Mechanosensing of Integrin α6 and EGFR Converges at Calpain 2

Cells sense and respond to mechanical cues from the extracellular matrix (ECM) via integrins. ECM stiffness is known to enhance integrin clustering and response to epidermal growth factor (EGF), but we lack information on when or if these mechanosensitive receptors converge intracellularly. Towards closing this knowledge gap, we combined a biomaterial platform with transcriptomics, molecular biology, and functional assays to link integrin-mediated mechanosensing and epidermal growth factor receptor (EGFR) signaling. We found that high integrin α6 expression controlled cell adhesion and motility on soft, laminin-coated substrates, and this mimicked the response of cells to EGF stimulation. Signaling pathways downstream of mechanosensitive cell adhesion and motility converged on calpain 2, an intracellular protease important for talin cleavage and focal adhesion turnover. Inhibiting calpain 2 shifted the biphasic dependence of cell migration on substrate stiffness. EGF stimulation enhanced both adhesion and motility on soft substrates, but required the presence of both integrin α6 and calpain 2. In sum, we identified a new role for integrin α6 mechanosensing, where high integrin α6 binding to laminin mimicked the effect of EGF stimulation. Downstream of both integrin α6 and EGFR, calpain 2 is known to control focal adhesion dynamics and motility, implicating integrin α6 and calpain 2 as potential targets to inhibit the migration of cancer cells in stiff tumor environments.


Introduction
Carcinoma progression is associated with deposition of ECM that stiffens the local microenvironment [1,2]. This tissue stiffening results in deposition of additional matrix protein, initiating a positive feedback loop between cells and the evolving stroma [3]. Cells sense and respond to the stiffness of their environment via RhoA GTPase activation, which feeds back to increase cell contractility via activation of myosin light chain kinase [4]. These ECM-driven changes in cytoskeletal tension also regulate motility in a cell-type specific manner [5]. In vitro, synthetic biomaterials have revealed that the type of material, stiffness, and biochemical surface modification alter the attachment and motility of cells [6,7]. We therefore hypothesized that independently tuning stiffness, integrin-binding sites, and chemokinetic factors could reveal new mechanosensitive proteins that drive cell adhesion and motility in different environments.
Several mechanically responsive genes and proteins have been implicated in breast cancer metastasis [8].
For example, extracellular mechanical forces in the tumor microenvironment alter nuclear stiffness and gene expression [9] and activate adhesion proteins, including focal adhesion kinase (FAK) and talin [10][11][12], leading to increased motility. One class of surface receptors, integrins, translate extracellular forces to downstream signaling cascades in a process called mechanotransduction. Increasing substrate stiffness increases integrin binding and clustering, which has implications for several pathways in breast cancer metastasis, including FAK and PI3K signaling [13,14]. However, most cancer mechanobiology research has focused on collagen-and fibronectin-binding integrins. Integrins that bind to other ECM proteins, such as laminin, have been neglected, despite the prevalence of laminin in the tumor ECM [15,16].
EGFR has recently emerged as mechanosensitive [17,18]. The mechanisms of EGFR mechanosensing are unknown, but understanding this process is critically relevant to cancer progression, given the known abundance of EGF in tumors [19] and frequent acquired resistance to EGFR inhibitors [20]. Residues on EGFR can be phosphorylated by α V β 3 integrin clustering [21], so we postulated that there could be a role for integrins in facilitating EGFR mechanosensing [22].
To uncover whether such a relationship exists, we created a biomaterial system to identify mechanoresponsive proteins working cooperatively downstream of EGFR phosphorylation and integrin engagement. Through a combination of transcriptomics, molecular biology, and quantification of cell adhesion and motility, we found that the intracellular protease calpain 2 is likely one of these links. Both calpain 2 and integrin α 6 expression had an inverse relationship with ECM stiffness. Engagement of integrin α 6 adhesion mimicked the effect of EGF stimulation in a stiffness-dependent manner. Integrin α 6 can directly associate with EGFR, which increased cell motility via calpain 2. In sum, this study highlights the utility of tunable biomaterials systems to uncover previously unrealized relationships between different classes of proteins in mechanobiology.

