Introduction
Osteosarcoma is an osteogenic malignant tumor originating in the bone tissue and is most frequent in adolescents[
1,
2]. It has a high degree of malignancy, low sensitivity to radiotherapy and chemotherapy, easy recurrence and metastasis, and a poor prognosis[
3]. The current main treatment includes a combination of neoadjuvant chemotherapy and extensive surgical resection, but it still has a low overall survival rate[
4]. Therefore, investigating novel early diagnosis and prognostic indicators is of great significance for patients with osteosarcoma.
Pyroptosis is a type of cell programmed inflammatory death different from apoptosis[
5]. It relies on the activation of some caspases and is accompanied by the lysis of GSDMD and the release of pro-inflammatory cytokines[
5,
6]. Finally, stimulating the innate immune mechanism, expands the inflammatory response, causing the cells to collapse and die[
7]. Pyroptosis is activated by the Caspase-1-mediated classical pyroptosis pathway activated by the inflammasomes and the non-Caspase-1-mediated pyroptosis pathway[
8‐
10]. Pyroptosis can form an inflammatory microenvironment through the pro-inflammatory effects, or affect certain signaling pathways to promote the growth of the malignant tumors[
11,
12]. However, many studies have confirmed that pyroptosis playing a key role in malignant tumor treatments by regulating the activity of a certain target or signal pathway[
13,
14]. For example, Hou et al. have found that PD-L1 can regulate the expression of gasdermin C, transforming apoptosis into pyroptosis, and promote tumor necrosis[
15]. However, the specific function of pyroptosis in the prognosis and treatment of osteosarcoma is still at its infancy.
In this study, a systematic study of the PRGs was conducted in osteosarcoma and 6 PRGs signature were identified to have powerful prognostic functions and verified in the GSE21257 cohort. The relationship between the PRGs risk model and the immune microenvironment has also been discussed. In addition, a 9 PRLs signature was also found to be related to the prognosis of osteosarcoma. Through functional enrichment analysis, the possible mechanism of action was discussed. Compared with osteoblasts, the expression level of CHMP4C in osteosarcoma cells was up-regulated, which might be a promising biomarker. Finally, overexpression of CHMP4C promoted the proliferation, migration and invasion of the osteosarcoma cell line U2OS. Our findings provide new evidence for exploring the prognostic biomarkers and therapeutic targets of osteosarcoma.
Materials and methods
Data collection
The RNA-seq data and clinical information of 85 osteosarcoma patients were screened from the TCGA database (TARGET-OS project). The gene expression data of the musculoskeletal samples from 396 healthy humans were collected from the GTEx (The Genotype-Tissue Expression) database. To eliminate the platform data difference between the TCGA and GTEx databases, the gene transcriptional expression data of each sample were transformed into log2 (FPKM value + 1).
The GSE21257 and GSE42352 dataset of osteosarcoma was obtained from the high-throughput microarray expression profile database (Gene Expression Omnibus database, GEO,
https://www.ncbi.nlm.nih.gov/geo/)). GSE21257 contained the gene expression data and related clinical information of 53 osteosarcoma patients, which was used as the verification cohort. GSE42352 contained 15 normal samples and 103 osteosarcoma samples for analyzing the differential expression.
Differential analysis
Using the “limma” package, FDR < 0.05 and logFC > 1 as screening criteria, the differences of PRGs expression between the osteosarcoma and normal samples in the combination of TARGET and GTEx cohorts were determined, and the differences in the PRGs expression were visualized. The expression levels of CHMP4C were visualized in several common cancers by the GEPIA online tool (
https://cistrome.shinyapps.io/timer/) and TIMER online tool (Gene Expression Profiling Interactive Analysis,
http://gepia.cancer-pku.cn /).
Construction and validation of the PRG-based prognostic signature
The univariate Cox regression analysis of 57 PRGs was carried out by using the “survival” R package in the TARGET cohort, where p < 0. 05 is considered to be related to prognosis. The genes obtained from the univariate Cox regression analysis were analyzed by “glmnet” R package for 1000 times iterative Lasso regression analysis, and the final key prognostic genes were determined.
To obtain the PRLs, the 57 PRGs were compared with the lncRNAs one by one to calculate the Pearson correlation coefficient in the TARGET database. The PRLs were screened according to the absolute value of correlation coefficient ≥ 0.4 and
p < 0.05. Then, to build a PRLs prognostic model, the differentially expressed PRLs were selected, the prognostic PRLs were screened by univariate Cox regression, and the PRLs prognostic signature was constructed including ten PRLs by LASSO Cox analysis, at the same time, the risk coefficient of each gene was obtained. A risk scoring equation based on the expression of the genes was constructed:
$$Risk \,score={\sum }_{i = 1}^{n}\left({Coef}_{i}*{x}_{i}\right)$$
Here, \({Coef}_{i}\)refers to the regression coefficient of the gene, and \({x}_{i}\) is the expression level of the gene.
