Introduction
Human osteosarcoma (OS) is the most frequent aggressive bone cancer in children and adolescents [
1]. Despite improvements in multimodal therapies, the prognosis of OS remains poor (20–30%) due to the delay in diagnosis and the development of metastasis [
2]. Therefore, it is urgent to reveal novel biomarkers and ensure effective OS treatment.
Although RNA modification has been recognized for more than half a century, its cell biology remains largely unexplored [
3]. RNA adenosine modification is the most common type of RNA modification, including m
1A and m
6A modification.
RNA editing includes
adenosine-to-inosine (A-to-I) editing and alternative
polyadenylation (
APA) [
3]. RNA methylation accounts for approximately 60% of RNA modifications, whereas m
1A and m
6A modifications are common and abundant in
RNA methylation modification by the methylation of the adenine base. These modifications regulate RNA stability and translation [
4]. m
1A and m
6A modifications play key roles in various cellular processes, thus leading to a variety of diseases, including cancer [
5,
6]. Polyadenylation (APA) is a phenomenon in which one gene contains multiple polyadenylation (pA) sites to produce transcript isoforms at either the 3′-untranslated region (UTR) or coding regions [
7]. Another common posttranscriptional mechanism is RNA editing, which alters gene expression by regulating the nucleotides of transcripts [
8]. The most widespread type of RNA editing is A-to-I, which is catalysed by ADAR enzymes and alters the coding, folding, splicing, or transport of transcripts. Dysregulated A-to-I editing can lead to various diseases, including tumorigenesis [
9]. Studies have shown that many of these modifications could be related to a complex network that intersects each other [
10]. Recently, these RNA modifications have been reported as biomarkers and play key roles in the progression of tumour prognosis [
11,
12]. However, their role in OS remains limited.
In this work, we identified RMW patterns and RNA modification-related DEG patterns that were associated with the prognosis of OS and related to immune infiltration. Next, we constructed and verified an RMW signature of OS, which was not only associated with prognosis but also with immune infiltration and the cell cycle. Finally, strophanthidin was predicted and verified as an effective drug for OS.
Materials and methods
Datasets
The expression of OS patients with clinical information was obtained from the GSE21257 and Target datasets. Eighty-five OS patients and 53 OS patients with survival follow-up information were collected from the Target and GSE21257 datasets, respectively.
Consensus clustering analysis
To determine the RNA modification patterns in OS, 26 RMWs were selected for consensus clustering analysis using the Consensus Cluster Plus package, and k = 2 seemed to be the most appropriate selection.
The DEGs identified between the two RMW patterns were analysed using the “limma” package with adjusted p < 0.05. Survival-related DEGs were analysed using univariate Cox regression analysis and subsequently selected for a consensus clustering analysis using the Consensus Cluster Plus package, and k = 2 seemed to be the most appropriate selection.
The survival analysis of patients from the two patterns was analysed using the packages “survminer” and “survival”.
The relationship between patterns and clinical characteristics was analysed using the “RColorBrewer” package.
The RMW prognostic signature
The survival-related RMWs were analysed using univariate Cox regression analysis. The prognostic signature was analysed using lasso regression and multivariate Cox regression analysis. The risk score was calculated as follows:
Risk score = ∑iCoefficient RMWs *Expression RMWs.
The OS patients were divided into high- and low-risk groups based on the best cut-off value of the risk score from the TCGA dataset. The R “survival” package was used to assess the relationship between survival and the RMW signature. Univariate and multivariate Cox regressions were used to reveal the independent risk factors for OS using R software. ROC curves were employed to reveal the prognostic value using the R software “ROC” package. The nomogram and calibration curve were analysed using the R “rms” package.
GSEA and GSVA were used to reveal the risk signature-related pathways using the Target dataset. GSEA software (Cambridge, MA, United States) was used to describe the pathways related to RMWs. For GSVA, the signalling pathway alterations between the high- and low-risk groups were analysed using the “GSVA” R package.
Immune infiltration in OS tissues
The immune infiltration of OS tissues from the Target database was evaluated using xCell (
https://xcell.ucsf.edu/). Next, the immune-related signalling and immune cell infiltration in different RMW patterns, DEG patterns and high/low risk groups were analysed using the R “ggpubr” package.
Clinical samples and qPCR analysis
OS tissues were obtained from 63 OS patients (September 2018 to January 2020) from the Third Xiangya Hospital, Central South University. Our study was approved by the ethics committee of the Third Xiangya Hospital, Central South University.
Total RNA was collected from OS tissues and cells using TRIpure reagent (BioTek, VT, USA) and then reverse transcribed using a HiScript Q RT SuperMix kit (Vazyme, Nanjing, China). Then, qRT–PCR was performed to assess mRNA expression using SYBR Green Master Mix (CWBIO, Jiangsu, China) in an Applied Biosystems QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific, MA, USA) as previously described [
13]. The nucleotide sequences of primers for
CSTF2, ADAR, WTAP and
β-actin are listed in Table
1.
