Quantitative Assessment of the Role of Apo(a) Kringle IV Types 2-10 in Particle Formation
We have developed a system for the quantitative assessment of the efficiency of lipoprotein(a) [Lp(a)] formation in vitro. Amino-terminally truncated derivatives of a 17-kringle form of recombinant apo(a) [r-apo(a)] were transiently expressed in human embryonic kidney cells. Equimolar amounts of r-apo(a) derivatives were incubated with a fourfold molar excess of purified human low density lipoprotein, and r-Lp(a) formation was assessed by densitometric analysis of Western blots. Although r-Lp(a) formation was observed with each r-apo(a) derivative, both the rate and extent of particle formation were greatly lower on removal of kringle IV type 7. Additional substantial decreases in these parameters were observed on removal of kringle IV type 8, thereby suggesting a major role for these two kringles in Lp(a) assembly. We directly demonstrated that the lysine-binding sites (LBSs) within kringle IV types 5-9 are “masked” in the context of the Lp(a) particle and are consequently unavailable for interaction with lysine-Sepharose. Using site-directed mutagenesis, we also demonstrated that the previously described LBS in kringle IV type 10 is not required for r-Lp(a) formation: r-Lp(a) formation using a mutated form of apo(a) that lacks this LBS is comparable in efficiency to that of wild-type r-apo(a) and can be inhibited to a similar extent by ε-amino-n-caproic acid. In summary, the results of our study indicate that apo(a) kringle IV types 7 and 8 are required for maximal efficiency of Lp(a) formation, likely by virtue of their ability to mediate lysine-dependent noncovalent interactions with apoB-100 that precede disulfide bond formation.
- Received May 3, 1995.
- Revision received May 2, 1996.
Lp(a) has become a major focus of study since elevated plasma levels of this lipoprotein have been identified as an independent risk factor for coronary heart disease1 2 3 and stroke.4 Marked inherited variability has been observed in plasma Lp(a) levels, which vary >1000-fold in the human population. With respect to lipid composition and the presence of apoB-100, Lp(a) closely resembles LDL. However, Lp(a) is clearly distinguishable from LDL by the presence of apo(a), which likely confers the unique structural and functional properties that have been attributed to Lp(a). Human apo(a) consists of multiple tandem repeats of a sequence that closely resembles plasminogen kringle IV, followed by sequences that exhibit a high degree of similarity to the kringle V and protease domains of plasminogen.5 All apo(a) isoforms contain 10 distinct classes of kringle IV sequences in apo(a); the kringle IV type 2 motif (also referred to as the major repeat kringle) is present in variable numbers and constitutes the molecular basis of Lp(a) isoform size heterogeneity.6 7 Of the kringle IV sequences in apo(a), the sequence of type 10 most closely resembles that of plasminogen kringle IV. Like plasminogen kringle IV, apo(a) kringle IV type 10 also has Lys-binding properties and has been postulated to mediate the interaction of Lp(a) with Lys residues in biological substrates such as fibrin.8 9 10 11
In the Lp(a) molecule, apo(a) is covalently linked to apoB-100 by a single disulfide bond that involves Cys4057, which is present in apo(a) kringle IV type 9.12 13 In studies that have utilized recombinant expression systems for human apo(a), strong evidence suggests that the assembly of Lp(a) particles occurs extracellularly in plasma.12 13 14 15 16 Several groups have assessed the role of specific apo(a) kringles in Lp(a): studies by Frank et al14 suggested an essential role for kringle IV type 6 in Lp(a) formation, and Ernst et al15 indirectly identified LBSs in apo(a) kringle IV types 5-8. After deletion of these sequences from an 18-kringle r-apo(a) species, a dramatic reduction in particle formation was observed. More recent data suggest a major role for kringle IV type 6 as a noncovalent LDL binding site.17 In general, studies to date suggest that Lp(a) formation is a two-step process involving an initial noncovalent interaction between apo(a) and apoB-100 that precedes specific disulfide bond formation. A large component of the noncovalent interactions is Lys dependent and can therefore be inhibited by addition of Lys analogues.14 15 16
In the current study, we used a recombinant expression system to quantitatively assess the role of apo(a) kringle IV types 2-10 in Lp(a) formation. Our results indicate that removal of kringle IV types 2-6 has little effect on the efficiency of Lp(a) assembly, whereas both kringle IV types 7 and 8 appear to be required for maximal Lp(a) formation.
Construction of Apo(a) Expression Plasmids
The r-apo(a) derivatives utilized in this study are shown schematically in Fig 1⇓; details of the construction of the corresponding expression plasmids are briefly described below. All constructs were based on the pRK5ha17 17-kringle r-apo(a) expression plasmid, which contains the cytomegalovirus promoter and simian virus 40 transcription-termination sequences.18
The 12-kringle–containing r-apo(a) species was constructed as previously described19 by fusion of a 385-bp EcoRI/Hha 1 fragment derived from pRK5ha17 (containing sequences that encode the 5′ untranslated and signal sequences and the first 291 bp of kringle IV type 1) to a 4828-bp Hha I/EcoRI fragment containing the last 51 bp of the kringle IV type 2 sequence followed by three complete kringle IV type 2 repeats, sequences corresponding to kringle IV types 3-10, and the kringle V and protease domains. These fragments were ligated into pRK5ha17, which had been digested with EcoRI to remove the apo(a) cDNA sequence; the resultant expression construct was designated pRK5ha12. Note that the first kringle IV repeat in this construct is a hybrid, consisting of the first 291 bp of kringle IV type 1 and the last 51 bp of kringle IV type 2 (see Fig 1⇑).
