Identification of efficiently cleaved substrates for HIV-1 protease using a phage display library and use in inhibitor development.

The recognition sequences for substrate cleavage by aspartic protease of HIV-1 are diverse and cleavage specificities are controlled by complex interactions between at least six amino acids around the cleavage site. We have identified 45 efficiently cleaved peptide substrates of HIV-1 protease (PR) using substrate phage display, an approach that can elucidate both context-dependent and context-independent preferences at individual subsites of a protease substrate. Many of the selected peptides were cleaved more efficiently and had lower K(m) values than physiologically relevant substrates of HIV-1 PR. Therefore, mutations occurring in the cleavage sites of the Gag and Gag-pol polyproteins of HIV-1 could significantly lower the K(m) values to better compete against drugs for protease binding while maintaining cleavage rates necessary for viral replication. The most efficiently cleaved peptide substrate derived from these phage, Ac-GSGIF*LETSL-NH(2), was cleaved 60 times more efficiently and had a K(m) approximately 260 times lower than a nine-amino-acid peptide based on the natural reverse transcriptase/integrase cleavage site when assayed at pH 5.6, 0.2 M NaCl. The peptide substrates selected served as frameworks for synthesis of tight binding reduced amide inhibitors of HIV-1 PR. The results show that the most efficiently cleaved substrates serve as the best templates for synthesis of the tightest binding inhibitors. Thus, defining changes in substrate preferences for drug-resistant proteases may aid in the development of more efficacious inhibitors.

Negative selectivity and the evolution of protease cascades: the specificity of plasmin for peptide and protein substrates.

BACKGROUND: Understanding the networks of selective proteolysis that regulate complex biological systems requires an appreciation of the molecular mechanisms used to maintain substrate specificity. Human plasmin, a serine protease that promotes the dissolution of blood clots and is essential in maintaining normal hemostasis, is usually described as having broad substrate specificity. Recent evidence that plasmin also plays a key role in a variety of other important biological and pathological processes, however, has suggested that this description might need to be re-evaluated. RESULTS: We used substrate phage display to elucidate optimal subsite occupancy for substrates of plasmin. We identified a peptide substrate that is cleaved 710,000-fold more efficiently by plasmin than a peptide containing the activation sequence of plasminogen. Plasmin achieves this unexpected, large differential activity even though both target sequences possess an arginine residue in the P1 position. We also demonstrate that proteolysis by plasmin can be targeted to an engineered protein substrate and that introduction of substrate sequences identified by phage display into plasminogen increases plasmin-mediated cleavage of the mutant 2000-fold. CONCLUSIONS: The specificity of plasmin is more tightly controlled than previously recognized; interactions with substrates at all subsites between S4 and S2' contribute to catalysis. Furthermore, in contrast to most enzymes that exhibit positive selectivity for substrate, the evolution of substrate specificity by plasmin has apparently been dominated by a strong negative selection against development of autoactivation activity. This 'negative selectivity' avoids short-circuiting regulation of the fibrinolytic system and other important biological processes, and might be an important general mechanism for controlling protease cascades.

Effect of plasminogen activators on human recombinant apolipoprotein(a) having the plasminogen activation cleavage site.

The serine-proteinase domain in human apolipoprotein(a) [apo(a)] and plasminogen exhibit 89% sequence identity including the catalytic triad. Cleavage of the Arg(561)-Val(562) activation site in plasminogen by either tissue- or urokinase-type plasminogen activator results in formation of the fibrinolytic enzyme plasmin. Apo(a) does not contain measurable amidolytic activity nor can it be activated by plasminogen activators. It has been suggested that the latter finding might be explained by the substitution of the plasminogen Arg-Val activation site by Ser-Ile in apo(a). To investigate if introduction of the Arg-Val activation site in apo(a) might result in sensitivity towards plasminogen activators, we expressed wild-type and Arg-Val mutant recombinant apo(a) [r-apo(a)] in human embryonic kidney and hepatocyte cell lines. Free r-apo(a) and lipoprotein-like particles [r-Lp(a)] were obtained in the culture supernatants of transfected 293 and HepG2 cells, respectively. Incubation of mutant r-apo(a)/r-Lp(a) with plasminogen activators produced neither plasmin-like activity nor cleavage at the Arg-Val activation site, even in the presence of various stimulators of plasminogen activation. Our data suggest that the high selectivity of activators for plasminogen activation requires interactions with regions in plasminogen distant from the activation disulfide loop which are not present in apo(a).

