TLR7 induces autophagy in non-small cell lung cancer tumor cells and influences anti-tumors responses in patients

Introduction

Autophagy, a critical self-degradative process essential for cellular energy balance, survival, differentiation, and development, particularly under conditions of nutrient stress, plays a fundamental role in maintaining cellular homeostasis. This process involves the lysosomal degradation of cellular components and is pivotal in clearing damaged proteins and organelles, thus acting as a defense mechanism against diseases, including cancer (1). Autophagy’s role in cancer is complex and paradoxical. Initially, it acts as a tumor suppressor, mitigating oxidative stress and inflammation. However, as tumors develop, autophagy supports tumor growth, proliferation, metastasis, and resistance to therapy (1,2). Autophagy can also modulate the anti-tumor immune response through various mechanisms, including inflammation inhibition and expression modulation of key molecules such as the major histocompatibility complex (MHC) and programmed death-ligand 1 (PD-L1) molecules (3-9).

The intricate regulation of autophagy at both transcriptional and post-translational levels involves a network of factors, including TFEB (transcription factor EB), FOXO (forkhead box), P53, and NF-κB (nuclear factor-kappa B), which respond to a variety of stresses such as metabolic, oxidative, and endoplasmic reticulum stresses (1). Moreover, external factors like pathogen infections can influence autophagy, either through direct interactions or by inducing cellular stress (10). Recent studies have shown that non-coding RNAs, including microRNAs (miRNAs) and long non-coding RNAs, also play a role in modulating autophagy (11-15). Despite the recognized influence of numerous factors, the full range of modulators affecting autophagy is not yet completely understood, leaving a substantial gap in our comprehension of how autophagy levels fluctuate within cancer cells, independently of nutritional or stress-related contexts.

In the context of non-small cell lung cancer (NSCLC), the most prevalent form of lung cancer, accounting for approximately 85% of all cases, autophagy plays a critical role in shaping tumor behavior (16). Indeed, autophagy has been shown to influence tumor progression by enabling cancer cells to adapt to metabolic stress, hypoxia, and nutrient deprivation, as well as to resist anticancer treatments, particularly platinum-based chemotherapy (16). These effects are tightly regulated by key molecular signaling pathways, including mTOR (mechanistic target of rapamycin) or AMPK (AMP-activated protein kinase). However, the impact of autophagy in NSCLC patients is still a subject of debate. Studies assessing autophagy levels in NSCLC patients have produced mixed results; some suggest a correlation with poorer patient survival, while others indicate a potential beneficial role (16-19). Building on these insights, our previous study, utilizing public transcriptomic databases, indicated that high levels of genes encoding autophagy in malignant cells correlate with reduced patient survival (20). We found that increased autophagy in these cancer cells not only enhances proliferation but also impairs the immune response, further promoting the expression of genes encoding immune checkpoints like PD-L1 and galectin 9 on tumor cells.

The current study takes a step further, revealing a novel pathway of autophagy modulation in tumor cells within the lung tumor microenvironment (TME). This modulation is driven by stimulation of toll-like receptor 7 (TLR7), a single-stranded RNA pattern recognition receptor. We revealed the presence of natural TLR7 ligands within the TME, secreted by living cancer cells. This activation of TLR7 triggers in vitro a complete autophagic flux in tumor cells. In patients with NSCLC, higher levels of autophagy, determined through light chain 3 (LC3) staining in tumor cells, were associated with poorer prognosis and low response to platinum-based chemotherapy. Intriguingly, our data also suggest that high autophagy levels might correlate with a better response to nivolumab, an anti-programmed death-1 (PD-1) immunotherapy. Furthermore, in vitro experiments revealed that the TLR7-dependent autophagy pathway contributes to chemoresistance and increased PD-L1 expression on cancer cells. We present this article in accordance with the MDAR reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-aw-1173/rc).

Methods

Weighted gene co-expression network analysis (WGCNA, RRID: SCR_003302) cluster analysis and genes expression correlation analyses in public cohorts of adenocarcinoma lung tumor patients

The co-expression network was constructed using The Cancer Genome Atlas (TCGA—https://www.cancer.gov/ccg/research/genome-sequencing/tcga) and Gene-Expression Omnibus (GEO—https://www.ncbi.nlm.nih.gov/geo/) databases. Lung adenocarcinoma (LUAD) gene expression data and associated clinical information were retrieved from TCGA and GEO under the following conditions: primary site “bronchus and lung”, program “TCGA”, disease type “adenomas and adenocarcinomas”, data category “transcriptome profiling”, data type “Gene Expression Quantification”, and workflow type “HTSeq-FPKM”. The TCGA-LUAD dataset included 59 non-tumoral and 517 tumoral samples. Additionally, RNA sequencing (RNA-seq) profiles of flow-sorted malignant, endothelial, immune, and stromal cells from resected primary NSCLC tumors were obtained from GEO (GSE111907—https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE111907).

The “WGCNA” R package was used to construct co-expression networks based on the top 20,000 genes. Data quality was assessed with the GoodSamplesGenes function. The adjacency matrix was transformed into a topological overlap matrix (TOM), and a dissimilarity matrix (dissTOM =1− TOM) was used for hierarchical clustering to identify gene modules, with a minimum module size of 40. Modules were defined using the DynamicTreeCut algorithm and merged if highly similar. Module eigengenes (MEs) and gene significance (GS) were calculated to identify modules correlated with autophagy-related genes from the Human Autophagy Database (HAD, 232 genes) and GeneCards (GC, 58 genes; score >10 for “autophagy”). The blue module, showing the highest absolute module significance (MS, defined as the average GS), was selected for further analysis. Module membership (MM) was calculated to assess gene connectivity within the module, and genes with |GS| >0.2 and |MM| >0.8 were defined as candidate hub genes. These were uploaded to the STRING database (https://string-db.org/, RRID: SCR_005223) to construct a protein-protein interaction (PPI) network (Figure S1), with the top 10 genes based on GS and node degree (>30) shown in Figure 1.

Figure 1 Identification of genes correlated with autophagy in lung tumor cells. (A) Strategy of our analysis. A public data cohort of 516 adenocarcinoma NSCLC patients from TCGA (TCGA-LUAD) was used, and a WGCNA analysis was applied to the cohort. Correlation studies were realized between the different module and the autophagy genes from the HAD (232 genes) and the autophagy GC (99 genes). (B) Assessment of Correlation Strength in Modules from HAD and GC: this evaluation ranges from weakly correlated modules, depicted in green, to strongly correlated modules, illustrated in red. (C) Representation of GS for autophagy relative to its MM in the blue module. (D) Association network of genes in the blue module with an MM >0.8, analyzed using STRING software. Nodes with a GS >0.5 or GS >0.6 are depicted in red and dark red, respectively. (E) Selection of the top 10 genes in the blue module with a node degree >30, ranked according to their GS score. (F) A public data cohort of 38 lung tumor patients (GSE111907), each with sorted cell types from the TME, was utilized. Only adenocarcinoma NSCLC patients were selected for analysis (n=22). Correlation studies between autophagy genes and TLRs genes expression were conducted for all cell types, including malignant, pan-immune, fibroblast, and endothelial cells. (G) Graph representing the level of TLRs expression in function of the cell type composing the TME. (H) Graph representing the level of correlation of autophagy genes with each TLRs genes expression in function of the cell type composing the TME. GC, GeneCard; GS, gene significance; HAD, Human Autophagy Database; MM, module membership; NSCLC, non-small cell lung cancer; TCGA-LUAD, The Cancer Genome Atlas-lung adenocarcinoma; TLR, toll-like receptor; TME, tumor microenvironment; WGCNA, weighted gene co-expression network analysis.

Culture of cell lines and transfection with green fluorescent protein (GFP)-LC3 and GFP-red fluorescent protein (RFP)-LC3

The human LUAD cell line A549 (ATCC CCL-185, RRID: CVCL_0023), lung squamous cell carcinoma (ATCC HTB-58, RRID: CVCL_0630) and the murine LUAD cell line LLC (ATCC CRL-1642, RRID: CVCL_0391) were cultured in DMEM/F-12 medium (Gibco, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Eurobio Scientific, Les Ulis, France), 1% non-essential amino acids, 1% HEPES buffer, 1% L-glutamine, and 1% sodium pyruvate (all from Gibco). Cells were maintained at 37 ℃ in a humidified incubator with 5% CO₂. The human lung squamous cell carcinoma cell line SK-MES (ATCC HTB-58) was maintained under similar conditions in EMEM/F-12 medium (Gibco). Human embryonic kidney cell lines HEK-Blue Null and HEK-Blue TLR7 (Invitrogen, RRID: CVCL_IM84) were cultured in DMEM with GlutaMAX (Gibco), supplemented with 10% FBS, 1% penicillin-streptomycin, and 100 µg/mL zeocin (InvivoGen). In addition, HEK-Blue TLR7 cells were cultured with 10 µg/mL blasticidin (InvivoGen) to maintain selective pressure for TLR7 expression. For GFP-LC3 expression, A549 cells were transduced with a lentiviral vector encoding GFP-LC3 (multiplicity of infection, MOI =1), kindly provided by Prof. Guido Kroemer’s laboratory. The infection was performed for 48 hours in FBS-supplemented medium, followed by cell sorting (BD Influx), clonal expansion, and selection. The presence of contamination was monitored biweekly for all cell lines. For transient expression, the RFP-GFP-LC3 tandem construct (ptfLC3, Addgene), also from Prof. Kroemer’s lab, was introduced into A549 cells 48 hours prior to treatment. Additionally, A549 knockout cell lines for BCN1 (beclin-1) and ATG5 (autophagy-related gene 5) were generated using CRISPR-Cas9 technology (Express Knockout Cell Pool, EditCo, USA).

In vitro stimulation or inhibition of TLR7

Cell lines were seeded in 12 or 24-well plates (100,000 or 50,000 cells per well, respectively) and cultured for 24 hours before being treated with 10% FBS medium containing either 1 mM CL264 (Invivogen) or 1 mM loxoribin (Loxo., Invivogen), both of which acting as TLR7 agonists and exhibiting similar effects on TLR7 activation. The duration of treatment with these agonists varied according to the specific requirements of each experiment. In contrast, for TLR7 inhibition, cells were pre-treated overnight with 2 µM ODN 20958 (Miltenyi Biotec).

