Identification and functional characterization of paxillin as a target of protein tyrosine phosphatase receptor T

Abstract

Protein tyrosine phosphatase receptor-type T (PTPRT) is the most frequently mutated tyrosine phosphatase in human cancers. However, the cell signaling pathways regulated by PTPRT largely remain to be elucidated. Here, we show that paxillin is a direct substrate of PTPRT and that PTPRT specifically regulates paxillin phosphorylation at tyrosine residue 88 (Y88) in colorectal cancer (CRC) cells. We engineered CRC cells homozygous for a paxillin Y88F knock-in mutant and found that these cells exhibit significantly reduced cell migration and impaired anchorage-independent growth, fail to form xenograft tumors in nude mice, and have decreased phosphorylation of p130CAS, SHP2, and AKT. PTPRT knockout mice that we generated exhibit increased levels of colonic paxillin phosphorylation at residue Y88 and are highly susceptible to carcinogen azoxymethane-induced colon tumor, providing critical in vivo evidence that PTPRT normally functions as a tumor suppressor. Moreover, similarly increased paxillin pY88 is also found as a common feature of human colon cancers. These studies reveal an important signaling pathway that plays a critical role in colorectal tumorigenesis.

Reversible tyrosine phosphorylation, which is coordinately controlled by protein tyrosine kinases (PTKs) and phosphatases (PTPs), governs numerous signaling pathways that regulate cell proliferation, apoptosis, adhesion, and migration. Over the last two decades, many PTKs have been found to be mutated in a variety of different tumor types (reviewed in ref. 1). In contrast to PTKs, the role of PTPs in tumorigenesis is underexplored. To systematically evaluate possible roles of PTPs in tumorigenesis, we used a high throughput molecular and bioinformatics approach to detect genetic alterations of the tyrosine phosphatase gene family in colorectal cancers (CRCs) (2). Among the six mutated PTPs that we identified, protein tyrosine phosphatase receptor-type T (PTPRT), also known as PTPρ, was the most frequently mutated (2). In addition, we and others found that PTPRT is also mutated in lung, stomach, and skin cancers (2, 3). The spectrum of mutations, which includes nonsense mutations and frameshifts, suggested that these mutations were inactivating (2). Biochemical analyses demonstrated that missense mutations in the catalytic domains of PTPRT diminished its phosphatase activity, whereas overexpression of PTPRT inhibited CRC cell growth (2). Taken together, these studies suggest that PTPRT normally acts as a tumor suppressor gene. In light of these data, it is important to identify the functionally significant substrates of PTPRT as well as to elucidate the signal transduction pathways regulated by this phosphatase. Here, we report that paxillin, an adaptor protein involved in cell adhesion, migration, proliferation, and apoptosis (4, 5), is a direct substrate of PTPRT and that phospho-paxillin has oncogenic properties. We further demonstrate that in an in vivo model, PTPRT is both a potent tumor suppressor gene and a key regulator of colonic phospho-paxillin levels.

Given its roles in cell adhesion and migration, paxillin is thought to play an important role in tumor migration, invasion, and metastasis (6). Moreover, a recent study found that paxillin is mutated and amplified in roughly 20% of lung cancers and that mutant paxillin promotes cancer cell growth and enhances cell adhesion, suggesting that paxillin is a proto-oncogene (7). Paxillin directly interacts with a variety of cellular proteins, including FAK, Gab1, p120RasGAP, and others (4, 8 –10). Some of these interactions are tyrosine phosphorylation (pY)-dependent, wherein the pY residues recruit SH2 domain-containing proteins (4). A recent mass spectrometric analysis indicates that tyrosine phosphorylation occurs at multiple sites on paxillin including tyrosine residues 31, 40, 76, 88, 118, 182, 377, and 436 (11). Although paxillin Y31 and Y118 phosphorylation has been implicated in cell adhesion and migration (12), the physiological relevance of tyrosine phosphorylation at other residues remains to be determined.

In this study, we demonstrate that PTPRT dephosphorylates paxillin specifically at Y88 residue both in vitro and in CRC cells, and, by using paxillin Y88F mutant knock-in CRC cells, that regulation of paxillin Y88 phosphorylation plays a critical role in colorectal tumorigenesis.

Results

Paxillin Is a Direct Substrate of PTPRT.

