Abl tyrosine kinases regulate cell–cell adhesion through Rho GTPases
-
Edited by Joan S. Brugge, Harvard Medical School, Boston, MA, and approved September 14, 2007 (received for review April 3, 2007)
Abstract
Adherens junctions are calcium-dependent cell–cell contacts that link neighboring cells through cadherin receptors. Coordinated regulation of the actin cytoskeleton by the Rho GTPases is required for the formation and dissolution of adherens junctions. However, the pathways that link cadherin signaling to cytoskeletal regulation remain poorly defined. Here we identify the Abl family kinases as critical mediators of cadherin-mediated adhesion. Endogenous Abl family kinases, Abl and Arg, are activated and required for Rac activation after cadherin engagement and regulate the formation and maintenance of adherens junctions in mammalian cells. Significantly, we show that Abl-dependent regulation of the Rho–ROCK–myosin signaling pathway is critical for the maintenance of adherens junctions. Inhibition of the Abl kinases in epithelial sheets results in the activation of Rho and its downstream target ROCK, leading to enhanced phosphorylation of the myosin regulatory light chain. These signaling events result in enhanced stress fiber formation and increased actomyosin contractility, thereby disrupting adherens junctions. Conversely, Arg gain of function promotes adherens junction formation through a Crk-dependent pathway in cells with weak junctions. These data identify the Abl kinases as a regulatory link between the cadherin–catenin adhesion complex and the actin cytoskeleton through regulation of Rac and Rho during adherens junction formation, and also reveal a functional link between Abl and Rho that is essential for adherens junction stability.
Dynamic regulation of the actin cytoskeleton is required for the formation and dissolution of intercellular adhesions during tissue morphogenesis (1) and pathological processes such as tumor invasion and metastasis (2, 3). The formation of cell–cell contacts is dependent on the assembly of adherens junctions (1), which are specialized structures that link transmembrane cadherins in neighboring cells to the actin cytoskeleton (1). The extracellular domain of the cadherins cluster in a calcium-dependent manner and trigger association of the cadherin cytoplasmic domain with β-catenin, which in turn binds to α-catenin, a protein that links the cadherin–catenin complex to the actin cytoskeleton (1). Disruption of the actin cytoskeleton, genetic inactivation of actin regulatory proteins, or disruption of the link between the cadherin–catenin complex and the actin cytoskeleton all result in a loss of adherens junctions (4–6). Adherens junction formation, stability, and dissolution are processes regulated by Rho family GTPases (7). However, little is known about the mechanisms that link cadherin-mediated signaling to regulation of Rho GTPases, leading to dynamic reorganization of the actin cytoskeleton.
Abl (Abl1) and Arg (Abl2) define a family of nonreceptor tyrosine kinases characterized by unique carboxy-terminal, actin-binding domains that can bundle actin (8). The Abl kinases regulate a variety of cytoskeletal processes downstream of receptor tyrosine kinases and integrins (9–14). We have shown that Abl kinases play a role in the regulation of intercellular signals at the neuromuscular junction (NMJ) and at sites of contact between invading bacterial pathogens and mammalian host cells (9, 15). Genetic studies in Drosophila and mice support a role for Abl kinases in the regulation of epithelial morphogenesis (16–18). Mice lacking Abl and Arg die before embryonic day 11 and display collapse of the neural tube, which is secondary to the disruption of the apical actin latticework in neuroepithelial cells (18). Based on the unique properties of the Abl kinases, we asked whether Abl and Arg regulate the formation and/or maintenance of adherens junctions in mammalian cells and sought to define a pathway whereby the Abl kinases modulate these processes. Here we identify a role for Abl kinases in the regulation of cell–cell adhesion through modulation of Rho family GTPases.
Results
Abl Kinases Are Required for Cell–Cell Adhesion.
