Alteration in apolipoprotein A-I 22-mer repeat order results in a decrease in lecithin:cholesterol acyltransferase reactivity.

Apolipoprotein A-I contains eight 22-amino acid and two 11-amino acid tandem repeats that comprise 80% of the mature protein. These repeating units are believed to be the basic motif responsible for lipid binding and lecithin:cholesterol acyltransferase (LCAT) activation. Computer analysis indicates that despite a fairly high degree of compositional similarity among the tandem repeats, significant differences in hydrophobic and amphipathic character exist. Our previous studies demonstrated that deletion of repeat 6 (143-164) or repeat 7 (165-186) resulted in a 98-99% reduction of LCAT activation as compared with wild-type apoA-I. To determine the effects of substituting one of these repeats with a more hydrophobic repeat we constructed a mutant apoA-I protein in which residues 143-164 (repeat 6) were replaced with repeat 10 (residues 220-241). The cloned mutant protein, 10F6 apoA-I, was expressed and purified from an Sf-9 cell baculoviral system and then analyzed using a number of biophysical and biochemical techniques. Recombinant complexes prepared at a 100:5:1 molar ratio of L-α-dimyristoylphosphatidylcholine:cholesterol:wild-type or 10F6 apoA-I showed a doublet corresponding to Stokes diameters of 114 and 108 Å on nondenaturing 4-30% polyacrylamide gel electrophoresis. L-α-Dimyristoylphosphatidylcholine 10F6 apoA-I complexes had a 5-6-fold lower apparent Vmax/apparent Km as compared with wild-type apoA-I containing particles. As expected, monoclonal antibody epitope mapping of the lipid-free and lipid-bound 10F6 apoA-I confirmed that a domain expressed between residues 143 and 165 normally found in wild-type apoA-I was absent. The region between residues 119 and 144 in 10F6 apoA-I showed a marked reduction in monoclonal antibody binding capacity. Therefore, we speculate that the 5-6-fold lower LCAT reactivity in 10F6 compared with wild-type apoA-I recombinant particles results from increased stabilization within the 121-165 amino acid domain due to more stable apoprotein helix phospholipid interactions as well as from conformational alterations among adjacent amphipathic helix repeats.

There is a significant, negative correlation between the plasma concentration of high density lipoproteins (HDLs) 1 and coronary artery disease in human populations and in animal models of atherosclerosis (1)(2)(3). Recent investigations using apolipoprotein A-I (apoA-I) transgenic and knockout animals have clearly demonstrated the protective role of HDLs in the prevention of atherosclerosis by "reverse cholesterol transport" (4 -8). The predominant protein constituent of HDL, apoA-I, solubilizes and organizes phospholipid, cholesterol, and cholesteryl ester in plasma and facilitates the distribution of cholesteryl ester between hepatic and extrahepatic tissue. Conversion of nascent HDLs or discoidal complexes into mature HDL particles is mediated by the plasma enzyme lecithin:cholesterol acyltransferase (LCAT). When activated by apoA-I, LCAT catalyzes the conversion of HDL-cholesterol to HDL-cholesteryl esters (4,5). However, the precise mechanism of LCAT activation by HDL-bound apoA-I remains largely unknown.
Investigations aimed at identifying how apoA-I organizes lipid and activates LCAT have focused on the unique structural properties of this protein (6 -14). ApoA-I contains 10 tandem repeats, each having substantial ␣-helical character that typically begin with a proline residue, a feature that is shared within the entire apolipoprotein supergene family. These repeats comprise approximately 80% of the total protein and are divided into two 11-mer units and eight 22-mer units (15,16), a feature that is highly conserved across species. When displayed on an Edmunson wheel, the ␣-helices show a distinctive amphipathic character with separate hydrophilic and hydrophobic faces. The amphipathic nature of these repeating units is believed to be responsible for the ability of apoA-I to organize lipid and to activate LCAT (17). A number of different models have been suggested to explain how apoA-I's amphipathic repeats associate with phospholipid and how these repeats activate LCAT (6 -14, 18 -25). Progress toward obtaining crystals of either lipid-free or lipid-bound apoA-I has been slow, and a three-dimensional crystal structure is not available to reconcile the different models.
