Machine learning reveals PANoptosis as a potential reporter and prognostic revealer of tumour microenvironment in lung adenocarcinoma

Lung adenocarcinoma (LUAD), a prominent lung cancer subtype, has an underexplored relationship with PANoptosis, a recently discovered mode of tumour cell death. This study incorporated iron death, copper death, scorch death, necrotizing apoptosis and bisulfide death into a pan‐death gene set (PANoptosis) and conducted single‐cell analysis of scRNA‐seq data from 11 LUAD samples. Differentially expressed genes were identified, and Gene Ontology and Kyoto Encyclopedia of Genes and Genomes analyses were performed. Univariate COX regression and least absolute shrinkage and selection operator (LASSO) regression were used to screen PANoptosis key genes for constructing an LUAD risk model. The model’s prognostic performance was evaluated using survival curves, risk scores and validation in the Gene Expression Omnibus database. The study also explored the correlation between risk scores, tumour biological function, immunotherapy, drug sensitivity and immune infiltration. The SMS gene in the PANoptosis model was silenced in two LUAD cell lines for cellular validation. Single‐cell analysis revealed eight major cell types and several PANoptosis genes significantly associated with LUAD survival. The risk model demonstrated strong prognostic performance and association with immune infiltration, suggesting PANoptosis involvement in LUAD tumour immunity. Cellular validation further supported these findings. The PANoptosis key risk genes are believed to be closely related to the tumour microenvironment and immune regulation of LUAD, potentially providing valuable insights for early diagnosis and clinical treatment, and broader applications in other tumours and complex diseases.

younger age and is relatively more common in women.The biology of LUAD is that it is generally slow to develop, and it is detected by chest imaging, sometimes with early hematogenous metastases. 4,5The detection of the particular pathological substances requires special staining, particularly in poorly differentiated tumours. 6,7If lung cancer is not controlled in time and deteriorates rapidly, it can lead to chest pain and pleural fluid with much pressure on the lungs, making it difficult for patients to breathe and making the disease more difficult to treat, posing a great threat to patients' lives.Early diagnosis of LUAD relies mainly on imaging and serum tumour marker tests.The longterm prognosis for patients with mid-to late-stage LUAD is poor, and many patients are lost to surgery at the time of presentation.Targeted therapies and immunotherapy are currently a hot topic in oncology research and have now been shown in clinical trials to significantly improve the five-year survival rate for some patients. 8,9Nonetheless, given the considerable tumour heterogeneity in LUAD patients, selecting appropriate candidates for targeted immunotherapy remains crucial.Consequently, investigating the molecular mechanisms underlying LUAD and identifying novel therapeutic targets through bioinformatics analysis constitute significant areas of focus for future research into innovative LUAD therapies.
Pyroptosis, apoptosis and necroptosis are three key programmed cell death pathways with certain molecular and genetic characteristics. 10PANoptosis is an emerging concept that highlights the crosstalk and coordination between these three pathways. 10,11It is able to eliminate intracellular pathogens and maintain homeostasis in the body. 12,13Originally proposed in 2019, although scorch death, apoptosis and necroptosis have been described as distinct and independent pathways throughout the history of research, there is growing evidence of extensive interactions between these programmed cell death pathways.The interactions between each pair of pathways have now been clarified (apoptosis vs. apoptosis, apoptosis vs. necroptosis, and apoptosis vs. necroptosis), and a growing number of studies have mechanistically defined the interactions between the three pathways.These findings led to the establishment of PANoptosis.Early studies of cell death focused on the distinct programmes and biochemical functions behind these separate mechanisms; however, more recent studies have highlighted the crosstalk and redundancy involved.In cells with certain infection conditions, such as influenza virus-infected macrophages, the three types of cell death are not activated independently of each other; instead, they act simultaneously. 14Thus, the mechanism of PANoptosis-induced regulated cell death has attracted considerable attention and has become a hot topic of research in the field of tumour therapy.
Unfortunately, there is a lack of reported LUAD-related biomarkers based on PANoptosis.
Taking into account that PANoptosis is presently a prominent subject in the field of tumour-associated cell death research, it is likely to be intricately linked to the progression of LUAD.Therefore, in this study, we obtained key PANoptosis-related genes in LUAD by bioinformatics and used them as the basis for constructing a prognostic model.The constructed PANoptosis-related risk scores were also validated in relation to each other in single cell samples of LUAD.
Specifically, based on the potential functions of the PANoptosis key genes, experimental targets can be identified for in-depth study of the mechanisms of action of PANoptosis associated with LUAD while providing effective solutions for the early diagnosis and clinical treatment of LUAD, as well as suggesting approaches that can be applied to the study of other tumours and complex diseases.

