OXIDATIVE MODIFICATION OF HDL BY LIPID ALDEHYDES IMPACTS HDL FUNCTION

Abstract

Reduced levels of high-density lipoprotein (HDL) cholesterol correlate with increased risk for atherosclerotic cardiovascular diseases and HDL performs functions including reverse cholesterol transport, inhibition of lipid peroxidation, and suppression of inflammation, that would appear critical for cardioprotection. However, several large clinical trials utilizing pharmacologic interventions that elevated HDL cholesterol levels failed to provide cardioprotection to at-risk individuals. The reasons for these unexpected results have only recently begun to be elucidated. HDL cholesterol levels and HDL function can be significantly discordant, so that elevating HDL cholesterol levels may not necessarily lead to increased functional capacity, particularly under conditions that cause HDL to become oxidatively modified, resulting in HDL dysfunction. Here we review evidence that oxidative modifications of HDL, including by reactive lipid aldehydes generated by lipid peroxidation, reduce HDL functionality and that dicarbonyl scavengers that protect HDL against lipid aldehyde modification are beneficial in pre-clinical models of atherosclerotic cardiovascular disease.

Overview

Plasma lipoproteins play a crucial role in whole body physiology as carriers of lipids, and disorders related to lipid and lipoprotein metabolism have life-threatening consequences. It is now well established that elevated levels of low-density lipoprotein cholesterol (LDL-C) is a causative risk factor for atherosclerotic cardiovascular diseases (ASCVD). In contrast, whether reduced levels of high-density lipoprotein cholesterol (HDL-C) or its major apolipoprotein, apoA1, have a causative role in ASCVD remains controversial. Reduced levels of HDL-C correlate with increased risk for ASCVD. Furthermore, high-density lipoprotein (HDL) performs functions including reverse cholesterol transport, inhibition of lipid peroxidation, and suppression of inflammation, that would appear critical for cardioprotection. However, several large clinical trials utilizing pharmacologic interventions that elevated HDL-C levels failed to provide cardioprotection to at-risk individuals. The reasons for these unexpected results have only recently begun to be elucidated. HDL-C levels and HDL function can be significantly discordant, so that elevating HDL-C levels may not increase HDL functional capacity, particularly under conditions that cause HDL to become oxidatively modified, resulting in HDL dysfunction. Here we review evidence that oxidative modifications of HDL, including by reactive lipid aldehydes generated by lipid peroxidation, reduce HDL functionality and that dicarbonyl scavengers that protect HDL against lipid aldehyde modification are beneficial in pre-clinical models of ASCVD.

Potentially cardioprotective role of HDL

The Framingham Heart Study reported an inverse relationship between plasma HDL-C and the incidence of coronary heart disease [1] and numerous studies since then have confirmed this inverse correlation [2, 3]. In animal models of atherosclerosis, infusion of HDL [4, 5] or transgenic expression of apoA1 [6, 7] decreases atherosclerotic lesion areas. Transfusion of apoA1 or reconstituted HDL into humans also conferred cardioprotection in clinical trials [8, 9]. At least three functions of HDL may contribute to its potential anti-atherosclerotic effects: its ability to facilitate reverse cholesterol transport, its ability to protect other lipoproteins from peroxidation (reviewed in [10]) and its ability to dampen inflammatory responses (reviewed in [11][12]) (Figure 2).

Figure 2.

Effects of oxidative modification on HDL function. Oxidative modifications of HDL include modification of amino acid residues by reactive oxygen species (ROS) formed by MPO, by isocyanate formed non-enzymatically by oxidation of urea, or by various reactive lipid aldehydes formed by lipid peroxidation. These lipid aldehydes include malondialdehyde (MDA), isolevuglandin (IsoLG), acrolein (ACR), 4-oxononenal (ONE), and 4-hydroxynonenal (HNE). The sites of apoA1 amino acid residue modification by each compound is shown as well as other known HDL protein modifications.

It is important to recognize that HDL is a collection of highly heterogenous lipoparticles. ApoA1 is the major protein constituent of HDL, comprising about 70% of its protein mass, and appears to play critical roles in all of three of the major HDL functions. ApoA1 serves as a scaffold for other proteins to associate with HDL and also interacts with cellular transporters that facilitate the transfer of lipids to and from HDL. More than 85 different proteins at least transiently associate with HDL including lecithin-cholesterol acyltransferase (LCAT), cholesterol ester transfer protein (CETP), phospholipid transfer protein (PLTP), and paraoxonase-1 (PON-1) [13–15].

Reverse cholesterol transport.

The various forms of HDL (lipid-poor apoA1, HDL2, and HDL3) serve as acceptors for cholesterol from peripheral tissue to transport cholesterol from these cells to the liver for excretion or redistribution [16]. Reverse cholesterol transport begins when apoA1 secreted from the liver and intestine interacts with ATP Binding Cassette Subfamily A Member 1 (ABCA1) on cells such as macrophages to accept phospholipid and free cholesterol (Figure 2). LCAT associated with apoA1 then converts free cholesterol to cholesterol ester (CE), forming spherical HDL particles. Spherical HDLs accept additional free cholesterol via interaction with ATP Binding Cassette Subfamily G Member 1 (ABCG1) and Scavenger Receptor BI (SRBI) on donor cells. CETP and PLTP exchange cholesterol ester (CE) from HDL for triglyceride (TG) from LDL and VLDL. Finally, HDL interacts with SRBI on the surface of hepatocytes to deliver CE to the liver [16].