We have continued our investigation into the structure-function relationships of apoA-I by constructing a 22-mer substitution mutant in a region previously shown to be necessary for LCAT activation. Sequential deletion of any one of the 10 tandem repeats results in a modest (30%) or dramatic (99%) reduction in LCAT reactivity compared with wild-type apoA-I (25). However, several studies (26,27) have shown that the region corresponding to residues 143-186 (22-mer, repeats 6 and 7) was absolutely essential for LCAT activation. In this paper, we report the properties of a mutant apoA-I protein in which repeat 6 (residues 143-164) was replaced with a copy of repeat 10 (residues 220 -241). The cloned mutant protein, called 10F6 apoA-I, was expressed and purified from an Sf-9 cell baculoviral system and used for biophysical characterization, monoclonal antibody epitope mapping, and LCAT reactivity measurements. In this report, we describe those structural features of the residue 121-186 domain that are important for optimal LCAT activation.

EXPERIMENTAL PROCEDURES
HPLC grade organic solvents were purchased from Fisher. All other chemical reagents were purchased from Sigma unless otherwise noted. [ 3 H]Cholesterol was purchased from DuPont NEN. Tissue culture reagents, restriction endonucleases, and other DNA modifying enzymes were purchased from Life Technologies, Inc.
Expression of Wild-type and Mutant pBlueBac III:ApoA-I Constructs-Wild-type and 10F6 apoA-I cDNA:pBlueBac III constructs were used in conjunction with the Autographa californica nuclear polyhedrosis virus linear viral DNA for co-transfection into Sf-9 cells, as described previously (28). Recombinant wild-type and mutant baculoviral clones were purified and used for the generation of high titer viral stocks as described previously (28). To prevent degradation of the expressed protein, pepstatin A and leupeptin (at a final concentration of 700 g/liter and 500 g/liter) were added to the culture medium 36 h postinfection (29).
Preparation and Purification of Recombinant Wild-type and 10F6 ApoA-I-Preparation and purification of recombinant wild-type and 10F6 apoA-I protein were carried out as previously reported (28). Briefly, at the time of harvest (50 -72 h), Sf-9 medium was spun at 4°C, 10,000 rpm for 10 min, and the supernatant was adjusted to 10% acetonitrile (v/v). The adjusted medium was loaded onto a C-18 reverse phase (50 ϫ 5-cm) HPLC column and washed with 10% acetonitrile, 0.1% trifluoroacetic acid. ApoA-I was eluted from the column in 1 h at a flow rate of 10 ml/min using a linear gradient of 10 -95% acetonitrile, 0.1% trifluoroacetic acid at ambient temperature. The partially purified apoA-I was adjusted to 5 mM Tris, pH 8.0, 8 M urea and then applied to a DEAE-Fast-Sepharose anionic exchange column (16 ϫ 3 cm). ApoA-I eluted at approximately 9 mM Tris, pH 8.0, 8 M urea using a linear gradient (200 ml each) that went from 5 mM Tris, pH 8.0, to 150 mM Tris, pH 8.0, 8 M urea at a flow rate of 144 ml/h. The final purified protein was dissolved in 1 mM ammonium bicarbonate, pH 7.4, and its concentration was determined by the method of Lowry (30). The molecular weight of the proteins was confirmed by electrospray mass spectrometry on a Quattro II mass spectrometer.
