Structure-based, multi-targeted drug discovery approach to eicosanoid inhibition: Dual inhibitors of mPGES-1 and 5-lipoxygenase activating protein (FLAP)
Graphical abstract
Introduction
The arachidonic acid cascade is perhaps the most studied and well characterized pathway associated with human disease [[1], [2], [3]]. For centuries, long before there was an understanding of the exact molecular species involved, humans have used intervention of this pathway to alleviate multiple conditions, including pyrexia, inflammation and pain. As shown in Fig. 1a, it is now well understood that Arachidonic Acid (AA) is converted to prostaglandins (PGs) via the cyclooxygenase (COX) arm of the pathway and to leukotrienes (LTs) via the 5-Lipoxygenase (5-LO) arm. The product PGs and LTs bind to their respective receptors (prostaglandin E2, leukotriene B4, and cysteinyl leukotriene receptors) to elicit biological responses associated with multiple maladies [4].
Inhibition of PGE2 formation, through the use of steroidal and non-steroidal drugs, has proved to be a very successful way of mitigating inflammation, as well as many of the associated symptoms [2]. Many non-steroidal anti-inflammatory drugs (NSAIDs) have demonstrated robust efficacy through the non-selective or selective inhibition of COX-2 [5]. These approaches have not been without side effects, such as increased risk of GI [6] and cardiovascular adverse events [7,8]. More recently, it has been demonstrated that PGE2 lowering can also be accomplished via inhibition of the downstream enzyme in the AA cascade, the microsomal prostaglandin E synthase-1 (mPGES-1) [9]. It has been hypothesized that inhibition of this step could provide efficacy similar to older NSAIDs that target COX-1 and COX-2, but with an improved safety profile [10].
Similarly, enzymes and receptors in the LT arm of the AA pathway have been explored for various indications associated with inflammation [11], with clinical emphasis on pulmonary inflammation [12]. It has been shown that inhibition of the enzyme 5-lipoxygenase (5-LO) and its partner protein, the 5-lipoxygenase activating protein (FLAP) results in lower levels of downstream LTs [13]. Since the purpose of FLAP is to deliver AA to 5-LO, it is not surprising that inhibition of either target results in a functionally similar decrease in LTs. More recently, several FLAP inhibitors have been investigated in the clinic [[14], [15], [16]] and new scaffolds have also been discovered through virtual screenings [17,18]. Findings from Werner et al. [19] revealed that inhibition of FLAP may be more beneficial than direct inhibition of 5-LO as FLAP inhibitor MK-886 was able to suppress the formation of LTs without affecting the biosynthesis of proresolving lipid mediators (lipoxins, resolvins, maresins, and protectins), which function to terminate inflammation and promote tissue repair.
The hypothesis that simultaneous inhibition of both PGs and LTs could provide superior anti-inflammatory efficacy has been explored for over 20 years and multiple tool compounds have been developed that inhibit both the COX and 5-LO pathways [20,21]. The most thoroughly studied agent is licofelone (ML3000) [22], which has been reported to be a dual COX + 5-LO pathway inhibitor, suppressing the formation of both PEG2 and LTB4 [23,24]. Researchers have reported the potential for this agent to not only provide pain efficacy, but also to result in disease modification in knee osteoarthritis (OA) patients [25]. Additionally, it has been reported that licofelone has a superior GI safety profile, relative to COX-2 inhibitors, such as rofecoxib [26].
In recent years, significant understanding of eicosanoid pathway redistribution has emerged, perhaps adding greater appreciation for potential benefits of a Multi-Targeted Drug Discovery (MTDD) approach to reducing PGs and LTs. Redistribution occurs in response to the inhibition of one enzyme in the pathway, which can result in a concentration increase in the relevant substrate. This increased substrate can be shunted towards other pathways, producing greater quantities of other pro-inflammatory species (Fig. 1B). In vitro, He et al. [27] demonstrated this phenomenon in a rat whole blood assay. For example, during a 120-min incubation of calcium ionophore-stimulated rat whole blood, inhibition of COX with indomethacin resulted in the expected decrease in PGE2 levels, but also a concomitant increase in LTB4 levels. This is best explained via substrate (AA in this case) redistribution. Similarly, when the same experiment was run in the presence of the 5-LO inhibitor Zileuton, decreases in LTB4 occurred in concert with a corresponding increase in PGE2. More specifically, Werner et al. [19] demonstrated shunting occurring in human macrophages; upon treatment with COX inhibitors (ibuprofen and celecoxib), increased level of 5-LO products such as LTB4 was observed.
