Knockdown of Cyclin-dependent Kinase 10 (cdk10) Gene Impairs Neural Progenitor Survival via Modulation of raf1a Gene Expression*

Background: CDK10 is involved in cell proliferation; however, the function of CDK10 in cell development remains unclear. Results: The knockdown of cdk10 results in the apoptosis of neural progenitor cells (NPCs) through enhanced raf1a expression. Conclusion: Zebrafish cdk10 exhibits an important function in neurogenesis by the modulation of the raf1a level during development. Significance: These findings emphasize the significance of cdk10 in NPC survival. In this study, we used zebrafish as an animal model to elucidate the developmental function of cdk10 in vertebrates. In situ hybridization analyses demonstrated that cdk10 is expressed throughout development with a relative enrichment in the brain in the late stages. Similar to its mammalian ortholog, cdk10 can interact with the transcription factor ETS2 and exhibit kinase activity by phosphorylating histone H1. Morpholino-based loss of cdk10 expression caused apoptosis in sox2-positive cells and decreased the expression of subsequent neuronal markers. Acetylated tubulin staining revealed a significant reduction in the number of Rohon-Beard sensory neurons in cdk10 morphants. This result is similar to that demonstrated by decreased islet2 expression in the dorsal regions. Moreover, cdk10 morphants exhibited a marked loss of huC-positive neurons in the telencephalon and throughout the spinal cord axis. The population of retinal ganglion cells was also diminished in cdk10 morphants. These phenotypes were rescued by co-injection of cdk10 mRNA. Interestingly, the knockdown of cdk10 significantly elevated raf1a mRNA expression. Meanwhile, an MEK inhibitor (U0126) recovered sox2 and ngn1 transcript levels in cdk10 morphants. Our findings provide the first functional characterization of cdk10 in vertebrate development and reveal its critical function in neurogenesis by modulation of raf1a expression.

In this study, we used zebrafish as an animal model to elucidate the developmental function of cdk10 in vertebrates. In situ hybridization analyses demonstrated that cdk10 is expressed throughout development with a relative enrichment in the brain in the late stages. Similar to its mammalian ortholog, cdk10 can interact with the transcription factor ETS2 and exhibit kinase activity by phosphorylating histone H1. Morpholino-based loss of cdk10 expression caused apoptosis in sox2-positive cells and decreased the expression of subsequent neuronal markers. Acetylated tubulin staining revealed a significant reduction in the number of Rohon-Beard sensory neurons in cdk10 morphants. This result is similar to that demonstrated by decreased islet2 expression in the dorsal regions. Moreover, cdk10 morphants exhibited a marked loss of huC-positive neurons in the telencephalon and throughout the spinal cord axis. The population of retinal ganglion cells was also diminished in cdk10 morphants. These phenotypes were rescued by co-injection of cdk10 mRNA. Interestingly, the knockdown of cdk10 significantly elevated raf1a mRNA expression. Meanwhile, an MEK inhibitor (U0126) recovered sox2 and ngn1 transcript levels in cdk10 morphants. Our findings provide the first functional characterization of cdk10 in vertebrate development and reveal its critical function in neurogenesis by modulation of raf1a expression.
Protein kinase signals mediate various cellular functions such as metabolism, cell proliferation, cell growth, and apoptosis (1,2). The cyclin-dependent kinase (CDK) 2 family of proteins is a group of serine/threonine kinases that regulate cell cycle progression specifically at the G 1 /S and G 2 /M checkpoints (3,4). Cell cycle progression requires the association of CDK with specific cyclins to form active complexes, which subsequently control the expression of downstream cell cycle genes (3,4).
A family of cdc2-related kinases, including PISSLRE (CDK10), PCTAIRE, KKIALRE (CDK-like-1; CDKL1), KKIAMRE (CDKlike-2; CDKL2), and NKIATRE (CDK-like-3; CDKL3), was recently identified based on their biochemical and genetic structures. Their names refer to their sequence, which corresponds to the cdc2 PSTAIRE motif (5). No known cyclins are associated with cdc2-related kinases. Northern blot analyses revealed that human CDK10 is broadly expressed in normal and tumor tissues (6). In vitro kinase assays also showed that CDK10 can phosphorylate histone H1 and myelin basic protein (6). ETS2 is a CDK10-interacting protein identified by yeast two-hybridization screening (7). CDK10 is associated with the N-terminal domain of ETS2, which harbors the transactivation domain and the highly conserved point domain of ETS2. However, CDK10 is not associated with ETS1 (7). When CDK10 binds to ETS2, CDK10 can inhibit the transactivation activity of ETS2 (7). Several reports have suggested that CDK10 is involved in cell proliferation and differentiation. In human U2OS cells, the abrogation of the CDK10 signal may be achieved by using a kinase-dead form of CDK10 or antisense RNA-arrested cell cycle progression at the G 2 /M checkpoint (8). CDK10 was significantly down-regulated in retinoic acid-induced differentiation of retinoblastoma cells (9). Using two-dimensional gel electrophoresis and mass spectrometry analysis, Leman et al. (10) demonstrated the down-regulation of CDK10 in seminomas. CDK10 was shown to be up-regulated in follicular lymphomas compared with normal germinal center B-cells (11). CDK10 was also reportedly involved in endocrine therapy resistance in breast cancer (12). Using siRNA screening, Iorns et al. (12) found that changes in CDK10 expression can increase ETS2 transcriptional activity, elevate c-raf expression, and enhance MAPK activation, resulting in resistance to tamoxifen treatment.
