olig1 expression identifies developing oligodendrocytes in zebrafish and requires hedgehog and notch signaling

Authors


Abstract

Myelin, the isolating sheath around large diameter axons, is formed in the central nervous system (CNS) by oligodendrocytes. We isolated the zebrafish ortholog of olig1, a bHLH transcription factor, and describe the origin and development of oligodendrocytes in the zebrafish brain. Olig1:mem-eGFP transgenic animals demonstrate the highly dynamic nature of oligodendrocyte membrane processes, providing a tool for studying in vivo oligodendrocyte development. Formation of oligodendrocytes and initiation of olig1 expression are under the control of long-range hedgehog and notch signaling while maintenance of olig1 expression only depends on hedgehog. Over-expression of olig1 did not affect myelin formation in the brain and combined over-expression of olig1 and olig2 could not rescue loss of hedgehog signaling, indicating that critical factors other than olig1 and olig2 are necessary. Lastly, knockdown of Olig1 in an Olig2-sensitized background did result in defects in CNS myelination, indicating a functional overlap between Olig1 and Olig2 proteins. Developmental Dynamics 238:887–898, 2009. © 2009 Wiley-Liss, Inc.

INTRODUCTION

Myelination is the process by which isolating sheaths of membrane form around large-diameter axons. It is accomplished by glial cells wrapping membrane extensions multiple times around axons. Although functionally similar, two different glial cell types carry out this task. In the central nervous system (CNS), oligodendrocytes are responsible for myelination while in the peripheral nervous system, neural crest–derived Schwann cells are responsible for this process (Sherman and Brophy,2005). In the spinal cord and hindbrain, motor neuron/oligodendrocyte progenitor cells give rise to motor neurons in a first wave of differentiation and later on, in a second wave, to oligodendrocytes (Miller,2002). Oligodendrocyte precursor cells are generated in defined locations and then migrate and populate the CNS (de Castro and Bribian,2005). Once they reach their target for myelination, they differentiate, ensheath large-diameter axons, and form compact myelin. Transcription factors such as olig1, olig2, nkx2.2, and sox10 define oligodendrocyte precursor cells in mammals while expression of genes coding for structural proteins involved in the formation of the myelin sheath, such as myelin basic protein (mbp) or proteolipid protein (plp), are characteristic for later stages of oligodendrocyte development.

OLIG1 and OLIG2 proteins are basic helix-loop-helix (bHLH) transcription factors, critically involved in mammalian oligodendrocyte formation (Lu et al.,2000; Zhou et al.,2000). Knockout studies in the mouse demonstrated that while the Olig2 gene is essential for the earliest stages of motor neuron/oligodendrocyte progenitor formation, Olig1 is important during later stages of oligodendrocyte development (Ligon et al.,2006). In zebrafish, the dependence of olig2 expression and oligodendrocyte formation on hedgehog signaling has been demonstrated in the spinal cord (Park et al.,2002; Lewis et al.,2005). Abolishing sonic hedgehog signaling or knocking down Olig2, eliminates the formation of oligodendrocytes defined by expression of early stage marker genes like sox10 or late stage marker genes like mbp and plp (Brosamle and Halpern,2002; Park et al.,2002). Nevertheless, ectopic expression of olig2 in embryos where hedgehog signaling is blocked cannot rescue oligodendrocyte formation indicating that the hedgehog pathway is also regulating other critically important factors in addition to olig2 (Park et al.,2002). Furthermore, notch signaling plays a dual role in oligodendrocyte development. Zebrafish mutants in notch signaling such as mindbomb or deltaA/deltaD double mutants completely lack mbp expressing oligodendrocytes, while ectopic activation of the notch pathway during late development (at 60 hr post fertilization) appears to block oligodendrocyte differentiation indicating that notch signaling is important for the formation of oligodendrocyte precursor cells but also needs to be turned off for their maturation (Park and Appel,2003).

In this study, we characterize the zebrafish ortholog of Olig1. We describe the origin and development of brain oligodendrocytes based on olig1, olig2, and sox10 expression. We show that expression of olig1 is restricted to oligodendrocytes in the brain and spinal cord and is regulated by hedgehog and notch signaling. Generating a transgenic zebrafish line expressing membrane-targeted eGFP under the control of the olig1 promoter demonstrated the highly dynamic nature of membrane processes in developing oligodendrocytes. Over-expression of olig1, while specifically ablating formation of the anterior brain, had no effect on oligodendrocyte differentiation and could not, in combination with olig2, overcome the lack of oligodendrocyte formation in the absence of hedgehog signaling. Finally, knockdown of Olig1 alone had no effect on oligodendrocyte formation, but dramatically reduced mbp+ oligodendrocyte formation when combined with partial Olig2 knockdown.

RESULTS

Zebrafish Have a Single Olig1 Ortholog

In the human genome, OLIG1 is located next to OLIG2 on chromosome 21. To identify zebrafish olig1, we searched the Zv7 genome assembly of the Sanger Center (The Welcome Trust Sanger Institute) and found an ortholog of Olig1 situated next to zebrafish olig2 on chromosome 9. As in the human genome, zebrafish olig1 and olig2 are flanked by a class II cytokine receptor cluster on one side maintaining the synteny between human and zebrafish, while c21orf59, synj1, and c21orf66, neighboring on the opposite side in the human genome, are located on chromosome 10 in zebrafish (Fig. 1A). The zebrafish genome is known to encode multiple orthologs of single mammalian genes. However, searching the Zv7 genome assembly did not reveal additional copies of olig1. Next, we cloned and sequenced the zebrafish olig1 gene and aligned translated protein sequences of zebrafish, human and mouse Olig1, Olig2, and Olig3. Indeed, zebrafish Olig1 clustered together with the mammalian OLIG1 proteins, away from Olig2 and Olig3, confirming that the zebrafish gene is the true ortholog of mammalian Olig1 (Fig. 1B). Detailed protein alignments between the Olig1 amino acid sequences of zebrafish, mouse, and human demonstrated a high conservation in the helix-loop-helix domain (Fig. 1C). Particularly, the basic domain is almost identical to its mammalian counterpart, indicating that both orthologs most likely bind the same DNA consensus sequence. Interestingly, the serine/threonine-rich domain from mammalian OLIG1 and OLIG2 proteins is not present in zebrafish Olig1.

