Novel pathways of apolipoprotein A-I metabolism in HDL of different sizes in humans

Abstract

Objective

A prevailing concept is that HDL is secreted into the systemic circulation as a small mainly discoidal particle; which expands progressively and becomes spherical by uptake and esterification of cellular cholesterol; and then contracts by cholesterol ester delivery to the liver, a process known as reverse cholesterol transport, thought to be impaired in people with low HDL cholesterol (HDLc). This metabolic framework has not been established in humans.

Approach and results

We studied the metabolism of apolipoproteinA-I in four standard HDL sizes by endogenous isotopic labeling in six overweight adults with low HDLc and in six adults with normal body weight with high plasma HDLc. Contrary to expectation, HDL was secreted into the circulation in its entire size distribution from very small to very large, similarly in both groups. Very small (prebeta) HDL comprised only 8% of total apoA-I secretion. Each HDL subfraction circulated mostly within its secreted size range for 1–4 days, and then was cleared. Enlargement of very small and medium to large and very large HDL, and generation of very small from medium HDL were minor metabolic pathways. Prebeta HDL was cleared slower whereas medium, large and very large HDL were cleared faster in the low HDLc group.

Conclusions

A new model is proposed from these results in which HDL is metabolized in plasma mainly within several discrete, stable sizes, across the common range of HDLc concentrations.

The high density lipoproteins (HDL) level, quantified by the concentration of its cholesterol component, i.e. HDL-cholesterol level (HDLc), or its apolipoprotein A-I (apoA-I) level, is indisputably an independent risk factor for coronary heart disease (CHD), based on meta-analysis or pooling of many prospective studies (1,2). Yet, genetic variation that is associated with differences in only HDLc and not in LDL-cholesterol or triglycerides is not associated with risk of CHD (3). This, and the failure of trials of HDLc-raising treatments to prevent CHD (4–6) is leading to a consensus that knowledge of HDL functioning in humans is needed to make progress on the clinical meaning of a low or a high HDL level and how to use information on HDL to develop treatments that can lower CHD (7–12).

Intricate molecular machinery has evolved to maintain homeostasis of cellular cholesterol content by the actions of HDL (8,10–11,13–14). The current canonical framework for HDL metabolism begins with secretion into plasma of nascent very small discoidal HDL (prebeta-1) that has a low content of unesterified cholesterol and no cholesterol ester; size expansion by uptake of unesterified cholesterol transferred from cells by ATP-binding cassette A1 (ABCA1) and other transporters; esterification of cholesterol by lecithin-cholesterol acyl transferase (LCAT) converting HDL from discoidal to spherical; and fusion of small HDL by the action of phospholipid transfer protein (PLTP). HDL can then transfer its cholesterol esters to apoB lipoproteins by CETP and to the liver by SR-B1 (15) and by holoparticle uptake. During these processes, discoidal HDL may be regenerated and returned to the circulation (13). Enlargement of HDL during its circulation has been demonstrated in vervet monkeys (16,17). However, progressive enlargement of HDL in humans, in vivo, has not been reported nor has regeneration of discoidal HDL.

HDL metabolism is commonly studied by labeling its principal protein, apolipoprotein A-I (apoA-I), either endogenously by stable isotopes of amino acids or by exogenous radioisotopes. Usually, the total apoA-I is analyzed yielding a fractional catabolic rate (FCR) and secretion rate representing all HDL sizes combined (18–20); in a few studies apoA-I in two sizes was studied (21–23). However, this information has limited interpretation pertaining to HDL function and by extension, atherosclerosis. For example, would a drug or diet that decreases the FCR of plasma total apoA-I be expected to affect atherosclerosis, and if so, in which direction? HDL metabolism has never been studied in humans through the range of common sizes. Knowing how very small discoidal HDL is metabolized, in vivo, is critical to validating the concept that this HDL species is the precursor in plasma to the spherical (alpha) HDL and is regenerated during its metabolism. Very small-discoidal HDL metabolism has never been studied for this purpose in humans. Thus, knowledge of HDL metabolism has lagged far behind that of VLDL and LDL for which a compelling picture exists of metabolic diversity involving healthful and unhealthful mechanisms (24–26).

