Although constitutively activated nuclear factor-κB (NF-κB), attenuated transforming growth factor β (TGFβ) signaling, and TP53 mutations frequently occur in human cancers, how these pathways interact and together contribute to malignancy remains uncertain. Here, we found an association between overexpression of NF-κB–related genes, reduced expression of TGFβ receptor (TβR) subunits and downstream targets, and TP53 genotype in head and neck squamous cell carcinoma (HNSCC). In response to recombinant TGFβ1, both growth inhibition and TGFβ target gene modulation were attenuated or absent in a panel of human HNSCC lines. However, in HNSCC cells that retained residual TGFβ signaling, TGFβ1 inhibited both constitutive and tumor necrosis factor α–stimulated NF-κB activity. Furthermore, HNSCC lines overexpressing mutant (mt) TP53 and human tumor specimens with positive TP53 nuclear staining exhibited reduced TβRII and knocking down mtTP53 induced TβRII, increasing TGFβ downstream gene expression while inhibiting proinflammatory NF-κB target gene expression. Transfection of ectopic TβRII directly restored TGFβ signaling while inhibiting inhibitor κBα degradation and suppressing serine-536 phosphorylation of NF-κB p65 and NF-κB transcriptional activation, linking these alterations. Finally, experiments with TβRII conditional knockout mice show that abrogation of TGFβ signaling promotes the sustained induction of NF-κB and its proinflammatory target genes during HNSCC tumorigenesis and progression. Together, these findings elucidate a regulatory framework in which attenuated TGFβ signaling promotes NF-κB activation and squamous epithelial malignancy in the setting of altered TP53 status. [Cancer Res 2009;69(8):3415–24]
The development and progression of cancer are the result of sequential genetic and biological events, including loss of tumor suppressor genes and gain of function in proto-oncogenes. These multiple defects lead to self-sufficient proliferation with limitless replicative potential, evasion of apoptosis, tissue invasion and metastasis, and sustained angiogenesis (1). Mounting evidence indicates that nuclear factor-κB (NF-κB) is a critical mediator in many of these processes through transcriptional regulation of several hundred downstream genes, the products of which promote the malignant phenotype (2).
In regulating this broad genetic program, NF-κB transcription factors are assembled through dimerization of five subunits: RelA(p65), p50(NF-κB1), p52(NF-κB2), c-Rel, and RelB. Each dimer is bound in the cytoplasm by an inhibitor κB (IκB), which prevents its nuclear translocation (3). Cell stimulation transiently activates the IκB kinase (IKK) complex, which is composed of two catalytic subunits (IKKα and IKKβ) and a regulatory subunit (IKKγ). Activated IKK phosphorylates IκB proteins, resulting in their proteosomal degradation and subsequent translocation of NF-κB dimers to the nucleus. NF-κB is constitutively activated in diverse human malignancies, including head and neck squamous cell carcinoma (HNSCC), wherein numerous activating signals have been elucidated, including proinflammatory cytokines tumor necrosis factor α (TNFα) and interleukin-1 (IL-1); growth factors epidermal growth factor (EGF) and hepatocyte growth factor; and intermediate kinases phosphoinositide 3-kinase (PI3K), casein kinase 2, and IKK (4). Whereas previous studies have focused on the positive signals contributing to NF-κB activation in HNSCC, signals that negatively regulate NF-κB activity remain poorly understood (5).
The transforming growth factor β (TGFβ) signaling pathway performs an essential regulatory role in maintaining normal epithelial homeostasis (6). TGFβ signals through three TGFβ receptor (TβR) subunits (TβRI, TβRII, and TβRIII), resulting in phosphorylation of Smad2 and Smad3, which, with Smad4, enter the nucleus and regulate transcription. TGFβ potently inhibits epithelial proliferation by up-regulation of cyclin-dependent kinase inhibitor genes p15INK4b, p21Cip1, and p57Kip2 and down-regulation of c-MYC and ID1 expression (6). Thus, TGFβ signaling shows potent tumor suppression, and transcriptional inactivation or mutations of TβRI and TβRII have been reported in human epithelial malignancies (7). Whereas mutation of TβRII is uncommon in HNSCC development, decreased TβRII expression occurs frequently and leads to a less differentiated, more aggressive phenotype (8, 9). However, the relationship between alterations in TGFβ signaling and NF-κB activation, and the mechanisms contributing to reduced TβRII expression in HNSCC have not been elucidated.
