Interaction of an Exchangeable Apolipoprotein with Phospholipid Vesicles and Lipoprotein Particles

Apolipophorin III (apoLp-III) from Locusta migratoria is an exchangeable apolipoprotein that binds reversibly to lipid surfaces. In the lipid-free state this 164-residue protein exists as a bundle of five elongated amphipathic α-helices. Upon lipid binding, apoLp-III undergoes a significant conformational change, resulting in exposure of its hydrophobic interior to the lipid environment. On the basis of x-ray crystallographic data (Breiter, D. R., Kanost, M. R., Benning, M. M., Wesenberg, G., Law, J. H., Wells, M. A., Rayment, I., and Holden, H. M. (1991) Biochemistry 30, 603–608), it was proposed that hydrophobic residues, present in loops that connect helices 1 and 2 (Leu-32 and Leu-34) and helices 3 and 4 (Leu-95), may function in initiation of lipid binding. To examine this hypothesis, mutant apoLp-IIIs were designed wherein the three Leu residues were replaced by Arg, individually or together. Circular dichroism spectroscopy and temperature and guanidine hydrochloride denaturation studies showed that the mutations did not cause major changes in secondary structure content or stability. In lipid binding assays, addition of apoLp-III to phospholipid vesicles caused a rapid clearance of vesicle turbidity due to transformation to discoidal complexes. L34R and L32R/L34R/L95R apoLp-IIIs displayed a much stronger interaction with lipid vesicles than wild-type apoLp-III. Furthermore, it was demonstrated that the mutant apoLp-IIIs retained their ability to bind to lipoprotein particles. However, in lipoprotein competition binding assays, the mutants displayed an impaired ability to initiate a binding interaction when compared with wild-type apoLp-III. The data indicate that the loops connecting helices 1 and 2 and helices 3 and 4 are critical regions in the protein, contributing to recognition of hydrophobic defects on lipoprotein surfaces by apoLp-III.

Exchangeable apolipoproteins are important plasma proteins that bind reversibly to lipoproteins, a process triggered by changes in lipoprotein particle composition. Binding of exchangeable apolipoproteins serves to maintain the structural integrity of lipoproteins as they undergo modification of their lipid content. Apolipoproteins also carry out other functions as follows: apolipoprotein (apo) 1 E serves as a ligand for cell surface lipoprotein receptors (1), apoC-II activates lipoprotein lipase, and apoA-I modulates lecithin-cholesterol acyltransferase activity (2). Exchangeable apolipoproteins are characterized by an abundance of amphipathic ␣-helices, which confer lipid binding properties to these proteins (3).
Apolipophorin III (apoLp-III) is an invertebrate exchangeable apolipoprotein that binds reversibly to neutral lipid-enriched lipoprotein particles (4 -6). The availability of (i) high resolution structural information (7), (ii) biophysical data (8,9), and (iii) a bacterial system for expression of recombinant apoLp-III (10) establishes this as an excellent model system to study structure-function relationships of amphipathic exchangeable apolipoproteins. X-ray crystallography data showed that the lipid-free form of apoLp-III from Locusta migratoria is composed of a bundle of five amphipathic antiparallel ␣-helices (7). For the most part, hydrophobic amino acid residues are buried inside the protein, stabilized by helix-helix interactions (9,11). This helix-bundle organization resembles that of the 22-kDa N-terminal domain of human apoE (12) as well as that of Manduca sexta apoLp-III (13).
Exchangeable apolipoproteins are predicted to undergo a conformational change during the transition from a lipid-free to a lipid-bound form (7,14,15). For apoLp-III, it has been hypothesized that the protein opens about hinged loops and spreads out onto hydrophobic surfaces (7). This conformational adaptation allows contact of hydrophobic faces of amphipathic ␣-helices with the lipid surface. Based on structural data, it has been proposed that apoLp-III recognizes potential lipid surfacebinding sites via one of its ends (16) or the loops connecting the helices (7). The loops connecting helices 1 and 2 (loop A) and helices 3 and 4 (loop C) possess hydrophobic character (see Fig.  1). Sequence alignment of apoLp-IIIs from different species indicates that the Leu residues in these loops are conserved (17). These data have led to the hypothesis that they function as a "hydrophobic sensor" (18). We have used site-directed mutagenesis to test this hypothesis by substituting hydrophobic residues in the putative "sensor" loops of L. migratoria apoLp-III. Characterization of lipid and lipoprotein binding properties of wild-type and mutant apoLp-IIIs provide evidence that the Leu residues in loops A and C function in initiation of lipoprotein binding.

