If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To whom correspondence should be addressed: Boehringer Mannheim GmbH, Molecular Biology, TF-MM, Sandhofer Straße 116, D-68305 Mannheim, Federal Republic of Germany. Tel.: 49-621-759-4330; Fax: 49-621-759-6168.
Affiliations
Department of Molecular Biology, Boehringer Mannheim GmbH, D-68305 Mannheim, Federal Republic of Germany
∗ This work was supported by Deutche Forschungsgemeinschaft Grant Ar161/1-3. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore by hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The well documented association between high plasma levels of lipoprotein(a) (Lp(a)) and cardiovascular disease might be mediated by the lysine binding of apolipoprotein(a) (apo(a)), the plasminogen-like, multikringle glycoprotein in Lp(a). We employed a mutational analysis to localize the lysine-binding domains within human apo(a). Recombinant apo(a) (r-apo(a)) with 17 plasminogen kringle IV-like domains, one plasminogen kringle V-like domain, and a protease domain or mutants thereof were expressed in the human hepatocarcinoma cell line HepG2. The lysine binding of plasma Lp(a) and r-apo(a) in the culture supernatants of transfected HepG2 cells was analyzed by lysine-Sepharose affinity chromatography. Wild type recombinant Lp(a) (r-Lp(a)) revealed lysine binding in the range observed for human plasma Lp(a). A single accessible lysine binding site in Lp(a) is indicated by a complete loss of lysine binding observed for r-Lp(a) species that contain either a truncated r-apo(a) lacking kringle IV-37, kringle V, and the protease or a point-mutated r-apo(a) with a Trp-4174 → Arg substitution in the putative lysine-binding pocket of kringle IV-37. Evidence is also presented for additional lysine-binding sites within kringles 32-36 of apo(a) that are masked in Lp(a) as indicated by an increased lysine binding for the point mutant (Cys-4057 → Ser), which is unable to assemble into particles. An important role of these lysine-binding site(s) for Lp(a) assembly is suggested by a decreased assembly efficiency for deletion mutants lacking either kringle 32 or kringles 32-35.
)is a cholesterol-rich particle composed of a low density lipoprotein (LDL) core and a single copy of the high molecular weight glycoprotein apolipoprotein(a) (apo(a))(
). There is increasing evidence, however, that Lp(a) may be involved in atherogenesis. Epidemiological studies have demonstrated a strong association between high Lp(a) plasma concentrations and cardiovascular disease(
) have reported conflicting conclusions on the association of elevated Lp(a) levels and the risk of coronary heart disease. In both the Lipid Research Clinics Coronary Primary Prevention Trial (
) elevated Lp(a) concentrations were an independent risk factor for coronary heart disease in hypercholesterolemic white men. The Physicians' Health Study(
), however, found no evidence for an association between Lp(a) levels and the risk for future myocardial infarction. It should be noted that in this latter study (
). The fibrinolytic proenzyme Pg consists of a preactivation peptide, five so-called “kringle” domains, numbered I-V, and a protease domain (reviewed in (
)). Several functions of Pg appear to be enhanced by its kringle structures. Lysine-binding sites (LBSs) within Pg kringles mediate the binding of Pg to fibrin(ogen)(
). The three-dimensional structure of Pg kringle IV and its complex with the lysine analog ε-amino-caproic acid (EACA) has been determined by high resolution x-ray crystallography (
). Apo(a) contains 10 distinct kringle repeats arranged in tandem all with high homology (61-75%) to kringle IV of Pg followed by single Pg-like kringle V and protease domains(
), more than 30 apo(a) isoforms have been found in human plasma differing in the number of kringle IV repeats. Based on their electrophoretic mobility apo(a) isoforms have been classified as F, B, S1, S2, S3, and S4(
Knowledge of the structural basis for the lysine- and fibrin(ogen)-binding of Lp(a) is paramount to our understanding of the role of this lipoprotein in fibrinolysis. A sequence comparison between human and rhesus monkey Lp(a), which differ in their lysine binding abilities led Scanu et al.(
) it has been predicted that not only kringle 37, containing the most highly conserved lysine-binding pocket relative to Pg kringle IV, but also kringles 32-35 comprise potential lysine-binding domains. Single apo(a) domains expressed in bacteria (
) revealed lysine binding activity in kringle 37 but not in kringles 1 and 2. Direct experimental evidence for the localization of LBSs in apo(a) and Lp(a) is still lacking.
