A novel sulfated structure in the carbohydrate-protein linkage region isolated from porcine intestinal heparin.

A preparation of porcine stage 14 intestinal heparin, which contains Ser as a predominant amino acid, was used for isolation of the carbohydrate-protein linkage region of heparin. Two glycoserines were isolated in a molar ratio of 96:4 after an exhaustive digestion with a mixture of bacterial heparinase and heparitinases. Their structures were determined by composition analysis, heparitinase digestion, co-chromatography with an authentic glycoserine on high performance liquid chromatography, and by 500-MHz one- and two-dimensional 1H NMR spectroscopy. The structure of the major one is delta GlcA beta 1-3Gal beta 1-3Gal beta 1-4Xyl beta 1-O-Ser and that of the minor is delta GlcA beta 1-4GlcNAc(6-O-sulfate) alpha 1-4GlcA beta 1-3Gal beta 1-3Gal beta 1-4Xyl beta 1-O-Ser. The novel 6-O-sulfated GlcNAc residue was demonstrated to occur in the vicinity of the carbohydrate-protein linkage region. The Gal residues were nonsulfated, in contrast to the sulfated Gal structures recently discovered in the carbohydrate-protein linkage region of chondroitin sulfate proteoglycans. The structural features are discussed in relation to biosynthetic mechanisms of the heparin glycosaminoglycans.

However, for other activities the relation with the fine structure of the glycosaminoglycan remains to be established.
The order of the modification reactions to form the antithrombin-binding site is well understood (9). There are two populations of heparin chains, namely one which carries and the other which does not carry the antithrombin-binding site. The biosynthetic selection mechanisms for these chains are unknown. Heparin is synthesized as a proteoglycan to which a number of heparin glycosaminoglycan chains are attached. Interestingly, heparin glycosaminoglycans share the common carbohydrate-protein linkage region GlcA@1-3Gal@1-3Gal@l-4Xylpl-0-Ser' with other sulfated glycosaminoglycans: heparan sulfate, chondroitin sulfate, and dermatan sulfate (10-12). The question arises how different glycosaminoglycans can be synthesized on the common structure, especially since the chain elongation proceeds stepwise and is governed largely by the substrate specificity of the glycosyltransferases involved.

EXPERIMENTAL PROCEDURES AND RESULTS'
Isolation of the Linkage Glycoserines-Purified stage 14 heparin from porcine intestine contains only Ser as the predominant amino acid (10). This preparation was exhaustively digested with a mixture of heparinase, heparitinases I and 11, and fractionated into fractions I-IV by gel filtration (Fig. 1). The recoveries of Ser in these fractions were 5.9,9.8,6.6, and 77.7%, respectively. Fraction I contained dermatan sulfate, The abbreviations used are: Xyl, xylose; UA, uronic acid; AUA, 4,5-unsaturated uronic acid; GalN, galactosamine; GalNAc, P-deoxy- Portions of this paper (including "Experimental Procedures," part of "Results," Figs. 1 and 2, and Tables 1-111) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.
whereas fraction I1 contained heparin fragments and dermatan sulfate as judged by cellulose acetate membrane electrophoresis (data not shown). Dermatan sulfate accounts for GalN in the heparin preparation and represents approximately 5% (w/w) of the total polysaccharides. It likely accounts for most of Ser in fractions I and 11, which were not analyzed further. Linkage regions were isolated from fractions I11 and IV by HPLC.
Glycoserine Z-Subfractionation of fraction IV by HPLC yielded glycoserine I with the following conventional structure. The isolation procedure and the structural analysis are described in the "Miniprint." A4~6GlcA@l-3Gal@1-3Gal~l-4Xyl~l-O-Ser Glycoserine IZ-Subfractionation of fraction I11 by HPLC yielded a number of UV-absorbing peaks (Fig. 3). Approximately 60% of the applied Ser was recovered in the fastest eluting compound, designated glycoserine 11, which represent 2.0% of the applied AUA. No appreciable amounts of Ser were recovered in the other peaks, most of which are unsaturated tetrasaccharides as judged from the ratio of uronic acid or GlcN to AUA. In addition, fraction I11 was radiolabeled with [3H]acetic anhydride and chromatographed by HPLC under the same conditions. It gave one appreciable radiolabeled peak that eluted 2 min later than the original glycoserine I1 (data not shown), being consistent with acetylation of its Ser residue. These results suggest the presence of only one major glycoserine in fraction 111. Preparative HPLC yielded 125 nmol (as Ser) of glycoserine II/100 mg of purified heparin.
Glycoserine I1 contains 1 mol each of AUA, UA, GlcN, and sulfate (Table 11). The susceptibility of the isolated glycoserine I1 to heparinase and various heparitinases was tested in order to dissect it into the subcomponents. The enzymes used have different substrate specificities (18): heparinase produces ADiHS-diS, and ADiHS-triS from heparin; heparitinases I and V have similar specificities producing ADiHS-OS, ADiHS-6S, and ADiHS-NS from bovine kidney heparan sulfate; heparitinase I1 has a broad specificity giving rise to every unsaturated disaccharide listed in Table I from heparin. The results showed that glycoserine I1 was resistant to heparinase, heparitinases I and 11, but susceptible to heparitinase V as judged by the 2-fold increase in UV absorption at 232 nm. Upon HPLC the digest gave rise to an equimolar amount of two components, one was eluting at the position of ADiHS-6s and the other at that of glycoserine I (Fig. 4). This indicates that glycoserine I1 is composed of AGlcAB1-4GlcNAc(6-0sulfate) and GlcA/3l-3Gal~l-3Gal~l-4Xyl/3l-O-Ser.
The 'H NMR spectrum of glycoserine I1 is depicted in Fig.  5, and the NMR data are given in Table 111  In the two-dimensional homonuclear Hartmann-Hahn (21) spectrum ( Fig. 6) the indicated assignment pathway, starting at the aGlcNAc H-1, leads to the H-2, H-3, H-4, and H-5 signals of aGlcNAc. The cross-peaks between the H-5 and the H-6 and H-6' atoms demonstrates that the signals shifted under influence of sulfate can be assigned to the H-6 and H-6' of aGlcNAc, bearing an 0-sulfate at C-6. Full assignment of Xyl and AGlcA and a partial assignment of the Ser, Gal-2, and Gal-3 residues is possible in the two-dimensional HO-HAHA spectrum (Fig. 5 , Table 111). The structure of glycoserine I1 is as follows.
The differences in the chemical shifts of the Ser and the Xyl H-1 and H-2 protons are probably due to pH variations. Similar differences are found in work on synthetic linkages (compare Refs. 27 and 28).

