Microtubule architecture connects AMOT stability to YAP/TAZ mechanotransduction and Hippo signalling

Abstract

Cellular mechanotransduction is a key informational system, yet its mechanisms remain elusive. Here we unveil the role of microtubules in mechanosignalling, operating downstream of subnuclear F-actin and nuclear envelope mechanics. Upon mechanical activation, microtubules reorganize from a perinuclear cage into a radial array nucleated by centrosomes. This structural rearrangement triggers degradation of AMOT proteins, which we identify as key mechanical rheostats that sequester YAP/TAZ in the cytoplasm. AMOT is stable in mechano-OFF but degraded in mechano-ON cell states, where microtubules allow AMOT rapid transport to the pericentrosomal proteasome in complex with dynein/dynactin. This process ensures swift control of YAP/TAZ function in response to changes in cell mechanics, with experimental loss of AMOT proteins rendering cells insensitive to mechanical modulations. Ras/RTK oncogenes promote YAP/TAZ-dependent tumorigenesis by corrupting this AMOT-centred mechanical checkpoint. Notably, the Hippo pathway fine-tunes mechanotransduction: LATS kinases phosphorylate AMOT, shielding it from degradation, thereby indirectly restraining YAP/TAZ. Thus, AMOT protein stability serves as a hub linking cytoskeletal reorganization and Hippo signalling to YAP/TAZ mechanosignalling.

Fig. 1. Microtubule organization serves as a determinant of YAP/TAZ mechanotransduction.

a, Left: representative IF images (3D sections) of MCF10A cells seeded on stiff (40 kPa, mechano-ON, hereafter Mech.ON) versus soft (0.7 kPa, mechano-OFF, hereafter Mech.OFF) hydrogels. Right: quantifications of γ-TURC number in cells seeded as in the left panels for n = 65 (Mech.ON) and n = 51 (Mech.OFF) pooled from two independent seedings. P < 0.0001. b, Left: representative IF images (3D sections) of MCF10A cells seeded on spread (unconfined, Mech.ON) versus small (100 μm² micropatterns, Mech.OFF) substrates. Right: quantifications (n = 50 cells for each condition, pooled from two independent seedings) of γ-TURC number in cells seeded as in the left panels. P < 0.0001. In a and b, MT are labelled by α-tubulin staining (αTub, magenta), γ-TURCs by γ-tubulin staining (γTub, yellow) and nuclei are counterstained with Hoechst (cyan). White arrowheads in Mech.ON panels indicate γ-TURCs convergence into a single MTOC. Scale bars (a,b), 5 μm. Extended Data Fig. 1a and Supplementary Video 1 provide 3D reconstructions of cells seeded as in a and b. Comparable results were obtained with HEK293 and U2OS cells (Extended Data Fig. 1c,d). c, Representative IF images (3D sections) of MCF10A cells seeded on stiff (40 kPa, Mech.ON) versus soft (0.7 kPa, Mech.OFF) hydrogels. MT architecture is labelled by α-tubulin staining (αTub, magenta) and the centrosome by pericentrin staining (PCNT1, grey) and highlighted by a white arrowhead. Nuclei are counterstained with Hoechst (cyan). Scale bars, 5 μm. Extended Data Fig. 1b presents representative IF images (3D sections) of MCF10A cells seeded on spread versus small substrates and stained as in c. Comparable results were obtained with HEK293 and U2OS cells (Extended Data Fig. 1c,d). d, Left: representative IF images of MTs and γ-TURCs (top panels) and EGFP-YAP (bottom panels) in MCF10A cells bearing a YAP-EGFP knock-in^(KI) allele seeded on hydrogels tuned to the indicated stiffness gradient. Images in the top panels are magnifications of the insets shown in the upper right corners. Scale bars, 5 μm. Right: quantifications of γ-TURCs (left, n = 49 cells for 40 kPa, n = 52 for 13 kPa, n = 53 each for all other conditions, pooled from two independent seedings) and YAP nuclear-to-cytoplasmic subcellular localization (N/C, right, n = 63 cells for 40 kPa, n = 51 for 3 kPa, n = 53 for each for all other conditions, pooled from two independent seedings) in cells seeded as in the left panels. P < 0.0001. e, Representative IF images of MTs (αTub) in HEK293 cells transfected with the indicated siRNAs. Nuclei are counterstained with Hoechst (cyan). Scale bar, 10 μm. f, Left: representative IF images of YAP/TAZ (Y/T, grey) in HEK293 cells transfected with the indicated siRNAs. Right: quantifications (n = 51 cells for siγTub#1, n = 50 cells each for all other conditions, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P < 0.0001. g, Left: representative IF images of MTs (αTub), γ-TURCs (γTub) and YAP/TAZ in HEK293 cells transfected with the indicated siRNAs. Nuclei are counterstained with Hoechst (cyan). The white arrowhead indicates MT convergence in the centrosomal MTOC in control cells. Scale bars (f,g), 10 μm. Right: quantifications (n = 50 cells for siCo., n = 52 cells each for all other conditions, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P < 0.0001. h, Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous targets CTGF (left) and Cyr61 (right) in MCF10A cells treated with the indicated siRNAs and seeded in Mech.ON conditions. Data are presented as mean + s.d. of n = 3 biologically independent samples. In the left graph, P = 0.0034 (siCo. versus siγTub), P = 0.0031 (siCo. versus siαTAT1); right graph P = 0.0087 (siCo. versus siγTub), P = 0.0073 (siCo. versus siαTAT1). Extended Data Fig. 2d presents similar results obtained with U2OS cells. i, Representative IF images of MTs (αTub) and γ-TURCs (γTub) in HEK293 cells transduced with empty or NLP1-encoding lentiviruses and treated with cytochalasin D (Cyto.D, 1 μM for 2 h) or DMSO as negative control (Co.). Nuclei are counterstained with Hoechst (cyan). Scale bar, 5 μm. j, Left: representative IF images of YAP/TAZ (grey) in HEK293 cells treated as in i. Right: quantifications (n = 52 cells for each condition, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P < 0.0001. Extended Data Fig. 2f shows controls demonstrating effective F-actin disruption by Cyto.D treatment. P values were determined by unpaired two-tailed Student’s t-test with Welch’s correction (a,b) or one-way analysis of variance (ANOVA) with Welch’s correction (d,f–h,j). Source data

Fig. 2. A mechanical continuum between stress fibres, the NE and MTs dictates YAP/TAZ mechanotransuction.