We now describe a mutational analysis aimed at the identification of lysine-binding domains in recombinant apo(a) (r-apo(a)). Lysine-Sepharose affinity chromatography was employed to compare the lysine binding activity of wild type and mutant r-apo(a)/r-Lp(a) in the medium of transfected HepG2 hepatocarcinoma cells. In addition, we performed immunoblotting experiments with wild type and mutant r-apo(a) samples to assess a possible function of lysine-binding kringles in apo(a) for the assembly of apo(a)•apoB complexes.
EXPERIMENTAL PROCEDURES
Plasmids and Oligonucleotides
The construction of plasmids pCMV-A18 (Fig. 1a), pCMV-A18 Δ32-P (Fig. 1e), and pCMV-A18_4057Ser (Fig. 1g) has been previously described(
). The following oligonucleotides were synthesized with a model 392 DNA/RNA synthesizer (Applied Biosystems Inc., Foster City, CA): 1, 5′-GTGTTTTACCATGGACCCCAGCATTCGGAGGGAGTACTGCAACC-3′; 2, 5′-CTAGGGCCTCCTTCTGAACAATGAG-3′; 3, 5′-TCGACTCATTGTTCAGAAGGAGGCC-3′; 4, 5′-TCCTAGAGACTCCCACTGTTGTTCCAGTTCCAAGCATGGAGGCTCATTCTGAAGCATGAG-3′; 5, 5′-TCGACTCATGCTTCAGAATGAGCCTCCATGCTTGGAACTGGAACAACAGTGGGAGTCTCT-3′; 6, 5′-CGGTTCCAAGCACAGAGGCTCCTTCTGAACAATGAG-3′; 7, 5′-TCGACTCATTGTTCAGAAGGAGCCTCTGTGCTTGGAACC-3′; 8, 5′-GGTCATCGATGACACCACACTC-3′; 9, 5′-CAGGGCTCTTCTCAGGTGGTGCTTGTTCAAAAAAAGCCTCTAGG-3′; 10, 5′-CCTAGAGGCTTTTTTTGAACAAGCACCACCTGAGAAGAGCCCTG-3′; 11, 5′-GGCCCTAGGCTTGGAACCTGGATG-3′.
Figure 1Schematic representation of wild type (a) and mutant (b-i) apo(a) encoded by cDNA expression plasmids. Amino acid substitutions in the point mutants pCMV-A18_4174Arg and pCMV-A18_4057Ser are indicated under the corresponding kringles. The presence of a B in kringle 36 indicates the ability of apo(a) entities to form a disulfide bridge with apoB-100 in LDL. Oligonucleotides(
) and restriction sites involved in vector constructions are indicated: A, AvrII; B, BsmAI; C, Cfr10I; Cl, ClaI; E, EcoRV; Ea, EarI; N, NciI; S, SalI; Sm, SmaI.
), single-stranded template DNA from papoa3-7, oligonucleotide 1, and the U-DNA mutagenesis kit (Boehringer Mannheim) were employed to change codon 4174 in kringle 37 from TGG(Trp) to AGG(Arg) and to introduce by silent mutations a BsmI restriction site in order to facilitate the identification of mutant plasmids. Following sequence verification of the apo(a) cDNA insert in the mutated phagemid papoa3-7_4174Arg, a 448-base pair Cfr10I/EcoRV fragment in the apo(a) expression plasmid pCMV-A18 (
) were employed to introduce deletions at the 3′ end of the apo(a) cDNA in pCMV-A18 by replacement of restriction fragments with synthetic linkers. A 700-base pair AvrII/SalI fragment of pCMV-A18 was replaced with the annealed oligonucleotides 2 and 3 to obtain plasmid pCMV-A18 ΔV-P encoding an apo(a) glycoprotein lacking kringle V and the protease domain (Fig. 1b). Substitution of a BsmAI/SalI fragment in pCMV-A18 by the annealed oligonucleotides 4 and 5 gave rise to the mutant plasmid pCMV-A18 Δ37-P with a deletion of 3′ apo(a) sequences encoding kringle 37, kringle V, and the protease (Fig. 1c). A deletion of 3′ sequences encoding kringle 36, kringle 37, kringle V, and the protease in pCMV-A18 Δ36-P (Fig. 1d) was achieved by cloning of the annealed oligonucleotides 6/7 into NciI/SalI-restricted pCMV-A18. SmaI restriction of pCMV-A18 and subsequent religation of the large vector fragment resulted in the deletion mutant pCMV-A18 Δ32 (Fig. 1h) lacking kringle 32 sequences as compared with the wild type plasmid. Another deletion mutant characterized by the absence of kringle repeats 32-35 was obtained by PCR mutagenesis. Plasmids containing nucleotides 10050-10900 and nucleotides 11330-13062 of the apo(a) cDNA (
) were used together with oligonucleotides 8/9 and 10/11 to amplify two overlapping fragments containing kringles 30 and 31 and kringles 36 to V, respectively. Following ligation of the two purified fragments by a second PCR with primers 8/11 a 1.25-kilobase fragment obtained by ClaI/AvrII restriction of the PCR ligation was cloned into pCMV-A18. The resulting mutant plasmid pCMV-A18 Δ32-35 was identified by a new EarI restriction site in kringle 36 (Fig. 1i). Synthetic linker sequences and PCR-derived sequences were verified by sequencing.
