Embryonic fibroblasts with a gene trap mutation in Ext1 produce short heparan sulfate chains.

Mutational defects in either EXT1 or EXT2 genes cause multiple exostoses, an autosomal hereditary human disorder. The EXT1 and EXT2 genes encode glycosyltransferases that play an essential role in heparan sulfate chain elongation. In this study, we have analyzed heparan sulfate synthesized by primary fibroblast cell cultures established from mice with a gene trap mutation in Ext1. The gene trap mutation results in embryonic lethality, and homozygous mice die around embryonic day 14. Metabolic labeling and immunohistochemistry revealed that Ext1 mutant fibroblasts still produced small amounts of heparan sulfate. The domain structure of the mutant heparan sulfate was conserved, and the disaccharide composition was similar to that of wild type heparan sulfate. However, a dramatic difference was seen in the polysaccharide chain length. The average molecular sizes of the heparan sulfate chains from wild type and Ext1 mutant embryonic fibroblasts were estimated to be around 70 and 20 kDa, respectively. These data suggest that not only the sulfation pattern but also the length of the heparan sulfate chains is a critical determinant of normal mouse development.

Heparan sulfate proteoglycans (HSPGs) 1 are composed of one or more linear HS polysaccharide chains covalently bound to selected serine residues in a protein core. HSPGs are ubiquitously present at cell surfaces and in the extracellular matrix, affecting a variety of biological processes, including specific signaling pathways (1)(2)(3). The constituent sulfated HS polysaccharide chains exert their biological functions by inter-acting with a multitude of proteins in differential and specific fashion. The critical roles of HS structure in normal growth and development have become evident using model animals including Drosophila and mouse (4,5).
The biosynthesis of HSPG is a multistep process involving the concerted action of several enzymes and includes (a) formation of the polypeptide core; (b) assembly of the polysaccharide-protein linkage region; (c) generation of a polymer consisting of alternating GlcA and GlcNAc residues; and (d) a series of modification reactions by which N-and O-sulfate groups are introduced and a fraction of the GlcA residues are C5-epimerized to L-iduronic acid (IdoUA) units. The HS-protein linkage region consists of the tetrasaccharide: glucuronic acid-galactose-galactose-xylose (GlcA␤1-3Gal␤1-3Gal␤1-4Xyl), where xylose is linked to a serine residue in the protein core. Separate enzymes catalyze the incorporation of the different sugar units (6). Transfer of a single GlcNAc unit to the tetrasaccharide linkage region initiates HS assembly. The alternating addition of GlcA and GlcNAc, from their respective UDP-derivatives, to the nonreducing terminus of the growing polymer thus forms the (GlcA-GlcNAc) n HS chain, which is modified through sulfation and GlcA C5-epimerization reactions (reviewed in Refs. [7][8][9]. The regulation of this process, as required to generate specific saccharide structures, is poorly understood but apparently involves arrays of enzyme isoforms that differ with regard to substrate specificities and kinetic properties. Hereditary multiple exostoses is an autosomal dominant skeletal disorder characterized by the presence of cartilage capped bony outgrowths mainly located at the juxtaepiphyseal region of the long bones (10). Genetic linkage studies in families with hereditary multiple exostoses disclosed two main loci, EXT1 on chromosome 8q24.1 and EXT2 on chromosome 11p11-p12 (11,12). Although linkage to another locus, EXT3, on chromosome 19p has been described (13), the EXT3 gene has not yet been isolated, and genetic linkage to this locus has only been detected in a few pedigrees to date. Together with EXT1 and EXT2, three additional members, designated EXTL1, EXTL2, and EXTL3, form the exostosin (EXT) gene family (14 -16). The EXT as well as EXTL proteins show sequence homology especially in the C-terminal regions. However, there is no evidence that defects in the EXTL genes result in hereditary multiple exostoses. Despite extensive genetic characterization, the function of the EXT proteins remained unknown until 1998, when two independent studies revealed the connection between HS-synthesizing glycosyltransferases and the EXT gene family (17,18). An ϳ70-kDa enzyme committed to HS polymer formation was isolated from bovine serum, and analysis of the corresponding cDNA showed that the isolated protein was identical to EXT2 (17). Transfection of EXT1 into an HS-deficient L-cell mutant restored the ability to synthesize HS, suggesting that chain elongation may involve both EXT1 and EXT2 (18).