The TP53 tumor suppressor represents another commonly altered target underlying the development of cancer, including HNSCC. Defects in TP53 function contribute to loss of cell cycle regulation, genomic instability, and therapeutic resistance, promoting the survival of malignant cells. Mutation of TP53 in HNSCC occurs with a frequency of ∼50%, resulting in altered TP53 expression and function (10). In HNSCCs that retain a wild-type (wt) TP53 genotype, approximately one-half exhibit deficient TP53 expression by immunohistochemistry (11, 12). Finally, TP53 mutations are not restricted to neoplastic cells but frequently affect adjacent, normal-appearing keratinocytes, implicating TP53 mutation in the initiation of HNSCC (13).
We recently identified distinct molecular signatures in HNSCCs differing in TP53 status, wherein genome-wide microarray and bioinformatic analysis revealed activation of NF-κB target gene expression with concomitant down-regulation of TGFβ pathway genes (14, 15). These observations prompted us to hypothesize that a potential relationship exists among TGFβ signaling, NF-κB activation, and TP53 status, which we define here using HNSCC-derived cell lines, human tissue specimens, and an animal model of HNSCC generated by TβRII deletion from murine head and neck epithelia (16).
Materials and Methods
Cell lines and mice. HNSCC cell lines from University of Michigan squamous cell carcinoma (UM-SCC) series were obtained from Dr. T.E. Carey (University of Michigan). UM-SCC cell lines were previously characterized and found to possess molecular and phenotypic alterations expressed in situ and important in the pathogenicity of HNSCC (15). The tumor and outcome characteristics of patients providing UM-SCC cell lines are shown in Supplementary Table S1. The TP53 mutation status of these cell lines was analyzed by bidirectional DNA sequencing of exons 2 to 9 (Supplementary Table S2). UM-SCC lines and primary human epidermal keratinocytes (HeKa) were cultured as previously described (12). All animal experiments were performed under protocols approved by the Institutional Animal Care and Use Committee at the Oregon Health and Science University. TβRII-floxed mice and K5.CrePR1 transgenic mice were established and maintained as described (16).
Reagents and transfection. Recombinant TNFα and TGFβ1 were from R&D Systems. Primary antibodies against the following proteins were used: TβRII (Santa Cruz), p-Smad2 (Cell Signaling), Smad2/3 (BD Biosciences), TP53 (Santa Cruz), serine-536 and serine-276 phosphorylated p65 (Cell Signaling), p65 (Santa Cruz), p50 (Santa Cruz), Oct-1 (Cell Signaling), and β-actin (Cell Signaling). All transfections were performed using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen). Transient transfection was performed using constructs for TβRII and pcDNA3 control vector. The expression vector containing the human TP53 open reading frame (pORF-hTP53) and the empty control vector (pORF-mcs) were obtained from InvivoGen. Control small interfering RNA (siRNA) was from Qiagen and TP53 siRNA was from Dharmacon.
Real-time quantitative reverse transcription-PCR. RNA isolation and cDNA synthesis were performed as previously described (12, 16). cDNA products were subjected to real-time quantitative reverse transcription-PCR (QRT-PCR) using TaqMan Assays-on-Demand probes (Applied Biosystems). An 18S rRNA probe was used as an internal control. Each sample was examined in triplicate. The relative mRNA expression levels were determined by normalizing with the 18S transcripts, the values of which were calculated using the comparative Ct method. Further details are available in the Supplementary Materials and Methods.
Immunohistochemistry and immunofluorescence. Formalin-fixed and paraffin-embedded HNSCC tissue arrays were obtained from Cybrdi as previously described (17). Each array contained HNSCC tumor tissues from 20 individuals spotted in triplicate plus normal mucosa tissues from six normal subjects spotted in duplicate. Immunohistochemistry and immunofluorescence analysis are described in Supplementary Materials and Methods.
