The Bro1-domain-containing protein Myopic/HDPTP coordinates with Rab4 to regulate cell adhesion and migration

Cell adhesion and cell migration are essential for the development and coordinated function of multicellular organisms. Aberrant regulation of these processes often results in the progression of many diseases, including cancer invasion and metastasis. Accumulating evidence has indicated that dynamic and reversible protein tyrosine phosphorylation is essential for the regulation of cell migration and cell adhesion (Zamir and Geiger, 2001; Burridge et al., 2006; Huveneers and Danen, 2009). While many studies have been devoted to the role of protein tyrosine kinases in these processes, the function of protein tyrosine phosphatases (PTPs) in cell adhesion and migration remains unclear.

The dynamic change of integrin-mediated focal adhesions plays a critical role in cell adhesion and migration. Many focal adhesion regulators such as focal adhesion kinase (FAK), Src, p130Cas and paxillin are tyrosine phosphorylated (Panetti, 2002). The tyrosine phosphorylation of these proteins affects focal adhesion dynamics. Phosphorylation of tyrosine 397 in FAK promotes its association with Src, and the activated FAK–Src complex subsequently regulates focal adhesion dynamics by signaling downstream targets (Mitra and Schlaepfer, 2006). Several PTPs have been implicated in integrin signaling, cell adhesion and motility. One study has shown that SHP-2 phosphatase influences FAK activity (Yu et al., 1998). SHP-2 also promotes Src kinase activation by inhibiting Csk (Zhang et al., 2004). Depletion of PTP-PEST has been found to lead to the hyperphosphorylation of p130Cas, FAK and paxillin, and a marked increase in focal adhesions (Angers-Loustau et al., 1999). Moreover, PTP1B and PTPα, have also been found to regulate Src phosphorylation and integrin-mediated adhesion (Harder et al., 1998; Liang et al., 2005).

In Drosophila, a total of sixteen putative classical PTPs have been identified (Andersen et al., 2005). Compared to mammalian PTPs, Drosophila PTP family members are relatively simple, most containing only one gene corresponding to each subtype (except for DPTP10D and DPTP4E, which share similar domain structures). Therefore, Drosophila can serve as an excellent model system for the study of the physiological and developmental function of PTPs. While much research has been devoted to the function of receptor PTPs, the role of non-transmembrane PTPs (NT-PTPs) in Drosophila development remains unknown. One of the most well studied Drosophila NT-PTPs is Corkscrew (Csw). Csw is the ortholog of human SHP-2 which has two SH2 domains at the N-terminus and a PTP domain at the C-terminus. Csw functions as a downstream effecter of Sevenless PTK and is essential for the development of the R7 photoreceptor (Allard et al., 1996). Phenotypic analysis showed that Csw can also act downstream of many receptor tyrosine kinases, such as the Drosophila epidermal growth factor receptor (DER) and the fibroblast growth factor (Breathless) (Perkins et al., 1996; Feng, 1999). Protein tyrosine phosphatase-ERK/Enhancer of Ras1 (PTP-ER) has been shown to function as a negative regulator downstream of Ras1 and to be involved in RAS1/MAPK-mediated R7 photoreceptor differentiation (Karim and Rubin, 1999). PTP61F, the Drosophila ortholog of human PTP1B and TCPTP, has been reported to interact with Dock, an adapter protein required for axon guidance (Clemens et al., 1996). PTP61F has recently been shown to coordinate with dAbl in regulating actin cytoskeleton organization via reversible tyrosine phosphorylation of Abi and Kette (Huang et al., 2007; Ku et al., 2009). Moreover, dPtpmeg, a FERM and PDZ domain-containing NT-PTP, is reported to be involved in the formation of neuronal circuits in the Drosophila brain (Whited et al., 2007), though its molecular function in this process is not known.

To explore the functional role of Drosophila NT-PTPs in cell adhesion and migration, we performed genetic analyses to identify NT-PTPs that could modulate border cell migration during oogenesis. We found that Myopic (Mop), the Drosophila homolog of the human His domain phosphotyrosine phosphatase (HDPTP), plays an important role in maintaining border cell cluster integrity. Depletion of Mop altered the normal distribution of integrin receptor. While Mop has recently been reported to regulate EGFR and Toll receptor signaling (Miura et al., 2008; Huang et al., 2010), its molecular mechanism has remained elusive. This study found that Mop interacts with Rab4 GTPase in controlling integrin distribution and cell adhesion. We further demonstrated that human HDPTP is essential for the intracellular positioning of Rab4, integrin trafficking and cell migration. These findings provide some insight into the mechanisms underlying HDPTP in the regulation of cell adhesion and migration.

