INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Heparan sulfate (HS)¹ proteoglycans occur on cell surfaces and in the extracellular matrix of loose connective tissue and in basement membranes. In connective tissue-type mast cells, the polysaccharide is stored intracellularly in granules. The mast cell polysaccharide is traditionally referred to as heparin but is composed of the same structural units as HS. HS has been implicated in cell adhesion processes, cytokine action, regulation of enzyme activities and in the maintenance of the permeability of basement membranes (reviewed in Refs. 1-3). The polysaccharide chain carries many negative charges that enables it to interact electrostatically with a wide variety of molecules. However, some of the interactions are highly specific (4-6).

During biosynthesis, the HS-chains are polymerized by sequential addition of glucuronic acid and N-acetylglucosamine units. Concomitantly, the polymer is modified through a series of reactions that include N-deacetylation and N-sulfation of glucosamine residues, epimerization of glucuronic acid to iduronic acid, and finally O-sulfation at various positions (3, 6, 7). The extent of these reactions varies, giving rise to enormous structural heterogeneity. Compared with HS, heparin is more highly sulfated and has a higher content of N-sulfated glucosamine and iduronic acid (3). The first modifying enzyme, a combined N-deacetylase/N-sulfotransferase has a prime regulatory role in determining the overall structure and charge density of HS as N-deacetylation followed by N-sulfation are required for all subsequent modifications (1).

Proteins, with features typical of Golgi proteins, expressing N-deacetylase/N-sulfotransferase activities have been purified and cloned from a mouse mastocytoma (8-10) and from rat liver (11-14). Recently, the cDNA sequence of a human counterpart to the rat liver transcript (15) as well as the predicted structure of the corresponding gene was published (16).

The enzyme purified from the mouse mastocytoma is encoded by an ~4-kb mRNA while the rat liver transcript contains ~8 kb. The major part of the protein sequences have ~70% identity. However, the most N-terminal 80 amino acids show a low level of structural homology, with approximately 30% identical amino acids. In addition, both the 5' and the 3' noncoding regions seem unrelated (9, 10, 14). In this study, we demonstrate that the two transcripts are widely distributed in mouse tissues and that they are encoded by different genes. The structure of the gene encoding the 4-kb transcript is reported and shown to share many features with the previously characterized human gene for the 8-kb transcript.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Southern Blot Analysis-- Isolated genomic DNA from adult mouse liver was digested with XbaI, HindIII, EcoRI, and BglII. Thirty-five µg of genomic DNA was used for each digest. The samples were fractionated by 1% agarose gel electrophoresis in 0.9 M Tris, 0.9 M boric acid, 0.020 M EDTA. After electrophoresis, the gel was soaked in 0.4 M NaOH for 30 min before transfer to nylon membrane (Hybond-N⁺, Amersham Pharmacia Biotech). The membranes were hybridized overnight at 65 °C with ³²P-labeled probes in 5× SSC (SSC is 0.15 M NaCl, 0.015 M sodium citrate buffer, pH 7.0), 5× Denhardt's solution, 2% SDS containing 100 µg/ml salmon sperm DNA, and washed in 0.1× SSC, 0.5% SDS at 65 °C. Four different cDNA probes, labeled with [-³²P]dCTP (NEN Life Science Products) in the random primed DNA labeling kit (Boehringer Mannheim), were used for hybridization. Probes A and C recognize the coding and 3'-untranslated region, respectively, of the 4-kb transcript, whereas probes B and D hybridize with the coding and 3'-untranslated region of the 8-kb transcript. Probe A is a HinfI cDNA fragment of the 4-kb mouse transcript (nt 1412-1900 in Fig. 1 of Ref. 10). Probe B was generated by RT-PCR of rat liver RNA using the same conditions as described for the 504-bp PCR-product (10), with a sense and an antisense primer corresponding to nt 1639-1660 and 2025-2004, respectively, in the rat liver cDNA-sequence (14). To the 5'-end of the primers was added a dCCGAATTC extension to create an EcoRI site in the PCR-product. Probe C is a 356-bp cDNA-fragment, corresponding to the 3'noncoding end of the 4 kb mouse mastocytoma transcript (nt 2951-3307; Ref. 10). Probe D is a HinfI fragment of the 504-bp PCR product generated by RT-PCR of rat liver RNA (10), corresponding to nt 3396-3679 in Ref. 14.

