Background
Ovarian cancer (OC) is the second leading cause of gynecological cancer death worldwide [
1]. Due to the lack of timely diagnosis at an early stage and the emergence of chemoresistance at a late stage, most patients with OC have a poor prognosis [
2,
3]. Therefore, distinguishing patients according to their treatment response and identifying the underlying mechanisms limiting the anticancer efficacy of chemotherapy drugs are crucial to ameliorate patient outcomes.
Recently, as the most promising treatment against cancers, immunotherapy has gradually revolutionized cancer treatment with the progression of research on tumor-immune interactions [
4]. An increasing number of studies have demonstrated that the immune system plays a significant role in cancer initiation, progression, and therapeutic responses [
5,
6]. Moreover, OC is an immunogenic tumor, the course of which can effectively be transformed by immunotherapy [
7]. Previous researches reported the prognostic value of the OC immune system and revealed the importance of tumor-related signaling pathways in the tumor immune microenvironment [
8]. Although several types of tumors exhibit effective responses to immunotherapy, especially immune checkpoint blockade, many patients fail to benefit from immunotherapy, which is related to the prognosis of various cancer types, and OC patients have a poor response to PD1/PD-L1 monotherapy [
9]. Therefore, screening out these patients and providing therapeutic targets to improve patient prognosis are of great significance. In addition, a clinical situation has emerged in which the tumor recurred despite seemingly lasting remission at initial treatment [
10,
11]. One of the possible explanations is an extra disruption of cancer-immune homeostasis. Consistent with this, mounting evidence has shown that the tumor immune microenvironment and systemic immune system strongly influence the efficacy of anticancer drugs [
12]. Local tumor immunity and systemic immunity can enhance or weaken the effect of anticancer treatment by regulating the composition and characteristics of the tumor microenvironment [
13]. This reminds us of the great significance of determining the mechanisms of immunity regulation in resensitizing patients with OC to anticancer therapy. As reported, high immune cell infiltration (ICI)-scoring OC patients with better clinical overall survival (OS), higher tumor mutation burden (TMB), higher immune checkpoint expression (PD1, PD-L1, PD-L2 and CTLA4) and higher sensitivity to two first-line chemotherapy drugs (paclitaxel and cisplatin) might benefit from immunotherapy, which means that the ICI score is an effective prognosis-related biomarker of OC and can provide valuable information on the potential response to immunotherapy [
14].
Interestingly, studies have revealed that long noncoding RNA (lncRNA) expression is reliably related to cancer prognosis. Additionally, lncRNAs involved in the infiltration, differentiation and function of immune cells, have great predictive value for the immune response [
15,
16]. And lncRNAs, play a key role in a wide range of biological processes, including the regulatory network of regulator gene, which plays a key role in cancer progression. Notably, an increasing number of clinical studies have also shown potential in diagnosis, prediction of prognosis, and therapeutic targets for OC [
17,
18]. However, only limited comprehensive investigations focused on the molecular regulatory mechanism between lncRNAs and immune infiltration have been conducted [
19]. Given the above factors, lncRNAs would be an excellent choice as immune biomarkers to predict OC prognosis. Furthermore, studies have indicated that lncRNAs are essential regulators in gene expression networks and participate in almost all biological processes, including tumorigenesis and progression, through diverse mechanisms at the transcriptional, posttranscriptional, and epigenetic levels [
20,
21]. As documented, a risk model based on 4 lncRNAs (CACNA1G-AS1, ACAP2-IT1, AC010894.3 and UBA6-AS1) involved in m6A regulation was identified to predict OS and therapeutic value in OC independently [
22].
