RESULTS AND DISCUSSION
Heparan sulfate proteoglycans (HSPGs) are constituents of the extracellular matrix and play important roles during embryogenesis and cancer formation. They exert their function by regulating various signaling pathways, including Wingless/Int (Wnt), fibroblast growth factor (FGF), transforming growth factor (TGF) -β and Hedgehog (Hh; Blackhall et al.,2001; Nybakken and Perrimon,2002; Kramer and Yost,2003). The synthesis of HSPGs depends on the addition of a glycosaminoglycan chain to a core protein. This reaction is catalyzed by glycosyltransferases belonging to the EXT family (Lind et al.,1998). In humans, mutations in these genes cause HME, benign tumors at the growths caps of the long bones (Zak et al.,2002). To date, several members of the EXT family have been cloned and identified in humans. These members include EXT1, EXT2, EXT3, and three EXT-like genes (Cook et al.,1993; Le Merrer et al.,1994; Hecht et al.,1995; Wise et al.,1997; Wuyts et al.,1997; Van Hul et al.,1998). In addition, orthologous genes for EXT1 and EXT2 were identified in mouse, Xenopus, Caenorhabditis elegans, and Drosophila, where the EXT1 homolog was named tout-velu (ttv; Clines et al.,1997; Lin and Wells,1997; Stickens and Evans,1997; Bellaiche et al.,1998; The et al.,1999; Han et al.,2001; Katada et al.,2002). So far, only one member of each family has been identified in all species examined. In contrast to this, we describe the identification of three members of the EXT1 family in the zebrafish genome.
Cloning of Zebrafish ext1a, ext1b, and ext1c
We identified zebrafish ext1 genes by performing a BLAST search of the publicly available zebrafish databases with mouse EXT1. From this search, we retrieved three sequences, which we amplified by polymerase chain reaction (PCR) and subsequent rapid amplification of cDNA ends (RACE) PCR to generate the full-length open reading frames (ORF; see Experimental Procedures section). The predicted amino acid sequences of these proteins showed a high homology to known EXT1 proteins of other species (Fig. 1B). EXT1a shows 81% identity to mammalian EXT1 proteins, whereas EXT1b shows 83% identity. In contrast to these high values of identity, EXT1c shows only 62% identity. Zebrafish EXT1a and EXT1b show 88% identity among themselves and only 61% identity to EXT1c. Of interest, these values are lower than the respective identities within the mammalian clade. Here, EXT1 proteins show 95% identity. These values differ greatly within the protein. After the first seven amino acids, which are conserved in all species, and the putative transmembrane domain, the N-terminus up to amino acid 115 is less well-conserved. This portion is thought to form a putative stem region, which separates the signal sequence from the catalytic portion of the protein (Wei et al.,2000). Several mutations have been characterized in EXT1 genes affecting the glycosyltransferase activity (Wei et al.,2000). The amino acids altered in these mutants are conserved in the zebrafish EXT1 paralogs (Fig. 1A, asterisks), underlining the evolutionary conservation of all three proteins. In addition, the phylogenetic tree (Fig. 1C) groups all three zebrafish EXT1 proteins in the EXT1 clade with high bootstrap support. The phylogenetic analysis furthermore suggests that ext1a and ext1b arose during a more recent duplication event in the zebrafish lineage, whereas ext1c is more diverged. Thus, these findings indicate that ext1a, ext1b, and ext1c are the orthologous zebrafish genes of other known EXT1 genes. To analyze their expression during embryonic development, we performed in situ hybridization with riboprobes for the respective genes. As a control, we performed in situ hybridizations with sense probes of each gene that did not produce a staining (data not shown).
