PEAK1 Y635 phosphorylation regulates cell migration through association with Tensin3 and integrins

Cell adhesion to the ECM is essential for the development and postnatal homeostasis of multicellular organisms. Cell–ECM adhesions are largely facilitated by the integrin family of adhesion receptors, which are formed through heterodimerization of an α and β subunit. Depending on the combination of the 24 α and 8 β subunits, integrins bind different ECM components, including laminins, collagens, and proteins containing an arginine (R), glycine (G), aspartic acid (D) tripeptide, such as fibronectin and vitronectin (Geiger and Yamada, 2011; Hynes, 2002). Integrins link the ECM to the intracellular cytoskeleton via a regulated network of adaptor proteins and recruit many cytoplasmic scaffolding and signaling proteins to generate an integrated signaling network with growth factor receptors (Morse et al., 2014). These protein complexes, known as the “integrin adhesome,” play key roles in a wide range of cellular processes, including cell migration, proliferation, differentiation, and survival. Failure to properly establish cell adhesion complexes can result in severe human diseases (Winograd-Katz et al., 2014).

Multiple studies have been undertaken to characterize the integrin adhesome and mechanistically understand how it drives a multitude of cellular functions in health and disease (Horton et al., 2016; Manninen and Varjosalo, 2017). These efforts have resulted in an extensive literature-based adhesome protein–protein interaction network (Winograd-Katz et al., 2014; Zaidel-Bar et al., 2007), and a large list of adhesome components are a result of multiple proteomic studies (Manninen and Varjosalo, 2017). Combining different proteomic studies generated a dataset of 60 components termed the consensus integrin adhesome (Horton et al., 2015; Humphries et al., 2009; Ng et al., 2014; Robertson et al., 2015; Schiller et al., 2011; Schiller et al., 2013). Remarkably, only half of the proteins in the consensus adhesome are described in the literature-curated adhesome (Manninen and Varjosalo, 2017), indicating the need for further refinement.

A powerful technology to study protein interactions at nanometer resolution in integrin adhesion sites is the proximity-dependent biotin identification (BioID) assay combined with mass spectrometry (MS)-based proteomics (Roux et al., 2012). BioID has been used to characterize the proximity interactors of the focal adhesion (FA) proteins Paxillin and Kindlin2 (Dong et al., 2016) and to map the composition and topological features of integrin adhesion complexes by using 16 commonly identified adhesome components as bait (Chastney et al., 2020). Integration of these BioID datasets revealed a network of 147 proteins, of which only 11 were identified by multiple bait proteins and most likely represent the core adhesome components (Chastney et al., 2020). We also successfully used BioID to characterize the proximitome of integrin β4 and β5 in epithelial cells (Te Molder et al., 2020; Wang et al., 2020; Zuidema et al., 2018).

In the present study, we characterized the proximitome of integrin β1 and β3 and found a set of five adhesome components also reported in other BioID datasets (Chastney et al., 2020; Dong et al., 2016). Among the established integrin interactors Talin1, Kindlin2, and Tensin3, we also identified Kank2 and pseudopodium-enriched atypical kinase 1 (PEAK1; SgK269). PEAK1 is a member of the PEAK family of pseudokinases, can form homo- or heterodimers with its family member inactive tyrosine-protein kinase PRAG1 (Skg223; Ha and Boggon, 2018; Patel et al., 2017; Patel et al., 2020), and associates with FAs and the EGFR signaling pathway (Bristow et al., 2013; Chastney et al., 2020; Dong et al., 2016; Wang et al., 2010; Zheng et al., 2013). How PEAK1 is integrated into FAs and whether it links FA and EGFR signaling is poorly understood. Several studies have reported elevated levels of PEAK1 and implicated PEAK1 in the regulation of tumor cell proliferation and invasion (Abu-Thuraia et al., 2020; Agajanian et al., 2015; Croucher et al., 2013; Ding et al., 2018; Fujimura et al., 2014; Kelber et al., 2012; Strnadel et al., 2017). Furthermore, it was shown that PEAK1 is phosphorylated by Src family kinases downstream of growth factor receptors, which promotes its localization in FAs (Agajanian et al., 2015; Bristow et al., 2013; Leroy et al., 2009; Wang et al., 2010; Zheng et al., 2013). PEAK1 also interacts with the scaffolding protein Shc1, which occurs upon EGF stimulation with a delayed kinetics and regulates the assembly of a protein cluster involved in cytoskeletal reorganization (Zheng et al., 2013). Finally, PEAK1 has been identified as a component of the canonical Src-p130Cas-CrkII-Paxillin and Erk signaling pathways (Wang et al., 2010). Our findings show that PEAK1 is a specific proximity interactor of RGD-binding integrins in FAs, and that phosphorylation of Tyr-635 in PEAK1 regulates cell migration through association with Tensin3. In turn, PEAK1 recruits Shc1 to FAs, which increases Shc1 phosphorylation and thereby links the integrin adhesion and EGFR signaling pathways.

