Zebrafish embryos used in this study were wild-type AB line, Tg (olig2:egfp) (Shin et al., 2003), Tg(sox10:egfp) (Carney et al., 2006) and raised at 28.5℃ in egg water or embryo medium (15 mM NaCl, 0.5 mM KCl, 1 mM CaCl₂, 1 mM MgSO₄, 0.15 mM KH₂PO₄, 0.05 mM NH₂PO₄, 0.7 mM NaHCO₃, EM). Embryos were staged according to days post fertilization (dpf) and morphological criteria (Kimmel et al., 1995).

Embryos were dechorionated and incubated in EM with histone-deacetylase inhibitor Trichostatin A (Tsai et al., 2006) at a concentration of 100 ng/ml.

Embryos were raised until appropriate stage and anesthetized with ethyl 3-aminobenzoate methanesulfonic acid on ice. A thousand embryos were chopped into several pieces and transferred into 1.5 ml eppendorf tube. Collected samples were rinsed with calcium-, magnesium-free Ringers solution for 20 min, then washed with Dulbecco's phosphate buffered saline (D-PBS) (GIBCO) three times. Tissue was digested with 100 µl of D-PBS containing 35 µl of Liberase Blendzyme 3, followed by incubation at 29℃ for 15 min with occasional mixing. After incubation, 1 ml of trypsin (GIBCO) was added and incubated at 29℃ for 15 min until complete dissociation. To inactivate trypsin, we added 2 ml of D-PBS containing 10% fetal bovine serum (FBS) and filtered the solution through a cell strainer (BD Falcon) to remove undissociated cell clusters. We collected dissociated cells by centrifugation at 300 g at 4℃ for 5 min and rinsed twice with D-PBS containing 10% FBS. Dissociated cells were collected by centrifugation and resuspended in PBS containing 1 mM EDTA, 25 mM HEPES (pH 7.0) and 1% FBS. Cell suspensions were separated into GFP+ and GFP- cells by BD FACSAria II (Becton Dickinson) performed at room temperature under sterile conditions. Isolated GFP+ and GFP- cells were collected in D-PBS/10% FBS pre-coated tubes with RNase inhibitor.

Isolated cells were collected by low-speed centrifugation and homogenized in TRIzol solution (Invitrogen) subsequently. Total RNA extraction was performed as described previously (Peterson and Freeman, 2009). Isolated total RNA was cleaned up with RNeasy midi kit (Quiagen) and qualified as a microarray target samples using Bioanalyzer 2100 system (Agilent) and NanoDrop spectrophotometer.

1 µg of qualified total RNA was amplified and labeled with Cy3, Cy5 dyes. Hybridization was performed with Zebrafish Agilent Gene Expression Microarray Chips (Agilent), followed by washing step. Hybridized chips were then scanned, normalized and analyzed.

cDNA was synthesized using ImProm-II reverse transcription system (Promega). The quantitative real-time PCR was performed using the Lightcycler system (Roche Applied Science) with the Lightcycler-FastStart DNA Master SYBR Green I (Roche Applied Science) according to the manufacturer's instructions. Each reaction was performed in a volume of 20 µl with final concentration of 1×Lightcycler-FastStart DNA Master SYBR Green 1, 3 mM MgCl₂, 0.5 µM of each primer and 2 µl of first strand cDNA mixture. Quantitative real-time PCR was carried out for 40 cycles and fluorescence readings were acquired at the end of each amplification cycle at 72℃. Melting curve analysis was performed with continuous fluorescence acquisition from 65 to 95℃ at a temperature transition rate of 0.1℃/s to determine the amplification specificity. All reactions were performed as technical duplicates.

Embryos were anesthetized until movement had ceased and fixed in 4% paraformaldehyde overnight. Fixed embryos were embedded in 1.5% agar with 5% sucrose blocks and equilibrated by 30% sucrose solution. Frozen blocks were sliced into 10 µm sections on glass slides using cryostat microtome. Sections were rinsed with PBS several times and then blocked in 2% bovine serum albumin with sheep serum. Embryos were processed for immunohistochemistry, following primary antibodies were used: rabbit anti-Sox10 (1 : 250), rabbit anti-MBP (1 : 100), rabbit anti-GFP (1 : 500, Abcam) and mouse anti-HuC/D (16A11 1 : 20, Molecular Probes, Eugene, OR). For fluorescent detection of antibody labeling, we used Alexa Fluor 568-conjugated goat anti-mouse IgG, Alexa Fluor 568-conjugated goat anti-rabbit IgG, and Alexa Fluor 488-conjugated goat anti-rabbit IgG (1 : 500, Molecular Probes). Photos were taken using a confocal lasers canning microscope (LSM 510 Pascal, Carl Zeiss).

