Influence of apolipoprotein A-I and apolipoprotein A-II availability on nascent HDL heterogeneity.

It is important to understand HDL heterogeneity because various subspecies possess different functionalities. To understand the origins of HDL heterogeneity arising from the existence of particles containing only apoA-I (LpA-I) and particles containing both apoA-I and apoA-II (LpA-I+A-II), we compared the abilities of both proteins to promote ABCA1-mediated efflux of cholesterol from HepG2 cells and form nascent HDL particles. When added separately, exogenous apoA-I and apoA-II were equally effective in promoting cholesterol efflux, although the resultant LpA-I and LpA-II particles had different sizes. When apoA-I and apoA-II were mixed together at initial molar ratios ranging from 1:1 to 16:1 to generate nascent LpA-I+A-II HDL particles, the particle size distribution altered, and the two proteins were incorporated into the nascent HDL in proportion to their initial ratio. Both proteins formed nascent HDL particles with equal efficiency, and the relative amounts of apoA-I and apoA-II incorporation were driven by mass action. The ratio of lipid-free apoA-I and apoA-II available at the surface of ABCA1-expressing cells is a major factor in determining the contents of these proteins in nascent HDL. Manipulation of this ratio provides a means of altering the relative distribution of LpA-I and LpA-I+A-II HDL particles.

are important for the effi ciency of the RCT pathway and the inverse relationship with the incidence of atherosclerosis. Plasma HDL comprises a heterogeneous assembly of different size particles, with variations in lipid and protein content ( 6,7 ). Apolipoprotein (apo)A-I and apoA-II are the major protein constituents, comprising 70% and 20% of total HDL protein, respectively ( 8 ). ApoA-I has received the most attention to this point, and its structure-function relationship in preventing atherosclerosis is relatively well understood ( 7 ). In comparison, less attention has been paid to apoA-II. Human apoA-II is an amphipathic protein synthesized and secreted by the liver and small intestine as an 82-amino acid proprotein that is proteolytically cleaved to the mature 77-amino acid protein ( 9 ). The concentration of apoA-II in plasma is 31 ± 6 mg/dl, much lower than the apoA-I level (123 ± 28 mg/dl) ( 10 ). ApoA-II circulates as a 17.4 kDa homodimer formed by two mature apoA-II molecules linked by a disulphide bond at residue 6 ( 11 ).
HDL heterogeneity originates at the point of biogenesis at which apoA-I and ABCA1 interact to create discoidal nascent particles; the relative available lipid/apoA-I ratio controls the size distribution ( 12 ). The heterogeneity of mature spherical HDL is also affected by remodeling by lipases and lipid transfer proteins in the plasma ( 6 ). It is important to understand HDL heterogeneity because different HDL subspecies exhibit different functionalities ( 13 ). The existence of LpA-I and LpA-I+A-II particles is another aspect of HDL heterogeneity that infl uences HDL functionality. For instance, LpA-I+A-II HDL is less effective than LpA-I HDL at promoting selective cholesteryl ester uptake via SR-B1 ( 14 ) and the transfer of cholesteryl ester by CETP ( 15 ). Also, compared with LpA-I particles, LpA-I+A-II particles are less effective at promoting cholesterol esterifi cation via LCAT ( 16 ), although the abilities of the two types of HDL to promote cellular cholesterol effl ux are similar ( 17,18 ).