A low HDLc concentration in patients with the common phenotype of overweight or obesity and high triglycerides is associated with a high plasma total apoA-I fractional catabolic rate (FCR) (27–31), and a shift in the distribution of HDL to smaller sizes. High risk of CHD in this phenotype is well established. The metabolic basis for the reduced HDL size is not known.

Using endogenous stable isotopic labeling and kinetic modeling of apoA-I in 4 common sizes of HDL, the aim of this study was to determine the main pathways of metabolism of apoA-I HDL in adults with low or high plasma HDL-cholesterol. Using endogenous isotopic labeling and kinetic modeling, we evaluated the hypothesis that HDL would be shown to undergo diminished enlargement and contraction while it circulates in those who have low HDLc levels, suggesting impaired reverse cholesterol transport.

Material and Methods

Materials and Methods are available in the online-only Supplemental Material.

Results

Participant characteristics are shown in Table 1. The participants were selected according to HDLc and body mass index. Mean HDLc was 39 mg/dL and 75 mg/dL in the low and high HDLc groups; apoA-I was 78 and 119 mg/dL; BMI was 30 kg/m2 and 24 kg/M2; and plasma triglyceride was 129 and 74 mg/dL, respectively.

Table 1.

Characteristics of study participants.

	Low HDLc males			Low HDLc females			Mean (SD) low HDL group	High HDLc males			High HDLc females			Mean (SD) low HDL group	P-value for low vs high HDLc group*
	Pt. 1	Pt. 2	Pt. 3	Pt. 1	Pt. 2	Pt. 3		Pt. 1	Pt. 2	Pt. 3	Pt. 1	Pt. 2	Pt. 3
Body weight (Kg)	143.5	96.6	103	82.6	70.5	66.9	93.9 (28.1)	92.6	61.1	73.4	58.8	64.3	63.9	69 (12.6)	0.022
BMI (Kg/m2)	35.4	29.8	29.4	28.9	29.3	27.3	30.0 (2.8)	24.8	24.3	24.5	23.5	23.3	22.9	23.9 (0.8)	0.003
Race	White	White	Non-white	Non-white	White	White	4 White 2 Non-white	White	White	White	White	White	Non-white	5 White 1 Non-white	0.45†
Age (years)	34	50	32	32	59	32	39.8 (11.7)	31	41	56	53	42	55	46.3 (10.0)	0.24
Fasting plasma glucose (mg/dL)	120	89	91	79	90	94	93.8 (13.8)	77	86	96	80	95	87	86.8 (7.7)	0.24
Total cholesterol (mg/dL)	176	150	167	146	247	211	182.8 (39.1)	174	190	202	198	180	214	193.0 (14.7)	0.19
LDL cholesterol (mg/dL)	117	95	101	97	174	124	118 (29.8)	100	78	123	94	92	129	102.7 (19.6)	0.15
HDL cholesterol (mg/dL)	35	38	35	32	47	47	39.0 (6.5)	48	100	52	97	77	78	75.3 (21.8)	0.003
Triglycerides (mg/dL)	120	83	155	87	129	200	129.0 (44)	129	59	136	33	54	35	74.3 (46.3)	0.055
Plasma apoA-I (mg/dL)	68	97	72	61	95	77	78.3 (14.7)	73	175	101	157	106	102	119.0 (38.7)	0.01

BMI: Body-mass index

^*From a Mann-Whitney test.

^†From a 2 × 2 Fisher exact test.

Model development

The process of model development is explained in detail in the online Materials and Methods section. The final model (Figure 1) showed excellent apoA-I enrichment and mass fits (Supplemental Figure VI), and improved objective overall goodness-of-fit measures over previous models (Akaike Information Criterion, Bayes Information Criterion, Supplemental Table I).

Figure 1. Optimized model structure for apoA-I metabolism in plasma.

White circles represent pools of apoA-I in HDL size subfractions. All 4 HDL sizes appear directly in plasma and may undergo holoparticle clearance. Very small-discoidal HDL may be enlarged to large and very large HDL; and medium HDL may be enlarged to large. Very small HDL may be regenerated from medium HDL through release of lipidated apoA-I (gray circle). All HDL size fractions can appear in plasma with or without an extravascular processing compartment, called a delay pool (gray rectangles).

