Apolipoprotein A-I and its mimetics for the treatment of atherosclerosis

Abstract

Although statin treatment leads consistently to a reduction in major adverse coronary events and death in clinical trials, approximately 60 to 70% residual risk of these outcomes still remains. One frontier of investigational drug research is treatment to increase HDL, the ‘good cholesterol’ that is associated with a reduced risk of coronary artery disease. HDL and its major protein apolipoprotein A-I (apoAI) are protective against atherosclerosis through several mechanisms, including the ability to mediate reverse cholesterol transport. This review focuses on the preclinical and clinical findings for two types of therapies for the treatment of atherosclerosis: apoAI-containing compounds and apoAI mimetic peptides. Both of these therapies have excellent potential to be useful clinically to promote atherosclerosis regression and stabilize existing plaques, but significant hurdles must be overcome in order to develop these approaches into safe and effective therapies.

Introduction

Heart disease remains the most frequent cause of death in the US, with over 50% of heart disease mortality attributed to coronary artery disease (CAD) [1]. Observational epidemiological studies have identified multiple independent risk factors for coronary artery disease, including age, gender, smoking, diabetes, hypertension, and total and LDL cholesterol (LDL-C) levels [2]. Since the Framingham Heart Study, elevated levels of HDL-cholesterol (HDL-C) have been recognized as an independent protective factor against CAD. Moreover, an early meta-analysis of four large prospective American studies (including Framingham) revealed that for every 1 mg/dl increase in HDL-C, a 2 and 3% reduction in CAD risk was observed for men and women, respectively [3].

Additional support for the protective effects of HDL comes from animal models and drug trials. ApoAI overexpression in mouse models of atherosclerosis increases HDL levels and reduces plaque formation and progression significantly [4,5]. Moreover, the Coronary Drug Project clinical trial revealed that participants treated with niacin, the most effective FDA-approved HDL-raising drug, had a 14% decrease in heart attacks or death as a result of CAD, a 27% decrease in non-fatal heart attacks, a 26% decrease in stroke and a 47% decrease in the incidence of heart surgery [6]. This finding has held up with time, and increasing HDL levels has been demonstrated to be effective in reducing CAD risk in subjects with low, medium, or high levels of LDL-C [7].

Although small-molecule drug therapies aimed at increasing HDL are mentioned in this review, the focus is on the administration of apoAI formulations and apoAI mimetic peptides, both of which may impart rapid effects on plaque regression and stabilization. Initial evaluations in a small clinical trial demonstrated that these types of therapies were efficacious [8]. Trial participants receiving injections of an apoAI variant, recombinant apoAI_Milano (r-apoAI_M), formulated with lipids into reconstituted HDL (rHDL) particles, demonstrated a small, but significant, decrease in coronary artery plaque volume after four weekly treatments, as assessed by intravascular ultrasound (IVUS) [8], thus proving the concept that apoAI therapy can reduce atheroma burden in humans directly. In vitro studies comparing the ability of apoAI_M and wild-type (WT) apoAI to act as cholesterol acceptors have reported mixed results [9-11]; thus, an opportunity exists for the development of more effective apoAI-based therapies. This review covers the mechanisms by which HDL is thought to protect subjects from CAD, and the development of apoAI and mimetic peptide therapeutics for the treatment of CAD.