As reported previously, we used a proteomic approach to identify potential substrates of PTPRT (13). We generated two HEK 293T cell lines: one expressing the intracellular part containing the two phosphatase domains of PTPRT and the other expressing the extracellular portion of the protein (13). We used the proteomic approach developed by Rush et al. (14) to globally profile tyrosine phosphopeptides in the two cell lines as well as parental HEK293T cells. Phosphopeptides that are present in parental HEK 293T cells and in cells expressing the extracellular portion of PTPRT, but not in cells expressing the intracellular portion, are potential PTPRT substrates. Through this approach, we identified five candidate substrates of PTPRT, including paxillin and STAT3. We validated that STAT3 is indeed a direct substrate of PTPRT (13). Given that PTPRT plays important roles in cell proliferation and adhesion (2, 15), two processes that paxillin is also involved, we set out to determine whether paxillin was a direct substrate of PTPRT. To test this hypothesis, we employed a substrate-trapping assay modified from that described for investigation of unrelated phosphatase (PTP1B) (16, 17). Three GST fusion proteins were constructed for this purpose as reported (13). All contained both phosphatase domains of PTPRT, but two of the three (TRAP-1 and TRAP-2) contained mutations predicted to result in substrate trapping (the D1074A single mutant and the C1106S D1074A double mutant, respectively). The recombinant GST fusion proteins were expressed in Escherichia coli and equal amounts of each were attached to beads (Fig. S1A). The beads were then incubated with lysates from CRC cells. Paxillin proteins bound abundantly to both of the two substrate trapping mutants, but only minimally to the wild-type (WT) PTPRT protein and not at all to the control GST protein (Fig. 1A).

Fig. 1.

Paxillin is a direct substrate of PTPRT. (A) Paxillin is pulled down by substrate-trapping mutants of PTPRT. Colon cancer cell lysates were incubated with beads bound to the indicated GST-fusion proteins. Western blots were performed with an anti-paxillin antibody. (B) PTPRT dephosphorylates paxillin at residue Y88. Phospho-paxillin proteins were immunoprecipitated from lysate of HEK 293 cells. The immunocomplexes were incubated with the indicated recombinant proteins with or without Na₃VO₄. (C) DLD1 cells were infected with adenoviruses expressing PTPRT or GFP and starved and then stimulated with PDGF-AA for the indicated times. The intensity of pY88 signals were quantified with NIH image and normalized against total paxillin levels. Fold change of pY88 paxillin for each time point over unstimulated PTPRT-infected cells was calculated.

The target site of PTPRT on paxillin, as predicted by our proteomic data (13), is the previously uninvestigated phospho-Y88 residue. To characterize phosphorylation at this tyrosine, we successfully generated an antibody specifically recognizing phospho-tyrosine 88 (pY88) paxillin as described in detail in Materials and Methods. This antibody is highly specific as it recognized a single pY88 paxillin band in lysates of parental CRC cells by Western blot analysis, but produced no signal in cell lysates of paxillin Y88F homozygous mutant cells (Fig. S2C). To test whether PTPRT directly dephosphorylates paxillin in vitro, we incubated phospho-paxillin immunoprecipitated from HEK 293T cells as a substrate with equal amounts of WT PTPRT, TRAP-2 mutant PTPRT, or GST protein, with or without the phosphatase inhibitor Na₃VO₄. Paxillin was dephosphorylated at the Y88 residue by WT PTPRT but not by the phosphatase-dead mutant (TRAP-2) or GST alone (Fig. 1B). This activity was also inhibited by Na₃VO₄, indicating that it was phosphatase-dependent (Fig. 1B). Interestingly, PTPRT failed to dephosphorylate paxillin at either the Y31 residue or the Y118 residue (Fig. S1B), suggesting that PTPRT is a paxillin pY88-specific phosphatase.

PTPRT Regulates Paxillin Y88 Phosphorylation in CRC Cells.

Although Y88 paxillin phosphorylation has been reported in several studies (11, 18 –21), the biological significance of this phosphorylation and the extracellular cues that activate it remain to be determined. To address these issues, we used the CRC cell line DLD1 to test whether growth factors and cytokines could stimulate paxillin Y88 phosphorylation. As shown in Fig. S3A, PDGF-AA robustly activated this phosphorylation, whereas PDGF-AB, PDGF-BB, PDGF-CC, EGF, interleukin-6 (IL-6), and FGF-stimulated paxillin Y88 phosphorylation to varying, lesser extents. No stimulation was observed with VEGF. Furthermore, PDGF-AA activated paxillin Y88 phosphorylation in a variety of CRC cell lines (Fig. S3B). Thus, we chose PDGF-AA for in-depth study.

