Ferroptosis-related gene signature for predicting prognosis and identifying potential therapeutic drug in EGFR wild-type lung adenocarcinoma

Zhang, Chuankai; Wu, Qi; Yang, Hongwei; Zhang, Hui; Liu, Changqing; Yang, Bo; Hu, Qingsong

doi:10.1038/s42003-024-07117-1

Download PDF

Article
Open access
Published: 30 October 2024

Ferroptosis-related gene signature for predicting prognosis and identifying potential therapeutic drug in EGFR wild-type lung adenocarcinoma

Communications Biology volume 7, Article number: 1416 (2024) Cite this article

3585 Accesses
2 Citations
1 Altmetric
Metrics details

Subjects

Abstract

Epidermal growth factor receptor wild type lung adenocarcinoma (EGFR^WT LUAD) still has limited treatment options and unsatisfactory clinical outcomes. Ferroptosis, as a form of cell death, has been reported to play a dual role in regulating tumor cell survival. In this study, we constructed a 3-ferroptosis-gene signature, FeSig, and verified its accuracy and efficacy in predicting EGFR^WT LUAD prognosis at both the RNA and protein levels. Patients with higher FeSig scores were found to have worse clinical outcomes. Additionally, we explored the relationship between FeSig and tumor microenvironment, revealing that enhanced interactions between fibroblasts and tumor cells in FeSig^high patients causing tumor resistance to ferroptosis. To address this challenge, we screened potential drugs from NCI-60 (The US National Cancer Institute 60 human tumour cell line anticancer drug screen) and Connectivity map database, ultimately identifying 6-mercatopurine (6-MP) as a promising candidate. Both in vitro and in vivo experiments demonstrated its efficacy in treating FeSig^high EGFR^WT LUAD tumor models. In summary, we develop a novel FeSig for predicting prognosis and guiding drug application.

Knowledge-based mechanistic modeling accurately predicts disease progression with gefitinib in EGFR-mutant lung adenocarcinoma

Article Open access 31 July 2023

Anti-EGFR aptamer exhibits direct anti-cancer effects in NSCLC cells harboring EGFR L858R mutations

Article Open access 21 November 2024

Targeting ferroptosis: a promising approach for treating lung carcinoma

Article Open access 29 January 2025

Introduction

With an estimated 235,760 new cases and 131,880 deaths in the United States in 2021, lung cancer has become the most common cancer and the leading cause of cancer-related deaths worldwide¹. Based on histopathological morphology, lung cancer can be classified into small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC), with NSCLC accounting for more than 80% of cases. According to the 2015 World Health Organization Lung Tumor Classification Guidelines, NSCLC comprises lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and large cell carcinoma (LCC), with LUAD being the most common subtype, accounting for approximately 40–45% of cases^2,3,4. Of note, the morbidity and mortality of LUAD are among the highest, making it a serious public health problem.

Epidermal growth factor receptor (EGFR) is a tyrosine kinase receptor that is associated with cell proliferation, migration, and apoptosis, and EGFR mutations are major drivers of NSCLC⁵. Approximately 15-30% of LUAD cases possess EGFR mutations, in which L858R missense mutations and exon 19 deletions accounting for 85% of EGFR-activating mutations. EGFR-targeted tyrosine kinase inhibitors (TKIs) have proven to be highly effective treatments for LUAD harboring the oncogenic EGFR mutation⁶. The successful use of EGFR-TKIs has shifted cancer treatment from empirical cytotoxic chemotherapy to molecularly targeted therapies. Nevertheless, TKIs treatments have limited efficacy in patients with EGFR^WT LUAD. Therefore, there is an urgent need to investigate new strategies to combat EGFR^WT LUAD.

Ferroptosis is an iron-dependent programmed cell death caused by massive iron accumulation and lipid peroxidation⁷. In recent years, the induction of ferroptosis has become a promising cancer treatment approach, which triggers the death of cancer cells, particularly in malignancies resistant to conventional therapies^8,9. Accumulating evidence suggests that activating ferroptosis can facilitate anti-tumor treatments for various tumors^10,11. For instance, erastin enhances the therapeutic effect of cisplatin in NSCLC by inducing ferroptosis, offering a new strategy for treating drug-resistant tumors¹². Combined treatment with erastin and acetaminophen offers a new strategy for ferroptosis induction in lung cancer¹³. Consequently, studying the expression of ferroptosis-related genes in NSCLC and their prognostic value could lead to new strategies for the treatment of NSCLC. However, a comprehensive understanding of the role of ferroptosis-related genes in EGFR^WT LUAD prognosis is still deficient, and the role of ferroptosis in EGFR^WT LUAD remains largely unexplored.

In this study, we developed a 3-gene ferroptosis-related prognostic signature by analyzing RNA-seq data and corresponding clinical information of EGFR^WT LUAD patients from TCGA (The Cancer Genome Atlas) database. This 3-gene model was further validated in 3 independent EGFR^WT LUAD cohorts. Patients were classified into high-risk and low-risk groups based on this model. We also validated the protein expression profiles of the core genes using clinical tissue microarrays. By integrating the FeSig model and single-cell sequencing data, we obtained a deeper understanding of tumor microenvironment (TME) of high-risk patients, which may assist future TME-targeted drug investigations and developments. Notably, to enhance the clinical applicability of the FeSig model, we screened 6-Mercaptopurine (6-MP) as a promising drug for treating high-risk patients with EGFR^WT LUAD.

Results

Identification of ferroptosis-associated genes as a prognostic feature of EGFR^WT lung adenocarcinoma

The workflow for identifying ferroptosis-associated genes and constructing signature is delineated in Fig. 1a. We collected ferroptosis-related genes from ferroptosis database (FerroDb)¹⁴, while gene expression matrix and clinical traits of EGFR^WT LUAD patients were extracted from The Cancer Genome Atlas (TCGA) lung adenocarcinoma database. Firstly, univariate cox analysis was performed, identifying 32 prognostic genes (Supplementary Fig. 1a). Subsequently, 11 prognostic genes were further refined by multivariate cox analysis, including 6 risk genes (PARK7, PRKCA, DUOX1, NOX4, DDIT4, GCLC) and 5 protective genes (SNCA, GDF15, PEBP1, HERPUD1, MFN2) (Fig. 1b). DDIT4, NOX4 and GCLC were significantly upregulated in tumor tissue (LogFC > 1 and p.adj < 0.05) and correlated with poor clinical outcomes (Supplementary Fig. 1b–d). After excluding protective genes, we established a 3-gene-based ferroptosis-related risk signature (FeSig) (Fig. 1c). The single-sample Gene Set Enrichment Analysis (ssGSEA)¹⁵ algorithm was applied to EGFR^WT LUAD patients to obtain FeSig scores, categorizing patients into two groups according to the median score of FeSig (FeSig^high: n = 143, FeSig^low: n = 143). To evaluate FeSig’s predictive capabilities of survival among TCGA EGFR^WT LUAD patients, Kaplan-Meier survival analysis, time-dependent receiver operating characteristic curve (ROC) and nomogram analysis were conducted. Results showed that FeSig^high patients had significantly shorter overall survival (OS) time than FeSig^low patients (Fig. 1d, p = 0.0048). In addition, the FeSig model can also predict survival across different tumor stages, with death events increasing as FeSig scores increased (Supplementary Fig. 1e–h). The area under curve (AUC) values of time-dependent ROC for OS were 0.71(1-year survival), 0.63(3-year survival) and 0.62(5-year survival) (Fig. 1e). Nomograms of 1 year, 3 years, and 5 years survival and corresponding calibration curves were constructed (Fig. 1f). Calibration curve indicated the favorable predictive accuracy of the nomogram model (Fig. 1g). To further validate predictive performance of FeSig, the FeSig was applied to 3 additional EGFR^WT LUAD validation cohorts: GSE31210¹⁶ (n = 68), oncoSG¹⁷ (n = 73) and GSE11969¹⁸ (n = 58, Fig. 1h–j). FeSig performed robustly in predicting survival across these 3 datasets (Supplementary Fig. 2a–f). Collectively, these results support the FeSig model’s ability to predict the prognosis of EGFR^WT LUAD patients.

