Structural Analysis of Glycosaminoglycans inDrosophila and Caenorhabditis elegans and Demonstration That tout-velu, a Drosophila Gene Related to EXT Tumor Suppressors, Affects Heparan Sulfate in Vivo *

We have devised a sensitive method for the isolation and structural analysis of glycosaminoglycans from two genetically tractable model organisms, the fruit fly, Drosophila melanogaster, and the nematode, Caenorhabditis elegans. We detected chondroitin/chondroitin sulfate- and heparan sulfate-derived disaccharides in both organisms. Chondroitinase digestion of glycosaminoglycans from adult Drosophilaproduced both nonsulfated and 4-O-sulfated unsaturated disaccharides, whereas only unsulfated forms were detected in C. elegans. Heparin lyases released disaccharides bearingN-, 2-O-, and 6-O-sulfated species, including mono-, di-, and trisulfated forms. We observed tissue- and stage-specific differences in both chondroitin sulfate and heparan sulfate composition in Drosophila. We have also applied these methods toward the analysis of tout-velu, anEXT-related gene in Drosophila that controls the tissue distribution of the growth factor Hedgehog. The proteins encoded by the vertebrate tumor suppressor genes EXT1and 2, show heparan sulfate co-polymerase activity, and it has been proposed that tout-velu affects Hedgehog activity via its role in heparan sulfate biosynthesis. Analysis of total glycosaminoglycans from tout-velu mutant larvae show marked reductions in heparan sulfate but not chondroitin sulfate, consistent with its proposed function as a heparan sulfate co-polymerase.

Proteoglycans consisting of core proteins with glycosaminoglycan chains are abundant molecules, found both in the extracellular matrix and on the cell surface. These diverse molecules serve a wide range of functions, from affecting the compressive properties of cartilage to growth factor reception. Until recently, proteoglycans were studied principally in vertebrate systems. However, genetic experiments in the fruit fly, Drosophila melanogaster, established that proteoglycans, and their associated glycosaminoglycans, are required for normal development of this invertebrate model organism (reviewed in Ref. 1). A Drosophila member of the glypican family, division abnormally delayed (dally) 1 (2,3), affects signaling mediated by two conserved growth factors, Wingless, a member of the Wnt family, and Decapentaplegic, a transforming growth factor-␤/bone morphogenetic protein-related protein (3,4). Wnts and transforming growth factor-␤/bone morphogenetic proteins are important patterning molecules in vertebrate and invertebrate species, and studies of Drosophila and Caenorhabditis elegans have identified many of the evolutionarily conserved components of these signaling systems (5,6).
Mutations affecting genes encoding proteins related to known glycosaminoglycan biosynthetic enzymes have also been described in Drosophila. sugarless shows striking homology to UDP-glucose dehydrogenase (7)(8)(9) and affects signaling mediated by multiple growth factors, including Wingless, Decapentaplegic, and the fibroblast growth factor receptor-related proteins Heartless and Breathless (10). sulfateless encodes a protein similar to N-deacetylase/N-sulfotransferase and is also required for Wingless-mediated and fibroblast growth factor receptor signaling (10,11). Both sugarless and sulfateless mutations disrupt glycosaminoglycan-modification of Dally in vivo, supporting their assignment as glycosaminoglycan biosynthetic enzymes (3,11). pipe, a gene required for establishing the embryonic dorsal-ventral axis, encodes a protein with significant homology to heparan sulfate 2-O-sulfotransferase genes in vertebrates (12). Finally, tout-velu (ttv), a gene related to the tumor suppressor genes, (EXT1 and 2) has been shown to affect events directed by Hedgehog, a Drosophila homolog of Sonic Hedgehog and Indian Hedgehog (13). It is not known, however, whether tout-velu, like the vertebrate EXT1 and 2 genes, encodes an enzyme with heparan sulfate co-polymerase activity (14).
