Identification of a Drosophila gene encoding xylosylprotein beta4-galactosyltransferase that is essential for the synthesis of glycosaminoglycans and for morphogenesis.

In mammals, the xylosylprotein beta4-galactosyltransferase termed beta4GalT7 (XgalT-1, EC ) participates in proteoglycan biosynthesis through the transfer of galactose to the xylose that initiates each glycosaminoglycan chain. A Drosophila cDNA homologous to mammalian beta4-galactosyltransferases was identified using a human beta4GalT7 cDNA as a probe in a BLAST analysis of expressed sequence tags. The Drosophila cDNA encodes a type II membrane protein with 322 amino acids and shows 49% identity to human beta4GalT7. Extracts from L cells transfected with the cDNA exhibited marked galactosyltransferase activity specific for a xylopyranoside acceptor. Moreover, transfection with the cloned cDNA restored glycosaminoglycan synthesis in beta4GalT7-deficient Chinese hamster ovary cells. In transfectant lysates the properties of Drosophila and human beta4GalT7 resembled each other, except that Drosophila beta4GalT7 showed a less restricted specificity and was active at a wider range of temperatures. Drosophila beta4GalT7 is expressed throughout development, with higher expression levels in adults. Reduction of Drosophila beta4GalT7 levels using expressed RNA interference (RNAi) in imaginal discs resulted in an abnormal wing and leg morphology similar to that of flies with defective Hedgehog and Decapentaplegic signaling, which are known to depend on intact proteoglycan biosynthesis. Immunohistochemical analysis of tissues confirmed that both heparan sulfate and chondroitin sulfate biosynthesis were impaired. Our results demonstrate that Drosophila beta4GalT7 has the in vitro and in vivo properties predicted for an ortholog of human beta4GalT7 and is essential for normal animal development through its role in proteoglycan biosynthesis.

Proteoglycans are polyanionic molecules consisting of different core proteins covalently attached to a variety of glycosaminoglycan (GAG) 1 chains. They are ubiquitously expressed not only on the cell surface but also in extracellular matrices (1) and are generated in a cell type-specific manner resulting in characteristic sulfation of GAG chains. The strictly regulated expression patterns of proteoglycans suggest that they play regulatory roles in development, cell proliferation, and differentiation (2) and in organogenesis (3). Mutational analyses in the mouse and in Drosophila have identified specific cell types and signaling pathways that are defective if proteoglycan synthesis is disrupted (4,5).
The biosynthesis of sulfated GAG chains on proteoglycans is initiated by the addition of Xyl to Ser or Thr residues in the core protein, followed by the addition of two Gal residues and a GlcUA residue to form the common core sequence (1). Addition of a GlcNAc or a GalNAc residue to the terminal GlcUA of the core leads to the formation of heparin/heparan sulfate or chondroitin/dermatan sulfate GAG chains, respectively. Glycosyltransferases involved in the sequential transfer of individual sugars have recently been cloned, including glucuronyltransferase I (6,7), galactosyltransferase I (xylosylprotein 4-␤-galactosyltransferase, ␤4GALT7, XGalT1) (8,9), galactosyltransferase II (10), and xylosyltransferase (11).
We previously isolated cDNAs of human galactosyltransferase I on the basis of BLAST searches using the sequence of Caenorhabditis elegans sqv-3 (8) and mapped the gene to the human ␤4GALT7 locus (12). sqv-3 is one of three genes isolated by Herman et al. (13,14) from C. elegans mutants defective in vulval epithelial invagination. They predicted that these proteins encode components of a conserved glycosylation pathway that assemble a C. elegans carbohydrate moiety. In fact, it was demonstrated that sqv-8 encodes GlcUA transferase I (15), and sqv-7 encodes a nucleoside-diphosphate sugar transporter (16). As for sqv-3, after we reported the cloning and characterization of human ␤4GalT7 (h␤4GalT7) (8), sqv-3 was shown to encode the C. elegans homolog of galactosyltransferase I (15).
