Integrin α6β4 requires plectin and vimentin for adhesion complex distribution and invasive growth

ABSTRACT

Integrin α6β4 binds plectin to associate with vimentin; however, the biological function remains unclear. Here, we utilized various integrin β4 mutants and CRISPR-Cas9 editing to investigate this association. Upon laminin binding, integrin α6β4 distinctly distributed peripherally as well as centrally, proximal to the nucleus. Upon fibronectin addition, integrin α6β4 was centrally recruited to large focal adhesions (FAs) and enhanced Fak (also known as PTK2) phosphorylation. Integrin β4 plectin-binding mutants or genetic deletion of plectin inhibited β4 recruitment to FAs and integrin α6β4-enhanced cell spreading, migration and three-dimensional invasive growth. Loss of the β4 signaling domain (but retaining plectin binding) blocked migration and invasiveness but not cell spreading, recruitment to FAs or colony growth. Immunostaining revealed that integrin α6β4 redistributed vimentin perinuclearly, where it colocalized with plectin and FAs. Depletion of vimentin completely blocked integrin β4-enhanced invasive growth, Fak phosphorylation and proliferation in three dimensions but not two dimensions. In summary, we demonstrate the essential roles of plectin and vimentin in promoting an invasive phenotype downstream of integrin α6β4.

This article has an associated First Person interview with the first author of the paper.

Summary: Integrin α6β4 is recruited with vimentin to perinuclear focal and fibrillar adhesions indicating how integrin α6β4 cooperates with other integrins to drive migration and invasive growth.

INTRODUCTION

Tumor cells have evolved complex interactions with their microenvironment, and these interactions govern malignant progression that is driven, in a large part, by integrins (Balkwill et al., 2012; Cooper and Giancotti, 2019). Integrins are a major class of αβ-heterodimeric extracellular matrix (ECM) receptors that recognize various ECM substrates, including laminins, fibronectin and collagens (Humphries et al., 2006). Integrins cluster and interact with cytoplasmic effectors that, in turn, recruit other signaling components to form large multiprotein integrin-associated complexes that can vary in their size and composition (Horton et al., 2015; Tsuruta et al., 2011). Integrin α6β4 is a specialized integrin for which the β4 subunit pairs only with the α6 subunit. Unlike other integrins, the β4 subunit has an unusually large cytoplasmic tail that consists of two sets of fibronectin type III repeats (FNIII) separated by a connecting segment and a small C-terminal tail. The β4 subunit contains two plectin-binding domains, one in the second FNIII domain and one in the C-terminal tail, that link the integrin to the cytokeratin (Walko et al., 2015), vimentin (Colburn and Jones, 2018) and actin cytoskeletons (Rabinovitz and Mercurio, 1997). Integrin α6β4 binding to plectin and cytokeratins is essential for the formation of hemidesmosomes, which are large adhesive plaques that connect epithelial monolayers to the basement membrane to provide essential epithelial integrity. Hemidesmosomes are highly organized structures formed at the basal layer of stratified epithelium, such as in the skin epidermis, where they link the epidermal and dermal layers, providing mechanical strength and durability (Walko et al., 2015). Although substantial effort has been put toward understanding hemidesmosome structure and function in keratinocytes owing to their involvement in the genetic disease epidermolysis bullosa, much less is known about how hemidesmosomes function in other simpler epithelia and carcinomas, including how integrin α6β4 binds to the various cytoskeletons in cells that have undergone epithelial-to-mesenchymal transition (EMT).

Both β1 and β4 integrins are involved in breast cancer development, prognosis and resistance to therapy (Nisticò et al., 2014; Stewart and O'Connor, 2015). β1 integrins, as well as αv integrins, which bind multiple β-subunits, form distinct adhesion complexes to that formed by integrin β4. These adhesions include F-actin-linked focal adhesions (FAs), which can be smaller and form along the leading edge of cells [also referred to as focal complexes (FCs); 0.25–2 μm], and larger FAs and fibrillar adhesions (>2 μm) that provide greater adhesion and traction (Geiger et al., 2001; Horzum et al., 2014). Fibrillar adhesions can mature from FAs when cells primarily adhere to fibronectin, and are centrally located in cells (Geiger et al., 2001). Hemidesmosomes crosstalk with F-actin-associated FAs (Hatzfeld and Magin, 2019; Tsuruta et al., 2011). Notably, plectin is a common component of these complexes (Hopkinson et al., 2014). Plectin is a giant protein found in nearly all mammalian cells that scaffolds adhesive molecules to various cytoskeletons, and provides crosslinking between intermediate filaments, microfilaments and microtubules (Wiche et al., 2015). The involvement and impact of plectin linking intermediate filaments to FAs have been shown to impact adhesive strength (Bhattacharya et al., 2009) and cell polarization (Burgstaller et al., 2010).

Integrin α6β4 associates with vimentin filaments through plectin binding (Homan et al., 1998, 2002). Vimentin, a type III intermediate filament, serves as a classical mesenchymal phenotype biomarker (Lamouille et al., 2014; Mendez et al., 2010). Like integrin α6β4, vimentin overexpression positively associates with enhanced cell motility, induction of EMT (Liu et al., 2015) and metastasis (Zelenko et al., 2017). In breast cancer, high vimentin expression correlates with poorly differentiated invasive ductal breast carcinoma, associates with a worse prognosis clinically and is characteristic of triple-negative breast cancer (TNBC) (Yamashita et al., 2013). Although the interaction between integrin α6β4 and vimentin has been defined, the biological impact of this interaction on breast cancer invasiveness has not been investigated.

In this study, we utilize two naturally occurring missense mutants in integrin β4 (R1225H and R1281W), found in epidermolysis bullosa patients, that impair integrin β4 binding to plectin (Koster et al., 2001), as well as integrin β4 signaling domain mutants, in conjunction with CRISPR-Cas9 editing to study the interaction of integrin β4, plectin and vimentin in the biology of poorly differentiated TNBC cells. Here, we uncover a spatially distributed array of complexes that involve integrin α6β4 and which assemble in a plectin-dependent manner. We further discover that integrin β4–plectin–vimentin complexes are critical for the integrin α6β4-dependent invasive growth of TNBC cells.

