A Cas-BCAR3 co-regulatory circuit controls lamellipodia dynamics

Integrin adhesion complexes regulate cytoskeletal dynamics during cell migration. Adhesion activates phosphorylation of integrin-associated signaling proteins, including Cas (p130Cas, BCAR1), by Src-family kinases. Cas regulates leading-edge protrusion and migration in cooperation with its binding partner, BCAR3. However, it has been unclear how Cas and BCAR3 cooperate. Here, using normal epithelial cells, we find that BCAR3 localization to integrin adhesions requires Cas. In return, Cas phosphorylation, as well as lamellipodia dynamics and cell migration, requires BCAR3. These functions require the BCAR3 SH2 domain and a specific phosphorylation site, Tyr 117, that is also required for BCAR3 downregulation by the ubiquitin-proteasome system. These findings place BCAR3 in a co-regulatory positive-feedback circuit with Cas, with BCAR3 requiring Cas for localization and Cas requiring BCAR3 for activation and downstream signaling. The use of a single phosphorylation site in BCAR3 for activation and degradation ensures reliable negative feedback by the ubiquitin-proteasome system.


Introduction
Animal cells migrate by adhesive crawling or amoeboid blebbing (Trepat et al., 2012). During crawling, transmembrane receptors called integrins provide attachment to the extracellular matrix and organize the actin cytoskeleton (Bachir et al., 2017, Legate et al., 2009. Integrin engagement stimulates protrusion of a dynamic leading lamellipodium. Inside the lamellipodium, rearward flowing actin engages with integrin-associated proteins such as talin and vinculin, forming catch bonds, clustering the integrins, and recruiting additional regulatory and scaffold proteins to form transient structures called nascent adhesions (Case et al., 2015, del Rio et al., 2009, Galbraith et al., 2002, Partridge et al., 2006, Puklin-Faucher et al., 2009, Tadokoro et al., 2003. Most nascent adhesions are short-lived, but some mature into focal adhesions at the base of the lamellipodium, anchoring actin stress fibers and resisting the rearward actin flow to increase lamellipodium protrusion. While many aspects of cell migration can be explained by biomechanics, integrin adhesions also activate biochemical signaling molecules, including focal adhesion kinase (FAK), Src-family kinases (SFKs) and small GTPases (Burridge et al., 1992, Huang et al., 1993, Miyamoto et al., 1995, Schaller et al., 1992. Some FAK and SFKdependent phosphorylations regulate adhesion assembly (Pasapera et al., 2010, Stutchbury et al., 2017, Zaidel-Bar et al., 2007, while others coordinate adhesion with lamellipodium dynamics and other aspects of cell biology, such as cell survival (Mitra et al., 2006).
While the consequences of Cas phosphorylation are well-understood, it is less clear how Cas phosphorylation is activated. Cas phosphorylation requires integrin-dependent actin polymerization and an intact actin cytoskeleton (Vuori et al., 1996, Vuori et al., 1995, Zhang et al., 2014, Zhao et al., 2016. In vitro, Cas phosphorylation by SFKs is increased when the substrate domain is physically extended, suggesting that Cas may be a mechanosensor (Hotta et al., 2014, Sawada et al., 2006, Tamada et al., 2004. Cas contains N-and C-terminal domains that associate with integrin adhesions and could be involved in extending the substrate domain (Branis et al., 2017, Donato et al., 2010, Harte et al., 1996, Nojima et al., 1995, Vuori et al., 1995 (Figure 1a). At the N terminus, an SH3 domain binds adhesion proteins vinculin and FAK (Harte et al., 1996, Janostiak et al., 2014, Polte et al., 1995, while at the C-terminus, a CCH domain binds to adhesion proteins paxillin and ajuba (Pratt et al., 2005, Yi et al., 2002. In addition, the Cas SH3 domain binds to N-WASP via FAK, and N-WASP stimulates actin polymerization and Cas phosphorylation (Kostic et al., 2006, Zhang et al., 2014. Thus, Cas may be activated by actin flow or when the substrate domain is extended by forces acting on the Cas N-and C-terminal domains.
Cas is also subject to negative feedback. Phosphorylation of a YDYV sequence near the CCH domain mediates pY-Cas degradation . This phosphosite binds to the SH2 domain of SOCS6 (suppressor of cytokine signaling 6), recruiting CRL5 (Cullin 5-RING-ligase) and targeting Cas to the ubiquitin-proteasome system. SOCS6 co-localizes with pY-Cas in adhesions at the leading edge of migrating cells where it inhibits adhesion disassembly (Teckchandani et al., 2016). Mutation of the YDYV sequence or knockdown of SOCS6 or the CRL5 scaffold, Cullin 5 (Cul5), stabilizes Cas and increases adhesion disassembly, lamellipodia protrusion and ruffling (Teckchandani et al., 2016, Teckchandani et al., 2014. Thus, while YxxP phosphorylation activates Cas, YDYV phosphorylation provides negative feedback, restraining leading edge dynamics and stabilizing adhesions. Cas receives additional regulatory input from NSP family proteins, including BCAR3 (breast cancer anti-estrogen resistance 3, also called AND-34, NSP2, Sh2d3b) (Wallez et al., 2012).
Cas forms a complex with BCAR3, with the Cas CCH domain bound to a CDC25H domain in BCAR3 (Figure 1a). BCAR3 and Cas cooperate in many biological assays, with each protein requiring the other. For example, BCAR3 increases Cas phosphorylation and Cas-dependent membrane ruffling, adhesion disassembly, cell migration, cell proliferation (Cai et al., 2003, Cross et al., 2016, Oh et al., 2013, Riggins et al., 2003, Roselli et al., 2010, Schrecengost et al., 2007, Schuh et al., 2010. Like Cas, BCAR3 protects estrogen-dependent breast cancer cells from inhibitory actions of anti-estrogens (van Agthoven et al., 1998). However, it remains unclear how Cas and BCAR3 cooperate. Several lines of evidence suggest that BCAR3 may be regulated by tyrosine phosphorylation. First, it contains an SH2 domain that binds pY proteins, including the epidermal growth factor (EGF) receptor (Oh et al., 2008) and PTPRA (RPTP), a protein phosphatase that activates SFKs and Cas , von Wichert et al., 2003, Zheng et al., 2000. Second, BCAR3 tyrosine phosphorylation is stimulated by adhesion or serum, although the significance of the phosphorylation is unknown (Cai et al., 1999). These studies suggest that BCAR3 interacts with tyrosine kinases and is phosphorylated, but leave open whether BCAR3 phosphorylation is involved in Cas activation.
Here, we report that we detected BCAR3 in a screen for pY proteins that are down-regulated by CRL5 and the proteasome. We identified BCAR3 Y117 as a phosphorylation site that recruits SOCS6 and leads to CRL5-dependent degradation. In addition, we found that BCAR3 is needed for single-cell migration and invasion, and for the increased lamellipodial ruffling and collective migration of Cul5-deficient cells. Using gene silencing and mutant analysis, we find that Cas localizes to adhesions independent of BCAR3 but BCAR3 localization to adhesions requires association with Cas. In the adhesion, BCAR3 activates SFKs, Cas phosphorylation, membrane ruffling and cell migration, dependent on both Y117 and the SH2 domain. BCAR3 and Cas thus form a signaling hub that localizes to active integrins and coordinates actin dynamics under negative control by the ubiquitin-proteasome system.

