Transcriptional autoregulation of zebrafish tbx6 is required for somite segmentation

During embryogenesis, somites, bilateral epithelial structures of the paraxial mesoderm, are formed by the segregation of cells from the anterior region of the presomitic mesoderm (PSM), subsequently giving rise to metameric structures in the vertebrate body. With the posterior extension of the embryonic body, somite boundaries are periodically established in the anterior-posterior direction.

Prior to the morphological formation of somite boundaries, the future intersomitic boundaries are established at the anterior PSM. As the expression domain of a T-box transcription factor, Tbx6 protein, exhibits a steep boundary in the position corresponding to the presumptive intersomitic boundaries in the PSM in mice and zebrafish, the anterior border of the expression domain of Tbx6 (hereafter referred to as Tbx6 domain) is thought to provide crucial positional information for formation of the somite boundary (Oginuma et al., 2008; Wanglar et al., 2014; Zhao et al., 2015). The functional importance of Tbx6 in somite segmentation has been well recognized in vertebrates. In mice, a hypomorphic allele of Tbx6 shows defective segmentation of somites (Watabe-Rudolph et al., 2002; White et al., 2003). In zebrafish, fused somites/tbx6 (previously designated tbx24) mutants lack all the apparent somite boundaries (Nikaido et al., 1997; van Eeden et al., 1996). In accordance with periodic somite segmentation, the anterior edge of the Tbx6 domain retreats with an interval of one somite length towards the posterior region in the PSM in a stepwise manner. This regression of the Tbx6 domain is mediated by Groucho/TLE-associated Ripply family proteins (Kawamura et al., 2005). Although the function of Ripply appears to be different depending on the cell context (Kawamura et al., 2008; Kondow et al., 2007; Zhao et al., 2018), Ripply in the anterior PSM mediates the degradation of Tbx6 proteins, leading to the posterior shift of the Tbx6 domain (Kinoshita et al., 2018; Oginuma et al., 2008; Wanglar et al., 2014; Zhao et al., 2018). In mice, the clear and reciprocal border between Ripply2 and Tbx6 proteins has been shown to correspond to the future somite boundary (Ninomiya et al., 2016; Zhao et al., 2015). In the absence of Ripply function, the anterior border of the Tbx6 domain is not properly terminated but instead is sustained in the anterior paraxial mesoderm, resulting in severe defects in somite segmentation in mice and zebrafish (Chan et al., 2007; Kawamura et al., 2005; Morimoto et al., 2007; Takahashi et al., 2010; Wanglar et al., 2014; Zhao et al., 2015). On the other hand, the expression of ripply genes is dependent on Tbx6, suggesting that a Tbx6-Ripply negative-feedback loop is operating to define the presumptive somite boundary at the anterior PSM (Windner et al., 2015; Zhao et al., 2015).

In zebrafish, the anterior border of the Tbx6 domain should be efficiently shifted toward the posterior in accordance with the relatively rapid segmentation of somites (Wanglar et al., 2014). The observation that the expression of zebrafish tbx6 remains specific to the PSM during the segmentation period suggests that the transcription of tbx6 is also controlled by the posterior shift of the Tbx6 domain. In this study, we have investigated the molecular mechanisms underlying the transcriptional regulation of zebrafish tbx6 in the PSM. We show here that the transcriptional autoregulatory loop of tbx6 plays essential roles in the maintenance of tbx6 expression in the PSM. This autoregulatory loop is dependent on a proximally located single enhancer: TR1. We suggest molecular mechanisms by which Ripply-mediated removal of Tbx6 proteins enables the efficient and synchronous posterior shift of the expression domain of tbx6 transcription and the Tbx6 domain during somite segmentation in zebrafish.

Wanglar et al. showed that zebrafish tbx6 mRNA is distributed more anteriorly than the anterior limit of the Tbx6 domain in the PSM (Wanglar et al., 2014). To clarify where de novo transcription of tbx6 occurs, we first attempted to examine the expression patterns of tbx6 nascent RNA in the PSM using an intron probe that can recognize premature transcripts prior to splicing. The intron probe detected one or two punctate signals representing tbx6 intronic nascent transcripts in nuclei of the PSM cells (Fig. S1A). Two-color in situ hybridization revealed that the anterior border of the expression domain of tbx6 nascent RNA was located more posteriorly than that of tbx6 mRNA in the PSM (Fig. 1A-C,G and Fig. S1B). To examine whether the anterior border of the tbx6 transcriptional region corresponds to that of the Tbx6 domain, we compared the expression patterns of tbx6 nascent transcripts and Tbx6 proteins in the PSM. We found that the anterior border of the tbx6 nascent transcription coincides well with that of Tbx6 proteins (Fig. 1D-G), although there is a small degree of mismatch between the anterior limit of expression of tbx6 nascent RNA and Tbx6 protein. These results suggest that the anterior border of the expression domain of tbx6 nascent transcripts and the Tbx6 domain are synchronously regressed in accordance with the somite segmentation, and further suggest a correlation between Tbx6 protein and transcription of tbx6 in the PSM.

