Immunochemical characterization of two antigenic sites on human apolipoprotein A-I; localization and lipid modulation of these epitopes

Biochimica

160

et Biophysics

Acta 959 (1988) 160-168 Elsevier

BBA 52768

Immunochemical characterization of two antigenic sites on human apolipoprotein A-I; localization and lipid modulation of these epitopes F. Pio a, H. De Loof

b, N. Vu Dac a, V. Clavey and M. Rosseneu b

’ SERLIA,

Department of Lipids-Lipoproteins and b Department of Clinical Biochemistry, (Received

Key words:

Monoclonal

a, J.C. Fruchart

a

and INSERM U 279, Institut Pasteur, Lille (France) Academisch Ziekenhuis St. Jan, Bruges (Belgium) 30 September

1987)

antibody; HDL; Dimyristoylphosphatidylcholine Competitive inhibition; Antigenic site; (Human

recombination; plasma)

Apolipoprotein

A-I;

Two monoclonal antibodies, Al7 and A30, were raised against human apolipoprotein A-I (apo A-I). They were studied by competitive inhibition of ‘25I-labeled HDL, with HDL subfractions, delipidated apo A-I, and complexes of dimyristoylphosphatidylcholine (DMPC) containing apo A-I and apo A-II. Immunoblotting located the Al7 antibody on CNBr fragment 4 of apo A-I and the A30 antibody on CNBr fragment 1. The Al7 antigenic determinant was expressed identically in all HDL subclasses, on delipidated apo A-I as well as all on the DMPC-apo A-I and DMPC-apo A-I/ape A-II complexes. In contrast, the apparent affinity constant of the A30 antibody for delipidated apo A-I was about 30-times less than for HDL, or for apo A-I/ape A-II-phospholipid complexes. These data suggest that the association of apo A-I with phospholipids improves the reactivity of the A30 monoclonal antibody towards apo A-I, and that this antigenic determinant has a different conformation in delipidated apo A-I compared to apo A-I complexed with phospholipids. Turbidimetric and fluorescence experiments monitoring the phospholipid-apo A-I association in the presence and in the absence of the Al7 and A30 antibodies were consistent with the competition experiments carried out by solid phase radioimmunoassay (RIA). After reaction of apo A-I with the A30 antibody, we observed an enhancement of the degradation kinetics of large multilamellar vesicles (LMV), while the Al7 antibody did not have a significant effect. Calcein leakage experiments carried out below the transition temperature of DPPC showed an enhancement of the degradation kinetics with both monoclonal antibodies, while the phase-transition release was independent of the reaction of apo A-I with the monoclonal antibodies. These data therefore suggest the existence of at least two different types of epitope on apo A-I, which might account for the differences in immunological reactivity of apo A-I that is either delipidated or present on HDL.

Abbreviations: BSA, bovine serum albumin; HDL, high-density lipoproteins; DMPC, dimyristoylphosphatidylcholine; DPPC, dipahnitoylphosphatidylcholine; Lh4V, large multilamellar vesicles; RIA, radioimmunoassay; ELISA, enzymelinked immunosorbent assay; Mab, monoclonal antibody; apo A-I, apolipoprotein A-I; apo A-II, apolipoprotein A-II. Correspondence: M. Rosseneu, chemistry, A.Z. St. Jan, Brugges OOOS-2760/88/$03.50

Department of Clinical 8000, Belgium.

0 1988 Elsevier Science Publishers

Bio-

Introduction Apolipoprotein A-I (apo A-I) is the major protein component of the human high-density lipoproteins (HDL). Its 243-residue sequence [l] contains several repeating sequences which can form amphipathic helices [2] and are involved in the

B.V. (Biomedical

Division)