HDL has larger-size distribution in the high HDL group

Participants in the high HDLc group exhibited larger pool sizes in the medium, large and very large HDL fractions, and smaller pool sizes in the small-discoidal fraction than those in the high HDL group (Figure 2- Panel C).

Figure 2. Apo A-I secretion rate into plasma (Panel A), and apoA-I fractional catabolic rate (FCR) (Panel B), and apoA-I mass (Panel C) of the 4 HDL size subfractions.

White bars represent the low HDLc group, gray bars represent the high HDLc group. P-values are from 2-way ANOVAs in which HDLc group and HDL mass, secretion rate or FCR were fixed factors.

Similar apoA-I secretion rates in low and high HDLc groups

Secretion rates of total apoA-I were not significantly different between the high- and low-HDLc groups (14.1 versus 12.0 mg/Kg/d, p=0.23). There were no significant differences in the percent of total apoA-I flux that went into each HDL size (Figure 2- Panel A). The mean percentage of total apoA-I secretion into each HDL size pool for the low and high HDLc groups was, respectively: very small-discoidal HDL: 8.4 versus 8.2%; medium HDL: 33.4 versus 29.4%; large HDL: 35.5 versus 37.3% and very large HDL: 22.7 versus 25.1%.

Origin of the HDL subfractions

About 17% of very small-discoidal HDL production came from medium HDL in the low HDL group compared to 60% in the high HDL group (p<0.001, Figure 3). The main source of all spherical HDL species (medium, large, and very large) was in both groups direct secretion, including its routing through an extravascular precursor (delay compartment). By model optimization, 100% of medium HDL was directly secreted, as there were no conversion pathways needed that lead to medium HDL. On average, 66.3% of large HDL in the low HDLc group, and 94.6% in the high HDLc group, came from secretion (p=0.004). Similarly, 85.7% of very large HDL in the low HDLc group, and 99.8% in the high HDLc group came from secretion (p=0.035) (Figure 3).

Figure 3. Sources of apoA-I in HDL of each size by group.

P-values are from 2-way ANOVA in which HDLc group and HDL source were fixed factors.

Faster clearance of large and very large HDL in the low HDLc group

The low HDLc compared to the high HDLc group had about double the FCR of large and very large HDL (Figure 2- Panel B). In the low HDLc group, mean rates of direct removal of large and very large HDL apoA-I were also about twice as fast as in the high HDLc group (large HDL 0.75 versus 0.32 pools/day, p=0.01; very large HDL 0.64 versus 0.27 pools/day, p=0.004). (Figure 4). The low HDLc group tended to have faster FCR and direct removal rates for medium-size HDL although not significantly.

Figure 4. Model-derived rates and pool sizes for the two study groups.

Numbers inside pools represent mean pool sizes in mg. Numbers above lines or after arrowheads represent mean rates in pools/24h. Percentages in parentheses after rates are the percentages of apoA-I metabolized by that pathway (percent of that pool´s total FCR). Numbers above the arrows out of the box labeled “ApoA-I production” represent the mean percent of apoA-I production going into each HDL subfraction. White circles represent pools that were actually separated and measured, and the number inside them represents mean measured pool size in mg apoA-I. Gray circles represent nonsampled pools whose size was estimated by the model. Gray rectangles represent nonsampled extravascular delay pools in which processing of HDL may occur. *p<0.05 for between-group difference.

Faster clearance of very small-discoidal HDL apoA-I in high HDL group

In the high HDLc group, the FCR for very small HDL apoA-I was more than double that of the low HDL group, 1.24 vs 0.51 pools per day (Figure 2- Panel B). Removal of very small HDL apoA-I from plasma also happened significantly faster than in the high HDLc group, 1.15 vs 0.05 pools per day (Figure 4). The flux of very small apoA-I clearance from circulation was 2.03 mg/kg/d for the high HDLc and only 0.06 mg/kg/d in the low HDLc group (p=0.0026). Clearance of very small HDL apoA-I represented 15% of total apoA-I flux out of circulation in the high HDLc group; this pathway was minimal or undetectable in the participants in the low HDLc group.

