Dose Dependent Activation of Retinoic Acid-Inducible Gene-I Promotes Both Proliferation and Apoptosis Signals in Human Head and Neck Squamous Cell Carcinoma

The retinoic-acid-inducible gene (RIG)-like receptor (RLR) family proteins are major pathogen reorganization receptors (PRR) responsible for detection of viral RNA, which initiates antiviral response. Here, we evaluated the functional role of one RLR family member, RIG-I, in human head and neck squamous cell carcinoma (HNSCC). RIG-I is abundantly expressed both in poorly-differentiated primary cancer and lymph node metastasis, but not in normal adjacent tissues. Activation of RIG-I by transfection with low dose of 5′-triphosphate RNA (3p-RNA) induces low levels of interferon and proinflammatory cytokines and promotes NF-κB- and Akt-dependent cell proliferation, migration and invasion. In contrast, activation of RIG-I by a high dose of 3p-RNA induces robust mitochondria-derived apoptosis accompanied by decreased activation of Akt, which is independent of the interferon and TNFα receptor, but can be rescued by over-expression of constitutively active Akt. Furthermore, co-immunoprecipitation experiments indicate that the CARD domain of RIG-I is essential for inducing apoptosis by interacting with caspase-9. Together, our results reveal a dual role of RIG-I in HNSCC through regulating activation of Akt, in which RIG-I activation by low-dose viral dsRNA increases host cell surviral, whereas higher level of RIG-I activation leads to apopotosis. These findings highlight the therapeutic potential of dsRNA mediated RIG-I activation in the treatment of HNSCC.


Introduction
The cellular innate immune response is the first line of host defense against viruses and other pathogens. As host cells, cancer cells and virus-infected cells share certain properties, such as the expression of specific antigens and the need to evade immune and non-immune control mechanisms in order to persist [1,2]. To suppress viral replication and spread, host cells often undergo premature cell death by triggering apoptosis. Apoptosis is therefore considered a potent antiviral defense mechanism by which infected cells are eliminated from the host. Tumor cells could be more susceptible to this kind of death signal than nonmalignant cells, many alterations required for tumor formation can also result in increased vulnerability to certain apoptotic stimuli [3]. Therefore, triggering anti-viral responses may be applied as an effective tumor therapy approach.
Double stranded RNA (dsRNA), generated during infection with both RNA and DNA viruses, is a strong inducer of host antiviral responses. Mammals have several families of pattern recognition receptors (PRRs), e.g., Toll-like receptors (TLRs), Retinoic acid-inducible gene (RIG) like receptors (RLRs), and Nod-like receptors (NLRs), to recognize viral dsRNA [4,5]. The innate immune responses to virus infection are often initiated by Toll-like receptors; alternatively, cytoplasmic dsRNA-recognizing RNA helicases RIG-I and melanoma differentiation-associated antigen 5 (MDA-5) can initiate antiviral signaling [5]. RIG-I is localized in the cytosol and recognizes 59-triphosphate RNA (3p-RNA) generated by viral RNA polymerases in the cytosol of cells [6,7]. Polyinosinic-polycytidylic acid [poly(I:C)], a synthetic and artificial mimic of long double-stranded RNA, is a strong activator of MDA-5. Upon recognition of RNA ligands, RIG-I and MDA-5 bind to the adapter protein interferon-b (IFN-b) promoter stimulator 1 (IPS-1) (also known as CARDIF, MAVS or VISA) located in the outer mitochondrial membrane [8,9]. The interactions of RIG-I or MDA-5 with IPS-1 initiate signaling pathways that elicit the activation of transcription factors including IFN regulatory factor 3 (IRF-3) and nuclear factor-kB (NF-kB), resulting in IFN production, activation of NF-kB target genes and the secondary induction of IFN-stimulated genes [5,10].
On the other hand, apoptosis has been known one of the important RLR activation-mediated antiviral responses in many cells. However, apoptotic mechanisms triggered by different virus in different cells appeared to be quite complex. For instance, it has been reported that the activation of RIG-I and MDA-5 using 3p-RNA and poly(I:C) leads to apoptosis of human melanoma cells, which proved to be independent of type I IFNs but dependent on upregulation of Puma and Noxa [11], while RIG-I-mediated activation of IRF-3 was demonstrated to be required for the apoptotic effect of adenovirus infection in the fibrosarcoma cells [12,13]. Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide, affecting 600,000 new patients each year. In the United States, 50,000 new cases are diagnosed, and nearly 10,000 deaths are attributable to this disease annually [14]. Despite advances in multimodality therapy, the overall 5-year survival rate is 40-50%, and has increased only incrementally in the past two decades [15,16]. Although RLR activation-mediated apoptosis is a potentially effective approach to tumor therapy, whether it also leads to apoptosis in HNSCC cells and/or the molecular mechanisms involved still remain largerly elusive. Developing new therapeutics targeting signaling molecules contributing to the evasion of apoptosis in HNSCC may effectively render those cancer cells susceptible to natural or induced programmed cell death.
To investigate whether RLRs also function in HNSCC, here we first examined the expression of RIG-I and MDA5 in both tumor biopsies and HNSCC cell lines. Then, we tested the impact of RLR activation by transfection with various doses of 3p-RNA or poly (I:C) on cell proliferation, migration and production of proinflammatory cytokines and type I interferon in above cell lines. Our results showed that activation of RLR with low dose of 3p-RNA promoted cell proliferation, whereas that with a high dose of 3p-RNA led to robust apoptosis in HNSCC cells. Furthermore, we demonstrated that the dual role of RIG-I in regulating the activation of Akt in proliferation and apoptosis.

