Summary and Introduction

Summary

It is well known that, in large populations, HDL-cholesterol levels are inversely related to the risk of atherosclerotic clinical events; however, in an individual, the predictive value of an HDL-cholesterol level is far from perfect. As a result, other HDL-associated factors have been investigated, including the quality and function of HDL in contradistinction to the level of HDL-cholesterol. Regarding their quality, HDL particles are highly heterogeneous and contain varying levels of antioxidants or pro-oxidants, which results in variation in HDL function. It has been postulated that HDL functions to promote reverse cholesterol transport. Recent studies support this role for HDL but also indicate that HDL is a modulator of systemic inflammation. In the absence of inflammation, HDL has a complement of antioxidant enzymes that work to maintain an anti-inflammatory state. In the presence of systemic inflammation, these antioxidant enzymes can be inactivated and HDL can accumulate oxidized lipids and proteins that make it proinflammatory. Under these conditions the main protein of HDL, apolipoprotein A-I, can be modified by reactive oxygen species. This modification impairs the ability of HDL to promote cholesterol efflux by the ATP-binding cassette transporter A-1 pathway. Animal studies and small-scale human studies suggest that measures of the quality and novel functions of HDL might provide an improved means of identifying subjects at increased risk for atherosclerotic events, compared with the current practice of only measuring HDL-cholesterol levels. The quality and function of HDL are also attractive targets for emerging therapies.

Introduction

It is well accepted that HDL-cholesterol levels are inversely related to the risk of clinical events due to atherosclerosis. Careful review of the literature, however, reveals that, even in the original Framingham study (which established the importance of HDL-cholesterol levels in predicting coronary events), more than 40% of events occurred in subjects with normal HDL-cholesterol levels.^[1-5] The Air Force–Texas Coronary Atherosclerosis Prevention Study tested the efficacy of statin therapy in a population with 'average' total cholesterol levels over a period of 5.2 years. In this study, the event rate in the placebo group was 2.1%, 2.9%, and 3.4% for those with HDL-cholesterol levels of >40, 35–39, and ≤34 mg/dl, respectively (>1.03, 0.91–1.00, and ≤0.88 mmol/l, respectively), where values less than 40 mg/dl (<1.03 mmol/l) were considered abnormal.^[4] The event rate in subjects with normal HDL-cholesterol levels who were given placebo was, therefore, approximately two-thirds of the event rate in the subjects with the lowest HDL-cholesterol levels.^[1]

Evolving knowledge that a significant number of cardiovascular events occur in subjects with normal levels of both LDL-cholesterol and HDL-cholesterol^[1,6]has fueled a search for additional biomarkers with better predictive value.^[3,4,7] This review will present evidence that HDL can mitigate or potentiate risk for any given LDL-cholesterol level, and will also consider the mechanisms by which HDL might exert these effects.

LDL and Atherosclerosis Risk

The focus of clinical attention for the past half-century has been on LDL-cholesterol levels. A quarter of a century ago it was reported that the oxidation of LDL injured cells in artery walls, and that HDL inhibited LDL-induced cytotoxicity (Box 1).^[8] Oxidized cholesterol products, especially cholesterol hydroperoxides, turned out to be important in mediating the LDL-induced cytotoxicity.^[9]

The LDL-receptor pathway is the major mechanism for mammalian cells to take in cholesterol. LDL carries not only cholesterol but also a great deal of phospholipid. The oxidation of phospholipid contained within LDL results in the formation of a series of oxidized lipids associated with LDL that might account for many of the early events in atherogenesis.^[10] The foam cell (a cholesterol-loaded cell that is usually a macrophage) is the hallmark of atherosclerotic lesions that are vulnerable to plaque erosion or rupture (Box 2).^[11] Foam cells result, in part, from the uptake of oxidized LDL via a series of receptors, which unlike the LDL-receptor pathway are not regulated by cellular cholesterol content. Some of these receptors (termed 'scavenger receptors') recognize oxidized phospholipids,^[10] a large amount of which has been found in LDL-containing lipoprotein (a) (Lp[a]), which has been associated with increased risk of atherosclerotic events.^[12]

The Role of HDL in Mitigating LDL-induced Risk

Although low HDL-cholesterol has been established as an independent risk factor for atherosclerotic clinical events, it is interpreted in relationship to LDL-cholesterol levels. HDL has been shown to have multiple antiatherosclerotic and anti-inflammatory properties.^[13] The ability of HDL to promote reverse cholesterol transport (i.e. from peripheral tissues to the liver for excretion in the bile) has been thought of as the major function of HDL for 4 decades.