| Data acquisition
The LUAD transcriptome data were sourced from The Cancer Genome Atlas Program (TCGA) and Gene Expression Omnibus (GEO) databases.The genome-wide expression profiling microarray dataset containing lung adenocarcinoma tissue with paired normal paracancerous tissue was downloaded.The TCGA training set contained a total of 503 files, each file corresponding to one sample, which included lung adenocarcinoma tissue and paracancerous tissue.The mRNA expression matrix data, encompassing 85 LUAD patients, as well as the associated clinical dataset (GSE30219), were procured from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) to serve as a validation set, facilitating the evaluation of the model's robustness and stability.Adhering to the rigorous standards of academic medical English, both TCGA and the GSE72094 datasets stringently conformed to identical inclusion and exclusion criteria, ensuring the consistency and comparability of the cohorts under investigation.
The inclusion criteria for the dataset were: (i) the number of samples was >30; (ii) tumour specimens were included; and (iii) clinical information and survival time were complete.Exclusion criteria were: (i) tumour pathological type other than lung adenocarcinoma; and (ii) incomplete clinical information such as relevant stage, survival time and gender.

| Single-cell spatial depiction of LUAD and differential annotation characterisation of PANoptosis
Data from 11 patients with LUAD from GSE131907 were acquired and subjected to single-cell transcriptomic spatial analysis.Utilizing the Seurat R package, the scRNA-seq analysis method encompasses conducting data quality control, including filtering low-quality cells, normalizing gene expression and mitigating batch effects for reliable downstream analyses.The 50 Hallmark gene sets were obtained from the Molecular Signatures Database.Gene Set Variation Analysis (GSVA) was employed to compute the enrichment scores for each pathway, providing a quantitative assessment of pathway activity.
The GSVA enriched pathway scores were calculated for eight cell populations using 50 Hallmark datasets.The AddModuleScore function of the Seruat package scored the PANoptosis pathway for gene set enrichment to obtain the PANoptosis score.Further, we wanted to obtain LUAD-associated PANoptosis differential genes.
The FindAllmarkers function of the Seruat package was applied to compare the two groups of high-panoptosis and low-panoptosis to screen for differentially expressed genes (DEGs), with adjusted p-values < 0.05 and absolute values of logFC > 0.585.The R package "clusterProfiler" (version 4.0.5) was used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of DEGs.

| PANoptosis risk gene screening and prognostic modelling of LUAD
Univariate COX regression was performed on the DEGs obtained above using the "survival" package, to assess the prognostic association between each risk gene and overall survival (OS).Hazard ratios (HRs) and 95% confidence intervals were calculated for each gene, with p < 0.01 being statistically significant.A value of HR > 1 indicates that the gene was a high-risk pathogenic gene for lung adenocarcinoma and a value of HR < 1 indicates that the gene was a protective prognostic gene for lung adenocarcinoma.The genes obtained from the univariate COX analysis were further included in the least absolute shrinkage and selection operator (LASSO) regression analysis, and the PANoptosis-related prognostic genes were screened and their regression coefficients were calculated to construct a prognostic risk model.All patients in the TCGA-LUAD cohort were divided into highand low-risk subgroups based on the median risk score of the model.
In order to assess the predictive value of the model for patients with lung adenocarcinoma, Kaplan-Meier survival curve analysis was first performed using the "survival" and "Survminer" packages to compare the difference in survival prognosis based on OS between the highand low-risk groups.
For the external validation of this prognostic model, the GSE30219 lung adenocarcinoma cohort from the GEO database was employed as the testing group.The test group was divided into lowand high-risk subgroups by applying the median risk score of the TCGA-LUAD cohort.Kaplan-Meier survival analysis was performed to validate the stability of the model.