Ethics Statement
This study was approved by the Ethics Committee of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine and carried out according to the recommendations of the Declaration of Helsinki. No informed consent (written or verbal) was obtained for use of retrospective tissue samples from the patients within this study, some of whom were deceased, since the Ethics Committee, who waived the need for consent, did not deem this necessary. All samples were anonymous.

Immunohistochemistry and mRNA Analysis of RIG-I Expression in HNSCC Tissues
Human HNSCC tissues for immunohistochemistry and mRNA expression were obtained from the tissue library of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine. Sixty-one samples of 34 patients with primary HNSCC who underwent surgery in our department from January 2008 to December 2011 were recruited in this study. The tissues prepared for immunohistochemistry were fixed with formalin, embedded in paraffin and immunostained with an anti-RIG-I antibody using the avidin-biotin peroxidase complex method. The intensity of staining seen were analyzed according to criteria described by Malik et al. [19]. The intensity is designated as negative when no tumor cells stain, weak staining when 10-20% of cells stain, moderate staining when 20-50% of cells stain, and strong staining when over 50% of cells stain. Tissues prepared for mRNA expression were immediately frozen in isopentane cooled by liquid nitrogen and stored until use at 280uC. Each tissue was pathologically confirmed (graded as 1 to 3 according to the WHO standard classification of HNSCC, 2005 version).

RNA Quantification
Quantitative real-time RT-PCR (Q-PCR) analysis was performed using a LightCycler (Roche, Basel, Switzerland) and SYBR RT-PCR kit (Takara, Dalian, China). Data were normalized to bactin expression.

Assays for Cell Proliferation, Migration and Invasion
For 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays, CAL27 cells or SCC4 cells (2,000 per well) were seeded in 96-well plates, cultured for 12 h, serum starved overnight and then transfected with 3p-RNA or BE-RNA for 16 h. Cell growth was assessed by the MTT assay (Promega, Madison, WI, USA), as described by the manufacturer. Each experiment consisted of three to four replicate wells for each treatment dose, and each experiment was performed at least three times. In vitro migration and invasion assays were performed using Matrigel coated and uncoated 24-well chambers/microfilters (5 mm pore polycarbonate filters), respectively, according to the manufacturer's instructions (BD Bioscience, Bedford, MA). Briey,

