Selective inhibition reveals the regulatory function of DYRK2 in protein synthesis and calcium entry

The dual-specificity tyrosine phosphorylation-regulated kinase DYRK2 has emerged as a critical regulator of cellular processes. We took a chemical biology approach to gain further insights into its function. We developed C17, a potent small-molecule DYRK2 inhibitor, through multiple rounds of structure-based optimization guided by several co-crystallized structures. C17 displayed an effect on DYRK2 at a single-digit nanomolar IC50 and showed outstanding selectivity for the human kinome containing 467 other human kinases. Using C17 as a chemical probe, we further performed quantitative phosphoproteomic assays and identified several novel DYRK2 targets, including eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1) and stromal interaction molecule 1 (STIM1). DYRK2 phosphorylated 4E-BP1 at multiple sites, and the combined treatment of C17 with AKT and MEK inhibitors showed synergistic 4E-BP1 phosphorylation suppression. The phosphorylation of STIM1 by DYRK2 substantially increased the interaction of STIM1 with the ORAI1 channel, and C17 impeded the store-operated calcium entry process. These studies collectively further expand our understanding of DYRK2 and provide a valuable tool to pinpoint its biological function.

Abstract: The dual-specificity tyrosine phosphorylation-regulated kinase DYRK2 has emerged as a critical regulator of cellular processes. We took a chemical biology approach to gain further insights into its function. We developed C17, a potent small-molecule DYRK2 inhibitor, through multiple rounds of structure-based optimization guided by several co-crystallized structures. C17 displayed an effect on DYRK2 at a single-digit nanomolar IC 50 and showed outstanding selectivity for the human kinome containing 467 other human kinases. Using C17 as a chemical probe, we further performed quantitative phosphoproteomic assays and identified several novel DYRK2 targets, including eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1) and stromal interaction molecule 1 (STIM1). DYRK2 phosphorylated 4E-BP1 at multiple sites, and the combined treatment of C17 with AKT and MEK inhibitors showed synergistic 4E-BP1 phosphorylation suppression. The phosphorylation of STIM1 by DYRK2 substantially increased the interaction of STIM1 with the ORAI1 channel, and C17 impeded the store-operated calcium entry process. These studies collectively further expand our understanding of DYRK2 and provide a valuable tool to pinpoint its biological function.

Introduction
Dual-specificity tyrosine phosphorylation-regulated kinases (DYRKs) belong to the CMGC group of kinases together with other critical human kinases, such as cyclin-dependent kinases (CDKs) and mitogen-activated protein kinases (MAPKs) (Aranda et al., 2011;Becker and Joost, 1999;Manning et al., 2002). DYRKs uniquely phosphorylate tyrosine residues within their activation loops in cis during biosynthesis, although mature proteins display exclusive serine/threonine kinase activities (Lochhead et al., 2005). There are five DYRKs in humans: DYRK1A, DYRK1B, DYRK2, DYRK3, and DYRK4. DYRK1A has been extensively studied due to its potential function in the pathogenesis of Down syndrome and neurodegenerative disorders (Becker and Sippl, 2011;Wegiel et al., 2011). DYRK3 has been shown to function as a central 'dissolvase' to regulate the formation of membraneless organelles (Rai et al., 2018;Wippich et al., 2013). On the other hand, DYRK2 is a crucial regulator of 26S proteasome activity (Guo et al., 2016).
The 26S proteasome degrades the majority of proteins in human cells and plays a central role in many cellular processes, including the regulation of gene expression and cell division (Collins and Goldberg, 2017;Coux et al., 1996). Recent discoveries have revealed that the 26S proteasome is subjected to intricate regulation by reversible phosphorylation (Guo et al., 2017;Guo et al., 2016;Liu et al., 2020). DYRK2 phosphorylates the Rpt3 subunit in the regulatory particle of the proteasome at Thr25, leading to the upregulation of proteasome activity (Guo et al., 2016). DYRK2 is overexpressed in several tumors, including triple-negative breast cancer and multiple myeloma, which are known to rely heavily on proteasome activity for progression, and perturbation of DYRK2 activity impedes cancer cell proliferation and inhibits tumor growth (Banerjee et al., 2018;Banerjee et al., 2019).
Our knowledge of the physiological functions of DYRK2 remains in its infancy, and DYRK2 likely has cellular targets in addition to Rpt3. Substrates of many kinases, especially Ser/Thr kinases, remain insufficiently identified. A major obstacle to discovering physiologically relevant substrates of a kinase is the lack of highly specific chemical probes that allow precise modulation of kinase function. Some DYRK2 inhibitors have been reported; however, these compounds also inhibit other kinases, mostly other DYRK family members, to various degrees (Chaikuad et al., 2016;Jouanne et al., 2017). We have recently identified LDN192960 as a selective DYRK2 inhibitor and showed that LDN192960 could alleviate multiple myeloma and triple-negative breast cancer progression by inhibiting DYRK2mediated proteasome phosphorylation (Banerjee et al., 2019). To obtain even more potent and selective DYRK2 inhibitors, we applied a structure-guided approach to further engineer chemical compounds based on the LDN192960 scaffold. One of the best compounds we generated, compound C17 (C17), displays an effect on DYRK2 at a single-digit nanomolar IC 50 with moderate to excellent selectivity against kinases closely related to DYRK2. Using this potent DYRK2 inhibitor as a tool, we treated U266 cells with C17. We performed quantitative phosphoproteomic analyses, which led to identifying several novel DYRK2 targets, including eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1) and stromal interaction molecule 1 (STIM1). These results demonstrate that DYRK2 plays critical regulatory roles in multiple cellular processes, including protein translation and storeoperated calcium entry, and indicate that C17 can serve as a valuable probe for the study of DYRK2 function. Wei

