There was initially no information related to tcof1 gene in zebrafish genomic databases. Therefore, we began our analysis using the tcof1 cDNA sequence from Xenopus laevis (GenBank accession number AY731504) to search for homologous sequences in the Danio rerio cDNA_ALL Ensembl database (tblastx, Assembly Zv9). Two alignments with a high percentage of identity were located on a region of the minus strand of chromosome 13 (Table S1 and Figure 1A; Chromosome13: 4,655,685–4,665,683). This region of chromosome 13 contains the Ensembl predicted genes B8JIY2_DANRE, Q7ZUM1_DANRE, and nucleolar and coiled-body phosphoprotein 1-like or nolc1l^[19]. Q7ZUM1_DANRE and nolc1l sequences partially overlap with the complete region covered by the B8JIY2 sequence (Figure 1A). On the contrary, B8JIY2_DANRE consists of 14 exons that span 19,250 bp in the zebrafish genome (Figure 1A). The predicted protein has 1,001 amino acid residues, a molecular weight of 102,031.33 Da, and an isoelectric point of 9.85. In a similar fashion than in treacle ^[20], ^[21], the most relevant features of the B8JIY2-derived polypeptide are the presence of an amino terminal LisH dimerisation motif, three TCS treacle-like domains and three histone H5 motifs (Figure 1B). More than 100 putative Ser-phosphorylation sites were predicted using the NetPhos 2.0 application ^[22], most of which are located within the clusters of serine-acidic residues present within the repeat units (Figure 1C). Finally, because of the high homology observed between mammalian TCOF1 and Nolc1 we assessed the percentage of similarity among them and the zebrafish B8JIY2. Data revealed higher similarity of B8JIY2 with Tcof1 than with Nolc1 (Table S2). Collectively, these results strongly suggest that B8JIY2 is the TCOF1 ortholog in zebrafish. The B8JIY2 sequence was submitted to Gene Bank as the zebrafish tcof1 gene under submission ID 1430387.

We designed primers to amplify a fragment specific to full-length zebrafish tcof1 cDNA. RT-PCRs combining different sets of oligonucleotides led to the amplification of two transcripts, which were named B8JIY2a and B8JIY2b. In silico analyses revealed that B8JIY2b lacks 103 b of exon 11 (Figure S1). This could be due to either differential pre-mRNA processing or, alternatively, RT-PCR artifacts because the sequence lost in B8JIY2b corresponds to a genome repetitive region.

The following challenge was to analyze the tcof1 expression pattern. By whole-mount in situ hybridization (Figure 2) Tcof1 mRNA was detected ubiquitously in early staged embryos (Figure 2A–C), indicating a maternally inherited origin. Expression was still ubiquitous in 15 hours post-fertilization (hpf) embryos; however, the optic primordium and prospective brain regions (dotted oval in Figure 2D) showed more intense labeling. At 24 hpf, expression was mainly detected in the prospective eye, midbrain, hindbrain, and the NCC population that later forms a portion of the pre-mandibular, mandibular, first and second branchial arches (Figure 2I–J). At 48 hpf embryos, tcof1 expression was detected in the retina, pectoral fin buds, pharyngeal arches, prospective craniofacial structures, and olfactory pits (Figure 2L–M). From 48 hpf onwards, the expression observed in brain was refined into a specific pattern along the anterior, dorsal, and posterior edges of the optic tectum (dotted lines in Figure 2L and P). At 72 hpf, tcof1 expression was observed in the edges of the optic tectum, lenses, and craniofacial structures (arrowheads in Figure 2M, O–P, and dotted oval in R). Later on, 120 hpf-larvae showed expression in the craniofacial structures (arrowheads and dotted circles), lenses, and developing gut (Figure 2S–T). Controls were performed using a tcof1 sense probe (Figure 2E–H, K, N and Q). Together, results revealed that tcof1 expression is restricted to the anterior-most regions of zebrafish developing embryos, similar to what happens in mouse embryos ^[13], ^[15]. Noteworthy, zebrafish tectal tcof1 expression profile was highly similar to those of genes involved in cell proliferation, including proto-oncogenes and cell cycle regulatory genes ^[23]–^[26]. This finding is in agreement with previous report showing a role for Tcof1 in cell proliferation control ^[16].