RESULTS

Plectin binding is required for spatial distribution of integrin α6β4

For this study, we utilize the TNBC cell line BT-549, which is one of the few established cell lines that do not express endogenous integrin β4 subunit but has ample integrin α6 to pair with β4 to form the α6β4 integrin receptor. These cells are also more representative of the TNBC than previously used MDA-MB-435 and represent the claudin low (Prat et al., 2013) and TNBC mesenchymal/mesenchymal stem-cell like (Bierie et al., 2017) breast cancer subtypes. To define how plectin impacts the pattern of integrin β4 distribution on the basal surface of the cells, we generated naturally occurring mutants of integrin β4 (R1225H or R1281W) in the plectin-binding domain, which reduce the affinity with plectin (Koster et al., 2004). We also created a novel 1355–1662 deletion mutant of integrin β4 (β4 Δ1355–1662), which retains the known N- and C-terminal plectin-binding sites (Rezniczek et al., 1998), but deletes the connecting segment and the third FNIII domain that control signaling (Nikolopoulos et al., 2005). Integrin β4-deficient BT-549 cells were transfected with EGFP-tagged integrin β4 [wild-type (WT) or various mutants], or an empty vector (EV) as a control (Fig. 1A). Cells were plated on laminin-1 and then treated briefly with epidermal growth factor (EGF) prior to immunostaining for plectin and lamin A/C. Using total internal reflection fluorescence (TIRF) microscopy, we observed that integrin β4 WT and β4 Δ1355–1662 were prominently localized on the basal surface beneath the nucleus as well as at the cell periphery within lamellae, where they frequently colocalized with plectin. However, integrin β4 R1225H and R1281W did not specifically concentrate near the nuclear or peripheral areas but were rather diffusely distributed throughout the cell basal surface and displayed a reduced colocalization with plectin (Fig. 1B). Pearson correlation coefficient of integrin β4 WT or β4 Δ1355–1662 colocalization with plectin was significantly higher than that for the plectin-binding mutants (R1225H or R1281W) (Fig. 1C). We also observed that cells expressing integrin β4 WT and β4 Δ1355–1662 spread over a greater surface area than did the EV, β4-R1225H mutant or β4-R1281W mutant cells (Fig. 1D).

Fig. 1.

Integrin β4 association with plectin determines the distribution of integrin α6β4. (A) Immunoblot analysis of EGFP–integrin β4 in BT-549 cells detected by rat anti-integrin β4 using actin as an internal control. Cells include BT-549 cells expressing empty vector (EV), or an EGFP-tagged wild-type integrin β4 (WT), β4 plectin-binding mutants (R1225H or R1281W), or β4 signaling domain deletion (Δ1355–1662). (B) Cells as noted in A were plated on laminin-coated cover slips and stimulated with EGF (5 ng/ml; 10 min). Cells were immunostained and imaged by TIRF microscopy to identify EGFP–integrin β4 (green), plectin (red), and lamin A/C (blue). Mean Pearson's correlation coefficients (PCC) for β4–plectin colocalization for the cells shown are indicated in merged images. (C) Mean±s.d. PCC from 60 cells collected from three separate experiments represented graphically. The box represents the 25–75th percentiles, and the median is indicated. The whiskers show the complete range. (D) Quantification of cell spreading of BT-549 cells expressing the indicated constructs and treated as in B. Data represent mean±s.e.m. size (μm²) collected from 20 cells from three separate experiments (n=60). (E) BT-549 cells expressing EGFP–integrin β4 WT and EV were plated on laminin for 4 h in the presence of 2% FBS, stained and then imaged by TIRF microscopy for EGFP–integrin β4 (green), plectin (red) and F-actin (blue). (F) BT-549 cells (EV or EGFP–β4 WT) were plated on fibronectin, stimulated with EGF, stained and imaged for EGFP–integrin β4 (green), paxillin (red), and F-actin (blue) as in B. Higher magnification of regions in white boxes in merged images are shown in the panels on the far right (B,E,F); white arrows indicate colocalization. Scale bars: 10 μm (whole-cell images); 2 μm (higher magnification images). All data are representative of at least three separate experiments. *P<0.05, **P<0.01 (two-tailed unpaired Student's t-test).

Immunocytochemical staining with an integrin β4 antibody showed similar integrin β4–plectin–F-actin structures to those seen with EGFP-tagged integrin β4 (Fig. S1A), suggesting that these observations are not artifacts of the tag, as recently demonstrated with an tdTomato tag (Elaimy et al., 2019). Interestingly, when cells were plated in the presence of serum, integrin β4 WT was recruited into numerous large, centrally localized puncta located under the nucleus, where it colocalized with plectin and thick f-actin fibers (Fig. 1E). These integrin β4–plectin–F-actin structures appeared to be similar in size to large FAs or fibrillar adhesions. However, when cells were plated on fibronectin alone and stimulated with EGF, these large complexes were not apparent (Fig. 1F).

Plectin binding is required for recruitment of integrin α6β4 into FAs

To clarify the relationship between integrin α6β4, plectin and FAs, we assessed the pattern of integrin β4 distribution on different matrices and used paxillin to mark FAs. Fibronectin is one of the major constituents of serum, and facilitates formation and elongation of FAs. On combined matrices of fibronectin and laminin-1, BT-549 β4 WT cells exhibited large adhesions similar to those found on laminin-1 in the presence of serum. These structures also stained strongly for paxillin, suggesting these structures are FAs (Fig. 2A). Integrin β4 WT and paxillin colocalization corresponded to the termini of actin stress fibers and were always found on the basal surface of the cell in close proximity to the nucleus, as well as along the leading edge of the cell (Fig. 2A). A similar distribution was also found with integrin β4 Δ1355–1662. The integrin β4 plectin-binding mutants (R1225H or R1281W), although they could form FAs, could not recruit β4 into large puncta near the nucleus (Fig. 2A). Quantitative analyses showed that integrin β4 WT- and Δ1355–1662-expressing cells formed more and larger adhesion complexes on the cell surface compared to the plectin-binding mutants (R1225H or R1281W), and these resembled FCs (2.5 μm²) and larger FAs (5 μm² or larger; Fig. 2B). Integrin β4 accumulation around FAs on the cell surface was not only identified by paxillin, but also by tenascin C (Fig. 2C), α6 integrin (Fig. S1B) and zyxin (Fig. S1C). These integrin α6β4-containing complexes, including FAs near the nucleus, were also formed in other TNBC cell lines, including MDA-MB-231, Sum159PT, 4T1, and MDA-MB-468 and, to a lesser extent, in immortalized MCF10A cells (Fig. 2D). Notably, when BT-549 cells were plated on laminin (Fig. S2) or fibronectin (Fig. 1F) alone, integrin β4 did not accumulate appreciably in paxillin-containing adhesions (Fig. S2). These data demonstrate that recruitment of integrin α6β4 to centrally localized FAs requires the cooperation between laminin and fibronectin adhesion receptors.