To investigate the transcriptional regulatory mechanism of tbx6, we tried to identify a cis-regulatory region(s) that drives tbx6 expression in the PSM. Using a BAC clone, CHORI211-279J3, that possesses a 139 kb genomic region encompassing the entire tbx6 locus, an EGFP reporter gene was inserted into the initiation codon of tbx6 by using homologous recombination in bacteria (Lee et al., 2001). Transient injection of the 279J3-BAC EGFP reporter into embryos showed a specific EGFP signal in the PSM and somites (Fig. S2), which includes the endogenous expression of tbx6 at the 15-somite stage (Nikaido et al., 2002). Owing to the stability of EGFP proteins and the rapid segmentation of somites in zebrafish, the EGFP signal tends to be persistently observed in the somites (Kawamura et al., 2016).

To identify the functional cis-regulatory region of tbx6 in the PSM, we dissected genomic regions surrounding the tbx6 locus in the 279J3-BAC EGFP reporter. Subsequent reporter gene analysis revealed that a 2.1 kb DNA fragment upstream of the translation start site of tbx6 fused to the EGFP gene (tbx6-2.1kb-EGFP) is capable of driving the expression in the PSM and somites (Fig. S2). To evaluate the enhancer activity, we generated the transgenic line Tg(tbx6:EGFP), which harbors the 2.1 kb upstream region of tbx6 fused to the EGFP reporter gene. Two transgenic lines isolated from different founder fish essentially showed the same expression patterns. In Tg(tbx6:EGFP) embryos, faint EGFP fluorescence was first observed at the bud stage in the paraxial mesoderm, and the EGFP signal was subsequently restricted to the PSM and somites during the segmentation period (Fig. 2A-F).

To precisely locate the distributions of EGFP mRNA in Tg(tbx6:EGFP) embryos, whole-mount in situ hybridization was performed. During gastrulation, no signal of EGFP mRNA was detected (Fig. 2A′), although initial expression of tbx6 was evident at both sides of the embryonic shield at 60% epiboly (Fig. 2A″) (Nikaido et al., 2002). At 80% epiboly, EGFP mRNA was first detected in the paraxial mesoderm of Tg(tbx6:EGFP) embryos (Fig. 2B′), which corresponds to the endogenous expression of tbx6 (Fig. 2B″) (Nikaido et al., 2002). Thereafter, the expression of EGFP mRNA was detected in the PSM and somites during somitogenesis (Fig. 2C′-F′). As the expression of the reporter gene in Tg(tbx6:EGFP) resembled the expression patterns of endogenous tbx6 in the PSM, we concluded that the 2.1 kb upstream genomic region of tbx6 possesses enhancer activities that drive the expression of tbx6 specific to the PSM from the bud stage to somitogenesis but does not possess the capacity to drive the initial expression of tbx6 during early gastrulation.

A search for the putative binding sites of transcription factors in this region revealed that three binding sites of T-box transcription factor (designated T1, T2 and T3) and one binding site of bHLH transcription factor (E-box; designated E1) are located in the genomic region proximal to the translational initiation site of tbx6 (Fig. S3A). Comparisons of the genomic sequences upstream of tbx6 revealed that T1 and E1 sites are highly conserved in teleosts (Fig. S3B). We next examined whether these putative binding sites are required for enhancer activity of the tbx6-3.7kb-EGFP reporter gene. By injecting reporters in which these binding sites were individually mutated, we found that substitution of T1 apparently reduced the EGFP signal compared with the intact reporter gene in embryos, and substitution of T2 slightly reduced the EGFP signal (Fig. 2G-M and Table S1). In contrast, the substitution of T3 or E1 alone did not appear to influence EGFP activity (Fig. 2G,J,L,M and Table S1). In addition, combined removal of T1 and T2 resulted in complete loss of EGFP signal (Fig. 2G-K and Table S1). These results suggest that the T1 and T2 sites activate the reporter gene in an additive manner and play important roles as core regulatory sequences in the expression of tbx6. Hereafter, the genomic region containing T1 and T2, which also includes E1, is designated as tbx6 regulatory region 1 (TR1).