| In-depth interpretation of the PANoptosis risk model for immune infiltration and assessment of drug sensitivity
Given that cell death modalities are generally linked to immune interactions within the tumour, two methods were employed to compute immune infiltration scores: single-sample Gene Set Enrichment Analysis (ssGSEA) and the xCell algorithm.These scores were visualized using box plots, heatmaps and scatter plots.ssGSEA calculates the enrichment scores for individual samples and gene set pairs to ascertain the extent of immune infiltration.Meanwhile, xCell quantifies the abundance of 67 immune cell types utilizing transcriptomic data.In addition, a mutational display of the TCGA-LUAD cohort was performed.Mutation data from the low-risk and high-risk groups were visualised using the R package "maftools" (version 2.12.0).In order to further enrich the clinical content of the model and increase its application value, we also calculated sensitivity scores for the drugs in the GDSC database based on the R package "oncoPredict".

2.
Relative expression of SMS gene-cells from each group were digested with trypsin after transfection for 48 h.Total protein was extracted using the protein extraction reagent and the protein concentration was determined using the bicinchoninic acid method.A 50 μg aliquot of protein was separated by gel electrophoresis and blocked with 5% skimmed milk powder for 2 h.The primary antibody (rabbit anti-human SMS, 1:1000 dilution) was added dropwise and incubated overnight at 4 C.The secondary antibody was added and incubated at room temperature for 2 h.Colour developer was added for 20 min in a dark room.GAPDH was used as the internal reference protein, and the expression of the target protein was expressed as the grey value of the target band/the grey value of the internal reference protein.
3. Cell cycle assay and apoptosis assay-after 48 h of culture, each group of cells was collected and stained using a cell cycle assay kit (Biosharp, China), and the cell cycle assay was performed by flow cytometry.Apoptosis was detected in a similar way using an apoptosis kit.Experiments were repeated more than 3 times.The results of the experiments were fitted using FlowJo software.Cell cycle analysis was performed in MultiCycle analysis software and the relative ratios of G0/G1, S and G2/M phases were calculated separately.

Clone formation-H1395 and NCI-H1975 cells at logarithmic
growth stage were inoculated in six-well plates and cultured for 24 h.Clone formation was compared between the SI-NC, SI-SMS-1 and SI-SMS-2 groups of the two cell lines.For fixing, 4% paraformaldehyde was used and 0.1% crystalline violet was used for staining.The numbers of clone formations in the SI-NC, SI-SMS-1 and SI-SMS-2 groups were compared in the two cell lines, respectively.

| Statistical analysis
All statistical analyses were performed through the R software (4.0.1).R package "survival" provides functions to draw Kaplan-Meier curves, estimate survival, and perform COX regression analysis.Univariate and multivariate COX regression analysis, along with a statistical test with a p-value less than 0.05, to determine significant prognostic factors.Western blot, flow cytometry and clonal formation assay were all repeated three times.

