Drosophila heparan sulfate 6-O-sulfotransferase (dHS6ST) gene. Structure, expression, and function in the formation of the tracheal system.

Heparan sulfate, one of the most abundant components of the cell surface and the extracellular matrix, is involved in a variety of biological processes such as growth factor signaling, cell adhesion, and enzymatic catalysis. The heparan sulfate chains have markedly heterogeneous structures in which distinct sequences of sulfate groups determine specific binding properties. Sulfation at each different position of heparan sulfate is catalyzed by distinct enzymes, sulfotransferases. In this study, we identified and characterized Drosophila heparan sulfate 6-O-sulfotransferase (dHS6ST). The deduced primary structure of dHS6ST exhibited several common features found in those of mammalian HS6STs. We confirmed that, when the protein encoded by the cDNA was expressed in COS-7 cells, it showed HS6ST activity. Whole mount in situ hybridization revealed highly specific expression of dHS6ST mRNA in embryonic tracheal cells. The spatial and temporal pattern of dHS6ST expression in these cells clearly resembles that of the Drosophila fibroblast growth factor (FGF) receptor, breathless (btl). RNA interference experiments demonstrated that reduced dHS6ST activity caused embryonic lethality and disruption of the primary branching of the tracheal system. These phenotypes were reminiscent of the defects observed in mutants of FGF signaling components. We also show that FGF-dependent mitogen-activated protein kinase activation is significantly reduced in dHS6ST double-stranded RNA-injected embryos. These findings indicate that dHS6ST is required for tracheal development in Drosophila and suggest the evolutionally conserved roles of 6-O-sulfated heparan sulfate in FGF signaling.

Heparan sulfate proteoglycans are ubiquitously present on the cell surface and in the extracellular matrix and are known to be involved in a variety of biological phenomena, including cell proliferation, differentiation, cell adhesion, angiogenesis, blood coagulation, lipid metabolism, and viral and bacterial infections. These diverse mechanisms of action are achieved by interactions between the specific structures of heparan sulfate and the binding proteins. The backbone of heparan sulfate is a polysaccharide chain composed of alternating D-glucuronic acid (GlcA) and N-acetyl-D-glucosamine units, and some GlcA residues are converted into L-iduronic acids. The microheterogeneity of the heparan sulfate structures is mainly produced by the nonrandom introduction of N-, 2-O-, 6-O-, and 3-O-sulfate groups (1,2). Thus, the biological functions of heparan sulfate proteoglycans are controlled by biosynthetic events, which define the fine structures of heparan sulfate. For many years, functional studies of structurally complex heparan sulfate have focused on vertebrate tissues and cultured cells. Recent studies, however, have shown that heparan sulfate is also found in simple model organisms such as Drosophila melanogaster and Caenorhabditis elegans (3). The structural features of heparan sulfate in Drosophila are similar to those found in vertebrates. In addition, the analyses of heparan sulfate in Drosophila mutants have shown that the biosynthesis apparatus is also conserved in this organism (4). Therefore, on the basis of their homology to vertebrate heparan sulfate-modifying genes, it is now possible to analyze the function of modified heparan sulfate in vivo using these genetically tractable organisms.
Recently, genetic studies on mutants for heparan sulfate sulfotransferases indicated that the regulated synthesis of heparan sulfate affects morphogenesis during development. Bullock et al. (5) reported that heparan sulfate 2-O-sulfotransferase (HS2ST) 1 knock-out mice showed renal agenesis as well * This work was supported in part by the Human Frontier Science Program and a grant-in-aid for scientific research on priority areas from the Ministry of Education, Science, Sports, and Culture of Japan. 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 reported in this paper has been submitted to the DDBJ/GenBank/EBI Data Bank with accession number AB051856.
§ Supported by research fellowships from the Japanese Society for the Promotion of Science for Young Scientists.
