The novel ECM protein SNED1 mediates cell adhesion via the RGD-binding integrins α5β1 and αvβ3 Open Access

The extracellular matrix (ECM), a fundamental component of multicellular organisms, is a complex 3-dimensional (3D) meshwork consisting of over 100 proteins (Hynes and Naba, 2012; Naba, 2024). The primary function of the ECM is to serve as a substrate for cell adhesion (Hynes, 2009). The adhesion of cells to their surrounding ECM is mediated by cell-surface receptors and is crucial for cell survival, as detachment from the ECM results in apoptotic cell death, known as anoikis (Frisch and Francis, 1994). In addition, cell–ECM interactions trigger molecular events that regulate a multitude of cellular phenotypes, including migration (Dzamba and DeSimone, 2018; Pally and Naba, 2024), proliferation (Hynes, 2009) and differentiation (Walma and Yamada, 2020). As a result, alterations of cell–ECM adhesions and downstream signaling pathways lead to developmental defects (Rozario and DeSimone, 2010) and pathologies like cancer (Cox, 2021; Pickup et al., 2014) and fibrosis (Herrera et al., 2018). However, only a small subset of the hundreds of proteins comprising the matrisome is the focus of active investigations, and the mechanisms by which they interact with cells and guide cell phenotype are known for an even smaller subset.

One such understudied ECM protein is ‘sushi, nidogen, and EGF like domains 1’ (SNED1). The murine gene Sned1 was cloned two decades ago (Leimeister et al., 2004); however, it took 10 years to identify its first function as a promoter of breast cancer metastasis (Naba et al., 2014). Beyond its role in breast cancer, we recently reported that Sned1 is an essential gene, as knocking it out resulted in early neonatal lethality and severe craniofacial malformations (Barqué et al., 2021). We further showed that knocking out Sned1 specifically from neural crest cells, the cell population that contributes to forming most craniofacial features (Mankarious and Goudy, 2010; Martik and Bronner, 2021; Trainor, 2005), was sufficient to recapitulate the craniofacial phenotype observed upon global Sned1 deletion, demonstrating a new role for SNED1 in craniofacial morphogenesis (Barqué et al., 2021). However, as of today, the mechanisms through which SNED1 interacts with cells to mediate its phenotypes remain unknown. Of note, metastatic breast cancer cells and neural crest cells share common features (Gallik et al., 2017), including their ability to remodel their adhesions to acquire increased migratory potential (Doyle et al., 2022; Gallik et al., 2017; Mayor and Theveneau, 2013; Tucker et al., 1988). This process is, in part, mediated by integrins, which are the main class of ECM receptors (Bökel and Brown, 2002; Campbell and Humphries, 2011; Hood and Cheresh, 2002; Hynes, 2002; Kanchanawong and Calderwood, 2023). For example, in vivo and in vitro experiments have demonstrated that β1 integrins at the surface of neural crest cells interact with ECM proteins, like fibronectin, to promote cell adhesion and subsequent migration (Alfandari et al., 2003; Duband et al., 1991; Leonard and Taneyhill, 2020; Pietri et al., 2004; Testaz et al., 1999; Yang et al., 1993). β1-containing integrin heterodimers expressed by breast cancer cells have been shown to interact with the vascular ECM to promote extravasation during metastasis (Chen et al., 2016) and be essential for every step of the metastatic cascade (Hamidi and Ivaska, 2018). Interestingly, SNED1 contains two putative integrin-binding motifs, an arginine-glycine-aspartic acid (RGD) triplet and a leucine-aspartic acid-valine (LDV) triplet, known in other proteins to mediate cell–ECM adhesion (Vallet et al., 2021; Lawler et al., 1988; Ruoslahti and Pierschbacher, 1987). We thus sought to determine whether SNED1 played a role in cell adhesion.

Here, we report that SNED1 mediates the adhesion of breast cancer cells and neural crest cells, two cell types of relevance to the in vivo functions of SNED1. Using a combination of genetic and pharmacological approaches, we further show that cell adhesion to SNED1 is mediated by its RGD motif and the engagement of α5β1 and αvβ3 integrins. Our study is thus the first to report the identification of SNED1 receptors and constitutes an important step toward the identification of the biochemical signaling events leading to SNED1-dependent breast cancer metastasis and craniofacial development.