161

association of apo A-I with phospholipids [3]. Apo A-I is an activator of lecithin-cholesterol acyltransferase [4], and some putative amphipathic domains of apo A-I have been proposed as activators [S]. Studies of the immunogenicity of apo A-I using monoclonal and polyclonal antisera have suggested an imrnunochemical heterogeneity of apo A-I in human HDL [6]. In view of the compositional differences between HDL subfractions, the heterogeneity was attributed to a variable expression of the apo A-I epitopes in the lipoprotein particles and to a masking of some antigenic sites by the lipids [6,7]. In order to check whether lipid-protein interactions might account for some of these observations, series of monoclonal antibodies were raised against human apo A-I. Among the 14 clones reacting against human apo A-I, two clones, reacting with different epitopes, were selected. The reactivity of these monoclonal antibodies with HDL subclasses and with lipid-apolipoprotein complexes was thoroughly studied. Since the amphipathic helices are not uniformly distributed along the apo A-I sequence, one can expect that the expression of the two epitopes, located in different regions of apo A-I, will differ in terms of their modulation by the apo A-I-lipid interaction. Their reactivity might also be affected by protein-protein interactions between apo A-I and apo A-II. These parameters were investigated both by conventional displacement measurements using a solid-phase radioimmunoassay (RIA) and by physicochemical techniques, including turbidimettic and fluorescence measurements. These techniques enabled the characterization of the antigen-antibody-lipid system in a liquid phase. These data should contribute to the further understanding of the structure-function relationship in apo A-I, especially concerning the modulation of the antigenic structure of the protein by lipids, and are also relevant to the metabolic function of apo A-I within HDL subclasses. Methods Preparation of lipoproteins and apolipoproteins.

Lipoproteins

were isolated

from

fresh human

plasma by sequential ultracentrifugation at respective densities of 1.063-1.12 for HDL, and 1.12-1.21 g/ml for HDL,. The fractions were dialysed against 0.15 M NaCl and subsequently delipidated with diethyl ether/ethanol (2 : 1, v/v). Apo A-I and apo A-II were isolated from delipidated HDL, by ion-exchange chromatography on a Mono-Q ion-exchange column in 6 M urea, using a fast pressure liquid chromatography (FPLC) system (Pharmacia, Uppsala, Sweden) [g]. Lipoproteins were quantitated by the method of Lowry et al. [9] and apo A-I was assayed by an enzyme-linked immunosorbent assay (ELISA) [lo]. Production of monoclonal antibodies. Male balb/c mice were immunized intraperitonealy with 500 pg of HDL emulsified in 300 ~1 of complete Freund’s adjuvant. The same dose was injected 21 days later in incomplete Freund’s adjuvant and a further injection was given 2 months later. 1 week following the final intraperitoneal injection, the animals received an intravenous booster of 300 pg HDL,. 3 days later, the spleens were excised and splenic cells were fused with myeloma sp,O [ll]. Hybrid cells were screened for the secretion of specific antibody by ELISA [12] on microtiter plates coated with either apo A-I or HDL. Antibodies were obtained from positive clones by intraperitoneal injection of (2-4) - lo6 hybridoma cells. Purification of antibodies was performed by ammonium sulfate precipitation and affinity chromatography on a protein A-Sepharose column (Pharmacia, Uppsala, Sweden). Radioiodination. HDL was labeled with Na’251 using the iodine monochloride method [13]. The specific activity of the ‘251-labeled HDL ranged from 500 to 700 cpm/ng protein in the different experiments. Binding and competitive displacement of 12jIlabeled HDL. Microtiter plates (Falcon) were

coated overnight at 25’C with 100 ~1 of the monoclonal antibody at a concentration of 20 pg/ml of protein in phosphate-buffered saline (pH 7.5) and subsequently blocked with BSA at a concentration of 10 g/l for 1 h. 100 ~1 of 1251labeled HDL,, containing a total of 25 ng protein, in the phosphate buffer containing 10 g BSA/l, was added to the wells, together with 50 ~1 of competitors (DMPC-apolipoprotein complex, HDL or apo A-I) at increasing amounts, ranging

162

from 0.01 to 1000 pg/ml of apo A-I. Incubation was performed overnight at room temperature. Results are expressed as log B/B, = log( B/B,)/ (1 - B/B,), where B is the concentration of competitor-bound ‘251-labeled HDL and BO is the concentration of free 12’1-labeled HDL. The concentration of competitor was expressed in pg/ml of apo A-I assayed by ELISA using a polyclonal antiserum [lo]. Apparent affinity constants for each antibody with the different competitors were calculated according to the formula of Muller [14]: K,=