We have recently reported on the identification of novel inhibitors of mPGES-1 [28]. In considering the potential to identify agents which might inhibit the formation of pro-inflammatory mediators in both pathways, we were interested in combining mPGES-1 and FLAP inhibition into a single molecule. This is an attractive approach for several reasons. First, it was demonstrated by He et al. [27] that this combination of activities gave one of the more preferred MTDD profiles in vitro. Additionally, since both targets are structurally enabled, the use of X-ray crystallography and SBDD should inform overlapping pharmacophores which may improve the chances of identifying a dual inhibitor. Finally, mPGES-1 and FLAP are phylogenetically similar, both being members of the membrane associated proteins in eicosanoid and glutathione metabolism (MAPEG) superfamily [29]. While most of the reported mPGES-1 inhibitors are not particularly similar in structure to the reported FLAP inhibitors (Fig. 2A), it is interesting to note that Riendeau et al. [30] have described how the FLAP inhibitor MK-886 was used as a hit for their mPGES-1 inhibitor program and was ultimately optimized into a potent mPGES-1 inhibitor 3 (Fig. 2B). (ref: The in vitro data shown in Fig. 2B as reported in Riendeau et al. [30] and is in agreement with our internal experiments as noted in Table 1) However, the group did not describe compounds that retained significant activity towards both targets. More recently, BRP-187 was reported as a potential multi-target mPGES-1/FLAP inhibitor with superior inhibition of mPGES-1 over MK-886 [[31], [32], [33]].
Structurally, both mPGES-1 and FLAP are trimeric integral membrane proteins with four transmembrane helices in each monomer (α1 – α4), but share <5% sequence identity. Distinguished by their function and cellular localization, mPGES-1 is an enzyme in the endoplasmic reticulum and microsomal membranes responsible for the synthesis of PGE2 from PGH2, and FLAP activates 5-LO in the nuclear membrane by presenting AA to 5-LO, but itself is devoid of any enzymatic activity.
X-ray crystal structures of inhibitor-bound FLAP [34] and mPGES-1 [28,[35], [36], [37], [38]] together with this report show that the two proteins do not share the same inhibitor binding pocket (Fig. 3–5). Ligands binding FLAP largely bury themselves in the non-polar center of the trimer, situated between α2 and α4 of one FLAP monomer and α1’ and α2’ of another FLAP monomer (Fig. 4a). Polar ends of the ligands project out to the phosphate-exposed protein surface, where the ligands can form salt-bridges with basic sidechains of FLAP, which are in turn engaging the anionic phosphate groups of the plasma membrane, on the side adjoining the aqueous cytosol. These ligands experience little exposure to the lipid bilayer when bound to FLAP. Ligands binding mPGES-1 trimers also require two distinct monomers, but the nature of the protein-ligand complex differs markedly from that in FLAP. While acidic ends of ligands can also form ionic bonds with basic sidechains of the phosphate-exposed region of transmembrane mPGES-1 protein, they do not penetrate into the trimeric core. Instead, the remainder of the ligands extend down the lipid-exposed protein surface, between α4 of one mPGES monomer and α1’ of another monomer (Fig. 5a). Moreover, unlike mPGES-1, FLAP activity is not dependent on glutathione and does not have a glutathione binding pocket, consistent with its lack of enzymatic activity. Interestingly, each FLAP monomer also possesses a long acidic C-terminal tail, which could serve as a handle for forming a complex with 5-LO.
While our work on mPGES-1 allowed us to solve many X-ray structures of ligands bound to mPGES-1, we had significantly less structural understanding of FLAP. Besides the relatively low resolution (~4 Å) ligand-bound FLAP crystal structures [34], we desired additional structures of FLAP to better understand potential overlapping pharmacophores of mPGES-1 and FLAP.
Section snippets
Crystal structure of FLAP and mPGES-1 bound to DG-031 (Bay X-1005)
A crystal structure of FLAP was previously solved with MK-591 to 4.0 Å [34]. While MK-591 is visible in the structure, side chain conformations and water molecules are not resolved at that resolution. Our FLAP/DG-031 crystal structure was solved to 2.4 Å with clearly defined side chain conformations and water molecules, providing strong evidence that the ligand binding site is in the phosphate-exposed region of FLAP, not in the lipid-exposed region. Comparing these two structures, the binding
Conclusion
In summary, we have described the solution of new X-ray crystal structures of protein-ligand complexes with FLAP at improved resolutions. By combining these findings with our previous understandings of mPGES-1 inhibitors, we demonstrate the value of structure-based drug design tools towards multi-targeted drug discovery efforts. By defining the conserved binding regions of both mPGES-1 and FLAP, we were able to, a priori, design compounds that combined the recognition elements of each
mPGES-1 enzyme inhibition assay
Expression of human mPGES-1 in HEK293E cells, preparation of microsomes and enzyme inhibition assays using PGH2 as substrate were performed as previously described [37]. Production of PGE2 was quantitated via LC/MS, converted to percent specific inhibition and plotted against compound concentration (62.5 μM to 3.2 nM). Data were fit to a four-parameter nonlinear logistic equation using Activity Base (IDBS, Guildford, UK) to determine IC50 values.