Quantification of autophagic vesicles via confocal microscopy

Sterilized round coverslips were placed in the wells of a 24-well plate. Cell lines were seeded on these coverslips and incubated under various conditions for 6, 24, or 48 hours (50,000 cells per well). To visualize autophagosome in LLC and SK-MES cell lines, cells were incubated in phosphate buffered saline (PBS) containing 5% bovine serum albumin (BSA) and anti-LC3 antibody (cells signaling, 1/1,000, LC3A/B #4108, RRID: AB_2137703) for 1 hour. After washing with PBS, cells were fixed using BD Cytofix solution (BD Biosciences) for 30 minutes at 4 ℃, stained with DAPI (1/500 dilution, Thermo Scientific) for 7 minutes at 4 ℃ in darkness and mounted on slides using glycergel (Dako). Autophagosomes (indicated by LC3 or GFP-LC3 dots) were visualized by confocal microscopy (LSM 710) using ZEN software (ZEISS, RRID: SCR_013672). For each experimental condition, ten random images (encompassing 100–150 cells) were captured, and the number of autophagosomes per cell quantified using a custom script in R software. For co-localization experiments, GFP-LC3 A549 cells were cultivated as described above. Lysosomes were labeled with lysotracker deep red (50 nM, Invitrogen) for 2 hours before fixation. Co-localization and correlation analysis (using Pearson’s R coefficient) between green (GFP-LC3, autophagosomes) and red (Lysotracker, lysosomes) pixels were conducted using the “coloc2” script in ImageJ software (RRID: SCR_003070). This method facilitated detailed assessment of the interaction between autophagosomes and lysosomes. For RFP-GFP-LC3 transfected cells, autophagosomes (RFP⁺ GFP⁺) and autolysosomes (RFP⁺ GFP⁻) dots were visualized by confocal microscopy (LSM 710) using ZEN software (ZEISS, RRID: SCR_013672), as previously described.

Modulating autophagy in lung tumor cell lines using pharmacological agents

To investigate the role of autophagy in various biological processes, we employed rapamycin (Sigma), a well-established autophagy inducer, as a consistent positive control. Lung tumor cell lines were initially seeded into culture plates of appropriate sizes and allowed to adhere for 24 hours. Cells were then cultured in DMEM/F-12 medium supplemented with 10% FBS and treated with 1 µM rapamycin for 6, 24, or 48 hours, depending on the experimental setup. To further dissect the role of autophagy, particularly following TLR7 activation, we used a panel of autophagy inhibitors that target different stages of the autophagic process. For inhibition of autophagy initiation via class III phosphatidylinositol 3-kinase (PI3K), we treated cells with 3-methyladenine (3-MA, 10 mM, Sigma), wortmannin (100 nM, Sigma), or SAR405 (10 µM, MedChemExpress). To block autophagosome maturation and fusion with lysosomes, cells were treated with bafilomycin (Baf.) A1 (100 nM, Sigma) or a combination of leupeptin hemisulfate (1 µg/mL, Invitrogen) and E64D (1 µg/mL, Invitrogen). All inhibitors were applied for 6, 24, or 48 hours. Since 3-MA and wortmannin can also inhibit type I PI3Ks, we further validated the specific role of PI3K-III in autophagy by using selective type I PI3K inhibitors: capivasertib (8 nM, MedChemExpress) and pictilisib (3 nM, MedChemExpress).

Quantification of TLR7 ligands in various tumor supernatants

We employed a gene reporter-based assay to quantify TLR7 ligands in tumor supernatants. HEK-blue null and HEK-blue hTLR7 cells (Invitrogen) were exposed to supernatants from A549 cells (non-treated, treated with chemotherapy and or specific cell death inducers), and supernatants derived from fresh NSCLC patient tumors. Resected tumors were cultured in RPMI medium supplemented with 10% FBS, and supernatants were collected and preserved at −80 ℃ after a 24-hour incubation. HEK-blue cells were treated with the supernatants, and after for 24 hours, a mixture of 20 µL of cell culture supernatant and 180 µL of QUANTI-Blue solution (Invivogen, hb-det2), which contains an alkaline phosphatase substrate, was incubated for 1.5 hours at 37 ℃. The optical density (OD) values, indicative of TLR7 ligand concentration, were subsequently measured at 620 nm using the Infinite 200Pro spectrophotometer.

Evaluation of cell death induction by chemotherapeutic agents and specific cell death inducers using flow cytometry

Our methodology involved assessing the efficacy of various chemotherapies and cell death inducers in the induction of cells across different time points (6, 24, 48, and 72 hours). An IC50 dose, the concentration causing 50% cell mortality, was established for each drug at 24 hours. We investigated the impact of chemotherapy-induced cell death using cisplatin (Cis., 50 µM) (sourced from Cochin Hospital), taxol (10 µM) (Cochin Hospital), and oxaliplatin (Oxali., 200 µM) (Sigma) for durations of 24 or 48 hours, contingent on experimental requirements. Additionally, to probe the effects of specific cell death types like ferroptosis, pyroptosis, and necroptosis, we utilized respective inducers: erastin (Era., 15 µM) (Sigma), polyphillin VI (Pol VI., 8 µM) (Selleckchem), and FTY720 (10 µM) (Sigma) for either 24 or 48 hours. Post a 24-hour culture period, cells were subjected to varying treatments as per experimental design: anti-cancer chemotherapy agents (Cis., Oxali.) or specific inducers of alternate cell death pathways (Era., Pol VI., and FTY720), for 6, 12, 24, or 48 hours. After treatment, cells were washed with PBS and stained with live dead (1/100), annexin V (1/200), and propidium iodide (PI) (1/100) (all from Invitrogen) for 30 minutes at 4 ℃ in the dark. Following another PBS wash, cells were fixed using BD Cytofix (BD Biosciences) for 30 minutes at 4 ℃ in darkness. Flow cytometry analysis (BD, LSR Fortessa) was then conducted to evaluate different cell markers. Data acquisition and analysis were performed using Diva (BD, Biosciences) and Kaluza (Beckman Coulter) software, respectively. Based on their labeling, cells were categorized as live (live dead⁻) or dead (live dead⁺). Further, the type of death induced by each treatment on the lung tumor cells was determined: apoptosis (Annexin V⁺/PI⁻) or non-apoptotic death (Annexin V⁻/PI⁺).

Assessment of immunogenic molecule expression by flow cytometry

Cell lines were cultured in 12-well plates for 24 hours, followed by the application of various treatments, including synthetic TLR7 agonists and chemotherapies, either alone or in combination, contingent on the experimental design. Treatments were conducted over 24 or 48 hours. Subsequently, cells were processed with a live/dead kit (1 µL/million cells) (Invitrogen) and Fc receptor blocker (Fc block, 1/100) (BD Biosciences) for 30 minutes at 4 ℃. This step was followed by the application of specific antibodies (1/100) to expose the membrane antigens of interest for 45 minutes. After antibody incubation, cells were washed with PBS and subjected to fixation and permeabilization using a fixation/permeabilization solution (BD Cytofix, BD Biosciences) for 30 minutes at 4 ℃. The cells were then incubated with various intracellular targeting antibodies, diluted in WashBuffer solution (1/100) (Invitrogen), for another 45 minutes. For each experimental run and time point, control wells containing untreated cells were prepared to incubate with all isotype controls corresponding to each antibody utilized. Post-incubation, cells were processed through a flow cytometer (BD, LSR Fortessa), where the intensity of the fluorochromes attached to the antibodies was quantitatively measured. Data acquisition and subsequent analysis were carried out using Diva (BD, Biosciences) and Kaluza (Beckman Coulter) software, respectively.

Analysis of retrospective NSCLC patient cohorts

This study utilizes three distinct retrospective cohorts of NSCLC patients, all sourced from the thoracic surgery department of Cochin Hospital. The first cohort consists of NSCLC patients who did not receive chemotherapy treatment (detailed in Table 1) (summary: demographics: Caucasian; age 62 years median; gender: 72% male and 60% female) and the patients were enrolled between 2001 and 2005. The second cohort includes NSCLC patients who underwent various combinations of neoadjuvant chemotherapy regimens (outlined in Table 2) (summary: demographics: Caucasian; age 60 years median; gender: 79% male and 23% female) and patients were enrolled between 2000 and 2007. The final retrospective cohort encompasses NSCLC patients prior to receiving PD-1 anti-immune checkpoint immunotherapy (presented in Table 3) (summary: demographics: Caucasian; age 64 years median; gender: 70% male and 30% female) and the patients was enrolled between 2010 and 2017. Clinical data for the NSCLC patients comprising these cohorts were meticulously gathered by clinicians. The clinical parameters characterizing each of these three NSCLC patient cohorts are concisely summarized in respective tables, providing a comprehensive overview of the patient profiles and treatment modalities for each group. To reduce bias, investigators were blinded during the analyses.

Immunohistochemical (IHC) staining and analysis of LC3 autophagy protein in NSCLC patient samples using halo software

For the IHC examination of LC3 autophagy protein in NSCLC patient tissue samples, slides prepared from patient blocks underwent a series of processing steps. Initially, they were incubated at 37 ℃ for deparaffinization in sequential clearing baths (Leica microsystems), followed by graded ethanol baths (100%, 90%, 70%, and 50%) (VWR). Antigen retrieval was performed using PT-link (Dako) in a heated alkaline solution (PT-link high). Tissue sections, delineated using the Dako Pen (Dako), were then processed in the Autostainer plus (Dako) to ensure uniform staining conditions across all slides. This automated system facilitated the initial incubation with H2O2 (Gilbert) to quench endogenous peroxidases, followed by blocking of nonspecific binding sites using a protein block solution (Dako). Subsequently, slides were manually labeled with the anti-LC3 antibody (Sigma, HPA053767, RRID: AB_2682253), diluted appropriately in IHC diluent (Enzo), and incubated at 4 ℃ overnight. Post three TTBS washes, slides were re-introduced into the automated system for secondary labeling. Here, they were incubated with an HRP-conjugated anti-rabbit secondary polymer antibody (Polyview plus HRP, ENZ-ACC 103-0000) for 30 minutes. LC3 detection was achieved via a 10-minute incubation with a DAB solution (Dako). Nuclei were counterstained with hematoxylin (Dako), followed by final TTBS washes. The prepared slides were then mounted with glycergel (Dako) and stored at 4 ℃ until scanned at 40x magnification using the nanozoomer (Hamamatsu). The specificity of staining was confirmed through negative controls using an isotype control (Negative control, Rabbit, Agilent, IR600), which showed no labeling. The scanned slides were analyzed using Halo software (IndicaLabs), with assistance from anapathologists. This software enabled the delineation of the tumor area within each sample, excluding stroma, necrotic regions, and voids from analysis. LC3 labeling intensity within tumor cells was assessed, categorizing them into LC3 negative (weak or no staining) and LC3 positive (strong staining) groups. For each patient in the cohorts, the proportion of LC3 negative versus LC3 positive tumor cells was determined. Based on the median proportion of these categories, patients were classified into LC3 low or LC3 high groups according to the median (n=108 per group). Additionally, an autophagy score was assigned to each patient (score 1: <25% of LC3 positive tumor cells; score 2: 25%< x <50%; score 3: 50%< x <75% and score 4: >75%), further quantifying the level of autophagy in the tumor samples.

