Gab1 mediates PDGF signaling and is essential to oligodendrocyte differentiation and CNS myelination

Oligodendrocytes (OLs) myelinate axons and provide electrical insulation and trophic support for neurons in the central nervous system (CNS). Platelet-derived growth factor (PDGF) is critical for steady-state number and differentiation of oligodendrocyte precursor cells (OPCs), but its downstream targets are unclear. Here, we show for the first time that Gab1, an adaptor protein of receptor tyrosine kinase, is specifically expressed in OL lineage cells and is an essential effector of PDGF signaling in OPCs in mice. Gab1 is downregulated by PDGF stimulation and upregulated during OPC differentiation. Conditional deletions of Gab1 in OLs cause CNS hypomyelination by affecting OPC differentiation. Moreover, Gab1 binds to downstream GSK3β and regulated its activity, and thereby affects the nuclear accumulation of β-catenin and the expression of a number of transcription factors critical to myelination. Our work uncovers a novel downstream target of PDGF signaling, which is essential to OPC differentiation and CNS myelination.


Introduction
In the central nervous system (CNS), oligodendrocytes (OLs) myelinate axons and provide electrical insulation and trophic support for neurons (Simons and Nave, 2015). The precursors of OLs, oligodendrocyte precursor cells (OPCs), are generated from the germinal regions of neural tube (Rowitch, 2004), and then proliferate and migrate throughout the CNS before differentiating into OLs. The proliferation, migration and differentiation of OPCs are coordinated in a predictable manner by numerous extrinsic and intrinsic factors (Miller, 2002;Emery, 2010), including axonally expressed ligands (Wang et al., 1998;Charles et al., 2000;Mi et al., 2005), nuclear transcription factors (Fu et al., 2002;Arnett et al., 2004;Battiste et al., 2007;He et al., 2007;Emery et al., 2009), and mitogens, for example, platelet-derived growth factor (PDGF) (Pringle et al., 1992), fibroblast growth factor (FGF) (Furusho et al., 2017), netrins and semaphorins (Spassky et al., 2002;Tsai et al., 2006), and chemokine CXCL1 (Filipovic and Zecevic, 2008). Among these molecules, PDGF is a major in vivo mitogen for OPC development (Pringle et al., 1992;Fruttiger et al., 1999). PDGF provided by neurons and astrocytes determines the steady-state number of OPCs in the developing CNS (van Heyningen et al., 2001) and negatively regulates OPC differentiation. In cultures, the withdrawal of PDGF from the medium rapidly stops the proliferation and initiates the differentiation of OPCs (Barres et al., 1993). Correspondingly, the inactivation of PDGFa receptor (PDGFRa), which is majorly expressed in OPCs (Pringle et al., 1992), results in a reduced number of OPCs and precocious OPC differentiation (Zhu et al., 2014), whereas the activation of PDGFRa facilitates OPC division and migration (Frost et al., 2009;Tripathi et al., 2017). Although these studies demonstrate that PDGF serves as a gate controller of OPC development, it is surprising that the downstream targets of PDGF/PDGFRa signaling participating in OPC proliferation and differentiation are poorly understood.
The growth factor receptor bound 2 (Grb2)-associated binders, Gab1 and Gab2, are scaffolding proteins that act downstream of cell surface receptors, and interact with a variety of cytoplasmic signaling proteins, such as Grb2, SH2-containing protein tyrosine phosphatase 2 (Shp2), and phosphatidylinositol 3-kinase (PI3K) (Liu and Rohrschneider, 2002). It is known that Gab1 functions in lung diseases, such as allergic asthma and idiopathic pulmonary fibrosis, by interacting with cytoplasmic signaling proteins (Wang et al., 2016;Zhang et al., 2016;Guo et al., 2017). In the CNS, Gab proteins interact with growth factors, including epidermal growth factor (EGF), FGF, and PDGF, and modulate the mitotic process of neural progenitor cells (Korhonen et al., 1999;Cai et al., 2002;Mao and Lee, 2005). Interestingly, Gab1 deletion in Schwann cells interrupts neuregulin-1 (NRG-1)induced peripheral nerve myelination (Shin et al., 2014). However, the functions of Gab proteins in OL development and CNS myelination are not understood.
In the present study, we sought to investigate the functions of Gab proteins in mediating OPC differentiation and CNS myelination, given the interaction between growth factors and Gab proteins in neural progenitor cells and the importance of PDGF signaling in OL development. Our study provides compelling evidence that Gab1 is an important downstream effector of PDGF signaling during OPC differentiation and regulates CNS myelination by modulating the activity of GSK3b and bcatenin.

