Molecular architecture of the human caveolin-1 complex.

Membrane-sculpting proteins shape the morphology of cell membranes and facilitate remodeling in response to physiological and environmental cues. Complexes of the monotopic membrane protein caveolin function as essential curvature-generating components of caveolae, flask-shaped invaginations that sense and respond to plasma membrane tension. However, the structural basis for caveolin's membrane remodeling activity is currently unknown. Here, we show that, using cryo-electron microscopy, the human caveolin-1 complex is composed of 11 protomers organized into a tightly packed disc with a flat membrane-embedded surface. The structural insights suggest a previously unrecognized mechanism for how membrane-sculpting proteins interact with membranes and reveal how key regions of caveolin-1, including its scaffolding, oligomerization, and intramembrane domains, contribute to its function.

INTRODUCTION

Biological membranes assume a wide range of morphologies and are actively remodeled to support the normal physiological functions of cells as well as cell growth and differentiation (, ). An important mechanism for shaping and remodeling membranes is through the activity of curvature-inducing proteins (–). Caveolins, a family of monotopic membrane proteins first identified more than 30 years ago, fall into this class of proteins based on their ability to bend membranes to form 50- to 100-nm flask-shaped invaginations of the plasma membrane known as caveolae (, ). Comprising up to 30 to 50% of the surface area of some cell types (), caveolae serve as mechanosensors and mechanoprotectors that sense and respond to changes in membrane tension at the cell surface by reversibly flattening (). The absence of caveolin-1 (Cav1) and, thus, caveolae has profound consequences for mice, including decreased life span, lipid metabolism disorders, vascular abnormalities, dilated cardiomyopathy, and pulmonary hypertension (). In humans, Cav1 has been linked to cardiovascular disease, cancer, lipodystrophy, and kidney disease (–). How Cav1 facilitates caveolae biogenesis is not fully understood but is thought to require the insertion of a highly hydrophobic region of the protein, termed the intramembrane domain (IMD), into the cytoplasmic face of the plasma membrane to induce membrane curvature via a wedging mechanism (, , , ). To induce curvature, Cav1 must also assemble correctly into oligomeric complexes 8S in size that undergo additional higher-order interactions with each other and the cavin proteins to form mature caveolae that are polygonal in shape (–). 8S complex formation occurs via a cooperative process mediated by its oligomerization domain (OD) and aided by its scaffolding domain (SD) and signature motif (SM) (–). However, the structural basis for Cav1’s membrane-bending activity remains unknown, as only low-resolution structures of caveolin complexes currently exist (–).

RESULTS AND DISCUSSION

Although caveolin proteins are found exclusively in metazoans (), when expressed in Escherichia coli, Cav1 forms 8S-like complexes () and induces the formation of heterologous caveolae (h-caveolae) (, ). Thus, E. coli–expressed Cav1 forms oligomers and sculpts membranes, two of its essential functions in mammalian cells. Using single-particle cryo–electron microscopy (cryo-EM) analysis, we determined the structure of the 8S complex expressed in E. coli (Fig. 1, figs. S1 to S3, and Materials and Methods) (). Two-dimensional (2D) class averages of en face views of the complex showed 11 spiraled α helices, and an ab initio 3D reconstruction with no enforced symmetry resulted in a low-resolution map with structural features indicating potential 11-fold symmetry (fig. S3, C and D). Applying C11 symmetry during 3D refinement steps led to a 3.5-Å-resolution structure (Fig. 1 and fig. S4). This agrees with previous reports suggesting 7 to 14 copies of Cav1 per 8S complex (, , ). The 11 Cav1 protomers organize into a disc-shaped complex with a diameter of ~140 Å and a height of ~34 Å (Fig. 1, A to F). To ensure we were not missing other oligomeric forms of the complex by enforcing 11-fold symmetry, we performed 3D refinements using symmetries spanning C8 to C14. None of these resulted in maps with apparent secondary structural features or improved resolutions (fig. S5). In addition, 3D classification did not yield models of any other oligomeric forms.

Fig. 1.

Cryo-EM structure of the 8S Cav1 complex.

