Lipoprotein‐associated phospholipase A2: The story continues

Abstract Inflammation is thought to play an important role in the pathogenesis of vascular diseases. Lipoprotein‐associated phospholipase A2 (Lp‐PLA2) mediates vascular inflammation through the regulation of lipid metabolism in blood, thus, it has been extensively investigated to identify its role in vascular inflammation‐related diseases, mainly atherosclerosis. Although darapladib, the most advanced Lp‐PLA2 inhibitor, failed to meet the primary endpoints of two large phase III trials in atherosclerosis patients cotreated with standard medical care, the research on Lp‐PLA2 has not been terminated. Novel pathogenic, epidemiologic, genetic, and crystallographic studies regarding Lp‐PLA2 have been reported recently, while novel inhibitors were identified through a fragment‐based lead discovery strategy. More strikingly, recent clinical and preclinical studies revealed that Lp‐PLA2 inhibition showed promising therapeutic effects in diabetic macular edema and Alzheimer's disease. In this review, we not only summarized the knowledge of Lp‐PLA2 established in the past decades but also emphasized new findings in recent years. We hope this review could be valuable for helping researchers acquire a much deeper insight into the nature of Lp‐PLA2, identify more potent and selective Lp‐PLA2 inhibitors, and discover the potential indications of Lp‐PLA2 inhibitors.

application of fragment-based lead discovery (FBLD) in developing Lp-PLA2 inhibitors provides a referential practice for other targets' lead discovery. Furthermore, the new knowledge of the role of Lp-PLA2 in certain diseases 36,37 and the substantial amounts of data generated from clinical trials 38 might help extend the applied range of Lp-PLA2 as a therapeutic target. Therefore, we hope that this updated, comprehensive review can help the interested researchers become better acquainted with Lp-PLA2 and attract more attention to Lp-PLA2 research.

| Biochemical properties
The human plasma Lp-PLA2 is a 45 kDa hydrophobic protein that functions in the absence of calcium. It possesses the ability to catalyze hydrolysis at the sn-2 position of phospholipids, while sequence analysis revealed a GXSXG motif, the characteristic fingerprint for neutral lipases and serine esterases, in the primary structure of Lp-PLA2. 25 Consequently, Lp-PLA2 exhibits characteristics of both PLA2s and neutral lipases. Even though hematopoietic cells might be chiefly responsible for the circulating Lp-PLA2 levels, a number of tissues, such as liver cells, 39 aorta cells, 40 and adipocytes, 41  F I G U R E 1 The proinflammatory role of Lp-PLA2 during LDL oxidization. This figure illustrates the proinflammatory effects of Lp-PLA2 at cellular level, focusing on the biological activities of its productions. Herein, the adhesive molecules include intercellular adhesive molecule 1 and vascular cell adhesive molecule 1, and the inflammatory mediators comprise the cytokines and proteases. 13,14 EC, endothelial cell; LDL, low-density lipoprotein; LPC, lysophosphatidylcholine; Lp-PLA2, lipoprotein-associated phospholipase A2; NEFA, nonesterified fatty acid; SMC, smooth muscle cell lipoproteins in plasma, whereby LDL and HDL carry 70% to 80% and 20% to 30% of the total plasma activity, respectively. 11 Lp-PLA2 acts on the substrate in the aqueous phase, as no signs of interfacial activation are associated with Lp-PLA2. 42 However, given that the surfaces of both LDL and HDL are enriched with phospholipids, we cannot rule out the possibility of Lp-PLA2 binding to its substrates from the lipid membrane phase, which is the case for all typical membrane-associating PLA2 enzymes. 11 Indeed, this speculation was subsequently supported by two studies, which implied that the substrates could enter the catalytic center of Lp-PLA2 from lipoprotein particles. 43,44 Nevertheless, this hypothesis remains to be established, as supporting evidence is limited.
The susceptibility to oxidative inactivation is another odd property of Lp-PLA2, in view of its elevated expression in response to oxidative stress. The inhibition of Lp-PLA2 activity, which apparently results from both physiological (heavy metals and oxygen radicals) and nonphysiological (cigarette smoke) oxidants, may involve irreversible modification of key amino acid residues. Peroxynitrite, one of the key oxidants produced in cellular oxidation in vivo, was identified to inactivate Lp-PLA2. 45 Using site-directed mutagenesis, MacRitchie et al 46 revealed that a primary target of oxidation in the protein was Met117, which was exposed on the protein surface; its oxidation would not only result in enzymatic inactivation but also affect the association with LDL. In addition, the tyrosine nitration of Tyr307 and Tyr335 also moderately contributed to the inactivation of the enzyme. Very recently, Gurung et al 45 used molecular dynamics simulation and essential dynamics in tandem with a molecular docking approach to elucidate the effect of the structural alteration of Lp-PLA2 upon the oxidation of the aforementioned three amino acids. The results revealed that Met117 oxidation induced enhanced flexibility, decreased compactness in the oxidized state, and insubstantial binding to the substrate PAF. 45 In addition to the decreased binding affinity of the substrate and high flexibility, nitration of Tyr307 and Tyr335 also led to the disorientation of the catalytic triad and a reduction in the molecular interactions of NT-Tyr307 and NT-Tyr335 with other residues of the protein. 47 Even though these findings offered some plausible mechanism for the reduction in enzymatic activity of Lp-PLA2 under oxidative stress, further insights into the molecular mechanisms are anticipated through studies using other biophysical techniques such as nuclear magnetic resonance, cryoelectron microscopy and X-ray crystallography.

| Substrate specificity
At the time of discovery, Lp-PLA2 was found to be capable of hydrolyzing PAF, a phosphatidylcholine containing an acetate group at the sn-2 position. Subsequent studies demonstrated that Lp-PLA2 possessed broad substrate specificity. A distinguishing feature of Lp-PLA2, compared with other members of the phospholipase A2 superfamily, is its specific recognition of the type of sn-2 residue to be hydrolyzed. Compared with the short acetyl chains at the sn-2 position of PAF, the extended sn-2 residue of PAF analogs are hydrolyzed by Lp-PLA2 with decreased efficiency; for instance, the C5 homolog is 60% as efficient as PAF, the C6 homolog is 20% as efficient, and the C9 homolog is only 2% as efficient. However, if the sn-2 residue of phospholipids contains a ω-terminal oxo function, such as an aldehyde or a carboxyl, this length restriction is greatly relaxed, and sn-2 acyl chains of up to nine carbon atoms in length can be tolerated. 11 In addition, Lp-PLA2 could efficiently hydrolyze short-chain diacylglycerols, triacylglycerols, and acetylated alkanols, signifying that a polar headgroup might not be required for Lp-PLA2 substrates. Furthermore, the enzyme shows a weak specificity at the sn-1 position, as it could not distinguish ester from ether at the sn-1 position of PAF or PAF analogs. 48 Apparently, glycerides with a sn-2 ester and a reasonably hydrophobic sn-1 residue would be efficient substrates of Lp-PLA2. Interestingly, Lp-PLA2 display phospholipase A1 activity that specifically removes the sn-1 acyl group of phospholipids under certain conditions. 31 Truncated phospholipids with shortened and oxidized sn-2 residues are mainly generated via the oxidative attack on phospholipid components of cellular membranes and lipoproteins. Considering the substrate selectivity, the identification of truncated oxidized phospholipids (oxPLs) as substrates of Lp-PLA2 is comprehensible. 49 Two analyses indicated that in vitro incubation of LDL with copper sulfate will generate oxidized modified LDL, in which lysoPC appeared to be the major product. However, its generation was strongly suppressed by the treatment of human LDL with a Lp-PLA2 inhibitor, suggesting that Lp-PLA2 activity is critical for lysoPC generation in LDL. 50,51 Further, mass spectrometry (MS)-based analyses were carried out to investigate the changes in the compositions of native and copper-oxidized LDL phosphatidylcholines in the presence and absence of an Lp-PLA2 inhibitor. As a result, the major oxidized PC (oxPC) species that accumulated after inhibitor treatment were a cleaved oxPC, 1-palmitoyl-2-(9-oxo-nonanoyl) PC, and a long-chain oxPC species with two double bonds, indicating possible hydrolysis by Lp-PLA2. 51 Both of these studies established that truncated oxPC was more efficient than long-chain oxPC as a substrate for Lp-PLA2. With a better understanding of the physiological functions of Lp-PLA2, more oxidized phospholipids with complicated substituents at the sn-2 position were defined as substrates for Lp-PLA2.
For example, peroxidized 1-palmitoyl-2-oleoyl phosphatidylcholine, 52 a phospholipid hydroperoxide, and esterified revealed that the preferable substrates might be PSox species whose sn-2 ester bonds were in proximity with Ser273 and oxygenated functional groups were hydroxy or hydroperoxy, particularly, at the C9 position of the sn-2 chain. 55 Furthermore, Lp-PLA2 possesses transacetylase activity, that is, it can transfer acetyl from PAF to ether-or ester-linked lysophospholipids. 31 In general, oxPC, PAF and its analogs are the major substrates for Lp-PLA2, while PSox and other types of oxPL constitute the minor ones. In both PC and PS, the oxygenated long-chain fatty acid residues may be the dominant products of oxidation and are less efficient Lp-PLA2 substrates, while truncated oxidized species with low abundance are preferentially hydrolyzed by Lp-PLA2. Interestingly, Greenberg et al 56 observed that the sn-2 fatty acid moiety of oxidatively fragmented phospholipids no longer remained in the hydrophobic core of the membrane, but rather protruded into the aqueous phase. Along with the finding that the enzyme possibly accessed its substrates from the aqueous phase, Kono et al 31 speculated that the substrate specificity of Lp-PLA2 might be related to the conformational orientation of the oxidized sn-2 fatty acid species in the membrane bilayer. In addition, oxPL species were shown to be chemically linked to naturally occurring human proteins such as apolipoprotein (a) and plasminogen; hence, the covalent lipid adducts might affect the hydrolysis of Lp-PLA2 by accessibility to the sn-2 acyl group. Since two studies obtained inconsistent results regarding whether these covalent adducts could protect oxPL from being metabolized by Lp-PLA2, 57,58 further studies are essential to address this issue.
distribution of Lp-PLA2 among lipoproteins was observed in patients with paroxysmal atrial fibrillation, chronic kidney disease (CKD), hyperlipidemia, and coronary artery disease (CAD). 60 The proposal that HDL-bound Lp-PLA2 is antiatherogenic while LDL-bound enzyme has the opposite effect, together with the diverse metabolic behavior between the LDL particles that carry Lp-PLA2 and those that lack this enzyme, apparently indicates that the dynamic distribution of Lp-PLA2 is likely to influence the pathophysiology, clinical significance, and behavior of pharmacological intervention in humans. 12 To identify the specific regions of Lp-PLA2 involved in binding to LDL and HDL, a number of studies were performed based on the apo form of Lp-PLA2 structures. Consequently, two short α-helices rich in hydrophobic residues are identified to form the interface binding surface (i-face) and presumably insert into the interfacial region of the membrane. Specifically, the residues Tyr205, Trp115, Leu116, and to a lesser extent, Met117, which are part of the N-terminal α-helix, are probably essential for binding to LDL, 23 whereas a part of the C-terminal α-helix (His367 to Lys370) accounts for the association with HDL. 61 Additionally, other factors can also affect the association between lipoproteins and Lp-PLA2. For example, N-linked glycosylation of plasma Lp-PLA2 was shown to hinder its binding to HDL in humans and Asn423 and 433 might be involved in this process. 62 Moreover, charged amino acids may also be potentially involved in the interaction with lipoproteins. The preference of Lp-PLA2 for small dense LDL, electronegative LDL and Lp(a) was proposed to be mediated by the altered conformation of apoB100, the major apolipoprotein present on LDL, in which the exposure of additional regions with positively charged residues may electrostatically interact with the negatively charged acidic patch of Lp-PLA2. In addition, specific basic patch amino acids could interact with either the apoproteins or the phospholipid component of LDL and HDL particles. As an example, the charged state of His114 present on the binding surface likely determines the pH-dependent binding of Lp-PLA2 to LDL. Seemingly, the interactions between residues of Lp-PLA2 and LDL or HDL are electrostatic in nature, and these electrostatic interactions may play a key role in modulating the distribution of the enzyme between lipoproteins. 23 The alteration of the structural properties of lipoproteins also affects the interaction with Lp-PLA2. Pande et al 43 demonstrated that increasing the amount of cholesterol in the phospholipid membranes has a significantly dissimilar effect on membrane binding, the membrane penetration, and the activity of purified Lp-PLA2, which varied with the diverse composition of membrane vesicles. 43 In addition, different species of oxPLs induce distinctive changes in the physicochemical properties of the lipid particles/membranes containing them, presumably depending on the chemical nature of polar functional groups in their oxidized fatty acyl chain. Therefore, the presence of oxPLs in LDL is predicted to regulate the recruitment and the catalysis of Lp-PLA2.
Notably, the nature of Lp-PLA2 distribution among lipoproteins is species-dependent. In species that express lower concentrations of LDL, such as guinea pigs, rats, and mice, Lp-PLA2 is carried exclusively by HDL in plasma. A structure-based study revealed that the lack of Trp and Leu in mouse Lp-PLA2 at the positions corresponding to Trp115 and Leu116 in the human protein might be responsible for the varying distribution.
Moreover, the levels of LP-PLA2 vary extensively among mammalian species, and basal Lp-PLA2 activity in humans is 10-fold lower than that in mice. These discrepancies might lead to difficulty in reproducing the results from animal experiments in humans, especially allowing for the different pathophysiological functions of Lp-PLA2 within diverse lipoproteins. 60 Thus, continued efforts to elucidate the physiological consequences resulting from these differences among species are essential. Another issue that needs to be taken into account is that Lp-PLA2 possibly utilizes the same regions to interact with both LDL and HDL, as determined based on its structure and hydrogen-deuterium exchange experiments. 44 Consequently, further studies to identify the HDL and LDL regions should be conducted with both lipoprotein particles in parallel, considering the fact that previous results were obtained from binding experiments focused exclusively on only one lipoprotein particle.

