Protein-tyrosine Phosphatase α Acts as an Upstream Regulator of Fyn Signaling to Promote Oligodendrocyte Differentiation and Myelination*

The tyrosine kinase Fyn plays a key role in oligodendrocyte differentiation and myelination in the central nervous system, but the molecules responsible for regulating Fyn activation in these processes remain poorly defined. Here we show that receptor-like protein-tyrosine phosphatase α (PTPα) is an important positive regulator of Fyn activation and signaling that is required for the differentiation of oligodendrocyte progenitor cells (OPCs). PTPα is expressed in OPCs and is up-regulated during differentiation. We used two model systems to investigate the role of PTPα in OPC differentiation: the rat CG4 cell line where PTPα expression was silenced by small interfering RNA, and oligosphere-derived primary OPCs isolated from wild-type and PTPα-null mouse embryos. In both cell systems, the ablation of PTPα inhibited differentiation and morphological changes that accompany this process. Although Fyn was activated upon induction of differentiation, the level of activation was severely reduced in cells lacking PTPα, as was the activation of Fyn effector molecules focal adhesion kinase, Rac1, and Cdc42, and inactivation of Rho. Interestingly, another downstream effector of Fyn, p190RhoGAP, which is responsible for Rho inactivation during differentiation, was not affected by PTPα ablation. In vivo studies revealed defective myelination in the PTPα−/− mouse brain. Together, our findings demonstrate that PTPα is a critical regulator of Fyn activation and of specific Fyn signaling events during differentiation, and is essential for promoting OPC differentiation and central nervous system myelination.

Myelination is an essential feature of the vertebrate nervous system. The myelin sheath provides electrical insulation to axons and facilitates transmission of nerve impulses. Other important roles of myelin are to contribute to neuronal survival and development, as well as neurotransmission and synaptic activity (1). Deficiencies in myelination during development, or demyelination that can occur following injury or in diseases such as multiple sclerosis, lead to neurological disorders (2)(3)(4).
The formation of the highly specialized multilamellar myelin sheath by oligodendrocytes (OLs) 3 in the CNS occurs early in development, following proliferation and migration of oligodendrocyte progenitor cells (OPCs) to their final axonal targets (5). The molecular mechanisms that regulate OPC differentiation and OL maturation and myelination remain poorly understood. Consistent with the physical juxtaposition of axons and enwrapping oligodendroglia, axonal signals have been identified that influence OPC differentiation and/or myelination, such the axonal ligands Jagged1 and contactin that engage the glial receptor Notch (6,7). Other signals, such as those described below that are provided by components of the extracellular matrix or the presence or absence of growth factors, are also important in these processes.
The Src family tyrosine kinase (SFK) Fyn is an essential participant and central coordinator of OL differentiation, maturation, and myelination. Although mice null for the SFKs Src, Yes, or Lyn do not exhibit defects in CNS myelination, mice with mutant Fyn or lacking Fyn exhibit hypomyelination (8,9). In vitro studies have linked Fyn activation or inhibition to several stimuli that, respectively, induce or inhibit OL differentiation and maturation. Fyn is required for and activated in OL differentiation by serum withdrawal, IGF-1, ␤1 integrin stimulation (for example, by laminin binding to ␣6␤1 integrin), netrin-1 interaction with the receptor Dcc, and antibody-mediated cross-linking of MAG or FcR␥ (8, 10 -15). Recently, co-stimulation of an integrin-contactin complex in OLs was found to amplify Fyn activation and promote myelination (16). Conversely, inhibition of OL differentiation by LINGO-1 or myelin protein extract mimicking the myelin debris generated by demyelination results in reduced Fyn activity (17,18). Fyn signals to several molecules that are important for OL morphological changes that require cytoskeletal rearrangement and process extension and elaboration, such as focal adhesion kinase (FAK), the Rho GTPases Rho, Rac1, and Cdc42, the Rho regulators p190 and p250 RhoGAP, Tau protein, and possibly via the kinase Cdk5 to paxillin (13, 19 -23). It also controls myelin production transcriptionally and post-transcriptionally (24,25).
SFK activation in various cell types and model systems is regulated by catalytic and non-catalytic mechanisms, and is manifested by altered SFK tyrosine phosphorylation. Dephosphorylation of the C-terminal tail inhibitory tyrosine residue by protein-tyrosine phosphatases (PTPs) is an important event in SFK activation (26). The potential roles of PTPs in coupling upstream signals, many of which involve engagement of catalytically inactive receptors, to Fyn activation in OL differentiation have not been extensively investigated. Indeed, although several PTPs have been implicated in OL differentiation and myelination, little is known of their specific actions in these processes. Investigation of regulated PTP expression during differentiation of the CG4 OL cell line revealed the most abundant of the 11 transcripts identified to be those of 4 receptortype PTPs: PTP␣, PTP⑀, PTP, and PTP␥ (27). Roles for several of these receptor PTPs in in vivo myelination have been described. PTP␣ and PTP⑀ are involved in peripheral nervous system myelination (28,29). CNS myelination appears normal in PTP-null mice, although myelin stability may be reduced (30). However, remyelination is impaired in PTP-null mice after experimental autoimmune encephalomyelitis-induced demyelination (31). Mice null for another receptor PTP, CD45, have reduced numbers of MAG-positive myelinating OLs and exhibit general, mild dysmyelination (32). FcR␥-stimulated CD45 Ϫ/Ϫ OPCs show impaired morphological differentiation and a lack of MBP up-regulation. In normal OPCs, CD45 is complexed with Fyn and the Fyn negative regulatory kinase Csk, suggesting that CD45 is a candidate activator of Fyn and Fyn-MBP signaling in response to FcR␥ engagement. Myelination defects are also detected in mice lacking the cytosolic PTP SHP-1, and MBP expression is reduced in OLs from the mice (33).