Analysis of the relationship between LC3 and TLR7 or PD-L1 expression in NSCLC patient cohorts

This study focuses on the correlation between LC3 and the expression of TLR7 or PD-L1 in NSCLC patients. For the first two cohorts, which include untreated patients and those receiving neoadjuvant chemotherapy, TLR7 expression in tumor samples was previously quantified through IHC labeling (doi: 10.1158/0008-5472.CAN-13-2698) (21). These samples were scanned at 40x magnification using the nanozoomer (Hamamatsu), and TLR7 expression was specifically analyzed in tumor cells. Based on the optimal cut-off of 78%, as established in prior research (doi: 10.1158/0008-5472.CAN-13-2698) (21), patients were classified into two groups: TLR7 low (less than 78% TLR7 expression) and TLR7 high (greater than or equal to 78% TLR7 expression). In parallel, for the cohort of NSCLC patients treated with anti-immune checkpoint immunotherapy (cohort 3), clinicians had previously quantified PD-L1 expression in lung tumor cells (doi: 10.1016/j.lungcan.2022.05.001) (22). Patients in this cohort were categorized based on the PD-L1 expression in their tumor cells, using median. Those with less than 40% (median) of lung tumor cells expressing PD-L1 were defined as PD-L1 low, while patients with more than 40% PD-L1 expression were classified as PD-L1 high. Following these classifications, detailed correlation analyses were conducted to explore the relationship between LC3 and PD-L1 or TLR7 expression levels across individuals in each NSCLC cohort.

Statistical tests analyses

R software (v4.0.3) was used for all bioinformatic statistical analyses, and PRISM software (RRID: SCR_002798) was employed for the analyses of in vitro and ex vivo experiments. The Wilcoxon test was used to compare the differences between the two groups. The Kruskal-Wallis test was utilized to compare the differences between three groups and above. Values P<0.05 were considered statistically significant (*). The survival time of the patient was evaluated by Kaplan-Meier survival analysis, and the different groups were compared by utilizing a log-rank test. Univariate and multivariate Cox regression analysis was used to investigate the independent prognostic factor, employing the “survival” R package. Survival curves were performed utilizing R package “survminer”. We employed “GOplot” R package to visualize the functional annotation enrichment analyses. Data visualization was performed using R package “ggplot2”. The R packages utilized in this study could be obtained from “bioconduction”.

Ethical consideration

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the regional ethics committee of “Comité de Protection des Personnes Île-de-France II” (Nos. 2007-A00845-48 and 2012-06-12) and individual consent for this retrospective analysis was waived.

Results

Identification of TLR7 as a potential regulator of autophagy in LUAD malignant cells

To investigate the molecular pathways regulating autophagy in tumor cells, we performed a WGCNA (23). This approach provided a comprehensive correlation between global gene expression and autophagy-related genes in 516 NSCLC patients from the TCGA-LUAD dataset (Figure 1A). Using average linkage and Pearson’s correlation methods, we clustered these patients and confirmed the absence of outlier samples. To construct a scale-free network, we selected a soft-thresholding power of β =6 (scale-free R² =0.9). Hierarchical clustering based on average linkage revealed 39 distinct gene modules, with the blue module (MEblue) exhibiting the strongest association with autophagy-related gene expression, as defined by GC and the HAD (Figure 1B). Within this module, we identified 131 high-connectivity hub genes (MM >0.8, Figure 1C). We then conducted a PPI network analysis using the STRING database to further investigate these 131 hub genes. Applying a node degree threshold of >30, we identified 35 key hub nodes (Figure 1D, genes in red), among which 18 exhibited significant GS for autophagy (GS >0.5). Notably, 4 of the top 10 hub nodes with the highest GS scores were linked to toll-like receptor (TLR) signaling pathway genes—TLR7, TLR4, IRF8, and TLR8 (Figure 1E)—suggesting a potential role for TLR pathways in autophagy regulation in lung cancer. Interestingly, correlation analysis of individual TLR gene members revealed that TLR7 (r=0.61), TLR4 (r=0.58), and TLR8 (r=0.56) exhibited the strongest correlations with autophagy gene expression, highlighting their potential involvement in modulating autophagy in NSCLC.

To advance our understanding of TLRs in modulating autophagy, particularly in malignant cells, we employed the GSE-111907 public cohort, encompassing 22 NSCLC adenocarcinoma patients (24). In this study, malignant cells [epithelial cell adhesion molecule (EPCAM)⁺ CD45⁻ CD31⁻], pan-immune cells (EPCAM⁻ CD45⁺), endothelial cells (CD31⁺ CD45⁻ EPCAM⁻), and fibroblasts (CD10⁺ CD45⁻ EPCAM⁻ CD31⁻) were sorted cells from the TME to perform a bulk RNA-seq per cell type. We performed a correlation analysis between the expression of autophagy genes and TLR-encoding genes across these cell types (Figure 1F). We observed that TLR gene expression varied substantially across different cell types (Figure 1G). Intriguingly, TLR7, although not the most prominent gene encoding TLR in any TME cell type, exhibited the strongest correlation with autophagy genes in both pan-immune and malignant cells (Figure 1H). This observation highlights TLR7’s potential as a key modulator of autophagy in malignant environments. Coupled with our previous findings that indicate TLR7’s expression on lung tumor cells correlates with poor NSCLC prognosis (21,25,26), these results propel further investigation into TLR7’s influence on autophagy modulation and its implications for patient outcomes.

TLR7 activation in lung tumor cells initiates a complete autophagy process

To elucidate the relationship between TLR7 and autophagy in lung tumor cells, A549 GFP-LC3 cells were treated with synthetic TLR7 agonists, CL264 and Loxo.. Autophagic vacuole quantification was conducted at 6, 24, and 48 hours post-treatment. Notably, at 6 hours, a significant increase in LC3-positive autophagic vacuoles was observed in treated cells compared to untreated controls, reflecting the response seen with the autophagy inducer rapamycin (1 µM) (Figure 2A). In line with the pattern observed during Rapamycin treatment, TLR7-induced autophagosome appeared to be a transient activation, peaking at 6 hours and then gradually decreasing, yet remaining elevated compared to untreated cells at 24 hours (Figure 2B). Comparable results were also observed in other human (SK-MES) and murine (LLC) lung tumor cell lines (Figure S1A,S1B).

Figure 2 Study the impact of TLR7 on the autophagy level in lung tumor cells. (A) Example of images of A549 stably transfected with the GFP-LC3 construct (A549 GFP-LC3) treated or not (NT) with an autophagy inducer, the rapamycin (Rapa., 1 µM) or by synthetic TLR7 agonists, CL264 (1 mM) and loxoribin (Loxo., 1 mM) for 6h. The graph represents the quantification of the GFP-LC3 dots per cell in each condition at 6 h post-treatment. The experiment was performed six times. (B) The same experiment as in A was performed with different kinetics: 6, 24, and 48h post-treatments. The graph represents the evolution of the GFP-LC3 dots per cell in each condition over time. The experiment was performed six times. (C) Example of images of A549 GFP-LC3 treated or not (NT) with either inhibitors of the autophagy maturation, Leupeptin and E64d (L + E, 1 ug/mL each), an autophagy inducer (Rapa., 1 µM) or synthetic TLR7 agonists (CL264 and Loxo., 1 mM) for 24 h. The graph represents the quantification of the GFP-LC3 dots per cell in each condition. The experiment was performed three times. (D) Example of images of A549 GFP-LC3 treated or not (NT) with either an inhibitor of the autophagy maturation, the Baf. (100 nM), an autophagy inducer (Rapa., 1 µM) or synthetic TLR7 agonists (Loxo., 1 mM) for 24 h. Lysosomes were marked in red by the addition of lysotracker 90 minutes before the end of the experiment. The graph represents the study of pixel colocalization (green versus red puncta) using the coloc2 script from the ImageJ software. The experiment was performed three times. (E) A549 cells were transfected with a plasmid encoding a dual-labeled LC3 probe, RFP-GFP-LC3. Cells were again treated with Rapa. or with Loxo. (1 mM) for 24 h. Representative images are shown and the number of total autophagic vacuoles (GFP⁺ RFP⁺ + GFP⁻ RFP⁺), autophagosomes (GFP⁺ RFP⁺), and autolysosomes (GFP⁻ RFP⁺) were enumerate. The inset permits visualization of yellow autophagosome (yellow arrow) and red autolysosomes (red arrow). The experiment was performed three times. (F) At the left: measurement of the proportion of lung tumor cells LC3 positive versus negative for both TLR7 low (proportion of lung tumor cells expressing TLR7 <78%) versus TLR7 high (proportion of lung tumor cells expressing TLR7 >78%) NSCLC patients of the cohort 1. In the right: implementation of an autophagy score (score 1, proportion of lung tumor cells LC3 pos <25%; 2, 25%< x <50%; 3, 50%< x <75% and 4, x >75%). Evaluation of the TLR7 expression level in function of the autophagy score. (G) The same analysis as in E was carried out on the NSCLC patients from cohort 2, those treated by neoadjuvant chemotherapies. Student’s t-test, *, P<0.05. Baf., bafilomycin; GFP, green fluorescent protein; LC3, light chain 3; Loxo., loxoribin; NSCLC, non-small cell lung cancer; Rapa., rapamycin; RFP, red fluorescent protein; TLR7, toll-like receptor 7.

To assess autophagy flux following TLR7 activation, we analyzed autophagy in A549 GFP-LC3 transfected cells in the presence of the lysosomal inhibitors leupeptin and E64D (Figure 2C). As expected, inhibition of autophagosome-lysosome fusion led to an increased number of LC3 puncta in both control cells (reflecting basal autophagy flux) and rapamycin-treated cells (induced autophagy). Similarly, leupeptin and E64D enhanced autophagosome accumulation in TLR7-stimulated cells, supporting the conclusion that TLR7 activation induces de novo autophagosome formation (Figure 2C). To confirm this finding, we directly assessed colocalization of LC3 and the lysosomal (Figure 2D). Employing rapamycin (1 µM) as a positive control for autophagy flux and Baf. (100 nM) to inhibit autophagy maturation, we analyzed pixel colocalization via ImageJ’s coloc2 script, generating Pearson coefficients (Figure 2D). The results indicated a colocalization pattern in TLR7-agonist treated cells comparable to rapamycin, contrasting with the Baf. profile, suggesting a complete autophagy flux initiation under TLR7 stimulation. As a final measure of autophagic flux, we transfected A549 cells with a construct encoding an RFP-GFP-LC3 polyprotein. In this way, it is possible to distinguish autophagosomes (RFP⁺ GFP⁺ puncta) and autolysosomes (RFP⁺ GFP GFP⁺ vesicles), with the latter being GFP negative as a result of the quenching of signal by the acidic microenvironment of the lysosome. TLR7 stimulation increased the number of both autophagosomes and autolysosomes compared with untreated cells (Figure 2E). Moreover, an equivalent ratio of autophagosomes and autolysosomes was measured in TLR7-treated cells and cells treated with rapamycin (Figure 2E). Altogether, the data presented in Figures 1,2 provide evidence for TLR7 stimulation inducing de novo autophagosome formation.