(A to C) Ninety-degree rotated views of the cryo-EM density of the 8S Cav1 complex at 3.5 Å with 11-fold symmetry. The complex is a disc-shaped structure composed of tightly packed α-helices and a cylindrical β-barrel. Magenta, Cav1 protomer. (D to F) Secondary structure model of the refined 8S Cav1 complex. Same views as shown in (A) to (C). Dimensions of complex are labeled in (E). (G and H) Secondary structure of the Cav1 protomer with secondary structural features and position of the N and C termini labeled. (I) Central slice of the density map (blue) with detergent micelle (light gray). Scale bar, 20 Å.

The structure reveals that Cav1 oligomerizes into a disc that contains an outer “rim” (~23 Å wide), a central β-barrel “hub” (~28 Å wide), and 11 curved α-helical “spokes.” Each Cav1 protomer is mostly α-helical (α1 to α5), consistent with previous reports (Fig. 1, D to H) (). Viewed from the side, helices α2 to α5 of the protomer align along the flat surface of the complex, and helix 1 forms a ~20° angle to this surface (Fig. 1H). Cav1 also contains two nonhelical regions located at both termini (Fig. 1, D to H). The N terminus, which makes up part of the rim region, consists of a loop with a short 310 helix (η1) (Fig. 1, G and H). This region makes a 180° turn at the outer edge of the complex, intercalating with the spokes and N-terminal regions of neighboring protomers (Fig. 1, A to H). Each protomer terminates in a C-terminal β strand angled by ~40° relative to the plane of the flat surface of the disc. These β strands interact to form an 11-stranded parallel β barrel hub (Fig. 1, A to H). On the basis of the position of the detergent micelle in the density map (Fig. 1I) and our previous negative stain analysis of h-caveolae (), the flat face of the complex corresponds to the membrane-facing surface, whereas the β barrel faces the cytoplasm. Notably, the central cavity of the barrel extends entirely through the complex and is open to solvent at the cytoplasmic-facing side (Fig. 1, A to F and I). The detergent micelle covers ~60% of the surface area of the 8S complex, including the membrane-facing surface, and extends ~14 Å around the sides of the outer rim of the disc (Fig. 1I).

From the density map, we built an atomic model for most of Cav1 (residues 49 to 177) (Fig. 2, A to D). However, residues 1 to 48, which fall within a predicted disordered region (, ) and are dispensable for caveolae biogenesis (), were not detected in the map (Fig. 2, A to D, and fig. S6). To correlate the extensive biochemical and functional data in the literature with our structural analysis, we mapped the key regions and residues of Cav1 onto the structure (Fig. 2, A to G; fig. S7; and movie S1). The OD (residues 61 to 101) () is located at the outer rim of the disc and contributes to extensive subunit interfaces. The SM (residues 68 to 75) (, ) and SD (residues 82 to 101) () both fall within the OD. Lying within the loop region, the SM forms tight contacts with two neighboring protomers, whereas the SD comprises most of helix α1 and encircles the periphery of the complex. The residues separating the SM and SD make two 90° turns, bringing these two motifs into proximity (Fig. 2D). The IMD (residues 102 to 134) (, ) begins on the C-terminal end of the helix α1 immediately adjacent to the SD and continues across helices α2, α1, and part of α3 (Fig. 2, A to D). Known phosphorylation and ubiquitination (, ) sites are accessible on the cytoplasmic face of the complex, whereas Cav1’s palmitoylation sites (Cys133, Cys143, and Cys156) () are located on the membrane-facing surface (fig. S8).

Fig. 2.

Structure of Cav1.

(A) The 3.5-Å-resolution structure of the 8S Cav1 complex as viewed from the cytoplasmic face. (B) The structure rotated 90°. (C and D) The structure of Cav1 rotated 90°. The positions of previously defined regions are labeled: SM, signature motif (red); SD, scaffolding domain (green); and IMD, intramembrane domain (purple). The OD, which contains the SM and SD, is indicated by the dashed box. Previously uncharacterized structurally defined motifs include the following: PM, pin motif (yellow), SR, spoke region (gray); and β strand (cyan). N terminus (NT) and C terminus (CT) are marked with arrows. (E to G) The space-filling model of the 8S Cav1 complex rotated 90°. Color scheme identical as in (A) to (D). (H to J) The space-filling model of the 8S Cav1 complex rotated 90°, showing the charge of the amino acids. Red, negative; gray, neutral; blue, positive.