Although PTP␣ mRNA is up-regulated in OL differentiation, a role for PTP␣ in this process and in CNS myelination has not been described. PTP␣ is a well characterized activator of SFKs (34). It is an essential component of several SFK-dependent signaling systems, and acts in physical and functional conjunction with various ligand-activated receptors; such as integrins, contactin, NCAM, and c-Kit, to catalyze Fyn dephosphorylation and activation (35)(36)(37)(38). We therefore investigated whether PTP␣ is required for OL differentiation and Fyn activation and signaling in this process using two model systems: the cultured CG4 OL cell line in which PTP␣ expression was ablated by siRNA-mediated silencing, and primary OPCs derived from wild-type and PTP␣ Ϫ/Ϫ mouse embryos. We report that PTP␣ is required for OL differentiation and morphological changes. It acts as an essential but not the only regulator of Fyn in differentiating OLs, and is required for Fyn signaling to several effectors that includes a distinct mechanism of Rho inhibition. Furthermore, we find that forebrain myelination is impaired in PTP␣-deficient mice.
Cell Line and Primary Cell Cultures-CG4 cells were kindly provided by Dr. Y. Feng (Emory University School of Medicine) and maintained in CG4 proliferation medium (Dulbecco's modified Eagle's medium, 1% FBS, 5 g/ml of insulin, 50 g/ml of transferrin, 30 nM sodium selenite, 100 M putrescine, 20 nM progesterone, 10 ng/ml of biotin, 10 ng/ml of platelet-derived growth factor, and 10 ng/ml bFGF). To promote differentiation, cells were seeded on poly-D-lysine (PDL, 10 g/ml)-coated dishes at a density of 1.5 ϫ 10 4 /cm 2 . After attachment (ϳ3 h), cells were gently washed and subsequently cultured in CG4 differentiation medium (Dulbecco's modified Eagle's medium, 0.5% FBS, 5 g/ml of insulin, 50 g/ml of transferrin, 30 nM sodium selenite, 50 nM triiodothyronine) for various times. Primary mouse OPCs were generated from neurospheres as described previously (40,41) with some modifications. In brief, after removal of meninges and cerebellum, cerebral cortex tissue from E14.5-E17.5 mouse embryos was mechanically triturated with a 1-ml Gilson pipette until the cell suspension had no or very few small clumps, filtered through a 70-m cell strainer, and plated at 5 ϫ 10 4 cells/ml in a six-well plate (4 ml/well of neural culture medium supplemented with 20 ng/ml of bFGF and 20 ng/ml of epidermal growth factor). Neural culture medium contains Dulbecco's modified Eagle's medium/F-12, 25 g/ml of insulin, 100 g/ml of apo-transferrin, 20 nM progesterone, 60 M putrescine, and 30 nM sodium selenite. After 3-4 days, floating neurospheres were passaged at a 1:3 ratio in the same medium every 3-4 days. Passage 2-6 neurospheres were mechanically dissociated into a single cell suspension and resuspended in neural culture medium supplemented with 20 ng/ml of platelet-derived growth factor-AA and 20 ng/ml of bFGF (oligosphere medium) to induce oligosphere formation. After 72 h, cell aggregates were passaged at a 1:2 ratio every 4 -6 days. Oligospheres (passage 2-6) were dissociated using the NeuroCult Chemical Dissociation kit (mouse) (StemCell Technologies, Alberta, Canada) and plated on poly-DL-ornithine (PDLO, 50 g/ml)-coated chamber slides or dishes at a density of 3 ϫ 10 4 /cm 2 in oligosphere medium for 2 days. To induce differentiation, medium was changed to neural culture medium supplemented with 5 g/ml of N-acetyl-L-cysteine and 50 nM triiodothyronine for 2-4 days.
Immunofluorescence Labeling of Cells and Tissues-Cells grown on PDL-or PDLO-coated coverslips or chamber slides (Nalgene Nunc International, Rochester, NY) were fixed with 4% paraformaldehyde for 15 min at room temperature and then washed three times with PBS. For other experiments, animals were anesthetized and perfused intracardially with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.2. The brain was removed, post-fixed in the same solution overnight at 4°C, and then cryopreserved in 30% sucrose in 0.1 M phosphate buffer and embedded in OCT media. Cells and tissue sections (10 m) were incubated with blocking buffer (0.1 M phosphate buffer, 0.1% gelatin, 1% bovine serum albumin, 0.02% sodium azide, 10% goat serum) for 30 min (0.5% Triton X-100 was added to the blocking buffer if permeablization was required), followed by incubation with primary antibodies overnight at 4°C. After washing three times with PBS, cells and tissues were incubated with secondary antibodies for 2 h at room temperature. The slides were washed three times with PBS followed by mounting in Prolong Gold Antifade Reagent (Invitrogen) with 4Ј,6-diamidino-2-phenylindole and viewed using an Axioplan2 fluorescence microscope (Carl Zeiss MicroImaging, Thornwood, NY).
Immunoblotting-Cells were harvested by washing twice with ice-cold PBS on ice. For preparation of lysates, cells were lysed on ice by adding RIPA lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% sodium deoxycholate, 1% Nonidet P-40, 0.1% SDS, 1 mM EDTA, 2 mM sodium orthovanadate, 50 mM sodium fluoride, 10 g/ml of aprotinin, 10 g/ml of leupeptin, and 1 mM phenylmethylsulfonyl fluoride) or Nonidet P-40 lysis buffer (RIPA lysis buffer without sodium deoxycholate and SDS) directly onto the cells. Cell lysates were then transferred to microtubes and incubated for 30 min on ice, centrifuged at 12,000 ϫ g for 10 min at 4°C, and the supernatants collected to obtain protein extracts. Protein concentration was determined with the BCA Protein Assay kit (Pierce). Protein extracts were resolved by SDS-PAGE and transferred to a polyvinylidene difluoride membrane, which was then blocked with 3% bovine serum albumin in PBS with 0.1% Tween 20 for 1 h at room temperature. The membranes were probed overnight at 4°C with the relevant antibodies, washed, and probed again with species-specific secondary antibodies coupled to horseradish peroxidase. Chemiluminescent reagents were then added for signal detection.
