Contribution of EXT1, EXT2, and EXTL3 to Heparan Sulfate Chain Elongation*

The exostosin (EXT) family of genes encodes glycosyltransferases involved in heparan sulfate biosynthesis. Five human members of this family have been cloned to date: EXT1, EXT2, EXTL1, EXTL2, and EXTL3. EXT1 and EXT2 are believed to form a Golgi-located hetero-oligomeric complex that catalyzes the chain elongation step in heparan sulfate biosynthesis, whereas the EXTL proteins exhibit overlapping glycosyl-transferase activities in vitro, so that it is not apparent what reactions they catalyze in vivo. We used gene-silencing strategies to investigate the roles of EXT1, EXT2, and EXTL3 in heparan sulfate chain elongation. Small interfering RNAs (siRNAs) directed against the human EXT1, EXT2, or EXTL3 mRNAs were introduced into human embryonic kidney 293 cells. Compared with cells transfected with control siRNA, those transfected with EXT1 or EXT2 siRNA synthesized shorter heparan sulfate chains, and those transfected with EXTL3 siRNA synthesized longer chains. We also generated human cell lines overexpressing the EXT proteins. Overexpression of EXT1 resulted in increased HS chain length, which was even more pronounced in cells coexpressing EXT2, whereas overexpression of EXT2 alone had no detectable effect on heparan sulfate chain elongation. Mutations in either EXT1 or EXT2 are associated with hereditary multiple exostoses, a human disorder characterized by the formation of cartilage-capped bony outgrowths at the epiphyseal growth plates. To further investigate the role of EXT2, we generated human cell lines overexpressing mutant EXT2. One of the mutations, EXT2-Y419X, resulted in a truncated protein. Interestingly, the capacity of wild type EXT2 to enhance HS chain length together with EXT1 was not shared by the EXT2-Y419X mutant.

Heparan sulfates (HSs) 4 are sulfated glycosaminoglycans (GAGs) distributed on the cell surfaces and in the extracellular matrices of most tissues. HS chains are synthesized covalently linked to various proteins in a proteoglycan structure. HS proteoglycans have been implicated as regulators in a diverse number of biological events related to intracellular signaling, cellcell interactions, and tissue morphogenesis (1). Most of these activities depend on the binding of the negatively charged HS chains to a variety of molecules including members of the fibroblast growth factor family, serine protease inhibitors, and extracellular matrix proteins (2,3).
The biosynthesis of HS is a complex process (reviewed in Refs. [2][3][4]. Chain elongation is initiated by the formation of a tetrasaccharide linkage region composed of glucuronic acid-galactose-galactose-xylose (GlcA␤1-3Gal␤1-3Gal␤1-4Xyl␤1-), where Xyl is attached to a serine residue in the core protein (5). After the addition of a single GlcNAc residue by GlcNAc transferase I (GlcNAc-TI), elongation proceeds by the action of glycosyltransferases (GlcNAc-TII and GlcA-TII), which add ␤1,4-GlcA and ␣1,4-GlcNAc units in alternating sequence to the nonreducing end of the growing polymer (6). The polymerization reactions are carried out by the exostosin proteins EXT1 and EXT2 (7)(8)(9). Concomitantly with chain elongation modification of the chain occurs, initiated by N-deacetylation and N-sulfation of glucosamine units, carried out by the bifunctional enzyme N-deacetylase/N-sulfotransferase (NDST). These reactions occur in a more or less blockwise fashion and create stretches of N-sulfated regions and nonmodified N-acetylated domains, interspersed with short mixed regions (N-sulfated/N-acetylated domains) (10,11). The N-deacetylation/N-sulfation reaction has been regarded as a key regulatory step because all subsequent modifications, epimerization of GlcA to iduronic acid and various sulfation reactions, only occur in the vicinity of N-sulfate groups. However, NDST1 Ϫ/Ϫ /NDST2 Ϫ/Ϫ embryonic stem cells that completely lack N-sulfation are still able to produce 6-O-sulfated HS (12). Subsequent modifications include C-5 epimerization of GlcA to iduronic acid, 2-O-sulfation of iduronic