ABCA1-Derived Nascent High Density Lipoprotein-Apo AI and Lipids Metabolically Segregate

Abstract

Objective

Reverse cholesterol transport (RCT) comprises cholesterol efflux from ABCA1-expressing macrophages to apo AI giving nascent high density lipoprotein (nHDL), esterification of nHDL-free cholesterol (FC), selective hepatic extraction of HDL-lipids, and hepatic conversion of HDL-cholesterol to bile salts, which are excreted. We tested this model by identifying the fates of nHDL-[³H]FC, [¹⁴C]PL, and [¹²⁵I]apo AI in serum in vitro and in vivo.

Approach and Results

During in vitro incubation of human serum, nHDL-[³H]FC and [¹⁴C]PL rapidly transfer to HDL and low density lipoproteins (LDL; t_1/2 = 2–7 min) while nHDL-[¹²⁵I]apo AI transfers solely to HDL (t_1/2<10 min) and to the lipid-free form (t_1/2 >480 min). Following injection into mice nHDL-[³H]FC and [¹⁴C]PL rapidly transfer to liver (t_1/2 ~ 2–3 min) whereas apo AI clears with t_1/2 = ~460 min. The plasma nHDL-[³H]FC esterification rate is slow (0.46%/h) compared to hepatic uptake. Phospholipid transfer protein enhances nHDL-[¹⁴C]PL but not nHDL-[³H]FC transfer to cultured Huh7 hepatocytes.

Conclusions

nHDL-FC, PL, and apo AI enter different pathways in vivo. Most nHDL-[³H]FC and [¹⁴C]PL are rapidly extracted by the liver via scavenger receptor class B member 1 and spontaneous transfer; hepatic PL uptake is promoted by phospholipid transfer protein. nHDL-[¹²⁵I]apo AI transfers to HDL and to the lipid-free form that can be recycled to nHDL formation. Cholesterol esterification by LCAT is a minor process in nHDL metabolism. These findings could guide the design of therapies that better mobilize peripheral tissue-FC to hepatic disposal.

Graphic Abstract

Visual Overview – An online visual overview is available for this article.

Atherosclerosis and its major sequela, cardiovascular disease, are major causes of mortality and morbidity in the United States.¹ Atherosclerosis, which is characterized by the accumulation of cholesterol in macrophages within the subendothelial space of the arterial wall, is suppressed by reverse cholesterol transport (RCT).^2–4 According to one model,^3–5 RCT is a three-step process whereby arterial macrophage-cholesterol transfers to the liver for disposal. The initial RCT step is the formation of nascent high density lipoproteins (nHDL) via the interaction of apo AI with the ABCA1 lipid transporter; the nHDL so-formed contains mainly apo AI, free cholesterol (FC), and phospholipid (PL).^6,7 The second step has both molecular and macromolecular consequences, respectively, esterification of nHDL-FC to cholesteryl ester (CE) by plasma lecithin:cholesterol acyltransferase (LCAT), and concurrent conversion of nHDL from a disc to a sphere with a central CE core. In the third step—cholesterol disposal—hepatic scavenger receptor class B member 1 (SR-B1) selectively extracts lipids from HDL via a nibbling mechanism in which apo AI and apo AII are excluded from uptake and HDL is reduced to a remnant while releasing lipid-free apo AI.⁸ Although cited to near-consensus,⁴ this RCT model has never been rigorously validated. Herein we provide data showing that the major components of ABCA1-derived HDL—FC, PL, and apo AI—segregate in serum in vitro and enter different pathways for clearance in vivo. Importantly, our results reveal that a major portion of the rapid hepatic nHDL-FC and PL clearance is independent of the reaction catalyzed by LCAT. These data are important in the context of the current segue from “HDL-C is the good cholesterol” to “HDL functionality is at least equally important.”^4,9–13