To gain further insights into the roles of the identified tbx6 enhancer in somite segmentation, genomic deletions were introduced in putative binding sites in TR1 by using the TALEN or CRISPR/Cas9 (Fig. 3A). Individual deletion of each binding site (tbx6^ΔT1/ΔT1, tbx6^ΔT2/ΔT2 and tbx6^ΔE1/ΔE1) did not lead to any apparent morphological defects (Fig. S4A-C). We also isolated mutants (designated tbx6^ΔTR1) lacking 28 bp that contains all of the putative binding sites in TR1. In contrast to the mutants lacking each binding site, tbx6^ΔTR1/ΔTR1 embryos clearly exhibited defective somite segmentation; the somite boundaries in the intermediate trunk region were severely impaired, although the first five to eight somites and the tail somites appeared to be segmented (Fig. 3B,C). By crossing tbx6^ΔTR1 with a tbx6 null allele, we confirmed that tbx6^ΔTR1 is a hypomorphic allele of tbx6 (Fig. S5). We then precisely examined which segment boundaries were disrupted in tbx6^ΔTR1/ΔTR1 by staining the segment boundary; we found that the 7th to 25th somite boundaries were severely affected in tbx6^ΔTR1/ΔTR1 mutants (Fig. 3D,E). These genetic results indicate that TR1 is essential for somite patterning in the intermediate region of zebrafish embryos and suggest that the T1 and T2 sites function in a cooperative manner in this process.

Next, to determine whether the TR1 influences the expression of tbx6, we analyzed the endogenous expression patterns of tbx6 in tbx6^ΔTR1/ΔTR1 embryos (Fig. 4A). At 80% epiboly, when the expression of tbx6 was observed in the dorsal blastoderm margin of wild-type embryos (Nikaido et al., 2002), a signal of tbx6 mRNA was undetectable in tbx6^ΔTR1/ΔTR1 (Fig. 4A). Subsequently, at the bud stage, the expression of tbx6 was restricted to the paraxial mesoderm in sibling embryos (Nikaido et al., 2002). In tbx6^ΔTR1/ΔTR1, tbx6 mRNA was similarly detected in the paraxial mesoderm, although the signal intensity was slightly weaker than that in sibling embryos (Fig. 4A). During the segmentation period, expression of tbx6 was specifically observed in the PSM of sibling embryos (Nikaido et al., 2002). In contrast, tbx6 mRNA was significantly reduced at the six-somite stage (Fig. 4A), when tbx6^ΔTR1/ΔTR1 embryos begin to exhibit segmental defects (Fig. 3B-E). At the 12-somite stage, tbx6 mRNA signal was substantially reduced in the PSM of tbx6^ΔTR1/ΔTR1(Fig. 4A). Additionally, immunostaining showed that homozygous deletion of TR1 decreased the amount of Tbx6 protein to an undetectable level at the 12-somite stage, whereas the expression of Tbx6-related Tbx16 in the posterior PSM and tailbud was indistinguishable from that in wild-type embryos (Fig. 4B). On the other hand, we investigated the expression patterns of tbx6 in tbx6^ΔT1/ΔT1, tbx6^ΔT2/ΔT2 and tbx6^ΔE1/ΔE1 embryos, which do not show abnormal somite segmentation. Whole-mount in situ hybridization showed that the expression patterns of tbx6 in tbx6^ΔT1/ΔT1, tbx6^ΔT2/ΔT2 and tbx6^ΔE1/ΔE1embryos were not significantly altered compared with those in tbx6^ΔTR1/ΔTR1 embryos at the 12-somite stage (Fig. S4D-F). Thus, these results indicate that TR1 plays an essential role in the maintenance of tbx6 expression in the PSM during the somite segmentation in the intermediate region.

Analysis of the expression patterns of several genes involved in somite patterning revealed that the defects in tbx6^ΔTR1/ΔTR1 are indistinguishable from those in tbx6 null mutants at the 12-somite stage. As observed in tbx6^−/− (Nikaido et al., 2002), the expression of segmentation genes such as mesp-ba, mesp-bb, her1 and papc was abolished in the anterior PSM of tbx6^ΔTR1/ΔTR1 (Fig. 4C). The decrease of these mRNAs in tbx6^ΔTR1/ΔTR1 was also supported by qPCR analysis (Fig. S6). In addition, the rostral-caudal polarity within a somite was perturbed in tbx6^ΔTR1/ΔTR1 embryos, as observed in tbx6 null mutants (Fig. 4C) (Nikaido et al., 2002). Expression of fgf8a and tbx18, which is normally confined to the rostral region of somites, was almost absent in tbx6^ΔTR1/ΔTR1. In contrast, the expression of myod and uncx4.1, which is restricted to the caudal region of somites, was rostrally expanded in tbx6^ΔTR1/ΔTR1. Furthermore, time-course analysis revealed that the abnormal expression patterns of segmentation genes and the perturbance of somite polarity in tbx6^ΔTR1/ΔTR1 at the 12-somite stage were not observed at early or late somite segmentation stages (Fig. S7), suggesting that the somites segmentation in tbx6^ΔTR1/ΔTR1 is transiently perturbed in the formation of intermediate embryonic body and subsequently recovered in the tail somites. Moreover, these results demonstrate that deletion of TR1 considerably decreases the expression of tbx6, leading to the almost complete loss of function, suggesting that the transcriptional regulation of tbx6 is totally dependent on a single enhancer, TR1, at the 12-somite stage.

