Prognostic biomarkers related to PANoptosis in esophageal cancer and their immune microenvironment: multi-omics analysis and therapeutic significance
Runda Jie, Yining Liu, Tianbiao Wang, Shabahaiti Wusiman, Hongda Ren, Wugelinisan Tuoheniyazi, Ling Liu
TL;DR
This study identifies PANoptosis-related biomarkers in esophageal cancer and explores their role in prognosis and the immune environment.
Contribution
The study introduces new prognostic biomarkers and a novel mechanism involving CCT6A in regulating PANoptosis and the immune microenvironment in ESCA.
Findings
74 PANoptosis-related differentially expressed genes were identified, with CCT6A, GMNN, and HSPB6 as key hub genes.
High-risk subgroups showed poor prognosis, immune exhaustion, and reduced immunotherapy response.
CCT6A may inhibit PANoptosis by promoting TAM-SPP1 macrophage differentiation, offering a new therapeutic target.
Abstract
Esophageal squamous cell carcinoma (ESCA) is one of the most common cancers worldwide. PANoptosis is an inflammatory programmed cell death pathway event regulated by the PANoptosome complex. Currently, there is limited research on the PANoptosis-related genes (PORGs) in ESCA. We aim to explore the prognostic biomarkers of PANoptosis in ESCA and their underlying mechanisms through comprehensive bioinformatics analysis. In this study, we analyzed transcriptome and single-cell RNA sequencing (scRNA-seq) data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. Weighted gene co-expression network analysis (WGCNA) and differential expression analysis were used to identify PANoptosis-related differentially expressed genes (POR-DEGs) in esophageal cancer. Hub genes were screened by univariate and multivariate Cox regression combined with machine learning models to…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11| Dependent:Surv(time,state) | all | HR(univariable) | HR(multivariable) | |
|---|---|---|---|---|
| age | Mean±SD | 64.9±12.2 | 1.00(0.98-1.02,p=0.884) | |
| gender | female | 20(17.9%) | ||
| male | 92(82.1%) | 1.17(0.46-2.96,p=0.749) | ||
| T | 1 | 27(24.1%) | ||
| 2 | 20(17.9%) | 1.30(0.55-3.08,p=0.552) | ||
| 3 | 63(56.2%) | 1.28(0.60-2.72,p=0.521) | ||
| 4 | 2(1.8%) | 3.67(0.78-17.33,p=0.101) | ||
| N Stage | 0 | 37(33.0%) | ||
| 1 | 58(51.8%) | 2.37(1.16-4.86,p=0.018) | 1.21(0.51-2.86,p=0.669) | |
| 2 | 9(8.0%) | 3.16(0.85-11.76,p=0.086) | 1.38(0.31-6.09,p=0.672) | |
| 3 | 8(7.1%) | 1.41(0.30-6.56,p=0.659) | 0.69(0.13-3.73,p=0.670) | |
| Stagel | 15(13.4%) | |||
| Stage ll | 41(36.6%) | 3.58(0.77-16.55,p=0.103) | 3.46(0.68-17.53,p=0.134) | |
| Stage III | 45(40.2%) | 6.75(1.41-32.35,p=0.017) | 6.36(1.06-38.16,p=0.043) | |
| Stage IV | 11(9.8%) | 8.64(1.67-44.86,p=0.010) | 8.28(1.31-52.21,p=0.024) | |
| riskScore | Mean±SD | 1.1±0.6 | 1.83(1.17-2.86,p=0.008) | 1.87(1.18-2.96,p=0.008) |
| Numerator ID | Name | Receptor protein | PDB ID | Binding abilitya |
|---|---|---|---|---|
| 5280343 | Quercetin | CCT6A | AF- | -8.3 kcal·mol-1 |
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsFerroptosis and cancer prognosis · Immune cells in cancer · Single-cell and spatial transcriptomics
Introduction
1
Esophageal cancer is a common malignant tumor of the digestive system, characterized by invasiveness, early spread, rapid tumor recurrence and poor prognosis (1). In recent years, the incidence of esophageal cancer has been on the rise. According to the latest research data from international and domestic research institutions, in 2022, its incidence ranked 11th among global tumors, and its mortality rate ranked 7th (2). The two main subtypes of esophageal cancer are esophageal squamous cell carcinoma (ESCC) and adenocarcinoma (EAC), which are different in epidemiology and biology (3). Patients with esophageal cancer usually present with main symptoms such as dysphagia, gastrointestinal bleeding, repeated aspiration or vomiting, and weight loss.
PANoptosis is a programmed cell death (PCD) pathway activated by specific stimuli and regulated by the PANoptosome complex. It integrates the molecular mechanisms of apoptosis, pyroptosis and necroptosis, forming a multi-pathway synergistic regulatory network (4). Its core features are reflected in: 1. Sharing the key regulatory nodes of apoptosis, pyroptosis and necrotic apoptosis at the molecular level; 2. Pathologically, it has both pro-inflammatory and non-pro-inflammatory dual modes. 3. The cell phenotype can simultaneously present mixed characteristics such as apoptotic body formation, membrane pore generation and content release (4). The assembly of the PANoptosome complex and the activation of PANoptosis occur simultaneously in autoinflammation, infection, neurodegenerative diseases and cancer (5). It is crucial for eliminating host pathogens after infection, cancer treatment, immunity and disease prevention (6). Recently reported cases such as breast cancer (7), clear cell renal cell carcinoma (8), intrahepatic cholangiocarcinoma (9), pancreatic cancer (10), lung adenocarcinoma (11), thyroid cancer (12), hepatocellular carcinoma (13), intestinal adenocarcinoma (14), cutaneous melanoma (15), and multiple myeloma (16) have all utilized PANoptosis-related genes to construct prognostic models or unsupervised classifications. However, the specific regulatory network of PANoptosis in ESCA and its interaction with the tumor microenvironment have not been fully elucidated, which restricts its potential for clinical transformation. Analyzing the mechanism of action of PANoptosis-related prognostic biomarkers in tumor cells and immune cells will help develop precise treatment strategies based on pan-apoptotic regulation.
The combined analysis of single-cell sequencing (scRNA-seq) is widely used in many studies as a powerful tool for discovering important cell types and diagnostic markers to identify relevant biomarkers for characterization (17). This joint analysis provides a more reliable theoretical basis for the study of disease pathogenesis, allowing for precise exploration of individual cell populations at an unprecedented resolution (18).