Zebrafish is a widely used animal model in the study of genetic and developmental biology (13). To date, the role of CDKs in zebrafish development remains poorly understood. Zebrafish cyclin H/cdk7 is ubiquitously expressed throughout all examined stages and tissues (14). Abrogation of cyclin H expression results in defective development and apoptosis (14). Previous reports revealed that zebrafish cdk5 is ubiquitously and abundantly expressed in the brain and the eyes (15). Knockdown of zebrafish cdk5 by using siRNA decreases the number of primary sensory Rohon-Beard (RB) neurons in the trigeminal ganglia and promotes apoptosis in the brain (15,16). Meanwhile, morpholino-based knockdown of cdk5 increases the number of supernumerary motor neurons (17). Recently, cdkl1 was reported to fulfill a function in zebrafish development. Interruption of cdkl1 expression decreases neurogenin-1 (ngn1) levels in the brain and sonic hedgehog expression in the floor plate region (18). In the present study, we used zebrafish as an animal model to investigate the developmental role of cdk10.

EXPERIMENTAL PROCEDURES
Reagents-All of the chemical reagents were purchased from Sigma or Mallinckrodt Baker. Antibodies against human CDK10, human ETS2, and GFP and HRP-conjugated goat antimouse IgG antibodies were purchased from Santa Cruz Biotechnology. The anti-␤-actin antibody was obtained from Sigma. Anti-sox2 antibody (ab97959) was purchased from Abcam.
Maintenance of Zebrafish-The zebrafish AB strain and transgenic Tg (huC:gfp) fish were kind gifts from the Taiwan Zebrafish Core Facility (Academia Sinica, Taiwan) and Dr. Chang-Jen Huang (Institute of Biological Chemistry, Academia Sinica, Taipei, Taiwan), respectively. All of the fish were raised and maintained under standard conditions at 28.5°C under a 14-h light/10-h dark cycle (19). The developmental stages of the embryos were determined according to methods described previously (20). All of the animal handling protocols were approved by the Chung Shan Medical University Animal Care Committee. Images of the fluorescent signals were captured using a Zeiss LSM510 confocal microscope.
RNA Extraction, Reverse Transcription-Polymerase Chain Reaction (RT-PCR), Quantitative Real Time PCR Analysis, and Plasmid Constructs-Total RNA was isolated from different tissues and stages in adult zebrafish using TRIzol (Invitrogen). First strand cDNA was synthesized from 3 g of total RNA by Moloney murine leukemia virus reverse transcriptase (Promega) according to the manufacturer's instructions. The full coding region of cdk10 was amplified by RT-PCR using zebrafish cDNA as a template and the following primer set: forward, 5Ј-ATGGAGAGCACAGCAGACAC-3Ј; reverse, 5Ј-GTCATACTTTGCTCCTCTTGG-3Ј. The PCR was performed under the following conditions: 94°C, 30 s; 52°C, 60 s; 72°C, 60 s for 30 cycles. The amplified products were then ligated into a pGEM-T-Easy vector (Promega) and transformed into competent cells. The pEGFP-cdk10 and pCS2-cdk10 were constructed by enzyme digestion and ligated into pEGFP-C1 and pCS2, respectively. Quantitative real time PCR was performed on an ABI 7000 Sequence Detection System (Applied Biosystems) with SYBR Green fluorescent label (Thermo Scientific). Primers for zebrafish raf1a (forward, 5Ј-AAGACATC-AGCTCCTGCACA-3Ј; reverse, 5Ј-AGACGATCACTCTTTG-AGGG-3Ј) were used. Gene expression levels were normalized to elongation factor 1␣ (ef1␣) (forward, 5Ј-TGCCTTCGTCCCAA-TTTCAG-3Ј; reverse, 5Ј-TACCCTCCTTGCGCTCAATC-3Ј) and assessed using comparative C T (40 cycles) according to the manufacturer's instructions (Applied Biosystems). The expression level was normalized to ef1␣ and calculated using the ⌬⌬C T method.
