Heat activation mechanism of TRPV1: New insights from molecular dynamics simulation

ABSTRACT

As a member of the transient receptor potential (TRP) channels superfamily, the TRPV1 channel undergoes a closed-to-open gating transition in response to various physical and chemical stimuli including heat. Thanks to recent progress in cryo-electron microscopy, high-resolution structures are becoming available for various TRP channels including TRPV1. This has enabled us to study the molecular mechanism of TRPV1 channel gating by using molecular simulation. Here we review recent progress in molecular simulations of TRPV1 channel by us and others, with focus on our molecular dynamics (MD) simulations of TRPV1 at different temperatures. While no consensus has been reached on the heat activation mechanism of TRPV1, the simulations have offered specific predictions and models for future experimental studies to test.

The transient receptor potential (TRP) channels are a superfamily of cation channels [1,2] which undergo a closed-to-open gating transition in response to various physical and chemical stimuli including heat [3,4] and cold [5–7]. These channels make promising drug targets [8,9] thanks to their involvements in various diseases [10,11]. As a representative TRP channel, TRPV1 forms a homo-tetramer. Each subunit is comprised of a six-helix (S1-S6) transmembrane domain (TMD) and an intracellular domain (ICD) (see Figure 1). The TMD consists of a S1-S4 module on the periphery and a S5-S6 module enclosing a central pore. The N-terminal portion of ICD forms an ankyrin repeats domain (ARD) [12]. The C-terminal domain (CTD) of ICD contains a highly conserved TRP helix [13]. At the TMD-ICD interface lies a membrane proximal domain (MPD), which was implicated in heat activation [14]. Other functional and mutational studies identified alternative heat-sensing sites in the CTD [15–17], the outer pore, and the pore domain [18–21]. It remains unknown whether these sites directly sense heat or indirectly mediate downstream gating events [22]. Several key linker domains may contribute to the allosteric couplings to the activation of TRPV1, including the MPD linker, the S2-S3 linker, and the S4-S5 linker (see Figure 1).

Figure 1.

Structural architecture of TRPV1 tetramer (a) in the side view and (b) in the top view. A representative subunit is colored by domain: the ARD (pink), the MPD linker (green), the S1-S2 linker (yellow), the S2-S3 linker (cyan), the S4-S5 linker (purple), the outer pore (red), and the TRP helix (blue). Residues G643 and I679 at the upper and the lower gate are shown as balls colored in light and dark gray, respectively. In (b), the heat-activated motions of the MPD linker, the S1-S2 linker, the S2-S3 linker, and the S4-S5 linker are marked by arrows colored by domain. Based on our MD simulations, we proposed the following sequence of domain motions during the heat activation: MPD linker and S1-S2 linker → S2-S3 linker → S4-S5 linker → opening of lower gate (I679).

High-resolution structures of full-length channels are required to elucidate the TRP-channel activation mechanism. Electron microscopy and X-ray crystallography were used to solve low-resolution structures of TRP channels [23–29] and high-resolution structures of their truncated domains [30–44]. Recently, the labs of Cheng and Julius solved and refined the first high-resolution (~3.3 Å) structures of a minimal functional construct of rat TRPV1 in detergents [45,46] and in lipid nanodisc [47] by cryo-electron microscopy (cryo-EM), capturing distinct conformations of TRPV1 with a closed or open channel. More TRP channels structures were later solved [48–50]. The new TRPV1 structures revealed a dual-gate channel pore with two constrictions (see Figure 1) – an upper gate (selectivity filter) near the outer pore (residues G643 and M644) and a lower gate in the S6 helix (residue I679). Both gates are closed (open) in the closed (open) structures [45–47]. While there is no functional evidence for the selectivity filter serving as a gate, both heat and capsaicin gate the lower gate as first shown by Salazar et al. [51]. Despite providing structural insights, these cryo-EM structures have their limitations owning to their non-physiological conditions (e.g. cryo temperature and activation by toxin/agonist rather than heat, etc), so they cannot illuminate the heat activation mechanism of TRPV1. It remains unknown what is the structural and energetic basis for the large enthalpy change that dictates the high temperature sensitivity of TRPV1 gating [52]. Various models were proposed based on solvation/de-solvation of non-polar/polar residues [53,54], or denaturation of heat-sensing domain [55]. Recently, a study by Sanchez-Moreno et al. provided evidence that denaturation or a similar irreversible mechanism might participate in TRPV1 temperature gating [56]. Another study by Sosa-Pagan et al. provided initial correlative support for the model that heat activation is driven by the exposure of hydrophobic residues to solvent [57].