This study, based on the TCGA and GEO databases, introduced the concept of PANoptosis into ESCA research and combined single-cell sequencing to analyze its cell-specific regulatory network. Build a multi-gene combined prognosis/diagnosis model to promote the development of ESCA precision medicine. To reveal the new mechanism by which quercetin may act on CCT6A, which promotes the differentiation of monocytes into TAM-SPP1, to inhibit PANoptosis, and to provide some references for the clinical treatment of ESCA patients.
Materials and methods
2
Data collection
2.1
The ESCA dataset collected from TCGA includes 13 normal tissue samples and 185 ESCA samples. The GSE17351 dataset includes 5 ESCC samples and 5 normal tissue samples, the GSE38129 dataset includes 30 ESCC samples and 30 normal tissue samples, and the GSE29001 dataset includes 21 ESCC samples and 24 normal tissue samples. The single-cell dataset GSE196756 includes 3 patients with esophageal squamous cell carcinoma and 3 patients with matched adjacent tissues. The tissue microarray was purchased from Shanghai Xinchao Biotechnology Co., LTD. Among them, the HEsoS060CS01 serial number contains 2 normal esophageal tissues and 30 pairs of esophageal cancer tissues and adjacent tissues, and the HEsoSqu060PG01 serial number contains 30 pairs of esophageal cancer tissues and adjacent tissues. Some PANoptosis-related genes are derived from previous studies (19).
Differential gene analysis
2.2
To detect TCGA_DEGs between the regular group and the ESCA group in the TCGA-ESCA dataset, we used the “wilcoxon” software (20). Set the threshold at p.AJ <0.05, >1. The ggplot2 and ggVolcano volcano distribution maps show TCGA_DEGs. The heat map shows the top50 TCGA_DEGs (top50 downregulated and top50 upregulated).
Weighted gene co-expression network analysis
2.3
The “GSVA” software calculates the ssGSEA scores of 14 PORGs. The SsGSEA score is a clinical feature. We grouped the samples and removed outliers to ensure the accuracy of the analysis. Then create the feature heat map, sample dendrite map and soft threshold. Generate phylogenetic trees by utilizing genetic similarity and adjacency. The gene module ssGSEA_WGCNA with the highest correlation with clinical features (or =0.15 and p<0.05, or =0.39 and p<0.0001) was taken as the key for subsequent analysis.
Single-cell data processing
2.4
The scRNA-seq data were derived from GSE196756, which included 3 ESCC specimens and 3 adjacent specimens. Quality control ensures that the number of expressed genes detected in each cell is greater than 300 and less than 7000, the proportion of mitochondrial gene expression in the total genes in each cell is less than 10%, and that of red blood cells is less than 3%. The UMI count content of sequencing in each cell is greater than 1000, and the top 3% cells with the largest are excluded. We executed the “Find Variable Characteristics” function to screen the top 2,000 high-variable genes for harmony analysis. The cells are clustered using consensus clustering and visualized using the UMAP algorithm based on harmony. Myeloid cell subsets were annotated using double-cell excision and marker genes of myeloid cell subtypes, while epithelial cell subsets were annotated using double-cell excision, marker genes of epithelial cell subtypes and the “infercnv” package. The expression of scDEGs was analyzed for differential gene expression using the Seurat FindMarker function, with the threshold set at |log2FC| >0.25.
Machine learning methods
2.5
Among these POR-DEGs, identify the POR-DEGs related to prognosis and conduct subsequent analyses. The univariate Cox proportional hazards regression model was used to analyze the correlation between POR-DEGs and overall survival (OS) in the internal validation of TCGA, and those with OS≥30 days were included in the analysis. The machine learning method employs LASSO penalized Cox proportional hazards regression, RF algorithm and SVM-RFE algorithm to screen POR-DEGs in the internal validation of TCGA. The prognostic genes were obtained by taking the cross-genes of the three machine learning algorithms through 10-fold cross-validation. Then, prognostic genes were used to construct a prognostic model related to ESCA.
The risk score was calculated using multivariate Cox regression coefficients, with the formula being: Risk score =[expression level of gene 1 *coef][expression level of gene 2 *coef] … [Gene n expression level *coef], and then divided into high-risk group and low-risk group according to the median. To evaluate the reliability of predicting the prognosis of POR-DEGs, we plotted the Kaplan-Meier (K-M) survival curve using the R software package “Survival” (21). To further reflect the sensitivity and specificity of POR-DEGs, we conducted a time-varying receiver operating characteristic (ROC) curve analysis using the R software package “survivalROC” (22). In addition, univariate and multivariate analyses of POR-DEGs OS and clinicopathological factors in the entire TCGA cohort were conducted using the R package “survival” (21). In addition, independent t-tests were used to analyze the correlation between POR-DEGs and different clinicopathological factors.
Construction of prognosis maps and diagnostic models
2.6
Based on the above results, we constructed a prognostic nomogram to quantitatively estimate the survival risk of patients with HNSC. In addition, the correction curve is used to compare the predicted survival possibility with the observed survival possibility. Draw Nomograms and calibrate curves through the R package “rms”. SHAP interpretable analysis further elaborates on the contribution of each gene to prognosis. The internal validation of TCGA and the external validations of GSE17351, GSE38129, and GSE29001, which were corrected and integrated with the “ComBat” batch, respectively detected the mRNA expression levels of POR-DEGs using transcriptomics. When calculating the sensitivity and specificity of the XGBoost model, the ROC curve and confusion matrix were used for evaluation.
Gene function enrichment analysis
2.7
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted via the “clusterProfiler package” (23). Take P < 0.05 as the standard. In addition, GSEA explored potential KEGG pathways (24) related to prognostic genes through the “clusterProfiler” software package.
Evaluation of the tumor immune microenvironment
2.8
CIBERSORT platform (https://cibersortx.stanford.edu) through the M22 data sets match the expression profile of this study, to help us identify immune cells infiltrating in esophageal cancer patients and normal person. The CIBERSORT algorithm was applied to calculate the relative abundance (25) of 22 immune cells infiltrating the ESCA microenvironment. Based on all immune infiltration algorithms, correlation analysis was used to further calculate and display the correlation between prognostic genes and differential immune cells. “IBOR-IPS” was used to analyze the immunotherapy responses of these high and low-risk groups. The “Estimate” algorithm further revealed the infiltration of tumor immune cells and stromal cells, and was used to evaluate the TME in high and low-risk groups.
