Expression of rib-1, a Caenorhabditis elegans Homolog of the Human Tumor Suppressor EXT Genes, Is Indispensable for Heparan Sulfate Synthesis and Embryonic Morphogenesis*

The proteins encoded by all of the five cloned human EXT family genes (EXT1, EXT2, EXTL1, EXTL2, and EXTL3), members of the hereditary multiple exostoses gene family of tumor suppressors, are glycosyltransferases required for the biosynthesis of heparan sulfate. In the Caenorhabditis elegans genome, only two genes, rib-1 and rib-2, homologous to the mammalian EXT genes have been identified. Although rib-2 encodes an N-acetylglucosaminyltransferase involved in initiating the biosynthesis and elongation of heparan sulfate, the involvement of the protein encoded by rib-1 in the biosynthesis of heparan sulfate remains unclear. Here we report that RIB-1 is indispensable for the biosynthesis and for embryonic morphogenesis. Despite little individual glycosyltransferase activity by RIB-1, the polymerization of heparan sulfate chains was demonstrated when RIB-1 was coexpressed with RIB-2 in vitro. In addition, RIB-1 and RIB-2 were demonstrated to interact by pulldown assays. To investigate the functions of RIB-1 in vivo, we depleted the expression of rib-1 by deletion mutagenesis. The null mutant worms showed reduced synthesis of heparan sulfate and embryonic lethality. Notably, the null mutant embryos showed abnormality at the gastrulation cleft formation stage or later and arrested mainly at the 1-fold stage. Nearly 100% of the embryos died before L1 stage, although the differentiation of some of the neurons and muscle cells proceeded normally. Similar phenotypes have been observed in rib-2 null mutant embryos. Thus, RIB-1 in addition to RIB-2 is indispensable for the biosynthesis of heparan sulfate in C. elegans, and the two cooperate to synthesize heparan sulfate in vivo. These findings also show that heparan sulfate is essential for post-gastrulation morphogenic movement of embryonic cells and is indispensable for ensuring the normal spatial organization of differentiated tissues and organs.

the genes encoding these enzymes and genetic analyses using knock-out mice, Zebrafish, fruit flies, and nematodes have led to unanticipated findings of interesting biological phenotypes (for a review see Ref. 5).
In humans, polymerization of the repeating disaccharide region in HS occurs through the actions of an enzyme complex consisting of EXT1 (6) and EXT2 (7) of the EXT (exostosin) gene family (8 -11). These genes were first identified as causative of a genetic bone disorder, hereditary multiple exostoses (HME), and subsequently demonstrated to function as tumor suppressor genes. HME is an autosomal dominant disorder characterized by the formation of a cartilage-capped tumor, caused by mutations in either EXT1 or EXT2 (12). The family of EXT genes has been extended to include three EXT-Like genes, EXTL1, EXTL2, and EXTL3, all of which have GlcNAc transferase activities, likely involved in HS synthesis (4,13,14).
HS is present in genetically tractable model animals as well (15)(16)(17), and EXT genes are well conserved in mammals, Zebrafish (Danio rerio) (ext1, ext2/dackel, and extl3/boxer), Drosophila melanogaster (ttv, sotv, and botv) and Caenorhabditis elegans (rib-1 and rib-2) (18 -28). In Zebrafish dackel and boxer mutants, which are defective in ext2 and extl3, respectively, some dorsal retinal ganglion cell axons inappropriately project into the optic tract (19). Even though Dackel and Boxer have not been demonstrated to have glycosyltransferase activities, the amount of HS was drastically reduced in the dackel and boxer mutants, suggesting that both genes are required for its production (19). In addition, Drosophila homologs of EXT genes, tout-velu (ttv), sister of tout-velu (sotv), and brother of tout-velu (botv), which correspond to vertebrate EXT1, EXT2, and EXTL3, respectively, are involved in Hh, Wg, and Dpp signaling (20 -23). Biochemical and immunohistochemical studies in Drosophila have revealed that HS levels are dramatically reduced in the absence of ttv, sotv, or botv (21)(22)(23). We recently demonstrated in vitro that the polymerization is performed by a complex of TTV and SOTV (TTV-SOTV) as an enzyme source (25). However, TTV-SOTV exhibited no GlcNAc transferase-I (GlcNAcT-I) activity required for the initiation of HS, indicating that BOTV, which corresponds to human EXTL3 and possesses GlcNAcT-I activity (24), is indispensable for the biosynthesis of HS chains in Drosophila. Thus, all three EXT members in Drosophila, TTV, SOTV, and BOTV, are required for the biosynthesis of a full-length HS.
In C. elegans, rib-1 and rib-2 are most homologous to EXT1 and EXTL3, respectively, among the human EXT genes (26). rib-2 mutants exhibit developmental delay and egg-laying defects, which are most likely caused by a reduction in HS (28). Although rib-2 encodes a GlcNAc transferase involved in initiating the biosynthesis of HS and in the elongation of HS chains (27), the involvement of the protein encoded by rib-1 in HS biosynthesis remains unclear. In this study, we investigated the polymerization of HS using recombinant soluble forms of RIB-1 and RIB-2 to elucidate the mechanism behind the biosynthesis of HS in C. elegans, and we demonstrated that the polymerization was performed by a complex of RIB-1 and RIB-2 as an enzyme source. In addition, to investigate functions of RIB-1 in vivo, we depleted the expression of rib-1 by deletion mutagenesis. Here we demonstrate that in addition to rib-2, rib-1 is indispensable for the biosynthesis of HS and embryonic morphogenesis in C. elegans.