Distinct effects of triiodothyronine and PDGF on Gab1 expression in OPCs
To investigate the roles of Gab proteins in OL development, we first assessed their expressions in oligodendrocyte linage cells and other types of neural cells. Using purified cultures, we uncovered a number of interesting findings: i) Gab1 and Gab2 were not uniformly expressed in neural cells. Gab1 was highly expressed in astrocytes and oligodendrocyte linage cells, whereas Gab2 was highly expressed in neurons, astrocytes and microglia ( Figure 1A); ii) Gab1 was absent from cortical neurons ( Figure 1A); and iii) Gab1 expression was remarkably elevated in mature OLs compared with OPCs ( Figure 1A), accompanying by the increased expression of myelin-specific proteins, myelin basic protein (MBP) and myelin oligodendrocyte glycoprotein (MOG) ( Figure 1A and B). The western blotting was corroborated by immunocytochemical staining, showing intense Gab1 signals in cell bodies and elaborated processes of mature OLs ( Figure 1C).
It has been shown that the differentiation of cultured OPCs is promoted by triiodothyronine, but suppressed by PDGF (Barres et al., 1993). Consistently, we found that external PDGF-AA (10 ng/ ml) treatment arrested OPC differentiation, as indicated by a much reduced increase in MBP expression ( Figure 1D). Interestingly, either 1-or 3 day treatment with PDGF-AA significantly decreased Gab1 expression in OPCs ( Figure 1D). To better evaluate the opposite effects of PDGF and triiodothyronine on Gab1, we administered PDGF-AA and triiodothyronine (40 ng/ml) simultaneously in OPCs. Our results demonstrated that PDGF-AA was sufficient to reverse the Gab1 expression augmented by triiodothyronine ( Figure 1D). To confirm in vitro results, we assessed Gab1 expression in Pdgfra conditional knockout (Pdgfra f/f ;Cnp-cre) mice, in which Pdgfra was specifically ablated in differentiating OLs. Indeed, the expression of Gab1 was significantly increased in the cortex and spinal cord ( Figure 1E). While these results demonstrated a suppressive effect of PDGF signaling on Gab1 Figure 1. Gab1 expression increased during OPC differentiation but was reduced by PDGF in vitro. (A) The expressions of Gab1, Gab2, myelin-related proteins, and cell-specific marker proteins in cultured neurons, astrocytes, microglia, OPCs, and OLs. (B) The blots of Gab1 and MBP were normalized to corresponding GAPDH and their ratios in OL vs. OPC (OL/OPC) were shown as the percentage changes of OPC group. Gab1: 100 ± 7% (OPC) and 169 ± 21% (OL), p=0.014. MBP: 100 ± 10% (OPC) and 459 ± 31% (OL), p=0.00002, n = 4/group, t-test, df = t(7). (C) The immunostaining of Gab1 and Figure 1 continued on next page expression, a remaining question was how PDGF signaling negatively regulates Gab1. We measured the mRNA levels of Gab1 and Gab2 in cultured OPCs treated with PDGF-AA. Our results showed that Gab1 mRNA was reduced after 1 day treatment with PDGF-AA, whereas Gab2 mRNA was not altered ( Figure 1F), implying that PDGF signaling affects Gab1 transcription.

