Nuclear export of OLIG2 in neural stem cells is essential for ciliary neurotrophic factor–induced astrocyte differentiation

OLIG2 is a basic helix–loop–helix (bHLH) transcription factor required for oligodendrocyte development (Lu et al., 2000; Takebayashi et al., 2000; Zhou et al., 2000, 2001). OLIG2 also blocks astrocyte differentiation, by inhibiting complex formation between STAT3 and the transcription coactivator p300 (Lu et al., 2002; Takebayashi et al., 2002; Zhou and Anderson 2002; Gabay et al., 2003; Fukuda et al., 2004). Despite the crucial role of OLIG2 in glial fate decision, little is known about how OLIG2 itself is regulated during glial development. Previous findings have suggested that when neural stem cells (NSCs) begin to differentiate into glial–fibrillary–acidic–protein (GFAP)-positive astrocytes in culture, OLIG2 disappears from the nucleus and is sometimes detected in the cytoplasm (Fukuda et al., 2004), raising the possibility that the translocation of OLIG2 from the nucleus to cytoplasm might be crucial for astrocyte differentiation.

Here we show that OLIG2 is detected in the cytoplasm of GFAP-positive astrocytes in the subventricular zone (SVZ) of adult mouse brain, which have been shown to act as NSCs that give rise to both neurons and glia. We demonstrate that Leptomycin B (LMB), which is a specific inhibitor of CRM1-dependent nuclear export (Fornerod et al., 1997; Yoshida and Horinouchi, 1999), blocks both the export of OLIG2 from the nucleus and astrocyte differentiation induced by ciliary neurotrophic factor (CNTF). Moreover, we show that an OLIG2 mutant lacking a putative CRM1 binding site is deficient in both CNTF-induced nuclear export and astrocyte differentiation. Finally, we show that AKT, which is activated by CNTF stimulation, phosphorylates OLIG2 in vitro and that an OLIG2 mutant lacking an AKT phosphorylation site is also deficient in both CNTF-induced nuclear export and astrocyte differentiation. These findings suggest that AKT-dependent nuclear export of OLIG2 plays a crucial part in astrocyte differentiation.

We first investigated whether OLIG2 localized in the cytoplasm of astrocyte in vivo. We immunolabeled sections of either newborn or adult mouse brain for both OLIG2 and a neural marker—SSEA1 for NSCs, neurofilament (NF) for neurons, or GFAP for astrocytes. In the SVZ of the adult brain, 44% of the cells with OLIG2 in the nucleus stained for SSEA1, and 38% stained for NF (Fig. 1, a and b). None of GFAP-positive cells had OLIG2 in the nucleus, although 20% of them had OLIG2 in the cytoplasm (Fig. 1 c). In the ventricular zone (VZ) of the newborn brain, we also found that 57% of the cells with OLIG2 in the nucleus stained for SSEA1, and 41% stained for NF; none of GFAP-positive cells had OLIG2 in the nucleus, although 4% of them had OLIG2 in the cytoplasm (unpublished data). It seems, therefore, that, in the VZ and SVZ, OLIG2 is mainly in the nucleus of both NSCs and neurons, by contrast, the few astrocytes contained OLIG2 exclusively in their cytoplasm.

We investigated further the relationship between OLIG2 localization and astrocyte differentiation in culture. We prepared NSCs from embryonic mouse telencephalon and expanded them in both basic FGF (bFGF) and EGF for 2 wk. Over 90% of the NSCs were OLIG2-positive by 2 wk (Fukuda et al., 2004). To induce astrocyte differentiation, we withdrew both the bFGF and EGF and added CNTF (Fukuda et al., 2004). We used LMB to block active export from the nucleus. After 2 d in CNTF, over 50% of the cells differentiated into GFAP-positive astrocytes; none of the GFAP-positive cells had OLIG2 in their nucleus (Fig. 2 a), but 10% had OLIG2 in their cytoplasm (Fig. 2, a and b). By contrast, the GFAP-negative cells still had OLIG2 in their nucleus (Fig. 2 a), and these cells stained for Nestin that is a marker for NSCs (Fig. 2 d). After 4 d, 90% of the cells were GFAP-positive and OLIG2-negative (not shown). After 2 d in CNTF plus LMB, compared with the NSCs cultured in CNTF alone for 2 d, NSC proliferation had significantly decreased, the cell bodies had become larger, over 95% of the cells were GFAP-negative and had OLIG2 in the nucleus (Fig. 2 c). These findings suggest that CNTF-induced astrocyte differentiation is associated with CRM1-dependent translocation of OLIG2 from the nucleus into the cytoplasm.