Recently, more detailed functional data on the role of the EXT protein family members in HS biosynthesis has been provided by several studies. EXT1 and EXT2 form a heterooligomeric complex in vivo that is accumulated in the Golgi apparatus (19,20). The Golgi-localized EXT1-EXT2 protein complex possesses substantially higher glycosyltransferase activity than EXT1 alone, suggesting that this complex represents the biologically relevant form of the HS polymer modification unit (20,21). Mutational analysis of CHO cells defective in EXT1 demonstrated that the GlcA-transferase catalytic domain is localized to the N-terminal region of the EXT1 protein (22). No mutations in EXT1 affecting only GlcNAc-transferase activity have yet been found. Less is known about the domain organization of EXT2 that has been shown to have both GlcAand GlcNAc transferase activities when expressed in COS-7 cells and in yeast (17,21). Interestingly, the three other members of the EXT gene family, EXTL1, EXTL2, and EXTL3 also encode glycosyltransferases, which are likely to be involved in HS biosynthesis (23,24). Furthermore, mutations in the Drosophila orthologs of EXT2 and EXTL3 (sister of tout-velu (sotv) and brother of tout velu (botv), respectively) seriously diminish HS synthesis (25,26).
An indispensable role for EXT1 in HS polymerization has been suggested by the observations that HS synthesis is abolished in Ext1-deficient embryonic stem cells (27) and in Drosophila bearing a mutation in tout-velu, the orthologue of human EXT1 (28). Ext1-deficient mice generated by gene targeting fail to gastrulate, lack HS, and die by embryonic day 8.5 (27). In contrast, Ext1 mutant mice generated by the gene trap method (Ext1 Gt(pGT2TMpfs)064Wcs , Ext Gt/Gt ) survive to embryonic day 14.5 and have a much less severe phenotype (29). 2 The longer survival time of the Ext1 Gt/Gt mice suggested that the gene trap insertion created a hypomorphic allele of Ext1. To study the effect of the gene trap disruption of Ext1 on HS polymerization, we have studied the structural properties of HS produced by fibroblasts from Ext Gt/Gt embryos. Our results show that sulfated HS chains are still produced, albeit much less than normally. The decrease in HS was mainly due to reduced chain length. The overall HS sulfation pattern was moderately affected by the mutation. HS is required for growth factor signaling and diffusion, and the embryonic lethal phenotype of Ext1 Gt/Gt may reflect the inability of the mutant HS to mediate some of the HS-dependent steps in embryogenesis. These findings may have important implications for our understanding of the structural requirements of HS in regulating growth factor signaling.  (30). Polysaccharide standards for estimations of molecular mass values were derived from heparin (3.3 and 8.6 kDa) and from hyaluronan (19,30,43, and 210 kDa) as described (31). The generation of Ext1-deficient mice was performed as described (29).

Materials-Heparitinase
Northern Blot Analysis-Total RNA from mouse embryonic fibroblasts was obtained using the RNeasy Mini kit (Qiagen) according to the manufacturer's instructions. Denatured RNA was electrophoresed on 1.2% agarose-formaldehyde, transferred to a nylon membrane (Hybond-N; Amersham Pharmacia Biotech), and hybridized at three occa-sions with probes labeled with [␣-32 P]dCTP using Ready-To-Go DNA labeling beads (Amersham Biosciences). Each lane of the blot contained 10 g of total RNA. The probes used were a 0.55-kb fragment (nt 862-1407) of the mouse Ext1 cDNA, a 0.53-kb fragment (nt 1938 -2464) of the mouse Ext1 cDNA, and a 0.39-kb fragment (nt 777-1163) of the mouse Ext2 cDNA. Unincorporated [␣-32 P]dCTP was removed with a Microspin S-300 HR column (Amersham Biosciences). The filters were hybridized at 68°C in ExpressHyb solution (Clontech) first with the ϳ550-bp Ext1 fragment, subsequently with the 530 bp Ext1 fragment, and then with the mouse Ext2 cDNA probe.