Figure 1.

Zebrafish has one ortholog of Olig1. A: The synteny between the zebrafish and the human OLIG1 gene locus is conserved. Mb, chromosomal locations in megabases. B: Phylogenetic sequence comparison between Olig1, Olig2, and Olig3 proteins from zebrafish (Danio), human (Homo), and mouse (Mus) shows that mouse and human OLIG1 are the closest relatives to zebrafish Olig1. C: Amino acid alignment of the Olig1 proteins from zebrafish, mouse, and human. The basic helix-loop-helix domain of Olig1 is highly conserved.

olig1 Is Expressed in Brain and Spinal Cord Oligodendrocytes

In zebrafish, spinal cord oligodendrocytes originate in the ventral neural tube (Brosamle and Halpern,2002; Shin et al.,2003). olig2-expressing cells in the neural tube and hindbrain define motor neuron/oligodendrocyte progenitors, cells that give rise to motor neurons and oligodendrocytes (Park et al.,2002; Shin et al.,2003). In contrast to olig2, which is already strongly expressed before 20 hr post-fertilization (hpf), olig1 expression was first detected at 36 hpf medially in the midbrain, in a region that also expressed olig2 and sox10 (Fig. 2A,C, red arrowhead). Hindbrain expression of olig1 and sox10 also started by 36 hpf (Fig. 2A, yellow arrowhead; Fig. 2D, red arrowhead). By 48 hpf, expression of olig1 increased significantly and was detected in the fore-, mid-, and hindbrain (Fig. 2A,C,D). olig1-, olig2-, and sox10-positive oligodendrocytes appear to rapidly spread and populate the brain between 36–48 hpf. The midbrain expression of all three markers has shifted laterally away from the initial medial onset (Fig. 2A,C, red arrowheads). At later stages, when mbp starts to be expressed in differentiated oligodendrocytes and Schwann cells, olig1 and olig2 remain expressed widely throughout the brain (Fig. 2B: 72, 96, and 120 hpf). When compared to olig2, olig1 is only expressed in individual cells of the anterior spinal cord by 48 hpf (Supp. Fig. 1B, which is available online).

Figure 2.

olig1 expression specifically marks differentiating oligodendrocytes. A: Onset of olig1 expression at 36 hpf in the brain was delayed compared to olig2 expression (from 22 hpf). Red arrowheads indicate midbrain oligodendrocytes characterized by olig1, olig2, and sox10 expression. Yellow arrowheads indicate earliest expression of olig1, olig2, and sox10 in hindbrain oligodendrocytes. Red brackets mark hindbrain expression of olig1 and olig2 at 48 hpf. Red asterisks indicate cerebellum expression of olig2. B: During late stages of embryogenesis, olig1- and olig2-expressing oligodendrocytes rapidly spread throughout the brain and accumulated in areas of mbp expression. C,D: Transverse sections of olig1 and olig2 in situ hybridizations at the level of the fore-midbrain (C) and hindbrain (D). Red arrowheads indicate co-expression domains of olig1 and olig2.

From the overlapping expression of olig1, olig2, and sox10, we conclude that oligodendrocytes originate in at least three distinct locations in zebrafish: medial in the midbrain, medial in the hindbrain, and ventral in the spinal cord. Upon differentiation, cells migrate away from these locations and to the prospective sites of myelination. Together, these results indicate that olig1 is expressed in differentiating oligodendrocytes prior to the expression of the late stage markers mbp and plp.

Transgenic olig1 Reporter Line Recapitulates olig1 Expression and Oligodendrocyte Plasticity

To study oligodendrocyte formation and myelination in vivo, we used 5.4 kb of the zebrafish olig1 promoter to drive membrane-targeted enhanced green fluorescent protein (mem-eGFP) expression. Stable transgenic lines (olig1:mem-eGFP) recapitulated olig1 expression. eGFP fluorescence was detected at 48 hpf in two lateral domains in the midbrain as well as medially in the hindbrain (Fig. 3A and Supp. Fig. 1A). Sections from olig1:mem-eGFP transgenic animals at the level of the midbrain and hindbrain showed a strong similarity to the olig1 in situ hybridization expression domains, confirming that eGFP fluorescence recapitulates endogenous olig1 expression (Fig. 3C,D). In the anterior spinal cord, eGFP fluorescence was detected in a few isolated cells by 48 hpf, again similar to the olig1 mRNA expression pattern (Supp. Fig. 1B). By 72 and 96 hpf, the number of eGFP-positive cells increased significantly and by 96 hpf a dense network of membrane processes extends throughout the spinal cord (Fig. 3B,F). In addition to oligodendrocytes, we also detected eGFP fluorescence in the area of the cerebellum, most likely indicating ectopic expression of the transgene compared to the olig1 mRNA expression pattern. Time-lapse microscopy on individual eGFP+ oligodendrocytes in the spinal cord showed multiple membrane processes extending and retracting, highlighting the plasticity of developing oligodendrocytes (Fig. 3E). Lastly, we blocked oligodendrocyte formation by injecting a morpholino (MO) targeted against olig2 (olig2-MO) into olig1:mem-eGFP transgenic animals and indeed observed that eGFP+ cells were mostly absent from the brain and spinal cord, but not the cerebellum, further supporting that eGFP+ cells are oligodendrocytes (Supp. Fig. 5). In summary, position, morphology, and overlap with olig1 expression indicates that indeed eGFP fluorescence in olig1:mem-eGFP fish marks developing oligodendrocytes and that this transgenic line provides a new tool to study in vivo and real time oligodendrocyte formation and myelination.