We first sought to determine whether SNED1 could mediate cell adhesion. To do so, we seeded highly metastatic MDA-MB-231 ‘LM2’ breast cancer cells (hereafter termed LM2) or O9-1 neural crest cells on surfaces coated with increasing concentrations of purified human SNED1 (Vallet et al., 2021) (Fig. S1A) or murine Sned1, respectively, and allowed cells to adhere for 30 min. Cell adhesion was assayed using a Crystal-Violet-based colorimetric assay, and we found that SNED1 mediated the adhesion of LM2 breast cancer cells and O9-1 neural crest cells in a concentration-dependent manner with maximum adhesion observed at 10 µg/ml concentration (Fig. 1A,B, respectively). At this concentration, SNED1 exerted the same adhesive properties as fibronectin toward these two cell populations (represented by 100% adhesion based on the standard curve). We further showed that the O9-1 cells, which are of murine origin, adhered in the same proportion to both murine Sned1 and human SNED1 (Fig. S1B), perhaps not surprisingly, as the two orthologs share an 84% sequence identity (Barqué et al., 2021).

Sned1 also mediated the adhesion of immortalized mouse embryonic fibroblasts derived from Sned1 knockout mice (Sned1^KO iMEFs) to the same extent as the other two cell lines (Fig. S1C). Altogether, these experiments demonstrate that SNED1 mediates cell adhesion and exerts adhesive properties toward a panel of cell types.

SNED1 is a modular protein composed of different domains, including a NIDO domain, a follistatin domain and a sushi domain (also known as complement control protein or CCP domain), in addition to multiple repeats of EGF-like (EGF), Ca²⁺-binding EGF-like (EGF-Ca) and fibronectin type 3 (FN3) domains, all of which are involved in protein–protein interactions (Fig. 2A). To assess which region(s) of SNED1 mediate cell adhesion, we generated and purified three truncated forms of SNED1 (Fig. 2A) – the SNED1^1-751 form lacks the C-terminal region that comprises three FN3 domains, shown to mediate cell adhesion in other ECM proteins, and the most C-terminal EGF-like domains; the SNED1^1-530 fragment additionally lacks the sushi domain and EGF-like domains after the follistatin domain; and the SNED1^1-260 construct, which encompasses the very N-terminal region, including a single NIDO domain, which is only present in four other proteins in the human and mouse proteomes (nidogen-1 and nidogen-2, α-tectorin and mucin-4) and its function remains unknown (Barqué et al., 2021). These synthetic constructs were designed to include adequate domain boundaries compatible with proper folding for the truncated proteins to be secreted (Fig. 2B).

Using these purified truncated proteins as substrates, we found that LM2 cells had a decreased ability to adhere to SNED1^1-751 and SNED1^1-530 as compared to full-length SNED1 (24.6% and 36.6% decrease, respectively; Fig. 2C). Similarly, we observed that O9-1 cells had a decreased ability to adhere to SNED1^1-751 and SNED1^1-530 compared to full-length SNED1 (28.7% and 32.5% decrease, respectively; Fig. 2D). Notably, although SNED1^1-530 further lacks the sushi domain and four EGF-like domains, along with the domains absent in SNED1^1-751, it mediated cell adhesion to the same extent as SNED1^1-751 for both cell lines (Fig. 2C,D). Interestingly, LM2 and O9-1 cells could adhere to the shortest N-terminal SNED1^1-260 fragment in a similar proportion to what was seen for full-length SNED1 (Fig. 2C,D, respectively). Given that the SNED1^1-260 fragment is six times smaller than full-length SNED1, we thought to perform the same experiment but using similar molar concentration (66.3 µM) rather than amount (10 µg/ml), and obtained a similar result, namely that LM2 cell adhesion on SNED1^1-260 is comparable to that seen for full-length SNED1 (Fig. S1D). Altogether, these results indicate that the adhesive property of SNED1 is primarily mediated by its N-terminal region. These results also suggest that the three FN3 domains and the EGF-like domains lacking in the SNED1^1-751 and SNED1^1-530 constructs are required for full-length SNED1 to adopt a conformation where the N-terminal adhesive site is fully accessible to cells, given that their absence resulted in decreased cell adhesion.