Immunoprecipitation-Cell lysates (50 -100 g of protein) prepared with RIPA buffer were immunoprecipitated with the indicated antibody at 4°C overnight, followed by incubation with 40 l of protein A/G-agarose (Santa Cruz Biotechnology) at 4°C for 3 h. After washing three times with lysis buffer, the immunoprecipitates were analyzed by immunoblotting.
Rho Family GTPase Activities-Rho activity was measured by GST-RBD (rhotekin-binding domain) pulldown assays per-formed using the Rho activation assay kit (Upstate, Temecula, CA). Rac1 and Cdc42 activities were measured by GST-PBD (PAK-binding domain) pulldown assays (42). Cells were lysed on ice by adding RIPA lysis buffer directly onto the cells. Cell lysates (50 -100 g) were incubated with 10 g of GST-PBD bound to glutathione-Sepharose beads. Samples were washed with lysis buffer and then immunoblotted with anti-Rac1 and Cdc42 antibodies. Lysates were directly immunoblotted to determine the total amount of Rho, Rac1, or Cdc42 proteins. Levels of active Rho, Rac1, and Cdc42 were normalized to those of total Rho, Rac1, and Cdc42.
Data Analysis-Densitometric quantification of immunoblots and cell differentiation data were statistically analyzed using ANOVA (single factor).

RESULTS
Characterization of CG4 Cell Differentiation-To study OL differentiation, we used the rat-derived CG4 cell line (A2B5positive OPCs) (43). When the cells were plated on PDL-coated dishes and incubated in CG4 differentiation medium for up to 6 days, they underwent differentiation from A2B5-positive bipolar OPCs into CNPase-positive OLs that extend multiple branched processes (Fig. 1A). To confirm that specific OL markers were up-regulated in these cells after differentiation was induced, cell lysates were subjected to immunoblotting with anti-CNPase antibody. CNPase protein levels were upregulated in CG4 cells during differentiation (Fig. 1B). It has been reported that PTP␣ mRNA expression increases during OL differentiation (27). As shown in Fig. 2A, we found that the PTP␣ protein level was up-regulated ϳ1.5fold in CG4 cells after differentiation for 2 days and was gradually up-regulated to ϳ2.5-fold during differentiation days 3-6, indicating that PTP␣ may play a role in this process.
Because PTP␣ can dephosphorylate and activate SFKs (34), we characterized the protein expression of the three SFKs present in OLs, Fyn, Lyn, and Src (25), in differentiating CG4 cells. Among these SFKs, Fyn plays a unique role in myelination, because myelin deficits are only found in Fyn Ϫ/Ϫ mice and not in Lyn Ϫ/Ϫ or Src Ϫ/Ϫ mice (9). As shown in Fig. 2B, Fyn protein level rapidly increased (2.8-fold) over the first 2 days of CG4 differentiation, and then increased slightly further and was maintained for the remainder of the 6-day differentiation period. Consistent with previous studies (10,12,24), we found that Lyn protein level increased by 1.7-fold over the first 2 days of differentiation, and then decreased over subsequent days to return to the starting level or somewhat lower by days 5-6. Src protein continually decreased during differentiation, and by day 6 was reduced to 35% of the starting level. In conjunction with these findings, and as Fyn is reported to be the only SFK with significant kinase activity in either cultured OPCs or OLs (10), we investigated Fyn activity in differentiating CG4 cells by determining its phosphorylation status. Phosphorylation of Fyn at the negative regulatory C-terminal tail residue Tyr 531 was significantly reduced during CG4 differentiation, especially over the first 2 days (Fig. 2C), indicative of Fyn activation. Overall, Fyn Tyr 531 phosphorylation per unit of Fyn protein decreased to 34% of the starting level by day 6.
PTP␣ Is Required for CG4 Differentiation-To investigate the role of PTP␣ in OL differentiation, we generated PTP␣-deficient CG4 cells using siRNA. Cells were cultured in proliferation medium and transfected, 24 h later they were seeded on PDL-coated plates. After 3 h of attachment in proliferation medium, the medium was changed to CG4 differentiation medium (differentiation day 0). Lysates of control siRNA-and PTP␣ siRNA-treated CG4 cells were prepared from cells maintained for 0 -3 days in differentiation medium, and examined by immunoblotting to determine the effectiveness of siRNAmediated knockdown (Fig. 3A). PTP␣ expression was reduced by more than 90% during the 24 -72 h following siRNA transfection (differentiation day 0 -2). To evaluate the differentiation of control siRNA-and PTP␣ siRNA-treated CG4 cells after 2 days in differentiation medium, they were immunostained with anti-A2B5 and anti-CNPase antibodies for microscopic visualization and quantitative measurements. The numbers of A2B5-positive cells were counted, and about 4 times more cells in the PTP␣-knockdown CG4 population were found to remain A2B5-positive (progenitor-like) compared with the control siRNA-treated cell population (Fig. 3, B and C). CNPase immunofluorescence revealed multiple branched processes that were formed by the control siRNA-treated cells, but that were lacking in the PTP␣-directed siRNA-treated cells (Fig. 3B, bottom panels). As cell process extension is an indicator of differentiation, we counted the number of processes per cell with a length greater than that of the cell body. After 2 days in differentiation medium, significantly more PTP␣-knockdown CG4 cells had a low number (2 or less) of extended processes per cell, and fewer PTP␣-knockdown CG4 cells had a high number (4 or 5) of extended processes per cell, compared with control siRNAtreated cells (Fig. 3, B and D). Moreover, CNPase protein . The band intensity of each protein was normalized to that of actin. C, Fyn phosphorylation at Tyr 531 was analyzed by immunoprecipitation (IP) followed by immunoblotting with anti-Src P-Tyr 527 antibody. Membranes were stripped and reprobed with anti-Fyn antibody. The band intensity of Fyn P-Tyr 531 was normalized to that of Fyn. Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D. *, p Ͻ 0.05; **, p Ͻ 0.01, ANOVA test. expression was lower in PTP␣-knockdown CG4 cells than in control siRNA-treated cells after differentiation (Fig. 3E). These results indicate that PTP␣ is required for the differentiation of CG4 cells into OLs.
