ABSTRACT
The vertebral column is a characteristic structure of vertebrates. Genetic studies in mice have shown that Hox-mediated patterning plays a key role in specifying discrete anatomical regions of the vertebral column. Expression pattern analyses in several vertebrate embryos have provided correlative evidence that the anterior boundaries of Hox expression coincide with distinct anatomical vertebrae. However, because functional analyses have been limited to mice, it remains unclear which Hox genes actually function in vertebral patterning in other vertebrates. In this study, various zebrafish Hox mutants were generated for loss-of-function phenotypic analysis to functionally decipher the Hox code responsible for the zebrafish anterior vertebrae between the occipital and thoracic vertebrae. We found that Hox genes in HoxB- and HoxC-related clusters participate in regulating the morphology of the zebrafish anterior vertebrae. In addition, medaka hoxc6a was found to be responsible for anterior vertebral identity, as in zebrafish. Based on phenotypic similarities with Hoxc6 knockout mice, our results suggest that the Hox patterning system, including at least Hoxc6, may have been functionally established in the vertebral patterning of the common ancestor of ray-finned and lobe-finned fishes.
INTRODUCTION
The vertebral column is a defining feature of the vertebrate body, consisting of repetitive units called vertebrae. Each vertebra is typically composed of the vertebral body (centrum) and the attached vertebral arches (Gadow, 1933; Hilderbrand and Goslow, 1998). Along the anterior-posterior axis, vertebrae in tetrapods are classified into cervical, thoracic, lumbar, sacral and caudal vertebrae based on their morphologies. The vertebral column is derived from the somitic mesoderm. Genetic studies in mice have shown that the nested expression of Hox genes in the somites is crucial for establishing distinct vertebral morphologies (Krumlauf, 1994; McIntyre et al., 2007; Ramirez-Solis et al., 1993; Wellik, 2007; Wellik and Capecchi, 2003). Hox genes, which encode homeodomain transcription factors, are conserved in bilaterians and are generally organized in gene clusters known as Hox clusters. A unique feature of Hox clusters is collinearity, meaning that the positional order of Hox genes within clusters correlates with their expression patterns in embryos along the anterior-posterior axis (Dolle et al., 1989; Duboule and Dolle, 1989; Graham et al., 1989). During the early stages of vertebrate evolution, two rounds of whole-genome duplication resulted in the quadruplication of the primitive Hox cluster (Dehal and Boore, 2005; Ohno, 1970). Mice have four Hox clusters (HoxA-D), each consisting essentially of 1-13 paralogous groups. Loss-of-function studies using knockout mice have demonstrated that Hox genes belonging to paralogous groups 3-11 in four Hox clusters define the anatomically different vertebrae of the spinal column along the anterior-posterior axis, with collinearity generally maintained (Condie and Capecchi, 1994; McIntyre et al., 2007; Wellik and Capecchi, 2003). Compared with the phenotypes of single or double Hox knockout mice, mice lacking all paralogous Hox genes have been shown to exhibit more pronounced vertebral abnormalities, known as the anteriorizing homeotic transformation, in which the more anterior vertebral features are maintained (Chen et al., 1998; McIntyre et al., 2007; van den Akker et al., 2001; Wellik and Capecchi, 2003), suggesting that Hox genes in a paralogous relationship function redundantly to define the identity of the vertebrae in mice.
In other vertebrates, expression pattern analyses using embryos of cartilaginous fish (Criswell et al., 2021), teleosts such as zebrafish (Ahn and Gibson, 1999; Morin-Kensicki et al., 2002; Prince et al., 1998) and various tetrapods (Burke et al., 1995; Cohn and Tickle, 1999; Mansfield and Abzhanov, 2010; Ohya et al., 2005; Woltering et al., 2009) have shown a strong correlation between the anterior boundaries of Hox expression domains in the somites and discrete vertebral morphologies. These results suggest that Hox-mediated patterning of the vertebral column is conserved across gnathostomes. However, although there is accumulated correlative evidence from the expression pattern analyses, loss-of-function analyses have been primarily restricted to mice. Therefore, it remains to be elucidated whether the Hox-mediated vertebral patterning system is indeed functionally conserved in other vertebrates. In addition, it remains unclear whether the loss of function of Hox genes in other vertebrates leads to anterior vertebral transformations, as observed in knockout mice.
In zebrafish, the first to fourth vertebrae, located anterior to the fifth vertebra with the anterior-most ribs, contain the Weberian apparatus. The Weberian apparatus is unique to fishes of series Otophysi (Cypriniformes, Characiformes, Siluriformes and Gymnotiformes) of the superorder Ostariophysi and facilitates the transmission of gas bladder vibrations to the inner ear (Bird and Mabee, 2003; Facey et al., 2022; Grande and Young, 2004; Hastings et al., 2015). In the otophysan fishes, the Weberian apparatus consists of a chain of four small bones (scaphium, claustrum, intercalarium and tripus; see Fig. 1A-D; Movie 1) located in the anterior-most region of the vertebral column. The Weberian ossicles, along with other attached bones, have distinct structures that distinguish each of the first to fourth vertebrae from each other in zebrafish. In a previous study, we created seven individual deficiencies in Hox clusters in zebrafish and provided the first genetic evidence that Hox-mediated vertebral patterning is functionally conserved in zebrafish (Yamada et al., 2021). However, because these mutants lack entire Hox clusters containing multiple Hox genes (Yamada et al., 2021), it remains unclear which specific Hox genes are responsible for vertebral patterning. In this study, we generated various zebrafish Hox mutants and analyzed the morphologies of the anterior vertebrae using micro-computed tomography (CT) scanning. We found the anteriorizing vertebral transformations in several Hox mutants that contribute to the anterior vertebral identities and morphological features of the first to fourth vertebrae. Furthermore, among the Hox genes responsible for the vertebral identities in zebrafish, we show that hoxc6a mutants also exhibit the anteriorizing vertebral transformations in medaka, an evolutionarily distant teleost from zebrafish, indicating that at least the Hoxc6-mediated vertebral patterning mechanism is functionally conserved among teleosts. Additionally, the similarities in phenotypes observed in zebrafish and medaka hoxc6a mutants, as well as Hoxc6 knockout mice, support the notion that Hox-mediated vertebral patterning might have been used in the common ancestor of ray-finned and lobe-finned fishes.