Analysis of the SNED1 sequence has revealed the presence of two putative integrin-binding motifs: RGD and LDV. These motifs were discovered in other ECM proteins, such as fibronectin and thrombospondin 1, and interact with integrin heterodimers at the cell surface to mediate cell adhesion (Lawler et al., 1988; Ruoslahti and Pierschbacher, 1987).

To determine whether the RGD and LDV motifs of SNED1 are required for cell adhesion, we mutated these sites alone or in combination (p40D>E in the RGD motif; p311D>A in the LDV motif; Fig. 3A). Similar mutations in other ECM proteins have been shown to disrupt their interaction with integrin heterodimers (Cherny et al., 1993; Pytela et al., 1985). We next expressed these constructs in HEK293T cells (hereafter termed 293T) and purified the corresponding secreted proteins from the conditioned culture medium using affinity chromatography (Fig. S2). Purified proteins were then used to perform cell adhesion assays. We observed that LM2 cells showed a statistically significant decrease in their ability to adhere to SNED1^RGE and SNED1^RGE/LAV as compared to wild-type SNED1 (SNED1^WT) (64.7% and 57.9% decrease, respectively; Fig. 3B). Similarly, O9-1 cells showed a significant decrease in their ability to adhere to SNED1^RGE and SNED1^RGE/LAV as compared to SNED1^WT (65.4% and 72% decrease, respectively; Fig. 3C). However, mutation of the LDV motif to LAV did not affect cell adhesion (Fig. 3B,C). These results demonstrate that the RGD but not the LDV motif in SNED1 is required for cell adhesion.

To complement this set of observations and determine whether integrins mediate adhesion to SNED1, we performed experiments aimed at targeting the ability of integrins to interact with SNED1. First, we performed adhesion assays in the presence of cyclic RGDfV (cRGDfV) peptide, a peptide known to bind with high affinity to integrins that engage with the RGD motif of ECM proteins (Aumailley et al., 1991). We observed reduced adhesion of both LM2 (Fig. 4A) and O9-1 cells (Fig. 4B) in a concentration-dependent manner in presence of cRGDfV as compared to vehicle-treated cells. Given that we have previously shown that the N-terminal region of SNED1 is sufficient to mediate cell adhesion and the RGD motif is located within this fragment, we evaluated the ability of cells to adhere to SNED1^1-260 in presence of 10 µM of cRGDfV peptide. We found that this integrin inhibitor fully abrogated LM2 and O9-1 cell adhesion (Fig. S3A,B). This result suggests that the RGD motif at the N-terminal end is essential for the adhesive property of the SNED1^1-260.

The cRGDfV peptide inhibits a panel of RGD-binding integrin heterodimers, such as αvβ3, αvβ6, α5β1, and αvβ5 (Kapp et al., 2017). Using computational molecular modeling, we previously predicted that SNED1 could potentially interact with 11 integrin subunits (α1, α4, α7, α10, α11, β1, β2, β3, β4, β5 and β7; Vallet et al., 2021) forming six functional heterodimers (α1β1, α4β1, α7β1, α10β1, α11β1 and α4β7; Hynes, 2002), including the RGD-binding integrin α5β1 (Fig. 4C). We thus sought to test whether cell adhesion to SNED1 was dependent on α5β1 integrin. We first confirmed that LM2 and O9-1 cells expressed α5β1 integrin (Fig. 4D). We then performed cell adhesion assays in presence of functional blocking antibodies targeting β1 and α5 integrins and demonstrated that the adhesion of LM2 cells (Fig. 4E) and O9-1 cells (Fig. 4F) to SNED1 was significantly reduced in presence of antibodies targeting β1 integrin (52% and 41% decrease, respectively, as compared to isotype controls) or α5 integrin (30% and 9% decrease, respectively, as compared to isotype controls).

