Active site competition is the mechanism for the inhibition of lipoprotein-associated phospholipase A2 by detergent micelles or lipoproteins and for the efficacy reduction of darapladib

Lipoprotein associated phospholipase A2 (Lp-PLA2) has been characterized for its interfacial activation as well as inhibition by detergent micelles and lipoprotein particles. The enzyme has been shown to bind on the surfaces of hydrophobic aggregates, such as detergent micelles, lipoprotein particles and even polystyrene latex nanobeads. Binding to hydrophobic aggregates stimulates the activity of Lp-PLA2 but may not be the necessary step for catalysis. However, at higher concentrations, detergent micelles, latex nanobeads or lipoprotein particles inhibit Lp-PLA2 possibly by blocking the access of substrates to the active site. The competition mechanism also blocks inhibitors such as darapladib binding to Lp-PLA2 and reduces the efficacy of the drug. Darapladib has very low solubility and mainly exists in solutions as complexes with detergents or lipoprotein particles. The inhibition of Lp-PLA2 by darapladib is dependent on many factors such as concentrations of detergents or lipoproteins, incubation time, as well as the order of mixing reaction components. The in vitro Lp-PLA2 activity assays used in clinical studies may not accurately reflect the residual Lp-PLA2 activity in vivo. Darapladib has been found mainly bound on HDL and albumin when it is incubated with human serum. However, Lp-PLA2 is more sensitive to darapladib when bound on LDL and relatively resistant to darapladib when bound on HDL. Therefore, high cholesterol levels may decrease the efficacy of darapladip and cause the drug to be less effective in high risk patients. Our study will help to design better inhibitors for Lp-PLA2. The discoveries also contribute to understanding the mechanism of interfacial activation and inhibition for Lp-PLA2 and provide a new concept for researchers in building better kinetic model for interfacial enzymes.


Effects of detergent micelles, latex beads and lipoproteins on the activation and inhibition of recombinant Lp-PLA 2 .
To investigate the effects of detergents on the activity of Lp-PLA 2 , recombinant Lp-PLA 2 was incubated with low concentrations (10 µM or lower) of artificial substrate, 1-myristoyl-2-(4nitrophenylsuccinyl)-sn-glycero-3-phosphocholine (14:0 NPS-PC), at pH 7.4. Figure 1a show the activation and inhibition of Lp-PLA 2 by four detergents with different critical micelle concentration (CMC). Like other lipases, Lp-PLA 2 has very low activity towards water soluble substrates in the absence of a detergent. When titrated with a detergent, starting around the CMC of each detergent, the activity of the recombinant Lp-PLA 2 was suddenly enhanced and then gradually decreased back to the baseline with increasing concentration of the detergent. The activation and subsequent inhibition of recombinant Lp-PLA 2 activity resulted in the formation of a peak in the titration curves of each detergent. The formation of the peak correlated to the CMC of each detergent. Inhibition of Lp-PLA 2 activity was also observed in reactions with higher concentrations (0.54 mM) of the substrate, 14:0 NPS-PC, for detergents with lower CMC such as Tween-20, digitonin and Triton X-100 but not for detergents with higher CMC such as CHAPS, deoxycholate and SNS (Fig. 1b). Studying the inhibition of Lp-PLA 2 by Tween-20 revealed that it was competitive between the substrate and detergent (Fig. 1c). The Michaelis-Menten kinetic parameters of the hydrolysis reaction of 14:0 NPS-PC by Lp-PLA 2 were also compared between Scientific Reports | (2020) 10:17232 | https://doi.org/10.1038/s41598-020-74236-0 www.nature.com/scientificreports/ the presence and absence of 10 mM CHAPS ( Table 1). The difference between two V max values was within the experimental error, but the difference between two K m values was about three times of the experimental error. Therefore, K m was affected in the presence of 10 mM CHAPS whilst the V max was constant. Detergents were also found to affect the efficacy of darapladib on the inhibition of Lp-PLA 2 . The inhibitory effects of darapladib were significantly reduced when the concentration of Tween-20 was increased tenfold (Fig. 1d). The IC 50 of darapladib was about 10 nM in 80 µM Tween-20 and 80 nM in 800 µM Tween-20.