Zebrafish hoxc6a−/−;hoxc6b−/− mutants show severe abnormalities in the third to fourth vertebrae, including the Weberian ossicles. (A-P) The anterior vertebrae were visualized by micro-CT scan in the following group: wild-type (A-D, n=4), hoxc6a−/− (E-H, n=3), hoxc6b−/− (I-L, n=3) and hoxc6a−/−;hoxc6b−/− (M-P, n=4) fish. Adult zebrafish of the same genotype subjected to micro-CT scanning showed essentially similar morphology. Therefore, representative images are shown here. Fish with the desired phenotype were obtained by multiple intercrosses between hoxc6a+/−;hoxc6b+/− fish, and the genotype analysis of the surviving juveniles is shown in Table S1A. The numerical values correspond to the positional order of each vertebra from the first vertebra. In the case of the wild-type zebrafish, unique ossicles attached to each centrum are indicated by a line in A-D. These ossicles attach symmetrically to both sides of the vertebral body. Specifically, the scaphium (sc) is present on the first centrum, and the laterally extending bones known as the lateral process (lp) and intercalarium (ic) are observed on the second centrum. The third centrum possesses a fan-like shaped tripus (tri), and the fourth centrum features a mid-ventrally extending bone known as the os suspensorium (os) and a lateral-ventrally extending bone known as the transverse process of vertebra 4 (tp4). The arrowheads highlight the deformed tripus on the third centrum, which resembles the lateral process on the second centrum in hoxc6a−/− and hoxc6a−/−;hoxc6b−/− mutants. The asterisk indicates the absence of the os suspensorium in hoxc6a−/− and hoxc6a−/−;hoxc6b−/−. The bracket in E,F,M,N indicates the flattened supraneural on the dorsal side of the third centrum in hoxc6a−/− and hoxc6a−/−;hoxc6b−/− mutants. Scale bars: 1 mm. For additional details, micro-CT scan 3D movies are provided in Movies 1,2,3,4. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxc6a−/−;hoxc6b−/− is shown in Fig. S3.
Zebrafish hoxc6a−/−;hoxc6b−/− mutants show severe abnormalities in the third to fourth vertebrae, including the Weberian ossicles. (A-P) The anterior vertebrae were visualized by micro-CT scan in the following group: wild-type (A-D, n=4), hoxc6a−/− (E-H, n=3), hoxc6b−/− (I-L, n=3) and hoxc6a−/−;hoxc6b−/− (M-P, n=4) fish. Adult zebrafish of the same genotype subjected to micro-CT scanning showed essentially similar morphology. Therefore, representative images are shown here. Fish with the desired phenotype were obtained by multiple intercrosses between hoxc6a+/−;hoxc6b+/− fish, and the genotype analysis of the surviving juveniles is shown in Table S1A. The numerical values correspond to the positional order of each vertebra from the first vertebra. In the case of the wild-type zebrafish, unique ossicles attached to each centrum are indicated by a line in A-D. These ossicles attach symmetrically to both sides of the vertebral body. Specifically, the scaphium (sc) is present on the first centrum, and the laterally extending bones known as the lateral process (lp) and intercalarium (ic) are observed on the second centrum. The third centrum possesses a fan-like shaped tripus (tri), and the fourth centrum features a mid-ventrally extending bone known as the os suspensorium (os) and a lateral-ventrally extending bone known as the transverse process of vertebra 4 (tp4). The arrowheads highlight the deformed tripus on the third centrum, which resembles the lateral process on the second centrum in hoxc6a−/− and hoxc6a−/−;hoxc6b−/− mutants. The asterisk indicates the absence of the os suspensorium in hoxc6a−/− and hoxc6a−/−;hoxc6b−/−. The bracket in E,F,M,N indicates the flattened supraneural on the dorsal side of the third centrum in hoxc6a−/− and hoxc6a−/−;hoxc6b−/− mutants. Scale bars: 1 mm. For additional details, micro-CT scan 3D movies are provided in Movies 1,2,3,4. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxc6a−/−;hoxc6b−/− is shown in Fig. S3.
RESULTS
Deciphering functional Hox code in the anterior vertebrae of zebrafish
In the previous study, we generated individual deficiencies in seven zebrafish Hox clusters and found that fish with hoxca cluster homozygous deletion showed an anteriorizing homeotic transformation, in which the third and fourth vertebrae exhibited a morphology similar to that of the second vertebra (Yamada et al., 2021). Given that the first two pairs of zebrafish somites do not contribute to the vertebral column (Morin-Kensicki et al., 2002), we postulated that the third vertebra primarily originates from the fifth somites, which coincide with the anterior boundary of hoxc6a expression domain (Morin-Kensicki et al., 2002; Prince et al., 1998). By using the CRISPR-Cas9 system, we generated frameshift-induced hoxc6a mutants (Fig. S2G) and found that hoxc6a−/− fish essentially recapitulated the vertebral defects observed in hoxca cluster mutants (Fig. 1E-H). In hoxc6a mutants, the tripus on the third vertebra, one of the Weberian ossicles, transformed into a bone resembling the lateral process on the second vertebra (Fig. 1F,G). Furthermore, the os suspensorium extending ventrally from the fourth centrum to the midline was completely absent, while the ventrolaterally extending transverse process of vertebra 4 exhibited the shortening (Fig. 1H). As hoxc6a and hoxc6b were duplicated through teleost-specific whole-genome duplication (Amores et al., 1998), we presumed functional redundancy between them. Accordingly, we generated hoxc6b mutants (Fig. S2H) and hoxc6a;hoxc6b double mutants. Although no obvious morphological defects were detected in hoxc6b mutants (Fig. 1I-L), the vertebral abnormalities in hoxc6a;hoxc6b mutants were enhanced in addition to the phenotypes observed in hoxc6a mutants (Fig. 1M-P). Specifically, a crest-like supraneural is present on the dorsal side of the third and fourth vertebrae in wild-type fish (Fig. 1A). In contrast, hoxc6a mutants showed a reduced supraneural (Fig. 1E), and the most severe case of hoxc6a;hoxc6b mutants displayed a flattened supraneural (Fig. 1M), resembling the dorsal region of the second vertebra. Moreover, in wild-type fish, the dorsal region of the third vertebra exhibited boundaries with laterally interconnected bones (Fig. 1B). However, in the hoxc6a;hoxc6b mutants, these boundaries were absent, and the dorsal bones transformed into a continuous single bone (Fig. 1N), similar to that of the second vertebra in the wild type. Together, hoxc6a;hoxc6b mutants exhibited an anteriorizing homeotic transformation, with the third and fourth vertebrae essentially resembling the second vertebra. Notably, the lateral process in the second vertebra appeared to be normal in hoxc6a;hoxc6b mutants (Fig. 1M-P). Collectively, our findings suggest that the identity of the third and fourth vertebrae in zebrafish is cooperatively determined by hoxc6a and hoxc6b.