Preparation of Discoidal Complexes Containing ApoA-I-A molar ratio of 100:5:1 phospholipid to cholesterol to apoA-I protein was used for making discoidal complexes as described previously (28). Briefly, 2.1 mg of L-␣-dimyristoylphosphatidylcholine (DMPC) in chloroform (30 mg/ml) was added to 60 g of cholesterol in ethanol (10 mg/ml) and 10 l of radiolabeled cholesterol (1,2-3 H) (50 Ci/mmol) in ethanol. Organic solvent was removed under a stream of argon, and the tubes were placed under vacuum for 30 min. 2.7 mg of sodium cholate was added, and the solution was vortexed and then incubated for 30 min at 39°C. The mixture was briefly vortexed three times during the 30-min incubation. To this mixture was added 0.9 mg of wild-type or mutant apoA-I, and the incubation was continued for an additional 1 h at 39°C. Sodium cholate was removed by dialysis against 10 mM Tris, pH 7.4, 140 mM NaCl, 0.25 mM EDTA, and 0.15 mM sodium azide at 39°C. Recombinant discoidal complexes were purified by passage through a Superose 12 (Pharmacia Biotech Inc.) column (55 ϫ 1.8 cm) at a flow rate of 1 ml/min. Phospholipid, protein, and cholesterol assays were performed to determine the final molar composition of the recombinant discoidal complexes (28).
Nondenaturating Gradient Gel Electrophoresis-Discoidal complex size was determined using nondenaturating gradient gel electrophoresis. Briefly, gels were run for 3,000 V-h and then fixed in 10% sulfosalicylic acid (31). Following fixation, gels were stained in Coomassie G-250 overnight and then destained in 7.5% acetic acid for 1-2 days. Gels were scanned using a Zeineh scanning densitometer model SL-504-XL. Discoidal particle size was determined by comparison with protein standards of known Stokes radii.
Circular Dichroism Spectroscopy-Circular dichroism spectra were recorded with a Jasco J720 spectropolarimeter at 25°C using a 0.1-cm path length cell. Ellipticity was measured at 222 nm. Five scans were recorded and averaged, and the background was subtracted. Mean molar residue ellipticity (⍜) is reported as degrees⅐cm 2 ⅐dmol Ϫ1 and calculated from the equation, [⍜] ϭ ⍜ obs ⅐115/10⅐l⅐c, where q obs is the observed ellipticity at 222 nm in degrees, 115 is the mean residue molecular weight of the protein, l is the optical path length in centimeters, and c is the protein concentration in g/ml. The percentage of ␣-helix was calculated from the formula of Chen et al. (32), [⍜] 222 ϭ Ϫ30,300f h Ϫ 2,340. Stability of the discoidal 10F6 and wild-type DMPC complexes was determined by plotting the mean residue ellipticity versus guanidine HCl concentration and expressed as the concentration of guanidine HCl (D1 ⁄2 ) that reduced the ellipticity by 50%.
Epitope Mapping Studies-Competitive solid phase immunoassays were used to assess the binding of monoclonal antibodies to lipid-free apoA-I and recombinant DMPC phospholipid complexes containing apoA-I. With two exceptions (antibodies AI-17 and AI-141.7), each antibody used in this study has been described, and its epitopes on wildtype apoA-I have been documented (32). The epitopes of antibodies AI-17 and AI-147.7 were localized to residues 143-165 and 220 -242, respectively, on the basis of their binding to numerous apoA-I mutant proteins described previously. 2 The immunoassays were performed as described (33). Briefly, isolated human plasma apoA-I or plasma HDL (0.05 ml of 5 g/ml) was immobilized onto 96-well Falcon 3911 Microtest III flexible assay plates. After coating the plates, increasing amounts of purified apoA-I or recombinant A-I discs (0.025 ml) diluted in phosphate-buffered saline containing 3% normal goat serum were added to wells in the presence of 0.025 ml of ascites fluid containing a limiting amount of monoclonal antibody (typically dilutions of 10 Ϫ4 to 10 Ϫ6 ). Competitor concentrations listed in the figures represent the final concentrations (g/ml) in the 0.05-ml reaction mixture. The plates were incubated overnight at 4°C. After washing the wells, mouse antibody binding to the immobilized antigen was detected by a second 1-h incubation at 37°C with 125 I-labeled goat anti-mouse IgG. All data were expressed as B/B o , where B is the cpm bound in the presence of competitor, and B o is the cpm bound in the absence of competitor. To compare the affinity of the antibodies for DMPC discs containing wildtype or mutant apoA-I's competitor, the slopes of logit-transformed B/B o ratios were obtained by linear regression and subjected to tests of equality.