MATERIALS AND METHODS
Site-directed Mutagenesis-Site-directed mutagenesis was performed using the pALTER system (Promega, Madison, WI). The cDNA for apoLp-III present in pET22b(ϩ) (apoLp-III/pET construct (10)) was digested with EcoRI and XbaI, isolated by extraction from 0.8% agarose gels, and subcloned into the pALTER-1 vector (which is ampicillinsensitive and tetracycline-resistant). The resulting apoLp-III/pALTER construct was used as template DNA for site-directed mutagenesis reactions. Primers, designed to substitute Arg for Leu-32, Leu-34, or Leu-95, were added to alkali-denatured template DNA together with ampicillin repair and tetracycline knockout primers. All primers were phosphorylated. The annealing reaction was carried out at 75°C for 5 min and cooled slowly to room temperature (ϳ45 min). T4 DNA polymerase and ligase were added, and the incubation was continued for 90 min at 37°C. High efficiency ES1301 mutS Escherichia coli cells were transformed with 2.5 l of the reaction mix and grown overnight in LB medium with 125 g/ml ampicillin. This culture was used for a plasmid miniprep, and the extracted DNA was used to transform JM109 cells. DNA was extracted from transformed JM109 cells and subsequently sequenced using the dideoxynucleotide chain termination method (19) to verify the desired mutations. DNA from a positive clone was digested with EcoRI and XbaI, and the apoLp-III insert was subcloned back into the original pET22b(ϩ) vector. Finally, the mutated apoLp-III/pET constructs were used to transform BL21 (DE3) E. coli cells.
Expression and Purification of ApoLp-III-ApoLp-III was expressed in E. coli BL21 cells harboring the apoLp-III/pET vector as described (10) and purified by reversed-phase high pressure liquid chromatography. The purity of the apoLp-III preparation was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and analytical reversed-phase high pressure liquid chromatography. 3 H-ApoLp-III was obtained by growing E. coli harboring the apoLp-III/pET construct in 50 ml of M9 minimal medium at 37°C. When the absorbance at 600 nm reached 0.6, apoLp-III expression was induced with isopropyl-␤-D-thiogalactopyranoside (2 mM final concentration). The cells were further grown at 30°C for 30 min after which 150 Ci of a tritiated amino acid mixture was added (Amersham Pharmacia Biotech, Oakville, Ontario, Canada). The cells were grown for another 4.5 h and pelleted by centrifugation. The cell-free supernatant, which is highly enriched in recombinant apoLp-III (10), was dialyzed against 100 mM sodium phosphate buffer, pH 7.5, and concentrated 10-fold by ultrafiltration. A typical 50-ml culture yielded 4 mg of protein with a specific activity of 6 ϫ 10 6 dpm/mg. CD Spectroscopy-A Jasco J-720 spectropolarimeter, interfaced to an Epson Equity 386/25 computer controlled by Jasco software, was used to analyze the ␣-helical content of mutant and wild-type (WT) apoLp-IIIs and to monitor temperature-or guanidine hydrochloride-induced denaturation (8). A modified Contin program by Provencher and Glöckner (20), which contains poly-L-glutamate as a helical reference standard, was used to estimate the ␣-helical content of the proteins.