CRL5 regulates BCAR3 protein turnover
We previously reported that CRL5 inhibits Src activity and Src-dependent transformation of MCF10A epithelial cells, in part by targeting pY proteins such as pYCas for degradation by the ubiquitin-proteasome system . Because overexpression of Cas alone did not phenocopy CRL5 inhibition , we infer that CRL5 downregulates additional pY proteins that become limiting when Cas is over-expressed. We sought to identify such pY proteins by screening for pY peptides whose abundance increases when Cul5 is inhibited. To this end, control and Cul5-deficient MCF10A cells were lysed under denaturing conditions, proteins were digested with trypsin, and peptides were labeled with isobaric TMT tags for quantitative pY proteomics (Zhang et al., 2007). In one experiment, samples were prepared from control and Cul5-deficient cells that were starved for epidermal growth factor (EGF) for 0, 24 or 72 hr. Starvation time had no systematic effect on peptide abundance, so, in a second experiment, we prepared biological triplicate samples from growing control and Cul5-  Table). We decided to focus on BCAR3 because Cul5 regulates Cas and Cas binds BCAR3 (Wallez et al., 2012).
The increased quantity of BCAR3 pY peptides in Cul5-deficient cells could result from increased phosphorylation, increased protein level, or both. We used immunoblotting to test whether BCAR3 protein level is regulated by Cul5. BCAR3 protein level increased approximately 4-fold in Cul5-deficient cells, under two different media conditions ( Figure 1c). However, RNA levels were unaltered (Figure 1c), consistent with altered protein synthesis or degradation. To measure degradation, we inhibited new protein synthesis with cycloheximide and monitored BCAR3 protein level. BCAR3 half-life was approximately 4 hours in control cells but greater than 8 hours in Cul5-deficient cells (Figure 1d). BCAR3 degradation was inhibited by Cullin neddylation inhibitor MLN4924 or proteasome inhibitor MG132, but not by lysosome inhibitor Bafilomycin (Fig 1e). These results suggest that CRL5 regulates BCAR3 turnover by the ubiquitin-proteasome system and that the increase in BCAR3 pY117 and pY266 in Cul5deficient cells is likely due to an increased availability of BCAR3 protein for phosphorylation rather than, or in addition to, an increase in kinase activity.