Figure 3.

The olig1 promoter is sufficient to drive membrane-targeted eGFP expression in developing oligodendrocytes. A: eGFP expression from the 5.4-kb zebrafish olig1 promoter recapitulates endogenous olig1 expression determined by in situ hybridization. Red arrowheads indicate midbrain oligodendrocytes and red bracket demarcates hindbrain oligodendrocytes. Red asterisk marks ectopic expression of eGFP in the region of the cerebellum. B: Developmental time course demonstrates increasingly complex network of oligodendrocyte processes in the spinal cord of olig1:mem-eGFP transgenic animals at 48, 72, and 96 hpf. C,D: Transverse sections of olig1 in situ hybridizations and eGFP fluorescence at the level of the fore-midbrain (C) and hindbrain (D). Red arrowheads indicate olig1 expression domains. E: Time-lapse microscopy images of an individual oligodendrocyte in the 48-hpf spinal cord demonstrating its highly motile, extending, and retracting, membrane processes (red arrows). F: Close-up of eGFP+ oligodendrocytes in the spinal cord of a 78-hpf transgenic animal.

Induction of olig1 Expression Is Regulated by Hedgehog and Notch Signaling

olig1 and olig2 were originally identified in the mouse as transcriptionally regulated by sonic hedgehog signaling (Lu et al.,2000). Therefore, we investigated whether this regulation is evolutionarily conserved. First, we performed in situ hybridization for olig1, olig2, and sox10 on 48-hpf embryos carrying mutations in two critical genes of the hedgehog pathway, smoothened and dispatched1. Smoothened is a central hedgehog regulator at the cell surface, which upon activation of the pathway initiates the signaling cascade (Alcedo et al.,1996; Varga et al.,2001). We analyzed olig1, olig2, and sox10 expression in smonv122, a new allele of slow muscle omitted (smo), carrying a premature stop codon at amino acid position 73 of zebrafish Smoothened, likely representing a null allele. Indeed, smonv122 homozygous embryos showed no olig1 expression at 48 hpf and no olig2 expression in brain or neural tube oligodendrocytes, while other olig2 expression domains such as the cerebellum (McFarland et al.,2008) or the eyes were not grossly affected (Fig. 4A). Dispatched1, in contrast to Smoothened, is not necessary for activation of the basic hedgehog signaling cascade, but functions to enable long-range signaling of secreted hedgehog proteins (Nakano et al.,2004). disp1nv108 is a new allele of chameleon, encoding a premature stop codon at amino acid position 757 of Dispatched1. As in the case of smonv122, disp1nv108 completely abolishes olig1 expression. While hindbrain and neural tube expression of olig2 is absent in disp1nv108 mutants, a small patch of cells did express olig2 in the medial forebrain at a location similar to where forebrain oligodendrocytes arise. Brain expression of sox10 is also absent in smonv122 and disp1nv108 mutants indicating that loss of long-range hedgehog signaling blocks oligodendrocyte development (Fig. 4A).

Figure 4.

Initiation of olig1 expression depends on hedgehog and notch signaling. A: Hedgehog pathway mutants showed no olig1 expression at 48 hpf and greatly reduced olig2 expression in oligodendrocytes. Also sox10 expression in the brain of hedgehog mutants is absent. DeltaA knockdown animals had little olig1 expression at 52 hpf but still strong olig2 expression and deltaD mutant embryos (dldtr233) showed some reduction of olig1, olig2, and sox10 expression. Compound DeltaA knockdown/deltaD mutant embryos showed no olig1 expression at 52 hpf and greatly reduced olig2 expression in the anterior spinal cord (yellow bracket). B: 100 μM cyclopamine treatment from 24–48 hpf abolished olig1 and sox10 expression in oligodendrocytes and greatly reduced olig2 expression as compared to control treated animals. DAPT treatment (100 μM) from 24–48 hpf significantly reduced but did not abolish olig1, olig2, or sox10 expression. Red bracket marks location of hindbrain oligodendrocytes and red asterisks mark location of cerebellum. Red arrowheads indicate location of midbrain oligodendrocytes.

Next, we tested whether notch signaling is necessary for olig1 expression. We examined 52-hpf embryos with reduced DeltaA and/or DeltaD, ligands in the notch signaling pathway. Morpholino knockdown of DeltaA resulted in greatly reduced olig1 and sox10 expression (Fig. 4A). In contrast, olig2 was still strongly expressed in areas of presumptive hindbrain and spinal cord oligodendrocytes, although in narrower and more medial locations compared to control-injected siblings. Next we looked at after eight (dldtr233), a mutant containing a point mutation in DeltaD (Holley et al.,2000). dldtr233 mutant embryos, identified by their somite abnormalities, showed some reduction of olig1 expression (Fig. 4A and Supp. Fig. 2). Injection of deltaA-MO into dldtr233 mutant embryos resulted in complete loss of olig1 expression and further reduction of olig2 and sox10 expression. In addition, olig2 expression in the neural tube, which was robust along its whole length in DeltaA knock-down animals, was absent in the anterior half of the neural tube of DeltaA/DeltaD-deficient embryos (Fig. 4A).

