Protein Kinase Casein Kinase 2 Mediates Inhibitor-κB Kinase and Aberrant Nuclear Factor-κB Activation by Serum Factor(s) in Head and Neck Squamous Carcinoma Cells

We showed previously that the signal transcription factor nuclear factor-κB (NF-κB) is aberrantly activated and that inhibition of NF-κB induces cell death and inhibits tumorigenesis in head and neck squamous cell carcinomas (HNSCC). Thus, identification of specific kinases underlying the activation of NF-κB could provide targets for selective therapy. Inhibitor-κB (IκB) kinase (IKK) is known to activate NF-κB by inducing NH₂-terminal phosphorylation and degradation of its endogenous inhibitor, IκB. Casein kinase 2 (CK2) was previously reported to be overexpressed in HNSCC cells and to be a COOH-terminal IKK, but its relationship to NF-κB activation in HNSCC cells is unknown. In this study, we examined the contribution of IKK and CK2 in the regulation of NF-κB in HNSCC in vitro. NF-κB activation was specifically inhibited by kinase-dead mutants of the IKK1 and IKK2 subunits or small interfering RNA targeting the β subunit of CK2. CK2 and IKK kinase activity, as well as NF-κB transcriptional activity, was shown to be serum responsive, indicating that these kinases mediate aberrant activation of NF-κB in response to serum factor(s) in vitro. Recombinant CK2α was shown to phosphorylate recombinant IKK2 as well as to promote immunoprecipitated IKK complex from HNSCC to phosphorylate the NH₂-terminal S32/S36 of IκBα. We conclude that the aberrant NF-κB activity in HNSCC cells in response to serum is partially through a novel mechanism involving CK2-mediated activation of IKK2, making these kinases candidates for selective therapy to target the NF-κB pathway in HNSCC. (Cancer Res 2006; 66(13): 6722-31)

The nuclear factor-κB (NF-κB/Rel) proteins, a family of nuclear transcription factors and their Inhibitors-κB (IκB), are normally involved in regulating a wide range of intracellular processes, including response to oxidative stress, inflammation, growth factors, injury, and programmed cell death (1). Aberrant NF-κB activation has been detected in a variety of cancers, including head and neck squamous cell carcinomas (HNSCC; refs. 1–3). We discovered that NF-κB1/RelA (p50/p65) activation is increased with tumor progression in murine SCC and in cell lines and tumor specimens from patients with HNSCC (4–8). Increased nuclear staining of the phosphoactivated form of p65 has recently been shown in the majority of high-grade squamous dysplasias and HNSCC specimens from patients and correlated with decreased survival, supporting the wider importance of NF-κB in tumor development and clinical outcome of HNSCC (9). We have also shown that NF-κB is an important modulator of the altered pattern of gene expression and malignant phenotype (10), indicating that NF-κB activation is an important target for therapy. Specific inhibition of NF-κB by overexpression of an IκBα with S32/S36 phosphorylation site mutations (IκBαM) restored altered gene expression and inhibited SCC cell proliferation, survival, migration, angiogenesis, tumorigenesis, and radiation resistance in experimental models (7, 10, 11). Bortezomib, an inhibitor of proteasome and IκB degradation, had similar effects in preclinical and in a recent phase I clinical study (8, 12). However, the extent of inhibition of the proteasome NF-κB and tumor response was limited by the dose of bortezomib achievable without significant side effects (8). Thus, identification of the specific kinases underlying the signal phosphorylation of IκBα and activation of NF-κB could provide more selective targets for therapy in patients with HNSCC.

The eponymous IκB kinase (IKK) plays a critical role in activating NF-κB-mediated events in cell survival and the immune response (2). The classic IKK complex and pathway is composed of IKK1 (IKKα), IKK2 (IKKβ), and NF-κB essential modulator (NEMO; IKKγ) subunits. An alternate pathway involving IKK1 has also been described (2). In normal cells, the classic IKK complex plays a critical role in integrating responses to various stimuli, resulting in IKK2-mediated phosphorylation of S32/S36 in the NH₂ terminus of IκBα and NF-κB activation. Casein kinase 2 (CK2) is another stress-activated protein kinase that participates in the transduction of signals that promote cell growth and survival (13) and has been implicated in NF-κB activation (14–18). In distinction to IKK2, CK2 has been shown to directly phosphorylate the COOH-terminal PEST domain of IκBα and has been recently shown to be a COOH-terminal IκBα kinase responsible for UV light-induced NF-κB activation (14–16). CK2 may also promote transactivation by phosphorylation of the RelA subunit of NF-κB (17, 18). CK2 is overexpressed and activated in HNSCC cell lines and tumor specimens and associated with poor prognosis (19, 20). Furthermore, antisense RNA targeting CK2α was found to inhibit proliferation of a HNSCC cell line (21), similar to our findings with inhibition of NF-κB. These observations suggested a potential relationship and a role for CK2 in upstream signaling and activation of NF-κB in HNSCC.