We analyzed the eight Drosophila non-transmembrane PTPs (NT-PTPs), including Ptp61F, dMeg2, CG7180, Myopic (Mop), dPtpmeg, dPez, PTP-ER and Csw, in Drosophila genome by multiple sequence alignment of their PTP domain. Based on their homology to the PTP domain of human PTP1B, 10 conserved motifs of the PTP domain were aligned (Fig. 1A). Among them, two catalytic essential residues, aspartic acid within the WPD loop (motif 8) and the active site cysteine within the signature motif HC(X)5R (motif 9), were identified. Interestingly, the critical aspartic acid residue (D) in the WPD loop of Mop, dPez and CG7180 were replaced with lysine (K), glutamic acid (E) and serine (S), respectively. Although the active site cysteine could be found in all NT-PTPs, the overall signature motif of Mop showed much greater sequence divergence than all the other NT-PTPs.

We next measured the tyrosine phosphatase catalytic activities of Ptp61F, dMeg2, dPtpmeg, dPez and Mop using in vitro phosphatase assays. The PTP domain of each NT-PTP was expressed in cultured Drosophila S2 cells, immunoprecipitated, and subjected to pNPP assay (Fig. 1B). PTP domain of dMeg2, dPtpmeg and Ptp61F showed detectable phosphatase activity by pNPP assay (Fig. 1B). It appears that dPtpmeg has much higher PTP activity than that of dMeg2 and PTP61F (Fig. 1C). However, the PTP activity of dPez and Mop were barely detectable by in vitro phosphatase assay (Fig. 1B). Our sequence and biochemical analysis suggested that dPez and Mop may not have tyrosine phosphatase enzyme activity.

To study the function of protein tyrosine phosphatases in collective cell migration, we used transgenic RNAi approach to systematically knockdown each of the Drosophila NT-PTPs, including Ptp61F, dMeg2, CG7180, Myopic, dPtpmeg, dPez, PTP-ER, and Csw, in migrating border cells during oogenesis. Border cells consist of a group of six to ten specialized migratory cells that are derived from the follicular epithelium in the developing egg chamber (Rørth, 2002; Montell, 2003). These cells adhere tightly to each other and migrate as a cluster toward oocyte. Using the GAL4-UAS targeted expression system, double-stranded RNA (dsRNA) targeting each NT-PTP was expressed under the control of the border cell-specific slbo–Gal4 driver. We found that RNAi-mediated downregulation of Drosophila NT-PTPs did not have an obvious effect on border cell migration, except in the case of dMeg2-RNAi, which resulted in a 23.3% (n = 219) migration delay (Fig. 2A). We next examined whether the NT-PTPs were involved in controlling border cell cluster integrity during collective migration. In control egg chambers (slbo–Gal4), border cells (marked by GFP) are tightly associated to each other during migration (Fig. 2B,C). Strikingly, Mop-RNAi knockdown resulted in a marked increase in cluster dissociation (26.5%, n = 276; Fig. 2B). Western blot analysis showed that Mop-RNAi could effectively reduce the expression levels of Mop (supplementary material Fig. S1A). Depletion of Mop caused the dissociation of one or two cells from border cell cluster and retained in nurse cells by stage 10; however, we did not observe obvious changes in the total number of border cells, compared with the control (Fig. 2C; supplementary material Fig. S1B). These results together suggest that Mop plays a role in cell adhesion, rather than cell invasiveness, during oogenesis. Moreover, the Mop-RNAi cluster dissociation defect was rescued by co-expression of full-length Mop, Mop-ΔPTP and Mop-C/S mutant (supplementary material Fig. S2), suggesting that the cell adhesion defect is PTP activity independent. Interestingly, overexpression of Mop with slbo–Gal4 disrupted the cluster integrity (30.7%, n = 189; supplementary material Fig. S2), indicating that a proper level of Mop expression is essential for maintaining border cell cluster integrity.

E-cadherin and integrin are adhesion-related molecules that have been reported to play a role in maintaining border cell cluster integrity (Llense and Martín-Blanco, 2008). Immunostaining analysis of slbo–Gal4 controls revealed that both DE-cadherin and βPS-integrin are enriched at cell-cell contacts of border cells (Fig. 3A,B). Depletion of Mop did not markedly affect the level and distribution of DE-cadherin at cell-cell contacts (Fig. 3B). However, depletion of Mop resulted in significant changes in integrin distribution (Fig. 3A). Most of the βPS-integrins appear to aggregate at the leading edges or cell periphery of migrating border cells rather than in cell-cell contacts. Clonal analysis of mop mutants further confirmed our observation. In mop homozygous mutant border cell clones (marked by the loss of GFP), βPS-integrins formed punctate aggregates in the peripheral region of mop mutant cells (Fig. 3C). We also found a slight reduction and diffuse localization of the endogenous DE-cadherin in mop mutant cells (Fig. 3C). Taken together, these findings show that Mop is involved in the control of cell adhesion during border cell cluster migration, possibly through the control of integrin localization. We next investigated the effect of Mop depletion in epithelial follicle cells surrounding the oocyte. DE-cadherin and βPS-integrin were enriched at cell-cell contacts in control cells (GFP positive; supplementary material Fig. S3A,C). In mop homozygous mutant follicle cell clones (GFP negative), βPS-integrin and DE-cadherin were diffusely distributed in the cell peripheral region (supplementary material Fig. S3A,C). A slight increase of DE-cadherin at cell-cell contacts was also observed in some mop mutant clones (supplementary material Fig. S3C,D). These results indicate that Mop is involved in regulating the distribution of integrin and DE-cadherin in non-migrating follicle cells.

