Integrin α11 cytoplasmic tail is required for FAK activation to initiate 3D cell invasion and ERK-mediated cell proliferation

Integrin α11β1 is a collagen-binding integrin, which is receiving increasing attention in the context of wound healing and fibrosis. Although α11β1 integrin displays similar collagen specificity to α2β1 integrin, both integrins have distinct in vivo functions. In this context, the contribution of α11 subunit cytoplasmic tail interactions to diverse molecular signals and biological functions is largely unknown. In the current study, we have deleted the α11 cytoplasmic tail and studied the effect of this deletion on α11 integrin function. Compared to wild-type cells, C2C12 cells expressing tail-less α11 attached normally to collagen I, but formed fewer focal contacts. α11-tail-less cells furthermore displayed a reduced capacity to invade and reorganize a 3D collagen matrix and to proliferate. Analysis of cell signaling showed that FAK and ERK phosphorylation was reduced in cells expressing tail-less α11. Inhibition of ERK and FAK activation decreased α11-mediated cell proliferation, whereas α11-mediated cell invasion was FAK-dependent and occurred independently of ERK signaling. In summary, our data demonstrate that the integrin α11 cytoplasmic tail plays a central role in α11 integrin-specific functions, including FAK-dependent ERK activation to promote cell proliferation.

integrin α11 cytoplasmic tail is required for FAK activation to initiate 3D cell invasion and ERKmediated cell proliferation Pugazendhi Erusappan 1,2 , Jahedul Alam 1 , Ning Lu 1 , Cédric Zeltz 1,3 & Donald Gullberg 1* integrin α11β1 is a collagen-binding integrin, which is receiving increasing attention in the context of wound healing and fibrosis. Although α11β1 integrin displays similar collagen specificity to α2β1 integrin, both integrins have distinct in vivo functions. In this context, the contribution of α11 subunit cytoplasmic tail interactions to diverse molecular signals and biological functions is largely unknown. In the current study, we have deleted the α11 cytoplasmic tail and studied the effect of this deletion on α11 integrin function. Compared to wild-type cells, C2C12 cells expressing tail-less α11 attached normally to collagen I, but formed fewer focal contacts. α11-tail-less cells furthermore displayed a reduced capacity to invade and reorganize a 3D collagen matrix and to proliferate. Analysis of cell signaling showed that FAK and ERK phosphorylation was reduced in cells expressing tail-less α11. Inhibition of ERK and FAK activation decreased α11-mediated cell proliferation, whereas α11mediated cell invasion was FAK-dependent and occurred independently of ERK signaling. In summary, our data demonstrate that the integrin α11 cytoplasmic tail plays a central role in α11 integrin-specific functions, including FAK-dependent ERK activation to promote cell proliferation.
Integrins are heterodimeric cell surface receptors composed of non-covalently associated α and β subunits, which act as cell surface links to the extracellular matrix (ECM) and to other cells in dynamic cell-cell linkages 1 . Integrin subunits are composed of different domains with different functions. The extracellular domain of collagen-binding α integrin chains contain an inserted α-I domain, which is responsible for collagen binding without direct involvement of the β subunit. Whereas different integrin β chains display conserved regions including their cytoplasmic tail, the cytoplasmic tails of integrin α chains show little sequence similarity except for the very proximal membrane sequence GFFXR 2 . It is interesting in this context to note that α11 integrin lacks the conserved GFFXR sequence, but instead the α11-tail contains the GFFRS sequence 3 . A number of proteins have been found to bind to the conserved GFFXR sequence without displaying specificity for any particular α chain 2 . The conserved GFFXR site has been demonstrated to bind, for example, SHARPIN 4 , which keeps integrin in an inactive conformation. Other proteins reported to bind to the conserved membrane proximal region include Rab21, Nischarin and PP2A 2 . Deletion of GFFXR or mutation of Arginine in the GFFXR sequence render integrins constitutively active, suggested to occur as a result of breakage of a salt linkage between α chains and β chain 5 . If the deletion occurs beyond the GFFXR sequence, effects vary depending on the nature of the α chain and the cellular background 6 .
