Molecular Cross-Talk between the NFκB and STAT3 Signaling Pathways in Head and Neck Squamous Cell Carcinoma

DNA Constructs and Reagents

The dominant-negative IκB A32/36S super-repressor mutant (kindly provided by Dr. Siebenlist) [16] was cloned in-frame with a green fluorescent protein (GFP) coding sequence into the pCEFL expression vector, generating a pCEFL-GFP-IκB A32/36S plasmid. The lentivirus-encoding dominant-negative GFP-IκB and GFP genes were generated using the gateway system from Invitrogen Life Technologies (Carlsbad, CA) and CSCG-based retroviral vectors, as previously reported [17,18]. The full-length IL-6 promoter and the point mutations in the AP-1 and NFκB regulatory elements of the IL-6 promoter (kindly provided by Dr. Libermann) [19] were inserted into the KpnI-XhoI sites of the pGL3 luciferase reporter vector by polymerase chain reaction (PCR), generating pGL3-IL-6-Luc and its mutants. For report assays, PGL3-NFκB-Luc, in which luciferase expression is controlled by five NFκB response elements, was also used. The plasmid pcDNA3-β-galactosidase (BD Biosciences, San Jose, CA) and pCEFL-GFP were used as controls. Recombinant human IL-6 was purchased from PeproTech, Inc. (Rocky Hill, NJ), and tumor necrosis factor α (TNF-α) was from Roche (Indianapolis, IN). Cells were treated with a single dose of IL-6 or TNF-α (10 ng/ml) for 10 minutes.

Cell Lines and Culture Conditions

The HNSCC cell lines HN6, HN12, HN13, and HN30 [20]; HaCaT cells [21]; an immortalized nontumorigenic human skin keratinocyte cell line; and HEK293T and HEK293FT cells were cultured in DMEM supplemented with 10% fetal calf serum, penicillin, and streptomycin. The cells were maintained in a 5% CO₂-humidified incubator. Stable cell lines were established by lentiviral infection using CSCG-based retroviral vectors and 293FT cells as packaging cells. HN13 and HN30 cells were infected with viral supernatants for 24 hours at 37°C in the presence of 8 µg/ml polybrene (hexadimethrine bromide; Sigma, St. Louis, MO). Conditioned media (CM) from HaCaT and HNSCC cell lines were prepared by incubating subconfluent cultured cells for 24 hours in DMEM without serum and supplements. Harvested CM were then filtered through a 0.22-µm low-protein-binding polyethylsulfonate membrane filter.

Transcription Factor Analysis

Nuclear extracts were prepared using a NE-PER Nuclear and Cytoplasmic Extraction reagent (Pierce Biotechnology, Inc.). The double-stranded oligonucleotide consensus sequence for NFκB (5′-AGTTGAGGGGACTTTCCCAGGC-3′) and activating protein 1 (AP1; 5′-CGCTTGATGACTCAGCCGGAA-3′) (Santa Cruz Biotechnology, Inc., Santa Cruz Biotechnology, CA) was labeled with [γ³²P]ATP using T4 polynucleotide kinase (Invitrogen Life Technologies), purified, and added to the reactions (20,000 cpm/reaction) for 15 minutes. Complexes were analyzed on nondenaturing (4.5%) polyacrylamide gels. For supershift assays, 1 µg of anti-NFκB (C-20) antibody (Santa Cruz Biotechnology, Inc.) was added to the binding reaction before the addition of a radiolabeled probe. In separate experiments, to identify the NFκB subunit proteins in the complex, 2 µl of p65 antibody was incubated with nuclear proteins before the addition of the radiolabeled probe to visualize any supershift-retarded bands in the NFκB complex. Protein-DNA complexes were separated by electrophoresis on 4% polyacrylamide gel in Tris-borate-EDTA buffer. NFκB DNA-binding activity assays were also performed using Trans-AM enzyme-linked immunosorbent assay (ELISA)-based kit from Active Motif (Carlsbad, CA), according to the manufacturer's protocol. Briefly, cell extracts were incubated in a 96-well plate coated with an oligonucleotide containing the NFκB consensus-binding site. Activated transcription factors from extracts specifically bound to the respective immobilized oligonucleotides were detected using antibodies to NFκB p65, followed by a secondary antibody conjugated to horseradish peroxidase (HRP) in an ELISA-like assay.