DNA Sequence Analysis
DNA sequencing by the dideoxynucleotide chain termination method (
). Sixty to sixty-five h after transfection, cell culture supernatants were harvested, adjusted to 1 mM phenylmethylsulfonyl fluoride, and centrifuged for 10 min each at 300 and 4000 × g to remove cells and cell debris, respectively.
) and anti-mouse Ig (Fab′) peroxidase conjugate (Boehringer Mannheim) were combined in a sandwich ELISA assay. Plates were washed three times each with phosphate-buffered saline (PBS), 0.05% Tween-20 (Merck) between the different incubation steps. MaxiSorb microtiter plates (Nunc, Wiesbaden, FRG) were coated with sheep anti-human Lp(a) diluted 1:2000 in PBS, 0.02% sodium azide (200 μl/well) overnight at 4°C. Nonspecific binding sites were blocked with PBS, 0.1% casein (300 μl/well) for 30 min at room temperature prior to the addition of samples and standards (200 μl/well) for 2 h at 37°C. Standard dilutions in PBS, 0.1% casein ranged from 15 to 1000 ng/ml. Immunodetection of bound apo(a)•Lp(a) was obtained by successive incubations with 200 μl/well of 1) antibody 1A2, diluted to 5 μg/ml in PBS, 0.1% casein (2 h at 37°C); 2) anti-mouse Ig (Fab′)-peroxidase conjugate, diluted 1:350 in 50 mM Tris/HCl, 150 mM sodium chloride, 0.05% bovine serum albumin, 0.05% Tween-20, pH 7.5 (1 h at 37°C); and 3) 2,2′-azino-di[3-ethylbenzthiazoline-6-sulfonic acid] substrate solution (Boehringer Mannheim). After 30 min of color development absorbance was measured at 405 nm. The data are expressed in ng/ml Lp(a) as deduced from a standard curve obtained with Lp(a) reference standard (Immuno AG, Heidelberg, FRG).
SDS-PAGE and Immunoblotting
Fractionation of cell culture supernatants by reducing and nonreducing SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequent immunoblotting was performed as described previously (
) and the enhanced chemiluminiscence Western blot detection kit (Amersham Buchler GmbH, Braunschweig, FRG). The relative intensities of the bands corresponding to apo(a) glycoprotein and apo(a)•apoB complex were determined by densitometry using a UVP E.A.S.Y. plus system (Herolab, Wiesloch, FRG).
Human Plasma Samples
Venous blood samples from healthy donors were collected in heparinized tubes and centrifuged for 10 min at 300 g to remove cells. Following ELISA determination of Lp(a) plasma levels and prior to lysine-Sepharose affinity chromatography, the integrity of plasma apo(a)s was analyzed by SDS-PAGE and immunoblotting.