To exclude the possibility that early developmental patterning defects caused by interfering with hedgehog or notch signaling are responsible for the loss of olig1 expression, we used the Smoothened inhibitor cyclopamine to block hedgehog signaling and the γ-secretase inhibitor DAPT (N-(N-(3,5-Difluorophenacetyl)-L-alanyl)-S-phenylglycine t-butyl ester) to block notch signaling, after early patterning had been completed. Cyclopamine treatment from 24 to 48 hpf greatly reduced olig1, olig2, and sox10 expression in oligodendrocytes, while not interfering with olig2 expression in the cerebellum and the eyes, clearly demonstrating that olig1 induction is controlled by hedgehog signaling (Fig. 4B). Treatment with DAPT between 24–48 hpf reduced olig1 expression in the brain, but did not abolish it (Fig. 4B). Similarly, olig2 and sox10 expression in oligodendrocytes was restricted to fewer cells after DAPT treatment, but not abolished, while expression in the eyes and cerebellum was clearly reduced. Together, these results indicate that long-range hedgehog signaling, as well as notch signaling, is important for onset of olig1 expression.

Maintenance of olig1 and olig2 Expression in Oligodendrocytes

We demonstrated that initiation of olig1 expression is dependent on hedgehog and notch signaling. To investigate if maintenance also depends on these pathways, we treated embryos with cyclopamine or DAPT at 38 hpf for 12 hr (38–50 hpf). Inhibition of hedgehog signaling only slightly reduced olig1 expression in fore- and midbrain and to a lager extent in hindbrain oligodendrocytes (Fig. 5C). Spinal cord expression of olig1 and olig2 was significantly reduced and only a few cells remained positive after cyclopamine treatment (Fig. 5A). DAPT treatment between 38–50 hpf abolished spinal cord expression of olig1 and olig2 (Fig. 5A), while brain expression in oligodendrocytes seemed less affected (Fig. 5C). In contrast, olig2 expression in non-oligodendrocyte tissues, such as the eyes and the cerebellum, was reduced by late DAPT treatment (Fig. 5C). To differentiate whether reduced in situ staining reflects a decrease in transcription or a lack of expressing cells, we took advantage of the olig1:mem-eGFP transgenics. GFP protein has a half life greater then 24 hr (Corish and Tyler-Smith,1999). During 12-hr inhibitor treatments, existing eGFP fluorescence will only be moderately affected even if mRNA levels are significantly reduced. eGFP fluorescence in olig1:mem-eGFP animals, treated from 38–50 hpf with cyclopamine or DAPT, recapitulated olig1 in situ patterns at 50 hpf, confirming that blocking hedgehog or notch signaling interfered with the formation of oligodendrocytes in the spinal cord (Fig. 5B). Interestingly, washing out the inhibitors and observing the embryos after 2 more days in regular embryo medium did not rescue the overall reduction of oligodendrocytes in the spinal cord demonstrating that 38–50 hpf inhibitor treatment irreversibly ablated oligodendrocyte progenitor cell formation (Fig. 5B). We also looked at sox10 expression in the brain after inhibitor treatment and washout and similarly observed a loss of sox10+ oligodendrocytes compared to control animals (Supp. Fig. 3). Reduction of olig1 mRNA levels was also confirmed by quantitative RT-PCR (Fig. 5D).

Figure 5.

Maintenance of olig1 expression requires hedgehog but not notch signaling. A: Spinal cord expression of olig1 and olig2 was greatly reduced after 100 μM cyclopamine or DAPT treatment from 38–50 hpf. C: In contrast, expression in the brain was only mildly reduced by inhibition of hedgehog or notch signaling. While cyclopamine treatment reduced hindbrain oligodendrocyte expression of olig1 and olig2, DAPT had little effect although strongly reducing olig2 expression in the cerebellum (red asterisk) and the eye. Red arrowheads indicate midbrain oligodendrocytes and red bracket demarcates hindbrain oligodendrocytes. B: eGFP fluorescence in the spinal cord of olig1:mem-eGFP transgenics recapitulates olig1 expression after cyclopamine or DAPT treatment. Washout of inhibitors for 46 hr after treatment from 38–50 hpf did not rescue oligodendrocyte formation indicating an irreversible effect. Cyclopamine or DAPT treatment between 50–62 hpf showed no morphological difference in oligodendrocyte formation in the spinal cord. D: Quantitative RT-PCR analysis of olig1 expression in whole embryos. Results are internally controlled for ef1α expression. Data represent averages of 4 biological replicates with 2 technical replicates each (Student's t-test: *P < 0.002; **P < 0.03).

Next, we treated embryos with cyclopamine or DAPT from 50–62 hpf, after the first olig1-positive cells could be detected in the spinal cord. In those animals, the formation of oligodendrocyte in the spinal cord was not affected at 62 hpf as determined by eGFP fluorescence (Fig. 5B). Nevertheless, when we performed quantitative RT-PCR at 62 hpf, we observed that cyclopamine treatment down-regulated olig1 expression, while DAPT had no effect (Fig. 5D). These results indicate that oligodendrocytes in different parts of the CNS have different temporal requirements for hedgehog and notch signaling and that initiation of olig1 and olig2 expression in the spinal cord is dependent on hedgehog and γ-secretase/notch activity while maintenance of olig1 expression is only sensitive to interference with hedgehog signaling.