Materials and Methods
Cell Culture. All cell culture reagents were purchased from Thermo Fisher Scientific ( seconds. 20wt% Irgacure 2959 (BASF, Florham Park, NJ) was dissolved in 70% ethanol and added to the polymer solution in a ratio of 40 µL Irgacure to 1 mL of polymer solution. 50 µL of this final hydrogel precursor solution was sandwiched between a 3-(trimethoxysilyl)propyl-methacrylate treated 15mm diameter coverslip and an untreated coverslip and UV polymerized for 20 minutes. Gels were swelled in PBS for at least 48 hours before functionalizing the surface using 0.3 mg/mL Sulfo-SANPAH (c1111-100mg, ProteoChem, Hurricane, UT) in 50 mM HEPES buffer at pH 8.5, under UV for 10 minutes. Gels were then flipped onto droplets to achieve a final concentration of either 10 µg cm -² collagen 1 (A1048301, Thermo Fisher Scientific), 10 µg cm -² collagen 1 + 0.5 µg cm -² β 1 -chain-containing laminin isoforms (AG56P, Millipore, Billerica, MA), or 10 µg cm -² collagen 1 + 20 ng cm -² EGF (R&D Systems, Minneapolis, MN). Gels were incubated with protein overnight in a hydrated chamber at room temperature, washed in PBS, and sterilized under germicidal UV for 1 hour before cell seeding. For mechanical testing, gels were formed in a cylindrical mold (5µm high, 5µm diameter), allowed to swell for at least 48 hours and compression rheology was performed using an AR2000 (TA Instruments, New Castle, DE), as previously described [26].
RNA-Seq. Total RNA was isolated using Gen Elute mammalian total RNA Miniprep kit (RTN70, Sigma Aldrich). The TruSeq stranded RNA LT kit (15032612, Illumina, San Diego, CA) was used to purify and fragment the mRNA, convert it to cDNA, and barcode and amplify the strands. Quality and length of the inserts was confirmed with an Agilent Genomics 2100 bioanalyzer, followed by single-end 75 bp reads on the MiSeq (Illumina) to generate a complete transcriptome from each sample. Transcripts were aligned to the hg19 human reference genome using the Tuxedo Suite pathway [27][28][29][30]. Cufflinks was used to determine statistically significant differential expression of genes (p<0.05) [31][32][33].
qRT-PCR. Total RNA was isolated from 2 biological replicates as previously described [34]. 50 ng cDNA was then amplified using 10 pmol specific primers (Table S1) and the Maxima Sybr green master mix (Thermo Fisher Scientific) on a Rotor-Gene Q thermocycler (Qiagen, Valencia, CA) as follows: 50°C for 2 minutes, 95°C for 10 minutes followed by 45 cycles at 95°C for 10 seconds, 58°C for 30 seconds, and 72°C for 30 seconds. Both β-actin and ribosomal protein S13 were included as reference genes to permit gene expression analysis using the 2 -ddCt method.
Cell Adhesion Quantification. MDA-MB-231 cells were seeded at a final density of 5,700 cells cm -2 . Cells were either seeded immediately, or pretreated for 30 minutes before seeding with a small molecule ERK inhibitor FR180204 (Sigma Aldrich), calpain inhibitor IV (208724, Millipore), or 100 ng mL -1 EGF. The plate was pre-incubated on the Zeiss Axio Observer Z1 microscope (Carl Zeiss AG, Oberkochen, Germany) for 1 hour, then cells were seeded onto gels functionalized with collagen 1, collagen 1 + laminin, or collagen 1 + 20ng cm -2 bound EGF and imaging was started within 10 minutes of seeding. Images were taken with a 20x objective every 5 minutes for at least 2 hours. Cells were manually traced using Image J (NIH, Bethesda, MD).

N≥ 50 cells per condition.
Cell Migration. Cells were seeded onto gels functionalized with 10 µg cm -² collagen 1 or 10 µg cm -² collagen 1 + 20ng cm -² EGF at a density of 5,700 cells cm -2 and allowed to adhere for 24 hours. For inhibitor conditions, cells were treated 2 hours prior to the start of imaging with calpain inhibitor IV, ERK inhibitor days, and the resulting matrix created by the cells was visualized by adapting a previously published protocol [35]. Briefly, cells were washed with warm PBS, then lysed with 500µL warm extraction buffer (PBS with 0.5% Triton-X with 20 mM ammonium hydroxide) for 10 minutes at room temperature. 1 mL of PBS was added and the gels were stored overnight at 4°C before staining as described above.
Statistical Analysis. One-way Analysis of Variance (ANOVA) with a Tukey post-test was performed on paired samples using Prism v5.04 (Graphpad software, La Jolla, CA). Data reported is the mean and reported error is standard deviation with significance values of p ≤ 0.05 as *, p ≤ 0.01 as **, and p ≤ 0.001 as ***.

