Targeting Autophagy Facilitates T Lymphocyte Migration by Inducing the Expression of CXCL10 in Gastric Cancer Cell Lines

Autophagy is a type of cellular catabolic degradation process that occurs in response to nutrient starvation or metabolic stress, and is a valuable resource for highly proliferating cancer cells. Autophagy also facilitates the resistance of cancer cells to antitumor therapies. However, the involvement of autophagy in regulating CXCL10 expression in gastric cancer (GC) cells and T lymphocyte migration remains unclear. In this study, we aimed to investigate the effect of autophagy inhibition on CXCL10 expression and T lymphocyte infiltration in GC and elucidate the underlying mechanism. Analysis of public databases revealed a positive correlation between CXCL10 expression and both prognosis of patients with GC and the expression profile of T lymphocyte markers in the GCs. Chemotaxis and spheroid infiltration assays revealed that CXCL10 induced T lymphocyte migration and infiltration into GC spheroids, an in vitro three-dimensional cell culture model. In addition, in vitro autophagy inhibition in GC cells increased CXCL10 expression under both normal and hypoxic culture conditions. Further investigation on the underlying mechanism showed that in vitro autophagy inhibition suppressed the JNK signaling pathway and further enhanced CXCL10 expression in GC cells. Collectively, our results provide novel insights for understanding the role of autophagy in regulation of intra-tumor immunity.

Gastric cancer (GC) is the fifth most common malignancy diagnosed worldwide, with 952,000 estimated new cases and 723,000 GC related-deaths in 2012 (19). Although immune check-point inhibitors have shown promising results for GC treatment, the objective response rates remain low (18,20). Thus, the effectiveness of this immunomodulatory strategy depends not only on the unleashing of pre-existing immunity but also on the presence of a baseline immune response (21). In fact, intratumor T lymphocyte recruitment is one of the potential ratelimiting steps in immunotherapy; therefore, many investigators have focused on the role of intra-tumoral chemokines in TIL recruitment into the tumor (22,23).
It is well-known that T lymphocyte infiltration into the tumor is always insufficient when the chemokine receptors expressed on T lymphocytes do not match to the tumorsecreted chemokines (24). CXCR3, a predominant chemokine receptor expressed on TILs, is expressed in several solid tumors, including melanoma (25), colorectal cancer (26), and breast cancer (27). Moreover, TILs in lymphocyte-rich GCs predominantly express CXCR3 (28). Among the CXCR3 ligands, CXCL10 was reported to be associated with T lymphocyte infiltration into tumors. For example, CXCL10 expression was associated with T lymphocyte recruitment in melanoma metastases (25). In addition, intra-tumor induction of CXCL10 enhanced the infiltration of CXCR3+ cytotoxic T lymphocytes, thereby improving the antitumor effect of other therapies in some rodent solid tumor models (29,30). However, the association between CXCL10 expression and T lymphocyte infiltration in GC remains poorly understood.
In recent years, autophagy in GC pathogenesis has been explored extensively, and autophagy inhibition is being considered as a potential strategy for GC treatment (31). Autophagy is critical for the digestion of intracellular contents and generation of energy to control cellular homeostasis (32). Autophagy was reported to play a pivotal role in GC cell survival and enhance tumor cell resistance to antitumor therapies (31). Therefore, autophagy inhibition may alter this tumor protective mechanism and potentiate anticancer drug-induced cell death in GC. In fact, an autophagy inhibitor chloroquine (CQ) was reported to improve the chemosensitivity of GC cells to platinum-based antitumor drugs (33,34). Li et al. demonstrated that treatment with 3-MA, an alternative autophagy inhibitor, enhanced the curcumin-induced antitumor effect (35). Interestingly, a recent study showed that autophagy inhibition could induce CCL5 expression in melanoma cells, resulting in tumor regression facilitated by NK cell migration into the tumor bed (36).
In this study, we investigated the effect of autophagy inhibition on CXCL10 expression in GC cells and T lymphocyte migration toward GC cells. We also attempted to elucidate the mechanism underlying the observed effects of autophagy inhibition on CXCL10 expression in GC cells.