We next test whether misexpressing integrin in migrating border cells may affect border cell cluster integrity. Strikingly, we found that expression of myospheroid (mys, encoding βPS-integrin) with slbo–Gal4 caused a marked increase in border cell dissociation, as compared with controls (Fig. 3D,E). Immunostaining analysis revealed that, unlike the control in which the βPS-integrin is mainly localized at cell-cell junctions, misexpression of mys resulted in an accumulation of βPS around the cell edges (supplementary material Fig. S4). Moreover, genetic analysis further showed that depletion of βPS-integrin (mys) could rescue the Mop-RNAi-induced border cell dissociation phenotype (Fig. 3E). These results together suggest that βPS-integrin mislocalization may contribute to the cluster dissociation caused by Mop depletion.

Integrin-related cell adhesion and migration is essential during many developmental processes. During wing development, wing imaginal discs fold into two layers of wing epithelia, dorsal and ventral wing epithelia, which will later adhere to each other via an integrin-mediated process (Bökel and Brown, 2002; Brower, 2003). Significantly, we found that ectopic expression of Mop under the control of engrailed–Gal4 (en–Gal4) induced a blistered-wing phenotype with high penetrance (90%, n = 165; compare supplementary material Fig. S5A,B). We further demonstrated that the Mop-induced wing cell adhesion defect could be rescued by specifically knocking down Mop at the same region (100%, n = 106; supplementary material Fig. S5C). These results indicate that Mop plays a role in integrin-mediated cell adhesion.

Recent studies have indicated that Mop plays a role in regulating cell surface receptor signaling. Miura et al. found that Mop acts as an endosomal protein in promoting EGF receptor signaling during photoreceptor differentiation (Miura et al., 2008). Moreover, Huang et al. has reported that Mop is required for Toll receptor innate immune signaling (Huang et al., 2010). They found the Bro1 domain of Mop to be required for Mop to function in Toll pathway activation, though the molecular mechanism remained unclear. We performed a yeast two-hybrid screen of a Drosophila embryo cDNA library using the Bro1 domain of Mop as bait (Fig. 4A). From a screen of about 3.6×10⁷ transformants, we identified several positive clones. Sequence analysis revealed that one of these clones encoded the Drosophila Rab4 GTPase (Fig. 4B). Rab4 has been implicated in the regulation of membrane receptor recycling from early endosomes in mammalian cells (Roberts et al., 2001; Grant and Donaldson, 2009). We then used GST pull down and co-immunoprecipitation assays to further examine the interaction between Mop and Rab4. GST–Mop–Bro1, but not GST, efficiently interacted with Rab4 (Fig. 4C). To determine whether full-length Mop interacts with Rab4, we transfected HA-tagged full-length Mop (HA–Mop), Bro1 domain of Mop (HA–Mop–Bro1), and Myc-tagged Rab4 GTPase (Myc–Rab4) into HEK293 cells, and Mop was immunoprecipitated with an anti-HA antibody. We found Rab4 to be co-immunoprecipitated with both full-length Mop and the Bro1 domain of Mop (Fig. 4D). Since Rab4 has been found to localize at early endosome and recycling endosome (van der Sluijs et al., 1992; Sönnichsen et al., 2000), we investigated whether Mop was also localized at these compartments. Immunofluorescence analysis showed Mop to be distributed at enlarged vesicle-like compartments and significantly colocalized with GFP–Rab4 in Drosophila S2R+ cells (Fig. 4E). We further tested the interaction between Rab4 and Mop by genetic analysis. Wing blistering defects induced by the overexpression of Mop could be rescued by knocking down Rab4 or co-expressing GDP-bound Rab4 (Rab4-SN; supplementary material Fig. S5D,E). Rab GTPases cycle between the GTP-bound active form and the GDP-bound inactive form (Stenmark, 2009). To find out whether Rab4 interacted with Mop in a guanine nucleotide-dependent manner, we examined the interaction of Mop with bacterially expressed GST–Rab4 in the presence or absence of nucleotide. As shown in supplementary material Fig. S6, Mop associated with nucleotide-free, GDP-bound and GTPγS-bound forms of Rab4. These data together indicate that Rab4 interacts with Mop in a guanine nucleotide-independent manner.