Effective reverse cholesterol transport requires HDL to penetrate the subendothelial space to interact with macrophages that have taken up cholesterol from a variety of sources including apoptotic cells. The small particle size of HDL (7.3–13 nm) compared to other lipoproteins like LDL (21.2–27 nm) and VLDL (27–200 nm) may be an important feature of HDL in this regard. Two mechanisms for transportation of HDL into subendothelial space have been reported: i) non-specific passive penetration through pores in the endothelial cell monolayer [17], and ii) active transcytosis through endothelial cells via active transport requiring ABCA1, ABCG1, SRBI, ecto-F1-ATPase, and caveolin-1 [18–20]. After collecting cholesterol and other toxic molecules from tissue, HDL enters the lymphatics to return to circulation and deliver cholesterol to the liver for metabolism and disposal. HDL facilitated cholesterol transport from peripheral tissues reduces cholesterol accumulation in the intima-media of arteries [21]. Measurements of whole body reverse cholesterol transport (RCT) capacity can be used to assess risk for development of ASCVD [22]. However, whole body RCT is too cumbersome for regular clinical use, so that surrogate measures such as measuring the capacity of HDL isolated from patients to facilitate the efflux of cholesterol from cholesterol-loaded macrophages are used instead. These ex vivo cholesterol efflux assays correlate well with whole body RCT [23] and the cholesterol efflux capacity of HDL inversely associates with ASCVD [24–26].

Antioxidation.

Oxidation of LDL phospholipids to form lipid peroxides (LOOH) and subsequent formation of lipid aldehydes that modify apoB (oxLDL) play critical roles in the development of atherosclerosis [27]. The evidence that HDL exerts potent antioxidative effects, especially to prevent oxidation of LDL, has been reviewed by Brites et al [10]. Isolated LDL is readily oxidized in vitro using copper, but coincubation with HDL or with PON1 largely inhibits to formation of LOOH or the breakdown of LOOH to reactive lipid aldehydes such as MDA under these conditions [28, 29]. Although PON1 is a relatively minor protein component of HDL, it appears to be essential for the antioxidative effects of HDL [10, 28, 30, 31]. PON1 competes with MPO for binding to apoA1 [32], which may protect HDL phospholipids from peroxidation, as MPO is a major source of reactive oxygen species that cause lipid peroxidation in circulation [33]. Other HDL proteins with potential antioxidant properties include apoA1, apoA2, apoA5, apoE, apoM, apoD, apoF, apoJ, apo L-1 and serum amyloid A [10]. HDL can also readily catalyze the transfer of CE peroxides and hydroxides from cells and from LDL [34–36], and cholesteryl ester transfer protein (CETP) appears to be critical to this process [37], as does HDL phospholipid composition and the rigidity of HDL surface [31].

Anti-inflammation.

OxLDL is pro-inflammatory, so that HDL inhibition of LDL oxidation exerts an anti-inflammatory effect. However, HDL inhibition of inflammation extends to other stimuli besides oxLDL. HDL inhibits lipopolysaccharide (LPS)-induced cytokine secretion, including reducing levels of tumor necrosis factor‐α (TNF‐α), interleukin‐6 (IL‐6) and interferon‐γ (IFN‐γ) [38]. HDL binding and neutralization of LPS only partially accounts for these effects [39]. HDL also inhibits induction of CCL2 by co-culture of endothelial and smooth muscle cells [41], inhibits endothelial cell expression of adhesion molecules in response to TNFα [42, 43], and suppresses the adhesion of monocytes to endothelial cells [44]. Both apoA1 and HDL phospholipids appear to contribute to these various anti-inflammatory effects [45, 46]. Some of the anti-inflammatory effects require interaction of apoA1 with ABCA1 and cholesterol efflux [21, 40]. Of note, the cholesterol efflux mediated by HDL can drive both anti-inflammatory and pro-inflammatory effects, although the anti-inflammatory effects predominate [40, 47].

HDL function versus HDL cholesterol Levels in Cardioprotection

The apparent cardioprotective effects of HDL led to development of pharmacological interventions that increase HDL-C levels in individuals with low HDL-C levels. Both inhibition of CETP and administration of niacin raise HDL-C levels, but large scale clinical trials of these interventions found no beneficial effect [48–50], calling into question the role of HDL in cardioprotection [51–53].

The premise that pharmacologically increasing HDL-C levels in persons with low HDL-C levels enhances their HDL function may not be valid (reviewed in [54]). For instance, CETP drives the net transfer of unoxidized cholesterol esters from HDL to LDL but also drives the net transfer of oxidized cholesterol esters from LDL to HDL [36, 37]. Thus, CETP inhibition may raise HDL levels of unoxidized cholesterol ester at the cost of reducing hepatic degradation of oxidized cholesterol esters via HDL. CETP inhibition using torcetrapib also appears to decrease PON1 activity [55], a critical antioxidative function of HDL.

