Abstract
Background/Aim: Renal cell carcinoma is one of the most common types of cancer worldwide. Understanding tumor pathogenesis is important in developing better treatment. Micro RNAs (miRNAs) are key players in controlling cancer behavior. Transcription factors (TFs) are potentially responsible for controlling miRNA expression and dysregulation in kidney cancer. The objective of this study was to better understand the TF–miRNA axis of interaction. Materials and Methods: We utilized publicly available databases to investigate miRNA–TF interactions, including ChipBase database for TFs that binds to the promoters of miRNAs which are dysregulated in renal cell carcinoma. Renal cancer-specific TFs were extracted from the list using the GENT Database. We assessed the prognostic significance of these TFs using cBioPortal. Results: We identified TFs which bind to miRNA promoters, including hepatocyte nuclear factor-4 alpha (HNF-4α), E2F transcription factor 4 (E2F4), signal transducer and activator of transcription 1 (STAT1), Sp1 transcription factor (SP1), GATA binding protein 6 (GATA6), and nuclear factor kappa B (NFκB). These TFs were positively correlated with their targeted miRNAs, including miR-200c, miR-15a, miR-146b, miR-155, and miR-223. We recognized unique patterns of interactions, including a divergent effect in which multiple miRNAs are simultaneously affected by the same TF. Conclusion: Our results show that miRNA–TF interaction is complex. Expression levels of these TFs were found to correlate with renal carcinoma prognosis and have potential utility as biomarkers for aggressive tumor behavior. Targeting these TFs may result in modulating the expression of their target genes and miRNAs, with subsequent therapeutic implications.
Renal cell carcinoma (RCC) is an aggressive disease with an overall 5-year relative survival rate of 75%. Survival drops significantly to 69% with regional spread of the disease and to only 12% in RCC with distant metastasis (1, 2).
It is well understood that the loss of the Von Hippel-Lindau tumor suppressor (VHL) gene contributes to the pathogenesis of RCC. Normally, VHL binds to hypoxia-inducible factors (HIFs) and induces their degradation in the proteasome via poly-ubiquitinylation. However, under hypoxic conditions or if VHL has been inactivated, high levels of HIFs accumulate and promote transcription of genes involved in angiogenesis, cell growth and proliferation. These genes include erythropoietin, vascular endothelial growth factor, platelet-derived growth factor, and transforming growth factor. Apart from the documented contribution of the VHL–hypoxia pathway to disease pathogenesis, other molecular factors that contribute to RCC pathogenesis have yet to be identified (3). There are very few therapeutic options for advanced RCC. To develop more effective targeted therapy, it is essential to understand the mechanisms that contribute to progression of disease.
An important level of eukaryotic molecular signaling involves short non-coding RNA sequences known as microRNAs (miRNAs, miRs). In an intricate manner, miRNAs are able to regulate the expression of their targets (4). A single gene can be controlled by numerous miRNAs, and a single miRNA can control numerous genes (5). Apoptosis avoidance, cell proliferation and migration have all been shown to be influenced by miRNAs, all of which play an important role in renal carcinogenesis (6, 7).
There is a wealth of literature showing the differential expression of miRNAs in RCC and their potential involvement in disease pathogenesis in addition to their potential utility as prognostic markers (8-13).
To identify RCC biomarkers and potential therapeutic targets, it is vital to understand the mechanisms that regulate miRNA expression and function. DROSHA and DICER processing are important regulatory steps in miRNA maturation. Through methylation, uridylation and adenylation, alteration of miRNA is achieved. In addition, non-canonical pathways can also be utilized (14).
To control the rate of transcription, transcription factors (TFs) bind to specific DNA sequences and regulate the genetic information of DNA being transcribed to mRNA. Recently, it has been shown that there is a TF–miRNA regulatory network that controls gene expression (15). Investigating this network in RCC can assist in better knowledge of how this contributes to renal carcinogenesis. E2F Transcription factor 4 (E2F4), nuclear factor kappa B (NFκB) and caudal type homeobox 2 CDX2 are among several TFs that were reported to be involved in RCC. The exact mechanism of this regulation remains unknown (16).
A wide variety of databases and literature have recently described gene regulatory relationships in the hope of developing a genome-wide regulatory network database to bring together regulatory relationships that are found in many different databases.
In this study, we analyzed a number of publicly available databases using different bioinformatics tools to elucidate TF–miRNA regulatory relationships and dissected their potential role in disease pathogenesis.
Materials and Methods
Differential miRNA expression in RCC was obtained from previously published data. For the purpose of this study, we compared the results of seven published articles (13, 17-22) and used miRNAs differentially expressed between RCC and normal kidney tissues as reported in at least three of these publications.