Although cRGDfV inhibits α5β1 integrin, αvβ3 is more sensitive to inhibition by this peptide (Gurrath et al., 1992; Pfaff et al., 1994). Given that LM2 cells also express αvβ3 (Fig. 4D), we sought to test whether αvβ3, could act as a receptor for SNED1, although it was not predicted by our modeling approach. Interestingly, we observed that the adhesion of LM2 cell was significantly decreased in presence of an antibody targeting αvβ3 integrin (35.5% decrease as compared to isotype control; Fig. 4E). This result is in line with our observation of decreased cell adhesion to SNED1^RGE. Altogether, these experiments identified α5β1 and αvβ3 integrin as the first SNED1 receptors.

It is worth noting that inhibition of α5, β1, or αvβ3 integrins did not fully abrogate cell adhesion to SNED1, suggesting that there are likely additional integrin (such as the RGD-integrin αvβ1 or the non-RGD integrin α1β1) and non-integrin receptors at the surface of these cells. Indeed, we have previously shown using computational prediction, that SNED1 can interact with 55 transmembrane proteins, in addition to integrins, including the basal cell adhesion molecule (BCAM) or dystroglycan 1 (DAG1), two known ECM receptors (Vallet et al., 2021). The identification of these receptors at the surface of breast cancer cells and neural crest cells will be the focus of future studies.

Although our results demonstrate that the RGD motif mediates breast cancer and neural crest cell adhesion, they also raise questions on the role of the LDV motif in SNED1. The LDV motif is primarily recognized by α4β1 and α4β7 integrins that are mainly expressed by leukocytes. Integrin α4 has been previously demonstrated to play a vital role in neural crest cell migration (Kil et al., 1998) and cancer cell adhesion to the vascular endothelium (Taichman et al., 1991). Here, we show that α4 is expressed by LM2 and O9-1 cells (Fig. S4A); however, functional blocking of integrin α4 did not affect the adhesion of LM2 cells to SNED1 (Fig. S4B), in line with our observation that these cells could adhere similarly to SNED1^LAV and SNED1^WT. This is also in line with the observation that the two truncated forms of SNED1, SNED1^1-530 and SNED1^1-751, which contain the LDV motif, are not sufficient to mediate cell adhesion. This opens the possibility that SNED1, via its LDV motif, could engage other cell populations in other pathophysiological processes that have yet to be discovered.

In addition to mediating cell adhesion, integrin heterodimers are involved in the early steps of ECM assembly; for example, α5β1 integrin is critical for the initiation of fibronectin fibrillogenesis (Mao and Schwarzbauer, 2005). We have previously shown that SNED1 forms fibers in the ECM (Vallet et al., 2021), yet we do not know the mechanisms required for this process. It will thus be interesting to determine whther α5β1 integrin also mediates SNED1 fiber assembly.

Finally, given the crucial role integrin–ECM interactions play in driving pathological processes, devising therapeutic strategies to prevent or disrupt these interactions is an active area of investigation (Hamidi et al., 2016; Pang et al., 2023; Raab-Westphal et al., 2017). To date, seven small-molecule inhibitors and biologics have been successfully marketed, including monoclonal antibodies specifically blocking α4β7 and α4β1 to treat inflammatory disorders such as ulcerative colitis, Crohn's disease and multiple sclerosis (Pang et al., 2023; Slack et al., 2022). Our study has thus the potential to pave the way to devise novel therapeutic strategies aimed at targeting SNED1–integrin interaction to prevent breast cancer metastasis.

The cDNA encoding full-length human SNED1 cloned into pCMV-XL5 was obtained from Origene (clone SC315884). 6×-His-tagged SNED1, as previously described (Vallet et al., 2021), was subcloned into the bicistronic retroviral vector pMSCV-IRES-Hygromycin (kindly gifted by the Hynes laboratory, MIT, USA) between the BglII and HpaI sites. FLAG-tagged SNED1 was previously described (Vallet et al., 2021).

