Combination of interferon-gamma and autophagy inhibitor as a therapeutic approach in oral squamous cell carcinoma

Zhi-hang Zhou Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine Tong-Chao Zhao Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine Si-yuan Liang Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine Dong-wang Zhu Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine Zhi-yuan Zhang Ninth People's Hospital, Shanghai Jiao Tong Unviersity School of Medicine Wu-tong Ju Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine Lai-ping Zhong (  zhonglp@hotmail.com ) Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine


Background
Oral squamous cell carcinoma (OSCC) is a common malignant tumor that occurs in the oral and maxillofacial region and has approximately 300,000 new cases worldwide each year [1,2]. Despite the progress achieved in radical surgical resection with proper reconstruction and postoperative radiotherapy/chemoradiotherapy, the 5-year survival rate has remained at 50% to 60% [3,4] and even lower in patients with locally advanced lesions. Consequently, there is an urgent need to explore effective OSCC therapeutic strategies. Macroautophagy (autophagy or 'self-eating') is a lysosome-mediated process whereby cells degrade organelles and macromolecules and recycle cellular waste [5]. Accumulating evidence suggests that autophagy can both suppress and promote tumor functions in cancer progression [6,7]. On the one hand, the initial stages of tumorigenesis can be inhibited by autophagy [8] and on the other, autophagy could nurture established cancers [9][10][11]. Yang et al. [12] revealed the mechanism of how autophagy promotes pancreatic tumor growth via p53 alternation. In research relating to OSCC, autophagy was considered as crucial for either nurturing cancer cells [13,14] or initiating programmed cell death [15,16]. Accordingly, the inhibition of autophagy may be a potential strategy for OSCC treatment.
Studies in interferon-gamma (IFNγ) have demonstrated an anti-tumor effect in several tumors, including those in colorectal [17], gastric [18], and cervical [19] cancers. Recent ndings suggest that members of the IFN family synergize with traditional anti-tumor treatment for OSCC through several mechanisms, including augmenting various immune functions [20]. However, IFNγ therapy alone does not achieve an ideal effect on established tumors [17][18][19], which may re ect an adaptive resistance. Studies of chemoresistance in brain, gastric, and ovarian cancers suggest that autophagy plays a key role [21][22][23].
Herein, autophagy co-activated by IFNγ is considered to be a cancer-promoting factor, which provides a potential strategy for combined inhibition therapy. In our present study, we rst demonstrate autophagy and apoptosis induced by IFNγ in OSCC cells. The ATG5 molecule was found to be necessary for IFNγinduced autophagy. Furthermore, IFNγ and chloroquine (CQ) exhibited clear synergistic anti-tumor effects in vitro and in vivo. This study supports the combined application of autophagy inhibitors and IFNγ in the clinical therapy of OSCC.

Materials And Methods
2.1. Culture system HN4, HN6, HB96, and CAL27 cell lines were used in our study. The HB96 cell line was established from our in vitro cellular carcinogenesis model of OSCC. HN4 and HN6 cell lines were provided as a gift from the National Institutes of Health (USA). The HN4 cell line was derived from tongue squamous cell carcinoma. Cell line HN6 was established from pharyngeal squamous cell carcinoma. The tongue squamous cell carcinoma cell line CAL27 was purchased from ATCC (Manassas, VA, USA). These cell lines were cultured in complete DMEM (Gibco, Carlsbad, CA, USA), supplemented with 10% FBS, 1% penicillin-streptomycin, and 1% glutamine. Cells were maintained in a humidi ed 5% CO 2 atmosphere at 37°C.
The plates were incubated in a humidi ed incubator at 37°C for 2 h. The absorbance was measured at 450 nm.

Colony formation assay
To evaluate the in uence of IFNγ on colony formation, HN4 and CAL27 cells were seeded in 24-well plates at approximately 1×10 3 cells/well and cultured in complete medium with 200 ng/ml IFNγ for 48 h.
Then the medium was replaced by complete DMEM for 6 d. Plates were nally washed with PBS twice, xed with 4% paraformaldehyde for 30 min, and stained with 0.5% crystal violet for 5 min. The counting of colonies was performed using Image J software.
2.6. Immuno uorescence analysis OSCC cells were xed and permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, St Louis, MO, USA) for 5 min. After blocking with 1% BSA-PBS for 1 h, 0.5 mg/ml of 4′,6-diamidino-2-phenylindole was used to stain the nuclei of cells. The BioTek Cytation 3 Cell Imaging Reader was used to visualize and acquire immuno uorescence images. The number of LC3B-I/II puncta was determined by Image J software.