Molecular Dynamics (MD) simulation is well suited for investigating protein dynamics and energetics under physiological conditions with atomic details [58]. Despite recent developments in computing hardware and software, it is still impractical for MD to access the μs – ms time scales relevant to the gating transition in TRPV1. So MD is largely limited to simulations of fast transitions (e.g. ion permeation) or equilibrium of stable states (e.g. closed and open state of a channel). To overcome the time-scale limit of MD simulation, coarse-grained modeling (e.g. the elastic network model [59–61]) has been developed using simplified protein representations and force fields [62,63].

In an early study [64], we combined coarse-grained modeling and atomistic MD simulation of the first TRPV1 structures [45,46]. Our normal mode analysis captured two key modes of domain motions involved in the TRPV1 gating transition, featuring a global twist motion of the ICD relative to the TMD as observed between the closed and open structure. Our transition pathway modeling predicted a sequence of domain movements that propagate from the ARD to the TMD via the MPDs and the CTDs, leading to final opening of the channel pore. Additionally, our MD simulation of the ICD fragment identified key residues in the MPD and the CTD that contribute differently to the nonpolar energy of the open and closed state, and these residues were predicted to control the temperature sensitivity of TRPV1.

In a subsequent study [65], we performed extensive MD simulations (total ~3 μs) using the closed and open structures of TRPV1 [45,46]. In the closed (C) state simulations at 30°C, we observed a stably closed channel constricted at the lower gate (residue I679), while the upper gate (residues G643 and M644) undergoes dynamic fluctuations to allow passage of water and ions. This is consistent with the proposal that the functional gate of TRPV1 channel is the lower gate and not the “upper gate”. In the open (O) state simulations at 60°C, we found higher conformational variation consistent with a large entropy increase required for the heat activation, and both the lower and upper gates undergo dynamic fluctuations. Through ensemble-based structural analyses of the two states, we found pronounced closed-to-open conformational changes involving the MPD linker, the outer pore, and the TRP helix, which are accompanied by breaking/forming of a network of C/O-state specific hydrogen bonds. By comparing the C-state simulations at 30°C and 60°C, we observed heat-activated conformational changes in the MPD linker, the outer pore, and the TRP helix that resemble the closed-to-open conformational changes, along with partial formation of the O-state specific hydrogen bonds.

Following the recent publication of improved TRPV1 structures in nanodisc [47], we conducted more MD simulations using the new structures and a new simulation strategy to accelerate the TRPV1 gating transition via mutations. In this recent study [66], we ran extensive MD simulations (total ~11 μs) for WT TRPV1 at three different temperatures (30°C, 60°C, and 72°C), and for a constitutively active TRPV1 double mutant. Our simulation aimed to uncover heat-activated conformational changes in various domains of the WT channel, and directly observe the accelerated gating transition in the TRPV1 mutant. Encouragingly, the mutant simulation revealed a series of domain motions leading to the lower-gate opening, which agree well with the heat-activated conformational changes observed in the WT simulation. Our MD simulations also supported the contribution of protein-water electrostatic interaction to the high temperature sensitivity of TRPV1 gating.

