The Caenorhabditis elegans Genes sqv-2and sqv-6, Which Are Required for Vulval Morphogenesis, Encode Glycosaminoglycan Galactosyltransferase II and Xylosyltransferase* 210

In mutants defective in any of eightCaenorhabditis elegans sqv (squashedvulva) genes, the vulval extracellular space fails to expand during vulval morphogenesis. Strong sqv mutations result in maternal-effect lethality, caused in part by the failure of the progeny of homozygous mutants to initiate cytokinesis and associated with the failure to form an extracellular space between the egg and the eggshell. Recent studies have implicated glycosaminoglycans in these processes. Here we report the cloning and characterization ofsqv-2 and sqv-6. sqv-6 encodes a protein similar to human xylosyltransferases. Transfection of sqv-6 restored xylosyltransferase activity to and rescued the glycosaminoglycan biosynthesis defect of a xylosyltransferase mutant hamster cell line. sqv-2 encodes a protein similar to human galactosyltransferase II. A recombinant SQV-2 fusion protein had galactosyltransferase II activity with substrate specificity similar to that of human galactosyltransferase II. We conclude that C. elegans SQV-6 and SQV-2 likely act in concert with other SQV proteins to catalyze the stepwise formation of the proteoglycan core protein linkage tetrasaccharide GlcAβ1,3Galβ1, 3Galβ1,4Xylβ-O-(Ser), which is common to the two major types of glycosaminoglycans in vertebrates, chondroitin and heparan sulfate. Our results strongly support a model in which C. elegans vulval morphogenesis and zygotic cytokinesis depend on the expression of glycosaminoglycans.


jor types of glycosaminoglycans in vertebrates, chondroitin and heparan sulfate. Our results strongly support a model in which C. elegans vulval morphogenesis and zygotic cytokinesis depend on the expression of glycosaminoglycans.
Glycosaminoglycans (GAGs) 1 are important in animal development, and defects in GAGs are responsible for certain human disorders. For example, mutations in the Drosophila melanogastger genes tout-velu (1) and sulfateless (2), which encode homologs of heparan sulfate co-polymerase and heparan sulfate N-deacetylase/N-sulfotransferase, respectively, cause zygotic lethality and defects in segmentation. Mutations in the mouse tout-velu homolog EXT1 disrupt gastrulation and the generation of mesoderm (3), while mutations in human EXT1 and EXT2 have been associated with hereditary multiple exostoses (reviewed in Ref. 4). Mutations in the human galactosyltransferase I have been associated with a progeroid variant of the connective-tissue disorder Ehlers-Danlos syndrome (EDS) (5)(6)(7). EDS is a group of heritable disorders characterized by hyperelasticity of the skin and hypermobile joints. Tout-velu, EXT-1, EXT-2, and Sulfateless affect the biosynthesis of heparan sulfate specifically, while galactosyltransferase I deficiency affects the biosynthesis of both chondroitin and heparan sulfate.
Eight sqv (squashed vulva) genes were genetically identified in a screen for Caenorhabditis elegans mutants defective in vulval morphogenesis (9). All sqv mutants fail to form a large fluid-filled vulval extracellular space and have a reduced separation of the anterior and posterior halves of the vulva from the early to middle phases of L4 larval development. Strong mutant alleles of all eight sqv genes also cause maternal-effect lethality. Most progeny of mothers homozygous for a strong sqv mutant allele arrest at the one-cell stage (9). The nuclei of the arrested progeny divide normally, but the extrusion of the polar bodies and the initiation of cytokinesis are impaired (10). These mutant eggs fail to form the normal fluid-filled extracellular space between the membrane of the egg and the eggshell. We have postulated that the sqv genes control the biosynthesis of GAGs that are secreted and become hydrated to form fluidfilled extracellular spaces (10,11).
The molecular identification of five sqv genes has led to a model implicating the biosynthesis of chondroitin and/or heparan sulfate in C. elegans development. sqv-1, -3, -4, -7, and -8 encode UDP-GlcA decarboxylase (10), galactosyltransferase I * This work was supported by National Institutes of Health (NIH) Grant GM24663 (to H. R. H.) and NIH Grant GM33063 (to J. D. E.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to GenBank™ (12), UDP-glucose dehydrogenase (13), UDP-GlcA/UDP-Gal/ UDP-GalNAc transporter (14), and glucuronosyltransferase I (12), respectively. sqv-3 was used to identify the human galactosyltransferase I, which has been implicated in the progeroid variant of EDS (7,15). In this paper, we show that sqv-6 encodes the xylosyltransferase that adds Xyl to the protein core, thus initiating GAG biosynthesis. sqv-2 encodes a galactosyltransferase that adds the second Gal residue to the linkage tetrasaccharide.