Fig. 2.

Plectin is required for recruitment of integrin β4 binding to FAs. (A) The indicated BT-549 cells were plated on coverslips coated with laminin and fibronectin in serum-free medium prior to stimulation with EGF (5 ng/ml; 10 min). Cells were stained and imaged as described in Fig. 1B to identify EGFP–integrin β4 (green), paxillin (red) and F-actin (blue). Higher magnification of regions around the nucleus and cell periphery from the merged images are from the areas highlighted by the white and red boxes, respectively, are shown in the far-right panels. Representative areas of colocalization are highlighted by white arrows. (B) Area distribution of various (top), >5 μm² (middle) or >2.5 μm² (bottom panel) sized complexes containing EGFP–integrin β4 that are found on central basal surface. Data indicate the mean±s.e.m. complex number from 60 cells. *P<0.05, **P<0.01; NS, not significant (two-tailed unpaired Student's t-test). (C) The indicated BT-549 cells were plated on coverslips coated with laminin and fibronectin in serum-free medium prior to stimulation with EGF (5 ng/ml; 10 min). Cells were stained and imaged as described above to identify EGFP–integrin β4 (green), tenascin C (red) and paxillin (blue). Higher magnifications of regions in the white boxes in the merged images are shown in the far-right panels and representative areas of colocalization are highlighted by white arrows. (D) FA distribution of integrin β4 in various TNBC cell models treated as in A with higher magnification (areas of white boxes) shown in lower panels. All data are representative of at least three separate experiments. Scale bars: 10 μm (whole-cell images); 2 μm (higher magnification images).

Plectin binding is critical for integrin β4-mediated cell migration and three-dimensional invasive growth

Next, we investigated the invasive growth in three-dimensional (3D) culture and migration of BT-549 cells stably expressing the various β4 subunit constructs (Fig. 3A). In 3D culture, integrin β4 WT- and β4 Δ1355–1662-expressing BT-549 cells had significantly accelerated cell growth, as evidenced by larger colony size, compared to the cells expressing EV or the integrin β4 plectin-binding mutants (R1225H or R1281W) (Fig. 3B,C). Importantly, integrin β4 WT-expressing cells displayed an invasive pattern with multiple cell protrusions extending from the colonies (Fig. 3B, arrows), which were absent in the other β4 mutants, including the mutant that lacked the β4 signaling domain (Δ1355–1662). Using a Transwell migration assay, we observed that integrin β4 WT expression accelerated cell migration of BT-549 cells toward EGF compared to that seen in cells with EV or expressing the integrin β4 mutants (R1225H, R1281W or Δ1355–1662) (Fig. 3D), despite the similar expression level of these constructs (Fig. 3A). These data indicate that integrin β4 N-terminal plectin-binding domain and the signaling domain are essential for effective enhancement of migration and invasive growth.

Fig. 3.

Integrin β4 requires plectin to promote invasive growth in 3D culture. (A) Immunoblot analysis of the integrin β4 subunit of the indicated stably transfected BT-549 cells using actin as an internal control. (B,C) Indicated BT-549 cells were cultured in 3D for 10 days, imaged by phase contrast and colony size calculated. Representative images (B) and mean±s.e.m. colony volume of 60 colonies collected from three different colonies (C) are shown. (D) Transwell migration assay of the indicated BT-549 cells toward EGF (5 ng/ml). Data represent mean±s.d. cell number migrated over 4 h per mm². (E) Immunoblot analysis of the integrin β4 subunit and plectin of the indicated stably transfected BT-549 cells using actin as an internal control. (F) Representative images of the indicated BT-549 populations cultured in 3D for 10 days. Scale bars: 50 μm (B,F). (G) The quantification of colony size of cells from E presented as mean±s.e.m. colony volume (μm³). (H) BT-549 and BT-549 plectin-knockout (BT-549 PLEC KO) clones with EGFP–integrin β4 WT were treated as in Fig. 2A and imaged by TIRF microscopy to identify EGFP–integrin β4 (green), paxillin (red) and F-actin (blue). Representative areas of colocalization are highlighted by white arrows. Scale bars: 10 μm (whole-cell images); 2 μm (higher magnification images). (I) Size distribution of EGFP–integrin β4 WT complexes in the indicated cells. Data represents mean±s.e.m. of 60 cells collected from three separate experiments. All data are representative of at least three separate experiments. *P<0.05, **P<0.01 (two-tailed unpaired Student's t-test).

To test the biological impact of plectin directly, knockout of plectin was performed using CRISPR-Cas9 technology in BT-549 cells. Selected BT-549 clones (PLEC KO-9 and PLEC KO-11) were chosen for further analysis and stable expression of integrin β4 WT was established in these plectin KO cells (Fig. 3E). In 3D culture, integrin β4 expression induced robust invasive growth, but not in the PLEC KO cells (Fig. 3F). Statistical analysis showed that depletion of plectin significantly reduced the invasive growth, with the greatest impact on cells expressing integrin β4 (Fig. 3G). These observations demonstrate that plectin is required for integrin β4 to drive BT-549 cell invasive growth.

To confirm the functional relevance of plectin binding to integrin β4 biology, BT-549 PLEC KO cells and BT-549 parental cells were transfected with EGFP-tagged integrin β4 WT and plated on laminin plus fibronectin matrices, stained for paxillin and F-actin, and imaged by TIRF microscopy. We found that BT-549 cells developed structures enriched in paxillin and F-actin with or without plectin (Fig. 3H). However, in the absence of plectin, integrin β4 distribution did not align with paxillin–F-actin structures (Fig. 3H). The number and size of integrin β4-containing puncta were significantly diminished in PLEC KO cells, compared with what was seen for cells expressing plectin (Fig. 3I). These data indicate that integrin α6β4 distribution on the basal surface of cells and its recruitment to large FAs specifically are regulated by plectin binding.