Table 1
The primers for qPCR
β-actin | CTGTCCCTGTATGCCTCTG | TGATGTCACGCACGATTT |
CSTF2 | CAGCGGTGGATCGTTCTCTAC | AACAACAGGTCCAACCTCAGA |
ADAR | ATCAGCGGGCTGTTAGAATATG | AAACTCTCGGCCATTGATGAC |
WTAP | CTTCCCAAGAAGGTTCGATTGA | TCAGACTCTCTTAGGCCAGTTAC |
Cell culture
Human OS cell lines (MG63, HOS and U2OS), L02 and HUVEC cell lines were purchased from American Type Culture Collection (ATCC) (Manassas, VA, USA). All cells were cultured as previously described [
14].
Cell viability assay
The MTT assay was used to detect cell viability. In short, the cells were cultured in 96-well plates overnight and then treated with strophanthidin (0.001 to 100 μM) (Santa Cruz, sc-215914A)/PBS for 24 or 48 h. Then, cell viability was detected using an MTT assay. The IC50 value was calculated using GraphPad Prism.
OS cells were incubated with strophanthidin for 48 h. Then, ~ 1000 cells were seeded in 6-well plates and cultured for 10 days for the colony forming assay.
Flow cytometric analysis
OS cells were incubated with strophanthidin for 48 h. Then, flow cytometry was used for the cell cycle assay as previously described [
14].
Western blot analysis
OS cells were incubated with strophanthidin for 48 h. Then, the protein was collected using RIPA buffer with a protease inhibitor cocktail (Roche Applied Science, Indianapolis, USA). The proteins were separated and transferred to PVDF membranes (Millipore, Massachusetts, USA). Primary antibodies against GAPDH (ab8245, Abcam), CDK4 (ab108357, Abcam), CCND1 (ab16663, Abcam) and p53 (ab32389, Abcam) and secondary antibodies were used to detect target proteins.
Animal experiment
Four- to six-week-old female BALB/c nude mice were maintained in SPF conditions. Nude mice were injected with HOS cells (1 × 106) into the left scapula subcutaneously. For the control group, the mice were administered 200 μl of saline po/day. After 10 days, nude mice were randomly assigned to 2 groups. For the treatment group, mice were treated with strophanthidin (0.5 mg/kg/day). The body weights and the tumour volume were measured every 4 days. After 24 days, all nude mice were sacrificed, and the tumour weights were detected. All animal experiments were approved by the ethics committee of Third Xiangya Hospital, Central South University.
Statistical analysis
All data from Target and GEO were analysed by R-3.6.1. The differential RMWs were evaluated using the limma and heatmap packages. Student’s t test was used for two groups of in vitro experiments, and chi-squared (χ2) tests were used for in vivo experiments. p < 0.05 was considered statistically significant.
Discussion
OS is a commonly aggressive tumour in children and adolescents with poor survival rates. Recently, prognostic models for improving the prognosis of OS have attracted considerable attention. In this study, we constructed and verified the RMW patterns and prognostic signature in OS and shed light on strophanthidin as a novel therapeutic drug for OS patients.
RNA adenosine modification is the most common type of RNA modification and is involved in various pathophysiological processes, including tumorigenesis [
17,
18]. m6A modification is an RNA modification with dynamic and reversible posttranscriptional characteristics that affects RNA processing, degradation and translation. Studies have shown that m6A modification levels are dysregulated in tumour tissues and are associated with the occurrence and progression of various tumours [
19‐
22]. The “writers” play a key role in these RNA modifications, and some of the writers were reported as therapeutic targets of tumours. For example, METTL3, the m
6A writer, is related to the prognosis and progression of acute myeloid leukaemia (AML), and METTL3 inhibition was reported as a therapeutic strategy for myeloid leukaemia [
23]. A-to-I RNA editing is another abundant RNA modification event affecting adenosines in mammals, playing a critical role in the pathogenesis of various tumours [
24]. Xu et al. revealed the role of A-to-I-edited miRNA cancer progression and highlighted the translational potential of edited miRNAs as a new class of cancer therapeutics [
25]. Moreover, other adenine-related RNA modifications (m1A methylation and APA) have recently also emerged as key players in cancer pathophysiology by regulating the expression of cancer-related genes [
26]. In OS, m
6A modification of YAP is involved in the progression of OS [
22]. The relationship between other adenine-related RNA modifications and OS remains unknown. RNA modification plays an important role in various biological processes through interactions with various “writers” (RMWs). Recent studies have focused on RNA modification signatures to evaluate the prognosis of various tumours. Our study explored RMW patterns and RNA modification DEG patterns that were associated with the survival of OS patients. Next, the RMW prognostic signature was constructed and identified as an independent prognostic factor for OS. In conclusion, these results indicated the important role of RMWs in OS. Although previous studies revealed several prognostic models (including ferroptosis-related gene signatures and autophagy-related prognosis models) for OS [
27‐
30], this is the first study of the RWM-related prognosis model and RMP patterns for OS. Moreover, the ROC of the RMW model was considerably increased compared with that of the ferroptosis-related gene signature (ROC = 8.06) and autophagy-related prognosis model (ROC = 8.38). These results indicated that the RMW signature is a more effective prognostic model for OS.
