1. Structural Biology and Molecular Biophysics
Download icon

Symmetry transitions during gating of the TRPV2 ion channel in lipid membranes

  1. Lejla Zubcevic
  2. Allen L Hsu
  3. Mario J Borgnia
  4. Seok-Yong Lee  Is a corresponding author
  1. Duke University School of Medicine, United States
  2. National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, United States
Research Article
  • Cited 3
  • Views 1,216
  • Annotations
Cite this article as: eLife 2019;8:e45779 doi: 10.7554/eLife.45779

Abstract

The Transient Receptor Potential Vanilloid 2 (TRPV2) channel is a member of the temperature-sensing thermoTRPV family. Recent advances in cryo-electronmicroscopy (cryo-EM) and X-ray crystallography have provided many important insights into the gating mechanisms of thermoTRPV channels. Interestingly, crystallographic studies of ligand-dependent TRPV2 gating have shown that the TRPV2 channel adopts two-fold symmetric arrangements during the gating cycle. However, it was unclear if crystal packing forces played a role in stabilizing the two-fold symmetric arrangement of the channel. Here, we employ cryo-EM to elucidate the structure of full-length rabbit TRPV2 in complex with the agonist resiniferatoxin (RTx) in nanodiscs and amphipol. We show that RTx induces two-fold symmetric conformations of TRPV2 in both environments. However, the two-fold symmetry is more pronounced in the native-like lipid environment of the nanodiscs. Our data offers insights into a gating pathway in TRPV2 involving symmetry transitions.

https://doi.org/10.7554/eLife.45779.001

Introduction

Transient Receptor Potential V (TRPV) channels are part of the larger TRP channel family which plays important roles in numerous physiological processes (Clapham et al., 2001). A subset of TRPV channels, including subtypes TRPV1-TRPV4, possess an intrinsic capability to sense heat and are therefore referred to as thermoTRPV channels (Cao et al., 2013a; Liu and Qin, 2016; Smith et al., 2002; Chung et al., 2003). TRPV1-TRPV4 are non-selective cation channels which play important physiological roles in sensing noxious heat (Bölcskei et al., 2010; Julius, 2013; Marwaha et al., 2016; Mitchell et al., 2014), maintaining cardiac structure (Katanosaka et al., 2014) and maintaining skin (Eytan et al., 2014; Imura et al., 2009; Kim et al., 2016), hair (Asakawa et al., 2006; Imura et al., 2007; Xiao et al., 2008) and bone physiology (Masuyama et al., 2008). A distinctive feature of TRPV1 and TRPV2 is their permeability to large organic cations (Chung et al., 2008), such as the cationic dye YO-PRO-1 and the sodium channel blocker QX-314. This feature has led to proposals to utilize these channels as conduits for delivering small molecules to intracellular targets (Puopolo et al., 2013). The non-conducting structures of TRPV1 and TRPV2 possess two restrictions, one at the selectivity filter (SF) and second one at the intracellular mouth of the pore (termed the common gate) (Liao et al., 2013; Zubcevic et al., 2016; Huynh et al., 2016). Both restrictions must open widely to accommodate the passage of large organic cations. However, the mechanism that enables such opening was long unclear. In order to study the permeation of both metal ions and large organic cations in TRPV2, we recently crystallized the rabbit resiniferatoxin (RTx)-sensitive (Zhang et al., 2016) TRPV2 channel with a truncation in the pore turret in the presence of the agonist RTx (Zubcevic et al., 2018a). This study led to the revelation that the binding of RTx leads to a two-fold symmetric (C2) opening at the selectivity filter that is wide enough to permeate YO-PRO-1. This unexpected result offered the first experimental evidence that the homotetrameric TRPV2 can adopt C2 symmetric conformations during the gating cycle. However, it was unclear if crystal contacts or the crystallization conditions (e.g. high concentration of Ca2+) played a role in stabilizing the C2 symmetry. In addition, the minimal TRPV2 construct used in the crystallographic study lacked the pore turret, a region that is not essential for function (Liao et al., 2013; Zubcevic et al., 2016; Zubcevic et al., 2018a; Yao et al., 2010) but had previously been shown to have a modulatory effect on gating in TRPV1 and TRPV2 (Jara-Oseguera et al., 2016; Dosey et al., 2019). It was uncertain if the absence of this region in our crystallographic study affected the symmetry of the channel.

In order to answer these questions and further study the role of two-fold symmetry in TRPV channel gating, we conducted cryo-electronmicroscopy (cryo-EM) studies of the full-length, RTx-sensitive rabbit TRPV2 (Zhang et al., 2016) channel reconstituted into nanodiscs and amphipol. We present four structures of the TRPV2/RTx complex, one obtained in nanodiscs (TRPV2RTx-ND) and three in amphipol (TRPV2RTx-APOL 1-3) determined to 3.8 Å, 2.9 Å, 3.3 Å and 4.2 Å resolution, respectively (Table 1, Figure 1). Our data shows that binding of RTx induces C2 symmetric conformations in TRPV2, but the extent of C2 symmetry depends on the environment in which the channel is reconstituted. C2 symmetry is particularly pronounced in the dataset collected from nanodisc-reconstituted TRPV2, which better approximates the physiological environment of the channel. Moreover, the data offers further insights into the allosteric coupling between the RTx-binding site and the activation gates in TRPV2, confirms the critical role of the S4-S5 linker π-helix (S4-S5π-hinge) in ligand-dependent gating of TRPV2, and provides a glimpse of the conformational landscape of TRPV2 gating.

Figure 1 with 7 supplements see all
Overview of TRPV2RTx-APOL and TRPV2RTx-ND structures.

(A) Orthogonal view of TRPV2RTx-APOL 1-3 and TRPV2RTx-ND structures. TM domains are colored in grey and the cytoplasmic domains (ARD and C-terminal domain) are colored in blue, violet, salmon and red, respectively. RTx is shown in stick and sphere representation and colored in yellow. Lines drawn between diagonally opposite ARDs (residue E95, shown in red, orange, blue and cyan spheres, respectively) illustrate the relative position of ARDs in the tetramer. The close-up shows that the ankyrin repeats of diagonally opposing subunits in TRPV2RTx-APOL 2 and TRPV2RTx-ND are positioned in different planes. (B) Top view of the channel (S5, S6 and PH are colored in blue, violet, salmon and red, respectively). Lines drawn between residues V620 in the S6 helix illustrate the symmetry within the pore domain. Distances and angles indicate the presence of two-fold symmetry.

https://doi.org/10.7554/eLife.45779.002
Table 1
Data collection and refinement statistics
https://doi.org/10.7554/eLife.45779.010
Data collection and processingTRPV2RTx-NDTRPV2RTx-APOL 1TRPV2RTx-APOL 2TRPV2RTx-APOL 3
 Electron microscopeTitan KriosTitan Krios
 Electron detectorFalcon IIIFalcon III
 Magnification75,000x75,000x
 Voltage (kV)300300
 Electron exposure (e–/Å2)4242
 Defocus range (μm)−1.25 to −3.0−1.25 to −3.0
 Pixel size (Å)1.081.08
 DetectorCountingCounting
 Total extracted particles (no.)1,407,292580,746
 Refined particles (no.)482,602470,760
Reconstruction
 Final particles (no.)112,622101,570109,62390,862
 Symmetry imposedC2C4C2C2
 Nominal Resolution (Å)3.82.93.34.19
 FSC 0.143 (masked/unmasked)3.7/3.92.9/3.053.2/3.54.0/4.3
 Map sharpening B factor (Å2)−30−78−92−133
Refinement
Model composition
Non-hydrogen atoms
Protein residues
Ligands

16,878
2396
RTx: 4

18,236
2404
RTx: 4

18,452
2440
RTx: 4

17,548
2440
RTx: 4
Validation
MolProbity score
Clashscore
Poor rotamers (%)

1.39
4
0

1.11
1.9
0

1.28
2.7
0

1.37
2.7
0
Ramachandran plot
Favored (%)
Allowed (%)
Disallowed (%)
96.5
3.5
0
97.1
2.9
0
96.6
3.4
0
95.5
4.5
0

Results

In order to capture the RTx-induced gating transitions in the rabbit TRPV2 channel, we conducted cryo-EM studies of the TRPV2/RTx complex reconstituted into amphipol (TRPV2RTx-APOL) and nanodiscs (TRPV2RTx-ND). Amphipols (Zoonens and Popot, 2014) have been a useful tool in structural studies of membrane proteins, and especially TRP channels (Liao et al., 2013; Zubcevic et al., 2016; Cao et al., 2013b; Paulsen et al., 2015; Yoo et al., 2018; Hirschi et al., 2017; Zubcevic et al., 2018b). Indeed, Amphipol A8-35 enabled the very first structural determination of the TRPV2 channel (Zubcevic et al., 2016). Nanodiscs, on the other hand, represent the closest in vitro approximation to the native lipid membranes used in structural studies (Denisov and Sligar, 2016). The data was processed using RELION (Scheres, 2012) (Materials and methods), with no symmetry imposed during particle classification and 3D reconstruction in order to avoid obscuring any classes with lower symmetry (C1 and C2) that might exist in the sample. Symmetry was only imposed in the last step of the refinement and only if the 3D reconstructions showed clear two-fold (C2) or four-fold (C4) symmetry (Figure 1—figure supplements 13). Classification of the TRPV2RTx-APOL sample revealed the presence of four classes: one low-resolution (~7 Å) class, which was excluded from further analysis, and three higher resolution classes which are representative of three different conformations. These include one C4 symmetric and two distinct C2 symmetric classes refined to 2.9 Å, 3.3 Å and 4.2 Å, respectively (Figure 1, Figure 1—figure supplement 1). By contrast, 3D classification of the TRPV2RTx-ND converged on a single C2 symmetric conformation resolved to 3.8 Å (Figure 1, Figure 1—figure supplements 23). All four maps were of sufficient quality to enable placement of individual structural motifs with confidence (Figure 1—figure supplements 47) and the models for all four structures were built to good overall geometry (Table 1).

