Correlation between Genes of the ceRNA Network and Tumor-Infiltrating Immune Cells and Their Biomarker Screening in Kidney Renal Clear Cell Carcinoma

This study aimed to using bioinformatics tools, qPCR, and the immunohistochemical analysis to find out factors related to the early diagnosis and prognosis of kidney renal clear cell carcinoma (KIRC). The expression profiles of lncRNA, miRNA, and mRNA of KIRC were downloaded from The Cancer Genome Atlas database. A ceRNA regulatory network was constructed based on the interaction between these three differentially expressed genes. The CIBERSORT deconvolution algorithm was used to analyze the differential distribution of 22 types of immune cells. The Kaplan–Meier survival and Cox analyses were used to screen genes of the ceRNA network and also immune cell subtypes related to the clinical and prognostic prediction of KIRC. Co-expression regulatory relationships were found among LINC01426, LINC00894, CCNA2, L1 cell adhesion molecule (L1CAM), and T follicular helper cells, which served as potential biomarkers. The results of quantitative reverse transcriptase-polymerase chain reaction showed that LINC01426 was upregulated while L1CAM was downregulated in KIRC, but no difference was found in the expression levels of LINC00894 and CCNA2 in cancer and adjacent samples. The immunohistochemical analysis showed that T follicular helper cells were more concentrated in core tissues and metastases of KIRC. In a word, co-expression relationships were found among LINC01426, L1CAM, and T follicular helper cells, and they may serve as biomarkers for early diagnosis and prognostic evaluation of KIRC.


Introduction
Kidney renal clear cell carcinoma (KIRC) originates from proximal tubular epithelial cells [1]. It is the most common and aggressive subtype of renal cancer, accounting for approximately 75%-80% [2]. Most patients are diagnosed in the advanced stage because the initial clinical symptoms and signs of KIRC are relatively hidden [3]. Compared with other subtypes of kidney cancer, KIRC has a higher recurrence rate and metastasis rate. Although surgical treatment, molecular-targeted therapy (sorafenib and sunitinib), immunotherapy (interleukin-2), and other treatments developed in recent years have greatly improved the survival time of patients; the 5-year survival rate is still less than 10% [4,5]. erefore, a biomarker that can detect KIRC early and predict its prognosis needs to be identified. Salmena et al. formulated a hypothesis about ceRNA in 2011; they believed that long noncoding RNAs (lncRNAs) use some core seed sequences to adsorb the corresponding miRNA, thereby interfering with the abundance of target gene mRNA and affecting gene expression [6]. A large number of studies have shown that ceRNA played a vital role in the occurrence, development, and prognosis of tumors [7]. For example, Wang et al. experimentally proved that lncRNA UCA1 was used as the ceRNA of miR-182-5p to positively regulate the expression of Delta-like4 (DLL4), thereby promoting the malignant phenotype of renal cancer cells and playing a carcinogenic role in the pathogenesis of renal cancer [8]. Human immune surveillance is an important immune function of the body to prevent tumors, and evading the destruction of the body's immune function is one of the important mechanisms of tumors [9,10]. In recent years, the distribution and density of local immune cells have received wide attention from scholars in tumor diagnosis and prognostic evaluation [11]. Studies have shown differences in infiltrating immune cells in different types of sarcoma [12]. Liang et al. found that Janus Kinase 3 (JAK3) moderately to strongly positively correlated with the abundance of B cells, CD8+ T cells, CD4+ T cells, neutrophils, and dendritic cells in KIRC, which may become potential biomarkers of KIRC [13]. Although a large number of studies have explored the correlation between infiltrating immune cells and tumor occurrence, development, prognosis, and so on, the specific mechanism of action in tumors has not yet been clearly elucidated.
is topic analyzed the potential roles of the ceRNA network and tumor-infiltrating immune cells in KIRC in tumorigenesis, metastasis, and prognosis. A flowchart explaining this process is given in Figure 1. In conclusion, this study might offer new ideas for prognostic monitoring of patients with KIRC and research on new treatment methods.