a, Representative 3D IF reconstructions of MCF10A cells seeded in Mech.ON (40 kPa) versus Mech.OFF (0.7 kPa) conditions. Scale bars, 10 μm. Supplementary Video 2 presents a wrap-around view and Extended Data Fig. 3a representative Z-planes of cells seeded as in a. The nuclear lamina was labelled by laminA staining (LmnA, cyan) and F-actin by phalloidin staining (F-actin, grey). NASFs are highlighted in orange in the 3D reconstruction (Methods). b, Representative 3D IF reconstructions of MCF10A cells treated with the indicated siRNAs. Scale bars, 10 μm. Extended Data Fig. 3d presents representative Z-planes of cells treated with the same siRNAs. The nuclear lamina is labelled by laminA staining (LmnA, cyan) and F-actin by phalloidin staining (F-actin, grey). NASFs are highlighted in orange in the 3D reconstruction (Methods). c, Left: representative IF images of YAP/TAZ (grey) in HEK293 cells transfected with the indicated siRNAs. Scale bar, 10 μm. Right: quantifications (n = 50 cells for each condition, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P < 0.0001. d, qRT–PCR assessing the expression levels of the YAP/TAZ endogenous target CTGF in HEK293 cells treated with the indicated siRNAs and seeded in Mech.ON conditions. Data are presented as mean + s.d. of n = 3 biologically independent samples. P = 0.0087 (siCo. versus siSUN2#1), P = 0.003 (siCo. versus siSUN2#2). Extended Data Fig. 3e presents qRT–PCR results assessing the expression levels of the YAP/TAZ endogenous target Cyr61 in the same experiment. e,f, Representative IF images of MTs (αTub) and γ-TURC (γTub) in HEK293 cells transfected with the indicated siRNAs. Nuclei are counterstained with Hoechst (cyan). The white arrowhead indicates MT convergence in a perinuclear MTOC only in control cells (siCo.). Scale bar, 10 μm. g, Left: representative IF images of YAP/TAZ (grey) in HEK293 cells transfected with the indicated siRNAs. Scale bar, 10 μm. Right: quantifications (n = 50 cells for each condition, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P < 0.0001. h, qRT–PCR assessing the expression levels of the YAP/TAZ endogenous target CTGF in HEK293 cells treated with the indicated siRNAs and seeded in Mech.ON conditions. Data are presented as mean + s.d. of n = 3 biologically independent samples. P < 0.0001. See Extended Data Fig. 4a for qRT–PCR assessing the expression levels of the YAP/TAZ endogenous target Cyr61 in the same experiment. i, Representative IF images of MTs (αTub) and γ-TURCs (γTub) in HEK293 cells transfected with the indicated siRNAs. The white arrowhead highlights MT convergence in a perinuclear MTOC only in control (siCo.) cells. Nuclei are counterstained with Hoechst (cyan). Scale bar, 10 μm. j, Left: representative IF images of YAP/TAZ (grey) in HEK293 cells transfected with the indicated siRNAs. Scale bar, 10 μm. Right: quantifications (n = 50 cells for each condition, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P < 0.0001. k, qRT–PCR assessing the expression levels of the YAP/TAZ endogenous target CTGF in HEK293 cells treated with the indicated siRNAs and seeded in Mech.ON conditions. Data are presented as mean + s.d. of n = 3 biologically independent samples. P = 0.015 (siCo. versus siEMD#1), P = 0.0183 (siCo. versus siEMD#2). Extended Data Fig. 4c presents qRT–PCR results assessing the expression levels of the YAP/TAZ endogenous target Cyr61 in the same experiment. P values were determined by one-way ANOVA with Welch’s correction (c,d,g,h,j,k). Source data

Fig. 3. AMOT acts as a cytoplasmic mechanical rheostat to control YAP/TAZ activity.

a, Representative IF images of HEK293 cells seeded in Mech.ON (stiff, 40-kPa hydrogels) versus Mech.OFF (soft, 0.7-kPa hydrogels or dense culture) conditions. The proteasome was labelled by 20S/PSMA5 staining (proteasome, yellow) and nuclei were counterstained with Hoechst (cyan). Scale bars, 10 μm. Extended Data Fig. 5a presents quantifications of proteasome localization in cells seeded under the same conditions. b, Representative IF images of HEK293 cells seeded in Mech.ON conditions and treated with the indicated siRNAs. The proteasome was labelled by 20S/PSMA5 staining (proteasome, yellow) and nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. Extended Data Fig. 5b presents quantifications of proteasome localization in cells treated in the same way. c, Representative AMOT immunoblot of MCF10A cells seeded on hydrogels of the indicated stiffness. GAPDH serves as loading control. The same experiment was repeated twice with comparable results. Extended Data Fig. 5d,f–h presents AMOT immunoblots from cells experiencing independent Mech.OFF conditions by treatment with F-actin inhibitors, and Supplementary Fig. 1a shows quantifications. d, Representative AMOT immunoblot of HEK293 cells treated with the indicated siRNAs. GAPDH serves as loading control. Supplementary Fig. 1b provides quantifications. The same experiment was repeated twice with comparable results. e, Left: representative images of PLAs showing exclusively cytoplasmic interaction (magenta) between endogenous AMOT and YAP/TAZ in HEK293 cells seeded in Mech.OFF conditions. Mech.ON conditions (absence of endogenous AMOT protein) serve as a negative control. Nuclei are counterstained with Hoechst (cyan). Scale bar, 10 μm. Right: quantification of the mean number of PLA dots per cell (50 cells were quantified for each n = 4 independent experiment). The boxplot shows the interquartile range, and whiskers represent min to max. P < 0.0001. f, Representative stills from live fluorescence images of RFP-AMOT-expressing MCF10A-YAP-EGFP^KI cells showing that AMOT overexpression is sufficient to cause YAP cytoplasmic retention in Mech.ON conditions. Cells without AMOT overexpression (*), showing nuclear YAP accumulation, serve as negative control. Cell and nuclear borders are outlined in yellow and blue dashed lines, respectively. Scale bar, 20 μm. g, Left: representative IF images of YAP/TAZ in control (wt) versus AMOT-130/AMOT-L1/AMOT-L2 triple KO (AMOT tKO) HEK293 cells in Mech.ON versus Mech.OFF conditions. Nuclei are counterstained with Hoechst (cyan). Scale bar, 20 μm. Right: quantifications (n = 51 for wt Mech.OFF, n = 50 cells each for all other conditions, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells seeded as in the left panels. P < 0.0001. h, AMOT is epistatic to LINC proteins and MTs. Left: representative IF images of YAP/TAZ in control versus AMOT tKO HEK293 cells treated with the indicated siRNAs. Scale bar, 10 μm. Right: quantifications (n = 45 for AMOTtKO siCo., n = 50 cells each for all other conditions, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P = 0.0587 for wt versus AMOTtko in siCo., P < 0.0001 for all other comparisons shown. P values were determined by unpaired two-tailed Student’s t-test with Welch’s correction (e) or one-way ANOVA with Welch’s correction (g,h). Source data

Fig. 4. Mechanical regulation of AMOT stability relies on retrograde transport along MTs in mechano-ON cells.

