Rabbit anti-SUN1 polyclonal antibody (pAb) (HPA008346) was purchased from Sigma Aldrich (St. Louis, MO, United States) and used at 1:100 to 1:200 dilution for immunofluorescence microscopy. Rabbit anti-SUN2 pAb (06-1038) was obtained from Merck Millipore (Temecula, CA, United States) and used at a 1:200 dilution for immunofluorescence microscopy. Mouse anti-β-actin monoclonal antibody (mAb) clone AC-15 (A1978) and clone 6D1 were obtained from Sigma-Aldrich and Fujifilm Wako Pure Chemical Corporation (Osaka, Japan), respectively. Mouse anti-vinculin mAb (clones VIN11-5 and V4505) was purchased from Sigma Aldrich and used at a 1:300 dilution for immunofluorescence microscopy. Mouse anti-β-tubulin mAb (clones TUB2.1 and T4026) was purchased from Sigma-Aldrich. Mouse anti-integrin β1 mAb (TS2/16) was procured from eBIoscience, Inc. (San Diego, CA, United States) and Santa Cruz (Dallas, TX, United States), and used at 1:50 to 1:100 dilution for immunofluorescence microscopy. Mouse anti-active integrin β1 mAb (HUTS-4, MAB 2079Z) was purchased from Merck Millipore and used at a 1:50 dilution for immunofluorescence microscopy. Mouse anti-active integrin β1 mAb (12G10) was purchased from BioRad (Hercules, CA, United States) and used at 1:100 dilution for immunofluorescence microscopy. Rabbit anti-FAK pAb (#3285) and rabbit anti-Tyr397 phosphorylated focal adhesion kinase (FAK-pY397) mAb (D20B1) were purchased from Cell Signaling (Danvers, MA, United States) and used at 1:500 dilution for western blotting. Rabbit anti-FAK-pY397 pAb (sc-11765-R) was from Santa Cruz Biotechnology and used at 1:50 dilution for immunofluorescence microscopy. Mouse anti-paxillin mAb (D-9 sc365174 and B-2 sc365379) was obtained from Santa Cruz Biotechnology and used at 1:100 dilution for immunofluorescence microscopy. Mouse anti-Tyr118 phosphorylated paxillin mAb (A-5 sc365020) was obtained from Santa Cruz Biotechnology and used at 1:100 dilution for immunofluorescence microscopy. Mouse anti-zyxin mAb (sc-136128) was obtained from Santa Cruz Biotechnology and used at a 1:200 dilution for immunofluorescence microscopy. Anti-YAP (D8H1X) XP rabbit mAb was purchased from Cell Signaling and used at a 1:100 dilution for immunofluorescent microscopy. Rat anti-histone H3 mAb was a gift from Dr. H. Kimura (Tokyo Institute of Technology). Fluorescent secondary antibodies were obtained from Jackson ImmunoResearch Laboratories (Soham, United Kingdom). Rhodamine phalloidin was purchased from Cytoskeleton Inc. (Denver, CO, United States).

HeLa cells were obtained from the Japanese Cancer Research Bioresources (JCRB) Cell Bank and grown in Dulbecco’s modified Eagle’s medium (low glucose, Fujifilm Wako Pure Chemical Corporation) supplemented with 10% fetal calf serum at 37°C in a 10% CO₂ atmosphere. HeLa cells were used in this study unless otherwise stated. The human mammary epithelial cell line MCF10A (CRL-10317) was obtained from the American Type Culture Collection (ATCC) and cultured as previously described (Yokoyama et al., 2014). SUN1-knocked out HeLa cells have been described previously (Nishioka et al., 2016) and cultured as previously described. Transfection was performed using Gene Juice (Merck Millipore) and described previously (Satomi et al., 2020).

