Biosynthesis of Heparin AVAILABILITY OF GLUCOSAMINYL 3-0-SULFATION SITES*

Heparin preparations isolated from pig intestinal mucosa and from bovine lung were fractionated with regard to affinity for antithrombin. The resulting fractions, with high (HA) or low (LA) affinity for the proteinase inhibitor, were analyzed by 13C NMR or by identification of di- and tetrasaccharides obtained through deaminative cleavage with nitrous acid. Structural differences between corresponding HA and LA fractions were essentially restricted to minor constituents, in particular 3-O-sulfated glucosamine units that occurred (1 or 2 residues/chain) in all HA preparations but were scarce or absent in LA heparin. The HA fractions also consistently showed higher contents of nonsulfated iduronic acid and, to a lesser extent, N-acetylated glucosamine units than the LA fractions. The two tetrasaccharide sequences, -IdoA-GlcNAc(6-OSO3)-GlcA-GlcNSO3- and -IdoA-GlcNAc(6-OSO3)-GlcA-GlcNSO3(6-OSO3)- , recently implicated as part of the acceptor site for glucosaminyl 3-O-sulfate groups (Kusche, M., Bäckström, G., Riesenfeld, J., Petitou, M., Choay, J., and Lindahl, U. (1988) J. Biol. Chem. 263, 15474-15484), were identified in mucosal LA heparin; it was calculated that the preparation contained approximately one potential acceptor site/polysaccharide chain. Yet this material did not yield any labeled HA components on incubation with adenosine 3'-phosphate 5'-phospho-[35S]sulfate in the presence of glucosaminyl 3-O-sulfotransferase, solubilized from a mouse mastocytoma microsomal fraction. The failure to incorporate any 3-O-sulfate groups could conceivably be explained by the occurrence of a D-glucuronic rather than L-iduronic acid unit linked at the reducing ends of the above tetrasaccharide sequences. Alternatively, 3-O-sulfation may be restricted by other, as yet unidentified, inhibitory structural elements that are preferentially expressed in polysaccharide sequences selected for the generation of LA heparin.


Heparin
preparations isolated from pig intestinal mucosa and from bovine lung were fractionated with regard to affinity for antithrombin. The resulting fractions, with high (HA) or low (LA) affinity for the proteinase inhibitor, were analyzed by 13C NMR or by identification of di-and tetrasaccharides obtained through deaminative cleavage with nitrous acid. Structural differences between corresponding HA and LA fractions were essentially restricted to minor constituents, in particular 3-O-sulfated glucosamine units that occurred (1 or 2 residues/chain) in all HA preparations but were scarce or absent in LA heparin. The HA fractions also consistently showed higher contents of nonsulfated iduronic acid and, to a lesser extent, Nacetylated glucosamine units than the LA fractions. The two tetrasaccharide sequences, -IdoA-GlcNAc(G-OS03)-GlcA-GlcNS03-and -IdoA-GlcNAc(6-OSO& GlcA-GlcNS0,(6-OSO,)-, recently implicated as part of the acceptor site for glucosaminyl 3-O-sulfate groups (Kusche, M., Backstrom, G., Riesenfeld, J., Petitou, M., Choay, J., and Lindahl, U. (1988) J. Biol.
Chem. 263,15474-15484), were identified in mucosal LA heparin; it was calculated that the preparation contained approximately one potential acceptor site/ polysaccharide chain. Yet this material did not yield any labeled HA components on incubation with adenosine 3'-phosphate 5'-phospho- [35S]sulfate in the presence of glucosaminyl3-0-sulfotransferase, solubilized from a mouse mastocytoma microsomal fraction. The failure to incorporate any 3-O-sulfate groups could conceivably be explained by the occurrence of a Dglucuronic rather than L-iduronic acid unit linked at the reducing ends of the above tetrasaccharide sequences.
Alternatively, 3-0-sulfation may be restricted by other, as yet unidentified, inhibitory structural elements that are preferentially expressed in polysaccharide sequences selected for the generation of LA heparin.
The key step in the blood anticoagulant action of the sulfated glycosaminoglycan heparin is binding of the serine  (1988) and by Lidholt et al. (1989). The-costs of publication of this article were defrayed in part by the payment of page charges.
