Role of the Arg123–Tyr166 Paired Helix of Apolipoprotein A-I in Lecithin:Cholesterol Acyltransferase Activation*

The Arg123–Tyr166central and Ala190–Gln243 carboxyl-terminal pairs of helices of apoA-I were substituted with the pair of helices of apoA-II, resulting in the apoA-I(Δ(Arg123–Tyr166),∇A-II(Ser12–Ala75)) and apoA-I(Δ(Ala190–Gln243),∇A-II(Ser12–Gln77)) chimeras, respectively. The structures of these chimeras in aqueous solution and in reconstituted high density lipoproteins (rHDL) and the lecithin:cholesterol acyltransferase (LCAT) activation properties of the rHDL were studied. Recombinant human apoA-I and the chimeras were expressed in Escherichia coli and purified from the periplasmic space. Binding of the apolipoproteins with palmitoyloleoylphosphatidylcholine was associated with a similar shift of Trp fluorescence maxima from 337 to 332 nm, from 339 to 334 nm, and from 337 to 333 nm, respectively. All rHDL had a Stokes radius of 4.8 nm and contained 2 apolipoprotein molecules/particle. Circular dichroism measurements revealed eight α-helices per apoA-I and per chimera molecule. The catalytic efficiencies of LCAT activation were 1.5 ± 0.33 (mean ± S.D.; n = 3), 0.054 ± 0.009 (p < 0.001 versusapoA-I), and 1.3 ± 0.32 (p = not significantversus apoA-I) nmol of cholesteryl ester/h/μm, respectively. The lower LCAT activity of the central domain chimera was due to a 27-fold reducedV max with unaltered K m . Binding of radiolabeled LCAT to rHDL of apoA-I and apoA-I(Δ(Arg123–Tyr166),∇A-II(Ser12–Ala75)) was very similar. In conclusion, although substitution of the Arg123–Tyr166 central or Ala190–Gln243 carboxyl-terminal pair of helices of apoA-I with the pair of helices of apoA-II yields chimeras with structure similar to that of native apoA-I, exchange of the central domain (but not the carboxyl-terminal domain) of apoA-I reduces the rate of LCAT activity that is independent of binding to rHDL.

The structures in apoA-I involved in phospholipid binding and/or LCAT activation remain largely unidentified. Reported differences in LCAT activity of apoA-I and deletion mutants may result from altered folding and/or organization of these molecules in rHDL rather than from deletion of a functional domain (15,16). Therefore, in this study, the apoA-I(⌬(Arg 123 -Tyr 166 ),ٌA-II(Ser 12 -Ala 75 )) and apoA-I(⌬(Ala 190 -Gln 243 ),ٌA-II(Ser 12 -Gln 77 )) chimeras were produced, in which the Arg 123 -Tyr 166 central or Ala 190 -Gln 243 carboxyl-terminal pair of ␣-helices of apoA-I was deleted (⌬) and substituted (ٌ) with the pair of ␣-helices of apoA-II. The average structural properties in solution and in reconstituted high density lipoprotein particles of the two chimeras were found to be unaltered, but the central domain chimera had a markedly reduced LCAT activity.