The transmembrane domains of TRPV2RTx-APOL are trapped in a closed conformation

Unexpectedly, the transmembrane domains (TM) of the three structures obtained from amphipol-reconstituted TRPV2, TRPV2RTx-APOL 1-3, show similarity to our previously solved cryo-EM structure of TRPV2 in its apo form (Zubcevic et al., 2016) (TRPV2APO) and adopt non-conducting conformations (Figure 2—figure supplement 1). While fully bound to RTx, the TM domains of TRPV2RTx-APOL 1 and TRPV2RTx-APOL 2 structures largely retain C4 symmetry (Figure 1 and Figure 3—figure supplement 1). However, the TMs of TRPV2RTx-APOL 3 exhibit a slight departure from C4 symmetry in the pore (Figure 3—figure supplement 2). The effects of RTx on the TRPV2RTx-APOL are particularly obvious in the ankyrin repeat domains (ARD) of the two-fold symmetric TRPV2RTx-APOL 2 and TRPV2RTx-APOL 3 which display pronounced broken symmetry and a range of rotational states (Figure 1A, Figure 3—figure supplements 13).

In order to determine the effect of RTx on the TRPV2RTx-APOL sample, we aligned TRPV2RTx-APOL 1 with TRPV2APO. The transmembrane helices S1-S6 of the two channels aligned remarkably well (Cα R.M.S.D = 0.86) (Figure 3—figure supplement 1). However, RTx binding induces a 5° clockwise rotation of the ARD when viewed from the extracellular space and a ~ 10 Å lateral widening of the cytoplasmic assembly (Figure 3—figure supplement 1). In addition, RTx causes a conformational change in the S4-S5 linker (Figure 3—figure supplement 1), as well as a displacement of the TRP domain (Figure 3—figure supplement 1). The conformational change in the S4-S5 linker is caused by the introduction of a π-helical turn at the junction of the S4-S5 linker and the S5 helix in the TRPV2RTx-APOL 1 structure (S4-S5π-hinge), which is absent in TRPV2APO (Figure 3—figure supplements 1 and 4). This observation concurs with our previous finding that RTx binding elicits a conformational change in the S4-S5 linker, and that the S4-S5π-hinge is critical for ligand-dependent gating in TRPV2 (Zubcevic et al., 2018a). In TRPV2RTx-APOL 3, slight C2 symmetry is observed in the TM domains and is evident in the SF, PH and the S4-S5 linker (Figure 3—figure supplement 2). Nevertheless, the RTx-induced conformational changes in the S4-S5 linker are not efficiently propagated to the TM in the TRPV2RTx-APOL structures, and they fail to open either of the two restrictions in the pore (Figure 2—figure supplement 1). Instead, RTx only effects changes in its immediate binding site above the S4-S5 linker and in the parts of the channel not bound by amphipol, strongly suggesting that the polymer constricts the TM and prevents conformational changes at the S4-S5 linker and the ARD from propagating to the TM domain. The fact that the TRPV2/RTx complex is stabilized in multiple distinct closed states with different arrangements of the ARD assembly (Figure 1, Figure 3—figure supplements 13) suggests that the conformational changes in the ARD might represent low-energy, pre-open states that can be achieved without substantial changes in the TM domains.

Interestingly, metal ions are not visualized in the pores of any of the TRPV2RTx-APOL structures, despite the high resolutions obtained in this study. Whether this is the result of cryo-EM experimental conditions is unclear, but thus far metal ions occupying the SF and the pores of thermoTRPV channels have only been captured in structures obtained by X-ray crystallography (Zubcevic et al., 2018a).

RTx induces a break in symmetry in TRPV2RTx-ND

In stark contrast to the amphipol-reconstituted channel, RTx binding induces C2 symmetry in the nanodisc-reconstituted TRPV2 which extends throughout the channel. The symmetry of the TRPV2RTx-ND map was assessed both visually, and by the Map Symmetry function in Phenix which yielded a CC = 0.9 and score of 1.28 for C2 symmetry. For comparison, C4 symmetry yielded a lower correlation coefficient (CC = 0.8). To further confirm the correctness of the symmetry assignment, we evaluated the fit of the TRPV2RTx-ND model built into the C2 symmetric map to the non-symmetrized C1 map (Figure 1—figure supplement 3). In addition, we evaluated the fit of the TRPV2RTx-ND model to the individually refined non-symmetrized classes 1 and 6, which constitute the TRPV2RTx-ND map (Figure 1—figure supplement 3). All FSC curves indicate that the two-fold symmetric model fits well into the C1 maps and the density of the C1 symmetric TRPV2RTx-ND map supports the model (Figure 1—figure supplement 3), showing that two-fold symmetry is truly present in the TRPV2RTx-ND sample.

The pore of TRPV2RTx-ND adopts a C2 symmetric arrangement (Figure 2A). The pore helices are arranged so that the carbonyl oxygens of the selectivity filter in subunits B and D line the entry to the pore while pore helices of subunits A and C tilt away from the permeation pathway. This arrangement creates a large C2 symmetric opening where the narrowest constriction between SF residues in diametrically opposing subunits A and C and B and D is ~11 Å and ~8 Å, respectively. This results in an SF with ample room to accommodate large organic cations (Figure 2B). A closer look at the pore helices reveals that this arrangement in the SF is achieved through a ~ 10° swivel of the subunit A pore helix, which brings the N-terminal part of the helix closer to S5 while distancing it from S6 (Figure 2C). The position of the pore helices controls the size and the shape of the SF, and appears to exert control over ion permeation in TRPV2. While the SF is widely open, the common gate adopts a putative intermediate conformation where two of the diagonally opposing subunits adopt a closed state, and the remaining two are open. In subunits A and C, the S6 helix adopts a straight α-helical, closed conformation, while the S6 of subunits B and D is bent and the common gate apparently open (Figure 2A). However, the overall functional state of the common gate is likely non-conductive as the gate residues from subunits A and C would presumably hinder ion permeation.

Figure 2 with 1 supplement see all
Overview of the pore in the TRPV2RTx-ND structure.

(A) S6 and pore helices of subunits A and C (left) and subunits B and D (right). Pore helices are shown in both cartoon and cylinder representation (grey). Dashed lines and values represent distances between the indicated residues. S6 helices in A and C are straight and α-helical, while the S6 in subunits B and D is bent. (B) Top view of the TRPV2RTx-ND pore, with pore helices shown in both cartoon and cylinder representation. Dashed lines illustrate the distances between residues G604 in the selectivity filter. (C) Overlay of the TRPV2RTx-ND pore domains (S5, S6 and pore helices). Subunit A is shown in red and subunit B in violet. The pore helix of subunit A swivels by ~10° relative to subunit B.

https://doi.org/10.7554/eLife.45779.011

In order to establish the origin of the C2 symmetry in the TRPV2RTx-ND structure, we aligned subunits A and B (Cα R.M.S.D = 0.96) (Figure 3—figure supplement 5). Similar to our previous findings, this alignment shows that the two subunits diverge at the S4-S5 linker and the PH and indicates that rotation of subunits around the S4-S5π-hinge appears to result in the distinct C2 symmetric arrangement observed in TRPV2RTx-ND (Figure 3—figure supplements 45).

When compared to the TRPV2APO, the TM domains of the TRPV2RTx-ND structure appear to contract in an asymmetric manner (Figure 3A), while the ARD assembly expands by ~10 Å and rotates by 3° (Figure 3B). The TM domains and the ARDs appear to move as a single rigid body, which is evident when individual subunits from TRPV2APO and TRPV2RTx-ND are superposed (Cα R.M.S.D = 1.9 Å) to reveal that only the S4-S5 linker and the pore helix deviate significantly in the two structures (Figure 3C). This coupled movement of the TM and ARD indicates that RTx-binding to TRPV2 in lipid membranes induces a rigid-body rotation of the entire subunit that originates at the S4-S5π-hinge (Figure 3D–E).

Figure 3 with 5 supplements see all
Comparison of TRPV2RTx-ND (red) and TRPV2APO (orange).