A large study including both healthy individuals and those with angiographically confirmed coronary artery disease examined correlates with carotid artery intima-media thickness. This study found that lower cholesterol efflux capacity strongly predicted coronary artery disease status, but that HDL-C and apoA1 levels account for less than 40% of the observed variation in measured cholesterol efflux capacity [26]. A second study followed individuals for a median time of 9.4 years after making initial measurements including HDL-C and cholesterol efflux capacity. This study again found a strong inverse relationship between cholesterol efflux capacity and coronary disease and poor correlation between efflux capacity and HDL-C levels [56].

Modifications of HDL alter its functional capacity.

Individuals with risk factors for ASCVD such as familial hypercholesterolemia, chronic kidney disease, and diabetes often have reduced HDL function that poorly correlates with changes in HDL-C levels [57–62] suggesting that under these conditions, HDL might undergo modificationsthat alter its functionality.

Such modifications could include changes in the protein and lipids content of HDL or covalent modifications of HDL proteins and phospholipids. A well-established example of the first type of modification is “acute-phase HDL”, where an infection or other inflammatory stimulus invokes wholesale changes in the proteins and lipid content of HDL. Notable changes in acute-phase HDL include reduction in apoA1 and marked increase in serum amyloid A; reduction in cholesterol esters with concomitant increases in free cholesterol, triglycerides, and free fatty acids; and reduced levels of CETP, LCAT, and PON1 with increased levels of secretory phospholipase A2 and LPS-binding protein (reviewed in [63] and [64]). The net effect of these changes is that acute-phase HDL has significantly altered function compared to normal HDL. Significant alterations in the HDL proteome, including enrichment in serum amyloid A, have been reported in patients with coronary artery disease and other chronic inflammatory conditions (reviewed in [14]).

We hypothesize that oxidative modification of HDL proteins and phospholipids also contributes to HDL dysfunction and evidence for this hypothesis will be the focus on the remainder of this review. In serum, the majority of cholesterol linoleate hydroperoxides [65], phosphatidylcholine hydroperoxides [66], and F₂-isoprostanes [67] (stable end-products of arachidonic acid peroxidation) are associated with HDL rather than the other lipoproteins. Mechanisms driving HDL enrichment with LOOH likely include LCAT-facilitated transfer of oxidized acyl chains from phospholipids to free cholesterol to form oxidized cholesterol esters [68], CETP-facilitated transfer of oxidized lipids from LDL to HDL [36, 37], and binding of extracellular MPO to apoA1 [33, 69]. The capture of oxidized lipids and MPO by HDL likely facilitates their removal and protects the organism under conditions of low oxidative stress. However, pathophysiological conditions may overwhelm the antioxidative capacity of HDL, leading to its oxidative modification and loss of function.

Multiple mechanisms for Oxidative modification of HDL

In vitro oxidation of apoA1 or HDL using MPO results in oxidative modification and loss of function that markedly resembles that seen in apoA1 or HDL extracted from atherosclerotic lesions [69–74], consistent with MPO being a major driver of HDL oxidative modification in vivo. After activation by hydrogen peroxide, MPO can utilize substrates including chloride (Cl^-), bromide (Br^-), thiocyanate (SCN^-), and nitrite ions (NO₂^-) to produce reactive oxygen species (ROS) such as hypochlorous acid (HOCl), hypobromous acid (HOBr), hypothiocyanous acid (HOSCN), isocyanate (CNO-), and nitrogen dioxide radical (^●NO₂) [75]. Other ROS, such as superoxide (O₂^●-), the hydroxyl radical (OH^●) and peroxynitrite (ONOO^-), can be generated by other enzymes including NADPH oxidases and xanthine oxidase. These ROS can directly oxidize amino acids including cysteine (Cys), methionine (Met), tyrosine (Tyr), tryptophan (Trp), or lysine (Lys) to generate modified amino acids including disulfide bonds (Cys-Cys crosslinks), oxidized Met (oxMet), oxidized Trp (oxTrp), 3-chlorotyrosine (3-Cl-Tyr), nitrotyrosine (NO₂-Tyr), and tyrosine dimers (Tyr-Tyr), N-chloroamines, and carbamyl-Lys.

In addition to directly oxidatively modifying HDL amino acids, ROS also indirectly modify HDL proteins by generating lipid peroxides (LOOH) that fragment into reactive lipid aldehydes. ^●NO2 is particularly adept at the peroxidation of polyunsaturated fatty acids [76, 77], as is ONOO^- [78]. LOOH undergoes secondary reactions to form reactive lipid aldehydes including malondialdehyde (MDA), 4-hydroxy-nonenal (HNE), 4-oxo-nonenal (ONE), isolevuglandins (IsoLG), acrolein (ACR), and methylglyoxal (MGO) (Figure 2). These reactive lipid aldehydes then modify amino acids including Lys, Cys, histidine (His), or arginine (Arg) (reviewed in [79]). MDA and IsoLG target primary amines including Lys and phosphatidylethanolamine (PE), while HNE preferentially targets cysteine (Cys) and histidine (His), then arginine (Arg), Lys, and PE. ONE shows similar preferences as HNE, except that ONE modifies Lys and PE to a greater extent. ACR preferentially modifies Cys, but also modifies Lys, PE, and to a lesser extent His. MGO preferentially modifies Arg.

