Effect of Arginine 172 on the Binding of Apolipoprotein E to the Low Density Lipoprotein Receptor*

The region of apolipoprotein E (apoE) that interacts directly with the low density lipoprotein (LDL) receptor lies in the vicinity of residues 136–150, where lysine and arginine residues are crucial for full binding activity. However, defective binding of carboxyl-terminal truncations of apoE3 has suggested that residues in the vicinity of 170–183 are also important. To characterize and define the role of this region in LDL receptor binding, we created either mutants of apoE in which this region was deleted or in which arginine residues within this region were sequentially changed to alanine. Deletion of residues 167–185 reduced binding activity (15% of apoE3), and elimination of arginines at positions 167, 172, 178, and 180 revealed that only position 172 affected binding activity (2% of apoE3). Substitution of lysine for Arg172 reduced binding activity to 6%, indicating a specific requirement for arginine at this position. The higher binding activity of the Δ167–185 mutant relative to the Arg172mutant (15% versus 2%) is explained by the fact that arginine residues at positions 189 and 191 are shifted in the deletion mutant into positions equivalent to 170 and 172 in the intact protein. Mutation of these residues and modeling the region around these residues suggested that the influence of Arg172 on receptor binding activity may be determined by its orientation at a lipid surface. Thus, the association of apoE with phospholipids allows Arg172 to interact directly with the LDL receptor or with other residues in apoE to promote its receptor-active conformation.

The region of apolipoprotein E (apoE) that interacts directly with the low density lipoprotein (LDL) receptor lies in the vicinity of residues 136 -150, where lysine and arginine residues are crucial for full binding activity. However, defective binding of carboxyl-terminal truncations of apoE3 has suggested that residues in the vicinity of 170 -183 are also important. To characterize and define the role of this region in LDL receptor binding, we created either mutants of apoE in which this region was deleted or in which arginine residues within this region were sequentially changed to alanine. Deletion of residues 167-185 reduced binding activity (15% of apoE3), and elimination of arginines at positions 167, 172, 178, and 180 revealed that only position 172 affected binding activity (2% of apoE3). Substitution of lysine for Arg 172 reduced binding activity to 6%, indicating a specific requirement for arginine at this position. The higher binding activity of the ⌬167-185 mutant relative to the Arg 172 mutant (15% versus 2%) is explained by the fact that arginine residues at positions 189 and 191 are shifted in the deletion mutant into positions equivalent to 170 and 172 in the intact protein. Mutation of these residues and modeling the region around these residues suggested that the influence of Arg 172 on receptor binding activity may be determined by its orientation at a lipid surface. Thus, the association of apoE with phospholipids allows Arg 172 to interact directly with the LDL receptor or with other residues in apoE to promote its receptor-active conformation.
There are two major structural and functional domains in apoE. (15,16). The 22-kDa amino-terminal domain, shown by x-ray crystallographic studies to be a four-helix bundle (17), contains the receptor-binding region (2, 18 -20). The 10-kDa carboxyl-terminal domain contains important lipoprotein-binding regions, although the amino-terminal domain is capable of binding phospholipids and remodeling them into discoidal lipoprotein particles (15,(21)(22)(23)(24). The two domains are joined by a hinge region (approximately residues 165-215) of unknown function (15,16). To date, this hinge region has been considered simply a linker between the two domains of apoE, and significant portions of it can be deleted (residues 186 -223) without affecting the binding of apoE to lipoproteins (25).
The original characterization of the receptor binding activity of apoE showed that arginine and lysine residues were critical (26,27). The LDL receptor-binding region was narrowed down to residues in the vicinity of 136 -150 by examining naturally occurring and genetically engineered point mutations (2,18) and proteolytic fragments of apoE (19) and by using monoclonal antibodies to inhibit apoE binding to the LDL receptor (20). However, residues outside the receptor-binding region can also affect receptor binding activity. ApoE2 (Cys 158 ) has only 1% of the LDL receptor binding activity of apoE3 (Arg 158 ) and apoE4 (Arg 158 ) (28). Studies involving chemical modification of cysteine groups (29), mutagenesis, and x-ray crystallography of the apoE2 22-kDa fragment (30,31) demonstrated that the presence of a cysteine at position 158 in apoE2, instead of an arginine, indirectly affected receptor binding activity by rearranging crucial salt bridges, reducing the positive electrostatic potential of the receptor-binding region.