Single Worm PCR-Single animals were picked up with a platinum wire and placed individually in a 10-l drop of lysis buffer (500 g/ml proteinase K in 25 mM Tris-HCl, pH 8.5, 50 mM KCl, 0.5% Tween 20, 1 mM EDTA) in a 200-l PCR tube. The drops were spun down to the bottom of the tube by brief centrifugation, frozen (Ϫ80°C, 10 min), and heated (50°C, 30 min followed by 95°C, 20 min). After cooling to room temperature, 1-2 l of the drops was used as a template for single worm PCR for the tm516 and tm710 deletion alleles with the primers described above.
Materials-UDP-[U- 14  Construction of Soluble Forms of RIB-1 and RIB-2-A cDNA fragment of a truncated form of RIB-1, lacking the first 17 amino-terminal amino acids, including the putative cytoplasmic and transmembrane domains, was amplified by reverse transcription-PCR with adult C. elegans total RNA as a template using a 5Ј-primer (5Ј-GAAGATCTGTTCGGTTACAAAAT-CCAACTATTG-3Ј) containing a BglII site and a 3Ј-primer (5Ј-GAAGATCTAGATCGGGATTGAACCTTATGG-3Ј) containing a BglII site located 22 bp downstream of the stop codon. In the case of RIB-2, the cDNA fragment encoding a truncated form of RIB-2, lacking the first amino-terminal 38 amino acids of RIB-2 (AB077851), was amplified by reverse transcription-PCR with adult C. elegans total RNA as a template using a 5Ј-primer (5Ј-CGGGATCCTCCTTCTCCGAGCCTTCCTG-3Ј) containing a BamHI site and a 3Ј-primer (5Ј-CGGGATC-CGTAGCACCATTCGA-3Ј) containing a BamHI site located 68 bp downstream of the stop codon. PCR was carried out with KOD polymerase (Toyobo, Osaka, Japan) for 32 cycles of 94°C for 30 s, 60°C for 30 s, and 68°C for 180 s in 5% (v/v) dimethyl sulfoxide. The PCR fragment was subcloned into the BamHI site of pGIR201protA (31), resulting in the fusion of the insulin signal sequence and the protein A sequence present in the vector as described previously (25). The nucleotide sequence of the amplified cDNA was determined in a 377 DNA sequencer (Applied Biosystems).
Expression of Soluble Forms of RIB-1 and RIB-2 and Enzyme Assays-The expression plasmid (6.0 g) was transfected into COS-1 cells on 100-mm plates using FuGENE TM 6 (Roche Applied Science) according to the manufacturer's instructions. For cotransfection experiments, the RIB-1 and RIB-2 expression plasmids (3.0 g each) (27) were cotransfected into COS-1 cells on 100-mm plates using FuGENE TM 6 as above. After a 2-day culture at 30°C, 1 ml of the culture medium was collected and incubated with 10 l of IgG-Sepharose (GE Healthcare) for 1 h at 4°C. The beads were recovered by centrifugation, washed with and then resuspended in each of the assay buffers described below, and tested for GlcNAcT-II transferase activity using 10 g of N-acetylheparosan oligosaccharides, GlcUA␤1-[4GlcNAc␣1-4GlcUA␤-] n , and for GlcAT-II activity using 10 g of GlcNAc␣1-[4GlcUA␤1-4GlcNAc␣1-] n as an acceptor, respectively, as described previously (25). The assay mixture for GlcNAcT contained 10 l of the resuspended beads, an acceptor substrate, 0. Western Blot Analysis-After 3 days of culture at 28°C, the culture medium was collected and incubated with 10 l of IgG-Sepharose (GE Healthcare) for 1 h at 4°C. The beads recovered by centrifugation were washed with phosphate-buffered saline and then resolved on 7.5% SDS-polyacrylamide gels, transferred to polyvinylidene difluoride membranes, and incubated for 1 h with mouse IgG antibody. The mouse IgG antibody was diluted 1:1,000 with 25 mM Tris-buffered saline. The bound antibody was detected with anti-mouse IgG conjugated to horseradish peroxidase. Proteins bound to the antibody were visualized with an enhanced chemiluminescence (ECL) advance kit (GE Healthcare).
Characterization of the Reaction Products-Products of polymerization reactions on N-acetylheparosan oligosaccharides, GlcUA␤1-[4GlcNAc␣1-4GlcUA␤-] n , or GlcUA␤1-3Gal␤1-O-C 2 H 4 NHCbz were isolated by gel filtration on a column of Superdex TM 75 or Superdex TM Peptide, respectively, with 0.2 M NH 4 HCO 3 as the eluent. The radioactive peak containing the cotransfection reaction product was pooled and evaporated dry. The [ 14 C]GlcUA-labeled oligosaccharide chains were exhaustively digested with 3 mIU of heparitinase I in a total volume of 20 l of 50 mM sodium acetate buffer, pH 7.0, containing 2 mM calcium acetate at 37°C overnight. The enzyme digest was analyzed using the same Superdex TM 75 or Superdex TM Peptide column as described above.