Methylation modification acts as an essential component of epigenetic modifications associated with multiple pathological processes [
23]. Recent studies have presented methylation modification, which exists in mRNAs and lncRNAs, as an emerging mechanism in gene regulation [
24,
25]. To date, the six most prominent types of methylation modifications in the human genome have been reported, comprising N1-methyladenosine (m1A), N6-methyladenosine (m6A), 5-methylcytosine (m5C), N7-methylguanosine (m7G), N6,2'-O-dimethyladenosine (m6Am), and 5-hydroxymethylcytosine (hm5C) [
26]. This reversible posttranscriptional modification is regulated by methylation regulators, which are usually classified as “writers”, “erasers”, and “readers” based on their functions. The “writers” (methyltransferases) catalyze methylation. The “erasers” (demethylase), remove methylation modifications from RNA. Readers, as methylation binding proteins, recognize methylation and generate functional signals SSSS [
27]. Increasing evidence has demonstrated that methylation modifications participate in the progression of cancers, such as glioma, breast cancers, hepatocellular carcinoma and ovarian cancer [
28‐
30]. In terms of the importance of methylation and immune infiltration in tumors, it is conceivable that immune status alterations caused by the methylation modification of lncRNAs could contribute to cancer treatment. Additionally, the deregulation of lncRNAs by epigenetic alterations has been implicated in cancer initiation and progression. For example, the lncRNA SNHG12 acts as a mediator of chemoresistance in OC via epigenetic mechanisms [
31], and the N-methyladenosine reader YTHDF2 mediates lncRNA FENDRR degradation, promoting endometrial cancer (EC) progression [
32].
Moreover, studies have indicated that m6A modification of lncRNA MIR155HG promotes immune escape of hepatocellular carcinoma cells to upregulate PD-L1 expression [
33]. Similarly, the methylation level of lncRNA FAM83H-AS1 was related to the gene expression of FAM83H-AS1, which was related to immune cell infiltration in OC patients [
34]. These findings further inspired us to conclude that methylation modification of lncRNAs can modulate the immune microenvironment and thus modulate tumor cell biology characteristics. However, a systematic analysis is still lacking. Therefore, we hypothesized that methylation modification could affect the immune status and prognosis of OC by regulating immune-related lncRNAs. Here, we aimed to not only provide a promising immune prognostic model for predicting outcome and immune response but also explore the regulatory mechanism of tumor immunity conditions, thus helping to overcome the obstacles in immune evasion and chemoresistance of OC.
Methods
Clinical and profiling data
Clinical and profiling data were downloaded from The Cancer Genome Atlas (TCGA) database (
https://portal.gdc.cancer.gov/cart; up to January 9, 2022). The inclusion criteria were as follows: (1) the primary disease was diagnosed as ovarian cancer, removing patients who have ever been affected by other malignant tumors; (2) Patients with complete follow-up information including survival time, survival status, outcome, age, sex, clinical stage, and grade were selected to match with their RNA seq data. The main outcome of our study was overall survival. Patients without survival information were removed for further evaluation. Following these criteria, 363 OC cases were included. Besides, we have provided more details about clinical information of OC cases in Supplementary Table
1. The transcription data were scale normalized.
The methylation-related genes are 50 recognized methylation regulators extracted from previous studies including “writers”, “erasers”, and “readers” [
28,
35‐
37]. We obtained the profiles of methylation-related genes from the TCGA database. the expression matrixes of 50 methylation-related genes were retrieved from TCGA, including the expression data of 32 writers, 14 readers and 7 erasers (Supplementary Table
2).
We obtained immune-related genes of OC from previous study [
38]. In this study, the authors divided the 308 ovarian cancer samples from TCGA into high- and low-abundance immune subtypes based on the abundance of immune cell infiltration. Differentially Expressed Genes (DEG) between these two groups were determined with the R package limma package (Bioconductor version 3.0). Meanwhile, the authors downloaded immune-related genes from the website (IMMPORT:
https://www.immport.org/) in which genes with different immune functions were included. Then, the intersection of the DEGs and immune-related genes were obtained, which include 98 differentially expressed immune-related genes of OC. In our study, we used these 98 immune-related genes to obtain immune-related lncRNA for follow-up research (Supplementary Table
3).
Selection of immune- and methylation-related lncRNAs
Based on the level 3 data provided in TCGA, according to the classification of gene type in the Ensembl (
http://www.ensembl.org/) database, non-coding protein genes were selected as ncRNAs, and those with lengths greater than 200 bp were selected as lncRNAs. The expression matrixes of 50 methylation-related genes and 98 immune-related genes were retrieved from TCGA.