LCAT Reaction Kinetics-The LCAT reaction was monitored by following the cholesterol to cholesteryl ester conversion using recombinant discoidal complexes containing either wild-type or 10F6 apoA-I. The complexes were assayed in duplicate using 0 -3.0 g of substrate cholesterol in a final concentration of 10 mM Tris, pH 7.4, 140 mM NaCl, 0.25 mM EDTA, and 0.15 mM sodium azide, 0.6% fatty acid-free bovine serum albumin, 2 mM ␤-mercaptoethanol, and 25 ng of purified human plasma LCAT (kindly provided by Dr. J. S. Parks). The reactions were carried out for 60 min at 37°C, and the conversion of [ 3 H]cholesterol to [ 3 H]cholesteryl ester was determined by lipid extraction of the incubation mixture followed by thin layer chromatography (28). The extent of cholesterol esterification was kept below 15% to maintain first order kinetics. Background values were determined by omitting LCAT from the reaction tube. The fractional cholesterol esterification rate was multiplied by the nmol of substrate cholesterol in the assay tube, corrected for the background, and converted to nmol of cholesterol ester formed/h/ml of LCAT. Apparent V max and K m values were determined from Hanes-Woolf plots (34) of discoidal cholesterol substrate concentration (M) divided by the cholesterol ester formation rate (nmol/h/ml of LCAT) versus discoidal cholesterol concentration (M).
The activation energy (E a ) for cholesterol ester formation was derived from Arrhenius plots of the LCAT reactivity of wild-type and 10F6 discoidal substrates. Reactions were performed at 37, 34, 31, 28, and 25°C using 0.55 g of discoidal cholesterol substrate and 25 ng of purified human LCAT. Arrhenius plots were constructed as the reciprocal of incubation temperature in Kelvin versus the log of cholesteryl ester formed (nmol/h) (34). Linear regression analysis was used to determine the activation energy according to the formula, E a (cal/mol) ϭ Ϫ2.3R(slope), where R is the gas constant (1.987 cal/deg Ϫ1 /mol Ϫ1 ).
Data Analysis-Values are given as the mean Ϯ S.D. Statistical comparisons were made using Student's t test.

RESULTS
To test whether substitution of a single 22-mer repeat located at 143-164 (repeat 6) would alter apoA-I's ability to bind lipid or to activate LCAT, we replaced repeat 6 of wild-type apoA-I with repeat 10 (corresponding to amino acids 220 -241). The boxed region in Fig. 1 shows the primary amino acid sequence of wild-type and 10F6 apoA-I within the residue 143-164 domain. The mutant protein 10F6 apoA-I contains one copy of repeat 10 inserted in the region normally occupied by repeat 6 and another copy of repeat 10 in its native position, 220 -241. Also shown in Fig. 1 (35,36). Replacing repeat 6 with repeat 10 increases the hydrophobicity of the 143-164 region in 10F6 apoA-I from Ϫ0.30 to ϩ0.24, reflecting the greater proportion of nonpolar and uncharged amino acids in this 22-mer.
The effect of the 10F6 apoA-I replacement mutation on the overall secondary structure of wild-type apoA-I protein was first examined by determining the percentage of ␣-helix content. Both lipid-free and lipid-bound wild-type and 10F6 apoA-I were analyzed by circular dichroism spectroscopy. As shown in Table I, a trend toward a higher ␣-helix content for both the lipid-free and lipid-bound 10F6 apoA-I compared with wildtype apoA-I was noted. However, these differences did not reach statistical significance at the level of p Ͻ 0.05. When examined in the lipid-bound state, the ␣-helix content for each of the apoA-Is studied increased compared to its lipid-free state (Table I).