Results
Structure-based optimization of DYRK2 inhibitors LDN192960 was identified as a DYRK2 inhibitor (Banerjee et al., 2019;Cuny et al., 2010;Cuny et al., 2012). It occupies the ATP-binding pocket of DYRK2 and mediates extensive hydrophobic and hydrogen bond interactions (Banerjee et al., 2019). Nevertheless, LDN192960 also inhibits other DYRK2-related kinases, especially Haspin and DYRK3 (Banerjee et al., 2019). To generate DYRK2 inhibitors with better selectivity, we synthesized a series of new compounds based on the same acridine core structure ( Table 1). The amine side chain was first changed to a protected amine (compounds 1-3), a cyano group (compound 4), or a cyclic amine (compounds 5-6) ( Figure 1-figure supplement 1A, Table 1). Among these candidates, compound 6 exhibited the most potent inhibitory effect towards DYRK2, with an in vitro IC 50 of 17 nM. In comparison, LDN192960 showed an IC 50 of 53 nM when the same protocol was used ( Table 1)-treating HEK293T cells transiently expressing DYRK2 with increasing concentrations of compound 6 efficiently inhibited Rpt3-Thr25 phosphorylation, with the maximal effect observed at an inhibitor concentration of less than 3 μM (Figure 1-figure supplement 1B). Notably, compound 6 also displays good selectivity towards DYRK2 than other kinases, including DRYK1A, DRYK1B, DYRK3, Haspin, and MARK3 (IC 50 values of 889, 697, 121305, 45, and 100 nM, respectively; Table 1). Therefore, compound 6 was chosen as the lead compound for further chemical modification. We subsequently crystallized DYRK2 in complex with compound 6 and determined the structure at a resolution of 2.2 Å ( Figure 1A, . Not surprisingly, compound 6 binds the ATP-binding site of DYRK2 like LDN192960. A water molecule is located deep inside the binding pocket acting as a bridge in the interactions between LDN192960 and the protein. The newly added amino side chain displays apparent densities and adopts an extended conformation. An in-depth analysis of the crystal structure revealed several additional sites for chemical expansion that may further strengthen its interaction with DYRK2 ( Figure 1-figure supplement 2A-B). First, a hydrophilic group can be introduced into the acridine core to functionally replace the aforementioned water molecule and maintain constant contact with DYRK2. Second, a bulky functional group can replace the methoxy groups to mediate other interactions with DYRK2. Finally, the amine side chain can be altered to stabilize its conformation ( Figure 1-figure supplement 2A -B). To this end, we synthesized 9 new compounds (compounds 7-15) and evaluated their inhibitory effects on DYRK2 and related kinases ( Figure 1-figure supplement 2C-D). We also determined the co-crystallized structures of several of these compounds with DYRK2 to visualize their detailed interactions ( Figure 1, Figure 1-figure supplement 3). Compound 7, introducing a hydroxymethyl group to the acridine core, inhibits DRYK2 efficiently as compound 6 while displaying better selectivity against other DRYK family members ( Table 1). The co-crystallized structure shows that the hydroxymethyl group directly contacts the main chain amide group of Ile367 and indirectly coordinates Glu266 and Phe369 via a water molecule ( Figure 1D). Compared to compound 7, compounds 8-10, which contain a carboxyl, aminomethyl, and fluoromethyl group, respectively, instead of a hydroxymethyl group, display reduced inhibition towards DYRK2. Compounds 11-15, designed to replace the methoxy group with a bulkier side chain, showed significantly decreased activity and selectivity and were not further pursued ( Table 1).
Further chemical modification was carried out based on compound 7. By changing the 6-membered ring to a straight-chain or smaller ring, we synthesized compounds 16-19 (Table 1, Figure 1-figure supplement 2E). Among these compounds, C17 with an (S)-3-methylpyrrolidine side-chain exhibited the best potency and selectivity among all the analogs (Table 1). Interestingly, we noticed that compound 18 containing an (R)-3-methylpyrrolidine side chain was not as good as C17, indicating that the chirality of the 3-methylpyrrolidine motif plays an essential role in both potency and selectivity. Further modification of compound 17 (leading to compound 20) to promote further hydrogen bond interactions with DYRK2 failed to improve the inhibitory effect. We also wondered whether acridine was the best aromatic core structure and synthesized two new compounds (compounds 21 and 22) by changing one side of the benzene group to a sulfur-containing thiazole structure (Figure 1-figure supplement 2F), which we thought might facilitate hydrophobic interactions with DYRK2 within the ATP-binding pocket; however, they did not have as effective an inhibitory effect as compound 17 (Table 1) The online version of this article includes the following source data for

C17 is a potent and selective DYRK2 inhibitor
We set to comprehensively characterize the inhibitory function of compound 17 (Figure 2A), referred to as C17 hereafter. In vitro, C17 displays an effect on DYRK2 at a single-digit nanomolar IC 50 value (9 nM) ( Figure 2B, . To further evaluate the selectivity of C17, we performed kinome profiling analyses. Among the 468 human kinases tested, C17 targeted only DYRK2, Haspin, and MARK3 at a concentration of 500 nM ( Figure 2C). Nonetheless, the in vitro IC 50 values of C17 for Haspin and MARK3 (26 nM and 87 nM, respectively) were 3-10-fold higher than that for DYRK2 ( Figure 2B, . Similarly, C17 also inhibited DYRK3 to a lesser extent (IC 50 of 68 nM). In contrast, LDN192960 inhibited DYRK3 and Haspin more than it inhibited DYRK2 (Banerjee et al., 2019;Cuny et al., 2010). Significantly, C17 also efficiently suppressed DYRK2 activity in the cell and abolished Rpt3-Thr25 phosphorylation at an inhibitor concentration of less than 1 μM ( Figure 2D). Taken together, these data demonstrate that C17 is a highly potent and selective DYRK2 inhibitor both in vitro and in vivo.       (1.5 σsurrounding compounds 5, 10, 13, 14, 19, and 20  Acridine derivatives have traditionally been used as antibacterial, antiparasitic, and anticancer agents since these compounds usually show strong DNA intercalating effects (Chaikuad et al., 2016;Jouanne et al., 2017). Considering the potential toxicity of C17 due to its possible DNAbinding capacity, we also assessed the DNA-binding effect of C17 ( Figure 2-figure supplement  2). Isothermal titration calorimetry revealed that C17 (Kd = 22.9 µM) binds to DNA with significantly lower affinity than LDN192960 binds to DNA (Kd = 198 nM), possibly because of the introduction of hydroxymethyl group on the acridine core, which is not present in LDN192960.