To determine whether the Abl kinases are involved in the regulation of cadherin-mediated cell–cell adhesion, we first examined the consequences of genetic inactivation of Abl and Arg on the formation of N-cadherin-dependent cell–cell junctions in mouse embryo fibroblasts (MEFs) (19). Abl−/−Arg−/− MEFs were reengineered to reexpress Abl and Arg (13, 15). Cells expressing Abl and Arg displayed continuous cell–cell junctions similar to wild-type (WT) MEFs as visualized by staining for N-cadherin, β-catenin, and α-catenin (Fig. 1 A Upper). In contrast, adherens junction proteins failed to accumulate at sites of cell–cell contact in Abl−/−Arg−/− MEFs (Fig. 1 A Lower). Loss of Abl and Arg did not alter the expression levels of adherens junction proteins (Fig. 1 B). To assess the role of Abl kinases in the formation of adherens junctions, cell monolayers were placed in low-calcium media to dissolve preexisting cell–cell junctions, and then were switched to high-calcium media to activate cadherin-mediated intercellular adhesion. Notably, >90% of cells expressing Abl/Arg formed adherens junctions after 4 h in high-calcium media (Fig. 1 C Upper). In contrast, Abl−/−Arg−/− cells displayed a disorganized N-cadherin-staining pattern and were unable to form discrete adherens junctions even 4 h after addition of high-calcium medium (Fig. 1 C Lower). These data reveal that Abl kinases are required for the formation of adherens junctions in fibroblasts.
Loss of Abl family kinases in fibroblasts disrupts N-cadherin-based adhesion. (A and B) Abl/Arg-null MEFs transduced with retroviruses encoding Abl and Arg or vector alone were grown to near confluency and either fixed and stained with antibodies to N-cadherin, β-catenin, or α-catenin (red) and counterstained with DAPI (blue) (A) or lysed to evaluate adherens junction protein expression by blotting with the indicated antibodies (B). (C) Abl/Arg- or vector-reconstituted null MEFs were grown to subconfluency and subjected to calcium switch to stimulate adherens junction formation. Cells were fixed and stained with antibody against N-cadherin (red) and counterstained with DAPI (blue).
We next examined whether Abl kinase activity plays a role in the regulation of epithelial cell–cell junctions. Confluent NBT-II rat epithelial cells were stimulated to form adherens junctions by calcium switch in the presence or absence of STI571, a pharmacological inhibitor of the Abl kinases (20, 21). In control cells, β-catenin accumulated at sites of cell–cell contact after incubation in high-calcium media (Fig. 2 A Upper). In contrast, cells treated with the Abl kinase inhibitor failed to induce β-catenin-containing protrusions and to form adherens junctions (Fig. 2 A Lower). Abl kinases were markedly inhibited by 10 μM STI571 as determined by the reduction of endogenous CrkL phosphorylation on Y207, the Abl-specific site (Fig. 2 B). To determine whether Abl/Arg proteins are required for intercellular adhesion in E-cadherin-expressing epithelial cells, Abl/Arg expression was down-regulated by using RNAi (Fig. 2 C). NBT-II cells transfected with Abl, Arg, or both Abl- and Arg-directed siRNAs showed decreased levels of β-catenin containing cell–cell junctions. To determine whether this effect is observed in other E-cadherin-expressing epithelial cells, Abl/Arg expression was down-regulated by RNAi in MCF-10A cells [Fig. 2 D and supporting information (SI) Fig. 8C]. MCF-10A cells transfected with Abl- and Arg-directed siRNAs showed decreased levels of E-cadherin and catenins at sites of cell–cell contact or lacked cell–cell contacts completely with no change in adherens junction protein expression (Fig. 2 D and SI Fig. 8). These data support a critical role for Abl family protein and kinase activity in the formation of epithelial adherens junctions.
Inhibition of Abl kinase activity or protein impairs adherens junction formation in epithelial cells. (A) Confluent NBT-II cells were subjected to calcium switch in the absence or presence of 10 μM STI571. Cells were fixed and stained with antibody against β-catenin (red) and counterstained with DAPI (blue). (B) Tyrosine phosphorylation of the endogenous Abl substrate CrkL at the Abl-specific site (Y207) is inhibited with 10 μM STI571. (C) Anti-β-catenin immunostaining of NBT-II cells transfected with control or Abl- and/or Arg-directed siRNAs. (D) Anti-β-catenin immunostaining of MCF-10A cells cotransfected with siGLO RISC-free siRNA (to identify transfected cells) and with control or Abl- and Arg-directed siRNAs. Images of β-catenin (green) and siGLO oligo (red) were merged. (Scale bars: 20 μm.)
Abl Kinases Are Required for the Maintenance of Adherens Junctions.
