Understanding HDL metabolism and biology through in vivo tracer kinetics

Abstract

High-density lipoprotein (HDL), owing to its high protein content and small size, is the densest circulating lipoprotein. In contrast to lipid-laden very low and low-density lipoproteins that promote atherosclerosis, HDL is hypothesized to mitigate atherosclerosis via reverse cholesterol transport, a process that entails the uptake and clearance of excess cholesterol from peripheral tissues. This process is mediated by apolipoprotein A-I, the primary structural protein on HDL, as well as by the activities of additional HDL proteins. Tracer-dependent kinetic studies are an invaluable tool to study HDL-mediated reverse cholesterol transport and overall HDL metabolism in humans when a cardiovascular disease therapy is investigated. Unfortunately, HDL-cholesterol raising therapies have not been successful at reducing cardiovascular events suggesting an incomplete picture of HDL biology. But as HDL tracer studies have evolved from radioactive isotope- to stable isotope-based strategies that in turn are reliant on mass spectrometry technologies, the complexity of the HDL proteome and its metabolism can be more readily addressed. In this review we outline the motivations, timelines, advantages, and disadvantages of the various tracer kinetics strategies. We also feature the metabolic properties of select HDL proteins known to regulate reverse cholesterol transport; that in turn underscore that HDL lipoproteins comprise a heterogeneous particle population whose distinct protein constituents and kinetics likely determine its function and potential contribution to cholesterol clearance.

Graphical Abstract:

INTRODUCTION

The high-density class of lipoproteins (HDL) has been of intense research interest since its inverse relationship with coronary heart disease risk was discovered.^1–6 Yet, HDL-cholesterol raising therapies have not matched the clinical success of their low-density lipoprotein (LDL)-lowering counterparts, the reasons for which are varied owing to HDL’s complex biology.^7–9 The initial proposed mechanism by which HDL exerts its beneficial effects is through its function to clear excess cholesterol from arterial macrophages and other peripheral tissues through a process known as reverse cholesterol transport.^{5, 10, 11} However, HDL has additional functions that are incurred by its varied lipid and protein constituents. HDL particles have been classified into subspecies in various ways including by their protein content, such as LpAI (apolipoproteinA-I/APOA1-containing particles without APOA2) versus LpAI/AII (APOA1-containing particles with APOA2) particles, by their density, such as HDL2 (lower density, larger) versus HDL3 (higher density, smaller) particles, and size/shape, such as very small discoidal prebeta versus small to very large spherical alpha particles.^12–17 Regardless of the subspecies classification, relative to LDL, which is composed of only 20% protein by weight, HDL is comprised of proportionately more proteins (50% by weight), whose functions are related to not only lipid metabolism but also to general vascular health through their anti-oxidative and anti-inflammatory properties.^{18, 19} Protein-defined HDL subspecies are also differentially associated with risk of coronary heart disease, stroke, and diabetes,^20–22 further emphasizing that not all HDL particles are created equal. HDL has thus been characterized as a heterogeneous population of lipoproteins that if increased simultaneously would not necessarily improve its cholesterol clearance and other protective properties, as would targeting a specific subpopulation related to a known function such as cholesterol clearance.^23–25

The cholesterol clearance function of HDL has been a driving force to monitor HDL particle metabolism in vivo. The physiological responses of HDL to a therapy can thus be determined, thereby also providing insight into HDL biology itself.^26–28 Since APOA1 is a major structural protein for HDL, APOA1 metabolism is often synonymous with HDL particle metabolism. HDL metabolism studies have been accomplished using tracer kinetics, a strategy that relies on labeling APOA1 to monitor its appearance and disappearance into and out of one or more HDL particle subspecies (e.g., large versus small) in steady-state conditions. Kinetic parameters such as the fractional synthesis rate or fractional catabolic rate (FCR), pools size, residence time, and production rate (PR) are compared between HDL subspecies and between subject groups or cohorts.²⁹ Changes to these parameters due to a treatment or study condition are what we use to understand HDL’s genesis and metabolism.

With the current knowledge that HDL is a heterogeneous particle class whose protein constituents in part dictate its metabolism and function, tracer kinetic studies have expanded to proteins other than APOA1. APOA2, lecithin-cholesterol acyltransferase (LCAT), and cholesteryl-ester transfer protein (CETP), as well as several other proteins, regulate HDL’s cholesterol clearing function and thus have been monitored in humans.^{26, 30–34} As we will outline below, despite the large number of HDL studies conducted to date, this lipoprotein class remains poorly understood. Tracer kinetic studies, however, are revealing increasingly more insight into HDL metabolism in circulation, including its interaction with the other lipoprotein classes such as LDL and chylomicrons. Tracer studies have transitioned from the reliance on radioisotopes to stable isotopes, that in turn have enlisted mass spectrometry technologies for tracer detection. Moreover, these studies have successfully united multiple disciplines pertaining to clinical research, biochemistry and molecular biology, analytical chemistry, mass spectrometry, and mathematics and modeling to expand our knowledge of lipoprotein and HDL biology.