Methylation gene-related lncRNAs and immune gene-related lncRNAs by Pearson’s correlation analysis were calculated in 363 samples. Then, the 789 immune-related lncRNAs were identified for high correlation with the immune score (|Pearson R|> 0.35&P < 0.05) as several published articles have constructed several immune related lncRNA models [
39,
40]. After screening methylation gene-related lncRNAs and immune gene-related lncRNAs by Pearson’s correlation analysis, 1688 methylation-related lncRNAs (mrlncRNAs) and 789 immune-related lncRNAs (irlncRNAs) were identified. The criteria of |Pearson R|> 0.35 and p < 0.05 were used in the process. We selected 326 lncRNAs of the intersection of methylation-related lncRNAs (mrlncRNAs) and immune-related lncRNAs.
Before construction of a prognostic model, 21 lncRNAs were screened out significantly correlated with OS (p < 0.05) from the 326 lncRNAs. Then, we used computer-generated random numbers to assigned the 363 TCGA-OV samples into a row and numbered them 1–363 randomly, took a 50% random number of 182 patients as the Training group, and the remaining 181 patients as the Validation group. The training set was utilized to construct mrlncRNA and irlncRNA risk models. Some published articles have also constructed and validated relevant risk prediction models based on this way [
37,
40].
Survival analysis
Univariate Cox analysis was implemented by the survival package to investigate the prognostic value of lncRNAs in TCGA-OV patients (
https://mirrors.tuna.tsinghua.edu.cn/CRAN/web/ packages/survival/index.html). Only lncRNAs with a
P value < 0.05 were considered to be significantly associated with survival. The coxpbc, nomogram, cph, calibrate, and cindex functions in R were used to construct, validate, and calibrate the nomogram.
Construction and validation of an immune-related prognostic model
To construct the prognostic model, TCGA ovarian cancer samples were randomly divided into a training set and a validation set at a ratio of 1:1. In the training set, lncRNAs related to prognosis were screened by univariate Cox regression, and then the model was simplified by least absolute shrinkage and selection algorithm (LASSO) regression analysis (tenfold cross validation). Finally, the optimal model was obtained by Cox proportional risk regression analysis. The model followed the Akaike information criterion (AIC), and the model with the lowest AIC value was selected as the optimal model, The survminer package of R software was used to analyze the best cutoff value of this model and group the training set (the number of samples in the group was greater than 30% of the total number of samples in the training set). The log⁃rank test was used to compare the differences in TCGA-OV in the training set and validation set and generate the survival curve; The receiver operating characteristic (ROC) curves of 1, 3 and 5 years were analyzed. The discrimination of the model was evaluated by the area under the curve (AUC) and the C⁃index. The survival of the training set and validation set were analyzed with the same cutoff value. Then, the relationship among age, clinical-stage, grade, risk-group and the prognosis of ovarian cancer was analyzed by univariate and multivariate Cox regression in the training set and validation set. Finally, the 1-, 3- and 5-year survival rates were predicted by constructing a nomogram, and evaluated by a calibration curve (followed by the bootstrap method). The calibration curve showed the fitting degree between the predicted survival rate and the actual survival rate of the nomogram, to evaluate the prediction accuracy of the nomogram. Finally, the benefit of the nomogram in guiding clinical decision-making was evaluated by 5-year decision-making curve analysis, and compared with other pathological parameters. The above analysis process was completed by R4.0.3 (
http://www.r-project.org) and SPSS 21.0 software. Unless otherwise specified, two-tailed
P < 0.05 was considered to indicate statistical significance.
Estimation of immune cell type fractions
As a method based on gene expression profiles to characterize the cell composition of complex tissues, CIBERSORT is highly consistent with the basic truth estimations in many cancers. LM22, as a leukocyte gene signature matrix consisting of 547 genes, was used to distinguish 22 immune cell types including myeloid subsets, natural killer (NK) cells, plasma cells, naive and memory B cells and seven T cell types. CIBERSORT was combined with the LM22 signature matrix to estimate the fractions of these 22 immune cell types between the risk score and immune infiltration degree or immune checkpoints. The sum of all estimated immune cell type fractions is equal to 1 for each sample.
