Introduction

In jawed vertebrates, paired appendages—such as the pectoral and pelvic fins in fish, and their homologous forelimbs and hindlimbs in tetrapods—develop at precise locations along the anterior-posterior axis of each species. These paired appendages arise from progenitor cells located in distinct regions of the lateral plate mesoderm (Murata et al., 2010; Nishimoto and Logan, 2016; Shimada et al., 2013). The positioning of these paired appendages has long fascinated researchers, yet our understanding of the molecular mechanisms underlying this process remains limited.

In bilaterian animals, Hox genes—encoding evolutionarily conserved homeodomain-containing transcription factors—provide positional information and developmental timing along the anterior-posterior axis (Iimura and Pourquie, 2007; Izpisua-Belmonte and Duboule, 1992; Krumlauf, 1994). A defining feature of Hox genes is their structural organization into Hox clusters, where multiple Hox genes are arranged in a precise order. Additionally, Hox clusters exhibit a distinctive phenomenon known as Hox collinearity, where the genomic arrangement of Hox genes correlates with specific developmental regions along the body axes (Dolle et al., 1989; Duboule and Dolle, 1989; Graham et al., 1989). In vertebrates, Hox clusters underwent divergence due to two rounds of whole-genome duplication early in vertebrate evolution (Dehal and Boore, 2005; Ohno, 1970), leading to the establishment of four distinct Hox clusters (HoxA, HoxB, HoxC, and HoxD), each consisting essentially of 1-13 paralogous groups in tetrapods. In contrast, teleost fishes experienced an additional teleost-specific whole-genome duplication, followed by the loss of a hox cluster, resulting in seven hox clusters in zebrafish (Amores et al., 1998; Woltering and Durston, 2006).

Previous studies in mice and chickens suggest that the initial position of limbs is likely regulated by Hox genes (Tanaka, 2013; Tickle, 2015). Supporting this hypothesis, earlier research has shown that the anterior boundaries of specific Hox gene expression domains align with the future positions of limbs (Burke et al., 1995; Cohn et al., 1997). Moreover, experiments with avian embryos, in which several Hox genes were overexpressed or functionally interfered with, demonstrated altered positions of forelimb buds (Moreau et al., 2019). Additionally, it has been established that Hox proteins bind to the enhancer of Tbx5 gene, which is crucial for the initial formation of forelimbs, thereby inducing Tbx5 expression in the limb buds (Minguillon et al., 2012; Nishimoto et al., 2014). However, despite the generation of numerous knockout mice for single and multiple Hox genes, Hoxb5 knockout mice exhibit only a rostral shift in the position of forelimb buds, with incomplete penetrance (Rancourt et al., 1995), and no significant abnormalities have been observed in the early stages of limb formation. This lack of genetic evidence for limb positioning by Hox genes sharply contrasts with results from genetic analyses of various knockout mice, which have shown that the paralogous 9-13 Hox genes in HoxA and HoxD clusters are cooperatively essential for proximal-distal patterning of limbs after limb bud formation (Boulet and Capecchi, 2004; Davis et al., 1995; Fromental-Ramain et al., 1996a; Fromental-Ramain et al., 1996b; Kmita et al., 2005). Consequently, the precise role of Hox genes in defining the initial position of limb formation along the anterior-posterior axis remains unclear.

Using zebrafish, which belong to the Actinopterygii class, distinct from the tetrapods of the Sarcopterygii class, we previously generated mutants lacking each of the seven zebrafish hox clusters using the CRISPR-Cas9 method (Yamada et al., 2021). In our recent genetic analyses of hoxaa, hoxab, and hoxda clusters—homologous to the mouse HoxA and HoxD clusters—we demonstrated that, similar to mice, these zebrafish hox clusters cooperatively play an essential role in the formation of the pectoral fins (Ishizaka et al., 2024), which are homologous to the forelimbs. In this study, we provide the first genetic evidence, to our knowledge, that Hox genes specify the positions of paired appendages in the vertebrates. The double-deletion mutants of hoxba and hoxbb clusters, derived from the ancient HoxB cluster, exhibit a complete absence of pectoral fins due to the failure to express tbx5a. Despite incomplete penetrance, we propose a model in which hoxb4a, hoxb5a, and hoxb5b cooperatively provide positional cues along the anterior-posterior axis within the lateral plate mesoderm, thereby specifying the initial positions for fin bud formation through the induction of tbx5a in pectoral fin field.