The signals mediating NF-κB activation in HNSCC have not been elucidated. CK2, IKK, and NF-κB may be activated in response to a variety of cytokines, growth, and serum factors (13, 17, 18, 22–26). In the present study, we examined the contribution of CK2, IKK, and serum to aberrant NF-κB activity in HNSCC.

Cell lines. HNSCC cell lines from the University of Michigan squamous cell carcinoma (UM-SCC) series were obtained from Dr. T.E. Carey (University of Michigan, Ann Arbor, MI) and described previously (27). All UM-SCC cells that are used for this study were maintained in MEM (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (FBS). Supplemented serum-free medium (SFM) was the same MEM supplemented with nonessential amino acids, insulin-transferrin-selenium-G, vitamin solution, fibronectin (Invitrogen), and trace elements (Biofluids, Camarillo, CA). Human keratinocytes (HEKa) were purchased and maintained in supplemented 154CF medium with 200 mmol/L CaCl₂ (Cascade Biologics, Inc., Portland, OR). All cells were maintained at 37°C in 5% CO₂ atmosphere.

Reagents. Anti-IKK2 and anti-CK2β antibodies were purchased from BD PharMingen (San Jose, CA). Anti-IκBα, anti-IKK1, and anti-NEMO antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Wild-type (WT) and mutant IκBα-glutathione S-transferase (GST) fusion proteins and pcDNA3-IKK vectors were kind gifts from Dr. Keith Brown (National Institute of Allergy and Infectious Disease, NIH, Bethesda, MD).

Western blot analysis. Protein (10 μg) of whole-cell lysates was used for each assay with standard immunoblotting procedures. Total proteins that were transferred onto the membrane were stained and quantified for equal loading control and for normalizing the intensities of the detected specific protein bands.

Electrophoretic mobility shift assay. Nuclear extracts were obtained by using a nuclear extract kit (Active Motif, Carlsbad, CA). Nuclear extract (5 μg) was used for electrophoretic mobility shift assay (EMSA) with ³²P-NF-κB response element oligonucleotide (sense, 5′-AGTTGAGGGGACTTTCCCAGGC-3′; antisense, 3′-TCAACTCCCCTGAAAGGGTCCG-5′; Promega, Madison, WI). ³²P-OCT-1 was used as a positive control for loading.

NF-κB reporter assay. Cells were transfected by jetPEI complexed with control or cis-pNF-κB-luc cis-reporter plasmids [PathDetect In vivo Signal Transduction Pathway cis-Reporting Systems NF-κB (5×); TGGGGACTTTCCGC; Stratagene, San Diego, CA].¹

After 24 hours, luciferase activities were measured with a dual-luciferase system from Promega (San Luis Obispo, CA) using a vector II spectrophotometer (Perkin-Elmer, Boston, MA). pCIS-CK plasmid was used as a negative control for the pNF-κB-luc cis-reporter plasmid, whereas pFC-MEKK plasmid (both obtained from Stratagene) was cotransfected with pNF-κB-luc cis-reporter plasmid to serve as a positive control. The firefly luciferase activities of triplicate samples were averaged following normalization by corresponding renilla luciferase activity cotransfected in the same wells. The NF-κB luciferase activity is presented as the ratio of NF-κB-luc to CK-luc.

CK2 small interfering RNA transfection. Cells were transfected with CK2β-Stealth-RNAi (sense, 5′-CCAGCAACUUCAAGAGCCCAGUCAA; antisense, 5′-UUGACUGGGCUCUUGAAGUUGACGG), a corresponding scramble stealth RNA interference (RNAi) control (sense, 5′-CCACAACUUGAACGAACCUGGCCAA; antisense 5′-UUGGCCAGGUUCGUUCAAGUUGUGG; both designed by Invitrogen BLOCK-iT RNAi Designer),²

or a universal negative control small interfering RNA (siRNA; Ambion, Austin, TX) and LipofectAMINE 2000 (Invitrogen) for 48 hours followed by evaluation assays. Transfection efficiency and cell viability were monitored by FITC-labeled RNA oligo, 4′,6-diamidino-2-phenylindole (DAPI), and dead cell dye, respectively (Invitrogen).

mRNA quantification. IκBα, NF-κB2/p100, and CK2β mRNA levels from the cells were quantified by QuantiGene assay kit with probe sets designed to target each of these mRNAs specifically (Genospectra, Fremont, CA). The results were normalized to corresponding 18S rRNA levels.