Materials and Methods

Materials and Methods are available in the online-only Data Supplement

Results

SEC Analysis of nHDL

nHDL was produced by incubation of apo AI with ABCA1-expressing BHK cells with and without methyl β-cyclodextrin (CDX)-mediated FC-loading. The media from each incubation were collected, concentrated, and analyzed. After the cells were FC-loaded with 0, 0.06, and 0.26 mM FC (0, 2.5, and 5 mM CDX respectively), the resulting cellular FC contents were 29 ± 2, 42 ± 6, and 52 ± 6 µg/mg cell protein. The size exclusion chromatography (SEC) profiles of the media (Figure 1) revealed two well-resolved peaks for protein absorbance corresponding to medium and small nHDL, nHDL-M and nHDL-S, respectively, which were collected and analyzed by SEC (Figure 1 A–C, chromatograms b, c). Based on past in vitro and in vivo studies^6,14 showing a large nHDL is formed from FC-loaded cells, we collected a third larger (L), less prominent, and earlier-eluting nHDL-L (Figure 1 A–C, chromatogram a). SEC analysis of the pooled fractions for nHDL-L, -M and -S showed that in the absence of CDX-mediated FC-loading, only two nHDL were formed (Figure 1 A, chromatograms b, c). However, the SEC profile of nHDL from CDX-mediated FC-loaded cells revealed three fractions (Figure 1 B, C, chromatograms a–c). The first two, nHDL-M and S, were similar to those found without CDX-mediated FC-loading; the third, larger nHDL-L, was unique to nHDL formed from CDX/FC-loaded cells. Based on SEC and comparison with the elution volumes of standard proteins, the respective diameters of nHDL-L, M, and S were 14.5 nm, 12.2 nm and 8.8 nm; the corresponding molecular weights were 510, 262, and 96 kDa. According to non-denaturing PAGE (Figure 1 D), the molecular weights were 482, 298 and 91 kDa respectively. Of note, while not detected by SEC, a minor amount of nHDL-L, detected by non-denaturing PAGE, was formed in cells without FC loading (Lane L, No CDX). As expected and shown in the SDS-PAGE (Figure 1 E), apo AI is the dominant protein in all three nHDL fractions but minor bands in the 60 – 200 kD range are also detected. The remainder of our studies were performed with the most abundant fraction, nHDL-M, which for brevity will from hereon be called nHDL. nHDL used in our study is, according to SEC, stable; repeat analytical SEC runs 1 and 3 months apart were identical.

Figure 1.

SEC Isolation of nHDL. nHDL was collected from the media after apo AI incubation of BHK cells over expressing ABCA1 with no FC-loading (A, 0 CDX) and with FC-loading with 2.5 mM (B) and 5.0 mM CDX (C) and analyzed by SEC. Grey-filled plots are the A280 SEC of the media; red bars denote the fractions from media that were collected and pooled as large (L), medium (M) and small (S) nHDL. Analytical A280 SEC of the pooled fractions L, M and S are shown in the chromatograms labeled a, b and c respectively. Non-denaturing PAGE (D) and SDS-PAGE (E) of the L, M and S pooled fractions from each of the three cell cholesterol loading conditions as labeled. Molecular weight standards are in the left lanes, and apo AI and BSA are as labelled.

nHDL Composition

The trends in nHDL composition as a function of particle size were similar for all three FC-loading conditions. PL/apo AI, FC/apo AI and FC/PL molar ratios and mol% FC in each particle increased with nHDL particle size (Table 1). The PC/SM ratios were 4.3 ± 0.2, 5.6 ± 0.2, and 7.0 ± 0.9 for the large, medium, and small nHDL respectively; the PC/SM ratio for the BHK-ABCA1 cells from which the nHDL were formed was 7.4 ± 0.2. The PC/SM ratios of the large and medium nHDL were significantly less than those of the small nHDL and the cells, p< 0.005 for all of these comparisons.

Table 1.

Nascent HDL Molar Compositions

CDX*	nHDL†	PL/apo AI mol/mol	FC/apo AI mol/mol	FC / PL mol/mol	mol% FC	mol% PL
0.0	Large	43.2 ± 1.8	76.7 ± 2.5	1.77 ± 0.05	64.0	36.0
0.0	Medium	34.2 ± 1.5	54.6 ± 2.3	1.60 ± 0.03	61.5	38.5
0.0	Small	17.9 ± 0.6	16.1 ± 0.5	0.90 ± 0.03	47.3	52.7
2.5	Large	39.4 ± 1.7	66.1 ± 1.0	1.68 ± 0.07	62.7	37.3
2.5	Medium	33.7 ± 1.4	49.6 ± 2.0	1.47 ± 0.00	59.5	40.5
2.5	Small	17.7 ± 0.4	13.7 ± 0.4	0.77 ± 0.02	43.6	56.4
5.0	Large	48.6 ± 0.8	83.8 ± 1.4	1.72 ± 0.01	63.3	36.7
5.0	Medium	37.6 ± 1.0	56.3 ± 1.6	1.50 ± 0.02	60.0	40.0
5.0	Small	16.2 ± 1.0	13.3 ± 1.0	0.82 ± 0.02	45.0	55.0

^*Concentration, mM

^†nHDL sizes are from Figure 1.