In addition to our own studies, other labs have performed molecular simulation studies to probe different aspects of TRP channel gating. In an early study [67], Domene and coworkers studied the permeation and ion selectivity based on MD simulation of the TRPV1 pore domain. Their subsequent study [68] used MD simulation enhanced by bias-exchange meta-dynamics to further probe different energetic barriers between Na⁺ and K⁺ permeation. Besides permeation, MD simulation was also used to study TRPV1-ligand interactions. In one study, Domene and coworkers employed unbiased and biased MD simulations to calculate the free energy landscape of capsaicin binding to TRPV1 [69]. Hanson et al studied the movement of capsaicin through the lipids, and proposed a possible pathway of capsaicin binding [70]. In other simulation studies, the conformational changes induced by PI(4,5)P₂ binding were investigated [71], and the binding sites of general anesthetics were identified [72]. In addition, Feng et al carried out MD simulations of human TRPV1 bound with agonist/antagonist to probe the ligand-binding-induced conformational changes and validate their homology model of human TRPV1 [73,74]. To elucidate the TRPV1 heat activation, one MD study simulated the transmembrane domain of TRPV1 at different temperatures, which observed transient partial opening of the channel [75]. The authors suggested the channel is more hydrophobic in the open state than in the closed state, which agreed with a previous model of TRPV1 thermal sensing [53]. However, in another recent study [76], Carnevale and coworkers suggested the open channel is more hydrophilic than the closed one, and they proposed a new model of TRPV1 heat activation based on a dewetting transition. In a subsequent paper [77], the same authors suggested a mechanism for TRPV1 activation, which involves rotation of a conserved asparagine in S6 from a position facing the S4-S5 linker toward the pore, and this rotation is associated with hydration of the pore and dehydration of the four peripheral cavities located between each S6 and S4-S5 linker. In another paper, the authors proposed another model of TRPV1 heat activation which involves lipid molecules escaping from the vanilloid-binding pocket [78]. Based on the NMR spectroscopy and MD simulation of a homology model of the C-terminal domain of TRPV1, Raymond et al found that the C-terminal domain is unstructured at room temperature, and higher temperature and PI(4,5)P₂ binding induce substantial conformational changes that may sensitize the channel [79]. In sum, although no consensus has emerged from various MD simulations of TRPV1 heat activation, the simulations made specific predictions for future experiments to test. Future MD simulation of longer time will be needed to directly observe the gating transition of the full-length WT channel.

In the remainder of this review, we will highlight key findings of our recent MD simulation study of TRPV1 heat activation [66].

MD simulation of WT TRPV1 shows temperature-dependent changes of dynamics

To explore the temperature-dependent conformational dynamics of TRPV1 in the C state, we conducted three sets of ten 200-ns MD simulations based on the new closed-channel structure (PDB id: 5irz) at 30°C, 60°C, and 72°C (named C30, C60, and C72). As in our previous study [65], we did not observe channel opening at 60°C and 72°C owning to limited simulation time. However, these simulations are still informative in elucidating the heat-dependent conformational dynamics of TRPV1 in the C state.

To assess the per-residue conformational flexibility, we calculated the root mean squared fluctuation (RMSF) for three C-state ensembles. The RMSF profiles exhibit several pronounced peaks (see Figure 2(a)) in the MPD linker, the S1-S2 linker, the S2-S3 linker, the outer pore, and the C-terminus of the TRP helix. In addition, We found heat-dependent RMSF increase in the MPD linker, the S1-S2 linker, the S2-S3 linker, the outer pore, the lower S6 helix, and the TRP helix (see Figure 2(b)). These temperature-dependent RMSF increases are larger than what is expected from an unspecific thermal effect. So these regions may be involved in sensing temperature change in TRPV1. As a control, further MD simulation of a structurally similar yet non-heat-activated channel will be able to assess if such heat-dependent RMSF increases indeed correlate with heat activation.

Figure 2.

