Mutation of lysine residues in apolipoprotein B-100 causes defective lipoprotein[a] formation

Lipoprotein[a] (Lp[a]) is assembled by a two-step process involving an initial lysine-dependent binding between apolipoprotein B-100 (apoB-100) and apolipopro-tein[a] (apo[a]) that facilitates the formation of a disulphide bond between apoB-100Cys4,326 and apo[a]Cys4,057. Previous studies of transgenic mice expressing apoB-95 (4,330 amino acids) and apoB-97 (4,397 amino acids) have shown that apoB-100 amino acids 4,330–4,397 are important for the initial binding to apo[a]. Furthermore, a lysine-rich peptide spanning apoB-100 amino acids 4,372–4,392 has re-cently been shown to bind apo[a] and inhibit Lp[a] assembly in vitro. This suggests that a putative apo[a] binding site exists in the apoB-4,372–4,392 region. The aim of our study was to establish whether the apoB-4,372–4,392 sequence was important for Lp[a] assembly in the context of the full-length apoB-100. Transgenic mice were created that expressed a mutant human apoB-100, apoB-100K 4 → S 4 , in which all four lysine residues in the 4,372–4,392 J. Res. 2004. 45: 63–70.

of similarity (19,20). Mouse apoB is of interest, because it is well documented that mouse apoB can associate noncovalently with apo[a] despite lacking the Cys4,326 residue (10,21,22). Interestingly, there are only five lysine residues within the human apoB-4,330-4,397 sequence. Four of these, Lys4,372, Lys4,379, Lys4,385, and Lys4,392, are clustered within a 21 amino acid region that is ␣ -helical in structure (20). Recently, a synthetic peptide spanning the human apoB-4,372 -4,392 sequence was shown to bind apo[a] and effectively inhibit Lp[a] assembly (20). Taken together, these observations have led us to hypothesize that the apoB-4,372 -4,392 region constitutes a putative apo [a] binding site that is important for Lp[a] assembly. Furthermore, we speculate that one or more of the lysine residues in the apoB-4,372 -4,392 sequence are important for apo[a] binding.
A line of human apoB-100 transgenic mice was generated that expressed a mutant apoB in which all four lysines in the apoB-4,372 -4,392 sequence were mutated to serine residues. The ability of the mutant apoB-100 to form Lp[a] was evaluated both in vitro and in vivo.

Generation of transgenic mice expressing a mutant human apoB-100
Four lysine-to-serine substitutions at human apoB-100 positions 4,372, 4,379, 4,385, and 4,392 were introduced into a 108 kb yeast artificial chromosome (YAC) spanning the apoB-100 gene using a previously reported "pop-in, pop-out" gene-targeting strategy (8). The point mutations were introduced into a yeast-integrating vector containing a 2.8 kb Xba I fragment spanning exon 29 of the apoB-100 gene (8). Site-directed mutagenesis was performed with the ExSite™ mutagenesis kit (Stratagene, La Jolla, CA) using the oligonucleotides K → S3 (5 Ј -CA GC AACCT-GTTAGTTGCgCTTA GC GACTTCCATTCTG-3 Ј ) and K → S4 (5 Ј -ATCAGATCGACTATC GA TTCTTCAAGTTCATAATA GC TCACT-GTCCAGCCAAC-3 Ј ). The nucleotides written in bold-italic signify mutations effecting the lysine-to-serine substitutions. Two new restriction sites (underlined) were generated in association with the mutations; Bpu1102 I in K → S3 and Cla I in K → S4. An additional silent substitution (shown in lower case) was needed to create the Bpu1102 I site. Clones containing the correct mutations were identified by DNA sequencing. The modified yeast integrating vector, apoB-100K 4 → S 4 , was then linearized at a unique Eco RI site in exon 29 and transformed into yeast spheroplasts harboring the 108 kb human apoB-100 YAC. Pop-in and pop-out transformants were selected using the uracil, tryptophan, and lysine yeast markers as previously described (8). High-molecularweight yeast DNA was prepared from selected pop-in and popout yeast clones (23), and the targeting event was analyzed on a 1% agarose pulsed-field gel. To confirm the incorporation of mutations, a 0.9 kb fragment of exon 29 spanning the introduced mutations was amplified by PCR from yeast DNA with the human apoB-specific primers 5SP2 (5 Ј -GGGTTAGAAATACT-GTTTTCC-3 Ј ) and 3SP1 (5 Ј -GTTTTGAATGGACAGGTCAATC-3 Ј ). The PCR products were digested with the Bpu1102 I and Cla I restriction enzymes and analyzed on 1% agarose gels.
To generate transgenic mice expressing the human apoB-100K 4 → S 4 mutant, the apoB-100K 4 → S 4 YAC DNA was purified from pulsed-field gels, adjusted to a concentration of 1 ng/ l, and microinjected into FVB/N fertilized mouse eggs. Transgenic founders were identified by PCR of genomic DNA using the 5SP2 and 3SP1 PCR primers, followed by direct sequencing of the PCR product. Plasma from transgenic founders was tested for the presence of the mutant apoB by Western blotting using the human-apoB-specific monoclonal antibody, 1D1 (24).
Other transgenic mice used in this study were human apo[a] transgenic mice (14)