Analysis of Glycosaminoglycans-Freshly cultured nematodes were sonicated with a GE-70 ultrasonic processor (Branson Ultrasonics) and freeze-dried. The dried samples (295.2 mg of wild type or 386.3 mg of mutant nematode) were extracted with acetone and then treated with 6 ml of 1.0 M NaBH 4 , 0.05 M NaOH at 4°C for 20 h. The reaction mixture was neutralized with acetic acid to stop the reaction. The samples were adjusted to 5% trichloroacetic acid and centrifuged. The soluble fraction was extracted with ether. As shown previously (15,32,33), the amount of HS in C. elegans was so small that 100 g of shark cartilage chondroitin 6-O-sulfate (Seikagaku Corp.), which contains a negligible proportion of nonsulfated disaccharides, was added as a carrier. The aqueous phase was adjusted to 80% ethanol. The resultant precipitate was dissolved in 50 mM pyridine acetate and subjected to gel filtration on a PD-10 column using 50 mM pyridine acetate as an eluent. The flow-through fraction was collected and evaporated dry. The dried samples were dissolved in water and applied to a column (7 ml) of cation-exchange resin AG 50W-X2 (H ϩ form; Bio-Rad) pre-equilibrated with water. The unbound fraction containing the liberated O-linked saccharides was neutralized with 1 M NH 4 HCO 3 . The purified glycosaminoglycan fraction was digested with chondroitinase ABC or a mixture of heparitinases I and II, and then the digests were derivatized with 2-aminobenzamide and analyzed by high performance liquid chromatography as described previously (34).
Pulldown Assays-The cDNA fragment of a truncated form of RIB-2, lacking the first 38 amino-terminal amino acids of RIB-2, was amplified using a 5Ј-primer (5Ј-CGGGATCCCTC-CTTCTCCGAGCCTTCC-3Ј) containing an in-frame BamHI site and a 3Ј-primer (5Ј-CGGGATCCGTAGCACCATTCGA-3Ј) containing a BamHI site located 68 bp downstream of the stop codon. The DNA fragment was inserted into the expression vector pcDNA3Ins-His, resulting in the fusion of the protein with the insulin signal sequence and His 6 sequence present in the vector. The rib-2 construct and the protein A-tagged RIB-1 expression vector were introduced into COS-1 cells on 100-mm plates using FuGENE TM 6 (Roche Applied Science) according to the manufacturer's instructions. Two days after the cotransfection, 1 ml of the culture medium was collected and incubated with 10 l of Ni-NTA-agarose (Qiagen) overnight at 4°C. The beads recovered by centrifugation were washed with Tris-buffered saline buffer containing Tween 20 three times and subjected to SDS-PAGE (7% gel), and proteins were transferred to a polyvinylidene difluoride membrane. The membrane, after blocking in phosphate-buffered saline containing 2% skim milk and 0.1% Tween 20, was incubated with IgG antibody and then treated with anti-mouse IgG conjugated with horseradish peroxidase (GE Healthcare). Proteins bound to the antibody were visualized with an ECL advance kit (GE Healthcare).
Transgenic Construct-The tph-1 (ZK1290.2) reporter gene plasmid was constructed using the vector pFX_DsRedXT (33,34). The translational fusion construct contained a 3.7-kb genomic fragment, including the 3.1-kb potential promoter region. The fragment was amplified with C. elegans genomic DNA as a template using a 5Ј-primer (5Ј-GAGTGCAAGAC-TATTAGGGAGT-3Ј) and a 3Ј-primer (5Ј-AGTACGAGTTG-GTGTGAAGAG-3Ј) located at the start of the 3rd exon. PCR was carried out with Platinum TaqDNA polymerase high fidelity (Invitrogen) for 28 cycles of 94°C for 30 s, 58°C for 30 s, and 68°C for 240 s followed by adenine addition with Taq polymerase at 72°C for 15 min. The PCR fragment was inserted into pFX_ DsRedXT using the TA cloning strategy to fuse the coding sequence region of DsRed.
Phenotypic Analysis-We determined the embryonic phenotypes and extent of lethality of the rib-1/rib-2 mutant worms by monitoring embryonic development with four-dimensional microscopy (a DMRXA full automatic microscope with differential interference contrast and fluorescence optics; Leica) (32) followed by single worm PCR to confirm the genotype. A total of 92 rib-1 Ϫ/Ϫ worms was monitored from the start of gastrulation until they had reached a terminal phenotype and scored. Phenotypes were also monitored and scored by using a high resolution dissecting microscope (Olympus SZX12) as described previously (36).
Transgenic Rescue Experiment-To rescue the rib-1 deletion mutant tm516, a transgenic rescue construct was created using the C. elegans expression vector (pFX_LVT-R03G5.1) constructed from PFX_venusT cloning vector (35), in which eft-4 (R03G5.1; elongation factor-1 ␣) promoter (3,616 bp from X:7820004 to X:7823619) was inserted into the XcmI site in the multicloning site, which is located upstream of the venus open reading frame. The rib-1 cDNA was reverse-transcribed with SuperScript II reverse transcriptase (Invitrogen) from the total RNA prepared from mixed stage worms (36) and amplified with PCR using a 5Ј-primer (5Ј-GTGGCGGCCGCATGCAAAAT-GTAATGAAAT-3Ј) containing the start codon and a 3Ј-primer (5Ј-CATGCGGCCGCCTTGATGATCTTCCAG-CTC-3Ј) located before the stop codon. PCR was carried out with Platinum TaqDNA polymerase high fidelity (Invitrogen) for 30 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 90 s. The full-length cDNA was inserted into a NotI site and placed under the control of the eft-4 promoter, and the sequence was verified by using Prism 3130 Genetic analyzer (Applied Biosystems, Inc.). The rescue of the rib-1 mutant strain (tm516) with the full-length cDNA was performed by microinjecting the rescue construct into distal gonads as described (36). The concentration of injected DNA was 39 g/ml, and no coinjection marker was used in the experiment.