More conversion of HDL from smaller to larger sizes in low HDL group

The results showed conversion of smaller into larger HDL particles in most study participants (6/6 low HDL and 4/6 high HDL). Three conversion pathways were identified as significantly contributing to the model fit: From very small-discoidal to large HDL, from very small-discoidal to very large HDL and from medium to large HDL (Figure 1). The low HDLc group had significantly greater flux of apoA-I through these size expansion pathways (mean sum of apoA-I flux through the three pathways: 2.67 mg/Kg/d for the low HDLc group, 0.29 mg/Kg/d for the high HDLc group, p=0.007). The mean proportion of apoA-I flux in the very small-discoidal HDL pool going towards these enlargement pathways was much larger in the low HDLc (81.2%) than in the high HDLc group (6.8%) (Figure 5). Similarly, there was a larger proportion of apoA-I flux from the medium HDL pool to large HDL in the low HDLc than in the high HDLc group (mean 35.0% and 1.6%, respectively).

Figure 5. Metabolism of very small-discoidal and medium HDL.

Numbers are percentage apoA-I flux. P-values pertain to the difference between the low and high HDLc groups.

More regeneration of very small-discoidal HDL from medium HDL in the high HDL group

The mean flux of apoA-I release from medium HDL, which serves as a precursor for circulating very small-discoidal HDL, was higher in the high HDLc than in the low HDLc group (1.31 versus 0.34 mg/kg/d, respectively, p=0.033; mean percentage of flux out of the medium HDL pool was 40.5% versus 9%, respectively) (Figure 4,5).

Cholesterol contents increase with size of HDL on native PAGE

Mean (SE) for molar ratios of cholesterol per apoA-I in the very small, medium, large, and very large HDLs were 17(6), 44(4), 50(5), and 80(7) showing that cholesterol content per apoA-I increases with increasing size (Supplemental Figure III). In addition, cholesterol ester comprised 32% of the cholesterol content in very small HDL, compared with 70–85% in the 3 larger subfractions (Supplemental Figure IV), consistent with very small HDL containing mainly discoidal HDL and small spherical HDL.

Discussion

This study of the metabolic behavior in plasma of size-defined HDL subfractions provides new insights into the physiology of an insufficiently understood lipoprotein. Strengths of the study include its applicability, as it was done in free-living humans who had typical phenotypes associated with low- or high-HDL-cholesterol, complete dietary control using controlled feeding of a commonly eaten diet, and the extensive testing by compartmental modeling of pathways directly implicated by the experimental results. The central finding is that the four HDL size subfractions are all secreted into plasma and circulate mainly within the secreted size for 1–4 days before they are removed from the circulation. These results are not consistent with stepwise size expansion and contraction as major pathways for HDL metabolism and especially not with a common metabolic framework that considers very small discoidal HDL as the main nascent particle that is a precursor the larger cholesterol-rich HDL (10, 13, 32). Still, the complex peaks of the tracer enrichment curves in apoA-I in HDL size subfractions, most clearly evident in the smallest discoidal HDL, and small but real differences among the subfractions in peak time are evidence for interconversion of size subfractions but as minor pathways. Therefore we propose that the processes of cholesterol uptake and transfer into and out of HDL occur mainly within rather than between size subfractions.

The tracer appeared in plasma apoA-I of four HDL sizes at similar times, the tracer enrichment time plots intersected the time-zero origin, and the major peaks of the apoA-I tracer enrichment curves did not progress in time from small to large HDL. These observations together provide a strong case for a considerable degree of secretion into plasma of all four HDL size-defined subfractions, and metabolism that occurs mainly within the secreted size. These tracer enrichments in apoA-I are strikingly unlike those in kinetic studies of apoB lipoproteins in which the peaks of the enrichment curves follow one another in order of size from large VLDL to small VLDL, IDL, and large and small LDL, and the appearance of tracer in LDL apoB starts several hours after the enrichment peak in VLDL thus demonstrating dominance of precursor-product metabolism of apoB lipoproteins according to lipoprotein size (24–26). Results of previous tracer studies of apoA-I in light and dense HDL are also discordant with a major precursor-product relation of HDL size subfractions in plasma (21–23).