Single-cell analysis
2.9
The processed GSE196756 dataset was used for further analysis of the hub gene. The “ggplot2” and “Scop” were used to draw the violin plot of hub gene expression, “AUCell” was used to score cells based on PORGs, the correlation heat map was used to evaluate the association between CCT6A and PORGs and the ssGSEA score of PORGs, and “SctenifoldKnk” was used to virtually knockout the hub gene in cells In addition to enrichment analysis, “Monocle3” and “CytoTRACE2” were used for cell trajectory analysis and quasi-temporal analysis, UMAP and bar charts were used to show the changes in the proportion of cells in different groups, “Ro/e” was used to analyze the distribution tendency of cell subpopulations, “Augur” was used to analyze the gene perturbation of cell subpopulations, and “ClusterGVis” was used for marking Enrichment analysis of genes and dynamic genes, “CellChat” inference and quantification of intercellular communication interactions involved in cell clusters. The processed GSE196756 dataset was also used for further analysis of the high and low-risk groups in the TCGA-ESCA cohort. “AUCell” was used to classify the cells into high-risk driver groups and low-risk inhibitory groups based on differences in gene expression, and “CellChat” was used to identify the differences in efferent and afferent signals of cell types as well as the relative contributions of ligand-receptors in the signaling pathway.
Molecular docking
2.10
Map the hub gene CCT6A that regulates macrophages to the Coremine medical ontology information retrieval platform (https://coremine.com/medical/), and select the top 8 traditional Chinese medicine drugs that may have a regulatory effect on the hub gene in patients with esophageal cancer. In addition, in order to obtain its target protein in a results file, in the PDB database (https://www.rcsb.org/) and AlphaFold database (https://alphafold.ebi.ac.uk/). Active ingredient from the PubChem database (https://alphafold.ebi.ac.uk/) retrieval file structure, and use the Open Babel 3.1.1 software to convert them into PDB format. Then, the receptor protein was carefully prepared by removing water molecules and ligands using PYMOL 3.1.1 software. Subsequently, the necessary modifications to the receptor protein were carried out using the AutoDockTools software, including hydrogenation and charge balancing. Subsequently, molecular docking simulation was conducted between the receptor protein and the small molecule ligand using AutoDock Vina 1.1.2. Finally, the docking results were visualized using Pymol, with a focus on the docking products that had good binding energy.
Immunohistochemistry
2.11
The 60-point esophageal cancer tissue microarrays (2 HEsoS060CS01 and 1 HEsoSqu060PG01) were placed in an oven preheated to 60°C for 1 hour of wax drying treatment. After completion, immediately take the chip out of the oven and place it in a container containing xylene I for dewaxing for 15 minutes. Then, transfer the chip to xylene II in another container and carry out dewaxing treatment for the same period of 15 minutes. Subsequently, the samples were respectively soaked in two different anhydrous ethanols for 5 minutes for dewaxing. And complete dewaxing treatment was carried out in the order of 95% ethanol (5 minutes), 95% ethanol (5 minutes), and 85% ethanol (5 minutes).
Subsequently, wash twice with PBS buffer and immerse the sample in EDTA antigen remediation solution prepared with a pH value of 9.0. Preheat in a microwave oven for 3 minutes and then heat on medium-high heat for 10–15 minutes. During the process, replenish the solution. After the sample was cooled at room temperature for 1 hour, it was washed three times again with PBS (each lasting 5 minutes), and then blocked with goat serum for 30 minutes. Next, the primary antibody was dropped onto the tissue surface in a wet box and incubated overnight in a 4 °C refrigerator.
The next day, take the sample out outdoors for rewarming for 40 minutes, then add the secondary antibody and incubate for 30 minutes. After washing with PBS once, observe the yellow color development reaction of the sample under a microscope. If there is a reaction, counterstain with hematoxylin for 1 minute and soak the sample in 1% hydrochloric acid ethanol differentiation solution for no less than 2 seconds.
Finally, the tissue microarray was subjected to dehydration treatment: first, it was soaked in 85% ethanol for 10 seconds, and then quickly transferred to two different 95% ethanols, two different anhydrous ethanols, as well as xylene I and xylene II for 10 seconds each. After the moisture on the sections has dried, seal the sections with neutral gum and observe the staining of the samples under a microscope.
Drug sensitivity analysis
2.12
Drug sensitivity analysis data from GDSC2 (Genomics of drug sensitivity in Cancer) database (https://www.cancerrxgene.org/). The relationship between the high and low-risk groups and drug sensitivity was detected using the “oncoPredict” R package.
Data analysis
2.13
All statistical analyses were conducted using R software (https://www.r-project.org/). These experiments were conducted at least three times. Statistical research was conducted using R 4.4.1 version and GraphPad Prism 10.0 version software. The statistical differences between the two groups were evaluated using the wilcoxon rank sum test. The significance level is expressed as:*, p < 0.05; **, p < 0.01 ***, p < 0.001.
Result
3
Identification of key genes for esophageal cancer
3.1
Under the adjusted P values of <0.05 and |log2FC| >1, a total of 3,425 DEGs were identified, including 2,975 up-regulated genes and 450 down-regulated genes (Figure 1A). The heat map shows the top 50 up-regulated genes and the top 50 down-regulated genes (Figure 1B). To identify the key modules related to ESCA PANoptosis, we conducted WGCNA. The score of PANoptosome-related genes (26) in ESCA samples was significantly different from that in normal samples, and thus it was used as a clinical feature (Figure 1C). The sample clustering results show that there are few abnormal samples, and the optimal soft threshold is 5. When β=5, the average connectivity tends to 0, and the ordinate scale-free fitting index approaches the threshold of 0.9 (Figure 1D). A total of five modules were obtained through the dynamic tree cutting algorithm (Figure 1E). The blue module is weakly positively correlated with the “PORGs score”, but strongly negatively correlated with “whether it is a tumor phenotype” (Figure 1F). Therefore, 683 basic module genes related to the PANoptosis score were obtained for subsequent analysis.
DEG identification and WGCNA analysis. (A) Volcano map of DEGs. (B) Heat map of DEGs. (C) Difference in PORGs scores between ESCA samples and controls. (D) Gene correlations best capture scale-free distributions at power 5 according to soft threshold analysis. (E) Cluster dendrogram of ESCA co-expression. (F) Heatmap of correlations between different gene modules and traits, with p.adj values in brackets and correlation coefficients for different modules outside brackets.