Immunohistochemistry
To retrieve antigen, tumor slices were heated in a water bath at 100°C with EDTA buffer (pH=10.0) for 20 min. The primary antibodies were as shown below: LC3B II (CST, Danvers, MA, USA), P62, and Ki67 (Abcam, Cambridge, MA, UK). Immunohistochemistry and image analysis were performed to assess the mean optical density for Ki67, LC3B II, and P62 in vivo. Tumor volume was calculated using the formula length × width 2 /2. After three weeks, the mice were killed and the tumor tissue was resected. Tumor tissue and organ parts were immobilized and embedded in para n. Tissue sections (4 mm) were stained with hematoxylin and eosin. Terminal deoxynucleotide transferase dUTP notch end marker (TUNEL) was used to detect apoptotic cells.

Statistical analysis
All data were presented as the means ± standard deviation (SD). GraphPad Prism version 7 (GraphPad Software, San Diego, CA, USA) was used to process the initial data and plot the results. Statistical analyses were performed with SPSS 13.0 software for Windows (SPSS Inc., Chicago, IL, USA). Student's ttest and one-way analysis of variance were used to evaluate the difference. The difference was considered signi cant at P<0.05.

IFNγ exhibited anti-proliferation activity and induces apoptosis in OSCC cells
To evaluate the function of IFNγ on OSCC cell proliferation and viability, MTT and colony formation assays were performed. IFNγ exerted cytotoxicity in a time-and dose-dependent manner in HB96, HN4n, and CAL27 cells (Fig. 1A). In addition, the treatment of IFNγ (200 ng/ml) for 8 days effectively attenuated the colony-forming capacity of HB96 and CAL27 cells (Fig. 1B). As shown in Fig. 1C and D, IFNγ upregulated the cleaved-PARP and caspase-3 expression in a time-dependent manner in the HB96, HN4, and CAL27 cells. Flow cytometry assays support the above results, given that the proportion of Annexin Vpositive cells increased in a time-and dose-dependent manner after IFNγ treatment ( Fig. 1E and F).

IFNγ-induced autophagy in OSCC cells
As shown in Fig. 2A and B, Beclin 1 and LC3B II expression increased in the IFNγ-treated HB96, HN4, and CAL27 cells in a time-and dose-dependent tendency. P62 level was analyzed to determine whether the autophagosome up-regulation observed after IFNγ treatment was caused by an increase of autophagic activity or a reduced turnover of autophagosomes. P62 was hydrolyzed when OSCC cells were treated with IFNγ, demonstrating IFNγ-induced autophagy. In accordance with western blot results (Fig. 2C), the distribution of LC3B II puncta also increased in the IFNγ-treated OSCC cells compared with untreated control. Furthermore, we examined the morphology of the HB96, HN4, and CAL27 cells after IFNγ induction by transmission electron microscopy (TEM). As shown in Fig. 2D, IFNγ activated autophagy ux by increasing the formation of the initial sequestering compartment (the phagophore), the number of autophagosomes often containing multivesicular and multilamellar structures, and autolysosomes.
Herein, our data indicated that IFNγ induced autophagy in OSCC cells.

IFNγ-induced autophagy via ATG5 signaling in OSCC cells
A set of autophagy-related genes (ATGs) was involved in the dynamic membrane-rearrangement reactions of autophagy. Among these genes, Beclin1, ATG5, and ATG7 represent the major regulators of the classical autophagy pathway in mammalian cells [24]. Using real-time PCR assays (Fig. 3A), we found that ATG5 mRNA expression increased signi cantly after IFNγ treatment in HB96, HN4, and CAL27 cells. Moreover, the suppression of ATG5 expression using siRNAs (Fig. 3C) in the context of IFNγ treatment ( Fig. 3B and D) decreased Beclin 1 and LC3B II expression and increased P62 protein levels.
These results suggested that ATG5 was required for IFNγ-induced autophagy in OSCC cells. Moreover, silencing ATG5, the key regulator of autophagy, signi cantly enhanced the anti-tumor effects of IFNγ using FACS (Fig. 3E), indicating that autophagy inhibitors might synergize with IFNγ in the treatment of OSCC.

