Portal de Periódicos da CAPES

Elements in the C terminus of apolipoprotein [a] responsible for the binding to the tenth type III module of human fibronectin

2005; Elsevier BV; Volume: 46; Issue: 12 Linguagem: Inglês

10.1194/jlr.m500239-jlr200

1539-7262

Celina Edelstein, Mohammed Yousef, Angelo M. Scanu,

Blood Coagulation and Thrombosis Mechanisms

In previous studies, we showed that the C-terminal domain, F2, but not the N-terminal domain, F1, is responsible for the binding of apolipoprotein [a] (apo[a]) to human fibronectin (Fn). To pursue those observations, we prepared, by both elastase digestion and recombinant technology, subsets of F2 of a different length containing either kringle (K) V or the protease domain (PD). We also studied rhesus monkey apo[a], which is known to contain PD but not KV. In the case of Fn, we used both an intact product and its tenth type III module (10FN-III) expressed in Escherichia coli. The binding studies carried out on microtiter plates showed that the affinity of F2 for immobilized 10FN-III was ∼6-fold higher than that for Fn (dissociation constants = 1.75 ± 0.31 nM and 10.25 ± 1.62 nM, respectively). The binding was also exhibited by rhesus apo[a] and by an F2 subset containing the PD linked to an upstream microdomain comprising KIV-8 to KIV-10 and KV, inactive by itself. Competition experiments on microtiter plates showed that both Fn and 10FN-III, when in solution, are incompetent to bind F2.Together, our results indicate that F2 binds to immobilized 10FN-III more efficiently than whole Fn and that the binding can be sustained by truncated forms of F2 that contain the catalytically inactive PD linked to an upstream four K microdomain. In previous studies, we showed that the C-terminal domain, F2, but not the N-terminal domain, F1, is responsible for the binding of apolipoprotein [a] (apo[a]) to human fibronectin (Fn). To pursue those observations, we prepared, by both elastase digestion and recombinant technology, subsets of F2 of a different length containing either kringle (K) V or the protease domain (PD). We also studied rhesus monkey apo[a], which is known to contain PD but not KV. In the case of Fn, we used both an intact product and its tenth type III module (10FN-III) expressed in Escherichia coli. The binding studies carried out on microtiter plates showed that the affinity of F2 for immobilized 10FN-III was ∼6-fold higher than that for Fn (dissociation constants = 1.75 ± 0.31 nM and 10.25 ± 1.62 nM, respectively). The binding was also exhibited by rhesus apo[a] and by an F2 subset containing the PD linked to an upstream microdomain comprising KIV-8 to KIV-10 and KV, inactive by itself. Competition experiments on microtiter plates showed that both Fn and 10FN-III, when in solution, are incompetent to bind F2. Together, our results indicate that F2 binds to immobilized 10FN-III more efficiently than whole Fn and that the binding can be sustained by truncated forms of F2 that contain the catalytically inactive PD linked to an upstream four K microdomain. Fibronectin (Fn) is a large (450 kDa) multidomain protein that interacts with a variety of macromolecules, including components of the cytoskeleton and the extracellular matrix and cell surface receptors particularly of fibroblasts, neurons, macrophages, and bacteria (1Schwarzbauer J.E. Alternative splicing of fibronectin: three variants, three functions.Bioassays. 1991; 13: 527-533Google Scholar). Fn occurs in two main forms: one is insoluble, exhibiting adhesive properties, and is synthesized by fibroblasts, chondrocytes, endothelial cells, epithelial cells, and macrophages; the other is soluble, a heterodimer present in the plasma, and is synthesized by hepatocytes. In both forms there are three internally homologous repeats, called modules (types I, II, and III), that are readily separable by the action of proteolytic enzymes and exhibiting various binding motifs (2Hynes R.O. The molecular biology of fibronectin.Annu. Rev. Cell Biol. 1985; 1: 67-85Google Scholar). In human soluble Fn, there are 12 type I, 2 type II, and 15 type III modules, each module representing independently folded units containing mostly β sheets and turns (Fig. 1). The first two modules contain four conserved cysteines comprising two disulfide bonds that are critical for module stability and function. In turn, type III modules are devoid of disulfide bonds because the two unpaired cysteines are buried (3Baron M. Norman D.G. Campbell I.D. Protein modules.Trends Biochem. Sci. 1991; 16: 13-17Google Scholar, 4Potts J.R. Campbell I.D. Fibronectin structure and assembly.Curr. Opin. Cell Biol. 1994; 6: 648-655Google Scholar). Although Fn is present in early atherosclerotic lesions and in atherosclerotic plaques, its overall role in the atherosclerotic process is unclear (5Wight T.N. The Vascular Extracellular Matrix.in: Fuster V. Topol E.J. Nabel E.G. Atherothrombosis and Coronary Heart Disease. Lippincott