Opus: University of Bath Online Publication Store Anaplastic Lymphoma Kinase Is Required for Neurogenesis in the Developing Central Nervous System of Zebrafish

This version is made available in accordance with publisher policies. Please cite only the published version using the reference above. Abstract Anaplastic Lymphoma Kinase (ALK) was initially discovered as an oncogene in human lymphoma and other cancers, including neuroblastoma. However, little is known about the physiological function of ALK. We identified the alk ortholog in zebrafish (Danio rerio) and found that it is highly expressed in the developing central nervous system (CNS). Heat-shock inducible transgenic zebrafish lines were generated to over-express alk during early neurogenesis. Its ectopic expression resulted in activation of the MEK/ERK pathway, increased cell proliferation, and aberrant neurogenesis leading to mis-positioning of differentiated neurons. Thus, overexpressed alk is capable of promoting cell proliferation in the nervous system, similar to the situation in ALK-related cancers. Next, we used Morpholino mediated gene knock-down and a pharmacological inhibitor to interfere with expression and function of endogenous Alk. Alk inhibition did not affect neuron progenitor formation but severely compromised neuronal differentiation and neuron survival in the CNS. These data indicate that tightly controlled alk expression is critical for the balance between neural progenitor proliferation, differentiation and survival during embryonic neurogenesis. Copyright: ß 2013 Yao et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.


Introduction
Anaplastic Lymphoma Kinase (ALK) was initially discovered as an oncogene in human anaplastic large cell lymphomas (ALCL), a subset of T-cell lymphomas [1,2,3]. Genetic dissection revealed an hybrid gene (NPM-ALK, also known as p80) at the t(2;5)(p23;q35) chromosomal translocation breakpoint, comprising a fusion of a nucleolar protein gene nucleophosmin (NPM) and a part of a gene coding for ALK, a novel tyrosine kinase. Moreover, another fusion of alk with the gene encoding echinoderm microtubule-associated protein-like4, encoding EML4-ALK hybrid protein, was identified in a subset of non-small-cell lung cancers (NSCLCs) [4]. To date, several more ALK hybrid proteins have been identified in various cancer types, such as TRK-fused gene (TFG)-ALK, tropomyosin 3 (TPM3)-ALK, tropomyosin 4 (TPM4)-ALK, clathrin heavy chainlike 1 (CLTCL1)-ALK, and moesin (MSN)-ALK [5]. In most, if not all, of these hybrid proteins the distal portion of ALK containing the tyrosine kinase domain is fused with proximal parts of the diverse proteins having oligomerization domains. Consequently, the hybrid proteins undergo spontaneous dimerization, leading to autophosphorylation and ultimately constitutive enzymatic activation of the ALK tyrosine kinase domain.
Overexpression of ALK is often observed in human neural tumors, primarily neuroblastoma [6]. Neuroblastomas are embryonic tumors of the peripheral sympathetic nervous system derived from neural crest tissues. It is one of the most frequent pediatric solid tumors, accounting for about 15% of childhood cancer mortality [7]. A clear correlation of ALK deregulation with neuroblastoma development has been established [8,9,10,11], and at least ten ALK mutations, all found in the tyrosine kinase domain, were identified in the neuroblastoma samples. Among these, R1275Q, F1174L and F1245L/V/C occur most frequently [8,11]. Increased copy numbers and gene amplifications of the ALK locus is also often detected in neuroblastoma patients [8,10]. These germ line mutations in ALK explain most of the inheritable neuroblastomas, but activating mutations can also be somatically acquired. These mutations lead to constitutive autophosphorylation of mutated ALK, which displays increased kinase activity and excess phosphorylation on downstream targets such as Akt, STAT3 and ERK1/2. Mutated ALK from neuroblastoma samples is capable of transforming interleukin-3-dependent haematopoietic Ba/F3 cells into cytokine-independent growth [9], as well as NIH3T3 fibroblasts into colony and tumors in nude mice [8]. Knock-down of ALK by siRNA [8,11], or ALK inhibition by small molecules PF-2341066/Crizotinib [10] and TAE684 [9] leads to suppressed cell growth, with decreased proliferation and increased apoptosis in neuroblastoma cells harboring mutated or amplified ALK. Together, these studies in neuroblastomas suggest that in a normal cellular context, appropriate ALK expression and activity levels need to be maintained for cell proliferation and survival.
While a pathogenic role for ALK has been demonstrated for many human cancers, little is known about the physiological function of Alk during development [12,13,14,15]. Expression of ALK has been reported in murine neural tissues [13,14,16] but its function, especially in the CNS, remains elusive. A recent report in chicken showed that Alk controls proliferation of sympathetic neurons in the peripheral nervous system (PNS) (Reiff et al., 2011). In zebrafish (Danio rerio), a closely related co-ortholog of Alk, leukocyte tyrosine kinase (ltk) encodes a 1530 amino acids (aa) long protein [17]. ltk is expressed in neural crest cells (NCCs) and iridophores, a pigment cell type widespread in anamniote vertebrates. ltk is mutated in zebrafish shady mutants, which are characterized by the complete lack of iridophores and the loss of a subset of NCCs by apoptosis. Further analysis revealed that zebrafish ltk controls fate specification in a distinct subset of NCCs derived from the multipotent premigratory NCCs [17].
In the present study, we identified the full length zebrafish alk sequence and show that alk is expressed in the developing zebrafish CNS, that its over-expression results in increased proliferation of neural progenitors and that it is required for neuronal differentiation and survival. This is the first study that investigates Alk function is zebrafish and shows a critical role for Alk in the embryonic CNS.