Aberrant expression of TGFβ and activated NF-κB genes in HNSCC cells with distinct TP53 status. Previous microarray profiling and bioinformatic analyses of gene expression in a panel of nine HNSCC cell lines (Supplementary Table S1; ref. 18) and primary HeKa cells predicted potential cross talk among NF-κB, TGFβ, and TP53 signaling pathways (14). We performed real-time QRT-PCR to assess the expression of critical components of these pathways and downstream effector genes generating a cluster-based heatmap normalized to the expression pattern in primary keratinocytes (HeKa; Fig. 1A).
Increased basal expression of an NF-κB activating kinase, subunit (IKKA and C-REL) and target proto-oncogenes and proinflammatory genes (CCND1, cIAP1, IL-6, and IL-8) were detected (Fig. 1A). Conversely, attenuated TGFβ signaling was indicated by down-regulation of at least one of three essential TGFβ receptor subunits and other downstream targets (Fig. 1A and B). Strikingly, reduction in receptor expression clustered closely with the TP53 genotype and expression signature (Supplementary Table S2; Fig. 1A, bottom row). HNSCC cells deficient in wtTP53 transcript and protein (UM-SCC-1, UM-SCC-6, UM-SCC-9, and UM-SCC-11A) underexpressed TβRI and TβRIII, whereas most lines expressing mutant (mt) TP53 (UM-SCC-11B, UM-SCC-22A, UM-SCC-22B, UM-SCC-38, and UM-SCC-46) showed reduced TβRII levels. Expression of downstream signaling components SMAD2, SMAD3, and SMAD4 varied but was not markedly reduced. Additionally, TGFβ target genes p15INK4b, p21Cip1, SMAD7, HPGD, PAI1, and MMP2 were strongly down-regulated in most UM-SCC lines. By contrast, ID1, previously implicated in metastatic aggressiveness of human SCC, was up-regulated in all UM-SCC lines (19). KRAS or HRAS overexpression was also detected in this panel of human HNSCC cell lines, as previously observed in ∼80% of human HNSCC tissue samples (16). Together, these data suggest that HNSCC exhibit aberrant NF-κB gene signatures in the context of multiple deficiencies in the TGFβ signaling network, which are linked to deficient wtTP53 or mtTP53 status.
HNSCC cells exhibit deficient TGFβ-induced growth inhibition and transcriptional activity. To determine if reduced expression of TGFβ receptor subunits affects TGFβ signal-induced growth arrest in HNSCC cells, we measured cell proliferation by a 5-day 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay using five UM-SCC lines of differing TP53 status (UM-SCC-6, UM-SCC-11B, UM-SCC-22A, UM-SCC-22B, and UM-SCC-46) and HeKa control cells under varying concentrations of rTGFβ1. TGFβ1 significantly inhibited HeKa but not HNSCC cell proliferation (Fig. 2A and Supplementary Fig. S1A for UM-SCC-11B and UM-SCC-22A). UM-SCC-11B, UM-SCC-22A, and UM-SCC-46 exhibited complete resistance to TGFβ-induced cytostatic effects. UM-SCC-6 and UM-SCC-22B showed a partial reduction in cell proliferation, suggesting attenuated TGFβ signaling and cytostatic activity.
We next examined whether a reduction in cellular response to TGFβ1 affects expression of TGFβ downstream target genes important in growth arrest and other biological functions. In HeKa, TGFβ1 exposure strongly induced mRNA expression of PAI1, MMP2, p15INK4b, p57Kip2, and SMAD7 and suppressed expression of c-MYC, an NF-κB target gene (ref. 20; Fig. 2B and Supplementary Fig. S1B for UM-SCC-11B and UM-SCC-22A; data not shown for p57Kip2). By contrast, all HNSCC lines exhibited diminished sensitivity to TGFβ-modulated gene expression. Collectively, these findings show that HNSCC cells with deficient expression of component(s) of the TGFβ signaling pathway exhibit functional attenuation of TGFβ signaling through loss of TGFβ-mediated growth arrest and transcriptional activity.
