Identification of a Necroptosis-Related Prognostic Signature and Associated Regulatory Axis in Lung Adenocarcinoma

Background Lung cancer is considered to be the second most aggressive and rapidly fatal cancer after breast cancer. Necroptosis, a novel discovered pattern of cell death, is mediated by Receptor-interacting serine/threonine-protein kinase 1 (RIPK1), Receptor-interacting serine/threonine-protein kinase 3 (RIPK3), and Mixed Lineage Kinase Domain Like Pseudokinase (MLKL). Methods For the purpose of developing a prognostic model, Least absolute shrinkage and selection operator (LASSO) Cox regression analysis was conducted. Using Pearson's correlation analysis, we evaluated the correlation between necroptosis-related markers and tumor immune infiltration. A bioinformatics analysis was conducted to construct a necroptosis-related regulatory axis. Results There was a downregulation of most of necroptosis-related genes in lung adenocarcinoma (LUAD) versus lung tissues but an increase in PGAM5, HMGB1, TRAF2, EZH2 levels. We also summarized the Single Nucleotide Variant (SNV) and copy number variation (CNV) of necroptosis-related genes in LUAD. Consensus clustering identified two clusters in LUAD with distinct immune cell infiltration and ESTIMATEScore. Genes related to necroptosis were associated with necroptosis, Tumor necrosis factor (TNF) signaling pathway, and apoptosis according to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. Four prognostic genes (ALDH2, HMGB1, NDRG2, TLR2) were combined to develop a prognostic gene signature for LUAD patients, which was highly accurate in predicting prognosis. Univariate and multivariate analysis identified HMGB1, pT stage, and pN stage as independent factors impacting on LUAD patients' prognosis. A significant correlation was found between the level of TLR2 and NDRG2 and clinical stage, immunity infiltration, and drug resistance. Additionally, the progression of LUAD might be regulated by lncRNA C5orf64/miR-582-5p/NDRG2/TLR2. Conclusion The current bioinformatics analysis identified a necroptosis-related prognostic signature for LUAD and their relation to immunity infiltration. This result requires further investigation.


Introduction
Lung cancer will account for d 228,820 new cases and 135,720 deaths in 2020 all globally [1]. Adenocarcinoma of the lung (LUAD) accounts for more than 40% of all cases of lung cancer [2]. Due to the absence of typical clinical symptoms, patients usually have disseminated metastatic tumors when initially diagnosed with LUAD. Worse still, LUAD is characterized by high aggression and rapidly fatality with overall survival (OS) less than 3 years [3]. Even though smoking has been identified as a risk factor for LUAD, the molecular mechanism is still not understood. And exploration and identification of the potential tumorigenesis molecular mechanism and prognostic markers for LUAD are urgent.
Necroptosis, a novel discovered pattern of cell death, is mediated by Receptor-interacting serine/threonine-protein kinase 1 (RIPK1), Receptor-interacting serine/threonineprotein kinase 3 (RIPK3), and Mixed Lineage Kinase Domain Like Pseudokinase (MLKL) [4,5]. Studies have implicated necroptosis in the pathogenesis of Parkinson, Alzheimer, vascular atherosclerosis, and infectious disease [5][6][7]. The death of T cells and cancer metastasis can also be accelerated by necroptosis, according to recent evidence [8]. On the other hand, when immunotherapy is used to treat malignancies,  CNV.frequency (%)   International Journal of Genomics necroptosis may trigger and amplify antitumor immunity [4]. Necroptosis regulators may also provide prognostic information for cancer and other diseases [9,10]. Moreover, some necroptosis-related prognostic signature had been identified for types of cancer, including stomach adenocarcinoma, breast cancer and cervical squamous cell carcinoma, and endocervical adenocarcinoma [11][12][13]. As such, necroptosis-related genes may also contribute to LUAD prognosis.
Big data mining has been proposed as a promising tool for examining tumorigenesis mechanisms, associated prognosis markers, and therapy targets, following the development of the Cancer Genome Atlas (TCGA). Herein, we systematically investigated the expression of genes associated with necroptosis, their prognostic significance, and their association with immune infiltration. The data may offer another evidence about the vital functions of necroptosis in LUAD.

Datasets and Preprocessing.
The RNA sequencing data (Fragments Per Kilobase of exon model per Million mapped fragments (FPKM) value) and copy number variation (CNV) data of LUAD patients were downloaded from TCGA database. After requiring the clinic characters of LUAD cohort from TCGA, we rearranged it and Supplementary Table 1 showed the detail. Using R (version 4.0.5) and R Bioconductor packages, dataset processing and further analysis were carried out. All the patients losing some clinical information were rejected and a total of 486 cases were obtained. In order to analyze the expression profile, we normalized it to transcripts per kilobase million values.