A549 epithelial carcinoma cell assay
Cell-based inhibition of PGE2 production after
Author contributions
J.D.H., J.G.L., and S.A. solved the crystal structures. J.D.H., B.C., C.T.R. J.S.P., M.M., and A.Z. expressed and purified the proteins. K.A. and M.J.H. crystallized the proteins. N.E.H. and B.H.N synthesized the compounds. S.C., A.V.S., K.G., A.H., X.Y. and S.D.K. performed the biological assays. M.R.L. performed the computational modeling. J.D.H., M.R.L., S.D.K., and B.H.N. design and supervised the research and wrote the manuscript.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Declaration of Competing Interest
The authors declare no competing financial interest.
Acknowledgement
The authors thank the staff at the Lilly Research Laboratories Collaborative Access Team (LRL-CAT) beamline Sector 31 of the Advanced Photon Source. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357.
References (43)
Prostanoids in health and disease
J. Lipid Res.
(2009)Effects of a FLAP inhibitor, GSK2190915, in asthmatics with high sputum neutrophils
Pulm. Pharmacol. Ther.
(2014)Inhibitors of the inducible microsomal prostaglandin E2 synthase (mPGES-1) derived from MK-886
Bioorg. Med. Chem. Lett.
(2005)Identification of multi-target inhibitors of leukotriene and prostaglandin E2 biosynthesis by structural tuning of the FLAP inhibitor BRP-7
Eur. J. Med. Chem.
(2018)BRP-187: a potent inhibitor of leukotriene biosynthesis that acts through impeding the dynamic 5-lipoxygenase/5-lipoxygenase-activating protein (FLAP) complex assembly
Biochem. Pharmacol.
(2016)Discovery and characterization of [(cyclopentyl)ethyl]benzoic acid inhibitors of microsomal prostaglandin E synthase-1
Bioorg. Med. Chem. Lett.
(2017)The minimum significant ratio: a statistical parameter to characterize the reproducibility of potency estimates from concentration-response assays and estimation by replicate-experiment studies
J. Biomol. Screen.
(2006)The eicosanoids and their biochemical mechanisms of action
Biochem. J.
(1989)Prostaglandins and leukotrienes: advances in eicosanoid biology
Science
(2001)- et al.
Prostaglandins and inflammation
Arterioscler. Thromb. Vasc. Biol.
(2011)
Anti-inflammatory drugs in the 21st century
Subcell. Biochem.
The coxibs, selective inhibitors of cyclooxygenase-2
N. Engl. J. Med.
Risk of cardiovascular events associated with selective COX-2 inhibitors
JAMA
Role of prostacyclin in the cardiovascular response to thromboxane A2
Science
Characterization of microsomal prostaglandin E synthase 1 inhibitors
Basic Clin. Pharmacol. Toxicol.
Cyclooxygenases, microsomal prostaglandin E synthase-1, and cardiovascular function
J. Clin. Invest.
Biosynthesis and metabolism of leukotrienes
Biochem. J.
Leukotrienes
N. Engl. J. Med.
The membrane organization of leukotriene synthesis
Proc. Natl. Acad. Sci. U. S. A.
Initial clinical experience with AZD5718, a novel once daily Oral 5-lipoxygenase activating protein inhibitor
Clin Transl Sci
Discovery and early clinical development of an inhibitor of 5-lipoxygenase activating protein (AZD5718) for treatment of coronary artery disease
J. Med. Chem.
Cited by (13)
Identification of selective 5-LOX and FLAP inhibitors as novel anti-inflammatory agents by ligand-based virtual screening
2024, European Journal of Medicinal ChemistryRotational constriction of curcuminoids impacts 5-lipoxygenase and mPGES-1 inhibition and evokes a lipid mediator class switch in macrophages
2022, Biochemical PharmacologyCitation Excerpt :The binding site was defined in a 12 Å radius around the coordinates 12.86, –22.259, −18.586, and ChemPLP (https://www.ccdc.cam.ac.uk/support-and-resources/support/case/?caseid = 5d1a2fc0-c93a-49c3-a8e2-f95c472dcff0) was used as a scoring function. The docking simulation on mPGES-1 was conducted on the pdb entry 6VL4 with the co-crystallized ligand DG-035 [33]. The monomeric structure was displayed as a trimer by creating symmetry mates in PyMOL (PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC).
Untangling the web of 5-lipoxygenase-derived products from a molecular and structural perspective: The battle between pro- and anti-inflammatory lipid mediators
2021, Biochemical PharmacologyCitation Excerpt :Mutagenesis studies have shown that the AA-binding site and the inhibitor binding sites partially overlap [35]. > 10 years later higher-resolution structures of FLAP, bound to MK-866 and to the quinoline DG-031 (2.4 and 2.6 Å), were published [36]. The significantly improved resolution for these complexes allowed for the detailed placement of side-chains, inhibitors, and the water molecules that mediate interactions with the protein.