Identification of microRNA editing sites in clear cell renal cell carcinoma

Clear cell renal cell carcinoma (ccRCC) is a malignant tumor originating from the renal tubular epithelium. Although the microRNAs (miRNAs) transcriptome of ccRCC has been extensively studied, the role of miRNAs editing in ccRCC is largely unknown. By analyzing small RNA sequencing profiles of renal tissues of 154 ccRCC patients and 22 normal controls, we identified 1025 miRNA editing sites from 246 pre-miRNAs. There were 122 editing events with significantly different editing levels in ccRCC compared to normal samples, which include two A-to-I editing events in the seed regions of hsa-mir-376a-3p and hsa-mir-376c-3p, respectively, and one C-to-U editing event in the seed region of hsa-mir-29c-3p. After comparing the targets of the original and edited miRNAs, we found that hsa-mir-376a-1_49g, hsa-mir-376c_48g and hsa-mir-29c_59u had many new targets, respectively. Many of these new targets were deregulated in ccRCC, which might be related to the different editing levels of hsa-mir-376a-3p, hsa-mir-376c-3p, hsa-mir-29c-3p in ccRCC compared to normal controls. Our study sheds new light on miRNA editing events and their potential biological functions in ccRCC.


Results
Summary of mutation and editing sites identified in miRNAs.The MiRME pipeline 43 was used to identify miRNA M/E sites from the collected 176 sRNA-seq profiles.In total, 1025 M/E sites with significant editing levels were identified (as listed in Supplementary Table S2.1).As mentioned above, these M/E sites were divided into 9 categories (Fig. 1A and Supplementary Table S2.2).We found that there were more 3 ′-addition sites than 5 ′ editing sites (Fig. 1A and Supplementary Tables S2. 3, S2.4).In addition, M/E sites in the central regions of mature miRNAs include 9 (0.88%) A-to-I, 30 (2.93%) C-to-U, 5 (2.54%)Other and 2 (0.20%) SNP sites.
We further investigated the miRNA editing events that occured in the central regions of mature miRNAs (Supplementary Table S2.5),i.e., A-to-I, C-to-U and Other.We counted the base changes of these three editing types (Fig. 1B), and found 9 A-to-I, 30 C-to-U, and 5 Other editing events.One insertion and two deletion events were also found (Fig. 1B).Furthermore, we also investigated the distributions of these events within different regions of mature miRNAs, including the 5 ′ ends, 3 ′ ends, and central regions.Our results indicated that 3 ′ www.nature.com/scientificreports/-addition was the most common type of RNA editing events of miRNAs.As shown in Fig. 1C, the numbers of editing events at the 3 ′ ends were significantly higher than those at the 5 ′ ends and the central regions.For the vast majority of miRNAs, only 1 or 2 editing events happened at the 5 ′ ends and the central regions.
Nine miRNA A-to-I editing sites in ccRCC .Totally, 9 A-to-I editing sites with significant editing levels were identified from the 176 samples (Fig. 2A).The editing levels of hsa-mir-376a-1 and hsa-mir-376a-2 were almost identical, probably due to their high sequence similarity.We then counted the nucleotides around these 9 A-to-I editing sites, and found that the 5 ′ and 3 ′ ends of these A-to-I editing sites had clear U and G base preference, respectively (Fig. 2B), which was consistent with results in literature 27,29,43 .As examples, the detailed information of two A-to-I editing sites in hsa-mir-411 and hsa-mir-7977 in two ccRCC samples (SRR11873730 and ERR4367258, respectively) were presented in Fig. 2C and D, respectively.As shown in Fig. 2E and F, a large number of sequencing reads in these samples supported the two A-to-I editing sites.Thirty miRNA C-to-U editing sites in ccRCC .We carefully studied the 30 C-to-U editing sites and found that their editing levels were generally low in the 176 sample (Fig. 3A).Furthermore, nucleotides on the 5 ′ sides of these C-to-U editing sites were biased toward C, while those on the 3 ′ sides were biased toward A (Fig. 3B).Examples of two C-to-U editing sites were shown in Fig. 3C and D, respectively.In addition, the detailed editing sites and editing levels of these two miRNAs were shown in Fig. 3E and F, respectively.These two C-to-U editing sites were supported by thousands of sequencing reads in the samples used, indicating good reliabilities of these two editing sites.
