To better understand the molecular mechanism underlying axis elongation during embryogenesis, we performed ENU mutagenesis screening for mutations affecting morphogenesis of the trunk and tail in zebrafish (Koshida et al., 2005). Through this screening, we obtained a mutant in which tail elongation is defective due to a single recessive allele. In homozygotes for this mutation, kt441b, the tail elongation defect appeared after the 20-somite stage (Figures 1A–F). Homozygous embryos were less transparent than normal siblings, but most of them survived until the high-pec stage (42 h at 28°C, 3 days at 24°C).

To examine morphogenesis and cell differentiation in kt441b homozygous embryos, expression of several tissue specific marker genes, foxa2 (the axial mesoderm, the floor plate, the hypochord, as well as their progenitors, the chordo-neural hinge (CNH); Figures 1G–L), tbx6 (posterior presomitic mesoderm; Figures 1M, N), and mespba (anterior presomitic mesoderm; Figures 1O, P), tbx16 (the tailbud; Figure 1Q) were analyzed. While no obvious difference was observed in the expression pattern of tbx16 at the 19-somite stage among embryos obtained by crossing between kt441b heterozygotes, misexpression of foxa2, tbx6, and mespba was found in some embryos (Embryos with misexpression were identified by genotyping as kt441b homozygotes after identification of the responsible gene. Penetrance of the phenotype of misexpression was 100%). This result suggests that a defect of the gene responsible for the kt441b mutation causes aberrant gene expression in the axial and paraxial mesoderm in developing zebrafish embryos.

To better understand the phenotype of kt441b mutant embryos, we next identified the gene responsible for this mutation. By linkage analysis, we mapped kt441b to chromosome 24. Further fine mapping revealed that kt441b was located between two simple sequence length polymorphism (SSLP) markers that were generated close to the EST marker fi13d03 and on the BAC clone CR385039 (Figure 2A). Because the genome sequence was not completely assembled at that time, we manually aligned contigs and BAC clones between the two SSLP markers by Blast search and examined the expression pattern of several predicted genes on the contigs. Among them, expression of mastl was detectable in somite stages and reduced in kt441b embryos (Figure 2B). We suppose that this reduction is probably due to nonsense-mediated mRNA decay (Chang et al., 2007). Sequencing of mastl mRNA prepared from kt441b homozygous embryos showed a remnant of intron 4, and genomic sequencing revealed a one-base substitution of the splice donor of intron 4 (Figure 2C). RT-PCR analysis with primers designed on exons 3 and 5 of mastl showed two splicing variants, both of which contain the STOP codon just downstream of exon 4. These were specifically expressed in kt441b homozygous embryos (Figure 2D). Both mis-spliced products encode Mastl proteins truncated around the N-terminus of the kinase domain (Figure 2E). Furthermore, by CRISPR/Cas9-mediated genome editing, we generated a mutant containing an 11-bp deletion (mastl ^kt3002 ) just upstream of the conserved Mastl kinase domain (Supplementary Figure S1). This mutant showed the same live phenotype and did not complement kt441b (Figure 2F). Taken together, we concluded that mastl is the gene responsible for kt441b mutation.

In situ hybridization analysis showed that zebrafish mastl mRNA is expressed ubiquitously during early developmental stages, including the pre-MBT (midblastula transition) period (Figure 2G), indicating that maternally supplied mastl mRNA and protein can compensate for lack of zygotic products in early developmental stages (Kane et al., 1996). This is consistent with our observation that the morphological defect in kt441b homozygous embryos became evident after the 20-somite stage.

Mastl is a kinase that functions in mitotic progression. Thus, we examined progression of mitosis in mastl ^kt441b homozygous embryos. Monitoring the number of cells positive for phosphorylated Histone H3 Ser10 (pHH3), a mitotic marker, showed an increase of mitotic cells in whole bodies of mastl ^kt441b homozygous embryos at the 14-somite stage, but not at the 6-somite stage (Figure 3A), perhaps because of depletion of maternal products.

