DYRK3 activation, engagement of protein kinase A/cAMP response element-binding protein, and modulation of progenitor cell survival.

DYRKs are a new family of dual-specificity tyrosine-regulated kinases with emerging roles in cell growth and development. Recently, we discovered that DYRK3 is expressed primarily in erythroid progenitor cells and modulates late erythropoiesis. We now describe 1) roles for the DYRK3 YTY signature motif in kinase activation, 2) the coupling of DYRK3 to cAMP response element (CRE)-binding protein (CREB), and 3) effects of DYRK3 on hematopoietic progenitor cell survival. Regarding the DYRK3 kinase domain, intactness of Tyr(333) (but not Tyr(331)) within subdomain loop VII-VIII was critical for activation. Tyr(331) plus Tyr(333) acidification (Tyr mutated to Glu) was constitutively activating, but kinase activity was not affected substantially by unique N- or C-terminal domains. In transfected 293 and HeLa cells, DYRK3 was discovered to efficiently stimulate CRE-luciferase expression, to activate a CREB-Gal4 fusion protein, and to promote CREB phosphorylation at Ser(133). Interestingly, this CREB/CRE response was also supported (50% of wild-type activity) by a kinase-inactive DYRK3 mutant as well as a DYRK3 C-terminal region and was blocked by protein kinase A inhibitors, suggesting functional interactions between protein kinase A and DYRK3. Finally, DYRK3 expression in cytokine-dependent hematopoietic FDCW2 cells was observed to inhibit programmed cell death. Thus, primary new insight into DYRK3 kinase signaling routes, subdomain activities, and possible biofunctions is provided.

Mammalian DYRKs 1 (1,2) and HIPKs (3) are recently described subfamilies of MAPK-related protein kinases that target Ser/Thr sites, yet also appear to be activated by tyrosine (auto)phosphorylation at a conserved YXY motif (or loop) between consensus kinase subdomains VII and VIII (hence, the nomenclature dual-specificity tyrosine-regulated kinases, DYRKs) (1,4). At least seven DYRK isoforms have been described (5) that appear to derive from four unique genes (dyrk1-4). 2 Within this family, DYRK1 has been best studied to date, is expressed at high levels in brain (6), maps to the critical region of the Down's syndrome locus (7), and precipitates learning and memory defects when expressed in transgenic mice (8). Also, mutation of a dyrk1 gene homolog in Drosophila (MNB, for minibrain kinase) disrupts neuroblast formation in the outer proliferation center and limits optic and central lobe development (9). By comparison, HIPKs contain a DYRK-type kinase domain, but, as a separable subfamily, differ in possessing N-terminal domains that interact with NK homeoproteins (3). HIPK2 has been best studied and recently has been discovered to play an important role in p53 regulation during radiation-induced apoptosis (10).
Other DYRKs have not been well studied. Recently, our laboratory (11) and Lord et al. (12) discovered that DYRK3 is selectively expressed at high levels in hematopoietic cells of erythroid lineage. Using an antisense oligonucleotide approach, it has also been demonstrated, in primary murine and human hematopoietic progenitor cells, that inhibition of DYRK3 expression significantly and specifically affects the production of colony-forming units-erythroid (the penultimate progenitors of erythroblasts). Based on apparently preferred arginine-and proline-directed substrate sequences (13) and on the in vitro ability of DYRK1 to phosphorylate eukaryotic synthesis initiation factor-2B⑀ and tau microtubule-associated protein at priming sites (14), DYRK1 has been suggested to act as a glycogen synthase kinase-priming kinase. Via in vitro kinase assays, DYRK1 has also been shown to be capable of phosphorylating forkhead transcription factor FKHR (forkhead in rhabdomyosarcoma) (15), CREB (16), and STAT3 (signal transducer and activator of transcription-3) (17). However, factors that regulate DYRK3 and the nature of DYRK targets in general remain otherwise unclear.