Site-directed mutagenesis to generate SNED1^RGE (c.C120A; p.D40E) and SNED1^LAV (c.A932C; p.D311A) was performed using the QuikChange kit (Agilent #200519) following the manufacturer's instructions using the cDNA encoding SNED1 from Origene as a template. To obtain the double mutant SNED1^RGE/LAV (c.C120A; p.D40E/c.A932C; p.D311A), we introduced the c.C120A mutation in the c.A932C mutant. These constructs were then subcloned into the bicistronic retroviral vector pMSCV-IRES-Hygromycin between the BglII and HpaI sites, and a 6×-Hix-tag was introduced by PCR in 3′ (C-terminus of the protein).

Truncated forms of human SNED1 were subcloned by PCR to generate the following fragments: SNED1^1-530, encompassing the N-terminal region of SNED1 until the follistatin domain, and SNED1^1-751, encompassing the N-terminal region of SNED1 until the sushi domain (Fig. 2A). These constructs were then subcloned into the bicistronic retroviral vector pMSCV-IRES-Hygromycin between the BglII and HpaI sites, and a FLAG tag (DYKDDDDK) was added at the C-terminus via PCR as previously described (Vallet et al., 2021). FLAG-tagged SNED1^1-260, encompassing the very N-terminal region of SNED1 was previously described (Vallet et al., 2021).

The sequences of the primers used to introduce point mutations or generate truncated fragments of SNED1 are listed in Table S1. All constructs were verified by sequencing.

Highly metastatic breast cancer cells, MDA-MB-231 ‘LM2’ (termed LM2 in the manuscript), were kindly gifted by Dr Joan Massagué (Memorial Sloan Kettering Cancer Center, New York, NY). LM2 cells, human embryonic kidney 293T cells (termed 293T in the manuscript) stably overexpressing different constructs of SNED1, and immortalized mouse embryonic fibroblasts isolated from Sned1^KO mice (Sned1^KO iMEFs) (Vallet et al., 2021) were cultured in Dulbecco's Modified Eagle's medium (DMEM; Corning, #10-017-CV) supplemented with 10% fetal bovine serum (FBS; Sigma, #F0926) and 2 mM glutamine (Corning, #25-005-CI); this formulation is termed ‘complete medium’ in this manuscript. The O9-1 mouse cranial neural crest cell line (Millipore Sigma, #SCC049) was cultured as per the manufacturer's instructions. Briefly, cells were cultured on dishes coated with Matrigel^® (Corning, #356234) prepared at 0.18 mg/ml in 1× Dulbecco's phosphate buffered saline (D-PBS) containing Ca²⁺ and Mg²⁺ (Cytiva, #SH30264.FS) in presence of complete ES cell medium containing 15% FBS and leukemia inhibitory factor (Millipore Sigma, #ES-101-B) and supplemented with 25 ng/ml fibroblast growth factor-2 (FGF-2, R&D systems, #233-FB). All cell lines were maintained at 37°C in a 5% CO₂ humidified incubator.

293T cells were plated at ∼30% confluency in a 6-well plate. Cells were transfected the following day using a Lipofectamine 3000 (Invitrogen, #L3000-008) mixture containing 1 µg of a retroviral vector with the construct of interest and 0.5 µg each of a packaging vector (pCL-Gag/Pol) and a vector encoding the VSVG coat protein prepared in Opti-MEM™ (Gibco, #31985070). After 24 h, the transfection mixture was replaced with fresh complete medium, and cells were cultured for an additional 24 h, after which the conditioned medium containing viral particles was collected, passed through 0.45 µ filter, and then either immediately used for cell transduction or stored at −80°C for later use.

293T cells were seeded at ∼30% confluency. The following day, cells were transduced with undiluted viral particles-containing conditioned medium. 24 h after transduction, the medium was replaced with fresh complete medium, and cells were allowed to grow for another 24 h before selection with hygromycin (100 µg/ml). Once stable cell lines were established, we assessed the production and secretion of recombinant proteins using immunoblotting of total cell extract (TCE) and conditioned medium, respectively, using either an anti-SNED1, anti-His or anti-FLAG antibody (see Table S2 for details).