| Single-cell spatial depiction of LUAD and functional characterisation of PANoptosisassociated genes
The scRNA-seq of LUAD patients was down-annotated and divided into different cell populations.umap methods were successfully performed for cell clustering and downstream analysis.Cells were annotated by clusters (Figure 1A) and the main cell types included B lymphocytes, myeloid and endothelial cells, NK cells, epithelial cells, fibroblasts, mast cells and T lymphocytes.T lymphocytes and myeloid cells were expressed at higher levels, and endothelial and NK cells were annotated at lower levels.In addition, we analysed the annotation share of single cell analysis in 11 LUAD patients (Figure 1B).The findings demonstrated a significant difference in overall survival between patients classified as high-risk and those in the low-risk groups.Survival timeline plots and scatter plots similarly demonstrated this overall difference in distribution between the two groups (Figure 4A).In the dataset of 85 patient with lung adenocarcinoma in GSE30219, we similarly verified a significant difference in prognosis between the two groups (Figure 4B).This suggests that the PANoptosis risk model is a valid stratification indicator for the prognosis of LUAD patients.The mechanism behind its sensitive stratification performance, however, needs to be further investigated.We also further described the prognostic assessment value of each key gene in the risk model for LUAD using forest plots (Figure 4C).The results showed that MS4A1, CRNDE and SMS had the most significant survival stratification value for LUAD patients.The correlation heat map demonstrated the expression correlation of PANoptosis genes obtained from the screening (Figure 4D), with a high positive correlation of expression for SFTPB, SFTA3 and CRNDE.In addition, we also plotted the nomogram to further enrich the value of the model for clinical application (Figure 4E).

| Chromosomal distribution and mutational analysis of PANoptosis risk genes
We further depicted the distribution of the six PANoptosis risk genes on chromosomes (Figure 5A).Interestingly, SMS is located on chromosome X.SFTPB and MS4A1 are located on chromosomes 2 and 11, respectively.KCTD12 and SFTA3 are located on chromosomes 13 and 14, respectively.CRNDE is located on chromosome 16.6C).This included activated B cells, immature B cells, eosinophil cells, activated CD4 T cells, mast cells and immature dendritic cells.The results showed that risk score showed a significant positive expression correlation with activated CD4 T cells and a significant negative expression correlation with several other cells.In addition, we used the xCell algorithm to clarify this association even more (Figure S1).The results showed significant expression correlations between all six genes constituting the PANoptosis risk model and multiple immune cells.This suggests that immune infiltration and immune modulation-related pathways are the main ways in which PANoptosis genes mediate alterations in the tumour microenvironment.
Immediately following this, we performed a more in-depth drug sensitivity assessment.Figure 7A shows a heat map of the correlation between the six modelled key genes and drug sensitivity.The results show that key genes for PANoptosis are significantly associated with drug treatment sensitivity.In particular, KCTD12 and MS4A1 were significantly positively correlated with the therapeutic sensitivity of   multiple chemotherapeutic drugs.Moreover, there was high concordance between the two genes.In addition, we also plotted the correlation points between several important chemotherapeutic drugs and PANoptosis scores (Figure 7B).The results showed that Bl-2536_1086, Tozasertib_1096, PF-4708671_1129, RO-3306_1052 and GSK269962A_1192 showed a significant negative correlation between the treatment sensitivity of these drugs and the risk score.A number of chemotherapeutic agents, including AZD2014_1441, Vori-nostat_1012, Crizotinib_1083, PFI3_1620 and Palbociclib_1054 showed a significant positive correlation with the risk score.

| Cellular intervention assay of SMS gene on LUAD propagation and invasion
The immunoblotting results showed (Figure 8A,B) that the protein expression levels of SMS were significantly lower in both Si-SMS-1 and Si-SMS-2 groups of both cell lines than in the blank knockdown control group ( p < 0.001).This demonstrated the successful construction of a cellular model of SMS silencing.It facilitated our subsequent further analysis of the effect of the SMS gene on the function of LUAD cells.The cell cycle alteration of LUAD cells after SMS silencing was detected by flow cytometry (Figure 8C,D).The G0/G1 phase ratio was significantly increased in the Si-SMS-1 and Si-SMS-2 groups compared with the control group, and the difference was statistically significant.The above experimental results indicate that silencing the expression of the SMS gene stalled the cell cycle in G0/G1 phase.In both cell lines, the silencing of SMS resulted in a higher percentage of apoptotic cells in the LUAD cell line (Figure 8E).The significance of this difference was further revealed by the histograms of the different groups (Figure 8F).The colony formation assay showed that the colony formation numbers of Si-SMS-1 and SI-SMS-2 were significantly lower than those of the Si-NC group (Figure 8G,H).This suggests that SMS can also affect the cell