IκB Kinase (IKK) Knockdown

Cells were seeded in 24-well or 6-well plates and, on reaching 70% confluence, the cells were washed twice in serum-free medium and transfected with 12.5 nM double-stranded RNA oligonucleotides directed against human IKKα (NM_001278, SI00605115; Qiagen, Valencia, CA) and with 25 nM against IKKβ (NM_001556, SI02777376; Qiagen) using HyperFect (Qiagen), following the manufacturer's instructions. Optimal concentrations of siRNA and time points were determined by performing a dilution curve of siRNA for each target, and knockdown was determined by Western blot analysis. After 72 hours, the cells were treated as indicated, and nuclear extracts were isolated as described above. The sequences of the negative siRNA (Qiagen) oligonucleotides used as controls were as follows: 5′-UUCUCCGAACGUGUCACGUdTdT-3′ and 5′-ACGUGACACGUUCGGAGAAdTdT-3′.

Immunoblotting

Cultured cells were harvested at the times indicated in each experiment, washed with phosphate-buffered saline, and lysed in 62.5 mM Tris, 2% sodium dodecyl sulfate (SDS), and 10% glycerol (SDS sample buffer) with aprotinin, leupeptin, pepstatin, and 4-(2-aminoethyl) benzenesulfonyl fluoride. Cell suspensions were briefly sonicated, and the protein concentration for each cell lysate was determined using DC protein assay (Bio-Rad, Hercules, CA). Twenty to 40 µg of total protein from whole cell lysates was loaded onto each lane for gel electrophoresis. Immunoblotting was performed using 0.1 M Tris (pH 7.5), 0.9% NaCl, 0.05% Tween20 with 5% nonfat dry milk as a blocking and antibody-dilution buffer, and working antisera for STAT3, phospho-STAT3 (pSTAT3)-tyrosine 705, IKKα, IKKβ, phospho-IKKα/β (Cell Signaling Technology, Beverly, MA), GFP (Covance, Inc., Denver, PA), and tubulin (Santa Cruz Biotechnology, Inc.). We used 62.5 mM Tris (pH 6.8), 2% SDS, and 100 mM β-mercaptoethanol to strip probed filters, as indicated in each experiment.

Immunohistochemistry

Individual paraffin blocks of formalin-fixed tissues from human squamous cell carcinomas of the oral cavity and HNSCC tissue arrays containing approximately 460 cases of oral SCC and normal control tissues were obtained from the National Institute of Dental and Craniofacial Research, National Institutes of Health Oral Cancer Tissue Array Initiative (Bethesda, MD) (Molinolo et al., in preparation). Briefly, individual paraffin-embedded HNSCC tissues from the oral cavity were arrayed and processed following the standard methodology of the Tissue Array Research Program (www.cancer.gov/tarp). Immunohistochemistry was performed as previously described [22]. In brief, the slides were dewaxed in xylene and hydrated through graded alcohols. Antigen retrieval was performed using 10 mM citrate buffer (pH 6.0) placed in a microwave oven for 20 minutes (2 minutes at 100% power and 18 minutes at 10% power). The slides were allowed to cool down for 30 minutes at room temperature, rinsed twice with tris-buffered saline solution (TBST), and incubated in 3% hydrogen peroxide for 30 minutes to quench the endogenous peroxidase. The sections were then incubated in blocking solution (5% bovine serum albumin) for 1 hour at room temperature, followed by treatment with anti-IL-6 primary antibody (clone 6708.11; Sigma), NFκB p65 (A) antibody (Santa Cruz Biotechnology, Inc.), and pSTAT3^Y705-specific antibody (Santa Cruz Biotechnology, Inc.), where indicated, overnight at 4°C. After washing with TBST, the slides were incubated with the labeled streptavidin biotin reagent (LSAB+system HRP; DAKO Corporation, Carpinteria, CA), following the manufacturer's instructions. The slides were developed in 3,3-diaminobenzidine (Sigma FASTDAB tablet; Sigma) and counterstained with Mayer's hematoxylin. Images were taken using a SPOT digital camera attached to a Zeiss Axiophot microscope (Carl Zeiss, Thornwood, NY). All immunostainings were assessed by two experienced pathologists. The tumors were classified as positive or negative (no staining). Statistical analysis was performed by two-tailed Fisher's exact test, and P ≤ .05 was considered statistically significant.