Adherens junctions are dynamic structures that undergo constant remodeling (22). To test the involvement of Abl kinases in adherens junction stability, Abl activity was inhibited with STI571 after formation of confluent monolayers of NBT-II epithelial cells. Treatment with STI571 for 3 h resulted in the relocalization of β-catenin from adherens junctions to the cytoplasm (Fig. 3 A). After 6 h of STI571 treatment, only 40% of the cells retained adherens junctions (Fig. 3 A and B). Adherens junction protein expression was unaffected by Abl kinase inhibition for ≤6 h (Fig. 3 C). Similarly, STI571 treatment of MCF-10A cells promoted mislocalization of adherens junction proteins (Fig. 3 D) with no change in E-cadherin and α-catenin protein levels (Fig. 3 E) or the integrity of the E-cadherin–α-catenin complex (Fig. 3 F). Moreover, we did not observe any changes in the tyrosine phosphorylation of the cadherin–catenin complex in the absence of Abl kinases or after inhibition of Abl kinase activity (data not shown).
Inhibition of Abl kinase activity impairs accumulation of E-cadherin–catenin complexes at cell–cell contacts. (A) Confluent NBT-II cells were treated with 10 μM STI571 for the indicated times, fixed and stained for β-catenin (red), and DAPI (blue). (B) β-catenin staining at sites of cell–cell contact was analyzed in four independent experiments. *, P = 0.0004, indicating a significant decrease in adherens junctions in STI571-treated cells versus controls. (C) Confluent NBT-II cells were untreated or treated with 10 μM STI571 for 6 h. Total lysates were examined by immunoblotting with antibodies against the indicated proteins. (D) Confluent MCF-10A cells were treated with 10 μM STI571 for the indicated times. Cells were fixed and stained for the indicated proteins. (E) Confluent MCF-10A cells were untreated or treated with 10 μM STI571 for the indicated times. Total lysates were examined by immunoblotting with antibodies against the indicated proteins. (F) To evaluate integrity of the cadherin–catenin complex in response to Abl kinase inhibition, the level of α-catenin bound to E-cadherin was assessed by immunoprecipitation of E-cadherin from 500 μg of lysate and immunoblotting for α-catenin. Blots were stripped and reprobed with an antibody against E-cadherin. (Scale bars: 20 μm.)
Cell–Cell Adhesion Leads to Abl Kinase Activation and Recruitment to Sites of Cell–Cell Contact.
To test whether Abl kinases are catalytically activated by cell–cell adhesion, we analyzed the phosphorylation of the Abl substrates CrkII or the related CrkL by using phosphospecific antibodies against the Abl-specific sites, which have been shown to be unphosphorylated in Abl−/−Arg−/− MEFs or in cells lacking Abl kinase activity (15, 23). Abl kinase activity was up-regulated by 5 min after calcium switch to induce cadherin-mediated cell–cell adhesion, and Abl activation was sustained for 30 min in WT MEFs (Fig. 4 A). Similar kinetics were observed in NBT-II cells by using in vitro kinase assays with GST-Crk to measure Abl/Arg kinase activity (Fig. 4 B) (24). To eliminate the possible contribution of intercellular adhesion-independent effects, we compared Abl activation in sparse versus confluent cultures and found that Abl kinase activity is unaltered in the absence or presence of calcium in sparse cultures (Fig. 4 C and D). In contrast, Abl kinase activity increases in a time-dependent manner after calcium switch in confluent cell monolayers (Fig. 4 C and D). Thus, regulation of Abl kinase activity after calcium switch is dependent on cadherin-mediated cell–cell contacts.
Cell–cell adhesion regulates the activity of endogenous Abl kinases. (A) Activation of Abl/Arg kinases in response to calcium switch in wild-type (WT) MEFs was assessed by immunoprecipitation of Crk and immunoblotting with a phosphospecific antibody to Y221. Blots were stripped and reprobed with an anti-Crk antibody to demonstrate equal loading. Results shown are representative of three independent experiments. (B) Confluent NBT-II epithelial cells were subjected to calcium switch, Abl or Arg protein was immunoprecipitated as indicated by using specific antibodies, and in vitro kinase assays were performed by using GST-Crk as a substrate. Abl/Arg protein expression in whole-cell lysates was assessed by immunoblotting with 8E9 monoclonal antibody. (C) Abl/Arg kinase activity in response to calcium switch was assayed in confluent versus sparse epithelial cell cultures. (D) Abl kinase activity was quantitated by using densitometry with pCrkL levels indexed against low calcium for each condition. Results represent three independent experiments. *, P < 0.05. Fold changes in pCrk or pCrkL levels are indicated.