Ectopic Expression of olig1 Causes Head Defects But Not Premature Oligodendrocyte Differentiation

While zebrafish olig2 is expressed very early, olig1 expression was detected later (36 hpf) indicating that this transcription factor could rather be involved in differentiation of oligodendrocytes. To test if premature expression of olig1 is sufficient to induce oligodendrocyte differentiation and expression of late stage markers, we injected olig1 mRNA into 1- to 2-cell stage embryos. By 24 hpf, embryos showed severe head defects and anterior structures such as the eyes were missing (Fig. 6A). At 72 hpf, those defects were maintained (Fig. 6B). Large parts of the anterior head including the eyes and the forebrain are absent while the trunk of the embryos looked normal. To test if olig1 mRNA injection induces changes in myelination, we examined mbp expression at 72 hpf. While olig1+/olig2+-expressing oligodendrocytes are widely dispersed in the brain at that stage, mbp-expressing cells are only present in the spinal cord and hindbrain (Fig. 2B). Injection of olig1 mRNA had no effect on the mbp expression pattern at 72 hpf (Fig. 6C) or at 96 hpf (data not shown). This result indicates that olig1 over-expression alone is not sufficient to induce ectopic or premature formation of mbp+ oligodendrocytes in the brain.

Figure 6.

olig1 over-expression interferes with anterior head formation but does not affect mbp expression. Injection of 75 ng/μl olig1 mRNA resulted in truncation of anterior head structures at 24 hpf (A) and 72 hpf (B). In contrast, early mbp expression in the CNS at 72 hpf was not affected by olig1 over-expression (C).

Ectopic Expression of olig1 and olig2 Cannot Overcome Blockade of Hedgehog Signaling

Blocking hedgehog signaling abolishes formation of oligodendrocytes even when olig2 was ectopically expressed (Park et al.,2004). To investigate if olig1 alone or together with olig2 can overcome this defect, we injected olig1 mRNA alone or in combination with olig2 mRNA into embryos and treated them with cyclopamine (Fig. 7). Neither ectopic olig1 expression alone nor together with olig2 could rescue formation of mbp+ oligodendrocytes in the presence of cyclopamine indicating that hedgehog signaling has additional critically important roles other than inducing olig1 and olig2 expression.

Figure 7.

olig1 over-expression alone or together with olig2 does not rescue the lack of mbp+ oligodendrocytes caused by inhibition of hedgehog signaling. Embryos were either injected with 75 ng/μl olig1 mRNA alone or in combination with 150 ng/μl olig2 mRNA and treated with cyclopamine from 22 to 72 hpf. Numbers in columns indicate numbers of observed embryos while n indicates total number of analyzed embryo per treatment/injection.

Olig1 Knockdown Perturbs Generation of mbp-Expressing Oligodendrocytes in an Olig2-Sensitized Background

To investigate the function of zebrafish Olig1 in vivo, we performed gene knockdown experiments using morpholino antisense oligonucleotides. Because the coding region of zebrafish olig1 lacks introns, we designed a morpholino (MO) against the translational start site. To assure functionality of the morpholino, we first looked at mRNA expression of olig1 after olig1-MO injection since binding of morpholinos to their targeting sequences may result in degradation of the mRNA via nonsense-mediated mRNA decay. Indeed, in situ hybridization of olig1-MO-injected embryos showed reduced olig1 expression at 48 hpf (Supp. Fig. 4A). To further validate the efficacy, we fused the olig1-MO targeting sequence to eGFP and co-injected olig1-MO with in vitro transcribed olig1-eGFP reporter mRNA. The presence of the olig1 targeting site in the eGFP mRNA abolished eGFP fluorescence at 24 hpf in olig1-MO injected embryos, compared to olig1-eGFP mRNA injection alone (Supp. Fig. 4B) or olig1-eGFP mRNA and olig2-MO co-injections (data not shown). Finally, we injected olig1-MO or olig2-MO into olig1:mem-eGFP transgenics. In the transgenic animals, eGFP protein is fused to the first eight amino acids of Olig1 (including the targeting sequence of the olig1-MO). Indeed, injection of olig1-MO completely blocked translation of eGFP in the transgenic animals (Supp. Fig. 5). In contrast, injection of olig2-MO specifically interferes with the development of oligodendrocytes, not affecting eGFP expression in the cerebellum (Supp. Fig. 5) Taken together, the results indicate that the Olig1 morpholino is functional. Therefore, we analyzed myelination in olig1-MO-injected embryos. In contrast to knocking-down Olig2, Olig1 knockdown did not affect mbp+ or plp+ oligodendrocytes (Fig. 8A). Next, we co-injected olig1-MO into embryos with Olig2 sensitized background (co-injection of a sub-optimal dose of the olig2-MO, olig2-MOLOW). Injection of olig2-MOLOW alone did not show any significant effect on mbp expression. Interestingly, co-injection of olig1-MO and olig2-MOLOW resulted in a significant reduction of mbp-expressing oligodendrocytes (Fig. 8B). These results demonstrate that knocking down Olig1 function alone does not affect the formation of mbp+ oligodendrocytes but sensitizes them to reduction of Olig2 function indicating partial redundancy of Olig1 and Olig2.

Figure 8.

Olig1 knockdown alone does not interfere with oligodendrocyte development unless combined with an Olig2 sensitized background. A: Injection of 750 μM olig1 morpholino (olig-1-MO) alone had no effect on mpb or plp expression at 96 hpf while 250 μM olig2-MO abolished mbp expression in the CNS. B: In contrast, co-injection of 750 μM olig1-MO with low concentrations of olig2-MO (75 μM), each morpholino individually having no phenotype, reduced mbp+ oligodendrocytes. Numbers in columns indicate numbers of observed embryos while n indicates total number of analyzed embryo per treatment/injection.

