Age-dependent Modulation of Heparan Sulfate Structure and Function*

Heparan sulfate interacts with growth factors, matrix components, effectors and modulators of enzymatic catalysis as well as with microbial proteins via sulfated oligosaccharide domains. Although a number of such domains have been characterized, little is known about the regulation of their formation in vivo. Here we show that the structure of human aorta heparan sulfate is gradually modulated during aging in a manner that gives rise to markedly enhanced binding to isoforms of platelet-derived growth factor A and B chains containing polybasic cell retention sequences. By contrast, the binding to fibroblast growth factor 2 is affected to a much lesser extent. The enhanced binding of aorta heparan sulfate to platelet-derived growth factor is suggested to be due to an age-dependent increase of GlcN 6-O-sulfation, resulting in increased abundance of the trisulfated l-iduronic acid (2-OSO3)-GlcNSO3(6-OSO3) disaccharide unit. Such units have been shown to hallmark the platelet-derived growth factor A chain-binding site in heparan sulfate.

Interactions of the sulfated glycosaminoglycan (GAG) 1 heparan sulfate (HS) with various proteins affect the biological activity, tissue localization, and turnover of the protein ligands (1)(2)(3). Such interactions, generally electrostatic in nature, regularly involve specific oligosaccharide domains generated by an elaborate biosynthetic machinery in the Golgi apparatus. HS formation starts by assembly of an initial (GlcA-GlcNAc) n polymer. Parts of the nascent polymer are subsequently modified by Ndeacetylation/N-sulfation of GlcNAc residues, and further modifications, including C-5 epimerization of GlcA residues into IdceA residues as well as O-sulfation at various positions, occur mainly in the vicinity of the previously incorporated N-sulfate groups (3,4). The O-sulfate groups are predominantly found at the C-2 position of IdceA residues and the C-6 position of GlcN residues. The protein-binding HS domains typically reside within the Nsulfated regions, their functional specificity being determined by the pattern of modification, particularly the positioning of sulfate groups. Although information regarding the structures of the recognition sites for individual proteins (5)(6)(7)(8)(9)(10)(11) and the general features of HS biosynthesis (3) is accumulating, the biological control of HS structure and function remains poorly understood. Nevertheless, HS species from various cells and tissues clearly differ in their structure and in some studies such differences have been correlated to differential protein-binding properties (12)(13)(14)(15). These and other findings (discussed in Refs. 1 and 3) suggest that the biosynthesis of HS is subject to regulation during development or aging. Control of the appropriate expression of functional HS domains in given organs or at particular developmental stages would appear essential, whereas, conversely, perturbed regulation could contribute to various pathologies. In the present study we have explored the aging aortic wall as a model to gain insight into the control of HS structure and function in humans. Aging is a strong predisposing factor to atherosclerosis, characterized by endothelial damage, lipid accumulation, and cell proliferation in arterial wall plaques (16). HS has been attributed multiple roles in these processes by interacting with lipoprotein lipase (17) and with growth factors such as basic fibroblast growth factor (FGF-2) and platelet-derived growth factors (PDGFs) (18 -21). HS binds to and enhances the mitogenic activity of FGF-2 (22) and regulates the tissue localization of PDGFs (20). PDGFs are homo-or heterodimers of two closely related, A and B, polypeptide chains. The PDGF-A chain exists as two variants due to alternative mRNA splicing. The longer variant (PDGF-A L ) contains a C-terminal polybasic sequence that serves as a cell retention signal and is associated with the localization of PDGF to the surface of the producer cell or to the extracellular matrix (21), presumably due to interactions with HS (20). A similar but not identical retention signal is found at the C terminus of the PDGF-B propeptide chain (PDGF-B L ) (21). This propeptide may undergo various N-and C-terminal proteolytic processing events that give rise to multiple forms of cell-associated PDGF-B species (23).