TUNEL Assay-A TUNEL assay was performed according to the manufacturer's instructions (Promega). Briefly, the embryos (24 or 48 HPF) were fixed with 4% paraformaldehyde at 4°C overnight, dehydrated by washing three times in methanol, and stored at Ϫ20°C for at least 30 min. After rehydration, the embryos were transferred to tubes containing 100 l of equilibration buffer and incubated at room temperature for 10 min. Next, the embryos were incubated with 100 l of labeling solution at 37°C for 1 h. After washing with 2ϫ SSC buffer and phosphate-buffered saline (PBS), the TUNEL signals were examined using an upright fluorescence microscope (Zeiss AXioskop2).
Cell Culture and Transfection-Mouse neuroblastoma Neuro-2a and human embryonic kidney (HEK) 293 cells were obtained from ATCC and maintained in minimum essential medium supplemented with 10% fetal bovine serum, 100 g/ml streptomycin, and 100 units/ml penicillin. The cells were maintained in a humidified atmosphere of 95% air and 5% CO 2 . Plasmid DNAs were transfected into Neuro-2a or HEK 293 cells using the Lipofectamine reagent (Invitrogen). Twenty-four hours after transfection, the cell lysates were collected.
Immunoblot Analysis and in Vitro Kinase Assay-Five hundred micrograms of total cell lysates derived from HEK 293 cells overexpressing GFP alone or the GFP-cdk10 fusion protein was subjected to immunoprecipitation using an anti-GFP antibody, and a kinase assay was performed as described previously (18,21). After washing three times with kinase buffer (25 mM Tris, pH 7.5, 150 mM NaCl, 2 mM dithiothreitol, and 0.1 mM sodium orthovanadate), the precipitates were incubated with H1 histone in kinase buffer supplemented with 0.1 mM ATP, pH 7.0, 5 Ci of [␥-32 P]ATP, and 10 mM MgCl 2 at 37°C for 10 min. After separation by electrophoresis with a 12.5% SDS-polyacrylamide gel, the phosphorylation status was determined by autoradiography of the dried gel.
Microinjection of Antisense Morpholino Oligonucleotides (MOs) and Synthetic mRNA-The following MOs were designed and synthesized by Gene Tools, LLC (Philomath, OR): cdk10 MO (5Ј-GGTCTGTGTCTGCTGTGCTCTCCAT-3Ј), which targets the start codon (from ϩ1 to ϩ25); 5-mismatch (5-mis) MO (5Ј-GGTGTCTGTCTCCTGTCCTGTCCAT-3Ј) where the underlines indicate the mismatched sites; and control MO (5Ј-CCTCTTACCTCAGTTACAATTTATA-3Ј), which corresponded to the human ␤-globin gene and was used as a negative control. MOs were dissolved in 1ϫ Danieau solution and injected into the embryos at the one-to four-cell stage. The pCS2-cdk10 plasmid was linearized by NotI, and the mRNA was synthesized by SP6 RNA polymerase using the mMESSAGE mMACHINE kit (Ambion) according to the manufacturer's recommendation. For the rescue experiments, morpholinos were co-injected with 11 pg of mRNA into one-to four-cell embryos.
Whole-mount in Situ Hybridization-Whole-mount in situ hybridization was performed as described previously (22). Digoxigenin-labeled antisense riboprobes were generated from a pGEM-T-Easy plasmid containing the full coding sequence and partial 3Ј-untranslated regions of cdk10. The plasmid was linearized using NotI. The RNA was transcribed in vitro by the SP6 RNA polymerase to generate antisense probes using the DIG-RNA labeling kit (Roche Applied Science).
Western Blot Analysis-Zebrafish embryos injected with the control MO, cdk10 MO, 5-mismatch MO, or cdk10 MO plus cdk10 mRNA were collected at 24 HPF. The total protein was extracted as described above. Thirty micrograms of protein was separated by SDS-PAGE and transferred onto PVDF membranes. The membranes were blocked with PBS containing 5% nonfat milk, washed with PBS buffer, and incubated with antihuman CDK10 antibody (Santa Cruz Biotechnology) overnight. Next, the membranes were incubated with an HRP-conjugated goat anti-mouse IgG antibody, and the signals were detected using an enhanced chemiluminescence kit (Millipore).