Western Blotting and Immunoprecipitation
After disrupting cells in lysis buffer, the cellular proteins were loaded and separated on SDS-PAGE and transferred to a nitrocellulose membrane (Amersham Biosciences, Piscataway, NJ, USA) using a standard electric transfer protocol. The membrane was blocked, probed with primary antibodies and then incubated with a horse radish peroxidase (HRP)-labeled secondary antibody (DAKO, Carpinteria, CA, USA). The protein signals were developed for visualization with the TMB Membrane Peroxidase Substrate (KPL, Gaithersburg, MD, USA).
For immunoprecipitation, CAL27 or SCC4 cell lysates were incubated with anti-RIG-I or anti-caspase-9 antibody cross-linked to protein A-Sepharose beads and prepared for analysis as previously described [11].

Statistical Analysis
Data are presented as means 6 standard deviation (S.D.) of more than three independent experiments. Statistical analysis was performed using the Chi-square test (only for biopsies) and Student's t-test. P values less than 0.05 were considered to be significant.

RIG-I Expression in HNSCC Biopsies and Cell Lines
To evaluate RIG-I expression in HNSCC, we first assayed the protein level in biopsy samples (Table 1). RIG-I was found to be substantially expressed in a high percentage of HNSCC tissues (82.4%, 28/34), whereas it was not detected in any of the adjacent normal tissues tested (0%, 0/11). Moreover, increased staining intensity of RIG-I was associated with the pathological grade of HNSCC. While none of the well-differentiated primary tumors (grade 1) and only 41% of moderately differentiated carcinoma cells (grade 2) exhibited strong staining intensities; however, about 80% of the poorly differentiated (grade 3) and 87% of lymph node metastatic tumors exhibited strong staining for RIG-I (*P,0.01, Chi-square test). Typical staining patterns are shown in Figure 1A. However, MDA5 was found to be expressed weakly in all primary tumor tissues (data not shown). To further confirm the relationship between expression of RIG-I and progress of HNSCC tumor progression, we detected the mRNA expression in HNSCC biopsy samples from Table 1 by Q-PCR. Our results indicated that the expression of RIG-I mRNA was also correlated with the progression of HNSCC (Fig. 1B). In addition, we also assayed the mRNA expression level in the HNSCC cell lines SCC4, CAL27 (Fig. 1C), SCC9 and SCC25 ( Figure S1), in which the basal expression level of RIG-I was much higher than that of MDA-5, and RIG-I was more potently induced than MDA-5 upon stimulation with the TLR4 ligand lipopolysaccharide (LPS), or transfection with 3p-RNA. Finally, we further confirmed the inducible expression of RIG-I at the protein level in CAL27 cells (Fig. 1D). Together, all these results revealed a positive correlation of RIG-I expression with development and progression of HNSCC.

RIG-I Activation by Low dose 3p-RNA Leads to Increased Cell Proliferation, Migration and Invasion
Given the predominant expression of RIG-I compared with MDA-5 in HNSCC, we then focused our study on the functional role of RIG-I in HNSCC. As shown in Figure 2A, we found that transfection of 3p-RNA into both CAL27 and SCC4 cells at doses between 10-500 ng/ml induced cell proliferation when compared with the BE-RNA control. However, the number of live cells transfected with high doses (1-10 mg/ml) of 3p-RNA were significantly decreased ( Fig. 2A). Low dose (100 ng/ml) of 3p-RNA transfection promoted CAL27 cell proliferation in time dependent manner (Fig. 2B). Furthermore, low dose of 3p-RNA transfection increased CAL27 cell number (Fig. 2C), which further indicates the pro-proliferative function of 3p-RNA. In addition, low doses of 3p-RNA also promoted cell migration and invasion of CAL27 (Fig. 2D) and SCC4 ( Figure S2) cells, and stimulated MMP-9 expression (Fig. 2E). Finally, siRNA knock-down of RIG-1 in CAL27 cells inhibited low-dose 3p-RNA-mediated cell proliferation (Fig. 2F), suggesting that low-dose 3p-RNA induced activation of RIG-I promotes proliferation, migration and invasion of HNSCC cells.

