Biosynthesis of Heparin/Heparan Sulfate

Glucuronyl C5-epimerases catalyze the conversion of d-glucuronic acid (GlcUA) tol-iduronic acid (IdceA) units during the biosynthesis of glycosaminoglycans. An epimerase implicated in the generation of heparin/heparan sulfate was previously purified to homogeneity from bovine liver (Campbell, P., Hannesson, H. H., Sandbäck, D., Rodén, L., Lindahl, U., and Li, J.-p. (1994) J. Biol. Chem. 269, 26953–26958). The present report describes the molecular cloning and functional expression of the lung enzyme. The cloned enzyme contains 444 amino acid residues and has a molecular mass of 49,905 Da. N-terminal sequence analysis of the isolated liver enzyme showed this species to be a truncated form lacking a 73-residue N-terminal domain of the deduced amino acid sequence. The coding cDNA insert was cloned into a baculovirus expression vector and expressed in Sf9 insect cells. Cells infected with recombinant epimerase showed a 20–30-fold increase in enzyme activity, measured as release of 3H2O from a polysaccharide substrate containing C5-3H-labeled hexuronic acid units. Furthermore, incubation of the expressed protein with the appropriate (GlcUA-GlcNSO3) n substrate resulted in conversion of ∼20% of the GlcUA units into IdceA residues. Northern analysis implicated two epimerase transcripts in both bovine lung and liver tissues, a dominant ∼9-kilobase (kb) mRNA and a minor ∼5-kb species. Mouse mastocytoma cells showed only the ∼5-kb transcript. A comparison of the cloned epimerase with the enzymes catalyzing an analogous reaction in alginate biosynthesis revealed no apparent amino acid sequence similarity.

Heparin and heparan sulfate are complex sulfated glycosaminoglycans composed of alternating glucosamine and hexuronic acid residues. The two polysaccharides are structurally related but differ in composition, such that heparin is more heavily sulfated and shows a higher ratio of L-iduronic acid (IdceA) 1 to D-glucuronic acid (GlcUA) units (2,3). Heparin is produced by connective tissue-type mast cells, whereas heparan sulfate has a ubiquitous distribution and appears to be expressed by most cell types. The biological roles of heparin and heparan sulfate are presumably largely due to interactions of the polysaccharides with proteins, such as enzymes, enzyme inhibitors, extracellular matrix proteins, growth factors/cytokines, and others (2)(3)(4)(5). The interactions tend to be more or less selective/specific with regard to carbohydrate structure and thus depend on the amounts and distribution of the various sulfate groups and hexuronic acid units. Notably, IdceA units are believed to generally promote binding of heparin and heparan sulfate chains to proteins, due to the marked conformational flexibility of these residues (6).
Heparin and heparan sulfate are synthesized as proteoglycans (for reviews, see Refs. 3-5 and 7). The process is initiated by glycosylation reactions that generate saccharide sequences composed of alternating GlcUA and GlcNAc units (8) covalently bound to peptide core structures. The resulting (GlcUA␤1,4-GlcNAc␣1,4) n disaccharide repeats are modified, probably along with chain elongation, by a series of enzymatic reactions that is initiated by N-deacetylation and N-sulfation of GlcNAc units, continues through C5-epimerization of GlcUA to IdceA residues, and is concluded by the incorporation of O-sulfate groups at various positions. The N-deacetylation/N-sulfation step has a key role in determining the overall extent of modification of the polymer chain since the GlcUA C5-epimerase as well as the various O-sulfotransferases all depend on the presence of N-sulfate groups for substrate recognition. While the GlcNAc N-deacetylation and N-sulfation reactions are both catalyzed by the same protein (9), the isolation and molecular cloning of N-deacetylase/N-sulfotransferases from mouse mastocytoma and rat liver implicated two distinct forms of the enzyme (10 -12). The two enzyme types differ with regard to kinetic properties, with the mastocytoma enzyme being more efficient in introducing N-sulfate groups in the growing polymer (13).
A GlcUA C5-epimerase was purified to apparent homogeneity (ϳ1 million-fold) from bovine liver (1). This report describes the molecular cloning, using a bovine lung cDNA library, as well as the functional expression of this enzyme.

