Structural insights into the gating mechanisms of TRPV Channels

Abstract

Transient Receptor Potential channels from the vanilloid subfamily (TRPV) are a group of cation channels modulated by a variety of endogenous stimuli as well as a range of natural and synthetic compounds. Their roles in human health make them of keen interest, particularly from a pharmacological perspective. However, despite this interest, the complexity of these channels has made it difficult to obtain high resolution structures until recently. With the cryo-EM resolution revolution, TRPV channel structural biology has blossomed to produce dozens of structures, covering every TRPV family member and a variety of approaches to examining channel modulation. Here, we review all currently available TRPV structures and the mechanistic insights into gating that they reveal.

1.1. Introduction

Transient receptor potential (TRP) ion channels are a diverse superfamily of integral membrane proteins which generally function as non-selective cation channels[1]. Six TRP channel subfamilies have been reported in mammals based on sequence homology: TRPC1-7 (canonical), TRPV1-6 (vanilloid), TRPM1-8 (melastatin), TRPA1 (ankyrin), TRPP1-3 (polycystin) and TRPML1-3 (mucolipin)[2, 3]. TRP channels are expressed abundantly in various excitable and non-excitable cell types in the human body and play an important role in human physiology[1, 2].

The founding member of the TRPV family, TRPV1, was first identified in 1997 in neurons based on its sensitivity to both heat and capsaicin[4], with the other five family members identified over the next few years[5–10]. Of the six family members, TRPV1-4 are further classified as thermosensitive (thermoTRPV channels) and are fairly well conserved (40-50% sequence identity)[11]. TRPV5 and TRPV6 are highly homologous to each other (75% sequence identity), are not sensitive to temperature, and are divergent from the thermoTRPVs in sequence conservation (~30% sequence identity)[12, 13].

Although initial work on the thermoTRPV channels determined that they are non-selective cation channels activated by temperatures ranging from innocuous warmth (<37°C) to noxious heat (>50°C) in heterologous systems[4–8, 10], subsequent studies have suggested that the primary function of these channels in humans may not be thermosensation[14, 15]. In the last 20 years, it has been shown that thermoTRPVs are expressed in many tissue types throughout the body and play an important role in neuropathic pain, inflammation, immunity, neuronal development, diabetes, cardiovascular disease and cancer[2, 16].

The remaining TRPV family members, TRPV5 and TRPV6, are highly selective Ca²⁺ channels expressed on the apical membrane of epithelial cells[12, 13]. They play key roles in calcium homeostasis in the kidney (TRPV5) or the colon, prostate, pancreas and other epithelial tissues (TRPV6)[12, 13]. Consequently, TRPV5 and TRPV6 dysfunction is implicated in osteoporosis and the formation of kidney stones[13, 17].

1.2. TRPV channel modulation

TRPV channel activity can be modulated by protein-protein interactions and a variety of ligands, including endogenous lipids and both natural and synthetic small molecules. Protein-protein interactions can range in function from direct modulation of channel activity to channel trafficking to and from the cell membrane. One well characterized protein-protein interaction that directly alters channel activity is calmodulin-based channel inactivation, which rapidly reduces ion permeation across the TRPV family in the presence of high intracellular Ca²⁺[18–20]. CaM has been shown to bind to the usually unstructured C-termini of TRPV channels, and several studies with NMR and crystallography have examined the interaction of C-terminal TRPV peptides with CaM[20–24]. Some protein interactions can fulfil both of these activities, for example Pacsin3 is thought to both regulate TRPV4 endocytosis and also directly alter channel activity by interaction with the unstructured N-terminus, recently examined by NMR[25–27].

As sensitive machines dwelling in the lipid bilayer TRPV channels are inevitably regulated by endogenous lipids, particularly the phosphatidylinositiols. Phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂) has been well established as an activator of both TRPV5 and TRPV6, but has a more contentious role in modulating the thermoTRPVs[19, 28–31]. As is frequently true in the TRPV family, lipid modulation has been chiefly studied in TRPV1. Early examination of the role of lipids in regulating TRPV1 suggested that PI(4,5)P₂ could function both to inhibit and activate the channel, while subsequent work showed only an activating role for PI(4,5)P₂, along with a few other phosphatidylinositols[32–36]. Studies of lipid modulation of the other thermoTRPVs are sparse, but PI(4,5)P₂ hydrolysis after calcium influx has also been linked to TRPV2 desensitization[37]. PI(4,5)P₂ seems to play an inhibitory role in TRPV3, where PI(4,5)P₂ hydrolysis has been linked to enhanced TRPV3 activity[38]. On the other hand, PI(4,5)P₂ has been reported to have both inhibitory and activating roles for TRPV4[27, 39–41]. Apart from phosphatidylinositols, cholesterol has been reported to be important for both TRPV1 and TRPV3 activity. In neuronal cells where TRPV1 is highly expressed natively, TRPV1 is found to localize in cholesterol-rich lipid rafts[42–45], and cholesterol has also been shown to directly bind to the channel and modulate the TRPV1 response to drugs and temperature[46–48]. Cholesterol has been shown to increase the sensitivity of TRPV3 to agonists and temperature[49].

Activation of TRPV1 by vanilloids has been well characterized[4], but the other TRPV channels are insensitive to vanilloids and much less is known about their pharmacological modulators[2]. 2-Aminoethoxydiphenyl borate (2-APB) has been shown to activate TRPV1-3[50], inhibit TRPV6, but have no effect on TRPV4 and TRPV5[51, 52]. Another modulator with well characterized effects is cannabidiol (CBD), one of several cannabinoids which activate TRPV1-4[53–56]. The anti-mycotic agent econazole has been shown to inhibit TRPV5 and TRPV6 channels[57]. It has also been shown that TRPV1 is prone to ligand and heat-based desensitization[4, 58, 59]. Another level of complexity is seen for TRPV2, TRPV3 and TRPV4, which require sensitization to temperature and certain ligands through repeated or prolonged exposure to achieve maximal activity[6, 10, 60–65].

1.3. Structural Studies of TRPV Channels

Despite interest in these channels, structural determination was initially limited by the difficulty in producing the channel in adequate quantities for X-ray crystallography[66]. As a result, the earliest structural information obtained for the TRPV subfamily were X-ray crystal structures of the isolated cytoplasmic ankyrin repeat domain from TRPV2 in 2006, soon followed by similar structures for the other TRPV channels[40, 67–73]. The reduced sample requirement for cryo electron microscopy (cryo-EM) made it possible to obtain the first reconstruction of TRPV1 with a full-length construct in 2008, though this structure was limited to low resolution due to the technology available at the time[74]. The cryo-EM resolution revolution led to the determination of the first high resolution TRPV1 channel structure in 2013 on a truncated construct[75, 76]. In the following years, high resolution structures using both cryo-EM and X-ray crystallography have been determined for all of the TRPV channels, the majority of which have been released since 2018 (Table 1). The accumulation of so many structures in such a short time frame makes this an exciting time to review these structures and the mechanistic insights they reveal. As a caveat, please remember that the overall 2.8-5.2Å resolution range seen for TRPV channels is not sufficient to accurately place every sidechain – even if they have been built into the model. Therefore, it is always good practice to examine the map in combination with the model built into it.

Table 1.

TRPV channel structures.