**Fig. 1: Construction of FeSig model and assessment of predictive capacity.**

Functional investigations of FeSig genes

The impacts of FeSig genes on lung adenocarcinoma cell lines were evaluated by knocking down DDIT4, NOX4 and GCLC in A549, H441 and H1299, respectively (Fig. 2a, Supplementary Fig. 3a). Cell viabilities of these 3 cell lines were significantly impeded after the knockdown of these FeSig genes (Fig. 2b, Supplementary Fig. 3d). Clone formation assays also confirmed that the FeSig genes are closely associated with cell proliferation, as the number of colonies in the knockdown group was reduced compared to the control group (Fig. 2c, d, Supplementary Fig. 3b, f). Cell scratch assay was performed to assess the cell mobilities and scratch width was measured at 0 h and 36 h after treatment. We observed that after knocking down FeSig genes in A549, H441 and H1299, all 3 cell lines exhibited slower healing capabilities compared to the control group (Fig. 2e, f, Supplementary Fig. 3c, e).

**Fig. 2: Functional validation of FeSig genes.**

Since FeSig genes were initially screened based on their ferroptosis-related functions, we subsequently evaluated the alteration in ferroptosis sensitivity of A549 and H1299 after silencing FeSig genes. Results suggested that in the FeSig-gene silenced group, the percentage of ferroptotic cells were significantly increased after treatment with ferroptosis inducer RSL3 (Fig. 2g, Supplementary Fig. 4a, b). All these findings confirmed that DDIT4, NOX4 and GCLC act as ferroptosis inhibitors, suggesting higher expression of these genes inhibits ferroptosis in cancer cell, leading to unsatisfactory clinical outcomes.

Protein level validation of FeSig on EGFR^WT lung adenocarcinoma microarray

Given the imperfect alignment between mRNA expression and protein levels¹⁹, we utilized a lung adenocarcinoma tissue microarray to validate survival states with FeSig at protein level (Fig. 3a). The mean fluorescence intensity of DDIT4 and GCLC was calculated by randomly selecting 5 views per sample. Samples were evaluated and categorized into FeSig^high and FeSig^low groups based on the median FeSig score. After excluding EGFR^MUT patients, we obtained a cohort containing 61 EGFR^WT LUAD patients (Supplementary Data). As expected, combining DDIT4 and GCLC fluorescence and clinical information, patients who fell into the FeSig^high group have significantly upregulated DDIT4 and GCLC protein expression (Fig. 3b–e). Meanwhile, a higher FeSig score predicted worse clinical outcomes (Fig. 3f), which is consistent with previous results derived from mRNA levels. We next combined the clinical traits of these 61 EGFR^WT LUAD patients with FeSig score. Notably, the FeSig score increased with the progression of the tumor stage (Fig. 3g). Consequently, the FeSig score demonstrates efficacy at both mRNA and protein levels, emphasizing the promising potential for clinical applications.

**Fig. 3: Protein level validation of FeSig model.**

FeSig closely related with fibroblasts infiltration in tumor microenvironment

To profile the relationship between TME and FeSig score, we leveraged immune and stromal cell feature genes derived from single-cell RNA sequencing data²⁰. Each cell type was scored using ssGSEA algorithm, and cell type scores were then correlated with FeSig score to decode TME composition differences between FeSig^high and FeSig^low patients (Fig. 4a). Notably, across all 4 independent datasets, fibroblasts exhibited significant positive correlation with FeSig score, and the expression of fibroblast feature genes such as COL1A1, COL1A2, LUM were upregulated with the increasing of FeSig score, underscoring fibroblasts may be critical in modeling FeSig^high TME or vice versa (Fig. 4a, b). To validate this phenomenon, we checked hematoxylin-eosin (HE) slides from the TCGA cohort (Fig. 4c), including 4 patients in the FeSig^high group and another 4 patients in the FeSig^low group. Interestingly, FeSig^high patients displayed abundant fibroblasts infiltration in their TME (Fig. 4c), while the TME in FeSig^low patients primarily consisted of tumor cells and microvessels (Fig. 4c). We also observed similar TME composition differences in our LUAD microarray samples: FeSig^high samples showing more α-SMA area compared to FeSig^low samples (Fig. 4d, e). Altogether, these data highlight that fibroblasts may contribute to ferroptosis regulation in EGFR^WT LUAD TME, potentially impacting the clinical outcomes of LUAD patients.

**Fig. 4: TME composition features of FeSig^high and FeSig^low cohort.**

Upregulation of fibroblast-cancer cell interaction in FeSig^high TME

To further elucidate relationships between ferroptosis state and fibroblasts, we queried a lung adenocarcinoma single-cell sequencing data from Gene Expression Omnibus (GEO) database, GSE131907²⁰, to explore the interaction between fibroblasts and tumor cells under different FeSig conditions at single-cell resolution. After dimensionality reduction using Uniform Manifold Approximation and Projection (UMAP) and combining with marker-based cell type identification and annotation, we grouped 33632 cells into 9 clusters, including endothelial (n = 257), fibroblast (n = 663), mast cells (n = 680), plasma cells (n = 1705), NK cells (n = 1786), T cells (n = 11,380), epithelial (n = 9824), B cells (n = 2040) and myeloid cells (n = 5297) (Fig. 5a, b, Supplementary Fig. 5a, b). FeSig scores were applied to each patient using pseudobulk algorithm from muscat package. As expected, an enrichment of fibroblasts in FeSig^high patient was observed (2.85% in FeSig^high group, 0.53% in FeSig^low group), consistent with our results mentioned above in bulk RNA-seq data (Fig. 5c). We then performed CellChat analysis on this single-cell data and verified enhanced fibroblast-cancer cell interaction than other cell types, and this interaction in FeSig^high group was greater than that in FeSig^low group (Fig. 5d). To find out potential mechanisms by which fibroblasts regulate ferroptosis status of EGFR^WT LUAD cells, we performed ligand-target analysis with Nichenetr package (Fig. 5e, top 80 targets were shown). Gene Ontology (GO) analysis of target genes enriched in insulin-like growth factor (IGF) binding-related pathways (Supplementary Fig. 5c). Interestingly, when examining specific Cellchat pathways in FeSig^high and FeSig^low groups, the IGF pathway was also identified in the FeSig^high cohort (Supplementary Fig. 5d). 4 independent datasets consistently demonstrated that patients with higher FeSig score exhibited higher IGF binding score (Supplementary Fig. 5e), indicating insulin-like growth factor binding-related pathways may contribute to tumor cell resistance to ferroptosis. To validate the correlation of these predicted ligand-receptor interactions with ferroptosis, we constructed a ligand-receptor-target (L-R-T) signature based on Nichenetr outputs. Results showed that this L-R-T signature is negatively correlated with ferroptosis state (TCGA-EGFR^WT-LUAD: Rs = −0.293, p < 0.0001; GSE31210: Rs = −0.405, p = 0.0006; oncoSG: Rs = −0.296, p = 0.0109. Figure 5f–i), except dataset GSE11969 (Rs = −0.033, p = 0.803. Figure 5i). In conclusion, the interaction between fibroblasts and tumor epithelial cells in EGFR^WT LUAD TME appears to regulate the ferroptosis state, stronger interaction and communication between fibroblasts and tumor epithelial cells leading to resistance to ferroptosis in tumor cells.