To better understand the origins of nascent HDL heterogeneity arising from the coexistence of LpA-I and LpA-I+A-II particles, we examined the particles formed when Abstract It is important to understand HDL heterogeneity because various subspecies possess different functionalities. To understand the origins of HDL heterogeneity arising from the existence of particles containing only apoA-I (LpA-I) and particles containing both apoA-I and apoA-II (LpA-I+A-II), we compared the abilities of both proteins to promote ABCA1-mediated effl ux of cholesterol from HepG2 cells and form nascent HDL particles. When added separately, exogenous apoA-I and apoA-II were equally effective in promoting cholesterol effl ux, although the resultant LpA-I and LpA-II particles had different sizes. When apoA-I and apoA-II were mixed together at initial molar ratios ranging from 1:1 to 16:1 to generate nascent LpA-I+A-II HDL particles, the particle size distribution altered, and the two proteins were incorporated into the nascent HDL in proportion to their initial ratio. Both proteins formed nascent HDL particles with equal effi ciency, and the relative amounts of apoA-I and apoA-II incorporation were driven by mass action. The ratio of lipid-free apoA-I and apoA-II available at the surface of ABCA1-expressing cells is a major factor in determining the contents of these proteins in nascent HDL. Manipulation of this ratio provides a means of altering the relative distribution of LpA-I and LpA-I+A-II HDL particles. -Alexander, E. T., and M. C. Phillips High density lipoprotein (HDL) cholesterol levels in plasma are inversely associated with the risk of cardiovascular disease ( 1 ). This antiatherogenic behavior is thought to arise in part from the central role of HDL in reverse cholesterol transport (RCT), the pathway by which excess cholesterol is removed from peripheral tissues and transported to the liver for excretion from the body (2)(3)(4)(5). The current consensus is that the structure and composition of HDL particles, not just the plasma HDL cholesterol levels, concentrated to 0.8-1.0 ml using Amicon Ultracel-10K centrifugal fi lter units, and stored at 4°C for further analysis. The gel fi ltration column was washed between runs with 30% isopropanol and 1 M NaOH, as recommended by the manufacturer. The following standards (Sigma-Aldrich, St. Louis, MO) were used to calibrate the column: cytidine, (total volume, V t ); thyroglobulin, 17.0 nm; apoferritin, 12.2 nm; lactic dehydrogenase, 8.16 nm; BSA, 7.1 nm; carbonic anhydrase, 4.4 nm; blue dextran, (void volume, V o ). The particles sizes corresponding to the various fractions were determined by comparing their K av values with those of the above proteins ( 20 ). K av was calculated using the following equation: K av = (V e -V 0 ) / (V t -V 0 ), where V e is the elution volume of the HDL fraction. The apparent particle size [hydrodynamic diameter (d) in nanometers] was derived from the following equation: log 10 d = Ϫ 1.103 K av + 1.372. An increase in V e of 1 ml corresponded to an increase in d of 0.7 nm.

HDL separation into LpA-I and LpA-I+A-II
As described earlier (24)(25)(26), chromatography on activated thiol-sepharose beads was used to separate LpA-I and LpA-I+A-II particles. In brief, an HDL sample was reduced in TBS with 10 mM dithiothreitol (DTT) and then dialyzed into TBS containing 1 mM DTT overnight. Thiopropyl-sepharose beads (0.25 g) were stirred into deionized water for 15 min and then washed extensively. The washed beads were resuspended in 1 ml of TBS and divided into two equal aliquots. One aliquot was stirred with the reduced HDL, and the other was transferred to a 1 × 6 cm chromatography column. After 2 h, the HDL slurry mixture was added to the column and allowed to settle. The unbound LpA-I was eluted with TBS and collected. The column was then washed with 30 ml of TBS. After this treatment, LpA-I+A-II was eluted by the addition of 20 mM DTT to the TBS (10 ml). The distributions of apoA-I and apoA-II were confi rmed by scintillation counting to detect both 14 C-apoA-I and 3 H-apoA-II.