APPROACHES TO CONDUCT HDL PROTEIN METABOLISM STUDIES IN HUMANS

HDL particles comprise a heterogeneous population of lipids (>200 species) and proteins (>200), of which the phospholipids and APOA1 predominate, respectively.^{18, 35, 36} HDL metabolism studies therefore entail a variety of approaches to study one or more of its constituents, as reflected in the 89,714 articles that were retrieved when the term “high-density lipoprotein metabolism” was queried in the National Library of Medicine’s PubMed database (Figure 1A). The term “metabolism” however is broad and not exclusive to the tracer kinetics required to monitor production, residence time, and fractional catabolism of HDL particles and its protein constituents. Such tracer kinetic studies first relied on labeling APOA1 and/or other HDL proteins with radioisotopes (exogenous labeling) but over time stable isotope amino acid tracers (endogenous labeling) became the label of choice (Figure 1B,C).

Figure 1. HDL protein metabolism studies.

A, PUBMED search results with “High-density lipoprotein metabolism”. B, Endogenous vs. exogenous labeling of HDL proteins in humans. C, Example studies featured in this Review highlighting the transition in HDL protein metabolism studies from radioisotopes (exogenous labeling) to stable isotope labeled amino acids (endogenous labeling), as well as the transition from GC-MS to LC-MS/MS for stable isotope measurements. MRM, multiple reaction monitoring; PRM, parallel reaction monitoring (HR/AM-MS/MS).

Exogenous vs Endogenous Labeling of HDL Proteins

The radioisotope strategy entails first labeling with a purified protein, such as APOA1,^{28, 37} or purified HDL^38–40 particles with iodine-125 or iodine-131 (Figure 1B).^41–43 The radiolabeled protein or HDL is injected into the research participant, and its movement into one or more lipoprotein classes can be monitored by collecting blood over the study time course for subsequent lipoprotein isolation and radiography. This method is referred to as exogenous labeling since labeling occurs in vitro. Various strategies have been employed to radiolabel HDL proteins, such radiolabeling native HDL, protein reconstituted with HDL or HDL-like particles, or radiolabeled protein alone. For instance, APOA1 and APOA2 injected as radiolabeled protein alone exhibited similar fractional catabolic rates as when monitored in radiolabeled HDL, and most of their respective radioactivity was located within, although distinct, HDL subfractions as opposed to less dense or lipid-poor fractions.⁴⁴ This study also established that while APOA1 and APOA2 PRs are similar, APOA1 is catabolized more rapidly than APOA2.⁴⁴ On the other hand, radiolabeled APOA4 as either a component of the lipid-poor fraction, or in association with HDL or chylomicrons, exhibited distinct metabolic profiles.⁴⁵ When injected on a chylomicron, APOA4 radioactivity rapidly transferred to the HDL and the lipid-poor fraction, but when injected as either HDL or the lipid-poor fraction, APOA4 transfer was not observed demonstrating that the protein was more stable on these fractions and likely cleared directly from these injected HDL fractions.⁴⁵

In vivo or endogenous labeling of proteins relies on the injection of a stable isotope labeled amino acid such as deuterium- (D/²H) or carbon-13-labeled (¹³C) leucine (Leu) or phenylalanine (Phe) (Figure 1B). The use of stable isotopes to label HDL proteins was first implemented in the early 1990’s,^{37, 46} and due to the safety concerns of radioisotopes, has become the labeling method of choice (Figure 1C).⁴⁷ The labeled amino acid is taken up by tissues and incorporated into nascent proteins such as liver-derived and small intestine-derived APOA1 that is then secreted into plasma. Stable isotopes are not radioactive, so they must be detected using mass spectrometry (more below).