Integrin family members assist in integrin β4 recruitment to FAs

Fibronectin facilitated the accumulation of large integrin β4-containing complexes in the presence of laminin, thus suggesting that non-laminin binding integrin family members may be required. Major fibronectin-binding integrins include αvβ3, αvβ1, αvβ5 and α5β1 integrins (Multhaupt et al., 2016). Accordingly, we investigated the localization of two fibronectin-binding integrin family members involved in FA development, namely the αv and β1 integrins. Immunocytochemistry analysis of cells seeded on laminin plus fibronectin showed that integrin αv and integrin β1 localized with integrin β4 on the basal surface in close proximity to the nucleus, and also localized with the ends of actin stress fibers, which coincided with the location of FAs (Fig. 4A,B). These observations suggest that integrin family members can collaborate to alter the distribution of integrin β4 to FAs.

Fig. 4.

Integrin β4 aggregates with αv and β1 integrins. (A,B) BT-549 cells transfected with EV and EGFP–integrin β4 WT were plated on laminin plus fibronectin matrix, treated with EGF as performed in Fig. 2A, immunostained and imaged with TIRF microscopy for EGFP-integrin β4 (green), αv (A) or β1 (B) integrins (red), and F-actin (blue). Higher magnification of regions in the white boxes in the merged images are shown in the far-right panes and representative areas of β4 and αv or β1 colocalization are highlighted by white arrows. (C) Immunoblot analysis demonstrating integrin β1 knockout in BT-549 cells. (D) BT-549 β4 WT cells treated as in A were processed for PLA, as described in Materials and Methods section. TIRF microscopy was performed to identify PLA product (red, arrow) that represent integrin β4–plectin or integrin β1–plectin interactions as indicated, and F-actin (blue). (E) Quantification of PLA particles from 20 cells are reported as mean±s.e.m. from three independent experiments. (F) BT-549 and BT-549 ITGB1 KO cells were transfected with EV and EGFP-integrin β4, treated as in A and imaged by TIRF microscopy for EGFP–integrin β4 (green), paxillin (red) and F-actin (blue) with higher magnification noted in right panels. Scale bars: 10 μm (in whole-cell images in A,B,D,F), 2 μm (higher magnification images). (G) Quantification of integrin β4 puncta from F for 60 cells collected from three separate experiments is presented as mean±s.e.m. number of complexes. All data represent or include at least three separate experiments. **P<0.01 (two-tailed unpaired Student's t-test).

To test this concept, integrin β1 knockout was conducted by CRISPR-Cas9 technology (Fig. 4C). Proximity ligation assays (PLAs) were performed to detect intermolecular interaction between integrin β4, plectin and integrin β1. Cells were counterstained with phalloidin, and the fluorescence products of proximity ligation were observed by TIRF microscopy. As expected, integrin β4 and plectin directly associated in the center of control cells, but not in PLEC KO cells (Fig. 4D,E). These interactions were not altered by loss of β1 integrin. Next, PLAs for integrin β4 and plectin or β1 integrins and plectin were assessed, which demonstrated that β1 and β4 integrins resided in close proximity to plectin, but not when PLEC was knocked out (Fig. 4D), thus demonstrating the specificity of the reaction. Comparing BT-549 integrin β1 (ITGB1) and PLEC KO cells (Fig. 4E), these data suggest direct interaction of both integrin β4 and integrin β1 with plectin. A parallel immunocytochemistry experiment showed that BT-549 ITGB1 KO cells had significantly fewer paxillin-associated integrin β4 complexes (Fig. 4F,G), due to the depletion of integrin β1-associated FAs and FCs. These data indicate that β1 integrin family members are involved in recruitment of integrin β4 to FAs.

Plectin is required for integrin β4-dependent vimentin redistribution and colocalization in FAs

Given that integrin α6β4 binds to intermediate filaments yet can be recruited to FAs, we postulated that integrin α6β4 could recruit intermediate filaments to FAs. To test our hypothesis, we assessed the distribution of intermediate filaments, including vimentin and cytokeratins, as well as microtubules for BT-549 β4 cells plated on laminin plus fibronectin. Using 3D reconstruction of confocal images, we found that microtubule distribution did not match the pattern of integrin β4 (Fig. S3A,E). Anti-pan-cytokeratin staining showed that cytokeratins localized throughout the cells with weak fluorescence detected in the central areas near the nucleus (Fig. S3B,F). Furthermore, these cells lacked the well-defined keratin filaments that are seen in other epithelial cells, such as the Clone A colon carcinoma cells, which displayed a more extensive cytokeratin structure with more networked cytokeratin filaments that included longer filaments (Fig. S3C). In contrast, vimentin developed well-established structures around cell nucleus in BT-549 cells (Fig. S3D). To investigate the impact of integrin α6β4 on the distribution of vimentin, BT-549 cells were co-transfected with Orange2–vimentin and empty vector (EV) or EGFP–integrin β4, and imaged by TIRF microscopy as performed above. We observed that cells without β4 integrin displayed a mesenchymal distribution of vimentin in which fibrils and puncta of vimentin could be seen throughout the cell (Fig. 5A, top panels). In the presence of integrin β4 (Fig. 5A, lower panels, B), the predominant localization of vimentin was found in the center of the cell proximal to the nucleus. We found that integrin β4 and vimentin substantially colocalized with plectin in this region (Fig. 5A, white spots), where they also overlapped with paxillin (Fig. 5B).

Fig. 5.

Integrin β4 overlaps with plectin and vimentin in FAs. (A) BT-549 cells were co-transfected with Orange2–vimentin and EV or EGFP–integrin β4 WT. Cells were plated on laminin and fibronectin for 12 h in T4 medium plus 2% FBS, fixed, stained and imaged by TIRF microscopy to identify EGFP–integrin β4 (green), vimentin (red) and plectin (blue). (B) BT-549 and BT-549 PLEC KO cells were co-transfected with Orange2–vimentin and EV or EGFP–integrin β4 WT. Cells were treated as in A and imaged by TIRF microscopy for EGFP–integrin β4 (green), vimentin (red) and paxillin (blue). (C) Various TNBC cell lines treated as in A and imaged for EGFP–integrin β4 (green), vimentin (red), and paxillin (blue) in merged images. Higher magnification of regions in the white boxes in the merged images are shown in the far-right panels and representative co-localization areas are highlighted by white arrows. Scale bars: 10 μm (whole-cell images); 2 μm (higher magnification images).