IKK kinase assay. Cells were lysed in 25 mmol/L HEPES, 150 mmol/L NaCl, 1% Triton X-100, 10% glycerol, 5 mmol/L EDTA, 2 mmol/L DTT, and protease and phosphatase inhibitor cocktails. IKK was immunoprecipitated from cell lysates by polyclonal anti-NEMO antibody (Santa Cruz Biotechnology). Kinase assay was done in kinase buffer [20 mmol/L HEPES (pH 7.5), 10 mmol/L MgCl₂, 1 mmol/L DTT, 1 mmol/L sodium orthovanadate, 25 mmol/L β-glycerol phosphate] by incubating IKK complex with either NH₂-terminal 72 amino acid of WT IκBα (S32/S36)-GST or mutated IκBα (S32G/S36A)-GST fusion proteins as substrate at 30°C for 30 minutes. The products from the kinase reaction were separated on a 4% to 12% SDS gradient gel and then transferred onto a nitrocellulose membrane. IKK kinase activity was quantified by the intensity of the ³³P-IκBα-GST band detected by autoradiography and normalized as a ratio to the corresponding quantified protein levels of IκBα, and IKK2 and IκBα were detected from immunoblot analysis on the same membrane.

CK2 kinase assay. Cell lysates were prepared as described above for IKK kinase assay. CK2 kinase activity was measured by using a CK2 kinase assay kit with a CK2-specific peptide substrate (Upstate, Charlottesville, VA).³

Total protein (20 μg) of the lysate from each reaction was incubated along with a protein kinase A (PKA) inhibitor and [γ-³³P]ATP in assay dilution buffer for 10 minutes at 30°C. The phosphorylated substrate is then separated by P81 phosphocellulose paper and quantified by a scintillation counter. Active CK2 enzyme was used as positive control. Spike assay was also used to ensure accuracy.

Aberrant activation of NF-κB in UM-SCC cell lines. We showed previously that NF-κB is activated in several human HNSCC cell lines from the UM-SCC series (6, 7). To select representative cell lines for further analysis, we compared NF-κB DNA-binding activity in a panel of nine UM-SCC cell lines derived from seven patients and in nonmalignant human keratinocytes (HEKa). Increased NF-κB DNA-binding activity was detectable in eight of nine UM-SCC cell lines from six of seven patients relative to HEKa (Fig. 1A), similar to the frequency of increased NF-κB activation observed in HNSCC tumor relative to normal squamous mucosa specimens (9). To determine if variations in DNA binding are reflected by variations in functional activation, NF-κB luciferase reporter activity was compared in UM-SCC-1, UM-SCC-6, UM-SCC-9, UM-SCC-11A, and UM-SCC-11B cell lines. Variable NF-κB reporter activities were observed in these cells corresponding to their DNA binding activities (Fig. 1B). Thus, UM-SCC cell lines showed aberrant NF-κB activation, with a prevalence and variability similar to that reported for HNSCC tumor specimens (8, 9).

UM-SCC-6 showed significantly higher NF-κB reporter activity when compared with the other cell lines (P < 0.01), whereas UM-SCC-9 showed significantly lower NF-κB reporter activity compared with the other cell lines (P < 0.05, between UM-SCC-11B and UM-SCC-9; P < 0.01, to the other three cell lines). Protein levels of IκBα, a protein both induced by and degraded with activation of NF-κB, were compared between UM-SCC-6 and UM-SCC-9 cells to determine the relative differences in NF-κB activation (Fig. 1C). Cells were treated with 50 μg/mL cycloheximide for variable times as indicated. IκBα levels were quantified by the intensity of the band on immunoblot and normalized to total protein levels detected on the membrane. Figure 1C shows that the endogenous IκBα protein expression level in UM-SCC-6 cells is ∼1.8-fold that in UM-SCC-9 cells, consistent with the relatively greater activation of NF-κB detected in UM-SCC-6 cells. The data are representative of two independent experiments. UM-SCC-6 cells also showed a ∼50% shorter IκBα half-life (t_1/2; ∼1.5 versus 3 hours; Fig. 1C) and 2-fold difference in IKK kinase activity for phosphorylation of S32/S36 IκBα compared with UM-SCC-9 cells (Fig. 1D). These data confirm that UM-SCC-6 cells exhibit greater NF-κB activity and signal kinase activation by IKK than UM-SCC-9 cells. Based on all of these observations, we chose UM-SCC-6 and UM-SCC-9 for further studies based on the range of strong and weak NF-κB activation detected in these two cell lines.