To identify the genes involved in oligodendrocyte differentiation and myelination, we isolated OPCs and differentiated oligodendrocytes expressing EGFP from the transgenic zebrafish and compared their transcriptional profiles. Therefore, the embryo stages suitable for collecting OPCs and differentiated oligodendrocytes were first identified. Previously, we have shown that olig2-expressing precursor cells in the pMN precursor domain of the ventral spinal cord first generate motor neurons during neurogenesis and later produce migrating OPCs, which express olig2 continuously to the mature oligodendrocyte stage (Park et al., 2002a; Park and Appel, 2003; Park et al., 2004). As soon as OPCs are specified in the pMN precursor domain, sox10 starts to be expressed specifically in the oligodendrocyte lineage cells. Larvae carrying the Tg(olig2:egfp) and Tg(sox10:egfp) transgenic reporters, which express EGFP under the control of the olig2 and sox10 promoters (Shin et al., 2003; Carney et al., 2006), respectively, were examined for use in the transcriptome analysis.

Expression of EGFP in OPCs and mature oligodendrocytes at several developmental stages in the Tg(olig2:egfp) and Tg(sox10: egfp) larvae was verified by fluorescence microscopy using transverse sections of the spinal cord. By 2 days post-fertilization (dpf), sox10:EGFP⁺ cells are located in the gray matter of the spinal cord, in which Hu⁺ neurons are located, indicating that sox10:EGFP⁺ cells are undifferentiated OPCs (Fig. 1A, arrow-heads). By 5 dpf, sox10:EGFP⁺ OPCs have migrated to the white matter of the spinal cord and differentiated into mature oligodendrocytes with multiple processes (Fig. 1B, arrows). Labeling of the spinal cord of Tg(olig2:egfp) larvae with anti-Sox10, a marker for OPCs and mature oligodendrocytes, revealed that Sox10⁺, olig2:EGFP⁺ cells are located in the gray matter of ventral spinal cord at 2 dpf, indicating that these cells are undifferentiated OPCs (Fig. 1C, arrowheads). Sox10⁺, olig2:EGFP⁺ cells migrated to the white matter and differentiated into mature oligodendrocytes possessing multiple process by 5 dpf (Fig. 1D, arrows). These data demonstrated that in the spinal cord of Tg(olig2:egfp) and Tg (sox10:egfp) larvae, EGFP-expressing cells are OPCs at 2 dpf and mature oligodendrocytes at 5 dpf. Thus, OPCs and mature oligodendrocytes can be obtained by sorting EGFP⁺ cells from Tg (olig2:egfp) and Tg(sox10:egfp) 2 dpf and 5 dpf larvae, respectively.

To identify the genes specific to oligodendrocytes, a comparison of transcriptional profiles of normal and oligodendrocyte-deficient larvae could be conducted. Previously, oligodendrocyte development was shown to require histone deacetylase (HDAC) activity (Marin-Husstege et al., 2002; Cunliffe and Casaccia-Bonnefil, 2006; Ye et al., 2009), and treatment with the HDAC inhibitor Trichostatin A can block formation of oligodendrocytes in zebrafish (Tsai et al., 2006). Consistent with previous reports, we found that treatment of Tg(olig2:egfp) embryos with TSA from 36 hours post fertilization (hpf) to 5 dpf blocked migration of Sox10⁺, olig2:EGFP⁺ oligodendrocytes, which are located in the white matter of the spinal cord in the normal 3 dpf larvae (Fig. 1E and F, arrows indicate mature oligodendrocytes). These data indicated that comparison of transcriptional profiles of EGFP⁺ cells from TSA-treated and untreated Tg(olig2:egfp) larvae can be used to identify genes expressed specifically in oligodendrocytes.

On the basis of the above data, OPCs and mature oligodendrocytes were purified from the dissociated cells of Tg(olig2:egfp) and Tg(sox10:egfp) larvae at 2 dpf and 5 dpf, respectively, using FACS. EGFP⁺ cells were also purified from TSA-treated and untreated 3 dpf Tg(olig2:egfp) larvae. Prior to sorting of EGFP⁺ cells, dissociated cells from the Tg(olig2:egfp) and Tg(sox10:egfp) larvae contained 20% EGFP⁺ cells (Fig. 2A~C). After sorting, an average of 98% of the sorted cells was EGFP⁺ OPCs or mature oligodendrocytes (Fig. 2D~F). To confirm the purity of the sorted cells, total RNA was isolated from EGFP⁺ mature oligodendrocytes and semi-quantitative RT-PCR was performed with several PCR primers specific to mature oligodendrocytes, neurons, and blood vessels. Mature oligodendrocyte specific genes were successfully amplified by RT-PCR, but neuron and blood vessel specific genes were not. These data indicate that oligodendrocyte lineage cells were specifically isolated by FACS (Fig. 2G).

To compare the transcriptional profiles of OPCs and mature oligodendrocytes, we purified RNA from the sorted cells and synthesized dye-labeled cDNAs, which were hybridized to microarray gene chips. A total of 19,956 transcripts were examined with the Agilent zebrafish gene expression microarray. Genes up-regulated by 2.5-fold in mature oligodendrocytes were identified as genes required for oligodendrocyte maturation and genes down-regulated to 0.3-fold or less in mature oligodendrocytes were identified as genes required for the maintenance of OPC status. Fig. 3A and B illustrate diagrammatically the 4348 up-regulated and 1808 down-regulated genes in mature oligodendrocytes compared to OPCs as analyzed by three different algorithms. In addition, a total of 32 up-regulated and 19 down-regulated genes were selected from the intersection of the three algorithms (Fig. 3A, B). To confirm the accuracy of the microarray screen, we searched the expression profile of well-characterized genes required for the oligodendrocyte differentiation, such as mpz, plp1b, mag, cldnk and flj13639, and found that the expression levels of these genes were highly elevated in mature oligodendrocytes (Fig. 3C), indicating that our microarray analysis was successful.

