Protein kinase CK2 activation is required for transforming growth factor β‐induced epithelial–mesenchymal transition

Transforming growth factor β (TGFβ) is overexpressed in advanced cancers and promotes tumorigenesis by inducing epithelial–mesenchymal transition (EMT), which enhances invasiveness and metastasis. Although we previously reported that EMT could be induced by increasing CK2 activity alone, it is not known whether CK2 also plays an essential role in TGFβ‐induced EMT. Therefore, in the present study, we investigated whether TGFβ signaling could activate CK2 and, if so, whether such activation is required for TGFβ‐induced EMT. We found that CK2 is activated by TGFβ treatment, and that activity peaks at 48 h after treatment. CK2 activation is dependent on TGFβ receptor (TGFBR) I kinase activity, but independent of SMAD4. Inhibition of CK2 activation through the use of either a CK2 inhibitor or shRNA against CSNK2A1 inhibited TGFβ‐induced EMT. TGFβ signaling decreased CK2β but did not affect CK2α protein levels, resulting in a quantitative imbalance between the catalytic α and regulatory β subunits, thereby increasing CK2 activity. The decrease in CK2β expression was dependent on TGFBRI kinase activity and the ubiquitin–proteasome pathway. The E3 ubiquitin ligases responsible for TGFβ‐induced CK2β degradation were found to be CHIP and WWP1. Okadaic acid (OA) pretreatment protected CK2β from TGFβ‐induced degradation, suggesting that dephosphorylation of CK2β by an OA‐sensitive phosphatase might be required for CK2 activation in TGFβ‐induced EMT. Collectively, our results suggest CK2 as a therapeutic target for the prevention of EMT and metastasis of cancers.

Protein kinase CK2 is a constitutively active, growth factor-independent serine/threonine-protein kinase with key roles in cell cycle regulation, cellular differentiation, proliferation and apoptosis regulation (Ahmad et al., 2005;Shin et al., 2005;Song et al., 2000). Changes in CK2 expression or activity have been reported in many cancers (Kim et al., 2007;Landesman-Bollag et al., 2001;Scaglioni et al., 2006;Shin et al., 2005) and overexpression of the catalytic subunit of CK2 can induce tumorigenesis (Landesman-Bollag et al., 2001). CK2 is also a positive regulator of Wnt signaling, which is important for metastasis (Seldin et al., 2005). Recently, we reported that an increase in CK2 activity induced the E-to N-cadherin switch (Ko et al., 2012). Although an increase in CK2 activity could induce the E-to N-cadherin switch, it is not known whether CK2 plays a role in TGFb-induced EMT. Because it is well known that TGFb induces EMT, the present study aimed to investigate whether TGFb signaling could activate CK2 and also whether the activation was essential for TGFbinduced EMT.

In vitro kinase assay
To evaluate intracellular CK2 activity, an in vitro kinase assay was performed as described previously with slight modification (Scaglioni et al., 2006). Bacterially expressed GST-CS (CK2 Substrate; GST-RRRDDDSDDD) (3 lg) was incubated with glutathione-Sepharose 4B beads for 60 min, and washed twice with kinase buffer (4 mM Mops, pH 7.2, 5 mM b-glycerophosphate, 1 mM EGTA, 200 lM sodium orthovanadate, and 200 lM dithiothreitol). The beads were incubated with 100 lg of cell lysates in a final volume of 50 lL of kinase reaction buffer (10 lL of 5 9 kinase buffer, 10 lL magnesium/ATP cocktail [90 lL of 75 mM MgCl 2 /500 mM ATP and 10 lL (100 lCi) of [c-32 P]-ATP]) for 20 min at 30°C. The reactions were stopped by washing twice with 1 9 kinase buffer. The samples were resuspended with 30 lL of 2 9 SDS/PAGE sample-loading buffer, subjected to 12% SDS/PAGE, stained with Coomassie Brilliant Blue, and dried on Whatman paper (GE Healthcare Life Sciences, Little Chalfont, UK). 32 P incorporation was detected by autoradiography.

