Developmental Biology 302 (2007) 683–693www.elsevier.com/locate/ydbio

Genomes & Developmental Control

Induction of oligodendrocyte differentiation by Olig2 and Sox10: Evidencefor reciprocal interactions and dosage-dependent mechanisms

Zijing Liu a, Xuemei Hu a, Jun Cai a, Ben Liu a,b, Xiaozhong Peng b,Michael Wegner c, Mengsheng Qiu a,⁎

a Department of Anatomical Sciences and Neurobiology, School of Medicine, University of Louisville, Louisville, KY 40292, USAb The National Laboratory of Medical Molecular Biology, Institute of Basic Medical Sciences,

Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100005, Chinac Institut fuer Biochemie, Universitaet Erlangen-Nuernberg, Erlangen, Germany

Received for publication 28 June 2006; revised 2 October 2006; accepted 4 October 2006Available online 10 October 2006

Abstract

Recent studies have suggested that oligodendrocyte development is likely to be under the control of a hierarchy of lineage-specifictranscription factors. In the developing mouse spinal cord, expression of Olig2, Sox10 and Nkx2.2 is sequentially up-regulated in cells ofoligodendrocyte lineage. These transcription factors play essential roles in oligodendrocyte specification and differentiation. However, theregulatory relationship and functional interactions among these transcription factors have not been determined. In this study, we systematicallyinvestigated the function and hierarchical relationship of Olig2, Sox10 and Nkx2.2 transcription factors in the control of oligodendrocytedifferentiation. It was found that over-expression of Olig2 is sufficient to induce Sox10, Nkx2.2 and precocious oligodendrocyte differentiation inembryonic chicken spinal cord. Sox10 expression alone is also sufficient to stimulate ectopic oligodendrocyte differentiation and weakly induceNkx2.2 expression. Although genetic evidence indicated that Sox10 functions downstream of Olig2, Sox10 activity can modulate Olig2expression. In addition, we presented evidence that the control of oligodendrocyte differentiation by Olig2, Sox10 and Nkx2.2 is a dosage-dependent developmental process and can be affected by both haploinsufficiency and over-dosage.© 2006 Elsevier Inc. All rights reserved.

Keywords: Spinal cord; Gene expression; In ovo electroporation; Mutation; Myelin genes; Nkx2.2

Introduction

Recent studies have identified a large number of transcriptionfactors that are selectively expressed in cells of oligodendrocyte(OL) lineage at specific stages of cellular development (Rowitch,2004). Mounting evidence suggests that oligodendrocyte devel-opment is tightly regulated by coordinated expression andbiochemical interactions of these lineage-specific transcriptionfactors. In the developing spinal cord, early oligodendrocyteprecursor cells (OPCs or OLPs) are induced from the Olig1/2+pMN domain of the ventral neuroepithelium by Sonic hedgehog(Shh) (Trousse et al., 1995; Pringle et al., 1996;Orentas et al., 1999;Lu et al., 2000), from which motor neurons and OPCs are

⁎ Corresponding author. Fax: +502 852 6228.E-mail address: m0qiu001@louisville.edu (M. Qiu).

0012-1606/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ydbio.2006.10.007

sequentially generated (Richardson et al., 2000; Spassky et al.,2000; Miller, 2002). At slightly later stages, a small number ofOPCs are also generated from the dorsal spinal cord independent ofShh signaling (Cai et al., 2005; Fogarty et al., 2005; Vallstedt et al.,2005). It is estimated that approximately 10–15%ofmatureOLs inthe spinal cord have a dorsal origin (Richardson et al., 2006).

During gliogenesis stages, early OPCs delaminate from theventricular zone and subsequently undergo extensive migration topopulate all regions of the spinal cord. Regardless of their origins,OPC cells initially express the Olig2 basic helix–loop–helix(bHLH) transcription factor and retain its expression even after theydifferentiate into mature oligodendrocytes (Lu et al., 2000;Takebayashi et al., 2000; Zhou et al., 2000). Soon after theirbirth,OPCcells progressively gain expression of other transcriptionfactors, notably the Sox10 HMG transcriptional regulator and theNkx2.2 homeodomain transcription factor. In embryonic spinalcord, expression of Sox10 is selectively up-regulated in OPC cells

684 Z. Liu et al. / Developmental Biology 302 (2007) 683–693

before their emigration into the surrounding regions and persists inmature OLs (Stolt et al., 2002). Nkx2.2 is initially expressed in thep3 domain and its derived V3 interneurons (Briscoe et al.,1999; Liu et al., 2003). However, at later stages, Nkx2.2expression is specifically up-regulated in Olig2+ OPC cellseither immediately before they migrate out of the ventralventricular zone in chicken (Xu et al., 2000; Soula et al., 2001;Zhou et al., 2001; Fu et al., 2002) or after they disperse into thewhite matter region in rodents (Fu et al., 2002). Following OLdifferentiation in the white matter, Nkx2.2 expression isgradually down-regulated in differentiated oligodendrocytes(Xu et al., 2000).

Recent molecular and genetic studies demonstrated thatOlig2,Sox10 andNkx2.2 play distinct roles in OL lineage progression. Ithas been shown that the Olig2 transcription factor is a primarydeterminant of oligodendrocyte fate specification. In Olig2 nullmutants, OPC generation was completely abolished in the spinalcord (Lu et al., 2002; Takebayashi et al., 2002; Zhou andAnderson, 2002). However, expression of Olig2 alone inembryonic chicken spinal cord did not promote OPC generation,and co-expression of Nkx2.2 was required to promote ectopicformation of OPC cells and precocious oligodendrocyte differ-entiation (Sun et al., 2001; Zhou et al., 2001). The Nkx2.2homeodomain transcription factor plays a crucial role in theterminal differentiation of oligodendrocytes. Null mutation ofNkx2.2 resulted in a dramatic delay and reduction of oligoden-drocyte differentiation although the initial generation of Olig2+OPCs in the mutants was not compromised (Qi et al., 2001).

