Heparin sequencing brings structure to the function of complex oligosaccharides

Heparin is a highly sulfated linear polysaccharide that was discovered in 1916 as an anticoagulant and has been used clinically since 1935 (see ref. 1 for review). The anticoagulant activity of heparin results from its binding to the serine protease inhibitor, antithrombin III (ATIII) (2, 3). Heparin binding causes a conformational change in ATIII that results in enhanced inhibition of thrombin and other serine proteases involved in the blood clotting cascade (Fig. 1 A and B). Whereas the clinical reagent heparin is found nearly exclusively in mast cells, it represents a specialized member of the widely distributed class of compounds known as heparan sulfate. Heparan sulfate is found throughout all tissues in virtually every animal species examined, most prominently on cell surfaces and within extracellular matrices. Heparin and heparan sulfate are characterized by repeating units of disaccharides containing a uronic acid (glucuronic or iduronic acid) and glucosamine, which is either N-sulfated or N-acetylated. The sugar residues may be further O-sulfated at the C-6 and C-3 positions of the glucosamine and the C-2 position of the uronic acid. Thus, there are at least 32 potential unique disaccharide units that together make this class of compounds one of the most information dense in biology. The high degree of structural complexity likely underlies the ability of heparan sulfate to play critical roles in a large and diverse number of biological processes (1). However, the excitement generated by the growing list of functions attributed to heparan sulfate is often met by the sobering reality that the methodology for carbohydrate structural analysis has lagged well behind that for proteins and nucleic acids.

Figure 1.

Hypothetical mechanisms for heparin/heparan sulfate regulation of protein–protein interactions. (A) The interaction of ATIII with the serine protease factor Xa (Xa) is catalyzed by a conformational change in ATIII by binding to a specific pentasaccharide sequence. (B) The interaction of ATIII with thrombin (IIa) is catalyzed by a heparin-induced conformational change in ATIII and the nonspecific binding of thrombin to heparin to bring the two proteins together (approximation model). (C) The interaction of heparan sulfate proteoglycans (HSPG) with FGF-2 and its receptor creates a high affinity multimeric complex.

It has been only in recent years that methods for structural analysis of heparan sulfate have begun to become widely available. Separation techniques such as HPLC and capillary electrophoresis, combined with the use of specific enzymatic and chemical methods to depolymerize heparan sulfate, have provided important information on composition and domain structure (4–6). Approaches using NMR and mass spectrometry have allowed accurate sequencing of small saccharides (7, 8). More recently, methods have been developed that use integrated strategies with chemical and enzymatic steps coupled to accurate separation methods (9–11). One of the most promising of these approaches is the recently described use of a property-encoded nomenclature in conjunction with matrix assisted laser desorption mass spectrometry (PEN-MALDI) (11). In two papers in this issue of PNAS, Sasisekharan and his colleagues put these methods to practice (12, 13). As a test of their method, they used PEN-MALDI to sequence a decasaccharide (AT10), derived from heparin, which was believed to be of known sequence (14). However, the PEN-MALDI analysis suggested a different sequence (Fig. 2) (13). To reconcile this apparent conflict, the authors used several analytical techniques, including integral glycan sequencing and one-dimensional proton NMR to converge on the structure of AT10. Armed with definitive knowledge of the structure of AT10, Shriver et al. (12) were then able to probe the anticoagulant mechanisms of a heparin fragment containing only a partial ATIII binding site, as well as to evaluate potential ways to generate low molecular weight heparin for clinical applications. Together, these papers provide an excellent paradigm for what will undoubtedly become a large number of studies aimed at defining the structure–function relationships for specific heparin- and heparan sulfate-derived oligosaccharides.

Figure 2.

Heparin sequences. (A) The consensus ATIII binding pentasaccharide sequence (red). I, α-l-iduronic acid; G, β-d-glucuronic acid; ΔU, 4,5 unsaturated uronic acid; H, glucosamine; NAc, N-acetylation; NS, N-sulfation; 2S, 3S, 6S, sulfation. (B) The sequence for AT10 determined in Shriver et al. (13). (C) Previously published sequence of AT10 containing an intact ATIII binding site.