Induction of residual TGFβ signaling inhibits NF-κB activation. As our previous bioinformatic analysis (14) and results in Fig. 1 suggested potential cross talk and an inverse relationship between the TGFβ and NF-κB gene expression, we examined if TGFβ could inhibit NF-κB activation. We selected the two HNSCC lines (UM-SCC-6 and UM-SCC-22B) that exhibited residual TGFβ-induced growth inhibition and UM-SCC-46 as a negative control line because of its deficient TGFβ signaling (Fig. 2). In Fig. 3A, UM-SCC-6 and UM-SCC-22B exhibited significant detectable TGFβ1-induced TGFβ luciferase reporter gene at 12 hours, whereas UM-SCC-46 was not responsive to TGFβ1 treatment. UM-SCC-6, UM-SCC-22B, and UM-SCC-46 were transfected with a NF-κB reporter plasmid and treated ±TGFβ1 and ±TNFα. TGFβ1 inhibited constitutive and/or TNFα-induced NF-κB transcriptional activity in UM-SCC-6 and 22B, whereas UM-SCC-46 was unresponsive to TGFβ1 treatment (Fig. 3B). Taken together, these findings show that TGFβ is able to suppress NF-κB activation in some UM-SCC cells.
Linkage between diminished TβRII expression and signaling, mtTP53 accumulation, and NF-κB target gene activation in a subset of HNSCC. Given that loss of TβRII gene expression and signaling was observed in a subset of HNSCC cell lines expressing mtTP53 (Fig. 1; Supplementary Table S1), we examined if a similar subset is found in HNSCC tumors in situ. Immunostaining for TβRII, phosphorylated (activated) Smad2, and TP53 protein expression was performed using a tissue array with 20 HNSCC specimens and six normal oral mucosa specimens (Fig. 4A). Sixteen of the 20 HNSCC samples (80%) displayed absent (−) or low TβRII protein, with 11 of these 16 showing absent (−) or low phospho-Smad2. Of nine scored completely negative for TβRII, eight showed decreased p-Smad2 staining (Supplementary Fig. S2A). Conversely, all normal oral mucosa samples exhibited both strong TβRII and activated p-Smad2 staining. Together, these observations confirm that deficient TβRII expression and p-Smad2 signaling is prevalent in human HNSCC (16).
As TP53 mutations often result in proteins highly resistant to degradation, detection of TP53 nuclear accumulation by immunohistochemistry is strongly associated with mutations in the TP53 gene (21). Consistent with an expected mutation rate of ∼50% in HNSCC (10), 11 of the 20 tumors (55%) showed nuclear staining for TP53. Nine of these 11 (82%) exhibited concomitantly reduced or absent TβRII staining (Supplementary Fig. S2A). Of the six HNSCC samples that were most strongly positive for TP53 nuclear staining, five (83%) showed a corresponding loss of TβRII staining within the tumor tissue. None of the normal mucosa specimens exhibited significant TP53 staining. Thus, decreased TβRII expression is seen in a subset of HNSCCs exhibiting increased accumulation of TP53 protein, as well as in a subset without TP53 staining.
We next investigated whether mtTP53 may have a suppressive role in TβRII expression by comparing TP53 and TβRII expression in HeKa with wtTP53, two cell lines with a defect in wtTP53 expression (UM-SCC-9 and UM-SCC-11A), and four cell lines expressing missense mtTP53s (UM-SCC-11B, UM-SCC-22A/B, and UM-SCC-46; Fig. 4B; Supplementary Table S2). Whereas similar levels of TβRII mRNA and protein expression were detected in either HeKa expressing or UM-SCC deficient for wtTP53 protein (Figs. 1B and 4B), diminished TβRII mRNA and protein levels were seen in most UM-SCC lines expressing mtTP53, consistent with the pattern observed in a subset of HNSCC tumors above (Figs. 1B and 4A and B). Transfection of wtTP53 into UM-SCC-11A cells deficient in endogenous wtTP53 promoted TβRII and target gene expression (Supplementary Fig. S3), suggesting that expression of wtTP53 has an enhancing effect on TβRII expression whereas endogenous mtTP53 in the other subset may have a dominant-negative effect.