	PDB ID	Description	Ligand	Reconstitution	Method/Resolution (Å)	EMDB	Reference	Comments
TRPV1	3J5P	rTRPV1mi	-	Amphipols	cryo-EM / 3.28	5778	[75]	Closed
	3J5Q	rTRPV1mi	DkTx+RTX	Amphipols	cryo-EM / 3.8	5776	[76]	Open
	3J5R	rTRPV1mi	Capsaicin	Amphipols	cryo-EM / 4.2	5777		Closed, Lower gate open
	5IRX	rTRPV1mi	DkTx+RTX	Nanodisc	cryo-EM / 2.95	8117	[78]	Open
	5IRZ	rTRPV1mi	-	Nanodisc	cryo-EM / 3.28	8118		Closed
	5IS0	rTRPV1mi	Capsazepine	Nanodisc	cryo-EM / 3.43	8119		Closed
TRPV2	5AN8	rbTRPV2mi	-	Amphipols	cryo-EM / 3.8	6455	[79]	Closed
	5HI9	rTRPV2	-	Detergent	cryo-EM / 4.4	6580	[80]	Closed
	6BWJ	rbTRPV2cryst RTX	RTX+Ca2+	Detergent	X-ray / 3.1	-	[81]	Closed, Selectivity filter open, C2 Symmetry
	6BWM	rbTRPV2cryst	Ca2+	Detergent	X-ray / 3.9	-		Closed, Selectivity filter possibly open, C2 Symmetry
	6BO4	rTRPV2mi	-	Detergent	cryo-EM / 4.0	7118	[84]	Open, Pore turrets resolved
	6BO5	rTRPV2miΔTurret	-	Detergent	cryo-EM / 3.6	7119		Closed, Lower gate open
	6OO3	rbTRPV2RTX	RTX	Amphipols	cryo-EM / 2.9	20143	[82]	Closed
	6OO4	rbTRPV2RTX	RTX	Amphipols	cryo-EM / 3.3	20145		Closed, C2 Symmetry
	6OO5	rbTRPV2RTX	RTX	Amphipols	cryo-EM / 4.2	20146		Closed, C2 Symmetry
	6OO7	rbTRPV2RTX	RTX	Nanodisc	cryo-EM / 3.8	20148		Closed, Selectivity filter open, C2 Symmetry
	6U84	rTRPV2	-	Nanodisc	cryo-EM / 3.7	20677	[83]	Closed, State 1
	6U86	rTRPV2	-	Nanodisc	cryo-EM / 4.0	20678		Closed, State 2, Selectivity filter open
	6U8A	rTRPV2	CBD	Nanodisc	cryo-EM / 3.2	20686		Closed, State 1
	6U88	rTRPV2	CBD	Nanodisc	cryo-EM / 3.4	20682		Closed, State 2
TRPV3	6DVW	mTRPV3	-	Detergent	cryo-EM / 4.3	8919	[86]	Closed
	6DVY	mTRPV3	2-APB	Detergent	cryo-EM / 4.0	8920		Closed
	6DVZ	mTRPV3Y564A	2-APB	Detergent	cryo-EM / 4.24	8921		Open
	6MHO	hTRPV3T96A	-	Amphipols	cryo-EM / 3.4	9115	[85]	Closed
	6MHS	hTRPV3T96A	-	Amphipols	cryo-EM / 3.2	9117		Closed, Sensitized
	6MHV	hTRPV3T96A	2-APB	Amphipols	cryo-EM / 3.5	9119		Closed, Sensitized
	6MHW	hTRPV3T96A	2-APB	Amphipols	cryo-EM / 4.0	9120		Closed, C2 Symmetry
	6MHX	hTRPV3T96A	2-APB	Amphipols	cryo-EM / 4.0	9121		Closed, C2 Symmetry
	6OT2	hTRPV3T96A, K169A	-	Amphipols	cryo-EM / 4.1	20192	[87]	Closed, Sensitized mutant
	6OT5	hTRPV3T96A, K169A	2-APB	Amphipols	cryo-EM / 3.6	20194		Closed, Sensitized mutant
	6PVL	mTRPV3	-	Detergent	cryo-EM / 4.4	20492	[88]	Closed, 42°C
	6PVM	mTRPV3	-	Detergent	cryo-EM / 4.5	20493		Closed, Sensitized, 42°C
	6PVN	mTRPV3Y564A	-	Detergent	cryo-EM / 4.07	20494		Closed, Sensitized, 4°C
	6PVO	mTRPV3Y564A	-	Detergent	cryo-EM / 5.18	20495		Closed, Sensitized, 37°C
	6PVP	mTRPV3Y564A	-	Detergent	cryo-EM / 4.48	20496		Open, 37°C
	6PVQ	mTRPV3Y564A	-	Detergent	cryo-EM / 4.75	20497		Closed, Intermediate state, 37°C, C2 Symmetry
TRPV4	6BBJ	xTRPV4miN647Q	-	Detergent	cryo-EM / 3.8	7075	[89]	Closed
	6C8F	xTRPV4miN647Q	Cs+	Detergent	X-ray / 6.5	-		Closed
	6C8G	xTRPV4miN647Q	Ba2+	Detergent	X-ray / 6.31	-		Closed
	6C8H	xTRPV4miN647Q	Gd3+	Detergent	X-ray / 6.5	-		Closed
TRPV5	6B5V	rbTRPV5	ECN	Detergent	cryo-EM / 4.8	7058	[57]	Closed
	6DMR	rbTRPV5	-	Detergent	cryo-EM / 3.9	7965	[19]	Closed
	6DMU	rbTRPV5	PI(4,5)P2	Nanodisc	cryo-EM / 4.0	7966		Open
	6DMW	rbTRPV5	CaM	Detergent	cryo-EM / 4.4	7967		Closed, C1 Symmetry
	6O1N	rbTRPV5Δ70	-	Nanodisc	cryo-EM / 2.9	0593	[17]	Closed
	6O1P	rbTRPV5	-	Nanodisc	cryo-EM / 3.0	0594		Closed
	6O1U	rbTRPV5W583A	-	Nanodisc	cryo-EM / 2.8	0605		Open
	6O20	rbTRPV5	CaM	Detergent	cryo-EM / 3.3	0607		Closed, C1 Symmetry
	6PBF	rbTRPV5	ZINC9155420	Nanodisc	cryo-EM / 4.2	20292	[90]	Closed
	6PBE	rbTRPV5	ZINC17988990	Nanodisc	cryo-EM / 3.78	20291		Closed
TRPV6	5IWK	rTRPV6cryst	-	Detergent	X-ray / 3.25	-	[91]	Closed
	5IWP	rTRPV6cryst	Ba2+	Detergent	X-ray / 3.65	-		Closed, No domain swap
	5IWR	rTRPV6cryst	Ca2+	Detergent	X-ray / 3.85	-		Closed, No domain swap
	5IWT	rTRPV6cryst	Gd3+	Detergent	X-ray / 3.8	-		Closed, No domain swap
	5WO6	rTRPV6cryst	-	Detergent	X-ray / 3.31	-	[92]	Closed, No domain swap
	5WO7	rTRPV6*	-	Detergent	X-ray / 3.25	-		Closed
	5WO8	rTRPV6*Δ477-480	-	Detergent	X-ray / 3.4	-		Closed, No domain swap
	5WO9	rTRPV6*	Ca2+	Detergent	X-ray / 3.7	-		Closed
	5WOA	rTRPV6*	Gd3+	Detergent	X-ray / 3.9	-		Closed
	6BO8	hTRPV6	-	Nanodisc	cryo-EM / 3.6	7120	[93]	Open
	6BO9	hTRPV6	-	Amphipols	cryo-EM / 4.0	7121		Open
	6BOA	hTRPV6R470E	-	Amphipols	cryo-EM / 4.2	7122		Closed
	6BOB	rTRPV6*	-	Nanodisc	cryo-EM / 3.9	7123		Closed
	6D7O	rTRPV6*	2-APB	Detergent	X-ray / 3.45	-	[51]	Closed
	6D7P	rTRPV6*Y466A	-	Detergent	X-ray / 3.37	-		Closed
	6D7Q	rTRPV6*Y466A	2-APB	Detergent	X-ray / 3.5	-		Closed
	6D7S	hTRPV6Y467A	-	Amphipols	cryo-EM / 4.34	7824		Open
	6D7T	hTRPV6Y467A	2-APB	Amphipols	cryo-EM / 4.44	7825		Closed
	6D7V	rTRPV6*	2-APB-Br	Detergent	X-ray / 4.3	-		Closed
	6D7X	rTRPV6*Y466A	2-APB-Br	Detergent	X-ray / 3.6	-		Closed
	6E2F	hTRPV6	CaM	Amphipols	cryo-EM / 3.9	8961	[18]	Closed, C1 Symmetry
	6E2G	rTRPV6	CaM	Detergent	cryo-EM / 3.6	8962		Closed, C1 Symmetry