Results

The absence of pectoral fins in zebrafish hoxba;hoxbb cluster-deleted embryos

In a previous study, we created seven individual hox cluster-deficient mutants in zebrafish using the CRISPR-Cas9 system (Yamada et al., 2021). Among these mutants, hoxba cluster-deleted embryos exhibited morphological abnormalities in their pectoral fins at 3 dpf (Figure 1A-C). To further investigate the pectoral fin phenotype in hoxba cluster mutants, we first analyzed the expression patterns of tbx5a, which plays a key role in the initial induction of pectoral fin buds in zebrafish (Ahn et al., 2002; Garrity et al., 2002; Ng et al., 2002). When comparing with tbx5a expression in wild-type embryos, we found that the tbx5a signal was reduced in the pectoral fin buds of hoxba cluster mutants (Figure 1H, I). Given that hoxba and hoxbb clusters originated from the ancestral HoxB cluster through teleost-specific whole-genome duplication (Amores et al., 1998), there may be functional redundancy between them. Surprisingly, we discovered that the simultaneous deletion of both hoxba and hoxbb clusters resulted in the complete absence of pectoral fins (Figure 1A, G). In contrast, pectoral fins were present in hoxba^-/-;hoxbb^+/- and hoxba^+/-;hoxbb^-/- mutants (Figure 1E, F), indicating that an allele from either hoxba or hoxbb cluster is sufficient for the pectoral fin formation. Moreover, all embryos lacking pectoral fins were identified as hoxba;hoxbb double homozygous mutants, with the expected penetrance (n=15/252; 5.9%) being consistent with predictions based on Mendelian genetics (1/16=6.3 %). Furthermore, while hoxba;hoxbb cluster double homozygous mutants are embryonic lethal around 5 dpf, we could not detect any trace of pectoral fin development. Alongside this phenotype, the expression of tbx5a was significantly reduced to nearly undetectable levels in hoxba;hoxbb cluster mutants at 30 hpf (Figure 1H-M). The absence of pectoral fins in zebrafish hoxba;hoxbb cluster mutants sharply contrasts with the results of a previous study, which showed that mice lacking all HoxB genes except for Hoxb13 of the HoxB cluster did not exhibit apparent abnormalities in their forelimbs (Medina-Martinez et al., 2000). Our genetic results suggest that the zebrafish hoxba and hoxbb clusters cooperatively play a crucial role in the pectoral fin formation.

Figure 1.

Lack of pectoral fins is specific to zebrafish hoxba;hoxbb cluster-deleted embryos

Multiple genetic studies in mice have demonstrated the functional redundancy of Hox genes across the four Hox clusters (Horan et al., 1995; McIntyre et al., 2007; van den Akker et al., 2001; Wellik and Capecchi, 2003; Wellik et al., 2002). To investigate whether the absence of pectoral fins is specific to hoxba;hoxbb cluster mutants, we created double-hox cluster mutants by crossing hoxba cluster mutants with six other hox cluster mutants in zebrafish. The combinatorial mutations of hoxba cluster with five other hox clusters, excluding hoxbb cluster, did not enhance the reduced expression of tbx5a (Figure 2A-H). Furthermore, we examined combined mutants of hox clusters, excluding hoxba and hoxbb. In hoxca^-/-;hoxcb^-/- cluster mutants, where the functions of HoxC genes are completely absent, the expression patterns of tbx5a in fin buds were indistinguishable from those of wild-type embryos (Figure 2I). Additionally, our previous study showed that tbx5a expression is unaffected in hoxaa^-/-;hoxab^-/-;hoxda^-/-cluster mutants (Ishizaka et al., 2024), which lack all HoxA and HoxD-related genes. The results of our deletion mutations of zebrafish hox clusters in various combinations emphasize that the diminished expression of tbx5a in the pectoral fin buds is evident only in hoxba;hoxbb cluster mutants. Taken together, in contrast to the functional redundancy of the four Hox clusters in mice, our genetic analysis indicates that the Hox clusters responsible for the specification of pectoral fin buds are restricted to the HoxB-derived hoxba and hoxbb clusters among the seven hox clusters in zebrafish.

Figure 2.