a, Representative AMOT immunoblot of HEK293 cells seeded in Mech.ON conditions and treated with proteasome inhibitor (lactacystin 10 μM) for the indicated timings. GAPDH serves as loading control. b, Representative 3D IF reconstructions of HEK293 cells seeded in Mech.ON conditions treated with proteasome inhibitor (lactacystin 10 μM, 6 h). Centrosome was labelled by pericentrin staining (PCNT1, magenta), AMOT in yellow, and the nucleus was counterstained with Hoechst (cyan). Scale bar, 5 μm. c, Pulldown of endogenous AMOT from proteasome-inhibited HEK293T cells seeded in Mech.ON versus Mech.OFF conditions, showing AMOT interaction with the proteasome component 20S exclusively under Mech.ON conditions. IgG pulldown serves as the negative control. The inputs of the same pulldown experiment are shown in Extended Data Fig. 6g. d, Representative AMOT immunoblots of HEK293 cells treated with the indicated siRNAs targeting components of the dynactin complex. GAPDH serves as loading control. e, Left: representative IF images of YAP/TAZ in HEK293 cells transfected with the indicated siRNAs. Nuclei are counterstained with Hoechst (cyan). Scale bar, 10 μm. Right: quantifications (n = 55 for siCo., n = 50 cells each for all other conditions, pooled from two independent seedings) of YAP/TAZ N/C subcellular localization in cells treated as in the left panels. P < 0.0001. f, Pulldown of endogenous AMOT from proteasome-inhibited Mech.ON HEK293T cell lysates showing AMOT interaction with the dynactin complex component DCTN1. IgG pulldown serves as negative control. g, Structural superposition between the 14-3-3 domain (grey) and intra-coiled-coil domain (ICD) of DCTN1/p150 (magenta). Extended Data Fig. 7j presents the predicted alignment error (PAE) plot for the complex shown here. h, Pulldown of FLAG-DCTN1 from Mech.ON HEK293T cells transiently transfected with HA-AMOT and the indicated FLAG-DCTN1 mutants and treated with dynarrestin (10 μM) for 6 h, showing impaired interaction between AMOT and a DCTN1 mutant bearing S659W/L662W/T835W/A839W substitutions in the ICD (ICD^4W mutant). The inputs of the same pulldown experiment are shown in Extended Data Fig. 7m. i, Pulldown of endogenous DCTN1 from Mech.ON HEK293T cells transiently transfected with the indicated AMOT mutants and treated with dynarrestin 10 μM for 6 h, showing impaired interaction between DCTN1 and the AMOTL178W mutant. The inputs of the same pulldown experiment are shown in Extended Data Fig. 7n. j, Representative IF images of MTs (αTub, magenta) and endogenous AMOT (yellow) in MCF10A cells seeded in Mech.ON conditions and treated with dynarrestin (10 μM) for 6 h. Nuclei are counterstained with Hoechst (cyan). Scale bar (left), 10 μm. The right panel shows a magnification of the dashed area in the left panel. Scale bar (right), 5 μm. Supplementary Video 3 shows live time-lapse videos showing AMOT retrograde transport along MTs in DMSO versus dynarrestin-treated MCF10A cells. P values (P < 0.0001 for siCo. versus all other samples) were determined by one-way ANOVA with with Dunnett T3 correction for multiple comparisons (e). Source data

Fig. 5. Hippo/LATS signalling regulates YAP/TAZ mechanotransduction through AMOT.

a, Representative immunoblots of Mech.OFF HEK293 LATS1/2 dKO cells showing impaired AMOT stabilization upon LATS1/2 double knockout. GAPDH serves as loading control. b, Pulldown of HA-AMOT from Mech.ON HEK293T cells transiently transfected with the indicated HA-AMOT mutants and treated with lactacystin (10 μM) for 6 h, showing lower and higher ability of the AMOT-S175E and AMOT-S175A mutants to bind to DCTN1, respectively. The inputs of the same pulldown experiment are shown in Extended Data Fig. 8c. c, Pulldown of endogenous DCTN1 from Mech.ON HEK293T cells transiently transfected with the indicated HA-AMOT mutants and treated with dynarrestin (10 μM) for 6 h, showing enhanced interaction between DCTN1 and the AMOT S175A mutant. The inputs of the same pulldown experiment are shown in Extended Data Fig. 8d. d, Representative AMOT immunoblots of HEK293 cells transiently transfected with the indicated AMOT mutants. GAPDH serves as loading control. The same experiment was repeated twice with comparable results. e, Luciferase assay of control (wt) or LATS1/2 dKO HEK293 cells transfected with a synthetic reporter for YAP/TAZ/TEAD-dependent transcription (8xGTIIC-Lux). The middle and right panels show luciferase assays performed with these same cells seeded in Mech.ON (sparse), Mech.OFF (dense) conditions or treated with Cyto.D (0.5 μM, 15 h). Data are presented as mean + s.d. of n = 3 biologically independent samples. Left panel, P = 0.0291; middle panel, P = 0.0012 (Mech.ON versus Mech.OFF), P = 0.0018 (Mech.ON versus CytoD); right panel, P = 0.002 (Mech.ON versus Mech.OFF), P = 0.0008 (Mech.ON versus CytoD). f, Luciferase assay of HEK293 cells treated with the indicated siRNAs and transfected with a synthetic reporter for YAP/TAZ/TEAD-dependent transcription (8xGTIIC-Lux) and with the indicated AMOT mutants. Data are presented as mean + s.d. of n = 3 biologically independent samples. P = 0.9972 (AMOTS175E siCo. versus siLATS1/2#1), P = 0.9981 (AMOTS175E siCo. versus siLATS1/2#2), P < 0.0001 for all other comparisons shown. g, Luciferase assay of HEK293 cells treated with the indicated siRNAs and transfected with a synthetic reporter for YAP/TAZ/TEAD-dependent transcription (8xGTIIC-Lux). Data are presented as mean + s.d. of n = 3 biologically independent samples. P = 0.2399 (siTNKS+siCo. versus siTNKS+siLATS1/2#1), P = 0.3032 (siTNKS+siCo. versus siTNKS+siLATS1/2#2), P < 0.0001 for all other comparisons shown. h, Top: schematics of the cytoskeletal organization of Mech.ON (left) versus Mech.OFF (right) cells. Bottom: schematic representation of the mechano-signalling events controlling the stability of AMOT proteins. In Mech.ON cells, AMOT protein is controlled by binding to the dynein/dynactin complex, allowing fast dynein-mediated transport of AMOT through the MT aster towards the pericentrosomal proteasome where AMOT is locally degraded. In Mech.OFF cells, absence of a MT aster allows accrual of AMOT protein levels. Here, LATS kinase further contributes to AMOT stability by direct phosphorylation of AMOT, averting it from its degradation route. In Mech.OFF cells, increase of AMOT levels serves as a cytoplasmic sink for YAP/TAZ, impeding nuclear accumulation. Created with BioRender. i, Schematic representation of factors and cytoskeletal components tipping the balance towards AMOT degradation and YAP/TAZ activation in Mech.ON cells (shown on the left) and those that drive AMOT stabilization in Mech.OFF cells (shown on the right). Created with BioRender. P values were determined by unpaired two-tailed Student’s t-test with Welch’s correction (e, left panel), one-way ANOVA with Tukey’s multiple comparison test (e, middle and right panels) or two-way ANOVA (f,g). Source data