Vesicle Clearance Assay-Binding of apoLp-III to dimyristoylphosphatidylcholine (DMPC) or dimyristoylphosphatidylglycerol (DMPG) vesicles was monitored by 90°light scattering. ApoLp-III has the ability to transform phospholipid vesicles to protein-lipid disc complexes. This process, which results in a significant reduction in size of the particles, can be monitored spectroscopically. To prepare multilamellar vesicles of DMPC or DMPG (Avanti Polar Lipids Inc., Alabaster, AL), the phospholipid was dissolved in chloroform:methanol (3:1, v/v) and dried under a stream of N 2 . Further dryness was achieved under vacuum for at least 4 h. The dried lipid sample was dispersed in pre-warmed buffer (10 mM Tris-HCl, pH 7.2; 150 mM NaCl; 0.5 mM EDTA) to a final lipid concentration of 10 mg/ml and vortexed for 1 min. From this solution, small unilamellar vesicles (ϳ200 nm in diameter) were prepared by extrusion using 200-nm filters (21). A Perkin-Elmer spectrofluorometer (model LS 50B) was used to monitor phospholipid vesicle clearance as a result of association with exchangeable apolipoproteins (22). Excitation and emission wavelengths were set at 600 nm with a slit width of 3 nm. The temperature of the cuvette holder was maintained at 23.9°C (DMPC) or 23°C (DMPG), and all solutions were preincubated at these temperatures. Lipid vesicles (250 g) were added to 1 ml of buffer and equilibrated for 10 min in the cuvette holder. Specified apoLp-IIIs (250 g of protein for DMPC assays and 10 g for DMPG assays) were then added, mixed for 10 s, and the change in light scattering monitored as a function of time.
Lipoprotein Binding Assay-Apolipoprotein binding to spherical lipoproteins was investigated using the assay system described by Liu et al. (23). Phospholipase C (PL-C)-mediated hydrolysis of the phosphatidylcholine component of human low density lipoprotein (LDL) results in accumulation of diacylglycerol in the monolayer, which leads to lipoprotein particle aggregation and development of sample turbidity (24). Although PL-C-mediated LDL aggregation is irreversible, it can be prevented by inclusion of amphipathic exchangeable apolipoproteins in the incubation. In the present study, LDL (50 g of protein, isolated by sequential density ultracentrifugation (25)) was incubated with 0.16 units of Bacillus cereus PL-C (Sigma) (1 unit liberates 1 mol of watersoluble organic phosphorus from egg yolk L-␣-phosphatidylcholine per min at pH 7.3 at 37°C) in the presence or absence of 40 g of apoLp-III in buffer (50 mM Tris-HCl, pH 7.5; 150 mM NaCl, and 2 mM CaCl 2 ) at 37°C. At indicated time points sample absorbance at 340 nm was determined.
Competition Binding Assay-A variation of the assay described above was used to compare the abilities of WT and mutant apoLp-III to initiate a stable binding interaction with PL-C-treated LDL. LDL (250 g of protein) was treated with 0.88 units of PL-C in the presence of 3 H-WT recombinant apoLp-III (250 g of total culture medium protein) and indicated amounts of unlabeled competitor (WT or mutant) apoLp-III for 40 min at 37°C. The reaction was stopped by the addition of KBr (2.8 M final concentration) and EDTA (3.5 mM final concentration) (26). The LDL fraction was re-isolated by KBr density gradient ultracentrifugation and 3 H-WT apoLp-III bound to LDL determined by liquid scintillation spectrometry. Protein concentration was determined by the bicinchoninic acid protein assay (Pierce) and the specific activity expressed as dpm/mg protein.

RESULTS
Biophysical Properties-Loop regions that connect helices 1 and 2 and helices 3 and 4 in the globular ␣-helix bundle conformation of L. migratoria apoLp-III span residues 32-37 (loop A) and residues 88 -96 (loop C). These loops possess hydrophobic character and have been suggested to interact with surfacelocalized diacylglycerol in lipoproteins to initiate stable binding (7). To test this hypothesis, specific leucine residues in the loops were converted to arginine, L32R, L34R, or L95R. 2 A fourth mutant, where all the three Leu residues were mutated to Arg (L32R/L34R/L95R, referred to as the triple mutant), was also designed. The precise location of the mutations are depicted in Fig. 1. Before conducting functional analysis of the mutant apolipoproteins, their structural properties were char- acterized. CD spectroscopy revealed that each of the mutants possess a high ␣-helical content (approximately 75%, results not shown). In temperature-induced denaturation studies monitored by CD, it was observed that WT apoLp-III showed an increase in content of random coil at 40°C with a maximum reached at 70°C. Upon cooling, the protein regains its helical structure indicating the process is reversible. Plots of ellipticity as a function of temperature revealed a 55°C midpoint of temperature-induced denaturation. A similar reversible denaturation was observed for the mutant apoLp-IIIs, although lower denaturation midpoints were found for L32R, L34R, and the triple mutant (Table I) Table I) and human apoA-II (⌬G D H 2 O ϭ 1.0 kcal/mol (28)). Thus, we conclude that the mutations introduced did not cause major changes in secondary structure content or stability.
