Potential value of expression of receptor accessory protein 4 for evaluating the prognosis of lower-grade glioma patients

Abnormal increase of REEP4 expression in LGG pathological process

Abnormally high expression of pathogenic genes is a common phenotype in the pathological progression of malignant tumors [17]. Therefore, to explore whether REEP4 is also abnormally overexpressed in the pathological process of LGG, we first searched the mRNA expression level of REEP4 in LGG in the GEPIA database based on unpaired t-test results. Our findings suggested that REEP4 expression is significantly higher in LGG than in normal brain tissue (Figure 1A). In addition, RT-qPCR revealed that the expression level of REEP4 in clinical tissue samples and LGG cell lines was significantly higher than that in the control group (Figure 1B, 1C). Finally, the translation of mRNA into protein is an important form of its regulatory effect. Therefore, we found that the protein expression level of REEP4 in LGG tissue samples was also significantly higher than that in the control group using immunohistochemical methods (Figure 1D, 1E). Collectively, these results suggest that REEP4 may play an important regulatory role in the malignant evolution of LGG.

Figure 1. Differential expression level of receptor accessory protein 4 (REEP4) in LGG and normal brain tissue. (A) REEP4 is significantly upregulated in 518 lower-grade glioma (LGG) and 207 normal brain tissues based on GEPIA. ^*p < 0·05. (B) Real-time quantitative polymerase chain reaction (RT-qPCR) of clinical samples showing elevated REEP4 expression in LGG tissues. ^**p < 0·01. (C) RT-qPCR showing elevated REEP4 expression in LGG cell lines. ^***p < 0·001. (D, E) Immunohistochemistry shows elevated REEP4 expression in LGG tissue compared with than that in normal brain tissue. ^**p < 0·01.

REEP4 has a positive expression relationship with the malignant clinical phenotype of LGG patients

As the prognosis of patients with LGG is significantly correlated with their clinical characteristics [18], this study attempts to reveal the relationship between REEP4 expression and the clinical phenotype of patients with LGG based on the results of CGGA microarray, CGGA RNA-seq, and TCGA RNA-seq. The mRNA expression level of REEP4 was significantly higher in patients with WHO grade III and recurrence than in those with WHO grade II and primary disease (Figure 2A, 2B). Besides, the mRNA expression level of REEP4 was significantly higher in patients with wild type isocitrate dehydrogenase (IDH) and 1p19q non-codeletion than in IDH-mutation-type and 1p19q-co-deletion-type patients (Figure 2C, 2D). Furthermore, the mRNA expression level of REEP4 was significantly higher in chemotherapy and radiotherapy groups than in non-chemotherapy and non-radiotherapy groups (Supplementary Figure 1A, 1B). The mRNA expression level of REEP4 is higher in mesenchymal types than in classical, proneural, and neural types (Supplementary Figure 1C). Finally, the mRNA expression level of REEP4 also differed significantly among groups classified according to histological type (Supplementary Figure 1D–1F). Collectively, these results suggest that the high REEP4 expression may be closely related to the poor prognosis of LGG.

Figure 2. Relationship between receptor accessory protein 4 expression, clinical features, and molecular typing. (A) World Health Organization grade. (B) Primary, Recurrent, Secondary (PRS) type. (C) Isocitrate dehydrogenase mutation status. (D) 1p19q co-deletion status.

Overexpression of REEP4 may be an independent risk factor for LGG prognosis

It is significant that pathogenic genes lead to reduced survival time in patients with cancer [19]. Therefore, we tried to reveal the impact of REEP4 on the survival time of patients with LGG. Firstly, the results of Kaplan-Meier curves showed that the overall survival time of patients with LGG in the high REEP4 expression group was significantly lower than that in the low REEP4 expression group based on the CGGA microarray, CGGA RNA-seq, and TCGA RNA-seq (Figure 3A). It is worth emphasizing that the impact of REEP4 on the disease free survival time of patients with LGG has similar results (Supplementary Figure 2A). Subsequently, ROC analysis showed that high REEP4 expression had good diagnostic value for the prognosis of patients with LGG based on the above three datasets (Figure 3B). These results suggest that the high expression of REEP4 was associated poor prognosis of LGG.