To further establish if the mtTP53 in UM-SCC lines contributes to deficient TβRII expression, we tested the effect of TP53 siRNA to inhibit TP53 mRNA expression in UM-SCC lines with mtTP53. TP53 siRNA efficiently inhibited endogenous TP53 expression in UM-SCC-11B and UM-SCC-22A over 96 hours (Fig. 4C; Supplementary Fig. S2B). Knockdown of mtTP53 resulted in a progressive increase in TβRII gene expression relative to control siRNA (Fig. 4C and Supplementary Fig. S2B). Similar results were seen in UM-SCC-22B and UM-SCC-46 (Fig. 4D; data not shown). Concomitant with enhanced TβRII expression, restoration of TGFβ signaling was evident by increased transcription of TGFβ downstream genes, SMAD7 and MMP2, at 72 hours (Fig. 4C). We further compared effects of TP53 siRNA on the expression of TP53, TβRII, and NF-κB REL(p65)-dependent proinflammatory genes IL-6 and IL-8 (14). Figure 4D shows that, after TP53 siRNA knockdown and increase in TβRII, there is a significant reduction in both IL-8 and IL-6 expression in the mtTP53 line UM-SCC-22B. Collectively, the results above suggest that mtTP53 may be a mechanism contributing to attenuated TβRII gene expression, TGFβ signaling, and enhanced NF-κB inflammatory gene expression in HNSCCs.
Ectopic TβRII expression restores TGFβ signaling and suppresses NF-κB activation in HNSCC cells. To directly examine the role of TβRII expression and signaling in inhibition of NF-κB, we determined if transient transfection of a plasmid expressing exogenous TβRII protein could restore TGFβ activity and suppress NF-κB signaling in HNSCC lines that showed deficient TβRII. We selected UM-SCC-46 because it had previously exhibited marked attenuation of TGFβ signaling (Fig. 2). Upon transfection, ectopic TβRII was successfully transiently expressed between 48 and 96 hours (Fig. 5A). Expression of TβRII dramatically induced TGFβ signaling in UM-SCC-46, illustrated by enhanced levels of activated phosphorylated Smad2 at 48 hours (Fig. 5B). This implicated TβRII expression as a critical defect in TGFβ signaling in these HNSCC cells.
We further examined the effect of restored TβRII signaling on established mechanisms required for NF-κB target gene transactivation. IKK-dependent signal phosphorylation of p65(RelA) serine-536 and PKA-inducible phosphorylation of serine-276 have been shown to enhance NF-κB gene transactivation after stimulation with TNFα (22). Increased expression of TβRII strongly suppressed TNFα-induced p65(RELA) serine-536 phosphorylation, but not serine-276 phosphorylation, or expression of total nuclear p65 or Oct-1, an unrelated control (Fig. 5B and Supplementary Fig. S4A). Together, these data are consistent with the hypothesis that TβRII in these HNSCC inhibits NF-κB transactivation by a signal-dependent mechanism at or above IKK. To further examine if TβRII modulates IKK-dependent IκBα degradation involved in upstream signal activation and nuclear translocation of NF-κB (2, 23), we used a plasmid expressing an IκB-luciferase fusion protein (24), which can serve as a nontranscriptional reporter of kinase activation and degradation. As expected, TNFα-enhanced degradation of the IκBα-luciferase fusion protein (Fig. 5C). Upon transfection of ectopic TβRII, restoration of TGFβ signaling resulted in a significant attenuation of TNFα-induced IκB-luciferase protein degradation (∼25%; P < 0.05), suggesting that a mechanism upstream of IKK-mediated IκB degradation could partially contribute to the attenuating effect of TGFβ on NF-κB signaling.