Activation of RIG-I by Low-dose 3p-RNA Induces Low Levels of Proinflammatory Cytokines and Type I Interferon
Given that RIG-I functions as a major virus RNA PRR, the effects of low-dose 3p-RNA on levels of type I interferon and proinflammatory cytokines in CAL27 cells were then assayed by Q-PCR. The mRNA expression levels of TNFa (Fig. 3A), IL-6 ( Fig. 3B) and interferon b (Fig. 3C) were all weakly induced [12 to 30-fold of the unstimulated basal level compared with over 200fold induction in macrophages (data not shown)]. We further assayed the function of IFN-b and TNFa on cell proliferation by knockdown of IFNRa and TNFR1 simultaneously. Knockdown of these two receptors did not affect the low-dose 3p-RNA-induced cell proliferation by MTT assay (data not shown), which may result from very low dose of IFN-b and TNFa induced by 3p-RNA

Activation of RIG-I by Low-dose 3p-RNA Induces Activation of NF-kB, MAPK and Akt
We next investigated the molecular signaling downstream of RIG-I activation that promotes cell proliferation of CAL27 cells. As shown in Figure 4A, transfection with a low dose of 3p-RNA strongly induced NF-kB and MAPK signaling pathways, while it had a modest effect on IRF3 activation. Consistent with this, low dose of 3p-RNA greatly induced nuclear translocation of p65, c-Jun/AP1 and IRF3 (Fig. 4B). In further support of pro-proliferation role of low-dose 3p-RNA, our results also showed that low dose of transfected 3p-RNA also strongly induced phosphorylation of Akt and p85 in the PI3K pathway (Fig. 4A).
To determine which above-mentioned signaling pathways are responsible for the induction of proliferation by low-dose 3p-RNA, we treated CAL27 cells with pharmacological inhibitors of NF-kB, MAPK and PI3K (upstream of Akt) (Fig. 4C). Individual inhibitors could only partially block the proliferation, while the combination of NF-kB and Akt inhibitors decreased the proliferation to a basal level. However, high dose of 3p-RNA transfection could barely activate NF-kB (Fig. 4D).The results indicate that the proliferation induced via RIG-I activation by transfection with low-dose 3p-RNA is dependent on both NF-kB and Akt activation.

Activation of RIG-I by High-dose 3p-RNA Induces Robust Apoptosis
As it was found above that high doses (1-10 mg/ml) of 3p-RNA reduced the number of viable cells ( Fig. 2A), we wanted to confirm whether apoptosis is involved. Transfection with high doses of 3p-RNA (1, 10 mg/ml) led to high levels of apoptosis in both CAL27 (Fig. 5A) and SCC4 ( Figure S3) cells. The cleavage of poly (ADPribose) polymerase (PARP), a prominent marker of apoptosis, was greatly induced by 3p-RNA at the concentration of 5 mg/ ml, when compared to at 1 mg/ml in both CAL27 and SCC4 cells (Fig. 5B), while low dose of 3p-RNA did not induce PARP cleavage (data not shown). The induction of apoptosis by high concentrations of transfected 3p-RNA was also confirmed by activation of caspase-3 (Fig. 5C). At the same time, we found that high-dose 3p-RNA (5 mg/ml) potently induced cleavage of PARP even 8 h after transfection (Fig. 5D), and that the mitochondrial  transmenbrane potential (DY m) was also damaged as early as 4 h after transfection, indicating that a high dose of 3p-RNA transfection led to mitochondria-derived apoptosis (Fig. 5E). The dependence on RIG-I of apoptosis induced by high-dose 3p-RNA was further confirmed by siRNA mediated inhibition of RIG-I (Fig. 5F). RIG-I knockdown nearly abrogated the 3p-RNA induced cleavage of PARP. Consistent with a previous report [11], knockdown of IFNRa did not affect the cleavage of PARP (data not shown). In addition, knockdown of IFNRa and TNFR1 simultaneously did not affect the high dose of 3p-RNA-induced PARP activation (data not shown), which may results from very low dose of IFN-b and TNFa induced by 3p-RNA transfection. These results indicate that the high dose of transfected 3p-RNA led to robust apoptosis in HNSCC cells, likely derived from the mitochondrial pathway.