**Fig. 5: Interaction of fibroblasts and tumor cells at single-cell resolution.**

Drug screening and testing based on FeSig

Given the significant difference in prognosis between FeSig^high and FeSig^low patients, we combined both the NCI-60 database (NCI-60 Human Tumor Cell Lines Screen) and the Connectivity Map database to identify compounds that could potentially improve the prognosis of FeSig^high patients. The NCI-60 cell line expression profile was acquired, and the FeSig score for each cell line was calculated using the ssGSEA algorithm. The FeSig scores were then correlated with half-maximal inhibitory concentration (IC50) of compounds, which finally resulted 17 compounds, including 4 sensitive drugs and 13 resistant drugs (Fig. 6a). To further refine the list of potential compounds for FeSig^high patients, top 150 overexpressed genes in FeSig^high patients of TCGA-EGFR^WT-LUAD were inputted to Connectivity map database to screen a secondary drug set. By intersecting the NCI-60 sensitive drugs with the Connectivity map drug sets, we identified 6-Mercaptopurine (6-MP), a purine antagonist, as a promising candidate (Fig. 6b). The effects of 6-MP on lung adenocarcinoma cell lines were subsequently tested both in vitro and in vivo.

**Fig. 6: Drug screening and validation based on FeSig.**

We first pulled lung adenocarcinoma cell line RNA-sequencing data from Genomics of Drug sensitivity in Cancer (GDSC) database and profiled FeSig score for each lung adenocarcinoma cell line. We then selected 2 cell lines, H3122 (FeSig = 0.993, FeSig^low), A549 (FeSig = 1.407, FeSig^high), to assess the impact of 6-MP on cancer cell elimination. In vitro, clone formation assay demonstrated the efficacy of 6-MP in killing cancer cells increased with higher compound concentration (Fig. 6c, d). Notably, the cell line with a higher FeSig score (A549) displayed greater sensitivity to 6-MP compared to those with a lower FeSig score cell line (H3122) (Fig. 6c, d). To assess the in vivo tumor elimination capacities, we inoculated A549 and H3122 cancer cell lines subcutaneously into 6-week-old Balb/c nude mice and intraperitoneally administrated 6-MP or control PBS (Phosphate Buffered Saline), respectively. Results showed that 6-MP had limited effects on H3122 tumors (FeSig^low), while A549 tumors (FeSig^high) were more sensitive to 6-MP, consistent with in vitro colony formation assay (Fig. 6e, f).

Given the critical role of fibroblasts in TME modeling, we quantified fibroblasts infiltration alteration with or without 6-MP treatment using immunohistochemistry (IHC) (Fig. 6g). Surprisingly, 6-MP treatment significantly reduced fibroblast infiltration in A549 tumor models (sensitive to 6-MP), whereas H3122 tumor models (resistant to 6-MP) did not show any changes (Fig. 6g, Supplementary Fig. 6a). These findings suggested that 6-MP not only directly targets tumor cells, but also remodels the TME to improve chemotherapy outcomes in FeSig^high patients, indicating combining drug administration with 6-MP may offer significant benefits for FeSig^high EGFR^WT LUAD patients.

Discussion

Ferroptosis, as a non-apoptotic and iron-dependent form of cell death, has been proven to play a regulatory role in carcinogenesis and cancer progression across various cancer types^{21,22,23,24,25}. Meanwhile, ferroptosis is closely related to outcomes of clinical treatment of cancer, such as radiotherapy^26,27, chemotherapy^28,29, and immunotherapy^30,31,32. Unlike EGFR^MUT LUAD, the prognosis of EGFR^WT LUAD patients is unsatisfactory due to the absent of targetable genes. To address this issue, many efforts have been paid to characterize specific subtypes of EGFR^WT LUAD^33,34,35. In this study, we integrated transcriptomic and ferroptotic features to identify potential ferroptosis-related subtypes of EGFR^WT LUAD patients. Through univariate and multivariate Cox regression analysis, we got 11 risk-associated and 5 protective ferroptosis genes. To further refine molecule signature, we focused on overexpressed ferroptosis genes in EGFR^WT LUAD and intersected them with the 11 risk genes, resulting in a three-gene signature: DDIT4, NOX4 and GCLC. This molecular signature, termed FeSig, was constructed based on these three genes. Survival analysis, time-dependent ROC of TCGA data and another three validation cohorts suggested FeSig can precisely predict the prognosis of EGFR^WT LUAD patients (Fig. 7).

DDIT4 (DNA Damage Inducible Transcript 4) is known to be overexpressed in various of solid tumors, including head and neck squamous carcinoma³⁶, triple-negative breast cancer³⁷, colon cancer³⁸, and is related to unsatisfactory clinical prognosis. Overexpression of DDIT4 in lung adenocarcinoma has been reported^39,40 and the silence of DDIT4 enhances apoptosis and sensitizes lung adenocarcinoma cell lines, H358 and H838, to chemotherapies like ispinesib and cabazitaxel³⁹. In addition, DDIT4 was included in many prognostic gene signatures for LUAD^41,42,43, further confirming its role in regulating tumor growth, proliferation and predicting the prognosis of lung cancer patients. However, the detailed mechanisms by which DDIT4 regulates ferroptosis of EGFR^WT-LUAD cancer cells still stay in opaque.

NOX4 (NADPH Oxidase 4) belongs to NOX family which functions as superoxide or hydrogen peroxide producer. Research has confirmed that NOX family members are closely related to cancer progression⁴⁴. NOX4 has been implicated in resistance to EGFR-TKI therapy in NSCLC via enhancing IL-8/PD-L1 pathways⁴⁵. Fan et al.⁴⁶ demonstrated that EFHD2 induces the expression of NOX4 and through the EFHD2-NOX4-ROS-ABCC1 axis, promotes drug efflux, leading to chemotherapy resistance in non-small cell lung cancer. Chen et al.⁴⁷ identified Fangchinoline (Fan) as a compound that reduces NSCLC metastasis by directly targeting NOX4, lowering cytoplasm ROS levels, and inhibiting the Akt-mTOR pathway, highlighting NOX4 as a promising therapeutic target in NSCLC. Besides affecting cancer cells directly, NOX4 also influences the TME by altering macrophages and fibroblast phenotypes. NOX4-derived ROS in the TME induces M2 macrophage polarization through ROS/PI3K signaling to facilitate cancer cell progression and immune evasion⁴⁸. Cancer-associated fibroblasts (CAFs), as a component of TME, maintain an immune desert microenvironment characterized by the exclusion of CD8⁺ T cells through the expression of NOX4. Silencing NOX4 in CAFs remodels CAF phenotype and TME, potentially enhancing the effectiveness of immunotherapy⁴⁹.