Public Dataset Mining
Kaplan Meier-plotter (http://kmplot.com/analysis/) is an online database that enables evaluation of the effect of over 54,000 genes on survival in several cancer types, including GC, breast cancer, ovarian cancer, and lung cancer (37). This database was used to obtain prognostic information on CXCL10. Survival information and gene expression data were from Gene Expression Omnibus (GEO), European Genome-phenome Archive (EGA), and The Cancer Genome Atlas (TCGA) database.
Gene Expression Profiling Interactive Analysis (GEPIA; http://gepia.cancer-pku.cn/index.html) is a customizable online tool developed by Zhang lab of Peking University to analyze gene expression data in both tumor and normal tissues on the basis of TCGA and Genotype-Tissue Expression (GTEx) data (38). GEPIA was used for correlation analysis and for investigating the expression levels of autophagy-related genes (ATGs) between GCs and the normal tissues.
Cell Sorting and Activation of CD3+ T lymphocytes CD3+ T lymphocytes were isolated from cryopreserved human peripheral blood mononuclear cells (PBMCs; StemExpress) using MACS microbeads (Miltenyi Biotec). After separation, T lymphocytes were stimulated using CD3/CD28 Dynabeads (Life Technologies) for 2 days, as described previously, and re-cultured without any external stimuli for another 2 days to induce the expression of CXCR3 (39). The primed T lymphocytes were used in the chemotaxis and spheroid infiltration assays.

Chemotaxis Assay
The chemotaxis assay was performed in CytoSelect TM 24-well cell migration assay kit (5 µm pore size; Cell Biolabs) per the manufacturer's instructions (Figure 2A). Briefly, the primed T lymphocytes were prepared at density of 3 × 10 6 cells/ml in serum-free RPMI 1640 medium containing 0.5% bovine serum albumin (BSA), 2 mM MgCl 2 , and 2 mM CaCl 2 . For each well, the cells were placed in upper chamber (3 × 10 5 cells/100 µl) and the medium was loaded in the lower chamber. The plate was then incubated in a 37 • C cell culture incubator for 5 h. The migrated cells were dissociated from the membrane, lysed, and detected using the patented CyQUANT R GR Dye (Life Technologies).

Tumor Spheroids and Spheroid Infiltration Assay
NCI-N87 spheroids were established using 96-well EZSPHERE SP micro-plates (Nacalai Tesque). The culture plate has a concave and ultra-low attachment bottom surface so that the cells adhere to each other, but not with the bottom surface of the plate. Therefore, the cells did not spread out on plastic, but formed spheroids. Here, the NCI-N87 cells were transfected with the pIREShyg3-CXCL10 plasmid; 1 day later, 8 × 10 4 CXCL10-transfected NCI-N87 cells were seeded with 200 µl medium in each well. The spheroids were formed 4 days after seeding. Then, 8 × 10 5 primed T lymphocytes were added into each well and incubated overnight ( Figure 2C). The spheroids were then washed three times with PBS to remove the loosely attached T lymphocytes, fixed in 4% paraformaldehyde for 2 h, and embedded into paraffin for immunohistochemistry analysis.

Immunohistochemistry
Paraffin blocks were sectioned using a microtome to obtain 4 µm thick sections for immunostaining. The paraffin sections were dewaxed in xylene and hydrated in decreasing concentrations of ethanol. Sections were then incubate in 1 × DIVA Decloaker antigen retrieval solution (Biocare Medical) at 110 • C for 15 min using the decloaking chamber (Biocare Medical). Following antigen retrieval, sections were incubated in peroxidazed 1 solution (Biocare Medical) at room temperature for 5 min to quench endogenous peroxidase activity. After blocked with background sniper at room temperature for 10 min, sections were incubated with a monoclonal rabbit anti-human CD3 antibody (0.3 µg/ml; Biocare Medical) in Dako REAL antibody diluent (Dako) at room temperature for 1 h. Sections were subsequently incubated with HRP-labeled goat anti-rabbit IgG polymer (Dako) at room temperature for 30 min. Finally, sections were exposed to liquid DAB+ substrate chromogen system (Dako) at room temperature for 5 min and counterstaining was performed using Gill's hematoxylin (Sigma).