Two miRNA SNP sites in ccRCC .The identified M/E sites were compared with SNPs in miRNAs previously reported 44 and NCBI dbSNP (v151) according to the criteria mentioned in Materials and Methods.Two SNPs (rs11614913 and rs2155248) were shown in Supplementary Fig. S1 and Supplementary Table S2.6, and the editing levels of these two SNPs in one normal and one ccRCC sample, respectively, were shown in Supplementary Fig. S1A.As shown in Supplementary Fig. S1B and S1C, these SNPs were supported by many sequencing reads in SRR11873716 and SRR11873719, respectively.Additionally, the editing levels of these two sites reached 100% (Supplementary Fig. S1D and S1E), consistent to the fact that these sites were SNPs.
122 M/E sites with significantly different editing levels in ccRCC .We obtained 122 miRNA M/E sites with significantly different editing levels in ccRCC samples (n = 154) compared to normal controls (n = 22) by using Mann-Whitney U-tests (corrected p < 0.05 ) (Supplementary Table S3).The 122 miRNA M/E sites were divided into 7 categories.112 of these 122 sites were 3 ′ -addition events (3 ′ -A, 3 ′ -U, and 3 ′ -Other).In addition, there were 5 A-to-I editing sites, 2 C-to-U editing sites, and the remaining two were 5 ′ end sites and one Pseudo site, i.e., potential false positive site (Fig. 4A).Most of the 122 M/E sites had increased editing levels (77 M/E sites, 63.1%) and 45 M/E sites (36.9%) had decreased editing levels in ccRCC samples when compared to normal controls (Fig. 4B).The categories of M/E sites with increased and decreased editing levels in ccRCC samples were shown in Fig. 4C.In M/E sites with increased editing levels in ccRCC, the 3 ′ -A editing events accounted for the largest portion, while the 3 ′ -U editing events accounted for the largest part in those with decreased editing levels in ccRCC compared to normal controls.
The expression levels of ADAR and APOBEC genes in ccRCC .As shown in Supplementary Fig. S2 and Supplementary Table S7, the expression levels of the ADAR genes that catalyzed A-to-I editing of miRNAs were examined.In the TCGA data, ADAR1 was downregulated in ccRCC compared to normal controls (Supplementary Fig. S2A) which was consistent with the decreased editing levels of hsa-mir-376a-1_49_A_g, hsamir-376a-2_55_A_g, and hsa-mir-376c_48_A_g.However, in the other two cohorts of gene expression profiles, ADAR1 was upregulated, suggesting that more researches were needed to reveal the role of ADAR1 in the editing of miRNAs in ccRCC.We found that ADAR2 was significantly upregulated in ccRCC samples compared to normal controls in the three selected cohorts of gene expression datasets (Supplementary Fig. S2B).Meanwhile, the editing level of hsa-mir-6503_59_A_g was significantly increased in ccRCC samples compared to normal controls (Fig. 4D), suggesting that ADAR2 mediated the editing of hsa-mir-6503 in ccRCC and contributed to its increased editing levels in ccRCC compared to normal controls.ADAR3 was downregulated in one cohort of gene expression profiles, and had no significant change in the other two datasets (Supplementary Fig. S2C).
The expression of the APOBEC genes that mediated C-to-U editing of miRNAs was also examined.As shown in Supplementary Fig. S3 and Supplementary Table S8, the expression levels of most APOBEC family members showed a significantly upward trend in ccRCC tumors, including AICDA, APOBEC3A, APOBEC3B, APOBEC3C, APOBEC3D, APOBEC3F, APOBEC3G and APOBEC3H genes (Supplementary Fig. S3A, D-J, respectively).In our results, the editing level of hsa-mir-29c_59_C_u was significantly increased in ccRCC samples (Fig. 4E).More researches are needed to clarify which APOBEC gene is the most promising enzyme that mediates the C-to-U editing of hsa-mir-29c_59_C_u and other miRNA C-to-U editing sites in ccRCC.