In the present study, we isolated cDNA clones of a Drosoph-ila homolog of h␤4GalT7, based on BLAST analysis of expressed sequence tags (EST) and demonstrated the galactosyltransferase activity of the cloned cDNA product in vitro and in mutant Chinese hamster ovary (CHO) cells lacking ␤4GalT7 activity. We show that Drosophila ␤4GalT7 (D␤4GalT7) has similar enzymatic properties to h␤4GalT7 and that down-regulation of this gene by RNA interference (RNAi) impairs GAG synthesis and alters the morphology of the Drosophila wing and leg in a manner consistent with the disruptions in proteoglycan synthesis.
Construction of Expression Vectors-A cDNA fragment encoding the open reading frame of D␤4GalT7 was prepared by PCR using a 5Ј primer containing a KpnI site, 5Ј-AGCGGTACCAAATGGTGAATATA-TCC-3Ј (nucleotides 87-112), and a 3Ј primer containing a KpnI site, 5Ј-AACGGTACCCATCAGGTTTGTACCGC-3Ј (nucleotides 1052-1077) and EST clone CK02622 (GenBank TM accession number AA142310) as a template. The PCR product was inserted into the KpnI sites of pGEM 3 vector (provided by Dr. Y. Nishida at Nagoya University). An XhoI-XbaI fragment from AA142310 clone was cut and inserted into an XhoI-XbaI site of pMIKneo vector (provided by Dr. K. Maruyama at Tokyo Medical and Dental University) to construct a mammalian cell expression vector.
Cell Culture-Drosophila Kc cells were grown in Shields and Sang M3 insect medium (Sigma) supplemented with 2% fetal calf serum (FCS) at 27°C. Mouse fibroblast L cells were grown in Dulbecco's modified minimum essential medium supplemented with 7.5% FCS at 37°C in a 5% CO 2 atmosphere. CHO mutant pgsB-761 (17) was obtained from the American Type Culture Collection and grown in F-12K medium (Invitrogen) supplemented with 10% FCS.
Preparation of Membrane Fraction-Kc cells were transiently transfected with an expression plasmid with CELLFECTIN Reagent (Invitrogen) according to the manufacturer's instruction. L cells were transiently transfected with an expression plasmid by the DEAE-dextran method (18). After 48 h of culture, the cells were harvested, and the membrane fraction was prepared as described (19).
Galactosyltransferase Assay-As a standard assay system, the galactosyltransferase activity was determined according to Lugemwa et al. (20) with modification. The assay mixture containing 1 l of Me 2 SO, 15 mM MnCl 2 , 50 mM KCl, 1% Triton X-100, 100 mM MES buffer, pH 6.5, 0.6 mM UDP-Gal, 5000 dpm/l UDP-[ 3 H]Gal, 20 g of the enzyme from Kc cells or 3 g of the enzyme from L cells, and substrates in a total volume of 25 l. After incubation at 37°C for 120 min, the reaction mixture was applied onto a Sep-Pak C18 cartridge (Waters), and the product was eluted with 5 ml of methanol. The radioactivity in the eluates was measured in a liquid scintillation counter (Beckman). To analyze effects of temperature, pH, cations, or detergents, the enzyme activity was measured at 17, 22, 27, 32, or 37°C for temperature, at pH 5.5 to 7.5 for acidity, in the presence of Ca 2ϩ , Mg 2ϩ , and Mn 2ϩ or EDTA for cations, and with Nonidet P-40, Triton CF-54, or Triton X-100 for detergents. Donor specificity was analyzed using UDP-Gal, UDP-Gal-NAc, or UDP-GlcNAc. Acceptor specificity was analyzed with p-Nph-␤- The enzyme activity was measured as described above unless otherwise described.
Flow Cytometry Analysis-CHO pgsB-761 cells were transfected with 10 g of plasmid DNA using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. Three days later, cells were trypsinized and prepared for flow cytometry using mAb 10E4 at a dilution of 1:25 (40 g/ml) and FITC-conjugated goat anti-mouse IgM ( chain specific) (Zymed Laboratories Inc.) on a FACSCalibur with Cell Quest version 3.1f software (Becton Dickinson).