Figure 3. The impact of REEP4 on the prognosis and diagnostic value of LGG patients. (A) Kaplan-Meier analysis based on the different expression levels of receptor accessory protein 4 (REEP4) in lower-grade glioma (LGG). (B) Diagnostic value of REEP4 in LGG by receiver operating characteristics.

To verify the above points, we first performed a meta-analysis to collect 1100 LGG patient samples from five different datasets. Each of the five independent datasets suggested that REEP4 was a risk factor for the overall survival of LGG, and the overall risk coefficient also suggested that REEP4 was a significant risk factor for the overall survival of LGG (Supplementary Figure 2B). The more important finding is that both univariate and multivariate analysis suggest that REEP4 can be used as an independent risk factor to affect the prognosis of patients with LGG (Supplementary Figure 2C–2H). In this part, various statistical methods and data sources are used to confirm that the high expression of REEP4 can be used as an independent risk factor as an ideal target for the diagnosis and treatment of LGG.

The high expression of REEP4 is negatively regulated by its methylation in LGG

Based on the effect of high expression of REEP4 on the prognosis of LGG, we aimed to reveal the reason for the increased expression of REEP4 in LGG. DNA methylation is an important epigenetic regulatory mechanism that has an important regulatory effect on mRNA expression [20]. Therefore, we analyzed the TCGA-RNA-seq and TCGA-DNA methylation data of REEP4 in LGG to reveal their relationship. First, we extracted the methylation status of 14 DNA methylation sites that regulate REEP4 expression and summarized them in Figure 4A. Second, co-expression analysis showed that the overall methylation level of the 14 DNA methylation sites exhibited a significant negative correlation with the mRNA expression level of REEP4 (Figure 4B). Third, co-expression analysis of single methylation sites showed that 6 of the 14 methylation sites (cg13265914, cg16311504, cg07664173, cg20582089, cg19738333, cg02498268) showed significant negative correlations with REEP4 expression (Figure 4C–4H). Only one methylation site (cg02399048) showed positive correlation (Figure 4I), and the remaining methylation sites showed no regulatory relationships. Fourth, the Kaplan-Meier curves showed that only one methylation site (cg16311504) affects LGG prognosis, and its hypermethylation status can prolong the overall survival of patients (Figure 4J). Meanwhile, clinical correlation analysis showed that the hypermethylation status of cg16311504 was negatively related to the clinical and molecular features associated with poor prognosis of patients with LGG (Supplementary Figure 3). This suggests that cg16311504 has a protective effect on LGG prognosis. Fifth, after the SHG-44 of LGG cell line was treated with SAM (methylating drug), RT-qPCR results revealed that REEP4 expression was significantly inhibited. Collectively, the hypothesis that the increased expression of REEP4 in the malignant process of LGG was negatively regulated by its methylation sites was confirmed.

Figure 4. Methylation level of receptor accessory protein 4 (REEP4) in lower-grade glioma (LGG) based on the TCGA database. (A) Methylation sites of REEP4 in LGG. (B) The expression level of REEP4 is negatively correlated with the total methylation level. (C–H) The expression level of REEP4 is negatively correlated with methylation sites cg13265914, cg16311504, cg07664173, cg20582089, cg19738333 and cg02498268. (I) The expression level of REEP4 is positively correlated with the methylation site cg02399048. (J) Effect of methylation site cg16311504 on the overall survival of patients with LGG. (K) Real-time quantitative polymerase chain reaction showing that the expression level of REEP4 was significantly inhibited by SAM. ^****P < 0·0001.

Co-expression analysis and enrichment analysis of KEGG cell signal pathway of REEP4 in LGG

To address how highly expressed REEP4 participates in the regulation of LGG pathological progression, we first determined that REEP4 was co-expressed with hundreds of genes through gene overexpression analysis and presented the top five genes with the most positive and negative correlations in Figure 5A. In addition, we found that REEP4 was primarily enriched in cancer-related cell signal pathways such as the cell cycle, MAPK signaling pathway, HIF-1 signaling pathway, NOD-like receptor signaling pathway, TNF signaling pathway, etc., (Figure 5B) through KEGG signal pathway analysis. Therefore, highly expressed REEP4 may play a regulatory role through the above signaling pathways in promoting the malignant progression of LGG. At the same time, further literature searches found that activating these cellular pathways not only promoted the proliferation and migration of tumor cells, but also played an important role in regulating tumor related immune cells.

