Targeting the pregnane X receptor using microbial metabolite mimicry

Abstract The human PXR (pregnane X receptor), a master regulator of drug metabolism, has essential roles in intestinal homeostasis and abrogating inflammation. Existing PXR ligands have substantial off‐target toxicity. Based on prior work that established microbial (indole) metabolites as PXR ligands, we proposed microbial metabolite mimicry as a novel strategy for drug discovery that allows exploiting previously unexplored parts of chemical space. Here, we report functionalized indole derivatives as first‐in‐class non‐cytotoxic PXR agonists as a proof of concept for microbial metabolite mimicry. The lead compound, FKK6 (Felix Kopp Kortagere 6), binds directly to PXR protein in solution, induces PXR‐specific target gene expression in cells, human organoids, and mice. FKK6 significantly represses pro‐inflammatory cytokine production cells and abrogates inflammation in mice expressing the human PXR gene. The development of FKK6 demonstrates for the first time that microbial metabolite mimicry is a viable strategy for drug discovery and opens the door to underexploited regions of chemical space.


Synthetic Procedures
Overview of the synthetic routes. The screened compounds are highlighted in blue (DMF -N,Ndimethyl formamide, THF = tetrahydrofurane, µw = microwave); initial synthetic targets in green, lead compounds in red. Interestingly, treatment with hydrochloric acid selectively removes the terminal ethyl ether from both EOM-protected indoles. The remaining hydroxymethyl groups can then be cleaved using lithium hydroxide in methanol (THF = tetrahydrofurane).
The mixture was stirred at 0 °C and followed TLC analysis of reaction aliquots (micro-workup: satd. aq. NH4Cl/EtOAc). After 3 h, complete conversion of the starting material was observed. 1 The milky reaction mixture was poured on 200 mL ice-cold satd. aq. NaHCO3. The mixture was transferred into a separatory funnel, extracted with ethyl acetate (300 mL in total). The organic layer was washed with satd. aq. sodium bicarbonate (100 mL), water (100 mL), satd. aq. NaHCO3 (100 mL), and brine (100 mL). The organic layer was dried (MgSO4), filtered and evaporated in vacuo to receive the crude product as pale yellow oil. Chromatography (silica; 0®15 % EtOAc in hexane) afforded 1-(ethoxymethyl)-1H-indole (3.17 g, 18.1 mmol, 71 %) as colorless oil. 1 In other runs, the mixture was stirred overnight, which turned out to be disadvantageous, as considerable decomposition was observed. It is hence recommended to follow the reaction closely and work up as soon as complete conversion is reached.
Isonicotinaldehyde (202 mg, 1.88 mmol, 1.10 equiv) was added at -78°C and the reaction mixture was allowed to slowly warm to RT overnight. TLC analysis of a reaction aliquot (micro-workup: satd. aq. NH4Cl/EtOAc) indicated complete conversion. 10 mL satd. aq. NH4Cl were added, and the resulting mixture was extracted with EtOAc (1 ´ 20 mL and 2 ´ 10 mL). The combined organic layers        3 Two columns were required to reduce contamination with triphenylphospine oxide. Compounds were used in the next step if the purity was 95% or better as judged by 1 H-NMR.    was used successfully to obtain 7b, which was subsequently cyclized using the same silver-catalyzed protocol as described above (µw = microwave) to obtain 8a.

Gram-Scale Synthesis of lead compounds FKK5 (S5b) and FKK6 (S6b)
Scheme 7: Improved synthesis of S5b and S6b: For a more efficient synthesis of the two lead compounds, the synthetic route was slightly modified. The addition of a lithium reagent, directly generated from S3b and n-BuLi, to the Weinreb amide of isonicotinic acid (S10) gave easy and scalable access to S5b in only one synthetic step, starting exclusively from commercially available starting materials and reagents. The previously utilized addition of a propargylic Grignard reagent to S5b proved scalable and delivered S6b in excellent yield. The mixture was warmed to RT with vigorous stirring. Water (40.0 mL) was added, followed by EtOAc (200 mL), the mixture was vigorously shaken, the layers separated. The aqueous layer was extracted with EtOAc (2 ´ 150 mL), the combined organic layers were dried (MgSO4), filtered and evaporated in vacuo. Crude NMR analysis indicated ca. 85 % purity of the crude product. Chromatography (silica; 0®15% EtOAc in CH2Cl2) afforded the desired product (13.1 g, 36.1 mmol, 80 %) as off-white solid.
X-ray Analysis. Single crystals of FKK5 and FKK6, respectively, were submitted for X-ray diffraction analysis. The data were collected on a Bruker X8 Kappa Apex II diffractometer using Mo Ka radiation.
Crystal data, data collection and refinement parameters are summarized in Table EV1 (FKK5) and   Table EV2 (FKK6), respectively. The structures were solved using a dual-space method and standard difference map techniques, and refined by full-matrix least-squares procedures on

