SRC and ERK cooperatively phosphorylate DLC1 and attenuate its Rho-GAP and tumor suppressor functions

The SRC gene is the prototypic member of the SRC family kinases (SFKs), whose nine members encode nonreceptor tyrosine-protein kinases that share a similar structure and have key roles in normal physiology and cancer (Sen and Johnson, 2011). Three of the SFKs—SRC, FYN, and YES—are widely expressed. Despite their partial genetic redundancy, inactivation of any two of these genes results in severe developmental defects (Lowell and Soriano, 1996). In neoplasia, where the demands of abnormal growth may be greater than during normal development, the requirement for endogenous WT SRC alone for full oncogenicity can be readily demonstrated (Guy et al., 1994). The SRC protein is implicated in several tumorigenic processes, including key roles in regulating integrin signaling, cytoskeletal organization, and adhesion dynamics, and increased SRC activity is associated with integrin activation, disruption of actin stress fibers, and disassembly of focal adhesions (Fincham et al., 1999; Klinghoffer et al., 1999; Frame et al., 2002; Webb et al., 2004). SRC also positively regulates the RhoA GTPase, which is frequently activated in advanced cancer (Ellenbroek and Collard, 2007). Many of the previously identified SRC and SFK substrates are pro-oncogenic proteins, such as FAK, p130Cas, Cortactin, and c-RAF, whose phosphorylation by SRC increases their activity, or substrates whose consequences for tumorigenesis have not been clearly established (Reynolds et al., 2014; Sulzmaier et al., 2014; Takahashi et al., 2017). In addition, SRC phosphorylation of other targets, such as PP2A and caveolin-1, can attenuate their tumor suppressor activities.

The serine/threonine kinases ERK1 and ERK2, which are key components of the RAS-RAF-MEK-ERK pathway, phosphorylate several substrates with diverse cellular functions (Yoon and Seger, 2006; Buscà et al., 2016). For example, ERK phosphorylation of some proteins, such as FOS and JUN, helps orchestrate the cell transcription program. In other proteins, such as SOS1, B-RAF, and MEK, the phosphorylation attenuates upstream activators of ERK1/2. In still others, such as BAD and caspase-9, it can reduce their pro-apoptotic activity.

DLC1 is a tumor suppressor gene that is required for embryonic development (Durkin et al., 2007) and is down-regulated in a variety of malignancies through genetic and epigenetic modifications (Durkin et al., 2007; Lukasik et al., 2011; Barras and Widmann, 2014; Ko and Ping Yam, 2014; Wang et al., 2016). The DLC1 protein possesses a Rho-GAP (GTPase-activating protein) activity, which negatively regulates Rho GTPases by catalyzing the hydrolysis of their active GTP-bound form to their inactive GDP-bound form (Wong et al., 2005). DLC1 localizes to focal adhesions, whose turnover is RhoA dependent. RhoA, which is required for embryonic development (Zhou and Zheng, 2013) and regulates cell proliferation, cytoskeletal dynamics, cell migration, and cytokinesis, is frequently activated in advanced cancer, where it contributes to maintenance of the oncogenic phenotype (Wong et al., 2005; Wang et al., 2016). In addition to its Rho-GAP activity, the DLC1 protein has scaffold functions, including the binding of tensins, talin, FAK, and caveolin-1. Both the catalytic and scaffold activities contribute to the tumor suppressor functions of DLC1.

There is no previously known relationship between DLC1 and SRC/SFKs or ERK, but we report here that DLC1 is a critical physiological substrate for SRC/SFKs and ERK, which directly phosphorylate DLC1 and attenuate its Rho-GAP and tumor suppressor activities. Our observations are noteworthy because the regulation of DLC1 by SRC/SFKs makes an important contribution to the physiology of SRC/SFKs and to the growth control of DLC1-expressing tumors, and may have translational implications.

In a survey of nontransformed and cancer-derived cell lines, we unexpectedly found an excellent correlation between levels of RhoA-GTP, total SRC protein, and SRC activity (as monitored by SRC-Y416 phosphorylation), an inverse correlation with DLC1 protein levels, and no correlation with p190–Rho-GAP (Fig. 1, A and B; and Fig. S1, A and B, for quantitation of DLC1 and SRC protein levels). To explore a possible mechanistic relationship between SRC, RhoA-GTP, and DLC1, we treated two DLC1-positive (H1703 and H157) and two DLC1-negative (H358 and A549) non–small cell lung cancer (NSCLC) lines with the SRC inhibitor saracatinib, which reduced RhoA-GTP in both DLC1-positive lines, but not in the DLC1-negative lines (Fig. 1, C–F). Similarly, treatment of the two DLC1-positive lines with the tyrosine kinase inhibitor bosutinib (Fig. S1, C and D) or SRC siRNAs (Fig. S1, E and F) led to decreased RhoA-GTP. Transfection of H1703 with WT SRC increased RhoA-GTP, while transfection with kinase-dead SRC decreased RhoA-GTP, presumably because of its dominant-negative activity (Fig. S1, G and H; Destaing et al., 2008). Taken together, these results indicate SRC kinase can increase RhoA-GTP in DLC1-positive, but not in DLC1-negative, lines.

To directly establish the role of DLC1, we tested the effect of DLC1 depletion by siRNAs on the ability of saracatinib to affect the RhoA-GTP level in two DLC1-positive lines, H1703 and H157. This reduction in DLC1, which increased RhoA-GTP as expected, abolished the ability of saracatinib to down-regulate RhoA-GTP in both lines (Fig. 1 G and Fig. S1 I). Conversely, the forced expression of DLC1 in a DLC1-negative line, H358, which reduced RhoA-GTP as expected, enabled saracatinib to further decrease RhoA-GTP (Fig. 1 H). Thus, SRC regulation of RhoA-GTP under these conditions depends upon DLC1. Although SRC can phosphorylate p190–Rho-GAP (Moran et al., 1991), this gene was not involved in the SRC-dependent regulation of RhoA-GTP identified here, as p190–Rho-GAP depletion by siRNAs did not affect the ability of saracatinib to reduce RhoA-GTP in a DLC1-positive line, H1703 (Fig. S1 J).

To determine if SRC and DLC1 form an endogenous protein complex, we performed coimmunoprecipitation (coIP) experiments in two NSCLC lines, H157 and H1703. SRC and DLC1 formed a protein complex in both lines, whether cell lysates were immunoprecipitated (IP) first with a DLC1 antibody and then immunoblotted (IB) for SRC (Fig. 2 A) or by the reciprocal coIP (Fig. 2 B). A majority of the SRC protein associated with DLC1 was enzymatically active, as saracatinib reduced most of the coIP total SRC (Fig. 2 C) and kinase-active SRC (Fig. 2 D). Complex formation between SRC and DLC1 was also observed in the nontransformed H2071 and human bronchial epithelial cell (HBEC) lines (Fig. 2 E), which implies the interaction is physiologically relevant.