We next explored the molecular mechanisms underlying the transcriptional regulation through the TR1 region. Sequences of T1 (5′-TCACACTT-3′) and T2 (5′-TCACACGT-3′) are similar to the consensus binding sequence of zebrafish Tbx6, revealed by the ChIP assay (Windner et al., 2015), and that of zebrafish Tbx6-related Tbx16, revealed by in vitro-binding selection assay (Garnett et al., 2009). These results suggest that Tbx6 binds to T1 and T2, and regulates its own transcription. To examine this possibility, we performed a luciferase assay in which the reporter genes are under the control of the 2.1 kb genomic fragment upstream of tbx6 (Fig. 5A). We found that zebrafish Tbx6 significantly activates reporter gene activity in HEK293T cells (Fig. 5B). The dose-dependent activation by Tbx6 was not seen in the case of a mutated luciferase reporter lacking TR1. ChIP experiments were then conducted to determine whether Tbx6 proteins are associated with TR1 in the PSM of zebrafish embryos. Using anti-Tbx6 serum, we found that Tbx6 proteins were preferentially associated with the genomic region containing the TR1, but not with another genomic region 3.1 kb upstream of the tbx6 locus (Fig. 5C). These results suggest that Tbx6 binds to TR1 and activates the transcription of tbx6 in the PSM.

To address whether Tbx6 activates its own transcription in embryos, we examined the expression patterns of tbx6 in the absence of tbx6 function. As the tbx6 allele fss^ti1 used in this study possesses the 2 bp deletion that results in the premature translation-termination codon (Nikaido et al., 1997; van Eeden et al., 1996), mutated tbx6 transcripts could be influenced by nonsense-mediated mRNA decay (Maquat, 2004). Therefore, we analyzed the nascent transcripts of tbx6 in the antisense morphoplino-mediated knockdown embryos using an intron probe. As shown in Fig. 1A, punctuate signals of tbx6 nascent RNA were detected in the PSM of control embryos. Compared with the expression domain of Tbx16, tbx6 nascent RNA was distributed more anteriorly (Fig. 5D,E). In contrast, knockdown of tbx6 showed that the expression of tbx6 de novo RNA was decreased although the initial transcription at the posterior PSM was not influenced (Fig. 5D,E). In particular, it is notable that tbx6 nascent transcripts were not detected in the anterior PSM. Tbx6 functions as a transcriptional activator, as observed in the luciferase assay (Fig. 5B); it is also transcriptionally autoregulated, which proceeds after the initiation of tbx6 expression. Our data are consistent with this prediction and demonstrate that tbx6 is actually required for the maintenance of its own transcription in the anterior PSM, although the transcription is initiated independently of tbx6 in the posterior PSM. Based on these results, we conclude that the transcriptional autoregulatory loop through TR1 maintains the expression of tbx6 in the anterior PSM.

Next, we investigated the functional roles of the autoregulatory loop of tbx6 in somite segmentation. In the anterior PSM, Ripply has been shown to regress the anterior border of the Tbx6 domain (the future intersomitic boundary) by degradation of the Tbx6 protein by one somite length toward the posterior (Oginuma et al., 2008; Takahashi et al., 2010; Wanglar et al., 2014; Zhao et al., 2018). In the absence of zebrafish ripply1 and ripply2 function, the Tbx6 domain is not properly terminated and expands toward the anterior paraxial mesoderm (Wanglar et al., 2014; Yabe et al., 2016). If Tbx6 upregulates its own expression, we speculate that tbx6 nascent transcription would be similarly expanded toward the anterior. In wild type, the expression of tbx6 nascent RNA was properly terminated at the anterior PSM (Fig. 6A). However, the expression of tbx6 nascent RNA was expanded toward the anterior paraxial mesoderm in accordance with the anterior expansion of the Tbx6 domain in ripply1^−/−; ripply2^−/− mutants (Fig. 6A). Then, to address whether a feedback loop through TR1 is involved in the anterior expansion of the expression domain of tbx6 transcription, we investigated the expression patterns of tbx6 in tbx6^ΔTR1/ΔTR1; ripply1^−/− embryos at the three- to five-somite stage when the expression of tbx6 is still detectable in tbx6^ΔTR1/ΔTR1. In tbx6^+/+;ripply1^−/−, the expression of tbx6 was ectopically expanded toward the anterior region (Fig. 6B). However, this anterior expansion of tbx6 expression was significantly suppressed in tbx6^ΔTR1/ΔTR1; ripply1^−/−, which resulted in comparable expression patterns of tbx6 in tbx6^ΔTR1/ΔTR1; ripply1^+/+ and tbx6^ΔTR1/ΔTR1; ripply1^−/− (Fig. 6B). These results suggest that the anterior expansion of tbx6 expression in the absence of ripply1 is totally dependent on TR1. Furthermore, TR1 is not only required for the maintenance of tbx6 expression at the anterior PSM but also acts as a key cis-regulatory module for the Ripply-mediated regression of the expression domain of tbx6 transcription in zebrafish embryos.