RESULTS
Infl uence of ABCA1 upregulation on cholesterol effl ux from HepG2 cells HepG2 cells were plated, allowed to reach confl uency, and labeled with [1,[2][3] H(N)]cholesterol, and then half was treated with 9-cis -retinoic acid and 22-hydroxy-cholesterol for 18 h to upregulate ABCA1. Fresh medium without an apolipoprotein acceptor was then added to both sets of cells and incubated for 4 h. The effl ux of cholesterol from nonupregulated cells was 1.9 ± 0.6% over 4 h, whereas the effl ux from upregulated cells was 3.2 ± 0.3% over 4 h. Effl ux with the latter cells was carried out for 24 h to generate suffi cient HDL to allow FPLC analysis ( Fig. 1 ). These HepG2 cells secreted nascent HDL particles that contained endogenously synthesized apoA-I and apoA-II as reported previously (27)(28)(29)(30). The effl ux medium was also analyzed by immunoblotting; the ratio of apoA-I to apoA-II was approximately 4/1 w/w (data not shown). The nascent HDL particles produced by HepG2 cells are heterogeneous ( 30 ). As shown in Fig. 1 , when ABCA1 is upregulated, the cholesterol released from the cell is incorporated into two populations of nascent HDL particles. The larger particles elute from the Superdex 200 column with a peak maximum of V e = 63 ml, and the smaller both apoA-I and apoA-II are available to ABCA1 on the surface of HepG2 cells, which were used because the liver is the primary source of nascent HDL in vivo ( 19 ).

Cell cholesterol effl ux assay
HepG2 cells were seeded in 24-well plates at a 1/10 dilution from a confl uent culture, allowed to attach overnight, labeled with 0.5 µCi/ml [1,[2][3] H(N)]cholesterol overnight, and upregulated for ABCA1 expression for 16-20 h as described above. The cells were then exposed to 10 µg/ml of either human apoA-I or apoA-II in MEM/50 µg/ml gentamicin for 4 h. The medium was collected and fi ltered through a 96-well fi lter plate (EMD Millipore, Billerica, MA); the radioactivity in a 100 µl aliquot of each sample was determined using a scintillation counter ( 23 ). Cell lipids were extracted with hexane-isopropanol (3:2, v/v), the solvent was evaporated, the lipids were dissolved in Scintiverse, and the radioactivity was determined with a scintillation counter. The percentage of cellular cholesterol released to apoA-I was calculated by dividing the [ 3 H] counts in the medium by the sum of [ 3 H] counts in the medium and cells and multiplying by 100.

Gel fi ltration chromatography of nascent HDL
A 1 ml aliquot of the 20× concentrated cell medium containing nascent HDL was resolved into 1 ml fractions on a calibrated HiLoad 16/60 Superdex 200 gel fi ltration column (GE Healthcare, Mickleton, NJ) using tris-buffered saline (TBS), pH 7.4, as the mobile phase. Each fraction was combined with 5 ml of Scin-tiVerse BD cocktail (Fisher Scientifi c, Pittsburg, PA) and read in a scintillation counter; when dual labeling was used, [ 3 H] and [ 14 C] counts were adjusted for the energy emission spectra overlap. Alternatively, the fractions containing the larger (>9 nm) or the smaller (<9 nm) nascent HDL particles were combined, from the peak V e values were 12 and 8 nm, which agrees with a prior report for HepG2 cells treated similarly ( 33 ). Effl ux to apoA-II gave rise to a 14 C-cholesterol FPLC profi le with one broad peak at V e = 67 ml ( Fig. 4 ). The equivalent 3 H-apoA-II profi le had peaks with V e values of 67, 71, and 81 ml, corresponding to LpA-II particle diameters of 11, 10, and 7 nm. particles elute at V e = 77 ml; these peak positions correspond to particle hydrodynamic diameters of 13 and 8 nm, respectively.

Cholesterol effl ux from HepG2 cells in the presence of exogenous apoA-I and apoA-II
We performed comparative experiments to assess the abilities of exogenous lipid-free apoA-I (molecular weight 28016) and apoA-II (molecular weight in the dimerized state 17414) to mediate effl ux of cholesterol from HepG2 cells in which ABCA1 was upregulated to enhance production of nascent HDL particles. ApoA-I and apoA-II were added to the media at a concentration of 2.5 µg/ml and incubated with the cells for 2, 4, and 6 h. There were no differences between apoA-I and apoA-II in the rate of cholesterol effl ux over the 6 h incubation period ( Fig. 2A ). The initial rate of cholesterol effl ux over the fi rst 4 h to apoA-I was 2.9 ± 0.15% per h and 3.2 ± 0.15% per h to apoA-II. To address the effect of protein concentration, either apoA-I or apoA-II was added to the effl ux media in the range of 2.5-20 µg/ml. Fig. 2B shows that apoA-I and apoA-II were similarly effi cient at promoting cholesterol effl ux with V max (percentage cholesterol effl ux over 4 h) values of 9.8 ± 1.1 and 9.5 ± 1.3 respectively, and K m (micrograms of apo per milliliter) values of 1.9 ± 1.0 and 1.5 ± 0.9, respectively. The catalytic effi ciencies (V max / K m ) were similar to values for ABCA1-mediated cholesterol effl ux in other cell types ( 31,32 ).