RESULTS
Coexpression and Characterization of RIB-1-A data base search revealed only two homologs, rib-1 and rib-2, of the mammalian EXT genes in the C. elegans genome (26). Although rib-2 encodes a GlcNAcT involved in initiating the biosynthesis and in the elongation of HS (27), the involvement of the protein encoded by rib-1 in HS biosynthesis remains unclear. The RIB-1 protein, composed of 382 amino acids with two potential N-glycosylation sites (Fig. 1A), is about half the size of the RIB-2 protein and the four mammalian EXT family members, EXT1, EXT2, EXTL1, and EXTL3. Interestingly, the RIB-1 protein shows significant homology to the amino termini of these family members, especially to EXT1 (45% amino acid identity) (Fig.  1A). In fact, a phylogenetic tree of the EXT family members of humans and C. elegans generated based on amino acid sequences showed that RIB-1 and EXT1 are related more closely to one another than to the other members of the family (Fig. 1B). Kyte-Doolittle hydropathy plot revealed one prominent hydrophobic segment of 18 amino acid residues, which begins at the amino terminus and lacks sufficient length for a membrane-spanning domain. In addition, the hydrophobic segment differs from a membrane anchor in that it contains two cationic residues and is not flanked by cationic residues. Moreover, an analysis using the SOSUIsignal system (37) for the prediction of signal peptides and membrane proteins showed that the protein has a signal peptide. These results suggest that RIB-1 appears to be a soluble protein. In this regard, HS D-glucosaminyl 3-O-sulfotransferase is also predicted to be a soluble protein because the protein has only one prominent hydrophobic segment of 18 amino acid residues, which begins at the amino terminus and differs from a membrane anchor in that it contains two glutamine residues and is not flanked by cationic residues (38).
To facilitate the functional analysis of RIB-1, a soluble, protein A-tagged form of the protein was generated by replacing the first 17 amino acids with a cleavable insulin signal sequence and a protein A IgG-binding domain as described under "Experimental Procedures." Then the soluble protein was expressed in COS-1 cells at 30°C as a recombinant protein fused with the protein A IgG-binding domain. The fusion protein secreted in the medium was adsorbed onto IgG-Sepharose beads for purification to eliminate endogenous glycosyltransferases, and then the protein-bound beads were used as an enzyme source. Although the bound fusion protein was assayed for glycosyltransferase activity at 26°C using N-acetylheparosan oligosaccharides, [GlcUA-GlcNAc] n , or GlcNAc-[GlcUA-GlcNAc] n , as a sugar acceptor and either UDP-GlcNAc or UDP-GlcUA as a sugar donor substrate, neither GlcNAcT-II nor GlcAT-II activity was detected (Table 1). However, coexpression of the soluble RIB-1 with the soluble RIB-2 clearly showed GlcAT-II activity, which had not been detected when only RIB-2 was expressed. Notably, these effects of coexpression were not because of differences in the levels of these proteins, as assessed by Western blot analysis (Fig. 2). These results are analogous to the findings previously made for human EXT1 and EXT2 (9). Hence, we examined whether the coexpression of RIB-1 and RIB-2 could result in the expression of polymer- izing activity. Incubation of the coexpressed proteins with N-acetylheparosan oligosaccharides, [GlcUA-GlcNAc] n , with the nonreducing terminal GlcUA as an acceptor in the presence of UDP-GlcNAc and UDP-[ 14 C]GlcUA yielded radiolabeled polymer N-acetylheparosan chains. The reaction products were analyzed by gel chromatography using a column of Superdex TM 75 as shown in Fig. 3A. The synthesized saccharide chains were around 20-mer in length. The radioactive peaks containing cotransfectant reaction products were digested with heparitinase I and subjected to gel chromatography as described under "Experimental Procedures." The radiolabeled products were completely digested, quantitatively yielding a 14 C-labeled peak at the position of [ 14 C]GlcUA␤1-3GlcNAc, indicating that the polymerization occurred on N-acetylheparosan when RIB-1 and RIB-2 were coexpressed (Fig. 3A).
To further examine the polymerization activity, the size-defined oligosaccharide, GlcUA␤1-3Gal␤1-O-C 2 H 4 NHCbz, was used. The linkage region analog, GlcUA␤1-3Gal␤1-O-C 2 H 4 NHCbz, was incubated with the coexpressed RIB-1 and RIB-2, and the reaction products were analyzed by gel chromatography using a column of Superdex TM Peptide as described under "Experimental Procedures." As shown in Fig. 3B, HS polymerization was demonstrated, and the synthesized saccharide chains were completely digested by heparitinase I. These results indicated that two C. elegans EXT proteins, RIB-1 and RIB-2, together could achieve polymerization on a linkage region.
Interactions between RIB-1 and RIB-2-As shown above, coexpression of RIB-1 and RIB-2 resulted in a marked augmentation of not only glycosyltransferase activities but also polymerase activity. In view of these results, physiological interactions between these molecules were expected. Thus, the interactions of these molecules were evaluated by pulldown assays. For this analysis, a soluble form of RIB-1 fused with protein A at its amino terminus (RIB-1-ProA) and a soluble form of RIB-2 tagged with the His 6 epitope at its amino terminus (RIB-2-His) were generated as described under "Experi-

TABLE 1 GlcNAcT-II and GlcAT-II activities of the fusion proteins secreted into the culture medium by transfected COS-1 cells
Values represent the averages of three independent experiments.