To determine whether Abl and Arg are recruited to sites of cell–cell contact to participate in signaling, we generated Madin– Darby Canine Kidney (MDCK) cell lines expressing GFP-tagged versions of the Abl kinases and examined the localization of these kinases during adherens junction formation. Arg accumulated at sites of cell–cell contact and in the cytoplasm in both confluent cells, and in cells stimulated to reform adherens junctions by switch to high-calcium medium (SI Fig. 9A). Similar results were observed for GFP-Abl (data not shown). In confluent MCF-10A cells, Arg also colocalized with β-catenin at sites of cell–cell contact (SI Fig. 9C), and E-cadherin and β-catenin coimmunoprecipitated with GFP-Arg (SI Fig. 9B). Disruption of cadherin-mediated cell–cell adhesion (low calcium) resulted in redistribution of Arg from adherens junctions to the cytoplasm (SI Fig. 9 A and C). After calcium switch, GFP-Arg relocalized in lamellipodial extensions and at sites of nascent cell–cell contacts (SI Fig. 9C). These data further support a role for Abl and Arg in adherens junction regulation because both proteins are recruited to sites of cell–cell contact, are associated with E-cadherin–β-catenin complexes, and are activated in response to cadherin engagement.
Abl Kinases Regulate Rac Activation in Response to Cadherin Engagement Through Crk/CrkL.
To test whether Abl kinases are sufficient to promote junction formation, we expressed a constitutively active mutant form of Arg (ArgPP) in HeLa cells (25). These cells express N-cadherin and exhibit weak, immature adherens junctions. ArgPP expression resulted in enhanced, continuous staining of β-catenin at sites of cell–cell contact (SI Fig. 10A). Thus, Abl kinases promote N-cadherin-mediated adhesion in cells with weak junctions.
Because formation of cell–cell junctions correlates with increased phosphorylation of Crk/CrkL proteins at the Abl-specific sites, we next examined whether Crk family proteins play a role in ArgPP-induced adherens junction strengthening by analyzing the consequences of Crk/CrkL siRNA-mediated knockdown. We observed a marked reduction in the integrity of adherens junctions in ArgPP-expressing cells transfected with Crk/CrkL siRNAs (SI Fig. 10 B and C). Similarly, junction integrity was markedly impaired in NBT-II cells transfected with Crk/CrkL siRNAs (SI Fig. 10D). These data reveal a critical role for the Abl kinases in adherens junction formation in mammalian cells and suggest that this effect may be mediated, in part, by Crk proteins.
The Rac GTPase has been implicated in the formation of adherens junctions (26, 27). In agreement with previous findings (27), we observed that Rac was activated by 5 min of cadherin engagement (Fig. 5 A). Basal levels of activated Rac were unchanged in confluent monolayers of WT or Abl−/−Arg−/− MEFs grown in calcium-rich media (data not shown). Significantly, Rac activation induced by adherens junction formation was delayed and markedly reduced in Abl−/−Arg−/− cells (Fig. 5 A and B) and in WT MEFs subjected to calcium switch in the presence of the Abl kinase inhibitor STI571 (Fig. 5 C). Previous studies have linked tyrosine phosphorylation of Crk to Rac activation after bacterial infection and integrin engagement (15, 28). We found that overexpression of a Crk mutant lacking the Abl-specific phosphorylation site in MEFs decreased Rac activation during adherens junction formation (Fig. 5 D and E). The inhibitory effect of the Crk mutant was similar to that induced by Abl kinase inhibition. These data suggest that Abl-mediated phosphorylation of Crk is required for maximal activation of Rac downstream of cadherin engagement.
Abl kinases regulate Rac activation in response to cadherin engagement. (A) Rac activation in serum-starved WT or Abl−/−Arg−/− MEFs grown to confluency and subjected to calcium switch. (B) Rac activation was quantitated by using densitometry with Rac-GTP levels indexed against low calcium in each condition. Results represent three independent experiments. *, P < 0.05. (C) Rac activation in serum-starved WT or STI571-treated MEFs grown to confluency and subjected to calcium switch. (D) Rac activation in serum-starved WT MEFs expressing vector (control) or CrkYF was analyzed in cells grown to confluency and subjected to calcium switch. (E) Rac activation was quantitated by using densitometry with Rac-GTP levels indexed against low calcium in each condition. Results represent four independent experiments. *, P < 0.05.
