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 anteroposterior 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).

Multiple studies in chick and mouse have suggested that the initial anteroposterior position of limbs is regulated by Hox genes (Tanaka, 2013; Tickle, 2015). For example, the anterior boundaries of Hox gene expression domains were shown to align with future limb positions in chick embryos (Burke et al., 1995; Cohn et al., 1997). Experimental manipulations in avian embryos, including overexpression or interference of specific Hox genes, led to altered positions of forelimb buds (Moreau et al., 2019). In mouse, Hoxb5 knockout mutants exhibit a rostral shift of forelimb buds with incomplete penetrance (Rancourt et al., 1995), while other subtle shifts in limb positioning have also been reported in Hox mutants (Royle et al., 2021). Furthermore, loss of Gdf11 in mice, which alters the expression of posterior Hox9–13 genes, causes posterior displacement of hindlimb buds, whereas ectopic Gdf11 induces anterior shifts of hindlimb position (Matsubara et al., 2017). At the molecular level, Hox proteins directly bind to the Tbx5 limb enhancer and regulate its expression, providing a mechanistic link between Hox activity and forelimb initiation (Minguillon et al., 2012; Nishimoto et al., 2014). Together, these studies provide strong evidence across different model organisms that Hox genes contribute to the regulation of limb positioning. Nevertheless, despite the generation of numerous single and compound Hox knockout mice, no severe defects in the initial positioning of limb buds have been documented, and the precise mechanisms by which Hox genes specify limb position remain incompletely understood. This stands in contrast to the well-established role of Hox genes in limb patterning after limb bud formation, where paralogous group 9–13 genes in the HoxA and HoxD clusters cooperatively control the proximal–distal axis of developing limbs (Boulet and Capecchi, 2004; Davis et al., 1995; Fromental-Ramain et al., 1996a; Fromental-Ramain et al., 1996b; Kmita et al., 2005). Consequently, the exact 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 vertebrate class Actinopterygii and are distinct from the tetrapods of the class Sarcopterygii, 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—orthologous 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 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 the pectoral fin field.

Results

Absence of pectoral fins in hoxba;hoxbb cluster-deleted zebrafish

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 Tbx5 orthologs. Zebrafish possess two paralogs. tbx5a and tbx5b (Boyle-Anderson et al., 2022). Among them, tbx5a plays a predominant 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 1G, H). 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, F). In contrast, pectoral fins were present in hoxba^-/-;hoxbb^+/- and hoxba^+/-;hoxbb^-/- mutants (Figure 1D, E), 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 1G-L). 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 pectoral fin formation.

Figure 1.

Pectoral fin loss specific to hoxba;hoxbb cluster-deleted zebrafish

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.

Loss of fin progenitors with posterior expansion of cardiac marker expression

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, abnormal expression patterns of tbx5a were evident at early stages in hoxba;hoxbb cluster mutants. Compared to wild-type embryos, the bilateral stripes of tbx5a expression were shortened 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, consistent with the loss 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 the formation of pectoral fin buds through the induction of tbx5a expression. In zebrafish, studies have shown a reciprocal relationship between pectoral fin and heart progenitor populations in the anterior lateral mesoderm, where retinoic acid signaling promotes pectoral fin formation while restricting cardiac progenitors (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) showed abnormal morphology relative to wild-type embryos (Figure 3I, J) (Yelon et al., 1999). Taken together, these results indicate that hoxba and hoxbb clusters are essential for establishing the pectoral fin field in zebrafish. In their absence, pectoral fin progenitor cells fail to emerge, accompanied by a posterior expansion of cardiac marker expression domain. Our results highlight a shift in the balance between fin and cardiac progenitor populations within the anterior lateral plate mesoderm.

Figure 3.

hoxba/bb clusters required for RA-mediated tbx5a induction 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 indicate that RA-dependent induction of tbx5a does not occur in the absence of 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.

hoxb5a/b5b expression in fin buds is RA-dependent

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). Although hoxc1a showed the most pronounced reduction, our analysis focused on hoxba and hoxbb clusters. 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 screen for hox genes specifying pectoral fin buds

To understand the molecular mechanisms underlying pectoral fin specification, we sought to identify hox genes essential for fin formation through a genetic approach. We anticipated that frameshift mutations in relevant hox genes would phenocopy 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 phenocopy the deletion of hoxbb cluster, recapitulating the absence of pectoral fins in hoxba;hoxbb cluster mutants.

Figure 6.

However, we encountered an unexpected result: 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. Consistent with this, 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 1L). 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 still formed pectoral fins (Figure 6J, M).