Comparison of nascent HDL formed by apoA-I and apoA-II
Although apoA-I and apoA-II promote similar cholesterol effl ux from HepG2 cells, it was important to compare the nascent HDL particles formed by the two proteins. To this end, conditioned effl ux media containing either apoA-I or apoA-II were analyzed by gel fi ltration chromatography ( Figs. 3 and 4 ). FPLC traces of both 14 C-apoA-I and 3 H-cholesterol ( Fig. 3 ) showed two HDL peaks with maxima at V e = 66 and 76 ml, respectively. The hydrodynamic diameters of these LpA-I HDL particles deduced   experiment using media containing 5 µg/ml each of unlabeled apoA-I and apoA-II and cells that had been labeled with 3 H-cholesterol showed one peak with a maximum V e at ‫ف‬ 66 ml and a noticeable shoulder centered at 76 ml ( Fig. 5B ). The hydrodynamic diameter of the primary HDL peak at V e = 66 ml, as indicated by all three FPLC profi les, was 12 nm. The hydrodynamic diameters of the particles in the secondary HDL peaks were 8 and 7 nm. Samples of nascent HDL formed with apoA-I, apoA-II, or a mixture of the two proteins that were purifi ed from conditioned medium were analyzed by negative stain electron microscopy and shown to contain discoidal HDL particles that were apparent in the electron micrographs as characteristic rouleaux (data not shown).
The presence of both apoA-I and apoA-II in the effl ux media resulted in a shift in the size distribution of the nascent HDL particles, refl ecting formation of HDL species containing both apoA-I and apoA-II (LpA-I+A-II) (see Figs. 3, 4, and 5A ). A similar effect on HDL size occurs in mice transgenic for both human apoA-I and apoA-II ( 34 ). To investigate this phenomenon in more detail, the molar ratio of 14 C-apoA-I to 3 H-apoA-II incubated with the HepG2 cells was varied from 1:1 to 16:1. The distributions of the two proteins between the various HDL species are summarized in Table 1 . It is apparent that apoA-I and apoA-II were incorporated into the nascent HDL particles in proportion to their levels in the original mixture incubated with the HepG2 cells. As shown in Figs. 3 and 4 , the size distributions of nascent LpA-I and LpA-II HDL particles were different, and as expected, the size distribution of the particles created from a mixture of apoA-I and apoA-II refl ected the presence of both proteins. Thus, as shown in Fig. 6 , the peak distribution of apoA-II in LpA-II particles was at V e = 71 ml ( Fig. 6B ), and the proportion of apoA-II eluting at this position was progressively decreased as apoA-II was mixed with increasing amounts of apoA-I ( Fig. 6C, D ). The addition of apoA-I led to formation of LpA-I+A-II particles that eluted at V e = 67 ml, causing the maximal distribution of apoA-II to shift to this position ( Fig. 6C, D ), which corresponds to the V e of nascent LpA-I particles ( Fig. 6A ).