This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. protease inhibitor antithrombin (AT)' to the polysaccharide chain (Bjork and Lindahl, 1982). Thus bound, AT inactivates the enzymes of the coagulation process much more efficiently than does the free inhibitor.
Functional binding of AT to the heparin chain requires the presence of a specific pentasaccharide sequence, the structure/function relationships of which have been explored in detail (Thunberg et al., , 1982Riesenfeld et al., 1981;Choay et al., 1981Choay et al., , 1983Lindahl et al., 1980Lindahl et al., , 1983Lindahl et al., , 1984Atha et al., 1985Atha et al., , 1987Petitou et al., 1988). The distinguishing structural feature of the AT-binding region is a 3-O-sulfated glucosamine residue ( Fig. 1) that is either absent or very rare elsewhere in the heparin molecule .
In commercial heparin preparations, only a fraction, usually 30-40% of the molecules contains the AT-binding sequence and has high affinity for AT (HA heparin). The majority of the chains lack the 3-O-sulfate group and thus show low affinity for the inhibitor (LA heparin) and only weak anticoagulant activity (see Bjijrk and Lindahl, 1982). Apart from the 3-O-sulfate group, no apparent qualitative structural difference between HA and LA heparin has been found. However, it has been reported that HA heparin contains more Nacetylated glucosamine, D-glucuronic acid, and nonsulfated Liduronic acid than does LA heparin (Rosenberg et al., 1978;Rosenberg and Lam, 1979;Lindahl et al., 1979;Radoff and Danishefsky, 1985). Reports by Radoff and Danishefsky (1984) and by Rosenfeld and Danishefsky (1988), proposing that the AT-binding sequence is preferentially located toward the nonreducing end of the HA heparin chain, were not confirmed by Oscarsson et al. (19891, who favored rather a random distribution of the binding region. It may be noted that an AT-binding region similar or identical to that of heparin has also been found in heparan sulfate (Lane et al., 1986;Marcum et al., 1986;Pejler et al., 1987).
Heparin is synthesized in connective tissue-type mast cells as a proteoglycan that consists of multiple polysaccharide chains joined by a polypeptide sequence of alternating serine and glycine units (Yurt et al., 1977;Robinson et al., 1978;Lidholt et al., 1988;Kjellen et al., 1989 presence of PAPS, this polysaccharide is carried through a series of modification reactions which is initiated by Ndeacetylation and N-sulfation of glucosamine units, proceeds through C-5 epimerization of D-glucuronic to L-iduronic acid units, and is concluded by incorporation of O-sulfate groups at various positions (for reviews, see Lindahl and Kjellen, 1987;. The structural heterogeneity typical of heparin (and heparan sulfate) reflects the fact that these reactions are generally incomplete, a variable proportion of the disaccharide units remaining at lower levels of modification.
For the formation of defined saccharide sequences, the selection of target units in each reaction must be strictly regulated. Although the mechanisms behind such regulation are only partly understood, they clearly depend to a large extent on the substrate specificities of the corresponding enzymes.
The formation of the AT-binding region highlights the problems of regulation in glycosaminoglycan biosynthesis. A previous study focused on the introduction of the specific marker group for this region, i.e. the glucosaminyl3-O-sulfate residue . It was found that 3-0-sulfation concludes the polymer modification process and is essentially restricted to saccharide sequences that already exhibit all other structural features required for high affinity binding to AT. These findings raised some specific questions regarding those polysaccharide chains that escape 3-0-sulfation and thus will be ultimately classified as LA heparin. Do these molecules lack the appropriate acceptor sequence required for substrate recognition by the 3-0-sulfotransferase, or do they simply not encounter the enzyme? These problems have been addressed in the present study, which involves a detailed structural comparison between heparin preparations with high and low affinity for AT as well as direct attempts to test LA heparin as a potential acceptor molecule in the glucosaminyl 3-0-sulfotransferase reaction.  (Lindahl et al., 1965). The product was "H-labeled by acetylating free amino groups with [3H]acetic anhydride (Hook et al., 1982). In addition, a preparation (lot 170. FW) of heparin derived from bovine lung was obtained from the Upjohn Co. The various preparations were fractionated by affinity chromatography on antithrombin-Sepharose, essentially as described (Laurent et al., 1978). respectively. An a-methylglycoside of a pentasaccharide with the structure GlcNS0~(6-OS0~)-GlcA-GlcNSO~(6-OSO~)-IdoA(2-OSO~)-Glc-NSO,(G-OSO,), i.e. a sequence corresponding to the antithrombinbinding region except for the lack of a 3-O-sulfate group at unit 4 (see Fig. l aManR, respectively, with O-sulfate groups in various positions, were as described (Pejler et al., 1987). Inorganic Na2"S0, (carrier-free) was purchased from Amersham Corp. and was used in the preparation of [35S]PAPS as reported (Jacobsson and Lindahl, 1980). Sephadex G-15, bovine liver fl-D-glucuronidase (contaminated with some a-L-iduronidase), and human antitbrombin covalently coupled to Sepharose 4B were obtained as described . A microsomal fraction was prepared from a transplantable mouse mastocytoma (Furth et al., 1957) according to Jacobsson et al. (1979a).