We subsequently tried to identify Hox genes responsible for the vertebral identity of the first and second vertebrae. Although hoxc4a or hoxc5a, included in the same hoxca cluster, could be potential candidates (Fig. S1), our previous study revealed that fish with hoxca cluster deletion did not exhibit any abnormalities in the first and second vertebrae (Yamada et al., 2021). Moreover, the other four surviving mutants deficient in Hox clusters (hoxaa, hoxbb, hoxcb and hoxda) also showed no evident abnormalities in the first and second vertebrae (Yamada et al., 2021). It should be noted that functional redundancy exists among mouse Hox paralogs (Chen et al., 1998; McIntyre et al., 2007; van den Akker et al., 2001; Wellik and Capecchi, 2003). Thus, zebrafish Hox genes in these Hox clusters may also have redundant functions with Hox genes in other clusters, even if no abnormalities are observed in mutants lacking the entire Hox cluster. On the other hand, homozygous mutants lacking hoxab or hoxba clusters could not be examined for the anterior vertebrae due to their lethality (Yamada et al., 2021). Notably, hoxba cluster, in contrast to hoxab cluster, contains the same paralog 6 as hoxc6a and hoxc6b (Fig. S1). Zebrafish hoxb6a is expressed in paraxial mesoderm, with its anterior expression boundary found around the fourth somite (Morin-Kensicki et al., 2002; Prince et al., 1998). Therefore, we generated hoxb6a and hoxb6b frameshift-induced mutants (Fig. S2E,F) and analyzed them with respect to the anterior vertebrae using a micro-CT scan. Although hoxb6b mutants did not manifest any abnormalities (Fig. 2I-L), hoxb6a mutants exhibited a short lateral process on the second vertebra in comparison to that of wild-type fish (Fig. 2A-H). In hoxb6a;hoxb6b mutants, the abnormalities were observed where the scaphium and the shortened lateral process coexisted on the same vertebra due to the fusion of the first and second vertebrae (Fig. 2M-P). However, the shortened lateral process observed in hoxb6a mutants was not considerably enhanced in hoxb6a;hoxb6b mutants. For the third and fourth vertebrae, in which substantial morphological abnormalities were detected in hoxc6a;hoxc6b mutants (Fig. 1M-P), no striking morphological abnormality was observed in hoxb6a;hoxb6b mutants. These results suggest that hoxb6a contributes to the identity of the second vertebra, but to a lesser extent, and that hoxb6a may function in concert with other Hox gene(s) in this process.
Zebrafish hoxb6a−/−;hoxb6b−/− mutants exhibit shortening of the lateral process in the second vertebra. (A-P) The anterior vertebrae were analyzed by micro-CT scan in wild-type (A-D, n=4), hoxb6a−/− (E-H, n=2), hoxb6b−/− (I-L, n=2) and hoxb6a−/−;hoxb6b−/− (M-P, n=2) fish. Adult zebrafish of the same genotype subjected to micro-CT scanning showed similar morphology; therefore, representative images are shown. Fish were obtained from multiple intercrosses between hoxb6a+/−;hoxb6b+/− fish, and genotype analysis of the surviving juveniles is shown in Table S1B. The numbering indicates the position of each vertebra from the first vertebra. The arrowheads indicate the reduced lateral process on the second centrum in hoxb6a−/− and hoxb6a−/−;hoxb6b−/− mutants. The brackets signify the vertebra with both scaphium and lateral process. Scale bars: 1 mm. For details, micro-CT scan 3D movies are provided in Movies 5,6,7. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxb6a−/−;hoxb6b−/− is shown in Fig. S3. ic, intercalarium; lp, lateral process; os, os suspensorium; sc, scaphium; tp4, transverse process of vertebra 4; tri, tripus.
Zebrafish hoxb6a−/−;hoxb6b−/− mutants exhibit shortening of the lateral process in the second vertebra. (A-P) The anterior vertebrae were analyzed by micro-CT scan in wild-type (A-D, n=4), hoxb6a−/− (E-H, n=2), hoxb6b−/− (I-L, n=2) and hoxb6a−/−;hoxb6b−/− (M-P, n=2) fish. Adult zebrafish of the same genotype subjected to micro-CT scanning showed similar morphology; therefore, representative images are shown. Fish were obtained from multiple intercrosses between hoxb6a+/−;hoxb6b+/− fish, and genotype analysis of the surviving juveniles is shown in Table S1B. The numbering indicates the position of each vertebra from the first vertebra. The arrowheads indicate the reduced lateral process on the second centrum in hoxb6a−/− and hoxb6a−/−;hoxb6b−/− mutants. The brackets signify the vertebra with both scaphium and lateral process. Scale bars: 1 mm. For details, micro-CT scan 3D movies are provided in Movies 5,6,7. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxb6a−/−;hoxb6b−/− is shown in Fig. S3. ic, intercalarium; lp, lateral process; os, os suspensorium; sc, scaphium; tp4, transverse process of vertebra 4; tri, tripus.