BCAR3 regulates epithelial cell migration
BCAR3 is required for the migration of cancer cells and fibroblasts in single-cell assays (Cross et al., 2016, Schrecengost et al., 2007 but its importance in single and collective epithelial cell migration is unknown. We inhibited BCAR3 expression in MCF10A cells using siRNA or BCAR3 gene disruption ( Moreover, inspection of the wound edge revealed that BCAR3 is also needed for the increased lamellipodia length and ruffling in Cul5-depleted cells (Figure 2e-g). This epistatic relationship is consistent with CRL5 inhibiting BCAR3-dependent migration and lamellipodia under collective conditions, as found before for Cas . We do not understand which CRL5 substrates regulate single cell migration, but can make use of single-cell assays to test the role of BCAR3 in normal cells, and collective assays to test the role of BCAR3 when it is over-expressed or activated by Cul5 depletion.

CRL5 directly targets BCAR3 through SOCS6
CRL5 promotes ubiquitination and degradation of substrate proteins bound to adaptor proteins (Okumura et al., 2016). CRL5 adaptors include SOCS family proteins, which contain SH2 domains for binding to pY. To test whether SOCS proteins bind BCAR3, we transiently expressed T7-tagged SOCS proteins and assayed binding to endogenous BCAR3 by immunoprecipitation and immunoblotting. BCAR3 specifically co-precipitated with SOCS6, the same adaptor that binds Cas ( Figure 3a) . Accordingly, SOCS6 loss of function in MCF10A cells increased BCAR3 steady-state protein levels and decreased the rate of BCAR3 turnover (Figure 3b,c). Together, these results suggest that CRL5 SOCS6 mediates BCAR3 turnover.
Since BCAR3 and Cas bind each other and both are bound and regulated by SOCS6, it is possible that SOCS6 binds BCAR3 indirectly, through Cas. We tested this possibility by two strategies (Figure 3d). First, we used a BCAR3 mutant, L744E/R748E, called here BCAR3 EE , that does not bind Cas (Wallez et al., 2014). This mutation inhibited binding to Cas but not SOCS6 ( Figure 3e). Second, we found that BCAR3-SOCS6 binding occurred in cells from which Cas had been depleted with siRNA ( Figure 3f). Collectively, these data suggest that SOCS6 binds to BCAR3 independently of Cas.

SOCS6 binds BCAR3 pY117
We considered that SOCS6 might bind BCAR3 through pY-dependent or -independent interactions. Pervanadate, a cell-permeable phosphotyrosine phosphatase inhibitor, increased BCAR3-SOCS6 association, suggesting pY dependence (Figure 4a). In addition, disrupting the SOCS6 SH2 domain by deletion (ΔC) or point mutation (R407K) inhibited binding to BCAR3 ( Figure 4b). This suggests that the SOCS6 SH2 domain binds pY-BCAR3. Serum or adhesion stimulates tyrosine phosphorylation of BCAR3 in mouse fibroblasts (Cai et al., 1999), but the specific sites have not been identified. High-throughput pY proteomics surveys have identified phosphorylation of BCAR3 at tyrosine residues 42, 117, 212, 266 and 429 in over 25 mouse and human cell lines (Hornbeck et al., 2019). We tested whether these sites were required for binding SOCS6 using site-directed mutagenesis ( Figure 4c). BCAR3 F5 , in which all five tyrosines were changed to phenylalanines, was unable to bind SOCS6 ( Figure 4d. By mutating each pY site individually, we found that Y117 is necessary to bind SOCS6 ( Figure 4d). Furthermore, mutating all sites except Y117 (42, 212, 266 and 429) had little effect on SOCS6 binding (BCAR3 F4 mutant, Figure 4e). These results support a model in which phosphorylation of BCAR3 at Y117 is both necessary and sufficient to bind the SOCS6 SH2 domain.