Lysine-Sepharose Affinity Chromatography
2 ml of hydrated lysine-Sepharose-4B (Pharmacia, Freiburg, FRG) was filled into a conical 0.8 × 4 cm polypropylene column (Poly-Prep, Bio-Rad). Chromatography was performed at a constant flow rate of 0.75 ml/min. Following equilibration of the matrix with 30 ml of washing buffer (0.1 M sodium phosphate buffer, 1 mM EDTA, 0.02% sodium azide, pH 7.4) 1-ml fractions were collected during all subsequent steps. Lp(a)/apo(a)containing samples (1 ml) were loaded onto the column either directly (culture supernatants of transfected HepG2 cells) or after a 10-fold dilution in HepG2 conditioned culture medium (human plasma samples). Unbound material was eluted from the column with 15 ml of washing buffer. Elution of lysine-bound material was achieved by subsequent application of 10 ml of washing buffer containing 50 mM of the lysine analog ε-amino caproic acid (EACA, Sigma). Lysine-binding chromatography was performed in the presence and in the absence of the nonionic detergent Tween-20. In the latter case, 1% (v/v) Tween-20 (Merck AG, Darmstadt, FRG) was added to the samples and to both washing and elution buffer. The amount of r-Lp(a)/r-apo(a) in each fraction was determined by ELISA. Lp(a)/apo(a) values were plotted against fraction numbers. For each sample, the percentage of lysine-binding Lp(a)/apo(a) was calculated as the amount of Lp(a)/apo(a) measured in fractions 16-25 (× 100) divided by the total amount of Lp(a)/apo(a) in fractions 1-25.
RESULTS
Characterization of HepG2-derived Lp(a)/apo(a)
Each wild type and mutant r-apo(a) plasmid was transfected at least three times. As measured by ELISA, the expression levels varied in the range between 300 ng/ml and 6.6 μg/ml. Following each series of transient transfections into HepG2 cells, r-apo(a)s were analyzed by nonreducing SDS-PAGE/immunoblotting to determine the amounts of apoB-complexed and free apo(a) glycoprotein in the medium of the transfected cells. Wild type r-apo(a) (A18), as well as the mutants ΔV-P, Δ37-P, and Arg-4174 were present primarily in the form of the apo(a)•apoB complex (Fig. 2A, lanes 1, 3, 4, and 5). Mutants Δ32-P and Δ36-P both lacking kringle 36, which mediates the covalent binding of apo(a) to apoB(
) has been replaced by serine, migrated as free apo(a) glycoprotein (Fig. 2A, lanes 2 and 6 and data not shown). Characterization of the deletion mutants Δ32-P, Δ37-P, and ΔV-P by reducing SDS-PAGE revealed the expected decrease in their apparent molecular weights (Fig. 2B, lanes 2, 3, and 4) as compared with the 18-kringle wild type and point mutants (Fig. 2B, lanes 1, 5, and 6).
Figure 2Immunoblot analysis of wild type and mutant r-apo(a)s. Following transient transfection of HepG2 cells with wild type and mutant apo(a) expression plasmids, culture supernatants were analyzed by SDS-PAGE under nonreducing (panel A) or reducing (panel B) conditions and subsequent immunoblotting. A phenotyping standard (St) containing the apo(a) isoforms S1, S2, S3, and S4 (
) has been included in panel B. Samples contained the following r-apo(a)s: lane 1, A18 (wild type); lane 2, Δ32-P; lane 3, ΔV-P; lane 4, Δ37-P; lane 5, Arg-4174; lane 6, Ser-4057. The positions of high molecular weight apo(a)•apoB complexes (c) and free apo(a) glycoproteins (A11 and A18) are indicated to the left of panel A.
In a first series of experiments, we compared the lysine binding of r-Lp(a)-like particles produced by HepG2 cells upon transfection with the wild type apo(a) plasmid pCMV-A18 (Fig. 1a) to the lysine binding of plasma-derived Lp(a). Based on a previous finding (
M. Helmhold and V. W. Armstrong, unpublished observations.
)of an increased binding of plasma Lp(a) to lysine-Sepharose upon addition of the nonionic detergent Tween-20, lysine binding activities were determined both in the absence and in the presence of 1% Tween-20.
HepG2-derived wild type r-Lp(a) obtained after three independent transfections revealed lysine binding of 30.6 and 92.2%, respectively, when analyzed in the absence and in the presence of Tween (Fig. 3, panels A and G, Table 1). Binding of r-Lp(a) to the lysine-Sepharose matrix could be completely blocked by the addition of 50 mML-lysine to sample and washing buffer, an effect that was not seen by using L-glycine at the same concentration (data not shown). Additional evidence for the specificity of the employed lysine-Sepharose method came from the estimation of HepG2-derived albumin and apolipoprotein A-I in the chromatography fractions. In contrast to r-Lp(a), both proteins were detected exclusively in the fall-through but not in the EACA-eluted fractions (data not shown). SDS-PAGE/immunoblotting and densitometric scanning of the r-apo(a) bands were employed to confirm the ELISA quantification of Lp(a) in the chromatography fractions and to demonstrate the integrity of r-apo(a) in all fractions of the lysine-Sepharose chromatography (data not shown). Variation in the expression rate of r-apo(a) led to some interexperimental variation in the amount of loaded r-Lp(a), which did not affect the lysine-binding data as indicated by the standard deviations in Table 1.