Fig. 6. AMOT plays a central role in biological responses driven by YAP/TAZ mechanical regulation.

a, Representative IF images of hMSCs seeded on spread (unconfined, Mech.ON) or small (1,000 μm², Mech.OFF) micropatterns and treated with the indicated siRNAs. Oil Red (O.red) staining is used to identify lipid vacuoles (red), F-actin is stained with phalloidin (green) and nuclei are counterstained with Hoechst (blue). Scale bars, 15 μm. Extended Data Fig. 9a provides quantifications. b, Top: stacked images of BrdU incorporation. The colour scale indicates the extent of cell proliferation in a given position of the monolayers. Bottom: spatial mapping of the proliferation rate (BrdU signal) along the x axis of each panel shown on top (Methods). Extended Data Fig. 9d presents images of controls showing comparable densities of cells in all conditions. c, Representative AMOT immunoblots of HEK293 cells transiently transfected with empty vector or with constitutively active oncogenes and seeded in Mech.OFF conditions. GAPDH serves as loading control. d, Representative images and quantifications of colonies formed by human mammary luminal differentiated cells undergoing oncogenic reprogramming when transduced with lentiviral vectors encoding for the indicated factors. HER-CA, constitutive active HER2 mutant (Methods); AMOT^L/PPXA, AMOT point mutant in the three L/PPXY YAP-binding motifs (Methods). Scale bar, 170 μm. Extended Data Fig. 9e provides quantifications. e, Representative images (left) and tumour volume quantifications (right, n = 14 mice for wt and n = 8 mice each for each AMOTtko condition) of orthotopic mammary tumours formed by control or AMOT tKO MII cells. Lines represent mean ± s.e.m. P = 0.0205 (wt versus AMOTtKO#1), P = 0.0252 (wt versus AMOTtKO#2). f, Tumour volume quantifications (n = 5 mice for each condition) of orthotopic mammary tumours formed by MDA-MB-231 cells transduced with lentiviral vectors encoding for the indicated doxycycline (Doxy)-dependent AMOT mutants. Lines represent mean ± s.e.m. P = 0.0105 (AMOT wt Doxy⁻ versus Doxy⁺), P = 0.7525 (AMOT wt Doxy⁻ versus AMOT L/PPXA Doxy⁻), P = 0.9875 (AMOT L/PPXA Doxy⁻ versus AMOT L/PPXA Doxy⁺). g, Representative immunohistochemical pictures of AMOT, YAP and TAZ proteins in chemo-naïve human TNBC samples or adjacent normal mammary tissue. Data are representative of n = 7 independent patient samples. Scale bar, 50 μm. Extended Data Fig. 10a presents lower magnifications of the same samples and Extended Data Fig. 10b quantifications of immunohistochemical staining in all tested samples. P values were determined by one-way ANOVA with Tukey’s multiple comparison test (e,f). Source data

Extended Data Fig. 1. Microtubules (MTs) centrosomal organization serves as determinant of YAP/TAZ mechanotransduction.

a) Representative 3D immunofluorescence reconstructions of MCF10A cells seeded the indicated Mech.ON versus Mech.OFF conditions. MTs were labelled by α-Tubulin staining (αTub, magenta) and nuclei were counterstained with Hoechst (cyan). Scalebar, 10 μm. See also Supplementary Video 1 for an all-around view of the same cells. b) Representative immunofluorescence images (3D sections) of MCF10A cells seeded on Spread (unconfined, Mech.ON) versus Small (100 μm² micropatterns, Mech.OFF) substrates. MTs were labelled by α-Tubulin staining (αTub, magenta), the centrosome was labelled by pericentrin (PCNT1, grey) and highlighted by white arrowhead. Nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. c, d) Representative immunofluorescence images of HEK293 cells (c) and U2OS cells (d) seeded in the indicated mechanical conditions. Images are magnifications of the insets shown in the upper right corner of each picture. MTs were labelled by α-Tubulin staining (αTub, magenta), γ-TURCs were labelled by γ-Tubulin staining (γTub, yellow), the centrosome was labelled by pericentrin (PCNT1, grey). γ-TURCs and centrosomes are highlighted by white arrowheads. Scale bar, 5 μm. The same experiment was repeated twice with comparable results. e) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous targets CTGF (left) and Cyr61 (right) in HEK293 cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. Left graph: P = 0.007 (siCo. vs siγTub#1), P = 0.0017 (siCo. vs siγTub#2); right graph: P = 0.0021 (siCo. vs siγTub#1), P = 0.0033 (siCo. vs siγTub#2). f) Left: representative immunofluorescence images of HEK293 cells treated with the indicated siRNAs. F-actin was labelled by Phalloidin staining (F-actin, orange), focal adhesions were labelled by Paxillin staining (Pax, grey) and nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. Right: quantifications (n = 30 cells for siCo., n = 71 cells for siγTub#1, n = 40 cells each for siγTub#2, pooled from two independent seedings) of focal adhesion length in cells treated as in left panels. P = 0.5302 (siCo. vs siγTub#1), P = 0.2378 (siCo. vs siγTub#2). P values were determined by one-way ANOVA with Welch’s correction (e, f). Source data

Extended Data Fig. 2. Microtubules acetylation is required for centrosomal radial MT sprouting and for YAP/TAZ activation in mechano-ON cells.