| Structural characteristics of Lp-PLA2
Lp-PLA2 is a secreted enzyme whose first 17 residues (Met1 to Ala17) presumably constitute a hydrophobic signal peptide. The N-terminus of endogenous Lp-PLA2 protein is heterogeneous and possesses Ser35, Ile42, or Lys55, while the C-terminus contains Asn441 and includes two potential N-linked glycosylation sites at the residues Asn423 and Asn433. 63 The primary structure of Lp-PLA2 is unique and homologous to a conserved Gly-X-Ser-X-Gly motif found in serine esterases, including many lipases. 11 Furthermore, elegant crystallographic analyses performed on Escherichia coli-expressed enzyme with 383 nonglycosylated residues  revealed that the enzyme contains the classic lipase/esterase α/β-hydrolase fold ( Figure 2). The ligand-free crystal belongs to the space group C2 and contains two structurally similar subunits A and B in the asymmetric unit 64 ; the structure between the A and B subunits is not significantly different, with an overall 0.3 Å C α RMSD (root mean square error). Since the enzyme is assumed to function as a monomer, this dimerization is merely due to crystal packing and plays no functional or physiological role. Noticeably, an atypical cis-peptide bond exists between Phe72 and Asp73 in each subunit of both the structures reported and the presence of this bond is further supported by the distinct electron density of this region. 65 A catalytic triad comprising Ser273, His351, and Asp296 and an oxyanion hole constituted by the backbone NH groups of Leu153 and Phe274 make up the binding site of Lp-PLA2 ( Figure 3A). In the active site, nucleophilic Ser273 is located at the N-terminus of a core α-helix and Asp296 and His351 are positioned suitably to activate Ser273 for catalysis ( Figure 2A). The Leu153 and Phe274 of the oxyanion hole stabilize the incipient negative charge present on the tetrahedral intermediate produced during the ester hydrolysis of a substrate. As observed in other lipases, the catalytic triad is located within a hydrophobic pocket that is oriented toward its substrate, facing the aqueous phase and lying just above the interface ( Figure 2B). As a result, the active site would theoretically allow substrates to enter from the aqueous phase as well as from both lipoprotein particle carriers. Additionally, the binding pocket of Lp-PLA2 is large enough to accommodate bulky substrates, while its predilection for substrates with short and polar sn-2 chains is possibly due to their aqueous phase solubility, which may influence their accessibility to the active site. 42 Furthermore, the positions of the main single-nucleotide polymorphism (SNP) sites are shown in Figure 3B. Mutations on residues Q281 and V279 might disrupt the active site, while those on residues R92, V379, and I198 are believed to affect the interactions between Lp-PLA2 and lipoproteins. were validated to covalently bind at the active site Ser273 by mass spectral data. The crystal structures of these covalent complexes revealed well-defined differences in electron densities with ordered side-chain positions for all the residues modeled. The side-chain positions of these OP complex crystal structures were fundamentally identical to those in the ligand-free protein structure, as the overall C α RMSD between each OP complex and the ligand-free structure ranges from 0.25 to 0.5 Å. Apart from the tabun adduct structure containing three subunits, the other four complexed structures were determined to be in the space group C2 with two subunits in the asymmetric unit. 66 Furthermore, the other four complexed structures showed similar positions of the active-site residues and neighboring residues, which was consistent with the tetrahedral intermediate of the reaction mechanism. Nevertheless, the structure of tabun complexed with Lp-PLA2 had an overall C α RMSD of 0.5 Å compared with the ligand-free structure, which resulted in differences in the contact between Ser273 and its adjacent residues. In addition to the covalent bond with Ser273, the double-bonded phospho-oxygen within the five OPs interacted with the backbone amide of Phe274 and Leu153 through hydrogen bonds located in the oxyanion hole of the enzyme. In the case of the active-site His351, paraoxon and DFP formed a hydrogen bond with the side chain nitrogen of the residue through the oxygen atoms in the alkoxy portion, which prevented the amino acid from reverting to its unprotonated form. However, sarin and soman had chiral centers at the P atom and showed hydrogen bonding with His351 only in specific P stereoisomers, while tabun, containing a chiral P atom, failed to generate an interaction with His351, probably due to the electronic effect of the N,N-dimethyl group displacing the active site His351 away from its native and functional conformation ( Figure 4A and 4B). In the complex of Lp-PLA2 with tabun, good binding stability might be attributed to a CH-π interaction of the N,N-dimethyl group with Phe322 and Trp298 of subunit A and Phe51 of subunit C. Notably, the tabun adduct on Lp-PLA2 additionally showed a stereoselective preference for the P R stereoisomer, whereas both the P R and P S stereoisomers of sarin and soman apparently bind to the enzyme with equal occupancy. 64 OP compounds prevent the degradation of the neurotransmitter acetylcholine through the inhibition of acetylcholinesterase (AChE), posing a fatal threat to human life. Subsequent to binding AChE, OP compounds undergo a rapid dealkylation process referred to as aging, which irreversibly prevents the enzyme from being reactivated by nucleophiles to regenerate the enzymatic activity. However, if the compounds do not age upon binding, the complex may become reactivated by the attack of an activated water molecule or an activated nucleophile. In addition, numerous serine hydrolases, including Lp-PLA2, have been shown to be inactivated by OP compounds. Efforts have focused on engineering diverse serine hydrolase enzymes to neutralize or scavenge nerve agents in the plasma before they reach the AChE targets in the peripheral and CNSs. Thus, Lp-PLA2 seems to be an excellent bioscavenger target due to its presence in human blood. 64 the asymmetric unit of both structures. Additionally, the frameworks of these reversible inhibitor-bound complexes were similar to those of the earlier covalent-bound structures as well as the ligand-free form. The binding pocket of these inhibitors primarily relies on a gorge that initiates from the catalytic triad and extends to two solventaccessible α-helices (residues 114-126 and 362-369), which were validated to participate in binding to the lipoprotein particles LDL or/and HDL ( Figure 5A and 5B). These inhibitors perfectly nestled into the binding gorge and usually contained three binding sites, including the catalytic center, hydrophobic region, and solvent-accessible area ( Figure 5C).
Our group first determined the crystal structures of human Lp-PLA2 bound with two potent inhibitors, darapladib ( Figure 6B) and compound 1 ( Figure 6A), which were diffracted to 2.7 and 2.37 Å, respectively.
Compound 1 was recently identified as an Lp-PLA2 inhibitor by our group. The pyrimidinone carbonyl of darapladib and compound 1 interacts with the oxyanion hole in the enzyme through two bonds to the backbone amides of Phe274 and Leu153. In addition, several residues (Leu371, Phe357, Trp298, Leu153, Leu111, and Phe110) display hydrophobic interactions with darapladib. The two benzene rings of the biphenyl moiety form an intramolecular edge-to-face π-stacking interaction with the fluorophenyl ring and extend into a predominantly lipophilic channel toward Phe125. Apart from two conserved H-bonds, the complex of Lp-PLA2 and 1 formed a third and unique H-bond between the oxygen linked to the pyrimidone ring and the side chain of Gln352. The residues that contribute to the hydrophobic interactions are identical to those presented in the Lp-PLA2 and darapladib complex, except that Gly152 and Phe322 are added while Leu111 and Leu369 are missing. Additionally, an edge-to-face π-π interaction is formed between the terminal benzene ring of compound 1 and Phe357. The superimposition of the two complex structures reveals that the bound darapladib and 1 greatly overlap except the diethylamine group of darapladib, as no corresponding group is found in 1. Further isothermal titration calorimetry measurements demonstrate that the enthalpic effects drive the binding of two inhibitors to Lp-PLA2, which is consistent with the multiple protein-ligand interactions revealed by the two complex structures. 32 Compound 2, a bioisostere of darapladib, showed improved physicochemical properties compared with darapladib ( Figure 6C). The crystal structure of bioisosteric analog 2 in the Lp-PLA2 protein was solved at 1.9 Å resolution and revealed a binding precludes the adjacent trifluorophenyl moiety extending as far toward Leu121 and Phe125, while showing no effect on the main interactions within the oxyanion hole. 68 Currently, FBLD has gradually matured to be a reliable and efficient approach for producing promising lead compounds for subsequent medicinal chemistry optimization, and more than 30 drug candidates derived from fragments have entered in clinical trials. 69 On the basis of the aforementioned crystal structures, both Astex in collaboration with GSK and our group applied the FBLD strategy to identify novel Lp-PLA2 inhibitors, consequently releasing numerous crystal structures for fragment hits or representative compounds. Typically, compounds 3 Concurrently, the sulfone moiety sits in the Leu153, Trp298, and Phe322 groove, where the electron withdrawing effect of sulfone might polarize the neighboring CH that then interacts with the electron-rich indolyl of Trp298. In particular, the linker between the phenyl and sulfone moiety is vital for maintaining the inhibitory potency, as it can orient the two parts to the respective binding pocket, and the ether oxygen forms a close contact with the side chain of Gln352. Compared with darapladib, compound 3 efficiently utilizes the groove above Trp298 to enhance the potency, as Trp298 appears to be essential for substrate interactions with Ser273. Since the groove extends toward the solvent, occupying the space can also modulate the physicochemical properties of the ligand. 34 hydrophobic interactions via the 4-chloro-3-(trifluoromethyl) phenyl moiety, which sits above the lipoproteinbinding helices on the other side. Remarkably, Lp-PLA2 rotates the Gln352 residue 90°from its position in the structure of the original fragment to that of compound 4, resulting in a newly ordered conformation mediated by two H-bonds with Lys370 and a water molecule. Nevertheless, this rotation does not seem to significantly influence the enzymatic potency or protein framework. 33 Unlike other reversible inhibitors, compound 5 does not directly interact with the catalytic residues of the enzyme. Furthermore, the terminal tolyl moiety of the bis-aryl group in compound 5 inserts into a new pocket and forms πstacking interactions with Phe125 and Phe357, resulting from a −22°rotation of the C α -C β bond in Phe357. The rest of the tolyl is in the van der Waals contact with the side chains of Leu111 and Leu371 and forms an edge-to-face πstacking interaction with Phe357. Additionally, two direct hydrogen bonds can be found between the amino of the aminoethyoxy moiety and the Leu369 carbonyl as well as between the hydroxyl of the hydroxylmethy moiety and the Gln352 amino. A small lipophilic subpocket formed by the side chains of Ala355, Tyr160, and Phe357 is appropriately filled by the methoxy moiety. In addition, a number of water-mediated contacts with a network of waters are found across the top of the pocket, which may also contribute to improving the affinity. 35 The structural information obtained from the complexes between Lp-PLA2 and several reversible inhibitors validate that the inhibitor-binding pocket is fairly open and large and those hydrophobic interactions play a major role in binding.
Except for compound 5, the other five compounds share common pharmacophore features, such as a carbonyl or sulfonyl occupying the oxyanion hole, a hydrophobic group located in the lipophilic pocket above two solvent-accessible α-helices, and a hydrophilic moiety extending toward the solvent area. Compound 5 is still effective in inhibiting Lp-PLA2 though it does not occupy the active site, further suggesting a larger inhibitor-binding pocket. Moreover, specific interactions with Trp298 or its adjacent residues might contribute to the improved potency as compound 3 did.
Altogether, the elucidation of the binding mechanism and detailed interactions of the reversible inhibitors with Lp-PLA2 provide valuable information for further design and development of novel Lp-PLA2 inhibitors. More importantly, the successful application of FBLD in identifying novel efficient Lp-PLA2 inhibitors with vastly improved physicochemical properties can be used for reference in other targets.