| DISCUSSION
Lung cancer is currently the leading cause of cancer death. 15,16The development of LUAD, the main type of non-small cell lung cancer, 17 is not only associated with abnormal expression of proto-oncogenes and oncogenes and gene mutations in the classical sense, 18 but is also closely related to the immune system.Many studies have proposed that the cause of the efficacy of anti-immunotherapy should be sought in the tumour tissue, 19 hence the urgent need to find new biomarkers.In addition, in lung cancer there may be a chronic inflammatory response leading to differentiation bias of immune cells, thus favouring immune escape as well as resistance to immunotherapy. 20,21[24] Immunotherapy not only improves the survival of tumour patients, but also has fewer side effects than conventional chemotherapy and can be combined with chemotherapy to enhance its benefits.Immunotherapy against immune checkpoint inhibitors such as PD-1/PD-L1 has been used as a first-and second-line treatment for non-small cell lung cancer. 25However, as the overall response rate to immunotherapy is poor, 26 there is a need to find biomarkers and predictive models for the diagnosis and treatment of LUAD.
We analysed the differences between TCGA-LUAD data and normal tissue data.The PANoptosis-related differential genes were obtained by screening and identifying the PANoptosis gene set.Fiftyeight genes associated with overall survival were selected by univariate COX analysis, and finally a PANoptosis-related prognostic model was constructed based on LASSO regression.We found that the overall survival of patients in the experimental group was significantly lower in the high-risk group compared with the low-risk group, and the same results were obtained in the independent validation group of the GEO database, demonstrating the stability of the model.The providing a strong link between polyamine metabolic pathways and oncogenic signalling pathways in tumourigenesis.However, reports of SMS genes promoting tumours are mainly in colorectal cancer, particularly its cooperation with MYC over a different regulatory pathway to suppress Bim expression.Its potential role for LUAD has yet to be reported in related mechanistic studies. 27Noptosis is a recently discovered inflammatory regulated cell death pathway regulated by a cytoplasmic polyprotein complex called PANoptosome. 28This complex has key features of apoptosis, pyroptosis and/or necroptosis, but cannot be explained by any of these three regulated cell death pathways alone.PANoptosis is commonly associated with cytokine storms and has been found to be involved in the pathogenic processes of a variety of human diseases, particularly in the context of microbial infections.However, the specific role of In conclusion, we constructed a prognostic model on PANoptosis consisting of six genes to achieve the survival stratification of LUAD.
The predictive and independent prognostic value of the model was also validated using the independent validation set of GEO, demonstrating that the model obtained in this study is a reliable biomarker for lung adenocarcinoma.In addition, the model was demonstrated not only to have prognostic value but also to be able to

