University of Huddersfield Repository Combinatorial Roles of Heparan Sulfate Proteoglycans and Heparan Sulfates in Caenorhabditis elegans Neural Development

Heparan sulfate proteoglycans (HSPGs) play critical roles in the development and adult physiology of all metazoan organisms. Most of the known molecular interactions of HSPGs are attributed to the structurally highly complex heparan sulfate (HS) glycans. However, whether a specific HSPG (such as syndecan) contains HS modifications that differ from another HSPG (such as glypican) has remained largely unresolved. Here, a neural model in C. elegans is used to demonstrate for the first time the relationship between specific HSPGs and HS modifications in a defined biological process in vivo. HSPGs are critical for the migration of hermaphrodite specific neurons (HSNs) as genetic elimination of multiple HSPGs leads to 80% defect of HSN migration. The effects of genetic elimination of HSPGs are additive, suggesting that multiple HSPGs, present in the migrating neuron and in the matrix, act in parallel to support neuron migration. Genetic analyses suggest that syndecan / sdn-1 and HS 6-O-sulfotransferase , hst-6 , function in a linear signaling pathway and glypican / lon-2 and HS 2-O- sulfotransferase , hst-2 , function together in a pathway that is parallel to sdn-1 and hst-6 . These results suggest core protein specific HS modifications that are critical for HSN migration. In C. elegans , the core protein specificity of distinct HS modifications may be in part regulated at the level of tissue specific expression of genes encoding for HSPGs and HS modifying enzymes. Genetic analysis reveals that there is a delicate balance of HS modifications and eliminating one HS modifying enzyme in a compromised genetic background leads to significant changes in the overall phenotype. These findings are of importance with the view of HS as a critical regulator of cell signaling in normal development and disease.


Introduction
Heparan sulfate (HS) and HS proteoglycans (HSPGs) critically regulate a range of biological processes in metazoan organisms from early events in gastrulation to adult physiology, disease and host response to pathogens (for reviews see [1,2,3]. The involvement of HS/HSPGs in such a broad range of biological activities is due to their ability to interact with a vast repertoire of proteins such as growth factors, morphogens, enzymes, extracellular matrix and plasma proteins. The molecular interactions of HSPGs occur predominantly via the negatively charged HS glycans attached to the proteoglycan protein core. The biosynthesis of HS occurs in the Golgi complex, where a set of polymerases first synthesize the polysaccharide backbone consisting of N-acetylglucosamine (GlcNAc) and glucuronic acid (GluA) residues. The linear HS backbone undergoes a complex pattern of modifications at selective positions. The first modification step is N-deacetylation and subsequent N-sulfation of some glucosamine units. Subsequent epimerization of some GluA units to iduronic acid (IdoA) and sulfation at C2 hydroxyl group of GluA/IdoA and sulfation at C6 or C3 of the glucosamine are actions of a glucuronyl C5-epimerase and heparan 2-, 6-and 3-O-sulfotransferases, respectively. These modifications generally occur in the Nsulfated regions creating domains of high sulfate density (NS-domains), intermediate sulfate density (NS/NA-domains) and low sulfate density (NA-domains). HS biosynthesis is not templatedriven and enzymatic reactions do not modify every sugar unit, resulting in huge structural complexity in one of the most heterogeneous molecules in biology.
Biochemical analyses have revealed developmental and disease related changes to HS composition. Gene knock-out studies have identified key signaling pathways and developmental processes that require the activities of specific HS biosynthetic enzymes. For example, genetic elimination of mouse Hs2st leads to complete lack of kidneys and neonatal lethality [11]. Targeted disruption of C5-epimerase in mice leads to renal agensis, lung defects and skeletal malformation [12]. Targeted disruption of HS biosynthetic enzymes is expected to affect all different HSPGs expressed in the targeted tissue or developmental stage. Similarly, most biochemical analyses to structural changes in HS have used whole animal or tissue samples containing a pool of HSPGs. There is limited data on the relationship between HS structures and specific core proteins and whether the core protein influences structural specificity of the HS. Mouse syndecan and glypican HSPGs ectopically expressed in immortalized cell lines are decorated with HS of similar structural composition [13] suggesting that at least in vitro, the core protein context is not critical for HS structure determination. Syndecan and glypican from rat embryonic fibroblasts carry HS chains of different molecular mass, but similar charge and structural composition, yet display differential binding to fibronectin, suggesting that the core protein may determine ligand specificity [14]. However, syndecan-3 purified from perinatal rat forebrain is decorated with HS that contains a higher degree of N-sulfation and IdoA units as compared to average tissue HS [15] suggesting that the structural regulation of specific HS is more complex in vivo.
The C. elegans hermaphrodite specific neurons (HSNs) are a pair of serotonergic neurons that innervate the vulval muscles and are necessary and sufficient to initiate egg laying [16]. The HSNs are born in the tail of the embryo and during embryonic development the HSNs migrate to mid body, to the immediate proximity of the developing vulva. Defects in HSN development often result in egg-laying defects, which can be readily observed as accumulation of embryos inside the mother.
This study sought to address the relation between a specific HSPG core protein and distinct HS motifs using a neural development model in C. elegans in vivo. C. elegans has in most cases a single orthologue of vertebrate genes encoding for HSPG core proteins and HS biosynthetic enzymes facilitating analysis without genetic redundancy [17,18,19,20,21,22,23,24]. Developmental genetic approach was used to assess the relation between specific HS structures on different HSPG core proteins for the migration of the HSNs. The HSNs undergo a significant developmental migration in relation to the C. elegans body plan, and were thus chosen as a model to assess HSPGs and HS in neural development. Multiple HSPG dependent neuron guidance pathways are required for HSN migration in parallel. These parallel pathways rely on different HS modifications. Genetic analysis also suggests an interdependence of HS biosynthetic enzymes during HS biosynthesis.