It is well established that heparan sulfate binds to bFGF and is required for the activation of the signal transduction pathway. However, the mechanism by which heparan sulfate promotes bFGF action is not clearly understood. It has been variously proposed that heparan sulfate induces a conformational change in bFGF (10), serves as a bridge between bFGF and FGFR (11), or promotes dimerization of FGFRs (12). It has been reported that specific fine structures of heparan sulfate have a critical role for binding to bFGF. Biochemical studies have shown that N-and 2-O-sulfation is required for binding of bFGF to heparan sulfate (13). On the other hand, studies using biological assays have shown that, in addition to 2-O-sulfation, 6-O-sulfation is responsible for the activation of bFGF signaling (14,15). This may reflect the fact that heparan sulfate interacts with not only FGFs but also their receptors. Recent crystallographic analysis also supported the idea that 6-Osulfation promotes the dimerization of FGFRs (16).
In Drosophila, one FGF ligand, Branchless (Bnl), and two FGFRs, Heartless (Htl) and Breathless (Btl) have been identified and characterized (17)(18)(19). btl and bnl are required for the normal development of the Drosophila respiratory system, branching trachea that deliver air to all tissues (19,20). The Drosophila tracheal system shares many features with the vertebrate vascular system and is therefore regarded as a simple model to study tubulogenesis (21). Branching morphogenesis in both Drosophila and mammals is regulated by growth factors FGF and vascular endothelial growth factor (22,23), cadherins (24,25), and hypoxia-inducible factor-1-like basic helix-loop-helix transcription factors (26 -28). Heparan sulfate is also critical for both Drosophila tracheogenesis and vertebrate angiogenesis (7,29).
In this study, we examined the structure and function of Drosophila 6-O-sulfotransferase (dHS6ST). We present here the cDNA and deduced amino acid sequences of dHS6ST and the characterization of its enzymatic activity. In situ hybridization of dHS6ST mRNA revealed that dHS6ST is expressed in specific patterns in several tissues. During embryogenesis, dHS6ST mRNA was detected in developing tracheal precursor cells that also express a Drosophila FGF receptor, btl. Blocking the dHS6ST activity by double-stranded RNA interference caused embryonic lethality and defects in the migration of the tracheal cells. In addition, Btl-dependent activation of MAPK, which is seen normally in tracheal cells, is significantly decreased in these embryos. These results suggested the possible involvement of 6-O-sulfated heparan sulfate in the modulation of FGF signaling during tracheogenesis in Drosophila.
Cloning of dHS6ST-The genomic sequence for a putative Drosophila 6-O-sulfotransferase (dHS6ST) was identified in the Drosophila genome data base of the Berkeley Drosophila Genome Project (BDGP). Based on the sequence, the 938-base pairs PCR-amplified genomic DNA was prepared using the 5Ј-primer (TCTGCCCATCCAAAAAGACG) and the 3Ј-primer (CTGAAAGAGCAAAGTCTTGG). Approximately 2 ϫ 10 5 plaques from a Drosophila larval cDNA library were screened using the above PCR product from dHS6ST genomic DNA as a probe. Hybridization was carried out as described by Sambrook et al. (33). Four positive clones were isolated, and the inserted cDNAs were sequenced using the BigDye Thermal Cycle Sequencing Ready Reaction Kit and the ABI PRISM 310 analyzer. Nucleotide and amino acid sequences were analyzed using a DNASIS program version 3.5 (Hitachi).
Assay for Sulfotransferase Activities and Analysis of the Enzymatic Reaction Products-To express dHS6ST in tissue culture cells, pFLAG-CMV2-dHS6ST was constructed by insertion of the PCR product containing the open reading frame of 1374 base pairs from positions 589 -1962 for dHS6ST into the HindIII/ClaI site of the pFLAG-CMV2 expression vector (Eastman Kodak Co.). COS-7 cells (5.5 ϫ 10 5 ) precultured for 48 h in a 60-mm culture dish were transfected with 5 g of pFLAG-CMV2-dHS6ST or pFLAG-CMV2 alone. The transfection was performed using a liposome-mediated method according to the manufacturer's recommended protocol (TransFast, Promega). The enzyme activities of HS6ST and HS2ST in the cell extract were determined using CDSNS-heparin as a substrate. In this experiment, HS2ST activity was specifically determined in the presence of 10 mM dithiothreitol as described previously (34). Dithiothreitol at this concentration had little effect on HS2ST activity, whereas HS6ST activity was substantially decreased. FLAG fusion proteins in the cell extract were isolated by anti-FLAG M2 (Kodak) affinity chromatography according to the method described by the manufacturer, and the activities of purified dHS6ST in the FLAG-bound fractions were measured using 25 nmol (as hexosamine, 0.5 mM) of acceptor glycosaminoglycans in the 50-l reaction mixture as described previously (35). The amounts of enzymes added to the reaction mixture were chosen to obtain the linear incorporation of [ 35 S]sulfate into CDSNS-heparin, which was 0.028 pmol/min (35). We measured 35 S incorporation activities into various glycosaminoglycans of the purified dHS6ST under those condition. Analysis of the reaction products derived from CDSNS-heparin and heparan sulfate (from a mouse EHS tumor) by expressed enzymes was performed as described previously (35,36).