Studies of carboxyl-terminal truncations of the 22-kDa fragment of apoE3 showed that residues 171-183 are also important for receptor binding activity (32). It was hypothesized that arginines in this region may interact directly with the LDL receptor or, like Arg 158 , indirectly affect the receptor-binding region of apoE. Since the electron density of residues 163-191 is uninterpretable in the x-ray structure of the 22-kDa fragment of apoE (17,23,31), we have taken a mutagenesis approach to determine which residues in this region are important in receptor binding activity. An internal deletion in apoE3 of residues 167-185 (apoE3⌬167-185) confirmed that this region had an effect on receptor binding activity. From the examination of point mutations in apoE3, we found that a single amino acid, Arg 172 , was responsible for this effect. Finally, sequence comparison of apoE3⌬167-185 with intact apoE3 implicated the importance of the presence of Arg 172 and its possible positioning on a lipid surface for LDL receptor binding activity.

Construction of ApoE Variants-
The cDNAs of human apoE3 and apoE4 were inserted into a modified thioredoxin (Trx) fusion expression vector (pET32a; Novagen) as described previously (33). Internal deletions and point mutations of apoE were made by overlapping polymerase chain reaction (PCR) (23) with Pfu polymerase (Stratagene) and appropriate oligonucleotides (Oligos Etc.) ( Table I). The DNA was then digested with the appropriate restriction enzymes, subjected to agarose gel electrophoresis, and purified with a gel extraction kit. The purified inserts were ligated into the expression plasmid using T4 Ready-To-Go ligase (Amersham Pharmacia Biotech). The sequence of the DNA insert was verified by double-stranded DNA sequencing.
Expression and Purification of ApoE and Variants-Plasmids containing cDNA of human apoE or variants were transformed into Escherichia coli (strain BL21 (DE3); Novagen). Several transformants were grown in LB medium at 37°C to an optical density of 0.5-0.6 (550 nm). Trx-apoE expression was induced by adding isopropyl-1-thio-␤-D-galactopyranoside (final concentration, 100 g/ml) to the culture medium, after which the culture was incubated for 60 min at 37°C. The transformants were screened for high expression of the Trx-apoE fusion protein by SDS-polyacrylamide gel electrophoresis followed by immunoblotting.
ApoE and variants were purified as described previously (33). Briefly, the transformed E. coli were cultured in LB medium at 37°C, and Trx-apoE expression was induced with isopropyl-1-thio-␤-D-galactopyranoside (final concentration, 100 g/ml) for 2 h. After the bacterial pellet was sonicated and the lysate was centrifuged to remove debris, the Trx fusion protein, which contains a His tag, was purified on a nickel affinity column. The fusion protein was then cleaved with thrombin to remove Trx from apoE. Before cleavage, the fusion protein was complexed with dimyristoylphosphatidylcholine (DMPC) to protect the protease-susceptible region of apoE. The thrombin was inactivated with ␤-mercaptoethanol, the mixture was lyophilized and delipidated, and the protein was solubilized in 6 M guanidine HCl (0.1 M Tris, pH 7.4, 0.01% EDTA, and 1% ␤-mercaptoethanol).
Protein was isolated by gel permeation chromatography on a Sephacryl S-300 column in 4 M guanidine HCl (0.1 M Tris, pH 7.4, 0.01% EDTA, and 0.1% ␤-mercaptoethanol). The fractions containing apoE or variants were pooled, dialyzed against 5 mM NH 4 HCO 3 , and lyophilized. The protein was solubilized in 100 mM NH 4 HCO 3 and stored at Ϫ20°C. All apoE variants were structurally similar to apoE3 as determined by circular dichroism measurements (data not shown).
Preparation of ApoE⅐DMPC Complexes and Determination of Receptor Binding Activity-Purified recombinant human apoE and variants were complexed with DMPC and isolated by KBr gradient cetrifugation as described (34). The apoE⅐DMPC discs were negatively stained on the surface of carbon fluid grids. Electron micrographs were made at a magnification of ϫ 200,000 and imported with a video camera into an Image 1/AT image-analysis system. Particle size was analyzed by au-tomated sizing and counting programs available on system software (mL version 4.03a; Universal Imaging Corp.). Multiple areas on a single grid were sampled. The complexes formed by the apoE variants and DMPC were the same size as normal apoE3 (disc diameter ranged from 15 to 17 nm). The receptor binding activities of the apoE⅐DMPC complexes were determined in a competition assay with 125 I-labeled human LDL and human primary fibroblasts in culture (34).