To validate the microarray data, we performed a qRT-PCR analysis of selected genes which exhibited highly elevated expression levels in the differentiated oligodendrocytes. We selected seven genes not previously known to be involved in oligodendrocyte differentiation. The expression levels for bhlhe40 (basic helix-loop-helix family, member e40, NM_212679), dlg1 (discs, large (Drosophila) homolog 1, NM_199526), rbp2a (retinol binding protein 2a, cellular, NM_153004), rock2b (rho-associated, coiled-coil containing protein kinase 2b, NM_001093747), sepp1a (selenoprotein P, plasma, 1a, NM_178297), zgc:110190 (NM_001020570) and zgc:112160 (NM_001017724) were evaluated by qRT-PCR using RNA purified from the sorted OPCs and mature oligodendrocytes. Myelin basic protein (mbp) was used as a positive control, and the expression levels of the selected genes are presented as the relative induction in differentiated oligodendrocytes compared to OPCs. Consistent with the microarray profiles, qRT-PCR demonstrated that the expression of these seven genes was highly elevated in differentiated oligodendrocytes (Fig. 4), suggesting that all seven genes may be involved in oligodendrocyte differentiation.

Basic helix-loop-helix family member e40 (bHLHB2, also known as Dec1/Eip1/Sharp2/Stra13/Clast5) is a novel cAMP-inducible bHLH protein family member. Stra13 is required for the proliferation and differentiation of nerve cells. Stra13 controls growth arrest, and represses the expression of cell proliferation associated genes through an HDAC1-independent mechanism (Sun and Taneja, 2000). Dec1 is expressed in the nuclei of endothelial, glial and neuronal cells in brain tissue (Preusser et al., 2005). Since the bHLHB2 expression level is elevated in differentiated oligodendrocytes, we hypothesize that bHLHB2 plays a role in oligodendrocyte differentiation by regulating the OPC cell cycle. Discs-large homolog 1 (Dlg1) is a multi-domain scaffolding protein that is involved in protein trafficking in polarized cells and membrane addition in Drosophila (Lee et al., 2003; Gorczyca et al., 2007). Dlg1 is expressed in myelin-forming Schwann cells, and enriched in Schwann cell paranodes where myelin outfoldings preferentially arise. Interestingly, dlg1 interacts with kinesin 13B (kif13B) and Sec8, which play a central role in oligodendrocyte membrane formation by regulating the recruitment of vesicles that transport myelin proteins, suggesting that dlg1 function may be required for oligodendrocyte differentiation (Bolis et al., 2009). Retinol binding protein 2a (Rbp2a/CRBP) is a cytoplasmic retinol-specific binding protein and is expressed in the lateral ganglionic eminences (LGE), where a second wave of OPCs originate in the telencephalon (Kessaris et al., 2008). Retinol is naturally converted into retinoic acid which regulates the timing of oligodendrocyte differentiation in vivo (Barres et al., 1994). Therefore, we hypothesize that Rbp2a may play a role in oligodendrocyte development together with retinol.

Rho-GTPase family proteins play important roles in coordinating the remodeling of the actin cytoskeleton during myelination. A key effecter of Rho-GTPase is Rho-kinase (ROCK), a serine/threonine kinase that regulates cell migration, proliferation and survival (Mueller et al., 2005). ROCK is expressed in Schwann cells at the onset of myelination, and pharmacological inhibition of ROCK activity causes Schwann cells form aberrant short myelin segments. However, after ROCK inhibition is removed, new myelin segments are formed normally (Melendez-Vasquez et al., 2004), indicating that ROCK plays an important role together with Rho-GTPase in the progression of the Schwann cell membrane encasement of the axon during myelination. Rho-associated coiled-coil containing protein kinase 2b (Rock2b) is a ROCK family protein and our microarray data showed that Rock2b expression was highly elevated in myelinating oligodendrocytes, suggesting the possibility that Rock2b function is involved in myelin formation in CNS oligodendrocytes. Selenoprotein P, plasma, 1a (Sepp1a) is a selenium binding protein. Selenium is an essential micronutrient necessary for normal growth and function of the brain, and selenium deficiency causes a CNS demyelinating disease. Selenium is required for the normal morphological development of oligodendrocytes and their survival in vitro (Eccleston and Silberberg, 1984), and is also required for the normal up-regulation of myelin genes in differentiating oligodendrocytes (Gu et al., 1997). Therefore, significant up-regulation of Sepp1a in mature oligodendrocytes indicates the possibility that Sepp1a is involved in selenium signaling in oligodendrocyte differentiation. We also examined the previously unannotated genes zgc:110191 and zgc:112160. These two novel factors are expressed at elevated levels in differentiated oligodendrocytes.