Mycophenolate suppresses inflammation by inhibiting prostaglandin synthases: a study of molecular and experimental drug repurposing

Chemicals, reagents, instruments, and software

The highest quality chemicals including λ-carrageenan, indomethacin, diclofenac sodium, celecoxib, MMF, MFA, tris(hydroxymethyl) aminomethane, hydrochloric acid, carboxymethyl cellulose sodium salt (NaCMC), EDTA, phenol, and hematin were purchased from Sigma–Aldrich, Inc. (St. Louis, MO, USA). Quercetin was obtained from APExBIO (Houston, TX, USA). The COX inhibitor screening assay kit and LOX inhibitor screening kit were purchased from Cayman Chemical (Ann Arbor, MI, USA). The vernier caliper was from SMIEC (Shanghai, China). The Schrodinger Maestro package release 2020.1 (Schrodinger, Inc., LLC, New York, USA) was used in compound and protein preparation and docking runs. Groningen Machine for Chemical Simulations (GROMACS) version 2018.8 software was available at (http://www.gromacs.org/). GraphPad Prism version 7 software was purchased from (GraphPad Software, Inc., San Diego, USA).

Retrieval and optimization of drug structures

The structures of meclofenamic acid (MFA) and MMF (Fig. 1) were retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/, accessed on February 10, 2020) and stored in SDF format. The structures were imported to Ligprep software (Schrodinger Maestro package) for 3D optimization using OPLS2005 force field at physiological pH.

COX-2 structure optimization and preparation

The Maestro software package was used to prepare the structure for molecular docking. The protein (PDB ID 5IKQ, https://www.rcsb.org/structure/5IKQ) was corrected for the missing atoms and residues, protonated, optimized at physiological pH, and energy minimized. The OPLS2005 force field was used. The docking grid was generated by forming a grid box of 20 Å around the bound ligand, meclofenamic acid.

Molecular docking

The high precision (XP) glide docking module of the Schrodinger Maestro package was used for molecular docking. The results were ranked by docking score. The accuracy of results was confirmed by comparing the docking of MFA with the coordinates of the bound ligand in the structure file. The molecular mechanics/generalized Born surface area (MM/GBSA) was calculated from the docked poses.

Determination of COX-1, COX-2, and 12-LOX inhibition

The inhibitory effects of the drugs MMF and MPA on COX-1 and COX-2 were measured using the COX inhibitor screening assay kit and LOX inhibitor screening kit. The assay was performed according to the manufacturer’s directions and previously described methods (Chowdhury et al., 2008; Lee et al., 2004). Briefly, the reaction buffer (Tris–HCl buffer, pH 8.0) containing 5 mM EDTA, 2 mM phenol, and 1 µM hematin was measured into test tubes. The test drugs were dissolved in dimethyl sulfoxide and added to the tubes to final concentrations of 0.05–200 mM. COX-1 or COX-2 was added to the test tubes and preincubated for 10 min at 37 °C. Arachidonic acid was then added, and the tubes were incubated for 2 min at 37 °C. The enzyme immunoassay was used to measure the concentration of the product prostaglandin F2α (PGF2α). The estimated values were determined after comparison to the predetermined standard curve. The IC₅₀ was determined after three different experiments. The selectivity index was calculated from Eq. (1):

(1) $$Selectivityindex=\frac{I{C}_{50}COX-1}{I{C}_{50}COX-2}$$

Evaluation of anti-inflammatory activity in rats with carrageenan-induced paw edema

The in vivo assay was performed by the rat paw edema inhibition test as previously described (Morris, 2003; Winter, Risley & Nuss, 1962). Forty-nine male adult albino rats weighing 220–180 g were obtained from the Animal Breeding Center at the College of Veterinary Medicine, King Faisal University, and were randomly divided into seven groups (n = 7/group). The animal groups included the nontreated control group, carrageenan group, and five groups receiving carrageenan and treated with either indomethacin, celecoxib, or MMF at three doses (12.5, 25, and 50 mg/kg). All rats were allowed free access to food and water. All cages were kept in controlled rooms at 25 °C and exposed to a 12-h light/dark cycle. MMF and two reference drugs, indomethacin and celecoxib, were suspended in 1% NaCMC in normal saline. A 1% carrageenan solution in normal saline was injected into the subplantar area of the right hind paw of rats. One hour before injection of carrageenan, rats were given either indomethacin or celecoxib at a dose rate of 10 mg/kg or MMF at three dose levels of 12.5, 25, and 50 mg/kg by oral administration. A vernier caliper was used to measure the rat paw thickness of all rats in all groups at 0, 1, 2, 3, 4, and 5 h after drug administration. The measurements were blind to avoid measurement bias. The King Faisal University Ethics Committee (approval no. KFU-REC/2020-10-11) approved all animal experiments. Since no samples were taken from rats, there were no anesthesia procedures. The size of the hind paw was only measured in non-anesthetized rats. At the end of the experiment, all rats were euthanized according to the regulations of the King Faisal University Ethics Committee by an overdose of pentobarbital (200 mg/kg IP) and decapitation.