(A) Overlay of TRPV2RTx-ND and TRPV2APO, top view. TRPV2RTx-ND is shown in cartoon representation and TRPV2APO as cylinders. Relative to TRPV2APO, the TM subunits of TRPV2RTx-ND exhibit contraction (red arrows). (B) Top view of the ARDs in TRPV2RTx-ND and TRPV2APO. TM helices are removed for ease of viewing. Dashed lines represent distances between residues T100, showing a 10 Å expansion (Δ 10 Å) and 3° rotation (Δ 3o) of the TRPV2RTx-ND ARD assembly relative to TRPV2APO. (C) A rigid-body rotation of TRPV2RTx-ND subunit B around the S4-S5 linker achieves alignment with the subunit B from TRPV2APO. Following alignment, only the S4-S5 linkers and the pore helices (PH) diverge in the two subunits (dashed box). (D) Cartoon illustrating how the movements of the TM and the ARD in TRPV2RTx-ND are coupled. The red and orange shapes represent a single subunit of TRPV2RTx-ND and TRPV2APO, respectively. The rotation of the subunit is manifested as ‘contraction’ in the TM domains and ‘expansion’ of the ARD. (E) RTx binding in the vanilloid binding pocket exerts force on the S4-S5 linker, changing the conformation of the junction from α- to π-helix, and induces the rotation of the subunit around the S4-S5π-hinge.

https://doi.org/10.7554/eLife.45779.013

Interestingly, the TRPV2RTx-ND structure exhibits different degrees of reduced symmetry from the previously determined crystal structure of TRPV2 in complex with RTx (TRPV2RTx-XTAL) (Zubcevic et al., 2018a). Compared to the TRPV2RTx-XTAL, the TM domains of TRPV2RTx-ND contract in an two-fold symmetric manner (Figure 4A). This conformational change, which stems from rotation of individual TRPV2RTx-ND subunits around the S4-S5π-hinge (Figure 4—figure supplement 1), results in an overall fold that is closer to C4 symmetry than that of the TRPV2RTx-XTAL (Figure 4B). However, while the TRPV2RTx-ND helices S1-S6 adopt a more C4 symmetric arrangement, the pore helices and the SF remain distinctly C2 symmetric (Figure 4C). Remarkably, the SF of TRPV2RTx-ND is wider than that of TRPV2RTx-XTAL, and the two structures display different C2 symmetric openings at the SF (Figure 4C). The two different conformations result from both the different arrangements of subunits and changes in the position and tilt angle of the pore helices (Figure 4D–E). In the TRPV2RTx-XTAL structure, the pore helices of subunits B and D, which assume a widened conformation, are free of interactions with the pore domain, while a network of interactions (presumably hydrogen bonds) between Y542-T602-Y627 in subunits A and C tethers the pore helices to S5 and S6. Our previous work showed that disruption of these interactions is detrimental to the permeation of large organic cations but has no effect on permeation of metal ions (Zubcevic et al., 2018a). Interestingly, the putative hydrogen bond triad is disrupted in all four subunits of the TRPV2RTx-ND structure (Figure 4—figure supplement 2). Nevertheless, the SF assumes a fully open state that can potentially accommodate passage of a large cation. This suggests that the putative hydrogen bond triad, while not a feature of the fully open SF, is an essential part of the transition between closed and open states of the channel.

Figure 4 with 6 supplements see all
Comparison of TRPV2RTx-ND (red) and TRPV2RTx-XTAL (cyan).

(A) Overlay of TRPV2RTx-ND and TRPV2RTx-XTAL, top view. TRPV2RTx-ND is shown in cartoon representation and TRPV2RTx-XTAL as cylinders. Relative to TRPV2RTx-XTAL, the TM domains of TRPV2RTx-ND are contracted (red arrows). (B) Comparison of two-fold symmetry in TRPV2RTx-ND and TRPV2RTx-XTAL. Dashed lines represent distances between residues A427. The distances between diagonally opposing subunits are indicated. (C) Top view of the SF in TRPV2RTx-ND and TRPV2RTx-XTAL. Pore helices are shown in both cartoon and cylinder representation. Dashed lines represent distances between residues G604 in the selectivity filter. (D–E) Overlay of the pore domains of TRPV2RTx-ND and TRPV2RTx-XTAL subunit A (D) and subunit B (E) shows that the pore helices A and B in TRPV2RTx-ND swivel by ~8° and ~10°, respectively, compared to TRPV2RTx-XTAL.

https://doi.org/10.7554/eLife.45779.019

It is interesting to point out that extensive rearrangements around the SF and the PH during gating have thus far only been observed in structural studies of TRPV1 (Cao et al., 2013b) and TRPV2 (Zubcevic et al., 2018a) channels. In the non-conductive state, the SFs of the remaining members of the TRPV subfamily (TRPV3-TRPV6 [Zubcevic et al., 2018b; Singh et al., 2018; Deng et al., 2018; Hughes et al., 2018; McGoldrick et al., 2018]) adopt a conformation that is wide enough to accommodate a semi-hydrated cation, and do not move appreciably during channel activation. This may indicate that TRPV1 and TRPV2 are the only members of the TRPV subfamily that possess a gate at the SF, and that the coupling of structural elements necessary for activation of these channels differs from that of TRPV3-TRPV6 (Zhang et al., 2019).

As observed in our previous study (Zubcevic et al., 2018a), RTx assumes different binding poses in subunits of the C2-symmetric structures, both in the amphipol and the nanodisc samples (Figure 4—figure supplement 3) which may lead to the distinct conformations observed in these channels.

Despite the use of a full-length rabbit TRPV2 construct in this study, we were not able to confidently resolve the entire loop connecting S5 to the pore helix known as the ‘pore turret’. Interestingly, a recent structure of rat TRPV2 with the pore turret resolved showed that this region, which contains a large number of charged and polar residues, occupies the space within the membrane plane between S5 and the Voltage Sensor-Like Domain (VSLD) (Dosey et al., 2019). While the density in our cryo-EM maps was not of sufficient quality to build the entire pore turret with confidence, we do observe density following the S5 helix and preceding the pore helix. However, the direction of this density is perpendicular to the membrane and does not agree with the structure reported for rat TRPV2 (Figure 4—figure supplement 4). Indeed, the pore turret is the least conserved region amongst the TRPV2 orthologs, and the variations in its sequence might indicate that the turret adopts different conformations in TRPV2 channels from different species. Nevertheless, our study clearly shows that the omission of this region from the construct used in the crystallographic study of the TRPV2/RTx complex is not the cause of the C2 symmetry.

While both TRPV2RTx-ND and TRPV2RTx-XTAL structures adopt C2 symmetry, the distinct arrangement of subunits within the two channels suggests that the structures represent different functional states. We propose that TRPV2RTx-XTAL precedes TRPV2RTx-ND in the conformational activation trajectory based on two observations. Firstly, the common gate is fully closed in the TRPV2RTx-XTAL while it adopts an apparently partially open conformation in TRPV2RTx-ND (Figure 4—figure supplement 5). Secondly, our previous studies have shown that the putative hydrogen bond network between S5 and S6 and the pore helix is essential for the channel’s ability to transition to a fully open SF that can accommodate large organic cations (Zubcevic et al., 2018a). Nevertheless, in TRPV2RTx-ND the pore helices do not interact with S5 and S6 and the SF is fully open. We therefore propose that the conformational step that requires the presence of the putative hydrogen bond triad precedes the open SF conformation seen in TRPV2RTx-ND.

Discussion

Here, we have conducted a study that reveals symmetry transitions associated with gating of the TRPV2 channel by RTx. Interestingly, our data shows that RTx induces C2 symmetric conformations of TRPV2 in both amphipol and nanodiscs, and it thereby negates the hypothetical role of crystallization artefacts and crystal packing bias in stabilising two-fold symmetry. Similarly, C2 symmetry in TRPV2 is independent of the presence or absence of the pore turret region, suggesting that this region does not play an essential role in the regulation of the SF in rabbit TRPV2. Our study, similar to a previously published study of the magnesium channel CorA (Matthies et al., 2016), also emphasizes the notion that careful inspection of the intermediate maps and conservative application of symmetry during refinement of cryo-EM data can result in valuable insights into gating transitions and intermediate states. In addition, we have also investigated how amphipols and nanodiscs affect the conformational space that can be accessed during ligand gating of TRPV2.

While both TRPV2RTx-APOL and TRPV2RTx-ND are C2 symmetric, the two-fold symmetry in TRPV2RTx-APOL is confined to regions that are not bound by the amphipol polymer. This is evident in the fact that the TM domains, which are in contact with the amphipol, largely retain four-fold symmetry and the common gate and the SF remain firmly closed, while the ARD exhibit symmetry breaking, rotation and lateral expansion. These data, while adding valuable data points to the conformational landscape of TRPV2, also illustrate the potential caveats of using amphipols in studies of conformational changes in the transmembrane domains of proteins, as they appear to constrict the TM domains and stabilize low-energy pre-open states (Figure 4—figure supplement 6). However, at this time we caution against any generalized conclusions about the effect of amphipols and look forward to more systematic studies that will address these issues in the future. The TRPV2RTx-ND dataset yielded a single, two-fold symmetric structure thus giving strong evidence that RTx stabilizes two-fold symmetric conformational states in the TRPV2 channel in lipid membranes. The ARDs in the TRPV2RTx-ND structure echo the conformational changes observed in TRPV2RTx-APOL. However, in nanodiscs TRPV2 is captured with its SF fully open and its common gate in an intermediate conformation where the gate is apparently open in one set of diagonally opposing subunits and closed in the other. In this structure, the opening of the SF occurs according to a mechanism previously observed in the crystallographic study of the TRPV2/RTx complex where RTx binding in the vanilloid pocket, above the S4-S5π-hinge, induces a rigid body rotation of the entire subunit. In turn, the rotation causes a break in the interaction network between the pore helix and helices S5 and S6, allowing the pore helices to reposition and the SF to open (Zubcevic et al., 2018a).