1 (C-

L)(1-1.5b+0.562)

where C is the molar concentration of competing antigen required for 50% inhibition of ‘251-labeled HDL, binding to the antibody, L is the molar concentration of ‘251-labeled HDL, added, and b the maximum binding of ‘251-labeled HDL, in the absence of competitor. Cyanogen bromide cleavage of apo A-I. CNBr fragments were prepared as described by Baker et al. [15]. Apo A-I was dissolved in 70% formic acid and CNBr was added at a 500: 1 molar ratio relative to the methionine content of apo A-I. The reaction was allowed to proceed for 24 h at 25 o C, in the dark, under N,. The CNBr-cleaved fragments were then lyophilized and redissolved in a Tris-HCl buffer comprising 0.01 M Tris-HCl (pH 8)/7 M urea. Gel electrophoresis, isofocusing and immunoblotting. SDS-polyacrylamide gel electrophoresis was performed according to Weber and Osbom [16], isofocusing as described by Menzel et al. [17], immunoblotting according to Towbin et al. [18], and two-dimensional gel electrophoresis according to O’Farell [19]. Preparation of liposomes. Liposomes were prepared with DMPC and DPPC (Sigma). The lipids were dissolved in chloroform, evaporated and sonicated at 37 o C for 30 min in 0.01 M Tris-HCl buffer/O.1 M NaCl (pH 8) using a Branson sonifier equipped with a microtip probe. For the in vitro complex formation, the unilamellar DMPC liposomes were first incubated with apo A-I at a 2 : 1 (w/w) ratio for 3 h at 24” C, followed by the addition of apo A-II at molar ratios of, respectively, 2 : 1 and 0.5 : 1, and further incubated un-

der the same conditions. Complex isolation and characterization were performed as described previously [20,21]. For preparation of large multilamellar vesicles (LMV), DMPC was dissolved in CHCl,, the solvent was evaporated to dryness under nitrogen and, finally this was lyophilized. 2 mM phosphate buffer (pH 7.4)/0.2 M NaCl/l mM EDTA/30 mM NaN, was added to the dried lipid, and the mixture was vortexed for 3 min. Turbidity measurements. The turbidity of the DMPC-LMV dispersions (0.15 M) was monitored at 325 nm with a Jasco UVIDEC-610 double beam spectrophotometer, equipped with a water bath (Julabo) able to generate temperature gradients. In all experiments, apo A-I was first incubated with a monoclonal antibody (Mab) directed either against apo A-I (sample) or against apo C-III (reference) at a 5-fold molar excess of Mab to apo A-I, for 2 h at 37” C prior to the measurements. The liposomes were then carefully mixed with the apolipoprotein-antibody complex at 17 o C, a temperature at which the reassembly of apo A-I with lipids occurs only very slowly [20]. Temperature scans between 17 and 27” C were carried out at a rate of 0.25 C”/min. The use of this temperature gradient allowed the observation of small differences between the kinetics of degradation of the liposomes in the two cuvettes and the simultaneous measurement of reference and sample using the same liposome preparation and thus avoiding problems due to mixing of the solutions and to sedimentation of the large liposomes. Calcein-release assay. The release of calcein encapsulated in the vesicles, induced by the interaction of DPPC with apo A-I, was followed using the techniques described by Weinstein et al. [22,23]. The release of the encapsulated dye was accompanied by an increase of the fluorescence intensity outside the liposomes. The calcein (Fluka AG) was purified on a Sephadex LH-20 column to remove hydrophobic impurities [22]. Encapsulation of the dye into the liposomes was achieved by vortexing and subsequent sonication of the DPPC in a 250 mM calcein solution three times for 7 min each at 43” C under N,. The nonencapsulated calcein was removed by two consecutive passages through a PD-10 gel-filtration column. The fluorescence was measured on an Aminco SP-500 spectrofluorimeter at an emission wavelength of