Methods
Hexuronic acid was determined by the carbazole reaction (Bitter and Muir, 1962) and radioactivity either in a Beckman model LS 3800 liquid scintillation spectrometer or by a flow detector connected to an HPLC apparatus as described (Pejler et al., 1987). Analysis of Heparin Deamination Products-The composition of heparin preparations was investigated by separation and quantification of Hlabeled fragments obtained by deaminative cleavage of the polysaccharides with nitrous acid followed by reduction of the products with NaB3HI.
Samples (300 pg) of polysaccharides (dissolved in 20 ~1 of H,O) were mixed with 200 ~1 of HNO, reagent (pH 1.5), and the reaction was allowed to proceed at room temperature for 10 min. Following the addition of 75 pl of 1 M Na&O,, the products were reduced with 5 mCi of NaB3H4 ( were related to those of disaccharide standards (see Fig. 3). The identification of disaccharides was ascertained by rerunning samples after digestion with fi-glucuronidase (Jacobsson et al., 197913). Before analysis of tetrasaccharides, "anomalous" deamination products containing an internal 2-deoxy-2-C-formyl-n-pentofuranosyl unit (Lindberg et al., 1975;Shively and Conrad, 1976) Fig. 2) allowing the separation of distinct groups of mono-, di-, and trisulfated species. The elution procedure was designed such that the two tetrasaccharides derived from the antithrombin-binding region, the di-O-sulfated IdoA-GLNAc(6-OSO~)-GlcA-aMans(3-OS0~) and the tri-O-sulfated IdoA-GlcNAc(G-OSOJ-GlcA-aMana(3,6&OSOs), both appeared as single peaks. These components could be identified directly by use of the corresponding standards (components a and b, respectively, in Fig. 7A in Thunberg et al., 1982) but nevertheless were, along with the other tetrasaccharide fractions, subjected to more detailed structural analysis, essentially according to methods described by Thunberg et al. (1982), Shaklee and Conrad (1984), and Bienkowski and Conrad (1985). In brief, the following procedure was employed.

Compositional Analysis of Heparin Fractions
Analysis of Deamination Products-Heparin preparations derived from pig intestinal mucosa (preparation A) and from bovine lung were fractionated by affinity chromatography on antithrombin-Sepharose, and the HA and LA fractions obtained were subjected to HNOz (pH 1.5)/NaB3H4 treatment as described under "Methods." The resulting labeled di-and tetrasaccharides were separated by gel chromatography (not shown), and the tetrasaccharide fractions were treated with dilute H&O4 followed by repeated gel chromatography to eliminate components due to "anomalous ring contraction" (see "Methods").