Transformation of pectoral fin progenitor cells into cardiac cells in hoxba;hoxbb cluster-deficient embryos

To understand the absence of pectoral fin formation in hoxba;hoxbb cluster mutants, we examined the expression patterns of tbx5a during embryogenesis (Figure 3A-F). In wild-type embryos at the 10-somite stage, tbx5a expression appeared as bilateral symmetric stripes in the anterior lateral mesoderm, extending from the posterior midbrain to the second somite (Figure 3A) (Ahn et al., 2002). These tbx5a-positive cells subsequently divided into two groups: the anteriorly migrating cells form the heart primordia, while the posteriorly migrating cells give rise to the future pectoral fin buds (Figure 3C, E). In contrast, we observed that abnormal expression patterns of tbx5a in the lateral mesoderm were evident at the early stages in hoxba;hoxbb cluster mutants. Compared to wild-type embryos, the bilateral stripes of tbx5a expression were shorter along the anterior-posterior axis, and the tbx5a-positive signal was absent from the posterior region where pectoral fin progenitor cells typically emerge (Figure 3B). Although anterior migration toward the presumptive heart occurred, posterior migration toward the fin buds was undetectable due to the lack of pectoral fin progenitor cells in hoxba;hoxbb cluster mutants (Figure 3D, F). These results underscore the early role of the hoxba and hoxbb clusters in providing positional information for pectoral fin buds through the induction of tbx5a expression. In both mice and zebrafish, there is a converse relationship between forelimb/pectoral fin and heart formation in the anterior lateral mesoderm (Keegan et al., 2005; Niederreither et al., 2001; Niederreither et al., 2002; Waxman et al., 2008). Therefore, we also analyzed cardiac development (Figure 3G-J). By examining the expression patterns of nkx2.5, an early cardiac progenitor marker (Chen and Fishman, 1996; Serbedzija et al., 1998), we found that the expression domain of nkx2.5 in hoxba;hoxbb cluster mutants extended more posteriorly compared to wild-type embryos (Figure 3G, H). Furthermore, the region positive for the differentiated cardiac cell-specific myosin light chain (cmlc2) was enlarged in hoxba;hoxbb cluster mutants (Figure 3I, J) (Yelon et al., 1999). These results indicate that hoxba and hoxbb clusters play an essential role in the initial establishment of the pectoral fin field in zebrafish. In their absence, cells that should form the pectoral fins instead differentiate into cardiac cells. The transformation of pectoral fin progenitor cells into cardiac cells observed in zebrafish hoxba;hoxbb cluster mutants is reminiscent of the anterior transformation typical of the loss-of-function phenotypes of Hox genes in Drosophila and mice (Le Mouellic et al., 1992; Lewis, 1978).

Figure 3.

hoxba and hoxbb clusters are indispensable for the retinoic acid-mediated induction of tbx5a expression in the fin buds

Retinoic acid (RA) is a well-known upstream regulator of Hox genes in vertebrates (Boncinelli et al., 1991; Langston and Gudas, 1994; Marshall et al., 1996). Loss of function of retinaldehyde dehydrogenase 2 (raldh2), an essential regulator in RA synthesis, leads to the absence of pectoral fins in zebrafish (Begemann et al., 2001; Grandel et al., 2002), a phenotype similar to that observed in hoxba;hoxbb cluster mutants. Previous studies have shown that exogenous RA treatments can rescue pectoral fin formation in raldh2^-/- embryos (Gibert et al., 2006; Grandel et al., 2002). It has been suggested that RA directly or indirectly induces tbx5a expression in zebrafish fin buds (Gibert et al., 2006; Grandel and Brand, 2011; Neto et al., 2012). The phenotypic similarities between hoxba;hoxbb cluster mutants and raldh2 mutants prompted us to investigate whether RA exposure could also rescue the absence of pectoral fins in hoxba;hoxbb cluster mutants. To explore this, we introduced a frameshift mutation in raldh2 using CRISPR-Cas9, confirming the phenotypic recapitulation of previously described raldh2^-/- mutants (figure-supplement 1). Consistent with prior results (Gibert et al., 2006; Grandel et al., 2002), RA treatment in our raldh2 mutants rescued the expression of tbx5a in pectoral fins (Figure 4A-D). In contrast, RA did not induce tbx5a expression in the fin buds of hoxba;hoxbb cluster mutants (Figure 4E-H), and no trace of pectoral fin formation was observed in the treated embryos even at 5 dpf. Furthermore, the endogenous expression patterns of raldh2 in hoxba;hoxbb cluster mutants seem indistinguishable from those in wild-type embryos (Figure 4I, J), suggesting that the response to RA is absent in the fin buds of hoxba;hoxbb cluster mutants, while RA synthesis appears intact. Although a model proposing that RA directly induces tbx5a expression has been suggested (Gibert et al., 2006; Grandel and Brand, 2011; Neto et al., 2012), our results favor the model that RA-mediated induction of tbx5a in the fin buds occurs through hoxba and hoxbb clusters. Given that hoxba;hoxbb cluster mutants retain the other five intact hox clusters, our findings further support the notion that the hoxba and hoxbb clusters are specifically responsible for the RA-mediated specification of pectoral fins.