To further investigate the nature of the activation and inhibition of Lp-PLA 2 under the condition of low substrate concentration, polystyrene latex nanobeads were used to mimic the roles of detergent micelles. Figure 1e shows that latex nanobeads can also activate and inhibit Lp-PLA 2 Fig. 1f. HDL and LDL were both found to activate at lower concentration and inhibit at higher concentration for recombinant Lp-PLA 2 (Fig. 1f). Kinetic difference for the hydrolysis of 14:0 NPS-PC by Lp-PLA 2 in the presence and absence of detergents. To further explore the kinetic difference of Lp-PLA 2 in the presence and absence of detergents, the hydrolysis reaction was carried out with and without 10 mM CHAPS (Fig. 3a). In the presence of 10 mM CHAPS, the reaction obeyed Michaelis-Menten kinetics but it did not do so in the absence of 10 mM CHAPS. Under these conditions, the Hill slope of the substrate titration curve was approximately 1.0 in the presence of 10 mM CHAPS and approximately 2.0 in the absence of 10 mM CHAPS. The Hill slope values were dependent on the concentration range of the substrate in the reaction. The lower of the concentration of substrate, the higher was the Hill slope values (Fig. 3b). The Hill slope decreased with the increase of substrate concentration and approached 1.0 at the substrate concentration approximately 0.5-1 mM. However, in the presence of 10 mM CHAPS, the Hill slope of substrate titration was kept at 1.0 and independent of the substrate concentration (Fig. 3b). The CMC of 14:0 NPS-PC was determined by the method of polarized fluorescence 21 using fluorescein as the probe in TBS, pH 7.4. When the concentration of lipid or detergent approaches its CMC, the fluorescent anisotropy ratio will increase due to the formation of micelles carrying fluorescein. Figure 3c shows that the CMC of 14:0 NPS-PC is about 0.5 -1 mM and the obtained CMC of CHAPS (control) by the same method is about 5-10 mM, which is in agreement with the literature. Interestingly, the determined CMC value of 14:0 NPS-PC is close to the substrate concentration (0.5-1 mM, Fig. 1b) when the Hill Slope approaches 1.0 and much higher than the substrate concentration when the reaction rate starts to increase along the sigmoidal curve (Fig. 3a), which is around 20-30 µM (log1.2-log1.5).

Effects of detergent micelles on the activation and inhibition of lipoprotein-bound
Inhibition of lipoprotein-bound Lp-PLA 2 by lipoprotein particles. The relationship between substrate and lipoprotein-bound Lp-PLA 2 was investigated by titration of lipoprotein under limiting substrate concentration (10 µM 14:0 NPS-PC). Lp-PLA 2 activity increased with lipoprotein concentration at the initial phase due to the increase of Lp-PLA 2 concentration and achieved a plateau value of at 0.008 mg/ml for HDL and 0.06 mg/ml for LDL respectively (Fig. 4a). Inhibition of Lp-PLA 2 bound on lipoprotein particles by darapladib. The inhibitory effects of darapladib over Lp-PLA 2 were analyzed under different conditions for LDL and HDL separately. Inhibition of lipases can be complicated and depends on how the inhibitors are delivered 22 . Figure 5 shows the responses of Lp-PLA 2 bound on LDL and HDL in the presence of darapladib under different experimental conditions. When LDL and HDL were assayed for Lp-PLA 2 activity by adding the mixture of 14:0 NPS-PC (final concentration of 10 µM) and varied concentrations of darapladib to start the reaction, only partial inhibition of the enzyme was achieved at the highest concentration of darapladib and the IC 50 was about 0.5 and 1 nM for LDL and HDL respectively (Fig. 5a,b). If preincubating the LDL and HDL with darapladib at ambient temperature for 30 min before adding 14:0 NPS-PC to start the reaction, the IC 50 was about 10 nM for both LDL and HDL. When 40 mM of CHAPS (final