Homologs of mammalian DYRKs interestingly also occur in Saccharomyces cerevisiae and Dictyostelium (Yak1p and YakA, respectively) (18,19); and recently, each of these DYRK-like kinases has been linked to PKA signaling pathways (20,21). In Dictyostelium, YakA is required for cAMP (and PKA)-directed differentiation to sporulating stalks due to nutrient withdrawal (21,22), whereas in S. cerevisiae, Yak1p may directly affect PKA function by phosphorylating Bcy1p, the single PKA regulatory subunit of budding yeast (23). Based on these observations, the prospect that DYRK3 might also somehow engage a PKA (and possibly CREB)-linked pathway was investi-* This work was supported by National Institutes of Health Grant RO1 DK40242. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
DYRK3 Constructs-The full-length murine DYRK3 cDNA used in these studies was prepared by expressing a bacterial artificial chromosome-derived dyrk3 gene fragment in COS cells. cDNAs generated by this process were cloned into a ZAP vector, excised from a derived phagemid library, adapted at 5Ј and 3Ј termini with EcoRI and XhoI sites, and cloned into the mammalian expression vector pEFNeo-Myc6. The DYRK3 point mutants Y331A, Y333A, Y331E/Y333E, and K202R were prepared from pEFNeo-Myc-wtDYRK3 using the Stratagene PCRbased XL site-directed mutagenesis kit and the following primers: 5Ј-CGA GTA TCA GAA GCT TGC CAC GTA TAT CCA GTC C-3Ј plus 5Ј-GGA CTG GAT ATA CGT GGC AAG CTT CTG ATA CTC G-3Ј (Y331A construct), 5Ј-GAA GCT TTA CAC GGC TAT CCA GTC CCG C-3Ј plus 5Ј-GCG GGA CTG GAT AGC CGT GTA AAG CTT C-3Ј (Y333A construct), 5Ј-GTA TCA GAA GCT TGA GAC GGA AAT CCA GTC CCG C-3Ј plus 5Ј-GCG GGA CTG GAT TTC CGT CTC AAG CTT CTG ATA C-3Ј (Y331E/Y333E construct), and 5Ј-CGG CAG TAC GTG GCC CTG AGA ATG GTG CGC AAT GAG AAA C-3Ј plus 5Ј-GTT TCT CAT TGC GCA CCA TTC TCA GGG CCA CGTBACT GCC G-3Ј (K202R FIG. 2. Tyr 333 (but not Tyr 331 ) is required for DYRK3 kinase activity. cDNAs encoding Myc epitope-tagged wtDYRK3, DYRK3-K202R (kinase-inactive type II mutant), DYRK3-Y331A, and DYRK3-Y333A were cloned into pEFNeo and expressed transiently in transfected 293 cells. At 48 h post-transfection, lysates and monoclonal antibody 9E10 immunoprecipitates were prepared. DYRK3 kinase activity in immunoprecipitates was then assayed using MBP and [␥-32 P]ATP. Shown in the upper panels are 32 P-labeled MBP products and Western blot controls for expression and immunoprecipitate equivalence. In the lower panel, the kinase activities of these DYRK3 constructs are compared quantitatively as normalized means Ϯ S.D. of triplicates. Results are representative of three independent experiments. K, kDa. Expression of DYRK3 Constructs-293 cells (American Type Culture Collection, Manassas, VA) were maintained in Opti-MEM I (Invitrogen) and 7% fetal bovine serum (FBS) plus PSF (100 units/ml penicillin, 100 g/ml streptomycin, and 0.25 g/ml amphotericin B). DYRK3 constructs in pEFNeo vectors were transfected and expressed using 15 g of plasmid DNA plus 30 l of FuGENE 6 (Roche Molecular Biochemicals) per 100-mm dish of 293 cells at 50% confluency. Lysates were prepared at 48 h post-transfection. In assays of phospho-CREB, the 293 cell medium was changed to 0.5% FBS in Opti-MEM I 2 h prior to transfections.
In Vitro Kinase Assays-In kinase assays, washed immunoprecipitates were reacted at 30°C with 40 l of a solution containing 1. were also tested as described for A for activity in activating this CREB/ CRE response pathway. Interestingly, significant activity was retained by kinase-inactive DYRK3-K202R. sium/unlabeled ATP mixture (catalog no. 20-113). At the indicated intervals, supernatants were recovered and denatured in SDS sample buffer. 32 P-Labeled MBP products were assayed by SDS-PAGE and phosphorimaging (Storm Scanner, Amersham Biosciences).
