A novel pyroptosis-related gene signature exhibits distinct immune cells infiltration landscape in Wilms’ tumor

Wilms’ tumor (WT) is the most common renal tumor and the second most common malignant abdominal tumor in childhood. The incidence of WT in general population is 0.5–7.5 per million and is lower in high-income areas [1]. Current treatment strategy for WT is based on genetic markers, histology, stage, and other risk factors, which spare children with low-risk tumors from intensive treatment and intensify treatment for children with high-risk tumors [2]. Outcomes and long-term survival have improved over the decades [1, 3].

WTs are divided into two histologies: the favorable histology (FH) and the unfavorable histology (UH) which includes anaplastic histology (AH), clear cell sarcoma of the kidney (CCSK), and malignant rhabdoid tumor (MRT) [4]. Despite the advances in multi-disciplinary treatment and risk-based management of WT, the current prognoses of patients with unfavorable histology remain dismal [5].

Pyroptosis is a type of gasdermin-mediated, inflammation-characterized, and immune-related programmed cell death, which has received increasing attention due to its association with immune response in neoplastic and non-neoplastic diseases [6]. The role of pyroptosis has been extensively elucidated in cardiovascular diseases, nervous system disorders, psychiatric disorders, infection diseases, periodontal diseases, etc. [7‐11]. Further, pyroptosis has been proven to be closely related to the biological behavior of multiple tumors [12].

The tumor microenvironment (TME) consists of miscellaneous cell types, such as cancer cells, cancer stem cells, immune cells, stromal cells, and vascular endothelial cells. Tumor-infiltrating immune cells are found generic in tumor tissues with complex tumor-antagonizing or tumor-promoting functions, which surprisingly can affect the hallmarks of the tumor [13]. Quantification methods of these cells are divided into two categories: methods of enrichment on marker genes and methods leveraging the deconvolution algorithm [14, 15]. Each of these methods has a property to allow intra- or inter-sample comparison. We used several authoritative methods in current research.

The RNA-sequencing technique has been extensively employed to identify transcriptome profiles of various tumors [16], based on which several gene signatures have been established for the prognosis of WT [17, 18]. We have previously constructed a ferroptosis-related lncRNA signature in WT [17]. To the best of our knowledge, no signature based on pyroptosis for WT was reported. Here, we established a pyroptosis-related gene signature from the RNA-seq data of WT and constructed a nomogram with the signature and clinical variables. Immune infiltration characteristics of the signature were then identified. We further used snRNA-seq data for validation of our signature. The overall workflow is presented in Fig. 1.

Wilms’ tumor is an embryonal malignancy derived from nephrogenic blastemal cells in nephrogenic rests [2]. Current treatment strategy for WT from Children’s Oncology Group (COG) is based on traditional risk factors, including histology, stage, age, tumor weight, response to therapy, and loss of heterozygosity at 1p and 16q [44]. These prognostic factors are derived from elaborate clinical trials and have been guiding operative therapy, chemotherapy, and radiotherapy of WT. Here, we have screened out gender and stage as prognostic clinical factors in WT. To acquire a more precise risk-classifying approach, exploring specific patterns of gene expression profiles in tumor tissue is necessary.

Characterized by inflammation, pyroptosis is a more intense pattern of cell death compared with apoptosis, leading to pore formation on cell membrane, chromatin fragmentation, cell swelling, and osmotic lysis of the cell [6]. Exact molecular relationship between pyroptosis and tumors’ bio-behavior remains unclear, but studies have found that enhancement of pyroptosis in tumors leads to inhibition of tumor progression [45‐49]. To the best of our knowledge, few studies have elucidated the role of pyroptosis in WT.

In this study, we have established a pyroptosis-associated gene signature with 14 PRGs, among which CARD8, GSDMD, TP63, TFAP2A, IRF2, and PCSK9 were negatively correlated with risk score, while NAIP, NLRP6, SDHB, MIR30C1, APIP, CASP9, MIR103A1, and GLMN were positively correlated with risk score.

Caspase recruitment domain family member 8 (CARD8) is an inflammasome sensor, which ultimately activates GSDMD and inflammatory cytokines, leading to pyroptosis [50]. Gasdermin D (GSDMD) is the main executioner of pyroptosis, and is considered as a tumor suppressor [51, 52]. It has been deeply studied in pyroptotic cell death. Transcription factor AP-2 alpha (TFAP2A) and interferon regulatory factor 2 (IRF2) were proven to transcriptionally induce GSDMD by binding to its promoter, which subsequently induces pyroptosis [53, 54]. The role of proprotein convertase subtilisin/kexin type 9 (PCSK9) in pyroptosis has been clarified in cardiomyocytes and vascular endothelial cells [55, 56]. These five genes are pyroptosis-executive or pyroptosis-promoting, and in the sight of the tumor-suppressing role of pyroptosis, these are reasonable to be negatively correlated with risk score.