| GENETIC VARIABILITY AND REGULATION OF EXPRESSION
The gene encoding Lp-PLA2 (PLA2G7), whose chromosomal location is in 6p12-21.1, is organized in 12 exons and encodes 441 amino acids. 70,71 Numerous variations were identified within PLA2G7, and the corresponding alterations in physiology and body homeostasis have been illustrated to varying degrees. Since elevated Lp-PLA2 levels were detected in patients with certain diseases, most researchers have focused their attention on variants that considerably weaken Lp-PLA2 activity or function, viz the loss-of-function alleles. The functional and clinical consequences associated with PLA2G7 polymorphisms were summarized in a review by Stafforini 24 in 2009, which mainly discussed the research status of Val279Phe, Arg92His, Ile198Thr, and Vla379Ala SNPs at that time. The clinical consequences were largely focused on cardiocerebrovascular diseases, which was considered the most promising field in which Lp-PLA2 was involved. In addition, several diseases that were rarely reported to be associated with PLA2G7 polymorphisms, such as asthma, schizophrenia, CKD, multiple sclerosis, ulcerative colitis, and Kawasaki disease. Notably, however, the results of most research mentioned in that review, regardless of functional or clinical consequences, were inconsistent. 24 In this review, we also described the research state of the PLA2G7 polymorphisms from two aspects, the functional and clinical consequences, while emphasis will be placed on recent results.

| Val279Phe mutation (rs76863441)
A base G to T transition at nucleotide position 994 in exon 9 of PLA2G7 leads to a Val to Phe substitution at residue 279, which is in proximity to the active site of the enzyme. The V279F mutant gene is codominantly inherited, and the mutant protein is almost undetectable in the plasma of a homozygous person (279FF), as the misfolded protein cannot be secreted normally. 24 The prevalence of 279FF was reported to be 0.4%, 1.2%, and 3% in the Chinese, Korean, and Japanese populations, respectively, while it is rare in the Middle Eastern population and almost completely absent in Europeans. 72 On the basis of the regional nature of the prevalence, the correlations between the V279F mutation and diseases was widely determined in the East Asian population and focused on the cardiocerebrovascular field.
The clinical consequences associated with the V279F polymorphism were summarized in separate reports by Stafforini 24 and Karasawa,73 in which almost all the results were obtained before 2010. However, in contrast to the distinct effect on Lp-PLA2 activity, the influence of the V279F mutation on specific diseases remains unclear. The results from the studies discussed in these two reviews were inconsistent, as positive, negative, and nonsignificant associations between V279F mutation and diseases were all described in different studies, even for the same type of disease. Similarly, such contradictory results were observed in several literature reports over the past couple of years (Table 1). Earlier, two meta-analyses aiming to determine the association between the V279F polymorphism and coronary heart disease (CHD) risk did not support a role of V279F in increased risk of CHD (the F allele vs the V allele: odds ratio (OR) = 1.14, 95% confidence interval (CI): 0.86-1.52; OR = 1.09, 95% CI: 0.88-1.35, respectively). 74,75 This observation was further substantiated by two recent large-scale human genetic studies.
One was in a Chinese population, and almost 90 000 individuals with the F279 allele 76 were randomly selected for identification the associations of V279F with various vascular diseases. However, neither positive nor negative correlations were demonstrated between V279F and any vascular diseases. Another large correlative study also illustrated that the V279F polymorphism was not related to CHD risk (OR = 0.95, 95% CI: 0.88-1.03). 77 Nonetheless, harboring the F279 allele was identified as protective against the development of CAD in South Korean males (OR = 0.80, 95% CI: 0.66-0.97, P = .02). 78 Interestingly, inconsistent results were obtained even for the same ethnic population and disease as well. For example, a case-control study demonstrated no significant association between the V279F mutation and ischemic stroke in a Chinese population (OR = 0.82, 95% CI: 0.44-1.24), 79 whereas another genetic study established an increased risk despite using the same study population and disease (OR = 2.51, 95% CI: 1.02-5.91, P = .037). 80 In the case of atherosclerosis, a significant inverse association of V279F polymorphism with clinical atherosclerosis was indicated in a recent meta-analysis (OR = 0.88, 95% CI: 0.81-0.95, P = .0007), 81 whereas another study did not reveal any significant associations between the V279F genotype and subclinical atherosclerosis using Mendelian randomization analyses. 82 Therefore, the dramatic reduction of Lp-PLA2 levels in plasma resulting from the V279F mutation is apparent, but the relationship between the V279F mutation and cardiocerebrovascular diseases is unclear. In this context, deeper The "n" column represents the number of participants in each study category and includes subjects harboring 0-1-2 polymorphic alleles. The "P" column represents whether the polymorphism incidence is statistically significant differences between control and diseased populations, where available/appropriate. Abbreviations: BSS, blood stasis syndrome; CHD, coronary heart disease; CVD, cerebrovascular disease; N/A, not applicable; N/S, not statistically significant. and more comprehensive research may be necessary to elucidate the actual role of the V279F mutation in the pathology of cardiocerebrovascular diseases.
Additionally, the potential contribution of the V279F mutation to the occurrence or progression of other diseases was tested in recent studies. For example, both genotype and allele frequencies of the V279F mutation in Chinese populations with polycystic ovary syndrome (PCOS) 83,84 and acute pancreatitis (AP) 85 were significantly higher than those in controls. Furthermore, multifactor logistic regression analysis revealed that the V279F mutation was an independent risk factor for AP (OR = 4.03, 95% CI: 1.27-12.7, P = .018). However, the V279F mutation was not associated with a reduced risk of AD in a Japanese population (OR = 0.98, 95% CI: 0.86-1.12, P = .81). 86 Interestingly, a recent study demonstrated that overweight Korean individuals harboring the F allele exhibited lower systolic and diastolic blood pressure than those harboring the V allele (both P < .05) at the end of the 3-year follow-up, though no significant differences were found at baseline. This result implied that the F allele might protect carriers against hypertension, even in the case of persistent overweight for longer than 3 years. 87 Nevertheless, some potential limitations existed in these studies, such as lower power in subsidiary analyses due to small sample size and incomplete conclusions regarding the effect of the V279F mutation on diseases resulting from specific inclusion criteria.

| Val379Ala mutation (rs1051931)
A systematic screening of the PLA2G7 gene revealed a base A to G alteration at position 1136 of exon 11, which resulted in a valine to alanine replacement at residue 379. 88 Peculiarly, the ancestral allele at this position is base A, but the frequency of mutant base G predominantly reaches approximately 80%. 77 Although the V379A mutation is relatively common in European and African populations, its effect on plasma Lp-PLA2 activity is unclear until recently. Earlier, Casas et al 89  with PCOS (P > .05), 84 AD (P = .33), 85 and CHD with (P = .21) or without blood stasis syndrome (BSS) (P = .98). 93 Interestingly, two studies demonstrated increased Lp-PLA2 activity in A379 allele carriers. The relatively smallscale study described dramatically increased activity in homozygous AA subjects (28.9 ± 6.9 nmol/mL) compared with homozygous VV subjects (5.6 ± 2.3 nmol/mL, P < .001), while the large one detected only a moderate increase. 38,94 However, functional studies revealed that the Val379 would lower the substrate affinity (namely, higher Michaelis-Menten constant, K M ) and increase the maximal velocity (V max ) in vitro. 95 Hence, the effect of the V379A mutation on plasma Lp-PLA2 activity is controversial thus far, but the degree of action may be limited for each direction.
Due to its potential impact on plasma Lp-PLA2 activity, the V379A mutation was explored in patients with a variety of diseases. Compared with the ancestral allele A (V379 allele), the derived allele G (A379 allele) presents at a much higher frequency. Therefore, most studies have concentrated on the influence of the minor V379 allele on diseases, in which this allele was regarded as the mutant variant. In the Chinese population, the frequencies of the A379 allele were not significantly different between patients with AP (P = .71) or PCOS (P = .88) and their respective healthy controls. Moreover, no associations were observed between the V379A genotype and the risks of AP or PCOS (both P > .05). 84

| Arg92His mutation (rs1805017)
In addition to the V379A mutation, Bell et al 88  including data from five community-based studies revealed a strong association of the R92H mutation with increased Lp-PLA2 mass (P = 2.4 × 10 −23 , significance threshold P < 5 × 10 −8 ), but a relatively weak association with Lp-PLA2 activity (P = 2.4 × 10 −6 ). 90 Furthermore, three studies successively discerned that a significantly increased Lp-PLA2 level was associated with the H92 allele in the Chinese population, and of these, two were about Lp-PLA2 mass and the third was about activity. 84 Lp-PLA2 mass and activity in high-risk CAD patients of European ancestry. In addition, a large genetic study revealed a trend toward an inverse association between Lp-PLA2 activity and the H92 allele (genotype HH vs RR, OR = −0.02, 95% CI: −0.05 to 0.00). 89 In a sizable pharmacogenetic analysis reported recently, Yeo et al 38 found that the R92H common variant was associated with a slightly reduced Lp-PLA2 activity. As the authors of the aforementioned three studies provided only data on Lp-PLA2 activity, we could not compare the effect on Lp-PLA2 mass among them. 38 Though divergent, the impact of the R92H polymorphism on plasma Lp-PLA2 levels might be modest, and the characteristic biochemical properties of the protein would not be significantly altered. 99 Comprehensive genetic analyses were carried out to determine the influence of the R92H variant on the susceptibility and severity of various diseases. Similar to the previous two variants, researchers also explored the associations of the R92H polymorphism with AP 85 and PCOS. 84 Consequently, a significant association was observed only in AP patients (OR = 1.96, 95% CI: 1.02-3.76, P = .043), in whom the frequency of the H92 allele was higher than that in the healthy controls. 85

