Cryo-EM structure of the cytoplasmic domain of murine transient receptor potential cation channel subfamily C member 6 (TRPC6)

The kidney maintains the internal milieu by regulating the retention and excretion of proteins, ions, and small molecules. The glomerular podocyte forms the slit diaphragm of the ultrafiltration filter, whose damage leads to progressive kidney failure and focal segmental glomerulosclerosis (FSGS). The canonical transient receptor potential 6 (TRPC6) ion channel is expressed in the podocyte, and mutations in its cytoplasmic domain cause FSGS in humans. In vitro evaluation of disease-causing mutations in TRPC6 has revealed that these genetic alterations result in abnormal ion channel gating. However, the mechanism whereby the cytoplasmic domain modulates TRPC6 function is largely unknown. Here, we report a cryo-EM structure of the cytoplasmic domain of murine TRPC6 at 3.8 Å resolution. The cytoplasmic fold of TRPC6 is characterized by an inverted dome-like chamber pierced by four radial horizontal helices that converge into a vertical coiled-coil at the central axis. Unlike other TRP channels, TRPC6 displays a unique domain swap that occurs at the junction of the horizontal helices and coiled-coil. Multiple FSGS mutations converge at the buried interface between the vertical coiled-coil and the ankyrin repeats, which form the dome, suggesting these regions are critical for allosteric gating modulation. This functionally critical interface is a potential target for drug design. Importantly, dysfunction in other family members leads to learning deficits (TRPC1/4/5) and ataxia (TRPC3). Our data provide a structural framework for the mechanistic investigation of the TRPC family.

Transient receptor potential (TRP) 5 channels form the second largest tetrameric cation-permeating ion channel superfamily. The 27 mammalian TRP channel subunits are classified into six subfamilies, TRPC (canonical), TRPM (melastatins), TRPV (vanilloid), TRPA (ankyrin), TRPML (mucolipins), and TRPP (polycystins) based on primary structure similarity (1). They are ubiquitously expressed throughout the body and involved in several physiological and pathophysiological processes, including but not limited to temperature detection, pain, neurotransmission, and vascular regulation (2,3). The transmembrane domain (TMD) forms the ion channel core of TRP channels and shares common structural features. Each subunit consists of six membrane-spanning segments (S), S1-S6, and a pore helix connecting to a re-entrant loop (4). The pore-forming helices (S5 and S6) adopt an inverted teepee structure similar to the architecture of potassium channels (5,6). TRP channels share a unique common motif, the TRP box (EWKFAR), immediately adjacent to the ion channel gate. The N and C termini co-assemble into a cytoplasmic domain, which is the most variable substructure among the TRP receptor family members whose structures have been solved (4,(7)(8)(9)(10). The observed structural diversity suggests that the cytoplasmic domain confers a subtype-specific function to the receptor, such as serving as an interface for protein interactions and allosteric gating modulation.
The canonical TRP subfamily is further divided into two groups based on evolutionary phylogeny, TRPC1/4/5 and TRPC3/6/7 (11). Homotetrameric and heterotetrameric assembly of subunits are observed among the members of this subfamily (12). Although the two subgroups are homologous, the architecture of individual TRPC channels in their homotetrameric and heterotetrameric arrangement may differ, such as seen in the case of AMPA-type ionotropic glutamate receptors (13). The architecture of TRP channels has been extensively studied using cryo-EM (14,15). TRPV1 was one of the first high-resolution single-particle cryo-EM structures of a membrane pro-tein to be determined by exploiting the power of direct electron detectors and image analysis software (4). Since then, a representative structure of each subfamily of the vertebrate TRP receptors has been solved. The TRPC channels were the last family to be characterized (16 -19).
The pore of TRPC6 is permeable to Ca 2ϩ and Na ϩ ions. Diacylglycerol is sufficient to activate the TRPC6 channel (23), and physiological gating occurs in synergy with the activation of a Gq/11-coupled receptor, such as the muscarinic receptors (24). The currents elicited by TRPC6 depolarize the cell membrane to subsequently activate voltage-dependent ion channels, as well as secondary signaling triggered by Ca 2ϩ ions (25). Its gating is also influenced by glycosylation, phosphorylation, lipids (e.g. diacylglycerol, phosphatidylinositol (3,4,5)-trisphosphate), and binding of other proteins such as calmodulin and inositol 1,4,5-trisphosphate receptors (26). The biological processes regulated by TRPC6 include endothelial permeation in blood vessels, growth of neuronal processes, and glomerular filtration (1).
Mutations in TRPC6 were identified in patients with both adolescent and adult-onset focal segmental glomerulosclerosis (FSGS) (27)(28)(29)(30)(31). TRPC6 channels carrying different FSGS mutations have been expressed in HEK cells and show varying gating phenotypes that span from gain-to loss-of-function (27,29,30). The TRPC6 gating modulation mechanism and its relation to disease is largely unknown. Locating and determining the structural impact of these disease-causing mutations in the channel would be an important step toward solving this problem. There are a large number of studies addressing the physiological role of TRPC6, and its molecular architecture is starting to become clear (17). Here we report a cryo-EM structure of the TRPC6 cytoplasmic domain at an overall resolution of 3.8 Å together with its atomic model that was built de novo.

