Background

Osteosarcoma (OS), with primitive mesenchymal cells as its origin, is one of the most found primary solid malignancy of the bone, showing an annual prevalence of around 4.8 per million [1,2,3]. OS, frequently arising in the metaphysis of long bone, features high disability rate. By reason that OS inclines to metastasizing to other sites, especially the lung, OS cases have been plagued by unsatisfactory prognosis [4]. In spite of a 60–70% five-year survival rate of localized OS cases following surgical excision, neoadjuvant chemotherapy and postoperative chemotherapy, the rate was under 30% in OS cases experiencing metastasis [5, 6]. As unveiled by a report, advancement in therapies for OS is limited by the complicated and unstable genome [7], which manifest a necessity for advances in early diagnosis, therapy, and prognosis prediction of OS regarding molecular genetics.

Manic Fringe (MFNG) encodes β-1,3-N-acetylglucosaminyltransferase manic fringe that modifies epidermal growth factor repeats in the extracellular domain of Notch [8]. Reports show that MFNG functions as an oncogene and exhibits a high expression in claudin-low breast cancer (CLBC) [9]. A lowered MFNG level limits neovascularization and the migration of renal carcinoma cells [10]. May wa's team reported that an elevation of MFNG was detectable in Ewing's sarcoma cells, and that NIH 3T3 cells with overexpressed MFNG display tumorigenicity in mice suffering severe combined immunodeficiency disease [11]. Ewing's sarcoma is one of the most frequently found malignancies of the bone besides OS, usually arise in childhood and adolescence [12, 13]. Nonetheless, the function of MFNG in OS prognosis is far from understood.

We found that MFNG was an independent prognosis-related gene (PRG) in OS by analyzing the GSE39058. Thus, with the aid of gene expression data in GSE39058, we delved into the interrelation between MFNG expression and OS progression, recurrence, and survival. It was unveiled that high MFNG expression was indicative of adverse outcomes of OS.

Methods

Data acquisition

We retrieved gene expression microarray of OS by simultaneously entering "osteosarcoma" and "survival" in the search box in the gene expression omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/), and singled out GSE39058 as a training dataset herein. Later, we directly filtered clinical prognostic data from the matrix file of microarray datasets in the GEO database, and the messenger RNA (mRNA) sequencing data were utilized for mRNA profiling.

Screening of PRGs

Kaplan–Meier (K-M) curves and Cox regression analysis (CRA) were adopted for filtering PRGs by use of "survival" package in R with the thresholds of K-M < 0.01 and p < 0.01 (Additional file 1: Table S1).

Target gene identification

After screening of PRGs, multivariate CRA was employed to perform independent prognostic analysis (p < 0.01) to obtain genes significantly associated with independent prognosis. Then, we analyzed the correlation between independent PRGs and clinical traits by wilcox.test or kruskal.test. As MFNG had the highest correlation with clinical traits, it was selected as the research object.

All OS patients fell into high- (> median, n = 21) and low- MFNG groups (≤median, n = 21) as per the MFNG expression level. Survival analysis of MFNG in the prognostic model was implemented with the aid of the K-M curve and log-rank test via the "survival" and "survminer" packages in R.

Statistical assessment

R-3.5.0 was utilized for assessment of all data. Hypothesis testing in categorical and continuous variables was implemented by the Fisher exact and Wilcoxon rank-sum tests, respectively. The limma package was employed for the assessment of differentially expressed genes (DEGs). Further, the survival was estimated by use of the K-M method and multivariate CRA, and the log-rank test was carried out for intergroup comparisons. Additionally, enriched Gene Ontology (GO) terms were filtered by use of "Clusterprofiler" package.

Results

Identification of MFNG as a target gene

We got GSE39058 from GEO database (https://www.ncbi.nlm.nih.gov/geo/). As per the probe information from Illumina HumanHT-12 WG-DASL V4.0 R2 Expression BeadChip (GPL14951), the annotation of 20,791 genes in 47 OS samples from GSE39058 was annotated. The results of screening of PRGs in OS with K-M curves (KM < 0.01) and CRA (coxPvalue < 0.01) by “survival” R package are shown in Additional file 1: Table S1. After multivariate CRA was adopted for independent prognostic analysis (p < 0.01), we obtained the genes significantly associated with independent prognosis. Next, wilcox.test or kruskal.test (p < 0.05) displayed the correlation between independent PRGs and clinical traits (age,gender,and recurrence) (Table 1). It was uncovered that MFNG was associated with two clinical traits (gender and recurrence), so we chose MFNG for follow-up study.