Williams & Williams, Philadelphia, PA2005: 421-437Google Scholar). Salonen et al. (6Salonen E-M. Jauhiainen M. Zardi L. Vaheri A. Ehnholm C. Lipoprotein(a) binds to fibronectin and has serine proteinase activity capable of cleaving it.EMBO J. 1989; 8: 4035-4040Google Scholar) were the first to show that lipoprotein [a] (Lp[a]) binds, via apolipoprotein [a] (apo[a]), to immobilized Fn using preparations isolated from human plasma. Moreover, by studying thermolysin digests of Fn, they observed that the strongest binding involved the C-terminal heparin binding 29 kDa fragment. Binding of apo[a] to immobilized Fn was also reported by van der Hoek et al. (7van der Hoek Y.Y. Sangrar W. Cote G.P. Kastelein J.P. Koschinsky M.L. Binding of recombinant apolipoprotein(a) to extracellular matrix proteins.Arterioscler. Thromb. 1994; 14: 1792-1798Google Scholar), who used Fn isolated from human plasma and a recombinant apo[a]. The binding on Fn was located to a 12 amino acid sequence in the N-terminal region of the overlapping 29–38 kDa thermolysin fragments. These two previous studies did not rule out the possibility that in its immobilized form, Fn has additional binding sites for apo[a]. Moreover, they did not provide information on the site(s) on apo[a] involved in the binding. In this context, we previously showed that the C-terminal domain of apo[a], referred to as F2, is critical for the binding of apo[a] to Fn (8Scanu A.M. Edelstein C. Klezovitch O. Dominant role of the C-terminal domain of apolipoprotein(a) to the protein core of proteoglycans and other members of the extracellular matrix.Trends Cardiovasc. Med. 2000; 9: 196-200Google Scholar). This domain comprises kringle (K) IV types 5–10, one copy of KV, and the protease domain (PD) (9Edelstein C. Italia J.A. Klezovitch O. Scanu A.M. Functional and metabolic differences between elastase-generated fragments of human lipoprotein[a] and apolipoprotein[a].J. Lipid Res. 1996; 37: 1786-1801Google Scholar, 10Scanu A.M. Edelstein C. Kringle-dependent structural and functional polymorphism of apolipoprotein (a).Biochim. Biophys. Acta. 1995; 1256: 1-12Google Scholar) (Fig. 2). This PD has a high degree of homology with that of plasminogen; however, unlike the latter, it is inactive as a result of the presence of serine instead of arginine at the site of activation by tissue plasminogen activator (11McLean J.W. Tomlinson J.E. Kuang W. Eaton D.L. Chen E.Y. Fless G.M. Scanu A.M. Lawn R.M. cDNA sequence of human apolipoprotein(a) is homologous to plasminogen.Nature. 1987; 330: 132-137Google Scholar) and possibly because of the influence of the deletion of a nonapeptide (present in plasminogen) between residues 4,483 and 4,491 of the apo[a] sequence (12Gabel B.R. Koschinsky M.L. Analysis of the proteolytic activity of a recombinant form of apolipoprotein(a).Biochemistry. 1995; 34: 15777-15784Google Scholar). The significance of this inactive protease in apo[a] has not been determined, and questions have been raised regarding possible enzyme-unrelated activities. The current studies were carried out to define the elements in F2 responsible for the binding of apo[a] to Fn. For this purpose, we used a series of truncated products obtained by either proteolytic digestion of naturally occurring human apo[a] or recombinant technology, an approach that we have found useful in investigating the complex structure-function relationships in Lp[a] (13Scanu A.M. Edelstein C. Learning about the structure and biology of human lipoprotein [a] through dissection by enzymes of the elastase family: facts and speculations.J. Lipid Res. 1997; 38: 2193-2206Google Scholar). In our binding systems, besides intact Fn, we also used the tenth type III module (10FN-III) that in preliminary studies in human carotid artery plaques was found to be associated with apo[a] (A. M. Scanu et al., unpublished observation). Here, we provide evidence that this module, known to participate in a number of cell adhesion events, is involved in the binding of Fn to human apo[a] and that in the latter case, the PD is a necessary but not sufficient element in the binding. Fn was purchased from Enzyme Research Laboratories (South Bend, IN). Human leukocyte elastase (EC 3.4.21.37), BSA, Tween 20, 2-mercaptoethanol, SDS, diiopropylfluorophosphate, ε-aminocaproic acid (EACA), goat anti-rabbit and anti-mouse IgG alkaline phosphatase conjugates, rabbit affinity-purified anti-human Fn, and p-nitrophenyl phosphate were all purchased from Sigma (St. Louis, MO). Kallikrein inactivator was purchased from Calbiochem (San Diego, CA). Rabbit affinity-purified antibodies to apo[a] were prepared as described previously (14Fless G.M. Snyder M.L. Scanu A.M. Enzyme-linked immunoassay for Lp[a].J. Lipid Res. 1989; 30: 651-662Google Scholar). Anti-apo[a] did not react against either LDL or plasminogen. A monoclonal antibody against a recombinant human apo[a] KV was a gift from Abbott Laboratories (Abbott Park, IL), and