Expression and purification of TRPC6
To determine the structure of the Mus musculus TRPC6 cytoplasmic domain, we used a construct lacking the initial 94 residues at the N terminus and the two glycosylation sites (N472Q and N560Q), referred to as TRPC6 hereafter (Fig. 1A, top). This construct displays higher amounts of protein after purification than the WT and recapitulates the lipid (OAG)mediated activation observed in WT TRPC6 (Fig. 1, A and B). For expression and purification, we generated a baculovirus construct consisting of a His 8 -maltose-binding protein (MBP) tag at the N terminus of the TRPC6 sequence (Fig. 1A, top). We expressed this construct in insect cells (Sf9), stably purified to homogeneity in dodecyl maltoside (DDM), transferred it to the amphipol PMAL-C8, and subsequently cleaved the MBP tag. Final size-exclusion chromatography resolved the TRPC6 as a single peak and to homogeneity as determined by SDS-PAGE gel (Fig. 1C). Following this strategy, we obtained biochemical quantities of amphipol reconstituted detergent-free proteins for negative stain and cryo-EM analysis.

Structural heterogeneity in TRPC6 particles
The initial negative stain and cryo-EM analysis of TRPC6 revealed an overall particle structure that consists of a combination of well-defined and flexible regions ( Fig. 2A and Fig. S1). In the vitrified sample, the amphipol-embedded TRPC6 particles were monodisperse in the absence of detergent (Fig. 2B). In the cryo-EM class averages, substructures corresponding to ␣-helices were clearly detectable and views representing a 4-fold symmetric architecture were observed within the well-defined region ( Fig. 2C and Fig. S2B). In contrast, the flexible region was splayed apart at a variety of angles in the negative stain class averages ( Fig. 2A), whereas in the cryo-EM class averages they were averaged out as diffuse densities ( Fig. 2C and Fig. S2B). Attempts were made to identify a biochemical condition that would remove the structural heterogeneity, but this proved unsuccessful. To solve the cryo-EM structure of the well-defined portion of the particle, while neglecting the heterogeneous remainder, we applied a mask to facilitate 3D classification in RELION2 (32,33) so that only the well-structured region would be taken into account in our analysis (see "Experimental Procedures" for details).