We interrogated the ChIPBase database (https://rna.sysu.edu.cn/chipbase/) which is an integrated resource and platform for TF binding maps, expression profiles and transcriptional regulation of protein-coding and non-coding RNAs using ChIP-Seq data. We searched for TFs located upstream of miRNA transcription start sites (up to 5 kb).
Tissue specificity of the TFs was obtained from GeneHub-GEPIS (https://bio.tools/genehub-gepis). The interactions between TFs, miRNAs, and miRNA target genes were obtained from the ChIPBase network version 1.1 (1 November 2012). Target prediction analysis of the miRNAs dysregulated in RCC was performed using a combination of algorithms, as previously published (12). Pathway analysis of miRNA targets was obtained from DIANA miRPath v.2.0 (https://dianalab.ece.uth.gr/). Survival data (overall survival) were obtained from cBioPortal (https://www.cbioportal.org/).
Results
TFs control miRNA expression in RCC. Top significantly dysregulated miRNAs in RCC are shown in Table I. To better understand the mechanisms controlling miRNA dysregulation in RCC, we hypothesized that TFs are key factors in controlling miRNA alteration during renal carcinogenesis. In order to identify TF-binding sites located upstream of intronic and intergenic miRNAs, we cross-examined the ChIPBase database. A total of 101 TFs that can potentially control RCC-related miRNAs were identified.
Partial list of most significantly differentially expressed microRNAs in renal cell carcinoma compared to normal renal tissue.
Multifarious interactions between TFs and miRNAs. Next, we were able to identify the presence of a complex relationship between TFs and miRNAs. A total of 101 TFs were identified with a divergent pattern of control on miRNAs where the same TF was able to control multiple miRNAs concurrently (suppression or activation). Clinically significant RCC-related TFs are shown in Table II. At the same time, a convergent pattern was also identified where the same miRNA was under the control of multiple TFs, as shown in Figure 1. Representative examples of TFs are shown in Table III.
Partial list of transcription factors that target the promotors of miRNAs that are differentially expressed in renal cell carcinoma.
The complex interaction between transcription factors, microRNAs (miRNAs) and their target genes. A representative miRNA, miR-200c, is shown at the center of the image. This miRNA is under convergent regulation by multiple transcription factors.
Representative miRNA clusters that share overlapping miRNA promotor regions.
An additional interesting observation was the presence of dual control, whereby the same TF was able to control a miRNA and concurrently control the target of this same miRNA, representing a more complex control pattern. For example, miR-200c is regulated by a number of TFs such as GATA binding protein 2 (GATA2), signal transducer and activator of transcription 1 (STAT1), RAD21 cohesin complex component (RAD21) and nuclear factor kappa B subunit (NFκB). miR-200c controls the expression of many target genes such as phosphatase and tensin homolog (PTEN), spindlin 1 (SPIN1), cyclin-dependent kinase inhibitor 1B (CDKN1B) and golgin A1 (GOLGA1). A few of these genes are directly under the control of the same TFs.
We also found that the number of DNA-binding sites for the same TF may vary from one miRNA promoter to another, indicating a potential ‘quantitative’ control of miRNAs by TFs. For instance, hepatocyte nuclear factor 4 alpha (HNF4A) has nine TF-binding sites in the promotor of miR-21 and seven in miR-451, while it has only one in miR-34a.
Furthermore, for the same miRNA, there was variation in the distances of TF-binding sites from the transcription start site between TFs. There was also variation of the location of the same TF relative to the promoters of different miRNAs. Recent studies implied that the smaller the distance from the TF-binding site to the transcription start site, the more likely the association is one of activation as opposed to repression, leading to potential functional implications (23).
Locus control of miRNA families. It has been documented that multiple miRNAs can share the same promoter region that contains TF clusters (5). We observed the same for RCC. Our results indicate that multiple miRNAs may be regulated by the same TF cluster, demonstrating a locus control phenomenon, whereby a TF cluster will concurrently affect the expression of a group of miRNAs with the same frequency. For example, cluster 3 of TFs on chromosome 21 (21C3), which includes HNF4A, STAT1, NFκB, CDX2, GATA6 and Sp1 transcription factor (SP1), is predicted to target a number of miRNAs, as shown in Table III.
TF–miRNA regulatory networks. Gene-regulatory networks are formed by network motifs, which are repeated patterns of network structures. These have the ability to function as autoregulation and feed-forward loops (24). Our results show three different patterns of TF–miRNA interaction networks:
i) Feedback loop: The TF affects a miRNA and at the same time is also a target of this miRNA. For instance, SP1 controls and is also controlled by miR-130b (Figure 2A).