1 .
Cell culture and SMS silencing-two lung adenocarcinoma cell lines, H1395 and NCI-H1975 from the American Type Culture Collection (ATCC) repository, were used.The culture conditions were 37 C and 5% CO 2 .When the cells grew to about 85% of the bottom of the bottle, they were digested with trypsin and passaged.The logarithmic growth phase cells were inoculated in multi-well culture plates and transfected using lipofectamine 2000.The cells were grouped according to the transfection sequence: (1) si-NC group, transfected with the negative control sequence; (2) si-SMS-1 group, transfected with SMS-specific knockdown sequence; and (3) si-SMS-2 group, transfected with SMS-specific knockdown sequence.Each group was incubated for 48 h.
lung_t34 was predominantly distributed in the epithelial cell region, while other patients possessed a mixed distribution of multiple cells.To further clarify the distribution of possible PANoptosis-related gene and pathway expression in LUAD samples, single cell samples were annotated by PANoptosis-specific markers, as shown in Figure 1C.The markers included were CD68, CD8A, CD79A, NKG7, TPSAB1, LUM, RAMP2 and SFN.The umap distribution of typical markers had unique characteristics, with CD68 highly overlapping with myeloid expression regions, CD8A being a significant enrichment marker for T lymphocytes, B lymphocytes mainly co-expressed with CD79A and NKG7 being significantly expressed in both T and NK cells.The ggrain package-based cloud and rain map (Figure 2A) visually represents the distribution of PANoptosis gene expression values across different samples, facilitating comparative analysis.PANoptosis gene expression was relatively high in myeloid as well as mast cells, while PANoptosis expression was low in B lymphocytes as well as NK cells.In addition, we also used stacked plots to depict the percentage of different cell populations in 11 lung adenocarcinoma patients (Figure2B).A similar cell expression pattern can be seen in several patients, with T-lymphocytes and myeloid cells being the predominant cell types, except in LUNG_T34, where epithelial cells are the predominant type.Further, we visualised the expression of PANoptosis gene set scores in different cell types using umap plots (Figure2C).Heat maps of GSVA scores based on hallmark pathways for the eight cell populations (Figure2D) showed that HALLMARK ALLOGRAFT REJECTION was the major enrichment pathway for B lymphocytes.Endothelial cells were associated with angiogenesis and fibroblasts were mainly enriched for angiogenesis and epithelial mesenchymal transition.Mast cells were associated with androgen response.Myeloid highly expressed IL6-JAK-STAT3-SIGNALING. NK cells and T lymphocytes were relatively less enriched in the pathway.To gain deeper insights into the relationship between PANoptosis and tumour biological function, as well as enrichment pathways, additional GO and KEGG analysis were conducted.Gene Ontology analysis (Figure3A) showed that the major pathways of PANoptosis were enriched in virus receptor activity, tertiary receptor activity, immune receptor activity, exogenous protein binding, cargo receptor activity antioxidant activity, vacuolar membrane, regulation of T cell activation, regulation of cell-cell adhesion, positive regulation of leukocyte activation and positive regulation of cell activation.A further KEGG (Figure 3B) analysis shows that the role of PANoptosis in LUAD may be related to those of viral myocarditis, th17 cell differentiation, phagosomes, lysosomes and human T-cell leukaemia virus infection.

3. 2 |
Screening of PANoptosis risk genes and construction of prognostic model for LUADBased on the univariate COX regression model ( p < 0.01), a total of 53 PANoptosis-related genes were found to be significantly associated with OS.The TCGA-LUAD cohort was used as the training set, and six genes with prognostic significance were selected from the 53 genes according to the LASSO regression algorithm and prognostic models were constructed.The feature screening process of LASSO regression is shown in Figure3C,D.The final PANoptosis-related genes that formed the risk model included SFTPB, MS4A1, KCTD12, SFTA3, CRNDE and SMS.The median risk score was used to distinguish the low-risk group from the high-risk group and to establish the risk score score.The risk score formula was: SFTPB * (À0.004) + MS4A1 * (À0.111) + KCTD12 * (À0.069) + SFTA3 * (À0.031) + CRNDE * (À0.188) + SMS * 0.237.The box plot shows (Figure 3E) that SFTPB, MS4A1, KCTD12, SFTA3 and CRNDE were all protective genes for LUAD, while SMS was the only risk gene in the model.In the TCGA-LUAD training cohort, the survival stratification performance of the model was evaluated utilizing Kaplan-Meier curves.

F I G U R E 1
Single-cell spatial depiction of lung adenocarcinoma (LUAD) and spatial annotation of PANoptosis-associated genes.(A) umap plots of the eight cell populations of LUAD obtained after descending fractionation; (B) umap plots of cell type occupancy of 11 lung adenocarcinoma patients; and (C) umap plots of PANoptosis typical marker genes, typical markers include CD68, CD8A, CD79A, NKG7, TPSAB1, LUM, RAMP2 and SFN.On this basis, Figure 5B,C shows mainly the base pair mutation patterns in the low-risk and high-risk groups.In the low-risk group for PANoptosis, C > A,C > T,C > G,T > C were the predominant mutation patterns; in patients in the high-risk group, mutations of C > A,C > T,C > G,T > A were more common.In addition, we analysed the major mutations in the low-risk and high-risk groups of PANoptosis (Figure5D,E).The results showed that TP53, TTN, MUC16 and CSMD3 were the most commonly mutated genes in both the lowand high-risk groups.