Molecular interactions of the integrin cytoplasmic tails can both regulate inside-out and outside-in signaling as well as strengthening the actin linkages 2,7 . The NPXY motifs located in the β subunits are important binding sites for talins and kindlins, both taking part in integrin inside-out signaling 8,9 . These important interactions in turn are regulated through binding of other proteins such as Dok1 and ICAP-1, to the same integrin β chain NPXY motifs. In addition, phosphorylation of the proximal NPXY motif appears to be a molecular switch to regulate tensin binding and localization of α5β1 to fibrillar adhesions 10 . More recent data have demonstrated that integrin α chains contribute to filamin A-, talin-, and kindlin-binding to the integrin β subunits [11][12][13][14] . Data are thus accumulating with indications that integrin α cytoplasmic tails take an active part in interactions of importance for integrin heterodimer function.
Careful analyses of mice lacking individual collagen-binding integrins show that the collagen-binding integrin receptors are dispensable for normal development, but suggest important roles for these receptors in tissue remodeling events occurring in wound healing, fibrosis and tumor-stroma interactions 15 . α11β1 integrin is a collagen receptor, which is the latest identified member of the integrin family 3,16,17 . Although α11 shows an overall sequence homology to other collagen-binding integrin α chains of the β1 subfamily and also displays similar collagen specificity as α2β1 integrin, α11β1 integrin in vivo has functions distinct from the other collagen-binding integrins [18][19][20][21][22] . This suggests that α11 cytoplasmic tail may regulate α11 functions. The role of cytoplasmic tails of collagen-binding integrins has been studied extensively in the 1990s by the group of Hemler et al. Deletion of the α2 integrin cytoplasmic tail in K562 and RD cells (both requiring integrin activation) demonstrated that the α2 -tail-less integrin showed reduced adhesive activity on collagen I, in a manner suggesting an activation defect 23 .
In chimeric experiments where again the α2 cytoplasmic tail was replaced with the tail of other integrins, it was demonstrated that chimeric α2 integrins with α5-tail (Xα2 Cα5) could mediate collagen gel contraction, whereas chimeric Xα2 Cα4 failed to mediate contraction, but instead promoted cell migration 24 . Already at this time it was speculated that "α subunit cytoplasmic domains, probably acting in concert with their associated β subunit, also have important but distinct roles and perhaps eventually will be shown to interact with distinct set of intracellular proteins" 24 . More recent data, using more sensitive assays, analyzing chimeras of all 12 α chains of the β1 integrin subfamily confirm that cytoplasmic tails of integrin α-chains do affect integrin inside-out activation, but that this varies greatly between different integrin α chains 25 . Deletion experiments are thus not easy to interpret since one also has to consider possible modulatory effects on integrin α/β chain interactions.
In the current study, we have deleted the α11 cytoplasmic tail and studied the effect of this deletion on α11 integrin function. Our data show that the integrin α11 cytoplasmic tail is dispensable for cell attachment but is essential for focal adhesion formation, ERK-dependent cell proliferation, cell migration and reorganization of 3D collagen matrices.

Results
Generation and expression of a human integrin α11-tail-less variant. In order to identify the role of the α11 cytoplasmic tail, a mutant variant ( Hu α11-1171) with a deletion of the terminal 17 amino acids in the cytoplasmic tail of human integrin α11 ( Hu α11) was generated. Since the antibodies used to detect human α11 react with the cytoplasmic tail 26 , we have EGFP-tagged the integrin carboxy terminus using a 10 amino acid linker to avoid interference of the EGFP tag. Horwitz et al. pioneered this strategy for integrin α5 and the resultant tagged α5 integrin was characterized in detail without any evidence of artifacts due to the EGFP tag 27 . Full-length ( Hu α11-WT) and tail-less ( Hu α11-1171) Hu α11 variants were tagged with enhanced green fluorescence protein (EGFP) and expressed in C2C12 mouse myoblasts, which do not express any collagen-binding integrins 28 (Fig. 1A). Based on EGFP intensity, the transfected cells were sorted by flow cytometry with uniform gating to obtain similar expression levels of EGFP. The expression of comparable levels of Hu α11-EGFP in the total protein lysates was confirmed by immunoblotting, either with an anti-Hu α11 polyclonal antibody or with an anti-EGFP antibody (Fig. 1B). Comparable expression levels of Hu α11-EGFP at the cell surface were also confirmed, using mock transfected cells (Mock, empty GFP vector) as a negative control (Fig. 1C).