In this study, we investigated the molecular mechanisms underlying the transcriptional regulation of zebrafish tbx6 in somite segmentation. We identified an essential regulatory module, TR1, in which two T-box binding (T1 and T2) sites and one E-box (E1) binding site are present. Among these binding sites, we presume that T1 and T2 play a crucial role in the maintenance of tbx6 expression in the anterior PSM. First, deletion of TR1 resulted in severe defects in somite segmentation (Fig. 3), although the deletion of either T1 or T2 did not lead to apparent embryonic defects (Fig. S4). Consistently, transient injection into embryos showed that simultaneous mutations of T1 and T2 significantly influence the reporter gene activity (Fig. 2K). Second, ChIP analysis revealed a preferential association of Tbx6 proteins with TR1 in the PSM (Fig. 5C). Third, knockdown of tbx6 abolished the transcription of tbx6 in the anterior PSM (Fig. 5D,E). Although we could not isolate mutants possessing mutations in both T1 and T2 sites due to technical difficulties, our results suggest that Tbx6 activates its own expression through the T1 and T2 sites. Besides T1 and T2, the E1 site may possess cis-regulatory activity, although deletion of the E1 site alone did not cause any apparent phenotype (Fig. S4). Interestingly, the initial expression of tbx6 was delayed in tbx6^ΔTR1/ΔTR1, suggesting that TR1 is also involved in the establishment of tbx6 transcription during gastrulation. It was shown that the bHLH transcription factor mesogenin 1 with the T-box transcription factor, no tail, is able to induce ectopic expression of tbx6 in zebrafish embryos (Yabe and Takada, 2012). These observations indicate the possibility that the E1 site cooperatively functions with neighboring T1 and/or T2 sites to induce the expression of tbx6.

The phenotype of tbx6^ΔTR1/ΔTR1 embryos showing that there is segmental defect only in the intermediate region of the embryonic body indicates that transcription of tbx6 is differentially regulated during somite segmentation. Indeed, the expression of tbx6 in tbx6^ΔTR1/ΔTR1 was detected, although its expression was weak compared with that in sibling embryos (Fig. 4A), suggesting that other cis-regulatory region(s) might be involved in the transcriptional regulation of tbx6 in a temporally specific manner. Molecular mechanisms underlying somite development are thought to be different along the anterior-posterior axis of zebrafish embryos. For example, there are differences between the control of anterior and posterior somitogenesis (Jülich et al., 2005; Kimmel et al., 1995; Koshida et al., 2005; van Eeden et al., 1996). Somite development in the trunk and tail regions is also differentially regulated (Fior et al., 2012; Yabe and Takada, 2012). Furthermore, homozygous embryos deficient in her7, which encodes one of oscillators in zebrafish segmentation clock, show regional disruption of segmentation between the 8th and 17th somites (Choorapoikayil et al., 2012), similar to the TR1 deletion. Thus, the differential temporal regulation of tbx6 could reflect different mechanisms operating in somite development along the anterior-posterior axis.

Developmental genes frequently possess multiple tissue-specific enhancers with overlapping spatial and temporal cis-regulatory activities. For example, genome-wide high-throughput screening in mice revealed that the enhancers outnumber the protein-coding genes by approximately one order of magnitude (ENCODE project Consortium, 2012). In Drosophila, transcriptional regulation is often ensured by a primary enhancer located near the basal promoter and ‘shadow’ enhancers located at remote positions (Cannavò et al., 2016; Hong et al., 2008; Perry et al., 2010). Although it has been reported that deletion of a single enhancer causes severe developmental defects in some cases (Gonen et al., 2018; Sagai et al., 2005; Yanagisawa et al., 2003), enhancer redundancy appears to be a remarkably widespread hallmark in the transcriptional regulation of developmental genes (Frankel et al., 2010; Osterwalder et al., 2018). This functional redundancy of enhancers has been proposed to contribute to the phenotypic robustness against genetic or environmental perturbations. In contrast to a number of developmental genes, we showed that the transcriptional regulation of zebrafish tbx6 is governed by a single proximal enhancer designated TR1 at least in the intermediate period of somite segmentation.