Given the extensive overlap in gel fi ltration FPLC profi les of LpA-I and LpA-II nascent HDL separated on the basis of particle size, it was diffi cult to resolve any LpA-I+A-II particles that formed at various initial molar ratios of

Nascent HDL produced by mixtures of apoA-I and apoA-II
To determine what types of HDL particles are generated when apoA-I and apoA-II are present in equal amounts, effl ux medium with 5 µg/ml of both apoA-I and apoA-II was analyzed by gel fi ltration chromatography ( Fig. 5 ). The 14 C-apoA-I profi le ( Fig. 5A ) had one well-defi ned peak with a maximum V e at ‫ف‬ 66 ml and a small peak centered around 75 ml, which preceded the lipid-free apoA-I peak at ‫ف‬ 81 ml. The corresponding 3 H-apoA-II profile had one large, broad peak with a maximum V e at ‫ف‬ 66 ml and a second, smaller peak with a maximum at 80 ml. The gel fi ltration profi le from a separate effl ux a Initial ratio of 14 C-labeled apoA-I to 3 H-labeled apoA-II in the effl ux media applied to upregulated HepG2 cells.
b Ratio of 14 C-apoA-I to 3 H-apoA-II in 11 individual HDL fractions (elution volume 60-71 ml) of FPLC profi le (e.g., Fig. 5A ). Average of two separate experiments, with the exception of 0.6 and 1 ratios, which were performed once each. Mean ± SD (n = 11-22). HepG2 cells were prepared as described in Fig. 3 . ABCA1-mediated cholesterol effl ux was induced by adding 10 µg/ml of apoA-II, and the conditioned medium was analyzed as described in Fig. 3 . ApoA-II (open circles) cholesterol (fi lled diamonds). particles containing very different relative amounts of the two proteins.

DISCUSSION
Plasma HDL consists of a diverse population of particles that have signifi cant differences in composition and size. It is important to understand this heterogeneity because different HDL subspecies exhibit different biological activities in vivo, and understanding their biogenesis is the fi rst step in understanding their functionalities ( 6,13 ) and identifying subspecies with desired cardioprotective properties. Nascent HDL particle heterogeneity is infl uenced by the ratio of cell lipids made available by ABCA1 activity to available apoA-I ( 12 ) and by the composition of the available cell lipids ( 35 ). Another potential source of nascent HDL heterogeneity is the composition of the available apolipoproteins during HDL biogenesis. To better understand the origins of nascent HDL heterogeneity arising from the existence of LpA-I and LpA-I+A-II particles, we examined the particles formed when both apoA-I and apoA-II are available to ABCA1 on the surface of HepG2 cells.
Before analyzing the nascent HDL particles produced by apoA-I and apoA-II, we fi rst quantifi ed the cholesterol effl ux from HepG2 cells elicited by the addition of exogenous apoA-I and apoA-II. There were no differences in the rate of cholesterol effl ux over time ( Fig. 2A ) or with the effect of protein concentration ( Fig. 2B ) between apoA-I and apoA-II. Given the increased hydrophobicity of apoA-II ( 36,37 ) and the somewhat enhanced ability of apoA-II to solubilize dimyristoyl phosphatidylcholine vesicles compared with apoA-I ( 38 ), we had anticipated that apoA-II may be a better acceptor of ABCA1-mediated cholesterol effl ux from HepG2 cells. However, despite the differences in physical properties, apoA-I and apoA-II promote cholesterol effl ux equally effi ciently.
Although apoA-I and apoA-II promoted ABCA1-mediated cholesterol effl ux similarly, the resulting LpA-I and LpA-II nascent HDL particles had somewhat different size distributions ( Figs. 3 and 4 ). Presumably, this variation is a consequence of the two proteins of different size requiring different interfacial areas to be optimally packed and cover the phospholipid acyl chains at the edge of the discoidal HDL particles ( 39,40 ). Both proteins adopt a belt-like molecular arrangement around the edge of the segment of phospholipid bilayer in such particles ( 39 ). When both apoA-I and apoA-II were present, the distribution of nascent HDL subspecies produced was altered, consistent with a mixture of LpA-I and LpA-I+A-II particles being produced. It is apparent that both proteins are incorporated into nascent HDL in proportion to their availability as lipid-free proteins at the surface of the ABCA1-expressing HepG2 cells ( Table 1 ). ABCA1 activity creates two types of apolipoprotein binding sites at the cell surface ( 41 ). Direct apolipoprotein/ABCA1 interaction creates a low capacity site that serves a regulatory role, and apolipoprotein/membrane lipid interactions create a much higher capacity site that is involved in the assembly of nascent apoA-I and apoA-II. To solve this problem, the nascent HDL was separated into LpA-I and LpA-I+A-II fractions by covalent chromatography using thiopropyl-sepharose. Analysis of three separate samples of nascent HDL containing an average apoA-I-to-apoA-II molar ratio of 2:1 demonstrated that 84 ± 5% of the total apoA-I resided on HDL particles that did not bind to the column. These particles contained an average apoA-I-to-apoA-II dimer molar ratio of 5 to 1. In contrast, the 16 ± 5% of apoA-I in HDL particles that bound to the thiopropyl-sepharose column and were subsequently eluted with DTT were present in apoA-II-enriched particles (1/4 mol/mol apoA-I/apoA-II). It is apparent that a nascent HDL preparation with an average apoA-I/apoA-II mol ratio of 2/1 comprises individual LpA-I/LpA-I+A-II HDL distributions and to potentially generating HDL populations with enhanced cardioprotective characteristics.