3. 4 |
Immune infiltration elucidation and drug sensitivity assessment in PANoptosis risk modelsRegardless of iron death, apoptosis or scorch death, changes in the mode of tumour death are inevitably accompanied by microscopic impacts on the tumour immune microenvironment.Therefore, we hypothesized that the effect of the PANoptosis risk model on LUAD may be mediated through altered tumour immunity.To this end, the ssGSEA and xCell algorithms were used to analyse the specific effects F I G U R E 2 Distribution exhibition and functional analysis of PANoptosis-related gene markers.(A) Cloud and rain map of PANoptosis gene set scores in different cell populations; (B) stacked map of the percentage of different cell populations in 11 lung adenocarcinoma patients; (C) umap of PANoptosis gene set expression; and (D) heat map of Gene Set Variation Analysis (GSVA) scores based on Hallmark pathway in eight cell populations. of PANoptosis scores and associated genes on the expression of tumour immune cells and immune pathways.The ssGSEA algorithm results are shown in Figure 6A for activated B cells, activated CD8 T cells, activated dendritic cells, eosinophil cells, MDSC, T follicular helper cells and Type1 helper T cells, which were significantly more highly expressed in the low-risk group than in the high-risk group for PANoptosis.In contrast, levels of activated CD4 T cells and Gamma delta T cells were significantly higher in the high-risk group.This suggests that the PANoptosis model can achieve stratified prediction of multiple tumour immune cells.This reinforces our hypothesis that immune cells of the tumour microenvironment may be the main point of action for PANoptosis, even though this may require more basic experimental validation in the future.Figure 6B demonstrates the distribution of immune infiltration abundance of 23 immune cells obtained by ssGSEA in different patients.Further, we also analysed the correlation between risk score and different immune cells of the immune microenvironment (Figure

F I G U R E 3
Biological pathway enrichment and model construction of PANoptosis-related genes.(A) Gene Ontology (GO) enrichment strip plot of PANoptosis-associated genes; (B) Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment point plot of PANoptosis-associated genes; (C) binominal deviation plot of LASSO for risk gene screening; (D) coefficient acquisition of LASSO for risk gene screening; (E) box line plots for expression of key genes in two groups of high-risk and low-risk.F I G U R E 4 Efficacy validation of risk models for PANoptosis-associated gene model.(A) Kaplan-Meier survival curves, line plots, and scatter plots demonstrating the prognostic stratification efficacy of the risk model in the TCGA-LUAD database; (B) Kaplan-Meier survival curves, line plots and scatter plots demonstrating the prognostic stratification efficacy of the risk model in the GSE30219 database; (C) forest plot of multifactorial COX analysis for six key genes; (D) correlation corrplot of six key genes; and (E) nomogram of prognostic models for PANoptosisrelated genes.

F I G U R E 5
Chromosomal distribution and mutation analysis of PANoptosis risk genes.(A) Chromosomal distribution of the six PANoptosisassociated key genes; (B) base pair mutation probability plot in the PANoptosis low-risk group; (C) base pair mutation probability plot in the PANoptosis high-risk group; (D) mutation waterfall plot and mutation-type distribution of top 20 genes in the PANoptosis low-risk group; and (E) mutation waterfall and distribution of mutation types of top 20 genes in the PANoptosis high-risk group.

F I G U R E 6
The single-sample Gene Set Enrichment Analysis (ssGSEA) method reveals tumour immune infiltration correlation in PANoptosis risk models.(A) Box line plots based on the ssGSEA immune infiltration algorithm for the high-risk and low-risk groups; (B) stacked plots of the distribution of the abundance of immune infiltration of 23 immune cells in different patients by ssGSEA; and (C) scatter plots of the correlations of ssGSEA with different immune cells of the immune microenvironment (including activated B cells, immature B cells eosinophil, activated CD4 T-cells, mast cells and immature dendritic cells).

F
I G U R E 7 Functional assessment of drug sensitivity stratification in PANoptosis risk models.(A) Heat map of correlations between the six modelled key genes and drug sensitivity-related drugs; (B) point plots of correlations between multiple chemotherapy drug sensitivity scores and risk score.