Construction or identification of markers associated with chemoresistance
The pRRophetic package was used to construct or identify markers associated with chemoresistance. Additionally, it was used to analyze the relationship between prognostic models and immune infiltration, checkpoint expression, drug resistance and IC50 of cisplatin.
Construction of ceRNA networks
Cell culture and treatment
Human ovarian cancer cell line A2780 was obtained from Shanghai Institute of Cell Biology,China Academy of Sciences.Cells were cultured in RPMI medium modified supplemented with 10% fetal calf serum (Gibco BRL, Grand Island, NY), 100 U/mL penicillin and 100 µg/mL streptomycin in a humidified incubator containing 5% CO2 at 37 °C.
Tissue samples
With the approval of the Jiangsu University ethics committee, epithelial ovarian cancer samples from patients with FIGO stage IIIC or IV were collected at Zhenjiang Maternal and Child Health Hospital (The Fourth Affiliated Hospital of Jiangsu University) and The Affiliated People's Hospital of Jiangsu University. All patients were treated with the standard care of platinum-based therapy after surgery, and informed consent was obtained from all patients. PFS was calculated from the time of surgery to the time of progression or recurrence. Platinum resistance or platinum sensitivity was defined as a relapse or progression within 6 months or 6 months after the last platinum-based chemotherapy, respectively. Each group had more than 12 patient samples. Clinical and pathological features are described in Table
1.
Table 1
Clinical and pathological features of EOC patients
EOC1 | 66 | serous mucinous carcinoma | IIIA1(i) | Low | > 6 |
EOC2 | 62 | serous carcinoma | IIIA2 | Moderate | > 6 |
EOC3 | 70 | mucinous carcinoma | IVA | High | > 6 |
EOC4 | 65 | serous carcinoma | IIIC | High | > 6 |
EOC5 | 63 | serous mucinous carcinoma | IIIC | Moderate | > 6 |
EOC6 | 64 | serous carcinoma | IVA | Moderate | > 6 |
EOC7 | 68 | serous carcinoma | IIIC | High | > 6 |
EOC8 | 67 | clear cell carcinoma | IVA | High | > 6 |
EOC9 | 69 | endometrioid carcinoma | IIIA1(ii) | Low | > 6 |
EOC10 | 71 | mucinous carcinoma | IIIC | Moderate | > 6 |
EOC11 | 65 | serous carcinoma | IVA | Low | > 6 |
EOC12 | 59 | serous carcinoma | IVA | High | > 6 |
EOC13 | 60 | serous carcinoma | IIIC | High | < 6 |
EOC14 | 68 | serous mucinous carcinoma | IIIC | Low | < 6 |
EOC15 | 68 | mucinous carcinoma | IVA | High | < 6 |
EOC16 | 65 | clear cell carcinoma | IIIC | High | < 6 |
EOC17 | 66 | serous carcinoma | IIIB | Moderate | < 6 |
EOC18 | 69 | serous carcinoma | IIIA2 | Moderate | < 6 |
EOC19 | 72 | serous carcinoma | IIIC | High | < 6 |
EOC20 | 66 | serous carcinoma | IIIB | High | < 6 |
EOC21 | 70 | serous carcinoma | IVB | High | < 6 |
EOC22 | 65 | serous carcinoma | IIIC | Moderate | < 6 |
EOC23 | 76 | serous carcinoma | IIIC | High | < 6 |
EOC24 | 60 | serous carcinoma | IVA | High | < 6 |
siRNA
The si-FTO and its corresponding negative control siRNA were purchased from Suzhou GenePharma Co.,Ltd (Suzhou, China). The details on siRNA are described in Table
2.
si-FTO | GenePharma Co.,Ltd | Human | UCGCAUCCUCAUUGGUAAUTT | AUUACCAAUGAGGAUGCGAGA |
Cisplatin
Cisplatin(cDDP) is provided by the central pharmacy of the Fourth People's Hospital of Zhenjiang City, and its diluent is phosphate-buffered saline (PBS). The details on siRNA are described in Table
3.