293T cells stably expressing constructs of interest were seeded in a 15 cm dish in complete medium and allowed to grow until reaching 100% confluency. The culture medium was aspirated, the monolayer was rinsed with 1× D-PBS containing Ca²⁺ and Mg²⁺, and the medium was replaced with serum-free DMEM medium supplemented with 2 mM glutamine for 48 h. The serum-free conditioned medium (CM) containing secreted SNED1 was harvested, and cells were allowed to recover for 48 h in complete medium before repeating the next cycle. Cells were discarded after five cycles. An EDTA-free protease inhibitor cocktail (0.067× final concentration; Thermo Fisher Scientific, #A32955) was added to the CM, and the CM was centrifuged at 2576 g for 10 min to remove any cell or cellular debris. Pre-cleared supernatants were collected and stored at −80°C until further processing.

6× His-tagged SNED1 proteins (wild-type or integrin-binding mutants) were purified via immobilized metal affinity chromatography (IMAC) using an AKTA Pure system for fast protein liquid chromatography (FPLC) at the UIC Biophysics Core Facility. In brief, conditioned medium (CM) containing the secreted protein of interest was thawed at 4°C, concentrated using a 100-kDa protein concentrator with a polyethersulfone membrane (Thermo fisher Scientific, #88537), buffer-exchanged against a binding buffer containing 20 mM Tris-HCl, 500 mM NaCl, 20 mM imidazole (pH 7.5), and filtered using a 0.2 μm filter. In parallel, a HisTrap HP column (column volume: 1 ml) was equilibrated with the binding buffer. The filtered CM was applied to the equilibrated HisTrap column to allow protein binding. The column was then washed with 20 column volumes of binding buffer and bound proteins were eluted with a buffer containing 20 mM Tris-HCl, 500 mM NaCl, 5 mM β-mercaptoethanol (pH 7.5), and a stepwise gradient of increasing imidazole concentration (12.5 mM, 50 mM, 125 mM, 250 mM, 375 mM and 500 mM). The elution fractions were concentrated, and a buffer exchange was performed with HEPES-buffered saline (HBS; 10 mM HEPES, 150 mM NaCl, pH 7.5).

Given that we could not achieve sufficient purity of His-tagged truncated forms of SNED1, we cloned FLAG-tagged versions of these proteins for purification (Fig. 2B). In brief, conditioned medium containing FLAG-tagged full-length or truncated forms of SNED1 was thawed at 4°C overnight and concentrated using protein concentrators with a 10 kDa molecular mass cut-off for the SNED1^1-260 and SNED1^1-530 fragments (Thermo Fisher Scientific, #88535), a 30 kDa molecular mass cut-off for the SNED1^1-751 fragment (Thermo Fisher Scientific, #88536), and a 100 kDa molecular mass cut-off for full-length SNED1 (Thermo Fisher Scientific, #88537). An anti-FLAG resin (Sigma, #A220), containing monoclonal M2 anti-FLAG antibodies coupled to agarose beads was washed with 20 column volumes of HBS twice. Concentrated CM was applied to the resin and incubated overnight under constant rotation at 4°C to allow protein binding. The following day, the unbound fraction was collected, and the resin was washed with 20 column volumes of HBS three times. Bound FLAG-tagged proteins were eluted by competing with 200 μg/ml of FLAG peptide in HBS in a stepwise manner, resulting in four elution fractions. Protein fractions were then pooled, and buffer exchanged with HBS using protein concentrators of appropriate molecular mass cut-off (see above) to remove the excess of unbound FLAG peptide.

Purified proteins were quantified by measuring the absorbance at λ=280 nm using a NanoDrop spectrophotometer. To assess their quality and purity, proteins were resolved by electrophoresis on polyacrylamide gels. Gels were stained overnight using Coomassie-based AquaStain (Bulldog, #AS001000) and imaged with a ChemiDoc MP™ imaging system (Bio-Rad).