Cloning of full-length zebrafish alk
The 59UTR and N-terminal zebrafish alk sequences were obtained by 59RACE (Rapid Amplification of cDNA Ends) using a kit from Invitrogen. The full length alk coding sequence was then PCR amplified with Phusion polymerase (Finnzymes), primers ALKatgXhoI (59-CCGCTCGAGCCACCATGTGTGA-TAACGCAGCAGAG -39) and ALKdownXbaI (59-TGCTCTA-GATTACAGCACAGTGGCGTTGT -39), and cloned into pCS2+ (for capped mRNA preparation) or alk:HSE:cfp (for generation of transgnic fish, see below) using XhoI and XbaI sites present in these vectors.

Generation and maintenance of transgenic fish
All animal experiments were performed in accordance with approved IACUC protocols of the National University of Singapore (protocol numbers 075/07; 082/10; BR19/10). Zebrafish inbred AB wild-type strains were crossed to obtain embryos that were raised in 30% Danieau's solution at 28uC and staged as described previously [18]. For generation of transgenic fish, a alk:HSE:cfp construct was generated that contains the full-length zebrafish alk and cfp under control of a bidirectional heat-shock promoter in addition to flanking I-SceI sites [19,20]. HSE:cfp was used as control. Plasmid DNA and I-SceI meganuclease (New England Biolabs) were co-injected into AB wild-type embryos at the one-cell stage as described [21]. On the next day, injected embryos were heat shocked for 1 hr at 39C. Embryos with strong CFP expression were raised to adulthood and crossed to WT AB fish. For screening and maintaining these lines, embryos were heat-shocked after 24 hpf to prevent early embryonic defects caused by alk overexpression. F2 or F3 embryos of two independent alk overexpressing lines (alk:HSE:cfp 1 and al-k:HSE:cfp 2 ) and the HSE:cfp control line were used in the experiments described below.
Capped mRNAs were synthesized in vitro using the mMES-SAGE mMACHINE SP6 Kit (Ambion). NPM-ALK mRNA was injected at 50 ng/ml. alkATG-egfp mRNA was injected at 50 ng/ ml. To test the translation blocking efficiency of the alk ATG-MO, a target construct alkATG-egfp was generated that contains the alk 59UTR and the ATG-MO target site, in addition to a 6 bp XhoI linker, in-frame with the egfp coding sequence.

RNA in situ hybridization, immunostaining and Western blotting
Whole mount RNA in situ hybridization and immunostaining were performed as described earlier [22]. An antisense riboprobe comprising the alk sequence from nt 6-885 was used to detect alk expression, and a sense probe covering the same sequence was used as control. The following primary antibodies were used: antiphospho-Histone 3 (anti-pH 3; Millipore); anti-HuC/HuD (anti-HuC/D; Molecular Probes); anti-ERK1/2 (Santa Cruz); and antiphospho-ERK1/2 (anti-pERK; Cell Signaling). For detection, either the VECTASTAIN Elite ABC Kit (Vector Labs) for diamino benzidine (DAB; Sigma) staining was used, or Alexa Fluor-488/568 coupled secondary antibodies (Invitrogen) for fluorescence detection. TUNEL assays were performed using the ApopTag Peroxidase In Situ Apoptosis Detection Kit (Millipore) or with an anti-DIG primary antibody (Roche) for fluorescence detection. For 5-bromo-2-deoxyuridine (BrdU) labelling of cells in S-phase, embryos were incubated in 10 mM BrdU (Sigma) for 15 min on ice before fixation, followed by detection with an anti-BrdU primary antibody (DSHB, University of Iowa) as described previously [23]. For Western analysis, zebrafish embryos were deyolked and proteins extracted as previously described [24]. Western blotting was performed with anti-ERK and anti-pERK antibodies using standard protocols.

