Integrative Analysis of Pyroptosis-Related Prognostic Signature and Immunological Infiltration in Lung Squamous Cell Carcinoma

Background Lung cancer is one of leading causes of human health threatening with approximately 2.09 million initially diagnosed cases and 1.76 million deaths worldwide annually. Pyroptosis is a programmed cell death mediated by Gasdermin family proteins. Pyroptosis could suppress the tumor oncogenesis and progression; nevertheless, pyroptosis could promote tumor growth by forming a suitable microenvironment. Methods LASSO Cox regression analysis was performed to construct prognostic pyroptosis-related gene (PRG) signature. A ceRNA was constructed to explore the potential lncRNA-miRNA-mRNA regulatory axis in LUSC. Results The expression of 26 PRGs were increased or decreased in LUSC. We also summarized simple nucleotide variation and copy number variation landscape of PRGs in LUSC. Prognosis analysis suggested a poor overall survival rate in LUSC patients with high expression of IL6, IL1B, ELANE, and CASP6. A pyroptosis-related prognostic signature was developed based on four prognostic PRGs. High-risk score LUSC patients had a poor overall survival rate versus low-risk score patients with an AUC of 0.565, 0.641, and 0.619 in 1-year, 3-year, and 5-year ROC curves, respectively. Moreover, the risk score was correlated with immune infiltration in LUSC. Further analysis revealed that pyroptosis-related prognostic signature was correlated with immune cell infiltration, tumor mutation burden, microsatellite instability, and drug sensitivity. We also constructed a ceRNA network and identified a lncRNA KCNQ1OT1/miR-328-3p/IL1B regulatory axis for LUSC. Conclusion A bioinformatics method was performed to develop a pyroptosis-related prognostic signature containing four genes (IL6, IL1B, ELANE, and CASP4) in LUSC. We also constructed a ceRNA network and identified a lncRNA KCNQ1OT1/miR-328-3p/IL1B regulatory axis for LUSC. Further in vivo and in vitro studies should be conducted to verify these results.


Introduction
Lung cancer is one of leading causes of human health threatening with approximately 2.09 million initially diagnosed cases and 1.76 million deaths worldwide annually [1]. Lung squamous cell carcinoma (LUSC) is one of most frequent pathological subtypes of lung cancer [2]. LUSC was characterized by large blood vessels and proximal bronchus invasion, leading to hemoptysis. Without the opportunity of radical surgery, over 50% the initially diagnosed LUSC patient would lose their lives within 12 months [3]. Moreover, molecular mechanism for LUSC had not been elucidated and the therapies were limited, resulting in a poor prognosis of LUSC and the average 5-year overall survival (OS) rate of 10-20% [4]. These data demonstrated the urgent need to clarify potential mechanism of LUSC, thus identifying the prognostic biomarkers and therapy targets for LUSC.
Pyroptosis is a programmed cell death mediated by Gasdermin family proteins [5]. Different from autophagy and apoptosis, pyroptosis was characterized by unique morphology and mechanism [6]. Releasing inflammatory cytokines IL-1β and IL-18, pyroptosis played a crucial function in the pathogenesis of many diseases, including atherosclerosis, sepsis, and Parkinson's disease [7][8][9]. Recent studies found that pyroptosis suppressed the tumor oncogenesis and progression; nevertheless, pyroptosis could promote tumor growth by forming a suitable microenvironment [10]. Increasing evidence suggested pyroptosis as a new frontier for tumor due to its effect on the proliferation, invasion, and metastasis of tumor [11]. Moreover, the pyroptosis-related signature could serve as a prognosis biomarker and predicts immune microenvironment infiltration in certain type of cancer, including ovarian cancer, gastric cancer, and lung adenocarcinoma [12][13][14].

Defining the Expression and Genetic Alteration
Landscape of Pyroptosis-Related Genes (PRGs). Based on a previous literature, we obtained 33 PRGs (Supplementary  Table 1) [13,15]. The "limma" package was applied for identification of differently expressed pyroptosis-related genes (PRGs) and a p value was set as 0.05. The mutation landscape of PRGs was visualized using waterfall function within the "maftools" package. By using "RCircos" package in R, we presented the location of CNV alteration of PRG human chromosomes.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes
and Genomes (KEGG) Pathway Analysis. Using differently expressed pyroptosis-related genes (PRGs) in LUCS, we then performed GO and KEGG pathway analysis with "ggplot2" package in R with a p < 0:05. Noteworthily, GO analysis comprises biological process (BP), cellular component (CC), and molecular function (MF).

