Unraveling the PAF-AH/Lp-PLA₂ controversy

The secreted or plasma form of platelet-activating factor acetylhydrolase (PAF-AH), also known as lipoprotein-associated phospholipase A₂ (Lp-PLA₂) or phospholipase A₂ group 7 (PLA2G7), is a member of the PLA₂ superfamily of enzymes that circulates in blood in association with lipoproteins, and is found in atherosclerotic lesions [reviewed in (2, 3)]. This enzyme was discovered in the early 1980s based on its ability to hydrolyze the pro-inflammatory glycerophospholipid PAF, and was thus proposed to have anti-inflammatory properties. In subsequent years, it was recognized that PAF-AH hydrolyzes glycerophospholipids containing short and/or oxidized functionalities at the sn-2 position, with no preference for the type of linkage at the sn-1 position, i.e., alkyl versus acyl. Substrate hydrolysis catalyzed by PAF-AH generates lysoPAF/lyso phosphatidylcholine (lysoPC) and short and/or oxidized fatty acids, many of which also have been reported to have pro-inflammatory and pro-oxidative activities (4). These observations fueled multiple investigations that led to controversial views regarding the role of PAF-AH in human physiology and disease (1, 4–6). Notably, the hypothesis that PAF-AH might actively contribute to vascular inflammation during atherogenesis owing to its ability to generate pro-inflammatory substances (7) led to the proposition that inhibition of the activity could offer vascular protection in addition to that afforded by cholesterol lowering agents. A number of reversible PAF-AH inhibitors were developed in the pharmaceutical industry and one of them, darapladib, has been extensively tested in vitro and in vivo (8–14). Moreover, GlaxoSmithKline sponsored three darapladib clinical trials that yielded relatively consistent results (11, 15–20).

In this issue of the Journal of Lipid Research, Marathe et al. (1) provide their views on the role of PAF-AH in inflammatory responses, with a focus on CVD. The authors make several key points and offer a seldom encountered perspective that takes into consideration the origin and wide range of substrates hydrolyzed by PAF-AH, the physiological meaning of studies involving one of the products of the reaction (lysoPAF/lysoPC), and the impact of receptors that recognize substrates and products on downstream signaling events. The authors present several lines of evidence arguing against a pro-atherogenic role for PAF-AH and its products, and suggest that elevated enzyme levels reflect a response to the pro-inflammatory/pro-oxidative stress that is typical of atherosclerosis. Their conclusions are timely and consistent with results from recent clinical trials in humans. In the Integrated Biomarker and Imaging Study 2 phase II trial involving patients with coronary heart disease, the PAF-AH inhibitor darapladib did not meet prespecified primary and secondary endpoints that included effects on coronary atheroma deformability, composition and size, CRP levels, and total atheroma volume (15). While darapladib administration inhibited necrotic core expansion, this conclusion was reached only by fine interpretation of imaging data (21). Nonetheless, the finding provided, in part, the basis to conduct two phase III trials. The recently published STABILITY (Stabilization of Atherosclerotic Plaque by Initiation of Darapladib Therapy) trial showed that darapladib did not affect the primary composite endpoint that included time to cardiovascular death, myocardial infarction, or stroke in patients with stable coronary heart disease (19). Similarly, results recently reported from SOLID-TIMI 52 (Stabilization of Plaques using Darapladib-Thrombolysis in Myocardial Infarction 52) revealed no reduction in major coronary events when added to standard of care after an acute coronary syndrome.

An important lesson learned from the outcome of studies using darapladib is related to the necessity of gathering rigorous scientific evidence supporting a solid rationale to justify launching clinical trials in humans. Arguably, this was not the case for darapladib. Mechanistically, the trials were largely based on laboratory studies that investigated pro-inflammatory functions of lysoPC and, to a lesser extent, oxidized fatty acids. Marathe et al. (1) appropriately describe two major potential problems associated with interpretation of studies using exogenous lysoPC. First, trace amounts of PAF and/or related phospholipids have been shown to contaminate numerous commercial lysoPC preparations. This problem may have affected multiple studies in which contaminating PAF could have accounted for responses incorrectly ascribed to lysoPC (22). Marathe et al. (1) also point out that lysoPC occurs naturally at very high concentrations in body fluids and atherosclerotic tissues (23). Its amphipathic nature and detergent-like properties can induce nonspecific cellular responses. In physiological settings, these effects are in large part prevented because lysoPC forms complexes with serum proteins, immunoglobulins, and plasma membranes. Thus, total lysoPAF/lysoPC concentrations do not relate to bioavailability and it is highly unlikely that the relatively small amounts of free lysoPAF/lysoPC generated by PAF-AH contribute to inflammatory responses in the vasculature.