A six-kringle–containing r-apo(a) (designated 6K) was constructed by fusion of a 194-bp EcoRI/HindIII fragment derived from pRK5ha17 that contained sequences encoding the 5′ untranslated and signal sequences and the first 100 bp of kringle IV type 1 to a 3003-bp HindIII/Sal I fragment that contained the last 242 bp of kringle IV type 5 and sequences encoding kringle IV types 6-10 and kringle V and protease domains. These fragments were ligated into pRK5ha17, which had been digested with EcoRI and Sal I; the resultant expression construct was designated pRK5haIV5-P. Note that the first kringle IV repeat in this construct is also a hybrid, consisting of the first 100 bp of kringle IV type 1 and the last 242 bp of kringle IV type 5 (see Fig 1⇑). A corresponding construct lacking kringle V and protease sequences was generated as follows. A 1678-bp Avr II fragment was isolated from the pRK5haIV5-P plasmid and replaced with a 580-bp Avr II fragment from the pRK5-SK10 construct [encoding single apo(a) kringle IV type 1020 ], which resulted in a stop codon following the kringle 5-10 sequence; the resulting r-apo(a) encoded by pRK5haKIV5-10 was designated 6KΔVP. To generate a plasmid encoding apo(a) kringle IV types 5-9, a stop codon was introduced by PCR with oligonucleotides 1a and 1b as primers (see the Table⇓) and pRK5haIV5-10 as the template. The 5′ primer was designed to span the Msc I site in kringle IV type 10 and contains a stop codon immediately upstream of this site. The 3′ primer spans the Avr II site at the 3′ end of the apo(a) kringle IV type 10 sequence. PCR amplification with these primers and subsequent digestion with Msc I and Avr II resulted in a 281-bp fragment that was used to replace the corresponding Msc I/Avr II fragment in the pRK5haKIV5-10 plasmid. The final construct (pRK5haKIV5-9) encodes kringle IV types 5-9 followed by the first 19 amino acids of kringle IV type 10. Note that as for pRK5haKIV5-10, the first kringle IV repeat in this construct is a hybrid of kringle IV types 1 and 5 (see Fig 1⇑).
To construct a plasmid encoding apo(a) kringle IV types 6-10 followed by kringle V and protease sequences, two oligonucleotides (2a and 2b, the Table⇑) containing 20 bp of a complementary sequence were annealed and extended with T4 DNA polymerase (GIBCO/BRL) as previously described.20 The resultant double-stranded fragment contained sequences corresponding to the apo(a) 5′-untranslated region, the signal sequence, and the first 18 bp of kringle IV type 6, with an EcoRI site at the 5′ end of the fragment and an Xma I site at the 3′ end. This fragment and a 1702-bp Xma I/Sal I fragment isolated from pRK5haKIV5-10 [containing the last 324 bp of kringle IV type 6 and thereby reconstituting the complete apo(a) kringle IV type 6 sequence followed by the kringle IV types 7-10 sequence; see above] were ligated into the pRK5ha17 vector that had been digested with EcoRI and Sal I. The kringle V and protease sequences were added by replacing a 580-bp Avr II fragment in this latter construct with a 1678-bp Avr II fragment from pRK5ha17. The final expression construct was designated pRK5haKIV6-P.
A construct encoding apo(a) kringle IV types 7-10 followed by kringle V and protease domains necessitated the generation of several intermediate constructs. Two synthetic oligonucleotides (3a and 3b, the Table⇑) containing 17 bp of a complementary sequence were annealed and extended as previously described. The resultant double-stranded fragment contained the apo(a) 5′-untranslated and signal sequences as well as the first 38 bp of kringle IV type 7, with an EcoRI site at the 5′ end and a BstXI site at the 3′ end. An intermediate plasmid was constructed by ligating this fragment and a 325-bp BstXI/HindIII fragment (obtained by digestion of a construct encoding single kringle IV type 9; M.L.K., unpublished data, 1995) into EcoRI/HindIII–digested pBluescript SK+ (Stratagene). A 684-bp fragment spanning kringle IV types 7-9 obtained by BstXI digestion of pRK5haKIV5-9 was ligated into the intermediate vector that had been digested with BstXI. Sequences encoding kringle IV types 7-9 were then excised from the intermediate plasmid by digestion with EcoRI and HindIII and ligated into the pRK5ha17 vector that had been digested with EcoRI and HindIII. The resulting intermediate expression construct was designated pRK5haKIV7-9. The construct encoding apo(a) kringle IV types 7-10 was obtained by digestion of pRKhaKIV7-9 with EcoRI and Alu I; the resultant 385-bp fragment (containing the 5′-untranslated and signal sequences followed by sequences corresponding to the first 325 bp of kringle IV type 7) was ligated to a 355-bp Alu I/Sal I fragment derived from pRK5haKIV6-10. The latter fragment contained the remainder of kringle IV type 7 and the complete sequence of kringle IV types 8-10 followed by a stop codon. These two fragments were ligated into pRK5ha17 digested with EcoRI and Sal I; the resultant expression construct was designated pRK5haKIV7-10. The kringle V and protease sequences were added by replacing a 580-bp Avr II fragment with a 1678-bp Avr II fragment that contained these sequences as described above; the final construct was designated pRK5haKIV7-P.
An expression plasmid encoding apo(a) kringle IV types 8-10 followed by kringle V and protease domain sequences was generated as follows. The intermediate construct encoding kringle IV types 7-10 (see above) was linearized with Sal I and subjected to partial digestion with Pst I. A 857-bp partial Pst I/Sal I fragment (containing the last 163 bp of kringle IV type 8 followed by sequences corresponding to kringle IV types 9 and 10) and a 236-bp EcoRI/Pst I fragment (isolated from an intermediate construct corresponding to kringle IV types 8 and 9 and containing the first 179 bp of kringle IV type 8) were ligated into EcoRI/Pst I–digested pRK5ha17. The 580-bp Avr II fragment in the resultant construct was replaced with a 1678-bp Avr II fragment from pRK5ha17 as described above for pRK5haKIV7-P; the final construct was designated pRK5haKIV8-P. Note that this construct contains the entire sequence of apo(a) kringle IV type 8 (see Fig 1⇑).
