Cross-striations protrude from rootlets and connect to intracellular membranes.

Schematic of photoreceptor cell isolation from mouse eyes. Insets show the interface between photoreceptor cells and retinal pigment epithelium (RPE). (B) Low magnification cryo-EM micrograph of the isolated sample. The black square depicts the location of acquisition of panel C. (C) Slice through a denoised and isotropically reconstructed cryo-electron tomogram. The vertical line indicates the position of the cross-section in panel D. (D) cross-section of panel C, with the position of panel C indicated by a vertical line. The rootlet is highlighted in orange. (E) Tomogram segmentation using Eman2 TomoSeg (Chen et al, 2017) with filaments displayed in orange on the left half and hidden on the right. (F) Example of a tomogram slice with membrane connections, and their corresponding segmentation, showing direct membrane connections and connections via membrane-associated proteins with grey and white arrows, respectively.

Cryo-ET analysis of rootlet striations.

Schematic depiction of rootlet purification by membrane removal and gradient centrifugation (not shown). (B) Low-magnification cryo-EM micrograph of a purified rootlet and associated ciliary cytoskeleton. (C) Negative-stain EM of a purified rootlet, highlighting features visible on the rootlet surface. (D) Cryo-ET projection image of a purified rootlet. The Fourier-filtered and thresholded striations are colored according to their appearance: D-bands in yellow, and A-bands in green. Mean values of their spacing and the location of the centriole are indicated below, based on Fig S2H–J. (E) Central slice in a denoised and isotropically reconstructed electron tomogram showing two rootlet sub-fibers. (F) Example of fine features of D bands in a cryo-ET slice and its segmentation Example where D1 aligns with D2 of a neighboring sub-fiber. (H) Segmentation of the striations in the tomogram from panel D. D1-D1 contact of two sub-fibers is indicated by a white arrow. (J) Segmentation of amorphous material on the rootlet surface. The side view is shown in FigS3H (H, J) The position of the A and D-bands is shown by lines in the background. (E,H,J) Black arrows indicate the space between sub-fibers. (F,G,I) features not picked up by the automated segmentation were drawn with dotted lines.

Rootlet filaments are highly flexible and occur as coiled-coil dimers.

(A) Semi-automated segmentation of a rootlet tomogram. The inset shows a 265 nm long model of stitched rootletin AlphaFold predictions. Filaments that show splaying and merging were manually highlighted in pink. (B, C) Tomogram slices and their corresponding segmentations. (B) Example of thick filaments splaying into thin filaments, each indicated by arrowheads as in the legend of panel C. (C) Example of filament melting pointed out by red arrowheads. (D) Schematic of the location along the rootlet where particles were extracted. (E) The initial average after alignment of particles with a wide spherical alignment mask. (F) The initial average of particles aligned with a narrower cylindrical mask. (G) A class average of particles aligned and classified with a narrow mask. The PDB structure (PDB:2TMA,) of two lamin tetramers is shown in red and fitted in the class average.

Model of rootlet organisation.

(A) Filaments are depicted as rootletin coiled-coils based on the 265 nm long AlphaFold prediction. Rootletin coiled-coils are shown in different shades of orange for clarity (B) Branchpoints are found where thicker filaments splay into thinner filaments, here indicated as single coiled coils. (C) Rootletin molecules are arranged in a polar manner based on the polarity we observed in the striations. The N-termini point away from the centriole, with an offset of 80 nm. We propose the N-termini align with the A bands (D) Staggered rootletin filaments may be supported via previously reported weak interactions supported by accessory proteins in the D bands (E) D-bands were observed as punctate laterally connected densities associated with filaments. (F) Amorphous densities of the A-bands were occasionally observed to contain two parallel lines. A-band accumulations in purified rootlet suggest they correspond to the membrane interaction sites in cellular tomograms.

Analysis of cross-striations in cellular tomograms.

(A) Slice through a denoised and isotropically reconstructed cryo-electron tomogram of rootlets surrounded by membranes. (B–D) Tomogram slices and corresponding segmentations of rootlet cross-striations connected to membrane directly (white arrowheads), or via membrane-associated proteins (grey arrowheads).

Cross-striation analysis of purified rootlets.

(A) low magnification negative-stain micrographs of purified rootlet sample with thicker and darker areas containing sample impurities and clustered rootlets. (B–D) Negative-stain micrographs that were local contrast normalized for visualisation of both thick and thin regions. Amorphous (A) and Discrete (DC, DA) striations are indicated with green and yellow respectively. (B, C) Examples of the rootlet connected to centrioles. (D) Example of the rootlet tip. (E) Cryo-ET projection image of a purified rootlet tomogram. (F) Fourier transform of panel E. (G) Inverse Fourier transform of dominant frequencies from panel F. The mean distance was obtained from analysis in panel H and I. (H) example of sinusoid fitting to fourier filtered and thresholded striations of rootlet tomograms such as in panel G. (I) Distribution of sinusoid-fit values of 10 rootlets for each type of striation. (J) phase offset in nanometer of the fitted sinusoid waves from each striation.

Segmentation of purified rootlets.

(A) Central slice in a denoised and isotropically reconstructed electron tomogram showing four rootlet sub-fibres. (B) Cartoon of sub-fibre arrangement in panel A. (C) Semi-automated segmentation of the striations in the tomogram from panel A by Eman2 TomoSeg. (D) Quantification of D-band alignment in 48 tomograms. (E) Full segmentation of the tomogram from panel A. (F) Segmentation of amorphous material on the rootlet surface. The side view is highlighted in blue. (G) Full segmentation of the tomogram from Fig 2. (H) Side view of amorphous densities on surface of the rootlet from Fig 2. (I, J) Tomogram slice and segmentation highlighting connections to amorphous surface material.

AlphaFold predictions of rootletin dimeric fragments.

(A) PAE plots of AlphaFold predictions with identical fragments chain A and chain B. (A, I) The green square highlights AA 1–60 with low dimerization confidence. Yellow indicates a confident coiled coil for region 65–175. Blue highlights the rest of the predicted fragment as a coiled coil. (A, II) Green consists of a predicted coiled coil (264–549), PAE scores in yellow show confidence in a short 4 helix coiled coil between residues 444–469 and 517–541 of both chains. Blue shows the rest of the fragment as a predicted 2 helix coiled coil. (A, VI) Green shows the initial predicted coiled coil. Yellow highlights the interactions that are predicted as a 4 helix bundle between two chains of the coiled-coils in green and blue (1438–1445, 1504–1510). (A, VII) green highlights the alpha helices at the C-terminus of rootletin that are predicted to interact (1932–1998). (B) Alphafold predictions coloured according to their pLDDT value. Chain highlights are coloured to correspond approximately to the boxes on the PAE plots

Subtomogram averaging and classification of rootlet filaments.

(A) Initial refined average and a top-view of a section through its filaments. Class averages of filaments within the blue classification mask are shown in the blue box. The surfaces of a cross-section through the filament classes are shown in orange. (B) Average of inplane/rotationally randomized particles originating from the alignments of fig S3A. (C) Class averages of a classification with alignment of particles from Fig S3B. (D, E) Crystal structures of lamin tetramers (Lilina et al, 2020) and rootletin 1108-1317 tetramer (Ko et al, 2020) fitted in the density of class 5 from panel C.