To characterize the integrin β1 and β3 proximitome by BioID (Fig. 1 A), we transfected β1-null mouse epithelial GE11 cells (Gimond et al., 1999) with integrin β1- or β3-BirA* cDNA and selected the bulk population of cells expressing the integrin-BirA* fusion proteins by FACS. Integrin β1- and β3-BirA* localized in FAs and β1-BirA* additionally in cell–cell contacts and in adhesion structures that resemble fibrillar adhesions/podosomes (Fig. 1 B). Cells were treated for 24 h with biotin before cell lysis, and untreated cells were used as negative controls. The use of untreated controls allowed us to normalize for any changes in protein composition resulting from expression of integrin β1 or β3. Purified proteins were run and excised from a stained polyacrylamide gel and analyzed by LC-MS/MS. We cannot rule out that the C-terminal fusion of the BirA* to the cytoplasmic tail of integrin β1 or β3 could interfere with the recruitment of some integrin-associated proteins; nevertheless this approach resulted in datasets of 170 and 61 proteins for β1 and β3, respectively (Fig. 1, C and D; Fig. S1; and Tables S1, S2, S3, and S4). Among the top integrin proximity interactors, we found the integrin-binding proteins Talin1/2, Kindlin2, and Tensin3. Furthermore, PEAK1 and Kank2 were also among the significantly enriched proteins (Fig. 1, C and D). Major differences between the β1 and β3 proximitomes were the integrin α subunits (α2, α3, α5, α6, and α7 for β1 and αV for β3) and the enrichment of CD151 in the proximity of β1 (Fig. 1 E). Further comparison of the β1 and β3 datasets revealed a common set of 46 proteins (Figs. 1 F and S1 A), which was further compared to the proximitomes of Paxillin and Kindlin2 (Dong et al., 2016) revealing an overlap with 6 adhesome components (Fig. 1 F). Among them were proteins associated with FAs (Talin1, Kindlin2, and Tensin3) and the cortical microtubule stabilizing complex (Kank2; Bouchet et al., 2016; Sun et al., 2016). Additionally, PEAK1 and EphA2 were identified in all four datasets. Talin1, Tensin3, Kank2, and PEAK1 were also identified by Chastney et al. (2020) using multiple FA proteins, including Kindlin2, as bait. Whereas Talin1 and Kindlin2 were reported in both the consensus adhesome (Horton et al., 2015) and the literature-curated adhesome (www.adhesome.org; Winograd-Katz et al., 2014; Zaidel-Bar et al., 2007), Tensin3 was reported only in the consensus adhesome, PEAK1 in the literature-curated dataset, and EphA2 and Kank2 in neither (Fig. 1 E).

The identification of Talin1, Kindlin2, Tensin3, Kank2, and PEAK1 as adhesome components in multiple studies, together with their close association to integrin β1 and β3 observed in our experiments, points to their crucial role in integrin function. Because the interaction between PEAK1 and integrins is poorly understood, we decided to study this interaction.

Cell adhesion to fibronectin was shown to induce PEAK1 tyrosine phosphorylation (Wang et al., 2010). The fibronectin dependence of this process suggests that PEAK1 associates with RGD-binding integrins. Immunofluorescence analysis of endogenous PEAK1 (Fig. 2 A) or overexpression of PEAK1-GFP (Fig. 2 B) in GE11/β1-BirA* cells shows that the protein localized to cell matrix adhesions.

Integrin αVβ3 is a receptor for RGD-containing ECM proteins, but β1 can heterodimerize with 12 different α subunits to additionally mediate adhesion to laminins and collagens. To study the specificity of the integrin–PEAK1 interaction, we overexpressed integrin α3-, α5-, or α6-BirA* fusion proteins in GE11 cells that also stably express integrin β1. These integrin α subunits are expressed at high levels in GE11 cells and are among the top hits in the β1 proximitome (Fig. 1 C). All overexpressed integrin α BirA* constructs were expressed at the cell surface and similar levels of Itg. β1 could be detected using FACS analysis (Fig. S1 B). Integrin α3β1 and α6β1 are laminin receptors, whereas α5β1 binds fibronectin. Overexpression of the α subunits resulted in an altered β1 subcellular distribution, with β1 localizing in FAs in the α5-BirA*–expressing cells and showing a more dispersed organization in the α3- and α6-BirA*–expressing cells (Fig. 2 C), likely owing to the ability of integrin α3β1 and α6β1, but not α5β1, to incorporate into tetraspanin webs through their interaction with CD151 (Charrin et al., 2014).

To examine whether PEAK1 associates primarily with the RGD-binding α5β1 integrin, we performed BioID assays combined with Western blot analysis to determine the levels of biotinylated PEAK1 in GE11 cells stably expressing integrin β1 and overexpressing integrin α3-, α5-, or α6-BirA* fusion proteins. Strong biotinylation of PEAK1 was detected in cells expressing α5-BirA* but was absent or hardly detected in cells expressing the α3- or α6-BirA* constructs (Fig. 2 D). Altogether, these findings demonstrate that PEAK1 localizes in FAs and acts downstream of the RGD-binding integrins α5β1 and αVβ3.