Although the 10F6 apoA-I substitution did not show a statistically significant change in its ␣-helix content compared with lipid-free or lipid-bound wild-type apoA-I, the DMPC 10F6 apoA-I recombinant complexes were found to be less susceptible to guanidine HCl denaturation than DMPC wild-type apoA-I complexes. Guanidine HCl denaturation of DMPC 10F6 apoA-I complexes gave a D1 ⁄2 ϭ 2.8 M, while DMPC wild-type complexes gave a D1 ⁄2 ϭ 2.1 M (data not shown). These results demonstrate that the lipid-protein interaction in DMPC 10F6 complexes was more stable than that for DMPC wild-type apoA-I complexes.
In the next set of experiments the effects of substituting apoA-I's repeat 6 with repeat 10 were evaluated with respect to the protein's ability to activate LCAT. Kinetic studies were conducted, and the K m and V max of the reactions are summarized in Table II. The apparent K m values are 1.2 and 6.0 M and the apparent V max values are 51.0 and 48.0 nmol/ml⅐h for DMPC wild-type and 10F6 apoA-I complexes, respectively (Table II). The overall efficiency for the LCAT reaction, referred to as the apparent V max /apparent K m , was calculated to be 42.5 and 8.0 nmol/ml⅐h•/M for DMPC wild-type and 10F6 apoA-I complexes, respectively. To probe the mechanistic basis for the lower LCAT efficiency observed with the 10F6 apoA-I-containing particles, the energy of activation was measured as a function of temperature. The Arrhenius plots shown in Fig. 3 produced activation energies of 16.0 and 25.3 kcal/mol for DMPC wild-type and 10F6 apoA-I complexes, respectively.
To further explore the conformational impact of the 22-mer replacement mutant on apoA-I secondary structure, epitope mapping studies were carried out using a panel of defined monoclonal antibodies (33). First, the binding capacity of each antibody for lipid-free wild-type or 10F6 apoA-I were compared in competitive immunoassays. The results are tabulated in Table III. Six antibodies that identify N-terminal epitopes between residues 1 and 126 and three antibodies that identify C-terminal epitopes between residues 178 and 242 bound well to both apoproteins. These antibodies gave binding ratios (10F6 to wild type) of 5 or less, indicating that these particular epitopes were expressed by both apoproteins and that there were only minimal differences in the extent of epitope expression between the two apoA-I proteins. In contrast, four antibodies that identify epitopes between residues 119 and 165 on wild-type apoA-I either did not bind to purified 10F6 apoA-I or   (Table III).
To probe the conformational changes caused by association of apoA-I with lipid, we studied the binding of the same 13 antibodies to the recombinant DMPC discoidal complexes containing wild-type or 10F6 apoA-I. The six N-terminal antibodies, AI-16, AI-1.2, AI-19.2, AI-11, AI-4, and AI-115.1, had comparable affinity for both apoprotein-lipid complexes and had 10F6 to wild-type apoA-I binding ratios of 1.1 or less. This indicates that each of these epitopes was fully expressed on both the wild-type and 10F6 apoA-I when complexed with lipid. Competitive binding curves for two of these six antibodies are shown in Fig. 4. These antibodies bind epitopes that are linearly adjacent to helix 6, the AI-4 epitope within residues 99 -121 and the AI-115.1 within residues 115-126. The superimposable curves confirm that AI-4 has both comparable binding capacity and binding affinity for the DMPC wild-type and 10F6 complexes. The displaced binding curves indicates that 115.1 has a greater binding capacity for DMPC wild-type apoA-I complexes. However, the similar slopes of the two AI-115.1 curves demonstrates that antibody AI-115.1 binds the two epitopes with comparable affinity (Fig. 4). The three Cterminal antibodies, AI-178.1, AI-187.1, and AI-144, also bound both lipid-associated apoproteins with ratios between 1 and 7, and this pattern of reactivity was consistent with the results obtained with the lipid-free apoproteins (data not shown). Antibodies AI-137.1 and AI-17, which did not bind the lipid-free 10F6 apoA-I, also did not bind to the DMPC 10F6 apoA-I (Fig.  5), although they bound the DMPC wild-type apoA-I complexes. These data are in complete agreement with those obtained with the lipid-free apoA-Is (Table III) and further confirm the absence of these epitopes within the 137-165 region of 10F6 apoA-I protein. Competition curves for the two antibodies