To validate the link between TLR7 expression and autophagy in NSCLC patients, we analyzed two retrospective cohorts: one untreated prior to surgery (cohort 1) and another treated with neoadjuvant chemotherapy (cohort 2). Clinical details for each cohort are summarized in Tables 1,2 respectively. TLR7 expression by tumor cells was already quantified (21) and LC3 expression on malignant cells was measured specifically in the tumor area using the Halo software. NSCLC patients were then divided into TLR7 high (>78% expression on tumor cells) versus TLR7 low (<78% expression on tumor cells), as previously published (21). In both cohorts, patients with high TLR7 expression consistently exhibited a greater number of LC3-positive tumor cells compared to those with low TLR7 expression (Figure 2F,2G—left part). This trend was coupled with a reduced presence of LC3-negative cells in the TLR7-high group, underscoring a notable association between TLR7 expression and increased autophagy levels in tumor cells. Based on the percentage of LC3-positive tumor cells within tumor areas, an autophagy score from 1 to 4 for each NSCLC patient of the cohorts was given (1: <25% of LC3 positive cells, 2: 25%< x <50%, 3: 50%< x <75% and 4: >75%). Interestingly, in both NSCLC patient cohorts, a consistent correlation was observed between TLR7 expression and the autophagy score (Figure 2F,2G—right parts), reinforcing the link between TLR7 expression and autophagy levels in NSCLC under various conditions.

Notably, our analysis revealed no significant correlation between LC3 expression and various patient characteristics, including age, size, tumor stage, tumor phenotype, or chronic obstructive pulmonary disease (COPD) status (Figure S1C-S1G), indicating that LC3 expression in NSCLC patients is independent of these clinical parameters. Similarly, our findings indicate no discernible relationship between immune cell infiltration and LC3 expression in tumor cells (Figure S1H), suggesting that autophagy levels within tumors do not evidently affect immune infiltration.

Physiological TLR7 ligands in NSCLC TME trigger autophagy in cancer cells

Building on our previous findings of a correlation between TLR7 and LC3 expression in NSCLC patients, we wanted to know if TLR7 could physiologically activate autophagy in tumor cells. Our approach involved investigating whether diffused molecules from the NSCLC TME had the capacity to induce autophagy. Fresh tumors or adjacent tissue from NSCLC patients were incubated for 24 hours to facilitate the diffusion of TME molecules into the medium, forming tumor and control supernatants. A549 GFP-LC3 cells were then incubated with supernatants for 6, 24, or 48h, and the average number of autophagosomes per cell was measured (Figure 3A). In comparison with the control conditions (RPMI or control supernatant), we overserved that lung tumor cells incubated with the tumor supernatant exhibited a significantly higher count of LC3 dots at 6 hours post-incubation, suggesting that diffused molecules within the TME are conducive to the induction of autophagy. Furthermore, the profile of autophagy levels in lung tumor cells exposed to tumor supernatant, while not as elevated as those treated with a synthetic TLR7 agonist, displayed a similar pattern (Figure 3A and Figure 2B). These observations imply that in a pathophysiological context of lung tumors, there are diffusible molecules in the TME that contribute to heightened autophagy in lung tumor cells. Notably, when lung tumor cells pre-treated with a TLR7 antagonist (ODN 20958, 2 µM) were incubated with tumor supernatant, the increase in autophagy levels was significantly diminished compared to cells without TLR7 blockage (Figure 3B). This result indicates that the enhanced autophagy seen with tumor supernatant incubation is, at least partially, attributed to the action of TLR7 ligands interacting with their receptors.

Figure 3 Measurement of the presence of TLR7 ligands in the TME and its impact on autophagy activation in lung tumor cells. (A) At the left: examples of images of A549 GFP-LC3 cells incubated for 6h with either an RPMI medium alone, an adjacent tissue supernatant (control supernatant), or a tumor supernatant (tumor supernatant). Medium: measurement of the LC3-GFP dot per cell in A549 GFP-LC3 in function of the medium in which the cells are incubated for 6 h. On the right: kinetics of LC3 GFP-LC3 dots per cell at 6, 24, and 48 h post-incubation in function of the medium, the cells are incubated. The experiment was performed three times. (B) Examples of images of A549 GFP-LC3 cells treated with either an RPMI medium or a tumor supernatant under two different conditions: an untreated condition (control) or a condition where cells were preincubated with a specific TLR7 antagonist (ODN 20958, 2 µM) overnight before and at the moment of incubation of the cells with the supernatant, for 6 h. Plot representing the average number of autophagosomes in cells pre-incubated or not overnight with the TLR7 antagonist. The experiment was performed three times. (C) Measurement of the presence of TLR7 ligands through the measurement of the Sepa activity of HEK-blue cells, following the treatment or not (NT) with different doses of TLR7 agonists (Loxo., 1 nM, 10 nM, 100 nM, and 1 mM), or after the incubation with a culture media alone (control media), with an adjacent tissue supernatant (control supernatant) or with a tumor supernatant (tumor supernatant). The experiment was performed three times in triplicates. (D) Graph describing the GFP dot per cell in A549 GFP-LC3 as well as their concentration in TLR7 ligands after incubation with tumors supernatant from different NSCLC patients (P1 to P10). Tumor supernatants were classified based on their richness in TLR7 ligands: highly concentrated in green, medium in orange, and less concentrated in yellow. The experiment was performed three times. (E) Graph depicting the median fold increase in autophagy induction relative to the concentration of TLR7 ligands in the tumor supernatant of each patient. This experiment was replicated three times, each in triplicate. (F) Measurement of the TLR7 ligands abundance following the treatment or not (NT) with different doses of TLR7 agonists (Loxo., 1 nM, 10 nM, 100 nM, and 1 mM), or after the incubation with a culture medium alone (control media), or with supernatant from A549 in culture for 24 h [A549 NT supernatant (24 h)] or 48 h [A549 NT supernatant (48 h)]. The experiment was performed three times in triplicates. Student’s t-test, *, P<0.05. GFP, green fluorescent protein; HEK, human embryonic kidney; LC3, light chain 3; Loxo., loxoribin; NSCLC, non-small cell lung cancer; TLR7, toll-like receptor 7; TME, tumor microenvironment.

To substantiate the presence of TLR7 ligands in tumor supernatants, we employed HEK-blue TLR7 reporter cells, measuring their secreted alkaline phosphatase (SEAP) activity as an indicator of TLR7 stimulation. Using HEK-blue Null cells as a negative control, we first validated our assay with varying concentrations of synthetic TLR7 ligands (Figure 3C). We then assessed TLR7 ligand presence in NSCLC tumor supernatants by incubating reporter cells with supernatant from tumor or adjacent tissues for 6 hours. The HEK-blue TLR7 cells exposed to NSCLC tumor supernatants exhibited notable SEAP activity as compared to HEK-Null cells, akin to the response seen with a 10 nM concentration of a synthetic TLR7 agonist (Figure 3C). Notably, we observed variability in the capacity of tumor supernatants from different NSCLC patients to induce autophagy in A549 GFP-LC3 cells (Figure 3D). Concurrently, HEK-blue cell assays demonstrated that this autophagy-inducing ability of the supernatants directly correlated with the concentration of TLR7 ligands present (Figure 3D,3E). Taken together, these findings confirm the presence of active TLR7 ligands in NSCLC tumor supernatants capable to stimulate autophagy in lung tumor cells.

To investigate the source of TLR7 ligands in the TME, we examined the ability of tumor cells themselves to secrete these ligands. Inspection of supernatants from A549 cell line, as well as other lung tumor cells, revealed that tumor cells were capable to secrete TLR7 ligands (Figure 3F, Figure S2A,S2B). To investigate TLR7 ligand release mechanisms, we explored if cell death could liberate RNA recognizable by TLR7. Oxali. was used to induce apoptosis in A549 cells, confirmed by Annexin V/PI staining (Figure S2C). Despite significant apoptosis, no increased SEAP activity in HEK-blue TLR7 cells was noted with Oxali.-treated A549 supernatants (Figure S2D). This contrasted with untreated A549 supernatants, which did enhance SEAP activity. While apoptosis in lung tumor cells doesn’t seem to release TLR7 ligands, we explored the effect of other cell death types like necroptosis, ferroptosis, and pyroptosis, known for immunogenic cell death. Lung tumor cells were treated with specific inducers for each death type (respectively FTY720, Era. and Pol VI.) and cell death types were confirmed (Figure S2E). However, no significant SEAP activity increase in HEK-blue TLR7 cells was observed with these treatments (Figure S2F), suggesting these cell death types might not be key in TLR7 ligand release.

Collectively, these findings indicate that physiological TLR7 ligands, secreted by living tumor cells in the TME, play an important role in modulating autophagy in lung malignant cells through TLR7 stimulation.

Elevated autophagy in lung tumor cells is linked to poorer prognosis in NSCLC patients

In light of autophagy’s relevance in cancer progression, we examined the impact of autophagy level on clinical outcome of NSCLC patients (cohort 1). We measured the ratio of LC3 positive and negative cells in tumors, categorizing patients based on this ratio into high and low LC3 groups according to the median of LC3 positive tumor cells (Figure 4A). Our survival analysis revealed that patients with higher LC3 levels had poorer survival outcomes and high cumulative hazard ratio (Figure 4B). Moreover, the survival outcomes of NSCLC patients were found to be significantly influenced by their autophagy scores (elaborated as previously explained). Patients with higher autophagy scores [3–4] exhibited worse overall survival compared to those with lower scores [1–2] (Figure S3A). Our univariate and multivariate analyses corroborated that the level of autophagy is an independent prognostic factor in NSCLC patient, including the T CD8 infiltration (Figure 4C). While TLR7 expression alone did not significantly impact survival outcome, we found that patients with high expression of both TLR7 and LC3 (classified as TLR7 high/LC3 high) exhibited the worst overall survival and highest cumulative hazard ratio. Conversely, those with low TLR7 and LC3 expression (TLR7 low/LC3 low) showed better overall survival (Figure 4D).

Figure 4 Influence of autophagy level in lung tumor cells on the survival outcomes of NSCLC patients. (A) Example of two typical LC3 autophagy protein staining patterns in lung tumor cells obtained from an LC3 high NSCLC patient (left) and an LC3 low NSCLC patient (right). Based on the median percentage of LC3 positive versus LC3 negative lung tumor cells in their tumor, each NSCLC patient was distributed either into LC3 high or LC3 low NSCLC patients, as depicted in the graph. (B) Left: Kaplan Meier curve illustrating the survival probability of patients over time, categorized based on their status as either LC3 low or LC3 high. Right: Kaplan Meier describing the cumulative hazard over the time based on the autophagy level of NSCLC patients. (C) Univariable and multivariable analyses of different biological and clinical parameters on survival of NSCLC patients. (D) Left: Kaplan Meier curve illustrating the survival probability of patients over time, categorized based on their status as either LC3 and TLR7. Right: Kaplan Meier describing the cumulative hazard over the time based on the autophagy level and TLR7 expression of NSCLC patients. Student’s t-test, *, P<0.05. ADK, adenocarcinoma; CI, confidence interval; LC3, light chain 3; NSCLC, non-small cell lung cancer; SCC, squamous cell carcinoma; TLR7, toll-like receptor 7.

The findings indicate a strong association between high autophagy levels and worse prognosis in NSCLC patients. Furthermore, they emphasize the possible significance of TLR7-mediated autophagy in influencing the survival outcomes of these patients.