integrin α11 cytoplasmic tail is dispensable for cell adhesion but mediates focal adhesion formation, collagen reorganization, cell migration and cell proliferation. To examine the role of integrin α11 cytoplasmic tail in cell adhesion, Hu α11-WT and Hu α11-1171 cells were allowed to attach to collagen I or fibronectin using BSA coated wells as negative control. Mock transfected cells did not adhere to collagen, whereas Hu α11-WT and Hu α11-1171 cells adhered equally well to collagen I and fibronectin. This suggests that the deletion of 17 amino acids in the cytoplasmic tail of Hu α11 had no apparent negative influence on integrin activation or cell adhesion to collagen I ( Fig. 2A). However, Hu α11-1171 cells displayed fewer focal adhesions after 2 hours, with a reduction of 50% in the total area of focal adhesions (Fig. 2C) and 35% reduction in cell spreading ( Supplementary Fig. 2). This indicates that cytoplasmic tail of integrin α11 is involved in mediating cell adhesion signaling. To confirm this hypothesis, we assessed the ability of Hu α11-1171 cells in mediating: collagen gel reorganization, cell migration and proliferation.
When cells were allowed to attach on collagen I for 24 hours in low serum conditions, Hu α11-1171 cells displayed a significant reduction in cell proliferation compared to Hu α11-WT cells, but not on fibronectin, suggesting that the α11 cytoplasmic tail-mediated signaling is involved in the regulation of cell proliferation (Fig. 2B). We further examined the ability of these cells to contract 3D collagen I lattices, a process previously shown to be α2β1and α11β1mediated 18,28,29 . Sixteen hours after the contraction was initiated, the Hu α11-1171 cells displayed 50% reduction in collagen contraction compared to Hu α11-WT cells (Fig. 2D).
We also investigated the role of the α11 cytoplasmic tail in cell migration using a spheroid migration model in a 3D collagen matrix. The radial cell density profile of the spheroid was analyzed from the center of the spheroid to quantify migrated cells in relation to their distance of migration. Hu α11-WT cells migrated out 50% more than Hu α11-1171 and mock transfected cells, beyond 400 µm from the center of the spheroid (Fig. 2E). Interestingly, the size of the spheroid core for the Hu α11-WT cells was smaller than that observed for spheroids formed from the Hu α11-1171 and Mock cells. These results indicate that the α11 cytoplasmic tail is indeed essential to mediate cell proliferation, collagen reorganization and cell migration.
