Heparan 2-O-sulfotransferase, hst-2, is essential for normal cell migration in Caenorhabditis elegans

Abstract

The importance of heparan sulfate proteoglycans has been highlighted by a number of human genetic disorders associated with mutations in genes encoding for heparan sulfate proteoglycan protein cores or biosynthetic enzymes required for heparan sulfate (HS) assembly. To study the functional role of HS in Caenorhabditis elegans development cosmid sequence C34F6.4 was identified as the C. elegans ortholog of vertebrate heparan 2-O-sulfotransferase (HS2ST) and the gene named hst-2. HS2ST activity is present in C. elegans and is completely absent in a deletion mutant of hst-2, ok595, and specifically reduced by hst-2 RNA interference. Expression of hst-2 in CHO cells deficient in HS2ST rescues enzyme activity and binding of FGF2 to cell surface HS. hst-2 expression is found in the hypodermis, muscle, distal tip cells (DTCs), and in neurons. A null mutation in hst-2 causes cell migration defects. This work demonstrates sulfotransferase activity in C. elegans and indicates that specific 2-O-sulfate modifications are critical for normal HS functions in controlling cell migration.

Heparan sulfate proteoglycans (HSPGs) have been implicated in a wide variety of biological processes, such as cell adhesion, wound healing, and host response to pathogens (1). HSPGs also modulate the activities of several growth factors, including members of the wingless/wnt, TGF-β and FGF families. The biosynthesis of heparan sulfate (HS) occurs in the Golgi complex, where a set of polymerases first synthesizes the polysaccharide backbone of N-acetylglucosamine and glucuronic acid residues. This reaction is followed by a complex pattern of modifications, especially by sulfotransferases, at selective positions. N-deacetylation and subsequent N-sulfation of glucosamine units are the first modification steps. Subsequent epimerisation of glucuronic acid to iduronic acid and sulfation at C2 hydroxyl group of hexuronic acids and sulfation at C6 or C3 of the glucosamine generally occur in these N-sulfated regions. Not every glucosamine unit will acquire a sulfate group and because the subsequent modifications occur predominantly in N-sulfated regions, HS has a domain structure of high sulfate density (NS domains), intermediate sulfate density (NS/NA domains), and low sulfate density (NA domains). HS synthesis is not template-driven, and enzymatic reactions do not proceed to completion, creating enormous structural complexity. Most of the molecular interactions of HSPGs are mediated by these highly divergent oligosaccharide sequences.

In vertebrates, HS biosynthetic activities involve in most cases multiple enzyme isoforms. Multiple EXT and EXT-like genes (HS copolymerases), four N-deacetylase/N-sulfotransferases, three glucosaminyl 6-O-sulfotransferases and six isoforms of glucosaminyl 3-O-sulfotransferase have been identified to date. Interestingly, only a single isoform of epimerase and uronyl 2-O-sulfotransferase have been identified to date. Many of these isoforms act on different precursor sequences, and their expression is temporally and spatially regulated (reviewed in ref. 2).

Recent genetic analysis of specific signaling pathways that control cell differentiation and tissue morphogenesis in mice, Drosophila, and C. elegans have implicated dedicated and highly specific roles for HS and HSPGs. Disruptions of N-deacetylase/N-sulfotransferase genes in mice (3–6) and Drosophila (sulfateless; ref. 7) cause several early developmental defects related to impaired wingless/Wg, FGF, and hedgehog (Hh) signaling. Mouse glypican mutants have skeletal abnormalities and overgrowth similar to that found in a rare human overgrowth syndrome Simpson-Golabi-Behmel caused by mutations in glypican 3 (8). Drosophila glypican (dally) mutants have defects in cell divisions in the central nervous system (9, 10). C. elegans mutants defective in rib-2, the worm ortholog of EXT1, exhibit developmental delay and egg-laying defects (11). Whereas the biological activities of C. elegans homologues of HS modification enzymes have not been demonstrated before, genes encoding for homologues of epimerase, hse-5, and 6-O-sulfotransferase (hst-6) have been shown to modulate activities of C. elegans homolog of Kallmann protein (anosmin), kal-1 (12). Mutations in these genes and in the homolog of 2-O-sulfotransferase, hst-2, show specific axon guidance defects (12, 13). Previous studies have shown that 2-O-sulfotransferase activity is required for normal development in mice and in Drosophila. Gene trap mutation of heparan 2-O-sulfotransferase (HS2ST) in mice causes renal agenesis, defects in the eye and skeleton, and neonatal lethality (14). A Drosophila sulfotransferase homolog called pipe is expressed in follicle cells and is required for dorso-ventral polarity (15, 16), but its relationship to HS formation is unclear.