CREB/CRE Reporter Experiments-In experiments using 293 cells, 6 ϫ 10 5 cells were plated (six-well plate format) and cultured overnight. Four hours prior to transfection, the medium was changed to 1% FBS in Opti-MEM I. Cells were then transfected (in triplicate) with pEFNeo-DYRK constructs (2 g), pCRE-Luc reporter vector (1 g; Stratagene), pSEAP (0.2 g; Tropix Inc.), and 10 l of FuGENE 6 per well. At 36 h post-transfection, supernatants were assayed for alkaline phosphatase activity (Phospha-Light system, Tropix Inc.), and lysates were prepared (Promega reporter lysis buffer, catalog no. E3971). Lysates were then assayed for protein content (BCA system, Pierce) and luciferase activity (Promega substrate, catalog no. E1483). CCL-2 HeLa PathDetect cells FDC Cells and Assays of Programmed Cell Death-FDC cells were maintained in Opti-MEM I containing 7% FBS plus 3.5% WeHI3 cellconditioned medium (as a source of IL-3). For stable expression of wtDYRK3 and Bcl-x L , FDCW2 cells were washed once with ice-cold Opti-MEM I medium and resuspended at 1 ϫ 10 7 cells/ml in Opti-MEM I. pEFNeo constructs encoding Myc-wtDYRK3 or Bcl-x L (60 g) were transfected into FDCW2 cells (0.8 ml) using a Gene Zapper 450/2500 electroporation cuvette (Bio-Rad). Transfected cells were then selected for 2 weeks in medium containing 1 mg/ml G418. Derived lines were screened for DYRK3 expression by Western blotting. In assays of programmed cell death, nonviable cells were assayed by propidium iodide staining. Exponentially growing cells were adjusted to 3 ϫ 10 5 cells/ml, expanded to 8 ϫ 10 5 cells/ml, and washed twice with and cultured at 5 ϫ 10 5 cells/ml in Opti-MEM I medium and 0.9% FBS in the absence of IL-3. For propidium iodide staining, cells (1 ml) were incubated with 5 l of propidium iodide (0.25 mg/ml in phosphate-buffered saline) for 15 min and analyzed by flow cytometry.

Roles for the DYRK3 YTY Motif in Kinase Activation-Stud-
ies of DYRK3 are limited, and little is known regarding the nature of activation mechanisms and/or targets. Experiments first sought to assess ways in which Tyr 331 and/or Tyr 333 within the DYRK3 unique predicted YTY activation loop might affect kinase activity. As shown in Fig. 1 (lower panel), this YTY motif lies between kinase subdomains VII and VIII and is related to motifs within PRP4 and ERK1/2 kinases, but occurs in DYRK3 as a SSCFEYQKYTYQSRFYR sequence (bounded by kinase DFG and APE signature sequences). To test how DYRK3 ki-nase activity might be affected by this motif, cDNAs were prepared that encoded the Y331A, Y333A, or Y331E/Y333E mutation (Fig. 1, upper panel). As an additional control, Lys 202 within the DYRK3 predicted ATP-binding site (kinase subdomain II) was mutated to generate kinase-inactive DYRK3 (K202R). Next, 293 cells were transfected with pEFNeo vectors encoding these DYRK3 forms, and DYRK3 kinase assays were performed using anti-Myc epitope immunoprecipitates plus [␥-32 P]ATP and MBP. These experiments revealed that intactness of Tyr 333 (but not Tyr 331 ) is important for DYRK3 kinase activity (Fig. 2, upper panels). Replicate quantitative analyses showed that DYRK3-Y333A retained little to no activity, whereas ϳ75% of the wild-type activity was retained by DYRK3-Y331A (Fig. 2, lower panel). cDNAs were also prepared (and cloned into a pEFNeo vector) that encoded N-terminal plus kinase domains (NK construct) or kinase plus C-terminal domains (KC construct). These forms were also expressed and assayed for in vitro kinase activities. Each form proved to possess activity approximating that of wtDYRK3 (data not shown), suggesting that these domains may not substantially affect kinase domain activity.
Based on the above findings, efforts were also made to assay the predicted (auto)phosphorylation of Tyr 333 (and possibly Tyr 331 ). In experiments using several distinct anti-phosphotyrosine antibodies, however, these events were not clearly detected (as compared with JAK2 (Janus kinase-2) kinase as a parallel and positive control). In addition, DYRK3 expression levels in stably transfected 293 lines and in Sf9 lines (baculovirus system) were low, and this complicated direct analyses of phosphorylated sites. We therefore chose to test the possible effects of YTY loop acidification on DYRK3 activity. Interestingly, acidification of this loop to Y331E/Y333E yielded a DYRK3 form that possessed ϳ75% of the wild-type activity (Fig. 3). This DYRK3-Y331E/Y333E mutant also phosphorylated MBP in a time course comparable with wtDYRK3 (Fig. 4). These outcomes are consistent with a primary role for Tyr 333 phosphorylation in DYRK3 activation. They also provide a novel DYRK3 gain-of-function mutant (Y331E/Y333E) that is active in the predicted absence of loop (auto)phosphorylation.

