We previously reported that CRL5 inhibits Src activity and Src-dependent transformation of MCF10A epithelial cells, in part by targeting pY proteins such as pY-Cas for degradation by the ubiquitin-proteasome system (Teckchandani et al., 2014). Because overexpression of Cas alone did not phenocopy CRL5 inhibition (Teckchandani et al., 2014), we infer that CRL5 down-regulates 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-deficient cells. Sixteen pY peptides increased significantly in Cul5-deficient cells in both experiments, including pY128 from Cas and pY117 and pY266 from BCAR3 (Figure 1b, Table 1, Supplementary File 1). 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 fourfold 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 hr in control cells but greater than 8 hr 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 (Figure 1e). These results suggest that CRL5 regulates BCAR3 turnover by the ubiquitin-proteasome system and that the increase in BCAR3 pY117 and pY266 in Cul5-deficient 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 is required for the migration of cancer cells and fibroblasts in single-cell assays (Cross et al., 2016; Schrecengost et al., 2007; Sun et al., 2012) 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 (Figure 2a, Figure 2—figure supplement 1a). BCAR3-deficient cells migrated slower than control cells in single-cell migration and invasion assays, regardless of Cul5, suggesting that BCAR3 and CRL5 regulate single-cell migration independently (Figure 2b,c, Figure 2—figure supplement 1b and c). In contrast, in a collective migration scratch wound assay, BCAR3 was not required unless Cul5 was depleted (Figure 2d). 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 (Teckchandani et al., 2014). We do not understand the differences between single-cell and collective 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 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; Teckchandani et al., 2014). Accordingly, SOCS6 depletion 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.

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.

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-V5-tagged 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 twofold higher level than BCAR3^WT at the same concentration of dox (Figure 5a). Moreover, depleting Cul5 increased the level of BCAR3^WT more than twofold while the level of BCAR3^Y117F was unchanged (Figure 5b). This suggests 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 site 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 and Boggon, 2020; Marengere and Pawson, 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.

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.

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; Sun et al., 2012) 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 and Basson, 2005; Sharma and Mayer, 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,b, Figure 5—figure supplement 1). Together, these results suggest that Cas localizes to adhesions independent of BCAR3 and that Cas brings BCAR3 to adhesions by direct binding.

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 (Figure 8—figure supplement 1). Notably, Cas activity was inhibited when BCAR3 was depleted (Figure 8—figure supplement 2a). As expected, BCAR3^WT and BCAR3^F4 rescued Cas activity in BCAR3-depleted cells (Figure 8a,c). However, Cas activity was not rescued by BCAR3^Y117F, BCAR3^R177K, or BCAR3^EE, even though Cas was still present (Figure 8a,c, Figure 8—figure supplement 3). 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 the latter is phosphorylated by SFKs (Mitra and Schlaepfer, 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.

pMXpuroII and pMXpuroII-shCul5 plasmids were made as previously described (Teckchandani et al., 2014). pLKO.1-nontarget small hairpin RNA (shRNA) control vector (SHC002) and pLKO.1-shSOCS6 (TRCN0000284351) were purchased (Sigma Aldrich). pCAG-T7-mSOCS1, pCAG-T7-mSOCS2, pCAG-T7-mSOCS3, pCAG-T7-mSOCS4, pCAG-T7-mSOCS5, pCAG-T7-mSOCS6, pCAG-T7-mSOCS7, and pCAG-T7-mCisH were made as previously described (Simó and Cooper, 2013; Teckchandani et al., 2014). All pCAG-T7-SOCS plasmids used in this work have LC-QQ mutations in the SOCS box to prevent binding to CRL5, and were made using PfuTurbo DNA polymerase to perform site-directed mutagenesis followed by DpnI digestion. pCAG-T7-mSOCS6 R407K was made as previously described (Teckchandani and Cooper, 2016). pCAG-T7-mSOCS6 ΔC was made by PCR amplifying codons 1–381 of mSOCS6 using a 5’ primer complementary to the T7 tag and a 3’ primer that inserts a stop codon and a NotI site after codon 381. The PCR product was inserted into pCAG-T7-SOCS6 by BamH1/NotI restriction digest and ligation.