(a) RMSF profiles at 30°C, 60°C, and 72°C (C30, C60, and C72). (b) The fractional change in RMSF from C30 to C60, and from C30 to C72, where a gray horizontal line corresponds to an increase of RMSF by 7% due to unspecific thermal effect. The residue positions corresponding to the MPD linker, the S1-S2 linker, the S2-S3 linker, the S4-S5 linker, the outer pore, the TRP helix, and the upper/lower gate are marked by horizontal bars colored in green, yellow, cyan, purple, red, blue, and black, respectively.

Channel pore analysis reveals a more dynamic upper gate than the lower gate

Following our previous study [65], we probed the channel pore dynamics by calculating the pore radius for each C-state ensemble at the upper and lower gate. The pore radius at the lower gate is 0.72 ± 0.14 Å at 30°C, 0.81 ± 0.22 Å at 60°C, and 0.80 ± 0.17 Å at 72°C, suggesting the lower gate is largely closed in the C state at all three temperatures. At the upper gate, the pore radius is 0.86 ± 0.25 Å at 30°C, 0.80 ± 0.27 Å at 60°C, and 1.06 ± 0.33 Å at 72°C. Therefore, the upper gate is more dynamic than the lower gate, allowing passage of water and ions through the upper gate (but not the lower gate) even in the C state [66]. No ion permeation event was observed during our MD simulations of the WT channel. Our finding is consistent with the lower gate being the main activation gate of TRPV1.

R_g analysis uncovers heat-activated expansion/contraction of TRPV1 domains

By comparing various cryo-EM structures [45,47] and MD-averaged structures of TRPV1 in the C/O state [65], we observed extensive conformational changes which may drive the channel activation. However, since the open-channel structures of TRPV1 were obtained in complex with agonist/toxin under the cryogenic condition [45,47], it is uncertain if those cryo-EM-observed conformational changes are relevant to the heat activation of TRPV1. To directly observe heat-activated conformational changes in TRPV1 under physiological conditions, we analyzed conformational changes from 30°C to 60/72°C in our MD simulations. We focused on those concerted inward/outward domain motions by calculating the radius of gyration (R_g) at every residue position of the core TRPV1 structure. Interestingly, we observed heat-activated contraction in MPD linker and S2-S3 linker, as well as heat-activated expansion in S1-S2 linker, S4-S5 linker and outer pore (see Figure 3(a)). These MD-observed changes are distinct from the cryo-EM-observed changes. Notably, in the S6 helix, we found no heat-activated change in R_g near the lower-gate residue I679. Instead, we observed a pronounced increase in R_g at N687 (see Figure 3(a)), suggesting a heat-activated dilation in the pore below the lower gate which may subsequently break the hydrophobic seal at I679. Additionally, in the TRP helix, we observed little heat-activated change in R_g, which is in contrast to the large motion of TRP helix observed between the closed and open structures from cryo-EM [45,47].

Figure 3.

The results of WT TRPV1 simulations. (a) A TRPV1 subunit colored by the heat-activated R_g change (with positive/negative change corresponding to red/blue). Outward/inward motions are indicated by block arrows colored in red/blue. Marker positions for key domains are shown as balls and labeled. (b) Key HBs formed at low temperature; (c) Key HBs formed at high temperature. In (b) and (c), the HB-forming residues are labeled, and colored by atom type (carbon: green, nitrogen: blue, oxygen: red); key domains are colored as follows: S2-S3 linker (cyan), S4-S5 linker (purple), outer pore (red), and TRP helix (blue).

Together, the above heat-dependent domain motions support the following heat activation pathway in TRPV1: upon temperature increase, the MPD linker and S2-S3 linker move inward to engage with the S4-S5 linker, causing the latter to move outward and allow the pore-forming S6 helix to expand for gating. Additional expansions in the S1-S2 linker and the outer pore may further contribute to gating. This putative pathway overlaps with a previously proposed vanilloid activation pathway involving the S2-S3 linker [80] and the S4-S5 linker [45].