GCLC (Glutamate-cysteine ligase catalytic subunit) has been demonstrated to have an anti-ferroptosis function, which promotes cancer cell survival and progression in gastric cancer⁵⁰, non-small-cell lung cancer⁵¹, and hepatocellular carcinoma⁵². GCLC regulates the synthesis of glutathione (GSH), which protects cancer cells from oxidative stress-induced ferroptosis⁵⁰. Apart from the classic way of ferroptosis inhibition, Kang et al. discovered GCLC synthesize γ-glutamyl-peptide in a GSH-independent manner, further preventing ferroptosis in NSCLC⁵¹. Epigenetically⁵³, Kras^LSL-G12D/+ mouse NSCLC models demonstrated that histone variant H1.2 stabilizes nuclear NRF2 levels and enhances NRF2 binding to GCLC promoter, leading to upregulated GCLC expression in tumor cells. Depletion of H1.2 significantly reduced GCLC expression and induced oxidative stress, resulting in NSCLC regression and increased sensitivity to cisplatin. In our gene functional assay, all three FeSig exhibited anti-ferroptosis properties, and knocking down these genes significantly increased the sensitivity to ferroptosis inducer. By staining FeSig genes using IHC or mIF (multiple color immunofluorescence) and quantifying mean fluorescence intensities, we could profile the ferroptosis state of each patient, allowing for tailored clinical treatment strategies that could lead to a better prognosis.

Resistance to immunotherapies poses challenges to current immunotherapeutic strategies. Tumor patients exhibit varying responses to immunotherapy. TME of non-responders often enriched in exhausted T cells (Tex)^54,55,56 and regulatory T cells (Treg), which are commonly targeted by mainstream immunotherapies^57,58. Additionally, some non-responders exhibit a lack of immune effector cells within TME, which is defined as an immune-desert TME^59,60,61. In such cases, even if T cells are targeted, they remain ineffective due to their exclusion from the tumor site. This type of TME is often enriched with tumor-associated fibroblasts. Liu et al. reported such cancer-fibroblast niche in hepatocellular carcinoma, where CAFs and SPP1 macrophages together form a barrier-like structure surrounding the tumor border, preventing the infiltration of T cells and other immune cells, ultimately leading to the failure of immunotherapy⁶².

Here, we utilized α-SMA to label fibroblasts. α-SMA, encoded by ACTA2, was reported as a marker of cancer-associated fibroblasts (CAFs)^63,64, and ACTA2⁺ fibroblasts have been reported closely related to worse prognosis of gastric cancer⁶⁵ and colorectal cancer⁶⁴. However, Hui Ding et al reported that alveolar epithelial cells could also express α-SMA under the induction of TGF-β⁶⁶. In the tumor microenvironment, TGF-β is expressed and secreted by multiple cell types, such as fibroblast, macrophage and tumor cell itself, which may lead heterologous expression of α-SMA out of fibroblasts. Also, α-SMA can be expressed by endothelial cells which undergo endothelial-mesenchymal transition (EndoMT) process under systematic sclerosis⁶⁷, but whether such transition occurs in tumor microenvironment still unknown which needs further investigation and validation. In our analysis, FeSig^high patients characterized a phenotype where a CAF-barrier surrounds tumor regions, dividing the tumor into nest-like structure, which may impede T cell infiltration. Therefore, patients with this phenotype might benefit from CAF-targeting therapies.

To apply our FeSig feature to clinical practice, we screened FDA (Food and Drug Administration)-approved compounds to identify those that could benefit FeSig^high patients, with the goal of improving their survival. By integrating NCI-60 database and Connectivity Map compound library, we identified the purine antagonist 6-MP as a potential candidate. Both in vitro and in vivo experiments verified its beneficial effects on EGFR^WT LUAD. However, our findings have certain limitations. 6-MP showed partial tumor elimination capacities in in vitro experiments. Notably, under the same concentration of 6-MP, the OD595 values of A549 were significantly lower than those of H3122. This suggests that A549 (FeSig^high) was more sensitive to 6-MP compared to H3122 (FeSig^low). The limited effects on FeSig^low in in vivo tumor models are likely due to the complex tumor microenvironments in vivo. Unlike in vitro culture systems, where tumor cells are directly exposed to the drug, in vivo tumor cells interact with multiple cell types, such as fibroblasts, macrophages, and immune cells. Cytokines, secretory proteins, and ligand-receptor interactions all contribute to complicated intracellular signaling pathways in tumor cells, affecting their response to drug treatment. Thus, in vitro experiments sometimes cannot perfectly simulate in vivo results.

The role of 6-MP as a purine antagonist in regulating tumor ferroptosis and affecting fibroblasts infiltration in TME remains unclear. Further experimental evidence is needed to determine whether 6-MP could enhance immunotherapeutic effects. In addition, the mechanisms by which interactions between fibroblasts and EGFR^WT LUAD cancer cells lead to resistance to ferroptosis require further investigation and explanation. Potential mechanisms involving IGF and associated pathways in regulating ferroptosis need to be explored, and whether IGF could serve as a promising target for the treatment of EGFR^WT LUAD also warrants detailed studies.

In summary, our study identified a 3-ferroptosis-inhibitor-gene signature, FeSig, which can be used to predict clinical prognosis at both mRNA and protein levels. FeSig can be conveniently applied to clinical use through techniques such as immunohistochemistry (IHC) or fluorescence in situ hybridization (FISH). Besides, FeSig can also characterize a fibroblast-enriched TME, aiding in the evaluation of immunotherapeutic responses in specific patients. Of note, we screened the purine antagonist, 6-MP, which shows promising potential in enhancing existing chemotherapies or neoadjuvant chemotherapies for FeSig^high EGFR^WT LUAD patients.

Methods

Data collection and processing

RNA expression data, mutation information, and clinical characteristics for EGFR^WT LUAD were downloaded from the UCSC-XENA database (https://xenabrowser.net/datapages/). Validation datasets GSE31210¹⁶ and GSE11969¹⁸ were obtained from GEO database, and the oncoSG cohort was downloaded from the cbioportal database^17,68. EGFR^MUT patients were excluded from subsequent analysis. Single-cell RNA sequencing data, GSE131907²⁰, was also downloaded from the GEO database. Ferroptosis-related genes were obtained from FerroDB¹⁴.

Risk gene screening

Univariate and multivariate Cox regression analysis were conducted on the TCGA-EGFR^WT LUAD cohort. Genes with significant prognostic properties in the univariate Cox regression analysis were selected as input for the multivariate cox regression analysis to narrow down candidate genes. Both univariate and multivariate Cox regression analysis were performed using survival and survminer R package.

FeSig scoring and survival analysis

GSVA R package was used to calculate FeSig score by setting methods as “ssgsva” and algorithm as “Gaussian” for array data or FPKM normalized data, detailed GSVA function and parameter settings as follow: gsva(as.matrix(Normalized gene expression matrix), gene signature, method = “ssgsea”, verbose = FALSE, kcdf = “Gaussian”). Gene signatures used in this manuscript were listed in Supplementary Data. Survival analysis of all four RNA sequencing data and one tissue array data were conducted using survival R package. Patients of TCGA-EGFR^WT LUAD cohort were split into FeSig^high or FeSig^low according to median FeSig score, and another 3 validation cohorts’ patients were divided by survival and survminer R package.

Time-dependent ROC and nomogram analysis

Time-dependent ROC analysis was applied to all four RNA sequencing datasets using survivalROC R package. For TCGA-EGFR^WT-LUAD, GSE31210 and oncoSG cohorts, ROC of 1 year, 3 years, and 5 years were evaluated, and for GSE11969, ROC of 3 years, 5 years, and 7 years were evaluated. Nomogram was conducted based on TCGA-EGFR^WT-LUAD cohort and calibration curve was also plotted by utilizing survival and rms R package.