Protein GlcNAcT-II activity a GlcAT-II activity b pmol/ml medium/h pmol/ml medium/h
RIB-1 ND c ND RIB-2 114.2 ND RIB-1/RIB-2 d 172.2 6.5 a N-Acetylheparosan oligosaccharides, ͓GlcUA-GlcNAc͔ n , with the nonreducing terminal GlcUA were used as an acceptor substrate. b N-Acetylheparosan oligosaccharides, GlcNAc-͓GlcUA-GlcNAc͔ n , with the nonreducing terminal GlcNAc were used as an acceptor substrate. c ND, not detected (Ͻ0.01 pmol/ml medium/h). d RIB-1/RIB-2 represents coexpressed RIB-1 and RIB-2. mental Procedures." To evaluate the interactions between RIB-1 and RIB-2, coexpression of RIB-1-ProA with RIB-2-His was carried out. In addition, to ensure specificity, we tried these assays with CPF-1 (PAR2.4; C. elegans chondroitin polymerizing factor), which is expected to interact with SQV-5 (cChSy; C. elegans chondroitin synthase) (32, 33) but not with RIB-1 or RIB-2. Ni-NTA-agarose was added to the culture medium to pull down His-tagged proteins, and the proteins were then subjected to SDS-PAGE followed by Western blotting using an IgG antibody as a primary antibody to protein A-tagged proteins and detected with an ECL advance kit. No band was detected for the transfectant of RIB-1-ProA alone (Fig. 4, lane 1) or for the cotransfectant of RIB-1-ProA and PFC-1 (PAR2.4)-His (Fig.  4, lane 3). In contrast, a single protein band with a molecular mass of ϳ80 kDa, corresponding to RIB-1-ProA, was detected for the cotransfectant of RIB-1-ProA and RIB-2-His (Fig. 4, lane  2). As expected, the coexpression of SQV-5 (cChSy)-ProA and CPF-1-His gave a single band with a molecular mass of ϳ130 kDa, corresponding to SQV-5-ProA (Fig. 4, lane 4). These results indicated that RIB-1 and RIB-2 physiologically interact with each other.
Knocking Out of rib-1 by Deletion Mutagenesis-To understand the functions of the rib-1 gene in vivo, we isolated a deletion mutant of rib-1, tm516, by screening a trimethylpsoralen/ ultraviolet-induced deletion mutant library of N2 worms, and we examined the development of the mutant worm by fourdimensional microscopy. The deletion mutant has a 486-bp deletion and a 32-bp insertion (nucleotides 37528/37529 to 38014/38015 in the cosmid F12F6) corresponding to a deletion of most of the 6th intron and 5Ј region of the 7th exon of the rib-1 gene (Fig. 1C). The homozygous deletion was lethal or caused sterility, so the mutant worms were maintained as heterozygotes by using a dpy-20(e1415) balancer. Phenotypes were scored by observing and isolating a worm and conducting PCR to determine genotypes of all the worms observed. All rib-1 ϩ/Ϫ animals were apparently wild type. The rib-1 Ϫ/Ϫ embryos hatched from rib-1 ϩ/Ϫ worms can develop into adult worms, and 98.5% of Ϫ/Ϫ worms developed into the adult stage (n ϭ 404). The successful development of adult worms from rib-1 Ϫ/Ϫ eggs in P0 worms can be attributed to maternal rib-1 mRNA and/or HS proteoglycans from heterozygous worms. The F1 adult worms showed defects in egg-laying (Egl phenotype) (53.5%, n ϭ 143) and accumulated dead embryos in the uterus and were unable to lay eggs from the vulva (Fig. 5, A and  B). Among adult worms defective in egg-laying, 65.2% (n ϭ 43) showed a protruding vulva (Pvl) phenotype. In this study, worms having a 10 m or longer protruding vulva were scored as Pvl worms, and 38.4% of them showed very severe Pvl phenotypes (see Fig. 5C).
Almost all (99.3% (n ϭ 1622)) of the rib-1 Ϫ/Ϫ fertilized eggs laid by rib-1 Ϫ/Ϫ adult worms showed embryonic lethality and died before the L1 stage. The remaining 0.7% worms hatched and showed abnormal head or tail structures and died just after hatching at L1 (Figs. 5D and 6C). In some cases, axon migration patterns of adult worms hatched from heterozygous worms were severely distorted (Fig. 5F), especially in hermaphroditespecific neurons (HSNs) (26.5%, n ϭ 34). Two axon neuronal processes from HSNs showed midline crossover defects, whereas in normal N2 worms, they run parallel with each other (Fig. 5G). This phenotype is similar to the HSN abnormality reported on the knock-out of the HS glucuronyl C5-epimerase gene (hse-5) in C. elegans (39). For the Egl phenotype, further detailed study is necessary to know whether the phenotype is related to defects in egg-laying neurons, muscle cell functions, or abnormal vulva structure as found in the protruding vulva.