Cryo-EM structure of TRPC6 cytoplasmic domain The structure of TRPC6 cytoplasmic domain
The primary structure of TRPC6 predicts that the N-terminal cytoplasmic segment consists of four ankyrin repeats (AR1-4) followed by the linker helical domain (LHD) comprising six additional ␣-helices connected by loops. In contrast, the C-terminal cytoplasmic segment is a shorter polypeptide and contains two long ␣-helices connected by a short linker, as shown in Fig. 3A (see also Fig. S3). The N-terminal and C-terminal portions of the polypeptides are known to fold into a cytoplasmic assembly in other TRP receptors (4,(7)(8)(9)(10).
The cryo-EM density map clearly resolved individual polypeptides, the secondary structures, as well as the large side chains (Fig. S4). At an overall resolution of 3.8 Å (Fig. 3E and Fig. S5), we were able to build a de novo atomic model (statistics in Table 1). The global tetrameric architecture resembles a baseless wine glass (Fig. 3, C and D). More precisely, the structure adopted a 4-fold symmetric inverted dome shape with a bundle of ␣-helices extending downward. The wall of the inverted dome was formed primarily from AR1-4, loops, and six additional ␣-helices, which correspond to the LHD, contributed by each subunit (Fig. 3B and Fig. S6). Hereafter, we refer to the individual helices in the LHD as LH1-6. Our results agree with secondary structure prediction, suggesting that the N-ter-minalcytoplasmicdomainfoldsintoAR1-4andmultiple␣-helices (Fig. S3). The structure in the EM density map lacked any characteristic transmembrane ␣-helices, and thus we concluded that the domain structure we solved is the cytoplasmic domain.

The cytoplasmic domain of TRPC6 is an independent module
Consistent with the majority of TRPC6 in PMAL-C8 adopting a splayed apart TMD, after 3D classification, the 3D map contained featureless density consisting mostly of noise around where the TMD would exist (Figs. S2 and S5). This density disappears when the map is viewed at an optimal threshold that resolves the cytoplasmic domain. The transmembrane ␣-helices were completely unresolved, whereas the overall architecture of the cytoplasmic domain was well-defined, suggesting that the cytoplasmic domain forms a robust stable module even when the TMD fold is disordered. This observation is in agreement with the idea that the C-terminal coiled-coil domain of the TRP channel is critical for assembly (7,34,35). The disordered TMD is unlikely to reflect a physiological state of the protein (see "Discussion").

Organization of the loops and helices in the cytoplasmic domain
Extensive co-assembly of the N-and C-terminal domains is observed in this structure. In detail, the C-terminal cytoplasmic segment exits from the TMD, bends 90 o toward the central axis, and forms a horizontal helix (HH) running parallel to the membrane ( Fig. 4A and Fig. S6C). As a result, in the tetrameric assembly, the HHs form a cross shape within a plane parallel to the membrane, in which the junction of the cross forms a right angle (Figs. 3B and 4A). The HH enters the inverted dome from the outside by penetrating an opening formed between the AR domain and the LHD of the adjacent subunit. Each HH connects to a vertical helix (VH) after bending 90°near the central axis. These VHs assemble into a coiled-coil (Fig. 3, C and D), which penetrates the inverted dome into the cytoplasm (Fig.  S6C).
Notably, subunit domain swaps occur when the four HHs merge near the central axis, characterized by a crossover of linkers that connect the HHs and VHs (Fig. 4, A-C). Immediately above these four crossover linkers, a density was found that appears to form a plug at the opening of the hollow coiledcoil tube (Fig. 4C). The exact identity of the plug cannot be determined at the resolution of our map; however, an ion could be coordinated at this position. The plug density was only seen when the reconstruction was calculated from the Titan Krios dataset and was absent when calculated with the Polara dataset. The reason for the difference is unclear. Because the biochemical preparation method was identical and highly reproducible,

Cryo-EM structure of TRPC6 cytoplasmic domain
we speculate that the difference may have emerged from contrasting electron doses applied to the specimen. We used 100 e Ϫ /Å 2 and 47 e Ϫ /Å 2 during data acquisition on the Polara and Titan Krios, respectively.
Other TRP channels such as TRPM4, TRPM8, and TRPA1 have a similar vertical coiled-coil domain but so far lack the C-terminal domain swap observed in TRPC6 (Fig. 4D). Interestingly, the degree of twist found in TRPC6 coiled-coil is weaker when compared with the other family members. The global arrangement of the C-terminal ␣-helices of TRPC6 is reminiscent of the stretcher helices in the TRPM4, which merge at the center and transition to a coiled-coil (8). However, the perfectly horizontal arrangement of the ␣-helices forming an X shape is a unique feature of TRPC6.

