Functional and structural insights into activation of TRPV2 by weak acids

The transient receptor potential vanilloid (TRPV) subfamily belongs to the large superfamily of transient receptor potential channels. TRPV1-TRPV6 are polymodal cation channels that are activated by numerous endogenous and exogenous ligands as well as by physical stimuli, including heat, cold, and pressure (Rosenbaum et al, 2022). While TRPV2 has long been the least well understood member of the TRPV subfamily, there is growing evidence for an eminent relevance of TRPV2 in several physiological processes. TRPV2 has been implicated in neuronal outgrowth (Cohen et al, 2015; Shibasaki et al, 2010; Sugio et al, 2017), mechano-sensitivity of sensory neurons (Katanosaka et al, 2018), myocardial structure and function (Entin-Meer and Keren, 2020; Katanosaka et al, 2014), innate immunity (Link et al, 2010), brown fat thermogenesis (Sun et al, 2016), endocrine secretion (Hisanaga et al, 2009) and endometrial development (De Clercq et al, 2017). From a pathophysiological perspective, TRPV2 not only seems to be highly relevant for growth and invasiveness of different tumors (Conde et al, 2021; Huang et al, 2022; Shoji et al, 2023; Siveen et al, 2020), but also for the permeation of several virus subtypes through the cell membrane of myeloid cells (Guo et al, 2022). A significant challenge in studying TRPV2 is the lack of potent and specific agonists and antagonists. While 2-aminoethoxydiphenyl borate (2-APB), cannabidiol (CBD) and probenecid are commonly used TRPV2 activators, they have many off target effects, including modulation of other TRPV channels (Chung et al, 2004; Etemad et al, 2022; Zhang et al, 2022). High heat exceeding >50 °C activates rodent TRPV2 (Caterina et al, 1999), but human TRPV2 is insensitive to heat (Neeper et al, 2007). This species difference raises the question of whether heat-induced activation of TRPV2 is endogenously relevant, and mice lacking TRPV2 do not have a phenotype in regard to heat-induced nociceptive behavior (Park et al, 2011). TRPV2 was also reported to be gated by mechanical or osmotic stress (Mihara et al, 2010; Muraki et al, 2003). More recent studies have identified putative endogenous modulators of TRPV2, including reactive oxygen species (ROS) (Fricke et al, 2019), a tyrosine-dependent phosphorylation (Mo et al, 2022) and cholesterol (Su et al, 2023).

Acidosis is a common endogenous modulator of several TRP channels, i.e., protons act as agonists or inhibitors on different TRP-channel subtypes. TRPV1 is directly activated and sensitized by extracellular protons (Caterina et al, 1997; Tominaga et al, 1998). The extracellular residues Glu600 and Glu648 have been validated as critical extracellular proton modulation sites of TRPV1 (Jordt et al, 2000; Zhang et al, 2021), but further residues were suggested to regulate activation of TRPV1 by protons as well (Aneiros et al, 2011; Ryu et al, 2007; Wang et al, 2010). Less attention has been drawn to the reports demonstrating that TRPV1 is inhibited by weak acids, and that protons induce a direct channel inhibition when applied to the cytosolic side (Chung et al, 2011; de la Roche et al, 2016; Wang et al, 2011). In contrast, TRPV3, as well as the irritant receptor TRPA1, are activated by weak acids (Cao et al, 2012; Gao et al, 2016; Wang et al, 2010, 2011). While both ion channels seem to be directly activated by intracellular protons, no distinct proton modulation sites have yet been identified for TRPV3 or TRPA1.

Given that TRPV2 displays a high sequence homology as well as overlapping functional properties with TRPV1 and TRPV3, it is notable that previous studies have shown that both extracellular and intracellular protons fail to activate TRPV2 (Cao et al, 2012; Caterina et al, 1999; Gao et al, 2016). In this study, we report on an as yet unappreciated property of TRPV2 allowing weak acids to activate and sensitize the channel. By combining patch clamp electrophysiology, cryo-EM in lipid nanodiscs, and molecular dynamics (MD) simulations, we demonstrate that previously identified intracellular binding sites for 2-APB are also critical for weak acid-sensitivity of rat TRPV2 (Pumroy et al, 2022). In brief, our report describes a detailed characterization of a novel and likely physiologically relevant acid-sensitivity of TRPV2.