Phospholipid Vesicle Clearance-When exchangeable apolipoproteins are mixed with unilamellar phospholipid vesicles at their transition temperature, discoidal protein-lipid complexes are formed (29). This process can be monitored by light scattering spectroscopy since the turbid vesicle solution (ϳ200 nm in diameter) is clarified upon transformation to discoidal particles (ϳ14 nm in diameter (8)). Two different phospholipid vesicles were examined as follows: negatively charged DMPG and zwitterionic DMPC. As shown in Fig. 2A, WT apoLp-III causes clearance of DMPG vesicle suspensions, whereas in the absence of apolipoproteins DMPG vesicles remain turbid. All single mutant apoLp-IIIs were able to cause DMPG vesicle clearance at 2-6-fold higher rates than WT apoLp-III, which required 95 s to achieve 50% of maximal clearance (t1 ⁄2 , see Table II). The triple mutant was 19-fold more efficient than WT apoLp-III in its ability to cause clearance (t1 ⁄2 ϳ5 s). Also, L32R and L34R apoLp-IIIs, with mutations in loop A, showed a more rapid interaction with DMPG vesicles than L95R apoLp-III (in loop C). To characterize further the effect of charge-charge interactions between phospholipid vesicles and the different mutants, clearance assays with zwitterionic DMPC vesicles were carried out. WT apoLp-III-induced clearance of DMPC vesicle turbidity is a slower process compared with DMPG, as judged by its higher t1 ⁄2 (455 s), and the higher protein:lipid ratio (Fig. 2B). A stronger interaction of apolipoprotein for  DMPG compared with DMPC has also been observed for human apoA-I (29). The clearance rates varied with each mutant, with L34R and the triple mutant displaying the most effective clearance activity. L32R and L95R apoLp-IIIs showed rates of clearance similar to WT apoLp-III.
Protection against PL-C-induced LDL Aggregation-The ability of WT and mutant apoLp-IIIs to bind to hydrophobic defects on a spherical lipoprotein surface was examined (Fig.  3). Treatment of isolated human LDL with PL-C creates surface-localized diacylglycerol, which destabilizes the LDL particle. This unstable, modified LDL tends to aggregate, resulting in sample turbidity which can be followed turbidimetrically. Inclusion of WT apoLp-III in the incubation prevents aggregation due to its ability to bind to the modified LDL surface (23). In the present study, WT and mutant apoLp-IIIs (L32R, L34R, and L95R) conferred complete protection against PL-C-induced LDL aggregation. The triple mutant, however, lost some of its ability to protect after extended incubation times. Control incubations of LDL without PL-C demonstrated no change in absorbance, indicating that LDL alone was stable under the conditions employed. We conclude that each of the four mutant apoLp-IIIs are capable of binding to lipolyzed LDL.
Competition Binding Assay-In order to compare the relative ability of WT and mutant apoLp-IIIs to initiate lipid binding, a competition binding assay was employed. In this case, PL-C treated LDL was incubated simultaneously with tritiated WT apoLp-III and a given unlabeled apoLp-III. These two "species" of apoLp-III then compete for binding sites created on the surface of LDL. The reaction was stopped after 40 min, a time point where WT apoLp-III fully protects LDL from aggregation. In experiments comparing the effect of increasing amounts of unlabeled WT competitor, a concentration-dependent reduction of LDL bound 3 H-apoLp-III was observed (Fig. 4). Significant reductions in the relative amount of 3 H-apoLp-III bound to LDL were observed using 50 -400 g of unlabeled recombinant WT apoLp-III. Thus, 250 g of unlabeled competitor apolipoprotein was used in subsequent assays comparing WT and mutant apoLp-IIIs.