The inverted dome-like chamber at the cytoplasmic domain
Given the topology predicted from the primary structure and homology to other members of the TRP channels, the ion channel gate opens into the chamber formed inside the dome. The wall of the dome is made by two stacked layers, the ARs (Fig.  S6A) and the LHD. AR4 connects to LH1-6, which links the dome to the TMD (Fig. S6B). LH1-6 interact with each other pairwise in an anti-parallel orientation, making contacts laterally along the helical axis (Fig. 3B, arrow). They contribute significantly to the structure of the interface between the rim of the inverted dome and the membrane. AR1-4 make residue contacts with nearby structures, creating a physical continuity across different substructures along the wall of the dome. Specifically, toward the membrane they contact the LHD and the HHs, and toward the central 4-fold symmetry axis they contact the VH (Fig. 4).
The small openings of the dome at various locations would allow the escape of ions passing through the channel (Fig. 5, yellow circle). Side chains that contain hydroxyl groups occupy the majority of the inner surface of the inverted dome-like chamber (Fig. 5B) and its openings to the outside (Fig. 5, C and  D). This arrangement would facilitate cations exiting from the chamber after entering the cytoplasm through the channel. The lower half of the dome's exterior exhibits an overall negative charge. On the other hand, a cluster of basic residues close to the transmembrane region is exposed, primarily stemming from the HH and the lower portion of the LHD (Fig. 5, A, B, and D).

The less rigid structural elements in the cytoplasmic domain
The resolution of the cryo-EM density was locally lower in subregions that consist of loops, as well as the very C terminus corresponding to the tip of the coiled-coil. The local resolution of the EM density map was calculated using ResMap (36) (Fig.  S7). Two long loops in the ankyrin repeats make long-range contacts with the C-terminal HH. The loop connecting AR1 and AR2 approaches the HH of the adjacent subunit from the Similarly, the loop connecting AR3 and AR4 consisting of 36 residues is in close proximity with the HH, while adopting a unique fold (Fig. 6, B and D). The resolution of the second loop was ϳ5 Å and thus only the ␣-carbon backbone could be reliably modeled. However, when we place the actual residues into the map it was clear that the amino acids with larger side chains could easily make contact with the nearby structural elements, including the HH and LHD, located closer to the membrane. Overall, we were able to interpret the subregions of our EM density map that had lower resolution (ϳ5 Å) as loops, because the adjacent ␣-helices were well-resolved.

The connection between the cytoplasmic domain and the TMD
The N terminus of S1 and the C terminus of the TRP box are geometrically close in the structures of other TRP channels (4,(7)(8)(9)(10). In TRPC6, the HH follows the TRP box and LH6 precedes S1. It follows that LH6 and the N terminus of the HH must be close in the tetrameric assembly. This is exactly what we find in our cytoplasmic domain structure. The existence of a large chamber in the cytoplasmic domain is predicted to occlude direct physical contact between the majority of the cytoplasmic domain and the ion channel gating machinery. The dynamics of the TMD must be transduced bidirectionally through the junction where the N-terminal tip of the HH and portions of the LHD colocalize. We suggest that subtle conformational rearrangement in the cytoplasmic domain can potentially transduce to the TMD through this junction and vice versa.

Locations of disease-causing mutations
Many mutations in the cytoplasmic domain of TRPC6 have been found in patients with FSGS (27)(28)(29)(30). The geometrical information of residue contacts provided by our cryo-EM structure and atomic model allows us to gain insights into how these mutations may function. We categorize the different mutations into three groups based on the substructures of TRPC6 that may be affected (Fig. 7A). The first group of mutations is clustered at the buried interface between the ARs and the VHs (Fig. 7, A and B, red). Because they are inaccessible from the outside, we postulate that the effect of these mutations on ion channel function is mediated by influencing the internal motion between the AR and coiled-coil, rather than altering TRPC6's interaction with other effectors. These mutations surround a strong point of contact between AR1 and the VH between Asn-109 and Tyr-895 (Fig. 7B, black).
The second group contains two mutations that are near or within the HHs (Fig. 7A, green). Met-131 and its adjacent residues located in the loop connecting AR1 and AR2 reach out to the HH and AR4 of the adjacent subunit (Fig. 7C, green and  black). Hence, the M131T mutation is predicted to alter this intersubunit interaction. An amber mutation, K873X, is located in the HHs at the end nearest the coiled-coil. These two mutations are positioned in a way that could potentially influence the dynamics of the HHs. Because the HHs and VHs are directly attached, it is conceivable that mutations in the two groups influence a similar underlying mechanism of TRPC6 gating. The third group of mutations are scattered in various locations, which did not allow us to deduce specific insights (Fig. 7A, magenta).