To see whether the increase of mitotic cells was caused by the delay of mitosis, we sparsely labeled nuclei of CNH cells by injection of Histone H2A-mCherry mRNA into one blastomere at the 8-cell stage and performed time-lapse imaging. While the average duration from prometaphase to the onset of anaphase, i.e., from the start of chromosome condensation to chromosome segregation, was 15 min (SD 2.38, n = 10, from three embryos) for cells in siblings, it was more than 84 min (SD 40.83, n = 18, from three embryos) for cells in mastl ^kt441b homozygous embryos. Distribution of the duration in each cell is shown in Figure 3B and Supplementary Figure S2. The average duration in cells in mastl ^kt441b homozygous embryos should be longer than indicated because data obtained from mastl ^kt441b homozygous embryos includes duration in cells that were already in prometaphase or metaphase at the start of time-lapse imaging, or those that were still in metaphase at the end of time-lapse imaging. This prolonged duration from prometaphase to the onset of anaphase is consistent with previous reports (Yu et al., 2004; Burgess et al., 2010; Voets and Wolthuis, 2010; Álvarez-Fernández et al., 2013; Diril et al., 2016).

To investigate the mitotic defect more precisely, we sparsely labeled cell membranes with Venus, along with labeling of nuclei with Histone H2A-mCherry. In this analysis, we focused on tailbud cells that entered prometaphase within the first 30 min of our time-lapse imaging and tracked them until the end of imaging, or as long as they could be tracked. Interestingly, in mastl ^kt441b homozygous embryos, many prometaphase cells progressed to cytokinesis after a prolonged metaphase-like state (Figures 3C, D). Two types of aberrant cytokinesis were observed. One was that chromosomes were scattered in the elongated cell, and subsequent cytokinesis was completed to pinch off some of the chromosomes (missegregation). The other type was that cytokinesis was completed without obvious chromosome segregation (no segregation). Missegregation was the major mitotic defect in mastl ^kt441b homozygous embryos. Additionally, in missegregated cells, after completion of cytokinesis, daughter cells often maintained condensed chromosomes for a long time, even after daughter cells were well separated. Similar defects have not been reported in previous studies of cells or organisms with impaired mastl function, although failure of cytokinesis and appearance of polyploid nuclei or double nuclei were observed. Prolonged condensation of chromosomes in separated daughter cells observed in mastl ^kt441b homozygous zebrafish embryos might be an anomaly highlighted in the context of development of tailbuds in zebrafish. Taken together, we concluded that mastl is required for proper progression of mitosis in zebrafish embryos, as reported in other organisms, and that cytokinesis without completion of chromosome segregation and delay of chromosome decondensation in separated daughter cells is prominent, especially in zebrafish embryos.

How does perturbed gene expression occur in mastl ^kt441b homozygous embryos? Is it related to observed mitotic defects? To address these questions, we focused on aberrant expression of the foxa2 gene. This is because abnormal foxa2 expression was observed beyond CNH, which is specified at the gastrulation stage.

To explain how aberrant foxa2 gene expression occurs, we considered two hypotheses. One is that CNH cells aberrantly move out, maintaining their own gene expression profiles. The other is that cells outside of the CNH start to express foxa2 mRNA aberrantly. To distinguish these two cases, we generated transgenic fish that express a photoconvertible fluorescent protein, Kaede, under control of the foxa2 promoter (TgBac(foxa2:Kaede)). As shown in Figure 4A, Kaede expressed in CNH cells was photoconverted by UV irradiation and embryos were examined 5.5–6 h after photoconversion. Due to the stability of photoconverted Kaede protein, all cells in the CNH under irradiation should maintain the photoconverted form of Kaede, Kaede-red, throughout the experiment. Thus, if some CNH cells moved out, they should still preserve Kaede-red outside the CNH. In contrast, if cells outside the CNH newly start to express foxa2 after UV irradiation, the unconverted form of Kaede, Kaede-green, should be expressed. Note that Kaede fluorescence of TgBac(foxa2:Kaede) can be detected after the 3-somite stage only in axial tissue in our experiments, even though foxa2 mRNA starts to be expressed in the axial hypoblast and endodermal precursors from the early gastrulation stage.