Developmental tcof1 mRNA distribution was also examined by semi-quantitative RT-PCR (Figure 3). Zebrafish elongation factor 1 α (ef1α) mRNA was used as a control for RNA quality and for setting linear amplification conditions. Semi-quantitative RT-PCR assays confirmed that tcof1 is maternally inherited and expressed at high levels during the first stages of zebrafish embryonic development. The highest tcof1 mRNA expression was observed at the 1-cell stage. Over the course of development, tcof1 mRNA expression decreased reaching a minimum at 12 hpf. From 12 hpf onwards, tcof1 expression increased, reaching a plateau at 48 hpf.

Together, results presented here revealed that the tcof1 spatiotemporal expression pattern is highly dynamic over the course of zebrafish embryonic development. Expression was restricted to regions and developmental stages wherein most of the critical morphogenetic events responsible for craniofacial structure formation take place, including the formation and blending of the branchial arches with the rest of the developing face. Notably, tcof1 expression decreases once the facial complex has formed, and by 48 hpf onwards, expression becomes reduced near to background levels.

TCS patients show mutations in the TCOF1 gene that result in proteins lacking the C-terminal region ^[4]–^[9]. The loss of the C-terminus leads to mislocalized proteins and consequently to treacle loss-of-function ^[20], ^[21]. Importantly, overexpression of C-terminal truncated forms of treacle did not affect the performance of cultured cells, suggesting that TCS results from treacle haploinsufficiency ^[8], ^[12]. In view of this, we reasoned that the most informative approach to assess the role of tcof1 during zebrafish embryonic development would be to decrease tcof1-encoded protein levels by generating C-terminal truncated proteins. To achieve this, we designed two morpholino oligonucleotides (MO; ^[27]) that specifically modify tcof1 pre-mRNA splicing during embryonic development. One of these MOs, in6:ex7-MO, was directed to the junction between intron 6 and exon 7 while the other one, ex7:in7-MO, was directed to the junction between exon 7 and intron 7. Both MOs targeted splice donor or acceptor sites in tcof1 exon 7, thereby preventing its inclusion during the pre-mRNA splicing process. In silico analysis predicted that the loss of exon 7 results in a premature stop codon, which generates C-terminal truncated treacle forms. This strategy had been successfully used in our laboratory to study the role of other proteins during craniofacial development ^[28]. Different amounts of MO were injected in 1-cell staged embryos to achieve detectable phenotypes along with the lowest embryonic lethality. Noteworthy, the injection of in6:ex7-MO alone produced higher lethality than the injection of ex7:in7 MO (not shown). The best results were obtained by co-injecting 10.7 ng of each MO per embryo or 21.4 ng of ex7:in7 MO per embryo (Table 1). Lower amount of MO did not result in detectable phenotypes. Controls were performed as described elsewhere ^[27]. Normal embryos without any phenotype after injection of both the MOs may reflect different genetic backgrounds among fish, similarly to what happens in humans.

Because MO injection may induce neural cell death by off-target activation of p53 and derivatives ^[29], we checked by real time quantitative RT-PCR (qRT-PCR) the transcriptional levels of p53, p21 and the N-terminal truncated p53 isoform (Δ113 p53) mRNAs in MO-treated embryos at 24 hpf. While full-length p53 RNA levels were not significantly increased in any of the MO-injected zebrafish embryos, the expression of p21 was upregulated in MO-injected embryos and virtually absent in controls (Figure 4). In contrast, the expression of Δ113 p53 isoform, the most reliable marker to diagnostic MO off-target effects ^[29], was significantly upregulated in in6:ex7-MO- and ex7:in7-MO- injected embryos but did not changed in embryos co-injected with both MOs or in controls (Figure 4). These findings ruled out a MO off-target effect for the latter experimental condition and, therefore, the following experiments were carried out using embryos injected with both MOs at a concentration of 10.7 ng each per embryo. Why a combination of in6:ex7-MO and ex7:in7-MO does not cause MO off-targeting remains completely unknown.