The amounts of N-acetylated glucosamine units in the HA and LA fractions of mucosal heparin preparation A, as estimated from the relative amounts of labeled tetra-and disaccharides, corresponded to 11 and 8.1%, respectively, of the total disaccharide units (see Table II; calculated from the data in Table I). The corresponding values for the lung heparin HA and LA fractions were 4.4 and 2.7%, respectively. Assuming an average M, for mucosal HA heparin of 15,000 (Oscarsson et al., 1989) and an M, of 600 for the average disaccharide unit (hence 25 such units for each polysaccharide chain), each heparin molecule would contain between 2 and 3 solitary N-acetylated glucosamine residues interspersed between N-sulfated sequences (see also below). By analogous reasoning, the lung HA heparin chain would contain at most 1 N-acetylated glucosamine unit. For both types of heparin, the HA fractions showed somewhat higher N-acetyl contents than the LA fractions. The occurrence of N-acetylated glucosamine units resistant to deamination resulted in the generation of tetrasaccharides with the general backbone structure HexA-GlcNAc-GlcA-[ l-3H]aManR, and the separation of these components by ionexchange HPLC is illustrated in Fig. 2. The two tetrasaccharides containing 3-O-sulfated glucosamine residues (IdoA-GlcNAc(G-OS03)-GlcA-aMana (3-OSOZ) and IdoA-GlcNAc(G-OSO&-GlcA-aManz(3,6-di-OS03)), derived from the antithrombin-binding region, were readily detected among the deamination products from both mucosal ( Fig. 2A) and lung (not shown) HA heparin but were not seen in the tetrasaccharides derived from LA heparin ( Fig. 2B and Table I). On the other hand, the corresponding "precursor" sequences, of identical structures but for the absence of the marker 3-0sulfate groups, were clearly present in significant amounts in the LA heparin fractions.' All additional tetrasaccharides identified occurred in the deamination products from HA as well as LA heparin although in somewhat variable quantities ( Table I).
The HexA-[ l-3H]aMans disaccharides were also analyzed by ion-exchange HPLC (Fig. 3). The identities of the various components as inferred from the elution positions of standard disaccharides were ascertained by repeated HPLC of the samples following digestion with a P-D-glucuronidase/cu-Liduronidase mixture. This treatment resulted in the degradation of disaccharides containing nonsulfated glucuronic or iduronic acid units along with the generation of the corresponding labeled anhydromannitol monosaccharide derivatives. Although the gross composition of disaccharide fractions obtained from the corresponding HA and LA preparations did not differ markedly (Table I) The minor labeled material indicated with an asterisk (*) was not analyzed further but probably represents the tetrasaccharide GlcA-GlcNAc(G-OSOB)-GlcA-aMana (Bienkowski and Conrad, 1985). ---, KH,PO, concentration.
units (Bienkowski and Conrad, 1985) occurred in all the heparin fractions studied, with no apparent correlation to affinity for antithrombin. On the other hand, nonsulfated iduronic acid recovered in the disaccharide IdoA-aManR(6-OSOa) was consistently more abundant in the HA than in the LA fractions (see also Lindahl et al., 1979). Moreover, both HA fractions, and particularly the lung material, generated appreciable amounts of GlcA-aManR(3,6-di-OSOJ disaccharides on deamination, presumably derived from antithrombinbinding regions containing an N-sulfated rather than Nacetylated glucosamine unit at position 2 (see Fig. 1 and Lindahl et al., 1984). Unexpectedly, significant albeit smaller amounts of this disaccharide were noted also on analysis of the mucosal LA heparin (Fig. 3B and Table I), suggesting that the -GlcA-GlcNS0,(3,6-di-OSOZ)-sequence is not exclusively linked to high affinity for antithrombin.
The presence of 3-O-sulfated glucosamine in the mucosal LA heparin preparation could not be due to residual HA heparin since no 3-O-sulfated tetrasaccharides of the types produced by HA heparin were detected ( Fig. 2B and Table I).
Calculations based on the data in Table I, again assuming  25 disaccharide units/polysaccharide chain, indicate a total of close to 2 3-O-sulfated glucosamine residues for each lung or mucosa HA heparin molecule. However, it cannot simply be assumed that the average HA chain contains two functional antithrombin-binding sites (see e.g. Rosenberg et al., 1979;Danielsson and Bjork, 1981). In fact, the occurrence of some glucosamine 3-O-sulfate in one of the LA preparations studied points to the possibility that a fraction of the 3-O-sulfate groups also in HA heparin may fall outside the actual binding regions. The relatively low yield of S-O-sulfated tetrasaccharides from the lung HA heparin (Table I) indicates that a predominant portion of the antithrombin-binding regions in these molecules carries N-sulfate rather than N-acetyl groups at unit 2. For the mucosal heparin, this ratio is likely to be reversed. It is notable, as indicated by previous studies (Lindahl et al., 1984), that the iduronic acid unit 1 appears to be nonsulfated when unit 2 is N-acetylated but 2-O-sulfated when unit 2 is N-sulfated. NMR Spectroscopy-Fractions of pig mucosal and beef lung heparins as well as the corresponding unfractionated heparins were also characterized by 13C NMR spectroscopy. The spectra shown in Fig. 4, obtained with mucosal heparin preparation A, consist of a set of major signals due to the prevalent trisulfated disaccharide units and of weak signals attributable to minor residues (see Gatti et al., 1979). Peaks associated with the binding site for antithrombin (especially the C-2 signal of 3-or 3,6-di-O-sulfated, N-sulfated glucosamine units, marked with an asterisk in the spectrum of unfractionated heparin (Casu et al., 1981;Meyer et al., 1981)) are seen in the sample of unfractionated heparin but are, as expected, stronger and better defined in the HA fraction. Also the weak C-l signal upfield to the major GlcNS03 signal and indicated by the arrow in the HA spectrum becomes clearly stronger in the HA fraction. This signal was tentatively attributed (Braud et al., 1985) to 3-O-sulfated GlcNS03 residues. These two signals as well as the C-l signal of nonsulfated iduronic acid are much weaker or barely observable in the spectrum of the LA fraction. Similar trends were observed for the corresponding fractions derived from mucosal heparin preparation B and from beef lung heparin (spectra not shown; results summarized in Table II).