Many adaptor proteins engage with the conserved membrane-proximal (MP) NPxY or membrane-distal (MD) NxxY motifs on the cytoplasmic tail of integrin β subunits (Calderwood et al., 2003). For example, the MP-NPxY is required for the interaction of Talin1 with the integrin β1/3 subunit, while the MD-NxxY motif is needed for Kindlin2 binding (Moser et al., 2009). PEAK1 does not contain an integrin-binding FERM or PTB domain found in other proteins that interact with the MP-NPxY or MD-NxxY motifs (Wang et al., 2010). Therefore, we hypothesized that the interaction between PEAK1 and integrin β subunits might be indirect, through an adaptor protein. To further examine the PEAK1–integrin interaction, we generated β1- and β3-BirA* fusion proteins carrying disruptive Y>A mutations in the MP-NPxY (termed N1), MD-NxxY (termed N2), or both (termed N1 + 2; Fig. 3 A). These mutants were stably expressed in β1-null GE11 cells; FACS analysis confirmed that all the mutants were expressed at the cell surface at similar levels (Fig. S1 C). The generated cell lines were used to perform BioID combined with MS (Fig. 3, B and C). Our experiments revealed that the proximity of Talin, Tensin3, and PEAK1 to β1/3 was strongly impaired when MP-NPxY was mutated and reduced when MD-NPxY was mutated. These results were confirmed by performing BioID assays in combination with Western blot analysis (Fig. 3, D and E).

To determine which integrin MP-NPxY motif interacting protein could mediate the PEAK1–integrin association, we depleted established interactors of the integrin MP-NPxY motif by using siRNAs and studied the effects on the integrin–PEAK1 interaction by BioID and Western blot. Interestingly, knockdown of Talin1/2 or Filamins (Calderwood et al., 2003; Kiema et al., 2006) did not reduce the integrin β1–PEAK1 proximity interaction (Fig. S2 D). We therefore wondered whether the proximity interaction between RGD-binding integrins and PEAK1 could be mediated by Tensin3, which contains a C-terminal PTB domain shown to interact with the integrin MP-NPxY domain (Calderwood et al., 2003; McCleverty et al., 2007) and was identified as an interactor of β1 and β3 in our proximity screen (Figs. 1 and 3). Unfortunately, we were unable to successfully reduce Tensin3 expression by using siRNAs/shRNAs or CRISPR/Cas9-mediated gene disruption in GE11 cells. As an alternative approach to investigate whether Tensin3 mediates the integrin–PEAK1 association, we generated PEAK1-deficient GE11/β1-BirA* cells using CRISPR/Cas9 and subsequently introduced WT or mutant forms of PEAK1 that carried a Y635F or Y1188F substitution. The Y635F mutation was selected based on the remarkable similarity of the ⁶³⁵YDNL sequence in PEAK1 to the YDNV sequence present in DLC1, which mediates an interaction with the Tensin SH2 domain that precedes the PTB domain (Liao et al., 2007; Yam et al., 2006). Moreover, YDNL was identified as a binding peptide motif in a pY peptide library screen using the SH2 domain of Tensin (Wavreille and Pei, 2007). On the other hand, Y1188 was shown to bind Shc1, which could represent an alternative connection between PEAK1 and the integrin β NPxY motif (Cowan et al., 2000; Deshmukh et al., 2010; Smith et al., 2006; Zheng et al., 2013). BioID experiments combined with Western blot analysis clearly indicated that phosphorylation of Y635 in PEAK1 establishes the proximity interaction with integrin β1, while Shc1 binding is not required for the PEAK1–β1 interaction (Fig. 4, A and B).

To demonstrate that PEAK1 interacts directly with Tensin3 in a phosphorylation-specific manner, we expressed GFP-tagged PEAK1 or PEAK1-Y635F in CRISPR/Cas9-generated PEAK1-knockout human colorectal SW480 cells (Fig. S2 A). GFP-PEAK1 was expressed at high levels and used for GFP pull-down experiments. Cells were serum-starved overnight and treated with pervanadate, an irreversible inhibitor of protein tyrosine phosphatases, before lysis. GFP pull-down experiments showed that Tensin3 is associated with PEAK1 in pervanadate-treated samples but hardly associated with PEAK1 in serum-starved PEAK1-WT– or PEAK1-Y635F–expressing cells (Fig. 4, C and D). Furthermore, PEAK1-Y635F showed reduced colocalization with Tensin3 compared with PEAK1-WT (Fig. S2 B). Intriguingly, the interaction between DLC1 and the SH2 domain of Tensin3/c-ten does not require phosphorylation of the YDNV motif in DLC1 (Cao et al., 2015; Liao et al., 2007) and is less dependent on the tyrosine phosphorylation of the Tensin3 SH2 domain (Qian et al., 2009). This indicates that direct competition between DLC1 and PEAK1 for Tensin3 binding is unlikely, because phosphorylation events upon growth factor stimulation would promote a switch from Tensin3-DLC1 to Tensin3-PEAK1 binding (Chumbalkar et al., 2011; Cui et al., 2004).

PEAK1 is recruited to FAs upon phosphorylation by Src family kinases (Bristow et al., 2013; Croucher et al., 2013; Leroy et al., 2009; Wang et al., 2010). In GE11 cells, where Src is already maximally active in the absence of EGF stimulation (i.e., its activity cannot be further increased by EGF stimulation; Fig. S2 C), PEAK1 is highly phosphorylated on Y635 (Fig. S2 D). To analyze whether PEAK1 phosphorylation by Src is required for the PEAK1–Tensin3–β1 interaction, we conducted a BioID assay using GE11/β1-BirA* cells treated with Saracatinib, an inhibitor of Src and Abl family kinases. Saracatinib treatment decreased tyrosine phosphorylation of PEAK1 and decreased the proximity interaction between PEAK1 and β1 (Fig. S2 E), in line with previous findings (Bristow et al., 2013). GFP pull-down experiments using PEAK1-deficient GE11 cells expressing GFP-PEAK1 treated with Saracatinib confirmed that phosphorylation of PEAK1-Y635 is decreased upon Src inhibition (Fig. S2 F).