integrin α11 cytoplasmic tail contributes to FAK and ERK activation. Localization of focal adhesion kinase (FAK) in focal adhesions and its autophosphorylation at Tyrosine residue 397 (FAK Y397 ) is a primary event in integrin signaling leading to cell migration and proliferation [30][31][32][33] . Analysis of FAK Y397 activation at different time points revealed that Hu α11-1171 cells demonstrated less FAK Y397 activation than Hu α11-WT cells at 30 and 45 minutes on collagen I (Fig. 3A). Since other collagen-binding integrins have previously been shown to activate MAPK pathways, we investigated activation of ERK, p38 and JNK following attachment of cells to collagen I [34][35][36] . Interestingly, Hu α11-1171 cells showed a strong reduction in ERK activation compared to Hu α11-WT cells on collagen I (Fig. 3A). In contrast, activation of p38 and JNK was not affected by the deletion of the α11 cytoplasmic tail. To confirm the role of integrin α11 in FAK and ERK activation in primary cells, human gingival fibroblasts (hGF) were transfected with control siRNA (Ctrl) or α11 siRNAs ( Fig. 3B and supplementary Fig. 3A). The knockdown of α11 did not affect expression of the collagen-binding integrins α1 and α2 chains ( Supplementary Fig. 3B,C). Phosphorylated FAK Y397 and ERK levels were only reduced in α11 siRNA-treated hGF ( Fig. 3B and supplementary Fig. 3A). These results demonstrate that cytoplasmic tail of integrin α11 contributes to FAK and ERK activation. FAK Y397 activation is involved in integrin α11 cytoplasmic tail-mediated cell proliferation and migration. We next examined if the reduced cell functions observed with the α11-tail-less cells was due to the defect in FAK and ERK activation. Inhibition of FAK Y397 activation by PF573228 inhibited cell proliferation in both Hu α11-WT and Hu α11-1171 cells, on collagen I. Similarly, inhibition of ERK activation by U0126 also inhibited α11β1-mediated cell proliferation on collagen I, implying that both FAK Y397 and ERK activation are required for α11β1-mediated cell proliferation (Fig. 4A). Inhibition of both FAK and ERK activation weakly decreased cell proliferation in Hu α11-WT cells on fibronectin, suggesting that cell proliferation mediated by fibronectin-binding integrins is less dependent on these signaling molecules than α11β1-mediated cell proliferation. In order to understand the overlapping function of FAK and ERK in α11β1-mediated cell proliferation, we investigated the relationship between FAK and ERK activation. Inhibition of FAK Y397 activation by PF573228 significantly inhibited ERK activation 30 minutes after cells attached to collagen (Fig. 4B). In contrast, inhibition of ERK activation by U0126 did not affect FAK Y397 activation, indicating that FAK Y397 phosphorylation is required for ERK activation.
In spheroid migration assays, inhibition of FAK Y397 phosphorylation, completely abrogated migration for both Hu α11-WT and Hu α11-1171 cells (Fig. 5A). Contrariwise, inhibition of ERK in spheroid assay did not inhibit  www.nature.com/scientificreports www.nature.com/scientificreports/ migration beyond 400 µm of both Hu α11-WT and Hu α11-1171 cells, suggesting that in contrast to cell proliferation, ERK activation is not involved in integrin α11-mediated cell migration in this 3D model. Furthermore, neither inhibition of FAK Y397 activation nor ERK activation inhibited serum-stimulated collagen gel contraction (data not shown).
Since FAK is localized in focal adhesions, we investigated the phosphorylation of FAK Y397 in focal adhesions. Similar to our previous results with immunoblotting, we observed significantly reduced amounts of phosphorylated FAK Y397 positive adhesions for Hu α11-1171 cells as compared to Hu α11-WT cells (Fig. 5B). These findings suggest that the reduced activation of FAK Y397 in Hu α11-1171 cells is due to fewer focal adhesions in these cells per se.
To summarize, our results strongly suggest that the cytoplasmic tail of integrin α11 is essential to stabilize focal adhesions that in turn mediate FAK Y397 activation involved in cell proliferation and cell migration.

Discussion
Although integrin cytoplasmic tails are relatively short, they are able to exert control of integrin activity and mediate a number of interactions of importance for integrin function 7,8 . There are in total more than 20 integrins and various research groups have independently deleted cytoplasmic tails and expressed the mutated variants with different results 6,37-39 . When integrin β-tails are expressed as chimeric proteins with non-integrin www.nature.com/scientificreports www.nature.com/scientificreports/ transmembrane-and extracellular parts, the β1 integrin tail sequence can direct chimeric proteins into focal adhesions 40 . When integrin α-tails are deleted, some α-tail-less heterodimers localize to focal adhesions in a ligand-independent manner [41][42][43] and often become constitutively activated.
In the current study, we have used EGFP-tagged α11 integrins, in a strategy where the tag is separated by a 10 amino acid linker. Like for EGFP tagged α4 and α5 integrins we find no evidence that addition of the tag interferes with integrin function 27,44 . It could be argued that addition of a tag on tail-less integrin would be more likely to interfere with integrin function than a tag added to full-length integrin. However, independent experiments suggest that the loss of α11 integrin activity seen in the tail-less variant can be duplicated by a specific point mutation in the α11-tail, excluding that non-specific interference of the EGFP-tag (data not shown).
