Prognostic and immunological potential of AC012236.1/hsa-miR-30d-5p CeRNA of AVEN by integrated analysis of single-cell and bulk RNA-seq in lung adenocarcinoma

Abstract

Lung adenocarcinoma (LUAD) is the most prevalent histological subtype of non-small cell lung cancer (NSCLC) and is characterized by high mortality and limited therapeutic efficacy in advanced stages. AVEN, an apoptosis inhibitor that interacts with Bcl-xL and Apaf-1 to suppress caspase activation, has been implicated in tumour progression and drug resistance in various cancers. However, its role in LUAD remains unclear. In this study, the prognostic importance, immune microenvironment association, and regulatory mechanisms of AVEN in LUAD were comprehensively investigated using bulk RNA sequencing (RNA-seq), single-cell RNA sequencing (scRNA-seq), and experimental validation. Analysis of the TCGA and GEO datasets revealed that AVEN expression was significantly upregulated in LUAD tissues compared with normal tissues and that high AVEN expression correlated with advanced T/N stage and pathological stage and was associated with poor overall survival (OS), disease-specific survival (DSS), and progression-free survival (PFS). Multivariate Cox regression identified AVEN expression as an independent prognostic factor, and a nomogram incorporating AVEN expression demonstrated high predictive accuracy for 1-, 3-, and 5-year OS. Functional enrichment analysis linked AVEN to keratinocyte differentiation, spliceosome activity, and cell cycle pathways, whereas the results of scRNA-seq highlighted its predominant expression in malignant epithelial cell subtypes (tS2), which is associated with aggressive proliferation and immune evasion. AVEN expression was positively correlated with Th2, NK CD56dim, and Tgd cell infiltration but negatively associated with TFH, eosinophil, and mast cell infiltration, suggesting its role in modulating the tumour immune microenvironment. Detection of clinical samples verified the high expression of AVEN in LUAD. In vitro, AVEN knockdown in A549 cells suppressed proliferation, migration, and invasion while promoting apoptosis. Furthermore, bioinformatics prediction and validation revealed that hsa-miR-30d-5p was an upstream regulator of AVEN, with its low expression in LUAD tissues inversely correlated with that of AVEN and predicting a favourable prognosis. Subsequent bioinformatics analysis further revealed that lncRNA-AC012236.1 functioned as an upstream regulator of hsa-miR-30d-5p. This lncRNA was found to be highly expressed in LUAD tissues, and its elevated expression was significantly associated with poor overall survival (OS) in LUAD patients. In conclusion, AVEN, as a promising diagnostic and prognostic biomarker in LUAD, affected tumour progression, immune infiltration and apoptosis resistance through the lncRNA-AC012236.1/hsa-miR-30d-5p-AVEN axis. These findings provided new insights into the pathogenesis of LUAD and highlighted potential therapeutic targets for improving patient prognosis.

Introduction

Lung cancer is among the common causes of cancer-related death worldwide and is associated with high morbidity and mortality, resulting in severe social and economic burdens¹. Non-small cell lung cancer (NSCLC) is the most common type of lung cancer, accounting for more than 85% of lung cancer cases. Among non-small cell lung cancers, lung adenocarcinoma (LUAD) is the dominant histological subtype, accounting for approximately 40% of cases². Despite new advances in immune checkpoint inhibitors and targeted therapies in recent years^3,4,5, the prognosis for lung cancer patients, especially those in advanced stages, remains poor. Therefore, discovering new core molecular network features and establishing new diagnostic and therapeutic models for early LUAD are crucial for reducing the damage caused by LUAD in the population. In recent years, research related to tumour gene therapy has increased. Gene therapy with specific small molecules, including genes, small interfering RNAs (siRNAs)/microRNAs (miRNAs) and oligonucleotides, has entered clinical trials and has shown promising results^6,7,8.

Apoptosis and caspase activation inhibitor (AVEN) is an intracellular apoptosis inhibitor that interacts with large B-cell lymphoma (Bcl-xL, an antiapoptotic Bcl-2 family member) and apoptotic protease activator 1 (Apaf-1) to inhibit caspase activation and thus apoptosis⁹. The antiapoptotic effect of AVEN is involved in various pathophysiological activities. A recent study of PNAS revealed that inhibition of AVEN increased myocardial damage after myocardial infarction¹⁰, and another study reported that AVEN inhibited autophagy and apoptosis in atherosclerosis¹¹. In addition, AVEN was reported to protect podocytes against the damage induced by high glucose in diabetic nephropathy¹². AVEN is considered an oncogene for a variety of cancers and is involved in the generation of tumour drug resistance. The overexpression of AVEN worsened the prognosis of patients with haematologic diseases. AVEN is involved in the progression of brain glioma and chemotherapy resistance in osteosarcoma. AVEN regulates tumorigenesis through autophagy, and its role in drug resistance is mediated by the Akt signalling pathway^{11,13,14,15,16}. To date, only one bioinformatics article has mentioned an AVEN-derived prognostic model in LUAD, but without experimental validation¹¹. These results suggested that AVEN might play important roles in the development of tumours and their progression to drug resistance. In addition to exploring AVEN as a prognostic biomarker, this study further investigated the mechanism of a ceRNA constructed from its upstream lncRNA/miRNA and its prognostic ability.

MicroRNAs (miRNAs) are short noncoding RNAs that are evolutionarily conserved and participate in the regulation of posttranscriptional protein expression by binding to the 3’-untranslated region (3’-UTR) of target mRNAs, which may regulate tumorigenesis at multiple levels¹⁷. Progress has been made in the development of molecular therapies targeting miRNAs, which are in early clinical trials. In the treatment of lung cancer, miRNA-based protocols could complement traditional chemoradiotherapy, immunotherapy, and targeted therapies to reverse drug resistance and improve the sensitivity of lung cancer cells⁶. Long noncoding RNAs (lncRNAs) can act as competing endogenous RNAs (ceRNAs) of miRNAs to regulate the target genes of miRNAs in a competitive manner. In the ceRNA mode, lncRNAs can reverse the inhibitory effect of miRNAs on mRNA expression¹⁸. However, the role of AVEN and its upstream ceRNA network in LUAD remain unclear.

In this study, the relationships of AVEN with LUAD prognosis and the tumour microenvironment were analysed through a bulk RNA sequencing (RNA-seq) database and scRNA-seq single-cell RNA sequencing (scRNA-seq) analysis, and the lncRNAs and miRNAs of AVEN and their prognostic correlations in LUAD were predicted.

Materials and methods

Patient datasets

The cancer genome atlas (TCGA) (https://portal.gdc.cancer.gov/) database and Genotype Tissue Expression (GTEx) ( http://commonfund.nih.gov/GTEx/ ) database were used to detected the AVEN expression levels in paired and unpaired of 33 human cancer samples¹⁹. We downloaded the mRNA expression data and clinical data information of LUAD (including 539 LUAD tumour samples and 59 normal samples) from the TCGA database. Download three LUAD datasets, namely GSE10072²⁰, GSE19188²¹ and GSE31210²² from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo). GSE19188 and GSE10072 were used to verify the expression level of AVEN in LUAD. GSE31210 was used to verify prognosis value.

The Human Protein Atlas (HPA) database (http://www.proteinatlas.org/) was used to validate the protein expression levels of AVEN in LUAD.