Dual-luciferase reporter assay
The cells were seeded in six-well plates and cotransfected with p3TP-Lux and pRL-TK using ViaFect TM (Promega Corp., Madison, WI, USA). Twenty-four hours after transfection, the cells were treated with TGFb for 24 h, washed with PBS and harvested. Cell lysates were prepared with 200 lL of Passive Lysis buffer (Promega). Aliquots (20 lL) of cleared lysate were analyzed for luciferase activity using a Dual-luciferaseÒ reporter assay system (Promega). The luciferase activity of p3TP-Lux was normalized to that of pRL-TK.

Cell fractionation
The cells were allowed to swell in buffer A comprising 10 mM Hepes (pH 7.9), 10 mM KCl, 0.1 mM EDTA, 1 mM dithiothreiltol, 1 mM phenylmethanesulfonyl fluoride, 1 9 protease inhibitor cocktail and 1 mM sodium orthovanadate. The samples were adjusted to 0.6% Nonidet P-40 (NP-40), and vortexed vigorously for 10 s. Nuclei were pelleted by centrifugation at 10 000 9 g for 30 s at 4°C. The supernatants were collected and used as the cytoplasmic fraction. After washing the pellets with PBS, they were lysed in buffer C comprising 20 mM Hepes, pH 7.9, 0.4 M NaCl, 0.1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethanesulfonyl fluoride, 1 9 protease inhibitor cocktail and 1 mM sodium orthovanadate by sonication. The lysates were cleared by centrifugation at 10 000 9 g for 20 min at 4°C. The supernatants were collected and used as the nuclear fraction.

Immunoprecipitation
The cells were collected and lysed with 1 mL of immunoprecipitation lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% NP-40) with cOmplete TM protease inhibitor cocktail (Roche Diagnostics). The cell lysates were pre-cleared and then incubated with the appropriate antibodies for 1 h at 4°C. The antibodyprotein complexes were precipitated with Protein A/G-Sepharose beads (Santa Cruz Biotechnology Inc.), washed, and resuspended in 40 lL of SDS/PAGE loading buffer.

Cell migration assay
A cell migration assay was conducted using specific wound-assay chambers purchased from ibidi GmbH (Munich, Germany). All experiments were performed in accordance with the manufacturer's instructions.

Statistical analysis
Statistical comparisons of groups were performed using Student's t-test. P < 0.05 was considered statistically significant.

CK2 activation was required for TGFb-induced EMT
An increase in CK2 activity by CK2a overexpression induced EMT in the cancer cells (Ko et al., 2012) and TGFb-induced EMT in A549 cells (Kasai et al., 2005). To investigate whether CK2 was activated during TGFb treatment, A549 cells were treated with TGFb and harvested at 0, 24, 48 and 96 h after treatment. Lysates were prepared, and an in vitro kinase assay and western blot analysis were performed. CK2 activity peaked at 48 h after TGFb treatment and the E-to N-cadherin switch was observed 24 h after TGFb treatment (Fig. 1A). To examine whether TGFBRI kinase activity was required for the CK2 activation, A549 cells were pretreated with the TGFBRI kinase inhibitor, SB431542. We found that, without TGFBRI activation, neither the increase in CK2 activity, nor the cadherin switch occurred (Fig. 1B). To examine whether the CK2 activation is required for TGFbinduced EMT, A549 cells were treated with the pharmacological CK2 inhibitor, emodin, and then with TGFb for 48 h. In the absence of emodin, A549 cells changed from a rounded, epithelial morphology to a spindle and fibroblast-like appearance (Fig. 1C) and the E-to N-cadherin switch (Fig. 1D) was observed. However, in the presence of emodin, morphological changes and cadherin switch were not observed (Fig. 1C,D). To confirm these results, we generated stable CSNK2A1 knockdown (CKD) A549 cells. Previously, we reported that CKD could decrease cellular CK2 activity (Ko et al., 2012). We found that CKD  cells showed neither morphological changes (Fig. 1E), nor the cadherin switch ( Fig. 1F) with TGFb treatment. To examine the effect of CKD on motility of the cells, migration assays were performed. Even in the presence of TGFb, CKD cells were not motile (Fig. 1G).