Whole Mount in Situ Hybridization-Fixation and pretreatment of embryos were performed as described by Tautz and Pfeifle (37). RNA probes were prepared as follows. dHS6ST cDNA was inserted into pBluescript, and purified plasmid DNA was linearized by appropriate restriction enzymes. Sense and antisense RNA probes were prepared by in vitro transcription from linearized cDNA templates using digoxigenin-11-UTP (Roche Molecular Biochemicals) and T7 or T3 RNA polymerase. After fragmentation into an average size of 400 nucleotides by heating at 60°C in a buffer containing 40 mM NaHCO 3 , 60 mM Na 2 CO 3 , and 50 mM dithiothreitol, labeled RNAs were used as probes for hybridization. Hybridization and detection of hybridized probes were carried out as described by Hauptmann and Gerster (38) and Tautz and Pfeifle (37), respectively.
RNA Interference-RNA interference was performed as previously described (39). RNAs were synthesized using T3 or T7 RNA polymerase and linearized plasmids as templates. For the annealing reaction, equimolar quantities of sense and antisense RNAs were mixed to a final concentration of 5 M each and incubated at 67°C for 10 min, followed by incubation at 37°C for 30 min. RNA aliquots were ethanol-precipitated and dissolved in an injection buffer (40) to a final concentration of 5 M. The average injection volume was 85 pl, ranging from 65 to 110 pl, as determined by measuring the diameter of droplets injected into diphosphorylated mitogen-activated protein kinase; MAPK, mitogenactivated protein kinase; CME, central midline element; dsRNA, double-stranded RNA. halocarbon oil. Early embryos of 1-eve-1, the trachealess (trh)-lacZ line, were collected within 2 h of egg laying and injected with dHS6ST double-stranded RNA (dsRNA) or a buffer only. Injected embryos were allowed to develop at 25°C for 11 h to observe tracheal phenotypes or for 7 h to detect MAPK activation, and subjected to the staining procedure.
Immunohistochemistry-␤-Galactosidase expressed in 1-eve-1 enhancer trap embryos was detected using rabbit polyclonal antiserum (Cappel) at a 1:500 dilution in PBT (5% goat serum, 0.3% Triton X-100 in phosphate-buffered saline). Anti-dpMAPK (Sigma), which recognizes activated diphospho-MAPK was used at 1: 200. After repeated washes with PBT, the primary antibody was visualized using the Vectastain ABC kit (Vector Laboratories) and horseradish peroxidase immunochemistry following the protocol recommended by the manufactures. Embryos were mounted in 80% glycerol in phosphate-buffered saline and photographed using Olympus BX50 microscopy.

RESULTS
Cloning of dHS6ST cDNA and Genomic Organization of the dHS6ST-A genomic sequence, AE003728, which has been mapped to 92C on the third chromosome (BDGP), was identified as a clone bearing sequences with significant homology to mouse heparan sulfate 6-O-sulfotransferase-1 (mHS6ST-1) (35). We named this gene presumptive Drosophila heparan sulfate 6-Osulfotransferase (dHS6ST). To isolate a dHS6ST cDNA, a Drosophila larval cDNA library was screened with the PCR-amplified genomic fragment as a probe. From ϳ2 ϫ 10 5 plaques, four independent positive clones were obtained and sequenced.