163

520 nm using and excitation wavelength of 490 nm. In order to monitor complete calcein release, 10 ~1 of 10% Triton X-100 was added at the end of each experiment. Two types of measurement were performed: isothermal kinetic measurements, where the leakage of calcein was monitored as a function of time after addition of apo A-I, and phase-transition release experiments, where the percentages of calcine release were measured at the passage through the transition temperature of DPPC. All experiments were carried out with apo A-I that had been preincubated with a monoclonal antibody raised either against apo A-I (sample) or apo C-III (reference). The isothermal kinetic measurements were carried out below the DPPC transition temperature (31S°C), to obtain a slow release of the dye. For the phase-transition release, the apo A-I-Mab complex was mixed with the vesicles at 4O C and rapidly heated up to 43°C in about 2 min in the cuvette of the spectrofluorimeter [22]. Rli!SUltS

Specificity of the two monoclonal antibodies A single fusion yielded 60 clones that secreted anti-apo A-I antibodies. Fourteen were screened for their reaction against apo A-I by ELISA using plates coated with apo A-I and HDL. By immunodiffusion, according to Ouchterlony, only the A30 antibody produced IgG,, immunoglobulins, while all other clones were of the IgG, subclass. Apolipoprotein specificity was tested against the apolipoprotein components of HDL by immunoblotting after SDS-polyacrylamide gel electrophoresis. The Al7 and A30 antibodies were specific only for apo A-I. Location of the antigenic determinants on the CNBr fragments of apo A-I The fragments obtained by CNBr cleavage were analysed by isofocusing and two-dimensional gel electrophoresis. A comparison of the charge and relative molecular weights with previous data [24] was used to identify the CNBr apo A-I fragments. The cleavage was incomplete, as some apolipoprotein A-I remained intact and as several bands with different molecular weights were

Fig. 1. Two-dimensional gel electrophoresis of apo A-I cleaved with CNBr: Isofocusing was used for separation in the first dimension (shown also in Fig. 2) of 100 pg of cleaved apolipoprotein. Separation in the second dimension used a slab gel of 25% acrylamide with 18% glycerol. Gel was stained with Coomassie blue G 250 and destained with methanol/acetic acid/H,0 (5 : 7.8 : 87.5, v/v). Data indicate charge heterogeneity of CNBr-cleaved apo A-I and incompletely cleaved apo A-I.

observed in cleaved apo A-I (Fig. l), possibly due to incomplete cleavage at the methionine-serine bond between CNBr fragments 1 and 2 [24]. The patterns coincided with previously published data [24], thus enabling the identification of native apo A-I and CNBr fragments 1, 1-2, l-2-3, 2-3-4 and 3-4 according to the nomenclature of Brewer et al, [l]. On Fig. 1, spots of identical molecular weight on two-dimensional gel electrophoresis indicate charge heterogeneity for both apo A-I and CNBr fragments 1, l-2, l-2-3 and 2-3-4 [25]. Immunoblotting, carried out after isofocusing of apo A-I CNBr-cleaved fragments (Fig. 2), shows that some fragments, as well as the uncleaved apo A-I isoforms, reacted with the two antibodies. This suggests that the charge heterogeneity of apo A-I isoforms does not coincide with the location of the antigenic sites reacting with our monoclonal antibodies. According to the electrophoretic and. immunological data, the A30 antibody recognizes the CNBR-cleaved fragments 1, l-2 and l-2-3 as well as native apo A-I, but does not react with the CNBr-cleaved fragments 2-3-4 and 3-4, suggest-

164

1

2,x,4

e

L apoA-l

01

0 Fig. 2. Immunoblotting of apo A-I fragments cleaved with CNBr. A sample of apo A-I (20 pg) or cleaved apo A-I (100 pg) was applied to the polyacrylamide gel containing 7 M urea and ampholytes (pH 3.5-10). The immunoreactive fragments are identified by arrows, according to Brewer [l]: A, native apo A-I; B, immunoblotting of CNBr fragments with the Al7 antibody at 20 ag/ml; C, immunoblotting with the A30 antibody at the same concentration; D, isofocusing of 100 pg of the CNBr fragments analysed also by two-dimensional gel elcctrophoresis. The fragments are indicated on the right-hand side of the figure.

