Involvement of the urokinase kringle domain in lipopolysaccharide-induced acute lung injury.

Urokinase plasminogen activator (uPA) plays a major role in fibrinolytic processes and also can potentiate LPS-induced neutrophil activation through interactions with its kringle domain (KD). To investigate the role of the uPA KD in modulating acute inflammatory processes in vivo, we cloned and then developed Abs to the murine uPA KD. Increased pulmonary expression of uPA and the uPA KD was present in the lungs after LPS exposure. Administration of anti-kringle Abs diminished LPS-induced up-regulation of uPA and uPA KD in the lungs, and also decreased the severity of LPS-induced acute lung injury, as determined by development of lung edema, pulmonary neutrophil accumulation, histology, and lung IL-6, MIP-2, and TNF-alpha cytokine levels. These proinflammatory effects of the uPA KD appeared to be mediated through activation of Akt and NF-kappaB. The present studies indicate that the uPA KD plays a major role in the development of TLR4-mediated acute inflammatory processes, including lung injury. Blockade of the uPA KD may prevent the development or ameliorate the severity of acute lung injury induced through TLR4-dependent mechanisms, such as would occur in the setting of Gram-negative pulmonary or systemic infection.

Antibodies to tissue-type plasminogen activator (t-PA) in patients with inflammatory bowel disease: high prevalence, interactions with functional domains of t-PA and possible implications in thrombosis.

BACKGROUND: Patients with inflammatory bowel disease (IBD) have an increased prevalence of thromboembolic events. The pathogenetic mechanisms of these events include reduced fibrinolysis, which may be caused by antibodies to tissue-type plasminogen activator (t-PA). OBJECTIVES: To evaluate anti-t-PA antibodies in patients with IBD, considering clinical, biochemical and functional characteristics. PATIENTS AND METHODS: We immunoenzymatically measured anti-t-PA antibodies in plasma from 97 consecutive IBD patients and 97 age- and sex-matched healthy controls. We also assessed the antibody interactions with different epitopes of t-PA, the antibody inhibition on t-PA activity and the correlations with clinical features and other serum antibodies. RESULTS: IBD patients had higher median anti-t-PA antibody levels (5.4 U mL(-1) vs. 4.0 U mL(-1); P < 0.0001): 18 patients were above the 95th percentile of the controls (OR 5.3; 95% CI 1.7-16.3; P < 0.003), and the six with a history of thrombosis tended to have high levels (6.9 U mL(-1)). Anti-t-PA antibody levels did not correlate with IBD type, activity, location or treatment, or with age, sex, acute-phase reactants or other antibodies. The anti-t-PA antibodies were frequently IgG1 and bound t-PA in fluid phase; they recognized the catalytic domain in 10 patients and the kringle-2 domain in six. The IgG fraction from the three patients with the highest anti-t-PA levels slightly reduced t-PA activity in vitro. CONCLUSIONS: The prevalence of anti-t-PA antibodies is high in IBD patients. By binding the catalytic or kringle-2 domains of t-PA, these antibodies could lead to hypofibrinolysis and contribute to the prothrombotic state of IBD.

Lupus-derived antiprothrombin autoantibodies from a V gene phage display library are specific for the kringle 2 domain of prothrombin.

Autoantibodies to prothrombin are common in patients with systemic lupus erythematosus. Although their presence is a risk factor for thrombosis, neither their origin nor their precise role in inducing the procoagulant state is known. We have developed a phage-display antibody library from patients with systemic lupus erythematosus with antiprothrombin antibodies, and we have selected two single-chain Fv antibody fragments (ScFvs) by panning on a prothrombin-coated surface. In prothrombin activation assays using purified components, these antibodies promoted prothrombin activation. These ScFvs, termed AN78 and AN129, bound to immobilized prothrombin in a concentration-dependent specific manner but not to other anionic phospholipid binding proteins such as beta2-glycoprotein I or annexin V. Phosphatidylserine-bound prothrombin, but not soluble prothrombin, inhibited the binding suggesting that the epitope is available only on immobilized prothrombin. To localize the epitope, prothrombin was treated with thrombin or factor Xa and various prothrombin activation fragments were subsequently isolated and tested in ELISA with the ScFvs. Both AN78 and AN129 bound to prethrombin I (the fragment lacking the Gla domain and the first kringle domain), to fragment 1.2 (containing Gla and the two kringle domains only) and to fragment 2 but not to thrombin, thus localizing the cognate epitope to the kringle 2 domain in prothrombin. Analysis of the cDNA sequences of these antibodies show clustered mutational patterns in the complementarity determining region, suggesting that variable domains are the products of antigen-driven B cell clonal maturation.