[GlcNAc-GlcA] n -aMan R acceptors were prepared by digestion of GlcA-[GlcNAc-GlcA] n -aMan R with bovine liver ␤-D-glucuronidase (Sigma). To assay GlcNAc and GlcA transferase activities, protein preparations from wild type or Ext1 Gt/Gt fibroblast cultures were incubated for 1 or 2 h at 37°C with 14 C-labeled UDP-GlcA and [GlcNAc-GlcA] n -aMan R oligosaccharide acceptors (measuring GlcA-transferase activity) or with 14 C-labeled UDP-GlcNAc and GlcA-[GlcNAc-GlcA] n -aMan R acceptors (measuring GlcNAc-transferase activity). Labeled oligosaccharides were isolated by gel chromatography on a Sephadex G-25 superfine column (1 ϫ 22 cm) eluted with 0.2 M NH 4 HCO 3 , and quantified by scintillation counting. The protein concentration of cell homogenate was quantified using the BCA Protein Assay Reagent (Pierce) with bovine serum albumin as a standard.
Isolation of Metabolically Labeled HS-Fibroblast cells obtained from wild type and homozygous Ext1 Gt/Gt animals (embryonic day 11.5) were maintained in Dulbecco's modified Eagle's medium (Invitrogen) containing 20% fetal calf serum and 1% penicillin G-streptomycin. At passage 3 or 4, cells were labeled with 200 Ci/ml Na 2 35 SO 4 , 50 Ci/ml [ 3 H]galactose, or 50 Ci/ml [ 3 H]glucosamine for 48 h. The medium was removed from each cell culture, and the cells were washed two times with ice-cold PBS. The cell layer was solubilized with 50 mM Tris-HCl, pH 8.0, containing 150 mM NaCl and 1% Triton X-100 under gentle rocking at 4°C for 4 h and centrifuged. To release the O-linked sugars from the protein core, the supernatants were treated with 0.5 M NaBH 4 , 0.5 M NaOH at 4°C overnight. The reactions were terminated by the addition of 4 M acetic acid to decompose excess NaBH 4 and neutralized with 1 M NaOH. After centrifugation at 13,000 rpm for 10 min, the supernatants were applied to a Sephadex G-50 superfine column (1 ϫ 145 cm) eluted with 0.2 M NH 4 HCO 3 . Fractions eluting at the void volume were recovered, galactosaminoglycans were eliminated by chondroitinase ABC digestion, and resistant HS chains were recovered by DEAE-Sephacel chromatography (33). Alternatively, after centrifugation, the supernatants were collected, treated with 0.5 M NaOH at 4°C overnight, and neutralized, and labeled glycosaminoglycans were isolated by anion exchange chromatography using DEAE-Sephacel chromatography. Galactosaminoglycans were eliminated by digestion with chondroitinase ABC, and resistant HS chains were recovered by gel filtration. The purity of labeled HS was determined by the sensitivity to enzyme digestion with a mixture of heparinase and heparitinase.
Embryonic fibroblasts obtained from wild type and homozygous Ext1 Gt/Gt animals were immortalized by transformation with recombinant retroviruses encoding simian virus 40 large T antigen (34). A producer cell line, ⌿ 2 , containing integrated copies of defective ectropic Moloney murine leukemia virus provirus, was cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and 1% penicillin G-streptomycin. The conditioned medium from confluent ⌿ 2 cell culture was collected and mixed with Polybrene (Aldrich) to a final concentration of 8 g/ml. Embryonic fibroblasts at ϳ60% confluence in 2 O. G. Kelly and W. C. Skarnes, manuscript in preparation. a 10-cm dish were incubated with 2 ml of the culture medium containing 0.2 ml of the viral supernatant at 37°C for 2 h, and then 8 ml of the growth medium was added. Colonies of transformed cells started to appear after 1.5-2 weeks. Immortalized cells were labeled with 50 Ci/ml [ 3 H]glucosamine or 200 Ci/ml Na 2 35 SO 4 for 48 h, and the resultant labeled HS was isolated as described above.