To determine at what point abnormal development is detected, we monitored 92 rib-1 Ϫ/Ϫ embryos by four-dimensional microscopy. Ingression of E-daughter cells seemed to be normal and blastocele was formed. The first abnormality was detected at the "closure of ventral cleft" stage, and worms with an abnormal ventral cleft (80%, n ϭ 10) were observed along with worms with a normal ventral cleft (20%). In severe cases (class I, Fig. 7A), oozing of cells and prominent bulges were observed in various regions of embryos. These phenotypes strongly suggest that normal epidermal enclosure was severely affected (40). All class I embryos were arrested at 1-1.25-fold stages and died (68.4% n ϭ 92). In some cases (class II, Fig. 7B), embryos continued to elongate to the 1.5-fold stage at which point they failed to elongate further and died (21.1%). Oozing of cells from ventral midline region was clearly observed in these embryos. In other cases (class III), embryos developed into 1.5-2.5-fold stages and died (10.5%).Oozing of a few cells was observed. In class I to class III rib-1 Ϫ/Ϫ F1 embryos, active apoptosis was observed after intestinal differentiation (Fig. 6, A and  B). Judging from active contractile movement (twitching) in these rib-1 Ϫ/Ϫ embryos (92.4%, n ϭ 92), at least some of the contractile muscle tissues seemed to differentiate normally. In some cases, the differentiation of a pharyngeal bulb-like structure in the head region was observed, and muscles frequently contracted (see Ref. 41 and data not shown). To examine the relative timing of neuronal differentiation and rib-1 Ϫ/Ϫ embryonic death, we crossed rib-1 ϩ/Ϫ worms with the pan-neuronal GFP reporter strain NW1229 (made and deposited to CGC by J. Culotti), and we examined the differentiation of neuronal cells in rib-1 Ϫ/Ϫ integrated Ex[F25B3.3::gfp; dpy-20(ϩ)] embryos. The rib-1 Ϫ/Ϫ embryos confirmed to be null mutants by single worm PCR with rib-1 primer pairs showed active neuronal GFP fluorescence (Figs. 5B and 6D). Similar results were obtained in transgenic worms expressing DsRed under the control of the tph-1 promoter. In the transgenic N2 worms, differentiated NSM serotonergic neurons were usually stained brightly, and in some worms the differentiated hermaphrodite-specific HSNs were also stained brightly (see Ref. 42 and data not shown). In the transgenic rib-1 Ϫ/Ϫ Ex[tph-1::DsRed] worms, bright neuronal fluorescence of possible NSM pair neurons (NSML/NSMR) and neuronal processes were observed in developmentally arrested embryos, suggesting the normal differentiation of these neuronal cells in rib-1 Ϫ/Ϫ worms (Fig. 6, E and F). In these null mutant embryos, morphogenic movement stopped in the early stages of elongation and development appeared to be arrested (Fig. 7, A-C). Although some of the neurons and muscle cells seemed to differentiate in the null mutant embryos as described above, further detailed study is necessary to know whether any other cells (including neuronal and muscle cells) show various degrees of defects in the embryos or not.
Similar phenotypes were observed in the rib-2 Ϫ/Ϫ mutant strain tm710, which was isolated in our laboratory and has a 1306-bp deletion (nucleotide 2128/2129 to 3434/ 3435) in the cosmid K01G5.6 ( Fig. 7, D and E). The tm710 mutant lacks first six exons and five introns and most of the sixth introns. A similar abnormality after gastrulation has been reported for another rib-2 null mutant (28). Examination of an in situ hybridization data base indicates that rib-2 mRNA is highly expressed in elongation stage embryos (see NEXTDB on line). These results altogether strongly suggest that HS is indispensable for morphogenesis, including ventral/ epidermal enclosure and elongation of embryos. In developmentally arrested embryos of the rib-1 null mutants, the differentiation of some of the muscles and neurons seemed to proceed normally, but in all abnormal embryos of the null mutants, the spatial organization of these tissues was severely distorted (Fig. 6C). Similar phenotypes (class III) have been observed in pps-1 (3-phosphoadenosine 5Ј-phosphosulfate) mutant worms (RNAi) deficient in the sulfation of HS (36) as FIGURE 5. Typical phenotypes of rib-1 null ؊/؊ mutant worms hatched from rib-1 ؉/؊ adult worms. A and B, egg-laying defective (Egl) phenotype in rib-1 Ϫ/Ϫ adults. A large number of fertilized eggs were packed into the uterus of the rib-1 Ϫ/Ϫ hermaphrodite (A, DIC image). In each egg, the neuronal differentiation marker (green fluorescence; for details see text) was detectable (B). C, protruding vulva phenotypes (Pvl) in rib-1 Ϫ/Ϫ adult worm. D, dead fertilized eggs dissected out of the uterus of an Egl, Pvl phenotype worm (C). E, cDNA rescue of tm710 (rib-1 null) worm. The rescued worm developed normally into 3-fold stage (inset) and into the adult worm which laid normal eggs. F and G, crossover defects of HSN axons in rib-1 Ϫ/Ϫ adult worm (F) and in wild type N2 worm (G). Fluorescence micrograph (left) and superimposed DIC image (right) are shown. The position of two motor neuron crossover defects is indicated (arrow). Scale bar, 30 m. well as pat (paralyzed arrest at 2-fold) mutant worms, including pat-2 (␣-integrin) and pat-3 (␤-integrin), which are essential for muscle-body wall integration (43), suggesting possible interactions with HS proteoglycans.
Rescue of Null Mutant Phenotypes with Wild Type Gene-To examine whether the above-mentioned phenotypes were because of the lack of rib-1 gene expression, we tried to rescue tm516 worms by introducing wild type rib-1 cDNA. Germ line rescue experiment was performed with full-length rib-1 cDNA plus venus construct driven by elongation factor-1 promoter. We obtained 20 lines. In all the 20 lines examined, complete rescue of phenotypes was confirmed. Egl and Pvl phenotypes were rescued, and lethal phenotypes were also rescued. Rescued tm516 worms laid eggs, which developed normally into adult worms (Fig. 5E).
