Discovery of MDI-114215: A Potent and Selective LIMK Inhibitor To Treat Fragile X Syndrome

Abstract

LIMKs are serine/threonine and tyrosine kinases responsible for controlling cytoskeletal dynamics as key regulators of actin stability, ensuring synaptic health through normal synaptic bouton structure and function. However, LIMK1 overactivation results in abnormal dendritic synaptic development that characterizes the pathogenesis of Fragile X Syndrome (FXS). As a result, the development of LIMK inhibitors represents an emerging disease-modifying therapeutic approach for FXS. We report the discovery of MDI-114215 (85), a novel, potent allosteric dual-LIMK1/2 inhibitor that demonstrates exquisite kinome selectivity. 85 reduces phospho-cofilin in mouse brain slices and rescues impaired hippocampal long-term potentiation in brain slices from FXS mice. We also show that LIMK inhibitors are effective in reducing phospho-cofilin levels in iPSC neurons derived from FXS patients, demonstrating 85 to be a potential therapeutic candidate for FXS that could have broad application to neurological disorders or cancers caused by LIMK1/2 overactivation and actin instability.

Figure 1

(a) Chemical structures of selected LIMK1/2 inhibitors. (b) Co-crystal structure of the allosteric inhibitor 7 (TH-300) bound to LIMK2 (PDB: 5NXD); (c) type II inhibitor 8 (TH-470) bound to LIMK1 (PDB: 7B8W, green) and LIMK2 (PDB:7QHG, magenta).

Scheme 1. Synthesis of Tertiary Amide Isomers 9 and 10

Reagents and conditions: (a) PhNH₂, Py, DCM, rt, 3 h, 41%; (b) BnNH₂ or BuNH₂, NaHCO₃, MeOH, rt, 18 h then NaBH₄, 0 °C, 4 h; (c) butyric acid or benzoic acid, HOBt, EDC.HCl, DCM, rt, 30 min then 13 or 14, rt, 18 h, 38–62% (over two steps).

Scheme 2. Synthesis of Reverse Sulfonamide 15 and Amide Analogue 17

Reagents and conditions: (a) N-benzylbutylamine, AlEt₃, 1,2-DCE, 0–80 °C, 18 h, 62–81%.

Scheme 3. Synthesis of Sulfone Derivative 29 and Aromatic Core Analogues 19–24 and 30

Reagents and conditions: (a) HOBt, EDC.HCl, DCM, rt, 30 min then N-benzylbutylamine, rt, 18 h, 78–97%; (b) 26a, K₂S₂O₅, NaHCO₂, Pd(OAc)₂, PPh₃, phen, TBAB, DMSO, 70 °C, 3 h; (c) PhNH₂, NCS, THF, 0 °C, 2 h, 7–24% (over two steps); (d) benzyl bromide, rt, 18 h, 62% (over two steps); (e) 27b, N-benzylbutylamine, T3P, Et₃N, DMF, rt, 1 h, 33%; (f) PhNH₂, THF, rt, 18 h, 66%; (g) LiOH, H₂O/MeOH/THF, rt, 18 h, 95%.

Scheme 4. Synthesis of Substituted N-Phenylsulfonamide Analogues 34–38 and N-Benzylbutylamide Analogues 39–65

Reagents and conditions: (a) SOCl₂, DMF (cat.), 70 °C, 3 h then N-benzylbutylamine, Et₃N, THF, −78 °C, 30 min, 77%; (b) R¹NH₂, THF, rt, 18 h, 17–78%; (c) R³NH₂, NaHCO₃, MeOH, rt, 18 h then NaBH₄, 0 °C, 1–4 h; (d) HOBt, EDC.HCl, DCM, rt, 30 min then 70, rt, 18 h, 5–97% (over two steps); (e) (COCl)₂, DMF (cat.), DCM, rt, 5 h then 70, Et₃N, 0 °C, 1 h, 56–77% (over two steps). Synthesized from methyl 3-(4-((butylamino)methyl)phenyl)propanoate precursor over four steps.