We have cloned the HS2ST homologue in C. elegans-designated hst-2. We show that hst-2 encompasses the enzymatic activity responsible for specific modification of HS at 2-hydroxyl groups of hexuronic acid. We show that a deletion allele of hst-2, ok595, lacks 2-O-sulfotransferase activity, resulting in specific cell-migration defects of the gonad leader cells (DTCs) and of hermaphrodite-specific neurons (HSNs). We suggest that the specific HSN cell migration and axon outgrowth defects underlie the egg-laying defects seen in hst-2 mutants. hst-2, which is expressed in the hypodermis and in the DTCs, specifically modifies proteoglycans in the extracellular matrix and on the cell surface that are essential for cell migrations. We propose that heparan 2-O-sulfation displays dual roles in cell adhesion and guidance in C. elegans.

Materials and Methods

Materials. C. elegans were cultured and manipulated by using standard methods (17). All strains were maintained at 20°C. Some strains were obtained from the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health National Center for Research Resources. The stains used were as follows: wild-type N2 var. Bristol, RB800 hst-2 (ok595). RB800 was outcrossed six times with respect to the X chromosome. All phenotypic analyses of ok595 were performed by using homozygous mutants. Two separate integrated GFP marker strains for serotonergic neurons were used, mgIs71 [tph-1::gfp, rol-6(d)] (18) and dzIs13 [tph-1::gfp] (19). jtEx10 contains extrachromosomal array of translational HST-2::GFP fusion and ttx-3::rfp (as injection marker) in ok595;zdIs13.

hst-2 Cloning and Expression Constructs. Standard molecular biological techniques were used (20). For additional information, see Supporting Methods, which is published as supporting information on the PNAS web site. The hst-2::gfp transcriptional reporter contains 3.7 kb of sequence upstream from the predicted ATG site cloned into SphI/SmaI sites of pPD95.75. A shorter version of hst-2::gfp reporter that contains 300 bp of sequence upstream from the predicted ATG site was generated by using PCR fusion (21). The short reporter contains a sequence up to the nearest gene predicted by the Genome Consortium. GFP constructs were injected at 30 μg/ml. pRF4 [rol-6(d)] was used as an injection marker at 100 μg/ml. Both promoter constructs showed the same expression pattern. hst-2 rescue construct contains a full-length hst-2 genomic clone driven by 3.7 kb of sequence upstream from hst-2 ATG cloned into SphI/SmaI sites of pPD95.75. This construct was injected at 10 μg/ml with ttx-3::rfp transcriptional reporter as an injection marker. Expression of GFP of the hst-2 rescue construct was beyond detection levels.

RNAi Interference (RNAi) of hst-2. Double-stranded RNA-mediated interference was performed by feeding double-stranded RNA as described in ref. 22. For modifications, see Supporting Methods.

Brood Size Counts. Single L4 hermaphrodites were plated on individual plates and left to mature and lay eggs. The mothers were transferred to new individual plates every 24 h until they did not lay any more eggs. The number of progeny for each mother was counted at larval stage L4 or later.

Cell Culture. CHO-K1 cells were obtained from American Type Culture Collection (CCL-61, Rockville, MD). The 2-O-sulfotransferase-deficient CHO cell line, pgsF-17, was characterized in ref. 23. Cells were grown under an atmosphere of 5% CO₂ in air and 10% relative humidity in Ham's F-12 growth medium (Life Technologies, Carlsbad, CA), supplemented with 10% (vol/vol) FBS (HyClone)/100 μg/ml streptomycin sulfate/100 units penicillin G.