Labeled HSPGs present at the cell surface and in the extracellular matrix were isolated after labeling of immortalized cells with 50 Ci/ml [ 3 H]glucosamine. After 24 h of labeling, the culture medium was removed and frozen. The cells were washed twice with PBS and then treated with 1 mg/ml trypsin (Sigma) for 10 min at 37°C. The trypsin was then neutralized with 2 mg/ml trypsin inhibitor (Sigma). Cells were then centrifuged at 1,000 rpm for 10 min at 4°C. The cell pellets were frozen for future use. The supernatants were treated with 0.5 M NaOH, and labeled HS chains were recovered by gel filtration as described above.
Enzymatic Digestions-Enzymatic digestion with a mixture of heparinase and heparitinase or chondroitinase ABC was carried out in 100 mM Tris-HCl buffer, pH 7.4, containing 3.3 mM CaCl 2 and 0.1 mg/ml bovine serum albumin or 50 mM Tris-HCl buffer, pH 8.0, containing 30 mM sodium acetate and 0.1 mg/ml bovine serum albumin, respectively. The digestion products were analyzed by gel filtration chromatography on a column of Sephadex G-50 superfine (0.55 ϫ 95 cm) using 0.5 M NH 4 HCO 3 as the eluent.
Structural Analysis of HS-The size of labeled HS chains prepared from wild type and Ext1 Gt/Gt mouse embryonic fibroblasts was analyzed by gel chromatography on a Superose 6 column (Amersham Biosciences) eluted with 50 mM Tris-HCl, pH 8.0, containing 1.0 M NaCl, at a flow rate of 0.5 ml/min. Fractions were collected at 1-min intervals, and the radioactivity was monitored by liquid scintillation counting.
Nitrous acid treatment at pH 1.5 was conducted with 0.5 M HNO 2 followed by reduction of the products with NaBH 4 , yielding an aMan R residue at the reducing end (32). Under these conditions, N-sulfated glucosamine units are selectively attacked, but N-acetylated glucosamine residues are not affected. The degradation products were sizefractionated by gel chromatography on a Sephadex G-15 column (1 ϫ 180 cm) eluted with 0.2 M NH 4 HCO 3 or on Superdex 30 in 0.2 M NH 4 HCO 3 . For essentially complete depolymerization to disaccharides, labeled HS was chemically N-deacetylated by treatment with 70% (w/v) aqueous hydrazine (Fluka) containing 1% (w/v) hydrazine sulfate at 96°C for 4 h, and the product was treated with nitrous acid at pH 1.5 and at pH 3.9 (35). Recovered labeled disaccharides were analyzed on a Whatman Partisil-10 SAX column eluted at a rate of 1 ml/min with KH 2 PO 4 solutions of stepwise increasing concentration as described in the legend to Fig. 6.
Alternatively, the disaccharide composition of [ 3 H]glucosamine-labeled HS was determined after depolymerization with a mixture of heparinase and heparitinase and separation on an amino-bound silica PA-03 HPLC column. Individual disaccharides were identified by co-chromatography with disaccharide standards as described previously (36).

