Cryo-electron tomography pipeline for plasma membranes

Abstract

Cryo-electron tomography (cryoET) provides sub-nanometer protein structure within the dense cellular environment. Existing sample preparation methods are insufficient at accessing the plasma membrane and its associated proteins. Here, we present a correlative cryo-electron tomography pipeline optimally suited to image large ultra-thin areas of isolated basal and apical plasma membranes. The pipeline allows for angstrom-scale structure determination with subtomogram averaging and employs a genetically encodable rapid chemically-induced electron microscopy visible tag for marking specific proteins within the complex cellular environment. The pipeline provides efficient, distributable, low-cost sample preparation and enables targeted structural studies of identified proteins at the plasma membrane of mammalian cells.

Fig. 1. Generating isolated plasma membranes on EM grids.

a Diagrams showing the workflow for isolating basal plasma membranes. A plasma-cleaned grid is placed onto a coverslip and secured with a PDMS stencil. Cells are then seeded onto the grid and incubated overnight. To generate isolated basal plasma membranes, the grid is placed under a pressurized fluid-delivering device and the grid is sprayed with unroofing buffer to wash away the apical portions of cells with a shearing force. b Diagrams showing the workflow for isolating apical plasma membranes. Cells are first seeded onto a coverslip and incubated overnight. A plasma-cleaned, poly-L-lysine coated grid is then brought into contact with the coverslip to pick up cells. Pressurized fluid is then applied to the grid to wash away the basal portions of cells to generate isolated apical plasma membranes.

Fig. 2. Evaluating isolated plasma membranes of HSC-3 cells on EM grids with platinum replica electron microscopy.

a A cartoon showing the generation of platinum replicas of the isolated plasma membranes on grids. b An isolated HSC-3 cell basal plasma membrane on an R2/1 Quantifoil grid. The inset shows an enlarged view of the white dashed square area. Structural classes of clathrin are color-coded: lilac = flat; cyan = dome; green = sphere. c An isolated HSC-3 cell apical plasma membrane on an R2/1 Quantifoil grid. The inset shows an enlarged view of the white dashed square area. Color-coding is as in (b). d A close-up view of the isolated basal plasma membrane showing the three classes of clathrin structures. The lilac arrow points to a flat clathrin structure, the cyan arrow points to a dome-shaped clathrin structure, and the green arrow points to a spherical clathrin structure. e The edge of the hole (white dashed circle) is used as the reference point to evaluate the distribution of the different classes of clathrin-coated structures across the changing grid surface (1000 nm range, yellow circles inside and outside of the hole denote a 500 nm distance from the hole edge). f, g Comparison of the distribution of flat, dome, and sphere clathrin-coated structures (mean ± SEM) with respect to the edge of the hole between basal (f) and apical (g) isolated plasma membranes. h, i are box and whisker plots (box: 25th–75th percentile; whiskers: min to max) with individual data points shown to the right. h Comparing the density of different classes of clathrin structures between isolated basal and apical plasma membranes, and (i) of the projected area of individual clathrin structures grouped by structural class (Mann-Whitney test, two-tailed). ***p = 0.0002; ****p < 0.0001. For (h, i), horizontal lines = median. Images are representative of N = 4 grids (basal) and N = 4 grids (apical). For (f–i), N = 16 membranes, 203 flat, 237 domed, and 404 spherical clathrin structures from 2 grids (basal) and N = 16 membranes, 139 flat, 206 domed, and 173 spherical clathrin structures from 1 grid (apical).

Fig. 3. Unroofed cells provide 100–200 nm thick plasma membrane samples for cryoET.

a A diagram highlighting unroofing, addition of fiducial markers, back-blotting, and plunging in liquid ethane for vitrification. b A 2250x magnification montage of a grid square containing an isolated HSC-3 basal membrane (outlined with an orange dotted line). c A 2250x magnification montage of a grid square containing an isolated apical HSC-3 membrane (outlined with an orange dotted line). d (top) shows a minimum-intensity projection along the Z-axis through 21 slices of a Gaussian-smoothed bin-8 tomogram acquired at the location of the arrow in (b). d (middle) shows a minimum-intensity projection of 101 slices through the Y-axis of the same tomogram with the measured thickness shown. d (bottom) shows a segmentation (mask-guided isosurface) of the above tomogram: gray = membrane, purple = ribosomes, blue = actin, light green = clathrin, orange=intermediate filaments. e Same as (d), but for the apical membrane tomogram acquired at the position of the black arrow in (c). f Histogram of tomogram thicknesses of HSC-3 basal and apical membranes imaged here. N_apical = 56, N_basal = 61, 2 grids represented for each.