Quantitative RT-PCR
Total RNA was extracted from the GC cell lines using RNeasy Plus Mini Kits (QIAGEN). Quantitative RT-PCR and data analysis were performed as described in our previous work (40,41). Briefly, the SuperScript TM IV First-Strand Synthesis System (Life Technologies) was used to synthesize cDNA. PCR was performed and quantified using Power SYBR Green PCR Master Mix (Life Technologies). Primers used in the realtime quantitative PCR were as follows: CXCL10 (accession no. NM_001565), sense primer 5 ′ -AAAAGAAGGGTGAGAAGAG-3 ′ and antisense primer 5 ′ -AAGAACAATTATGGCTTGAC-3 ′ ; ATG5 (accession no. NM_004849), sense primer 5 ′ -GCAACTCTGGATGGGATTGC-3 ′ and antisense primer 5 ′ -AGGTCTTTCAGTCGTTGTCTGAT-3 ′ ; ATG7 (accession no. NM_ 006395), sense primer 5 ′ -CATGGTGCTGGTTTCCTTGC-3 ′ and antisense primer 5 ′ -GCTACTCCATCTGTGGGCTG-3 ′ ; GAPDH (accession no. NM_002046), sense primer 5 ′ -CGGATTTGGTCGTATTGGG-3 ′ and antisense primer 5 ′ -CTGGAAGATGGTGATGGGAT-3 ′ . The relative target gene mRNA level was calculated using the Ct method. The threshold cycle (Ct) values of the target gene mRNAs were initially normalized to the Ct values of the internal control GAPDH in the same samples: Ct = Ct (the target gene) -Ct (GAPDH). These values were further normalized to the control group: Ct = Ct (sample group) -Ct (control group). The fold change was then determined (2 − Ct ). The relative target gene mRNA level represents an average fold calculated from separate experiments. PCR was performed at least three times, and similar results were observed.

Luminex Assay
The protein level of CXCL10 in the cell culture supernatant was assessed using the human Magnetic Luminex Assay (R&D Systems), which was performed per the manufacturer's instructions. Briefly, all the samples and standards were first mixed with the CXCL10 antibody coated magnetic microparticles and incubated for 2 h at room temperature on a horizontal orbital microplate shaker set at 800 rpm. After washing the microparticles, biotinylated detector antibodies were added and incubated for 1 h at room temperature on the shaker set at 800 rpm. Following a wash to remove any unbound biotinylated detector antibody, streptavidin-phycoerythrin conjugates were added and incubated for 30 min at room temperature on the shaker set at 800 rpm. Finally, the protein level of CXCL10 in the cell culture supernatant was analyzed using the Bio-Plex TM 200 system (Bio-Rad).

Western Blot
Cell lysis, protein extraction, and western blot analyses were performed as described in our previous work (40). Proteins were dissolved in a lysis buffer and separated using SDS/PAGE for western blot analyses. Primary antibodies included rabbit anti-Phospho-SAPK/JNK (Thr183/Tyr185), anti-SAPK/JNK, anti-Phospho-c-Jun (Ser73), anti-c-Jun, anti-ATG5, anti-LC3B and anti-GAPDH (Cell Signaling Technology). Secondary antibody was HRP-conjugated anti-rabbit IgGs (Life Technologies). The densitometric analyses of western blotting images were performed using ImageJ software (National Institutes of Health).

Cell Viability Assay
Cell counting kit-8(CCK-8, Dojindo) was used to evaluate cell viability based on the dehydrogenase activity. AGS cell suspensions were first dispensed in a 96-well plate (1 × 10 4 in 100 µL/well) and cultured in DMEM with 10% FBS at 37 • C for 24 h, and then were treated with vehicle, 10 and 20 µM CQ, respectively. After incubation for 0, 1, 2, and 3 days, 10 µl CCK-8 solution was added to each well and the plate was incubated at 37 • C for 1 h. Finally, the absorbance at 450 nm was measured by using a SpectraMax M5 microplate reader (Molecular Devices).

