Glioblastoma multiforme (GBM), the most common malignant primary brain tumor, represents a significant disease burden. GBM tumor cells disperse extensively throughout the brain parenchyma, and the need for tumor-specific drug targets and pharmacologic agents to inhibit cell migration and dispersal is great. The receptor protein tyrosine phosphatase μ (PTPμ) is a homophilic cell adhesion molecule. The full-length form of PTPμ is down-regulated in human glioblastoma. In this article, overexpression of full-length PTPμ is shown to suppress migration and survival of glioblastoma cells. Additionally, proteolytic cleavage is shown to be the mechanism of PTPμ down-regulation in glioblastoma cells. Proteolysis of PTPμ generates a series of proteolytic fragments, including a soluble catalytic intracellular domain fragment that translocates to the nucleus. Only proteolyzed PTPμ fragments are detected in human glioblastomas. Short hairpin RNA–mediated down-regulation of PTPμ fragments decreases glioblastoma cell migration and survival. A peptide inhibitor of PTPμ function blocks fragment-induced glioblastoma cell migration, which may prove to be of therapeutic value in GBM treatment. These data suggest that loss of cell surface PTPμ by proteolysis generates catalytically active PTPμ fragments that contribute to migration and survival of glioblastoma cells. [Cancer Res 2009;69(17):6960–8]
- protein tyrosine phosphatase
- cell migration
Gliomas are malignancies of glial supporting cells of the central nervous system, including astrocytes and oligodendrocytes ( 1, 2). These neoplasms are categorized by their putative cell of origin based on morphologic similarities to various types of normal glia ( 2, 3). They are graded histologically between 1 and 4 according to the WHO classification system of tumor cellularity, proliferation, angiogenesis, and invasiveness ( 4). Glioblastoma multiforme (GBM), a WHO grade 4 glioma, has a poor prognosis with a mean survival time of <1 year ( 5). The lethality of GBM can be attributed to the dispersive phenotype where cells migrate and develop foci throughout the brain ( 3, 6, 7). We recently showed that the receptor protein tyrosine phosphatase μ (PTPμ) negatively regulates GBM cell migration, and full-length PTPμ protein is lost in human GBM tumors in comparison with low-grade astrocytomas ( 8).
PTPμ is the prototype of the type IIb subfamily of receptor PTPs (RPTP). PTPμ has been shown to participate in homophilic binding. PTPμ on the extracellular surface of one cell binds to PTPμ on the surface of an adjacent cell ( 9– 11). As a transmembrane adhesion receptor, PTPμ has the ability to sense an extracellular signal via its extracellular segment and transduce this signal intracellularly via its phosphatase activity ( 12– 14). The PTPμ extracellular domain is composed of a MAM (meprin/A5-protein/PTPμ) domain, an immunoglobulin-like (Ig) domain, and four fibronectin type III (FNIII) repeats ( 12, 15, 16). The intracellular domain of PTPμ contains a juxtamembrane sequence with homology to cadherins and two phosphatase domains of which only the membrane proximal is catalytically active ( 17, 18). The juxtamembrane portion contains a helix-loop-helix wedge-shaped motif ( 14) that was targeted in the design of a peptide inhibitor of PTPμ function. This wedge peptide inhibitor specifically blocks PTPμ function in migration assays ( 19, 20).
PTPμ is expressed as a 200-kDa protein that is proteolytically cleaved in the fourth FNIII repeat, resulting in a 100-kDa extracellular fragment (E-subunit) that remains associated with the 100-kDa transmembrane and intracellular portion (P-subunit) through a noncovalent interaction ( 11, 21, 22). This cleavage is mediated by a furin-like protease in the endoplasmic reticulum during intracellular trafficking ( 21). Another type IIb RPTP, PTPκ, is also cleaved by a furin-like protease and further processed by an α-secretase of the ADAM (a disintegrin and metalloproteinase domain) family and a γ-secretase ( 23). The extracellular ADAM cleaves the P-subunit adjacent to the membrane to generate PΔE and shed the ectodomain ( 23). This cleavage primes PTPκ PΔE to be cleaved by γ-secretase, which releases the intracellular portion of PTPκ containing the active phosphatase domain from the membrane ( 23). The intracellular fragment of PTPκ translocates to the nucleus and controls β-catenin transcription ( 23). We previously observed a similar fragment of PTPμ containing the catalytically active intracellular domain (ICD) in the nucleus of a lung cell line ( 24).