Sulfotransferase Assays. pgsF-17 (23) cells were transiently transfected with pcDNA3.1-hst-2 by using Lipofectamine (Life Technologies). After 48 h of incubation at 30°C, the cells were scraped from the culture dish in 50 μl of a solution containing 0.25 M sucrose, 20 mM Tris, pH 7.5, with protease inhibitors (Sigma), and cell lysates were kept frozen at -80°C. Briefly, the 2-O-sulfotransferase assay (25 μl) contained 50 mM Mes (pH 6.5), 1% TX-100, 10 mM MgCl₂, 10 mM MnCl₂, 5 mM CaCl₂, 87.5 μM NaF, ≈4,000 cpm/nmol [³⁵S]3′-phosphoadenosine 5′-phosphosulfate (PAPS), 50 μg of N,O-desulfated re-N-sulfated heparin as an acceptor substrate and 25 μg of cell lysate proteins as the enzyme source. The reaction was incubated for 4 h at 25°C with occasional mixing and stopped by heat inactivation for 1 min at 95°C. Standard chondroitin sulfate was added (2 mg) as a carrier and the volume was increased to 500 μl with water. The sample was applied to 0.25-ml column of DEAE-Sephacel in disposable polypropylene tips as described in ref. 24. The column was washed with 15 ml of 0.25 M NaCl, 20 mM sodium acetate (pH 6.0), and eluted with 2.5 ml of a 1 M NaCl/20 mM sodium acetate, pH 6.0. An aliquot (1 ml) was counted by liquid scintillation (Ultima Gold XR, Packard). Penicillium chrysogenum adenosine 5′-phosphosulfate kinase was a kind gift from Irwin Segel, University of California, Davis, and was used to prepare [³⁵S]PAPS as described in ref. 25.

Sulfotransferase assays of worm lysates were performed as described in refs. 26 and 27 with modifications that are described in Supporting Methods.

Flow Cytometry. pgsF-17 and transiently transfected cells were screened by flow cytometry for binding of biotinylated FGF2 to HS as described in ref. 28.

Analysis of Egg Laying. Egg-laying rates were measured by counting the number of eggs laid after 1 h on nematode growth medium-agar plates seeded with E. coli OP50 as a food source in the presence or absence of drugs. Serotonin was applied to the plates at a final concentration of 7.5 mM, and imipramine was applied at a final concentration of 2.5 mM.

Microscopy. Microscopy of living animals was performed by mounting the animals on a 4% agarose pad in a drop of M9 containing 30 mM NaN₃ as an anesthetic. Fluorescent and differential interference contrast images were obtained by using a Nikon Eclipse TC300 microscope equipped with an ORCA-ER digital camera (Hamamatsu, Middlesex, NJ) and openlap software. Confocal images were obtained by using a Bio-Rad Radiance 2000MP with a Nikon Eclipse E600 microscope, Nikon Plan Apo 60× A/1.4 Oil IR DIC objective, and lasersharp2000 software and with Zeiss LSM 510 Meta with Zeiss C-Apochromat 40×/1.2W and Fluar 10×/0.5 objectives. Images were processed by using photoshop 7.0 software (Adobe Systems, San Jose, CA).

Results and Discussion

Identification and Cloning of C. elegans hst-2. A blast search of the C. elegans database identified a single sequence, C34F6.4, as an ortholog of the vertebrate and Drosophila HS2STs. We cloned and sequenced the corresponding cDNA and genomic DNA and named the gene hst-2 (for heparan sulfotransferase 2; ref. 29). The hst-2 cDNA has a 972-bp ORF as predicted in the database. The genomic structure of hst-2 consists of eight exons (Fig. 6, which is published as supporting information on the PNAS web site).