Statistical Analysis
Data represent mean ± SE. Experimental data were subjected to statistical analyses using one-way ANOVA followed by Tukey post-hoc test or student's t-test with a significance level of P < 0.05.  (Figures S1A,D), lung cancer (Figures S1B,E), or ovarian cancer (Figures S1C,F). In addition, correlation analysis in GEPIA showed strong positive correlation between CXCL10 expression and several T lymphocyte markers such as CD3D (Figure 1C

CXCL10 Recruited T lymphocytes in the Chemotaxis and GC Spheroid Infiltration Assay
Binding specificities of chemokines to their specific receptors are well-defined (42), and high expression of CXCR3 (the receptor of CXCL10) on effector T lymphocytes has been reported (43). Therefore, to confirm whether CXCL10 induces T lymphocyte infiltration, CXCR3+ T lymphocytes were required for the chemotaxis and spheroid infiltration assays. Because of the difficulties in detecting CXCR3 on most of the T lymphocytes freshly isolated from PBMCs of normal donors (Figure S2), CD3/CD28 Dynabeads were used to activate the T lymphocytes and induce the expression of CXCR3. After activation, over 90% of CD3/CD28 Dynabeads treated T lymphocytes were CXCR3+ (Figure S2). Chemotaxis assays revealed that CXCL10 recruited the primed T lymphocytes in a dose-dependent manner ( Figure 2B).
In addition, to further confirm whether CXCL10 facilitates T lymphocyte infiltration in GC, GC spheroids were established using NCI-N87 cells transfected with CXCL10 or control plasmid (Figures S3, S4). Compared with the control vector-transfected spheroids, the CXCL10-overexpressing GC spheroids showed significantly high infiltration of T lymphocytes (Figures 2D-F).

Autophagy Inhibition Enhanced CXCL10 Expression in AGS Cells
It is well-known that ATG proteins are critical for the formation of autophagosome and the activity of autophagy (44,45). ATG5 and ATG7 are two of the most important components of the ATG family; therefore, ATG5 or ATG7 ablation is sufficient to impair autophagic functions (46)(47)(48)(49)(50)(51)(52). In this study, we aimed to induce ablation of ATG5 or ATG7 in AGS cells, as AGS cells showed the highest endogenous CXCL10 expression level among the available GC cell lines (Figure S3). ATG5 siRNA transfection in AGS cells significantly suppressed ATG5 expression at both mRNA ( Figure 4A) and protein levels (Figures 7A,F). Such ATG5 knockdown inhibited autophagy, as demonstrated by decreased LC3II/LC3I ratio (Figures 7A,E). In addition, ATG5 knockdown significantly induced CXCL10 mRNA expression in AGS cells ( Figure 4B) and significantly increased CXCL10 secretion by AGS cells (Figure 4C). Similarly, ATG7 knockdown significantly induced CXCL10 expression at both mRNA and protein levels (Figures 4E,F).
CQ inhibits autophagic flux by decreasing the fusion of autophagosome-lysosome (53). Therefore, we used CQ to further confirm whether autophagy inhibition could induce CXCL10 expression in AGS cells. Treatment with 20 µM CQ significantly induced the accumulation of LC3-II in a time-dependent manner (Figures 5A,B), as reported previously (53)(54)(55). Furthermore, 20 µM CQ significantly induced CXCL10 mRNA expression in a time-dependent manner in AGS cells without affecting the cellular viability (Figure 5C, Figure S5). The maximal induction effect was observed at day 3. When incubation time was fixed for 3 days, Treatment with 10 and 20 µM CQ significantly induced CXCL10 mRNA in AGS cells (Figure 5D). In addition, CXCL10 secretion by AGS cells treated with 20 µM CQ was significantly higher than that by control cells (Figure 5E).

Autophagy Inhibition Facilitated T lymphocyte Migration by Inducing CXCL10 Secretion
Chemotaxis assay revealed that T lymphocyte recruitment by culture supernatant of ATG5-knockdown AGS cells was significantly higher than that by culture supernatant of control cells (Figures 6A,B). This T lymphocyte recruitment was effectively blocked in the presence of neutralizing anti-CXCL10 antibody ( Figure 6B).