The IKK1 and IKK2 subunits contribute to NF-κB activation in UM-SCC cells. The IKK complex has previously been shown to mediate phosphorylation of IκBα and regulate canonical activation of NF-κB by proinflammatory cytokines (13, 28). The role of IKK in aberrant NF-κB activation in HNSCC was examined by cotransfection of amino acid–substituted dominant-positive (S176E/S180E, IKK1^EE; S177E/S181E, IKK2^EE), dominant-negative (S176A/S180A, IKK1^AA; S177A/S181A, IKK2^AA), kinase-dead IKK (K44A, IKK1^KA, and IKK2^KA), or WT IKK1 or IKK2 vectors together with cis-pNF-κB-luc in UM-SCC-6 cells (28). Figure 2A and B shows that constitutively active IKK1^EE and IKK2^EE induced over a 20-fold increase in NF-κB reporter activity compared with that of control cells (P < 0.001). Cotransfection and expression of additional WT IKK1 or IKK2 induced an ∼1.5-fold (P < 0.05) and 4-fold (P < 0.001) increase of NF-κB activity, respectively. Thus, IKK^EE and WT both increased NF-κB activity, serving as a positive control for the transfection and responsiveness of the cells. In contrast, the IKK2^KA partially reduced the NF-κB reporter activity by ∼65% (P < 0. 001), whereas IKK1^KA reduced NF-κB reporter activity by ∼50% (P < 0.001). The partial inhibition of NF-κB reporter activity observed with either kinase-dead IKK1^KA or IKK2^KA in UM-SCC-6 cells is consistent with the partial ∼50% to 80% inhibition we have observed in repeated experiments with IKK1^KA or IKK2^KA, IKK1 or IKK2 siRNA, or IKK2 inhibitor PS-1145 in UM-SCC-11A and UM-SCC-11B.⁴

The contribution of both IKK1 and IKK2 to NF-κB activation is also supported by their independent roles in processing of p100 to p52 (29) and degradation of IκBs (28), both of which were observed after cycloheximide treatment in UM-SCC-6 cells (Fig. 1C; data not shown). In addition, cotransfection of the phosphoacceptor mutants IKK2^AA did not inhibit NF-κB activity (P > 0.10) and that of IKK1^AA increased NF-κB activity (P < 0.05) in UM-SCC-6 (Fig. 2A and B), suggesting that activation of these IKKs and NF-κB may occur via a mechanism other than that previously reported for classic NF-κB activation by tumor necrosis factor and interleukin (IL)-1 (28).

CK2 contributes to aberrant NF-κB activation in UM-SCC cells. CK2 has been reported to induce NF-κB activation (14–18). Elevated CK2 activity has been shown in HNSCC and linked to cell proliferation and patient survival (19–21). Treatment of UM-SCC cells with apigenin, an agent that inhibits CK2 kinase activity (30), strongly inhibited NF-κB reporter activity in all four UM-SCC cell lines (Fig. 3A; UM-SCC-6 and UM-SCC-11A, P < 0.001; UM-SCC-9 and UM-SCC-11B, P < 0.05). Depletion of the regulatory subunit CK2β has previously been shown to inhibit CK2-mediated activation of NF-κB by UV light (16). To examine whether CK2 specifically contributes to aberrant NF-κB activity, siRNA targeting CK2β was used. Transfection of UM-SCC-6 and UM-SCC-9 cells with human CK2β siRNA resulted in an ∼90% decrease of CK2β mRNA in UM-SCC-6 cells and ∼75% decrease of that in UM-SCC-9 cells [Fig. 3B for UM-SCC-6 cells (P < 10⁻⁹); Fig. 3C for UM-SCC-9 cells (P < 10⁻⁵)]. The decreased expression of CK2β resulted in a corresponding 82% reduction in NF-κB reporter activity in UM-SCC-6 (Fig. 3D; P < 0.05). We also observed an ∼60% reduction of NF-κB reporter activity in UM-SCC-9 cells, but this did not reach the level of statistical significance, with the variance observed at the low level of activation detected in this cell line (Fig. 3E; P = 0.09). The inhibition of NF-κB reporter activity observed was not due to nonspecific inhibition of mRNA expression by siRNA or transfection, as no significant reduction was observed with unrelated or scrambled control siRNAs. Consistent with CK2β siRNA inhibitory effect on NF-κB reporter activity, we also observed a decrease in expression of NF-κB-inducible genes IκBα (58%; P < 10⁻⁶) and NF-κB2 (68%; P < 10⁻⁷) in anti-CK2β siRNA-transfected UM-SCC-6 cells when compared with that of CK2β scramble control siRNA-transfected cells (Fig. 3F). Comparable cell viability among different siRNA treatments during the assay was monitored by DAPI and propidium iodide stainings to exclude the possibility that effects were the result of cell death (data not shown). Together with data using inhibitor apigenin, these siRNA data indicate that CK2 is involved in increased NF-κB activation and expression of its target genes in UM-SCC-6 cells.