Some variability is noted with regard to the signals attributed to the binding site for antithrombin.
For instance, they are somewhat stronger in the mucosal heparin preparation A than in most other preparations of unfractionated heparin" (I$ Gatti et al., 1979;Choay et al., 1980), in accord with the unusually high proportion of HA fraction in this material (see "Materials").
Further, whereas a weak peak in the C-2 region of 3-O-sulfated glucosamine residues (at about 58 ppm) is still observable in the spectrum of the LA fractions derived from mucosal heparin preparation A (Fig. 4), there is no evidence of these signals in the LA fraction of preparation B. This is clearly illustrated by the partial spectra relating to preparation B, shown in Fig. 5. This latter figure also shows that the C-2 signal of the 3-O-sulfated glucosamine residues has a low field, still unassigned, component (indicated by an arrow in the figure), which is much less prominent for the HA fraction of preparation A but is well evident in other HA heparin fractions.3 The relative contents of the variously substituted monosaccharide units, as determined by 13C NMR (values in parentheses), are compared in Table II with the corresponding  data  obtained  through  identification  of di-and tetrasaccharide  deamination products (values calculated from Table I). The HPLC and the NMR data are generally in good agreement for the prevalent residues such as 2-O-sulfated iduronic acid and total 6-O-sulfated glucosamine.
of the minor residues, notably total N-acetylated glucosamine and 3-O-sulfated glucosamine, as estimated by NMR, are consistently higher than those determined by HPLC. The discrepancy regarding N-acetylglucosamine units may be rationalized by assuming that the NMR approach will not discriminate between such units located in the polysaccharide-protein linkage region (Lindahl, 1966) and those situated in the more peripheral regions of the polysaccharide chain, whereas the HPLC (HN02/NaB3H4) technique will recognize only the latter type of residues (which will be recovered in labeled tetrasaccharides).
Moreover, since the instrumental integration of the area of weak signals located close to strong ones (as in the case of the C-2 signal upfield to the major C-2 signal of N-sulfated glucosamine) usually overestimates the minor component, the contents of total 3-O-sulfated glucosamine are probably more accurately represented by the HPLC data. It also should be noted that the still unassigned minor signal of 3-O-sulfated glucosamine contributes up to 30% of the area measured for estimating total 3-O-sulfated glucosamine.
Should this minor peak turn out not to be associated with S-O-sulfated glucosamine units, the appropriate correction would bring the NMR data closer to the HPLC values. NMR studies are in progress to resolve and assign this and other minor heparin signals.
Enzymatic Sulfation of Heparin

Fractions
The analyses described above conformed in showing that the HA and LA fractions of the various heparin preparations differed primarily with regard to the presence or absence of 3-O-sulfated glucosamine units. We therefore decided to test whether LA heparin may indeed serve as a substrate in the glucosaminyl 3-0-sulfotransferase reaction, thereby being converted into HA heparin. The enzyme source in these experiments was a detergent-solubilized mouse mastocytoma microsomal fraction, which was found previously to catalyze the 3-0-sulfation of appropriate oligosaccharide substrates . Incubation of the microsomal enzymes with [35S]PAPS, as described under "Methods," in the absence of any exogenous sulfate acceptor, yielded labeled polysaccha- Spectra (75 MHz, D,O) are shown for the unfractionated pig mucosal heparin preparation A (UH) and for its subfractions with high affinity (HA) and low affinity (LA) for antithrombin. Assignments (Gatti et al., 1979;Casu et al., 1981) are given in the spectrum of UH for signals used in determining the relative contents of the various residues (Table II). The C-2 signal of GlcNSO,(3,6-di-OSO,) (probably coinciding with that of G~cNSO~(~-OSO~)) is labeled with an asterisk. The C-l signal labeled with an arrow in the HA spectrum (upfield to the major GlcNS03(6-OS03) signal) was attributed (Braud et al., 1985) to 3-O-sulfated GlcNS03. ent in the microsomal preparation.