Characterization of human apoB-100K 4 → S 4 in transgenic mouse plasma
The ability of the apoB-100K 4 → S 4 protein to bind to multiple human apoB-specific monoclonal antibodies was examined by Western blot analysis. For these studies, we used the following antibodies: 1D1, which binds between apoB-100 amino acids 474 and 539 (24); MB47, which binds near amino acid 3,500 (28); Bsol16, which binds between amino acids 4,157 and 4,189 (24); and 605, which binds between amino acids 4,235 and 4,536 (29). Plasma samples from transgenic mice that express human apoB-100 or human apoB-90 were used as controls. To evaluate the distribution of the mutant apoB-100 among the plasma lipoproteins, apoB-100K 4 → S 4 transgenic mouse plasma was size fractionated by fast protein liquid chromatography (FPLC) on a Superose 6 10/30 column (Pharmacia Biotech, Uppsala, Sweden) (27). The lipoprotein fractions were analyzed for cholesterol content by an enzymatic assay (Cholesterol CHOD-PAP, Boehringer Mannheim, Mannheim, Germany), and apoB content was analyzed by Western blot analysis with the 1D1 antibody.

Assessing the ability of the apoB-100K 4 → S 4 protein to form Lp[a]
The ability of apoB-100K 4 → S 4 to form Lp[a] was assessed using a modification of the assay described by McCormick et al. (26). Plasma (10 l) from an apo[a] transgenic mouse was incubated with plasma samples from either human apoB-100 transgenic mice (20 l) or human apoB-100K 4 → S 4 transgenic mice (34 l) in 0.15 M NaCl in a total volume of 400 l at 37 Њ C for 2 h. Similar incubations containing apo[a] only and apo[a] with human apoBCys4,326Gly, a mutant apoB that noncovalently binds apo[a] but cannot make the covalent linkage, were included as negative controls. The amount of Lp[a] formed in each incubation was measured by an Lp[a]-specific sandwich ELISA (30) using MAb a-6 as the capture antibody and horseradish peroxidase (HRP)-labeled 1D1 as the detection antibody (20). The amounts of human apoB in each incubation were measured using the same ELISA procedure, only in this case, the capture antibody was the human apoB-specific antibody, MB47 (28). A human serum control (Precipath ® , Roche Diagnostics Corporation, Indianapolis, IN) was used as the apoB standard. All steps were performed at room temperature unless otherwise stated. A time course assay of Lp[a] formation was also performed on similar incubations in which samples (40 l) were removed from the incubations at 10 min, 30 min, 60 min, 120 min, and 240 min, and were immediately snap frozen and stored at Ϫ 80 Њ C.  (30) using MAb a-6 as the capture antibody and MAb a-1 as the detection antibody. The total amount of human apoB in double transgenic mouse plasma was quantified using the apoB ELISA described above.

Generation of the mutant apoB-100K 4 → S 4 YAC
To test the importance of apoB-100 amino acids 4,372 -4,392 for Lp[a] assembly, we generated a line of transgenic mice expressing a mutant apoB-100, apoB-100K 4 → S 4 , in which serines were substituted for the lysine residues at amino acid positions 4,372, 4,379, 4,385, and 4,392. The base substitutions required to effect the lysine-to-serine mutations were targeted into exon 29 of the apoB-100 gene contained in a 108 kb YAC by "pop-in, pop-out" gene-targeting ( Fig. 1A ). Successful integration of the 7 kb targeting vector in the pop-in step was visualized on a pulsed-field gel as an increase in YAC size (Fig. 1B, PI), compared with the wild-type apoB-100 YAC (Fig. 1B, Wt). Successful incorporation of all mutations in the pop-in step was confirmed by PCR amplification of a 900 bp fragment from YAC DNA and cleavage with the Bpu1102 I and Cla I restriction enzymes (Fig. 1C, PI lanes). Following the pop-out step, the targeted apoB-100 YAC showed a return to its original size (Fig. 1B, PO), and retention of the mutations was documented by complete cleavage of amplified YAC DNA with Bpu1102 I and Cla I (Fig. 1C, PO lane).