Only a few HDL kinetic studies have been conducted to compare exogenous and endogenous labeling methods in the same group of individuals.^{37, 48} When ¹²⁵I-APOA1 and ¹³C-Phe were simultaneously administered to the participants, the APOA1 fractional catabolism resulting from the endogenous labeling was significantly lower than that of APOA1 determined by exogenous labeling when the APOA1 input from the intestine was assumed to contribute 20% or more (with 80% or less coming from the liver) to the total plasma APOA1 pool; but tended to be slightly but not significantly lower when the intestinal input was estimated at only 10%.³⁷ The results from the simultaneous administration of ¹³C-Phe and ¹³¹I-APOA2 were more distinct.⁴⁸ Endogenous labeling indicated that APOA2 fractional catabolism was significantly slower than determined by exogenous labeling in all seven subjects analyzed.⁴⁸ The differences between endogenous versus exogenous labeling outcomes likely reflect differences in the posttranslational (endogenous) or artificial modification (exogenous) of proteins between the labeling methods. The fractional catabolism of radiolabeled APOA1 incorporated into HDL by in vitro incubation is significantly higher compared to in vitro labeling of intact whole particles of HDL,⁴⁹ indicating that APOA1 on intact HDL particles is more stable and less exchangeable than purified and labeled APOA1 that has been reassociated with HDL. Additional insights into these labeling techniques have been detailed previously.^{50, 51}

Gas Chromatography vs Liquid Chromatography Mass Spectrometry Applied to Endogenous Labeling

Mass spectrometry is required to detect stable isotope tracer in proteins. Subjects are usually administered a stable isotope-labeled amino acid (e.g., trideuterated leucine, D3-Leu) by constant infusion or by bolus, that is in turn taken up by tissues and then incorporated into nascent proteins, some of which are secreted into circulation. Circulating proteins such as those associated with different size HDL particles (Figure 2A) are thus readily accessible by venous blood draw. But tracer detection in HDL proteins, even with the most advanced mass spectrometers, can be challenging since these proteins are very slowly metabolized and the corresponding incorporation of the tracer is very low (<1% of a given HDL protein pool).³⁴ In order to increase the likelihood of detecting the tracer, protein and/or particle enrichment steps are typically needed.

Figure 2. Endogenous labeling of HDL coupled to mass spectrometry for tracer detection.

A, Example particle isolation using native PAGE that separates HDL based on size (alphas and prebeta). The ultracentrifugation fractions (HDL2b etc.) are added for ease of comparison across studies. Excised protein can be prepared for GC-MS using hydrolysis and derivatization methods, or LC-MS using proteolysis. B, Schematic for selected ion monitoring (SIM) of unlabeled-M0 and labeled derivatized-M3 (D3-Leu*). The difference between M0 and M3 is 3 Da. C, Schematic of HR/AM-MS/MS for tracer detection. Unlabeled (M0) and labeled (M3) peptide are co-isolated for fragmentation. Fragments containing leucine are monitored for tracer (zoom). D, A further zoom of the y4 DELR fragment when scanned at two resolutions, R=35,000 vs R=140,000. The latter differentiates D3-Leu’s 2HM3 ion from other M3 isotopes comprising ¹³C, ¹⁵N, etc., improving peak accuracy and quantification. Panel D corresponds to a real y4 fragment ion from APOA1’s peptide THLAPYSDELR (Singh et al., 2016)

Gas chromatography mass spectrometry (GC-MS) predates liquid chromatography MS (LC-MS), thus most HDL metabolism studies to date have relied on the former for stable isotope tracer detection.^52–54 GC-MS entails isolating the protein of interest by SDS-PAGE for subsequent hydrolysis and derivatization of the amino acids for mass spectrometric analysis (e.g., D3-Leu) (Figure 2A). The measurement of an amino acid in GC-MS benefits from the accumulated signal of all (for example) leucines in the protein, increasing the signal for tracer detection. The unlabeled (H3-Leu) and labeled (D3-Leu, Δ3 Da) amino acids are detected independently as the M0 and M3 ion peaks, respectively, using selected ion monitoring (SIM) (Figure 2B). However, the reliance on chemical derivatization to stabilize the amino acid for mass spectrometric analysis lends to contamination and thermal degradation of the analyte, and one cannot completely rule out contamination from other labeled proteins.⁵³

LC-MS, on the other hand, detects the D3-Leu in the protein’s peptide fragments acquired from tandem mass spectrometry (MS/MS); thus the resulting amino sequence confirms protein specificity (Figure 2A,C). Fragments that contain leucines are monitored for tracer (Figure 2C).^{55, 56} As detailed by others, using fragment peaks (MS/MS) versus the intact peptide (MS1) peak is more accurate for quantitative proteomics.⁵⁷ A higher background signal in the MS1 that interferes with low-abundance (<1%) tracer detection, is the primary reason why tracer is measured in the peptide fragment scan.³⁴