that bound residues 119 -144, a region that both adjoins and slightly overlaps the deleted repeat 6 (residues 143-164), are shown in Fig. 6. These antibodies needed significantly higher concentrations of lipid-free 10F6 apoA-I to achieve 50% inhibition of binding (Fig. 6, top panels), as compared with wild-type apoA-I. These results strongly suggest that replacement of repeat 6 with repeat 10 changed the conformation of epitopes in repeat 5 (residues 121-142) but had little effect on the conformation of epitopes in repeat 7. However, these same repeat 5 epitopes were more similar when the apoA-Is were complexed with DMPC (Fig. 6, bottom panel). Taken together, these results imply that repeats 5 and 6 of apoA-I interact with phospholipid in a manner that normalizes and stabilizes a native conformation at the lipid interface. DISCUSSION In this report we describe the biophysical and biochemical properties of the structural mutant, 10F6 apoA-I. This mutant protein was designed to investigate the effect of a single 22-mer repeat substitution on apoA-I structure and function. In previous studies (25), we showed that deletion of repeat 6 (residues 143-164) or repeat 7 (165-186) reduces the overall LCAT reaction velocity by 98 -99% relative to wild-type apoA-I (25). From these observations and from those of others (26,27), we proposed that the reduced LCAT reactivity observed with the ⌬6 apoA-I 22-mer deletion mutants resulted from disruption of interactions between adjacent helical repeats within apoA-I. We also suggested that electrostatic and hydrophobic side chain interactions were responsible for the proper alignment of the phospholipid substrate for LCAT catalysis (25). In the present studies, we have further refined our working model by characterizing apoprotein helix-lipid interactions and the apoprotein helix-helix interactions that are critical for optimal LCAT reactivity.
Recombinant DMPC complexes containing 10F6 apoA-I displayed a particle size distribution that was similar to wild-type and human plasma apoA-I-containing complexes. These data

FIG. 3. Arrhenius plots of the initial velocity of the LCAT reaction using discoidal complexes composed of DMPC cholesterol wild-type apoA-I (f--f) or DMPC-cholesterol-10F6
apoA-I (q--q) prepared at a 100:5:1 starting molar ratio. The energy of activation, E a , was calculated from the slope of the line as described under "Experimental Procedures." a Amount of soluble competitor (g/ml) required for 50% inhibition of binding to isolated and immobilized plasma apoA-I. agree well with previous reports describing the size, composition, and ␣-helix content of DMPC plasma apoA-I discoidal complexes (37). Despite the absence of a significant difference in particle size or composition, DMPC complexes containing 10F6 apoA-I showed a profound reduction in their ability to co-activate LCAT. The reactivity of the DMPC 10F6 apoA-I complexes was determined by measuring the initial velocity of the LCAT reaction as a function of cholesterol concentration. The kinetic parameters are shown in Table II and demonstrate that the overall catalytic efficiency (apparent V max /apparent K m ) of the LCAT reaction catalyzed by 10F6 apoA-I was reduced 5-6-fold compared with wild-type apoA-I. This difference was primarily due to a 5-6-fold increase in the apparent K m for 10F6 apoA-I complexes. These results suggest that the reduced binding affinity of LCAT to DMPC 10F6 apoA-I complexes was responsible for the large decrease in catalytic efficiency (38). Thus, we conclude that substitution of a single 22-mer repeat within the 143-165 domain resulted in a sub-stantial reduction in LCAT binding to the recombinant substrate and ultimately in a lower catalytic efficiency. The decrease in LCAT catalytic efficiency was accompanied by a 9 kcal/mol increase in the activation energy for 10F6 compared with wild-type apoA-I recombinant complexes (Fig. 3). This increase in activation energy for catalysis of DMPC 10F6 complexes may reflect the interaction of hydrophobic residues (residue 143-165 domain) with boundary phospholipid, in addition to altering side chain interactions between adjacent helices in this region (repeats 5, 6, and 7). This hypothesis is strengthened by the guanidine denaturation studies in which DMPC 10F6 complexes were denatured at a higher molar concentration of guanidine hydrochloride (2.8 M) than DMPC wild-type complexes (2.1 M).
