Background
Gastric cancer is a prevalent malignant tumor of the digestive tract and remains the third leading cause of cancer-related death worldwide. An estimated 951,600 newly diagnosed cases and 723,100 deaths of gastric cancer occurred in 2012 [
1]. The 5-year survival rate for patients with advanced gastric cancer is only 5–20% with 10 months of median overall survival [
1]. Therefore, new molecular targets and therapeutic approaches are urgently needed.
Autophagy is a highly conserved homeostatic process that involves the formation of a double-membrane structure, autophagosome, which subsequently fuses with the lysosome to generate autolysosome leading to the degradation of the cellular proteins and damaged organelles. This catabolic pathway plays a pivotal role in cell survival, cellular metabolism and immune responses. Growing evidence reveals that the role of autophagy in tumorigenesis is complex and context-dependent [
2]. On one hand, autophagy can inhibit tumor formation by reducing oxidative stress and DNA damage in normal tissues [
3]. On the other hand, autophagy can promote tumor cell survival by providing cells with energy and vital compounds upon various stress stimuli in developed cancers [
2,
4]. Also, autophagy can be activated in response to cytotoxic chemotherapeutics, acting as a mechanism of drug resistance [
5‐
7]. Thus, modulating autophagy is an attractive option, which allows us to harness this process for improving the disease management in cancers.
Programmed cell death-1 (PD-1) with its ligand (PD-L1) are important immune checkpoint proteins. Elevated expression of PD-L1 receptors on cancer cell membranes has been observed in many cancer types. PD-L1 can interact with PD-1 and CD80 on the surface of T-cells, thus protecting cancer cells from immune-mediated rejection by inhibition of T effector functions [
8,
9]. PD-L1 expression can be induced by inflammatory cytokines, such as interferon (IFN)-γ [
10] secreted by infiltrating lymphocytes or induced by tumor-cell intrinsic signaling, including nuclear factor (NF)-κB, mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase (PI3K), mammalian target of rapamycin (mTOR) and Janus kinase/signal transducers and activators of transcription (JAK/STAT) [
11]. In addition, PD-L1 is regulated by the tumor suppressor genes
PTEN and
LKB1 as well as epithelial-mesenchymal transition-related molecules [
12,
13]. More recently, evidences suggest that PD1 receptor and its ligand PD-L1 can have crosstalk with autophagy in cancer cells. In mouse melanoma and human ovarian cancer, tumor cell-intrinsic PD-L1 upregulates mTOR complex 1 signaling to inhibit autophagy and sensitizes tumor cells to clinically available autophagy inhibitors [
14]. Recent work shows that CMTM6 co-localizes with PD-L1 at the cell membrane and in endosome, where it protects PD-L1 from lysosome-mediated degradation in a broad range of cancer cells [
15]. Defective autophagy has also been shown to promote PD-L1 expression in cerulein-treated Atg5
L/L mice with pancreatitis [
16]. The link between autophagy and PD-L1 in gastric cancer is unclear. Here, we investigated if tumor-intrinsic PD-L1 could be regulated by autophagy in gastric cancer. To test our hypothesis, we determined if inhibition of autophagy could increase PD-L1 levels in human gastric cancer cells.
Methods
Gastric cancer cell lines
Eight gastric cancer cell lines (AGS, BGC823, HGC27, MGC803, MKN28, MKN45, NCI-n87 and SGC7901) and a human normal gastric epithelial cell line (GES-1) were used in this study. Cell lines were maintained in RPMI-1640 medium or DMEM medium with 10% fetal bovine serum.
Human sample collection
One hundred and thirty-seven primary gastric cancer samples were collected during surgical resection at Peking University Cancer Hospital in Beijing, China. None of these patients received preoperative chemotherapy or radiotherapy. The diagnoses of gastric cancer were all histologically confirmed and all subjects provided informed consent for obtaining the study specimens. The study protocol was approved by the Clinical Research Ethics Committee of Peking University Cancer Hospital and Institute.
Reagents, antibodies and commercial kits
RPMI1640 medium (72400) and DMEM medium (10564) are products from Life Technologies. 3-methyladenine (M9281), bafilomycinA1 (B1793), chloroquine (C6628), rapamycin (R0395) and phytohemagglutinin-M (PHA, L8902) are from Sigma-Aldrich. BMS 345541 (S8044) is from Selleck. The following primary antibodies were used: microtubule-associated light chain 3 (LC3B, NB100–2220, Novus Biologicals), LC3A/B (13,082, Cell Signaling), p62/SQSTM1 (H00008878-M01, Novus Biologicals), PD-L1 (NBP1–76769, Novus Biologicals), PD-L1 (59,949, Cell Signaling), PD-L1 (Spring Bio, SP142), ATG5 (12,994, Cell Signaling), ATG7 (SAB4200304, Sigma-Aldrich), β-actin (4967, Cell Signaling), CD45 (368,508, Biolegend), CD8a (301,041, Biolegend), CD4 (357,408, Biolegend), FITC Mouse IgG1(400,110, Biolegend), PD-L1 (329,708, Biolegend), APC Mouse IgG2b (300,907, Biolegend), and 7-AAD (420,404, Biolegend).