PTP␣ Is Required for Activation of Fyn and Fyn Effectors FAK, Rac1, and Cdc42 during CG4 Differentiation-To determine whether PTP␣ is required to dephosphorylate and activate Fyn in differentiating CG4 cells, the phosphorylation of Fyn at Tyr 531 was determined in control siRNA-and PTP␣ siRNAtreated CG4 cells that were induced to differentiate for 2 days. Fyn immunoprecipitates from PTP␣-directed siRNA-treated cells contained higher levels of phospho-Tyr 531 -Fyn than those from control siRNA-treated cells (Fig. 4A), indicating that PTP␣ is required to dephosphorylate this tyrosine residue and activate Fyn in differentiating CG4 cells.
The effect of silencing PTP␣ on the activation of several downstream effectors of Fyn was examined. FAK activation involves the phosphorylation of FAK Tyr 576 (44), and this is reported to be up-regulated in a Fyn-dependent manner during differentiation of CG4 cells (19). We confirmed that phosphorylation of FAK Tyr 576 increases during CG4 differentiation (Fig. 4B, upper panels). This required PTP␣, because compared with control siRNA-treated CG4 cells, the PTP␣ siRNA-treated cells displayed significantly reduced (by ϳ40%) phospho-Tyr 576 -FAK after induction of differentiation (Fig. 4B, lower panels).
The Rho family GTPases Rac1 and Cdc42 play important roles in cytoskeleton rearrangement and are crucial for morphological differentiation of OLs and myelination (13,45). It has also been reported that activation of Rac1 and Cdc42 is dependent on the activity of Fyn and FAK in differentiating OLs (13,19). We therefore investigated whether the activities of Rac1 and Cdc42 were affected by PTP␣ silencing in CG4 cells following the induction of differentiation. Using GST-PBD pulldown assays to measure the levels of active GTP-bound Rac1 and Cdc42, we found that both Rac1 and Cdc42 were activated in CG4 cells during differentiation (Fig. 4C, upper  panels). However, the differentiation-induced activity of both Rac1 and Cdc42 was significantly reduced by more than 50% in PTP␣-knockdown CG4 cells placed in differentiation medium for 2 days (Fig. 4C, lower panels). These results indicate that PTP␣ is required for Fyn-mediated signaling to FAK, Rac1, and Cdc42 in differentiating CG4 cells.
PTP␣ Is Not Required for Fyn-mediated Signaling to p190RhoGAP, but Is Required for Rho Inactivation during CG4 Differentiation-During OL differentiation, the Fyn-interacting protein and substrate p190RhoGAP is tyrosine phosphorylated, resulting in increased p190RhoGAP activity that promotes Rho inhibition and OL differentiation (13,20). In accord with these findings, we observed that p190RhoGAP co-immunoprecipitated with Fyn in both progenitor and differentiating CG4 cells and that tyrosine phosphorylation of p190RhoGAP increased after differentiation for 2 days (Fig. 5A). To investigate if PTP␣ is an upstream regulator of differentiation-induced Fyn signaling to p190RhoGAP, we determined Fyn and p190RhoGAPassociationandp190RhoGAPtyrosinephosphorylation in CG4 cells that were treated with control siRNA or PTP␣-directed siRNA and induced to differentiate for 2 days. Surprisingly, both the Fyn-p190RhoGAP interaction and tyrosine phosphorylation of p190RhoGAP were not affected by PTP␣ siRNA treatment (Fig. 5B). These results suggest that PTP␣ does not act upstream of Fyn-mediated regulation of p190RhoGAP, and that PTP␣ thus regulates specific aspects of Fyn signaling in differentiating CG4 cells.
The Rho family GTPase Rho plays important roles in controlling cellular morphology. Overexpression of constitutively active Rho inhibits process extension in OLs, whereas overexpression of dominant-negative Rho results in a hyperextension of OL processes (20). Because the primary function of p190RhoGAP is to inactivate Rho, we determined Rho activities to determine whether, like p190RhoGAP, this was unaffected upon PTP␣ silencing. Using GST-RBD pulldown assays to measure the levels of active GTP-bound Rho, we found that Rho was inactivated in CG4 cells during differentiation (Fig. 5C). Compared with control CG4 cells, Rho activity was significantly increased by ϳ3-fold in PTP␣- knockdown CG4 cells placed in differentiation medium for 2 days (Fig. 5C). Taken together, these results indicate that PTP␣ is required for Rho inactivation in differentiating CG4 cells, but in a manner independent of Fyn-p190RhoGAP signaling.
Isolation and Characterization of OPCs from Mouse Embryonic Neural Progenitors-To extend our findings from CG4 cells where PTP␣ expression was transiently silenced, we investigated the role of PTP␣ in OPC differentiation in a different model system where PTP␣ expression was permanently ablated, using primary OPCs isolated and cultured from PTP␣null mouse embryos. Mouse OPCs are reportedly more difficult to isolate as cultured cells than rat OPCs, because they do not share all of the cell surface antigens with rat, such as A2B5, and they tend to differentiate in in vitro mixed glial cultures (40,46,47). Several studies have determined that self-renewing OPCs can be generated from neural progenitor/stem cells of different species (40, 41, 47-50), and we utilized these procedures with some modifications (see "Experimental Procedures") to isolate and culture OPCs from wild-type (WT) and PTP␣ Ϫ/Ϫ mouse embryos.