EXPERIMENTAL PROCEDURES
Peptide Purification and Sequencing-The 52-kDa epimerase protein (ϳ1 g), purified from a detergent extract of bovine liver by chromatography on O-desulfated heparin-Sepharose, red-Sepharose, phenyl-Sepharose, and concanavalin A-Sepharose (1), was subjected to direct N-terminal sequencing using an Applied Biosystems Model 470A protein sequenator equipped with an on-line Model 120 phenylthiohydan-* This work was supported by Swedish Medical Research Council Grants 2309 and 6525, European Commission Grant BIO4-CT95-0026, the Konung Gustaf V:s 80-Årsfond, and Polysackaridforskning AB (Uppsala, Sweden). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF003927.
§ To whom correspondence should be addressed. Tel.: 46-18-4714241; Fax: 46-18-4714209; E-mail: Jin-Ping.Li@medkem.uu.se. 1 The abbreviations used are: IdceA, L-iduronic acid; GlcUA, D-glucu-toin analyzer (14). Another sample (ϳ1 g) was applied to preparative (12%) SDS-polyacrylamide gel and was then transferred to a polyvinylidene difluoride membrane. After staining the membrane with Coomassie Brilliant Blue, the enzyme band was excised. Half of the material was subjected to direct N-terminal sequence analysis, whereas the remainder was digested with Lys-C (0.0075 unit; Waco) in the presence of 1% reduced Triton X-100, 10% acetonitrile, and 100 mM Tris-HCl, pH 8.0. The generated peptides were separated on a reverse-phase C 4 column, eluted at a flow rate of 100 l/min with a 6-ml 10 -70% acetonitrile gradient in 0.1% trifluoroacetic acid, and detected with a Waters Model 990 diode array detector. Selected peptides were then subjected to sequence analysis as described above.
Probes for Screening-Total RNA was isolated from bovine liver by the LiCl/urea/SDS method (15). Single-stranded cDNA was synthesized by incubating ϳ5 g of bovine liver total RNA (denatured at 65°C for 3 min) in 10 mM Tris-HCl, pH 8.3, with 1 unit of RNase inhibitor (Perkin-Elmer), 1 mM each dNTP, 5 M random nucleotide hexamer, and 1.25 units of murine leukemia virus reverse transcriptase (Perkin-Elmer). The mixture was kept at 42°C for 45 min and then at 95°C for 5 min. Degenerated oligonucleotide primers were designed based on the amino acid sequence determined for one of the internal peptides derived from the purified epimerase (see Table I). Single-stranded bovine liver cDNA was subjected to PCR together with 100 pmol of primers 1 (sense) and 3 (antisense) in a total volume of 100 l of 10 mM Tris-HCl, pH 9.0, containing 0.1% Tween 20, 6 mM MgCl 2 , 1 mM each dNTP, and 2.5 units of Taq polymerase (Pharmacia Biotech Inc.). The reaction products were separated on a 12% polyacrylamide gel. An ϳ100-bp band was cut out from the gel and reamplified using the same PCR conditions. After an additional polyacrylamide gel electrophoresis, the product, a 108-bp fragment, was isolated, inserted into a pUC119 plasmid, and sequenced. The DNA fragment cleaved from the plasmid was labeled with [ 32 P]dCTP (NEN Life Science Products) using a random-primed DNA labeling kit (Boehringer Mannheim).
Screening of cDNA Library-A bovine lung cDNA library constructed in a gt10 vector (CLONTECH) was screened with the 108-bp PCR fragment as hybridizing probe. The nitrocellulose filters were prehybridized in 6 ϫ SSC (1 ϫ SSC is 0.15 M NaCl and 0.015 M sodium citrate, pH 7.0) and 5 ϫ Denhardt's solution containing 0.1% SDS and 0.1 mg/ml denatured salmon DNA for 2 h at 65°C. Hybridization was carried out at 42°C in the same solution containing 32 P-labeled probe for 16 -18 h. The filters were washed two times with 2 ϫ SSC and 0.5% SDS and two times with 0.5 ϫ SSC and 0.1% SDS at the same temperature. Finally, the cDNA phage library was subjected to PCR amplification using an epimerase cDNA-specific primer (5Ј-GCTGATTCTTT-TCTGTC-3Ј, antisense) together with gt10 forward or reverse primers (CLONTECH).