A second issue to reflect upon is the importance of interpreting correlative studies appropriately. Strong positive correlations between plasma PAF-AH and LDL cholesterol levels were established approximately 30 years ago [reviewed in (4)]. Plasma PAF-AH associates with cholesterol-containing LDL particles (24, 25); as expected, enzyme levels decrease in response to statin treatment (26, 27). Many prospective population-based studies confirmed tight links between increased plasma PAF-AH levels and increased cardiovascular risk (28–34). Predictably, the strength of this association is highly reduced after adjustment for baseline concentrations of lipids and apolipoproteins, particularly apoB levels (35). These observations demonstrate that elevated circulating levels of PAF-AH (Lp-PLA₂, the FDA-cleared diagnostic test is called PLAC®) are associated with atherosclerosis. But such observations support neither causal nor protective roles for PAF-AH in the disease process. In contrast, the fact that partial inhibition of PAF-AH with darapladib did not prevent adverse events in coronary heart disease patients argues against active contribution of the enzyme to plaque vulnerability and major adverse cardiovascular complications such as heart attack, stroke, and death.

A third consideration is that the relationship between PAF-AH, PAF, and PAF-like substrates and products generated to various extents in settings of inflammation and oxidant stress is incompletely understood. In this regard, Marathe et al. (1) present a hypothetical model suggesting that the relative abundance of alkyl versus acyl PAF may determine whether PAF-AH has pro- or anti-inflammatory/atherogenic functions. While alkyl and acyl PAF and PAF-like lipids have been reported to be continuously generated, alkyl-linked species are functionally more potent owing to their higher affinity for the PAF receptor (PAF-R), a G protein-coupled receptor that transduces PAF signals (36, 37). According to Marathe et al. (1), acyl-linked PAF-like analogs are less potent and could potentially behave as relative PAF-R antagonists, although this remains to be shown experimentally. If this scenario is correct, hydrolysis of acyl-linked PAF-like lipids by PAF-AH could effectively decrease the levels of “natural” PAF-R antagonists, potentially increasing pro-inflammatory/atherogenic activities. This interesting and provocative model will require future pharmacologic and molecular studies, including targeted silencing of PAF biosynthetic (38) and hydrolytic (36) pathways. It is important to also consider, however, that PAF-AH substrates often elicit biological activity in PAF-R-independent manners. Moreover, it will be important to establish whether the truncated/oxidized sn-2 fatty acids released by PAF-AH also contribute to its physiologic function. Adding to the complexity is that in this process truncated oxidized phospholipids (OxPLs), which are potent pro-inflammatory molecules, are degraded by this enzymatic activity.

While darapladib failed to reach all of its primary endpoints, the inhibitor showed some efficacy when administered in experimental animal models and humans. Indeed, darapladib and related compounds significantly decreased atherosclerotic coronary lesion development, reduced macrophage content in vascular lesions, and attenuated plaque inflammation in various animal models [recently reviewed in (4)]. In interpreting these studies, it is important to establish whether darapladib exerted biologic effects by inhibiting PAF-AH activity or by some other mechanism. Careful analysis of changes in lipid metabolites suggests the possibility that darapladib may have antioxidant and/or other “off target” effects (4). Although darapladib treatment was associated with reduced content of lysoPC in pig atherosclerotic lesions (14), it did not affect the levels of truncated OxPL species known to be metabolized by PAF-AH (39), and did not alter serum PAF levels in two murine models of atherosclerosis (12, 40). Besides its effects on lipid metabolism, darapladib treatment decreased caspase-3 and caspase-8 activity in vivo (15, 41), and a compound related to darapladib (SB222657) inhibited macrophage apoptosis induced by oxidized LDL in vitro (42). These observations raise the possibility that darapladib has activities in addition to its inhibitory effects on PAF-AH. It is thus conceivable that these effects may contribute to its in vivo activities. Regardless, darapladib has taught us that partial inhibition of PAF-AH does not appear to have major impact on vascular events and that the enzyme is unlikely to be the sought-after cholesterol-independent biomarker and target whose inhibition would further decrease morbidity and mortality of patients with vascular disease. The fact that darapladib-mediated partial inhibition of PAF-AH did not reduce coronary events does not, however, suggest that the future of anti-inflammatory heart drugs is dimmed, as recently suggested (43). The activity of PAF-AH has not been shown to lead to a net increase in pro-inflammatory lipid mediator levels in vascular settings. Regrettably, the enzyme is often referred to as a pro-inflammatory protein despite the fact that it has anti-inflammatory properties, as demonstrated in the original article describing cloning of the gene and characterization of the enzyme (44).

Deciphering the physiologic roles of PAF-AH continues to be a challenge for investigators across academia and the private sector alike, and a number of issues remain to be resolved. For example, it is not clear whether PAF-AH functions in the circulation, in atherosclerotic plaques and other tissues, or both. A study using mice lacking PAF-AH expression suggested that the enzyme may not function in the circulation and that substrate transport to the intracellular compartment may be required before hydrolysis occurs (45). In addition, the relationship between circulating and tissue PAF-AH and OxPLs has not been critically evaluated, and this issue raises important questions regarding the impact of enzyme, substrate, and products in different biologic compartments. The article by Marathe et al. (1) discusses some of these variables in the intricate biochemistry and biology of PAF-AH, PAF, and related lipids.