1.4. General TRPV channel architecture

As is true for all TRP channels, TRPV channels are tetrameric and each monomer features the classic 6 transmembrane helix (S1-S6) architecture of voltage gated ion channels in its transmembrane domain (TMD) (Fig. 1a,b)[77]. S1-S4 form a bundle similar to the voltage sensing domain in voltage sensitive ion channels, earning it the name voltage sensing-like domain (VSLD) in the relatively voltage insensitive TRPV channels. S5 and S6 extend away from the S1-S4 bundle to interact with the S1-S4 bundle of a neighboring monomer and form the tetrameric pore (Fig. 1a, b). This domain-swapped pore domain consists of S5 and S6 forming the central pore and lower gate, while a short loop and helix between S5 and S6, termed the pore helix (PH), forms the upper gate or selectivity filter (Fig. 1c). TRPV1, TRPV2, and TRPV4 have an extended unstructured loop of 15-25 residues between S5 and the PH, named the pore turret.

Figure 1.

Domain architecture and 3-dimensional organization of TRPV channels. (a) Cartoon representation of a TRPV monomer. The N-terminal domain (NTD) is depicted in grey, the ankyrin repeat domain (ARD) in yellow, the N-linker in orange, the pre-S1 helix in pink, the S1-S4 helices in magenta, the S4-S5 linker in purple, S5-PH-S6 in purple, the TRP helix in blue, and the C-terminal domain (CTD) in green. (b) The same domains colored on a monomer of a representative TRPV channel (TRPV5, PDB 6DMR). (c) The tetrameric assembly of a representative TRPV channel (TRPV5, PDB 6DMR) from the side (left) and from the extracellular face (right), with helices depicted as tubes.

A unique and essential feature of TRP channels is the TRP domain immediately C-terminal to S6, consisting of a helix running parallel to the membrane and wedged intimately into the intracellular side of the S1-S4 bundle (Fig. 1a, b). A short helical linker between S4 and S5 (S4-S5 linker) interacts extensively with the TRP domain, the S1-S4 bundle, and the pore domain from an adjacent monomer, which puts it in a key position for regulating channel gating (Fig. 1a, b).

The defining structural feature of the TRPV subfamily is that each monomer of all six family members have six N-terminal ankyrin repeats (AR) on their cytoplasmic face, named the ankyrin repeat domain (ARD) (Fig. 1a, b, c). The ARDs oligomerize with the β-sheet region of neighboring monomers, creating a skirt-like enclosure on the intracellular face of the channel essential for regulatory protein-protein interactions (Fig. 1a, b, c). The β-sheet region consists of 1 β-sheet from the C-terminus and 2 β-sheets from the N-linker region between the ARD and S1. The N-linker also features a helix-turn-helix motif and a short helix immediately before S1 (pre-S1) (Fig. 1a, b).

2.1. TRPV1

The first high-resolution structures of truncated TRPV1 were obtained by the Julius & Cheng groups in 2013 by cryo-EM[75, 76]. In order to obtain high-resolution structures, the authors created a minimum functional rat TRPV1 construct (rTRPV1_mi) by removing 109 residues of the N-terminus, 24 residues of the unstructured pore turret between S5 and the PH, and 74 residues of the C-terminus of the channel[75, 76].

The structure of rTRPV1_mi stabilized in amphipols was resolved in an apo state (PDB 3J5P) as well as in two different ligand bound states: with capsaicin (PDB 3J5R) or with a combination of resiniferatoxin (RTX) and double-knot toxin (DkTx) (PDB 3J5Q)[75, 76]. These structures revealed the assembled architecture of the pore for the first time, and the addition of ligand activators allowed the authors to trap the channel in three different functional states. They found that the vanilloids, capsaicin and RTX, both bound in a pocket located above the S4-S5 linker and between S3 and S4 of one monomer and S6 of an adjacent monomer, earning it the name ‘vanilloid binding pocket’. The activator DkTx is a 75 amino acid long peptide which forms two almost identical globular domains joined by a short linker. A single globular domain was observed to bind at the junction between monomers at the PH and S6 on the extracellular face of the channel, for a ratio of two molecules of DkTx to one tetramer of TRPV1.

The apo structure revealed two gates in the ion permeation pathway: the selectivity filter as the upper gate (Gly643-Met644-Gly645) and a hydrophobic seal at Ile679 as the lower gate, both of which were closed (Fig. 2a). Capsaicin binding to TRPV1 left the selectivity filter closed, but partially opened the lower gate. The combination of RTX and DkTx opened both gates (Fig. 2b), which the authors attribute to DkTx wedging open the pore which would otherwise be expected to flicker between open and closed states. The density for the ARDs in these structures was of poor resolution and incomplete, so the authors fit the TRPV1 ARD crystal structure model (PDB 2PNN) into this part of their maps to assemble the complete structures[69].

Figure 2.

Pore diagrams of TRPV1-4. (a) rTRPV1_mi closed (PDB 5IRZ). (b) rTRPV1_mi open (PDB 5IRX). (c) rbTRPV2_mi closed (PDB 5AN8). (d) rTRPV2 semi-open (PDB 6U86). (e) mTRPV3 closed (PDB 6DVW). (f) mTRPV3_Y564A open with 2-APB (PDB 6DVZ). (g) hTRPV3_T96A sensitized (PDB 6MHS). (h) xTRPV4_miN647Q closed (PDB 6BBJ). Residues of interest depicted as sticks.

In 2016, the authors followed up with new structures of the same rTRPV1_mi construct, this time reconstituted in the more natural environment of nanodiscs[78]. They obtained three new high-resolution structures: apo (PDB 5IRZ), with RTX and DkTx (PDB 5IRX), and with the antagonist capsazepine (PDB 5IS0), which they found also binds in the vanilloid binding pocket[78]. While the apo and RTX/DkTx structures were essentially identical to the structures previously obtained stabilized in amphipols, the higher resolution and lipid environment made it possible to resolve several densities for lipids interacting with the channel. The most notable of these was observed in the apo state, where a lipid headgroup was wedged between S3 and the S4-S5 linker, with the tails of the lipid filling the vanilloid binding pocket. The authors identified the lipid in the vanilloid binding pocket as phosphatidylinositol (PI) and proposed a mechanism of TRPV1 channel activation by vanilloids where ligands would displace PI, allowing the S4–S5 linker to move away from the pore to accommodate the opening of the channel.

2.2. TRPV2

Since 2016, several TRPV2 structures have been independently obtained by the Lee group in collaboration with the Lander group and the Moiseenkova-Bell and Wensel groups[79–84]. Seven different constructs have been used for these structural studies: 1) a minimal truncated rabbit TRPV2 construct, which is missing 55 residues of the N-terminus, 21 residues of the pore turret between S5 and the PH, and 40 residues of the C-terminus of the channel (rbTRPV2_mi)[79]; 2) a minimal truncated rabbit TRPV2 construct optimized for crystallography, which is missing 56 residues of the N-terminus, 22 residues of the pore turret, and 64 residues of the C-terminus (rbTRPV2_cryst)[81]; 3) a minimal truncated rabbit TRPV2 optimized for crystallography with four additional mutations in the vanilloid binding pocket to acquire sensitivity for RTX (rbTRPV2_crystRTX)[81]; 4) full-length rabbit TRPV2 with four mutations in the vanilloid binding pocket to acquire sensitivity for RTX (rbTRPV2_RTX)[82]; 5) truncated rat TRPV2 construct (rTRPV2_mi), which is missing 73 residues of the N-terminus and 35 residues of the C-terminus but with the first 31 residues of the yeast TRPY1 channel added to the N-terminus to improve expression[84]; 6) truncated rat TRPV2 with an additional 25 residues of the pore turret between S5 and the PH excised (rTRPV2_miΔTurret)[84]; and 7) full-length rat TRPV2 (rTRPV2)[80, 83].