We first challenged HEK293T cells expressing rat TRPV2 (rTRPV2) with pH 5.0 titrated either with the strong hydrochloric acid (HCl), or the weak acetic acid (HOAc). While HCl fully dissociates and is therefore membrane-impermeable, the undissociated form of HOAc can permeate through the membrane and induce intracellular acidosis (Fig. EV1A). Solution titrated to pH 5.0 with HCl did not induce any inward currents except for the rapidly activating and inactivating currents known to be mediated by endogenously expressed ASIC1a channels in some cells (Figs. 1A and  EV1B) (Gunthorpe et al, 2001). In contrast, application of 30 mM HOAc titrated to pH 5.0 evoked slowly activating inward currents that were reversible upon washout (Fig. 1B). These currents were not observed in non-transfected cells (Fig. EV1B). In voltage ramps ranging from −100 to +100 mV, 30 mM HOAc at pH 5.0 induced large membrane currents with a prominent inward and outward rectification (Fig. 1C, n = 5). This strong outward rectification was also observed on smaller membrane currents evoked by 10 mM HOAc at pH 5.0 examined by means of 200 ms long voltage steps applied from −60 to +120 mV in steps of 20 mV (Fig. 1D,E, n = 6). HEK293 cells were recently described to endogenously express proton-activated chloride channels (PAC or TMEM206) which produce large proton-activated outward chloride currents at positive membrane potentials (Ullrich et al, 2019; Yang et al, 2019). In agreement with this, pH 5.0 titrated with HCl evoked large PAC-like outward currents that were blocked by the PAC-inhibitor pregnenolone sulfate in non-transfected HEK293T cells (Fig. EV1C,D). Surprisingly, these PAC-like currents were almost completely inhibited by 10 or 30 mM HOAc at pH 5.0 (Fig. EV1E,F), and no outwardly rectifying currents were evoked by 10 mM HOAc at pH 5.0 when non-transfected cells were examined with voltage steps (Fig. EV1G,H). Furthermore, the TRP-channel inhibitor ruthenium red (10 µM) strongly inhibited the HOAc-evoked inward currents in cells expressing rTRPV2 (Fig. 1F,G, n = 9, 98 ± 1% inhibition). These data allow us to conclude that only TRPV2 accounts for the observed HOAc-induced inward currents. This notion also seems to apply for HOAc-induced outward currents recorded at positive potentials, but it is possible that endogenous PAC channels to a limited degree contribute to these currents.

We next asked if HOAc also induces potentiation of TRPV2-mediated currents induced by other stimuli. The most potent TRPV2-agonist available to date is 2-APB. However, the potency of 2-APB on TRPV1, TRPV2, and TRPV3 was demonstrated to be increased by protons due to a modification of its chemical structure (Gao et al, 2016). Accordingly, we observed that pH 5.0 titrated with either HCl or 10 mM HOAc induced a strong potentiation of 2-APB-induced currents (Fig. EV1I–K). These data either implicate that 2-APB may not be suitable to study effects induced by weak acids, or that extracellular protons sensitize TRPV2. Therefore, we explored the effects of pH 5.0 titrated with HCl as well as 10 mM HOAc at pH 5.0 when co-applied with heat or CBD. Because rTRPV2 exhibits a strong auto-sensitization when repeatedly activated by heat (Caterina et al, 1999; Liu and Qin, 2016), we applied subthreshold heat ramps ( ~ 45 °C) that do not induce activation of rTRPV2 in control solution (Fig. 1H) (Fricke et al, 2019; Mo et al, 2022; Zhang et al, 2022). Application of 10 mM HOAc at pH 5.0 induced small inward currents that exhibited a strong potentiation upon application of heat ramps (Fig. 1H, n = 6). In contrast, pH 5.0 titrated with HCl did not provoke heat-induced currents (Fig. 1I, n = 5). HOAc also induced a strong potentiation of inward currents induced by 20 µM CBD in rTRPV2-expressing cells (Fig. 1J,L, n = 6, one-way ANOVA with Bonferroni correction, DF = 2, F value: 9.86266). In contrast, pH 5.0 titrated with HCl did not potentiate CBD-induced currents (Fig. 1K,L, n = 9, P = 1). Together, these data demonstrate that HOAc can both activate and sensitize TRPV2.

In order to determine if gating of rTRPV2 by HOAc is an effect limited to HEK293T cells, we examined neuroblastoma ND7/23 cells transfected with rTRPV2. HOAc induced activation of inward currents as well as sensitization of heat-evoked currents on rTRPV2 expressed in ND7/23 cells (Appendix Fig. 1A,B). As TRPV2 displays considerable species-specificities in regard to activation by heat and 2-APB (Neeper et al, 2007), we also examined the mouse (m) and human (h) orthologues of TRPV2. Similar to rTRPV2, cells expressing mTRPV2 produced prominent inward as well as outward currents induced by 30 mM HOAc at pH 5.0 (Appendix Fig. 1C,D). They also exhibited large heat-evoked currents when challenged with 10 mM HOAc at pH 5.0 (Appendix Fig. 1E). In contrast, cells expressing hTRPV2 failed to produce inward currents when treated with 30 mM HOAc at pH 5.0 (Appendix Fig. 1F, G). These cells produced small outward currents at positive membrane potentials, but these currents may at least in part be generated by PAC channels. Furthermore, even the application of 30 mM HOAc at pH 5.0 only induced small heat-evoked currents on hTRPV2-expressing cells (Appendix Fig. 1H). CBD is one of few TRPV2-agonists that robustly activates hTRPV2 (Qin et al, 2008). When we examined CBD-induced currents of hTRPV2, an HOAc-induced potentiation was only observed following washout of HOAc (Appendix Fig. 1I,J). These data show that HOAc can activate and sensitize rodent TRPV2 orthologues, but also that hTRPV2 seems to be both inhibited and activated by at least HOAc. We also examined whether activation of TRPV2 is an effect specific for HOAc, or if it applies to weak acids in general. Indeed, the weak acids carbon dioxide, lactic acid, propionic acid and formic acid activated and sensitized rTRPV2 (Fig. EV2A–O). In agreement with a recent study on TRPV2 (Catalina-Hernandez et al, 2024), we also observed an activation and sensitization of rTRPV2 by benzoic acid (Fig EV2P–S). However, the non-deprotonatable β-hydroxybutyric acid failed to activate rTRPV2 (Appendix Fig. 1K,L).