Implicating oxidative modification of HDL as a key contributor to HDL dysfunction and ASCVD requires the following evidence: i) demonstration that oxidative modification of HDL occurs in vivo; ii) demonstration that in vitro oxidation or direct modification of HDL by lipid aldehydes causes loss of HDL function; iii) demonstration that selective elimination of amino acid oxidation (e.g. by substitution of oxidation-resistant amino acids for oxidation-prone amino acids) or aldehyde modification (e.g. by using aldehyde scavengers that block modification) protects HDL function under conditions that normally result in loss of function; and iv) that restoration of HDL function in this manner decreases disease. We examine the current state of such evidence below.

HDL is oxidatively modified in vivo.

A substantial body of evidence demonstrates that amino acids of HDL proteins undergo oxidative modification of HDL during atherosclerosis. Approximately 20% of apoA1 isolated from atherosclerotic lesion includes oxTrp72, and oxTrp72 levels correlate with increased cardiovascular disease risk [80]. Zheng et al found apoA1 3-Cl-Tyr levels to be about 2.5-fold higher in CVD patients compared to controls, and apoA1 NO₂-Tyr levels to be 1.5-fold higher [70]. Bergt et al also found that HDL 3-Cl-Tyr levels are higher in patients with coronary artery disease than in control individuals, and that HDL isolated from atherosclerotic lesions had higher 3-Cl-Tyr levels compared to HDL isolated from serum [81]. Similarly, HDL NO₂-Tyr is higher in HDL isolated from atherosclerotic lesions than HDL isolated from serum or plasma [70, 81, 82]. The carbamyl-Lys content of HDL isolated from the atherosclerotic lesion correlates with its 3-Cl-Tyr content and with lesion severity [83]. In apoA1 isolated from human aortic atheroma, 16 of the 21 Lys residues show some amount of carbamylation [84]. This pattern is consistent with the carbamylation reaction primarily occurring with lipid-poor apoA1 rather than HDL and with the isocyanate for the carbamylation reaction being formed non-enzymatically from urea rather than enzymatically by MPO from thiocyanate [84].

In individuals with familial hypercholesterolemia (FH), a strong risk factor for early onset CAD, levels of HDL IsoLG-protein adducts, HDL IsoLG-PE adducts, MDA-apoAI adducts, and HDL ONE-protein adducts are elevated from two to five-fold in plasma compared to that of control individuals [85–87]. When HDL was isolated from patients with coronary artery disease, those with acute coronary syndrome showed 40% higher levels of MDA adducts than those with stable angina [88]. In patients undergoing carotid endarterectomy, arterial lesion HDL levels of MDA were 3.6-fold higher than plasma HDL levels of MDA [89]. Individuals with chronic kidney disease (CKD) have elevated risk for atherosclerosis and HDL HNE-adducts are increased in CKD patients compared to controls [90]. Immunohistochemistry showed colocalization of ACR adducts and apoAI in atherosclerotic lesions [91], but to the best of our knowledge, increased levels of ACR-apoAI or ACR-HDL in patients with ASCVD or at risk for ASCVD have not been reported.

Does oxidative modification of HDL correlate with reduced HDL function?

In vitro oxidation of isolated HDL, especially when HDL is oxidized enzymatically using MPO, leads to reductions in HDL function including impairments in cholesterol efflux, anti-apoptotic effects, anti-peroxidation effects, and anti-inflammatory effects [32, 69–74, 92]. Modification of apoA1 or isolated HDL with reactive lipid aldehydes typically shows similar effects as MPO oxidation.

Reduced Cholesterol efflux.

In vitro modification of HDL with MPO reduces its capacity to facilitate ABCA1-mediated cholesterol efflux from macrophages [70, 91]. Early studies demonstrated that both MPO/H₂O₂/Cl^- and MPO/H₂O₂/NO₂^- mediated oxidation of apoA1 led to Tyr modification, with Tyr192 being the primary Tyr modified [91]. Of interest, only 3-Cl-Tyr192 modification and not NO₂-Tyr192 modification correlated with reduced cholesterol efflux [91]. Oxidation of HDL with peroxynitrite (ONOO^-) had no significant effect on cholesterol efflux despite resulting in significant apoA1 nitration of Tyr [70, 91]. Subsequent studies showed that mutation of apoA1 Tyr residues to oxidation-resistant phenylalanine (Phe) residues did not protect apoA1 from the reductions in ABCA1-mediated cholesterol efflux capacity induced by MPO/H₂O₂/Cl^- [93], suggesting that 3-Cl-Tyr may simply serve as a surrogate indicator for other oxidative modifications. Follow-up studies showed that MPO/H₂O₂/Cl^- oxidation of HDL also results in Trp72 oxidation, and mutation of Trp72 to Phe protected apoA1 from MPO-mediated reduction in its ABCA1-mediated cholesterol efflux capacity [80]. Carbamylation of HDL causes cholesterol to accumulate in macrophages as lipid droplets by increasing SR-BI dependent uptake, but carbamylation of HDL does not alter ABCA1-dependent cholesterol efflux from cholesterol loaded macrophages [83].