Figure 4.

Expression patterns of hoxb5a and hoxb5b in pectoral fin buds are dependent on RA

The absence of pectoral fins, commonly observed in hoxba;hoxbb cluster mutants and raldh2 mutants, suggests that candidate hox genes may be downregulated in raldh2 mutants. To investigate this, we performed RNA-seq analysis to compare the expression profiles between sibling and raldh2 homozygous embryos. As expected, we found that most of the 3’-located hox genes are downregulated in raldh2^-/- embryos (Figure 5A, B, figure supplement 2). Notably, the levels of hoxb5b transcripts were significantly reduced, along with those of hoxb5a among hoxba genes, in raldh2 mutants. Previous studies have shown that zebrafish hoxb5a and hoxb5b are expressed in the lateral plate mesoderm corresponding to presumptive pectoral fin buds (Waxman et al., 2008). Additionally, knockout mice for Hoxb5, which is homologous to zebrafish hoxb5a and hoxb5b, only exhibit altered positioning of forelimbs (Rancourt et al., 1995), among various other Hox knockouts that have been generated. Therefore, we examined whether RA regulates the expression of hoxb5a and hoxb5b in fin buds. In wild-type embryos, hoxb5a expression was observed in the neural tube, somites, and lateral plate mesoderm (Figure 5C). However, in raldh2 mutants, hoxb5a expression was significantly reduced, notably in the lateral plate mesoderm, which differentiates into the fin buds, where it was reduced to the point of being undetectable (Figure 5D). Since RA treatments in raldh2 mutants can rescue pectoral fin development, we found that hoxb5a expression in raldh2 mutants is also rescued by RA exposure (Figure 5C-E). Regarding the expression of hoxb5b, which diverged from the same ancestral gene as hoxb5a, RA-dependent expression was also confirmed, mirroring the observations made with hoxb5a (Figure 5F-H). These results suggest that both hoxb5a and hoxb5b are involved in the formation of the pectoral fins in zebrafish.

Figure 5.

Genetic screening for hox genes responsible for the specification of pectoral fin buds in zebrafish

To understand the molecular mechanisms underlying pectoral fin specification, we aimed to identify the hox genes essential for pectoral fin formation through a genetic approach. By introducing frameshift mutations into the relevant hox genes, we anticipated that the resulting mutants would replicate the absence of pectoral fins observed in hoxba;hoxbb mutants. Since hoxbb cluster contains fewer hox genes than hoxba cluster (Figure 5A), we focused our investigation on hoxbb cluster. Although hoxb5b is expressed in the presumptive fin buds (Figure 5F), other hoxbb genes may also be involved. Therefore, we simultaneously injected three gRNAs targeting hoxb5b, hoxb6b, and hoxb8b into hoxba hemizygous mutants, expecting various germline mutations in these hoxbb genes in the injected fish (Figure 6A). Subsequently, F1 fish with a hemizygous deletion of hoxba cluster and mosaic mutations were crossed with hoxba;hoxbb hemizygous mutants. Among the progeny, several embryos lacking pectoral fins were identified. Following genotyping and DNA sequencing of the target hoxbb genes, we found that all embryos without pectoral fins possessed frameshift mutations in hoxb5b, with hoxba^-/-;hoxbb^+/- being common among them (Figure 6B-E, table-supplement 1, figure-supplement 3). Similarly, we injected each crRNA for hoxb6b and hoxb8b into hoxba^+/- mutants and confirmed that neither frameshift mutation in hoxb6b nor hoxb8b in hoxba^-/-;hoxbb^+/- led to the loss of pectoral fins (Figure 6F, G, figure-supplement 3). These results suggest that the frameshift mutation of hoxb5b alone can compensate for the deletion of hoxbb cluster, recapitulating the absence of pectoral fins in hoxba;hoxbb cluster mutants.

Figure 6.

However, we encountered a problem: hoxba^-/-;hoxb5b^-/- embryos, which have a homozygous frameshift mutation in hoxb5b and a homozygous deletion of hoxba cluster, exhibited severely truncated but visibly present pectoral fins (Figure 6H, I). After analyzing hundreds of embryos from several crossings, we did not find any hoxba^-/-;hoxb5b^-/- embryos without pectoral fins. We also confirmed that the expression of tbx5a was detectable in the fin buds of hoxba^-/-;hoxb5b^-/- embryos (Figure 6K, L), in contrast to its absence in hoxba;hoxbb cluster mutants (Figure 1M). To examine the potential contributions of other hoxbb genes, we observed the phenotypes of hoxba^-/-;hoxb5b^-/-; hoxb6b^-/- embryos; however, these mutants did not exacerbate the abnormalities and did not also exhibit the absence of pectoral fins (Figure 6J, M).