| Ile198Thr mutation (rs1805018)
The I198T mutation, first termed the Iso195Thr polymorphism, was another common frequency variant found by Bell et al 88 in 1997. The impact of the I198T variant on the plasma Lp-PLA2 level remained inconclusive. 88 Qi et al 92 found an evident reduction in both Lp-PLA2 mass and activity when carriers homozygous for the minor allele (TT) were compared with the major allele homozygotes (II) in a Chinese population. 92  protein by lowering the substrate affinities rather than its hydrolytic activity.
Since its discovery together with the V379A and R92H mutations more than 20 years ago, a number of studies have been carried out to clarify the involvement of the I198T variant in a variety of diseases. A recent study revealed that T198 allele showed no significant effect on aging in Sicilian octogenarians and either risk or age of onset of AD in a Japanese population. 86 Similarly, Sankararaman et al 106

| Other SNPs
A number of other missense mutations were explored in relation to their effect on either plasma Lp-PLA2 level or disease progression but were usually of lower prevalence in the studied population. The Gln281Arg mutation (rs201256712) could lead to a Gln to Arg replacement at residue 281, which occurs rarely in exon 9 of PLA2G7 (minor allele frequency (MAF) < 0.5%). 110 Three studies determined the impact of the variant on plasma Lp-PLA2 activity and revealed its effect on the reduction in enzymatic activity, though to a different extent among the three investigations. 77,110,111 Furthermore, two studies using computational modeling found that the Q281R variant was deleterious to the structure and function of Lp-PLA2. 112,113 To identify rare loss-of-function mutations of the  plasma Lp-PLA2 mass or activity, whereas a significant association was detected with CAC. In addition, a number of other intronic variants in the PLA2G7 gene were also explored, and their impact on the function and expression of Lp-PLA2 was preliminarily demonstrated with respect to the susceptibility and severity of diseases. 90,105 Furthermore, numerous variants located in the 5′ upstream, 5′-untranslated region (5′-UTR), and 3′-UTR might be associated with the plasma Lp-PLA2 level and even certain diseases. Rs1421378, an intergenic variant, might impair plasma Lp-PLA2 activity with a gene-dose effect, 89 although two studies did not observe any significant association. 98 Usually, a significant association between elevated plasma Lp-PLA2 activity and increased risk of CHD could be detected, and certain SNPs were also found to be responsible for the altered activity. However, a number of large genetic studies failed to demonstrate any association between SNPs and CHD risk, while several other reports described either a positive or a negative association. In particular, in a recent large-scale study that included more than 260 000 total participants, the researchers genotyped four rare loss-of-function mutations (c.109 + 2T > C, R82H, V279F, and Q287X) and one common modest-impact variant (V379A) in PLA2G7. They compared the effects of darapladib treatment and Lp-PLA2-lowering alleles on plasma Lp-PLA2 activity, conventional cardiovascular risk factors, and CHD risk. As a result, none of the five functional alleles, which induced a widely differing degree of reduction in Lp-PLA2 activity, was connected to CHD risk. Thus, the authors concluded that Lp-PLA2 is unlikely to be a causal risk factor in CHD. 77 In contrast, Ferguson et al 105 121 observed that the minor allele of rs7756935 was associated with increased platelet activity in Caucasians with type 2 diabetes and a Chinese population with ischemic stroke, respectively, signifying a higher risk of aspirin resistance in carriers than noncarriers. In addition, the I198T variant might be responsible for the interaction effects between genotype and omega-3 polyunsaturated fatty acid (n-3 PUFA) supplementation on plasma triglyceride (TG) levels. 122 However, a recent pharmacogenetic study revealed that none of the variants associated with Lp-PLA2 activity, including the PLA2G7 gene, was related to any efficacy endpoints of darapladib treatment. 38 In summary, although much progress has been made in this field, the inconsistent results require further comprehensive and powerful research to eliminate divergence. Moreover, the knowledge acquired from the studies on vascular diseases could be applied to other diseases, which might uncover the unanticipated impact of PLA2G7 variants on certain diseases or specific treatments.

| Transcriptional regulation of Lp-PLA2 expression
Hematopoietic stem cell-derived cells, such as monocytes, macrophages, mast cells, and T-lymphocytes, are the main sources of the Lp-PLA2 protein. 25,123 Messenger RNA (mRNA) for Lp-PLA2 is extensively detected in several kinds of tissues, which is most likely the result of the expression of macrophages in these tissues. 123 Analogous to the housekeeping genes, the PLA2G7 promoter lacks a TATA box but harbors a set of GC-rich motifs adjacent to the transcription start site; however, it seems more like a tissue-specific gene that has the same TATA-less promoter. 11 Stafforini et al determined the transcriptional initiation site and identified the essential elements of the promoter region that afforded maximal basal promoter activity. Additionally, they observed negative regulation when reporter constructs consisted of specific additional 5′ upstream sequences. 123  that the state of cellular differentiation was also able to affect the responses to exogenous stimulation. 123,127 Additionally, enzymatic substrates, such as PAF and oxidized low-density lipoprotein, were also identified to upregulate Lp-PLA2 expression at the cellular level. 25,41,128,129 In vivo, Miyaura et al 130  the greatest increase was detected in the lung followed by the spleen, liver, kidney, and thymus. In addition, a twofold increase in plasma Lp-PLA2 activity was detected at 24 hours following LPS exposure, but no change was observed at 12 hours. In male SD rats, Svetlov et al 135 described a maximal four-to fivefold increase in bile Lp-PLA2-specific activity at approximately 2.5 hours after LPS infusion, followed by a gradual decline to baseline levels within 18 hours. However, plasma Lp-PLA2 activity was significantly elevated 18 hours after LPS exposure, whereas no difference was detected at 5 hours compared with saline injection. 135 In Syrian hamsters, the liver, spleen, lung, and small intestine exhibited strikingly increased Lp-PLA2 mRNA levels after a 16 hours exposure to LPS, and the increase in plasma peaked at 24 hours and was sustained for at least 48 hours. 136 The increased plasma Lp-PLA2 activity was also determined in two other studies after LPS administration in C57BL/6 mice. 136,137 Nevertheless, when a lethal dose of LPS was injected into Swiss mice, plasma Lp-PLA2 activity was found to decrease as observed in sepsis patients. 138 Similarly, two studies reported a diminished plasma Lp-PLA2 activity in response to LPS challenge in a sepsis mouse model, 137 and one of them resulted in a noticeable decline that persisted for up to 48 hours followed by a modest recovery. 139 Treating healthy volunteers with an approved dose of LPS produced a dynamic plasma Lp-PLA2 response according to unrelated studies, where the expression of Lp-PLA2 mRNA displayed an initial decline, followed by an increase and finally recovered to the initial levels. 140  dynamic range, namely, the "expression window," was a crucial determining factor in response to endotoxemia, and out of the "window" in either direction might be detrimental. 60 In addition, recent research revealed quite a few potential regulators of Lp-PLA2 expression. Nitro-oleic acid (OA-NO2), a nitrated fatty acid, might downregulate Lp-PLA2 expression in THP-1-derived macrophages through multiple signaling pathways. 142 Apolipoprotein CIII (Apo CIII) is one of the major components of TG-rich lipoproteins and is chiefly synthesized in the liver. In vitro, Apo CIII could induce Lp-PLA2 expression via MAPK and NFĸB pathways. Moreover, blood from Apo CIII transgenic pigs displayed a fourfold increase in plasma Lp-PLA2 activity and a 10-fold increase in Lp-PLA2 mRNA in macrophages. 143

| BIOLOGICAL FUNCTIONS AND DISEASE IMPLICATIONS
Lp-PLA2 was shown to have the ability to degrade the PAF and its analogs and the oxLDL species containing short, oxidized, and/or truncated sn-2 chains, without any threat to the integrity of phospholipid components within cellular membranes and lipoproteins. 48 In this context, Lp-PLA2 might exert anti-inflammatory effects, as PAF and its analogs and truncated oxLDL potentially induce inflammatory responses. [148][149][150] However, increasing evidence suggests that Lp-PLA2 seems to be a promoter in the development and progression of inflammation, primarily due to its proinflammatory products, lysoPC and oxNEFA, as previously described. 11,13 In addition, the binding of Lp- Furthermore, many large-scale epidemiological studies have found that elevated plasma Lp-PLA2 levels are associated with an increased risk of coronary disease, stroke, and/or mortality, further substantiating the proatherogenic role of Lp-PLA2. 152 In addition, downregulating the expression of the PLA2G7 gene by RNA interference could induce ameliorative inflammation and atherosclerosis in apolipoprotein E-deficient mice. 153 Taken together, these findings indicate that Lp-PLA2 might be crucial during the development of atherosclerosis and in determining plaque instability via inflammatory pathways.
Considering its potential proatherogenic role, inhibition of Lp-PLA2 might retard atherogenic progression.
Consequently, a number of Lp-PLA2 inhibitors were discovered, among which, darapladib (developed by GSK) is the most advanced and most widely studied Lp-PLA2 inhibitor. 154 As a potent oral inhibitor of Lp-PLA2, darapladib was used as a candidate drug to elucidate the influence of Lp-PLA2 inhibition on a range of cardiovascular and CVDs, including stroke, myocardial ischemia, subclinical atherosclerosis, CHD, and heart failure. 155 Several limitations were proposed to explain the failure of the previous clinical trials. For example, nearly 95% of the patients received statins as the basic treatment, which was associated with two indefinite concerns. One concern was that the efficacy of Lp-PLA2 inhibitors in improving the outcomes in well-managed statin-treated patients was limited, as statins could not only inhibit plasma Lp-PLA2 levels but also could reduce the occurrence of HUANG ET AL. | 101 CHD events. 158 A different concern was the rationality of the inclusion criteria for the population under study.
Even after intensive statin therapy, many atherosclerosis patients suffered high rates of recurrent cardiovascular events. These patients could be classified into two categories: one represents a group with residual cholesterol risk (high sensitivity C-reactive protein (hsCRP) <2 mg/L and low-density lipoprotein cholesterol (LDL-C) ≥70 mg/dL), and the other presented a residual inflammatory risk (hsCRP ≥2 mg/L and LDL-C <70 mg/dL). 159 Recently, canakinumab, an anticytokine agent targeting IL signaling pathways, was demonstrated to significantly reduce cardiovascular event rates following statin therapy, signifying a proof of principle that reducing inflammation in the absence of lipid lowering could diminish vascular event rates. In particular, the studied population in that clinical trial comprised stable postmyocardial infarction patients with hsCRP levels ≥2mg/L. 160 However, the inclusion criteria of the SOLID-TIMI 52 and STABILITY trials did not involve the hsCRP levels. Consequently, we supposed that patients with the residual inflammatory risk might benefit from Lp-PLA2 inhibition. Furthermore, the possibility that Lp-PLA2 might not be a suitable target to address the residual inflammatory risk could not be excluded, based on the observation that broad-spectrum anti-inflammatory therapy by methotrexate failed to reduce cardiovascular events while specific cytokine inhibition by canakinumab could. 161 Additionally, the study failed to identify the beneficial effects of treatment on patients with higher baseline levels of Lp-PLA2 activity, as patients were screened based on their Lp-PLA2 activity levels. 19,20 Also, the effects of prolonged Lp-PLA2 inhibition could not be determined since only 3 to 4 years of median follow-up were recorded. 77 Nevertheless, the failure of these clinical trials definitely dims the future of Lp-PLA2 inhibitors as therapeutics for atherosclerosis but does not limit their potential to address other diseases.
Currently, Lp-PLA2 is considered a reliable biomarker of cardiovascular risk, and many reports have established Lp-PLA2 activity as an independent predictor of CHD outcomes in the general population. [162][163][164][165][166][167][168][169][170][171][172] Since the vascular specificity of Lp-PLA2 is different from conventional risk factors, in 2014, the Food and Drug Administration approved the Lp-PLA2 test for patients without existing coronary disease measure a person's risk of heart disease, cardiac arrest, and the potential for other heart problems. 166 Furthermore, the European Society of Cardiology recommended Lp-PLA2 as part of a refined risk assessment in patients at high risk of a recurrent acute atherothrombotic event, and the American Heart Association/American Stroke Association recommended that measurement of the Lp-PLA2 in patients without CVD might be helpful in identifying patients at an increased risk of stroke. 169 Very recently, a pharmacogenetic (PGx) study was conducted to assess the influence of genetic variants on efficacy and tolerability in the aforementioned two trials. An analysis of darapladib efficacy endpoints was performed on data from more than 23 000 subjects referring to approximately 9.3 million common and low frequency variants. However, the efficacy analysis identified only one common locus (rs181937009), which conferred risk in the placebo arm but was protective in the darapladib treatment arm. Moreover, none of the variants that were significantly associated with baseline Lp-PLA2 activity was associated with efficacy endpoints. 38 This finding was validated in another genetic study with 261 950 total participants, which identified that both Lp-PLA2-lowering alleles and darapladib treatment were not related to CHD risk. 77 Altogether, this evidence indicates that Lp-PLA2 is a reliable biomarker rather than a causal risk factor for CVDs, but this view remains to be confirmed. 173