Data Acquisition and Differential Expression Analysis of
Genes. Metadata files, manifest files, and cart files of KIRC transcriptome and miRNA and patient clinical information were downloaded from e Cancer Genome Atlas (TCGA) database. After decompressing the cart file, the Perl script was run to obtain the original transcriptome and miRNA matrix files. e gene names were converted using the human. gtf file downloaded from the Ensembl database and the mature. fa file downloaded from the miRBase database. DESeq2 package in R4.0.2 software was used for differential expression analysis to obtain differentially expressed lncRNAs, miRNAs, and mRNA ((false discovery rate, FDR) < 0.05, |log (fold change)| > 2).

2.2.
Construction of the ceRNA Network. LncRNA-miRNA and miRNA-mRNA interactions predicted from the miRcode [14] and StarBase [15] databases, respectively, showing significant results in hypergeometric testing and correlation analysis, were selected for the visualization of the ceRNA network using the Cytoscape 3.7.2 software.

Clinical Significance of the ceRNA Network in KIRC.
Single-factor Cox regression, lasso regression, and multifactor Cox regression analyses were performed on all genes in the ceRNA network, and a risk scoring model was built for the selected genes. e diagnostic value of the model was assessed through the risk survival curve and receiver operating characteristic (ROC) curve.
e Kaplan-Meier survival method was employed to perform the survival analysis of all genes in the network in batches.

Abundance Analysis and Differential Expression Analysis of Infiltrating Immune Cells.
e gene expression feature set of 22 types of immune cell subtypes was downloaded from the CIBERSORT website. Based on the gene expression profile, the e1071 package was run to obtain the abundance of infiltrating immune cells and statistical accuracy (P value) of 22 types of immune cells in each sample (the number of permutations was set to 1000). e samples with P < 0.05 were retained for subsequent analysis. e difference in immune cells between KIRC tissue and adjacent tissues was analyzed by a two independent-sample t test.

Survival Correlation Analysis of Infiltrating Immune Cells in KIRC.
Single-factor Cox regression, lasso regression, and multifactor Cox regression analyses were conducted on infiltrating immune cells to build a risk assessment model. e risk survival curve and ROC curve were drawn to evaluate the diagnostic value of the model. e correlation between immune cell subtypes and clinical metastasis was predicted using the Wilcoxon rank-sum test ( e clinical research objects are all T staging in TNM). e Kaplan-Meier survival method was used to analyze the survival of all immune cells with different distributions.

Co-expression Analysis of Genes in the ceRNA Network and Immune Cells.
e relationship between ceRNAs and 22 types of immune cells was investigated using Pearson's correlation coefficient.

Quantitative Reverse Transcriptase-Polymerase Chain
Reaction. Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) was used to quantitatively express key genes in the ceRNA network. Clinical tissue cDNA chips were purchased from Shanghai Outdo Biotech Co., Ltd.
e chip lot number was cDNA-HKidE030CS01 (15 cases of renal clear cell carcinoma, 1 spot on the cancer/ adjacent, the RNA of the frozen sample was reversetranscribed into cDNA and spotted on a 96-well plate, and the samples covered clinical stage 1, stage 2, and stage 3.). e relative expression levels of lncRNA and mRNA in cancer and adjacent cancer samples of KIRC were detected using the PerfectStart Green qPCR SuperMix (TransGen Biotech, China) following manufacturer's instructions. Primers for lncRNA and mRNA are shown in Table 1. e reaction conditions were predenaturation at 94°C for 30 seconds; 94°C for 5 seconds, 60°C for 30 seconds, 40 cycles; finally, the temperature was lowered to 37°C for 20 min until 2 Journal of Oncology the reaction was completed.
e gene expression profiles downloaded from TCGA database were statistically analyzed in the R software (GDCRNATools, ggplot2, DESeq2, survival, glmnet, survminer). e results of qRT-PCR were analyzed by 2 −△△Ct and independent t test in IBM SPSS statistics 25.0.