DYRK2 substrate profiling by quantitative phosphoproteomic analyses
Quantitative phosphoproteomic approaches have significantly expanded the scope of phosphorylation analysis, enabling the quantification of changes in thousands of phosphorylation sites simultaneously (Álvarez-Salamero et al., 2017). To obtain a comprehensive list of potential DYRK2 targets, we  Source data 1. Raw data of C17 Kinome profiling list for Figure 2C.
Source data 2. Raw data of western blot for Figure 2D.  treated the myeloma U266 cells with C17 and carried out quantitative phosphoproteomic analyses Hogrebe et al., 2018). We prepared lysates of U266 cells treated with C17 or the DMSO control and trypsinized them. Phosphorylated peptides were then enriched using Ti 4+immobilized metal ion affinity chromatography (IMAC) tips and analyzed by LC-MS/MS ( Figure 3A). A total of 15,755 phosphosites were identified, among which 12,818 (81%) were serine, and 2,798 (18%) were threonine. A total of 10,647 (68%) phosphosites were Class I (localization probability >0.75), 2557 (16%) were Class II (0.5 < localization probability ≤0.75), and 2401 (16%) were Class III (0.25 < localization probability ≤0.5) ( Figure 3B). This is a very comprehensive phosphoproteomic dataset prepared for DYRK2 substrate profiling by treating the U266 cells with 10 μM of C17. A good Pearson correlation coefficient of 0.9 was obtained for the phosphosite intensities among the treatment and control samples ( Figure 3-figure supplement 1), and the coefficient of variance of the intensities of the majority of the phosphosites was lower than 20% ( Figure 3-figure supplement 2), demonstrating the high quantification precision of our label-free phosphoproteomic analysis. Remarkably, C17 treatment led to significant downregulation of 373 phosphosites ( Figure 3C), including pThr37 of the eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1), as well as pSer519 and pSer521 in the stromal interaction molecule 1 (STIM1) ( Figure 3D, Figure 3-figure supplement 3). Interestingly, another 445 phosphosites were upregulated ( Figure 3C), suggesting that DYRK2 likely inhibited some downstream kinases or activated phosphatases, and suppressing its activity reversed these effects. Together, these data demonstrate that DYRK2 is involved in a network of phosphorylation events and can directly or indirectly regulate the phosphorylation status of many proteins. The top pathways with which DYRK2 may participate were revealed by a global analysis of the significantly up-and down-regulated phosphoproteins ( Figure 3E).

4E-BP1 is a direct cellular target of DYRK2
We set out to determine whether some of the 373 downregulated phosphosites are genuine DYRK2 targets. We first examined 4E-BP1 for several reasons. First, C17 treatment decreased the pThr37 level in U266 cells ( Figure 3C). Second, a previous study showed that Ser65 and Ser101 in 4E-BP1 can be phosphorylated by DYRK2 in vitro, indicating that 4E-BP1 is a potential DYRK2 substrate (Wang et al., 2003). And lastly, several phosphosite-specific antibodies for 4E-BP1 are commercially available. 4E-BP1 is a master regulator of protein synthesis. It has been well established that its phosphorylation by other kinases such as mTORC1 leads to its dissociation from eukaryotic initiation factor 4E (eIF4E), allowing mRNA translation (Laplante and Sabatini, 2012;Ma et al., 2009).
Using an antibody that detects 4E-BP1 only when it is phosphorylated at Thr37 and Thr46, we found that C17 treatment significantly reduced the level of pThr37/pThr46 of endogenous 4E-BP1 in HEK293T cells ( Figure 4A), consistent with our mass spec analyses in U266 cells. Further investigations using two other 4E-BP1 phosphosite-specific antibodies showed that C17 also decreased the phosphorylation of Ser65 in endogenous 4E-BP1 ( Figure 4A) by a previous study Wang et al., 2003; as well as Thr70. Knockdown of endogenous DYRK2 using a short hairpin RNA (shRNA) also significantly reduced the phosphorylation of these sites ( Figure 4B). Successful knockdown is demonstrated by quantitative RT-PCR analysis (Figure 4-figure supplement 1). Similarly, C17 suppressed DYRK2mediated phosphorylation of 4E-BP1 when overexpressed in the HEK293 cells ( Figure 4C). To ascertain whether DYRK2 can directly phosphate 4E-BP1, we performed an in vitro kinase assay using purified DYRK2 and 4E-BP1 proteins. DYRK2 efficiently phosphorylated 4E-BP1 at multiple sites, including Thr37/Thr46, Ser65, and Thr70, whereas the kinase-deficient DYRK2 mutant (D275N) displayed no activity ( Figure 4D). C17 suppressed the phosphorylation of these sites in a dose-dependent manner ( Figure 4E). These results demonstrate that DYRK2 effectively phosphorylated 4E-BP1 on multiple sites in vivo and in vitro.
4E-BP1 is targeted by multiple kinases (Qin et al., 2016). Indeed, C17 or DYRK2 shRNA decreased but did not abolish the phosphorylation of 4E-BP1 ( Figure 4A and B). A previous study showed that combined inhibition of AKT and MEK kinases suppressed 4E-BP1 phosphorylation and tumor growth (She et al., 2010). We observed similar results when we treated the HEK293A cells with AKTi (an AKT1/AKT2 inhibitor) and PD0325901 (a MEK inhibitor). Significantly, knockdown of DYRK2 in the presence of these compounds further markedly diminished 4E-BP1 phosphorylation ( Figure 4B). To assess whether C17 can also elicit a synergistic effect with these kinase inhibitors, we treated HEK293A, HCT116, and U266 cells with these molecules, either alone or in combination, and examined the  The presence of C17 potentiated the inhibitory effect of AKTi and PD0325901 in all these cells. Together, these results confirm that 4E-BP1 is a direct cellular target of DYRK2 and suggest the potential use of DYRK2 inhibitors in combination with other kinase inhibitors for cancer therapy.