Lp(a) particles mold fibrin-binding properties of apo(a) in size-dependent manner: a study with different-length recombinant apo(a), native Lp(a), and monoclonal antibody.

OBJECTIVE: Small-sized apolipoprotein(a) [apo(a)] isoforms with high antifibrinolytic activity are frequently found in cardiovascular diseases, suggesting a role for apo(a) size in atherothrombosis. To test this hypothesis, we sought to characterize the lysine (fibrin)-binding function of isolated apo(a) of variable sizes. METHODS AND RESULTS: Recombinant apo(a) [r-apo(a)] preparations consisting of 10 to 34 kringles and a monoclonal antibody that neutralizes the lysine-binding function were produced and used in parallel with lipoprotein(a) [Lp(a)] particles isolated from plasma in fibrin-binding studies. All r-apo(a) preparations displayed similar affinity and specificity for lysine residues on fibrin regardless of size (K(d) 3.6+/-0.3 nmol/L) and inhibited the binding of plasminogen with a similar intensity (IC50 16.8+/-5.4 nmol/L). In contrast, native Lp(a) particles displayed fibrin affinities that were in inverse relationship with the apo(a) kringle number. Thus, a 15-kringle apo(a) separated from Lp(a) and a 34-kringle r-apo(a) displayed an affinity for fibrin that was higher than that in the corresponding particles (K(d) 2.5 versus 10.5 nmol/L and K(d) 3.8 versus 541 nmol/L, respectively). However, fibrin-binding specificity of the r-apo(a) preparations and the Lp(a) particles was efficiently neutralized (IC50 0.07 and 4 nmol/L) by a monoclonal antibody directed against the lysine-binding function of kringle IV-10. CONCLUSIONS: Our data indicate that fibrin binding is an intrinsic property of apo(a) modulated by the composite structure of the Lp(a) particle.

Production and characterization of monoclonal antibodies directed to the kringle V and protease domains of human apolipoprotein(a).

Production and use of anti-apolipoprotein(a) monoclonal antibodies (MAbs) specific to single copy regions in the polymorphic lipoprotein(a) (Lp(a)) has been emphasized to be important for the standardization of measurements of the coronary heart disease risk factor, Lp(a). Here, mouse MAbs were prepared against the kringle V (V) and protease (P) domains of human apolipoprotein(a) (apo(a)), which domains are present in single copy in the apo(a) molecule. The cDNA for apo(a)VP was cloned from human liver cDNA library, and the V-P recombinant protein overexpressed in Escherichia coli was used as an antigen for the antibody production. Two antibodies named as MAb(a)20 and MAb(a)23 were finally produced, and they were characterized for their binding specificity and epitopes. The specificity of the antibodies was confirmed by an immunoblotting procedure and an enzyme-linked immunoassay (ELISA). It was shown that the antibodies had little, if any, cross-reactivity with human plasminogen, which is relatively abundant in human serum and is highly homologous (85%) with apo(a) in amino acid (aa) sequence. For epitope analysis, 3'-deletional series of apo(a)VP cDNA were constructed, and expression products of them were analyzed for the binding MAb(a)20 and MAb(a)23 do. It has been revealed that distinct epitopes were recognized by the two MAbs: MAb(a)23 (gamma2b, kappa) bound to the V region about 60 aa downstream from the N-terminal, and MAb(a)20 (gamma1, kappa) bound to the P region close to the C-terminal. A one step-sandwich ELISA system for Lp(a) was developed using MAb(a)20 as a capturing antibody and horseradish peroxidase (HRP)-coupled MAb(a)23 as a detecting antibody. The assay was found to be sensitive and useful for detecting Lp(a) in the range of 4-150 microg/dL (80 pM-3 nM).

A battery of monoclonal antibodies that induce unique conformations to evolve cryptic but constitutive functions of plasminogen.