RT-PCR Using a Drosophila Expression Panel-A Drosophila
Rap-idScan gene expression panel was purchased from OriGene Technologies, Inc. (Rockville, MD). The product contains first-strand cDNAs prepared from different Drosophila tissues and developmental stages. Forward primer 331 F (5Ј-ACTGCTGGTGCCGTTTCGAG-3Ј) and the reverse primer 730 R (5Ј-TCCTCTAATCCCCAGCCCCA-3Ј) were selected to avoid introns and to give a fragment of about 400 bp. PCR conditions were: denature of the template at 95°C for 1 min, followed by 95°C for 30 s, 60°C for 30 s, 68°C for 30 s, and 68°C for 1 min for either 25 or 35 cycles.
RNA Interference Constructs-Based on the predicted genomic structure of D␤4GalT7 (CG11780), the second exon and second intron of the ␤4GalT gene were cloned from Drosophila genomic DNA by PCR using Taq and Pwo DNA polymerases (Expand TM Long Template PCR System, Roche Molecular Biochemicals) and primers (5Ј-CAAGAAT-TCACTGCGTGTGCCCGCTGTC-3Ј and 5Ј-CCCAGATCTCTGAGGTT-GAATGACAATA-3Ј) with EcoRI and BglII restriction sites. This fragment was cloned into the pUAST vector (21) using the EcoRI and BglII sites in the polylinker. The sequence of the insert in the resulting construct was confirmed. The second exon of the D␤4GalT7 gene was cloned from genomic DNA by PCR using primers (5Ј-CAATCTAGAACT-GCGTGTGCCCGCTGTC-3Ј and 5Ј-CAGCTCGAGCTGAATGTATCAT-TAGTGC-3Ј) that contain an XbaI or XhoI restriction site, and, after confirming the sequence, the PCR product was ligated into the same plasmid in the reverse orientation to generate the double-stranded RNA (dsRNA)-mediated RNAi construct (pUAST-i␤4GalT7), containing two copies of exon 2 in reverse orientation separated by intron 2 and 22 bp of vector sequence (AGATCTGCGGCCGCGGCTCGAG). The ligation reaction from this second step was transformed into SURE-2 cells (Stratagene) to prevent recombination between the inverted repeats. Two independent transgenic lines for pUAST-i␤4GalT7, UAS-i␤4GalT7 [A], and UAS-i␤4GalT7 [B], both on the second chromosome, were generated by P-element-mediated transformation.

Molecular
Cloning of the D␤4GalT7 Gene-The gene mutated in the sqv-3 mutant of C. elegans is similar in amino acid sequence to members of the ␤4GalT gene family of humans and other vertebrates (15) and is required for vulval invagination and oocyte development in C. elegans (13). The human gene corresponding to sqv-3 is now termed ␤4GALT7 (12). To find a Drosophila ortholog, the NCBI Data Bank and Berkeley Dro-sophila Genome Project of EST cDNA clones were probed with the deduced amino acid sequence of h␤4GalT7 cDNA. One cDNA was obtained from the databases (CK02622 contig, accession number AA142310). This cDNA is encoded by a gene predicted by the genomic sequence of Drosophila, CG11780, located at 96B16 on chromosome 3. Based on the biochemical and functional studies outlined below, we have named this Drosophila gene ␤4GalT7 (symbol ␤4G7). The cDNA contains a complete open reading frame encoding a protein of 322 amino acids with a molecular mass of 36,468 Da (Fig. 1A). It has two potential N-linked glycosylation sites. The position of the AUG start codon was determined according to the Kozak consensus sequence (26). Hydropathy analysis (27) indicates one prominent hydrophobic segment of 26 residues in length in the amino-terminal region, predicting that the protein has type II transmembrane topology characteristic of many other glycosyltransferases cloned to date (Fig. 1B). Comparison of the primary structure of the identified cDNA and h␤4GalT7 (Gen-Bank TM accession number AB028600 (6)) revealed that 144 out of the 322 amino acids (49%) are identical (Fig. 2). Approximately 30% identity was found between the newly cloned gene and other human ␤4GalTs (now termed ␤4GalT-1 through ␤4GalT-6). In addition, D␤4GalT7 contains the three short amino acid sequence motifs conserved among all members of the ␤4GalT family (28 -33), i.e. FNRA, DVD, and WG-WGREDDE as indicated in Fig. 2 (34,35).