Reporter Assays

The NFκB and IL-6 promoter reporter constructs containing the firefly luciferase cDNA (0.1–0.3 µg/well) and the pRL-null normalization construct (0.01–0.03 µg/well) containing Renilla luciferase from Renilla reniformis were transfected into HaCaT, HNSCC, and HEK293T cells (40% confluent in six-well plates). The total amount of plasmid DNA was adjusted with pcDNA3-β-galactosidase. When indicated, the cells were also transfected with pCEFL-p65 plasmid and pretreated with 10 ng/ml TNF-α or left untreated for 10 minutes. The measurement of firefly and Renilla luciferase activities present in cellular lysates was carried out using the dualluciferase reporter assay (Promega, Madison, WI) 24 hours after adding DNA. Light emission was quantitated using a Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego, CA). Data were presented as firefly luciferase activity normalized by the Renilla luciferase activity present in each sample, and the values plotted were the average ± SE of triplicate samples from typical experiments, which were repeated at least three to five times with nearly identical results.

Chromatin Immunoprecipitation (ChIP) Assays

ChIP assays were performed using the ChIP Assay Kit (Upstate Biotechnology, Waltham, MA). Briefly, HN cancer cells were plated on two 100-mm dishes; cross-linking was performed by adding formaldehyde directly to a tissue culture medium to a final concentration of 1% and by incubating for 10 minutes at room temperature; and nuclear extracts were isolated and sonicated to generate DNA fragments. Transcription factors bound to chromatin were immunoprecipitated with the specific antibody anti-NFκB (C-20; Santa Cruz Biotechnology, Inc.), protein-DNA cross-linking was reversed, and isolated genomic DNA was amplified by PCR, using specific primers encompassing the NFκB region of the IL-6 promoter. PCR reactions were performed using 2 µl of a 50-µl DNA extraction in Tris-EDTA buffer with Hi-Fi Taq polymerase (Invitrogen Life Technologies). PCR mixtures were amplified for 1 cycle at 94°C for 2 minutes; followed by 32 cycles at 94°C for 30 seconds, annealing temperature of 55°C for 30 seconds, and 68°C for 1 minute; and then subjected to a final elongation at 68°C for 5 minutes. The primers used were hIL-6P-NFκB-F3 (5′-GCTAGCCTCAATGACGACCT-3′) and hIL-6P-NFκB-R3 (5′-GCCTCAGACATCTCCAGTCC-3′), which amplify 227 bp of the hIL-6 promoter surrounding the NFκB site. hGAPDH (sense: 5′-CCCCACACACATGCACTTACC-3′; antisense: 5′-CCTAGTCCCAGGGCTTTGATT-3′), which amplified a fragment with an expected size of 96 bp, served as a control.

Protein Secretion Analysis

CM from stably infected HNSCC cell lines and HaCaT cells were prepared by incubating subconfluent cultured cells for 24 hours in DMEM without supplements. Harvested CM were then filtered through a 0.22-µm low-protein-binding polyethylsulfonate membrane filter. IL-2, IL-5, IL-6, IL-8, IL-10, IL-12(p70), IL-13, IL-17, GM-CSF, and G-CSF proteins in the supernatant were assayed by the customized quantitative multiplexed sandwich ELISA Searchlight (Pierce Biotechnology, Inc.)