The cloned hst-2 cDNA encodes for a 324-aa protein with 42% identity (62% similarity) to vertebrate HS2ST (Fig. 6). The encoded HST-2 protein contains the two 5′ phosphoadenosine 3′ phosphosulfate binding sites conserved in all sulfotransferases (30). Northern analysis identified a single mRNA for hst-2 (Fig. 6D), suggesting that there is only one splice form of hst-2 in C. elegans. The encoded HST-2 protein is predicted to be a type II transmembrane protein as would be expected for a Golgi enzyme. This cDNA confers HS2ST activity, providing further evidence that it represents the hst-2 cDNA (see below).

C. elegans Heparan 2-O-Sulfotransferase, hst-2, Rescues Enzyme Activity in 2-O-Sulfotransferase-Deficient CHO Cells. The ability of hst-2 to act as a HS2ST was analyzed by using pgsF-17 cells, a mutant CHO cell line deficient in HS2ST activity (23). Lack of heparan 2-O-sulfation in the mutant abolished binding of FGF2 to cell surface HS, as measured by flow cytometry with biotinylated FGF2 (Fig. 1B). Transient transfection of the mutant with full-length hst-2 cDNA restored expression of FGF2 binding sites at the cell surface almost to the levels observed in wild-type cells (Fig. 1 A and C). To determine directly whether hst-2 can catalyze the transfer of sulfate group to the 2-O position of appropriate substrates, in vitro extracts were prepared from transfected pgsF-17 cells and incubated with N- and O-desulfated re-N-sulfated heparin as the acceptor substrate. hst-2-expressing cells showed an enhanced activity (69 ± 1.6 pmol/mg per min) when compared with parental pgsF-17 cells (19 ± 2.2 pmol/mg per min), indicating that hst-2 encodes for a protein with authentic HS2ST activity. The high background level of transferase activity in the parental mutant cells reflects N-deacetylase/N-sulfotransferase and 6-O-sulfotransferase activity in the extracts (23). The highly conserved motifs diagnostic of the PAPS binding site, type II transmembrane orientation, complementation of a CHO mutant defective in HS 2-O-sulfotransferase, and demonstration of transfer of sulfate from PAPS to the C-2 hydroxyl groups of iduronic acid in exogenous substrates strongly support the view that hst-2 encodes the HS 2-O-sulfotransferase of C. elegans.

Fig. 1.

hst-2 rescues FGF2 binding to cell surface in HS2ST-deficient CHO cells. A mutant CHO cell line, pgsF-17, deficient in HS2ST activity, was transiently transfected with hst-2. FGF2 binding to cell-surface HS as assayed by flow cytometry (outlined curves). (A) Wild-type CHO cells. (B) Mutant pgsF-17 cells. (C) Mutant pgsF-17 transfected with hst-2. The filled curve represents cell sorts done on cells without added biotinylated FGF2. The y axis indicates the relative frequency of cells exhibited by a given level of binding.

2-O-Sulfotransferase Activity Is Absent in an hst-2 Deletion Allele, ok595. A deletion allele of hst-2, ok595, was obtained from the C. elegans Knockout Consortium. This allele contains a 1346-bp deletion in the hst-2 genomic region, causing a frame shift in the ORF and a premature stop in the protein translation. Furthermore, the ok595 deletion abolishes the second part of the conserved sulfotransferase domain, which interacts with the 3′ phosphate group of the sulfate donor (PAPS; 3′ PBS; ref. 30) and thus is expected to lack sulfotransferase activity. Further evidence for the identity of hst-2 was obtained by direct assay of endogenous HS2ST activity in ok595 and wild-type C. elegans (N2) in the presence or absence of double-stranded RNA to silence the hst-2 gene. Total protein lysates from C. elegans were incubated with [³⁵S]PAPS as a sulfate donor and exogenous sugar acceptors as substrates. Sulfate incorporation was monitored and the acceptor substrates were analyzed for their disaccharide composition (Fig. 2). Polysaccharide substrate incubated with ok595 protein lysate displayed a 48% decrease in overall sulfate incorporation as compared with wild-type C. elegans (n = 2; data not shown). Analysis of the disaccharide composition revealed a complete loss of 2-O-sulfotransferase activity in ok595 as observed by the absence of disaccharides representing uronic acid 2-O-sulfate-N-sulfated glucosamine units (ΔUA2S-GlcNS; Fig. 2 A; 904 and 85,899 dpm/mg of protein for ok595 and N2, respectively). hst-2 RNAi caused an ≈35% decrease in overall sulfate incorporation in exogenous substrates (control, n = 4; hst-2 RNAi, n = 8; P < 0.01). Analysis of the resulting disaccharides showed that hst-2 RNAi caused an ≈50% reduction in ΔUA2S-GlcNS (Fig. 2B; P < 0.01). The residual 2-O-sulfotransferase activity is likely attributed to incomplete penetrance of RNAi.