Co-immunoprecipitation-HEK 293 cells were transfected with 2 g of pEGFP-cdk10 or pEGFP vector, and the cell lysates were collected as described above. Four hundred micrograms of proteins was subjected to immunoprecipitation using an anti-GFP antibody. The immunoprecipitated complexes were separated by electrophoresis on a 10% polyacrylamide gel followed by immunoblot analysis using anti-human ETS2 and anti-GFP antibodies. The signal was detected using an enhanced chemiluminescence kit (Millipore).
Anti-acetylated ␣-Tubulin Antibody Staining-Briefly, embryos at the indicated stages were collected, fixed in 4% paraformaldehyde at 4°C overnight, and stored in methanol at Ϫ20°C. After rehydration with PBST (PBS containing 0.1% Tween 20), the embryos were blocked with blocking solution (5% goat serum, 2% bovine serum albumin, and 1% DMSO in PBST) for 1 h and then incubated with monoclonal anti-acetylated ␣-tubulin at 4°C overnight. Samples were washed with PBST and incubated with goat anti-mouse FITC-conjugated antibodies (1:500) in blocking buffer at 4°C overnight in the dark. After washed with washing buffer (2% BSA and 1% DMSO in PBST), the images were captured by a scanning laser confocal microscope (Zeiss 510 META).
Statistical Analysis-Statistical analysis was performed by Student's t test using SPSS software (SPSS, Inc.). A p value Ͻ0.05 was considered statistically significant.  with an anti-GFP antibody. NS denotes nonspecific signal. The arrowhead indicates ETS2 signals. The arrow indicates GFP-cdk10 signals. B, in vitro kinase assay. pEGFP or pEGFP-zcdk10 was transfected into HEK 293 cells. GFP fusion proteins were immunoprecipitated with an anti-GFP antibody and subjected to an in vitro kinase assay using histone H1 as substrate. Left panel, immunocomplexes were analyzed using Western blot and anti-GFP antibody. Right panel, phosphorylated signals were detected by autoradiography.

Identification of Zebrafish cdk10-We performed Basic Local
Alignment Search Tool program analysis using the human CDK10 protein sequence to study the function of cdk10 in zebrafish. A gene (NCBI Reference Sequence accession number NM_001017622) that encodes the full-length zebrafish cdk10 was identified. Zebrafish cdk10 contains 1298 base pairs; encodes a polypeptide of 360 amino acid residues; and displays 78, 77, 77, and 76% identity to human, mouse, rat, and Xenopus orthologs, respectively. The kinase domain extended from residues 48 to 300, and the conserved PISSLRE motif was localized at amino acids 81-87 (Fig. 1A). On the basis of the phylogenetic analysis, zebrafish cdk10 was closer to Xenopus cdk10 (Fig. 1B).
Functional Characterization of cdk10-A previous report showed that human CDK10 interacts with the transcription factor ETS2 (7). We performed a co-immunoprecipitation assay to investigate whether cdk10 interacts with human ETS2. A strong ETS2 signal was detected in the complexes immunoprecipitated with GFP-cdk10 but not in those immunoprecipitated solely with GFP ( Fig. 2A). These results demonstrated that cdk10 interacts with ETS2 and are similar to the results obtained with human and murine CDK10.
We performed an in vitro kinase assay to confirm whether cdk10 exhibits kinase activity. The GFP-cdk10 fusion protein was overexpressed in the HEK 293 cell line and purified with an anti-GFP antibody. Histone H1 was phosphorylated by cdk10 in the presence of an isotope-labeled ATP (Fig. 2B). Similar to its mammalian ortholog, cdk10 was shown to interact with ETS2 and exhibit kinase activity during the phosphorylation of histone H1.
Temporal and Spatial Expression Patterns of cdk10-We conducted RT-PCR and whole-mount in situ hybridization to analyze the developmental expression patterns of cdk10. RT- PCR analysis showed that cdk10 was expressed in all examined zebrafish stages. Furthermore, cdk10 was predominantly expressed in the adult brain, intestine, liver, ovary, and testis. Whole-mount in situ hybridization revealed that cdk10 mRNA was ubiquitously expressed from the two-cell to sphere stages. At 24 and 36 HPF, cdk10 expression was relatively enriched in the eye and brain regions. At 48 HPF, cdk10 was predominantly expressed in the eye, tectum, hindbrain-midbrain boundary, and telencephalon (Fig. 3, A and B).