Cells that have ABCA1, including hepatocytes, produce nascent HDL with considerable heterogeneity of size from 7.0 to 12 nm, similar to the range of sizes in the present study. In baby hamster kidney cells, apoA-I solubilizes membrane cholesterol and phospholipid, when ABCA1 is available, producing nascent HDL with variable lipid content (33). The variability in the ratio of membrane cholesterol to phospholipid in the microdomain that engages in HDL formation is responsible for the heterogeneity in size of the nascent HDL, a high cholesterol to phospholipid ratio correlating with large size (33). Specific acyl chains of the membrane phospholipid also affect the size of nascent HDL. In cultured primary mouse hepatocytes, cholesterol loading of the cells and treatment with an LXR agonist increased formation of all sizes of nascent HDL from 7 – 12 nm, especially the larger ones (34). ApoA1 is able to acquire phospholipid and unesterified cholesterol intracellularly and from the plasma membrane (35). Accordingly, primary mouse hepatocytes transfected with the human apoA1 gene secrete lipidated apoA1 that contains both phospholipid and cholesterol and is in the common size range of HDL2 and HDL3 (36, 37).

In human embryonic kidney cells expressing ABCA1, the nascent HDL particles that were produced from exogenous apoA-I were mostly spheroidal or spherical in shape from <6 nm to 12 nm in diameter and 90% of the cholesterol was unesterified (38). Kinetic studies of HDL formation in this cell system (39) and previous evidence in J774 macrophages (40) and human fibroblasts (41) showed that each size subfraction is formed simultaneously at the cell surface and released into the culture media with no evidence of conversion of small to large HDL. All told, this considerable body of evidence on HDL formation and secretion in cultured hepatocytes and other cells showing direct large and small HDL production by cells gives important mechanistic support to our in vivo kinetic study.

After a nascent HDL detaches from the hepatocyte surface, in vivo, LCAT in the Space of Disse, or plasma LCAT in hepatic sinusoids, pulmonary circulation, or systemic arterial circulation can rapidly convert unesterified to esterified cholesterol to create the cholesterol ester filled core of plasma HDL (42), which may protect HDL from abnormally rapid renal catabolism. It is possible that LCAT action may not affect the size of HDL particles but rather their shape (38,43,44), and that the usual size distribution and spherical shape of plasma HDL mostly may be established in nascent HDL before it reaches the systemic venous circulation where it is sampled.

The shallow decline in the apoA-I tracer enrichment curves after their peaks required delay compartments that represent extravascular residence before entry into the circulating HDL compartment, as reported previously in nonhuman primates (16,17,45). The extravascular compartments metabolized a variable percentage of the apoA-I secretion from minimal to 37%. These nonsampled delay pools could represent the hepatic lymphatics or sinusoids, other lymphatics, or the pulmonary or systemic small vessels and capillaries and adjacent tissue such as vascular intima. The time of appearance of a late peak of tracer enrichment in apoA-I of large HDL of 5–20 hours is consistent with estimates of residence time of HDL in human lymphatics (46,47). Remodeling of nascent HDL could occur in these extravascular spaces.

According to the results of the present study, an HDL that appears in the systemic venous circulation in a specific size range remains mainly or entirely in that range throughout its lifetime of 1 to 4 days. This is especially so for large HDL with diameter between 8.2 to 9.5 nm, which is not enlarged to very large HDL, 9.6 to 12.0 nm, or contracted to medium HDL, 7.0 to 8.2 nm; and for very large HDL which does not contract to a smaller size category during circulation. Only a small percentage of medium HDL is converted to large HDL, 35.0% in the low HDL and 1.6% in the high HDL group; and no medium HDL is converted to very large HDL. This suggests that HDL particles have structural stability while they circulate. Recent studies of native human HDL favor a trefoil model having 4 apoA-I molecules per particle in most spherical HDL sizes and 5 in the largest subfraction, HDL2b, which corresponds to very large HDL (48). The trefoil's capacity for cholesterol ester can decrease by twisting and increase by untwisting. The trefoil of a medium HDL can untwist enough to produce a large HDL. However, fully untwisted, the 4-apoA-I trefoil of large HDL may not be able to accommodate sufficient additional cholesterol ester molecules to enlarge to very large HDL unless it acquires a fifth apoA-I. However, the incorporation of extra apoA-I molecules to an already stable large HDL may be energetically unfavorable in vivo, because it would require disruption of the bonds between 2 of the 4 apoA-I molecules in the trefoil.