Having established that weak acids activate and sensitize rTRPV2, we next aimed to study the activation mechanism in more detail. Activation of TRPV3 and TRPA1 by weak acids seems to be mediated by an intracellular accumulation of protons driven by permeation of the undissociated form of the acid through the cell membrane (Cao et al, 2012; Wang et al, 2011). In order to examine if this property also applies to rTRPV2, we analyzed the pH-dependency (pH 7.4, 6.0, or 5.0) for activation of rTRPV2 by 10, 20, and 30 mM HOAc. Given that HOAc has a pK_a-value of 4.76, this experimental approach results in an increasing amount of intracellular entry of HOAc at lower pH values and high concentrations. To standardize experimental conditions between each experimental group, peak current amplitudes were determined after 3 min application of the test solutions (Fig. 2A). Indeed, amplitudes of HOAc-induced inward currents increased with higher concentrations of HOAc and with lower pH values (Fig. 2B, n = 5–11 for each condition). Using the pH-indicator BCECF, we found that application of 10, 20, or 30 mM HOAc at pH 7.4, 6.0, or 5.0 on HEK293T cells resulted in a reduction of the intracellular pH value that depended on both concentration of HOAc and the pH value (Fig. 2C). As HOAc at pH 7.4 both failed to activate TRPV2 and to induce a substantial intracellular acidosis, it seems likely that membrane permeability of HOAc and the resulting intracellular acidosis are required for activation of TRPV2. Accordingly, channel activation was evoked by application of 30 mM HOAc at pH 5.0 in the cell-attached configuration (Fig. 2D, n = 5). In all, 10 mM HOAc at pH 5.0 also induced a strong potentiation of CBD-induced membrane currents in this configuration (Fig. 2E,F, n = 6). Overall, 30 mM HOAc at pH 5.0 also induced channel activation in cell-free inside-out multi-channel recordings at −60 mV (Fig. 2G, n = 5). When single-channel recordings were performed on inside-out patches at +60 mV, the application of 30 mM HOAc at pH 5.0 induced robust channel openings with an amplitude of 4.3 ± 0.1 pA and a single-channel conductance of 71.6 ± 5.2 pS (Fig. 2H–K, n = 6). All patches displaying sensitivity to HOAc also generated single-channel openings following the application 500 µM 2-APB, confirming expression of TRPV2 (Fig. 2H; Appendix Fig. 2A,B). Of note, the single-channel amplitudes of currents induced by 500 µM 2-APB were larger than those induced by HOAc (Appendix Fig. 2B, 8.1 ± 0.1 pA). Moreover, experiments with 100 µM 2-APB or 10 µM CBD resulted in single-channel currents with amplitudes around 5–6 pA (Appendix Fig. 2C–F). These data suggest that gating properties of rTRPV2 vary between different agonists, and that they are concentration-dependent. In agreement with previous reports that cytosolic protons do not activate TRPV2 (Cao et al, 2012; Gao et al, 2016), application of pH 5.0 titrated with HCl to the cytosolic side did not induce membrane currents in multi-channel inside-out recordings (Appendix Fig. 2G, n = 5). Application of pH 5.0 on both sides of the membrane as well as 30 mM potassium acetate at pH 7.4 also failed to activate currents in TRPV2-expressing inside-out patches (Appendix Fig. 2H,I, n = 5 and 4). The potentiation of CBD-induced currents by HOAc observed in whole cell as well as cell-attached experiments was very strong, probably similar to the degree of the supra-additive potentiation we and other groups recently described for the combination of CBD and 2-APB on TRPV2 (Gochman et al, 2023; Pumroy et al, 2022). This HOAc-induced potentiation of CBD-induced currents was not observed at pH 7.4 in the whole-cell recordings (Appendix Fig. 2J,K, n = 6), again indicating that intracellular mechanisms are involved in HOAc-induced gating of TRPV2. Considering that the combination of CBD and HOAc gives rise to a much stronger channel activation as compared to HOAc alone, we examined the effects of HOAc or protons on CBD-induced membrane currents in cell-free inside-out patches. As expected, 10 mM HOAc at pH 5.0 induced a strong potentiation of CBD-induced currents in inside-out recordings (Fig. 2L). In contrast, an inhibition rather than a potentiation of CBD-induced currents was observed upon application of pH 5.0 titrated with HCl on the cytosolic side or with the combined acidification of the cytosolic and extracellular solutions (Fig. 2M,N, n = 5 each). Taken together, these data indicate that the effect of HOAc on TRPV2 cannot be reproduced by application of protons or acetate. Thus, the mechanism for weak acid-induced activation of TRPV2 seems to be more complex than a sole binding of protons on specific intracellular or extracellular binding sites (Fig. 2O).

TRPV2 was reported to interact with the actin cytoskeleton, and that this interaction renders TRPV2 mechano-sensitive (Sugio et al, 2017). Even though HOAc was found to gate TRPV2 in cell-free inside-out patches, we examined if weak acid-sensitivity is modulation following disruption of the cytoskeleton. Application of cytochalasin-D and latranculin A (inhibitors of actin polymerization) or nocodazole (a microtubule inhibitor) did not reduce HOAc-induced inward currents in TRPV2-expressing cells (Appendix Fig. 2L–N). These data suggest that weak acid-sensitivity of TRPV2 does not depend on the cytoskeleton.