a) Representative immunofluorescence images of MTs (αTub) and acetylated Tubulin (ac. Tubulin) in HEK293 cells seeded in the indicated mechanical conditions. Nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. b) Representative immunofluorescence images of MTs (αTub) and acetylated Tubulin (ac. Tubulin) in HEK293 cells seeded in Mech.ON conditions and treated with the indicated siRNAs. Nuclei were counterstained with Hoechst (cyan). Scale bar, 5 μm. c) Representative immunoblots of acetylated Tubulin (ac. Tub) in HEK293 cells seeded in mechano-ON conditions and treated with the indicated siRNAs. αTubulin (αTub) serves as loading control. d) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous targets CTGF (left) and Cyr61 (right) in U2OS cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. Left graph: P = 0.0270 (siCo. vs siγTub), P = 0.0169 (siCo. vs siαTAT1); right graph: P = 0.009 (siCo. vs siγTub), P = 0.0222 (siCo. vs si siαTAT1). e) Left: representative immunofluorescence images of HEK293 cells treated with the indicated siRNAs. F-actin was labelled by Phalloidin staining (F-actin, orange), focal adhesions were labelled by Paxillin staining (Pax, grey) and nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. Right: quantifications (n = 30 cells for siCo., n = 77 cells for siαTAT1#1, n = 50 cells each for siαTAT1#2, pooled from two independent seedings) of focal adhesion length in cells treated as in left panels. P = 0.2372 (siCo. vs siαTAT1#1), P = 0.5225 (siCo. vs siαTAT1#2). f) Representative immunofluorescence of F-actin (orange, Phalloidin labelling) in HEK293 cells transduced with empty or NLP1-encoding lentiviruses and treated with Cytochalasin D (Cyto.D, 1 μM for 2 h) or DMSO as negative control, as in Fig. 1i, j. Nuclei were counterstained with Hoechst (cyan). Scale bar, 5 μm. P values were determined by one-way ANOVA with Welch’s correction (d, e). Source data

Extended Data Fig. 3. Loss of NE-NASFs tethering impairs YAP/TAZ mechano-activation.

a) Representative Z-planes of immunofluorescence stacks of MCF10A cells seeded in mechano-ON (40 kPa) versus -OFF (0,7 kPa) conditions, showing NASFs contacting the basal side of the nuclear envelope of mechano-activated, but not of mechano-inhibited cells. The nuclear lamina was labelled by LaminA staining (cyan) and F-actin was labelled by Phalloidin staining (orange). Basal to apical planes are shown in bottom to top panels. Scale bar, 10 μm. See also Fig. 2a for 3D reconstructions of cells seeded in the same conditions. b) Representative (n = 3 independent replicates) immunofluorescence YZ-sections of MCF10A cells seeded on stiff (Mech.ON) versus soft (Mech.OFF) hydrogels. Lamin was labelled by a conformation-sensitive LmnA antibody (LmnA C-C, cyan), which is unable to bind the basal side of the NE of mechano-ON cells due to LaminA stretching. The corresponding intensity profiles of LmnA conformation-sensitive staining along the Z-axis is shown below each immunofluorescence picture. Scale bar, 10 μm. c) Representative (n = 3 independent replicates) immunofluorescence YZ-sections of MCF10A cells treated with the indicated siRNAs. A conformation-sensitive LmnA antibody (LmnA C-C, cyan), which is unable to bind stretched LaminA was used to show loss of basal nuclear tension in SUN2-depleted cells. The corresponding intensity profiles of LmnA conformation-sensitive staining along the Z-axis is shown below each immunofluorescence picture. Scale bar, 10 μm. d) Left: Representative Z-planes of immunofluorescence stacks of MCF10A cells treated with indicated siRNAs, showing abolished NASFs formation in SUN2-depleted cells. The nuclear lamina was labelled by LaminA staining (cyan) and F-actin was labelled by Phalloidin staining (orange). Basal to apical planes are shown in bottom to top panels. Scale bar, 10 μm. See also Fig. 2b for 3D sections of cells treated as in (d). Right: quantifications of NASFs/total F-actin ratios in the same cells (n = 30 for each condition, pooled from two independent seedings). P < 0.0001. e) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous target Cyr61 in HEK293 cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. P = 0.0488 (siCo. vs siSUN2#1), P = 0.0221 (siCo. vs si siSUN2#2). f, g) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous targets CTGF and Cyr61 in MCF10A (f) and U2OS (g) cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. P < 0.0001. h) Loss of NE-MTs tethering impairs YAP/TAZ mechano-activation, independently of F-actin and focal adhesions architecture. Left: representative immunofluorescence images of HEK293 cells treated with the indicated siRNAs. F-actin was labelled by Phalloidin staining (F-actin, orange), focal adhesions were labelled by Paxillin staining (Pax, grey) and nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. Right: quantifications (n = 30 cells for siCo., n = 42 cells for siSUN2#1, n = 65 cells each for siSUN2#2, pooled from two independent seedings) of focal adhesion length in cells treated as in left panels. P = 0.2621 (siCo. vs siSUN2#1), P = 0.0802 (siCo. vs si siSUN2#2). P values were determined by unpaired two-tailed Student’s t-test with Welch’s correction (d) one-way ANOVA with Welch’s correction (e, h), or with two-way ANOVA (f, g). Source data

Extended Data Fig. 4. Loss of Nesprin or Emerin impairs YAP/TAZ mechano-activation.

a) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous target Cyr61 in HEK293 cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. P = 0.0007 (siCo. vs siNesp4#1), P = 0.0029 (siCo. vs siNesp4#2). b) Left: representative immunofluorescence images of HEK293 cells treated with the indicated siRNAs. F-actin was labelled by Phalloidin staining (F-actin, orange), focal adhesions were labelled by Paxillin staining (Pax, grey) and nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. Right: quantifications (n = 30 cells for siCo., n = 39 cells each for all other conditions, pooled from two independent seedings) of focal adhesion length in cells treated as in left panels. P = 0.0732 (siCo. vs siNesp4#1), P = 0.0972 (siCo. vs siNesp4#2). c) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous target Cyr61 in HEK293 cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. P = 0.0009 (siCo. vs siEMD#1), P = 0.0012 (siCo. vs siEMD#2). d) Loss of NE-MTs tethering impairs YAP/TAZ mechano-activation, independently of F-actin and focal adhesions architecture. Left: representative immunofluorescence images of HEK293 cells treated with the indicated siRNAs. F-actin was labelled by Phalloidin staining (F-actin, orange), focal adhesions were labelled by Paxillin staining (Pax, grey) and nuclei were counterstained with Hoechst (cyan). Scale bar, 10 μm. Right: quantifications (n = 32 cells for siCo., n = 51 cells for siEMD#1, n = 30 cells each for siEMD#2, pooled from two independent seedings) of focal adhesion length in cells treated as in left panels. P = 0.0824 (siCo. vs siEMD#1), P = 0.2492 (siCo. vs siEMD#2). e, f) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous targets CTGF and Cyr61 in MCF10A (e) and U2OS (f) cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. In (e): P = 0.0002 (Cyr61 siCo. vs siEMD), P < 0.0001 for all other comparisons shown; in (f): P = 0.0002 (CTGF siCo. vs siNesp4), P = 0.0003 (CTGF siCo. vs siEMD); P = 0.0038 (Cyr61 siCo. vs siNesp4); P < 0.0001 (Cyr61 siCo. vs siEMD). g) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous target CTGF (left) and Cyr61 (right) in HEK293 cells treated with the indicated siRNAs and seeded in mechano-OFF conditions. Data are mean + s.d. of n = 3 biologically independent samples. Left graph: P = 0.0055 (siCo. vs siADF/CFL); P = 0.0634 (siCo. vs siADF/CFL+ siαTAT1); P = 0.6626 (siCo. vs siADF/CFL+ siNesp4). Right raph: P = 0.0329 (siCo. vs siADF/CFL); P = 0.2094 (siCo. vs siADF/CFL+ siαTAT1); P = 0.2102 (siCo. vs siADF/CFL+ siNesp4). P values were determined by one-way ANOVA with Welch’s correction (a–d, g), or with two-way ANOVA (e, f). Source data