ing that this particular epitope is located on the CNBr fragment 1. The antibody Al7 recognizes apo A-I and fragments 2-3-4 and 3-4 but not the CNBr fragments 1, l-2 or l-2-3, indicating that the Al7 epitope is located on CNBr fragment 4. Immunoreactivity

of apo A-I in HDL subfractions

In order to determine whether the expression of apo A-I epitopes was different in the HDL, and HDL, subfractions and in delipidated apo A-I, the inhibition of the two monoclonal antibodies was tested with several competitors (Fig. 3). With the Al7 antibody, the protein concentration required to achieve 50% displacement of 1251labeled HDL, was identical, whether apo A-I was present on HDL, or HDL,, suggesting a similar reactivity of this antibody with the two competitors. However, a 25-fold higher concentration of delipidated apo A-I was needed for a 50% inhibition of HDL, binding to the A30 antibody. The 50% inhibition of HDL, binding to the A30 anti-

1 pg/ml

10

100

lOcx3

OfapoA-I

Fig. 3. Competitive inhibition of the binding of the A30 ) and Al7 (- - -) antibodies with different compe(titors. (A) DMPC-apolipoprotein complexes and native apo A-I: apo A-I (0); DMPC-apo A-I (v); DMPC-apo A-I/ape A-II (2:l) (0) and (1:2) (+) (M/M). (B) HDL, (0); HDL, (v), apo A-I (0). Competitors were expressed as pg/ml of apo A-I, quantified by an ELISA assay. The displacement curves are depicted as Y = log( B/Be) as a function of X = log apo A-I concentration expressed in gg/ml. The slopes did not differ significantly from those obtained with native apo A-I, and the correlation coefficients were around 0.97-0.99.

body required an even higher (400-fold) protein concentration. These data indicate that the reactivity of the A30 antibody is different with delipidated apo A-I compared to apo A-I present in the HDL subclasses. The apparent affinity constants for apo A-I, listed in Table II, are higher with the Al7 than with the A30 Mab. The affinity constant of the A30 antibody is 40-times lower for delipidated apo A-I than for apo A-I in HDL,. Immunoreactivity complexes

of apo A-I in lipid-apolipoprotein

The influence of the apo A-I-lipid association on the immunoreactivity of the apo A-I epitopes was studied with apo A-I-DMPC complexes. The effect of apo A-II was also investigated with DMPC-apo A-I/A-II complexes at apo A-I/A-II molar ratios of 2 and 0.5. As seen in Fig. 3B and Table I, the protein concentration of competitor, expressed as apo A-I concentration required to

165 TABLE

I HDL

A-I A-I

01

SUBCOM-

Apparent Concentration affinity of competitor required for 50% of constants antibody binding (10*/M)

Apo A-I HDL, HDL, DMPO-apo DMPC-apo /A-II 2:l 1:2

I

A

EXPRESSION OF APO A-I EPITOPES IN CLASSES AND IN APOLIPOPROTEIN-DMPC PLEXES

( -b rl

w

Al7

A30

Al7

A30

0.45 0.58 0.58 0.4

300 0.7 12.5 10

14.3 15 15 14.3

0.04 7 1.5 1.2

0.9 0.9

15 3.8

5 5

0.9 4.2

/

-01

t

.

a

I

L

--_L 25

achieve 50% displacement of labeled HDL,, as well as the apparent affinity constant, are not significantly different for all the competitors tested with the Al7 antibody. These data agree with the results obtained with HDL, and show that the reactivity of this antibody is comparable with apo A-I and with the various complexes. In contrast, the 50% inhibition concentration and the affinity constant for A30 binding vary greatly with the competitor tested. The concentration of native apo A-I required is 30-times higher than that of apo A-I complexed with DMPC; moreover it is 25and 80-times higher than that of the mixed DMPC-apo A-I/A-II complexes, at molar ratios of apo-I to apo-II of 2 and 0.5, respectively. These results show that the reactivity of the A30 epitope varies to a large extent with the competitors, either HDL subfractions, or apo A-I-DMPC complexes (Table I). A comparison of the reactivity of the two DMPC-apo A-I/A-II mixed complexes suggests that apo A-II increases the imrnunoreactivity of apo A-I with the A30 Mab, possibly through protein-protein interactions affecting the conformation of this particular epitope. Turbidity measurements

The results of the temperature scans, shown in Fig. 4, monitoring the degradation of the liposomes, indicate that the turbidity difference occurs around the DMPC transition temperature.