To test the importance of plectin in integrin β4 contact with vimentin, BT-549 PLEC KO cells (Fig. 3E) were co-transfected with empty vector (EV) or EGFP–integrin β4 and Orange2–vimentin, treated as above, and imaged by TIRF microscopy. The resulting images demonstrated the colocalization of vimentin and paxillin in cells with and without integrin β4 (Fig. 5B). However, silencing plectin expression disrupted the colocalization of integrin β4 with vimentin as well as the redistribution of vimentin mediated by integrin β4 (Fig. 5B, lower panels). Notably, integrin β4–plectin–vimentin complexes were found in MDA-MB-231 cells (Fig. S4A). We further observed integrin β4–paxillin–vimentin complexes in MDA-MB-231, SUM159, 4T1 and MDA-MB-468 cells, and to a lesser extent in MCF10A cells (Fig. 5C). Depletion of plectin in MDA-MB-231 resulted in a dramatic reduction in integrin β4–vimentin puncta at the basal cell surface (Fig. S4B). These observations indicate that integrin β4 associates with vimentin on the cell basal surface near the nucleus, and plectin and integrin α6β4 association regulates the distribution of vimentin spatially. Furthermore, plectin is important for integrin β4–vimentin complex localization in FAs; however vimentin can localize to FAs in the absence of plectin and β4 integrin.

Vimentin is required for integrin β4-enhanced migration and invasive growth

To address the biological relevance of vimentin to integrin β4 signaling, we used CRISPR-Cas9 genome editing to knockout vimentin in BT-549 cells (BT-549 VIM KO) and then expressed integrin β4 in select clones (Fig. 6A). In the presence of vimentin, integrin α6β4 dramatically increased the activation of Fak (also known as PTK2), as indicated by Fak phosphorylation on Y397. In the absence of vimentin expression, this enhanced activation was completely blocked (Fig. 6A,B). Next, we assessed the impact of vimentin on the distribution of integrin β4 and paxillin using EGFP–integrin β4 WT-transfected BT-549 VIM KO cells. As visualized by TIRF microscopy, BT-549 VIM KO cells displayed normal paxillin–F-actin-containing FA structures (Fig. 6C compared to Fig. 2A). Integrin β4 condensed in paxillin–F-actin-associated FAs in BT-549 VIM KO cell; however, with the loss of vimentin, much of integrin α6β4 was shuttled into FAs (Fig. 6C). In 3D culture, loss of vimentin expression significantly diminished the size of BT-549 β4-expressing cell colonies but did not substantially impact colony size of the BT-549-EV cells (Fig. 6D,E). VIM knockout did not inhibit BT-549 cell motility towards EGF in a Transwell migration assay, but rather promoted a two-fold increase in migration. Expression of integrin β4 dramatically enhanced cell migration, but this effect was dramatically reduced with vimentin knockout (Fig. 6F). Interestingly, we did not observe significant differences between the growth of cells with and without vimentin under normal culture conditions, thus showing that vimentin is not required for integrin α6β4-accelerated growth in two dimensions (Fig. 6G).

Fig. 6.

Vimentin is required for integrin β4-enhanced migration and 3D invasive growth but not proliferation. (A) Immunoblot analysis of BT-549 EV, β4 WT and VIM KO (clones KO-1 and KO-5) cells with either EV or integrin β4 WT for expression of vimentin and integrin β4 subunit using actin as an internal control (top panels). Alternatively, cells were treated with EGF for 10 min prior to harvest and assessed for phospho-Y397 Fak, total Fak and actin with quantification of phospho-Fak compared to total Fak noted and normalized to EV value. (B) Statistical analysis of phospho-Fak intensity representing mean±s.d. of intensities collected from three separate experiments. (C) BT-549 VIM KO-1 cells were transfected with EV or EGFP-integrin β4 WT, plated on a laminin plus fibronectin matrix, stimulated briefly with EGF, stained and imaged by TIRF microscopy for EGFP-integrin β4 (green), paxillin (red) and F-actin (blue). Scale bars: 10 μm (whole-cell images); 2 μm (higher magnification images). (D) Representative images of the indicated cells grown in 3D culture for 10 days. (E) Quantification of colony size in C, represented as mean±s.e.m. colony volume (μm³). (F) Transwell migration assay of BT-549 VIM KO cells expressing either empty vector (EV) or integrin β4 WT toward EGF (5 ng/ml; 4 h). (G) Cell viability assay for cells from A. *P<0.05, **P<0.01; NS, not significant (two-tailed unpaired Student's t-test).

Finally, we tested the impact of reconstitution of vimentin in the VIM KO cells (Fig. 7A) on 3D invasive growth. We found that overexpression of vimentin increased colony growth in both BT-549 EV and β4 cells and reconstitution of vimentin effectively recovered the invasive growth of the VIM KO cells with integrin β4 (Fig. 7B,C). Ki67 staining also confirmed that reconstitution of vimentin restored the colony growth of VIM KO cells with integrin β4 (Fig. 7D,E). Collectively, these data indicate that the association of integrin β4 with vimentin is critical for integrin α6β4 to enhance migration and 3D invasive growth of breast cancer cells, but loss of this association does not interrupt integrin β4-enhanced cell growth under normal two-dimensional (2D) culture conditions.

Fig. 7.

Reconstitution of vimentin restores integrin β4-enhanced 3D invasive growth in VIM KO cells. (A) Immunoblot analysis of select BT-549 populations for integrin β4, vimentin and actin expression. (B) Representative images of the indicated cells grown in 3D culture for 7 days. Scale bars: 50 μm. (C) Quantification of colony size from B, represented as mean±s.e.m. colony volume (μm³). (D) Indicated cell lines were cultured in 3D for 5 days, and fixed, stained and imaged by confocal microscopy for tubulin (green), Ki67 (red), and DAPI (blue). White arrows indicate Ki67 positive cells. Bars represent 10 μm. (E) Statistic analysis of the Ki67 positive cell counts from (D). Data represents mean±s.d. of total 300 cells collected from three separate experiments. *P<0.05, **P<0.01 (two-tailed unpaired Student's t-test).