EXPERIMENTAL PROCEDURES
C. elegans Maintenance-Strains were cultured as described (16) and were grown at 20 -22°C unless indicated otherwise.
Molecular Biology-Standard molecular biological techniques were used (17). The sequences of all PCR-amplified DNAs used for cloning were confirmed to exclude unintended mutations. Oligonucleotide sequences used for amplification or mutagenesis of DNA are shown in Supplementary Materials and Methods.
SQV-2 Galactosyltransferase II Assay-A sequence encoding amino acids 25-330 of SQV-2, thus lacking the presumptive transmembrane domain at the amino terminus, was cloned into pDEST-CMV-protA. This plasmid was designed to express a secreted fusion protein containing protein A and SQV-2 amino acids 25-330. COS7 cells were transiently transfected with pDEST-CMV-protA-sqv-2 using Lipofect-AMINE (Invitrogen) according to the manufacturer's instructions. After 72 h of incubation, the fusion protein was recovered from the cell culture supernatant by affinity chromatography using IgG-agarose (20). Galactosyltransferase II activity was assayed as described by Bai et al. (21).
Rescue of the Xylosyltransferase Defect in Chinese Hamster Ovary (CHO) pgsA-745 Cells by sqv-6 -The xylosyltransferase-deficient CHO pgsA-745 cells (22) were transfected with sqv-6 ORF, which was cloned into pcDNA3.1. Stable transfectants were selected with 400 g/ml geneticin (Invitrogen). Several drug-resistant colonies were isolated and screened by flow cytometry for sqv-6 expression based on binding of biotinylated FGF-2 as described (23). Incorporation of 35 SO 4 into GAG chains of wild-type CHO or pgsA-745 cells with or without sqv-6 was assayed essentially as described by Bame and Esko (24), labeling cells overnight at 30°C with 50 Ci/ml 35 SO 4 (PerkinElmer Life Sciences).

RESULTS AND DISCUSSION
Molecular Identification of sqv-2-sqv-2 was previously mapped to the left of lin-31 on LGII (25). We further mapped sqv-2 to an interval between sup-9 and lin-31 (see Supplementary Materials and Methods). We assayed 27 cosmids in this interval for the ability to rescue the sqv-2 mutant phenotype, but none showed rescuing activity (Fig. 1A).
We examined the DNA sequence corresponding to the gaps between the cosmids in this interval and found a predicted gene, Y110A2AL.14, that is weakly similar to galactosyltransferases. Because all previously cloned sqv genes are implicated in the biosynthesis of chondroitin and/or heparan sulfate, we suspected that sqv-2 also encodes a protein involved in GAG biosynthesis. Specifically, it seemed plausible that Y110A2AL.14 encodes the galactosyltransferase II involved in the formation of the protein core linkage tetrasaccharide and that had not been identified molecularly in any organism at the time.
We identified three molecular lesions corresponding to three of the four identified alleles of sqv-2 in the ORF of

sqv-2 and sqv-6 Act in Glycosaminoglycan Biosynthesis
Y110A2AL.14 ( Fig. 1B). The two stronger alleles of sqv-2, n3037 and n3038, cause a maternal-effect lethal phenotype and are an opal nonsense mutation at arginine 225 and a methionine-to-isoleucine missense mutation of the predicted start codon, respectively. A weak allele, n2826, that results in live progeny is a missense mutation causing a glycine-toarginine substitution at amino acid position 99. The molecular lesion of the weakest allele, n2840, has not been identified yet.
We determined the sequences of two cDNA clones, yk94e4 and yk292g2 (see Supplementary Materials and Methods), that correspond to Y110A2AL.14. The yk292g2 clone contains 990 bases of ORF, 17 bases of 5Ј-untranslated region (UTR), and 121 bases of 3Ј-UTR. The 5Ј end contains three bases that correspond to the sequence of 5Ј SL1 trans-spliced leader, which is found at the 5Ј end of many C. elegans transcripts (26). The 3Ј end contains a poly(A) sequence. The longest ORF in this cDNA is identical to Y110A2AL.14 and is predicted to encode a protein of 330 amino acids. The yk94e4 clone lacks the 5Ј end of Y110A2AL.14. Expression of the longest ORF in yk292g2 under the control of the C. elegans heat-shock promoters (19) rescued the defect in sqv-2 vulval morphogenesis in all five isolated lines.
sqv-2 Encodes a Protein Similar to Galactosyltransferase II-The predicted SQV-2 protein contains a putative transmembrane domain near the amino terminus, suggesting it may be a type II transmembrane protein (Fig. 1B). Most known glycosyltransferases that act in the lumen of the ER and the Golgi apparatus are type II transmembrane proteins. Of 330 amino acids of SQV-2, 93 (28%) and 133 (40%) are identical to the Drosophila and human homologs, respectively (Fig. 1B). Recently, the human homolog of SQV-2 was identified as galactosyltransferase II (21).
SQV-2 Has Galactosyltransferase II Activity-We assayed recombinant protein A-SQV-2 fusion protein expressed in COS7 cells for galactosyltransferase II activity (see "Experimental Procedures"). The SQV-2 fusion protein specifically catalyzed the addition of galactose to a disaccharide acceptor, Gal␤1,4Xyl␤-O-benzyl that had been used to demonstrate the acceptor substrate specificity of the human galactosyltransferase II (21) ( Table I).