Temperature

(-‘Cl

Fig. 4. Differential turbidity measurements monitoring the degradation of DMPC-LMV by apo A-I complexed with the Al7 (A) and the A30 (B) antibody. In this experiment, DMPCLMV were mixed with apo A-I complexed with either the Al7 or A30 antibody, and compared to the same vesicles mixed with apo A-I plus anti-apo C-III. In (A), an enhancement of the degradation of the DMPC-LMV was observed with A30 Mab-apo A-I complex while in (B), no effect was induced by the Al7 Mab-apo A-I complex.

Fig. 4A depicts a differential measurement of the turbidity of apo A-I complexed either with the A30 antibody (sample) or with anti-apo C-III (reference), and subsequently incubated separately with the same preparation of DMPC liposomes. The relative turbidity decrease suggests faster kinetics of association of the apo A-I-A30 Mab complex with DMPC compared to uncomplexed apo A-I. Fig. 4B shows a similar experiment, where apo A-I is complexed with the Al7 antibody, and no significant difference was observed between the kinetics of degradation of the liposomes by apo A-I with or without reaction with the Al7 Mab. The small baseline drift measured around the transition temperature was due to extremely small temperature differences between the two cuvettes. Moreover, the passage through the phase-transition temperature did not affect the relative turbidity. These data therefore support the results obtained by competitive immunoassay, suggesting that the lipid-‘apo A-I association affects only slightly the binding to the Al7 antibody, whereas the A30 antibody is very sensitive to the state of relipidation of apo A-I.

166

01

b‘

II

2730 33 36

39

42 temperature

L5 O%

0

20 5

0

10

15

'20 hlI"f

time

Fig. 5. Calcein-release experiments monitoring the fluorescence intensity increase at 520 nm on mixing DPPC-calcein vesicles with apo A-I complexed with the Al7 and A30 antibodies. Isothermal kinetic measurements at 31.5“ C for (0) DPPC+ (apo A-I + Al7 Mab); ( X) DPPC+ (apo A-I +A30 Mab); (0) DPPC + (apo A-I + anti-apo C-III Mab).

Fluorescence measurements The kinetics of calcein release from the DPPC vesicles at 31.5”C (Fig. 5) show a slow release of the dye induced by apo A-I complexed with the Al7 and A30 monoclonal antibodies as well as with the anti-apo C-III. Under these experimental conditions, apo A-I associates only with the external layer of the phospholipid vesicle, but does not penetrate deeply enough to induce a breakdown of the vesicles which would lead to massive calcein release [22,23]. Compared to the total release induced by Triton X-100, the percentage of release by apo A-I amounted to only 8.8%. The two

TABLE

II

REACTION PARAMETERS LEASE EXPERIMENTS The kinetics 31.5 o c.

experiment

FOR

was carried

THE

CALCEIN-RE-

out at a temperature

Release (%) apo A-I

Kinetics h/2

apo A-I + AI7 Mab

apo A-I + A30 Mab

7.4 11.5

7.3 10.5

42.2

43

expt.

Onin)

% release Temperature % release

9.2 8.8 jump expt. 42.2

of

Fig. 6. Calcein-release experiments monitoring the fluorescence intensity increase at 520 nm on mixing DPPC-calcein vesicles with apo A-I complexed with the Al7 and A30 antibodies. Phase-transition release of calcein was in a temperature scan ), DPPC + (apo A-I + Al7 from 4 to 43OC in 2 min. (Mab); (......) DPPC+(apo A-I+A30 Mab); (---) DPPC + (apo A-I + anti-apo C-III Mab).