DISCUSSION

Our study highlights the importance of the microenvironment on the distribution and composition of integrin α6β4-containing adhesion complexes. Specifically, we demonstrate that the cooperation of laminin and fibronectin ECM components in the formation, distribution and signaling of integrin α6β4 receptor complexes. On laminin alone, integrin α6β4 forms smaller complexes throughout the cell and colocalizes with paxillin predominantly at the leading edge (Fig. S2). This observation is in line with previous studies on integrin α6β4 (Elaimy et al., 2019; Lotz et al., 1997) and a prevailing concept that laminin integrins tend to form smaller FCs (DiPersio et al., 1995). Previous work in keratinocytes suggests that hemidesmosomes and FAs associate but treadmill as separate entities, particularly at the leading edge of a migrating cell (Pora et al., 2019). We find that with the addition of fibronectin, a matrix protein that supports larger FA formation through β1 and αv integrins, integrin α6β4 is recruited to large FAs that predominate in the center of the cell proximal to the nucleus, resemble fibrillar adhesions and colocalize with the fibillar adhesion marker tenascin C. Notably, fibronectin can be found in the basement membrane in vivo, and is encountered by epithelial cells during invasion through the stroma or with the induction of EMT, where carcinoma cells subsequently deposit fibronectin (Insua-Rodríguez and Oskarsson, 2016). As a result, carcinoma cells, and especially those that have undergone EMT, are capable of displaying a variety of integrin α6β4 associations including FAs, fibrillar adhesions and intermediate filament-associated complexes. These carcinomas include the mesenchymal subtype of TNBC, which the BT-549 and MDA-MB-231 cell lines used here represent. Notably, the Weinberg group identified integrin β4 as a marker of cancer stem cell-rich populations within the mesenchymal subtype that associates with a worse 5-year progression-free survival (Bierie et al., 2017).

The distribution of integrin α6β4 depends on the ability of the β4 subunit to bind plectin. We find that integrin β4 with the R1225H and R1281W mutations, as well as PLEC knockout, reduce integrin α6β4 distribution to a random pattern, prevent its accumulation in FAs, and severely stunt the migration and invasiveness of breast cancer cells. Whereas integrin α6β4 in normal epithelial cells binds to plectin to initiate and solidifies the formation of hemidesmosomes, these observations suggest that this integrin β4–plectin association might not be synonymous with the presence of mature hemidesmosomes in carcinoma cells. We find that plectin association defines the patterning of integrin α6β4 complexes throughout the cell, leading to small FAs in the lamellipodium of cells as well as large FAs or fibrillar adhesions that form on the basal surface peripheral to the nucleus. These complexes might involve F-actin and other integrins, or also incorporate vimentin intermediate filaments. A recent study by the Jones group demonstrated that integrin α6β4 associates with vimentin in lamellae of lung carcinoma cell line A549 and that these complexes increase the activation of Rac1 (Colburn and Jones, 2018). Our previous studies demonstrated that although integrin α6β4 enhances Rac1 activation, it cooperates with β1 integrins to amplify Rac1 activity (O'Connor et al., 2012). These observations suggest that integrins cooperate between species but also cooperate between cytoskeletons to promote migratory signals.

Integrin α6β4 has been previously shown to bind vimentin through its association with plectin in endothelial cells (Homan et al., 1998, 2002) and keratinocytes (Geerts et al., 1999). Despite the importance of both integrin α6β4 and vimentin to breast cancer progression, the biological importance of this association has been underappreciated. Vimentin, like fibronectin, is upregulated during EMT by epithelial cells and carcinoma cells. Hemidesmosomes are dismantled during the EMT process leading to the concept that integrin α6β4 promotes many of the phenotypes attributed to EMT. This concept is supported by our previous study that demonstrated that EMT-associated S100A4, also known as fibroblast specific protein-1 (FSP-1), is epigenetically regulated by integrin α6β4 (Chen et al., 2009). This study demonstrates that, in poorly differentiated cells, integrin α6β4 bound to vimentin through plectin binding is recruited to FAs in the presence of fibronectin, thus providing a mechanism of how integrin α6β4 can cooperate with β1 and αv integrins to promote Fak phosphorylation, as depicted in Fig. 8. Previous work in fibroblasts has demonstrated that vimentin intermediate filaments can increase mechanotransduction through FAs and the actomyosin network, which in turn impacts activation of Fak and directed cell migration (Gregor et al., 2014). Our data demonstrate that vimentin is not of particular importance to breast cancer cell migration and invasive growth in the absence of integrin α6β4. In fact, in the absence of integrin α6β4, loss of vimentin actually promotes a two-fold increase in migration of the BT-549 cells. However, integrin α6β4 alters the overall distribution of vimentin and requires vimentin to enhance cell migration and invasive growth of TNBC cells. We also note that vimentin is dispensable for integrin α6β4-enhanced growth in 2D but not in 3D, and that the interaction of these two molecules through plectin is essential for cell migration and invasion. Although BT-549 cells do not display robust cytokeratin networks under the conditions utilized in this study, one cannot underestimate the impact of integrin α6β4 to connect with these structures and provide overall architecture to migrating cells.

Fig. 8.

Proposed mechanism for plectin-dependent integrin α6β4 and vimentin recruitment to focal/fibrillar adhesions. Integrin α6β4, through its binding to plectin, recruits vimentin to FAs and fibrillar adhesions (marked by tenascin C) in cooperation with other laminin- and fibronectin-binding integrins, which results in enhanced Fak phosphorylation, migration and invasive growth. Although the integrin α6β4-mediated recruitment of vimentin to focal and fibrillar adhesions can occur at the cell periphery, these complexes are particularly concentrated on the plasma membrane near the nucleus.

It is interesting to note that our novel signaling domain β4 mutant, which retains both the N and C-terminal plectin-binding domains, is capable of supporting cell spreading and patterned distribution of integrin α6β4. This mutant differs from the previously reported signaling domain mutant (Nikolopoulos et al., 2005) in that our mutant retains the C-terminal plectin-binding domain. The utilization of both plectin-binding domains is suggested to fold integrin β4 into a closed conformation that restructures the connecting segment and blocks signaling (Frijns et al., 2012; Pereda et al., 2009). The deletion of this region also eliminates important binding sites for BP180 (collagen XVII) and BP280 (Romagnoli et al., 2019). Notably, hemidesmosomes can modulate force generated from FAs (Wang et al., 2020) suggesting that disassembly of hemidesmosomes can increase mechanotransduction. These observations suggest that the integrin β4 subunit promotes cell spreading and its patterned distribution through mechanical mechanisms or through the release of BP180 and/or BP230 components. Alternatively, there might be signaling components of β4 that reside outside of the currently recognized signaling domain. How these plectin-binding domains cooperate in mechanotransduction to drive the signaling and biological properties downstream of integrin α6β4 will require further investigation.

In summary, we demonstrate a critical cooperation between integrin α6β4 and vimentin in driving the migration and invasive growth of TNBC cells. This event leads to restructuring of the vimentin network and recruitment of integrin α6β4 and vimentin to FAs and fibrillar adhesions in a manner that is dependent on plectin binding. This co-dependency between integrin α6β4 and vimentin suggests that these molecules cooperate and drive the aggressive biological properties downstream of integrin α6β4 in the progression of TNBC.