Multiple reaction monitoring (MRM) and parallel reaction monitoring (PRM) are the specific LC-MS strategies.^{18, 34, 57, 58} MRM is performed on a triple quadrupole that entails low resolution (~1 Da resolution) readouts, whereas PRM is performed on a quadrupole Orbitrap or quadrupole time-of-flight that entails high resolution readouts also known as, high resolution/accuracy (HR/AM) MS/MS (mDa resolution).^{18, 57} Similar to SIM (Figure 2B), MRM first entails isolation of each unlabeled and labeled peptide separately (Δ3 Da), but differs by subsequent fragmentation and scanning of a single leucine-containing fragment at a time. Multiple fragments (2 to 3) per M0 and M3 peak are often monitored.⁵⁷ PRM, on the other hand, can be used to co-isolate and co-fragment the unlabeled and labeled peptides simultaneously (e.g., +/− 5 Da centered on the average of the M0 and M3 peaks), providing all fragment ions to be monitored in parallel (Figure 2C).³⁴ As the MS/MS resolution increases, the ability to resolve the 2HM3 deuterium isotope itself (535.2906 m/z) from other M3 isotopologs (e.g., 535.2859 m/z; Δ11 mDa) comprising combinations of ¹³C, ¹⁴N, etc. increases (Figure 2D), thereby increasing quantification precision and accuracy.⁵⁵

For a variety of reasons including the reliance on lower protein inputs and the ability to conduct tracer kinetics in multiple proteins simultaneously, LC-MS is increasing in use over GC-MS.^{27, 33, 54, 56, 59, 60} Furthermore, due to its HR/AM scans, PRM will likely become more frequent than MRM in future tracer studies.

EFFECTS OF HDL ISOLATION METHODS ON TRACER ENRICHMENT

HDL can be isolated from plasma using a variety of methods: ultracentrifugation isolates HDL particles based on their density, fast protein liquid chromatography (FPLC) based on particle size, native two-dimensional gel electrophoresis by charge and size, and immunoaffinity chromatography based on APOA1 content.^12–17 However, a detailed investigation of the effects of each isolation method on the resulting tracer curves of APOA1 and other HDL proteins has not been well defined. One study found that HDL isolated by FPLC had a lower APOA1 tracer-to-tracee ratio, respectively, relative to HDL isolated by ultracentrifugation.⁶¹ Upon further analysis by two-dimensional electrophoresis, the ultracentrifugation-isolated HDL was found to only contain some of alpha HDL and free APOA1, but not the prebeta-1, prebeta-2, or the smaller alpha particles included in the FPLC-isolated fractions.⁶¹ This loss of prebeta and small alpha particles was also seen when comparing ultracentrifugation-isolated HDL to plasma.¹⁷ In addition to isolating only a subset of HDL sizes, ultracentrifugation has also been shown to result in a differential loss of HDL proteins, up to a 35% loss of APOA1 with only a minimal loss of APOA2.⁶² Overall, ultracentrifugation as well as the other isolation methods are each biased towards the isolation of specific HDL particles and their corresponding proteins, as reviewed in detail previously.⁶³ These biases likely affect the apparent metabolism of the HDL particles and proteins as well. More detailed studies on the effect of isolation method on the resulting protein metabolism are needed to fully understand its impact on HDL protein metabolism.

ENDOGENOUS LABELING OF HDL PROTEINS IN DISEASE CONDITIONS, INTERVENTION STUDIES, & CLINICAL TRIALS

Thus far we have highlighted various approaches to characterize the metabolism of HDL and its proteins. Some of these studies have analyzed HDL metabolism in the context of a disease condition, or drug or lifestyle intervention. To date, most studies have used endogenous labeling and GC-MS to monitor the metabolism of APOA1, and occasionally other proteins such as CETP, in total plasma or total HDL.

Obesity and Diabetes

Low plasma concentrations of HDL-cholesterol and APOA1 are associated with obesity and type 2 diabetes. Kinetic studies have shown that obese subjects have a higher HDL APOA1 FCR (57% increase) and PR (40% increase), relative to normal weight controls.⁶⁴ The higher percentage increase in FCR relative to PR in obese individuals results in a net lower plasma APOA1 concentration. In patients with impaired glucose tolerance and type 2 diabetes, lower levels of HDL APOA1 were also due to a higher APOA1 FCR.^{65, 66}