Given that PDGF plays direct roles in development of gastrointestinal villi (22) and in promoting the growth of CRC cells (23), it is important to determine whether PTPRT regulates paxillin Y88 phosphorylation in CRC cells activated by PDGF-AA. As shown in Fig. 1C, infection of DLD1 CRC cells with an adenovirus expressing PTPRT resulted in reduced pY88 paxillin levels in comparison with cells infected with a control GFP adenovirus at each of 45, 60, and 120 min after PDGF-AA stimulation. Similar results were also observed with HCT116 CRC cells (Fig. S3C). Interestingly, phosphorylation of paxillin Y31 and Y118 was not altered by PDGF-AA or PTPRT (Fig. 1C). Note that only the pY88 paxillin, not its total levels, was affected by PTPRT, thereby excluding protein degradation as a cause of these effects (Fig. 1C).

Generation of Paxillin Y88F Knock-In Colon Cancer Cells.

To rigorously test whether regulation of paxillin Y88 phosphorylation is critical to colorectal tumorigenesis, we set out to engineer paxillin Y88F knock-in (KI) CRC lines. The adeno-associated virus (AAV) targeting system was used to generate the KI cell lines because of its high homologous recombination frequency in somatic cells (24, 25). We first chose to knock in the paxillin Y88F mutant allele into the human colon cancer cell line HCT116, because we have shown that paxillin Y88 phosphorylation could be stimulated by PDGF-AA in HCT116 cells and because HCT116 has been widely used for successful gene targeting by homologous recombination (26). The exon 3 sequences of the paxillin gene, which encodes the Y88 residue, were replaced by mutant alleles by homologous recombination (Fig. S2A). This exon was sequenced in the targeted clones and the parental cells. As expected, both WT paxillin alleles were replaced by the paxillin Y88F mutants in the homozygous KI clones (Fig. S2B). Furthermore, Western blot analyses showed that paxillin proteins were expressed in the homozygous KI (Y88F/Y88F) cells, but they remained unphosphorylated at residue 88 after PDGF-AA stimulation. Moreover, phosphorylation at Y88 was heavily induced in the parental cells, but it was attenuated in the heterozygous KI (WT/Y88F) cells in response to PDGF-AA stimulation (Fig. S2C). These data indicated that we had successfully engineered paxillin Y88F mutant cells. To ensure that what we observe with HCT116 cells can be generalized to other colon cancer cells, we used the same method to generate paxillin Y88F KI mutant DLD1 cells (Fig. S2C).

Paxillin Y88F Mutant Cells Exhibit Impaired Anchorage-Independent Growth.

When grown under normal tissue culture conditions (McCoy's 5A supplemented with 10% FBS), there was no obvious proliferation difference between the heterozygous KI clones and the parental cells; however, the homozygous Y88F mutant cells grew a little slower (Fig. S4A). Consistently, we observed a subtle increase of apoptotic cell population in the homozygous Y88F mutant clones (Fig. S5). To test whether the paxillin Y88F mutant affects transformation, we performed colony formation and soft agar assays with the paxillin mutant KI cells. Compared to the parental cells, homozygous paxillin Y88F KI HCT116 cells displayed reduced colony-formation ability (Fig. S4B). In the case of DLD1 cells, however, only subtle differences were observed between parental and paxillin Y88F KI cells (Fig. S4B). In contrast, all homozygous Y88F KI clones formed ≈3-fold (P < 0.001) fewer foci in the soft agar assay than their WT counterparts in both HCT116 and DLD1 CRC cells (Fig. 2A). Interestingly, all of the heterozygous KI clones also displayed a significant (P < 0.05) reduction in the number of soft-agar foci with respect to WT cells (Fig. 2A), suggesting that paxillin Y88F protein may act as either a dominant-negative or haploid-insufficient mutant. Taken together, these results demonstrated that regulation of paxillin Y88 phosphorylation plays an important role in transformation.

Fig. 2.