CK2 activation independent on SMAD4
Because the increase in CK2 activity depended on TGFBRI kinase activity, we then examined whether canonical SMAD signaling was required for activation. To disrupt canonical SMAD signaling, we generated stable SMAD4-knockdown (SKD) A549 cells using shRNA. When SKD cells were treated with TGFb, EMT was induced and CK2 was activated ( Fig. 2A). Next, we examined whether CK2 was required for SMAD signaling. When CKD cells were treated with TGFb, SMAD2 was phosphorylated (Fig. 2B, lane 2 vs. lane 4) and SMAD4 was translocated into the nucleus (Fig. 2C, lane 2 vs. lane 4) even in the absence of CK2 activation (Fig. 2B, lane 2 vs. lane 4). There was no difference in p3TP-Lux (Wrana et al., 1992) luciferase activity between the control and CKD cells by TGFb treatment (Fig. 2D). Collectively, these results suggested that CK2 activation and EMT did not require SMAD4.

CK2b degradation by TGFb signaling
Because TGFb-induced CK2 activation depended on TGFBRI kinase activity, TGFBRI CA was used for TGFb signaling instead of TGFb treatment (Wieser et al., 1995). Because unbalanced protein levels of CK2 subunits may drive EMT (Deshiere et al., 2013), we then examined whether TGFb signaling could alter the protein level of CK2a or CK2b. Western blot analysis using lysates from HEK 293 cells cotransfected with TGFBRI CA and either with CK2a or CK2b showed that CK2b expression was decreased CK2b but did not affect CK2a protein levels (Fig. 3A). To determine the effect of CK2b downregulation on CK2 activity, CSNK2B-knockout (bKO) A549 cells were generated using the CRISPR/Cas9 gene knockout system. Western blot analysis and in vitro kinase assay showed that with bKO, the E-to N-cadherin switch was induced (Fig. 3B, top) and CK2 activity was increased (Fig. 3B, bottom) even in the absence of Effect of CKD on TGFb-induced SMAD2 phosphorylation. Control or CKD cells were treated, or not, with TGFb for 48 h. Western blot analysis was performed with the indicated antibodies (top) and an in vitro kinase assay was performed (bottom). (C) Effect of CKD on TGFbinduced nuclear localization of SMAD4. Control or CKD cells were treated, or not, with TGFb for 48 h. Cells were fractionated, and western blot analysis was performed with the indicated antibodies using the nuclear fraction. (D) Effect of CKD on p3TP-lux-promoter activation by TGFb. The luciferase activity of p3TP-Lux was normalized to that of pRL-TK. Data represent the mean AE SD of one experiment performed in triplicate. Similar results were obtained from two independent experiments. TGFb treatment. A nigration assay showed that bKO cells became motile even without TGFb treatment (Fig. S1). To confirm the effect of TGFb signaling on the protein level of CK2a and CK2b, A549 cells were treated with TGFb for 1, 3, 6, 12 and 24 h. Western blot analysis confirmed that the protein level of endogenous CK2a was not altered by TGFb treatment; however, CK2b was decreased by the treatment (Fig. 3C, top). An in vitro kinase assay using the same lysates showed that CK2 activity was increased and (B) Effect of bKO on CK2 activity and EMT. bKO cells were generated using the CRISPR/Cas9 system. Western blot analysis was performed with the indicated antibodies (top) and an in vitro kinase assay was performed (bottom). (C) Effect of TGFb on the protein levels of endogenous CK2 subunits and CK2 activity. A549 cells were treated with TGFb for the indicated time periods. Western blot analysis was performed with the indicated antibodies (top) and an in vitro kinase assay was performed (bottom). (D) TGFBRI kinase activity-dependent (top) and proteasome-dependent (bottom) CK2b degradation. HEK 293 cells were cotransfected with His-TGFBRI CA or His-TGFBRI KD along with Myc-CK2b, or cotransfected with His-TGFBRI CA along with Myc-CK2b, followed by MG132 (10 lM) treatment. Western blot analysis was performed with the indicated antibodies. (E) TGFBRI kinase activity-dependent and proteasome-dependent degradation of CK2b and EMT. A549 cells were transfected with Myc-CK2b and pretreated with dimethylsulfoxide, SB431542 or MG132 for 12 h. The cells were treated with TGFb for the indicated time periods. Western blot analysis was performed with the indicated antibodies. maintained until 24 h after TGFb treatment (Fig. 3C,  bottom). To examine whether TGFBRI kinase activity is required for the decrease in CK2b expression, TGFBRI kinase dead was used (lysine 232 is replaced with arginine). Unlike TGFBRI CA, TGFBRI kinase dead did not decrease the protein level of CK2b, indicating that the decrease in the CK2b protein level was dependent on TGFBRI kinase activity (Fig. 3D, top). To examine whether CK2b is degraded by the ubiquitin-dependent proteasome pathway, cells cotransfected with TGFBRI CA and CK2b were treated with MG132; in the presence of MG132, CK2b was not degraded by TGFb signaling (Fig. 3D, bottom). To confirm these results, A549 cells were pretreated with SB431542 or MG132 before TGFb treatment. In the absence of SB431542 or MG132, CK2b was rapidly degraded, and the E-to N-cadherin switch was induced ( Fig. 3E; dimethylsulfoxide). When SB431542 was pretreated, the CK2b was degraded more slowly than in dimethylsulfoxide treated cells, and the E-to N-cadherin switch was not induced (Fig. 3E; SB431542). When MG132 was pretreated, CK2b was not degraded, and the E-to N-cadherin switch was not induced (Fig. 3E; MG132).