In silico experiments
A hybrid structure-based method has been adapted for modeling pharmacophore at PXR ligand binding pocket. Further, considering the dynamic nature of the protein, a conformational-based docking experiment has been conducted with identified FKK compounds at multiple sites in the PXR LBD (i.e., conventional ligand binding pocket, AF2 site as well as α8-pocket that was identified from earlier simulation study of PXR LBD 6 ). The methodology for pharmacophore modeling and PXR ensemble-based docking are described below in sections (a) and (b). was then used to screen our library of vendor available small molecules and 5 hit molecules that strictly obeyed the pharmacophore were docked into the LBD of PXR using docking program GOLD and the complexes were scored using goldscore and chemscore functions adopted in GOLD program. FKK999 and BAS451 were chosen for in vitro testing and based on the results, FKK999's core was chosen for further medicinal chemistry optimization. Based on the core structure, ten molecules designated as FKK1-10 were designed and synthesized. All FKK molecules were modeled and docked as described previously and FKK5 and FKK6 were the lead ranking molecules.
(b) Ensemble based-molecular docking to multiple sites. Ensemble-based molecular docking of ten FKK compounds were performed using GOLD suite version 5.5.0 (CCDC, Cambridge, UK) 14 .
GOLD uses a genetic algorithm (GA) to explore the conformational flexibility of the ligand and receptor side chains in the binding pocket. Thirty centroid conformations of apo hPXR generated using a RMSD-based clustering algorithm, obtained from previous work, were used for the docking 6 .
In all the protein conformations, water and ions were removed prior to docking. For the docking purpose, a binding site was defined by considering all atoms within 12 Å from the geometrical center of the docking site. For each of the 30 independent GA runs, a maximum number of 200 GA operations were performed. The docked complexes were ranked with goldscore and then rescored using a chemscore fitness function 14 . All the FKK compounds were docked at all the known binding sites in hPXR, including α8 pocket, to evaluate the relative affinity at different sites. The scoring functions account for the hydrogen bonding, vdW interactions and steric complementarity between the ligand and receptor. For each ligand, the best-ranked docked pose with corresponding chemscore was considered. Along with the scores, analysis was performed by visualizing the residues interacting with the ligand, using LigPlot + software 15 .
Immunoblotting. The cells were centrifuged (1500 rpm for 3 min), the medium was removed, and the pellet was resuspended in 150 µl of ice-cold lysis buffer (150 mM NaCl; 50 mM HEPES; 1% (v/v) Triton X-100; 5 mM EDTA; anti-protease cocktail, anti-phosphatase cocktail). The mixture was vortexed and incubated for 10 min on ice and then centrifuged (15000 rpm/13 min/4 °C). Supernatant was collected and the protein content was determined using the Bradford reagent. SDS-PAGE gels (10%) were run on a BioRad apparatus according to the general procedure followed by the protein  nM to 25 µM. Dimethyl sulfoxide (DMSO; 1% v/v) and GW4064 (500 nM) were used as a negative and positive control, respectively. The reaction mixture was incubated at room temperature for 1 hour in a dark, and then fluorescent signals were measured at 495 nm and 520 nm, with the excitation filter 340 nm, on Infinite F200 microplate reader (Tecan Group Ltd, Switzerland). Finally, the TR-FRET ratio was calculated by dividing the emission signal of 520 nm by that at 495 nm. All experiments were done in tetraplicates and as two independent experiments. Final EC50 values were obtained by processing the data with GraphPad Prism 6 using standard curve interpolation (sigmoidal, 4PL, variable slope).
PPARg reporter assay. Transcriptional activity of PPARγ was studied in stably transfected reporter cell line PAZ-PPARg 20 . Cells were seeded in 96-well plates, stabilized for 24 h, and then incubated with tested compound. Dimethyl sulfoxide (DMSO; 0.1% v/v) and 15-deoxy-δ12,14-prostaglandin J2 (15d-PGJ2; 40 µM) were used as a negative and positive control, respectively. After the treatments, cells were lysed and luciferase activity was measured on Tecan Infinite M200 Pro plate reader (Schoeller Instruments, Prague, Czech Republic). The data are expressed as fold induction ± SD of luciferase activity over the control cells. Differences were tested using one-way ANOVA with Dunnett's post hoc test, p < 0.05, was considered significant (*).