HDL particles. The latter site is an exovesiculated domain of plasma membrane lipids to which apoA-I and apoA-II can bind ( 42 ). The phospholipid bilayer in such a domain is destabilized by the penetration of apolipoprotein amphipathic ␣ -helices ( 22 ), leading to solubilization and formation of discoidal nascent HDL particles. When both apoA-I and apoA-II are present, the two proteins bind simultaneously and become incorporated into the HDL particles formed. The relative degree of binding of apoA-I and apoA-II to the membrane lipid surface is driven by mass action effects, which explains why the degree of incorporation of the two proteins into HDL particles is in proportion to their availability (on a molar basis). Consequently, the apoA-I/apoA-II ratio in nascent HDL particles at the moment of formation is controlled by the relative concentrations of the two lipid-free proteins at the HepG2 cell surface where ABCA1 is active. The relative apoA-I/ apoA-II content of these particles can then modifi ed by subsequent HDL remodeling events due to the activity of plasma factors such as LCAT ( 43 ). In our experiments, the lipid-free apoA-I and apoA-II were added exogenously to the exterior of the HepG2 cells. However, the process of mass-action driven incorporation of the two proteins into nascent HDL particles is of physiological relevance because HepG2 cells ( 29,(44)(45)(46)(47)(48)(49)(50) and hepatocytes ( 51,52 ) secrete signifi cant proportions of apoA-I and apoA-II as lipid-free proproteins that are rapidly converted to the lipid-free mature proteins.
Although the overall degree of incorporation of apoA-I and apoA-II into nascent HDL particles when both proteins are present is driven by mass action effects, the apoA-I/apoA-II ratio in individual HDL particles can diverge signifi cantly from the population average. Thus, HDL fractions containing apoA-I/apoA-II molecular ratios of 5/1 and 1/4 were separated by thiol-affi nity chromatography from a nascent HDL preparation with an average apoA-I/apoA-II molar ratio of 2/1. It follows that a given size of nascent HDL particle can contain different ratios of apoA-I and apoA-II. Consistent with this fi nding, alterations of the apoA-I/apoA-II ratio in either discoidal or spherical HDL particles can have little effect on particle size ( 15,(53)(54)(55). Furthermore, Segrest and colleagues recently isolated individual HDL subspecies using immunoaffi nity chromatography and identifi ed seven populations with different stoichiometries of apoA-I and apoA-II, many of which were similar in size ( 40 ).
In summary, the present comparison of nascent HDL formation with apoA-I and apoA-II revealed the following fi ndings: i ) ApoA-I and apoA-II promoted similar levels of ABCA1-mediated cholesterol effl ux, but the resulting nascent HDL particles were of different sizes; ii ) both proteins formed nascent HDL with equal effi ciency, and the relative amounts of apoA-I and apoA-II incorporation were driven by mass action; iii ) LpA-I+A-II HDL particles of a given size can contain different ratios of apoA-I and apoA-II; and iv ) the ratio of available apoA-I and apoA-II at the cell surface is a major factor in determining the contents of these proteins in nascent HDL. Manipulation of this ratio may provide a novel approach to altering the