Distinct contributions of residue 192 to the specificity of coagulation and fibrinolytic serine proteases.

Archetypal members of the chymotrypsin family of serine proteases, such as trypsin, chymotrypsin, and elastase, exhibit relatively broad substrate specificity. However, the successful development of efficient proteolytic cascades, such as the blood coagulation and fibrinolytic systems, required the evolution of proteases that displayed restricted specificity. Tissue-type plasminogen activator (t-PA), for example, possesses exquisitely stringent substrate specificity, and the molecular basis of this important biochemical property of t-PA remains obscure. Previous investigations of related serine proteases, which participate in the blood coagulation cascade, have focused attention on the residue that occupies position 192 (chymotrypsin numbering system), which plays a pivotal role in determining both the inhibitor and substrate specificity of these enzymes. Consequently, we created and characterized the kinetic properties of new variants of t-PA that contained point mutations at position 192. These studies demonstrated that, unlike in coagulation serine proteases, Gln-192 does not contribute significantly to the substrate or inhibitor specificity of t-PA in physiologically relevant reactions. Replacement of Gln-192 with a glutamic acid residue did, however, decrease the catalytic efficiency of mature, two-chain t-PA toward plasminogen in the absence of a fibrin co-factor.

The antifibrinolytic effect of lipoprotein(a) in heterozygous subjects is modulated by the relative concentration of each of the apolipoprotein(a) isoforms and their affinity for fibrin.

Individuals heterozygous for the apolipoprotein(a) [apo(a)] trait have phenotypes combining two different lipoprotein(a) [La(a)] particle suspecies that are present in plasma at a different concentration. Evaluation of the ability of each of these isoforms to bind to fibrin and affect plasminogen binding is essential to assess the pathogenic role of Lp(a) in these subjects; therefore, fractions containing different ratios of Lp(a) with distinct apo(a) isoforms (e.g. B/S3, S1/S4) were prepared by density gradient ultracentrifugation of plasma, and tested. Lp(a) fractions containing mainly small apo(a) isoforms (either B or S1) showed the highest affinity for fibrin (Kd approximately 150 nmol L-1) and the best competitor activity for plasminogen, whereas fractions containing mainly the high molecular mass isoforms (either S3 or S4) showed the lowest affinities (Kd > or = 500 nmol L-1). An increase in Kd was observed as a function of the relative content in isoforms of high molecular mass in these fractions. This inverse relationship between affinity for fibrin and apo(a) size indicates that Lp(a) subspecies in heterozygotes may have different pathogenic potential. Thus, the antifibrinolytic effect of Lp(a) in heterozygous subjects would depend on the relative concentration of the isoform with the highest affinity for fibrin.

Multiple binding with identical linkage: a mechanism that explains the effect of lipoprotein(a) on fibrinolysis.

We have previously shown that both recombinant apo(a) and native Lp(a) inhibit the binding of Glu-plasminogen to fibrin surfaces [Fleury & Anglés-Cano (1991) Biochemistry 30, 7630-7638; Rouy et al. (1992) Biochemistry 31, 6332-6339]. The aim of the present study was to characterize the mechanism of this inhibition and to define the parameters governing binding when two different Lp(a) species compete with plasminogen for fibrin, a situation that may be found in vivo in subjects heterozygous for the apo(a) trait. The Kd for the binding of plasminogen to fibrin was 660 nM whereas the affinity of Lp(a) was inversely related to apo(a) size (Kd range: 50 to > 500 nM). To determine the effect of plasminogen on Lp(a) binding and reciprocally, competition experiments were performed. The Kd of either Lp(a) or plasminogen for fibrin remained unchanged in the presence of the other competitor whereas Bmax, the maximal amount bound, was importantly decreased. In a similar fashion, competition for fibrin binding among Lp(a) isoforms was shown with the use of Lp(a) density fractions containing varying proportions of isoforms B (approximately 460 kDa) and S3 (approximately 640 kDa); variations in Kd values (from 141 nM to 460 nM) as a function of the relative content in isoform S3 were observed. Altogether, these results are indicative of multiple binding by ligands that bind with different affinities to equivalent but independent sites. Thus, in plasma from heterozygous subjects, the influence of each Lp(a) isoform on fibrinolysis will depend on their affinity for fibrin and on their concentration relative to each other and to plasminogen.