nHDL-[³H]FC Rapidly Transfers to HDL and Low Density Lipoproteins (LDL) In Vitro

nHDL-[³H]FC was incubated with serum under several conditions of time and temperature, and analyzed by SEC (Figure 2). nHDL-[³H]FC elutes as a single SEC peak (Figure 2 A). When added to HDL, nHDL-[³H]FC rapidly transfers to HDL. After rapid mixing at 4 °C followed by immediate injection into the FPLC, SEC shows all the [³H]FC is already associated with HDL (Figure 2 B). When incubated with whole serum, nHDL-[³H]FC transfers to all lipoproteins. Initially, most is associated with HDL, but lesser amounts co-elute with HDL and more are associated with the apo B-containing lipoproteins with increasing incubation time at 37° C. (Figure 2 C). The kinetics of transfer of nHDL-[³H]FC to other lipoproteins were too fast to quantify by SEC. [³H]FC accumulated in LDL with a t_1/2 = 15 min (Figure 2D). Parallel studies of the transfer of native human HDL-[³H]FC to serum gave remarkably similar results with the majority of HDL-[³H]FC transferring to LDL within 2 h (Supplementary Figure I). Given time limitations of SEC, the kinetics of nHDL-[³H]FC transfer to LDL were determined by heparin-Mn⁺² precipitation of LDL^15,16 after its incubation with nHDL-[³H]FC at 37° C for various times; the halftime for this process was 2.1 ± 0.5 min (Figure 2E). Notably, transfer of nHDL- and HDL-FC to LDL occurred over a time frame in which less than 5% of nHDL- and HDL-[³H]FC are converted to CE by mouse plasma LCAT activity (Supplementary Figure II).

Figure 2.

Transfer of nHDL-[3H]FC to Serum Lipoproteins. nHDL-[3H]FC was prepared from BHK-ABCA1 cells labeled with [3H]FC as in Figure 1. nHDL-[3H]FC were tested under several conditions and analyzed by SEC, which was monitored by β-counting of the collected fractions. A) nHDL-[3H]FC; B) nHDL-[3H]FC incubated with HDL at 4 °C and immediately analyzed by SEC (0 min). C) nHDL-[3H]FC incubated with serum at 4 °C or 37oC for various times as labeled and analyzed by SEC. D) Kinetics of [3H]FC accumulation in LDL (fractions encompassed by grey box in Panel C). E) The kinetics of nHDL-[3H]FC transfer to isolated LDL determined by heparin-Mn+2 precipitation of LDL.

nHDL-[¹⁴C]PL Transfers to HDL and LDL In Vitro

Experiments similar to those of Figure 2 were conducted with nHDL-[¹⁴C]PL. Like nHDL-[³H]FC, the elution volume of nHDL-[¹⁴C]PL was distinct from those of the major native serum lipoproteins (Figure 3 A). SEC analysis following incubation of nHDL-[¹⁴C]PL with serum revealed a rapid transfer of [¹⁴C]PL to LDL (peak elution volume 22 – 23 mL) but not to very low density lipoproteins (VLDL; Figure 3, peak elution volume 16 mL); the former had a transfer halftime = ~6.5 min and reached ~33% of total DPM at t ≥ 30 min (Figure 3 C). In whole serum, the halftime for nHDL-[¹⁴C]PL transfer to HDL appeared to be on a time scale similar to that for transfer to LDL but could not be quantified because the elution volumes of nHDL-[¹⁴C]PL and HDL overlap (Figure 3). The halftime for HDL-[¹⁴C]PL transfer to serum lipoproteins was slightly faster, t_1/2 ≤5 min (data not shown). Halftimes for nHDL-[¹⁴C]PL transfer to lipoproteins in the absence of non-lipidated proteins were slower: for nHDL-[¹⁴C]PL transfer to LDL in total lipoproteins, to isolated LDL, and to HDL respectively t_1/2 = 3.4, 4.1, and < 0.5 h. (Supplementary Figures III–V). Transfer of nHDL-[¹⁴C]PL to total lipoproteins, which do not contain lipid transfer factors, was enhanced by the addition of PLTP (Supplementary Figure III, Panel A, right insert).

Figure 3.

Transfer of nHDL-[14C]PL to Serum Lipoproteins: A) Control nHDL-[14C]PL, no serum (O). B) nHDL-[14C]PL was incubated with serum at 37 °C for various times (as labeled) and separated by SEC after which collected aliquots were β-counted (●). The grey rectangle circumscribes the regions of LDL-[14C]PL that were summed to calculate kinetics of nHDL-[14C]PL transfer to LDL, which is shown in Panel C.

nHDL-[¹²⁵I]Apo AI Transfers to HDL In Vitro

Experiments similar to those in Figures 2 and 3 were also conducted with nHDL-[¹²⁵I]apo AI. The SEC elution profile of nHDL-[¹²⁵I]apo AI was distinct from those of the native serum lipoproteins (Figure 4 A) and from that of lipid-poor [¹²⁵I]apo AI (Figure 4 B). According to SEC analysis, nHDL-[¹²⁵I]apo AI rapidly transferred to HDL, as seen after only brief incubation with serum at 4 °C (Figure 4C). Changes in the nHDL-[¹²⁵I]apo AI distribution occurred in two time frames. At early times (0.5 – 2 h), nearly all nHDL-[¹²⁵I]apo AI coeluted with HDL (Figure 4 D–F). At later times, ≥ 4 h, some [¹²⁵I]apo AI co-eluted with lipid-poor apo AI (Compare Figures 4 B and I–K). A trace of [¹²⁵I]apo AI was detected in the LDL and none appeared in the VLDL fractions. A direct comparison of the SEC distribution of nHDL-[³H]FC, [¹⁴C]PL, and [¹²⁵I]apo AI before and after short incubations with HDL or serum is shown in Supplementary Figure IX.