Single-cell atlas of esophageal cancer
3.2
After cell filtering, we conducted cell clustering and annotation analysis on GSE196756 (Figure 2A). All highly variable genes and the first 15 are shown in the figure (Figure 2B). We divided the cells into 16 clusters at a resolution of 0.2 (Figure 2C). Based on the “Single R” package and marker genes, they were further clustered into T cells, B cells, Epithelial cells, Endothelial cells, and Mast cell, Fibroblast, Neurons, Myeloid and Neutrophils (Figures 2D, E), the top5 markers of different cell types further verified the accuracy (Figure 2F). After removing the 0-cluster immune and epithelial two-cell (Figures 3A, B), based on the marker genes (27) of known epithelial cell subtypes and the identification of tumor cells by “infercnv” (Figures 3C, D), we further clustered the epithelial cells into five subtypes expressing different programs. Including quiescent progenitor cells (QP), mucosal defense cells (MD), terminally differentiated cells (TD), reactive oxygen-associated stress cells (RS), and tumor cells (Figure 3E). Using the same method, after removing the two clusters of endothelial and epithelial double cells (Figures 3F, G), we further clustered the macrophages related to PANoptosis into eight subtypes expressing different programs based on the marker genes of myeloid cell subtypes (Figure 3H). Including cDC3-LAMP3, Macro-PLTP, Mast-TPSAB1, Mono-CD14, Monolike-FCN1, pDC-LILRA4, TAM-SPP1 and Unknow (Figure 3I).
Quality control of single-cell data and main cluster clustering and annotation. (A) The number of genes, the number of cells, and the percentages of mitochondria and red blood cells sequenced in the dataset GSE196756. (B) 2,000 highly variable genes in the GSE159677 dataset. (C) The dataset GSE196756 is divided into 16 clusters. (D, E) Cell annotations and 9 types of cells. (F) Top5 Markers of different cell types.
Sub-cluster clustering and annotation of single cell data. (A, B) Epithelial cell subpopulation double cell expulsion (C–E) Epithelial cell subpopulation cell annotation and 5 cell types (F, G) Myeloid cell subpopulation double cell expulsion (H, I) Myeloid cell subpopulation cell annotation and 8 cell types.
Identification of PANoptosis-related differentially expressed genes in esophageal cancer
3.3
Considering that PANoptosis is a unique innate immune inflammatory regulated cell death (RCD) pathway (7), Cibersort analysis indicated that the infiltration levels of 10 types of immune cells in the TCGA database showed statistical differences in tumor tissues. These immune cells include the resting states of B cells naive, T cells CD4 memory activated, NK cells activated, Mast cells resting, and the activated states of T cells CD4 naive, T cells follicular helper, Macrophages M0, Macrophages M1, Dendritic cells activated, Mast cells activated (Figures 4A, B).
Screening of core genes. (A) Comparison of ESCA and normal samples using heat maps of 22 immune cell subsets. (B) Expression of different immune cells in ESCA and normal individuals. (C) Crossover genes of module genes, single-cell differential genes, and TCGA differential genes. (D, E) GO and KEGG enrichment analysis. ESCA, esophageal cancer; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.
Further exploration identified 10,330 ScDEGs at the single-cell level between ESCA and normal samples. In addition, 74 PANoptosis-related differentially expressed genes (POR-DEGs) associated with ESCA were retained through TCGA_DEGs, basic module genes, and ScDEGs (Figure 4C). We continued to conduct functional enrichment analysis to reveal the potential mechanism of POR-DEGs related to ESCA. The results of GO enrichment analysis indicated that POR-DEGs mainly participated in biological processes such as extracellular matrix, apoptosis regulation, and immune cell differentiation, were distributed in cellular components such as extracellular matrix and muscle fibers, and were involved in molecular functions related to tumor cell proliferation such as growth factors and DNA replication (Figure 4D). The results of KEGG enrichment analysis indicated that POR DEGs were mainly involved in signaling pathways such as ECM-receptor interaction, pyrimidine metabolism, and necroptosis (Figure 4E).
Screening of prognostic biomarkers for esophageal cancer
4
To explore the close relationship between POR-DEGs and the survival prognosis of ESCA, univariate Cox regression analysis was used. The results indicated that the four POR-DEGs were significantly associated with the overall survival of ESCA patients (P < 0.05). Among them, CCT6A and GMNN were risk factors associated with poor tumor prognosis, while HSPB6 and TNXB were protective factors associated with good tumor prognosis. These genes were selected for subsequent machine learning (Figure 5A). Using LASSO, RF and SVM, we evaluated the significance of four central genes, namely CCT6A, GMNN, HSPB6 and TNXB, in the prognosis of ESCA.
Screening of hub genes. (A) Univariate Cox regression analysis of risk and protective genes related to survival time (B, C) Cross-validation curves and regression coefficient path diagrams of the LASSO-Cox algorithm. (D, E) The training error, validation error of the model and the importance of each gene in the RF algorithm (F, G) The variation curves of the predicted true value and error value of each gene in the SVM-RFE algorithm. (H, I) Evaluation of the machine learning algorithms in 3. (J) The Venn diagram shows the intersection of the above three analyses. LASSO-Cox, least absolute shrinkage and selection operator penalized Cox proportional risk regression; SVM-RFE, support vector machine recursive feature elimination; RF, random forest; ROC, receiver operating curve.
Further establish the machine learning algorithm model to extract the key genes that are more prognostic for ESCA from POR-DEGs. The characteristics of four genes were retained through LASSO-Cox regression and RF algorithm (Figures 5B–E). In SVM-RFE, the classifier with three genes has the smallest error and the highest classifier accuracy (minimum error = 0.0592, maximum accuracy = 0.941) (Figures 5F, G). The stability and AUC of the three selected machine learning models are good (Figures 5H, I). Subsequently, we combined the three overlapping prognostic genes among these three methods, including CCT6A, GMNN, and HSPB6 (Figure 5J), as the hub genes for subsequent analysis.
Evaluation of the predictive ability of prognostic models for esophageal cancer
3.5
We calculated the risk score based on the relative expression levels of three hub genes and their corresponding multivariate Cox regression coefficients, and then divided them into the high-risk group and the low-risk group according to the median (Figure 6A). As expected, low-risk patients showed better survival rates than high-risk patients (Figure 6B, P < 0.05). The 1-year, 3-year and 4-year mortality rates of the prognostic model were explored using the time-varying ROC curve (Figure 6C). The results show that the area under the curve (AUC) at 1 year, 3 years and 4 years is all greater than 0.65, indicating that the prediction model has good potential in monitoring mortality.