In our present study, we found that deletion of α11 cytoplasmic tail led to reduced focal adhesion formation, reduced cell spreading, reduced collagen gel contraction, reduced cell migration in a 3D context and reduced ERK-dependent cell proliferation, but that the α11-tail was dispensable for cell adhesion under the conditions used. Deletion of α1 cytoplasmic tail in fibroblastic 3T3 cells leaves cell adhesion to collagen IV unaffected but results in ligand-independent focal contact localization 45 , whereas in endothelial cells α1-tail deletion leads to reduced adhesion to collagen IV 36 . Deletion of α2 cytoplasmic tail reduces cell adhesion to collagen I in RD rhabdomyosarcoma cells, but this effect can be overcome by ions or ligand concentration, suggesting that a major function of the cytoplasmic α2-tail is to mediate intracellular inside-out activation events, and that α2 deletion effects on cell attachment can be overcome by activating the extracellular domain 23 . The reduced number of focal contacts seen in α11-tail-less expressing cells probably reflects a contribution of α11-tail to a cytoskeletal linkage, Attached cells were fixed, stained with 0.1% crystal violet and absorbance was read at 595 nm. Results were expressed as mean ± standard deviation of at least three replicates pooled from three independent experiments. (B) Integrin α11 tail-mediated ERK activation is dependent on FAK Y397 activation. Serum starved Hu α11-WT and Hu α11-1171 cells were treated with DMSO or U0126 or PF573228 and allowed to attach on collagen I for 30 minutes. After 30 minutes, cells were lysed, and the lysates were analyzed for total and phosphorylated levels of FAK Y397 and ERK by Western blotting. Protein bands were quantified by densitometry analysis and data shown are pooled from at least three independent experiments (Full size immunoblots are shown in supplementary). Statistical significance was assessed by two tailed, unpaired t-tests and P-values are expressed as ***P < 0.001; **P < 0.01 and *P < 0.05. (2019) 9:15283 | https://doi.org/10.1038/s41598-019-51689-6 www.nature.com/scientificreports www.nature.com/scientificreports/ whose nature remains to be established, rather than need for α11-tail in integrin activation. In vivo integrin α2β1 is expressed in platelets and hematopoietic cells 46 where integrin activation is essential, whereas α11β1 is mainly expressed on fibroblastic cells 26 where β1 integrins are constitutively activated 47 .
Here we showed that interaction of α11β1 with collagen I mediated ERK signaling. This signaling is thus similar to that observed for α1 (although the preferred ligand for α1β1 is collagen IV 48 ), but is different than for α2β1-mediated signaling, which occurs mainly via p38 in 3D collagen I matrix 34 . Interestingly, in mouse endothelial cells, limited α2-dependent p38 signaling is observed 36 . These data suggest for collagen-binding integrins that the presence of cell-dependent factors influence which MAPK signaling pathway will be activated upon collagen ligation. siRNA knockdown of α11 reduced FAK and ERK activation, supporting that α11-mediated ERK to attach on collagen I for 120 minutes. Cells were fixed with 4% PFA and stained for phospho-FAK Y397 . Focal adhesions positive for phospho-FAK Y397 were quantified using ImageJ. Scale bar: 10 µm. Results were expressed as mean ± standard deviation of at least three replicates from one representative experiment of at least three independent experiments. Statistical significance was assessed by two tailed, unpaired t-tests and P-values are expressed as ***P < 0.001; **P < 0.01 and *P < 0.05.