Table 1 The correlation between independent prognostic-related genes and clinical traits
Full size table

Survival outcomes in high-MFNG group and low-MFNG group

All OS patients fell into high- (> median, n = 21) and low- MFNG groups (≤median, n = 21) as per the MFNG expression level. Survival analysis of MFNG in the prognostic model was implemented with the aid of the K-M curve and log-rank test via the "survival" and "survminer" packages in R. As shown in Fig. 1, OS patients in high-MFNG group had worse outcome than those in low-MFNG group.

Fig. 1
figure 1

Survival outcomes in high-MFNG group and low-MFNG group. Osteosarcoma patients in MFNG high group had worse outcome than that in MFNG low group

Full size image

In univariate and multivariate CRA, MFNG exhibited notable relevance to the prognosis of OS and was proven to be an independent prognostic factor (PF) (Fig. 2A, B; Additional file 2: Table S2).

Fig. 2
figure 2

Cox regression analysis in high-MFNG group and low-MFNG group. A Univariate Cox regression analysis and B multivariate Cox regression analysis screened out the independent prognostic-related factor

Full size image

Differences in clinical data of OS cases between high-MFNG group and low-MFNG group

A report shows that worldwide, 15–19-year-old boys and 10–14-year-old girls displayed the highest OS prevalence, in line with age of puberty [14]. Age groups include < 10-year-old group, 10–20-year-old group, and > 20-year-old group. In the current study, the differences in clinical data of OS cases between high-MFNG group and low-MFNG group were analyzed through "ggpubr" package. There was no significant difference in the age between high-MFNG group and low-MFNG group (Fig. 3A), while there were significant differences in the gender and recurrence (Fig. 3B, C). It can be concluded that male OS patients have a higher tendency of low MFNG expression, and OS patients with no recurrence have a higher tendency of low MFNG expression.

Fig. 3
figure 3

Differences in clinical data of OS cases between high-MFNG group and low-MFNG group. Subgroups analysis of MFNG in accordance with different A age, B sex, and C recurrence

Full size image

Analysis of DEGs and pathways in high-MFNG group versus low-MFNG group

With the aim for determining MFNG-associated genes, the expression of DEGs was compared between high-MFNG group and low-MFNG group. The results showed 19 up-regulated genes and 3 down-regulated genes (adj.P < 0.05, (FC, log2) > 0.5 or < -0.5, Fig. 4A, B, Additional file 3: Table S3). Later, the enriched GO terms were assessed by use of DEGs. It was uncovered that a majority of DEGs were enriched in GO terms involving the biological process, including Notch signaling pathway (GO:0007219), positive regulation of phosphatidylinositol 3-kinase signaling pathway (GO:0008593), regulation of Notch signaling pathway (GO:0008593), regulation of phosphatidylinositol 3-kinase (PI3K) signaling pathway (GO:0014066) and regulation of mRNA processing (GO:0050684) (Fig. 4C, Additional file 4: Table S4).

Fig. 4
figure 4

Analysis of DEGs and pathways in high-MFNG group versus low-MFNG group. A Heatmap and B volcano plot show DEGs between high-MFNG group versus low-MFNG groups. Cut-off criteria for DEGs significance was adj.P < 0.05 and the absolute value of the log2 fold change > 0.5. C GO result for differential expression genes

Full size image

Module screening from the protein–protein interaction (PPI) network

We measured the correlation among the top 22 DEGs in high-MFNG group and low-MFNG group (Fig. 5A). Additionally, the PPI network of 22 DEGs was built with the use of the String database, which unveiled interactions among most of the up-regulated genes and all down-regulated genes (Fig. 5B).

Fig. 5
figure 5

Module screening from the PPI network. A The correlation analysis of DEGs with the Pearson correlation coefficient. B PPI network of top DEGs

Full size image

MFNG promotes the cell proliferation and inhibits the cell apoptosis of U2OS cells

In Fig. 5B, we can see that hematopoietic cell signal transducer (HCST) is co-expression with MFNG. Reports showed that HCST affects cell proliferation and survival by participating in the natural killer and T cell responses and in the activation of PI3K-dependent signaling pathway [15]. Thus, we want to explore the role of MFNG in OS cells. We overexpressed MFNG in OS cell line U2OS and found that MFNG overexpression (MFNG OE) promoted the cell proliferation and inhibited the cell apoptosis of U2OS cells (Fig. 6A, B).