APAF1 interacting protein (APIP) is validated to inhibit pyroptosis and apoptosis [57, 58]. Glomulin, FKBP Associated Protein (GLMN) is a negative regulator of pyroptosis via regulating cIAP-mediated inflammasome activation [59, 60]. These two genes have the pyroptosis-inhibiting effect, which is easy to explain the positive correlations with risk score.

NLR family apoptosis inhibitory protein (NAIP) is necessary for inflammasome assembly which subsequently cleaves caspase-1 and leads to pyroptosis [61]. Like CARD8, NLR family pyrin domain containing 6 (NLRP6) is also an inflammasome sensor that mediates inflammasome activation and promotes recruitment of effector proinflammatory caspases [62]. Overexpression of succinate dehydrogenase complex iron sulfur subunit B (SDHB) has been proven to enhance pyroptosis in vascular endothelial cells [63]. Caspase 9 (CASP9) has the ability to cleave and activate caspase-3, which subsequently activates Gasdermin E (GSDME) [64]. These four genes are pyroptosis-executive or pyroptosis-promoting genes. Providing the negative correlation of pyroptosis with tumor progression, the four genes are supposed to be negatively correlated with risk score, which contradicts our observations.

Here are possible explanations. First, the predominant role of pyroptosis in the bio-behavior of tumors is tumor suppressor, but the adverse effect may exist in certain tumor microenvironments. Second, except for pyroptosis, the functions of genes are most likely to relate to multiple biological processes. Such as NAIP can also act as an anti-apoptotic protein by inhibiting caspase-3, and caspase-7 [65, 66], which may promote tumorigenesis and tumor invasion. Third, few studies have explored the exact roles of certain genes (NLRP6 and SDHB) in WT. Although NLRP6 was reported to serve as a tumor suppressor in colorectal cancer, hepatocellular carcinoma, and gastric cancer [67‐69], it was reported to restore immune evasion and radio-resistance in glioma through ASC/caspase-1/IL-1β axis [70]. The SDHB gene encodes the iron-sulfur protein subunit of the succinate dehydrogenase enzyme complex which plays a critical role in respiratory electron transport and tricarboxylic acid cycle [71]. Higher gene expression of SDHB may provide more energy for tumor cells in WT. The roles of these genes in WT need further investigation. Last but not the least, genes may have complex interactions with the tumor microenvironment. For example, CASP9 inhibition triggers immunogenic cell death, increases tumor-intrinsic innate sensing, and induces remarkable anti-tumor effects in chemotherapy-induced anti-tumor immunity [72]. Higher gene expression of CASP9 may exhibit less immunogenic cell death and worse prognosis in WT. The above-mentioned theoretical deductions need further experiment-based validation.

Tumor protein p63 (TP63) has two isoforms, in which TAp63 is thought of as a tumor suppressor, while ΔNp63 is considered as an oncogene [73]. In the sample set of our study, TAp63 may be functionally predominant over ΔNp63. MicroRNA 30c-1 (MIR30C1) is known to inhibit the progression of prostate cancer and the invasion of melanoma [74, 75]. MicroRNA 103a-1 (MIR103A1) was found to have dual effects on different tumor types [76‐78]. In our results of bioinformatics analysis, MIR103A1 is most likely to negatively regulate the progression of WT.

Solid tumor tissue does not only have cancer cells but also immune inflammatory cells, endothelial cells, etc. Tumor-promoting and tumor-antagonizing immune cells exist simultaneously in tumors [13]. In our observations of immune cell infiltration, the expression of CD8( +) T cells, B cells, and Th2 cells are positively correlated with risk score, while the expression of dendritic cells and type 2 macrophages are negatively correlated with risk score.

Dendritic cells can either suppress tumor progression or drive tolerance in the TME [79]. As for our results, dendritic cells are most likely to play a tumor suppressor role in WT. B cells are considered to have dual effects in the TME, and Th2 cells are the intrinsic helper for B cells [80]. They are both positively related to risk score in our research, based on which we can assume that the tumor-promoting effect overweighs the tumor-suppressing effect.