A total of 14 LUAD samples from March 2022 to December 2024 were collected from 7 male patients and 7 female patients (average age 62) and the adjacent lung tissues which distance from tumour tissue more than 2 cm were collected as negative controls. Inclusion criteria: (1) First diagnosed as lung adenocarcinoma by histopathological examination and without any treatment; (2) The patient’s clinical data is complete, and the expected survival period is more than 6 months. (3) The patient has no serious infections, autoimmune diseases or other complications, and does not have any other malignant tumours. Patients with severe heart, liver or kidney diseases, incomplete clinical data (with missing data exceeding 20%), or multiple primary tumours were excluded from the study. Samples that met the inclusion criteria were included in this study to avoid selection bias.The studies were conducted in accordance with the ethical standards according to The Declaration of Helsinki, as well as national and international guidelines. The present study was approved by the Institutional Research Ethics Committee of Yantai Affiliated Hospital of Binzhou Medical University (protocol no. 20220215001; Yantai, China). The patients provided written informed consent to participate in this study.

To evaluate the statistical power of the current sample size, we conducted a post hoc power analysis by G*Power 3.1²³. Based on the results of the Western blot experiment, the observed effect value (Cohen’s d) is 2.1, the significance level is set to 0.05, and the statistical power of the current sample size (n = 14) is 0.99. This value is higher than the conventional threshold of 0.8. The current sample size is sufficient to detect significant differences between groups.

Single-gene difference analysis of AVEN

Single-gene difference analysis of AVEN gene in LUAD based on TCGA database were performed using the XIANTAO platform (www.xiantaozi.com)²⁴. To avoid the batch effect caused by cross-dataset analysis, we independently conducted differential expression analysis within each dataset. The raw RNA-seq count data were standardized using “limma” packages (TMM standard method) before entering downstream analysis to correct for differences in library size and sequencing depth. Subsequently, we evaluated through PCA whether there were significant batch effects within the dataset to ensure that the samples were mainly based on the biological state (tumour vs. normal tissues) rather than technical factors. According to the median AVEN expression level, LUAD samples were divided into high expression group and low expression group. Volcano plots were drawn based on the single gene difference analysis results, DEG cutoff values were set to∣Log FC∣ < 1 and adjusted p value < 0.05.

Functional enrichment analysis gene set enrichment analysis (GSEA)

To explore potential signaling pathways based on differential expression analysis (AVEN high expression versus AVEN low expression samples) in TCGA LUAD project. Adjusted p-value < 0.05 was considered to be significantly enriched. After screening DEGs based on truncation value (Log FC > 1 and adjusted p value < 0.05), Gene ontology (GO) ( http://geneontology.org/ ) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (https://www.kegg.jp/) pathway enrichment analyses of DEGs were performed using DAVID software (http://david.ncifcrf.gov/)^25,26. P < 0.05 was set as the cut-off criterion.

Protein-protein interaction (PPI) network analysis

Visual analysis of DEGs was performed using the STRING database²⁷. network analysis and protein-protein interactions (PPI) of DEGs using Cytoscape software, and then 21 HUB genes were identified using the MCODE plugin. Co-expression heat maps and correlation heat maps of 21 HUB genes were drawn using the “ggplot2” R package.

Variation and mutation analysis of AVEN in LUAD

We obtain the variation type and the frequency of the AVEN in LUAD from cBioPortal database (http://www.cbioportal.org/). We further obtain the mutation type and the probability of the AVEN in LUAD from the Catalogue of Somatic Mutations in Cancer (COSMIC) (https://cancer.sanger.ac.uk).

Structure of AVEN

The structure of AVEN was constructed by uniport database (https://www.uniprot.org). AlphaFold produces a per-residue confidence score (pLDDT) between 0 and 100 by different colors. Blues, very high (pLDDT > 90); cyans, confident (90 > pLDDT > 70); yellows, low (70 > pLDDT > 50); oranges, very low (pLDDT < 50).

Survival analysis

The Kaplan Meier - plotter (http://kmplot.com/analysis/)²⁸was used to evaluate the correlation between the AVEN expression and different clinical features of LUAD patients in TCGA database. We selected RNAseq of LUAD patients with complete clinical data.

Receiver operating characteristic (ROC) analysis

ROC curves were plotted to evaluate the diagnostic value of AVEN in LUAD.

Analysis of nomograms to assess prognosis

Based on Cox regression results, T stage, N stage, M stage, Primary therapy outcome and AVEN expression level were selected to construct prognostic nomogram for evaluating 1-year, 3-year and 5-year OS rates of LUAD patients in TCGA database, and calibration plots were drawn to verify its accuracy.

Correlation analysis between AVEN, immune cell infiltration, immune checkpoints and cytokines

Single-sample gene set enrichment (SSGSE) analysis were performed to analyze the association between AVEN and immune cell infiltration, immune checkpoints and cytokines in LUAD on the XIANTAO platform. P < 0.05 was set as the cut-off criterion.

scRNA-seq analysis

R-packet Seurat was used to filter and further analyze the gene expression matrix of GSE131907, and cells were identified according to criteria of mitochondrial gene expression ≤ 20% and gene counts greater than 200 to less than 10,000. Cell type recognition based on the specific cells obtained from Cell Marker database (http://biocc.hrbmu.edu.cn/CellMarker/). The Average Expression function was used to calculate the average expression value of the cells contained in the cell subpopulation or different groups, then calculate the score of the corresponding pathway for each cell subpopulation or group according to the GSVA package, and finally visualize by P heatmap. Pseudo-time trajectories were analized by Monocle 2.

ENCORI database analysis

Encyclopedia of RNA Interactomes (ENCORI) database is a database for the discovery of links between miRNA-lncRNA and miRNA-mRNA²⁹. ENCORI was used to predict upstream potential miRNAs interacting with AVEN and upstream potential lncRNAs interacting with hsa-miR-30d-5p. Meanwhile, we used Cytoscape software to build the ceRNA network of LncRNA/hsa-miR-30d-5p/AVEN.

Cell culture and treatment

A549 cells were acquired from the National Collection of Authenticated Cell Cultures (Shanghai, China). Cells were cultured in DMEM medium (HyClone, Thermo Fisher Scientific Inc) supplemented with 10% FBS and 100 U/ml penicillin/streptomycin at 37°C in a 5% CO₂ incubator. The transfection of siRNA was carried out by Lipo8000™ (cat. no. C0533, Beyotime). After 48 h of transfection, the silencing efficiency was examined by qRT-PCR and subsequent experiments were performed. siRNA was designed and constructed by Wuhan Genecreate Biological Engineering CO., LTD. The sequence of siAVEN: 5’-GAGCAACCAAGUACCUCCAAATT-3’.

Western blot analysis of AVEN expression

Tissues and cells were homogenized with RIPA (with 1% PMSF) tissue lysate. The concentration of the extraction was measured by a BCA protein concentration kit. Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGEs) were performed, following by membrane transferring and blocking in 5% skim milk powder. The primary antibody against AVEN (product number: sc-136455, 1:200, Santa Cruz) was incubated overnight at 4 °C. The secondary antibody was incubated at room temperature for 2.5 h. The blots were visualized using an enhanced chemiluminescence detection system (Tanon, Shanghai, China) with Millipore Immobilon ECL (Millipore). GAPDH (product number: 60004-1-Ig, 1:5000, ProteinTech) was used as an internal control. Image J was used to analyze the gray values of the western blot bands. The second antibodies included HRP-conjugated Affinipure Goat Anti-Rabbit IgG (H + L) (product number: SA00001-2, 1:3000, ProteinTech) and HRP-conjugated Affinipure Goat Anti-Mouse IgG (H + L) (product number: SA00001-1, 1:3000, ProteinTech).