acid, and 6-O-sulfation of N-acetylated and N-sulfated GlcN residues. In addition, a small number of N-sulfated D-glucosamine residues become O-sulfated at C-3, and a few GlcA units become O-sulfated at C-2. Five members of the EXT family of glycosyltransferases are known in humans: EXT1, EXT2, EXTL1 (for EXT-like 1), EXTL2, and EXTL3. Members of the EXT gene family encode glycosyltransferases that are suggested to be involved in HS biosynthesis, and most of the members seem to possess more than one in vitro glycosyl transferase activity (5,9). EXT1 and EXT2 appear to have dual enzyme activities, GlcA-TII and GlcNAc-TII, although those of EXT2 are weak and have been questioned. Even though EXT1 alone is able to polymerize the HS backbone structure in vitro, all evidence suggests that both EXT1 and EXT2 are necessary for chain elongation (13,14). All three EXTLs catalyze GlcNAc transferase reactions. EXTL1 adds GlcNAc to the nonreducing end of the growing polysaccharide (GlcNAc-TII activity); EXTL3 adds GlcNAc to the GAG protein linkage region (GlcNAc-TI activity) and to the growing chain (GlcNAc-TII activity), thus harboring enzyme activities involved in both the initiation and elongation of HS chains (15); and EXTL2, the shortest member of the EXT family, exhibits GlcNAc-TI activity (16). The biological roles of EXTL1 and EXTL2 in HS biosynthesis have not yet been demonstrated, and in contrast to the other members of the EXT family, orthologs of the mammalian EXTL1 and EXTL2 are absent in Drosophila, suggesting that they are not essential for HS biosynthesis.
Hereditary multiple exostoses (HME), an autosomal dominant hereditary disorder, is one of the most common benign skeletal conditions affecting 1:100 000, with a risk for malignant transformation. The clinical signs are shorter stature and skeletal deformities caused by cartilage capped bony outgrowths at the epiphyseal end of the long bones (17). The mechanism by which exostosis develops is poorly understood, but mutations in either EXT1 or EXT2 and the resulting reduction or absence of HS in the exostosis cartilage cap has been implicated in disturbed signaling response in exostosis chondrocytes (18,19). A number of mutations in EXT1 and EXT2 have been reported in HME patients, and the majority of cases involve mutations in EXT1 (60 -70% of the patients) (9,20). The mutations are randomly distributed over the entire EXT1 gene, whereas EXT2 mutations appear to be concentrated toward the N-terminal part of the protein. Most mutations result in premature termination of translation and loss of function. Less common are missense mutations affecting single amino acids. The amino acid substitutions are believed to alter the functional properties of the EXT proteins. However, although it is generally believed that the mutations result in abnormal HS synthesis, this is not always the case. Some of the reported EXT1 mutant proteins retain their ability to synthesize HS (21).
Using the human embryonic kidney (HEK) 293 cells, we evaluated the contribution of EXT1, EXT2, and EXTL3 to HS chain elongation by up-regulation of their expression by overexpression and down-regulation of their expression by small interfering RNAs (siRNA). We also analyzed the effect on HS synthesis of EXT2 with two different mutations found in HME patients.
Mutant bases were introduced into EXT2 cDNA using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. EXT2-D227N point mutation was constructed using forward primer 5Ј-CTTACC-GGCAGGGCTACAATGTCAGCATTCCTG-3Ј and complementary reverse primer 5Ј-CAGGAATGCTGACATTGTAG-CCCTGCCGGTAAG-3Ј. The EXT2-Y419X deletion mutant was generated using forward primer 5Ј-ATCAATGACAGGA-TCTAACCATATGCAGCCATCTCC-3Ј and complementary reverse primer 5Ј-GGAGATGGCTGCATATGGTTAGATC-CTGTCATTGAT-3Ј (mutated nucleotides are underlined and in bold). The wild type and mutated inserts were excised with EcoRI and subcloned into the corresponding site of the pcDNA-3 expression vector (Invitrogen). Plasmids with inserts containing the coding region in 5Ј to 3Ј direction were selected by sequencing.