Reduction of HS Synthesis in rib-1 Knock-out Worms-We next examined the amount of HS and chondroitin in mass cultures of tm516 worms. Because it was impractical to isolate large quantities of rib-1 Ϫ/Ϫ worms because of their lethal con-dition, plate culture samples of tm516 worms possibly containing rib-1 ϩ/ϩ , rib-1 ϩ/Ϫ , and rib-1 Ϫ/Ϫ worms were used for biochemical analysis. The expected ratio of worms containing these genotypes is 1:2:1. Because rib-1 Ϫ/Ϫ worms are expected to die in the later stages of embryogenesis, the rib-1 gene dose in the mass culture is estimated to be 66% that in wild type N2 worms. As shown in Table 2, direct measurements revealed that the amount of HS in the mass culture was reduced to 63% that in the wild type mass culture, whereas the amount of chondroitin was increased to 143%. Although the change in chondroitin was unexpected, comparable increases have been  A and B). C and D, the spatial organization of differentiated tissues and organs was severely distorted in rib-1 Ϫ/Ϫ embryos (C, DIC image). Neuronal differentiation in the null mutant worm was confirmed by neuronal specific fluorescence (D, arrow) in the transgenic worm (see text for details). By using transgenic rib-1 Ϫ/Ϫ worms expressing the DsRed gene under the control of a neuron-specific tph-1 promoter, differentiation of NSM neuron pairs (arrows) was detected in dying rib-1 Ϫ/Ϫ embryos (E and F). Scale bar, 20 m. FIGURE 7. Arrest of embryonic development after gastrulation stage in rib-1 ؊/؊ and rib-2 ؊/؊ null mutants. All subsequent images were collected by four-dimensional microscopy. A, in rib-1 class I animals (ϳ68% of animals, see text), cells oozed from head, tail, and ventral midline region (arrows) and were arrested at the 1-fold stage with muscular twitching. Prominent bulges in the epidermis were visible (arrowheads). Variable defects in cell movements following gastrulation were observed, and some animals resulted in rupture of the embryo possibly because of enclosure failure (data not shown). Images of the same embryo were acquired at intervals of 56 and 152 min, respectively (left to right). B, rib-1 class II animals (ϳ21% of animals) initially appeared normal and started elongation and then internal cells oozed along the ventral midline and were arrested at the 1.5-fold stage. Images of the same embryo were acquired at intervals of 28 and 54 min, respectively (left to right). C, in wild type embryos, ventral enclosure, dorsal intercalation, and epidermal enclosure occurred normally after gastrulation and elongated to the 2-fold stage (430 -450-min post-fertilization). No oozing of cells was observed. Images of the same embryo were acquired at intervals of 43 and 49 min, respectively (left to right). D, tm710 rib-2 Ϫ/Ϫ embryo from a rib-2 Ϫ/Ϫ adult worm. Like rib-1 class I animals, cells oozed, and the embryo was arrested at the 1-fold stage. The middle and the right images were acquired at a 38-min interval at one focal plane, and the left image at the different focal plane was acquired 46-min earlier than the middle image. Arrows show oozed cells. E, terminal phenotype of tm710 rib-2 Ϫ/Ϫ embryo. Enclosure failure resulted in rupture of the embryo. Arrows show that internal cells seemed to be ejected through the opening in the epidermis. Scale bar, 20 m. observed in rib-2 mutants (28), and an analogous increase in HS was observed in sqv-5 (cChSy)-or pfc-1-RNAi-treated worms deficient in chondroitin biosynthesis (32,33), suggesting levels of HS and chondroitin are interdependent in vivo. In addition, the disaccharide composition of the HS in the mass culture was comparable with that in the wild type mass culture (Table 3). These findings were similar to those obtained for the rib-2 deletion mutant (28) and strongly suggest that the rib-1 mutant is specifically defective in the synthesis of HS and not chondroitin in the nematode C. elegans.

DISCUSSION
Recent genetic analyses using the nematode C. elegans together with cell biological and biochemical studies have led to reports exploring the mechanism behind the biosynthesis of glycosaminoglycans and the functions of glycosaminoglycans during development (for a review see Ref. 5). In this study, we demonstrated for the first time that RIB-1 and RIB-2 are indispensable for the biosynthesis of HS chains in vivo. In addition, we showed that in C. elegans, HS is essential for embryonic morphogenesis in the later stages of development.
Although five EXT family genes (EXT1, EXT2, EXTL1, EXTL2, and EXTL3) have been identified in the human genome, only two homologs, rib-1 and rib-2, have been found in the C. elegans genome. Previously, we revealed that rib-2 encodes GlcNAcT involved in initiating the biosynthesis and in the elongation of HS but not GlcAT-II required for lengthening the HS chains (27). Because HS is found in C. elegans (15,16), it was predicted that rib-1 encodes the GlcAT-II. However, when a soluble recombinant form of RIB-1 was produced and assayed for GlcAT-II and also GlcNAcT activities, no such glycosyltransferase activities were detected. In this study, despite little glycosyltransferase activity by RIB-1, we showed that the polymerization of HS in C. elegans was achieved by an enzyme complex composed of RIB-1 and RIB-2, which is analogous to the findings made previously for the human EXT1 and EXT2 protein complex (9). In this regard, it should be noted, however, that RIB-1 is unique among members of the EXT family identified to date in that the protein, composed of 382 amino acids, is about half the size of RIB-2 and shows significant homology only to the amino termini of EXT family members, especially to EXT1 (45% amino acid identity, see Fig. 1A). Moreover, RIB-2 is most homologous to human EXTL3 (not to EXT2) (see Fig. 1B). In fact, as mentioned above, RIB-2 harbors both GlcNAcT-I and GlcNAcT-II activities involved in the chain initiation and elongation of HS, and its acceptor specificity resembles that of human EXTL3 (12,27), suggesting that rib-2 is a C. elegans EXTL3 ortholog (Fig. 1B). These results together indicate that the mechanism of HS biosynthesis in C. elegans is similar but distinct from that in humans and that both RIB-1 and RIB-2 are indispensable for the biosynthesis of HS chains in the nematode.
