Heat-responsive homotetrameric thermosensitive transient receptor potential vanilloid 1–4 (TRPV1-4) channels have a specific activation threshold and a high temperature sensitivity Q₁₀, which is the ratio of activation rates or open probabilities (P_o) of an ion channel measured at two temperatures 10 °C apart^1–8. A recent study revealed that reduced or oxidized mouse TRPV3 (mTRPV3) in the absence or presence of the C612-C619 disulfide bond can be activated by heat-induced unfolding of the least-stable K614-N647 H-bond in the pore domain (PD) (S5-S6) or the least-stable R416-D519 H-bond at the interface between the pre-S1 domain and voltage sensor-like domain (VSLD) (S1-S4) at the specific melting temperature threshold (T_m,th) of 52 °C or 40 °C, respectively. Furthermore, the specific temperature coefficient or sensitivity Q₁₀ of mTRPV3 is closely related to a change in molar heat capacity (ΔC_p). For a gating transition from a reduced and closed state to an oxidized and open state, ΔC_p is about 8.68 kcal/mol-K. This value decreases to 0.762 kcal/mol-K for an oxidized closed mTRPV3⁹. In this case, based on a single Gibbs–Helmholtz equation, both heat and cold unfolding-induced channel activations of TRPV3 should be observed¹⁰. However, the latter activation may not be directly detected if it occurs at a temperature below the freezing point of water¹¹.

Previous studies have shown that capsaicin-induced release of a phosphatidylinositol (PI) lipid from an active vanilloid site decreases the heat activation threshold of TRPV1^12–14. Therefore, when the phosphatidylcholine (PC) lipid at the corresponding vanilloid site is outcompeted by a ligand such as the natural cannabinoid tetrahydrocannabivarin (THCV), TRPV3 is expected to be activated above the freezing point of water. Given that both cold and heat activations of the minimal thermosensitive rat TRPV1 (rTRPV1) channel without the unstructured pore turret (604–626) upon the PI release from the same active vanilloid site share a similar structural or functional temperature coefficient (Ω₁₀ or Q₁₀, respectively)¹⁵, the same case should be observed for TRPV3 if both cold and heat activations start from the PC release from the same active vanilloid site.

To test this hypothesis, cryogenic electron microscopy (cryo-EM) structures of oxidized human TRPV3 (hTRPV3) activated by THCV at 4 °C were compared with those of heat-activated oxidized mTRPV3 at 42 °C^16,17. In addition, since the removal of the unstructured pore turret (604–626) does not significantly affect the heat activation of rTRPV1^8,18, the cryo-EM structures of rTRPV1 in the PI-bound closed state at 4 °C, the PI-free activated state at 4 °C and 42 °C were used as positive controls. After identifying two different temperature-dependent quaternary structures in the cold- and heat-evoked open states, the thermoring structures of THCV-gated oxidized hTRPV3 at 4 °C were characterized using a highly sensitive thermodynamic tool developed and examined in several thermosensitive proteins^{9,13–15,19–28}, and then compared with those of heat-gated oxidized mTRPV3 at 42 °C. As inactivation is involved in both heat activation of rTRPV1 and cold activation of hTRPV3^14,17, cryo-EM structures of homotetrameric inactivated and homopentameric pore-dilated hTRPV3 at 4 °C were also analyzed.

The results showed that the two open states of oxidized hTRPV3 or mTRPV3 at low and high temperatures had similar thermosensitivity but distinct quaternary and tertiary thermoring structures for different thermostabilities. This led to inactivation below 30 °C and no inactivation above 30 °C even in the presence of THCV at the active vanilloid site. These temperature-dependent gating pathways demonstrated that even if subsequent inactivation was involved, the initial hot and cold sensing abilities of the thermosensitive TRPV3 channel can still be linked and explained by the single Gibbs–Helmholtz equation for a significant ΔC_p, supporting the heat capacity model. In contrast, the initial inactivation in homotetrameric oxidized hTRPV3 induced unpredictable pore dilation in pentameric oxidized hTRPV3.