Extended Data Fig. 5. AMOT acts as cytoplasmic mechanical rheostat to control YAP/TAZ activity.

a) Quantifications (n = 30 cells for each condition, pooled from two independent seedings) of proteasome (20-S staining) in cells seeded as in Fig. 3a. Preferential accumulation of the proteasome in a single perinuclear area is set to 1 in mechano-ON conditions. Conversely, cells seeded in mechano-OFF displayed multiple proteasome signals scattered throughout the cytoplasm. P < 0.0001. b) Quantifications (n = 30 cells for each condition, pooled from two independent seedings) of proteasome (20-S staining) in cells seeded as in Fig. 3b. Preferential accumulation of the proteasome in a single perinuclear area is set to 1 in siCo. conditions. Conversely, cells depleted of SUN2 or Emerin displayed multiple proteasome signals scattered throughout the cytoplasm. P < 0.0001. c) Representative immunoblot of YAP/TAZ cytoplasmic interactors in HEK293 cells seeded in mechano-ON (stiff) versus mechano-OFF (0.7 kPa hydrogels) conditions. GAPDH serves as sample processing control. See also Supplementary Table 1 for a summary of these results. d) Representative AMOT immunoblot of HEK293 cells seeded in mechano-ON (sparse) versus mechano-OFF (dense) conditions. GAPDH serves as loading control. e) Quantitative real-time PCR (qRT–PCR) showing that the endogenous expression levels of AMOT-130 and AMOT-L1 (HEK293 cells) and AMOT-L2 (MCF10A cells) are unaffected by Mech.ON (stiff) versus Mech.OFF (0.7 kPa hydrogels) conditions. Data are mean + s.d. of n = 3 biologically independent samples. P = 0.4309 (left), P = 0.3940 (middle), P = 0.6024 (right). f) Representative AMOT immunoblot of HEK293 cells treated with increasing doses of the Rho inhibitor 1 (Exoenzyme C3 Transferase, 5 μg ml⁻¹ and 10 μg ml⁻¹, 4 h). GAPDH serves as loading control. g) Representative AMOT immunoblot of HEK293 cells treated with increasing doses of LatrunculinA (Lat.A, 343 nM and 686 nM, 2 h). GAPDH serves as loading control. h) Representative AMOT immunoblot of HEK293 cells treated with the indicated mechano-inhibitory drugs (Das., Dasatinib 100 nM; Def., Defactinib 5 μM; Fasudil 10 μM; Cyto.D, Cytochalasin D 1 μM for 2 h). GAPDH serves as loading control. i) Left: representative immunoblot showing quantitation of filamentous (F) versus free globular (G) actin in MCF10A cells seeded on stiff (40 kPa) versus soft (0.7 kPa) hydrogels. Right: quantitation of F/G-actin ratios of the samples shown on the left (see Methods). Results are representative of two independent experiments. j) Representative AMOT immunoblot of HEK293 cells treated with the indicated inhibitors of microtubule polymerization (Nocodazole 5 μM; Vincristine 5 μM, for 1 h). GAPDH serves as loading control. k) Luciferase assay of HEK293 cells seeded in mechano-ON conditions and transfected with a synthetic reporter for YAP/TAZ/TEAD-dependent transcription (8xGTIIC-Lux) and with the indicated AMOT mutants. Data are mean + s.d. of n = 3 biologically independent samples. P = 0.0210 (Empty vs AMOTwt), P = 0.0733 (Empty vs AMOT L/PPXA), P = 0.0137 (AMOTwt vs AMOT L/PPXA). l) Luciferase assay of HEK293 cells treated with siRNAs targeting YAP/TAZ and reconstituted with either YAPwt of YAP-WW mutant constructs (see Methods). Cells were concomitantly transiently transfected with AMOT-expressing or empty vectors, and with a synthetic reporter for YAP/TAZ/TEAD-dependent transcription (8xGTIIC-Lux), and after 24 h seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. P < 0.0001 (YAPwt AMOT- vs AMOT+), P = 0.8023 (YAPwt AMOT- vs YAPWWmut AMOT+), P = 0.0139 (YAPwt AMOT+ vs YAPWWmut AMOT+). m) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous targets CTGF (left) and Cyr61 (right) in control (wt) or AMOT tKO M2 (see Methods) cells seeded in Mech.ON (stiff) versus Mech.OFF (0.7 kPa hydrogels) conditions. Data are mean + s.d. of n = 3 biologically independent samples. Left graph: P < 0.0001 (wt Mech.ON vs Mech.OFF), P = 0.3100 (wt Mech.ON vs AMOTtKO#1 Mech.OFF), P = 0.9137 (wt Mech.ON vs AMOTtKO#2 Mech.OFF); right graph: P < 0.0008 (wt Mech.ON vs Mech.OFF), P = 0.0977 (wt Mech.ON vs AMOTtKO#1 Mech.OFF), P > 0.9999 (wt Mech.ON vs AMOTtKO#2 Mech.OFF). n) Left: representative immunofluorescence images of YAP/TAZ in hMSCs treated with the indicated siRNAs and seeded in Mech.OFF (1000 μm² micropatterns) versus Mech.ON (unconfined) conditions. Nuclei were counterstained with Hoechst (cyan). Scale bar, 20 μm. Cell borders are outlined by orange dashes. Right: quantifications (n = 54 cells for Mech.ON siCo., n = 51 cells for Mech.OFF siAMOT#2, n = 50 cells each for all other conditions, pooled from two independent seedings) of YAP/TAZ nuclear-to-cytoplasmic subcellular localization (N/C) in cells seeded as in left panels. P < 0.0001. o) Luciferase assay of HEK293 cells treated with the indicated siRNAs, seeded in mechano-ON versus mechano-OFF (Cyto.D treated) conditions and transfected with a synthetic reporter for YAP/TAZ/TEAD-dependent transcription (8xGTIIC-Lux). Data are mean + s.d. of n = 3 biologically independent samples. P = 0.0027 (siCo. Mech.ON vs Mech.OFF), P = 0.5249 (siCo. Mech.ON vs siAMOT#1 Mech.OFF), P = 0.9874 (siCo. Mech.ON vs siAMOT#2 Mech.OFF). P values were determined by unpaired two-tailed Student’s t-test with Welch’s correction (e), one-way ANOVA with Welch’s correction (a, b, k, l, n), or with two-way ANOVA (m, o). Source data