Cells Maximize Expression of Integrin α 6 and Extracellular Laminin on Soft Substrates
To identify ECM and integrin-associated genes involved in breast cancer mechanosensing, we performed whole transcriptome sequencing (RNA-Seq) on MDA-MB-231 cells cultured on collagen 1-functionalized PEG-PC hydrogels ranging in stiffness from 1 to 41 kPa for either 24 hours or 6 days (Fig. S1). The stiffness range chosen spans that of several tissues in the body where breast cancer metastasizes (1 kPa-brain [36], 4 kPa-bone marrow [37], 8 kPa-lung [38]), as well as a supra-physiological stiffness (41 kPa) that approximates extremely stiff breast tumors [39] and a tissue culture polystyrene (TCPS) control (Fig. S1).
Because the ECM is known to regulate the ability of cancer cells to migrate and metastasize, we focused our in-depth analysis of this RNA-Seq data set on genes that regulate interaction with the ECM. We found that the expression of integrins, other focal adhesion complex genes, and genes downstream of EGFR phosphorylation were sensitive to stiffness (Fig. S1, Fig. 1a-b). Short-term (24 hour) decreases in expression were clear in integrin α 6, calpain 2, and integrin β 4 (Fig. 1a). Integrin β 4 expression was 3-fold higher on gels by day 6 (Fig. 1b). This likely induces a switch in dimer pairs, as integrin α 6 preferentially dimerizes with integrin β 4 over integrin β 1 [40], and is known to regulate cancer cell invasion and motility [41].
Given that integrin α 6 β 4 is laminin-binding and we saw mechanosensitive changes in gene and protein expression (Fig. S2a), we hypothesized that we could tune integrin α 6 mechanosensing on these surfaces by supplementing the collagen-functionalized hydrogel surfaces with β 1 -chain-containing laminins. On soft gels, integrin α 6 expression was higher at 24 hours with added laminin compared to collagen 1 only and, interestingly, it was also high after 6 days of culture, even without laminin (Fig. 1c). This mechanosensitivity appears to be specific to α 6 as other laminin-binding integrins (e.g. integrin α 3 ) did not follow this trend (Fig.   S2b).
We hypothesized that cells must be producing their own laminin to initiate integrin α 6 expression at the longer culture times. Immunofluorescent staining revealed that laminin, but not collagen 1, was upregulated on the softer gels at 6 days (Fig. 1d). We separately confirmed that differential staining was not due to ECM proteins provided to the cells either via covalent crosslinking or from serum (Fig. S3). Cultures that were decellularized after 6 days of culture showed the highest amount of extracellular laminin on the softest, 1 kPa gels (Fig. 1d), concordant with the increase in integrin α 6 expression in Figure 1c. To understand the impact of laminin and integrin α 6 expression on cell adhesion to the gels, we quantified cell spreading during initial adhesion. We found cells spread to similar extents across the 4 stiffnesses examined here when only collagen 1 was present (Fig. 1e). When laminin was added to the collagen on the gel surfaces, cells spread to a smaller extent, on soft, but not stiff gels (Fig. 1f). In sum, cells upregulate integrin α 6 expression on soft substrates, which is correlated with secretion of extracellular laminin. This coordination appears to take several days, but the mechanosensitivity effect on cell spreading can be accelerated by adding β 1 -chain-containing laminin to the culture substrata.