The phenotype of hoxba^-/-;hoxb5b^-/- embryos indicates that a frameshift mutation in hoxb5b does not fully mimic the complete deletion of the entire 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). To further clarify which Hox genes could potentially contribute to fin bud formation, we analyzed the expression of PG4 and PG6–8 genes from hoxba and hoxbb clusters using whole-mount in situ hybridization. Among these, only hoxb4a showed detectable expression in the fin buds, whereas hoxb6a, hoxb6b, hoxb7a, hoxb8a, and hoxb8b were not detectable (Figure 7M-R). 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), 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 7H), some hoxb4a-b5a^del/del;hoxb5b ^del/del embryos did lack pectoral fins, although at low penetrance (Figure 7I, n=3/397; 0.7%). The majority of hoxb4a-b5a^del/del;hoxb5b ^del/del embryos displayed truncated pectoral fins (Figure 7J). The lack of pectoral fins was not detected in hoxb4a-b5a^del/del and hoxb4a^del/del;hoxb5b^del/del mutants (Figure 7K, L, figure-supplement 4). These results suggest that hoxb4a, hoxb5a, and hoxb5b may cooperatively contribute to the 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 direct 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.

hoxba/bb clusters (possibly hoxb4a, b5a, b5b) define the pectoral fin field

We showed that hoxba;hoxba mutants exhibit a lack of pectoral fins. Abnormal expression of tbx5a in the lateral plate mesoderm, where pectoral fin precursor cells normally 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 intact hoxba and hoxbb clusters are required for RA-dependent induction of tbx5a expression. One possible explanation is a linear pathway in which RA induces hox expression, which then activates tbx5a. Alternatively, RA may induce hox expression and, together with Hox proteins, act cooperatively to activate tbx5a. Our data are consistent with both models. Through genetic studies, albeit with low penetrance, we further 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). Moreover, in chickens, Tbx5 expression in the forelimb field is restricted by posterior Hox genes, preventing posterior expansion (Moreau et al., 2019; Nishimoto et al., 2014). In contrast, our combinatorial deletions of zebrafish hox clusters did not produce posterior expansion of tbx5a expression. We therefore propose that, in zebrafish, hoxb4a, hoxb5a, and hoxb5b, induced by RA in the paraxial mesoderm, act primarily to promote tbx5a expression in the pectoral fin field.

Partial phenocopy of locus-deleted hox mutants implies 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 and the evolutionary origin of pectoral fins

According to fossil records from the Cambrian period, the earliest vertebrates did not possess paired fins (Janvier, 1996; Shubin et al., 1997). Later in evolution, vertebrates acquired primitive pectoral fins, which served as evolutionary precursors to tetrapod forelimbs, before diverging into ray-finned and lobe-finned fishes. Our results align well with this evolutionary scenario. Hox genes, which are conserved across bilaterian animals, were already present in primitive vertebrates without paired fins, suggesting that their functional diversification may have facilitated the origin of new appendages. The early fossils of vertebrates with pectoral fins were discovered in Ordovician and Silurian jawless fishes (Janvier, 1996; Shubin et al., 1997). Our finding that hoxba;hoxbb mutants completely lack pectoral fins implies that molecular evolution within the ancestral HoxB cluster contributed to the emergence of pectoral fins. Although the precise molecular modifications remain unclear, one plausible model is that HoxB genes acquired the capacity to induce Tbx5 expression in the lateral plate mesoderm, establishing the RA–HoxB–Tbx5 gene network essential for fin initiation. We presume that the acquisition of pectoral fins was driven by multiple genomic changes accumulated over time, among which molecular evolution within the HoxB cluster may have represented one of the principal factors contributing to the origin of this novel appendage.

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 generated multiple cluster deletions and demonstrated that hox genes required for establishing pectoral fin buds are restricted to hoxba and hoxbb clusters. In contrast, analyses of knockout mice have revealed extensive functional overlap among the four Hox clusters. For instance, mice lacking all HoxB genes except for Hoxb13 do not lose forelimbs (Medina-Martinez et al., 2000), and forelimbs are still present in Hoxa5;Hoxb5;Hoxc5 triple knockout mice, which lack all Hox genes of the paralog group 5 (Xu et al., 2013). Thus, while zebrafish hoxba;hoxbb mutants exhibit a striking phenotype of absent pectoral fins, forelimbs are preserved in comparable mouse mutants. We speculate that paired fins emerged after the quadruplication of primitive Hox clusters, and that novel functions related to pectoral fin formation arose within the HoxB cluster. In zebrafish, these functions remain confined to hoxba and hoxbb clusters, making them indispensable for pectoral fin development. Interestingly, a similar cluster-specific confinement has been observed in median fins: in both zebrafish and medaka, Hox genes responsible for positioning dorsal and anal fins are primarily located in HoxC-related clusters (Adachi et al., 2024). In contrast, in mice, functional redundancy among Hox clusters—likely accumulated during over half a billion years of vertebrate evolution—may mask the requirement of HoxB genes for forelimb formation. While the precise evolutionary mechanisms remain to be clarified, our results suggest that cluster-specific specialization in teleosts contrasts with broader redundancy across Hox clusters in mammals.

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 sperm 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.

Data availability

All data necessary to support the conclusions of this article are provided within the main text, figures, and supplementary data. The other data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

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 for in situ hybridization. This work was supported by KAKENHI Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology, Japan (18K06177, 23K05790 to AK).

Additional files

Supplementary Figure and Table

Additional information

Funding

KAKENHI (18K06177)

• Akinori Kawamura

KAKENHI (23K05790)

• Akinori Kawamura