F I G U R E 8
Cellular intervention assay of SMS gene on LUAD propagation invasion.(A) Western Blotting (WB) bands of SMS gene and internal reference expression, corresponding to two LUAD cell lines of the SI-NC, SI-SMS-1 and SI-SMS-2 groups; (B) histogram of SMS gene expression obtained by WB; (C) flow chart of cell cycle; (D) proportion of cell cycle in different groups; (E) flow chart of apoptosis; (F) histogram of percentage of apoptotic cells in different groups; (G) representative colony formation experiments for different groups; and (H) colony formation histogram for different groups.proliferation of lung adenocarcinoma cells.In summary, inhibition of SMS expression in LUAD cells inhibited cell proliferation activity and reduced cell migration and invasion ability, which laid the foundation for the specific regulatory mechanism of PANoptosis intervention in LUAD.
genes comprising the model could be used as independent risk markers for LUAD.Meanwhile, immune infiltration analysis revealed that the risk scores of the PANoptosis risk model were significantly correlated with the expression of multiple immune cells and immune pathways.The ssGSEA and xCell algorithms are computational methods for characterizing immune cell composition within bulk gene expression data.ssGSEA offers a comprehensive, unbiased approach to quantify pathway activity, while xCell employs a gene signaturebased strategy to estimate cell-type abundance.Each algorithm has unique advantages, with xCell providing more accurate cell-type deconvolution and ssGSEA enabling broader, context-specific pathway assessment.These differences render them complementary tools for investigating immune infiltration in various disease contexts.This suggests that the PANoptosis risk model may regulate tumour development in a way that reveals alterations in the immune microenvironment of the tumour.The model includes six genes with prognostic value, of which the positive correlation with risk factors is SMS and the negative correlations are SFTPB, MS4A1, KCTD12, SFTA3 and CRNDE.SMS is mainly associated with polyamine synthesis.Polyamine biosynthesis is frequently dysregulated in human cancers, particularly in relation to early tumour development.Recent studies have brought polyamine metabolism back into the spotlight, PANoptosis in LUAD remains to be undiscovered.Our understanding of the molecular mechanisms of PANoptosis and PANoptosome assembly is still very limited, but several important regulatory targets have been identified, including absence in melanoma 2 (AIM2), members of the Caspase s family (including CASP3, CASP6 and CASP8), Z-DNAbinding protein 1 (ZBP1), receptor interacting proteinkinase1/3 (RIPK1/3) and interferon regulatory factors (IRF1).29,30The discovery of these markers will provide breakthroughs in the optimization of subsequent cancer treatment strategies.This study presents several limitations.Firstly, the limited number of LUAD samples analysed may reduce the generalizability of the results.Secondly, the construction and validation of the PANoptosis risk model relied solely on retrospective data, which may introduce biases.Thirdly, the functional roles and mechanisms of the identified key PANoptosis genes in LUAD remain unclear, necessitating further experimental validation.Lastly, the study's focus on PANoptosis, a recently discovered cell death mechanism, may overlook potential interactions with other established cell death pathways and their influence on LUAD progression and immune regulation.The relationship between the PANoptosis risk model and therapeutic sensitivity of chemotherapeutic agents is crucial to its potential clinical application, as it may aid in predicting patients' response to treatment, optimizing drug selection and improving personalized therapy strategies for lung adenocarcinoma patients.
predict the level of immune infiltration in lung adenocarcinoma patients by xCell as well as ssGSEA scores.The PANoptosis risk model's relationship with chemotherapeutic sensitivity could enhance personalized therapy for lung adenocarcinoma patients.It provides a new research direction and expands ideas for the diagnosis and treatment of lung adenocarcinoma.AUTHOR CONTRIBUTIONS Chunyan Han was the main person in charge of this paper.Quzhen Danzeng, Liang Li, Shanwang Bai and Chunyan Han participated in the writing, data analysis and figure drawing of this paper.Professor Cuixia Zheng was responsible for the supervision and revision of this article.