its specificity was verified in our laboratory. All chemicals were of reagent grade. The subjects used for the preparation of Lp[a] were healthy donors with plasma Lp[a] protein levels in the range of 15–43 mg/dl with a known apo[a] phenotype. Their plasma was obtained by plasmapheresis performed in the blood bank at the University of Chicago. All of the subjects in the study gave written informed consent according to protocols approved by the Institutional Review Board of the University of Chicago. The steps for Lp[a] isolation were carried out immediately after blood drawing using a combination of ultracentrifugation and lysine-Sepharose affinity chromatography as described previously (15Fless G.M. Snyder M.L. Furbee J.W.J. Garcia-Hedo M.T. Mora R. Subunit composition of lipoprotein(a) protein.Biochemistry. 1994; 33: 13492-13501Google Scholar). For the current study, we used Lp[a] having 15 type IV Ks. In those subjects having two apo[a] isoforms, the Lp[a] species containing the desired isoform was separated from the other by CsCl density gradient ultracentrifugation as described (15Fless G.M. Snyder M.L. Furbee J.W.J. Garcia-Hedo M.T. Mora R. Subunit composition of lipoprotein(a) protein.Biochemistry. 1994; 33: 13492-13501Google Scholar). Apo[a] was isolated from Lp[a] essentially as described by Edelstein et al. (16Edelstein C. Mandala M. Pfaffinger D. Scanu A.M. Determinants of lipoprotein(a) assembly: a study of wild-type and mutant apolipoprotein(a) phenotypes isolated from human and rhesus monkey lipoprotein(a) under mild reductive conditions.Biochemistry. 1995; 34: 16483-16492Google Scholar) in the presence of 1.25 mM dithioerythreitol. Apo[a] phenotyping was performed on plasma samples by SDS-PAGE followed by immunoblotting using a monospecific anti-human apo[a] antibody. The mobility of the individual apo[a] bands was compared with that of isolated apo[a] isoforms of known molecular weights, and the number of Ks was compared with those provided by Dr. Angles-Cano (Institut National de la Santé et de la Recherche Médicale Unit 143, Paris, France). Lp[a] was isolated from rhesus plasma as described previously (17Scanu A.M. Miles L.A. Fless G.M. Pfaffinger D. Eisenbart J. Jackson E. Hooverplow J.L. Brunck T. Plow E.F. Rhesus monkey lipoprotein(a) binds to lysine Sepharose and U937 monocytoid cells less efficiently than human lipoprotein(a). Evidence for the dominant role of kringle 4(37).J. Clin. Invest. 1993; 91: 283-291Google Scholar). Apo[a] was separated from Lp[a] by the same techniques used for the human Lp[a] and had an apparent mass of 325 kDa. We have shown previously that both products are recognized by the human antibodies (17Scanu A.M. Miles L.A. Fless G.M. Pfaffinger D. Eisenbart J. Jackson E. Hooverplow J.L. Brunck T. Plow E.F. Rhesus monkey lipoprotein(a) binds to lysine Sepharose and U937 monocytoid cells less efficiently than human lipoprotein(a). Evidence for the dominant role of kringle 4(37).J. Clin. Invest. 1993; 91: 283-291Google Scholar). A schematic representation of the apo[a] fragments used in our studies is shown in Fig. 2. Human leukocyte elastase (1 unit = 1 nm p-nitrophenol/s from N-1 T-butyloxycarbonyl-L-Ala p-nitrophenol ester) was diluted 1,000-fold in 50 mM Tris-HCl, 0.1 M NaCl, pH 8.0. One microliter of the diluted enzyme was incubated per 15 μg of apo[a] at 37°C for 30 min. The reaction was terminated with 5 mM diiopropylfluorophosphate for 20 min at 22°C. The purification of the fragments was carried out as described previously, and each fragment gave a single band on SDS-PAGE (18Klezovitch O. Edelstein C. Scanu A.M. Stimulation of interleukin-8 production in human THP-1 macrophages by apolipoprotein(a): evidence for a critical involvement of elements of its C-terminal domain.J. Biol. Chem. 2001; 276: 46864-46869Google Scholar). The fragments behaved as monomers when examined by sedimentation velocity in the analytical ultracentrifuge. The rIII recombinant protein lacking KIV-6, -7, -8, and PD (6KΔKIV6-8ΔPD) was prepared by digesting the 6 K expression plasmid with ClaI and EcoRV (19Edelstein C. Pfaffinger D. Hinman J. Miller E. Lipkind G. Tsimikas S. Bergmark C. Getz G.S. Witztum J.L. Scanu A.M. Lysine-phosphatidylcholine adducts in kringle V impart unique immunological and potential pro-inflammatory properties to human apolipoprotein(a).J. Biol. Chem. 2003; 26: 52841-52847Google Scholar). Subsequently, the 2.4 kb ClaI-EcoRV insert was subjected to digestion with BamHI to remove the BamHI-BamHI internal DNA domains coding for apo[a] KIV-6 through -8. The 0.32 kb ClaI-BamHI and the 1.03 kb BamHI-EcoRV DNA fragments were then ligated back into the 6 K expression plasmid that was digested with ClaI and EcoRV (vector part). rIII (6KΔKIV6-8ΔPD) (Fig. 3)was sequenced and transfected into human embryonic kidney 293 cells. The clone produced significant amounts of recombinant product (0.5–10 mg/l of the culture medium) that was purified