As noted in Box 2, the macrophage has a key role in the accumulation of LDL-derived cholesterol in the artery wall. As also indicated in Box 2, to achieve regression in atherosclerosis, the macrophage-LDL-derived cholesterol must be transported out of the artery wall into the plasma, and then to the liver where it is excreted in the bile. The major mediators of this reverse cholesterol transport from macrophages are apolipoprotein A-I (apoA-I) and HDL. The reverse cholesterol transport hypothesis, therefore, is based on the putative role of HDL in returning excess LDL-derived cholesterol to the liver for excretion. Despite widespread acceptance of this hypothesis, however, the ability of HDL and its major protein, apoA-I, to promote reverse cholesterol transport from macrophages in vivo was only established by Rader and colleagues in 2003.^[14]

This same research group later generated triple-knockout mice, which lack the LDL receptor, apoA-I, and an editing enzyme whose absence results in the accumulation of full-length apoB in LDL.^[15] The HDL-cholesterol levels of mice lacking apoA-I were not significantly different from double-knockout mice that expressed apoA-I but not the LDL receptor or the editing enzyme.^[15] The mice lacking apoA-I had increased atherosclerosis and showed both impaired reverse cholesterol transport and increased inflammation with impaired HDL-mediated, vascular anti-inflammatory function.^[15] The anti-inflammatory properties of HDL and its ability to promote reverse cholesterol efflux, therefore, appear to be linked. This relationship will be discussed later in detail.

In surveying a large number of mouse models, Navab et al.^[1] found that mice that were genetically resistant to developing atherosclerosis had anti-inflammatory HDL (i.e. HDL that inhibited LDL from inducing monocyte chemotactic activity, for example, via monocyte chemotactic protein 1, in cultures of human artery-wall cells; Boxes 1 and 2). In contrast, mice that were genetically susceptible to develop atherosclerosis had proinflammatory HDL (i.e. HDL from these mice enhanced the ability of LDL to induce monocyte chemotactic activity, via monocyte chemotactic protein 1, in cultures of human artery-wall cells). Navab et al.^[1] also reported that there was an inverse relationship between the ability of human HDL to inhibit LDL-induced monocyte chemotactic activity in cultures of human artery-wall cells and its ability to promote cholesterol efflux from human-monocyte-derived macrophages.

HDL, therefore, seems to have a role in removing excess LDL-derived cholesterol from peripheral tissues—as was long thought to be the case—but also seems to be involved in mitigating LDL-induced inflammation. It should be noted that solid, direct clinical evidence is needed to prove convincingly that changes in functional properties of HDL affect atherosclerosis.

Antioxidant Enzymes Associated With HDL

There are a number of enzymes associated with HDL that have antioxidant properties,^[16] including paraoxonase, platelet-activating factor acetylhydrolase, and glutathione peroxidase. These enzymes have the ability to prevent the formation of the proinflammatory oxidized phospholipids described above and to block the activity of those already formed;^[10] however, these oxidized lipids negatively regulate the activities of the HDL-associated enzymes.^17,18] During an acute-phase response in rabbits, mice, and humans, there seems to be an increase in the formation of these oxidized lipids that results in the inhibition of these HDL-associated enzymes and an association of acute-phase proteins such as ceruloplasmin with HDL that renders HDL proinflammatory.^19,20]