The phenotype of hoxba^-/-;hoxb5b^-/- embryos indicates that the frameshift mutation in hoxb5b cannot fully compensate for the complete deletion of hoxbb cluster, suggesting that other mechanisms may be involved. A novel genetic compensation mechanism known as transcriptional adaptation has been identified: mRNA containing a premature termination codon (PTC) can induce increased expression of structurally related genes, thereby compensating for the function of mutated genes (El-Brolosy et al., 2019; Rossi et al., 2015). To avoid producing transcripts with PTC (Sztal and Stainier, 2020), we generated a full locus-deleted allele of hoxb5b using CRISPR-Cas9 with two crRNAs targeting both ends of the target locus (figure-supplement 4). In contrast to the phenotype of hoxba^-/-;hoxb5b^-/-, we found that hoxba^-/-;hoxb5b ^de/dell mutants lack pectoral fins and exhibit a significant reduction in tbx5a expression when intercrossing hoxba^+/-;hoxb5b^+/del fish (Figure 7A, B, D, E). However, the occurrence rate of embryos lacking pectoral fins (n=3/120; 2.5 %) was lower than predicted based on Mendelian genetics (1/16; 6.3 %). Other hoxba^-/-;hoxb5b^del/del embryos did exhibit shortened pectoral fins (Figure 7C). Additionally, concerning hoxba cluster, we also generated frameshift-induced hoxb5a^-/-;hoxb5b^-/- and a full locus-deleted hoxb5a^del/del;hoxb5b^del/delembryos (figure-supplement 4). hoxb5a^-/-;hoxb5b^-/-larvae consistently displayed normal pectoral fins (Figure 7F). In contrast, hoxb5a^del/del;hoxb5b^del/del mutants exhibited a slightly more pronounced phenotype of shortened pectoral fins, but no instances of missing pectoral fins were identified (Figure 7G). Furthermore, due to the functional redundancy observed between hoxb4a and hoxb5a observed in zebrafish vertebral patterning and the RA-dependent expression of hoxb4a (Grandel et al., 2002; Maeno et al., 2024), as well as its expression in fin buds (Figure 7H), we generated frameshift-induced hoxb4a^-/-;hoxb5a^-/-;hoxb5b^-/- embryos and an allele lacking the entire genomic region encompassing hoxb4a and hoxb5a (figure-supplement 4). Although hoxb4a^-/-;hoxb5a^-/-;hoxb5b^-/- embryos did not lack pectoral fins (Figure 7I), some hoxb4a-b5a^del/del;hoxb5b ^del/del embryos did lack pectoral fins, although at low penetrance (Figure 7J, n=3/397; 0.7%). The majority of hoxb4a-b5a^del/del;hoxb5b ^del/del embryos displayed truncated pectoral fins (Figure 7K). The lack of pectoral fins was not detected in hoxb4a-b5a^del/del and hoxb4a^del/del;hoxb5b^del/del mutants (Figure 7L, M, (figure-supplement 4). These results suggest that hoxb4a, hoxb5a, and hoxb5b may cooperatively contribute to anterior-posterior positioning of pectoral fin buds in zebrafish. Furthermore, our findings reveal that transcriptional adaptation may partially play a role in genetic compensation for frameshift-induced hox mutations during pectoral fin formation, while also suggesting the potential involvement of other unknown mechanisms or additional genomic regions in the initial induction of zebrafish pectoral fin buds.

Figure 7.

Discussion

Since the establishment of gene targeting (Mansour et al., 1988), numerous Hox knockout mice have been generated. However, neither single nor compound Hox mutants, nor Hox overexpression approaches, have been reported to cause significant defects in limb positioning (Jurberg et al., 2013; Moreau et al., 2019). Consequently, the role of Hox genes in the positioning of vertebrate paired appendages remains unclear due to a lack of genetic evidence. In this study, we provide genetic evidence using zebrafish, demonstrating that specific combinatorial deletions of hoxba and hoxbb clusters result in the absence of pectoral fins and a lack of induction of tbx5a expression in the pectoral fin field. Based on our results, we propose a model in which Hox expressions in the lateral plate mesoderm, regulated by RA synthesized in the paraxial mesoderm, provide positional information and induce tbx5a expression in the presumptive fin buds. Our genetic results shed light on longstanding fundamental questions regarding the molecular mechanisms by which the positioning of paired appendages is established in vertebrates.