A truncated, soluble form of human EXTL3, lacking the first 51 amino acids, was amplified using a human placenta cDNA (Clontech) with the primers 5Ј-ATATGCGGCCGCAACCAC-TCTGGATGAG (forward) and 5Ј-GGGCTCTAGAGATGA-ACTTGAAGCACTTGGT (reverse), and the PCR products were cloned in-frame with the preprotrypsin leader sequence and the N-terminal 3-FLAG tag p3XFLAGCMV9 expression vector (Sigma). The truncated forms of EXT1 and EXT2 were as described (13).
Metabolic Labeling, Isolation, and Analysis of HS Chains from Cell Culture-Subconfluent transfected cell cultures were incubated with 200 Ci/ml [ 35 S]sulfate or 50 Ci/ml [6-3 H]glucosamine HCl for the time periods indicated in the text. After incubation at 37°C the medium was removed and frozen for future use, the cells were washed twice with phosphate-buffered saline (PBS) and then treated with 1 mg/ml trypsin (Sigma) for 10 min at 37°C, and the cell suspensions were centrifuged. The supernatants were collected and treated with 0.5 M NaOH at 4°C overnight to release the O-linked saccharides from the protein core. The samples were applied to 300-l DEAE-Sephacel columns, equilibrated in 150 mM NaCl, 50 mM Tris/ HCl, pH 7.5, 0.1% Triton X-100. The columns were washed with 10 bed volumes of 1) the equilibration buffer; 2) 50 mM sodium acetate, pH 4.0, 150 mM NaCl, 0.1% Triton X-100; 3) 50 mM Tris/HCl, pH 7.5, 150 mM NaCl; or 4) H 2 O. The GAGs were finally eluted with 2 M NH 4 HCO 3 and lyophilized. Galactosaminoglycans were digested with chondroitinase ABC (Seikagaku) in 50 mM Tris/HCl, pH 8.0, 30 mM sodium acetate and 0.1 mg/ml bovine serum albumin, and the reaction mixtures were separated by gel filtration on a Superose 6 HR10/30 column (Amersham Biosciences) eluted with 0.5 M NH 4 HCO 3 . HS was digested with heparitinase I and II (Seikagagu) in 50 mM Tris/HCl buffer, pH 8.0, containing 30 mM sodium acetate and 0.1 mg/ml bovine serum albumin.
To isolate cellular HS, the cells were solubilized in 50 mM Tris/HCl, pH 7.5, 1% Triton X-100, and protease inhibitors (2 mM EDTA, 2 mM N-ethylmaleimide, 1 mM Pefabloc, and 10 g of pepstatin/ml) for 1 h at 4°C. After centrifugation at 2000 rpm for 20 min, the supernatant containing the radiolabeled proteoglycans was recovered and isolated by DEAE anion exchange chromatography. After desalting on PD-10 columns (Sephadex G-25) (Amersham Biosciences), GAG chains were released from the peptide cores and isolated using PD-10 gel filtration columns eluted with 0.2 M NH 4 HCO 3 and lyophilized. The samples were digested with chondroitinase ABC and analyzed as described above. For analysis of medium-derived HS chains, NaOH to a final concentration of 0.5 M was added to the medium. After 24 h at 4°C the medium was neutralized and diluted to a final concentration of 150 mM NaCl. Sulfated GAG chains were isolated on 500-l DEAE-Sephacel columns essentially as described above.
Compositional Analysis of Labeled HS-Labeled HS chains were depolymerized with nitrous acid at pH 1.5 followed by reduction with NaBH 4 (25). The labeled deamination products were fractionated by gel chromatography on Sephadex G-15 (1 ϫ 180 cm) in 0.2 M NH 4 HCO 3 . Fractions corresponding to oligo-and disaccharides were collected and desalted by lyophilization. Labeled disaccharides were analyzed by anion exchange HPLC using a Whatman Partisil 10-SAX column eluted with aqueous KH 2 PO 4 of stepwise increasing concentration at a rate of 1 ml/min (26).