Intervention Studies

Several interventions increase HDL APOA1 concentrations by diverse kinetic means. For instance, oral estrogen increased APOA1 concentration in HDL2 and HDL3 size particles by increasing its PR.⁶⁷ Dietary saturated fat, when replacing carbohydrate, also increased total plasma APOA1 by increasing its PR.^68–70 Fibrates elevated not only APOA1 PR, but also to a lesser extent its FCR, thereby increasing APOA1 levels.⁷¹ HDL-cholesterol and APOA1 levels are increased during weight loss due to a decrease in APOA1 FCR.⁷² In contrast, although statin therapy, such as atorvastatin, significantly increased the catabolism of APOB in very-low-density lipoprotein (VLDL), intermediate-density lipoprotein (IDL), and LDL, statins did not significantly alter APOA1 metabolism.^{71, 73}

CETP Inhibitor Studies

The major function of CETP is to exchange cholesteryl-ester in HDL for triglyceride in APOB-lipoproteins.⁷⁴ Although CETP inhibitor trials have yet to meet clinical expectations, insights into their mechanisms of action have been made possible with endogenous labeling conducted in various studies. Patients deficient in CETP exhibited high levels of HDL-cholesterol and have a lower APOA1 and APOA2 FCR compared to control subjects.⁷⁵ Pharmacological inhibition of CETP with torcetrapib also increased plasma HDL-cholesterol, APOA1, and APOA2 concentrations and decreased APOA1 and APOA2 FCRs without altering PRs.^76–78 HDL2 (alpha1 and alpha2, Figure 2A) exhibited the most significant increase in cholesterol content,⁷⁸ and APOA1 and APOA2 pool sizes increased in alpha1 and alpha2, respectively.^{77, 78} Similar results were seen for subjects receiving anacetrapib, with the exception that anacetrapib did not alter the FCR of HDL APOA2.²⁷ The anacetrapib study was also unique in that, while GC-MS was used to determine APOA1 and APOA2 metabolism, MRM was used to monitor the metabolism of immunopurified total plasma CETP. Kinetic analysis revealed that the increase in circulating CETP levels after anacetrapib treatment was due to a decrease in its FCR.²⁷

HDL PROTEIN KINETICS VIEWED ACROSS HDL SIZE FRACTIONS

We now understand that HDL is a highly heterogeneous population of various protein constituents and protein-defined HDL subspecies that associates with the overall health and disease risk of the individual.^{24, 79, 80} Moreover, these proteins associate with specific HDL particle sizes,^{26, 32} indicating distinct functions.

HDL Proteins Exhibit Distinct Tracer Enrichment Profiles

The majority of HDL protein kinetic studies in humans have focused on the metabolism of APOA1 in total plasma HDL, while only a handful of studies have looked at APOA1 metabolism in two or more HDL sizes: large HDL2 and small HDL3;^{67, 81} prebeta and alpha HDL;⁸² and prebeta, alpha3, 2, and 1 HDL.^{26, 30, 34, 83–85} A major advantage of measuring tracer using LC-MS, as opposed to GC-MS, is that the study is not limited to APOA1, but can be employed to detect tracer in multiple proteins in a given HDL size simultaneously.³⁴

Our group has capitalized on PRM (HR/AM-MS/MS)-enabled stable isotope tracer kinetics to profile the metabolism of 11 HDL proteins – APOA1, APOA2, APOA4, APOC3, APOE, APOL1, APOM, APOJ, CETP, LCAT, and phospholipid transfer protein (PLTP) – across 5 or more HDL sizes over four studies.^{26, 30, 32, 34} These proteins have all been reported to play a functional role in reverse cholesterol transport, and thus we hypothesized that studying their metabolism would give us a better understanding of how this process is occurring in humans in vivo. Additionally, these HDL proteins are 10- to 1000-fold less abundant than APOA1. The bolus administered D3-Leu tracer enrichment profiles for each protein are highlighted (Figure 3A). These tracer enrichment curves underscore two main points: 1) peak enrichment is less than 1% for most HDL proteins except, APOE, CETP, and APOA4 on small HDL; and 2) no two HDL proteins exhibit the same tracer enrichment profiles, indicating distinct metabolism and potentially unique functions for each protein (Figure 3A). The relative ascending and descending slopes demonstrate relative rates of appearance in and catabolism out of plasma circulation, respectively. For example, relative to APOA1, APOE and CETP appearance in and clearance from circulation are faster (Figure 3A). The following sections highlight how the unique metabolic properties of featured proteins provide insight into general HDL biology.

Figure 3. HDL protein metabolism determined by endogenous labeling and HR/AM-MS/MS.