Zebrafish hoxba and hoxbb clusters, possibly hoxb4a, hoxb5a, and hoxb5b, cooperatively define the pectoral fin field in the lateral plate mesoderm

We showed that hoxba;hoxba mutants exhibit a lack of pectoral fins. The abnormal expression of tbx5a in the lateral plate mesoderm, where pectoral fin precursor cells arise, was evident from its initial expression in the mutants. We presume that hox genes in hoxba and hoxbb clusters, induced by RA, define the region along the anterior-posterior axis where pectoral fin formation can occur and promote the expression of tbx5a in the lateral plate mesoderm. Importantly, tbx5a expression in the fin bud was not induced even by RA exposure in hoxba;hoxbb mutants (Figure 4E-H). In zebrafish, it has been suggested that RA synthesized in the paraxial mesoderm acts directly on the lateral plate mesoderm, or that RA induces tbx5a expression via wnt2b in the intermediate mesoderm (Gibert et al., 2006; Grandel and Brand, 2011; Neto et al., 2012). However, our results suggest that the main pathway integrates these signaling cascades through hoxba and hoxbb clusters to induce tbx5a expression. Through genetic studies, despite low penetrance, we showed that hoxb4a, hoxb5a, and hoxb5b play a significant role in the induction of pectoral fin buds. The homologous Hoxb4 and Hoxb5 proteins in mice have been shown to bind directly to the enhancer of Tbx5 (Minguillon et al., 2012). Furthermore, it has been demonstrated that the Tbx5 expression in the forelimb field is restricted by the posterior Hox genes, preventing posterior expansion in mice and chickens (Moreau et al., 2019; Nishimoto et al., 2014). However, our several combinatorial deletions of zebrafish hox clusters did not exhibit any abnormalities in which the expression of tbx5a expanded to the posterior. Therefore, in zebrafish, we propose that hoxb4a, hoxb5a, and hoxb5b, which are induced by RA in the lateral plate mesoderm, simply upregulate the expression of tbx5a.

Partial recapitulation of genomic locus-deleted hox mutants suggests the other unidentified mechanisms

In this study, we observed that hox mutants with deletions of genomic loci exhibited more pronounced pectoral fin phenotypes than frameshift-induced hox mutants. The frameshift mutants used in this study are likely to be loss-of-function alleles, as we demonstrated several defects in vertebral patterning in these mutants (Maeno et al., 2024). These results suggest that transcriptional adaptation may occur during the formation of pectoral fins in frameshift-induced hox mutants, potentially compensating for the loss of function. In our previous studies, which focused on dorsal and anal fin patterning, vertebral patterning, and swim bladder formation, we demonstrated that the phenotypes observed in zebrafish hox cluster-deleted mutants could essentially be replicated by introducing frameshift mutations into hox genes within the cluster (Adachi et al., 2024; Maeno et al., 2024; Satoh et al., 2024; Yamada et al., 2021). Therefore, creating a mutant with a deleted locus that does not induce transcriptional adaptation was unnecessary. One factor that may explain the differences in results between this study and our prior findings is that even a small amount of hox gene products essential for pectoral fin positioning may be sufficient to induce fin buds, as shown in Figure 1, potentially leading to the manifestation of gene compensation through transcriptional adaptation in frameshift-induced mutants. Alternatively, transcriptional adaptation may function differently depending on various developmental processes or stages. Interestingly, all the aforementioned phenotypes reproducible by frameshift mutations were observed at later developmental stages than pectoral fin formation. Additionally, in hox locus-deleted mutants, only some hoxba^-/-;hoxb5b ^de/dell or hoxb4a-b5a^del/del;hoxb5b ^del/del embryos were able to reproduce the loss of the pectoral fin, suggesting the presence of other compensatory mechanisms. The organized expression patterns of Hox genes are thought to be regulated by complex regulatory controls, with their expression regulated according to the genomic order within the cluster. In the absence of Hox function, neighboring intact Hox genes could compensate for the loss of function. It is also possible that unidentified non-coding regions in hoxba and hoxbb clusters are involved in pectoral fin positioning. These issues should be clarified in future studies.