Correlation analysis

To correlate the FeSig score and cell types of TME, TME cell signature genes were obtained from GSE131907. TME cell scores were calculated using GSVA R package. In addition, for testing the relationship between ferroptosis state and ligand-receptor (L-R-T) interaction of all four RNA sequencing cohorts, ferroptosis signature score and L-R-T score were calculated and scored using GSVA package and cor.test function, respectively. Spearman correlation analysis was performed on all four datasets and |Rs| > 0.3 and p value < 0.05 was set to filter cell types that significantly correlated with FeSig.

Single-cell RNA sequencing data processing

Single-cell sequencing data was obtained from GEO database, and data normalization and dimensional reduction were performed by utilizing Seurat V4⁶⁹. Patients who harbor mutated EGFR were screened out and EGFR^WT patients were retained for the subsequent analysis. To evaluate patients’ FeSig score, we employed pseudobulk algorithm from muscat⁷⁰ R package and the yielded pseudobulk data matrix was subjected to the GSVA algorithm to calculate corresponding FeSig score and patients were categorized into FeSig^high and FeSig^low two groups based on the mean FeSig scores. Cell type annotation was based on canonical marker genes: Endothelial cells (PECAM1, CLDN5, FLT1), fibroblasts (DCN, COL1A1, LUM), mast cells (KIT, MS4A2, GATA2), plasma cells (IGHM, IGHG3, IGHA2), NK cells (NKG7, GNLY, NCAM1), T cells (CD3D, CD3E, TRAC), epithelial cells (EPCAM, KRT19, KRT18), B cell (CD79A, MS4A1, CD19), myeloid cells (LYZ, MACRO, CD68). To elucidate the interaction alterations under single-cell resolution, cell-cell interactions were predicted and evaluated by R package Cellchat⁷¹ and Nichenetr⁷² with default settings.

Drug screening

To screen potential clinical applicable drugs based on FeSig, drug information of IC50 and corresponding cell line gene expression profiles were downloaded from NCI-60 database. Spearman correlation analysis was applied to all drug IC50 and cell line FeSig score which was calculated by GSVA R package. The drugs with Rs < −0.3 and p < 0.05 were seen as sensitive drugs, while the drugs with Rs > 0.3 and p < 0.05 were resistant drugs. Meanwhile, FeSig^high patient’s overexpression genes were inputted into the Connectivity Map database for screening another drug set. The intersection of the NCI-60 drug set and the Connectivity Map drug was employed to narrow down potential drugs.

Cell culture

Human Embryonic Kidney (HEK) 293T, A549, H1299, H441, H3122 (A549 and H3122 were kindly provided by Guangya Zhu Lab) and indicated knock-down (KD) cells were maintained in DMEM/RPMI-1640 (Servicebio, Wuhan, China) supplemented with 10% FBS (Lonsera, catalog S711-001) and 1% penicillin-streptomycin (Gibco, catalog 15140148) in a 5% CO2 incubator at 37°C. HEK293T, A549, H1299, H441, and H3122 were authenticated by short tandem repeat (STR) genotyping. All cells were tested routinely to avoid infection of Mycoplasma.

Construction of DDIT4/NOX4/GCLC KD cells

To make stable KD cells (sequences are listed in Supplementary Data), sgRNA were cloned into a lentiCRISPR vector. 6 μg lentiviral plasmid DNA was co-transfected with 3 μg pMD2.G, 6 μg psPAX2 to a 10 cm dish of HEK293T cells with around 80% confluence of cells. After 48 hours transfection, culture medium containing lentivirus was collected and centrifuged for 10 min at 3,000 g at room temperature, lentivirus supernatant was then filtered using 0.22 μm Millipore Express PES membrane filter unit. A549/H441/H1299 cells were cultured in infection medium (culture medium: lentivirus supernatant: polybrene = 1 ml : 1 ml : 2 μl), infection medium was then replaced with normal culture medium after 12 h-infection. Two days after incubation, infected cells were selected with puromycin (Merck, catalog 540411-25MGCN, working concentration, 1 μg/mL) for 3 generations. Knocking down the efficacy of corresponding target protein was tested by immunoblotting. For siRNA mediated knocking down of genes, cells were seeded and cultured in 6-well plates (10,000 cells per well) with RPMI-1640 medium with 10% FBS and cultured until around 60% of confluence. 9 μl RNAiMAX regent (Invitrogen, catalog 13778150) was diluted in 150 μl Opti-MEM (Gibico, catalog 31985070) medium (solution A), 30 pmol siRNA was diluted in 150 μl Opti-MEM (solution B). Mix solution B to solution A at a ratio of 1:1 gently and incubated under room temperature for 5 min, then add the mixture solution to the corresponding well and cultured for 48 h. After 48-h culturation, culture medium was replaced with a fresh RPMI-1640 medium.

Cell viability

Cell viability was measured using the Cell Counting Kit (CCK-8) (Yeasen Biotechnology, Shanghai, China, catalog 40203ES60). In brief, cells were seeded into 96-well plates at a density of 1000 cells/well. Cells were then cultivated at 37 °C for 72 h. Upon measurement, 10 μl of the CCK8 reagent was added to each well and incubated for 60 min. The supernatant’s optical density (OD) values were then measured at wavelengths of 450 nm and compared within different groups. Each experiment was independently repeated three times.

Cell colony formation assay

The colony formation assay was performed, and indicated cells were plated into 6-well plate at 1,500 cells per well and cultured for 10 days. Colonies were fixed with 4% polyoxymethylene (Servicebio, Wuhan, China, catalog G1101-500ML) for 30 min and stained with 2% crystal violet (Sangon, Shanghai, China, catalog E607309-0100) for 30 min for colony visualization. To quantify clone formation, crystal violet stained cell clones were eluted by 10% acetic acid solution (Sinopharm, Shanghai, China). The value was measured using a microplate reader at a wavelength of 595 nm. The relative colony number stands for fold change in relation to the cell number of control group.

Wound-healing assay

After cells were cultured to a confluence around 90%, a small area was disrupted by scratching the monolayer with a 200 μl plastic pipette tip. Afterward, cells were cultured for 36 h with low serum medium (1%) and then taken images by microscope. The migration rate was calculated and compared within different groups. Each experiment was independently repeated three times.

Flow cytometric analysis

A549 and H1299 cells were plated into 6-well plate and cultured to confluence around 80%, ferroptosis was induced by RSL3 (Selleck, Shanghai, China) for 24 hours. After treatment of RSL3, cells were washed with cold PBS and stained with Invitrogen™ BODIPY™ 581/591 C11 at room temperature for 30 minutes. When staining finished, cells were washed with cold PBS twice and resuspended in flowcytometry buffer (PBS containing 1% FBS). The fluorescence of stained cells was captured and analyzed at the FITC channel.

Multi-fluorescence staining

Formalin-fixed and paraffin-embedded tumor samples were deparaffined, rehydrated, and pre-treated with 3% hydrogen peroxidase (Sangon, Shanghai, China) for 10 min. Antigen retrieval with 1× Citrate Antigen Retrieval Solution buffer (Beyotime, Shanghai, China) was conducted in a steam cooker for 20 min at 100 °C, followed by slowly cooling down to room temperature. After blocking with 10% Bovine Serum Albumin (BSA, Beyotime, Shanghai, China), sections were incubated with anti-DDIT4 antibody (1:200, Proteintech, 10638-1-AP) for overnight at 4°C. After incubation with signalstain(R) Boost IHC Det. (HRP, Rab), positive signals were visualized by AF 555 TYRAMIDE (Invitrogen, Shanghai, China). Then, sections were incubated with anti-GCLC antibody (1:200, Proteintech, 12601-1-AP) overnight at 4 °C. After incubation with signalstain (R) Boost IHC Det. (HRP, Rab), positive signals were visualized by AF 647 TYRAMIDE (Invitrogen, Shanghai, China). Last, sections were incubated with α-SMA antibody (1:200, CST,19245S) overnight at 4 °C. After incubation with signalstain (R) Boost IHC Det. (HRP,Rab), positive signals were visualized by AF 488 TYRAMIDE (Invitrogen, Shanghai, China). After staining of DAPI (Sangon, Shanghai, China), slides were sealed and imaged with microscope (TG TissueFAXS PLUS).