The sequences of siRNA pools against SUN1 (UNC84A) and SUN2 (UNC84B) have been described previously (Nishioka et al., 2016). The siRNAs were obtained from Nippon Gene (Tokyo, Japan). Cells were transfected with the indicated siRNAs or a non-targeting siRNA pool (siNC, Thermo Fisher Scientific, Waltham, MA) as a negative control using Lipofectamine RNAiMAX reagent (Invitrogen, CA, United States) as previously described (Hieda et al., 2021). Briefly, all siRNAs were used at a final concentration of 10 nM, and cells were fixed or harvested 48 h after transfection unless otherwise stated.

The cells were fixed with 4% paraformaldehyde and immunostaining was mostly performed as described previously using appropriate primary and secondary antibodies (Hieda et al., 2008) unless stated otherwise. For HUTS4 mAb staining, cells were fixed and stained without Triton X-100 permeabilization. Actin filaments were stained with 50 nM rhodamine–phalloidin for 30 min. For YAP staining, cells were cultured on type I collagen-coated cover glass (#4910-010, Iwaki, Japan). For Triton X-100 permeabilization before fixation, cells were washed twice with ice-cold transport buffer (TB, 20 mM HEPES, pH 7.3; 110 mM potassium acetate; 2 mM magnesium acetate; 5 mM sodium acetate; 0.5 mM EGTA; Adam et al., 1990) and subsequently incubated with TB containing 0.5% Triton X-100 for 5 min on ice, followed by fixation. Cells were viewed and captured with Olympus IX81 with Plan Apo 60×/NA1.4 or Olympus BX53 with a UPlanS Apo 40×/NA 0.95 objective lens using an Olympus DP-73 camera, Olympus U-HGLGPS light source, and Olympus U-FBNA filter (excitation 470–495, emission 510–550) and Olympus U-FGW (excitation 530–550, emission).

Quantification of the number, area, and fluorescent intensity was performed using ImageJ software (https://imagej.nih.gov/ij/). The number of FAs was quantified after thresholding and segmentation. The integrated density (i.e., the sum of all pixels in the ROI, region of interest) was measured as “RawIntDen”.

The total cell extract was collected using 2× sample buffer and sonicated, or syringe sampled 15 times using a 1 ml syringe fitted with a 24 G needle to ensure lysate homogenization and genomic DNA shearing. Afterward, the protein concentration was analyzed using Ionic Detergent Compatibility Reagent (Thermo Fisher Scientific) and Pierce 660 nm Protein Assay Reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. The total cell lysate was analyzed by western blotting using the indicated antibodies.

Cell adhesion assays were performed as described previously (Hieda et al., 2015). Briefly, 96-well plates were coated with 100 μL of 5 μg/ml laminin (AGC Inc., Tokyo, Japan), 5 μg/ml vitronectin (Wako Pure Chemical, Osaka, Japan), 5 μg/ml fibronectin (AGC Inc.), 5 μg/ml collagen type IC (Nitta Gelatin, Osaka, Japan), or 3% bovine serum albumin (BSA) and blocked with 3% BSA. Next, cells were added to the plates. After 2 h of incubation at 37°C, plates were washed with phosphate-buffered saline (PBS) and cells were stained with crystal violet. The absorbance was measured using measurement filter 595 nm and reference filter 630 nm. Experiments were repeated at least four times.

The internalization and recycling assays of integrin β1 were performed as described previously (Maekawa et al., 2017). Briefly, integrin β1 on the cell surface was labeled with Alexa 488-conjugated TS2/16 antibody in the growth medium containing 30 mM HEPES (pH 7.6) on ice for 1 h. Next, the cells were washed with ice-cold PBS and the medium was replaced with a fresh growth medium containing 30 mM HEPES (pH 7.6). Cells were incubated at 37°C for the indicated time (mentioned in the figure caption) to allow the internalization of fluorescent integrin β1. After internalization, the remaining fluorescence on the cell surface was quenched with an anti-Alexa 488 antibody. For the internalization assay, cells were fixed, and the signals inside the cells were imaged. To monitor the recycling of integrin β1, cells were re-incubated at 37°C for the indicated time points. After re-incubation, the surface fluorescence signal of integrin β1 was quenched again. Cells were subsequently fixed, and the signals inside the cells were imaged. For both internalization and recycling assays, images were quantified using the ImageJ software.