Reporter constructs
Standard molecular biological techniques were used. The sdn-1::gfp transcriptional reporter contains 1.5 kb of sequence upstream from predicted ATG site cloned into SphI/SmaI sites of pPD95.75. The hst-3.1::gfp and hst-3.2::gfp reporters contain a 2 kb of sequence upstream from corresponding predicted ATG site cloned into SphI/XbaI and SphI/BamHI restriction sites of pPD95.75, respectively. DNA constructs were injected at 30 mg/ ml. pttx-3::mCherry or ptph-1::mCherry were used as injection markers at 50 mg/ml and pBluescript was used as a filler DNA.

Neuron scoring
HSN neuron migration was scored using cell specific fluorescent reporters that are expressed in the HSNs. Each C. elegans has two HSNs but sometimes it was not possible to see both HSNs due to the position of the animal. Neuron scores are thus presented as number of total neurons scored, rather than number of animals scored, unless otherwise stated. Statistical significance was tested using either Student's t-test or Fisher's test.

Microscopy
Microscopy of living C. elegans was performed by mounting the animals on a 4% agarose pad in a drop of M9 buffer containing 30 mM NaN 3 as an anesthetic. HSN migration was scored under fluorescent microscopy using GFP expression in all serotonergic neurons including the HSNs. Compound fluorescent images were acquired using Zeiss AxioCam MRm camera mounted on Zeiss Axioskop2 microscope equipped with x20, x40 and x63 optics. Images were captured using Axiovision. Confocal images were acquired using Leica AOBS SP2 system and images were captured using Leica software. Microscopy images were further cropped and scaled using Adobe Photoshop CS3 (Adobe Systems, San Jose, CA).

Role of heparan sulfate biosynthetic enzymes for HSN migration
zh20 is a null allele of sdn-1/syndecan [26], whereas ok449 is an in-frame deletion of the second and third exons [20]. This deletion abolishes two of the three consensus heparan sulfate attachment sites (Ser-Gly) of the SDN-1 protein. Truncated SDN-1 protein is present in ok449 mutants, but it does not contain any HS chains as detected by monoclonal antibodies against HS ( Figure S1 and Methods S1). zh20 and ok449 alleles exhibit identical HSN migration defects, suggesting that the role of SDN-1 in HSN migration is predominantly mediated via the HS chains. rib-1 and rib-2 are HS polymerases (EXT/EXTL orthologues) required for the biosynthesis of HS chains [23,24,31]. Homozygous mutations in rib-1 (ok556) are lethal or sterile [24] and although some homozygous rib-2 (tm710) mutants from hetero-zygous mothers become adults their progeny dies shortly after gastrulation [31]. Homozygous rib-2 (tm710) mutants from  heterozygous mothers were analyzed but were found to have no defects in HSN migration (n animals = 50). The presence of maternal rib-2 mRNA in the embryo at the time of HSN migration may provide enough RIB-2 protein to mask any defects that result from complete lack of RIB-2. hst-1 encodes an orthologue of ndst, heparan N-de-acetylase-N-sulfotransferase. Deletion mutation in ndst/hst-1 (ok1068) is also homozygous lethal preventing analysis of homozygous offspring from homozygous mutant mothers. Hence the focus on the study was on enzymes modifying HS further in the biosynthetic pathway, mutants of which are all viable. Single mutants in C. elegans heparan C5-epimerase, hse-5 (tm472), and heparan 2-O-sulfotransferases, hst-2 (ok595) have 15% and 30% defects in HSN migration, respectively, whereas single mutants in heparan 6-O-sulfotransferase, hst-6 (ok273), are wild type for HSN migration (Fig. 3 and [19,21,26]. C. elegans has two orthologues of heparan 3-O-sulfotransferases, hst-3.1 and hst-3.2. Single mutants in hst-3.1 (tm734) are wild type for HSN migration, whereas hst-3.2 (tm3006) mutants have 17% defects in HSN migration (Fig. 3).