RNA interference
The expression of ATG5, ATG7, PD-L1, SQSTM1 and p65 was lowered using target-specific small interfering RNA (siRNA) molecules purchased from Qiagen as follows: Control siRNA (SI03650318), ATG5 siRNA (SI02655310), ATG7 siRNA (SI02655373), PD-L1 siRNA (SI03093076, SI03021158, SI00103250, SI00103243), SQSTM1 siRNA (SI00057596), RELA siRNA (SI02663101, SI02663094, SI00301672, SI05146204). Two hundred picomoles of gene-specific or control siRNA was transfected into cells at 40–60% confluence using Lipofectamine™ 3000 reagent (Invitrogen, 30,000–15) according to the manufacturer’s instructions.
Animal experiments
MKN45 cells (1 × 107 cells in 0.1 ml phosphate-buffered saline) were injected subcutaneously into the dorsal left flank of 4-week-old male BALB/c nude mice (n = 5 per group). Tumor diameter was measured every 2 days for 3 weeks. Pharmacological modulation of autophagy was achieved by intraperitoneal administration of chloroquine (50 mg/kg) every other day for 3 weeks. Tumor volume (mm3) was estimated by measuring the longest and shortest diameter of the tumor and calculating as previously described. All experimental procedures were approved by the Animal Ethics Committee of Peking University Cancer Hospital and Institute.
Lymphocyte preparation
Peripheral blood mononuclear cells (PBMC) were isolated from heparinised peripheral blood samples obtained from gastric cancer patients by Ficoll-Paque (GE Healthcare Life Sciences) density gradient centrifugation. To induce production of PD-1, PBMC were resuspended in RPMI-1640 containing 5 mg/mL PHA, 5% heat inactivated human AB serum, 1% penicillin, streptomycin, and amphotericin (Gibco), and incubated for 48 h [
17]. This PBMC culturing method was used to induce proliferation of activated T lymphocytes by mitogen activation and precondition them to express PD-1. Then cells were rested overnight in the same growth condition minus the PHA. These cells were then co-cultured with the gastric cancer cells.
Drug treatment of melanoma cells and coculture with lymphocytes
Gastric cancer cells were plated on two sets in 12-well plates, and on the following day, treated either with DMSO, CQ (chloroquine), 3-MA (3-methyladenine), Baf (bafilomycin A1) or Rap (rapamycin). After another 24-h period, one set was treated with media and the other set with INF-γ. The final concentrations of the drugs were 16 μmol/L for CQ, 10 mmol/L for 3-MA, 10 nmol/L for Baf, 100 nmol/L for Rap, and 200 U/mL for INF-γ. In the case of cocultures, all the steps and conditions were the same, and on the next day, a suspension of lymphocytes (primed as described in the above) was added to each well. The final concentration of lymphocytes was 550,000 cells/mL. Each assay was repeated at least twice.
Histology and immunohistochemical staining
Formalin-fixed and paraffin-embedded blocks were sectioned at 5 μm and stained with hematoxylin and eosin. Immunohistochemistry was performed on paraffin sections of gastric cancer tissues using anti-LC3B antibody (1:2000), anti-p62/SQSTM1 antibody (1:2000) or anti-PD-L1 antibody (1:100). The immunostaining score was estimated based on the positive cell and the staining intensity, as described previously [
18]. The percentage of positively stained cells was graded as follows: grade 0, < 5%; grade 1, 5–25%; grade 2, 25–50%; grade 3, > 50%. Immunostaining intensity was rated as follows: 0, negative; 1, weak; 2, moderate; and 3, strong. The total expression score was the product of the aforementioned factors, which ranged from 0 to 9. The expression was grouped into low expression (scores of 0–3) and high expression (scores of 4–9).
Real-time PCR
RNA was extracted using Trizol reagent (15596–026, Life technologies) and reverse-transcribed using SuperScript® III Reverse Transcriptase (18080–093, Life technologies). Real-time PCR (Applied Biosystems 7500 Fast Real-Time PCR System, Life technologies) was performed using the Power SYBR® Green PCR Master Mix with gene-specific primers: PD-L1, 5′-CAATGTGACCAGCACACTGAGAA-3′ and 5′-GGCATAATAAGATGGCTCCCAGAA-3′.