In 2016, cryo-EM structures of rbTRPV2_mi stabilized in amphipols (PDB 5AN8) from the Lee & Lander groups and rTRPV2 stabilized in detergent (PDB 5HI9) from the Moiseenkova-Bell group revealed the channel in an apo state and with well resolved ARDs[79, 80]. Like the apo rTRPV1_mi channel, the apo rbTRPV2_mi structure was closed at both the selectivity filter and lower gate. However, a major deviation from the rTRPV1_mi structure was observed: S6 of rbTRPV2_mi lacked a π-helix turn near the top of the S6 helix, and so the residue in TRPV2 equivalent to Ile679 in TRPV1 was not oriented into the pore (Fig. 2c). Instead, Met643 formed a hydrophobic seal at the lower gate of the channel (Fig. 2c). The structure of rTRPV2 was determined at ~5 Å resolution and had a wider selectivity filter (formed by Gly606-Met607-Gly608) and lower gate compared with the rTRPV1_mi structures but had the same S6 orientation as the rbTRPV2_mi structure. Although the rTRPV2 construct was full-length, the N-terminus, C-terminus, and pore turret region were not resolved in the structure.

The Lee group next obtained structures of rbTRPV2_crystRTX with 2-APB (PDB 6BWJ) and of rbTRPV2_cryst without 2-APB (6BWM) stabilized in detergent and in the presence of Ca²⁺ by X-ray crystallography, followed by cryo-EM studies of rbTRPV2_RTX in the presence of RTX stabilized in both amphipols and nanodiscs (PDB 6OO3, 6OO4, 6OO5, 6OO7)[81, 82]. Based on these structural studies, the authors proposed that TRPV2 channels could adopt several intermediate two-fold symmetric (C2) semi-open states upon RTX binding. While an open TRPV2 state was not resolved in these studies, the authors suggested that the channel would come back to the four-fold symmetry (C4) in the fully open state[82]. These studies also identified a π-helix at the junction between the S4-S5 linker and S5 and observed that it acts as a hinge, allowing the S4-S5 linker and S5 to move relative to one another and accommodate the changes in symmetry which the authors predict are required for channel opening[81, 82].

Structures of rTRPV2_mi (PDB 6BO4) and rTRPV2_miΔTurret (PDB 6BO5) stabilized in detergent were determined by the Wensel group by cryo-EM[84]. While they proposed a mechanism of channel opening and suggested that the TRPV2 pore turrets lay parallel to the lipid membrane, the quality of the provided cryo-EM data did not allow independent confirmation of the channel conformations. While structures of full-length TRPV2 constructs obtained by the Lee or Moiseenkova-Bell groups were not able to resolve the full pore turrets[80, 82, 83], they were able to show that they are oriented perpendicular to the membrane regardless of whether the channel was stabilized in detergent, amphipols or nanodiscs.

Structures of rTRPV2 in apo and CBD-bound states stabilized in nanodiscs were recently determined by the Moiseenkova-Bell group by cryo-EM[83]. The apo rTRPV2 dataset yielded two distinct structures – state one, closed at both the selectivity filter and the lower gate of the channel (PDB 6U84) and state two with an open selectivity filter (PDB 6U86) (Fig. 2d). The structure of apo rTRPV2 state one (PDB 6U84) is quite different from both the initial apo rbTRPV2_mi structure stabilized in amphipol (PDB 5AN8) and the rTRPV2 structure stabilized in detergent (PDB 5HI9). In this structure, a break in the helix between the S4-S5 linker and S5 produces an overall outward shift of the S1-S4 bundle as well as the ARDs. Additionally, the C-terminal domain (CTD) ends in a short helix and a 16 residue segment (residues Gln30-Asn45) of the N-terminal domain (NTD) binds to a portion of the ARDs and the β-sheet region of the adjacent monomer. The opening of the selectivity filter in apo rTRPV2 state two (PDB 6U86) seems to be permitted by the same break in the helix between the S4-S5 linker and S5, as the pore domain from each monomer rotated slightly away from the pore.

The CBD-bound rTRPV2 dataset also produced two distinct structures, each with strong density for CBD in a novel ligand binding pocket located between S5 and S6 of adjacent TRPV2 monomers[83]. While CBD densities were well resolved in these structures, CBD binding did not trap an open conformation of the channel and both gates stayed closed in these structures. CBD-bound rTRPV2 state one (PDB 6U8A) resembled apo rTRPV2 state one (PDB 6U84), while CBD-bound rTRPV2 state two (PDB 6U88) resembled the initial apo rbTRPV2_mi stabilized in amphipol structure (PDB 5AN8)[79]. Apart from the 16 residue segment of the NTD resolved, the remaining residues of the N-terminus are not seen, along with the pore turrets and the C-terminus.

2.3. TRPV3

Several TRPV3 cryo-EM studies were presented independently by the Sobolevsky group and the Lee group in collaboration with the Lander group starting in 2018[85–88]. The Sobolevsky group started by obtaining a cryo-EM structure of full-length mouse TRPV3 (mTRPV3) stabilized in detergent (PDB 6DVW)[86]. As in TRPV2, TRPV3 was well resolved throughout the TMD and in the ARDs, but the N- and C- termini could not be resolved. They observed the apo state to be closed at the lower gate by Met677, equivalent in position to the Met643 that gates TRPV2, though with a wider selectivity filter than TRPV2 formed by Gly638-Leu639-Gly640 (Fig. 3e). The structure of mTRPV3 purified in the presence of the activator 2-APB (PDB 6DVY) was identical to the apo mTRPV3 structure (PDB 6DVW) and the 2-APB binding site could not be identified[86]. The introduction of a Tyr564Ala mutation in S4 increased the efficacy of 2-APB by 20-fold in mTRPV3, which allowed the authors to capture an open state of the full-length mouse TRPV3-Tyr564Ala mutant channel (mTRPV3_Y564A) in the presence of 2-APB (PDB 6DVZ) (Fig. 2f). The authors suggested that TRPV3 has three 2-APB binding sites in this mutant channel structure (PDB 6DVZ): 1) in a pocket above the TRP helix between S1 and S2 and adjacent to the Tyr564Ala mutation; 2) in a pocket below the TRP helix and adjacent to the pre-S1 helix; and 3) in a pocket on the extracellular face of the channel wedged into the S1-S4 bundle. The opening of the lower gate in mTRPV3_Y564A required the formation of a π-helix in the upper portion of S6, causing Met677 to rotate out of the path of the ion permeation pathway and the lower portion of S6 to tilt away from the central axis of the pore (Fig. 2f). This rotation was accompanied by an extension of the S6 helix by two turns and a shortening of the TRP helix by two turns.

Figure 3.

Pore diagrams of TRPV5-6. (a) rbTRPV5 closed (PDB 6DMR). (b) rbTRPV5 open (PDB 6DMU). (c) rTRPV6* closed (PDB 5WO7). (d) rTRPV6 open (PDB 6BOB). Residues of interest depicted as sticks.