Figure 3Elution profiles from the characterization of r-Lp(a)/r-apo(a) by lysine-Sepharose affinity chromatography. Lysine binding of r-Lp(a)/r-apo(a) contained in the medium of transfected HepG2 cells was studied in the absence (panels A-F) and in the presence of 1% Tween-20 (panels G-M). The Lp(a) content of the fractions collected before and after the addition of EACA was determined by ELISA. The following samples were analyzed: panels A and G, A18 (wild type); panels B and H, mutant ΔV-P; panels C and I, mutant Δ37-P; panels D and K, mutant Δ32-P; panels E and L, mutant Arg-4174; panels F and M, mutant Ser-4057. The addition of EACA to the elution buffer is indicated by an arrow.
Plasma samples from 10 healthy donors (see “Experimental Procedures”) contained between 4 and 460 mg/liter Lp(a) without significant degradation of apo(a) isoforms as analyzed by immunoblotting (data not shown). Prior to affinity chromatography plasma samples were diluted 10-fold in HepG2-conditioned culture medium to mimic the binding conditions employed during the characterization of r-Lp(a). When the samples were analyzed in the absence of Tween, lysine binding expressed as percentage of Lp(a) bound to lysine-Sepharose (see “Experimental Procedures”) varied between 5.2 and 55.8% (mean = 26.2%). The addition of Tween caused a strong increase in the lysine binding of each plasma Lp(a) sample to values between 40.8 and 94.6% (mean 69.8%). There was no correlation between lysine binding and the level of plasma Lp(a), nor was the lysine-binding ability correlated with the apo(a) isoform size (data not shown).
Lysine Binding Activity of Lp(a) Is Mediated by Carboxyl-terminal Kringle IV Repeats
To delimit the structural domains of apo(a) that mediate lysine binding we constructed the plasmids pCMV-A18 Δ32-P (Fig. 1e), pCMV-A18 Δ36-P (Fig. 1d), pCMV-A18 Δ37-P (Fig. 1c), and pCMV-A18 ΔV-P (Fig. 1b) encoding carboxyl-terminal apo(a) deletion mutants with 11-, 15-, 16-, and 17-kringle domains. Upon transient expression in HepG2 cells the mutant ΔV-P, which lacks kringle V and the protease domain, revealed “wild type-like” lysine binding of about 30 or 90% when analyzed in the absence or in the presence of 1% Tween-20, respectively (Fig. 3, panels B and H, and Table 1). To assess the lysine-binding ability of the amino-terminal apo(a) kringles 1, 2A, 30, and 31 we transfected the plasmid pCMV-A18 Δ32-P encoding a 3′ truncated apo(a) lacking the carboxyl-terminal domains from kringle 32 to the protease domain (Fig. 1e). The mutant apo(a) glycoprotein did not associate in Lp(a)-like particles (Fig. 2A, lane 2) and revealed an almost complete loss of lysine binding both in the absence (2.2%) and in the presence (0.9%) of Tween (Fig. 3, panels D and K,Table 1).
) have suggested that kringle 37 of Lp(a) has retained the lysine binding properties observed in Pg kringle IV. Therefore, we analyzed r-Lp(a) containing an apo(a) without kringle 37, kringle V, and the protease domain for its lysine binding activity. Following transfection of plasmid pCMV-A18 Δ37-P (Fig. 1c) r-Lp(a) containing culture medium was characterized by lysine-Sepharose chromatography. In the absence of Tween, we detected only 3.7% lysine binding (Fig. 3, panel C, and Table 1). When analyzed in the presence of Tween, however, 68.6% of the Δ37-P mutant r-Lp(a) bound to lysine-Sepharose (Fig. 3, panel I, and Table 1). This value, intermediate between the binding of wild type (92.2%) and Δ32-P (0.9%) could be explained by the existence of additional LBSs within kringles 32-36 of apo(a), which might be masked in Lp(a). Direct evidence for this additional lysine binding activity independent of kringle 37 was obtained by the characterization of the mutants Ser-4057 and Δ36-P (Table 1). Because of a Cys → Ser substitution of the single unpaired cysteine in kringle 36 the mutant Ser-4057 apo(a) (Fig. 1g) is unable to form Lp(a) particles (
) as seen from the analysis by nonreducing SDS-PAGE (Fig. 2A, lane 6). Analyzed in the absence of Tween, this mutant revealed 53.4% lysine binding, which increased to 84.2% in the presence of Tween (Fig. 3, F and M, and Table 1). Mutant Δ36-P (Fig. 1d) lacking not only the ability to form Lp(a) particles (data not shown) but also the kringle 37 LBS exhibited 35.2 and 96% lysine binding in the absence and in the presence of Tween, respectively (Table 1).