PANoptosis-related prognostic model of ESCA. (A) The risk score distribution, survival status and three hub expression patterns of patients in the high-risk group and the low-risk group. (B) Kaplan-Meier analysis of OS in ESCA patients. (C) Time-dependent ROC curve analysis of hub. (D) Nomograms for predicting 1-year, 3-year and 4-year survival rates. (E) The calibration curve of the Nomogram shows the consistency between the cohort predictions and the observed 1-year, 3-year, and 4-year results. The dotted line at 45° indicates a perfect prediction. The actual performance of the Nomogram is shown in the red, green and blue lines. (F) SHAP interpretable analysis of the prognostic model.
To quantitatively predict the survival probability of individual patients, we further established a prognostic nomogram integrating a prognostic model and multiple clinical variables (Figure 6D). In addition, by applying the prognostic nomogram correction curve, the predicted 1-year, 3-year and 4-year survival rates in the entire TCGA cohort were in good agreement with the observed survival rates (Figure 6E). To compare the predictive capabilities of the predictive model with various clinicopathological factors, we conducted univariate and multivariate analyses of OS for the entire TCGA cohort. Overall, the risk score is more sensitive than other clinicopathological factors and can serve as an independent prognostic predictor (Table 1). SHAP further explained the contribution of the three hub genes to the prognostic model (Figure 6F).
Construction of diagnostic models for esophageal cancer
3.6
The three genes all showed consistent changes in the internal validation of TCGA and the external validation of GSE17351, GSE38129, and GSE29001 in combined batches. CCT6A was highly expressed in tumors (p<0.0001,p<0.0001), GMNN was highly expressed in tumors (p<0.0001,p<0.0001), and HSPB6 was lowly expressed in tumors (p<0.001,p<0.001) (Figures 7A–F). Therefore, a PANoptosis-related diagnostic model was further established and evaluated in ESCA. The confusion matrix indicates that the model can well distinguish ESCA from the control samples (Figures 7G–J), indicating that the accuracy of its classification model establishment is satisfactory. The AUC values of CCT6A, GMNN, and HSPB6 in internal validation were 0.913, 0.868, and 0.782 respectively (Figure 7K), and those in external validation were 0.907, 0.890, and 0.691 respectively (Figure 7L), indicating that the model has good accuracy.
*PANoptosis-related diagnostic model of ESCA. (A–C) The expression levels of CCT6A, GMNN and HSPB6 in the ESCA group and the control group in the internal validation of TCGA. (D–F) Expression levels of CCT6A, GMNN and HSPB6 in the ESCA group and the control group in the combined de-batch external validation of GSE17351, GSE38129 and GSE29001. (G, H) Confusion matrix for the internal validation performance on TCGA dataset. (I, J) Confusion matrix for external validation across three merged cohorts (GSE17351, GSE38129, GSE29001) with batch-effect correction. (K, L) ROC curves of hub genes for diagnostic performance evaluation. ***p<0.001, ***p<0.0001.
Immune infiltration, functional enrichment and distribution of hub genes in single cells
3.7
To further investigate the potential roles of CCT6A, GMNN and HSPB6 in ESCA, we conducted immune microenvironment correlation analysis and single-gene GSEA on prognostic genes. The results showed that CCT6A and GMNN jointly promoted tumor proliferation and migration functions such as “odorant binding” and “olfactory receptor activity”, and inhibited immune-related functions such as “immune receptor activity”. This also explains why high expression of CCT6A and GMNN inhibits the activation of various immune cells (Figures 8A–D); HSPB6 promotes the normal functions of smooth muscle cells such as “muscle system process” and “multivesicular body”, and inhibits cell proliferation functions such as “structural components of ribosome”. The low expression of HSPB6 as a protective factor may indirectly inhibit the immune microenvironment (Figures 8E, F).
*GSEA analysis of hub genes. (A, C, E) Correlation analysis of CCT6A, GMNN and HSPB6 expression with immune infiltration (B, D, F) Differential gene GSEA with high and low expression of CCT6A, GMNN and HSPB6. The distribution of (G, H) CCT6A, GMNN and HSPB6 in each cell. GSEA, gene set enrichment analysis. *<0.05, **p<0.01, ***p<0.0001.
Based on the single-cell atlas, we determined the distribution of three hub genes in nine cell populations (Figures 8G, H). Among them, CCT6A is distributed in various cells, and the expression of CCT6A in most cells shows significant differences between normal tissues and tumor tissues (p < 0.05). GMNN is mainly distributed in epithelial cells and fibroblasts. The expression of GMNN in the two types of cells also shows significant differences in normal tissues and tumor tissues (p < 0.05). HSPB6 is mainly distributed in fibroblasts, and the expression of HSPB6 also shows significant differences between normal tissues and tumor tissues (p < 0.05).