Inclusion of 250 g of unlabeled WT apoLp-III caused reduction of LDL bound 3 H-WT apoLp-III from 100 to 16%. However, equivalent amounts of L32R, L34R, or L95R apoLp-IIIs were less effective competitors (Fig. 5). With these proteins LDLassociated radioactivity was reduced to 41 Ϯ 7% (L32R), 43 Ϯ 6% (L34R), and 53 Ϯ 5% (L95R) under the conditions employed. These values are significantly different (p Ͻ 0.01) from that observed for WT apoLp-III, with little difference among the single mutants. The triple mutant showed a strongly decreased ability to compete with 3 H-WT apoLp-III for binding sites created by PL-C, suggesting a cumulative effect of the mutations on its ability to initiate lipoprotein binding. DISCUSSION When the x-ray structure of L. migratoria apoLp-III was solved (7), one of the questions that arose related to the mode of interaction of this globular ␣-helix bundle protein with lipoproteins. It was postulated that neutral lipid accumulation on the surface of lipoproteins triggers binding of apoLp-III via confor-  3. Effect of apolipoproteins on PL-C-induced turbidity of isolated human LDL. Fifty g of LDL protein was incubated at 37°C with 160 milliunits of PL-C in the presence or absence of WT or mutant apoLp-III (40 g each). Sample absorbance was monitored at 340 nm. PL-C treated LDL was incubated with the following: no apolipoprotein (q); WT apoLp-III (E); L32R apoLp-III (ϫ); L34R apoLp-III (f); L95R apoLp-III (Ⅺ); L32R/L34R/L95R apoLp-III (triple mutant) (OE). Values are mean Ϯ S.D. of three determinations.

FIG. 4. Effect of unlabeled WT apoLp-III on PL-C-induced binding of 3 H-WT apoLp-III to isolated human LDL.
LDL (250 g of protein) was incubated at 37°C with 0.88 units of PL-C, 3 H-WT apoLp-III (250 g of culture medium protein), and specific amounts of unlabeled WT apoLp-III. After 40 min incubation LDL was re-isolated by density gradient ultracentrifugation, and the amount of LDL-associated radioactivity was determined by liquid scintillation spectrometry. The amount of radioactivity bound to LDL in incubations containing 3 H-WT apoLp-III in the absence of unlabeled WT apoLp-III was taken as 100% (ϳ5 ϫ 10 5 dpm). mational opening of the bundle (7). Such an opening (helices 1, 2, and 5 moving away from helices 3 and 4) was proposed to involve putative hinge domains (loops B and D in Fig. 1). In this model, apoLp-III exposes its interior to interact with a lipid surface but can re-adopt the closed conformation following release from the lipoprotein surface. A similar mechanism has been proposed for the N-terminal domain of human apoE where lipid binding is essential for LDL receptor recognition (1,30). This open conformation model has been supported by a series of experiments reported over the last 5 years (8,(31)(32)(33)(34). Direct experimental evidence came from engineering a disulfide bridge in apoLp-III, locking the helix bundle in the lipidfree closed conformation that resulted in abolition of lipoprotein binding (33). This leads to the intriguing question as to how apoLp-III initiates lipoprotein binding. Whereas hydrophobic residues are likely involved in this role, their predominant localization within the interior of the helix bundle would imply that partial unfolding of the protein has to occur. However, loops A and C, located at one end of the protein ( 29 HETL-GLPTPD 38 and 87 SIHDAATSLN 96 ), were noted to contain hydrophobic amino acids. 3 Alignment of apoLp-III from several insect species suggested the presence of conserved Leu residues at or around positions 34 and 95 (17). These Leu residues were suggested to function in recognition of potential binding sites on the lipoprotein surface. This initial contact may be considered the first step in lipoprotein binding (18), followed by conformational opening of the helix bundle and spreading out on the lipid surface. In this study, we have investigated the role of such hydrophobic sensor residues in lipoprotein binding by replacing candidate Leu residues with Arg, thereby dramatically altering the hydrophobic character of the loop segment in question. The hydrophobicity of the loops of the single mutants, according to the normalized Eisenberg consensus (35), decreased from 0.70 to Ϫ2.89 (loop A) and from 1.19 to Ϫ2.40 (loop C). On the other hand, loops B ( 67 HQG 69 ) and D ( 125 SAQEAWAPV 133 ), located at the opposite end of the helix bundle (Fig. 1), are postulated to function as hinge domains, about which the helix bundle opens. Unlike the Leu residues in loops A and C, the two hydrophobic residues present in loop D (Trp-130 and Val-133) are not conserved across species (17). Loop B lacks hydrophobic residues. Thus, based on these considerations, we focused on Leu residues located in loops A and C.