Alk inhibitor treatment
Dechorionated embryos were incubated with the pharmacological Alk inhibitor CEP-26939 [Teva Pharmaceuticals, also known as cmpd 13; 25] at 40 mM with 1% DMSO in Danieau's solution. 1% DMSO in fish medium was used as control. Embryos were treated from 24 hpf to 60 hpf, with changes of medium every 12 hours, and fixed at 60 hpf for further processing.

Microscopy, cell counting and statistical analysis
Confocal images were taken with a Zeiss LSM 510 Meta laser scanning microscope. All other images were taken with a Nikon SMZ1000 or a Nikon ECLIPSE 90i microscope. pH 3 positive cells were counted in confocal projections of 20 continuous optical sections representing a 50 mm thick area within rhombomeres r4 to r6. For cell counting, BODIPY-Texas-Red (Molecular Probes, Invitrogen) was used to counterstain cell membranes of HuC/D positive neurons. Embryos were stained in 20 mM BODIPY-TR for 1 h prior to imaging [26,27]. Statistical analysis was done with GraphPad Prism 5. Unpaired two-tailed Student's t-test was used in comparisons between treated and control groups, with p,0.05 indicating statistical significance. Bar graphs were plotted with GraphPad Prism 5, showing means with standard error means (SEM) as error bars.

Identification of full-length zebrafish alk
A partial zebrafish anaplastic lymphoma kinase (alk) sequence was available in the NCBI database (GenBank: XM_686872.2). It is located on chromosome 17 (Ensembl: ENSDARG00000095833) and comprises a putative 4164 base pairs (bp) transcript encoding 1387 amino acids (aa). We used 59RACE to identify a 216 bp 59 UTR that contains several STOP codons in all three reading frames, and the predicted translation start site of zebrafish alk. The full-length zebrafish alk coding region contains 4167 bp encoding a 1388 aa protein. The zebrafish Alk tyrosine kinase (TK) domain (aa 795-1062) has 89.6% aa identity to that of human ALK (aa 1116-1383; GenBank: NM_004304.4). Key tyrosine residues are present in zebrafish Alk (Y957, Y961 and Y962) that correspond to the human ALK kinase domain activation loop, a YXXXYY motif. Notably, this motif presents as YRASYY in all vertebrate Alk proteins, and slightly modified as YRSDYY in the fruit fly [28]. Moreover, three other residues in the TK domain: F1174, F1245 and R1275 that are most frequently affected by gain-offunction mutations in human neuroblastoma [8,11] are also conserved in zebrafish Alk (F853, F924 and R954, respectively). Together, these data suggest a critical role for the highly conserved Alk residues in zebrafish.

Zebrafish alk expression during embryonic neurogenesis
We examined Alk mRNA transcript expression by RT-PCR at developmental stages from 6 hours post fertilization (hpf) to 72 hpf. Alk transcripts are first detected at 12 hpf, and expression levels continuously increased with the most profound increase at 48 hpf (Fig. 1A). The appearance of alk expression shortly after gastrulation corresponds to the initiation of segmentation and neurulation (10 hpf; [18]. Neurogenesis in the zebrafish hindbrain peaks between 24 hpf and 48 hpf, when the neuron to progenitor ratio changes most significantly [29]. The increase of alk mRNA levels from 12 hpf to 48 hpf therefore coincides with neurogenesis in zebrafish. In adult tissues, alk is highly expressed in brain, with much lower expression in heart, caudal fin, as well as testis, and none detectable in liver (Fig. 1B). Abundant expression of alk in adult brain suggests that it may function in adult neurogenesis and neural regeneration, which is prominent in zebrafish [30].
To analyze the spatial control of alk expression, whole mount in situ hybridization was performed with two antisense alk riboprobes representing either the alk N-(nt 6-885) or C-terminus (nt 3393-4071) in embryos at 24, 48 and 72 hpf. Both probes showed identical patterns (Fig. 1C-K; Fig. S1; and data not shown). Sense probes were used as negative control and showed no staining (Fig. S1). As early as 24 hpf, alk transcripts were broadly expressed in the developing central nervous system (CNS) including the spinal cord, with elevated levels in the brain (Fig. 1C). At 48 hpf, strong expression was found in the mid-and hindbrain (Fig. 1H), which remained high at 72 hpf (Fig. S1H,I). Transverse sections of embryos at 24 ( Fig. 1D-G) and 48 hpf ( Fig. 1I-K) at different brain levels showed that alk is expressed in the entire brain, with higher expression in ventral than dorsal regions of the early forebrain (Fig. 1D) and midbrain (Fig. 1E). In the hindbrain, expression levels were higher in regions where differentiated neurons locate rather than in regions with newborn progenitors (Fig. 1F,K), especially at 48 hpf [29].