MATERIALS AND METHODS

Plasmid construction

The full-length wild-type (WT) integrin β4 (Dans et al., 2001) cDNA was obtained from pcDNA3.1/Myc-His-beta4 (Addgene #16039) and then, subcloned into pBABE-puro, pBABE-hygro or pEGFP-N1 plasmid using EcoRI+Sal1 or XhoI+KpnI sites. Integrin β4 R1225H, R1281W, and Δ1355-1662 were generated from the integrin β4 WT constructs using Pfu-Ultra-based mutagenesis PCR (Agilent Technologies). Mutagenesis primers include 5′-GAGCCAGGGCATCTGGCCT-3′ and 5′-AGGCCAGATGCCCTGGCTC-3′ for β4 R1225H, 5′-GACAACCCTAAGAACTGGATGCTGCTTATTGAG-3′ and 5′-CTCAATAAGCAGCATCCAGTTCTTAGGGTTGTC-3′ for β4 R1281W, 5′-ACCATAGAGTCCCAGGATGGAGGAC-3′ and 5′-GCGTAGAACGTCATCGCTGT-3′ for β4 Δ1355-1662. The vimentin construct (mOrange2–vimentin-7) was obtained from Addgene (#57977). Plasmids were sequenced by Eurofins MWG Operon (Louisville, KY).

Cell culture, transfection and stable expression of integrin β4

BT-549 cells (ATCC, HTB-122) were cultured in RPMI 1640 with 10% fetal bovine serum (FBS; Sigma-Aldrich, St Louis, MO) and insulin (1 µg/ml; Sigma, 11070-73-8). MDA-MB-231 cells (ATCC, CRM-HTB-26) were grown in Dulbecco's modified Eagle's medium (DMEM; low glucose) with 10% FBS. 4T1 cells (ATCC, CRL-2539) were cultured in RPMI 1640 with 10% FBS. MDA-MB-468 (ATCC, HTB-132) and Sum159PT (Asterand Bioscience) cells were cultured in DMEM-F12 with 10% FBS. MCF10A cells (ATCC, CRL-10317) were grown in DMEM-F12 medium with 5% horse serum, 20 ng/ml EGF (PeproTech, Inc, AF-100-15), 10 μg/ml insulin, 0.5 mg/ml hydrocortisone (Sigma, 50-23-7) and 100 ng/ml cholera toxin. Media were supplemented with 1% penicillin and 1% streptomycin (GIBCO by Life Technologies, Grand Island, NY). Cells were transfected with mammalian expression plasmids, as indicated, using TransIT^®-BrCa Transfection Reagent (Mirus) according to the manufacturer's instructions. Experiments were conducted at 24–48 h after transfection. Retroviral expression system was used to deliver stable expression of integrin β4. Retrovirus constructs encoding integrin β4 (ITGB4) and its mutants were packaged in 293 T cells (ATCC catalogue # CRL-3216), infected into BT-549 cells and selected with puromycin (1 μg/ml) or hygromycin B (50 μg/ml) for 14 days. Then, cells were sorted to obtain the same level of integrin β4 by Flow Cytometry and Immune Monitoring Core Facility at the University of Kentucky using iCyt-Sony Cell Sorter System (iCyt Mission Technology, Inc.). Integrin β4 WT and the mutants were detected by rat anti-integrin β4 (BD, 439-9B, 1:1000). All cell lines, including all derivative cells (ITGB4 CRISPR gene editing and retroviral expression) were authenticated by Laboratory Corporation of America Holdings (LabCorp) using short tandem repeats (STRs) profiling.

CRISPR-Cas9 gene engineering

CRISPR gene engineering was performed using pSpCas9(BB)-2A-Puro (PX459) V2.0 (Addgene #62988). Guide RNA (gRNA) design and knockout cell lines were generated as described previously (Ran et al., 2013). Vimentin target gRNAs were 5′-CACCGTGGACGTAGTCACGTAGCTC-3′ and 5′-AAACGAGCTACGTGACTACGTCCAC-3′. Plectin target gRNAs were 5′-CACCCGAATGACCGCCCGCTCAGC-3′ and 5′-AAACGAGCTACGTGACTACGTCCAC-3′. Integrin β1 gRNAs were 5′-CACCGAAGCAATAGAAGGTACGGTG-3′ and 5′-AAACCACCGTACCTTCTATTGCTTC-3′. gRNAs were subcloned into PX459 vector in the BBS1 restriction site. CRISPR constructs and one scramble construct were transfected respectively with TransIT^®-BrCa Transfection Reagent (Mirus). After 24 h, the cells were detached and divided into 10 cm cell culture dishes. Selection was performed using puromycin (0.75 μg/ml) for 3 days, followed by 2–3 weeks of normal cell culture before individual clones were validated by immunoblot analysis for the targets. All CRISPR plasmids were sequenced by Eurofins MWG Operon (Louisville, KY).

Immunoblotting

Cells were washed twice with cold PBS and collected with 400 μl cold RIPA buffer (1 M Tris-HCl pH 7.4, 5 M NaCl, 20% NP-40, 10% sodium deoxycholate, 20% SDS and 0.5 mM PMSF) supplied with protease inhibitor cocktail (Sigma-Aldrich, P2714). Extracts were collected by centrifugation (16,000 g for 5 min), separated by SDS-PAGE (8–10% gels), transferred and immunoblotted for integrin β4 (BD, 439-9B, 1:10,000), plectin (BD, Clone 31, 1:1000), integrin β1 (Millipore AB1952, 1:1000), Fak (Cell Signaling Technologies, #D5O7 U, 1:3000), phospho-Y397 Fak (Invitrogen, #44-624G, 1:10,000), vimentin (Invitrogen, #MA5-11883, 1:10,000) and β-actin (Sigma-Aldrich, 1:10,000), as a loading control.