RNA Purification-- Hepatocytes were isolated after collagenase perfusion of a rat liver and purified by density gradient centrifugation in Percoll (17). Total RNA was isolated from these cells and from mouse mastocytoma (18) and mouse liver tissue using the LiCl/urea/SDS method (19).

Northern Blot Analysis-- A single filter containing poly(A)-selected RNA from several adult mouse tissues (CLONTECH) was hybridized at 65 °C in 5× SSPE (1× SSPE is 0.15 M NaCl, 10 mM NaH₂PO₄, 1 mM EDTA, pH 7.4), 10× Denhardt's solution, 2% SDS, 100 µg/ml salmon sperm DNA, at three occasions with probes labeled with [-³²P]dCTP (NEN Life Science Products) as described above. The probes used were a 356-bp cDNA-fragment, identical to the 356-bp cDNA fragment (Probe C) used in Southern blotting (see above), a full-length cDNA corresponding to an mRNA encoding a protease specific for connective tissue-type mast cells, kindly provided by Dr. Lars Hellman, University of Uppsala (MMCP-4; Ref. 20), and a 2-kb cDNA recognizing -actin mRNA (CLONTECH).

Cloning and cDNA Analysis of Mouse Liver 8-kb Transcript-- An RT-PCR-based strategy was employed to clone the mouse homologue of the rat liver N-deacetylase/N-sulfotransferase. Primers were selected in regions corresponding to conserved sequences in the rat (14) and the human (15) transcripts. One pair of primers (1) corresponding to nt 1-18 (sense) and 1199-1182 (antisense), respectively, in the rat liver cDNA-sequence (14) were used to amplify the 5' half of the cDNA. An overlapping clone extending to the 3'-untranslated region was obtained using two primers (2) corresponding to nt 1147-1163 (sense) and nt 2717-2700 (antisense), respectively, in Ref. 14. Random hexamer primed cDNA was synthesized from 1 µg of mouse liver total RNA using the GeneAmp RNA PCR kit (Perkin-Elmer). The PCR reaction for primer pairs 1 was performed under conditions of: 1 cycle of 94 °C for 1 min, 30 cycles each of 94 °C for 15 s, 50 °C for 15 s, 72 °C for 1 min, and a final extension at 72 °C for 2 min. The amplified 1199-bp fragment was subcloned into the pUC 119 vector and sequenced. Conditions used to amplify the second part of the cDNA using primer pairs 2 were: 1 cycle of 94 °C for 1 min, 35 cycles each of 94 °C for 20 s, 45 °C for 20 s, 72 °C for 2 min plus 1 s for each cycle, and a final extension at 72 °C for 5 min. The amplified product (1571 bp) was subcloned into the pCR II vector (Invitrogen) and sequenced. The sequence identities between the mouse liver and rat liver 8-kb transcript are 96% for both the 1199 and 1571-bp fragments (EBI/GenBank^TM accession number AF049894). To obtain a full-length clone, a unique NdeI site at nt 1178 was used. The insert in the pCR II plasmid was excised with NdeI and EcoRI and ligated into the same site in the pUC vector containing the PCR product of primer pairs 1. The sequence around the NdeI site in the resulting full-length clone was verified by nucleotide sequence analysis as described above, using synthetic oligonucleotide primers. Sequences were analyzed with the aid of the Lasergene software package (DNAStar Inc., Madison).

Isolation of Genomic Clones-- A mouse FIX II genomic library (Stratagene, La Jolla, CA) was screened using as probes a mixture of a 1.7-kb cDNA fragment (10), corresponding to a region entirely within the coding region of the 4-kb transcript, and "probe B" (see above), corresponding to 366 nt of the coding region of the rat 8-kb transcript. The probes were labeled with [-³²P]dCTP (NEN Life Science Products) as described above. About 1× 10⁶ plaques were screened, and approximately 60 clones were plaque purified. Of those, two different clones for the 4-kb transcript and 4 different clones for the 8-kb transcript were obtained. The genomic clones were further analyzed by restriction mapping and Southern blotting using as probes specific cDNA oligonucleotide primers (17 nt) that had previously been used for nucleotide sequence analysis of the mouse mastocytoma 4-kb cDNA (10). Genomic DNA fragments obtained from the phage DNA after digestion with SacI and/or BamHI were subcloned into the plasmid vector pUC 119 for characterization. The gene encoding the 4-kb transcript was further characterized, using other restriction fragments of the  clones that were subcloned into pUC 119 and sequenced. The size of each intron was determined by nucleotide sequence analysis.