Immunohistochemistry in Clinical Tissues.
Clinical tissue chips were purchased from Shanghai Outdo Biotech Co., Ltd. e chip lot number was KidE085CS01 (26 cases of renal clear cell carcinoma, one spot on the carcinoma/adjacent/distal, and 7 metastases, one site per metastases, and the samples covered clinical stage 1, stage 2, and stage 3). Immunohistochemical (IHC) staining was performed using a Leica BOND-MAX auto-stainer (Leica Instrument Co., Ltd., Germany), and the CD4 (EP204) rabbit mAb (48274, Cell Signaling Technology, China) was diluted to 1 : 200. e marker of T follicular helper cells was CD4, respectively [16]. IHC was performed as follows. Briefly, 4 μm thick tissue sections were cut with a microtome, deparaffinized in xylene, dehydrated through graded ethanol (100% and 95%), and rinsed with water. Subsequently, the sections were subjected to heat-induced antigen retrieval and finally loaded onto the Benchmark auto-stainer, and the detection was performed using a bond polymer refine detection kit (Leica Instrument Co., Ltd.).

Immunohistochemical Digital Pathological Analysis.
e expression levels of CD4 were estimated by QuPath (open source software for Quantitative Pathology, version 0.2.0) [17]. Each slice included 26 cores of tumor tissues, corresponding peritumor normal renal tissue, distal normal renal tissue, and 7 cores of metastatic renal carcinoma. Digitized IHC microarrays of CD4 were acquired at 100× magnification using an Olympus slide scanner (Olympus motorized BX61VS). e annotation of each core was manually delineated on the pathological slice, while the peritumoral region of the tumoral core and nonspecific staining was excluded. en, cells within the annotations were detected. e positive cell ratio, Allred score, and H score of each core were calculated to assess the expression levels of CD4. e process of digital analysis of IHC is shown in Supplemental Material 3.  Journal of Oncology 3     256  197  221  157  184  121  157  86  123  53  89  32  63  19  43  11  30  6  25  0  13  0  3  0  1 Risk High risk Low risk aforementioned analysis was performed using the IBM SPSS statistics 25.0 software. Only the two-sided P value <0.05 was considered to be of statistical significance.      Journal of Oncology

Composition of Immune Cells in KIRC.
e CIBERSORT algorithm was used to obtain the immune cell infiltration abundance of all samples, and 223 samples with P < 0.05 were retained for subsequent analysis. e heatmap and the violin map showed the difference in the distribution of immune cells between cancer and adjacent cancer samples ( Figure 5).

Clinical Correlation Analysis of Immune Cells in KIRC.
ree potential prognostic biomarkers (T-cell CD4 memory activated, T follicular helper cells, and resting mast cells) were regarded as key members among 22 types of immune cells and were integrated into a new multivariable model (Figures 6(a)-6(c)). e risk survival curve suggested that the survival rate of the high-risk group was considerably higher than that of the low-risk group (P � 0.006) (Figure 6(d)). e ROC curve (AUC of 1-, 3-, and 5-year survival was 0.587, 0.642, and 0.616, respectively) demonstrated the sensitivity and specificity of the model (Figure 6(e)).
e Wilcoxon rank-sum test suggested that resting mast cells had significant differences in T stage and stage (Figures 7(a) and 7(b)).
e results of the Kaplan-Meier survival analysis showed that plasma cells (Figure 7(c)), T follicular helper cells (Figure 7(d)), and regulatory T cells (Figure 7(e)) correlated with the survival of patients with KIRC.

