Structure–function study of a novel inhibitor of the casein kinase 1 family in Arabidopsis thaliana

Abstract Casein kinase 1 (CK1) is an evolutionarily conserved protein kinase family among eukaryotes. Studies in non‐plants have shown CK1‐dependent divergent biological processes, but the collective knowledge regarding the biological roles of plant CK1 lags far behind other members of the Eukarya. One reason for this is that plants have many more genes encoding CK1 than do animals. To accelerate our understanding of the plant CK1 family, a strong CK1 inhibitor that efficiently inhibits multiple members of the CK1 protein family in vivo (i.e., in planta) is required. Here, we report a novel, specific, and effective CK1 inhibitor in Arabidopsis. Using circadian period‐lengthening activity as an estimation of the CK1 inhibitor effect in vivo, we performed a structure–activity relationship study of analogues of the CK1 inhibitor PHA767491 (1,5,6,7‐tetrahydro‐2‐(4‐pyridinyl)‐4H‐pyrrolo[3,2‐c]pyridin‐4‐one hydrochloride). A propargyl group at the pyrrole nitrogen atom (AMI‐212) or a bromine atom at the pyrrole C3 position (AMI‐23) had stronger CK1 inhibitory activity than PHA767491. A hybrid molecule of AMI‐212 and AMI‐23 (AMI‐331) was about 100‐fold more inhibitory than the parent molecule PHA767491. Affinity proteomics using an AMI‐331 probe showed that the targets of AMI‐331 inhibition are mostly CK1 kinases. As such, AMI‐331 is a potent and selective CK1 inhibitor that shows promise in the research of CK1 in plants.

In the model plant Arabidopsis thaliana (Arabidopsis), CK1 family kinases regulate stomatal closure (Zhao et al., 2016), bluelight signaling (Tan, Dai, Liu, & Xue, 2013), cortical microtubules (Ben-Nissan et al., 2008), and ethylene production (Tan & Xue, 2014). A subset of CK1 proteins in Arabidopsis known as CASEIN KINASE 1 LIKE (CKL) are known to phosphorylate substrate proteins, and phosphorylation by CKLs triggers one of two distinct effects: degradation of substrates through the ubiquitination pathway, or modification of substrate activity. Phosphorylation of CRYPTOCHROME (CRY), a protein involved in the blue-light signaling pathway as mediated by CKL3 and CKL4, is related to CRY degradation (Tan et al., 2013). CKL8 is involved in controlling the degradation of 1-AMINOCYCLOPROPANE-1-CARBOXYLIC ACID SYNTHASE 5 (ACS5) by phosphorylation during ethylene synthesis (Tan & Xue, 2014). CKL2 regulates F-actin disassembly of ACTIN-DEPOLYMERIZING FACTOR 4 (ADF4) by phosphorylation (Zhao et al., 2016). CKL6 controls microtubule dynamics by phosphorylating tubulin (Ben-Nissan et al., 2008). Rice Hybrid breakdown 2 (Hdb2) belongs to the CK1 family and is known to be involved in regulating reproductive isolation or hybrid breakdown (Yamamoto et al., 2010), root development, and hormone sensitivity (Liu, Xu, Luo, & Xue, 2003), although the specific substrates of rice CK1 have not been identified.
Extensive genetic redundancy among multiple members of the CKL subfamily (e.g., the 13 CKLs in Arabidopsis) may make further delineation of the biological processes regulated by the CK1 family challenging because of the difficulty in eliminating kinase function by knocking out or knocking down gene expression of one or combinations of CK1 family genes. To meet this challenge, small molecule inhibitors of CK1 can be employed to determine whether or not CK1 enzymes are involved in a given biological process (Uehara et al., 2019). The small molecule IC261 has mostly been used for this purpose, and more recent studies used PF-670462, which is a more potent and specific inhibitor of plant CK1 enzymes Uehara et al., 2019). Chemical screening combined with target identification of the target molecule indicated that PHA767491, a mammalian CDC7 (Cell division control protein 7) inhibitor, also targets plant CK1 (Uehara et al., 2019)