Table 3
Details on Cisplatin
Cisplatin | QiLu Pharmaceutical (Hainan) Co., Ltd | H20073652 | 10 mg | phosphate-buffered saline (PBS) | 1, 2, 4, 8, 16, 32 |
Transfection
Cells were seeded into 6-well plates (6 × 105 cells/well) and transfected with si-FTO and its negative control (Suzhou GenePharma Co.,Ltd, Suzhou, China) at a final concentration of 100 nM using Lipofectamine 2000 which concentration was 4 μl per well (Invitrogen, CA, USA). The time course is as follows: 1) 1.5 ml basic culture medium (serum-free and triple antibody free) was used to starve experimental cells for 1 h; 2) 250 μl of basic culture medium was mixed with 1 μl of si-FTO and its negative control; 3) 250 μl of basic culture medium was mixed with 4 μl of Lipofectamine 2000; 4) Fully mixed 2) and 3); 5) After incubating in incubator for 4–6 h, replaced with a complete medium (including serum and triple antibody) and incubated for another 24 h; 6) Collect samples and extract RNA.
The cells were plated in 6-well dishes (500 cells/well) and exposed to a specific dose of cDDP(0.5 µg/ml), and subsequently grown for 14 days. Next, the cells were fixed with 4% Paraformaldehyde (PFA), which were then stained with 0.3% crystal violet. Colonies containing more than 50 cells were identified using densitometry software (Image J) and scored as survivors.
Evaluation of proliferation and apoptosis
To evaluate cell proliferation and apoptosis, 6 × 105 cells were seeded into 6-well plates. After transfection, EOC cells were treated with cDDP and PPAR inhibitor for 24 h. Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) (In situ cell death detection kit, Fluorescein, Roche, Basel, Switzerland) was used to assess apoptosis according to the instructions. Also, the proliferative ability of ECO cells were analyzed by the the EDU (5-Ethynyl -2’- deoxyuridine) cell proliferation test kit(Ribobio, Guangzhou, China). The weaker the proliferation signal, the stronger the apoptotic ability of the tested cells, and vice versa.
Assessment of chemosensitivity to cDDP
The control group and FTO knockdown cells were separately plated into 96-well plates (5 × 103 cells/well) and exposed to various doses of cDDP (1, 2, 4, 8, 16 and 32 µg/ml). Then, 10 μl of CCK-8 solution (Vazyme, Nanjing, China) was added to each well, and the plate was incubated for 2 h in a humidified incubator. The absorbance of each well was measured at 450 nm using a Model 550 series microplate reader (Bio-Rad Laboratories). Cell viability was expressed as the ratio of treated cells to untreated controls at each dose or concentration. The IC50 value for each cell line was determined by nonlinear regression analysis using GraphPad Prism (GraphPad Software Inc., San Diego, CA).
Real-time Quantitative PCR
Total RNA was isolated using Trizol reagent. cDNA as synthesized using a FastQuant RT Kit (with gDNase) (#KR106, Tiangen, Shanghai, China) according to the manufacturer’s instructions. Quantitation of RNAs was carried out using a miRcute Plus RNA qPCR Detection Kit (#FP411, Tiangen). The raw qRT-PCR RNAs data were normalized to the spiked GAPDH or U6 levels as described previously. The quantitative PCR procedures were carried out with real-time PCR SYBR Green q-PCR Super-mix. The RNA expression levels were analyzed and quantified by calculating using the 2 −
ΔΔCt method. The PCR cycle parameters are as follows: 1) Pre denaturation at 95℃ for 15 min; 2) 95℃ denaturation for 10 s; 3) Annealing at 48℃ for 20 s; 4) 72℃ extension for 30 s; 5) a total of 45 cycles. The primers of RNA are listed in Table
4. The primer concentrations were provided in Table
5.