Dynamic hydrogen bonds form and break to enable heat-activated conformational changes

To probe dynamic interactions that drive the heat-activated conformational changes in TRPV1, we focused on those hydrogen bonds (HBs) specifically formed in three C-state ensembles of different temperatures (30°C, 60°C, and 72°C). Following our previous study [65], we analyzed those intra-subunit HBs that couple between key domains of TRPV1 (including the ARD, the MPD, the S2-S3 linker, the S4-S5 linker, the S5 helix, the outer pore, the S6 helix, and the TRP helix, see Figure 1). We found heat-activated formation of HBs (i.e. with higher occupancy at 60/72°C than at 30°C, see Figure 3(c) and Table 1 of ref [66]) between the following residue pairs: E326-R367, R355-L365, E356-T370, S510-E570, Y511-E570, S512-E570, and T641-Y666. Some of them couple the S2-S3 linker (S510, Y511, and S512) to the S4-S5 linker (E570), which may drive the observed heat-activated motions between these linkers. We also observed heat-induced breaking of HBs (i.e. with higher occupancy at 30°C than at 60/72°C, see Figure 3(b)) between Q560-R701 and D576-T685. These HBs may anchor the N/C-termini of S4-S5 linker on the TRP helix and the S6 helix in the C state, so their breaking could enable the heat-activated motion of the S4-S5 linker relative to the TRP helix and the S6 helix. Together, the above dynamic HBs may facilitate heat-activated motions of the S2-S3 linker and the S4-S5 linker, thereby mediating the aforementioned heat activation pathway in TRPV1. Additionally, we also found some high-occupancy HBs which are temperature-insensitive (including R409-D509, K425-E709, R499-D707, and R575-E692), and they may stabilize the allosteric network for channel activation.

For validation, we have surveyed previous mutational studies of the above HB-forming residues. The Q560H/R mutations caused gain of function to TRPV1 when expressed in yeast [81], which is consistent with its predicted role in stabilizing the C state. Mutations in E570 and D576 compromised heat activation [82]. The T641S mutant showed large constitutive channel activation when expressed in yeast [81]. The Y666A mutation resulted in a non-functional channel [83]. The R701A mutation caused strongly reduced PI(4,5)P₂-dependent activation [15], suggesting that PI(4,5)P₂ binding may destabilize the R701-Q561 interaction. The mutations Y511A and S512Y resulted in wild-type heat response [80], which is consistent with our finding because the MD-observed HBs between Y511/S512 and E570 involve their backbone amide nitrogen. Other mutations that change the charge of Y511, S512, or E570 will be desirable to test the functional importance of their interactions.

MD simulation of a TRPV1 mutant observes fast gating transition

To directly observe the gating transition of TRPV1 within limited simulation time (e.g. hundreds of ns), we conducted multiple MD simulations of a F640L/T641S double mutant of TRPV1 at 60°C. The F640L mutant is known to be constitutively active and hyper-sensitive to thermal stimulus while preserving an intact gating machinery [81]. Similarly, the T641S mutant also exhibited constitutive channel activation [81]. These mutations likely shift the thermodynamic equilibrium from the C state to the O state without altering the gating transition mechanism [81], making it feasible for MD simulation to directly probe an “accelerated” gating transition. Indeed, we successfully observed channel opening at the lower gate in six mutant MD trajectories, with the pore radius increasing to ~2 Å at the lower gate (see Figure 4(a)). Away from the pore, we observed asymmetric outward motions of the S4-S5 linkers accompanied by inward motions of the MPD linkers and the S2-S3 linkers relative to the closed structure (see Figure 4(b)). These gating motions resemble the heat-activated expansions/contractions of these domains observed in the WT MD simulations.

Figure 4.