The PANoptotic mechanism of the hub gene CCT6A is related to TAM-SPP1
3.8
Given that myeloid cells (Supplementary Figure 1A) and monocyte/macrophages (Supplementary Figure 1B) exhibited elevated PORG scores driving PANoptosis, and emerging therapeutic strategies are investigating NEI + IFN-induced macrophage PANoptosis to suppress tumorigenesis and tumor growth (28), we further analyzed CCT6A, which is highly expressed in myeloid cells (Figure 8H). Based on the significant increase of TAM-SPP1 and Monolike-FCN1 and the significant decrease of Mono-CD14 in tumor tissues (29) (Figure 9A, Supplementary Figure 1C), the differential expression of CCT6A in TAM-SPP1, Mono-CD14 and Monolike-FCN1, Mono-CD14 in tumor tissues (Figure 9B), the increased infiltration of TAM-SPP1 and Monolike-FCN1 in the high expression group of CCT6A in tumor tissues (Figure 9C), and the fact that CCT6A promotes the M2-like phenotypic change of macrophages by regulating the PI3K-AKT signaling pathway (30). We speculate that this gene may play a key role in regulating the formation of TAM-SPP1 and Monolike-FCN1, and conducted a pseudotemporal analysis of the expression of CCT6A in the cell trajectory of the monocyte/macrophages lineage. To explore the actual situation of tumor tissues, after removing normal tissues, we found that the cell trajectory Mono-CD14 was taken as the starting point (Supplementary Figure 1D), CCT6A sharply increased during the differentiation of Mono-CD14 to TAM-SPP1, and remained stable during the differentiation of TAM-SPP1 to Monolike-FCN1. This indicates that CCT6A may play an important regulatory role in inducing the differentiation of Mono-CD14 to TAM-SPP1 (Figures 9D, E), thereby promoting the occurrence and development of ESCA (31). The Monolike-FCN1 PORGs score was relatively high and functionally presented as M1 type, while the TAM-SPP1 PORGs score was relatively low and functionally presented as M2 type (Supplementary Figures 1B, E, F). Meanwhile, TAM-SPP1 responded strongly to transcriptome disturbance (Supplementary Figure 1G), further confirming that the cancer-promoting function of CCT6A is concentrated on inducing the differentiation of Mono-CD14 into TAM-SPP1. In terms of single-cell mechanism, CCT6A mainly affects the collagen metabolism process, extracellular matrix decomposition and collagen decomposition process during the differentiation of TAM-SPP1, and is a key driving force for shaping the tumor microenvironment required for TAM differentiation (Figure 9F, Supplementary Figure 1H); In terms of transcriptome mechanisms, CCT6A is also correlated with PORGs scores. CCT6A is negatively correlated with the expression of the key promoter of PANoptosis, ZBP1, and positively correlated with the inflammasomal PSTPIP2 and the inflammatory signaling hub TAK1 (Supplementary Figure 1I). The strongest ligand receptor pathway for tumor cells formed by the differentiation of TAM-SPP1 is SPP1/CD44 (Figure 9G). Similarly, in ESCA, SPP1 is mainly secreted by TAM-SPP1 and acts on tumor cells (Figure 9H). The potential prediction of traditional Chinese medicines and molecular docking will import the hub gene CCT6A that regulates macrophages into the Coremine database, and select the top 8 corresponding traditional Chinese medicines. A binding energy value less than -4.25 kcal·mol ^−1^ indicates a certain level of binding ability between the two entities, while a value less than -5.0 kcal·mol ^−1^ indicates a better binding ability. The results indicated that the core target CCT6A had a good binding ability with quercetin (Figure 9H), with a binding energy of -8.3 (Table 2).
*Quercetin targets CCT6A to inhibit TAM-SPP1 formation and may be involved in ESCA PANoptosis through SPP1/CD44. (A) Distribution of myeloid cells in tumor tissues and normal tissues (B) Expression of CCT6A in myeloid cells (C) Proportion of immune infiltration in high and low expression groups of CCT6A (D) Differentiation of Mono-CD14 into Cell trajectory of Monolike-FCN1 (E) Quasi-temporal analysis of CCT6A in cell trajectories (F) Enrichment analysis of perturbed genes after virtual knockout of CCT6A in TAM-SPP1 (G) Ligand-receptor communication between TAM-SPP1 and tumor cells (H) Sending and receiving cells of the SPP1 signaling pathway (I) Molecular docking of CCT6A protein. ***p<0.001, ***p<0.0001.
Analysis of riskscore subgroups
3.9
Further explore the influence of risk score on treatment response. Compared with the low-risk group, the high-risk group demonstrated a lower response to immunotherapy, as reflected in a significant reduction in AZ_IPS scores and a trend of IPS reduction (Figure 10A). According to the Estimate algorithm, the stromalscore, immunescore, and estimatescore of high-risk patients were all lower (Figure 10B), indicating that high-risk patients had less immune infiltration, which was consistent with immune cells. Interestingly, the matrix score also decreased in the high-risk group, indicating that the stroma of patients in the high-risk group was squeezed or replaced by tumor cells. According to the GSEA enrichment analysis of the high and low-risk groups, the immune response regulation of cell surface receptor signaling pathways, white blood cell chemical migration, monocyte migration and other immune responses showed significant enrichment differences between the high and low-risk groups. Patients in the high-risk group inhibited the above immune responses (Figure 10C). In addition, drug sensitivity analysis revealed that patients in the high-risk group could benefit from most chemotherapy drugs, with only five drugs being resistant to them (Figure 10D).
*The impact of risk subgroups on the immune microenvironment. (A) Immunotherapy response of patients in the high and low risk groups (B) Degree of matrix infiltration, degree of immune infiltration and tumor purity of patients in the high and low risk groups (C) Functional enrichment analysis of patients in the high and low risk groups (D) Drug sensitivity analysis of patients in the high and low risk groups (E) Signal afferent and efferent intensity of cells (F) Overview of important cell afferent signals (left: High-risk driver group, right: low-risk inhibition group) (G) Overview of important cell-transmitted signals (left: high-risk driver group, right: low-risk inhibition group) (H) APP signal strength network. (I) GALECTIN signal strength network. *<0.05, **p<0.001.
To further identify the key subgroups causing the differences, we used “AUCell” in the “Seurat” R software package to calculate the score (32) of POR-DEGs in single-cell tumor samples. The cell population was divided into the high-risk driver group and the low-risk inhibition group based on the critical value between the two peaks. We observed that frequent interactions of pDC-LILRA4 were only observed in the high-risk driver group, while this was not the case in the low-risk inhibition group (Figure 10E). Compared with the low-risk inhibition group, the pDC-LILRA4 in the high-risk driver group exhibited more frequent interactions, among which the most important input signal was APP and the most important output signal was GALECTIN (Figures 10F, G). By comparing the differences of two signals between the high-risk driver group and the low-risk inhibition group, we found that in the high-risk driver group, pDC-LILRA4 received more amyloid precursor proteins (APP) released by fibroblasts, Endothelial cells, cancer, etc. (Figure 10H) More GALECTIN is released to communicate with T cells, B cells, TAM-SPP1, etc. (Figure 10I), and the ligand-receptor pairs that contribute the most are APP-CD74 and LGALS9-CD44, respectively. APP-CD74 has been confirmed in recent years to be a key communication between non-immune cells and immune cells in various disease states, participating in amplifying inflammation (33), tumor drug resistance (34), etc. LGALS9-CD44 plays a key role in tumor microenvironment remodeling and immune escape, inhibiting T cell activation (35) and influencing TAM cells (36).