RNA extraction and real-time quantitative PCR (RT-qPCR)

Total RNA was extracted from cells using Ultrapure RNA Kit (cat. no. CW0581; Jiangsu CoWinBiotech Co., Ltd.). The HiFiScript cDNA Synthesis Kit (cat. no. CW2569M) was used to reverse transcribe RNA into cDNA, and the Hieff UNICON Universal Blue qPCR SYBR Green Master Mix (cat. no. 11184ES, Yeasen Biotechnology (Shanghai) Co., Ltd.) was used for qPCR. The qPCR was performed at 95°C for 30 sec, followed by 40 cycles at 95°C for 3 sec, 60°C for 20 sec on the QuantStudio 1 Real-Time PCR System (Thermo Fisher Scientific Inc). The relative mRNA expression levels were measured using the 2^−∆∆Cq method³⁰. The specific primer sequences were as follows: AVEN, 5’- GAGCAACCAAGUACCUCCAAATT-3’ and reverse 5’-UUUGGAGGUACUUGGUUGCUCTT-3’; GAPDH, forward 5’- CGGAGTCAACGGATTTGGTCGTAT-3’ and reverse 5’- AGCCTTCTCCATGGTGGTGAAGAC-3’, has-miR-30d-5p stem loop primers, CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCTTCCAGT-3’; has-miR-30d-5p, forward 5’-ACACTCCAGCTGGGTGTAAACATCCCCGAC-3’ and reverse 5’-TGGTGTCGTGGAGTCG-3’; U6 stem loop primers, 5’-AACGCTTCACGAATTGTGCGT-3’; U6, forward 5’-CTCGCTTCGGCAGCACA-3’ and reverse 5’-AACGCTTCACGAATTGTGCGT-3’.

Cell counting Kit-8 (CCK-8) assay

After siRNA transfection, cells of each group were collected and seeded into 96-well plates with 2000 cells/ 100 µl per well. CCK‐8 solution (cat. no. CK04, Dojindo) at 10% of the total volume of culture medium was added to each well at 0 h, 24 h, 48 h, respectively. After adding the working solution, the 96‐well plate was cultured in the incubator for 2 h. The absorbance of each well at 450 nm was detected by SpectraMax Mini (Molecular Devices, USA).

Transwell invasion assay

After siRNA transfection, 60 µL of diluted matrix gel (Matrigel, 1:8 diluted in serum-free medium) was spreaded on the upper surface of the transwell chamber and incubated at 37 °C for 2 h to polymerize the gel and form a basement membrane barrier. 200 µL of serum-free cell suspension (density 2 × 10⁵ cells/ mL) was added to the upper chamber and 600 µL of medium containing 10% FBS was added to the lower chamber at 37 °C for 24 h. Remove the culture medium in the chamber. The non-migrated cells on upper side of the membranes were removed by cotton swabs. The cells on lower side of the membranes in the chamber were fixed with 4% paraformaldehyde for 30 min, and then in 0.1% crystal violet for 20 min at RT. The floating color was removed by PBS. The migrated cells were observed MF53-N microscope (Guangzhou Mingmei Optoelectronic Technology Co., LTD) at a magnification of 400 ×. The number of invasion cells was counted in three random fields and statistically analysed by Image J software.

Scratch assay

The scratch assay was performed to evaluate the effect of the migration of A549 cells after siRNA transfection. When the cells grown to 80% confluence, a straight scratched wound was made by a sterile 200 µL micro-pipette tip in the monolayer of the cells. The floating cells and cellular debris were removed by PBS. Then, cells were cultured in serum-free DMEM at 37 °C for 24 h. The distance between the two sides of the wound was monitored and photographed at 0 and 24 h by MF53-N microscope (Guangzhou Micro-shot Technology Co.Ltd.) at a magnification of 200 ×. The cell migrate rate was calculated by Image J software.

Flow cytometry for apoptosis analysis

Annexin V/propidium iodide (PI) staining was conducted to quantitatively analyze apoptosis after siRNA transfection. Cells were harvested and centrifuged at 300 × g for 5 min at 4 °C. The cell pellet was resuspended in 500 uL 1× binding buffer at a density of 1 × 106 cells/ml, and the cells were then treated with 5 uL Annexin V-EGFP and 5uL Propidium lodide (PI) in the dark at room temperature for 15 min. The samples were analyzed using a FACSCanto II flow cytometer (BD Biosciences).

Statistical analysis

The expression profile of AVEN in patients with LUAD was illustrated in violin plots and scatter plots. The data from TCGA database were analyzed by Mann-Whitney U test for the unpaired data and the paired Student’s t-test for the paired data. The unpaired data from the GEO database were analyzed by Mann-Whitney U test. The association between clinical parameters and AVEN expression was analyzed by Kruskal-Wallis’ test and Dunn’s test. The prognostic potential of AVEN in patients with LUAD was evaluated via univariate and multivariate Cox regression analyses. The association between the expression levels of AVEN and the overall survival (OS) of patients with LUAD were analyzed with the Kaplan-Meier method and log-rank test P-values were calculated. The contingency table (age exclusion) was analyzed by χ² test when satisfying the assumption that the expected count in < 20% of the cells of the analyzed contingency table was ≤ 5, when the expected count violated this assumption, Fisher’s exact test was used to compare the groups; Fisher’s exact test was used to analyze primary therapy outcome. The patients were split into high and low expression groups by the median. Finally, the correlation between AVEN and immune characteristics was analyzed by Spearman’s rank correlation coefficient. All the data in vitro experiments were expressed as means ± SEM from three independent experiments in all the experimental verification part. The statistical analysis was performed by t tests or one-way ANOVA followed by Dunnettʼs multiple comparison tests using GraphPad Prism 9.4 software. P < 0.05 was considered to indicate a statistically significant difference.

Results

Evaluation of AVEN expression in different cancers and normal tissues

To determine the role of AVEN in different tumours, we summarized the expression levels of AVEN in 33 different types of cancer through the TCGA database and the GTEX database. Compared with normal tissues, AVEN was significantly upregulated in 17 tumour tissues, including lung adenocarcinoma, adrenal cortical carcinoma, cholangiocarcinoma, colon adenocarcinoma, diffuse large B-cell lymphoma, glioblastoma multiforme, head and neck squamous cell carcinoma, kidney clear cell carcinoma, kidney papillary cell carcinoma, acute myeloid leukaemia, low-grade glioma, hepatocellular carcinoma, lung squamous cell carcinoma, prostate adenocarcinoma, rectum adenocarcinoma, thyroid cancer, and thymoma. AVEN expression was downregulated in 4 tumour types, namely, invasive breast carcinoma, oesophageal carcinoma, ovarian serous cystadenocarcinoma, and skin cutaneous melanoma (Fig. 1A). We also examined the expression of AVEN in paired tumour tissue and normal tissue from 23 different cancers in the TCGA database. The results revealed that AVEN was highly expressed in 10 of these tumours, including lung adenocarcinoma, cholangiocarcinoma, colon adenocarcinoma, head and neck squamous cell carcinoma, kidney clear cell carcinoma, kidney papillary cell carcinoma, hepatocellular carcinoma, lung squamous cell carcinoma, prostate adenocarcinoma, and thyroid cancer (Fig. 1B). Combining the above expression results for AVEN, our study demonstrated the pattern of AVEN expression across multiple types of cancer.