Even though our functional data did not define a clear role of extracellular or intracellular protons as important mediators of weak acid-sensitivity of TRPV2, we performed two independent approaches with the aim to identify proton modulation sites of TRPV2 being required for HOAc-induced activation and sensitization of rTRPV2. Similar to the approach that allowed for the identification of extracellular proton modulation sites of TRPV1 (Jordt et al, 2000), we first probed the outer pore domain of rTRPV2 by performing an unbiased mutational screen. All targeted residues were conservatively replaced by similar amino acids lacking a titratable moiety. Some residues (Lys566, Glu569, Asp570, Glu577, and Glu586) were also more severely mutated to alanine. Among the thirteen conservative mutants, only two (rTRPV2-E561Q and -K602Q) displayed obvious deficits for both HOAc-induced inward currents and HOAc-induced sensitization of heat-evoked currents (Fig. EV3A,B). The general functionality of all mutants was examined by testing their responses to a high concentration (1 mM) of 2-APB. Only rTRPV2-K602Q displayed markedly reduced 2-APB-induced responses (Fig. EV3C–E). Therefore, the almost complete loss of HOAc-sensitivity observed for rTRPV2-K602Q may be due to a general loss of function. Surprisingly, the alanine mutants displayed a clear gain-of-function property for activation by HOAc (Fig. EV3F–H). However, this property was not observed for potentiation of heat-induced currents or for activation by 2-APB (Fig. EV3I,J). Of note, similar effects were previously observed on pore-mutants in TRPV1 (Jordt et al, 2000). Together, this unbiased mutational screen did not deliver conclusive evidence for or against an important role of proton modulation sites in outer pore domain for weak acid-sensitivity of rTRPV2.

We next employed cryo-EM to examine the structural effects of adding 30 mM HOAc at pH 5.0 to rTRPV2. We first obtained a higher resolution apo rTRPV2 in nanodisc dataset at pH 8, yielding the same classes observed previously: TRPV2_Apo1 at 2.85 Å and TRPV2_Apo2 at 2.68 Å, as well as an additional state, TRPV2_Apo3 at 3.0 Å, similar to the CBD-bound rTRPV2_CBD2 (PDB 6U88) or the apo rabbit TRPV2 (PDB 5AN8) states (Appendix Table 1; Appendix Figs. 3–5) (Huynh et al, 2016; Pumroy et al, 2019, 2022; Zubcevic et al, 2019). The overall conformations of TRPV2_Apo1 and TRPV2_Apo2 are quite similar—closed at the lower gate and gated by Met645, N-terminus in a helical conformation, His651 constricting the bottom of the pore (Figs. 3A,B and EV4A,B,D). The major difference between these states is the status of the selectivity filter, which is tightly closed in TRPV2_Apo1 and opened wide in TRPV2_Apo2 (Fig. EV4). While the pore of TRPV2_Apo3 is otherwise very similar to that of TRPV2_Apo1, His651 is rotated out of the ion permeation path (Fig. EV4E,F). This state also features a downward rotation of the ankyrin repeat domains (ARD) and the conformational change of the C-terminus from a helix inside the ARD skirt to a loop which wraps around to interact with the N-terminal end of the ARDs (Fig. EV4G–I). Work on TRPV3 suggests that the loop conformation of the C-terminus indicates an inactivated or desensitized state, while the helix conformation is required for an activated or sensitized state (Zubcevic et al, 2018). Therefore, we propose that TRPV2_Apo1 and TRPV2_Apo2 are in pre-open states and TRPV2_Apo3 is a desensitized channel.

We then prepared grids of rTRPV2 briefly exposed (<1 min) to 30 mM sodium acetate with a final pH of 5.0. This dataset yielded a single state, TRPV2_HOAc at 3.14 Å, which featured an open selectivity filter and a helical C-terminus (Appendix Table 1; Fig. EV4C,F,J; Appendix Figs. 5 and 6), resembling TRPV2_Apo2. Beyond the shift in equilibrium to favor the open selectivity filter, there are other notable changes in TRPV2_HOAc. First, we examined the extracellular surface of the channel. Our initial mutagenesis data suggested two residues, Glu561 and Lys602 as possible proton modulation sites. We noticed that the voltage sensing-like domain (VSLD) of TRPV2_HOAc shifts compared to TRPV2_Apo1 and TRPV2_Apo2 (Fig. EV5A,B). In addition, we see an inward movement of S5 (Fig. EV5A,B) and a corresponding change at the bottom of S6 compared to TRPV2_Apo2 (Fig. EV4A,B). While these changes did not lead to the opening of the lower gate at Met645, these movements coincide with a 180° rotation of Glu495 at the top of S4, away from the membrane interface towards S5 and swapping positions with Arg617 (Fig. 3A–C). We used PROPKA3 to determine that Glu495 and Glu561 are likely to be protonated at pH 5.0 in TRPV2_HOAc (Appendix Table 2), which could allow for the formation of a carboxyl-carboxylate interaction between the two glutamic acids when in close proximity. In order to determine if Glu495 and Glu561 are relevant for activation by weak acids, we examined the mutants rTRPV2-E495Q (Fig. 3D, n = 9) and rTRPV2-E561Q (Fig. 3En = 8). Indeed, both mutants displayed significantly reduced HOAc-induced inward currents (Fig. 3F, n = 9, one-way ANOVA, Bonferroni correction, DF = 2, F value: 8.30592). HOAc-induced potentiation of heat-evoked currents was also reduced for both mutants (Fig. 3G–I, n = 5 for both mutants, one-way ANOVA, DF = 2, F value: 36.28001). Of note, these residues are highly conserved in mammalian TRPV2, but not across the TRPV family (Appendix Fig. 3). We then ran molecular dynamic (MD) simulations starting from TRPV2_Apo1 both with and without protonation of key residues, including Glu495 and Glu561. Consistent with the cryo-EM predictions, during MD simulations with the protonated residues (simulating pH 5) the Glu561–Arg617 contact is present in 67.15% of frames and the Glu561–Glu495 contact is present in 16.57% of frames compared to 87.07% for Glu561–Arg617 and 0.003% for Glu561–Glu495 in the MD simulation with the unprotonated residues (simulating pH 7) (Fig. 3J–L). These data suggest that Glu495 and Glu561 may be relevant, but not required for activation of TRPV2 by weak acids. The structures also suggest a reason for the loss of function of the K602Q mutant: in the open selectivity filter conformations, TRPV2_Apo2 and TRPV2_HOAc, Lys602 is engaged in a cation-π interaction with Phe612, which is likely essential for stabilizing the open conformation of the selectivity filter (Fig. EV5C–E).