The secondary structure content, denaturation properties, and ⌬G D H 2 O values of the mutants were similar to that of WT protein and in the range observed for other exchangeable apolipoproteins (⌬G D H 2 O M. sexta apoLp-III: 1.68 kcal/mol (27); apoA-I: 4.2 kcal/mol (36); apoA-II: 1.0 kcal/mol (28); and apoC-II: 2.8 kcal/mol (37)). The midpoint of temperature-induced denaturation of L. migratoria apoLp-III is similar to that of M. sexta apoLp-III (55 and 52°C, respectively). As described earlier for M. sexta apoLp-III (27), temperature-induced denaturation of locust WT and mutant apoLp-III is completely reversible. Of the three Leu residues in question, Leu-95 is the most solvent-accessible residue (88%), 3 and its replacement by Arg did not result in any change in the measured biophysical properties of apoLp-III. The observed small differences in the biophysical properties are likely due to local effects in the region surrounding the mutations. Therefore, changes in the function of the protein are likely a result of these local changes.
DMPG vesicle clearance data showed that substitution of Leu for Arg in either loop A or C enhanced clearance rates. The triple mutant displayed the highest rate, followed by both L32R and L34R apoLp-IIIs. L95R apoLp-III also displayed an increased clearance rate, although the level of enhancement was less than that observed with "loop A" mutant apolipoproteins. Apparently, serial introduction of Arg results in a cumulative increase in the ability to transform vesicles into discoidal complexes. Therefore, the higher DMPG clearance rates observed with mutant apoLp-IIIs are likely a result of charge effects that facilitate electrostatic attraction of apoLp-III to the negatively charged DMPG vesicle surface. In addition, clearance studies using zwitterionic DMPC vesicles showed that L34R apoLp-III and the triple mutant interacted more rapidly with DMPC than WT apoLp-III. Taken together, these results support the view that it is the end of the apoLp-III bearing loops A and C that initiates contact with these phospholipid surfaces.
Formation of a stable apoLp-III-lipoprotein interaction is thought to result from replacement of helix-helix contacts in the bundle conformation (Fig. 1) by helix-lipid interactions at the lipoprotein surface through conformational opening (7). Since the Leu residues replaced in the present study are localized at one end of the molecule, their substitution by Arg is not expected to result in abolition of lipoprotein binding. Indeed, single mutants and the triple mutant (bearing three additional positive charges in the loops) were able to bind to lipolyzed LDL, although the triple mutant demonstrated a slightly decreased ability to protect over an extended period of time (Fig.  3). Since this turbidimetric binding assay is qualitative, we developed a competition assay to quantify the interaction of exchangeable apolipoproteins with PL-C-treated LDL. LDL was co-incubated with 3 H-and unlabeled apoLp-III (WT or mutant), which compete for binding sites created on the surface of LDL by the action of PL-C. This assay showed that the three single Arg mutants were poor competitors compared with unlabeled WT protein. Under the conditions employed, the ability of the single mutants to compete with WT apoLp-III for binding sites on lipolyzed LDL was reduced 3-fold, with the triple mutant eliciting a 5-fold reduction. These data indicate that replacing Leu with Arg has a profound effect on the ability of the protein to compete with WT apoLp-III for binding to hy-  3 H-WT apoLp-III (250 g of culture medium protein), and 250 g of unlabeled competitor apoLp-III. After 40 min incubation LDL was re-isolated by density gradient ultracentrifugation, and the LDL-associated radioactivity was determined by liquid scintillation spectrometry. The amount of radioactivity bound to LDL in the absence of unlabeled competitor apoLp-III was taken as 100% (ϳ5 ϫ 10 5 dpm). Values are mean Ϯ S.E. of five independent determinations. Values obtained with mutants L32R, L34R, and L95R differed significantly with that from WT apoLp-III and the triple mutant (p Ͻ 0.01). drophobic defects created on the surface of LDL. These results corroborate data from M. sexta apoLp-III wherein mutation of Val-97, located in the short helix (helix 3Ј) connecting helices 3 and 4 (13), resulted in a strong reduction in lipoprotein binding ability (38). The location of helix 3Ј is similar to that of loop C in locust apoLp-III. Importantly, this mode of exchangeable apolipoprotein interaction may also apply to human apolipoproteins. In the case of human apoE, its N-terminal fourhelix bundle domain possesses a short connecting helix at one end of the bundle whose sequence is conserved across species. This helix is well suited to function as a recognition site to initiate lipid association of the bundle, an event that is required for manifestation of apoE's LDL receptor recognition properties (1). Furthermore, based on evidence that apoA-I adopts a helix bundle conformation (39,40), together with conformational changes upon lipid binding wherein the protein is postulated to adopt an extended, continuously curved ␣-helical conformation (41,42), it is conceivable that a similar recognition motif exists in apoA-I as well.