Colocalization of SRC and DLC1 was confirmed by confocal microscopy in H1703 and H157 cells, with colocalization coefficients of 0.66 and 0.62, respectively (Fig. 2 F and Fig. S2 A), and with positive proximity ligation assays (PLAs; Fig. 2 G and Fig. S2 B). Colocalization of kinase-active SRC at focal adhesions, as monitored with vinculin, was confirmed in H1703 and H157 cells, with colocalization coefficients of 0.72 and 0.66, respectively (Fig. 2 H and Fig. S2 C), and with positive PLA (Fig. 2 I and Fig. S2 D). Consistent with this finding, SRC and vinculin were not colocalized in a DLC1-negative line (H358; data not shown). Thus, colocalization of SRC at focal adhesions depends on the presence of DLC1.

The above results, conducted under steady-state growth conditions, led us to explore whether a physiological, inducible situation that can stimulate SRC activity and increase RhoA-GTP might also depend on DLC1. To test this possibility, we placed six different cell lines on culture dishes that had been coated with the integrin ligand fibronectin, which can activate SRC (Harburger and Calderwood, 2009) and increase RhoA-GTP (Ren et al., 1999). To determine if the response would be relevant to nontransformed cells and cancer cells, two of the DLC1-positive lines, H1634 and H2071, were nontransformed and expressed DLC1, while the other four were NSCLC lines, the two DLC1-positive and two DLC1-negative ones studied above. Plating the cells on fibronectin led to SRC activation in all six lines as expected, but the increase in RhoA-GTP was limited to the four DLC1-positive lines (Fig. 3, A–F). The inability of SRC to increase RhoA-GTP in the two DLC1-negative lines was not attributable to their RhoA-GTP levels being maximal, as lysophosphatidic acid (LPA), which increases RhoA-GTP through activation of large G protein–coupled receptors (Xiang et al., 2013), increased RhoA-GTP in both lines when plated on fibronectin, as well as in the DLC1-positive H1634 line (Fig. S2, E and F).

The increased RhoA-GTP depended on the SRC activation by fibronectin, as saracatinib prevented SRC activation and increased RhoA-GTP (Fig. 3 G). Plating cells on fibronectin can also increase FAK activity (Meng et al., 2009), but treatment with a FAK kinase inhibitor, FAK-14, did not interfere with either fibronectin-induced SRC activation or increased RhoA-GTP, although it did inhibit FAK activity, as monitored by reduced FAK-Y397 phosphorylation (Fig. 3 G). The increased RhoA-GTP required DLC1 protein, as siRNAs depletion of DLC1 abrogated the ability of fibronectin to increase RhoA-GTP (Fig. 3 H). SRC activation by fibronectin treatment led to increased tyrosine phosphorylation of DLC1, which was markedly reduced by saracatinib treatment, but not by FAK-14 treatment (Fig. S2, G–J). Fibronectin and saracatinib treatment did not alter SRC protein level, and fibronectin treatment did not alter the FAK activity. Thus, the fibronectin-induced increase in RhoA-GTP requires SRC activation and the presence of DLC1, and is associated with the phosphorylation of one or more tyrosines in DLC1.

Kinetic analysis of the fibronectin-induced activities indicated that SRC-Y416 phosphorylation began within 15 min of plating, while the increased RhoA-GTP and tyrosine phosphorylation of DLC1 began at ∼30 min (Fig. 3, I and J). The latter two increases were correlated with increased SRC/DLC1 complex formation between 15 and 30 min (Fig. S2 K).

DLC1 has four recognized regions: an N-terminal SAM domain, a linker region, a Rho-GAP domain, and a C-terminal START domain (Fig. 4 A). To identify the regions of DLC1 required for SRC binding, lysates from human embryonic kidney (HEK) 293T cells expressing various GFP-tagged DLC1 fragments (Fig. 4 A) were IP with SRC antibody followed by IB with GFP antibody. Interestingly, SRC bound two nonoverlapping DLC1 fragments, one spanning amino acids 1–492 and the other 500–1,091, implying that SRC interacts with more than one DLC1 region (Fig. 4 B). The N-terminal binding region of DLC1 was mapped more precisely to DLC1 amino acids 80–200 (Fig. 4 C). DLC1 amino acids 800–850, in the Rho-GAP domain, were required for SRC binding the C-terminal region, as both DLC1(800–900) and DLC1(609–850) were positive, while DLC1(850–1,091) was negative (Fig. 4, D and E).

As had been true following the plating of cells on fibronectin, tyrosine in DLC1 was phosphorylated in a SRC-dependent manner under steady-state growth conditions, as saracatinib reduced DLC1 tyrosine phosphorylation in H1703 and H157 lines (Fig. 4 F). To map the phosphorylated DLC1 tyrosine residues, liquid chromatography–mass spectrometry (LCMS) was performed with partially purified DLC1 protein from HEK 293T cells transfected with DLC1 alone or cotransfected with constitutive kinase-active or kinase-dead SRC (Fig. 4 G and Fig. S3 A). The LCMS results support the conclusion that DLC1-Y451 and -Y701 are phosphorylated in a SRC-dependent manner, as LCMS detected strong phosphorylation that was fully localized on Y451 and Y701 in the relevant DLC1-WT peptides in the samples from cells expressing constitutively kinase-active SRC, but not from cells expressing kinase-dead SRC.

To establish that SRC directly phosphorylates DLC1, we performed an in vitro kinase assay using ³²P labeling. Partially purified GFP-DLC1-WT was strongly phosphorylated by recombinant SRC in vitro, unlike the GFP control (Fig. 4 H, lanes 2 and 3), and saracatinib treatment abolished the phospho-signal in GFP-DLC1-WT (Fig. S3 B). Thus, DLC1 is a direct SRC substrate. Furthermore, DLC1 fragments spanning amino acids 1–492 and 500–1091, which contain Y451 and Y701, respectively, were strongly phosphorylated (Fig. S3 C). Recombinant SRC also phosphorylated the other two members of DLC family, DLC2 and DLC3, but less strongly than DLC1 (Fig. S3 D).

To confirm that Y451 and Y701 are the major SRC phosphorylation sites in DLC1, both tyrosines were mutated to the phospho-deficient phenylalanine (DLC1-2F), which drastically reduced, but did not abolish, the phospho-signal (Fig. 4 H, lane 6). When tyrosine-to-phenyalanine single mutants were analyzed, the phospho-signal was reduced, but was greater than the DLC1-2F mutant (Fig. 4 H, lanes 4 and 5). The two other widely expressed SFKs, FYN and YES, also directly phosphorylate DLC1 (Fig. 4 I). As with SRC, the phospho-signal by FYN kinase was much lower when the DLC1-2F mutant was tested in an analogous in vitro kinase assay (Fig. S3 E).