Results

Asymmetric intersubunit interactions of rTRPV1 during cold and heat activations

Previous studies have shown that the release of PI from the active vanilloid site is required for the activation of homotetrameric rTRPV1 by heat^12,13. Therefore, it is important to investigate whether any intersubunit interactions change during heat activation. In the PI-bound resting closed state at 4 °C, R579 in the PD of one subunit formed a cation-π interaction with Y565’ in the S4-S5 linker from a neighboring subunit (PDB: 7LP9). These swapping R579-Y565’ cation-π interactions may be crucial for homotetrameric channel closure, as the equivalent Y602-R616 cation-π interactions in TRPV4 promote the closed state²⁹. Additionally, E600 in the PD formed a salt bridge with R455’ in the VSLD of another neighboring subunit. Finally, the W549-F589’ swapping π interactions were also observed at the PD/VSLD’ interfaces (PDB: 7LP9) (Fig. 1)¹².

When PI at the vanilloid site was displaced by capsaicin at 48 °C, the Y565’-R579 π interactions were replaced by Y565’-F580 π interactions and the E600-R455’ bridges were broken in the open state (PDB: 7LPE). In contrast, the Y565’-R579 π bridges were disrupted but the E600-R455’ bridges remained in the inactivated state (PDB: 7LPD) (Fig. 1)^12–14. On the other hand, although the same intersubunit interactions were present in the open state along with an exchange between resiniferatoxin (RTx) and PI at the vanilloid site at 4 °C (7RQW)³⁰, the hybrid swapping interactions as shown in the heat-evoked open and inactivated states appeared in the RTx-free open state below 25 °C (8U3J or 8U3L) (Fig. 1)^12–14,31. Thus, despite mirrored cold and heat sensitivity upon PI release from the same vanilloid site, both tertiary and quaternary structures were different during cold and heat activations from the same starter¹⁵.

Heat-evoked channel opening at 42 °C rearranges the intersubunit interactions of oxidized mTRPV3

In the presence of the PC lipid at the active vanilloid site of the closed homotetrameric mTRPV3 channel with the C612-C619 disulfide bonds in cNW11 nanodiscs at 4 °C, two critical intersubunit interactions were identified along the PC-dependent minimal gating pathway from D396 to K705⁹. One was a π–π interaction between F625 from the pore helix of the nth subunit and Y460’ from the VSLD of the (n + 1)th subunit; the other was a CH-π interaction between F666 from the PD of the nth subunit and I637’ from the pore helix of the (n + 1) subunit. Thus, each PD was linked with the neighboring VSLD and PD (Fig. 2)¹⁶.

In contrast, in the heat-evoked open state at 42 °C, those swapping interactions mentioned above were replaced by the swapping L599-W559’ and A549-Y650’ π interactions, as well as the intersubunit R464-S607’ H-bonds (Fig. 2)¹⁶. Hence, the quaternary structure underwent significant changes during channel opening at high temperature (Fig. 2).

THCV-evoked channel gating transitions at 4 °C rearrange the intersubunit interactions of oxidized hTRPV3 differently

The same intersubunit interactions were observed in the oxidized closed hTRPV3 channel in cNW30 nanodiscs at 4 °C (Fig. 3). When THCV outcompeted the PC lipid from the active vanilloid site, along with the intact Y460-F625’ intersubunit interactions serving as an anchor, the I637-F666’ swapping π interactions were replaced by the swapping L599-W559’ π interactions in an open state or by N671-Y575’ π interactions in an inactivated state (Fig. 3)¹⁷. Therefore, THCV-induced gating transitions of oxidized hTRPV3 at 4 °C were completely different from those of oxidized mTRPV3 at 42 °C, even starting from the same closed state.

Closed oxidized hTRPV3 in cNW30 nanodiscs at 4 °C is unstable in the absence of THCV

Oxidized closed hTRPV3 in cNW30 nanodiscs at 4 °C exhibited some thermoring structures similar to oxidized closed mTRPV3 in cNW11 nanodiscs at 42 °C⁹. However, the distinct PC binding at the active vanilloid site resulted in some differences along the PC-dependent minimal gating pathway from D396 to K705. In cNW11 nanodiscs, one PC lipid in the VSLD is linked by S444 and K500 and the other at the vanilloid site is sandwiched by W521 and Q695⁹. In contrast, PC in the VSLD was packaged by S444, W493, K500 and H523 but PC at the vanilloid site only attached to W521 in cNW30 nanodiscs (Fig. 4A, B)¹⁷. In addition, the E610-N647-K614-E610 bridges in the PD of closed oxidized mTRPV3 at 42 °C were absent in closed oxidized hTRPV3 at 4 °C. Finally, the Q570-W692-R696 π interactions were absent in closed oxidized mTRPV3 at 42 °C but present in closed oxidized hTRPV3 at 4 °C (Fig. 4A)⁹. In this case, in order to understand the temperature-dependent gating pathways of oxidized TRPV3 upon a ligand exchange, it is necessary to evaluate the thermal stability of TRPV3 in different gating states. This can be achieved by using the grid thermoring model that has been recently developed and examined^{9,13–15,19–28}.