Developmental Role of cdk10-Antisense MOs that overlap with the translation start codon were designed to block the translation of the cdk10 gene and thereby explain the function of cdk10 during zebrafish development. To confirm whether cdk10 MO affects endogenous cdk10 expression, protein extracts from embryos receiving control MO, cdk10 MO, and 5-mis MO were subjected to Western blot analysis using antihuman CDK10 antibody. As shown in Fig. 4B, 1 ng of cdk10 MO reduced the amount of cdk10 protein expression by 44%. However, no clear changes in cdk10 expression were observed in embryos that received control MO or 5-mis MO. These results indicated that cdk10 MO specifically blocks the translation of the endogenous cdk10 gene. A previous report found that the microinjection of specific MOs induces off-target effects by increasing p53 activity, resulting in cell death (23). These offtarget effects were avoided in this study by using specific MOs that were co-injected with 2-fold amounts of p53 MO in all MO experiments. To determine whether p53 MO contributes to the phenotype of cdk10 morphants, we injected the embryos with p53 MO alone. Indeed, no distinct phenotype was observed between the p53 morphants and the control embryos (data not shown). At 24 HPF, embryos receiving 1 ng of cdk10 MO exhibited significant developmental abnormalities such as small brain, undersized eye, and somite curvature. These embryos also lacked the isthmus region. The percentages of normal, moderate, and severe phenotypes (characterized by smaller brain size, undersized eyes, and curved somites) are presented in Fig. 4C. Developmental defects observed in cdk10 morphants as well as endogenous cdk10 protein level could be rescued by co-injection of cdk10 mRNA, demonstrating the specificity of this morpholino (Fig. 4, A and B). Dark cell masses observed in cdk10-depleted embryos indicated that these morphants underwent cell death. To verify this hypothesis, we performed a TUNEL assay to examine the embryos for apoptosis. Compared with the control MO-injected embryos, depletion of cdk10 sharply increased the number of TUNEL-positive cells in the dorsal neuronal regions at the three-somite stage (in the brain at 24 HPF as well as the forebrain and the midbrain regions at 48 HPF). Co-injection with cdk10 mRNA significantly reduced apoptosis (Fig. 5, A-C). These findings show the significance of cdk10 during development in zebrafish.
Knockdown of cdk10 Abrogates Several Central Nervous System (CNS) Marker Expression-We examined the expression of sox2 (a known neural stem cell marker) and ngn1 (a marker of differentiating neurons at the three-somite stage) to evaluate the effect of cdk10 on neural development. Depletion of cdk10 resulted in expansion and reduction of sox2 expression compared with wild-type control embryos (Fig. 6A). Loss of cdk10 reduced ngn1 expression in dorsal neuron precursor cells (Fig.  6B). At 24 HPF, morphant embryos showed a significant disruption of ngn1 expression in the forebrain and the cranial ganglia (Fig. 6C). Significant reductions in zath3 (a late proneural gene) and nes (a marker for stem cells) were observed in cdk10 morphants (Fig. 6D). In addition, wnt1 (alar plate midbrain region) expression was significantly reduced in the dorsal midline neurons of cdk10-depleted morphants at 24 HPF (Fig. 6E). We subsequently determined whether forebrain patterning events were disrupted in cdk10 mutants. The expression of emx1 (a marker involved in telencephalic patterning) was markedly reduced in cdk10 morphants (Fig. 6F). The expression of islet2 (a dorsal root ganglion marker) was significantly reduced in RB neurons, whereas motor neurons were not affected in cdk10 morphants (Fig. 6G). These findings suggest the significance of cdk10 in the formation of forebrain and midbrain as well as RB neurons.
To determine the role of cdk10 during neurodevelopment, we conducted acetylated ␣-tubulin (AcTub) staining to determine whether disruption of cdk10 affects neuronal differentiation. At 24 HPF, AcTub staining revealed the presence of anterior commissure, postoptic commissure, supraoptic tract, nucleus of the tract of anterior commissure, dorsal-ventral diencephalic tract, epiphyseal cluster, and tract of the posterior commissure in the forebrain and the midbrain in the embryos injected with control and 5-mis MOs (Fig. 7A). Axonal scaffolds in cdk10 morphants visibly deteriorated. Reduced arborization was evident in several neurons, including those in the spinal cord (Fig. 7B). Embryos co-injected with cdk10 mRNA resembled those of the control group. Similar results were also observed at 48 HPF at which time the staining patterns of the tectal neurons, hindbrain neurons, and tubulin-labeled axons were severely reduced in cdk10 morphants. Co-injection with specific cdk10 mRNA restored these defects (Fig. 7C). Therefore, depletion of cdk10 caused abnormalities in the number of neurons and in axonal pathfinding, indicating that cdk10 is involved in the proper maintenance of neurons.