The Lee & Lander groups used full-length human TRPV3 with a Thr96Ala mutation (hTRPV3_T96A) stabilized in amphipols to look at 2-APB interaction with the channel by cryo-EM[85]. The hTRPV3_T96A structure in an apo state (PDB 6MHO) was very similar to that obtained by the Sobolevsky group for apo mTRPV3 (PDB 6DVW)[86]. They next exposed hTRPV3_T96A to 2-APB repeatedly during purification to sensitize the channel and obtained a sensitized hTRPV3_T96A structure (PDB 6MHS). This structure did feature a π-helix in the S6 helix, but the lower gate was now closed by Ile674 (Fig. 2g). Next, they determined the structure of sensitized hTRPV3_T96A in the presence of 2-APB (PDB 6MHV, 6MHW, 6MHX). While they were not able to identify a 2-APB binding site, they identified three states of the channel that largely resembled apo hTRPV3_T96A . One state had C4 symmetry (PDB 6MHV), while the others had C2 symmetry. (PDB 6MHW, 6MHX).

Both studies observed that the CTD of TRPV3 in the apo states wraps around the β-sheet region, making contacts with the ARD of an adjacent monomer[85, 86]. In follow-up work, the Lee group probed this interaction based on a study showing that a Lys169Ala mutation[65], part of the CTD-ARD interface, relieved the need for sensitization of TRPV3 by both heat and ligands[87]. They obtained two cryo-EM structures of full-length human TRPV3 with mutations Thr96Ala and Lys169Ala (hTRPV3_T96A,K169A) in the presence and absence of 2-APB. The structure of apo hTRPV3_T96A,K169A (PDB 6OT2) diverged significantly from the apo structure previously obtained for hTRPV3_T96A, with shifts throughout the S1-S4 bundle, TRP helix and ARDs. The authors observed that the CTD of each monomer in the apo hTRPV3_T96A,K169A structure was no longer wrapped around the β-sheet region, but instead formed a helix on the inside of the ARD skirt. They also found additional density in the region the CTD had vacated that was weakly connected to the most N-terminal portion of the ARDs, which they assigned to an unidentified portion of the NTD. On the addition of 2-APB to hTRPV3_T96A,K169A (PDB 6OT5), the channel remained in the NTD switched position and they observed an α- to π-helical transition in S6 as they had previously seen in the hTRPV2_T96A sensitized state (PDB 6MHS), though this movement was still not sufficient to fully open the pore. They identified a single binding site for 2-APB per monomer in one of the sites identified previously by the Sobolevsky group: site 3 located beneath the TRP helix. Although the Sobolevsky group had not observed the CTD/NTD switch in their open 2-APB-bound structure[86], examination of that map in light of the Lee group’s results reveals density consistent with the CTD/NTD switch.

Most recently, the Sobolevsky group examined temperature-based activation of mTRPV3[88]. They found that at 42°C mTRPV3 had an increased probability of opening, so they exposed mTRPV3 to 42°C briefly and produced a dataset which yielded two states. The first state (PDB 6PVL) was identical to the apo structure of mTRPV3 (PDB 6DVW) (Fig. 2e) and the second state (PDB 6PVM) resembled the sensitized state the Lee group obtained for hTRPV3_T96A (PDB 6MHS) (Fig. 2g). They observed that the Tyr564Ala mutation, which they had previously used to increase the efficacy of 2-APB, also significantly increased the open probability of the channel while reducing temperature sensitivity, so the authors used mTRPV3_Y564A to obtain structures at 4°C (PDB 6PVN) and 37°C (PDB 6PVO, 6PVP, 6PVQ). At 4°C mTRPV3_Y564A (PDB 6PVN) resembled the sensitized states of mTRPV3 at 42°C (PDB 6PVM) and hTRPV3_T96A (PDB 6MHS). The dataset for mTRPV3_Y564A at 37°C yielded three distinct states. State one was identical to the sensitized state for mTRPV3_Y564A at 4°C (PDB 6PVO) (Fig. 2g) and state two (PDB 6PVP) was identical to the open state previously determined of 2-APB-bound mTRPV3_Y564A (PDB 6DVZ) (Fig. 2f). In this second state they identified the CTD/NTD switch previously described by the Lee group[87], which they assigned to the 30 residues immediately before the start of the ARD region. The third state was labelled as an intermediate (PDB 6PVQ) as it had two monomers with the CTD wrapped around the β-sheet region and the other two monomers with the NTD. This structure deviated from the four-fold symmetry observed in all other structures in this study, exhibiting very strong C2 symmetry, but with poorly resolved pore domains.

2.4. TRPV4

In 2018 structures of the truncated Xenopus tropicalis TRPV4 channel (xTRPV4_miN647Q) were determined by the Hite & Yuan groups using cryo-EM and X-ray crystallography[89]. This TRPV4 construct was truncated by 133 residues at the N-terminus and by 71 residues at the C-terminus, had an Asn647Gln point mutation, and displayed poor channel activity. A structure of apo xTRPV4_miN647Q was determined in detergent and revealed intriguing architectural divergences from other TRPV channels. Compared to the other TRPV channels, the S1-S4 bundle of xTRPV4_miN647Q displayed an unusual conformation, with S2 moved away from S1 towards the vanilloid binding pocket, causing S3 and S4 to swivel away from S2 and around S1. This resulted in a straight helix being formed by the joined S4-S5 linker and S5, with the S4-S5 linker making minimal contacts with the TRP helix. Another consequence of the movement of S2 was a distortion of the vanilloid binding pocket that has been observed for all other TRPV channels. In the apo state, the xTRPV4_miN647Q selectivity filter was open (formed by Gly675-Met676-Gly677), but the lower gate was sealed by Met714, in the same position as in TRPV2 (Fig. 2h).

2.5. TRPV5

The Moiseenkova-Bell group contributed the first structure of the TRPV5 channel in 2018[57]. They obtained the structure of a full-length rabbit TRPV5 construct (rbTRPV5) in the presence of the inhibitor econazole and stabilized in detergent (PDB 6B5V) by cryo-EM. This structure was in a closed state with a π-helix in S6 and revealed that the econazole binding site is located in the vanilloid binding pocket of the channel. The selectivity filter is also quite different from that observed for the thermoTRPVs, formed by Thr539-Ile540-Ile541-Asp542 rather than Gly643-Met644-Gly645 as in TRPV1 (Fig 2, Fig. 3). The N- and C-termini of the channel were not resolved, as is usual for TRPV channels. Next, the Moiseenkova-Bell group again used cryo-EM to examine the modulation of rbTRPV5 by lipids and calmodulin (CaM)[19]. The soluble form of the activating lipid PI(4,5)P₂ was added to rbTRPV5 stabilized in either detergent or nanodiscs. The rbTRPV5 structure with PI(4,5)P₂ in detergent (PDB 6DMR) had several non-protein densities likely to be lipids, though none of these densities could be positively identified as PI(4,5)P₂. The structure was captured in the closed state, with a π-helix in the S6 helix and the lower gate formed by Ile575, equivalent in position to Ile679 in the apo rTRPV1_mi structure (PDB 5IRZ) (Fig. 3a). The rbTRPV5 structure obtained in the presence of PI(4,5)P₂ in nanodiscs revealed an open channel (Fig. 3b) and allowed the authors to identify a PI(4,5)P₂ binding site located between the N-linker (Arg302, Arg305), S4-S5 linker (Lys484), and the S6 helix (Arg584) of the channel. The opening of the channel was facilitated by interaction of Arg584 with the 4’ and 5’ phosphates of PI(4,5)P₂, causing it to rotate away from the pore by almost 90° and turning and tilting the lower part of the S6 helix with the π-helix as the pivot point (Fig. 3b). The rbTRPV5 structure bound to rat calmodulin and stabilized in detergent (PDB 6DMW) revealed that binding only one copy of CaM is necessary to inactivate the channel. CaM is composed of two lobes (C-lobe and N-lobe), each of which can bind two Ca²⁺ ions with different affinity as well as a peptide substrate. CaM bound inside the ARD skirt, with the N-lobe bound to a helical extension from the visible C-terminus of one monomer of TRPV5 (residues Leu642-Lys652) and the C-lobe bound to an unconnected helix from the distal C-terminus (residues His699-Thr709). This binding scheme brought the C-lobe towards the base of the pore, where Lys116 from the C-lobe extended into the base of the pore and interacted with a ring of Trp583 resides, physically blocking ion passage.