CRL5-dependent BCAR3 turnover requires Y117 and Cas association, but not the SH2 domain or other tyrosine residues
To test whether Y117 or other domains of BCAR3 are required for CRL5-dependent BCAR3 protein turnover, we measured the effect of various mutations on the level of tagged BCAR3 protein in control and Cul5-deficient cells. To avoid possible artifacts due to over-expression, we used a doxycycline-inducible promoter (Baron et al., 1997). MCF10A cells were first transduced to express the reverse tet transactivator (rtTA), and then transduced to express SNAP-V5tagged wildtype or mutant BCAR3 under control of the tet operator. Cells were treated with doxycycline (dox) to induce wildtype or mutant BCAR3 expression, with or without knocking down endogenous BCAR3 with an siRNA targeting the 3' UTR.
We first examined the role of Y117 in BCAR3 turnover. BCAR3 Y117F was expressed at approximately two-fold higher level than BCAR3 WT at the same concentration of dox ( Figure 5a). Moreover, depleting Cul5 increased the level of BCAR3 WT more than 2-fold while the level of BCAR3 Y117F was unchanged (Figure 5b), suggesting that CRL5 regulates BCAR3 protein level dependent on Y117. BCAR3 F4 , which contains Y117 but not four other tyrosine phosphorylation sites, was also regulated by CRL5 (Figure 5c). These results are consistent with SOCS6 binding to pY117 and targeting BCAR3 for CRL5-dependent degradation.
We extended this approach to test the importance of the BCAR3 SH2 domain and Cas binding for degradation. To inactivate the BCAR3 SH2 domain, we created an arginine to lysine at residue 177 (R177K), which lies in the consensus FLVRES motif and is required to bind the pY phosphate (Jaber Chehayeb et al., 2020, Marengere et al., 1994. Cul5-depletion increased the level of BCAR3 R177K (Figure 5d), suggesting this mutant is still phosphorylated at Y117 and targeted by CRL5. In contrast, BCAR3 EE , which binds SOCS6 (Figure 3e) but not Cas ( Figure   3e and Figure 5 figure supplement 1), was not regulated by CRL5. Taken together, these results suggest that pY117 and Cas association are required for CRL5-dependent turnover of BCAR3 expressed at near endogenous level.

Lamellipodial ruffling and cell migration require BCAR3 Y117, SH2 domain and Cas association
Over-expression of BCAR3 in fibroblasts and breast cancer cells stimulates Cas-dependent functions, such as lamellipodia ruffling (Cai et al., 2003, Wallez et al., 2014, Wilson et al., 2013. Similarly, BCAR3 WT increased membrane ruffling when over-expressed in MCF10A cells ( Figure   6a). Since BCAR3 Y117F accumulates to higher levels, we suspected it may be more active in biological assays. However, contrary to expectations, ruffling was induced by over-expressing BCAR3 F4 but not BCAR3 Y117F , BCAR3 R177K or BCAR3 EE (Figure 6a). Similarly, when endogenous BCAR3 is depleted, BCAR3 WT and BCAR3 F4 , but not BCAR3 Y117F , BCAR3 R177K or BCAR3 EE , were able to rescue single-cell migration ( Figure 6b) and increase collective migration and lamellipodia ruffling of Cul5-deficient cells ( Figure 6c). This suggests that Y117 has two roles.
First, Y117 is the main phosphorylation site for BCAR3 down-regulation. Second, it is also the main phosphorylation site for BCAR3 function, cooperating with the SH2 domain and Cas to promote single-cell and collective migration and lamellipodial dynamics in the presence and absence of Cul5.

Cas recruits BCAR3 to integrin adhesions
These findings raise the question of how BCAR3 Y117, SH2 domain, and Cas binding cooperate to regulate MCF10A cell motility. Previous studies in cancer cells and fibroblasts found that BCAR3 localizes to integrin adhesions (Cross et al., 2016 and that BCAR3 over-expression can increase Cas in membrane ruffles (Riggins et al., 2003). Increased Cas could then activate lamellipodia dynamics through the established Cas/Crk/DOCK180/Rac pathway (Klemke et al., 1998, Sakai et al., 1994, Sanders et al., 2005, Sharma et al., 2008. These findings suggest that BCAR3's Y117 and SH2 domain may be needed for BCAR3 to correctly localize Cas. To explore this possibility, we monitored the effect of depleting BCAR3 on Cas localization and the effect of depleting Cas on BCAR3 localization. We were unable to detect endogenous BCAR3 by immunofluorescence with available antibodies, so we expressed SNAP-V5-BCAR3 at near endogenous levels and detected the fusion protein with SNAP ligand (Grimm et al., 2015, Keppler et al., 2003. Both Cas and BCAR3 WT localized to adhesions in the leading edge of collectively migrating cells. Cas remained in adhesions when BCAR3 was depleted ( Figure 7a). In contrast, BCAR3 was absent from adhesions when Cas was depleted ( Figure 7b). Thus, BCAR3 requires Cas to localize in adhesions, and not vice versa.
Consistently, all BCAR3 mutants that bound Cas were present in adhesions (Figure 8a