EXPERIMENTAL PROCEDURES
Oligonucleotide-directed Mutagenesis and DNA Sequencing-All DNA manipulations were carried out essentially as described by Maniatis et al. (17). Oligonucleotide-directed mutagenesis was performed by the gapped-duplex method of Kramer et al. (18) using the pMa/c vector system of Stanssens et al. (19). This system employs phasmid (i.e. phage/plasmid hybrid) vectors, allowing cloning, sitedirected mutagenesis, and sequencing using the same vector without recloning. Oligonucleotides were obtained by custom synthesis (Pharmacia, Brussels, Belgium). DNA sequences were determined using a primer walking strategy on an ALF DNA sequencer (Pharmacia, Uppsala). Template DNA was purified by alkaline hydrolysis followed by a polyethylene glycol precipitation step. The sequencing reactions were carried out using T7 DNA polymerase (Pharmacia) and a fast denaturation protocol as described (20). The reaction products were sized on 6% Hydrolink Long Ranger gels (AT Biochem, Malvern, PA) containing 1 ϫ * This work was supported by Fonds voor Wetenschappelijk Onderzoek-Vlaanderen Project 3.0063.94. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Expression and Purification of ApoA-I and the ApoA-I/ApoA-II Chimeras-The pMc-5-apoA-I transfection vector, encoding wild-type recombinant apoA-I, for expression in Escherichia coli under control of the tac promoter and the PhoA signal peptide sequence was constructed starting from plasmid pA1-3 (21), the pUC-19 vector (22), and the pMa/c mutagenesis vector (19). Plasmid pA1-3, obtained by insertion of the apoA-I cDNA into the PstI site of pBR322 (21), was a kind gift of Dr. L. Chan (Baylor College of Medicine, Houston, TX).
Apolipoproteins were expressed in the periplasmic space of E. coli WK6 host cells as described (16,23). Standard apoA-I was isolated from normolipemic human plasma as described previously (1). The purity of proteins was established by SDS gel electrophoresis (24) and immunoblotting (25).
Preparation of Discoidal Apolipoprotein-POPC-Cholesterol Complexes-Complexes of the apolipoproteins with POPC (Sigma) and cholesterol, at an apolipoprotein/POPC/cholesterol ratio of 1:3:0.15 (w/w/ w), were prepared using the cholate dialysis procedure (11,26). The mixture was incubated at 43°C for 16 h, and cholate was removed by extensive dialysis. Complexes were isolated by gel filtration on a Superdex 200 HR column equilibrated with 20 mM Tris-HCl, pH 8.1, containing 0.15 M NaCl and 0.02 mg/ml sodium azide in a Waters fast performance liquid chromatography system. One-ml fractions were col-lected. The sizes of these complexes were estimated by comparison of their migration positions on native 8 -25% gradient polyacrylamide gels with those of standard proteins: thyroglobulin (Stokes radius of 8.5 nm), apoferritin (6.1 nm), catalase (5.2 nm), and lactate dehydrogenase (4.1 nm) (Pharmacia).
The number of apolipoprotein molecules/rHDL particle was determined following cross-linking of apolipoproteins with bis(sulfosuccinimidyl) suberate (final concentration of 1 mM) for 6 h, as described previously (27), and separation of cross-linked proteins by SDS-PAGE on 10 -15% gradient gels. The extent of oligomer formation was estimated by comparison with plasma cross-linked apoA-I (27). The levels of phospholipid and cholesterol were determined using commercial enzymatic kits (Biomérieux for phospholipid and Boehringer Mannheim for cholesterol), and the concentrations of apoA-I were determined according to Bradford (28).
Circular Dichroism Spectrometry-Circular dichroism spectra of lipid-free apolipoproteins in solution and of apolipoproteins in discoidal apolipoprotein-POPC-cholesterol complexes were measured with a Jasco J600 spectropolarimeter (Japan Spectroscopy, Tokyo) at wavelengths between 200 and 250 nm, using 6 M sample solutions and a 1-mm path length cuvette. Backgrounds were measured at 250 nm for 5 min, followed by measurements of the ␣-helical content at 222 nm for 5 min. The fraction of ␣-helices in the secondary structure of the apolipoproteins was estimated from the molar ellipticities at 222 nm Fluorescence Analysis-Corrected steady-state fluorescence spectra and intensities were measured as described previously (16). Excitation was carried out at 295 nm, where the contribution of tyrosyl fluorescence to the total intensity is minimal. Fluorescence was measured with a 2-mm slit width in quartz cuvettes with optical path lengths equal to 1 and 0.4 cm and wavelength resolutions of 7.2 and 3.6 nm for the emission and excitation wavelengths, respectively. Excitation was vertically polarized, and fluorescence was detected with a polarizer at a 54°a ngle to reduce the influence of fluorescence depolarization and brownian motion on the detected intensity. The spectra were corrected for the wavelength dependence of the emission monochromator and the photomultiplier and for background intensities as described previously (30). In the fluorescence-quenching experiments, the spectra were recorded at increasing concentrations of KI. The results were analyzed by the Stern-Volmer equation as modified by Lehrer (31): where F 0 and F are the fluorescence intensities observed in the absence and presence of a given concentration of KI, respectively, and f a is the fraction of quenchable fluorescence (32). The Stern-Volmer constant (K SV ) is an index of the accessibility of trypto-phan residues to the aqueous phase.