Aberrant NF-κB activation by UM-SCC cells occurs in response to serum. Aberrant NF-κB activation in response to growth factor receptors has previously been observed in breast cancer (31, 32). To explore the contribution of serum to the aberrant activation of NF-κB in UM-SCC cells, we examined the effects on NF-κB reporter activity of replacing FBS with supplemented SFM. A panel of four UM-SCC lines newly changed to SFM (SFM-new) or serially adapted to SFM (SFM-adp) before transfection, each showed a progressive reduction of NF-κB reporter activity when compared with cells grown in 10% FBS that is not heat inactivated (FBS-NHI; Fig. 4A; P < 0.01, for both SFM-new and SFM-adp compared with FBS-NHI; P < 0.01, between SFM-new and SFM-adp). Accordingly, ∼50% decrease of endogenous IκBα mRNA level was also observed in SFM-adp UM-SCC-6 cells compared with that in FBS-NHI cells (P < 10⁻⁵; data not shown). Heat inactivation of FBS (FBS-HI) to 56°C also showed a significant decrease in NF-κB reporter activity (Fig. 4A; P < 0.01, when compared with FBS-NHI). The NF-κB reporter activities from the cells that were grown in FBS-HI are also significantly higher than those in SFM-adp (P < 0.01) but were not statistically different when compared with those grown in SFM-new (P > 0.05). Cotransfection and expression of mitogen-activated protein kinase kinase kinase, an upstream activator of NF-κB, increased NF-κB reporter activity under each of the conditions, indicating that NF-κB in UM-SCC cells remained responsive to signal activation (data not shown). The results shown here represent the average of four experiments. Two different sources of serum were used to exclude the possibility of single batch effects of the serum. Together, these results indicate that NF-κB activation is increased in UM-SCC cells in response to serum and that these serum components are temperature sensitive.

Increased CK2 kinase activity in response to serum in UM-SCC-6 cells. CK2 is known to mediate signal responses to serum (22–26), and we observed that the extent of reduction of NF-κB reporter activity, in the absence of serum, was similar to that observed following inhibition of CK2 by either apigenin or CK2β siRNA. This led to the hypothesis that CK2 may mediate the activation of NF-κB by serum component(s). To examine this hypothesis, we first compared CK2 kinase activities by a CK2-specific peptide substrate from UM-SCC-6 cells that were cultured either in FBS or SFM. CK2 activity from cell lysate that was cultured in SFM was significantly reduced by 50% compared with that in FBS (P < 0.001; Fig. 4B). Applying apigenin resulted in significant further reduction of CK2 kinase activity in both of the cultures (P < 0.05; Fig. 4B) and strongly inhibited a positive control with recombinant CK2α (rCK2α) alone. A PKA inhibitor was used to block nonspecific kinase activity in these assays. Spiking cell lysates with a known amount of rCK2α showed an ∼70% rate of recovery of CK2 activity by the assay (data not shown). Additionally, we also detected a statistically significant ∼30% decrease of CK2β (P < 0.01; Fig. 4C) and an ∼50% decrease of IκBα (P < 10⁻⁶; data not shown) mRNA expression level in the cells that were cultured under SFM compared with that under FBS-NHI, suggesting that a serum-dependent decrease in CK2 expression contributes to reduced NF-κB activity and expression of target gene IκBα. The differences in CK2 kinase activity between SFM and FBS indicate that CK2 activity is increased in response to serum in UM-SCC-6 cells and suggest that CK2 may play an important role in mediating the increased NF-κB activity in response to serum.

Increased IKK activity in response to serum in UM-SCC-6 cells. In Fig. 2, we showed that cotransfection of kinase-dead IKK2^KA inhibited NF-κB activity in UM-SCC-6 cells by ∼65% (Fig. 2A) and that SFM also reduces NF-κB activity in these cells (Fig. 4A). We next examined the effect of serum on IKK kinase activity in UM-SCC-6 cells. IKK kinase activity from whole-cell lysate of UM-SCC-6 cells cultured with FBS or SFM was measured as described in Materials and Methods. An ∼50% decrease of IKK kinase activity was observed from cells that were cultured in SFM when compared with that in FBS in two independent experiments (Fig. 4D; data not shown). The change in IKK kinase activity is consistent with the magnitude of contribution of IKK2 to NF-κB activation observed in Fig. 2. Specific phosphorylation of S32/S36 IκBα, the phosphoacceptor sites for IKK2, was ensured by comparison with a negative control S32G/S36A IκBα mutant substrate (Fig. 4D). Thus, our data indicate that IKK mediates signal phosphorylation of IκBα and activation of NF-κB in response to serum by UM-SCC-6 cells.