A, Each protein tracer enrichment curve is shown for the HDL size fraction in which that protein is most abundant (data from Singh et al., 2016; Andraski et al., 2019; Singh et al., 2021; Andraski et al., 2023). B, APOA1 tracer data across the HDL sizes (data from Andraski et al., 2019) and APOB tracer data in VLDL, IDL, and LDL (data derived from Zheng et al., 2010). C, The major and minor production pathways for APOA1-HDL; also featuring minor pathways for APOA2 (References: Singh et al., 2016; Andraski et al., 2019). D, A summary of distinguishing metabolic properties for PLTP, CETP, and LCAT on HDL. PLTP on medium alpha2 transfers to larger alpha1 and alpha0 HDL; CETP on medium alpha2 and small alpha3 remains on each HDL size until it is cleared from circulation; and once LCAT appears on very small prebeta, small alpha3, and medium alpha2 HDL, it remains on each HDL size until it is cleared (Reference: Singh et al., 2021). E, APOA4 metabolism on small and large HDL are distinct, indicating that each may have a distinct origin and function. APOA4 on small HDL may be directly secreted by the small intestine while APOA4 on large HDL may originate from chylomicron transfer (Reference: Andraski et al., 2023).

APOA1 HDL Particle Metabolism is Complex

Tracer kinetics analysis of APOA1 across multiple HDL sizes provided a unique perspective on HDL particle metabolism. APOA1 tracer across all HDL sizes appears rapidly in circulation (by 30 minutes) and peaks in circulation around the same time (between 8–12 hours).^{34, 83} This enrichment profile contrasts that of APOB in VLDL, IDL, and LDL in which APOB appears and peaks first on VLDL, followed by IDL, and finally LDL (Figure 3B), indicating different metabolic structures between APOB and APOA1 particle systems. The APOB precursor product tracer enrichment curve pattern indicates two major pathways. Not only are all size fractions directly secreted into plasma but also larger sizes progressively are converted to smaller sizes.⁸⁶ In contrast, the similar time of appearance and peak enrichment of APOA1 across the HDL sizes indicates that the major pathway is the direct secretion of multiple HDL sizes (Figure 3B), consistent with studies in cultured cells showing that primordial HDL particles can vary in size, and their size is a result of the amount of cholesterol and phospholipid in the cell membrane domains utilized to synthesize HDL.^87–91 However, compartmental modeling of these APOA1 enrichment profiles also confirmed APOA1 flux from prebeta to medium, large, and very large alpha particles and from small to medium alpha particles; contributing to ~2% to 20% of total APOA1 flux into the HDL sizes that in turn may mark the HDL particle size expansion involved in reverse cholesterol transport. ^{34, 83}

APOA2 Metabolism Provides Insight into Reverse Cholesterol Transport

Of all the protein tracer enrichment profiles monitored to date by PRM, APOA1 and APOA2 on medium-size HDL look the most similar (Figure 3A). Given that 60% of APOA1 particles contain APOA2, it is likely that these similar enrichment profiles reflect the colocalization of APOA1 and APOA2 on many of the same particles.⁹² Also similar to APOA1, a flux pathway from small to medium spherical HDL was detected for APOA2 (Figure 3A).^{26, 34} This small to medium alpha size expansion pathway may reflect in vivo particle fusion, as small alpha HDL are more unstable and participate more readily in particle fusion compared with larger HDL.^{93, 94} Additionally, this pathway may also reflect size expansion via ABCA1-mediated cholesterol efflux, as APOA1-APOA2-containing HDL (LpAI-AII) have been shown to be significantly better activators of efflux relative to APOA1 particles that do not contain APOA2.⁹⁵ APOA1-APOA2 HDL particles are also enriched in other lipid transport proteins,⁹⁵ similar to what we see in small and medium HDL, where APOA2 dominates. Taken together, these findings suggest that a subspecies of small and medium spherical HDL particles that contain APOA1, APOA2, and other reverse cholesterol transport proteins may be at least partially responsible for mediating cholesterol metabolism and particle size expansion, and that not all HDL particles are participants in this process.

PLTP & CETP Metabolism and Size Expansion

Of all the HDL proteins that have been monitored to date across the different HDL sizes, PLTP is the only one to exhibit substantial flux from smaller to larger HDL sizes.³² This precursor product relationship from smaller to larger particles was reflected in PLTP’s enrichment curves. The majority of PLTP reside on medium and large HDL with a small amount on very large HDL.^{32, 96} Compartmental modeling revealed that approximately 75% of PLTP on medium HDL was transferred to large and very large particles, while only 25% was directly catabolized from circulation (Figure 3D); suggesting that the majority of PLTP’s phospholipid transfer activity in vivo may be confined to medium HDL, as reported previously.^{32, 97, 98} Thus, this PLTP transfer pathway from medium to large and very large HDL likely represent particles size expansion via PLTP, potentially via particle fusion, and that PLTP remains on the particle as it expands in size.