The Rab family of small GTPases acts as molecular switches that spatially and temporally regulate vesicle transport in the cell (Stenmark, 2009; Hutagalung and Novick, 2011). The binding of Mop to Rab4 in our study raised the question of whether Mop also interacts with other Rab proteins. Because Mop has been shown to colocalize with Rab5 on early endosomes (Miura et al., 2008), we investigated whether Mop and Rab5 also interacted with each other. Co-immunoprecipitation assays revealed that Mop also interacted with Rab5 but not with Rab7 and Rab11 (supplementary material Fig. S7). Because Rab GTPase-mediated endocytic trafficking has been reported to play an important role in regulating border cell migration (Assaker et al., 2010), we wanted to investigate whether Rab4, Rab5, Rab7 and Rab11 were involved in maintaining border cell cluster integrity. Expression of the dominant-negative Rab5-S43N, Rab7-T22N or Rab11-S25N mutant in the migrating border cells using slbo–Gal4 impaired border cell migration; however, none of them affected the border cell cluster integrity (Fig. 5A,B). Strikingly, expression of the dominant negative GDP-bound Rab4-S22N mutant, but not Rab4-WT or the GTPase-deficient Rab4-Q67L mutant, resulted in enhanced dissociation of border cells (Fig. 5B). Moreover, unlike other Rab GTPases we tested, Rab4-S22N mutant did not markedly affect border cell migration (more than 80% of border cells migrated normally; Fig. 5A). We further investigated the distribution of βPS-integrin in the control and Rab4-S22N expressing border cells. Similar to depletion of Mop, the expression of dominant negative Rab4 in border cells caused a redistribution of βPS-integrin from cell-cell junction to cell peripheral in migrating border cells (Fig. 5C). Taken together, our results further confirm the role that Rab4 and Mop play in regulating border cell association and integrin distribution during oogenesis.

To find out whether Mop–Rab4-mediated integrin trafficking is conserved in mammalian cells, we further investigated whether the Mop human ortholog HDPTP would interact with Rab4A GTPase. Co-immunoprecipitation assay on co-transfected cells demonstrated that FLAG-tagged HDPTP interacted with mammalian Rab4A (Fig. 6A), and immunofluorescence analysis revealed that HDPTP and Rab4 colocalized with each other (Fig. 6B), indicating that the interaction between HDPTP and Rab4 GTPase is evolutionarily conserved. It has been shown that HDPTP plays an essential role in the regulation of cargo sorting and endosome morphogenesis (Doyotte et al., 2008). We therefore investigated the effect of HDPTP depletion on Rab4 localization in HeLa cells (Fig. 6C,D). In control cells, GFP–Rab4A and the early endosome antigen 1 (EEA1) were dispersed throughout the cytoplasm (Fig. 6D). However, in the stable HDPTP knockdown cells, GFP–Rab4A and EEA1-positive endosomes were accumulated in the perinuclear region (Fig. 6D). The shHDPTP-induced change in Rab4A distribution could be reversed by the expression of wild-type HDPTP (data not shown). We also noted that the localization of the late endosomal marker Rab7 and the lysosomal marker LAMP-1 appeared to be normal when HDPTP was depleted (data not shown). These results together indicate that HDPTP may play a role in regulating the proper distribution of Rab4A in endosomal compartments.

Because mop mutation affects Drosophila βPS-integrin distribution during oogenesis, we checked whether knocking down HDPTP would have any influence on integrin receptor distribution. As shown in Fig. 7A, β1 integrin was dispersed over the cell surface and in the peripheral region of control cells. Significantly, knocking down HDPTP resulted in an increased accumulation of β1 integrins to the perinuclear region of HeLa cells (Fig. 7A). We then investigated the role of HDPTP in internalization and recycling of β1 integrin in HeLa cells. Quantitative internalization assay by flow cytometry revealed that the amount of β1 integrin endocytosed in HDPTP knockdown cells was slightly reduced compared to control cells (Fig. 7B). We further analyzed the effect of HDPTP knockdown on the recycling of integrins from endosomes to the plasma membrane. To monitor recycling of integrins from Rab4-mediated early endosomes, cell surface integrins were biotin-labeled and allowed to be endocytosed at 22°C for 15 min. After we removed the integrins remaining on the cell surface, we chased cells at 37°C for various time periods to permit recycling of internalized integrins. In shLuc cells, ∼60% of the internalized α5β1 recycled to the membrane after 5 min (Fig. 7C). Depletion of HDPTP significantly reduced the amount of α5β1 recycled to the plasma membrane (Fig. 7C). We then investigated the effect of knocking down HDPTP on the recycling of integrin from the perinuclear recycling compartments. Cells were surfaced labeled and integrins were internalized at 37°C for 30 min. The recycling rate was slightly reduced in HDPTP knockdown cells (Fig. 7D). Taken together, our results indicate that HDPTP plays an important role in the recycling of α5β1 integrin from early endosomes.