Biochemical studies of proteoglycans from both Drosophila and C. elegans have been reported. In addition to the glypican, Dally, a Drosophila syndecan has been identified and shown to be heparan sulfate-modified (15). Heparan sulfate and chondroitin sulfate polymers have been detected in extracellular matrix preparations of Drosophila, material that could be radiolabeled with 35 SO 4 and degraded with either chondroitinase ABC or nitrous acid (16). Other proteoglycan or proteoglycanlike molecules have been described in Drosophila, including DROP-1 (17), Papilin (18), and macrophage-derived proteoglycan-1 (a hemocyte/macrophage-derived protein of the extracellular matrix) (19). A gene encoding a protein related to perlecan, unc-52, has been studied in some detail in C. elegans and shown to affect muscle attachment and sarcomere organization (20,21).
Although these findings collectively show that proteoglycans exist in Drosophila and C. elegans, performing critical functions during development, very little is known about the different glycosaminoglycan structures found in these organisms. A detailed understanding of proteoglycan and glycosaminoglycan functions in these systems will require structural information that can be used in conjunction with genetic and molecular data. We therefore devised a method for structural analysis of glycosaminoglycan-derived disaccharides suitable for the relatively small samples that can be easily obtained from these animals. We have applied these methods toward identifying tissue-specific and developmental stage-specific distributions of glycosaminoglycans, as well as to characterize glycosaminoglycans in animals bearing mutations in ttv, a gene proposed to affect heparan sulfate biosynthesis.
Determination of Unsaturated Disaccharides from Heparan Sulfate and Chondroitin Sulfate-Unsaturated disaccharides produced enzymatically from heparan sulfate and chondroitin sulfate were determined by a reversed-phase ion-pair chromatography with sensitive and specific postcolumn detection (23). A gradient was applied at a flow rate of 1.1 ml/min on a Senshu Pak Docosil (4.6 ϫ 150 mm) at 55°C. The eluents used were as follows: A, H 2 O; B, 0.2 M sodium chloride; C, 10 mM tetra-n-butylammonium hydrogen sulfate; D, 50% acetonitrile. The gradient program was as follows: 0 -10 min, 1-4% eluent B; 10 -11 min, 4 -15% eluent B; 11-20 min, 15-25% eluent B; 20 -22 min, 25-53% eluent B; 22-29 min, 53% eluent B; equilibration with 1% B for 20 min. The proportions of eluent C and D were constant at 12 and 17%, respectively. To the effluent were added aqueous 0.5%(w/v) 2-cyanoacetamide solution and 0.25 M sodium hydroxide at the same flow rate of 0.35 ml/min by using a double plunger pump. The mixture passed through a reaction coil (internal diameter, 0.5 mm; length, 10 m) set in a dry reaction temperature controlled bath at 125°C and a following cooling coil (internal diameter, 0.25 mm; length, 3 m). The effluent was monitored fluorometrically (excitation, 346 nm; emission, 410 nm).
Preparation and Enzymatic Digestion of Chondroitin Sulfate-Adult worms were obtained by collection from agar plates, washed, and separated from E. coli on a 0.25% Ficoll gradient as described (24). For the analysis of both C. elegans and Drosophila, whole animals or tissues were first lyophilized to dryness. Approximately 20 mg of lyophilized sample was then homogenized with 1.0 ml of acetone. The homogenate was washed with acetone and dried. The pellet was extracted in 1.0 ml of 0.5% SDS, 0.1 M NaOH, 0.8% NaBH 4 for 16 h at room temperature with constant stirring. Two hundred l of 1.0 M sodium acetate and 300 l of 1 M HCl were then added, the solution was filtered, and 200 l of 1 M HCl was added to the filtrate. Insoluble material was removed by centrifugation at 2500 ϫ g for 10 min at 4°C. Seven ml of ethanol was added to the supernatant and chilled for 2 h at 0°C, and the crude glycosaminoglycan fraction collected by centrifugation at 2500 ϫ g for 10 min at 4°C. The resulting precipitate was dissolved in 250 l of water. A 20-l portion of the crude glycosaminoglycan solution was diluted to 100 l with water and used for the determination of chondroitin sulfate. For chondroitinase digestion, 5 l of 0.2 M Tris-acetate buffer (pH 8.0) and 10 l of an aqueous solution containing chondroitinase ABC (50 mIU) and chondroitinase ACII (50 mIU) were added to a 20-l portion of the sample solution and incubated at 37°C for 3 h. An 8-l portion of this mixture was loaded onto the high performance liquid chromatograph.