It has been recently shown that Sox10 also possesses importantfunctions in promoting terminal differentiation of oligodendro-cytes. In the absence of Sox10 activity, oligodendrocytedifferentiation in the spinal cord was dramatically inhibited ordelayed (Stolt et al., 2002; Lu et al., 2002). Conversely, over-expression of Sox10 protein in heterologous cells lines can inducegene expression from the myelin basic protein (MBP) promoter(Stolt et al., 2002). However, it is not known whether the in vivoexpression of Sox10 has similar effects in the developing CNS.

Although the roles of Olig2, Sox10 and Nkx2.2 inoligodendrocyte development have been extensively investi-gated, the epistatic relationship and functional interactions ofthese three transcription factors remain to be determined. In thepresent study, we investigated their functional relationship andpresented the first in vivo evidence for the reciprocal regulationof Olig2 and Sox10 expression in their induction of OL fate anddifferentiation. Sox10 and Nkx2.2 function in parallel anddownstream of Olig2 to regulate OL maturation. The control ofOL differentiation by Olig2, Sox10 and Nkx2.2 is a dosage-dependent developmental process, as suggested by the sig-nificant delay of OL maturation in all three heterozygous mice.

Results

Expression of Olig2 is sufficient for inducing Sox10 and Nkx2.2expression and precocious oligodendrocyte differentiation

Previous studies showed that Olig2 is primarily involved inthe early fate specification of OLs (Lu et al., 2002;

Takebayashi et al., 2002; Zhou and Anderson, 2002).However, Olig2 is continuously expressed in differentiatingand mature OPCs, raising the possibility that Olig2 may alsobe involved in OL differentiation and maturation. Toinvestigate this possibility, we ectopically expressed Olig2 inembryonic chicken spinal cords by in ovo electroporation andthen examined its effects on myelin gene expression and OLdifferentiation after various time periods.

In contrast to the previous reports that Olig2 expressionalone was not sufficient to induce ectopic oligodendrocytedevelopment (Sun et al., 2001; Zhou et al., 2001), we foundthat Olig2 expression was able to induce expression of theearly OPC marker Sox10 4 days after electroporation (atcE6) (Figs. 1A–B). However, ectopic expression of Nkx2.2,myelin basic protein (MBP) and myelin-associated glyco-protein (MAG) was not observed at this stage (Figs. 1C–D;data not shown). Intriguingly, following 1 more day ofOlig2 expression (at cE7), ectopic expression of Nkx2.2 andmyelin genes MBP, and MAG was then detected in theelectroporated side of the cord (Figs. 1E–J). Moreover,nearly all the induced MAG+ oligodendrocytes co-expressedOlig2 (Figs. 1I, J), suggesting a cell autonomous mechanismfor Olig2 induction of oligodendrocyte differentiation.Therefore, Olig2 expression alone in the developing spinalcord is capable of inducing both the early specification andthe maturation of OLs.

Expression of Sox10 is sufficient for inducing precociousoligodendrocyte differentiation and ectopic expression of Olig2and Nkx2.2

The induction of mature OL markers by Olig2 closelyresembled the ectopic expression of Sox10 but not of Nkx2.2(Fig. 1) implies that Sox10 is directly mediating the effects ofOlig2 on OL differentiation. To test this possibility, we nextexamined the effects of Sox10 over-expression on OLdevelopment in embryonic chicken spinal cord. Starting at3 days following Sox10 electroporation (at cE5), induction oflate differentiation markers MAG, MBP and PLP wasobserved in the electroporated side of the spinal cords (Figs.2A–E). The induction of myelin gene expression by Sox10occurred about 2 days earlier than that induced by Olig2.Surprisingly, Sox10 also induced the expression of early OPCmarkers Olig2 and Nkx2.2. Double labeling of MAG withOlig2 or Nkx2.2 revealed that some of the ectopically inducedMAG+ oligodendrocytes did not co-express Olig2 (Fig. 2D) orNkx2.2 (Fig. 2E), suggesting that Sox10 can directly inducemyelin gene expression in the absence of Olig2 and Nkx2.2expression. Moreover, Olig2 and Nkx2.2 were induced bySox10 over-expression in distinct populations of cells (Figs.2G–H). The ectopic induction of Olig2+ and Nkx2.2+ cellscould be further confirmed by the dorsal restricted expressionof Sox10 (Fig. 2I–J; data not shown), ruling out the possibilitythat the ectopic Olig2+ or Nkx2.2+ cells resulted from theSox10-induced dorsal migration of ventrally derived OPCs.Together, there results suggested that MAG, Olig2 and Nkx2.2can be induced by Sox10 independently of each other.

Fig. 1. Induction of oligodendrocyte cells by Olig2 expression in embryonic chicken spinal cords. Stage 12–13 chicken embryos were electroporated with RCABSP–Olig2 expression vector and allowed to developfurther for 4 days (A–D) or 5 days (E–J). Spinal cord sections were then subjected to in situ hybridization with Sox10 (B, F) orMBP (D, H) riboprobes, or subjected to immunofluorescent staining with anti-Olig2 (A, E),anti-Nkx2.2 (C, G), or anti-Olig2 and anti-MAG simultaneously followed by laser confocal scanning (I, J). Panel J is a higher magnification of the outlined area in panel I. Induction of Nkx2.2 by Olig2 in the white matteris represented by arrows in panel G.

685Z.Liu

etal.

/Developm

entalBiology

302(2007)

683–693

Fig. 2. Induction of both early and late oligodendrocyte markers by ectopic expression of Sox10 for 3 days in embryonic chicken spinal cords. (A–F) Induction ofoligodendrocyte markers after Sox10 expression in one experiment. Note that many MAG+ cells in the dorsal spinal cord did not co-express Olig2 or Nkx2.2 (D,E). (G, H) Independent induction of Olig2 and Nkx2.2 by Sox10 over-expression. Sox10 transgene expression was detected by Myc. (I, J) Induction of Olig2 bydorsal restricted expression of Sox10.