Similar to PLTP, we also found that CETP is primarily localized on medium and large HDL.³² Given CETP’s primary function is to exchange cholesteryl ester in HDL for triglyceride in APOB-lipoproteins, it is not surprising that CETP dominates in these larger HDL sizes, as these particles have higher amounts of cholesteryl ester compared with smaller HDL.³⁵ Unlike PLTP, however, CETP was not found to transfer between medium and large HDL (Figure 3D). Instead, all of CETP remained on its given size fraction until it was removed from circulation or transferred to a non-HDL compartment. These findings suggest that CETP may not alter the size of the HDL particle in humans in vivo, as has been suggested by some⁹⁹ but not all in vitro studies, ^{99, 100} or that CETP is most stable on medium and large HDL and thus does not remain on the particle if it expands or shrinks out of this preferred size range.

LCAT’s Appearance on HDL is Delayed

LCAT is primarily localized on small spherical HDL.^{26, 32, 55} LCAT esterifies free cholesterol and alters the shape (discoidal to spherical) and increases the size of HDL.^101–103 Thus, it is likely that LCAT activity accounts at least partially for the size expansion pathways detected in the APOA1 metabolism model (Figure 3C). However, LCAT itself may not remain on HDL as it expands in sizes, as no pathways between sizes were detected in the LCAT metabolism model. ³² Unlike other HDL proteins, LCAT metabolism is unique in that it is the only protein monitored to date that has a very delayed appearance on HDL in circulation, up to 6 hours post tracer infusion (Figure 3A, D).³² There are several mechanisms that may account for the delayed appearance of LCAT on circulating HDL, and not all of these possibilities can be delineated using tracer studies alone. First, the LCAT delay may be due to mechanisms controlling protein synthesis, processing, and secretion from the hepatocyte. Second, LCAT may be secreted at a similar rate as other HDL proteins, but spend time outside of systemic circulation, such as in the space of Disse, hepatic sinusoids, interstitial space, or lymphatic vessels, before attaching to circulating HDL.^{83, 104, 105} Third, LCAT may be secreted unattached to HDL particles¹⁰⁶ and enter circulation, where it may interact with other HDL particles, changing their size, before attaching to a circulating HDL particle. In addition to circulating LCAT bound to HDL, future studies of LCAT metabolism would benefit from considering these additional compartments – extravascular LCAT and lipoprotein-free LCAT – to further delineate the mechanisms underlying LCAT metabolism and function.

APOA4’s Distinct Metabolism on Small and Large HDL

APOA4 is unique in that it is the only apolipoprotein in humans that is primarily synthesized by the small intestine, with only minor amounts synthesized by the liver.^107–109 Thus, APOA4 on HDL may serve as a marker of small intestine-derived HDL in plasma.³⁰ Unlike the other HDL proteins studied, APOA4’s enrichment curves and corresponding metabolism are drastically different between smaller sized (prebeta, small HDL) and larger sized (medium, large, very large spherical HDL) HDL particles (Figure 3A).³⁰ APOA4 on small HDL appears in circulation rapidly (by 30 minutes), and is rapidly catabolized, relative to APOA4 on large HDL, whose appearance in circulation was delayed and much more slowly catabolized from circulation (Figure 3A, APOA4 small vs. large). Physiologically, the delayed appearance of APOA4 on large HDL may account for the time necessary for APOA4 on chylomicrons to transfer to large HDL in circulation, while the rapid, early appearance of APOA4 on small HDL suggests that these particles may be directly synthesized and secreted by the small intestine (Figure 3E).

DISADVANTAGES AND ADVANTAGES OF STABLE ISOTOPE KINETIC STUDIES

Disadvantages

One notable disadvantage of tracer kinetic studies in humans is that we can only monitor the metabolism of HDL proteins once they reach plasma circulation. We cannot directly assess the secretion of particles by the liver or other organ systems. Once synthesized and secreted by the hepatocyte, HDL enter the Space of Disse, followed by the hepatic sinusoids and systemic arterial circulation before entering the systemic venous circulation where it is sampled.⁸³ During this trajectory, HDL can interact with LCAT as well as other enzymes and peripheral tissues which can potentially modify its size, shape, and cholesterol content before it is sampled. Thus, kinetic studies need to be complemented with in vitro human tissue studies to directly assess the secretion of HDL particles and the interaction of HDL with peripheral tissues, such as arterial macrophages. Additionally, it would be most advantageous and physiologically relevant for these complementary studies to use whole HDL particles isolated from human plasma, as opposed to synthesized particles that contain only one or two HDL constituents (i.e., APOA1 and phospholipid), as the complex heterogeneous lipid and protein constituents of in vivo HDL likely influence particle stability, metabolism, and function.