CD8( +) T cells, also called cytotoxic lymphocytes (CTLs), have long been known to relate to better survival in tumors [81‐83]. But in our results, it is positively related to risk score. Type 2 macrophages are known to be tumor-promoting [84, 85], which contradicts our observations, too. There are several explanations. First, it is the enrichment scores or fractions of immune cells that stand for expression levels, but not the real quantity of them. Second, despite our best efforts, only one dataset in TARGET database was included in our research, and no validating dataset of good quality was harnessed. Thereby, the sample number of tumors is only 126, which may be too small to get completely correct results. Third, it may be reasonable to assume that the more invasive the tumor is, the more activation of immune system to fight against it. Based on this assumption, the role of immune infiltrating cells in WT is more of a defensive reactor than a prognostic indicator.

Hence, the distinct immune cells infiltration landscape between high- and low-risk groups was mainly made up of CD8( +) T cells, B cells, Th2 cells, dendritic cells, and type 2 macrophages, which were validated by various algorithms.

WT classically consists of three elements: blastemal, stromal, and epithelial tubules, which are normal components of developing kidneys. Less commonly, skeletal muscle, cartilage, osteoid, or adipose tissue can also be found in WT [2, 4]. In our study, GO BP analysis showed DEGs between two groups were mainly enriched in tissue/organ development, the top five of which were pattern specification process, muscle tissue development, connective tissue development, cartilage development, and ossification. These observation indicates that the extent of development of embryonal components in WT is correlated with tumor progression and prognostic risk. KEGG analysis showed pathway enrichment in signaling pathways regulating pluripotency of stem cells, proteoglycans in cancer, and Wnt signaling pathway. WT has similar components to fetal kidneys, where pluripotent stem cells (PSCs) play a key role in tissue formation and organ development. Our results suggest the signaling pathway that regulates PSCs is correlated with tumorigenesis of WT. Proteoglycans (PGs) and the Wnt signaling pathway have been extensively studied in multiple cancers, both of which are closely related to tumor progress and invasion [37‐43].

Therefore, the extent of development of embryonal components in WT together with the status of proteoglycans and Wnt signaling pathway might be intermediate contributors to tumor progression.

Single-cell RNA-seq (scRNA-seq) and single-nucleus RNA-seq (snRNA-seq) have received increasing attention over the years. Both techniques have remarkable advantages over bulk RNA-seq. We have employed a recently published snRNA-seq dataset of WT to validate our gene signature. In accordance with a former authoritative study [36], cell clusters are similar to normal constituents in fetal kidneys, which is a distinct characteristic of WT. Combined with the results of GO/KEGG analysis, we might infer that the capacity of tumor cells to generate various renal cells isn’t lost or is required in WT.

Thus, the snRNA-seq dataset has facilitated the validation of signature-establishing genes and has supported the renal lineage-generating ability of tumor cells in WT.

The tumor microenvironment has complex gene interactions and cell cross-talks, which form latent networks. Identifying key modules and hub genes by weighted gene co-expression network analysis might provide a deeper understanding of the roles of signature-constructing genes and tumor-infiltrating immune cells like another study [86], which needs future exploration.

Several limitations are supposed to be acknowledged. First, as mentioned before, only one bulk RNA-seq dataset of good quality was used in our research, which might not draw completely correct conclusions. Second, due to the particularities of constituents of WT, the annotation process for each cluster in snRNA-seq analysis was mainly based on one literature of authority, which might lack rigor. Last, compared with clinical factors, there is still a long way to go to perform risk classification based on tumor transcriptome.

The current study is mainly based on retrospective data analysis and in silico experiments. Further validation is essential to confirm the exact roles of signature-constructing genes and tumor-infiltrating immune cells. For future investigation, we aim to perform gene-overexpression or knockdown in WT cell lines and observe changes in proliferation, migration, invasion, and other malignant behavior in vitro and in vivo. Moreover, a long-term follow-up of patient cohorts grouped by the gene signature will provide validation for the current study.

Here, we have established a pyroptosis-related gene signature of moderate to high predicting capacity with 14 PRGs and have constructed a prognostic-predicting nomogram with risk score and two clinical variables. Moreover, immune cell infiltration analysis, differential expression analysis, and functional enrichment analysis were performed to further explore underlying mechanisms contributing to the differences in survival states. Our research provides insights into the role of pyroptosis and possible therapeutic targets in Wilms’ tumor.