Paxillin Y88F mutant CRC cells are less tumorigenic. (A) Anchorage-independent growth. CRC cells of the indicated clones were mixed in 0.4% soft agar and plated in six-well plates in triplicates. Cells were grown for 30 days, and colony foci were counted (*, P < 0.001; **, P < 0.05). (B) Athymic nude mice were injected s.c. and bilaterally with cells of the indicated genotypes and killed 35 days after cell injection. Mice that formed xenografts of the indicated genotypes were counted. (C) Tumor sizes of the indicated clones were measured weekly for 5 weeks, and the average volume at each time point was plotted.

Paxillin Y88F Mutant Cells Fail to Form Xenograft Tumors in Nude Mice.

To test tumorigenicity of the paxillin Y88F KI cells, paxillin Y88F homozygous clones, heterozygous clones, or parental HCT116 and DLD1 cells were injected s.c. into nude mice. Tumor formation and size were assessed by weekly caliper measurements. After 35 days of growth, WT cells formed tumors in all injected mice, yielding an average size of 2,300 mm³ in HCT116 cells and 1,800 mm³ in DLD1 cells. In contrast, homozygous HCT 116 KI cells failed to form tumors in any of the five mice (Fig. 2B and Fig. S4C), and only one of the five mice injected with DLD1 paxillin Y88F homozygous mutant cells formed a tumor, but it was tiny, only 18 mm³ (Fig. 2 B and C). Notably, the average tumor sizes of paxillin Y88F heterozygous KI clones were also significantly (P < 0.05) smaller than those formed by the parental cells (Fig. 2C).

Paxillin Y88F Mutant Cells Exhibit Reduced Cell Migration.

It is well documented that paxillin is involved in cell adhesion and migration (4). The paxillin Y88F KI cells showed no defect in cell adhesion. To test the ability of paxillin Y88F KI cells to undergo migration, a Boyden chamber assay was employed. As shown in Fig. S6A, both HCT116 and DLD1 Y88F homozygous mutant cells exhibited a ≈4-fold decreased ability to migrate through a porous membrane coated with fibronectin in comparison with parental cells (P < 0.001). Consistent with this result, the paxillin Y88F homozygous mutant cells also migrated slower in wound healing assays (Fig. S6 B and C). Compared to parental cells, heterozygous paxillin KI cells also displayed a significant defect in cell migration in these assays (P < 0.05).

Paxillin Y88F Mutant Affects AKT, p130CAS, and SHP2 Signaling.

We demonstrated that phosphorylation of the paxillin Y88 residue plays a critical role in colorectal tumorigenesis. To gain insights into the effects of this phosphorylation on downstream signaling, we examined how the paxillin Y88F KI affects phosphoyrlation of signaling molecules in CRC cells after PDGF-AA stimulation. It is well documented that PDGF receptors (PDGFRs), once they are engaged by their ligands, activate multiple well-characterized signaling pathways, including Ras-MAPK, PI3K-AKT, and PLC-γ (22). We tested the phosphorylation status of 27 sites on 17 proteins that could be potentially modulated by PDGF signaling (Table S1). In both parental HCT116 and DLD1 cells, PDGF-AA activated AKT phopshorylation at both threonine residue 308 and serine residue 473, two posttranslational modifications that are known to be critical for AKT activation (27). However, no activation of these AKT phosphorylations could be induced in the paxillin Y88F homozygous mutant CRC cells (Fig. 3). PDGFRs also crosstalk with integrins and modulate cell adhesion signaling (22). Among the cell adhesion molecules tested, PDGF-AA stimulated the phosphorylation of p130CAS at tyrosine 165 (Y165) (Fig. 3). This stimulation of pY165 p130CAS was also attenuated in paxillin Y88F mutant CRC cells. Lastly, phospho-Y542 of SHP2 could also be moderately stimulated by PDGF-AA and the levels of this phosphorylation were consistently lower in the paxillin Y88F mutant CRC cells than in the parental cells. The paxillin Y88F mutant did not affect the phosphorylation of MAP kinase (Thr-202/204), PLC-γ (Tyr783 and Ser1248), several STAT proteins (STAT1, STAT3, and STAT5), FAK, and CRKL in response to PDGF-AA in CRC cells (Fig. 3 and Table S1). These results suggest that phosphorylation of paxillin Y88 plays a role in transducing the PDGF-AA signal to pathways involving AKT, p130CAS, and SHP2.

Fig. 3.