Abl Kinases Regulate the Architecture of the Actin Cytoskeleton in Epithelial Cell Sheets by the Rho–ROCK Pathway.
The Abl kinases regulate multiple cytoskeletal processes such as membrane ruffling, chemotaxis, and cell spreading (29). Dynamic regulation of the actin cytoskeleton is required for the formation, stability, and dissolution of adherens junctions (5, 6, 29, 30). Therefore, we tested whether Abl/Arg kinase inhibition could alter the cytoskeletal architecture of confluent cells. Epithelial cell clusters display thick cortical bundles of actin at cell–cell borders (Fig. 6 A). In contrast, cells treated with the Abl kinase inhibitor showed decreased cortical actin, increased stress fiber formation (accompanied by the appearance of gaps between cells), and increased numbers of spiky membrane protrusions (Fig. 6 A). The enhanced formation of stress fibers in STI571-treated cells suggested that inhibition of Abl kinases may result in increased Rho activity. Indeed, cellular levels of Rho-GTP increased after STI571 treatment, which correlated with the appearance of stress fibers (Fig. 6 A–C). To determine whether loss of Abl/Arg proteins could similarly affect cadherin-mediated regulation of Rho activity, we examined Rho activation in epithelial cells where Abl/Arg proteins were knocked down by RNAi. In agreement with previous findings (27), cadherin engagement suppressed Rho activity in control epithelial cells (Fig. 6 E and F). In contrast, Rho inhibition after cadherin engagement was not observed in the Abl/Arg siRNA-treated confluent monolayers (Fig. 6 E and F). Thus, the Abl kinases are required for cadherin-dependent inhibition of Rho activity.
Abl kinases regulate the actin cytoskeleton and Rho activity in epithelial cell monolayers. Confluent NBT-II cells were treated with 10 μM STI571 for the indicated times. (A) Staining with AlexaFluor 488-conjugated phalloidin (green) and DAPI (blue). (B) Lysates were analyzed for Rho activity by using the EZ-Detect Rho activation kit. Total Rho protein levels were analyzed by immunoblotting. (C) Rho activity was quantitated by using densitometry with Rho levels indexed against the untreated condition. Results represent three independent experiments. *, P < 0.05. (D) Cells were fixed and stained with a phosphospecific antibody to pMLC (red) and counterstained with DAPI (blue). (E) MCF-10A cells were transfected with control or Abl/Arg-directed siRNA oligos, grown to confluence, and serum-starved overnight. Cells were subjected to calcium switch, and Rho activity was assayed. (F) Rho activity was quantitated by using densitometry with Rho levels indexed against the untreated condition. Results represent three independent experiments. *, P < 0.05. (Scale bars: 20 μm.)
Rho activation may result in increased actomyosin contractility mediated through phosphorylation of the myosin regulatory light chain (MLC) of myosin II by Rho-activated kinases (31). Inhibition of Abl family kinases resulted in a time-dependent increase of MLC phosphorylation and enhanced accumulation of phospho-MLC at the cell periphery, which correlated with increased formation of stress fibers (Fig. 6 D). Similarly, Abl kinase inhibition resulted in enhanced phosphorylation of MLC as assessed by immunoblotting to detect phosphoserine 19 of MLC (data not shown) (32).
Rho signaling has been shown to regulate both the formation and dissolution of adherens junctions, and activation of the Rho effector ROCK has been shown to disrupt adherens junctions in epithelial cells (2, 3, 30, 33–36). To test whether the observed effects of Abl/Arg kinase inhibition are dependent on ROCK kinase activity, we examined whether the phenotypes induced by loss of Abl kinase activity could be reversed by inhibition of ROCK. Consistent with previous findings (3), cells treated with the ROCK kinase inhibitor Y27632 for 1 h did not display adherens junctions defects, but did show a marked decrease of stress fibers at the cell center (Fig. 7 A). Significantly, cells treated with both STI571 and Y27632 failed to display any of the aberrant morphology characteristic of STI571-treated cells, with β-catenin and actin remaining localized to sites of cell–cell contacts. Also, stress fiber formation was inhibited in these cells (Fig. 7 A–C). Notably, whereas inhibition of the Abl kinases resulted in a 60% decrease in the percentage of cells with mature, continuous adherens junctions, this effect was blocked by the inhibition of ROCK (Fig. 7 C). Further, dual inhibition of both Abl and ROCK kinases abolished the increase in pMLC and the enhanced localization of MLC protein to the cell membrane (data not shown). Thus, inhibition of Abl kinase activity leads to up-regulation of Rho–ROCK signaling, enhanced MLC phosphorylation, and increased actomyosin contractility, thereby promoting the dissolution of cell–cell adhesions. Together these data support a model whereby endogenous Abl kinases normally function to modulate Rho–ROCK activity and maintain cell–cell contacts in epithelial cell sheets (SI Fig. 11).