Gab1 is specifically regulated by PDGF signaling
As an adaptor molecule, Gab1 is suggested to interact with a number of growth factors in neural progenitor cells (Korhonen et al., 1999;Cai et al., 2002;Mao and Lee, 2005). Our next question was whether the regulation of Gab1 in OLs is controlled by other growth factors besides PDGF. Therefore, we administered EGF (10 ng/ml), insulin-like growth factor-1 (IGF-1, 10 ng/ml), NRG-1 (50 ng/ml), and PDGF (10 ng/ml) individually to OPC cultures for 1 day prior to 3-day treatment with triiodothyronine. Our results showed that only PDGF was able to decrease Gab1 expression augmented by triiodothyronine, whereas EGF, NRG-1 and IGF-1 had no effect (Figure 2A), suggesting that Gab1 is specifically regulated by PDGF.
We next compared the regulatory effects of PDGF on Gab proteins in cultured OPCs and astrocytes, since Gab1 and Gab2 appeared to be expressed more or less in both cell types ( Figure 1A). To do so, we administered PDGF or triiodothyronine to cultured OPCs and astrocytes, which could be distinguished by the specific molecular markers (CNP, PDGFRa, and glial fibrillary acidic protein, GFAP). Because the shaking could not completely exclude astrocytes from OPCs and vice versa, it was not surprising that weak bands of GFAP and PDGFRa were shown in OPCs and astrocytes, respectively ( Figure 2B). PDGF and triiodothyronine again induced opposite effects on Gab1 expression in OPCs, but not in astrocytes ( Figure 2B). In contrast, Gab2 expression was not affected by PDGF in both astrocytes and OPCs. These results suggest that the expression of Gab1 is selectively regulated by PDGF signaling in OPCs.

Deletion of Gab1 in OLs impairs myelination in CNS
Since Gab1 expression was elevated in mature OLs, we investigated its role in CNS myelination using Gab1 conditional knockout mice generated by mating Gab1 f/f mice with various cre transgenic lines, Olig1-cre, Cspg2-cre, Campk2a-cre and Nestin-cre. It is acknowledged that the conditional deletion mediated by Olig1-cre or Cspg2-cre mainly affects OL lineage, whereas conditional deletion mediated by Nestin-cre or Campk2a-cre affects neural stem cells or excitatory neurons. Western blots revealed that Gab1 was richly expressed in the cerebral cortex derived from Gab1 f/f and Gab1 f/f ;Campk2a-cre mice at P21 ( Figure 3A). However, Gab1 was absent in the cortex from Gab1 f/ f ;Nestin-cre, Gab1 f/f ;Olig1-cre and Gab1 f/f ;Cspg2-cre mice ( Figure 3A). Gab2 mutation had no additional effects on Gab1 expression in various double mutant mice ( Figure 3A). It should be noted that, compared with in vitro results that might be affected by culture purity ( Figure 1A), in vivo data from Gab1 f/f ;Olig1-cre conditional mutants provide convincing evidence showing that Gab1 is solely expressed in OLs.
Myelin-specific proteins appear in OLs prior to the onset of myelination and are continually produced by OLs during the anabolism and catabolism of myelin sheath (Sternberger et al., 1978). Myelin proteins are not only the major components of myelin but also the characteristic indicators of myelination capacity. Thus, myelin-specific proteins were examined to define the effects of Gab1knockout on myelination. Our results showed that the expression of MBP, CNP and MOG was attenuated in the cerebral cortex, the hippocampus, the cerebellum, spinal cord, corpus callosum, and optic nerves from Gab1 f/f ;Olig1-cre and Gab1 f/f ;Cspg2-cre mice at P21 compared to those from Gab1 f/+ ;Olig1-cre and Gab1 f/f controls ( Figure 4A and B). The myelin-related proteins were also examined in 3-month-old Gab1 f/+ ;Olig1-cre and Gab1 f/f ;Olig1-cre mice. Similar to the mice at P21, the expression of MBP, CNP and MOG significantly decreased in the cerebral cortex and the corpus callosum of Gab1 f/f ;Olig1-cre mice compared to Gab1 f/+ ;Olig1-cre mice at this age (Figure 4-figure supplement 1). To examine the difference in myelin components, myelin fractions were isolated and purified from the brain and myelin-specific proteins were examined (Saher et al., 2005). Similar to their total expressions, MBP, PLP and MOG in myelin fractions also remarkably decreased in Gab1 f/f ;Olig1-cre mice at P21 ( Figure 4C).
Taken together, our TEM, immunohistochemistry, and protein assay demonstrate that Gab1 deletion in OLs leads to myelin deficits in the CNS. We compared the expression of myelin-specific proteins in Gab2-knockout mice as well, and our results showed no difference in the expression of MBP, CNP and MOG proteins between control and Gab2 -/mice ( Figure 4D).
The action of Gab1 on OPC differentiation was further confirmed by Gab1-knockdown experiments using lentiviral transfection of GFP-tagged Gab1 shRNA in cultured OPCs. In order to compare the levels of maturation between naive control and Gab1 shRNA groups, OPCs were grown in the same density in two groups, as indicated by Olig2 staining ( Figure 6A). This strategy allowed us to focus on OPC differentiation without considering the proliferation. In the control group, triiodothyronine treatment yielded a large number of MBP+ cells ( Figure 6A). In contrast, Gab1 down-regulation by shRNA resulted in markedly fewer MBP+ cells after triiodothyronine treatment ( Figure 6A). Furthermore, MBP protein was dramatically decreased by Gab1 shRNA, which was verified by western blots ( Figure 6B). Hence, in vitro evidence supports the conclusion that Gab1 regulates the differentiation of OPCs.