Expression of ext1a
Ext1a message is maternally provided (data not shown) and present at sphere stage (Fig. 2A). At approximately 80% of epiboly, ext1a is expressed in two ectodermal wings animal to the germ ring. These areas exclude the most dorsal structures, such as the axial mesoderm (Fig. 2B,C). At the three-somite stage, ext1a message is present in prospective forebrain regions, while in putative midbrain regions, it marks the outline of the neural plate (Fig. 2G). Furthermore, three stripes of ext1a expressing cells can be detected in the forming hindbrain. More posteriorly, ext1a is expressed in the tail bud and in the axial mesoderm (Fig. 2G). From the five-somite stage onward, ext1a RNA can be detected in the somitic mesoderm. This expression continues during somitogenesis (Fig. 3A). Cross-sections at the 16-somite stage reveal staining in two ventromedial domains of the somites (Fig. 3B, black arrowheads) and in a dorsomedial compartment, adjacent to the neural tube (Fig. 3B, black arrows). In addition to the somitic expression, ext1a message can be detected in a segmental manner in the dorsal neural tube (Fig. 3C). Double in situ hybridization with Krox20 reveals that rhombomeres 4 and 6 express ext1a (Fig. 3C, inset). Furthermore, the anterior hindbrain (Fig. 3C, black arrow) and the dorsal diencephalon (Fig. 3C, white asterisk) are positive for ext1a message. In the eye, expression consists of two domains: the distal retina (Fig. 3C′, white arrowhead) and the optic stalk region (Fig. 3C′), as marked by double in situ hybridization with pax2.1 in red (Fig. 3C′, inset shows fluorescent pax2.1 signal). Most posteriorly, Kupffer's vesicle exhibits elevated ext1a expression (Fig. 3A, black arrow). At 24 hpf, strong cerebellar expression is detected (Fig. 4A, black arrow, 4D, black arrow) as well as expression in the optic stalk (Fig. 4A, black arrowhead, 4F, black arrowhead) and the dorsal diencephalon (Fig. 4A, black asterisk). At this stage, the anterior part of the epiphysis shows a stronger signal compared with the posterior part (Fig. 4E, outlined with white dotted line). In general, ext1a expression is confined to the dorsal part of the neural tube, as revealed by a cross-section at the level of the anterior spinal cord (Fig. 4C). Furthermore, expression can be detected in the tail somites. At approximately 48 hpf, expression continues in the dorsal diencephalon and in the cerebellum. In addition, mRNA can be detected in the otic vesicle (Fig. 4B, black arrow).
Expression of ext1b
Ext1b is also provided maternally (data not shown) and transcripts can be detected at sphere stage (Fig. 2D). During gastrulation, expression consists of two domains: the germ ring and the axial mesoderm (Fig. 2E, asterisk). Cross-section at the level of the germ ring confirms mesendodermal expression (Fig. 2F). During somitogenesis stages, expression is strongest in the tail bud and the posterior somites (Fig. 3D). Cross-section reveals transcripts in the myotomal part of the somites, while no expression is detected in the sclerotomal part and in adaxial tissues (Fig. 3E). In addition, transcripts can be detected in the forming eye (Fig. 3D). At the 24 hpf stage, cells of the neural tube up to the posterior limit of the hindbrain (Fig. 4G, black arrow) are positive for ext1b transcripts. In addition, the posterior somites and the tail bud (Fig. 4G, black arrowhead) continue to express ext1b. Another mesodermal expression domain comprises cells in the region of the anterior pronephric duct (Fig. 4I, black arrow). In the forming fin buds (Fig. 4J), ext1b can also be detected. At approximately 48 hpf, the fin buds (Fig. 4H, black arrow) as well as the brain are positive for ext1b transcripts.
Expression of ext1c
Ext1c message can be detected first at the beginning of somitogenesis, where it is expressed in the forming somites. Later, at approximately the 16-somite stage (Fig. 3F), expression is confined to adaxial cells in the posterior part of the embryo (Fig. 3G). More anteriorly, cells adjacent to the neural tube and in ventromedial regions of the somite express ext1c (Fig. 3H, black arrowheads). At the end of somitogenesis, neural expression domains of ext1c appear, whereas the posterior somites continue to express ext1c. Neural domains become more elaborate at the 24 hpf stage (Fig. 4K): transcripts can be detected in the ventral rhombomeres (Fig. 4K, white arrowhead) and the telencephalon (Fig. 4K, black arrow). Furthermore, close up of the telencephalon (Fig. 4M, white asterisk) reveals that also cells of the olfactory bulbs contain ext1c transcripts (Fig. 4M, black arrow). At 48 hpf, transcripts can be detected in the brain (Fig. 4L). Cross-section at the level of the otic vesicle shows strong expression in the white matter (Fig. 4O, black arrow) as well as in the otic vesicles (white arrowhead) and in more dorsally located nascent neurons. At this stage, also the fin buds (data not shown) and the retina express ext1c (Fig. 4N).
Taken together, our results reveal the existence of at least three different members of the EXT1 family in zebrafish that most likely arose during a genome duplication event in the teleost lineage. Their expression throughout embryogenesis comprises partially overlapping and mutually exclusive domains. In addition, the expression patterns show a spatial shift over time: at somitogenesis stages, transcripts are present in the somites, while at later stages of development, the brain and the fin buds express ext1 genes.