Integrins are adhesion receptors that connect the extracellular matrix (ECM) and the actin cytoskeleton to control cell movement (DeMali et al., 2003). Upon binding to ECM molecules, integrins induce the formation of focal complexes or focal adhesions which are found at the ends of actin stress fibers (Carragher and Frame, 2004; Romer et al., 2006). Since HDPTP plays an essential role for integrin distribution and trafficking, we asked whether HDPTP would be important for the organization of focal adhesions and actin cytoskeleton. The MDA-MB231 breast cancer cells stably expressing shHDPTP (or shLuc) were stained with anti-paxillin antibody and phalloidin to visualize focal adhesion and actin structure, respectively. Interestingly, in contrast to the prominent actin stress fibers in the control cells, the F-actin was organized into short and thin filaments in the HDPTP-depleted cells (Fig. 8A). Notably, knocking down HDPTP also resulted in a redistribution of paxillin from large focal complexes/adhesions at the cell periphery to small punctate staining at the leading edge and in the cytoplasm (Fig. 8A). These results suggest that HDPTP is required for the actin cytoskeleton organization and focal adhesion formation. We next wanted to investigate what effect HDPTP depletion would have on cell motility using the in vitro wound-healing and Transwell migration assays. Control and shHDPTP MDA-MB231 cells were grown to confluence and then wounded by scratching in the presence of mitomycin-C, an inhibitor of cell proliferation. We found that shHDPTP cells migrated across the wound in greater numbers than that of control cells over 24 hours (Fig. 8B). Time-lapse analysis revealed that HDPTP knockdown dramatically increased cell migration speed in MDA-MB231 cells (supplementary material Fig. S8A,B). In the Transwell assay, compared with control cells, there was a marked increase in the number of HDPTP knockdown cells that migrated across the Transwell membrane to the underside of the inserts (Fig. 8C). Together, our findings indicate that HDPTP depletion altered the assembly of focal adhesions and the organization of actin cytoskeleton within the cells, which may account for the increased motility of these cells.

Accumulating evidence has indicated that vesicular trafficking regulates the distribution of plasma membrane content as well as the localization of cytoskeletal proteins during cell adhesion and migration. Drosophila border cells migrate as a cluster of strongly adherent cells during the development of the egg chamber. During this process, JNK signaling and endocytosis-mediated spatial distribution of receptor tyrosine kinases play a critical role (Llense and Martín-Blanco, 2008; Assaker et al., 2010), though mechanisms involved in this process have remained elusive. In this study we identified Mop, the Drosophila homolog of human HDPTP, as a regulator of integrin trafficking. Mop is essential for proper integrin localization and for maintaining border cell integrity during oogenesis. We further demonstrated that Mop and HDPTP interacts with Rab4 GTPase in both Drosophila and mammals. Rab4 has been shown to regulate integrin recycling and cell migration (Roberts et al., 2001; White et al., 2007). Our findings indicate that Mop/HDPTP-mediated endocytic trafficking plays an essential role in integrin-mediated cell adhesion and migration.

Mop has been predicted as a nontransmembrane PTP (Andersen et al., 2005). However, amino acid sequence analysis revealed that Mop displays several differences from conserved PTP motifs within the phosphatase domain. For example, the catalytic essential aspartic acid (D) within motif 8 (WPDXGXP) is replaced by a lysine residue (K). Although the active site cysteine (C) in the catalytic motif 9 (VHCSAGXGR[T/S]G) could be found, the overall signature motif of Mop was much more divergent compared to other PTPs. Moreover, we could not detect Mop tyrosine phosphatase activity using in vitro phosphatase assays. These results suggest that Drosophila Mop may not be enzymatic active. Alternatively, Mop may exhibit weak phosphatase activity which can not be detected using either pNPP or in gel phosphatase assay. A recent study by Lin et al. indicated that human PTPN23/HDPTP exhibits relatively low activity that is comparable with the specific activity of PTP1B D181E mutant (Lin et al., 2011). We also found that expression of Mop-C/S mutant, in which the catalytic cysteine in the active site is replaced by serine, or Mop phosphatase domain deletion mutant rescued the Mop-RNAi-induced border cell dissociation defects as effectively as the wild-type Mop, indicating that the putative tyrosine phosphatase activity is not essential for maintaining border cell cluster integrity.