Fig. 2.

Loss of hst-2 activity in the deletion mutant ok595 and by hst-2 RNAi. C. elegans protein lysates were incubated with acceptor sugar (N,O-desulfated, re-N-sulfated heparin) and [³⁵S]PAPS as a sulfate donor. The labeled heparin polysaccharide was enzymatically digested, and the resulting disaccharides were separated by HPLC by using a strong-anion-exchange column (44). (A) Disaccharide analysis of ³⁵S-labeled reaction products. 2-O-sulfated disaccharides (ΔUA2S-GlcNS) are absent in the ok595 sample (open bars) (n = 2); wild-type (N2) (filled bars) (n = 2). (B) hst-2 RNAi also causes a reduction in 2-O-sulfation. Open bars, hst-2 RNAi (n = 7); filled bars, control RNAi (n = 3).

The relative amount of disaccharide representing ΔUA-GlcNS6S was not significantly affected, suggesting no changes in 6-O-sulfotransferase activity. Treatment of the trisulfated disaccharide (ΔUA2S-GlcNS6S) (Fig. 2 A) present in ok595 samples with iduronate-2-sulfatase released no free [³⁵S]O₄ and resulted in a shift of the ³⁵S-detectable disaccharide to the standard position of ΔUA-GlcNS6S (data not shown). This result indicates that the trisulfated disaccharide present in ok595 samples is generated by 6-O-sulfotransferase activity on a small amount (<10%) of residual UA2S-GlcNS sequences in the heparan substrate. These results strongly suggest that hst-2 is responsible for HS2ST activity in C. elegans.

hst-2 Is Expressed in the Hypodermis and in the Nervous System. hst-2 expression was analyzed by using a transcriptional GFP reporter gene driven by sequences upstream of the first exon of hst-2 (Fig. 3). The onset of hst-2 expression coincides with the start of morphogenesis in mid-embryonic (comma) stages where hst-2::GFP is first detected in developing pharyngeal cells (Fig. 3A). hst-2::GFP expression in the pharynx persists through larval development and remains in adults (Fig. 3 B and C). In larvae, hst-2 is also expressed in the hypodermis (epidermis) and in neurons in both the dorsal and ventral nerve cords (Fig. 3B). In adults, hst-2::GFP is widely expressed in the pharynx, muscle and in several neurons in the nerve cords (Fig. 3C).

Fig. 3.

Expression pattern of hst-2::GFP transcriptional reporter. (A) hst-2::GFP is expressed in the developing pharynx (ph) in a 1.5-fold embryo. (B) hst-2::GFP expression in L2 larva. GFP is detected in the ph, in neurons in the head (arrows), and in ventral and dorsal nerve cords (arrowheads). (C) In adults, hst-2::GFP is widely expressed in the pharynx, in neurons of the nerve cords (nc), and in the muscle (m). Fluorescence (D, F, and H) and differential interference contrast (E, G, and I) images show developing vulva and gonad. (D and E) In L3 larva, hst-2::GFP is strongly expressed in the four descendants of the P6.p vulval precursor cell. Anchor cell (AC), which does not express hst-2::GFP, induces P5.p, P6.p and P7.p to adopt vulval fate. (F and G) hst-2::GFP expression persists in vulval cells in L4 larva. (H and I) hst-2::GFP expression in DTC during gonad development in L3 larva. (A) Anterior is top, dorsal right. (B–I) Anterior left, ventral down. [Scale bars: 10 μm(A), 200 μm(B), 100 μm(C), and 10 μm (D–I).]