BCAR3 activates Cas, dependent on BCAR3 Y117 and SH2 domain
Since BCAR3 does not regulate Cas localization, what is the function of Y117 and the SH2 domain? We considered that BCAR3 may activate Cas. To monitor Cas activity, we used antibodies to pY165, one of the repeated YxxP motifs in Cas that recruit Crk (Fonseca et al., 2004, Sakai et al., 1994, Songyang et al., 1993. Cas pY165 in leading edge adhesions was abolished by SFK inhibition or Cas depletion, consistent with Cas activity ( Cas activation in adhesions correlated with the rescue of ruffling and migration (Figures 6b and   8). This suggests that BCAR3 not only has to be bound to Cas but also needs Y117 and its SH2 domain to activate Cas and promote downstream signaling. To test whether BCAR3 activates Cas specifically or phosphorylation more generally, we also stained for FAK pY397 and pY861.
The former is a FAK autophosphorylation site and is phosphorylated by SFKs (Mitra et al., 2006). We found that BCAR3 depletion did not inhibit FAK autophosphorylation at pY397 but did inhibit FAK pY861 (Figure 8 - Figure Supplement 2b and c), consistent with a general role of BCAR3 in stimulating SFKs in adhesions. Taken together, these results suggest that BCAR3 is brought to integrin adhesions by binding to Cas, where it uses its Y117 and SH2 domain to activate SFKs, Cas and downstream signaling, leading to lamellipodial ruffling and migration.