In addition to having a C-terminus phosphatase domain, Mop has a sequence similar to that of yeast Bro1 at the N-terminus. The Bro1 domain consists of a folded core of about 370 residues and has been found in many proteins, including Bro1, Alix, and Rim20, known to regulate endosome trafficking (Kim et al., 2005). The Bro1 domain has been shown to bind with multivesicular body components (ESCRT-III proteins) such as yeast Snf7 and mammalian CHMP4B for targeting Bro1-domain-containing proteins to endosomes (Ichioka et al., 2007). Interestingly, mutational analysis revealed that CHMP4B binding is not required for HDPTP function, suggesting that the Bro1 domain may have other functions (Doyotte et al., 2008). Here we identified Rab4 as an interactor with Mop and HDPTP through the Bro1 domain. The Rab GTPases cycle between the GTP-bound active and the GDP-bound inactive forms and are necessary for efficient membrane vesicle transport between different subcellular compartments (Stenmark, 2009). Among them, Rab4 and Rab11 have been implicated in controlling membrane trafficking through the endocytic recycling pathways (Grant and Donaldson, 2009). Several lines of evidence suggest that Rab4 is involved in Mop-mediated integrin distribution. First, we found that expression of dominant-negative Rab4S22N or depletion of Rab4 in migrating border cells resulted in integrin redistribution and border cell dissociation. Second, genetic analysis revealed that Rab4S22N and Rab4-RNAi suppressed Mop-induced wing blistering phenotypes. Third, in mammalian cells, Rab4 regulated integrin recycling from early endosomes and was required for cell adhesion and spreading (Roberts et al., 2001). We found that downregulation of HDPTP disrupted integrin distribution, focal adhesion formation, actin organization, and enhanced cancer cell motility. Moreover, since Mop interacts with Rab4 in a nucleotide-independent manner, Mop may act as an adaptor rather than an effector of Rab4 during endosomal trafficking processes.

In mammals, Rab4 has been shown to regulate fast integrin recycling and persistent cell migration (Roberts et al., 2001; White et al., 2007). We also noticed that dominant-negative Rab4 caused integrin redistribution to the perinuclear region. However, we found that expression of HDPTP could not rescue the dominant-negative Rab4-induced integrin redistribution (data not shown). This is consistent with our genetic data that Rab4 acts downstream of Mop (supplementary material Fig. S5). We propose that Mop/HDPTP may act as an adaptor to keep Rab4 in proper endosomal domains. Depletion of HDPTP resulted in Rab4 mislocalization, changes in the integrin dynamics and increases in cell migration. Indeed, we found that misexpression of βPS-integrin resulted in a marked increase in border cell cluster dissociation and βPS-integrin knockdown suppressed Mop^RNAi-induced border cell dissociation phenotype. Mop/HDPTP is likely to function via its association with Rab4 on early endosomes to regulate integrin sorting and recycling.

Integrins activate multiple signaling pathways involved in regulating cell proliferation, survival and migration (Giancotti and Ruoslahti, 1999). One major signaling event stimulated by integrins is mediated by the FAK and Src tyrosine kinases (Mitra and Schlaepfer, 2006; Harburger and Calderwood, 2009). FAK and Src are crucial regulators of cell adhesion and motility which they regulate by controlling the formation and turnover of focal adhesions (Mitra et al., 2005). Several reports have shown interactions between HDPTP and FAK–Src complex. HDPTP has been proposed to act as a molecular bridge between FAK and Src in regulating endothelial and bladder carcinoma cell motility (Mariotti et al., 2009b; Mariotti et al., 2009a). Moreover, Lin et al. identified PTPN23/HDPTP as a negative regulator of cell invasion in mammary epithelial cells (Lin et al., 2011). In those studies, FAK and Src were found to be substrates of HDPTP, suggesting that loss of HDPTP may increase FAK–Src activity to promote cell motility. Intriguingly, in Drosophila, we showed that Mop regulates the border cell association in a PTP domain-independent manner. Recent studies in Drosophila and mammalian cells also found that the phosphatase activity of Mop and HDPTP is not required for its biological function in receptor signaling and tumor suppression (Miura et al., 2008; Gingras et al., 2009; Huang et al., 2010). Our findings on Mop/HDPTP–Rab4 interaction and the regulation of integrin recycling provide new insights into the role of Mop/HDPTP in the regulation of cell adhesion and migration, and they are not mutually exclusive from previous findings.

Flies were raised at 25°C following standard procedures unless otherwise noted. The following Drosophila strains were used: mop^T612 and UAS-Mop-C/S (Miura et al., 2008), UAS-mys and UAS-mys^RNAi (Bhandari et al., 2006). The UAS-Mop transgene was generated by subcloning the Mop from SD03094 into the pUAST vector. The UAS-MopΔPTP mutant (amino acids 1–1496), in which C-terminus PTP domain of Mop was deleted, was obtained by PCR and subcloned into the pUAST vector. UAS-Rab4^RNAi (v106651), UAS-FAK^RNAi (v108608), UAS-CSW^RNAi (v21756), and UAS-CG7180^RNAi (v34369) were obtained from the Vienna Drosophila RNAi Center. The mop mutant clones were generated using the Flp/FRT-mediated mitotic recombination system (Xu and Rubin, 1993). For transgenic RNAi lines target sequences were amplified from PCR and inserted as inverted repeats with pRISE vector as the destination vector (Kondo et al., 2006). To construct an inducible RNA interference allele of NT-PTPs, we used the following specific primers: Mop-RNAi (forward: 5′-caccTTATCGCGAGAGTTCCAGAAA-3′ and reverse: 5′-CTTAACGCTCTGCATCCTTTG-3′), dPez-RNAi (forward: 5′-caccAACACATCAGCTTCCACATCC-3′ and reverse: 5′-ACTCGAGACGCTGATGACTGT-3′), dPtpmeg-RNAi (forward: 5′-caccCGTTCAGGTGTCAAAAGTGGT-3′ and reverse: 5′-TCCATTACGGTCGTTAGCATC-3′), PTP-ER-RNAi (forward: 5′-caccGAATCGGAACTGTCAAT-3′ and reverse: 5′-TCCACTGTTCATTAGGTTGCC-3′), dMeg2-RNAi (forward: 5′-CACCGCATGATCTGGGAACAACATT-3′ and reverse: 5′-TACACGTTGTTTTCTCGTCCC-3′). Other stocks were obtained from the Bloomington Stock Center.