concentration in the assay mixture was 20 mM) were included in the preincubation mixture of darapladib and LDL or HDL, the IC 50 was about 1 nM for both LDL and HDL. Also, higher activity and more complete www.nature.com/scientificreports/ inhibition of Lp-PLA 2 were observed in the presence of CHAPS. For HDL, the associated Lp-PLA 2 activity is relatively less sensitive to the inhibition by darapladib in the absence of CHAPS. The IC 50 of darapladib for both LDL-and HDL-associated Lp-PLA 2 was dependent on the concentration of LDL and HDL ( Fig. 5c-e). When lipoprotein concentration was reduced 100-fold, Lp-PLA 2 activity in LDL became much more sensitive to darapladib and was completely suppressed at less than 0.1 nM of darapladib. In the presence of 20 mM CHAPS, the IC 50 of darapladib remained the same for both concentration levels of LDL and HDL (Fig. 5e). On the other hand, Lp-PLA 2 activity associated with HDL was less sensitive to the decrease of lipoprotein concentration especially under the conditions where the lipoproteins were preincubated with darapladib ( Fig. 5c,d). Such difference disappeared when CHAPS was included in the incubation mixture and the Lp-PLA 2 activity in both LDL and HDL decreased significantly (Fig. 5e). The IC 50 of darapladib for Lp-PLA 2 was the same for both LDL and HDL and not affected by lipoprotein concentration in the presence of CHAPS. The resistance of HDL-bound Lp-PLA 2 to darapladib was further confirmed by excessive inhibition of the enzyme in human serum with 150 µM of darapladib. After incubation with high concentration of darapladib and removal of the bulk content by size exclude chromatography (SEC) using a TSKgel G3000SW XL (7.8 mm × 30 cm, 5 µm) in TBS, pH 7.4, the fractions of human serum were assayed for residual Lp-PLA 2 activity. The column was calibrated with purified HDL and LDL in TBS, pH 7.4. LDL was eluted in fraction 7 and HDL was eluted in fractions 8-10 when 40 µl of HDL or LDL at the concentration of 1 mg/ml were injected for resolution (Fig. 5f). When the same volume of human serum was resolved by the same column under the same conditions, majority of Lp-PLA 2 activity was fractionated in fraction 7 (Fig. 5g, blue). The distribution pattern did not change when the serum was incubated with 150 µM darapladib for 1 h at ambient temperature and fractionated (Fig. 5g, red). When the same serumdarapladib mixture was continued to incubate for 24 h at ambient temperature before fractionation, Lp-PLA 2 activity in fraction 7 (LDL) disappeared completely indicating that the inhibitor was tightly bound. However, the Lp-PLA 2 activity in the HDL fractions of the same mixture was intact and seemingly increased slightly (Fig. 5g, green). The same results were obtained when the experiment was repeated multiple times. www.nature.com/scientificreports/ buffered saline (PBS). The retention shifting of apolipoprotein A1 (ApoA1), apolipoprotein B (ApoB) and Lp-PLA 2 were analyzed by ELISA and enzyme activity (Fig. 6). Consistent with earlier studies 11,12 , the majority of Lp-PLA 2 mass (yellow curve) or activity (green curve) was associated with ApoB (LDL, red curve) and eluted in fraction 8-14 (Fig. 6a). A small portion of the enzyme was associated with ApoA1 (blue curve) and eluted in fraction 15-19 (larger size HDL, Fig. 6a). No bulk phase distribution of Lp-PLA 2 activity or mass was found. When including 10 mM CHAPS in the fractionation buffer, however, all of the Lp-PLA 2 mass and activity was dissociated from LDL and HDL (Fig. 6b). The peak fraction of ApoB was shifted from fraction 11 to fraction 9 in the presence of 10 mM CHAPS while no significant changes were observed for the elution pattern of ApoA1 under the same conditions. Similar results were obtained when serum samples were analyzed by non-denatured gel electrophoresis for ApoA-1 and ApoB-100 (Fig. 6c). Mixing serum with 10 mM CHAPS resulted in the aggregation of ApoB but less damage of HDL was observed except for some de-lipidation of ApoA1.  