DISCUSSION

olig1 Expression Identifies Onset of Oligodendrocyte Formation

CNS myelination is achieved by oligodendrocytes. To further characterize oligodendrocyte development in zebrafish, we cloned the ortholog of mammalian Olig1 and compared its expression to olig2 and sox10. In contrast to mouse Olig1, whose onset of expression coincides with that of Olig2 (Lu et al.,2000), zebrafish olig1 is expressed later starting around 36 hpf medial in the midbrain and expanding to the fore- and hindbrain by 48 hpf, while olig2 is already expressed by 22 hpf in the hindbrain. Since olig2 is expressed in the common motor neuron/oligodendrocyte progenitor cell, and commitment to the motor neuron lineage occurs first, the onset of olig1 expression around 36 hpf most likely indicates the time when the common progenitor cell becomes determined to the oligodendrocyte lineage also reflected by the coinciding onset of sox10 expression in those cells at 36 hpf. Interestingly, and in contrast to olig2 and sox10, we observed zebrafish olig1 expression only in oligodendrocytes and not in any other tissue, while murine olig1 is also expressed in the nasal placode, the otic and the optic vesicle during development (Lu et al.,2000). To our knowledge, olig1 is therefore the earliest oligodendrocyte specific marker in zebrafish and valuable for identification and/or isolation of early stage oligodendrocytes.

To further analyze oligodendrocyte development, we generated a stable transgenic line expressing membrane targeted eGFP from the olig1 promoter. Recapitulating olig1 expression, eGFP+ oligodendrocytes form in an anterior to posterior temporal gradient, first in the brain, then in the anterior spinal cord, and finally in the posterior spinal cord, reflecting the general anterior-to-posterior maturation sequence during zebrafish development. Membrane processes of eGFP+ oligodendrocytes form an increasingly complex network in the brain and spinal cord. To our knowledge, three other transgenic fish lines expressing GFP in oligodendrocytes have been characterized: olig2:eGFP, nkx2.2a:meGFP, and plp:eGFP (Shin et al.,2003; Yoshida and Macklin,2005; Kirby et al.,2006). In contrast to our transgenics, GFP expression in olig2:eGFP and nkx2.2a:meGFP, reflecting endogenous olig2 and nkx2.2a, is also expressed in other neuronal tissues. In addition to eGFP expression in oligodendrocytes, we also detected eGFP fluorescence in the area of the cerebellum. Since neither we nor Li et al. (2007) have reported this expression of olig1 by in situ hybridization, we conclude that it is artificially induced by our 5.4-kb promoter fragment. Given that the olig2 and olig1 loci are adjacent, it is possible that our promoter fragment contains a regulatory element for olig2 expression in the cerebellum (compare Fig. 2A with 3A). The olig1:mem-eGFP transgenic line provides a valuable new tool for studying myelination in vivo and in real time as indicated by time lapse microscopy demonstrating the high dynamic nature of processes in newly formed oligodendrocytes.

Hedgehog and Notch Signaling Requirements

Examination of olig1 expression in hedgehog pathway mutants demonstrated the requirement for long-range hedgehog signaling as short-range signaling is unaffected in dispatched1 mutants (Caspary et al.,2002). Interestingly, in mammals a second, hedgehog-independent, population of oligodendrocytes has been characterized (Miller,2005). It has also been speculated that this hedgehog-independent population of oligodendrocytes is evolutionarily younger than the hedgehog-dependent population (Richardson et al.,2006). Consistent with this hypothesis, we could not detect any expression of oligodendrocyte markers such as olig1, olig2, or mbp in hedgehog mutant embryos indicating that all zebrafish oligodendrocytes depend on hedgehog signaling. Since early obstruction of the hedgehog pathway blocks motor neuron/oligodendrocyte progenitor formation (Park et al.,2002), these data can not directly show the involvement of hedgehog signaling on olig1 expression in oligodendrocytes. To overcome this problem, we used pharmacological pathway inhibitors. Treatment of embryos with the Smoothened inhibitor cyclopamine between 24–48 hpf completely abolished olig1 expression. Since motor neuron/oligodendrocyte progenitor cells are already formed and expressing olig2 by 24 hpf, these results show that hedgehog signaling is necessary for the induction of olig1 expression. Interestingly, sox10 expression was also lost in the brain suggesting that the differentiation of oligodendrocytes was blocked by cyclopamine treatment. Later cyclopamine treatment (38–50 hpf or 50–62 hpf) had less effect on olig1 and olig2 expression in the brain. In the spinal cord, cyclopamine treatment at 38–50 hpf strongly reduced olig1 expression. By analyzing cyclopamine treatments of olig1:mem-eGFP transgenic animals, we found that the reduced expression of olig1 and olig2 is partially due to an irreversible lack of oligodendrocytes, indicating that hedgehog signaling is crucial for the formation of oligodendrocyte progenitor cells. In addition, cyclopamine treatment at 50–62 hpf does not interfere with oligodendrocyte formation, but does result in reduced expression of olig1, confirming that in addition to hedgehog's function in the formation of oligodendrocytes, hedgehog signaling also plays a role in maintaining olig1 expression.

In addition to hedgehog, we also looked at notch signaling and its role in olig1 expression. Mutations in the notch signaling pathway (mindbomb or deltaA/deltaD) lack differentiated mbp+ oligodendrocytes (Park and Appel,2003). While dldtr233 mutants showed only a slight reduction in olig1, olig2, and sox10 expression, DeltaA knock-down animals almost completely lacked olig1 expression. Interestingly, olig2 was strongly expressed in DeltaA knock-down animals, albeit in a fashion resembling a developmental delay. Hence, the loss of olig1 expression could be due to a developmental delay of oligodendrocyte differentiation rather than to a direct effect of the loss of this notch ligand on olig1 expression. Combined deficiency in DeltaA and DeltaD resulted in a complete loss of olig1 expression and a loss of olig2 expression in the anterior spinal cord. To test if developmental delay is the cause for the lack of olig1 expression, we treated embryos with the γ-secretase inhibitor DAPT to interfere with notch signaling. Treatment between 24–48 hpf clearly reduced but did not abolish olig1, olig2, or sox10 expression in the brain. Treatment from 38–50 hpf has little effect on oligodendrocytes in the brain but completely and irreversibly ablated spinal cord oligodendrocytes. In contrast to blocking hedgehog activity, blocking γ-secretase activity/notch signaling only interfered with initial formation of olig1+ oligodendrocytes but not with maintenance of olig1 expression.