Typical morphant phenotypes obtained by tcof1 loss-of-function are shown in Figure 5. In 15 hpf morphants, the distinctive optic primordium was malformed, the four prominent subdivisions of the brain, the telencephalon, diencephalon, midbrain and hindbrain, as well as the rhombomeres 2 to 6 were undetectable in a dorsolateral view (Figure 5A–B, black dotted line). At 20 hpf, the otic vesicle was barely detectable, and dark apoptotic zones started to appear in the retina and midbrain-hindbrain ventricles (Figure 5C–D, black dotted arch). Morphants at 25 hpf showed eyes reduced in size and opaque apoptotic zones in all brain structures (Figure 5E–F). The typical internal structures seen for the forebrain, midbrain and hindbrain regions in control embryos were severely misarranged in morphants. Besides, morphants showed the dorsal tip of the developing head reduced in extension, being less prominent than in controls (see red lines and arrowhead in Figure 5E–F). At 5 days post-fertilization (dpf), morphant larvae exhibited eyes reduced in size and a shortened and rounded rather than triangular head. Moreover, morphants showed frontonasal hypoplasia manifested by the absence of the typical protruding mouth in front of the eyes (Figure 5G–J). Larvae exhibited heart edema, gut malformations, and swim bladder not fully inflated (see labels in Figure 5J). It should be noted that during all the developmental stages the somites and posterior-most regions were not affected by the knockdown of tcof1 expression.

We next assessed whether the observed phenotypes were due to the generation of shorter tcof1 mRNAs analyzing by RT-PCR and cDNA sequencing the pattern and identity of tcof1 mRNAs present in 15 hpf morphants and controls. RT-PCRs were performed using specific oligonucleotides that hybridized with exons 6 and 8. Specific oligonucleotides for ef1α were used to confirm sample quality (not shown). Tcof1 mRNA variants that lacked exon 7 were detected in MO-injected embryos. The in silico translated polypeptide consisted of 235 amino acids that perfectly matched with the first 224 amino acids of the tcof1-derived peptide but lacked the C-terminal region (Figure S2). These data strongly suggest that the observed phenotypes result from the reduction of functional protein level due to the generation of a protein isoform that lacks the C-terminal domain.

Studies made in Tcof1^+/− mice indicate that craniofacial abnormalities observed in TCS patients result from defects in NCC generation, proliferation, and/or viability ^[13]. Therefore, we employed whole-mount in situ hybridization to evaluate the expression patterns of the typical NC-specifier genes foxd3 and sox9b in tcof1-knockdown zebrafish embryos (Figure 6). FoxD3 is one of the earliest NC genes to be expressed in murine, zebrafish, frog, and chick embryos ^[30], and sox9b is involved in iridophores and craniofacial skeleton differentiation ^[31]. At 10 hpf, foxD3 expression pattern was clearly reduced in MO-treated embryos compared with controls (Figure 6A–B), although that difference was not so critical at 11 hpf (Figure 6C–D). In 14 hpf morphants, foxD3 expression was reduced in the somites (see dotted lines in Figure 6K–L) and almost undetectable in the dorsal and anterior brain regions (see arrowheads in Fig. 6K–L, and compare to G–H; 80% of morphant-embryos; n = 25). At 11 hpf, sox9b expression pattern was similarly detected in control and MO-treated embryos (Figure 6E–F). In 14 hpf controls, sox9b is expressed in NCCs of the diencephalic, midbrain, and hindbrain regions, as well as in the otic vesicle (Figure 6I–J). Tcof1 knockdown reduced sox9b expression, especially in brain regions, in 73% of analyzed embryos (n = 26). Moreover, expression of sox9b in the otic vesicles was hardly detected in morphants (Figure 6M–N). These results indicate that the protein encoded by tcof1 indeed plays a role in the specification and/or establishment of NCCs in zebrafish.

