High Selective Azadipeptide Nitriles Inhibitors For Cathepsin K: Design, Synthesis, Activity Assay And Mechanism Study

Cathepsins have been described as lysosomal proteolytic enzymes due to theirlocalization in the lysosomes. The cysteine protease, which represent the largest andbest characterized group of cathepsins, degrade proteins that have entered thelysosomal system, and consequently play a significant role in pathological orphysiological processes, such as bone remodeling, antigen presentation, osteoporosis,autoimmune disorders, and cancer.Since cathepsins K, B, L and S are involved in several importantpathophysiological conditions, such as tumor growth and invasion, autoimmunediseases, and osteoporosis, they represent promising therapeutic targets.Typically, cathepsin inhibitors are derived from peptides or peptidomimeticstructures, which generally contain electrophilic entities prone to covalently interactwith the cysteine-thiol moiety in enzymes. With the rapid development of the potentand selective inhibitors for human cathepsins, the dipeptide-derived nitriles haveattracted attention in recent years.The selectivity of the inhibitors towards cathepsins is mainly determined by theinteraction of the peptidic residues of the inhibitor (P2, P3), as well as the linkerbetween them, P2-P3, with the nonprimed subsites of each different enzyme. Startingwith these recently described efficient strategies, we envisioned the possibility ofusing these substructures as building blocks to generate new cathepsin inhibitors.We have developed a series of azadipeptide nitriles with different P3groups.After purification and characterization by1H NMR, infrared and mass spectroscopy, all the compounds were tested on human cathepsins K, B, L and S based on thereported and modified enzymatic assay methods. Before that, the enzymatic assay ofthe reference compound against cathepsin K was measured in supplying a Ki=0.024nM, being closed to the reported values (Ki=0.029nM) and validating the modifiedprotocols we used.All the azadipeptide nitrile inhibitor were found to strongly inhibit Cat K and CatL, however, they were more or less inactive against Cat B and Cat S. Thesecompounds were two orders of magnitude more active towards Cat K than Cat B andCat S. On push-pull electronic effect perspective, methoxyl(2) is a electron-donatinggroup, nitro(3) is a electron-withdrawing group and bromine(1) is between the two.3(Ki=0.022nM)<1(Ki=0.027nM)<2(Ki=0.031nM), maybe this is becauseelectron-withdrawing s group make C atom in the CN group more active.One of the important strategies in drug design is the identificationoffluorophilic/fluorophobic environments within the target proteinforfavorable/unfavorable interactions with covalently bound fluorine atoms intheligand. The introduction of trifluoromethyl group in compound4was not toleratedby cathepsin K despite trifluoromethyl is a strong electron-withdrawing group,although it more or less accepted by cathepsins B, L and S.A triaryl meta-phenyl derivative, compound13, in the series was not only apotent inhibitor for cathepsin K (Ki=0.0031nM) but also highly selective over bothcathepsin B and S (~1000-fold).As the inhibition potencies toward other three cathepsins were kept at almost thesame level as other inhibitors, the compound13exhibited an improved selectivityprofile versus the other cathepsins. Of note was that, the present achievement wasintrinsically different from a similar compound19had been reported, where the highselectivity toward cathepsin K over other three cathepsins were achieved somewhatby sacrificing the potency to enzymes (Ki=260pM for compound19). That is, the~1000-fold selectivity of13for cathepsin K over cathepsin B and S here wasachieved essentially by improving the potency of it against cathepsin K (3.1pM vs260pM). To elucidate the binding mode, especially, the location and orientation of the P3group in compound13, we performed molecular docking studies with compound7,11,12and13as representative in the active site of cathepsin K. Covalent docking wasperformed using the GOLD (Genetic Optimization for Ligand Docking)4.12program.From the atom-based scoring, it was observed that the overall predictedGoldScore fitness with cathepsin K was lower for12than that for13, while that for7was higher than11, being consistent with the experimentally obtained inhibitionpotency (Ki) of them toward cathepsin K.However, although the tested inhibition potency was better for7(Ki=0.029nM)than12(Ki=0.124nM) against cathepsin K, the Score fitness of former was relativelylower (26.5141vs27.4432). For all reported inhibitors, a clear correlation ofGoldScore fitness score with the experimental Kimay not exist because the scorevalues are not the actual binding free-energies. Further improvement would beperformed in future.