Scheme 5. Synthesis of MDI-65658 (74), MDI-114215 (85) and Their Derivatives

Reagents and conditions: (a) CPrCH₂NH₂, NaHCO₃, MeOH, rt, 18 h then NaBH₄, 0 °C, 1 h; (b) 4-(phenylsulfamoyl)benzoyl chloride, Et₃N, 0 °C, 1 h, 54–77% (over two steps); (c) R⁴NHR⁵, CuI, l-proline, K₂CO₃, DMSO, 80–100 °C, 18 h, 10–51%; (d) LiO^tBu, ethylene glycol, rt, 5 min then 91a, CuI, 110 °C, 18 h, 50%; (e) acrylonitrile, Pd(OAc)₂, NaHCO₃, TBAB, DMF, 110 °C, 4 h then Et₃SiH, Pd–C, MeOH, rt, 24 h, 39%; (f) NaN₃, NH₄Cl, DMF, 120 °C, 18 h, 63%; (g) PhNHMe, THF, rt, 18 h, 57%; (h) (COCl)₂, DMF (cat.), DCM, rt, 5 h then 90a, Et₃N, 0 °C, 1 h, 21% (over two steps).

Figure 2

New structural insights of LIMK1 highlighted by novel compound 85. (a) 85 docked into the homology model of LIMK1 (generated from PDB code 5NXD). Interactions involving key residues are labeled and drawn using dashed lines. The protein surface and nearby allosteric residues have been hidden for clarity. (b) Ligand interaction diagram of 85. Shading represents the following: hydrophobic region (green), charged interaction (positive, blue; negative, red), polar (teal).

Figure 3

Kinome selectivity and safety profiling of 85. (a) Chemical structure of 85 and kinome screen data illustrated using the TREEspot interaction map (DiscoverX). (b) hERG, CYP450 and CEREP panel profiling data of 85.

Figure 4

MDI-114215 (85) was well tolerated in male CD-1 mice (30 mg/kg/day i.p. for 28 days). (a) No significant change in bodyweight was detected between animals dosed with vehicle or 30 mg/kg q.d. i.p. with 85 for 28 days. (b) Food consumption over course of treatment. A small difference (*P < 0.05) in consumption was detected between days 15–22 only but had resolved by the end of the study. (c) Weights of key male sexual organs after 29 days dosing of 85 and, (d) after a further two-week recovery period. Data is represented as mean ± SEM. One control animal was euthanised on day 14 due to convulsions.

Figure 5

MDI-114215 (85) decreases phosphorylated cofilin levels ex vivo and reverses the deficit in LTP of mouse Fmr1 KO hippocampal CA1 pyramidal neurons. (a) Western blot analysis of treated brain slices isolated from young WT or Fmr1 KO (P7–9) showing significant reductions (*P < 0.05, **P < 0.01, ***P < 0.001 determined by one-way ANOVA with Dunnett’s multiple comparisons test) in p-cofilin upon incubation with 3 μM of DMSO control (P7 WT n = 4, Fmr1 KO n = 4), 2 (P7 WT n = 2, Fmr1 KO n = 2), 4 (P7 WT n = 2, Fmr1 KO n = 3), 85 (P7 WT n = 2, Fmr1 KO n = 2). Quantification of p-cofilin to cofilin ratio are also presented. (b–d) Illustrated plots of the field excitatory postsynaptic potential (fEPSP) slope against time (mean ± s.e.m). All fEPSPs were recorded from the hippocampal CA1 dendritic field region of neonatal (P7–9) wild type (WT) and Fmr1 KO mice. LTP was expressed as a percentage of the control normalized mean fEPSP slope and was determined between 50 and 60 min postdelivery of the 4-TBS. For each plot representative traces of fEPSPs obtained at baseline and 55 min after the TBS are shown overlaid.

Figure 6

Dose–response curves showing reduction in p-cofilin by TH-257 (6) in stem cell-derived neurons from two control individuals: (a) KYOU 1 week neurons, (b) AIW002 1 week neurons, and two FXS individuals: (c) FX11–7 and (d) FX8–1. Levels of p-cofilin were measured using the AlphaLISA assay. Data are reported as mean ± SEM of at least 3 independent experiments.