Higher autophagy in lung tumor cells is linked to chemotherapy resistance and poorer response in NSCLC patients

In the second cohort of NSCLC patients treated with neoadjuvant chemotherapy, an analysis of LC3 levels revealed that non-responders had significantly more LC3-positive cells in tumor areas compared to responders (Figure 5A). Additionally, non-responders had fewer LC3-negative cells. The autophagy score was closely linked with response rates: 50%, 60%, 75%, and 95% of non-responders were observed in autophagy scores 1, 2, 3, and 4, respectively (Figure 5B). Notably, a higher percentage of non-responders (56%) had higher autophagy scores (score 4) compared to responders (24%). Separate analyses of different chemotherapy combinations supported these findings (Figure S3B-S3D). Univariate and multivariate analyses confirmed that higher autophagy levels, whether independently or in conjunction with high TLR7 expression, was associated with a good prognosis of patients undergoing neoadjuvant chemotherapy (Figure 5C), further underlining autophagy’s role in chemotherapy resistance.

Figure 5 Impact of autophagy level in lung tumor cells on the resistance to chemotherapies in NSCLC patients. (A) Left: graph representing the percentage of LC3 positive versus LC3 negative lung tumor cells in NSCLC patients treated with various neoadjuvant chemotherapies in function of their responsiveness to these treatments. (B) Left: distribution of the proportion of NSCLC patients who respond (yes) versus do not respond to chemotherapies (no) in function of their autophagy score. Right: proportion of the different autophagy scores between responders (yes) versus non-responders to chemotherapies (no). (C) Univariable and multivariable analyses of different biological and clinical parameters on the survival of NSCLC patients treated with neoadjuvant chemotherapy. (D) Graph representing the mortality rate of A549 cells treated or not (NT) with different chemotherapies: Cis. (50 µM), Oxali. (200 µM), and taxol (10 µM) for 48 hours. In addition to treatment with chemotherapies, A549 were treated or not with a synthetic TLR7 agonist (Loxo., 1 mM) alone or with different autophagy inhibitors, 3-MA (10 mM), SAR-405 (10 µM) or Baf. (100 nM). The experiment was performed at least three times in triplicates. (E) Graph representing the mortality rate of A549 cells deficient for beclin-1 (BECN KO) or ATG5 (ATG5 KO), treated or not (NT) with different chemotherapies: Cis. (50 µM), Oxali. (200 µM), and taxol (10 µM) for 48 hours. The experiment was performed at three times in triplicates. (F) Top: graph representing the average number of autophagosomes per A549 GFP-LC3 cell in function of whether the cells were treated (NT) with the different chemotherapies (Cis., Oxali., or taxol) combined or not with a synthetic TLR7 agonist, Loxo. (1 mM) for 6 hours. The experiment was performed three times. Bottom: A549 GFP-LC3 were treated or not (NT) with either an inhibitor of the autophagy maturation, the Baf. (100 nM), an autophagy inducer (Rapa., 1 µM) or with the different chemotherapies (Cis., Oxali., or taxol) combined or not with a synthetic TLR7 agonist, Loxo. (1 mM) for 6 hours. Lysosomes were marked in red by the addition of lysotracker 90 minutes before the end of the experiment. The graph represents the study of pixel colocalization (green versus red puncta) using the coloc2 script from the ImageJ software. The experiment was performed three times. Student’s t-test, *, P<0.05; NS, not significant. 3-MA, 3-methyladenine; ADK, adenocarcinoma; Baf., bafilomycin; CI, confidence interval; Cis., cisplatin; LC3, light chain 3; Loxo., loxoribin; NSCLC, non-small cell lung cancer; Oxali., oxaliplatin; Rapa., rapamycin; SCC, squamous cell carcinoma; TLR7, tolllike receptor 7.

To understand TLR7-mediated autophagy’s effect on chemoresistance, A549 lung tumor cells were exposed to various chemotherapies. These included DNA-targeting agents Cis. and Oxali., and taxol, known for inhibiting microtubule depolymerization. Through experiments with different concentrations and time points, the IC50 dose for each chemotherapy was established at 24 hours post-treatment: 50 µM for Cis., 200 µM for Oxali., and 10 µM for taxol. A549 cells were then subjected to different chemotherapies while being simultaneously stimulated with the TLR7 synthetic agonist Loxo. (1 mM) (Figure 5D). TLR7 stimulation in cells treated with Cis. or Oxali. resulted in a decreased number of cell death as compared to unstimulated treated cells, indicating a TLR7-induced chemoresistance as supported by our previous report (21). Interestingly, this effect was not seen in cells treated with taxol. To investigate the role of autophagy in this chemoresistance, various autophagy inhibitors were introduced alongside the chemotherapy and Loxo. treatments. These included 3-MA and SAR405, which inhibit the initiation of autophagy flux, and Baf., which delays autophagy by preventing autophagosome-lysosome fusion. The addition of these inhibitors significantly increased the mortality of chemotherapy-treated cells, suggesting an important effect of autophagy in chemoresistance. Importantly, inhibition of autophagy prevented the TLR7-stimulated decrease in cell mortality in both Cis.- and Oxali.-treated cells (Figure 5D). Considering that 3-MA and wortmannin can inhibit both type III PI3K (required for autophagy) and type I PI3K, we extended our investigation to include PI3K-I inhibitors, specifically Capivasertib (8 nM) and Pictilisib (3 nM), to evaluate their effect on TLR7-induced chemoresistance (Figure S4A). Nevertheless, these PI3K-I inhibitors did not alter the chemoresistance prompted by TLR7 stimulation, supporting the specificity of autophagy process in this context. Consistent findings were observed when examining additional lung tumor cell lines (Figure S4B,S4C). To further validate these findings, we conducted similar experiments in cells deficient for BCN1 (BCN1 KO) and ATG5 (ATG5 KO), two key components of the autophagy machinery involved in the initiation and elongation phases of autophagy, respectively (Figure 5E). As expected, TLR7 stimulation had no effect in BCN1 KO and ATG5 KO cells compared to their respective control cells, confirming that autophagy is essential for TLR7-mediated resistance to salt-based chemotherapies.

To clarify why TLR7 stimulation leads to chemoresistance against Cis. and Oxali., but not Taxol, we examined the autophagy levels under various treatment conditions. Our initial observation revealed that treating tumor cells with these chemotherapies alone increased autophagy levels and autophagy flux, aligning with existing literature on autophagy induction in chemotherapy-treated tumors (Figure 5F). Further analysis confirmed that TLR7 activation enhanced autophagy in lung tumor cells. Notably, when TLR7 agonist was added to Cis. or Oxali. treatments, there was a more pronounced increase in autophagy flux compared to treatments with chemotherapy or TLR7 agonist alone, indicating a synergistic effect on autophagy induction (Figure 5F). This enhanced autophagy was not seen in taxol-treated cells, where Taxol alone induced a high level of autophagy. This differential response to TLR7 stimulation in Cis. or Oxali. versus Taxol treatments may account for the observed chemoresistance in tumor cells treated with the former drugs (Figure 5F). Importantly, no difference in the induction of complete autophagic flux was observed for all chemotherapy treatments, regardless of whether the cells were treated with TLR7 or not, suggesting that differences observed in taxol-treated cell is not due to a defect in autophagy flux (Figure 5F—bottom).

Autophagy level within tumor cells did not impact immune infiltration

Having established that a high level of autophagy in lung tumor cells correlates with poorer overall survival in NSCLC patients, we next focused on a critical factor in tumor progression: immune infiltration within the TME. Utilizing our non-treated NSCLC patient cohort, for which staining for markers of various immune cells [NK cells, B cells, T lymphocytes (CD8), DC, and macrophages (CD68)] had been previously conducted (reference), we explored the relationship between LC3 staining in tumor cells and immune infiltration. Our analysis revealed no apparent correlation between LC3-positive tumor cells and the different types of immune cells examined (Figure S5A). Additionally, no significant differences in the density of the tested immune cells were observed when comparing patients with high versus low LC3 levels, as previously categorized (Figure S5B). This analysis was also extended to consider the autophagy score of NSCLC patients. Consistent with earlier findings, no significant variation in immune infiltrate was associated with different autophagy scores (Figure S5C). Supporting these observations, a transcriptomic analysis of malignant cells, using public datasets (GSE_111907), revealed no distinct cytokine or chemokine signatures between tumors highly expressing autophagy genes and those with low autophagy gene expression (Figure S5D). In summary, these findings suggest that the level of autophagy within tumor cells was not significantly associated with the immune infiltration.

TLR7-induced autophagy, in combination with salt-based chemotherapy, significantly enhanced the expression of immune checkpoint proteins on tumor cells

Given that the immune infiltration seems unaffected by the level of autophagy in lung tumor cells, we next explored the potential impact of autophagy on the expression of immune modulators of the tumor cells themselves. This includes their ability to be recognized and eliminated by the immune system. Utilizing the public cohort of NSCLC patients with comprehensive transcriptomic analysis performed on sorted malignant cells (GSE-111907 dataset), we assessed the expression of various genes encoding proteins that influence the immunogenicity of tumor cells (Figure S6A). Among the genes encoding proteins with pro-tumoral effects (indicated in red), we identified a significant expression of genes related to negative immune checkpoints, including those coding for CD47, galectin-9 (LGAS9), B7-H3 (CD276), IDO (IDO1) and PD-L1 (CD274). Conversely, genes with anti-tumoral effects (indicated in green) that were highly expressed included those coding for the interferon gamma receptor 1 (IFNGR1) and the pro-death molecule Fas (FAS) (Figure S6A). To investigate the association of autophagy with the expression of these genes, we examined the correlation between the expression levels of these protein-encoding genes and autophagy genes (using GC database). Our analysis revealed a strong correlation between autophagy genes and those encoding the interferon gamma (IFNG) receptor, galectin-9, B7-H3, PD-L1, and Fas, suggesting a potential role of autophagy in modulating their expression (Figure 6A,6B). A moderate correlation was observed with genes encoding CD47 (r=0.43), while no significant correlation was detected with genes encoding IDO (r=0.12).