Evaluation and verification of the risk model
The risk score of the osteosarcoma samples was calculated, ranked from low to high, and the osteosarcoma samples were divided into the low-risk and high-risk groups according to the median. The Kaplan—Meier curve was used to analyze the difference in the prognosis between the groups. The time-dependent ROC curve was drawn by the “survival” R package. The univariate Cox and multivariate Cox regression analyses were used to explore the independent prognostic factors, including age, gender, and metastasis. The “rms” package was used to establish a nomogram, and draw calibration curves to assess the consistency of the predicted results with the actual results.
Immune cell infiltration and immune score analysis
The ssGSEA was used to evaluate the immune cell infiltration in each sample. Based on the ESTIMATE algorithm, the ESTIMATE score, immune score, and stromal score of the osteosarcoma patients were calculated by the “estimate” R package.
Functional enrichment analysis
The DEGs between the low-risk and high-risk groups were determined using the “limma” package, and the “clusterProfiler” R package was used for Gene Ontology (GO) and KEGG analysis. The hallmark gene sets (h.all.v7.4.symbols) of the high- and low-risk groups were further analyzed by the GSEA software, and a gene enrichment map was drawn. The GSEA software was downloaded from (
http://www.gsea-msigdb.org/).
Cell lines and reagents
The hFOB1.19 and the 143B, SaOS2, and U2OS osteosarcoma cell lines were purchased from the National Collection of Authenticated Cell Cultures (Shanghai, China). The TRIzol reagent and penicillin/streptomycin were purchased from Thermo Fisher Scientific, USA. The RT-qPCR kit was purchased from Takara Company, Japan. The Dulbecco’s modified Eagle’s medium (DMEM) and fetal bovine serum (FBS) were purchased from Gibco, USA. The primers (CHMP4C, GAPDH) were purchased from Sangon Biotech Shanghai, China. The Primers are listed in Additional file
1: Table S4.
Cell culture
The osteosarcoma cell lines were grown in complete DMEM (containing 10% FBS and 1% penicillin/streptomycin) at 37℃ in a humidified atmosphere containing 5% CO2. The osteoblast cell lines were grown in the same complete medium at 34℃ in a humidified atmosphere containing 5% CO2.
Clinical specimens
We collected 3 osteosarcoma tissues and 3 matched adjacent normal tissues. The samples came from patients who underwent surgery at The Second Affiliated Hospital of Nanchang University and were pathologically diagnosed with osteosarcoma. All patients signed an informed consent form, and the study was approved by the Research Ethics Committee of the Second Affiliated Hospital of Nanchang University.
RNA extraction and RT-qPCR
Add TRIzol to the cells to extract total RNA, and obtain cDNA after reverse transcription. The qPCR kit was used to detect the expression of CHMP4C using the relative quantification method according to the instructions and GAPDH as an internal control.
Lentivirus infection
Lentiviruses containing pFBLV-CHMP4C-Puro and controls were purchased from Focus Bioscience Company (Nanchang, China), and U2OS cells were infected according to the manufacturer’s protocol. Puromycin (1.0 µg/mL) was used to select stably transfected cells. Overexpression of CHMP4C was confirmed via western blotting.
Western blotting
Proteins were extracted from cells using RIPA lysis buffer and quantified using the BCA method. The proteins were separated by 10% SDS-PAGE gel electrophoresis and transferred to PVDF membrane. Block with 5% skim milk and incubate with anti-CHMP4C (Abcam) overnight at 4℃. The next day, the membrane was rinsed twice with PBST and incubated with horseradish peroxidase secondary antibody (1:20000) for 1 h. Finally, expression of the corresponding protein was observed via chemiluminescence and analyzed using ImageJ software.
The proliferation of osteosarcoma cells was detected by CCK-8 and colony formation assays. For the CCK-8 assay, CHMP4C overexpressing cells and control U2OS cells were seeded in 96-well plates at a density of 1 × 103 cells/well in 5 replicates. CCK-8 reagent was added to the wells at the indicated time points. Plates were incubated at 37 °C for 1.5 h before recording optical density (OD) at 450 nm.