Notably, Src has also been implicated in the phosphorylation of the SH2 domain of Tensin3, and tyrosine phosphorylation of the SH2 domain contributes to complex formation with several proteins (Qian et al., 2009). To determine the importance of tyrosine phosphorylation of the SH2 domain for PEAK1 binding, we transiently expressed SH2-WT and SH2 double (SH2-2F; Y1206 and Y1256F) and SH2 triple (SH2-3F; Y1173F, Y1206, and Y1256F) mutants, engineered as GST fusion proteins together with PEAK1 in COS-7 cells. Cells were treated or not with pervanadate for 10 min. GST pull-down confirmed that the Tensin3 SH2-WT binds to PEAK1 and showed that tyrosine phosphorylation of the SH2 domain greatly enhanced the binding of Tensin3 to PEAK1 (Fig. 4 E). Compared with Tensin3 SH2-WT, binding of the SH2-2F mutant to PEAK1 was only slightly reduced, but that of the SH2-3F mutant was reduced by >50% (Fig. 4, E and F).

Next, we analyzed the PEAK1-Tensin3-integrin interaction in human U251MG glioblastoma cells, which express Tensin3 at high levels. PEAK1 and Tensin3 colocalized in FAs, as shown by immunofluorescence analysis (Fig. 4 G). In line with our biochemistry results, PEAK1 was absent from cell matrix adhesions in U251MG cells spread on fibronectin in the presence of the Src inhibitor Saracatinib (Fig. S2 G). Integrin αVβ5 is the main RGD-binding integrin found to colocalize with Tensin3 and PEAK1 in FAs, whereas integrin β1 shows a more dispersed subcellular distribution, and integrin β3 could not be detected (Fig. 4 G). We performed BioID assays by expressing β5-BirA* in β5-deficient U251MG cells that were additionally depleted of Tensin3 by shRNA-mediated gene knockdown. Depletion of Tensin3 by two different shRNAs reduced the PEAK1–integrin β5 proximity interaction (Fig. 4, H and I), while the Talin1-integrin β5 interaction was not affected (Fig. S2 H). Furthermore, PEAK1 was absent from cell matrix adhesions in Tensin3-depleted cells (Fig. S2 I). Finally, we compared the localization of GFP-PEAK1 WT versus Y635F in FAs by immunofluorescence analysis. PEAK1-Y635F showed significantly reduced colocalization with the FA marker Vinculin compared with WT PEAK1 (Fig. 4, J and K). Taken together, these results indicate that Src-mediated phosphorylation of PEAK1 and the SH2 domain of Tensin3 promote their binding and facilitate PEAK1 interaction with the integrin β MP-NPxY motif.

To identify additional proteins that interact with PEAK1, we used a GFP-Trap–based affinity purification method combined with MS using PEAK1-deficient GE11 cells that express GFP-PEAK1, or the PEAK1-deficient cells as a negative control. This screen identified Shc1 and many proteins involved in the EGFR/Shc1 signaling network (Zheng et al., 2013), including Asap1/2, Dab2ip, Grb2, Lrrk1, Ppp1ca/cc, Prag1, and Rasal2 as major PEAK1 interacting proteins (Fig. 5 A and Table S5). Both human and mouse PEAK1 were enriched in the GFP-PEAK1 sample, indicating that endogenous PEAK1 is not completely deleted in our GE11 control sample and dimerizes with the GFP-PEAK1 fusion protein (Patel et al., 2020). Tensin3 was not found as a significant interactor of PEAK1 in this analysis (Fig. 5 A). However, in a GFP-Trap pull-down assay using SW480 cells, a physical interaction between Tensin3 and PEAK1 could be demonstrated and was found to be dependent on the phosphorylation of PEAK1 at Y635 (Fig. 4 C). Interestingly, Asap1/2 and Grb2 were recently identified as binding partners of the smaller PEAK family member PEAK3 (Hou et al., 2022), suggesting that these proteins may interact with conserved domains in PEAK1–3.

Because PEAK1 is a scaffolding protein that functions in late EGFR-mediated signaling responses (Zheng et al., 2013), we investigated whether the PEAK1 interactome changes upon 20-min EGF stimulation. The major interactors of PEAK1 remained unchanged after EGF treatment (Fig. S3), which is in line with our finding that Src is already maximally phosphorylated and PEAK1 phosphorylation is high in GE11 cells. However, Ptpn13 and Shroom2 were identified as additional, EGF-induced PEAK1-binding proteins (Fig. S3). The interaction of these proteins with PEAK1 likely depends on non–Src-mediated posttranslational modifications that occur downstream of EGFR activation.