Autophagy Inhibition Enhanced CXCL10 Expression by Suppressing the Inhibitory Effect of JNK Signaling
Next, we investigated the mechanism underlying the induction of CXCL10 expression via autophagy inhibition. Here, we demonstrated that ATG5 knockdown was sufficient to inhibit autophagy (Figures 7A,E,F) and investigated the levels of components of the JNK signaling pathway in AGS cells. ATG5 knockdown significantly decreased the levels of phospho-JNK (Figures 7A,B), phospho-c-Jun (Figures 7A,C), and c-Jun (Figures 7A,D), thereby suppressing JNK signaling. Treatment with the JNK inhibitor SP600125 resulted in a dose-dependent increase in CXCL10 mRNA expression in AGS cells, and 20 and 40 µM SP600125 showed a significant increase in CXCL10 mRNA levels ( Figure 7G). In addition, treatment with 100 ng/ml anisomycin, a JNK activator, significantly inhibited CXCL10 mRNA expression in control-vector transfected AGS cells and significantly suppressed the ATG5 knockdown-induced increase in CXCL10 mRNA expression ( Figure 7H). Collectively, these data suggest that autophagy inhibition induced CXCL10 expression via suppression of the inhibitory effects of JNK signaling.

Autophagy Inhibition Induced CXCL10 Expression in CoCl 2 -Treated AGS Cells
Intra-tumor hypoxia is an important characteristic of 50-60% malignant tumors (56). Moreover, GEPIA showed that mRNA level of HIF1α, the hypoxia marker, in GCs was significantly higher than that in normal gastric tissues ( Figure 8A). Therefore, we investigated the effect of autophagy inhibition on CXCL10 expression under hypoxia mimetic conditions. Treatment with CoCl 2 , a hypoxia mimetic reagent, significantly increased HIF1α protein level in AGS cells (Figures 8B,C). Treatment with 200 µM CoCl 2 significantly increased the LC3II/LC3I ratio, indicating increased autophagic activity in AGS cells (Figures 8B,D). Furthermore, CoCl 2 decreased CXCL10 expression in a dose-dependent manner, and both 50 and 200 µM CoCl 2 significantly decreased CXCL10 mRNA levels in AGS cells (Figure 8E). ATG5 knockdown significantly increased CXCL10 expression in CoCl 2 treated AGS cells at both mRNA and protein levels (Figures 8F,G).