DYRK3 Engages a CREB/CRE Response Pathway via Kinase Domain-and C-terminal Domain-dependent Mecha-
nisms-In S. cerevisiae and Dictyostelium, the DYRK kinase homologs Yak1p and YakA have recently been revealed to function within PKA-integrated signaling routes (20,21), and human DYRK1A has been suggested to phosphorylate CREB (16). These reports prompted tests of the ability of DYRK3 to possibly modulate a CREB/CRE signaling pathway. This was investigated initially by cotransfecting 293 cells with a pEFNeo vector encoding wtDYRK3 (or empty pEFNeo as a control) plus a pCRE-Luc transcriptional reporter. pCRE-Luc was reproducibly stimulated 4 -5-fold by wtDYRK3, and DYRK3-Y331E/ Y333E likewise proved to efficiently activate pCRE-Luc (Fig.  5A). Interestingly, yet somewhat unexpectedly, pCRE-Luc was also stimulated significantly by the kinase-inactive construct DYRK3-K202R (Fig. 5B). To examine these effects in an independent and CREB transactivation domain-specific system, the activities of DYRK3 constructs were also tested in CCL-2 HeLa cells. These cells stably express a CREB-Gal4 fusion protein that is activated upon phosphorylation of the CREB kinase-inducible activation domain and report activation via a single stably integrated Gal4-binding element-luciferase reporter cassette. To control for transfection efficiencies (and to analyze DYRK3 activities within transfected cell populations), CCL-2 HeLa cells were transfected with pEFNeo-DYRK3 vectors plus an EYFP expression vector (pEYFP-C1). At 60 h post-transfection, EYFP-positive cells were retrieved by fluo-rescence-activated cell sorting, and luciferase activities due to DYRK3 expression were assayed (Fig. 6). These analyses revealed CREB kinase-inducible transcription domain activation by wtDYRK3 and DYRK-Y331E/Y333E and again demonstrated significant activation by kinase-inactive DYRK-K202R.
Next, to investigate possible bases for the above apparent CREB/CRE targeting of kinase-inactive DYRK3-K202R, cDNAs encoding the unique N-and C-terminal subdomains of DYRK3 were prepared, cloned into pEFNeo vectors, and tested in 293 cells for possible activation of this response pathway. Luciferase assays revealed that the DYRK3 C-terminal domain (but not the N-terminal domain) possessed activity essentially equivalent to that of DYRK3-K202R (Fig. 7, upper panels). Western blot analyses of cell lysates also showed that each Myc epitope-tagged form was expressed at a comparable level (Fig.  7, lower panel). Together, these findings indicate that DYRK3 can affect this signaling pathway via not only kinase domain-, but also C-terminal subdomain-mediated routes.
Roles for PKA in DYRK3 Activation of CREB/CRE Signaling-As indicated above, evidence has recently been provided that the DYRK-related Yak1p kinase of S. cerevisiae may interact with PKA (especially via the PKA regulatory subunit Bcy1) (23). Mammalian PKAs are more complex and include at least four regulatory subunits and three catalytic subunits (and useful antibodies to specific PKA subunits also are limiting) (26). Therefore, to test possible roles for PKA in our discovered DYRK3/CREB/CRE response pathway, we employed the specific PKA inhibitors (R p )-cAMP-S and H-89 (27,28). In pilot experiments, concentrations/doses of each that efficiently inhibited PKA (but did not compromise 293 cell viability) were determined in cells transfected transiently with a PKA catalytic subunit expression vector (pFC-PKA). Based on reported Upper panels, 293 cells were transfected with an empty pEFNeo vector (negative control), pFC-PKA (positive control), or pEFNeo-Myc-wt-DYRK3, and cells were cultured as described under "Experimental Procedures." Lysates were then prepared, and levels of endogenous phospho-Ser 133 CREB (P-SER133-CREB), total CREB, and ectopically expressed DYRK3 (as indicated) were assayed by Western blotting. Lower panels, in this system, the ability of the PKA inhibitor H-89 (at 0, 20, or 40 M) to inhibit the DYRK3-dependent formation of phospho-Ser 133 CREB was also assessed. K, kDa. administered at these doses to 293 cells transfected with pEF-Neo-wtDYRK3 (and pCRE-Luc), H-89 and (R p )-cAMP-S each efficiently blocked the ability of DYRK3 to activate a CREB/ CRE response (Fig. 8). In addition, each inhibitor also efficiently blocked the activities of the DYRK3-Y331E/Y333E, DYRK3-K202R and DYRK3-CT constructs (Fig. 9). These outcomes (plus the above-detailed experiments) support a model for DYRK3 action whereby DYRK3 can interact via not only its kinase domain, but also its unique C-terminal region, with PKA in a way that activates CREB/CRE responses.