Preparation and Enzymatic Digestion of Heparan Sulfate-To 230 l of crude glycosaminoglycan sample, 50 l of 0.3 M sodium phosphate buffer (pH 6.0) was added, and the solution was applied on an Ultrafree-MC DEAE membrane, which had been equilibrated with sodium phosphate buffer (pH 6.0) containing 0.15 M NaCl. The fractions eluted with 1.0 M NaCl in the same buffer were collected, desalted with Biomax-5, evaporated, and resuspended in 12 l of water in preparation of heparin lyase digestion. For the analysis of Drosophila heparan sulfate, 5 l of 0.1 M acetate buffer (pH 7.0) with 10 mM calcium acetate and 15 l of an aqueous solution containing heparin lyase mixture (Seikagaku America), heparin lyase I (1 mIU), heparin lyase II (1 mIU), and heparin lyase III (1 mIU) were added to a 5-l portion of sample. The mixture was incubated at 37°C for 16 h, and an 8-l aliquot was loaded onto the high performance liquid chromatograph. For the analysis of heparan sulfate from C. elegans, chondroitin was removed from the crude glycosaminoglycan solution prior to heparin lyase digestions by chondroitinase treatment followed by separation with Ultrafree-MC DEAE membrane. Enzymatic digestion with a heparin lyase mixture was then carried out as described above, with the exception that Sigma enzymes were used.
Genetic Analysis of tout-velu-We used tout-velu l(2)00681 mutants for our analysis, the only allele for which genetic studies have been described (13). ttv/CyO, P[wt GFP] green fluorescent protein animals were self-crossed, and ttv/ttv third instar larvae were identified by the lack of green fluorescence derived from the green fluorescent protein marked CyO balancer chromosome.

RESULTS
Sample Preparation and HPLC for the Determination of Glycosaminoglycans-We established a protocol for highly reproducible and sensitive HPLC analysis of unsaturated disaccharides from chondroitin sulfate and heparan sulfate in C. elegans and Drosophila. The coefficient of variation for each unsaturated disaccharide was less that 5% (adult Drosophila, n ϭ 5). The lower determination limits of the HPLC for chondroitin sulfate and heparan sulfate were approximately 0.5 and 1.5 ng, respectively. All results described below were reproduced in at least duplicate experiments.
Analysis of Glycosaminoglycans in C. elegans-The complete digestion of material from C. elegans with both chondroitinase ABC and ACII released disaccharides that co-chromatographed with ⌬Di-0S using reverse phase ion-pair chromatography (Fig.  1B). The identity of ⌬Di-0S was established using graphitized carbon chromatography, which can resolve ⌬Di-0S and ⌬Di-HA, showing that chondroitin is found in this invertebrate organism (data not shown) (25). Digestion with chondroitinase ACII alone generated the same chromatographic profile obtained with ACII plus ABC, indicating that the majority of chondroitin contains glucuronic acid and not iduronic acid (data not shown). We did not detect ⌬Di-4S, ⌬Di-6S, and other over-sulfated disaccharides found in vertebrate chondroitin sulfate in these unfractionated glycosaminoglycan preparations from C. elegans, nor were we able to detect ⌬Di-HA in our preparations. The compositions of unsaturated disaccharides produced from chondroitin in worms is listed in Table I.
Digestion of the glycosaminoglycans isolated from C. elegans with a heparin lyase mixture generated disaccharides represented in vertebrate samples such as bovine kidney (Fig. 2B). The disaccharide profile of worm heparan sulfate we observed is relatively simple, with ⌬UA-GlcNS, ⌬UA2S-GlcNS, and ⌬UA2S-GlcNS6S species in nearly equal amounts and nonsulfated ⌬UA-GlcNAc representing about 50% of the total. The composition of disaccharides produced from heparan sulfates in worms, compared with that derived from bovine kidney, is provided in Table II. The identities of all disaccharides have been confirmed by comparison with standards using two HPLC separation methods, reversed phase ion-pair (23) and graphi-  tized carbon column chromatography (25).
