Structure of High Density Lipoprotein THE IMMUNOLOGIC REACTIVITIES OF THE COOH-AND NH,-TERMINAL REGIONS OF APOLIPOPROTEIN A-I*

Only 5 to 10% of the apolipoprotein A-I (ApoA-I) of intact high density lipoprotein (HDL) is detectable by radioimmunoassay. In addition, when isolated ApoA-I is recombined with lipids in vitro, its immunologic reactivity is decreased by 30 to 95%. Thus, ApoA-I is less reactive immunologically in the presence of lipids. Our aim was to ascertain whether the COOH- or NH2-terminal regions of ApoA-I were equally reactive in intact HDL2. CNBr fragments of ApoA-I were produced by the method of Baker et al. (Baker, H.N., Jackson, R.L., and Gotto, A.M. (1973) Biochemistry 12, 3866-3871) and iodinated with lactoperoxidase. Double-antibody radioimmunoassays were set up using anti ApoA-I antisera and 125I-CNBr I (COOH-terminal region) or 125I-CNBr II (NH2-terminal). Both labels were bound by the antisera. Affinity columns were prepared by binding CNBr I or CNBr II to Sepharose 4B. Antibodies specific against CNBr I or CNBr II were isolated by means of these columns, suggesting that ApoA-I had at least two antigenic sites. In other assays using labeled fragments and anti ApoA-I antisera, 125I-CNBr I was displaced by CNBr I, ApoA-I , and HDL2 but not CNBr II. Conversely, 125I-CNBr II was displaced by CNBr II, ApoA-I, and HDL2 but not by CNBr I. Thus the assays were region-specific. The reactivities of isolated ApoA-I and the ApoA-I in intact HDL2-ApoA-I) were compared in these assays. On a molar basis, HDL2-ApoA-I was consistently more reactive (2- to 5-fold) in the 125I-CNBr I than in the 125I-CNBr II assays. The findings suggest (a) that the two terminal regions of ApoA-I are immunologically distinct, (b) that the two regions can be assayed independently of each other in intact HDL2, and (c) that the COOH-terminal region is more reactive immunologically than is the NH2-terminal. The results are compatible with a more "exposed" position for the COOH-terminal region on the surface of HDL2.

Only 5 to 10% of the apolipoprotein A-I (ApoA-I) of intact high density lipoprotein (HDL) is detectable by radioimmunoassay.
In addition, when isolated ApoA-I is recombined with lipids in vitro, its immunologic reactivity is decreased by 30 to 95%. Thus, ApoA-I is less reactive immunologically in the presence of lipids. Our aim was to ascertain whether the COOH-or NH,-terminal regions of ApoA-I were equally reactive in intact HDL,.
CNBr fragments of ApoA-I were produced by the method of Baker et al. (Baker, H. N., Jackson, R. L., and Gotto, A. M. (1973) Biochemistry 12, 3866-3871) and iodinated with lactoperoxidase. Double-antibody radioimmunoassays were set up using anti ApoA-I antisera and "'I-CNBr I (COOH-terminal region) or 'Y-CNBr II (NH,-terminal). Both labels were bound by the antisera. Affinity columns were prepared by binding CNBr I or CNBr II to Sepharose 4B. Antibodies specific against CNBr I or CNBr II were isolated by means of these columns, suggesting that ApoA-I had at least two antigenic sites. In other assays using labeled fragments and anti ApoA-I antisera, 'Y-CNBr I was displaced by CNBr I, ApoA-I, and HDL, but not CNBr II. Conversely, 'ZSI-CNBr II was displaced by CNBr II, ApoA-I, and HDL, but not by CNBr I. Thus the assays were region-specific.
The reactivities of isolated ApoA-I and the ApoA-I in intact HDL, (HDL,-ApoA-I) were compared in these assays. On a molar basis, HDL,-ApoA-I was consistently more reactive (2-to 5-fold) in the ""I-CNBr I than in the ""I-CNBr II assays.