Neuronal markers are not detected in OPC cultures derived from oligospheres (41), suggesting that these OPCs are mainly OL type-2 astrocyte (O-2A) progenitors that can differentiate into OLs or type-2 astrocytes. WT mouse OPCs from P2 oligospheres were seeded on PDLO-coated dishes in OPC proliferation medium for 2 days. Cells were fixed and immunostained with antibodies against the progenitor marker NG2, the pre-OL marker O4, and the astrocyte marker GFAP. As shown in Fig. 6A, Ͼ95% of WT mouse OPCs were positive for NG2 and negative for either O4 or GFAP. The cells were then induced to differentiate by mitogen withdrawal and thyroid hormone (triiodothyronine) exposure (51). In addition, N-acetyl-Lcysteine was added to support cell survival (52). Two growth factors, IGF-1 (53, 54) and ciliary neurotrophic factor (CNTF) (55) can promote OPC differentiation and survival. Therefore, we also examined the effects of IGF-1 and CNTF on the differentiation of mouse OPCs. We found that after differentiation for 2 days, the numbers of NG2ϩ cells decreased and O4ϩ cells increased in all conditions tested (Fig. 6A). In addition, O4 immunostaining revealed that more cells extended highly branched processes in the presence of IGF-1, and that there were less O4ϩ cells but more GFAPϩ cells in the presence of CNTF. Moreover, very few GFAPϩ cells were detected in OPC differentiation medium with or without IGF-1. These results are consistent with previous studies showing that IGF-1 induces OPCs to differentiate into OLs (56), whereas CNTF also induces OPCs to differentiate into type-2 astrocytes (57). Therefore, to enrich the OL population after OPC differentiation, we used OPC differentiation medium with or without IGF-1 for further studies.
To characterize the role of PTP␣ in primary mouse OLs, we first examined PTP␣ protein expression during WT OPC differentiation. Consistent with our finding in CG4 cells, PTP␣ protein expression, as well as that of Fyn and CNPase, increased during the differentiation of primary mouse OPCs (Fig. 6B).
PTP␣ Is Required for Primary Mouse OPC Differentiation-To investigate the role of PTP␣ in the differentiation of primary mouse OPCs, WT and PTP␣ Ϫ/Ϫ OPCs were induced to differ-

. PTP␣ is required for activation of Fyn and its downstream effectors FAK, Rac1, and Cdc42 in CG4 cells. A, reduced activity of Fyn in PTP␣-knockdown CG4 cells at differentiation day 2. Fyn phosphorylation
at Tyr 531 was analyzed by immunoprecipitation (IP) with anti-Fyn antibody followed by probing with anti-Src P-Tyr 527 and anti-Fyn antibodies. The band intensity of Fyn P-Tyr 531 was normalized to that of Fyn. Data obtained from three independent experiments are expressed as the mean Ϯ S.D. *, p Ͻ 0.05, ANOVA test. B, CG4 cells were differentiated for 0 -6 days. Tyrosine phosphorylation of FAK Tyr 576 and protein expression of FAK were determined by immunoblotting with anti-FAK P-Tyr 576 and anti-FAK antibodies, respectively (top panels). Control siRNA or PTP␣ siRNA-treated cells were differentiated for 0 -2 days. FAK tyrosine phosphorylation at Tyr 576 and FAK amounts were determined as described above (bottom panels). The band intensity of FAK P-Tyr 576 (differentiation day 2) was normalized to that of FAK. Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D. **, p Ͻ 0.01, ANOVA test. C, cell lysates from CG4 cells grown on PDL-coated dishes in CG4 differentiation medium for 0 -6 days were incubated with a GST-PBD fusion protein to pull down the GTP-bound forms of Rac1 and Cdc42. The amount of Rac1 and Cdc42 in the pulldown assay as well as in the whole cell lysates (WCL) were determined by immunoblotting with anti-Rac1 and anti-Cdc42 antibodies, respectively (top panels). Control siRNA or PTP␣ siRNA-treated cells were differentiated for 2 days. Active Rac1 and Cdc42 were determined as described above. The band intensity of GTPbound Rac1 and Cdc42 was normalized to that of Rac1 and Cdc42 in whole cell lysates. Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D. *, p Ͻ 0.05; **, p Ͻ 0.01, ANOVA test. entiate for 2 days in OPC differentiation medium with or without IGF-1. After differentiation in both conditions, PTP␣ Ϫ/Ϫ OLs expressed less of the OL marker CNPase than did WT OLs (Fig. 7A), suggesting that their differentiation was impaired. This was confirmed by immunostaining with antibodies against the progenitor marker NG2 and the pre-OL marker O4. As shown in Fig. 7B, after differentiation for 2 days in medium with or without IGF-1, fewer WT than PTP␣ Ϫ/Ϫ cells were NG2ϩ (top panels), and more WT than PTP␣ Ϫ/Ϫ cells were O4ϩ (lower panels). The NG2ϩ and O4ϩ cells of each genotype were quantified (Fig. 7C), demonstrating that about 3-4-fold higher populations of O4ϩ WT cells than PTP␣ Ϫ/Ϫ cells, and about 3-3.5-fold lower populations of NG2ϩ WT cells than PTP␣ Ϫ/Ϫ cells, were present after 2 days in differentiation medium Ϯ IGF-1. These results further confirm that PTP␣ is required for OPC differentiation.