In considering possible explanations for the data presented in this study, it is important to consider the stark contrast between results of phospholipid vesicle clearance assays versus lipoprotein binding assays. In the former, there was a dramatic increase in binding upon mutagenesis of Leu to Arg, whereas the opposite was true for lipoprotein binding. Although we did not detect major changes in protein stability or structure, it is conceivable that the mutations caused an increase in the flexibility of the protein, specifically in and around the loop regions. Such a change in local structure could cause an increase in vesicle clearance, similar to the effect of pH on apoLp-IIIinduced DMPC vesicle clearance reported by Soulages and Bendavid (43). These authors proposed that formation of a partially folded conformation (possibly a "molten globule" state) is responsible for increased vesicle binding as a function of decreasing pH. If the Leu to Arg mutations induced formation of such a partially folded conformation, however, it would be expected that their lipoprotein binding abilities would also be enhanced. This was not the case and, in fact, the mutant apoLp-IIIs showed a marked decrease in lipoprotein binding compared, compared with WT apoLp-III. By contrast, the divergent results obtained in these two assay systems may be explained by considering the fundamental basis of the binding interactions. In the case of DMPG and DMPC vesicles, the increased positive charge introduced by the mutations at one end of the protein increases the electrostatic attraction of the protein for the vesicle surface, facilitating vesicle disruption. In the case of lipoprotein binding, however, binding is dictated by the presence of hydrophobic sites on the particle surface. Elimination of the Leu residues at the end of the molecule that contacts the lipid surface diminishes this attraction and is manifest by decreased binding ability. This effect is accentuated in the present study by the fact that Leu residues were replaced by the positively charged Arg. The presence of charged side chains at this location in the protein likely decreased the attraction for hydrophobic binding sites created on the surface of the LDL particles. In this manner, the present results provide support for the concept that hydrophobic residues in loops A and C of apoLp-III play a role in binding to lipoprotein surfaces. Furthermore, the data on phospholipid clearance corroborate this interpretation and provide additional evidence that the end of apoLp-III bearing loops A and C initiate contact with potential lipid-binding sites.
Phospholipid vesicle clearance assays are widely used in studies of apolipoprotein structure and function (44). However, the present study documents that apolipoprotein interaction with phospholipid vesicles is an essentially different process from interactions with hydrophobic surface defects on spherical lipoproteins. The initial encounter between apolipoprotein and vesicle is followed by penetration of the apolipoprotein into the bilayer, disrupting it to form disc-like particles. In these complexes apolipoprotein helical segments circumscribe the disc perimeter, covering otherwise exposed acyl chains (30,31,44). Unlike its interaction with DMPG or DMPC bilayer vesicles, apoLp-III does not form a stable binding interaction with the zwitterionic monolayer surface of LDL. Binding takes place in vitro, however, when hydrophobic binding sites (diacylglycerol) are created on the particle surface by the action of PL-C. Importantly, diacylglycerol is the physiologically relevant substrate for apoLp-III, and this lipid is known to exist in the surface monolayer of low density lipophorin (4,(45)(46)(47)(48). Further study is ongoing to assess whether the present conclusions are applicable to helix bundle apolipoproteins in general.