Figure 4.

Transfer of nHDL-[125I]Apo AI to Serum Lipoproteins. A) SEC Profile of nHDL-[125I]apo AI (O). B) SEC of [125I]Apo AI(●) and HDLA280 nm (grey fill). C–K) SEC profiles of [125I]apo AI after incubation with serum at various temperatures and times as labeled.

Serum Factors Segregate nHDL-FC, PL, and Apo AI

Figure 5 compares the distribution of nHDL-FC, PL, and apo AI at the end of 2 and 24 h incubations with serum (from the data of Figures 2, 3, and 4). These data show that each component of nHDL has a different fate. After a 2 h incubation, nHDL-FC distributes as LDL > HDL >> VLDL; nHDL-PL distributes as HDL > LDL > lipid-poor (LP) apo AI > VLDL; nHDL-apo AI distributes as HDL > LP apo AI >> LDL. The distribution of nHDL analytes is only slightly different after a 24 h incubation with serum.

Figure 5.

Distributions of nHDL-[3H]FC, [14C]PL, and [125I]apo AI into Serum Lipoproteins and Lipid-Poor Apo AI. Bars represent the percent distribution of each radio-analyte from Figures 2–4 at 2 h. The 24 h values for [125I]apo AI are shown as a red line above the 2 h bar.

In vivo nHDL Metabolism

The plasma clearance of the constituents of nHDL and their appearance in major organs was followed in groups of mice after injection of nHDL-[³H]FC, [¹⁴C]PL, and [¹²⁵I]apo AI. The respective halftimes for disappearance of nHDL-[³H]FC and nHDL-[¹⁴C]PL from plasma were fast, t_1/2 = 5.2 ± 0.07 and 2.0 ± 0.4 min (Figure 6 A, B). The in vivo kinetics for nHDL-[¹²⁵I]apo AI were profoundly different from those for FC and PL (Figure 6 C). These data best fit a bi-exponential curve having fast and slow components with half times of 2.1 ± 1.0 min and 7.7 ± 1.3 h; the respective pre-exponential factors, which reflect the relative contributions of the two components to the total decay, were 0.68 and 1.9, i.e., 25 and 75% of the total. In the same experiment, we collected tissues at various times and determined their associated [³H]FC, [¹⁴C]PL and [¹²⁵I]apo AI. The majority of the tissue [³H]FC- and [¹⁴C]PL-uptake was hepatic (Figure 6 D—F; Supplementary Figure VI); the corresponding halftimes for hepatic uptake were 2.6 ± 0.7 and 1.4 ± 0.1 min (Figure 6 D, E inserts). These data complemented the plasma kinetic data by showing the rapid disappearance of [³H]FC and [¹⁴C]PL coincided with rapid hepatic uptake. Similarly, the much slower disappearance of plasma [¹²⁵I]apo AI coincided with little hepatic uptake, which declined with time (4.5 → 0.5 % over 24 h); the initial rapid apo AI clearance has been attributed to transfer from plasma to extravascular spaces.^17–19 At later times, hepatic [³H]FC and [¹⁴C]PL declined while accumulating in other tissue sites, especially in erythrocytes, which respectively contained 13 and 7% of injected [³H]FC and [¹⁴C]PL at 30 min and 8% and 10% at 24 h (Supplementary Figures VI and VII). In a second set of in vivo experiments we compared the plasma kinetics and organ uptake of nHDL- vs. HDL-[³H]FC and [¹⁴C]PL. Rates of plasma clearance of FC and PL from the two lipoproteins were similarly rapid (t_1/2 <3 min; Supplementary Figure VIII) and organ uptake of nHDL- and HDL-FC and PL was mostly hepatic (data not shown). Kinetic constants for the plasma clearance and liver accretion of nHDL-FC, PL, and apo AI are summarized in Table 2.

Figure 6.

nHDL-FC, PL and Apo AI Kinetics for Plasma Clearance and Hepatic Uptake in Mice. Plasma and tissue were collected and analyzed at various times following the injection of nHDL-[3H]FC, -[14C]PL, and -[125I]Apo AI; details are in Material and Methods. A – C) Plasma decay plots for each analyte. Data are plotted as smoothed curves with the ordinates adjusted to the same scale for comparison. For C, data were fitted to a four-parameter bi-exponential equation having the form, Weight-Normalized Plasma DPM = ae-k1t + be-k2t where t is time, k1 and k2 are rate constant, and a and b are their respective pre-exponential constants that reflect injected dose. D – E) Hepatic uptake plots for each analyte. Inserts in D and E show the kinetics for the initial (0 – 30 min) rapid hepatic uptake of [3H]FC and [14C]PL.