signaling is central in fibroblasts, which is the major cell type expressing α11. Previous studies have demonstrated α11-dependent ERK and PI3K phosphorylation in mesenchymal stem cells expressing multiple collagen-binding integrins 49 . However, in our cell system (C2C12 cells lacking other collagen receptors than the overexpressed α11β1), we failed to detect α11-dependent PI3K activation (data not shown). Blocking α11-dependent cellular signaling in C2C12 and human gingival fibroblasts cells also blocked ERK-dependent cell proliferation. A majority of the α11-dependent ERK signaling appeared to be dependent on FAK, since FAK inhibition also attenuated the α11-dependent ERK signaling. In the case of α1, FAK independent ERK signaling via Shc has been noted 50 . Later studies have demonstrated that FAK may enhance and prolong integrin-mediated activation of ERK through p130 (CAS), Crk, and Rap1 in cells expressing B-Raf 51 . α2-mediated p38 activation has been suggested to depend on specific residues within the α2 integrin subunit cytoplasmic domain 52 , and independent experiments failed to record FAK activation in 3D collagen gel under conditions of α2-mediated p38 activation 34 .
To analyze cell migration in 3D collagen gel we used a spheroid assay. Cell migration 53 , MMP-induction 54 and collagen gel remodeling 55 has been shown to depend on ERK signaling in some conditions. In our study, ERK inhibition did not impair cell migration in a collagen matrix. ERK inhibition could attenuate G-protein dependent integrin inhibition as has been reported for α2β1 integrin-dependent cell migration in smooth muscle cells 56 .
Finally, the collagen gel contraction was not affected by ERK or FAK inhibition suggesting that an alternative signaling pathway is operative in the C2C12 cells overexpressing α11. We have previously demonstrated that TGF-β-dependent contraction of floating collagen lattices by dermal fibroblasts depends on α11and JNK-signaling 19 . This signaling pathway might be restricted to dermal fibroblasts or depend on relative levels of crucial components in non-canonical TGF-β signaling pathway being present in the cells. Previous studies have demonstrated that thrombospondin 1 in scleroderma fibroblasts can activate TGF-β to stimulate ERK-dependent collagen contraction 57 . Since αvβ3 signals via ERK, it is possible that αvβ3 mediates this collagen gel contraction under these conditions 58 . ERK activation has been shown to stimulate phosphorylation of MLC and in this way contribute to collagen lattice contraction 53 , but in our experiments pharmacological inhibition of ERK in α11-C2C12 cells failed to inhibit contraction.
In summary, our data suggest that the unique functions of α11 that separates it from other collagen-binding integrins is in part due to its cytoplasmic tail, which is needed for efficient focal contact formation, cell spreading, cell proliferation, cell migration and collagen remodeling.

Materials and Methods
Cell culture. Mouse C2C12 mouse satellite cells were provided by Prof. Anna Starzinski-Powitz (Goethe-Universität, Frankfurt am Main, Germany) and Phoenix 293 cells were provided by Prof. James Lorens, University of Bergen. Primary human gingival fibroblasts (hGF) were isolated from healthy gingival tissues as described earlier 59 . MRC5 human lung fibroblasts (American Type Culture Collection) were obtained from Robert Lafyatis laboratory (University of Pittsburgh Medical Center, Pittsburgh, PA, USA). Cells were cultured at 37 °C in Dulbecco's modified Eagle's medium (DMEM; Gibco ® , Invitrogen) with 10% fetal bovine serum (FBS; Gibco ® , Invitrogen), 1% penicillin-streptomycin (PEST; Sigma-Aldrich) and 5 µg/ml plasmocin (InvivoGen). Human gingival fibroblasts were grown from biopsies obtained during oral surgery after obtaining informed consent and in accordance with guidelines and regulations at the Department of Prosthetic Dentistry, Karolinska Institute, Stockholm in the 1990s following approval of experimental protocols by local ethics committee at faculty of Odontology, Karolinska institute and were kindly provided by Prof. Kamal Mustafa (University of Bergen) 60 .

Generation and expression of integrin α11 variants in C2C12 mouse satellite cells.