Paxillin Y88F mutant affect cell signaling through altering phosphorylation of AKT, p130CAS, and SHP2. Parental (WT) and paxillin Y88F homozygous KI CRC cells were serum-starved for 16 h and stimulated with PDGF-AA for the indicated times. Western blot analyses were performed with the indicated antibodies.

Paxillin Is an in Vivo Substrate of PTPRT.

To determine whether paxillin is an in vivo substrate of PTPRT, we generated PTPRT knockout mice. Exon 22 of PTPRT, which encodes the phosphatase catalytic core motif of the first catalytic domain, was targeted (Fig. S7A). This strategy produced a truncated PTPRT protein devoid of the two phophatase domains in 129/Ola mouse embryonic stem (ES) cells. Three clones were aggregated with eight-cell stage embryos. Chimeras were mated with C57BL/6J females, and germ-line transmission of the disrupted PTPRT allele was achieved. Interbreeding of F₁ PTPRT^+/− mice yielded F₂ offspring at the normal Mendelian ratio; a representative Southern blot of genomic DNA from PTPRT^+/+, ^+/−, and ^−/− mice is shown in Fig. S7B. Furthermore, PTPRT^−/− mice showed no obvious developmental abnormalities.

To test whether PTPRT regulates paxillin Y88 phosphorylation under in vivo physiological conditions, Western blots were performed with colon tissue lysates from WT and PTPRT^−/− mice. As shown in Fig. 4A, pY88 paxillin, but not pY31 or pY118, was up-regulated in the knockout mouse colons in comparison with colons from the WT littermates, suggesting that PTPRT is a key regulator of paxillin Y88 phosphorylation in vivo.

Fig. 4.

Increased tyrosine phosphorylation of paxilin correlates with azoxymethane (AOM) induces colon tumors in PTPRT knockout mice. (A) PTPRT regulates pY88 paxillin in vivo. Colon lysates were made from PTPRT^+/+ (WT) and PTPRT^−/− mouse littermates. Western blots were performed with the indicated antibodies. (B) Numbers of colon tumors developed in each of the PTPRT^+/+ and ^−/− C57BL/6J mice (n = 20 in each group). Each circle or square represents one mouse. (C) Gross morphology of two tumors (arrowheads) in a PTPRT^−/− mouse (Right) compared with a colon of a PTPRT^+/+ (WT) mouse. (D and E) Representative images of hematoxylin and eosin staining (D) and Ki67 IHC (E) of AOM-induced tumors.

PTPRT Suppresses Azoxymethane-Induced Colon Tumors.

Our biochemical and cellular data indicate that PTPRT normally acts as a tumor suppressor gene. To test this hypothesis in vivo, we bred the PTPRT knockout allele for more than 10 generations onto the C57BL/6J mouse strain, which has been shown to be highly resistant to colon tumor induced by azoxymethane (AOM) (28, 29). Intercrossing heterozygous PTPRT knockout mice generated littermates of PTPRT genotypes that were ^+/+ and ^−/−. Twenty littermates of each genotype were treated with six doses of AOM by i.p. injection. As reported previously, the ^+/+ mice were highly resistant to AOM-induced colon cancer and none developed any colon tumor (Fig. 4B). In contrast, 17 of the 20 PTPRT homozygous knockout mice developed visible colon tumors, averaging 1.5 tumors per mouse. Representative tumors that developed in a PTPRT^−/− mouse colon are shown in Fig. 4C. Histological analyses indicated that those tumors were adenomas (Fig. 4D) and that the dysplastic cells were highly proliferative as indicated by Ki-67 immunohistochemistry (IHC) staining (Fig. 4E).

Paxillin Y88 Phosphorylation Is Up-Regulated in Human Colon Cancer Tissues.

Our data demonstrated that PTPRT is a key regulator of murine colonic pY88 paxillin and that dephosphorylation of paxillin pY88 is strongly tumor suppressive. To explore the role of this pY88 paxillin pathway in human colon cancers, we performed IHC staining of formalin-fixed paraffin-embedded (FFPE) human colon carcinoma sections and their matched normal colon tissues with our anti-pY88 paxillin antibody. These tumors were all low-grade moderately differentiated adenocarcinomas. As shown in Fig. 5A Upper, pY88 paxilin was highly expressed in tumor epithelial cells but not in the matched normal colon epithelial cells. Interestingly, total paxillin levels were comparable between the tumor and normal cells (Fig. 5A Lower), indicating that paxillin Y88 phosphorylation was specifically up-regulated in human colon cancers. Of the 10 colon tumor and normal pairs examined, pY88 paxillin was up-regulated in 8 of the tumors but not the matched normal tissues (Fig. 5B). Staining of colon cancer cells could be completely blocked by competition with pY88 paxillin peptides, but not by competition with the corresponding nonphosphorlated peptides (Fig. S7C). In stroma, we noted some nonspecific staining that could not be competed by the pY88 paxillin peptides. Nonspecific staining that could not be competed was also noted in mouse epithelial cells, thus restricting IHC analysis to human samples.