Comparison of the dHS6ST cDNA and the AE003728 genomic sequences revealed that dHS6ST spans 79.0 kilobase pairs of the genome and contains six exons separated by five introns, with the second intron being more than 67 kilobase pairs in size (Fig. 1A). Around the genomic region corresponding to the 5Ј end of the cloned dHS6ST cDNA, we identified the sequence TCAGTT, which matches the Drosophila consensus sequence for an initiator "TCA(G/T)T(T/C)" (ϩ1 site is underlined, Fig. 1B) (41)(42)(43). Although a typical TATA box was not found in this region, the sequence GGACGTT is highly similar to the downstream promoter element (DPE), which is contained in many TATA-less promoters in Drosophila, at a typical position from the initiator sequence (ϩ28 to ϩ34 in Fig. 1B) (44). The occurrence of these sequences predicted the transcription start site as the A residue shown by ϩ1 in Fig. 1B. The cDNA clone that we have sequenced therefore appears to be nearly full-length. Northern blot analysis showed that dHS6ST encodes a single major transcript of ϳ3.1 kilobase pairs (data not shown).
BDGP suggested the existence of several genes around the dHS6ST locus (Fig. 1A). miranda, which encodes a cytoskeletal structural protein affecting the asymmetric division of neuro-blasts, was contained within in the fifth intron of dHS6ST (45,46). An EST clone, GH26202, derived from an adult head cDNA library, was mapped to the second intron of the dHS6ST gene. Furthermore, the 5Ј sequence of an EST clone, LD07688, was found to start at 499 base pairs upstream of the 5Ј end of dHS6ST in a reverse direction.
Primary Structure of dHS6ST-The complete cDNA and predicted amino acid sequences for dHS6ST are presented in Fig.  2A. The amino-terminal sequence contains three in-frame ATG codons. Out of three ATGs, the third one appears to be the most likely initiation codon, judging from the Drosophila consensus translation initiation sequence (47). The pattern of the hydropathy plot for the amino acid sequence started from the third ATG also matches those for mammalian homologues of dHS6ST (Fig. 2B) (35,48). The predicted protein was composed of 432 amino acid residues with two putative N-linked glycosylation sites. A hydropathy plot of dHS6ST showed the type II transmembrane structure of a Golgi resident protein with one prominent hydrophobic segment in the amino-terminal region that extends from amino acid residue 9 to 20.
A comparison of the primary structures of mammalian and Drosophila HS6STs revealed that the amino acid sequence of dHS6ST is 53%, 54%, 46%, and 53% identical to the sequences of human HS6ST (hHS6ST), mouse HS6ST-1, -2, and -3 (mHS6STs), respectively. Although relatively low levels of similarity were observed in both the amino-and carboxyl-terminal regions among HS6STs, the amino acid sequences of the central region are highly conserved (Fig. 3). Putative PAPS binding sites were present in these regions of all these proteins (49,50). Furthermore, all HS6STs bear two N-glycosylation sites at conserved positions. The predicted amino acid sequence of dHS6ST therefore shows all the structural features of HS6STs reported previously.
Characterization of Heparan Sulfate 6-O-Sulfotransferase Activity of dHS6ST-To determine whether the isolated dHS6ST cDNA encodes a protein with HS6ST activity, we examined the activity of dHS6ST expressed in tissue culture cells. dHS6ST cDNA was inserted in a mammalian expression vector, pFLAG-CMV2, and COS-7 cells were transfected with this construct or with pFLAG-CMV2 alone as a control. HS6ST activity in the cells transfected with dHS6ST was approximately twice as high as that with the control vector, whereas the dHS6ST cDNA insert did not increase the HS2ST activity (Table I). These results confirmed that recombinant dHS6ST produces HS6ST activity.