CHIP and WWP1 are E3 ubiquitin ligases for CK2b degradation
Because CK2b is polyubiquitinated in TGFb signaling (Fig. S2), we examined which E3 ligase (s) is involved in the ubiquitination of CK2b. Among the E3 ubiquitin ligases known to be involved in TGFb signaling and that we used for screening (De Boeck and ten Dijke, 2012), CHIP and WWP1 lowered the CK2b protein level (Figs S3 and 4A). MG132 protected CK2b from CHIP-and WWP1-mediated degradation (Fig. 4A) and CK2b interacted with these E3 ligases (Fig. 4B). Both CHIP and WWP1 increased CK2b ubiquitination (Fig. 4C) and, together with TGFb signaling, CHIP and WWP1 efficiently degraded CK2b (Fig. 4D). To examine the effect of CHIP or WWP1 knockdown on the CK2b protein level during TGFb signaling, siRNA against CHIP or WWP1 was used. The CK2b protein level was not decreased by TGFb treatment in the absence of either CHIP or WWP1 expression (Fig. 4E).

Dephosphorylation-dependent CK2b degradation
As reported previously , CK2b was autophosphorylated by CK2a and stabilized (Fig. 5A). To examine whether TGFb signaling could degrade phosphorylated CK2b, phosphomimetic CK2b 3E mutant was used. We found that phosphorylated CK2b was not degraded by TGFb signaling (Fig. 5B). Based on these results, we assumed that dephosphorylation of CK2b preceded the degradation of CK2b. Because it was reported that TGFb signaling could activate OAsensitive protein phosphatase (Petritsch et al., 2000), HEK 293 cells cotransfected with TGFBRI CA and Myc-CK2b were treated or untreated with 2 nM OA. In the presence of OA, CK2b was no longer degraded by TGFb signaling, indicating that the degradation required the activation of OA-sensitive phosphatase (Fig. 5C). To confirm these results, A549 cells were pretreated or untreated with OA for 12 h and then treated with TGFb for the indicated time periods. Western blot analysis showed that OA treatment protected endogenous CK2b from degradation (Fig. 5D). To examine whether CHIP binds to dephosphorylated CK2b, HEK 293 cells were cotransfected with CHIP and CK2b in the presence or absence of CK2a or CK2b 3E mutant. IP and western blot analysis revealed that CHIP could bind more selectively to unphosphorylated CK2b (Fig. 5E). CHIP and WWP1 efficiently degraded wt CK2b but did not degrade CK2b 3E mutant (Fig. 5F).