pCMV-SPORT6-mBCAR3 was purchased from the Harvard Plasmid Database (MmCD00318547). pcDNA5-3xFlag-mBCAR3 FRT/TO was made by Gateway cloning. mBCAR3 was PCR amplified with flanking attB sites from pCMV-SPORT6-mBCAR3 and inserted into pDONR221 with a BP reaction (Thermo Fisher Scientific). pDEST-5’−3xFlag-pcDNA FRT/TO was a kind gift of Dr. Anne-Claude Gingras. pDONR221-mBCAR3 was moved into pDEST-5’−3xFlag-pcDNA FRT/TO with an LR reaction (Thermo Fisher Scientific). pLX304-hBCAR3-V5 was obtained from the human ORFeome version 8.1 Entry clone collection (Yang et al., 2011). pLX304-hBCAR3-F5-V5 was made by Gibson Assembly (New England Biolabs). A gBlock (Integrated DNA Technologies) containing BCAR3 Y42F, Y117F, Y212F, Y266F, and Y429F mutations was assembled into pLX304-hBCAR3-V5.

rtTA-N144 was a gift from Andrew Yoo (Addgene plasmid # 66810). pLTRE3G-SNAP-V5-hBCAR3 was made as follows. A gBlock containing sequence for SNAP-V5 tags was inserted into pLenti CMVTRE3G eGFP Blast (w818-1) (gift of Eric Campenau, Addgene #27568) at the AgeI restriction site using Gibson Assembly to make pLTRE3G-SNAP-V5-eGFP. hBCAR3 was PCR amplified with flanking AgeI and XbaI sites and inserted into AgeI and XbaI-digested pLTRE3G-SNAP-V5-eGFP to make pLTRE3G-SNAP-V5-hBCAR3. pLTRE3G-SNAP-V5-hBCAR3 mutants were made as follows. Single mutants Y42F, Y212F, Y266F, Y429F, R177K, and double mutant EE (L744E/R748E) were made by site-directed mutagenesis with Q5 polymerase (New England Biolabs) and Dpn1 digestion. The Y117F mutant was made by Gibson Assembly. A gBlock containing the BCAR3 Y117F mutation was assembled into pLTRE3G-SNAP-V5-hBCAR3 WT. pLTRE3G-SNAP-V5-hBCAR3 F5 was made by BamHI restriction digest of pLTRE3G-SNAP-V5-hBCAR3 Y42F, discarding the region between codon 107 and 763 and inserting the corresponding region from pLX304-hBCAR3-F5-V5. pLTRE3G-SNAP-V5-hBCAR3 F4 was made by PshAI restriction digest of pLTRE3G-SNAP-V5-hBCAR3 F5, discarding the region between codon 67 and 163 inserting the corresponding region from pLTRE3G-SNAP-V5-hBCAR3 WT.

Parental cell lines were originally obtained from ATCC and confirmed by DNA fingerprinting using short tandem repeat, or STR, typing. Cell lines were screened periodically for mycoplasma using the MycoProbe kit from R and D Systems.

MCF10A cells were cultured in growth medium (DMEM/F12 (Thermo Fisher Scientific) supplemented with 5% horse serum (Thermo Fisher Scientific), 0.1 μg/ml cholera toxin (EMD Millipore), 10 μg/ml insulin (Thermo Fisher Scientific), 0.5 μg/ml hydrocortisone (Sigma-Aldrich), and 20 ng/ml EGF (Thermo Fisher Scientific)). Indicated experiments used assay media (DMEM/F12, 2% horse serum, 0.1 μg/ml cholera toxin, 10 μg/ml insulin, 0.5 μg/ml hydrocortisone, and 0 ng/ml EGF). The cells were passaged using 0.02% trypsin in PBS/EDTA. Trypsin was inactivated with an equal volume of DMEM/10% FBS and cells were harvested by centrifugation before resuspending in growth media. HeLa and 293 T cells were grown in DMEM/10% FBS and passaged with trypsin as above.