Glycosyltransferase Assays-GlcNAc and GlcA transferase activities were measured essentially as described (13). Glc-NAc-T activities of crude cell extracts were determined by incubating 60 -90 g of lysate protein with 0.125 Ci of UDP-[ 14 C]GlcNAc and 50 g of [GlcA-GlcNAc] n oligosaccharide (the reducing terminal sugar is modified to an 2,5-anhydromannitol unit). After 1 h at 37°C, the mixtures were applied to columns (28 ϫ 0.5 cm) of Sephadex G-25 in 0.2 M NH 4 HCO 3 . Effluent fractions were analyzed by scintillation counting. GlcA-T activities were assayed in similar fashion, except that UDP-[ 14 C]GlcA was used as a sugar donor, and GlcNAc-[GlcA-GlcNAc] n oligosaccharides were used as acceptors.
Immunofluorescence-For immunocytochemical analysis, HEK 293 cells were transiently transfected with expression plasmids on coverslips as described above. Twenty-four hours after transfection, the cells were washed twice with PBS, fixed with 4% paraformaldehyde in PBS for 10 min, washed three times with PBS, permeabilized in 0.1% Triton X-100 for 15 min in room temperature, and blocked with 10% bovine serum albumin in PBS. The cells were then incubated with the polyclonal primary antibodies, goat anti EXT2-N15 (Santa Cruz Biotechnology), and the Golgi marker mouse-anti GM130 (BD Biosciences, provided by Dr. Jaakko Saraste, University of Bergen) at dilutions of 1:100 and 1:200, respectively, for 1 h at room temperature. They were then rinsed three times in PBS and further incubated for 2 h with the secondary antibodies Alexa Fluor 488 donkey anti-goat and Alexa Fluor 555 donkey antimouse IgG, respectively. Finally, they were mounted in VectaShield with 4Ј,6Ј-diamino-2-phenylindole (Vector Laboratories). The images were visualized under a Zeiss Axioscope microscope equipped with optics for observing fluorescence and captured using a digital AxioCam MRm camera.

RESULTS
siRNA-mediated Silencing of EXT1, EXT2, and EXTL3-To investigate the role of EXT1, EXT2, and EXTL3 in HS chain elongation, we transiently transfected HEK 293 cells with siRNA targeting EXT1 (siEXT1), EXT2 (siEXT2), EXTL3 (siEXTL3), and complement C1r (control siRNA). As deter-mined by real time PCR, the respective EXT mRNA expression levels were reduced to Յ10% of that of the control siRNAtreated cells. The cells transfected with either siEXT1 or siEXT2 exhibited decreased GlcNAc and GlcA transferase activities relative to the control siRNA-transfected cells. By contrast, treatment of cells with EXTL3 siRNA exhibited slightly increased transferase activities (Fig. 1). We were interested to determine whether there was a possible coregulation of EXT1, EXT2, and EXTL3 at the mRNA level. Therefore, we also examined whether silencing of EXT1, EXT2, or EXTL3 would affect the transcript levels of the other EXTs examined. Analysis by real time PCR revealed no detectable effect of silencing of one EXT on the mRNA levels of the other EXTs (data not shown).
Effect of siRNA-mediated Silencing of EXT1, EXT2, and EXTL3 on HS Chain Length-To investigate the effect of siRNA on HS synthesis, HEK 293 cells transiently silenced with siEXT1, siEXT2, siEXTL3, or the control siRNA were metabolically labeled for 24 h with [ 35 S]sulfate or [ 3 H]glucosamine. Proteoglycans were isolated from the cell surface/matrix and from the culture medium, and the GAGs were released from the core proteins.
The cells transfected with EXTL3 siRNA displayed an increased chain length, and those transfected with EXT1 and EXT2 siRNA produced HS chains that were shorter than those synthesized by the control cells (Fig. 2). siEXT1 and siEXT2 cells labeled with [ 3 H]glucosamine exhibited a ϳ50% lower HS/chondroitin sulfate ratio relative to the control transfected cells, whereas the ratio in siEXTL3 cells was largely unaffected. Interestingly, the effect of reduced EXT1 or EXT2 mRNA levels on HS chain elongation is analogous to our previous findings regarding HS produced by mouse embryonic fibroblasts carrying a hypomorphic mutation in EXT1 so that they still contained small amounts of normal EXT1 transcript (27). The mutant cells produce about 18% HS compared with wild type cells, and the mutations mainly affect HS chain length, although some reduction in the number of chains may also have occurred (27).