Increased neural proliferation and aberrant neurogenesis in embryos ectopically expressing alk
We generated transgenic alk:HSE:cfp zebrafish lines that simultaneously co-express CFP and wild-type Alk under control of a bi-directional heat-shock promoter (Fig. S2) [19]. A control line with only the cfp sequence was also established (Fig. S2A).
Embryos showing no CFP signal after heat shock (transgenic negative siblings (Sib); arrows in Fig. S2) were used as negative controls for comparison. In situ hybridization at different time points after heat shock at 24 hpf showed elevated levels of alk mRNA, detectable as soon as 1 hour post heat shock (hph; Fig.  S3A), degraded over time (Fig. S3B), and eventually returned to the baseline alk concentration at 22 hph (Fig. S3C). A second heat shock performed at 48 hpf reactivated the transgene (Fig. S3D). CFP negative embryos showed no additional alk RNA (arrows in Fig. S3), confirming that they were transgenic negative siblings. Notably, after a 1.5 hour heat shock at 39.5uC at 10.5 hpf, all transgenic alk expressing embryos in both lines showed marked morphological malformations of the brain (Fig. S2D,E), similar to embryos injected with human NPM-ALK mRNA (data not shown). In contrast, HSE:cfp control embryos showed no defects with the same treatment (Fig. S2C), suggesting that the observed defects are caused by overexpressed Alk.
Defects in embryos heat shocked at 10.5 hpf were investigated at 24 hpf. For quantification of cell numbers in the experiments outlined below, we analyzed the hindbrain between rhombomeres r4 and r6, using the otic vesicles as landmarks in transverse sections (Fig. 2). Importantly, similar findings were also observed in other regions of the CNS where endogenous alk is expressed at 24 hpf ( Fig. 1C; data not shown). In alk-overexpressing embryos (Tg+), the 4 th ventricle, a normally T-shaped opening in the hindbrain (asterisks in Fig. 2A,B,D), was completely absent. Instead, several small cavities were observed (arrows in Fig. 2E,G, with high magnification in the inset E9). This suggested early neural tube defects, since the 4 th ventricle, like other brain ventricles, is formed by proliferation and differentiation in the early neuroepithelium [18]. Accordingly, proliferation is abundant at this stage in the zebrafish hindbrain [29], and cell divisions occur almost exclusively at the most apical ventricular surface [29,31]. In control siblings, proliferating cells positive for phosphohistone 3 (pH 3) were observed at the apical surface as expected ( Fig. 2A   complementary to pH 3 and HuC/D positive cells (Fig. 2H). In alk:HSE:cfp 1 Tg+ embryos, S-phase cells were not restricted to these positions but scattered throughout the hindbrain (Fig. 2I). The total area with cells in S-phase was drastically expanded, which made cell counting impossible. Interestingly, an increase of cells in S-phase has also been reported in primary sympathetic ganglia neuron cultures from embryonic chicken transfected with human ALK gene [15].
To test whether the observed increased proliferation is associated with upregulation of the cell cycle machinery, we evaluated the expression of cyclin D (ccnd1), which is known to promote proliferation by regulating cyclin dependent kinases CDK4 and CDK6 [32]. cyclin D is overexpressed in many cancers [32] including ALK positive neuroblastomas [5]. In non-transgenic control embryos, ccnd1 expression was observed in two stripes of the dorsal hindbrain (Fig. 2J). In contrast, alk:HSE:cfp 1 Tg+ embryos exhibited an expansion of ccnd1 expression into other regions of the hindbrain (Fig. 2K).