Transfection of High-dose 3p-RNA Leads to Mitochondria-derived Apoptosis
To investigate the apoptotic mechanisms, induced by high-dose 3p-RNA, we next examined the activation of caspase-8, caspase-9 and caspase-3. Transfection of high-dose 3p-RNA induced strong activation of caspase-9, but weak activation of caspase-8 (Fig. 6A), indicating that the high dose of 3p-RNA led to mitochondriaderived apoptosis. The activation of caspase-9 could not be inhibited by the protein synthesis inhibitor cycloheximide (CHX) (Fig. 6B), suggesting that the 3p-RNA induced apoptosis is not dependent on new protein synthesis. By immunoprecipitating endogenous RIG-I, we further found that transfection with highdose 3p-RNA led to the interaction of RIG-I with both procaspase-9 and cleaved caspase-9, while a low dose of transfected 3p-RNA did not. As the antibody for total caspase-9 was not suitable for immunoprecipitation, we used a cleaved caspase-9 (Asp330) antibody in the reciprocal immunoprecipitation. Cleaved caspase-9 could also interact with RIG-I transfected with a high dose of 3p-RNA, but not in those transfected with a low dose of 3p-RNA in both CAL27 and SCC4 cells (Fig. 6C). We further overexpressed the myc-tagged full-length or CARD domaindepleted mutant of RIG-I in HEK293 cells. The CARD domaindepleted RIG-I mutant could not interact with caspase-9 (Fig. 6D). Overexpression of the CARD domain-depleted RIG-I mutant also blocked 3p-RNA induced PARP cleavage in CAL27 cells (Fig. 6E). Taken together, these results demonstrate that the high dose of transfected 3p-RNA induced apoptosis by promoting interaction of RIG-I with caspase-9.

Apoptosis Induced by High-dose 3p-RNA can be Rescued by Constitutively Active Akt
To understand why 3p-RNA concentration-dependent activation of RIG-I can lead to both proliferation and apoptosis, we examined the Akt activation. While a high dose of transfected 3p-RNA induced weaker Akt activation, it led to a higher expression of Akt phosphatase PTEN and PP2A than did the low dose of 3p-RNA (Fig. 7A). To further demonstrate the role of Akt activation, we overexpressed the HA-tagged constitutively active Akt (CA-Akt) before transfection with a high dose of 3p-RNA in CAL27 cells. As shown in Figure 7B, the cleavage of PARP induced by high-dose 3p-RNA was abrogated by CA-Akt. Consistently, the interaction of caspase-9 with RIG-I was also blocked by CA-Akt (Fig. 7C). These results demonstrate that the apoptosis induced by high-dose 3p-RNA is dependent on decreased activation of Akt.