Once we showed that exogenous TβRII expression could inhibit NF-κB phosphorylation and IκB degradation, we sought to confirm if transient TβRII expression and signaling suppressed aberrant basal and TNFα-enhanced NF-κB transcriptional activity, using three UM-SCC lines expressing mtTP53. UM-SCC-46 exhibited the strongest restoration of TGFβ-induced luciferase activity at 48 hours upon ectopic expression of TβRII, followed by UM-SCC-22B and UM-SCC-22A (Fig. 5D and Supplementary Fig. S4B). Restoration of TβRII signaling inhibited both constitutive and TNFα-induced NF-κB transcriptional activity in UM-SCC-46 (Fig. 5D). UM-SCC-22A and UM-SCC-22B also exhibited a significant reduction in NF-κB transactivation, but this was most clearly shown after exposure to TNFα (Supplementary Fig. S4B). These findings show that restoration of TβRII signaling not only reduces activated RELA(p65) serine-536 phosphorylation but can also suppress NF-κB reporter gene transactivation in HNSCC lines expressing different mtTP53s.
Loss of TGFβ signaling promotes NF-κB activation in murine HNSCCs with conditional deletion of TβRII. To further examine the role of attenuated TGFβ signaling in the aberrant activation of NF-κB, we investigated mucosa and tumors from a murine model in which K5-targeted genetic deletion of TβRII from mucosal stratified epithelia, in combination with 7,12-dimethylbenz (a)anthracene (DMBA) initiation caused the development of primary and metastatic HNSCCs histologically and molecularly similar to human HNSCC (16). Tissue samples were collected from buccal mucosa of five TβRII+/+, TβRII−/−, and HNSCCs from five TβRII−/− mice and compared. Representative H&E–stained tumor sections showed atypical hyperproliferative epithelia by 3 months and HNSCC by 6 months in TβRII−/− mice when compared with their wt littermates (ref. 16; Supplementary Fig. S5). Immunostaining of TβRII−/− mucosa and HNSCCs confirmed the loss of TβRII protein and the absence of phosphorylated Smad2, showing abrogation of TGFβ signaling compared with wt controls (Supplementary Fig. S5). As classical NF-κB signal activation involves the nuclear translocation of p65 and p50, we examined the aforementioned tissues for the presence and localization of these subunits using immunofluorescence microscopy (Fig. 6A). As in human HNSCC lines above, concomitant with loss of TGFβ signaling, a corresponding increase was seen in detectable nuclear RELA(p65) phosphorylated serine-536, as well as p50 staining in five of five TβRII−/− buccal mucosa and five of five TβRII−/− basilar or HNSCC specimens. In contrast, normal TβRII+/+ mucosa revealed basilar or no NF-κB activation in the five specimens studied. Interestingly, regions in HNSCCs with the strongest NF-κB activation exhibited reduced K14 expression, consistent with previous studies implicating aberrant NF-κB activity and down-regulation of K14 in the epithelial-to-mesenchymal transition of high-risk HNSCCs (25).
Given the critical role of NF-κB in inflammation and our findings in Fig. 6A, we investigated the expression of NF-κB–regulated proinflammatory and angiogenic genes in these tissues by QRT-PCR (Fig. 6B). TβRII−/− HNSCCs exhibited a significant induction of NF-κB–inducible genes IL-1α, Vegf, Cxcl1/Gro1, and Cxcl5/Ena-78 when compared with both TβRII−/− buccal mucosa and normal TβRII+/+ mucosa (P < 0.05). Increased expression levels of Cxcl1 and Cxcl5 was also detected in TβRII−/− preneoplastic lesions compared with TβRII+/+ buccal tissues (Fig. 6B). Potentially further reflective of an inflammatory state, we observed previously (16) and here in the TβRII−/− underlying stroma the presence of numerous infiltrating cells, which are strongly positive for nuclear NF-κB (p50) immunofluorescence (Fig. 6A). Together, these data strongly support a model where abrogation of TGFβ signaling in stratified epithelia promotes the aberrant activation of NF-κB, target cytokines, enhanced inflammation, and development and malignant progression of HNSCC (Fig. 6C).