Fig. 1

The mRNA expression of AVEN in pan-cancer. (A) The mRNA expression of AVEN in 33 tumours in TCGA_GTEx samples. (B) Expression of AVEN in paired samples of 23 tumours in TCGA database.

AVEN expression was upregulated in lung adenocarcinoma tissue compared with normal tissue

We detected the expression level of AVEN in paired LUAD tumour tissue and normal lung tissue (normal = 59 and tumour = 539; p = 3.7e-23; Fig. 2A) and in unpaired tumour tissue and normal lung tissue (normal = 58 and tumour = 58; p = 1.4e-16; Fig. 2B) via the TCGA database. The results indicated that the expression level of AVEN in LUAD tissue was greater than that in normal lung tissue. In addition, we selected two GEO datasets, GSE19188 and GSE10072, to further verify the expression level of AVEN in LUAD. The results also revealed that AVEN expression was significantly greater in LUAD tissues than in normal lung tissues (Fig. 2C,D). Western blotting revealed the upregulation of AVEN expression in LUAD tissues compared with normal lung tissues (Fig. 2E).

Fig. 2

Expression profile and genetic changes of AVEN in LUAD. (A) Differences in AVEN expression levels between tumour tissues and normal tissues in TCGA LUAD database. (B) Expression differences in AVEN between tumour patients and paired adjacent normal samples. (C, D) In the GSE19188 and GSE10072 datasets, the expression level of AVEN in LUAD tissues was higher than that in normal tissues. (E) Representative western blot analysis showing the change of the expression of AVEN in normal tissues and LUAD tissues. N, normal tissues; T, tumours. Original blots/gels are presented in Supplementary Fig. S1. (F) Summary of AVEN changes in the TCGA LUAD dataset. (G) AVEN’s type of mutation in LUAD. (H) The mutation frequency of AVEN in LUAD.

Mutation of AVEN in LUAD

Analysis of the cBioPortal database revealed that 3.91% of LUAD patients had AVEN mutations, with deep deletion being the most common variation (Fig. 2F). Therefore, we further extracted likely mutation types from the COSMIC database. The results revealed that 78.57% of the LUAD samples had missense substitutions and that 10.71% of the LUAD samples had synonymous substitutions (Fig. 2G). Additionally, the base substitutions were mainly G > T (39.13%), C > T (13.04%) and G > A (13.04%) (Fig. 2H).

Correlation analysis between the expression of AVEN and clinical features of lung adenocarcinoma patients

The relationships between AVEN expression and different clinical features of LUAD patients, including T stage, N stage, pathological stage, sex, overall survival (OS) events, disease-specific survival (DSS) events, and progression-free interval (PFI) events, were analysed. High expression of AVEN was significantly correlated with T stage (p < 0.05), N stage (p < 0.01), pathological stage (p < 0.01), sex (p < 0.001), OS events (p < 0.001), DSS events (p < 0.001) and PFI events (p < 0.05) (Fig. 3A–G; Table 1).

Fig. 3

The expression of AVEN in patients with lung adenocarcinoma was evaluated according to different clinical features. (A) Pathologic T stage, (B) Pathologic N stage, (C) Pathologic stage, (D) Gender, (E) OS event and (F) DSS event, (G) PFI event.

Table 1 Relationship between AVEN expression and clinical characteristics in lung adenocarcinoma.

Prognostic and diagnostic value of AVEN in LUAD

To evaluate the relationship between the AVEN expression level and the prognosis of LUAD patients, LUAD patients in the TCGA database were divided into a high-AVEN expression group and a low-AVEN expression group according to the median. Kaplan–Meier survival analysis revealed that high AVEN expression was associated with poor OS (HR = 1.61; p = 0.002), DSS (HR = 1.81; p = 0.002) and PFS (HR = 1.41; p = 0.007) in LUAD patients (Fig. 4A–C). We also validated the relationship between AVEN expression and prognosis in LUAD patients using the external GEO dataset GSE31210, and Kaplan‒Meier survival analysis revealed that patients with increased AVEN expression in the external dataset GSE31210 had shorter OS (HR = 3.87; p = 0.000472) (Fig.  4D). We subsequently analysed the relationship between AVEN expression levels and OS in LUAD patients in each subgroup on the basis of clinical features. High expression of AVEN was significantly correlated with poor OS in LUAD patients with T1-T2 (HR = 1.41; p = 0.038) (Fig. 5A), N0-N1 (HR = 1.66; p = 0.003) (Fig. 5B), pathological stage I-II (HR = 1.54; p = 0.020) (Fig. 5C), smoking status (HR = 1.71; p = 0.002) (Fig. 5D), age ≤ 65 years (HR = 1.66; p = 0.024) (Fig. 5E), and female sex (HR = 1.79; p = 0.008) (Fig. 5F). Univariate Cox regression analysis was conducted on the basis of the clinical characteristics of LUAD patients in the TCGA database, and the results revealed that T2, T3, T4, N1, N2, N3, M1, SD, CR, and high AVEN expression were independent risk factors for poor OS in LUAD patients (Table 2; Fig. 6A). Multivariate Cox regression analysis revealed that N2, N3, SD and CR were independent risk factors for poor OS in LUAD patients (Table 2). An OS prediction model was constructed on the basis of the AVEN expression level, T stage, N stage, M stage, and primary therapy outcome and was visualized with a nomogram (Fig. 6B). Calibration curves were used to evaluate the accuracy of the nomogram (Fig. 6C). An ROC curve was constructed to evaluate the diagnostic value of AVEN expression in LUAD, and the area under the ROC curve (AUC) was 0.893 (Fig. 6D), suggesting that AVEN has high diagnostic value in LUAD.

Fig. 4

Kaplan-Meier survival curves of patients with high and low AVEN expression of LUAD. (A) OS (B) DSS (C) PFI (D) Kaplan-Meier survival curve for progression-free survival in the GSE31210 dataset.

Fig. 5

Kaplan-Meier curve for overall survival in LUAD. Subgroup analysis for patients with (A) T1&T2, (B) N0&N1, (C) Stage Ⅰ& stage Ⅱ, (D) smoker, (E) Age ≤ 65, (F) male patients.

Table 2 Univariate and multivariate cox regression analyses of the clinical characteristics associated with OS in lung adenocarcinoma in the cancer genome atlas.

Fig. 6

ROC curve analysis and nomogram model of AVEN. (A) Cox regression analysis of AVEN in LUAD. (B) Nomogram model, combining clinical variables and AVEN level to predict 1-, 3- and 5-year survival probability. (C) Calibration curve of the nomogram. (D) ROC curve analysis of AVEN expression in diagnosis of LUAD.

Enrichment analysis of the AVEN expression phenotype

We screened a total of 6708 differentially expressed genes (DEGs) (|log FC| > 1, adjusted p value < 0.05) on the basis of high and low expression of AVEN in the LUAD cohort of the TCGA database. There were 503 genes whose expression was high and 6205 genes whose expression was low (Fig. 7A). GO and KEGG enrichment analyses of the 6708 DEGs revealed that 10 related signalling pathways were enriched. As shown in Fig. 7B, the enriched pathways were closely related to keratinocyte differentiation, keratinization, structural constituents of the skin epidermis, small nuclear ribonucleoprotein complex, the spliceosomal snRNP complex, the cornified envelope, pre-mRNA 5’-splice site binding, and spliceosomal tri-snRNP complex assembly. In addition, to further explore the biological function of AVEN in LUAD, the data of 6708 DEGs were subjected to GSEA. The results revealed that the formation of the cornified envelope, keratinization, developmental biology, NABA matrisome-associated, and NABA matrisome were enhanced (Fig. 7C) and that the sensory perception of taste and Class C 3 metabotropic glutamate pheromone receptors were inhibited (Fig. 7D).