Figure 5. Exploring the mechanism of REEP4 in the pathological progression of LGG. (A) Gene co-expression analysis enumerating the first five genes that were positively and negatively correlated with receptor accessory protein 4 (REEP4) in lower-grade glioma (LGG). (B) KEGG signal pathway analysis showed the pathway of significant enrichment of REEP4 in LGG.

Highly expressed REEP4 participates in the formation of LGG inhibitory immune microenvironment

The tumor-specific immunosuppressive tumor microenvironment is closely related to the abnormal expression of transcriptomics [21]. Based on the above results, we tried to reveal whether abnormal REEP4 expression is involved in the formation of the LGG immune microenvironment. Therefore, using the TIMER database, we first found that REEP4 was negatively correlated with the purification of cancer cells and exhibited a high positive correlation with the degree of infiltration of five different immune cells (B cell, CD4+ T cell, macrophage, neutrophil, and dendritic cell) (Supplementary Figure 4A). Further survival analysis showed that the overall survival of patients with LGG in the high infiltration group with six infiltrating immune cells was significantly reduced (Supplementary Figure 4B). In addition, it is interesting that the copy number variation of REEP4 can also affect the degree of immune cell infiltration (Supplementary Figure 4C). These findings suggest that the adverse effect of REEP4 on the prognosis of patients with LGG may be partly induced by effects on the immune microenvironment.

Due to the lack of subtype classification of immune cells in the TIMER database, we further used ESTIMATE and CIBERSORT to supplement and verify the impact of REEP4 on immune cell microenvironment based on TCGA-RNA-seq results. We first performed three groups of immune classification grouping (stromal score, immune score, and ESTIMATE score) and scoring, and found that the high REEP4 expression had higher scores than the low expression group (Figure 6A). In further immune cell typing, the expression level of REEP4 differed significantly among various immune cells, especially in M2 macrophages (Figure 6B). Subsequently, we focused on the relationship between REEP4 expression and the markers of 16 immune cells, and found that all immune cell markers were positively correlated with REEP4 in LGG except NOS2 (M1 Macrophage), which had a negative correlation (Table 1).

Figure 6. The effect of REEP4 on the immune microenvironment of LGG. (A) The expression level of receptor accessory protein 4 (REEP4) was positively correlated with the stromal, immune, and ESTIMATE scores. ^***P < 0·001. (B) The correlation between REEP4 and 22 immune cell subtypes. ^*P < 0·05. ^**P < 0·01. ^***P < 0·001. (C–K) REEP4 expression is positively correlated with 8 immune checkpoints, including CD274, PDCD1, KLRB1, LAG3, CD96, CD276, CTLA4, HAVCR2, CD163.

Finally, anti-tumor immunotherapy is a current research hotspot. Identifying new immune checkpoints will help researchers develop new drugs for anti-tumor immunotherapy [22]. Therefore, we tried to reveal the relationship between REEP4 expression and known immune checkpoints based on TCGA-RNA-seq results and found that REEP4 had a high positive expression relationship with the eight most common immune checkpoints (Figure 6C–6K). Therefore, revealing the relationship between REEP4 and immune checkpoints will help to improve the progress of anti-tumor immunotherapy.

Data collection

Gene Expression Profiling Interactive Analysis (GEPIA: http://gepia.cancer-pku.cn/) is an online analysis data platform that integrates the transcriptome sequencing of tumor tissues in The Cancer Genome Atlas (TCGA) database and the normal tissue samples from the GTEX projects to explore the changes in expression and prognosis of target genes in tumor tissues [53]. In the present study, we used the GEPIA database to search the expression changes of REEP4 and its impact on disease-free survival in LGG.