Discussion
Cell migration requires biophysical and biochemical signaling between the leading edge and adhesion complexes to coordinate adhesion assembly and disassembly with actin polymerization, anchorage of stress fibers and generation of traction forces (Trepat et al., 2012).
Our results address one aspect of this complicated process: the linkage between integrin engagement and lamellipodial extension. Previous studies have found that Cas and BCAR3 synergize for lamellipodium ruffing and cell migration, but they also synergized for protein expression, making the mechanism unclear (Cai et al., 1999, Cross et al., 2016, Riggins et al., 2003, Schrecengost et al., 2007, Wallez et al., 2014. Here, using loss-of-function approaches, we find that there is positive feedback between Cas and BCAR3 phosphorylation, and dual negative feedback on both proteins by the ubiquitin-proteasome system. Our results support a multi-step model ( Figure 9). First, Cas associates with active integrin adhesion complexes near the leading edge of migrating cells through its N and C terminal localization signals. BCAR3 is recruited to adhesions by Cas. Since most BCAR3 in the cell is complexed with Cas (Cross et al., 2016), the two proteins are likely recruited jointly, via localization signals on Cas. BCAR3 then stimulates SFKs and Cas phosphorylation in a mechanism that requires BCAR3 SH2 and Y117 but not four other phosphorylation sites. Cas phosphorylation then drives membrane ruffling, presumably utilizing the well-documented mechanisms involving Crk, DOCK180 and Rac (Cheresh et al., 1999, Klemke et al., 1998, Sakai et al., 1994, Sanders et al., 2005, Schaller et al., 1995, Sharma et al., 2008. Thus, the BCAR3 Y117 and SH2 provide a critical switch to activate Cas. Our results also suggest that phosphorylation at Y117 promotes SOCS6 binding and targets BCAR3 for ubiquitination and degradation by the CRL5 E3 ubiquitin ligase. pY-Cas is targeted by CRL5 SOCS6 binding to its pYDYV motif . Ubiquitination likely occurs within the adhesion (Teckchandani et al., 2016). Ubiquitination and degradation of either protein is expected to terminate signaling until new Cas and BCAR3 molecules are recruited from the cytosol, providing double insurance against excess activity.
Where is Y117 phosphorylated and how does it activate SFKs? We suspect that phosphorylation may occur in adhesions, because BCAR3 EE , which doesn't localize to adhesions, is not subject to pY117-dependent turnover by CRL5. However, we have been unable to generate a phospho-specific antibody with the sensitivity and specificity to localize pY117 BCAR3 in cells. After Y117 is phosphorylated, we infer that BCAR3 activates SFKs in adhesions to phosphorylate Cas and FAK. Indeed, previous studies reported SFK activation by BCAR3 (Riggins et al., 2003, Schuh et al., 2010. One mechanism involves pY-PTPRA binding to the BCAR3 SH2 domain and bringing the BCAR3-Cas complex to adhesions where Cas is phosphorylated by SFKs . Since PTPRA is able to activate SFKs (Ponniah et al., 1999, Su et al., 1999, Zheng et al., 2000, this is an attractive model. However, in our studies, the BCAR3 SH2 domain is not needed to localize either BCAR3 or Cas, suggesting that Cas-BCAR3 would bring PTPRA to adhesions rather than vice versa. In addition, PTPRA depletion does not inhibit Cas Y165 phosphorylation (data not shown). An alternative is that pY117 binds the SH2 domain of an SFK or an unidentified protein that activates SFKs. The Y117 sequence is conserved across vertebrates and a homologous residue is present in Shep1, a second NSP family protein that activates Cas (Roselli et al., 2010). However, the sequence does not suggest which SH2 domain may bind other than SOCS6  ( Figure 9 -Figure Supplement 1). The presence of a positively charged residue two residues after the phosphosite would inhibit binding to SFKs (Songyang et al., 1993). Therefore, pY117 may bind another SH2 protein that activates SFKs, or may stimulate Cas phosphorylation by an allosteric mechanism or by altering binding to other charged molecules, such as membrane phospholipids.
Our results reveal that BCAR3 Y117 is required for both signaling and degradation. This means that BCAR3 activation triggers BCAR3 degradation in a negative feedback loop. The dual use of a single site for activation and degradation resembles Y198 of the neuron migration protein Dab1, which binds downstream signaling molecules and also recruits SOCS6/7 for CRL5dependent degradation . Mutating the SOCS-binding site in BCAR3 or Dab1 generates a non-functional protein that accumulates at higher level. In contrast, the SOCS6binding site in Cas is not needed for downstream signaling, so mutating the site causes increased levels of an active protein and a dominant gain-of-function phenotype .
Our finding that BCAR3 is subject to complex post-translational regulation may be helpful in reconciling the observation that BCAR3 over-expression in cell culture increases transformed phenotypes whereas BCAR3 RNA expression correlates with favorable outcomes in patients (Guo et al., 2014, Wallez et al., 2012, Zhang et al., 2018. In addition to the phosphorylationdependent mechanism investigated here, a previous study showed that TGFβ stimulates proteasomal degradation of BCAR3 (Guo et al., 2014), and over-expressed BCAR3 and Cas appear to stabilize each other in breast cancer cells, dependent on their mutual binding (Wallez et al., 2014). Changes in BCAR3 protein stability or phosphorylation state might influence BCAR3 activity in patients, but that such regulation at the protein level might not be accounted for when correlating patient outcomes with gene expression. Thus, further study is required to understand mechanisms of BCAR3 regulation and how mis-regulation may contribute to disease progression.
MCF10A empty vector (EV) and shCul5 cells were made as previously described . MCF10A shControl and shSOCS6 cells were made as follows.
Validation of BCAR3 knockout was done through genomic DNA isolation, PCR, and sequencing, as well as Western blotting.
MCF10A dox-inducible SNAP-V5 BCAR3 cells were made as follows. Viruses containing rtTA-N144 were packaged using 293T cells and used to infect MCF10A cells. A stable line was selected using 50 µg/mL hygromycin. The MCF10A cell line stably expressing rtTA was then infected with viruses containing one of the pLTRE3G-SNAP-V5 constructs (pLTRE3G-SNAP-V5-eGFP, pLTRE3G-SNAP-V5-BCAR3 WT, Y42F, Y117F, Y212F, Y266F, Y429F, F5, F4, R177K, or EE). Stable lines were selected using 10 µg/mL blasticidin. Each cell line was hygromycin resistant and blasticidin resistant, however not all cells within each line expressed the SNAP-V5-containing construct following induction with 50-100ng/mL dox for 48-72 hours. To remove non-inducible cells, dox-treated cells were treated with 100 nM JaneliaFluor cp646JF-SNAP-ligand (Lavis Lab, Janelia Farms) (Grimm et al., 2015) for 1 hour, washed three times with PBS, incubated in ligand-free growth media for an hour and positive cells were sorted and harvested by FACS using the APC channel to detect cp646JF.