Reactive lipid aldehydes can also cause loss of efflux capacity by HDL. Modification of apoA1 with 0.3- and 3-molar equivalents of IsoLG per apoAI molecule reduces cholesterol efflux from cholesterol loaded macrophages by about 20% and 30%, respectively [87]. Modification of HDL with 5- to 50-molar equivalent MDA reduces efflux from 13% to 75% respectively [89]. Modification of apoA1 with 20- or 50-molar equivalent ACR reduced efflux about 25% and 50%, respectively [94, 95]. In contrast, incubation of apoA1/HDL with various concentrations of HNE, ONE, MGO, or glyoxal has no effect on cholesterol efflux [62, 89].

Why IsoLG, MDA, and ACR modification of apoA1 inhibits cholesterol efflux, while apoA1 modification with other lipid aldehydes like HNE, ONE, and MGO do not, remains unclear. Simply crosslinking apoA1 appears insufficient to reduce cholesterol efflux, as ONE, HNE, and MGO can also crosslink apoA1 when sufficient concentrations are used. The two most C-terminal alpha helices of apoA1, Helix 9 (residues 209–219) and helix 10 (residues 220–243), are critical for its interaction with ABCA1 [96–98] and these two helices include three Lys (Lys226, Lys238, Lys239), one Arg (Arg215) but no His residues residues that could be targets for aldehyde modification. Of note, the loss of cholesterol efflux with ACR modification was directly proportionate to the extent of Lys226 ACR modification [94]. Modifying apoA1 with 20 mol equivalents of MDA leads to extensive crosslinking of Lys226 to Lys206 and to Lys182 [89], and of Lys239 to Lys208 and Lys182, while also forming MDA monoadducts on Lys118, Lys133, and Lys195 [89]. The specific Lys residues of apoA1 modified by IsoLG have not yet been reported. While incubation of HNE with apoA1 modifies all five His residues, no modification of Lys were detected [87, 89]. Incubation of MGO with apoA1 leads to extensive Arg modification but not Lys modification [89]. These results suggest that the extent of apoA1 Lys modification, especially Lys226 modification, may be a key determinant of altered cholesterol efflux capacity. However, ONE modifies Lys226 (as well as Lys12, Lys23, and Lys96), yet fails to inhibit cholesterol efflux [62]. Furthermore, carbamylation of apoA1 in vitro results in extensive Lys226 modification, along with modification of Lys23, Lys40, Lys94, and Lys 118 [84], but carbamylation of apoA1 does not result in altered cholesterol efflux [83]. Thus, the mechanisms underlying reduced efflux after apoA1 modification with specific lipid aldehydes, and the extent to which amino acid oxidation and aldehyde modification interact to reduce cholesterol efflux needs further elucidation.

Reduced antioxidation.

Although HDL (and especially HDL-associated PON1) protects LDL from in vitro oxidation by copper or free radical initiators, early studies showed that HDL-associated PON1 begins to lose its activity as oxidation continued. This inhibitory effect can be recapitulated by addition of peroxidized lipid preparations to PON1 [99].

In vitro oxidation of HDL with MPO/H₂O₂/NO₂^- or MPO/ H₂O₂/Cl^- leads to a drop in PON1 activity, as well as oxidation of Tyr71 of PON1 to generate NO₂-Tyr or 3-Cl-Tyr adducts, respectively [32]. These two events appear mechanistically related, as Tyr71 interacts with cholesterol within HDL to stabilize PON1 binding to HDL and substitution of Tyr71 with other amino acids such as alanine, aspartic acid, or Lys reduces PON1 activity [32]. In samples from individuals presenting for diagnostic coronary angiography and matched healthy controls, the levels of HDL 3-Cl-Tyr inversely correlated with PON1 activity, further supporting a causative role of Tyr71 modification in reducing PON1 activity in vivo [32].