To further analyze the scaffolding function of PEAK1, we expressed different PEAK1 mutants in PEAK1-depleted GE11 cells and subsequently performed GFP pull-down experiments. Western blot analysis indicated that the Y1188F mutation almost completely abrogated the binding of PEAK1 to Shc1, but not to Rasal2 and only slightly to Grb2 (Fig. 5 B). Grb2 binding to PEAK1-Y635E/F decreased by ∼50%, whereas Shc1 was unchanged. Interestingly, however, binding of tyrosine-phosphorylated Shc1 to PEAK1-Y635E/F was strongly decreased (Fig. 5 B). In line with these biochemical findings, immunofluorescence analysis revealed that PEAK1 recruits Shc1 to FAs (Fig. 5 C). Recruitment of Shc1 to FAs in PEAK1-deficient or PEAK1-Y1188F mutant cells shown by colocalization with phospho-Paxillin was significantly reduced (Fig. 5, C and D). Because PEAK1-Y635F cannot bind Tensin3, we speculate that the phosphorylation of Shc1 is increased upon its recruitment to FAs, via PEAK1–Tensin3–integrin binding. Grb2, which also binds PEAK1-Y635, is not found in FAs and therefore does not play a major role in the localization and phosphorylation of Shc1 in FAs. In line with our findings, Shc1 interacts with the FA-forming integrins, including α5β1 and αVβ3, and is phosphorylated by integrin β1 crosslinking or when cells are plated on fibronectin/vitronectin (Wary et al., 1996). Shc1 controls various cellular processes in association with integrins and growth factor stimulation, including cell survival, proliferation, and migration (Mauro et al., 1999; Wary et al., 1996). Therefore, we tested the contribution of PEAK1 phosphorylation at either Y635 or Y1188 to cell motility. GE11 cells depleted of endogenous PEAK1 migrated ∼40% more slowly compared with WT GE11 cells; reexpression of PEAK1-GFP was able to rescue this reduction in cell velocity (Fig. 5 E). The motility defect of the PEAK1-depleted GE11 cells was also rescued by expression of PEAK1-Y1188F-GFP, but not PEAK1-Y635F-GFP (Fig. 5 E).

We report here that Tensin3 mediates the proximity interaction between PEAK1 and RGD-binding integrins through binding of its SH2 domain to PEAK1. Our findings indicate that PEAK1 mainly interacts with Shc1 and its associated protein cluster (Zheng et al., 2013) and recruits Shc1 to FAs by interacting with Tensin3 and RGD-binding integrins (Fig. 5 F). Moreover, integrins and growth factor receptors synergize to regulate cellular functions, including cell motility (Miyamoto et al., 1996; Morello et al., 2011). In line with these findings, expression of PEAK1-Y635F, which showed decreased binding to tyrosine-phosphorylated Shc1, was unable to restore the migration velocity of GE11 PEAK1 KO cells, unlike expression of PEAK1 WT (Fig. 5 E). PEAK1 could be an important player in regulating integrin–growth factor receptor synergism, as it connects RGD-binding integrins to the protein cluster regulating late EGFR/Shc1 signaling (Zheng et al., 2013).

Primary antibodies used are listed in Table S6. Secondary antibodies were as follows: goat anti-rabbit Alexa Fluor 488, goat anti-mouse Alexa Fluor 488, goat anti-mouse Texas Red, goat anti-mouse Alexa Fluor 568, donkey anti-rabbit Alexa Fluor 594, goat anti-rabbit Alexa Fluor 647, and goat anti-mouse Alexa Fluor 647 (Invitrogen); PE-conjugated donkey anti-rabbit antibody (#406421; BioLegend); and stabilized goat anti-mouse HRP-conjugated and stabilized goat anti-rabbit HRP-conjugated (Pierce; Bio-Rad).

The phospho-peptide KVPIVINPNAY(PO3H2)DNLAIYKS with a cysteine added at the C-terminus was synthesized. The phospho-peptide was chemically conjugated with mariculture keyhole limpet hemocyanin (mcKLH) using Imject Maleimide-Activated mcKLH Spin Kit (77666; Thermo Fisher Scientific) according to the manufacturer’s protocol. Briefly, 1 mg phospho-peptide dissolved in 500 µl conjugation buffer was mixed with 2 mg mcKLH dissolved in water for 2 h at room temperature, and the conjugate was desalted and collected by passing through the Zeba Spin Desalting Column. Two rabbits were immunized with 500 µg mcKLH-conjugated peptide in TiterMax Gold-Adjuvant (T2684; Sigma-Aldrich) and subsequently boosted with the same mcKLH conjugated phospho-peptide in Freund’s incomplete adjuvant (F5506; Sigma-Aldrich) at 4, 8, and 12 wk. After the final immunization, blood of the animals was collected, and IgG was purified on a Protein A Sepharose column (Pharmacia LKB Biotechnology). The phospho-specific antibodies were then purified by positive and negative affinity purification. For positive affinity purification, the phosphorylated peptide was covalently coupled to SulfoLink Coupling Gels from Pierce according to the manufacturer’s protocol. Bound antibodies were eluted with 1 M glycine-HCl, pH 2.9, and neutralized with an equal volume of 1 M Tris-HCl, pH 8.5. Protein-containing fractions were pooled and concentrated with Amicon Ultra-4 10K centrifugal filter devices (Millipore). After washing with PBS using the same filter devices, the concentrated protein fraction was added to a second column in which the corresponding nonphosphorylated peptide was covalently bound. All flow-through fractions containing antibodies were collected and stored at 4°C.