Due to their genomic proximity to hoxb6a and hoxb6b (Fig. S1), our next focus was directed toward hoxb5a and hoxb5b, and we successfully generated their frameshift mutants (Fig. S2C,D). In comparison with the wild type (Fig. 3A-D), a significant reduction in the length of the lateral process on the second vertebra was observed in hoxb5a mutants (Fig. 3E-H), whereas no discernible morphological abnormalities were apparent in hoxb5b mutants (Fig. 3I-L). In the case of hoxb5a;hoxb5b double mutants, enhanced phenotypes were manifested: the lateral process was almost absent, and both the tripus in the third vertebra and the transverse process of the fourth vertebra exhibited a severe asymmetrically deformed morphology (Fig. 3M-P). In addition, the fusion of the third and the fourth centra was detected in some mutants. These phenotypes at the third and fourth vertebrae resemble the phenotypes of hoxc6a;hoxc6b mutants (Fig. 1M-P). In mouse vertebral patterning, it has been observed that mutants with combined Hox mutations exhibit phenotypes similar to posterior Hox mutants (Horan et al., 1995b). These results suggest that hoxb5a and hoxb5b cooperatively participate in determining vertebral identity of the second to fourth vertebrae in zebrafish.
Severe defects in the second to fourth vertebrae of zebrafish hoxb5a−/−;hoxb5b−/−mutants. (A-P) The anterior vertebrae were examined using micro-CT scan in the following groups: wild-type (A-D, n=4), hoxb5a−/− (E-H, n=4), hoxb5b−/− (I-L, n=3) and hoxb5a−/−;hoxb5b−/− (M-P, n=4) adult fish. Representative images are shown. Fish of the desired genotype were obtained by several intercrosses between hoxb5a+/−;hoxb5b+/− fish and the genotype analysis of the surviving juveniles is shown in Table S1C. The numerical values correspond to the positional order of each vertebra from the first vertebra. The arrowheads indicate a significantly shortened lateral process on the second centrum in hoxb5a−/− and hoxb5a−/−;hoxb5b−/− mutants. The brackets indicate malformation of the ossicles on the second to fourth centrum in hoxb5a−/−;hoxb5b−/− mutants. The asterisk indicates the absence of the os suspensorium in hoxb5a−/−;hoxb5b−/− mutants. Scale bars: 1 mm. For additional details, micro-CT scan 3D movies are provided in Movies 8,9,10. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxb5a−/−;hoxb5b−/− is shown in Fig. S3. ic, intercalarium; lp, lateral process; os, os suspensorium; sc, scaphium; tp4, transverse process of vertebra 4; tri, tripus.
Severe defects in the second to fourth vertebrae of zebrafish hoxb5a−/−;hoxb5b−/−mutants. (A-P) The anterior vertebrae were examined using micro-CT scan in the following groups: wild-type (A-D, n=4), hoxb5a−/− (E-H, n=4), hoxb5b−/− (I-L, n=3) and hoxb5a−/−;hoxb5b−/− (M-P, n=4) adult fish. Representative images are shown. Fish of the desired genotype were obtained by several intercrosses between hoxb5a+/−;hoxb5b+/− fish and the genotype analysis of the surviving juveniles is shown in Table S1C. The numerical values correspond to the positional order of each vertebra from the first vertebra. The arrowheads indicate a significantly shortened lateral process on the second centrum in hoxb5a−/− and hoxb5a−/−;hoxb5b−/− mutants. The brackets indicate malformation of the ossicles on the second to fourth centrum in hoxb5a−/−;hoxb5b−/− mutants. The asterisk indicates the absence of the os suspensorium in hoxb5a−/−;hoxb5b−/− mutants. Scale bars: 1 mm. For additional details, micro-CT scan 3D movies are provided in Movies 8,9,10. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxb5a−/−;hoxb5b−/− is shown in Fig. S3. ic, intercalarium; lp, lateral process; os, os suspensorium; sc, scaphium; tp4, transverse process of vertebra 4; tri, tripus.
Notably, the scaphium on the first vertebra appeared to be normal across the aforementioned Hox mutants (Figs 1-3). One explanation is that the first vertebra in zebrafish is in a Hox-free default state, as was observed in the anterior-most region of the rhombomeres and pharyngeal arches (Lumsden and Krumlauf, 1996; Minoux et al., 2009). Alternatively, considering that the identity of the first vertebra is defined by Hox genes in mice (Condie and Capecchi, 1994), it is plausible that Hox also plays a defining role in zebrafish first vertebra. The presence of Hox genes with the anterior boundary of expression in the anterior somites (Prince et al., 1998) supports the hypothesis that the identity of the first vertebra in zebrafish is determined by other Hox gene(s).
Given its genomic proximity to hoxb5a (Fig. S1), we turned our attention to hoxb4a as a potential candidate responsible for the first vertebra. Although hoxb4a mutants were isolated (Fig. S2A), no morphological abnormalities were observed in the anterior vertebrae (Fig. 4A-D). Considering the functional redundancy between hoxb4a and hoxb5a due to their structural similarities, we generated hoxb4a;hoxb5a double mutants (Fig. S2B). Micro-CT scan analysis of hoxb4a;hoxb5a mutants clearly revealed the absence of the scaphium (Fig. 4E-H), with the space caused by its absence seemingly occupied by bone elongation originating from the second vertebra (Fig. 4E). Taken together with results showing that the scaphium was not significantly affected in hoxb5a mutants (Fig. 3E-H), our findings suggest that the identity of the first vertebra is cooperatively defined by hoxb4a and hoxb5a in zebrafish, although the contribution of other paralogous Hox genes remains to be elucidated. Additionally, we observed that the anterior-most ribs on the fifth centrum were abbreviated in hoxc8a mutants (Fig. 4I,J; Fig. S2I), suggesting that hoxc8a is implicated in defining the fifth vertebra along its foremost pleural ribs. In summary, our loss-of-function-based phenotypic analyses have delineated a framework for Hox code activity in determining the identities of the first through to fifth vertebrae in zebrafish (Fig. 6A,B).