The reaction of heparin with antithrombin and proteases has been studied in great detail. The two principle mechanisms of action involve the binding of ATIII to a specific pentasaccharide sequence that is found in only about one third of the chains in commercial heparin preparations (Fig. 2A). The initial binding of heparin to ATIII causes a conformational change that results in further interactions between ATIII and heparin. The heparin-ATIII complex is competent to interact with and inhibit factor Xa. Inhibition of thrombin, on the other hand, depends on thrombin also binding to the heparin chain at a site proximate to the pentasaccharide ATIII binding site (Fig. 1 A and B). Thus, neutralization of factor Xa can be achieved with a pentasaccharide whereas the minimum length for effective neutralization of thrombin is 18 saccharide units (2, 3). Whereas the binding of ATIII to heparin shows extreme structural constraints, with a particular requirement for the unusual 3-O-sulfate on the internal glucosamine residue, the binding of thrombin to heparin shows very little sequence specificity. From this single example of heparin–protein interactions, it is clear that specific and nonspecific binding interactions can have important biological consequences. This issue is likely to have wide-ranging implications, because the list of proteins that bind heparin has grown to well over 100 (1). With heparan sulfate representing a major component of the extracellular matrix and being widely distributed on cell surfaces, it is well positioned to play a prominent role in orchestrating the flow of information to and from cells within multicellular organisms.

On a practical level, the large number of heparin-binding interactions can cause significant unwanted side effects of pharmacologically active heparin (e.g., thrombocytopenia). For these reasons, low molecular weight fractions of heparin have been developed by controlled chemical or enzymatic breakdown of heparin (15). Useful low molecular weight fractions would ideally retain the pentasaccharide binding site for ATIII, yet show limited interactions with other proteins. To this end, the enzymes heparinase I, II, and III from Flavobacterium heparinum have been used to generate low molecular weight heparin fragments (15). Heparinase I was used to generate the model decasaccharide AT10 (14). Thus, the unexpected finding that AT10 contains only a portion of the ATIII pentasaccharide sequence (12, 13) suggests that linkages with a 3-O-sulfated glucosamine are not resistant to cleavage, as previously thought. The further characterization of all three heparinases with model oligosaccharides containing complete and partial ATIII binding sites indicates that heparinases I and II are not likely to be as useful as previously believed for the generation of low molecular weight heparin fragments containing an intact ATIII binding site. In contrast, the data presented demonstrate that heparinase I and II may be ideal agents for the neutralization of the anticoagulant activity of pharmacological doses of heparin (16). The functional analysis of the partial ATIII site contained in AT10 (12) provides independent support for the proposed mechanism for the ATIII-binding pentasaccharide, where the two saccharide residues on the reducing end of the sequence are not involved in the initial binding to native ATIII but, instead, greatly affect conversion of ATIII to the high-heparin-affinity state (17). The power of this approach lies in the ability to relate the function of an oligosaccharide derived from a biological source, heparin, to its structure. There are specific clinically relevant conclusions from these studies, and, at the same time, they represent a general move toward the chemical mechanism-based analysis of heparin–protein interactions.

With the advent of new methods for the structural analysis of complex oligosaccharides, it is now possible to begin to define the biological functions of heparan sulfate–protein interactions at a mechanistic level. Whereas the importance of heparin as a pharmacological reagent has been clear for some time, recent years have witnessed an explosion in the number of studies aimed at identifying the biological consequences of heparan sulfate–protein interactions. Probably the best example of a critical heparan sulfate interaction is with the fibroblast growth factors (FGFs) (see ref. 1 for review). The FGFs are a family of proteins that affect the growth, migration, differentiation, and survival of nearly all cell types throughout the animal kingdom. One of the prominent characteristics of FGFs is their ability to bind to heparin and heparan sulfate. Whereas this characteristic dramatically facilitated the purification of these growth factors and made the proteins more resistant to denaturation and proteolytic degradation, it is now widely accepted that the binding of FGFs to heparan sulfate has significant functional impact.

The primary findings with FGF-2 showed that high affinity binding of this growth factor to its receptor is dramatically enhanced by heparin and heparan sulfate (18, 19). Whereas the mechanistic details of this process remain to be defined, there are clear similarities to, and distinctions from, that of the ATIII/thrombin/factor Xa systems. As with ATIII, FGF-2 binding sites on heparin and heparan sulfate appear to have important general properties, such as dependence on particular sulfate groups. A current model for heparan sulfate-mediated FGF-2 receptor binding includes a minimum length requirement for heparan sulfate to potentially allow it to bridge two FGF-2 monomers with receptors (Fig. 1C) (20, 21). It appears that the heparan sulfate chain functions by binding to both the growth factor and its receptor in an approximation model reminiscent of the ATIII/thrombin system. As such, the minimum fragment length required for FGF-2 binding, a hexasaccharide, is insufficient for receptor activation. Furthermore, 2-O-sulfation and not 6-O-sulfation is required for binding to FGF-2, yet 6-O-sulfation appears necessary for receptor activation (22, 23). The specific sequences of heparan sulfate expressed on different cell types can have varying capabilities for binding FGF-2 and potentiating receptor activation, suggesting the possibility that heparan sulfate acts as an exquisitely sensitive tuning device to modulate cellular response. As one considers the possibilities of hundreds of proteins vying for binding sites of varying specificity on heparan sulfate, the complexity becomes massive. The development of physical methods to analyze heparan sulfate structure, in combination with the new methods for complex systems analysis being established, within the nascent field of bioinformatics (24), portends a quantum shift in glycobiology, bringing it to the forefront of biology along side genomics and proteomics.