by lysine-Sepharose chromatography. The recombinant product, 10FN-III, was the tenth unit of the human Fn type III domain containing 96 amino acids (Fig. 1). The steps involved in the gene construction, phage display, expression in Escherichia coli, and purification of 10FN-III from cell culture medium were described previously (20Koide A. Bailey C.W. Huang X. Koide S. The fibronectin type III domain as a scaffold for novel binding proteins.J. Mol. Biol. 1998; 284: 1141-1151Google Scholar). By SDS-PAGE, it migrated as a single band in the expected position (Fig. 4). Sedimentation velocity experiments were conducted using the Beckman XL-A analytical ultracentrifuge. Approximately 0.4 mg/ml 10FN-III in TBS was placed in a two sector centerpiece and centrifuged at 60,000 rpm for 15 h at 22°C. The data were analyzed using SEDFIT (21Schuck P. Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling.Biophys. J. 2000; 78: 1606-1619Google Scholar). A single symmetrical peak was observed with an apparent molecular weight of 10,400. All binding experiments were conducted at room temperature. Microtiter plates (Beckman Instruments, Fullerton, CA) were coated with 100 μl of either Fn or 10FN-III, each at 10 μg/ml, in TBS buffer (50 mM Tris-HCl, pH 7.5, 0.15 M NaCl) for 24 h. Nonspecific binding sites were blocked with 1% BSA in TBS for 1.5 h. After three washes with TBST (TBS supplemented with 0.1% BSA and 0.02% Tween 20), various concentrations of either apo[a] or apo[a] fragments were added to the wells in TBS buffer and incubated for 2 h. After incubation, the wells were washed three times with TBST. The bound protein was detected using a monospecific polyclonal anti-apo[a] antibody in TBST for 1 h. In the specific case of fragment F7 (KV-PD), a monoclonal anti-KV antibody was used because this fragment was not recognized by our anti-apo[a] antibody. At this time, the wells were washed three times with TBST, and goat anti-rabbit IgG, or in the case of KV-PD, anti-mouse IgG, both conjugated to alkaline phosphatase, were added for 1 h. After washing with TBST, p-nitrophenyl phosphate (1 mg/ml in diethanolamine buffer, pH 9.8) was added, and the color development was followed at 405 nm on a Versamax microplate reader (Molecular Devices, Sunnyvale, CA). In another set of experiments, we coated the wells with apo[a] and after washing the excess we examined the binding of either Fn or 10FN-III using a polyclonal antibody specific for human Fn. Analysis of the binding data was performed on the assumption of single-site binding. Dissociation constants (Kd) were derived using the program Origin version 7.03 (Origin Lab Co., Northhampton, MA) fitting the data to a one-site model represented by the equation [Y]=B[X](K+[X])where [Y] represents the absorbance at 405 nm, which is proportional to the amount of ligand bound; [X] represents the concentration of free ligand; B represents the maximum absorbance at saturation; and K represents the association constant. In some experiments, a constant amount (50 nM) of either apo[a] or fragments was incubated with immobilized Fn in the presence of various concentrations of EACA (0–200 mM) or NaCl (0–2 M). After incubation for 1 h, the bound protein was detected with anti-apo[a] antibody or anti-KV antibody, as described above, depending on the protein under study. A range of concentrations of either 10FN-III or Fn were mixed with 50 nM apo[a] or in some cases a range of apo[a] concentrations and incubated in wells of microtiter plates coated with Fn or 10FN-III. Bound apo[a] was detected using a monospecific anti-apo[a] antibody. The secondary antibody and color development reaction were as described above. Lp[a] protein was quantified by a sandwich ELISA as described previously (14Fless G.M. Snyder M.L. Scanu A.M. Enzyme-linked immunoassay for Lp[a].J. Lipid Res. 1989; 30: 651-662Google Scholar). The concentration of apo[a] was determined either by ELISA or using an extinction coefficient (e278 = 1.31 ml/mg/cm) established previously (22Edelstein C. Italia J.I. Scanu A.M. Polymorphonuclear cells isolated from human peripheral blood cleave lipoprotein(a) and apolipoprotein(a) at multiple interkringle sites via the enzyme elastase: generation of mini Lp(a) particles and apo(a) fragments.J. Biol. Chem. 1997; 272: 11079-11087Google Scholar) for apo[a]. Protein concentrations of the fragments were determined by the Bio-Rad DC protein assay. As shown in Fig. 5A, apo[a] bound in a saturable manner. The apparent Kd value was 10.25 ± 1.62 nM (Table 1). To determine which portion of apo[a] was responsible for this binding, we examined F1 and F2, the N- and C-terminal fragments, respectively, generated by limited digestion of apo[a] with leukocyte elastase (Fig. 2). F2 bound in a saturable manner, whereas F1 exhibited no detectable binding (Fig. 5, Table 1). Concentrations of EACA up to 200 mM or NaCl up to 2 M had no effect on the binding of either apo[a] or F2 (data not shown), suggesting a hydrophobic mode of interaction. These results indicated that F2 has the necessary elements for binding to Fn and that this binding was stronger when F2 was an integral part of apo[a], suggesting an enhancing effect by the attached F1.TABLE 1Summary of the binding parameter Kd of apo[a] and fragments to immobilized Fn and 10FN-IIISampleFn10FN-IIInMApo[a], 15 kringle IV10.25 ± 1.62 (6)1.75 ± 0.31 (3)F228.75 ± 6.10 (4)5.64 ± 0.46 (3)F432.16 ± 1.91 (3)15.16 ± 1.57 (4)Rhesus apo[a]7.72 ± 0.26 (3)3.42 ± 0.62 (3)apo[a], apolipoprotein [a]; Fn, fibronectin; 10FN-III, tenth type III module of fibronectin; Kd, dissociation constant. Kd was calculated by fitting the data from replicate experiments globally to a one-site model as described in Methods. The data shown are means ± SEM. The numbers of independent experiments conducted in triplicate are shown in parentheses. Open table in a new tab apo[a], apolipoprotein [a]; Fn, fibronectin; 10FN-III, tenth type III module of fibronectin; Kd, dissociation constant. Kd was calculated by fitting the data from replicate experiments globally to a one-site model as described in Methods. The data shown are means ± SEM. The numbers of independent experiments conducted in triplicate are shown in parentheses. To define the region within F2 responsible for the binding to Fn, we studied the elastase digests of apo[a] shown in Fig. 2. Only F4 (KIV-8, -9, -10, KV, and PD) bound, whereas F3 (KIV-5, -6, -7, -8, -9, -10), F5 (KIV-8, -9, -10), and F6 (KIV-5, -6, -7) were inactive (Fig. 5B). These results suggested that either KV or PD or both were participants in the binding. To resolve this issue, we used the recombinant protein rIII, containing KIV-9, -10, and KV but not the PD. We found that rIII exhibited no binding (Fig. 5B), suggesting that the PD is important for the binding of apo[a] to Fn. To corroborate this conclusion, we used apo[a] isolated from rhesus Lp[a]. This nonhuman primate apo[a] is structurally similar to human apo[a] but lacks KV. As shown in Fig. 5C, the binding of this apo[a] to Fn was comparable to that of its human counterpart, with apparent Kd values of 7.72 ± 0.26 nM and 10.25 ± 1.62 nM, respectively (Table 1). We next tested the small fragment, F7 (KV-PD), for its ability to bind to Fn. For detection, we used, as a primary antibody, a monoclonal antibody directed against KV after establishing that it was suited to recognize both KV-PD and apo[a]. We observed binding with apo[a] but not with KV-PD (Fig. 5D). Together, all of the information gathered provided evidence that the PD required upstream sequences longer than KV for Fn binding, In these studies, we coated the wells of the microtiter plates with 10FN-III. As shown in Fig. 6, human apo[a], rhesus apo[a], F2, and F4 exhibited binding, whereas F1 did not. In all cases, the binding affinity to 10FN-III was markedly higher than that exhibited by Fn. This was particularly true for apo[a] and F2, which exhibited a 5- to 6-fold difference from the data obtained with Fn (Table 1). To assess the ability of apo[a] to bind Fn in solution, we coated the microtiter plates with Fn at 10 μg/ml and then incubated the wells with solutions containing variable amounts (0–80 μg/ml) of Fn and a constant amount (50 nM) of apo[a]. As shown in Fig. 7, when in the solution phase, Fn was unable to compete for the binding of apo[a]. Similar results were observed for 10FN-III. In addition, in a system in which apo[a] was coated on microtiter plates at 10 μg/ml or 45 nM, binding of either Fn or 10FN-III was not observed (data not shown). The current studies have identified five elements in the C terminus of F2 that together were required for the binding of human apo[a] to immobilized Fn. Of these, the PD proved to be necessary but not sufficient for binding, given the need for the three upstream Ks, KIV-8 to -10, which by themselves were functionally inert. This conclusion is based on the combined results summarized in Fig. 5 obtained by studying proteolytic digests of apo[a] and recombinant products. KV was not required based on the results with the truncated products and those for rhesus monkey apo[a], which is known to contain PD but not KV (23Tomlinson J. McLean J. Lawn R. Rhesus monkey apolipoprotein(a).J. Biol. Chem. 1989; 264: 5957-5965Google Scholar), that proved to be as effective as a human counterpart in binding Fn. Our conclusion is also in keeping with the lack of binding exhibited by the small KV-PD unit. With reference to the type of binding, the lack of effect by EACA suggests a lysine-independent mode despite the fact that the PD, a 236 amino acid residue structure, contains 14 of the 23 lysines present in the whole apo[a]. Of note, the PD, although containing the histidine/asparagine/serine triad, a requisite for serine protease activity, is catalytically inactive because the arginine required for cleavage by tissue plasminogen activator is replaced by serine (11McLean J.W. Tomlinson J.E. Kuang W. Eaton D.L. Chen E.Y. Fless G.M. Scanu A.M. Lawn R.M. cDNA sequence of human apolipoprotein(a) is homologous to plasminogen.Nature. 