The LCMS analysis of DLC1 phosphorylations indicated that S129 was phosphorylated (Fig. 4 G). Inspection of the sequences surrounding S129 indicated this serine might be part of overlapping consensus motifs for ERK1/2 phosphorylation (127-PXSP-130) and SH3 domain binding (127-PXXP-130). As SRC binding had been mapped to DLC1 amino acids 80–200 (Fig. 4 C), we investigated the possibility that the SRC SH3 domain might bind to this fragment, that S129 might be phosphorylated by ERK, and that this phosphorylation might affect the putative SH3 binding. CoIP of extracts from the H157 and H1703 lines determined that ERK and DLC1 formed an endogenous protein complex (Fig. 5, A and B). Furthermore, the SRC SH3 domain did bind DLC1 amino acids 80–200 in vivo, and it bound the full-length DLC1-WT more strongly than it bound the full-length DLC1-S129A mutant (Fig. 5, C and D). Similarly, an in vitro ERK kinase assay gave a strong phospho-signal with full-length DLC1-WT or DLC1 amino acids 80–200, but a greatly attenuated phospho-signal for the S129A mutant in both the full-length and 80–200 DLC1 fragment (Fig. 5, E and F). Importantly, the in vitro SRC kinase reaction gave a stronger DLC1 tyrosine phosphorylation signal with DLC1-WT than with the DLC1-S129A mutant, indicating the ERK-dependent phosphorylation of S129 increases the efficiency of SRC-dependent phosphorylation of Y451 and Y701 (Fig. 5 G). Thus, ERK-dependent phosphorylation of S129 increases the efficiency of SRC binding via its SH3 domain, which in turn increases phosphorylation of Y451 and Y701.

The DLC1 scaffold functions include its binding of talin, FAK, and tensin near residue Y451. Tensin binding requires a DLC1 segment that includes residues 440–445 (Liao et al., 2007; Qian et al., 2007), while both FAK and talin bind DLC1 residues 467–489 (Li et al., 2011; Zacharchenko et al., 2016). To determine if phosphorylation of Y451 affects tensin binding, GST-tagged full-length tensin was cotransfected with a DLC1 fragment encoding WT amino acids 80–550, with the same fragment encoding Y451F phospho-deficient mutant, or with the phosphomimetic Y451D mutant, and GST pull-downs were analyzed for binding the expressed DLC1 segment. Compared with tensin binding to DLC1-WT, the DLC1-Y451F mutant bound tensin more efficiently, while the DLC1-Y451D mutant bound it less efficiently (Fig. 5 H). Analogous results were seen when the tensin fragment (amino acids 1508–1786) shown previously to bind DLC1 was transfected with DLC1 WT or the Y451 mutants (Fig. 5 I). These observations strongly suggest that phosphorylation of Y451 decreases tensin binding. By contrast, an analogous evaluation of talin protein indicated that its binding to DLC1 was not affected by the Y451F or the Y451D mutation (Fig. 5, J and K).

Our finding that SRC could increase RhoA-GTP in a DLC1-dependent manner suggested that SRC-dependent phosphorylation of DLC1 would attenuate its Rho-GAP function. Therefore, we compared the Rho-GAP activity for DLC1-WT, for the tyrosine to phenylalanine single mutants (DLC1-Y451F and DLC1-Y701F), the combined tyrosine to phenylalanine double mutant (DLC1-2F), the phosphomimetic tyrosine to aspartate single mutants (DLC1-Y451D and DLC1-Y701D), and the combined tyrosine to aspartate double mutant (DLC1-2D), and included a “Rho-GAP–dead” mutant (DLC1-R718A) as a control.

In A549 cells that expressed similar levels of the DLC1 mutants, RhoA-GTP levels in the DLC1-Y701D and DLC1-2D transfectants were similar to DLC1-R718A, while RhoA-GTP in the DLC1-2F mutant was even lower than that induced by DLC1-WT (Fig. 6 A). The RhoA-GTP phenotypes mapped to Y701 phosphorylation, as RhoA-GTP was high and low, respectively, for the DLC1-Y701D and DLC1-Y701F single mutants, while there was little difference in RhoA-GTP between the DLC1-Y451D and DLC1-Y451F single mutants and the DLC1-WT. Analogous results were seen when the DLC1 mutants were analyzed for two RhoA-GTP downstream effectors, Rho kinase (ROCK) activity (Fig. 6 B) and phosphorylation of myosin regulatory light chain (pMRLC), as analyzed by IB (Fig. 6 C) and immunofluorescence (Fig. S4, A and B). Consistent with these results, cells transfected with GFP, DLC1-Y701D, DLC1-2D, or DLC1-R718A had well-formed stress fibers, unlike cells transfected with DLC1-WT, DLC1-Y701F, or DLC1-2F (Fig. S4 C).

As Y701 is located in the Rho-GAP domain, we asked whether the Rho-GAP activities that were mainly attributable to amino acid encoded at residue 701 would map to the Rho-GAP domain, which has previously been genetically mapped to DLC1(609–878; Kim et al., 2008). Consistent with this possibility, the in vivo RhoA-GTP phenotypes for the Y701 mutants of the isolated DLC1 Rho-GAP domain were similar to those with the same Y701 mutation in full-length DLC1 (Fig. 6 D). Analogous results were observed when the DLC1 full-length or isolated Rho-GAP domain mutants were analyzed for their in vitro Rho-GAP activity (GTP hydrolysis), with Y701F being most active and Y701D least active (Fig. 6, E and F). Compared with DLC1-WT, the in vitro RhoA-GTP binding to the full-length DLC1-2D mutant was as low as that of the Rho-GAP–dead DLC1-R677A mutant, whose RhoA-GTP binding is known to be reduced (Fig. 6 G; Jaiswal et al., 2014). Thus, the mutant analysis indicates that SRC phosphorylation of DLC1 attenuates its Rho-GAP activity, and this phenotype maps to Y701 and the Rho-GAP domain.

To examine the biological significance of DLC1 phosphorylation by SRC, we evaluated the phosphodeficient and phosphomimetic DLC1 single and double mutants in several cancer-related bioassays. Stable DLC1-WT transfectants in the H358 line reduced the following functions: anchorage-independent cell growth (Fig. 7, A and B), transwell cell migration (Fig. 7, C and D), and xenograft tumors in immunodeficient mice (Figs. 7, E–G; and Fig. S4, D and E). In these bioassays, the DLC1-2D mutant was as deficient as the Rho-GAP–dead DLC1-R718A, while the DLC1-2F mutant was even more active than DLC1-WT.

Given the reversibility of the SRC-dependent DLC1 phosphorylation, we evaluated whether SRC inhibitors might have therapeutic efficacy in a DLC1-positive tumor that had high SRC activity. First, we used the MMTV-PyMT breast cancer model, where the MMTV promoter drives the polyomavirus middle T antigen, which stimulates SRC activity (Guy et al., 1994). 5 d of oral treatment with saracatinib reduced breast tumor weight by >50% (Fig. 8 A), which was correlated with decreased SRC activity and RhoA-GTP levels (Fig. 8, B and C).