The maintenance of size of spherical HDL subfractions as they circulate does not mean that reverse cholesterol transport is not occurring. Flux of cholesterol into HDL from cells during circulation may be balanced by flux of cholesterol transferred to apoB lipoproteins by CETP and by selective uptake of cholesterol ester by the liver. Thus, reverse cholesterol transfer could be very active but mainly within rather than among the HDL sizes, as constrained by trefoil untwisting and twisting.

While expansion of large to very large HDL may not be common under normal metabolic circumstances, as suggested by the results of our study, we speculate that it could become important in genetically-based CETP deficiency or pharmacological inhibition of CETP, in which HDL is especially large. The load of cholesterol ester accumulating in HDL could overcome the energy required to accommodate more apoA-I molecules to convert large to very large HDL. In addition, or alternatively, an increase in HDL size could result from expansion of large HDL that has 4 apoA-I molecules and very large HDL which has 5 apoA-I molecules to their maximum sizes.

Metabolism of HDL size subfractions in vivo has been studied in two distinct experimental models in nonhuman primates (16, 17). In one study, small-size LpA-I HDL was prepared by immunoaffinity chromatography, radiolabelled, reinjected, and its metabolism followed (16). Some of the small HDL was converted directly to large HDL and some to medium size HDL. This conversion took place in a noncirculating extravascular compartment. The medium-size HDL formed from small HDL was not itself converted to large HDL. The size of medium HDL in the vervet study is similar to large (alpha-2) HDL in the present study. In the present study, large HDL is in part a product of medium HDL and is not converted to very large HDL, analogous to the findings in vervets. Furthermore, in these vervet monkeys, substantial amounts of medium and large size HDL did not arise from small size HDL (16). For example, 50% of large HDL in the monkeys was not accounted for by conversion from small or medium HDL, and thus must be directly secreted into the circulation, supporting the finding in the present study that large HDL is secreted into plasma. In the other study, plasma HDL was partially delipidated to convert some of the large and very large HDL to small and medium HDL, and a large amount of the partially delipidated HDL was reinjected without prior labeling (17). Medium HDL was converted to large and very large HDL directly, whereas large HDL was not converted to very large, analogous to the earlier tracer study and the present study. However, in contrast to the present study, in both nonhuman primate studies, small HDL was quantitatively converted to larger sizes.

We found that very small-discoidal (prebeta-1) HDL plays a minor role in the formation of larger spherical HDL, as indicated by the major peak tracer enrichment of very small-discoidal apoA-I occurring at or after the peaks in the larger sizes. This is surprising since very small-discoidal HDL is often thought of as the major secreted type of HDL that serves as the precursor for the larger spherical HDL formed during circulation. In fact, only 8% of total apoA-I secretion into plasma is the very small-discoidal pool. Very small-discoidal HDL was not converted to medium HDL in either group, and it contributed a small percentage of the production of large HDL, 14% and 4%, and of the production of very large HDL, 14% and 0.2%, in the low and high HDLc groups, respectively. However, we speculate that a large infusion of prebeta HDL could stimulate its conversion into large HDL sizes, as found in the HDL partial delipidation experiments in vervets (17) and in humans after infusion of apoA-I phospholipid disks (47).

The main synthetic pathway for plasma very small-discoidal HDL in the high HDL group is from medium HDL: 60% of small-discoidal HDL synthesis comes from medium HDL, compared to 17% in the low HDL group. In the high HDL group, most of the small-discoidal HDL is then cleared from the circulation. This suggests that formation of small-discoidal HDL from medium HDL followed by clearance of the resulting small-discoidal particle(s) is a potentially more meaningful pathway in the high than the low HDL group (11% of total apoA-I flux compared to <1%). Generation of small-discoidal HDL from medium HDL could occur by LCAT- or PLTP-mediated fusion of medium HDL to form a larger HDL while shedding small-discoidal HDL during remodeling (49). CETP activity can trigger HDL fusion, with a reduction in the size of preexisting HDL and formation of new, smaller particles (50). Joint CETP and hepatic lipase activity can lead to shedding of free or lipid poor apoA-I from alpha HDL, which promptly becomes small-discoidal HDL (51). Smaller spherical HDL, like medium HDL in our study, are more dynamic and unstable, and tend to participate in this fusion process more than larger HDL (52). This is consistent with our kinetic findings that medium but not large or very large HDL are metabolized to small-discoidal HDL.