TCGA database (https://portal.gdc.cancer.gov/) is a professional data platform for cancer research of individuals of Caucasian and African descent and includes various data entries on cancer patients, e.g., transcriptome sequencing, DNA methylation, somatic mutation, etc., [28]. In the present study, we collected 503 RNA-seq and 511 DNA methylation data entries and corresponding clinical and molecular characteristics of the patients. Additionally, detailed patient information has been included in Supplementary Table 1. The above data were downloaded through the Xena portal (https://xenabrowser.net/) and used to explore the correlation between REEP4 and the prognosis, clinical features, DNA methylation, and immune microenvironment of patients with LGG.

The CGGA database is a public data platform for the pathological mechanism of human gliomas of Asian descent [54]. In this database, we obtained two types of data on LGG, namely CGGA microarray (142 cases) and CGGA RNA-seq (403 cases), as well as the corresponding patient clinical information (Supplementary Tables 2 and 3). The above data were used to verify the prognostic impact of REEP4 on LGG in TCGA database.

In addition to the above public database, we also collected brain tissue samples from 5 patients with epilepsy and 7 with LGG, which were used to perform RT-qPCR to evaluate the change in REEP4 mRNA expression level. Additional tissue samples from 3 patients with epilepsy and 3 with LGG were used to detect the protein expression levels of REEP4 using immunohistochemistry. All tissue samples were obtained from the operating room of the Henan Provincial People’s Hospital and placed into a −80° refrigerator for storage until use. The collection of samples was approved by the Ethics Committee of Henan Provincial People’s Hospital (2020107).

Cell culture and treatment

To validate the expression patterns of REEP4 at the cellular level, human astrocyte (HA) cell lines and the SHG-44 of LGG cell line were selected for in vitro experiments. Cells were cultured using a complete medium formulated with 89% high glucose medium (Cat PM150210, Procell, China), 10% fetal bovine serum (FBS) (Cat 10099141, Gibco, USA), and 1% penicillin and streptomycin (Cat P1400, Solarbio, China). They were then incubated in a constant temperature incubator at 37°C with 5% CO₂. When cell fusion approached 100%, the cells were passaged at 50% and cultured for subsequent experiments. For the methylation assay, SHG-44 cells were treated with 100 uM of ademetionine disulfate tosylate (SAM) when their cell fusion approached 70% (Cat 97540-22-2, Topscience, China) for 10 h. The treated cells were harvested to identify the REEP4 RNA level.

RNA isolation and RT-qPCR

Total RNA was collected in accordance following the instructions of Total RNA Kit I (R6834-02, Omega, USA) and the concentration thereof was measured using NanoDrop (Thermo Fisher Scientific, USA). The cDNA was synthesized on the T100 Thermal Cycler (Bio-Rad, USA) as per the manufacturer’s instructions of NovoScript Plus All-in-one 1st Strand cDNA Synthesis SuperMix (Novoprotein, China) and quantitatively analyzed to determine the expression of REEP4 mRNA on StepOne Plus Real-Time PCR System equipment (Thermo Fisher Scientific) in accordance with the manufacturer’s instructions of NovoStart^® SYBR qPCR SuperMix Plus (Novoprotein). An internal reference gene, 18S, was used to standardize the mRNA expression of REEP4. The primer sequences of 18S and REEP4 are as follows. The results of cell and tissue RT-qPCR were calculated using the log₂^−ΔΔCT and ^−ΔCT modes, respectively. The forward and reverse primers of 18S are as follows: 5′-GTAACCCGTTGAACCCCATT-3′ and 5′-CCATCCAATCGGTAGTAGCG-3′, respectively. The forward and reverse primers of REEP4 are as follows: 5′-TCGTGCTGTGGCTGCTCTCA-3′ and 5′-CGATCTCCTTCTCATGGCGG-3′, respectively.