sqv-2 and sqv-6 Act in Glycosaminoglycan Biosynthesis
further mapped sqv-6 to the left of the cosmid W07B8 and within about 0.2 map units of unc-34 (see Supplementary Materials and Methods). We assayed 11 cosmids to the right of unc-34 for the ability to rescue the sqv-6 mutant phenotype, but none showed rescuing activity ( Fig. 2A).
We examined the DNA sequences in the gaps in the cosmid coverage near the cosmid W07B8 and unc-34 and found a gene, Y50D4C.d, that is similar to two human xylosyltransferases (27). Using DNA from the only allele of sqv-6, n2845, we identified in the ORF of Y50D4C.d an amber nonsense mutation causing the elimination of the last 42 amino acids of the predicted protein product (Fig. 2B).
We determined the sequence of PCR-amplified cDNA and 5Ј-rapid amplification of cDNA ends products corresponding to Y50D4C.d (see Supplementary Materials and Methods). We found that this cDNA contains a 5Ј SL1 trans-spliced leader, 23 bases of 5Ј-UTR, and 2418 bases of ORF, including two additional 5Ј exons not in Y50D4C.d. The longest ORF in this cDNA including the additional exons is predicted to encode a protein of 806 amino acids. Expression of this ORF under the control of the C. elegans heat-shock promoters (19) prior to the start of vulval morphogenesis rescued the sqv-6 vulval morphogenesis defect in all animals (n ϭ 13) and the maternal-effect lethality of the progeny of sqv-6 homozygotes generated by ϩ/sqv-6 heterozygous parents for three of 13 sqv-6 homozygotes studied.
sqv-6 Encodes a Protein Similar to Xylosyltransferases-Of the 806 amino acids of the SQV-6 protein, 182 (23%) and 193 (24%) are identical to human xylosyltransferases I and II, respectively (Fig. 2B). Both the predicted SQV-6 protein and the human xylosyltransferase II contain a putative transmembrane domain near the amino terminus and are likely to be type II transmembrane proteins. Neither the start codon nor a presumptive transmembrane domain has been defined for human xylosyltransferase I (27).
sqv-6 Can Correct a Xylosyltransferase Defect in CHO Cells-We tested the ability of sqv-6 to act as a xylosyltransferase by testing its ability to complement GAG-deficient CHO mutant cells lacking this enzymatic activity (22). Mutant pgsA-745 cells were transiently transfected with a plasmid containing sqv-6 under the control of a cytomegalovirus (CMV) promoter. These cells showed partial rescue of the defect, as assayed by the restored abilities to incorporate 35 SO 4 into GAGs (16 -27% that of the wild type) and to bind biotinylated FGF-2, which binds cell surface heparan sulfate as assayed by flow cytometry (data not shown). From these transiently transfected cells, we obtained a clonal cell line stably expressing sqv-6. This cell line showed full restoration of FGF-2 binding to heparan sulfate on the cell surface (Fig. 3A). Stable expression of sqv-6 in pgsA-745 cells enhanced the incorporation of 35 SO 4 into GAGs to ϳ50% of wild-type levels, compared with 1% for the untreated mutant or mutant transfected with empty vector (Fig. 3B). The 35 SO 4 incorporation into GAGs was similar in wild-type CHO cells and pgsA-745 cells transfected with sqv-6: 30 -40% was released by treatment with chondroitinase ABC and 55-65% by a heparin lyase mixture in both cells, indicating that the composition of chondroitin and heparan sulfate was comparable. Expression of sqv-6 also resulted in restoration of xylosyltransferase activity, as measured by the transfer of xylose from UDP-xylose to a soluble silk acceptor (22), whereas pgsA-745 cells transfected with empty vector had virtually no activity (Fig. 3C).
The sqv-2 and sqv-6 Genes Act in the C. elegans Chondroitin and Heparan Sulfate Biosynthesis Pathway-Our findings indicate that sqv-2 and sqv-6 encode galactosyltransferase II and xylosyltransferase, respectively. With the previously identified sqv-3 galactosyltransferase I and sqv-8 glucuronosyltransferase I, all four C. elegans genes responsible for the biosynthesis of the proteoglycan core protein linkage tetrasaccharide of chondroitin and heparan sulfate have now been defined (Fig.  4). Three previously identified genes, sqv-4 UDP-glucose dehydrogenase, sqv-1 UDP-GlcA decarboxylase and sqv-7 UDP-GlcA/UDP-Gal/UDP-GalNAc transporter, act in earlier steps of GAG biosynthesis. All sqv genes identified to date affect the biosynthesis of both chondroitin and heparan sulfate. Based upon these observations, we conclude that in C. elegans early embryonic cytokinesis and epithelial invagination during vulval development depend on the expression of GAGs.