We have previously shown that PTPμ protein is down-regulated in glioblastoma ( 8). Here, we show that overexpression of full-length PTPμ in glioblastoma cells suppresses cell migration and growth factor–independent cell survival. In addition, we propose that PTPμ down-regulation in glioblastoma is the result of sequential cleavage of full-length PTPμ protein to generate the fragments PΔE and ICD. In support of this hypothesis, the intracellular fragments of PTPμ are present in human glioblastoma samples and glioblastoma xenograft flank tumors. Surprisingly, short hairpin RNA (shRNA)–mediated down-regulation of PTPμ fragments decreases cell migration and growth factor–independent survival in glioblastoma cells. Furthermore, peptide inhibition of the function of PTPμ fragments inhibits cell migration. These data suggest that proteolytic cleavage of full-length PTPμ generates PTPμ fragments that regulate cell migration and growth factor–independent survival in glioblastoma. These PTPμ fragments can be targeted to develop novel therapeutic agents for glioblastoma patients.
Materials and Methods
Cell lines. The human GBM cell lines U-87 MG and LN-229 were obtained from the American Type Culture Collection. Human Gli36Δ5 glioblastoma cells have been described ( 25).
Lentiviral transduction. A human full-length PTPμ cDNA construct in pMT2 has been described ( 26). Full-length PTPμ was ligated into the lentiviral expression vector pCDH-MCS2 (System Biosciences). A full-length PTPμ-green fluorescent protein (GFP) fusion construct has been described ( 27). The PTPμ-GFP cassette was subcloned into pCDH-MCS2. An intracellular PTPμ-GFP fusion construct corresponding to PTPμ ICD has been described ( 24). Lentiviral shRNA constructs and the production of VSV-G–pseudotyped lentiviral particles have been described ( 8).
Immunoblotting. Cell lysates were prepared and immunoblotted as described ( 8) using normalized samples of ∼20 μg protein detected with monoclonal antibodies recognizing the intracellular segment of PTPμ (SK-7 or SK-18; ref. 28). An antibody against vinculin was from Sigma-Aldrich. The GFP antibody JL-8 was from Clontech.
Reverse transcription-PCR. Reverse transcription-PCR (RT-PCR) was performed as described ( 8). The PCR primers were as follows: extracellular, CGCGAATTCTAGAGACGTTCTCAGGTGGC (forward) and CCCGCAAGCTTACTTCTTCTCGCACTTG (reverse); intracellular, CGCGGATCCAAAGAGACCATGAGCAGCACCCGA (forward) and CCGGAATTCTCATCTGTTC-TCATCTTTCTTAGCCGA (reverse).
Scratch wound assay. Scratch wound assays were performed as described ( 8). Confluent monolayers of cells were scratched to induce a wound and analyzed by microscopy for the distance migrated by the leading edge of the wound at 0 and 24 h.
Colony formation assays. Growth factor–independent clonogenic colony assays were performed as described ( 29). Crystal violet–stained colonies were imaged with the Quantity One imaging software of the Gel Doc imaging system (Bio-Rad). Images were quantitated using MetaMorph software (Molecular Devices) by measuring the thresholded area of each well to include only colonies. For the soft agarose assay, cells were seeded at a concentration of 75,000/mL in 0.4% agarose and plated on an underlay of 0.8% agarose in a six-well plate. Colonies were analyzed after 4 wk by imaging Z-stacks of 20 random 10× fields using a Leica DMI6000B automated inverted microscope (Leica Microsystems GmbH) attached to a Retiga EXi camera (QImaging). The number of colonies in minimized Z-stacks from each microscope field was recorded.
Biotinylation of cell surface proteins. Cell surface biotinylation was performed using a Sulfo-NHS-SS-Biotin kit (Pierce). Biotinylated proteins were isolated and resolved by SDS-PAGE on 6% gels followed by immunoblotting with an antibody to PTPμ (SK-18) as described ( 30).