monoclonal antibodies enhanced slightly the kinetics and the amplitude of the calcein release (Table II). The phase-transition release (Fig. 6) observed through the passage at 41 o C, the transition temperature of DPPC, induced a 4-fold larger calcein release. This reaction was very rapid, and the association of apo A-I with the monoclonal antibodies did not significantly affect either the amplitude or the kinetics of the calcein release. Discussion

Immunoblotting of apo A-I CNBr-cleaved fragments located the epitope for the A30 Mab on the CNBr fragment 1 (residues l-85) and the epitope for the Al7 Mab on the CNBr fragment 4 (residues 147-243). The characterization of the two monoclonal antibodies by competitive RIA showed that the Al7 antibody has the same reactivity whatever the competitor studied. In contrast, the reactivity of the A30 antibody is largely dependent upon the competitor. The strong difference observed with the latter antibody between native apo A-I and apo A-I present in HDL subclasses suggests that the maximal reactivity of this antibody involves an apo A-I-lipid association. Our reassembly studies indicate that phospholipids increased the reactivity of delipidated apo A-I to a level close to that of apo A-I in HDL. Incorporation of apo A-II into the liposome-apolipoprotein

167

complexes further increased slightly the reactivity of apo A-I with the A30 antibody. These data suggest that phospholipids are the major determinants which optimize the conformation of this epitope to enable the A30 antibody to recognize apo A-I. No such conformational change seems to occur for the Al7 antigenic determinant, or at least it is not sufficient to modify the immunoreactivity of apo A-I when associated either with lipids or with apo A-II. In agreement with the results of Marcel et al. [7], we observed that the apparent affinity constant of the Al7 Mab, reacting both with native and relipidated apo A-I, was higher than that of the A30 Mab whose epitope is dependent upon the lipid environment. Secondary structure predictions and analysis of the repetitive segments in human apo A-I [26] have suggested that the NH,-terminal domain (CNBr fragment 1) of apo A-I contains fewer amphipathic helical segments than the COOHterminal domain (CNBr fragment 4). Upon binding with phospholipids, the increase in helicity might therefore be more pronounced in CNBr fragment 1, accounting for the different exposure of the A30 epitope. The tmbidity measurements with apo A-I complexed with the A30 antibody showed an unexpected enhancement of the degradation of the liposomes by apo A-I. As native apo A-I self-associates readily in aqueous solutions, and has a limited content of tertiary structure, binding to the A30 Mab prior to its association with lipids might induce a conformational change of an apo A-I domain [27], resulting in an enhancement of the liposome breakdown. The inverse case, an inhibition of liposome degradation, was also observed e.g., with antibodies AlO, A05 (data not shown). This can, however, be easily explained by steric hindrance of the antibody for binding of lipids to their epitopes. The Cal&n-release experiments carried out both below and through the phase-transition temperature of DPPC indicate that the interaction of apo A-I with the monoclonal antibodies does not significantly modify the interaction of apo A-I with the phospholipids at low temperature where micellar complexes are generated [22]. Since several lipid-binding domains exist on apo A-I [5], the presence of an IgG molecule does impair the

interaction of the protein with the lipids and even slightly enhances the kinetics of interaction. At the passage through the phase transition, the calcein release occurs very rapidly due to the insertion of apo A-I into the bilayer, enhanced by the presence of crystal-lattice defects inside the bilayer structure [22,23]. The association of apo A-I with the monoclonal antibody does not affect this process to a significant extent. In contrast, the breakdown of the vesicles, monitored by turbidimetric measurements, is a slower process, scanned at 0.25 Co/mm, compared to 20 C”/min for the calcein leakage. This process is sensitive to the conformation of apo A-I, and can be enhanced by the interaction with the A30 Mab. Our results therefore support the concept of the conformational adaptability of apo A-I and of its immunological heterogeneity in different particles reported previously [ll]. Difficulties encountered in the quantification of apo A-I might arise from the fact that some antigenic determinants adopt a different conformation as a function of their lipid and protein environment, such as the epitope reacting with the A30 antibody, whereas others, e.g., that for the Al7 Mab, do not follow the same behavior. Further studies with monoclonal antibodies reacting with different epitopes, and a more precise mapping of the antigenic sites of apo A-I, should prove valuable for the elucidation of the relationship between lipid and antibody-binding domains on apo A-I. Acknowledgements