Overexpression of alk activates MAPK signaling in vivo
Activation of several signaling pathways has been reported in human ALK-related cancers including MEK/ERK pathway involved in cell proliferation [5]. In Drosophila, Alk functions through ERK activation during visceral mesoderm development [33,34,35,36]. To test whether the MEK/ERK pathway is affected by the alk overexpression in zebrafish, we examined phosphorylation of ERK1/2 (pERK) and expression of total ERK levels using anti-pERK and anti-ERK antibodies, respectively ( Fig. 3; Fig. S4).
In 24 hpf WT embryos without heat shock, increased pERK concentration could only be found in forebrain, mid-hindbrain boundary, caudal-most hindbrain, spinal cord (Fig. 3A), as well as the tail bud (image not shown). There were only few individual pERK positive cells in the hindbrain (Fig. 3A). This distribution was found in both Sib and Tg+ embryos of the HSE:cfp line, with no additional pERK signals (Fig. 3B,C). In cross sections at the level of r5, usually no pERK positive cells could be seen (Fig. 3A9,B9,C9). This demonstrated that neither heat shock nor overexpression of CFP activated ERK phosphorylation. In contrast, in Tg+ embryos of alk:HSE:cfp lines, pERK was strongly expressed across the entire CNS albeit in a mosaic pattern (Fig. 3E,E9; Fig. S4D,D9), while the total ERK concentration and distribution were similar to the control Sib line (Fig 3F,F9,G,G9). Consistently, Western blot analysis showed increased pERK levels but comparable ERK expression in Tg+ alk:HSE:cfp embryos at 22 hpf when compared to Sib controls (Fig. 3H). The fusion protein NPM-ALK is a known highly oncogenic form of ALK capable of activating MEK-ERK pathway (Chiarle et al., 2008). Accordingly, embryos at 11 hpf injected with human NPM-ALK mRNA showed increased pERK compared to WT controls, with similar levels of total ERK protein (Fig. 3H). Taken together, overexpression of alk causes up-regulation of pERK while total ERK levels remain unchanged. Ubiquitous MEK/ERK activation in the CNS upon alk overexpression is consistent with the observed deregulation of cell proliferation and aberrant neurogenesis.
Increased apoptosis but normal cell proliferation in the hindbrain after alk knock-down Morpholino oligonucleotides capable of blocking zebrafish alk translation (ATG-MO) and a pair of splice blocking MOs (Spl-MO) were designed to knock-down alk (Fig. S5A). As shown in   . S5B, the alk ATG-MO suppressed GFP expression from an alkATG-egfp target construct, while a corresponding alk ATGmismatch-MO with 5 nt substitutions failed to block GFP expression (Fig. S5C-F). Two splice MOs were designed to prevent splicing of intron 3, which results in a 289 bp insertion leading to frame shift with nine possible premature STOP codons (Fig. S5G). RT-PCR with RNA from alk Spl-MO injected embryos at 24 hpf showed a size increase as expected (Fig. S5H).
alk ATG-MO or Spl-MO were injected into zebrafish embryos at the one-or two-cell stage, and apoptosis was examined at 22 hpf using a transferase mediated dUTP nick end labeling (TUNEL) assay. This showed increased apoptosis in the CNS (Fig. 4B,C,E;  Fig. S6C,F) when compared to uninjected WT ( Fig. 4A; Fig.  S6A). Also, the size of the neural tube appeared smaller after alk knock-down when compared to WT (Fig. 4A). Neither a standard control MO nor the alk ATGmismatch-MO resulted in increased apoptosis (Fig. S6B,E), suggesting that the increased apoptosis was specific to alk deficiency. To further exclude MO off-target effects due to p53 activation [37], a p53 MO [38] was co-injected with alk MOs. This did not rescue the apoptosis defects caused by alk ATG-MO or alk Spl-MO (Fig. S6D,G) indicating that the alk MO knock-down effects are specific. Thus, alk depletion seems to promote apoptosis in the CNS independent of the p53 pathway. Interestingly, increased apoptosis was almost exclusively observed in the CNS, where endogenous alk is expressed broadly during early stages (24 hpf; Fig. 1C) thus further supporting the specificity of the observed defects.
We also tested whether alk inhibition affected cell proliferation. In embryos injected with either alk ATG-MO or Spl-MO, pH 3 positive cells were positioned normally (Fig. 4B,C). Counting of pH 3 positive cells within the r4 -r6 hindbrain region (Fig. 4D) revealed that neither alk ATG-MO (23.80 6 0.86) nor alk Spl-MO (24.60 6 0.93) resulted in a significant difference when compared to WT (25.00 6 0.84). In contrast, within the same samples, counting of TUNEL positive cells (Fig. 4E) showed that both alk ATG-MO (40.60 6 2.93) and alk Spl-MO (44.00 6 4.56) injected embryos had significantly increased number of apoptotic cells as compared to controls (5.40 6 1.44). These findings suggest that alk deficiency may not affect dividing cells in M-phase. Since the majority of cell divisions at this stage represent proliferative neural progenitors, the increased apoptosis observed in alk morphants most likely does not affect these progenitor cells, but rather other, possibly more mature neuronal cell types.