Herein, we show that altered expression of TGFβ receptor subunits and deficient TGFβ signaling are common in human HNSCC lines and tissues, which exhibit enhanced activation of NF-κB. We show that restoration of TGFβ signaling inhibits such aberrant NF-κB signaling and transcriptional activity, indicating that deficient TGFβ signaling is a critical event leading to the activation of NF-κB in HNSCC. Our data further suggest that TP53 mutation may be one mechanism that contributes to attenuated TGFβ signaling in a subset of HNSCCs through its ability to suppress TβRII expression. Finally, TβRII−/− conditional knockout mice that develop HNSCC with enhanced inflammation exhibited increased NF-κB activation and proinflammatory gene expression with tumor development, marking a significant connection between deficient TGFβ signaling and NF-κB activation in tumorigenesis. Whereas we have previously shown that sustained induction of NF-κB plays a critical role in the pathogenesis of HNSCC, we now implicate loss of a negative regulatory signaling pathway, TGFβ, as an integral step in this activation.
We previously reported that NF-κB–regulated gene signatures are differentially up-regulated in HNSCCs with distinct TP53 status and that overexpression of either K-RAS or H-RAS at the transcriptional level frequently occurs in human HNSCC tissues with concomitant reduction of TβRII mRNA levels (14, 16, 17). Here, we found reduced TβRII gene expression with concurrent RAS overexpression in a panel of human HNSCC cell lines and report additional TGFβ receptor and biological alterations, including those involving the NF-κB and TP53 pathways (Fig. 1A and B). Notably, expression of certain genes clustered according to TP53 status, including the TGFβ receptor subunits, SMAD3, SMAD7, TGFB1, TGFB2, HPGD, p21Cip1, and MMP2, as well as the NF-κB–related genes C-REL, IL-8, and cIAP1, suggesting that differential regulatory mechanisms that exist are directly or indirectly affected by TP53 genotype. Furthermore, the aforementioned genes serve as critical effectors commonly involved in the progression of aggressive HNSCCs (4, 16, 26, 27).
Although it has been suggested that inactivation of TβRI and TβRII expression occurs more frequently than genetic disruption in human HNSCC tissues (16, 28), the mechanisms by which this inhibition takes place in HNSCC have not been well defined. Our findings provide evidence that overexpression of mtTP53 can contribute to decreased expression of TβRII and signaling in a subset of HNSCC. Other mechanisms independent of mtTP53 accumulation may contribute to the deficient expression of TβRII in other HNSCC, such as those detected in tumor tissue arrays that lack TP53 staining. Here, we also detected reduced TβRI or TβRIII expression in a subset of HNSCC tumor lines deficient in wtTP53, representing other potential defects that merit further investigation. Down-regulation of TβRIII transcript levels has been recently reported in breast and prostate cancers corresponding to more aggressive phenotypes and metastatic dissemination (29).
A connection between TP53 and TGFβ signaling was suggested in thyroid epithelial cells, wherein inactivation of wtTP53 induced a loss of responsiveness to TGFβ1 treatment, accompanied by a partial reduction in TGFβ receptor expression (30). Subsequent studies revealed that ectopic overexpression of mtTP53 proteins induced resistance of squamous carcinoma cells to TGFβ-mediated growth inhibition (31, 32). Here, we provide further evidence for these relationships in distinct subsets of the same tumor type. Although TβRII expression is not solely dependent on nor significantly attenuated in the UM-SCC cell lines deficient in endogenous wtTP53, transfection of wtTP53 into these cells further enhanced TβRII expression and TGFβ downstream signaling (Supplementary Fig. S3) whereas, in UM-SCC lines with mtTP53, RNA interference–mediated TP53 inhibition enhanced TβRII and downstream SMAD7 gene expression, supporting a dominant-negative role of mtTP53 in TβRII signaling (Fig. 4C).