We used the following strategy to assemble an expression construct encoding apo(a) kringle IV types 9 and 10 followed by sequences encoding the kringle V and protease domains. A 273-bp EcoRI/Pst I fragment was isolated by digestion of a construct encoding a single copy of apo(a) kringle IV type 9; this fragment was used to replace the corresponding EcoRI/Pst I fragment in a previously described construct that encodes apo(a) kringle IV type 10 (pRK5-SK10; see Reference 20). A 684-bp Pst I fragment spanning kringle IV types 9 and 10 was obtained by PCR amplification of pRK5haKIV5-10 (see above) by using oligonucleotides 4a and 4b (the Table⇑); this PCR fragment was cloned into the Pst I site of the intermediate vector described above, resulting in an expression construct encoding apo(a) kringle IV types 9 and 10. To obtain a construct containing kringle IV types 9 and 10 as well as kringle V and protease sequences, a 580-bp Avr II fragment in the intermediate vector was replaced with a 1678-bp Avr II fragment derived from pRK5ha17 as described above; the final construct was designated pRK5haKIV9-P.
To disrupt the LBS in kringle IV type 10 (in the context of the 17-kringle construct), a Trp→Arg substitution was introduced at position 72 by PCR-mediated mutagenesis. With pRK5-SK10 as the template, primers 5a and 5b (the Table⇑), and primers that flank the multiple cloning site in pRK5, two overlapping PCR products were generated. The mismatches were incorporated into oligonucleotides 5a and 5b (boldface type in the Table⇑) and mutate the Trp (TGG) to an Arg (CGC) codon and create an Hha I restriction site (underlined type in the Table⇑). The PCR products were digested with EcoRI/Hha I and Hha I/Sal I, respectively, and inserted into the pRK5-SK10 expression plasmid that had been digested with EcoRI and Sal I. A 273-bp Msc I/Avr II fragment was isolated from this plasmid and inserted into the pRK5ha17 plasmid that had been digested with these enzymes to replace the corresponding wild-type sequence. The mutant expression construct was designated pRK5ha17ΔTrp.
All constructs were verified by DNA sequence analysis, and the corresponding protein was analyzed by transient transfection of human embryonic kidney cells and Western blot analysis of CM from transfected cells (see below).
Human embryonic kidney (293) cells21 were cultured in 100-mm dishes in MEM (GIBCO/BRL) supplemented with 5% fetal calf serum. Cells were transiently transfected by calcium phosphate precipitation22 with 10 μg plasmid DNA per culture dish. The precipitate was left on the cells for 7 hours, after which time the transfected cells were washed thoroughly with PBS and incubated for 48 hours in 6 mL Optimem (GIBCO/BRL). After incubation, the CM was harvested and clarified by brief centrifugation at 1845g. With the exception of 17KΔTrp, 293 cells were stably transfected with each of the r-apo(a) derivatives shown in Fig 1⇑. For generation of stably expressing cell lines, expression plasmids (10 μg) were cotransfected with 1 μg of a plasmid encoding the neomycin resistance gene23 by calcium phosphate coprecipitation. Transfectants were selected by culturing the cells with 800 μg/mL G418 (GIBCO/BRL) as previously described,18 and expression levels were determined by ELISA.
For generation of metabolically labeled 17-kringle r-apo(a), cells were transfected with the pRK5ha17 plasmid as described above. After overnight recovery, the cells were preincubated for 1 hour in Met/Cys–depleted MEM (GIBCO/BRL) without fetal calf serum. [35S]Cys (ICN) was then added (30 μCi/mL media). CM containing labeled r-apo(a) was harvested 3.5 hours after labeling and clarified by brief centrifugation.
CM from 293 cells stably expressing the various r-apo(a) derivatives was loaded onto a 50-mL Lys-Sepharose (Pharmacia) column. The column was washed with PBS containing 0.5 mol/L NaCl, and protein was eluted with 0.2 mol/L ε-ACA in this buffer. Protein-containing fractions were pooled, dialyzed against PBS, and precipitated overnight with (NH4)2SO4. The precipitate was pelleted by centrifugation at 12 000g for 20 minutes at 4°C, dissolved in HEPES-buffered saline (20 mmol/L HEPES and 150 mmol/L NaCl, pH 7.4), and dialyzed against this buffer. The protein concentration was determined by measurement of absorbance at 280 nm. Extinction coefficients for each r-apo(a) protein were determined as previously described for the 17-kringle r-apo(a).18
LDL within the 1.006<d<1.05 g/mL density range was isolated from human plasma by sequential flotation.24 In brief, plasma (containing 1 mmol/L PMSF, 1 mmol/L EDTA, and 0.02% NaN3) was centrifuged at 436 000g for 2 hours at 15°C. The d<1.006 g/mL fraction was removed and the infranatant density adjusted to d<1.05 g/mL with NaBr and centrifuged for another 2 hours. At this time, the d<1.05 g/mL fraction was isolated and centrifuged at d=1.05 g/mL for another 2 hours under the conditions described above. LDL isolated from this centrifugation step contained no contaminating Lp(a), as determined by Western blot analysis and ELISA.
Determination of r-Apo(a) Expression Levels in CM of Transfected Cells
The apo(a) protein content of CM harvested from transiently transfected 293 cells was measured by sandwich ELISA. Microtiter wells were coated with an anti-apo(a) sheep polyclonal antibody (100 μL of a 10 μg/mL solution in 0.1 mol/L NaHCO3 buffer; Affinity Biologicals) and left overnight at 4°C. Nonspecific binding sites were blocked by incubation with 150 μL of a 5% solution of BSA in HEPES-buffered saline for 1 hour at 22°C. Microtiter wells were then incubated for 2 hours at 22°C with either CM harvested from transfected cells (diluted in HEPES-buffered saline containing 2.5% BSA and 0.1% Tween-20) or known amounts of corresponding purified r-apo(a) derivatives with the exception of the 17ΔTrp derivative, for which purified 17-kringle r-apo(a) was used as the standard. Then the wells were washed and incubated for 1 hour with 100 μL of a 1:3000 dilution of the monoclonal antibody a-34 (1 mg/mL stock concentration). This antibody belongs to a monoclonal series raised against human Lp(a)25 and is specific for the protease region of apo(a) (S.M.M. and M.L.K., unpublished data, 1995). For detection, the wells were incubated with 100 μL of a 1:5000 dilution of sheep anti-mouse IgG horseradish peroxidase–conjugated antibody (Amersham) for 1 hour, after which 100 μL of development buffer containing O-phenylenediamine dihydrochloride (0.4 mg/mL) was added. The reaction was stopped by adding 2 mol/L H2SO4, and the absorbance at 492 nm was determined with a Titertek plate reader.