Table 2.

Kinetic Constants for the Plasma Clearance and Liver Accretion of nHDL-FC, PL, and Apo AI

	Plasma Decay Kinetics*		Liver Uptake,
Plasma Analyte	k, min−1	t1/2, min	k, min−1; uptake max, (%)	t1/2, min
HDL-[3H]FC	0.68 ± 0.09	1.0 ± 0.09	0.91 ± 0.7 (25 ± 6)	0.77 ± 0.33
HDL-[14C]PL	0.46 ± 0.16	1.5 ± 0.4	0.46 ± 0.23 (12 ± 2)	1.5 ± 0.5
nHDL-[3H]FC	0.22 ± 0.12	3.2 ± 2.1	0.17 ± 0.06 (18 ± 3)	4.1 ± 1.1
nHDL-[14C]PL	0.52 ± 0.23	1.3 ± 0.7	0.71 ± 0.2 (11 ± 2)	1.4 ± 0.05
nHDL-[125I]Apo AI†	0.34 ± 0.29	2.1 ± 1.1	---	---
	(1.5 ± 0.32) × 10−3	460 ± 80

^*From Figure 6 and Supplementary Figure VIII.

^†apoAI kinetics were conducted on a longer time scale (Supplementary Figure VI) from which short and long components of the kinetics were calculated.

PLTP Increases Hepatocyte nHDL-PL Uptake

Given the rapid removal of nHDL-[¹⁴C]PL by liver in vivo (Figure 6), its faster in vitro transfer to lipoproteins in serum (Figure 3) compared to TLP (Supplementary Figure III), and the PLTP-mediated increase in nHDL-[¹⁴C]PL transfer to TLP (Supplementary Figure III), we tested the effects of PLTP on nHDL- and HDL-FC and PL uptake by cultured human hepatocyte Huh7 cells. Our data revealed similar rates of nHDL- and HDL-[³H]FC uptake, which were not affected by addition of PLTP to the incubation media (Figure 7 A and C). The rates of nHDL- and HDL-[¹⁴C]PL uptake were also similar. However, addition of PLTP to the incubation media increased nHDL-[¹⁴C]PL uptake (+85%; p = 0.0016) but not that of HDL-[¹⁴C]PL (Figure 7 B and D). When uptake was calculated as percent of input, FC uptake from both nHDL and HDL was 30% at 180 min, while PL uptake, in this case PC uptake, in the absence of PLTP was 5%, and increased to 10% from nHDL in the presence of PLTP. Greater uptake of FC vs PC has also been reported for fibroblasts.²⁰

Figure 7.

Comparison of the rates of Hepatic nHDL- and HDL-[14C]PL Uptake In Vitro. nHDL- and HDL-[14C]PL (20 µg/mL protein) were incubated with Huh7 cells in the presence or absence of PLTP (10 µg/mL) and the cell-associated radioactivity determined at various times as shown. Additional details are provided in the Materials and Methods. Slopes in panel B were different, p = 0.0016. Slopes in the other panels were not significantly different.

Discussion

Formulation of the RCT Hypothesis

Discovery of the role of LCAT in HDL metabolism provoked the hypothesis that cholesterol esterification is part of a broader cholesterol transport process that prevents accumulation of cytotoxic FC levels in peripheral tissue.^21–23 This process, now known as RCT, was hypothesized decades before the transporters involved in the first step, nHDL production mainly by ABCA1,^24,25 and the selective hepatic lipid-uptake receptor, SR-B1,²⁶ involved in the final disposal step, were identified. In the context of atheroprotection by HDL, RCT evolved into a three-step process—cholesterol efflux from ABCA1-expressing macrophages in the subendothelial space of the arterial wall to apo AI or preβ HDL giving rise to nHDL, esterification of nHDL-FC by plasma LCAT producing more mature, spherical HDL, and selective hepatic extraction of HDL-CE by SR-B1, which “nibbles” a few CE from many HDL.⁸ In the liver, CE is converted to FC, some of which is converted to bile salts, both of which enter the bile and then transfer to the feces for excretion.²⁷

nHDL Composition, Size and Structure

ABCA1-derived nHDL is thought to be a precursor to the mature forms of HDL that occur in plasma. The composition, size and structure of ABCA1-derived nHDL^6,7 as well as the metabolism of nHDL-protein derived from several cell types, including macrophage foam cells, have been reported and are similar.^18,28,29 Cell-ABCA1-derived nHDL is both similar and different from human plasma pre-β HDL. Both contain FC, PL and apo AI. However, the human plasma pre-β HDL is smaller and is ~90% protein, mostly apo AI,³⁰ and has been proposed to be a terminal nascent HDL product,¹⁸ a conclusion that is consistent with the likely rapid removal of lipids from nascent HDL soon after its production as reported here and by others.^19,31 This difference may limit the transferability of findings with ABCA1-derived nHDL to human pre-β HDL metabolism.

nHDL-lipid metabolism has not been reported so that we have advanced the state of knowledge of the role of nHDL-analytes in HDL metabolism and are the first to study the plasma clearance and tissue uptake of nHDL-FC, PL, and apo AI derived from the same cell system and using the same mouse model, C57BL/6J mice. The major novelty of our contribution is the observed rapid, quantitative in vivo metabolism of the nHDL-FC and PL in a background of slower apo AI metabolism in which LCAT plays a minor role in the nHDL-FC clearance and tissue uptake. These findings should guide the design of therapies that promote the transfer of peripheral tissue-FC to the liver for disposal.