Discussion
The cytoplasmic domain of TRPC channels is a site for protein interaction with regulatory factors, such as calcium/calmodulin and inositol 1,4,5-trisphosphate receptors (25). These protein interactions are known to modulate channel function. Determining the structure of the cytoplasmic domain of TRPC6 would be an effective first step toward revealing the mechanism by which these interactions modulate channel gating as well as trafficking. We report here the overall architecture of the cytoplasmic domain of TRPC6 at 3.8 Å resolution. The domain forms a stable modular architecture in the absence of a structured TMD. Disease-causing mutations that produce changes in ion channel gating properties were mapped onto the structure, providing insight into their action.
The 94 residues at the N terminus that were deleted in this study are unique to TRPC6 and absent in other members of the TRPC subfamily; TRPC6 is functional in the absence of these residues (Fig. 1A). In the tetramer, they would add a significant mass (a total of ϳ40 kDa) to the surface of the ankyrin repeat. This stretch has no predicted secondary structures, contains multiple prolines and charged residues. If we assume that they

Cryo-EM structure of TRPC6 cytoplasmic domain
do not form any secondary structure, the fragment may potentially have access to various surfaces of the cytoplasmic domain. Alternatively, this fragment may serve as an interface for inter-acting with other cytoplasmic proteins. More data are needed to address the role of this fragment. Given that the mutations located at the distal portion of the dome influence channel gating (27,29,30), it is conceivable that yet unidentified conformational changes occur to the dome during the gating cycle. Because of the proximity of many mutations to interfaces between the N and C termini, they are likely to destabilize the cytoplasmic domain, which may cause the

Cryo-EM structure of TRPC6 cytoplasmic domain
entire channel to become more unstable and permeable. In particular, we postulate that the C-terminal horizontal and vertical helices would be a dynamic module that undergoes rearrangement of residue contacts with the surrounding ankyrin repeats and their connecting loops, structures that form the wall of the inverted dome. Conformations of TRPC6 in different states would provide answers to these questions.
The cytoplasmic domain retained its structural integrity even when the TMDs were distorted. Therefore, the tetrameric assembly of the channel does not require the intact TMD. The stability of the coiled-coil has been reported before (34,35), but our data extend the previous finding by demonstrating that the combined global fold of the ankyrin repeats, together with the coiled-coil, form a stable modular unit. The connections between the TMD and the linker helical domain, which is positioned more proximal to the membrane relative to the ankyrin repeats, make extensive contacts with various elements of the dome, potentially functioning to bidirectionally transduce the effect of gating and subtle alteration of the residue contacts within the cytoplasmic domain.
An intriguing question arises about how the subunit assembly takes place and embeds the horizontal and vertical coiledcoil into the core of the tetramer, because these C-terminal elements of the polypeptide are synthesized last during translation by ribosomes in the rough endoplasmic reticulum. We speculate that the four ankyrin repeats are flexible during assembly and the incorporation of the C terminus provides the final stability of the domain. An additional conformational rearrangement we postulate, assuming that a state in which the ankyrin repeats detached from the coiled-coil exists, is the rotation of the coiled-coil that causes the crossover linkers, con-necting the horizontal helix to the vertical helix, to unwind and adopt an arrangement seen in other members of the TRP channels.
Our structure lacks the TMD and to demonstrate the veracity of the structure of cytoplasmic domain in the context of the full-length ion channel architecture it would be essential to investigate the cryo-EM structure of the TRPC6 with its TMD intact. We adjusted the detergent conditions to attempt and stabilize a physiological arrangement of the TMD. This showed a high degree of TMD homogeneity of protein seemingly embedded in a uniform detergent micelle in negative stain. When taken to cryo-EM, none of these preparations were able to give ice conditions amenable to determining a high-resolution structure of full-length TRPC6. We predict that an optimal biochemical preparation preserving the structural integrity of the TMD is achievable, but it happens to be outside the experimental parameter space we had explored. Not only the types of detergents but also the lipid composition surrounding the ion channel may also influence the stability of the TMD. Reconstitution of the receptor in nanodiscs would be particularly preferred for future structural studies of the TRPC6 bound to drugs and modulators because of its stability (17).
Before structural biology-guided drug design was a conceivable notion, the active ingredients of St. John's Wort (Hypericum perforatum, plant) were known to have anecdotal antidepressant effects. The active ingredient was hyperforin, a bicyclic polyprenylated acylphloroglucinol derivative that increases ion flux into cells via the TRPC6 channel (37). Efforts have been made to develop synthetic small molecules that would specifically target TRPC6, with an end goal of treating FSGS or hypoxia-induced pulmonary vasoconstriction. Larixyl carbamate (38)