1987; 330: 132-137Google Scholar) and possibly also because of conformational issues relating to the nine amino acid deletion in the PD (12Gabel B.R. Koschinsky M.L. Analysis of the proteolytic activity of a recombinant form of apolipoprotein(a).Biochemistry. 1995; 34: 15777-15784Google Scholar). It is unclear whether this catalytic inactivity has meaning regarding Fn binding. In this regard, plasminogen, which is highly homologous to apo[a] and has a catalytically active PD (11McLean J.W. Tomlinson J.E. Kuang W. Eaton D.L. Chen E.Y. Fless G.M. Scanu A.M. Lawn R.M. cDNA sequence of human apolipoprotein(a) is homologous to plasminogen.Nature. 1987; 330: 132-137Google Scholar), has been shown to bind to Fn via a lysine-dependent mechanism (6Salonen E-M. Jauhiainen M. Zardi L. Vaheri A. Ehnholm C. Lipoprotein(a) binds to fibronectin and has serine proteinase activity capable of cleaving it.EMBO J. 1989; 8: 4035-4040Google Scholar, 7van der Hoek Y.Y. Sangrar W. Cote G.P. Kastelein J.P. Koschinsky M.L. Binding of recombinant apolipoprotein(a) to extracellular matrix proteins.Arterioscler. Thromb. 1994; 14: 1792-1798Google Scholar), indicating a mode of Fn binding divergent from that of apo[a]. The other novel finding of our work is the identification of the tenth type III module, referred to as 10FN-III, as a major site on Fn involved in apo[a] binding. This is an interesting model from a structural standpoint in that it represents a relatively small monomeric unit, one of the few members of the immunoglobulin superfamily with no disulfide bonds (3Baron M. Norman D.G. Campbell I.D. Protein modules.Trends Biochem. Sci. 1991; 16: 13-17Google Scholar, 4Potts J.R. Campbell I.D. Fibronectin structure and assembly.Curr. Opin. Cell Biol. 1994; 6: 648-655Google Scholar). Moreover, the three-dimensional structure of 10FN-III has been determined by NMR (24Main A.L. Harvey T.S. Baron M. Boyd J. Campbell I.D. The three dimensional structure of the tenth type III module of fibronectin: an insight into RGD-mediated interactions.Cell. 1992; 71: 671-678Google Scholar) and X-ray crystallography (25Dickinson C.D. Veerapandian B. Dai X-P. Hamlin R.C. Xuong N-H. Ruoslahti E. Ely K.R. Crystal structure of the tenth type III cell adhesion module of human fibronectin.J. Mol. Biol. 1994; 236: 1079-1092Google Scholar), and its stability and folding (26Plaxco K.W. Spitzfaden C. Campbell I.D. Dobson C.M. Rapid refolding of proline-rich beta-sheet fibronectin type III module.Proc. Natl. Acad. Sci. USA. 1996; 93: 10703-10706Google Scholar, 27Plaxco K.W. Spitfaden C. Campbell I.D. Dobson C.M. A comparison of the folding kinetics and thermodynamics of two homologous fibronectin type III modules.J. Mol. Biol. 1997; 270: 763-770Google Scholar) and conformational dynamics (28Carr P.A. Erickson H.P. Palmer A.G.R. Backbone dynamics of homologous fibronectin type III cell adhesion domains from fibronectin and tenascin.Structure. 1997; 5: 949-959Google Scholar) are also well established. The structure is best described as a β-sandwich with seven β-strands containing the integrin binding RGD sequence involved in cell adhesion (24Main A.L. Harvey T.S. Baron M. Boyd J. Campbell I.D. The three dimensional structure of the tenth type III module of fibronectin: an insight into RGD-mediated interactions.Cell. 1992; 71: 671-678Google Scholar). Among the already established properties of 10FN-III, we now show that it also binds to critical elements of the C-terminal domain of apo[a]. It remains to be established which site(s) on 10FN-III are responsible for apo[a] binding. In previous studies, van der Hoek et al. (7van der Hoek Y.Y. Sangrar W. Cote G.P. Kastelein J.P. Koschinsky M.L. Binding of recombinant apolipoprotein(a) to extracellular matrix proteins.Arterioscler. Thromb. 1994; 14: 1792-1798Google Scholar) identified a 12 amino acid sequence in Fn involved in the binding to a recombinant form of apo[a]. As for 10FN-III, the binding was not lysine-mediated and was unaffected by high salt concentrations. According to the published data on Fn, the sequence reported by van der Hoek et al. (7van der Hoek Y.Y. Sangrar W. Cote G.P. Kastelein J.P. Koschinsky M.L. Binding of recombinant apolipoprotein(a) to extracellular matrix proteins.Arterioscler. Thromb. 