686 Z. Liu et al. / Developmental Biology 302 (2007) 683–693

Oligodendrocyte differentiation is dependent on the genedosage of Olig2, Sox10 and Nkx2.2 transcription factors

The induction of OL differentiation by a prolonged activationof Olig2 pathway in chicken embryos implicates that OLdevelopment may be sensitive to the dosage or level of Olig2expression in the developing spinal cord. To investigate thispossibility, we first compared the expression of mature OLmarkers in E18.5 mouse spinal cords with different dosages ofOlig2 expression. At this stage, many MBP+ and PLP+oligodendrocytes were found in the white matter region of thewild-type embryos, but none in the Olig2 homozygous mutants(Figs. 3A–C; Lu et al., 2002, Takebayashi et al., 2002). In theheterozygous animals, the number of MBP+ /PLP+ mature OLswas significantly reduced compared to that in the wild-type tissues(Figs. 3A, B, M). The number of Olig2+, Sox10+ and Nkx2.2+OPCs was similarly reduced in the heterozygous embryos (Figs.3D–M). The reduction in the expressions of Olig2, MBP, PLPand Sox10 in heterozygous embryoswas further confirmed byRT-PCR (Figs. 3N, O). The decrease in the number of OPCs and OLsin Olig2+/− embryos did not appear to result from an increasedapoptosis, as cell death detected by Caspase 3 immunostaining(Figs. S2A–C) orTUNEL staining (data not shown)was similar inall three genotypes. These observations demonstrated that theOlig2 activation of OL differentiation in the spinal cord isdependent on the gene dosage of Olig2.

The observation that Olig2 controls the expression of Sox10and Nkx2.2 in oligodendrocyte cells (Figs. 1 and 3) raises thepossibility that regulation of oligodendrocyte differentiation bythese two transcription factors could also be a dosage-dependentdevelopmental process. To investigate this possibility, wesystematically studied the expression of myelin genes inE18.5 spinal cords of Sox10+/− and Nkx2.2+/− animals (Stoltet al., 2002; Qi et al., 2001; Lu et al., 2002). At this stage,

expression of MBP and PLP was observed in the white matter ofwild-type (Fig. 4A) but not Sox10 and Nkx2.2 homozygousmutant tissues (Figs. 4E–F). As in Olig2+/− embryos, thenumber of MBP+ /PLP+ mature OLs in these two heterozygoteswas significantly smaller than that in the wild-type tissues (Figs.4B, C, G). The reduction of MBP expression was previouslynoticed in Sox10+/− embryos (Stolt et al., 2004). In addition, inthe Sox10+/− and Nkx2.2+/− double heterozygous animals, thenumber of MBP+ /PLP+ oligodendrocytes was furtherdecreased, suggesting that Sox10 and Nkx2.2 mutations havean additive or synergistic effects on OL differentiation.

Sox10 mutation leads to a reduced number of Olig2+ andNkx2.2+ OPCs in the spinal cord

The observation that Sox10 over-expression resulted in anincrease of Olig2+ cells and later Nkx2.2+ cells (Fig. 2) raisesthe possibility that Sox10 may regulate the expression of Olig2and Nkx2.2. To investigate this possibility, we first examined theexpression of Olig2 in Sox10mutant spinal cords. At the onset ofoligodendrogenesis (E12.5), Olig2+ OPCs started to migrate outof the ventricular zone (VZ) in all three genotypes. However, thenumber of Olig2+ cells in the ventral ventricular zone (pMNdomain) and the mantle zone was slightly decreased in bothSox10 heterozygous and homozygous tissues (Figs. 5A–C, J).BrdU pulse for two hours revealed a similar cycling rate ofOlig2+ cells in all three genotypes (Figs. 5A–C, K). The reduc-tion of Olig2+ OPCs in the mutant tissues remained apparent atE13.5 (Fig. S1) and E18.5 (Figs. 5D–F, L). The decrease in thenumber of OPCs in Sox10mutant embryos could not be account-ed for by differential survival, as a similar apoptosis rate wasobserved in all three genotypes (Figs. S2D–F). Together, theseresults suggested that Sox10 may play a role in up-regulating orsustaining Olig2 gene expression in oligodendrocyte cells.

Fig. 3. Delayed oligodendrocyte differentiation inOlig2 heterozygous animals. (A–L) Transverse spinal cord sections at the thoracic level from E18.5 wild-type (A, D,G, J), Olig2+/− (B, E, H, K) andOlig2−/− (C, F, I, L) embryos were subjected to in situ hybridization with MBP (A–C) riboprobe, or immunostaining with anti-Olig2(D–F) and anti-Sox10 (G–I) and anti-Nkx2.2 (J–L). Expression of these markers was significantly reduced in the heterozygous tissues compared to that in the wild-type animals. (M) Statistical analyses of MBP+ cells, Sox10+ cells, and Nkx2.2+ cells in the white matter (those in the gray matter likely represent V3 interneurons) inOlig2 heterozygous embryos relative to wild-type embryos. (N) Semi-quantitative RT-PCR from two different groups of wild-type and heterozygous P0 spinal cordtissues. (O) Quantitation of RT-PCR products in Olig2+/− tissues relative to the wild-type tissues.