Experiments were next performed to more directly demonstrate DYRK3-dependent phosphorylation of CREB at a critical site (Ser 133 ) within the CREB kinase-inducible transactivation domain. Here, 293 cells were transfected with pEFNeo-Myc-wtDYRK3 (or empty pEFNeo as a negative control), and possible effects on the phosphorylation of endogenous CREB were assayed using an anti-phospho-Ser 133 CREB antibody (Fig. 10). Under the conditions specified, endogenous CREB phosphorylation proved to be clearly stimulated upon wtDYRK3 expression. In addition, this event (DYRK3-dependent CREB phosphorylation) also proved to be inhibited by the PKA inhibitor H-89 (Fig. 10, lower panels). Using the singularly available antibody to one phospho-PKA regulatory subunit, whether PKA reg ␣II might be a target of DYRK3 was also tested, but with negative results. Whether this outcome relates to antibody sensitivity or possibly PKA reg isoform specificity is presently unresolved.
DYRK3 Inhibits Apoptosis Due to Cytokine Withdrawal in FDC Hematopoietic Progenitor Cells-In a final series of experiments, the possible effects of DYRK3 on progenitor cell growth, cell cycle, and/or survival were tested via the ectopic stable expression of wtDYRK3 in hematopoietic FDCW2 cells. FDCW2 cells depend strictly upon IL-3 as a growth and survival factor (30) and correspond closely to a myeloid progenitor cell population, but retain the capacity to activate erythroid gene expression in response to ectopically expressed GATA-1 (30). Also, FDCW2 cells express only low level endogenous DYRK3 (11). FDCW2 cells were electrotransfected with pEF-Neo-Myc-wtDYRK3 (or empty vector as an initial control), and stably transfected lines were selected in G418. In FDCW2-Myc-wtDYRK3 cells, expression of Myc-DYRK3 was confirmed (Fig.  11). Intact DYRK3 (70 kDa) as well as proteolyzed forms (30

DYRK3 Regulates CREB and Apoptosis
and 40 kDa) were observed (here and in repeated independent experiments). The possible effects of ectopically expressed Myc-DYRK3 on either proliferation or apoptosis due to cytokine (IL-3) withdrawal were then assayed. In [ 3 H]dThd incorporation assays (Fig. 11, upper left panel), no significant effects of DYRK3 on proliferation were detected. In contrast, ectopically expressed DYRK3 interestingly proved to significantly attenuate cell death due to cytokine withdrawal (Fig. 11, upper right and lower left panels). To better gauge the magnitude of this effect, FDCW2 cells were electrotransfected with the pEFNeo-Bcl-x L construct, and lines expressing Bcl-x L were isolated and assayed in parallel for inhibition of apoptosis due to cytokine withdrawal. In parallel assays, the effects of DYRK3 on this process were gauged to be strong (i.e. ϳ2-fold greater than those exerted by Bcl-x L ) ( Table I). DISCUSSION The main goals of this investigation were to analyze ways in which the YTY motif of DYRK3 kinase might contribute to kinase activation, to initially define possible effects of DYRK3 on progenitor cell growth and/or survival, and to attempt to discover intracellular signaling routes that DYRK3 might engage. As discussed above, the DYRK3 YXY loop together with several unique kinase domain sequence components structurally best define DYRKs as a novel group of dual-specificity kinases. These additionally conserved components include an SSC motif in subdomain VII; lack of an arginine that is highly represented in subdomain VIB in other kinases; and unusual DYRK characteristic cysteine, NLY, and glycine residues in subdomains IV-V, V, and VIII-IX, respectively (1,4). Regarding the DYRK3 YTY motif, our observation that intactness of Tyr 333 (but not Tyr 331 ) within the DYRK3 activation loop is important for catalysis is consistent with a primary role for Tyr 333 in phosphorylation-induced activation. In support of this notion (and while this work was in progress), Himpel et al. (31) reported that in DYRK1 kinase, the (auto)phosphorylation of the second Tyr residue of a corresponding YTY loop (Fig. 1) correlates with kinase activation. In our study, catalytic roles for the YTY loop of DYRK3 were also assessed by acidification to mimic predicted phosphorylation. Mutation