Cryo-EM structure of TRPC6 cytoplasmic domain
and two other small molecules, GSK255B and GSK503A (39), have been identified as TRPC6-specific inhibitors. With a structural analysis pipeline of TRPC6 using cryo-EM, it would become possible to reveal the effects of these drugs on the conformational states of the channel, providing mechanistic insights, and potential for future structure-based drug design for this subfamily of the TRP channels.

Experimental procedures
Cell culture and electrophysiology HEK293 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen), supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin at 37°C and 5% CO 2 . Transfections were performed in 6-well plates using Lipofectamine ® (Invitrogen) with Opti-MEM I Reduced Serum Media (Invitrogen). WT TRPC6 and ⌬TRPC6 cloned in pcDNA3 vector with the ␤-globin gene were used for transfection. For whole-cell recordings, the extracellular solution contained 140 mM NaCl, 2.8 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , and 2 mM HEPES (pH 7.4). The pipette solution contained 140 mM CsCl, 5 mM EGTA, and 10 mM HEPES (pH 7.2). Currents were recorded with an Axoclamp 200A amplifier (Molecular Devices) using a 1-s ramp from Ϫ80 mV to 80 mV. Pipettes were made of glass capillaries (Sutter Instruments) and fire-polished before use until a resistance between 2.8 and 4 megohms was reached. TRPC6 channel agonist OAG (Avanti Polar Lipids) in chloroform was dried under a gentle stream of N 2 , dissolved in DMSO, sonicated for 20 min, and freshly dissolved in bath solution to the indicated concentration.