GE11 β1 null cells were isolated from β1 chimeric embryos, as previously described (Gimond et al., 1999). SW480 and U251MG cell lines were obtained from American Type Culture Collection. COS-7 cells have been described previously (De Melker et al., 1997). All cell lines were cultured in DMEM containing 10% heat-inactivated FBS and antibiotics. All cells were cultured at 37°C in a humidified, 5% CO₂ atmosphere.

The target sgRNA against mouse Peak1 (exon 4; 5′-AAG​CAA​TTC​TTG​CAT​TCG​CCA​GG-3′) or human PEAK1 (exon 4; 5′-TGC​CCG​TGT​TCC​TGA​TGC​GG-3′) was cloned into pX330-U6-Chimeric_BB-CBh-hSpCas9 (a kind gift from Feng Zhang, Broad Institute of MIT and Harvard, Cambridge, MA [Cong et al., 2013]; plasmid #42230; Addgene). GE11 cells were transfected with this vector in combination with a blasticidin cassette, as previously described (Heijink et al., 2015), and selected with 4 μg/ml blasticidin (Sigma-Aldrich) for 4 d after transfection. Transfected SW480 cells were selected with 2.5 μg/ml puromycin for 3 d.

The integrin β1 (mutants), generated as previously described (Margadant et al., 2012), were subcloned into the EcoRI and BamHI sites of the pcDNA3.1 MCS-BirA(R118G)-HA vector (a gift from Kyle Roux [plasmid #36047; Addgene; Roux et al., 2012]). The integrin α3-BirA* construct was cloned by 3-point ligation of an N-terminal PCR fragment containing an EcoRI site and a C-terminal part of 2.3 kb into EcoRI and BamHI sites of the pcDNA3.1 MCS-BirA(R118G)-HA vector. The integrin α5-BirA* construct was cloned by 3-point ligation of a N-terminal part of 2.7 kb and a C-terminal PCR fragment lacking the stop codon into the BamHI site of the pcDNA3.1 MCS-BirA(R118G)-HA vector. The integrin α6-BirA* construct was described previously (Te Molder et al., 2020). The integrin β1-BirA* (mutants), α5-BirA*, α3-BirA*, and α6-BirA* constructs were cloned into the SnaBI site of the LZRS-IRES-Zeo vector.

The pRetroX-DSRed-SGK269 plasmid encoding human PEAK1 cDNA was kindly provided by Dr. Roger Daly. The PEAK1 construct was subcloned by 3-point ligation of a N-terminal part of 4.9 kb and a C-terminal PCR fragment creating a stop codon and NotI site into EcoRI and NotI sites of pUC19 vector. Point mutants of PEAK1 Y635E, Y635F, and Y1188F were generated by site-directed mutagenesis with the PCR-based overlap extension method using Pfu DNA polymerase (Promega) and subcloned into the EcoRI and NotI sites of the pLZRS-EGFP vector, as previously described (Ketema et al., 2013).

Retroviral vectors were introduced into Phoenix packaging cells using the calcium phosphate method. Virus-containing supernatant was collected 48 h later. After retroviral transduction, infected cells were selected with 0.2 mg/ml zeocin (Invitrogen).

Short hairpins against human Tensin3 (cloned into pLKO.1) were obtained from the MISSION TRC shRNA collection and transfected into HEK 293FT cells using Lipofectamine 2000 (Invitrogen). U251MG β5-BirA*–expressing cells were generated as previously described (Zuidema et al., 2018), infected with the produced recombinant lentivirus, and selected with puromycin for 3 d. Target sequences included: 5′-CCG​GAC​GCA​TAG​GAG​TGG​TCA​TAT​CCT​CGA​GGA​TAT​GAC​CAC​TCC​TAT​GCG​TTT​TTT​G-3′ (shRNA 1) and 5′-CCG​GCT​CCC​AGC​AAA​GCG​TTC​AAA​CCT​CGA​GGT​TTG​AAC​GCT​TTG​CTG​GGA​GTT​TTT​G-3′ (shRNA 4).

Talin1 (M-040902-01-0020), Talin2 (M-065877-01-0020), Filamin-A (M-058520-02-0020), and Filamin-C (M-055631-01-0020) siGENOME SMARTpool siRNAs were purchased from Dharmacon. GE11 cells were transiently transfected using lipofectamine 2000 (Invitrogen). Lipofectamine (20 μl/ml) and siRNA (0.5 μM) solutions in Opti-MEM were mixed (1:1) and incubated for 30 min at room temperature. Cells were incubated with the transfection solution overnight.

Cells were treated as indicated, trypsinized, and washed twice in PBS containing 2% FCS, followed by primary antibody (1:200 dilution) incubation for 1 h at 4°C. Then, cells were washed three times in PBS containing 2% FCS and incubated with secondary PE-conjugated antibodies (1:200 dilution) for 1 h at 4°C. After subsequent washing steps, desired cell populations were isolated by FACS sorting using a Becton Dickinson FACSAria IIu or Beckman Coulter Moflo Astrios cell sorter.

Cells stably expressing the integrin subunit-BirA* fusion proteins were grown on 100-mm or 3 × 145-mm plates (for Western blot or MS, respectively). Plates were coated with a laminin-332–rich matrix obtained by growing RAC-11P cells (Delwel et al., 1994) to complete confluence, after which the plates were washed with PBS and incubated with 20 mM EDTA in PBS overnight at 4 °C. The RAC-11P cells were then removed by extensive washing with PBS.