Extended Data Fig. 6. AMOT is constitutively tagged for proteasomal degradation under both mechano-ON and mechano-OFF conditions.

a) Representative AMOT immunoblot of HEK293 cells treated with siRNAs targeting the poly-ADP-ribosyltransferase TNKS1/2 or the E3 ubiquitin ligase RNF146 and seeded in mechano-ON conditions. GAPDH serves as loading control. The same experiment was repeated twice with comparable results. b) Representative AMOT immunoblot of HEK293 cells treated with the indicated independent Tankyrase1/2 inhibitors (10 μM, 24 h). GAPDH serves as loading control. The same experiment was repeated twice with comparable results. c) Representative RNF146 immunoblot of HEK293 cells seeded in Mech.ON (sparse) versus Mech.OFF (dense) conditions. GAPDH serves as loading control. The same experiment was repeated twice with comparable results. d) Representative immunofluorescence images of RNF146 in HEK293 cells seeded in Mech.ON (spread) versus Mech.OFF conditions (dense). Nuclei were counterstained with Hoechst (cyan). Scale bar, 25 μm. The same experiment was repeated twice with comparable results. e) Representative PARylation assay (see Methods) in HEK293 cells seeded in Mech.ON (sparse) versus Mech.OFF (dense) conditions and treated with proteasome inhibitors showing that AMOT PARylation levels are unaffected by the mechanical state of the cell. GAPDH serves as loading control for inputs. The same experiment was repeated twice with comparable results. f) Representative Ubiquitination assays (see Methods) in HEK293 cells seeded in Mech.ON (sparse) versus Mech.OFF (dense) conditions and treated the indicated siRNAs, showing that Tankyrase1/2-dependent AMOT poly-ubiquitination occurs irrespective of the the mechanical state of the cell. The same experiment was repeated twice with comparable results. g) Inputs for the co-immunoprecipitation experiment shown in Fig. 4c. h) Cycloheximide (CHX) pulse-and-chase experiment (see Methods) in HEK293 cells showing that in mechano-ON conditions, AMOT is rapidly degraded after proteasome inhibitor (Prot.i) washout. Cells were treated with CHX 50 μg/mL. GAPDH serves as loading control. The same experiment was repeated twice with comparable results. Source data

Extended Data Fig. 7. AMOT delivery to proteasomal condensates by dynein-mediated retrograde transport.

a) Graph depicting the most significant GO terms emerging from the Gene Ontology analyses of AMOT proximal proteins, as reported in a publicly available dataset. A full list of retrieved AMOT proximal proteins and associated significant GO terms (-Log₁₀[p-value]) ≥ 2.6), and associated p-values are provided in Supplementary Table 2. b) Representative AMOT immunoblot of HEK293 cells seeded in mechano-ON conditions and treated with the indicated siRNAs. GAPDH serves as loading control. c) Representative AMOT immunoblot of HEK293 cells seeded in mechano-ON conditions and treated with DMSO (Co.) or Dynarrestin (10 μM, 6 h). GAPDH serves as sample processing control. d) Representative AMOT immunoblot of HEK293 cells seeded in mechano-ON conditions and treated with DMSO (Co.) or Ciliobrevin D (Cilio.D 10 μM, 5 h). Cilio.D treatment was followed by 2 or 4 h of washout. GAPDH serves as loading control. e) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous target CTGF in HEK293 cells treated with the indicated siRNAs and seeded in mechano-ON conditions. Data are mean + s.d. of n = 3 biologically independent samples. P < 0.0001. f, g) Luciferase assay of HEK293 cells treated with the indicated siRNAs, seeded in mechano-ON conditions and transfected with a synthetic reporter for YAP/TAZ/TEAD-dependent transcription (8xGTIIC-Lux). Data are mean + s.d. of n = 3 biologically independent samples. P = 0.0031 for siCo. vs siDYNLL2#2), P < 0.0001 for all other comparisons shown. h) Pulldown of endogenous AMOT (left) or endogenous YAP (right) from proteasome-inhibited HEK293T cell seeded in Mech.ON versus Mech.OFF conditions, showing that AMOT bound to DCTN1 in Mech.ON cells is unable to bind to YAP. IgG pulldown serves as negative control. i) Cartoon representations of the X-ray structure (PDB code 7nma) of the 14-3-3 domain (grey) in complex with a short AMOT peptide phosphorylated at S175 (pS175). The AMOT peptide is shown as stick representation in green with pS175 and L178 highlighted. j) AlphaFold2 predicted alignment error (PAE) plot for the complex shown in Fig. 4g, indicating that the ICD structure is predicted with very high confidence. k) DCTN1-ICD as surface representation with highlighted in white the four residues (S659, A662, T835, A839) lining the putative binding cleft. These have been replaced by tryptophans for co-immunoprecipitation (co-IP) assays of Fig. 4h (see main text for details). l) Schematic illustrations of the domain structures and motifs of DCTN1/p150 (top) and AMOTp130 (bottom). Created in BioRender. m) Representative immunoblots of the inputs of the pulldown experiment shown in Fig. 4h. n) Representative immunoblots of the inputs of the pulldown experiment shown in Fig. 4i. P values were determined by one-way ANOVA with Welch’s correction (e–g), or with multiple testing correction using the default g:SCS method (a). Source data

Extended Data Fig. 8. Hippo/LATS signaling indirectly feeds on YAP/TAZ mechanoregulation through AMOT.

a) Representative AMOT immunoblot of HEK293 cells treated with the indicated siRNAs and seeded in mechano-OFF conditions. GAPDH serves as loading control. The same experiment was repeated twice with comparable results. b) Representative phospho-AMOT-S175 immunoblot of HEK293 cells transfected with the indicated siRNAs, seeded in mechano-ON conditions and treated with proteasome inhibitor (Lactacystin, 10 μM, 8 h). GAPDH serves as loading control. The same experiment was repeated twice with comparable results. c) Representative immunoblots of the inputs of the pulldown experiment shown in Fig. 5b. d) Representative immunoblots of the inputs of the pulldown experiment shown in Fig. 5c. e) Representative immunoblots of HEK293 cells seeded in Mech.ON (stiff) versus Mech.OFF (0.7 kPa hydrogels) conditions treated with proteasome inhibitor (Lactacystin, 10 μM, 8 h), showing that AMOT-S175 and LATS1-T1079 phosphorylation is unaffected by cell mechanics. GAPDH serves as loading control. The same experiment was repeated twice with comparable results. f) Quantitative real-time PCR (qRT–PCR) assessing the expression levels of the YAP/TAZ endogenous targets CTGF (left) and Cyr61 (right) in YAPwt or YAP5SA reconstituted MCF10A cells (see Methods) seeded in Mech.ON (sparse), Mech.OFF (dense) conditions or treated with Lat.A (343 nM, 15 h). Data are mean + s.d. of n = 3 biologically independent samples. P < 0.0001. P values were determined by two-way ANOVA (f). Source data