Discussion
The present study shows that TGFb activated CK2 and activation was required for TGFb-induced EMT. We observed that TGFb signaling decreased the CK2b Fig. 4. CHIP and WWP1 as E3 ubiquitin ligases for TGFb-induced CK2b degradation. (A) CHIP-and WWP1-mediated degradation of CK2b. HEK 293 cells were cotransfected with Flag-CHIP and Myc-CK2b (left) or with Flag-WWP1 and Myc-CK2b (right) and then treated or not with MG132 for 12 h. Western blot analysis was performed with the indicated antibodies. (B) CK2b interaction with CHIP or WWP1. HEK 293 cells were cotransfected with Flag-CHIP and Myc-CK2b (left) or with Flag-WWP1 and Myc-CK2b (right). Immunoprecipitation was performed using anti-Myc Ab followed by western blot analysis. The expression controls were given in the Input. (C) CHIP-or WWP1induced polyubiquitination of CK2b. HEK 293 cells were cotransfected with indicated plasmids and then treated with MG132 for 12 h. Immunoprecipitation was performed using anti-Myc Ab. Western blot analysis was performed with anti-HA Ab. The expression controls were given in the Input. (D) CHIP-or WWP1-mediated CK2b degradation during TGFb signaling. HEK 293 cells were cotransfected with indicated plasmids. Western blot analysis was performed with the indicated antibodies. (E) CHIP-or WWP1-mediated CK2b degradation in TGFb signaling. A549 cells were transfected with either siRNA against CHIP (left) or siRNA against WWP1 (right) and then treated with TGFb for the indicated time periods. Western blot analysis was performed with the indicated antibodies.
protein level, thereby resulting in an imbalance between the protein levels of the catalytic a and regulatory b subunits, leading to CK2 activation. This decrease was TGFBRI kinase activity-dependent and proteasomedependent. We also observed that the E3 ubiquitin ligase involved in CK2b degradation was CHIP, and that OA-sensitive phosphatase-mediated dephosphorylation was required for CHIP-mediated degradation.
Although CK2 is known to be a ligand-independent, constitutively active serine/threonine kinase, EGF could activate CK2 (Ackerman et al., 1990;Ji et al., 2009). Apart from EGF, TGFb also could activate CK2 (Fig. 1). Although CK2 activity peaked at 50 min post EGF treatment, and returned to baseline by approximately 120 min (Ackerman et al., 1990), CK2 activity peaked approximately at 48 h post TGFb treatment (Fig. 1) suggesting that EGF and TGFb might operate with different mechanisms for CK2 activation. Although EGF activated CK2 via ERK2-mediated CK2a phosphorylation (Ji et al., 2009), TGFb might activate CK2 by inducing an imbalance between the levels of catalytic a and regulatory b subunits through b subunit degradation (Figs 3  and 4). The results of the present study were supported by previous studies reporting that the imbalance between CK2 subunit levels caused by the reduction of b regulatory subunit is linked to increase in molecular target levels related to EMT in tissue samples from breast cancer patients, and that CK2b-depleted epithelial cells exhibited EMT-like morphological changes, as well as enhanced migration and anchorage-independent growth (Deshiere et al., 2011(Deshiere et al., , 2013. Although CK2b knockdown could induce EMT phenotye and strongly elevate TGFb2 expression, blocking the TGFb signaling pathway did not counteract the EMT phenotype (Deshiere et al., 2011). Consistent with these results, we demonstrated that bKO A549 cells showed EMT phenotypes even in the absence of TGFb treatment (Figs 3B and