MCF10A empty vector (EV) and shCul5 cells were made as previously described (Teckchandani et al., 2014). MCF10A shControl and shSOCS6 cells were made as follows. Viruses containing pLKO.1-control and pLKO.1-shSOCS6 were packaged using HEK 293 T cells and MCF10A cells were infected. Stable cell lines were selected using 4 µg/mL puromycin.

BCAR3-/- MCF10A cells were made as follows. Early passage MCF10A cells were serially diluted and subclones were grown and characterized. Clone J8 was selected for its epithelial morphology. The J8 clonal cell line was infected with lentiCRISPR v2 (a gift from Feng Zhang; Addgene plasmid # 52961), lacking or containing guide RNA against BCAR3. pLCRISPRv2-sgRNA-hBCAR3(30) and (31) were made by cloning annealed oligos (5’-CACCGTCAGAGAGCTACCTGCCGAT-3’ or 5’-CACCGCCCGAAACATACCAATCGGC-3’) into the BsmBI site of plentiCRISPRv2. Potential knockouts were isolated by single cell expansion. 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 293 T 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 50–100 ng/mL dox for 48–72 hr. 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 hr, 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.

Antibody	Supplier		Catalog	Dilution
Rabbit anti-BCAR3	Bethyl	A301-671A-M		Western (1:2000)
Rabbit anti-Cas (C-20)	Santa Cruz	sc-860		Western (1:1000)
Mouse anti-Cas	BD Biosciences	610271		Western (1:5000)
Mouse anti-Cas	Santa Cruz	sc-20029		IF (1:100)
Rabbit anti-pY165 Cas	Cell Signaling	4015S		IF (1:200)
Rabbit anti-Cullin5	Abcam	ab184177		Western (1:1000)
Rabbit anti-pY397 FAK	Fisher	44–624G		IF (1:200)
Rabbit anti-pY861 FAK	Fisher Scientific	44–626G		IF (1:200)
Rabbit anti-GAPDH (0411)	Santa Cruz	sc-47724		Western (1:1000)
Mouse anti-vinculin	Sigma	V9131		Western (1:10000)
Mouse anti-paxillin	BD Biosciences	610051		IF (1:200)
Sheep anti-paxillin	R and D Systems	AF4259		IF (1:200)
Mouse anti-FLAG	Sigma	F1804		Western (1:1000)
Mouse anti-T7	EMB Biosciences	69522–4		Western (1:5000) IP
Rabbit anti-V5	Bethyl	A190-120A		Western (1:5000) IP
AlexaFluor 488 goat anti-rabbit IgG (H+L)	Invitrogen	A11008		IF (1:1000)
AlexaFluor 647 goat anti-mouse IgG (H+L)	Invitrogen	A28181		IF (1:1000)
AlexaFluor 647 goat anti-sheep IgG (H+L)	Invitrogen	A21448		IF (1:1000)

Reagents used: cycloheximide (Sigma), epidermal growth factor (Invitrogen), MLN4924 (Active Biochem), MG132 (Fisher Scientific), bafilomycin A, doxycycline (Fisher Scientific).

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 6plex, Themo Fisher Scientific) was performed according to the manufacturer’s instructions, using 400 μg of each sample. After reaction, the six samples were combined and concentrated under vacuum. For immunoprecipitation, protein G agrarose (60 μl, Millipore) was incubated with anti-phosphotyrosine antibodies (12 μg 4G10 [Millipore], 12 μg PT66 [Sigma] and 12 μg PY100 [Cell Signaling Technology] for 8 hr at 4°C with rotation, then washed with 400 μl IP buffer [100 mM Tris HCl, 100 mM NaCl, and 1% Nonidet P-40, pH 7.4]). The TMT-labeled peptides were resuspended in 400 μl IP buffer and pH adjusted to 7.4, then mixed with the antibody-conjugated protein G agarose overnight at 4°C with rotation. The antibody beads were spun down and the supernatant saved. The beads were washed with 400 μl IP buffer then four rinses with 100 mM Tris HCl pH 7.4 before elution with 70 μl of 100 mM glycine, pH 2.0 for 30 min at 25°C. Offline immobilized metal affinity chromatography (IMAC) was used to further enrich for phosphotyrosine peptides (Zhang et al., 2005).