In addition to these previously characterized functional regions, we identified three previously undescribed motifs in Cav1 that are critical for 8S complex stability. First, residues 49 to 60 form a loop extending over the SM of the neighboring protomer (Figs. 2, A to D, and 3). This previously unnamed region is required for caveolae assembly () and localization in migrating fibroblasts (). The structure shows that these residues appear to “lock” the interaction between protomers, leading us to designate this region as the “pin motif” (PM) (Figs. 2, A to G, and 3; and fig. S7). Second, residues 135 to 169, a region required for caveolae formation (), oligomerization, and exit of the protein from the Golgi complex (), generate a spoke region (SR) organized parallel to the membrane plane (Fig. 2, A to D, and fig. S7). The structure shows that these residues form the planar hydrophobic surface on the membrane-facing side of the complex while also creating a highly negatively charged surface on the cytoplasm-facing side of the complex (Fig. 2, G and J). Helix α5 (residues 143 to 155) distorts this flat surface by bending toward the cytoplasm-facing side of the complex where it connects to a β strand (residues 170 to 176) (Figs. 1, G and H, and 2, C and D). β strands from adjacent protomers assemble into the third previously undescribed motif, an 11-stranded parallel β barrel with a hydrophilic exterior and hydrophobic interior (Fig. 2, A to G). To our knowledge, this represents the largest reported example of a parallel β barrel to date; the most similar β barrel reported in the Protein Data Bank (PDB) (2AO9) was observed in a soluble bacteriophage protein of unknown function and contains nine parallel β strands (fig. S9A). The barrel is capped by Lys176, introducing a highly positively charged layer separating the hydrophobic interior of the barrel from the cytoplasm. The narrowest accessible region of the interior of the β barrel has a diameter of 15 Å (fig. S9B), making the channel large enough to accommodate small molecules such as lipids. However, no density was detected inside the β barrel, perhaps because of the imposed C11 symmetry.

Fig. 3.

Cav1 8S complex is stabilized by extensive interactions along the length of the protomers.

(A) Overall structure of the 8S complex highlighting five distinctly colored protomers labeled i − 2 to i + 2. (B) Zoomed-in view of the rim region [box in (A)] highlighting key interacting residues. (C) Packing of two protomers. Secondary structure elements are labeled.

The structure also unveils atomic details of how protomers oligomerize to form 8S complexes, a process essential for caveolae biogenesis. Classically, the OD has been proposed to function as the main region driving protomer-protomer interactions that lead to 8S complex formation (). The structure reveals that, instead, extensive networks of interactions occur along the entirety of each protomer, including the rim, spoke, and hub regions (Fig. 3 and fig. S9, C to F). Starting at the rim, the N terminus of one protomer (i) engages four neighboring protomers (i − 2, i − 1, i + 1, and i + 2) (Fig. 3, A and B). The ODs from neighboring protomers are arranged adjacent to each other, but only slightly overlap (Fig. 3B). Instead, the OD of protomer i stacks onto the cytoplasmic side of the IMDs of the neighboring protomers i + 1 and i + 2 (Fig. 3, A and B). Within OD, SM is sandwiched between α1, α2, α2, α3, π1, and PM. Arg54 of PM intercalates into a pocket formed by His79 and Trp85, pinning OD in place (Fig. 3, A and B). To test the importance of this previously unknown molecular “pin,” we mutated Arg54, a residue predicted to play an important role in facilitating protomer-protomer interactions, to alanine. This mutation severely disrupts the 8S complexes, confirming that the PM interaction with the SM of the neighboring protomer is important for the 8S complex formation (fig. S10). The highly hydrophobic IMD helices (α2, π1, and part of α3) lay underneath the OD of the neighboring protomers, contributing to oligomerization (Fig. 3, A and C). α2 forms a helical bundle that crosses α1 facilitated by the small side chains of residues Ala112, Gly116, and Ala120. The side chain of Trp128 extends from π1 into the cavity formed by OD and the loop between α1 and α2, suggesting that it plays a critical role in oligomerization (Fig. 3, A and C). The remaining helices forming the SR (C-terminal portion of α3, α4, and α5) make pairwise contacts across protomers (Fig. 3C and fig. S9, C to F). These interactions involve residues with larger side chains, increasing the separation of the helices compared to those of the α1-α2 crossing, which ultimately converge to form the C-terminal β barrel, the final major region of protomer-protomer interactions. The symmetry dictates a two-residue offset between neighboring protomers, creating a (, ) β barrel. Several disease-associated mutations in Cav1 localize on the connecting regions between the IMD and SR or the SR and β barrel (fig. S8), suggesting that this region of the complex is sensitive to mutations that destabilize the complex (, ).