Apolipoprotein A-I and Cholesterol Efflux

Most HDL particles contain both apoA-I and apoA-II, and most HDL is said, on the basis of electrophoretic assay, to be αHDL; however, some HDL particles contain only apoA-I and have preβHDL mobility, as assessed by electrophoresis. These particles with preβHDL are considered to be the most active in removing cholesterol from cells.^[21] Cavallero et al.^[22] reported that plasma from fasting patients with well-controlled type 2 diabetes contained reduced levels of preβHDL. Additionally, preβHDL from the diabetic patients was less able to induce cellular cholesterol efflux than similar particles taken from normal controls.^[22] It must be noted that HDL particles with apoA-II also can have preβ mobility on electrophoresis.

Gowri et al.^[23] subsequently reported that HDL from patients with poorly controlled type 2 diabetes exhibited decreased protection against LDL oxidation. Gowri et al.^[23] found that serum triglyceride concentrations were negatively correlated with protection by HDL2 (a subfraction of HDL) in diabetic subjects, but not in controls. HDL2-associated platelet-activating factor acetylhydrolase activity was positively correlated with protection by HDL2 in controls, but not in diabetic subjects. The authors concluded that compositional alterations in HDL2 from subjects with poorly controlled type 2 diabetes might reduce its antiatherogenic properties.^[23]

Some 5 years after these findings, Hazen and colleagues^[24] reported that apoA-I in human serum and in atherosclerotic lesions is selectively nitrated or chlorinated by myeloperoxidase—one of the enzymes released by macrophages in atherosclerotic lesions. Since apoA-I is different from apoB in that it does not bind to the matrix in the artery wall, apoA-I and HDL move in and out of atherosclerotic lesions, whereas apoB and LDL are trapped inside them. Changes in apoA-I were identified in serum taken from patients with atherosclerosis, and in specimens taken from atherosclerotic lesions.^[24] The authors suggested that these changes in apoA-I might be produced by the effect of myeloperoxidase in the artery wall. The result of the nitration or chlorination of apoA-I by myeloperoxidase was a reduced ability of apoA-I to promote cholesterol efflux from macrophages via the ATP-binding cassette transporter A-1 (ABCA-1) pathway.^[24] Specific tyrosine residues in apoA-I were nitrated or chlorinated by myeloperoxidase and were associated with this decreased ability of apoA-I to induce efflux, which suggests that these residues are critical for this process.^[25]

Subsequently, it was learned that apoA-I prepared by recombinant techniques not to contain tyrosine was equally susceptible to damage by myeloperoxidase, indicating that the specific residues were actually a marker of myeloperoxidase modification, but were not necessary for the loss of function.^[26]

When comparing apoA-I tyrosine modification in human atherosclerotic intima and in circulating HDL, Heinecke and colleagues reported sixfold enrichment in the intima and colocalization of the nitrotyrosine with myeloperoxidase.^[27] The HDL from patients with coronary heart disease (CHD) contained twice as much 3-nitrotyrosine as did that from healthy controls.^[27] Since myeloperoxidase was found associated with HDL in atherosclerotic lesions, it was concluded that myeloperoxidase and HDL directly interact in lesions.^[28] It should be noted, however, that it is not clear if chlorinated or nitrated HDL is a marker of this association, rather than a culprit.

Proinflammatory HDL

Navab et al.^[29] studied 27 subjects with angiographically documented coronary atherosclerosis and compared them with 31 controls matched for age and sex. The patients had normal lipid levels and their HDL-cholesterol levels were not different to those seen in the controls. When HDL from the patients was added at equal concentrations to cultures of human artery wall cells, together with a standard control LDL, the patient HDL actually increased the resulting monocyte chemotactic activity.

In contrast, HDL from the controls significantly inhibited the LDL-induced monocyte chemotactic activity. In a cell-free assay (in which the ability of the LDL to induce a fluorescent signal, dependent on oxidative changes, was measured in the presence of a test HDL), HDL from the patients enhanced the oxidative signal, and HDL from the controls inhibited the signal. The patients did not have evidence of an acute illness that could explain an acute-phase response.^[7] Navab et al.^[29,30] postulated that the inflammatory properties of HDL were a form of 'chronic acute-phase response', similar to that characterized by C-reactive protein levels in the top tertile of the normal range (i.e. the range that is seen in humans without obvious infection or inflammation such as active rheumatoid arthritis).

