Morpholino Analogues of Fingolimod as Novel and Selective S1P1 Ligands with In Vivo Efficacy in a Mouse Model of Experimental Antigen-Induced Encephalomyelitis

Multiple sclerosis (MS) is a chronic, inflammatory, autoimmune disease of the central nervous system (CNS) which is associated with lower life expectancy and disability. The experimental antigen-induced encephalomyelitis (EAE) in mice is a useful animal model of MS, which allows exploring the etiopathogenetic mechanisms and testing novel potential therapeutic drugs. A new therapeutic paradigm for the treatment of MS was introduced in 2010 through the sphingosine 1-phosphate (S1P) analogue fingolimod (FTY720, Gilenya®), which acts as a functional S1P1 antagonist on T lymphocytes to deplete these cells from the blood. In this study, we synthesized two novel structures, ST-1893 and ST-1894, which are derived from fingolimod and chemically feature a morpholine ring in the polar head group. These compounds showed a selective S1P1 activation profile and a sustained S1P1 internalization in cultures of S1P1-overexpressing Chinese hamster ovary (CHO)-K1 cells, consistent with a functional antagonism. In vivo, both compounds induced a profound lymphopenia in mice. Finally, these substances showed efficacy in the EAE model, where they reduced clinical symptoms of the disease, and, on the molecular level, they reduced the T-cell infiltration and several inflammatory mediators in the brain and spinal cord. In summary, these data suggest that S1P1-selective compounds may have an advantage over fingolimod and siponimod, not only in MS but also in other autoimmune diseases.

Life Technologies Limited (Paisley, UK). Fetal bovine serum (FBS) was purchased from PAA Laboratories (Catalogue No. A15-101, Cölbe, Germany). All oligonucleotide primers were from Eurofins Genomics GmbH (Ebersberg, Germany). The chemical synthesis of the starting compounds (1) and (2) has been performed according to previously optimized procedures [1]. Compounds (3) and (4) were synthesised in one-pot two step procedure with amide formation and ring closure. The final compounds ST-1893 and ST-1894 have been obtained after reduction with complex hydrids.

Synthesis of compounds 5 (ST-1893) and 6 (ST
All starting materials have been obtained from Sigma Aldrich and Apollo Scientific and been used without further purification. Analytical thin-layer chromatography (TLC) was carried out on precoated
The reaction mixture was cooled to 0 °C and chloroacetyl chloride (1.2 eq, 2.50 mmol, 210 µL) were added dropwise. The reaction mixture was allowed to stir at RT for 2h, evaporated to dryness and extracted with EtOAc. Organic phase was washed with NaHCO3 (sat.) and water, dried with MgSO4 and evaporated. Crude product was dissolved in THF (20 mL) and 3 eq (6.24 mmol, 250 mg) of 60 % NaH suspension on oil were added. After 21 h at RT additional 5 eq. of NaH were added (

Protein modeling of the S1P1 receptor:
A homologue model of an activated S1P1 receptor was developed, using the protein sequence of S1P1 and corresponding templates. Templates applied for the modeling of S1P1 were X-ray structures deposited in the Protein Data Bank (PDB) under the PDB codes 3V2W and 3V2Y [5,6]. The fusion protein T4L of the X-ray structure [7] was removed and the resulting fragmentary S1P1 protein was used as template to generate a full-length S1P1 protein via the Fasta sequence (UniProt KB -P21453). The The model with the ligand S1P was also examined by MD simulation. The stereochemistry quality aspects of the resulting model were checked using MOE (Ramachandran plot). To generate an activated S1P1 homology model, the intermediate model was fused with Arrestin and the corresponding complex was subjected to a comprehensive MD investigation. Thus, structural stability of the overall model could be achieved. From these, the S1P1 receptor was separated and used as a novel homologous model for docking studies.

Ligand structure optimization and docking:
The initial structures of the ST-1893, FTY720-phosphate and ST-1894 were sketched with Chemdraw Suite 2016 and transferred to a MOE database. The phosphorylated compounds were designed with MOE as 3D structures. All structures were minimized until a root mean square deviation (RMSD) of 0.01 kcal/mol Å was reached. The Energy minimization was performed using the MMFF94 force field option [11] with the restriction to preserve original chirality of the molecules. In Molegro, the docking algorithm called MolDock was used to find the correct poses [2].
No usable results were found with the plant algorithm. The further development of Molegro Plant in the CLC Drug Discovery Workbench 3 (test license) was also used, but here too no valid results could be obtained. In MOE we used the algorithm MOE Dock. The docking protocols for rigid receptor and induced fit were used.

Molecular dynamics (MD):
The best complex for each ligand and receptor was then subjected to MD simulations (Refinement) using Amber 14. A simulation cell was constructed around the models (2 x 7.5 Å larger than the model) with a 7.9 Å real space cut-off for the electrostatic force calculated used the Particle Mesh Ewald method. The pKa values of the ionizable groups were predicted and assigned protonation states based on pH 7.4 (temperature= 298K, density=0.997). The cell was filled with water and the Yamber electrostatic potential was evaluated at all water molecules, the molecules with the lowest or highest potential were turned into sodium or chloride counter ion until the cell was neutral. A short steepest descent minimization was performed to remove severe bumps followed by simulated annealing minimizations at 298 K. Then, MD simulations were run with Amber 14 force field at 298 K and 0.9% NaCl in the simulation cell for 500 ps to refine the models. For further analysis simulation snapshots were captured every 25 ps. For the intermediate and final models, the MD simulation was carried out over a period of 100 ns.

Results:
Table S1: Interaction energies between ligand and the S1P1 receptor.
Interactions energies were calculated by Molegro as described in the supplementary Methods Section. E-inter (protein-ligand) indicates the energy released upon ligand binding and is indicated as kcal/mol. S1A S1B Fig. S1: Flow of the building of the activated S1P1 model: New approach to model development.
(A) After the development of the Standard Model (1), it is fused with the binding protein arrestin (2). This complex is optimized by MD studies until it leads to a stable valid structure. From this complex the S1P1 is cut out and the final model is used (3). (B) Ramachandran plot for the quality control of the models.  into the binding pocket of S1P1.