The absence of the scaphium in the first vertebra in hoxb4a−/−;hoxb5a−/− mutants. (A-H) The anterior vertebrae were examined by micro-CT scan in hoxb4a−/− (A-D, n=4) and hoxb4a−/−;hoxb5a−/− (E-H, n=5) adult fish. Representative images are shown. For hoxb4a−/− and hoxb4a−/−;hoxb5a−/− mutants, fish with the desired genotypes were obtained by multiple intercrosses between each heterozygous fish, and the genotype analysis of the surviving juveniles is shown in Tables S1D,E. The arrowheads indicate the absence of scaphium on the first centrum in hoxb4a−/−;hoxb5a−/−. The asterisk indicates the elongated bone dorsal to the second centrum. Digital dissection of each vertebra in hoxb4a−/−;hoxb5a−/− is shown in Fig. S3. (I,J) Trunk region of wild-type (n=3) and hoxc8a−/− fish (n=5). Adult fish were obtained by several intercrosses between hoxc8a+/− fish, and genotype analysis of the surviving juveniles is shown in Table S1F. The arrowheads indicate the tip of the anterior-most rib. The rib shortening was observed in three of the five hoxc8a mutants. Such rib shortening was not detected in the other Hox mutants isolated in this study. Scale bars: 1 mm in A-H; 2 mm in I,J. For detailed information, micro-CT scan 3D movies are provided in Movies 11,12,13. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxb4a−/−;hoxb5a−/− is shown in Fig. S3. ic, intercalarium; lp, lateral process; os, os suspensorium; sc, scaphium; tp4, transverse process of vertebra 4; tri, tripus.
The absence of the scaphium in the first vertebra in hoxb4a−/−;hoxb5a−/− mutants. (A-H) The anterior vertebrae were examined by micro-CT scan in hoxb4a−/− (A-D, n=4) and hoxb4a−/−;hoxb5a−/− (E-H, n=5) adult fish. Representative images are shown. For hoxb4a−/− and hoxb4a−/−;hoxb5a−/− mutants, fish with the desired genotypes were obtained by multiple intercrosses between each heterozygous fish, and the genotype analysis of the surviving juveniles is shown in Tables S1D,E. The arrowheads indicate the absence of scaphium on the first centrum in hoxb4a−/−;hoxb5a−/−. The asterisk indicates the elongated bone dorsal to the second centrum. Digital dissection of each vertebra in hoxb4a−/−;hoxb5a−/− is shown in Fig. S3. (I,J) Trunk region of wild-type (n=3) and hoxc8a−/− fish (n=5). Adult fish were obtained by several intercrosses between hoxc8a+/− fish, and genotype analysis of the surviving juveniles is shown in Table S1F. The arrowheads indicate the tip of the anterior-most rib. The rib shortening was observed in three of the five hoxc8a mutants. Such rib shortening was not detected in the other Hox mutants isolated in this study. Scale bars: 1 mm in A-H; 2 mm in I,J. For detailed information, micro-CT scan 3D movies are provided in Movies 11,12,13. A summary of the phenotypes of these mutants is shown in Table S2. Digital dissection of each vertebra in hoxb4a−/−;hoxb5a−/− is shown in Fig. S3. ic, intercalarium; lp, lateral process; os, os suspensorium; sc, scaphium; tp4, transverse process of vertebra 4; tri, tripus.
Medaka hoxc6a mutants exhibit vertebral transformation
The Weberian apparatus is a unique vertebral structure that has developed in fishes of series Otophysi of the superorder Ostariophysi (Bird and Mabee, 2003; Facey et al., 2022; Grande and Young, 2004). Therefore, it cannot be concluded from our zebrafish results alone whether the Hox patterning system in the anterior vertebrae is conserved among teleosts. To investigate this further, we turned to medaka (Acanthopterygii), a teleost fish that is evolutionarily distant from zebrafish and estimated to have diverged from the Ostariophysi (zebrafish) over 200 million years ago (Hastings et al., 2015; Hughes et al., 2018; Kasahara et al., 2007). Both medaka and zebrafish possess seven Hox clusters due to the teleost-specific whole-genome duplication. However, zebrafish lack the hoxdb cluster, whereas medaka lack the hoxcb cluster (Amores et al., 1998; Kuraku and Meyer, 2009; Kurosawa et al., 2006), indicating the very different histories of duplicate resolutions in the lineage leading to two teleosts. In our study with medaka, we focused on hoxc6a, homologous to one of the Hox genes responsible for the vertebral identity of the anterior vertebrae in zebrafish. Notably, we observed severe abnormalities, including part of the Weberian apparatus, in the anterior vertebrae of zebrafish hoxc6a;hoxc6b mutants (Fig. 1M-P). Taking advantage of the absence of hoxcb clusters in medaka, we generated medaka hoxc6a mutants (Fig. S4). Micro-CT scan analysis revealed that wild-type medaka have a simpler structure in their anterior vertebrae compared with zebrafish (Fig. 5A,B). In medaka, the epipleural, a slender bone extending from the centrum and parapophyses attached to the centrum was observed from the first vertebra, whereas the pleural ribs were observed from the second vertebra. Interestingly, in medaka hoxc6a mutants, the pleural ribs were absent in the second vertebra but appeared from the third vertebra, leading to the anterior vertebral transformation from the second to the first vertebra (Fig. 5C,D). As hoxc6a homologs in both zebrafish and medaka are essential for anterior vertebral patterning, our results suggest that hoxc6-mediated vertebral patterning is functionally conserved among the teleost fishes.
The second vertebra of medaka hoxc6a−/− mutants shows abnormalities without ribs, similar to the first vertebra. (A-D) Micro-CT scan analysis of the anterior vertebrae in wild-type (n=3) and hoxc6a−/− (n=3) adult medaka. As similar phenotypes were observed for all mutants of the same genotype, a representative individual is shown. The arrowheads indicate the anterior-most pleural ribs attached to vertebrae. In wild-type medaka, the ribs are present from the second vertebra (n=3), but in hoxc6a mutants, the ribs are observed from the third vertebra (n=3). Micro-CT scan 3D movies are provided in Movies 14,15. Scale bars: 1 mm. ep, epipleurals; pp, parapophysis.