RNA Expression Analysis Using RT-PCR-- One µg of total RNA from each tissue in a total volume of 20 µl was used for the generation of single-stranded cDNA in the GeneAmp RNA PCR kit. After dilution to 40 µl, 1 µl of each sample was used for PCR in a Rapidcycler (Idaho Technology), with Taq DNA polymerase (MBI-Fermenta) and TaqStart Antibody (CLONTECH). The sense primer corresponds to nt 1301-1317 in the sequence of the mouse mastocytoma 4-kb transcript (10), identical to nt 1593-1609 in the sequence of the rat liver 8-kb transcript (14). Also, the antisense primer corresponds to identical regions in the mouse 4-kb (nt 2299-2282; Ref. 10) and in the rat 8-kb transcript (nt 2594-2577; Ref. 14). The 10 µl reactions were done in heat-sealed glass capillaries in 50 mM Tris buffer, pH 8.3, containing 3 mM MgCl₂ and 0.25 mg/ml bovine serum albumin. After an initial hold for 2 min at 95 °C, each cycle included denaturation at 95 °C for 0 s, annealing at 56 °C for 0 s, and extension at 72 °C for 30 s. After 20, 25, or 30 cycles, the capillaries were emptied into Eppendorf tubes, and to the tubes were added 10 µl of "restriction enzyme mix, " containing 4 units of EcoRI or HindIII, 50 ng of pUC 119, 20 µg/ml bovine serum albumin in the appropriate restriction enzyme buffer (MBI-Fermenta). After incubation for 2 h at 37 °C and addition of 4 µl of sample dye, 20 µl of the samples were electrophoresed in 1.5% agarose gels. Samples subjected to 30 PCR cycles and subsequently incubated without any restriction enzyme were also analyzed. The plasmid vector pUC 119 served as a control of complete restriction enzyme cleavage, which was checked after ethidium bromide staining of the gel. After blotting to Hybond-N⁺, the nylon membrane was pre-hybridized at 42 °C for 30 min in ExpressHyb hybridization solution (CLONTECH) and incubated in the same solution at 42 °C for 1 h with a ³²P-labeled oligonucleotide IE 16 (dAACTATGGAAATGACCG). This sequence is found in both transcripts from rat and mouse, and hybridizes with the intact rat and mouse 1-kb products as well as the 829-nt EcoRI and the 668-nt HindIII fragment generated from the 1-kb band corresponding to the rat and mouse 8-kb transcript, respectively. The membrane was finally washed with 2× SSC, 0.05% SDS twice for 10 min at 42 °C, before exposure to x-ray film.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Southern blots of genomic DNA isolated from mouse liver was hybridized with ³²P-labeled cDNA-probes from the coding and from the 3'-untranslated regions of the 4-kb mouse and the 8-kb rat transcript, respectively (Fig. 1). While the coding regions of the 4-kb mouse and the 8-kb rat transcript are fairly similar (~70% identity at the amino acid level), the untranslated regions are unrelated (9, 10). The hybridization with mouse DNA of the probe corresponding to the 3'-untranslated region of the 8-kb rat transcript (Fig. 1D) clearly demonstrates that also the mouse genome contains sequence information for this transcript. The cDNA probe corresponding to the coding region of the 8-kb transcript, hybridized to other genomic fragments (Fig. 1B), demonstrating the presence of restriction sites for XbaI, HindIII, EcoRI, and BglII in this gene. Notably, the hybridization pattern was distinct from that obtained with the probes corresponding to the 4-kb transcript (Fig. 1, A and C), indicating that the two transcripts are encoded by different genes.

View larger version (83K): [in this window] [in a new window]   Fig. 1.   Southern blot analysis of mouse genomic DNA. Mouse genomic DNA was cut with XbaI (X), HindIII (H), EcoRI (E), and BglII (B). The resulting fragments were separated on a 1% agarose gel. After blotting, the filters were hybridized with probes recognizing the coding and 3'-untranslated region of the 4-kb transcript (panels A and C, respectively) and with probes hybridizing with the coding and 3'-untranslated region of the 8-kb transcript (panels B and D).