Fig. 7

Enrichment analysis and correlation analysis of DEGs function and pathway of AVEN in LUAD. (A) A total of 503 up-regulated and 6205 down-regulated genes were identifed as being statistically signifcant between AVEN high expression and low expression groups. (B) Gene Ontology term and Kyoto Encyclopedia of Genes and Genomes pathway analyses of AVEN-related genes in LUAD. (C, D) GSEA function and pathway enrichment analysis of AVEN-related genes in LUAD. (E) Enrichment network of 21 HUB-related genes. (F) Heat map of co-expression of AVEN and 21 HUB genes in LUAD. (G) Heat map of correlation between AVEN and 21 HUB genes in LUAD. (H) Structure of AVEN.

Analysis of genes associated with AVEN in LUAD

We used the MCODE plugin in Cytoscape software to identify 21 HUB genes associated with AVEN in LUAD and construct a PPI network (Fig. 7E). Then, heatmaps of gene coexpression were constructed (Fig. 7F). Correlation heatmaps of the 21 HUB genes revealed that most of the genes were positively correlated with each other (Fig. 7G). The structure of AVEN was constructed via the UniProt database (Fig. 7H).

Correlation between AVEN expression and immune infiltration

We subsequently investigated the correlation between AVEN expression in LUAD and the tumour immune response. An evaluation of the distribution of immune cells in patients with high and low expression of AVEN revealed that the numbers of helper T cells (Th) 2, Tgd cells, and natural killer CD56dim cells (NK CD56dim cells) were significantly lower in LUAD patients with low AVEN expression than in patients with high expression of AVEN (Fig. 8A–C). In addition, compared with those in patients with low AVEN expression, the numbers of follicular T helper cells (TFH), mast cells, eosinophils, syndactyly cells (iDC), natural killer CD56bright cells (NK CD56bright), and central memory T cells (Tcm) in LUAD patients with high expression of AVEN decreased significantly (Fig. 8D–I).

Fig. 8

Differential distribution of immune cells in patients with high and low AVEN expression. (A) Th2, (B) Tgd, (C) NK CD56dim, (D) TFH, (E) Mast cells, (F) Eosinophils, (G) iDC, (H) NK, (I) CD56bight. Tcm Th, T helper; Tgd, γδ-Τ; NK, natural killer; TFH, T follicular helper; iDC, immature DC; Tcm, central memory T.

Furthermore, we examined the correlation between AVEN expression and the level of immune infiltration in LUAD (Fig. 9A). The expression of AVEN was positively correlated with Th2 cells (ρ = 0.360; p < 0.001) (Fig. 9B), NKCD56dim cells (ρ = 0.224; p < 0.001) (Fig. 9C), Tgd (ρ = 0.211; p < 0.001) (Fig. 9D), neutrophils (ρ = 0.141; p = 0.001) (Fig. 9E) and Th1 cells (ρ = 0.114; p = 0.008) (Fig. 9F). In contrast, the expression of AVEN was negatively correlated with that of TFH cells (ρ = -0.271; p < 0.001) (Fig. 9G), eosinophils (ρ = -0.173; p < 0.001) (Fig. 9H), Tcm (ρ = -0.162; p < 0.001) (Fig. 9I), and mast cells (ρ = -0.174; p < 0.001) (Fig. 9J).

Fig. 9

Relationship between AVEN expression and immune infiltration in LUAD. (A) Association of all 24 immune cell types with AVEN. (B–F) The expression of AVEN was positively correlated with the enrichment of Th2 cells, NKCD56dim cells, Tgd, Neutrophils and Th1 cells. (G–J) The expression of the AVEN and TFH cells, Eosinophils, Tcm, Mast cells concentration showed a negative correlation. LUAD, lung adenocarcinoma; Th, T helper; Tgd, γδ-Τ; NK, natural killer; DC, dendritic cell; Tem, effector memory T; pDC, plasmacytoid DC; Tcm, central memory T; TFH, T follicular helper; iDC, immature DC; aDC, activated DC.

Correlations between AVEN expression and cytokine and immune checkpoint expression

The association between the expression of AVEN and that of multiple cytokines in LUAD revealed that the expression of AVEN was positively correlated with that of IL-6 (ρ = 0.227; p < 0.001; Fig. 10A). The expression of AVEN was negatively correlated with that of IL-2 (ρ = -0.207; p < 0.001; Fig. 10B) and IL-12B (ρ = -0.232; p < 0.001; Fig. 10C). There was no relationship between AVEN and IL-10, IL-12 A or IFN-γ (Fig. 10D–F).

Fig. 10

Relationship between AVEN expression and cytokines and immune checkpoints in LUAD. (A) The expression of AVEN was positively correlated with that of IL-6. (B, C) The expression of AVEN was negatively correlated with that of IL-2 and IL-12B. (D–F) There was no relationship between AVEN and IL-10, IL-12 A and IFN-γ. (G) The results showed that there was no relationship between AVEN and PDCD1. (H) The expression of AVEN was negatively correlated with that of CTLA-4. (I) The expression of AVEN was positively correlated with that of PD-L1.

Programmed cell death 1 (PDCD1), cytotoxic T lymphocyte antigen-4 (CTLA-4) and programmed death ligand 1 (PD-L1) are important immune checkpoints involved in tumour immune escape. The associations between AVEN and PDCD1, CTLA-4 and PD-L1 were further investigated by SSGSE analysis. The results revealed that there was no relationship between AVEN and PDCD1 (Fig. 10G). The expression of AVEN was negatively correlated with that of CTLA-4 (ρ = -0.128; p = 0.003; Fig. 10H). The expression of AVEN was positively correlated with that of PD-L1 (ρ = 0.087; p = 0.044; Fig. 10I).

AVEN is closely related to the tumour microenvironment landscape by single-cell RNA seq

We analysed the immune microenvironment of LUAD using the single-cell sequencing dataset GSE131907. According to the CellMarker database, we annotated a total of 8 cell types: T lymphocytes, myeloid cells, B lymphocytes, epithelial cells, NK cells, mast cells, fibroblasts, and endothelial cells (Fig. 11A). The eight cell types can be further subdivided into 47 cell subtypes (Fig. 11B). Next, we compared the proportions of various types of cells in lung cancer tissue (tLung) and normal lung tissue (nLung), and the results revealed that the relative proportions of T lymphocytes and B lymphocytes significantly increased, while the relative proportions of NK cells and myeloid cells decreased (Fig. 11C,D). A comparison of the proportions of lymphocytes and myeloid cells in lung cancer tissue and normal tissue revealed that the proportions of lymphocytes, such as follicular B cells, exhausted CD8 + T cells, and Tregs, clearly increased in lung cancer tissues (Fig. 11E,F). Compared with that in normal tissues, the proportion of alveolar Macs in cancer tissues was significantly lower, while the proportion of mo-MACs was significantly greater (Fig. 11G,H).