DYRK2 promotes STIM1-ORAI1 interaction to modulate SOCE
In addition to 4E-BP1, another potential target of DYRK2 is STIM1, as the phosphorylation levels of both Ser519 and Ser521 in endogenous STIM1 were significantly reduced upon DYRK2 inhibition in our mass spectrometry analyses ( Figure 3C). STIM1 is a single-pass transmembrane protein residing in the endoplasmic reticulum (ER) and plays a vital role in the store-operated calcium entry (SOCE) process (Collins et al., 2013). The luminal domain of STIM1 senses calcium depletion in the ER and induces protein oligomerization and puncta formation (Liou et al., 2005;Prakriya and Lewis, 2015;Zheng et al., 2018). Oligomerized STIM1 then travels to the ER-plasma membrane contact site and activates the ORAI1 calcium channel. The cytosolic region of STIM1 contains multiple phosphorylation sites, and it has been shown that the function of STIM1 is regulated by several kinases, including ERK1/2 .
Purified wild-type DYRK2, but not the kinase-dead mutant D275N, induced mobility changes of the cytosolic region of STIM1 (STIM1 235-END ) in SDS-PAGE gel ( Figure 5A). As increasing amounts of DYRK2 lead to greater shifts of STIM1 235-END , there are likely multiple DYRK2 phosphorylation sites in STIM1. Consistently, DYRK2 induced a mobility shift of STIM1 when they were co-expressed in the HEK293A cells ( Figure 5B). To further pinpoint DYRK2-specific phosphorylation sites, we co-expressed DYRK2, Orai1, and STIM1 in HEK293A cells, treated the cells with C17, isolated STIM1 using Anti-FLAG agarose, and then subjected it to label-free quantitative mass spectrometry analyses. The phosphorylation levels of at least eight phosphosites on four peptides of STIM1 were significantly reduced upon treatment with C17 compared with the untreated sample ( Figure 5-figure supplement 1A-B), including Ser519 and Ser521 that were identified in the U266 phosphoproteome analysis ( Figure 3C). In a separate mass spec experiment, phosphorylation of Ser608 and Ser616 were also reduced by C17. Together, these results demonstrate that DYRK2 can phosphorylate multiple sites in the cytosolic region of STIM1. STIM1 puncta formation indicates its oligomerization and activation (Liou et al., 2005;Prakriya and Lewis, 2015;Zheng et al., 2018). To assess the functional outcome of STIM1 phosphorylation by DYRK2, we co-expressed STIM1 and DYRK2 in an Orai-deficient (Orai-KO) cell line, which has all three Orai genes genetically ablated (Zheng et al., 2018). DYRK2 induced the appearance of STIM1 puncta under resting conditions, indicating that DYRK2 promotes STIM1 oligomerization ( Figure 5C). In contrast, the STIM1 puncta were not observed in the presence of C17. DYRK2 also failed to promote the punctate formation of STIM1-10M, a STIM1 variant with all ten potential DYRK2 phosphorylation sites mutated to Ala ( Figure  The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. Raw data of the significantly up-and down-regulated phosphosites after U266 cells treated with C17 for Figure 3C.        To further understand the importance of STIM1 phosphorylation, we examined the interaction between STIM1 and Orai1 using co-immunoprecipitation. Expression of WT DYRK2 significantly increased the interaction between STIM1 and Orai1, whereas expression of DYRK2-KD exerted no such effect ( Figure 5D). Treating cells with C17 effectively abolished the DYRK2-dependent STIM1-Orai1 interaction. Notably, both STIM1-1-491, a C-terminal truncated STIM1 ( Figure 5C, Figure 5figure supplement 1C), and STIM1-10M displayed significantly reduced interaction with Orai1 even in the presence of WT DYRK2 ( Figure 5E), suggesting that DYRK2-mediated phosphorylation is essential to promote the binding between STIM1 and Orai1. C17 also decreased the interaction between STIM1 and Orai1 without exogenously expressing DYRK2 ( Figure 5F).
We examined fluorescence resonance energy transfer (FRET) between STIM1-YFP and CFP-Orai1 to validate the regulatory function of DYRK2 on STIM1-Orai1 interaction. The FRET signals between STIM1-YFP and CFP-Orai1 were significantly increased in HEK293 cells in the presence of WT DYRK2 ( Figure 5G). To exclude the influence of endogenous STIM1, we performed further analyses in a STIM1-STIM2 DKO cell line (Zheng et al., 2018). The FRET signals between STIM1-1-491 and Orai1 were unaltered by DYRK2 ( Figure 5H), indicating that the effect of DYRK2 is dependent on the C-terminal region of STIM1. Furthermore, the FRET signals between STIM1-10M and Orai1 were unaffected by DYRK2 ( Figure 5I). These results are consistent with the co-immunoprecipitation results and demonstrate that DYRK2 can promote the STIM1-Orai1 interaction via STIM1 phosphorylation.
Lastly, to examine the physiological relevance of the STIM1-Orai1 interaction regulated by DYRK2, we performed SOCE analyses in HEK293A cells expressing GCaMP6f, a genetically encoded calcium sensor (Nakai et al., 2001). Treating cells grown in a calcium-free medium containing thapsigargin resulted in a transient increase in GCaMP6f fluorescence due to calcium release from the ER to the cytosol ( Figure 5J, black line). Subsequent addition of calcium to the cell culture medium resulted in a marked increase in GCaMP6f signaling, indicating calcium entry into the cells, further augmented by STIM1 overexpression ( Figure 5J, blue line). Pre-treating cells with C17 for 1 hr substantially reduced SOCE in cells with either endogenous ( Figure 5J, green line) or overexpressed STIM1 ( Figure 5J, red line). Quantifications of these results are present in Figure 5K. Taken together, our results strongly suggest that DRYK2 can directly enhance SOCE by phosphorylating STIM1 and promoting its interaction with ORAI1, which can all be effectively inhibited by C17.