Subcloning and Sequencing of cDNA Inserts-cDNA inserts, isolated by preparative agarose gel elctrophoresis (15) after EcoRI restriction cleavage of recombinant bacteriophage DNA, were subcloned into a pUC119 plasmid. The complete nucleotide sequence was determined independently on both strands using the dideoxy chain termination reaction either with 35 S-dATP and the modified T7 DNA polymerase (Sequenase Version 2.0 DNA sequencing kit, U. S. Biochemical Corp.) or with the ALF™ system (Pharmacia Biotech Inc.). DNA sequences were compiled and analyzed using the DNASTAR™ program (Lasergene).
Polyclonal Antibodies and Immunodetection-A peptide corresponding to residues 77-97 of the deduced epimerase amino acid sequence (see "Results") was chemically synthesized by Åke Engström (Department of Medical and Physiological Chemistry, Uppsala University, Uppsala, Sweden) and was then conjugated to ovalbumin using glutaraldehyde (16). A rabbit was immunized with 240 g of the peptide conjugates emulsified in complete Freund's adjuvant. After six booster injections (each with 240 g of conjugated peptide), blood was collected, and the antiserum was recovered. IgG was isolated from the antiserum by affinity chromatography on a protein A-Sepharose column (Pharmacia Biotech Inc.), and used for immunoblotting.
Samples of GlcUA C5-epimerase were separated under denaturing conditions by 12% SDS-polyacrylamide gel electrophoresis and were then transferred to a nitrocellulose membrane (Hybond™ ECL). ECL immunoblotting was performed according to the protocol of the manufacturer (Amersham Corp.). Briefly, the membrane was first treated with blocking agent, then incubated with purified antibody, and finally incubated with the peroxidase-labeled anti-rabbit antibody. After adding the ECL reagent, the light emitted by the chemical reaction was detected by exposure to Hyperfilm™ ECL for 30 -60 s.
Northern Blot Hybridization-Bovine liver and lung total RNA was prepared by the LiCl/urea/SDS method according to Sambrook et al. (15), and mouse mastocytoma total RNA was extracted as described by Chomczynski and Sacci (17) from mastocytoma cells established in culture after passage through an ascites stage (18). Total RNA from each tissue (ϳ20-g samples) was denatured in 50% (v/v) formamide, 5% formaldehyde, and 20 mM Mops buffer, pH 7.0, at 65°C for 5 min. The denatured RNA was separated by electrophoresis on 1.2% agarose containing 5% (v/v) formaldehyde and was then transferred to a Hybond N ϩ nylon membrane (Amersham Corp.). The RNA blot was prehybridized in ExpressHyb hybridization solution (CLONTECH) at 65°C for 1 h and subsequently hybridized in the same solution with a [ 32 P]dCTPlabeled DNA probe of 2460 bp (bp 13-2472 in Fig. 1). The membrane was washed in 2 ϫ SSC and 0.5% SDS at the same temperature for 2 ϫ 15 min and in 0.5 ϫ SSC and 0.5% SDS for 2 ϫ 15 min. The membrane was exposed to Kodak x-ray film at Ϫ70°C for 24 h.
In Vitro Translation-The 3-kb GlcUA C5-epimerase clone inserted into a pcDNA3 expression vector (Invitrogen) was linearized at the 3Ј-end by cleavage with XbaI. In vitro translation was carried out with a Linked T7 transcription-translation system (Amersham Corp.) according to the instructions of the manufacturer. The mRNA generated by incubation of 0.5 g of linearized plasmid DNA with a T7 polymerase transcription mixture (total volume, 10 l; 30°C for 15 min) was mixed with an optimized rabbit reticulocyte lysate containing 50 Ci of [ 35 S]methionine (total volume, 50 l) and further incubated at 30°C for 1 h. A sample (5 l) of the product was subjected to 12% SDS-polyacrylamide gel electrophoresis. The gel was directly exposed to Kodak x-ray film. After exposure, the applied protein molecular standards (LMW molecular calibration kit, Pharmacia Biotech Inc.) were visualized by staining the gel with Coomassie Brilliant Blue.
