. Predictions were then made as to the heparln-blnding domains In endothellal cell growth factor, purpurln, and antithrombln-lll. Many of the natural sequences conforming to these consensus motifs show prominent amphlpathlc periodicities having both a-hellcal and 0-strand conformations as determined by predictive algorithms and circular dlchroism studies. The heparln-blnding domain of vttronectln was modeled and formed a hydrophlllc pocket that wrapped around and folded over a heparln octasaccharide, yielding a complementary structure. We suggest that these consensus sequence elements form potential nucleation sites for the recognition of potyanions In proteins and may provide a useful guide In identifying heparln-blnding regions In other proteins. 1112 However, the structural heterogeneity of the acidic mucopolysaccharides with respect to size, carbohydrate composition and charge, 13 and the variety of proteins that are known to bind to these molecules have complicated a detailed molecular analysis of protein-GAG interactions. The most studied system, binding of heparin to a specific domain or site on human antithrombin-lll (AT-III), involves a unique carbohydrate structure representing only a fraction of the total heparin. 14 The exact amino acid residues on AT-III or other proteins which comprise their heparin-binding domains have not been fully elucidated.To understand the molecular recognition of heparin by proteins, we recently determined the structure of the heparin-binding regions of apolipoprotein (apo) B-100, the major protein constituent of human plasma low density Received June 10, 1988; revision accepted September 14, 1988. lipoproteins (LDL). The structure of these regions was of particular interest as we hypothesized that they could mediate the interaction of LDL with acidic mucopolysaccharides of arterial tissue. We showed that LDL contains five to seven heparin-binding sites on the lipoprotein surface. 15 The heparin-binding cyanogen bromide peptides of apo B-100 were isolated, and their amino acid sequences were determined. 16 These peptides account for five domains of high positive charge density that we suggest mediate the lipoprotein's interaction with glycosaminoglycans. These same regions were also identified by Weisgraber and Rail. 17 We noticed 16 that these domains in apo B-100 have a sequence similarity to the heparinbinding regions of apo E 1819 and human vitronectin (Vn) 20 with respect to the organization of basic and hydropathic residues. The present study extends our initial observations to other proteins whose functions are modified by heparin. We show that all these proteins contain unique consensus sequences of amino acid residues that are potentially involved in heparin binding.
MethodsSecondary structure predictions of Vn peptide conformations were performed by the MSEQ program 21 based on Chou-Fasman algorithms 22 and by the method of Gamier et al. 23 Both methods gave comparable predictions of predominant /3-strand and 0-tum structures, and circular dichroism studies on...