To accomplish this, the tertiary structure of the TRPV3 subunit was depicted as a systematic grid-like mesh network of noncovalent interactions. Once this network was constrained by graph theory, each noncovalent bridge was found to correspond to the shortest cyclic path, known as a grid or a thermoring. In addition to the direct path represented by the bridge, other nearby noncovalent bridges and segments of the polypeptide chain could create the shortest reverse path. The minimum number of free and unreactive amino acid side chains involved in this process was defined as the grid size (s). By assigning a unique grid size to each noncovalent bridge to symbolize the minimum folding energy, the energy landscape of the tertiary folding of TRPV3 could be visualized using this grid thermodynamic model so that the biggest grid could be identified to control the weakest or least-stable noncovalent bridge and denoted as “Grid_s”.

When the gating pathway was extended from D396 to K318 in the N-terminal and from K705 to G754 in the C-terminal, the total numbers of grid sizes and tertiary noncovalent interactions were 143 and 82, respectively (Fig. 4A). Thus, their ratio, referred to as the systematic thermal instability (T_i), was 1.74, higher than the 1.21 of closed oxidized mTRPV3 in cNW11 nanodiscs at 42 °C (Table 1)⁹.

PDB ID	8GKA	9DIJ	8V6K	8V6L	8V6M
Construct	hTRPV3
Lipid at the active vanilloid site	PC	free	PC	THCV
Redox state	Oxidized
Lipid environment			cNW30
Sampling temperature, °C	4
Gating state	Inactivated	Pore-dilated	Closed	Open	Inactivated
# of the biggest Grids	Grid9	Grid11	Grid17	Grid16	Grid18
grid size (s)	9	11	17	16	18
# of energetically equivalent basic H-bonds (n) controlled by Grids	1.2	0.8	1.0	1.0	1.9
Total non-covalent interactions (N)	97	75	82	82	73
Total grid sizes (S), a.a	94	121	143	137	114
Calculated Tm,th °C	48	40	30	32	37
Systemic thermal instability (Ti)	0.97	1.61	1.74	1.67	1.56
Calculated Ω10,cold at E = 1 kcal/mol				3.0
Experimental Q10,heat				2.32
Experimental Tth, °C			33
Refs for Tth and Q10			(3, 9, 16, 32)	(9, 32)

The comparative parameters are highlighted in bold.

Notably, the least-stable R416-D519 salt bridge in the biggest Grid₁₇ of closed oxidized mTRPV3 in cNW11 nanodiscs at 42 °C was accompanied by the H471-Y540-Y547-H471 π interactions⁹. In contrast, the same R416-D519 salt bridge of closed oxidized hTRPV3 in cNW30 nanodiscs at 4 °C was present along with the least-stable E467-K545 salt bridge controlled by the new biggest Grid₁₇ in the VSLD (Fig. 4A, C)¹⁷. It had a 17-residue thermoring from K545 to Y547, Y540, V531, F527, F526, Y448, F449, W559, T456, Y460, Y461, E467 and back to K545 (Fig. 4D). With 1.0 basic H-bond energetically equivalent to this least-stable E467-K545 salt bridge at the external interface between S1-S2 and S3-S4 linkers, the calculated T_m,th for its heat unfolding was about 30 °C (Table 1), which was similar to the measured slow warm activation threshold of 33 °C of TRPV3 but lower than the calculated T_m,th of 40 °C of the oxidized closed mTRPV3 channel in cNW11 nanodiscs at 42 °C^3,9,16,32. Therefore, oxidized hTRPV3 in cNW30 nanodiscs can remain closed until above 30 °C. Of special note, the same critical swapping Y575-K589’ π interactions as shown in reduced closed hTRPV3 (PDB: 6UW4) were observed between the S4-S5 linkers and the PDs for homotetrameric channel closure (Fig. 4E)^15,33.