During the development of the hermaphrodite vulva, hst-2::GFP is expressed in the vulval precursor cells. At larval stage L3, hst-2::GFP is seen in the four descendants of the P6.p vulval precursor cell (Fig. 3 D and E). The expression of hst-2::GFP in vulval hypodermis persists through larval stage L4 (Fig. 3 F and G) to adults. During the hermaphrodite gonad morphogenesis, the gonad leader cells or DTCs also express hst-2::GFP (Fig. 3 H and I).

Gonad Leader Cell Migration Defects in hst-2 Mutants. In mutants defective in any of the eight C. elegans squashed vulva (sqv) genes, the vulval extracellular spaces fail to expand during vulval morphogenesis. Seven of the eight sqv genes encode for enzymes involved in the biosynthesis of the common tetrasaccharide linkage of HS/CS and in the synthesis and transport of nucleotide sugars (31–35). One of the genes, sqv-5, encodes the chondroitin polymerase, indicating a role for chondroitin in epithelial invagination (36, 37). Although hst-2::GFP is expressed in the vulval precursor cells, we could not detect any defects in vulval morphogenesis in homozygous hst-2 (ok595) animals, suggesting that 2-O-sulfation is dispensable for vulval development. However, homozygous hst-2 mutants displayed defects in hermaphrodite gonad development. The gonad leader cell DTC migrated abnormally, resulting in mispositioned gonad arms (Fig. 4). The C. elegans hermaphrodite gonad develops based on patterned cell migrations of the DTC during the third and fourth larval stages. Beginning at the ventral midbody, the anterior and posterior DTCs follow mirror-image patterns of migrations in three stages. In the first stage the DTCs migrate toward the head or tail along the ventral muscles, followed by ventral to dorsal migration along the lateral hypodermis and finally migrating back toward the midbody along the dorsal muscles (38). The incidence of the DTC migration defects was 33% for the anterior DTC (27 of 83 DTCs scored) and 23% for the posterior DTC (22 of 96 DTCs). hst-2 affected mainly the third DTC migration stage (25 of 83 anterior DTCs and 19 of 96 DTCs scored), and the gonad arms frequently migrated back to the ventral side instead of following the dorsal muscle band (Fig. 4C). In only a very few cases (2% of all DTCs scored), the DTC failed to migrate circumferentially from the ventral muscle band to the dorsal side. A similar phenotype was observed with hst-2 RNAi (Fig. 4D).

Fig. 4.

Loss of HST-2 activity affects DTC migration (A) The C. elegans hermaphrodite gonad develops based on patterned cell migrations of the DTC. The anterior and posterior DTCs follow mirror-image patterns of migrations in three sequential phases along the ventral bands of body wall muscle from ventral to dorsal along the hypodermis and toward the midbody along the dorsal muscle bands. (B) Tip of the wild-type gonad arm. The pathway of the gonad arm is marked with a dashed line. Genetic deletion in hst-2 (ok595) (C) or loss of hst-2 by RNAi (D) leads to DTCs migrating back to the ventral side from the dorsal muscle band. Anterior is left, ventral down. v, position of the vulva. (Scale bar, 20 μm.)

As hst-2 is expressed in the DTCs during gonad development, HS on the surface of the DTCs may mediate interactions with guidance molecules in the extracellular matrix; loss of 2-O-sulfation may compromise these interactions. Furthermore as hst-2 is also expressed in the muscle, HS on the muscle surface may provide cues for migrating DTCs. The migratory routing of the anterior DTCs is more frequently affected than the posterior DTCs in hst-2 mutants. Several mutants have been identified where the migration pathways of the anterior and posterior DTCs are differentially affected, suggesting that the migratory pathways of the DTCs are differentially regulated (39, 40). The molecular nature of the growth factor and cell adhesion molecules affected in the hst-2 mutants warrants further work, but recent studies suggest several possible candidate genes. The C. elegans perlecan homologue, unc-52, which is expressed in the muscle (41), does not itself cause DTC migration defects but does in combination with the UNC-5/netrin receptor and with growth factor mutations that cause aberrant DTC migrations (39). Unc-129/TGF-β which is normally expressed in dorsal muscles, does not normally affect DTC migrations, but ectopic expression of unc-129 in ventral body wall muscles causes DTC migration defects (42). Furthermore, unc-129 partially suppresses the enhancement by class I unc-52 alleles of unc-5(e152) DTC migration defects, suggesting genetic interactions of unc-52 with growth factors (39).