Disruption of adherens junctions in response to inhibition of Abl kinase activity is reversed by inhibition of ROCK kinase activity. (A) Confluent NBT-II cells were treated with 10 μM STI571, 10 μM Y27632, or both STI571 and Y27632 for 1 h; fixed and stained with phalloidin (green) or anti-β-catenin (red); and counterstained with DAPI (blue). (B) Stress fiber formation was analyzed by phalloidin staining. Results represent the mean of four independent experiments. *, significant increase (control vs. STI571 treatment, P < 0.001) or decrease (control vs. Y27632 treatment, P < 0.05; control vs. STI571 + Y27632 treatment, P < 0.05) in stress fibers. **, significant decrease (P < 0.001) in stress fibers in cells treated with both STI571 and Y27632 together compared with STI571-treated cells. (C) The number of cells with mature adherens junctions marked by continuous β-catenin staining at sites of cell–cell contact was quantified in four independent experiments, in which nine random fields were examined per condition. *, significant decrease (P < 0.001) or increase (P < 0.001) in adherens junctions in cells treated with STI571 in the absence or presence of Y27632. (Scale bars: 20 μm.)
Discussion
Here we identify the Abl family of tyrosine kinases as critical mediators of cadherin-dependent adhesion in mammalian cells through regulation of Rho family GTPases. We have used genetic ablation and pharmacological inhibition to show that Abl kinases are required for the formation of adherens junctions and that loss of Abl kinases impairs Rac activation induced by cadherin engagement. Additionally, we show that Abl gain of function promotes adherens junction formation in cells with weak junctions and that Crk/CrkL proteins are required for this strengthening phenotype.
Significantly, we show that Abl kinases are activated by cadherin engagement, are recruited to nascent junctions, and are present in E-cadherin–β-catenin complexes. Additionally, we have shown that Abl kinase activity is required for the maintenance of adherens junctions in epithelial cell sheets because inhibition of these kinases results in the redistribution of adherens junction components from sites of cell–cell contact to the cytoplasm. Abl kinases are required for maximal Rac activation induced by cadherin-mediated adhesion, and modulate cell–cell adhesion through other pathways. Inhibition of the Abl kinases in epithelial sheets results in activation of Rho and its downstream target ROCK, leading to enhanced phosphorylation of the MLC. Activation of this signaling pathway leads to enhanced stress fiber formation and increased actomyosin contractility, thereby disrupting adherens junctions. These findings identify a role for the Abl kinases as regulators of intercellular adhesion through their ability to modulate both Rac and Rho.
Notably, we showed that Abl kinases activated by cadherin engagement phosphorylate Crk/CrkL proteins, and that both Abl and Crk family proteins are required for adherens junction integrity in a pathway upstream of Rac. We and others have shown that Crk phosphorylation on the Abl-specific site is required for Rac activation and membrane localization (Fig. 5 E) (15, 28). The mechanism by which Crk phosphorylation by Abl results in Rac activation may involve the recruitment of the Crk-binding protein DOCK180, a guanine nucleotide exchange factor (GEF) for Rac (37). Alternatively, Abl may activate Rac by phosphorylating Sos-1, which was reported to stimulate the Rac-GEF activity of Sos-1 (14).