The second vertebra of medaka hoxc6a−/− mutants shows abnormalities without ribs, similar to the first vertebra. (A-D) Micro-CT scan analysis of the anterior vertebrae in wild-type (n=3) and hoxc6a−/− (n=3) adult medaka. As similar phenotypes were observed for all mutants of the same genotype, a representative individual is shown. The arrowheads indicate the anterior-most pleural ribs attached to vertebrae. In wild-type medaka, the ribs are present from the second vertebra (n=3), but in hoxc6a mutants, the ribs are observed from the third vertebra (n=3). Micro-CT scan 3D movies are provided in Movies 14,15. Scale bars: 1 mm. ep, epipleurals; pp, parapophysis.
DISCUSSION
In this loss-of-function study focusing on Hox genes in HoxB- and HoxC-related clusters, we have identified several Hox genes responsible for patterning the anterior vertebrae of zebrafish. By generating frameshift-induced mutants for each of eight Hox genes and a series of compound mutants, we demonstrated that zebrafish Hox genes, such as hoxb4a, hoxb5a, hoxb5b, hoxb6a, hoxc6a and hoxc6b, which are homologous to mouse Hox genes responsible for vertebral patterning, also play a crucial role in patterning zebrafish anterior vertebrae (summarized in Fig. 6A,B; Fig. S3; Table S1). As the zebrafish anterior vertebrae, including the Weberian ossicles, are highly specialized in otophysan fishes (Bird and Mabee, 2003; Facey et al., 2022; Grande and Young, 2004), it is necessary to investigate whether the Hox genes responsible for the identities of the zebrafish anterior vertebrae are conserved among teleost fishes. To investigate this further, we used medaka (Acanthopterygii) and demonstrated that at least hoxc6a is also important for patterning the anterior vertebrae in medaka (Fig. 5), as well as in zebrafish, suggesting that hoxc6a-mediated vertebral patterning is conserved in teleosts. On the other hand, in Hoxc6 knockout mice, the anterior thoracic vertebrae have been shown to exhibit anterior vertebral transformation (Garcia-Gasca and Spyropoulos, 2000). This phenotype is similar to the zebrafish and medaka phenotypes in terms of the anterior vertebrae, revealed in this study. Furthermore, Hoxc6 expression has been shown to correspond to the cervical-thoracic transition region in several tetrapods (Burke et al., 1995; Mansfield and Abzhanov, 2010; Ohya et al., 2005; Woltering et al., 2009). Combining our loss-of-function analysis using zebrafish and medaka with previous knockout studies in mice, we propose that the Hox-mediated vertebral patterning system, including at least Hoxc6, was functionally established in a common ancestor of ray-finned and lobe-finned fishes, despite several differences in vertebral components and ossification in several aspects (Fleming et al., 2015).
Hox genes responsible for the anterior vertebral identities in mouse, zebrafish and medaka. (A) Diagrams illustrating the typical morphologies of the anterior vertebrae in the zebrafish Hox mutants isolated in this study. The bones that are observed to be abnormal in Hox mutants are highlighted in red, and the missing bones are indicated by dotted lines. Ventral view. (B) Comparisons of Hox genes responsible for the anterior vertebrae in mouse, zebrafish and medaka. Each vertebra is represented by a box, and a triangle attached to the box indicates thoracic vertebrae with ribs. In mice, Hox genes that define the vertebral identities are shown based on malformed vertebrae in Hox knockout mice. In mice, Hox genes from paralog 3 are responsible for vertebral patterning. The specific Hox gene is shown below the vertebrae that are abnormal in individual Hox knockout mice: Hoxb3 (Manley and Capecchi, 1997), Hoxd3 (Condie and Capecchi, 1993), Hoxa4 (Horan et al., 1994; Kostic and Capecchi, 1994), Hoxb4 (Ramirez-Solis et al., 1993), Hoxc4 (Saegusa et al., 1996), Hoxd4 (Horan et al., 1995a), Hoxa5 (Jeannotte et al., 1993), Hoxb5 (Rancourt et al., 1995), Hoxa6 (Kostic and Capecchi, 1994), Hoxb6 (Rancourt et al., 1995) and Hoxc6 (Garcia-Gasca and Spyropoulos, 2000). Multiple paralogous Hox knockout mice showed more severe vertebral abnormalities than single knockout mice. For paralog 3, which has three paralogs, no triple Hox mutants have been reported, but double mutants combining two of the three mutations showed vertebrae with abnormalities (Condie and Capecchi, 1994; Manley and Capecchi, 1997). For paralog 4, triple mutants were reported among the four paralogs, showing C2-C5 vertebrae with abnormalities (Horan et al., 1995b). For paralogs 5 and 6, triple mutants lacking all paralogs have been reported with abnormal vertebrae (McIntyre et al., 2007). In zebrafish, Hox genes that define specific vertebrae, as revealed in this study, are shown below the vertebrae. The frame of the box is black where vertebral abnormalities were observed in double Hox mutants. In medaka, hoxc6a is responsible for the second vertebra.