Fig. 5.

Paxillin Y88 phosphorylation is up-regulated in human colon cancers. (A) Representative images of IHC staining of human colon carcinomas and matched normal colon tissues with the indicated antibodies. (Scale bar: 100 μm.) (B) Quantitative IHC staining of 10 pairs of tumors and matched normal colon tissues with anti-pY88 paxillin antibody (*, P < 0.001).

Discussion

These results provide critical in vivo evidence that PTPRT normally functions as a tumor suppressor gene. Moreover, they establish paxillin as a functionally important direct substrate of PTPRT. Specifically, we showed that regulation of paxillin phosphorylation at the Y88 residue, the target site of PTPRT, plays an important role in CRC tumorigenesis and CRC cell migration, and that phosphorylation of the paxillinY88 residue affects signaling through AKT, p130CAS, and SHP2.

Although paxillin Y88 phosphorylation was observed more than a decade ago (18), it was thought that this was a minor phosphorylation site of SRC kinase (20). Phosphorylation of Y31 and Y118 residues on paxillin was recognized as the major tyrosine phosphorylation events (18 –20). Here, we demonstrated that pY88 paxillin was strongly induced by multiple growth factors and cytokines including PDGFs, EGF, FGF, and IL-6. Consistent with our observation that EGF activated paxillin Y88 phosphorylation in CRC cells, a recent proteomic study showed that EGF stimulates pY88 paxillin in a head and neck squamous carcinoma cell line (21). Among the tested growth factors, PDGF-AA activates robust paxillin Y88 phosphorylation in multiple CRC cell lines. In fact, stimulation of paxillin tyrosine phosphorylation by PDGF was first observed more than 15 years ago (30), but the identity of the pY sites remained elusive. Our data clearly demonstrated that PDGF-AA predominantly activated paxillin phosphorylation at Y88 residue but not the Y31 or Y118 residues (Fig. 2C). Although a body of literature demonstrated the involvement of paxillin Y31 and Y118 phosphorylation in cell adhesion (4), we did not observe any major defects of paxillin Y88F mutant CRC cells in cell adhesion. Our studies showed that the unphosphorylatable paxillin Y88F mutant CRC cells display significantly reduced anchorage-independent growth, although the proliferation rate of these mutant cells is only slightly decreased compared to their parental cells when grown on plastics. Surprisingly, both paxillin Y88F mutant HCT116 and DLD1 CRC cells largely failed to form xenograft tumors in nude mice, suggesting that paxillin Y88 phosphorylation plays a prominent role in tumor growth in the rich in vivo environment. All these results support the hypothesis that PTPRT-regulated paxillin Y88 phosphorylation is critical for colorectal tumorigenesis.

Increasing evidence suggests that receptor protein tyrosine phosphatases (RPTPs) play critical roles in tumorigenesis. Since our first identification of RPTPs, including PTPRT, PTPRF, and PTPRG as potential tumor suppressor genes, several laboratories recently discovered that PTPRD was mutationally inactivated in brain, colon, head and neck, lung, and skin cancers (31, 32). Interestingly, PTPRD also dephosphorylates STAT3 (33), another proven PTPRT substrate (13). This study suggests that PTPRT and PTPRD may have overlapping tumor suppression functions. The target site of PTPRT on STAT3 is the Y705 residue, and phosphorylation of Y705 STAT3 is a key event to its activation. Interestingly, neither STAT3 nor paxillin forms stable complexes with PTPRT. Besco and colleagues (34) found that PTPRT associated with adhesion molecules such as E-cadherin and different catenin isofroms. It is yet to be determined whether paxillin or STAT3 cross-talks with those adhesion proteins.