THCV binding stabilizes oxidized hTRPV3 in the open state at 4 °C

The closed oxidized hTRPV3 channel in cNW30 nanodiscs was unstable when the PC lipid at the active vanilloid site only attached to W521. When THCV outcompeted the PC lipid from the active vanilloid site at 4 °C, it was wrapped by W521, F522 and N561 via two π interactions and an H-bond. In addition, as expected for mirrored cold and heat activations^10,15, the E467-K545 salt bridge in the biggest Grid₁₇ was disrupted along with the broken Y575-K589’ swapping bridges (Fig. 5A, B)¹⁷. Although the stimulatory D519-R567 H-bond in the open state of oxidized mTRPV3 in cNW11 nanodiscs at 42 °C was also present in the open oxidized hTRPV3 in cNW30 nanodiscs at 4 °C^9,17, the presence of the new K500-E702, E405-K705, T411-D512, Y451-Q529 and other bridges allowed the new least-stable Q346-N394 H-bond to be controlled by the biggest Grid₁₆ for channel opening in cNW30 nanodiscs at 4 °C (Fig. 5A)¹⁷. It featured a 16-residue thermoring from W331 to Q346, N394, D391, R729, and back to W331 (Fig. 5C, D). With a basic H-bond energetically equivalent to the least-stable Q346-N394 bridge at the internal interface between the N-terminal linker domain (NLD) and ARD (Fig. 5C), the calculated T_m,th to unfold it was about 32 °C (Table 1). In contrast, the different least-stable N412-D512 H-bond is governed by the biggest Grid₂₃ for T_m,th of 23 °C of the open K169A mutant¹⁵. Thus, despite the difference in the tertiary thermoring structure between open oxidized hTRPV3 in cNW30 nanodiscs at 4 °C and open oxidized mTRPV3 in cNW11 nanodiscs at 42 °C (Fig. 5A)⁹, the replacement of PC with THCV at the active vanilloid site slightly stabilized the cold-evoked open state of oxidized hTRPV3 in cNW30 nanodiscs.

In that case, although the cold-induced structural transition of globular proteins like ubiquitin and Trp-cage miniprotein below water’s freezing point is non-cooperative, fragmented and locally limited at the protein-water interface^11,34, a global conformational change during THCV-induced cold activation at low temperature above the freezing point of water was cooperative (Figs. 4A, 5A). Along the same PC-dependent gating pathway from K318 to G754, such a change decreased the total number of grid sizes from 143 to 137 (Figs. 4A, 5A). Thus the systematic thermal instability (T_i) was 1.67, similar to the 1.74 of closed oxidized hTRPV3 but lower than the T_i of 2.55 in the open K169A mutant (Table 1)¹⁵. Notably, after comparing a change in the total grid sizes and noncovalent bridges between closed and open states at low temperatures, the structural cold sensitivity (Ω_10,cold) was calculated as 3.0., similar to the heat sensitivity (Q_10,heat = 2.32) of mTRPV3⁹. Therefore, the removal of PC from the same active vanilloid site did trigger both cold activation by THCV and heat activation of oxidized TRPV3^10,15.

Inactivation further stabilizes oxidized hTRPV3 at 4 °C

In the THCV-induced inactivated state, the least-stable Q346-N394 H-bond at the NLD/ARD interface was disrupted. Meanwhile, the stimulatory D519-R567 H-bond at the S4-S5 linker/VSLD interface was also broken but the inhibitory K581-E679 salt bridge at the S4-S5 linker/S6 interface was present as the least-stable in the biggest Grid₁₈ (Fig. 6A)¹⁷. It had an 18-residue thermoring from L670 to E679, K581, H585, D586, Y594,Y661, T665, F666, and back to L670 (Fig. 6B, C). For 1.9 basic H-bonds to be energetically equivalent to the least-stable K581-E679 bridge, the calculated T_m,th to unfold it was about 37 °C, higher than the 32 °C of the open state (Table 1).

Of special note, in addition to the broken Y575-K589’ intersubunit bridges (Fig. 6D), the non-native structure was present in the inactivated state due to the disconnection of the highly conserved D586-T680 and F590-L673 bridges in the normal closed and open states of oxidized mTRPV3 in cNW11 nanodiscs at 42 °C (Fig. 6A)⁹. Furthermore, along the expanded PC-dependent gating pathway from K318 to G754, after reducing the total number of grid sizes and noncovalent interactions from 137 and 82 to 114 and 73, respectively, the systematic thermal instability (T_i) also decreased from 1.67 to 1.56 (Table 1). Thus, both the increased T_m,th and decreased T_i indicated that the inactivated state was more stable than the open state in the presence of THCV.