MATERIALS AND METHODS
Isolation and Radiolabeling of Heparan Sulfate-Tissue samples from human abdominal aorta were obtained at autopsy and stored at Ϫ70°C until processed for HS isolation. The surrounding connective tissue was removed, and the sample, encompassing the entire thickness of the vessel wall, was cut into fine pieces with a scalpel. The samples were defatted essentially as described (15) with the exception that ethyl ether was replaced by ethanol. Defatted samples (dry weight, 0.15-2.0 g) were subjected to protease digestion with papain (Sigma; 5 mg/g of defatted tissue) in 25 ml of 0.05 M Tris, pH 5.5, 0.01 M EDTA, 2 M NaCl, 0.01 M cysteine/HCl at 60°C for 18 h and centrifuged (10 min at 2000 ϫ g), and the supernatants were applied to columns of DEAE-Sephacel (1.5 ϫ 5 cm; Amersham Pharmacia Biotech) equilibrated with 0.05 M Tris, pH 7.2. The column was washed with 0.05 M sodium acetate, pH 4.0, followed by elution of the DEAE-bound material (containing sulfated GAGs) by a linear gradient of LiCl (0.15-2.0 M) in the acetate buffer (15). 2-ml fractions were collected and analyzed for uronic acid content by the carbazole reaction (24). Fractions containing GAGs were pooled, dialyzed against water, lyophilized, and digested with 1 unit of chondroitinase ABC (Seikagaku Corporation) and 125 units of endonuclease (Benzonase, Benzon Pharma A/S) in 0.05 M Tris-HCl, pH 8.0, 1 mM MgCl 2 , 0.05 M sodium acetate at 37°C overnight (25). The digest was heated for 2 min and applied to a DEAE-Sephacel column (1.5 ϫ 5 cm) equilibrated with 0.2 M NH 4 HCO 3 and eluted by a linear gradient of NH 4 HCO 3 (0.2-2.0 M). Fractions containing HS were identified by the carbazole reaction, pooled, and freeze-dried, and the material was stored at Ϫ20°C until further use. Treatment of the purified HS preparations with HNO 2 at pH 1.5 resulted in quantitative degradation of the material into lower molecular weight species as demonstrated by chromatography of intact and HNO 2 -treated samples on a column of Superose 12 (data not shown), indicating that the purification procedure yielded pure HS.
PDGF Affinity Chromatography-5 mg of recombinant PDGF-AA L was mixed with an equimolar amount of heparin and immobilized to 3 ml of CH-Sepharose CL4B (Amersham Pharmacia Biotech) according to the manufacturer's instructions (8). The column was equilibrated with Tris-buffered saline (50 mM Tris-HCl pH 7.4, 150 mM NaCl) prior to the application of [ 3 H]HS. Subsequently, the column was washed with 10 ml of Tris-buffered saline, and the bound material was eluted using a linear NaCl gradient (0.15-2.0 M in 50 mM Tris-HCl, pH 7.4). Fractions of 1 ml were collected and analyzed for radioactivity in a liquid scintillation counter.
To calculate the degree of N-sulfation of GlcN residues the [ 3 H]a-Man R end group-labeled HS oligosaccharides resulting from the HNO 2 pH 1.5/NaB 3 H 4 treatment were separated by chromatography on a column of Bio-Gel P-10 (1 ϫ 160 cm, Bio-Rad) in 0.5 M NH 4 HCO 3 . Fractions of 1 ml were collected and analyzed for radioactivity. The degree of N-sulfation was calculated from the elution profiles using the following formula.
N-sulfate groups total N-substituents ϭ nϭ2 nϭϱ a n nϭ2 nϭϱ a n ϫ n/2 (Eq. 1) where n is the number of monosaccharide units and a is the total radioactivity in a given oligosaccharide species.
In order to isolate 3 H-labeled tetrasaccharides, aliquots of the HS oligosaccharides resulting from cleavage with HNO 2 at pH 1.5 were separated by gel chromatography on a Superdex 30 (Amersham Pharmacia Biotech) column in 0.5 M NH 4 HCO 3 . Fractions corresponding to tetrasaccharides were pooled and desalted. Tetrasaccharide species were separated by high voltage paper electrophoresis on Whatman number 3MM paper in 6.5% HCOOH, pH 1.7.

RESULTS
We first examined the binding of radiolabeled human aorta HS from a young and an old individual to FGF-2 and to dimers of the long isoforms of PDGF A or B chains (designated PDGF-AA L and PDGF-BB L below) (30). A filter trapping procedure was employed, involving interaction of labeled HS and the protein ligand in solution followed by rapid passage of the mixture through a nitrocellulose filter. Protein and proteinbound HS chains (but not free HS chains) were retained on the filter (8). We found that the binding capacity for PDGF-AA L and PDGF-BB L of HS from a 76-year-old individual was 4 -5 times higher than that of HS from a 21-year-old individual (Fig. 1A), whereas the binding to FGF-2 differed only marginally between the two HS samples. To assess the inter-individual variation in these interactions, we next compared the binding of six HS preparations, from three young and three old subjects, to PDGF and FGF-2. The results of this experiment (Fig. 1B) indicated a virtually invariable level of binding of HS to a given protein ligand within each of the two age groups but marked differences in binding characteristics between these groups. The effect of age thus was selectively expressed for different proteins, in accord with the data shown in Fig. 1A. These findings clearly point to the occurrence of age-dependent differences in HS structure that affect the binding of HS to PDGF and FGF-2 in distinct ways. The effect of this structural transition with regard to PDGF binding was further examined by affinity chromatography of HS from young and old subjects on a column of immobilized PDGF-AA L (Fig. 1C). Both types of HS contained material that remained bound to the growth factor at physiological pH and ionic strength, as well as a fraction of unbound material. The PDGF-binding chains amounted to 50.0 Ϯ 2.4% versus 81.8 Ϯ 1.5% (mean Ϯ S.E. from three and two samples, respectively) of the total HS from young 2 F. Lustig, J. Hoebeke, C. Simonson, G.Östergren-Lundén, G. Bondjers, U. Rü etchi, and G. Fager, manuscript in preparation. and old subjects, respectively. These results thus confirm not only the correlation between individual age and PDGF binding ability of aortic HS but also the striking inter-individual similarity within each age group.
Binding of HS to PDGF-AA L and FGF-2 involves structurally distinct oligosaccharide domains. The former domain is comprised by N-sulfated ϳ8-mer sequences with at least one trisulfated IdceA(2-OSO 3 )-GlcNSO 3 (6-OSO 3 ) disaccharide unit (8), whereas the minimal FGF-2-binding site contains an essential IdceA(2-OSO 3 ) residue but no 6-O-sulfate groups (5,7,31). Given these data, we wanted to examine whether the differential protein binding of aorta HS from young and old subjects would correlate with the O-sulfate substitution pattern of the N-sulfated regions of HS. We therefore determined the disaccharide composition of these regions in a total of 15 HS preparations from subjects aged 20 -84 years (including the six preparations used in the binding experiments). The results of this analysis (Table I) demonstrated an age-dependent increase in the proportion of the IdceA(2-OSO 3 )-GlcNSO 3 (6-OSO 3 ) unit (Fig. 2). Calculation of the overall extent of 2-O-and 6-O-sulfation indicated that this increase was due to an approximate doubling of the level of 6-Osulfate substitution of GlcNSO 3 residues in the old subjects (Fig.  2), whereas the IdceA 2-O-sulfation remained high and essentially unchanged (Fig. 2). The 6-O-sulfation of GlcNSO 3 residues showed an almost linear increase between the age of 20 and 40 years; in the still older subjects the levels of 6-O-sulfation were somewhat scattered but consistently higher than in the young individuals (Fig. 2).
We further determined the degree of N-sulfate substitution of GlcN residues from the pattern of HS depolymerization following cleavage with HNO 2 (at pH 1.5; see "Materials and Methods"). Five of the HS samples (subject aged 20, 21, 74, 76, and 80 years) were analyzed (Fig. 3) (Table I)