Two groups of anti-plasminogen monoclonal antibodies, whose epitope was either in the kringle 1 + 2 + 3 domain (F3P2, F11P5, F11P6, and F12P18) or the kringle 5 domain (F1P6 and F12P16), were isolated and their effects on the conformation of plasminogen were explored. All antibodies except F1P6 had 3- to 10-fold higher affinity toward Lys-plasminogen than Glu-plasminogen. F1P6 exhibited a comparable affinity to Glu- and Lys-plasminogen. Among these, only F11P5 binding was inhibited by epsilon-amino-nu-caproic acid (EACA) in a concentration-dependent manner, with half maximal inhibition at 3 mM. From a competition assay, we concluded that the epitopes of F11P5, F11P6, and F12P18 should be very close, and located at or near the low affinity lysine binding site on the kringle 2 + 3. These three antibodies dramatically enhanced the binding of Glu-plasminogen to the other antibodies, except to F1P6. Interestingly, F3P2, whose non-overlapping epitope was in the kringle 2 + 3 domain, also augmented the binding of Glu-plasminogen to the other antibodies. In contrast, we did not observe enhanced binding of Lys-plasminogen to one antibody in the presence of the other antibodies, and the binding of Glu-plasminogen to these antibodies did not increase in the presence of 10 mM EACA. In the presence of these antibodies, including F1P6, Glu-plasminogen bound more efficiently to immobilized degraded fibrin, with a binding profile similar to Lys-plasminogen. All antibodies except F1P6 enhanced the conversion rate of plasminogen to plasmin remarkably. Taken together, we propose that these two groups of monoclonal antibodies can dissociate the intramolecular interactions of Glu-plasminogen and induce the conformational transition of Glu-plasminogen to Lys-plasminogen. In addition, the kringle 2 + 3 and kringle 5 structures of Glu-plasminogen liganded with EACA are distinct from the Lys-plasminogen structure.

LDL-unbound apolipoprotein(a) and carotid atherosclerosis in hemodialysis patients.

High lipoprotein(a) [Lp(a)] plasma concentrations, which are genetically determined by apo(a) size polymorphism, are directly associated with an increased risk for atherosclerosis. Patients with end-stage renal disease (ESRD), who show an enormous prevalence of cardiovascular disease, have elevated plasma concentrations of Lp(a). In recent studies we were able to show that apo(a) size polymorphism is a better predictor for carotid atherosclerosis and coronary artery disease in hemodialysis patients than concentrations of Lp(a) and other lipoproteins. Less than 5% of apo(a) in plasma exists in a low-density lipoprotein (LDL)-unbound form. This "free" apo(a) consists mainly of disintegrated apo(a) molecules of different molecular weight, ranging from about 125 to 360 kDa. LDL-unbound apo(a) molecules are elevated in patients with ESRD. The aim of this study was therefore to investigate whether the LDL-unbound form of apo(a) contributes to the prediction of carotid atherosclerosis in a group of 153 hemodialysis patients. The absolute amount of LDL-unbound apo(a) showed a trend to increasing values with the degree of carotid atherosclerosis, but the correlation of Lp(a) plasma concentrations with atherosclerosis was more pronounced. In multivariate analysis the two variables were related to neither the presence nor the degree of atherosclerosis. Instead, the apo(a) phenotype took the place of Lp(a) and LDL-unbound apo(a). After adjustment for other variables, the odds ratio for carotid atherosclerosis in patients with a low molecular weight apo(a) phenotype was about 5 (p<0.01). This indicates a strong association between the apo(a) phenotype and the prevalence of carotid atherosclerosis. Finally, multivariate regression analysis revealed age, angina pectoris and the apo(a) phenotype as the only significant predictors of the degree of atherosclerosis in these patients. In summary, it seems that LDL-unbound apo(a) levels do not contribute to the prediction of carotid atherosclerosis in hemodialysis patients. However, this does not mean that "free", mainly disintegrated, apo(a) has no atherogenic potential.

High plasma levels of apo(a) fragments in Caucasians and African-Americans with end-stage renal disease: implications for plasma Lp(a) assay.