Fig. 11

Immune infiltration model based on single cell RNA sequencing (scRNA-seq) analysis. (A, B) 8 cell types and 47 cell subtypes from 11normal lung tissues and 11 LUAC tissues. (C, D) Ratios of 8 cell types in 11 normal lung tissues and 11 LUAC tissues. (E, F) Ratio of lymphocytes in 11normal lung tissues and 11 LUAC tissues. (G, H) Ratio of myeloid cells in 11normal lung tissues and 11 LUAC tissues. nLung, normal lung tissues; tLung, LUAC tissues.

Next, we analysed the difference in the distribution of AVEN between normal lung tissues and LUAD tissues, which revealed significant differences in the expression of AVEN in epithelial cells (Fig. 12A–C). Analysis of the expression of AVEN in different epithelial cell subtypes revealed that the expression of AVEN in epithelial cells was concentrated mainly in the tS2 cell subtype (Fig. 12D,E). Next, we focused on the analysis of the pathway differences between normal lung tissues and LUAD tissues in terms of cell subtypes, such as follicular B cells, exhausted CD8 + T cells, Tregs, alveolar Mac cells, and mo-MAC cells, and found that in the five cell subtypes, there were obvious differences in the pathways in the different groups (Fig. 13A–E). Correlation analysis of pathways in the seven subtypes of epithelial cells revealed significantly increased levels of glycolysis, mitosis, E2F, and G2M in tS2 cells (Fig. 13F).

Fig. 12

Expression of AVEN based on scRNA-seq analysis. (A, B) Expression of AVEN in 8 cell types. (C) Differentiation of AVEN expression in 8 cell types between normal lung tissues and LUAC tissues. (D, E) Differentiation of AVEN expression in 4 normal pulmonary epithelial cells and 3 lung cancer epithelial cells.

Fig. 13

Corresponding pathways of different cell subtypes (subpopulations) or groups based on scRNA-seq analysis. (A–E) Differentiation of pathways in folicular B cells, exhausted CD8 + T, Treg, alveolar Mac, mo-MAC between normal lung tissues and LUAC tissues. (F) Differentiation of pathways in 4 normal pulmonary epithelial cells and 3 lung cancer epithelial cells. A red color means a higher score, and a blue color means a lower score.

Finally, we performed a pseudotime analysis of the seven epithelial cell subtypes. Pseudotime analysis of the epithelial cells revealed that the tS1, tS2 and tS3 subtypes of lung cancer epithelial cells had different cell states, which were significantly different from those of the four normal lung epithelial cell subtypes. Ciliated and AT2 cells were clearly located at both ends of two of the normal epithelial cell types, and tS2 was clearly located at both ends of the other two lung cancer epithelial cell lines, tS1 and tS3 (Fig. 14A,B). AVEN was expressed mainly in tS2 near the terminal (Fig. 14C). The results of the scRNA-seq analysis revealed that the immune microenvironments of cancer tissue and normal tissue significantly differed. The high expression of AVEN in tS2 indicates that AVEN is distributed mainly in tumour cells with a high degree of malignancy, which is consistent with the previous findings that AVEN expression is higher in LUAD tissues than in normal lung tissues.

Fig. 14

Pseudo-time analysis of different epithelial cells based on scRNA-seq analysis. (A) Pseudotime trajectory of epithelial cells. (B) Pseudotime trajectory of seven epithelial cell subtypes, colored by cell subtype. (C) Expression of AVEN in seven epithelial cells based on pseudo-time analysis.

Effects of AVEN on the biological behaviour of LUAD

Lung adenocarcinoma cells (A549) were cultured and transfected with siRNA targeting AVEN. The qRT‒PCR results revealed that the mRNA level of AVEN in the si-AVEN group was significantly lower than that in the si-NC group after siRNA transfection (Fig. 15A). The proliferation ability of A549 cells significantly decreased after AVEN expression was downregulated (Fig. 15B). The migration and invasion abilities of A549 cells were reduced significantly in the si-AVEN group (Fig. 15C,D). We further detected the apoptosis of A549 cells by flow cytometry. The flow cytometry results revealed that the number of apoptotic cells significantly increased after the knockdown of AVEN (Fig. 15E).

Fig. 15

Effect of silencing AVEN on the biological behavior of LUAD. (A) qRT-PCR to detect the knockdown efficiency of AVEN. (B) Effect of down-regulation of AVEN on cell proliferation ability. (C) Effect of down-regulation of AVEN on cell migration ability. (D) Effect of down-regulation of AVEN on cell invasion ability. (E) Effect of silencing AVEN on cell apoptotic ability detected by flow cytometry. *p < 0.01, **p < 0.01, ***p < 0.001, ***p < 0.0001.

Correlation analysis of the ceRNA targeting of AVEN

Many studies have shown that noncoding RNAs (ncRNAs) can influence and regulate the expression of target genes through several different mechanisms^31,32. A lncRNA‒miRNA‒mRNA competitive endogenous RNA (ceRNA) network plays a key regulatory role in the development of multiple tumour types^33,34.

The ENCORI database was used to predict the upstream miRNAs of AVEN to analyse the ncRNAs that might affect AVEN, and a total of 13 miRNAs were identified. In terms of the mechanism underlying the negative regulation of mRNA at the posttranscriptional level, AVEN and upstream miRNAs should be negatively correlated. Therefore, we used the TCGA-LUAD database to determine the correlation between AVEN and the 13 miRNAs. The results revealed that the expression of AVEN was negatively correlated with that of hsa-miR-30d-5p (ρ = -0.502, p < 0.001), hsa-miR-30a-5p (ρ = -0.245, p < 0.001), hsa-miR-374a-5p (ρ = -0.100, p = 0.024) and hsa-miR-30e-5p (ρ = -0.145, p = 0.001), while the relationships between the expression of AVEN and that of the other miRNAs were not statistically significant in LUAD (Fig.  16A). Next, we detected the expression levels of hsa-miR-30d-5p, hsa-miR-30a-5p, hsa-miR-374a-5p and hsa-miR-30e-5p in LUAD in the TCGA database, and the results revealed that the expression levels of hsa-mir-30d-5p (p = 9.6e-12), hsa-miR-30a-5p (p = 3.2e-13) and hsa-miR-374a-5p (p = 7e-17) in LUAD tissues were lower than those in normal lung tissues. However, the expression level of hsa-miR-30e-5p in LUAD tissues was greater than that in normal lung tissues (Fig.  16B). Finally, we used the TCGA database to determine the correlation between the expression levels of hsa-miR-30d-5p, hsa-miR-30a-5p, hsa-miR-374a-5p and hsa-miR-30e-5p and the OS of LUAD patients. The results suggested that high expression of hsa-miR-30d-5p (p = 0.002) and hsa-miR-30e-5p (p = 0.002) was significantly related to a good prognosis of patients with LUAD (Fig.  16C). On the basis of the results of the above analysis, we believe that the most likely upstream regulatory miRNA of AVEN in LUAD is hsa-miR-30d-5p. RT‒qPCR verified that the level of miR-30d-5p increased significantly after transfection with miR-30d-5p mimics (Fig.  16D), and the western blotting results revealed that the protein expression level of AVEN decreased significantly after transfection with an miR-30d-5p inhibitor (Fig. 16E).