HS Formation in Ext1 Mutant
Embryos-It has been demonstrated that EXT1 plays an essential role in the HS polymerization process and that the formation of HS is abolished in mice with a targeted mutation in Ext1 (27). However, it has never been clarified whether the HS chain elongation process stops after the formation of the HS protein linkage region (GlcNAc-GlcA-Gal-Gal-Xyl) or at some later stage of HS synthesis. We have generated mutant mice with a lethal insertional mutation in the gene encoding Ext1 (Ext1 Gt/Gt ) (29) using a gene trap strategy and generated fibroblasts from embryonic day 11.5 wild type and gene trap mutated Ext1 Gt/Gt embryos. 2 To examine the effect of the gene trap mutated Ext1 on the sequential addition of GlcA and GlcNAc contiguous to the HS-protein linkage region, wild type fibroblasts and Ext1 Gt/Gt fibroblasts were metabolically labeled with [ 3 H]galactose. During glycosaminoglycan biosynthesis, galactose is primarily incorporated into the linkage region. Galactose is also converted to UDP-GlcA and incorporated into the uronic acid units of glycosaminoglycans although less efficiently. Cell extracts were treated with alkali to liberate Oglycans including glycosaminoglycan-related polysaccharides from core proteins, and samples were subjected to gel filtration chromatography on a Sephadex G-50 column, to separate di-, tetra-, hexa-, and octasaccharide from large polysaccharides (data not shown). Both wild type and Ext1 Gt/Gt fibroblasts incorporated very small amounts of [ 3 H]galactose (less than 1% of isolated labeled O-linked sugars) at the elution position of a linkage pentasaccharide. Instead, the majority of the radioactivity was incorporated into larger polysaccharides containing material susceptible to specific HS-degrading enzymes, indicating that HS chain elongation had occurred also in the mutant cells.
To determine whether the Ext1 Gt/Gt mice indeed synthesize HS chains, embryonic fibroblasts were metabolically labeled with [ 35 S]sulfate or [ 3 H]glucosamine. The yields of [ 35 S]sulfatelabeled glycosaminoglycans were comparable between wild type cells (2.3 ϫ 10 5 cpm/mg protein) and Ext1 Gt/Gt cells (2.5 ϫ 10 5 cpm/mg protein). Isolated 35 S-labeled polysaccharides from wild type and Ext1 Gt/Gt fibroblasts were subjected to selective degradation of chondroitin sulfate (chondroitinase ABC digestion) or HS (nitrous acid deamination), and the respective degradation products were analyzed by gel chromatography (Fig. 1). The mutant sample contained material susceptible to nitrous acid treatment, showing that Ext1 Gt/Gt embryonic fibroblasts synthesize sulfated HS chains. The proportion of HS (i.e. material resistant to digestion by chondroitinase ABC but susceptible to nitrous acid deamination) decreased from ϳ65% of the 35 S-labeled glycosaminoglycans in the wild type cells to ϳ35% in the Ext1 Gt/Gt cells. When wild type and Ext1 Gt/Gt cells were labeled with [ 3 H]glucosamine, the incorporation of label into glycosaminoglycans was slightly reduced in Ext1 Gt/Gt fibroblasts (2.65 ϫ 10 5 cpm/mg protein) as compared with the wild type cells (3.1 ϫ 10 5 cpm/mg protein). The proportion of HS in Ext1 Gt/Gt and wild type cells was determined as 12 and 55%, respectively, of the 3 H-labeled glycosaminoglycans (data not shown). Thus, the amount of 3 H-labeled HS synthesized by Ext1 Gt/Gt cells was ϳ18% of the HS produced by wild type cells.
The presence of HS in Ext1 Gt/Gt cells was confirmed by indirect immunofluorescent staining using the HepSS-1 antibody that recognizes sulfated HS chains (37) and subsequent confocal microscopic examination. A dense staining for HS was observed (data not shown).