Animals

We have complied with all relevant ethical regulations for animal use. Mouse studies were approved by the Institutional Animal Care and Use Committee of The First Affiliated Hospital of USTC (2023-N(A)-0167). Four-week-old BALB/c-nu/nu male mice were purchased from SLAC Laboratory Animal Co. Ltd. (Shanghai, China). Mice were bred in a pathogen-free animal facility in The First Affiliated Hospital of USTC for 1 week before experiments. Mice were randomly grouped and injected subcutaneously with 1 × 10⁶ A549 cells and NCI-H3122 cells within Matrigel-Matrix (Corning, catalog 354234). Experiments were terminated before the tumor volumes reached 1000 mm³. The volume of tumors was measured with calipers and calculated by the formula: Volume = 1/2 × length × width². 6-MP (MedChemExpress, Shanghai, China) was applied 20 mg/kg, i.p., every 2 days when tumor volume reached 100 mm³. The maximal tumor volume (1000 mm³) was permitted by Institutional Animal Care and Use Committee of The First Affiliated Hospital of USTC and none of experiments exceeded the maximal limit.

Statistics and reproducibility

Data are presented as mean ± standard deviation. Experiments were conducted and at least three biological replicates were set. Paired t test was performed on paired-sample data (Fig. 1c). For a two-group comparison, the Wilcoxon test or t test were used depending on the distribution of data (Figs. 2b, d, f, g, 3c–e, 4e, 6d, f, h, Supplementary Fig 3b–f, 4a, b, 5e). Spearman correlation analysis was performed to determine the correlations between two variants (Figs. 4a, 5f–i, 6a). Fisher exact test was performed to determine the relationship of variants (Fig. 3g). Sample sizes of each graph were described in figures and figure captions or presented as dots in figures. Data analyses and visualization were performed by R version 4.2.2. p < 0.05 was considered a statistically significant difference.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Public datasets used in this study can be found and downloaded with accession code: UCSC-XENA-TCGA-LUAD, GSE31210, GSE11969, oncoSG (data downloaded from cBioPortal), GSE131907. Uncropped images including western blots, scratch assays, clone formation assays and flow cytometry gating strategy can be found in Supplementary Information (Supplementary Fig. 7–9). Original source data supporting the results of this study are listed in Supplementary Data, which contains numerical data for Figs. 2b, d, f, 3c–e, 4e, 6d, f, h and Supplementary Fig. 3b–f. Other original image data supporting this study also available upon reasonable request from the corresponding authors.

Code availability

Analysis codes are deposited in https://github.com/wuqi12138/FeSig .