Analysis of Glycosaminoglycans from Drosophila-As for the analysis of glycosaminoglycan from C. elegans, the identities of Drosophila disaccharides were established by comparison with standards using two distinct separation methods, reverse phase ion-pair (23) and graphitized carbon chromatography (25). Digestion of unfractionated glycosaminoglycans from adult flies with chondroitinases yields disaccharides that cochromatograph with ⌬Di-0S and ⌬Di-4S (Fig. 1C). As we observed for material from C. elegans, the equivalent release of disaccharides with chondroitinase ACII compared with digestion with both ACII and ABC indicates that the majority of polymers are glucuronate containing and hence derived form chondroitin sulfate, not dermatan sulfate (data not shown). The identity of the 4S species was confirmed by its conversion to ⌬Di-0S with chondro-4-sulfatase. Chondroitin sulfate from adult flies has a relatively low degree of sulfation, and as in the human serum protein bikunin, only 4-O-sulfated disaccharides are represented (26). ⌬Di-HA from Drosophila was not detectable using these methodologies. The disaccharide composition of adult Drosophila chondroitin sulfate is listed in Table I.
Heparan sulfate disaccharides are also found in glycosaminoglycan preparations from adult flies (Fig. 2C). Fly heparan sulfate shows a greater degree of complexity than samples from C. elegans, with ⌬UA-GlcNAc, ⌬UA-GlcNS, ⌬UA-GlcNAc6S, ⌬UA-GlcNS6S, ⌬UA2S-GlcNS, and ⌬UA2S-GlcNS6S species generated by digestion with a mixture of heparin lyases I, II, and III. To determine whether the material we detected in Drosophila is a typical heparan sulfate polymer, we identified the disaccharides released by treatment with heparin lyase I or III. Heparin lyase I generated ⌬UA-GlcNS6S, ⌬UA2S-GlcNS disulfated disaccharides, and ⌬UA2S-GlcNS6S trisulfated disaccharide (data not shown). Heparin lyase III generated ⌬UA-GlcNAc, ⌬UA-GlcNS, ⌬UA-GlcNAc6S, and ⌬UA-GlcNS6S disaccharides (data not shown). These findings demonstrate that material from Drosophila has heparin lyase disaccharide profiles typical of heparan sulfate and is not the unusual glycosaminoglycan found in the snail Achatina futica that is resistant to heparin lyase I and III digestion (27). Overall, the proportion of sulfated disaccharides is high, representing about 69% of total compared with 47% in bovine kidney. The composition of unsaturated disaccharides produced from adult Drosophila heparan sulfates is given in Table II.
Analysis of Drosophila Tissues and Developmental Stages-In vertebrates, structural variants of glycosaminoglycans show tissue-specific distributions (28,29). To determine whether Drosophila tissues also show reproducible differences in glycosaminoglycans we examined different tissues and developmental stages in parallel samples from whole adult bodies, ovaries, embryos, and third instar larvae. We detected reproducible differences in the disaccharide profiles of heparan sulfates isolated from Drosophila ( Fig. 3 and Table III). For example, third instar larvae show a reduced percentage of ⌬UA-GlcNS, compared with embryos and adults. Embryos show higher relative amounts of ⌬UA-GlcNS6S compared with larvae, adults, or ovaries. Ovaries show a significantly higher proportion of ⌬UA2S-GlcNS. The ratios of heparan sulfate to chondroitin sulfate also vary widely in Drosophila (Table VI). The ovary showed the greatest proportion of heparan sulfate (heparan sulfate:chondroitin sulfate, 0.74), with larvae showing the lowest (0.06). The degree of 4-O-sulfation of chondroitin sulfate also differed among tissues, ranging from 36% in the ovary to 11% in whole adult flies ( Table I).