Figure 7 shows that Lp-PLA 2 loaded on LDL can be transferred to HDL during mixing of the two lipoproteins (Lane HDL-0/LDL-1 and HDL-0/LDL-2 in Fig. 7). However, very low levels of Lp-PLA 2 loaded on HDL can be transferred to LDL during the mixing (Lane HDL-1/LDL-0 and HDL-2/LDL-0 in Fig. 7).  Figure 8b shows the distribution of darapladib in TBS, pH 7.4, in the absence and presence of detergents. Since the concentration of darapladib was low (0.2 µM) and not easy to monitor, fractions were collected and assayed for its ability to inhibit spiked Lp-PLA 2 . The results indicated that, in the absence of detergents, darapladib at 0.2 µM was not detectable in TBS buffer, possibly due to the low solubility (Fig. 8b, blue). The compound was solubilized in the presence of either 10 mM CHAPS (Fig. 8b, red curve) or 0.6 mM Tween-20 (Fig. 8b, green curve). However, it eluted in very late fractions possibly due to the interaction of darapladib-detergent complex with the column packing materials. When mixtures of darapladib with HDL or LDL were fractionated, inhibition of spiked Lp-PLA 2 was observed with the fractions of the lipoprotein correspondingly (Fig. 8c). Darapladib only associated with LDL, HDL and albumin and was not present in bulk phase under the experimental conditions. Figure 8d shows the distribution of darapladib in human serum and further confirms the association of darapladib with HDL and albumin. Darapladib was mixed with human serum at the final concentration of 150 µM and incubated at ambient temperature as indicated. Forty µL of the mixture were subjected to fractionation by the same TSKgel G3000SW XL column. Inhibition of spiked Lp-PLA 2 was only observed with the HDL and albumin fractions (Fig. 8d, green curve). Again, no bulk distribution for darapladib was found. Also, darapladib inhibition of spiked Lp-PLA 2 was not observed in the LDL fraction (fraction 7, green curve in Fig. 8d) albert the endogenous Lp-PLA 2 activity of LDL was completely inhibited as shown in Fig. 5g.

Discussion
Lipases are enzymes that catalyze the hydrolysis of lipids at the ester bond and belong to a subclass of esterases. However, the main difference between regular esterases and lipases is that esterases prefer substrates in solution and lipases prefer substrates at the phase interfaces 23 . Like other lipases, Lp-PLA 2 catalyzes the hydrolysis of phospholipids at the lipid-water interfaces, involving interfacial adsorption and subsequent catalysis. Catalytical www.nature.com/scientificreports/ reactions in Fig. 1a demonstrate that Lp-PLA 2 has low hydrolytic activity when the concentration of artificial substrate,14:0 NPS-PC, at 10 µM is too low to form aggregates or micelles. Figure 8a shows that 14:0 NPS-PC elutes from the TSKgel G3000SW XL column as monomers at 10 µM. Titration of four detergents with different critical micelle concentration (CMC) results in the stimulation of Lp-PLA 2 hydrolytic activity around the CMC of each detergent. This correlation strongly suggests that the formation of water-micelle interface is critical for the activation of Lp-PLA 2 . On the other hand, the CMC of 14:0 NPS-PC is found to be near 0.5-1 mM (Fig. 3c). A plot of the reaction rate against the substrate concentration in the presence and absence of 10 mM CHAPS shows that the kinetics of the reaction are different under the two situations. While the reaction is typical Michaelis-Menten kinetics with Hill slope of 1.0 in the presence of 10 mM CHAPS, it becomes a positively cooperative sigmoidal curve with Hill slope of 2.0 in the absence of detergent. In the absence of detergent, the reaction is significantly