Abl kinases are required for both activation of Rac and inhibition of Rho at adherens junctions, and it is possible that these events may be coordinately linked. Abl may down-regulate Rho signaling indirectly by activating a Rac–ROS pathway (38) or directly by regulating the activities of a RhoGAP or RhoGEF. In this regard, activated RacV12 was reported to inhibit Rho by p190 RhoGAP (38), and p190 RhoGAP was shown to be phosphorylated in an Arg-dependent manner in response to integrin engagement in neuronal cells (10). However, we have not detected any changes in p190 RhoGAP tyrosine phosphorylation in the absence of Abl/Arg kinases or upon STI571 treatment in epithelial cells or fibroblasts (data not shown). Moreover, in agreement with published data (39), we found that p190 RhoGAP does not localize to adherens junctions in the cells analyzed here (data not shown). Thus, our data suggest that Abl substrates other than p190 RhoGAP are likely to regulate adherens junction formation and maintenance.
Together our findings implicate the Abl family of tyrosine kinases as regulators of intercellular adhesion in epithelial cells and fibroblasts and support the model depicted in SI Fig. 11. These data and our previous results demonstrating that Abl kinases are required for the formation of the NMJ suggest that Abl kinases may function to modulate adhesive interactions between the same or different cell types. Changes in the activation of the Abl kinases also may modulate cell–cell adhesion during tumor progression and metastasis. In this regard, enhanced expression of Abl or Arg has been reported in a subset of metastatic colon carcinomas, pancreatic ductal carcinoma, and renal medullary carcinoma (40–42). Excessive Abl/Arg kinase activity may disturb the proper balance of Rho and Rac activation and induce altered cytoskeletal reorganization, thereby affecting cell–cell adhesion.
Materials and Methods
Antibodies and Reagents.
The following antibodies and chemical reagents were used: N-cadherin, β-tubulin, and actin (Sigma–Aldrich, St. Louis, MO); pCrkL Y207, pCrk Y221, and pMLC polyclonal (Cell Signaling Technology, Danvers, MA); E-cadherin, β-catenin, p120 catenin, and Crk (BD Biosciences Transduction Laboratories, Lexington, KY); α-catenin (Zymed Laboratories, South San Francisco, CA); Abl K12 and CrkL (Santa Cruz Biotechnology); Abl 8E9 (BD Biosciences PharMingen, San Diego, CA); Rac and Rho (Pierce Chemical, Rockford, IL); GFP (Roche Diagnostics, Indianapolis, IN); CY3-donkey anti-mouse, CY3-donkey anti-rabbit, and CY2-donkey anti-mouse (Jackson ImmunoResearch Laboratories, West Grove, PA); Alexa488-conjugated phalloidin, protein A-Sepharose, protein G-Sepharose, HRP-linked protein A, and ECL Western blotting reagents (Amersham Biosciences, Piscataway, NJ); HRP-linked goat anti-mouse IgG (Santa Cruz Biotechnology); HRP-linked goat anti-rabbit IgG and DAPI (Molecular Probes, Eugene, OR); and Y27632 (Calbiochem, San Diego, CA). Antibody to Arg was produced by immunizing rabbits with a peptide specific for a unique C-terminal region (9). STI571 was a gift from B. Druker (Oregon Health Sciences University, Portland, OR). The pK1 Arg GFP plasmid was obtained from A. Koleske (Yale University, New Haven, CT). pK1 ArgPP (25) and MigR1 CrkYF (15) plasmids were generated as described.
Cell Lines.
Abl/Arg double-null MEFs were obtained from A. Koleske (Yale University) and reconstituted with Abl/Arg as previously described (9, 24). MEFs, HeLa, NBT-II, and MDCK cells were maintained in DMEM with 10% FBS (18, 24). MCF-10A human epithelial cells were maintained in DMEM/F12 with 5% donor horse serum, 20 ng/ml hEGF, 500 ng/ml hydrocortisone, 10 μg/ml insulin, and 100 ng/ml cholera toxin.
Retroviral Transduction.
Transduction of MDCK, HeLa, MCF-10A, and MEF cells was performed as described (43, 44). Cells transduced with GFP-Arg-infected cells were subjected to FACS sorting to select a heterogeneous GFP-positive population.
Cell Lysis, Immunoblotting, in Vitro Kinase Assays, and RhoGTPase Activation Assays.