| Alzheimer's disease
AD, pathologically characterized by the deposition of amyloid β (Aβ) peptide and neurofibrillary tangles formed by tau proteins in the brain, is the most common cause of dementia in elderly individuals. 174 Breakdown of the BBB, a highly specialized brain endothelial structure of the fully differentiated neurovascular system, seems to be involved in the etiopathology of AD. In conjunction with astrocytes and microglia, the BBB elaborately regulates the exchange of substances between the blood and brain to uphold the microenvironment homeostasis of the CNS. 175 In particular, the tight junctions between the brain vascular endothelial cells (BVECs), comprising occludin proteins, claudin proteins, and junctional adhesion molecules, are vital for the structural and functional integrity of BBB. 176 Therefore, BBB breakdown, due to disruption of the tight junctions, inflammatory responses, or other risk factors, may enhance its permeability and induce the leakage of harmful substances from blood vessels into the brain, contributing to the process of AD. 175,177 Very recently, Nation et al 178 demonstrated that individuals with early cognitive dysfunction develop brain capillary damage and BBB breakdown in the hippocampus and established BBB breakdown as an early biomarker of human cognitive dysfunction in individuals with and without Aβ or phosphorylated tau (pTau) positivity. Therefore, the authors concluded that neurovascular dysfunction might be a previously underappreciated promotor of AD, independent of the classic pathophysiological hallmarks. 178 In this context, Lp-PLA2, a specific marker of vascular inflammation, was investigated to determine its role in AD.
The epidemiological investigations on the relationship between Lp-PLA2 activity and AD yielded inconsistent results. Two studies demonstrated no significant association between plasma Lp-PLA2 mass and AD. 179,180 In addition, Davidson et al 181 observed that plasma Lp-PLA2 activity was not associated with a diagnosis of AD, since no strong correlations were found between Lp-PLA2 activity and cerebrospinal fluid (CSF) markers of AD.
Furthermore, for the PLA2G7 null mutation (V279F), Lp-PLA2 activity was not associated with a diminished risk of AD in a Japanese population. 86 However, several limitations might be responsible for the lack of association, such as the small size of AD cases, age at which the markers were tested, inadequate or overlong duration of follow-up (this effect has been discussed in two studies), 182 classification, we believed that this investigation was effective to in detecting the association between Lp-PLA2 and the risk of AD. Overall, though inclusive, Lp-PLA2 is likely to be a risk factor for AD.
Initially, darapladib, a selective and orally effective Lp-PLA2 inhibitor, was investigated for its ability to inhibit the development of advanced atherosclerotic lesions in a diabetic-hypercholesterolemic (DMHC) porcine model.
The researchers unexpectedly discovered that darapladib treatment resulted in improved alertness and activity in DMHC animals compared with untreated DMHC animals. In consideration of the increased BBB permeability of DMHC animals, they hypothesized that the observed behavioral improvements in darapladib-treated animals were probably due to the beneficial effects on the functional integrity of the BBB. To explore this speculation, they monitored the variation in BBB permeability in DMHC pigs after darapladib treatment for 6 months. Compared with the untreated group, the darapladib-treated group showed significantly alleviation of the deleterious effects induced by prolonged diabetes mellitus (DM) and hypercholesterolemia (HC) in the brain of DMHC pigs. In particular, darapladib improved the compromised BBB permeability and reduced the influx of plasma components into the brain tissue, which occurred principally in the arterioles within the microvasculature. 186 A subsequently published study revealed that claudin-5 and occludin, tight junction proteins in BVECs, might be involved in the altered BBB permeability. 36 Using immunoglobulin G (IgG) as a visual indicator of local BBB breakdown and vascular leak, they established that the leaky microvascular IgG bound preferentially to pyramidal neurons in the cerebral cortex. Moreover, the authors also detected enhanced deposition of the Aβ42 peptide predominantly on the pyramidal neurons within the brain parenchyma of DMHC pigs, which are the same neurons in human AD brains that accumulate excessive amounts of Aβ42-positive material. These two alterations within neuronal cells potentially induced functional disruption and pathological changes, which could be recovered to nearly normal levels using darapladib. 186 A phase II trial, which was carried out to reproduce these findings in a relevant AD population, also demonstrated that Lp-PLA2 inhibition could offer beneficial effects on AD progression. The participants with AD and CVD were randomized to treatment with rilapladib, another potent Lp-PLA2 inhibitor developed by GSK, or placebo once daily for 24 weeks on top of their stable background therapy. Compared with the placebo, rilapladib treatment resulted in low levels of plasma CSF neurodegenerative markers, including albumin quotient (AlbQ), total HUANG ET AL. On the basis of the proliferative status of the retinal vasculature, DR has been divided into two stages: a nonproliferative stage (NPDR), characterized by microaneurysms, retinal hemorrhages, vascular tortuosity, and hard exudates; and a proliferative stage (PDR), characterized by the creation of new but leaky blood vessels. 188 In addition, DME is a unique condition of DR, which is characterized by fluid accumulation in the neural retina DME can occur in both the NPDR and PDR phases and is considered the clinical feature that is most strongly associated with vision loss. 189 Even though the pathology and molecular mechanism of DR are complicated, it is probably not only a vascular disease but also a neurodegenerative disease. Moreover, increasing evidence suggests that inflammation is a critical contributor to the development of DR via both vascular and neurosensory pathways. 188,190 Considering the potential proinflammatory effect of Lp-PLA2, a number of studies were carried out to determine the role of this enzyme in DR. A recent epidemiologic study showed that the activity of plasma Lp-PLA2 is higher in diabetic patients with PDR than in healthy individuals and diabetic patients with NPDR. In addition, the increase in Lp-PLA2 was correlated with the severity of DR. 191 Very recently, a longitudinal disease progression study was reported, aiming to explore the association among Lp-PLA2 activity levels, incident DR, and change in retinopathy grade. In each analysis, the participants were divided into four equal-sized groups based on their Lp-PLA2 activity levels, and the lowest quartile (Q1) was regarded as the reference. Consequently, the authors established that the type Though Lp-PLA2 inhibition showed limited benefit for macroangiopathy in clinical trials, it might efficiently improve certain diseases with microvascular abnormalities, such as DR and AD. The results from the above two phase II clinical trials support the hypothesis that agents with vascular anti-inflammatory properties can ameliorate the progression of diseases characterized by impaired BBB and BRB permeability. Undeniably, these findings need to be confirmed in large clinical trials of long duration, along with comprehensive studies on the pathogenic mechanism, pharmacogenetics, racial specificity, or disease subtype specificity. In general, these findings encourage researchers to develop new therapeutic approaches for chronic neurodegenerative disorders of the BBB.

| Cancers
The role of Lp-PLA2 was also investigated in various cancer types. Aberrant Lp-PLA2 expression exists in a variety of cancers, and most tumor tissues show higher level compared with normal tissues. Moreover, the invasive and metastatic tumor samples exhibited greater PLA2G7 expression compared with the primary tumor samples, which were derived from the breast, colon, kidney, liver, or lung. 18 In terms of pathogenic mechanisms, both pro-and HUANG ET AL.

| 105
antitumorigenic effects have been described for Lp-PLA2. Seemingly, the net effects of Lp-PLA2 on tumors and tumor microenvironments result from the combined actions of its substrates, products, and secondary products.
Previously, Stafforini 195  were detected after silencing PLA2G7. In addition, the PLA2G7 gene was found to be significantly associated with altered phospholipid metabolism in prostate cancer, signifying a key role in the pathogenesis. 198 Overall, these observations indicate that Lp-PLA2 might be a valuable biomarker or therapeutic target for prostate cancer, particularly, in ERG-positive prostate cancers.
Lp-PLA2 activity and mRNA levels in tumor tissues and plasma from colon cancer patients were higher than those in healthy controls. The increased PLA2G7 expression might be synergistically controlled by loss-of-function p53 and Ras activation in murine and human colon cells, and silencing PLA2G7 could decrease the size of the tumor by 42% in vivo. 199  to relapse-free survival. Protumorigenic tumor-associated macrophages (TAMs) seemed to be involved in this process since they express PLA2G7 at higher levels than tumor cells. 203 Interestingly, lysoPC16:0, the major product of Lp-PLA2, was increased in the plasma of early-stage epithelial ovarian cancer patients while other lysoPCs were decreased, suggesting that Lp-PLA2 might play an important role in the development of ovarian cancer. 204 Similarly, TAM-derived Lp-PLA2 was found to enhance the migration of nasopharyngeal carcinoma cells, possibly stimulating tumor invasiveness or metastasis. 205 Altogether, these results indicate that TAMs might exert a tumor-promoting function in various cancers and targeting TAMs, such as inhibiting the induction of PLA2G7 expression, is beneficial for addressing the tumor process. In addition, PLA2G7 gene expression in peripheral blood might be able to predict the response to immunotherapy among melanoma patients, in combination with the expression of three additional genes including cathepsin D (CTSD), thioredoxin reductase 1 (TXNRD1), and interleukin 1 receptor-associated kinase 3 (IRAK3). The ability of PLA2G7 to improve the predictive power of the gene signatures probably results from the regulation of the tumor microenvironment, and an elevated level of PLA2G7 was a predictor of increased survival following anti-cytotoxic T lymphocyte-associated antigen-4 directed therapy. 206 Owing to the autoimmune toxicity induced by immunotherapy, the discovery of the biomarkers, especially blood biomarkers such as PLA2G7, is valuable and could aid the identification of the subset of patients who can benefit from immunotherapy and simultaneously exhibit low risk for autoimmune toxicity. 207,208 Furthermore, PLA2G7 was also identified as a diagnostic marker via a microarray technique for profiling the transcriptome of esophageal cancer cells. It was significantly overexpressed in earlier tumor stages of both histological subtypes, namely, esophageal squamous cell carcinoma and esophageal adenocarcinoma. 209 Due to its potential in promoting cancer cell migration and invasion, the authors speculated that Lp-PLA2 might be a promising anticancer target.
Although significant efforts are needed for establishing the protumorigenic role of Lp-PLA2 in some types of cancer, the results of the existing studies proposed a potential approach to managing cancers through Lp-PLA2 inhibition. Certainly, the approach is challenging since Lp-PLA2 might have a beneficial role in certain cancers such as melanoma, multiple myeloma, and glioblastoma. In this context, precision medicine might be helpful for identifying specific cancer subtypes that could be alleviated by inhibiting Lp-PLA2. 210

| Other diseases
In addition to the impact of Lp-PLA2 on traditional CVDs, its role is also determined for various diseases in which CVD is a complication. Similar to DR, significantly altered Lp-PLA2 levels were observed in diabetic macroangiopathy. Lp-PLA2 was significantly associated with fasting plasma glucose and cardiovascular outcome in diabetic patients. 211  recently. 213 Lin et al 214 proposed that Lp-PLA2 might affect the incidence of atherosclerosis in diabetic patients and the decline in Lp-PLA2 levels through intensive insulin therapy might contribute to protection from diabetic atherosclerotic complications. 214 Additional studies are essential to clarify the role of Lp-PLA2 in modifying the atherosclerotic process of T2DM patients.
HUANG ET AL.