To extend the study of morphants phenotype, we assessed the consequences of tcof1 knockdown on zebrafish craniofacial cartilage development. Alcian blue staining revealed that morphant larvae developed reduced pharyngeal arches and neurocranial cartilages (Figure 7A–B). Compared to controls (Figure 7A, C, and E), MO-injected larvae were not able to extend anteriorly the medial edge of Meckel's cartilage or to achieve the typical rounded shape, and ceratobranchial arches 1–5 were difficult to distinguish (Figure 7B, D and F). The ceratohyal cartilage was malformed, and the hyosymplectic and palatoquadrate cartilages, including the pterygoid process, were reduced in size. A variation in the angle formed by ceratohyal cartilages was clearly detected. Compared to controls, this angle was more obtuse in MO-injected fish (Figure 7C–D, red lines). In the neurocranium, the anterior edge of the ethmoid plate appeared smaller and dysmorphic in morphants. The injection of higher concentrations of MOs resulted in the complete loss of either all cartilage elements or Meckel's cartilage (not shown). Further examination at higher magnifications revealed that craniofacial chondrocytes of tcof1 knocked down larvae have abnormal morphology. In controls, chondrocytes of the ethmoid plate and trabeculae have an ordered organization, with the appearance of a stack of coins (Figure 7E), whereas in tcof1 morphants, chondrocytes were significantly larger and more rounded in shape (Figure 7F). Taken together, these results demonstrate a phenocopy of the craniofacial abnormalities observed in the Tcof1^+/− mouse and TCS patients, including hypoplasia of the jaw cartilages.

By microarray and real time RT-PCR studies, several genes were found to be positively or negatively regulated by Tcof1 in the mouse NB N1E-115 cell line ^[32]. Among them, ndrg1 was up-regulated while tbx2b and cnbp were down-regulated by Tcof1 knockdown ^[32]. Two genes Mapk14 and Ddx42 were found to be downregulated with knockdown of Tcof1 but this was not further confirmed by quantitative real time RT-PCR ^[32]. In view of this, we wondered if the tcof1 loss-of-function affects the expression of such target genes in zebrafish developing embryos. The expression of putative tcof1 targets was assessed by qRT-PCR using total RNA purified from control and tcof1-depleted embryos staged at 24 hpf. Specific oligonucleotides were designed for each putative target as indicated in Material and Methods. In agreement with previous results ^[32], ndrg1 expression significantly increased, and tbx2b, ddx42 and cnbp expression significantly decreased in tcof1-depleted embryos (Figure 8). In our experimental conditions, mapk14b expression was the only one that did not significantly changed ^[32].

Adult zebrafish were maintained at 28°C on a 14-h light/10-h dark cycle as previously described (Westerfield et al 1995). All embryos were staged according to development in hpf or dpf at 28°C (Kimmel et al. 1995), as well as handled in compliance with relevant national and international guidelines. Protocols were approved by the Commission of Bioethics, Facultad de Cs. Bioquímicas y Farmacéuticas-UNR, which has been accepted by the Ministerio de Salud de la Nación Argentina (http://www.saludinvestiga.org.ar/comites.asp?num_prov=13).