Finally, our findings are significant for human disease, as demonstrated by the up-regulation of pY88 paxillin in ≈80% of human colon carcinomas. Unfortunately, currently available PTPRT antibodies do not work for immnuohistochemistry, which prohibits us from determining whether there is inverse correlation between PTPRT levels and paxillin pY88 levels in human colon tumors. Even though PTPRT is mutated in ≈11% of colon cancers, epigenetic silence could be another mechanism by which tumors inactivate PTPRT, which may provide potential explanation to the up-regulation of paxillin pY88 in a majority of colon cancer specimens. These results have potential therapeutic and diagnostic implications. Given that paxillin Y88F mutant CRC cells largely failed to form xenograft tumors (Fig. 2), inhibition of the corresponding kinase(s) should mimic this mutation and, thus, lead to arrest of in vivo tumor growth. It will be of great interest to identify such kinase(s).

Materials and Methods

Generation of Anti-Paxillin pY88 Antibody.

Rabbits were immunized with the qpqssspv(phos)ygssakts peptide conjugated to BSA. The antiserum was first absorbed to a BSA-Sepharose 4B column and then passed twice through BSA-conjugated unphosphrylated peptide (qpqssspvygssakts) columns.

Phosphatase Substrate Trapping and in Vitro Phosphatase Assay.

Those assays were performed as described (13). Detail procedures are described in SI Materials and Methods.

Cell Culture and Reagents.

HCT116, DLD1, RKO, and HEK 293T cells (ATCC, Manassas, VA) were grown as described (13). Cells were serum-starved for 15 h before stimulation with 10 ng/mL of PDGF-AA, PDGF-AB, PDGF-BB, PDGF-CC, FGF, VEGF, IL-6, or 200 ng/mL EGF, respectively. Cell proliferation was assayed by using the Cell Counting Kit-8, which quantifies the number of viable cells. (Dojindo Molecular Technologies, Rockville, MD).

Targeted Knock-In of the Paxillin Y88F Mutant Allele.

The approach for generating targeted cells with AAV was performed as described (25). The targeting vector was constructed by PCR using HCT116 genomic DNA as the templates for the homologous arms. Constructs and primer sequences are available upon request. Stable G418-resistant clones were selected in the presence of either 0.4 mg/mL or 1 mg/mL for HCT116 and DLD1 cells, respectively. After the first allele was targeted, the neomycin resistance gene was excised by Cre-recombinase and the targeted clones were retargeted to obtain homozygous knock-in cells.

Xenografts.

Five million cells were injected s.c. and bilaterally into 4- to 6-week-old female nude mice (5 nude mice in each group). Tumor formation and size were assessed by weekly caliper measurements. After 21 days, the mice were killed and tumors were harvested.

Generation of PTPRT Knockout Mice.

A genomic clone encompassing exons containing the catalytic domains of PTPRT was isolated from the 129/SvJ genomic library (Stratagene). The pPGK-NeobpA vector and a DTA cassette from pMC1DTpA were used to construct the targeting vector in which the sequences for PGK-NeobpA replaced the exon containing the essential cysteine and the surrounding core motif. The homologous regions at 5′ and 3′ ends were 1.3 kb and 7.9 kb, respectively (Fig. S7). The targeting vector was linearized by NotI digestion and electroporated into E14-1 ES cells. Three targeted clones were injected aggregated with eight-cell stage embryos. Chimeras were mated with C57BL/6J females, and germ-line transmission of the disrupted PTPRT allele was verified.

AOM Treatment.

Six-week-old mice were injected i.p. once weekly for 6 weeks with 10 mg/kg AOM (Sigma Chemical, St. Louis). Mice were killed 24 weeks after the last AOM injection. Animal protocols were designed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by either the Institutional Animal Care and Use Committee at Case Western Reserve University or the Committee of Animal Experiments, Institute of Development, Aging and Cancer, Tohoku University.

Immunohistochemistry.

Paraffin-embedded mouse and human tissues were deparaffinized in xylene and antigen retrieved by boiling the sample for 20 min. Samples were incubated with primary antibodies at 4 °C overnight. The sections were stained with secondary antibody for 30 min at room temperature and then stained with an EnVision-HRP kit (Dako). Stained sections were classified according to the intensity of staining and the percentage of cells showing paxillin Y88 phosphorylation staining. The intensity of pY88 paxillin staining was assessed in a semiquantitative manner with assignment of staining ranging from 0 to 4+.