The expression of the 8-kb transcript in mouse liver was established using RT-PCR followed by nucleotide sequence analysis of the isolated clones. The nucleotide sequence of the coding region (GenBank^TM accession number, AF049894) was highly homologous to that of the corresponding region in the rat 8-kb transcript (96% identity). On the protein level, only a few amino acids differed between the two species (Fig. 2). According to Northern blotting of different mouse tissues, the 8-kb transcript was widely distributed, with lung, liver, and kidney containing the highest amounts (Fig. 3).

View larger version (117K): [in this window] [in a new window]   Fig. 2.   Alignment of the N-deacetylase/N-sulfotransferase amino acid sequences. Deduced amino acid sequence of mouse liver (ML) N-deacetylase/N-sulfotransferase 8-kb transcript and alignment with mouse mastocytoma (MM) 4-kb and rat liver (RL) 8-kb transcript homologues. The numbers indicate the amino acid residues of the proteins. All residues in the sequence corresponding to the 8-kb mouse liver transcript (ML 8 kb), as well as identical residues in homologs, are boxed. Positions where amino acids are identical in two of the sequences are shaded. Gaps introduced to optimize alignment, using the Clustal algorithm with PAM250 residue weight table (Lasergene software, DNAStar), are shown as dashes ().

View larger version (50K): [in this window] [in a new window]   Fig. 3.   Distribution of the 8-kb N-deacetylase/N-sulfotransferase mRNA in adult mouse tissues. A mouse multiple tissue Northern blot was hybridized with a 216-bp 32P-labeled probe recognizing the coding region of the 8 kb transcript (A) and a 2-kb cDNA recognizing -actin (B). The filter was washed at 50 °C in 0.1× SSC, 0.1% SDS. Sizes of RNA-markers in kilobases are indicated.

Isolation of Two Mouse N-Deacetylase/N-Sulfotransferase Genes-- To isolate genomic clones containing the N-deacetylase/N-sulfotransferase genes, two cDNA probes corresponding to parts of the coding region of the mastocytoma and the rat liver cDNA transcripts, receptively (see "Experimental Procedures"), were mixed and used for screening of a 129/sv mouse genomic DNA library. Several positives clones were isolated and subsequently identified by Southern hybridization with cDNA fragments and oligonucleotides specific for the 4 and the 8-kb transcript, respectively (see "Experimental Procedures"). Two clones, 2:3 and 4:3 in Fig. 4, were derived from the gene encoding the 4-kb transcript, but these clones did not cover the 5'-part of the gene. An ~400-bp SacI fragment from the most 5'-terminal region of the 4:3 genomic clone was therefore used as a probe in another screening of the library. This approach yielded two more clones (8:2 and 10:1). Together with clones 2:3 and 4:3, approximately 30 kb of genomic sequence were now covered (Fig. 4).

View larger version (7K): [in this window] [in a new window]   Fig. 4.   Schematic representation of the organization of the mouse gene encoding the 4-kb N-deacetylase/N-sulfotransferase transcript. The four genomic clones used in determining the sequence of the N-deacetylase/N-sulfotransferase gene are shown (top). The clones covered together approximately 30 kb extending in the 3' direction (- - -). Below is a schematic view of the gene where exons are represented by boxes while horizontal lines depict introns. The portions of the exons that contain protein coding sequences are filled. The box representing the left-most exon is open to indicate that the 5'-boundary of this exon has not been localized.

Structure of the Gene Encoding the 4-kb Transcript-- A crude map of the gene was constructed based on hybridization of restriction fragments with oligonucleotides, previously used as specific primers in nucleotide sequence analysis of cDNA corresponding to the mouse mastocytoma 4-kb transcript (10). Genomic fragments generated by digestion of the clones with SacI and/or BamHI were subcloned into pUC 119. The exon-intron organization and the size of exons and introns were determined by nucleotide sequence analyses of these fragments. The data available suggest that the mouse 4-kb N-deacetylase/N-sulfotransferase gene consists of 14 exons distributed within 8 kb of genomic DNA. However, it cannot be excluded that additional introns may exist in the extreme 5'- and 3'-ends of the gene. The exons contain between 88 and 1322 nucleotides, and all of the donor and acceptor splicing sites agree to the consensus sequences for eukaryotic genes (21). These data are summarized in Fig. 4 and in Table I. Exon 1 contains part of the 5'-untranslated region, whereas the remainder of the 5'-untranslated sequence is located in exon 2, which contains the translation start site. This exon, which is much larger than any of the other exons, also encodes both the short N-terminal cytoplasmatic domain and the transmembrane part of the protein in addition to a large portion of the luminal domain of the N-deacetylase/N-sulfotransferase. While most of the intron junctions are of the 0 type (splicing occurs between codons), introns 3 and 8 are type 1 introns (where splicing occurs after the first base of the codon). Introns 9 and 12, finally, are of type 2 (splicing occurs after the second base of the codon).