Gab1 binds to GSK3b and modulates its activity
Although the results above show that Gab1 deficiency interrupts OPC differentiation and CNS myelination, the downstream effectors of Gab1 are not understood. It occurred to us that GSK3b and bcatenin signaling is critical for proper CNS myelination (Azim and Butt, 2011;Zhou et al., 2014), but their upstream factor is unclear. Thus, an interesting question was whether Gab1 affects GSK3b and b-catenin in OLs. We explored this possibility by assessing the binding capacity between Gab1 and GSK3b. Interestingly, in vivo Co-IP in cortical tissues showed that GSK3b was robustly precipitated by Gab1 and vice versa ( Figure 7A). Akt1, a signaling hub of growth factors in many biological processes and an upstream regulator of GSK3b, bound to Gab1 as well ( Figure 7A). Moreover, both GSK3b and PDGFRa were precipitated by Gab1 in OPCs and OLs in vitro ( Figure 7B), confirming the binding between Gab1 and GSK3b. These results indicate that Gab1 is a mediator between PDGFRa and GSK3b.
Once again, the regulation of GSK3b-S9 phosphorylation by Gab1 was investigated in cultured OPCs infected with Gab1 shRNA lenti-virus. In consistent with control group, GSK3b-S9 phosphorylation significantly increased in Gab1 shRNA group, whereas GSK3b-Y216 phosphorylation was unchanged ( Figure 8A).

Gab1 controls b-catenin nuclear accumulation and expression of transcription factors
Activated GSK3b causes the degradation of b-catenin (Aberle et al., 1997), which participates in the development of OLs (Fancy et al., 2009). Our previous work also shows that GSK3b inhibition promotes the nuclear accumulation of b-catenin in OPCs (Zhou et al., 2014). Since GSK3b activity was decreased by Gab1 ablation (Figure 7C), we investigated whether conditional knockout of Gab1 changes the nuclear accumulation of b-catenin. As shown in Figure 9A, nuclear b-catenin significantly increased in Gab1 f/f ;Olig1-cre mice P21 while its total was unchanged. These results affirm that GSK3b controls OPC differentiation through regulating the nuclear accumulation of b-catenin.

Discussion
In the present work, we revealed previously unidentified roles of Gab1: it is a downstream effector of PDGF signaling and promotes OL differentiation by modulating the activity of GSK3b and b-catenin. The functions of Gab1 in the mitotic processes of neural progenitor cells have been reported Korhonen et al., 1999;Mao and Lee, 2005), but this is the first report regarding the functions of Gab1 in OL development and CNS myelination. We showed that (i) Gab1 is specifically expressed in OLs and oppositely regulated by triiodothyronine and PDGF; (ii) Gab1 is regulated by PDGF but not other growth factors in OLs; (iii) Gab1 deletion in OLs causes hypomyelination in the CNS by reducing OPC differentiation; (iv) Gab1 binds to GSK3b and regulates its activity; and (v) Gab1 affects nuclear accumulation of b-catenin and regulates the expression of a number of factors critical to the transcription of myelin proteins. In summary, our work reveals a novel downstream target of PDGF signaling and an intrinsic cascade essential for OL development. were normalized to corresponding GAPDH and percent changes are summarized. Gab1: 100 ± 4% (NC) and 20 ± 5% (shRNA) (p=0.00005). MBP: 100 ± 3% (NC) and 37 ± 7% (shRNA) (p=0.000075). n = 4/group. t-test, df = t(7). Gray dots indicate individual data points. ***p<0.001. Figure 7. Gab1 binds to GSK3b and modulates its activity. (A) Precleared cortical lysates from wild-type mice (P21-23) were immunoprecipitated with mouse anti-Gab1 and anti-GSK3b antibodies. Immunoprecipitates were probed with antibodies to Gab1 (rabbit polyclonal antibody), GSK3b, Akt and GAPDH. The experiment was repeated three times. Rabbit IgG was used as the negative control. (B) The lysates of cultured OPCs and OLs were immunoprecipitated with mouse anti-Gab1 antibody. Immunoprecipitates were probed with antibodies to Gab1 (rabbit polyclonal antibody), GSK3b, PDGFRa, and GAPDH. The experiment was repeated three times. (C) The Figure 7 continued on next page