Effect of siRNA-mediated Silencing of EXT1, EXT2, and EXTL3 on HS Domain Organization-For structural analysis we selected 3 H samples with more than 90% reduction in their mRNA levels (as determined by real time PCR) and with a marked effect on HS chain length. The various samples were treated with chondroitinase ABC to degrade any galactosami-noglycans present in the samples. Chondroitinase ABC-resistant labeled polysaccharides were isolated, and the N-substitution patterns of HS from siRNA-treated cells were analyzed by treatment of [ 3 H]glucosamine-labeled polysaccharides with nitrous acid at pH 1.5 that results in cleavage of the chains at the sites of N-sulfated glucosamine units. Under these conditions, N-acetylated units remain intact. Contiguous N-sulfated sequences will be degraded to disaccharides, whereas alternating N-sulfated and N-acetylated glucosamine residues will give rise to tetrasaccharides. Spaced sequences with solitary N-sulfate groups will yield oligosaccharides of at least hexasaccharide size. The gel chromatography profiles of the products revealed very similar disposition of the sulfated regions with a range of differently sized saccharide fragments (Fig. 3). However, it was noted that the silencing of EXT1 and EXT2 resulted in more pronounced peaks of ϳ26-mers and larger oligosaccharides than HS from control cells.
Detailed information regarding the distribution of O-sulfate groups was obtained by analysis of the disaccharides generated by low pH HNO 2 /NaBH 4 treatment of 3 H-labeled HS fractions (see "Experimental Procedures"). The resultant disaccharides were separated by anion exchange HPLC into different O-sul-   NOVEMBER 9, 2007 • VOLUME 282 • NUMBER 45 JOURNAL OF BIOLOGICAL CHEMISTRY 32805 fated species. There was a small increase 6-O-sulfated disaccharides in siEXT2 and siEXTL3 HS compared with the control ( Table 1).

EXTs and Heparan Sulfate Chain Elongation
Effect of Overexpression of EXT1, EXT2, and EXTL3 on HS Chain Length-We next investigated the effect of overexpression of EXT1 and EXT2 on HS chain length. HEK 293 cells were stably transfected with the vector alone (mock), EXT1, EXT2, or coexpressed EXT1 and EXT2 (EXT1/EXT2). The specific GlcNAc and GlcA transferase activities of the EXT1 overexpressing cells expressed on a cellular protein basis were increased compared with the activities obtained for mock transfected cells (Fig. 4, A and B). Similar to previous observations (8) coexpression of EXT1 and EXT2 further augmented the two glycosyltransferase reactions. Mock transfected, EXT1, EXT2, and EXT1/EXT2 overexpressing cells were metabolically [ 35 S]sulfate-labeled for 60 min, and cell-associated 35 Slabeled HS proteoglycans were isolated by gel chromatography on Superose 6, and the HS chains were released from the core protein by alkali treatment (28). As shown in Fig. 5A, EXT1 overexpression resulted in a significant increase in chain length, which was even more enhanced in EXT1/EXT2-overexpressing cells. HS from EXT2 or EXTL3 overexpressing cells showed no striking differences in chain length from control cells, even though some variations in chain length were observed between different labeling experiments. The effect on chain elongation was not affected by the labeling time because the effect on HS chain elongation was similar also when cells were labeled for 6 h or overnight. That the chondroitinase ABC-resistant chains indeed represent HS was ascertained by the complete degradation of labeled material to di-and smaller oligosaccharides by digestion with a combination of chondroitinase ABC and heparitinase I and II (data not shown).