Inhibitors. The furin inhibitor I (Dec-RVKR-CMK; Calbiochem) was used at 50 μmol/L for 17 to 20 h. The γ-secretase inhibitors DAPT (Sigma-Aldrich) and L685,458 (Sigma-Aldrich) were used at 2 and 5 μmol/L, respectively, for 17 to 20 h. The proteasome was inhibited with MG132 (Sigma-Aldrich) at 20 μmol/L or epoxomicin (Calbiochem) at 5 μmol/L for 4 h. GM6001 (Calbiochem) was used at 50 μmol/L as a matrix metalloproteinase (MMP)/ADAM inhibitor for 17 to 20 h. Inhibitors were reconstituted in DMSO, which was used as a vehicle control. An inhibitor of PTPμ function targeting the helix-loop-helix wedge domain has beenshown to inhibit PTPμ function ( 19, 20). The PTPμ wedge peptide and a scrambled control peptide were synthesized to include a membrane-penetrant Tat-derived sequence at the COOH terminus to promote cellular uptake. Peptides synthesized by Genemed Synthesis or GenScript were reconstituted in water and added to cells at a final concentration of 5 μmol/L.
Immunoprecipitations. Cells were grown to confluence, treated with inhibitors, and lysed in 20 mmol/L Tris-HCl (pH 7.5), 1% Triton X-100, 150 mmol/L NaCl, 2 mmol/L EDTA, 1 mmol/L benzamidine, 5 μg/mL aprotinin, 5 μg/mL leupeptin, and 1 μg/mL pepstatin. Samples were sonicated and centrifuged at 10,000 rpm for 5 min. Immunoprecipitations from ∼400 μg total protein were performed as described ( 27) using a PTPμ antibody (SK-18) and resolved by SDS-PAGE on 8% gels followed by immunoblotting with an antibody to PTPμ (SK-7).
Immunocytochemistry. Immunofluorescent cell staining was performed as described ( 24). Fixed cells were probed with SK-7 or SK-18, which recognize intracellular PTPμ, and detected with goat anti-mouse Alexa Fluor 488 secondary antibody (Molecular Probes, Invitrogen). Slides were mounted with Citifluor antifadent mounting medium (Electron Microscopy Sciences) and imaged using the Leica system described above.
Tumor specimens. Fresh human brain and tumor tissues were obtained from surgical resections in accordance with an approved protocol from the University Hospitals Case Medical Center Institutional Review Board. GBM specimens of ∼100 mg each were obtained for protein extraction. Noncancerous, noneloquent, cortical brain was also collected.
GBM xenograft tumors were grown in NIH athymic nude female mice in accordance with an approved protocol from the Case Western Reserve University Institutional Animal Care and Use Committee. LN-229 or Gli36Δ5 cells (2 × 106) were resuspended in a 1:1 dilution of Matrigel (BD Biosciences) in PBS and injected s.c. in the right flank region of the mouse. Tumors were harvested between 9 and 28 d after injection. Lysates of human and xenograft tumor specimens were prepared as described ( 8). Tumor samples were homogenized using a tissue tearor homogenizer or a 2 mL Dounce homogenizer. Cleared lysates (∼20 μg from human samples and ∼50 μg from xenograft samples) were analyzed by immunoblot on 8% gels with an antibody to PTPμ (SK-18).
Statistics. Data presented represent at least three independent experiments. Replicates were normalized as a percent of the control, and the means were plotted using Microsoft Excel. Error bars indicate SE. Data were analyzed for statistical significance using an unpaired Student's t test.
PTPμ protein is down-regulated in the human glioblastoma cell line LN-229. We recently showed that PTPμ is endogenously expressed in the human GBM cell line U-87 MG and that shRNA-mediated down-regulation of PTPμ in U-87 MG cells promotes cell migration and dispersal ( 8). Furthermore, PTPμ protein is down-regulated in human GBM tumors and the migratory human GBM cell line LN-229. In the current study, PTPμ was overexpressed in LN-229 cells via a lentiviral construct, and both the full-length and normally produced P-subunit were detected by immunoblotting with an intracellular antibody to PTPμ ( Fig. 1A ). Lentiviral overexpression of PTPμ generated doublets at molecular weights corresponding to both full-length and P-subunit PTPμ ( Fig. 1A). These doublets likely are due to posttranslational modifications. mRNA expression of PTPμ was examined by RT-PCR in both U-87 MG and LN-229 cells. U-87 MG cells expressed PTPμ transcript as expected. Surprisingly, PTPμ transcript was also detected in LN-229 cells despite their lack of PTPμ protein expression ( Fig. 1B). PTPμ shRNA down-regulated PTPμ transcript but did not affect control glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Fig. 1C). These data suggest that the down-regulation of PTPμ in glioblastoma is due to a posttranscriptional mechanism.