of Tyr 331 and Tyr 333 to glutamate proved to activate DYRK3 kinase function to ϳ75% of wild-type DYRK3 levels (Figs. 3 and 4). In contrast, deletion of either the unique N-or C-terminal domain of DYRK3 did not appreciably affect activity. Together, these results indicate that YTY loop phosphorylation may constitute a primary mechanism of DYRK3 activation. Tyr 333 of DYRK3 corresponds in position to Tyr 185 in the TXY activation loop of ERK2. In ERK2, however, acidification of Tyr 185 fails to substantially affect catalysis (24). This differs from Thr 183 of ERK2 (or Ser 218 and Ser 222 of the ERK kinase MKK1 (MAPK kinase-1) as a second example), for which acidification leads to kinase activation (24,32). To our knowledge, DYRK-Y331E/ Y333E represents the first case among MAPK-related kinases for efficient Tyr acidification loop activation, and this outcome adds a further functional uniqueness to DYRKs (especially DYRK3).
Regarding signaling pathways, the ability of DYRK3 to stimulate a CREB/CRE response pathway is also described in this study . Evidence that this response depends upon phosphorylation of the CREB transactivation domain (33) was provided using CCL-2 HeLa cells, which contain a CREB-Gal4 fusion protein and an integrated single-copy Gal4 elementluciferase reporter cassette, and by direct analyses of DYRK3dependent CREB phosphorylation. Aspects of these experiments that merit discussion are 1) a partial (yet clear) activation of this CREB/CRE response pathway by a kinaseinactive form of DYRK3 (K202R) as well as the DYRK3 unique C-terminal domain and 2) the blocking of this response pathway by PKA-specific inhibitors (Figs. 7-11). Specifically, the ability of DYRK3-K202R and DYRK3-CT to engage a CREB/ CRE response argues that this response likely depends (at least in part) upon kinase domain-independent protein-protein interactions. In a minimal model, this might involve interactions of DYRK3 with CREB and/or CREB-interacting proteins (see Fig. 12 for a model). However, in tests of the possible effects of PKA inhibitors on DYRK3-dependent CREB/CREluciferase activation, the PKA catalytic subunit inhibitor H-89 (29) and the regulatory subunit inhibitor (R p )-cAMP-S (27,28) were each shown essentially to block this DYRK3 action (Figs.  8 and 9). This result, together with the recently reported ability of Yak1p of S. cerevisiae to engage the Bcy1 regulatory subunit of PKA (23), suggests a regulatory route through which DYRK3 (and possibly other DYRKs) may likewise directly engage mammalian PKAs. Tests of this model in mammalian systems are complicated by the occurrence of multiple PKA regulatory and catalytic isoforms and genes (26,34). Nonetheless, a primary route for PKA activation is known to involve PKA-interacting factors, including AKAPs, that modulate not only subcellular targeting (35), but also PKA kinase activity (36). The C-terminal domain of DYRK3 comprises a unique region with no obvious homologies to known proteins or protein motifs; yet in future experiments, it should be interesting to discover how direct, or indirect, the presently discovered linkage between DYRK3 (and possibly other DYRKs) and PKA (and AKAPs) might be. Both PKA (37,38) and CREB (38,39) are known to play important roles in cell growth, survival, and/or differentiation events in several specific tissues and lineages (including erythroid progenitor cells) (40,41). The specific ways by which DYRK3 modulates these regulators should likewise be interesting to discover.
Finally, DYRK3 has been shown to attenuate apoptosis due to cytokine withdrawal in IL-3-dependent hematopoietic FDCW2 cells (Fig. 11 and Table I), and this occurred independently of any detectable effect on mitogenic potential. Roles for DYRKs in cell survival have not previously been well studied or described. However, the DYRK3-related kinase HIPK2 has recently been discovered to mediate p53 activity following radiation-induced damage (10,42), and p53 is well known to affect cell survival pathways (43,44). By comparison, the specific mechanisms that DYRK3 may use to modulate progenitor cell survival should likewise be interesting to uncover in future studies.