1994; 14: 1792-1798Google Scholar) is located downstream of 10FN-III in the junction between the 11th and 12th modules (29Kornblihtt A.R. Umezawa K. Vibe-Pedersen K. Baralle F.E. Primary structure of human fibronectin: differential splicing may generate at least 10 polypeptides from a single gene.EMBO J. 1985; 4: 1755-1759Google Scholar). This site was identified by submitting soluble Fn to thermolysin digestion followed by a trypsin step resulting in the unmasking of a site buried in undigested Fn. It is difficult to compare those data with ours because the authors neither provided quantitative binding parameters nor identified the elements in the recombinant apo[a] used. Overall, we believe that our current results make a strong case for 10FN-III being a major site for apo[a] binding. This conclusion is corroborated by the observation that the binding affinity of this module was 6-fold higher than that exhibited by the whole Fn, suggesting that in the intact molecule the binding site on 10FN-III is partially buried. Past studies from this laboratory have underscored the value of proteolytic dissection in the investigation of the structure and biology of Lp[a] (13Scanu A.M. Edelstein C. Learning about the structure and biology of human lipoprotein [a] through dissection by enzymes of the elastase family: facts and speculations.J. Lipid Res. 1997; 38: 2193-2206Google Scholar, 22Edelstein C. Italia J.I. Scanu A.M. Polymorphonuclear cells isolated from human peripheral blood cleave lipoprotein(a) and apolipoprotein(a) at multiple interkringle sites via the enzyme elastase: generation of mini Lp(a) particles and apo(a) fragments.J. Biol. Chem. 1997; 272: 11079-11087Google Scholar). Since then, the use of proteolytically derived fragments alone or in combination with recombinant products has provided and is continuing to provide evidence for the structural heterogeneity of apo[a] associated with a functional diversity in which Ks play an important role. For instance, KIV-9, which is known to be engaged in disulfide linkage with the C-terminal domain of apolipoprotein B-100 (11McLean J.W. Tomlinson J.E. Kuang W. Eaton D.L. Chen E.Y. Fless G.M. Scanu A.M. Lawn R.M. cDNA sequence of human apolipoprotein(a) is homologous to plasminogen.Nature. 1987; 330: 132-137Google Scholar), was implicated recently in the stimulating action of apo[a] on the migration and proliferation of vascular smooth muscle cells (30O'Neil C.H. Boffa M.B. Hancock M.A. Pickering J.G. Koschinsky M.L. Stimulation of vascular smooth cell proliferation and migration by apolipoprotein(a) is dependent on inhibition of transforming growth factor-beta activation and on the presence of kringle IV type 9.J. Biol. Chem. 2004; 279: 55187-55195Google Scholar). KIV-10 contains the high-affinity lysine binding site important for critical apo[a] functions (31Scanu A.M. Fless G.M. Lipoprotein (a): heterogeneity and biological relevance.J. Clin. Invest. 1990; 85: 1709-1715Google Scholar), whereas KV-5 to -8 contain some lysine residues that are linked covalently to oxidized phospholipids that appear to impart a proinflammatory function to apo[a] (19Edelstein C. Pfaffinger D. Hinman J. Miller E. Lipkind G. Tsimikas S. Bergmark C. Getz G.S. Witztum J.L. Scanu A.M. Lysine-phosphatidylcholine adducts in kringle V impart unique immunological and potential pro-inflammatory properties to human apolipoprotein(a).J. Biol. Chem. 2003; 26: 52841-52847Google Scholar). Moreover, a recombinant form of KV has been reported to exhibit an antiangiogenic function (32Kim J-S. Yu H-K. Ahn J-H. Lee H-J. Hong S-W. Jung H-H. Chang S-I. Hong Y-K. Joe Y-A. Byun S-M. et al.Human apolipoprotein(a) kringle V inhibits angiogenesis in vitro and in vivo by interfering with the activation of focal adhesion kinases.Biochem. Biophys. Res. Commun. 2004; 313: 534-540Google Scholar). Very recently in mouse models, the KIV-5 to -8 peptide was implicated in the delayed chylomicron remnant removal from the plasma and also immunochemically identified in the atherosclerotic area of the aortic root examined (33Devlin C.M. Lee S-J. Kuriakose G. Spencer C. Becker L. Grosskopf I. Ko C. Huang L-S. Koschinsky M. Cooper A.D. et al.An apolipoprotein(a) peptide delays chylomicron remnant clearance and increases plasma remnant lipoproteins and atherosclerosis in vivo.Arterioscler. Thromb. Vasc. Biol. 2005; 25: 1704-1710Google Scholar). The PD has received relatively little attention, being functionally inert from an enzymatic standpoint, although recently it was shown to be one of the elements involved in the binding of human apo[a] to fibrinogen (34Xue S. Madison E.L. Miles L.A. The Kringle V-protease domain is a fibrinogen binding region within human apo(a).Thromb. Haemost. 