Extended Data Fig. 9. AMOT acts as mechanical rheostat to control biological responses driven by YAP/TAZ.

a) Quantifications of the percentage of adipogenic differentiation (based on Oil Red staining) of hMSCs seeded on spread (unconfined, Mech.ON) or small (1000 μm², Mech.OFF) micropatterns and treated with the indicated siRNAs, as in Fig. 6a. Data are representative of n = 5 biologically independent experiments, with 100 cells quantified each. Boxplot shows interquartile range and whiskers represent min to max. P = 0.1920 (siAMOT#1 Mech.ON vs Mech.OFF), P = 0.8271 (siAMOT#2 Mech.ON vs Mech.OFF), P < 0.0001 for all other comparisons shown. b) Left: Representative Oil Red staining of hMSCs transfected with the indicated siRNAs, seeded in mechano-ON conditions and treated with DMSO (Co.) or with Cytochalasin D (Cyto.D, see Methods). To verify Cyto.D effectiveness, F-actin was labelled with phalloidin (F-actin, green). Nuclei were counterstained with Hoechst (blue). Right: quantifications of the percentage of adipogenic differentiation (based on Oil Red staining) of hMSCs treated as in left panels. Data are representative of n = 6 biologically independent experiments, with 100 cells quantified each. Boxplot shows interquartile range and whiskers represent min to max. P = 0.0101 (Co. vs CytoD), P = 0.0125 (siCo. vs siAMOT#1 CytoD), P = 0.0176 (siCo. vs siAMOT#2 CytoD). c) Left: Representative Oil Red staining of hMSCs seeded in mechano-ON conditions, transduced with Empty or AMOT-encoding lentiviral vectors and subjected to adipogenic differentiation protocol (see Methods). Nuclei were counterstained with Hoechst (blue). Right: quantifications of the percentage of adipogenic differentiation (based on Oil Red staining) of hMSCs treated as in left panels. Data are representative of n = 6 biologically independent experiments, with 100 cells quantified each. Boxplot shows interquartile range and whiskers represent min to max. P = 0.0086. d) Representative colorimetric stacked images of Hoechst counterstained nuclei of MCF10A cells treated with the indicated siRNAs and seeded as cell monolayers of defined dimensions as in Fig. 6b. The colour scale indicates the density of Hoechst labelled nuclei, showing a uniform distribution of cells within the monolayers in all the experimental conditions. e) Quantifications of the number of colonies formed by human mammary luminal differentiated cells undergoing oncogenic reprograming when transduced with lentiviral vectors encoding for the indicated factors, as in Fig. 6d (n = 3 biologically independent experiments): HER-CA, constitutive active HER2 mutant (see Methods); AMOT^L/PPXA, AMOT point mutant in the three L/PPXY YAP-binding motifs (see Methods). Lines represent mean ± s.e.m. P < 0.0001. f) Left: Quantifications of the number of mammospheres (n = 6 biological independent samples) formed by control (wt) or AMOT tKO MII cells treated with the indicated siRNAs. Right: Quantifications of the number of soft agar colonies (n = 3 biological independent samples) formed by control (wt) or AMOT tKO MII cells treated with the indicated siRNAs. Lines represent mean ± s.e.m. Left graph: P > 0.9999 (wt siCo. siY/T#1), P = 0.9408 (wt siCo. siY/T#2), P < 0.0001 for all other comparisons shown. Right graph: P > 0.9999 (wt siCo. vs siY/T#1), P = 0.9903 (wt siCo. vs siY/T#2), P = 0.0002 (wt siCo. vs AMOTtKO#1 siCo.), P = 0.0029 (AMOTtKO#1 siCo. vs AMOTtKO#1 siY/T#1), P < 0.0001 for all other comparisons shown. g) Left: Quantifications of the number of mammospheres (n = 6 biological independent samples) formed by control (wt) or AMOT tKO MII cells treated with the TEAD inhibitor VT107 (see Methods), or with DMSO (Co.). Right: Quantifications of the number of soft agar colonies (n = 3 biological independent samples) formed by control (wt) or AMOT tKO MII cells treated with the with the TEAD inhibitor VT107 (see Methods), or with DMSO (Co.). Lines represent mean ± s.e.m. Left graph: P = 0.8249 (wt Co. vs TEADi), P < 0.0001 for all other comparisons shown. Right graph: P = 0.1024 (wt Co. vs TEADi), P < 0.0001 for all other comparisons shown. h) Quantifications of the number of soft agar colonies (n = 3 biological independent samples) formed by MDA-MB-231 cells transduced with lentiviral vectors encoding for the indicated doxycycline (Doxy)-inducible AMOT mutant constructs. Lines represent mean ± s.e.m. P = 0.0047 (AMOTwt -Doxy vs AMOTwt +Doxy), P = 0.0138 (AMOTwt +Doxy vs AMOTL/PPXA -Doxy), P = 0.9256 (AMOTL/PPXA -Doxy vs AMOTL/PPXA +Doxy). P values were determined by two-way ANOVA (a, f–h), or by one-way ANOVA with Welch’s correction (b, e), or by unpaired two-tailed Student’s t-test with Welch’s correction (c). Source data

Extended Data Fig. 10. Role of AMOT as candidate tumour suppressor.

a) Representative large field immunohistochemical pictures of AMOT, YAP and TAZ proteins in chemo-naïve human triple negative breast cancer (TNBC) samples or adjacent normal mammary tissue. The enlargements (top and bottom panels) are the same pictures shown in Fig. 6g. Data are representative of n = 7 independent patient samples. Scale bars, 50 μm. b) Digital-pathology based quantifications (Z-scores, see Methods) of AMOT abundance and YAP/TAZ nuclear-to-cytoplasmic subcellular localization based on immunohistochemical signals in the indicated cell populations (Tumour, Tum.; adjacent normal luminal, Lum.; and adjacent normal basal mammary cells, Basal) in independent TNBC samples (#1-#7), as shown in representative pictures of Fig. 6g and Extended Data Fig. 10a. n = 702261 tumour and n = 138509 normal cells were quantified for AMOT, n = 197876 tumour and n = 3264 normal cells were quantified for YAP, n = 144099 tumour and n = 5205 normal cells were quantified for TAZ. P = 0.0325 (TAZ IHC Tum. vs Lum. #3), P = 0.0043 (YAP IHC Basal vs Lum. #3), P < 0.0001 for all other comparisons shown. P values were determined by unpaired Student’s t-test with Welch’s correction (b). Source data