Total RNA from embryos at different developmental stages was obtained using TRIZOL® Reagent (Invitrogen) following the manufacturer's instructions. Purified RNA was incubated with RQ1 DNAse (Promega) and used to perform RT-PCR. Total RNA was retro-transcribed with SuperScript II enzyme (Invitrogen) and either poly-dT or sequence-specific oligonucleotides. For B8JIY2 cDNA cloning, combinations of seven forward primers and six reverse primers were employed. The sequences of the oligonucleotides were: Fwd1_ B8JIY2 ([gene: AGCCACGTCCAAACCAGTG]), Fwd2_ B8JIY2 ([gene: TCAGACGAGGAACCGAAAAAG]), Fwd3_ B8JIY2 ([gene: CTCAGAAGAAGCAGGACAGCAG]), Fwd4_ B8JIY2 ([gene: CTCCGATGAGGAAGATGCTC]), Fwd5_ B8JIY2 ([gene: GTACTCGTTTCTCGTGGAAAACA]), Fwd6_ B8JIY2 ([gene: CAGGACAGCACGGTTCCTA]), Fwd7_ B8JIY2 ([gene: GGCGGAGGAATCTTCATCTT]), Rvs1_ B8JIY2 ([gene: GGTGTTTTAGACGGCTCCTCTTC]), Rvs2_ B8JIY2 ([gene: CTGAACTACTGCTCTCCGCTTTC]), Rvs3_ B8JIY2 ([gene: TAGGAGTACTCACTGGTTTGGAC]), Rvs4_ B8JIY2 ([gene: CTGTCCGAGGCTGAGCTAGA]), Rvs5_ B8JIY2 ([gene: ATCACTAGAGCTGTCCTCACTGC]), Rvs6_ B8JIY2 ([gene: GACTGGAGCTTGAGCCGTCAG]). The isoforms generated by MO treatment were analyzed with a specific B8JIY2 exon 6 forward primer ([gene: CTCTAGAAGAAGCAGGACAGCAG]) and an exon 8 reverse primer ([gene: GACTGGAGCTTGAGCCGTCAG]). Products were resolved in 1.2% (w/v) agarose gels stained with Gel Green, cloned into pGEM-T Easy Vector System (Promega), and their sequences obtained by University of Maine DNA Sequencing Facility. Ef1α RT-PCR was performed as a control and for setting conditions ^[47], which were established at 24 amplification cycles. For qRT-PCR, first strand cDNA was reverse-transcribed from total RNA purified from 24 hpf embryo as described above and using oligo(dT) primer. Reactions were performed using four different RNA purifications and three independent experiments using an Eppendorf realplex2 and standard temperature protocol were performed. Primer sequences for ef1α, rpl14, p53, p21 and Δ113 p53 were used as described elsewhere ^[29]. The resting primer sequences were obtained using Primer3 program, as follows: Ndrg1f, [gene: GACCAAAACCACACTGCTAAAGAT]; Ndrg1r, [gene: GAATGAAGTACTTGAAAGCCTCTG]; Tbx2bf, [gene: CTTTCATCTGTCTCAACACATGCT]; Tbx2br, [gene: GTGTAGGGGTACGGGAAAAGTC]; Mapk14bf, [gene: CTCATCCGTATTTCGCTCAGTATC]; Mapk14br, [gene: ATACGTCAGACTTTTCCACTCCTC]; Ddx42f, [gene: AGCTCTTGGCAACTATGAAACCT]; Ddx42r, [gene: CGACAAAGTGGTTTTTGTACTGAG]. Expected PCR products span at least one intron (except the Δ113 p53 fragment), to ensure amplification solely from the cDNA and not from genomic DNA. Ef1α and rpl14 were used as endogenous control for normalization analysis. Acquired data were analyzed using qBase software version 2.2, and p-values were obtained from t-student analysis.

cDNAs were sequenced by the DNA Sequencing Facility of University of Maine. Complete mRNA and gene organization analysis were obtained from the www.ensembl.org website. Sequence alignments and similarity indexes calculations were carried out using ClustalW2 version 2.0.12 (http://www.ebi.ac.uk/Tools/msa/clustalw2/). The B8JIY2 sequence was submitted to Gene Bank as the zebrafish tcof1 gene under submission ID 1430387.

Embryos were staged and fixed overnight in 4% (w/v) paraformaldehyde (PFA) in 1× PBS at 4°C. After washing, embryos were stored in methanol at −20°C until use. Whole-mount in situ hybridization was carried out as previously described ^[48]. Digoxigenin-UTP-labeled sense and anti-sense riboprobes were prepared according to the manufacturer's instructions (Roche Diagnostics, Mannheim). B8JIY2 exon 11 was used as the detection probe. FoxD3 and sox9b expression patterns were analyzed as described elsewhere ^[25].

Embryos were obtained by natural mating and injected at the one-cell stage into the yolk immediately below the blastomeres using a gas-driven microinjection apparatus (MPPI-2 Pressure Injector, Applied scientific Instrumentation; Eugene, OR, USA). MOs were designed and synthesized by Gene Tools (Philomath, OR, USA). The in6:ex7-MO sequence is: [gene: 5′-CGCTCTGTGAGAAGAGTAAGAATAT-3′], and the ex7:in7-MO sequence is: [gene: 5′-ATGAGAAAGTTAGGCAATACCAGTC-3′]. Embryos were injected with 5 nl of the optimum concentration of MO prepared in 1× Danieau ^[27].

Four dpf larvae were fixed in 4% (w/v) PFA in 1× PBT, washed in 1× PBT four times, and treated as described elsewhere ^[49].