To determine mechanistically if DLC1 reactivation contributes to the antitumor activity from SRC kinase inhibition, we examined the response of saracatinib treatment to tumor xenografts derived from the H358 NSCLC line in immunodeficient NOD-SCID mice by comparing four isogenic H358 cell lines that differed in the DLC1 allele they expressed: DLC1-WT, DLC1-2F, DLC1-2D, or DLC1-negative (GFP control). Saracatinib treatment for 5 d reduced tumor weight of the DLC1-WT line by 75%, in contrast to a 20% reduction for the DLC1-negative control line (Fig. 8, D and E). Similar to the DLC1-negative control, saracatinib treatment of the DLC1-2F and DLC1-2D tumors, whose DLC1 activity would be predicted to not be altered by a SRC inhibitor, reduced tumor weight by 18% and 21%, respectively. The results with the DLC1-2F and DLC1-2D mutants, which should not be reactivatable by the SRC inhibitor because their tyrosine residues have been mutated, strongly imply that the higher antitumor activity in the DLC1-WT tumors is attributable to reactivation of DLC1 by the SRC inhibitor. Biochemical analyses of the tumors directly support this inference, as saracatinib treatment reduced RhoA-GTP only in the DLC1-WT line, although SRC activity was inhibited in all lines (Fig. 8, F and G).

We recently reported that AKT, by directly phosphorylating three specific serines in DLC1, attenuates its tumor suppressor activity and showed that this attenuation could be reversed by the AKT inhibitor MK-2206 (Tripathi et al., 2017). As these effects of AKT and its inhibition were qualitatively similar to those of SRC and its inhibition, but were clearly attributable to different phosphorylations on DLC1, we asked whether combined inhibition of AKT and SRC might have cooperative effects on tumor inhibition in a cell line that expressed DLC1. To examine this possibility, we compared the antitumor activity of the H358 cell line expressing DLC1-WT with the isogenic DLC1-negative line. In the DLC1-positive line, 5 d of combined treatment with MK-2206 and saracatinib resulted in an 89% decrease in tumor weight, while single-agent treatment reduced tumor weight by 67% for MK-2206 alone and 64% for saracatinib alone (Fig. 9, A and B). These results were correlated with the combination treatment reducing RhoA-GTP to a greater degree than did single-agent treatment (Fig. 9, C and D). In contrast, combined treatment of the DLC1-negative line reduced tumor weight by only 28% and did not result in decreased RhoA-GTP, although inhibition of the AKT and SRC—as determined by pAKT-S473 and pSRC-Y416, respectively—was similar in both lines. The combination treatment induced β-galactosidase and annexin V expression, which are markers of cellular senescence and apoptosis, respectively, to a greater degree than single-agent treatment (Fig. S5, A–D). We also determined that the antitumor activity of two other AKT and tyrosine kinase inhibitors, perifosine and bosutinib, respectively, when used in combination against the two H358 lines, was similar to that found for MK-2206 and saracatinib (Fig. S5, E and F).

This study reports several noteworthy observations. We have determined that DLC1 is a major biological and biochemical target of SRC, identified a DLC1-dependent relationship between SRC and RhoA, and demonstrated that enzymatically active SRC binds and phosphorylates two tyrosines in DLC1. These observations are relevant both to normal physiology and cancer, as they were made in nontransformed lines and in tumor cell lines. In addition, DLC1-S129 lies within amino acids that form both a consensus site for SH3 binding domains and for ERK1/2 phosphorylation; our experimental data confirm that ERK1/2 phosphorylation of DLC1-S129 facilitates the binding of the SRC SH3 domain and full-length SRC to DLC1, which increases the SRC-dependent phosphorylation of DLC1-Y451 and -Y701. Thus, SRC and ERK1/2 signaling converge on DLC1 to cooperatively regulate it, with SRC-dependent phosphorylation of DLC1 attenuating the Rho-GAP– and tensin-binding activities of DLC1 and decreasing its tumor suppressor functions. Although most of our experiments have been focused on SRC, the results are likely to be relevant to other SFKs, as in vitro kinase assays were also positive for FYN and YES. Finally, we began to explore the translational implications of the potential reversibility of the SRC-induced attenuation of DLC1 functions. These experiments show the antitumor activity of a SRC kinase inhibitor is much more potent in DLC1-positive tumors, because the inhibitor reactivates the tumor suppressor activity of DLC1. This observation also underscores the importance of DLC1 as a key biological SRC target in DLC1-positive tumors. Furthermore, we demonstrated cooperative antitumor activity from the combined inhibition of SRC and AKT, which phosphorylates and attenuates the tumor suppressor function of DLC1 by a mechanism distinct from that of SRC (Tripathi et al., 2017). A schematic model of the various effects of SRC on DLC1 is shown in Fig. 10.

SRC has been previously reported to regulate RhoA, but its ability to do so has often been attributed to the activation of Rho-specific GEFs or to inactivation of p190–Rho-GAP (Fincham et al., 1999; Huveneers and Danen, 2009; Pullikuth and Catling, 2010). Under both normal steady-state growth conditions and acute integrin activation by fibronectin, the regulation of RhoA-GTP by the SRC kinase was found to be determined primarily by whether or not DLC1 was expressed, rather than by these other regulators, and to depend on DLC1 phosphorylation by SRC. Kinetic analysis of the fibronectin-dependent increase in RhoA-GTP indicated it occurred within minutes of SRC binding and phosphorylating DLC1. Thus, attenuation of DLC1 by its SRC-dependent phosphorylation should be considered when SRC is found to regulate RhoA-GTP. These results complement those we recently reported for AKT, which is downstream of receptor tyrosine kinases, in phosphorylating serines in DLC1 that attenuate its function, including increased RhoA-GTP (Tripathi et al., 2017).