Analysis of Ext1 Gene Disruption-To investigate whether the gene trap insertion leads to a deficiency of the normal Ext1 transcript, the mRNA level was analyzed by Northern blotting. The gene trap mutation is inserted in the intron between exons 1 and 2 in the Ext1 gene (Fig. 2). As a result of the insertion, a fusion transcript is generated containing part of the Ext1 mRNA (exon 1) fused to the gene trap vector sequence (Fig. 2). To characterize the Ext1 mRNA expression, two different probes A and B (corresponding to nucleotides 862-1407 and 1938 -2464 in the mouse Ext1 cDNA sequence) were hybridized to RNA from wild type and Ext1 Gt/Gt cells (Fig. 3). A band of ϳ3.5 kb was detected in wild type fibroblasts (Fig. 3, A and B), consistent with the previously reported size of the Ext1 mRNA (38). In the Ext1 Gt/Gt cells, the putative fusion transcript (Ͼ9 kb) was detected by probe A (Fig. 3A). However, no transcript was detected using probe B, indicating that the gene was disrupted prior to the sequence corresponding to probe B (Fig. 3B). No differences in the expression levels of Ext2 were observed between wild type and Ext1 Gt/Gt fibroblasts. To use a more sensitive method, an RT-PCR amplification experiment was performed using primers F1 and R1 shown in Fig. 2. The expected 500-bp product, corresponding to nucleotides 1939 -2468, was generated in the wild type but initially not in the Ext1 Gt/Gt sample (data not shown). However, using the PCR product mixture from the initial PCR as a template in an additional PCR resulted in the amplification of a 500-bp product also from the Ext1 Gt/Gt sample, suggesting that the mutant cells still produced small amounts of wild type transcript.
HS GlcNAc-and GlcA-transferase Activities-The alternating addition of GlcNAc and GlcA from their respective UDPsugars generates HS chains. The two transferase activities were separately assayed at several different occasions using freshly prepared cell extracts from wild-type and EXT Gt/Gt embryonic fibroblast cell cultures (see "Experimental Procedures"). The specific GlcNAc-and GlcA-transferase activities, expressed on a cellular protein basis, varied between different cell extracts. However, even with variations between single extracts, the ratio of the activities in Ext1 Gt/Gt versus wild type cells was repeatedly ϳ0.15 for GlcNAc transfer and 0.1 for GlcA transfer ( Table I).
Effect of Ext1 Gene Trap Mutation on HS Structure-The molecular size of the HS chains was analyzed by gel chromatography on Superose 6. A clearly reduced chain length of HS in Ext1 Gt/Gt fibroblasts was evident, with a complete lack of longer chains (Fig 4, A and B). Compared with the elution positions of size-defined hyaluronan and heparin standard polysaccharides, the peak elution positions of the rather broad peaks correspond to ϳ50 -100 kDa for wild type HS and ϳ20 kDa for Ext1 Gt/Gt HS.
To obtain a stable source of embryonic fibroblasts, wild type and Ext1 Gt/Gt fibroblasts were immortalized by transformation with recombinant retroviruses encoding simian virus 40 large T antigen (see "Experimental Procedures") (34). Analysis of [ 35 S]sulfate-(data not shown) or [ 3 H]glucosamine-labeled glycosaminoglycans from immortalized wild type and Ext1 Gt/Gt fibroblasts gave results virtually identical with those from the primary cells both regarding the level of HS synthesis and chain length (Fig. 4C).
During development, numerous secreted factors bind HS, and HS is suggested to play a vital role in morphogen signaling and/or transport. Since these molecules encounter HS associated with the cell membrane or the extracellular matrix, we analyzed trypsin-released HS from the surface of wild type and Ext1 Gt/Gt fibroblasts. Similar to the results obtained with HS isolated from whole cell extracts, the Ext1 Gt/Gt HS chains lacked the longer HS chains present on wild type cells (Fig. 4D).
The N-substitution patterns of HS from wild type and Ext1 Gt/Gt cells were analyzed by treatment of [ 3 H]glucosaminelabeled 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 (Fig. 5). Calculations based on the distribution of radiolabel between oligosaccharides of different sizes indicated that ϳ40% of the glucosamine units in HS from Ext1 Gt/Gt cells and from wild type cells were N-sulfated. Similar analysis using labeled immortalized cells gave comparable results. Thus, the gene trap mutation did not significantly change the N-sulfation content of the polysaccharide.