Fig. 16

Expression and prognosis of ceRNA targeting AVEN in LUAD. (A) Correlation between hsa-miR-30d-5p, hsa-miR-30a-5p, hsa-miR-374a-5p and hsa-miR-30e-5p and AVEN in TCGA LUAD database. (B) Expression levels of hsa-miR-30d-5p, hsa-miR-30a-5p, hsa-miR-374a-5p and hsa-miR-30e-5p in LUAD in TCGA database. (C) Correlation between the expression levels of hsa-miR-30d-5p, hsa-miR-30a-5p, hsa-miR-374a-5p and hsa-miR-30e-5p in the TCGA database and OS in LUAD patients. (D) RT-qPCR showed the effect of transfection of miR-30d-5p mimics. (E) Western blotting showed the expression of AVEN after the transfection of miR-30d-5p mimics. *p < 0.01, ***p < 0.0001. (F) ceRNA network of lncRNA/hsa-miR-30d-5p/AVEN. (G) The expression level of AC012236.1 in LUAD tissues was significantly higher than that in normal tissues. (H) High expression of lncRNA-AC012236.1 was significantly associated with poorer OS in LUAD patients.

Potential upstream lncRNAs that may bind to hsa-miR-30d-5p were predicted by the ENCORI database. We ultimately selected 35 potential lncRNAs that bind to hsa-miR-30d-5p and constructed a ceRNA network of lncRNA/hsa-miR-30d-5p/AVEN using Cytoscape software (Fig. 16F). The expression levels of the 35 lncRNAs in LUAD were explored via the TCGA dataset. The results revealed that the expression level of AC012236.1 in LUAD tissues was significantly greater than that in normal tissues (Fig. 16G). The TCGA database was subsequently used to analyse the relationship between the expression level of AC012236.1 and the prognosis of patients with LUAD. High expression of AC012236.1 was significantly correlated with poor OS of patients with LUAD (Fig. 16H). On the basis of the results of the above analysis, we selected lncRNA-AC012236.1 as the most likely upstream lncRNA of the AVEN/hsa-miR-30d-5p axis in LUAD.

Discussion

In this project, the expression of the hub gene AVEN in LUAD was identified by bioinformatics analysis technology and verified in tumour samples, after which AVEN was identified as a molecular diagnostic marker of LUAD by bulk RNA-seq, scRNA-seq and experimental validation. ScRNA-seq analysis further verified the high expression of AVEN in LUAD, which was closely related to the immune microenvironment landscape of LUAD. Furthermore, bioinformatics prediction and validation revealed that hsa-miR-30d-5p was an upstream regulator of AVEN, with its low expression in LUAD tissues inversely correlated with that of AVEN and its high expression being associated with a favourable prognosis in patients with LUAD. Moreover, an upstream lncRNA–AC012236.1–hsa-miR-30d-5p ceRNA network targeting AVEN was constructed to identify more accurate targets for tumour gene therapy.

AVEN is an inhibitor of apoptosis and caspase activation that binds to the antiapoptotic Bcl-2 family member Bcl-xL and the caspase-regulating factor Apaf-1. The synergistic effect of the AVEN protein enhances the antiapoptotic function of Bcl-xL, and in addition, AVEN interferes with the self-binding ability of Apaf-1, suggesting that AVEN independently affects APAF1-mediated caspase activation⁹. In addition to its antiapoptotic function, AVEN has been shown to activate ATM in response to DNA damage³⁵ and to inhibit DNA damage-induced apoptosis in breast cancer cells by stabilizing Bcl-xL protein levels³⁶. In addition, AVEN was found to be overexpressed in patients with acute lymphoblastic leukaemia but was expressed at significantly higher levels in patients who experienced relapse in the standard risk group. These results suggest that AVEN expression can be used to predict the outcome of ALL in children¹⁵. Baranski Z et al. reported that AVEN expression was negatively correlated with metastasis-free survival in osteosarcoma patients and was increased in metastatic tumours compared with primary tumours¹³. In this study, we performed bioinformatics analyses using multiple online databases and scRNA-seq analysis and showed that AVEN expression was closely related to LUAD immune cell infiltration, the cell cycle, and prognosis.

Cell proliferation is strictly regulated by orderly transitions of the cell cycle. Alterations in cell cycle mechanisms are among the main ways in which normal cell proliferation is transformed into uncontrolled cell division and, eventually, carcinogenesis and cancer progression³⁷. Guo JY et al. reported that AVEN acts as an activator and substrate of ATM, thereby regulating the DNA damage response and the G(2)/M cell cycle³⁸. AVEN is also involved in ATR–Chk1 signalling to control and regulate proliferation and chemotherapy resistance in osteosarcoma¹³. In this study, through enrichment analysis of KEGG pathways and GO functional annotation of AVEN-related genes, it was found that AVEN was positively correlated with the cell cycle-related genes UBE2T, CEP55, SKA3, NCAPG and CCNA2. However, no clear AVEN coexpressed gene has been identified in studies of other tumours; therefore, validation of the genes coexpressed with AVEN, especially cell cycle-associated proteins, should be the goal of future work.

The tumour microenvironment (TME) is a complex, evolving network composed of different types of cancer cells, a variety of resident and infiltrating host cells, secreted soluble factors, and extracellular matrix (ECM) proteins, among which host cells include adipocytes and stromal cells, such as cancer-associated fibroblasts (CAFs), tumour endothelial cells (tec), and immunoinflammatory cells. The TME plays important roles in tumour growth, progression, and prognosis³⁹. Immune cells constitute the cellular basis of immunotherapy, which results in a more durable anti-cancer response by enhancing the antitumour response of the immune system compared to conventional treatments. In some highly immune invasive tumours, the percentage of tumour-infiltrating white blood cells can reach more than 40%⁴⁰. So, understanding immune infiltration in the TME is therefore key to improving response rates and developing new cancer immunotherapy strategies. Growing evidence highlights the potential importance of CD4 T helper cells (Th) and B cells in cancer and immunotherapy, with enrichment of TFH cell transcriptional signatures associated with GC B-cell signatures and extended survival in patients with lung adenocarcinoma (LUAD)⁴¹. Xu Z et al. reported that TAGAP affects the differentiation and function of CD4 + T cells in LUAD through the STAT pathway, promoting immune invasion and cytotoxicity⁴².

In this study, immune infiltration analysis revealed that the high expression of AVEN was significantly positively correlated with multiple immune cell subsets involved in immunosuppression (Th2, Tgd, etc.), and negatively correlated with immune cell subsets of anti-tumour immunity (TFH, Tcm, mast cells, etc.). It indicates that the high expression of AVEN may shape an immunosuppressive microenvironment, which is conducive to tumour immune escape^43,44. Correlations analysis of cytokines and immune checkpoints revealed that AVEN was positively correlated with IL-6. IL-6 can promote tumour growth and inflammation⁴⁵. AVEN was negatively correlated with IL-2 and IL-12B. IL-2 and IL-12 usually enhance anti-tumour immunity^46,47. Meanwhile, AVEN was positively correlated with PD-L1. PD-L1 is a key immune checkpoint molecule, and its upregulation can lead to T-cell exhaustion and immune evasion⁴⁸. It indicates that high expression of AVEN may suggest greater sensitivity to PD-1/PD-L1 inhibitors. These results further support the potential of AVEN as a predictor of immunotherapy response.