Hox genes responsible for the anterior vertebral identities in mouse, zebrafish and medaka. (A) Diagrams illustrating the typical morphologies of the anterior vertebrae in the zebrafish Hox mutants isolated in this study. The bones that are observed to be abnormal in Hox mutants are highlighted in red, and the missing bones are indicated by dotted lines. Ventral view. (B) Comparisons of Hox genes responsible for the anterior vertebrae in mouse, zebrafish and medaka. Each vertebra is represented by a box, and a triangle attached to the box indicates thoracic vertebrae with ribs. In mice, Hox genes that define the vertebral identities are shown based on malformed vertebrae in Hox knockout mice. In mice, Hox genes from paralog 3 are responsible for vertebral patterning. The specific Hox gene is shown below the vertebrae that are abnormal in individual Hox knockout mice: Hoxb3 (Manley and Capecchi, 1997), Hoxd3 (Condie and Capecchi, 1993), Hoxa4 (Horan et al., 1994; Kostic and Capecchi, 1994), Hoxb4 (Ramirez-Solis et al., 1993), Hoxc4 (Saegusa et al., 1996), Hoxd4 (Horan et al., 1995a), Hoxa5 (Jeannotte et al., 1993), Hoxb5 (Rancourt et al., 1995), Hoxa6 (Kostic and Capecchi, 1994), Hoxb6 (Rancourt et al., 1995) and Hoxc6 (Garcia-Gasca and Spyropoulos, 2000). Multiple paralogous Hox knockout mice showed more severe vertebral abnormalities than single knockout mice. For paralog 3, which has three paralogs, no triple Hox mutants have been reported, but double mutants combining two of the three mutations showed vertebrae with abnormalities (Condie and Capecchi, 1994; Manley and Capecchi, 1997). For paralog 4, triple mutants were reported among the four paralogs, showing C2-C5 vertebrae with abnormalities (Horan et al., 1995b). For paralogs 5 and 6, triple mutants lacking all paralogs have been reported with abnormal vertebrae (McIntyre et al., 2007). In zebrafish, Hox genes that define specific vertebrae, as revealed in this study, are shown below the vertebrae. The frame of the box is black where vertebral abnormalities were observed in double Hox mutants. In medaka, hoxc6a is responsible for the second vertebra.
This study provides genetic evidence that Hox genes regulate the characteristic skeletons in the anterior vertebrae of zebrafish, including the Weberian apparatus, which is acquired in the Otophysi lineage. In zebrafish hoxc6a;hoxc6b mutants, the tripus, one of the Weberian ossicles attached to the third centrum, and the os suspensorium on the fourth centrum, are severely deformed or lacking (Fig. 1). Similarly, medaka hoxc6a mutants exhibit vertebral transformation in the anterior-most thoracic vertebrae (Fig. 5). These results provide molecular evidence supporting the idea that the tripus and os suspensorium have developed their own morphology through the deformation of the rib or parapophysis in the thoracic vertebrae. With this morphological transition, the vertebra with the anterior-most ribs was changed to be defined by hoxc8a in zebrafish, different from medaka using hoxc6a. It is intriguing to understand the evolution of the molecular mechanism by which otophysan fishes acquired these highly distinct bones. In contrast to fishes of the Otophysi, Gonorynchiformes, which also belong to the superorder Ostariophysi, are known to have a primitive Weberian apparatus (Hastings et al., 2015). One possibility is that the gene network downstream of hoxc6 underwent gradual changes during the evolution of ostariophysan fishes, leading to the acquisition of these unique bones. Alternatively, Hox genes may have been newly integrated into the vertebrate patterning system in the ostariophysan lineage, leading to the formation of unique bones. To clarify this, it would be beneficial to investigate the conserved and fundamental Hox patterning mechanism in the anterior vertebrae of teleosts, using medaka or other ostariophysan fishes as a comparison with zebrafish.
Compared with zebrafish, the structure of the medaka anterior vertebrae is simple. Based on the zebrafish results, we generated medaka hoxc6a mutants and observed an anteriorizing vertebral transformation, in which ribs that normally develop from the second centrum instead develop from the third centrum (Fig. 5). However, the parapophyses, which enlarge from the second vertebra, remain almost unchanged in medaka hoxc6a mutants. Additionally, the morphology of the neural arch on the dorsal side of the second vertebra differs from that of the first vertebra. However, it is also unaffected in medaka hoxc6a mutants. These results indicate that medaka hoxc6a mutants exhibit only a partial vertebral anteriorization, specifically limited to the ribs, and do not significantly affect other characteristic morphologies. Therefore, it is possible that other Hox genes are responsible for the identity of the medaka second vertebra, and they are likely to be homologous to Hox genes identified in zebrafish in this study. Using the zebrafish results as a clue, the functional identification of Hox genes essential for the patterning of medaka vertebrae, which are more similar to typical teleost fish vertebrae, is crucial for understanding the molecular mechanism of vertebral patterning in teleosts.
Previous analyses using knockout mice have demonstrated that all of the paralogous 3-11 Hox genes in the four Hox clusters contribute to vertebral patterning (Condie and Capecchi, 1994; McIntyre et al., 2007; Wellik and Capecchi, 2003). It is well established that there is functional redundancy among paralogous Hox genes in mice (Chen et al., 1998; McIntyre et al., 2007; van den Akker et al., 2001; Wellik and Capecchi, 2003). In this study, we show that zebrafish Hox genes of the HoxB- and HoxC-related clusters play important roles in the anterior vertebral identity. However, it should be noted that the contribution of zebrafish HoxA- and HoxD-related clusters in vertebral patterning remains unknown. In our previous study, we isolated zebrafish mutants lacking the entire region of hoxaa or hoxda clusters, but no abnormalities were found in the anterior vertebrae (Yamada et al., 2021). Furthermore, zebrafish hoxab cluster mutants are lethal, and vertebral analysis was not performed. However, the hoxab cluster does not contain paralogs 3-8 (Fig. S1), which are involved in anterior vertebral patterning and are therefore unlikely to contribute. Therefore, it is possible that functional redundancy among paralogs, as observed in mice, may explain the absence of vertebral abnormalities in zebrafish hoxaa or hoxda cluster-deficient mutants. Further investigation is required to elucidate whether HoxA/D-related clusters play a role in zebrafish vertebral patterning.
The present study lays the groundwork for future investigations of vertebral patterning in teleosts. Understanding the complete picture of the functional Hox code for vertebral patterning in zebrafish is crucial. It will shed light on whether the Hox code for vertebral patterning, which has been accumulated by many genetic studies in mice, was established before the divergence of lobe-finned and ray-finned fishes and maintained in their subsequent lineages. Alternatively, the possibility of changes in the ancient Hox code for vertebral patterning during vertebrate evolution will be explored. Of particular interest is the question of whether Hox genes responsible for cervical vertebrae in mice also play a role in vertebral patterning in ray-finned fishes, as there are a few teleost vertebrae that could be potentially homologous to tetrapod cervical vertebrae. A recent study examining the expression patterns of several Hox genes in the skate, a cartilaginous fish, suggests that Hox-mediated vertebral patterning was established before the divergence of bony and cartilaginous fishes (Criswell et al., 2021). Integrating these studies will be important in determining when the primitive Hox code responsible for vertebrate patterning emerged during vertebrate evolution, the state of the primitive Hox code, and whether the Hox vertebral patterning system was highly constrained or modifiable during vertebrate evolution.