Heparan 2-O-and 6-O-sulfotransferase mutants have additive defects in HSN migration
Genetic elimination of multiple O-sulfotransferases has very robust effects on HSN migration in C. elegans. Removal of both HS 6-and 2-O-sulfotransferases in hst-6 hst-2 double mutants leads to 83% (n = 86) HSN migration defect compared to 30% defect in hst-2 single mutants ( Fig. 4B; P,0.001). In hst-2 single mutants only 2% of the HSNs failed to migrate at all and remained positioned in the tail. In contrast, in hst-6 hst-2 double mutants 17% of all HSNs failed to initiate migration suggesting that removing both HS 2-and 6-O-sulfation severely disrupts molecular interactions required for neuron migration. Intriguingly, further removal of one of the 3-O-sulfotransferases, HST-3.1, from hst-6 hst-2 double mutants suppresses the HSN migration defect (Fig. 4B). In hst-3.1; hst-6 hst-2 triple mutants 53% (n = 176) of the HSNs are posteriorly misplaced, with only 2% failing to migrate at all. Removal of HST-3.1 in the hst-3.1 hst-2 double mutant also weakly suppresses the HSN migration defects of single hst-2 mutants to 21% (n = 83; P,0.05). However, genetic elimination of the other 3-O-sulfotransferase, HST-3.2, from hst-2 mutant slightly increases the HSN migration defects to 35% (n = 268) in the hst-3.2 hst-2 double mutant.

HS modification dependence on HSPG core protein
Specific HS modifications are critical for defined developmental processes in vertebrates. For example, mice lacking heparan 2-Osulfation have total renal agenesis [11] suggesting that 2-O-sulfates are critical for the growth factors acting in kidney development [32]. HS 2-O-sulfates regulate growth factors involved in the induction and differentiation of kidney mesoteric mesenchyme [33], whereas 6-O-sulfates regulate growth factors required for ureteric bud branching later in kidney development [34]. The HSPGs carrying these HS modifications have not been identified. Neither is it known if the same HSPG undergoes developmental changes in its HS or whether different HSPGs carry the  structurally different HS chains. Gene knock-out of a HS biosynthetic enzyme such as HS2ST is anticipated to affect all HSPG core proteins.
To gain insight into relationship between HSPG protein cores and specific HS modifications, effects of eliminating HS biosynthetic enzymes in different core protein mutant backgrounds were tested. Eliminating 3-or 6-O-sulfates using hst-3.2 (tm3006) or hst-6 deletion alleles in sdn-1 null background did not alter the severity of the HSN phenotype (53% and 52% defect, respectively) as compared to sdn-1 null mutants alone. In contrast, genetic removal of 2-O-sulfates in the sdn-1 null background (sdn-1 hst-2 double mutant) lead to 87% (n = 119) failure of HSN migration to wild type position, with 37% of HSNs failing to migrate at all (Fig. 5). Similarly, double mutants of hse-5 epimerase and sdn-1 display more severe defects with 61% of HSNs (n = 81) failing to migrate to wild type position and 33% failing to migrate at all (compared with 17% in sdn-1 single mutants; P,0.01). In contrast, additional removal of 2-O-sulfates from glypican/lon-2 and collagen XVIII/cle-1 mutant backgrounds did not significantly alter the neuron defects observed in a single mutant. In lon-2 hst-2 double mutants 37% (n = 91) and in cle-1; hst-2 double mutants 24% of all neurons (n = 129) failed to migrate to wild type position. The pattern of migration defects was not significantly changed either.