Hox genes may have contributed to the acquisition of pectoral fins during vertebrate evolution

According to fossil records from the Cambrian period, the most primitive vertebrates did not possess paired fins (Janvier, 1996; Shubin et al., 1997). It is believed that, during subsequent evolution, vertebrates acquired primitive pectoral and pelvic fins, which served as evolutionary precursors to tetrapod limbs, before diverging into ray-finned and lobe-finned fishes. Our results align well with this evolutionary scenario. Hox genes, which are conserved in bilaterian animals, are thought to have been present in primitive vertebrates without paired fins. The early fossils of vertebrates with pectoral fins were discovered in Ordovician and Silurian jawless fish (Janvier, 1996; Shubin et al., 1997). Since it is estimated that two rounds of whole-genome duplications occurred early in the Cambrian period (Dehal and Boore, 2005), the first vertebrate with pectoral fins likely possessed four sets of Hox clusters, similar to those found in modern tetrapods. Hox genes have long been fundamental tools for animal development, even before vertebrates acquired paired fins. The acquisition of new functions in specific Hox genes may have contributed to the innovation of novel morphological features. Our findings that hoxba;hoxbb mutants completely lack pectoral fins suggest that molecular evolution within the ancestral HoxB cluster may have contributed to the development of pectoral fins in vertebrates. Although the specific molecular modifications remain unclear, one possible model suggests that HoxB gene(s) acquired a new capacity to induce the expression of Tbx5 in the lateral plate mesoderm. This may have played a role in establishing the RA-HoxB-Tbx5 gene cascade, which is essential for pectoral fin formation. Notably, the ancestral Tbx4/5 in ascidians, which do not have paired fins, already possesses the ability to rescue the loss of forelimbs in mice (Minguillon et al., 2009). Consistent with our idea, a model has been proposed in which the induction of Tbx5 in the lateral plate mesoderm may have contributed to the acquisition of pectoral fins.

Since hoxba;hoxbb mutants are embryonic lethal around 5 dpf, we cannot observe the effects of these clusters on the pelvic fins, which develop around one month after fertilization in zebrafish (Parichy et al., 2009). However, based on the composition of zebrafish hox genes and their expression patterns in hoxba and hoxbb clusters, we presume that these clusters have functions specific to the positioning of pectoral fins. The presence of hox genes responsible for the position of pectoral fins implies that there may be other hox genes that determine the position of pelvic fins and possibly contribute to their acquisition. This remains a critical area for future research.

The positioning of pectoral fins is restricted to specific hox clusters in zebrafish, in contrast to the functional overlap seen in mice

Taking advantage of our availability of seven individual hox cluster mutants in zebrafish, we created mutants with multiple cluster deletions. Through genetic analysis, we demonstrated that hox genes responsible for positioning pectoral fin buds are restricted to hoxba and hoxbb clusters. In contrast, analyses of knockout mice reveal significant functional overlap among the four Hox clusters. Mice lacking all HoxB genes except for Hoxb13 have been generated, yet there are no reports of a phenotype characterized by absent forelimbs (Medina-Martinez et al., 2000). In Hoxb5 knockout mice, the position of the forelimbs shifts rostrally by one vertebra, and forelimbs are not absent in Hoxa5;Hoxb5;Hoxc5 triple knockout mice, which lack all Hox genes of the paralog 5 group (Xu et al., 2013). In comparison, zebrafish hoxba;hoxbb mutants exhibit a striking phenotype of absent pectoral fins. This phenomenon, where function is confined to specific hox clusters, has also been observed in other developmental systems of teleosts. In both zebrafish and medaka, multiple hox genes responsible for positioning dorsal and anal fins are primarily located in HoxC-related clusters (Adachi et al., 2024). As mentioned earlier, it is believed that paired fins in vertebrates emerged after the quadruplication of the primitive Hox clusters. Therefore, Hox genes that acquired new functions related to pectoral fin formation likely originated from specific Hox clusters among the four, maintaining restricted functions that diverged after the teleost-specific whole-genome duplication. The validity of this assumption requires further verification through a clearer understanding of the overall functions of the 49 hox genes in zebrafish. However, the genetic results thus far suggest that, unlike in mice, the functions of hox genes—especially those related to development uniquely acquired in vertebrates—may be confined to specific hox clusters in zebrafish. While teleosts have gained functional redundancy through subsequent teleost-specific whole-genome duplication, it is postulated that, in mice, which have maintained four Hox clusters for over half a billion years, functional overlap among Hox genes may ensure the robustness of the developmental system.