Interestingly, however, the TRPV2RTx-ND structure differs from the previously obtained TRPV2RTx-XTAL. While both structures assume C2 symmetric conformations, the TRPV2RTx-ND channel appears to make a return toward C4 symmetry. Because the SF in TRPV2RTx-ND is fully open, and the common gate adopts an apparently partially open conformation, we reason that TRPV2RTx-ND follows the TRPV2RTx-XTAL structure in the conformational trajectory of the channel. Therefore, it is possible that TRPV2, as it travels toward the final open state where both the SF and the common gate are fully open, would adopt further conformations that increasingly approximate C4 symmetry (Figure 5). However, it is interesting to note that while the overall fold of TRPV2RTx-ND indeed is more C4 symmetric than that of TRPV2RTx-XTAL, the extent of C2 symmetry is not diminished in its SF. Because the symmetry of the SF does not appear to be dictated by the symmetry of the overall channel, we cannot exclude the possibility that the final open state might indeed possess a C2 symmetric SF while otherwise adopting a nearly C4 symmetric conformation. Our previous functional studies have suggested that C2 symmetric states play a role in the permeation of large organic cations and consequently in the full opening of the SF (Zubcevic et al., 2018a). Hence, the channel might be utilizing C2 symmetric states as means to achieve full opening in a step-wise manner. Similar C2 symmetric states elicited by ligand binding have been observed in TRPV3 (Zubcevic et al., 2018b) and TRPM2 (Yin et al., 2018) channels, which opens up the possibility that C2 symmetry might be widely associated with gating in members of the TRP channel superfamily. Intriguingly, a recent cryo-EM study of the human BK channel reconstituted in liposomes showed that this channel also enters C2 symmetric states (Tonggu and Wang, 2018), suggesting that two-fold symmetry might also play a role in the molecular mechanisms of other tetrameric ion channels.

Conformational states associated with RTx-mediated gating of TRPV2.

(A) TRPV2 subunit rotation upon binding of RTx. Rotation axis and direction are indicated in dashed line and circular arrow in apo TRPV2 (left). The rotation results in contraction of the TM domains and widening of the cytoplasmic assembly (right). (B) Hypothetical trajectory of TRPV2 gating with associated conformational states. Upon addition of RTx, TRPV2 first enters low-energy pre-open states that are characterized by rotation, widening and symmetry breaking in the ARD (TRPV2RTx-APOL 1-3, models shown in cartoon and surface representation). In the next step, the channel assumes C2 symmetric state with an open SF, but closed common (S6) gate (TRPV2RTx-XTAL, model shown in cartoon and surface representation). This is followed by a less C2 symmetric state with an open SF and semi-open common (S6) gate (TRPV2RTx-ND, model shown in cartoon and surface representation). Finally, we propose that the channel assumes a high-energy fully open state that is C4 symmetric but might have C2 symmetry in the SF. The SF is indicated in green in models and cartoons.

https://doi.org/10.7554/eLife.45779.026

Two-fold symmetry is a well-stablished feature of mammalian Na+ selective Two Pore Channels (TPCs) and Voltage Gated Sodium channels (NaV) (She et al., 2018; Shen et al., 2017; Pan et al., 2018; Shen et al., 2018). Interestingly, the arrangement of pore helices in TRPV2RTx-ND resembles that observed in TPC and NaV (Figure 6) and the selectivity filters in all three channels form a 'coin-slot' (Hille, 1971) opening. However, while the selectivity filters of TPC and NaV remain static during channel gating in order to maintain the structure necessary for Na+ selectivity, the SF of TRPV2 displays a large degree of plasticity. Moreover, the two-fold symmetry observed in TRPV2 is unique in that it arises in response to conformational changes in the TM domains induced by ligand binding. By contrast, the two-fold symmetry in TPC and NaV stems from the arrangement of their respective homologous tandem repeats.

Comparison of TRPV2RTx-ND (red), TPC1 (PDB 6C96, purple) and NaV1.4 (PDB 6AGF, blue).

Top view, pore helices are indicated.

https://doi.org/10.7554/eLife.45779.027

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Cell lineDH10Bac E. coliThermoFisher
Scientific
10361012
Cell lineSf9ATCCCRL-1711RRID:CVCL_0549
Recombinant
DNA reagent
rabbit TRPV2GenscriptPubmed Accession
No. XM_017349044
Recombinant
DNA reagent
Bac-to-Bac
Baculovirus Expression
System
ThermoFisher
Scientific
10359016
Recombinant
DNA reagent
MSP2N2 scaffold
protein
Stephen Sligar
laboratory
Addgene:Cat#29520PMID:20817758
Chemical
compound, drug
n-dodecyl-β-d-
maltopyranoside(DDM)
AnatraceD310
Chemical
compound, drug
Cholesteryl
Hemisuccinate
AnatraceCH210
Chemical
compound, drug
Amphipol A8-35AnatraceA835
Chemical
compound, drug
TRISFisher ScientificBP152
Chemical
compound, drug
NaClFisher ScientificS271
Chemical
compound, drug
CaCl2Fisher ScientificC70
Chemical
compound, drug
leupeptinGoldBioL-010
Chemical
compound, drug
pepstatinGoldBioP-020
Chemical
compound, drug
aprotininGoldBioA-655
Chemical
compound, drug
DNase IGoldBioD-301
Chemical
compound, drug
β-mercapto ethanolSigma AldrichM3148
Chemical
compound, drug
PMSFSigma AldrichP7626
Chemical
compound, drug
anti-FLAG resinSigma AldrichA4596
Chemical
compound, drug
ResiniferatoxinSigma AldrichR8756
Chemical
compound, drug
Bio-Beads SM-2BioRad152–8920
Chemical
compound, drug
1,2-dimyristoyl-
sn-glycero-3-
phosphocholine
Avanti Polar Lipids850345P
Chemical
compound, drug
1-palmitoyl-2-
oleoyl-sn-glycero-3-
phosphocholine (POPC)
Avanti Polar Lipids850457C
Chemical
compound, drug
1-palmitoyl-2-oleoyl
-sn-glycero-3-
phosphoethanolamine (POPE)
Avanti Polar Lipids850757C
Chemical
compound, drug
1-palmitoyl-2-oleoyl-
sn-glycero-3-phospho-
(1'-rac-glycerol) (POPG)
Avanti Polar Lipids840457C
OtherWhatman No.
one filter paper
Sigma AldrichWHA1001325
OtherUltrAuFoil R1.2/1.3
300-mesh grid
Electron Microscopy
Sciences
Q350AR13A
Software,
algorithm
MotionCor2Zheng et al. (2017)http://msg.ucsf.edu/em/software/motioncor2.htmlRRID:SCR_016499
Software,
algorithm
GCTFZhang (2016)https://www.mrc-lmb.cam.ac.uk/kzhang/RRID:SCR_016500
Software,
algorithm
RELION 3.0Zivanov et al. (2018)https://www2.mrc-lmb.cam.ac.uk/relion/RRID:SCR_016274
Software,
algorithm
CootEmsley and Cowtan (2004)https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/RRID:SCR_014222
Software,
algorithm
PhenixAdams et al. (2010)http://phenix-online.org/RRID:SCR_014224
Software,
algorithm
MolprobityChen et al. (2010)http://molprobity.biochem.duke.edu/index.phpRRID:SCR_014226
Software,
algorithm
UCSF ChimeraPettersen et al. (2004)https://www.cgl.ucsf.edu/chimera/RRID:SCR_004097
Software,
algorithm
PymolShrödinger LLChttps://pymol.org/2/RRID:SCR_000305
OtherCryo-electron microscopy
structure of rabbit TRPV2
ion channel
Zubcevic et al. (2018b)PDB ID 5AN8PMID:26779611
OtherCryo-electron microscopy
structure of rabbit TRPV2
ion channel
Zubcevic et al. (2018a)EMDB ID EMD-6455PMID:26779611
OtherCrystal structure of the
TRPV2 ion channel in
complex with RTx
Zubcevic et al. (2018a)PDB ID 6BWJPMID:29728656

Protein expression and purification

Request a detailed protocol

The construct for the RTx sensitive, full-length rabbit TRPV2 (TRPV2RTx) was prepared by introducing four point mutations (F470S, L505M, L508T and Q528E) into the synthesized full-length rabbit TRPV2 gene (Zhang et al., 2016). The construct was cloned into a pFastBac vector with a C-terminal FLAG affinity tag and used for baculovirus production according to manufacturers’ protocol (Invitrogen, Bac-to-Bac). The protein was expressed by infecting Sf9 cells with baculovirus at a density of 1.3 M cells ml−1 and incubating at 27oC for 72 hr in an orbital shaker. Cell pellets were collected after 72 hr and resuspended in buffer A (50 mM TRIS pH8, 150 mM NaCl, 2 mM CaCl2, 1 μg ml−1 leupeptin, 1.5 μg ml−1 pepstatin, 0.84 μg ml−1 aprotinin, 0.3 mM PMSF, 14.3 mM β-mercaptoethanol, and DNAse I) and broken by sonication (3 × 30 pulses).

For the amphipol-reconstituted TRPV2 (TRPV2RTx-APOL) sample, the lysate was supplemented with 40 mM Dodecyl β-maltoside (DDM, Anatrace), 4 mM Cholesteryl Hemisuccinate (CHS, Anatrace) and 2 μM RTx and incubated at 4oC for 1 hr. Insoluble material was removed by centrifugation (8000 g, 30 min), and anti-FLAG resin was added to the supernatant for 1 hr at 4oC.