Flow cytometry
Gastric cancer cells were probed with phycoerythrin-conjugated anti-PD-L1antibody. In the coculture experiments to distinguish gastric cancer cells from the immune cells, samples were stained with both anti-PD-L1 and anti-CD45 (Fig.
5a). All live/dead discrimination was performed with 7-aminoactinomycin D (7AAD). Lymphocytes were stained with CD45, CD8a, and CD4. All samples were run on a BD Accuri™ C6 Plus flow cytometer. Cells were gated according to the following schema: morphology was determined by using the area of the forward scatter emission peak (FSC-A) versus the area of the side scatter emission peak (SSC-A). Segregation of single cells was determined using SSC-A versus the width of the side scatter emission (SSC-W). Comparing the 7AAD with the APC emission peak allowed analysis of PD-L1 for the gastric cancer cells. For coculture assays, live/dead and lymphocyte discrimination was determined by comparing area of the 7AAD emission peak with the area of the CD45 emission peak and then 7AAD with APC for gastric cancer cell. Median fluorescent intensity (MFI) of PD-L1 was taken from plots of PD-L1.
Statistical analysis
The results are expressed as mean ± standard deviation (SD). Differences between two groups were compared by the Mann-Whitney U test or Student’s t test where appropriate. Multiple group comparisons were made by the Kruskal-Wallis test or one-way analysis of variance (ANOVA) where appropriate. The χ2 test was used for comparison of patient characteristics and distributions of expression and covariates by vital status. Crude relative risks (RRs) of death associated with expression of autophagy markers and other predictor variables were estimated by univariate Cox proportional hazards regression model. The difference in tumor growth rate between the two groups of nude mice was determined by repeated-measures analysis of variance. P values < 0.05 were taken as statistically significant.
Discussion
Autophagy has opposing, context-dependent roles in cancer and perturbations in autophagy are found in gastric cancer [
24]. As one of the most important survival mechanisms, autophagy helps tumor cells to adjust and adapt to an unfavourable environment, to escape from immune surveillance, and hence to promote tumor growth. Recent studies have delineated the mechanism underlying autophagy and the intricate involvement of PD-L1/PD1 axis in cancer cells. A study by Clark et al. found that tumor-intrinsic PD-L1 signals regulate cell proliferation and autophagy in ovarian cancer and melanoma. Tumor cells with high levels of PD-L1 expression are more sensitive to autophagy inhibitors than cells with lower PD-L1 levels in murine melanoma cells and human ovarian cancer cells [
14]. Melanoma cell-intrinsic PD-1 cooperates with PD-L1 to promote tumorigenesis and modulates downstream effectors of mTOR signaling [
17]. Blockade of PD-L1 in sarcoma cells inhibits mTOR activity and dampens glycolysis, thereby restoring glucose in tumor microenvironment [
25]. Depletion of glucose also induces autophagy through the mTOR complex 1 pathway [
26]. Until now, the link between autophagy and the immune checkpoint molecule PD-L1 is not quite well understood in gastric cancer. Here, we demonstrated that inhibition of autophagy by pharmacological or RNA interference approach could induce the expression of PD-L1, unveiling the unreported intrinsic regulation of PD-L1 by autophagy.
As a ligand of PD-1, PD-L1 is a transmembrane protein that is expressed on a wide variety of cells including tumor cells to inhibit CD8
+ T cell activities and suppress antitumor immunity. So, the PD-L1 protein on cell membrane mainly exerts its antitumor effect. Thus, we detected the expression of surface PD-L1 by flow cytometry in accordance with many of the published papers studying PD-L1 to determine its functional proportion [
14,
15,
27‐
29]. Also, we evaluated the total expression of PD-L1 protein to demonstrate the upregulation of PD-L1 by autophagy inhibition through Western blots [
30]. We demonstrated that blockade of autophagy increased the mRNA levels of PD-L1 as well as protein expression in gastric cancer cells. Accordantly, Yang et al. found that defective autophagy with deletion of
Atg5 by using a mouse model of cerulein-induced pancreatitis, activated the IκB kinase-related kinase TBK1 and promoted PD-L1 upregulation. These findings hinted at novel beneficial effects of autophagy inhibitors and their possible synergy with drugs targeting the PD-L1/PD-1 axis [
16].
We noticed that the basal protein levels of PD-L1 are higher in NCI-N87 and AGS cells than other gastric cancer cells as shown in Fig.