Reactive lipid aldehydes may also contribute to PON1 inactivation. Treating HDL with the peroxynitrite generator SIN1 induces lipid peroxidation, as measured by the formation of PC with fragmented acyl chains with omega aldehyde moieties (“core aldehydes”), and causes a rapid drop in PON1 activity [100]. Direct addition of PC with core aldehydes only minimally inhibited PON1 activity [100], as would be expected given the poor reactivity of these simple aldehydes. Whether SIN1 treatment of HDL generates IsoLG or MDA adducts has not been determined but would be expected. Treatment of HDL with 5 mol equivalent of IsoLG (relative to apoA1) results in approximately 50% inhibition of HDL PON1 activity [101]. ApoA1 allosterically activates PON1 [101, 102], raising the possibility that IsoLG modification of apoA1 caused the reduced PON1 activity. However, IsoLG modified-HDL stimulates the activity of subsequently added recombinant PON1 to nearly the same extent as unmodified HDL does, suggesting that IsoLG-modification of apoA1 has minimal effect on apoA1 allosteric activation of PON1 [101]. In contrast, IsoLG modification of recombinant PON1 markedly inhibits its activity and treatment of HDL with MPO/H₂O₂/NO₂^- results in IsoLG modification of PON1 [101], suggesting that the reduced HDL PON1 activity after exposure to oxidants results at least in part from IsoLG modification of PON1. Treatment of HDL modified by 5 mol equivalents of MDA (relative to apoA1) results in approximately 50% inhibition of HDL-mediated activation of PON1 subsequently added to the HDL, suggesting that MDA indirectly inhibits PON1 by blocking apoA1-mediated activation [85]. Whether MDA can also inhibit PON1 by direct modification of its Lys was not determined. Treatment of HDL with ACR results in dose dependent inhibition of HDL PON1 activity [103]. Whether this inhibition of PON1 activity results from ACR directly modifying PON1 or modifying apoA1 to prevent its allosteric activation of PON1 is unknown. These same investigators showed that in individuals with end-stage renal disease, hemodialysis treatment decreased plasma ACR levels and increased PON1 activity [103], suggesting that ACR (and other reactive lipid aldehydes removed by hemodialysis) play an important role in determining PON1 activity rates in vivo [103]. To the best of our knowledge, no studies have looked at whether levels of reactive lipid aldehyde adducts detected for HDL proteins or PON1 are inversely correlated to activity of the sample. Such studies would be important validation of a causal role for lipid aldehyde modification of HDL in reducing PON1 activity.

Reduced suppression of inflammation and pro-inflammatory effects.

Oxidation of HDL with MPO/H₂O₂/Cl^- not only blocks the ability of HDL to inhibit TNFα stimulated VCAM-1 surface expression by endothelial cells, but actually further potentiates this response [69]. MPO-oxidized HDL can stimulate NFκB activation and VCAM-1 surface expression on its own, even in the absence of TNFα, while non-oxidized HDL has no such effect [69]. Mutation of various Trp, Tyr, and Met residues of apoA1 to oxidation resistant residues (Phe and Val) do not protect apoA1 from MPO-mediated conversion to its proinflammatory form, suggesting that oxidation of these amino acids does not play a role in this conversion [69].

While native HDL suppresses LPS-induced secretion of Tnfα and Il-1β, HDL modified by 1 molar equivalent IsoLG not only fails to suppress LPS-induced inflammatory cytokine secretion, but in fact significantly potentiates LPS-induced IL-1β secretion [87], suggesting that IsoLG-HDL could play a similar role in driving inflammation as oxidized LDL. ONE modification of HDL also ablates the ability of HDL to suppress LPS-induced Tnfα, IL-1β, and IL-6 production, but ONE-HDL does not further potentiate LPS-induced cytokine secretion [62]. Modification of HDL by ACR, MDA, or HNE did not alter its ability to suppress LPS-induced macrophage inflammatory cytokine secretion [104].

One component of IsoLG-HDL, IsoLG modified phosphatidylethanolamine (IsoLG-PE), is able to directly stimulate TNFα secretion in a NFκB-dependent manner via activation of the receptor for advanced glycation endproducts (RAGE) [86] and induces monocyte adhesion to endothelial cells [105]. Interestingly, IsoLG modifies PE to a greater extent than lysine [106], which may be why even low concentrations of IsoLG suffice to turn HDL into an inflammatory particle. PE modified in vitro with MDA, HNE, and ONE, but not with ACR or MGO also induces monocyte adhesion to endothelial cells [105]; however, whether such PE adducts form when HDL is exposed to these aldehydes or to oxidants has not yet been investigated.

Oxidized LDL promotes the trapping of macrophages in atherosclerotic lesions by inhibiting macrophage migration from the plaque [107]. HDL normally promotes macrophage migration in vitro, but HDL modified with ACR or MDA fails to promote migration [104]. HNE modification has no effect. Whether IsoLG-modified HDL alters macrophage migration is untested.

In vivo correlations between oxidative modification and HDL function.

As noted previously, individuals with ASCVD or with strong risk factors for ASCVD have reduced HDL function [57–62]. Although increased HDL oxidative modification and reduced HDL function both correlate with increased cardiovascular diseases, whether the extent of HDL oxidative modification in a sample inversely correlates with HDL functional capacity measured in the same sample needs further elucidation. Zheng et al showed strong inverse relationships between ABCA1-dependent cholesterol efflux and apoA1 3-Cl-Tyr levels or apoA1 NO2-Tyr levels in a small group of patients from a U.S. cardiology clinic [70]. However, Wang et al found that increased levels of HDL 3-Cl-Tyr and HDL NO₂-Tyr did not correlate with reduced efflux capacity in individuals with coronary artery disease in a small Chinese cohort [108]. To the best of our knowledge, there have been no studies that have examined correlations between the extent of apoA1 or HDL modification by MDA, ACR, or IsoLG and cholesterol efflux capacity or other HDL functions in patient samples. Therefore, studies examining the relationships between various oxidative modifications and HDL function in large cohorts of relevant individuals are urgently needed to resolve these questions.