After binding, the anti-FLAG resin was loaded onto a Bio-Rad column and a wash was performed with 10 column volumes of buffer B (50 mM TRIS pH8, 150 mM NaCl, 2 mM CaCl2, 1 mM DDM, 0.1 mM CHS, 0.1 mg ml−1 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC, Avanti Polar Lipids), 2 μM RTx) before elution in five column volumes of buffer C (50 mM TRIS pH8, 150 mM NaCl, 2 mM CaCl2, 1 mM DDM, 0.1 mM CHS, 0.1 mg ml−1 DMPC, 2 μM RTx, 0.1 mg ml−1 FLAG peptide).

The eluate was concentrated and further purified by gelfiltration on a Superose six column. The peak fractions were collected, mixed with Amphipol A8-35 (Anatrace) in a 1:10 ratio and incubated for 4 hr at 4oC. Subsequently, Bio-Beads SM-2 (Bio-Rad) were added to a 50 mg ml−1 concentration and incubated at 4oC overnight to remove detergent.

After reconstitution, the protein was subjected to a second round of gelfiltration on a Superose six column in buffer D (50 mM TRIS pH8, 150 mM NaCl, 2 μM RTx), the peak fractions were collected and concentrated to 2–2.5 mg ml−1 for cryo-EM.

For the nanodisc reconstituted TRPV2 (TRPV2RTx-ND), the lysate was supplemented with 40 mM Dodecyl β-maltoside (DDM, Anatrace) and 2 μM RTx and incubated at 4o C for 1 hr. The solution was cleared by centrifugation (8000 g, 30 min), and anti-FLAG resin was added to the supernatant for 1 hr at 4oC.

After binding, the anti-FLAG resin was loaded onto a Bio-Rad column and a wash was performed with 10 column volumes of buffer BnoCHS (50 mM TRIS pH8, 150 mM NaCl, 2 mM CaCl2, 1 mM DDM, 0.1 mg ml−1 DMPC, 2 μM RTx) before elution in five column volumes of buffer CnoCHS (50 mM TRIS pH8, 150 mM NaCl, 2 mM CaCl2, 1 mM DDM, 0.1 mg ml−1 DMPC, 2 μM RTx, 0.1 mg ml−1 FLAG peptide).

The eluate from the anti-FLAG resin was concentrated to ~500 μl. A 10 mg ml−1 3:1:1 mixture of lipids 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1'-rac-glycerol) (POPG) was dried under argon, resuspended in 1 ml 50 mM Tris pH8, 150 mM NaCl and clarified by extrusion, before being incubated for 1 hr with 10 mM DDM. The membrane scaffold protein MSP2N2 was prepared as previously described (Ritchie et al., 2009). The concentrated TRPV2 was combined with MSP2N2 and the prepared lipid mixture in a 1:3:200 ratio and incubated at 4oC for 1 hr. After the initial incubation, 50 mg ml−1 Bio-Beads SM-2 were added and the mixture was incubated for another hour at 4oC, following which the reconstitution mixture was transferred to a fresh batch of Bio-Beads SM-2 at 50 mg ml−1 and incubated overnight at 4oC. Finally, the reconstituted channels were subjected to gelfiltration on Superose six in buffer D, the peak fractions collected and concentrated to 2–2.5 mg ml−1 for cryo-EM.

Cryo-EM sample preparation

Request a detailed protocol

TRPV2RTx-APOL and TRPV2RTx-ND were frozen using the same protocol. Before freezing, the concentrated protein sample was supplemented with 300 μM RTx and incubated ~30 min at 4°C. 3 μl sample was dispensed on a freshly glow discharged (30 s) UltrAuFoil R1.2/1.3 300-mesh grid (Electron Microscopy Sciences), blotted for 3 s with Whatman No. one filter paper using the Leica EM GP2 Automatic Plunge Freezer at 23°C and >85% humidity and plunge-frozen in liquid ethane cooled by liquid nitrogen.

Cryo-EM data collection

Request a detailed protocol

Data for both TRPV2RTx-APOL and TRPV2RTx-ND were collected using the Titan Krios transmission electron microscope (TEM) operating at 300 keV using a Falcon III Direct Electron Detector operating in counting mode at a nominal magnification of 75,000x corresponding to a physical pixel size of 1.08 Å/pixel.

For the TRPV2RTx-APOL 1293 movies (30 frames/movie) were collected using a 60 s exposure with an exposure rate of ~0.8 e-/pixel/s, resulting in a total exposure of 42 e-/Å (Cao et al., 2013a) and a nominal defocus range from −1.25 µm to −3.0 µm.

For TRPV2RTx-ND, 2254 movies were collected (30 frames/movie) with 60 s exposure and exposure rate of ~0.8 e-/pixel/s. The total exposure was of 42 e-/Å (Cao et al., 2013a) and a nominal defocus range from −1.25 µm to −3.0 µm.

Reconstruction and refinement

Request a detailed protocol

TRPV2RTx-APOL MotionCor2 (Zheng et al., 2017) was used to perform motion correction and dose-weighting on 1293 movies. Unweighted summed images were used for CTF determination using GCTF (Zhang, 2016). Following motion correction and dose-weighting and CTF determination, micrographs which contained Figure of Merit (FoM) values of <0.12 and astigmatism values > 400 were removed, leaving 1207 micrographs for further analysis. An initial set of 1660 particles was picked manually and subjected to reference-free 2D classification (k = 12, T = 2) which was used as a template for automatic particle picking from the entire dataset (1207 micrographs). This yielded a stack of 580,746 particles that were binned 4 × 4 (4.64 Å/pixel, 64 pixel box size) and subjected to reference-free 2-D classification (k = 58, T = 2) in RELION 3.0 (Zivanov et al., 2018). Classes displaying the most well-defined secondary structure features were selected (470,760 particles) and an initial model was generated from the 2D particles using the Stochastic Gradient Descent (SGD) algorithm as implemented in RELION 3.0. 3D auto-refinement in RELION 3.0 was performed on the 470,760 particles with no symmetry imposed (C1), using the initial model, low-pass filtered to 30 Å, as a reference map. This resulted in an 8.9 Å 3D reconstruction, which was then used for re-extraction and re-centering of 2 × 2 binned particles (2.16 Å/pixel, 128 pixel box size). 3D classification (k = 4, T = 8) without imposed symmetry (C1) was performed on the extracted particles, using a soft mask calculated from the full molecule. Classes 2–4 (90,862, 109,623 and 101,570 particles, respectively) all possessed well-defined secondary structure, but visual inspection of the maps suggested that the classes represented distinct conformational states. Therefore, each class was processed separately. For each class, the particles were extracted and unbinned (1.08 Å/pixel, 256 pixel box size), and soft masks calculated. 3D auto-refinement of the individual classes without symmetry imposed (C1) yielded 4.7 Å (class 2), 3.6 Å (class 3) and 3.2 Å (class 4) 3D reconstructions. Inspection of these volumes revealed that classes 2 and 3 adopted two-fold (C2) symmetry, while class four was four-fold symmetric (C4). Particles from class 2 were subjected to particle movement and dose-weighting using the ‘particle polishing’ function as implemented in RELION 3.0. The shiny particles were input into 3D auto-refinement with a soft mask and C2 symmetry applied, resulting in a 4.19 Å reconstruction (TRPV2RTx-APOL 3). Similarly, particles from class 3 were subjected to polishing, and the following 3D auto-refinement with a soft mask and C2 symmetry applied resulted in a 3.3 Å final reconstruction (TRPV2RTx-APOL2). Particles from class 4 were first subjected to CTF refinement using the ‘CTF refine’ feature in RELION 3.0. Particle polishing was then performed, followed by 3D auto-refinement with a soft mask and C4 symmetry applied, yielding a 2.91 Å reconstruction (TRPV2RTx-APOL 1). All resolution estimates were based on the gold-standard FSC 0.143 criterion (Scheres and Chen, 2012; Chen et al., 2013).

TRPV2RTx-ND The 2254 collected movies were subjected to motion correction and dose-weighting (MotionCor2) and CTF estimation (GCTF) in RELION 3.0. Micrographs with FoM values < 0.13 and astigmatism values > 400 were removed, resulting in a selection of 1580 good micrographs. From these, 2015 particles were picked manually, extracted (without binning, 1.08 Å/pixel, 256 pixel box size) and subjected to reference-free 2D classification (k = 12, T = 2) that was used as a template for autopicking. This resulted in a 1,407,292 stack of particles that were binned 4 × 4 (4.32 Å/pixel, 64 pixel box size) and subjected to reference-free 2D classification (k = 100, T = 2). Classes exhibiting the most well-defined secondary structure features were selected, resulting in 482,602 particles. These were re-extracted (2 × 2 binned, 2.16 Å/pixel, 128 pixel box size) and put into 3D auto-refinement, using the previously obtained map of apo TRPV2 (EMD-6455) filtered to 30 Å as a reference with no symmetry applied (C1). The 3D auto-refinement yielded a 5.4 Å reconstruction. The particles were then subjected to 3D classification (k = 6, T = 8), with a soft mask and the 5.4 Å volume as a reference without imposed symmetry (C1). Only two of the six classes (classes 1 and 6) contained significant density in the TM domains. They were selected (112,622 particles), re-extracted, re-centered and unbinned (1.08 Å/pixel, 256 pixel box size) before being input into 3D auto-refinement without symmetry imposed (C1) and with a soft mask and the previous 5.4 Å reconstruction filtered to 30 Å as a reference. The 3D auto-refinement resulted in a 4.12 Å map, which was then subjected to Bayesian particle polishing. 3D auto-refinement was then performed on the resulting shiny particles with no symmetry applied (C1), resulting in a 4 Å reconstruction. The particles were then subjected to CTF refinement, yielding a 3D reconstruction resolved to 4 Å (C1). However, visual inspection of the map revealed a strong tendency towards two-fold symmetry. Therefore, 3D auto-refinement was repeated with C2 symmetry applied, resulting in a map resolved to 3.84 Å as estimated by gold-standard FSC 0.143 criterion.