Consensus Cluster Analysis.
We then conducted consensus cluster analysis using the "ConsensusClusterPlus" package, which was calculated 1,000 times [27]. In the following step, survival and gene expression were analyzed using the packages "survival" and "pheatmap". ESTIMATE algorithm was applied for evaluating the difference of Immunoscore, StromaScore, ESTIMATEScore, and immune cell infiltration in each cluster of LUAD. The immune cell landscape in each cluster of LUAD was generated with "vioplot" package.

Development of Necroptosis-Related Prognostic Gene
Signature. The log-rank test was used to calculate the p -values, and hazard ratio for the necroptosis-related prognostic gene found by Kaplan-Meier analysis. In the following step, the development of the prognostic model was counted on the Least absolute shrinkage and selection operator (LASSO) Cox regression analysis based on necroptosis-    5 International Journal of Genomics related prognostic genes. The risk score of each LUAD patient was calculated by a computational equation (sum of coefficients × necroptosis-related gene expression). TCGA-LUAD patients were separated into two subgroups (lowand high-risk), and the cut-off was the median value of the gene expression. We used Kaplan-Meier analysis to determine the OS curve, and time ROC analysis to determine how well this prognostic signature predicted outcomes. The prognostic risk factor was further analyzed using univariate and multivariate Cox analysis using clinical characteristics and gene signatures. A predicted nomogram was developed to evaluate the predictive performance in OS rate (1-, 3-, and 5-year).

Prognostic Gene Individual Analysis.
Wilcox test was utilized to evaluated expression difference of prognostic signature in different pathological stage of LUAD patients. A correlation was then determined between prognostic signature and immune cell infiltration using Tumor Immune Estimation Resource (TIMER) (https://cistrome.shinyapps .io/timer/). This was followed by tumor mutation burden (TMB) and microsatellite instability (MSI) analysis using Spearman's correlation method. We then collected the IC50 of 265 small molecules in 860 cell lines, and its corresponding necroptosis-related prognostic gene expression from Genomics of Drug Sensitivity in Cancer (GDSC). Using Pearson correlation analysis, the relation between gene expression and drug IC50 was determined.

GO and KEGG Analysis.
We then further confirmed whether these genes were associated with necroptosis in LUAD by performing GO and KEGG pathways analysis. Accordingly, these genes were widely associated with programmed necrotic cell death, necrotic cell death, necroptotic process and NF-kappa B signaling, membrane raft, CD40 receptor complex, transcription coregulator activity, cytokine binding, and Tumor necrosis factor (TNF) receptor superfamily binding in GO analysis (Supplementary Figure 2(a)). Furthermore, these necroptosis-related genes showed widespread association with necroptosis, TNF signaling pathway, NF-kappa B signaling pathway, Nucleotide oligomerization domain (NOD)-like receptor signaling pathway, and apoptosis in KEGG pathway analysis (Supplementary Figure 2(b)).

International Journal of Genomics
Based on these five necroptosis-related prognostic genes, we then developed a prognostic gene signature with LASSO Cox regression analysis. The coefficients of each LUAD case was calculated with the followed computational equation: risk score = sum of coefficients × the expression of necroptosisrelated genes. Finally, based on the result of LASSO Cox regression analysis, the best module was obtained. TLR4 was ejected, and the risk score of patients was calculated by including four other genes in this prognostic signature (Riskscore = (−0.1017) × ALDH2 expression + (0.1559) × HMGB1 expression + (−0.0698) × NDRG2 expression + (−0.0845) × TLR2 expression). The coefficient and partial likelihood deviance of prognostic signature were shown in Figures 4(a) and 4(b). LUAD cohort was divided into high-and low-risk group, and the riskscore, survival status of patients, and gene expression were shown in Figure 4(c). Compared with low-risk group, high-risk group had a worse OS rate (p = 0:000338) with the area under the curves in 1-year, 3-year, and 5-year periods being 0.684, 0.592, and 0.584, respectively (Figures 4(d) and 4(e)).

Predictive Nomogram Based on Prognostic Signature and Clinical Characters.
Considering clinical characters and above four necroptosis-related prognostic genes, we identi-fied HMGB1, pT stage, and pN stage as independent factors impacting on LUAD patients' prognosis in further analysis (univariate and multivariate analysis) (Figures 5(a) and 5(b)). These factors were used to construct a predictive nomogram to predict survival probability, which showed a good prediction ability (Figures 5(c) and 5(d)).