S1). We also showed that CK2 activation and TGFb-induced EMT were blocked by TGFBRI kinase inhibitor (Fig. 1B) and also that EMT was not induced in the absence of CK2 activation (Fig. 1D,F) and CK2b downregulation (Fig. 3E), suggesting that the CK2 activity increase resulting from downregulation of regulatory CK2b subunit is required for TGFb-induced EMT. These results suggest that the roles of TGFb signaling in EMT induction might comprise CK2b degradation-dependent CK2 activation through a non-canonical SMAD signaling pathway and thus CK2b depleted cells no longer required TGFb signaling for EMT induction.
An increase in CK2 activity by the overexpression of CK2a catalytic subunit induced EMT in cancer cells even in the absence of TGFb-dependent canonical SMAD signaling (Ko et al., 2012), indicating that CK2 activation might be necessary and sufficient to induce EMT. TGFb induces the expression of EMTrelated transcription factors, such as SNAIL1 or ZEB1 through SMAD3-dependent transcription (Hoot et al., 2008;Postigo, 2003;Vincent et al., 2009). The SMAD pathway is a canonical TGFb signaling pathway and involves receptor-regulated SMADs (SMAD2 or SMAD3) and a common partner SMAD (SMAD4). Because SMAD4 is a common partner SMAD, SKD could abolish TGFb-mediated SMAD signaling by preventing SMAD2 or SMAD3 from forming a complex with SMAD4. In the absence of SMAD4, CK2 was activated and EMT was induced by TGFb, indicating that SMAD4 was not required for TGFbinduced EMT ( Fig. 2A). Our results are supported by a previous study reporting that SMAD4 is necessary for TGFb-induced cell-cycle arrest and migration, although it is not in TGFb-induced EMT (Levy and Hill, 2005). By contrast to our observations, it was reported that SMAD4 is indispensable for EMT. RNA interference-mediated SMAD4 knockdown or expression of a dominant negative SMAD4 mutant resulted in preserved E-cadherin expression (Deckers et al., 2006;Takano et al., 2007). Although the involvement of SMAD4 in EMT is controversial, we showed that TGFb could not induce EMT in A549 CKD cells (Fig. 1E,D) with no alterations in canonical SMAD signaling ( Fig. 2B-D). These results suggest that CK2 activation-dependent downstream signaling events could be dominant over SMAD signaling-dependent transcriptional induction of EMT-related transcription factors in TGFb-induced EMT. CK2 could stabilize Snail (MacPherson et al., 2010) or b-catenin (Polakis, 2007;Song et al., 2003) by phosphorylation. Stabilized and nuclear localized b-catenin subsequently upregulates Axin2 expression, upregulated Axin2 shuttles GSK3b out from the nucleus, and thus nuclear Snail can be stabilized (Yook et al., 2006). Collectively, we argued that CK2b subunit might mainly act as a regulatory subunit and unbalanced expression of CK2 subunits by signaling mediated CK2b depletion could increase intracellular CK2 activity for downstream signaling event such as EMT.
CK2b is ubiquitinated and degraded through a proteasome-dependent pathway . In the present study, we report that TGFb induced the ubiquitination and degradation of CK2b (Fig. 4). Many E3 ubiquitin ligases participate in the ubiquitindependent degradation of molecules involved in TGFb signaling (De Boeck and ten Dijke, 2012). We screened some of them and observed that the CK2b protein level was decreased by CHIP or WWP1 expression (Fig. S3). CHIP belongs to the group of really interesting new gene (RING) and RING-related E3 ligases, and it contains a tetratricopeptide repeat domain involved in