Restoration of GAG Synthesis by a Cloned D␤4GalT7 cDNA Expressed in a GAG-deficient Mutant Cell-To investigate
whether D␤4GalT7 is involved in the biosynthesis of GAGs in vivo, CHO mutant pgsB-761 (17) cells that lack ␤4GalT7 activity were transiently transfected with pMIKneo-D␤4GalT7. pgsB-761 cells are mutant for ␤4GalT7 and, consequently, are defective in GAG biosynthesis. The ability of CHO cells to synthesize GAGs can be monitored with an antibody that recognizes heparan sulfate, mAb10E4. As shown in Fig. 3, about 50% of the cells transfected with D␤4GalT7 cDNA reacted with mAb10E4, as did a similar number of cells transfected with a h␤4GalT7 cDNA, whereas mock-transfected cells were negative. These results show that the D␤4GalT7 corrected the inability of pgsB-761 mutant cells to synthesize proteoglycans, indicating that the cloned cDNA encodes a Drosophila ortholog of h␤4GalT7.
Galactosyltransferase Activity of D␤4GalT7-To analyze the galactosyltransferase activity of D␤4GalT7, the expression vectors pMIKneo-D␤4GalT7 or pMIKneo-h␤4GalT7 were transfected into L cells, and 3 days later cell extracts were prepared for galactosyltransferase assay to compare the activities of h␤4GalT7 and D␤4GalT7 under various conditions. Back- ground values obtained with transfectants of irrelevant expression constructs were subtracted. Significant activity was observed for p-Nph-␤-D-Xyl, a primer for glycosaminoglycanchain formation but not with other acceptors. To analyze the donor specificity and effects of individual reaction components on enzyme activity, extracts from L cells or Kc cells transfected with pGEM3-D␤4GalT7 or pMIKneo-h␤4GalT7, respectively, were used. Among three nucleoside-sugar donors examined, only UDP-Gal was efficiently transferred to p-Nph-␤-D-Xyl (Fig. 4). D␤4GalT7 enzyme from Kc transfectants showed near maximum activity at around 30°C and stayed similarly active up to 37°C. Almost no activity was detected in extracts from mock-transfected cells (Fig. 5A). The activity of D␤4GalT7 was maximal at pH 6.5, which was similar to the pH optimum of h␤4GalT7 (Fig. 5B). Under standard assay conditions, Mn 2ϩ exhibited the highest activity compared with other cations. Generally, divalent metal cations are essential for the ␤4GalT reaction (36), and 10 mM EDTA completely abolished D␤4GalT7 activity (Fig. 5C). The effects of three different nonionic detergents were compared, and the results showed that, although each gave good activity, Triton X-100 was the best under the conditions tested (Fig. 5D).
Spatio-temporal Expression Pattern of the D␤4GalT7 Gene-To analyze the expression of D␤4GalT7, RT-PCR was performed using a Drosophila RapidScan gene expression panel (Fig. 6). Although the D␤4GalT7 gene was ubiquitously expressed, relatively higher expression levels were observed in RNA from adults compared with embryos at early developmental stages.

RNAi Identifies a Role for D␤4GalT7 in Morphogenesis-
Proteoglycans aid the transport and reception of a number of signaling molecules in Drosophila, including Hedgehog (HH), Wingless (WG), fibroblast growth factor, and Decapentaplegic (DPP) (4,5). Disruption of these signaling events leads to mis-patterning and loss of tissues during development. If D␤4GalT7 is involved in the generation of proteoglycans in vivo, a reduction in the function of this enzyme should result in phenotypes in the fly similar to those observed when proteoglycan-dependent signaling pathways are disrupted. As D␤4GalT7mutants are not available, we used RNA interference (RNAi) to reduce the levels of the ␤4GalT7 transcript (37,38). A construct (i␤4GalT7) containing two copies of exon 2 of the ␤4GalT7 gene in an inverted repeat, such that it could fold back on itself to generate a 622-bp double-stranded RNA (dsRNA) stem with an 89-bp loop, was expressed using the UAS-Gal4 technique (21).