687Z. Liu et al. / Developmental Biology 302 (2007) 683–693

To study the effect of Sox10 mutation on Nkx2.2 up-regulation in Olig2+ OPCs, we examined the expression anddistribution of Nkx2.2+ OPCs in E18.5 spinal cord. At thisstage, Nkx2.2+ differentiating OPCs were predominantlylocated in the white matter region of wild-type tissue (Qi et

al., 2001), whereas the Nkx2.2-immunoreactive cells in the graymatter likely represented V3 interneurons (Liu et al., 2003). Asmaller number of Nkx2.2+ differentiating OPCs was observedin the white matter of Sox10+/− spinal cord and a furtherreduction was detected in the homozygous mutants (Figs. 5G–I,

Fig. 4. Delayed oligodendrocyte differentiation in Sox10+/− and Nkx2.2+/− heterozygous animals. (A–F) Transverse spinal cord sections at the thoracic level fromE18.5 various mutant embryos were subjected to in situ hybridization with MBP riboprobe. (G) Statistical analyses of MBP+ and PLP+ oligodendrocytes in Sox10 andNkx2.2 heterozygous and homozygous embryos. The number of MBP+ and PLP+ cells was significantly reduced in single heterozygous animals, and further reducedin Sox10+/−; Nkx2.2+/− double heterozygotes.

688 Z. Liu et al. / Developmental Biology 302 (2007) 683–693

L). Double labeling experiments revealed that the percentage ofOlig2+ cells that co-expressed Nkx2.2 was markedly reduced(Figs. 6A–C), suggesting that Sox10 also modulates Nkx2.2expression in undifferentiated Olig2+ OPC cells.

Nkx2.2 mutation did not affect Olig2 and Sox10 expression inOPC cells

We next investigated whether Nkx2.2 activity can regulateOlig2 and Sox10 expression at later stages of OL development.At E18.5, a slightly larger number of Sox10+ and Olig2+ OPCcells were observed in Nkx2.2 mutant spinal cord (Figs. 6D–F),likely due to an expanded pMN domain and subsequentlyincreased production of OPCs at earlier stages (Qi et al., 2001).Double immunostaining revealed that the percentage of Olig2+cells that co-expressed Sox10 was not significantly altered by

Nkx2.2 mutation (Figs. 6D–F), indicating that Nkx2.2 does notregulate Sox10 expression in Olig2+ OPCs. Since myelin geneexpression was greatly inhibited in Nkx2.2 mutant spinal cordsat this stage (Fig. 4F), it also suggested that during normaldevelopment, expression of Olig2 and Sox10 in Nkx2.2 mutantsis not sufficient for driving OL differentiation and the Nkx2.2activity is additionally required to fully achieve the differentia-tion process.

Discussion

In this study, we presented the first in vivo evidence thatectopic expression of Olig2 or Sox10 alone was sufficient forinducing myelin gene expression in the developing spinal cord.Sox10 and Nkx2.2 appear to function in parallel to regulate OLdifferentiation process. In addition, we demonstrated that

Fig. 5. Reduced expression of Olig2 and Nkx2.2 in Sox10 mutant embryos. (A–C) E12.5 wild-type (A), Sox10+/− (B) and Sox10−/− (C) embryos were pulse-labeled with BrdU for 2 h and transverse spinal cord sections were subjected to double immunolabeling with anti-BrdU (green) and anti-Olig2 (red). (D–I) Spinalcord sections from E18.5 wild-type (D, G), Sox10+/− (E, H) and Sox10−/− (F, I) embryos were subjected to immunostaining with anti-Olig2 (D–F) or anti-Nkx2.2(G–I). Expression of these markers was significantly reduced in the heterozygous tissues compared to that in the wild-type animals. (J) Number of Olig2+ cells inE12.5 Sox10+/− and Sox10−/− embryos relative to the wild-type embryos. (K) Percentage of Olig2+ cells that incorporated BrdU at E12.5 in all three genotypes.(J) Number of Olig2+ and Nkx2.2+ cells in E18.5 Sox10+/− and Sox10−/− embryos relative to the wild-type embryos.

689Z. Liu et al. / Developmental Biology 302 (2007) 683–693

transcriptional control of OL differentiation by Olig2, Sox10and Nkx2.2 is a dosage-dependent developmental process.

Sox10 mediates the Olig2 induction of oligodendrocytedifferentiation and myelin gene expression

Olig2 plays an essential role in the fate specification of OLlineage, and null mutation of Olig2 resulted in a complete lossof cells of the oligodendrocyte lineage in the spinal cord (Lu etal., 2002; Takebayashi et al., 2002; Zhou and Anderson, 2002).However, it was also demonstrated that expression of Olig2

alone was not sufficient for induction of OL development, andco-expression of Nkx2.2 was required for terminal differentia-tion of OLs (Sun et al., 2001; Zhou et al., 2001). In the presentstudy, we demonstrated that expression of Olig2 alone for4 days was able to induce the early OPC marker Sox10, and aslightly prolonged expression of Olig2 (for 5 days) can furtherinduce expression of the late OPC marker Nkx2.2 and maturemarkers MAG, MBP and PLP (Fig. 1). The discrepancybetween the present and previous over-expression studies couldbe accounted for by both the expression level and duration ofOlig2 expression. In our studies, we employed a higher

Fig. 6. Cross-regulation of Olig2, Sox10 and Nkx2.2 expression in oligodendrocyte cells. (A–C) Sox10 mutation affects the expression of Nkx2.2 in Olig2+ OPCs.E18.5 spinal cord sections from wild-type (A) and Sox10−/− (B) embryos were simultaneously immunostained with antibodies against Olig2 and Nkx2.2, and thepercentage of Olig2+ cells that co-express Nkx2.2 in wild-type and Sox10 mutants was calculated (C). (D–F) Nkx2.2mutation does not reduce the expression of Olig2and Sox10. E18.5 spinal cord sections from wild-type (D) and Nkx2.2−/− (E) embryos were simultaneously immunostained with antibodies against Olig2 and Sox10,and the percentage of Olig2+ cells that co-expressed Sox10 in wild-type and Nkx2.2 mutants was calculated (F).

690 Z. Liu et al. / Developmental Biology 302 (2007) 683–693

concentration (2 μg/μl) of RCASBP retroviral expressionvectors to achieve a robust and sustained expression of Olig2for a longer time period. Ectopic expression of mature OLmarkers was difficult to detect in embryos electroporated with1 μg/μl or lower concentration of Olig2 expression vector (datanot shown), consistent with the concept of dosage-dependentactivation of OL differentiation by oligodendrocyte lineagetranscription factors (as discussed below).