Overexpression of PTPμ suppresses cell migration and growth factor–independent cell survival. We showed recently that shRNA-mediated down-regulation of endogenous PTPμ in U-87 MG cells promotes cell migration ( 8). Based on these data, we hypothesized that overexpression of PTPμ in LN-229 cells would suppress cell migration. We evaluated this hypothesis using a scratch wound assay. Confluent monolayers of LN-229 cells overexpressing either vector or PTPμ were scratched to form a wound. After 24 hours, control LN-229 cells at the leading edge of the wound migrated an average of 150 μm ( Fig. 2A ). However, LN-229 cells overexpressing PTPμ had impaired migration with a 3-fold reduction in the distance migrated ( Fig. 2A). Additionally, overexpression of PTPμ induced a morphologic change in LN-229 cells and made the cells noticeably elongated ( Fig. 2A). Because this assay occurred over 24 hours, it was possible that changes in cell proliferation could account for the difference in wound size. To rule out this possibility, LN-229 cells infected with vector or PTPμ were labeled with propidium iodide and analyzed by flow cytometry. Flow cytometry revealed no significant changes in cell proliferation between the vector- and PTPμ-infected cells (data not shown). Therefore, we concluded that the difference in wound size was due to a decrease in migration resulting from PTPμ overexpression, indicating that PTPμ suppresses migration of LN-229 glioblastoma cells.
Growth factor–independent survival is a hallmark of tumorigenesis. To assess the effect of PTPμ overexpression on growth factor–independent survival, a colony formation assay was used. After 2 weeks of growth factor deprivation, control LN-229 cells formed abundant colonies ( Fig. 2B). In contrast, overexpression of PTPμ reduced colony formation by 2-fold ( Fig. 2B). Therefore, PTPμ overexpression suppresses migration in two-dimensional culture and reduces growth factor–independent survival in three-dimensional culture of glioblastoma cells.
Proteolysis of PTPμ contributes to its down-regulation in glioblastoma. Other receptor tyrosine phosphatases are sequentially cleaved by a furin-like protease, an ADAM-type MMP, and a γ-secretase to release a soluble intracellular fragment ( 23, 31, 32). Because GBMs are known to have up-regulated proteases ( 33), we hypothesized that constitutive proteolysis of PTPμ may be the mechanism of PTPμ down-regulation in GBM. We first determined whether full-length PTPμ could be detected in parental LN-229 cells. Because we cannot detect PTPμ in a total cell lysate of parental LN-229 cells, we biotinylated cell surface proteins and used avidin resin to enrich the pool of biotinylated cell surface proteins. Despite the lack of PTPμ in the total cell lysate, the biotinylated cell surface fraction contained trace amounts of PTPμ ( Fig. 3A ). PTPμ is cleaved by a furin-like protease to generate the E- and P-subunits of PTPμ ( 11, 21, 22). As expected, treatment of cells with an inhibitor of furin activity resulted in an accumulation of full-length PTPμ (200 kDa) at the cell surface. These data imply that there is a trace amount of endogenous PTPμ in LN-229 cells that is processed by proteolysis. Biotinylation of cell surface proteins from LN-229 cells overexpressing PTPμ showed a similar pattern of full-length PTPμ accumulation at the cell surface on furin inhibition ( Fig. 3A).
After furin cleavage, PTPκ, another PTPμ subfamily member, is subsequently cleaved by α- and γ-secretases ( 23). We hypothesized that PTPμ is cleaved similarly. To test this hypothesis, LN-229 cells were treated with inhibitors of α- and γ-secretases. Proteasome inhibitors were used for biochemical detection to prevent rapid degradation of these fragments ( 23). Because we cannot detect PTPμ in whole-cell lysates, the PTPμ fragments were immunoprecipitated from LN-229 cells treated with inhibitors using antibody to the intracellular domain of PTPμ. The γ-secretase inhibitor DAPT stabilized a fragment that corresponds by molecular weight to a membrane-tethered truncated P-subunit termed PΔE ( Fig. 3B). Treatment with the proteasome inhibitor MG132 led to the accumulation of both PΔE and a soluble fragment termed PTPμ ICD ( Fig. 3B). The MMP inhibitor GM6001 limited the formation of PTPμ PΔE and ICD fragments, indicating that cleavage by a MMP is required for subsequent processing ( Fig. 3B). MG132 has been reported to inhibit γ-secretase activity in addition to proteasome activity, leading to the accumulation of α- and γ-secretase products ( 23, 34). Subsequent experiments included a more specific proteasome inhibitor, epoxomicin, to distinguish these events. Overall, these data support our hypothesis that the endogenous PTPμ expressed in LN-229 cells is constitutively cleaved to generate PTPμ PΔE and ICD. Consequently, little full-length PTPμ is present to function at the cell surface in LN-229 cells.