Taken together, these results underline the anterior-to-posterior temporal gradient of oligodendrocyte formation. As oligodendrocytes develop in the CNS along the anterior-posterior axis, from the midbrain to the posterior spinal cord, their formation is sensitive to loss of hedgehog and notch signaling. Initial maintenance of olig1 gene expression is sensitive to loss of hedgehog signaling. During later stages of development (which occurs earlier in the brain and later in the spinal cord), olig gene expression in oligodendrocytes becomes independent from hedgehog signaling. Since oligodendrocytes are migratory at these later stages, their migration away from the source of signaling molecules (e.g., hedgehogs from the floorplate of the spinal cord) would deprive them from these molecules and alternative mechanisms have to assure olig1 and olig2 expression. This temporal-spatial gradient of development also explains the only modest effect of inhibitor treatments on olig1 expression by quantitative RT-PCR at a later stage of development since the earlier developing strong brain expression will largely be immune by 38–62 hpf to inhibitor treatments.

As recently reported by McFarland et al. (2008), olig2 expression in the cerebellum is regulated by hedgehog and wnt signaling. Our observations show that γ-secretase activity/notch signaling is also a key regulator of olig2 expression in the cerebellum. Therefore, it will be interesting to analyze this finding in more detail in future experiments.

olig1 Function in Zebrafish Myelination

Constitutive over-expression of olig1 caused a dramatic loss of anterior brain structures while the trunk of the embryos developed normally. Despite this activity, no difference in mbp expression could be observed in olig1 mRNA-injected embryos. This contrasts with retroviral ectopic olig1 expression in the mouse CNS, where it promotes oligodendrocyte formation (Lu et al.,2001). Olig1/2 bHLH transcription factors are believed to have repressor activity and potentially can form heterodimers with other E proteins (Ross et al.,2003). Therefore, the deleterious but specific effect of ubiquitous olig1 over-expression on anterior brain development could be explained either by Olig1 directly repressing or by forming heterodimers and sequestering genes critical for anterior brain development. In contrast, olig2 over-expression did not show a similar brain phenotype (data not shown), nor did over-expression of neurogenin1, another bHLH transcription factor involved in neurogenesis (Jeong et al.,2006). Li and colleagues have published that zebrafish olig1 in combination with sox10 can activate mbp expression in the spinal cord (Li et al.,2007). Consistent with our results, olig1 over-expression alone had no effect on mbp expression. We showed that olig1, olig2, and sox10 are co-expressed in the mid- and hindbrain, from around 36 hpf onwards while mbp expression can only be detected by 72–96 hpf. This clear delay between onset of olig1/sox10 and mbp expression would indicate that additional factors are necessary to initiate mbp transcription, a conclusion that is also supported by the observation of Li et al. that ectopic mbp expression appeared to be restricted to the spinal cord in olig1/sox10 over-expressing embryos.

Furthermore, we examined whether mRNA injection of olig1 alone or in combination with olig2 is sufficient to overcome the loss of oligodendrocyte formation when hedgehog signaling is pharmacologically blocked (Park et al.,2002). No rescue could be observed in cyclopamine-treated embryos by injection of olig1 or olig1/olig2 mRNA indicating that development of oligodendrocytes is strictly dependent on effects of hedgehog signaling other than expression of olig1 and olig2. Potential candidates for this task are nkx2.2a and nkx2.2b, which are controlled by sonic hedgehog signaling and have been implicated in oligodendrocyte formation (Briscoe et al.,1999; Zhou et al.,2001; Park et al.,2004).

Lastly, we examined the effect of loss of Olig1 on myelination. Consistent with Li et al. (2007), Olig1 knock-down animals exhibited normal mbp and plp expression (Li et al.,2007). Interestingly, the first published report of Olig1 knockout mice showed no developmental defect in myelination (Arnett et al.,2004). Later results demonstrated that this lack of developmental phenotype could potentially be explained by a compensatory mechanism of up-regulated Olig2 expression induced by the inserted gene targeting selection cassette into the neighboring Olig1 locus. Removal of this cassette resulted in a clear differentiation defect in Olig1 knockout oligodendrocytes (Xin et al.,2005). Therefore, we tested whether knocking down Olig1 in an Olig2 sensitized background would affect formation of mbp+ oligodendrocytes and observed that indeed this was the case. These results support that under circumstances when Olig1 function is impaired, Olig2 can functionally replace Olig1. Future experiments such as knocking Olig1 into the Olig2 locus, or vice versa, should reveal the complete extent of this interchangeability.

In conclusion, we established olig1 as a valuable marker for zebrafish oligodendrocytes. Our olig1:mem-eGFP transgenics provide a new tool for specific observation of oligodendrocytes in vivo and in real time. In addition, it will be interesting to generate transgenics expressing fluorescent protein of a different color and crossing them into other transgenic lines such as olig2:eGFP, nkx2.2a:meGFP, and plp:eGFP. This would provide the tool to observe the temporal and spatial overlay of the different reporter lines and to differentiate possible subpopulations of oligodendrocytes. Recently, the importance of neurotransmitters for myelin formation and maintenance has become a focus of studies (Karadottir and Attwell,2007). Oligodendrocyte-specific promoters offer the possibility to express controllable neurotransmitter receptors exclusively in those cells and not in neurons providing tools to study in vivo the role of neurotransmitters for oligodendrocyte formation and maintenance (Banghart et al.,2006).