Construction of a Pyroptosis-Related Prognostic
Signature. A Kaplan-Meier method was used to identify the prognostic PRGs with p values and hazard ratio (HR) with 95% confidence interval (CI) calculated by log-rank test. The LASSO Cox regression model was applied to isolate the candidate genes and construct a pyroptosis-related prognostic signature based on these prognostic PRGs. The risk score of each LUSC patients was calculated using the formula as follows: risk score =∑ 4 i Xi × Yi (X: coefficients, Y: candidate gene expression). With the median risk score as the cutoff, all LUSC cases were divided into low-and highrisk subgroups. The difference of OS rate in two subgroups of LUSC was calculated with Kaplan-Meier survival curves, and time-dependent receiver operating characteristic curves (ROC curves) were applied to evaluate the efficiency of the prognostic signature with "timeROC" packages. Considering the clinical characters and prognostic signature, we performed univariate and multivariate cox regression and the result was drawn in forest with "forestplot" R package. Based on the result, we constructed a predicted nomogram to predict the 1-, 3-, and 5-year OS rate of LUSC patients.
2.5. Immune Infiltration, Drug Sensitivity, TMB, and MSI Analysis. The abundance of immune cells was isolated from TIMER database (https://cistrome.shinyapps.io/timer/). We collected the IC50 of 265 small molecules in 860 cell lines, and its corresponding mRNA gene expression from the Cancer Therapeutics Response Portal (CTRP). The mRNA expression data and drug sensitivity data were merged. The microsatellite instability (MSI) score of LUSC patients was calculated as described previously [16]. After obtaining the masked somatic mutation file (varscan. Somatic. Maf) of LUSC patients from TCGA, we calculated the TMB score of LUSC patients with the "maftools" package in R. Pearson's correlation analysis was used to performed to calculate the correlation between pyroptosis-related prognostic signature and immune infiltration, drug IC50, TMB, and MSI score. A p value < 0.05 was considered statistically significant difference.
2.6. Construction of Potential Regulatory Axis. We then constructed a PPI network with String (https://string-db.org/) to identify hub gene among pyroptosis-related prognostic signature. This was followed by identification of the miRNA target of hub gene using miRDB (http://mirdb.org/), Star-Base (http://starbase.sysu.edu.cn/), and miRWalk (http:// mirwalk.umm.uni-heidelberg.de/). After identifying miRNA targets of hub genes, we used StarBase (http://starbase.sysu .edu.cn/) and LncBase module of DIANA tool (http://    BioMed Research International presented with simple nucleotide variation and NLRP3 was the gene with the highest frequency of mutation followed by NLRP7 and NOD2. We found that missense mutation ranked the top variant classification and C>A was the most common SNV class (Figure 1(c)). In CNV analysis, the data suggested that more than half of 33 PRGs had copy number amplification while the other had a widespread CNV deletion (Figure 1(d)). The location of CNV alteration of PRGs on human chromosomes is shown in Figure 1(e).

GO and KEGG Pathway
Analysis. GO and KEGG pathways were performed with above 26 differently expressed PRGs. The result of GO analysis in Figure 2(a) indicated that these PRGs were enriched in positive regulation of cytokine production, pyroptosis, endocytic vesicle, cysteine-type endopeptidase activity, cytokine receptor binding, and CRAD binding. As for KEGG pathways analysis, the data in Figure 2(b) suggested that these PRGs were enriched in NOD-like receptor signaling pathway, cytosolic DNAsensing pathway, NF-KB signaling pathway, and apoptosis.    3.6. The Correlation between PRG Expression and TMB, MSI, and Drug Sensitivity. Increasing evidences suggested TMB as a predictive marker for immunotherapy efficacy in lung cancer [17,18]. MSI was also referred as a predictive marker for cancer immunotherapy [19]. To clarify the important role of PRGs in LUSC, we then explored its correlation with TMB and MSI in LUSC. As shown in Figure 7(a), the TMB score decreased as the expression of IL1B (p = 0:038) and ELANE (p = 0:003) increased. Similarly, the MSI score decreased as the expression of IL6 (p = 0:048), IL1B (p = 0:004), and CASP4 (p = 6:15e − 10) increased (Figure 7(b)). To develop a therapy target, one of vital way is to analyze its correlation with exited drugs. Interestingly, drug sensitivity analysis indicated high expression of IL6, CASP4, and IL1B was correlated with drug resistance of CTRP (Figure 7(c)), suggesting that IL6, CASP4, and IL1B may serve as the potential biomarkers for drug scanning. We also explored PRGs expression in different TNM stages of LUSC patients. However, only ELANE showed positive correlation with TNM stage (Figure 8(c), p = 0:049). There is no significant difference between the expression of IL6 (Figure 8(a)), IL1B (Figure 8(b)), and ELANE (Figure 8(d)) and different TNM stage LUSC patients (all p > 0:05).