The functional role of heparan sulfate as a universal modulator of protein signals is likely much more significant than at first believed. In addition to the above types of direct protein-protein interactions that heparin and heparan sulfate can modulate, it has also become apparent that heparan sulfate can have other effects. In particular, heparan sulfate within the extracellular matrix might function as a general modulator of the movement of heparin-binding proteins (25). In this way, heparan sulfate–protein interactions can provide an intriguing mechanism for the generation of growth factor and cytokine gradients of varying pitch and duration to direct the coordinated development of complex organisms. These same processes might then be recapitulated to facilitate tissue repair after injury or disrupted to contribute to pathologic events. As one example, angiogenesis, the process of new blood vessel growth, is critical for the repair of damaged tissue. However, when angiogenesis is coopted by malignant tumors, it dramatically increases the aggressiveness and severity of disease. In recent years, several new angiogenic stimulators and inhibitors have been discovered, and, remarkably, almost all of these proteins possess heparin-binding activity (26). Although the role of heparin binding has not been delineated for most angiogenic factors, it is tempting to speculate that new blood vessel growth is controlled by a complex interplay between heparan sulfate and the combination of stimulatory and inhibitory heparin-binding proteins. The dream of pharmacologically controlling angiogenesis to stimulate repair and eradicate tumors will likely require a more complete understanding of the structure/function relationships between angiogenic factors and their heparan sulfate binding sites.

Heparan sulfate has also been implicated in modulating intracellular events. It has been suggested that binding endocytosis and metabolism of lipoproteins is mediated, in part, through interactions of lipases and apolipoproteins with cell surface heparan sulfate proteoglycans (1). In the FGF-2 system, it appears that endocytosis can occur directly via heparan sulfate proteoglycans, and may be fated to different catabolic pathways than when internalized on its receptors alone (27, 28). Interestingly a large fraction of internalized FGF-2 is translocated to the nucleus, correlating with mitogenesis (29). Consistent with a possible role for heparan sulfate in this process, there is considerable evidence that heparan sulfate is also translocated to the nucleus (1). Thus, one can speculate that heparan sulfate plays critical roles in regulating specific transcriptional events, or on a more general scale, these highly sulfated polysaccharides might have pleiotropic effects on nucleosome and chromatin structure.

The ability of heparin/heparan sulfate to bind to so many different proteins with high affinity is somewhat unique to this class of glycosaminoglycans. This characteristic is likely a reflection of the high iduronic acid content, which presents a certain amount of structural flexibility and leads to a high density of charged sulfate groups (1). However, the specific molecular contacts between these binding partners have been defined for relatively few cases. Hence, attempts to define general properties of “heparin-binding domains” have been of limited success (30, 31). In these instances, a major limitation was the lack of information available on the structure of the heparan sulfate chains. Thus, the search, to date, has been for generic protein motifs that would bind to uniform heparin-like polysaccharides. As the structure of each interaction is defined, from the standpoint of both the protein and polysaccharide, a more complete picture will be revealed. The example used in the present studies of ATIII/heparin might provide a valuable lesson, when one considers the findings that a partial binding site retained significant ATIII binding and anti-factor Xa activity, yet was clearly less active when compared with the complete sequence (12, 13). Considering the ATIII/heparin/thrombin and FGF-2/heparin interactions together, one begins to get a picture of the range of possibilities with respect to sequence specificity and the possible molecular mechanisms in which heparin and heparan sulfate might participate.

In summary, the field of heparin–protein binding has grown dramatically from its anticoagulant beginnings to one where protein–polysaccharide interactions are rapidly becoming recognized as critical controlling elements throughout biology. The major limitation in this field has been the lack of analytical tools. The recent development and use of new approaches for the analysis of heparin structure is a critical advance in this area. Periods of great discovery in science are almost always preceded by the development of new tools. The challenge will be in the use of these new tools to decode the complexities of nature.