The EXT proteins are type II membrane-bound proteins, and to determine whether the observed effects on HS chain elongation were dependent on the membrane-bound status of the enzymes, we next investigated the effect on HS chain length of overexpression of soluble EXT1, EXT2, and EXTL3. HEK 293 cells were stably transfected with 3-FLAG-tagged human EXT1, 3-FLAG-tagged EXTL3, and 3-FLAG-tagged EXT2, or cotransfected with 3-FLAG-EXT1/MycHis-EXT2 (EXT1/ EXT2) constructs. All of the constructs lacked the transmembrane domains, thus yielding soluble fusion proteins that were released into the culture medium. The expressed proteins were affinity captured on anti-FLAG-agarose, and the bound fusion proteins were analyzed for glycosyltransferase activities. The glycosyltransferase activities of immunopurified EXT1, EXT2, and EXT1/EXT2 have been described previously (13). The immunopurified EXTL3 exhibited weak GlcNAc-TII activity and no detectable GlcA-TII activity (data not shown). The weak  a aMan R , the 2,5-anhydromannitol deamination products of N-sulfated D-glucosamine residues; IdoA, iduronic acid. activity may be explained by the fact that, although EXTL3 has been shown to efficiently transfer GlcNAc to GlcA-[GlcNAc-GlcA] n -aMan R acceptors (measuring GlcNAc-TII), its activity toward acceptors mimicking the HS protein linkage region (measuring GlcNAc-TI) is much higher (15). Overexpression of secreted EXT1 and EXT1/EXT2 again resulted in a significant increase in chain length, whereas overexpression of EXT2 or EXTL3 did not significantly affect HS chain elongation (Fig. 5B). However, in contrast to the result observed with the full-length constructs, coexpression of soluble EXT2 and soluble EXT1 did not enhance HS chain length more than EXT1 alone. These findings suggest that the catalytic activities of EXT1 not depend on the transmembrane region of the expressed EXT1 protein. Western blotting of transfected cells showed that the expressed soluble proteins were also present in the cell extracts. Thus, soluble EXT1 protein apparently is able to interact with the biosynthetic machinery in the Golgi apparatus.
Effect of Overexpression of Mutated EXT on HS Chain Length-To evaluate the effect of EXT2 mutations in this system, we generated HEK 293 cell clones stably transfected with EXT2-D227N and EXT2-Y419X cDNAs. Previously, HME patients have been grouped according to several clinical parameters such as onset of exostoses growth, number of exostoses, stature, and functional rating (20). One of the mutations that we introduced, EXT2-D227N, is considered to give a mild phenotype. The other mutation, EXT2-Y419X, a nonsense mutation resulting in a truncated protein, gives a moderate severe phenotype.
Lysates of transfected cells were analyzed for GlcA and GlcNAc transferase activities. The enzyme activities of the different cell lines expressing the mutated EXT2 were similar to those of mock transfected cells (data not shown), indicating that overexpression of mutated or wild type EXT2 did not affect the in vitro glycosyltransferase activities. Stable clones expressing high levels of EXT2, EXT2-D227N, and EXT2-Y419X proteins, respectively, were selected by Western blotting and used to investigate the effect of mutated EXT2 expression on HS synthesis. The HEK 293 cell clones were metabolically labeled for 24 h with [ 35 S]sulfate and radiolabeled GAGs were isolated from solubilized cells and from the culture medium and quantified. Expression of wild type or mutated forms of EXT2 did not change the total amount of [ 35 S]GAGs produced nor the HS/CS ratio as compared with mock transfected cells (data not shown). Similar to the EXT2 overexpressing cells, the cells overexpressing mutated forms of EXT2 displayed no extensive changes in chain length compared with HS in mock transfected cells (Fig. 5C), indicating that the mutated EXT2 proteins did not exhibit a dominant negative effect on HS synthesis. We next considered whether the EXT2 mutation would affect the enhancing capacity of EXT2 on HS chain elongation when coexpressed with EXT1. HEK 293 cells were transiently transfected with EXT1 or EXT2-Y419X alone or cotransfected with EXT1 and EXT2 or with EXT1 and EXT2-Y419X (EXT1/ EXT2-Y419X). Analysis of 35 S-labeled HS from cells transiently transfected with EXT1 and EXT1/EXT2 again showed that the most extended HS chains were made in cells coexpressing EXT1 and EXT2 (Fig. 5D). In contrast, EXT1/EXT2-Y419X coexpression did not stimulate chain elongation more than EXT1 alone (Fig. 5D). Similar results were obtained after transient transfection of EXT2-Y419X into stably transfected EXT1 cell clones (data not shown). Western blotting of the transfected cells demonstrated that similar amounts of EXT2 and EXT2-Y419X, respectively, were expressed in the transiently transfected cells (data not shown). Thus, the lack of effect of the EXT2 mutant was not due to less EXT2 protein available. Analogous to the effect on chain elongation, transient expression of EXT1/EXT2-Y419X did not influence the transferase activities (Fig. 4, C and D).