HNSCC Cell Lines Constitutively Express IL-6: Correlation with NFκB Activity

To begin exploring the mechanism(s) responsible for the elevated IL-6 release from HNSCC, we first examined a representative set of HNSCC for IL-6 secretion. As shown in Figure 1A, only a limited amount of IL-6 was found in CM from an immortalized nontumorigenic epithelial cell line, HaCaT. In contrast, three representative HNSCC lines, HN12, HN13, and HN30, secreted remarkable levels of IL-6. Only one HNSCC from the many cells studied did not secrete IL-6 to the cultured medium HN6, thus serving as an internal control. IL-6 mRNA levels paralleled the pattern of IL-6 secretion in these cells (not shown). Because of the proposed role of NFκB in IL-6 expression and secretion in other cellular systems [23], we next determined the activity of NFκB in these cell lines. Under normal cultured conditions, the nuclear extracts from HN12, HN13, and HN30 cells exhibited constitutively active NFκB activity, as judged by its ability to bind to NFκB consensus oligonucleotides (Figure 1B), whereas HaCaT and HN6 showed nearly undetectable levels of protein-DNA complexes. As a control, the treatment of HaCaT and HN6 cells with TNF-α led to enhanced NFκB activity. However, only a slight activation was observed in HN12, HN13, and HN30 on TNF-α stimulation, aligned with the fact that these cells exhibit elevated levels of NFκB activity already under basal conditions. To confirm these observations, we assessed the activity of NFκB in each cell line. Indeed, as shown in Figure 1C, using an electrophoretic mobility shift assay (EMSA), we observed that nuclear extracts from HN12, HN13, HN30, and TNF-α-treated HaCaT retarded the mobility of NFκB consensus oligonucleotides in acrylamide gels. The identity of NFκB in these protein-DNA complexes was confirmed by the supershift in the apparent mobility of the complex through the addition of antisera against the p65 subunit of NFκB (Figure 1C). Thus, the remarkable secretion of IL-6 in this panel of HNSCC cell lines correlated with their constitutive activity of NFκB.

Deregulated IL-6 Secretion in HNSCC Cells Correlates with Enhanced Activity of the IL-6 Promoter: A Role for the NFκB and AP1 Response Elements

To investigate whether the underlying mechanism leading to deregulated IL-6 expression involves enhanced transcription from the IL-6 promoter, we inserted the wild-type (WT) 1.2-kb IL-6 promoter (Figure 2A) and an IL-6 promoter containing a point mutation in its NFκB response element (Figure 2B) upstream of the firefly luciferase gene in the pGL3 reporter plasmid. As a control, we also used an IL-6 promoter, including a point mutant in its AP1 site and in both NFκB and AP1 sites. When we transfected these constructs in HEK293T cells, the IL-6 promoter was greatly stimulated by the coexpression of the p65 NFκB subunit. In contrast, as expected, the IL-6 promoter harboring a point mutation in its NFκB site did not respond to p65. Of interest, the AP1mutated IL-6 promoter still responded to p65, but the level of luciferase expression was slightly reduced, suggesting a potential interaction between these two transcriptional response elements.

The reporter plasmids under the control of the WT IL-6 promoter and its mutants were then transiently transfected in the HNSCC lines. For these experiments, we selected HN13 and HN30 HNSCC cells, as HN12 cells behaved very similarly to HN30, and we used HaCaT cells as controls. When compared with luciferase expression from the reporter plasmid without the IL-6 promoter, HaCaT cells showed a fourfold activation of the IL-6 promoter. In contrast, HN13 and HN30 both showed a 16-fold activation of the IL-6 promoter. Mutations in the NFκB-binding site drastically reduced IL-6 promoter activity in all cell lines, indicating a critical role for this transcription factor in IL-6 promoter activity. A less pronounced reduction in IL-6 promoter activity was observed with the point mutant in the AP1 site, particularly in HN13. These observations suggested that NFκB is critical for the activation of IL-6 in all cells, but that, in HN13cells, AP1 may cooperate with NFκB to regulate IL-6 expression (Figure 2C). To examine this possibility, we analyzed the activity of AP1 in HNSCC cell lines by EMSA. AP1 activity was particularly highly active in HN13 cells, as demonstrated by its ability to bind to AP1 consensus oligonucleotides, whereas most cells showed low levels of AP1 protein-DNA complexes. As a positive control, HaCaT cells showed a high AP1 nuclear activity on TNF-α stimulation (Figure 2D).