The carboxyl-terminal region or repeats 9 and 10 (residues 209 -243) of apoA-I has been proposed to be important in cell binding (39), lipoprotein association (40,41,42), LCAT activation (25)(26)(27), and in vivo metabolism of HDL (40,42). Recent studies using truncation mutants of the carboxyl-terminal region demonstrated that the domain corresponding to residues 227-243 was critical in modulating the association of apoA-I with lipoproteins as well as in the in vivo metabolism of apoA-I (40,42). Other investigations show that substitution of the carboxyl-terminal region of apoA-I with a helical domain derived from apoA-II increases its hydrophobicity but does not restore the mutant protein's lipoprotein affinity or its ability to associate with HDL 3 (42). It is possible that in vivo apoA-I's carboxyl-terminal domain interacts cooperatively with the amphipathic helices located in the middle of the protein, thus allowing a conformation that can conform to particles of varying size and composition (42). The mechanistic role of this region in LCAT activation appears to be secondary to its role in lipid-lipoprotein association. Several studies have shown that deletion of repeat 9 or 10 results in at least a 60 -90% reduction in wild-type LCAT activation (25)(26)(27)41), while in one study, truncation of the residue 193-243 domain resulted in a 245% increase in wild-type LCAT activation (41). Thus, although this Microtiter plates were coated with isolated plasma HDL at 5 g/ml and incubated with a predetermined limiting amount of each antibody in the presence of increasing amounts of either DMPC wild type (E--E) or 10F6 (q--q) apoA-I discoidal complexes as described under "Experimental Procedures." To compare the affinities of each antibody for the two complexes, the B/B o binding ratios were subjected to logit transformation where logit(y) ϭ ln(y/1 Ϫ y) and y ϭ B/B o . The slopes were calculated by linear regression analysis of the logit-transformed data and were subjected to test of equality. Similar slopes indicted a similar affinity of the antibodies for wild-type and 10F6 apoA-I. The displacement of the 10F6 curve to the right of the wild-type curve observed with antibody AI-115.1 indicates that the antibody had comparable affinities but a reduced binding capacity for 10F6 apoA-I. As shown in Table III, antibodies AI-4 and AI-115.1 reacted with both lipid-free wild-type and 10F6 apoA-I.

FIG. 5. Binding of monoclonal antibodies to recombinant DMPC complexes containing wild-type (E--E) and 10F6
(q--q) apoA-I. Competition assays were performed, and the data were analyzed as described in Fig. 4 and under "Experimental Procedures."

FIG. 6. Binding of monoclonal antibodies to lipid-free wildtype (E--E) or lipid-free 10F6 (q--q) apoA-I (top panel) and binding to recombinant DMPC complexes containing wild-type (E--E) or 10F6 (q--q) apoA-I complexes (bottom panel).
Competition assays were performed and the data were analyzed as described in the legend to Fig. 4 and under "Experimental Procedures." region may play some role in LCAT activation, it appears that the contribution of the carboxyl-terminal region of apoA-I to LCAT activation is secondary to its intrinsic lipid binding affinity and particle formation properties.