The phylogenetically conserved tyrosines Y451 and Y701 phosphorylated by SRC are strategically located in DLC1 and are conserved in DLC family members DLC2 and DLC3, which we found are also SRC substrates. Y451 lies just downstream from amino acids 440–445, which form part of the docking site for the SH2 domains of tensins and includes Y442, which is required for tensin binding (Liao et al., 2007; Qian et al., 2007). Y451 is also located upstream from the amino acids needed for binding the R8 domain of talin, which have been mapped to DLC1 residues 467–489 (Li et al., 2011; Zacharchenko et al., 2016). Mutating DLC1-Y451 to the phosphomimetic DLC1-D451 led to decreased tensin binding but did not affect talin binding. Y701 lies within the Rho-GAP domain, and the DLC1-Y701D mutant had decreased RhoA-GTP binding and decreased Rho-GAP activity, which strongly implies that these phenotypes are attributable to conformational changes in the Rho-GAP domain induced by Y701 phosphorylation. Consistent with this interpretation, these phenotypes did not require sequences beyond the Rho-GAP domain, as the decreased Rho-GAP activity of Y701D mutant of the isolated Rho-GAP domain (DLC1[609–878]) was similar to that of the same mutant in full-length DLC1. The profound impact of Y701 phosphorylation on Rho-GAP activity contrasts markedly with the phosphorylation by PDK of S807 in the DLC1 Rho-GAP domain, which is reported not to affect Rho-GAP activity (Scholz et al., 2011). The small degree of residual tyrosine phosphorylation seen in the in vitro SRC kinase assay with the DLC1-2F mutant suggests there might be another tyrosine in DLC1 that is phosphorylated by SRC. However, this putative phosphorylation by SRC would be less efficient than that of Y451 and Y701, and does not by itself appear to alter the Rho-GAP activity of DLC1, as SRC inhibition did not affect this parameter in cells expressing the DLC1-2F and DLC1-2D mutants.

SRC protein in complex with DLC1 is enzymatically active and binds two distinct regions of DLC1, an N-terminal segment of the linker region (amino acids 80–200) and the Rho-GAP domain. The DLC1 N-terminal segment binds the SH3 domain of SRC, which is only available when SRC is active, as this domain forms an intramolecular complex with the SRC kinase domain when SRC is inactive (Espada and Martín-Pérez, 2017). Under our steady-state growth conditions, the colocalization of active SRC with the focal adhesion protein vinculin depended on the presence of DLC1.

The potential reversibility of DLC1 phosphorylation by SRC may have translational implications. SRC kinase inhibitors have been approved by the US Food and Drug Administration for cancer treatment (Roskoski, 2015), but what accounts for the heterogeneity of the clinical responses is frequently unclear. Our results indicate that SRC kinase inhibition can have strong antitumor activity in two DLC1-positive tumor models with high SRC activity, the MMTV-PyMT breast cancer model and tumor xenografts from a NSCLC line. The responses were associated with decreased tyrosine phosphorylation of DLC1 and decreased RhoA-GTP in tumor extracts. By contrast, if the NSCLC line did not express DLC1, SRC inhibition did not change RhoA-GTP levels and had lower antitumor activity, as was also true of isogenic DLC1 mutants that were not activatable by SRC inhibition because they carried mutations in Y451 and Y701. These results confirm that DLC1 activation in this line is required for the strong antitumor activity of SRC inhibition. Thus, despite there being many SRC substrates, DLC1 can be a sufficiently important one that its reactivation can make a critical contribution to the antitumor activity of SRC inhibitors.

We also demonstrated cooperative antitumor activity from combined treatment of the DLC1-positive NSCLC line with SRC and AKT inhibitors. These results were correlated with the combination inducing lower levels of RhoA-GTP and higher levels of senescence and apoptosis markers than treatment with either inhibitor alone. These observations represent a rigorous proof-of-principle that it might be clinically useful to combine drugs that use distinct mechanisms to reactivate the tumor suppressor activity of DLC1.

Although it is of course critical for inhibitors to hit their target, our data highlight that clinical efficacy can also be critically dependent on the downstream targets affected by the inhibition. Using reactivation of DLC1 as a possible biomarker of clinical efficacy incorporates the impact of oncoprotein inhibition on reactivation of a tumor suppressor, rather than focusing the effects exclusively on the inhibition of pro-oncogenic factors. Such considerations have been usefully applied to the development of CDK4/6 inhibitors, thanks to the basic observation that CDK4/6-dependent phosphorylation of pRB can attenuate its tumor suppressor functions (Sherr et al., 2016). pRB can serve as a biomarker whose decreased phosphorylation and concomitant reactivation are associated with an increased likelihood of clinical response to CDK4/6 inhibitors.

For DLC1, a limitation of the specific therapeutic approach described here is that it is likely to benefit only those tumors that express moderate to high levels of DLC1. However, the down-regulation of DLC1 in many tumors appears to be attributable to epigenetic changes that may be reversible, although DLC1 down-regulation in some tumors arises because of DLC1 mutation or deletion, which is irreversible (Durkin et al., 2007; Lukasik et al., 2011; Barras and Widmann, 2014; Ko and Ping Yam, 2014; Wang et al., 2016). One possible way to increase the proportion of tumors for which the therapeutic targeting of DLC1 could be clinically beneficial might be to use a suitable inhibitor to reverse an epigenetic change that has resulted in reduced or silenced DLC1 expression at the mRNA or protein level, thus increasing the steady-state level of DLC1 protein, and to combine this treatment with inhibition of SRC and/or AKT kinase activities. This approach could be conceptually relevant for targeting other tumor suppressor genes that are inactivated by several epigenetic mechanisms.

GFP-tagged DLC1 WT (GFP-DLC1-WT), two DLC1 Rho-GAP–dead mutants (GFP-DLC1-R718A and GFP-DLC1-R677A), and GFP-tagged DLC1 fragments encoding DLC1 residues 1–492, 500–1,091, 1–110, 80–200, 80–300, 80–400, 400–500, 500–700, 609–850, 609–878, 623–1,091, 800–900, 850–1,091, 899–996, 996–1,091, and GFP-DLC3 were constructed by PCR and subcloned into a modified pEGFP-C1 vector through Kpn1-NotI sites, as described (Qian et al., 2007). GST-tagged tensin and talin fragments were described previously (Qian et al., 2007; Li et al., 2011). GFP-DLC2 was a gift from M. Mowat (University of Manitoba, Winnipeg, Canada). A series of individual and combined tyrosine-to-phenylalanine (Y-F), tyrosine-to-aspartate (Y-D), serine-to-alanine (S-A), and serine-to-aspartate (S-D) mutations were introduced into full-length DLC1-WT and into DLC1 fragments using a site-directed mutagenesis kit (Agilent Technologies). Full-length DLC1 and DLC1(80–550) linker region fragments with or without Y451F or Y451D mutation were engineered into the PEBG vector by BamHI and NotI, resulting in GST-tagged DLC1 constructs. All PCR regions were confirmed by sequencing. The sequence for each primer was as follows: DLC1S129A, forward 5′-GAC​CTG​GTC​CCT​GGG​GCC​CCA​GAC​GAC​TCC-3′; DLC1S129A, reverse 5′-GGA​GTC​GTC​TGG​GGC​CCC​AGG​GAC​CAG​GTC-3′; DLC1S129D, forward 5′-GAC​CTG​GTC​CCT​GGG​GAC​CCA​GAC​GAC​TCC-3′; DLC1S129D, reverse 5′-GGA​GTC​GTC​TGG​GTC​CCC​AGG​GAC​CAG​GTC-3′; DLC1Y451F, forward 5′-GGC​TCC​ATC​CTC​TTC​TCC​AGT​TCA​GGG-3′; DLC1Y451F, reverse 5′-CCC​TGA​ACT​GGA​GAA​GAG​GAT​GGA​GCC-3′; DLC1Y451D, forward 5′-GGC​TCC​ATC​CTC​GAC​TCC​AGT​TCA​GGG-3′; DLC1Y451D, reverse 5′-CCC​TGA​ACT​GGA​GTC​GAG​GAT​GGA​GCC-3′; DLC1Y701F, forward 5′-GAC​TGT​GTC​AAC​TTC​GAA​GGA​CAG-3′; DLC1Y701F, reverse 5′-AGA​CTG​TCC​TTC​GAA​GTT​GAC​ACA​GTC-3′; DLC1Y701D, forward 5′-GAC​TGT​GTC​AAC​GAC​GAA​GGA​CAG​TCT-3′; and DLC1Y701D, reverse 5′-AGA​CTG​TCC​TTC​GTC​GTT​GAC​ACA​GTC-3′.