PTP␣ Selectively Regulates Fyn Activation and Signaling in Primary Mouse
OPCs-To confirm that PTP␣ is required for Fyn dephosphorylation at its negative regulatory site, lysates from WT and PTP␣ Ϫ/Ϫ progenitors or differentiating OLs were immunoprecipitated with anti-Fyn antibody followed by immunoblotting. Fyn phospho-Tyr 528 (the equivalent of rat Fyn Tyr 531 ) was enhanced 2-fold in PTP␣ Ϫ/Ϫ OPCs and differentiating OLs compared with the WT group, suggesting that PTP␣ is required for Fyn dephosphorylation at Tyr 528 in both progenitors and OLs (Fig. 8A). Fyn phospho-Tyr 528 decreased by ϳ40% in both WT and PTP␣ Ϫ/Ϫ cells after differentiation Ϯ IGF-1 for 2 days (Fig. 8A), suggesting that there are also other regulators of Fyn activation during OPC differentiation. Despite the decreased phosphorylation of Fyn Tyr 528 that occurs in both WT and PTP␣ Ϫ/Ϫ cells during differentiation, the level of Fyn Tyr 528 phosphorylation in the PTP␣ Ϫ/Ϫ cells after 2 days differentiation was not significantly lower than that in undifferentiated WT cells (Fig. 8A). In conjunction with the impaired differentiation of PTP␣ Ϫ/Ϫ cells that we observe, this suggests that non-PTP␣-mediated Fyn Tyr 528 activation is insufficient to promote mouse OPC differentiation.

FIGURE 5. PTP␣ is not required for tyrosine phosphorylation of p190RhoGAP (p190) and Fyn-p190 interaction in CG4 cells.
A, p190-Fyn interaction and tyrosine phosphorylation of p190 in CG4 cells. CG4 cells were induced to differentiate for 0 -2 days. Immunoprecipitation (IP) complexes isolated using antibodies against Fyn (top two panels) or p190 (bottom two panels) were evaluated by immunoblotting with anti-p190, Fyn, or P-Tyr (4G10) antibodies. B, p190-Fyn interaction and tyrosine phosphorylation of p190 are not altered by PTP␣-knockdown in differentiating CG4 cells. Control and PTP␣ siRNA-treated CG4 cells were induced to differentiate for 2 days. Immunoprecipitation complexes isolated using antibodies against Fyn or p190 were evaluated by immunoblotting with anti-p190, Fyn, or P-Tyr (4G10) antibodies. The band intensity of p190 was normalized to that of Fyn in Fyn immunoprecipitates (top graph). The band intensity of P-Tyr (4G10) was normalized to that of p190 in p190 immunoprecipitates (bottom graph). Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D. No significant differences were found between control siRNA and PTP␣ siRNA-treated cells. C, cell lysates from CG4 cells grown on PDLcoated dishes in CG4 differentiation medium for 0 -6 days were incubated with a GST fusion containing the RBD to pull down the GTP-bound form of Rho. The amount of Rho in the pulldown assay as well as in the whole cell lysate (WCL) was determined by immunoblotting with anti-Rho antibody. Control siRNA or PTP␣ siRNA-treated cells were differentiated for 2 days. Rho activity was determined as described above. The band intensity of GTPbound Rho was normalized to that of Rho in WCL. Data obtained from three independent experiments are expressed as the mean Ϯ S.D. *, p Ͻ 0.05, ANOVA test. FIGURE 6. Characterization of primary mouse OPC differentiation. A, WT mouse P2 oligospheres were dissociated and seeded on PDLO-coated chamber slides for 2 days in OPC proliferation medium followed by incubation for the indicated times in OPC differentiation medium alone (Ϫ), or with 100 ng/ml of IGF-1 or 10 ng/ml of CNTF. Cells were immunostained with antibodies against NG2, O4, and GFAP. Scale bar, 20 m. B, WT mouse P2 oligospheres were dissociated and seeded on PDLO-coated dishes for 2 days in OPC proliferation medium followed by incubation for the indicated times in OPC differentiation medium. The levels of PTP␣, Fyn, CNPase, and actin were determined by immunoblotting cell lysates. The band intensity of PTP␣ (differentiation day 0 and 2) was quantified and normalized to that of actin. Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D. **, p Ͻ 0.01, ANOVA test. DAPI, 4Ј,6-diamidino-2-phenylindole.
To investigate whether PTP␣ is required for FAK activation in primary mouse OPCs, we examined FAK Tyr 576 phosphorylation in WT and PTP␣ Ϫ/Ϫ progenitors and differentiating OLs. As shown in Fig. 8B, FAK phospho-Tyr 576 was significantly reduced by 60 and 50%, respectively, in PTP␣ Ϫ/Ϫ progenitors and differentiating OLs compared with WT cells. Although FAK phosphorylation at Tyr 576 increased in both WT and PTP␣ Ϫ/Ϫ cells after differentiation was induced, the FAK phosphorylation level in the PTP␣ Ϫ/Ϫ cells only increased to a level equivalent to that in undifferentiated WT cells (Fig. 8B). Thus the differentiation-induced modulation of FAK in both WT and PTP␣ Ϫ/Ϫ cells correlated closely with that of Fyn, further indicating that PTP␣ is required for Fyn-mediated FAK activity in primary mouse OPCs and differentiating OLs.
To determine whether PTP␣ is required for Fyn-p190RhoGAP signaling in the primary mouse cell system, p190RhoGAP immunoprecipitates were prepared from lysates of WT and PTP␣ Ϫ/Ϫ progenitors and differentiating OLs and analyzed. Increased amounts of p190RhoGAP immunoprecipitated from both WT and PTP␣ Ϫ/Ϫ cells that had been induced to differentiate for 2 days compared with the undifferentiated OPCs (Fig. 8C, middle panel), but determination of the phosphotyrosine incorporated into the p190RhoGAP protein revealed that there were no differentiation-induced changes in tyrosine phosphorylation of p190RhoGAP in either cell type or between the cell types (Fig. 8C, top panel and graph). It has been reported that p190RhoGAP associates with p120RasGAP in OLs (20), and that Src homology 2 domain of p120RasGAP binds tyrosine-phosphorylated p190RhoGAP (58). We investigated whether PTP␣ might affect the p190RhoGAP⅐p120RasGAP complex formation upon differentiation. Although p120RasGAP was detected in p190-RhoGAP immunoprecipitates, no difference in their extent of association was apparent between WT and PTP␣ Ϫ/Ϫ cells or between progenitors and OLs of each genotype (Fig. 8C,  bottom panel), indicating that PTP␣ is not required for Fynmediated p190RhoGAP-p120RasGAP signaling in progenitors and OLs.