Tissue specific expression of HS biosynthetic enzymes and HSPGs
Reporter constructs were used to assess expression patterns of hst-3.1 and hst-3.2 and revealed complementary expression patterns for the two 3-O-sulfotransferases. hst-3.2 reporter is expressed predominantly in neurons (Fig. 5C) whereas hst-3.1 reporter is expressed predominantly in pharyngeal and body wall muscle (Fig. 5D). Similar complementary expression patterns have previously been shown for the other HS biosynthetic enzymes and HSPGs. hst-6 is predominantly expressed by neurons, hst-2 is expressed in the muscle and hypodermis, and hse-5 is expressed in the hypodermis [19,21]. sdn-1 is predominantly expressed in neurons ( Fig. 5B; [26], whereas lon-2 and cle-1 are expressed in muscle and hypodermis [18,22]. Heparan sulfates are required for HSN migration and axon guidance but not process outgrowth Deletion mutations in genes encoding for HSPG core proteins and for HS biosynthetic enzyme, led to defects in HSN migration and axon guidance. However, process outgrowth per se was not affected in any of the mutants (Fig. 6). In other words, irrespective of whether the HSN cell body had migrated to wild type or aberrant position or failed to migrate at all, the HSNs extended axonal projections, suggesting that the process of axon outgrowth is independent of HS/HSPGs (Fig. 6). The defects in HSN migration and axon guidance were also independent of each other. HSN cell bodies which had failed to migrate at all often extended axons that appeared projecting correctly (Fig. 6A). Conversely, HSN cell bodies which had migrated to correctly to the vulval proximity frequently projected aberrant axons which failed to reach their appropriate targets. HS/HSPGs thus play independent roles in HSN neuron migration and in axon guidance.