All cells expressing Kaede fluorescence outside the CNH maintained Kaede-red in mastl homozygous embryos (n = 47 cells in 3 kt441b homozygous embryos, n = 5 in 11 sibling embryos; Figures 4B, C). Cells expressing only Kaede-green were not detected outside the CNH. These results support the former hypothesis, clearly showing that some CNH cells were abnormally detached in mastl ^kt441b homozygous embryos.

To examine the process of cell detachment more precisely, we next performed time-lapse imaging of mastl ^kt441b homozygous embryos by labeling CNH cells with Kaede fluorescent protein using TgBac(foxa2:Kaede). By reverse-tracking of Kaede-positive cells located outside the CNH during time-lapse imaging, the process of detachment of these cells was examined. To make this tracking easier, nuclei were labeled sparsely by injection of Histone H2A-mCherry mRNA into one blastomere of the 8-cell stage.

Twenty-one cells in 3 individual mastl ^kt441b homozygous embryos were tracked and 12 cells were successfully tracked back to the point of detachment from the CNH (Figure 5A). The other 9 cells were already detached at the beginning of time-lapse imaging (Supplementary Figure S3). On the other hand, in 5 sibling embryos, only one detached cell, which died soon after detachment, was detected (Figure 5B, Supplementary Material S1). In mastl ^kt441b homozygous embryos, all 12 cells successfully traced from the CNH detached during mitosis, suggesting that cell detachment was related to mitosis. Importantly, all 12 cells exhibited aberrant chromosome segregation (Figures 5C, D, Supplementary Material S2–S6). While chromosome segregation appeared to have started, some chromosomes remained in the metaphase plane following division into two unequal clumps, which sometimes fused and then separated again. This aberrant chromosome segregation was similar to the missegregation shown in Figure 3D. These results suggests that aberrant mitosis caused by mastl depletion promotes cell detachment from the CNH.

Interestingly, all tracked cells after “missegregation” type cell division maintained round cell shapes and condensed chromosomes for prolonged periods (Figures 5C, D, F) compared to normal cell division in siblings (Figure 5E). Because some detached cells eventually decondensed their chromosomes (Figure 5C), completion of mitosis seems to be delayed in mastl ^kt441b mutant embryos. We confirmed the delay of mitotic exit in detached cells by staining other mastl ^kt441b homozygous embryos with an antibody that specifically recognizes phosphorylation at-Ser10 of Histone H3 (pHH3); all detached cells with condensed chromosomes were pHH3 positive, while all detached cells with decondensed chromosomes were pHH3 negative, as predicted (Figure 5G; Supplementary Figure S4). Of note, double staining of foxa2 mRNA and Kaede revealed that foxa2 mRNA in detached cells, i.e., Kaede-positive cells, was not detected in cells with condensed chromosomes, but was mostly detected in cells with decondensed chromosomes (Figure 6A; Supplementary Figure S5). This result suggests that foxa2 expression ceased during aberrant mitosis in detached cells, but restarted after mitotic exit. This is consistent with our observation that the proportion of foxa2 mRNA-positive cells among detached cells, increased as development progressed.

Zebrafish with the TL2E background were used as the wild type, as described previously (Kishimoto et al., 2004). Zebrafish were maintained at 28 °C under a 14-h light/10-h dark cycle. Embryos were grown at 28.5 °C or 23.5 °C and their developmental stages were determined according to morphological criteria (Kimmel et al., 1995). mastl ^kt441b mutants were obtained with ENU-based mutant screening performed in our laboratory (Koshida et al., 2005).

Zebrafish lines generated in this study have been deposited in the National BioResource Project in Japan.