Results of Clinical Tissue Specimen Verification
e qRT-PCR results showed that lncRNA LINC01426 was upregulated while mRNA L1CAM was downregulated in kidney cancer tissues, which was consistent with the expression pattern in the TCGA database (P < 0.05) (Figures 9(a) and 9(b)). However, there was no difference in expression levels of lncRNA LINC00894 and CCNA2 mRNA in renal cancer tissue and adjacent tissue (P > 0.05) (Figures 9(c) and 9(d)). e IHC results showed that the level of T follicular helper cells (CD4 marker positive) was the highest in the core of tumor tissues, which was significantly different from the corresponding normal renal tissue adjacent to cancer, distal normal renal tissue and metastatic renal cancer core tissue (P < 0.05). e level of T-follicular helper cells is the second highest in the metastatic renal cell carcinoma core tissue, which was a significant difference between adjacent normal renal tissues and distal normal renal tissue (P < 0.05) (Figures 9(e) and 9(f)). e results demonstrated

Conclusions
At present, many patients with KIRC whose diagnoses were mainly based on the clinical symptoms and imaging methods have already developed distant metastases at this time; the recurrence rate after surgical radical treatment was high [18]. In recent years, a large number of researchers have focused on exploring the mechanism of genes, tumorinfiltrating immune cells, and the interaction between the two in the occurrence, development, metastasis, and prognosis of KIRC, indicating that genes and immune cells were closely related to tumors, and provided direction for   the diagnosis and treatment of KIRC in the future [19][20][21][22]. For example, Zhengyan Chang et al. separately constructed the risk scoring model of the ceRNA network and infiltrating immune cells in colon cancer and found that T follicular helper cells and hsa-miR-125b-5p, macrophages M0 and hsa-miR-125b, and macrophages M0 and FAS might become potential biomarkers through co-expression analysis, and this conclusion was verified in clinical tissues [23]. is research model based on bone metastatic melanoma, gastric cancer, breast cancer bone metastasis, mesothelioma bone metastasis, and other tumors has been adopted by various studies [24][25][26]. e present study also used this model and used bioinformatics analysis to identify co-expression regulation relationships among LINC01426, LINC00894, CCNA2, L1CAM, and T follicular helper cells. ese key members might become KIRC diagnostic and therapeutic potential biomarkers. LINC01426 was upregulated in renal clear cell carcinoma tissues and its overexpression was correlated with a disappointing prognosis [27]. So far, the data on L1CAM expression in renal clear cell carcinoma were contradictory; studies have shown that cell adhesion, metastasis, and invasion abilities were significantly increased with the upregulation of L1CAM expression in KIRC, and in turn, the downregulation of LICAM expression decreased the proliferation of renal cancer cell and reduced the expression of cyclin D1 [28,29]. However, this just illustrated the importance of L1CAM in the progression of KIRC. T follicular helper cells are a specialized subset of CD4+ T cells that were first identified in tonsils in humans. ey play an essential role in forming germinal centers, and Xiaoliang Hua et al. found that tumors from high-risk patients had a higher relative abundance of T follicular helper cells [30]. e present study confirmed the high expression of LINC01426, L1CAM, and tumor infiltration of T follicular helper cells because these cells were closely related to the clinical and prognostic prediction of KIRC. us, these cells were found more likely to be KIRC biomarkers.
In conclusion, the present bio-report analysis indicated a relationship among LINC01426, L1CAM, and T follicular helper cells, which was meaningful. As it is difficult to detect the patient's immune cells, the abundance of T follicular helper cells in KIRC was determined by detecting the expression levels of LINC01426 and L1CAM, which have a coexpression relationship to provide new prospects for the early diagnosis of KIRC so as to develop new therapeutic drugs.
Data Availability e datasets used and/or analyzed in the present study are available from the corresponding author on reasonable request.

Ethical Approval
is study was approved by the Ethics committee of General Hospital of Ningxia Medical University adhering to the tenets of the Declaration of Helsinki. All tissue and cDNA microarrays purchased by Shanghai Outdo Biotech Co., Ltd. meet ethical requirements.

Conflicts of Interest
ere are no conflicts of interest regarding the publication of this paper.