Separation of this product by affinity chromatography on antithrombin-Sepharose produced a major fraction of labeled material (Fig. 6B) which appeared just before the elution position of standard LA heparin (Fig. 6A), a minor portion trailing into the HA region. Similar incubation with 200 wg of HA heparin added as an exogenous substrate showed the same distribution of background material as in the control but in addition, a distinct peak of 35S-labeled polysaccharide at the elution position of HA heparin (Fig. 6C). Structural characterization of this material by HPLC analysis of deamination products

indicated 0-[?S]sulfation
almost exclusively in the 6-O position of glucosaminyl residues but failed to show any incorporation of additional 3-O-sulfate groups (data not shown). The appearance of labeled HA heparin in this case thus simply represents sulfation of residual unoccupied 6hydroxyl groups. When 200 fig of LA heparin was substituted for the HA heparin, no labeled HA material was generated; in fact, the resulting affinity chromatogram (Fig. 6D) was indistinguishable from that produced by the endogenous sulfate acceptor alone (Fig. 6B). The occurrence in the microsomal preparation of functional 3-0-sulfotransferase was ascertained by incubating 50 pg of the synthetic pentasaccharide GlcNSO.?(6-OSOa)-GlcA-GlcNSO~(6-OSO~)-IdoA(2-OSO~)-GlcNS03(6-OS03) as exogenous sulfate acceptor. Previous experiments have shown that this molecule selectively incorporates a single sulfate group at the 3-O position of the internal glucosamine residue and that the 3-O-sulfated product, in contrast to the pentasaccharide substrate, has high affinity for antithrombin . Chromatography of the pentasaccharide incubation products on antithrombin-Sepharose showed a distinct, well retarded, labeled component (Fig. 6E) that was not detected in control incubations lacking the pentasaccharide substrate (Fig. 6B). The mastocytoma microsomal fraction used in these experiments thus expressed the 3-0-sulfotransferase activity required to conclude the formation of HA heparin. However, the LA heparin preparation tested did not function as a substrate for this enzyme. Similarly, incubation of LA heparin chains of high molecular weight, isolated after alkaline p-elimination of rat skin heparin proteoglycan (provided by Dr. Alan A. Horner, Toronto, Canada) (see Jacobsson et al., 1986), failed to produce any detectable amounts of 35S-labeled HA heparin product (data not shown).

DISCUSSION
The immediate product of heparin biosynthesis is a multichain proteoglycan with extended polysaccharide chains, M, 60-100 x lo3 Lindahl and Kjellen, 1987;. Within hours after completed formation, this proteoglycan undergoes intracellular degradation, the polysaccharide chains being cleaved by an endoglucuronidase into fragments of M, 5-30 X lo3 (Jacobsson and Lindahl, 1987). Such fragments constitute the bulk of commercial heparin preparations.
Analysis of heparin proteoglycan isolated from rat skin showed that the antithrombin-binding sequences are not randomly distributed but are accumulated in a restricted fraction of the polysaccharide chains (Jacobsson et al., 1986;Horner et al., 1988). Chain fragments containing the binding sequence will be designated HA heparin. Which factors determine whether a heparin sequence will become 3-O-sulfated, and thus ultimately recovered in HA heparin, or escape 3-0sulfation and remain an LA component?
Recent studies on the biosynthesis of heparin in a cell-free mastocytomal system have yielded some information concerning the relation between 3-0-sulfation and other polymer modification reactions. Experiments using an iV/O-desulfated, re-N-sulfated HA octasaccharide (isolated after partial deaminative cleavage of heparin) as [?S]sulfate acceptor thus indicated that the distinctive structural feature of HA heparin, i.e. the glucosaminyl3-O-sulfate group, is introduced at a late stage of the modification process ; see also the Introduction).