Because it was shown previously that STAT3 accumulates in the nucleus when NSCs are induced by CNTF/LIF to express GFAP (Bonni et al., 1997; Nakashima et al., 1999), we examined STAT3 localization in NSCs cultured in either CNTF alone or CNTF plus LMB. After 2 d in CNTF alone, STAT3 was mainly in the nucleus, and >50% of the cells stained for GFAP, and many looked like astrocytes (Fig. 2 e). By contrast, after 2 d in CNTF plus LMB, STAT3 strongly accumulated in the nucleus, but none of the cells were GFAP positive and acquired an astrocytic morphology (Fig. 2 f). Thus, nuclear accumulation of STAT3 is apparently not enough to induce astrocyte differentiation; nuclear export of transcriptional repressors such as OLIG2 may also be required.

Because it seemed likely that OLIG2 export from the nucleus would depend on the nuclear export signal receptor CRM1, we looked for CRM1-binding sites in OLIG2. We found one putative nuclear export signal sequence (NES; 146-LSKIATLLL-154). To examine whether this sequence was a functional NES, we replaced the leucines at 152 and 154 to alanines and examined the localization of the mutant protein in Cos7 cells. When FLAG-tagged wild-type OLIG2 (OLIG2-w) was overexpressed in Cos7 cells, it was diffusely distributed (Fig. 3 a), although in the presence of LMB it accumulated in the nucleus (Fig. 3 b), suggesting that OLIG2 is normally exported from the nucleus by CRM1 in Cos7 cells. By contrast, when FLAG-tagged mutant OLIG2 (OLIG2–NES) was expressed in Cos7 cells, it accumulated in the nucleus even in the absence of LMB (Fig. 3 c), confirming that 146-LSKIATLLL-154 is the relevant CRM1-dependent export signal in OLIG2.

Because it has been shown that enforced expression of OLIG2-w does not block CNTF-induced astrocyte differentiation completely (Fukuda et al., 2004), we examined whether OLIG2–NES would be more effective at blocking CNTF-induced astrocyte differentiation. We overexpressed OLIG2–NES in NSCs, cultured them in the presence of CNTF for 2 d, and then immunolabeled them for GFAP. Whereas >50% of the cells transfected with a control vector and 35% of the cells transfected with OLIG2-w were labeled for GFAP (Fig. 3, d and f; and data not depicted), <10% of the cells transfected with OLIG2–NES were labeled for GFAP (Fig. 3, e and f). Together with the previous findings (Fukuda et al., 2004), these results suggest that CRM1-dependent OLIG2 translocation from the nucleus is crucial for CNTF-induced astrocyte differentiation.

It has been shown that the nuclear corepressor N-CoR, which inhibits astrocyte differentiation, is also exported from the nucleus, following CNTF stimulation (Hermanson et al., 2002), raising the possibility that OLIG2 and N-CoR may be physically associated. When we immunolabeled cultured NSCs with anti N-CoR antibodies, ∼10% of the cells showed nuclear staining (Fig. 4 a). CNTF stimulation caused the N-CoR to translocate to the cytoplasm (Fig. 4 b), as reported previously (Hermanson et al., 2002). However, we detected OLIG2 in the nucleus of >90% of the cultured NSCs (Fukuda et al., 2004), making it unlikely that OLIG2 is associated with N-CoR in the cells. To test this conclusion further, we transfected either a control vector expressing GFP or a vector expressing OLIG2–NES, cultured them in CNTF for 2 d, and then immunolabeled them for N-CoR and either GFP or FLAG. In the cells transfected with either the control vector or the OLIG2–NES vector, N-CoR was detected only in the cytoplasm (Fig. 4, c and d), whereas OLIG2–NES remained in the nucleus (Fig. 4 d). It seems, therefore, that OLIG2 and N-CoR independently regulate astrocyte differentiation.

It was shown previously that CNTF can activate PI3K, as well as STAT3 (Alonzi et al., 2001). Moreover, the PI3K inhibitor LY294002 blocks both CNTF-induced astrocyte differentiation and CNTF-induced N-CoR translocation from the nucleus (Hermanson et al., 2002). As we could not detect N-CoR in the nucleus of most of our cultured NSCs (Fig. 4 a), we tested whether PI3K signaling also regulates OLIG2 localization. We cultured NSCs in the presence of either CNTF alone or CNTF plus LY294002 for 3 d and then immunolabeled the cells for both OLIG2 and an astrocyte marker (either GFAP or S100β). In CNTF alone, >70% of NSCs differentiated into GFAP-positive and S100β-positive astrocytes, which had lost OLIG2 from the nucleus (Fig. 5, a and c; and

We are grateful to M. Raff for a critical reading of the manuscript. H. Takebayashi for anti-OLIG2 antibodies, and Y. Hooks for mouse brain sections.

T. Setoguchi was supported by the Yamada Science Foundation. T. Kondo was supported by Merck, Sharp, and Dohme.