Mus musculus TRPC6 expression and purification (NP_038866.2)
A DNA construct containing His 8 -MBP-N terminus truncated TRPC6 (94) and putative glycosylation sites mutated (N472Q and N560Q) (40) was cloned into the pFastBac1 expression vector. Recombinant baculovirus was obtained following the manufacturer's protocol (Bac-to-Bac expression system; Invitrogen). Sf9 cells were infected with recombinant baculovirus and harvested by centrifugation 72 h after infection. Cell pellet from 0.8 liter of culture was resuspended and lysed with a high-pressure homogenizer (Avestin) in a hypotonic buffer (36.5 mM sucrose, 50 mM Tris, 4 mM tris(2-carboxyethyl)phosphine (TCEP), pH 8) and supplemented with protease inhibitors (1 mM PMSF, 3 mg/ml aprotinin, 3 mg/ml leupeptin, and 1 mg/ml pepstatin). Cell debris was collected by low-speed centrifugation (8000 ϫ g for 15 min). Membranes were collected by ultracentrifugation (100,000 ϫ g for 30 min at 4°C) and solubilized in Buffer A (150 mM NaCl, 4 mM TCEP, 10% glycerol, 50 mM HEPES, pH 7.4) supplemented with protease inhibitors. Protein was extracted with 20 mM DDM (Anatrace) with gentle stirring for 2 h. The detergent-insoluble material was removed by centrifugation (150,000 ϫ g for 45 min), and the supernatant was incubated with amylose resin (New England Biolabs) with gentle stirring for 3 h. After loading onto the column and collecting the flow-through, the resin was washed with 10 column volumes of Buffer B (150 mM NaCl, 4 mM TCEP, 10% glycerol, 1 mM DDM, 50 mM HEPES, pH 7.4). Afterward, the protein was eluted with Buffer B supplemented with 20 mM maltose. The eluted protein was then mixed with PMAL-C8 (Anatrace) at 1:3 (w/w) with gentle agitation for 2 h at 4°C. Then, protein was digested with ProTEV Plus protease (Promega) at 4°C overnight to remove the MBP tag. Cleaved protein was further purified by size-exclusion chromatography on a Superose 6 10/300 GL column (GE Healthcare) pre-equilibrated with Buffer C (150 mM NaCl, 4 mM TCEP, 20 mM HEPES, pH 7.4). Peak fractions corresponding to the tetrameric channel were collected and concentrated to ϳ0.3 mg/ml for cryo-EM analysis.

Negative stain data collection and analysis
400 mesh copper grids were coated with carbon. 4 l of 0.05 mg/ml TRPC6 was applied to each glow discharged grid and allowed to absorb for 30 s. The excess buffer was blotted on filter paper, washed twice in Milli-Q water, and negatively stained with 0.75% (w/v) uranyl formate (41). Images were recorded on a 4000 ϫ 4000 charge-coupled device camera using an FEI F20 transmission electron microscope operated at 200 keV. All images were taken at 50,000 magnification in lowdose mode at a defocus of Ϫ1.5 m. Image processing was performed using the SPIDER (48) software package. Images were converted to SPIDER format and particles were picked manually using the WEB display program. Images were rejected if they had a large amount of astigmatism or deviated by more than 0.1 m from Ϫ1.5 m defocus as determined by processing with CTFFIND3 (42). Once picked, the particles were windowed in 100 ϫ 100 pixel boxes. 100 2D class averages were determined using eight iterations of multireference classification alignment (Fig. S1).

Random conical tilt specimen and data collection
4 l of 0.05 mg/ml TRPC6 was applied to carbon-coated C-flat holey carbon 2/1 200 mesh grids and allowed to absorb for 30 s. Excess buffer was blotted on filter paper, washed twice in Milli-Q water, and negatively stained with 0.75% (w/v) uranyl formate (41). Images were recorded on a 4000 ϫ 4000 chargecoupled device camera using an FEI F20 transmission electron microscope operated at 200 keV. Tilted images were taken at 28,000 magnification in low-dose mode at 50°and a defocus of Ϫ1.8 m. Paired untilted images were taken at Ϫ1.5 m defocus (Fig. S1).

Initial 3D model generation
An initial 3D model was created using the SPIDER software package. Micrographs were converted to SPIDER format and tilt pairs were manually picked using the WEB display program. The picked particles were windowed into 100 ϫ 100 pixel boxes and bad particles were manually discarded from both the untilted and tilted data sets. 100 2D class averages were generated from the untilted particles using eight iterations of multireference classification alignment. Eight iterations of SPIDER backprojection were done to create initial models of TRPC6 in a closed and open conformation from two classes (classes 55 and 83), respectively. Initial models were converted to mrc format with box and pixel size adjusted to match the cryo-EM data using EMAN2 (Fig. S1).