To determine whether PTP␣ is required for Rac1 and Cdc42 activation and Rho inactivation in the primary mouse cell system, we examined Rac1, Cdc42, and Rho activities in WT and PTP␣ Ϫ/Ϫ OLs. As shown in Fig. 8, D and E, Rac1 and Cdc42 activities were reduced, and Rho activity was increased in PTP␣ Ϫ/Ϫ OLs compared with WT cells.

Ablation of PTP␣ Results in Decreased MBP Protein Expression in Primary Mouse OLs and Leads to Defective Myelination-
Fyn directly stimulates the promoter activity of the MBP gene and is involved in post-transcriptional regulation of MBP mRNA (24,25). Therefore, we investigated if PTP␣ is required for Fyn-MBP signaling. The WT and PTP␣ Ϫ/Ϫ OPCs were induced to differentiate for 2 days in OPC differentiation medium with or without IGF-1. After differentiation in both conditions, PTP␣ Ϫ/Ϫ OLs expressed less MBP than WT OLs (Fig. 9A), suggesting that PTP␣ is also required for Fyn-mediated up-regulation of MBP expression.
In support of the above in vitro results, myelinated fibers in WT and PTP␣ Ϫ/Ϫ mouse brains were examined for MBP immunoreactivity. Fewer myelinated fibers could be observed in the corpus callosum of P18 PTP␣ Ϫ/Ϫ mouse brains (Fig. 9B) and in the cortex and striatum of P10 and P18 PTP␣ Ϫ/Ϫ mouse brains (Fig. 9C). Taken together, these results demonstrate that PTP␣ is involved in regulating MBP expression during OPC differentiation and thus is required for proper myelination in the brain.

DISCUSSION
In this study, we have demonstrated that PTP␣ is required for OPC differentiation using two distinct model cell systems. The siRNA-mediated silencing of PTP␣ in the rat CG4 OPC cell line results in impaired differentiation to OLs as evidenced by the prolonged maintenance of a high population of A2B5positive population of progenitor cells, the inhibition of process extension, and the reduced expression of the maturation marker CNPase that is localized to cell bodies and processes. Oligosphere-derived OPCs isolated from PTP␣ Ϫ/Ϫ mouse embryos likewise exhibit an OL differentiation defect as deter- mined by elevated NG2-positive and reduced O4-positive populations, the appearance of few OLs with a mature morphology of multiple/branched processes, and reduced CNPase and MBP expression as compared with the cells isolated from WT mouse embryos. Furthermore, defective differentiation of PTP␣-deficient OPCs correlates with a physiological defect in CNS myelination, because relative to WT mice, PTP␣ Ϫ/Ϫ mice have a readily apparent overall reduction in myelin in forebrain sections detected by immunostaining of MBP.
Fyn is activated during OPC differentiation, and this is critical for morphological differentiation, maturation, and CNS myelination (9,10,25,59). Several upstream molecules stimulate Fyn activity in this process, including the ligand-receptor interactions of extracellular matrix components like vitronectin and fibronectin with ␤1 integrins, laminin 2 binding to ␣6␤1 integrin, and the laminin family member netrin 1 and its receptor Dcc (12)(13)(14). Other receptors such as FcR␥ (upon cross-linking of bound immunoglobulin G) (11) and the PTP CD45 (32) can also promote Fyn activation during OL differentiation/myelination. Our findings identify PTP␣ as an additional upstream activator of Fyn in OL differentiation. In CG4 cells, the differentiation-associated activation of Fyn, as measured by reduced phosphorylation of its inhibitory tyrosine residue, is reduced by siRNAmediated silencing of PTP␣ expression. Likewise, differentiating primary mouse OPCs lacking PTP␣ contain less activated Fyn than do WT mouse OPCs, irrespective of whether differentiation was induced by platelet-derived growth factor/bFGF withdrawal in the presence or absence of IGF-1. Interestingly, despite the reduced level of activated Fyn in differentiating PTP␣-null OPCs, differentiation cues still stimulated some Fyn activation in PTP␣ Ϫ/Ϫ cells. Thus, Fyn activation during this process is not exclusively regulated by PTP␣ but is also controlled by PTP␣-independent mechanisms. These could involve inhibition of the C-terminal Src kinase (Csk) that phosphorylates the negative regulatory site of SFKs, or dephosphorylation by other PTPs (32,60,61). Nevertheless, in the absence of PTP␣-activated Fyn, the Fyn activation that is mediated by these other mechanisms is insuf- FIGURE 8. PTP␣ is a regulator of Fyn-FAK signaling, but not Fyn-p190 signaling in OPCs and OLs. WT and PTP␣ Ϫ/Ϫ mouse oligospheres were dissociated and seeded on PDLO-coated dishes for 2 days in OPC proliferation medium followed by incubation for the indicated times in OPC differentiation medium without (Ϫ) or with IGF-1. A, Fyn immunoprecipitates were probed with anti-Src P-Tyr 527 to detect Fyn phospho-Tyr 528 and with anti-Fyn antibodies. The band intensity of Fyn P-Tyr 528 was normalized to that of Fyn. Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D., *, p Ͻ 0.05; **, p Ͻ 0.01, ANOVA test. B, cell lysates were immunoblotted with anti-FAK P-Tyr 576 and anti-FAK antibodies, respectively. Actin was used as a loading control. The band intensity of FAK P-Tyr 576 was normalized to that of FAK. Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D., *, p Ͻ 0.05; **, p Ͻ 0.01, ANOVA test. C, p190 immunoprecipitates were probed with anti-p190, anti-p120, or P-Tyr (4G10) antibodies. The band intensity of tyrosine-phosphorylated p190 was normalized to that of p190 protein. Data obtained from three independent experiments are shown in the graph as the mean Ϯ S.D. No significant differences were found between WT and PTP␣ Ϫ/Ϫ OPCs and OLs. D, OPCs were induced to differentiate for 2 days in the presence of IGF-1. The amounts of activated Rac1 and Cdc42 isolated from cell lysates by GST-PBD pulldowns (left panels), or of total Rac1 and Cdc42 in whole cell lysates (WCL) (right panels), were determined by immunoprobing with anti-Rac1 and anti-Cdc42 antibodies. E, as in D, but the active Rho isolated by GST-RBD pulldown (left panel), and the Rho in whole cell lysates (left panel), was detected with anti-Rho antibody. KO, knockout. ficient to promote OL differentiation. This may be because other upstream activators cannot stimulate Fyn activity to a level required for differentiation, as supported by our finding that the level of active Fyn detected in PTP␣ Ϫ/Ϫ cells after 2 days of differentiation was similar to that in undifferentiated WT OPCS, and/or because PTP␣ has other unique actions that are required for differentiation.