Total cell lysates from LN-229 cells overexpressing PTPμ showed a similar pattern of cleavage products on inhibitor treatment ( Fig. 3C). Stabilization of PTPμ ICD with treatment of epoxomicin confirmed that this fragment is labile and can only be seen when stabilized by the addition of a proteasome inhibitor. Treatment with MG132 and γ-secretase inhibitors (DAPT and L685,458) showed accumulation of PTPμ PΔE and ICD ( Fig. 3C). To verify if the cleavage products include the COOH terminus of the ICD of PTPμ, we overexpressed a PTPμ construct with a COOH-terminal GFP-tag (PTPμ-GFP) in LN-229 cells. Cells expressing PTPμ-GFP were treated with inhibitors as above, and total cell lysates were immuno-blotted with GFP to detect the PTPμ-GFP fragments. A GFP antibody detected a similar pattern of fragments, suggesting that PTPμ PΔE and ICD fragments include the COOH terminus of PTPμ ( Fig. 3C). These data support the model depicted in Fig. 3D. Full-length PTPμ is cleaved by a furin-like protease to generate the E- and P-subunits in “normal” proteolytic processing. Cleavage by an ADAM-type MMP (α-secretase) in GBM cells generates PTPμ PΔE. Subsequently, PΔE is cleaved by γ-secretase to generate PTPμ ICD.
PTPμ ICD is a soluble fragment that translocates to the nucleus in another cell type ( 24). To determine the subcellular localization of PTPμ ICD in glioblastoma cells, we performed immunocytochemistry on LN-229 cells. Antibodies recognizing the juxtamembrane (SK-7) and first phosphatase (SK-18) domains of PTPμ detected an endogenous PTPμ species with a nuclear pattern of localization similar to DAPI (4′,6-diamidino-2-phenylindole)–stained nuclei ( Fig. 4A ). The epitopes of these antibodies suggest that this species is PTPμ ICD. Overexpression of GFP-tagged PTPμ ICD also localized to the nucleus and confirmed these findings. In contrast, overexpression of GFP-tagged full-length PTPμ resulted in a cell-cell contact and filopodial staining pattern as reported previously ( 35). Full-length PTPμ likely senses extracellular adhesive cues to suppress migration by contact inhibition, whereas PTPμ ICD distributes to the cytoplasm and nucleus. These data suggest that full-length PTPμ and PTPμ ICD have distinct localization patterns, potentially leading to differences in their downstream signaling.
Intracellular fragments of PTPμ are expressed in human glioblastoma tumors and glioblastoma xenograft tumors. We previously showed that PTPμ protein expression is down-regulated in human GBM tumor samples ( 8). However, immunoblotting fresh GBM tumor tissue lysates on higher percentage gels indicated that fragments of PTPμ corresponding to PTPμ PΔE and ICD are expressed in human GBM tumor samples in comparison with normal brain tissue from the same patient ( Fig. 4B). Full-length PTPμ was undetectable in these GBM tumor samples ( Fig. 4B). PTPμ PΔE and ICD were identified in normal tissue samples that retain significant expression of full-length PTPμ ( Fig. 4B). Therefore, it is the expression of full-length PTPμ that differs between normal brain and GBM tumor tissue. Normal brain tissue expresses full-length PTPμ, whereas GBM tumor tissue does not express full-length PTPμ but retains PTPμ PΔE and ICD.
Neither full-length PTPμ nor PTPμ PΔE and ICD are detectable in LN-229 total cell lysates by immunoblot. We assessed human GBM cell line tumor xenografts grown in mouse flanks to determine if the three-dimensional architecture of the tumor would stabilize PTPμ fragments in the GBM cells. Flank tumor lysates from LN-229 xenografts expressed little detectable full-length PTPμ but expressed abundant PTPμ PΔE and ICD ( Fig. 4C). Similar results were obtained using xenografts prepared with another glioma cell line, Gli36Δ5 ( Fig. 4C). These data suggest that the three-dimensional human glioblastoma tumors and in vivo glioblastoma tumor models favor PTPμ proteolysis and stabilize PTPμ ICD and its precursor, PΔE, in vivo.