| In vitro phosphorylation assays of Arabidopsis CKL4
In vitro phosphorylation assays using recombinant CKL4 were performed as described previously (Uehara et al., 2019), with synthetic small molecules. IC261 and PF-670462 were purchased (Sigma-Aldrich catalog numbers I0658 and SML0795, respectively).
PHA767491 was synthesized as previously described (Uehara et al., 2019). All chemical compounds were dissolved in dimethyl sulfoxide (DMSO) as 10 mM stock solutions. Stock solutions were diluted with DMSO to the working concentration and added to assays in kinase reaction buffer (Uehara et al., 2019).

| Synthesis of PHA767491 analogues (AMI molecules)
Synthesis of PHA767491 analogues is described in Supporting information. AMI-331 for basic plant research is now commercially available (Tokyo Chemical Industry, product No. A3352).

| Western blotting
Four-day-old seedlings grown under 12-hr light/12-hr dark (LD) conditions were transferred into a 96-well plate with a dropper.
Seedlings were treated with 20 µl of MS liquid containing 2% sucrose and AMI-331 at 2, 10, or 50 µM with a final concentration of 5% (v/v) DMSO. As a control experiment, MS containing 2% sucrose and 5% DMSO was used to treat the seedlings. Seedlings were kept under constant light (L) or constant dark (D) for 24 hr, harvested, and kept frozen until proteins were extracted. Frozen samples were crushed with zirconia beads (Tomy ZB-50) in a Tissue Lyser II (Qiagen). Detection of PRR5-and TOC1-fusion proteins was performed using a 10%-20% gradient acrylamide gel (198-15041, Wako) as previously described (Nakamichi et al., 2012). Anti-FLAG antibody (F3165, Sigma) and anti-VP antibody (ab4808, Abcam) were used to detect FLAG-fusion and VP-fusion proteins, respectively.

| Screening of proteins bound to AMI-329 beads
Screening for proteins bound to AMI-329 beads was done by a method similar to what has previously been described (Uehara et al., 2019). Briefly, two-week-old seedlings grown under LD con-  (Uehara et al., 2019). Peptides were analyzed with a Q Exactive hybrid quadrupole-orbitrap mass spectrometer (Thermo Fisher Scientific), as described previously (Uehara et al., 2019).
Searches were performed using SEQUEST (Thermo Fisher Scientific) against the Arabidopsis thaliana (TAIR TaxID = 3,702) peptide sequence database.
Once the two technical replicates were shown to have similar results, spectra data of the technical replicates were merged.
Proteins whose digested peptides' spectra were over "2" in the 0 µM AMI-331 sample were selected to ensure data integrity. We then selected proteins whose relative spectra (spectra in 0 µM/summed spectra in 5 and 50 µM) were over "10," as AMI-331-bound proteins ( Figure 5c). To provide overview spectra for potential PHA767491target proteins (Uehara et al., 2019), spectra of these proteins were obtained from AMI-329 bead-bound samples, and relative spectra against 0 µM of AMI-331 samples were shown ( Figure 6).

| Gene expression analysis
Arabidopsis Col-0 seedlings that had been grown under con-

| Accession numbers
Sequence data for the genes described in this article are found in the Arabidopsis Information Resource under following numbers: (At3g30180), and PRR7 (At5g02810).