Expression of Gab proteins in the CNS
One important result in the present work is the expression locations of Gab1 and Gab2 in the CNS. Western blots indicated that Gab1 was absent from multiple brain regions of Gab1 f/f ;Olig1-cre and Gab1 f/f ;Cspg2-cre mice, suggesting its specific expression in OLs (Figures 3 and 4). Differently, Gab1 was found in both astrocytes and OLs in cultures ( Figure 1A). These results appear contradictory but actually not, because the shaking procedure used in the purification of astrocytes and OLs does not completely separate two types of cells. Thus, cultured astrocytes might occur along with OLs and vice versa. Indeed, we were also able to detect Olig2, a marker protein of OLs, in astrocytic cultures ( Figure 1A). The comparative western blots from Gab1 f/f ;Olig1-cre and Gab1 f/f ;Campk2acre mice also excluded the presence of Gab1 in neurons ( Figure 3B), which was confirmed by the absence of Gab1 in cultured cortical neurons ( Figure 1A).
The expression of Gab2 appeared different from that of Gab1, as it was found in neuronal, astrocytic and microglial cultures ( Figure 1A). The differential expressions of two Gab proteins imply that they play distinct roles in the CNS, for example, Gab2 is not required for myelination ( Figure 4D). It will be of interest to define the early expression of Gab1 and Gab2, two isoforms with similar structures, in neural progenitor cells. In fact, previous work has shown that Gab1 and Gab2 function differently in interacting with growth factors and modulating mitotic processes in neural progenitor cells (Korhonen et al., 1999;Cai et al., 2002;Mao and Lee, 2005).

Relationships between Gab proteins and growth factor receptors in OLs
Gab1 and Gab2 are recognized as docking/scaffolding proteins of tyrosine kinase receptors (Gu and Neel, 2003). Growth factors receptors, one group of these receptors, play important roles in multiple cellular processes, including cell-cycle progression, differentiation, metabolism, survival, adhesion, motility, and migration, some of which are mediated by interacting with Gab proteins. For example, EGF, but not IGF and PDGF, increases the tyrosine phosphorylation of Gab1 and promotes the activity of Shp2 in epidermal cells Buonato et al., 2015); Gab2 facilitates FGFinduced activation of Akt and decreases retinoic acid-induced apoptosis in embryonic stem cells (Mao and Lee, 2005). In OLs, we found that PDGF treatment induced a reduction in Gab1 expression, while other growth factors were ineffective (Figure 2). In addition, Akt phosphorylation was not changed by the conditional deletion of Gab1 in OLs (Figure 7). These results imply that Gab1 is controlled by previously unknown machinery initiated by PDGF/PDGFRa signaling in OLs. PDGF activates PI3K and mitogen-activated protein kinase (MAPK)/extracellular regulated protein kinases (ERK) in OPC survival and migration (Ebner et al., 2000;Vora et al., 2011). While PDGF/PDGFRa signaling is important for migration, proliferation and myelination of OPCs, it remains unclear how it and its downstream targets, such as ERK, induce diverse effects. In this sense, our finding that PDGF reduces the expression of Gab1 provides new insight into the functions of PDGF signaling in OLs. We speculate that ERK might mediate PDGF-induced reduction in Gab1 transcription in cultured OPCs, as ERK is known to affect many transcription factors.