These expression domains partially contrast with those found in other species. For instance, the Drosophila EXT1 homolog, ttv is present ubiquitously (The et al.,1999) and in Xenopus EXT1 is expressed without tissue specificity during early embryogenesis and at adult stages (Katada et al.,2002). Of interest, the mouse EXT1 expression pattern is more complex. In early embryonic development (5.5 to 7.5 days post coitum), EXT1 expression is reported to be ubiquitous (Lin et al.,1998). Later, transcripts accumulate in limb buds, regions of ossification, tail, and brain (Lin et al.,1998; Stickens et al.,2000; Inatani and Yamaguchi,2003). As we report for zebrafish ext1a, expression of Ext1 was strongest in the cerebellum, which continues to express Ext1 at postnatal stages (Rubin et al.,2002; Inatani and Yamaguchi,2003). In addition, the mouse forebrain contains Ext1 transcripts (Inatani and Yamaguchi,2003). In zebrafish, we find ext1c present in these structures (Fig. 4K, black arrow, M), while ext1a expression is lost in forebrain regions (Fig. 4A). Thus, different zebrafish ext1 genes are expressed in structures that both express the single mouse Ext1 gene. Another site of mouse Ext1 expression are the limb buds, homologous to fin buds in zebrafish. Here, ext1b and ext1c transcripts can be found, but no ext1a expression, which might have become dispensable during evolution due to the continued expression of the other two ext1 genes. From these findings, we speculate that the diversification of ext1 gene expression might provide another example of the duplication–degeneration–complementation model (Force et al.,1999).
Dependence of ext1a and ext1c on Hedgehog Signaling
Previous studies in Drosophila have shown that EXT1 genes are necessary for the spreading of molecules belonging to the hh family (Bellaiche et al.,1998). Furthermore, a study in mice suggests that EXT1 proteins might negatively regulate shh signaling by synthesizing HSPGs, which sequester the ligand (Koziel et al.,2004). This negative regulation might suggest the existence of a feedback loop. For instance, overexpression of shh induces expression of the SHH receptor patched (Concordet et al.,1996), which also sequesters the ligand (Chen and Struhl,1996), whereas loss of shh signaling leads to an absence of patched expression (Chen et al.,2001; Varga et al.,2001). We therefore asked whether ext1 genes respond to shh in a similar way. To this end, we examined the effects of shh overexpression on zebrafish ext1 genes using synthetically transcribed shh mRNA. We also addressed the effects of loss of hedgehog signaling by analyzing the expression of ext1 genes in known mutants of the hedgehog pathway in fish. These mutants included smu (slow muscle omitted; Chen et al.,2001; Varga et al.,2001), which encodes for zebrafish smoothened, the canonical receptor for all hedgehog signals, and syu (sonic you; Schauerte et al.,1998), a deletion mutant for the zebrafish shh gene. The effect of altered hedgehog signaling was assessed at the 14-somite stage. In a dorsal view of this stage, ext1a is more strongly expressed in tissue closer to the midline (Fig. 5B). Increased hedgehog signaling led to ectopic expression of ext1a throughout most of the somite (Fig. 5B,C, compare white brackets). However, different expression intensities of ext1a could still be detected. More medially located areas of the somites showed a stronger staining, which might suggest that additional factors are required for the induction of strong ext1a expression. In the case of ext1c, the effect of shh overexpression was more pronounced. Upon overexpression of shh, a large proportion of the somitic cells expressed ext1c at uniformly high levels (Fig. 5E,F, compare white brackets). Of interest, expression of ext1b was found to be essentially normal in a gain of shh function situation (data not shown).
In contrast to this finding, loss of shh signaling (shown for smu mutant embryos) led to a nearly complete loss of ext1a expression in the somites (Fig. 5A). Similarly, ext1c expression could only be detected in the more anterior somites with a progressive loss of expression toward posterior regions (Fig. 5D). Again, expression of ext1b was not altered in smu mutant embryos (data not shown). Similar results were obtained when analyzing syu mutant fish. However, we noted a less severe reduction in gene expression of both ext1a and ext1c, probably reflecting a possible redundancy with other members of the Hh family (data not shown). Therefore, we conclude that expression of ext1a and ext1c in somites depends in part on functional hh signaling and responds to shh overexpression by an increase in transcription. We speculate that a putative regulatory link between zebrafish ext1a and ext1c genes and hh signaling exists during vertebrate somite patterning. Alternatively, the observed effects may be due to an indirect effect of slow muscle respecification, which has been shown to occur upon shh overexpression (Blagden et al.,1997). The finding that ext1b does not respond to elevated or reduced levels of shh signaling underscores the divergence of different members of the ext1 family in zebrafish. Interestingly, we noted ectopic expression of ext1a in the notochord of smu mutant zebrafish at the 14-somite stage (Fig. 5A). At this stage, shh is produced in notochord cells (Krauss et al.,1993). In wild-type, axial mesoderm also expresses ext1a at the beginning of somitogenesis (Fig. 2G), but this expression is lost during later development (Fig. 3B). The prolonged expression of ext1a in the notochord of smu embryos might reflect a differential response of the ext1a promoter in notochord and somitic cells.