The immune microenvironment (TME) is related to the prognosis of various cancers, including OS [
31‐
33]. The recruitment of stromal cells was reported to be associated with a poor patient prognosis. Infiltrating myeloid cells are common and important components of the TME, and they are reported to drive OS progression [
34]. Tumour-infiltrating myeloid cells include monocytes, dendritic cells (DCs), tumour-associated macrophages (TAMs), and neutrophils. In OS lesions, monocytes and macrophages are the most common myeloid cells, and DCs account for < 5% of myeloid cells [
35]. M2 TAMs are anti-inflammatory macrophages and are often associated with a worse prognosis [
36]. M1 TAMs are antitumor immune cells that express proinflammatory cytokines. Anti-PD-1 is an effective therapeutic strategy for OS due to the regulation of the infiltration of M1 and M2 macrophages in tumour tissue [
37]. Recently, DCs were identified as therapeutic targets of immunotherapy due to their powerful antigen-presenting features. Zhou et al. showed that the infiltration of CD1c
+ DCs was significantly increased in metastatic OS [
35]. In this study, we used xCell to reveal the correlations between the risk signature and immune infiltration. We found that M1 macrophages, M2 macrophages and cDCs were significantly increased in OS patients at high risk. Recent studies shed new light on the regulation of RNA modification on the immune system [
38]. Furthermore, RNA adenosine modifications are associated with immunoregulation in tumour tissues [
38]. METTL3-mediated m6A modification promotes tumour growth and metastasis by macrophage reprogramming [
39]. METTL3-mediated m6A modification also affects resistance to chemotherapy by regulating M2-TAM infiltration [
40]. METTL3-mediated m6A methylation regulates dendritic cell activation [
41]. These results highlighted that RNA adenosine modifications might affect the prognosis of OS partly by regulating the immune microenvironment.
DGIdb and CMap, drug prediction databases, are widely used to screen drugs that can regulate certain target genes. In this study, we utilized these two databases to obtain the possible RMW-related drugs for OS. Ten drugs were identified, including dyclonine, resveratrol, zaprinast, abamectin, thioguanosine, cefotaxime, mesoridazine, clioquinol, physostigmine and strophanthidin. Dyclonine, an ALDH3A1 inhibitor, is widely used as a topical antipruritic agent [
42]. Recent studies reported that dyclonine enhances MG132-induced cell cytotoxicity in breast cancer [
43]. Resveratrol, an oestrogenic compound, was reported as a treatment for osteoporosis with anti-inflammatory and antioxidant properties [
44]. Recent studies also revealed the therapeutic potential of resveratrol on various tumours, including OS [
45‐
47]. Zaprinast, a synthetic GPR35 agonist, was reported to rescue OVX-induced bone [
48]. It also exhibits antitumor activity in colon cancer, lung cancer, and melanoma [
49‐
51]. Abamectin, an anthelmintic agent in animals, induces oxidative stress in cerebral and hepatic tissues [
52]. Thioguanosine has antitumor activities against hepatocellular carcinoma [
53]. Mesoridazine, an antipsychotic drug, has cardiac conduction side effects with fatal consequences in patients [
54]. Clioquinol has been reported as an antitumor drug for OS [
54]. Physostigmine has been reported to repress pancreatic cancer growth with anticholinergic toxicity [
55]. Strophanthidin, a cardiac glycoside, was reported to be a promising anticancer agent by regulating the cell cycles of breast, lung and liver cancer cells [
15]. It is also an immune cell activator that activates ROR T receptors [
16]. Therefore, we selected strophanthidin for further functional analysis. In this study, in vivo and in vitro experiments showed that strophanthidin elicited antitumor activity against OS by repressing cell proliferation and the cell cycle. As shown in Fig.
8, Smad3 was predicted to be the target of strophanthidin. Smad3 is a key mediator of TGF-β signalling, an important signalling pathway in tumorigenesis [
56]. Smad3 was also reported to be involved in the regulation of the cell cycle via the induction of CDK inhibitors [
57,
58]. Based on the above research, we hypothesize that strophanthidin regulates the cell cycle of OS cells by targeting Smad3.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.