Apolipoprotein(a) [apo(a)] is a plasma glycoprotein that is highly polymorphic in size due to differences in the number of a tandemly arrayed cysteine-rich repeat called kringle (K)4 at its N-terminus. Most plasma apo(a) is covalently attached to apolipoprotein B-100 and circulates as part of lipoprotein(a) [Lp(a)]. A fraction of apo(a) circulates free of lipoproteins. Almost all of the free apo(a) consists of fragments containing variable numbers of K4 repeats derived from the N-terminal region. Previously we provided evidence suggesting that the apo(a) fragments present in human plasma are the source of the apo(a) fragments in human urine. If this were the case, it would be expected that plasma levels of fragments would be higher in subjects with end-stage renal disease (ESRD). In this paper we quantified the levels of apo(a) fragments and plasma Lp(a) in 26 Caucasian and 26 African-American subjects with ESRD and 52 healthy subjects matched for race, sex and the size of the apo(a) isoforms. The plasma levels of apo(a) fragments and Lp(a) were both higher in the ESRD subjects. In addition, the ratio of apo(a) fragments to total immunodetectable apo(a) was increased in ESRD. To determine how much the increase in the apo(a) fragments contributed to the increase in plasma Lp(a) in ESRD, the plasma Lp(a) levels were measured employing two different anti-apo(a) enzyme-linked immunoabsorption assays (ELISA). One assay detected both free and bound apo(a), whereas the other assay detected only bound apo(a). Although the plasma levels of apo(a) in the ESRD subjects tended to be higher using the assay that detected both fragments and full-length apo(a), the increase was modest. Thus, although a greater proportion of the apo(a) in ESRD plasma circulates as fragments, most of the elevation in plasma levels of Lp(a) associated with renal insufficiency is due to an increase in intact Lp(a).

Kringle-containing fragments of apolipoprotein(a) circulate in human plasma and are excreted into the urine.

Apolipoprotein(a) [apo(a)] contains multiple kringle 4 repeats and circulates as part of lipoprotein(a) [Lp(a)]. Apo(a) is synthesized by the liver but its clearance mechanism is unknown. Previously, we showed that kringle 4-containing fragments of apo(a) are present in human urine. To probe their origin, human plasma was examined and a series of apo(a) immunoreactive peptides larger in size than urinary fragments was identified. The concentration of apo(a) fragments in plasma was directly related to the plasma level of Lp(a) and the 24-h urinary excretion of apo(a). Individuals with low (< 2 mg/dl) plasma levels of Lp(a) had proportionally more apo(a) circulating as fragments in their plasma. Similar apo(a) fragments were identified in baboon plasma but not in conditioned media from primary cultures of baboon hepatocytes, suggesting that the apo(a) fragments are generated from circulating apo(a) or Lp(a). When apo(a) fragments purified from human plasma were injected intravenously into mice, a species that does not produce apo(a), apo(a) fragments similar to those found in human urine were readily detected in mouse urine. Thus, we propose that apo(a) fragments in human plasma are derived from circulating apo(a)/Lp(a) and are the source of urinary apo(a).

Monoclonal antibody F1 binds to the kringle domain of factor XII and induces enhanced susceptibility for cleavage by kallikrein.

In a previous study we have shown that monoclonal antibody F1 (MoAb F1), directed against an epitope on the heavy chain of factor XII distinct from the binding site for anionic surfaces, is able to activate factor XII in plasma (Nuijens JH, et al: J Biol Chem 264; 12941, 1989). Here, we studied in detail the mechanism underlying the activation of factor XII by MoAb F1 using purified proteins. Formation of factor XIIa was assessed by measuring its amidolytic activity towards the chromogenic substrate H-D-Pro-Phe-Arg-pNA (S-2302) in the presence of soybean trypsin inhibitor and by assessing cleavage on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Upon incubation with MoAb F1 alone, factor XII was auto-activated in a time-dependent fashion, activation being maximal after 30 hours. Factor XII incubated in the absence of MoAb F1 was hardly activated by kallikrein, whereas in the presence of MoAb F1, but not in that of a control MoAb, the rate of factor XII activation by kallikrein was promoted at least 60-fold. Maximal activation of factor XII with kallikrein in the presence of MoAb F1 was reached within 1 hour. This effect of kallikrein on the cleavage of factor XII bound to MoAb F1 was specific because the fibrinolytic enzymes plasmin, urokinase, and tissue-type plasminogen activator could not substitute for kallikrein. Also, trypsin could easily activate factor XII, but in contrast to kallikrein, this activation was independent of MoAb F1. SDS-PAGE analysis showed that the appearance of amidolytic activity correlated well with cleavage of factor XII. MoAb F1-induced activation of factor XII in this purified system was not dependent on the presence of high-molecular-weight kininogen (HK), in contrast to the activation of the contact system in plasma by MoAb F1. Experiments with deletion mutants revealed that the epitopic region for MoAb F1 on factor XII is located on the kringle domain. Thus, this study shows that binding of ligands to the kringle domain, which does not contribute to the proposed binding site for negatively charged surfaces, may induce activation of factor XII. Therefore, these findings point to the existence of multiple mechanisms of activation of factor XII.