accelerated at substrate concentration above 20 µM (log1.3, Fig. 3a) which is much lower than its determined CMC, 0.5-1.0 mM (Fig. 3c). This suggests that Lp-PLA 2 can hydrolyze monomeric or low aggregated form of 14:0 NPS-PC without the formation of micelles. Higher Hill slope in the absence of detergents indicates that 14:0 NPS-PC probably needs first to bind to the regulatory site of Lp-PLA 2 and to change its conformation before the enzyme can bind and hydrolyze the substrate at the active site (similar to interfacial activation) 24 . This leads to the positive substrate cooperation when concentrations of 14:0 NPS-PC are lower than 0.5 mM and explains the deviation of Michaelis-Menten mechanism in the absence of detergents (substrate activation instead of interfacial activation) 25 . When concentration of 14:0 NPS-PC reaches its CMC, the Hill slope of the hydrolytic reaction decreases to 1.0 due to the formation of substrate micelles. Detergents play roles in the formation of aggregate and activation of Lp-PLA 2 so that the enzyme can bind and hydrolyze the substrate. Since the enzyme has been activated in the presence of detergents, the Hill slope remains constant at 1.0 in all substrate concentration range (Fig. 3b). This supports the model first proposed by Wieloch et al. that there is a hydrophobic region in lipases which regulates the activation of the enzyme and is different from the enzyme active site 26,27 . Several decades of extensive studies have built models for interfacial activation of various PLA 2 28,29 . Interfacial activation of certain enzymes involves conformational changes or displacement of a lid which covers the active or catalytical site slot, while the others become activated via yet unidentified mechanisms 28 . It is generally in common that the membrane or a hydrophobic surface acts as an allosteric ligand, shifting the conformation of a PLA 2 from the closed form in water to the open form on the surface of the membrane or hydrophobic aggregate 29 . This process enables the enzyme to bind a phospholipid molecule in the active site (ES·M), where it is converted into product (EP·M) 29 . Figure 8a demonstrates that 14:0 NPS-PC forms mixed micelles with CHAPS or Tween-20, two represented detergents in the study, and has very low level of distribution in the bulk phase. Lp-PLA 2 has been shown to have the similar distribution in detergents 12 . Lp-PLA 2 may extract substrate from bulk phase, another micelle or the same micelle where it resides. If Lp-PLA 2 extracts substrate within the same micelle, bringing enzyme and substrate into close approximate is also an important factor to accelerate the reaction by detergents.
Unlike classical interfacial activation, further increase of detergent concentration does not enhance, but rather inhibits, the Lp-PLA 2 lipase activity. Lineweaver-Burk plot of the inhibition by Tween-20 reveals the competitive mechanism between substrate and detergent (Fig. 1c). Comparing the Michaelis-Menten kinetic parameters in the presence and absence of 10 mM CHAPS confirmed the competitive mechanism because competitive inhibition only affects K m and not V max 25 . The hydrophobic regulatory site of Lp-PLA 2 binds not only to the surface of detergent micelles but also to the surface of polystyrene latex beads. The interfacial activation can also be induced when Lp-PLA 2 binds to the surface of polystyrene latex nanobeads with diameter of 100 nm or smaller (Fig. 1e). Higher concentrations of polystyrene latex nanobeads also turn to inhibition of the Lp-PLA 2 activity (Fig. 1e). The smaller polystyrene nanobeads are slightly more potent in both the activation and inhibition of Lp-PLA 2 probably due to higher particle number (at 0.1% w/w, 25 nm beads = 1.2 × 10 14 particles/ml, 50 nm beads = 1.4 × 10 13 particles/ml and 100 nm beads = 1.8 × 10 12 particles/ml). Negatively charged polystyrene beads are more potent than the same size beads without charges. These phenomena support that the interfacial Lp-PLA 2 probably extracts substrates from bulk phase or different micelles because nanobeads can absorb Lp-PLA 2 on surface but contain no substrate. Therefore, surface dilution 30 cannot explain the inhibition. That carboxylated latex beads have better binding affinity for Lp-PLA 2 suggests that the hydrophobic binding site on the enzyme composes of hydrophobic and cationic residuals and therefore prefer the binding of negatively charged amphiphiles. This is similar with other PLA 2 28 . The same pattern in activation and inhibition of recombinant Lp-PLA 2 is also observed for HDL and LDL which are devoid of endogenous Lp-PLA 2 (Fig. 1f). Based on the results of our studies, it can be proposed that binding of Lp-PLA 2 onto the membrane or a hydrophobic surface stabilizes the substrate binding domain. The substrate binding structure of Lp-PLA 2 must be quite open in order to accommodate lipid aggregate or cell membrane, from which phospholipids can be extracted or diffuse to the catalytical site of the enzyme. High concentration of detergents, polystyrene nanobeads or lipoprotein particles may fill in the substrate binding structure and block the access of substrates to the catalytical site resulting in inhibition of the enzyme. This explains why materials with hydrophobic surface such as detergent micelles, lipoprotein particles or polystyrene latex nanobeads can be both activator and inhibitor for Lp-PLA 2 in a concentration dependent manner.
Earlier studies in our laboratory have shown that Lp-PLA 2 has a hydrophobic binding site which can be disrupted and cause collapsing of the enzyme structure if exposed to the aqueous milieu 31 . Thus, monomeric Lp-PLA 2 is unstable in aqueous bulk phase without substrates and therefore forms self-aggregates. Binding of Lp-PLA 2 to lipoproteins or other hydrophobic surfaces protects the enzyme from denaturation in addition to the activation of the enzyme 31 . In human circulation system, Lp-PLA 2 mainly associates with and is stabilized by LDL or HDL in an active conformation. This means that the substrate binding domain is in open conformation. What are the physiological substrates for Lp-PLA 2 is not completely clear. However, from our data, the concentration of lipoprotein particles may affect the access of substrates to the active site of Lp-PLA 2 because they can block the substrate binding domain of the enzyme and act as inhibitors. Lipoprotein particles or detergent micelles not only block the access of substrates but also block the access of inhibitors such as darapladib as demonstrated in Fig. 1d.
Scientific Reports | (2020) 10:17232 | https://doi.org/10.1038/s41598-020-74236-0 www.nature.com/scientificreports/ In Fig. 5a,b, Lp-PLA 2 is shown to be more sensitive to darapladib when the inhibitor is added with substrate together. If darapladib is preincubated with the lipoproteins before the addition of substrate, the inhibitory effect is reduced about ten-fold. The IC 50 of darapladib is between that of the above situations when the lipoproteins are preincubated with darapladib in the presence of 40 mM CHAPS. Under the experimental conditions, concentration of Lp-PLA 2 is in low double digit nM concentration range 32 and the substrate 14:0 NPS-PC is 10 µM. Darapladib is at 0.1 to 200 nM. One mg/ml of LDL is about 1.1 µM and one mg/ml of HDL is about 67.3 µM of particles according to NMR studies 33 . Because the concentrations of Lp-PLA 2 and darapladib are much lower than that of lipoprotein particles or detergent micelles, the odds for Lp-PLA 2 to meet substrate or inhibitor can be affected by many different factors such as preincubation and lipoprotein