The expression levels of TENT genes in ccRCC . Terminal nucleotidyltransferases (TENTs) modify
RNA at post-transcriptional level, which regulates the stability and activity of RNA 45 .TENT2 and TENT4B proteins catalyze 3 ′ end adenylation of miRNAs 21,46-50 .TUT4 and TUT7 proteins are terminal uridine transferases that belong to the TENT3 subfamily, which have the ability to catalyze the 3 ′ uridylation of miRNAs 23,51-53 .The levels of these TENT genes in ccRCC and normal controls were examined (Supplementary Table S9).In ccRCC, the level of TENT2 was significantly higher (Supplementary Fig. S4A-C), while TENT4B was significantly downregulated (Supplementary Fig. S4A and C).There are more 3 ′ -A sites with increased editing levels in ccRCC than 3 ′ -A sites with decreased editing levels in ccRCC (Fig. 4C), suggesting that TENT2 might be the major enzyme for the 3 ′ -A editing of miRNAs in ccRCC, while the downregulation of TENT4B could be responsible for the decreased 3 ′ -A editing levels of some miRNAs in ccRCC.

Identification of targets for A-to-I and C-to-U edited miRNAs.
Three M/E sites (hsa-mir-376a-1_49_A_g and hsa-mir-376c_48_A_g and hsa-mir-29c_59_C_u) in seed regions of mature miRNAs were selected for subsequent analysis, because these sites had significantly different editing levels in ccRCC and the levels of their corresponding edited miRNAs were also significantly different and in the same directions of changes when compared their editing levels in ccRCC with normal controls.
To decrease false positives and increase the reliabilities of predicted miRNA targets, we used 11 PAR-CLIP sequencing profiles.The principle of PAR-CLIP (Photoactivatable Ribonucleoside-Enhanced Crosslinking and Immunoprecipitation) technology is used to detect the crosslinking of RNA-binding proteins (RBPs) to RNA molecules, followed by immunoprecipitation and sequencing to identify the binding sites 54 .Through PAR-CLIP experiments, the interaction between RISC (including AGO proteins and miRNAs) and mRNA can be identified 54 .The sequences of the miRNA binding sites were obtained through PAR-CLIP experiments (targeting one of the AGO proteins) and then were sequenced 54 .Next, the MiCPAR algorithm 55 was used to identify the targets of original and edited miRNAs by analyzing 11 PAR-CLIP profiles of AGO proteins (see Methods for details).We found that the original and edited hsa-mir-376a-1-3p had 102 common target genes, and 484 new target genes of hsa-mir-376a-1_49g were found (Fig. 5A and Supplementary Tables S4.1-S4.3).The original and edited hsa-mir-376c-3p had 107 common target genes, and hsa-mir-376c_48g had 568 new target genes (Fig. 5B and Supplementary Tables S5.1-S5.3).The original and edited hsa-mir-29c-3p had 107 identical target genes, and hsa-mir-29c_59u had 480 new target genes (Fig. 5C and Supplementary Tables S6.1-S6.3).Next, we compared the new target genes of hsa-mir-376a-1_49g, hsa-mir-376c_48g and hsa-mir-29c_59u to deregulated genes in ccRCC.We found that in the three gene expression datasets selected (GSE151419, GSE126964 and TCGA), 88 and 82 target genes of hsa-mir-376a-1_49g and hsa-mir-376c_48g were commonly significantly upregulated in ccRCC (Fig. 5D,E and Supplementary Tables S4.4,S5.4), respectively.Additionally, 51 target genes of hsa-mir-29c_59u were commonly significantly downregulated in ccRCC (Fig. 5F and Supplementary Table S6.4).