Although transcriptional regulation by a single enhancer is susceptible to genetic disturbances, this simple transcriptional regulation may provide a functional significance for somite segmentation in zebrafish. As mentioned above, Ripply removes Tbx6 proteins at the anterior PSM, leading to a posterior shift of the Tbx6 domain of one somite length. We showed that the expression domain of tbx6 nascent RNA is also expanded toward the anterior accompanied by an anterior expansion of the Tbx6 domain in ripply gene-deficient embryos (Fig. 6A). Importantly, this ectopic expression of tbx6 transcripts was significantly suppressed by additional mutation of TR1 (Fig. 6B). These results indicate that Ripply-mediated removal of Tbx6 proteins simultaneously leads to transcriptional shutdown of the autoregulatory loop through TR1. Consistent with these observations, the anterior edge of Tbx6 proteins and tbx6 transcription coincides well at the position of S-II in wild-type embryos (Fig. 1A-C). The transcriptional autoregulatory loop, which is regulated by a single enhancer, enables an efficient posterior shift of both of the expression domains of Tbx6 protein and tbx6 transcription by Ripply-mediated degradation of Tbx6 proteins. In zebrafish, the segmentation of somites proceeds rapidly (30 min). This observation implies that the transcription of tbx6, which is a source of Tbx6 proteins, has to efficiently be terminated within a limited time at the anterior PSM. Our finding of total dependence of the autoregulatory loop on a single enhancer may provide a dexterous strategy for an efficient and rapid shift of the expression domains of Tbx6 protein and tbx6 transcription during the zebrafish segmentation period.

The transcriptional regulation of zebrafish tbx6 appears to be distinct from that of mouse Tbx6, although the functional relationship between Ripply and Tbx6 in somite segmentation is conserved (Kinoshita et al., 2018; Oginuma et al., 2008; Takahashi et al., 2010; Wanglar et al., 2014; Windner et al., 2015; Zhao et al., 2018). Our study showed that Tbx6 maintains its own transcription through the two T-box binding sites in TR1, which is located proximal to the promoter. Although a 2.3 kb genomic fragment upstream of mouse Tbx6 is able to drive the reporter gene expression to the PSM, as observed in zebrafish, the reporter gene activity is not influenced in Tbx6 knockout mice but is influenced by base substitution of RBPJk binding sites (White et al., 2005). In addition, the absence of mouse Ripply2 function resulted in the expansion of Tbx6 proteins toward the anterior in a manner similar to that in zebrafish; however, the distribution of Tbx6 mRNA was still restricted to the PSM (Oginuma et al., 2008; Takahashi et al., 2010). These observations suggest that an autoregulatory loop is not operating to regulate the expression of Tbx6 in mice.

In contrast to the single Tbx6 gene in mice, three structural tbx6 homologs (tbx6, tbx16 and tbx16l) were identified in zebrafish (Griffin et al., 1998; Hug et al., 1997; Nikaido et al., 2002). In zebrafish, tbx6 was originally isolated and named tbx24 (Nikaido et al., 2002), and tbx16l was first named as tbx6 (Hug et al., 1997) and subsequently renamed as tbx6l (Morrow et al., 2017). As medaka also possesses three tbx6-related genes, tbx6 genes in teleosts must have diverged by teleost-specific gene duplication. Although duplicated genes generally retain the cis-regulatory elements of the parent locus, our results reveal that the genomic regions upstream of tbx6 were conserved in teleosts but not in mice. We presume that the regulatory elements of duplicated tbx6 loci are drastically changed in the teleost lineage. Indeed, the expression patterns of tbx6 in zebrafish and medaka are restricted to the entire PSM (Nikaido et al., 2002; Terasaki et al., 2006), whereas those of Tbx6 in mice are observed in the posterior PSM and tailbud (Chapman et al., 1996). Thus, gene duplication and alteration of the cis-regulatory module of tbx6 may have lead to subfunctionalization of tbx6 genes in teleosts. Taken together, our results suggest that the transcriptional feedback loop of tbx6 through TR1 is a teleost-specific transcriptional regulatory mechanism, which may contribute to the relatively rapid segmentation of somites in zebrafish.

Zebrafish (Riken WT; RW) were maintained at 27°C with a 14 h light/10 h dark cycle. Embryos were incubated at 28.5°C until they reached the appropriate stages. The embryonic stages were determined according to the morphological features and hours post-fertilization (hpf) (Kimmel et al., 1995). Zebrafish fss^ti1/tbx6 (Nikaido et al., 2002), ripply1^sud101 (Kinoshita et al., 2018) and ripply2^kt1034 (Yabe et al., 2016) mutants were used in this study. All the experiments using live fish complied with the protocols approved by the Committee for Animal Care and Use of Saitama University.