To further analyze this enzyme, we purified the epitope- tagged dHS6ST protein and examined the sulfotransferase activity using several heparin derivatives, heparan sulfates, and chondroitin as substrates (Table II). The dHS6ST was able to transfer sulfate to CDSNS-heparin, NS-heparosan, heparin, and heparan sulfate but was marginally active using CDSNAcheparin and chondroitin as substrates. The ratio of the activity of dHS6ST toward NS-heparosan to that toward CDSNS-heparin was 0.36. Since NS-heparosan contains only glucuronic acid and not its epimer iduronic acid, this result suggests a preference of dHS6ST for iduronic acid adjacent to the targeted N-sulfoglucosamine.
We analyzed the structures of the 35 S-labeled products obtained by incubation with purified dHS6ST and CDSNS-heparin or heparan sulfate from a mouse EHS tumor (Fig. 4). In both products, most of the radioactivity in the disaccharide fractions was at the position of ⌬UA-GlcNS6S. We also found a small peak at the position of ⌬UA2S-GlcNS6S from heparan sulfate. These results showed that dHS6ST transfer sulfate preferentially to position 6 of GlcNSO 3 residues. Since it has been reported that the substrate specificity of mHS6ST-2 is affected by the concentration of the substrates (35), we analyzed the dependence of the substrate preference of dHS6ST on substrate concentration. The ratio of the sulfotransferase activity toward NS-heparosan to that toward CDSNS-heparin was low at all concentrations tested (data not shown), suggest-ing the substrate specificity of dHS6ST is insensitive to the concentrations of acceptor substrate. Taken together, our findings indicated that dHS6ST shows HS6ST activity with preference for specific substrates.
Expression Patterns of dHS6ST mRNA in Embryos and Imaginal Discs-The expression patterns of dHS6ST mRNA were determined by in situ hybridization of whole-mount embryos and imaginal discs (Fig. 5). We observed high levels of accumulation of dHS6ST mRNA in syncytial blastoderm stage embryos indicating that dHS6ST is a maternally supplied product (Fig. 5A). During early gastrulation, zygotic expression of dHS6ST was detected in many tissues, including the invaginating mesoderm (Fig. 5B). In the stage 10 embryos, we detected highly specific expression of dHS6ST mRNA in tracheal precursor cells (Fig. 5C). dHS6ST expression in the invaginating tracheal precursor cells is maintained through stage 12 (Fig. 5, D and E). This spatial and temporal expression pattern of dHS6ST in tracheal precursor cells was strikingly similar to that of breathless (btl), a Drosophila FGF receptor (18). By stage 16, however, dHS6ST became uniformly expressed in various tissues (Fig. 5F). The restricted expression pattern of dHS6ST may reflect the requirements of the dHS6ST function in tracheal development.
In Drosophila, another FGF receptor homologue is encoded by heartless (htl) (17). htl is expressed in the invaginating FIG. 2. Structure of dHS6ST cDNA. A, complete nucleotide sequence of dHS6ST cDNA and the primary structure of dHS6ST. Three NH 2 -terminal ATG codons and the presumptive polyadenylation signal are underlined, and the deduced amino acid sequence from the third ATG codon is represented with single letters below the cDNA sequence. B, hydropathic analysis of the predicted dHS6ST protein. Hydrophobicity values were obtained according to the algorithm of Kyte and Doolitle with a window of 6 amino acid residues (56). Positive values represent increased hydrophobicity. mesoderm during embryogenesis, in the morphogenetic furrow of the eye disc and surrounding the outer proliferative center (OPC) of the larval CNS (51). The expression pattern of dHS6ST partially overlapped with that of htl in these tissues (Fig. 5, B, H, and I). In third instar larvae, dHS6ST mRNA was detected in actively dividing neuroblasts, including cells anterior to the morphogenetic furrow of the eye disc (Fig. 5H) and OPC of the CNS (Fig. 5I). This observation suggests the possible involvement of dHS6ST in htl-mediated neuronal develop-ment. We also observed high levels of expression of dHS6ST mRNA throughout the wing disc (Fig. 5G) and in the lobula of the CNS (Fig. 5I). The similarity of expression patterns between dHS6ST and Drosophila FGFRs raises the possibility that 6-O-sulfated heparan sulfate is required for all FGF signaling during Drosophila development.