We next examined the intracellular surface of the channel and observed that His651 is rotated out of the path of the central channel along with the shifts in S5 and S6 (Fig. EV4C). We also observed a major disruption at the S4–S5 linker, where His521 rotates away from S5 (Fig. 4A–C). His651 is unlikely to be the primary source of weak acid-sensitivity as this residue is only found in rat TRPV2 (Appendix Fig. 3). The MD simulations validate this notion as the protonation of His651 does not significantly alter its motility and orientation relative to the central channel pore (Appendix Fig. 7). His521 is highly conserved in most TRPV2 orthologues, but it is not observed in this position in other TRPV channels (Appendix Fig. 3). His521 is found at the N-terminal end of the S4–S5 linker, exposed to the lipid interface. In all three apo TRPV2 structures, His521 forms a cation-π interaction with Arg539 from the adjacent TRPV2 monomer (Fig. 4A,B). In TRPV2_HOAc, His521 is rotated away from Arg539, possibly due to protonation (Fig. 4C). The combined effect of protonation and rotation exposes a positively charged pocket comprised of His521, Arg539, and Arg535. In an intriguing parallel, we recently observed that the activator 2-APB bound in the same pocket (Pumroy et al, 2022), inducing an outward rotation of His521 similar to what we observe in TRPV2_HOAc (Fig. 4D). PROPKA3 predicted that His521 would be protonated in the TRPV2_HOAc structure (Appendix Table 2). We also observed that protonation of His521 caused a splaying of the S4–S5 linker and S5 helices in a subset of subunit interfaces (Fig. 4E,F). This splaying was similar in extent to that observed in the rTRPV2 inactivated 2-APB-bound structure (PDB ID: 7N0M, Fig. 4E,F). In contrast, the helices remain closer together at a distance observed in TRPV2_Apo1 with a neutral His521 (Fig. 4F). We next performed patch clamp experiments on the mutant rTRPV2-H521A as the putative proton modulation site, but also rTRPV2-R535A and rTRPV2-R539A as critical residues of this positively charged pocket. When challenged with 30 mM HOAc at pH 5.0, rTRPV2-H521A failed to generate inward currents (Fig. 4G, n = 6). Only minimal HOAc-induced currents were observed in cells expressing rTRPV2-R535A (Fig. 4H, n = 6) and rTRPV2-R539A (Fig. 4I, n = 8). All three mutants are functional as they generated large inward currents following application of 1 mM 2-APB (Appendix Fig. 8A), and we previously demonstrated that they are activated by heat (Pumroy et al, 2022). When normalizing the amplitudes of HOAc-induced currents with currents induced by 1 mM 2-APB, all three mutants displayed an almost complete loss in HOAc-sensitivity (Fig. 4J, one-way ANOVA, Bonferroni correction, DF = 4, F value: 15,57216). A replacement of His521 with an asparagine instead of an alanine also showed minimal HOAc-induced currents (rTRPV2-H521N, Fig. 4J, n = 8). TRPV2-H521A also completely failed to generate membrane currents when exposed to 30 mM or even 100 mM HOAc at pH 5.0 (Fig. 4K,L, n = 5 for each experiment). Furthermore, 10 mM HOAc at pH 5.0 induced negligible heat-evoked currents on cells expressing rTRPV2-H521A, rTRPV2-R535A or rTRPV2-R539A (Fig. 4M,N, n = 6–8, one-way ANOVA, Bonferroni post hoc test). Finally, rTRPV2-H521A displayed a reduced potentiation of CBD-induced currents by HOAc (Appendix Fig. 8B,C). Together, these data clearly indicate that the positively charged pocket comprised of His521, Arg539 and Arg535 is critical for weak acid activation and sensitization of TRPV2. As the exchange of H521 resulted in an almost complete loss of weak acid-sensitivity, it seems justified to argue that this residue is a critical proton modulation site.