Fig. 4. Subtomogram averaging and contextual analysis show sub-nanometer detail is preserved in isolated plasma membranes.

a Projection of 21 z-slices from a tomogram of an isolated plasma membrane representative of the 111 tomograms used in this analysis. 80S ribosomes (black arrows) are frequently found in unroofed HEK293 cells overexpressing dynamin-1(K44A). b Rotated views (top) and a clipped view (bottom) of a consensus subtomogram average filtered according to the local resolution. c Fourier shell correlation (FSC) profiles obtained from subtomogram averages. The nominal resolution is reported at FSC = 0.143. d, e Classification of the set of well-aligning particles obtained subtomogram averages of the 80S ribosome in non-rotated and rotated states. In (d), tRNA occupying the P, P/E, and A/P sites are indicated in orange, pink, and blue, respectively. A subset of 446 membrane-bound ribosomes is shown (two views, (e)). f A view from a tomogram with the top and bottom air-water interfaces (AWIs) indicated (black arrows). g Distances of putative ribosomes were measured from both AWIs. The number of particles within successive 5 nm bins from the bottom and top AWIs (left and right panels, respectively) is plotted for the set of particles obtained immediately following picking (cyan) and the set of well-aligning particles obtained from classification (purple). h The fraction of particles found with particle picking that constitute the well-aligning class are plotted with respect to their distance to the bottom and top AWIs. Dashed lines in (g, h) indicate 25 nm distances from each AWI. This condition is represented in grids 5-6, all ribosome data are from grid 5 (Table 1).

Fig. 5. CLEM finds sites of arrested clathrin-mediated endocytosis (CME).

a Select portion of grid shown as a low magnification cryoEM image registered with cryo-fluorescence images of Dyn1(K44A)-GFP (green) and 500 nm fiducial markers (red). b The grid square highlighted in (a) (white square) is shown in a higher resolution map. Orange dots indicate the outline of the isolated plasma membrane. c The fluorescence overlay is shown. d The black box from (c) is shown enlarged. Black boxes indicate the location of tilt series acquisition for the tomograms shown in (e). e Examples of tomograms in XY (left) and XZ (right). f Examples of arrested CME sites with Dynamin 1 (K44A) tubules. In (e, f), orange and black arrows point to putative clathrin and dynamin densities, respectively. g Examples of arrested CME sites with large clathrin-decorated clusters. XY tomogram images in (e–g) are minimum intensity projections (mIPs) over 21 Gaussian-smoothed XY slices while XZ images are mIPs of 101 XZ slices. h A subtomogram average of clathrin (EMD-46973) is shown with two representative thresholds (left and middle columns, 0.14; right column, 0.08). Labels indicate putative proteins and domains contributing to observed densities. A rigid body fit of PDB ID:6YAI is shown (bottom row) within the subtomogram average showing the clathrin heavy chain domains proximal to the central vertex (blue), clathrin light chain (yellow), clathrin heavy chain N-terminal domain with distal leg (pink), and the β2 appendage of the adapter AP2 (orange). Magnified views of the heavy chain N-terminal domain (bottom-middle) and the β2 appendage are shown (bottom-right). i The clathrin vertex reconstruction is superimposed onto refined subtomogram positions reconstructing a clathrin coated pit (green) representative of those found in the 20 tomograms used in this analysis. Several lattice arrangements are visible within the single structure. The image is an average of 10 XY slices from a denoised tomogram. This condition is represented in grids 5-6 (Table 1). All clathrin data are from grid 6.

Fig. 6. Iron-Free FerriTag is specific, efficient, and visible in cryoET.

a TIRF microscopy of a HEK293 cell expressing FerriTag (FRB-mCherry-FTH1, magenta, and FTL) and Hip1R-GFP-FKBP (green) shown before (left) and 30-60 seconds after (right) rapamycin addition. Representative of 2 coverslips, 15 regions, 24 cells. b Cryo-fluorescence microscopy of HEK293 isolated plasma membrane on a grid expressing FerriTag (FRB-mCherry-FTH1, magenta; and FTL) and Hip1R-GFP-FKBP (green) with combined colors on the right. c The same grid in low-magnification CryoEM (left), reflection image used for fluorescence registration (middle), and combined EM and fluorescence image (right). Orange dots = membrane outline, arrows = position of (d) acquisition. d Tomogram of Hip1R/FerriTag labeling (right). Empty FerriTag structures are visible as 12 nm circles (arrow) (Supplementary Movie 1). e Selected tomogram from HEK293 cells with labeled FerriTag on GFP-FKBP-LCa (right)(Supplementary Movie 2). The cartoons depict membrane (gray), clathrin (green), and FerriTag (purple). f A histogram (5-nm bins) of the FerriTag distance from clathrin-coated membrane (shown at 0–600 nm and 0–80 nm) for clathrin light chain (GFP-FKBP-LCa, grid 7-8, N = 784 FerriTags) and Hip1R (Hip1R-GFP-FKBP, grid 9, N = 2057 FerriTags). Molecular models of Ferritin (PDB-1fha), mCherry (PDB-2h5q), FRB/FKBP (PDB-3fap), GFP (PDB-5wwk), and EM density of the clathrin cage (emdb-21608) at scale with the zoomed-in histogram with cartoons of Hip1R, actin, and membrane as putative models. g An example of FerriTag Hip1R labeling around a clathrin structure (Supplementary Movie 3). Segmentation of membrane (gray), clathrin (green), FerriTag (purple), and actin (blue) are shown to the right. h Two examples (side-by-side) showing Hip1R density adjacent to FerriTag (Supplementary Movie 4-5). The top image displays the clathrin structure with clathrin (orange arrow) and a FerriTag (black arrow) highlighted and enlarged below with cartoons of membrane (light-gray), clathrin (green), Hip1R (dark gray), FerriTag (purple), and actin (blue) shown to aid image interpretation. Tomogram images are minimum intensity projections (mIPs) of 21 (d, e), 10 (g), 20 (h) Gaussian-smoothed XY-slices.