We also used the Tg (huC:gfp) model to track the generation of terminally differentiated neurons. A lateral view of the embryos at 24 HPF revealed that the control, 5-mis MO, and cdk10 MO plus cdk10 mRNA groups exhibited high levels of huC expression in the telencephalic cluster and neural tube. We observed that cdk10 morphants exhibited abrogated huC expression in the forebrain region and a significantly decreased huC population in the neural tube throughout the spinal cord axis (Fig. 8, A and B). At 48 HPF, only a small number of huCpositive neurons were present in the tectum and hindbrain regions of cdk10 morphants (Fig. 8C). A dorsal view of the embryos revealed that cdk10 morphants also exhibited severe defects of neuronal organization in the olfactory bulb, telencephalic cluster, and retinal ganglion (Fig. 8D). Thus, cdk10 knockdown in zebrafish embryos affects neuronal development in the brain.
Zebrafish cdk10 Did Not Directly Contribute to the Cell Cycle-The results showed that cdk10 knockdown induced cell apoptosis (Fig. 5) and hindered CNS development (Figs. 7 and  8). This phenomenon of cell death may result in compensatory proliferation, thereby influencing cell cycle distribution and abrogation of differentiation. This finding is consistent with the results of previous studies on Drosophila (24), chicken (25), and zebrafish embryos (26). We then performed in vitro and in vivo assays to assess cell cycle progression using gain and loss of cdk10 function. HEK 293 cells with or without cdk10 overexpression were collected and then analyzed for populations of cell cycle by flow cytometry. Our findings revealed no apparent change in cell cycle distribution in the cells transfected with the control or cdk10 (Fig. 9A). Despite the cdk10 knockdown, the transcript levels of cdk inhibitors such as cdkn1b, cdkn1c, ccnd1, and PCNA (a proliferation marker), were similar to the control in the CNS region (Fig. 9B). This result indicates that zebrafish cdk10 did not elicit apoptosis-associated compensatory proliferation.

Decrease in Differentiation of Retinal Cells in cdk10
Morphants-Using the Tg (huC:gfp) model, we observed fewer retinal ganglion cells (RGCs) in cdk10 MO-injected embryos. We analyzed the expression of zath5, which is expressed in a wavelike fashion in the inner retina before the retinal ganglion cell differentiation (27). The expression of zath5 in the control retina was found around the retina to the ventrotemporal region at 36 HPF, whereas the zath5 signal was observed in differentiated RGCs near the ciliary region in normal retinas at 48 HPF. However, the spread of zath5 expression from the ventral to the dorsal region was significantly repressed at 36 HPF. Although retinas of cdk10 morphants seemed to have high levels of zath5 expression, most zath5-expressing cells were located adjacent to the optic stalk even at 48 HPF. Embryos co-injected with cdk10 mRNA resembled those of the control group. In addition, brn3, which is the downstream factor of zath5 (28), exhibited severely diminished expression in the ganglion cell layer (Fig. 10, A and B). This result indicates that the onset of zath5 expression was not delayed. However, fewer neurons were subsequently generated. Thus, cdk10 is required during RGC production.
Zebrafish cdk10 Protects Neural Progenitor Survival-We evaluated whether the apoptotic cells observed in cdk10 morphants were NPCs. We collected embryos at 80% epiboly and conducted immunostaining with fluorescent sox2 antibody prior to TUNEL assay. In both early and late stages, most apoptotic cells were co-localized with sox2-positive cells in the anterior neural plate and throughout the axis of CNS in cdk10 morphants. Co-injection with specific cdk10 mRNA evidently reduced apoptosis, indicating that the apoptotic cells in cdk10 morphants are NPCs (Fig. 11, A and B). By contrast, p53 morphants presented no obvious TUNEL-positive signals in the early stages. These results show that the abrogation of cdk10 may cause severe damage during NPC maintenance and subsequently disrupted neurogenesis.