CQ inhibits IFNγ-induced autophagy in OSCC cells and synergizes with the anti-tumor effects of IFNγ
Based on previous results, CQ treatment caused the accumulation of both P62 and LC3B II. Consistent results are shown in Fig. 4A. Furthermore, an autophagy ux assay (Fig. 5) was performed in the HN4 and CAL27 cells treated with IFNγ using a tandem uorescent-tagged LC3B reporter plasmid (GFP-mRFP-LC3B) [25]. The yellow uorescence puncta represented the merging image of green and red uorescence in autophagosomes, which indicated impaired autophagy. The red uorescence puncta alone after fusion represented complete autophagic ux. Quanti cation of red (mRFP+ GFP-) and yellow (mRFP+ GFP+) puncta per cell indicates that IFNγ increased autophagy ux (red and yellow puncta). CQ resulted in the accumulation of yellow puncta (hence autophagosomes) induced by IFNγ.
CQ could also synergize IFNγ-mediated anti-tumor effects in OSCC cells. As shown in Fig. 4A, compared with the IFNγ-treated group, the combined application of IFNγ and CQ signi cantly up-regulated the expression of caspase-3 and cleaved-PARP proteins. Flow cytometry assays support the above results (Fig. 4B). The proportion of Annexin V-positive cells increased after the combination treatment. Thus, the combined application of IFNγ and CQ had a synergistic effect in OSCC cells.

Enhanced anti-tumor effect of IFNγ treatment when combined with CQ in vivo
OSCC xenografts in nude mice were used to con rm the in vitro results. As shown in Fig. 6A and B, compared with the group treated with a single agent, the tumor volume in the combination treatment group was signi cantly reduced (P<0.01). According to the TUNEL assay (Fig. 6C), IFNγ up-regulated apoptosis in vivo, especially when combined with CQ. Furthermore, immunohistochemistry (Fig. 6D) showed a reduction of Ki67, a protein relative to proliferation. In accordance with the in vitro results, CQ resulted in the accumulation of both LC3B and P62 proteins. In summary, IFNγ can induce autophagy and exhibit a synergistic effect with CQ on suppressing tumor growth in vivo (Fig. 6E).

Discussion
In this study, we demonstrated that IFNγ simultaneously induced apoptosis and autophagy, and the attenuation of autophagy also synergized the OSCC cell apoptosis mediated by IFNγ. IFNγ-induced autophagy might partly explain the limited effect of IFNγ in solid tumors. Moreover, we demonstrated a synergistic anti-tumor effect in the combination of IFNγ and CQ. This nding might provide the experimental evidence to support the clinical use of combination treatment with IFNγ and autophagy inhibitors in OSCC.
IFNγ is a crucial cytokine in anti-tumor immunity, although a pro-tumorigenic function is also found under certain circumstances [26,27]. In accordance with former research [26,27], our in vitro experiment also con rmed the cytotoxicity of IFNγ and the effective suppression of colony formation in OSCC cells. IFNγ may up-regulate the function of immune cells (increased MHC class I expression, antigen responsive genes, and costimulatory molecules). IFNγ is also capable of inducing the apoptosis of cancer cells via different mechanisms. Notably, IFNγ, either alone or in combination with other cytokines, can induce cellular stress in selected tumor cells, leading to cell death or senescence [28][29][30]. Accordingly, IFNγ has been taken into clinical trials for cancer treatment [31][32][33]. In our study, results of western blot analysis reveal IFNγ-induced up-regulation of caspase-3 and cleaved-PARP protein expression, in a dose-and timedependent manner in vitro. FACS also indicate increasing apoptosis in the OSCC cells, especially with Annexin V. This evidence suggests that IFNγ might be a potential drug in OSCC treatment.
However, former clinical trials of IFNγ have failed to support the effectiveness of IFNγ in malignant tumors. Schiller et al. have reported a good prognosis in patients with melanoma in phase II/III clinical trials of IFNγ [34]. Unfortunately, they failed to detect the e cacious effects of IFNγ, as the response rate was only 5%, with signi cant side-effects [34]. This evidence suggests that only patients with early-stage or disseminated cancer could bene t from IFNγ treatment [34]. It seems that this phenomenon shares something in common with autophagy, which is important in tumor progression. In accordance with our hypothesis, a former study demonstrates that IFNγ induces both apoptosis and autophagy in Atf6-/-mice [35]. However, there is no evidence testifying this phenomenon in human cancer lines in vitro or in vivo. In our present study, in vitro and in vivo experimental processes were performed to verify IFNγ-induced autophagy in OSCC. Since IFNγ has not yet been approved in the therapy of most solid tumors, the autophagy activation in OSCC tissues has not been improved. Therefore, our study is the rst to demonstrate that IFNγ induces autophagy in OSCC, providing a possible mechanism to explain the limited effect of IFNγ therapy in solid tumors.
Autophagy can both promote and inhibit tumor growth and the roles of autophagy vary in different contexts [36]. A cluster of ATGs is involved in the dynamic membrane-rearrangement reactions of autophagy [37]. Former research reported that the knockdown of ATGs, such as ATG5, synergized with chemotherapy in the e cient elimination of cancer cells [38,39]. Herein, autophagy inhibition via the genetic silencing of ATG5 may contribute to the therapy of advanced cancer. According to our results, genetic silencing of ATG5 could inhibit IFNγ-mediated autophagy via the LC3B pathway. Moreover, FACS also reveals that the silencing of ATG5 synergizes IFNγ-induced apoptosis in vitro, demonstrating the cytoprotective role of autophagy during IFNγ treatment and potential strategy of OSCC treatment by IFNγ combined with an autophagy inhibitor.
Despite the complex process of autophagy and the challenges of the treatment strategy, some pharmacologic autophagy inhibitions have been used in clinical trials [36]. Bryant et al. [40] have con rmed that CQ, as an autophagy inhibitor, could promote the effect of ERK inhibitor in pancreatic cancer treatment by attenuating the resulting autophagy of ERK-inhibition. Accordingly, CQ is chosen in our study, and the results demonstrate its synergistic effect with IFNγ in anti-tumor activity. The strong evidence provided by the CQ results supports the launch of a clinical trial to assess the e ciency of IFNγ in combination with an autophagy inhibitor in OSCC. Furthermore, our present results demonstrate that CQ sensitizes the OSCC cells to enhance IFNγ induced apoptosis both in vitro and in vivo.
It is worth noting that increasing evidence has shown a positive relationship between IFNγ and PD-L1 expression [41]. Kim et al. [41] have demonstrated that IFNγ produced by T-cells could up-regulate exosomal PD-L1 expression, which promotes tumor growth through immune escape in non-small cell lung cancer. Unfortunately, the side-effect of IFNγ in immune escaping has not been considered in this research. We are looking forward to exploring the relative tasks in future investigations.