Proteomic IHC validation of hub genes
3.10
The IHC results were consistent with our analysis. CCT6A was distributed in various cells, and its expression in cancer tissues was significantly higher than that in adjacent tissues (p < 0.0001, Figures 11A, B). GMNN is mainly distributed in epithelial cells, and its expression in cancer tissues is significantly higher than that in adjacent tissues (p < 0.05, Figures 11C, D). HSPB6 is mainly distributed in interstitial cells, and its expression in cancer tissues is significantly lower than that in adjacent tissues (p < 0.0001, Figures 11E, F).
*Expression levels of three hub proteins. (A, C, E) Expression of CCT6A, GMNN and HSPB6 proteins in esophageal cancer tissues and adjacent tissues (upper left: adjacent tissues *100, upper right: adjacent tissues *400, lower left: esophageal cancer tissues *100, lower right: esophageal cancer tissues *400). (B, D, F) Statistical differences in the expression of CCT6A, GMNN and HSPB6 proteins between esophageal cancer tissues and adjacent tissues. *p < 0.05, ***p < 0.0001.
Discussion
4
According to the 2022 Global Cancer Observatory (GLOBOCAN) report, esophageal cancer is the 11th most prevalent cancer and the 7th leading cause of cancer death worldwide (2). Despite advancements in treatment methods, the mortality rate of esophageal cancer remains high and the overall prognosis is poor. Multidisciplinary approaches, personalized treatment and innovative technologies can improve the prognosis of patients, so exploring more prognostic biomarkers has clinical value. The assembly of PANoptosome and the activation of PANoptosis occur simultaneously in autoinflammation, infection, neurodegenerative diseases and cancer, and it is crucial for cancer treatment (37). Evidence suggests that targeting PANoptosis can impede immune escape and provide a positive feedback immune activation channel, thereby overcoming immune resistance to refractory tumors (38).
In this study, a series of differentially expressed genes at the transcriptome level and single-cell level were obtained in ESCA by mining the TCGA and GEO databases. These genes were crossed with the genes of the basic modules of PANoptosis to obtain a gene set composed of 74 POR-DEGs. Machine learning algorithms process large amounts of biological data, group it and build models, and are widely used to predict disease markers and therapeutic targets (26). By further leveraging machine learning screening and univariate Cox regression, we ultimately identified three prognostic biomarkers, namely CCT6A, GMNN, and HSPB6.
The Chaperonin Containing TCP-1 complex (CCT) is a cytoplasmic chaperonin complex composed of eight highly conserved subunits (CCT1-CCT8), which plays a key role in maintaining protein homeostasis (39, 40). This complex not only participates in the conformational regulation of cytoskeletal proteins such as actin and tubulin, but also plays a significant pathological role in the malignant transformation and progression of tumors by regulating the folding process of key regulatory proteins such as STAT3 signal transduction factor and VHL tumor suppressor protein (39–44). Functional enrichment confirmed that CCT6A in ESCA also promotes tumor proliferation and inhibits immune responses. As a core regulatory subunit of the CCT complex, CCT6A plays a promoting role in the occurrence and development of tumors such as esophageal cancer (45), colon cancer (46), and lung adenocarcinoma (47). In ESCA, CCT6A is widely distributed in various cells such as T cells, B cells, Epithelial cells, and Myeloid, and the expression of its mRNA and protein is upregulated in tumor tissues. Clinical analysis indicates that high expression of CCT6A in ESCA tissues is significantly associated with poor prognosis in patients, which is consistent with previous research results in ESCC (45).
GMNN is located on chromosome 6 and is an important gene that regulates the cell cycle. GMNN encodes a protein that inhibits the histone acetyltransferase activity of KAT7/HBO1 in a CDT1-dependent manner, suppressing DNA replication permission and histone H4 acetylation to maintain genomic stability and participate in cell proliferation control. Functional enrichment confirmed that GMNN in ESCA also promotes tumor proliferation. As a key proliferation marker, the high expression of GMNN shows significant clinical relevance (48–55) in various malignant tumors. This study found that GMNN was specifically highly expressed in ESCA epithelial cells and fibroblasts, and was significantly associated with a shortened overall survival period of patients. The protein expressed in GMNN was also higher than that in adjacent tissues.
As a member of the sHSP family, HSPB6 is compositively expressed in various tissues, with high expression levels in bones, the heart, and smooth muscle (56). It mainly participates in physiological events such as smooth muscle relaxation (57), platelet function regulation, and cardiac protection (58). Functional enrichment confirmed that HSPB6 in ESCA not only maintains the normal function of smooth muscle but also inhibits abnormal cell proliferation. More and more evidence supports HSPB6 as an inhibitor of tumor growth (59–62). Studies (63–65) have confirmed its protective effect on certain digestive tract tumors, but its role in ESCA remains unclear. In glioma (66), primary lung cell adenocarcinoma (67), high-grade squamous intraepithelial lesion and cervical cancer (68), HSPB6 is significantly downregulated. As an example of a consistently low expression pattern in cancer, HSPB6 seems to have prognostic value in a wide range of cancers (69). The expression of HSPB6 is negatively correlated with the grade of malignant tumors (61). Our study found that in ESCA, the mRNA and protein of HSPB6 expression are significantly decreased in tumor tissues, and are specifically lowly expressed in fibroblasts. Its low expression is associated with a good prognosis for patients.
Stability is crucial to the model (70). The prognostic model established based on three genes was strictly validated using Kaplan-Meier survival analysis, time-dependent ROC curve, and univariate and multivariate Cox regression analysis, demonstrating good prognostic predictive value. Due to the existence of some clinicopathological factors related to OS, we further compared the prognostic risk score with their predictive ability and found that the prognostic risk score can serve as a good and independent prognostic predictor. XGBoost, ROC curve plotting, and analysis of gene expression levels in clinical patients suggest that it has significant diagnostic efficacy.
The rapid development of single-cell technology has provided an important tool for deciphering the interactions among different cell populations within the TME (32). To deeply explore the differences between high-risk and low-risk groups and provide personalized treatment for patients, we predicted the responses to chemical and immunotherapy and compared the types and intensities of cell communication. The association between stronger immune response activation and a favorable prognosis in the low-risk subgroup, as well as the enhanced pDC-LILRA4 immunosuppressive communication and antigen presentation inhibition in the high-risk subgroup, all suggest that the immune status plays a key role in the occurrence, development and prognosis of ESCA. These findings not only provide a theoretical basis for the application of immunotherapy in ESCA, but also lay the foundation for future exploration of personalized treatment strategies based on immune status. Although the sensitivity of patients in the high-risk group to chemotherapy drugs provides support for the position of chemotherapy in ESCA treatment, the limitations of chemotherapy also prompt us to further consider how to optimize the treatment plan to improve the long-term prognosis of patients.