Cell Response to EGF Stimulation Mirrors Mechanosensing on Laminin
Given the prevalence of EGF in the tumor microenvironment and its known role in promoting cancer cell motility, we posited that the mechanosensitive expression of EGFR alongside the laminin-binding integrin α 6 could hint that they mechanotransduce under a related pathway. Similar to previous reports, we found that MDA-MB-231 cells were smaller on both 1 and 41 kPa gels upon the addition of EGF prior to and during adhesion [34], achieving a similar size to cells on soft, laminin-coated gels (Fig. 2a). EGFR is known to associate with the α 6 β 4 integrin dimer to enhance clustering and response to EGF [42], and we found that the addition of soluble EGF increased the association of integrin α 6 with EGFR (Fig. S4a). We separately analyzed the protein expression of integrin α 6 and EGFR in tropic variants of the MDA-MB-231 cell line, which were selected in vivo for their ability to metastasize specifically to the brain, bone, and lung. We noted that the brain tropic cells, which metastasize to the softest of these sites, have the highest expression of both integrin α 6 and EGFR (Fig. S4b).
To test the necessity of integrin α 6 in the observed response to EGF, we generated a stable shRNA knockdown cell line (Fig. S4c-e). We then quantified adhesion of the integrin α 6 knockdown cells (shITGA6) to collagen-coated gels, and found that they remained large during EGF stimulation, compared to the scramble control (shSCR), suggesting that integrin α 6 is required for the typical EGF response (Fig. 2b).

Substrate-Bound EGF Restrains EGFR to Focal Adhesions and Maximizes Cell Motility on Soft
Substrates EGF in solution is rapidly internalized within 10 minutes of administration in vitro (Fig. 2c-d). To maximize the likelihood of sustained interactions between EGFR and integrin α 6 necessary for downstream signaling [42], we covalently linked both collagen 1 and EGF to the surface of the PEG-PC gels (Fig. 2e). This EGF presentation minimized EGFR internalization (Fig. 2e) and localized both pEGFR(Y1068) and total EGFR to regions of vinculin staining (Fig. S5a-b), suggesting this surface bound EGF facilitates cooperative signaling between integrins and EGFR.

EGFR receptor internalization was required to alter adhesion, as only cells seeded with soluble EGF
showed measurable differences in cell area (Fig. 2f). However, both bound and soluble EGF had large effects on cell motility. First, we observed a biphasic trend in cell speed with respect to stiffness, as previously described [7]. The addition of soluble or bound EGF increased overall motility on both soft and stiff substrates (Fig. 2g). In fact, we were forced to expand the mechanical range of the hydrogels to much softer conditions to capture the full biphasic curve during EGF stimulation (Fig. S5c) [26]. We also observed a role for integrin α 6 in regulating this EGF-dependent motility, as the integrin α 6 knockdown population (shITGA6) was significantly less responsive to EGF than the scramble control cells, which peaked at 34.2 µm/hr on 1 kPa gels ( Fig. 2g, grey line, Fig. S5d).

Calpain 2 is Downstream of EGFR and Integrin α 6 Mechanosensing
Inspired by the RNAseq data (Fig. S1), we wondered whether calpain 2 was an intracellular protein downstream of both EGFR and integrin α 6 mechanosensitivity. PCR data at 24 hours of culture on different substrates uncovered that ITGA6 and EGFR expression clustered with the intracellular protease CAPN2 (Fig.   3a). Calpain 2, which has been implicated in focal adhesion formation and disassembly, is thought to play an important role in breast cancer cell motility and metastasis [43]. Because calpain 2 is activated by ERK [44], we performed phospho-analysis of multiple kinases in integrin-and EGFR-dependent pathways (Fig. S6a). Only (Fig. S6b). Steady-state phosphorylation of EGFR, MEK, and ERK increased on stiffer surfaces, as previously demonstrated [17, 45,46]. However, we observed the opposite trend with the activation of calpain 2, and the cleavage of its substrate, talin (Fig. S6c).
We observed more sustained calpain 2 activity (autolysis) during adhesion to hydrogels compared to tissue culture plastic (Fig. 3b). This was only a slight trend; however, there was an obvious functional role for calpain 2 during mechanosensing on the gels. A pharmacological inhibitor to either calpain 2 or ERK had no effect on cell spreading on 1 kPa gels, but resulted in significantly larger cells on 41 kPa gels (Fig. 3c-d). To confirm the necessity of calpain 2 in responding to EGF stimulation, we generated a CAPN2 knockdown cell line (Fig. S7a-c), and observed that those cells were less responsive to EGF during adhesion, independent of stiffness ( Fig. S7d).
We further hypothesized that calpain 2 could disturb the typical biphasic relationship between cell migration speed and matrix stiffness given its known role in mediating focal adhesion turnover [44]. Inhibiting ERK activity entirely eliminated the durokinesis effect, and calpain 2 inhibition shifted the migration curve maximum to lower stiffnesses (Fig. 3e). The limiting step during motility on stiff surfaces is retraction of the rear edge [47]. Here we demonstrate that motility on stiff substrates is sensitive to calpain 2 inhibition, likely because cells on stiffer surfaces have decreased basal calpain 2 activity (Fig. S6c). After 24 hours of culture on PEG-PC gels with collagen 1 or collagen 1 and bound EGF, cells adopted a normal morphology with no noticeable protrusions (Fig. 3f). However, in the presence of lapatinib (dual kinase EGFR-HER2 inhibitor), ERK inhibitor, or calpain 2 inhibitor, we observed that cells left behind long, stable protrusions at the rear of the cell, indicative of their inability to release adhesion sites (Fig. 3g) [48]. While we demonstrated earlier that integrin α 6 is necessary for both mechanosensitive and EGFR-mediated adhesion, this data suggests that EGFR-ERK-calpain 2 signaling contributes to mechanosensitive motility necessary for breast cancer metastasis.