Golgi Localization of the Overexpressed EXT Proteins-In an attempt to identify the mechanism responsible for this differential response of HS chain elongation of HEK 293 cells transfected with EXT1/EXT2 versus EXT1/EXT2-Y419X, we stained the transfected cells with an antibody against EXT2 to determine the cellular localization of the overexpressed proteins. When expressed in HEK 293, both EXT2 and EXT2-Y419X proteins colocalized with the Golgi marker GM130 (Fig. 6). Therefore, the lack of effect after transfection of EXT1/ EXT2-Y419X is not caused by retention of the mutated protein in the ER.

DISCUSSION
Our knowledge of how the synthesis of HS is regulated is surprisingly limited. The mechanism responsible for regulation of HS chain elongation is not known, and the exact role of each particular EXT and EXTL protein is still an enigma. All of them have been shown to possess in vitro glycosyltransferase activities relevant for HS chain initiation or elongation (5,9). A prerequisite to understand the cellular functions of these proteins is to know the molecular mechanism of their actions. Thus, we addressed here the role of EXT1, EXT2, and EXTL3 in HS biosynthesis by means of overexpression studies and gene silencing. The most striking results were the effects of decreased amounts of the EXT proteins on HS chain elongation.
We have shown previously that embryonic fibroblasts isolated from mice carrying a hypomorphic mutation in Ext1 synthesize significantly shorter HS chains than do wild type fibroblasts (27). Consistent with this, silencing of EXT1 by EXT1-targeted siRNA in human cells resulted in lower transferases activities (Fig. 1) and the formation of shorter HS chains (Fig. 2). Interestingly, similar inhibition of EXT2 also resulted in the synthesis of shorter chains, confirming that EXT2 is indeed a partner of EXT1 in chain elongation. To gain more insight into the role of EXT2 in HS synthesis, we also compared the effect on HS chain elongation of cell lines overexpressing wild type and mutated forms of EXT2 alone or coexpressed with EXT1. Mutational defects in either EXT1 or EXT2 cause HME. Because all evidence suggests that both EXT1 and EXT2 are essential for HS polymerization, it is generally believed that defects in HS synthesis cause the abnormalities related to HME. Both EXT1 and EXT2 are ubiquitously and abundantly expressed in mammalian tissues. However, the effects of mutations seem to be limited to the growing bone. Despite the relatively high expression of the mutated EXT2 constructs, they did not alter HS synthesis, indicating that the mutated EXT2 constructs (EXT2-D227N with an Asp to Asn exchange, and EXT2-  Y419X, encoding a 419-amino acid protein lacking the 300 most C-terminal amino acid residues) did not exhibit a dominant negative effect. Interestingly, introduction of mutant EXT1 into Chinese hamster ovary cells also did not suppress HS synthesis (29). Taken together, these results indicate that the function of endogenous EXT proteins is not affected by the presence of increased levels of mutant forms of EXT1 or EXT2.