Knock-down of alk impairs neuronal differentiation and reduces neuron numbers in the hindbrain
As direct targets of Notch signaling, Hes genes (Hes1 and Hes5 in mammals) are abundantly expressed in undifferentiated neural progenitors and essential for maintaining stem cell identity [39]. We found no expression change of hairy-related 6 (her6), the zebrafish Hes1 homolog [40], in alk morphants (Fig. 5B,B9,B0; Fig. S7B) suggesting that proliferation of undifferentiated neural progenitors is not affected by alk knock-down. In differentiating neural cells, proneural gene expression is up-regulated as Notch-Hes signaling is switched off. Combined expression of different proneural genes restricts neuronal lineages, forces cell cycle exit, initiates neuronal differentiation and promotes neuron survival [41,42,43]. Expression of several zebrafish proneural genes was analyzed to study the effect of alk knock-down on neuronal precursor differentiation. Neurogenin 1 (neurog1, neurod3) expression was unchanged or only slightly reduced in alk morphants (Fig. 5D,D9,D0; Fig. S7D). On the other hand, expression of the zebrafish atonal ortholog, neurogenic differentiation 4 (neurod4, zath3) was significantly reduced throughout the CNS, including the hindbrain (Fig. 5F,F9,F0; Fig. S7F). The achaete-scute complex like (ascl or ash) genes belong to another proneural gene family that can act independently from atonal-related genes in vertebrates [44]. Expression of ascl1b was also reduced in alk morphants (Fig. 5H,H9,H0; Fig. S7H). Moreover, delta A (dla), a Notch ligand highly expressed in differentiating neurons, was also reduced (Fig. 5J,J9,J0; Fig. S7J). Finally, glial fibrillary acidic protein (gfap) expression was not changed in alk morphants (Fig. 5L,L9,L0;  Fig. S7L). Of note, gfap is expressed in radial glia cells that overlap with neural progenitors at this embryonic stage. All of the genes mentioned above were expressed normally in alk ATGmismatch-MO injected embryos (Fig. S7A,C,E,G,I,K). Taken together, we observed a down-regulation of several proneural genes in alk morphants. At the same time, progenitor marker expression was unchanged and cell proliferation was not altered. This indicates that progenitor cells do not require alk activity. Instead, we propose that alk knock-down affects early neuronal precursors that fail to differentiate and instead undergo programmed cell death.
MOs elicit their effects from earliest embryonic stages onward. To achieve alk deficiency at later embryonic stages, a pharmacological ALK inhibitor, CEP-26939 was used. Both CEP-26939 and its closely related analog ALK inhibitor, CEP-28122, have been reported to lead to growth inhibition in a variety of cultivated human ALK-related cancer cell lines, including neuroblastoma, through caspase 3/7 activation and apoptosis [25,45]. In our experiment, zebrafish embryos were incubated in fish medium containing CEP-26939, or DMSO alone as a negative control, starting at 24 hpf. Embryos were fixed at 60 hpf and examined for her6, neurog1, neurod4 and ascl1b expression, as well as apoptosis (Fig. 6). Similar to the findings in Alk morphants, TUNEL staining revealed increased apoptosis in the CNS of the ALK inhibitor treated embryos (Fig. 6A,B), which was also confirmed by Acridine Orange staining in living embryos at 48 hpf (data not shown). Consistent with the MO experiments, no difference in her6 expression was seen between inhibitor-and DMSO-treated groups (Fig. 6C,D). Furthermore, only a weak reduction of neurog1 expression was observed with no major change in its overall pattern (Fig. 6E,F). On the other hand, expression of neurod4 and ascl1b was strongly reduced and almost completely absent in the mid-and hindbrain in CEP-26939 treated embryos (Fig. 6G,H,I,J). Significantly, in all regions where neurog1, neurod4 or ascl1b expression was reduced, the number of apoptotic cells was strongly increased. Thus, this pharmacological ALK inhibitor fully recapitulated the effects observed after MO-mediated ALK depletion providing further evidence that ALK activity is required for the differentiation and survival of early neural precursors in the developing CNS.
The observed reduction of proneural gene expression suggested compromised neuronal differentiation and survival. To test to what degree alk deficiency affects the final steps of neurogenesis, i.e. the formation of fully differentiated neurons, expression of the pan-neuronal marker HuC/D [46,47] was analyzed. In the hindbrain at 31 hpf, areas where postmitotic HuC/D+ neurons are located were significantly reduced in size in alk morphants (Fig. 7B,B9,C,C9) when compared to WT (Fig. 7A,A9) or ATGmismatch-MO injected embryos (Fig. 7D,D9). In contrast, proliferative ventricular and subventricular zones appeared normal, when BODIPY-Texas Red was used to label cells (Fig. 7E), an average of 92.25 6 2.94 HuC/D positive neurons was counted per confocal single plane in controls (representing a 5 mm optical section). Cell numbers in the alk ATGmismatch-MO injected group (85.14 6 2.71) were not significantly different from WT. In contrast, numbers of HuC/D positive neurons were reduced to 50.50 6 3.43 in alk ATG-MO injected embryos, and 46.36 6 2.64 in alk Spl-MO injected embryos. In conclusion, alk ATG-MO or alk Spl-MO injected embryos show a significant reduction in the number of differentiated neurons in the hindbrain when compared to WT. Thus, alk is required for differentiation and survival or early neural precursors in order to maintain the correct number of fully differentiated hindbrain neurons at later stages.