Discussion
We used a structure-based approach to design, synthesize and evaluate a series of new analogs based on the acridine core structure and eventually identified C17 as a potent and selective DYRK2 inhibitor. We showed that C17 affects DYRK2 at a single-digit nanomolar IC50 and inhibits DYRK2 more potently than closely related kinases such as DYRK3, Haspin, and MARK3. The crystal structure of C17 for 1 hr. The cells were lysed, and immunoblotting was carried out with indicated antibodies. (D) DYRK2 directly phosphorylated 4E-BP1 at multiple sites. (E) C17 inhibited DYRK2-mediated 4E-BP1 phosphorylation in a concentration-dependent manner. (F-H) C17 displayed a synergistic effect with AKT and MEK inhibitors to suppress 4E-BP1 phosphorylation in HEK293A (F), HCT116 (G), and U266 cells. (H) The cells were treated with indicated concentrations of PD032590, AKTi-1/2, and C17 alone or in combination for 1 hr. Cell lysates were immunoblotted with indicated antibodies.
The online version of this article includes the following source data and figure supplement(s) for figure 4: Source data 1. Raw data of Western blot for Figure 4A, B.
Source data 2. Raw data of western blot for Figure 4C, D.
Source data 3. Raw data of western blot for Figure 4E, F.
Source data 4. Raw data of western blot for Figure 4G, H.    of DYRK2 bound to C17 revealed critical interactions that explain its high selectivity, including a hydrogen bond between the (S)-3-methylpyrrolidine ring and Glu352 in DYRK2. C17 provided us with a unique tool to interrogate the physiological functions of DYRK2. We treated U266 cells with C17 and performed quantitative phosphoproteomic analyses. We found that the cellular phosphorylation pattern is significantly altered by C17, suggesting that DYRK2 likely has multiple cellular targets and is involved in a network of biological processes. We then identified several leading phosphosites that are downregulated and demonstrated that 4E-BP1 and STIM1 are bona fide substrates of DYRK2. We showed that DYRK2 efficiently phosphorylated 4E-BP1 at multiple sites, including Thr37, and combined treatment of C17 with AKT and MEK inhibitors resulted in marked suppression of 4E-BP1 phosphorylation. Therefore, DYRK2 likely functions synergistically with other kinases to regulate protein synthesis.
For the first time, we also discovered that DYRK2 could efficiently phosphorylate STIM1, and phosphorylation of STIM1 by DYRK2 substantially increased the interaction between STIM1 and ORAI1. Treating cells with C17 suppressed SOCE, validating the critical role of DYRK2 in regulating calcium entry into cells. These data allow us to present a hypothetical model showing how DYRK2 triggers the activation of STIM1 ( Figure 5L). Under resting conditions, the cytosolic portion of STIM1 likely adopts an inactive conformation. DYRK2 can phosphorylate STIM1 and induce its oligomerization, which then interacts with the Orai1 channel and leads to its opening. One inadequacy of our study is the lack of further insight into the regulation mechanism of this process. In particular, what is the upstream signal that triggers DYRK2 activation? Nevertheless, our data offer a valuable model that allows further investigation of the relationship between DYRK2 and SOCE.
Source data 6. Raw data of Store-operated Ca2+ entry (SOCE) analyses for Figure 5J.   processes, and the selective DYRK2 inhibitor we developed here will serve as a valuable tool in dissecting its complex downstream pathways.