Similar results were also obtained by the Julius, Cheng, and van der Wijst groups recently using cryo-EM[17]. The authors used rabbit TRPV5, either full-length (rbTRPV5) or a construct truncated at the C-terminus by 70 residues (rbTRPV5_Δ70), stabilized either in detergent or nanodiscs. The apo structures for rbTRPV5 (PDB 6O1P) and rbTRPV5_Δ70 (PDB 6O1N) stabilized in nanodiscs were very similar to the rbTRPV5 structure with PI(4,5)P₂ stabilized in detergent reported by Moiseenkova-Bell group (PDB 6DMR) [19]. To examine the opening of the channel, they obtained the structure of a full-length rabbit TRPV5-Trp583Ala mutant (rbTRPV5_W583A) stabilized in nanodiscs (PDB 6O1U), which strongly resembled the PI(4,5)P₂ bound state obtained by the Moiseenkova-Bell group (PDB 6DMU)[19]. They also obtained a structure of rbTRPV5 stabilized in detergent and bound to bovine CaM (PDB 6O20), with the same binding mode seen previously by the Moiseenkova-Bell group for TRPV5 (PDB 6DMW)[19]. One difference from the Moiseenkova-Bell group work was an observation for a second molecule of CaM binding in the skirt region of the channel, yet only one copy of the C-lobe was able to bind closely to the bottom of the pore.

The Moiseenkova-Bell group in collaboration with the Rohacs and Filizola groups recently used in silico screening followed by functional testing to identify a novel, potent and specific inhibitor of TRPV5, ZINC 17988990[90]. They obtained cryo-EM structures of rbTRPV5 stabilized in nanodiscs in the presence of ZINC 17988990 (PDB 6PBE) and in the presence of another novel, but less specific, inhibitor ZINC 9155420 (PDB 6PBF). The structure of TRPV5 in the presence of the ZINC 1788990 revealed that the binding site for this novel and specific compound is located in the VSLD pocket.

2.6. TRPV6

The first TRPV6 channel structure was presented by the Sobolevsky group in 2016 and was determined using X-ray crystallography[91]. They created a construct of rat TRPV6 optimized for X-ray crystallography with four mutations, including Leu495Gln, and with 59 residues of the C-terminus truncated (rTRPV6_cryst). rTRPV6_cryst stabilized in detergent yielded a structure of the channel without the domain swapped architecture for the pore domain (PDB 5IWK)[91] seen previously in the first structures of TRPV1[75, 76]. The authors soon discovered that one of the mutations they incorporated to increase protein expression was responsible for this aberrant conformation: Leu495Gln located in the S4-S5 linker[91, 92]. On returning it to the native residue, they obtained an apo structure of the fixed rat TRPV6 construct (rTRPV6*) (PDB 5WO7), this time with the expected domain swap architecture[92]. The apo structure of rTRPV6* was closed, with the lower gate sealed by Met577 in a similar position as Met643 in TRPV2 (Fig. 2c, 3c). The selectivity filter is similar to that of TRPV5 and is formed by Thr538-Ile539-Ile540-Asp541. The N- and C-termini of these structures were not resolved.

After obtaining these initial structures with X-ray crystallography, the Sobolevsky group turned to cryo-EM to examine TRPV6 in a more native context. They obtained two structures of the full-length human TRPV6 construct (hTRPV6) stabilized in both amphipols (PDB 6BO9) and nanodiscs (PDB 6BO8) as well as a structure of rTRPV6* stabilized in nanodiscs (PDB 6BOB)[93]. The rTRPV6* structure stabilized in nanodiscs (PDB 6BOB) was almost identical to the structure of the same construct obtained by X-ray crystallography (PDB 5WO7)[92]. Both hTRPV6 structures were open at their lower gates, accomplished by the formation of a π-helix near the top of S6 which rotated the sealing Met577 out of the path of ion permeation (Fig. 3d). This transition also tilted the lower portion of S6 away from the central axis, which was paired with an outward movement of the S4-S5 linker. The open hTRPV6 structures had very strong non-protein density in the vanilloid binding pocket, but this density was almost absent in the closed rTRPV6* structure. The authors attributed this density to an activating lipid, and reasoned that by mutating a key residue of the lipid binding site (Arg470Glu) they could obtain a closed structure of the hTRPV6 structure. The authors obtained a structure of full-length human TRPV6 with an Arg470Glu mutation (hTRPV6_R470E) stabilized in amphipols (PDB 6BOA). This yielded a closed state, almost identical to rTRPV6* and also with poor lipid density in the vanilloid binding pocket.

The Sobolevsky group next examined TRPV6 inhibition by 2-APB and obtained several structures of TRPV6 with 2-APB by both cryo-EM and X-ray crystallography[51]. They used rTRPV6* to obtain a structure of the channel stabilized in detergent and bound to 2-APB using X-ray crystallography (PDB 6D7O) and confirmed the assignment of 2-APB by obtaining a duplicate structure with brominated 2-APB (PDB 6D7V). They found that 2-APB binds in the VSLD pocket where they had previously seen strong lipid density[93]. Since the rat TRPV6* channel is already closed, however, they saw little change in the overall structure. Due to low binding affinity the authors were not able to obtain a structure of hTRPV6 with 2-APB, but they were able to find a single mutation that increased the affinity of 2-APB for hTRPV6: Tyr467Ala. This allowed them to obtain two cryo-EM structures of a full-length human TRPV6-Tyr467Ala construct (hTRPV6_Y467A) with (PDB 6D7T) and without (PDB 6D7S) 2-APB and stabilized in amphipols. hTRPV6_Y467A without 2-APB was open like apo hTRPV6, showing that the channel was still functional. hTRPV6_Y467A with added 2-APB was closed and had density for 2-APB in the same location as observed in the rTRPV6* crystal structures. 2-APB binding caused the S1-S4 bundle to shift slightly towards the central pore, along with the S4-S5 linker and TRP helix. This shifted the density for the activating lipid partially out of the vanilloid binding pocket, which the authors reasoned was responsible for closing the pore.

Recently, the Sobolevsky group examined the interaction between TRPV6 and CaM[18]. They obtained cryo-EM structures of hTRPV6 (PDB 6E2F) and a full-length rat TRPV6 construct (rTRPV6) (PDB 6E2G) bound to human CaM. They observed CaM binding inside the ARD skirt, with the N-lobe bound to a helical extension from the visible C-terminus of one monomer of TRPV6 (residues Arg641 -Arg654) and the C-lobe bound to an unconnected helix from the distal C-terminus (residues Arg686-Arg705). As in TRPV5[19], this binding scheme brought the C-lobe towards the base of the pore with Lys115 interacting with a ring of Trp582 residues at the base of the pore to physically block the channel. Interestingly, the pore was in neither the open nor closed states previously observed but took on a new intermediate conformation, which looked very much like the apo state observed for rbTRPV5[17]. This CaM binding state had π-helices in S6 like the open state, but the lower portions of the S6 helix were drawn towards the center of the pore by cation-π interactions between CaM Lys115 and TRPV6 Trp582.