r-Apo(a) concentration in the CM from transfected cells was subsequently determined with ELISA standard curves corresponding to the respective purified r-apo(a) derivatives. The standard curves generated from these data were found to be linear up to 250 pmol/L for each purified r-apo(a) derivative (r2>.994).
Analysis of r-Lp(a) Particle Formation
To determine the effect of incubation time on r-Lp(a) formation, purified human LDL (50 nmol/L) was incubated with each r-apo(a) derivative (12.5 nmol/L contained in CM harvested from transfected 293 cells) at 37°C in a total volume of 150 μL. At selected times (0, 0.5, 1, 2, 4, 8, and 20 hours), a 15-μL aliquot was removed from the incubation mixture, added to an equal volume of 2× Laemmli sample buffer26 without reducing agents, and stored at −20°C. After assay, r-Lp(a) assembly was assessed by Western blot analysis. In brief, samples were boiled and resolved by SDS-PAGE on 3% to 12% gradient gels, which were subsequently transferred to Immobilon-P nylon membranes (Millipore) at 100 volt·hours in buffer containing 25 mmol/L Tris, 192 mmol/L Gly, and 10% methanol, pH 8.3. The membrane was incubated with the apo(a)-specific antibody a-34 (1:4000 dilution of a 1 mg/mL stock) and visualized with a horseradish peroxidase–conjugated sheep anti-mouse IgG antibody and chemiluminescence (Amersham).
To determine the effect of a variety of inhibitors on Lp(a) formation, 1 mL of [35S]Cys-labeled 17-kringle r-apo(a) was incubated with 2.5 μg of purified human LDL with increasing concentrations of l-Pro, l-Lys, N-α-acetyl-l-Lys, N-ε-acetyl-l-Lys (all from Aldrich), or ε-ACA (Sigma) for 3 hours at 37°C. Then r-Lp(a) formation was terminated by adding ε-ACA to a final concentration of 0.3 mol/L in all cases. Reaction mixtures were subsequently immunoprecipitated with a monoclonal antibody specific for apo(a) (2G7, see Reference 27) as previously described.18 Immunoprecipitates were analyzed by SDS-PAGE under nonreducing conditions on 2.5% to 15% gels; the gels were treated with Enlightning (DuPont), dried under vacuum, and exposed to Hyperfilm (Amersham).
To quantify Lp(a) formation, either Western blots or autoradiograms were scanned with a Hewlett-Packard ScanJet 3c flatbed scanner and analyzed using Corel PhotoPaint (version 5.0, Corel Corp) and SigmaGel (version 1.0, Jandel Scientific) software. With these methods, the ratio of r-Lp(a) particles to total r-apo(a) [total r-apo(a)=r-apo(a)+r-Lp(a)] was determined (ie, percent r-Lp(a) formed at various times or in the presence of different inhibitors of particle assembly).
Analysis of the Role of Apo(a) Lys-Binding Motifs in Lp(a) Formation
r-Lp(a) species formed in vitro as described above using KIV5-9 and 6KΔVP r-apo(a) derivatives were applied to 1-mL Lys-Sepharose (Pharmacia) columns. Columns were washed with 8 mL PBS, and specifically bound protein was eluted with 0.2 mol/L ε-ACA in PBS containing 0.5 mol/L NaCl. One-milliliter fractions were collected, immunoprecipitated with the apo(a)-specific polyclonal antibody, and analyzed by SDS-PAGE and fluorography as described above.
Contribution of Specific Kringles to the Time Course and Extent of Lp(a) Formation
To delineate the structural requirements in apo(a) for Lp(a) formation, a series of r-apo(a) derivatives that lacked specific domains was constructed (Fig 1⇑). For quantitation of r-Lp(a) formation, the following derivatives were expressed transiently in human embryonic kidney cells: 17K, 17KΔTrp, 12K, 6K, KIV6-P, KIV7-P, KIV8-P, and KIV9-P. The amount of each derivative in the CM harvested from transfected cells was determined by ELISA as described in “Methods.” Equimolar amounts (25 nmol/L) of each r-apo(a) derivative contained in the CM harvested from transfected cells were subjected to Western blot analysis (Fig 2⇓) using the anti-apo(a) monoclonal antibody a-34, which recognizes an epitope in the apo(a) protease-like domain.
We examined the effect of amino-terminal deletion of apo(a) on the time course of r-Lp(a) formation. CM containing r-apo(a) derivatives (final concentration, 12.5 nmol/L) was incubated with purified human LDL (final concentration, 50 nmol/L; ie, a fourfold molar excess) at 37°C. Aliquots of the reaction mixtures were removed at specific times and subjected to SDS-PAGE and Western blotting. The percentage of r-apo(a) incorporated into r-Lp(a) particles was assessed by densitometric analysis. The results of these experiments are shown in Fig 3⇓. Comparison of the time course of r-Lp(a) assembly from the 17-kringle form of apo(a) and one of its smaller derivatives (12K) that contains only three copies of apo(a) kringle IV type 2 (see Fig 1⇑) suggested that removal of multiple copies of kringle IV type 2 resulted in more efficient r-Lp(a) formation. Note that at the 20-hour time point, ≈89% of 12K r-apo(a) was incorporated into r-Lp(a) compared with only ≈70% of the 17-kringle species. Additionally at each time point examined, the amount of r-Lp(a) observed for the 12-kringle–containing apo(a) derivative was greater than that observed for 17K r-apo(a) (Fig 3A⇓). Calculated initial rates of r-Lp(a) formation, as determined by regression analysis over the linear portion of the time course curve, were also quite different for these two r-apo(a) species [1.8 and 0.9 nmol·L−1·h−1 for 12K and 17K r-apo(a) derivatives, respectively].