PTPμ fragments contribute to glioblastoma cell migration and both growth factor–independent and anchorage-independent cell survival. PTPμ ICD is a soluble fragment generated from PΔE that translocates to the nucleus ( Fig. 4A). PTPμ ICD contains the catalytic domain of PTPμ and has the potential to signal differently than that of membrane-bound, cell surface–associated PTPμ due to changes in substrate availability in different cellular compartments. Overexpression of membrane-bound, cell surface–associated PTPμ suppressed GBM cell migration and growth factor–independent survival ( Fig. 2). We hypothesized that PTPμ ICD and its precursor, PΔE, may signal differently and affect the migration and growth factor–independent survival of GBM cells. First, the effect of PTPμ fragments on cell migration was analyzed using a scratch wound assay.
PTPμ mRNA is expressed in LN-229 cells, but the only detectable proteins are PTPμ fragments ( Fig. 4). Therefore, we were able to use shRNA to down-regulate PTPμ fragments. Confluent monolayers of LN-229 cells expressing either control or two different PTPμ shRNA constructs were scratched and allowed to migrate ( Fig. 5A ). Down-regulation of PTPμ fragments by both shRNA constructs suppressed cell migration by 2-fold ( Fig. 5A). To rule out changes in cell proliferation, LN-229 cells infected with control or PTPμ shRNA were labeled with propidium iodide and analyzed by flow cytometry. No significant changes in cell proliferation were detected (data not shown).
Both PTPμ PΔE and ICD are partially stabilized by the γ-secretase inhibitor DAPT and are not formed when ADAMs are inhibited ( Fig. 3B). These inhibitors were used in a scratch wound assay to analyze their effects on PTPμ fragment–mediated cell migration. Stabilization of PTPμ fragments with DAPT increased migration, and prevention of PTPμ fragment formation by GM6001 decreased migration (Supplementary Fig. S1). These data suggest that proteolysis of PTPμ promotes LN-229 cell migration.
Because PTPμ overexpression affected growth factor–independent cell survival, we hypothesized that PTPμ fragments may also affect cell survival. To test this hypothesis, LN-229 cells expressing control or PTPμ shRNA were seeded at low density and allowed to form colonies over 2 weeks ( Fig. 5B). Down-regulation of PTPμ fragments via shRNA reduced the number of colonies in comparison with control cells by 3-fold ( Fig. 5B). These findings were confirmed in a soft agarose assay for anchorage-independent survival. PTPμ shRNA reduced the number of colonies in this assay by 5-fold ( Fig. 5C). These data suggest that PTPμ fragments promote both cell migration and growth factor–independent survival of glioblastoma cells.
Catalytic activity of PTPμ fragments is required for glioblastoma cell migration. Soluble intracellular PTPμ has been shown to retain catalytic activity ( 24, 28). To examine whether the catalytic activity of PTPμ fragments is important in the regulation of cell migration, PTPμ function was inhibited using a PTPμ-specific peptide inhibitor ( 19). Confluent monolayers of LN-229 cells were treated with a membrane-penetrant PTPμ wedge peptide or a control scrambled peptide before scratching to induce a wound ( Fig. 6 ). The PTPμ wedge peptide significantly reduced migration of LN-229 cells ( Fig. 6). This suppression is likely due to inhibition of the signaling of the PTPμ fragments as they are the only detectable PTPμ protein stabilized in LN-229 cells ( Fig. 4). These data suggest that PTPμ fragments must be catalytically active to induce GBM cell migration. Therefore, the wedge peptide inhibitor of PTPμ may have therapeutic value in the treatment of human glioblastoma.
Down-regulation of PTPμ in a human glioblastoma cell line that expresses PTPμ was reported to induce cell migration and dispersal ( 8). In this study, we show that overexpression of PTPμ suppresses migration and growth factor–independent survival of glioblastoma cells. Furthermore, down-regulation of PTPμ in GBM is due to proteolytic processing into a series of fragments. Human glioblastoma tumor samples selectively retain PTPμ fragments, both ICD and its precursor, PΔE, in comparison with patient-matched normal brain tissue. In the absence of full-length PTPμ, this PTPμ fragment signal promotes cell migration and growth factor–independent survival. The balance of full-length PTPμ and PTPμ fragment signaling is likely important in regulating the contact inhibition switch between cell adhesion and cell migration.