Discussion
This work demonstrates integrin α 6 and calpain 2 as mechanosensitive proteins important for stiffnessdriven breast cancer cell adhesion and migration (Fig. 4a). The mechanosensitivity of calpain 2 is a new finding, as well as the required physical association of EGFR with integrin α 6 to facilitate cell adhesion (Fig. 4b). EGF is known to enhance breast cancer cell adhesion and motility, and here we demonstrate that the mechanosensitive integrin α 6 -calpain 2 signaling axis can mimic the effect of EGF-stimulated cell adhesion and motility (Fig. 4c).
Cancer metastasis initiates with individual or collective groups of cells leaving the primary tumor site [49][50][51]. This motility is driven in part by cell-ECM adhesion via integrins and the creation and turnover of focal adhesions, both of which are known to be sensitive to the stiffness of the surrounding ECM [52]. Integrins have variable responses to mechanical forces derived from the microenvironment. For example, on 2D gels, integrin β 3 stiffness-dependence requires surface bound BMP-2 [21]. Expression of integrins α 2 , α 3 , and β 1 are higher on TCPS in mouse mammary epithelial cells, when compared to plating on a soft basement membrane-like matrix [53]. Integrin α 5 expression was also found to increase 5-fold with stiffness between 1 kPa and 10 kPa [54].
While most previous work has focused on the mechanical regulation of collagen-and fibronectin-binding integrins, one group has examined integrin α 6 in fibroblasts, where expression increases with stiffness to cause an increase in invasion [55]. We demonstrate here the integrin α 6 expression response in breast cancer cells, and its mechanosensitivity in cell adhesion, migration, and response to EGF (Fig. 1c, Fig. 2b). Integrin α 6 is a laminin-binding subunit, and so translating its mechanosensitive expression to altered mechanotransduction requires the addition of laminin to the culture substrate. On soft surfaces, cells produce extracellular laminin to engage integrin α 6 in the absence of exogenously provided ligand.
Previous work has established the role of integrin α 6 in key aspects of the metastatic cascade. Low integrin α 6 expression inhibits the migration and proliferation necessary for establishing metastases at distant secondary sites [56]. Additionally, integrin α 6 expression can protect cells from radiotherapy treatment both through PI3K and MAPK activity [57]. We observed that during cell adhesion, integrin α 6 engagement with laminin decreases cell size and facilitates adhesion similarly to EGF (Fig. 2a-b). Smaller cells can be correlated with a cancer stem cell and invasive phenotype [58], and our work suggests integrin-mediated mechanotransduction contributes to that behavior. Our breast cancer cells produce laminin intracellularly within the first 24 hours and this laminin is (a) Breast cancer cells on soft gels, or any stiffness gel supplemented with EGF, were small and motile. Cells on soft gels produced their own laminin to engage α 6 , or we added it exogenously. Our data suggests this increases activation of ERK and Calpain 2, which feeds back to increase turnover of focal adhesions. (b) During adhesion to substrates, cells were more responsive to laminin on soft substrates but did not adhere as well on stiff substrates with a calpain 2 inhibitor. Cell adhesion was facilitated in the presence of EGF on both soft and stiff gels. (c) After 24 hours on gels, calpain 2 activity decreased with increasing stiffness, while EGFR phosphorylation increased with stiffness without added EGF. Cell motility had a biphasic dependence on substrate stiffness. secreted by 6 days of culture with the most laminin produced on the softest substrate (Fig. 1d). Interestingly, this phenomenon is not maintained in other secreted ECM proteins, such as collagen 1 (Fig. 1d) or fibronectin (data not shown), suggesting a key role for integrin α 6 -mediated adhesion to laminin on soft surfaces. This is possibly mediated through the mechanically responsive transcription factor, TAZ [59], which has been shown to regulate laminin production [60].