Hsp70 and Hsp90 association (Ballinger et al., 1999). Hsp90 exists as a complex with Hsc70 and the a and b subunits of CK2 (Suttitanamongkol et al., 2002). We showed that CHIP interacted with  CK2b (Fig. 4B) and that CHIP preferentially bound to dephosphorylated CK2b (Fig. 5E). Unlike b-transducin repeat-containing proteins (b-TrCP), which specifically ubiquitinate phosphorylated substrates (Laney and Hochstrasser, 1999), CHIP does not require post-translational substrate modification for ubiquitination. WWP1 belongs to the C2-WW-Homologous to E6AP C Terminus (HECT) type E3 ubiquitin ligase family (Verdecia et al., 2003). We showed that WWP1 interacted with CK2b (Fig. 4B), although we could not detect preferential binding of WWP1 to dephosphorylated CK2b (data not shown). Instead, we observed that both CHIP and WWP1 degraded wt CK2b but did not degrade CK2 b 3E mutant, suggesting that CHIP and WWP1 might preferentially bind to dephosphorylated CK2b. Our results were partially supported by a previous study reporting that dephosphorylation induces the ubiquitination and degradation of FMRP (fragile X mental retardation protein) in dendrites (Nalavadi et al., 2012).
In non-canonical TGFb signaling, TGFBRI kinasedependent activation and interaction of phosphatase 2A with p70-S6 kinase could result in the dephosphorylation and inactivation of the kinase, thereby inducing G1 arrest (Petritsch et al., 2000). OA is a potent, selective inhibitor of protein phosphatases, completely inhibiting PP2A at 1 nM and PP1 at higher concentrations (IC 50 = 10-15 nM). In the present study, we treated cells with 2 nM OA and thus PP2A could be completely inhibited; however, this might not be the case for PP1. OA treatment protected CK2b from TGFb-induced degradation (Fig. 5), suggesting that PP2A was the phosphatase involved in TGFb-induced CK2b degradation. However, we could not inhibit TGFb-induced CK2b degradation in the PPP2CA-, PPP2CB-or PPP2R2A-knockout A549 cells generated using the CRISPR/Cas9 system (S. Kim & K. Kim, unpublished observation). Further experiments, including the generation of PPP2CA and PPP2CB double knockout A549 cells, are required to identify the phosphatase involved in TGFb-induced CK2b dephosphorylation.
TGFb is highly expressed in many cancers (Friedman et al., 1995;Levy and Hill, 2006;Picon et al., 1998). In advanced cancers, TGFb promotes tumorigenesis via EMT induction, and thus cancer cells become more invasive and metastatic. Sustained TGFb signaling could induce sustained CK2 activation, eventually resulting in metastasis.

Conclusions
In summary, the results of the present study show that TGFb activated CK2 and activation was required for TGFb-induced EMT. TGFb signaling decreased CK2b expression, thereby causing an imbalance between the protein levels of the catalytic a and regulatory b subunits, resulting in CK2 activation. The decrease in CK2b protein level was dependent on TGFBRI kinase activity and the ubiquitin-proteasome pathway. The E3 ubiquitin ligases responsible for TGFb-induced CK2b ubiquitination were CHIP and WWP1. Dephosphorylation of CK2b by OA-sensitive phosphatase might be required for CK2 activation in TGFbinduced EMT. Therefore, CK2 could be a good therapeutic target for inhibiting metastasis in cancers with high CK2 activity.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Effect of CSNK2B knockout (bKO) on motility. Fig. S2. Polyubiquitination of CK2b by TGFb signaling. Fig. S3. Screening of E3 ubiquitin ligases for CK2b degradation.