DISCUSSION
The present study demonstrates that human aging is accompanied by specific alterations in the fine structure of HS in the aortic wall. The increased degree of 6-O-sulfate substitution of GlcNSO 3 residues in old subjects is reflected by an elevated proportion of trisulfated, "heparin-type" IdceA(2-OSO 3 )-Glc-NSO 3 (6-OSO 3 ) disaccharide units and enhanced binding of the HS to PDGFs. Such progressive, age-dependent structural alteration represents a previously unrecognized type of functional modulation of a mammalian macromolecule. Unlike rapid and reversible modifications such as phosphorylation, the alterations in sulfation described here occur during an extended time span of several decades. The remarkable structural and functional similarities of HS specimens from different age-matched subjects suggest a strictly controlled process of HS assembly.
Our findings point to 6-O-sulfation of GlcNSO 3 units as a key regulator of the biological properties of HS in the human aortic wall. The 2-O-and 6-O-sulfation reactions appear to be separately regulated during HS biosynthesis, because the former sulfation type is largely confined to the contiguous N-sulfated sequences, whereas the latter modification occurs within as well as outside these domains (4). Intriguingly, the enhanced 6-O-sulfation appeared to selectively involve the contiguous N-sulfated domains, because the sulfation of alternating domains (recovered as tetrasaccharides after cleavage of HS with HNO 2 at pH 1.5) was essentially similar in HS from young and old subjects. This finding suggests the involvement of two (or more) HS GlcN 6-O-sulfotransferase species that differ in substrate specificity and mode of regulation. Unfortunately, our knowledge regarding the regulatory mechanisms in HS biosynthesis is still fragmentary (3). Could the abundance of a given sulfate group simply reflect the expression level for a corresponding sulfotransferase?
Atherosclerotic lesions are frequently encountered in human aorta, and HS-binding growth factors such as FGFs and PDGFs are thought to contribute to the pathological smooth muscle cell migration and proliferation characterizing the disease (16). The expression level of the PDGF-A L chain is high in human arterial smooth muscle cells and increases markedly during the conversion of monocytes into macrophages (32). It has been shown that PDGF isoforms containing the A L or B chains are retained at cell surfaces or in the extracellular matrix by HS (20,21). In the arterial wall, increased binding of PDGF to HS could thus result in extracellular accumulation of PDGF. Such early changes might facilitate later pathophysiological processes such as aberrant smooth muscle cell migration and growth in individuals prone to develop atherosclerotic disease. Moreover, we note that the trisulfated IdceA(2-OSO 3 )-GlcNSO 3 (6-OSO 3 ) disaccharide units found to promote binding of PDGF to HS has also been implicated in binding of lipoprotein lipase (6), a cell surface-bound enzyme that catalyzes the breakdown of triglycerides and affects the cellular uptake of lipids (17). The observed change in HS structure thus is likely to have more widespread functional implications than those emphasized in this study. Samples of aortic HS from a young (20 years) and an old (76 years) subject were subjected to cleavage with HNO 2 at pH 1.5. The cleavage products were radiolabeled by reduction with NaB 3 H 4 and separated by chromatography on Bio-Gel P-10 as described under "Materials and Methods."