Expression of the 4-kb N-Deacetylase/N-Sulfotransferase Transcript-- It has generally been assumed that the 4-kb transcript encodes the N-deacetylase/N-sulfotransferase involved in heparin biosynthesis and thus that this transcript is confined to mast cells. We previously noted a low expression of the 4-kb transcript in rat liver (10). Since mast cells in the liver capsule may have been the source of this transcript, we decided to look for the presence of the 4-kb transcript in purified hepatocytes using a PCR approach. In this method, a primer pair is used that will amplify a 1-kb band from cDNA generated by reverse transcription of either of the 4 or 8-kb transcripts². (The nucleotide sequence of the 4- and 8-kb transcripts is identical in the regions where the primers anneal.) After the PCR, the 1-kb band is incubated with a restriction enzyme that cleaves the band derived from the 8-kb transcript but leaves the band generated from the 4-kb transcript intact. For RNA derived from rat tissues, EcoRI is used, while HindIII cleaves the 1-kb band generated from the mouse 8-kb transcript (See Fig. 5 where exon 8 of the gene encoding the 4-kb transcript is compared with the corresponding exon in the mouse gene for the 8-kb transcript.). As shown in Fig. 6, most of the 1-kb PCR product obtained using rat hepatocyte cDNA as template is cleaved by EcoRI, but small amounts of the 4-kb transcript are present in these cells, represented by the weak intact 1-kb band (Fig. 6A). Mouse liver, in contrast, appears to contain similar amounts of the 4- and 8-kb transcripts (Fig. 6B), and mouse mastocytoma contains almost exclusively the 4-kb transcript (Fig. 6C). Small amounts of the HindIII cleavage product, not evident in the figure, can be detected after additional PCR cycles (data not shown). In Fig. 6, the large difference in expression of N-deacetylase/N-sulfotransferase mRNA in mastocytoma and liver is also evident.

View larger version (27K): [in this window] [in a new window]   Fig. 5.   Comparison of the nucleotide sequence of corresponding exons in the mouse genes encoding the 4- and 8-kb transcripts. Exon sequences are in capital letters, while intron sequences are in lowercase letters. The HindIII site, present in the exon from the gene encoding the 8-kb transcript but absent from exon 8 of the gene encoding the 4-kb transcript, is underlined. y = t or c.

View larger version (45K): [in this window] [in a new window]   Fig. 6.   Expression of 4- and 8-kb transcripts in rat hepatocytes, mouse liver, and mouse mastocytoma. After reverse transcription of total RNA, PCR was performed with primers selected to give a 1-kb band when cDNA corresponding to both 4- and 8-kb transcripts from either murine or rat tissues was used as template (see "Experimental Procedures"). After 20, 25, and 30 cycles, the samples were incubated with EcoRI (rat hepatocyte samples) or HindIII (mouse mastocytoma and liver samples). This treatment selectively cleaves the 1-kb band corresponding to the 8-kb transcript. The enzyme-treated samples, uncleaved controls, and size markers were electrophoresed in 1.5% agarose and subsequently blotted to nylon membrane. After hybridization with a 32P-labeled oligonucleotide, recognizing intact mouse and rat 1-kb product as well as the 829-nt EcoRI and 668-nt HindIII fragments, the membrane was washed and exposed to x-ray film.