CK2 promotes IKK kinase activity. Because the data above indicate that both CK2 and IKK mediate the response to serum and contribute to NF-κB activity in UM-SCC-6 cells, we wondered if CK2 could modulate IKK2-mediated phosphorylation of IκBα. In this regard, protein sequence analysis revealed 11 putative CK2 phosphorylation consensus motifs in IKK2. Based on prediction algorithms, 6 of these motifs had a score (probability to be phosphorylated by CK2) >0.92 on a 0 to 1 scale (data not shown). This analysis suggested that IKK2 could be a potential substrate of CK2; hence, IKK kinase activity could be regulated by CK2. We examined whether IKK2 is a substrate of CK2 in an in vitro CK2 kinase assay by using rCK2α and recombinant IKK2 (rIKK2; Fig. 5A). In this assay, we used 0.5 and 5 ng of rCK2α with rIKK2 (200 ng/reaction). Because CK2 can use both ATP and GTP, to ensure the CK2 specificity, we used GTP rather than ATP as a phosphodonor. Our result showed a low level of autophosphorylation by IKK2, which may be caused by the high concentration of rIKK2 used. This was confirmed by comparing GTP with ATP as a phosphodonor, which showed more dramatic autophosphorylation by IKK2 itself (data not shown). On top of this IKK2 autophosphorylation, we detected a rCK2α dose-dependent increase in phosphorylation of rIKK2. In the presence of 0.5 and 5 ng rCK2α, an ∼40% and a 2.2-fold increase of normalized phosphorylated IKK2 was observed when compared with control. Apigenin completely suppressed CK2α phosphorylation of IKK2. The result shown here is an average of two parallel measurements. Thus, these data provide direct evidence that IKK2 is a substrate of CK2.

To determine if CK2 could modulate IKK2 kinase activity as part of the IKK complex, anti-NEMO antibody immunoprecipitated IKK complexes were incubated with different amounts of catalytically active rCK2α and GTP under the conditions optimized for CK2 kinase assay followed by washing out of CK2 before the IKK kinase assay. Because the substrate IκBα-GST fusion protein used only contains the NH₂-terminal 72 amino acids, the assay excludes that CK2 phosphorylation is due to the IκBα COOH-terminal PEST domain. Again S32G/S36A mutant IκBα-GST was used as a negative control to ensure IKK-specific phosphorylation of the S32/S36 of the NH₂-terminal IκBα. The kinase activities were normalized to protein levels of IκBα and IKK2. Figure 5B shows the rCK2α dose-dependent increase of IKK kinase activity detected as IKK-mediated phosphorylation of WT but not S32G/S36A mutant IκBα-GST. A 13% and 60% increase of IKK kinase activities from those that were preincubated with 0.5 and 5 ng rCK2α, respectively, was observed compared with that of nontreated immunoprecipitated IKK complexes. When the immunoprecipitated IKK complex was incubated with 5 ng rCK2α in the presence of 40 μmol/L apigenin, the CK2α-induced IKK kinase activity was completely blocked. An ∼70% increase in IKK kinase activity was detected with saturating amounts of rCK2α (10 and 20 ng) in an independent experiment (data not shown). Because IKK2 phosphorylation of IκBα is dependent on complexing with the NEMO subunit, these data provide evidence supporting the hypothesis that CK2 may directly regulate IKK2 activity for phosphorylation of IκBα in UM-SCC-6 cells.

We discovered previously that NF-κB is aberrantly activated and promotes tumorigenesis in human HNSCC and murine SCC (4–8, 10). NF-κB activation has been broadly shown and associated with progression in intraepithelial premalignant and malignant squamous neoplasms of the head and neck as well as uterine cervix (9, 33). We have shown that expression of an S32A/S36A IκBα mutant unresponsive to IKK phosphorylation strongly inhibited NF-κB activation in HNSCC and murine SCC (7, 10). Tamatani et al. (34) detected increased IKK-mediated phosphorylation of IκBα and NF-κB activation in three HNSCC cell lines relative to that observed in five primary gingival keratinocyte cultures, indicating that IKK and NF-κB are aberrantly activated together in HNSCC. Gapany et al. (19) showed that cytosolic CK2 expression and activity are also increased in HNSCC tumor specimens and cell lines. In the present study, we directly examined the hypothesis that CK2 promotes IKK-mediated aberrant activation of NF-κB in HNSCC. NF-κB activation was specifically inhibited by siRNA targeting the β subunit of CK2 and kinase-dead mutants of the IKK1 and IKK2 subunits, implicating both the alternative and the classic IKK pathways in NF-κB activation. We found that CK2 contributes to the activation of IKK and NF-κB in response to serum factor(s). rCK2α was shown to phosphorylate rIKK2 as well as to promote immunoprecipitated IKK complex from HNSCC to phosphorylate the NH₂-terminal S32/S36 of IκBα. We conclude that the aberrant NF-κB activity in HNSCC cells in response to serum is partially through a novel mechanism involving CK2-mediated activation of IKK2, making these kinases candidates for selective therapy to target the NF-κB pathway in HNSCC. This is the first study to identify a potential mechanism linking the independent clinical pathologic observations that CK2 is an upstream regulator of IKK and NF-κB activation in HNSCC.