The following antibodies were obtained from Cell Signaling Technology: SRC rabbit (2108), SRC mouse (2110), phospho-SRC-pY416 (2101), phospho-SRC-pY527 (2105), AKT mouse (2920), AKT rabbit (4691), phospho-AKT-pS473 rabbit (4060), phospho-AKT-pT308 rabbit (13038), phospho-MRLC-Thr18/Ser19 rabbit (3674), MYPT1 (2634), pMYPT1-T853 (4563), FAK rabbit (3285), and GAPDH rabbit (2118). Two DLC1 antibodies, which gave similar results, were used: one generated in our laboratory (DLC1 antibody; clone 428; 24) and the other, DLC1 mouse (612021) antibody, from BD Biosciences. FAK mouse (610088), phospho-FAK-Y397 (611723), and phospho-serine mouse (612547) were purchased from BD Biosciences. EGFR rabbit (ab2430), phospho-EGFR-Y845 rabbit (ab5636), MRLC mouse (ab11082), GFP mouse (ab1218), GFP rabbit (ab290), ERK1 (ab137766), phospho-ERK (ab24157), annexin V (ab14196), and β-galactosidase (ab1047) antibodies were purchased from Abcam. RhoA mouse (ARH04) and phospho-MRLC goat (sc-12896) antibodies were obtained from Cytoskeleton and Santa Cruz Biotechnology, respectively. The phospho-tyrosine mouse (4G10) antibody was obtained from Millipore. The p190–Rho-GAP mouse (R3150) and vinculin (V9131) antibodies were obtained from Sigma-Aldrich. Anti-rabbit (NA934V) and anti-mouse (NXA931V) IgG horseradish peroxidase–linked secondary antibodies were purchased from GE Healthcare. Alexa Fluor 568 anti-rabbit IgG, Alexa Fluor 488 anti-mouse IgG, Alexa Fluor 488 phalloidin, and DAPI were purchased from Invitrogen.

HEK 293T, human skin epithelial H2071, human fibroblastic H1634, and HBECs were cultured in DMEM supplemented with 10% FBS. Human NSCLC lines (H1703, H157, A549, and H358) were cultured in RPMI-1640 supplemented with 10% FBS. Transient transfections were performed with Lipofectamine 3000 (Invitrogen) and cultured for 48 h. Stable clones expressing GFP or DLC1 mutants were made by transfecting H1703 or H358 cells with Lipofectamine 3000 followed by G418 selection (0.9 µg/ml). Knockdown of protein expression, with two different siRNAs, was confirmed by IB. Validated siRNAs for human DLC1 (Hs_DLC1 siRNA_5, SI03219909; and Hs_DLC1 siRNA_11, SI04952213), p190–Rho-GAP (Hs_GRLF1 siRNA_7, SI02664025; and Hs_GRLF1 siRNA_8, SI02664032), and SRC (Hs_SRC siRNA_7, SI02223928; and Hs_SRC siRNA_10, SI02664151) were from obtained from QIAGEN, as were negative control siRNAs (control siRNA 1, 1027280; and control siRNA 2, 1027310).

The sequence for each DLC1 siRNA was as follows: Hs_DLC1_5 sense sequence: 5′-CGA​UGU​CGU​AAU​UCC​UAU​ATT-3′; Hs_DLC1_5 antisense sequence: 3′-CGG​CUA​CAG​CAU​UAA​GGA​UAU-5′; Hs_DLC1_11 sense sequence: 5′-GGA​GUG​UAG​GAA​UUG​ACU​ATT-3′; and Hs_DLC1_11 antisense sequence: 3′-GAC​CUC​ACA​UCC​UUA​ACU​GAU-5′.

The sequence for each GRLF1 (p190 Rho-GAP) siRNA was as follows: Hs_GRLF1_7 sense sequence: 5′-GGA​UGU​UCU​GGG​AGA​GGA​ATT-3′; Hs_ GRLF1_7 antisense sequence: 3′-GTC​CUA​CAA​GAC​CCU​CUC​CUU-5′; Hs_ GRLF1_8 sense sequence: 5′-GCC​UAA​GGA​GGA​ACA​CUA​ATT-3′; and Hs_ GRLF1_8 antisense sequence: 3′-GTC​GGA​UUC​CUC​CUU​GUG​AUU-5′.

The sequence for each SRC siRNA was as follows: Hs_SRC_7 sense sequence: 5′-GCU​UGU​GGG​UGA​UGU​UUG​ATT-3′; Hs_ SRC_7 antisense sequence: 3′-GCC​GAA​CAC​CCA​CUA​CAA​ACU-5′; Hs_ SRC_10 sense sequence: 5′-CCA​UGU​GCG​UCC​AUA​UUU​ATT-3′; and Hs_ SRC_10 antisense sequence: 3′-GAG​GUA​CAC​GCA​GGU​AUA​AAU-5′.

To reduce the indicated protein expression, cells were transfected with indicated siRNAs (160 nM) or with scrambled control siRNAs and harvested 48 h later. The final concentration of fibronectin (Sigma-Aldrich) was 100 ng/ml. SRC inhibitors saracatinib, bosutinib, PP1, and FAK inhibitor FAK-14 (used at 10 µM each) were purchased from Selleckchem. After overnight incubation in serum-free media, cells were treated with indicated ligand or inhibitors for 15–120 min.

Lysates from transfected cells were IP with GFP antibody, and immunopellets were sequentially washed once with high-salt HNTG buffer (20 mM Hepes, 500 mM NaCl, 0.1% Triton X-100, and 10% glycerol), twice with low-salt HNTG buffer (20 mM Hepes, 150 mM NaCl, 0.1% Triton X-100, and 10% glycerol), and once with kinase reaction buffer (35 mM Hepes, pH 7.4, 10 mM MgCl₂, 1 mM EGTA, 1% Tween 20, 0.1 mM sodium vanadate, and 1 mM DTT). The kinase reaction was performed in 30 µl of reaction buffer containing 15 µM cold ATP, 2.5 µCi [³²P]γ-ATP, and 100 ng of recombinant active SRC, YES, FYN, or ERK1/2 (EMD Millipore) at 30°C for 45 min. The reaction was stopped by adding 10 µl of 4× Laemmli sample buffer and heating at 95°C for 5 min. Proteins were separated by gel electrophoresis and autoradiographed to detect ³²P incorporation.