View larger version (50K): [in this window] [in a new window]   Fig. 7.   Distribution of the 4-kb N-deacetylase/N-sulfotransferase mRNA in adult mouse tissues. A mouse multiple tissue Northern blot was hybridized with 32P-labeled probes (see "Experimental Procedures") recognizing the 3' noncoding region of the 4-kb transcript (the filter was washed at 50 °C in 2× SSC, 0.05% SDS) (A), mRNA for the mast cell-specific protease MMCP-4 (the filter was washed at 50 °C in 1× SSC, 0.05% SDS) (B), -actin mRNA (the filter was washed at 65 °C in 0.1 × SSC, 0.05% SDS) (C). Sizes of RNA-markers in kilobases are indicated.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

Southern blotting of mouse genomic DNA (Fig. 1) showed that the two different N-deacetylase/N-sulfotransferase enzymes, encoded by the 4 and 8-kb transcript, respectively, originate from related but distinct genes. Nucleotide sequence analysis of cDNA corresponding to the 8 kb mRNA (Fig. 2), established that mice also express this transcript. The nucleotide sequence of the rat and human transcripts have previously been published (14, 15). The mouse protein is very similar to its rat counterpart with 99% identity on the protein level (Fig. 2). As demonstrated by Northern blotting, the 8-kb transcript is widely distributed and was detected in all tissues analyzed (Fig. 3).

The gene encoding the human 8-kb transcript has previously been characterized (16). Characterization of the mouse gene encoding the 4-kb transcript, which spans at least 7.4 kb and is split into 14 exons (Fig. 4), revealed a high resemblance between the two genes. The exons have the same size, except for exons 1, 2, and 14. In addition, the exon-intron phases from exon 2 and downstream are of the same type (eight type 0, two type 1 and two type 2 boundaries). The structural similarities between these genes strongly suggest that they have evolved from a common ancestral gene. However, some differences between the two genes may be noted. 1) The introns of the human gene encoding the 8-kb transcript are much larger (16). This gene was estimated to span approximately 35 kb of genomic DNA. 2) In the murine gene encoding the 4-kb transcript, the initiation codon is found in the large exon 2, which contains 1005 nucleotides of coding sequence, while the corresponding region in the other gene (1008 nucleotides) is divided between the two first exons.³ 3) In the human gene, the splice donor site of exon 8 is a GC rather than the expected GT.³ This rare variant was not found in the murine gene encoding the 4-kb transcript. 4) Exon 14 in the mouse gene has 123 nucleotides of coding region, whereas the corresponding exon in the human gene contains 117 nucleotides. While so far only two different genes encoding glucosaminyl N-deacetylase/N-sulfotransferases have been identified, we cannot exclude the presence of additional genes.

It was previously suggested that the enzyme encoded by the 4-kb transcript, first recognized in mouse mastocytoma (8), was restricted to connective tissue mast cells and to the biosynthesis of heparin (9, 10). However, the tissue distribution of the 4-kb N-deacetylase/N-sulfotransferase mRNA reported in this paper (Figs. 6 and 7), demonstrates that this transcript is widely distributed and that it is present in cells other than mast cells. Accordingly, the enzyme encoded by the 4-kb transcript appears to take part both in heparin and heparan sulfate biosynthesis. The high level of expression of the 4-kb transcript in mastocytoma compared with its expression in e.g. liver (Fig. 6), may be the reason why this transcript was not previously recognized in other cells (9). As indicated by overexpression of the enzymes in vitro (12, 22), the two enzymes may have different catalytic properties; transfection of cDNA corresponding to the 4-kb transcript in the human kidney cell-line 293 resulted in a dramatic increase in N-sulfation of the heparan sulfate produced by the cells (22), while overexpression in COS cells of cDNA for the 8-kb mRNA did not cause any significant change in the overall N-sulfation of the polysaccharide (12). The enzyme encoded by the 4-kb transcript thus appears to participate in the production of HS with higher N-sulfate content. Tentatively, different locations of the 4- and 8-kb transcripts in a certain tissue may result in a local variation of HS structure. Such local variation of HS structure was recently reported for kidney, using a monoclonal antibody recognizing N-unsubstituted glucosamine residues, found in HS in the glomerular basement membrane but not in the basement membranes of the tubules (23). These results imply that the fine structure of heparan sulfate may be carefully regulated. Expression of the 4- or 8-kb transcript may be one of the means for the cell to control the structure of the heparan sulfate produced, which in turn will be important for the ability of the proteoglycan for functionally important interactions with effector molecules such as cytokines, enzymes, and enzyme inhibitors (4, 6).