To construct pBABE ITGA11 retroviral expression constructs, pBABE-puro-Itga11 plasmid, pBJ1-Hu α11-WT-EGFP and pBJ1-Hu α11-1171-EGFP (for detail, see Supplementary information) were used as templates. The ITGA11-EGFP cDNAs from pBJ1-Hu α11-WT-EGFP and pBJ1-Hu α11-1171-EGFP were excised with XhoI (blunted) and EcoRI and subcloned into pBABE-puro-Itga11 plasmid at BamHI (blunted) and EcoRI sites. The constructs were transfected into Phoenix 293 packaging cell line with X-tremeGENE 9 transfection reagent (Roche Diagnostics GmbH), according to manufacturer's instructions. The viral supernatant medium was collected after 48 hours post transfection. C2C12 cells, cultured on 6-well plates were infected with viral supernatants containing polybrene at 5 µg/ml by spinfection at 1200 g for 90 minutes. After 36 hours, the culture medium was changed to a selection medium containing 2 µg/ml of puromycin. In addition, the cell populations with similar levels of EGFP intensity were sorted by Fluorescence-activated cell sorting (FACS).

Estimation of cell surface protein expression by FACS. FACS was performed at The Molecular
Imaging Centre (MIC), University of Bergen. Cells were detached with Trypsin-EDTA (0.05% Trypsin and 0.02% EDTA; Gibco ® , Invitrogen) and neutralized with DMEM containing 10% FBS. The cell suspension was filtered with a 40 µm syringe filter and 2 × 10 6 cells were used for the analysis. Cells were washed twice with PBS for 5 minutes at 210 × g with phosphate buffered saline (PBS) and the cell pellet was fixed with 2% paraformaldehyde/ PBS for 10 minutes. After fixation, cells were washed three times with PBS and blocked with 2% Bovine serum albumin (BSA/PBS) for 30 minutes in room temperature (RT). Cells were incubated with mouse anti-Hu α11 IgG (mAb 203E3) 61  www.nature.com/scientificreports www.nature.com/scientificreports/ Cell adhesion assay. Forty-eight-well plates were coated with fibronectin (1 μg/cm 2 : Sigma-Aldrich) or collagen type I (5 μg/cm 2 : Bovine PureCol ® , Advanced BioMatrix) and incubated for 2 hours at 37 °C. After washing with PBS, the plates were blocked with 2% BSA for 1 hour at 37 °C. Cells were washed three times with DMEM and 1 × 10 5 cells/well were cultured for 50 minutes at 37 °C. Unattached cells were removed carefully by washing three times with PBS containing Ca 2+ and Mg 2+ . Cells were then fixed with 96% ethanol for 10 minutes at room temperature followed by staining with 0.1% crystal violet for 20 minutes at room temperature. Plates were washed three times with distilled water and the cells were lysed with 1% Triton X-100 for 5 minutes. The lysates were transferred to a 96-well plate and absorbance was read at 595 nm.
Cell proliferation assay. Cells were seeded on 24-well plates coated with fibronectin (1 μg/cm 2 ) or collagen type I (5 μg/cm 2 ) and cultured for 24 hours in DMEM containing 1% FBS. Cells were washed with PBS and fixed with 96% ethanol for 10 minutes at room temperature followed by staining with 0.1% crystal violet for 20 minutes at room temperature. After washing three times with distilled water, the cells were lysed with 1% Triton X-100 for 5 minutes and lysates were transferred to a 96-well plate. The absorbance was read at 595 nm. For inhibition experiments, cells were incubated with U0126 (20 µM; Sigma-Aldrich) or PF573228 (10 µM; Sigma-Aldrich). Data were normalized considering proliferation of wild-type cells as 100% on collagen and fibronectin.
Collagen gel contraction. Collagen gel contraction was performed as described earlier 28,59 . In brief, 24-well plates were blocked with 2% BSA, overnight at 37 °C and then washed three times with PBS. Immunocytofluorescence. Glass bottom dishes (3.5 mm, MatTek) were coated with collagen type I (50 μg/ cm 2 ) for 60 minutes at 37 °C. Dishes were washed three times with PBS and 2 × 10 5 cells were cultured for 2 hours in DMEM with 10% FBS. Cells were then fixed with 4% paraformaldehyde/PBS for 10 minutes at RT and followed by washing three times with PBS for 5 minutes per wash. Focal adhesion images were captured using Nikon Total Internal Reflection Microscope (TIRFM).