aldehyde scavengers represent an alternative approach To inhibit oxidative modification and provide cardioprotection

Validating a causal role for lipid aldehyde modification of HDL in development of cardiovascular disease requires demonstrating that selectively eliminating these modifications rescues HDL function and prevents or reverses disease. While dietary antioxidants can be used to reduce the formation of lipid aldehydes, the pharmacological doses of dietary antioxidants needed for even modest reductions in aldehyde levels have pluripotent effects including interfering with critical cell signaling pathways and blocking formation of other non-reactive lipid products, making it difficult to interpret outcomes. A more relevant approach uses small molecule nucleophiles serving as sacrificial targets for reactive lipid aldehydes, thereby sparing proteins and PE from aldehyde modification. A number of such reactive lipid aldehyde scavengers have been developed and show potential for treatment of cardiovascular disease.

α,β-unsaturated aldehydes such as HNE and ACR can be scavenged by thiol-based scavengers such as 2-mercaptoethanesulfonate (MESNA), amifostine, glutathione, and taurine or by imidazole-based scavengers such as carnosine (reviewed in [109]). MESNA is used as an adjunctive treatment for chemotherapy [110–112], and experimentally for ulcerative colitis [113, 114] and liver injury [115]. Carnosine, but not MESNA, also effectively scavenges MGO and carnosine has shown utility in treatment of diabetes [116, 117], atherosclerosis [118, 119], heart failure [120, 121], and ischemic brain damage[122, 123].

MESNA and carnosine are not effective for scavenging aldehydes that preferentially target primary amines like IsoLG. Furthermore, the extreme reactivity of IsoLG (>50% reaction of free aldehyde to form protein adduct in 20 sec compared to > 60 min required for 50% reaction of HNE to form protein adduct [124]) represented a significant challenge for identifying compounds with sufficient reactivity to prevent protein adduction by IsoLG. Fortunately, the screening of a series of bioavailable primary amines led to the identification of 2-aminomethylphenols (2-AMPs) as highly effective IsoLG scavengers [125, 126].

Mechanistically, the enhanced reactivity of 2-AMPs is due to the phenolic group stabilizing the initial formation of the imine adduct on the methylamine moiety, catalyzing the formation of the stable pyrrole adduct [125] (Figure 3). This leads to a >1000-fold faster reaction of IsoLG with 2-AMPs than the reaction of IsoLG with Lys [125]. In cellular assays, the more hydrophobic 2-AMP analogs such as 2-hydroxylbenzylamine (2-HOBA) and pentylpyridoxamine (PPM) show far greater ability to block the formation of IsoLG adducts than hydrophilic 2-AMP analogs such as pyridoxamine (PM) [126], presumably due to their greater hydrophobicity allowing them to penetrate into cellular membranes where IsoLG are generated [127]. 2-HOBA, PPM, and other 2-AMPs are also highly effective at scavenging other lipid dicarbonyls such as ONE and MDA [128]), but do not effectively scavenge α,β-unsaturated alkenals such as HNE that lack a second carbonyl moiety [125]. 2-HOBA and PPM reduce levels of lipid dicarbonyl protein adducts in a variety of cellular assays where oxidative stress is implicated and also inhibit key adverse effects [129–137]. Several compounds closely related to 2-AMPs only poorly react with IsoLGs including regioisomers of 2-AMPs such as 4-hydroxybenzylamine (4-HOBA), substituted amine analogs such as N,N-dimethyl-2-HOBA, or analogs where a hydroxyl group replaces the amine such as pentylpyridoxine (PPO) (Figure 3). Since these compounds are ineffective scavengers, they can be used as control compounds [138].

Figure 3.

2-aminomethylphenols (2-AMPs) including 2-hydroxybenzylamine (2-HOBA), pentylpyridoxamine (PPM) or pyridoxamine (PM) react with IsoLG or other lipid dicarbonyls to scavenge these dicarbonyls and thus blocking the modification of proteins by these lipid dicarbonyls. The high rate of reactivity of 2-AMPs is the result of their phenolic hydrogen helping to stabilize the initial imine adduct through hydrogen bonding and thereby helping to catalyze pyrrole formation. Inactive analogs of 2-AMP cannot participate in the same reaction and so can be used a control compounds for cellular and in vivo studies.

2-AMPs effectively protect HDL from oxidative modification and provide cardioprotection.