DISCUSSION
In this study, we demonstrated that intra-tumor CXCL10 is an important chemokine that contributes to intra-tumor infiltration of T lymphocytes in GC. We also showed that autophagy inhibition could effectively facilitate T lymphocyte migration into the tumor microenvironment by inhibiting the JNK pathway and further inducing the expression of CXCL10 (Figure 9). This might represent a novel therapeutic strategy to enhance the effectiveness of solid tumor immunotherapies such as immune check-point blockade.
It is well-known that the levels of T lymphocyte infiltration into the tumor determine the efficacy of immunotherapy. Primed T lymphocytes gain the expression of certain homing molecules (such as CXCR3) on their surface and thus obtain the capability to migrate toward the tumor site (24). In our study, CXCL10, the well-accepted CXCR3 ligand, functioned as a chemoattractant for T lymphocytes (Figures 2A,B) and recruited T lymphocytes to GC spheroids (Figures 2C-F). Moreover, CXCL10 expression was positively correlated with overall survival (Figure 1A) and relapse-free survival ( Figure 1B) in patients with GC. Consistent with our observations, Barash et al. indicated that CXCL10 administration not only induced the infiltration of T cells FIGURE 9 | Schematic representation of the mechanism underlying the increased CXCL10 expression in autophagy inhibited GC cells. In GC cells, autophagy inhibition suppressed JNK signaling and subsequently induced CXCL10 expression. As a result, increased CXCL10 recruited more T lymphocytes into gastric tumors. and NK cells into myeloma tumors but also reduced the accumulation of Treg cells at the tumor site, thereby suppressing tumor progression (57). In addition, CD26 inhibition was reported to enhance T lymphocyte trafficking into melanoma tumor by inducing the intra-tumor expression of CXCL10, further improving the efficacy of immunotherapy (58). In addition to being a potent chemoattractant for T lymphocytes, CXCL10 also inhibits tumor growth via suppressing angiogenesis (59)(60)(61)(62)(63). Furthermore, CXCL10 overexpression improved the radiosensitivity of tumors in a rodent cervical cancer model (64). In total, the evidence suggests that CXCL10 could be a potential novel candidate for the GC targeted therapy.
Considering the fact that autophagy was not measurable, the indicators for autophagy activation were judged by expression of ATGs. In our study, GEPIA indicated that the expression of some key autophagy genes in GC were significantly higher than that in normal tissue (Figure 3). These results were consistent with previous observations in established solid tumors (32,65). However, previous findings on the regulatory effect of autophagy inhibition on CXCL10 expression are not consistent. For instance, two studies showed that ATG5 knockdown significantly suppressed influenza-virus induced CXCL10 expression in macrophages (66,67). Two other studies reported that deletion of some other key autophagy genes, FIP200 or BECN1, led to increased CXCL10 production in mammary tumor cells (68) or melanoma cells (36). Nevertheless, the regulation of CXCL10 expression in GC cells has not yet been reported.
We next investigated the mechanism underlying the induction of CXCL10 expression by autophagy inhibition. We found that JNK activator decreased and JNK inhibitor increased CXCL10 expression in AGS cells. In addition, autophagy inhibition significantly decreased the activity of JNK signaling pathway. Thus, these data suggest that autophagy inhibition induces CXCL10 expression by suppressing the inhibitory effect of JNK signaling in AGS cells. In contrast, Mgrditchian et al. reported that BECN1 deletion induced CCL5 expression by activating the JNK signaling pathway, which in turn recruited more NK cells into melanoma tumors (36). This difference in the effect of autophagy inhibition on JNK signaling may be associated with tumor types.
Next, we investigated whether autophagy inhibition also induced CXCL10 expression under hypoxia mimetic conditions. Because of the inadequate oxygen supply and increased energy consumption within the tumor microenvironment, hypoxia is one of the most important characteristics of solid tumors, especially in the advanced stages (77). In the hypoxic microenvironment, autophagy flux is enhanced along with increased tumor growth (78). Advanced tumors have been shown to use autophagy to promote tumor survival (79,80). Our current observations that ATG5 knockdown induced CXCL10 expression in CoCl 2 -treated AGS cells support a scientific basis of autophagy inhibition as a potential combinational therapy strategy for immunotherapy.
Apart from recruiting T lymphocytes into solid tumors and enhancing the sensitivity to anti-tumor therapy, autophagy deficiency was also reported to cause some cancer related pathology (81,82). For instance, the mutation of ATGs was reported in tumor cells (83). Because of the function of autophagy in counteracting cellular stress, some ATGs were considered as tumor suppressors in rodent tumor models (45,(84)(85)(86). In addition, Yang et al. indicated that fluorouracil inhibited the growth of GC cells via ATG6 activation (87). In this case, autophagy also sometimes seems as a protective mechanism in tumor initiation period. Overall, autophagy might regulate tumorigenesis in a tumor stage-specific manner.
In summary, to the best of our knowledge, this is the first report on the regulatory effects of in vitro autophagy inhibition on CXCL10 expression in GC cells and its potential mechanism in recruiting T lymphocytes into the tumor. These findings provide novel insights into understanding the functions of autophagy in immunotherapy. Furthermore, our results highlight the potential of autophagy inhibition to be used in combination with immunotherapy approaches such as immune checkpoint blockade. Our findings also suggest CXCL10 as a potential novel candidate for targeted therapy against GC.

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed in this study. Kaplan Meier-plotter can be found here: http://kmplot.com/analysis/. The GEPIA (Gene Expression Profiling Interactive Analysis) can be found here: http://gepia.cancer-pku.cn/index.html.

AUTHOR CONTRIBUTIONS
QM and LH designed the study. QM and YZ performed the experiments and statistical analysis. QM drafted the manuscript. QM, YZ, and LH revised the manuscript. All authors read and approved the final manuscript.