Several observations from this and other studies in our laboratory suggested an alternative mechanism of activation and role of IKK in NF-κB activation in HNSCC. First, we observed that both IKK1 and IKK2 contribute to NF-κB activation. The inhibition of NF-κB reporter activity in different UM-SCC cell lines by specific inhibition of either IKK2 with kinase-dead IKK2^KA, IKK2 siRNA, or IKK2 inhibitor PS-1145 or by that of IKK1 with IKK1^KA or IKK1 siRNA along was significant, reproducible, but incomplete. Although IKK2 has been shown to mediate activation of NF-κB1/RelA (p50/p65), IKK1 has been reported to promote processing of p100 to p52 (NF-κB2), an alternative NF-κB activation pathway (29). Consistent with this, we found evidence for higher expression levels of endogenous NF-κB-inducible genes IκBα and NF-κB2 (data not shown), increased degradation rate of IκBs, and processing of p100 following cycloheximide treatment in UM-SCC-6 cells (data not shown). Furthermore, the IKK2^AA mutant did not inhibit NF-κB activation, and the IKK1^AA mutant enhanced activation of NF-κB activity. These results suggested that the signaling mediated by these IKK subunits in HNSCC may result from signal(s) and/or mechanism(s) that are distinct from those mediating classic activation of IKK and NF-κB (28).

CK2 is a highly conserved pleiotropic and ubiquitous serine and threonine kinase with a wide range of substrates involved in carcinogenesis and tumor progression. More than 300 CK2 substrates have been identified (13, 35), and the majority of which are proteins that are involved in transcription, cell cycle regulation, cell proliferation, cell survival, gene expression, and signal transduction. CK2 phosphorylation has been shown to inhibit apoptosis and favor cell proliferation and oncogenic transformation (13, 36). In this study, we provide evidence that CK2 is a key mediator of overall NF-κB activation and functions to enhance IKK kinase phosphorylation of IκBα. The CK2 inhibitor apigenin or specific siRNA targeting CK2β was sufficient to inhibit NF-κB activity, indicating that the overexpression of CK2 and increased CK2 activity found in prior studies (19, 20) may be responsible for mediating the aberrant activation of NF-κB in HNSCC.

CK2 has previously been shown to phosphorylate multiple sites in the COOH-terminal PEST domain of IκBα, including the response involved in UV light-induced NF-κB activation (14, 16). CK2 has also been reported to phosphorylate S529 of the RelA/p65 subunit (17, 18). In addition to these targets, we have found that catalytically active rCK2α can phosphorylate both rIKK2 and rIKK1 (data not shown) in vitro and that incubation of IKK complex with rCK2α resulted in increased IKK2 kinase activity for the phosphorylation of S32/S36 NH₂ terminus of IκBα, a novel and distinct function from that of CK2 as a known COOH-terminal PEST domain IKK. This finding is also supported by protein sequence analysis of IKK1 and IKK2, which reveals multiple CK2 phosphorylation motifs. Additionally, the decreased IKK kinase activity in UM-SCC-9 and SFM-cultured UM-SCC-6 cells were associated with lower expression levels of CK2β in these cells, adding another line of evidence that CK2 is an upstream regulator of IKK and NF-κB activation.

CK2 has been linked to aberrant NF-κB activity in cancer cells derived from human breast, hepatic, and colon carcinomas (30, 37–41). However, the relative contribution of CK2 to NF-κB activation seems to vary among different cancers. Compared with CK2 activation of NF-κB in other type of cancer cells (30, 37–41), our CK2β siRNA data indicate a dominant effect of CK2 as a holoenzyme on NF-κB activation in UM-SCC cells. On the other hand, our data do not rule out the possibility of direct effects of CK2 on activating the NF-κB pathway at other levels, as multiple CK2 phosphorylation motifs have been identified in every member of the NF-κB and IκB families. Thus, the presence of CK2 target sites in IκB and p65 as well as the potential sites in IKK and other NF-κB and IκB family members could explain why we observed nearly complete inhibition of NF-κB activity by blocking CK2 activity. More detailed characterization of the molecular basis of the interaction between CK2 and IKK is needed and may lead to additional specific targets for cancer treatment.