Traction force was visualized using wrinkle formation assay as described previously (Ichikawa et al., 2017; Kang et al., 2020). Briefly, silicone substrates CY 52-276 (Dow Corning Toray, Tokyo, Japan) were mixed at a weight ratio of 1.2:1 and spread on a coverslip to have a final elastic modulus of 5.4 kPa. To measure the elastic modules in a separate experiment with the Hertz contact model (Ichikawa et al., 2017), stainless beads were placed onto the substrate where the surface was coated in advance with fluorescent microbeads through silane coupling. For this process, the silicone surface was treated with 2% (3-aminopropyl) trimethoxysilane (Sigma Aldrich) in 90% EtOH for 30 min and then with 20% glutaraldehyde (Wako) for 5 min. By taking 3-dimensional images of the fluorescently labeled substrate using a confocal laser scanning microscope (FV-1000; Olympus), the indentation depth associated with the elastic modulus was measured. The coverslip was exposed to 4 mA oxygen plasma for hydrophilization for 1 min using a plasma generator (SEDE-GE, Meiwafosis, Tokyo, Japan), put on a 6-well culture plate, and coated with 10 µg/ml fibronectin (Sigma-Aldrich), on which HeLa cells (parental wild-type or SUN1-depleted) were cultured at 37°C in a 5% CO₂ stage incubator mounted on an inverted microscope (IX71; Olympus, Tokyo, Japan). 48 h after incubation, cells with wrinkle formation were imaged with phase-contrast microscopy using a ×10 semi-apochromat objective lens (NA 0.3). The acquired images were analyzed to automatically detect cellular traction force-generated wrinkles using a custom-made program written in Fiji software. Briefly, images were processed with a two-dimensional fast Fourier transformation and then with a band-pass filter to extract the wrinkles, which were skeletonized into line segments and integrated to finally obtain their total length (the number of pixels) per cell as traction index.

SUN1 is required for directional cell migration (Nishioka et al., 2016; Imaizumi et al., 2018); however, the underlying mechanism remains unknown. Both SUN1 and SUN2 proteins promiscuously interact with nesprin-2 and form the LINC complex (Stewart-Hutchinson et al., 2008; Ostlund et al., 2009; Haque et al., 2010; Sosa et al., 2012). Because nesprin-2G physically connects with the actin cytoskeleton (Zhen et al., 2002; Luxton et al., 2010), and its depletion reduces the ability to generate traction force (Woychek and Jones, 2019), we examined the involvement of SUN1 and SUN2 in the organization of the actin cytoskeleton using siRNAs. Depletion of SUN1 and SUN2 proteins was confirmed by immunofluorescence microscopy and western blotting (Figures 1A,B). Quantitative analysis showed that more than 90% of SUN1 and SUN2 expressions were depleted in each siRNA-targeted knockdown cell. Interestingly, cytoplasmic actin staining in the SUN1- but not SUN2-depleted cells decreased compared with the control cells (Figures 1A,C and Supplementary Figure S1A). Reduced staining intensity in the SUN1-depleted cells was rescued by the expression of mouse SUN1, which is siSUN1 resistant (Supplementary Figure S1B). In addition, some of the SUN1-depleted cells potentially have increased actin ruffling at their periphery (Figure 1A and Supplementary Figure S1A arrows), while obvious stress fibers and sub-nuclear actin structures were observed in the SUN2-depleted cells but not in the control cells (Figure 1A and Supplementary Figure S1C). However, the relative expression of β-actin protein remained the same in the control cells and the SUN1- or SUN2-depleted cells (Figure 1B and Supplementary Figure S1D). This is consistent with previous reports (Thakar et al., 2017). Thakar et al. reported that the mRNA levels of β-actin remained unaffected by the depletion of SUN1 in HeLa cells (Thakar et al., 2017). These results suggest that SUN1 depletion affects the organization of filamentous actin. Thus, we focused on the function of SUN1 in subsequent experiments.