dHS6ST Double-stranded RNA Inhibits FGF Signaling during Tracheal Formation-To elucidate the function of dHS6ST, we performed RNA interference (RNAi), a gene silencing strategy mediated by dsRNA (39). Approximately 75% of the control embryos, which had been injected with buffer only, hatched normally, while the remaining control embryos died owing to mechanical trauma (Table III). Most of the dHS6ST dsRNAinjected embryos (94%) died prior to the point immediately before hatching, although they survived to stage 17, the final stage of embryogenesis (Table III). This observation suggested that dHS6ST is essential for viability at late embryonic stages. Since whole mount in situ hybridization revealed expression of dHS6ST in the tracheal precursor cells, we tested whether disruption of dHS6ST activity affects tracheogenesis. The tracheal system of the Drosophila embryo form by a sequential series of branching steps that can be visualized by following the expression of 1-eve-1, an enhancer trap line for the trachealess (trh) gene (26,52). We injected dsRNA into the 1-eve-1 embryos at the syncytial blastoderm stage and collected embryos at stage 13, when primary branches continue to grow toward their targets. In 58% of dHS6ST dsRNA-injected embryos, tracheal branch formation was disordered while the levels of 1-eve-1 expression seemed to be comparable to the control embryos (Table III and Fig. 6). Twenty-five percent of these embryos exhibited stalled branches and disconnected dorsal trunks in all segments. The remaining 33% of dHS6ST dsRNA-injected embryos showed migration defects of the dorsal branch or lateral trunk in one to several segments. In buffer-injected embryos, 21% showed minor tracheal defects, and no control embryos showed severe tracheal disruption as described for dHS6ST dsRNA-injected animals (Table III). These results indicated that blocking dHS6ST activity disrupted the branching of tracheal precursor cells without affecting tracheal cell differentiation.

TABLE I
Overexpression of dHS6ST in COS-7 cells COS-7 cells were transfected with a vector alone or a plasmid containing dHS6ST, and the enzyme activities of HS6ST and HS2ST in the cell extract were determined as described under "Experimental Procedures." Values represent means Ϯ S.D. of triplicate independent experiments. The activities in the absence of exogenous CDSNS-heparin were usually less than 0.1 pmol/min/mg of protein.  anti-FLAG antibody affinity column chromatography Sulfotransferase activities were assayed using various glycosaminoglycans as acceptors. Sulfotransferase fractions were prepared from COS-7 cells transfected with pFLAG-CMV2-dHS6ST as described under "Experimental Procedures." Recombinant proteins were purified on an anti-FLAG antibody column. The values indicate the relative rate of the incorporation of 35 (53,54). When buffer was injected as a control, most of the embryos showed normal MAPK activation in tracheal precursor cells, while a modest reduction of MAPK activation was observed in 26% of embryos (Table III and Fig. 6C). In contrast, in 58% of dHS6ST dsRNAinjected embryos, anti-diphospho-MAPK immunoreactivity around tracheal pit was below the level of detection (Fig. 6D).
Our findings strongly suggest that 6-O-sulfated heparan sul- . dHS6ST is selectively expressed in tracheal precursor cells at stage 10 (C) and stage 11 (D and E). At stage 16, dHS6ST expression was ubiquitously detected again (F). G-I, expression patterns in wing disc, eye disc, and CNS from the third instar larvae. dHS6ST is detectable uniformly in the wing disc (G). Specific expression patterns were observed in cells anterior to the morphogenetic furrow of the eye disc (H) and OPC and lobula of the CNS (I). Arrowheads indicate tracheal placode (C), tracheal pit (D), and morphogenetic furrow (H). TPL, tracheal placode; TP, tracheal pit; MF, morphogenetic furrow.