The roles of HH, DPP, and WG signaling pathways have been well studied in the developing imaginal discs of Drosophila (39), and, to examine requirements for ␤4GalT7, dsRNA was expressed under the control of Gal4 lines that drive expression in these tissues. The patched (ptc) promoter drives expression in cells that receive the HH signal. The distance between veins three and four was significantly reduced in the wings of ptc-GAL4 UAS-i␤4GalT7 flies (Fig. 7B). A weak reduction in the vein three to vein four distance was also observed when dsRNA for ␤4GalT7 was expressed throughout the wing using the scalloped-Gal4 driver (data not shown). The loss of tissue in this region of the wing is typical of wing phenotypes of weak mutations in genes in the HH signaling pathway (39 -41). The apterous (ap) and engrailed (en) promoters drive expression in the dorsal and posterior compartments of the wing, respectively. When dsRNA for ␤4GalT7 was expressed under ap-Gal4 or en-Gal4 control, the corresponding regions of the wing were reduced in size (Fig. 7, C and D); this reduction can also be detected in the developing wing imaginal disc (Fig. 8B). In the case of en-Gal4 UAS-i␤4GalT7 flies, the loss of tissue appeared most pronounced between the posterior edge of the wing and the most posterior wing vein, vein five (Fig. 7C). In the case of ap-Gal4 UAS-i␤4GalT7 flies, tissue loss was also more pronounced toward the edges of the wing, manifest most obviously in the complete deletion of vein two from the dorsal surface of the wing (Fig. 7D). Because growth of the wing along the anterior-posterior axis is regulated primarily by DPP (39), these phenotypes can best be explained as resulting from impairment of DPP signaling. Notably, the bristles at the edge of the wing, the wing margin, were unaffected under all conditions examined. These structures are established through WG and Notch signaling, and we infer that these pathways were relatively unaffected. The Distal-less (Dll) promoter drives expression in the distal regions of all appendages. When dsRNA for ␤4GalT7 was expressed under Dll-Gal4 control, the most dramatic phenotypes were observed in the legs, which were truncated distally (Fig. 7F). Similar distal truncations were also occasionally (ϳ15% of legs) observed when i␤4GalT7 was expressed under the control of da-Gal4, which drives expression ubiquitously (data not shown). Distal-proximal outgrowth of the leg is dependent upon the combined action of the HH, DPP, and WG signaling pathways, and impairment of any of these pathways could cause truncations similar to those observed (39).

D␤4GalT7 Is Required for Heparan Sulfate and Chondroitin Sulfate Synthesis in Vivo-
The linker region that ␤4GalT7 is involved in synthesizing is required for both heparan and chrondroitin biosynthesis. We used immunohistochemical methods to directly examine the influence of i␤4GalT7 expression on GAG synthesis in the developing tissues that are phenotypically affected and to assess the impact of i␤4GalT7 expression on heparan sulfate versus chondroitin sulfate. The mAb3G10 recognizes a hexuronate that remains attached to the core linker after Heparinase III treatment and can thus be used as a specific marker of heparan sulfate. Similarly, the mAb 1B5 specifically recognizes the unsulfated product of Chondroitinase ABC treatment and thus can be used as a marker of chondroitin sulfate. Both antibodies specifically stain wild-type wing imaginal discs after the appropriate enzyme treatment (Fig. 8, E and F). Although the effects were slight and somewhat variable, in en-Gal4 UAS-i␤4GalT7 discs, staining with both antibodies is reduced in the posterior, where i␤4GalT7is expressed, relative to the anterior, where it is not expressed (Fig. 8, G and H). Thus the synthesis of both types of GAGs can be impaired by RNAi-mediated repression of ␤4GalT7 expression. DISCUSSION The results presented here show that D␤4GalT7 encodes the ortholog of h␤4GALT7. Based on predicted amino acid sequence, D␤4GalT7 is more similar to mammalian ␤4GalT7 than to the six other mammalian ␤4GalTs (28 -33). In vitro assays showed that it encodes a ␤4GalT with a preference for xylose as acceptor rather than GlcNAc or Glc. In addition, expression of D␤4GalT7 restores GAG synthesis to the CHO mutant pgsB-761, which lacks ␤4GalT7 activity due to a mutation in the hamster ␤4GalT7 gene (17), whereas RNAi of D␤4GalT7 impairs GAG synthesis in Drosophila tissues. Therefore, D␤4GalT7 functions as h␤4GalT7 in vivo and in vitro indicating that it is required for the biosynthesis of GAG core sugars in the linkage region of proteoglycans in the fly.