raf1a Is Involved in cdk10-dependent Neuron Progenitor Survival-Previous studies reported that loss of CDK10 induces ETS2-dependent RAF1 gene expression, thereby activating the mitogen-activated protein kinase/extracellular signal-regulated kinase kinase (MEK)-extracellular signal-regulated kinase 1/2 (ERK1/2) pathway (12). This finding suggests that the zebrafish raf1a ortholog is also regulated by cdk10. To verify this hypothesis, we analyzed the transcriptional level of raf1a in cdk10-deficient embryos. The depletion of cdk10 markedly elevated the transcription of raf1a (Fig. 12A). Co-injection with cdk10 mRNA reduced raf1a expression to the basal level. No apparent alternation of raf1a expression was found in the embryos that received 5-mis MO. Embryos were co-treated with the MEK inhibitor U0126 in the presence of cdk10 MO, and the expression of sox2 and ngn1 were determined to assess whether the raf1a involved in cdk10 depletion induced NPC loss and aberrant neural differentiation. The knockdown of cdk10 sharply decreased sox2 expression at 80% epiboly. In addition, ngn1 expression was significantly lower compared with the control at the tail bud stage and 24 HPF. Co-injecting cdk10 mRNA restored sox2 and ngn1 expression. Interestingly, treatment with U0126 in cdk10 morphants recovered sox2 and ngn1 expression to the same level as that of the control groups. By contrast, treatment with U0126 alone reduced sox2 and ngn1 expression, suggesting that basal levels of ERK1/2 activities were still required for normal neurogenesis (Fig. 12B). Our collective findings revealed that raf1a rescued the impaired NPC development caused by cdk10 deficiency.

DISCUSSION
CDK10 is a serine/threonine protein kinase from the CDK family (5). Recent studies suggested that human CDK10 functions as a tumor suppressor gene, preventing cell proliferation in breast cancer (12), biliary tract cancer cells (29), and hepatoma (30). However, the developmental function of CDK10 remains unclear. We present evidence that zebrafish cdk10 is involved in NPC survival. The disruption of cdk10 expression by morpholinos caused defects in the brain by triggering apo-  ptosis in the brain and the dorsal neurons at 24 and 48 HPF, respectively. Whole-mount in situ hybridization revealed the down-regulation of several neuronal markers such as ngn1, emx1, wnt1, and islet2 in morphants. Co-injection with cdk10 mRNA rescued these phenotypes. Acetylated tubulin staining and Tg (huC:gfp) zebrafish demonstrated that the disruption of cdk10 expression alters telencephalic patterns and decreases RB neurons and RGCs. The results of this study show that cdk10 is an essential gene for zebrafish during normal embryogenesis.
The cdk10 mRNA transcripts were diffusedly expressed in the early stages (prior to the sphere stage) and more abundant in the brain at 24 and 48 HPF. The progressively restricted pattern of cdk10 expression was similar to that of other cdk and cdk-related genes in zebrafish. The cdk7/cyclin H and cdkl1 genes are ubiquitously distributed before gastrulation (14,18). At 24 HPF and thereafter, cdk7 and cyclin H are primarily  expressed in the brain, eye, and heart (14). The expression of cdkl1, which is another cdc2-related protein kinase, is restricted to the floor plate, pronephric ducts, and hypochord at 12 HPF (18). These cdks may exhibit redundant functions during the early developmental stage and perform specific functions during late developmental stages.
Specific morpholinos were designed to knock down cdk10 to assess the developmental role of cdk10 in zebrafish. A previous report showed that morpholinos may cause off-target effects that result in neuronal apoptosis because of p53 activation (23). To prevent off-target effects, p53 MO was co-injected with cdk10 MO into zebrafish embryos. The knockdown of cdk10 caused no obvious abnormalities. However, small brains and eyes as well as slight pericardial edema were observed in the morphants. Interestingly, impaired cdk10 expression resulted in apoptosis in the dorsal regions at the tail bud stage as well as the dorsal neurons and anterior brain regions where cdk10 is abundantly expressed. These findings were consistent with the whole-mount in situ hybridization data. The expression of sox2, which is a neural progenitor marker, was found to have expanded and decreased in cdk10 morphants. Reduced expression of ngn1, emx1, and wnt1 in cdk10 morphants was evident. Expression of a dominant negative form of cyclin H, which decreases cdk7 activities, results in apoptosis in the early stage of zebrafish embryos (14). Previous reports demonstrated the down-regulation of zebrafish cdk5 by using siRNA, which leads to apoptosis in the brain and a subsequent reduction in the number of primary sensory neurons in the trigeminal ganglion (15,31). The present study shows that cdk10 is essential for cell survival in zebrafish.