Immunohistochemistry staining

The frozen brain tissue samples were removed from the liquid nitrogen, and after being fixed, dehydrated, embedded, and sliced, the 4-μm-thick paraffin sections were used to detect the protein expression of REEP4. Firstly, after being dried in an oven at 55°C for 30 min, the sections were dewaxed in xylene twice for 10 min and dehydrated in 85% absolute ethanol, 95% absolute ethanol, 100% absolute ethanol and 100% absolute ethanol for 10 min one by one. The sections were then immersed in EDTA antigen repair solution (ZSGB-BIO, China) and heated in a microwave for 20 min, and the endogenous peroxidase was inactivated and 10% serum solution was used to block nonspecific antigen for 30 min. Lastly, the sections were incubated overnight in the 4°C refrigerator with the primary antibodies of REEP4 (1:50, Proteintech, China), respectively, The next day, after the secondary antibody was incubated for 1 h and the chromogenic solution was used to develop color, the IHC staining results were photographed under the 200× microscope and calculated by ImagePro-Plus software (version 6.0).

Meta-analysis of REEP4 in LGG

To confirm whether REEP4 is a risk factor for the prognosis of patients with LGG, more data were collected through meta-analysis to improve the reliability of the analysis results. Five datasets from different sources (CGGA microarray: 142 cases; CGGA RNA-seq: 403 cases; TCGA RNA-seq: 503 cases; GSE43378: 18 cases; GSE50025: 34 cases) were included in this analysis. The GSE43378 and GSE50025 datasets were obtained from GEO database, and the gene symbols were transformed based on GPL570 and GPL13938 respectively [55, 56]. The above data sets first used the Cox method to obtain the risk coefficient of REEP4 on the overall survival of patients with LGG in each independent dataset, then collected five different risk coefficients through meta-analysis using the “Meta” package in R software (Version number: R x64 4·0·3). Since I₂ statistics of >50 existed among the five datasets, a random-effects model was selected to complete the meta-analysis.

Difference analysis and KEGG enrichment analysis of REEP4

To reveal the effect of REEP4 on cell signaling pathway in LGG, we classified all samples into high and low expression groups according to the median expression level of REEP4 based on TCGA RNA-seq. Hundreds of differentially expressed genes between the two groups were then filtered out based on the criteria with a log fold change of >1 and P-value of < 0·05 [28]. The subsequent enrichment analysis of KEGG cell signal pathway was completed using the “clusterprofiler” package in R software based on a q-value of < 0·05. The “ggplot2” package in R software was used to graph meaningful results.

Immune correlation analysis of REEP4 in LGG

The TIMER database (https://cistrome.shinyapps.io/timer/) is an online analysis platform for a series of algorithms to reveal the impact of target genes on tumor immune microenvironments [57]. Firstly, this study searched the TIMER database to determine the relationship between the expression or variation of REEP4 and six different degrees of immune cell infiltration, as well as the impact of high infiltration of immune cells on patient prognosis.

ESTIMATE is a method used to assess tumor purity by estimating the proportion of mesenchymal and immune cells in tumor samples using gene expression signatures [58]. To study the relationship between REEP4 and tumor purity, we divided LGG samples into two groups with high and low expression according to the median expression level of REEP4 based on TCGA RNA-seq. Afterwards, the differences in the stromal, immune, and estimated scores of the two groups of patients were compared using the ESTIMATE package in R software. To further explore the relationship between REEP4 and the LGG tumor microenvironment, we used CIBERSORT, the current most widely used immune cell infiltration algorithm tool, to conduct further research [59]. In this study, we used the CIBERSORT package of R software to evaluate differences in immune cell infiltration for each sample in two groups of samples with high and low REEP4 expression.

Statistical analysis

The results of RT-qPCR and immunohistochemistry were statistically analyzed using the unpaired t-test. The chi-squared test, Pearson’s correlation coefficient, Kaplan-Meier curves, and Cox expression models were used to perform clinical correlation, co-expression, survival, and multivariate analyses, respectively. Time-dependent receiver operating characteristic (TD-ROC) curves were plotted to assess the diagnostic value of REEP4 in the prognosis of patients with LGG. All statistical results with p-values of < 0·05 are considered significant.

Role of the funding source

Transcriptomic data and clinicopathological characteristics of LGG patients were obtained from the following datasets: CGGA microarray: 142 cases; CGGA RNA-seq: 403 cases; TCGA RNA-seq: 503 cases; GSE43378: 18 cases; GSE50025: 34 cases.

Availability of data and materials

Most of the patient samples in the Bioinformation Analysis part are from various public databases, but the patient samples in the validation part are from the operating room of Henan Provincial People’s hospital.