Model building

Request a detailed protocol

The TRPV2RTx-APOL and TRPV2RTx-ND models were built into the cryo-EM electron density map in Coot (Emsley and Cowtan, 2004), using the structures of TRPV2 (PDB 5AN8 and 6BWJ) as templates. The structures were real-space refined in Coot, and iteratively refined using the phenix.real_space_refine as implemented in the Phenix suite (Adams et al., 2010). Structures were refined using global minimization and rigid body, with high weight on ideal geometry and secondary structure restraints. The Molprobity server (Chen et al., 2010) (http://molprobity.biochem.duke.edu/) was used to identify problematic areas, which were subsequently manually rebuilt. The radius of the permeation pathways was calculated using HOLE (Smart et al., 1996). All analysis and structure illustrations were performed using Pymol (The PyMOL Molecular Graphics System, Version 2.0) and UCSF Chimera (Pettersen et al., 2004).

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
    Coot: model-building tools for molecular graphics
    1. P Emsley
    2. K Cowtan
    (2004)
    Acta Crystallographica Section D Biological Crystallography 60:2126–2132.
    https://doi.org/10.1107/S0907444904019158
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
    TRP channels and pain
    1. D Julius
    (2013)
    Annual Review of Cell and Developmental Biology 29:355–384.
    https://doi.org/10.1146/annurev-cellbio-101011-155833
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59

Decision letter

  1. Kenton Jon Swartz
    Reviewing Editor; National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
  2. Richard Aldrich
    Senior Editor; The University of Texas at Austin, United States
  3. Richard K Hite
    Reviewer; Memorial Sloan Kettering Cancer Center, United States
  4. Raimund Dutzler
    Reviewer; University of Zürich, Switzerland

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Symmetry transitions during gating of the TRPV2 ion channel in lipid membranes" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Kenton Swartz as the Reviewing Editor and Richard Aldrich as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Richard K Hite (Reviewer #3); Raimund Dutzler (Reviewer #4).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

In this report by Zubcevic and colleagues, single-particle cryo-EM structures of TRPV2 in amphipols and lipid nanodiscs in complex with the agonist resiniferatoxin (RTx) are presented. Three-dimensional classification of the amphipol-reconstituted channels reveal three distinct conformations, two of which display C2 symmetry and a third which displays C4 symmetry. Classification of the nanodisc-reconstituted channels reveal just a single conformation which displays C2 symmetry. Previously, the same group had reported a ligand-free, closed cryo-EM structure of TRPV2 which displayed C4 symmetry and a RTx-bound C2 symmetric crystal structure with an open selectivity filter gate that displayed two-fold symmetry and a closed common gate that displayed four-fold symmetry. Comparison of the ampiphol-reconstituted structures with the previous structures indicated that in all three conformations the TM domain adopts a nearly four-fold symmetric structure that resembles the ligand-free, closed state, while the ARD is flexible and adopts a range of different conformations. In contrast, large conformational changes are seen throughout the nanodisc-reconstituted structure. The selectivity filter gate is opened in a 2-fold symmetric fashion by alternating rotations of the pore helices in adjacent subunits. The common gate also adopts a 2-fold symmetric conformation, with two of the subunits adopting a closed state and two adopting an open state. Comparison of the nanodisc-reconstituted sample with the previous RTx-bound crystal structure suggested that they represent distinct conformational states with the crystal structure which contained a fully closed common gate potentially representing an earlier state in the transition from closed to an open state. This work provides evidence that RTx binding to TRPV2 induces two-fold symmetric pre-activated states in a lipid bilayer environment and will be of general interest. We request that you address the following issues in revision.

Essential revisions:

1) Some of the conformational differences observed between the structures or between subunits from a single structure are small (see Figure 3—figure supplement 1C, D; Figure 3—figure supplement 2B, C), or reflect changes in the patterns of hydrogen bonding such as the α- to π-helix transitions. Given that the resolution of the structures is not very high, the authors should provide structural comparisons including the experimental density maps to support their proposed conformational differences, rather than simply using the structural models. The density maps should be used to provide stronger evidence that the pattern of main-chain hydrogen bonds has indeed been disrupted in the S4-S5 and S6 helices to give rise to π-helical bulges. As symmetry is a major feature of this work, it would helpful to compare the model refined using the symmetrized reconstructions in density maps calculated without symmetry. This is particularly important for the selectivity filter and common gates of the four subunits in the nanodisc-reconstituted reconstruction.

2) Why is map-to-model fit so poor for TRPV2RTx-APOL1? The map-to-model fit has a 0.5 correlation at approximately 4Å while the FSC between half-maps is approximately 2.9 Å. How was this model refined into the density map? What does nominal resolution mean? Why is it different from FSC=0.143 in Table 1 and why are the unmasked resolution estimates higher than the masked resolution estimates? Any masks used for classification or refinement should be displayed in the figure supplements.

3) The authors suggest that the ampiphol restrict the flexibility of the TM domain. It would be helpful to compare the belt of ampiphol-reconstituted samples with those of the nanodisc-reconstituted samples to see if sufficient space is available for the channel to adopt the conformation of the nanodisc-reconstituted channels. Were lipid molecules resolved in the nanodisc structures that may provide the flexibility required to adopt the different conformation?

4) It would be interesting to know how many particles contribute to the different 3D classes displayed Figure 1—figure supplement 2. Do the particles in class 1 shown in Figure 1—figure supplement 1 really display and unsymmetrical (C1) channel and can we learn anything from this low-resolution structure. While it is possible that the X-ray structure of TRPV2-RTx complex could be an intermediate on the activation pathway, the authors might also consider the possibility that the pronounced C2-symmetric features observed in this structure would reflect the conditions in the crystalline environment and might thus not be an intermediate on the activation path.

5) Even though RTx activates both TRPV1 and TRPV2 with an extremely high affinity and efficacy, the kinetics of activation are distinctly slow, even when compared to other vanilloids with lower affinity such as capsaicin in the case of TRPV1. This raises the possibility that RTx can interact with the pocket in multiple orientations, with only some of them leading to channel activation. At present, it is not possible to determine whether the structures observed in this study are indeed reflecting RTx-channel complexes that are components of the activation pathway, or whether they reflect off-pathway intermediates that are not conducing to activation. It is intriguing that no open state has been determined even though RTx activates TRPV2-QM with very high affinity and efficacy. Are the binding poses of RTx inferred from the densities different in all structures? Are they different between subunits in a given structure?

6) The text makes constant reference to a selectivity filter activation gate in these channels. However, there is currently no functional evidence indicating that the selectivity filter does indeed function as an activation gate. This should be reflected in the text.

7) In several points throughout the manuscript (subsection “RTx induces a break in symmetry in TRPV2RTx-ND”, last paragraph), the authors mention that their previous work demonstrated that a network of hydrogen bonds between the pore helix, the S5 and the S6 is essential for activation. However, the results in the previous publication do not provide conclusive evidence for the involvement of hydrogen bonds, but instead show that alanine mutations in that region affect cation permeation through an unknown mechanism that could involve a network of hydrogen bonds. This should be clearly stated in the manuscript.

8) In the third paragraph of the Discussion, the authors mention that their previous work has shown that C2 symmetric states are critical for permeation of large organic cations. We think it is more accurate to state that their data suggests that C2 symmetry could be involved in channel activation by RTx, but it does not provide conclusive evidence that C2 symmetries are directly connected to channel activation or to the mechanism of cation permeation in an open channel. No experiments were done in that study to directly probe the functional consequences of the proposed asymmetries in the tetrameric channel.

https://doi.org/10.7554/eLife.45779.046

Author response

Essential revisions:

1) Some of the conformational differences observed between the structures or between subunits from a single structure are small (see Figure 3—figure supplement 1C, D; Figure 3—figure supplement 2B, C), or reflect changes in the patterns of hydrogen bonding such as the α- to π-helix transitions. Given that the resolution of the structures is not very high, the authors should provide structural comparisons including the experimental density maps to support their proposed conformational differences, rather than simply using the structural models. The density maps should be used to provide stronger evidence that the pattern of main-chain hydrogen bonds has indeed been disrupted in the S4-S5 and S6 helices to give rise to π-helical bulges.

We have now included figure supplements showing the density around the π-helices in all structures (Figure 3—figure supplement 4). However, we agree with the reviewer that the resolution of the TRPV2RTx-ND of the map is not very high enough to define π-helix on S6 unambiguously and thus we have removed references in the text and figures to the S6 π-helix in structure throughout the text, and refer them to as bent helices instead.

As symmetry is a major feature of this work, it would helpful to compare the model refined using the symmetrized reconstructions in density maps calculated without symmetry. This is particularly important for the selectivity filter and common gates of the four subunits in the nanodisc-reconstituted reconstruction.