The human EXT1 and EXT2 genes are involved in HME, an autosomal dominant disorder characterized by the formation of a cartilage-capped tumor, caused by mutations in either EXT1 or EXT2 (12). The family of EXT genes has been extended with the identification of three additional EXT-like genes, EXTL1, EXTL2, and EXTL3 (4). Although the EXT-like genes are predicted to be involved in the synthesis of HS on the basis of the enzymatic activities of their products in vitro (13,14), these genes are not linked with HME and have not been associated with any disease pathology, and so their functions in vivo have not been determined. In fact, the polymerization of HS on the linkage analog is achieved by an enzyme complex of EXT1 and EXT2, without the aid of EXTL proteins (8 -10). Because the polymerization of HS was achieved with low efficiency by a complex of RIB-1, the EXT1 counterpart in C. elegans, and RIB-2, the EXTL3 counterpart in C. elegans, and the amount of HS in C. elegans was much smaller than that of chondroitin (15,16) (see Table 2), it is likely that EXT1 and EXTL3 can also cooperate to polymerize HS chains albeit with low efficiency in mammals.
EXT1-and EXT2-deficient mice generated by gene targeting failed to undergo gastrulation and died by embryonic day 8.5, pointing to the early essential role for HS in mammalian embryonic development. In fact, embryonic stem cells from EXT1and EXT2-deficient mice showed a loss of HS (44,45). Interestingly, it has been reported that EXT1 mutant mice generated by the gene trap method (EXT1 Gt/Gt ) survive to embryonic day 14.5 (46). In addition, Yamada et al. (47) showed that the embryonic fibroblasts established from mice with the gene trap mutation in EXT1 produce short HS chains as compared with those from wild type mice. The gene trap cassette is inserted between exons 1 and 2 in the EXT1 gene. Exon 1 encodes 43% of all the amino acids (i.e. 320 of 746), indicating that the mutant EXT1 still contains almost half of the amino-terminal side of the protein. Thus, they suggested that the presence of HS in the gene trap mouse could be a result of residual transferase activity in the amino-terminal part of the protein. As mentioned above, the RIB-1 protein, composed of 382 amino acids, is about half the size of EXT1 and shows significant homology only to the amino termini of the EXT family members, especially to EXT1. Therefore, it is possible that the mutant EXT1 protein formed  in the gene trap mouse can interact with EXTL3 and that these two proteins might cooperate to produce short HS chains. The mechanism for the biosynthesis of HS is reminiscent of that for chondroitin in C. elegans. The repeating disaccharide region of chondroitin is synthesized by glycosyltransferases encoded by sqv-5 (cChSy) (32,48) and pfc-1 (PAR2.4) (33), which form an enzyme complex (chondroitin polymerase) (see Fig. 3, lane 4). RNAi depletion of SQV-5 or PFC-1 in C. elegans resulted in a reduction of chondroitin, and the RNAi-treated worms showed similar phenotypes, such as cytokinetic defects in early embryogenesis (32,33). Thus, it has been suggested that sqv-5 and pfc-1 cannot have redundant roles in the biosynthesis of chondroitin, which is similar to the observation that rib-1 or rib-2 null mutant worms showed reduced synthesis of HS and similar phenotypes, such as post-gastrulation and egg-laying defects. Therefore, both RIB-1 and RIB-2 have nonredundant functions as do SQV-5 (cChSy) and PFC-1 (PAR2.4). In addition, these observations suggest that the mechanism for the biosynthesis of HS is similar to that for chondroitin in C. elegans and clearly indicate that the functions of HS are different from those of chondroitin in C. elegans.
Several recent studies in C. elegans have greatly expanded our understanding of how specific modifications to HS can affect different signaling pathways in different tissues (2). In this study and in the related study of Morio et al. (28), mutations in rib-1and rib-2-encoded proteins involved in the polymerization of HS affected post-gastrulation morphogenic movement and the subsequent elongation in F1 embryos as well as egg-laying and other phenotypes in P0 worms. We detected defects in ventral enclosure and epithelial morphogenesis in rib-1 Ϫ/Ϫ F1 worms, and the results suggest the possible involvement of HS in epidermal morphogenesis as well. In this respect, Franks et al. (41) reported genetic interactions between pyr-1 and rib-1/rib-2 system. They also showed that the rib-1 and rib-2 system play vital roles in body elongation and pharyngeal isthmus morphogenesis. These results strongly indicate the indispensable roles of HS in post-gastrulation and the subsequent morphogenesis.
In contrast, HS-modifying enzymes are not required for viability in C. elegans. Bülow and Hobert (39) isolated mutants of the HS 6-O-sulfotransferase, the glucuronyl C5-epimerase, and the HS 2-O-sulfotransferase, and they found each to be viable and fertile, indicating a more specialized function of the modification to HS. Further analyses showed that each of the mutant worms exhibits distinct yet overlapping axonal and cellular guidance defects in specific classes of neurons. Hence, it has been proposed that individual modifications to HS in a given area or on specific neurons give rise to a "sugar code" that regulates axonal guidance by modulating specific receptor-ligand interactions (39). Additional regional specificity in HS modification patterns can be brought about by modifying core proteins, such as syndecan (49 -51). However, no studies have yet addressed the fundamental questions of whether signaling molecules interact differently with specifically modified HS sequences and how these sequences affect particular signaling pathways. The development of robust and highly sensitive sequencing methods for HS chains will provide definitive answers to these questions in the future.