Discussion
Activation of PRR leads to production of inammatory cytokines, including TNF-a, IL-6 and IFN-b, which are essential for both innate and adaptive immunity to clear pathogens [10,20]. Our results here show that as a PRR that functions in innate immunity against virus infections, RIG-I is also highly expressed on HNSCC cells and its activation contributes to tumor growth and progression in response to a low dose of 3p-RNA, whereas leading to apoptosis in response to a high dose of 3p-RNA.
RIG-I is localized in the cytosol and recognizes 3p-RNA generated by viral RNA polymerases in the cytosol of cells. 3p-RNA is also generated physiologically in the nucleus, but it lose its RIG-I ligand activity before it is released to the cytosol due to processing such as capping of the 59-triphosphate end or nucleoside modification of RNA [6,7]. Active 3p-RNA can be generated by in vitro transcription and used as synthetic ligands for RIG-I [11]. 3p-RNA shares certain structural features with siRNAs, and the requirements for cytosolic delivery are similar. Several siRNA molecules are in clinical studies, and short 3p-RNA may benefit from formulations developed for the clinical application of siRNAs [11].
The successful survival of a virus relies upon the evolution of strategies that modulate host cell signaling pathways, particularly those governing apoptosis and cell survival [21]. Infected host cells undergo apoptosis to eliminate invading viruses, but virus-encoded anti-apoptotic proteins can also counter this response to allow completion of the virus replication cycle before destruction occurs [22]. Activation of PI3K/Akt signaling is believed to contribute to the maintenance of the latent state by suppressing apoptosis and hence the elimination of virus-infected cells [23]. PI3K/Akt signaling promotes cell survival by multiple mechanisms, such as phosphorylating and inactivating caspase-9 or the BH3 domain containing family member BAD, or activation of PRL-3 [24]. Viral products, such as LMP1 of Epstein Barr Virus (EBV), can activate PI3K either through direct interaction with the catalytic or adaptor subunits or by facilitating the association of PI3K with receptor or non-receptor tyrosine kinases [25,26]. Our results show that activation of the PI3K/Akt and NF-kB pathways by transfection with the double-stranded RNA virus mimic 3p-RNA enhances cellular proliferation and functions to benefit virus survival, suggesting that virus dsRNA can also promote host cell survival. Apoptosis is a common mechanism used by the host to block viral spread. For example, virus-induced apoptosis of peripheral neuronal cells can be a protective host response that blocks transmission of herpes simplex virus 2 (HSV-2) to the central nervous system [27]. However, late in infection, apoptotic death of the infected cell may facilitate virus egress and spread of the infection [28]. Thus, the timing of apoptosis is a critical factor in determining whether it is pro-or anti-viral. Our results showing that transfection of low-dose 3p-RNA led to proliferation of infected cells may represent the early infection stage, during which the virus promotes host cell survival. Conversely, the induction of apoptosis by a high dose of transfected 3p-RNA may represent the late infection stage, during which the virus requires the cell lytic process to spread. Our finding that viral dsRNA also functions in the progression of infection gives new insight for future molecular research on virus infections.
Hanahan and Weinberg suggested essential alterations in cell physiology that collectively dictate malignant growth, and evasion of apoptosis or resisting cell death is an important hallmark of tumor cells [2]. Based on growing understanding of the molecular mechanisms underlying apoptosis resistance to tumors, many studies have attempted to induce tumor cell apoptosis by targeting anti-apoptotic proteins [29,30]. We found that low dose of 3p-RNA lead to cell proliferation by activating NF-kB signal pathway, which suggests that virus infection-activated RIG-I is an oncogenic factor at early stage of HNSCC initiation. This viewpoint adds new explanation of how HPV infection frequently leads to head and neck cancers [31]. This work demonstrated that RIG-I activation could lead to apoptosis by interaction with caspase-9 in CAL27 cells, in which Akt activation was shown to be down-regulated. Additionally, overexpression of constitutively active Akt could inhibit the interaction of RIG-I and caspase-9, indicating that activation of Akt mediates dsRNA-induced apoptosis. We now speculate that the structure of two CARD domains containing RIG-I favoring interacting with caspase-9 upon high-dose 3p-RNA transfection, which is different from that upon low dose of 3p-RNA or common dose of virus infection. Why transfection of high-dose 3p-RNA does not induce Akt activation with upregulation of Akt inhibitor phosphatase PTEN or PP2A will require further investigation [32]. 3p-RNA delivered by PEI in vivo has been shown to induce apoptosis of melanoma tumor cells and strongly activate NK cells at the same time [33]. Our results suggest that 3p-RNA delivered in vivo may also yield promising results in the HNSCC model.
In conclusion, we identified a signaling pathway which results in either proliferation or apoptosis, depending upon the dose of 3p-RNA, and described an additional physiological function for viral dsRNA in the HNSCC model. Our findings support that effective in vivo delivery of potent synthetic RNA ligands may be a promising approach for the treatment of HNSCC.