| Three CK1 inhibitors lengthen the circadian period
The activities of three known CK1 inhibitors were measured in vitro by a previously reported method (Uehara et al., 2019). Recombinant CKL4 kinase, casein, 32 P-ATP, and different concentrations of the inhibitors were combined in a reaction buffer and kept at 37°C for 2 hr. Resulting samples were separated on a polyacrylamide gel by electrophoresis, and 32 P phosphorylation of casein was measured as an indicator of CKL4 kinase activity. IC 50 (half-maximal inhibitory concentration) was determined by calculating the results from at least two independent experiments. IC 50 of IC261, PF-670462, and PHA767491 were 6.7, 0.8, and 5.9 µM, respectively (Figure 1a-c).
The stronger in vitro CK1 inhibitory activity shown by PF-670462 relative to PHA767491 was consistent with previous reports (Uehara et al., 2019).
To estimate CK1 inhibitor activity in vivo, we choose to measure circadian period-lengthening activity in Arabidopsis seedlings. The assay was made more efficient by monitoring the circadian period PF-670462 PHA767491 , which was consistent with previous work (Uehara et al., 2019). Thus, PHA767491 was the strongest in vivo CK1 inhib- itor among the three inhibitors tested. This contrasts with in vitro CK1 inhibitory activity, where PHA767491 was less effective than PF-670462.

| Pyrrole ring derivatives of PHA767491 have strong period-lengthening activities
We sought to create a more potent CK1 inhibitor by modifying the structure of PHA767491, since the in vitro CK1 inhibitor activity of PHA767471 was not as high as PF-670472. We applied previously published synthetic methods to make further derivatives of Relative luminescence Relative luminescence lengthened the period by about 2 and 5 hr, respectively (Figure 2c), or at about one-hundredth the concentration of PHA767491.  (Figures 3c, S2). The effective concentrations of AMI-331 for PRR5 or TOC1 accumulation (10-50 µM)

| AMI-331 has strong CK1 inhibitory activity in vitro
The IC 50 value for CKL4 kinase activity in vitro with AMI-212 treatment was 1.2 µM, and AMI-23, and AMI-331 had IC 50 values 0.7 µM ( Figure 4). The IC 50 for AMI-331 on CKL1 activity was 1.4 µM ( Figure   S3). The IC 50 for AMI-331 was about five times lower than for PHA767491 ( Figure 1). These results suggest that the strong in vitro CK1 inhibitory activity of AMI-331 is responsible for the correlated CK1 inhibitory activity in vivo, based on period-lengthening activity and accumulation of PRR5 and TOC1. However, the extensive period-lengthening activity that results from AMI-331 treatment compared to PHA767491 at one-hundredth the concentration, as well as PRR5 accumulation activity at one-fiftieth the concentration, suggests that there are pharmacological properties of AMI-331 which have not been accounted for that contribute to its strong in vivo period-lengthening activity.