Ansell et al.^[3] calculated an 'HDL inflammatory index' for 26 patients with CHD or CHD equivalents who met guidelines for statin therapy but who were not on a statin at the time of enrollment. Blood lipids in these patients prior to therapy were relatively normal (total cholesterol 202 ± 28 mg/dl [5.22 ± 0.70 mmol/l]; HDL cholesterol 57 ± 13 mg/dl [3.87 ± 0.10 mmol/l]; LDL cholesterol 118 ± 24 mg/dl [3.05 ± 0.62 mmol/l]; and triglycerides 125 ± 64 mg/dl [1.41 ± 0.72 mmol/l]).

The HDL inflammatory index was calculated by comparing the monocyte chemotactic activity generated by a standard control LDL in the absence and presence of the test HDL. In the absence of the test HDL, the monocyte chemotactic activity was normalized to 1.0. If the monocyte chemotactic activity increased upon addition of the test HDL, the HDL inflammatory index was >1.0 and the test HDL was classified as proinflammatory. If the monocyte chemotactic activity decreased, the inflammatory index was <1.0 and the HDL was classified as anti-inflammatory.

After the initial blood sample was taken the patients were started on simvastatin (40 mg daily) and, after 6 weeks, a second blood sample was collected. The plasma lipid levels after treatment were significantly improved (total cholesterol 154 ± 22 mg/dl [3.98 ± 0.57 mmol/l]; HDL-cholesterol 61 ± 14 mg/dl [1.58 ± 0.36 mmol/l]; LDL-cholesterol 73 ± 24 mg/dl [1.89 ± 0.62 mmol/l]; triglycerides 99 ± 22 mg/dl [1.12 ± 0.25 mmol/l]). The HDL inflammatory index prior to starting simvastatin was 1.38 ± 0.91 in the patients and 0.38 ± 0.14 in the controls (P = 0.000015). The HDL inflammatory index in the patients decreased from 1.38 ± 0.91 to 1.08 ± 0.71 (P = 0.002) 6 weeks after taking simvastatin. After simvastatin treatment, therefore, the patient's HDL inflammatory index significantly improved, but remained proinflammatory (on average), despite a very significant improvement in plasma lipid levels.^[3]

Ansell et al.^[3] also studied a group of 20 subjects with elevated HDL-cholesterol levels, all of whom were documented to have CHD. The average HDL-cholesterol level for this group was 95 ± 14 mg/dl (2.46 ± 0.36 mmol/l) and the lowest HDL-cholesterol value was 84 mg/dl (2.17 mmol/l). The average total cholesterol level for each patient in this group was 217 ± 35 mg/dl (5.61 ± 0.91 mmol/l); the average LDL-cholesterol level was 108 ± 34 mg/dl (2.80 ± 0.88 mmol/l); and average triglyceride levels were 89 ± 44 mg/dl (1.00 ± 0.50 mmol/l).

The HDL inflammatory index in this high-HDL group was 1.28 ± 0.29, compared with 0.35 ± 0.11 (P = 1.7 × 10^-14) in controls matched for age and sex (a different control group to the one mentioned above). The patients with CHD and high HDL-cholesterol levels, therefore, had proinflammatory HDL. It should be noted that some experimental studies in mice have suggested that serum amyloid A peptides in mice improve efflux capacity of HDL and have antiatherosclerotic effects.^[31]

Clinical Implications

Although 3 decades of lipid-altering therapy have targeted LDL-cholesterol levels, the focus of research in this area is now shifting to HDL.^{[32,33,34,35]}