A labeled HA product was obtained which accounted for only -2% of the total incorporated Osulfate but for >90% of the glucosaminyl 3-O-sulfate groups. The 3-O-sulfate groups appeared at the appropriate position for the antithrombin-binding region (unit 4 of the structure shown in Fig. l), additional O-sulfate groups occurring at C-6 of units 2 and 6. The essential lack of 3-O-sulfate groups in fractions of lower antithrombin affinity suggested that the 3-0-sulfation reaction concludes the formation of the antithrombin-binding region and hence occurs only when all other structural requirements for binding have been fulfilled. These requirements include in particular the 6-O-sulfate group on unit 2 and the two N-sulfate groups on units 4 and 6. The Nsubstituent on unit 2, on the other hand, may be either acetyl or sulfate. Moreover, the D-gluco and ~-do configurations of units 3 and 5, respectively, are essential, as shown in recent experiments using GlcNS03-HexA-GlcNS03-HexA-GlcNSOZ pentasaccharides, with various permutations of the hexuronic components of the C-2 signal of GlcNS0,(3,6di-OSOJ + GlcNS03(3-OSOa) in the HA spectrum and the essential absence of both signals in the LA spectrum. acid units, as O-sulfate acceptors." The role of other O-sulfate groups appears to be less decisive, since 3-0-sulfation can apparently take place in the presence as well as in the absence of O-sulfate groups at C-6 of the target unit 4 and at C-2 of the adjacent unit 5 (Kusche et aZ., 1988). The importance of 6-0-sulfation at unit 6 is unclear. The lack of 3-O-sulfate groups in a fraction of heparin molecules could conceivably reflect the topographical organization of the biosynthetic apparatus, both the enzymes and the proteoglycan substrate involved in polymer modification being membrane bound . The location of the 3-0-sulfotransferase molecules could be such as to preclude the interaction of the enzyme with some of the polysaccharide chains. On the other hand, chains predestined to 4M. Kusche, R. Reynertson, L. Roden, and U. Lindahl, unpublished results. remain LA type might lack a functional acceptor site for 3-0sulfation.
The predominant acceptor sequence (in mucosal heparin), as predicted from the results summarized above, would have the structure (-IdoA)-GlcNAc(G-OS03)-GlcA-GlcNS03(6-OX)-IdoA(2-OX)-GlcNS03(6-OX)-(corresponding to units l-6 in Fig. l), where X stands for either -H or -SO,. The target for 3-0-sulfation is the internal glucosamine unit, which may be either unsubstituted or sulfated at C-6. The N-acetylated unit 2 may be replaced by N-sulfated glucosamine and then appears to be preceded by a 2-O-sulfated rather than unsubstituted iduronic acid unit (Lindahl et al., 1984); this latter arrangement is seen especially in lung heparin.
The present analysis of heparin preparations from various sources, using either HPLC of deamination products or NMR spectroscopy, indicated that the overall composition of HA and LA fractions was very similar; apart from the glucosa-minyl3-O-sulfate marker group, only minor quantitative differences were detected (Table II). LA heparin isolated from intestinal mucosa (preparation A) was selected for a more detailed structural analysis, aimed at identifying potential 3-O-sulfate acceptor sites. Special attention was given to the tetrasaccharides obtained after deamination, since these components would contain the N-acetylated glucosamine unit 2 and adjacent sugar residues, in all units l-4 of the acceptor sequences.
Two tetrasaccharides, IdoA-GlcNAc(G-OSO& GlcA-aManR and IdoA-GlcNAc(G-OS03)-GlcA-aManR(G-OSOs), implicated as markers for the 3-O-sulfate acceptor site , were found in appreciable amounts (Table I) together accounting for approximately 0.8 potential acceptor site/polysaccharide chain. Since at least one-third of the antithrombin-binding regions in mucosal HA heparin contain N-sulfated rather than N-acetylated glucosamine at position 2 (Lindahl et al., 1984), the occurrence of 3-O-sulfate acceptor sites that are undetectable by the present approach appears highly probable.
It thus seemed plausible that the majority of the LA chains might contain a target site for the glucosaminyl3-0-sulfotransferase.
This assumption was tested directly in attempts to use LA heparin as an exogenous acceptor for a mastocytoma micro-