Table 4
Primers used for qRT-PCR
FTO | ACTTGGCTCCCTTATCTGACC | TGTGCAGTGTGAGAAAGGCTT |
RP5-991G20.1 | TGTTGCTCTTCTTCATGGCTCGTG | AGTGGATGGCTTCAATCTCGGTATG |
hsa-miR-1976 | CTCCTGCCCTCCTTGCTG | - |
MEIS1 | CTTCCCTCTCTTAGCACTGATT | AAAATAGAGGTTTTTCTGCGCG |
GAPDH | AATGCATCCTGCACCACCAA | GTAGCCATATTCATTGTCATA |
U6 | CTCGCTTCGGCAGCACA | AACGCTTCACGAATTTGCGT |
Table 5
The qRT-PCR reaction system
2xSuperReal Color PreMix | 5 μl |
Forward(10 μM) | 0.3 μl |
Reverse(10 μM) | 0.3 μl |
cDNA templet | 2 μl |
Rnase Free ddH2O | Fill in to 10 μl |
Statistical analysis
All numeric data are presented as the means ± standard deviations (SDs) of at least three independent experiments. The experimental results were analyzed by analysis of variance or two-tailed Student’s t test at a significance level of P < 0.05 (*P < 0.05, **P < 0.01 and ***P < 0.001) using Prism 5 software (GraphPad Software, San Diego, CA, USA). A P value < 0.05 was considered to indicate statistical significance. The remaining statistical analysis were performed with R version 4.0.3 software (package: GDCRNATools DEseq2, edgeR, ggplot2, clusterProfiler, glmnet, preprocessCore, survminer, survival, timeROC, rms, pheatmap, corrplot, and vioplot).
Discussion
The immune system is an important barrier to disease surveillance and clearance. Immune silencing is recognized as a vital hallmark of tumor development, which is associated with multiple characteristics of cancer, such as chemoresistance, transfer and invasion [
45]. Early studies indicated that alteration of immune cell infiltration status, modification of PD1/PD-L1 expression patterns, and polarization of immune cells contributed to the development of OC [
46]. However, we have not yet found effective signaling pathways to improve prognosis and immunotherapy response in OC, and our study was designed to solve this problem.
With the rapid development of bioinformatics analysis technology, we attempted to explore the potential value of the combination of immunity and methylation in OC and signaling pathways that regulate tumor immune responses based on the maximum transcriptomic sample data that could be acquired from TCGA and GEO. Growing evidence suggests that the expression of immune-related genes shows great prognostic value regarding tumor progression, and the methylation modification of noncoding RNA plays an important regulatory role in a series of biological functions [
47]. Thus, methylation and immune infiltration will greatly affect the outcome of patients with malignancy [
48,
49]. Similarly, it will inevitably affect the prognosis and treatment response of patients with OC.
As we known, the ICI score for predicting the prognosis of OC patients and providing the potential response to immunotherapy, painting a novel picture of regulation of immune response and immunotherapy, confirmed its association with clinical outcome [
14]. Here, Our study constructed a methylation- and immune-related risk model and ceRNA network. Moreover, we also explored the prognostic predictive ability of the risk model and its association with immune cell infiltration and assessed the reactivity of OC patients to chemotherapy. In this study, 4 mrlncRNAs and irlncRNAs were incorporated into a risk model. Kaplan‒Meier curves, ROC curves, Cox regression analysis, and nomograms showed that the risk model possessed excellent prediction ability and was an independent predictor of OC prognosis.
Emerging evidence has demonstrated that lncRNAs play an essential role in regulating immune cell infiltration, and some studies have also reported that methylation modification of lncRNAs regulates immune cell function in the tumor immune microenvironment [
50,
51]. In addition, we investigated the fundamental effects of the risk score on the regulation of immune cell infiltration. Consistent with previous reports [
52], the distribution of immune cell infiltration was significantly different between the low-risk and high-risk groups. Our study showed several interesting findings regarding the prognostic relevance of several well-known immune cells (e.g., CD8 + T cells, CD4 + T cells, and Tregs). CD8 + T cells and CD4 + T cells have been reported to be associated with better outcomes, whereas there are two different scenarios for Treg cells in OC. Our study indicated that CD8 + T cells and CD4 + T cells were associated with better outcomes among patients with low-risk OC but were associated with worse outcomes among patients with high-risk OC. Furthermore, CD8 + T cells are known to signify a favorable clinical outcome in a variety of tumors, including OC [
53]. Therefore, combined with previous studies, we found that CD8 + T cells can predict the prognosis of OC patients [
41]. The study also showed that the risk score was negatively correlated with the proportions of resting immune cells but positively associated with immunosuppressive cells, indicating that patients with low-risk scores were immunologically resting, while those with high-risk scores represented an immunosuppressive tumor microenvironment.