particle numbers, as well as the affinities of substrate and inhibitor to lipoprotein particles as shown in Fig. 5. At 1 mg/ml concentration, HDL has much higher particle concentration (67.3 µM) than that of LDL (1.1 µM) and, therefore, darapladib is less inhibitory to HDL-associated Lp-PLA 2 (Fig. 5b,d) because it is more diluted by HDL. As shown in Fig. 5d, inhibition of LDL Lp-PLA 2 by darapladib was enhanced when the lipoprotein concentration was reduced 100-fold. However, reducing HDL concentration by 100-fold made no difference because HDL particle concentration was still much higher than that of darapladib. Furthermore, LDL-bound Lp-PLA 2 may have higher affinity to darapladib than HDL-bound Lp-PLA 2 . While majority of darapladib is associated with HDL (Fig. 8d) and majority of Lp-PLA 2 is associated with LDL (Fig. 6a), LDL-bound Lp-PLA 2 activity can be completely inhibited but HDL-bound Lp-PLA 2 activity remains intact when human serum is incubated with excess concentration of darapladib and subjected to separation by size exclusion chromatography. It is interesting to note that HDL-bound Lp-PLA 2 is the indication of cardiovascular hazards but LDL-bound Lp-PLA 2 may be not 12,32 although the facts are difficult to be accepted by the research community due to that HDL is considered as "good" cholesterol. If HDL-bound Lp-PLA 2 does play roles in cardiovascular diseases, resistance of HDL-bound Lp-PLA 2 to darapladib may also contribute to the failure of the clinical studies.
Lp-PLA 2 bound on HDL or LDL is already in activated state and can also be inhibited by detergents (Fig. 2). When lipoproteins are treated with detergents, Lp-PLA 2 is usually extracted into detergent micelles ( Fig. 6b and reference 12 ). Also, darapladib is more soluble in the presence of detergents. These explain why darapladib has better inhibitory effect when lipoproteins are preincubated in the presence of CHAPS which forms micelles and improves the solubility of darapladib. Aggregation of ApoB indicates that LDL may be stripped off phospholipids by detergent micelles and collapsed (Fig. 6). However, majority of HDL seems still intact (Fig. 6) indicating that it is partially resistance to detergents. Preincubation of lipoproteins with darapladib in the absence of detergents is more closely resemble to in vivo physiological conditions. Under these conditions, darapladib may have limited inhibitory effects on HDL Lp-PLA 2 as shown in Fig. 5b,d,g. The reason why darapladib is more effective when delivered with the substrate of Lp-PLA 2 together is not clear. Mixing darapladib with 14:0 NPS-PC may improve the solubility of darapladib and therefore increases the inhibitory effects. The in vitro Lp-PLA 2 activity assays used in clinical studies to assess the residual Lp-PLA 2 activity usually includes detergents or does not preincubate serum samples with darapladib. Therefore, the in vitro Lp-PLA 2 activity assays used in clinical studies may overestimate the in vivo suppression of Lp-PLA 2 activity.
Lp-PLA 2 has been shown mainly associating with LDL and HDL in an average ratio about 3:1 respectively 10-13 . In our experiments, Lp-PLA 2 is only detected to transfer from LDL to HDL but not from HDL to LDL (Fig. 7). This suggests that Lp-PLA 2 on HDL may come from LDL and the transfer process is probably regulated because majority of Lp-PLA 2 is still distributed on LDL despite the fact that HDL particles out-number LDL particles. Darapladib has higher affinity with HDL and albumin, which carry less or no Lp-PLA 2 (Fig. 8d). The difference in distribution of Lp-PLA 2 and darapladib suggests that low efficacy of the drug can be expected. There may be a possibility that darapladib induces the migration of Lp-PLA 2 from LDL to HDL as shown in Fig. 5g. However, further studies are necessary to confirm.