The findings suggest (a) that the two terminal regions of ApoA-I are immunologically distinct, (b) that the two regions can be assayed independently of each other in intact HDL,, and (c) that the COOH-terminal region is more reactive immunologically than is the NH,-terminal. The results are compatible with a more "exposed" position for the COOH-terminal region on the surface of HDL,.
In the last few years, several models of high density lipoprotein structure have been proposed (l-3). All of these postulate that the apoproteins, ApoA-I' and ApoA-II,. are found on or near the surface of the HDL molecule.
However, recently published data show that only about 10% of the mass of ApoA-I, the major protein of HDL, can be detected in intact HDL by radioimmunoassay (4, 5). In addition, when delipidated ApoA-I or HDL protein are recombineu with egg lecithin vesicles or HDL lipid, and the recombined particles are assayed for their ApoA-I contents, far less ApoA-I is found than is known to be there (6). These findings suggest that not all of the ApoA-I molecules or all regions of the molecules may be immunologically reactive on the surface of intact HDL. We wished to ascertain whether certain regions of ApoA-I were more reactive than others. Specifically, we wished to explore the immunoreactivities of the COOH-and NH,-terminal regions of ApoA-I in intact HDL. To do this, it was necessary to ascertain whether the COOH-and NH,-terminal regions of the and yielded a single peak which contained CNBr II (Fig. 1C) (13, using crystalline bovine serum albumin standards; reaction tubes were extracted with diethyl ether before being analyzed in the spectrophotometer. (The Lowry/amino acid mass ratios for ApoA-I, ApoA-II, and ApoC, the proteins which comprised >95% of HDL proteins, were each >0.9 (la)*; therefore the Lowry procedure probably gives results for HDL protein mass which are accurate to within <lo%.) The ApoA-I content of Apo-HDL, was taken as 65% of Apo-HDL, mass. This was based on published data for HDL (4, 11, 19) and our own column chromatographic results with HDL,. 2

RESULTS
We have previously shown that '251-ApoA-I prepared by the chloramine-T method is a satisfactory label for radioimmunoassay purposes (4). 12SI-ApoA-I prepared by the lactoperoxidase method seems to be equally satisfactory in that, following repurification on Sephadex G-75, the label eluted as a single peak 2 to 3 ml earlier than chymotrypsinogen (molecular weight 25,000, Fig. 2A). This suggests that under assay conditions this label is in its monomeric form. Furthermore, >90% of '251-ApoA-I was precipitable by 10% trichloroacetic acid. A similar percentage was precipitable by antisera directed against ApoA-I (  (Table  II). Similar precipitabilities were obtained by both aggregated and monomeric labels. The 'Y-CNBr II purified by column chromatography on Sephadex G-50 in albumin-barbital or barbital-Triton yielded single peaks which had elution volumes within 2 to 3 ml of that of cytochrome c (Fig. 2), suggesting that the label was present as a monomer in both albumin-barbital and barbital-Triton assays. From 80 to 85% of this label was precipitated by 10% trichloroacetic acid and 70 to 76% by antisera directed against ApoA-I.
The specific radioactivities of '251-CNBr I and lz51-CNBr II ranged from 12 to 15 mCi/pg. Iodinations were carried out every 15 to 20 days.
labeled CNBr I and ApoA-I, but not Y-CNBr II. Further evidence of the immunologic distinctiveness of the two ends of the ApoA-I molecule was provided by using the labeled CNBr fragments in radioimmunoassays. Fig. 3A shows the displacement of 'Y-CNBr I by unlabeled CNBr I and ApoA-I. Over 97% of the added label was displaced by both unlabeled CNBr I and by ApoA-I. However, the unlabeled CNBr II fragment produced virtually no displacement of counts (ApoA-II and ApoC, too, were nonreactive). The converse situation is depicted in Fig. 3B, i.e. the displacement of "Y-CNBr II by unlabeled CNBr II. More than 90% of the added label was displaced in this assay. The curve produced by ApoA-I closely approximated that of the CNBr II curve, while CNBr I (and ApoA-II and ApoC) produced 'no displacement of counts. These CNBr I and CNBr II assays were run with antiserum R130-3. Similar results were obtained with antiserum R131-2. Furthermore, the assays exhibited the appropriate specificities whether run in albumin-barbital (as shown) or in barbital-Triton (not shown).