References

Siegel, R. L., Miller, K. D., Fuchs, H. E. & Jemal, A. Cancer Statistics, 2021. CA Cancer J. Clin. 71, 7–33 (2021).
Article PubMed Google Scholar
Marentakis, P. et al. Lung cancer histology classification from CT images based on radiomics and deep learning models. Med. Biol. Eng. Comput. 59, 215–226 (2021).
Article PubMed Google Scholar
Su, L. et al. CircRNAs in lung adenocarcinoma: diagnosis and therapy. Curr. Gene Ther. 22, 15–22 (2022).
CAS PubMed Google Scholar
Bade, B. C. & Dela Cruz, C. S. Lung Cancer 2020: epidemiology, etiology, and prevention. Clin. Chest Med. 41, 1–24 (2020).
Article PubMed Google Scholar
Tang, J. C. et al. Shikonin enhances sensitization of gefitinib against wild-type EGFR non-small cell lung cancer via inhibition PKM2/stat3/cyclinD1 signal pathway. Life Sci. 204, 71–77 (2018).
Article CAS PubMed Google Scholar
Kleczko, E. K. et al. Novel EGFR-mutant mouse models of lung adenocarcinoma reveal adaptive immunity requirement for durable osimertinib response. Cancer Lett. 556, 216062 (2023).
Article CAS PubMed PubMed Central Google Scholar
Li, J. et al. Ferroptosis: past, present and future. Cell Death Dis. 11, 88 (2020).
Article PubMed PubMed Central Google Scholar
Hassannia, B., Vandenabeele, P. & Vanden Berghe, T. Targeting ferroptosis to iron out cancer. Cancer Cell 35, 830–849 (2019).
Article CAS PubMed Google Scholar
Liang, C., Zhang, X., Yang, M. & Dong, X. Recent progress in ferroptosis inducers for cancer therapy. Adv. Mater. 31, e1904197 (2019).
Article PubMed Google Scholar
Ma, S., Henson, E. S., Chen, Y. & Gibson, S. B. Ferroptosis is induced following siramesine and lapatinib treatment of breast cancer cells. Cell Death Dis. 7, e2307 (2016).
Article CAS PubMed PubMed Central Google Scholar
Lou, J. S. et al. Ginkgetin derived from Ginkgo biloba leaves enhances the therapeutic effect of cisplatin via ferroptosis-mediated disruption of the Nrf2/HO-1 axis in EGFR wild-type non-small-cell lung cancer. Phytomedicine 80, 153370 (2021).
Article CAS PubMed Google Scholar
Yamaguchi, H. et al. Caspase-independent cell death is involved in the negative effect of EGF receptor inhibitors on cisplatin in non-small cell lung cancer cells. Clin. Cancer Res. 19, 845–854 (2013).
Article CAS PubMed PubMed Central Google Scholar
Gai, C. et al. Acetaminophen sensitizing erastin-induced ferroptosis via modulation of Nrf2/heme oxygenase-1 signaling pathway in non-small-cell lung cancer. J. Cell Physiol. 235, 3329–3339 (2020).
Article CAS PubMed Google Scholar
Zhou, N. et al. FerrDb V2: update of the manually curated database of ferroptosis regulators and ferroptosis-disease associations. Nucleic Acids Res. 51, D571–d582 (2023).
Article CAS PubMed Google Scholar
Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinform. 14, 7 (2013).
Article Google Scholar
Okayama, H. et al. Identification of genes upregulated in ALK-positive and EGFR/KRAS/ALK-negative lung adenocarcinomas. Cancer Res. 72, 100–111 (2012).
Article CAS PubMed Google Scholar
Chen, J. et al. Genomic landscape of lung adenocarcinoma in East Asians. Nat. Genet. 52, 177–186 (2020).
Article CAS PubMed Google Scholar
Takeuchi, T. et al. Expression profile-defined classification of lung adenocarcinoma shows close relationship with underlying major genetic changes and clinicopathologic behaviors. J. Clin. Oncol. 24, 1679–1688 (2006).
Article CAS PubMed Google Scholar
Anurag, M. et al. Proteogenomic markers of chemotherapy resistance and response in triple-negative breast cancer. Cancer Discov. 12, 2586–2605 (2022).
Article CAS PubMed PubMed Central Google Scholar
Kim, N. et al. Single-cell RNA sequencing demonstrates the molecular and cellular reprogramming of metastatic lung adenocarcinoma. Nat. Commun. 11, 2285 (2020).
Article CAS PubMed PubMed Central Google Scholar
Gao, R. et al. YAP/TAZ and ATF4 drive resistance to Sorafenib in hepatocellular carcinoma by preventing ferroptosis. EMBO Mol. Med. 13, e14351 (2021).
Article CAS PubMed PubMed Central Google Scholar
Chen, Q. et al. SOCS2-enhanced ubiquitination of SLC7A11 promotes ferroptosis and radiosensitization in hepatocellular carcinoma. Cell Death Differ. 30, 137–151 (2023).
Article CAS PubMed Google Scholar
Yang, W. H. et al. The hippo pathway effector TAZ regulates ferroptosis in renal cell carcinoma. Cell Rep. 28, 2501–2508.e2504 (2019).
Article CAS PubMed PubMed Central Google Scholar
Jiang, K. et al. STC2 activates PRMT5 to induce radioresistance through DNA damage repair and ferroptosis pathways in esophageal squamous cell carcinoma. Redox Biol. 60, 102626 (2023).
Article CAS PubMed PubMed Central Google Scholar
Qiu, S. et al. Mitochondria-localized cGAS suppresses ferroptosis to promote cancer progression. Cell Res. 33, 299–311 (2023).
Article CAS PubMed PubMed Central Google Scholar
Lang, X. et al. Radiotherapy and immunotherapy promote tumoral lipid oxidation and ferroptosis via synergistic repression of SLC7A11. Cancer Discov. 9, 1673–1685 (2019).
Article CAS PubMed PubMed Central Google Scholar
Zhang, Z. et al. Holo-lactoferrin: the link between ferroptosis and radiotherapy in triple-negative breast cancer. Theranostics 11, 3167–3182 (2021).
Article CAS PubMed PubMed Central Google Scholar
Niu, B. et al. Application of glutathione depletion in cancer therapy: Enhanced ROS-based therapy, ferroptosis, and chemotherapy. Biomaterials 277, 121110 (2021).
Article CAS PubMed Google Scholar
Wang, Y. et al. Overcoming cancer chemotherapy resistance by the induction of ferroptosis. Drug Resist Updat. 66, 100916 (2023).
Article CAS PubMed Google Scholar
Yang, F. et al. Ferroptosis heterogeneity in triple-negative breast cancer reveals an innovative immunotherapy combination strategy. Cell Metab. 35, 84–100.e108 (2023).
Article CAS PubMed Google Scholar
Gao, W., Wang, X., Zhou, Y., Wang, X. & Yu, Y. Autophagy, ferroptosis, pyroptosis, and necroptosis in tumor immunotherapy. Signal Transduct. Target Ther. 7, 196 (2022).
Article PubMed Google Scholar
Wang, W. et al. CD8(+) T cells regulate tumour ferroptosis during cancer immunotherapy. Nature 569, 270–274 (2019).
Article CAS PubMed PubMed Central Google Scholar
Zhao, F., Wang, M. & Zhu, J. Hypoxia-related lncRNAs to build prognostic classifier and reveal the immune characteristics of EGFR wild type and low expression of PD-L1 squamous and adenocarcinoma NSCLC. Cancer Med. 10, 6099–6113 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wang, M., Zhu, J., Zhao, F. & Xiao, J. Transcriptome analyses identify a metabolic gene signature indicative of antitumor immunosuppression of EGFR wild type lung cancers with low PD-L1 expression. Front. Oncol. 11, 643503 (2021).
Article CAS PubMed PubMed Central Google Scholar
Planck, M., Isaksson, S., Veerla, S. & Staaf, J. Identification of transcriptional subgroups in EGFR-mutated and EGFR/KRAS wild-type lung adenocarcinoma reveals gene signatures associated with patient outcome. Clin. Cancer Res. 19, 5116–5126 (2013).
Article CAS PubMed Google Scholar
Yu, F. et al. RBM33 is a unique m(6)A RNA-binding protein that regulates ALKBH5 demethylase activity and substrate selectivity. Mol. Cell 83, 2003–2019.e2006 (2023).
Article CAS PubMed PubMed Central Google Scholar
Chen, X. et al. Identification of DDIT4 as a potential prognostic marker associated with chemotherapeutic and immunotherapeutic response in triple-negative breast cancer. World J. Surg. Oncol. 21, 194 (2023).
Article PubMed PubMed Central Google Scholar
Fattahi, F. et al. High expression of DNA damage-inducible transcript 4 (DDIT4) is associated with advanced pathological features in the patients with colorectal cancer. Sci. Rep. 11, 13626 (2021).
Article CAS PubMed PubMed Central Google Scholar
Zhang, Y., Wang, Y., Chen, J., Xia, Y. & Huang, Y. A programmed cell death-related model based on machine learning for predicting prognosis and immunotherapy responses in patients with lung adenocarcinoma. Front. Immunol. 14, 1183230 (2023).
Article CAS PubMed PubMed Central Google Scholar
Song, X. et al. Streptococcus pneumoniae promotes migration and invasion of A549 cells in vitro by activating mTORC2/AKT through up-regulation of DDIT4 expression. Front. Microbiol. 13, 1046226 (2022).
Article PubMed PubMed Central Google Scholar
Guo, Q. et al. Clinical roles of risk model based on differentially expressed genes in mesenchymal stem cells in prognosis and immunity of non-small cell lung cancer. Front. Genet. 13, 823075 (2022).
Article CAS PubMed PubMed Central Google Scholar
Shi, Y. et al. TKI resistant-based prognostic immune related gene signature in LUAD, in which FSCN1 contributes to tumor progression. Cancer Lett. 532, 215583 (2022).
Article CAS PubMed Google Scholar
Tang, B. et al. Identification of critical ferroptosis regulators in lung adenocarcinoma that RRM2 facilitates tumor immune infiltration by inhibiting ferroptotic death. Clin. Immunol. 232, 108872 (2021).
Article CAS PubMed Google Scholar
Skonieczna, M., Hejmo, T., Poterala-Hejmo, A., Cieslar-Pobuda, A. & Buldak, R. J. NADPH oxidases: insights into selected functions and mechanisms of action in cancer and stem cells. Oxid. Med. Cell Longev. 2017, 9420539 (2017).
Article PubMed PubMed Central Google Scholar
Liu, W. J. et al. Elevated NOX4 promotes tumorigenesis and acquired EGFR-TKIs resistance via enhancing IL-8/PD-L1 signaling in NSCLC. Drug Resist Updat. 70, 100987 (2023).
Article CAS PubMed Google Scholar
Fan, C. C. et al. EFHD2 contributes to non-small cell lung cancer cisplatin resistance by the activation of NOX4-ROS-ABCC1 axis. Redox Biol. 34, 101571, (2020).
Article CAS PubMed Google Scholar
Chen, B. et al. Fangchinoline inhibits non-small cell lung cancer metastasis by reversing epithelial-mesenchymal transition and suppressing the cytosolic ROS-related Akt-mTOR signaling pathway. Cancer Lett. 543, 215783 (2022).
Article CAS PubMed Google Scholar
Zhang, J. et al. Tumoral NOX4 recruits M2 tumor-associated macrophages via ROS/PI3K signaling-dependent various cytokine production to promote NSCLC growth. Redox Biol. 22, 101116 (2019).
Article CAS PubMed PubMed Central Google Scholar
Ford, K. et al. NOX4 inhibition potentiates immunotherapy by overcoming cancer-associated fibroblast-mediated CD8 T-cell exclusion from tumors. Cancer Res 80, 1846–1860 (2020).
Article CAS PubMed PubMed Central Google Scholar
Yang, Z. et al. ACTL6A protects gastric cancer cells against ferroptosis through induction of glutathione synthesis. Nat. Commun. 14, 4193 (2023).
Article CAS PubMed PubMed Central Google Scholar
Kang, Y. P. et al. Non-canonical glutamate-cysteine ligase activity protects against ferroptosis. Cell Metab. 33, 174–189.e177 (2021).
Article CAS PubMed Google Scholar
Cui, Z. et al. Dihydroartemisinin enhances the inhibitory effect of sorafenib on HepG2 cells by inducing ferroptosis and inhibiting energy metabolism. J. Pharm. Sci. 148, 73–85 (2022).
Article CAS Google Scholar
Chen, Y. et al. An antioxidant feedforward cycle coordinated by linker histone variant H1.2 and NRF2 that drives nonsmall cell lung cancer progression. Proc. Natl Acad. Sci. USA 120, e2306288120 (2023).
Article CAS PubMed PubMed Central Google Scholar
Chow, A., Perica, K., Klebanoff, C. A. & Wolchok, J. D. Clinical implications of T cell exhaustion for cancer immunotherapy. Nat. Rev. Clin. Oncol. 19, 775–790 (2022).
Article PubMed PubMed Central Google Scholar
Wang, Q., Qin, Y. & Li, B. CD8(+) T cell exhaustion and cancer immunotherapy. Cancer Lett. 559, 216043 (2023).
Article CAS PubMed Google Scholar
Bell, H. N. et al. Microenvironmental ammonia enhances T cell exhaustion in colorectal cancer. Cell Metab. 35, 134–149.e136 (2023).
Article CAS PubMed Google Scholar
Kang, J. H. & Zappasodi, R. Modulating Treg stability to improve cancer immunotherapy. Trends Cancer 9, 911–927 (2023).
Article CAS PubMed Google Scholar
Moreno Ayala, M. A. et al. CXCR3 expression in regulatory T cells drives interactions with type I dendritic cells in tumors to restrict CD8(+) T cell antitumor immunity. Immunity 56, 1613–1630.e1615 (2023).
Article CAS PubMed Google Scholar
Liu, Y. T. & Sun, Z. J. Turning cold tumors into hot tumors by improving T-cell infiltration. Theranostics 11, 5365–5386 (2021).
Article CAS PubMed PubMed Central Google Scholar
Hornburg, M. et al. Single-cell dissection of cellular components and interactions shaping the tumor immune phenotypes in ovarian cancer. Cancer Cell 39, 928–944.e926 (2021).
Article CAS PubMed Google Scholar
Bagaev, A. et al. Conserved pan-cancer microenvironment subtypes predict response to immunotherapy. Cancer Cell 39, 845–865.e847 (2021).
Article CAS PubMed Google Scholar
Liu, Y. et al. Identification of a tumour immune barrier in the HCC microenvironment that determines the efficacy of immunotherapy. J. Hepatol. 78, 770–782 (2023).
Article CAS PubMed Google Scholar
Jain, S. et al. Single-cell RNA sequencing and spatial transcriptomics reveal cancer-associated fibroblasts in glioblastoma with protumoral effects. J. Clin. Invest. 133, e147087 (2023).
Article CAS PubMed PubMed Central Google Scholar
Kobayashi, H. et al. The Origin and Contribution of Cancer-Associated Fibroblasts in Colorectal Carcinogenesis. Gastroenterology 162, 890–906 (2022).
Article CAS PubMed Google Scholar
Park, S. et al. ACTA2 expression predicts survival and is associated with response to immune checkpoint inhibitors in gastric cancer. Clin. Cancer Res 29, 1077–1085 (2023).
Article CAS PubMed PubMed Central Google Scholar
Ding, H., Chen, J., Qin, J., Chen, R. & Yi, Z. TGF-β-induced α-SMA expression is mediated by C/EBPβ acetylation in human alveolar epithelial cells. Mol. Med. 27, 22 (2021).
Article CAS PubMed PubMed Central Google Scholar
Manetti, M. et al. Endothelial-to-mesenchymal transition contributes to endothelial dysfunction and dermal fibrosis in systemic sclerosis. Ann. Rheum. Dis. 76, 924–934 (2017).
Article CAS PubMed Google Scholar
Cerami, E. et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2, 401–404 (2012).
Article PubMed Google Scholar
Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587.e3529 (2021).
Article CAS PubMed Google Scholar
Crowell, H. L. et al. muscat detects subpopulation-specific state transitions from multi-sample multi-condition single-cell transcriptomics data. Nat. Commun. 11, 6077 (2020).
Article CAS PubMed PubMed Central Google Scholar
Jin, S. et al. Inference and analysis of cell-cell communication using CellChat. Nat. Commun. 12, 1088 (2021).
Article CAS PubMed PubMed Central Google Scholar
Browaeys, R., Saelens, W. & Saeys, Y. NicheNet: modeling intercellular communication by linking ligands to target genes. Nat. Methods 17, 159–162 (2020).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