We found that the most salient metabolic characteristics of individuals with low HDLc are the significantly faster clearance rate from plasma of large and very large HDL, and the markedly and significantly slower clearance of small-discoidal HDL. Secretion rates of apoA-I into plasma of each size of HDL and of total apoA-I were not significantly different between the groups. These results extend the findings in prior studies that had demonstrated a higher clearance rate of total HDL apoA-I in people who have low HDL levels, obesity and the metabolic syndrome (27–31).

The proposed model of HDL physiology, motivated by the results of the present study, integrates structure, cell biology, and human metabolism (Figure 6). This model of HDL metabolism applies to participants who have low or high HDL-cholesterol. Major enlargement and contraction of HDL size has been considered to typify the process of reverse cholesterol transport, a concept that may need to be reconsidered. More so, phenomena taking place within each size subfraction may be responsible for the biological properties of HDL.

Figure 6. Integrated proposal for the in vivo metabolism of plasma HDL in humans.

ApoA-I is secreted by hepatocytes variably lipidated with phospholipid and unesterified cholesterol, and interacts with ABCA1 to take up more unesterified cholesterol and form nascent HDL in a range of sizes. Exposure to LCAT in the hepatic lymphatics and sinusoids esterifies the cholesterol to form an HDL core of cholesterol ester while the size of most of the HDL does not enlarge enough to become a larger subfraction. HDL in these discrete sizes enters the systemic circulation, and then escapes to the interstitial space or lymphatic system. In these extravascular-extrahepatic spaces, very small, medium, large and very large HDL take up and esterify free cholesterol from cells. Simultaneously or right after HDL goes back to the systemic circulation, CETP catalyzes the exchange of recently formed cholesterol esters for triglycerides present in apoB lipoproteins. Circulating HDL may come into contact with SRB1 on hepatocytes, which selectively takes up cholesterol ester from HDL. HDL particles do not experience major changes in size because the influx of cholesterol from cells is balanced by the efflux of cholesterol esters from each HDL particle by the CETP and SRB1 pathways. Hepatic lipase, acting on the triglyceride from the CETP reaction, also may maintain stability of size. This cycle of HDL circulation between plasma and the extravascular space can be repeated many times over several days. In the vascular or extravascular space, small-discoidal HDL can, in a minor pathway, take up and esterify cholesterol from cells, becoming large or very large HDL. In a process facilitated by plasma PLTP, HL, CETP and LCAT, medium size HDL particles experience fusion to generate new small-discoidal HDL and perhaps new medium or large HDL. All HDL size subspecies can be cleared from circulation via holoparticle uptake. Reverse cholesterol transport can be accomplished by coupling cholesterol uptake from cells with CETP-mediated cholesterol ester transfer to apoB lipoproteins and SRB1-mediated selective CE uptake by the liver, without transforming the size of the individual HDL.

ABCA1: ATP-binding cassette transporter A1, FC: Free cholesterol, CE: Cholesterol esters, TG: Triglycerides, CETP: Cholesterol-ester transfer protein, PLTP: Phospholipid transfer protein, HL: Hepatic lipase, LCAT: Lecithin-cholesterol acyltransferase, SRB1: scavenger receptor B1. VS: very small discoidal; M: medium; L: large; VL: very large size HDL subfractions.

Significance.

This study of the metabolic behavior of apoA-I in plasma of size-defined HDL subfractions provides new insights into the physiology of an insufficiently understood lipoprotein. The central finding, applicable to those with low or high HDL-cholesterol, is that the four standard HDL size subfractions are each secreted into plasma and circulate mainly within the secreted size for 1–4 days before they are removed from the circulation. These results are not consistent with stepwise size expansion and contraction as major pathways for HDL metabolism and especially not with a common metabolic framework that considers very small discoidal HDL as the main nascent particle that is a precursor the larger cholesterol-rich HDL. These results by no means disfavor cholesterol efflux from cells to HDL as an important biological process. Rather, these and other phenomena taking place within each size subfraction may be responsible for the biological properties of HDL.

Abbreviations

Apo

apolipoprotein

CETP

cholesterol ester transfer protein

CHD

coronary heart disease

FCR

fractional catabolic rate

HDLc

HDL cholesterol

LCAT

lecithin cholesterol acyl transferase