A zebrafish BAC clone, CHORI211-279J3, containing the tbx6 locus was purchased from BACPAC Resources Center (bacpac.chori.org). An EGFP reporter and a kanamycin-resistant gene were amplified by PCR with the following primers (5′-TTCATTTAATATTCGATAAACAGAAACGTGAAGAAAGAGCAGACCGAGACATGGTGAGCAAGGGCGAGGA-3′ and 5′-CTGTAATAGCAGTCGCTCAATCTCTGAGGTCCCAGAGCCAAACCAGGGTACAGTTGGTGATTTTGAACTT-3′) as described previously (Kawamura et al., 2016; Kimura et al., 2006; Lee et al., 2001). Using homologous recombination in SW102 bacteria, the PCR product was inserted into the translation initiation codon of the first exon of the tbx6 gene. The correct integration of EGFP reporter gene in the target sites was confirmed by PCR and DNA sequencing. For the deletion analysis of the CHORI211-279J3 BAC-based EGFP reporter, the genomic region of interest in the BAC was replaced with a beta-lactamase gene using homologous recombination. DNA sequencing was performed to confirm the deletions of target loci.

To examine the reporter gene activity, the 1-2 nl of BAC- or plasmid-based EGFP reporters were injected at a concentration of 10 ng/µl into the blastomeres of fertilized embryos. At the 15- to 18-somite stage, EGFP fluorescence was observed under the fluorescent stereomicroscope (Leica, MFZFIII) and representative images were photographed with a digital camera (Leica, DFC310 FX). At least three independent injections for each reporter gene were repeated to confirm the results.

For knockdown of tbx6, verified antisense morpholino oligo (MO) specific to the 5′UTR and initiation codon of tbx6 (5′-CATTTCCACACCCAGCATGTCTCGG-3′) were used (Jahangiri et al., 2012; Kawamura et al., 2008). Antisense MO (1-2 nl at a concentration of 3 µg/µl) was injected into the blastomeres of the zebrafish embryos at the 1- to 2-cell stage.

Transgenic zebrafish lines were generated using the Tol2 transposon system (Kawakami, 2005). Briefly, a DNA fragment corresponding to the 2.1 kb genomic region upstream of the initiation codon of tbx6 was amplified by PCR from CHORI211-279J3 BAC clone and inserted upstream of the EGFP reporter gene in the pTol2-EGFP-SV40pA plasmid. Constructed pTol2-tbx6-2.1kb-EGFP plasmid at a concentration of 10 ng/µl was injected with Tol2 transposase mRNA (25 ng/µl) and the injected embryos were raised to adulthood. After the F0 founders were mated with the wild-type fish, F1 embryos showing the EGFP fluorescence in the PSM and somites were isolated. F3 progeny possessing a single transgene in the genome were used for the analysis.

To generate tbx6 enhancer mutants, several deletions were introduced in the TR1 region using TALEN and CRISPR-Cas9 methods (Bedell et al., 2012; Gagnon et al., 2014; Irion et al., 2014). For TALEN, the target sites were designed with TAL Effector Targeter (tale-nt.cac.cornell.edu) and the TALENs (left; NN NG NN NI NI NN NI HD NG HD HD NI HD HD NI, right; NI NG NI NG NI NN NN NG NN NI NN HD NN HD NG) were constructed according to the Golden Gate assembly method described previously (Engler et al., 2009). Capped mRNAs synthesized by in vitro transcription with mMESSAGE mMACHINE Sp6 kit (Ambion) were injected at a concentration of 150 ng/µl for each mRNA into fertilized embryos. For CRISPR-Cas9, the Alt-R CRISPR-Cas9 system (Integrated DNA Technologies) was used according to the manufacturer's instructions. Briefly, CRISPR RNA (crRNA) was incubated with transactivating crispr RNA (tracrRNA) and Cas9 protein to form the gRNA-Cas9 complex. A solution containing the gRNA-Cas9 complex was injected into embryos. After sexual maturation of the TALEN- or CRISPR-Cas9-injected fish, potential F0 founders were mated with wild-type fish to generate heterozygous F1 offspring. The heterozygous F1 fish were genotyped by the fin clip assay using the heteroduplex mobility assay (Ota et al., 2013), and individuals with deletions were identified by DNA sequencing. Male and female siblings carrying the same mutation were mated to generate homozygous mutants.

Genomic DNA was extracted as described (Meeker et al., 2007) and was used as a template for PCR-based genotyping. The genotype of tbx6^ti1 mutants was determined by PCR with the following primers: 5′-TGTGCCGTTGTACCCGTCCACATG-3′ and 5′-CGACTGTGCTGAACTGCTTCCACAG-3′. The genotype of tbx6^ΔT1, tbx6^ΔT2, tbx6^ΔE1 and tbx6^ΔTR1was identified by PCR with the following primers: 5′-CCCCAAAAGATGGACCCTGAAGAA-3′ and 5′-CCCCTTCTCTCTCTGTGTGTTTCT-3′. After the reactions, PCR products were separated on 15% polyacrylamide gel or 2% agarose gel in 0.5× TBE buffer. The genotype of ripply1^sud101 and ripply2^kt1034 mutants was determined as described previously (Kinoshita et al., 2018).

