Investigating bisubstrate inhibitors of PRMT1 and CARM1

Protein arginine methyltransferases (PRMTs) are gaining traction as a novel drug target class for a wide range of diseases. The high sequence conservation within the active sites of the nine human PRMTs, particularly in the co-factor (S-adenosyl methionine [SAM]) binding site, makes the design of isozymeselective PRMT inhibitors challenging. Linking guanidine or substrate peptides to SAM mimics has shown potential as an approach to designing isozymeselective inhibitors, but the peptidic nature of such compounds limits their in vivo utility. Herein, the binding of a novel series of non-peptidic bisubstrate inhibitors to CARM1 and PRMT1 was evaluated. The bisubstrate nature of the inhibitors was confirmed by crystal structures of the inhibitors bound to CARM1. Isothermal titration calorimetry (ITC) and activity assays showed that varying the alkyl linker length between the guanidine and adenosine moieties of the inhibitors tested does not confer selectivity for PRMT1 or CARM1. These findings were supported by co-crystal structures of the inhibitors bound to CARM1, which showed that the guanidinium group is thermally mobile within the CARM1 active site. Conversely, isosteric replacement of the substrate-mimic with an N-alkyl-2- aminopyridine group was found to confer selectivity for CARM1 over PRMT1 (Kd values of 1.1 μM and 43.7 μM, respectively, for the most selective inhibitor). A CARM1 N265Y mutant displayed a significant reduction in affinity for the CARM1-selective aminopyridine inhibitor (judged by ITC), suggesting that this residue, located in the substrate binding channel, may contribute to the CARM1-selectivity of this inhibitor. The substitution of ribose for a morpholino group in PRMT bisubstrate inhibitors was found to abrogate inhibitor binding to PRMT1 and CARM1 in the majority of cases. Additionally, two putative FRET probes designed to bind to PRMT1 and CARM1 were designed and partially synthesised.