Conclusion
In summary, we have demonstrated that IFNγ signi cantly induces autophagy in OSCC.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and analyzed during the current study available from the corresponding author on reasonable request       Expression of apoptosis and autophagy relative protein, including caspase-3, cleaved-PARP, P62, and LC3B II was assessed by a western blot assay. Quanti cation of the proteins relative to beta-actin OD values is presented. (B) IFNγ-induced apoptosis was synergized by CQ treatment in the HB96, HN4, and CAL27 cells, which was measured by ow cytometry analysis of Annexin V and PI staining. All the P values were compared with the control. *: P<0.05, **: P<0.01. Expression of apoptosis and autophagy relative protein, including caspase-3, cleaved-PARP, P62, and LC3B II was assessed by a western blot assay. Quanti cation of the proteins relative to beta-actin OD values is presented. (B) IFNγ-induced apoptosis was synergized by CQ treatment in the HB96, HN4, and CAL27 cells, which was measured by ow cytometry analysis of Annexin V and PI staining. All the P values were compared with the control. *: P<0.05, **: P<0.01.

Figure 5
Autophagy ux assay was performed in the HN4 and CAL27 cells treated with IFNγ (200 ng/ml for 48 h) using a tandem uorescent-tagged LC3B reporter plasmid (GFP-mRFP-LC3B). The yellow uorescence puncta represented the merging image of green and red uorescence in autophagosomes, which indicated impaired autophagy. The red uorescence puncta alone after fusion represented complete autophagic ux. Quantitative analysis of red and yellow LC3 puncta was reported as mean ± SD. All the P values were compared with the control. *: P<0.05, **: P<0.01.

Figure 5
Autophagy ux assay was performed in the HN4 and CAL27 cells treated with IFNγ (200 ng/ml for 48 h) using a tandem uorescent-tagged LC3B reporter plasmid (GFP-mRFP-LC3B). The yellow uorescence puncta represented the merging image of green and red uorescence in autophagosomes, which indicated impaired autophagy. The red uorescence puncta alone after fusion represented complete autophagic ux. Quantitative analysis of red and yellow LC3 puncta was reported as mean ± SD. All the P values were compared with the control. *: P<0.05, **: P<0.01.