EXPERIMENTAL METHODS

Sequence Information

Genomic sequence data were obtained from the Zebrafish Sequencing Group and the Human Chromosome 21 Sequencing Group at the Welcome Trust Sanger Institute (ftp://ftp.ensembl.org/pub/assembly/zebrafish/Zv7release/ and http://www.sanger.ac.uk/HGP/Chr21). The olig1 gene sequence is also annotated as ENSDARG00000040948. Protein alignments were performed using Vector NTI 9.1 (Invitrogen). For cloning of the full-length coding sequence of olig1, we performed RT PCR on 3-day-old embryos using the following primers: 5′-GCGGAATTCGCCACCATGCAGGCTGTGTCTGGTGTT-3′ and 5′-GCGTCTAGACGGAAACCTCATCCTCAAAA-3′. The following primers were used for cloning of in situ probes by RT-PCR: mbp: 5′-GCGAATCAGCAGGTTCTTCGGAGGA-3′ and 5′-GCGGGCATACAATCCAAGCCACA-3′; olig1: 5′-GCGATCAGGAGGACCAGGCTCCAAT-3′ and 5′-GCGGGCAAGTTTGGTCCTGGAGA-3′; olig2: 5′-GCGATGGACTCTGACACGAGCCGA-3′ and 5′-GCGGGCTGAGGAAGGTTTGCCAT-3′; plp: 5′-GCGTCCTCTATGGACTGTTGCTGCTG-3′ and 5′-GCGACAATCACACACAGGAGGACCAA-3′; sox10: 5′-GCGACCTACCGAAGTCACCTGTGG-3′ and 5′-GCGGTTTGTGTCGATTGTGGTGC-3′. The following primers were used for quantitative RT-PCR: olig1: 5′-CGGACTGAAAGTTTGAAGAATGC-3′ and 5′-TCCTGTTACCCGTACCATTCTTG-3′; ef1α: 5′-TCACTGGTACTTCTCAGGCTGACTG-3′ and 5′-TCCCAGGGTGAAAGCCAGGA-3′. The following morpholino oligonucleotide sequences were used for gene knockdowns: olig1: 5′-ATCTAACACCAGACACAGCCTGCAT-3′; olig2: 5′-CGTTCAGTGCGCTCTCAGCTTCTCG-3′ (Park et al.,2002); deltaA: 5′-CGCCGACTGATTCATTGGTGGAGAC-3′ (Amoyel et al.,2005). To generate the olig1:mem-eGFP transgenic construct, we isolated 5.4 kb of zebrafish olig1 promoter sequence and cloned it upstream of membrane-targeted eGFP, generating a fusion protein of the first eight amino acids of the Olig1 protein and mem-eGFP. The following primers were used to amplify the olig1 promoter sequence introducing ApaI restrictions sites for subcloning: 5′-GCGGGCCCTTGTGTATGAAGCCTCTTGGCACA-3′ and 5′-GCGGGCCCATCTAACACCAGACACAGCCTG-3′. To target eGFP to the membrane, we fused the membrane targeting domain of zebrafish Neuromodulin/GAP43 (CTGTGCTGTATGAGAAGAACCAAACAGGTTGAAAAGAATGATGAGGACCAAAAGATCGAT) in-frame to the N terminus of eGFP. The olig1-eGFP mRNA reporter construct was generated by fusing the targeting sequence of the olig1 morpholino in frame to the N-terminus of eGFP.

Fish Care, Injection, In Situ Hybridization, and Compound Treatment

Zebrafish were maintained using standard methods (Westerfield,1995; Nusslein-Volhard and Dahm,2002). The following mutants were used: smonv122 (this study), disp1nv108 (this study), dldtr233 (Holley et al.,2000). Transgenic olig1:mem-eGFP lines were generating by injecting purified plasmid DNA into single-cell-stage embryos. Potential founders were raised to adulthood and their progeny was screened for GFP fluorescence. For mRNA injections, linearized plasmid containing olig1, olig2, or olig1-eGFP coding sequences were used as templates for in vitro transcription using mMESSAGE mMACHINE (Ambion). Diluted, capped mRNAs were injected underneath blastomeres of 1–2-cell-stage embryos at the indicated concentrations. In all co- versus single-injection experiments, the total amount of injected material (mRNA or MO) was kept constant using control RNA or MO. For in situ hybridization, DIG-labeled antisense probes were generated and used according to standard protocols (in situ hybridization using a single digoxigenin-labelled probe, p41 in Nusslein-Volhard and Dahm,2002). For compound treatments, embryos in their chorions were incubated in E3 media with 100 μM cyclopamine (Sigma Cat. No. C4116) or 100 μM DAPT (Sigma Cat. No. D5942) for the indicated time intervals.

Sectioning and Microscopy

Whole mount in situ stained embryos were embedded in JB4 and sectioned using a glass knife. For sections of olig1:mem-eGFP embryos, animals were fixed in 4% PFA overnight at 4°C and embedded in 4% low melt agarose. Sections were generated using a vibratome. For time-lapse microscopy, live transgenic embryos were anesthetized in tricaine and embedded in 1% low melt agarose in E3. After embedding, the agarose containing the immobilized embryo was overlaid with regular E3, diluting out the anesthetic, and fluorescent images were acquired at the indicated times using a water immersion objective.

Acknowledgements

We thank Daniel Curtis, Rajeev Sivasankaran, and Elizabeth Wiellette for suggestions and comments on the manuscript, Wanhong Dai, Humberto Urquiza, and Gerlinde Wussler for excellent fish care, and May Shawi, Stephanie Wiessner, and Emma Oster for technical support. Mutants were kindly provided by Scott Holley, Brigitte Walderich, and Christiane Nüsslein-Volhard.

Ancillary