Discussion
Multiple HSPGs were found to play a role in the migration of the HSNs. Glypican/LON-2, collagen XVIII/CLE-1 and syndecan/SDN-1 are required in parallel neuron guidance pathways as genetic elimination of two or all of the HSPGs increases the severity of the HSN migration phenotype. In the absence of all three HSPGs only 20% of HSNs migrate normally. Although mutations in the gpn-1/glypican alone do not lead to defects in HSN migration, genetic elimination of gpn-1 in lon-2 sdn-1 background significantly increased the proportion of neurons that failed to migrate at all suggesting that in the absence of LON-2, GPN-1 plays a redundant role resulting in increased severity of defects when both glypicans are genetically eliminated.
sdn-1 is expressed by predominantly in neurons including the HSNs [26], whereas lon-2 and cle-1 are expressed in the hypodermis and body wall muscle [18,22]. This suggests that SDN-1/syndecan is required in the migrating neuron, the HSNs, whereas LON-2/glypican and CLE-1/collagen XVIII are required in the matrix that supports HSN migration and all three HSPGs act in concert. When LON-2/glypican is present, GPN-1/ glypican is not required for HSN migration. Moreover, SDN-1 HS chains are important for HSN migration as sdn-1 null allele and 'HS-less' allele display identical phenotypes.
Simultaneous genetic removal of both hst-2 and hst-6 results in almost complete loss of HSN migration suggesting that heparan O-sulfation is needed for HSN migration. Biochemical HS disaccharide analysis has shown increased N-sulfation in hst-2 hst-6 double mutants [35], however, as suggested by this study, glucosamine N-sulfation is not able to functionally compensate for the lack of O-sulfation. Genetic removal of hst-6 and hst-3.1 in hse-5 epimerase mutant background significantly enhances defects in HSN migration whereas removal of hst-2 in hse-5 background does not. Biochemical HS disaccharide analysis has shown compensatory 6-O-sulfation in hse-5; hst-2 double mutants retaining higher total 2-and 6-O-sulfation per 100 disaccharides than that seen in hse-5; hst-2 double mutants [35]. Heparan 2-Osulfation occurs predominantly in IdoA residues that have undergone C5-epimerisation (by HSE-5 epimerase) from GluA although some 2-O-sulfation occurs also on GluA residues. It is also plausible that removing both HSE-5 and HST-2 affects the same sugar units in HS, whereas removing HSE-5 and HST-6 or HST-3.1 affects both hexuronic acid and glucosamine units of HS. Hexuronic acids have very flexible conformation, and 2-Osulfation further increases this flexibility [36]. Changes to this flexibility may affect HS-ligand interactions. In contrast, 3-and 6-O-sulfate groups do not affect conformation of HS to the extent that 2-O-sulfates do [37], which could partly explain why eliminating either HST-3.1 or HST-6, or both, does not affect HSN migration in vivo provided other HS modifications are present. Biochemical HS analysis has shown that hst-6 mutants have increased compensatory 2-O-sulfation, retaining the combined 2-and 6-O-sulfation per 100 disaccharides at wild type levels [35], which may explain why single mutants of hst-6 display wild type neuron migration phenotype.
3-O-sulfation is the rarest of HS modifications, occurring in less than 5% of HS derived disaccharides. Findings that elimination of hst-3.1 suppress the HSN migration defects in hst-2 single and hst-2 hst-6 double mutant backgrounds suggest that in the absence of 2-O-sulfates or 2-and 6-O-sulfates the presence of 3-O-sulfates is unfavorable for the biological function of the HS. Alternatively, in the absence of 2-O-or 2-and 6-O-sulfation, 3-O-sulfation may be up-regulated. NMR studies have shown that 3-O-sulfates can in some cases influence the conformation of the uronic acid adjacent to the 3-O-sulfated glucosamine [38]. Up-regulation of 3-Osulfation could thus change the overall HS saccharide conformation and alter ligand binding properties.
Compensatory mechanism in HS biosynthesis has been demonstrated in vertebrates. The classical view of HS biosynthesis is that N-deacetylation/N-sulfation of glucosamine residues precedes other HS modifications that occur primarily in previously N-sulfated domains. However, some 6-O-sulfation is observed in mouse embryonic stem cells deficient of NDST1 and NDST2 and which completely lack N-sulfation [39]. The compensatory modifications that occur when one or more HS modification enzymes are eliminated together with the conformational changes to the resulting HS may alter ligand binding properties leading to altered phenotypic outcomes.
sdn-1 hst-2 double mutants exhibit HSN migration defect that resembles combined elimination of CLE-1, SDN-1 and LON-2 HSPGs. Similarly, genetic elimination of HS 2-and 6-O-sulfation resembles loss of SDN-1 and HST-2, or combined loss of CLE-1, SDN-1 and LON-2 HSPGs. hst-6 sdn-1 double mutants appear like sdn-1 single mutants. Genetic elimination of hst-3.2, leads to weak defects in HSN migration, but sdn-1 hst-3.2 double mutants mirror sdn-1 single mutants. hst-6 and hst-3.2 are predominantly expressed by neurons whereas hst-2 and hst-3.1 are expressed in the muscle and hypodermis. Combining the normal expression patterns of these genes and the HSN phenotypes observed in the mutants, suggest a model in which SDN-1, which is expressed in the migrating neuron, contains HS chains decorated with 3-and 6-O-sulfates, and the SDN-1 HS chains are important for normal HSN migration. Removal of HST-2 from LON-2 or CLE-1 mutants, did not alter HSN migration defect of the single mutants, suggesting that LON-2/glypican and CLE-1/collagen XVIII, present in the matrix that supports migrating neurons, contain 2-O-sulfated HS required for normal HSN migration. Removal of both HS 2-and 6-O-sulfotransferases resembles loss of all three HSPGs, suggesting that the role of all these HSPGs in neuron migration is predominantly mediated by the HS chains. Based on these findings, it is also plausible that in C. elegans, tissue specific expression of different HSPGs and HS O-sulfotransferases in part determines core protein specificity of HS modifications.
In conclusion, a neural development model in C. elegans was used to address relationship between different HSPG core proteins and distinct HS modifications for biological function in vivo. Genetic analysis demonstrated that SDN-1/syndecan, LON-2/ glypican and CLE-1/collagen XVIII are required in parallel guidance pathways and genetic elimination of all three HSPGs results in almost complete failure of neural migration. The different HSPG core proteins require distinct HS modifications to mediate in vivo functions. The core protein is thus not a carrier of HS chains with any or random modifications. In C. elegans this core protein-HS specificity may in part be imposed by tissue specific expression of the genes encoding for HSPGs and HS biosynthetic enzymes. This is the first demonstration of relationship between HS modifications and distinct HSPG core proteins in a defined biological process and provides a novel insight into the biological functions of HS and HSPGs. Proteoglycan core protein specificity should be considered when interpreting data or considering therapies relating to distinct HS structures.