Human GFP tagged Rab4A-WT and Rab4A-S22N were gifts from J. Norman (The Beatson Institute). Drosophila Rab4-WT, Rab4-Q67L and Rab4-S22N were kindly provided by M. Scott (Stanford University). Drosophila Rab5, Rab7 and Rab11 cDNAs were gifts from M. Gonzales-Gaitan (University of Geneva). Human HDPTP cDNA was obtained from Open Biosystems (Thermo). The lentiviral shRNA clones used to knockdown human HDPTP were obtained from the National RNAi Core Facility of Academia Sinica. The targeted sequences for these clones are HDPTP shRNA #A: (Clone ID: TRCN0000080843) 5′-GCTCAGGCAAAGATGATTATA-3′, HDPTP shRNA #E: (Clone ID: TRCN0000003047) 5′-CCGCCAGATCCTTACGCTCAA-3′, and HDPTP shRNA #G: (Clone ID: TRCN0000003049) 5′-GACAACGACTTCATTTACCAT-3′. Luciferase shRNA was used as a control. Lentiviral production and infection were performed as previously described (Tang et al., 2011). Antibodies used for the study were: anti-Mop (Abcam), anti-βPS (CF.6G-11, DSHB) anti-DE-cadherin (DCAD2, DSHB), anti-HDPTP (Proteintech), anti-α5 and anti-β1 (BD Biosciences), anti-paxillin (BD Biosciences), anti-EEA1 (Cell Signaling), anti-LAMP1 (Abcam).

Drosophila S2 cells were cultured at 25°C in Schneider's Drosophila medium (Invitrogen) containing 10% fetal bovine serum (FBS) and 1× penicillin/streptomycin antibiotics (Invitrogen). Mammalian cells were cultured at 37°C in DMEM (Invitrogen) medium supplemented with 10% FBS and antibiotics. For immunoprecipitations, HEK293T cells that were transiently transfected with the indicated plasmids were scraped from dishes in lysis buffer [50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP40, 0.25% sodium deoxycholate, 1 mM EDTA, 10 mM NaF, 2 mM Na₃VO₄, 1 mM PMSF, and protease inhibitor cocktail (Roche)]. Cell lysates were immunoprecipitated with anti-Myc or anti-HA antibody and protein-G–Sepharose beads (GE Healthcare) at 4°C for 2 h. These beads were washed three times with the lysis buffer. After resolution by SDS-PAGE, the immunoprecipitates were subjected to western blot analysis.

For border cell migration experiment, ovaries were dissected in Schneider's medium with 10% FBS, fixed with 4% paraformaldehyde in PBS for 15 min at room temperature. After a brief washing with PBS, egg chambers were permeabilized with 0.3% Triton X-100/PBS for 10 min. Samples were then blocked with 5% normal goat serum in PBST (PBS + 0.1% Triton X-100) for 1 h, and incubated with primary antibodies overnight at 4°C. On the following day, samples were washed three times with PBST and incubated with fluorescent-labeled secondary antibodies for 2 h at room temperature. Egg chambers were mounted in anti-fade reagent containing 0.1 M n-propyl gallate (pH 7.4) and 90% glycerol in PBS. For immunostaining, cells grown on coverslips were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-1000/PBS. They were then incubated with primary antibodies for at least 1 h, followed by incubation with secondary antibodies. F-actin was stained using TRITC-conjugated phalloidin (Sigma) and nucleus was stained using DAPI (1 µg/ml). Samples were visualized under an epifluorescence microscope (Olympus BX51) or a confocal laser scanning microscope (Zeiss LSM510).