This work was supported by the Natural Science Foundation of Anhui Province (2208085MH241, 2208085MH237), research funds of the Center for Advanced Interdisciplinary Science and Biomedicine of IHM (QYPY20220005), and Anhui Provincial Health Research Project (AHWJ2023BAa20192).

Author information

These authors contributed equally: Chuankai Zhang, Qi Wu, Hongwei Yang.

Authors and Affiliations

Department of Thoracic Surgery, The First Affiliated Hospital of University of Science and Technology of China (USTC), Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei, China
Chuankai Zhang, Changqing Liu & Qingsong Hu
Department of Hepatobiliary Surgery, The First Affiliated Hospital of USTC, Center for Advanced Interdisciplinary Science and Biomedicine of IHM, Division of Life Sciences and Medicine, University of Science and Technology of China, 230001, Hefei, China
Qi Wu, Hongwei Yang, Hui Zhang & Qingsong Hu
The First Affiliated Hospital of University of Science and Technology of China (USTC), Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei, China
Bo Yang

Authors

Chuankai Zhang
View author publications
Search author on:PubMed Google Scholar
Qi Wu
View author publications
Search author on:PubMed Google Scholar
Hongwei Yang
View author publications
Search author on:PubMed Google Scholar
Hui Zhang
View author publications
Search author on:PubMed Google Scholar
Changqing Liu
View author publications
Search author on:PubMed Google Scholar
Bo Yang
View author publications
Search author on:PubMed Google Scholar
Qingsong Hu
View author publications
Search author on:PubMed Google Scholar

Contributions

Q.H. C.Z., and B.Y. conceived and supervised the project. C.Z., H.Y., and H.Z. designed and performed the research. Q.W. performed data analysis. C.Z., C.L., and H.Y. developed mouse models and related experiments. C.Z., B.Y., and Q.H. interpreted the results. C.Z., Q.W., and Q.H. wrote the manuscript with input from all other authors.

Corresponding authors

Correspondence to Chuankai Zhang, Bo Yang or Qingsong Hu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Communications Biology thanks Bo Chu, and the other anonymous reviewer for their contribution to the peer review of this work. Primary Handling Editor: Johannes Stortz. [A peer review file is available.]

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Transparent Peer Review file

Supplementary Information

Description of Additional Supplementary Files

Reporting Summary

Supplementary Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Zhang, C., Wu, Q., Yang, H. et al. Ferroptosis-related gene signature for predicting prognosis and identifying potential therapeutic drug in EGFR wild-type lung adenocarcinoma. Commun Biol 7, 1416 (2024). https://doi.org/10.1038/s42003-024-07117-1

Download citation

Received: 16 February 2024
Accepted: 22 October 2024
Published: 30 October 2024
DOI: https://doi.org/10.1038/s42003-024-07117-1