In colony formation experiments, U2OS cells were seeded into 6-well plates at a density of 1 × 103 cells/well. Cells were cultured for 2 weeks, and the medium was changed every 3 days. After 2 weeks, Colonies were fixed and stained with 1% crystal violet. The plates were photographed and the number of cell colonies in each well was counted.
Wound healing and transwell invasion assays
The migration and invasion abilities of osteosarcoma cells were evaluated by wound healing and transwell migration and invasion assays. In wound healing assays, transfected osteosarcoma cells were seeded into six-well plates. When the cell density reached about 90%, the cells were scratched with a 10 µL sterile pipette tip to allow intercellular space to form, and cultured in serum-free medium for 48 h. An inverted microscope was used to observe the gaps at 0, 24 h, and 48 h and take pictures.
Transwell invasion assays were performed using Falcon® Cell Culture Inserts (NY, USA). Transfected osteosarcoma cells were digested and resuspended in serum-free medium at a density of 105 cells/mL. 400 µL of cell suspension was added to the upper chamber and 700 µL of medium (10% fetal bovine serum) was added to the lower chamber. After 24 h, the cells in the bottom cavity were fixed, stained with 1% crystal violet, and photographed with an inverted microscope.
Immunohistochemical staining
To further verify the expression of CHMP4C, immunohistochemistry was performed on paraffin sections following the standard protocol (Abcam, ab272638). All slides were observed and photographed under XSP-C204 microscope (CIC).
Statistical analysis
Analyzed the data with the R Software (v4.0.4) and GraphPad Prism (v9.0). The student’s t-test was used to compare the differences between the two groups. p < 0.05 indicated that the difference was statistically significant.
Discussion
This study analyzed the differential expression of the PRGs between the osteosarcoma and healthy tissues. Then, 10 prognostic PRGs were preliminarily screened out through the univariate Cox regression analysis. By the Lasso Cox regression analysis, 6 key PRGs were screened for constructing the optimal model, namely CHMP4C, GZMA, BAK1, CASP1, CASP6, and GSDMA, and a six PRGs signature was successfully constructed for osteosarcoma. Compared with low-risk patients, the survival rate of high-risk patients is significantly lower. The results of the validation cohort also showed that the model has good prognostic significance. In addition, the risk scores and other clinicopathological factors (including age, gender, and metastasis) were used to construct an excellent nomogram for predicting the survival rates. In summary, these results confirmed that in our study, the six PRGs signature have a strong prognostic value in the patients with osteosarcoma and can be extended to other cohorts.
Pyroptosis is a new mechanism of programmed cell death, also known as gasdermin-mediated programmed necrotic cell death [
5,
22,
23]. Recent studies have shown cell pyroptosis to be closely related to the occurrence and development of cancer [
11,
24]. However, the role of pyroptosis in osteosarcoma remains unclear. Although some studies have recently reported pyroptosis-related signatures [
25,
26], there are still some shortcomings in experimental verification, which affects the widespread application of signatures. This study identified 6 key PRGs related to the prognosis of osteosarcoma, and their role in tumors has been studied. The GZMA (Granzyme A) belongs to serine proteases, which are abundant in the cytotoxic T and NK cells [
27,
28]. When GZMA is delivered to the target cells through the immunological synapse, it can activate pyroptosis [
29,
30]. This immune effect mechanism promotes the cytotoxic T cell-mediated tumor clearance in the mice [
29]. BAK1 (BCL2 Antagonist/Killer 1) belongs to the BCL2 family, which is located in the mitochondria and induces apoptosis [
31,
32]. Recent studies have reported BAK1 to be involved in the caspase-3-GSDME mediated pyroptosis pathway, the knockdown of BAK1 can reduce cell pyroptosis [
33]. CASP1 (caspase-1) and CASP6 (caspase-6) are both members of the cysteine-aspartic acid protease (caspase) family. The activation of caspase plays a central role in programmed cell death. The low expression of CASP1 is related to the poor prognosis of lung adenocarcinoma, and CASP1 inhibits the invasion and migration of the non-small cell lung cancer (NSCLC) cells [
34]. Emerging pieces of evidence have indicated that CASP6 mediates the activation of innate immunity and inflammasomes, and can promote the activation of programmed cell death, including pyroptosis, apoptosis, and necroptosis [
35]. GSDMA can act as a regulator of programmed cell death [
36,
37]. Studies have reported that GSDMA may be a tumor suppressor gene [
38‐
40], which is generally suppressed in esophageal squamous cell carcinoma and gastric cancer. CHMP4C (chromatin-modifying protein 4 C) plays a role in cell division, which prevents the accumulation of DNA damage by delaying abscission [
41‐
43]. The polymorphism of CHMP4C increases the susceptibility to cancer and might promote genome instability, thereby inducing cancer [
20]. Li et al. found that CHMP4C can increase the NSCLC cells’ survival ability after ionizing radiation, and its silencing can increase the sensitivity of the cells to radiation [
44]. Compared to the normal tissues, CHMP4C is up-regulated in cervical cancer and lung squamous cell carcinoma, the knockdown of CHMP4C inhibits the proliferation of the cancer cells [
21,
45]. Similar to the results of our study, the high expression of CHMP4C might be related to the poor prognosis of osteosarcoma. Through the analysis of multiple public databases, CHMP4C was found to be up-regulated in a variety of tumors, including osteosarcoma. Consistent with this, RT-qPCR was performed to validate the high expressed CHMP4C in the osteosarcoma cell lines. We found that overexpression of CHMP4C enhanced the migratory and invasive abilities of osteosarcoma cells. These results indicate that PRGs play an important role in tumors, promoting or inhibiting metastasis and progression. Moreover, CHMP4C might act as a cancer-promoting factor, which is expected to become an effective target for cancers.