Expression of GlcUA C5-Epimerase-GlcUA C5-epimerase was expressed using a BacPAK8™ baculovirus expression system (CLON-TECH) according to the instructions of the manufacturer. Two oligonucleotides, one at the 5Ј-end of the cDNA clone (5Ј-GCCACCATGTCCA-AGCTGAATTCTCATAGCTATTCCAAAG-3Ј, sense) and the other at the 3Ј-end of the coding sequence (5Ј-CTAGTTGTGCTTTGCCC-3Ј, antisense), were used to PCR amplify the C5-epimerase cDNA clone. The resulting fragment was cloned into the BacPAK8 vector. Sf9 insect cells, maintained in Grace's insect medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum and penicillin/streptomycin, were then cotransfected with the C5-epimerase construct along with viral DNA. Control transfections were performed with a ␤-glucuronidase cDNA construct included in the expression kit and a mouse cDNA coding for the GlcNAc N-deacetylase/N-sulfotransferase implicated in heparin biosynthesis (12,13). Single plaques of each cotransfected recombinant were picked and propagated. Two 60-mm Petri dishes of Sf9 cells were infected by each recombinant virus stock and incubated at 27°C for 5 days. The cells from one dish were used for total RNA extraction followed by Northern blot analysis, performed as described above. Cells from the other dish were lysed in a buffer of 100 mM KCl, 15 mM EDTA, 1% Triton X-100, and 50 mM HEPES, pH 7.4, containing 1 mM phenylmethylsulfonyl fluoride and 10 g/ml pepstatin A. Supernatants of cell lysates as well as conditioned media were analyzed for epimerase activity. Protein contents of the cell lysates were estimated by the method of Bradford (19) or by the BCA reagent procedure (20).
Demonstration of GlcUA C5-Epimerase Activity-Epimerase activity was assayed by detecting 3 H release by using a biphasic liquid scintillation procedure essentially as described by Enzymatic conversion of D-glucuronic to L-iduronic acid was demonstrated using the metabolically 1-3 H-labeled substrate (N-deacetylated and N-sulfated capsular polysaccharide from E. coli K5) and the analytical procedure described by Campbell et al. (1). A sample (ϳ20 g, 200,000 cpm 3 H) of the modified polymer was incubated with 250 l of cell lysate in a total volume of 300 l of epimerase assay buffer at 37°C for 6 h. The incubation was terminated by heating at 100°C for 5 min. The sample was mixed with 50 g of carrier heparin and reacted with nitrous acid at pH 1.5 (21), followed by reduction of the products with NaBH 4 . The resultant hexuronylanhydromannitol disaccharides were recovered by gel chromatography on a column (1 ϫ 200 cm) of Sephadex G-15 in 0.2 M NH 4 HCO 3 , lyophilized, and subjected to paper chromatography on Whatman No. 3MM paper in ethyl acetate/acetic acid/ water (3:1:1).

RESULTS
Generation of Probe and Screening of cDNA Library-Amino acid sequence data for the ϳ52-kDa protein were obtained by digesting highly purified epimerase with lysine-specific protease, followed by separation of the generated peptides on a reverse-phase column. The five most prominent peptides were isolated and subjected to amino acid sequencing ( Table I). One of the peptides (peptide 1) was found to correspond to the N-terminal sequence of the purified protein. The sequence of the largest peptide obtained (peptide 5 in Table I) was used to design two sense and one antisense degenerate oligonucleotide primers, as shown in Table I. A DNA probe was produced by PCR using primers 1 and 3 with bovine liver cDNA as template. The resultant ϳ100-bp DNA fragment was purified by polyacrylamide gel electrophoresis, reamplified using the same primers, and finally isolated by electrophoresis. The identity of the product was ascertained by "nested" PCR, using primers 2 and 3, which yielded the expected ϳ60-bp fragment (data not shown). Moreover, sequencing of the larger (108 bp) DNA fragment gave a deduced amino acid sequence identical to that of the isolated peptide (Table I).