Characterization of the apoB-100K 4 → S 4 transgenic mice
The mutant YAC was purified from pulsed-field gels and microinjected into mouse zygotes to generate transgenic mice. Analysis of the genomic DNA from 27 founders by PCR and Bpu1102I and ClaI restriction digestion identified two animals harboring the mutant human apoB transgene. Both founder animals expressed the full-length mutant human apoB-100 as detected by Western blots of mouse plasma using the anti-human-apoB-specific monoclonal antibody, 1D1. To verify that the introduced mutations had not had any major effect on the conformation of the mutant apoB, four different human apoB-specific monoclonal antibodies were tested for their ability to bind the mutant apoB ( Fig. 2A). The apoB-100K 4 →S 4 mutant bound all antibodies tested, including the 605 antibody that binds an epitope in the same region as the mutated site. The distribution of the mutant apoB among the plasma lipoproteins was assessed by FPLC fractionation of mouse plasma and Western blot analysis. The distribution of apoB-100K 4 →S 4 was almost identical to that of wild-type apoB-100, with both proteins residing mainly in the LDL fractions (Fig. 2B).

Lp[a] formation with the mutant apoB-100K 4 →S 4 in vitro
To assess the ability of the apoB-100K 4   forming Lp[a], but the amount of Lp[a] formed in a 2 h incubation was less than that formed in a control incubation containing wild-type human apoB-100 (Fig. 3A). Both incubations contained similar amounts of human apoB, as measured by a human apoB-specific ELISA (Fig. 3B) (Fig. 4).

Lp[a] formation with the mutant apoB-100K 4 →S 4 in vivo
To test the ability of the apoB-100K 4 →S 4 protein to form Lp  Western blot analysis of the apoB-100K 4 →S 4 mutant in the plasma of transgenic mice. A: The proteins in plasma samples from mice expressing human apoB-100, apoB-90 (4,084 amino acids), and apoB-100K 4 →S 4 were separated on SDS/4%-polyacrylamide gels under reducing conditions. Separated proteins were detected by Western blot analysis with four different human apoB-specific monoclonal antibodies, 1D1, MB47, Bsol16, and 605. The epitope for each antibody is given above each blot. B: Distribution of apoB-100K 4 →S 4 among the plasma lipoproteins. Plasma samples (200 l) from transgenic mice expressing human apoB-100 or the apoB-100K 4 →S 4 mutant were size fractionated on a Superose 6 10/30 column, and the cholesterol content of each fraction measured to identify the major lipoprotein peaks. The apoB content of each fraction was assessed by Western blot analysis with the 1D1 antibody.   Fig. 5 and appear to be similar between animals.

DISCUSSION
The current study sought to define the structural features of apoB that facilitate its association with apo[a] to form Lp [a]. Previously, we demonstrated that apoB-100Cys4,326 was the site of covalent attachment to apo[a] (8). It was obvious, however, even before these studies, that Lp[a] assembly relied on more than just two free cysteines. A widely accepted two-step model of Lp[a] assembly depicts a noncovalent association between apoB and apo[a] preceding the formation of the disulphide bond (5,6). The exact amino acids required for the noncovalent association between apo[a] and apoB are not well characterized, although a growing body of evidence suggests that apoB lysine residues may be important (11)(12)(13)(14)(15)(16).
Our original hypothesis was that the apoB sequences involved in the noncovalent binding to apo[a] would be in close proximity to the Cys4,326 residue. Previous studies of two truncated apoB proteins suggested the presence of a putative apo[a] binding site in the region between apoB amino acids 4,330 and 4,397 (19). An alignment of the human apoB-4,330-4,397 region against pig, mouse, and rat apoB is shown in Fig. 6A. These other apoB species are of interest inasmuch as it has been shown that pig, mouse, and rat apoB still bind apo[a] despite lacking the Cys4,326 residue required for the disulphide link with apo[a] (10,31). The alignment shows clusters of amino acids within the 4,330-4,397 region that are highly conserved. Also of note are four well-conserved lysine residues at apoB amino acids 4,372, 4,379, 4,385, and 4,392. All four lysine residues are contained within a sequence (amino acids 4,365-4,396) that is predicted to form a Class A amphipathic helix (10, 32) with lipid binding properties. Indeed, circular dichroism studies of a syn-  show the sequence to be of ␣-helical structure and capable of interacting with lipid (20). Furthermore, the same study showed that the apoB 4,372-4,392 peptide was able to bind apo[a] and effectively inhibit Lp[a] formation (20). This led us to hypothesize that one or more of the four lysines in the 4,372-4,392 sequence were important for the interaction with apo[a].
To test the physiological importance of the apoB lysine residues at positions 4,372, 4,379, 4,385, and 4,392 for the interaction with apo[a], we generated a line of transgenic mice expressing a mutant apoB in which all four lysines were substituted for serines. Serine was selected, because its polar side chain would not be expected to perturb the local ␣-helical structure. Evaluation of the apoB-100K 4 →S 4 protein by Western blot analysis with a panel of monoclonal antibodies showed no major conformational changes compared with wild-type apoB-100. In vitro experiments showed that the mutant apoB-100 was impaired in its ability to form Lp[a] compared with wild-type human apoB-100. The apoB-100K 4 →S 4 protein showed a slower rate of Lp[a] formation than did the wild-type apoB-100. These results were reiterated in vivo in double transgenic mice expressing both the apoB-100K 4 Figure 6B shows the predicted structure of the ␣-helix spanning the apoB amino acids 4,365-4,396 sequence. Fig. 6. Alignment of apoB amino acids 4,326 to 4,397 from human, pig, mouse, and rat and structural analysis of a predicted ␣-helix in the region. A: Partial sequences spanning apoB-100 amino acids 4,326 to 4,395 from human, pig, and rat were obtained from the NCBI database. The mouse apoB sequence was obtained from Dr. Stephen Young. Sequences were aligned in FASTA format to determine sequence homology. Identical amino acids are indicated by dashed lines. A sequence (amino acids 4,365 to 4,396) predicted by the LOCATE program (32) to form a class A ␣-helix is highlighted in gray. B: A helical net diagram of the amphipathic ␣-helix formed by residues 4,365 to 4,396 centered on the polar face as generated by the HELNET/SNORKEL program (37). The dashed lines represent the position of the predicted water-lipid interface. The residues that are predicted to be buried in the lipid phase are given in boldface circles, while the residues in nonboldface circles in the center of the helix are predicted to project into the aqueous phase. The four lysine residues within the helix are highlighted in gray.