Discussion
This is the first study reporting structure and functional characterization of zebrafish alk. It is also the first analysis of the physiological role of Alk in the vertebrate central nervous system (CNS). We show that zebrafish alk is expressed in the developing CNS and is required for differentiation and survival of neural cells. The full length zebrafish alk transcript encodes a deduced 1388 amino acids (aa) protein that matches a predicted partial zebrafish alk sequence in the NCBI database (GenBank: XM_686872.2), except an additional Glutamine (Q1280) in a Q (13) repeat located at the C-terminal end, and a Serine (S309) to Alanine (A309) replacement in a S (4) repeat located in its N-terminus. These differences most likely reflect polymorphisms between different zebrafish wild-type strains. Interestingly, the N-terminus of zebrafish Alk (aa 1-708) shares only 25.3% aa identity with its human counterpart (aa 1-1029) and seems to be significantly shorter, while the TK domain shows 89.6% identity. Similarly, chicken Alk (GenBank: XM_419364.3) also has a conserved TK domain (aa 970-1237; 95.5% identical to human), but a much shorter N-terminus (aa 1-883) with only 53.3% identity. When Alk sequences from 17 vertebrate species and Drosophila melanogaster were compared, the TK domain was the most conserved part among all Alk's. In contrast, the Alk N-terminus showed much more variation, consistent with the idea that the N-terminus of Alk has undergone extensive divergence during vertebrate evolution. In mammals, ALK and LTK can be clearly distinguished by their extracellular domains, with respect to its length (1030 aa vs. 424 aa) and the presence or absence of MAM domains, respectively [48,49]. Phylogenetic analyses suggested that the MAM domain is a feature of a common Alk/Ltk ancestor, and has been retained in Ltk in zebrafish and chicken but not mammals [17]. It is likely that Ltk and Alk evolved their N-termini differently in mammals and in teleosts. As a consequence, mammalian Ltk lost some of the fish specific Ltk functions, such as for iridophore development [17].

Expression of alk in the central nervous system
In mouse, Alk is prominently expressed in the diencephalon, the neuroepithelium surrounding the fourth ventricle, the ventricular and subventricular zone of the cortex, the midbrain, and the medulla oblongata from day 10.5 to day 16.5 [14]. Similar to mouse, alk is expressed in most of the central nervous system (CNS) in zebrafish. However, in mouse and chicken, Alk expression is also detected in the spinal cord [12,14]. In contrast, zebrafish alk could not be detected in the spinal cord under the experimental conditions used, especially after 24 hpf (Fig. 1C,H). This could possibly reflect species-specific differences in Alk expression among different vertebrates. In mouse, rat, and chicken, Alk is also expressed in the peripheral nervous system (PNS) including the trigeminal ganglia, facial ganglia, vestibulocochlear ganglia, inferior ganglia of the vagus nerve, as well as dorsal root ganglia and sympathetic ganglia [12,13,14]. In our study, we could not detect expression of zebrafish alk in the PNS. On the other hand, the similar spatiotemporal expression patterns of Alk in the early CNS in mouse [14], chicken [12], and zebrafish (this study) suggest that the activity and function of the highly related Alk tyrosine kinase domain is conserved during CNS neurogenesis.