Molecular dynamics simulations

GROMACS software was used in all molecular dynamics (MD) simulations (Abraham et al., 2015; Van Der Spoel et al., 2005). AMBERFF14SB force field and general AMBER force field (GAFF) were used to handle the ligand parameter, topology, and restraint and the protein, respectively. The COX-2-drug complexes were solvated in a cubic box filled with a single point-charge (SPC) water model. The boundaries of the box were 1.0 nm from the protein to the box edge. The solvated complexes were minimized for 5,000 steps. During water and ion coupling, the heavy atoms of protein and ligand were restrained. The system equilibration comprised 50 ps NVT at 300 K followed by 1 ns NPT ensemble at 300 K. Production stages were extended for 50 ns with NPT ensemble. The constant 1 bar pressure and 300 K temperature were maintained by Parrinello-Rahman and V-rescale thermostat algorithms, respectively. The PME method was used to manage long-range electrostatics. The time step was 2 fs, and snapshots were collected every 10 ps.

Post-dynamic analysis

The GROMACS MD simulation toolkits performed the post-dynamic trajectory analysis. RMSD and RMSF were calculated by the g_rms and g_rmsf functions. The number of hydrogen bonds and hydrogen bond length were calculated.

Statistical analysis

The effect of MMF at different doses on relieving rat paw edema was compared with the control and standard NSAID-treated groups. Significant differences were determined by one-way analysis of variance (ANOVA) with Tukey’s post hoc test. All steps were implemented using GraphPad prism software (GraphPad Software, La Jolla, CA, USA).

Molecular docking

The results of drug docking with COX-2 are shown in Table 1. The output parameters include the docking scores, ligand efficiencies, and lipophilic and hydrogen bonding interactions. The docking score of MMF was approximately 78% of the score of MFA. This result was supported by favorable hydrogen bonding and electrostatic and hydrophobic interactions. Notably, these parameters were lower than the standard anti-inflammatory ligand, MFA. The generalized Born model and solvent accessibility (MM/GBSA) binding energy and its decomposition (Table 2) had similar profiles to the estimated docking. The binding of MMF showed lower MM/GBSA binding energy as a result of higher hydrogen bonding and lipophilic interaction scores and lower electrostatic forces.

The binding mode was analyzed (Fig. 2) to gain deeper insights into the interaction of drugs with COX-2. MFA formed two hydrogen bonds with TYR385 and SER530 (Fig. 2B). MMF formed similar interaction with the presence of bidentate hydrogen bonds with SER30 (Fig. 2C).

Inhibition of cyclooxygenases and lipoxygenase by in vitro enzyme assays

The inhibitory potency of MMF and MPA against COX-1, COX-2, and 12-lipooxygenase (12-LOX) was measured and compared to standard inhibitors for these enzymes. MMF and MPA inhibited the three enzymes with MMF IC₅₀ values of 5.53, 0.19, and 4.47 µM for COX-1, COX-2, and 12-LOX, respectively. The IC₅₀ for MPA was 4.62, 0.14, and 4.49 µM, respectively. The selectivity index for MMF and MPA was at least 29.1 and 6.5-fold higher than indomethacin and diclofenac, respectively. In contrast, it was 11.4-fold lower than celecoxib. Both MMF and MPA were weaker than quercetin in inhibition of 5-LOX with IC₅₀ values of 4.47 and 4.49 µM, respectively (Table 3).

Carrageenan-induced paw edema

Injection of carrageenan into rat paws resulted in a marked increase in paw size from 0.3 to 0.8 cm. Treatment with 12.5, 25, and 50 mg/kg MMF resulted in a significant decrease in rat paw edema 2–5 h after drug administration (Fig. 3). Measuring rat paw size at 3, 4, and 5 h after treatment revealed a dose-dependent decrease in edema, which was highly comparable to treatment with indomethacin and celecoxib (Fig. 4). Statistical analysis (Table 4) of the treated groups indicated. MMF at 12.5, 25, and 50 mg/kg significantly relieved the rat paw edema (p < 0.001) compared to the control group. There was no statistical difference between the effect of MMF and the standard COX inhibitors, indomethacin and celecoxib. Slight differences were observed in the average hourly results in the 3- to 5-h range after treatment (Table 4). Analyzing hours 3 and 4 separately revealed that 50 mg/kg MMF treatment results were significantly different from the other two doses and were comparable to indomethacin and celecoxib. Therefore, the 50 mg/kg dose was the only dose that provided significant relief of paw edema.

Molecular dynamics simulation

MD simulation provides more comprehensive insights into drug-receptor dynamics. These results, in combination with docking runs, help to rank the drugs based on binding affinity and dynamics. Docking followed by MD simulation can provide a complete picture of drug-protein complex stability and structural changes associated with their binding. We subjected ApoCOX-2 or COX-2 bound with MMF and MFA to MD simulations for 50 ns (Fig. 5). The structure RMSD indicated more stable COX-2-drug complexes over the Apo enzyme, indicating stable complexes with the drugs. The drug complexes were evidently stabilized early in the first 5 ns to the end of the simulation. Furthermore, the average RMSD was low (<1.5 Å), indicating good structural stability (Fig. 5A). The RMSF values indicated conserved profiles for all structures with markedly low drug complex values at the C-terminal of COX-2 (Fig. 5B). A high RMSF peak was observed in the residue range 336–338. We traced the number of hydrogen bonds during simulation time (Fig. 6). MFA formed 1–3 hydrogen bonds, while MMF formed 0–4 bonds. In addition, MFA formed a stable and strong H-bond with TYR385 (Fig. 6B). MMF did not show a similarly stable profile with distance ranges from 0.5–1.2 nm. This result might be driven by the high number of rotatable bonds in MMF (11 bonds), compared with three bonds in MFA.