Cell culture, transfection, and infection
Mammalian cells were all grown in a humidified incubator with 5% CO 2 at 37 °C. HEK293T (RRID:CVCL_0063), HEK293A (Thermo Fisher, R70507), and HEK293 (RRID:CVCL_0045) cells were grown in Dulbecco's Modified Eagle Media (DMEM, Gibco) supplemented with 10% FBS, 4 mM L-glutamine, 100 U/mL penicillin, and 100 mg/mL streptomycin (Gibco). U266 (RRID:CVCL_0566) cells were grown in RPMI 1640 (Gibco) supplemented with 10% FBS, 4 mM L-glutamine, 100 U/mL penicillin, and 100 mg/mL streptomycin (Gibco). HCT116 cells (China Infrastructure of Cell Line Resources, 1101HUM-PUMC000158) were grown in Iscove's Modified Dulbecco's Medium (IMDM, Gibco) supplemented with 10% FBS, 4 mM L-glutamine, 100 U/mL penicillin, and 100 mg/mL streptomycin (Gibco). All cell lines were confirmed by STR (short tandem repeat) profiling and tested negative for mycoplasma contamination. All cell lines are not in the list of commonly misidentified cell lines maintained by the International Cell Line Authentication Committee (version 11). Transient transfection of HEK293T, HEK293A cells were carried out using Lipofectamine 2000 (Thermo Fisher Scientific) or X-tremeGENE 9 DNA Transfection reagent (Roche) as recommended by the manufacturer, and transfected cells were used in experiments 24-48 hr later. In Lipofectamine transfection, the cells were cultured to ~70-80% confluency in 10 cm dishes, followed by transfection with 10-12 μg plasmid. The cells were changed with fresh DMEM after 12 hr and incubated for 36 hr before further experiments. In X-treme-GENE 9 DNA transfection, the cells were cultured to ~50 confluency in 35 mm glass bottom dishes coated with poly-D-lysine, followed by transfection with 1-3 μg plasmid. produced using the psPAX2 and pMD2.G packaging vectors. Viral media were passed through a prewetted 0.45 μm filter and mixed with 10 μg mL -1 Polybrene (Sigma) before being added to recipient cells. Infected cells were selected with puromycin (1-2 μg mL -1 , Life Technologies) to generate stable populations.
DYRK2 protein purification and co-crystallization DYRK2 208-552 with an N-terminal 6 × His affinity tag and TEV protease cleavage site which expressed in E. coli BL21 (DE3). Bacterial cultures were grown at 37 °C in LB medium to an OD600 of 0.6-0.8 before induced with 0.5 mM IPTG overnight at 18 °C. Cells were collected by centrifugation and frozen at -80 °C. For protein purification, the cells were suspended in the lysis buffer (50 mM HEPES, pH 7.5, 500 mM NaCl, 20 mM imidazole, 5% glycerol, 5 mM β-mercaptoethanol, and 1 mM phenylmethanesulfonylfluoride) and disrupted by sonication. The insoluble debris was removed by centrifugation. The supernatant was applied to a Ni-NTA column (GE Healthcare). The column was washed extensively with the wash buffer (50 mM HEPES, pH 7.5, 500 mM NaCl, 50 mM imidazole, 5% glycerol, and 5 mM β-mercaptoethanol) and bound DYRK2 protein was eluted using the elution buffer (50 mM HEPES, pH 7.5, 500 mM NaCl, 500 mM imidazole, 5% glycerol, and 5 mM β -mercaptoethanol). After cleavage with TEV protease, the protein sample was passed through a second Ni-NTA column to separate untagged DYRK2 from the uncut protein and the protease. Final purification was performed using a Superdex 200 gel filtration column (GE Healthcare), and the protein was eluted using the final buffer (25 mM HEPES, pH 7.5, 400 mM NaCl, 1 mM DTT, and 5% glycerol). Purify the DYRK2-D275N using the same method as shown above. Purified DYRK2 and DYRK2-D275N were concentrated to 10 mg mL -1 and flash-frozen with liquid nitrogen. DYRK2 208-552 was incubated with 200 µM compounds on ice before crystallization. The proteincompounds mixture was then mixed in a 1:1 ratio with the crystallization solution (0.36 M-0.5 M sodium citrate tribasic dihydrate, 0.01 M sodium borate, pH 7.5-9.5) in a final drop size of 2 µl. The DYRK2-compounds crystals were grown at 18 °C by the sitting-drop vapor diffusion method. Cuboidshaped crystals appeared after 4-7 days. Crystals were cryoprotected in the crystallization solution supplemented with 35% glycerol before frozen in liquid nitrogen.
The X-ray diffraction data were collected at Shanghai Synchrotron Radiation Facility (SSRF) beamline BL17U. The diffraction data were indexed, integrated, and scaled using HKL-2000 (HKL Research). The structure was determined by molecular replacement using the published DYRK2 structure (PDB ID: 3K2L) (Soundararajan et al., 2013) as the search model using the Phaser program (McCoy et al., 2007). Chembiodraw (version 13.0) was used to generated the. cif files for compounds, and then compounds were fitted using the LigandFit program in Phenix (Adams et al., 2010). The structural model was further adjusted in Coot (Emsley et al., 2010) and refined using Phenix. The quality of the structural model was checked using the MolProbity program in Phenix. The crystallographic data and refinement statistics are summarized in Figure 1-source data 1.

KINOMEscan kinase profiling
The KINOMEscan kinase profiling assay was carried out at The Largest Kinase Assay Panel in the world for Protein Kinase Profiling (https://www.discoverx.com). C17 kinase selectivity was determined against a panel of 468 protein kinases. Results are presented as a percentage of kinase activity in DMSO control reactions. Protein kinases were assayed in vitro with 500 nM final concentration of C17 and the results are presented as an average of triplicate reactions ± SD or in the form of comparative histograms.

Quantitative phosphoproteomic analysis
Triplicate U266 cells treated with/without C17 were lysed by the lysis buffer containing 1% (v/v) Triton X-100, 7 M Urea, 50 mM Tris-HCl, pH 8.5, 1 mM pervanadate, protease inhibitor mixture (Roche), and phosphatase inhibitor mixtures (Roche). The cell lysates were firstly digested with trypsin (Promega, USA) by the in-solution digestion method . After desalting, the Ti 4+ -IMAC tip was used to purify the phosphopeptides. The phosphopeptides were desalted by the C18 StageTip prior to the LC MS/MS analysis . An Easy-nLC 1200 system coupled with the Q-Exactive HF-X mass spectrometer (Thermo Fisher Scientific, USA) was used to analyze the phosphopeptide samples with 1 hr LC gradient. The raw files were searched against Human fasta database (71,772 protein entries, downloaded from Uniprot on March 27, 2018) by MaxQuant (version 1.5.5.1). The oxidation (M), deamidation (NQ), and Phospho (STY) were selected as the variable modifications for the phosphopeptide identification, while the carbamidomethyl was set as the fixed modification. The false discovery rate (FDR) was set to 0.01 on PTM site, peptide, and protein level. Label-free quantification (LFQ) and match between runs were set for the triplicate analysis data. The MaxQuant searching file 'Phospho ( STY) Sites. txt' was loaded into the Perseus software (version 1.5.5.3) to make volcano plots using student's t-test and cutoff of 'FDR < 0.05 and S0 = 2'. The pathway analysis was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database with cutoff of adjusted p-value < 0.05.