Figure 6 Impact of TLR7-induced autophagy on immune phenotype of malignant cells and on the response to anti-PD-1 immunotherapy in NSCLC patients. (A) Analysis of correlations between the expression of autophagy genes (GC dataset) and the most prominently expressed immune checkpoint genes. The gene correlation was evaluated using sorted NSCLC malignant cells (GSE-111907 dataset). In red, genes coding for proteins which exert pro-tumoral effects, and in green, genes coding for proteins which exert pro-tumoral effects. (B) Correlation between gene coding for PD-L1 and the expression of autophagy genes (GC dataset) using sorted NSCLC malignant cells (GSE-111907 dataset). (C) Graph representing the ratio of the relative PD-L1 expression (from NT condition) measured by flow cytometry in TLR7 synthetic agonist-treated A549 cells alone (Loxo. 1 mM) or in combination with different chemotherapies at 48 h post-treatment. The chemotherapies used are Cis. (50 µM), Oxali. (200 µM), and taxol (10 µM). In addition, different autophagy inhibitors (3-MA, 10 mM, and SAR405, 10 µM, Baf., 100 nM) were added to these combinatory treatments. The experiment was performed at least three times in triplicates. (D) Graph representing the ratio of the relative PD-L1 expression (from NT condition) of A549 cells deficient for beclin-1 (BECN KO) or ATG5 (ATG5 KO), treated or not (NT) with different chemotherapies: Cis. (50 µM), Oxali. (200 µM), and taxol (10 µM) for 48 hours. The experiment was performed at three times in triplicates. (E) Graph representing the percentage of LC3 positive versus LC3 negative lung tumor cells in PD-L1 low (proportion of NSCLC tumor cells expressing PD-L1<40%) versus PD-L1 high (proportion of NSCLC tumor cells expressing PD-L1 >40%) NSCLC patients (cohort 3). (F) Left: graph showing the average autophagy score in PD-L1 low versus PD-L1 high NSCLC patients in the same cohorts. Right: distribution of the PD-L1 expression level in NSCLC patients in function of their autophagy score. (G) Graph representing the percentage of LC3 positive versus LC3 negative lung tumor cells in NSCLC patients treated with nivolumab (cohort 3) in function of their responsiveness to the immunotherapy. (H) Left: distribution of the proportion of NSCLC patients who respond (yes) versus do not respond (no) to nivolumab in function of their autophagy score. Right: proportion of the different autophagy scores of NSCLC patients between responders (yes) versus non-responders to the immunotherapy, nivolumab (no). Student’s t-test, *, P<0.05; NS, not significant. 3-MA, 3-methyladenine; Baf., bafilomycin; Cis., cisplatin; GC, GeneCard; LC3, light chain 3; Loxo., loxoribin; NSCLC, non-small cell lung cancer; Oxali., oxaliplatin; PD-1, programmed death-1; PD-L1, programmed death-ligand 1; Rapa., rapamycin; TLR7, toll-like receptor 7.

Next, we examine how TLR7-induced autophagy affects the modulation of these immune protein expression in lung tumor cells. Our studies revealed that TLR7 stimulation augmented PD-L1 expression at 24 and 48 hours post-stimulation (Figure 6C, Figure S6B). Treatment with a TLR7 agonist also elevated the expression of MHC-I and Fas, while leaving the expression levels of CD47, interferon IFNG receptor, B7-H3, and TLR7 itself unchanged (Figure S6C). Interestingly, combining the TLR7 agonist with Cis. resulted in a more pronounced increase in PD-L1, Fas, CD47, MHC-I, IGNRy and B7-H3 but not TLR7 (Figure 6C, Figure S6D). A similar increase in PD-L1 expression was observed with Oxali. treatment (Figure 6C). In contrast, taxol alone induced a strong upregulation of PD-L1 in lung tumor cells, with no further enhancement upon additional TLR7 stimulation (Figure 6C). To assess the role of autophagy in this upregulation of PD-L1 on tumor cells, we utilized various autophagy inhibitors (3-MA, SAR405, and Baf.). Our observations indicated that the increase in PD-L1 expression, mediated by the combination of Cis. or Oxali. with TLR7 agonist, was markedly diminished by these autophagy inhibitors (Figure 6C). Notably, PIK3-I inhibitors did not affect the enhanced PD-L1 expression seen after the combinatory treatment (Figure S6E). Comparable results were confirmed upon analysis of other lung tumor cell lines (Figure S6F,S6G). Finally, these findings were confirmed in A549 cells deficient for BCN1 and ATG5 (ATG5 KO), where autophagy inhibition prevented the upregulation of PD-L1 in Cis.- or Oxali.-treated cells following TLR7 stimulation (Figure 6D). These findings establish that TLR7, in synergy with salt-based chemotherapy, significantly increases the expression of key immune checkpoint proteins like PD-L1 on tumor cells through the upregulation of autophagy.

Autophagy seems to be associated with increased responsiveness to anti-PD-1 immunotherapy

To explore the relationship between autophagy and immunotherapy efficacy, we analyzed a retrospective cohort of 20 NSCLC patients treated with nivolumab, an anti-PD-1 therapy, using biopsies collected post-treatment (cohort 3). As with the earlier cohorts, we measured autophagy levels in lung tumor cells using LC3 staining. Initially, we examined the correlation between LC3 and PD-L1 expression within these cells, finding no significant link (r=0.31) (Figure S6H). However, when classifying patients based on median PD-L1 expression in tumor cells (PD-L1 low: <40% expression, PD-L1 high: >40% expression), we noted a distinct pattern: while the ratio of LC3-positive to LC3-negative cells in tumor areas was similar in PD-L1 low patients, those with high PD-L1 expression showed a notably greater prevalence of LC3-high cells (Figure 6E). Furthermore, while no significant, the autophagy score seems to be higher in PD-L1 high patients compared to those with lower PD-L1 levels, and patients with an autophagy score of 4 tended to express PD-L1 more intensely than those with a score of 1 (Figure 6F), suggesting a link between autophagy levels in lung tumor cells and PD-L1 expression. We then aimed to understand if the autophagy level in lung tumor cells could correlates with the response of NSCLC patients to nivolumab treatment by comparing the ratio of LC3-positive to LC3-negative cells in tumor areas between responders and non-responders. Notably, responders exhibited a higher proportion of LC3-positive cells and a correspondingly elevated autophagy score compared to non-responders (Figure 6G,6H). These findings collectively seem to indicate that autophagy levels in lung tumor cells are a potential determinant of the response to nivolumab, likely mediated through changes in PD-L1 expression within the lung cancer cells.

Discussion

In this study, we explored a novel pathophysiological pathway inducing autophagy in NSCLC tumor cells, focusing on the role of TLR7. Our correlation studies between lung tumor cell gene expression and autophagy genes, supplemented by in vitro and ex vivo investigations, identified the presence of TLR7 ligands within the TME, involving a TLR7-dependent autophagy induction. Furthermore, our retrospective analysis of various NSCLC patient cohorts revealed that higher autophagy levels in tumor cells correlate with poorer prognosis, reduced chemotherapy response, and, conversely, a potential enhanced response to the anti-PD-1 immunotherapy nivolumab. These findings are supported by our in vitro results, where we demonstrated TLR7-induced chemoresistance to platinum-salt based chemotherapies and an increase in PD-L1 expression, both dependent on autophagy.

Our previous research has shown a link between TLR7 expression in NSCLC and both poor patient prognosis and reduced chemotherapy response, but the mechanisms behind this relationship remained unclear (21,25,26). Our current study suggests TLR7-induced autophagy as a possible underlying mechanism. However, the exact process by which TLR7 stimulates autophagy induction is not yet fully understood. It could involve direct induction through PPIs between TLR7 signaling elements and autophagy-related proteins. This is exemplified in studies like Delgado et al., where rapid autophagy induction in a murine macrophage model involved MyD88, BCN1, and ATG5, with a noted physical interaction between MyD88 and BCN1 (27,28). On the other hand, TLR7 might induce autophagy indirectly by upregulating autophagy gene expression through the activation of NF-kB and MAP kinases pathways, or by inducing the secretion of cytokines such as IFN-I, which are known to regulate autophagy (29-31). Our observations of rapid autophagy induction following TLR7 stimulation support the direct induction hypothesis. However, our preliminary results suggest that MyD88 is not involved in TLR7-induced autophagy (data not shown). Addressing this question will require additional experiments and more extensive mechanistic analyses to determine how TLR7 triggers autophagy in the lung tumor context.

Another aspect to consider is the nature of the TLR7 ligand in the TME and its release mechanism from cancer cells. Based on our results showing that viral infection does not affect TLR7 ligand release (unpublished data) and that cell death halts this release (Figure S2C-S2F), we hypothesize that these ligands are likely cell single-stranded non-coding RNAs, naturally released into the TME by NSCLC tumor cells via extracellular vesicles. Numerous studies have highlighted the secretion of extracellular vesicles containing non-coding RNAs in both non-tumoral and tumoral contexts (32-35). These RNAs can distinctly stimulate TLR7 and induce autophagy. For instance, in systemic lupus erythematosus, cellular RNA release and its TLR7 recognition play a crucial role in disease development (36). In septic mice, plasma extracellular vesicles have been shown to induce inflammation via miRNA- and TLR7-dependent mechanisms (37). In lung and pancreatic cancers, tumor cells secrete miR-21-containing vesicles, promoting JNK-dependent cell death through TLR7 interaction in murine myoblasts (38). In lung cancer, the passive release of miR-574-5p-containing small extracellular vesicles stimulates TLR7 in cancer cells and regulates PGE2-biosynthesis (39). Furthermore, extracellular vesicles can initiate autophagy stimulation, as seen in cases where vesicles from LMP1-activated CAFs promote tumor progression through autophagy and stroma-tumor metabolism coupling (40,41). Furthermore, the phenomenon of autophagy-mediated chemoresistance being transferred through the secretion of extracellular vesicles is documented in various tumor models (42,43). Specifically, NSCLC cells resistant to Cis. have been observed to secrete miR-425-3p via extracellular vesicles, which in turn promotes resistance to chemotherapy by inducing autophagy (44). This mechanism of transferring chemoresistance through autophagy, facilitated by extracellular vesicles, corroborates with our present results, underscoring the critical role of TLR7-mediated autophagy in this context (Figure 5). Importantly, TLR7 expression extends beyond lung tumor cells, encompassing various cells within the TME, particularly immune cells (Figure 1G). Additionally, our analysis reveals a strong correlation between TLR7 expression and autophagy genes in pan-immune cells (Figure 1D). This suggests that TLR7-mediated autophagy not only influence tumor cells directly but also significantly impact other cell types within the TME. While no significant changes in immune infiltration were observed between malignant cells with high versus low levels of autophagy, it is important to analyze the phenotype of the immune cells. Indeed, considering the significant impact of autophagy on immune cells, TLR7 ligands could modulate anti-tumor immune response either positively or negatively, contingent upon the specific ligand and cell type involved (45,46). Future research will focus on elucidating the precise nature of TLR7 ligands and the mechanisms underlying their secretion.

In this study, we found that TLR7 stimulation enhances chemoresistance to platinum-based chemotherapies via an autophagy-dependent mechanism, yet does not induce cytoprotective autophagy in cells treated with taxol (Figure 5D,5E and Figure S4B,S4C). The cytoprotective role of autophagy aligns with existing literature underscoring autophagy’s protective role against the cytotoxicity of these chemotherapeutic agents in lung cancers (16,47-49). Specifically, TLR7-mediated chemoresistance necessitates autophagy maturation (Figure 5F). Our research indicates that all chemotherapies, whether applied independently or in conjunction with TLR7 agonists, trigger a complete autophagy flux, corroborating existing studies (50,51) (Figure 5F). Consequently, the differential response observed in taxol-treated cells is not attributable to autophagy dysfunction. Instead, TLR7 stimulation increases the autophagy level in cells treated with Cis. or Oxali., but not with taxol, where autophagy levels are inherently high with chemotherapy alone (Figure 5F). These results could explain the lack of impact of supplementary TLR7 stimulation on the chemoresistance of taxol-treated cancer cells. Additionally, the impact of taxol on microtubules and its documented role in hindering endolysosomal trafficking might inhibit TLR7 trafficking and activation (52). Although this mechanism has not been explicitly demonstrated for TLR7, taxol-induced disruption of the endolysosomal system has been shown to reduce epidermal growth factor receptor (EGFR) trafficking rates in lung tumor cell models (53). Further research is required to elucidate the distinct effects of TLR7 stimulation in taxol-treated cells. It is noteworthy that patients not responding to combination therapy of Cis. and paclitaxel (taxol) exhibit elevated autophagy levels (Figure S3). This pattern is also observed in patients treated with Cis. in combination with gemcitabine or vinorelbine, indicating no unique impact of taxol when used in conjunction with Cis. in NSCLC patients.