Swapping Y575-K589’ bridges prime TRPV3 closure before predictable opening

A recent investigation demonstrated that the highly conserved intersubunit interaction between a Tyr residue in the S4-S5 linker from one subunit and an Arg/Lys residue on S5 must be finally disrupted for TRPV1-4 opening¹⁵. This study further indicated that the swapping Y575-K589’ bridges were formed in the closed state but disrupted in the open state and the subsequent inactivated state (Figs. 4E, 5B, 6D)¹⁷. Therefore, it is of special interest to test whether disrupting swapping Y575-K589’ bridges can directly open TRPV3.

In the absence of cNW30 nanodiscs that prevent membrane diffusive protomer exchange, the PC lipid in the VSLD was linked by S444, W493, K500, and Y564. However, the PC lipid at the vanilloid site of oxidized hTRPV3 was tightly packaged by W521, F522, Y564, R567, S576, Q695 via noncovalent bridges (PDB: 8GKA) (Fig. 7A, B)³⁵. When the E467-K545 salt bridge was disconnected, D519 H-bonded not only with R416 in the pre-S1 domain but also with R698 in the TRP domain. Along the similar extended PC-dependent gating pathway from K318 to G754, the total numbers of grid sizes and tertiary noncovalent interactions changed from 143 and 82 to 94 and 97, respectively (Figs. 4A, 7A). Therefore, the systematic thermal instability (T_i) dramatically decreased from 1.74 to 0.97 (Table1). Furthermore, the highly conserved D586-T680 H-bond in the pore domain, which is necessary for normal TRPV3 gating^9,36, became the weakest (Fig. 7A, C). It was governed by the biggest Grid₉ via a thermoring from D586 to F590, L673, T680 and back to D586 (Fig. 7D). When this weakest bridge was energetically equivalent to 1.2 basic H-bond (1.2 kcal/mol), the calculated T_m,th was 48 °C, higher than 30 °C of oxidized closed hTRPV3 in the presence of cNW30 nanodiscs (Table 1). Thereafter, both lower T_i and higher T_m,th suggested a very stable subunit in the initial gating state when the cNW30 nanodiscs were removed.

On the other hand, the critical Y575-K589’ swapping bridges were found to be broken, but the channel pore was still closed (Fig. 7E)³⁵. Given that disrupting the weakest D586-T680 bridge disables normal TRPV3 activation^9,36, the isolated whole hTRPV3 channel could not be opened normally without cNW30 nanodiscs. In other words, the channel was actually inactivated. When the crucial D586-T680 H-bond is broken along with PC release from the vanilloid site (Fig. 8A), this homotetrameric channel partially transitions to the homopentameric one^35,37. While the weakest W521-L584 bridge was formed between the VSLD and the S4-S5 linker, together with the broken Y575-K589’ bridge (Fig. 8B), favoring pore dilation (Fig. 8A)³⁷, it was controlled by the biggest Grid₁₁ via a thermoring from W521 to F522, Y564, Y565, R567, Q695, R693, E689, S688, M574, L584 and back to W521 (Fig. 8C, D). When the least-stable W521-L584 bridge was energetically equivalent to 0.8 basic H-bond (0.8 kcal/mol), the T_m,th was calculated as 40 °C, lower than the 48 °C of the homotetrameric hTRPV3 (Table 1). Furthermore, along the similar extended PC-dependent gating pathway from K318 to G754, the total numbers of grid sizes and tertiary noncovalent interactions changed from 94 and 97 to 121 and 75, respectively (Figs. 7A, 8A). Therefore, the systematic thermal instability (T_i) increased from 0.97 to 1.61 (Table 1). In other words, the homopentameric hTRPV3 channel was relatively unstable. Taken together, the swapping Y575-K589’ bridges must remain in the initial resting closed state for normal TRPV3 function. Once they were broken, the whole channel would become unstable and unpredictable, with a smaller fraction of pentamers and a background of lower oligomers³⁵, regardless of the initial stable subunits.