Protein-lipid hydrophobic interactions are believed to play a critical role in the activation of LCAT. Because LCAT is a surface active protein that binds to the discoidal lipid-protein interface, it has been suggested that apoA-I's amphipathic ␣-helices serve to disrupt the aqueous to lipid interface and expose phospholipid to LCAT (4,5). Recently, NMR studies using an apoA-I fragment, residues 166 -185 (12), as well as a synthetic peptide activator of LCAT, i.e. LAP-20, have shown that hydrophobic interactions between the nonpolar amino acids and the phospholipid acyl chains play an important role in stabilizing lipid-protein complexes (43). Furthermore, the NMR studies suggest that intermolecular salt bridges and "snorkeling" (44) of basic amino acid side chains play a less important role compared with hydrophobic interactions in stabilizing lipid-apoprotein interactions (43). Recent studies using synthetic peptides corresponding to each of the eight 22-mer repeats of apoA-I have shown that repeat 10 (220 -241) has the highest lipid binding affinity of all eight repeats (45) and the greatest calculated depth of penetration into phospholipid vesicles (11). However, the relative lipid binding affinity of an amphipathic helix does not directly relate to a peptides' ability to activate LCAT (13).
Modeling studies have suggested that the mode of assembly of adjacent amphipathic helical repeats around the edge of a discoidal complex is determined by both the hydrophobic character of the residues and by the charge complementary along the edge of the helices (46). From crystallography studies, helix-helix interactions inside lipid bilayers have been shown to include interhelical salt bridges, hydrogen bonds, or precise packing interactions (47). Thus, these structural features may determine the overall stability and the relative orientation of the adjacent helices. Therefore, we suggest from our data that substitution of repeat 10 for repeat 6 increases the hydrophobic nature of the 121-165 region both with respect to apoproteinphospholipid interaction and to adjacent helix-helix interaction. Taken together, these highly stabilized interactions most likely allow the helices in the 121-165 domain to penetrate more deeply into the phospholipid bilayer, which then restricts LCAT's access to the boundary phospholipid acyl chains and results in reduced overall catalytic efficiency. Our hypothesis is consistent with a proposed mechanism for LCAT activation in which the central helices of apoA-I are believed to be displaced from contact with lipid by an apoE-like segment within LCAT during the binding and activation process (5).
From the data presented in this report, it appears that the substitution of repeat 10 within the residue 143-165 region had diverse effects on the expression of apoA-I antibody epitopes. First, most epitopes N-terminal of the substituted repeat remained unchanged. However, the substitution of repeat 6 altered the expression of at least one repeat 5 epitope, the epitope between residues 115-126, defined by antibody AI-115.1. The reduced (but not abolished) binding capacity of this antibody for the DMPC 10F6 apoA-I complexes suggested that at least a portion of the DMPC 10F6 apoA-I complexes did not express an AI-115.1 epitope. Second, two epitopes between residues 119 and 144, the region of apoA-I that both adjoins and overlaps the deleted repeat 6, had more comparable reactivity for both lipid-bound apoproteins. Third, as expected, the repeat 6 epitopes defined by antibodies AI-137.1 and AI-17 were lost. Finally, the three antibodies that identified epitopes that were C-terminal to repeat 6 reacted similarly with both lipid-free and lipid-bound wild-type and 10F6 apoA-I. Although the 10F6 apoA-I protein contains two copies per molecule of the AI-141.7 epitope (residues 220 -242), greater reactivity of antibody AI-141.7 for this mutant apoA-I was not observed with either lipid-free or lipid-bound 10F6 apoA-I. Overall, these results suggest that substitution of repeat 6 for repeat 10 resulted in alteration of native wild-type apoA-I epitopes in both repeats 5 and 6 (residues 121-165). The conformational alterations in repeats 5 and 6 were found to become less apparent when the apoprotein was complexed with lipid, presumably as a result of lipid-apoprotein interactions that dominated apoA-I's helix-helix interactions. In summary, these studies have aided our understanding of the mechanism of apoA-I's activation of LCAT by clarifying the nature of the interaction between phospholipid and apoA-I's amphipathic ␣-helices and the intermolecular helix-helix interactions critical for optimal catalytic efficiency.