Generation of a Stable Cell Line Expressing Human LCAT-The pUC-LCAT.10 plasmid, covering the entire coding region of human LCAT, was obtained from Dr. J. Mc Lean (Department of Cell Biology, Genentech Inc., San Francisco, CA). The LCAT cDNA was cloned into the pcDNA3 expression vector (Invitrogen, San Diego, CA), which contains the enhancer-promoter sequences from the immediate-early region of human cytomegalovirus and a neomycin resistance gene. After trypsinization, 8 ϫ 10 6 293 cells were resuspended in 600 l of Dulbecco's modified Eagle's medium (Gibco, Gent, Belgium) supplemented with 10% fetal bovine serum (Gibco) and transfected with 75 g of pcDNA3-LCAT, linearized with XbaI, by electroporation. After electroporation, cells were plated at densities of 5 ϫ 10 6 , 10 6 , 5 ϫ 10 5 , and 10 5 /10-cm dish. One day after transfection, cells were placed on Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 600 g/ml Geneticin (Gibco). Cell colonies, appearing 14 days after transfection, were picked, replated, and screened for LCAT expression by spectrophotometric assessment of butyryl ester of pnitrophenol (33). Transfected 293 cells were cultured in serum-free Dulbecco's modified Eagle's medium for 48 h, and conditioned medium was collected and concentrated on a Centricon-10 filter (Amicon, Inc., Beverly, MA).
Kinetics of LCAT Activation by Discoidal Apolipoprotein-POPC-Cholesterol Complexes-LCAT activation by discoidal apolipoprotein-POPC-cholesterol complexes was monitored by incubation of the equivalent of 5-40 M unesterified cholesterol with a final concentration of 50 nM LCAT. The reaction was carried out in 10 mM Tris-HCl, pH 8.0, containing 5 mM EDTA, 0.15 M NaCl, 4 -16 mg/ml delipidated bovine serum albumin, and 6 mM ␤-mercaptoethanol (LCAT buffer). After preincubation for 20 min at 37°C, LCAT was added. The mixture was incubated at 37°C for 10 -180 min, after which the reaction was arrested by extraction of cholesteryl esters with hexane/isopropyl alcohol (3:2, v/v) containing cholesteryl heptadecanoate as an internal standard. Under these reaction conditions, Ͻ20% of the cholesterol was esterified. Generated cholesteryl esters were quantified by isocratic high performance liquid chromatography on a reversed-phase ZORBAX ODS column, eluted with acetonitrile/isopropyl alcohol (50:50, v/v), as described previously (34). Initial activation rates were obtained from plots of the concentration of generated cholesteryl esters versus time, and the kinetic parameters K m (expressed in M cholesterol) and V max (expressed in nmol of cholesteryl esters generated per h per M of cholesterol (cf. Table III)) were determined by linear regression analysis from Lineweaver-Burk plots.