Discussion

The thermosensitive TRPV1-4 channels are well-known for their specific activation thresholds and high temperature sensitivity (Q₁₀). The former can be explained by the melting temperature threshold (T_m,th) of the least-stable noncovalent interaction along the lipid-dependent minimal gating pathway from the pre-S1 domain to the TRP domain^9,13–15. However, the energy origin of the high Q₁₀ is still not well understood when inactivation is involved. Although a single Gibbs–Helmholtz equation has been predicted to link both cold and hot sensations of TRPV1-4 channels for a significant ΔC_p¹⁰, cold activation has not been examined for the native construct with inactivation. Since the heat capacity difference (ΔC_p) underlies heat sensing of TRPV3⁹, cold activation should be detected. In this computational study, symmetric initial cold and heat activations from the same starter were used to further examine this model^10,15. Quaternary and tertiary thermoring structural analyses demonstrated that different gating states had distinct tertiary thermoring structures for various thermostabilities, leading to temperature-dependent gating pathways of oxidized TRPV3. However, the mirrored thermosensitivity during initial cold and heat activations, once triggered by the PC release from the same active vanilloid site, still confirmed this heat capacity model, regardless of the subsequent inactivation at low temperature. On the contrary, the initial inactivation induced an unpredictable change in the quaternary structure during the pore dilation.

Gating state-dependent thermostability

Thermodynamic analyses revealed that each gating state of oxidized hTRPV3 in cNW30 nanodiscs at 4 °C exhibited different thermostability. Even with the PC lipid present at the active vanilloid site, the closed state had a calculated T_m,th of 30 °C (Table 1). When PC was replaced by THCV at the vanilloid site, the T_m,th values of the open and inactivated states increased to 32 °C and 37 °C, respectively (Table 1). In contrast, the closed and open states of oxidized mTRPV3 in cNW11 nanodiscs have T_m,th values of 40 °C and 61 °C, respectively⁹. Notably, while the final inactivated state of oxidized hTRPV3 in cNW30 nanodiscs at 4 °C showed the lower systematic thermal instability (T_i) of 1.56 than the common T_i around 1.71 in closed and open states (Table 1), the closed and open states of oxidized mTRPV3 in cNW11 nanodiscs at 42 °C have a lower T_i of 1.20⁹. In this regard, these gating state-dependent thermal stabilities at low and high temperatures may lead to asymmetric temperature-dependent gating pathways.

Asymmetric temperature-dependent gating pathways

Oxidized hTRPV3 can remain closed in cNW30 nanodiscs when the temperature is below 30 °C. If the same heat-evoked open state of oxidized mTRPV3 in cNW11 nanodiscs is applied to oxidized hTRPV3 in cNW30 nanodiscs, the release of PC from the active vanilloid site after heat stimuli could disrupt the weakest E467-K545 bridge above 30 °C for channel opening (Fig. 9). In agreement with this proposal, oxidized mTRPV3 can be activated above 30 °C after prolonged heat exposure^3,9,16,32. Consequently, if a ligand such as THCV displaces the PC lipid from the active vanilloid site, oxidized hTRPV3 could potentially be activated by both heat and cold stimuli through different gating pathways.

Given that oxidized hTRPV3 had a similar T_i around 1.71 in the PC-bound closed and THCV-opened states at 4 °C (Table 1), these two gating states may coexist at low temperature as recently proposed¹⁷. However, their distinct T_m,th values were 30 °C and 32 °C, respectively, together with the lower T_i of 1.56 and the higher T_m,th of 37 °C in the inactivated state (Table 1), may shift the balance from the close state to the open one and then the inactivated one below 30 °C (Fig. 9). On the other hand, oxidized mTRPV3 has the T_m,th of 61 °C and the T_i of 1.20 (Table 1)⁹. Therefore, an increase in thermostability may allow oxidized TRPV3 in cNW30 nanodiscs to remain open above 30 °C without any inactivation even in the presence of THCV (Fig. 9). In this regard, the structural and functional characterization of ligand or drug binding to a target is necessary at human body temperature. However, the reported cases are still limited^17,38–41.

Asymmetric quaternary structures during initial cold and heat activations

Given that homotetrameric oxidized TRPV3 channels had different tertiary thermoring structures in various temperature-dependent gating states (Figs. 3A, 4A, 5A)⁹, it is reasonable that they also exhibited the distinct quaternary structural changes during initial cold and heat activations (Figs. 2–3). Given that oxidized hTRPV3 in cNW30 nanodiscs at 4 °C shared the same R416-D519 and Y594-T636-Y661 bridges with oxidized mTRPV3 in cNW11 nanodiscs at 4 °C in the closed state with PC bound at the same active vanilloid site (Fig. 4A)⁹, it is reasonable that they had the same swapping Y460-F626’ and I637-F666’ π interactions (Figs. 2–3, 9). Furthermore, both closed states exhibited the same swapping K169-D752’ and Y575-K589’ bridges (Fig. 9), in agreement with previous reports^15,33,42.