As both NFκB and AP1 were required to stimulate the IL-6 promoter in HN13 cells, we wanted to confirm whether endogenous NFκB indeed binds to the IL-6 promoter in vivo in these cells. Therefore, we immunoprecipitated NFκB cross-linked to chromatin and amplified the human IL-6 promoter using PCR primers surrounding the NFκB-binding site (ChIP assay). As shown in Figure 2E, whereas PCR amplification of total genomic DNA from control, HaCaT, and HN13 cells yielded a fragment of 217 bp, no amplification was observed in NFκB immunoprecipitates from HaCaT cells, unless they were stimulated with TNF-α. In contrast, IL-6 promoter sequences were readily amplified in NFκB immunoprecipitates from HN13 cells (Figure 2E), demonstrating that this transcription factor is constitutively bound to the IL-6 promoter in vivo.

IKKα and IKKβ Contribute to p65 Constitutive Activation in HNSCC

The basal activity of p65 NFκB is repressed by its association with IκB, an ankyrin repeat-containing protein that binds to NFκB and masks its nuclear localization signal, thus retaining NFκB in its inactive state in the cytosol. The activation of NFκB involves the phosphorylation of IκB on two serine residues, Ser32 and Ser36, which triggers the rapid ubiquitination and degradation of phosphorylated IκB in the proteosome, and the consequent nuclear translocation and activation of NFκB [16]. IKK includes a regulatory subunit, NEMO (IKKγ), and two catalytic kinases subunits, IKKα (IKK1) and IKKβ (IKK2), which are lysates readily detected from HaCaT, HN12, HN13, and HN30 cells (Figure 3A). Whereas no phosphorylated IKKα/β could be detected in HaCaT cells, the levels of phosphorylated (active) IKKα/β increased dramatically in these cells on TNF-α stimulation (Figure 3A). In contrast, HN12, HN13, and HN30 showed an expression of phosphorylated IKKα/β under basal conditions. To analyze the contribution of IKKs in the constitutive activation of p65 in these cells, we interfere with endogenous IKKa and IKKβ expression using siRNA. Using the interference in HN13 as an example, we observed that IKKα- and IKKβ-specific siRNA effectively knocked down the expression of corresponding kinases without altering the expression of the untargeted IKK isoform (Figure 3B). Similar results were obtained in other cell lines (not shown). When siRNA were transfected into HaCaT cells, the inhibition of IKKα and IKKβ expression drastically reduced p65 activity (P < .001). Similarly, the knockdown of these IKKs reduced the activation of p65 by TNF-α (P < .05) (not shown). Furthermore, the constitutive activation of NFκB was also repressed in the HNSCC cell lines HN13 and HN30 by the knockdown of IKKs, although, in general, the most striking results were obtained by the interference of IKKβ. Indeed, IKKβ knockdown reduced the constitutive activation of p65 in both HN13 (P < .05) and HN30 (P < .01) (Figure 3C).

Inhibition of NFκB Diminishes IL-6 Promoter Activity and IL-6 Protein Expression

To further analyze the contribution of NFκB to IL-6 secretion from HNSCC cells, we asked whether interference with NFκB function affected the production and release of IL-6 from these cells. As both IKKα and IKKβ can contribute to NFκB signaling in HNSCC, and (as previously described) [24–26] as IKKα and IKKβ have unique activities but can partially compensate for each other's function, we chose to explore the role of NFκB through the use of the super repressor IκB. The super repressor IκB harbors a substitutionκ of Ser32 and Ser36 for alanine, causing the blockade of IκB proteolysis and, consequently, the sequestration of NFκB in the cytoplasm, thereby preventing the translocation of NFκB to the nucleus and its transcriptional activity [16]. Initially, we observed that this IκB repressor inhibited transcription from an NFκB reporter plasmid in HaCaT cells in a doseαdependent matter, both under basal conditions or on exposure to TNF-α, using a plasmid encoding the β-galactosidase gene as a control (Figure 4A). The IκB repressor also inhibited the activation of the IL-6 promoter by TNF-α in HaCaT cells (Figure 4B). Similarly, this repressor of NFκB inhibited the expression of luciferase when driven by the isolated NFκB response element or the WT IL-6 promoter in both HN13 and HN30 cells (Figure 4, C and D).