Kringle 4 of human apolipoprotein[a] shares a linear antigenic site with human catalase.

Monoclonal antibody (mab) 1A2, directed against human apolipoprotein[a] (apo[a]), revealed a strong reaction with peroxisomes as shown by immuno-gold labeled cryosections of human liver biopsies. This reactivity was not due to the presence of apo[a] in peroxisomes but to a cross-reactivity of mab 1A2. Immunoblot analysis of peroxisomal fractions and purified human catalase demonstrated that mab 1A2 reacts with catalase. Conversely, an anti-catalase antibody also recognized apo[a]. By sequence comparison we identified a 4-amino acid motif (Y-Y-P-N) that is shared between the highly repetitive kringle 4 motif of apo[a] and the carboxy-terminal third of the peroxisomal marker enzyme catalase. No other identical sequences were identified in these proteins. Results from the following experiments indicated that 1A2 recognizes this short linear epitope. i) Mab 1A2 reacted only with the 4 amino acid peptide sequence in a pin-ELISA using immobilized overlapping peptides. ii) A synthetic peptide including this sequence completely inhibited the 1A2 immunoreactivity to apo[a] and catalase. iii) A recombinant fusion protein tagged with the putative epitope was recognized by mab 1A2. Our findings demonstrate that unknown linear epitopes in native proteins can be identified by sequence comparison between known proteins. The practical implication is that antibodies against apo[a] must be controlled for this cross-reactivity before using them for immunohistochemical studies of intracellular apo[a] in tissues or cells.

Influence of apolipoprotein(a) phenotype on lipoprotein(a) quantification: evaluation of three methods.

Three commercially available assays (an enzyme-linked immunosorbent assay ELISA, an immunoradiometric assay, IRMA, and a nephelometric assay) for the determination of lipoprotein(a) [Lp(a)] were compared with respect to the dependency of these assays on the various apolipoprotein(a) [apo(a)] isoforms. Although there was a strong correlation between the three methods, a significant difference between the absolute values (mg/L) was observed (p < 0.001). Using purified Lp(a) preparations, we showed that the ELISA assay quantifies the Lp(a) concentration on a molar basis, independently of the apo(a) isoform size. The IRMA and the nephelometric assay however are apo(a) isoform size dependent and overestimate the Lp(a) concentration of large apo(a) isoforms whereas the amount of small apo(a) isoforms is underestimated. In general, the isoform dependency of the Lp(a) quantification is of limited clinical relevance. In this study, inconsistent risk assignments are made in approximately 3% of the cases, when the Lp(a) concentrations obtained with the apo(a) isoform dependent assays are compared with the isoform independent ELISA.

A kringle-specific monoclonal antibody.

Kringle domains are found in several plasma proteins of blood coagulation and fibrinolysis. A murine monoclonal antibody, designated alpha HII-5, was produced against a synthetic peptide representing residues 216-231 of human prothrombin kringle 2. The sequence of the hexadecapeptide (Glu-Asn-Phe-Cys-Arg-Asn-Pro-Asp-Gly-Asp-Glu-Glu-Gly-Val-Gly-Cys) is conserved in several kringle-containing proteins, represents a predicted region of high local hydrophilicity in prothrombin kringle 2, and contains the anionic (Asp-223 and Asp-225) residues that contribute to lysine binding by plasminogen kringle 4. In a solution-phase immunoassay, antibody alpha HII-5 bound prothrombin and the kringle 5 light chain fragment of plasminogen (miniplasminogen), but not plasminogen or plasmin. In contrast, using a solid-phase assay with antigen immobilized onto a surface (polystyrene microtiter plates, glass, or nitrocellulose) antibody alpha HII-5 specifically bound prothrombin, plasminogen, recombinant tissue plasminogen activator (tPA), and the apo(a) subunit of lipoprotein(a). By immunoblotting analysis antibody alpha HII-5 bound determinants on prothrombin fragment 2 and plasminogen kringle 5. These observations suggest that a subset of kringle domains on plasma proteins, including prothrombin kringle 2 and plasminogen kringle 5, contains a homologous antigenic determinant in the region of the kringle lysine-binding site. In contrast to prothrombin kringle 2, the homologous peptide site on plasminogen is not available for antibody binding except when plasminogen is adsorbed to a nonphysiological surface, or when kringles 1-4 are removed.