Fig 2: Active RHO and FAM65A mediate Golgi reorientation by inhibiting MST3 and MST4 function. (A) RHO inhibition impairs Golgi reorientation towards the direction of cell migration. Mock or TAT-C3 (C3)-treated migrating HeLa cells in a wound healing assay were fixed and immunostained with an anti-GM130 antibody (red) and DAPI (blue) before confocal analysis. Transmission light microscopy was used to visualise the boundaries of the cells. Scale bars: 10 µm. Arrows indicate the direction of the Golgi. (B) Quantification of Golgi orientations from A. The percentage of the edge cells with their Golgi reoriented towards the wound were calculated from five wound healing assays (n=5). 111–136 cells were quantified per condition. Error bars=s.d. Significance P-values were calculated using two-tailed heteroscedastic t-test analysis. (C) FAM65A loss mimics the RHO inhibition effect on Golgi reorientation, whereas double loss of MST3 and MST4 (MST3/4 KO) rescues this effect. Golgi reorientation in wild-type (WT), FAM65A, MST3, MST4 and double MST3 and MST4 CRISPR knockout (KO) HeLa cells was analysed as described in A. Scale bars: 10 µm. Arrows indicate the direction of the Golgi. (D) Quantification of Golgi orientation from C. The percentage of the edge cells with their Golgi reoriented towards the wound were calculated from five wound healing assays per condition (n=5). 85–153 cells were quantified per condition. Error bars=s.d. Significance P-values were calculated using two-tailed heteroscedastic t-test analysis; n.s., not significant (P>0.05).

Fig 3: RHO activation leads to relocation of MST4 from the Golgi to cytoplasmic punctae in a FAM65A-dependent manner. (A) A fraction of MST4 localises to the Golgi in serum-starved cells but relocates from the Golgi upon serum stimulation. HeLa cells were seeded at low density and serum-starved for 24 h before being stimulated as indicated with 10% FBS (15 min). Cells were then fixed and immunostained with anti-GM130 (Golgi marker protein) and anti-MST4 antibodies before analysis with confocal microscopy. Scale bars: 10 µm. (B) FAM65A colocalises with MST4 in the Golgi of serum-starved cells and relocates to cytoplasmic punctae along with MST4 upon stimulation. HeLa cells were seeded at low density and serum-starved for 24 h before being stimulated as indicated with 10% FBS (15 min). Cells were then fixed and immunostained with the indicated antibodies before confocal analysis. Scale bars: 10 µm. (C) MST4 relocation from the Golgi is dependent on RHO activity. Wild-type (WT) HeLa cells were seeded at low density and serum-starved for 24 h before being treated with TAT-C3 (C3) for 4 h and stimulated with 10% FBS (15 min) as indicated. Cells were fixed and immunostained with the indicated antibodies before confocal analysis. Scale bars: 10 µm. (D) RHO-induced MST4 relocation from the Golgi is dependent on FAM65A. FAM65A CRISPR knockout (KO) HeLa cells were seeded at low density and serum-starved for 24 h before being treated with TAT-C3 for 4 h and stimulated with 10% FBS (15 min), as indicated. Cells were fixed and immunostained with the indicated antibodies before confocal analysis. Scale bars: 10 µm. (E) Quantification of colocalisation in C and D. Pearson correlation coefficients between green (MST4) and red (GM130) channels were calculated and averaged from a minimum of five independent fields of view (n=5) from three independent experiments, each comprised 2–12 cells per field. Error bars=s.d. Significance P-values were calculated using two-tailed heteroscedastic t-test analysis; n.s., not significant (P>0.05).

Fig 4: CCM3 acts as an adaptor linking MST kinases to FAM65A. (A) Depletion of CCM3 does not fully abrogate MST activity. HeLa cells were transfected with non-targeting control (NT) or the indicated siRNA pools. At 72 h post transfection, cells were lysed and analysed by immunoblotting with indicated antibodies. Despite a reduction in the total levels of MST3 and MST4, active phosphorylated MST kinases (pGCKIII) were still present in CCM3-depleted cells. (B) RHO-induced MST4 relocation from the Golgi is dependent on CCM3. HeLa cells that had been transfected with CCM3-specific or non-targeting control siRNA pools were seeded at low density, before being serum-starved for 24 h. Cells were treated with TAT-C3 (C3) for 4 h and stimulated with 10% FBS (15 min) as indicated, before being fixed and immunostained with the indicated antibodies for confocal analysis. Scale bars: 10 µm. (C) Quantification of colocalisation from experiments shown in B. The Pearson correlation coefficients between green (MST4) and red (GM130) channels were calculated and averaged from a minimum of three independent fields of view (n=3) from three experiments, each comprised 2–11 cells per field. Error bars=s.d. Significance P-values were calculated using two-tailed heteroscedastic t-test analysis; n.s., not significant (P>0.05). (D) Interaction of MST3 and MST4 with FAM65A is CCM3 dependent. HeLa cells were serially transfected with CCM3-specific or non-targeting control siRNA pools followed by expression vectors for GFP–FAM65A or GFP-only as control and subjected to lysis and immunoprecipitation (IP) with anti-GFP antibody. Input lysates (WCL) as well as anti-GFP immunoprecipitation eluates were analysed by immunoblotting (IB) using the indicated antibodies. Depletion of CCM3 abrogates the MST3 and MST4 (MST3/4) interaction with FAM65A. Quantification of MST3 and MST4 (MST3/4) levels in each immunoprecipitation condition relative to the input is displayed on the right-hand side of the blots (arbitrary units). Quantification was performed on three independent experiments. Error bars=s.d. Significance P-value was calculated using two-tailed heteroscedastic t-test analysis. (E) The proposed mechanism for regulation of Golgi reorientation by RHO. When RHO proteins are inactive, the FAM65A–CCM3–MST complex is localised to the Golgi, where MST proteins act to inhibit reorientation. Upon RHO activation, the FAM65A–CCM3–MST complex is relocated away from the Golgi owing to its interaction with RHO, thus relieving the inhibitory effect of MST on Golgi reorientation.

Fig 5: FAM65A mediates directional migration by inhibiting MST3 and MST4 function. (A) Loss of FAM65A impairs directional migration, whereas double loss of MST3 and MST4 does not affect directional migration. Migrating wild-type (WT), FAM65A and double MST3 and MST4 (MST3/4) CRISPR knockout (KO) HeLa cells in a wound healing assay were then analysed for 24 h by using time-lapse microscopy. Images show the wound area at indicated timepoints. (B) Quantification of the wound closure from experiments shown in A. The average distances migrated by the wound edges were calculated from three wound healing assays per condition (n=6). Error bars=s.d. Significance P-values were calculated using two-tailed heteroscedastic t-test analysis; n.s., not significant (P>0.05). (C) Impairment of directional migration upon FAM65A depletion is rescued by double loss of MST3 and MST4. Migration of WT or MST3/4 CRISPR knockout (KO) HeLa cells that had been transfected with FAM65A-specific or non-targeting (NT) siRNA pools was analysed as described in A. Images show the wound at indicated timepoints. (D) Quantification of the wound closure from the experiments shown in C. The average distances migrated by the wound edges were calculated from three wound healing assays per condition (n=6). Error bars=s.d. Significance P-values were calculated using two-tailed heteroscedastic t-test analysis; n.s., not significant (P>0.05). NTi, non targeting siRNA; FAM65Ai, siRNA targeting FAM65A.