Mass spectrometry
MCF10A EV and shCul5 cells were grown in 15-cm plates, two plates per condition, to approximately 80% confluence, washed in PBS, and lysed in 3 mL per plate of ice-cold 8 M urea containing 1 mM sodium orthovanadate. Proteins were reduced, alkylated and digested with trypsin as described (Zhang et al., 2005). Peptide labeling with iTRAQ reagents (TMT variable modifications were methionine oxidation, tyrosine, serine and threonine phosphorylation. TMT quantification was obtained using Proteome Discoverer and isotopically corrected per manufacturer's directions, and normalized to the mean relative protein quantification ratios obtained from the total protein analysis (0.2% of the unbound peptides from the phosphotyrosine peptide immunoprecipitation). Mascot peptide identifications, phosphorylation site assignments and quantification were verified manually with the assistance of CAMV (Curran et al., 2013). Validated data were analyzed in Excel using a one-sided twosample equal variance Student t-test.

siRNA Transfection
MCF10A cells were trypsinized and seeded in growth medium into a 12-well plate so as to be 50% confluent after attaching. A mixture of 50 pmol siRNA, 1.25 l Lipofectamine 2000 (Invitrogen) in Optimem medium (Invitrogen) was added to the newly plated MCF10A cells and left for the cells to attach overnight. The next day the media was changed to fresh growth media. A second siRNA transfection was done 48 hours after the first transfection using the same protocol scaled up to use a 6-well plate (125 pmol siRNA) or scaled down to use a 96-well plate (for migration assays). One day later, cells from the 6-well dish were transferred to 4-well plates containing 12-mm coverslips for microscopy, or 12-well plates for protein analysis.
Doxycline was added as needed at 10 ng/mL for 2-3 days before analysis. Cells were either lysed 48 hours after the second transfection or assays were performed as described.
Lysates were adjusted to SDS sample buffer, heated to 95 o C, resolved by SDS-PAGE using 10% polyacrylamide/0.133% bis-acrylamide gels, and transferred onto nitrocellulose membrane.
The membrane was blocked in Odyssey blocking buffer (TBS) (LI-COR Biosciences) with 5% BSA for phosphotyrosine antibodies or 5% non-fat dry milk for all other antibodies. Following blocking, the membrane was probed with a primary antibody followed by IRDye 800CW goat anti-rabbit or 680RD goat anti-mouse conjugated secondary antibodies. Membranes were visualized using the Odyssey Infrared Imaging System (LI-COR Biosciences). Bands were quantified using ImageJ.

DNA transfections and Immunoprecipitation
HeLa cells were plated in 6-well plates the day before transfection such that cells were 50% confluent on the day of transfection. A mixture of DNA, Lipofectamine 2000 (Thermo Fisher Scientific) and Optimem (Invitrogen) was made according to manufacturer's protocol, added to the cells and removed after 5 hours.
Immunoprecipitation experiments were conducted 24-48 hours after transfection. For indicated experiments, cells were incubated with 1 mM sodium pervanadate for 30 minutes. Cells were lysed on ice in X-100 buffer with fresh protease and phosphatase inhibitors (see above).
Lysates were cleared by centrifugation for 10 minutes at 14,000 g. Lysates were rotated with 1 g antibody at 4C for 3 hours, after which Protein A/G plus agarose beads (Santa Cruz Biotechnology) were added for 1 hour at 4C. Beads were collected by centrifugation and washed three times with X-100 buffer. The beads were resuspended in SDS sample buffer, boiled, lightly mixed to release bound protein and centrifuged. Immunoprecipitation samples were resolved by SDS-PAGE using 10% polyacrylamide gels and bound proteins detected by Western blotting as above. Samples of total cell lysate typically contained 5% of the protein used for immunoprecipitation.

Cycloheximide chase
Cells were grown to 80% confluency and treated with 25g/mL cycloheximide by adding it directly to the conditioned media on the cells. Cells were treated with cycloheximide for 0, 2, 4, or 8 hours and lysed. Quantification of Western blots was done in ImageJ and BCAR3 protein levels were normalized to the loading control (GAPDH or vinculin).

Scratch wound assay
The desired proteins were knocked down using two siRNA transfections, as previously described. Cells were plated in ImageLock 96-well dishes (Essen BioScience) for the second siRNA transfection and transfection materials were scaled down proportional to dish sizes. Cells were plated at 30% confluence at the time of the second transfection. For migration assays where doxycycline (dox)-induction was required, the media was replaced with growth media containing dox (10 ng/mL except where noted) 6 hours after transfection and after two washes with PBS. For all migration assays, the confluent monolayers were placed in EGF-free assay media (with dox when required) 48 hours after the second transfection transfection. Monolayers were scratched using an Incucyte WoundMaker (Essen BioScience) 8 hours after being placed in assay media, the wells were washed with PBS three times to remove debris and cells were placed back in assay media (with dox when required). Scratch wounds were imaged once every 2 hours on an IncuCyte S3 and images were analyzed using the scratch wound function on the IncuCyte image analysis software. Overall migration was measured at 12 hours using the relative wound density calculated by the analysis software. Lamellipodium length was measured using the ruler in the IncuCyte image analysis software. Membrane ruffling was scored at 6 hours by counting the number of cells with ruffles relative to the total cell number. Ruffles were visualized as dark contrast at the front of the protrusion.