2001; 86: 1229-1237Google Scholar). Our current studies now provide evidence that in apo[a], the PD plays a role in the binding of human apo[a] to Fn, but only in cooperation with at least four Ks located upstream. The current studies also bring attention to the fact that both Fn and 10FN-III bind to apo[a] when immobilized but not in solution. This observation, which in the case of Fn is in keeping with previous studies by Salonen et al. (6Salonen E-M. Jauhiainen M. Zardi L. Vaheri A. Ehnholm C. Lipoprotein(a) binds to fibronectin and has serine proteinase activity capable of cleaving it.EMBO J. 1989; 8: 4035-4040Google Scholar), is not unique to the Fn-apo[a] system, because immobilization of Fn has been shown to be required for the binding of Fn to plasminogen, tissue-type plasminogen activator (35Salonen E-M. Saksela O. Vartio T. Vaheri A. Nielsen L.S. Zeuthen J. Plasminogen and tissue-type plasminogen activator binds to immobilized fibronectin.J. Biol. Chem. 1985; 260: 12302-12307Google Scholar), and acute-phase C-reactive protein (36Salonen E-M. Vartio T. Hedman K. Vaheri A. Binding of fibronectin to the acute phase-reactant C-reactive protein.J. Biol. Chem. 1984; 259: 1496-1514Google Scholar). In this regard, recent studies have shown that Fn undergoes a conformational transition from a closed form to an open form when moving from a solution to a solid phase (37Halter M. Antia M. Vogel V. Fibronectin conformational changes induced by adsorption to liposomes.J. Control. Release. 2005; 101: 209-222Google Scholar, 38Kowalczyñska H.M. Norwak-Wyrzykowska M. Kolos R. Dobkowski J. Kamiñski J. Fibronectin adsorption and arrangement on copolymer surfaces and their significance in cell adhesion.J. Biomed. Mater. Res. 2005; 72A: 228-236Google Scholar). Noteworthy, this notion explains why Fn and apo[a] do not interact with each other in the circulating plasma. Regarding the biological relevance of the current findings, we have previously shown that apo[a], via F2, binds to fibrinogen and to the protein core of the proteoglycans decorin and biglycan (8Scanu A.M. Edelstein C. Klezovitch O. Dominant role of the C-terminal domain of apolipoprotein(a) to the protein core of proteoglycans and other members of the extracellular matrix.Trends Cardiovasc. Med. 2000; 9: 196-200Google Scholar); in addition, we confirmed the critical role of F2 in this binding by subjecting it to limited proteolysis (either pancreatic elastase or metalloproteinase-12), and apo[a] bound to either decorin, biglycan, or Fn immobilized onto microtiter plates (39Scanu A.M. Chellamma S. Cleavage of human apolipoprotein(a) coated on the surface of constituents of the vascular extracellular matrix by pancreatic elastase and metalloproteinase-12. Different partitions between cleaved N-terminal and C-terminal domains.Vascular Disease Prevention. 2004; 1: 59-63Google Scholar). Moreover, we have observed that F2 undergoes further fragmentation under more extensive proteolytic conditions (22Edelstein C. Italia J.I. Scanu A.M. Polymorphonuclear cells isolated from human peripheral blood cleave lipoprotein(a) and apolipoprotein(a) at multiple interkringle sites via the enzyme elastase: generation of mini Lp(a) particles and apo(a) fragments.J. Biol. Chem. 1997; 272: 11079-11087Google Scholar). Together, these results, in keeping with the lipoprotein retention hypothesis (40Williams K.J. Tabas I. The response-to retention hypothesis of early atherogenesis.Arterioscler. Thromb. Vasc. Biol. 1995; 15: 551-561Google Scholar), suggest that the trapping by the vascular extracellular matrix elements favors the fragmentation of apo[a], the extent of which is dependent on the activity of the proteolytic enzymes, an expression of the chronic inflammatory milieu of the vessel wall. In this context, we recently provided evidence for an association between proteolytic activity and inflammation in plaques from endarterectomy segments of human carotid arteries and also showed the presence of fragments of apo[a], decorin, biglycan, versican (41Formato M. Farina M. Spirito R. Maggioni M. Guarino A. Cherchi G. Biglioli P. Edelstein C. Scanu A.M. Evidence for a proinflammatory and proteolytic environment in plaques from endarterectomy segments of human carotid arteries.Arterioscler. Thromb. Vasc. Biol. 2004; 24: 129-135Google Scholar, 42Fortunato J.E. Bassiouny H.S. Song R.H. Kocharian H. Glagov S. Edelstein C. Scanu A.M. Apolipoprotein(a) fragments in relation to human carotid plaque instability.J. Vasc. Surg. 2000; 32: 555-563Google Scholar), and, more recently, Fn (A.M. Scanu et al., unpublished observations). In the latter case, we have identified the 10FN-III module associated with apo[a] using both immunohistochemical and immunoprecipitation techniques. The notion emerging from these findings is that in the inflammatory microenvironment of the atheromatous plaque, fragments of blood-derived lipoproteins become linked to elements of the extracellular matrix, generating entities likely exhibiting unique functions. For instance, apo[a] binding may affect the 10FN-III-mediated interactions of Fn regarding the atherosclerotic process; in turn, immobilized apo[a] would become more amenable to the action of proteolytic enzymes, generating some potentially bioactive fragments. The authors thank Dr. Shohei Koide of the Department of Biochemistry and Molecular Biology at the University of Chicago for providing the recombinant 10FN-III module and for helpful discussions; the authors also thank Abbott Laboratories for the monoclonal antibody against human KV. This work was supported by grants from the National Institutes of Health (HL-63209 and HL-63115 to A.M.S.).