Null Mutants of hst-2 Display Egg-Laying Defects. Unlike mutants in the enzymes involved in heparan sulfate linkage region biosynthesis or in the HS chain elongation (11, 31–35), homozygous hst-2 (ok595) animals are viable and fertile although they have significantly reduced brood size (153 ± 38; n = 31) as compared with wild-type animals (203 ± 29; n = 20; P < 0.01). The ok595 deletion leads to a variable degree of egg-laying defects. Embryos frequently develop to the 3-fold stage (>500 min after first cleavage) before being laid, whereas wild-type C. elegans normally lay eggs during gastrulation (120–180 min after first cleavage).

Migration of HSNs Is Defective in hst-2 Mutants. Egg laying is initiated when two serotonergic motor neurons, the HSNs (HSNL/R), stimulate the contraction of vulval muscles. The HSNs are born during mid-embryonic stages in the posterior part of the embryo, and during embryonic development they migrate to the mid region of the animal just posterior to the vulva (43). The HSNs send axons anteriorily along the ventral midline to the nerve ring in the pharynx. A branch of these axons innervates the vulval muscles. Using cell-specific GFP marker strains to visualize the hermaphrodite-specific neurons, we show that in homozygous hst-2 mutants the HSNs fail to migrate to their wild-type positions. HSN migration and axon outgrowth were analyzed by using tryptophan hydroxylase GFP marker strains, mgIs71, and zdIs13 that express GFP in serotonergic neurons, including the HSNs (Fig. 5; ref. 18 and 19). HSNs migrate to the immediate vicinity of the vulva 96% (48 of 50) of the time in mgIs71 and 100% of the time in zdIs13 (n = 50) (Fig. 5 A and D and Fig. 7, which is published as supporting information on the PNAS web site). In homozygous hst-2 (ok595) mutants, 26% of the HSN cell bodies (55 of 210 HSNs scored) failed to migrate to their correct position (Fig. 5 B and D). We measured the percentage of HSN migration failure by using wild-type position as a reference for 100% migration (Fig. 5A). Of 210 HSNs scored, 2 (1%) HSNs failed to migrate at all and the HSN cell bodies remained in the tail, 6 (3%) of the HSNs migrated only 25%, 16 (8%) migrated 50%, and 31 (15%) migrated 75% as compared with the wild-type (Fig. 5D). This migration defect is caused by the loss of hst-2 activity in the ok595 mutant as transgenic expression of hst-2 in ok595;zdIs13 completely rescued HSN migration (Fig. 5 C and D). Two independent transgenic lines displayed the wild-type HSN migration pattern (70 of 73 and 29 of 30 HSNs scored; Fig. 5D). The marker strain zdIs13 was chosen for rescue studies because it does not contain rol-6. ok595;zdIs13 displayed similar HSN migration defects as compared with ok595;mgIs71 (30% of the 73 HSNs scored failed to migrate to wild-type position, see Fig. 7).

Fig. 5.

Loss of HST-2 activity causes defects in HSN migration. The serotonergic HSNs innervate vulval muscles and are required for egg laying. The HSNs are born in the tail and migrate to the midbody, posterior to the vulva, during development. (A) Visualization of HSNs in adult hermaphrodites by using HSN-specific GFP expression in the mgIs71 strain, which carries an integrated tph-1::gfp array. (B)In hst-2 (ok595) mutants the HSNs fail to migrate from the posterior body to their correct location. In some cases axon branching of HSN is also defective (arrowhead). (C) Transgenic expression of hst-2 in ok595 rescues HSN migration. Anterior is left, ventral view (A), and ventrolateral view (B and C). v, position of the vulva. (D) Migration of HSNs (asterisks) from tail to the vulval vicinity (v). Numbers correspond to HSNs scored and distance they migrated from the tail.