The structure of the human Cav1 8S complex provides a molecular framework for understanding the overall organization of Cav1 oligomers; the exact roles that OD, SD, SM, and IMD play in complex formation; the importance of previously unrecognized regions of the protein; and the impact of disease-associated mutations such as P132L (fig. S8, H and I) () on the structure. It also suggests unconsidered models for how Cav1 packs in membranes and controls caveolae architecture (Fig. 4, C to E). In contrast to the prevailing model suggesting that the IMD forms a hairpin-like structure that inserts into the membrane, creating a wedge that bends membrane (, , , ), the structure of the 8S Cav1 complex reveals that the IMD contributes to the formation of a flat membrane-facing surface while simultaneously stabilizing contacts between protomers. The outside of the outer rim of the complex is also primarily hydrophobic. Together, these features suggest a model in which the membrane-associated side of the complex embeds deeply within the cytoplasmic leaflet, interacting with the terminal carbons of the lipids of the opposing leaflet rather than sitting at the interface between the head groups and acyl chains as typically would be expected for amphipathic helices (Fig. 4B). In this model, the 8S complex, by displacing lipids from the cytoplasmic leaflet, could create an ordered membrane nanodomain composed of protein on one membrane leaflet and lipids on the other. How posttranslational modifications of Cav1 such as palmitoylation and phosphorylation influence the structure of the 8S complex remains to be investigated. Other important questions for the future include the role that specific lipids, such as cholesterol, play in regulating the architecture of the complex as well as how the structure of Cav1 might differ when embedded in a membrane versus a detergent micelle.

Fig. 4.

Proposed model for Cav1 8S complex association with membranes and packing on caveolae.

(A) Classical model of Cav1 oligomer organization and membrane interaction. (B) Structure-based model. See the main text for details. Coloring scheme matches Fig. 2 (A to D). (C to E) Packing of 8S complexes on dodecahedrons of the characteristic size of h-caveolae and caveolae. h-Caveolae and caveolae are depicted as regular dodecahedrons assuming a circumradius of 15 nm for h-caveolae () and 30.5 nm for mammalian caveolae (). 8S complexes are partially embedded in the membrane at a depth consistent with the model shown in (B). (C) A single 8S complex fits on each face of an h-caveola. (D) Representative packing density of 8S complexes on a mammalian caveola assuming a single 8S complex per face. (E) Each face of a mammalian caveola can potentially fit up to three 8S complexes. Images are drawn to scale in (C) to (E).

A single 8S complex fits snugly on each face of a dodecahedron of the characteristic size of h-caveolae (30 nm average diameter) (Fig. 4C) (). This model predicts that a single h-caveola would be expected to contain 132 copies of Cav1, within the range of 144 ± 39 copies per caveola reported from quantitative microscopy measurements (). However, more than one 8S complex can be accommodated on each surface of a mammalian caveolae (61 nm average diameter) assuming dodecahedral symmetry (fig. S4, D and E) (), thus raising the possibility that caveolae may contain more copies of Cav1 than previously estimated or that some of the space on each face is filled by additional proteins (). Together, these findings suggest that caveolin complexes may function primarily by stabilizing flat membrane surfaces of polyhedral structures rather than imposing continuous membrane curvature, defining a previously unrecognized mechanism for how integral membrane proteins sculpt cell membranes to form functional domains.