RESULTS AND DISCUSSION
When complexed with DMPC, the 22-kDa fragment of apoE3 binds to the LDL receptor with high affinity (19). LDL receptorbinding studies of carboxyl-terminal truncations of the apoE3 22-kDa fragment showed that a mutant consisting of residues 1-183 of apoE3, apoE3-(1-183), had essentially normal binding activity, but further truncation to residue 170, apoE3-(1-170), reduced binding activity to less than 1% of that of full-length apoE3 (32). Since arginine residues are important in binding activity (2,26), it was hypothesized that one or more arginines in this region interact directly with the receptor or act to stabilize the receptor-binding region (32). To confirm that the region within residues 171-183 was indeed significant for apoE receptor binding activity and that the previous findings were not influenced by the absence of the major lipid-binding region in apoE (the 10-kDa carboxyl-terminal fragment), we constructed internal deletions of apoE, rather than truncations. The first internal deletion (apoE3⌬167-185) was designed to eliminate four arginine residues (167, 172, 178, and 180) in this region. There are no lysines in the deleted region. The second variant (apoE4⌬186 -223), in which residues adjacent to 167-185 were deleted, was constructed previously for studies of lipoprotein binding (25).
The binding of apoE4⌬186 -223 to the LDL receptor was essentially identical to that of apoE3 and apoE4. (In this assay, apoE3 and apoE4 have identical binding to the LDL receptor (28).) However, apoE3⌬167-185 bound to the LDL receptor with only 15% of the binding activity of apoE3 (Fig. 1, Table II). These results are consistent with the data from the carboxylterminal truncation experiments, which suggested that part of the hinge region affects receptor binding activity.
To determine if the arginines in this region are important for receptor binding activity, we created four mutants in which an arginine was replaced with an alanine (apoE3 R167A, apoE3 R172A, apoE3 R178A, and apoE3 R180A). Of these four mu-TABLE I List of primers used to produce mutants of apoE R at the end of an internal primer name indicates it is the reverse primer of the one listed above it. In the primary PCRs, the 5Ј-flanking primer is used with the reverse internal primer, and the 3Ј-flanking primer is used with the forward internal primer. The two resulting PCR products have overlapping ends and are thus primer and template in the secondary PCR. (The 3Ј and 5Ј flanking primers are also used in the secondary PCR.) ACGCCACAGTGGCGGCTGCCACTGCGCCCTGTTCG tants, only apoE3 R172A had defective receptor binding activity (2% of apoE3) (Fig. 2, Table II). A mutant with an arginine to lysine change at position 172 (apoE3 R172K) had just 6% of the binding activity of apoE3, indicating the importance and specificity of an arginine at position 172 (Fig. 3, Table II). Thus, the effect of this region on receptor binding activity could be narrowed down to a single amino acid, Arg 172 . However, there were some paradoxes in the results. Specifically, apoE3⌬167-185 had consistently higher binding activity than apoE3 R172A (15 versus 2%, p ϭ 0.018). As a result, we decided to compare this section of the hinge region, residues 167-185, in apoE3 and apoE3⌬167-185. In apoE3⌬167-185, residues 186 -204 are essentially shifted down to replace the deleted residues 167-185; therefore, residue 186 can be referred to as 167Ј in the apoE3⌬167-185 variant and residue 204 as 186Ј. The other residues between 186 and 204 may be similarly renamed. When residues 186 -204 (167Ј-185Ј) were aligned with 167-185 (Fig. 4), it was revealed that an arginine at position 172 in apoE3 was replaced with an arginine in apoE3⌬167-185. (There are two arginines within the sequence from 186 -204, at positions 189 (170Ј) and 191 (172Ј).) Furthermore, at position 183, a proline, which disrupts ␣-helical structure, was replaced with another proline, residue 202 (183Ј). Instead of clarifying the difference in receptor binding activity between the two mutants, this analysis produced more questions. If Arg 172 is important, why was the receptor binding activity of apoE3⌬167-185 defective?
To explain this contradiction, we examined the predicted secondary structure of residues 167-204. Segrest et al. (35) predicted that residues 167-182 would form a class A am- FIG. 1. Ability of apoE3⅐DMPC (q), apoE4⌬186 -223⅐DMPC (f), and apoE3⌬167-185⅐DMPC (ࡗ) to compete with human 125 I-LDL for binding to LDL receptors on normal human fibroblasts. Cells incubated in medium containing 10% human lipoprotein-deficient serum received the same medium with 2 g/ml of 125 I-LDL and the indicated concentrations of apoE-DMPC complexes. After a 2-h incubation on ice, the cells were extensively washed, and the amount of 125 I-LDL bound to cells was determined. Data presented are from one of at least three experiments.   4. Comparison of residues 167-185 with 186 -204 of apoE3. In apoE3⌬167-185, amino acid residues 186 -299 are essentially shifted down so that residue 186 is now in position 167. Thus, in apoE3⌬167-185, residue 186 can be renamed as residue 167Ј, and residue 204 is 185Ј. The arginines and proline present in this region of apoE3 and apoE3⌬167-185 are highlighted.
phipathic ␣-helix, and algorithms by Garnier et al. (36) and Gibrat et al. (37) also predict that these residues are helical. However, Rall et al. suggested that residues 165-200 contain mostly random structure and ␤ turns (1), and Nolte and Atkinson predicted only a short helical segment from residues 178 -183 (38). Only the algorithm by Garnier et al. (36) suggested that residues 186 -195 form a helical structure.