It has been reported that the response to anti-checkpoint blockade is affected by intertumoral infiltration of immune cells [
54]. With the development of immune checkpoint inhibitors, immune-checkpoint blockade (ICB) immunotherapy has generated promising therapeutic results in tumors [
55]. Unfortunately, the majority of OC patients do not respond to ICB treatment. Thus, we investigated the correlation of the risk scores, which are positively associated with immunosuppressive cells, and the expression of immune checkpoints to predict OC patients’ responses to immunotherapy. It has been reported that increased levels of immune checkpoints, such as PD-1 and CTLA-4, indirectly indicate preexisting T-cell activation, and patients might be more sensitive to ICI treatment [
56]. Consistent with this conclusion, we achieved the same results in the training and validation sets: high expression of PDCD1, LAG3, ICOS, CTLA4, and CD274 indicated lower risk scores and better outcomes of OC. Consistently, when patients had lower risk scores and high expression of immune checkpoint genes, the infiltration of protective immune cells such as CD8 + T cells and CD4 + T cells was obviously enhanced. This result suggests that we can improve the immune evasion of tumor cells caused by the overexpression of immune checkpoints by altering immune cell infiltration in the tumor microenvironment.
Furthermore, in our research, a methylation and immune-related ceRNA regulatory network including 208 mRNAs, 24 miRNAs, and 3 lncRNAs was constructed to investigate the potential molecular mechanism of tumor immunity. Combined with the results of previous studies and the data from the GEPIA online database, we surprisingly found that the lncRNA RP5-991G20.1 within the network was significantly downregulated in OC. The findings echoed previous results that the high expression of RP5-991G20.1 indicated a longer OS of patients with OC. Moreover, the m6A demethylase FTO, which is positively correlated with lncRNA RP5-991G20.1, was reported to inhibit OC progression and stem cell self-renewal through demethylase activity [
57]. Moreover, most of the participants within ceRNA networks are involved in drug resistance and immunomodulation of tumors [
58]. Among them, the target gene of lncRNA RP5-991G20.1, MEIS1, is reported to promote the migration and chemotaxis of CD8 + T cells and indicate a favorable prognosis for patients with OC [
59]. Therefore, we believe that the signaling pathway regulating the expression of RP5-991G20.1 and MEIS1 through the alteration of FTO will have a significant impact on the prognosis of OC.
Then, we validated the FTO/RP5-991G20.1/hsa-miR-1976/MEIS1 pathway in A2780 cells, and the results showed that FTO positively regulates the expression of RP5-991G20.1 and MEIS1 and is negatively correlated with the expression of hsa-miR-1976. Moreover, low-grade epithelial ovarian cancer (EOC) and high-grade EOC are two very different types of cancer, which is not well described in the TCGA database. To compensate for this deficiency, we selected low-grade EOC as the sensitive group (PFS > 6) and high-grade EOC as the drug-resistant group (PFS < 6). And then, we found that the expression levels of FTO, RP5-991G20.1 and MEIS1 were clearly increased in the sensitive group. However the hsa-miR-1976 had an opposite trend. A previous study reported that MEIS1 promoted the migration and chemotaxis of CD8 + T cells in OC [
59], while RP5-991G20.1 was significantly downregulated within the tumor tissue of OC. These results suggest that these signaling pathways can be a new tool to improve the chemoresistance and immune therapeutic response of OC.
Our study is the first to investigate the molecular mechanism affecting the prognosis of OC from the perspective of methylation and immunity. Additionally, mrlncRNA and irlncRNA prognostic models possessing high predictive ability for survival in OC were constructed, and a ceRNA network was developed based on mrlncRNA and irlncRNA that may provide novel ideas for the study of OC.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.