MATERIALS AND METHODS

Construct preparation

The Cav1-MycHis plasmid was constructed by site-directed mutagenesis using the QuikChange Lightning Kit (Agilent, #210518). The template was cysteine-less Cav1-MycHis (a gift from K. Jebrell Glover, Lehigh University). Primers were used to introduce native cysteines in the positions 133 (primer: GGCGGTTGTTCCGTGCATCAAATCTTTCCTGATCG), 143 (primer: CCTGATCGAAATCCAGTGCATCTCTCGTGTTTACTC), and 156 (primer: CGTTCACACCGTTTGCGACCCGCTGTTCG). Gene sequences were confirmed by sequencing (Genewiz, NJ).

Expression and purification of the 8S Cav1 complex

Protein expression and purification were conducted as described before () with minor modifications. Cav1 was expressed in E. coli BL21 using the autoinduction expression system (). The MDG starter of monoclonal bacteria cultured at 37°C and 250 rpm for 20 hours was transferred into the autoinducing ZYM-5052 medium and incubated at 25°C and 300 rpm for 24 hours. The harvested cells were first washed with 0.9% NaCl and then resuspended in a buffer composed of 200 mM NaCl, 20 mM tris-HCl (pH 8.0), 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1 mM dithiothreitol (DTT). The cell suspension was homogenized using a French press pressure homogenizer and centrifuged at 9000 rpm for 15 min at 4°C to remove large cell debris. The supernatant was then centrifuged at 40,000 rpm (Ti-45 rotor) to pellet the membrane fractions. Membrane pellets were homogenized in a buffer composed of 200 mM NaCl, 20 mM tris-HCl (pH 8.0), and 1 mM DTT using a Dounce tissue grinder. Solubilization of the membrane was performed by adding n-dodecyl β-d-maltoside (C12M, Anatrace, Ohio) to a final concentration of 2% and by incubating with gentle agitation for 2 hours at 4°C. After centrifugation at 42,000 rpm (Ti-50.2 rotor) for 35 min to remove insoluble material, the supernatant was used for nickel Sepharose–based affinity purification. Specifically, the supernatant was incubated with nickel-charged Sepharose (Chelating Sepharose Fast Flow, GE Healthcare, Illinois) that was pre-equilibrated with buffer of 200 mM NaCl, 20 mM tris-HCl (pH 8.0), 0.05% C12M, 10 mM imidazole, and 1 mM DTT for 2 hours at 4°C. After incubation, the resin-supernatant mixture was poured into a column and was washed with 30 mM imidazole in the above-described buffer. The protein was then eluted using the buffer supplemented with 300 mM imidazole. The elutions with caveolin proteins were concentrated and further separated by size exclusion chromatography (SEC) by using a Superpose 6 Increase 10/300 GL column (GE Healthcare, Illinois) equilibrated with a buffer composed of 200 mM NaCl, 20 mM tris-HCl (pH 8.0), 1 mM DTT, and 0.05% C12M. Similar to our previous report (), two major peaks were observed in the SEC profile, both of which were enriched with Cav1 proteins (fig. S1). The P1 fractions containing the fully oligomerized 8S complexes were diluted from ~1.1 mg/ml to a final concentration of ~0.11 mg/ml for cryo-EM experiments.

Electrophoresis and Western blotting

SDS–polyacrylamide gel electrophoresis and Western blotting were performed as described previously (). Mouse anti-Cav1 polyclonal antibody (catalog number 610406, BD Biosciences, San Jose, CA) and rabbit anti–His-Tag (catalog number A1138-50, BioVision, California) were used with 1:2000 times dilution. Secondary antibodies and blocking buffer were obtained from LI-COR, and imaging was performed using the LI-COR Odyssey system (LI-COR Biosciences, Nebraska).

