Identification of a DYRK1A Inhibitor that Induces Degradation of the Target Kinase using Co-chaperone CDC37 fused with Luciferase nanoKAZ

The protein kinase family includes attractive targets for drug development. Methods for screening of kinase inhibitors remain largely limited to in vitro catalytic assays. It has been shown that ATP-competitive inhibitors antagonize interaction between the target kinase and kinase-specific co-chaperone CDC37 in living cells. Here we show a cell-based method to screen kinase inhibitors using fusion protein of CDC37 with a mutated catalytic 19-kDa component of Oplophorus luciferase, nanoKAZ (CDC37-nanoKAZ). A dual-specificity kinase DYRK1A, an importance of which has been highlighted in Alzheimer’s disease, was targeted in this study. We established 293T cells stably expressing CDC37-nanoKAZ, and analyzed interaction between CDC37-nanoKAZ and DYRK1A. We revealed that DYRK1A interacted with CDC37-nanoKAZ. Importantly, point mutations that affect autophosphorylation strengthened the interaction, thus improving signal/noise ratio of the interaction relative to non-specific binding of CDC37-nanoKAZ. This high signal/noise ratio enabled screening of chemical library that resulted in identification of a potent inhibitor of DYRK1A, named CaNDY. CaNDY induced selective degradation of DYRK1A, and inhibited catalytic activity of recombinant DYRK1A with IC50 value of 7.9 nM by competing with ATP. This method based on a mutant target kinase and a bioluminescence-eliciting co-chaperone CDC37 could be applicable to evaluation and development of inhibitors targeting other kinases.


Supporting Figures
. Development of a bioluminescent co-chaperone CDC37.
(a) Luminescence kinetics of total cell lysate (about 7 ng diluted in 5 µL of HENG buffer) from transfected 293T cells. Points are means ± SD (n = 5).
(b) Amounts of 3xFLAG-DYRK1A and 3xFLAG-DYRK4 bound on an antibody-coated well, as used in Figure 2c. Bound proteins were quantified with HRP-conjugated antibody against FLAG.
Absorbance values are shown. Amounts of bound 3xFLAG-DYRK1A proteins were almost the same between the samples (within ± 2.0%). Bar graphs show means ± SD (n = 8). (c) Amounts of 3xFLAG-DYRK1A and its mutants bound on an antibody-coated well, as used in Figure 2e.
Bound proteins were quantified, and absorbance values are shown. Amounts of bound 3xFLAG-DYRK1A proteins were almost the same between the samples (within ± 3.5%). Bar graphs show means ± SD (n = 5). (d) DYRK1A proteins did not interact with nanoKAZ. The luminescence intensities of the complexes between nanoKAZ and the 3xFLAG-tagged DYRK1A Targeting the DYRK1A/CDC37 complex 4 proteins bound on a 96-well plate coated with antibody against FLAG are shown as fold-changes relative to that with EGFP. Bar graphs show means ± SD (n = 5).
Amounts of 3xFLAG-DYRK1A proteins bound on an antibody-coated well, as used in Figure 3c.
Bound proteins were quantified with HRP-conjugated antibody against FLAG. Absorbance values are shown. Amounts of bound 3xFLAG-DYRK1A proteins were almost the same between the samples (within ± 4.0%). Bar graphs show means ± SD (n = 4).
Targeting the DYRK1A/CDC37 complex 6 Figure S3. DYRK1A inhibitors and a HSP90 inhibitor did not inhibit the luminescent activity of CDC37-nanoKAZ.  293T cells were transiently transfected with an expression vector for 3xFLAG-DYRK1A. At 24 h after transfection, the cells were pre-treated with 2 µM epoxomicin for 1 h to inhibit proteasome irreversibly, and then treated with/without CaNDY (2 µM) for 8 h. The total cell lysates were subjected to SDS-PAGE followed by Western blot analysis.
Targeting the DYRK1A/CDC37 complex  Tables   Table S1. Average of two replicates is shown as percent activity of kinase activity in the presence of CaNDY (1 µM) relative to solvent control (DMSO) DYRK1A, DYRK1B, CLK1, CLK2, and Haspin were inhibited by over 90%. in hertz (Hz). The abbreviations s, d, t and br signify singlet, doublet, triplet, and broad, respectively.