With the possibility that segments within residues 167-204 form ␣-helices, we modeled residues 167-182 and 186 -201 (167Ј-182Ј) as ␣-helices using the computer program Insight II (Fig. 5). Comparison of the helices suggested an explanation for the paradoxical receptor binding of apoE3⌬167-185, specifically that positioning of Arg 172 on a lipid surface, with respect to the LDL receptor, could be critical. The ␣-helix modeled from residues 167-182 is amphipathic, with its opposing polar and nonpolar faces oriented along the long axis of the helix (35). Arginines 167, 178, and 180 are clustered at the polar-nonpolar interface, typical of a class A amphipathic ␣-helix (35). It has been hypothesized that the arginine residues at the polarnonpolar interface interact with the phospholipid head groups at the lipoprotein surface, while the hydrophobic face inserts into the phospholipid layer (35). This model of association is consistent with recent studies examining the interaction of class A amphipathic peptides with phospholipid bilayers (39) and thus suggests that arginines 167, 178 and 180 would interact with the phospholipid head groups on the phospholipid disc. However, Arg 172 is positioned in the center of the polar face of the helix, where it may be available to interact with other residues in apoE or other proteins (e.g. the LDL receptor) rather than phospholipid head groups. When residues 186 -201 (167Ј-182Ј) are modeled as an ␣-helix, Arg 189 (170Ј) is on the opposite side of the helix from Arg 191 (172Ј), and one could predict that the position of Arg 191 (172Ј) would shift by approximately 90°relative to Arg 172 in apoE3, so that Arg 189 (170Ј) and Arg 191 (172Ј) would lie at the phospholipid interface (Fig.  5). Thus, the potential interactions of Arg 191 (172Ј) with other amino acid residues would be altered compared with Arg 172 in apoE3. Additionally, this stretch of residues is not very amphipathic. While one side of the putative helix contains polar and charged residues, and Arg 189 (170Ј) and Arg 191 (172Ј) may define the polar-nonpolar interface, the nonpolar face is not very hydrophobic, containing only one valine and a few alanine residues; therefore, its potential orientation on or association with phospholipid is questionable. Alternatively, it is possible that residues 186 -201 are not helical and that the receptorbinding results may be explained by a disruption of secondary structure in this region.
The results of these mutagenesis studies show that Arg 172 is important for LDL receptor binding activity; they also imply that its proper orientation, relative to the receptor-binding region, is critical. The structure of the receptor-active conformation of apoE is not known, but it does require association FIG. 5. Residues 167-182 and 186 -201 in apoE3 modeled as ␣-helices. A, residues 167-182 fit a model of a class A amphipathic helix, with leucine, isoleucine, and valine residues on the hydrophobic face of the helix and negatively charged glutamic acids on the hydrophilic face. As is typical for class A amphipathic helices, positively charged residues, in this case arginine, are located at the polar-nonpolar interface. However, Arg 172 is located in the center of the hydrophilic face of the helix. The model of interaction of a class A amphipathic helix with phospholipid suggests that the hydrophobic face inserts into the lipid layer while the hydrophilic face is exposed to the aqueous environment, and the positively charged residues at the polar-nonpolar interface interact with the phospholipid head groups. B, the helix on the left is oriented to align Arg 191 (172Ј) with Arg 172 in A. Arg 189 (170Ј) is on the opposite side of the helix from Arg 191 (172Ј). Thus, if the face of the helix that inserts into the phospholipid layer contains only hydrophobic residues, then the position of Arg 191 (172Ј) in apoE3⌬167-185 would rotate by approximately 90°(helix on the right), altering its potential interactions compared with Arg 172 in apoE3. The hydrophobic and hydrophilic faces are indicated. The helices were generated using Insight II (Molecular Simulations, San Diego). Residues are as follows: yellow, arginine; red, glutamic acid; brown, valine, leucine, and isoleucine; blue, glutamine; gray, glycine, serine, alanine, and threonine. with phospholipids (40). One hypothesis is that the four-helix bundle of the apoE 22-kDa fragment opens up when associating with phospholipid, promoting proper conformation of the receptor-binding region in apoE (4). These current studies also suggest the possibility that phospholipids promote the correct positioning of Arg 172 so that it can interact directly with the ligand-binding region of the LDL receptor or with other residues of apoE to maintain an active conformation of the apoE receptor-binding region. However, precisely how Arg 172 affects receptor binding activity cannot be established until the structure of apoE3 on phospholipid has been determined.