Binding of 125 I-Labeled LCAT to Discoidal Apolipoprotein-POPC-Cholesterol Complexes-Recombinant LCAT was labeled with 125 I by the IODOGEN method. Following the reaction, radiolabeled LCAT was separated from free 125 I by passage over a Sephadex G-25M PD-10 gel filtration column (Pharmacia) pre-equilibrated with LCAT buffer. Specific activity was typically 225 cpm/ng of protein. Radiolabeled LCAT (final concentration of 3 g/ml) was incubated with rHDL (final apolipoprotein concentrations ranging between 4.4 and 140 g/ml) for 20 min, and free LCAT was separated from bound LCAT by filtration on a Centricon-100 filter (Amicon, Inc.).   (Fig. 1). g The phospholipid surface was calculated from the circumference of the disc, using the measured diameter minus 3 nm, multiplied by a disc height of 3.8 nm. The area covered by an ␣-helix containing 16 amino acids was estimated to be 4 nm 2 . Fig. 1 is a schematic representation of the predicted amphipathic helical regions in apoA-I, apoA-II, apoA-I(⌬(Arg 123 -Tyr 166 ),ٌA-II(Ser 12 -Ala 75 )), and apoA-I(⌬(Ala 190 -Gln 243 ),ٌA-II(Ser 12 -Gln 77 )). The predicted number of amphipathic helices, according to Brasseur et al. (12), was eight for apoA-I and for the two chimeras. Wild-type apoA-I and the apoA-I/apoA-II chimeras were expressed in the periplasmic space of E. coli cells and purified to homogeneity as evidenced by their migration as single bands on 10 -15% SDS-polyacrylamide gels (Fig.  2). The molecular masses of the apolipoproteins, calculated from a plot of the logarithm of the molecular masses of the standard proteins versus the migration distance, were 28.3 kDa for apoA-I and 30.4 and 29.8 kDa for the respective chimeras and thus were in agreement with those calculated on the basis of the respective amino acid compositions. Wild-type apoA-I migrated in the same position as plasma apoA-I. The identity of each band was confirmed by immunoblot analysis using polyclonal rabbit anti-human apoA-I antibodies (data not shown).

RESULTS
The ␣-helical contents of the apolipoproteins in aqueous solution, as determined by circular dichroism scanning, were 48% for apoA-I and 43 and 41% for the respective chimeras. The recovery of apolipoproteins in discoidal apolipoprotein-POPCcholesterol complexes after gel filtration was 90% for wild-type apoA-I and the chimeras. The Stokes radius of all rHDL was 4.8 nm (Fig. 3 and Table I). Cross-linking of the apolipoprotein molecules within all the discoidal apolipoprotein-POPC-cholesterol particles revealed 2 apolipoprotein molecules/particle ( Fig. 4 and Table I). The phospholipid surface was calculated from the circumference of the disc using the measured diameter minus 3 nm, i.e. 2 ϫ radius of an ␣-helix, multiplied by a disc height of 3.8 nm. The number of phospholipid molecules was calculated from the calculated surface divided by 0.45 nm 2 , the surface area/condensed phospholipid molecule. The calculated apolipoprotein/phospholipid molar ratios were 1:90, 1:83, and 1:89, respectively, and were in agreement with the ratios calculated on the basis of the composition of the rHDL.
The ␣-helical contents of the apoA-I proteins in discoidal apolipoprotein-POPC-cholesterol complexes were 74% for apoA-I and 75 and 72% for the respective chimeras. The number of ␣-helices, calculated from the ␣-helical content determined by circular dichroism scanning and from the protein length assuming a length of 22 amino acids/␣-helical repeat (2,12), was eight per apoA-I or per chimera molecule (Table I) and was thus in agreement with the predicted values (12). The fraction of the phospholipid surface that was covered by the ␣-helix was calculated as the total phospholipid surface in nm 2 divided by 4 nm 2 , the surface area of an ␣-helix, that contains 16 amino acids. The calculated fraction was 0.77 for apoA-I and the chimeras ( Table I).