Advantages

The biggest advantage of stable isotope kinetic studies is that they allow researchers to label and monitor the metabolism of proteins in humans in vivo; an invaluable resource to learn about biological mechanisms in humans. The proteins and HDL particles in these studies are not subject to modification or alteration by potential isolation, labeling, and reinjection, and the composition – both lipid and protein – are thus more physiologically consistent with in vivo particles. Stable isotope kinetic studies can also label proteins produced by any organ system, such as HDL produced by the liver and the small intestine, and thus can be used to not only monitor the effect of an intervention on liver-derived HDL, but also on small intestine-derived HDL. In addition to HDL proteins, stable isotopes also label proteins on other lipoprotein classes, such as APOB-100, the primary protein on VLDL, IDL, and LDL; APOB48, the primary protein on chylomicrons and remnants; and apo(a), the primary protein on lipoprotein (a) allowing researchers to monitor the metabolism of a given protein across multiple lipoprotein classes.

FUTURE DIRECTIONS

As is the case with any lipoprotein, HDL metabolism and biology cannot be fully understood without context from the other lipoprotein classes. Chylomicrons, VLDL, IDL, LDL, lipoprotein (a), and HDL, although distinct in many functions and physicochemical properties, share lipid and protein constituents indicating potential common or interacting metabolic pathways. Similar to HDL, the majority of APOB lipoprotein kinetic studies have focused on monitoring the metabolism of APOB, but not of other minor proteins. However, as we have demonstrated with HDL, the mass spectrometry technologies are positioned to study not only minor proteins on HDL, but on other lipoprotein classes as well. For example, the metabolism of APOA4 can be monitored on chylomicrons and remnants and across the HDL sizes to determine whether APOA4 on large HDL does in fact originate from chylomicron transfer.

We can also monitor to what extent other proteins such as APOC1 and APOE on VLDL and LDL transfer to HDL and determine which HDL subspecies may be involved in this exchange. It would also be of interest to apply endogenous labeling to monitor the metabolism of APOC3 on HDL as well as VLDL, IDL, and LDL, in the context of the APOC3 antisense trials currently underway to lower plasma triglycerides.¹¹⁰ HDL protein kinetic studies can also be implemented to increase our understanding of the role of different protein and size-based HDL subspecies in HDL biology and function. Protein labeling can be paired with cholesterol labeling to monitor whether there are specific size and protein-based subspecies (such as small and medium-sized HDL with APOA2) that may be more active in cholesterol transport compared to other protein-defined HDL subspecies. Finally, despite the setbacks of CETP inhibitor trials, this HDL protein target remains a pursued target to lower plasma LDL.¹¹¹ Ongoing inhibitor trials should thus take advantage of established endogenous labeling workflows, to monitor and clarify CETP’s metabolic responses to pharmacological agents,^{27, 32} that in turn will contribute to the growing and general knowledge of HDL and lipoprotein biology.

Highlights.

• HDL protein tracer kinetic studies in vivo provide insight into HDL biology

• The transition from radioisotope to stable isotope labeling strategies for tracer kinetics opened access to liquid chromatography-mass spectrometry (LC-MS) technologies

• High resolution/accuracy MS enables the monitoring of tracer in multiple HDL proteins across multiple HDL sizes

• HDL protein tracer data are used to determine the production and catabolism of HDL proteins and the metabolic behavior of these proteins across the HDL sizes, such as whether a protein transfers from smaller to larger HDL particles

• HDL is a heterogeneous particle system comprised of several protein constituents, all of which exhibit a unique metabolism and likely a distinct function in mediating HDL biology and overall heath and disease

NONSTANDARD ABBREVIATIONS AND ACRONYMS

HDL

high-density lipoprotein

APO

apolipoprotein

FCR

fractional catabolic rate

PR

production rate

LCAT

lecithin-cholesterol acyltransferase

CETP

cholesteryl-ester transfer protein

PLTP

phospholipid transfer protein

Leu

leucine

Phe

phenylalanine

D3-Leu

trideuterated leucine

GC-MS

gas chromatography mass sectrometry

LC-MS

liquid chromatography mass spectrometry

SIM

selected ion monitoring

MS/MS

tandem mass spectrometry

MRM

multiple reaction monitoring

PRM

parallel reaction monitoring

HR/AM

high resolution/accuracy

FPLC

fast protein liquid chromatography

VLDL

very-low-density lipoprotein

IDL

intermediate-density lipoprotein

LDL

low-density lipoprotein