Negative stain imaging and data processing

Negative stain EM was done using established methods (). Briefly, 200-mesh copper grids covered with carbon-coated collodion film (Electron Microscopy Sciences, Hatfield, PA, USA) were glow-discharged for 30 s at 10 mA in a PELCO easiGlow glow discharge unit (Fresno, CA, USA). Aliquots (3.5 μl) of purified CAV1-myc-his (50 μg/ml) or Cav1-R54A-myc-his (30 μg/ml) were adsorbed to the grids and incubated for 1 min at room temperature. Samples were then washed with two drops of water and stained with two successive drops of 0.7% (w/v) uranyl formate (EMS, Hatfield, PA, USA) followed by blotting until dry. Samples were visualized on a Morgagni transmission electron microscope equipped with a field emission gun operating at an accelerating voltage of 100 keV (Thermo Fisher Scientific Inc., Waltham, MA, USA) at a nominal magnification of ×22,000 (2.1 Å per pixel). The Cav1-R54A-myc-his negative stain dataset was collected using a Tecnai Spirit T12 transmission electron microscope operated at 120 keV (Thermo Fisher Scientific, Waltham, MA, USA) at a nominal magnification of ×26,000 (2.34 Å per pixel). Sample data were collected using Leginon software on a 4 k × 4 k Rio complementary metal-oxide semiconductor camera (Gatan, Pleasanton, CA) at −1.5-μm defocus value. Images were manually curated. All data processing was performed in Relion 3.1.0 (). About 1000 particles were picked manually and 2D classified. Clear resulting classes were selected as references for particle selection on all images. Particles were extracted with a 128 pixel box size (30 nm by 30 nm) and then underwent 2D classification. The Cav1-R54A-myc-his dataset consisted of 54,874 particles.

Cryo-EM sample preparation

For cryo-EM, 4 μl of the protein sample (~0.11 mg/ml) was applied to Quantifoil 400-mesh Au grids (Electron Microscopy Sciences) that were glow-discharged for 30 s at 5 mA. The sample was then vitrified by plunge-freezing in liquid ethane slurry using a Vitrobot Mark IV robotic plunger (Thermo Fisher Scientific). The chamber was set to 100% humidity, and sample grids were blotted for 5 s with a blot force of 10.

Cryo-EM data collection

All images were collected on a Glacios transmission electron microscope (Thermo Fisher Scientific) operated at 200 keV and equipped with a K2 Summit direct electron detector (Gatan) at a nominal pixel size of 0.98 Å. The total exposure time was 6 s, and frames were recorded every 0.2 s, giving an accumulated dose of 55.5 e− Å−2 using a defocus range of −1.5 to −2.2 μm. Images were acquired using Leginon software ().

Image processing

All image processing, classification, and map refinements were done in cryoSPARC (). After image acquisition, 984 movie frames were dose-weighted and locally corrected for beam-induced drift using the patch motion correction. The contrast transfer function (CTF) parameters were estimated locally over each micrograph using the patch CTF estimation procedure. We first performed non–template-based particle picking on 100 images with the Blob Picker. The resulting 47,892 particles were used as input into the Topaz Train routine () to generate a neural network model of the particle data. Following the training procedure, the particle model was then input into the Topaz Extract routine over all micrographs, which identified 176,936 particle locations. Particles were extracted into 400-pixel2 boxes (392 Å by 392 Å). The number of the particles dropped to 140,242 after discarding particles at the micrograph edges. Following two successive rounds of 2D classification, 95,261 particles were selected and used for ab initio reconstruction, searching for two classes. One of the classes consisting of 60,615 particles produced a low-resolution map with features agreeing with the previous negative-stain EM studies (, ). The ab initio reconstruction of Cav1 along with several 2D class averages displayed a C11 symmetrical structure of the 8S complex (fig. S3). For this reason, the ab initio reconstruction and its associated particles were then submitted to the nonuniform refinement procedure, followed by local refinement using a mask encapsulating the protein only and imposing C11 symmetry (). The masked Fourier shell correlation (FSC) resolutions from both nonuniform and local refinement were 3.8 and 3.5 Å, respectively. The final refined B-factors from the nonuniform and local refinements were 174.5 and 154.6 Å2, respectively. Gold standard FSC calculations using the FSC Server (European Molecular Biology Laboratory - European Bioformatics Institute) of the local refinement half-maps indicated a global resolution of 3.5 Å at a 0.143 cutoff (table S1).