In a cohort of 80 patients, integrin β 4 was co-expressed with integrin α 6 and laminin in primary tumors, and that co-expression of α 6 β 4 with laminin production was significantly correlated with breast cancer relapse and death [40,61]. We found that both integrin β 4 expression and laminin production were enhanced after 6 days of culture on soft gels (Fig. 1b, Fig. 1d), indicating these cells have adopted a more aggressive phenotype.
Downstream of integrin α 6 engagement and EGFR phosphorylation, cells activate calpain 2, which facilitates focal adhesion turnover, and therefore, motility (Fig. 3e). Recent work has also demonstrated that calpains are recruited to protrusions containing integrin β 4 [62]. The α 6 β 4 integrin dimer associates with EGFR, which enhances receptor clustering and increases cellular capacity to respond to EGF (Fig. 2b) [42]. Here, we contribute that in breast cancer cells, integrin α 6 shows increased expression on the softest substrates, where adhesion to laminin mimics the phenotype induced by EGF. This provides insight to the effect of mechanics on adhesion formation in the tumor microenvironment.
Previous work has shown that EGF enhances cell speed while decreasing persistence [63], and here we demonstrate that during migration the effect of EGF on migration speed is significant on soft and stiff, but not intermediate substrates, where speed is already maximized (Fig. 2g). Motility is preferentially driven by surface localized EGFR, and we demonstrate that bound EGF restrains EGFR at the cell membrane (Fig. 2e) to enhance phosphorylation of Y1068 on EGFR (Fig S5a). Triggered by either ligand binding or integrin clustering, EGFR activates ERK [64] and subsequent ERK activity is localized to newly formed focal adhesions to propagate a local downstream response [65,66].
Localized EGFR phosphorylation is necessary for activity in downstream effectors, including MEK, ERK, and calpain 2 at the cell membrane [67,68], to cleave focal adhesion proteins vinculin, talin, paxillin, and FAK [69,70]. During adhesion, calpain 2 activity was sustained on the hydrogels compared to TCPS, which was consistent with RNA levels and steady state cleavage of calpain 2 and talin ( Fig. 1a-b, Fig. S6c). Calpain 2 inhibition during adhesion resulted in larger, less motile cells, particularly on stiff substrates. Cells were smaller at both 1 and 41 kPa in the presence of soluble EGF (Fig. 2a). Together, these data suggest that EGFR phosphorylation could be a stronger activator of calpain 2 than stiffness. Previous work has shown that calpain 2 has a significant correlation with an epithelial-to-mesenchymal transition [71] and lymphatic or vascular invasion, where patients with basal-like tumors and high calpain 2 expression had a significantly worse prognosis [72]. Our work demonstrates a mechanosensitive role for calpain 2, providing further evidence for its potential as a druggable target for metastatic breast cancer. In sum, both integrin α 6 and EGF regulate cell motility, and these pathways converge at stiffness-dependent calpain 2 activity.

Conclusions
Mechanical forces have long been thought to play a role in tumor progression, and here we shed light on two mechanoresponsive proteins that influence breast cancer cell motility. First, enhanced integrin α 6 expression on soft substrates is associated with adhesion on a basement membrane-like environment and laminin production. Second, clustering and cooperative signaling between EGFR and focal adhesion proteins, such as integrin α 6 , increases activity of calpain 2 on soft substrates, which mediates the destabilization and turnover of focal adhesions necessary for motility. Others have suggested targeting calpains to inhibit cancer cell invasion [73], and here we provide a mechanistic understanding of the role of calpain 2 in stiffer environments that may better recapitulate the tumor microenvironment.