Osteopontin (OPN), encoded by the secreted phosphorylated protein 1 (Spp1) gene, is a multifaceted stromal cell glycoprotein secreted by various immune cells such as macrophages and T cells (71–73). The deficiency of OPN in macrophages is associated with homeostasis imbalance by affecting multiple downstream signaling pathways, such as down-regulation of UCHL1-UPS, ROS production, oxidative stress, mitochondrial-related dysfunction, and subsequent apoptosis (74). Previous studies have shown that SPP1-CD44 and SPP1-PTGER4 have positive interactions with immune cells and mediate crosstalk between tumor cells and macrophages (75, 76). Our research found that CCT6A may promote the differentiation of Mono-CD14 into TAM-SPP1 through matrix remodeling, and this effect may be blocked by quercetin targeting CCT6A. Therefore, we speculate that the high expression of OPN by differentiated TAM-SPP1 reduces oxidative stress and inhibits its own PANoptosis. Meanwhile, paracrine SPP1 activates downstream PI3K/Akt, NF-κB (77) and other signaling pathways through the SPP1-CD44 ligand receptor pathway to inhibit mitochondrial oxidative stress, thereby inhibiting PANoptosis of tumor cells (78). This also explains the source of the high proportion of M2-like TAMs (79, 80) infiltration that can be detected in ESCC. The specific mechanism awaits further discussion through experiments.
Although recent studies have explored PANoptosis (81) in esophageal squamous cell carcinoma and esophageal adenocarcinoma, our work has advanced this field through three key innovations. Firstly, unlike the previous methods of searching for genes, we use WGCNA to identify pan-apoptotic bodies, focusing more on the mechanism of PANoptosis itself. Secondly, the early models mainly stratified prognosis, while we directly linked the high-risk group to the activation of pDC-LILRA4 cells, demonstrating the intercellular interactions behind the immune differences between the high-risk and low-risk groups. Finally, through multi-omics integration, we delved into the cellular mechanisms by which prognostic markers induce PANoptosis and how to treat it. We found that the traditional Chinese medicine component quercetin targeting CCT6A might inhibit PANoptosis by promoting the differentiation of monocytes into TAM-SPP1. Only a few articles explored the mechanism (10).
Although we conducted IHC on each protein in 30 cases of esophageal cancer tissues and 30 cases of adjacent tissues of esophageal cancer, more in-depth mechanism studies are still needed, and some limitations are worth noting. Firstly, the potential inhibition of Monocyte differentiation into TRAMs-SPP1 by quercetin targeting CCT6A remains to be further verified through experiments, which is also the focus of our subsequent work. In addition, the expression level and predictive ability of the hub gene in other tumors should also be taken into account, which is crucial for the subsequent application of the predictive model composed of these three hub genes.
Conclusion
5
In summary, in this study, we systematically identified three prognostic biomarkers related to PANoptosis in ESCA, namely CCT6A, GMNN, and HSPB6, constructed a prognostic and diagnostic model with clinical predictive value, and explored the mechanisms of action of these three prognostic biomarkers. A new mechanism by which CCT6A regulated by quercetin inhibits PANoptosis by promoting the formation of TAM-SPP1 was revealed. In terms of cell experiments, we used a 60-point tissue microarray for IHC and protein expression determination, which confirmed our previous analysis. These findings provide new theoretical basis and transformational directions for the disease prediction, immunotherapy and targeted intervention of ESCA.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Laursen L Schønau MN Bergenholtz HM Siemsen M Christensen M Missel M . Table in the corner: a qualitative study of life situation and perspectives of the everyday lives of esophageal cancer patients in palliative care. BMC Palliat Care. (2019) 18:60. doi: 10.1186/s 12904-019-0445-2, PMID: 31331302 PMC 6647132 · doi ↗ · pubmed ↗
- 2Laursen L Schønau MN Bergenholtz HM Siemsen M Christensen M Missel M . Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. (2024) 74:229–63. doi: 10.3322/caac.21834, PMID: 38572751 · doi ↗ · pubmed ↗
- 3Yang H Wang F Hallemeier CL Lerut T Fu J . Oesophageal cancer. Lancet (London England). (2024) 404:1991–2005. doi: 10.1016/S 0140-6736(24)02226-8, PMID: 39550174 · doi ↗ · pubmed ↗
- 4Malireddi RKS Kesavardhana S Kanneganti T-D . ZBP 1 and TAK 1: master regulators of NLRP 3 inflammasome/pyroptosis, apoptosis, and necroptosis (PAN-optosis). . Front Cell infection Microbiol. (2019) 9:406. doi: 10.3389/fcimb.2019.00406, PMID: 31850239 PMC 6902032 · doi ↗ · pubmed ↗
- 5Yan W-T Zhao W-J Hu X-M Ban X-X Ning W-Y Wan H . PA Noptosis-like cell death in ischemia/reperfusion injury of retinal neurons. Neural regeneration Res. (2023) 18:357–63. doi: 10.4103/1673-5374.346545, PMID: 35900430 PMC 9396479 · doi ↗ · pubmed ↗
- 6Chen W Gullett JM Tweedell RE Kanneganti T-D . Innate immune inflammatory cell death: PA Noptosis and PA Noptosomes in host defense and disease. Eur J Immunol. (2023) 53:e 2250235. doi: 10.1002/eji.202250235, PMID: 36782083 PMC 10423303 · doi ↗ · pubmed ↗
- 7He P Ma Y Wu Y Zhou Q Du H . Exploring PA Noptosis in breast cancer based on sc RNA-seq and bulk-seq. . Front Endocrinol (Lausanne). (2023) 14:1164930. doi: 10.3389/fendo.2023.1164930, PMID: 37455906 PMC 10338225 · doi ↗ · pubmed ↗
- 8Jiang Z Wang J Dao C Zhu M Li Y Liu F . Utilizing a novel model of PA Noptosis-related genes for enhanced prognosis and immune status prediction in kidney renal clear cell carcinoma. Apoptosis. (2024) 29:681–92. doi: 10.1007/s 10495-023-01932-3, PMID: 38281281 · doi ↗ · pubmed ↗