Effects of lipoprotein(a) on the binding of plasminogen to fibrin and its activation by fibrin-bound tissue-type plasminogen activator.

Molecular assembly of plasminogen and tissue-type plasminogen activator (t-PA) at the surface of fibrin results in the generation of fibrin-bound plasmin and thereby in the dissolution of a clot. This mechanism is triggered by specific interactions of intra-chain surface lysine residues in fibrin with the kringle domains of plasminogen, and is further amplified via the interaction of plasminogen kringles with the carboxy-terminal lysine residues of fibrin that are exposed by plasmin cleavage. By virtue of its marked homology with plasminogen, apo(a), the specific apolipoprotein component of Lp(a), may bind to the lysine sites available for plasminogen on the surface of fibrin and thereby interfere with the fibrinolytic process. A sensitive solid-phase fibrin system, which allows the study of plasminogen activation at the plasma fibrin interface and makes feasible the analysis of products bound to fibrin, has been used to investigate the effects of Lp(a) on the binding of plasminogen and its activation by fibrin-bound t-PA. Plasma samples from human subjects with high levels of Lp(a) were studied. We have established that Lp(a) binds to the fibrin surface and thereby competes with plasminogen (Ki = 44 nM) so as to inhibit its activation. We have further shown that Lp(a) blocks specifically carboxy-terminal lysine residues on the surface of fibrin. To further explore the role of apo(a) on the Lp(a) fibrin interactions, we have performed ligand-binding studies using a recombinant form of apo(a) that contains 17 kringle 4-like units. We have shown that recombinant apo(a) binds specifically to fibrin (Kd = 26 +/- 8 nM, Bmax = 26 +/- 2 fmol/well) and that this binding increases upon treatment of the fibrin surface with plasmin (Kd = 8 +/- 4 nM, Bmax = 115 +/- 14 fmol/well). Altogether, our results indicate clearly that binding of native Lp(a) through this mechanism may impair clot lysis and may favor the accumulation of cholesterol in thrombi at sites of vascular injury.

Does apolipoprotein(a) heterogeneity influence lipoprotein(a) effects on fibrinolysis?

High plasma levels of lipoprotein(a) [Lp(a)] are considered to be an independent risk factor for premature cardiovascular disease and are inversely associated with apolipoprotein(a) [apo(a)] isoform sizes. The contribution of apo(a) polymorphism to the inhibition of fibrinolysis, a mechanism that may favor thrombus development, was therefore evaluated by measuring the ability of Lp(a) particles of distinct apo(a) isoform content to compete with plasminogen for fibrin binding during plasminogen activation by fibrin-bound tissue-type plasminogen activator. The rate of plasmin generation was most efficiently inhibited by an isoform with a molecular weight (M(r)) of approximately 540 Kd. An isoform with M(r) approximately 590 Kd produced a less pronounced effect, whereas the isoform with M(r) approximately 610 Kd failed to inhibit plasminogen activation. These effects were directly proportional to the amount of Lp(a) bound to the carboxy-terminal lysine residues of degraded fibrin. The relative affinity of the binding (kd range, 16 to 180 nmol/L) reflected the ability of individual Lp(a) isoforms to inhibit the binding of plasminogen (kd, 600 nmol/L). The question of the influence of kringle sequence variability on the binding to fibrin was not addressed by the present work. These data suggest that apo(a) isoform types with high affinity for fibrin may influence the ability of Lp(a) to interfere with fibrinolysis and contribute thereby to the association of elevated levels of Lp(a) with atherosclerotic and thrombotic risks.