Mechanisms underlying the HDL-mediated reduction in CAD risk

The reverse cholesterol transport (RCT) hypothesis, first described by Glomset [12], proposes that HDL acts as a shuttle to retrieve cholesterol from the periphery and deliver it to the liver, where it can be excreted directly into the bile or metabolized into bile salts before excretion. Both lipid-poor apoAI and HDL are now known to act as acceptors for cellular lipids, and lipid efflux from cells can be mediated via the cell surface proteins ABCA1, ABCG1 and SR-BI [13]. However, there was little direct evidence for the RCT hypothesis in vivo until a study conducted by Rader and colleagues was published in 2003 [14]. In this study, a macrophage cell line was loaded with [³H]cholesterol in vitro and these cells were then injected into the peritoneal cavity of mice. Increased levels of [³H]cholesterol mobilization from the macrophages to the plasma, liver and fecal compartments were observed in animals that were also treated with recombinant human apoAI adenovirus, thus demonstrating that increased levels of apoAI/HDL can mobilize cholesterol from peripheral tissues for excretion [14]. Subsequent to the initial apoAI/HDL-mediated removal of cholesterol from peripheral cells, there are several more steps in the RCT pathway: the maturation of nascent HDL through cholesterol esterification mediated by the plasma enzyme lecithin-cholesterol acyltransferase (LCAT); the uptake of HDL-C by the liver mediated by SR-BI; the alternate pathway of transferring HDL-C to apoB-containing lipoproteins via the plasma enzyme cholesterol ester transfer protein (CETP) that allows cholesterol trafficking to the liver via the LDL receptor; the metabolism of hepatic cholesterol into bile acids, and the excretion of liver sterols and bile acids into the bile and feces.

HDL/apoAI might exert their anti-atherogenic effects through several other mechanisms. For example, Navab et al demonstrated that HDL has anti-inflammatory activity and can inhibit LDL-induced monocyte attachment and transendothelial migration in an endothelial cell vascular smooth muscle cell co-culture system [15]. Moreover, this anti-inflammatory activity could be mediated by the direct antioxidant activity of HDL or minor HDL-associated proteins, such as paraoxonase and platelet-activating factor acetylhydrolase. In addition, Barter and colleagues were the first to demonstrate that HDL has anti-inflammatory activity that inhibits cytokine-induced adhesion molecule expression in endothelial cell cultures [16]. HDL has also been observed to neutralize the proinflammatory activity of C-reactive protein [17]. Some of these effects may be mediated by the HDL-induced stimulation of endothelial cell nitric oxide synthase [18]. For example, Seetharam et al reported that HDL promotes endothelial cell migration and vessel re-endothelialization that is mediated by SR-BI [19].

Although HDL is generally protective against CAD, not all HDL is equivalent, leading to the concept of dysfunctional HDL [20,21]. For example, HDL from patients with CAD who had high levels of HDL was determined to exhibit less anti-inflammatory activity and, in some cases, less proinflammatory activity, as compared with HDL derived from healthy control participants, when assayed in the co-culture system, thus indicating that HDL from patients with CAD may be dysfunctional [22]. Conversely, animal models of inflammation exhibit impaired RCT and HDL remodeling [23,24]. Moreover, apoAI is known to be a selective target for oxidative modification by the enzyme myeloperoxidase, which is located in neutrophils, monocytes, plasma and atheroma, creating a covalently modified apoAI with greatly diminished ABCA1-dependent cholesterol acceptor activity [25]. Thus, future apoAI therapies may be designed to be resistant to myeloperoxidase-mediated loss of function.

HDL-targeted therapies

Although not the main focus of this review, it is relevant to mention HDL-targeted therapies other than apoAI-based therapies. Currently available drugs that increase HDL-C levels include niacin, fibrates and, to a lesser extent, statins (as reviewed in reference [26]). However, niacin has an adverse facial flushing side effect, limiting the use of this inexpensive therapy. CETP inhibition is being developed as another HDL-raising therapy; however, the ILLUMINATE clinical trial of torcetrapib, an oral CETP inhibitor and HDL cholesterol enhancer, was terminated early as a result of increased death and cardiac events observed in the treated patients [27]. Notably, torcetrapib has off-target effects, including increasing blood pressure and altering plasma levels of aldosterone, potassium, sodium and bicarbonate; thus, the failure of this drug cannot be attributed to any potential class effects (common to all CETP inhibitors) on generating dysfunctional HDL. A new small-molecule inducer of apoAI production, RVX-208, is under development by Resverlogix Corp, which induces apoAI mRNA and synthesis in hepatoma cells [28,29]. A phase I study in 18 human subjects found that RVX-208 treatment with varying doses over seven days yielded 10% increases in serum apoAI and HDL-C, a 42% increase in serum prebeta-apoAI (lipid free and lipid poor apoAI),and an 11% increase in the ABCA1-mediated cholesterol acceptor activity of 2% serum (assayed in vitro) [28]. Another HDL-targeted therapy developed by Lipid Sciences Inc involved the selective delipidation of plasma HDL and the reinfusion of the lipid-poor HDL and apolipoproteins; this procedure increased the ABCA1-dependent cholesterol acceptor activity of HDL from human plasma [30]. Moreover, atherosclerotic monkeys treated with 12 weekly infusions of selectively delipidated monkey plasma exhibited a 7% reduction in atheroma volume, as determined by IVUS [30]; however, Lipid Sciences filed for bankruptcy in 2008, and development activities were assumed to be discontinued [31].