We also established a PRLs prognostic signature for osteosarcoma patients. Firstly, to determine the PRLs, we performed Pearson correlation analysis between the PRGs and lncRNA. By differential expression analysis, we get the differentially expressed PRLs. Next, the differentially expressed PRLs related to the prognosis were selected, and a 9 PRLs signature was developed using the LASSO Cox analysis. As shown by the risk model, the prognosis of the high-risk patients was found to be significantly lower than that of the low-risk patients. The GSEA results suggest that the immune-related functions are enriched in the low-risk patients, suggesting that immune regulation might be related to the improvement of prognosis.
LncRNAs usually do not encode proteins, but they are important in gene regulation and cell metabolism [
46]. Recent studies have shown that lncRNAs are involved in the pathological progression of cardiovascular diseases, tumors, neurological diseases, and other diseases by directly or indirectly acting on the pyroptosis-related pathways [
47‐
50]. Nevertheless, the research on lncRNA related to pyroptosis in cancer, especially osteosarcoma, is very inadequate. We have identified 9 PRLs for constructing the risk model, some of which have been reported to be related to tumors. FOXD2-AS1 is up-regulated in a variety of cancers and has been identified as an oncogene [
51‐
53]. The knockdown of FOXD2-AS1 in osteosarcoma has been found to inhibit tumor growth and invasion in vitro and vivo [
54,
55], and inhibit its resistance to cisplatin [
56]. AL035446.1 might serve as a pro-cancer factor for clear cell renal cell carcinoma patients in the lncRNA risk signature constructed by Yang et al[
57]. The UNC5B-AS1 functions similarly to FOXD2-AS1, and its expression is up-regulated in hepatocellular carcinoma, papillary thyroid cancer, and prostate cancer [
58‐
60]. The silencing of UNC5B-AS1 inhibits tumor growth [
61,
62], but it has not been reported in osteosarcoma. SENCR has been extensively studied in the vascular smooth muscle cells and endothelial cells [
63,
64], but recent studies have showed that it also has a role in cancer. Cheng et al. reported that SENCR promotes the cell proliferation and progression of the NSCLC cells through sponge miR-1-3p [
65]. According to the prognostic model constructed by Guo et al., AC090559.1 is considered to be related to ferroptosis and is a favorable prognostic factor in lung adenocarcinoma [
66]. The functions of AC010894.2, AC018904.1, BX322562.1, AC016596.1 have not been reported in the literature. Our study proved their relationship with the prognosis of patients with osteosarcoma and inferred their role in osteosarcoma through enrichment analysis. The role of these lncRNAs in osteosarcoma needs to be further explored in the experimental studies.
This study still has certain limitations. Firstly, there are currently few public gene expression databases containing prognostic information for the patients with osteosarcoma, resulting in a small number of tumor samples in our study. In the future, a more accurate prognostic model should be built using a larger sample size. Secondly, the clinical information of the data set is not complete, and more abundant clinical data are needed to evaluate the relationship between the model and the clinic. Finally, the exact mechanism underlying how CHMP4C promotes proliferation, invasion and migration also requires further exploration. Hence, further functional experimental research is warranted in the future.
In summary, this study constructed a pyroptosis-related markers signature in osteosarcoma, which is of great significance in determining the prognosis of osteosarcoma patients. The results of this study have emphasized the importance of pyroptosis-related markers to osteosarcoma and provided important evidence for revealing the pathogenesis of osteosarcoma and guiding the future treatment of osteosarcoma.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.