Mutant TP53 knockdown and enhanced TβRII expression was also followed by attenuation in expression of prototypical NF-κB target genes IL-6 and IL-8 (Fig. 4D). In a murine squamous cell carcinoma model, we previously observed increased expression of mtTP53 protein accompanied by a corresponding reduction in TβRII expression, TGFβ1 responsiveness, and increased activation of NF-κB and target genes as tumors progressed to a metastatic phenotype (33, 34). Furthermore, in a TP53-null cell system, a recent study showed that ectopic mtTP53 can interact with the promoter of TβRII, contributing to diminished TβRII gene expression (35). Together, these observations suggest that mtTP53 represents one mechanism that can reduce the expression of TβRII and TGFβ signaling in a subset of SCCs and other cancers. However, as there are likely genetic and epigenetic mechanisms contributing to reduced or absent TP53 and TβRII expression in HNSCCs, it is not surprising that certain HNSCC specimens summarized in Supplementary Fig. S2, which exhibit reduced TβRII expression, do not also exhibit nuclear TP53 staining.
Although this is the first demonstration of a relationship between TGFβ and NF-κB signaling in HNSCC, these findings are supported by other studies showing cross talk. In human intestinal mononuclear cells, pretreatment with TGFβ1 diminished TNFα-induced activation of p65 (36), whereas in a transgenic mouse model of renal inflammation, TGFβ target gene Smad7 promoted induction of IκBα, inhibiting an NF-κB–driven inflammatory response (37). Here, we show that restored TGFβ signaling partially reduces degradation of IκBα and strongly inhibits IKKβ-dependent serine-536 phosphorylation of p65, localizing such cross talk at or above the level of IKK. In HNSCC, IL-1–IL-1R, TNF-TNFR1, TGFα-EGFR, and PI3K-CK2-IKK signaling pathways have been shown to promote IKK and NF-κB activation (38–40), and these pathways have been reported to include components that may be modulated by TGFβ signaling, such as TGFβ activating kinase (TAK1), TAK1 binding kinase, CK2, and ID1-mediated PI3K-Akt activation (23, 41, 42). Studies are underway to determine if these or other mechanisms enhance cytokine and growth factor activation of NF-κB and inflammation in HNSCC.
The potential relationship between reduced expression of TβRII and NF-κB activation was of particular interest because of our previous demonstration that conditional targeting of TβRII in murine aerodigestive epithelia results in HNSCCs exhibiting dramatic inflammation (16). Here, we show that both TβRII−/− preneoplastic mucosa and HNSCCs show sustained induction of nuclear activation of classic NF-κB1/RelA (p50/p65) and its proinflammatory target genes (Fig. 6A and B). Most of these factors are significant mediators of the human NF-κB–regulated genetic program, critical in inflammation, angiogenesis, tumorigenesis, and progression of human cancers (2, 26, 43–46). Together, our findings suggest a novel regulatory framework whereby abrogation of TGFβ signaling in epithelial cells enhances NF-κB activation, tumorigenesis, and a deleterious inflammatory signaling circuit through recruitment of infiltrating immune and angiogenic cells. Furthermore, we show that mtTP53 may serve as an upstream repressor of TβRII expression, contributing to TGFβ inhibition (Fig. 6C). It is notable that alterations in TP53, abrogated TGFβ signaling, and activation of NF-κB are facets of many cancers, raising the possibility that the mechanisms we have uncovered may extend beyond HNSCCs. Indeed, agents such as heat shock protein 90 inhibitors, which simultaneously inhibit both NF-κB and mtTP53, are under investigation and show promise in numerous preclinical models (47). Thus, NF-κB and mtTP53 may represent concomitant pharmacologic targets in tumors wherein compromised TGFβ signaling leads to enhanced NF-κB activation and resultant malignant progression.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Z. Chen and C. Van Waes contributed equally to this work.
Grant support: NIDCD intramural projects Z01-DC-00073-01 and 74-01 (C. Van Waes) and grants DE15953 and CA79998 (X-J. Wang).
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 Lalage Wakefield, Stuart Yuspa, and David Gius for their critical review and comments on the manuscript; Ning Yeh, Liesl Nottingham, and Jay Friedman for technical assistance; Anjali Shukla and Paul Albert for their valuable discussions and advice; and Melissa Stauffer for editorial assistance.