The natural product harmine has been identified as the most effective DYRK1A inhibitor scaffold due to its good selectivity and potent inhibitory effect on DYRK1A [
100,
108]. On this basis, Weber
et al. [
109] developed a variety of new DYRK1A inhibitor scaffolds, which greatly enriched the types of DYRK1A inhibitors. First of all, the selectivity mainly attributes to the interaction between inhibitor and the backbone carbonyl of Leu241. The scaffold was introduced into the thiazole to obtain compound
15 with nanomolar potency (IC
50 = 1.0 nmol/L), unfortunately, the sulfur of
15 blocked further growth at this position. But replacing S atom by N atom to form a pyrazole ring (compound
20 IC
50 = 2.0 nmol/L) breaks this limitation. The five-membered ring is further expanded into a six-membered ring, which is expected to contact the hinge region sterically accommodated to bring better inhibitory effect. The quinoline derivative and benzodioxane derivatives bring IC
50s of 10 and 11 nmol/L, respectively. The above results confirm that the previous design strategy, providing a heteroatom to accept a hydrogen bond from the hinge amide nitrogen of Leu241 to improve selectivity, is correct. More than one amino group on the pyridine is required to form an interaction with the Glu203 to increase the binding affinity to DYRK1A. The addition of the non-polar groups on the five-membered ring seems to maintain a high affinity to DYRK1A, high selectivity
versus CDK9, and potent inhibition of DYRK1A autophosphorylation in the cellular setting (
36 IC
50 = 2 nmol/L,
38 IC
50 = 8 nmol/L), which consolidates the choice of the imidazopyridine core for future optimization. The next optimization of inhibitors focused on the improvement of cellular activity.
46 was obtained by introducing 2-PhO-ethyl into imidazole, which provides the highest cellular inhibitory activity (IC
50 = 35 nmol/L), and strong inhibitory effect against DYRK1A (IC
50 = 1 nmol/L). It is deduced that the 2-PhO-ethyl side chain of
46 sits alongside the imidazo-pyridine core removes this apolar substituent from water, and rigidifies the linker to confine the phenyl to its DYRK1A-bound orientation. Based on this, replacing benzene with pyridine and introducing halogen atoms into pyridine was well tolerated. The compounds with bromine (
56) or chlorine (
57) was well tolerated by DYRK1A giving 2.0 nmol/L-ar or 3.0 nmol/L-ar inhibitors, respectively. In addition, the cellular most potent inhibitory effect of
56 and
57 were observed at 32 and 26 nmol/L, respectively. In the kinase selectivity experiment, DYRK1A and DYRK2 were inhibited to 10% activity by compound
40 at a lower dose, and
40 induced a 52% decrease of hERG activity at 10 µmol/L.
In vitro,
40 have favorable absorption, distribution, metabolism, and excretion (ADME) properties including the complete predicted oral absorption, moderate microsomal stability and a high unbound (free) fraction in plasma.
In vivo, the cytostatic effects of
40 on A270 cell line still exist after stopping treating, which suggests that DYRK1A inhibitors, at least in this ovarian model, shows good anti-tumor activity. But it might not be efficient to induce tumor regressions as single agents. Further work is needed to study whether these DYRK1A inhibitors is effective in tumor by combined with other agents (
Fig. 3).