TABLE III
Effects of dHS6ST dsRNA injection on lethality, tracheal morphology, and MAPK activation Embryos were injected with a buffer or dHS6ST dsRNA. Approximately 240 embryos were tested for each sample. "Lethality" indicates the percentage of embryos that could not hatch in total injected embryos. Tracheal phenotype and patterns of MAPK activation were observed in embryos that developed by stage 13 and stage 11, respectively. fate is required for normal Drosophila FGF signaling during tracheal development. DISCUSSION In this study, we cloned and characterized the dHS6ST gene. dHS6ST shows similarity in amino acid sequence with human and mouse HS6STs, and the substrate specificity of dHS6ST revealed that dHS6ST is comparable to mHS6ST-1; dHS6ST preferred the iduronosyl N-sulfoglucosamine unit as a substrate. Recent biochemical studies of heparan sulfate from Drosophila revealed disaccharides found in vertebrates; ⌬UA-Glc-NAc, ⌬UA-GlcNS, ⌬UA-GlcNAc6S, ⌬UA-GlcNS6S, ⌬UA2S-GlcNS, and ⌬UA2S-GlcNS6S disaccharides (3). Consistent with this biochemical evidence, our finding of an enzymatically active HS6ST in this organism suggests an equivalent function of 6-O-sulfated heparan sulfate in Drosophila to that of vertebrates.
We have shown here that dHS6ST is expressed in embryos and imaginal tissues. The expression of dHS6ST mRNA during development coincides with FGF-responsive cells known to synthesize Drosophila FGFRs, breathless (btl) and heartless (htl). Particularly in embryonic stages, the pattern of dHS6ST expression in the tracheal cells is analogous to those of btl. Activation of several genes required for tracheal tubular formation, including btl, is regulated by trachealess (trh), which encodes a transcription factor belonging to the basic helix-loophelix-PAS (Per-ARNT-sim) protein family (26,55). The heterodimers of Trachealess and Drosophila aryl hydrocarbon receptor nuclear translocator (dARNT) bind to the specific sequence, TACGTG, in the btl regulatory region and regulate its expression in tracheal cells as well as midline precursor cells (55). We identified a Trachealess/dARNT binding site in the first intron of dHS6ST, suggesting that the expression of dHS6ST might be regulated in the same fashion as that of btl by these transcription factors (Fig. 7).
RNAi experiments showed that dHS6ST is required for normal tracheal formation. The expressivity of the phenotypes observed in dHS6ST dsRNA-injected embryos was variable, from stalled primary branches in all segments to migration defects of the dorsal branch or lateral trunk at one segment. These phenotypes are reminiscent of the defects reported in null and hypomorphic mutants of btl and bnl (18,19). Loss of FGF-dependent MAPK activation in dHS6ST dsRNA-injected embryos strongly suggests that 6-O-sulfated heparan sulfate functions in FGF signaling, guiding tracheal cell migration. RNAi inhibition of dHS6ST function resulted in failure of hatching for 96% of injected embryos, while 58% of the embryos showed visible defects of the tracheal formation. This observation implies that dHS6ST may have functions in other developmental events essential for embryonic viability.
Previously, several biochemical studies of vertebrate cells/ tissues have suggested the possibility that 6-O-sulfated heparan sulfate controls FGF signaling (14,15,29). Pye et al. (15) showed that there is a direct correlation between the 6-Osulfate content of heparan sulfate oligosaccharides and their ability to promote bFGF activity. Lundin et al. (29) demonstrated that bFGF-induced angiogenesis in chick embryos was inhibited by 6-O-desulfated heparin, which binds to bFGF but fails to bind the receptor. Furthermore, the recent crystallographic analysis of two sets of FGF-FGFR-heparin complex showed that 6-O-sulfate groups interact with FGFR in the adjoining complex to promote FGFR dimerization (16). These findings suggest that 6-O-sulfation of heparan sulfate is required for a cellular response to FGF.
Our study provides direct in vivo evidence that 6-O-sulfation of heparan sulfate is critical for FGF signaling during development. Without 6-O-sulfation, heparan sulfate may be able to bind FGF but not mediate FGF signaling. Once HS6ST transfers sulfate groups along the chain, the dimerization of FGFR can occur, and cells acquire the ability to respond to FGF. Thus, spatially and temporally regulated expression of HS6ST is may serve as a key step for cellular responses to FGF. Since mammalian vascular development is also dependent on FGF (22,23), our findings support the possibility that HS6ST will also have a critical role in vascularization.