An important finding of the present study is the demonstration that CK2 and IKK mediate the altered activation of NF-κB in response to serum factor(s). The increase in CK2 and NF-κB activation in the majority of HNSCC specimens relative to normal mucosa (9, 20) indicates that the response of these pathways to serum factor(s) is altered and precedes culture rather than being a mere consequence of serum induction in culture. Indeed, we have shown that the increased NF-κB activity in murine SCC cells occurs with tumor progression in vivo and favors tumorigenesis and metastasis in the host environment (5). Consistent with this, the increased nuclear localization of NF-κB observed in human squamous dysplasias is associated with higher risk of progression in a recent clinical pathologic study (9).

The factor(s) in serum and/or the host environment that contribute to induction of CK2, IKK, and NF-κB remain to be elucidated and may provide additional targets for therapy. Autocrine factors produced by HNSCC (42, 43) or paracrine factors produced by tumor stromal fibroblasts (44) have been shown to enhance activation of NF-κB in HNSCC. Several components contained in serum, including cytokines, growth factors, different types of albumins, zinc, copper, and free thiols, have been reported to affect NF-κB activity (13, 24–26, 31, 43–49). Because the serum factors contributing to NF-κB activation in UM-SCC were heat sensitive, labile protein factors seem the most likely candidates.

We have previously evaluated epidermal growth factor (EGF) and cytokine IL-1 as possible candidates (42, 43) because they are factors that are known to activate NF-κB in nonmalignant and malignant cells. Although NF-κB was found to be inducible by EGF, C225, the specific antagonist of anti-EGF receptor (EGFR) antibody, inhibited only inducible but not aberrant NF-κB activity, making EGFR an unlikely candidate for the aberrant NF-κB activation in HNSCC (42). We have previously obtained evidence that the IL-1/IL-1 receptor (IL-1R) pathway may be one of the stimuli contributing to the aberrant activation of NF-κB in HNSCC. We found that transient expression of an intracellular form of the IL-1R antagonist could significantly inhibit NF-κB reporter activity and cytokine IL-8 gene expression in UM-SCC-9 and UM-SCC-11B cell lines (43). Recently, interruption of the IL-1 signal pathway by expression of siRNA knocking down expression of IL-1R1, or a dominant-negative mutant of the essential Toll receptor linker MyD88, was found to strongly inhibit NF-κB activation in five of six UM-SCC lines, including the UM-SCC-6, UM-SCC-9, UM-SCC-11A, and UM-SCC-11B cell lines.⁵

Interestingly, a recent study in malignant keratinocyte lines suggests that the IL-1 pathway and intracellular IL-1R antagonist may play a role in regulating signal degradation of IκBα and other transcription factors by the COP9 (CSN) signalosome, such as CK2 (50). Such an alternative pathway for activation of NF-κB by IL-1 and possibly other factors could explain the enhancement of NF-κB reporter activity by IKK1^AA and lack of inhibitory effect we observed with IKK2^AA, both of which are deficient in the phosphoacceptor sites mediating classic activation of NF-κB. It would be interesting if this signalosome containing CK2 interacts with the IKK signalosome and other NF-κB pathway components. Dissection of the role of CK2 or other intermediate kinases as possible oncogenes mediating aberrant activation of IKK and NF-κB in response to exogenous factors in HNSCC may also be important to development of therapy.

In summary, we found that CK2 contributes to the activation of IKK and NF-κB in response to serum factor(s), which suggests that CK2 and IKK2 are key candidates for targeting the NF-κB pathway in HNSCC. This is the first study to identify a potential mechanism linking the independent clinical pathologic observations that CK2 and NF-κB are activated and associated with decreased prognosis in HNSCC. Further clinical and molecular studies are indicated to confirm the role and further elucidate the factor(s) and mechanisms involved in CK2 activation of IKK and NF-κB in HNSCC.

Grant support: National Institute on Deafness and Other Communication Disorders Intramural Research Project Z01-DC-00016 (C. Van Waes) and Howard Hughes Medical Research Institute-NIH Scholars Program (J. Yeh).

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.

We thank Keith Brown (National Institute of Allergy and Infectious Disease) and David Gius (National Cancer Institute, Bethesda, MD) for their critical review and comments.