MATERIALS AND METHODS
Zebrafish and medaka
Riken wild-type (RW) zebrafish were obtained from the National BioResource Project Zebrafish (NBRP Zebrafish) and reared under controlled conditions at 27°C with a 14 h light/10 h dark cycle. Embryos were collected by natural spawning and larvae were reared at 28.5°C. All experiments using live zebrafish were performed in accordance with the regulations of the Animal Care and Use Committee of Saitama University.
Medaka Oryzias latipes (Hd-rR), provided by National BioResource Project Medaka (NBRP Medaka), were maintained at 25°C with a 14 h light/10 h dark cycle. All experiments using medaka were conducted under the guidelines of the Institutional Animal Care and Use Committee of Utsunomiya University.
Generation of zebrafish and medaka Hox mutants using CRISPR-Cas9
Hox mutants were isolated using essentially the same procedures in zebrafish and medaka. The Alt-R CRISPR-Cas9 system (Integrated DNA Technologies) was used to introduce the frameshift mutations into the coding sequence of the target Hox genes. For zebrafish, the crispr RNA (crRNA) library, which was pre-designed by Integrated DNA Technologies, was used to select specific crRNAs for each Hox gene. For medaka, crRNA against hoxc6a was selected using custom crRNA design algorithms (Integrated DNA Technologies). The target-specific sequences of the crRNAs used in this study are listed in Table S3. Before microinjection, Hox-specific crRNA was incubated with common tracrRNA, followed by Cas9 nuclease. Approximately 1 nl of the crRNA:tracrRNA-Cas9 RNA-protein complex was injected into fertilized wild-type zebrafish or medaka embryos, and the injected embryos were carefully reared to the juvenile stage. Candidate juvenile fish harboring the mutation were selected using the heteroduplex mobility shift assay (Ota et al., 2013). Upon reaching sexual maturity, candidate founder fish were mated with wild-type fish to produce heterozygous F1 progeny. Fish carrying frameshift mutations were then identified using the heteroduplex mobility shift assay (Ota et al., 2013) and DNA sequencing. For zebrafish hoxb4a;hoxb5asud144, the frameshift mutation in hoxb4a was introduced into the same allele of hoxb5asud124. Frozen sperm from all the Hox mutants described in this study have been deposited in the NBRP Zebrafish or in the NBRP Medaka and can be requested upon completion of the material transfer agreements.
Genotyping of zebrafish and medaka Hox mutants
PCR-based genotyping of zebrafish and medaka Hox mutants was performed. Genomic DNA was extracted from a partially dissected tail fin of an anesthetized zebrafish and medaka and used as a template. Genotyping of frameshift-induced Hox mutants was determined by PCR using the primers listed in Table S4. After the reactions, the PCR products were separated by electrophoresis on 2% agarose gel or 15% polyacrylamide gel in 0.5× TBE buffer. For the adult zebrafish undergoing X-ray CT scan analysis, the homozygous mutations identified by electrophoresis were further validated by DNA sequencing.
X-ray micro-CT scans of the anterior vertebrae of zebrafish and medaka
X-ray micro-CT scanning of the anterior vertebrae was performed using essentially the same procedures in adult zebrafish and medaka. Specimens were fixed in 4% paraformaldehyde in PBS overnight and then transferred to 70% ethanol. An X-ray micro-CT (ScanXmate-E090S105; Comscantechno) was used to scan the fixed specimens at a tube voltage peak of 85 kV and a tube current of 90 µA. For scanning of the anterior vertebrae in zebrafish and medaka, the specimens were rotated 360° in 0.24° steps, generating 1500 projection images of 992×992 pixels. Micro-CT data were reconstructed using coneCTexpress software (Comscantechno) with an isotropic resolution of 5.0 μm for zebrafish and 3.8 μm for medaka. Three-dimensional image analysis was performed using OsiriX MD software (Pixmeo) for zebrafish and medaka. Finally, the movies were edited using Adobe Premiere Pro (Adobe) and DaVinci Resolve (Blackmagic Design). In our observations of the anterior vertebrae, the bone morphology of adult zebrafish and medaka showed no discernible differences between males and females.
Digital dissection of the zebrafish anterior vertebrae
Based on the data of the first to fourth vertebrae obtained from micro-CT scans, each zebrafish anterior vertebra was manually dissected using the Imaris contour surface feature (in the Measurement Pro) and displayed in a 3D image using Imaris v10.0 (Bitplane).
Acknowledgements
We thank the NBRP zebrafish and NBRP medaka for providing fish and preserving the mutant lines isolated in this study.
Footnotes
Author contributions
Conceptualization: A.K.; Methodology: A.M.; Validation: A.M., R.K., H.N., R.F., K.Y., S.O., T.T., M.I., K.S., A.K.; Formal analysis: A.M., A.K.; Investigation: A.M., R.K., H.N., R.F., K.Y., S.O., T.T., M.I., K.S., A.I., T.S., U.A., M.K., N.I., M.M., A.K.; Resources: M.M.; Writing - original draft: A.K.; Visualization: A.M., R.F.; Supervision: A.K.; Project administration: A.K.; Funding acquisition: A.K.
Funding
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 A.K.), by the National Institute of Genetics under the Joint Research and Research Meeting (NIG-JOINT) program (38A2019, 7A2020, 66A2021, 18A2022, 31A2023 to A.K.) and by the Narishige Zoological Science Award 2021 to A.K.
Data availability
All relevant data can be found within the article and its supplementary information.
References
Competing interests
The authors declare no competing or financial interests.