Materials and methods

Zebrafish

Riken Wild-type (RW) zebrafish, obtained from the National BioResource Project in Japan, were maintained at 27°C with a 14-hour light/10-hour dark cycle. Embryos were collected from natural spawning, and the larvae were raised at 28.5°C. Developmental stages of the embryos and larvae were determined based on hours post-fertilization (hpf), days post-fertilization (dpf), and developmental stage-specific features (Kimmel et al., 1995; Parichy et al., 2009). The alleles of hox cluster-deleted mutants used in this study were previously generated using CRISPR-Cas9 (Yamada et al., 2021). The following frameshift-induced hox mutants were created using CRISPR-Cas9: hoxb5a^sud125, hoxb5b ^sud136, hoxb6b ^sud137, and hoxb4a;hoxb5a ^sud144 (Maeno et al., 2024). All experiments involving live zebrafish were approved by the Committee for Animal Care and Use of Saitama University.

Generation of mutants by the CRISPR-Cas9 system

All the mutants used in this study were generated using the Alt-R CRISPR-Cas9 system (Integrated DNA Technologies). To create frameshift-induced mutants, gene-specific crRNAs were incubated with common tracrRNA, followed by Cas9 nuclease. For the generation of mutants that delete large genomic regions, two crRNAs targeting both ends of the regions were incubated with common tracrRNA and Cas9 nuclease. Approximately one nanoliter of the crRNA:tracrRNA-Cas9 RNA-protein complex was injected into fertilized embryos, which were subsequently raised to juvenile fish. Candidate founder fish were selected using a heteroduplex mobility shift assay for frameshift mutants (Ota et al., 2013) and by amplifying genomic deletions with specific primers. After sexual maturation, candidate founder fish were mated with wild-type fish to produce heterozygous F1 offspring. Among the F1 offspring, mutant fish carrying the same mutation were identified through PCR-based genotyping followed by DNA sequencing. The target-specific sequences of the crRNAs used in this study are listed (table-supplement 2). The frozen sperms from the mutants used in this study have been deposited in the National BioResource Project Zebrafish in Japan (https://shigen.nig.ac.jp/zebra/) and are available upon request.

Genotyping of mutants

For the phenotypic analysis of embryos and the maintenance of mutant fish lines, PCR-based genotyping was performed. Briefly, genomic DNA was extracted from the analyzed embryos or the partially dissected caudal fins of anesthetized larvae or fish using the NaOH method (Meeker et al., 2007), which served as the template for PCR. The sequences of the primers used for genotyping frameshift-induced mutants are listed in the supplementary Table 3. For genotyping deletion mutants of hox genes, PCR was conducted using a combination of three primers, with their sequences also provided (table-supplement 3). Genotyping of hox cluster-deleted mutants was performed using PCR as previously described (Yamada et al., 2021). After the reactions, the PCR products were separated by electrophoresis. Based on differences in the lengths of PCR products derived from wild-type and mutated alleles, either a 2 % agarose gel in 0.5x TBE buffer, a 15 % polyacrylamide gel in 0.5x TBE buffer, or direct DNA sequencing was utilized to determine the genotype.

Whole-mount in situ hybridization

Whole-mount in situ hybridization was carried out as described (Thisse and Thisse, 2014). After the staining, the embryos mounted in 70-80 % glycerol were captured under the stereomicroscope (Leica M205 FA) with a digital camera (Leica DFC350F). After taking images, PCR-based genotyping was carried out as described above.

RNA-seq analysis

For RNA-seq analysis, embryos were obtained by the intercrosses between raldh2 heterozygous fish. At the 18-20-somite stages, sibling and homozygous embryos were separated based on the phenotype, and 20 embryos per tube were collected. Then, total RNA was extracted by ISOGENE (Nippon Gene) and followed by DNase I treatment (Takara). After the quality check of the isolated RNA, RNA-seq analyses (n=2 for each) were carried out by Genewiz, Azenta Life Science.

Treatment of retinoic acid in the embryos

Treatment of retinoic acid on the embryos was performed as previously described, with minor modifications (Gibert et al., 2006; Grandel et al., 2002). Briefly, embryos were obtained from intercrosses between raldh2 heterozygous mutants. At the 70 % epiboly stage, the embryos were soaked in a solution containing DMSO or 10^-8M RA and incubated at 28.5°C. Embryos were fixed at 30 hpf for analysis of tbx5a expression and at the 10-somite stage for examination of hox gene expression.

Acknowledgements

We thank the NBRP zebrafish for providing fish and preserving the mutant lines generated in this study. We also appreciate Drs. Koji Tamura and Gembu Abe for providing the zebrafish tbx5a plasmid. This work was supported by KAKENHI Grants-in-Aid for Scientific Research 18K06177, 23K05790 to AK.

Additional files

Supplementary information