| Target identification of AMI-311
The strong in vitro and in vivo CK1 clock-related AMI-331 inhibitory activity demonstrated in these experiments suggested that clock modulation is by direct inhibition of CK1. However, another possible mechanism for the strong AMI-331 biological activity in vivo is that it targets proteins other than the CK1 family for period lengthening. To test this possibility, we synthesized molecular probes to screen direct target proteins of AMI-331.
Analogues substituted with an alkyl group at the nitrogen atom of the pyrrole ring of PHA767491 retain period-lengthening activity ( Figure 2, Figure S1, Uehara et al., 2019); therefore, an alkyl linker was attached at the pyrrole N of AMI-331, generating AMI-329 ( Figure 5a). AMI-329 retained weak but significant period-lengthening activity (Figure 5a). AMI-329 was then covalently bound to agarose beads and mixed with Arabidopsis seedling protein lysates, with or without AMI-331 (0, 5 or 50 µM) as competitor ( Figure 5b). The resulting peptide spectra showing "1" in the MS analysis may be due to false-positive or background noise. Therefore, we used only proteins with digested peptide spectra that were > "2" in the no-competitor (0 µM) samples to ensure data integrity (Table 1). We further selected proteins from this group with relative spectra (spectra in 0 µM/summed spectra in 5 and 50 µM) were > "10," as AMI-331-bound proteins. These criteria resulted in a set of 23 proteins that included all members of the CK1 family and an additional ten non-CK1 proteins ( Table 1). Spectra of these proteins ranged from 0% to 7% in the  (Table 1). It is noteworthy that kinase AT4G08800 is very similar to the CKL family, but the ATP-binding pocket of mammalian CK1 as determined by its crystal structure (Shinohara et al., 2017) was absent from AT4G08800, as shown in an Araport 11 model (https ://www.arapo rt.org). Therefore, we were not able to conclude that AT4G08800 is a CKL protein, per se, in this study.
Spectra of a reductase C (AT2G41680) and an unknown protein (AT5G42765) in the input fraction were higher than in the 0 µM sample. Collectively, this analysis suggests that AMI-331 is most specific for the CKL family, but that it also has some binding affinities to HYDRA1 (HYD1), LUPEOL SYNTHASE 1 (LUP1), YEAST YAK1-RELATED GENE 1 (YAK1), NRPB3, RNA-binding protein AT3g15010, two possible kinases (AT4G08800 and AT4G34500), and membrane protein AT5G40670.
To examine other potential targets of AMI-331, we analyzed the spectra for PHA767491 target proteins (Uehara et al., 2019). Spectra of ATSK (GSK3) family proteins were "2" to "31" in the 0 µM AMI-331 sample ( Figure S3). These spectra were 14 to 53% in the 5 µM sample, F I G U R E 5 Procedure for target identification of AMI-331. (a) Structure and period-lengthening activity of AMI-329. Asterisks indicate significant period changes compared to control samples (Bonferroni correction p < .05 (*) and 0.01 (**), respectively). (b) Procedure for identification of AMI-331 targets, which should be enriched in "0 µM" compared to "5 µM," "50 µM," since free AMI-331 and AMI-329 beads competitively bind to AMI-331 targets and 0%-13% in the 50 µM sample (Figure 6), suggesting that AMI-331 has some binding affinity for ATSKs. Except for MPK5, spectra of the MPK family were present at a reliable range (2-52 PSMs) in the 0 µM sample ( Figure S4). MPK spectra were 13%-85% in the 5 µM samples, and 0%-31% in the 50 µM samples, suggesting that AMI-331 binds weakly to MPKs (Figure 6). Although CPK6 and CPK26 are possible targets of PHA767491 (Uehara et al., 2019), not all of the CPK family members were enriched in the 0 µM sample compared to the 5 or 50 µM samples, indicating that CPK is not a target of AMI-331 ( Figure 6). Three other protein kinases (AT2G32850, AT3G61160, and AT3G58640), that are candidates as targets of PHA767491, were also enriched in the 0 µM sample compared to 5 and 50 µM samples, suggesting that these proteins are also targets of AMI-331 ( Figure 6).
These quantitative data suggest that although AMI-331 binds to some PHA767491 targets, the specificity of AMI-331 for CK1 family kinases was much greater than PHA767491.

| Effect of AMI-311 on expression of CK1 and ATSK downstream genes
Because the ATSK family was identified by the molecular probe assay as potential targets of AMI-331, even though the affinities between AMI-331 and ATSKs were not as high as between AMI-331 and CK1, was stronger than on GR60ox2 and CPD. These results suggest that AMI-331 targets CKLs most effectively, but also targets ATSKs in vivo (Figure 7c).

| CK1 inhibitory activity of AMI-331 in vivo
In this work, we present a new and potent inhibitor of CK1 family kinases, derived from the lead or seed molecule PHA767491.
Generally, uptake by roots or through above-ground epidermal tissues, solubility, transport in and out of cells, metabolic turnover, and inhibitory activity of targets can affect and restrict the activity of pharmacologically active molecules in plants. One reason for anticipating the strong in vivo inhibitory activity of AMI-331 was its in vitro CK1 inhibitory activity. However, the IC 50 of AMI-331 on CKL4 kinase activity was only sevenfold less than PHA767491, but the period-lengthening activity of AMI-331 was 100 times greater than for PHA767491. In addition, the effective concentrations for increasing PRR5 and TOC1 expression by AMI-331 were about 10-50 µM, or 10-50 times lower than required for PHA767491. These results suggest that factors such as cell membrane permeability, metabolic turnover, and the intracellular location of AMI-331 all may contribute to the strong AMI-331 activity in vivo.