For the detection of Tbx6 and Tbx16 proteins in the PSM, embryos were fixed in 4% paraformaldehyde in PBS overnight. After the washing with PBS/0.25% Triton-X three times, the embryos were briefly stored in methanol at −20°C. Embryos were then transferred into the Can Get Signal solution (Toyobo) and reacted with the following primary antibodies: rabbit polyclonal anti-Tbx6 serum #2 (1:500 dilution; Wanglar et al., 2014) and mouse monoclonal anti-Tbx16 antibody (1:500 dilution, anti-Tbx16 antibody was obtained from Zebrafish International Resource Center). After washing three times with PBS(−)/0.25% Triton-X, fluorescent signals were visualized by Alexa-Fluor 488-conjugated anti-mouse IgG antibody (Abcam, ab150113, 1:800) and Alexa-Fluor 647-conjugated anti-rabbit IgG antibody (Abcam, ab150079, 1:800). After the removal of the yolk, flat-mounts were imaged using a laser-scanning confocal microscope (Olympus FV1000).

Whole-mount in situ hybridization was performed as described previously (Thisse and Thisse, 2014). Fluorescence in situ hybridization was performed essentially as described previously (Brend and Holley, 2009). Images were captured using a confocal microscope (Olympus FV1000). For the tbx6 intron probe, the 1.4 kb genomic region corresponding to the first intron of tbx6 was amplified by PCR from the zebrafish RW genomic DNA (5′-GCATAGTGGCAGATTCAAAAACAAG-3′ forward primer and 5′-AGCGAGTTAGCATTTTAGCACTTCC-3′ reverse primer) and hybridization was carried out at 60°C. For the simultaneous detection of mRNA and protein, fluorescence in situ hybridization was conducted followed by immunohistochemistry.

Luciferase reporter gene, tbx6-2.1kb-luc, was constructed by inserting a 2.1 kb upstream genomic region from the translation initiation of tbx6 into the upstream of firefly luciferase gene of the pGL3 basic vector (Promega). To construct the tbx6-2.1kb^ΔTR1-luc, inverse PCR was carried to delete 28 bp corresponding to the TR1 from tbx6-2.1kb-luc. For luciferase assay, human embryonic kidney 293T (HEK293T) cells were transiently transfected with the reporter genes along with the indicated expression vectors using polyethylenimine. The pRL-TK-Luc vector (Promega), under the control of the thymidine kinase promoter, was used for normalization of the transfection efficiencies. The total amount of DNA in each transfection was kept the same by supplementing with the empty expression vector. After the 24 h incubation, luciferase activity was assayed using a Dual-Glo luciferase assay system (Promega).

qPCR was carried out as essentially previously described (Fujino et al., 2018). Total RNA was extracted from wild-type, tbx6^ΔTR1/ΔTR1 and tbx6^−/−embryos at the 10- to 12-somite stage and subject to qPCR analysis. The sequences of the primers used in this study are listed in Table S2.

Posterior regions including the PSM and tailbud were dissected from dechorionated wild-type embryos at the 18- to 22-somite stage as previously described (Taminato et al., 2016). Dissected tissues were frozen by immersion in liquid nitrogen and stored at −80°C until use. Approximately 1500 dissected tissues were collected and subjected to ChIP analysis. ChIP experiments were carried out by using a ChIP-IT High Sensitivity kit (Active motif) according to the manufacturer's protocol. In brief, the dissected posterior tissues were fixed with 1% formaldehyde for 15 min at room temperature. After homogenization in the presence of a proteinase inhibitor cocktail, sonication treatment was carried out using BIORUPTOR UCD-250 (Tosho Denki). Fragmentation of genomic DNA of 150-500 bp in length was confirmed by electrophoresis. The fragmented chromatins were reacted with anti-Tbx6 serum #2 (Wanglar et al., 2014), or normal rabbit serum as a negative control at 4°C overnight. Then the solutions were reacted with Protein G Agarose Beads (Active motif). After washing five times, the precipitated chromatins were reverse crosslinked by heating at 80°C and DNA fragments were purified. For evaluation of the precipitated genomic DNA, qPCR was performed in triplicate with Thunderbird SYBR qPCR mix (Toyobo) on a StepOne plus Real-time PCR system (Applied Biosystems). The sequences of the primers are listed in Table S3. Precipitated DNA was quantified using the slope of standard curves, and fold enrichment was calculated relative to the non-specific normal serum signals.

Skeleton of juvenile zebrafish (∼1.5 months old) was visualized using 0.2% calcein (Wako) as described previously (Du et al., 2001).

We thank Dr Bernard Thisse for providing xirp2a/cb1045 plasmid and members of Developmental Biology Laboratory at Saitama University for their support. The MF-20 monoclonal antibody was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa. Zebrafish Tbx16 antibody (anti-VegT) was obtained from the Zebrafish International Resource Center (ZIRC).