The 108-bp PCR product was labeled with [ 32 P]dCTP and used for screening of a bovine lung gt10 library. One hybridizing clone, containing a 3-kb insert, was identified. Repeated screening of the same library yielded two additional positive clones, both of which were of smaller size. Subsequent sequencing showed both of the latter clones to be contained within the 3-kb species (data not shown). The 3-kb clone was sequenced through both strands and found to contain altogether 3073 bp; an additional 12-bp sequence was added at the 5Ј-end through characterization of a separate clone obtained by PCR amplification of the phage library (see "Experimental Procedures").
Characterization of cDNA and Predicted Protein Structure-The total cDNA sequence identified, in all 3085 bp, contains an open reading frame from the first in-frame ATG codon, corresponding to 444 amino acid residues (Fig. 1). Since no in-frame stop codon is present upstream of the assigned initiating ATG codon, it cannot be excluded that the identified cDNA still lacks the complete 5Ј-terminal sequence. The coding region (1332 bp) is flanked toward the 3Ј-end by a larger (1681 bp) noncoding segment. The deduced amino acid sequence corresponds to a 49,905-Da polypeptide. All five peptides isolated after endopeptidase digestion (Table I) were recognized in the primary struc-ture deduced from the cDNA (Fig. 1). One of these peptides (peptide 1) is identical to the N terminus of the isolated liver protein. This peptide was found to match residues 74 -86 of the deduced polypeptide sequence. The enzyme isolated from bovine liver thus represents a truncated form of the native protein.
Generation of mRNA from an expression vector inserted with the 3-kb cDNA clone followed by incubation of the product with rabbit reticulocyte lysate in the presence of [ 35 S]methionine resulted in the formation of a distinct labeled protein with an estimated molecular mass of ϳ50 kDa (Fig. 2). This product was recognized in immunoblotting (data not shown) by polyclonal antibodies raised against a synthetic peptide corresponding to residues 77-97 (see Fig. 1) of the deduced amino acid sequence. The same antibodies also reacted with the isolated ϳ52-kDa bovine liver protein (data not shown). These observations establish that the 3-kb cDNA is derived from the transcript that encodes the isolated ϳ52-kDa bovine liver protein.
The cDNA structure indicates the occurrence of three potential N-glycosylation sites (Fig. 1). A potential transmembrane region is underlined in Fig. 1. The predicted protein contains two cysteine residues, only one of which occurs in the isolated (truncated) protein.
Functional Expression of GlcUA C5-Epimerase-The cDNA corresponding to nucleotides 1-1407, with an added ATG codon at the 5Ј-end, was cotransfected with baculovirus into Sf9 insect cells. The expressed protein thus is larger, by a 25-amino acid sequence, than the predicted 444-amino acid-long protein (Fig. 1). In two separate experiments, the lysates from cells infected with the same recombinant epimerase virus stock showed Ͼ10-fold higher enzyme activities, on a mg of protein basis, than the corresponding fractions from cells infected with control recombinant virus stock (Table II). The conditioned media of cells infected with recombinant epimerase showed 20 -30-fold higher enzyme activities than the corresponding fractions from cells infected with control plasmid virus stock. Transfections with cDNA encoding other enzymes, such as a ␤-glucuronidase or the mouse mastocytoma GlcNAc N-deacetylase/N-sulfotransferase involved in heparin biosynthesis (12), did not significantly increase the epimerase activity beyond control levels. Notably, the higher 3 H 2 O release recorded for control samples as compared with heat-inactivated expressed enzyme (Table II) suggests that the insect cells constitutively produce endogenous C5-epimerase, although the amount of enzyme activity is less than that in mammalian cells (data not shown). Furthermore, isolation of the 3 H-containing fraction from the C5-labeled polysaccharide substrate after incubation with the recombinant epimerase followed by evaporation of the isolated fraction confirmed that 3 H 2 O was indeed formed during incubation (data not shown).