Analysis of tout-velu, a Gene Related to the Vertebrate Heparan Sulfate Co-polymerases EXT1 and EXT2-One of the utilities of structural analyses of glycosaminoglycans in Drosophila and C. elegans is to examine the effects of removing specific gene functions on the sugar polymers synthesized in an intact animal. We have used the analytical methods we developed to examine glycosaminoglycans in animals bearing mutations in ttv, a gene with 56% amino acid identity to the vertebrate tumor suppressor gene, EXT1 (13). Both EXT1 and 2 have been shown to encode enzymes with heparan sulfate co-polymerase activity, suggesting that ttv may also affect the synthesis of heparan sulfate (14). We examined glycosaminoglycans in third instar larvae homozygous for a null allele of ttv. ttv mutant larvae show a marked reduction in heparan sulfate, to levels at least 10-fold less than wild type ( Fig. 4 and Tables III and IV). The chromatograph shown in Fig. 4 represents material greater than 5000 Da that dissociates from DEAE membranes between 0.15 and 1.0 M NaCl. Heparan sulfate glycosaminoglycans in this fraction or with molecular mass less than 5000 Da (data not shown) are below the detection limits of our methods. Animals heterozygous for ttv show a slightly reduced amount of heparan sulfate with the same disaccharide composition as wild type. Chondroitin sulfate, however, is unaffected by ttv, consistent with ttv encoding a heparan sulfatespecific co-polymerase.

Drosophila and C. elegans as Model Organisms for Studying
Proteoglycan and Glycosaminoglycan Functions-Many studies in Drosophila have documented the important role of proteoglycans in developmental patterning. Yet at the structural level, little is known about glycosaminoglycans in Drosophila or C. elegans, another model organism that offers a powerful array of genetic and molecular tools. We describe here a sensitive method for analysis of disaccharides derived from chondroitin and heparan sulfates in these model organisms.
Analysis of Chondroitin Sulfate-Chondroitin polymers are found in adult Drosophila and C. elegans. Chondroitin-derived disaccharides in C. elegans were not sulfated, and only ⌬Di-4S was detected in Drosophila. This is in contrast to chondroitin sulfate found in cartilage from a wide range of animals, including the squid, in which the vast majority of disaccharides released by chondroitinase treatment are sulfated at either the 4 or 6-O position (30). Equivalent release of disaccharides with chondroitinase ABC and ACII, or ACII digestion alone, suggests that dermatan sulfate is either not found, or represented at very modest levels in Drosophila and C. elegans.
It is interesting that the profiles of disaccharides generated by chondroitinase treatment of C. elegans and Drosophila ma- terial resemble those generated from treatment of human bikunin (26), an abundant serum protein component of the inter-␣-trypsin inhibitor family of protease inhibitors (reviewed in Ref. 31). This suggests that the chondroitin-modified proteins in these invertebrates may include protease regulators. In fact, a gene with striking homology to mouse bikunin is found in C. elegans, showing greater than 40% amino acid identity over a stretch of 100 amino acids (GenBank TM accession number U64857). The C. elegans gene encodes a protein most similar to tissue factor pathway inhibitor, a member of the bovine pancreatic trypsin inhibitor/Kunitz family of protease inhibitors. Analysis of Heparan Sulfate-Heparin lyase treatment of glycosaminoglycans from Drosophila and C. elegans releases disaccharides found in vertebrates. Our analysis provides the first direct evidence for disaccharides bearing N-, 2-O-, and 6-O-sulfations in these organisms. We did not detect ⌬UA2S-GlcNAc or ⌬UA2S-GlcNAc6S in either of these organisms using the methods that we developed for microdetermination of glycosaminoglycans, but it remains possible that these forms exist, albeit at levels below our current detection limits. Our methods, using heparin lyase digestion of small quantities of crude glycosaminoglycans, is not suitable for detection of 3-O-S sequences. We plan further characterization of heparan sulfate from Drosophila and C. elegans using larger scale preparation of purified material and NMR spectroscopy.
Our findings show that enzymes required for biosynthesis and modification of heparan sulfate must exist in Drosophila and C. elegans. Indeed, genes encoding proteins with significant homology to EXT1 (13), N-deacetylase/N-sulfotransferase (10), C5 glucuronyl epimerase (GenBank TM accession number P46555), and heparan-sulfate sulfotransferase (12,32) enzymes from vertebrates are represented in Drosophila and C. elegans. Our analysis of ttv (see below) indicates that like its vertebrate homolog, EXT1, Ttv affects heparan sulfate biosynthesis. Structural studies of glycosaminoglycans from animals bearing mutations in these genes will show which genes are required for the generation of specific glycosaminoglycan forms in vivo.