The preceding results suggested that removal of amino-terminal kringles might enhance the rate and extent of r-Lp(a) formation. Accordingly, the effect of removal of kringle IV types 1-6 on r-Lp(a) formation was examined. The 6K, KIV6-P, and KIV7-P r-apo(a) derivatives exhibited comparable time courses for the formation of r-Lp(a) (Fig 3B⇑). For each of these r-apo(a) species, >40% of apo(a) was incorporated into r-Lp(a) particles within the first 2 hours of the reaction. The rate of r-Lp(a) formation was 3.1, 4.3, and 3.7 nmol·L−1·h−1 for 6K, KIV6-P, and KIV7-P derivatives, respectively. The amount of r-Lp(a) observed at the 20-hour time point was >80% for each of these apo(a) derivatives. Furthermore, these derivatives exhibited higher rates of r-Lp(a) formation relative to 17K and 12K apo(a) (compare 17K, 12K, and KIV7-P in Fig 3A⇑).
Removal of sequences for apo(a) kringle IV types 7 and 8 resulted in a considerable decrease in the amount of r-Lp(a) at each time point (Fig 3A⇑; compare KIV7-P, KIV8-P, and KIV9-P). Removal of KIV7 resulted in relative decreases of ≈40% to 50% at each time point. Furthermore, the estimated initial rate of r-Lp(a) formation decreased from 3.7 nmol·L−1·h−1 for KIV7-P to 1.63 nmol·L−1·h−1 for KIV8-P. Removal of apo(a) kringle IV type 8 resulted in an even more pronounced effect on r-Lp(a) formation: the initial rate of particle formation dropped to 0.3 nmol·L−1·h−1 with the KIV9-P derivative. The total amount of r-Lp(a) formed after 20 hours with the latter apo(a) species was only 22% compared with 57% for KIV8-P and 89% for KIV7-P.
On the basis of time course curves for r-Lp(a) formation shown in Fig 3⇑, we elected to examine the extent of r-Lp(a) formation after a 20-hour incubation at 37°C [note that at this time point, Lp(a) formation is maximal for all r-apo(a) derivatives in this study; see Fig 3⇑]. In Fig 4⇓, representative Western blots are shown from experiments in which equimolar amounts of each apo(a) derivative (12.5 nmol/L) were incubated for 20 hours at 37°C with a fourfold molar excess of LDL (50 nmol/L). As shown in Fig 4⇓, all apo(a) species formed r-Lp(a) particles under these conditions, although the total amount formed was dependent on the composition of the r-apo(a) derivative. For each r-apo(a) derivative, the percentage of r-apo(a) incorporated into r-Lp(a) was assessed by densitometric analysis of the Western blots (Fig 5⇓). The amount of r-Lp(a) for 12K, 6K, KIV6-P, and KIV7-P derivatives all exceeded 80% and the differences were not significantly different (P>>.05 by paired Student's t test). The amounts of r-Lp(a) for KIV8-P (≈57.1%) and KIV9-P (≈19%) were significantly less than those for other r-apo(a) species (P<.05 by paired Student's t test). The apo(a) derivative 17KΔTrp, containing a Trp→Arg substitution at position 72 in the kringle IV type 10 sequence, behaved identically to wild-type 17K r-apo(a) with respect to maximal r-Lp(a) formation and the time course thereof (Fig 5⇓ and data not shown). This result demonstrated that Lys-binding properties of apo(a) kringle IV type 10 do not contribute to Lp(a) assembly in vitro.
Role of Apo(a) Lys-Binding Motifs in r-Lp(a) Formation
CM harvested from metabolically labeled cells transfected with expression plasmids encoding either 6KΔVP or KIV5-9 (see Fig 1⇑) was incubated with 2.5 μg of purified human LDL for 1 hour. After incubation, reaction mixtures were directly applied to Lys-Sepharose columns that were washed with PBS; specifically bound proteins were eluted with 0.2 mol/L ε-ACA in PBS containing 0.5 mol/L NaCl. Column fractions were immunoprecipitated and analyzed by SDS-PAGE and fluorography (Fig 6A⇓). We observed that r-Lp(a) with KIV5-9 did not bind to Lys-Sepharose, whereas most r-Lp(a) containing 6KΔVP bound specifically to this resin. In contrast, free (uncoupled) r-apo(a) corresponding to 6KΔVP and KIV5-9 bound specifically to Lys-Sepharose columns (Fig 6A⇓). This suggests that Lys-binding motif(s) in KIV5-9 are involved in Lp(a) formation and are therefore inaccessible for binding to Lys-Sepharose when associated with LDL. In r-Lp(a) species containing 6KΔVP, the kringle IV type 10 sequence is available to interact with Lys-Sepharose, suggesting that this kringle does not interact specifically with Lys residues in apoB-100 in the context of the Lp(a) particle. These results imply that apo(a) kringle IV type 10 confers Lys-binding properties to Lp(a) particles.
A mutant form of 17-kringle apo(a) containing a Trp→Arg substitution at position 72 in the kringle IV type 10 sequence was used in Lp(a) assembly studies. This mutation removes one of the residues that, by analogy with plasminogen kringle IV, is required for Lys binding by this kringle.28 CM harvested from cells transfected with the mutant 17-kringle plasmid (pRK5ha17ΔTrp) or wild-type 17-kringle construct (pRK5ha17) was incubated with purified human LDL, and the effect of adding either NaCl or ε-ACA to r-Lp(a) formation was assessed (Fig 6B⇑). Both wild-type and mutant forms of 17-kringle r-apo(a) were able to efficiently couple with LDL, as shown in the fluorogram in Fig 6B⇑ and in the quantitative assay in which the extent of particle formation with either of these r-apo(a) derivatives was essentially identical (Fig 5⇑). In both cases, addition of NaCl (2 mol/L) had no effect on particle formation, whereas addition of ε-ACA (0.2 mol/L) essentially abolished Lp(a) particle formation in both cases. This observation combined with the data presented in Figs 4 and 5⇑⇑ clearly demonstrate that Lys-binding properties of apo(a) kringle IV type 10 are not required for Lp(a) formation.