1a, which may attribute to the specific genomic mutations harbored by the cells with
SMAD4 and
TP53 mutations in NCI-N87 cells and
CDH1, CTNNB1, KRAS and
PIK3CA mutations in AGS cells. In this regard, PD-L1 expression was significantly higher in tumors with
TP53 mutation in lung cancer while
KRAS mutation could induce PD-L1 expression in lung adenocarcinoma [
31,
32]. Also, the oncogenic activation of the AKT-mTOR pathway could upregulate the expression of PD-L1 in non-small cell lung cancer [
33]. Both pharmacological agents and siRNA targeting non-lysosomal components of autophagy could up-regulate PD-L1 expression in gastric cancer cell lines, and the induction of IFN-γ further increased PD-L1 levels as shown in Figs.
2 and
3. 3-MA could effectively block an early stage of autophagy by inhibiting the class III PtdIns3K, but also non-selectively inhibit the class I PI3K and affect cell survival through AKT and other kinases which may in turn inhibit PD-L1 expression in particular settings. It is therefore likely that the overall upregulation of PD-L1 expression by autophagy inhibition is alleviated upon treatment by 3-MA compared to other autophagy inhibitors in AGS cells with
PIK3CA mutations (Fig.
2b). Recently, the precise mechanism of how CQ blocks autophagy was firmly demonstrated – CQ mainly inhibits autophagy by impairing autophagosome fusion with lysosome but not affecting the acidity of this organelle [
34]. The mutant p53 proteins was reported to counteract the formation of autophagic vesicles and the fusion with lysosomes via the repression of autophagy-related proteins and enzymes in pancreas and breast cancer cells [
35]. Concordantly, we found that CQ induced a lower increase on PD-L1 expression compared to 3-MA and bafilomycin A1 group in NCI-N87 cells with
TP53 mutation (Fig.
2b). Thus, in studies where the effect of autophagy inhibition is being investigated, it is important to confirm results by inhibiting autophagy at different stages with several pharmacological inhibitors. We found that IFN-γ significantly induced PD-L1 expression through activation of STAT1 signaling independent of autophagy levels in AGS and NCI-n87 cells (Additional file
1: Figure S5A), which is in accordance with others in several types of cancer [
36‐
38]. Upon autophagy inhibition, the levels of p-p65 was upregulated in AGS and NCI-N87 cells treated with or without IFN-γ (Additional file
1: Figure S5B). These results indicated that autophagy inhibition upregulated the levels of PD-L1 protein via NF-κB signaling whereas the IFN-γ induced PD-L1 expression through STAT1 signaling.
Clinical interventions to manipulate autophagy mainly by pharmacological inhibitors, including chloroquine and hydroxychloroquine, with other chemotherapeutics in search of synergistic interactions in cancer are already underway [
39]. Due to the lack of our understanding of the interplay between autophagy and the immune response, a study has sought to elucidate their relationship and demonstrated that the antitumor adaptive immunity is not adversely impaired by autophagy inhibition in immune-competent mouse models of melanoma and mammary cancer [
40]. Such findings are corroborated by our findings that autophagy inhibition had minimal effect on T cell function and PD-L1 levels were still upregulated in the cocultures of gastric cancer cells and PBMC (Fig.
5b). The increase in basal levels of PD-L1 expression in co-cultures makes the fold change reduced (Fig.
5b) compared to the gastric cancer cells alone group (Fig.
2a). The reduced fold change could be explained that the co-culture with lymphocytes itself already had an inducing effect on the expression of PD-L1 than cells without lymphocytes. In this respect, an inducing effect of lymphocytes on the expression of PD-L1 was found in cocultures with melanoma cells [
29]. In addition, the pharmacological inhibitors may also have some effects on the lymphocytes, which has to be evaluated in our future study. In our study, we did not observe the inducing effect of IFN-γ on the levels of PD-L1 in co-cultured condition (Fig.
5b). The cytokines including IFN-γ secreted by the lymphocytes may have inducing effect on the levels of PD-L1 in co-culture conditions, which therefore attenuated the effect of exogeneous IFN-γ added to the co-cultures [
41].
Our work suggests that autophagy inhibition plus anti-PD-L1 is an attractive combination for further investigation, particularly for tumors with high levels of autophagy, and provides potential biomarkers and mechanisms to assess clinical efficacy. However, contradictory evidence also exists in the literature. Peng et al. reported that the loss of PTEN decreased T cell infiltration in tumors, inhibited autophagy and was correlated with inferior outcomes with PD-1 inhibitor therapy [
42]. PD-L1 expression can be induced by inflammatory cytokines or tumor-cell intrinsic signaling, including NF-κB, MAPK, PI3K, mTOR and JAK/STAT. Pearson correlation analysis as shown in Additional file
1: Figure S6C and Fig.
6f suggests that the expression of PD-L1 in gastric cancer is in part correlated with high levels of LC3 and p62/SQSTM1. These findings indicate that additional factors must be considered to discern the various scenarios in which blockade of autophagy would be beneficial in cancer therapy.