Further Characterization of Kringle 37 Lysine-binding Site
The LBS of Pg kringle IV is formed by seven amino acids (
); two basic and two acidic residues interacting with the α-carboxyl and with the ε-amino group of the ligand are located on the opposite sides of a hydrophobic trough formed by a phenylalanin and two tryptophan residues. Of the 10 distinct kringle IV repeats in apo(a), kringle 37 has the highest probability of containing a functional LBS based on the highest overall amino acid identity (85%) and the presence of only a single conservative Lys → Arg substitution within the 7 residues that form the LBS of Pg kringle IV(
). Using site-directed mutagenesis (see “Experimental Procedures”) we changed codon 4174 in pCMV-A18 from TGG to AGG. This leads to a Trp → Arg substitution in apo(a) kringle 37 at a position homologous to Trp-426 of Pg (Fig. 1f). The mutation did not affect the ability of the mutant apo(a) to form a covalent complex with apoB in LDL as shown by nonreducing SDS-PAGE (Fig. 2A, lane 5). However, the lysine binding of r-Lp(a) containing Arg-4174 mutant apo(a) was almost abrogated (4.4%) when analyzed in the absence of Tween (Fig. 3, panel E, and Table 1). This result confirms the important role of the ultimate kringle IV repeat in apo(a) for the binding of Lp(a) particles to lysine. Furthermore, it indicates a conserved lysine-binding pocket between Pg kringle IV and kringle 37 of human apo(a). Analyzed in the presence of Tween, the Arg-4174 mutant r-Lp(a) retained 64.9% lysine binding (Fig. 3, panel L, and Table 1).
A Role for Lysine Binding during the Assembly of r-Lp(a) Particles?
From our binding studies with apo(a) mutants we are now able to assign lysine binding activity to the region between kringles 32 and 36 of apo(a). In order to assess a functional role of these domains during apo(a)•apoB complex formation we constructed two further apo(a) mutants A18 Δ32 (Fig. 1h) and A18 Δ32-35 (Fig. 1i) characterized by the absence of a single (kringle 32) or four (kringles 32-35) domains. Following transient expression in HepG2 cells we employed nonreducing SDS-PAGE and immunoblotting to analyze the abilities of these two deletion mutants to associate with apoB. As seen in Fig. 4, both mutants (lanes 4 and 5) revealed a decreased formation of covalent apo(a)•apoB complex when compared with wild type r-apo(a) (lane 3). The effect was more pronounced for the Δ32-35 mutant (lane5, 91% apo(a), and 9% complex) than for the Δ32 mutant (lane 4, 65% apo(a) and 35% complex).
Figure 4Immunoblot analysis of wild type and mutant r-apo(a)s expressed by HepG2 cells. Culture supernatants from transfected HepG2 cells were fractionated by SDS-PAGE under nonreducing conditions followed by immunoblotting with apo(a) specific monoclonal antibody 1A2. Lane 1, mutant Ser-4057; lane 2, apo(a) phenotyping standard (Immuno); lane 3, A18 (wild type); lane 4, mutant Δ32; lane 5, mutant Δ32-35. The positions of free r-apo(a) glycoprotein (a) and the r-apo(a)•apoB complex (c) are indicated to the right.