Model building, refinement, and validation

The density map of the 8S caveolin complex was of sufficient quality for building a de novo model of protein. The regions with well-resolved secondary structures (residues 80 to 175) for one of the protomers were built using iterative cycles of DeepTracer (), RosettaRelax (, ), and RosettaES (). Specifically, we first manually segmented the map to obtain the putative monomeric region that corresponds to Cav160–178. The initial prediction of the amino acid coordinates was performed with DeepTracer () using the human Cav1 sequence range 70 to 178. Next, the model generated by DeepTracer was relaxed with RosettaRelax (, ) using cryo-EM densities as restraints to build the side chains. Two helical regions that had the most distinctive cluster of aromatic amino acids based on the sequence (Cav189–107 and Cav1111–130) were built manually as ideal α helices. The modeled structures were relaxed into the density domains corresponding to these regions with RosettaRelax with cryo-EM restraints. The two helices were combined in a single PDB file, and this file was used as the template for the following RosettaES () calculations. RosettaES was used to build the Cav180–178 model in two steps. First, the input consisting of the two helices was used as the template to build the three missing segments of the structure with predominantly helical secondary structures (Cav180–89, Cav1108–110, and Cav1130–163). Second, the input from the first RosettaES modeling step was used to build the residues between 163 and 178 that have less well-defined secondary structures using the same parameters. The resulting structure corresponded to Cav180–175.

The remainder of the model (residues 49 to 79) was built manually using Coot (v0.9.5) (). The model for the protomer was expanded into undecamer using the C11 symmetry, followed by the refinement of the model using Phenix Real Space Refinement () and Phenix-Rosetta (). We repeated build-refine iterations until a satisfactory model was obtained. The resolution of the model was estimated by FSC against the map used to construct it within the Phenix Cryo-EM Validation tool. MolProbity scores, Clashscores, and Ramachandran plots were calculated using MolProbity (table S1) (). Programs used for structure determination and refinement were accessed through SBGrid (). Figures were prepared using Chimera (v1.15) (), ChimeraX (), and PyMOL (v2; Schrödinger, LLC).

Search for structurally homologous proteins

Four servers were used to investigate similar motifs to that of Cav1 using the Cav149–178 structure as the input: the PDBeFold server (), DALI (distance matrix alignment) server (), RUPEE (run position encoded encodings of residue descriptors) server (), and FATCAT server (). In addition to these analyses, membrane protein structures from the OPM (Orientations of Proteins in Membranes) database () were investigated. A total of 508 entries classified as “bitopic membrane proteins” and 1574 entries classified as “monotopic/peripheral” were visually inspected to identify motifs similar to Cav1. The only match found by the PDBeFold server was toxofilin (2Q97.T) () with a Q value of 0.051. When the sequences were compared with pBLAST, no similarity was found even at a threshold level of 1000, suggesting that the scaffold similarity is not related to the amino acid sequence. The DALI server search results yielded long α helices or helix-turn-helix motifs, but visual inspection of these hits showed no other similarity to the unique structural motifs of Cav1. The RUPEE server search resulted in 400 structures, but none of these structures had TM (template modeling)-score values above 0.5 that is accepted as a cutoff to consider two structures similar. The FATCAT server search resulted in 4222 hits, but almost all structures consisted of a single long α helix with no sequence similarity. The exception was 6VQ6.M (), which showed two helices connected by a break and a coiled C terminus, but it had no sequence similarity to Cav1.

The SCOP (Structural Classification of Proteins) database (, ) was manually inspected through a keyword search for proteins that have barrel motifs. Of the resulting 89 families, the 2AO9 entry that belongs to the “BC1890-like family” showed similarity to the Cav1 structure in terms of the presence of a parallel β barrel formed by multiple protomeric units and, thus, was considered for our structural comparisons.