Structure of High Density Lipoprotein
The immunologic activity of HDL, was tested in these assays (Figs. 4 and 5). On a molar basis, the ApoA-I in intact HDL, (HDL,-ApoA-I) had about 10% of the activity of isolated ApoA-I in competing with '*SI-ApoA-I for reactive sites in anti ApoA-I antisera (4) (Fig. 4A). When the assay was repeated in barbital-Triton (Fig. 4B) instead of albumin-barbital, the reactivity of HDL,-ApoA-I was doubled (to -20% of the activity of ApoA-I). Where 'Y-CNBr I was used as the labeled moiety (Fig. 4C), displacement curves were produced by CNBr I (not shown), ApoA-I, and HDL,. HDL,-ApoA-I had 37% of the effectiveness of isolated ApoA-I in displacing 'Y-CNBr. With 'Y-CNBr II as the label, displacement curves were obtained by CNBr II, ApoA-I, and HDL,, but HDL,-ApoA-I was only 16% as effective as isolated ApoA-I in displacing the label (Fig. 40).
With 1251-CNBr II as the labeled moiety (Fig. 5B), the curves were farther apart; in fact, HDL,-ApoA-I appeared to be only 20% as effective as isolated ApoA-I in displacing 'Y-CNBr II. The experiments with R130-3 were repeated three times with two different CNBr fragment preparations with comparable results. To assess the effect of the state of aggregation of the CNBr I label, assays were performed in barbital-Triton Anti-CNBr II, antibody fraction bound to CNBr II affinity column; anti-CNBr I, antibody fraction bound to CNBr I affinity column. Each antibody was titrated two to three times.  Baker et al. (7,8) (Table I). Most important, there was virtually no cross-reactivity between the two fragments in radioimmunoassays (Fig. 3). Other apoproteins also did not react. Thus, the assays were specific. Reproducibility was adequate within each assay system (Figs. 3 and 5).
The immunologic activities of HDL,-ApoA-I relative to isolated ApoA-I were compared in these assays. HDL,-ApoA-I was consistently more reactive relative to isolated ApoA-I in   II (B) by curves produced with Y-CNBr I were much closer together than those the indicated proteins, using anti-ApoA-I antiserum R130.3 (diluted for '*4-CNBr II. Each point was run in triplicate; intraassay coefficient 1:500). Results are given in picomoles of isolated ApoA-I or HDL,. of variation was 5 + 2% (mean l 1 S.D.). Assays were repeated on ApoA-I and picomoles of the appropriate fragments. The displacement three other occasions (interassay coefficient of variation was 9 * 3%).
the CNBr I than in the CNBr II assays (Figs. 4 and 5). These may be that the COOH-terminal region is more reactive findings imply that the conformations of Y-CNBr I and of the because it is more "accessible" to antibody, perhaps because it COOH-terminal of ApoA-I in HDL, are more nearly alike than is less involved in lipid-protein or protein-protein interactions. are the conformations of '*"I-CNBr II and the NH,-terminal Perhaps other factors may be responsible but it is clear that the region. The reasons for this are not known, but one explanation findings are not due to the differing states of aggregation of by guest on March 17, 2020 http://www.jbc.org/ Downloaded from of High Density Lipoprotein '261-CNBr I and "'I-CNBr II because the differences between HDL,-ApoA-I and isolated ApoA-I, although diminished in magnitude, persisted even in assays run in barbital-Triton buffer, where both 'z61-CNBr I and "'1-CNBr II appeared to be monomeric.
The ability of Triton X-100 to increase the reactivity of HDL,-ApoA-I in each of the assays is an interesting finding, which suggests that the detergent, either by binding to  or by some other unknown mechanism, perturbs the structure of HDL so as to make the antigenic sites of APO-I more "accessible." It is anticipated that the use of region-specific radioimmunoassays during controlled perturbations of HDL structure may yield still more information on the structure of HDL.