Cells were washed in cold PBS, lysed in radioimmunoprecipitation assay buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 4 mM EDTA, pH 7.5, 1% NP-40, 0.1% SDS, and 0.5% sodium deoxycholate) supplemented with a protease inhibitor cocktail (Sigma-Aldrich), and cleared by centrifugation at 14,000 g for 30 min at 4°C. Lysates were incubated with Streptavidin Sepharose High Performance beads (GE Healthcare) overnight at 4°C. Beads were washed three times with NP40 buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 4 mM EDTA, pH 7.5, and 1% NP-40) and twice with PBS, and the isolated biotinylated proteins were analyzed by MS or Western blotting.

Cells were washed in cold PBS, lysed in NP40 lysis buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 4 mM EDTA, pH 7.5, and 1% NP-40) supplemented with a protease inhibitor cocktail (Sigma-Aldrich), 1.5 mM Na₃VO₄, and 15 mM NaF (Cell Signaling Technology), and cleared by centrifugation at 14,000 g for 30 min at 4°C. Lysates were incubated with GFP-Trap Agarose (Chromotek) according to the manufacturer’s instructions. Beads were washed three times with NP-40 lysis buffer and twice with PBS, and the isolated proteins were analyzed by MS or Western blotting.

COS-7 cells were cotransfected with plasmids expressing GST and GST fusion proteins of Tensin3 WT (SH2-WT) and SH2-domain mutants (SH2-2F; Y1206F/Y1256F and SH2-3F; Y173F/Y1206F/Y1256; a kind gift from Douglas Lowy [Qian et al., 2009]), and a PEAK1 expressing plasmid using the DEAE-Dextran method (De Melker et al., 1997). 2 d after transfection, the cells were treated with or without pervanadate and lysed in NP-40 lysis buffer. After clearance of the lysates by centrifugation, equal amounts of proteins were used for pull-down assays by adding 25 µl glutathione Sepharose 4B beads (GE-Healthcare) and rotating for 4 h at 4°C. The beads were washed three times with NP-40 lysis buffer and two times with PBS, and bound proteins were eluted with Laemmli sample buffer containing 2% β-mercaptoethanol at 95°C for 5 min. Proteins were separated by electrophoresis using Bolt Novex 4–12% gradient Bis-Tris gels (Invitrogen), transferred to Immobilon-P transfer membranes (Millipore Corp.), and blocked for ≥30 min in 2% BSA in TBST buffer (10 mM Tris, pH 7.5, 150 mM NaCl, and 0.3% Tween-20). Primary antibody (diluted 1:1,000 in 2% BSA in TBST buffer) incubation took place overnight at 4°C. After washing twice with TBST and twice with TBS buffer, blots were incubated for 1 h at room temperature with HRP-conjugated goat anti-mouse IgG or goat anti-rabbit IgG (diluted 1:3,000 in 2% BSA in TBST buffer). After subsequent washing steps, the bound antibodies were detected by enhanced chemiluminescence using SuperSignal West Dura Extended Duration Substrate (Thermo Fisher Scientific) or Clarity Western ECL Substrate (Bio-Rad) as described by the manufacturer. Signal intensities were quantified using ImageJ.

For BioID experiments, peptide mixtures were prepared, measured, and analyzed as previously described (Wang et al., 2020), with the following exceptions. Peptide mixtures (33% of total digest) were loaded directly on the analytical column and analyzed by nanoLC-MS/MS on an Orbitrap Fusion Tribrid mass spectrometer equipped with a Proxeon nLC1200 system (Thermo Fisher Scientific). Solvent A was 0.1% formic acid/water and solvent B was 0.1% formic acid/80% acetonitrile. Peptides were eluted from the analytical column at a constant flow of 250 nl/min in a 115-min gradient, containing a 100-min linear increase from 5 to 24% solvent B, followed by a 15-min wash at 90% solvent B.

Raw data were analyzed by MaxQuant (v2.0.1.0; Cox et al., 2014) using standard settings for label-free quantitation (LFQ). MS/MS data were searched against the Mus musculus Swissprot database (17,082 entries, release 2021_07) complemented with a list of common contaminants and concatenated with the reversed version of all sequences. LFQ intensities were log₂-transformed in Perseus (v1.6.15.0; Tyanova et al., 2016), after which proteins were filtered for at least three valid values (of four total). Differentially expressed proteins were determined using Student’s t test (threshold: false discovery rate [FDR]: 5% and S0: 0.1).

GFP pull-down samples were loaded on an SDS-PAGE gel. Gel bands were excised into ∼1 × 1-mm pieces and subjected to destaining and reduction alkylation of cysteines, followed by overnight digestion with Trypsin (Promega). After digestion, the peptides were desalted and concentrated via StageTips. Peptides were separated on a 40-min gradient on a 15-cm analytical column (75 µm inner diameter) packed with 1.9 µm ReproSil Gold C18 reversed phase particles (Dr. Maisch). The column was maintained at 50°C, and peptides were directly sprayed into a Q Exactive HF mass spectrometer via a nano-electrospray ionization source. The mass spectrometer was operated in data-dependent mode, and ≤15 precursors were selected for fragmentation. Raw data were processed using MaxQuant (v1.6.0.15), and all identifications were filtered at 1% FDR.