Immunocytochemistry and imaging

Cells were plated on coverslips coated with laminin-I (10 μg/ml; Trevigen) with or without fibronectin (10 μg/ml; BD Biosciences) in RPMI1640 with 2% BSA and allowed to adhere for 4–6 h and then stimulated with EGF (5 ng/ml) for 15 min. Cells were then fixed for 15 min with 4% paraformaldehyde containing 10 mM PIPES, pH 6.8, 2 mM EGTA, 2 mM MgCl₂, 7% sucrose and 100 mM KCl at room temperature, and then permeabilized with 0.05% Triton X-100 in PBS, as described previously (Chen et al., 2009). Cells were blocked for 1 h with 3% BSA plus 1% goat serum in PBS. The following primary antibodies were used at indicated concentrations and incubated in block solution at 4°C overnight: rabbit anti-lamin A/C (Abcam, EPR4100, 1:2000), mouse anti-plectin (BD, clone 31, 1:100), mouse anti-paxillin (BD, clone 349, 1:300), rabbit anti-tenascin C (Cell Signaling #33352, 1:600), rabbit anti-integrin β1 (Millipore, clone AB1952, 1:300), rat anti-integrin α6 (Santa Cruz Biotechnology, GOH3, sc-19622, 1:300), rabbit anti-zyxin antibody (Abcam, EPR4302, ab109316, 1:300), and mouse anti-integrin αv antibody (Santa Cruz Biotechnology, clone P2W7, 1:300). For pan cytokeratin staining, cells were treated with cold methanol for 5 min, and incubated with mouse anti-pan keratin (Sigma C2562, 1:400) overnight. Secondary antibodies, including Cy3-conjugated goat anti-rabbit-IgG and Cy2-conjugated goat anti-mouse-IgG (Jackson Immuno Research, 1:400), and Alexa Fluor 647–phalloidin counterstain (Invitrogen, 1:100) were incubated in block solution for 1 h at room temperature in dark. Coverslips were mounted on glass slides using 50% glycerol solution and sealed with clear nail polish. For 3D cell colony staining, Matrigel containing 3D cultures were spread on glass slides, fixed with ice-cold methanol plus acetone (1:1) for 20 min on ice, blocked for 1 h with 3% BSA in immunofluorescent (IF) buffer (130 mm NaCl, 7 mm Na₂HPO₄, 3.5 mm NaH₂PO₄, 7.7 mm sodium azide, 0.1% BSA, 0.2% Triton X-100, and 0.05% Tween-20), and incubated with rabbit anti-Ki67 (Millipore AB9260, 1:700) and rat anti-tubulin (Novusbio NB100-1639, 1:5000) overnight at 4°C. Secondary antibodies, including Cy3-conjugated goat anti-rabbit-IgG and Cy2-conjugated goat anti-rat-IgG (Jackson Immuno Research, 1:400), and co-stained with DAPI (Invitrogen, 1:100) were incubated in 3% BSA in IF buffer for 1 h at room temperature in dark.

TIRF images were acquired using a Nikon Eclipse Ti microscope, Nikon A1 camera, 60× oil objective, and numerical aperture of 1.49. Images were processed for colocalization analysis and Pearson's correlation coefficient (PCC), cell spreading or integrin β4 complex size using NIS Elements AR 3.2 software. For enumeration of integrin α6β4-containing complexes, a cut-off of 0.5 μm was used to reduce non-specific associations. For all quantifications, 20 randomly chosen cells from three different experiments (final n=60) were used to obtain unbiased results.

Serial confocal images were acquired using a Nikon Eclipse Ti2 confocal microscope with Nikon A1 Plus camera, 100× oil objective, numerical aperture 1.45, Z-step 0.02 µm and Nikon NIS Elements AR 4.6 software acquisition software. 3D reconstitutions and all image preparations were performed with Nikon NIS Elements AR 4.6. 3D volume rendering options utilized high resolution, maximum intensity projection blending and smooth volume.

Proximity ligation assay

For the PLA, cells were fixed and permeabilized as described above. Primary antibodies used were mouse anti-integrin β4 (1:100), rabbit anti-plectin (1:100), mouse anti-plectin (1:100) and rabbit anti-human integrin β1 (1:300; Millipore Sigma, AB1952). PLA was carried out with Duolink^® In Situ Detection Reagents Orange (#DUO92007, Sigma Aldrich), Duolink^® In Situ PLA^® Probe Anti-Rabbit PLUS/MINUS (#DUO92002/DUO92005, Sigma Aldrich) and Duolink^® In Situ PLA^® Probe Anti-Mouse PLUS/MINUS (#DUO92001/DUO92004, Sigma Aldrich). PLA was detected using TIRF microscopy as described above.

Migration assay

Transwell inserts (24-well insert, pore size, 8 mm; Corning) were coated with laminin-I (10 μg/ml) for 30 min, and then washed three times with serum-free medium. Cells (5×10⁴) were plated in the top chamber, and EGF (5 ng/ml) diluted in RPMI 1640 medium was placed in the lower chamber. After incubation for 4 h, cells that did not migrate through the pores were removed with a cotton swab. Cells on the lower surface of the membrane were fixed with methanol and stained with 1% Crystal Violet in 2% ethanol. Migrated cells from five different fields were counted under microscope and averaged, as done previously (Harrison et al., 2013).

3D culture assay

The 3D culture assay was performed as previously described (Chen et al., 2009; Lee et al., 2007). Briefly, the culture surface of 24-well plate (prechilled) was coated with a thin layer of growth factor reduced Matrigel (BD Biosciences, Bedford, MA, USA) and incubated for 15–30 min at 37°C to allow the Matrigel to gel. Then, 2.5×10³ cells mixed with 0.5 ml T4 medium (containing 250 ng/ml insulin, 10 μg/ml transferrin, 2.6 ng/ml sodium selenite, 10⁻⁷ M estradiol, 5 μg/ml prolactin, and 20 ng/ml EGF) were plated on Matrigel surface, and cultured at 37°C for 1 h to let cells adhere, then medium was removed and replaced by 0.5 ml 3% Matrigel mixed with T4 medium plus 2% FBS. Culture was maintained for 10 days with medium changed every 2 days.

Viability assay

Viability assays were performed using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) as described previously (Chen et al., 2011). Briefly, cells (2×10³) were seeded in triplicate using a 96-well plate and cultured normally. MTT (5 mg/ml) was added at designated culture times and incubated at 37°C for 3 h. Formazan precipitate was solubilized with 100 μl of stop-mix solution (90% isopropanol, 10% DMSO) and assessed for optical density (OD) at 570 nm.

Statistics

Data were compared and analyzed using a two-tailed unpaired Student's t-test. All experiments were performed at least three times to confirm reproducibility. Quantification of Pearson's correlation, cell spreading and adhesion complex size was performed on a minimum of 20 cells from at least two technical replicates. Colony size quantification was determined from a minimum of 50 colonies while migration assays were performed with quadruplicate technical replicates. Data are presented as mean±s.d. or ±s.e.m., as noted.

Peer review history

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.258471.