Our findings are consistent with previous nHDL studies: Interaction of apo AI with the cellular ABCA1 transporter produces multiple nHDL species with dimensions that are a function of cellular cholesterol content^6,7,32,33 and cell type³⁴ with lower ratios of FC/PL in nHDL of J774 macrophage cells and human fibroblasts. A high FC/PL molar ratio, with FC > PL (Table 1) has been reported previously for large and medium nHDL formed by BHK-ABCA1 cells^6,32 as well as HEK293-ABCA1 cells.⁷ Our data confirm that nHDL-SM content increases with increasing nHDL size³⁵ and that erythrocytes are FC and PL reservoirs (Supplementary Figures VI and VII).³⁶

Most variously-sized nHDL are discoidal,³⁷ and according to electron microscopy studies, both 9 nm and 12 nm diameter nHDL from J774 macrophage cells are discoidal.³⁴ Our preliminary electron cryo-microscopic studies of nHDL from BHK-ABCA1 cells also show a discoidal structure (J. Michael Bell, et al., unpublished). In contrast, nHDL from HEK293-ABCA1 cells were reportedly mostly spherical.⁷ We confirmed^6,7 that the nHDL from BHK-ABCA1 cells is FC-rich and contains PL and apo AI and then studied the in vitro and in vivo fates of [³H]FC, [¹⁴C]PL and [¹²⁵I]apo AI of the major nHDL-M species.

In Vitro Fates of nHDL-FC, PL, and Apo AI

The rate-limiting step for spontaneous lipid transfer from particle surfaces is desorption,^38,39 which follows Kelvin’s law stating that desorption halftimes are a positive function of particle radius.^40,41 Differences in particle shape—HDL is a sphere and nHDL is a disc, and composition—nHDL and HDL are 60 (Table 1) and 10 mol% FC,⁴² play unknown roles. However, with these caveats we conclude that their similar Stokes radii would dictate similar halftimes for FC desorption from nHDL and HDL, i.e., ~2–4 min.¹⁵ Consistent with this, the halftimes for nHDL-[³H]FC transfer to LDL is 2.1 ± 0.5 min (Figure 2 E). Given that FC desorption rates are independent of the identity of the acceptor,¹⁵ the rate of nHDL-[³H]FC transfer to HDL is the same as its transfer to LDL, ~2–4 min. In whole serum the number of HDL particles is ten times that of all other lipoprotein particles combined⁴³ so that based on collision theory the first encounter of desorbing nHDL-[³H]FC is expected to be HDL, the same as that found for plasma acceptors of macrophage FC efflux.⁴⁴ nHDL- and HDL-[³H]FC are esterified to [³H]CE by LCAT.²¹ However, the FC esterification rate is slow compared to the rate of transfer such that >90% of the [³H] activity transferred to LDL in the initial 10 min is [³H]FC, not [³H]CE (Supplementary Figure II). Slow FC esterification vs. rapid hepatic uptake in mice has also been observed for HDL-FC.¹⁹

The in vitro transfer halftime of dipalmitoylphosphatidylcholine (DPPC) from HDL₃ to LDL is longer than that of FC, i.e., 5 h,¹⁵ which is similar to what we observed for nHDL-PL transfer to total lipoproteins (3.4 h), LDL (4.1 h), and HDL (2.1 h) (Supplementary Figures III–V). In contrast, nHDL-[¹⁴C]PL transfer to lipoproteins in whole serum is faster (t_1/2 = 6.5 min; Figure 3C). We attribute the faster rate of PL transfer in whole serum to PLTP activity⁴⁵ for several reasons as discussed below. nHDL-apo AI transfers to HDL within 10 min in vitro (Figure 4) and between 2 and 24 h, some HDL-associated apo AI transfers to the lipid-free or lipid-poor form, an effect that is likely a function of HDL instability;^46,47 HDL is disrupted by many HDL-modifying factors–LCAT, CE and PL transfer proteins, and endothelial and hepatic lipases.^48–52 The distribution of nHDL-[³H]FC, [¹⁴C]PL, and [¹²⁵I]apo AI into lipoproteins in serum at 2 h (Figure 5), which is near equilibrium, is similar to the natural distribution of FC, PL, and apo AI among serum lipoproteins.^42,53 Most FC is in LDL, while more PL is in HDL than LDL, and nearly all apo AI is in HDL.⁵³