Cells were lysed in either Triton lysis buffer [1% Triton X-100, 150 mM NaCl, 50 mM Tris (pH 7.5)] or RIPA buffer [0.5 M NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris (pH 8.0)] plus inhibitors (1 mM phenylmethylsulfonyl fluoride, 2 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 25 mM β-glycerophosphate, and 1 μg/ml each of leupeptin, aprotinin, and pepstatin A). In vitro kinase assays were performed as previously described (13). Rac/Rho activation assays were performed according to the manufacturer's instructions by using the EZ-Detect Rac or Rho activation kit (Pierce Chemical). ImageJ Software was used to quantify relative band intensity normalized to the time 0 time point. Control and experimental conditions at each time point were compared and analyzed for statistical significance by using ANOVA, followed by Student's paired t test (two-tailed analysis, Excel).
Immunofluorescence.
Cells plated on glass coverslips were fixed with 4% (wt/vol) paraformaldehyde for 10 min, permeabilized by the addition of 0.25% Triton X-100 for 5 min, and incubated in 2% BSA block in PBS. Antibodies were diluted in block as follows: N-cadherin (1:100), E-cadherin (1:200), β-catenin (1:300), α-catenin (1:100), 488 phalloidin (1:1,000), pMLC pAb (1:100), CY3 donkey anti-mouse secondary (1:1,000), CY3 donkey anti-rabbit secondary (1:1,000), CY2 donkey anti-mouse secondary (1:600), and DAPI (1:25,000). Fluorescence microscopy was carried out at 40× or 63× by using a Zeiss Axioskop microscope equipped with a Hamamatusu ORCA-ER digital camera (Carl Zeiss, Thornwood, NY) or at 40× or 63× by using a Zeiss Axiovert 200M equipped with AxioCamMRm high resolution. Images were analyzed by using Metamorph software (Universal Imaging Corporation, Downington, PA) or AxioVision 4.5 (Carl Zeiss). Statistical analysis of adherens junction integrity in untreated versus STI571-treated cells was performed by using Student's paired t test (two-tailed analysis, Excel or Graphpad software). All other statistical analyses were performed by using one-way ANOVA analysis followed by Bonferroni's multiple comparison test (Graphpad software).
Calcium Switch Experiments.
MEFs, NBT-II, MCF-10A, or MDCK cells were grown to confluency and washed with PBS, and DMEM containing 10% FBS and 2 mM EGTA was added overnight. The following day, medium was replaced with DMEM containing 10% FBS. For Abl kinase activity assays, NBT-II cells were seeded onto glass coverslips, grown to subconfluency, and washed with PBS, and medium was replaced with KBM-2 containing 30 μM calcium. The following day, medium was replaced with KBM-2, as described earlier, containing 1.8 mM calcium.
siRNA.
Cells were transfected with Oligofectamine (Invitrogen, Carlsbad, CA) and control or the indicated duplex oligonucleotides according to the manufacturer's instructions. siControl Non-Targeting siRNA Pool (D-001206-13), siGENOME SMARTpool reagents human Abl1(M-003100), rat Abl1 (M-090649-00), human Abl2(M-003101), rat Abl2 (M-082758-00), and ON-TARGETplus Duplex reagents human CrkL (J-012023-07), human Crk (J-010503-07), rat CrkL (J-087587-12), and rat Crk (J-090657-09) were purchased from Dharmacon RNA Technologies (Lafayette, CO). Cells were lysed 48 h after transfection. For immunostaining, siGLO RISC-free siRNA (Dharmacon RNA Technologies) was cotransfected to identify cells containing siRNAs.
Acknowledgments
We thank Mike Cook for FACS-based cell sorting, Drs. Anthony Koleske and Jeff Settleman for generously providing reagents, and Sam Johnson for microscopy advice. This work was supported by National Institutes of Health Grant CA70940 (to A.M.P.) and PhRMA Foundation and Department of Defense Breast Cancer Research Program Grant BC050405 (to N.L.Z.).
Footnotes
- †To whom correspondence should be addressed. E-mail: pende014{at}mc.duke.edu
-
Author contributions: A.M.P. designed research; N.L.Z. and M.P. performed research; N.L.Z. and A.M.P. analyzed data; and N.L.Z. and A.M.P. wrote the paper.
-
↵*Present address: National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD 20982.
-
The authors declare no conflict of interest.
-
This article is a PNAS Direct Submission.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0703077104/DC1.
- Abbreviations:
- GAP,
- GTPase-activating protein;
- GEF,
- guanine nucleotide exchange factor;
- MEF,
- mouse embryo fibroblast;
- MLC,
- myosin regulatory light chain;
- NMJ,
- neuromuscular junction.
- © 2007 by The National Academy of Sciences of the USA