From the data presented here, we propose the existence of two functionally distinct LBSs within the apo(a) glycoprotein (Fig. 5). LBS I is situated in kringle 37 of apo(a) and is responsible for the 30.6% lysine binding of wild type r-Lp(a) particles observed in the absence of Tween. This is concluded from a comparison of the lysine binding activities determined for r-Lp(a)s containing wild type apo(a) with r-Lp(a) containing either of the apo(a) mutants, Δ37-P (Fig. 1c) or Arg-4174 (Fig. 1f). The mutant apo(a)s retained the ability to assemble into r-Lp(a) particles as shown by immunoblotting (Fig. 2A, lanes 4 and 5) and density gradient centrifugation (data not shown). In contrast to wild type Lp(a) both mutants failed to bind to the lysine matrix in the absence of Tween. This result is in agreement with predictions based on sequence comparisons and molecular modeling(
) that the failure of rhesus monkey Lp(a) to bind to this matrix is caused by the amino acid substitution Trp → Arg at the position homologous to Trp-4174 in human apo(a). The deletion mutant Δ32-P did not bind to lysine-Sepharose either in the absence or in the presence of Tween, indicating that the N-terminal apo(a) kringles 1, 2A, 30, and 31 do not contribute directly to the lysine binding activity of Lp(a). In fact, these domains might interfere with the overall lysine- and/or fibrin(ogen) activity of Lp(a) because of intramolecular interactions with LBS I in kringle 37. Crystals of tissue plasminogen activator kringle 2 represent an example where the lysine-binding pocket of one kringle is occupied by an internal lysine from another kringle(
). Such intramolecular masking (Fig. 5) could explain both our finding of 30% lysine binding activity for wild type r-Lp(a) and the significant interindividual variation in lysine binding of plasma Lp(a) ranging from 5 to 55%. The observation might result from differential masking by apo(a) isoforms present in the plasma samples. This hypothesis is supported by a recent report of apo(a) isoform-dependent fibrin(ogen) binding of plasma Lp(a)s(
). Clearly, further experiments are required to evaluate the contribution of intra- and intermolecular masking of LBS I within the Lp(a) particle.
Figure 5Model for the arrangement of LBS I and II of apo(a) within r-Lp(a). LBS I (I), localized in kringle 37 of apo(a) is exposed in 30% of wild type r-Lp(a). Masking of LBS I by either intramolecular kringle-kringle interactions or intermolecular interactions of apo(a) with apoB-100 and/or other components in the culture medium might explain the failure of 70% of r-Lp(a) particles to bind to the lysine matrix as indicated in the right panel. LBS II (II) has been mapped to kringles 32-36. The number and exact localization of the kringles that form LBS II remains to be resolved. Within r-Lp(a) particles these site(s) are masked but can be exposed in the presence of Tween-20. Kringles 1-31 of r-apo(a), kringle V, and the protease domain do not contain detectable lysine binding activity(-).
We could also localize substantial lysine binding activity to the region between kringles 32 and 35 (LBS II) in apo(a). In contrast to LBS I, LBS II is only operationally defined and may in fact represent lysine-binding sites within one or more of the respective kringles. On the basis of molecular modeling, Guevara et al.(
) proposed that in addition to kringle 37 the kringles 32-35 but not kringle 36 also contain potential lysine-binding sites. Further mutants will be required to accurately map LBS II. LBS II is masked in the native r-Lp(a) particle since neither the point mutant Arg-4174 nor the deletion mutant Δ37-P demonstrated significant binding to the lysine affinity matrix in the absence of Tween. In the presence of Tween 64.9% (Arg-4174) and 68.6% (Δ37-P) lysine binding was observed with these two mutants. Unmasking of LBS II in the presence of the detergent presumably explains the increased lysine binding of both plasma Lp(a) and r-Lp(a) species. Considering only wild type r-Lp(a) (Fig. 3, panels A and G), the effect of Tween might be caused by an increased accessibility of LBS I, e.g. by dispersal of aggregates that render LBS I inaccessible. This cannot be the reason, however, for our observations with the point and deletion mutants, specifically with the 35.2% lysine binding for the deletion mutant Δ36-P in the absence of Tween, and a second LBS must be inferred. We can only speculate about the molecular basis of the increase in lysine binding of Lp(a) and r-Lp(a). Tween-sensitive intra- and/or intermolecular interactions between apo(a) kringle domains might normally mask LBS II. Alternatively or additionally, noncovalent interactions between kringles 32-36 comprising LBS II and apoB in LDL might cause the relatively low lysine binding of wild type apo(a) (30.6%) and of the mutants Δ36-P (35.2%) and ΔV-P (33.6%) in the absence of Tween, which increased to more than 90% in the presence of the detergent. Proline-dependent interactions have been implicated in the apo(a)•apoB complex (
) and might be one target for the Tween effect. A Tween-sensitive association of apo(a) with components in HepG2 conditioned culture medium might be another explanation for the observed increase in lysine binding in the presence of the detergent.