| 107
Macrophages and phospholipases are assumed to play a key role in immune activation and inflammation in the context of human immunodeficiency virus (HIV) infection. 215,216 Notably, HIV-infected patients exhibit increased arterial inflammation that persists even after effective antiretroviral therapy (ART), suggesting the need for adjunctive, anti-inflammatory strategies to prevent and reduce CVD in virally suppressed HIV sufferers. 217,218 The Lp-PLA2 level is unusually high in HIV-infected patients and is further elevated after ART or protease inhibitorbased treatment. Furthermore, Lp-PLA2 is associated with several risk factors for HIV-associated CVD, such as carotid intima-media thickness (cIMT) and CAC. 219 Among HIV-infected individuals, statin therapy that can reduce the Lp-PLA2 level could lead to a decline in specific markers of immune activation and arterial inflammation, noncalcified plaque volume, and high-risk coronary plaque features. 220 Hence, Lp-PLA2 might be a suitable predictor of subclinical atherosclerosis or even a possible pharmacologic target to prevent CVD in HIV-infected individuals, but comprehensive studies are obligatory to determine the effectiveness of Lp-PLA2 in the clinical setting for HIV-infected patients.
Lp(a) appears to be a causal risk factor for CVD and calcific aortic valve disease (CAVD), 221 and Lp-PLA2 and oxidized phospholipids might be important determinants of the functionality and pathophysiological role of Lp(a). 151 Therefore, targeting the Lp(a)-Lp-PLA2-OxPL axis might be a viable approach for treating CAVD. 222 Chronic periodontitis (CP), which can lead to elevated Lp-PLA2 levels, is most likely associated with atheroma formation and cardiovascular events among CP subjects. 223 In addition, periodontal treatments that In recent years, the role of Lp-PLA2 as a prognostic or diagnostic biomarker has also been evaluated for many diseases. For "biomarker orphan" diseases, such as Rett syndrome and autism spectrum disorders (ASDs), Lp-PLA2, which showed a significant inverse association only with individuals with ASDs compared with healthy controls, was identified as a novel serum biomarker for their differential diagnosis. 229 The increased Lp-PLA2 level in both obstructive sleep apnea (OSA) and obesity might adversely affect endothelial integrity, which decreased after OSA treatment. Thus, plasma Lp-PLA2 activity levels may provide a reliable biomarker for children at risk for underlying atherosclerosis and vascular dysfunction atherosclerosis in the presence of obesity or OSA. 230 Similarly, plasma Lp-PLA2 activity is predominantly upregulated in patients with antiphospholipid antibodies, suggesting a prognostic biomarker role for Lp-PLA2 in managing the antiphospholipid syndrome. 231 In a cohort of rheumatoid arthritis (RA) patients, the baseline Lp-PLA2 mass was significantly associated with both disease severity and subclinical atherosclerosis including IMT and flowmediated dilation at both baseline and follow-up. Consequently, Lp-PLA2 might be a biomarker of vascular damage among patients with RA. 232 In addition, a prospective longitudinal cohort study identified the plasma Lp-PLA2 activity and mass, which were independently and significantly associated with incident abdominal aortic aneurysm (AAA), as markers of AAA risk. 17 Certain relationships between Lp-PLA2 and these diseases were already detected; however, further research is needed for a deeper understanding of the associations. Comprehensive pathophysiological, epidemiological, and genetic studies would be unquestionably helpful for determining the diagnostic and therapeutic utility of Lp-PLA2 in various diseases.

| IDENTIFICATION OF SMALL-MOLECULE INHIBITORS OF LP-PLA2
Synthetic Lp-PLA2 inhibitors have been developed for more than 20 years to explore their potential application in treating various diseases. Since being granted by Human Genome Sciences in 1993, GSK has consistently led the discovery of small-molecule inhibitors of Lp-PLA2, and three compounds have successively progressed to clinical research. Likewise, our group has focused on the identification of novel Lp-PLA2 inhibitors and their role in various disease models as well. Remarkably, utilizing the FBLD approach, both Astex cooperating with GSK and our group have recently discovered novel Lp-PLA2 inhibitors with new skeleton and promising prospects for research.
Interest in the development of Lp-PLA2 inhibitors has since become a topic of intense interest. In addition, many other groups also showed interest in the development of Lp-PLA2 inhibitors and reported numerous compounds with diverse structures in past decades.
PLA2-VIIB, also referred to as PAF-AH II, is a 40 kDa intracellular phospholipase possessing an N-terminal myristoylation site. Because of its 41% sequence homology and conserved catalytic site with Lp-PLA2, PLA2-VIIB shows similar substrate specificity. 11 Although the toxicological consequences of PLA2-VIIB inhibition remain unclear, a preferable Lp-PLA2 inhibitor is predicted to possess over 100-fold selectivity against PLA2-VIIB. 35

| PAF analogs
As PAF can be effectively metabolized by Lp-PLA2, an early study focused on the identification of PAF analogs that bind to Lp-PLA2 and compete with its endogenous substrates. In the general molecular formula, R represents an acetamide or trifluoroacetamide, Z contains a high polar moiety including a quaternary ammonium group, and a XR1 is a long-chain fatty ether (Figure 8). Among the published compounds, 6 showed the best inhibitory activity with an IC 50 value of 0.08 μM for the purified plasma Lp-PLA2, and an IC 50 value of 30 μM for the inhibition of elevated lysoPC during Cu 2+ -induced LDL oxidation. 233  2 IC 50 = 0.08 u uM F I G U R E 8 The structure of PAF analogs. IC 50 values were determined by measuring the rate of turnover of the artificial substrate 1-decanoyl-2-(4-nitrophenylglutaryl) phosphatidylcholine. Assays were performed in 96-well microtitre plates using purified human plasma Lp-PLA2. Lp-PLA2, lipoprotein-associated phospholipase A2; PAF, platelet-activating factor at physiological pH. As a result, 12 exhibited not only a high solubility (>15 mg/mL) in normal saline but also a favorable profile following infusion dosing, including a promising concentration and enzymatic inhibition in the blood and enhanced CNS penetration. 235 Subsequently, to further improve the druggability of these pyrimidine derivatives, the long lipophilic chain of the N-1 substituent was replaced with other inhibitors including piperazine amide Several darapladib analogs have now been reported ( Figure 10). Bayer Company used the triazinone structure to mimic the pyrimidine in darapladib and obtained a series of compounds with good inhibitory in vitro potency. In particular, the IC 50 of compound 14 against rhLp-PLA2 was 0.4 nM. 241 Our group also developed two types of darapladib analogs through bioisosteric replacement of the amide group of darapladib with an imidazole (15) or a triazole (16). Following further optimization, we identified a number of orally bioactive Lp-PLA2 inhibitors in C57BL/6 mice. 242,243 Recent studies reported that a bicyclo[1.1.1]pentane moiety replacement of a phenyl ring in darapladib and rilapladib maintained inhibitory potency and displayed improved physicochemical properties compared to their respective parent compounds. This substitution resulted in improved solubility and an enhanced overall PK profile. 68 However, although darapladib can be used orally, it displayed poor plasma exposure (data from GSK 238  GSK has successively patented a series of novel pyrimidone derivatives as Lp-PLA2 inhibitors. [245][246][247][248][249][250][251][252][253] Compared with darapladib, these newly reported compounds exhibit a simplified structure, small molecular weight, and low lipophilicity (Figure 11), likely resulting in improved PK properties. Among them, GSK2647544 was a potent and specific inhibitor of Lp-PLA2 disclosed in an early patent. 245 In human subjects, GSK2647544 administration was well tolerated for single oral doses up to 750 mg and displayed no significant adverse events. 254 Remarkably, oral single doses of 125 mg GSK2647544 resulted in improved PK performance compared with single doses of 160 mg darapladib (area under curve [AUC] (0-∞) : 2506 ng·h/mL of GSK2647544 vs 645 ng·h/mL of darapladib, ng·h/mL; C max : 457 ng/mL of GSK2647544 vs 18 ng/mL of darapladib, ng/mL). In addition, the inhibition of plasma Lp-PLA2 activity was dose-dependent, and 80 mg twice per day (160 mg daily dose) inhibited greater than or equal to 80% of plasma Lp-PLA2 activity at trough, which is necessary for an enzyme inhibitor to produce a clinical effect. 254 Furthermore, GSK2647544 could likely penetrate into the brains of preclinical species, which was verified by found that a twice-daily dose of 102 mg generated approximately 80% trough inhibition of brain Lp-PLA2 activity at steady state according to an exploratory analysis. 255 Taken together, preclinical and preliminarily clinical studies showed that GSK2647544 acted as a useful tool for determining the effects of brain Lp-PLA2 inhibition. However, another study reported that GSK2647544 might be a moderate to strong CYP3A4 inhibitor, suggesting potential adverse drug-drug interaction. This finding led to the early termination of this study. 254 Compounds 19 to 22 are specialized pyrimidine inhibitors with one or two fused heterocycles. 247 253 Collectively, both the early and later pyrimidone derivatives mentioned above possessed a more than 100-fold selectivity for Lp-PLA2 over PLA2-VIIB. Additionally, the later pyrimidone derivatives released by GSK focused on the identification of novel Lp-PLA2 inhibitors harboring rational physicochemical properties and desirable PK performance.

