The beta-form of antithrombin, lacking a carbohydrate side chain on Asn-135, is known to bind heparin more tightly than the fully glycosylated alpha-form. The molecular basis for this difference in affinity was elucidated by rapid-kinetic studies of the binding of heparin and the antithrombin-binding heparin pentasaccharide to plasma and recombinant forms of alpha- and beta-antithrombin. The dissociation equilibrium constant for the first step of the two-step mechanism of binding of both heparin and pentasaccharide to alpha-antithrombin was only slightly higher than that for the binding to the beta-form. The oligosaccharide at Asn-135 thus at most moderately interferes with the initial, weak binding of heparin to alpha-antithrombin. In contrast, the rate constant for the conformational change induced by heparin and pentasaccharide in the second binding step was substantially lower for alpha-antithrombin than for beta-antithrombin. Moreover, the rate constant for the reversal of this conformational change was appreciably higher for the alpha-form than for the beta-form. The carbohydrate side chain at Asn-135 thus reduces the heparin affinity of alpha-antithrombin primarily by interfering with the heparin-induced conformational change. These and previous results suggest a model in which the Asn-135 oligosaccharide of alpha-antithrombin is oriented away from the heparin binding site and does not interfere with the first step of heparin binding. This initial binding induces conformational changes involving extension of helix D into the adjacent region containing Asn-135, which are transmitted to the reactive-bond loop. The resulting decreased conformational flexibility of the Asn-135 oligosaccharide and its close vicinity to the heparin binding site destabilize the activated relative to the native conformation. This effect results in a higher energy for inducing the activated conformation in alpha-antithrombin, leading to a decrease in heparin binding affinity.

The interaction between cystatin C variants, in which the evolutionarily conserved Gly-11 residue was substituted by Ala, Glu or Trp, and the cysteine proteinases, papain, ficin, actinidin and cathepsin B, was characterized. The substitutions reduced the affinity of binding in a manner consistent with the Gly residue of the wild-type inhibitor, allowing the N-terminal region to adopt a conformation that was optimal for interaction with target proteinases. Replacement of Gly-11 by Ala resulted in only a 5- to 100-fold reduction in binding affinity. Comparison with the affinities of wild-type cystatin C lacking the N-terminal region indicated that even this small structural change affects the conformation of this region sufficiently to largely abolish its interaction with the weakly binding proteinases, actinidin and cathepsin B. However, the substitution allows interactions of appreciable strength between the N-terminal region and the tightly binding enzymes, papain or ficin. Replacement of Gly-11 with the larger Glu and Trp residues substantially decreased the affinity of binding to all enzymes, from 10(3)- to 10(5)-fold. These substitutions further affect the conformation of the N-terminal region, so that interactions of this region with papain and ficin are also essentially eliminated. The decreased affinities of the three cystatin C variants for papain, ficin and actinidin were due exclusively to increased dissociation rate constants. In contrast, the decreased affinity between cathepsin B and the Ala-11 variant, the only one for which rate constants could be determined with this enzyme, was due almost entirely to a decreased association rate constant. This behaviour is analogous to that observed for forms of cystatin C lacking the N-terminal region and supports the conclusion that the mode of interaction of this region with target proteinases varies with the enzyme as a result of structural differences in the active-site region of the latter.

The single Trp of human cystatin C, Trp-106, is located in the second hairpin loop of the proteinase binding surface. Substitution of this residue by Gly markedly altered the spectroscopic changes accompanying papain binding and reduced the affinity for papain, actinidin, and cathepsins B and H by 300-900-fold. The decrease in affinity indicated that the side chain of Trp-106 contributes a similar free energy, -14 to -17 kJ.mol-1, to the binding to all four cysteine proteinases, corresponding to about 20-30% of the total binding energy. Replacement of Trp-106 by Phe led to a smaller (30-120-fold) decrease in affinity for the four enzymes than Gly substitution. The binding energy of the Phe residue corresponded to 20-45% of that of Trp, showing that a phenyl group can only partly substitute for the indole ring. The reduced affinities of the cystatin C Trp-106 variants for all proteinases studied were due almost exclusively to increased dissociation rate constants. The second hairpin loop thus contributes to the binding primarily by keeping cystatin C anchored to the proteinase once the complex has been formed. This role is partly in contrast to that of the N-terminal region, which increases the affinity of cystatin C for cathepsin B by increasing the association rate constant. Removal of the N-terminal region of the Trp-106-->Gly variant by proteolytic cleavage substantially weakened the binding to papain and cathepsin B. The resulting affinity indicated that the first hairpin loop (the "QVVAG-region"), which is the only region of the proteinase binding surface remaining intact in the truncated variant, contributes 40-60% of the total free energy of binding of cystatin C to both proteinases.