast TRPY1 provides a model to scrutinize the relationship between force and Ca²⁺ activation. In theory, these two gating principles can work in series or in parallel. In a serial model, force could be detected by steady-state force sensors and exerts its effect by modifying the Ca²⁺ sensor conformations and thereby increasing the Ca²⁺ binding affinity, or alternatively, Ca²⁺ binding could result in subsequent formation of transient force sensors that could further perceive the addition of membrane stretch. In such possibilities of the serial model, Ca²⁺ activation can be viewed as a prerequisite step before force activation. In contrast, a parallel model describes different stimuli being sensed and transmitted to the gate independently and different forms of energy can be summed as Boltzmann factors at the channel gate to produce a synergistic effect. Although these two types of models may oversimplify the true gating mechanism, discriminating between them can provide insights. In our study, we found evidence supporting that force and Ca²⁺ act in parallel and these two gating stimuli operate independently and synergistically on TRPY1. Ca²⁺ activates this channel without added stretch force. Strong stretch force can also induce channel activities in the virtual absence of cytoplasmic Ca²⁺. Note, unlike Ca²⁺, the application of stretch force is limited by the strength of patch seals. Within the range we can test, stretch-force responses in the presence of cytoplasmic Ca²⁺ fit well to an a priori Boltzmann equation by summing the two energy exponents (Fig. 1). It is of interest to evaluate the binding constant (K _m) and the number of bound Ca²⁺ (n) from existing data. However, because of membrane–glass adhesion as well as surface tension and other forces innate to the lipid bilayer, it is likely that there is mechanical energy impinging on the gate even when no added tension is applied to the patch. Because the inability to experimentally set the mechanical energy term to zero, the mechanical contribution to gating energy cannot be practically sorted out from that of Ca²⁺ binding and, therefore, the K _m and n of Ca²⁺ activation cannot be accurately extracted from existing data. The energetic analysis fits well the observation that the deletion disrupting a presumed Ca²⁺-binding motif not only diminishes Ca²⁺ activation but also weakens the observed mechanosensitivity, as expected from the dramatic decrement of the Ca²⁺ activation energy term in the sum (Fig. 3). The mutants with strategic amino acid insertions selectively attenuate the Ca²⁺ activation to a degree that enough synergism can be preserved to manifest the observed near normal mechanosensitivity (Fig. 4). Since we do not yet have the crystal structure of TRP channels, it seems premature to envision how such insertions attenuate Ca²⁺ activation. Experiments are needed to examine other insertions toward understanding how the gating force of Ca²⁺ binding may reach the gate. Strategic insertions reducing gating force are known in the BK linker study (Niu et al. 2004). Regardless of the mechanism, the key observation here is that such insertions further support the parallel model of Ca²⁺ activation and stretch-force activation. We think the energy of Ca²⁺ binding simply converges at the gate together with mechanical energy rather than sensitizing the perception and transmission of the mechanical force to the gate. In order to respond to mild environmental mechanical stimuli, yeast cells may have evolved to enhance the TRPY1 response by Ca²⁺ activation through CICR feedback.

Since TRPY1 can be activated instantly by membrane stretch force in excised patches, energy- or enzyme-dependent pathways are probably not required for its force activation. Furthermore, multiple lines of high-throughput screens, including yeast two-hybrid screens as well as yeast single-gene deletion collection screens, in search for possible interacting partner proteins of TRPY1, did not find any candidate (Loukin et al. 2008). Thus, it seems likely that TRPY1 itself directly senses the stretch force through the membrane, as is the case with MscL, the bacterial mechanosensitive channel (Kung 2005; Martinac et al. 2008). A recent cryo-electron microscope structural study of TRPV1 has shed light on the general organization of TRP channels (Moiseenkova-Bell et al. 2008). In this structure, the transmembrane domains appear to be buried in the lipid bilayer like a balloon and the cytoplasmic domains form a gondola underneath in the cytoplasm. By analogy to the crystal structure of MthK, the balloon and the gondola are tethered through C-terminal peptide linkers (Jiang et al. 2002). Taking together our analysis of gating energetics, our mutational analyses as well as the cryo-electron microscopic image of TRP channels (Moiseenkova-Bell et al. 2008), we propose a mechanistic model for TRPY1 in which the mechanical force acts through the membrane-associated domains (the balloon) and Ca²⁺ binds to the negative charge residues in the cytoplasmic domains (the gondola), both generating energy that reaches the gate in parallel (Fig. 5).

Fig. 5

A mechanistic model of TRPY1 gating by force and by Ca²⁺. This channel senses membrane stretch force through membrane-buried domains and senses Ca²⁺ via the cytoplasmic gondola. The two gating principles act on the gate in parallel

TRPY1 is being developed as a model to understand TRP channels in general. Sorting out the interaction of various gating forces in these polymodal channels is a challenge. Of particular interest is mechanosensitivity, given its importance and the current state of molecular understanding. Clarifying how Ca²⁺ and stretch force gate TRPY1 should facilitate the molecular dissection of this molecule through forward and reverse genetics (Su et al. 2007; Zhou et al. 2007), for which the yeast system is especially suitable.