Metabolic Fates of nHDL-FC, PL, and Apo AI In Vivo

The disappearance halftime for plasma nHDL-[³H]FC in vivo (t_1/2 = 5.2 min; Figure 6 A) is comparable to its in vitro transfer to plasma lipoproteins, primarily HDL (t_1/2 = 2.1 min; Figure 2 E); the longer t_1/2 observed by SEC (Figure 2 D) is due to the intrinsic limitation of this method to t_1/2 >15 min. Rapid plasma clearance of nHDL-[³H]FC coincided with that of [³H]FC accretion by the liver, initially the main tissue site of uptake (Figure 6 D). nHDL-[³H]FC is likely cleared by multiple pathways. In one, nHDL-[³H]FC transfers directly to liver cells. In the other, nHDL-[³H]FC transfers to HDL from whence [³H]FC transfers to the liver. Two likely mechanisms underlie these pathways. One is mediated by SR-B1, which plays a role in reversible HDL-FC flux.^54–56 SR-B1 overexpression in CHO cells increases their uptake of HDL-FC,⁵⁷ and compared to WT mice, hepatic removal of HDL-FC is faster in mice overexpressing SR-B1.¹⁹ Given the size and compositional similarities of nHDL and HDL, we propose that some nHDL-[³H]FC is also cleared by hepatic SR-B1. The rates of desorption of both nHDL- and HDL-FC are rapid so that direct receptor-independent, spontaneous transfer to the plasma membranes of liver cells is a likely second mechanism underlying the accretion of [³H]FC by the liver and other major tissue sites at later times post nHDL-[³H]FC injection (Supplementary Figures VI and VII). Both mechanisms could support the role of HDL as the shuttle between the liver and extrahepatic tissues. The halftime for nHDL-FC transfer to the liver is 2.6 min during which mouse plasma only esterifies 2.6×0.46% FC/min =~1% of nHDL-FC (Supplementary Figure II). Thus, most hepatic nHDL-FC uptake is independent of LCAT activity; a similar finding has been reported for HDL-FC in mice¹⁹ and humans.⁵⁸

Similarly, t_1/2 for the in vivo disappearance of nHDL-[¹⁴C]PL coincided with its rapid appearance in liver (Figure 6 B, E). However, unlike nHDL-[³H]FC, plasma clearance and hepatic accretion of nHDL-[¹⁴C]PL was faster—t_1/2 = 2.0 and 1.4 min respectively (Figure 6 B, E)—than its in vitro desorption to isolated lipoproteins (t_1/2 = 2– 4 h; Supplementary Figure III – V) but comparable to nHDL-[¹⁴C]PL transfer to lipoproteins in whole serum, t_1/2 = 6.5 min (Figure 3). Thus, transfer of nHDL-[¹⁴C]PL from plasma to liver must involve a plasma factor or a hepatic receptor. This factor is likely PLTP, a conclusion supported by four observations: nHDL is a PLTP target;^18,28 injection of PLTP into rat doubles the rate of clearance of an rHDL-PC ether;⁵⁹ transfer of nHDL-[¹⁴C]PL to LDL in serum is faster than transfer in total lipoproteins in the absence of the lipoprotein-deficient fractions (6.5 min vs. 3.4 h; Figure 3, Supplementary Figure III); addition of PLTP increases transfer to LDL in a dose-dependent manner (Supplementary Figure IIIA, right insert); and PLTP increases the rate of transfer of nHDL-[¹⁴C]PL to Huh7 hepatocytes (Figure 7). nHDL-[¹⁴C]PL could, like nHDL- and HDL-FC,^19,57 also be cleared by hepatic SR-B1 although the much slower rate of selective uptake compared to FC (12%),⁶⁰ could make this mechanism quantitatively less important. Hydrolysis could also play a role but only at much longer times. For example, in rat, the plasma t_1/2 for the disappearance of [³H]PC is ~1 h. This process is mediated, in part, by LCAT and hepatic lipase-mediated hydrolysis, which account for ≥90% of PC clearance.⁶¹ In contrast to the nHDL-lipids, the plasma t_1/2 for nHDL-[¹²⁵I]apo AI is longer, 460 min, and after fusion with HDL and subsequent its release as lipid-free or poor, apo AI is available for additional efflux cycles.