3. Structural insights into TRPV channel mechanisms

The thermoTRPV channels are non-selective cation channels and have even been observed to allow for the passage of bulky organic cations, while TRPV5 and TRPV6 have a strong preference for calcium ions and have not been implicated in organic cation permeation[80, 81, 94, 95]. The thermoTRPV selectivity filter consensus sequence (Gly-Met-Gly) is very different from the TRPV5/TRPV6 selectivity filter consensus sequence (Thr-Ile-Ile-Asp), which likely contributes to the difference in selectivity. X-ray crystallography studies of TRPV4 from the Yuan and Hite groups and TRPV6 from the Sobolevsky group in the presence of anomalously scattering cations explored this difference in selectivity[89, 91, 92]. The X-ray crystal structures of xTRPV4_miN647Q in the presence of Cs⁺ (PDB 6C8F), Ba²⁺ (PDB 6C8G) and Gd³⁺ (PDB 6C8H), though at low resolution, each show anomalous density for a single cation in the selectivity filter - coordinated by the backbone carbonyls of Ile674 and Gly675[89]. By contrast, in the presence of Ca²⁺ the X-ray crystal structure of rTRPV6* (PDB 5WO9) shows two anomalous densities for Ca²⁺ in the selectivity filter – one coordinated on the extracellular side by the side chain carboxylic acid of Asp541 and the other coordinated on the interior by the backbone carbonyl and sidechain hydroxyl of Thr538[92]. There is a third weak anomalous density for Ca²⁺ in the vestibule below the selectivity filter, coordinated by Met569 on S6. They also obtained a structure of rTRPV6* in the presence of Gd³⁺ (PDB 5WOA) where they saw only a single density in the outermost position of the selectivity filter[92]. Yuan and Hite predicted that the single cation density observed binding to xTRPV4_miN647Q accounts for a lack of selectivity as whatever ion binds has an equal chance of exiting on either side of the selectivity filter[89]. On the other hand, the Sobolevsky group predicted that the multiple Ca²⁺ binding sites observed for rTRPV6* are indicative of a ‘knock-off mechanism where the affinity of cations for multiple binding sites tunes the channel’s selectivity[92]. Both groups concluded that the single density seen for Gd³⁺ ions was due to high affinity for the site and was likely responsible for channel blocking observed under high Gd³⁺ concentrations[89, 92].

While TRPV5 and TRPV6 have a very tight selectivity filter to maintain Ca²⁺ selectivity (Fig. 3), the thermoTRPV channels seem to have much more dynamic selectivity filters (Fig. 2)[94, 96]. Although apo TRPV1 (PDB 5IRZ) and the initial apo TRPV2 structures (PDB 5AN8, 5HI9) show seemingly closed selectivity filters, apo TRPV3 (PDB 6DVW, 6MHO) and TRPV4 (PDB 6BBJ) have much more open selectivity filters (Fig. 2)[78–80, 85, 86, 89]. The open DkTx/RTX-bound state of TRPV1 (PDB 5IRZ) shows a significant opening of the selectivity filter and the recent semi-open state obtained of apo TRPV2 in nanodiscs (PDB 6U86) also shows a significantly widened selectivity filter (Fig. 2)[78, 83]. This all agrees well with a recent report that the selectivity filter does not function as a gate in TRPV1-3 (TRPV4 was not tested)[97]. The authors observed that Ca²⁺ sized cations could penetrate the selectivity filter in the absence of channel stimulus, and that even large organic cations could enter, if not pass through, the non-conducting apo pore. They suggested that selectivity filter widening may aid structural changes that allow the channel to open at the lower gate and that the selectivity filter is not a major impediment for ion permeation.

So far, two different modes have been observed for the opening of the lower gate of TRPV channels. The first is a stable π-helix in the upper region of S6 which acts as a hinge to allow the lower portion of the S6 helix to move away from the pore, as seen for TRPV1 and TRPV5 (Fig. 2a–b, 3a–b, 4a)[19, 78, 96, 98]. The second is a switch from an inactive continuous α-helix to an activating π-helix, causing the residues forming the hydrophobic seal of the lower gate to rotate away from the pore, as seen in TRPV3 and TRPV6 (Fig. 2 e–f,3 c–d,4a)[86, 88, 93]. The apo states of TRPV2 and TRPV4 feature continuous α-helices and it seems likely that the open states will also follow the π-helix mode of gating (Fig. 2c–d, h, 4a)[83, 89]. TRPV channels have a highly conserved methionine residue near the bottom of S6, which is oriented into the pore in apo TRPV2, TRPV3, TRPV4, and TRPV6, and a highly conserved isoleucine residue three positions before, which is oriented into the pore in open TRPV3 and TRPV6 and persistently in TRPV1 and TRPV5 (Fig. 2 c–e, h, 3c, 4a)[19, 78, 83, 86, 88, 89, 93]. While the rotation of S6 to turn the methionine out of the pore and the isoleucine into the pore is not sufficient to open the channel, as observed in the sensitized TRPV3 states (PDB 6MHS, 6PVM, 6PVN, 6PVO), based on the currently resolved open TRPV structures it does seem to be a necessary precursor of channel opening (Fig. 2g, 4a)[85, 88].

Figure 4.

TRPV mechanisms revealed by structural biology. (a) All open TRPV structures determined so far have a π-helix (purple) in S6 (green). Some of these channels already have a π-helix in their apo states, which acts as a hinge allowing the lower portion of S6 to open. Other channels have a totally α-helical S6 in their apo states (grey), which requires the lower portion of S6 to rotate by ~100° to form the π-helix. TRPV3 has been captured in a sensitized state (yellow) where the channel has transitioned from an α- to π-helix in S6, but remains closed, (b) The CTD/NTD switch model for proposed by the Lee group and seen in TRPV2 and TRPV3. The ‘off’ state sees the CTD on one monomer (pink) wrapped around the beta-sheet region and making contacts with the ARD of an adjacent monomer (grey). The ‘on’ state has the CTD form a short helix on the interior of the ARD skirt (dark pink), while the NTD of the adjacent monomer (grey) occupies the CTD vacated. The interaction of the helical CTD (dark pink) with the interior of the adjacent ARD (grey) may contribute to channel opening. (c) At low Ca²⁺ levels, only the C-Lobe of CaM (orange) is Ca²⁺ (yellow) bound and thought to be persistently bound to the distal CTD (blue) of TRPV5 and TRPV6. When Ca²⁺ levels increase, the N-Lobe (orange) binds and is activated by Ca²⁺ and can then in turn bind to the proximal CTD (light blue), bringing the C-Lobe in close enough to block the bottom of the pore.

Another highly conserved residue in the S6 helix is an asparagine residue directly below the site of π-helix formation (Asn671 in human TRPV3). The Lee & Lander groups predicted that this asparagine can stabilize the π-helix by interacting with the backbone carbonyl exposed by π-helix formation[85], an interaction observed in most TRPV channel structures resolved so far with π-helices in S6[17, 19, 51, 76, 78, 86–88, 93]. They also observed in their apo TRPV3 structure (PDB 6MHO) that in the absence of a π-helix in S6 this residue could interact with the S4-S5 linker, thereby potentially coupling π-helix formation with movements at the S4-S5 linker[85]. The rotation of Asn671 into the pore in the π-helix conformation also reduces the hydrophobicity of the central pore, which could aid ion permeation. This Asn residue is highly conserved in this position across the TRPV family, and its importance in TRPV1 channel function has also been analyzed by molecular dynamics[99, 100].

Both structural and functional work have implicated endogenous lipids as a critical factor in channel gating. While structures of channels stabilized in nanodiscs reveal many non-protein densities thought to be lipids, even structures obtained in detergents and amphipols have seen strong non-protein densities predicted to be lipids – likely co-purified with the protein. The two most consistent and apparently critical lipid binding sites identified across the TRPV family are the vanilloid binding pocket and VSLD binding pocket.

While the vanilloid binding pocket of TRPV1 was initially observed with only ligands bound, the higher resolution apo structure of rTRPV1_mi obtained in nanodiscs (PDB 5IRZ) revealed lipid density in this pocket that they assigned to PI[78]. The authors suggested that the lipid functioned to trap the channel in a closed state, and that displacement of this lipid by drugs freed the channel to open[78]. While weak lipid density has been observed here in multiple TRPV2 structures, no lipid density has been observed in this pocket in TRPV3 and TRPV4[79, 80, 82–88]. TRPV5 and TRPV6 both see very strong lipid density in the vanilloid pocket, though of a shape which suggests cholesterol rather than a PI Iipid[4, 17–19, 51, 57, 90–93]. The Sobolevksy group observed much stronger density for this lipid in their open hTRPV6 structure (PDB 6BO8) compared to their closed structure of rTRPV6* in nanodiscs (PDB 6BOB), so they hypothesized that it might function to activate the channel rather than inhibit it as in TRPV1[93]. Both closed and open structures of TRPV5 have strong lipid density in the vanilloid binding pocket, and the TRPV5 inhibitor econazole was also observed to bind here[19, 57].