The polysaccharide substrate used for routine assays of epimerase activity was obtained by chemically N-deacetylating and N-sulfating the capsular polysaccharide (GlcUA␤1,4-GlcNAc␣1,4) n ) of E. coli K5 that had been grown in the presence of [5-3 H]glucose. The data in Table II thus reflect the release of 3 H 2 O from 5-3 H-labeled GlcUA units in the modified polysaccharide, due to enzyme action (22,23). More direct evidence for the actual conversion of GlcUA to IdceA residues was obtained by incubating the expressed enzyme with an analogous substrate, obtained following incubation of E. coli K5 with [1-3 H]glucose. This substrate will retain the label through the epimerization reaction and can therefore be used to demonstrate the formation of IdceA-containing disaccharide units. Following incubation with the recombinant epimerase, 21% of the glucuronic acid residues were converted to IdceA, as dem- onstrated by paper chromatography of disaccharide deamination products (Fig. 3). The composition of the incubated polysaccharide thus approached the equilibrium ratio of IdceA to GlcUA, previously determined to be ϳ3:7. FIG . 1. Nucleotide sequence and predicted amino acid sequence of C5-epimerase. The predicted amino acid sequence is shown below the nucleotide sequence. The numbers on the right indicate the nucleotide residues and the amino acid residues (boldface italic) in the   FIG. 2. In vitro transcription-translation. The epimerase cDNA was inserted into a pcDNA3 expression vector and linearized with XbaI at the 3Ј-end. It was then subjected to in vitro transcription-translation in a rabbit reticulocyte lysate system in the presence of [ 35 S]methionine as described under "Experimental Procedures." The translation product of epimerase cDNA (Epi) has a molecular mass of ϳ50 kDa by comparison with the low-molecular mass protein standard. A control sample of ␤-galactosidase (C; 118 kDa), expressed in the same system, is shown for comparison.

TABLE II Expression of hexuronyl C5-epimerase in Sf9 cells
Sf9 cells (1 ϫ 10 6 in 4 ml of medium) were seeded in 60-mm Petri dishes and allowed to attach for 3 h at 27°C. The medium was then removed, and 200 l of recombinant virus stock was added to infect the cells. After incubation at room temperature for 1 h, the virus suspension was aspirated off, and 4 ml of medium was added to each dish. The cells were incubated at 27°C for 5 days. The medium was then transferred into a sterile tube and centrifuged. The supernatant was saved for analysis ("Medium"), whereas the pellet was combined with the cells collected from the dish. After washing twice with phosphate-buffered saline, the cells were lysed with 300 l of homogenization buffer as described under "Experimental Procedures." Aliquots (25 l) of cell lysate and medium were assayed for epimerase activity. Enzyme activity is expressed as release of 3 H from the K5 polysaccharide substrate per h of incubation and per mg of protein or ml of medium. The data are the means Ϯ S.D. of three independent assays. and mouse mastocytoma were analyzed by hybridization with a 2460-bp DNA fragment from the epimerase cDNA clone as a probe. Both bovine liver and lung gave identical hybridization patterns, with a dominant band at ϳ9 kb and a weak ϳ5-kb band (Fig. 4). By contrast, the mastocytoma RNA contained only the ϳ5-kb transcript.

DISCUSSION
The structural diversity of heparin and heparan sulfate is generated through selective modification of the (GlcUA␤1,4-GlcNAc␣1,4) n polymer formed during the initial stage of the biosynthetic process (3,5,7). Most of these reactions, i.e. Ndeacetylation, N-sulfation, and the various O-sulfation reactions, can be reproduced, albeit in a poorly controlled fashion, through chemical modification (24,25). In fact, the conversion of GlcUA to IdceA units, in the intact polymer, is the only reaction that cannot as yet be conducted without the aid of a biological catalyst, i.e. the C5-epimerase. The molecular cloning and functional expression of this enzyme are important steps toward elucidating the mechanism of an intriguing reaction that has so far been elusive.