Adhesion assays were performed as described previously (Humphries, 1998). In brief, wells of a 96-well plate were coated with 50 µl of different concentrations, ranging from 0.1 µg/ml to 10 µg/ml, of purified human SNED1 proteins (full-length, fragments or integrin-binding mutants) or murine Sned1 (R&D systems, 9335-SN) for 180 min at 37°C. Fibronectin (5 µg/ml; Millipore, #FC010) was used as a positive control. The adsorbed protein was immobilized with 0.5% (v/v) glutaraldehyde for 15 min at room temperature (RT). To prevent cell adhesion to plastic, wells were blocked using 200 µl of 10 mg/ml heat-denatured bovine albumin serum (BSA; Sigma, #A9576). Uncoated wells blocked with BSA (10 mg/ml) alone were used as a negative control.

Single-cell suspension of LM2 or O9-1 cells containing 5×10⁵ cells/ml were prepared in complete medium. 25,000 cells were added to each well and allowed to adhere for 30 min at 37°C. Loosely attached or non-adherent cells were removed by gently washing the wells with 100 µl of D-PBS⁺⁺ three times and fixed using 100 µl of 5% (v/v) glutaraldehyde for 30 min at RT. Cells were stained with 100 µl of 0.1% (w/v) Crystal Violet (Alfa Aesar, B21932) in 200 mM 2-(N-morpholino)ethanesulfonic acid (MES), pH 6.0 for 60 min at RT. After washing the excess of unbound Crystal Violet, the dye was solubilized in 10% (v/v) acetic acid. Absorbance values were measured at λ=570 nm using a Bio-Tek Synergy HT microplate reader. Cell adhesion was determined by interpolating the absorbance values from a standard curve that was generated by seeding cells at several dilutions (10–100%) from the single cell suspension on poly-L-lysine (0.01% w/v; Sigma, P4707) coated wells and fixed directly by adding 5% (v/v) glutaraldehyde.

Integrin-blocking experiments were performed by seeding cells on a SNED1 substrate (10 µg/ml) in presence of a cyclic RGDfV peptide (cRGDfV; Sigma, #SCP0111) at concentrations ranging between 1.25 µM and 10 µM or in presence of 10 µg/ml of anti-β1, anti-α5, anti-α4, or anti-αvβ3 integrin-blocking antibodies (see Table S2 for detailed description of all antibodies used in this study).

LM2 and O9-1 cells were cultured for 3 days in complete medium and lysed in 3× Laemmli buffer (0.1875 M Tris-HCl, 6% SDS and 30% glycerol) containing 100 mM dithiothreitol. Cell lysates were passed through a 26.5-gauge needle to ensure complete cell lysis, and samples were heated at 95°C for 10 min. Lysates were resolved by gel electrophoresis on polyacrylamide gels at constant current (20 mA for stacking, 25 mA for resolving). Proteins were transferred onto nitrocellulose membranes at constant voltage (100 V) for 180 min at 4°C. Membranes were incubated in 5% (w/v) non-fat milk prepared in 1× PBS+0.1% Tween-20 (PBST) for 60 min at room temperature to prevent non-specific antibody binding and then incubated in the presence of primary anti-β1 integrin, anti-α5 integrin, anti-αV integrin, anti-β3 integrin or anti-α4 integrin antibodies (see Table S2 for details) in 5% (w/v) non-fat milk in PBST overnight at 4°C. Membranes were washed and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 60 min at RT. Immunoreactive bands were detected using chemiluminescence (Thermo Scientific, #E32109) and imaged with a ChemiDoc MP™ imaging system (Bio-Rad). Images of uncropped blots are presented in Fig. S5.

All experiments were performed with at least three technical replicates (n) and, unless noted otherwise, repeated independently three times (N; biological replicates). Data is represented as a mean±s.d. from three biological replicates. Unpaired Student's two-tailed t-test with Welch's correction or Welch and Brown–Forsythe one-way ANOVA with Dunnett's T3 correction for multiple comparisons was performed to measure statistical significance. Plots were generated using PRISM (GraphPad).

We would like to thank Dr Hyun Lee, director of the Biophysics Core facility at UIC, and her team for their help with protein purification. We would also like to thank Dr Sandra Pinho for recommendations on anti-α4 integrin antibodies and all the members of the Naba lab for insightful discussions.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.263479.reviewer-comments.pdf