For linkage analysis, kt441b heterozygous fish (TL2E background) were mated with wild-type India fish to generate F1 families. Homozygous kt441b mutant embryos were raised from the F1 cross and selected by morphological criteria (defect of tail elongation). Pools of genomic DNA from 10 homozygotes or siblings were subjected to segregation analyses using simple sequence polymorphism (SSLP) markers of the MGH or LN54 panels. BAC or contig sequences were obtained from the Sanger Institute Zebrafish Genome Browser (http://www.ensembl.org/) and UCSC Genome Browser (https://genome.ucsc.edu/). The GENSCAN web server at MIT (http://hollywood.mit.edu/GENSCAN.html) was used for prediction of genes near the mutation point.

Total RNA was purified from 50 18-somite-stage embryos using Trizol (Thermo Fisher), and ssDNA was prepared by reverse transcription using SuperScript III (Thermo Fisher) with a mastl specific primer 5′-TTT​TCC​GAT​AAC​GGA​TGA​GC-3′ either from the homozygotes or their siblings. Partial fragments of mastl cDNA were amplified using a primer pair, 5′-GAT​GGA​GTA​TCT​GAT​TGG​AGG-3′, 5′- GAC​CAA​GAG​AGC​TGA​TCA​GTG​A-3′, designed from exons 3 and 5, respectively. Amplified PCR fragments were separated by electrophoresis and each band of mastl ^kt441b homozygotes was subcloned using TOPO-TA cloning (Thermo Fisher). Four clones of each band were sequenced.

To generate foxa2-promoter-driven Kaede transgenic fish, BAC clone DKEY-208C12, which contains the foxa2 genomic region, was used for bacteria-mediated homologous recombination, basically following standard protocols (Bussmann and Schulte-Merker, 2011), but using Escherichia coli strain SW102 for recombination. Briefly, two Tol2 long terminal repeats of Oryzias latipes Tol2 transposon in opposing directions flanking an ampicillin resistance cassette were inserted into the BAC vector backbone. A cassette containing the Kaede open reading frame (from pKaede-S1; MBL) with an SV40 polyA signal and a reversely oriented kanamycin-resistance gene was inserted at the start ATG of the foxa2 gene. (sequences of primers are in Supplementary Table S1). Successful recombinants were confirmed by PCR analysis. Recombinant foxa2:Kaede BAC was then injected with synthesized tol2 transposase capped-mRNA into wild-type embryos at the 1-cell stage, and transgenic lines were established by screening for Kaede expression. Even though foxa2 mRNA starts to express in axial hypoblast and endodermal precursors from the early gastrulation stage, Kaede fluorescence of the Bac(foxa2:Kaede) transgene can be detected after the 3-somite stage.

Construction of guide RNA vectors and preparation of single-guide RNA (sgRNA) and Cas9 mRNA were performed as described previously (Cong et al., 2013; Hwang et al., 2013). The sgRNA sequence for targeting mastl was as follows: 5′-TGA​AAG​CTC​CTT​CCA​TCG-3’. Mutation efficiency was assessed by performing a T7 endonuclease assay (Chen et al., 2012) with the following primers: 5′-GAC​TTA​CCA​TCA​GCC​AGT​TG’, 5′-TGT​AAC​GTT​AGC​GGA​GTA​CG-3′. Details on mastl mutagenesis and sequences are provided in Supplementary Figure S1. For microinjections, 1 nL of injection solution (25 ng/μL sgRNA, 0.4 μg/μL Cas9 mRNA, 0.2 M KCl and 0.05% Phenol Red) was injected into one-cell-stage zebrafish embryos using an IM300 micro injector (Narishige).

For genotyping of mastl ^kt441b embryos, genomic DNA fragments amplified by PCR using mastl-genoF: 5′- TGA​TTT​GGA​GTG​TTT​TGT​TTG​TC-3′ and mastl-genoR: 5′- TAT​TAA​TTA​TTA​GGA​AAT​AGA​AAG​AT-3′ were digested with MboI, which can digest DNA fragment from mastl ^kt441b , but not from mastl ^wt .