Cells were transiently cotransfected with plasmids expressing GST or the indicated GST fusion constructs together with GFP or the indicated GFP-DLC1 constructs. 48 h after transfection, cells were lysed with golden lysis buffer (20 mM Tris, pH 7.9, 137 mM NaCl, 10% glycerol, 1% Triton X-100, 5 mM EDTA, 1 mM EGTA, 1 mM Na₃VO₄, 10 mM NaF, 1 mM sodium pyrophosphate, 0.5 mM β-glycerophosphate, protease inhibitor mixture tablet, and phosphatase inhibitor). The cleared supernatants were collected, and a small portion of supernatants was taken to determine the protein concentration using the DC protein assay (Bio-Rad). For the pull-down assay, 1.0 mg of total protein from each cell extracts were used, to which 30 µl of glutathione Sepharose-4B slurry (GE Healthcare) was added, with continuous rotation for 3 h at 4°C. The pellets were sequentially washed once with golden lysis buffer, once with high-salt HNTG buffer, and twice with low-salt HNTG buffer. The beads were incubated with 30 µl Laemmli sample buffer, subjected to 10% SDS-PAGE, transferred onto nitrocellulose membranes (Invitrogen), and detected by IB using specific antibodies. A portion of the cell extracts was used as a loading control to verify expression of the GFP fusion proteins and the GPF control. For coIP experiments, equal amounts of protein from each cell lysates were precleared with protein G slurry (Thermo Fisher Scientific) and then incubated with the indicated antibodies or control IgG for 1 h at RT. After incubation, 30 µl of protein G slurry was added to each immune reaction and rotated at 4°C overnight. The immunopellets were washed three times as above. coIP proteins were eluted by boiling for 5 min in 30 µl Laemmli sample buffer containing 5% (vol/vol) 2-mercaptoethanol. Eluted proteins were resolved on a NuPage 4–12% BisTris gel and detected by IB using specific antibodies. Immunoreactive bands were detected by enhanced chemiluminescence (ECL-Plus; GE Healthcare) using horseradish peroxidase–linked anti-rabbit or anti-mouse secondary antibodies (1:5,000 dilutions).

Transiently or stably transfected cells were seeded onto glass chambers, incubated for 24 h, and fixed with 4% paraformaldehyde for 20 min. Fixed cells were permeabilized with 0.25% Triton X-100 in PBS and then blocked with 3% BSA in PBS for 2 h. The cells were incubated with a 1:200 dilution of the indicated primary antibodies at 4°C overnight. After being thoroughly washed in PBS, the cells were incubated with the appropriate 1:250 Alexa Fluor–conjugated secondary antibodies for 1 h. To visualize actin or nuclei, cells were incubated with phalloidin (1:50) or DAPI (1:2,500) for 1 h. After staining, the cells were thoroughly washed with PBS and mounted with gel mounting solution (Biomeda).

Confocal microscopy of fluorescent-labeled cells was performed using a microscope (LSM 780; Carl Zeiss) with an excitation wavelength of 488 nm to detect transfected GFP fusion proteins. Alexa Fluor probes were viewed with excitation wavelengths of 488 nm (Alexa Fluor 488) and 568 nm (Alexa Fluor 568). Images were made at RT using photomultiplier tubes with a Plan-Apochromat 63×/1.4 NA oil differential interference contrast objective lens with a 2× magnifier to produce a 125× magnification. The colocalization of two proteins was analyzed by confocal software (ZEN 2012; Carl Zeiss). For quantification of representative morphology in each group, ∼15 cells per condition randomly selected from several fields were analyzed. The average Mander’s overlapping colocalization coefficient ± SD was calculated and is shown in each panel. The overlapping colocalization coefficients can range from 0 to 1, where 0 means no colocalization and 1 means full colocalization of the two proteins. The images were minimally processed for levels/contrast adjustment in DAPI panels, and the adjustment was done for all images using Adobe Photoshop CC software. The adjustments do not enhance, erase, or misrepresent any information present in the original images.

A Rho activation assay kit (EMD Millipore) was used to measure GTP-bound RhoA, as described (Tripathi et al., 2014). In brief, equal amounts (1,000 µg) of each cell lysate were incubated with 30 µg GST-Rhotekin Rho-binding domain coupled to glutathione-agarose beads for 45 min. Beads were washed three times with washing buffer. Washed samples were subjected to 4–12% SDS-PAGE, transferred onto nitrocellulose membranes, and detected by IB, using RhoA antibody (ARH04 from Cytoskeleton; and 05-778, clone 55, from EMD Millipore).

Cells were fixed and harvested in 10% trichloroacetic acid containing 10 mM DDT. Pellets were dissolved in 10 µl of 1 M Tris base and mixed with 100 µl of extraction buffer (8 M urea, 2% SDS, 5% sucrose, and 5% 2-mercaptoethanol). Equal amounts of protein from each cell extract were subjected to 10% SDS-PAGE, transferred onto nitrocellulose membranes, and incubated with antibody specific for phospho-myosin binding subunit (phospho-MYPT1-Thr853) or myosin binding subunit (MYPT1), and protein bands were visualized by enhanced chemiluminescence. ROCK activity was expressed as the ratio of phospho-MYPT1 to total MYPT1.

GFP-tagged DLC1 constructs were purified by IP using GFP antibody from transfected cells in a high-stringency buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 5 mM MgCl₂, NP-40 [0.5%], 1 mM DTT, and protease and phosphatase inhibitor). Highly purified Rho was bound to γ-labeled [P³²]GTP. The GTPase accelerating activity (Rho-GAP activity) of indicated DLC1 mutants was compared by incubating the DLC1 mutants with GTP-bound RhoA at 18°C with shaking while removing samples at the indicated time points. The guanidine nucleotides were separated by chromatography on cellulose filter paper dissolved in buffer. The γ-P³² signal was then determined, and the net GTP hydrolysis was calculated.

PLA was used to visualize proximity colocalization (<40 nm) of DLC1 and SRC, and kinase-active SRC and vinculin, in NSCLC lines using the Duolink Detection kit (Olink Bioscience). The cells were fixed with 4% PFA for 20 min at RT and then incubated with 0.25% Triton X-100 for 5 min. After blocking with 3% BSA, cells were incubated overnight at 4°C with indicated primary antibodies. After washing, cells were incubated with secondary antibodies with PLA probes (MINUS probe-conjugated anti-rabbit IgGplus PLUS probe-conjugated anti-mouse IgG). Circularization and ligation of the oligonucleotides in the probes were followed by an amplification step. A complementary fluorescent-labeled probe was used to detect the product of rolling circle amplification. Slides were mounted with Duolink II Mounting Medium containing DAPI. Images were obtained with an LSM 780. The number of colocalization PLA dots in cells was counted using ImageJ software.