| High selectivity of AMI-331 for CK1
A previous study suggested that PHA767491 targets the CKL family, resulting in clock dysregulation (Uehara et al., 2019). However, PHA767491 can also bind to ATSK, CPK, MPK, and other kinases,  a confounding factor for interpreting the binding selectivity of PHA767491 for CKL or other proteins involved in clock period regulation (Uehara et al., 2019).
In this study, target identification using AMI-329-bound beads suggested that the most specific targets of AMI-331 are in the CK1 family. Addition of 5 µM AMI-331 mostly eliminated binding between AMI-329 beads and CKL proteins (Figures 5 and   6). AMI-331 binds to each CKL member (Table 1), as would be predicted from the similarity of amino acid sequences within the CKL kinase domains. Although AMI-331 allows a greater ability to separate the CK1 family kinases from other kinases for clock activity, in the absence of finer levels of specificity, the dissection of F I G U R E 6 Relative spectra of potential PHA767491 target proteins in AMI-329 bead binding. Relative peptide spectrum matches (PSMs) for CKL family (a), ATSK family (b), MPK family (c), CPK family (d), and the other proteins that are possible targets of PHA767491 (e). Spectra normalized to "0 µM" values are shown from AMI-329 bead binding assays. Although we used only proteins with digested peptide spectra that were > "2" in the no-competitor (0 µM) samples to ensure data integrity in Table 1 to find targets of AMI-331, we used data include whose peptide spectra = 1 in 0 µM samples in Figure 6. Actual PSMs are shown in Figure S4 0 0  (Figure 7). MPKs are not so specifically bound by AMI-331 ( Figure 5). AT2g32850 (protein kinase) and MTK1, which were highly enriched by PHA767491-bound beads, were not enriched by the AMI-329 beads at all. These lines of evidence suggest that the selectivity of AMI-331 is greater than PHA767491 toward the CKL family. Although AMI-331 binds to ATSKs and modulates gene expression downstream of ATSKs (Figures 6 and 7), inhibition of ATSKs by Bikinin did not result in period lengthening (Uehara et al., 2019), suggesting that ATSKs are not involved in clock regulation.
Collectively, the use of the novel, highly potent, and specific CK1 inhibitor AMI-331 makes it possible to propose a new model in which inhibition of only the CKL family, but not ATSKs, MPKs, CPKs, or MTK1, is responsible for circadian period lengthening. Because the circadian clock controls many physiological processes, such as stress responses and flowering time regulation, clock modulators potentially become agricultural regents (Uehara et al., 2019). However, PHA767491, the parent molecule of AMI-331, reportedly inhibits mammalian CK1 and CDC7 proteins that are involved in essential roles in development and DNA replication, respectively (Montagnoli et al., 2008;Uehara et al., 2019). Thus, using AMI-331 itself for agricultural purposes does not seem to hold much promise, unless it is shown that AMI-331

| Possible uses for AMI-331 in plant biology
does not modulate CK1 and CDC7, or other enzymes of nonplant organisms. It is worthy to note that higher concentrations of AMI-331 may also modulate physiological processes through CKL-independent pathways (i.e., off-target effects). Consequently,

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest associated with the work described in this manuscript.

AUTH O R CO NTR I B UTI O N S
JY and NN designed the research plan; ANS and JY synthesized small molecules; HM, AO, and NN performed the experiments; KK performed proteomics analysis; HM, KK, AO, TK, and NN analyzed data; and JY and NN wrote the paper.