The Julius & Cheng groups first identified a lipid in the VLSD pocket as phosphatidylcholine with their rTRPV1_mi in nanodisc structures[78]. Since then, most other TRPV structures published have had prominent lipid density in this pocket, though no lipid density has been observed in TRPV4, likely because of structural divergence in this region. In addition to binding lipids, the VSLD pocket has also been identified as a binding site for the TRPV6 inhibitor 2-APB (PDB 6D7T, 6D7Q) and TRPV5 inhibitor ZINC17988990 (PDB 6PBE)[51, 90]. The Sobolevsky group found that by weakening lipid binding in the VSLD pocket in TRPV6 and TRPV3, they could increase the efficacy of 2-APB inhibition of TRPV6 or activation of TRPV3[51, 86].

The opening of TRPV5 and TRPV6 requires only modest movements of S6 and the TRP helix, with very little movement propagated to the rest of the channel[19, 93]. In stark comparison, the opening of TRPV1 and TRPV3 feature movements of the entire channel, with the ARDs moving relative to the TMD and movements throughout S1-S6[76, 78, 85, 86]. One component of this overall movement is a swap between the CTD and NTD at the β-sheet and ARD interface between monomers that has been recently observed for both TRPV2 and TRPV3[83, 87, 88]. TRPV2 and TRPV3 require sensitization by either repeated or prolonged exposure to stimuli to reach a fully active state. Based on their recent structures of TRPV3 with 2-APB, the Lee group proposed a switch model where the CTD wrapping around the β-sheet region would constitute an ‘off’ or inactive state, a switch that could be turned ‘on’ by swapping the CTD for the NTD (Fig. 4b)[87]. This ‘on’ state would allow the CTD to form a short helix on the interior of the ARD skirt where interactions with a loop of AR5 would induce rotation of the ARDs that propagate to the TMD and make π-helix formation and channel opening more energetically favorable. They argued that this switch may be responsible for the sensitization requirement of TRPV3, and that only after extensive exposure to stimuli would enough of the channel be in the ‘on’ state to give a full stimulus response[87]. While the original TRPV2 apo structures (PDB 5AN8, 5HI9) were both obtained in the ‘off’ state, the recent apo structures of rTRPV2 in nanodiscs (PDB 6U84, 6U86) show the apo channel in the ‘on’ state in a lipid environments[79, 80, 83]. The Lee & Lander groups originally proposed that their apo rbTRPV2_mi structure (PDB 5AN8) was desensitized based on a comparison to the TRPV1 structures available at the time, which is consistent with this model[79]. Additionally, the two CBD-bound TRPV2 states recently obtained are in both the ‘on’ (PDB 6U8A) or ‘off’ (PDB 6U88) state, which may correspond to a preopen and desensitized state, respectively, based on this model[83]. One difficulty in examining this switch across the thermoTRPV family is that many structural studies, including all of the TRPV1 and TRPV4 studies, were done using constructs with the N-terminus truncated up to the start of the ARDs. Since thermoTRPV constructs are still functional with this truncation, it seems likely that the change in conformation of the CTD is sufficient to turn the switch ‘on’ without coordination with the NTD.

Several structures of TRPV2 and TRPV3 have recently been reported with C2 symmetry rather than C4 symmetry[81, 82, 85, 88]. Most of these reductions in symmetry have been modest and have not revealed significantly different states of the protein. The recent TRPV3 structure from the Sobolevsky group labelled as an intermediate state (PDB 6PVQ) has the most dramatic C2 symmetry and features two opposite monomers with the CTD/NTD switch ‘on’ and two with it ‘off’[88]. Unfortunately, the pore region of this structure is poorly resolved and so it is impossible to determine the state of the pore.

Three cryo-EM papers have now been published examining the interactions between fully assembled TRPV5/TRPV6 and CaM[17–19], adding to previous electrophysiological, biochemical and NMR work identifying and characterizing the interaction of a portion of the TRPV5 and TRPV6 C-termini with CaM[20, 22–24, 101]. These previous studies had already determined that the CaM C-lobe binds to the distal TRPV5/TRPV6 C-terminal binding site at endogenous Ca²⁺ levels and that a second binding event to the N-lobe at high Ca²⁺ levels would be required for channel inhibition. By combining the functional model determined by NMR with the recent cryo-EM data, a consensus mechanism for CaM inhibition of TRPV5 and TRPV6 has emerged. The intracellular Ca²⁺ sensor CaM is an exquisitely tuned sensor for Ca²⁺ concentrations and each lobe can only bind to TRPV5 and TRPV6 when Ca²⁺ bound. The CaM C-lobe has a higher affinity for Ca²⁺ than the N-lobe and is effectively always Ca²⁺ bound at basal Ca²⁺ levels, while the N-lobe only becomes Ca²⁺ bound under high intracellular Ca²⁺ levels. The far distal end of the TRPV5/6 C-terminus would therefore be constitutively bound to the C-lobe of CaM, and under high Ca²⁺ levels the N-lobe could bind the more proximal end of the TRPV5 and TRPV6 C-termini, thus reeling in the C-lobe of CaM to block the base of the pore with a lysine residue (Fig 4c). TRPV5 and TRPV6 are thought to effectively be constitutively active due to their stimulation by PI(4,5)P₂, so this method of Ca²⁺-based inhibition may be a way to rapidly stop Ca²⁺ influx without the need to alter membrane lipids[4, 17–19, 22].

CaM has also been reported to regulate some of the thermoTRPV channels, but the C-termini of these channels do not align well to the C-termini of TRPV5 and TRPV6. A structure of CaM bound to a peptide from the TRPV1 C-terminus was solved with X-ray crystallography (PDB 3SUI), where a single helical peptide (Gly784-Asp798) bound between both CaM lobes[21]. It remains to be seen whether this alternative mode of binding yields a similar pore blockage since the thermoTRPV channels also lack the ring of tryptophan residues which interact with CaM Lys115 in the TRPV5 and TRPV6 structures. It would also be interesting to see how this mechanism interacts with the proposed CTD/NTD switch, where the CTD moves between the inside and outside of the ARD skirt.

Where does this leave the TRPV structural field moving forward? The mechanisms sketched out by the structures published so far suggest a variety of immediate follow-ups – such as further study of the NTD/CTD switch in the thermoTRPVs, more investigation into lipid modulation, and examination of the interaction between CaM and the thermoTRPVs. Structures with small molecule modulators have also been a fruitful area of research, having yielded several novel and unique drug binding sites in 2019 alone. The channel termini have been shown to play an important role in channel regulation but due to the flexible nature of these regions, the assembled structures published so far have yet to resolve the N- and C-termini of the channels beyond short isolated stabilized peptides. It is exciting to contemplate the possibilities for resolving the TRPV termini in the context of the fully assembled channel as techniques for handling and imagine membrane proteins advance. Beyond this, there is a great deal of information about channel movements to be mined from cryo-EM datasets, especially as data processing programs become more sophisticated. Most of the TRPV cryo-EM structures published were of a single state from a single dataset, but in the last two years a handful of papers have reported multiple unique structures found in a single dataset – a step on the road towards harnessing the potential of cryo-EM to reveal the full range of channel movement.

Highlights.

• Structural Biology of TRPV channels

• Binding pockets for ligand activators and inhibitors

• Mechanisms of endogenous channel modulation