3.7.
Construction of a lncRNA-miRNA-mRNA Regulatory Axis. A PPI network was constructed and revealed IL1B as the hub gene among pyroptosis-related prognostic signature (Figure 9(a)), and we selected IL1B for further analysis. The miRNA target of IL1B was obtained from miRDB, miRWalk, and StarBase. As a result, miR-328-3p was suggested as the miRNA target for IL1B (Figure 9(b)). Moreover, we also found that miR-328-3p was upregulated in LUSC (Figure 9(c), p = 0:017). We further explored the lncRNA target of miR-328-3p, and the data from lncBase and StarBase suggested lncRNA KCNQ1OT1 as a lncRNA target interacting to miR-328-3p   Figure 9(d)). Further analysis demonstrated upregulation of KCNQ1OT1 in LUSC versus normal tissues (p = 3:3e − 9, Figure 9(e)). Thus, lncRNA KCNQ1OT1/miR-328-3p/IL1B regulatory axis may play a vital role in the progression in LUSC. Further in vivo and in vitro studies should be conducted to verify this hypothesis.

Discussion
Pyroptosis was Gasdermin-mediated programmed necrosis associated with pathogenesis of many diseases [20]. Pyroptosis can be chemically induced in tumor cells in the absence of any bacterial or viral infection [21]. Pyroptosis was involved in all stages of carcinogenesis, suggesting it as one of most promising directions for cancer research [22]. Moreover, the pyroptosis-related signature could serve as a prognosis biomarker and predicts immune microenvironment infiltration in a certain type of cancer, including ovarian cancer and gastric cancer [12,13]. However, the role of pyroptosis in LUSC was not fully established.
Based on the above four prognostic PRGs, LASSO Cox regression analysis was conducted to develop a pyroptosisrelated prognostic signature, which could serve as a prognosis biomarker in LUSC and predict the OS rate with medium to high accuracy. We also constructed a predictive nomogram, which could predict relatively well in the 3-year and 5-year OS rates compared with an ideal model in the entire cohort. Actually, a previous study revealed that pyroptosis-related signature could serve as a prognosis biomarker in a certain type of cancer. Ying et al. constructed a 7-pyroptosis-related signature, which could predict the prognosis of ovarian cancer [13]. Another pyroptosis-related signature could serve as a biomarker in gastric cancer, for predicting prognosis and immune microenvironment infiltration [12]. In our study, we firstly identified a pyroptosis-related prognostic gene  Our results revealed that risk score and PRGs (IL6, IL1B, ELANE, and CASP4) were significant correlated with immune infiltration. Previous study had revealed the vital role of these PRGs in tumor microenvironment and immune infiltration. Combined blocking of IL-6 and PD-1/PD-L1 signals can eliminate their immunosuppressive effects in the tumor microenvironment [26]. In colorectal cancer, macrophage-derived IL6 in immune microenvironment was associated with chemoresistance [27]. In lung cancer, IL6 was involved in cell autonomous propensity for metastasis and establishing the metastatic niche [28]. Interestingly, our study also found that high IL6 expression was correlated with drug resistance in LUSC. Thus, the immune-microenvironment may confer chemoresistance of LUSC through immune cell derived IL6. A further study should be conducted to clarify the molecular mechanism of this regulatory axis.
Another vital finding of the current study was that we identified a lncRNA KCNQ1OT1/miR-328-3p/IL1B regulatory axis in LUSC by developing a ceRNA network. Zhang et al. found that KCNQ1OT1 was upregulation in non-SCLC and associated with clinicopathology [29]. Another study revealed that KCNQ1OT1 was a prognostic biomarker in non-SCLC and could accelerate tumor progression by the regulation of miR-204-5p/ATG3 axis [30,31]. Moreover, miR-328-3p could accelerate the occurrence and progression of lung cancer via NF2-mediated Hippo axis [32]. High expression of miR-328-3p was correlated with radiotherapy sensitivity in non-SCLC [33]. Our study revealed another lncRNA KCNQ1OT1/miR-328-3p/IL1B regulatory axis in LUSC, which may also play a vital function in the tumor progression.
Pyroptosis was a double-edged sword and played a vital function in both tumorigenesis and antitumor immunities at all stages of tumor development [34]. Different from autophagy and apoptosis and necroptosis, pyroptosis was characterized by unique morphology and mechanism [6,35]. Increasing evidences revealed that pyroptosis was involved in host defense and highly correlated with bridging innate and adaptive immunity [36]. Moreover, targeting pyroptosis and developing related drugs may provide another immunotherapy strategy for cancer [36]. Our study clarified a significant correlation between pyroptosis-related prognostic signature and immunological infiltration in LUSC, providing some data for the pyroptosis-related immunotherapy of LUSC.
In conclusion, a bioinformatics method was performed to develop a pyroptosis-related prognostic signature containing four genes (IL6, IL1B, ELANE, and CASP4) in LUSC. We also constructed a ceRNA network and identified a lncRNA KCNQ1OT1/miR-328-3p/IL1B regulatory axis for LUSC. Further in vivo and in vitro studies should be conducted to verify these results.

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