Transwell migration and invasion assays
Cells were grown in EGF-free assay media for 24 hours before the assay. Migration assays were performed in 24-well chemotaxis chamber with an 8 m pore size polyethylene terephthalate filter that separated the upper and lower wells (Thermo Fisher Scientific). Invasion assays were performed in 24-well Matrigel invasion chambers with an 8 m pore size polyethylene terephthalate filter coated in Matrigel matrix that separated the upper and lower wells (Thermo Fisher Scientific). In both assays, the lower wells were filled with MCF10A growth media. 80,000 cells were resuspended in EGF-free assay media and added to the top well.
After 24 hours, cells on the top of the membrane were removed and the cells on the bottom of the membrane were fixed with methanol and stained with SybrSafe at 1:10,000 for 15 minutes.
Membranes were rinsed, imaged and nuclei were counted.

Immunofluorescence
For experiments that required protein knockdown, cells were transfected with siRNA as previously described. On the day following the second siRNA transfection, cells were plated on 12-mm #1.5 coverslips at 30% confluency in 4-well (1.9 cm 2 ) plates (with dox when required).
Confluent monolayers were transferred to assay media and incubated overnight before scratching with a P200 pipette tip. To detect the SNAP tag, JaneliaFluor cpJF549 SNAP-ligand (Lavis Lab, Janelia Farms) (Grimm et al., 2015) was added to the cells at 100 nM 4 hours after forming the wounds. Cells were incubated for 1 hour with the SNAP-ligand, washed three times with PBS and placed in fresh, ligand-free assay media for 1 hour to remove unreacted SNAPligand. Cells were fixed and permeabilized a total of 6 hours after the start of migration. Antifade Mountant and left to cure overnight in the dark before imaging or storing in the cold.

Imaging and image quantification
Coverslips were imaged using 63x, 1.40 NA or 100x, 1.40 NA oil objectives on a Leica SP8 confocal microscope and deconvolved using Leica LAS acquisition software. In order to ensure Image quantification was done in ImageJ. A line was drawn ~6 µm from the front of the migrating cells, as described (Teckchandani et al., 2016), to isolate the leading edge. Using the threshold tool on ImageJ, paxillin-containing adhesions were identified and mean intensity of SNAP or other antigens (pY-Cas, Cas, pY397-FAK, pY-861-FAK and paxillin) was measured.
The relative abundance of each antigen in adhesions was calculated by dividing the mean antigen intensity by the mean intensity of paxillin. Because expression of SNAP-V5-BCAR3 varied slightly cell by cell, we corrected for expression level as follows. SNAP intensity was measured in the leading edge adhesions (SNAPad), at the rear of the cell (SNAPrear) and in nondox induced cells(SNAPbg). Corrected SNAP intensity in adhesions (SNAPad-SNAPbg)/(SNAPrear-SNAPbg) was divided by the mean paxillin intensity.         Step 1: A preformed BCAR3-Cas complex translocates to integrin adhesions through localization signals in the N and C terminal domains of Cas.
Step 2: BCAR3, phosphorylated at Y117 by an unknown kinase, stimulates Cas phosphorylation in a process that also requires the SH2 domain, leading to activation of signaling pathways that stimulate lamellipodial protrusion and ruffling.
Step 3: Signaling is terminated by the CRL5 SOCS6 -dependent ubiquitination and degradation of BCAR3, phosphorylated in their respective phosphodegrons. Remaining questions include the identity of the kinase that phosphorylates BCAR3 Y117, the timing of Y117 phosphorylation, and the mechanism by which BCAR3 pY117 and SH2 domain stimulate Cas phosphorylation.

Supplementary Material
Supplementary Table 1. Quantification of phosphotyrosine peptides in control and Cul5deficient MCF10A cells. Phosphopeptides whose abundance significantly differed between control and Cul5-deficient MCF10A cells in two separate experiments. Sheet 1: Experimental conditions, numbers of phosphopeptides quantified, and number of significant changes. Sheet 2: Phosphopeptides significantly increased in Experiment 1. Sheet 3: Phosphopeptides significantly increased in Experiment 2. Sheet 4: Phosphopeptides significantly increased in both Experiments (see also Figure 1 - Figure Supplement 1).