Our findings reveal that autophagy levels significantly influence the expression of molecules that modulate interactions between lung tumor cells and immune cells, ultimately affecting the anti-tumor immune response and the progression of NSCLC. We demonstrated a positive correlation between the expression of genes encoding PD-L1, IFNγ, Galectin 9, B7-H3, and Fas and genes associated with autophagy in malignant cells (Figure 6A,6B). This association between autophagy and these immune protein expression is supported by previous studies (54-57). Furthermore, our study corroborates that chemotherapy elevates the levels of certain proteins (PD-L1, FAS, CD47, MHC-I, B7-H3), reinforcing findings from previous research (58). Intriguingly, our findings reveal that combining a TLR7 agonist with salt-based chemotherapy markedly boosts the expression of these molecules (Figure S6D). Specifically, the enhanced expression of PD-L1 is reliant on autophagy, as blocking autophagy at either its initiation or maturation stages impedes the PD-L1 expression increase post-treatment (Figure 6C). It is important to highlight that the inhibition of autophagy within this combinatorial approach appears to exclusively curtail the additional upregulation of PD-L1 expression induced by TLR7 activation. This is evidenced by observing comparable levels of PD-L1 expression in cells treated with autophagy inhibitors to those subjected to chemotherapy alone (Figure 6C). As overserved for chemoresistance, the synergistic effect on PD-L1 expression from TLR7 stimulation combined with chemotherapy was not evident with Taxol treatment, likely due to previously mentioned mechanisms. The role of autophagy in regulating PD-L1 expression remains a topic of debate, with varying conclusions across studies. While several reports have indicated that autophagy facilitates the degradation of PD-L1 by directing it to lysosomes (59), other studies suggest that autophagy can actually promote PD-L1 expression in tumor models. For instance, a recent work found that autophagy enhances PD-L1 expression in lung tumor cell models by activating STAT3 signaling (6). In human bladder cancer, autophagy was shown to decrease miR-145 expression, leading to increased PD-L1 mRNA stability and protein expression (9). Potentially, autophagy could also facilitate the trafficking of PD-L1 to the plasma membrane, rather than to induce its de novo expression, as mentioned for GLUT1 or EGFR proteins (60,61). Our observations suggest that autophagy contributes to the increased expression of PD-L1 by tumor cells, enhancing the effectiveness of nivolumab treatment in NSCLC patients. However, this observation needs to be confirmed in a larger patient cohort, and the underlying molecular pathways require further investigation.

This study also highlights the potential existence of a functional interplay between TLR7 signaling, autophagy, chemoresistance, and the regulation of PD-L1 expression. Indeed, accumulating evidence has reported reciprocal interactions between autophagy and chemoresistance, between autophagy and PD-L1 expression, as well as between PD-L1 expression levels and chemoresistance (59,62,63). One plausible molecular axis connecting TLR7, autophagy, chemoresistance, and immune checkpoint regulation involves activation of the STAT3 signaling pathway. Autophagy has been shown to activate STAT3 in response to DNA damage, thereby contributing to genome repair mechanisms (64). In addition, the NF-κB pathway (activated downstream of TLR7) has also been reported to promote STAT3 activation (65). Moreover, STAT3 activation has been shown to promote resistance to Cis. and to modulate MHC-I expression in lung tumor cells (66,67). In a similar context, STAT3-driven signaling enhances PD-L1 expression in lung cancer cells, thereby further contributing to chemoresistance (68). Notably, esophageal tumor cells have been reported to secrete PD-L1-containing exosomes, which can in turn promote chemoresistance in recipient cells through STAT3 activation (69). Further investigations are required to elucidate the interconnections between TLR7-induced autophagy, chemoresistance, and PD-L1 expression, with particular emphasis on the role of STAT3 in this context.

Conclusions

To sum up, this study elucidates a novel and critical pathway by which TLR7 stimulation drives autophagy in NSCLC tumor cells, significantly impacting chemotherapy resistance and anti-P1 immunotherapy. We identified TLR7 ligands within the TME as pivotal inducers of autophagy, thereby unveiling a complex interplay between TLR7 activation, autophagy, and anti-tumor therapies. Our findings suggest that higher autophagy levels in tumor cells are associated with poor clinical outcomes, reduced chemotherapy efficacy, and potentially improved response to anti-PD-1 immunotherapy. This study not only advances our understanding of the molecular dynamics within the NSCLC TME but also highlights the potential of TLR7 and/or autophagy as predictive biomarkers to improve therapeutic efficacy. Moreover, a novel therapeutic strategy could be envisaged, notably through the use of a TLR7 antagonist, to selectively reduce autophagic activity in tumor cells while sparing other components of the TME.

Acknowledgments

We are most grateful for excellent technical assistance of the microscopy and imaging platform of the Centre de Recherche des Cordeliers for their support with images analyses. We also thank the Guido’s Kroemer team to help us for the analyses of GFP-LC3 puncta with R. This work partially contains content from Lucas Leonardi’s PhD thesis, submitted to Sorbonne Université in January 2023, and made available under a public license (CC-BY: CC0 1.0).

Footnote

Reporting Checklist: The authors have completed the MDAR reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-aw-1173/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-aw-1173/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-aw-1173/prf