Fig. 6
figure 6

MFNG promotes the cell proliferation and inhibits the cell apoptosis of U2OS cells. A CCK-8 detected the proliferation of U2OS cell after transfection. B Cell apoptosis of U2OS cell after transfection was detected. *P < 0.05

Full size image

Discussion

OS is the most found malignancy of the bone arising in childhood and adulthood, showing a high tendency of invasion and metastasis [1, 16, 17]. Since 2000, threefold elevation of the OS incidence rate has been observed in 0–24-year-old cases [2]. Up-to-now therapies for OS patients are neoadjuvant chemotherapy, surgery, radiotherapy and chemotherapy [18, 19]. Nonetheless, the high malignancy of OS results in poor overall survival of OS patients, especially those in the advanced stage [2, 20, 21]. Hence, in-depth deciphering of the molecular pathomechanism by which OS occurs and progresses appears to be a necessity, which benefits filtering of pivotal molecules or biomarkers for early diagnosis, targeted therapy, and prognosis assessment of OS.

Our research demonstrated that lower MFNG expression in OS cases showed relevance to satisfactory prognosis and low recurrence. MFNG has also been reported to be associated with the recurrence of ovarian carcinoma [22]. As unveiled by a study of Zhang et al., a high MFNG expression was detectable in CLBC and served as an oncogene by activating Notch signaling, thus hastening tumor cell migration, tumorsphere formation, and epithelial-to-mesenchymal transition (EMT) [9]. Beyond that, MFNG turns out to be a determinant player in optimal B and T cell development. In detail, it leverages a facilitating role in Th1 cell development and a suppressing role in Th2 cell development [23, 24].

Likewise, enrichment of the GO pathways is mainly detectable in the Notch signaling pathway and regulation of PI3K signaling pathway. Notch signaling pathways are a class of highly conserved signaling pathways in multicellular organisms that mediate the influence and regulation of the external environment on cells by participating in intercellular interactions, and programmatically manipulate cell fate and tissue differentiation in the early development of organisms [25]. Notch signaling pathways modulate cell invasion, adhesion, proliferation, apoptosis, and differentiation by virtue of cell–cell interactions [26, 27]. Beyond that, the growth and progression of human malignant tumors including OS appear to be subjected to influences by the Notch signaling pathway [28, 29]. For example, the Notch signaling pathway probably drives the differentiation of bone marrow mesenchymal stem cells into osteoblasts, thus facilitating ectopic bone formation [30]. DLX5 activates the Notch signaling pathway to benefit OS progression [31]. Suppressing the Notch signaling pathway put a brake on the progression of cell motility, metastasis, and EMT-like phenomena resulting from low-level cisplatin in OS [32]. Overexpression of lncRNA CEBPA-AS1 keeps a rein on OS cell proliferation and migration and elicits their apoptosis via the Notch signaling pathway [33]. On the other hand, activating the Notch signaling pathway can control the differentiation of macrophages into M1 and benefits inflammation and antitumor activity, but impeding the Notch signaling pathway polarizes macrophages to M2, so as to hinder inflammation and drive tumor growth [34].

PI3K, a family of lipid kinases, modulates a cascade of physiological cell processes such as metabolism, motility, exocytosis differentiation, proliferation, apoptosis, exocytosis and autophagy [35]. Subsequent to activation at the cell surface, the PI3K signaling pathway integrates signals from cytokines, growth factors and environmental signals and delivers them to effector molecules controlling protein synthesis, growth, survival and proliferation via protein kinase B (AKT) and mammalian target of rapamycin (mTOR) [36, 37]. AKT is the vital mediator of the PI3K signaling pathway, and its abnormal activation is interrelated to a multifold of malignant tumors [38, 39], including OS. For example, LINC00968 activates the PI3K/AKT/mTOR signaling pathway to exert an oncogene action [40]. PI3K inhibitors impair tumor progression and enhance sensitivity to anlotinib in anlotinib-resistant OS [41]. MDA19 inhibits OS via suppressing the PI3K/Akt/mTOR signaling pathway [42].

The PPI network also revealed that hematopoietic cell signal transducer (HCST) was co-expressed with MFNG. Reports show that HCST participates in the activation of PI3K-dependent signaling pathway and in the natural killer and T cell responses, which affect cell survival and proliferation [15]. Besides, HCST is interrelated to the poor prognosis of clear cell renal cell carcinoma [43]. We also performed CCK-8 and cell apoptosis assay in U2OS cells which were transfected with vector or MFNG OE. We found that MFNG OE promoted the cell proliferation and inhibited the cell apoptosis of U2OS cells.

However, there still some limitations in our study. Firstly, the current research is only based on public databases, we will further collect clinical specimens to verify our conclusion. Secondly, the specific physiopathologic mechanism developed by MFNG to function in OS cells has not yet been fully understood. In the future, we will conduct more detailed molecular mechanism research and gain a deeper understanding of MFNG at the OS genome level.

Conclusions

In closing, the low MFNG expression is a good PF in OS cases, while the high MFNG expression is associated with recurrence of OS. Additionally, enriched GO terms and PPI networks pertinent to OS shed light on the delving into the pathogenesis of high and low MFNG expressions.