HEK 293T cells were cotransfected GFP-tagged DLC1-WT constructs with Flag tag, constitutive kinase-active SRC, or kinase-dead SRC, or DLC1-WT–transfected cells were either untreated or treated with the SRC inhibitor saracatinib. Lysates from the cotransfected or treated cells were IP with GFP antibody, and the immunopellets were resolved on a NuPage 4–12% BisTris gel. The Coomassie-stained DLC1 gel band was destained, and proteins were reduced, alkylated, and digested with trypsin or LysC, as described (Shevchenko et al., 2006). Digested peptides were further desalted using C18 Ziptip (EMD Millipore) and analyzed by an EASY-nLC 1000 nanoHPLC with a C18 Nano Trap Column and an C18 Nano analytical column with a stainless-steel emitter on Nanospray Flex Ion Sources, coupled online with a Q-Exactive mass spectrometer (Thermo Fisher Scientific). Liquid chromatography–tandem mass spectrometry data were searched against DLC1 protein sequence and human protein database using Proteome Discoverer 1.4 with variable modifications of carbamidomethylation, oxidation of methionine, and phosphorylation of serine, threonine, and tyrosine residues. For legitimate peptide identifications, we applied protein percolator with 1% false discovery rate cutoff (Elias and Gygi, 2010).

For soft agar assays, a 0.6% agar (BD Biosciences) base in RPMI-1640 medium was placed in 60-mm dishes for 1 h at RT. 10⁵ cells were mixed with complete medium containing 0.4% agar and placed over 0.6% basal agar in 60-mm dishes. Cells were grown in 0.9 µg/ml G418 RPMI-1640 medium with 10% FBS for 3 wk, and colonies were fixed, stained, photographed microscopically, and quantified with a colony counter.

Cell migration was measured by 6.5-mm-diameter Falcon cell culture inserts (8 µm pore size; BD Biosciences). Transiently transfected cells or stable clones were trypsinized, resuspended in serum-free RPMI-1640 medium, and transferred to the upper chamber (7.5 × 10⁴ cells in 300 µl). 600 µl of 10% FBS in RPMI-1640 was placed in the lower chamber. After an 18-h incubation, the cells remaining on the upper surface of insert were removed three times with a cotton swab moistened in PBS. Migrated cells on the lower surface were fixed in methanol for 20 min at RT followed by staining with 2% crystal violet (Sigma-Aldrich) in methanol for 30 min, distained, examined, and photographed by microscopy. For quantification, migrated cells were solubilized with 1% Triton X-100 and counted in a spectrophotometer at OD 590 nm.

The mouse studies were approved by the National Cancer Institute Animal Care and Use Committee and conducted in compliance with the approved protocols. For tumor xenograft studies, H358 stable clones expressing GFP, GFP-tagged DLC1-WT, DLC1-Y451F, DLC1-Y451D, DLC1-Y701F, DLC1-Y701D, DLC1-2F, DLC1-2D, and DLC1-R718A were trypsinized, washed with cold PBS, diluted to 10⁸ cells/ml with serum-free medium/Matrigel basement membrane matrix (BD Biosciences) at a ratio of 3:1, and injected subcutaneously into NOD-SCID mice (10⁷ cells/injection). The animals were monitored for tumor growth, and tumor masses were weighed (in grams) 6 wk after injection.

For the treatment of mice with xenograft tumors, H358 stable clones expressing indicated GFP-tagged DLC1 constructs were trypsinized, washed with cold PBS, diluted to 10⁸ cells/ml with serum-free medium/Matrigel basement membrane matrix (BD Biosciences) at a ratio of 3:1, and injected subcutaneously into NOD-SCID mice (10⁷ cells/injection for GFP control and DLC1-2D groups; 5.0 × 10⁷ cells/injection for DLC1-WT and DLC1-2F groups). When tumors were 1.0–1.5 cm, mice were treated with SRC inhibitor, AKT inhibitor, and a combination of both drugs orally with 50 mg/kg or vehicle control for 5 d, and the remaining tumor tissues were then excised, weighed, and processed for biochemical assays. For MMTV-PyMT mice, tumors were randomly divided into two groups. Mice were treated orally with 50 mg/kg SRC inhibitor or vehicle control for five consecutive days. The remaining tumor tissues were then excised, weighed, and processed for biochemical assays.

At least two independent experiments were performed for all in vitro experiments. IBs were quantified by densitometric scanning using ImageQuant software. Results are expressed as mean densities ± SD from two or three experiments. All experiments were designed with matched control conditions within each experiment. Data distribution was assumed to be normal, but this was not formally tested. For statistical analysis, parametric two-tailed t tests were performed using Prism software (version 7.0a; GraphPad), and P < 0.05 was considered statistically significant.

Fig. S1 shows that SRC activity increases RhoA-GTP through DLC1. Fig. S2 shows that SRC colocalizes with DLC1, kinase-active SRC (pSRC-Y416) colocalizes with focal adhesion protein vinculin, and LPA increases RhoA-GTP independently of DLC1. Fig. S3 shows the SRC-dependent phosphorylation of the two tyrosines in DLC1 (Y451 and Y701) by mass spectrometry, and all three members of the DLC gene family, DLC1, DLC2, and DLC3, are SRC substrates. Fig. S4 shows that RhoA-GTP and its downstream effectors are regulated by SRC phosphorylation of DLC1. Fig. S5 shows that saracatinib and MK-2206 treatment induces cellular senescence and apoptosis in DLC1-WT tumors, and combined treatment with bosutinib and perifosine suppressed tumor growth more strongly than the individual inhibitors.

We thank the National Cancer Institute Center for Cancer Research Imaging Core Facility for confocal microscopy; Michael Mowat (University of Manitoba, Winnipeg, Canada) for the GFP-DLC2 construct; Joan Brugge (Harvard Medical School, Boston, MA) and Curt Harris (National Cancer Institute, Bethesda, MD) for H1703 and H358 cell lines; and Kent W. Hunter (National Cancer Institute, Bethesda, MD) for the MMTV-PyMT transgenic mice.

This research is supported by the Intramural Research Program, National Institutes of Health, National Cancer Institute, and Center for Cancer Research.

The authors declare no competing financial interests.

Author contributions: B.K. Tripathi and D.R. Lowy conceived the project, designed the experiments, analyzed the data, and wrote the manuscript. B.K. Tripathi performed most of the experiments with the help of M. Anderman. X. Qian engineered the DLC1 plasmid constructs. M. Zhou performed mass spectrometric analyses. D. Wang performed bioinformatics analyses of proteomics data. A.G. Papageorge helped with in vitro RhoA-GTP binding assay.