Individual Analysis of Necroptosis-Related Prognostic
Signature. As shown in Figure 6(a), a noteworthy correlation was obtained between pathological stage and TLR2 expression (p = 0:0277) and NDRG2 expression (p = 0:00581), suggesting that TLR2 and NDRG2 may be correlated with the progression of LUAD. And we select TLR2 and NDRG2 for further study. Previous study had suggested that immune infiltration was involved in tumor development and progression in LUAD [31]. The current result demonstrated a positive correlation between the expression of TLR2 and NDRG2 and the immune infiltration level of B cells, CD8+ T cells, CD4+ T cells, macrophage, neutrophils, and dendritic cells (Figure 6(b), all p < 0:05). Moreover, some somatic copy number alterations of TLR2 and NDRG2 could inhibit immune cell infiltration level (Figure 6(c)). TMB and MSI were suggested as predictive marker for 15 International Journal of Genomics cancer immunotherapy, including lung cancer [32][33][34]. Interestingly, the expression of TLR2 and NDRG2 showed significant correlation with TMB score (p = 6:08 × 10 −11 ) and MSI score (p = 9:07 × 10 −18 ) (Figure 6(c)). In MSI analysis, MSI score decreased as NDRG2 expression increased (p = 0:042, Figure 6(d)). To identify cancer immunotherapy target, a vital way is to evaluate the correlation between gene expression and existed drug targets. In order to clarify whether TLR2 and NDRG2 could serve as potential drug screening targets, we explored the correlation between TLR2 and NDRG2 and existed drug targets in GDSC database. Interestingly, the expression of TLR2 and NDRG2 showed positive or negative correlation with GDSC drug sensitivity, including methotrexate and vinblastine ( Figure 6(e)).

Discussion
Previous study had suggested the involvement of necroptosis migration and invasion regulation of tumor [35]. The necroptosis mechanism was proposed as an effective way for eradicating cancer cells [36]. The identification of prognostic value and potential regulatory axis of necroptosisrelated genes will allow necroptosis to be leveraged for therapeutic benefits and prognosis improvement of LUAD.
To confirm whether these necroptosis-related genes were associated with necroptosis in LUAD, we then performed GO and KEGG pathways analysis. As expected, these necroptosis-related genes showed widespread association with necroptosis, programmed necrotic cell death, necroptotic process, and TNF signaling pathway. These functions or pathways could mediate necroptosis and cancer progression. NF-κB signaling may influenced inflammation and the progression of tumor [37]. Signaling mediated by TNF is also crucial for homeostasis and immunity in mammals [38]. Interestingly, TNF was referred as a key mediator in balancing cell survival and necroptosis [39]. LUAD patients were differentiated using consensus clustering based on gene patterns and we identified two clusters, which showed conspicuous difference in pM stage and immune cell characterization. Cluster 1 of LUAD was linked to high Immunoscore, StromaScore, and ESTIMATEScore and abundant immune cell infiltration, referring to hot   [40]. Further moreover, high Immunoscore was significantly correlated with better prognosis in LUAD [40].
Our study also developed a prognostic model based on four prognostic necroptosis-related genes (ALDH2, HMGB1, NDRG2, TLR2), which was highly accurate in predicting LUAD prognosis. Univariate and multivariate analysis identified HMGB1, pT stage, and pN stage as independent factors impacting on LUAD patients' prognosis. Though certain prognostic signatures had been identified for LUAD, our study firstly developed a prognostic model with necroptosis-related markers in LUAD, providing another biomarker in LUAD. A machine learning strategy had constructed and validated a prognostic signature using 12 immune-related genes for LUAD [41]. Another prognostic signature showed good performance in predicting prognosis and reflecting tumor immune microenvironment in LUAD [42]. Jin et al. also constructed a 7-lncRNA prognosis signature and a predictive nomogram in LUAD [43].
The result of our study identified a lncRNA C5orf64/ miR-582-5p/NDRG2/TLR2 regulatory axis in the progression in LUAD. Interestingly, previous study demonstrated lncRNA C5orf64 as a novel biomarker associated with tumor microenvironment and mutation pattern remodeling in LUAD [44]. Moreover, miR-582-5p served as a prognostic biomarker in LUAD and inhibited tumor cell proliferation and invasion [45]. High mRNA level of TLR2 could accelerate tumor progression in LUAD [46]. NDRG2 acted a prognostic marker in LUAD and associated with depth of invasion, vascular invasion, and better OS [47]. Thus, lncRNA C5orf64/miR-582-5p/NDRG2/TLR2 regulatory axis may be involved in the progression in LUAD. This result requires further investigation.
Our study also had some limitations. Not all of 17 necroptosis-related genes were specific to necroptosis. Moreover, the result of consensus clustering analysis is barely satisfactory. This result requires further investigation.

Conclusion
The current bioinformatics analysis identified a necroptosisrelated prognostic signature for LUAD and their relation to immunity infiltration. This result requires further investigation.

Data Availability
The analyzed data sets generated during the study are available from the corresponding author on reasonable request.