Inhibition of Lp(a) Assembly by Lys, Lys Analogues, and Pro
We compared the relative abilities of Lys, Lys analogues, and Pro to inhibit Lp(a) assembly using 17-kringle r-apo(a). In brief, purified human LDL was incubated with CM containing metabolically-labeled 17K r-apo(a) with concentrations of the inhibitors listed in Fig 7⇓. After a 3.5-hour incubation at 37°C, reaction mixtures were immunoprecipitated and analyzed by SDS-PAGE and fluorography. Lp(a) formation was determined by densitometric analysis of the ratio of r-Lp(a) to total r-apo(a) (see above). An IC50 value for each potential inhibitor was determined graphically from the curves in Fig 7⇓. The IC50 values were as follows: N-α-acetyl-l-Lys (9.5 mmol/L) <ε-ACA, l-Lys (11 mmol/L) <Pro (20 mmol/L)<<N-ε-acetyl-l-Lys (66 mmol/L).
We used a recombinant expression system to identify sequences in apo(a) that are necessary for the formation of Lp(a) particles. Our assay differs from previous studies15 17 in that the former involves addition of equivalent amounts of r-apo(a) derivatives to purified human LDL, thereby allowing direct comparison of the rates and extents of r-Lp(a) formation. Initially we sought to determine the effect of amino-terminal truncation of apo(a) on the rate of formation of r-Lp(a) particles. We found that removal of kringle sequences up to kringle IV type 5 corresponded to marginally higher rates of r-Lp(a) formation (compare initial rates of particle formation for 17K, 12K, 6K, KIV6-P, and KIV7-P in Fig 3⇑). However, removal of kringle IV types 7 and 8 resulted in a greatly reduced rate of particle formation (1.63 and 0.3 nmol·L−1·h−1 for KIV8-P and KIV9-P, respectively). This trend paralleled that of the percentage of r-Lp(a) formed with these species (see below) and suggested that additional kringle sequences (ie, other than those in core sequences required for particle formation) reduced the efficiency of r-Lp(a) formation. This may occur because of more nonproductive interactions between apo(a) and apoB-100 or alterations in the accessibility of apo(a) sequences that are required for interaction with apoB-100. Interestingly, the data of Frank et al14 also suggest that larger apo(a) constructs may couple less efficiently to LDL, although this result was not discussed. Our data suggest that increasing numbers of kringle IV type 2 sequences (ie, the major repeat kringle) that correspond to larger apo(a) isoforms may decrease the extent to which these isoforms form Lp(a) particles. This trend would be compatible with the general inverse relationship that has been observed between apo(a) isoform size and plasma Lp(a) levels.29 In terms of identifying the minimum number of apo(a) sequences that are required for Lp(a) assembly, it appears that neither the kringle IV type 2 sequence (the major repeat kringle) nor sequences that encode kringle IV types 3 and 4 are required. These conclusions are based on the data in Fig 3⇑, which demonstrate that r-apo(a) species 12K and 6K form r-Lp(a) particles to a similar extent during a 20-hour incubation (note that these constructs differ from each other in that 6K lacks kringle IV types 2, 3, and 4; see Fig 1⇑).
The present study indicates that kringle sequences that are involved in the interaction between apo(a) and apoB-100 and are therefore directly involved in Lp(a) assembly are present within kringle IV types 5-9. This is suggested by several lines of evidence. First, comparable amounts of r-Lp(a) were observed when 6K r-apo(a) (containing kringle IV types 5-10 followed by kringle V and protease domains) and 12-kringle r-apo(a) were used. Second, direct interaction between apo(a) kringle IV types 5-9 and apoB-100 is clearly demonstrated in Fig 6A⇑. We found that r-Lp(a) particles formed with apo(a) kringle IV types 5-9 were unable to bind to Lys-Sepharose, despite previous demonstrations by others that kringle IV types 5-8 contain LBSs.15 This clearly suggests that the LBSs in kringle IV types 5-8 are masked in the context of the Lp(a) particle. With respect to identifying those sequences within kringle IV types 5-9 that are necessary for Lp(a) formation, our study suggests that no sequences within these kringles are required for Lp(a) formation. This contrasts with the findings of Frank et al14 and Trieu and McConathy,17 whose data suggest an essential role for apo(a) kringle IV type 6 in Lp(a) formation [in the case of the data reported by Frank et al,14 no r-Lp(a) was formed with r-Lp(a) species that lacked kringle IV type 6]. In the present study, we found little or no role for kringle IV type 6 in Lp(a) assembly [compare ≈91% r-Lp(a) with KIV6-P and ≈89% r-Lp(a) with KIV7-P; this difference is not statistically significant]. The reasons for this discrepancy are unclear. However, it is important to note that in the studies of Frank et al,14 their r-apo(a) derivatives contained a number of hybrid kringle sequences that may have grossly affected the overall structure and/or binding specificity of r-apo(a). In our study, the cloning strategy for 6K, 6KΔVP, and KIV5-9 created a hybrid kringle at the N-terminus that contained the first 34 amino acids of kringle IV type 1 and the last 80 amino acids of kringle IV type 5. As such, this hybrid kringle would be expected to maintain the putative Lys-binding properties of kringle IV type 5, since all of the residues thought to be important for Lys binding were maintained in the hybrid kringle. All other deletion constructs were designed to maintain the intact kringle sequences. In the recent study by Trieu and McConathy,17 the role of apo(a) kringle IV type 6 was assessed by insertional inactivation (addition of 4 amino acids) that may also have affected conformational and binding properties of the corresponding r-apo(a) derivative. Additionally, the latter study examined noncovalent association [binding of biotinylated LDL to immobilized r-apo(a) species] rather than formation of bona fide covalent Lp(a) particles.