Normal lung epithelial cells mainly consist of four different subtypes, including alveolar type I (AT1) and type II (AT2), as well as club cells and ciliated cells. The epithelial cells from lung cancer consist of three subtypes, including tS1, tS2, and tS3. The tS1 and tS3 states represent dysregulation of the normal differentiation program, whereas the tS2 tumour cell state completely deviates from the normal transcriptional program. Moreover, tS2-related genes exhibit clear tumour-oriented features, such as aggressive cell movement and abnormal proliferation or apoptosis⁴⁹. The results of our analysis are consistent with those in the literature. The high expression of AVEN in the malignant epithelial subtype of tS2 is associated with the immune evasion mechanism. Single-cell RNA sequencing analysis highlighted the specific expression of AVEN in the tS2 epithelial cell subtype, which exhibits malignant characteristics in LUAD, with significantly enhanced glycolytic, mitosis, E2F and G2M pathways in this subtype. These pathways are closely related to cell proliferation, metabolic reprogramming, and cell cycle progression, and are commonly found in highly malignant tumour cells^38,50. It indicates that AVEN may promote the growth and metastasis of tS2 subtype tumours by driving these processes. The high expression of AVEN in the tS2 subtype is directly related to changes in the immune microenvironment. The high expression of AVEN is positively correlated with PD-L1, which may lead to T cell function inhibition and immune tolerance. Meanwhile, AVEN is positively correlated with Th2 cells, further weakening anti-tumour immunity. This microenvironment is conducive to immune evasion of tumour cells, suggesting that AVEN may serve as a driver of immune evasion in the tS2 subtype. Therefore, targeting AVEN or combining it with immune checkpoint inhibitors (such as anti-PD-L1 therapy) may reverse immunosuppression and improve therapeutic outcomes.

MicroRNAs are involved in almost all carcinogenic processes as negative regulators of genes involved in cellular pathways⁵¹. LncRNAs are defined as RNAs that are longer than 200 nucleotides without protein-coding capabilities and they can be involved in cell differentiation and growth processes as well as the pathogenesis of tumours, including lung cancer⁵². LncRNAs are thought to interact with miRNAs as “sponges” or ceRNAs, weakening the inhibition of mRNA activity by miRNAs⁵³. In this study, we used the ENCORI database to predict the upstream ceRNAs of AVEN and explored the correlation between ceRNAs and AVEN, as well as the prognostic correlation of ceRNAs in LUAD. LncRNA‒miRNA‒mRNA axes play key roles in the occurrence and development of cancer. A ceRNA axis can regulate the PI3K/mTOR pathway to induce epithelial–mesenchymal transition and radiation resistance⁵⁴ and participate in metastasis related to chemotherapy resistance in NLUD through Wnt/β-catenin and TGF-β signalling⁵⁵ and ferroptosis in LUAD⁵⁶. A ceRNA network is involved in immune cell infiltration by influencing and/or interacting with macrophages in LUAD^57,58 and participates in the type I IFN pathway and antitumour immunity in colorectal cancer⁵⁹.

In this study, high expression of hsa-miR-30d-5p was significantly correlated with a good prognosis in patients with LUAD and it was found to regulate the expression of AVEN in LUAD, which was consistent with the findings of other studies. MiR-30d-5p is significantly downregulated in lung squamous carcinoma cells and inhibits tumour cell growth and movement by targeting cyclin E2 in non-small cell lung cancer⁶⁰. In addition, miR-30d-5p is involved in sepsis-related acute lung injury, indicating that miR-30d-5p is closely related to the lung⁶¹. In addition, miR-30d-5p plays multiple roles and is broadly applicable as a biomarker and therapeutic target in human cancer. It can act either as an inhibitor hindering the development of tumours or as a promoter accelerating the occurrence of tumours. The role of miR-30d-5p in tumorigenesis is noteworthy⁶². After constructing a ceRNA network of lncRNA/hsa-miR-30d-5p/AVEN, the expression level of lncRNA-AC012236.1 in LUAD tissues was found to be significantly greater than that in normal tissues. LncRNA-AC012236.1 was identified as a novel prognostic immune-related lncRNA, whose biological function and specific roles in cancer have not been investigated widely and deeply. The potential function of lncRNA-AC012236.1 was enriched in regulation of humoral immune response biological process and involved in JAK-STAT signaling pathway in clear cell renal cell carcinoma⁶³. At present, a large number of lncRNAs have been reported in LUAD, such as MALAT1, HOTAIR, XIST, etc⁶⁴. However, the role of lncRNA-AC012236.1 in LUAD has never been clarified. The bioinformatics analysis showed that the high expression of lncRNA-AC012236.1 was significantly correlated with poor OS in LUAD patients in this study, which was consistent with the reported role of lncRNA-AC012236.1.

Although bioinformatics prediction models provide important tools for prognosis assessment, molecular typing and personalized treatment, their specificity and effectiveness in biology have not been fully verified. Corresponding functional verification is indispensable. In this article, while bulk and scRNA-seq analyses were performed, western blot assays for the expression of AVEN in clinical samples, CCK-8 assays for proliferation, transwell assays for migration/invasion, flow cytometry for apoptosis analysis and RT‒qPCR assays for the expression of miRNAs in vitro experiments were also performed. Immune cell functional experiments to evaluate the impact of AVEN on the activation of T cells or NK cells will be conducted in future work. In future experiments, a luciferase dual-reporter assay will be used to provide the most direct experimental evidence validate whether lncRNA-AC012236.1 adsorbs hsa-miR-30d-5p to construct ceRNA network to regulate AVEN expression. We are already planning to construct reporter gene plasmids containing wild-type and mutant lncRNA-AC012236.1 binding sites. Besides, we will focus on exploring the functions of AVEN in vivo, including the effects of AVEN on tumour progression and immune microenvironment alterations and the efficacy of combined treatment with AVEN and checkpoint inhibitors.

This study for the first time revealed and experimentally verified a brand-new competitive endogenous RNA regulatory axis in LUAD, lncRNA-AC012236.1/miR-30d-5p/AVEN. Although lncRNA-AC012236.1 was mentioned as an immune-related lncRNA in clear cell renal cell carcinoma, its role as a “molecular sponge” to adsorb miR-30d-5p and thereby regulate the complete pathway of the apoptosis suppressor gene AVEN has been reported for the first time in LUAD and other cancers. The discovery of this axis provides a new perspective for understanding the pathogenesis of LUAD. High expression of lncRNA-AC012236.1, low expression of miR-30d-5p and high expression of AVEN together constitute a molecular tag strongly associated with poor prognosis. We should consider developing multi-gene combined detection kits to facilitate its clinical applications in the future. At the same time, based on the above mechanisms, we propose multiple treatment strategies. First targeting this axis, we consider using lncRNA-AC012236.1-specific siRNA delivered by lipid nanoparticles or antisense oligonucleotide (ASO) to directly silence lncRNA-AC012236.1. thereby to relieve its inhibition on miR-30d-5p. Alternatively, by administering the synthesized miR-30d-5p mimics, its tumour suppressive function could be restored, thereby simultaneously inhibiting AVEN and other potential oncogenic targets. Develop novel AVEN small molecule inhibitors that directly target the final effector proteins of this axis to induce tumour cell death. Further, consider combined treatment with chemotherapy or immunotherapy. The combination of miR-30d-5p mimics or ASO-lncRNA-AC012236.1 with platinum-based drugs could sensitize chemotherapy by down-regulating AVEN and overcome apoptotic drug resistance. Combining drugs targeting this axis with anti-PD-1/PD-L1 antibodies could reverse the tumour immunosuppressive microenvironment, transforming “cold” tumours into “hot” ones, thereby enhancing the efficacy of immunotherapy.

In summary, bulk and scRNA-seq analyses verified the high expression of AVEN in LUAD, which was closely related to the immune microenvironment landscape of LUAD. It was further concluded that lncRNA-AC012236.1/hsa-miR-30d-5p ceRNA-AVEN had potential as a new prognostic biomarker and therapeutic target for LUAD patients.