Discussion
We identified 1025 M/E sites with significant editing levels by analyzing 176 sRNA-seq profiles of ccRCC patients (n = 154) and healthy controls (n = 22).The M/E sites at the 3 ′ end accounted for the majority, which was consistent with previous work 37,40,43,[56][57][58][59] .We identified 122 editing events with significantly different editing levels in ccRCC samples compared to healthy control samples.The editing levels of 63.1% and 36.9% of these 122 miRNA editing events were increased and decreased in ccRCC patients, respectively, which may be related to the pathogenesis of ccRCC.
Subsequently, we focused on A-to-I and C-to-U miRNA editing events that occurred in the seed regions and were dysregulated in ccRCC.The editing levels of hsa-mir-376a-1_49g and hsa-mir-376c_48g were significantly lower in ccRCC than those in normal samples.In addition, their levels also showed the same significant downregulation in ccRCC samples compared to normal controls.The editing level of hsa-mir-29c_59_C_u and level of  www.nature.com/scientificreports/hsa-mir-29c_59u showed significant increases in ccRCC compared to normal controls.We compared the targets of original hsa-miR-376a-3p, hsa-miR-376c-3p and hsa-miR-29c-3p with the targets of edited miRNAs, hsa-mir-376a-1_49g, hsa-mir-376c_48g and hsa-mir-29c_59u.We found that hsa-mir-376a-1_49g, hsa-mir-376c_48g and hsa-mir-29c_59u could target 484, 568 and 480 additional novel targets, respectively.C-to-U editing of miRNA was not widely reported.Interestingly, we detected 30 C-to-U editing events, much more than A-to-I editing events identified, in ccRCC.The more prevalence of C-to-U miRNA editing in ccRCC may be related to the high levels of APOBEC gene family in ccRCC (Supplementary Fig. S4).For examples, APOBEC3C, APOBEC3F and APOBEC3G were highly expressed in ccRCC samples compared to normal controls (as shown in Supplementary Figs.S4F, H and I, respectively).The fifth position of hsa-miR-100-5p was specifically edited in Treg 60 .C-to-U edited hsa-miR-100-5p repressed a new target SMAD2, which reduced the fraction of Treg in peripheral blood mononuclear cells 60 .The C-to-U editing of hsa-mir-29c_59_C_u in ccRCC might represent another case of functional C-to-U editing sites in miRNAs, because its editing levels significantly increased in ccRCC and the C-to-U edited hsa-mir-29c-3p potentially repressed RASSF8 in ccRCC.As 11 of these 30 C-to-U editing sites were in seed regions (Fig. 3 and Supplementary Table S2.5), more of these C-to-U editing might be functional in tissue-or time-specific manners.
In our study, BCAT1 and MTHFD2 were upregulated in ccRCC and identified as target genes of hsa-mir-376a-1_49g and hsa-mir-376c_48g in ccRCC, respectively.BCAT1 is overexpressed in many cancers and has been proposed as a marker for predicting cancer prognosis 61 .For example, overexpression of BCAT1 promotes tumor growth in renal clear cell carcinoma 62 , lung cancer 63 , glioma 64 , ovarian cancer 65 , breast cancer 66 , hepatocellular carcinoma 67,68 , myeloid leukemia 69 , gastric cancer 70,71 , endometrial cancer 72 , non-small cell lung cancer 73 and pan-cancer 74 .Moreover, BCAT1 and its metabolites participate in the metabolism of cancer cells through different mechanisms.For example, the PI3K/AKT/mTOR signaling pathway is activated by BCAT1 to promote the proliferation and angiogenesis of gastric cancer cells in vitro 70 .At the same time, in the study of lung adenocarcinoma, BCAT1 promotes lung adenocarcinoma progression by enhancing mitochondrial function and NF-κ B pathway 75 .