Sox10 appears to directly mediate the effects of Olig2 onoligodendrocyte differentiation based on the present and someprevious studies. Firstly, in Olig2-transfected spinal cord, thepattern of induced MBP and PLP expression closely resembledthat of ectopic Sox10 expression (Fig. 1). Secondly, expressionof Sox10 alone in chicken spinal cord was able to induceexpression of myelin genes MAG, MBP and PLP in a muchshortened time frame (Fig. 2), as compared to the Olig2induction of myelin gene expression. Thirdly, loss of Sox10activity resulted in a severe reduction of myelin geneexpression and OL differentiation (Stolt et al., 2002). Inlight of the findings that Sox10 is the only transcription factorthat is known to be expressed in myelin-forming cells in bothCNS and PNS and its mutation affects myelin gene expressionin both tissues (Wegner and Stolt, 2005), it is possible thatSox10 may be directly responsible for stimulating myelin-specific gene expression and cell differentiation in oligoden-drocytes and Schwann cells. Previous studies demonstratedthat over-expression of Sox10 and its related SoxE memberSox9 also induced neural crest-like phenotype at the expenseof neurogenesis (Cheung and Briscoe, 2003; McKeown et al.,

2005). Thus, Sox10 functions to promote glial genesis andsuppress neurogenesis during neural development. At earlystages, Sox10 is expressed in the dorsal tips and promotes theformation and migration of neural crest precursor cells,whereas at later stages, it is expressed in OPCs (Stolt et al.,2002) and promotes the terminal differentiation of oligoden-drocytes in the CNS. Ectopic and sustained expression ofSox10 in young embryonic spinal cord appeared to recapitulateboth functions.

Sox10 and Nkx2.2 function in parallel to controloligodendrocyte differentiation

Sox10 and Nkx2.2 are both required for OL differentiationand maturation, and mutations in these two transcription factorsexhibit similar phenotypes with delayed and retarded OLdifferentiation (Qi et al., 2001; Stolt et al., 2002). In theirheterozygous embryos, the differentiation of oligodendrocytesis significantly decreased and delayed. Interestingly, there isa further reduction of oligodendrocyte differentiation in theSox10+/− and Nkx2.2+/− double heterozygous animals (Fig.4), suggesting that Nkx2.2 and Sox10 have synergistic effectson OL differentiation.

Despite the observations that Sox10 over-expression candirectly stimulate myelin gene expression (Fig. 2; Stolt et al.,2002; Wei et al., 2004), normal expression of Sox10 in OPCsdid not appear to be sufficient for OL differentiation in vivo. InE18.5 Nkx2.2 null mutants, expression of both Sox10 and Olig2was not compromised (Fig. 6) but OL differentiation was

691Z. Liu et al. / Developmental Biology 302 (2007) 683–693

severely retarded (Qi et al., 2001), implying that during normalCNS development, Sox10 activation of MBP/PLP genes isnegatively regulated by an unknown inhibitory molecule. Giventhat Nkx2.2 primarily functions as a transcriptional repressorduring neural development (Muhr et al., 2001; Zhou et al.,2001; Wei et al., 2005), it is possible that Nkx2.2 may regulatethe terminal differentiation of oligodendrocytes in an indirectmanner. Conceivably, Nkx2.2 may repress the expression oractivity of an inhibitory factor that prevents OL differentiationand myelin gene expression.

Taken into account all these observations, we propose thatoligodendrocyte differentiation is controlled by two parallelpathways, a positive regulatory pathway mediated throughSox10 and an inhibitory pathway mediated by an unknownfactor whose expression or activity can be repressed by Nkx2.2(Fig. 7). While Sox10 may directly stimulate gene expressionfrom myelin gene promoters, Nkx2.2 might repress theexpression or activity of the inhibitory factor. It is conceivablethat in OPC cells, this hypothetical inhibitory factor can bind toSox10 protein and therefore interfere with Sox10 activity.Nkx2.2 may repress its expression as a transcription repressor orits activity by physical association with this inhibitory factor. Inthis regard, Nkx2.2 protein may facilitate Olig2- and Sox10-mediated OL differentiation and function to control the timingof OL differentiation. However, a constitutive high level ofSox10 or Olig2 expression can obviate the requirement ofNkx2.2 in OL differentiation and lead to ectopic myelin geneexpression in the absence of Nkx2.2 activity (Fig. 2).

Regulation of Olig2 and Nkx2.2 expression by Sox10 in OPCcells

Although genetic studies have established that Sox10functions downstream of Olig2 (Lu et al., 2002; Takebayashiet al., 2002; Zhou and Anderson, 2002), both gain- and loss-of-function studies suggested that there is a reciprocal regulation ofOlig2 and Sox10 expression in oligodendrocyte cells. Inembryonic chicken spinal cord, over-expression of Sox10 canweakly induce Olig2 expression (Fig. 2). Conversely, mutation

Fig. 7. A hypothetical hierarchy and functional interaction of transcriptionfactors that control oligodendrocyte development. Dashes represent weakregulations.

of Sox10 gene resulted in a decreased expression of Olig2 in thespinal cord (Figs. 5 and 6). The small reduction of Olig2+ OPCswas previously unnoticed by in situ RNA hybridization (Stolt etal., 2002) possibly due to its lower detection sensitivity. Thepositive feedback on Olig2 expression by Sox10 also impliesthat Sox10may have an important function in up-regulating and/or maintaining Olig2 expression during oligodendrocytedevelopment.