We thank reviewers for the great suggestion. During processing, we merged classes 1 and 6, which displayed a high overall similarity (CC=0.98) and then proceeded with refinement with C1 imposed. Because visual inspection indicated that this reconstruction possesses two-fold symmetry, we further refined the combined particles with C2 symmetry imposed. We have now also calculated the symmetry of the map combining classes 1 and 6 using the “Map Symmetry” function implemented in the Phenix suite. The best solution is C2 symmetry (score 1.28, CC=0.90). (Score reflects the square root of the number of elements in the symmetry multiplied by the map correlation for that symmetry, so the max score for C2 symmetry is 1.414 and 2 is the max score for C4 symmetry). We also calculated the score for C4 symmetry (score=1.6, CC=0.80) which indicates that C2 symmetry is more appropriate for this map. We have now also made a remark in the manuscript describing this analysis (subsection “RTx induces a break in symmetry in TRPV2RTx-ND”, first paragraph).

For the purpose of this review, we have now included an FSC curve which estimates the fit of the model, originally built into the C2 map, into the C1 reconstruction. In addition, we have now also performed separate refinements of classes 1 and 6 with no symmetry imposed and estimated the fit of the TRPV2RTx-ND atomic model, built into the C2 symmetry, to these non-symmetrized maps. This data is shown in Figure 1—figure supplement 3. Figure 1—figure supplement 3 now also shows the density around the critical regions (pore helices and S6 helices) in the class 1+class 6 non-symmetrized map. A mention of this is now included in the manuscript (see the second paragraph of the aforementioned subsection).

2) Why is map-to-model fit so poor for TRPV2RTx-APOL1? The map-to-model fit has a 0.5 correlation at approximately 4Å while the FSC between half-maps is approximately 2.9 Å. How was this model refined into the density map? What does nominal resolution mean? Why is it different from FSC=0.143 in Table 1 and why are the unmasked resolution estimates higher than the masked resolution estimates? Any masks used for classification or refinement should be displayed in the figure supplements.

We thank the reviewers for pointing this out. A mistake was made in the plotting of the APOL FSC curves during our initial manuscript preparation. This has now been corrected and the new FSC curves placed in Figure 1—figure supplement 1.

3) The authors suggest that the ampiphol restrict the flexibility of the TM domain. It would be helpful to compare the belt of ampiphol-reconstituted samples with those of the nanodisc-reconstituted samples to see if sufficient space is available for the channel to adopt the conformation of the nanodisc-reconstituted channels. Were lipid molecules resolved in the nanodisc structures that may provide the flexibility required to adopt the different conformation?

We have now included a figure (Figure 4—figure supplement 6) showing the amphipol and nanodisc clouds surrounding the channels in the two different reconstitution conditions. We used the common nanodisc scaffold protein MSP2N2, which has been successfully used in many structural studies of TRP channels. MSP2N2 has an estimated diameter of ~17 nm and is large enough to accommodate TRPV2. The size of nanodisc is largely defined by the scaffold protein while amphipol polymers surround membrane proteins in a way that is similar to detergents. The quality of EM density for the nanodisc cloud is low and neither the scaffold protein nor the bound lipid molecules are well resolved. It is therefore difficult to say whether the lipids stabilize the captured conformation. However, a side-by-side comparison of the amphipol and nanodisc structures indicates that the nanodisc adopts a distinctly oval shape to accommodate the TRPV2-RTx complex, while amphipol retains a squarer form.

4) It would be interesting to know how many particles contribute to the different 3D classes displayed Figure 1—figure supplement 2.

This has now been included in the figure.

Do the particles in class 1 shown in Figure 1—figure supplement 1 really display and unsymmetrical (C1) channel and can we learn anything from this low-resolution structure.

We did indeed try to refine this class, but the resolution never got past ~7 Å. We therefore did not feel comfortable presenting this data or commenting on its symmetry and conformational state. We have now included a sentence stating this in the manuscript.

While it is possible that the X-ray structure of TRPV2-RTx complex could be an intermediate on the activation pathway, the authors might also consider the possibility that the pronounced C2-symmetric features observed in this structure would reflect the conditions in the crystalline environment and might thus not be an intermediate on the activation path.

This point was addressed in the paper published in NSMB (Zubcevic et al., 2018). In brief, we conducted electrophysiological experiments to probe the validity of the structure and found that the interactions that were uniquely found in the TRPV2-RTx crystal structure were of critical importance for channel function and that the structure was therefore not an artifact of crystallization. We also discussed the possibility of crystallization artifacts in the Introduction of the current manuscript.

5) Even though RTx activates both TRPV1 and TRPV2 with an extremely high affinity and efficacy, the kinetics of activation are distinctly slow, even when compared to other vanilloids with lower affinity such as capsaicin in the case of TRPV1. This raises the possibility that RTx can interact with the pocket in multiple orientations, with only some of them leading to channel activation. At present, it is not possible to determine whether the structures observed in this study are indeed reflecting RTx-channel complexes that are components of the activation pathway, or whether they reflect off-pathway intermediates that are not conducing to activation. It is intriguing that no open state has been determined even though RTx activates TRPV2-QM with very high affinity and efficacy. Are the binding poses of RTx inferred from the densities different in all structures? Are they different between subunits in a given structure?

We have observed that RTx can bind to TRPV2 in different modes, which we believe contributes significantly to the observed C2 symmetry. We have addressed this point in our previous publication as well as this study, but our clarification may have been somewhat insufficient. We have now included a figure supplement to show the binding poses of RTx in the different structures as well as in the subunits of different structures. In brief, in the C4 TRPV2RTx-APOL 1 structure, the RTx binding poses are the same in all subunits. They are also nearly identical in TRPV2RTx-APOL 2. However, they differ from the ones captured in the C4 structure. In both TRPV2RTx-APOL 2 andTRPV2RTx-ND,the binding poses are different within the subunits and they also differ from the poses observed in both TRPV2RTx-APOL 1 and TRPV2RTx-APOL 2. Nevertheless, given the resolution of TRPV2RTx-APOL 3 and TRPV2RTx-ND, the position of RTx in these structures should be taken with caution.

The point concerning the slow activation kinetics of RTx is excellent. In order to allow ample time between adding RTx to freezing the grids, we included RTx in our buffers from the very beginning of the purification. This is in contrast to the Julius-Cheng TRPV1/RTx/DkTx complex where the toxin was added to the sample ~30 minutes before freezing.

The point questioning whether the RTx-channel complexes observed here are components of the activation pathway is equally valid. The Swartz lab study that originally transplanted RTx-sensitivity into TRPV2 (Zhang et al., 2016) conducted recordings of single RTx-activated TRPV2 channels and found that these can achieve a high (~0.9) open probability. We hypothesize that the absence of a fully open state may point to the absence of unknown factors in our experimental conditions (e.g. the presence of the membrane bilayer rather than nanodisc). This is in line with observations made by many groups that agonist-bound ion channel structures often do not adopt fully open states.

However, we believe that the absence of the open state does not dismiss the rest of our findings as “off-pathway” conformations – the conformational changes that we observe in both the amphipol and nanodisc structures are in line with conformations associated with activation of closely related channels: the ARD rotation seen in TRPV1 (Cao et al., 2013; Liao et al., 2013) and the α-to-π transition in the S4-S5 linker in TRPV2 (Zubcevic et al., 2018). Notably, voltage-gated cation channels can assume multiple closed states before opening, and it is possible that TRPV2 channels also traverse a number of intermediate non-conducting states before entering the open conformation. It’s also worth pointing out that all of our data on the TRPV2/RTx complex, both crystallographic and cryo-EM, contains the same hot spots for conformational change: the S4-S5 linker and the pore helix. By contrast, pore of related channels TRPV3 and TRPV6 do not exhibit any change in the conformation of the pore helix during gating. And in addition, the pore of TRPV6 is preserved even in the structure of the point mutant which induces a non-domain swapped conformation in the channel, indicating that in TRPV6 the pore is a stable unit- as expected for a highly selective channel. In light of this, our TRPV2/RTx data points to the inherent flexibility of the pore helix in this channel.

Furthermore, we believe it unlikely that two-fold symmetry is an experimental artifact, as it’s only been observed in a few select TRP channels. Crystal and cryo-EM structures of K+ channels have never observed two-fold symmetry in the channel pores, and even the recent cryo-EM study of the KV channel in nanodiscs (Matthies et al., 2018) is identical to the four-fold symmetric crystal structure solved by the MacKinnon lab. Along the same lines, the structures of Ca2+-selective TRPV5 and TRPV6, which have been solved in many different conditions (e.g. crystallization, cryo-EM, agonist and antagonist-bound, detergent, nanodisc, amphipol, mutations) have never been captured in non-C4 symmetric arrangements. These, like K+ channels, possess seemingly rigid pore structures (meaning extensive interactions between pore helices and the rest of pore), which is consistent with their observed strong preference for four-fold symmetry.

We think that the ability of TRPV2 to enter different two- and four-fold symmetric states is a reflection of the intrinsic flexibility in the TRPV2 channel pore and especially the pore helix. As we discussed in our previous paper, we think that this flexibility might be unique to TRPV1 and TRPV2 amongst TRPV channels and that it may endow them with the ability to permeate large organic cations.

At its core, the question posed here applies universally to all structural data: how well do cryo-EM or crystallization conditions sample the conformational landscape of proteins in their native environments? In this manuscript, we have evaluated our cryo-EM structures based our structural and functional data as well as the available literature and proposed a hypothetical sequence of events that may lead to TRPV2 channel opening.

6) The text makes constant reference to a selectivity filter activation gate in these channels. However, there is currently no functional evidence indicating that the selectivity filter does indeed function as an activation gate. This should be reflected in the text.