The results of TRPV1 mutant simulations. (a) Pore radius at the lower gate (red) and the upper gate (green) in a representative mutant MD trajectory. (b) An open-channel conformation (opaque) superimposed on the closed structure of TRPV1 (transparent), where key domains are colored as follows: MPD linkers (green), S2-S3 linkers (cyan), S4-S5 linkers (purple), and S6 helices (gray). Marker residues of these domains (E405, S510, and A566) and the lower-gate residue I679 are shown as balls and labeled. Motions of these domains are indicated by arrows colored by domain. (c) to (f) show time-dependent shift of R_g averaged over six mutant MD trajectories at eight marker residue positions (E405 of the MPD linker, V469 of the S1-S2 linker, S510 of the S2-S3 linker, A566 of the S4-S5 linker, I679 and N687 of the S6 helix, G643 and E651 of the outer pore). The vertical impulses indicate the standard errors of R_g shift due to averaging over six MD trajectories.

The mutant simulations revealed the following sequence of domain motions during the gating transition: contraction of MPD linker (E405) and expansion of S1-S2 linker (V469) → contraction of S2-S3 linker (S510) → expansion of S4-S5 linker (A566) and S6 (N687) → expansion at lower gate (I679) (see Figure 4(c–f)). This is consistent with our proposed heat-activation pathway connecting the MPD linker to the lower gate via the S2-S3 and S4-S5 linkers (see Figure 1(b)).

To further probe the energetic basis of the temperature-sensitive gating transition in TRPV1 [52], we calculated gating-associated energy changes in the protein-membrane-water system based on the mutant simulations. We focused on those nonbonded energy terms that increase upon the gating transition, which may contribute to the large enthalpy increase (~100 kcal/mol) in TRPV1 gating [52]. Interestingly, we obtained a large increase in the protein-water electrostatic interaction energy (total 5978 ± 509 kcal/mol, reduced to ~70 kcal/mol due to solvent screening), which hints for significant de-solvation of polar/charged residues during the gating transition. Similar energy increase was observed from 30°C to 60/72°C in the WT simulations, supporting its role in heat-activated TRPV1 gating. Owning to such large de-solvation, 111 ± 24 protein-water HBs were lost after the gating transition, which involve polar/charged residues of key domains including the MPD linker and the S4-S5 linker. This de-solvation is extensively distributed over several domains and some of which overlap with the peripheral cavities identified by another study [77]. Additionally, we found an increase in the intra-protein vdW energy (157 ± 64 kcal/mol), which is mainly contributed by the ICD (150 ± 38 kcal/mol) in agreement with our previous simulation of an isolated ICD [64]. In contrast to the above two energy terms, all the other non-bonded energy terms (e.g. protein-water vdW energy, protein-membrane electrostatic/vdW energy, and intra-protein electrostatic energy) decrease after the gating transition, so they do not make positive contributions to the observed enthalpy increase.

Long MD simulation of WT TRPV1 observes gating transition

We are in the process of running multi-microseconds MD simulations of WT TRPV1 at 60^◦ using Anton – a special purpose supercomputer for biomolecular simulation (https://www.psc.edu/resources/computing/anton), aiming to observe the heat-activated gating transition in WT TRPV1. Encouragingly, in one trajectory, we observed lower gate opening around 2 μs (see Figure 5) which is accompanied by the MPD linker closing and the S4-S5 linker opening similar to the mutant simulation (see Figure 4). Surprisingly, the S2-S3 linker also opens which could result from its coupled motion with the adjacent S4-S5 linker. Further analysis is ongoing to reveal detailed similarities and differences between the gating transitions of WT vs mutant channel.

Figure 5.

Preliminary result of long MD simulation of WT TRPV1 by Anton: panels (a) to (d) show time-dependent R_g for a 3-μs MD trajectory at four marker residue positions (E405 of the MPD linker, S510 of the S2-S3 linker, A566 of the S4-S5 linker, and I679 of the lower gate in S6). The red arrows indicate a gating transition event.

Funding Statement

This work was supported by the American Heart Association [17GRNT33690009].

Acknowledgments

We thank funding from American Heart Association (17GRNT33690009). Computational support was provided by the Center for Computational Research at the University at Buffalo.

Disclosure statement

No potential conflict of interest was reported by the authors.