Similarly, compound 6 protected against the LRRK2:G2019SCinduced decreases in MMP in a dose dependent manner where 10 M compound 6 returned MMP dissipation to untreated levels (Figure 6B). both familial and idiopathic PD. A dual inhibitor is preferred over combined, individual JNK and LRRK2 inhibitors because it eliminates complications of drug-drug interactions and the need to optimize individual inhibitor doses for efficacy. The major challenge in developing kinase inhibitors is to gain high selectivity in order to diminish off-target side effects, which is especially important for non-oncogenic targets such as for CNS applications. Some type-II and type-III kinase inhibitors have given high selectivity since these compounds bind to protein pockets that are unique for a specific kinase, such as the allosteric site for type-III inhibitors20 and the hydrophobic pocket occupied originally by the Phe residue in the DFG-in conformation for type-II inhibitors.21 Others, such as BIRB-796 have not been as selective as some type I inhibitors.18 Additionally, AMG 487 application of type-II and type-III inhibitors can be limited because many kinases cannot assume a DFG-out conformation and allosteric binding sites have been discovered for only a few kinases. The majority of kinase inhibitors developed so far are ATP-competitive and their AMG 487 selectivity can be low due to binding in the highly conserved ATP-binding pocket. Despite this, very selective type-I inhibitors have indeed been developed however22C26. Here we present a strategy for identifying small molecule kinase inhibitors that combine the advantages of both type-I inhibitors (for easy access to kinase inhibitors with high affinity) and type-II/III inhibitors (for high selectivity). Specifically, our strategy is to design prepared a series of potent and selective bidentate-binding JNK1/2 inhibitors composed of the 11-mer JIP-peptides and a hinge binder connected through various linkers.27, 28 After optimizations, they were able to reduce the size of the peptide portion from 11-mers to the tri-peptide LNL (compound 1, Figure 1A).27 Compound 1, which utilized a 3-phenyl-indazole as the hinge binder and a moiety of propyl-1,3-diamine coupled with a di-Gly as the linker, still had good potency in both kinase activity Rabbit Polyclonal to OR13H1 assay and JIP displacement assay (Table 1, data were obtained in-house at Scripps). However, compound 1 and its analogs (with longer peptide moieties) are still peptide-like and possess all the major drawbacks associated with peptide-based drugs. Table 1 Enzyme assay data for bidentate inhibitors motif. Sub-pocket-1 bound conserved Leu/Val whereas sub-pocket-2 holds the aliphatic side chain of the Leu/Arg/Lys residues. It is interesting to note that similar hydrophobic interactions occurred between JNK3 and compound 6, although H-bonding interactions were also observed. Thus, this binding motif explains why 6 could displace the JIP-peptide with an IC50 value of 336 nM (Table 1), despite utilizing binding energy from less than half of the JIP binding sites. Open in a separate window Figure 4 Ribbon representation of the crystal structure of JNK3 in complex with AMP-PCP (PDB ID 1JNK) (A). JIP peptide derived from PDB ID 4H39 is placed to show JIP binding pockets, Pocket1, and Pocket2. N-lobe and C-lobe are colored wheat and light blue, respectively. Ligand binding pockets are shown as a transparent surface with yellow and cyan respectively representing hydrophobic and polar regions. Only the boxed area of each protein is shown in panels B and C. Docking poses of compound 6 to the crystal structure of human JNK3 (B) and a homology model of LRRK2 (C) are demonstrated in sticks with the ligand binding pouches overlaid with transparent surface. Key residues involved in H-bond relationships with 6 are labeled. H-bonds are demonstrated as dashed lines. Compound 6 was docked into a AMG 487 homology model of human being LRRK2 by using Phyre2, a web based server to model human being LRRK2 kinase website (Number 4C). The server produced 20 models with 100 % confidence where the sequence alignment between template and query ranged from 22 to 30%. We retrieved six top rated models and assessment of the models showed an average r.m.s.d. of 1 1.1 ? over core kinase domains indicating the similarity among the models. We further inspected the models using PyMol to inspect the hinge binding region and the adjacent surface binding pouches. We chose the top ranked model based on the crystal structure of c-Abl Tyrosine Kinase with the PDB ID of 2FO0 (sequence identity of 23 % with human being LRRK2) for the docking experiment. Importantly, the chosen model structure had hydrophobic pouches similar to the JIP binding site of JNK3. As demonstrated in Number 4C, the indazole moiety bound, as expected, to the hinge with.