For genotyping of mastl ^kt3002 , the following primers were used to amplify each allele of genomic DNA independently. For mastl ^kt3002 : 5′-GTG​GAC​TGA​AAG​GAA​GAC​TC-3′ and 5′-AGC​ACG​AAG​TCC​TCG​CTT​C-3′, for mastl ^wt : 5′-AAC​AGT​GCA​GTG​AAA​GCT​CC -3′ and 5′- AAC​GTT​AGC​GGA​GTA​CGT​AC-3’.

Photoconversion was carried out using the epifluorescence setup on a Zeiss Axio Imager Z1 microscope. Kaede-expressing embryos at the 6-somite stage were mounted on glass slides in 1/3 Ringer’s solution without removing the chorion. Directions of embryos were adjusted manually by moving the coverslip. A GFP filter set was first used to identify the region for photoconversion, and the size of the target field was adjusted to include the chordo-neural hinge region by adjusting the diameter of the diaphragm in the epifluorescence light path. Once target cells were in focus, a DAPI filter set was chosen, and UV light was switched on for 30 s at maximum intensity. After photoconversion, embryos were returned to the dishes with 1/3 Ringer’s solution and kept in the incubator at 24°C for 5.5–6 h. Embryos were mounted as described below and imaged on a LEICA SP8 confocal microscope.

Male and female mastl ^kt441b/wt ; TgBac(foxa2:Kaede) ^+/− fish were crossed and resulting embryos were labeled with Histone H2A-mCherry by injecting RNA into a blastomere at the 1-cell stage or into a blastomere at the 8-cell stage. Embryos that had Histone H2A-mCherry labeling in the tailbud region were selected for imaging. Embryos at the 6-somite stage were embedded in 0.5% low-melting-point agarose in 1/3 Ringer’s solution and imaged using a Leica SP8 confocal microscope with a 40x water immersion lens. Z-stacks were collected at 3–4 min intervals for 3–4 h. Cell movements were tracked manually, checking the orthogonal view with Las (Leica) or ImageJ. Images were compiled to movies with ImageJ.

Whole-mount in situ hybridization was performed as previously described (Koshida et al., 2005). Briefly, dechorionated embryos were fixed with a 2-h incubation at room temperature in 4% paraformaldehyde (PFA)/PBS. After fixation, embryos were dehydrated by treatment with methanol and kept at −20 °C until use. Hybridization was performed at 65 °C and signals were detected using anti-Digoxigenin-AP (Roche; 111093274910) and BM Purple AP Substrate, precipitating (Roche; 11442074001). For simultaneous detection of Kaede fluorescent protein with detection of foxa2 mRNA, anti-Digoxigenin-POD antibody (Roche; 11207733910) and a TSA Plus Cyanine3/Fluorescein System (Perkin Elmer; NEL753001KT) were used.

Plasmids for probes of foxa2, tbx6, and mespba were kindly donated by Drs. Phillip W. Ingham (Strähle et al., 1993), Masataka Nikaido (Nikaido et al., 2002), and Hiroyuki Takeda (Sawada et al., 2000). Primers for cloning zebrafish mastl probe were as follows: 5′-TAT​GCC​ATC​AAG​GTG​GTG​AA-3′, 5′-AGG​CCA​TGA​GGA​TGA​TCA​AG-3′

Embryos were fixed with 4% paraformaldehyde (PFA)/PBS for 2 h at room temperature. Permeabilization was done with 0.5% Triton X-100 at room temperature for 1.5 h or acetone at −20°C for 5 min (for anti-phosphorylated Histone H3 Ser 10 antibody) and then incubated with 1/500 diluted rabbit anti-Kaede antibody (MBL; PM012) or 1/1000 diluted rabbit anti-phosphorylated Histone H3 Ser10 antibody (Millipore; 06–570) overnight at 4 °C. Signals were detected with 1/1,000 diluted anti-rabbit IgG Alexa 488 (Thermo Fisher; A11008) or 555 (Thermo Fisher; A21429). For simultaneous detection of Kaede and pHH3, Kaede-green fluorescence was detected without staining.