We next engineered lentiviral constructs encoding the IκB super repressor fused to GFP (Figure 5A) to analyze the consequences of stably inhibiting NFκB in HNSCC cells. As depicted in Figure 5B, the protein product of the GFP-IκB super repressor was readily detected in HaCaT cells infected with corresponding lentiviruses, using GFP virus as control. Similarly, these constructs were highly expressed in HNSCC cells, as visualized by the fluorescent detection of GFP and its fusion protein (Figure 5, C–F). As expected, GFP-IκB S32/36A was restricted primarily to the cytoplasm, whereas GFP also exhibited strong nuclear distribution. Infection of HNSCC cells with the lentivirus encoding the super suppressor IκB S32/36A dramatically diminished the level of IL-6 secreted in the supernatants of both HN30 and HN13 HNSCC cell lines, further supporting the notion that NFκB is crucial for IL-6 promoter activity and protein expression.

Inhibition of NFκB Diminishes Cytokine Release from HNSCC Cells

As NFκB plays a key role in the regulation of the expression of numerous inflammatory cytokines, we next explored the contribution of NFκB to the release of cytokines known to be expressed in HNSCC cells, in addition to IL-6. Indeed, on infection with GFP-IκB, we observed a dramatic decrease in the release of key cytokines from HNSCC cells, including IL-2, IL-6, IL-8, IL-10, IL-12, GM-CSF, and G-CSF (Figure 6). Interestingly, there appears to be a more marked inhibition of certain cytokines in HN13 cells than in HN30 cells, as, for example, IL-8, IL-12, and GM-CSF were inhibited by less than 30% in HN30 cells, whereas they were greatly reduced in HN13 cells.

Inhibition of NFκB Blocks the Autocrine/Paracrine Activation of STAT3

The remarkable effect of NFκB inhibition on the secretion of IL-6 and other cytokines from HNSCC cells prompted us to examine the contribution of NFκB to the potential autocrine/paracrine activation of the STAT3 signaling pathway in HNSCC cells. As shown in Figure 7A, inhibition of NFκB did not affect the expression levels of STAT3 in these cells, but diminished dramatically the levels of tyrosine-phosphorylated active STAT3 (pSTAT3^Y705). Furthermore, whereas CM from HN13 and HN30 promote the accumulation of activated pSTAT3^Y705 when added to HaCaT cells, we observed that the inhibition of NFκB in HN13 and HN30 HNSCC cells and the consequent reduction in IL-6 and other cytokine expressions diminished the ability of their CM to activate STAT3 (Figure 7B).

Nuclear Localization of NFκB Correlates with IL-6 Expression and STAT3 Activation in HNSCC

These findings prompted us to explore whether the activation status of NFκB correlates with the expression of IL-6 and the phosphorylation of STAT3 in HNSCC tumors using an immunohistochemical approach. We detected a robust immunohistochemical stain for p65 in the cytoplasm and the nucleus (Figure 8A), which is more clearly seen at a higher magnification (Figure 8B). The nuclear staining of p65 was present in all areas of the carcinoma, including invading islands. Indeed, NFκB was present in most of the cases analyzed (n = 196) (Figure 8F). pSTAT3^Y705 was also expressed in the majority of tumor cells (Figure 8C). A close-up view demonstrates the specific nuclear staining of their phosphorylated form of STAT3 (Figure 8D). IL-6 presented a generalized cytoplasmic immunoreactivity (Figure 8E), which was positive in a large fraction of cases (n = 144) (Figure 8F). Immunohistochemical analyses of human HNSCC tissue array samples demonstrated an excellent correlation in vivo between the expressions of NFκB, IL-6, and pSTAT3 in the same tissue tumor cores (Table 1). Indeed, a significant correlation was present between NFκB and IL-6 (P ≤ .02), between IL-6 and pSTAT3^Y705 (P ≤ .001), and between NFκB and pSTAT3^Y705 (P ≤ .005).

Together, these data are aligned with our observations in HNSCC cells and support the existence of the autocrine/paracrine activation of STAT3, which can be initiated by the NFκB-dependent release of IL-6 and other inflammatory cytokines from HNSCC cells (Figure 9).