Fig 2: Loss of TBC1D5 expression enhances endosomal levels of the retromer CSC. (A) HeLa cells were transiently transfected with empty GFP vector, GFP-TBC1D5 wild type (WT) or GFP-TBC1D5 R169A/Q204A (RQ) mutant. After fixation, the cells were stained with antibodies against VPS35 and VPS26. Transfected cells are marked with an asterisk. Overexpression of the wild-type TBC1D5 can displace the retromer CSC from membranes. (B) HeLa cells were treated with siRNA to silence TBC1D5 expression. The knockdown cells were mixed with control cells and seeded onto coverslips. After fixation, cells were labelled with anti-TBC1D5 and antibodies against either VPS35 or VPS26. Loss of TBC1D5 expression (in cells marked with an asterisk) results in brighter staining of the retromer CSC proteins. (C,D) HeLa cells treated with siRNA to silence TBC1D5 expression were labelled with antibodies against VPS26, TBC1D5, CIMPR or TGN46 and then imaged using an automated microscope. Loss of TBC1D5 results in ~40% increase in VPS26 fluorescence intensity but does not markedly increase the number of VPS26-positive spots (D). No spots were counted for TGN46 as the morphology of the TGN is not punctate but ribbon-like. P-values for TBC1D5 knockdown versus control: VPS26, 1.2×10-4; CIMPR, 0.0095; TGN46, 0.0037 for total intensity values. The P-values for spot numbers are: 0.08 for VPS26 and 0.23 for CIMPR. Scale bars: 20 µm.

Fig 3: Knockdown of TBC1D5 can enhance retromer function. (A) Three different cell lines were treated with siRNA to knockdown TBC1D5 expression. Following fixation, cells were labelled with antibodies against CIMPR and TGN46 and then imaged using an automated microscope. The fraction of CIMPR present in a TGN46 mask is shown graphically. For each of the cell lines, knockdown on TBC1D5 enhances the colocalisation of CIMPR with TGN46 but only the GFP-VPS35 cells demonstrate statistical significance. Values are mean±s.d. and P-values for knockdown versus control are shown for each cell line. (B) TBC1D5 expression was silenced in cells expressing GFP-VPS35 wild type or the D620N mutant. Following fixation, the cells were labelled with antibodies against Atg9A and TGN46 and then imaged using an automated microscope. The Pearson correlation coefficient for Atg9A-TGN46 mask is shown graphically. For each of the cell lines, knockdown of TBC1D5 enhances the colocalisation of Atg9A with TGN46 but only the GFP-VPS35 wild-type cells demonstrate statistical significance. Values are mean±s.d. and P-values for knockdown versus control are shown for each cell line. (C) HEK293 cells stably expressing APPswedish were treated with siRNA to silence VPS35 or TBC1D5. Cell culture medium was collected and analysed for the Aß peptide by western blotting. Knockdown of VPS35 increases APP processing to Aß but loss of TBC1D5 expression has the opposite effect. The data shown are from two independent experiments that were highly reproducible. Values are mean±s.d. For both the VPS35- and TBC1D5-knockdown conditions, P<0.01 using Student's t-test compared with control. (D) Representative blots of media (for Aß and sAPPß) and lysates (for APP, VPS35, TBC1D5 and the loading controls, GAPDH and tubulin) from C showing the reduction in Aß detected when TBC1D5 is silenced. There is also a reduction in sAPPß.

Fig 4: Tollip translocates to a LAMP1 compartment during mitochondrial stress AQuantitation of Tollip vesicular clusters in SH-SY5Ys treated with 5 µM antimycin A/10 µM oligomycin (AO) for 2 h. The percentage of cells with vesicular clusters was quantified from 7 to 23 cells for each condition from four independent experiments. Statistical significance was assessed using an unpaired t-test. Bars indicate the mean value, and error bars represent SEM.BQuantitation of the prevalence of Tollip clusters in HEK293 WT or Tom1 KO cells treated with AO for 2 h. The percentage of cells with vesicular clusters was quantified from an average of 50 cells for each condition from five independent experiments. Statistical significance was determined using a two-way ANOVA (Sidak). Bars indicate the mean value, and error bars represent SEM.CEndogenous Tollip and LAMP1 colocalise in SH-SY5Y cells treated with AO for 2 h. Antibodies specific to LAMP1 (green) and Tollip (red) were used. Zoom insets represent 2.5× magnification.D, ESH-SY5Y cells were transfected with either wild-type GFP-Tollip (WT), GFP-Tollip lacking its N-terminus (?Nterm) or GFP-Tollip containing a CUE domain mutation (CUEmut) and then treated with AO for 2 h. Cells were fixed and immunostained with antibodies specific to GFP (green) and LAMP1 (red). Zoom insets represent 2.5× magnification.FThe extent of colocalisation between GFP-Tollip and LAMP1 from untreated and AO-treated SH-SY5Y cells was quantitated from immunofluorescence images and represented as Pearson's correlation (2–3 cells per experiment, from two independent experiments). Statistical significance was determined using an unpaired t-test. Bars indicate the mean, and error bars represent the SEM.G–IThe extent of colocalisation between GFP-Tollip and LAMP1 in Tom1 KO (G), ATG5 KO (H) and VPS35 KO (I) compared with wild-type (WT) HeLa cells was quantified from immunofluorescence images immunostained for GFP and LAMP1 and represented as Pearson's correlation (2–3 cells per experiment, from three independent experiments). Statistical significance was determined using a two-way ANOVA (Sidak). Bars indicate the mean, and error bars represent SEM. Western blot analysis of lysates harvested from wild type (WT) alongside either Tom1 KO (G) or VPS35 KO (I) HeLa cells was performed using antibodies specific to the indicated proteins.Data information: Images were captured at 100× magnification. Scale bars represent 10 µm. Source data are available online for this figure.

Fig 5: VPS35 D620N and WASH1 depletion impair ATG9A trafficking to autophagosomes.(a) HeLa cells stably expressing GFP-VPS35 WT and D620N were transfected with mRFP-LC3 for 24 h, immunostained for endogenous ATG9A, and imaged by confocal microscopy. (b) Colocalization is expressed in terms of Mander’s coefficient M1 to indicate the proportion of LC3 intensities that also contain ATG9A intensities. A representative experiment of three independent experiments is shown, in which at least 33 cells were analysed per condition. Error bars indicate s.e.m., and ***P<0.001 by 2-tailed Student’s t-test. (c) HeLa cells depleted of WASH1 were transfected with GFP-LC3 for 24 h, immunostained for endogenous ATG9A, and imaged by confocal microscopy. (d) Colocalization expressed as M1, as in b. A representative experiment of two independent experiments is shown, in which at least 24 cells were analysed per condition. Error bars indicate s.e.m., and ***P<0.001 by 2-tailed Student’s t-test. Scale bars in (a,c), 10 µm. In b,d, colocalization was measured for the entire area of the cell slice in each image. In order to ensure that results were not confounded by diffuse nuclear staining often observed in the LC3 channel, the analyses were repeated with the nuclei excluded, and all of the results remained statistically significant.