Detailed information regarding the distribution of sulfate groups was obtained by analysis of the disaccharides generated by low pH HNO 2 /NaBH 4 treatment of 35  (see "Experimental Procedures"). The resultant disaccharides were separated by anion exchange HPLC (Fig. 6) into three different mono-O-sulfated and one di-O-sulfated species. The HS from wild type and Ext1 Gt/Gt embryos yielded similar patterns (Fig. 6, Table II). There was, however, a small increase in mono-6-O-sulfated disaccharides in the mutant HS compared with the wild-type.  The overall disaccharide composition of HS from wild type and Ext1 Gt/Gt cells was determined following N-deacetylation and complete deaminative cleavage of [ 3 H]glucosamine-labeled HS to disaccharides (see "Experimental Procedures"). Separation of the products by anion exchange HPLC again revealed increased 6-O-sulfation in HS from Ext1 Gt/Gt cells (Table III). To confirm the sulfation patterns, we determined the disaccharide composition of medium-derived [ 3 H]glucosamine-labeled HS after digestion with a mixture of heparinase and heparitinase. The resultant disaccharides were analyzed by HPLC on an amine-bound silica PA-03 column. The composition of the medium-derived mutant HS was very similar to that of the wild type, and contrary to the analysis of cell-derived HS, no differences in 6-O-sulfation were observed (Table III). Although the 6-O-sulfation of Ext1 Gt/Gt HS may differ to some extent, taken together, our results indicated that the overall domain organization and disaccharide pattern is similar to the wild type HS.

DISCUSSION
Targeted mutagenesis or gene trap mutations of genes encoding enzymes essential for normal HS biosynthesis result in severe developmental abnormalities (27, 39 -43). The most severe effect was obtained by the targeted deletion of Ext1, which results in failure of the embryos to gastrulate, demonstrating the importance of HS for early signaling events (27).
In this study, we describe the molecular properties of HS in an Ext1 Gt/Gt gene trap mutant mouse. The Ext1 Gt/Gt mice die around embryonic day 14, and in contrast to the previous report of the complete loss of HS in embryonic stem cells from Ext1-deficient mice generated by the gene targeting method (27), our gene trap mice still produce some HS. The longer survival time for the Ext1 gene trap mouse may reflect the ability of its short but apparently normally modified HS chains to mediate some HS-dependent signaling events.
We are intrigued by the fact that the gene trap mice still produce HS. If the hetero-oligomeric complex of EXT1 and EXT2 is the unique enzyme complex responsible for chain elongation, the complete disruption of the Ext1 or Ext2 gene would result in the interruption of HS biosynthesis after the formation of the linkage pentasaccharide. We did detect very small amounts of normal Ext1 transcript in the Ext1 Gt/Gt fibroblasts by PCR. The presence of HS in the gene trap mouse could be a result of residual transferase activity in the Nterminal part of the protein or indicate that splicing around the large gene trap vector has generated a small amount of normal Ext1 transcript. 3 The mutation mainly affected HS chain length, although some reduction in the number of chains may also have occurred. Our findings raise intriguing questions regarding the regulation of chain elongation. By what mechanism does the mutation in Ext1 result in shorter chains and not in fewer chains of normal length? How does HS formation due to defective Ext1 influence HS-protein interactions? HSPGs may also regulate morphogen gradients by binding to ligands and thereby influence their diffusion. A, HS chains from wild type mice are able to bind to different signaling molecules and thereby facilitate the clustering of the high affinity receptor, and a signal is transduced into the cells. B, the truncated HS chains in the Ext1 Gt/Gt mouse may still functionally interact with some growth/differentiation factors but not with other factors. Growth/differentiation factors are shown as circles and diamonds. Highly sulfated regions (N-sulfated GlcN repeats) in HS chains are indicated by black dots.  H]Glucosamine-labeled samples were completely degraded to disaccharides either chemically (cell-HS, N-deacetylation, deamination at pH 3.9 and pH 1.5, reduction of products with NaBH 4 ) or enzymatically (medium-HS, lyase digestion with a mixture of heparinase and heparitinase) that were analyzed by anion exchange HPLC (see "Experimental Procedures"). The HS polysaccharide chains are constructed on the HSprotein linkage region by the alternating addition of GlcNAc and GlcA to the nonreducing end of the growing chain. Coexpression studies of the EXT1 and EXT2 proteins have been shown to result in increased glycosyl transferase activities, and the polymerase reaction has been ascribed to a hetero-oligomeric complex of EXT1 and EXT2 (20,21). Homozygous null embryos of Ext2 have similar developmental abnormalities as the Ext1-deficient mice, and diminished Ext2 expression in cultured mammalian cells blocks HS synthesis, further implicating the requirement of both proteins in the chain elongation process (9,44).