The enzymatic activity of the ␤4GalT that adds Gal to Xyl was originally studied by Roden's group (43,44), and was further characterized using CHO mutants defective in that activity (17) and in the study of clinical cases (45) with a genetic defect in the biosynthesis of dermatan sulfate proteoglycan. The identification of human ␤4GalT7 as the gene responsible for Xyl ␤4GalT activity (8) revealed that this enzyme almost exclusively utilizes xylose as an acceptor substrate. In the present study, we show that D␤4GalT7 has a slightly broader specificity in acceptor utilization. Intriguingly, the acceptor specificity of the C. elegans homolog, sqv-3 (16), is even broader, because transfer of Gal to p-Nph-␤-D-GlcNAc or p-Nph-␣-D-GlcNAc was 69.5% and 45.1% compared with p-Nph-␤-D-Xyl, respectively (16). Thus, the acceptor specificity of ␤4GalT7 enzymes from different species varies, with that of humans being most restricted. Although a few studies have characterized the chemical structures of Drosophila proteoglycans (46,47), and glycolipids (48), we need to accurately and comprehensively investigate the existing carbohydrate structures in flies with biochemical approaches to clarify the potential significance of differences in the acceptor specificity of ␤4GalT7 among species.
We also investigated optimal conditions for in vitro activity of ␤4GalT7 from Drosophila and human, and found that, with the exception of temperature, they prefer similar assay conditions. Thus, the biochemical properties of D␤4GalT7 are well Both transfectants were harvested, and the membrane fractions were prepared as a source of the enzyme. Enzyme assays proceeded as described under "Experimental Procedures." conserved. The difference in temperature optimum correlates with the difference in typical body temperature: because flies are cold-blooded, their body temperature is variable and is typically 10 -20°C lower than that of humans.
Ever since the activating and inhibitory effects of proteoglycans on cell proliferation were elucidated (49,50), important biological roles for GAG chains on proteoglycans have been increasingly recognized. Their importance for human health was emphasized by the discovery that two different mutations in the human ␤4GalT7 gene were identified in a case of Ehlers-Danlos syndrome (a progeroid variant), revealing that the patient was a compound heterozygote of ␤4GalT7 mutant alleles (51), which generate ϳ10% of the enzyme activity compared with that of normal individuals. By analyzing the phenotypes of flies with RNAi-mediated reduction of ␤4GalT7 levels, we have found that ␤4GalT7 is also required for growth and patterning during Drosophila development and have gained insight into the basis for this requirement.
When dsRNA for D␤4GalT7 was expressed under ptc-Gal4 control in the developing wing, tissue was lost from the middle of the wing, between veins three and four. Experiments involving conditional alleles of hh have shown that tissue between the third and fourth veins of the wing is particularly sensitive to HH signaling and that the level of HH activity defines the distance between veins three and four (40). Similar phenotypes are observed with mutants for the HH pathway gene fused (41) and the gene collier/knot, a transcriptional target of HH sig-naling that is required for the formation of the three to four intervein region (42). The wing phenotypes of animals with RNAi of ␤4GalT7 thus imply that HH signaling is impaired in these animals and, hence, that the ␤4GalT7 gene is required for normal HH signaling in vivo.
When dsRNA for ␤4GalT7 is expressed under ap-Gal4 or en-Gal4 control, wing tissue loss appears to be greatest at the anterior or posterior edges of the wing. Growth of the wing along the anterior-posterior axis is promoted by DPP, which is produced by cells in the middle of the wing, along the anteriorposterior compartment boundary, and then spreads from its site of synthesis. DPP is distributed in a concentration gradient, and cells at the edge of the wing receive the lowest amounts. Wing tissue loss may thus occur primarily at the edge of the wing in these flies, because these cells normally receive the lowest amounts of DPP and, hence, are particularly sensitive to a further reduction in DPP signaling. Alternatively, the preferential loss of tissue at the edge of the wing could also be consistent with an influence on the spread of DPP, because these cells are the farthest from the DPP source.