Gab1 promotes OPC differentiation and CNS myelination
We found that the density of Olig2+PDGFRa+ OPCs did not differ between Gab1 f/f ;Olig1-cre and Gab1 f/+ ;Olig1-cre mice ( Figure 5). Moreover, Gab1-shRNA caused significant fewer MBP+ cells and lower MBP expression in cultured OLs than in naive control after triiodothyronine treatment (Figure 6). Therefore, we conclude that Gab1 ablation impairs OPC differentiation with little impact on the steady-state numbers of OPCs. Gab proteins provide a docking site for SH2 domain-containing signaling proteins, such as Shp2 and PI3K (Gu and Neel, 2003;Nishida and Hirano, 2003). Gab1 and Shp2 are required for normal differentiation in developing brain (Ahrendsen et al., 2018), but there are some differences between them. First, Shp2 mutants display phenotypes in controlling OPC proliferation (Zhu et al., 2010), whereas Gab1 mutation only affects OPC differentiation. Second, Shp2 mutants have thicker myelin sheaths (Ahrendsen et al., 2018) but Gab1 mutants did not ( Figure 3D). Third, Shp2 loss leads to a delay in OPC differentiation (Ehrman et al., 2014;Ahrendsen et al., 2018), while Gab1 deficiency had more sustained effects. As both Gab1 and Shp2 transduce PDGFRa signal, these differences imply that OL development is subject to dynamic and multilevel regulations.
In keeping with our previous report (Zhou et al., 2014), we here show that Gab1 activates GSK3b and subsequently affects the expression of key transcription factors during OPC differentiation. Conditional knockout of Gab1 in OLs decreased the expressions of Sox10, Olig2, Mrf, and YY1, but increased the expression of Sox6 (Figures 7 and 8). It has been shown that Sox10, Olig2, Mrf and YY1 promote OPC differentiation (Emery, 2010), whereas Sox6 arrests it (Stolt et al., 2006). Thus, Gab1 functions in OPC differentiation and myelination by acting on both positive and negative transcription factors, similar to the functions of GSK3b in OPCs (Zhou et al., 2014). Shin et al. (2014) reported that tyrosine phosphorylation of Gab1 in the sciatic nerves is up-regulated during the myelination period and conditional removal of Gab1 from Schwann cells (SCs) results in hypomyelination. The myelin defects in SC-specific conditional mutant (Gab1-SCKO) are not same as those in Gab1 f/f ;Olig1-cre mice: Gab1-SCKO mice have fewer myelinated fibers, but the thickness of the myelin sheath is reduced and axonal diameter is unchanged (Shin et al., 2014). NRG-1, which is ineffective on Gab1 expression in cultured OLs (Figure 2), is responsible for inducing Gab1 effects in SCs (Shin et al., 2014). In consistent with our results, Shin et al. (2014) also found that Akt activity is not affected by Gab1 deletion in SCs. By comparison, we speculate that mitogens may act on axonal myelination through distinct mechanisms in the central and peripheral nervous systems.
It should be noted that Gab1 deletion in OLs causes partial hypomyelination: only approximately 50% decrease in the number of either myelinated axons ( Figure 3B) or Olig2-positive OLs ( Figure 5C) was found in Gab1 f/f ;Olig1-cre mice. This phenotype suggests that, although Gab1 is important, other factors may play similar roles mediating PDGF/PDGFRa signaling during OPC development. Alternatively, not all OPCs require Gab1 for their differentiation. Indeed, Zheng et al. (2018) find that the development of a type of OPCs is independent of PDGFRa. If so, Gab1 may not regulate the development of these OPCs and thereby Gab1 deletion cannot eliminate myelin formation.

Materials and methods
All animal experiments were carried out in a strict compliance with protocols approved by the Animal Care and Use Committee at Zhejiang University School of Medicine.

Antibodies and reagents
Astrocyte culture Cortical astrocytes were cultured from embryonic SD rats (E18) according to Ji et al. (2013). Cortices were dissected and incubated with trypsin-EDTA for 20 min at 37˚C. Tissue was triturated and suspended in 10% DMEM. OLs and microglia were removed by shaking at 200 rpm for 2 hr at 37˚C. Astrocytes were plated at a uniform density of 2 Â 10 5 cells ml À1 .

Neuronal culture
Cortical neurons from E16 SD rats were cultured according to previous work (Wang et al., 2015;Zhou et al., 2018). Dissociated neurons were plated and cultured in neurobasal supplemented with B-27 and L-alanyl-glutamine. Cultures were maintained at 37˚C in a humidified incubator gassed with 95% O 2 and 5% CO 2 .

OPC culture
OPCs from SD rats (E18) were cultured according to previous work (Zhou et al., 2014;Xie et al., 2018). OPCs were collected from glial cultures by shaking for 1 hr at 200 rpm, incubating in fresh medium for 4 hr, and shaking at 250 rpm at 37˚C for 16 hr. Collected OPCs were re-plated onto poly-D-lysine-coated plates and grown in neurobasal supplemented with 2% B27. PDGF-AA (10 ng/ ml) was added to keep OPCs undifferentiated or triiodothyronine (40 ng/ml) was added for 3 days to allow differentiation.