ApoAI-based therapies

Preclinical studies of apoAI-based therapies

Since HDL is protective against CAD, a series of studies were performed by several investigators to determine the effects of direct injections of HDL or rHDL, made with different apoAI preparations, on atherosclerosis in various animal models. Badimon et al demonstrated that four weekly infusions of rabbit HDL into rabbits fed an atherosclerosis-inducing high cholesterol diet could decrease the progression of aortic atherosclerotic lesions [32]. In a separate study, the same treatment regimen was used on older rabbits with more advanced lesions; despite the continued feeding of the high cholesterol diet, a regression of approximately 50% in the surface area of aortic atherosclerosis and in aortic total cholesterol content was observed [33]. In another study, Nicholls et al infused rabbit HDL or rHDL, prepared from palmitoyl-linoleoyl phosphatidylcholine and either rabbit apoAI or human apoAII, to cholesterol-fed and aortic-injured rabbits, and demonstrated that each of these treatments reduced the lesion area [34]. Other investigators have confirmed the effect of HDL infusions on retarding lesion progression in cholesterol-fed rabbits [35]. However, the effect on lesion regression was not reproducible, which is not surprising in the context of continued feeding of the high cholesterol diet that maintains high levels of LDL-C.

A series of preclinical studies using r-apoAI_M, a naturally occurring Arg¹⁷³Cys variant, have been conducted. Carriers of this variant have lower levels of apoAI and HDL-C, and are protected from carotid atherosclerosis compared with individuals with hypoalphalipoproteinemia [36]. However, apoAI_M carriers have been demonstrated to have a 57% increased prevalence of carotid plaques compared with healthy controls, although the difference was not statistically significant in this study because of the small sample size analyzed [36]. Shah et al investigated four groups of apoE-deficient mice, with the baseline group fed a high cholesterol western-type diet (WTD) for 20 weeks, and the control group maintained on this diet for 25 weeks. The treatment groups were fed the WTD from week 20 for 5 weeks and received tail vein injections every other day with either 40 mg/kg apoAI_M in a phospholipid complex (for simplicity, all apoAI phospholipid mixtures are referred to as rHDL) or the equivalent amount of phospholipids [37]. A 59% increase in lesion surface area in the control group (ie, progression) was observed, as compared with a 32% increase in lesion surface area in the phospholipid treatment group. Lesion progression was halted completely in the rHDL-treated mice, and was accompanied by a decrease in the lipid-staining area of the aortic root lesions [37]. In a follow-up study, Shah et al delivered a high dose of r-apoAI_M (400 mg/kg), formulated into rHDL particles, as a single infusion to WTD-fed apoE-deficient mice [38]. This treatment led to an acute > 2-fold increase in plasma free cholesterol levels and an increase in HDL-C levels. Aortic root lesions were investigated 2 days after treatment using histology-based assays; treatment with rHDL led to an approximately 40 and 30% decrease in lipid and macrophage content of aortic root lesions, respectively [38]. Another study by this research group demonstrated that r-apoAI_M rHDL injections administered to apoE-deficient mice led to improved endothelium-dependent vasodilation in perfused arteries analyzed ex vivo. In addition, the in vitro treatment of rabbit arteries with this rHDL also protected against endothelial dysfunction induced by lysophosphatidylcholine [39]. More recently, ETC-216, an r-apoAI_M rHDL formulation made by Esperion/Pfizer was evaluated in cholesterol-fed carotid artery-injured rabbits in a dose-response study of the carotid plaque, as assessed in vivo by IVUS [40]. Rabbits were treated every 4 days; after two treatments, the lower doses led to reduced lesion progression, while the higher doses led to lesion regression and a significant reduction in markers associated with plaque instability [40,41]. For undisclosed reasons, Pfizer Inc discontinued the development of ETC-216, and sold the apoAI_M license to The Medicines Company in 2009 [42].