The receptor tyrosine phosphatases PTPκ, PTPζ/β, and LAR are regulated by sequential proteolysis ( 23, 31, 32). Furthermore, other transmembrane receptors, such as Notch, are similarly cleaved. Notch signaling is regulated by sequential cleavage by furin, ADAMs, and γ-secretase that ultimately generates an intracellular fragment ( 36). This fragment translocates to the nucleus and regulates the CBF1 transcription complex to control cellular processes such as differentiation and tumorigenesis ( 37). This regulation of cell surface receptors by proteolysis during development might be recapitulated during tumorigenesis as GBM cells have dedifferentiated, stem cell–like characteristics ( 3).
Differences in full-length PTPμ and PTPμ fragment signaling likely depend on the availability of PTPμ binding partners and downstream effectors. Furthermore, as a homophilic cell adhesion molecule, it may be that cell surface PTPμ and PTPμ fragment signaling pathways regulate the adhesive versus migratory switch of contact-inhibited or dispersive cells, respectively. Cell surface PTPμ binds and regulates cadherins and catenins ( 12), key components of classic adherens junctions. Four classic cadherin subtypes, E-, N-, R-, and VE-cadherin, associate with PTPμ ( 35, 38– 40). The cadherin binding partner p120-catenin (p120) has been implicated as a PTPμ binding partner and substrate ( 41) and contributes to tumorigenesis by regulating cell migration ( 42). p120 can translocate to the nucleus and associate with the transcription factor Kaiso ( 43). Interestingly, a proteolytically cleaved intracellular fragment of E-cadherin requires p120 for its nuclear translocation ( 44). The cytoplasmic domain of N-cadherin can also be proteolytically processed and translocate to the nucleus ( 45). p120 is involved in the recruitment of γ-secretase to N-cadherin for its cleavage ( 46). It is interesting to speculate that PTPμ fragments generated from the proteolytic cleavage of PTPμ may regulate a nuclear complex of N-cadherin and p120 given that PTPμ interacts with cadherins and p120 via its ICD ( 35, 41). Computer-based searches for a canonical nuclear localization sequence (NLS) in PTPμ were unsuccessful. However, both p120 and another PTPμ-interacting protein, BCCIP ( 24), contain NLS motifs ( 47, 48). The yeast homologue of BCCIP has been shown to regulate nuclear export ( 49). Therefore, p120 and BCCIP may aid in the shuttling of PTPμ ICD in and out of the nucleus.
Migration and dispersal of glioblastoma cells remains a clinical problem due to the lack of effective specific therapies ( 1– 3). Individual glioblastoma cells migrate and disperse throughout the brain parenchyma to form new foci. These cells must have elevated growth factor–independent survival signaling to evade anoikis-mediated cell death and to clonally expand. Therefore, it is interesting that both migration and growth factor–independent survival pathways are regulated by PTPμ fragments. Furthermore, a peptide inhibitor targeting PTPμ fragment function reduces cell migration. A small-molecule inhibitor that mimics this peptide will be developed to target PTPμ fragments and suppress glioblastoma cell migration and dispersal in vivo. Such an advance in the field of targeted therapeutics would fulfill a vast need for specific therapy in glioblastoma treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: NIH grant R01-NS051520 (S.M. Brady-Kalnay, S. Robinson, and R.H. Miller); National Cancer Institute grants K08-CA101954 and R01-CA116257, Ivy Brain Tumor Foundation, and Cancer Genome Atlas Project (A.E. Sloan); and NIH grants T32-GM007250 (Medical Scientist Training Program) and T32-CA059366 (A.M. Burgoyne). Additional support was obtained from the Visual Sciences Research Center Core Grant P30-EY11373 and the Case Comprehensive Cancer Center Core Grant P30-CA043703 from the NIH.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Moonkyung Caprara, Carol Luckey, and Theresa Gates for technical support; Sara Lou and Scott Howell for help with figures and graphs; and members of the Brady-Kalnay lab for insightful discussions.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
This article is dedicated to Tabitha Yee-May Lou who recently lost her battle with glioblastoma.
- Received March 5, 2009.
- Revision received June 17, 2009.
- Accepted June 23, 2009.
- ©2009 American Association for Cancer Research.