Extended Data Fig. 2: Mechanisms of AKNA localization.

a, b, Electron microscopy micrographs showing AKNA immunogold-labelling at SDAs in the SVZ (a) and ventricular zone (b) of E13 cerebral cortex sections; b′, a magnification of the boxed area showing an SDA in b (n = 3 embryos). c, Stimulated emission depletion nanoscopy image showing AKNA immunofluorescence signal surrounding ODF2 immunofluorescence signal; this reveals the more-distal localization of AKNA at the SDA, as compared to ODF2, which localizes proximally with respect to the centriole in E14 brain NSCs (summarized in the schematic below; AKNA, green, ODF2, magenta) (n = 2 independent experiments). d, Micrographs showing DNA points accumulation for imaging in nanoscale topography super-resolution images of AKNA and TUBA, illustrating that AKNA is present along microtubules in E14 cortical cells (n = 3 independent experiments). e, Coomassie staining of a sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) gel showing samples of purified in vitro polymerized tubulin, with and without addition of purified recombinant AKNA as indicated (n = 2 independent experiments). Note that full-length AKNA co-precipitates with polymerized tubulin (lane 4), showing that AKNA is able to bind microtubules in vitro. f–h, E14 primary cortical cells treated with DMSO (f), nocodazole (g) or DCTN2 overexpression (h) show AKNA immunofluorescence remaining at centrosomes after microtubule depolymerization (n = 4 (f, g) and 3 (h) independent experiments). i, Micrographs showing AKNA immunofluorescence signal at centrosomes at the apical, ventricular surface in sections of E14 wild-type, but not Sas4^−/−p53^−/− (Sas4 is also known as Cenpji, and p53 is also known as Trp53) mice that lack centrioles⁷ (n = 2 independent experiments). Note that PCNT⁺ pericentriolar material is present in the absence of centrioles in Sas4^−/−p53^−/− mice. j, k, AKNA immunofluorescence of wild-type (j) or CRISPR–Cas9-generated Odf2 knock-out (k) mammary epithelial cells, showing that AKNA is lost from centrioles lacking SDAs (n = 3 independent experiments each). l, Schematic of different truncated AKNA variants used to analyse sub-cellular localization, as indicated to the right. m, n, Micrographs showing examples of the localization of truncated AKNA forms expressed in primary E14 cortical cells. Constructs containing the last 370 amino acids (n) localize to centrosomes, whereas AKNA(1–1080) is not localized at the centrosome (square) and is mainly cytoplasmic (m) (n = 3 independent experiments each). Notably, the clone used⁶ to first characterize AKNA lacked the C-terminal centrosome-targeting part. o, Representative images obtained from total internal reflection fluorescence (TIRF) time-lapse recordings of microtubules assembled in vitro from bovine brain tubulin, labelled with Cy5 in the presence of AKNA(1–1080)–GFP. BRB80 buffer with high salt concentration (100 mM KCl) was used to rule out unspecific binding to microtubules (n = 2 independent experiments). These data demonstrate that AKNA is an integral component of SDAs, is able to bind microtubules and is not recruited to centrosomes by microtubule or dynein–dynactin motors. Scale bars, 0.1 μm (a–c), 2 μm (d), 3 μm (o), 5 μm (f–h, m), 10 μm (i–k, n).