Downloaded from
The four lysine residues, Lys4,372, Lys4,379, Lys4,385, and Lys4,392, contained within the helix are located along the lipid-water interface. We hypothesize that one of the lysine side chains interacts with a kringle IV lysine binding pocket [possibly kringle IV 6 (33) (34) bound to lysine analogs show an anionic center that interacts with the amide group of the lysine side chain. Adjacent to the anionic center is a hydrophobic trough consisting of two tyrosine and two tryptophan residues that interact with the hydrocarbon backbone of the free lysine residue. In the physiological setting, where the lysine backbone is constrained as part of a polypeptide chain, it is possible that the hydrophobic trough in apo[a] is available for a hydrophobic interaction with apoB. The interaction of apo[a] kringle IV structures with hydrophobic residues such as phenylalanine, praline, and leucine has been predicted by computer modelling of these amino acids in the ligand binding domain of kringle IV 9 (36). Interestingly, there are several highly conserved hydrophobic residues within the apoB-4,372-4,392 sequence (Fig. 6A), and mutation of hydrophobic residues in the apoB 4,372-4,392 peptide does disrupt its ability to bind apo[a] (R. Sharp and S. McCormick, unpublished observations). It is possible that a hydrophobic interaction between the apoB-100K 4 →S 4 mutant and apo[a], combined with a weak interaction of the substituted serine with the anionic center, still allows for some Lp[a] formation.
An alternative explanation for the Lp[a] formation seen with the apoB-100K 4 →S 4 protein is that other apoB sequences remote from the 4,372-4,392 sequence are able to facilitate the binding of apoB to apo[a] to allow Lp[a] formation, albiet more slowly. A recent study by Becker et al. (18) of a truncated apoB, apoB18, identified a 25 amino acid sequence in the amino terminus of apoB (amino acids 680-704) that interacts with apo[a] in vitro. The apoBLys680 was found to be important for the binding of apoB18 to apo[a], suggesting the presence of an N-terminal site that may also be necessary for the proper anchoring of apoB to apo[a] to facilitate Lp[a] formation.
Our study is the first to document defective Lp[a] assembly in vivo after the introduction of point mutations into a full-length apoB-100 molecule. We have demonstrated that mutation of amino acids in the apoB-4,372-4,392 sequence significantly impairs Lp[a] formation. Lysine residues in the sequence are important for the in-teraction, although it is likely that other amino acids in this sequence play a role in binding to apo [a]. Binding of the apoB-4,372-4,392 sequence onto apo[a] kringle IV structures, combined with further mutagenesis studies of the 4,372-4,392 sequence, will be required to establish the exact amino acids in the 4,372-4,392 sequence involved in the binding to apo [a]. A more complete understanding of the protein sequences involved in Lp[a] assembly may ultimately lead to the design of novel inhibitors to prevent Lp[a] formation and hence provide a potential new class of Lp[a]-lowering agents.