Spheroid preparation and migration assay in 3D collagen gel. Homospheroids were made with C2C12 cells using hanging drop method as described earlier 62 . In short, C2C12 cells were harvested and suspended in culture medium to have a final concentration of 1 × 10 6 cells/ml. Approximately, 28 drops of cell suspension (25 µl/drop; 2.5 × 10 4 cells) were made on the lid of a 10 cm Petri dish, containing cell culture medium. The lid was carefully inverted over the Petri dish bottom, without disturbing the drop form. The spheroids were cultured for 3 days under regular cell culture conditions. Collagen I solution was prepared as described in collagen gel contraction assay and 100 µl of collagen I solution was added onto a 96-well plate and incubated for 15 minutes at 37 °C. One spheroid was embedded per well and the spheroid-collagen gel was allowed to polymerize for 90 minutes at 37 °C. After polymerization, 100 µl DMEM was added to each well and cultured for 24 hours. Spheroids were visualized under an inverted light microscope (Leica DMIL) and images were captured. When indicated, DMSO or 20 µM U0126 or 10 µM PF573228 were added to the medium.
integrin α11 silencing with siRNA. Primary hGFs were harvested and 5 × 10 5 cells were plated on 10 cm culture dishes 30 min prior to transfection. Cells were transfected with SMARTpool ON-TARGET plus ITGA11 siRNA (L-008000-00-0005, Dharmacon) or Individual ON-TARGET plus ITGA11 (J-008000-10, Dharmacon) or ON-TARGET plus Non-Targeting siRNA (D-001810-02-05, Dharmacon) at a final concentration of 20 nM with HiPerfect transfection reagent. After 48 hours, cells were serum starved for overnight. Cells were prepared as described for signaling experiments. Six-well plates were coated with thin film fibrillar collagen I gel prepared using the collagen I solution described in collagen gel contraction section and allowed to polymerize for 60 min at 37 °C. Cells were harvested and 5 × 10 5 cells were plated on to each 6-well and lysed as described above for SDS-PAGE and Western blotting analysis. Western blots stained with anti-phospho p44/42 MAPK IgG were reprobed with a custom-made mouse monoclonal anti-Hu α11 IgG, mAb 210F4 ( Supplementary Fig. 1) to confirm the silencing of ITGA11. In addition, membranes were analyzed with rabbit monoclonal anti-human α2 (EPR 5788, Abcam), mouse monoclonal anti-human α1 antibody (MAB 5676, R&D Systems), mouse GAPDH antibody 6C5 (sc-32233, Santa cruz biotechnology) and mouse anti-β-actin IgG (AC-74, Sigma-Aldrich) to confirm the unchanged levels of integrin α1 or α2 protein. MRC5 protein lysates were used as positive controls. Image analysis. Focal adhesions were quantified as described previously 63 . Briefly, raw images were subjected to background correction with a rolling ball radius of 50 using ImageJ. Image contrast was enhanced using ImageJ plugin CLAHE and threshold adjusted. The number of focal adhesions, area of individual focal adhesion and total area of focal adhesions were quantified for a single cell. Cell spreading was quantified by measuring the cell surface area, which was in turn calculated by drawing the cell boundary using ImageJ. Spheroid images were also subjected to background correction and contrast enhancement using CLAHE plugin in ImageJ. Threshold adjusted images were used to quantify the radial cell density profile from the center of the spheroid, using the Radial Profile plugin from ImageJ. The cell densities of migrated cells at different distant points were used to quantify the distance of migration from the center of the spheroid as described earlier 64 . Statistical analysis. Statistical significance was assessed by using two tailed, unpaired t-tests as indicated in the figure legends and P < 0.05 considered statistically significant. Statistical analysis and all graphs were done using GraphPad Prism 5 software (GraphPad Inc, USA). Data normalization was done based on maximum value in each experiment and data from three independent experiments was pooled together, and average ± standard deviation was calculated for each cell type and condition. For cell adhesion assay, data normalization was done based on the average absorbance values of the fibronectin-coated wells in each experiment.

Data availability
No datasets were generated or analyzed during the current study. Published: xx xx xxxx