In vivo activity of over-expressed, non-mutated Alk
Receptor protein tyrosine kinases (RTKs) can undergo spontaneous dimerization, transphosphorylation of monomers, and activation. This process is promoted and stabilized by the presence of RTK ligands [50,51]. In our alk overexpression experiments, Alk proteins could have spontaneously formed active dimers, resulting in RTK activation. Such autophosphorylation of ALK in absence of any ligand was also reported in cell culture [52]. The observed activation of MEK/ERK pathway confirmed that the transgenic Alk proteins are expressed and enzymatically active. Interestingly, we observed a high degree of mosaicism of MEK/ ERK activation although the Alk transgene was ubiquitously expressed. While the exact reason for this remains unclear, it suggests a dynamic regulation of intracellular signaling when Alk is activated in this in vivo context. Overexpression of alk in these transgenic lines resulted in enhanced cell proliferation with increased numbers of cells in M-phase and up-regulation of ccnd1 expression. Furthermore, aberrant neurogenesis patterns were observed with ectopic localization of differentiated neurons. We therefore propose that ectopic activation of Alk leads to excess MAPK activity resulting in increased proliferation of neuronal progenitors, randomized cell cycle exit and differentiation, and consequently mis-positioning of differentiating neurons. Triggering alk overexpression at a later stage (24 hpf) also enhanced proliferation but strikingly to a much lesser extent and in a spatially more restricted fashion (data not shown) when compared to overexpression at 10.5 hpf (Fig. 2). Recent studies suggested that overexpression of mutated ALK works cooperatively with MYCN overexpression to promote the severity of induced neuroblastoma [53,54]. Interestingly, zebrafish nmyc1, the ortholog of human MYCN, is expressed ubiquitously at high levels in the CNS before 24 hpf, but becomes restricted thereafter [55]. This could explain why proliferative effects of overexpressed Alk were higher at earlier embryonic stages.

Alk is required for differentiation and survival in the CNS but not proliferation
We show that an alk knock-down does not affect cell proliferation in the hindbrain at 24 hpf. Interestingly, decreased cell proliferation was reported in in vitro cultured neuroblastoma cell lines where ALK has been depleted by siRNA [8,11]. Also, a recent report using cultured chicken sympathetic neurons derived from PNS sympathetic ganglia showed that the knock-down of endogenous chicken Alk results in reduced proliferation [15]. This suggests possible differences of Alk function in different tissues and cell types. Alternatively, it could also reflect differences between in vivo and in vitro cell culture conditions suggesting more complicated scenarios that underlay Alk mediated cell cycle control, such as impact of the microenvironment or functional redundancy with other kinases that cannot be fully mimicked in cell culture settings.
In alk deficient embryos, generated by MO-mediated ALK depletion or pharmacological inhibition of its enzymatic activity, a universal increase in neural cell apoptosis and reduction in expression of selected proneural genes was observed. Since progenitor marker expression was unchanged and cell proliferation was not affected in hindbrain, this indicates that progenitor cells are not impaired by alk deficiency and that the observed apoptosis does not originate from neural progenitors. On the other hand, reduced proneural gene expression in differentiating neuronal precursors suggests defects in differentiation eventually leading to cell death. There are several possibilities how alk deficiency could affect cell survival. First, differentiating neural cells could undergo apoptosis because Alk is a survival factor and functions as a repressor of cell death. Thus, reduced proneural gene expression could simply reflect a reduction in the numbers of differentiating cells. Alternatively, Alk could be a specification factor, which is required to promote proneural gene expression. In morphants, cells that have exited the progenitor state would therefore fail to differentiate and consequently commence apoptosis. It remains unclear at present whether Alk induced apoptosis is the cause or consequence of impaired neurogenesis. However, our study showed that the extent of apoptosis in embryos deficient for Alk is relatively mild compared to the extensive changes in neurogenic gene expression. This observation favours a scenario where Alk deficiency leads to impaired neurogenesis, which subsequently results in apoptosis.
Unfortunately, Alk knock-out animals other than the embryos presented here are lacking in order to address which of the scenarios explains increased apoptosis after alk knock-down. Interestingly and in agreement with a role as specification factor, the Alk co-ortholog Ltk is involved in cell fate specification of multipotent precursors in the iridophore lineage (Lopes et al., 2008). On the other hand, siRNA-based approaches recently reported increased apoptosis upon Alk knock-down in neuroblastoma cell lines [56] supporting the idea of Alk as a survival factor. It is well established that proneural genes are required for cell survival during neuronal differentiation [57,58,59]. Based on this and our own observations, a combination of both scenarios therefore seems likely. We propose that Alk is required for the survival of neural precursors that have exited the progenitor state and are about to enter different neural differentiation fates.  Figure S3 In situ hybridization to detect alk overexpression in alk:HSE:cfp transgenic embryos. A heat shock was performed at 24 hpf. All in situ hybridizations were done using an alk antisense RNA probe, at 1 hour post heat shock (hph) (A), 6 hph (B), 22 hph (C), and shortly after a second HS performed at 48 hpf (D). Inserts show representative transgenic positive embryos. As soon as 1 hour after heat shock, high levels of exogenous alk RNA could be detected in the entire embryo. Reduced staining at 6 hours after HS indicated RNA degradation. Staining almost completely disappeared at 22 hours after HS. A second HS re-activated transcription of overexpressed alk. Arrows points to transgene negative siblings where no exogenous alk was detected. Endogenous alk expression was not detected because of short staining time.