In accord with the notion of there being insufficient Fyn activation in differentiation-induced PTP␣-depleted or -null OPCs to effectively promote differentiation, we observed significantly impaired regulation of the Fyn downstream effectors FAK and the RhoGTPases Rac1, Cdc42, and Rho. FAK has been implicated in CNS myelination, and is proposed to regulate OL process outgrowth and/or remodeling (19,62). Fyn-mediated signaling through activation of Rac1 and Cdc42 and inhibition of Rho is important for cytoskeletal alterations involved in process extension and branching that occur during morphological differentiation of OLs (13,20,45). Diverse Fyn-FAK signaling mechanisms that regulate these RhoGTPases during OL differentiation have been described. For example, laminin stimulation induces Fyn-FAK-Rac1/Cdc42 signaling in OL differentiation (19), whereas netrin-1 stimulates recruitment of Fyn to the netrin-1 receptor Dcc that is complexed with FAK and thereby promotes the inhibition of Rho without affecting Rac1/ Cdc42 (14). The ablated Fyn-FAK to RhoGTPase signaling in PTP␣-deficient cells, comprising impaired activation of Rac1 and Cdc42 and defective inhibition of Rho, is likely a major defect contributing to their reduced process extension and maturation.
Fyn-mediated inhibition of Rho in differentiating OLs is regulated by Fyn phosphorylating and activating p190RhoGAP, and is essential for process extension and differentiation (13). The differentiation-induced phosphorylation of p190RhoGAP, as well as its interaction with Fyn and p120RasGAP are not altered by the lack of PTP␣ in differentiation-induced CG4 cells or primary mouse OLs. This indicates that although essential, p190RhoGAP activation is not sufficient to inhibit Rho or promote process extension. Furthermore, our results suggest that OL differentiation involves distinct pathways that regulate Rho: one that appears to involve Fyn-mediated p190RhoGAP activation that is PTP␣-independent, and another that requires PTP␣ and may be Fyn-dependent but is p190RhoGAP-independent. The latter may represent a distinct, possibly specific action of PTP␣, and could utilize other Fyn-regulated RhoGAPs, such as the p250RhoGAP implicated in OL differentiation (21). Pending the identification of the specific Rho regulator(s) involved it is, nonetheless, clear that it is critical for optimal PTP␣-mediated Rho inhibition during OL differentiation.
The development of OPCs into mature OLs is a complex process that requires exit from the cell cycle, expression of OLspecific genes, and extension of processes and myelin sheets. Another possible role for PTP␣ in regulating OL differentiation is that PTP␣ functions in progenitor cells to regulate survival, proliferation, or cell cycle exit. Indeed, we observe an increased number of PTP␣ Ϫ/Ϫ primary mouse cells compared with WT cells when they are grown as oligospheres, but not with cells grown as neurospheres (data not shown), suggesting that PTP␣ functions to control proliferation or survival of OL lineage cells but not cells that are at an earlier stage of development. Whether PTP␣ is necessary for these or other OPC processes that position the progenitor cells to respond appropriately to differentiation stimuli requires further investigation.
In summary, we have identified PTP␣ as a novel regulator of OL differentiation and in vivo CNS myelination. We propose that the major function of PTP␣ in promoting these processes is through activation of Fyn, in accord with the well characterized role of PTP␣ as an activator of SFKs (34) and with the overlapping phenotypes of defective forebrain myelination in PTP␣ Ϫ/Ϫ and Fyn Ϫ/Ϫ mice. This study reveals PTP␣ to be an essential, but not the sole, regulator of Fyn in differentiating OPCs. Furthermore, we demonstrate that PTP␣ is required for activation of Fyn effectors FAK, Rac1, and Cdc42 and for Rho inhibition during OL differentiation, and mediates the latter through a p190RhoGAP-independent mechanism. This suggests that upstream regulators such as PTP␣ are differentially coupled to various Fyn-effector signaling modules to provide stimulus-specific responses that determine aspects of the profound changes in gene expression and morphology that occur during OL differentiation. FIGURE 9. Decreased MBP expression in PTP␣ ؊/؊ OLs and PTP␣ ؊/؊ mouse brain. A, WT and PTP␣ Ϫ/Ϫ mouse oligospheres were dissociated and seeded on PDLO-coated dishes for 2 days in OPC proliferation medium followed by incubation for 2 days in OPC differentiation medium without (Ϫ) or with IGF-1. Cell lysates were probed for MBP and actin. The numbers at the bottom show quantified MBP/actin expression in arbitrary units. B, brain coronal sections of P18 WT and PTP␣ Ϫ/Ϫ mice were immunostained with anti-MBP antibody. There were fewer myelinated fibers in PTP␣ Ϫ/Ϫ corpus callosum (cc) compared with WT. Scale bar, 100 m. C, there are fewer myelinated fibers in the P10 and P18 cortex and striatum of PTP␣ Ϫ/Ϫ mice compared with WT mice, as demonstrated by MBP immunoreactivity. Scale bar, 20 m. KO, knockout.