Recent reports have shown that MTHFD2 is also highly expressed in many cancers, including renal clear cell carcinoma 76,77 , pancreatic cancer, 78 breast cancer 79 , myeloid leukemia 80 , non-small cell lung cancer 81,82 , hepatocellular carcinoma 83 , colorectal cancer 84 , breast cancer 85,86 , ovarian cancer 87 , bladder cancer 88 , glioblastoma 89 , nasopharyngeal carcinoma 90 , oral squamous cell carcinoma 91 , and pan-cancer 92 .In addition, studies have reported the role of MTHFD2 in tumor immune evasion.MTHFD2 is induced by IFN-γ and promotes basal and IFN-γ -induced PD-L1 expression.Mechanistically, MTHFD2 drives folate circulation to maintain intracellular UDP-GlcNAc and cMYC O-GlcNAcylation, which promotes PD-L1 transcription 93 .Huang et al. found that MTHFD2 promoted breast cancer cell proliferation through AKT signaling pathway, and high level of MTHFD2 reduced the survival rate of breast cancer patients 79 .
In summary, the upregulation of BCAT1 and MTHFD2 in ccRCC may be due to the lower levels of hsa-mir-376a_49g and hsa-mir-376c_48g in ccRCC, respectively, which may contribute to the initiation and/or progression of ccRCC.Reduced A-to-I editing level of miRNAs were also noticed previously 38,39 .The editing levels of hsa-mir-376a-1_49_A_g and/or hsa-mir-376c_48_A_g were also reduced in many other cancers, such as glioblastoma 94 and lung cancer 95 .The increased expression of BCAT1 and MFTHD2 might be due to the reduced editing levels of hsa-mir-376a-1_49_A_g and hsa-mir-376c_48_A_g in many cancers.Furthermore, it has been shown that the higher the editing level of hsa-mir-376c_48_A_g site in ccRCC, the better the prognosis and the longer the survival time of patients 38 .Therefore, reduced editing levels of hsa-mir-376a-1_49_A_g and hsa-mir-376c_48_A_g might represent a general mechanism in the initiation and/or progression of different cancers.
RASSF8 has been reported to be a tumor suppressor gene with lower expression in gastric cancer 96,97 , colorectal cancer 98,99 , cervical cancer 100 , ovarian cancer 101 , melanoma 102 , esophageal squamous cell carcinoma 103 , lung cancer 104,105 and osteosarcoma 106 .In addition, RASSF8 is a potential therapeutic target for the prevention of many cancers 97,100,103,105 .In this study, RASSF8, the target gene of hsa-mir-29c_59u, was downregulated in ccRCC.Therefore, RASSF8 may also play similar tumor suppressor function in ccRCC, and the higher level of hsa-mir-29c_59u in ccRCC may lead to the decreased expression of RASSF8 in ccRCC, which consequently promotes the initiation and/or progression of ccRCC.
As shown in Fig. 5G-I, some genes with many PAR-CLIP reads were identified as targets of hsa-mir-376a-1_49g and hsa-mir-376c_48g, and hsa-mir-29c_59u, respectively.Although there currently are limited evidences about their functions in ccRCC, our results suggest that these genes represent promising directions for revealing the mechanisms of ccRCC in the future.
In summary, we presented a comprehensive view of miRNA editing and/or modification events in ccRCC, which promoted our understanding of miRNA editing and/or modification events in ccRCC.More researches are needed in the future to clarify the functional roles of different miRNA editing patterns in the pathogenesis of ccRCC.Our results also provided new clues for the clinical treatment of ccRCC, such as to promote the editing levels of hsa-mir-376a_49_A_g and hsa-mir-376c_48_A_g and to repress the editing level of hsa-mir-29c_59_C_u.