Our data strongly suggest that the Lys-binding motif in apo(a) kringle IV type 10 is responsible for the Lys-binding properties of Lp(a) particles. This is directly demonstrated by our results in which r-Lp(a) particles formed from r-apo(a) derivative KIV5-9 were incapable of interacting with Lys-Sepharose, whereas r-Lp(a) species containing 6KΔVP bound specifically to this resin (Fig 6A⇑). A similar result was shown by Ernst et al,15 in which r-Lp(a) formed from r-apo(a) that lacked sequences carboxyl to kringle IV type 9 was unable to bind to Lys-Sepharose. This is further supported by our site-directed mutagenesis data for apo(a) kringle IV type 10. Substitution of Trp for Arg at position 72 of the KIV10 sequence of 17-kringle r-apo(a) had no effect on the efficiency of r-Lp(a) formation relative to wild-type 17-kringle r-apo(a). However, r-Lp(a) formed from r-apo(a) containing the Trp→Arg substitution failed to interact with Lys-Sepharose (data not shown); this finding concurs with previously published data.15 Our results suggest that if the KIV10 sequences play a role in Lp(a) formation, it is not a consequence of their Lys-binding properties. Rather, the KIV10 sequence as a whole may be necessary for maintaining in the correct conformation those sequences in apo(a) that directly interact with apoB-100 [ie, apo(a) kringle IV types 5-9]. Alternatively, the KIV10 sequence may contribute hydrophobic or electrostatic interactions that are not Lys dependent. In support of these latter hypotheses, we observed reduced r-Lp(a) formation with an apo(a) derivative that contained KIV6-9, compared with its counterpart that contained apo(a) kringle IV type 10 (B.R.G. and M.L.K., unpublished data, 1994).
In light of the aforementioned findings that suggest an essential role for sequences within KIV5-8 in Lp(a) formation, our observation of an Lp(a) particle with KIV9-P apo(a) was somewhat unexpected. However, a similar result was also indirectly obtained by Ernst et al,15 who used a construct in which apo(a) kringle IV types 5-8 had been removed. This result implies that there are potential non–Lys-dependent interactions of the C-terminus (kringle V and protease domains) of apo(a) with apoB-100; this is also supported by our observation that an apo(a) derivative with kringle IV types 9 and 10 only does not form r-Lp(a) particles (data not shown). Observation of r-Lp(a) formation with KIV9-P contradicts the findings of Trieu and McConathy,17 who observed no interaction when they used a construct identical to ours and human LDL. The reason for this apparent discrepancy is unclear but may be due to their use of an immobilized system. However, we have observed binding of KIV9-P to LDL in an immobilized system using purified protein corresponding to this r-apo(a) derivative (data not shown).
In our studies, a significant proportion of r-Lp(a) formed from apo(a) derivatives 17K (data not shown) and 6KΔVP (Fig 6A⇑) did not bind to Lys-Sepharose. This phenomenon may correspond to the Lys-binding heterogeneity that has been previously reported for plasma-derived Lp(a), in which a consistent, donor-dependent fraction of Lp(a) was unable to bind to Lys-Sepharose.30 We have conclusively demonstrated that single apo(a) kringles are differentially N-glycosylated in our mammalian expression system.20 As such, heterogeneity in apo(a) glycosylation may alter its Lys-binding properties in the context of the Lp(a) particle. However, it is interesting to note that uncomplexed 6KΔVP r-apo(a) binds completely to Lys-Sepharose (see Fig 6A⇑). Thus, it is more likely that Lp(a) may not bind to Lys-Sepharose if the particles are complexed with either LDL or other Lp(a) particles, thereby blocking the Lys affinity site in apo(a) kringle IV type 10. Such complexes have been described previously.31 32
We determined that r-Lp(a) formation in our system was most effectively inhibited by N-α-acetyl-l-Lys (IC50 ≈9.5 mmol/L) and equally by ε-ACA and l-Lys (IC50 ≈11 mmol/L). These data are in fairly good agreement with those reported by Frank et al,33 except that in their study, N-α-acetyl-l-Lys was found to be a much more effective inhibitor of Lp(a) formation than either ε-ACA or l-Lys. We determined an IC50 of ≈66 mmol/L for N-ε-acetyl-l-Lys, which is in good agreement with the studies of Frank et al.33 The reduced specificity of this modified amino acid is in keeping with its decreased affinity for the LBS in plasminogen kringle IV.34 The similar efficacy of Lys and ε-ACA as inhibitors of r-Lp(a) formation was somewhat unexpected, given that the affinity of Lys for apo(a) kringle IV type 10 has been reported to be much lower than that of ε-ACA.35 However, there are a number of amino acid substitutions in kringle IV types 5-8 compared with kringle IV type 10 (McLean et al)5 that may yield altered specificities of the LBSs within these kringles. Finally, Pro also was able to effectively inhibit Lp(a) assembly (IC50 ≈20 mmol/L), but the mechanism underlying this observation remains to be determined.
Our data demonstrate that core sequences within apo(a) kringle IV types 7-9 are required for maximal Lp(a) formation. The Lys-dependent component of this interaction involves Lys affinity sites within apo(a) kringle IV types 7-8. The previously defined LBS in apo(a) kringle IV type 10 is likely responsible for the binding of Lp(a) to Lys residues in biological substrates such as fibrin and does not appear to contribute to essential apo(a)–apoB-100 interactions. In contrast to previously published work, the present study suggests that sequences within apo(a) kringle IV types 7-8 act in concert to mediate noncovalent interactions with apoB-100 that are required for Lp(a) assembly. The nature of this interaction will require detailed analysis of the LBS(s) within each of these kringles by site-directed mutagenesis combined with detailed analyses of Lp(a) formation and noncovalent apo(a)/LDL interactions using purified r-apo(a) derivatives. Additionally, the role of apo(a) kringle V and protease domains in Lp(a) assembly needs to be more clearly elucidated. Such studies will be essential to our understanding of the mechanism of Lp(a) formation and are ongoing in our laboratory.