Directional cell migration requires continuous turnover of FAs along the direction of cell movement (Lock and Strömblad, 2008). Thus, the effects of SUN1 depletion on FAs were examined by observing the localization of vinculin, a cytoskeletal adaptor protein in FAs (Carisey and Ballestrem 2011). Depletion of SUN1 was confirmed by immunofluorescence microscopy (Figure 1D). Accumulated vinculin signal was observed in the periphery of both control and SUN1-depleted cells (Figure 1D) indicating that vinculin is recruited to the plasma membrane in both control and SUN1-depleted cells. Of note, we did not observe an elevated level of plasma membrane-localized vinculin in the SUN1-depleted cells, while Thakar et al. (2017) showed that SUN1-depleted HeLa cells have increased vinculin staining at the plasma membrane as well as an increased level of GTP-bound RhoA. This disparity could be caused by activation by fibronectin on the coverslips they used. We did not observe an elevated level of GTP-RhoA in the SUN1-depleted cells (Supplementary Figure S1E).

Because the SUN1-depleted cells show many long cellular processes and the accumulation of actin at the tips of these processes that are not present in the control cells (Figure 1A, Supplementary Figure S1A, and Supplementary Figure S1C), we analyzed the effects of SUN1 depletion on FA turn over using a microtubule-induced FA disassembly assay (Ezratty et al., 2009). The result showed that SUN1-depleted cells retain the ability to disassemble their FAs (Supplementary Figure S2A). Vinculin has two distinct conformations, namely “open form” and “closed form” (Bays and DeMali, 2017). A significant amount of vinculin is a closed form at steady-state and can be washed out by Triton X-100 treatment, whereas a certain amount of vinculin is Triton X-100 insoluble because it is incorporated into FAs and binds to actin filaments (Lee and Otto, 1997; Sawada and Sheetz, 2002; Yamashita et al., 2014). To explore the incorporation of vinculin into FAs, cells were treated with Triton X-100 before fixation and stained with an anti-vinculin antibody. Triton X-100 treatment caused the dispersion of vinculin signals into small punctate patterns in SUN1-depleted cells, whereas condensed vinculin signals in siNC-transfected cells were resistant to Triton X-100 treatment (Figure 1E). The dispersion of the vinculin signal in the SUN1-depleted cells was rescued by the expression of siSUN1-resistant mouse SUN1 (Supplementary Figure S2B). Quantified data indicated that the integrated density of Triton X-100-resistant vinculin was decreased in SUN1-depleted cells compared with that in the control cells (Figure 1F), whereas the number of vinculin-positive dots was increased in SUN1-depleted cells (Figure 1G). The protein expression of vinculin in the SUN1-depleted cells remained unaltered (Figure 1H and Supplementary Figure S2C). Therefore, these results indicate that SUN1 depletion does not interfere with the recruitment of vinculin to the plasma membrane; however, it suppresses the conformational change in vinculin to a Triton X-100-resistant form (i.e., incorporation into FAs), which requires cytoskeletal forces (Omachi et al., 2017). Of note, SUN1 was resistant to Triton X-100 treatment (Figure 1E, upper panel), probably because SUN1 associates with nesprins, lamins, and/or chromatin.