Similarly, Sox10 expression can also regulate Nkx2.2expression, and this regulatory relationship may contribute tothe up-regulation of Nkx2.2 protein in differentiating OPCs(Fig. 7). In the absence of Sox10 function, Nkx2.2 expression isdramatically down-regulated but not completely inhibited (Fig.6). Conversely, over-expression of Sox10 can promote Nkx2.2expression in embryonic spinal cord independently of Olig2induction (Fig. 2), suggesting that the induction of Nkx2.2 andOlig2 by Sox10 is mediated by distinct molecular pathways ormechanisms.

Dosage-dependent regulation of oligodendrocyte developmentby transcriptional factors

The haploinsufficiency in OL differentiation was previouslyreported for the Sox10 transcriptional regulator (Stolt et al.,2004). Here we showed that the delayed maturation of OLs wasalso observed in Olig2 and Nkx2.2 heterozygous animals,indicating that the OL differentiation process is particularlysensitive to the gene dosage of these transcriptional factors. Thedosage-dependent OL development could also explain thedelayed OL differentiation phenotype in other mutants (e.g.Nkx6.1−/− and Gli2−/− mutants) in which the Olig2 expressionin the pMN domain was markedly reduced at the stage ofgliogenesis (Liu et al., 2003; Qi et al., 2003).

The developmental defects caused by haploinsufficiencyhave recently been noticed for a number of other transcriptionfactors in a diversity of developmental processes (Aoki et al.,2004; Bland et al., 2000; Kalinichenko et al., 2004). Forinstance, a decrease in Sox9 dosage by haploinsufficiency inneural crest-derived tissues resulted in delayed frontal cranialsuture closure (Sahar et al., 2005). Mice heterozygous for Pax6mutations display small eyes (Sey) phenotype with a small lens(van Raamsdonk and Tilghman, 2000). Although manytranscription factor genes display semi-dominant loss-of-function heterozygous phenotypes, the underlying mechanismfor this dosage requirement is not known. One plausibleexplanation is that a critical threshold of transcription factors isrequired for cell differentiation and that the time in developmentfor reaching this threshold level is delayed in heterozygotes.Consistent with this hypothesis, oligodendrocyte differentiationand maturation is simply delayed in animals heterozygous forOlig2, Sox10 and Nkx2.2. However, the initial delay of OLdifferentiation in these heterozygous animals is at least in partcompensated during further development, given the more orless normal oligodendrocyte phenotype in adult heterozygousmice. In further support of the threshold theory, a strong orprolonged activation of the Olig2 or Sox10 pathway is capableof promoting ectopic and precocious OL differentiation and

692 Z. Liu et al. / Developmental Biology 302 (2007) 683–693

myelin gene expression (Figs. 1 and 2). Therefore, OLdifferentiation appears to be exquisitely sensitive to the dosageof Olig2/Sox10 gene expression and can be affected by bothhaploinsufficiency and overdosage. A similar mechanism mayoperate in the control of skeletal growth by the SHOXhomeodomain transcription factor in human development(Ogata et al., 2001).

Experimental procedures

Genotyping of Olig2, Sox10 and Nkx2.2 mutant mice

The Olig2, Sox10 and Nkx2.2 homozygous null embryos were obtained bythe interbreeding of double heterozygous animals. Genomic DNA extractedfrom embryonic tissues or mouse tails was used for genotyping by Southernanalysis or by PCR. Genotyping of Olig2, Sox10 and Nkx2.2 loci was describedearlier (Lu et al., 2002; Stolt et al., 2002; Qi et al., 2001).

In situ RNA hybridization and immunofluorescent staining

Spinal cord tissues at the thoracic level were isolated from E10.5 to E18.5mouse embryos and then fixed in 4% paraformaldehyde at 4°C overnight.Following fixation, tissues were transferred to 20% sucrose in PBS overnight,embedded in OCT media and then sectioned (20 μm thickness) on a cryostat.Adjacent sections from the wild-type and mutant embryos were subsequentlysubjected to in situ hybridization (ISH) or immunofluorescent staining. ISHwas performed as described in Schaeren-Wiemers and Gerfin-Moser (1993)with minor modifications, and the detailed protocol is available upon request.Double immunofluorescent procedures were previously described in Xu et al.(2000). The dilution ratio of antibodies is as follows: anti-Olig2 (1:6000), anti-MAG (Chemicon, 1:500), anti-Myc (Sigma, 1:50), anti-Nkx2.2 (DSHB, 1:50),anti-Sox10 (1:3000; Stolt et al., 2003) and anti-caspase 3 (Cell Signaling,1:400).

Semi-quantitative RT-PCR

Total RNA was extracted from spinal cord of P0 wt and Olig2+/− mice byusing Trizol (Invitrogen). The concentration of total RNA was assessed byspectrophotometry and adjusted roughly to the same concentrations amongsamples. First-strand cDNAwas generated by using SuperScript III first-strandsynthesis system (Invitrogen). To ensure that PCR cycles ended beforesaturation, generally, the cycle number for each primer was first tried at 27cycles and then decreased or increased 2–5 cycles, depending on the intensity ofthe initial PCR products. Therefore, each group of samples were run for PCRwith at least two or three different cycle numbers to select the reaction showingclear bands with the lowest cycle number. Three different groups of sampleswere used. Primers for a housekeeping gene (GAPDH) were used as a control.Primer sequences used for analyses are as follows: GAPDH, forward 5′-ggccttccgtgttcctac-3′, reverse 5′-ctgttgctgtagccgtattca-3′; MBP, forward 5′-ccctcagagtccgacgagctt-3′, reverse 5′-tcccttgtgagccgatttat-3′; Nkx2.2, forward5′-gaccttccggacaccaacgat-3′, reverse 5′-gcgagggcagaggcgtcac-3′; Olig2, for-ward 5′-ccgaaaggtgtggatgcttat-3′, reverse 5′-tcgctcaccagtcgcttcat-3′; PLP,forward 5′-gacatgaagctctcactggtac-3′, reverse 5′-catacattctggcatcagcgc-3′;Sox10, forward 5′-aagagcaagccgcacgtc-3′, reverse 5′-acgttgccgaagtcgatgtgg-3′.