Methods
Small RNA-Seq data used.We collected 176 small RNA-Seq (sRNA-seq) profiles of ccRCC patients and healthy controls, including 154 samples of cancer tissue and 22 normal kidney tissue samples.These 176 sRNAseq profiles were obtained from previous studies 107,108 , and their accession numbers were shown in Supplementary Table S1.1.The qualities of these sRNA-seq sequence data were evaluated with FastQC (v0.11.9) 109 .

Genome and annotation of miRNAs used.
The human genome sequence used was GRCh38, and downloaded from the UCSC Genome Browser 110 .The pri-miRNA sequences in GFF3 format and mature miRNA annotation files were derived from miRBase (v21) 111 .

Identification of M/E sites with significantly different editing levels in ccRCC .
The Mann-Whitney U-tests were used to analyze the differences between the editing levels of 1025 miRNA M/E sites in ccRCC (n = 154) and normal samples (n = 22).The p-values were corrected with the Benjamini-Hochberg correction method 114 .The M/E sites with corrected p-values smaller than 0.05 were defined as having significantly different editing levels in ccRCC.
DESeq2 120 was used to analyze different expression levels (TPTM) of edited miRNAs between ccRCC and normal control samples.The miRNAs with corrected p-values smaller than 0.05 were defined as having different expression levels in ccRCC and normal control samples.
Identification of targets for original and edited miRNAs.Among the editing sites with significant differences in editing levels between ccRCC tumors and normal tissues, a total of five editing events occurred in seed regions of mature miRNAs.We chose two A-to-I editing sites (hsa-mir-376a-1_49_A_g and hsa-mir-376c_48_A_g) and one C-to-U editing site (hsa-mir-29c_59_C_u), because their editing levels and the levels of their corresponding edited miRNAs had significant differences in ccRCC tumor tissues compared to normal controls.Then, we predicted the targets of original and edited miRNAs using the MiCPAR algorithm 55 .As shown in Supplementary Table S1.2, 11 PAR-CLIP sequences were used in the identification of miRNA targets and were downloaded from the NCBI SRA database.Seven PAR-CLIP profiles of HEK293 cell lines stably expressing FLAG/HA-tagged AGO1, AGO2, AGO3, and AGO4 proteins were reported by 54 .Another study included four PAR-CLIP profiles derived from HEK293 cell lines stably expressing HIS/FLAG/HA-tagged AGO1 and AGO2 proteins 121 .In order to obtain reads of acceptable quality, the raw reads of these 11 PAR-CLIP profiles were filtered to keep reads with sequencing scores of at least 30 for their first 25 nucleotides from 5 ′ ends.The qualified reads were combined and the targets of miRNA were identified by the MiCPAR algorithm.The targets with at least one PAR-CLIP read with T-to-C variation were retained for further analysis.
Analyzing expression of targets of edited miRNAs.In order to understand the expression of targets of edited miRNAs in ccRCC tumor tissues, we identified genes that were significantly differentially expressed in ccRCC samples compared to normal samples.We collected three batches of gene expression profiles of ccRCC and controls (Supplementary Table S1.3), of which two batches were obtained from 107,122 .Another set of gene expression profiles was obtained from TCGA (https://portal.gdc.cancer.gov/).The edgeR (v3.34.1) package 123 was used for differential analysis of target genes.The glmFit and glmLRT functions in edgeR were used to build generalized linear models and to perform likelihood ratio tests, respectively.Genes with corrected p-values smaller than 0.05 were defined as having significantly different expression levels in ccRCC.www.nature.com/scientificreports/new targets of edited miRNAs.For edited miRNAs with increased editing levels in ccRCC, we compared their targets with genes that were downregulated in ccRCC because miRNAs normally repressed their target genes, and vice versa.We also analyzed the numbers of PAR-CLIP reads for the targets of edited miRNAs.We manually examined the functional relevance of some selected targets by reading the literature of the genes in the NCBI Gene database.The targets with more PAR-CLIP reads and functional relevance in ccRCC were preferred in our analysis.KOBAS3.0 124 was used to perform enrichment analysis of GO terms and KEGG pathways for new target genes of edited miRNAs.Significantly enriched GO terms and KEGG pathways were filtered with multiple test corrected p-values smaller than 0.05.

Figure 1 .
Figure 1.The number of significant M/E sites in miRNAs and their categories in the analyzed sRNA libraries.(A) The categories of significant M/E sites in miRNAs.(B) The editing sites and numbers in the central region of mature miRNAs.(C) The distribution of the numbers of pre-miRNAs with different numbers of 3 ′ -, 5 ′ -editing and Central editing sites.

Figure 2 .
Figure 2. The details of 9 identified A-to-I editing sites in miRNAs.(A) The editing levels of the 9 A-to-I editing sites in the 176 selected sRNA profiles.(B) The percentages of nucleotides around of 9 A-to-I sites.(C) The MiRME map of hsa-mir-411 in one of the ccRCC samples selected (SRR11873730).(D) The MiMRE map of hsa-mir-7977 in one of the ccRCC samples selected (ERR4367258).(E) The details of hsa-mir-411_20_A_g in SRR11878730.(F) The details of hsa-mir-7977_6_A_g in ERR4367258.In Part (C) to (D), The three panels from top to bottom show the total number of reads (Tags Per Ten Million sequencing reads, tptm), the number of M/E reads with variations (tptm), and the p-values of M/E events, respectively.In Part (E) to (F), the blue nucleotides are original and edited.And the three numbers on the right indicate the frequency of the read, the length of the read, and the weight of the read at this locus, respectively.