At 24 HPF, in situ hybridization of embryos by using an RNA probe against islet2 indicated that reduction in the number of RB neurons was more significant in embryos injected with cdk10 MO than in those injected with control MO. AcTub staining revealed a large reduction in the number of RB neurons along the body axis in cdk10 morphants. RB neurons are primary sensory neurons that gradually die within the first 2 days of development (32). A previous report showed that cdk5 may mediate RB neuronal survival. The knockdown of zebrafish cdk5 by specific siRNAs decreased the number of RB neurons in the spinal cord (15). Phosphorylation of dihydropyrimidinaselike proteins 2 and 3 (DPYSL2 and DPYSL3) by cdk5 and the dual specificity tyrosine-phosphorylated and -regulated kinase (DYRK2) were essential in the proper localization of RB and neural crest cells during neurulation in zebrafish development (33). Conversely, cdk5 knockdown using specific MOs induced motor neuron development (17). Exposure to perfluorooctane sulfonate resulted in motor neuron defects in zebrafish that may be attributed to increased cdk5 expression (34). However, no apparent changes in motor neuron expression were observed in cdk10 morphants. Our findings show that cdk10 exhibits a significant function in the development of the spinal sensory but not the motor nervous system. However, further investigation must be conducted to determine whether DPYSL2 and DPYSL3 are also downstream targets of cdk10.
Human CDK10 significantly affects the progression from the G 2 /M phase of the cell cycle (8). Overexpression of human CDK10 inhibits the proliferation of human biliary tract cancer (29) and hepatocellular carcinoma (30). By contrast, zebrafish cdk10 is not involved in G 2 /M phase progression. No apparent change in G 2 /M phase population was observed in cells overexpressing rodent cdk10 (35). An RNA probe of cdkn1b, cdkn1c, and ccnd1 revealed no considerable effects in cdk10-deficient embryos. These observations show that cdk10 exhibits various functional activities in cell cycle progression that differ between human and zebrafish.
RGCs, the output neurons of the retina, contribute to a topographical spatial code that corresponds with their position within the retinal compartments and in higher visual centers of the brain (36). A recent study suggested that elevation of phospho-NR2A(Ser-1232) by Cdk5/p35 contributes to the apoptotic death of RGCs in experimental glaucoma rats (37). In an in vivo optic nerve crush model, the combined treatment with the cdk5 inhibitor indolinone A and the Rho-associated protein kinase inhibitor Y-27632 produced a strong stabilizing effect on neuronal survival (38). Administration of the CDK4 inhibitor reduced ischemia-reperfusion injury-induced neural apoptosis (39). By contrast, cdk5 was found to support neuronal survival by Bcl-2 phosphorylation (40) and significantly influence the extension and maintenance of axons as well as the stability and steering of RGCs (41). Administration of the pan-CDK inhibitor AG-012986 also induced apoptosis in retina cells (42).
A previous report showed that depletion of CDK10 expression led to elevated Raf1 levels, consequently activating the MEK-ERK1/2 pathway (12). According to its duration and magnitude, Raf-MEK-ERK1/2 signaling controls diverse cell responses such as proliferation, differentiation, and cell death (29). In the nervous system, the Raf-MEK-ERK1/2 pathway affects the genesis of neural progenitors (29). Treatment with the ERK1/2 inhibitor PD98059 prevented traumatic brain injury in the hippocampus by suppressing ERK1/2 activation (30). Similarly, treatment with U0126 rescued the glutamateinduced neuronal injury by suppressing ERK1/2 activation (43). Death of striatal neurons by dopamine induction has been attributed to ERK1/2 activation (44). Yang et al. (45) recently demonstrated that Raf-MEK-ERK1/2 enhancement causes neurogenesis defects in cultured cortical neurons. Similarly, the present study demonstrated that the knockdown of cdk10 expression triggered raf1a expression and decreased expression of neuron progenitor markers such as sox2 and ngn1. Nevertheless, the exposure of embryos to U0126 alone under a nonlethal dosage severely diminished the transcriptional level of sox2 and ngn1, indicating the significance of raf1a in maintaining normal neurogenesis. Moreover, Raf1 exhibited a neuroprotective ability during serum withdrawal (46). Inhibition of Raf activities led the otic epithelia to undergo apoptosis, thereby reducing the neurogenesis of the acoustic-vestibular ganglion (47). The results of the present study collectively suggest that fine-tuning the modulation between cdk10 and the Raf-MEK-ERK1/2 axis is necessary for NPC survival.
Thus, the results of ours study emphasize that cdk10 may positively regulate NPC development. Changes in cdk10 expression led to NPC apoptosis in the early stages and subsequent impairment in differentiating neurons. Whole-mount in situ hybridization revealed a significant reduction in the number of RB neurons in cdk10 morphants. Tg (huC:gfp) zebrafish revealed a decrease in the number of RB neurons and RGCs at 24 and 48 HPF. The blocked cdk10 expression elicited an increase in the raf1a level, consequently decreasing NPC survival, whereas the MEK inhibitor reversed these phenomena. In summary, our findings show that cdk10 significantly affects NPC survival by Raf-MEK-ERK1/2 pathway modulation.