What is the physiological relevance of weak acid-sensitivity of TRPV2? This study does not include more intact in vivo or ex vivo experimental preparations required to address this question, but we examined the effects of weak acids on the properties of TRPV2 that seem to be of physiological relevance. TRPV2 displays a high permeability for extracellular Ca²⁺ (Caterina et al, 1999), and an increase of cytosolic Ca²⁺ is decisive for several processes depending on TRPV2 (Conde et al, 2021; Guo et al, 2022; Maksoud et al, 2019; Shoji et al, 2023). We examined if activation of TRPV2 by weak acids is associated with a Ca²⁺-influx. To our surprise, HEK293T cells expressing rTRPV2 did not exhibit a robust elevation of intracellular Ca²⁺ following the application of weak acids. HOAc-induced effects in rTRPV2-expressing cells (n = 237) were only marginally larger than the effects observed in non-transfected cells (n = 422) (Fig. 5A). A similar effect was obtained with 100 mM propionic acid at pH 6.0 (Fig. 5B, n = 153 transfected and 351 non-transfected). The patch clamp data obtained on TRPV2 in this study include experiments performed with nominal Ca²⁺-free solutions in order to avoid a Ca²⁺-dependent desensitization (Mercado et al, 2010). Therefore, these Ca²⁺ imaging data indicate that extracellular calcium inhibits weak acid-induced activation of TRPV2. Indeed, patch clamp experiments with 2 or 10 mM Ca²⁺ in the extracellular solution revealed that HOAc-induced inward currents through rTRPV2 were significantly reduced (Fig. 5C–E, n = 6 and 8, one-way ANOVA with Bonferroni correction, DF = 2, F value: 15.88024). Furthermore, inward currents induced by HOAc in Ca²⁺-free extracellular solution were strongly inhibited when 2 mM Ca²⁺ was co-applied (Fig. 5E, n = 7). A similar Ca²⁺-induced inhibition was previously described for TRPV3 (Xiao et al, 2008). Therefore, we also explored if similar effects of Ca²⁺ are observed on inward currents induced by CBD and 2-APB, which both evoke a strong TRPV2-mediated Ca²⁺-influx when applied alone (Neeper et al, 2007; Qin et al, 2008). Different from HOAc-induced currents, inward currents induced by both 20 µM CBD or 500 µM 2-APB in Ca²⁺-free solution exhibited a transient rapid current increase upon application of 2 mM Ca²⁺ (Appendix Fig. 9A,B). These data indeed indicate that a strong inhibition of TRPV2 by extracellular Ca²⁺ only applies for weak acid-induced activation. In TRPV3, both calcium- and proton-induced inhibition are dictated by the pore residue Asp641 (Wang et al, 2021; Xiao et al, 2008). The equivalent residue in TRPV2 is Glu609, i.e., one of the pore residues that we already examined as a possible proton modulation site (Fig. EV4). We found that Ca²⁺-induced inhibition of HOAc-induced currents was reduced for rTRPV2-E609Q (Fig. 5F,G, n = 6, P < 0.001 unpaired t test). However, the biphasic effect of Ca²⁺ on 2-APB-induced currents was maintained on rTRPV2-E609Q (Appendix Fig. 9C). We also observed that application of pH 5.0 titrated with HCl induced a partial inhibition of CBD-induced currents generated by wild-type rTRPV2 (Appendix Fig. 9D). This proton-induced inhibition was significantly increased on rTRPV2-E609Q (Appendix Fig. 9D–F, P < 0.01, unpaired t test). These data show that the pore residue Glu609 in rTRPV2 is an important determinant for the inhibition of weak acid-induced currents by extracellular Ca²⁺. In contrast to Asp641 in TRPV3 however, Glu609 does not seem to dictate inhibition of TRPV2 by Ca²⁺ or protons in general.

The failure of weak acids to induce a robust TRPV2-mediated Ca²⁺-influx raised the critical question as to whether weak acids are likely to be relevant endogenous modulators of TRPV2 expressed in the cell membrane or not. We examined if an HOAc-induced potentiation of TRPV2 activation induced by other agonists is observed in the presence of extracellular Ca²⁺. 10 mM HOAc at pH 5.0 induced a strong potentiation of both heat-induced currents (Fig. 5H, n = 6) and CBD-induced currents (Fig. 5I,J, n = 8) in the presence of 2 mM Ca²⁺. Furthermore, Ca²⁺-imaging experiments demonstrated that 10 mM HOAc at pH 5.0 strongly potentiates CBD-induced Ca²⁺-influx in rTRPV2-expressing cells, but not in non-transfected HEK293 cells (Fig. 5K, n = 597 and 388, respectively). These data demonstrate that weak acids can sensitize TRPV2 at physiological concentrations of extracellular calcium.

We also asked if the inhibition by extracellular Ca²⁺ is associated with a lower permeability of rTRPV2 to Ca²⁺ when it is activated by HOAc as compared to 2-APB or CBD. Of note, the frequently cited high Ca²⁺-permeability of rTRPV2 was only determined following activation by heat exceeding 50 °C (Caterina et al, 1999). As is demonstrated in Fig. 5L, we determined a high permeability for both Ca²⁺ and Mg²⁺ when rTRPV2 was activated by 2-APB and CBD. As postulated, activation by HOAc was associated with lower permeabilities for both Ca²⁺ and Mg²⁺. Thus, both divalent ion permeabilities and single-channel conductances of TRPV2 seem to vary between agonists.

Both ASIC1a and PAC channels, but also TRPV3 were demonstrated to mediate cytotoxicity induced by acids (Cao et al, 2012; Sherwood et al, 2011; Ullrich et al, 2019; Yang et al, 2019). Therefore, we finally asked if the expression of rTRPV2 in HEK293 cells reduces cell vitality following incubation with HOAc. While incubation in pH 5.0 titrated with HCl for 2 h did not result in a substantial cytotoxicity (Fig. 6E), treatment with 10 mM and 30 mM HOAc at pH 5.0 for 2 h resulted in an increased staining with propidium iodide in both non-transfected HEK293 cells and in cells expressing rTRPV2 (Fig. 6A–E). However, an increased cell death in rTRPV2-expressing cells was only observed at 30 mM HOAc (Kruskal–Wallis with Dunn’s correction, n = 4 repeats of 5 analyzed high power fields each).