One approach has been to use oral apolipoprotein mimetic peptides. Navab et al.^[36] reported that oral administration of an apoA-I mimetic peptide (D-4F) reduced atherosclerosis in mice without changing plasma cholesterol levels. In these studies, LDL-receptor-null mice on a Western diet (high cholesterol and high fat) and apoE-null mice (which lack apoE and, even on a low-fat diet, have hyperlipidemia and atherosclerotic lesions closely resembling human lesions) on a chow diet were found to have proinflammatory HDL that was converted to anti-inflammatory HDL after oral administration of D-4F. This process did not involve significant changes in HDL-cholesterol or total plasma cholesterol levels.^[36] The mechanism of action of D-4F in mice was reported to involve the formation of preβHDL with improvement in HDL-mediated cholesterol efflux and improvement in reverse cholesterol transport from macrophages in vivo.^[37] Treatment of apoE-null mice and monkeys with a combination of oral D-4F and pravastatin synergistically rendered HDL anti-inflammatory and caused lesion regression in old apoE-null mice.^[38]

The physical characteristics of apoA-I mimetic peptides that cause the sequestration of proinflammatory lipids—which inhibit antioxidant enzymes associated with HDL—are thought to have a key role in their efficacy.^[39] Ou et al.^[40] recently reported that oral D-4F restored endothelium dependent and endothelial nitric oxide synthase dependent vasodilation in direct relationship to the duration of treatment. It was also found to reduce the wall thickness of small arteries in as little as 2 weeks, in LDL-receptor-null mice with pre-existing disease fed a Western diet. D-4F had no effect on total cholesterol or HDL-cholesterol concentrations but it did reduce proinflammatory HDL levels. In addition, plasma myeloperoxidase concentrations were not altered by D-4F, but myeloperoxidase association with apoA-I was reduced, as were the levels of 3-nitrotyrosine in apoA-I.^[40] In LDL-receptor-null mice that also lacked apoA-I, D-4F increased endothelium dependent and endothelial nitric oxide synthase dependent vasodilation, but it did not reduce the thickness of the walls of small arteries as it did in mice with apoA-I.^[40,41]

In addition to apoA-I mimetic peptides, an oral peptide with a sequence from another HDL-associated protein (apoJ) has been shown to render HDL anti-inflammatory in mice and monkeys, and to reduce atherosclerosis in apoE-null mice significantly.^[42] Oral peptides, which are too small to form helical structures, have also been shown to exert the same effects.^[43] The common mechanism by which these different peptides render HDL anti-inflammatory has been postulated to be their common ability to sequester and remove oxidized lipids that inhibit enzymes associated with antioxidant HDL.^[44] This common mechanism has been shown to operate in peptides that are apoA-I mimetics or apoJ mimetics, and also in peptides that are too small to contain a helix like those found in apoA-I and apoJ, and which are not mimics of HDL-associated proteins. Shah and Chyu^[44] recently reviewed the potential use of apolipoprotein mimetic peptides in atherosclerosis management.

Conclusions

Promising HDL-based therapies, beyond the use of statins to lower LDL levels relative to HDL levels, include raising HDL-cholesterol levels by inhibition of cholesterol ester transfer protein, raising HDL-cholesterol levels by treatment with fibrates or peroxisome proliferative activating receptor agonists, and—for acute treatment—infusion of apoA-I_Milano ( Table 1 ). Examination of the quality and function of HDL is an emerging area of research, important for improving the detection of people at risk for atherosclerotic events and for targeting novel therapies. Additionally, it might be more important to improve HDL metabolism, composition, and, therefore, function as opposed to concentrating on efforts to merely increase the HDL-cholesterol levels. This can be done in at least three ways: by increasing the ability of HDL to sequester and destroy proinflammatory lipids; by protecting apoA-I against oxidant damage by myeloperoxidase and other oxidant pathways so that apoA-I is maximally able to promote reverse cholesterol efflux; and by increasing the levels of preβHDL formation and cycling, improving the balance between proinflammatory and anti-inflammatory HDL. It is likely that improving HDL function will produce even more favorable outcomes in patients treated with therapies that raise their HDL-cholesterol levels.