Funding: This work was supported by the “Institut National de la Santé et de la Recherche Médicale” (INSERM), Sorbonne université, Université de Paris Cité, the French foundation ARC (foundation ARC pour la recherche contre le cancer) (No. 20161204806), Cancéropole (No. R23149DD) and the LabEx Immuno-Oncology.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-aw-1173/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the regional ethics committee of “Comité de Protection des Personnes Île-de-France II” (Nos. 2007-A00845-48 and 2012-06-12) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Aman Y, Schmauck-Medina T, Hansen M, et al. Autophagy in healthy aging and disease. Nat Aging 2021;1:634-50. [Crossref] [PubMed]
2. Amaravadi RK, Kimmelman AC, Debnath J. Targeting Autophagy in Cancer: Recent Advances and Future Directions. Cancer Discov 2019;9:1167-81. [Crossref] [PubMed]
3. Liu D, Li M, Zhao Z, et al. Targeting the TRIM14/USP14 Axis Enhances Immunotherapy Efficacy by Inducing Autophagic Degradation of PD-L1. Cancer Res 2024;84:2806-19. [Crossref] [PubMed]
4. Lei Y, Zhang E, Bai L, et al. Autophagy in Cancer Immunotherapy. Cells 2022;11:2996. [Crossref] [PubMed]
5. Zeng H, Zhang W, Gong Y, et al. Radiotherapy activates autophagy to increase CD8(+) T cell infiltration by modulating major histocompatibility complex class-I expression in non-small cell lung cancer. J Int Med Res 2019;47:3818-30. [Crossref] [PubMed]
6. Liu Y, Zhang H, Wang Z, et al. 5-Hydroxytryptamine1a receptors on tumour cells induce immune evasion in lung adenocarcinoma patients with depression via autophagy/pSTAT3. Eur J Cancer 2019;114:8-24. [Crossref] [PubMed]
7. Maher CM, Thomas JD, Haas DA, et al. Small-Molecule Sigma1 Modulator Induces Autophagic Degradation of PD-L1. Mol Cancer Res 2018;16:243-55. [Crossref] [PubMed]
8. Booth L, Roberts JL, Poklepovic A, et al. [pemetrexed + sildenafil], via autophagy-dependent HDAC downregulation, enhances the immunotherapy response of NSCLC cells, et al. Cancer Biol Ther. 2017;18:705.
9. Zhu J, Li Y, Luo Y, et al. A Feedback Loop Formed by ATG7/Autophagy, FOXO3a/miR-145 and PD-L1 Regulates Stem-Like Properties and Invasion in Human Bladder Cancer. Cancers (Basel) 2019;11:349. [Crossref] [PubMed]
10. Leonardi L, Sibéril S, Alifano M, et al. Autophagy Modulation by Viral Infections Influences Tumor Development. Front Oncol 2021;11:743780. [Crossref] [PubMed]
11. Ma S, Xu Y, Qin X, et al. RUNX1, FUS, and ELAVL1-induced circPTPN22 promote gastric cancer cell proliferation, migration, and invasion through miR-6788-5p/PAK1 axis-mediated autophagy. Cell Mol Biol Lett 2024;29:95. [Crossref] [PubMed]
12. Kim EH, Choi J, Jang H, et al. Targeted delivery of anti-miRNA21 sensitizes PD-L1(high) tumor to immunotherapy by promoting immunogenic cell death. Theranostics 2024;14:3777-92. [Crossref] [PubMed]
13. Yang L, Wang H, Shen Q, et al. Long non-coding RNAs involved in autophagy regulation. Cell Death Dis 2017;8:e3073. [Crossref] [PubMed]
14. Yao H, Han B, Zhang Y, et al. Non-coding RNAs and Autophagy. Adv Exp Med Biol 2019;1206:199-220. [Crossref] [PubMed]
15. Akkoc Y, Gozuacik D. MicroRNAs as major regulators of the autophagy pathway. Biochim Biophys Acta Mol Cell Res 2020;1867:118662. [Crossref] [PubMed]
16. Guo W, Du K, Luo S, et al. Recent Advances of Autophagy in Non-Small Cell Lung Cancer: From Basic Mechanisms to Clinical Application. Front Oncol 2022;12:861959. [Crossref] [PubMed]
17. Karpathiou G, Sivridis E, Koukourakis MI, et al. Light-chain 3A autophagic activity and prognostic significance in non-small cell lung carcinomas. Chest 2011;140:127-34. [Crossref] [PubMed]
18. Überall I, Gachechiladze M, Joerger M, et al. Tumor autophagy is associated with survival outcomes in patients with resected non-small cell lung cancer. Lung Cancer 2019;129:85-91. [Crossref] [PubMed]
19. Langer R, Neppl C, Keller MD, et al. Expression Analysis of Autophagy Related Markers LC3B, p62 and HMGB1 Indicate an Autophagy-Independent Negative Prognostic Impact of High p62 Expression in Pulmonary Squamous Cell Carcinomas. Cancers (Basel) 2018;10:281. [Crossref] [PubMed]
20. Leonardi L, Siberil S, Alifano M, et al. Autophagy-Related Gene Signature Highlights Metabolic and Immunogenic Status of Malignant Cells in Non-Small Cell Lung Cancer Adenocarcinoma. Cancers (Basel) 2022;14:3462. [Crossref] [PubMed]
21. Chatterjee S, Crozet L, Damotte D, et al. TLR7 promotes tumor progression, chemotherapy resistance, and poor clinical outcomes in non-small cell lung cancer. Cancer Res 2014;74:5008-18. [Crossref] [PubMed]
22. Velut Y, Decroix E, Blons H, et al. SMARCA4-deficient lung carcinoma is an aggressive tumor highly infiltrated by FOXP3+ cells and neutrophils. Lung Cancer 2022;169:13-21. [Crossref] [PubMed]
23. Yuan Q, Zhou Q, Ren J, et al. WGCNA identification of TLR7 as a novel diagnostic biomarker, progression and prognostic indicator, and immunotherapeutic target for stomach adenocarcinoma. Cancer Med 2021;10:4004-16. [Crossref] [PubMed]
24. Gentles AJ, Hui AB, Feng W, et al. A human lung tumor microenvironment interactome identifies clinically relevant cell-type cross-talk. Genome Biol 2020;21:107. [Crossref] [PubMed]
25. Dajon M, Iribarren K, Petitprez F, et al. Toll like receptor 7 expressed by malignant cells promotes tumor progression and metastasis through the recruitment of myeloid derived suppressor cells. Oncoimmunology 2018;8:e1505174.
26. Cherfils-Vicini J, Platonova S, Gillard M, et al. Triggering of TLR7 and TLR8 expressed by human lung cancer cells induces cell survival and chemoresistance. J Clin Invest 2010;120:1285-97. [Crossref] [PubMed]
27. Delgado MA, Elmaoued RA, Davis AS, et al. Toll-like receptors control autophagy. EMBO J 2008;27:1110-21. [Crossref] [PubMed]
28. Shi CS, Kehrl JH. MyD88 and Trif target Beclin 1 to trigger autophagy in macrophages. J Biol Chem 2008;283:33175-82. [Crossref] [PubMed]
29. Tian Y, Wang ML, Zhao J. Crosstalk between Autophagy and Type I Interferon Responses in Innate Antiviral Immunity. Viruses 2019;11:132. [Crossref] [PubMed]
30. Luo W, Sun R, Chen X, et al. ERK Activation-Mediated Autophagy Induction Resists Licochalcone A-Induced Anticancer Activities in Lung Cancer Cells in vitro. Onco Targets Ther 2020;13:13437-50. [Crossref] [PubMed]
31. Zhu X, Zhou M, Liu G, et al. Autophagy activated by the c-Jun N-terminal kinase-mediated pathway protects human prostate cancer PC3 cells from celecoxib-induced apoptosis. Exp Ther Med 2017;13:2348-54. [Crossref] [PubMed]
32. Hasan H, Sohal IS, Soto-Vargas Z, et al. Extracellular vesicles released by non-small cell lung cancer cells drive invasion and permeability in non-tumorigenic lung epithelial cells. Sci Rep 2022;12:972. [Crossref] [PubMed]
33. Chang WH, Cerione RA, Antonyak MA. Extracellular Vesicles and Their Roles in Cancer Progression. Methods Mol Biol 2021;2174:143-70. [Crossref] [PubMed]
34. Doyle LM, Wang MZ. Overview of Extracellular Vesicles, Their Origin, Composition, Purpose, and Methods for Exosome Isolation and Analysis. Cells 2019;8:727. [Crossref] [PubMed]
35. Verweij FJ, Balaj L, Boulanger CM, et al. The power of imaging to understand extracellular vesicle biology in vivo. Nat Methods 2021;18:1013-26. [Crossref] [PubMed]
36. Satterthwaite AB. TLR7 signaling in lupus B cells – new insights into synergizing factors and downstream signals. Curr Rheumatol Rep 2021;23:80. [Crossref] [PubMed]
37. Xu J, Feng Y, Jeyaram A, et al. Circulating Plasma Extracellular Vesicles from Septic Mice Induce Inflammation via MicroRNA- and TLR7-Dependent Mechanisms. J Immunol 2018;201:3392-400. [Crossref] [PubMed]
38. He WA, Calore F, Londhe P, et al. Microvesicles containing miRNAs promote muscle cell death in cancer cachexia via TLR7. Proc Natl Acad Sci U S A 2014;111:4525-9. [Crossref] [PubMed]
39. Donzelli J, Proestler E, Riedel A, et al. Small extracellular vesicle-derived miR-574-5p regulates PGE2-biosynthesis via TLR7/8 in lung cancer. J Extracell Vesicles 2021;10:e12143. [Crossref] [PubMed]
40. Colletti M, Ceglie D, Di Giannatale A, et al. Autophagy and Exosomes Relationship. in Cancer: Friends or Foes? Front Cell Dev Biol 2020;8:614178.
41. Wu X, Zhou Z, Xu S, et al. Extracellular vesicle packaged LMP1-activated fibroblasts promote tumor progression via autophagy and stroma-tumor metabolism coupling. Cancer Lett 2020;478:93-106. [Crossref] [PubMed]
42. Xavier CPR, Belisario DC, Rebelo R, et al. The role of extracellular vesicles in the transfer of drug resistance competences to cancer cells. Drug Resist Updat 2022;62:100833. [Crossref] [PubMed]
43. Salimi L, Akbari A, Jabbari N, et al. Synergies in exosomes and autophagy pathways for cellular homeostasis and metastasis of tumor cells. Cell Biosci 2020;10:64. [Crossref] [PubMed]
44. Ma Y, Yuwen D, Chen J, et al. Exosomal Transfer Of Cisplatin-Induced miR-425-3p Confers Cisplatin Resistance In NSCLC Through Activating Autophagy. Int J Nanomedicine 2019;14:8121-32. [Crossref] [PubMed]
45. Xia H, Green DR, Zou W. Autophagy in tumour immunity and therapy. Nat Rev Cancer 2021;21:281-97. [Crossref] [PubMed]
46. Jiang GM, Tan Y, Wang H, et al. The relationship between autophagy and the immune system and its applications for tumor immunotherapy. Mol Cancer 2019;18:17. [Crossref] [PubMed]
47. Sumkhemthong S, Prompetchara E, Chanvorachote P, et al. Cisplatin-induced hydroxyl radicals mediate pro-survival autophagy in human lung cancer H460 cells. Biol Res 2021;54:22. [Crossref] [PubMed]
48. Datta S, Choudhury D, Das A, et al. Autophagy inhibition with chloroquine reverts paclitaxel resistance and attenuates metastatic potential in human nonsmall lung adenocarcinoma A549 cells via ROS mediated modulation of β-catenin pathway. Apoptosis 2019;24:414-33. [Crossref] [PubMed]
49. Li X, Zhou Y, Li Y, et al. Autophagy: A novel mechanism of chemoresistance in cancers. Biomed Pharmacother 2019;119:109415. [Crossref] [PubMed]
50. Škubník J, Svobodová Pavlíčková V, Ruml T, et al. Autophagy in cancer resistance to paclitaxel: Development of combination strategies. Biomed Pharmacother 2023;161:114458. [Crossref] [PubMed]
51. Lin JF, Lin YC, Tsai TF, et al. Cisplatin induces protective autophagy through activation of BECN1 in human bladder cancer cells. Drug Des Devel Ther 2017;11:1517-33. [Crossref] [PubMed]
52. Shin GJ, Pero ME, Hammond LA, et al. Integrins protect sensory neurons in models of paclitaxel-induced peripheral sensory neuropathy. Proc Natl Acad Sci U S A 2021;118:e2006050118. [Crossref] [PubMed]
53. Li H, Duan ZW, Xie P, et al. Effects of paclitaxel on EGFR endocytic trafficking revealed using quantum dot tracking in single cells. PLoS One 2012;7:e45465. [Crossref] [PubMed]
54. Sudhakar JN, Lu HH, Chiang HY, et al. Lumenal Galectin-9-Lamp2 interaction regulates lysosome and autophagy to prevent pathogenesis in the intestine and pancreas. Nat Commun 2020;11:4286. [Crossref] [PubMed]
55. Schulz D, Severin Y, Zanotelli VRT, et al. In-Depth Characterization of Monocyte-Derived Macrophages using a Mass Cytometry-Based Phagocytosis Assay. Sci Rep 2019;9:1925. [Crossref] [PubMed]
56. Jung KT, Oh SH. Polyubiquitination of p62/SQSTM1 is a prerequisite for Fas/CD95 aggregation to promote caspase-dependent apoptosis in cadmium-exposed mouse monocyte RAW264.7 cells. Sci Rep 2019;9:12240. [Crossref] [PubMed]
57. Sharma G, Ojha R, Noguera-Ortega E, et al. PPT1 inhibition enhances the antitumor activity of anti-PD-1 antibody in melanoma. JCI Insight 2020;5:e133225. [Crossref] [PubMed]
58. Hassanian H, Asadzadeh Z, Baghbanzadeh A, et al. The expression pattern of Immune checkpoints after chemo/radiotherapy in the tumor microenvironment. Front Immunol 2022;13:938063. [Crossref] [PubMed]
59. Cui Y, Shi J, Cui Y, et al. The relationship between autophagy and PD-L1 and their role in antitumor therapy. Front Immunol 2023;14:1093558. [Crossref] [PubMed]
60. Roy S, Leidal AM, Ye J, et al. Autophagy-Dependent Shuttling of TBC1D5 Controls Plasma Membrane Translocation of GLUT1 and Glucose Uptake. Mol Cell 2017;67:84-95.
61. Fraser J, Simpson J, Fontana R, et al. Targeting of early endosomes by autophagy facilitates EGFR recycling and signalling. EMBO Rep 2019;20:e47734. [Crossref] [PubMed]
62. Pu Y, Wang J, Wang S. Role of autophagy in drug resistance and regulation of osteosarcoma Mol Clin Oncol 2022;16:72. (Review). [Crossref] [PubMed]
63. Özen Eroğlu G, Erol Bozkurt A, Yaylım İ, et al. PD-1/PD-L1 Inhibitors and Chemotherapy Synergy: Impact on Drug Resistance and PD-L1 Expression in Breast Cancer-Immune Cell Co-Cultures. Int J Mol Sci 2025;26:6876. [Crossref] [PubMed]
64. Xu F, Li X, Yan L, et al. Autophagy Promotes the Repair of Radiation-Induced DNA Damage in Bone Marrow Hematopoietic Cells via Enhanced STAT3 Signaling. Radiat Res 2017;187:382-96. [Crossref] [PubMed]
65. Soutto M, Bhat N, Khalafi S, et al. NF-kB-dependent activation of STAT3 by H. pylori is suppressed by TFF1. Cancer Cell Int 2021;21:444.
66. Morelli AP, Tortelli TC Jr, Mancini MCS, et al. STAT3 contributes to cisplatin resistance, modulating EMT markers, and the mTOR signaling in lung adenocarcinoma. Neoplasia 2021;23:1048-58. [Crossref] [PubMed]
67. Kida H, Ihara S, Kumanogoh A. Involvement of STAT3 in immune evasion during lung tumorigenesis. Oncoimmunology 2013;2:e22653. [Crossref] [PubMed]
68. Tang D, Zhao D, Wu Y, et al. The miR-3127-5p/p-STAT3 axis up-regulates PD-L1 inducing chemoresistance in non-small-cell lung cancer. J Cell Mol Med 2018;22:3847.
69. Wang H, Qi Y, Lan Z, et al. Exosomal PD-L1 confers chemoresistance and promotes tumorigenic properties in esophageal cancer cells via upregulating STAT3/miR-21. Gene Ther 2023;30:88-100. [Crossref] [PubMed]