The other members of the exostosin family, the EXT-like proteins (EXTL1-3) have all been demonstrated to have Glc-NAc transferase activity (23,24). EXTL2 possibly catalyzes the addition of the first GlcNAc unit onto the GlcA-Gal 2 -Xyl-Ser linkage region, thus committing the process toward generation of an HS polymer. If instead an N-acetyl-D-galactosamine is first added to the GlcA residue, the same linkage structure will serve to initiate chondroitin sulfate formation (6,45). EXTL1 transfers GlcNAc to growing HS chains, whereas EXTL3 seems to have dual functions catalyzing the transfer of GlcNAc both to the HS-protein linkage region and to the growing chain (24). GlcA-transferase activity has so far only been demonstrated for EXT1 and EXT2 proteins.
The gene trap cassette is inserted between exons 1 and 2 (Fig. 2). Exon 1 encodes 43% of total amino acids (i.e. 320 amino acids of 746), indicating that the mutant EXT1 protein still contains almost half of the N-terminal side of the protein. This part of the protein has been proposed to contain the GlcAtransferase activity (22). The C-terminal half of the EXT1 protein is conserved among the EXT family members and is postulated to harbor the catalytic site of the GlcNAc-transferase activity. However, so far it has not been shown that the two activities constitute two independent sites. In the Ext1 Gt/Gt fibroblasts, small but detectable levels of both GlcNAc and GlcA transferase activities were observed. The GlcA transferase activity may remain in the chimeric protein of EXT1/␤-Gal/ neo and/or EXT2, whereas the GlcNAc transferase activity, which was significantly higher than the GlcA transferase activity, may in part be the result of EXT2, EXTL1, and/or EXTL3 activities.
Several signaling molecules, including members of the fibroblast growth factor, Wingless/Int, and Hedgehog families, require HSPG at the cell surface for optimal signaling (1). Whereas biochemical studies using cell culture systems and in vitro binding assays have provided data concerning the minimal growth factor binding epitope on HS, the influence of HS chain length for in vivo growth factor binding in tissues is not known. Mutation in the Drosophila homologue of Ext1 (toutvelu) was shown to affect long distance signaling by Hedgehog in the wing imaginal disc but not to affect fibroblast growth factor or Wingless signaling pathways (46). Staining of heparitinase-treated tout-velu mutant embryos with the 3G10 antibody, which recognizes the heparitinase-generated unsaturated uronic acid linked to the core protein, resulted in a reduced diffuse staining, indicating the presence of some HS (46). It is tempting to speculate that the tout-velu mutant embryos synthesize HS similar to that found in our Ext1 Gt/Gt mice. However, structural analysis of glycosaminoglycans from tout-velu mutant larvae showed no detectable levels of HS (28). The HS chains of HSPGs might play critical roles as co-receptors of signaling molecules and as regulators of morphogen gradients (47). Based on the findings presented in this study, one might speculate that the HS chains produced in the Ext1 Gt/Gt mice are insufficient in their functions as regulators of some of the HS-dependent cell-cell signaling in development. The mutant chains could be too short to bind the protein ligand, or the chains may not be able to properly present the growth factor to its high affinity kinase receptor (Fig. 7).
In summary, our results show that not only the sulfation pattern of HS but also its chain length is important for early embryonic development. These findings may have important implications for the understanding of HS signaling and also how mutations in EXT1 or EXT2 correlate with the development of exostoses. Further analysis such as growth factor signaling assays and growth factor phosphorylation assays on wild type and Ext1 Gt/Gt mutant embryonic fibroblasts will probably provide insight into the structural requirements of HS in regulating growth factor binding and signaling.