Although r-apoAI_M was demonstrated to be effective in preclinical atherosclerosis studies and in many other infusion studies that measured non-atherosclerosis endpoints (eg, see references [39,43]), whether r-apoAI_M is superior to WT human apoAI remains unclear. In head-to-head comparisons of RCT in mice after virus-mediated gene transfer, essentially no difference was observed between the two forms of apoAI, although r-apoAI_M led to lower LCAT activity compared with WT human apoAI [44]. Cho and Kim reported a greater decline in lesion areas in WTD-fed apoE-deficient mice 48 h after the infusion of a single dose of pro-apoAI_M rHDL, as compared with WT pro-apoAI; treatment with pro-apoAI_M rHDL was associated with increased plasma apoAI levels, reduced plasma peroxidase levels and a decreased susceptibility of apoAI_M to oxidation in vitro [45]. Notably, apoAI_M binds lipids less efficiently than WT apoAI and has a reduced capacity to accept membrane cholesterol [9,46]. In addition, in transgenic mouse models, apoAI_M was secreted less efficiently by the liver than WT apoAI [47,48]. Furthermore, in atherogenic mice, lesions in apoAI_M transgenic female animals were larger than in WT apoAI transgenic female animals; this effect was not observed in males [47,48]. Other studies compared virus-mediated expression of WT apoAI and apoAI_M directly in apoE-deficient or LDL receptor-deficient mouse hosts and demonstrated that they were equivalent in reducing plaque progression [49,50]. However, one study using apoAI-transduced macrophages reported that apoAI_M reduced lesion progression to a greater extent than WT apoAI in apoE/apoAI double-knockout mice [51].

Clinical trials of apoAI-based therapies

Single infusions of high-dose rHDL, derived from ~1.6 g of recombinant human pro-apoAI, into four patients with low HDL-C led to a 20% increase in HDL-C that persisted for approximately 4 days, and demonstrated a plasma residence time of ~5.5 days [52]. This residence time was similar to that observed in human apoAI tracer studies in which radiolabeled apoAI, which was not in a rHDL format, was mixed with autologous plasma and infused into human participants [53]. The net increase in plasma apoAI levels was similar following infusion of apoAI as rHDL or as a lipid-free preparation [54,55]. Thus, it is unclear whether there is any pharmacokinetic advantage of rHDL formulations compared with lipid-free apoAI. However, there are several potential advantages of the lipid-free form of apoAI, including easier preparation and excellent ABCA1-dependent cholesterol acceptor activity, which rHDL lacks [56].

A series of elegant papers from Miller and colleagues demonstrated that rHDL infusion into humans increases the levels of small preβ-HDL, HDL-C, plasma and lymph apoAI, and plasma CETP, as well as the activities of LCAT and phospholipid transfer protein [54,57-59]. These rHDL infusions were accompanied by an increase in fecal bile acid excretion [60], thus demonstrating increased RCT. Several additional studies with rHDL infusion in humans have demonstrated improvements in glucose metabolism in diabetic patients [61], endothelial function [62-64], anti-inflammatory responses [65-67] and platelet function [68-70].