Because HH and DPP signaling requires proteoglycans (4,5), the ␤4GalT7 RNAi phenotype fits with its biochemical activity, and we infer that its requirement for HH and DPP signaling stems from its role in the synthesis of proteoglycans that in-  fluence transport or reception of these signaling molecules. However, the phenotypes of ␤4GalT7 RNAi flies are less severe than would be expected if HH or DPP signaling were completely eliminated, presumably because RNAi is only partially effective at reducing D␤4GalT7 levels. The lack of WG phenotypes in the adult wing, despite the known requirement for proteoglycans in WG signaling, likely also results from the incomplete impairment of GAG synthesis. Nonetheless, the relatively greater effect on DPP as opposed to WG signaling is unexpected.
In Drosophila, mutations have previously been isolated in four different genes involved in the elongation and modification of GAG chains: sugarless (sgl, UDP-glucose dehydrogenase), fringe connection (frc, UDP-sugar transporter), tout-velu (ttv, EXT, GlcUA/GlcNAc polymerase), and sulfateless (sfl, N-deacetylase/N-sulfotransferase), and in one of the core proteins to which GAG chains are attached, dally (Glypican) (4,5). Genetic analysis of these mutations has revealed the key role of proteoglycans in signaling pathways that are essential for growth and morphogenesis during development. However, WG signaling requires frc, sfl, sgl, and dally, but not ttv; HH signaling requires frc, sfl, ttv, and possibly sgl, but not dally; and DPP signaling requires dally and possibly sgl but has not been reported to require sfl, ttv, or frc (4,5). In part, the apparent lack of requirements for certain genes may stem from genetic redundancy, i.e. the Drosophila genome contains homologs of ttv and dally, and the full complement of UDP-sugar transporters is not known. However, the particular sensitivities of different signaling pathways to mutations at different steps of proteoglycan biosynthesis imply that different pathways have distinct requirements for specific proteoglycans.
Notably, ␤4GalT7 is the first gene to be examined phenotypically in Drosophila that participates in the synthesis of the core linker region to which GAG chains are attached. The genomic sequence predicts only a single ␤4GalT7 gene in Drosophila. The two other putative ␤4GalTs in Drosophila, encoded by CG8536 and CG14517, are more closely related to mammalian ␤4GalTs that utilize GlcNAc acceptors, and RNAi of these genes does not generate the wing phenotypes generated by expression of i␤4GalT7. 2 Thus, D␤4GalT7 should in principle be required for all processes that require GAGs attached to core proteins through the GlcUA-Gal-Gal-Xyl linker, which would encompass all processes dependent upon heparan sulfate or chondroitin sulfate proteoglycans. Consistent with a broad role for D␤4GalT7, gene expression analysis by RT-PCR revealed that this gene is expressed throughout development, including the adult, and immunohistochemical staining has confirmed that it is required for the normal synthesis of both heparan and chondroitin sulfate.
Although both heparan and chondroitin sulfate are affected in flies expressing i␤4GalT7, analysis of other genes suggests that its phenotypic effects can be largely accounted for by the influence on heparan sulfate. That is, biochemical analysis implies that sfl and ttv specifically influence heparan sulfate synthesis, and Dally is expected to be modified exclusively by heparan sulfate (like other glypicans). Because these genes influence HH or DPP signaling, there is in principle no need to invoke an influence on chondroitin sulfate in explaining the i␤4GalT7 phenotypes. The actual requirements for chondroitin sulfate during Drosophila development remain to be determined, because genes that exclusively affect this GAG have not yet been analyzed phenotypically. frc and sgl mutants are expected to impair both heparan and chondroitin sulfate syn-thesis, but aside from a requirement for frc in the Fringe-dependent modulation of Notch signaling (52,53), the phenotypes of these mutations are similar to those of sfl. Nonetheless, it remains possible that the apparent differences in the sensitivities of different pathways to mutation or down-regulation of these genes reflect distinct requirements for different GAGs by signaling pathways. Importantly, the utility of reverse genetic approaches such as RNAi to investigating the requirements for proteins involved in GAG biosynthesis, such as we demonstrate here, should make it possible in the future to elucidate the specific requirements for GAG subtypes in different processes during Drosophila development.