Once THCV outcompeted the PC lipid from the active vanilloid site as heat does⁹, although the open state of oxidized hTRPV3 at 4 °C displayed the same stimulatory D519-R567 bridges along with broken R416-D519 and K169-D752’ and Y575-K189’ bridges as the heat-opened oxidized mTRPV3 at 42 °C (Fig. 5A)⁹, the Y460-F625’ swapping bridges in the open state at 4 °C were substituted by the A549-Y650’ ones in the open state at 42 °C, along with the same intersubunit W559-L599’ interactions (Figs. 2, 3 and 9). The same quaternary structure is also present in oxidized and activated hTRPV3 with citronellal at the vanilloid site in the presence of membrane scaffold protein 2N2 (MSP2N2) nanodiscs⁴¹.

Symmetric temperature sensitivity during initial cold and heat activations

Since inactivation is the final step of hTRPV3 after cold activation by THCV below 30 °C, a comparison of thermosensitivity between initial cold and heat activations upon the release of PC from the same active vanilloid site could be used to examine the heat capacity model^10,15. For THCV-induced cold activation of oxidized hTRPV3, a small decrease in total grid sizes from 143 to 137 would result in a lower mean structural cold sensitivity (Ω_{10, cold}) of 3.0 (Table 1), similar to the functional heat sensitivity (Q_{10, heat}) of 2.32 in mTRPV3⁹. Given that the same rTRPV1 channel exhibits cold and heat activations with symmetric thermosensitivity upon the release of PI from the same active vanilloid site (Fig. 1)¹⁵, the similar results between cold activation of oxidized hTRPV3 below 30 °C and heat activation of oxidized mTRPV3 above 40 °C upon the release of PC from the same active vanilloid site demonstrate that both cold and hot sensing abilities in either TRPV1 or TRPV3 can be linked and explained by the single Gibbs–Helmholtz equation for a temperature-independent ΔC_p¹⁰.

In contrast, the human TRP ankyrin 1 (hTRPA1) channel has been identified as an intrinsic redox state-dependent bidirectional thermosensor with a temperature-dependent gating pathway. Unlike cold activation below 22 °C, its heat activation above 22 °C is accompanied by inactivation or desensitization. More importantly, its initial cold and heat activations are asymmetric^43,44. Therefore, separate channel structures for cold and heat sensitivity suggest that different quaternary structural rearrangements may be involved in the dual but asymmetric cold and heat sensation of thermosensitive TRPA1 channels to generate a temperature-dependent ΔC_p^44,45.

Because Xenopus oocytes expressing frog TRPV3 (fTRPV3) also exhibit asymmetric cold activation below 20 °C and heat activation above 50 °C^46,47, further structural investigations are necessary to determine if the quaternary structures in cold- and heat-evoked open states are different even though they share the same closed fTRPV3 channel.

Timing of inactivation also triggers various gating pathways

In the heat-evoked inactivation of rTRPV1 from the same pre-open closed state above 43 °C, the R557-E570, D576-T685 and F580-L678 bridges remain conserved. However, the H410-E692/N695 bridges inactivate the rTRPV1, resulting in the absence of swapping Y565-F580’ bridges but the presence of intersubunit R455-E600’ bridges (Fig. 1)^14,15.

In contrast, in the THCV-induced inactivated state at 4 °C, the same swapping of the K169-D752’ H-bond and the Y460-F625’ π interaction in closed oxidized hTRPV3 was also observed. However, the swapping of the N671-Y575’ π interaction differed from the swapping of the I637-F666’ π interaction (Figs. 2, 3). This discrepancy could be attributed to the presence of the K581-E679 salt bridge in the biggest Grid₁₈ and the absence of the D586-T680 and F590-L673 bridges in the inactivated state (Figs. 4A). In any case, the swapping of the Y575-K589’ bridges in hTRPV3 were finally disrupted in the inactivated state (Figs. 6D)^9,15. Therefore, the initial normal heat or cold activation was not affected. However, the initial inactivation upon breaking the swapping of the Y575-K589’bridges caused hTRPV3 to be pore-dilated along with a significant change in the quaternary structures (Figs. 7E, 8B). Hence, inactivation timing also plays a critical role in regulating the gating pathways of thermosensitive TRP channels.