Fig 2: SEPT1 facilitates microtubule nucleation at the Golgi. (A) Staining of a-tubulin and GM130 in control and in SEPT1-depleted cells. (B) SEPT1 localization to GM130-positive Golgi objects in RPE1 cells is not affected by nocodazole administration. Scale bar: 10 µm. (C) GM130- and a-tubulin staining in control and SEPT1-depleted RPE1 cells, fixed at the indicated time points after nocodazole washout. Scale bar: 10 µm. (D) The Pearson's correlation coefficient of a-tubulin and GM130 staining decreases upon depletion of SEPT1, demonstrating that a-tubulin becomes less abundant at early Golgi membranes. Data are represented as mean±s.e.m., unpaired two-tailed Student's t-test, **P<0.01 (N=3; siScr: n=176; siSEPT1: n=96; t=6.613; d.f.=4). (E) Number of microtubules nucleated in 45 s upon nocodazole washout as determined for individual GM130-positive objects in control and SEPT1-depleted RPE1 cells (see also Fig. S7H). Data are represented as mean±s.e.m., unpaired two-tailed Student's t-test, **P<0.01, *P<0.05, (microtubule group size 1: t=4.478; d.f.=4; microtubule group size 2: t=4.652; d.f.=4; microtubule group size 3: t=3.65; d.f.=4; microtubule group size =4: t=3.085; d.f.=4). (F) Model for SEPT1 function at the Golgi. GM130 (in blue) recruits SEPT1 (in red) to cis-Golgi membranes. SEPT1 is organized in a filament-like fashion and serves as an anchor for the association of CEP170 (brown) and ?-tubulin (pink) to nucleate Golgi-derived microtubules (green).

Fig 3: SEPT1 is required for release of secretory cargo at the Golgi. (A) Secretion of the FKBP(F36M)–GFP reporter (Gordon et al., 2010) upon induction with D/D solubilizer ligand for the indicated times in control and SEPT1-depleted cells. HeLaM C1 cells were fixed and analyzed by fluorescence microscopy. Scale bar: 10 µm. Time-course of FKBP(F35M)–GFP secretion as quantified by loss of GFP fluorescence over time relative to control cells. Data are represented as mean±s.e.m., unpaired two-tailed Student's t-test, *P<0.05, **P<0.01 (N=3; 20 min: n=277-291; P-value=0.6534; t=0.4858; d.f.=4; 60 min: n=290-329; t=3.912; d.f.=4; 90 min: n=276-307; t=6.175; d.f.=4). See also Fig. S5. (B) Correlative light and electron microscopy analysis of HeLaM C1 cells transfected with siCtrl or siSEPT1#1, and incubated with D/D solubilizer for 20 min to allow for trafficking of the reporter protein to the Golgi. Left panels indicate light microscopy images (green, GFP-tagged reporter protein; blue, nuclei), right panels depict TEM images, and middle panels show overlays. Scale bars: 10 µm for upper panels and 1 µm for bottom panels. (C) Immunostaining of GM130 in control and SEPT1-depleted 3T3 L1 adipocytes. Depicted are two representative cells for each condition. Scale bar: 10 µm. (D) Western blot analysis of adipocyte lysates upon control or SEPT1 depletion (left). Adiponectin secretion from control and SEPT1-depleted adipocytes, as quantified by ELISA (right). Data were normalized to that in control cells (mean±s.e.m.), paired t-test, *P<0.05 (N=4; t=7.656; d.f.=2).

Fig 4: SEPT1 localizes predominantly to cis-Golgi membranes. (A) HeLa cells were fixed with 2% PFA and immunolabeled for SEPT1 and GM130. Scale bar: 10 µm. See also Fig. S1. (B) SEPT1 is enriched in the Golgi-fraction of HeLa cells (GM130, Golgi; EEA1, endosomes; GAPDH, cytosol). (C) gSTED microscopy analysis of HeLa cells co-immunolabeled with antibodies recognizing SEPT1 and GM130 (cis-Golgi). Scale bar: 500 nm. See also Fig. S2. (D,E) Line scan of SEPT1 localizations along Golgi membranes (D), and fluorescence intensity profile (E). (F) Nearest neighbor analysis of SEPT1 localizations. Histogram depicting the frequencies of distances observed along Golgi membranes (blue bars) or within the nucleus (unspecific staining, red bars). (G) Pearson's correlation coefficients determined from STED microscopy images upon staining for SEPT1 and various Golgi marker proteins (n=10–11). GM130, 0.47; Golgin 97, 0.33; COPI, 0.29; AP-1, 0.17; GGA1, 0.13; ERGIC53, 0.28.

Fig 5: SEPT1 depletion impairs the ultrastructural organization of the Golgi. (A) 3D-SIM images of GM130 (green) and golgin97 (red) labeling in HeLa cells after control or SEPT1 depletion. Scale bar: 5 µm. See also Fig. S5 and Movies 1 and 2. (B) Average number of GM130-positive objects observed after control or SEPT1 depletion. Data are represented as mean±s.e.m., unpaired two-tailed t-test, ****P<0.0001 (n=10; t=5.583; d.f.=18). (C) The average nearest neighbor (NN) distance between GM130-positive objects and the closest Golgin97-positive object remains unaffected upon depletion of SEPT1 (0.149 nm in control and 0.140 in SEPT1-depleted cells). Data are represented as mean±s.e.m., unpaired two-tailed t-test, ns, not significant (P-value=0.7148). (D) Electron microscopy images derived from control or SEPT1-knockdown cells reveal enlarged vacuole-like structures, instead of Golgi ribbon stacks. Scale bar: 2 µm. (E) The Golgi object diameter increases from 33 to 443 nm upon depletion of SEPT1. Data derived from images as depicted in E (siCtrl: n=10; siSEPT1: n=13; for each condition 15 cisternae/vacuoles were analyzed per cell). The box represents the 25–75th percentiles, and the median is indicated. The whiskers show the range.