| Pyrimidone derivatives
In the past few years, our group reported two series of novel Lp-PLA2 inhibitors based on the pyrimidone scaffold, with a favorable selective inhibition toward Lp-PLA2. Previous studies revealed that GSK2647544 had an improved PK profile compared with darapladib in human subjects. 254 Indeed, we detected greater plasma exposure of GSK2647544 than that of darapladib when dosed orally to SD rats (GSK2647544: C max , 0.78 μg/mL; AUC (0-∞) , 3.38 μg·h/mL at 25 mg/ kg and darapladib: C max , 0.15 μg/mL; AUC (0-∞) , 2.11 μg·h/mL at 50 mg/kg). However, GSK2647544 showed a limited elimination half-life in vivo (t 1/2 = 2.08 hours) and poor metabolic stability (CL int = 74.4 ml/min/kg) in further evaluation with rat liver S9 fractions. Therefore, we utilized a structural restriction strategy that might contribute to greater metabolic stability to identify a new imidazo[1,2-a]pyrimidine scaffold as an Lp-PLA2 inhibitor ( Figure 12). Subsequently, systematic structure-activity relationship (SAR) explorations generated a number of Lp-PLA2 inhibitors with high activity.
Compound 26 was selected for further studies of the PK profile and inhibitory activity in vivo. As a result, a longer halflife (t 1/2 = 13.34 hours) was revealed, which was consistent with good metabolic stability in rat liver S9 fractions and a higher C max (2.6 μg/mL) and AUC (0-∞) (78.1 μg·h/mL). In addition, at a single oral dose of 25 mg/kg, compound 26 could generate over 50% inhibition even 24 hours after administration. The overall inhibition was also comparable to 50 mg/kg F I G U R E 1 2 The discovery of a series of imidazo[1,2-a]pyrimidine compounds as Lp-PLA2 inhibitors. The IC 50 values were determined by measuring the rate of turnover of the substrate 2-thio-PAF. Assays were performed in 96-well microtitre plates using recombinant human Lp-PLA2. Lp-PLA2, lipoprotein-associated phospholipase A2; PAF, platelet-activating factor; SAR, structure-activity relationship darapladib and higher than that for 25 mg/kg GSK2647544. Taken together, these results suggest that the imidazo [1,2-a] pyrimidine derivatives harbor reasonable physicochemical properties, and are potent, orally bioavailable Lp-PLA2 inhibitors. 244 To obtain novel Lp-PLA2 inhibitors with simplified structures, we also made changes to the pyrimidone scaffold, focusing on the right-hand side of the structure. Unlike the compounds reported by GSK, we unfolded the fused heterocycle to determine the resulting effects on PK and PD properties. Through three rounds of intensive SAR studies, we obtained numerous inhibitors with high potency against rhLp-PLA2. Gratifyingly, compound 24 showed decent PK profiles and demonstrated robust inhibitory potency against plasma Lp-PLA2 in male SD rats after oral administration. Even a relatively low dose of 3 mg/kg led to a greater degree of Lp-PLA2 inhibition than that of a single oral dose of 25 mg/kg darapladib ( Figure 11). Furthermore, 24 significantly inhibited retinal thickening in STZ-induced diabetic SD rats, a model of DR, after oral dosing for 4 weeks, while darapladib showed similar efficacy at the same oral dose of 25 mg/kg/day. These results further confirm that Lp-PLA2 is an efficacious therapeutic target for DR. 194

| Biaryl inhibitors
In primary in vitro assays (fluorescence intensity assays), darapladib showed high activity against rhLp-PLA2 with an IC 50 = 0.049 nM. However, the IC 50 value dramatically declined to 35 nM when assessing the inhibitory potency of darapladib in whole human plasma. To minimize the divergence from the Lp-PLA2 biochemical and plasma assays, Astex Pharmaceuticals in cooperation with GSK used the FBLD strategy to obtain a potent inhibitor with improved physicochemical properties and speculated that the drop off between the biochemical and plasma assays might result from the high lipophilicity of darapladib. Initially, screening a core fragment set by an Astex Pyramid fragment screen generated 16 X-ray validated hits. A subset of fragment hits, such as bis-aryl 27, were shown to bind to a novel pocket generated by a rotation of the Phe357 side chain, illustrated by an overlay of the binding modes of fragment hits 27 and 28 ( Figure 13B). A novel inhibitor that did not interact with the oxyanion hole generated improved physicochemical properties. Together with the suitable growth vector of 27 to an area of the protein where polar interactions were observed in other fragment hits ( Figure 13A), the researchers selected 27 as a starting hit. To occupy the binding site of the amino group in amide 28, compound 29, with an aminothiazole moiety, was selected for further assessment, primarily due to the superior growth vector from C-4 and the easier replication of the conformation with nonsulfur-containing heterocycles ( Figure 13C). Under the guidance of the crystal structures of inhibitors bound to Lp-PLA2, they designed a hydroxypropyl group to occupy a narrow channel formed by Leu153 and Gln352 (compound 30, Figure 13D), a methoxy group from C-2 of the n-propyl to fill the small pocket formed by the side chain of Ala355 (compound 31), and an aminoethoxy moiety from position 6 of the central aryl ring that points toward the solvent to modulate the physicochemical properties and reduce the lipophilicity of the template (compound 5, Figure 13E). Compound 5 was able to maintain potency and selectivity for Lp-PLA2, simultaneously with desirable physicochemical properties (MW: 411, Chrom-LogD pH 7.4 : 3.4, and solubility: 302 μM). In addition, 5 exhibited a lower drop off between the Lp-PLA2 biochemical (IC 50 = 0.0014 μM) and plasma assays (IC 50 = 0.014 μM) relative to darapladib. Notably, these compounds were the first potent Lp-PLA2 inhibitors that did not directly interact with catalytic residues. However, the work stalled, as neither 31 nor 5 showed satisfactory PK performance consistent with once-daily dosing in humans. Nonetheless, the identification of novel Lp-PLA2 inhibitors with improved physicochemical properties and new binding pockets highlighted the success of FBLD in lead identification. 35

| Lactam inhibitors
The discovery of biaryl inhibitors suggests that fragment-based technology could accomplish tighter control of the physicochemical properties of lead compounds that are supportive of further advancements. To obtain novel Lp-PLA2 inhibitors with improved in vivo parameters during the process of optimization against plasma potency, HUANG ET AL. Utilizing a suitable substituent that was located in the groove above Trp298 facing the solvent, they identified an alternative low molecular weight, potent, selective inhibitor of Lp-PLA2 termed compound 3. The compound showed comparable inhibitory potency to darapladib in whole human plasma assays (3, IC 50 = 32 nM and darapladib, IC 50 = 35 nM) and the drop off between the Lp-PLA2 assay and plasma assay was small. Furthermore, PK and safety evaluations demonstrated that compound 3 provided an attractive PK profile and excellent physicochemical properties, suggesting a promising starting point for further lead optimization. 34

| Sulfonamide inhibitors
As previously mentioned, our group disclosed the crystal structures of Lp-PLA2 in complex with two reversible inhibitors, darapladib and compound 1, and shed light on the binding mechanisms of different inhibitors with Lp-PLA2, providing an important prerequisite for the accomplishment of a successful FBLD project. 32 To enrich the structural types of Lp-PLA2 inhibitors, we recently reported the discovery of novel sulfonamide inhibitors from a weak binder to a series of nanomolar inhibitors. First, we screened our in-house 500-fragment library in PAF enzymatic assays and selected sulfonamide fragment 37 as a promising starting point, which efficiently bound to the oxyanion hole of Lp-PLA2 ( Figure 15A). Second, similarity searching on this fragment followed by molecular docking led to the discovery of a micromolar inhibitor 38 with a 300-fold potency improvement (37, IC 50 =~1 mM and 38, IC 50 = 34 μM). Superimposing the X-ray crystal structures of 37 and 38 bound with Lp-PLA2 showed that compound 38 conserved almost all the aforementioned protein-ligand interactions of 37, suggesting a promising hit ( Figure 15B). Third, we conducted three rounds of SAR exploration on hit 38, including filling a subpocket with a cyano group, replacing the naphthyl group that weakly bound to the enzyme The arrows in each structure indicate that the suitable growth vector of 27 (blue circle), the rotated location of F357 compared with that in the structure of darapladib, and the superior growth vector from C-4 of thiazole group (red circle). The pocket of compound 30 is shown with gray surface to clarify that C6 of biaryl moiety (purple circle) is facing to the solvent and C2 of n-propyl group (yellow circle) contains an unoccupied pocket. The crystal structure of compound 5 with Lp-PLA2 is shown in cartoon model. Interacting residues of compound 5 are shown in lines, and inhibitors are rendered in green sticks. The hydrogen bonds are labeled with yellow dotted lines. The figure was prepared using the program PyMOL. The IC 50 value of compound 28 was determined by measuring the rate of turnover of the substrate PED6, while other compounds were 2-thio-PAF. Assays were performed in 96-well microtitre plates using recombinant human Lp-PLA2. FBLD, fragment-based lead discovery; Lp-PLA2, lipoprotein-associated phospholipase A2; PAF, platelet-activating factor; PED6, N-((6-(2,4-dinitrophenyl)amino)hexanoyl)-2-(4,4-difluoro-5,7dimethyl-4-bora-3a,4a-diaza-s-indacene-3-pentanoyl)-1-hexadecanoyl-sn-glycero-3-phosphoethanolamine, triethylammonium salt [Color figure can be viewed at wileyonlinelibrary.com] with a phenyl group containing both chlorine and trifluoromethyl substitutes, and exploring the hydrophilic substitutions on the solvent-exposed benzene ring. Ultimately, we obtained compound 4 with an IC 50 = 5 nM, indicating approximately 200 000-fold improved affinity from the initial sulfonamide fragment. Subsequently, through preliminary PK assessment in vitro, we identified another sulfonamide inhibitor, 39, with high inhibitory activity (IC 50 = 14 nM), decent stability and good permeability. Furthermore, 39 had favorable oral bioavailability in male SD rats (F = 35%) and maintained inhibitory activity for 24 hours post oral administration, which was superior to that of darapladib (F = 11%). 33

| Covalent inhibitors
Previously, several OP compounds were confirmed by X-ray crystallography and were able to inhibit Lp-PLA2 in a covalent manner, including paraoxon, sarin, soman, and tabun, which may be helpful in developing catalytic bioscavengers for OP nerve agents. 64,65 In addition, other chemicals may covalently bind to Lp-PLA2 to inhibit its physiological functions ( Figure 16). SB-253514 is a natural product isolated from the culture broths of Pseudomonas fluorescens strain DSM11579 and a potent inhibitor of Lp-PLA2. SB-253514 could form a 1:1 covalent complex with Lp-PLA2 confirmed by liquid chromatography-MS experiments, and further kinetic analysis indicated very slow reactivation (t 1/2 > 24 hours). In addition, SB-253514 showed good potency against Lp-PLA2 in isolated rabbit and human plasma and favorable selectivity over other protease enzymes except for PLA2-VIIB. Furthermore, SB-253514 significantly inhibited plasma Lp-PLA2 in WHHL rabbits when dosed intravenously, but no inhibition related to oral administration was observed. 256 On the basis of the SB-253514, several cyclic enol-carbamates were synthesized as potent Lp-PLA2 inhibitors, but no further information was disclosed. 257  manner and was first identified as an activity-based probe for another serine hydrolase DAGLβ. In addition to the inhibition of mouse Lp-PLA2 with an IC 50 = 39 nM, HT-01 could serve as a tailored activity-based probe for the detection of this enzyme in native proteomes according to gel-based activity-based protein profiling assays. 264 Recently, Abide Therapeutics disclosed a patent for a series of dual inhibitors of both Lp-PLA2 and α/β-hydrolase domain containing 6 (ABHD6). As an example, compound 41 possessed a carbamate moiety and could efficiently inhibit 50% of Lp-PLA2 and ABHD6 at concentrations lower than 100 nM. 265 However, although these noncanonical carbamates may provide a covalent resource to Lp-PLA2, their actual binding modes require verification in future studies.

| Other inhibitors
Compound 42 is natural fatty acid glycerol isolated from Saururus chinensis roots that showed narrow inhibitory activity against Lp-PLA2 with an IC 50 value of 45 μM (Figure 16). The amides of xanthurenic acid were other specific scaffolds of Lp-PLA2 inhibitors. 266 268 Furthermore, the results also highlighted quinolone-metal chelators as potential enzyme inhibitors.
To summarize, although a number of Lp-PLA2 inhibitors can be used at present, the aforementioned compounds have been used sparingly with the exception of pyrimidinone derivatives. Seemingly, the major factor that limits further exploration is the difficulty in balancing the inhibitory potency and physicochemical properties.
As mentioned above, Lp-PLA2 is a hydrophobic protein, and its hydrophobic interaction mainly contributes to the affinity between the ligand and protein.  271 Due to its ability to exhibit phospholipase A2 characteristics and circulate in the blood via binding to LDL and HDL, pPAF-AH is also called Lp-PLA2. Subsequently, Stafforini et al made great efforts to determine the biological properties and functions of Lp-PLA2. Since those studies, the effects of Lp-PLA2 on inflammatory processes have been widely studied, mainly due to the ability of inflammatory cells, especially macrophages, to secrete the majority of circulating Lp-PLA2, particularly, during the differentiation of monocytes to macrophages, indicating the occurrence of inflammation. 24,25,60 Coupled with a high distribution of Lp-PLA2 in the blood, the association between Lp-PLA2 and vascular inflammatory-related diseases, such as atherosclerosis, DR, AD, and HUANG ET AL.