Because of a substitution of Cys-4057, which is required for the assembly of r-Lp(a) (
) the expression of the mutant Ser-4057 in HepG2 cells yields free apo(a) glycoprotein. A 53.4% lysine binding of this mutant in the absence of Tween as compared with 30.6% for complexed wild type apo(a) suggests noncovalent interactions between the LBS of apo(a) and lysine residues of apoB in LDL. As hypothesized previously (
), such interactions might be involved in the positioning of cysteine residues in apo(a) and apoB prior to disulfide bridging of the two proteins. This hypothesis is supported but not proved by the decreased Lp(a) assembly observed for the deletion mutants Δ32 and Δ32-35 (Fig. 4). Additionally or alternatively to lysine binding, other noncovalent interactions between the deleted domain(s) and LDL might be involved in Lp(a) complex formation although the dose-dependent inhibition by EACA of the in vitro association between transgenic apo(a) and human LDL (
) argues in favor of a role of lysine binding during Lp(a) assembly. An involvement of LBS I in the assembly of r-Lp(a) can be excluded from our data since the mutants Δ37-P and Arg-4174 were assembled into r-Lp(a) particles as effectively as wild type apo(a).
Lysine-Sepharose affinity chromatography has been widely used to study in vitro the lysine binding characteristics of kringle proteins. Our mutational analysis provides an understanding of the structural elements determining the lysine binding properties of r-apo(a) and r-Lp(a). At this point, it should be stated that we cannot exclude the existence of additional low-affinity LBS in apo(a) that could not be detected with our model system because of the rather low concentrations of r-Lp(a)/r-apo(a) in the medium of transfected cells.
The substantial increase of lysine binding in the presence of Tween that we observed for both recombinant and plasma Lp(a) raises some questions about the physiological relevance of previous fibrin(ogen)-binding studies with plasma Lp(a)(
), which have been performed in the presence of Tween. Although the Tween concentrations in these studies were much lower than the one that we used in our lysine binding studies we were able to show that even the lowest concentration (0.01%), which has been used by Rouy et al.(
), results in a 1.6-fold increase in lysine binding of wild type r-Lp(a) from 30 to 48% (data not shown). Tween might lead to the exposure of fibrin-binding sites that are not necessarily identical with the postulated LBS. The identification of lysine binding activity in the kringle 32-36 domains of apo(a) raises the possibility that the described fibrin binding of Lp(a) is (in part) secondary to the “Tween effect.” In the absence of a physiological “Tween analog” it may be important to re-evaluate the fibrin-binding activity of plasma Lp(a) without demasking LBS II by addition of Tween to the assay buffer.
Apart from the well documented size variation of apo(a) isoforms in human plasma, recent data indicate a high degree of conservation in the structure of apo(a) isoforms(
), suggesting that our results obtained with r-Lp(a) can be extrapolated to human plasma Lp(a). In human atherosclerotic lesions Lp(a) has been co-localized with fibrinogen(
). In order to elaborate the hypothesis that lysine binding activity might contribute to the atherogenicity and/or thrombogenicity of Lp(a) it will now be important to use more physiological models to compare wild type and mutant r-Lp(a) for its binding to cells and/or fibrin(ogen) in the arterial wall. Without the availability of clinically defined Lp(a) mutants differing in their lysine binding properties a comparison of transgenic animals with circulating wild type or mutant Lp(a) seems to be most appropiate to investigate the impact of lysine binding for the proatherogenic and prothrombotic properties that have been attributed to Lp(a) particles. However, the reported binding of r-apo(a) to the arterial wall of apo(a) transgenic mice (
) with a masked LBS II. Double transgenic mice expressing both human apo(a) and human apoB-100 contain most of their apo(a) as components of Lp(a) particles (
) and should therefore represent a better model to investigate the putative atherogenic properties of Lp(a). Our present work is directed to use wild type and mutant Lp(a) transgenic animals to evaluate a putative implication of the described LBS I and II for the interaction of Lp(a) with components of the arterial wall.
81485-5&id=gr1.jpg){kind=link}
81485-5&id=gr2.jpg){kind=link}
81485-5&id=gr3.jpg){kind=link}
81485-5&id=tbl1.jpg){kind=link}
81485-5&id=gr4.jpg){kind=link}
81485-5&id=gr5.jpg){kind=link}