Relevance to Human Pathophysiology

It is interesting to compare our data with similar human kinetic studies of plasma HDL-[³H]FC. If one considers that human metabolism is slower than that of mouse, there are striking similarities. Among human volunteers, only a small fraction of intravenously injected HDL-[³H]FC is converted to HDL- and LDL-[³H]CE; the majority of FC is rapidly cleared from plasma into tissue without esterification with t_1/2 ~7– 9 min.^31,58 Moreover, as with our observed rapid in vivo hepatic nHDL-[³H]FC uptake data, human HDL-[³H]FC rapidly appears in bile reaching a maximum within 40 min.⁶²

A Model for nHDL Metabolism

nHDL, with its high FC content, and the transient nature of its lipid components, appears to be a key modulator of in vivo cholesterol homeostasis and phospholipid exchange among lipoproteins and tissues. Our data support a model of hepatic nHDL metabolism as follows (Figure 8, Steps 1 – 8): Apo AI interaction with cellular ABCA1 forms nHDL containing FC, PL, and apo AI. nHDL-FC rapidly transfers to the liver either directly or via initial transfer to HDL, followed by transfer to the liver; this occurs via both SR-B1 and spontaneous transfer. nHDL-PL transfer to liver occurs directly or via HDL and although our cell data implicate PLTP in this process, it is not clear whether PLTP is a direct mediator or that PLTP simply modifies nHDL in a way that accelerates PL uptake; the latter hypothesis is supported by the observation that PLTP is essential to nHDL remodeling.¹⁸ Ultimately, FC and PL equilibrate among all lipoproteins giving steady-state plasma concentrations that are the targets of enzymes and transfer proteins. Although our model also shows hepatic LDL-lipid uptake via the LDLR, this process is not as important during the early rapid hepatic uptake but contributes to hepatic removal at longer times when lipoprotein concentrations have equilibrated to a stationary state. Notably, these fast lipid transfer processes are independent of LCAT activity, which acts on a slower time scale. Over longer time intervals, erythrocytes accrue ~10% of nHDL-FC and PL deposition.

Figure 8.

Model of nHDL Remodeling and Metabolism In Vivo. Cells extrude FC and PL via the interaction of apo AI with ABCA1 giving nHDL (1). nHDL-FC, PL, and apo AI rapidly transfer to HDL, t1/2 < 2 min (2); concurrently nHDL-FC and PL transfer to LDL (3). Over the same time frame, nHDL- and HDL-FC and PL transfer mainly to the liver (4, 5, 6) while some nHDL-apo AI is recycled to ABCA1 (7). Over time, FC and PL equilibrate among erythrocytes (8), extrahepatic tissues, and lipoproteins, which achieve a steady state concentration that is the target of enzymes, transfer proteins, and hepatic receptors. nHDL- and HDL-FC and PL accretion in the liver occurs via spontaneous transfer and SR-B1 (4, 5, 6), with the latter being promoted by PLTP (4). Symbols for FC, PL, and apo AI are labeled as shown in the legend at the top of the figure.

In vitro, nHDL-apo AI associates with HDL with some appearing lipid-free at longer incubation times with serum. In vivo, nHDL-apo AI also transfers to HDL where it remains with a halftime of 460 min (Figure 6 C). No lipid-free apo AI was detected in vivo suggesting that soon after plasma factors convert nHDL-apo AI to the lipid-free form, it immediately reassociates with HDL or initiates another cycle of tissue cholesterol-efflux. Clearly, nHDL structure is disrupted by plasma factors in vitro, and in vivo its components—FC, PL, and apo AI—are cleared by independent mechanisms that do not involve holo-particle uptake. nHDL-apo AI persists in plasma longer than nHDL-FC and PL, which rapidly move through the plasma compartment while maintaining a stationary state of HDL-FC, PL, and apo AI. Given the similarities between HDL and nHDL metabolism, we conclude by paraphrasing from Schwartz et al. who studied human HDL-FC metabolism, i.e., “nHDL-FC and PL constantly exchange between HDL and tissue-membranes, while maintaining a steady state HDL composition.”⁶³

Highlights.

• Nascent (n) HDL-FC, PL and apo AI have distinct destinations and serum kinetics in vitro and enter different pathways for clearance in vivo.

• HDL is an early acceptor of nHDL-FC, which later transfers to LDL but mainly to the liver.

• Hepatic nHDL-FC uptake is rapid and faster than its esterification by LCAT.

• The terminal RCT step for most FC and PL respectively occurs by spontaneous and protein-mediated transfer to tissue, mainly liver.

• nHDL, with its high FC content and labile lipids, is a key modulator of in vivo cholesterol homeostasis and phospholipid exchange among lipoproteins and tissues.

Nonstandard Abbreviations and Acronyms

Apo

apolipoprotein

CDX

methyl β-cyclodextrin

CE

cholesteryl ester

CPM

counts per minute

DPM

disintegrations per minute

FC

free cholesterol

HDL

high density lipoprotein

LCAT

lecithin:cholesterol acyltransferase

LDL

low density lipoprotein

nHDL

nascent HDL

S-, M-, and L-nHDL

small, medium, and large nascent high density lipoprotein

PC

phosphatidylcholine

PL

phospholipid

PLTP

phospholipid transfer protein

RCT

reverse cholesterol transport

SEC

size exclusion chromatography

SM

sphingomyelin

SR-B1

scavenger receptor class B member 1

TLP

total lipoprotein

VLDL

very low density lipoproteins