Figure 3 .
Figure 3.The details of 30 identified C-to-U editing sites in miRNAs.(A) The editing levels of the 30 C-to-U sites in the 176 selected sRNA-seq profiles.(B) The percentages of nucleotides around the 30 C-to-U sites.(C) The MiRME map of hsa-mir-126 in one of the ccRCC samples selected (ERR4367258).(D) The MiRME map of hsa-mir-451a in one of the ccRCC samples selected (ERR4367208).(E) The details of hsa-mir-126_58_C_u in ERR4367258.(F) The details of hsa-mir-451a_21_C_u in ERR4367208.In Part (C) to (D), The three panels from top to bottom show the total number of reads (Tags Per Ten Million sequencing reads, tptm), the number of M/E reads with variations (tptm), and the p-values of M/E events, respectively.In Part (E) to (F), the blue nucleotides are original and edited.And the three numbers on the right indicate the frequency of the read, the length of the read, and the weight of the read at this locus, respectively.

Figure 4 .
Figure 4.The 122 miRNA M/E sites that have significantly different editing levels in ccRCC samples when compared to normal samples.(A) The categories of the 122 M/E sites.(B) The change of editing levels of 122 M/E sites in ccRCC samples compared to normal samples.(C) The percentages of different categories of 122 M/E sites with significantly different editing levels in ccRCC samples.(D) Comparison of editing levels of hsa-mir-376a-1_49g, hsa-mir-376c_48g and hsa-mir-6503_59g in ccRCC and normal samples.(E) Comparison of editing levels of hsa-mir-29c_59u in ccRCC and normal samples.(F) Comparison of expression levels of hsa-mir-376a-1_49g, hsa-mir-376c_48g and hsa-mir-6503_59g in ccRCC and normal samples.(G) Comparison of expression levels of hsa-mir-29c_59u in ccRCC and normal samples.(H) The MiRME map of hsa-mir-376a-1 in one of the ccRCC samples (SRR11873721).(I) The MiRME map of hsa-mir-376c in one of the normal samples (SRR11873716).(J) The MiRME map of hsa-mir-29c in one of the normal samples (ERR4367209).(K) The details of hsa-mir-376a-1_49_A_g in SRR11873721.(L) The details of hsa-mir-376c_48_A_g in SRR11873716.(M) The details of hsa-mir-29c_59_C_u in ERR4367209.In Part (D) to (G),

Figure 5 .
Figure 5. Comparisons of targets of edited miRNAs and deregulated genes in clear cell renal cell carcinoma.Venn diagrams were prepared with Venny (https://bioinfogp.cnb.csic.es/tools/venny/).(A-C) The venn diagrams depict the targets of the original and edited hsa-mir-376a-1, hsa-mir-376c and hsa-mir-29c, respectively.(D) The overlap between the new target genes of hsa-mir-376a-1_49g and genes upregulated in ccRCC in three gene expression datasets.(E) The overlap between the new target genes of hsa-mir-376c_48g and genes upregulated in ccRCC in three gene expression datasets.(F) The overlap between the new target genes of hsa-mir-29c_59u and genes downregulated in ccRCC in three gene expression datasets.(G)-(I) The numbers of PAR-CLIP reads for the 88, 82 and 51 overlapped targets of hsa-mir-376a-1_49g, hsa-mir-376c_48g, and hsamir-29c_59u in Part (D) to (F), respectively.
target genes of edited miRNAs.We kept targets with at least one PAR-CLIP read with T-to-C variation.Then, the targets of original and edited miRNAs were compared to obtain the Vol:.(1234567890)Scientific Reports | (2023) 13:15117 | https://doi.org/10.1038/s41598-023-42302-y