In ovo electroporation

Full-lengths mouse Olig2, Sox10 and Nkx2.2 cDNAs were subcloned intothe replication-competent retroviral vector RCASBP(B) (Morgan and Fekete,1996). For in ovo electroporation, about 1.5 μl (2 μg/μl) of expression vectorswas injected into stage 12–13 white horn chicken embryos with the aid ofPicospritzer III instrument. The injected embryos were then subjected to threeshort-pulses of electrical shock (25 V, 50 ms for each pulse) and allowed todevelop for various time periods before they were fixed in 4% PFA for geneexpression studies.

Acknowledgments

We are very grateful to Drs. Charles Stiles, David Rowitchand Richard Lu for generously providing the Olig1/Olig2mutantmouse lines and the anti-Olig2 antibody and for their criticalreading of the manuscript. This work was supported by NIH(NS37717) and by National Multiple Sclerosis Society (RG3275).

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.ydbio.2006.10.007.

References

Aoki, Y., Mori, S., Kitajima, K., Yokoyama, O., Kanamaru, H., Okada, K.,Yokota, Y., 2004. Id2 haploinsufficiency in mice leads to congenitalhydronephrosis resembling that in humans. Genes Cells 9, 1287–1296.

Bland, M.L., Jamieson, C.A., Akana, S.F., Bornstein, S.R., Eisenhofer, G.,Dallman, M.F., Ingraham, H.A., 2000. Haploinsufficiency of steroidogenicfactor-1 in mice disrupts adrenal development leading to an impaired stressresponse. Proc. Natl. Acad. Sci. U. S. A. 97, 14488–14493.

Briscoe, J., Sussel, L., Serup, P., Hartigan-O'Conner, D., Jessell, T.M.,Rubenstein, J.L., Ericson, J., 1999. Homeobox gene Nkx-2.2 andspecification of neuronal identity by graded sonic hedgehog signaling.Nature 398, 622–627.

Cai, J., Qi, Y., Hu, X., Tan, M., Liu, Z., Zhang, J., Li, Q., Sander, M., Qiu, M.,2005. Generation of oligodendrocyte precursor cells from mouse dorsalspinal cord independent of Nkx6-regulation and Shh signaling. Neuron 45,41–53.

Cheung, M., Briscoe, J., 2003. Neural crest development is regulated by thetranscription factor Sox9. Development 130, 5681–5693.

Fogarty, M., Richardson, W., Kessaris, N., 2005. A subset of oligodendrocytesgenerated from radial glia in the dorsal spinal cord. Development 132,1951–1959.

Fu, H., Qi, Y., Tan, M., Cai, J., Hirohide, T., Nakafuku, M., Richardson, W., Qiu,M., 2002. Dual origin of spinal oligodendrocyte progenitors and evidencefor the cooperative role of Olig2 and Nkx2.2 in the control ofoligodendrocyte differentiation. Development 129, 681–693.

Kalinichenko, V., Gusarova, G., Kim, I., Shin, B., Yoder, H., Clark, J.,Sapozhnikov, A., Whitsett, J., Costa, R., 2004. Foxf1 haploinsufficiencyreduces Notch-2 signaling during mouse lung development. Am. J. Physiol.:Lung Cell Mol. Physiol. 286, L521–L530.

Liu, R., Cai, J., Hu, X., Tan, M., Qi, Y., German, M., Rubenstein, J., Sander, M.,Qiu, M., 2003. Region-specific and stage-dependent regulation of Olig geneexpression and oligodendrogenesis by Nkx6.1 homeodomain transcriptionfactor. Development 130, 6221–6231.

Lu, Q., Yuk, D., Alberta, J., Zhu, Z., Pawlitsky, I., Chan, J., McMahon, A.,Stiles, C., Rowitch, D., 2000. Sonic Hedgehog-regulated oligodendrocytelineage genes encoding bHLH proteins in the mammalian central nervoussystem. Neuron 25, 317–329.

Lu, Q., Sun, T., Zhu, Z., Ma, N., Garcia, M., Stiles, C., Rowitch, D., 2002.Common developmental requirement for Olig function indicates a motorneuron/oligodendrocyte connection. Cell 109, 75–86.

McKeown, S.J., Lee, V.M., Bronner-Fraser, M., Newgreen, D.F., Farlie, P.G.,2005. Sox10 overexpression induces neural crest-like cells from alldorsoventral levels of the neural tube but inhibits differentiation. Dev.Dyn. 233, 430–444.

Miller, R.H, 2002. Regulation of oligodendrocyte development in the vertebrateCNS. Prog. Neurobiol. 67, 451–467.

Morgan, B., Fekete, D., 1996. Manipulating gene expression with replication-competent retroviruses. In: Bronner-Fraser, M.E. (Ed.), Methods in AvianEmbryology. Academic Press, San Diego, pp. 185–218.

Muhr, J., Andersson, E., Persson, M., Jessell, T., Ericson, J., 2001.

693Z. Liu et al. / Developmental Biology 302 (2007) 683–693

Groucho-mediated transcriptional repression establishes progenitor cellpattern and neuronal fate in the ventral neural tube. Cell 104,861–873.

Ogata, T., Matsuo, N., Nishimura, G., 2001. SHOX haploinsufficiency andoverdosage: impact of gonadal function status. J. Med. Genet. 38, 1–6.

Orentas, D., Hayes, J., Dyer, K., Miller, R., 1999. Sonic hedgehog signaling isrequired during the appearance of spinal cord oligodendrocyte precursors.Development 126, 2419–2429.

Pringle, N., Yu, W., Guthrie, S., Roelink, H., Lumsden, A., Peterson, A.,Richardson, W., 1996. Determination of neuroepithelial cell fate: inductionof the oligodendrocyte lineage by ventral midline cells and sonic hedgehog.Dev. Biol. 177, 30–42.