Quantitative RT-PCR
Total RNA from cells was extracted using the RNeasy Mini Kit (Qiagen) and reverse-transcribed with the PrimeScript Real Time reagent Kit (with genomic DNA Eraser, TAKARA). The product of reverse transcription was diluted five times then subjected to quantitative rtPCR reaction in Applied Biosystems ViATM7 Real-Time PCR System (Applied Biosvstems). The 20 μl quantitative rtPCR reaction contained 2 μl of the reverse-transcription reaction mixture, 2 × Hieff quantitative rtPCR SYBR Green Master Mix (Yeasen), 0.2 μM quantitative rtPCR forward primer, 0.2 μM quantitative rtPCR reverse primer (Figure 4-figure supplement 2)

STIM1 and 4EBP1 protein purification
The cytosolic domain of STIM1 (bases 235-END) with an N-terminal GST-tag and TEV protease cleavage site which expressed in E. coli BL21 (DE3). Bacterial cultures were grown at 37 °C in LB medium to an OD600 of 0.6-0.8 before induced with 0.5 mM IPTG overnight at 18 °C. Cells were collected by centrifugation and frozen at -80 °C. For protein purification, the cells were suspended in the lysis buffer 50 mM Tris-HCl (pH 7.5), 500 mM NaCl, 5 mM β-mercaptoethanol, and 1 mM phenylmethanesulfonylfluoride and disrupted by sonication. The insoluble debris was removed by centrifugation. The supernatant was applied to a glutathione-Sepharose column (GE Healthcare) and eluted in lysis buffer containing 20 mM glutathione. Purify the GST-4EBP1 using the same method as shown above. Purified STIM1 and 4EBP1 were flash-frozen with liquid nitrogen. In vitro kinase assays DYRK2 kinase assays were performed in 50 mM HEPES, pH 7.5, 100 mM NaCl, 10 mM MnCl 2 , 10 mM ATP using STIM1 or 4EBP1 as substrate. The kinase reactions were initiated by the addition of DYRK2 with indicated concentration. Assays (25 μl volume) were carried out at 30 °C for 30 min, and terminated by addition of SDS-PAGE buffer containing 20 mM EDTA and then boiled. The reaction mixtures were then separated by SDS-PAGE and visualized by Coomassie Blue staining or analyzed by immunoblot using primary antibodies as indicated throughout.

Co-immunoprecipitation and western blotting
HEK293A cells were cultured and transfected as described above. After transfection, the cells were washed three times with Ca 2+ -free buffer containing 10 mM HEPES, 10 mM D-glucose, 150 mM NaCl, 4 mM KCl, 3 mM MgCl 2 and 0.  . Membranes were blocked in 5% non-fat milk solution in TBS-T for 1 hr at room temperature, incubated with indicated primary antibodies overnight at 4 °C, and incubated with secondary antibodies (horseradish peroxidase-linked anti-mouse or anti-rabbit) for 1 hr at room temperature. Blots were developed with AMERSHAM ImageQuant 800 (GE Healthcare) and exposed to film.

Quantitative analysis of phosphorylation sites on STIM1
Triplicate HEK293A cells co-transfected with GFP-ORAI1, FLAG-DRYK2 and FLAG-STIM1 for 36 hr was treated with 1 μM and 10 μM C17 respectively for 1 hr. After collected, cells were washed with the Ca 2+ -free buffer to remove excess Ca 2+ and then lysed by the lysis buffer. For immunoprecipitations, lysates containing equal protein amounts were incubated with FLAG-beads for 1 hr at 4 °C, which were washed three times with the lysis buffer afterwards. The proteins were eluted from the FLAG-beads by addition of 500 μg FLAG peptides (Smart Lifesciences). Then the eluted proteins were digested with trypsin by the FASP digestion method (Wiśniewski et al., 2009). The peptides were analyzed on a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) with 1 hr LC gradient. The raw files were searched against Human fasta database (downloaded from Uniprot) by MaxQuant (version 1.6.3.4). The oxidation (M), deamidation (NQ), and Phospho (STY) were selected as the variable modifications for the phosphopeptide identification, while the carbamidomethyl was set as the fixed modification. The false discovery rate (FDR) was set to 0.01 on PTM site, peptide, and protein level. Label-free quantification (LFQ) and match between runs were set for the triplicate analysis data.

Fluorescence imaging
Fluorescence signals were recorded using a ZEISS obersever-7 microscope equipped with an X-Cite 120-Q (Lumen dynamics), brightline filter sets (Semrock Inc), a 40 × oil objective (NA = 1.30), and a Prime 95B Scientific CMOS (sCMOS) camera (Teledyne Imaging). This system was controlled by Slide book6 software (3i). For fluorescence resonance energy transfer (FRET) measurements, three-channelcorrected FRET include cyan fluorescent protein (CFP), yellow fluorescent protein(YFP) and FRET raw were collected with corresponding filters, F CFP (438 ± 12E x / 483 ± 16 Em ), F YFP (510 ± 10E x /542 ± 13.5 Em ) and FRET raw (438 ± 12E x /542 ± 13.5 Em ), every 10 s. Calibration of bleed through from FRET donor or acceptor to FRET channel (0.20, and 0.36, correspondingly), as well as the system-dependent factor, G (2.473) were done as described earlier (Ma et al., 2015). These parameters were then used to generate calculate FRET efficiency (Eapp) values from raw fluorescent signals, similar to those previously described (Ma et al., 2017). At least 16 cells were collected for each repeat. Corresponding results were calculated with Matlab 2014a software and plotted with GraphPad Prism 8.4.0 software. Representative traces of at least three independent experiments are shown as mean ± SEM.