Selected Abbreviations and Acronyms
|ELISA||=||enzyme-linked immunosorbent assay|
|IC50||=||median inhibitory concentration|
|MEM||=||minimal essential medium|
|PAGE||=||polyacrylamide gel electrophoresis|
|PCR||=||polymerase chain reaction|
This study was supported by grants from the Medical Research Council of Canada (to M.L.K.) and the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md (grant HL-30086 to S.M.M.). B.R.G. holds a research traineeship award from the Heart and Stroke Foundation of Canada.
Dahlen GH, Guyton JR, Attar M, Farmer JA, Kautz JA, Gotto AM Jr. Association of levels of lipoprotein Lp(a), plasma lipids, and other lipoproteins with coronary artery disease documented by angiography. Circulation. 1986;74:758-765.
Zenker G, Koltringer P, Bone G, Niederkorn K, Pfeiffer, K, Jurgens, G. Lipoprotein(a) is a strong indicator for cerebrovascular disease. Stroke. 1986;17:942-945.
van der Hoek YY, Wittekoek ME, Beisiegel U, Kastelein JJP, Koschinsky ML. The apolipoprotein(a) kringle IV repeats which differ from the major repeat kringle are present in variably-sized isoforms. Hum Mol Genet. 1993;2:361-366.
Lackner C, Cohen JC, Hobbs HH. Molecular definition of the extreme size polymorphism in apolipoprotein(a). Hum Mol Genet. 1993;2:933-940.
Hervio L, Chapman MJ, Thillet J, Loyah S, Angles-Cano E. Does apolipoprotein(a) heterogeneity influence lipoprotein(a) effects on fibrinolysis? Blood. 1993;82:392-397.
Harpel PC, Gordon BR, Parker TS. Plasmin catalyzes binding of lipoprotein(a) to immobilized fibrinogen and fibrin. Proc Natl Acad Sci U S A. 1989;86:3847-3851.
Loscalzo J, Weinfeld M, Fless GM, Scanu AM. Lipoprotein(a), fibrin binding and plasminogen activation. Arteriosclerosis. 1990;10:240-245.
Rouy D, Koschinsky ML, Fleury V, Chapman J, Anglés-Cano E. The binding of human recombinant apolipoprotein(a) and plasminogen to fibrin surfaces. Biochemistry. 1992;3:6333-6339.
Koschinsky ML, Côté GP, Gabel B, van der Hoek YY. Identification of the cysteine residue in apolipoprotein(a) which mediates extracellular coupling with apolipoproteinB-100. J Biol Chem. 1993;268:19819-19825.
Brunner C, Kraft H-G, Utermann G, Muller H-J. Cys4057 of apolipoprotein(a) is essential for lipoprotein(a) assembly. Proc Natl Acad Sci U S A. 1993;90:11643-11647.
Frank S, Durovic S, Kostner GM. Structural requirements of apo-a for the lipoprotein-a assembly. Biochem J. 1994;304:27-30.
Ernst A, Helmhold M, Brunner C, Petho-Schramm P, Armstrong VW, Muller H-J. Identification of two functionally distinct lysine-binding sites in kringle 37 and in kringles 32-36 of human apolipoprotein(a). J Biol Chem. 1994;270:6227-6234.
Chiesa G, Hobbs HH, Koschinsky ML, Lawn RM, Maika SD, Hammer RE. Reconstitution of lipoprotein(a) by infusion of human low density lipoprotein into transgenic mice expressing human apolipoprotein(a). J Biol Chem. 1992;267:24369-24374.
Trieu VN, McConathy WJ. A two-step model for lipoprotein(a) formation. J Biol Chem. 1995;270:15471-15474.
Sangrar W, Marcovina SM, Koschinsky ML. Expression and characterization of apolipoprotein kringle IV types 1, 2 and 10 in mammalian cells. Protein Eng. 1994;7:723-731.
Graham FL, Smiley J, Russell WC, Nairn R. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J Gen Virol. 1977;36:59-77.
Gorman C, Padmanahban R, Howard B. High efficiency DNA-mediated transformation of primate cells. Science. 1983;221:551-553.
Havel RJ, Eder HA, Bragdon JD. The distribution and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Clin Invest. 1955;34:1345-1353.
Marcovina SM, Albers JJ, Gabel B, Koschinsky ML, Gaur VP. Effect of the number of apolipoprotein(a) kringle 4 domains on immunochemical measurements of lipoprotein(a). Clin Chem. 1995;41:246-255.
Wong WLT, Eaton DL, Berloui A, Fendly B, Hass PE. A monoclonal-antibody-based enzyme-linked immunosorbent assay of lipoprotein(a). Clin Chem. 1990;36:192-197.
Utermann G, Kraft HG, Menzel HJ, Hopferwieser T, Seitz C. Genetics of the quantitative Lp(a) lipoprotein trait: relation of Lp(a) glycoprotein phenotypes to Lp(a) lipoprotein concentrations. Hum Genet. 1993;78:41-46.
Ye SQ, Trieu VN, Stiers DL, McConathy WJ. Interactions of low density lipoprotein2 and other apolipoproteinB-containing lipoproteins with lipoprotein(a). J Biol Chem. 1988;263:6337-6343.
Frank S, Durovic S, Kostner K, Kostner GM. Inhibitors for the in vitro assembly of Lp(a). Arterioscler Thromb Vasc Biol. 1995;15:1774-1780.
LoGrasso PV, Cornell-Kennon S, Boettcher BR. Cloning, expression and characterization of human apolipoprotein(a) kringle IV37. J Biol Chem. 1994;269:21820-21827.