In summary, Lp-PLA 2 ′s substrate binding site (active site) is modulated by a regulatory site which has affinity towards hydrophobic surfaces of membrane or aggregates. First, the hydrophobic surfaces of membrane or aggregates bind to the regulatory site of the enzyme and act as an allosteric ligand to change and stabilize the conformation of the active site (interfacial activation). Subsequently, it promotes the substrate binding on to the active site and facilitates the catalysis. Lp-PLA 2 also hydrolyzes high concentration (> 20 µM but < 0.5 mM) of the artificial substrate, 14:0 NPS-PC, in solution without interfacial activation by hydrophobic surfaces of membrane or aggregates. However, kinetics of the reaction is positively cooperative with Hill slopes > 1.0 under such conditions, possibly due to that the monomeric or lamellar substrate acts as the allosteric ligand to bind on the regulatory site of the enzyme. The positively cooperative Hill slopes decrease as the concentrations of the phosphocholine substrate increase and reach to 1.0 when substrate micelles form around the CMC of 14:0 NPS-PC (0.5-1 mM). Detergent micelles, polystyrene latex nanoparticles or lipoprotein particles not only activate Lp-PLA 2 by binding onto the regulatory site but also inhibit Lp-PLA 2 by blocking the active site. Blocking the active site by lipoproteins in circulatory system actually protects Lp-PLA 2 from inhibition by potent inhibitors such as darapladib. Darapladib has very low solubility and mainly associates with HDL and albumin. High concentration of lipoproteins will dilute the in vivo concentrations of darapladib and result in the efficacy variability for patients. Furthermore, In vitro activation and inhibition of Lp-PLA 2 are very much dependent on the buffer components because it affects the affinities of enzyme to inhibitors and the availability of the inhibitors. In vitro Lp-PLA 2 activity assays usually involve detergents and therefore may mislead the estimation of the in vivo efficacy of drugs due to that detergent micelles can extract the lipoprotein-associated enzymes from lipoprotein particles. Preincubation of lipoprotein particles with darapladib more closely resembles to the in vivo conditions but shows much less effective for darapladib in the inhibition of Lp-PLA 2 . Although darapladib mainly associated with HDL and albumin, HDL-associated Lp-PLA 2 has lower affinity to bind the inhibitor. It is also worth to note that HDL-associated Lp-PLA 2 predicts CVD events but LDL-associated Lp-PLA 2 does not correlate with the diseases 12 31 . Briefly, the purchased lipoproteins were thawed and subjected to inactivation of Lp-PLA 2 by incubation with 20 mM Pefabloc SC (Roche Applied Science, Indianapolis) in PBS, pH 7.2, at 2-8 ºC overnight. The Pefabloc SC inactivated lipoproteins were then dialyzed extensively with a 10 kD cutoff membrane in 1000-fold volume excess of buffer containing 50 mM phosphate, pH 7.2, and 150 mM sodium chloride with 3 exchanges at 2-8 ºC. The inactivated lipoproteins were found to have less than 10% of the original endogenous Lp-PLA 2 activity by the PLAC activity assay. Both lipoproteins were further diluted to the desired concentrations before used in each experiment. where I VV indicates the intensity with vertically polarized excitation and vertical polarization on the detected emission. I VH indicates the intensity when using a vertical polarizer on the excitation and horizontal polarizer on the emission. G is a grating factor used as a correction for the instrument's differential transmission of the two orthogonal vector orientations.

Reagent preparation.
Darapladib was dissolved in dimethyl formamide (DMF) as 15 mM solution and stored at − 70 °C. The solution was further diluted in 50% isopropanol to 100 µM or 2 mM before mixing with buffers, HDL or LDL at the final concentration of 0.2 or 20 µM. For inhibiting human serum, darapladib in DMF was diluted in 50% isopropanol to 3 mM. To 5 µl of the 3 mM solution, human serum was added and mixed. The final concentration of darapladib was150 µM. 14:0 NPS-PC was dissolved in isopropanol or DMF at the concentration of 145 or 290 mM and stored at − 70 °C. The stock solution was directly diluted into buffers at varied concentration. Experiment protocols related to human or animal samples were carried out in accordance with relevant guidelines and regulations. Methods were reviewed and approved by Research Ethical Review Board of diaDexus Inc.