Peptides were loaded on a precolumn and separated by reverse phase HPLC using an EASY-nLC1000 (Thermo) over a 140 min gradient from 100% solvent A (H₂O/CH₃COOH, 99:1 (v/v)) to 30% A: 70% B (H₂O/CH₃CN/CH₃COOH, 29:70:1 (v/v)). Eluted peptides were injected using nanoelectrospray into a QExactive Pluss mass spectrometer (Thermo) in data-dependent mode. The parameters for full scan MS were: resolution of 70,000 across 350–2000 m/z, AGC 3e⁶, and maximum IT 50 ms. The full MS scan was followed by MS/MS for the top 10 precursor ions in each cycle with a NCE of 34 and dynamic exclusion of 30 s. Raw mass spectral files (.raw) were searched using Proteome Discoverer (Thermo) and Mascot version 2.4.1 (Matrix Science). Mascot search parameters were: 10 ppm mass tolerance for precursor ions; 15 mmu for fragment ion mass tolerance; two missed cleavages of trypsin; fixed modifications were carbamidomethylation of cysteine and TMT 6plex modification of lysines and peptide N-termini; 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 two-sample 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 μl 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 hr after the first transfection using the same protocol scaled up to use a six-well plate (125 pmol siRNA) or scaled down to use a 96-well plate (for migration assays). One day later, cells from the six-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 hr after the second transfection or assays were performed as described.

siRNA	Target sequence	Source	Notes	Catalog
Control siRNA	5’-AATTCTCCCGAACGTGTCACGT-3’	Qiagen		1027310
SOCS6 siRNA Pool	5’-TAGAATCGTGAATTGACATAA-3’ 5’-CAGCTGCGATATCAACGGTGA-3’ 5’-TTGATCTAATTGAGCATTCAA-3’ 5’-CGGGTACAAATTGGCATAACA-3’	Qiagen		SI00061383 SI03068359 SI00061369 SI00061376
Cul5 siRNA Pool	5’-GACACGACGTCTTATATTA-3’ 5’-CGTCTAATCTGTTAAAGAA-3’ 5’-GATGATACGGCTTTGCTAA-3’ 5’-GTTCAACTACGAATACTAA-3’	GE Dharmacon		M-019553-01-0005
BCAR1 siRNA Pool	5’-AAGCAGTTTGAACGACTGGAA-3’ 5’-CTGGATGGAGGACTATGACTA-3’ 5’-CAGCATCACGCGGCAGGGCAA-3’ 5’-CAACCTGACCACACTGACCAA-3’	Qiagen		SI02757741 SI02757734 SI04438280 SI04438273
BCAR3 siRNA Pool	5’-CCGGAACTCTGGCGTCAACTA-3’ 5’-CCGAGCGGCCACTCTGAGTAA-3’ 5’-GCCCAACGAGTTTGAGTCAAA-3’ 5’-AAGGTATCAGTTATATGATAT-3’	Qiagen		SI03081603 SI03080196 SI00053102 SI00053095
BCAR3 siRNA	5’-GGUAACUACUGCUAAUGUUTT-3’	Life Technologies	Targets 3’UTR	AM16708

An RNeasy Plus Mini kit (Qiagen) was used to extract RNA and an iScript reverse transcription supermix (BioRad) was used to make cDNA. Control reactions lacked reverse transcriptase. cDNA abundance was measured using an iTaq Universal SYBR Green Supermix kit (BioRad). Samples were run on the QuantStudio 5 Real-Time PCR System.