Binding of apoA-I and the chimeras to POPC was associated with a shift of Trp fluorescence maxima to lower wavelengths from 337 to 332 nm for apoA-I (both for plasma apoA-I and recombinant apoA-I), from 339 to 334 nm for the central domain chimera, and from 337 to 333 nm for the carboxyl-terminal domain chimera (Table II), suggesting that lipid binding is associated with a translocation of Trp residues to a more apolar environment. The quenching parameters K SV and f a are summarized in Table II. For totally exposed Trp residues, in the absence of electrostatic or viscosity effects, K SV ϭ 12 M Ϫ1 and f a ϭ 1, whereas for totally protected Trp residues, K SV ϭ 0 and f a ϭ 0. The quenchable fluorescence of all apolipoproteins in the respective rHDL represented, on average, 60% of the total fluorescence (Table II).
Recombinant LCAT was obtained in serum-free conditioned medium of transfected 293 cells. The homogeneity of the recombinant LCAT preparation is illustrated in Fig. 5. LCAT activation by the discoidal apolipoprotein-POPC-cholesterol complexes obeyed Michaelis-Menten kinetics, as shown by linear Lineweaver-Burk plots of the inverse of the initial activation rate (1/V 0 ) versus the inverse of the cholesterol concentrations (1/[C]). The apparent kinetic parameters V max and K m and the V max /K m ratios for the different apolipoprotein-POPCcholesterol complexes are summarized in Table III. Exchange of the Arg 123 -Tyr 166 paired helix of apoA-I with the pair of helices of apoA-II reduced the LCAT activity of apoA-I 27-fold due to a reduction of V max and not of K m . In contrast, exchange of the Ala 190 -Gln 243 carboxyl-terminal domain helices of apoA-I with the pair of helices of apoA-II had no effect on the LCAT activity of apoA-I. Fig. 6 illustrates the binding of radiolabeled LCAT to rHDL of apoA-I and the apoA-I(⌬(Arg 123 -Tyr 166 ),ٌA-II(Ser 12 -Ala 75 )) chimera. Fifty % of maximal binding was obtained with 34 g/ml apoA-I and 27 g/ml apoA-I(⌬(Arg 123 -Tyr 166 ), ٌA-II(Ser 12 -Ala 75 )). DISCUSSION Reported differences in LCAT activity of apoA-I variants may be due to defective interaction with phospholipids, structural changes in rHDL, and/or deletion of functional domains. ) chimeras were produced. The extent of in vitro phospholipid binding of these chimeras was similar to that of apoA-I, as demonstrated by comparable disc formation after mixing the apolipoproteins and phospholipids at equal weight ratios. This was evidenced by a shift of the maximum Trp fluorescence to a shorter wavelength and by a decreased accessibility of the Trp residues to I Ϫ . The sizes of rHDL reconstituted with apoA-I and the chimeras were identical: the respective rHDL contained 2 apolipoprotein molecules/particle, and circular dichroism scanning revealed eight ␣-helices per intact apoA-I molecule and per chimera molecule, in agreement with the predicted numbers according to Brasseur et al. (12). The calculated apolipoprotein/phospholipid molar ratios of the different rHDL particles were very similar. Thus, substitution of the Ala 123 -Tyr 166 central or Ala 190 -Gln 243 carboxyl-terminal domain helices of apoA-I with the pair of helices of apoA-II did not affect the size and the composition of rHDL, and the conformation and helical distribution in the different apolipoproteins in these particles were very similar. Substitution of the carboxyl-terminal domain of apoA-I with the helices of apoA-II did not reduce LCAT activity, but substitution of the central domain resulted in a 27-fold reduction of LCAT activity, suggesting that the Ala 123 -Tyr 166 segment is critical for LCAT activation. Binding experiments revealed that the reduced LCAT activity of the apoA-I(⌬(Arg 123 -Tyr 166 ),ٌA-II(Ser 12 -Ala 75 )) chimera was not due to reduced binding of LCAT to rHDL. Based on data obtained with synthetic peptides, it was concluded that the LCAT-activating domain of apoA-I resides in a 22-mer tandem repeat located between residues 66 and 121 (35). This was further supported by the finding that monoclonal antibodies directed against an epitope that spanned residues 95-121 inhibited the LCAT activation with apoA-I (36). Binding of antibodies to an epitope in the amino-terminal domain of apoA-I may, however, induce conformational changes in the central domain of apoA-I that may be responsible for the reduction of LCAT activity (37). Deletion of the Leu 44 -Leu 126 amino-terminal domain of apoA-I indeed did not reduce its LCAT activity (16), suggesting that the amino-terminal domain of apoA-I is not critical for LCAT activation.