The para-nitrophenyl phosphate (pNPP) phosphatase assays were carried out as described previously (Montalibet et al., 2005). Briefly, S2 cells transiently expressing HA-tagged PTP domain of PTP61F (amino acids 1–310), dMEG2 (amino acids 505–815), dPTPMEG (amino acids 497–856), dPEZ (amino acids 957–1252) and Mop (amino acids 1497–1803) were lysed and immunoprecipitated with anti-HA antibody. The immunocomplexes were incubated with 20 mM pNPP (Sigma) in pNPP assay buffer (50 mM Hepes, pH 7.5, 1% NP-40, 10 mM DTT) at 37°C in the dark for 20 min. The absorbance of the product, pNPP, was measured at 405 nm with a spectrophotometer.

The DNA fragment encoding the Bro1 domain of Mop (amino acids 1–434) was cloned to pGBKT7 (Clontech) to generate a fusion to the Gal4 DNA-binding domain (pGBKT7-Bro1). The yeast strain AH109 expressing the pGBKT7-Bro1 was transformed with a Drosophila embryonic cDNA library fused to the Gal4 activation domain (Clontech). The double transformants were plated onto selective medium lacking adenine, histidine, leucine, tryptophan and further assayed for β-galactosidase activity by a colony lift filter assay. The positive clones were isolated and sequenced.

To study the internalization of the integrins, cells were cultured in 10 cm plate until they formed monolayers and the cells were washed twice with PBS and then detached with the Trypsin-EDTA buffer. Cell surface integrins were labeled with anti-α5 antibody (BD Biosciences) for 20 minutes at 4°C and washed twice in cold PBS. We then allowed endocytosis to occur at 37°C for various intervals (0, 3, 6 and 12 minutes). Cells were subsequently fixed with 4% paraformaldehyde for 20 minutes at 4°C and washed with cold PBS. Washed cells were labeled with APC-conjugated secondary antibody (BD Biosciences) for 20 minutes at 4°C to the primary antibody coated on the cell surface. Antibodies conjugated cells were washed twice with PBS before flow cytometry analysis (BD FACSCanto II).

Integrin recycling assay was performed as described previously (Roberts et al., 2001). Surface-labeled cells were incubated with serum-free DMEM at 22°C for 15 min or 37°C for 30 min to allow integrin internalization. Biotinylated surface proteins were removed by incubation of 20 mM 2-mercaptoethanesulfonate (Mesna, Sigma) in 50 mM Tris (pH 8.6) and 100 mM NaCl for 30 min at 4°C. Recycling of internalized biotin-labeled integrin was stimulated by 10% FBS containing DMEM at 37°C. At indicated time, medium was aspirated and cells were washed twice with cold PBS. Surface protein biotinylation was removed by a second reduction with MesNa incubation, followed by IAA incubation, cell lysis as described above. The percentage of recycled integrin was calculated from control cells maintained on ice. Biotinylated integrins were detected by capture ELISA as described (Roberts et al., 2001). In brief, Nunc Maxisorp 96-well plates were coated with 2.5 µg/ml anti-integrin antibodies in 0.05 M Na₂CO₃ (pH 9.6) at 4°C overnight and blocked for 1 h at room temperature with 5% BSA in PBS containing 0.1% Tween-20 (PBST). Integrins were captured by overnight incubation of cell lysates with equivalent concentration at 4°C. After extensive washing with PBST, color was developed with TMB substrate (Sigma) and stopped with 1 N HCl. The absorbance was measured at 450 nm using a microplate reader (Tecan).

MDA-MB-231 breast cancer cells expressing HDPTP siRNA (shHDPTP) or luciferase-knockdown control (shLuc) were plated onto six-well culture plates in DMEM containing 10% FBS. After 24 h, the cell monolayers were wounded manually by scratching with a pipette tip, and rinsed with PBS. Fresh DMEM with 1% FBS and 10 µg/ml mitomycin C was added to follow healing for 24 h. Cells were photographed at 0 h, 6 h and 24 h after wounding using a phase contrast microscopy and an Olympus IX70 camera. For time-lapse microscopy and cell tracking analysis, cells were observed with a 20× phase-contrast objective and images were collected every 10 min using a Cool SNAP CCD camera. Cell tracks were analyzed with ImageJ software. For Transwell migration assays, 1×10⁵ cells were seeded into and grown in low serum medium (1% FBS) on the top of Transwell chambers (8 µm pores, Millipore). The lower chamber was filled with medium containing 10% FBS. Mitomycin C (Sigma) was added to inhibit cell proliferation. After 24 h, the migrated cells on the lower surface of membrane were fixed and stained with 0.5% Crystal Violet (Sigma). The absorbance was measured at 570 nm using a microplate reader (Tecan).

We thank M. Gonzales-Gaitan, M. Grotewiel, Y. Kageyama, J. Norman, L-M Pai, H.-W. Pi, M. Scott, J. Treisman, the Bloomington Stock Center, Vienna Drosophila RNAi Center, Fly Core Taiwan and the Developmental Studies Hybridoma Bank (DSHB) for reagents. We thank S.Y. Tsai for the yeast two-hybrid screen and C.-C. Hung for the confocal microscopy assistance.