Primer	Sequence
BCAR3 forward	5’ – AATCGCTTCTCCAAACAGAGC – 3’
BCAR3 reverse	5’ – ATTCACCGGCATGTTTCTGG – 3’
Cul5 forward	5’ – TTTTATGCGCCCGATTGTTTTG – 3’
Cul5 reverse	5’ – TTGCTGGGCCTTTATCATCCC – 3’
GAPDH forward	5’ – CAGCCTCAAGATCATCAGCA – 3’
GAPDH reverse	5’ – TGTGGTCATGAGTCCTTCCA – 3’

Cells were washed three times in phosphate-buffered saline (PBS) before lysis. Cells were lysed in X-100 buffer (1% Triton X-100, 150 mM NaCl, 10 mM HEPES pH 7.4, 2 mM EDTA, 50 mM NaF) or RIPA buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 20 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 5 mM EGTA) with fresh protease and phosphatase inhibitors (10 μg/mL Aprotinin, 1 mM PMSF, 1 mM sodium vanadate) added before use.

Lysates were adjusted to SDS sample buffer, heated to 95°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.

HeLa cells were plated in six-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 hr.

Immunoprecipitation experiments were conducted 24–48 hr after transfection. For indicated experiments, cells were incubated with 1 mM sodium pervanadate for 30 min. Cells were lysed on ice in X-100 buffer with fresh protease and phosphatase inhibitors (see above). Lysates were cleared by centrifugation for 10 min at 14,000 g. Lysates were rotated with 1 μg antibody at 4°C for 3 hr, after which Protein A/G plus agarose beads (Santa Cruz Biotechnology) were added for 1 hr 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.

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

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 hr 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 hr after the second transfection transfection. Monolayers were scratched using an Incucyte WoundMaker (Essen BioScience) 8 hr 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 hr 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 hr 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 hr 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.

Cells were grown in EGF-free assay media for 24 hr before the assay. Migration assays were performed in 24-well chemotaxis chamber with an 8 μm 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 μm 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. A total of 80,000 cells were resuspended in EGF-free assay media and added to the top well. After 24 hr, 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 min. Membranes were rinsed, imaged and nuclei were counted.

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 four-well (1.9 cm²) 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 hr after forming the wounds. Cells were incubated for 1 hr with the SNAP-ligand, washed three times with PBS and placed in fresh, ligand-free assay media for 1 hr to remove unreacted SNAP-ligand. Cells were fixed and permeabilized a total of 6 hr after the start of migration.

Fixation and immunostaining was done at room temperature. Cells were washed with PBS, fixed and permeabilized with 0.1% Triton X-100, 4% paraformaldehyde (PFA) in PBS for 2 min, and further fixed with 4% PFA/PBS for 15 min. Cells were blocked with 2% BSA, 5% normal goat serum in PBS for 30 min or overnight at 4°C. Primary antibodies were diluted in block solution (typically 1:200 dilution) and added for 3 hr. Coverslips were washed three times with PBS. AlexaFluor-labeled secondary antibodies, diluted 1:1000 in PBS, were added for 1 hr. Coverslips were washed three times with PBS, mounted in Prolong Glass Antifade Mountant and left to cure overnight in the dark before imaging or storing in the cold.

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 unbiased data collection, 8–14 cells along the leading edge of the wound were imaged in the paxillin channel before imaging channels showing other antigen(s) of interest. The same scale and image settings were used for all conditions within an experiment. The number of cells and biological replicates is provided in each figure legend.

Image quantification was done in ImageJ. A line was drawn ~6 µm from the front of the migrating cells, as described (Teckchandani and Cooper, 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 (SNAP_ad) and rear (SNAP_rear) of dox-induced cells and in non-induced cells (SNAP_bg). Corrected SNAP intensity in adhesions (SNAP_ad-SNAP_bg)/(SNAP_rear-SNAP_bg) was divided by the mean paxillin intensity.