To explore the maturation of FAs in SUN1-depleted cells, we next focused on integrin β1, which is a ubiquitously expressed FA molecule that drives the establishment of nascent adhesion sites (Schiller et al., 2013). To examine the availability of integrin β1 at the plasma membrane, we stained integrin β1 in SUN1-depleted cells using an anti-integrin β1 mAb, TS2/16 (Sanchez-Madrid et al., 1982), which recognizes both active and inactive forms of integrin β1. The intensity of total integrin β1 (i.e., the sum of active and inactive forms) at the plasma membrane in SUN1-depleted cells was obviously increased (Figures 2A,B). In addition, increased intensity of cell surface integrin β1 was observed following SUN1 depletion in both the non-cancerous breast epithelial cell line MCF10A (Supplementary Figure S3A) and unfixed cells (Supplementary Figure S4A; 0 min). Western blotting showed more than a 1.5 times increase in integrin β1 protein expression in the SUN1-depleted cells (Figure 2C).

The activation of integrins promotes the recruitment of several adaptor proteins to form submicrometer clusters and trigger FA maturation (Bouvard et al., 2013). Because the balance between active and inactive integrins is dynamically regulated (Khan and Goult, 2019), we analyzed the active form of integrin β1 using an anti-integrin β1 mAb, HUTS4 (Luque et al., 1996), which recognizes only the active form of integrin β1. Intriguingly, the intensity of cell surface-active integrin β1 was decreased in SUN1-depleted cells (Figure 2D). HUT4-positive large structures recognized in siNC-transfected cells disappeared in siSUN1-transfected cells, and weak filamentous structures were observed (Figure 2D). Quantification of staining intensity showed a significant reduction in active integrin β1 intensity in SUN1-depleted cells (Figure 2E). In addition, a ligand-binding form of integrin β1 in the control cells was colocalized with vinculin and these signals were Triton X-100 resistant (Supplementary Figure S3B). The decreased active integrin β1 in the SUN1-depleted cells was recovered by the expression of siSUN1-resistant mouse SUN1 (Supplementary Figure S3C).

Cell adhesion to the ECM activates integrins (Shattil et al., 2010). Because SUN1-depleted cells showed diminished cell surface-active integrin β1 (Figures 2D,E), the depletion of SUN1 could attenuate the adhesion activity of the cells. However, in contrast to our expectations, SUN1-depleted cells showed slightly but reproducibly increased adhesion activity (Figure 2F). This is in agreement with the upregulation of integrin β1 expression at the plasma membrane (Figures 2A–C) and indicates that impaired adhesion is not responsible for the reduction of activated integrin β1 at the cell surface.

We next studied the effect of SUN1-depletion on the intracellular trafficking of integrin β1. Integrins continuously cycle between the plasma membrane and internal compartments with low lysosomal degradation rates (Lobert et al., 2010; Nader et al., 2016). The amount of active integrin β1 at the plasma membrane is regulated by the rate of endocytosis from the plasma membrane and recycling from endosomes to the plasma membrane (De Franceschi et al., 2015). Thus, facilitated internalization of integrin β1 from the plasma membrane or its aberrant recycling to the cell surface could decrease the cell surface-active integrin β1. However, depletion of SUN1 affected neither the internalization efficiency of integrin β1 (Figure 2G) nor the recycling efficiency to the cell surface (Figure 2H). Therefore, these data indicate that SUN1 is not involved in the trafficking of integrin β1. Based on these findings, we assume that the reduction of active integrin β1 in SUN1-depleted cells could be related to the impaired actin cytoskeleton (Figure 1A) because physical forces exerted by actin fibers are transmitted to the cytoplasmic domain of integrins, thus activating it.