Using deletion mutants of apoA-I, Minnich et al. (38) found that deletion of the Met 148 -Gly 186 segment resulted in decreased LCAT activity, whereas Sorci-Thomas et al. (39) found that deletion of the Pro 143 -Ala 164 segment reduced LCAT activity. In previous studies, the conformation of deletion mutants in their respective rHDL was not investigated. However, it has been demonstrated that the decreased LCAT activity of deletion mutants of apoA-I may be due to differences in folding (decreased ␣-helical content) and/or organization of the apolipoproteins in rHDL (3 or 4 molecules/particle as compared with 2 for intact apoA-I) rather than to the removal of specific domains for LCAT activity (16). Indeed, it has been demonstrated that rHDL discs containing wild-type apoA-I may have discrete sizes, compositions (with 2, 3, or 4 protein molecules/ particle), and apoA-I conformations (with six, seven, or eight ␣-helices/apoA-I molecule in contact with lipid) and that differences in the apoA-I structure in these particles correlate with their ability to activate LCAT (40).
Chimeras in which ␣-helical segments of apoA-I are substituted with helical segments of apoA-II, which does not activate LCAT, in such a way that the average secondary structure of the apolipoprotein molecule as well as the organization of the apolipoprotein molecules in rHDL are not affected may be preferable reagents to address the function of a particular structural domain in LCAT activation. Indeed, rHDL containing apoA-II have a very low LCAT activity (41), and the addition of apoA-II together with apoA-I to liposomes reduces LCAT activity by 70% (42). Thus, substitution of sequences in apoA-I that are critical for the interaction with LCAT with sequences derived from apoA-II would result in decreased LCAT activity. Substitution of the carboxyl-terminal domain of apoA-I with helices of apoA-II was found not to affect LCAT activity, but substitution of the central domain resulted in a 25-fold reduction of LCAT activity, suggesting that the central domain (but not the carboxyl-terminal domain) is essential for LCAT activation.
The differences in LCAT activity were due to differences in apparent V max values, which reflect the activated enzyme concentration, and not to differences in apparent K m values, which reflect the affinity of LCAT for rHDL (43). Similar K m values are indeed in agreement with similar binding of LCAT to rHDL of apoA-I and the central domain chimera. Thus, the Ala 123 -  Tyr 166 segment of apoA-I appears to contain structures that are required for optimal phospholipid and cholesterol presentation to LCAT (44) that cannot be mimicked by the apoA-II segment. It is possible that substitution of the central domain affects the conformation of a hinged domain that is crucial for LCAT activation because antibodies that interfere with the mobility of a hinged domain in the central part of apoA-I inhibit LCAT activation (37). Previous data obtained with deletion mutants supported the existence of such a hinged domain overlapping either the Asn 102 -Lys 140 or Ala 124 -His 162 segment of apoA-I (16). Because conformational changes elsewhere in the molecule could not be excluded, these data were, however, not conclusive. The present study strongly suggests that this hinged domain most likely overlaps the Ala 124 -His 162 domain of apoA-I.
In conclusion, substitution of the central or carboxyl-terminal pair of helices of apoA-I with the helices of apoA-II does not affect its average structure in rHDL. Substitution of the central domain (but not the carboxyl-terminal domain) results in a significant reduction of the rate of LCAT activation, although the binding of LCAT to rHDL is not reduced.