Subconfluent cells were fixed with 2% PFA for 10 min, permeabilized with 0.2% Triton X-100 for 5 min, and blocked with PBS containing 2% BSA (Sigma-Aldrich) for ≥30 min. Next, cells were incubated with the primary antibodies for 1 h at room temperature. Cells were washed three times before incubation with the secondary antibodies for 1 h. Additionally, the nuclei were stained with DAPI, and F-actin was visualized using Alexa Fluor 488– or 647–conjugated phalloidin (BioLegend; AAT Bioquest). After three washing steps with PBS, the coverslips were mounted onto glass slides in Mowiol. Images were obtained at room temperature using a Leica TCS SP5 confocal microscope with a AOBS scan head (158001107; Leica), controlled using Leica LAS AF SP5 software (v2.7.4), with a 63×/1.4 Oil CS HC PL APO objective (11506350; Leica), with filter cubes for 488/eGFP (15525302; Leica), 568/mCherry (15525303; Leica), and DAPI (15525301; Leica).

GE11 Tet-ON integrin β1 cells (WT, PEAK1 KO, and PEAK1 KO stably expressing PEAK1-GFP, PEAK1-Y635F-GFP, or PEAK1-Y1188F-GFP) were cultured for ≥48 h with 1 µg/ml doxycycline to induce β1 integrin expression, and then plated onto fibronectin-coated (10 µg/ml in PBS) plastic 12-well plates (20,000 cells per well) in DMEM containing 0.5% FBS, antibiotics, and doxycycline (1 µg/ml). After culture overnight, the medium was exchanged for DMEM containing 0.5% FBS, antibiotics, and doxycycline (1 µg/ml), and cells were transferred to the heated stage (37°C with 5% CO₂) of a Zeiss AxioObserver Z.1 microscope. Phase-contrast images were acquired with a 10×/0.30 EC Plan-Neofluar Ph1 objective (Zeiss), captured with a Zeiss AxioCam MRm. Images were acquired every 10 min for 20 h. Cells were tracked manually using ImageJ; tracking data were analyzed using the Chemotaxis Tool plugin (Ibidi).

Image analysis was performed using Fiji (ImageJ; Schindelin et al., 2012; Schneider et al., 2012). To quantify Shc1 clustering in FAs (based on Paxillin staining), background was subtracted in both channels using a bilateral filter, and the region of interest was selected at the cell periphery. Colocalization of Shc1 and FAs was determined using the Image Calculator (command “multiply”) on both channels and calculating the area of overlapping fraction as a percentage of the total Shc1 area per cell using the Analyze Particle function. The amount of GFP-PEAK1 (WT or Y635F) present in FAs was calculated in the same manner, using vinculin staining as a marker for FAs.

Mann–Whitney U test (two-tailed P value) was performed using GraphPad Prism (v7.0c). In figures, statistically significant values are shown as ****, P < 0.0001. Graphs were made in GraphPad Prism and show all data points. Data distribution was assumed to be normal but was not formally tested.

Fig. S1, a table of proteins identified as interactors of both integrin β1 and β3 in the BioID experiments, shows surface expression of the integrin BirA* constructs, assessed using FACS. Fig. S2 is related to data showing that PEAK1 is recruited to Tensin3 at cell matrix adhesions via a process requiring phosphorylation driven by Src. Fig. S3 shows interactors of GFP-PEAK from GFP-Trap–based affinity purification followed by MS in cells treated with EGF. Tables S1, S2, S3, S4 and S5 are Excel files with the MS data for proximity interactors of Integrin β1-BirA* (Table S1, complete data; Table S2, top 60 interactors), Integrin β3-BirA* (Table S3, complete data; Table S4, top 60 interactors), and GFP-Trap based affinity purification MS of GFP-PEAK expressed in PEAK1-null SW480 cells (Table S5). Table S6 is a list of antibodies used in this study.

We thank Anne Cress (University of Arizona Cancer Center, Tucson, AZ), David Critchley (University of Leicester, Leicester, UK), Roger Daly (Monash University, Victoria, Australia), Marina Glukhova (Institut Curie, PSL Research University, CNRS, Paris, France), Simon Goodman (Merck KGaA, Darmstadt, Germany), Douglas Lowy (National Institutes of Health, Bethesda, MD), Jacques Neefjes (Leiden University Medical Centre, Leiden, The Netherlands), Ellen van der Schoot (Sanquin Research, Amsterdam, The Netherlands), and NKI colleagues for sharing cells and reagents.

This work was supported by the Dutch Cancer Society (project 12143). Liesbeth Hoekman and Onno B. Bleijerveld are supported by the Dutch NWO X-omics Initiative. A. Sonnenberg and R. Fässler thank the Alexander von Humboldt Foundation for supporting A. Sonnenberg during his sabbatical year at the Max-Planck Institute.

The authors declare no competing financial interests.

Author contributions: A. Zuidema and A. Sonnenberg conceived the study, designed experiments, interpreted data, and wrote the manuscript with contributions from P. Atherton. A. Zuidema, P. Atherton, M. Kreft, L. Hoekman, O.B. Bleijerveld, N. Nagaraj, N. Chen, and A. Sonnenberg performed and analyzed the experiments. R. Fässler provided resources and reviewed and edited the manuscript.