Contrary to the results obtained with the mutated EXT2, coexpression of the wild type EXT2 with EXT1 in mammalian cells had a dramatic effect on both enzyme activity and HS chain length (Figs. 4 and 5). The effect of EXT2 on chain elongation is intriguing. EXT2 is necessary for HS polymerization as evident from mutational analysis in Drosophila, mouse, and zebrafish orthologs of EXT2 (30 -34). Previous in vitro data indicate that EXT1 has readily detectable GlcA and GlcNAc transferase activities, whereas the enzyme activities of EXT2 are less manifest (7,8,13). The lack of effect of EXT2 overexpression could be due to the fact that EXT2 may have a different function in HS polymerization from EXT1. If both EXT1 and EXT2 were catalytically active glycosyltransferases, one would expect gene silencing and overexpression to give similar results for both proteins. EXT2 forms a complex with EXT1, which facilitates the transfer of both proteins to the Golgi apparatus and modulates the activity of the latter protein (7,8,13); therefore, the EXT1/EXT2 hetero-complex is believed to be the functional HS polymerase. Because of the low transferase activities of EXT2, it is tempting to speculate that EXT2 is not involved in the actual elongation of the HS backbone but instead the function of EXT2 could be in assisting the folding and transport of EXT1 to the Golgi complex. One hypothetical explanation of our results could be that an excess of EXT2 protein is normally synthesized in the cell. Thus, increased levels of EXT2 will be without effect and will not affect polymerization. If instead the amounts of EXT1 are increased, more EXT1/ EXT2 heterocomplexes will form and thus affect chain elongation. Overexpression of both EXT1 and EXT2 will generate even more complexes that can participate in chain elongation. It is not known how the complex of EXT1 and EXT2 is formed. Although the mutant localized to the Golgi (Fig. 6), it is possible the mutant EXT2 failed to associate with EXT1 and thus was not able to further promote chain elongation. Further studies are needed to establish the role of EXT2 in HS polymerization. From our results we can conclude that chain length rather than the initiation of new HS chains is affected by the amounts of EXT1 and EXT1/EXT2 complex. Interestingly, NDST1 Ϫ/Ϫ / NDST2 Ϫ/Ϫ embryonic stem cells, which lack N-sulfation, showed increased HS chain length without affecting the amount of HS produced by the cells, indicating a more complex regulation of chain elongation (12).
In contrast to the silencing of EXT1 or EXT2, silencing of EXTL3 resulted in the synthesis of longer HS chains. This protein has been shown to be involved in HS chain initiation (Glc-NAc-TI activity), catalyzing the incorporation of the first Glc-NAc onto the polysaccharide protein linkage region (15). The increase in HS chain length after siRNA-mediated EXTL3 silencing could be explained by the fact that the reduction in EXTL3 results in fewer linkage regions that contain the first GlcNAc necessary to start the HS elongation process (-GlcNAc-GlcA-Gal-Gal-Xyl-O-Ser, where the first GlcNAc is indicated in bold). With less acceptor substrates available, more extensive polymerization can occur in the chains that are being synthesized. The observed lack of effect of EXTL3 overexpression could be due to the fact that the untransfected cells produce saturating levels of EXTL3, which is enough to catalyze the incorporation of GlcNAc residues in all of the available linkage regions.
It has also been proposed that EXTL3 may be involved in chain elongation or chain termination (15). Reducing the amounts of a chain terminator would naturally lead to longer HS chains, but overexpression of EXTL3 did not significantly alter the HS chain length as would have been expected if EXTL3 was a terminator of HS chains. In addition, Chinese hamster ovary cells lacking GlcNAc-TII activity still display GlcNAc-TI activity but lack HS chains (35) Moreover, this mutant accumulates a pentasaccharide intermediate with the structure Glc-NAc-GlcA-Gal-Gal-Xyl (36), favoring the concept that EXTL3 catalyzes the initiation rather than the termination of HS chain elongation. It has been proposed in a recent report that a GlcA residue is present at the nonreducing end of the HS chains (37). EXTL3 harbors no GlcA-T activities (15), so it would be unlikely to catalyze chain termination.
In summary, our results show that siRNA silencing of EXT1, EXT2, or EXTL3 will affect HS chain elongation, which may be one reason behind the formation of exostoses, because the shorter chains produced by the hypomorphic EXT1 mouse affect HS-dependent growth factor signaling. Furthermore, these three proteins may play specific and distinct roles in HS biosynthesis in human cells. EXT1 and EXT2 are together responsible for chain elongation, and the levels of the individual proteins affect the polymerization process. The level of EXTL3 also affects chain elongation, but the changes indicate that or EXTL3 must be an initiator of HS chains.