Confocal imaging and intracellular Ca 2+ measurement
Intracellular Ca 2+ measurement was performed on a Zeiss LSM 700 laser scanning confocal microscope equipped with a 63 × oil immersion objective lens (N.A. = 1.4) controlled by ZEN software. GCaMP6f fluorescence was excited using a 488 nm line of solid-state laser and fluorescence emission was collected with a 490-to 555 nm band-pass filter; mCherry fluorescence was excited using a 555 nm line of solid-state laser and fluorescence emission was collected with a 580 nm long-pass filter. Two high-sensitivity PMTs were used for detection. Cells were imaged at 10 s intervals for up to 20 mins. All live cell imaging experiments were performed at room temperature. Data were processed and analyzed using Zen and ImageJ software. For intracellular Ca 2+ measurement, HEK293A cells were plated on glass-bottom 35 mm dishes and transfected as described above. Cells were washed with Ca 2+ free buffer 3 times 24 hr after transfection. For DYRK2 inhibition, cells were treated with DMEM containing 1 μM of C17 at 37 for 1 hr before Ca 2+ free buffer rinse. Depletion of Ca 2+ -stores was triggered by incubating cells with 1 μM thapsigargin in Ca 2+ -free buffer, and Store-operated Ca 2+ entry (SOCE) was induced by addition of 2 mM CaCl 2 to thapsigargin containing buffer. One μM C17 was added for DYRK2 inhibition assay. The intracellular free calcium concentration was measured by monitoring the fold change of GCaMP6f fluorescence, the data were shown as the mean ± SEM.

Statistics and data presentation
Most experiments were repeated three times with multiple technical replicates to be eligible for the indicated statistical analyses, and representative image has been shown. All results are presented as mean ± SD unless otherwise mentioned. Data were analysed using Graphpad Prism 8.4.0 statistical package.
The following dataset was generated: Author ( General information for chemical synthesis NMR spectra were recorded on a Varian 400 MHz spectrometer, Bruker 400 MHz NMR spectrometer (ARX400), Bruker 400 MHz NMR spectrometer (AVANCE III), Bruker-500M Hz NMR spectrometer (500 M) and Bruker-600M Hz NMR spectrometer (600 M) at ambient temperature with CDCl 3 as the solvent unless otherwise stated. Chemical shifts are reported in parts per million relative to CDCl 3 (1 H, δ 7.26; 13 C, δ 77.16) and MeOD (1 H, δ 3.31; 13 C, δ 49.00). Data for 1 H NMR are reported as follows: chemical shift, integration, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, quint = quintet, sixt = sixtet, m = multiplet) and coupling constants. High-resolution mass spectra were obtained at Peking University Mass Spectrometry Laboratory using a Bruker APEX Flash chromatography. The samples were analyzed by UPLC/MS on a Waters Auto Purification LC/ MS system (Waters C18 5 μm 150 × 4.6 mm SunFire separation column) or prepared by HPLC/ MS on a Waters Auto Purification LC/MS system (ACQUITY UPLC BEH C18 17 μm 2.1 × 50 mm column). Analytical thin layer chromatography was performed using 0.25 mm silica gel 60 F plates, using 250 nm UV light as the visualizing agent and a solution of phosphomolybdic acid and heat as developing agents. Flash chromatography was performed using 200-400 mesh silica gel. Yields refer to chromatographically pure materials, unless otherwise stated. All reactions were carried out in oven-dried glassware under an argon atmosphere unless otherwise noted. Synthetic Procedures: Appendix 1-scheme 1. Synthesis route of C1-C4.
9-chloro-2,7-dimethoxyacridine (1b). Compound 1 a (2.58 g, 9.45 mmol) was added in a sealed tube, and 30 mL of phosphorus oxychloride was added under argon atmosphere. The reaction was heated at 130 °C for 8 h. The resulting slurry was poured onto ice with vigorous stirring, and a large amount of a yellow solid was precipitated. After filtration, the precipitate was washed with water and dried to give compound 1b (2.58 g, quant.) as an orange solid. 1b was used in next step without further purification.

Compounds 2-4
By employment of the above-described procedure, starting from 1b and using suitable bromides, compounds 2-4 were prepared.

Compounds 6 a
By employment of the above-described procedure, starting from 1b and using suitable bromide, compound 6 a were prepared.

2,7-bis(benzyloxy)-9-chloroacridine (15 a)
To a solution of 1d (94.5 mg, 0.364 mmol) in 10 mL of anhydrous DMF, potassium carbonate (150.8 mg, 1.092 mmol) silver oxide (253.0 mg, 1.092 mmol) and benzyl bromide (100 μl) was added. The reaction was allowed to react at 40 °C until full conversion. The reaction solution was spin-dried under reduced pressure, and water / dichloromethane was separated. The solvent was then evaporated and the residue was dissolved with dichloromethane and washed with water. The combined organic extracts were dried over anhydrous Na 2 SO 4 and concentrated in vacuo. The residue was purified by chromatography on a silica gel column (Petroleum ether / ethyl acetate = 9/1) to give 15 a (163.9 mg, quant.) as a yellow solid. 1  24' (150 mg, 0.66 mmol) in 1 mL of anhydrous tetrahydrofuran. The reaction was performed at -60 °C for 1.5 h and then slowly warmed up to room temperature and reacted overnight. The reaction was quenched by adding saturated ammonium chloride, diluted with dichloromethane, washed twice with brine, dried over anhydrous Na 2 SO 4 and concentrated in vacuo. The residue was purified by chromatography on a silica gel column (pure dichloromethane) to give compound 22 a (69.2 mg, 28%) as a yellow solid. 1 1, 163.9, 154.3, 153.6, 147.0, 137.0, 125.6, 121.0, 117.9, 117.4, 115.8, 61.1, 55.9, 52.7, 17.7, 14.7 Methyl-9-chloro-7-methoxy-2-(methylthio)thiazolo[5,4-b]quinoline-5carboxylate (22b) Compound 22 a (50 mg, 0.14 mmol) was added in a sealed tube, and 1 mL of phosphorus oxychloride was added under argon atmosphere. The reaction was heated at 130 °C overnight. The resulting slurry was poured onto ice with vigorous stirring to make full quenching. The mixture was diluted with dichloromethane, washed twice with brine, dried over anhydrous Na 2 SO 4 and concentrated in vacuo. The residue was purified by chromatography on a silica gel column (pure dichloromethane) to give compound 22b (15 mg, 32%) as a light yellow solid. 1