| 119
certain cancers, [19][20][21][22]195 has been evaluated. In this context, several compounds with diverse scaffolds have been identified as potent Lp-PLA2 inhibitors, providing several excellent probe molecules. In addition, benefiting from the FBLD strategy, recently reported Lp-PLA2 inhibitors exhibit favorable physicochemical properties and PK profiles. 34,35 In terms of their clinical applications, although Lp-PLA2 inhibitors failed to reduce the incidence of CVD events, they indeed showed significant therapeutic effects on microvascular diseases such as DR and AD.
Nonetheless, we could not ignore the neutral and negative reports regarding Lp-PLA2 inhibition, which may be attributed to its anti-inflammatory roles in some conditions. To date, the effects of Lp-PLA2 on various diseases remain inconsistent. Therefore, to motivate further research within the Lp-PLA2 field, we would like to offer several personal insights into the achievements from the past few decades in the following text.

| Genetic studies
GSK has heavily invested in the development of Lp-PLA2 inhibitors, particularly, into two unsuccessful phase III trials. Several hundreds of million dollars were spent and over 30 000 patients were recruited and followed up for a number of years. These failures not only dimmed the future of the anti-inflammatory strategy in atherosclerosis drug research, but also led to significant financial losses to the company. Genetic studies, particularly, those of rare loss-of-function variants, have been considered a promising approach to promote the development of targeted drugs by validating or invalidating novel therapeutic targets. Recently, two large-scale genetic studies established that PLA2G7 variants that reduced Lp-PLA2 activity to levels comparable to darapladib had no effect on the risk of CHD and outcomes. These results deny, to a certain extent, a causal role of Lp-PLA2 during atherosclerotic progression and are consistent with an early genetic study on PLA2G7 variants that resulted in a modest reduction in Lp-PLA2 activity. 77,89,114 Therefore, we speculate that a comprehensive genetic study before the clinical trial may be conducive to the identification of the fundamental disease pathways. In addition, genetic studies contribute to differentiating between responding and nonresponding patients. Very recently, a pharmacogenetic meta-analysis identified a potentially efficacious subgroup within the large global trials of darapladib. Among them, one common variant (rs181937009) conferred risk in the placebo arm but reduced the risk in the darapladib arm, suggesting of an improved outcome with the minor allele following treatment with darapladib compared with placebo. 38 Similarly, another SNP of PLA2G7 (rs1805018) was reported to affect plasma TG levels in response to supplementation with fish oil-derived long-chain n-3 PUFAs, signifying the interindividual variability in plasma TG levels following n-3 PUFA supplementation. 122 Furthermore, ischemic stroke patients carrying the rs1051931 and rs7756935 variants (AA-CC haplotype) were found to have a considerably higher risk of aspirin resistance than noncarriers (OR = 8.23, 95% CI: 1.59-42.63). 121 Accordingly, through the determination of the consequences of mutations, particularly, loss-of-function variants, within potential therapeutic target genes by systematic and largescale human genetic studies, predicting the potential efficacy and safety of the compounds aimed at these targets is possible. Since impressive advances in DNA analysis technology aid the implementation of systematic and largescale genetic studies, we need to take advantage of the data mining opportunities generated from the clinical trials to make full sense of the consequences of Lp-PLA2 inhibition.

| Nominal "safety window"
Divergence seems to be ubiquitous among Lp-PLA2 studies, including the results of epidemiologic studies, the impacts of genetic mutations, and the role of Lp-PLA2 in inflammation and disease progression. The causes of divergence are manifold, including complex experimental systems, different objects under study, and inconsistent methods of measurement and analysis. Nevertheless, we speculate that the contradiction of Lp-PLA2 in modulating inflammation might be a key causative factor for these divergences. Both the substrates (PAF and oxLDL) and products (lysoPC) of Lp-PLA2 hydrolytic action were demonstrated to be proinflammatory mediators. 13,148 Consequently, the accumulation of any of these mediators resulting from a fluctuation in Lp-PLA2 activity could theoretically promote the inflammation process. Apparently, Lp-PLA2 expression should lie within a rational "window" to maintain body homeostasis. Indeed, Stafforini 60 proposed an "expression window" of Lp-PLA2 that is a key determining factor in response to endotoxemia, based on the dynamics of Lp-PLA2 in patients most with severe sepsis and other critical illnesses. In this case, the survivors could maintain Lp-PLA2 activity within a range comparable to that of healthy controls, while the nonsurvivors displayed significant deviation from this "expression window" in either direction. 272 In fact, we agree with this concept to a degree, as we believe in the existence of the top edge of such a window but not the bottom edge. We hypothesize that extended exposure to a high level of Lp-PLA2 activity may result in the development of various diseases, while inadequate or even absent Lp-PLA2 activity may not be detrimental. The primary basis for our hypothesis might be that Lp-PLA2 inhibition showed limited impacts on the accumulation of PAF and oxLDL but significantly decreased the lysoPC level, suggesting a dominant proinflammatory effect of Lp-PLA2. 50,51,273 Furthermore, the loss-of-function variants of Lp-PLA2 were confirmed not to be associated with certain diseases in large-scale studies. 76,77 Moreover, in the clinical trials, serious adverse effects associated with intensive Lp-PLA2 inhibition by darapladib or rilapladib were similar to those in the placebo group, indicating the absence of target-specific toxicity. [19][20][21][22] Altogether, we attribute the detrimental consequences to an elevated Lp-PLA2 level. Since the direct receptor-mediated signaling events of Lp-PLA2 were not demonstrated, its biological effects are tightly associated with the upstream and downstream molecules. In this regard, more attention should be paid to Lp-PLA2-derived lysoPC, as it accumulates with an increase in Lp-PLA2.
Furthermore, lysoPC, particularly the 16:0 lysoPC, a main product of Lp-PLA2, was associated with several diseases, including NASH, 228 DR, 37 CAVD, 274 and CNS diseases. 14 Therefore, we assume that alterations in the lysoPC level, the ratio between Lp-PLA2 and lysoPC, or even the proportion among Lp-PLA2, lysoPC, and oxLDL may be an alternative monitoring metric for determining the role of Lp-PLA2 in various diseases. In addition, we anticipate that the multifactor evaluation can diminish the occurrence of inconsistent results among studies.

| Biological functions
Although certain progress has been achieved, the biological functions of Lp-PLA2 require further investigation due to conflicting results being reported. Moreover, the scope of Lp-PLA2 studies should be expanded from the blood to other potential secretory tissues. Recently, Jackisch et al 41 revealed that human adipose tissue appeared to be an active source of Lp-PLA2 and was differentially regulated by fat depot and metabolic status. For example, the abdominal subcutaneous but not the omental, adipose tissue displayed enhanced Lp-PLA2 expression, and type 2 diabetes status contributed to the upregulation of Lp-PLA2 in adipose tissue. Owing to its association with inflammation and atherosclerotic risk, Lp-PLA2 action within adipocytes may represent a potential therapeutic target to prevent the development of cardiometabolic complications in type 2 diabetes. 41 On the basis of this study, the perivascular adipose tissue (PVAT), which surrounds most large blood vessels, is probably a potential source of Lp-PLA2 as well. Moreover, the role of PVAT in aortic aneurysms and vasculitic syndromes was demonstrated.
Therefore, determining the biological functions of adipose-derived Lp-PLA2 is essential. 275,276 In addition, the level of Lp-PLA2 derived from the brain tissue remains unclear. Earlier, Cao et al 123 determined that Lp-PLA2 mRNA was expressed in all regions of the brain, 123 while a recent study failed to detect Lp-PLA2 expression within autopsy brain tissue through histopathologic analyses. 184 Moreover, rilapladib is not considered to be a brain penetrant based on the preclinical data of GSK and that of our group but shows therapeutic effects on AD, suggesting a mediating mechanism by peripheral tissue. 21 Therefore, whether brain tissue can secrete Lp-PLA2 and what the biological functions of brain-derived Lp-PLA2 are remain to be established. In addition to the source, the pathophysiological role and clinical significance of Lp-PLA2 bound to LDL and HDL should be determined. HDLassociated Lp-PLA2 is highly acknowledged to be anti-inflammatory, while LDL-associated Lp-PLA2 is proinflammatory; however, the underlying mechanisms have yet to be elucidated. Alterations in the distribution of Lp-PLA2 between LDL and HDL were detected in patients with various diseases, including atherosclerosis, dyslipidemias, 12 DM, 277 and PCOS. 278 However, the clinical consequences of these alterations were inadequately HUANG ET AL.

| 121
determined. 12 Furthermore, no data are available regarding the effects of selectively inhibiting LDL-or HDLassociated Lp-PLA2, as almost all published Lp-PLA2 inhibitors focus on inhibition of total plasma Lp-PLA2 activity, which primarily concerns LDL-associated Lp-PLA2. Furthermore, the pathological role of Lp(a) has gradually attracted more attention, especially in CVD. Since Lp-PLA2 was identified as an important determinant of Lp(a) functionality and pathophysiology, the potential of Lp(a)-associated Lp-PLA2 as a therapeutic target needs to be elucidated. 12,221,222 6.4 | Inhibitor discovery The discovery of inhibitors of a hydrophobic protein is challenging, as it is difficult to balance the potency and physicochemical properties. The FBLD strategy assists us in addressing this issue. 69 The successful application of FBLD to the identification of novel Lp-PLA2 inhibitors provides a practical example for the discovery of drugs for other targets. [33][34][35] Furthermore, the possibility of developing LDL-Lp-PLA2-or HDL-Lp-PLA2-specific inhibitors using the FBLD strategy or other emerging technologies are worthy of investigation. Additionally, to determine the efficacy accurately, various diseases may require the tested compound to have specific properties; thus extending the structural types of Lp-PLA2 inhibitors is necessary.
In summary, Lp-PLA2-targeted drug development has always been full of controversy from the beginning, due to the rather complicated functions of the enzyme and its considerable involvement in a wide variety of diseases.
The "door" to atherosclerosis was almost shut after the failure of darapladib in phase III clinical trials, while a "window" seemed to be opened for microvascular diseases. Clinical and preclinical studies not only provided basic evidence for the efficacy of Lp-PLA2 inhibitors in the treatment of AD and DME but also elucidated specific mechanisms of action that were independent of the most studied risk factors, such as Aβ and VEGF. In view of the high attrition rate of Aβ-based anti-AD drugs and the poor compliance of DME patients for anti-VEGF biologics, Lp-PLA2 inhibitors can yet be regarded as a potential good choice. With a series of eligible compounds available at present, scientists would be admired if they exerted more effort in exploring the role of Lp-PLA2 in diseases other than atherosclerosis, particularly, in microvascular diseases that may be more sensitive to Lp-PLA2 inhibition. In addition, the large pool of data generated from clinical trials provides numerous opportunities for comprehensive data mining, which may be beneficial to understanding the biological consequences of Lp-PLA2 inhibition and help us to learn a profound and valuable lesson from the failure of darapladib. [279][280][281][282][283][284][285][286][287][288] To conclude, we anticipate that the story of Lp-PLA2 will continue.