The key study in this area was the phase II, double-blind, randomized clinical trial of five weekly infusions of placebo or two dosages of the apoAI_M rHDL product ETC-216 in patients (n = 47) with acute coronary syndrome [8]. Atheroma burden was assessed at the beginning and the end of the treatment period by IVUS. Pooling data from the two dosages, ETC-216 treatment led to a significant 4% decrease in atheroma volume, while the atheroma volume increased by 0.14% in the placebo-treated group [8]; this result is noteworthy, considering the short treatment period.

The phase II, randomized, placebo-controlled ERASE clinical trial evaluated four weekly infusions of two doses of CSL-111, a rHDL product prepared from WT plasma apoAI that is being developed by CSL Behring, in patients (expected n = 180) with atherosclerosis [71]. The high dose regimen (80 mg/kg) was discontinued because of abnormal liver function results. IVUS was used to measure effects on plaque volume; although plaque volume decreased by 3.41% in the treated groups, this result was not significantly different from the 1.62% decrease in plaque volume observed in the placebo group [71]. A separate trial investigated the administration of one dose of CSL-111 or placebo to patients with femoral artery claudication (n = 10), 5 to 7 days prior to femoral artery atherectomy [72]. The recovered atheromas were examined histologically for lipid content, macrophage size and VCAM-1 expression, with all three parameters being lower in the rHDL treatment group compared with the placebo group [72]. Although the results suggested that CSL-111 infusion may lead to plaque stabilization, given that the study was limited by a lack of longitudinal data, these effects could not be demonstrated conclusively. CSL Limited is now recruiting subjects for a phase I safety study of CSL-112, an alternative rHDL formulation [73].

ApoAI mimetic peptides

Segrest et al first demonstrated the prominent pattern formed by apoAI residues 41 to 243 in helical wheel plots [74]. In this structure, termed the class A amphipathic α-helix, the hydrophobic residues are on one side, with positively charged residues flanking to, and negatively charged residues opposite of, the hydrophobic residues. This research team also developed a series of synthetic peptides, which were not based on the sequence of apoAI, but on the class A amphipathic helix structure that apoAI residues 41 to 243 formed; 18A is the prototype 18-mer peptide, and 37pA is a dimer of two 18A peptides with a proline spacer [75]. Similar to apoAI, these peptides can spontaneously clarify a phospholipid emulsion of dimyristoylphosphatidylcholine (DMPC), forming disc-like rHDL particles [75]. Additional interest was focused on a series of more hydrophobic analogs with an increasing number of phenylalanine substitutions (referred to as 2F, 3F, 4F and 5F) [76]. Several of these analogs exhibited increased lipid association activity and, when tested in the co-culture assay for LDL-induced monocyte chemotaxis, the 4F peptide demonstrated the strongest anti-inflammatory activity, which appears to be related to the ability of these peptides to absorb oxidized fatty acids [77]. Similar to apoAI, many of these peptides could act as ABCA1-dependent acceptors for cellular cholesterol; however, at high concentrations, these peptides displayed detergent-like properties and could extract cholesterol from cells independent of ABCA1 [78]. Notably, Remaley et al demonstrated that the 37pA peptide, which is composed entirely of d-amino acids, retains its ABCA1-dependent cholesterol acceptor activity, while substituting only two d-amino acids for l-amino acids led to a loss of the class A structure and the cholesterol acceptor activity [78]. This result implies that ABCA1 does not need to interact with these peptides in a chiral manner, but that ABCA1 modifies the plasma membrane that permits a productive interaction with class A amphipathic helical peptides resulting in cholesterol efflux.

The 5F peptide (20 μg/day ip) was injected for 16 weeks into C57BL/6 mice fed a cholate-containing high cholesterol atherogenic diet [79]. This treatment significantly reduced aortic root atherosclerosis, while injection with mouse apoAI (50 μg/day ip) was ineffective. To circumvent the need to inject the mimetic peptides and to facilitate oral administration, Navab et al prepared the 4F peptide entirely out of d-amino acid, referred to as APP-018 (d-4F); following oral delivery, the peptide could be detected intact in the plasma 4 h later [80]. The oral administration of APP-018 to LDL receptor-deficient or apoE-deficient mice led to substantial reductions in aortic root lesion areas [80]. In addition, oral APP-018 treatment increased small preβ-HDL levels and led to increased anti-inflammatory activity of HDL, with reductions in lipid hydroperoxides [81]. Combined oral APP-018 and statin treatment of apoE-deficient mice with established lesions resulted in the regression of lesions in the aortic root and the entire aorta [82]; however, Li et al reported that orally or intraperitoneally administered APP-018 without statin did not inhibit the progression of established lesions in apoE-deficient mice [83].