As discussed above, the native molecular size of 49,905 Da of the epimerase remains somewhat uncertain due to the lack of an in-frame stop codon upstream of the assigned initiating ATG codon (Fig. 1). Expression of this protein in the baculovirus system with 25 amino acids added at the N terminus, as encoded by the nucleotide sequence upstream of the assigned initiating ATG codon, resulted in a catalytically active enzyme. However, the extra 25 amino acids and the adjacent region corresponding to residues 1-73 of the deduced protein (including the putative transmembrane domain; see Fig. 1) are not needed for activity. The N-terminal proline residue of the enzymatically active epimerase protein initially isolated from bovine liver thus corresponds to residue 74 of the deduced polypeptide sequence shown in Fig. 1.
The molecular mass of a protein corresponding to residues 74 -444 of the deduced sequence would be 41,935 Da. The apparent molecular size of the isolated bovine liver epimerase, estimated by SDS-polyacrylamide gel electrophoresis, was ϳ52 kDa (1). Following deglycosylation by treatment with peptide N-glycosidase, the protein appeared as an ϳ47-kDa band, thus still larger than the calculated ϳ42 kDa. This discrepancy may be due to the presence of additional bound carbohydrate, such as O-linked oligosaccharides. Besides, it is also recalled that proteins differ in their ability to bind SDS, resulting in differences in mobility in SDS-polyacrylamide gel electrophoresis (26).
The structural difference between heparin and heparan sulfate is best explained in terms of the various biosynthetic polymer modification reactions, which are consistently more extensive in the case of heparin (3,5,7). The N-acetylglucosaminyl N-deacetylase/N-sulfotransferase is a key regulatory enzyme, which initiates polymer modification and commits the subsequent series of reactions toward formation of either heparin or heparan sulfate. Recent studies have revealed the occurrence in most cells of two N-deacetylase/N-sulfotransferase transcripts, ϳ4 and ϳ8 kb in size (11,12), that encode distinct enzyme forms with different catalytic properties and that are derived from different genes. 3 Transcription of GlcUA C5-epimerase appears to involve similar diversity, as Northern blotting of bovine lung and liver (using C5-epimerase cDNA as a probe) resulted in a predominant ϳ9-kb hybridizing band and a faint ϳ5-kb band, whereas murine mastocytoma RNA contained the ϳ5-kb transcript only (Fig. 4). RNAs from a number of other cell types, such as human embryonic kidney cells, Chinese hamster ovary cells, and COS-1 cells, all reproduced the lung/liver pattern (data not shown). Whether also the C5epimerase occurs in genetically and/or catalytically distinct forms remains to be elucidated. Interestingly, the C-terminal portion of a hypothetical Caenorhabditis elegans protein (27) shows amino acid sequence similarities to the cloned GlcUA C5-epimerase (data not shown).
Enzymes catalyzing C5-inversion of hexuronic acid residues in polysaccharides occur also in other biosynthetic systems. Thus, C5-epimerases converting D-mannuronic acid to L-guluronic acid residues in alginate have been described in brown algae and in bacteria (28). A C5-epimerase isolated from Azotobacter vinelandii generates alginates composed of guluronic acid blocks in addition to mannuronic acid blocks and mixed mannuronic acid/guluronic acid blocks, whereas an epimerase from Pseudomonas aeruginosa yields a product with mannuronic acid blocks and mannuronic acid/guluronic acid blocks, 3 M. Kusche-Gullberg, I. Eriksson, and L. Kjellén, personal communication. but no guluronic acid blocks (29,30). The two types of enzymes showed no significant amino acid sequence homology (31). Contrary to the GlcUA C5-epimerase(s) involved in the biosynthesis of heparin/heparan sulfate, which show no requirement for divalent cations, the algal and bacterial mannuronate C5-epimerases all depend on Ca 2ϩ for catalytic activity. Finally, an additional GlcUA C5-epimerase has been implicated in the biosynthesis of dermatan sulfate and thus also acts on a glycosaminoglycan substrate, i.e. chondroitin (GlcUA␤1,3-GalNAc␤1,4) n (32,33). Notably, this latter enzyme again requires divalent cations for activity. The two GlcUA C5-epimerases committed to glycosaminoglycan biosynthesis show no cross-reactivity with regard to polysaccharide substrates (chondroitin versus N-sulfoheparosan) and thus are believed to be distinct entities. Further information regarding the relationship between these two enzymes will emerge when also the dermatan epimerase has been subjected to molecular cloning.