HSN axon guidance was also affected in hst-2 (o595) animals, and axon branches frequently projected posteriorly as well as anteriorly from the HSN cell bodies (Fig. 5B). In addition to HSNs, cells that undergo long-range migrations during C. elegans development include the anterior mechanosensory neurons (ALM) and the canal-associated neurons (CAN). We could not detect abnormalities in ALM or CAN positions, suggesting that the migration defects are specific to the HSN.

Serotonin Rescues hst-2 Egg-Laying Defects. The egg-laying-defective (egl) phenotype can arise from defects in the migration and function of hermaphrodite-specific neurons that control egg laying or of sex myoblasts that generate the vulval muscles. The neuronal defects can be studied by analyzing responses to pharmacological agents that have specific effects on neurons (serotonin and imipramine). The HSNs innervate the vulval muscles and release serotonin, which initiates egg laying (43). Imipramine blocks presynaptic uptake of serotonin and, thereby, potentiates endogenous serotonin.

Availability of food influences serotonin levels and hence has an effect on egg laying. We tested the response of hst-2 (ok595) animals to both serotonin and imipramine in the presence of food. Without any drug treatment there was a clear reduction in the number of eggs laid by homozygous hst-2 animals as compared with the wild-type. hst-2 (ok595) animals laid on average 4.8 eggs per worm per h (±3.2, n = 39) as compared with wild-type worms 7.5 eggs per worm per h (±3.6, n = 37, P < 0.01). Treatment of hst-2 (ok595) animals with serotonin completely rescued egg laying to wild-type levels (9.4 ± 3.5, n = 47, for hst-2 and 8.9 ± 4.1, n = 40, for wild-type). Imipramine did not have a significant effect on egg laying of hst-2 animals (hst-2, 5.6 ± 3.1, n = 34; wild-type, 7.6 ± 3.7, n = 32, P < 0.05). The egg-laying defect of hst-2 is rescued by the addition of exogenous serotonin, suggesting that the failure of the HSN to migrate to their correct position in hst-2 mutants and, thereby, to innervate the vulval muscles, is the underlying mechanism to cause the egg-laying defect seen in hst-2 (ok595) animals. This finding is further supported by the finding that transgenic expression of hst-2 in ok595 mutants, which rescues HSN migration, also rescues egg laying almost to wild-type levels (6.8 ± 1.8 eggs per worm per h, n = 20).

Conclusions

Unlike vertebrates, C. elegans generally contains one isoform of each HS biosynthetic enzyme. Mutations in C. elegans genes encoding for the enzymes required for the biosynthesis of the HS/CS backbone are homozygous lethal. Deletion mutants of C. elegans heparan copolymerase rib-2, have reduced levels of HS, and exhibit egg-laying and elongation defects in F2 progeny and embryonic lethality in F3 progeny (11). Mutations in C. elegans heparan N-deacetylase/N-sulfotransferase–hst-1 or 3-O-sulfotransferase (hst-3) have not been described to date. Mutations in HS2ST as described in this and another recent study (13), and mutations in 6-O-sulfotransferase (hst-6) and C-5 epimerase (hse-5), are dispensable for viability, whereas they cause very specific cell migration and axon outgrowth defects (refs. 12 and 13; T.K. and J.E.T., unpublished results). The finding that the rib-2 mutation is homozygous lethal suggests that HS chains are critical for organogenesis, whereas specific HS sulfation patterns are required for fine tuning of developmental processes.

In summary, we have shown that C. elegans hst-2 encodes for enzyme activity responsible for substituting uronic acids of HS with sulfate groups. This report demonstrates sulfotransferase activity in C. elegans. Our results show that this specific substitution, although dispensable for viability, is critical for specific cell and axon migrations. Many growth factors, extracellular matrix components, and cell surface molecules controlling cell migrations and organogenesis that are known to interact with HSPG are also present in C. elegans. We anticipate that elucidation of the genetic and signaling networks influenced by 2-O-sulfation in C. elegans will reveal general principles applicable to higher organisms.