Depletion of SUN1 suppressed the incorporation of vinculin into FAs and the activation of integrin β1. The maturation of FA involves a stereotypical sequence of protein recruitment (Kuo et al., 2011). The initial stages of adhesion assembly occur in the actin-rich region at the cell periphery. Then, clustering of integrins occurs to form nascent adhesion, which is myosin-II independent. Also, additional FA proteins such as vinculin, phosphorylated paxillin, and FAK are recruited in a process termed maturation (Zaidel-Bar et al., 2003; Oakes and Gardel, 2014). To examine which step in FA maturation is disrupted in SUN1-depleted cells, we visualized FAs using antibodies against several FA-resident proteins, such as Tyr397 phosphorylated FAK (FAK-pY397), paxillin, Tyr118 phosphorylated paxillin (paxillin-pY118), and zyxin. FAK is a key tyrosine kinase involved in integrin signaling (Schlaepfer et al., 1999). The binding of integrin to the ECM and its clustering trigger the autophosphorylation of FAK at Tyr397 (Schlaepfer et al., 1999), which is critical for the maturation of FA; however, it is independent of mechanical tension (Horton et al., 2016). Depletion of SUN1 did not influence the FAK-pY397 staining intensity although the staining pattern was slightly altered (Figure 3A). Moreover, western blotting showed no significant differences in the expression of FAK protein and FAK Tyr397 phosphorylation between SUN1-depleted and control cells (Figure 3B). In addition, SUN1 depletion did not affect the protein expression or staining intensity of paxillin (Figures 3B,C), which is a scaffold protein in FAs and recruited to newly formed adhesions in a tension-independent manner. In contrast, the level of paxillin-pY118, which depends on intracellular traction force (Pasapera et al., 2010), moderately decreased in the SUN1-depleted cells (Figures 3B–D). Moreover, zyxin staining largely disappeared in SUN1-depleted cells, whereas zyxin staining patterns in control cells showed a relatively large dotted distribution (Figures 3E,F). The decreased zyxin signal in the SUN1-depleted cells was rescued by the expression of siSUN1-resistant mouse SUN1 (Supplementary Figure S3C). Because intracellular traction forces drive FA growth and recruitment of FA proteins such as zyxin (Zaidel-Bar et al., 2003), these results suggest a reduction in intracellular forces in SUN1-depleted cells.

Depletion of SUN1 did not prevent the recruitment of inactive integrin β1 and vinculin at the plasma membrane or inhibit the autophosphorylation of FAK at Tyr397. In contrast, the loss of SUN1 reduced the number of FAs that contain vinculin and active integrin β1. These results suggest that the depletion of SUN1 suppresses the force-dependent step of FA maturation. Thus, to investigate the intracellular forces, we first visualized the nuclear localization of a transcriptional co-regulator, Yes-associated protein (YAP). Because YAP enters the nucleus in an intracellular force-dependent manner in several cells including HeLa (Dupont et al., 2011; Finch-Edmondson and Sudol, 2016), it can be used as an indicator of intracellular forces. As reported previously (Panciera et al., 2017), YAP proteins were mostly present in the nucleus in control cells cultured on a glass coverslip (Figure 4A, left panel) because of the rigidity of glass as a substrate. In contrast, the depletion of SUN1 decreased the YAP signal in the nucleus and increased it in the cytoplasm (Figure 4A, right panel). The average staining intensity of YAP in the SUN1-depleted nucleus was decreased as compared with that in siNC-transfected cells (Figure 4B), supporting a reduction in cytoskeletal forces.

Next, we directly assessed the effects of depletion of SUN1 on the generation of traction force, which can be assayed by assessing the ability of cells to induce the formation of wrinkles on deformable silicone substrates (Ichikawa et al., 2017; Kang et al., 2020), because the contractile forces within a cell correlate with the length of wrinkles (Burton and Taylor, 1997; Fukuda et al., 2017). The control cells exhibited extensive wrinkles on the substrate, whereas the loss of SUN1 suppressed the formation of wrinkles, indicating the significantly decreased contractile forces in SUN1-depleted cells (Figures 4C,D). Because phase contrast images show differences in cell spreading between the SUN1-depleted and control cells (Figure 4C), the areas of cell spreading were quantified. The SUN1-depleted cells on the silicone substrates showed less spreading than the control cells; the SUN1-depleted cells on the glass coverslips also had less spreading, but the effect was moderate (Supplementary Figure S5), suggesting a possible potential issue with mechanosensing.