There is growing evidence for a critical involvement of TRPV2 in several physiological and pathophysiological processes. This emphasizes the need for a more detailed understanding of the functional properties of this yet poorly understood ion channel, including the identification of endogenous modulators. Recent studies have made some progress on this point by describing a modulation of TRPV2 by oxidants, cholesterol and a tyrosine-dependent phosphorylation (Fricke et al, 2019; Mo et al, 2022; Su et al, 2023). In this study we describe weak acids as possible endogenous modulators of TRPV2, and our comprehensive functional and structural data give detailed insights into the mechanism accounting for weak acid-sensitivity of TRPV2.

We and other laboratories have recently identified distinct intracellular binding sites for the non-selective TRPV2-agonists 2-APB, cannabidiol and the cannabinoid C16 (Pumroy et al, 2019, 2022; Zhang et al, 2022). Although TRPV1, TRPV2 and TRPV3 have homologous structures and share ~40% sequence identity, it has become clear that 2-APB interacts with distinct binding sites to activate the three ion channels (Boukalova et al, 2010; Hu et al, 2009; Pumroy et al, 2019; Singh et al, 2019). 2-APB induces a concentration-dependent intracellular acidosis when applied on HEK293 cells (Chokshi et al, 2012), possibly suggesting that 2-APB and weak acids target common binding sites to modulate TRP channels. Indeed, His426 seems to be a common intracellular activation site for both 2-APB and intracellular protons in TRPV3 (Cao et al, 2012; Hu et al, 2009). We previously determined that His521 is important for activation by 2-APB (Pumroy et al, 2022), and in this work the structure of TRPV2_HOAc guided us to His521 as a candidate intracellular proton modulation site. Both MD simulations and the almost complete loss of weak acid-sensitivity of the rTRPV2-H521A and rTRPV2-H521N mutants perfectly confirmed this interpretation for His521. The structure of TRPV2_HOAc also identified Glu495 and Glu561 as possible extracellular proton modulation sites, and an exchange of Glu495 and Glu561 indeed resulted in reduced sensitivities to HOAc. However, an altered HOAc-sensitivity was also determined for rTRPV2-E609Q and the alanine mutants at the outer pore of rTRPV2. Thus, the exchange of any residue in the outer pore region may result in altered functional properties, making it challenging to define the role of distinct residues for weak acid-sensitivity. Our functional data more or less discard a simple activation mechanism only requiring binding of extracellular and/or intracellular protons to distinct residues of rTRPV2. In fact, both intracellular and extracellular application of pH 5.0 titrated with HCl even inhibited CBD-induced membrane currents. Thus, in contrast to the weak acid-sensitive channels TRPV3 and TRPA1 which are directly activated by intracellular protons (Cao et al, 2012; Gao et al, 2016; Wang et al, 2011), the mechanism accounting for weak acid-sensitivity of TRPV2 seems to be more complex.

Although HOAc-sensitivity was almost abolished in the rTRPV1-H521A mutant, further binding sites may be involved. To this end, a recent study suggested that activation of TRPV2 by probenecid and benzoic acid derivatives may be due to an interaction with the proposed binding sites of the TRPV2-inhibitor piperlongumine, including the residue Arg539 (Catalina-Hernandez et al, 2024; Conde et al, 2021). While 2-APB-sensitivity of rTRPV2 only partially depends on His521, a stronger reduction of the 2-APB-sensitivity was observed in the double mutant rTRPV2-H521A/R539K (Pumroy et al, 2022). We suggested that 2-APB is likely to interact with a positively charged pocket formed by His521, Arg535, and Arg539 (Pumroy et al, 2022), and our data may indicate that the total charge of this pocket is decisive for weak acid-sensitivity as well. Whether or not this also applies for probenecid needs to be analyzed in more detail. Although His521, Arg535 and Arg539 are conserved in most other mammalian orthologues including hTRPV2, it is intriguing that both 2-APB and weak acids almost completely fail to activate hTRPV2 (Fricke et al, 2019; Neeper et al, 2007). Our data also show that HOAc-induced potentiation of CBD-induced currents in hTRPV2-expressing cells only becomes demasked following washout of HOAc. This effect may be explained by a pore block of hTRPV2 by protons, or by HOAc itself. The residues encoding for this property of hTRPV2 are yet to be identified, and in general the function of hTRPV2 is still poorly understood. As our data leave little doubt that weak acids gate TRPV2 only when they permeate the cell membrane, it is possible that the gating mechanism involves a modification of TRPV2-independent membrane properties that result in TRPV2 sensitization and activation. HOAc and other weak acids have been shown to induce several effects when permeating through the membrane, including an increase of lipid solubility, alterations of membrane elasticity and stiffness and a dissipating of the proton motive force across the membrane (Angelova et al, 2018; Guldfeldt and Arneborg, 1998; Zhou and Raphael, 2005; Zivanov et al, 2020). It is also possible that weak acid-sensitivity of TRPV2 requires an accumulation of the intracellular anion of the weak acid. We excluded that intracellular acetate activates TRPV2, but at this point we cannot say if a simultaneous interaction of protons and anions with intracellular or extracellular residues of TRPV2 is relevant.