Immunohistochemistry and immunocytochemistry
Sagittal sections (20 mm) were prepared and placed in a blocking solution (1% BSA, 0.3% Triton, 10% goat serum) for 1 hr at room temperature (RT). After washing with phosphate-buffered saline (PBS), sections were incubated sequentially with primary antibodies overnight at 4˚C and secondary antibodies for 1 hr at RT. The secondary antibodies were diluted at 1:1000. The sections were mounted using ProLong Gold Antifade Reagent (Invitrogen). Cultured cells were fixed in 4% paraformaldehyde for 15 min at RT, washed with PBS and permeabilized in 0.2% Triton X-100 for 10 min, blocked in 10% BSA for 1 hr, and labeled with primary antibodies overnight at 4˚C. Cells were then incubated with secondary antibodies (1:1000) for 1 hr at RT. All antibodies were diluted in PBS containing 1% BSA and 1% normal goat serum. The dilution ratios of primary antibodies were 1:1000 for MBP and 1:100 for Gab1, Olig2, CC1, and PDGFRa. For cell counts, four animals per genotype were used to examine cellular markers. Only the images of the midline of the corpus callosum were acquired in in vivo examination.

Black-gold staining
Brain tissue was fixed in formalin and cut at 30 mm on a freezing sliding microtome. Tissue sections were hydrated and incubated in Black-gold solution at 60˚C for 12 min. Staining was complete when the finest myelinated fibers turned to black. The sections were then rinsed in water, dehydrated in alcohols, and coverslipped with mounting medium.

Transmission electron microscopy (TEM)
TEM was performed according to previous work (Zou et al., 2011;Xie et al., 2018). Ultra-thin sections were obtained using Ultracut UCT (Leica) and stained with uranyl acetate and lead citrate. Micrographs were captured in a Philips CM100 microscope (FEI).

g-ratio analysis
TEM images containing large numbers of myelinated axons in cross-section were selected for g-ratio analysis. g-ratio analysis was performed with a threshold to identify axons and calculate their crosssectional area , from which axon diameters were calculated using the formula for the area of a circle, A = pr 2 . An experimenter blinded to genotype then measured myelin sheath thickness of each axon, and excluded any improperly detected or obliquely cut axons from analysis.

Coimmunoprecipitation (co-IP)
Cortices were lysed in RIPA buffer plus protease inhibitor. Protein concentrations were measured using BCA assays (Bio-Rad) after centrifugation at 16,000 Â g at 4˚C for 10 min. Decimus supernatant was used for input and the remainder was used for IP. Precleared preparations were incubated with mouse anti-Gab1 antibody, which was precoupled to protein A-Sepharose beads (GE Healthcare) at 2-4 mg antibody/ml of beads for 2 hr in 50 mM Tris-HCl. Proteins on the beads were extracted with 2 Â SDS sample buffer and boiled for 5 min before western blotting.

Myelin fraction isolation
According to previous work (Saher et al., 2005), crude myelin was obtained from brain homogenates by centrifuging at 25,000 rpm for 30 min and re-suspended in ice-cold water. The pellet was subjected to repeated centrifugations at 25,000 and 10,000 rpm, each for 15 min. The myelin pellets were then suspended in sucrose (0.32 and 0.85 M in order) and centrifuged at 25,000 rpm for 30 min. Myelin layers were suspended in 10 mM HEPES buffer (pH 7.4) with 1% Triton-X-100 for further experiments.

Statistics
The investigators who quantify western blots and immunostainings were blinded to the genotype. Data were analyzed using Excel 2003 (Microsoft), Igor Pro 6.0 (Wavemetrics), and SPSS 16.0 (SPSS). Sample sizes were constrained by availability of cohorts of age-matched mice and were not determined in advance. Statistical differences were determined using unpaired two-sided Student's t-test for two-group comparison or one-way ANOVA followed by Tukey's post hoc test for multiple group comparisons. For all analyses, the accepted level of significance was p<0.05. 'n' represents the number of animals or cultures tested. Data in the text and figures are presented as the mean ± SEM. The degree of freedom (df) was presented as df = t(x) for t test or df = F(v1, v2) for ANOVA.