Qi, Y., Cai, J., Wu, Y., Wu, R., Lee, J., Fu, H., Rao, M., Sussel, L., Rubenstein,J., Qiu, M., 2001. Control of oligodendrocyte differentiation by the Nkx2.2homeodomain transcription factor. Development 128, 2723–2733.

Qi, Y., Tan, M., Hui, C.-C., Qiu, M., 2003. Gli2 activity is required for normalShh signaling and oligodendrocyte development. Mol. Cell. Neurosci. 23,440–450.

Richardson, W., Smith, H., Sun, T., Pringle, N., Hall, A., Woodruff, R., 2000.Oligodendrocyte lineage and the motor neuron connection. Glia 29,136–142.

Richardson, W.D., Kessaris, N., Pringle, N., 2006. Oligodendrocyte wars. Nat.Rev., Neurosci. 7, 11–18.

Rowitch, D., 2004. Glial specification in the vertebrate neural tube. Nat. Rev.,Neurosci. 5, 409–419.

Sahar, D., Longaker, M., Quarto, N., 2005. Sox9 neural crest determinant genecontrols patterning and closure of the posterior frontal cranial suture. Dev.Biol. 280, 344–361.

Schaeren-Wiemers, N., Gerfin-Moser, A., 1993. A single protocol to detecttranscripts of various types and expression levels in neural tissue andcultured cells: in situ hybridization using digoxigenin-labeled cRNA probes.Histochemistry 100, 431–440.

Soula, C., Danesin, C., Kan, P., Grob, M., Poncet, C., Cochard, P., 2001.Distinct sites of origin of oligodendrocytes and somatic motoneurons inthe chick spinal cord: oligodendrocytes arise from Nkx2.2-expressingprogenitors by a Shh-dependent mechanism. Development 128,1369–1379.

Spassky, N., Olivier, C., Perez-Villegas, E., Goujet-Zalc, C., Martinez, S.,Thomas, J., Zalc, B., 2000. Single or multiple oligodendroglial lineages: acontroversy. Glia 29, 143–148.

Stolt, C., Rehberg, S., Ader, M., Lommes, P., Riethmacher, D., Schachner, M.,Bartsch, U., Wegner, M., 2002. Terminal differentiation of myelin-formingoligodendrocytes depends on the transcription factor Sox10. Genes Dev. 16,165–170.

Stolt, C., Lommes, P., Sock, E., Chaboissier, M., Schedl, A., Wegner, M., 2003.The Sox9 transcription factor determines glial fate choice in the developingspinal cord. Genes Dev. 17, 1677–1689.

Stolt, C., Lommes, P., Friedrich, R., Wegner, M., 2004. Transcription

factors Sox8 and Sox10 perform non-equivalent roles during oligoden-drocyte development despite functional redundancy. Development 131,2349–2358.

Sun, T., Echelard, Y., Lu, R., Yuk, D., Kaing, S., Stiles, C., Rowitch, D., 2001.Olig bHLH proteins interact with homeodomain proteins to regulate cell fateacquisition in progenitors of the ventral neural tube. Curr. Biol. 11,1413–1420.

Takebayashi, H., Yoshida, S., Sugimori, M., Kosako, H., Kominami, R.,Nakafuku, M., Nabeshima, Y., 2000. Dynamic expression of basic helix–loop–helix Olig family members: implication of Olig2 in neuron andoligodendrocyte differentiation and identification of a new member, Olig3.Mech. Dev. 99, 143–148.

Takebayashi, H., Nabeshima, Y., Yoshida, S., Chisaka, O., Ikenaka, K.,Nabeshima, Y., 2002. The basic helix–loop–helix factor olig2 is essential forthe development of motoneuron and oligodendrocyte lineages. Curr. Biol.12, 1157–1163.

Trousse, F., Giess, M., Soula, C., Ghandour, S., Duprat, A., Cochard, P.,1995. Notochord and floor plate stimulate oligodendrocyte differentiationin cultures of the chick dorsal neural tube. J. Neurosci. Res. 41,552–560.

Vallstedt, A., Klos, J., Ericson, J., 2005. Multiple dorsoventral origins ofoligodendrocyte generation in the spinal cord and hindbrain. Neuron 45,55–67.

van Raamsdonk, C.D., Tilghman, S.M., 2000. Dosage requirement and allelicexpression of PAX6 during lens placode formation. Development 127,5439–5448.

Wegner, M., Stolt, C., 2005. From stem cells to neurons and glia: a Soxist's viewof neural development. Trends Neurosci. 28, 583–588.

Wei, Q., Miskimins, W.K., Miskimins, R., 2004. Sox10 acts as a tissue-specific transcription factor enhancing activation of the myelin basicprotein gene promoter by p27Kip1 and Sp1. J. Neurosci. Res. 78,796–802.

Wei, Q., Miskimins, W., Miskimins, R., 2005. Stage-specific expression ofmyelin basic protein in oligodendrocytes involves Nkx2.2-mediatedrepression that is relieved by the Sp1 transcription factor. J. Biol. Chem.280, 16284–16294.

Xu, X., Cai, J., Hui, F., Qi, Y., Modderman, G., Liu, R., Qiu, M., 2000. Selectiveexpression of Nkx2.2 transcription factor in the migratory chickenoligodendrocyte progenitor cells and implications for the embryonic originof oligodendrocytes. Mol. Cell. Neurosci. 16, 740–753.

Zhou, Q., Anderson, D., 2002. The bHLH transcription factors OLIG2 andOLIG1 couple neuronal and glial subtype specification. Cell 109, 61–73.

Zhou, Q., Wang, S., Anderson, D., 2000. Identification of a novel family ofoligodendrocyte lineage-specific basic helix–loop–helix transcriptionfactors. Neuron 25, 331–343.

Zhou, Q., Choi, G., Anderson, D., 2001. The bHLH transcription factor Olig2promotes oligodendrocyte differentiation in collaboration with Nkx2.2.Neuron 31, 791–807.