DISCUSSION
In the present study the isolation and characterization of two glycoserines (I and 11) from the linkage region of porcine intestinal heparin is described. Glycoserines I and I1 are derived from two populations that differ in structure in the vicinity of the linkage region. Interestingly, two types of heparin chains occur, namely those which carry and those  The assignment of the other residues is indicated on the cross-sections (----).
which do not carry the antithrombin-binding region. This gives rise to the challenging question whether 6-0-sulfation of GlcNAc is a marker which corresponds to the presence or absence of the antithrombin high affinity heparin. The proportion of the anithrombin high affinity heparin in the stage 14 preparations is approximately 48% as shown by affinity chromatography? Thus there appears to be no direct correlation between 6-0-sulfation of the first GlcNAc residue and segregation of the low and high affinity binding sequences. It remains to be investigated whether the 6-0-sulfated GlcNAc is a signal for synthesis of any other domain structure in the outer region of the chain. The sulfate group first transferred to the vicinity of the linkage region may have an influence on the subsequent biosynthetic modification of the rest of the U. Lindahl, personal communication. chain in analogy to the possible effect of the sulfate groups attached to the Gal residues on the hexosamine transfer as discussed below.
Another structural aspect of glycoserines I and I1 is that the Gal residues are not sulfated, which is in contrast to the sulfated structures, GlcA@l-3Gal(4-O-sulfate)/31-3Gal~1-4Xyl (13,14) or GlcA@l-3Gal(6-O-sulfate)@l-3Gal(6-O-sulfate)/31-4Xyl(l5), demonstrated to occur in the linkage region isolated from chondroitin 4-or 6-sulfate, respectively. Thus, there are distinct differences in the structure of the linkage region between chondroitin sulfate and heparin. No modification by sulfation of the Gal residues may lead to glucosaminoglycans. The committing biosynthetic step in sorting could be the transfer of the first N-acetylhexosamine unit to the linkage tetrasaccharide core. The N-acetylglucosaminyltransferase may recognize the nonsulfated Gal residues while the sulfated Gal structures may be the specific recognition signals for the N-acetylgalactosaminyltransferase.
Phosphorylation of the Xyl residue in the linkage region has been demonstrated in chondroitin sulfate (29, 30), heparan sulfate (31), and also suggested in bovine lung heparin (32). Neither glycoserine I nor I1 is phosphorylated. It remains to be established whether the lack of phosphate is an authentic feature or an artifact resulting from the isolation procedure.
One of the structural features unique to the linkage region of heparin and heparan sulfate is that the glucosamine units adjacent to the linkage region are regularly N-acetylated (33). It has been suggested that glucosamine units may become 6-0-sulfated only when they are located next to an N-sulfated disaccharide unit (9). If so, the second glucosamine unit which had been bound to glycoserine I1 would have been N-sulfated. Based on the susceptibility to nitrous acid treatment, the amino group of the second glucosamine residue of bovine lung heparin has been claimed to be N-sulfated (32). Alternatively, a specific GlcNAc 6-0-sulfotransferase may exist which catalyzes 6-0-sulfation of only the innermost GlcNAc next to the linkage region.   Table I) The double p e a k ObSeNed tor ADIHS-OS and ADIHS-~S are due Io 0 and P anomers. The compound tn the extra peak conlalned approxlmafely 1 mol of Ser per mol 01 AUA, and IS referred to as Glycoserlne I, representlng 90 and 3 2% of the applled Sar and AUA, respectively Preparative chromatography yleided 3 08 pmol (as Ser) of Glycoserlne I per 100 mg 01 the purltied heparin Glycosenne I dld not conla~n GlcN, sullale or phosphate (Table 11) It co-chromalographed upon HPLC on an amlne-bound s111ca column with AGlcApt-3Gal~t-3Gatpt-4Xylpl-O-Ser (data not shown), whlch was prepared by chondmhnase ACll d~gssl~on 01 AGlcAPl-3GalNAcP1-4GlcAPI-3Galpl.3Galpl.4Xylpt-OSer #solated from chondrom 4-sulfate proteoglycans (tractton D-l In ret summarlred 10 Table 111 Comparison 01 the spectral data of Glycoserme I wlth those of the 13) To ldent8fy the compound a IH-NMR spectrum was recorded and the obtained data are that Glycosenne I contams a smlar strwlure, but mlsstng the GalNAc-GlcA dlsaccharlde Unlt SlnCe