That is an excellent point for discussion. To make our long answer short, we think that the SF acts as an activation gate in TRPV1 and TRPV2, but not in other TRPV members (TRPV3-TRPV6). The references to the “selectivity filter gate” originated with the very first TRPV1 structure which showed that the channel possesses two restrictions in the non-conductive state: the methionine in the SF which seals the selectivity filter (dubbed the “upper gate”) and the isoleucine in S6 which creates a hydrophobic seal at the common gate (dubbed the “lower gate”). In the RTx/DkTx-bound TRPV1, both of these restrictions are removed: the common gate opens via a bend in the S6 and the SF via a tilt of the pore helix. While it is possible that the SF opening is due to DkTx binding or simply reflects a conformational change that is coupled to the opening of the “lower gate”, the structural data points to the existence of the SF gate. We agree with the reviewer that the functional evidence for the SF activation gate is lacking. From a structural standpoint, we reason that TRPV2, like TRPV1, possesses a SF activation gate: the architecture of TRPV2 is very similar to TRPV1 and shows two restrictions in the closed state and binding of RTx induces an open conformation of the SF gate (that the pore helices tilt significantly and move away from the ion permeation pathway).

However, we think that other TRPV channels (TRPV3-6) only possess one restriction in their pores: the common gate. Structural studies of these channels have not observed significant movement around the PH or the SF during gating (e.g. TRPV3), which is consistent with the absence of a SF gate. We have now included a paragraph (subsection “RTx induces a break in symmetry in TRPV2RTx-ND”, last paragraph) to address this point.

7) In several points throughout the manuscript (subsection “RTx induces a break in symmetry in TRPV2RTx-ND”, last paragraph), the authors mention that their previous work demonstrated that a network of hydrogen bonds between the pore helix, the S5 and the S6 is essential for activation. However, the results in the previous publication do not provide conclusive evidence for the involvement of hydrogen bonds, but instead show that alanine mutations in that region affect cation permeation through an unknown mechanism that could involve a network of hydrogen bonds. This should be clearly stated in the manuscript.

We believe that our reference to the hydrogen bond network between the threonine in the pore helix and tyrosine residues in S5 and S6 is chemically sound. The donor/acceptor are within 3Å of each other in our 3.1 Å crystal structure, and therefore classified as a hydrogen bond of moderate strength (like most hydrogen bonds in proteins). While it is true that hydrogens are not resolved at ~3 Å, this assumption is widely used in structural biology (i.e. the hydrogen bonding in α-helices).

No other type of major interaction can exist between threonine and tyrosine side chains that are within 3Å beside hydrogen bonding. It is true that we performed alanine mutagenesis studies our previous studies (Zubcevic et al., 2018) instead of more subtle mutations such as threonine to valine (isosteric mutation) or tyrosine to phenylalanine (hydroxyl group removal). However, if we apply the same strict standards, introducing above-mentioned mutations to probe H-bonding would also be insufficient evidence for the involvement of H-bonding, and one would at least need to conduct an unnatural amino acid mutagenesis study of TRPV2 to properly probe the H-bonds. In response to the reviewer’s request, throughout the text, we have clarified this point by stating that the interaction between the Thr and Tyr residues (which includes H-bonds), is important for activation.

8) In the third paragraph of the Discussion, the authors mention that their previous work has shown that C2 symmetric states are critical for permeation of large organic cations. We think it is more accurate to state that their data suggests that C2 symmetry could be involved in channel activation by RTx, but it does not provide conclusive evidence that C2 symmetries are directly connected to channel activation or to the mechanism of cation permeation in an open channel. No experiments were done in that study to directly probe the functional consequences of the proposed asymmetries in the tetrameric channel.

This is a great point to bring up, as it allows us to explain our reasoning in more detail. In our NSMB paper we performed a series of experiments to address the role of two-fold symmetry. Our structures showed that RTx-bound TRPV2 adopts a two-fold symmetric state, where the SF takes on two distinct conformations: a narrow conformation, which coordinates Ca2+ ions, and a wide conformation, which is wide enough to permeate YO-PRO-1. In this structure, an interaction exits between the pore helix residue Thr602 and S5 helix residue Tyr542 in the subunits with the narrow SF conformation, but it is absent in the subunits in the wide conformation. This interaction is also absent in the ligand-free cryo-EM structure of TRPV2 in C4 symmetry, which adopts a closed conformation of the SF gate. We showed that eliminating the Thr602-Tyr542 interaction, which exists only in the 2-fold symmetric structure, results in channels that have normal Na+ currents but are unable to conduct YO-PRO-1. In addition, mutations Thr602A and Tyr542A both decrease the NMDG+/Na+ permeability ratio (these experiments were done in inside-out patches to avoid ion accumulation artifacts). This was also shown to be true for the corresponding residues in TRPV1. Interestingly, the Thr602-Tyr542 interaction is absent from all four subunits of the RTx/DkTx-bound TRPV1, consistent with fact that the pore of RTX/DkTx-bound TRPV1, due to its narrow profile, is incompatible with permeation of large organic cations.

These experiments showed that:

1) Disruption of the interactions between the SF and the S5 preferentially reduce the permeation of large organic cations;

2) This interaction has only been observed in the channel with C2-symmetry;

3) Intriguingly, this interaction exists in the contracted subunit, not in the widened subunit, further supporting the involvement of reduced symmetry in the ability of the channel to permeate large organic cations.

Therefore, based on our data we believe that reduced symmetry is involved in large organic cation permeation. In our manuscript, we have now changed the text to state that “our previous studies suggest that C2 symmetric states play an important role for permeation of large organic cations”.

Two-fold symmetry has now been observed in a number of TRP channels in a variety of different experimental conditions, but almost always in the presence of agonists: TRPV3 (Zubcevic et al., 2018), TRPV2 (Zubcevic et al., 2018; Pumroy et al., 2019) and TRPM2 (Yin et al., 2019). We believe that further exploration of these experimentally observed symmetry transitions may help us refine our understanding of TRP(V) channel activation and physiology.

References:

Matthies D, Bae C, Toombes GES, Fox T, Bartesaghi A, Subramaniam S, Swartz KJ. Single-particle cryo-EM structure of a voltage-activated potassium channel in lipid nanodiscs. eLife. 2018. doi: 10.7554/eLife.37558.

Pumroy RA, Samanta A, Liu Y, Hughes TET, Zhao S, Yudi Y, Huynh KW, Zhou ZH, Rohacs T, Han S, Moiseenkova-Bell VY. Molecular mechanism of TRPV2 channel modulation by cannabidiol. bioRxiv, May 24, 2019. Doi: 10.1101/521880

Yin Y, Wu M, Hsu AL, Borschel WF, Borgnia MJ, Lander GC, Lee SY. Visualizing structural transitions of ligand-dependent gating of the TRPM2 channel. bioRxiv, Jan 9, 2019. Doi: 10.1101/516468.

https://doi.org/10.7554/eLife.45779.047

Article and author information

Author details

  1. Lejla Zubcevic

    Department of Biochemistry, Duke University School of Medicine, Durham, United States
    Contribution
    Conceptualization, Data curation, Validation, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1884-9235
  2. Allen L Hsu

    Genome Integrity and Structural Biology Laboratory, National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, Research Triangle Park, United States
    Contribution
    Data curation, Validation, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2065-3802
  3. Mario J Borgnia

    1. Department of Biochemistry, Duke University School of Medicine, Durham, United States
    2. Genome Integrity and Structural Biology Laboratory, National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, Research Triangle Park, United States
    Contribution
    Supervision, Funding acquisition, Validation, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9159-1413
  4. Seok-Yong Lee

    Department of Biochemistry, Duke University School of Medicine, Durham, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Validation, Writing—original draft, Writing—review and editing
    For correspondence
    seok-yong.lee@duke.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0662-9921

Funding

National Institute of Neurological Disorders and Stroke (R35NS097241)

  • Seok-Yong Lee

National Institute of Environmental Health Sciences (ZIC ES103326)

  • Mario J Borgnia

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

Cryo-EM data were collected at the Shared Materials Instrumentation Facility at Duke University as part of the Molecular Microscopy Consortium, and screening was performed at National Institute of Environmental Health Sciences (NIEHS). We thank Alberto Bartesaghi for developing a pre-processing interface, which enabled on-the-fly monitoring of cryo-EM image quality during data collection. Funding: This work was supported by the National Institutes of Health (R35NS097241 to S-YL) and by the National Institutes of Health Intramural Research Program; US National Institute of Environmental Health Sciences (ZIC ES103326 to MJB). The EM maps and atomic models have been deposited with the Electron Microscopy Data Bank (accession numbers EMD-20143, EMD-20145, EMD-20146, and EMD-20148) and the Protein Data Back (entry codes 6OO3, 6OO4, 6OO5, and 6OO7), respectively.

Senior Editor

  1. Richard Aldrich, The University of Texas at Austin, United States

Reviewing Editor

  1. Kenton Jon Swartz, National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States

Reviewers

  1. Richard K Hite, Memorial Sloan Kettering Cancer Center, United States
  2. Raimund Dutzler, University of Zürich, Switzerland

Publication history

  1. Received: February 4, 2019
  2. Accepted: May 14, 2019
  3. Accepted Manuscript published: May 15, 2019 (version 1)
  4. Version of Record published: May 31, 2019 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 1,216
    Page views
  • 277
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Structural Biology and Molecular Biophysics
    Zhijie Chen et al.
    Research Article Updated
    1. Structural Biology and Molecular Biophysics
    Babette E de Jong, John van Noort
    Insight