As previously noted, incubation of isolated HDL with MPO generates IsoLG-protein adducts, ONE-protein adducts, and MDA-protein adducts [85, 87, 101, 105]. Preincubation of isolated HDL with PPM blocks MPO-mediated IsoLG-protein adduct formation [87]. Pre-incubation of HDL with PPM also blocks the ability of IsoLG to alter HDL function when added to isolated HDL including blocking IsoLG-induced crosslinking of apoA1, IsoLG-induced inhibition of cholesterol efflux, and IsoLG-induced loss of HDL-mediated LPS-induced cytokine secretion [87]. Preincubation of isolated HDL with PPM (and to a somewhat lesser extent with 2-HOBA, fluoro-2-HOBA, or chloro-2-HOBA) blocks crosslinking of apoA1 when ONE is incubated with HDL [62]. Preincubation of HDL with PPM (but not PPO) prior to incubation with ONE blocks the loss of HDL-mediated inhibition of LPS-induced cytokine secretion normally induced by ONE [62]. Preincubation of HDL with either PPM or 2-HOBA blocked MDA-mediated crosslinking [85].

Both 2-HOBA and PPM are orally bioavailable and can be administered to rodents in drinking water [85, 127]. Administration of 2-HOBA to female Ldlr^−/− mice fed a high-cholesterol diet markedly reduces total aorta atherosclerotic lesion area, reduces the necrotic area of lesions, and enhances macrophage efferocytosis, without significantly altering either total plasma cholesterol or triglyceride levels [139]). Consistent with its mechanism of action being lipid dicarbonyl scavenging, 2-HOBA treatment reduced MDA- and IsoLG-protein adduct levels in the lesions and IsoLG and MDA adducts of 2-HOBA were detected in various tissues of 2-HOBA treated mice [139]. HDL isolated from 2-HOBA treated mice showed reduced MDA levels compared to HDL isolated from mice treated with vehicle or 4-HOBA (other lipid dicarbonyl adducts were not measured due to limited sample size) and this isolated HDL showed increased ex vivo cholesterol efflux capacity compared to vehicle or 4-HOBA treated mice [139]. 2-HOBA treatment also increased HDL PON1 activity [85]. Similar to 2-HOBA, administration of PPM to Ldlr^−/− mice also reduced HDL dicarbonyl adducts and enhanced the ex vivo cholesterol efflux capacity and PON1 activity of HDL isolated from PPM-treated mice [85].

The anti-atherosclerotic effects of dicarbonyl scavenger treatment are unlikely to be due exclusively to their effects on HDL function, as scavengers also reduce levels of LDL adducts [85, 139]. Furthermore, treatment with either 2-HOBA or PPM markedly reduces hypertension (a major contributor to cardiovascular disease) in several rodent models [132, 133]. The anti-hypertensive effects of dicarbonyl scavengers are primarily attributed to their ability to block the formation of dicarbonyl modified neo-antigens presented by dendritic cells and recognized by cytotoxic T cell [132]. Given this evidence, the remarkable anti-atherosclerotic effects of 2-HOBA and PPM most likely result from multiple beneficial effects of reducing lipid dicarbonyl adducts, including the enhancement of HDL function seen with this treatment.

Conclusions and future directions

The failure of clinical trials with dietary antioxidants and with HDL-C raising drugs research initially appeared to foreclose any significant role for oxidative modification of lipoproteins in the development of atherosclerotic cardiovascular disease and for HDL in protecting against disease. However, recent studies suggest a different conclusion. Because dietary antioxidants have little impact on the extent of LOOH formation and oxidative modification in vivo and because raising HDL-C levels do not necessarily improve HDL functionality in at risk individuals, these previous clinical studies do not preclude oxidative modifications of apoAI and other HDL components making significant contributions to development of ASCVD. In vitro studies clearly demonstrate the potential for oxidative modifications of HDL to markedly alter HDL functionality, either by oxidation of amino acid residues such as tyrosine and tryptophan or by lipid aldehyde modification of nucleophilic amino acids such as lysine. That various studies have found increased levels of oxidized HDL amino acids and HDL lipid aldehyde modification in diseased individuals support a causative role in disease. However, whether the extent of oxidative modification is sufficient to drive loss of function and is in fact causal remains to be demonstrated convincingly. In this regard, the early mouse studies with the 2-AMP class of lipid dicarbonyl scavengers such as 2-HOBA and PPM that showed both enhanced HDL functionality and reduced atherosclerosis are highly promising. Even so, additional studies are needed to determine whether these effects are the result of protecting HDL from oxidative modifications, rather than blocking modifications at other sites such as LDL. Furthermore, given that successful treatments in animal models of disease often fail to translate to improved human outcomes, whether oxidative modifications of HDL significantly contribute to development of atherosclerotic cardiovascular disease in humans will remain an open question until dicarbonyl scavengers show success in clinical trials.

Figure 1.

Major anti-atherosclerotic functions of HDL. HDL has a critical role in reverse cholesterol transport, as HDL acquires free cholesterol (FC) from macrophages and other cells in peripheral tissues, esterifies FC to cholesterol esters, and then transports cholesterol esters to the liver for excretion by hepatocytes. Antioxidative effects of HDL include inhibiting the oxidation of LDL and the formation of lipid peroxides (LOOH). PON1 plays a major role in HDL’s antioxidative effect both through its catalytic activity and by competing with MPO for binding to apoA1. Other proteins associated with HDL have also been shown to exert antioxidative effects. HDL exerts potent anti-inflammatory effects including suppression of cytokine secretion by endothelial cells and macrophages in response to inflammatory stimuli.