One of the most striking features of glycosaminoglycans is their structural diversity, with discrete structural variants found in different tissues (28,33). Specific forms of heparan sulfate are also associated with different disease states and ages (29,34). These findings suggest that different forms of heparan sulfate are important for influencing the biological function of the associated proteoglycan. To determine whether Drosophila could provide a model system for exploring the function of different heparan sulfate structural variants, we examined glycosaminoglycans from different developmental stages and tissues. Indeed, Drosophila does show tissue-and stage-specific modifications of both heparan and chondroitin sulfate. For example, levels of ⌬UA-GlcNS6S are relatively higher in embryos compared with larvae and adults. Given the importance of GlcN 6-O-sulfate groups for binding several growth factors (reviewed in Ref. 33) the levels of these disaccharides are potentially important in regulating growth factor signaling throughout development (36,37).
The differences in 2-O-sulfated disaccharides in the ovary compared with other tissues and stages is worthy of note. Recently it has been shown that pipe, a gene required for establishing dorsal-ventral polarity in the embryo, encodes a protein related to heparan sulfate 2-O-sulfotransferase (12). pipe is expressed only in the ventral follicle cells of the ovary that surround the developing oocyte, and is required for the proteolytic activation of the protein ligand, Spä tzle. Spä tzle in turn activates Toll, a Drosophila homologue of the vertebrate interleukin 1 receptor that specifies ventral cell fates in the future embryo (reviewed in Ref. 38). We demonstrate here that 2-O-sulfated heparan sulfate disaccharides are found in the ovary, in a proportion distinct from that found in embryos, whole adults, or larvae. These findings show that a heparan sulfate 2-O-sulfotransferase must exist in Drosophila, and the homology of pipe to vertebrate proteins with this activity suggests that the pipe could provide this function.
Analysis of ttv, an EXT-related Gene in Drosophila-One of the utilities of organisms like Drosophila and C. elegans is that they provide the means of determining the effects of specific genes on glycosaminoglycan biosynthesis in whole animals. Using the methods we devised for disaccharide analysis in Drosophila, we examined glycosaminoglycans in animals bear-ing mutations in ttv, a gene encoding a protein with 56% amino acid identity to EXT1, a tumor suppressor with heparan sulfate co-polymerase activity (14). We isolated glycosaminoglycans from third instar larvae homozygous for a null allele of ttv. These animals showed a normal level of chondroitinase-sensitive disaccharides but markedly reduced levels of heparan sulfate-derived disaccharides. The heparan sulfate fraction that elutes from DEAE between 0.15 and 1.0 M NaCl and has a molecular mass of greater than 5000 Da was reduced to below detectable limits in ttv. Nor could we detect any material in the same DEAE fraction that passed through a 5000 Da filter (data not shown). Further study using larger quantities of purified glycosaminoglycans will be required to characterize exactly what heparan sulfate forms, if any, remain in ttv mutants. Our findings clearly show that the methods of microdetermination of heparan sulfate-derived disaccharides that we developed can be applied to the analysis of mutants, and demonstrate directly that ttv ϩ is required for normal heparan sulfate levels in larvae. These data support the proposal that this gene does indeed encode a heparan sulfate glycosyltransferase (13).
This finding has implications toward understanding the biological functions of EXT genes in vertebrates. Although it is clear that loss of EXT gene function promotes cartilaginous tumor formation, the molecular mechanism of EXT-mediated tumor suppression in not known. Based on the finding that ttv affects the distribution of Hedgehog in the developing wing, it has been hypothesized that loss of EXT function affects the levels of Indian Hedgehog at the growth plate (13). Indian Hedgehog limits the rate of chondrocyte differentiation, and loss of EXT could disrupt cartilage formation via abnormalities in Indian Hedgehog distribution (39 -41). Our finding that ttv affects heparan sulfate levels provides a link between EXT functions in vertebrates and Drosophila. Recently, analysis of ttv mutants using a monoclonal antibody directed against an epitope generated by heparinase III digestion supports the conclusion that these animals are defective in heparan sulfate biosynthesis (35).