This work was supported by E.E.C. grant BCR 292 and INSERM-C.N.A.M.T.S. grant 863427 E. H.D.L. is a recipient of an Institut voor Wetenschappelijk Onderzoek in Nijverheid en Landbouw Fellowship. References 1 Brewer, H.B., Jr., Fairwell, T., Larue, A., Ronan, R., Houser, A. and Bronzert, T.J. (1978) B&hem. Biophys. Res. Commun. 80,623-640. 2 Fitch, W.M. (1977) Genetics 86, 623-644. 3 Sparrow, J.T. and Gotto, A.M., Jr. (1981) Crit. Rev. Biothem. 12, 87-107. 4 Glomset, J.A. (1968) J. Lipid Res. 9, 155-167. 5 De Loof, H., Rosseneu, M., Brarseur, R. and Ruysschaert, J.M. (1987) B&him. Biophys. Acta 911,45-52.

168

6 Marcel, Y.L., Weech, P.K., Milthorp, P., Terce, F., Vesina, C. and Milne, W.R. (1985) Prog. Lipid Res. 23, 169-195. 7 Marcel, Y.L., Jewer, D., Vesina, C., Milthorp, P. and Weech, P.K. (1987) J. Lipid Res. 28, 768-776. 8 Mezdour, H., Clavey, V., Kora, I., Koffigan, L., Barkia, A. and Fruchart, J.C. (1987) J. Chromatogr. 414, 35-45. 9 Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J. (1951) J. Biol. Chem. 193, 265-275. 10 Bury, J. and Rosseneu, M. (1985) CIin. Chem. 31,247-251. 11 Curtiss, L.K. and Edington, T.S. (1982) J. Biol. Chem. 257, 15213-15221. 12 Piechaczyk, M., Chardes, T., Cot, M.C., Pau, B. and Bastide, J.M. (1985) Hybridoma 4, 361-367. 13 Bilheimer, D.W., Eisenberg, S. and Levy, RI. (1972) Biochim. Biophys. Acta 260, 212-221. 14 Muller, R. (1980) J. Immunol. Methods 34, 345-352. 15 Baker, H.N., Jackson, R.L. and Gotto, A.M., Jr. (1973) Biochemistry 12, 3866-3871. 16 Weber, K. and Osbom, M. (1969) J. Biol. Chem. 244, 4406-4412. 17 Menzel, H.J., Assman, G., Rall, SC., Weissgraber, K.H. and Mahley, R.W. (1984) J. Biol. Chem. 259, 3070-3076. 18 Towbin, H., Staehelin, T. and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354.

19 G’Farell, P.H. (1975) J. Biol. Chem. 250, 4007-4021. 20 Van Tomout, P., Vercaemst, R., Caster, H., Taveime, M. and Rosseneu, M. (1979) Arch. Int. Physiol. Biochim. 87, 1011-1012. 21 Van Tomout, P., Vercaemst, R., Lievens, M.J., Rosseneu, M. and Assman, G. (1980) Biochim. Biophys. Acta 601, 509-523. 22 Weinstein, J.N., Blumenthal, R. and Klausner, R.D. (1986) Methods Enzymol. 128, 657-668. 23 Klausner, R.D., Blumenthal, R., KIausner, R.D., Innerarity, T. and Weinstein, J.N. (1985) J. Biol. Chem. 260, 13719-13727. 24 Weech, P.K., Milne, R.W., Milthorp, P. and Marcel, Y.L. (1985) B&him. Biophys. Acta 835, 390-401. 25 Ghiselli, G., Schaefer, E.J., Light, J.A. and Brewer, H.B., Jr. (1983) J. Lipid Res. 24, 731-736. 26 Bogusky, M.S., Freeman, M., Elshourbagy, N.A., Taylor, J.M. and Gordon, J.I. (1986) J. Lipid Res. 27, 1011-1035. 27 Colman, P.M., Laver, W.G., Varghese, J.N., Baker, A.T., Tulloch, P.A., Air, G.M. and Webster, R.G. (1987) Nature 326, 358-363.