Cryo-EM structure of TRPC6 cytoplasmic domain
Cryo-EM data collection 2.0 l of TRPC6 in PMAL-C8 amphipol in 150 mM NaCl, 4 mM TCEP, 20 mM HEPES, pH 7.4 was applied to a 200 mesh C-flat holey carbon 2/1 grid that was glow discharged for 2 min at 25 milliamperes. The grid was blotted for 8 s before being plunged into liquid ethane using an FEI Vitrobot Mark III at 8°C and 100% humidity. The TRPC6 micrographs were collected using an FEI Polara F30 microscope (at Vanderbilt) operated at 300 keV. TRPC6 WT data were also collected using an FEI Titan Krios microscope (at Washington University in St. Louis). In the data collection sessions using the Polara, micrographs were recorded using super resolution mode on a K2 Summit direct electron detector (Gatan) at a nominal magnification of 31,000ϫ in a defocus range of Ϫ1.8 to Ϫ4.8 m at a low-dose condition. Specimens were exposed for 8 s at ϳ12.5 e Ϫ /pixel/second over 40 frames resulting in a total dose of 85 to 100 e Ϫ /Å 2 . Data were collected manually and with automation using SerialEM. The calibrated magnification after binning the image by a factor of 2 was 1.24 Å/pixel. The Titan Krios was equipped with a Cs aberration corrector, an energy filter (GIF), and a K2 Summit direct electron detector (Gatan). Images were taken at a nominal magnification of 81,000ϫ using EPU software-aided automation over 36 h using the counting mode on the detector. Under the low dose conditions, the specimen was exposed at ϳ11 e Ϫ /pixel/second for 9 s over 30 frames accounting for a total dose of ϳ47 e Ϫ /Å 2 . The calibrated magnification of the images was 1.41 Å/pixel.

Image processing of the Polara datasets
All images were motion corrected and binned 2 ϫ 2 using motioncor2 (43) with a dose weighting parameter of 2.5 e Ϫ /angstrom/frame. The contrast transfer function for each image was determined using Gctf (44) and used to remove micrographs with low maximum resolution or high astigmatism. Further processing was done using RELION2 (32). Putative particles were identified by autopicking, using representative class averages obtained from 2000 manually picked particles as templates. Particles were extracted at a box size of 256 ϫ 256 pixels. 2D class averages were determined using 25 iterations of classification. Our initial model, from average 55, described above was filtered at 60 Å and used as a reference for 3D classification. The dome-like feature was clearly defined in the initial model but the TMD was splayed apart and adopted variable conformations. Similar features were obvious from the cryo-EM class averages. A 3D classification into six classes was conducted with C4 symmetry applied using a mask excluding the splayed apart TMD but containing the dome-shaped feature. A single class that clearly resolved the polypeptide backbone was selected and subjected to 3D autorefine. Postprocessing was done by applying a B-factor that was calculated by RELION2 based on Rosenthal and Henderson's method (45) (Fig. S2). Detailed statistics are provided in Table 1.

Image processing of the Titan Krios datasets
All images were motion corrected without binning using motioncor2 with dose weighting parameter of 1.56 e Ϫ /angstrom/frame. Cs aberration parameter was set at 0.001. We followed an identical procedure as the Polara dataset, except that the model produced from the Polara data, filtered to 40 Å, was used as a reference for 3D classification into 12 classes with the same mask and C4 symmetry imposed. (Fig. S5) The detailed statistics are provided in Table 1.

Model building, refinement, and validation
A polyalanine model was first built manually with COOT (46) and subsequent amino acid assignment was performed by defining densities of the aromatic residues (Fig. S4). The fullatom model for the tetrameric assembly was constructed using UCSF Chimera (47). This model was optimized using Phenix's real-space refinement tools (20). To prevent overfitting of the model into the density, refinement was run for two cycles with mid-range geometric restraints of 0.0075 and 0.75 while non-crystallographic symmetry was imposed. To conserve helix assignments, strict secondary structure restraints were included to ignore outliers. These refinement parameters were iterated four times with manual adjustment of the structure to correct for Ramachandran and rotamer outliers and bond angle deviations of more than 4. We performed a final round of refinement incorporating only morphing, global minimization, secondary structure restraints, and non-crystallographic symmetry. Validation of the final model was performed using MolProbity (21).

Data deposits
The EM density map and atomic coordinates were deposited to the Protein Data Bank and EM Data Bank under accession codes 6CV9 and EMD-7637, respectively.