These and other preclinical studies of APP-018 led Bruin Pharma Inc to a conduct a phase I clinical trial in patients (n = 50) with CAD, who were administered a single oral dose of placebo or APP-018 (30, 100, 300 or 500 mg) [84]. The agent was detected in plasma at low levels, suggesting limited bioavailability; however, the two highest doses were associated with an increased anti-inflammatory activity of patient-derived HDL [84]. Novartis AG licensed APP-018 from Bruin Pharma in 2005 [85], but no subsequent development has been reported.

Various other amphipathic peptides have been developed based on apoAI, apoE and apoJ, or on purely structural constraints, similar to peptide 18A. Although many of these peptides appear to be promising, most have not advanced significantly in animal models. For example, Sethi et al developed a 37pA analog referred to as 5A [86]. The acute administration of a 5A lipid complex to mice increased RCT, and chronic intravenous or intraperitoneal administration of the 5A lipid complex to apoE-deficient mice significantly decreased the aortic surface atherosclerosis lesion area [87]. Another peptide, designated ATI-5261, is being investigated by Artery Therapeutics Inc. Daily intraperitoneal injection of this apoE-derived peptide for 6 weeks into LDL receptor-deficient mice led to a 30% reduction in aortic root lesion area, without affecting plasma cholesterol [88]. Similarly, the treatment of apoE-deficient mice with this peptide by intraperitoneal injection every other day led to a 45% decrease in lesion area [88]. In addition, Navab et al recently demonstrated that niclosamide complexed to 4F or an apoJ peptide protected the compounds from degradation and allowed oral delivery [89]. Oral treatment of the niclosamide-4F complex and a statin for 6-months resulted in lesion regression in aged apoE-deficient mice [89].

Conclusion

HDL is protective against CAD by several proposed mechanisms including reverse cholesterol transport, which has led to great interest in the development of therapeutics based upon apoAI and mimetic peptides. Preclinical studies have demonstrated the proof of principal that infusion of rHDL or mimetic peptides can lead to increased levels of HDL and reverse cholesterol transport, and decreased atherosclerosis. A landmark clinical study demonstrated that weekly infusions of ETC-216 (rHDL made from apoAI_M) into human CAD subjects led to small but significant reduction in atheroma volume. A clinical trial using CSL-111 (rHDL made with wild type apoAI) also trended towards decreased atheroma volume, although these results were not statistically significant. Although, apoAI_M has demonstrated the most therapeutic benefit, the preponderance of evidence suggests that this variant is no better than WT apoAI, thereby creating an opportunity for the development of new apoAI variants that are resistant to becoming dysfunctional. Another challenge in this area is to determine the optimal apoAI formulation for use as a therapeutic; for example, whether rHDL is superior to lipid-free apoAI has not been clearly demonstrated. An additional barrier to the use of apoAI as a therapeutic is the difficulty in producing endotoxin-free recombinant apoAI. Several apoAI mimetic peptides have been developed as potential therapeutics, and although there have been no clinical trials looking at atheroma volume with these peptides, there is a precedent for the use of peptides as drugs with several peptides having been FDA-approved as drugs in other disease areas. Given that these mimetics have more ABCA1-indpendent detergent-like properties than apoAI, these peptides may be more prone to non-specific side effects than apoAI therapeutics. In conclusion, apoAI and mimetic peptides offer great promise for the treatment of CAD. These developing therapies may be used together with statins to shrink, remodel, and stabilize existing plaques.