Another relevant finding in this study is the Ca²⁺-induced inhibition of HOAc-induced activation of TRPV2. A similar property was described for TRPV3 (Wang et al, 2021; Xiao et al, 2008), and our data indicate the pore residue Glu609 dictates this property. This strong inhibition induced by extracellular calcium may give an indication of the physiological role of TRPV2 as a sensor for weak acids. TRPV2 seems to be functionally expressed in intracellular membranes such as endosomes (Cohen et al, 2015; Saito et al, 2007), or in case of the heart in the intercalated discs (Katanosaka et al, 2014). Considering that the intracellular concentration of Ca²⁺ is very low (~100 nM), it is possible that the regulation of TRPV2-activity by weak acids is more relevant in intracellular membranes. Saying this, our data also clearly indicate that weak acids can strongly sensitize TRPV2 at physiological concentrations of extracellular calcium. The physiological meaning of the acid-sensitivities of TRPV channels is generally poorly understood, and in vivo models specifically exploring acid-sensitivity are scarce. TRPV1 is thought to mediate acid-evoked pain, but aside from in vitro electrophysiological data, this assumption is probably best supported by studies using selective TRPV1-inhibitors in the acetic-acid writhing test in mice as well as in a model of acid-evoked pain in human volunteers (Caterina et al, 2000; Heber, Ciotu et al, 2020; Ikeda et al, 2001; Leffler et al, 2006). While weak acid-sensitivity of TRPV3 was suggested to account for a cytotoxic effect on keratinocytes leading to an exfoliation of the skin (Cao et al, 2012), activation of TRPA1 may explain the pungency of some weak acids (Wang et al, 2011). Our data indicate that activation of TRPV2 by weak acids can contribute to cell injury as well, a property that might be relevant in conditions associated with severe lactic or hypercapnic acidosis. The global TRPV2-knockout mouse suffers from a high perinatal lethality (Park et al, 2011), and to date selective modulators of TRPV2 are not available. Thus revealing the physiological relevance of weak acid-sensitivity of TRPV2 will require innovative strategies, and our mechanistic data offers a framework for future studies.

To conclude, we have identified weak acids as endogenous substances which are able to sensitize and even directly activate TRPV2. Our functional and structural data strongly suggest that the pocket formed by the positively charged residues His521, Arg535, and Arg539, located between S5 and the S4–S5 linker of two monomers, is likely to be an important interaction site for weak acids and further exogenous and endogenous TRPV2-ligands. This reinforces the title of the S4–S5 linker as a “gearbox” of TRP channels (Hofmann et al, 2017), and our data may facilitate the development of selective compounds targeting TRPV2.

The datasets produced in this study are available in the following databases: The atomic coordinates and cryo-EM density maps generated in this study have been deposited in the Protein Data Bank and Electron Microscopy Data Bank under the accession codes PDB 8EKP and EMD-28209 for TRPV2Apo1, PDB 8EKQ and EMD-28210 for TRPV2Apo2, PDB 8EKR and EMD-28211 for TRPV2Apo3, and PDB 8EKS and EMD-28212 for TRPV2pH5. The MD simulation datasets generated for this study can be accessed under: https://doi.org/10.5281/zenodo.10018876.

The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44318-024-00106-4.

Ferdinand M Haug: Formal analysis; Investigation; Methodology; Writing—original draft. Ruth A Pumroy: Data curation; Formal analysis; Investigation; Visualization; Writing—original draft. Akshay Sridhar: Formal analysis; Investigation; Visualization; Writing—original draft. Sebastian Pantke: Formal analysis; Visualization. Florian Dimek: Formal analysis; Investigation; Visualization. Tabea C Fricke: Formal analysis; Investigation; Visualization. Axel Hage: Formal analysis; Investigation; Visualization. Christine Herzog: Resources; Investigation; Methodology. Frank G Echtermeyer: Investigation; Visualization; Methodology. Jeanne de la Roche: Formal analysis; Investigation; Visualization; Methodology; Writing—original draft. Adrian Koh: Resources; Software. Abhay Kotecha: Resources; Software. Rebecca J Howard: Formal analysis; Validation; Investigation; Visualization. Erik Lindahl: Resources; Supervision. Vera Moiseenkova-Bell: Formal analysis; Validation; Visualization; Methodology; Writing—review and editing. Andreas Leffler: Conceptualization; Resources; Data curation; Formal analysis; Supervision; Funding acquisition; Validation; Investigation; Visualization; Methodology; Project administration; Writing—review and editing.

Source data underlying figure panels in this paper may have individual authorship assigned. Where available, figure panel/source data authorship is listed in the following database record: biostudies:S-SCDT-10_1038-S44318-024-00106-4.

The authors declare no competing interests.

The authors thank Mrs. Heike Bürger, Mrs. Kerstin Gutt, and Ms. Marina Golombek (all Hannover Medical School) for excellent technical assistance. The authors acknowledge the use of instruments at the Electron Microscopy Resource Lab and at the Beckman Center for Cryo-Electron Microscopy at the University of Pennsylvania Perelman School of Medicine. We also thank Stefan Steimle for assistance with Krios microscope operation at the University of Pennsylvania. We thank Sabine Baxter for assistance with hybridoma and cell culture at the University of Pennsylvania Perelman School of Medicine Cell Center Services Facility. AS was supported by Marie–Sklodowska–Curie grant 898762. EL and RJH were supported by the Swedish Research Council (2019-02433, 2021-05806), the Swedish e-Science Research Centre, and the BioExcel Center of Excellence (EU-823830). Computational resources were provided by the Swedish National Infrastructure for Computing. This work was supported by grants from the National Institute of Health (R01GM103899 and R01GM129357 to VYM-B). AL was supported by the Friedrich und Alida Gehrke-Stiftung.