Figure 2.The different types of transcription factor–microRNA (miRNA) interactions. A: Feedback loop in which the transcription factor affects the miRNA whilst being itself a target of this miRNA. B: An example of a feed-forward loop, in which a transcription factor affects a miRNA and its target gene. C: Upstream cross-talk, where multiple transcription factors located in the promoter of the same miRNA can affect each other and also affect other transcription factors that control other miRNAs which have the same target as the miRNA. E2F4: E2F transcription factor 4; GATA6: GATA binding protein 6; SP1: Sp1 transcription factor; STAT1: signal transducer and activator of transcription 1; YY1: YY1 transcription factor; UNG2: uracil DNA glycosylase.
ii) Feedforward loop: The TF directly regulates both miRNAs and the targets of these miRNAs. For instance, NFκB controls miR-200c and its target gene A-Kinase Anchoring Protein 11(AKAP11) (Figure 2B).
iii) Up-stream cross-talk: Recently, cross-talk between TFs has been recognized (24, 25). Our results indicate potential interaction between TFs in the promoter of the same miRNA. For example, GATA6, YY1 and E2F4 are both located in the promoter region of the same miRNA 34a and affect each other as well as affecting another TF (STAT1) that controls miR-15b, which shares the same target gene (uracil DNA glycosylase, UNG2) (Figure 2C).
TFs can have multi-level control over miRNAs. Another interesting phenomenon discovered during analysis is that TFs utilize a multi-level control of miRNA functionality. TFs can control miRNA expression concurrently at different levels (Figure 3). We discovered that some TFs may: i) Facilitate miRNA maturation by binding to promoters of the RNA-induced silencing complex and other molecules involved in miRNA processing; ii) bind to a miRNA promotor resulting in underexpression or overexpression of miRNA (26, 27); iii) control the expression of the same predicted miRNA targets. Fascinatingly, both TFs and the mature miRNAs that are direct targets of the TFs may simultaneously control RNA-induced silencing complex molecules.
Transcription factors exert multi-step control over microRNA (miRNA) functionality. A schematic showing control of miRNA expression through transcription factors that bind to the miRNA promoter leading to under/overexpression of the miRNA. In the meantime, these transcription factors bind to promoters of the RNA-induced silencing complex (RISC) including DICER1 and other molecules involved in miRNA processing, thus facilitating miRNA processing and maturation. In certain situations, the transcription factors also control the target genes of the miRNAs. AGO2: Argonaute RISC catalytic component 2; AKAP11: A-kinase-anchoring protein 11; E2F4: E2F transcription factor 4; ETS1: ETS protooncogene 1 transcription factor; c-JUN: JUN proto-oncogene, AP-1 transcription factor subunit; MLL3: lysine methyltransferase 2C (KMT2C); SOX4: SRY-box transcription factor 4; VEGF: vascular endothelial growth factor A.
TF–miRNA–gene interactions can predict survival in RCC. To better understand the clinical implications of the interaction between miRNA and TFs, we assessed the influence of TF–miRNA–target interaction networks on the survival of a patient cohort using a database from cBioPortal. Our results showed that alterations in a selected group of miRNAs and TFs correlated with a significantly worse prognosis (overall survival) (Figure 4). The same trend was found for disease-free survival, but the results were not statistically significant (data not shown).
The transcription factor–miRNA–gene interactions network can predict prognosis in patients with kidney cancer. Kaplan–Meier survival curve analysis with log-rank test showing that cases with alteration in selected groups of miRNA and transcription factors had significantly worse prognosis (overall survival) compared to those without alterations.
Discussion
We have shown that TFs have a particular level of transcriptional regulation of miRNA expression. However, the association between the two is more intricate than originally understood. Through publicly available databases, we identified significant complex interactions between TFs and miRNAs which have potentially important roles in the pathogenesis of RCC. ChIPBase data experimental results were the basis of our analysis, and our discoveries were similarly validated. Nonetheless, it is important to note that the interaction between miRNAs and TFs can be tissue-specific and experimental validation is required due to the possibility of variation between disease conditions. A synergetic action between miRNAs can be revealed by the presence of locus control on a cluster of miRNAs. This synergy may result in a greater effect on a biological process, which may have an impact on drug resistance mechanisms. The clinical utility of our study is promising. We pursued this study to understand the disease course for RCC. Our major objective is to identify potential mechanisms to halt the growth of tumors through direct targeted therapy. This can be achieved by in-depth knowledge of the TF–miRNA– target network of pathways. Understanding biology also implies looking for complex interactions which can be used to develop multi-parametric biomarkers for RCC, which has the potential to provide prognostic information.
With the advent of publicly available databases, the world of research has been transformed (28). However, it is important to be cautious when analyzing public databases as there are many factors which can affect the utility of their data. Contamination, technical experimental variance, and suboptimal experimental protocols are among a few that should be considered. Nonetheless, our study shows great potential for utilizing publicly available databases, with its advantages significantly outweighing its possible limitations, including the need for experimental validation. Future experimental research can be assisted by in-silico analysis by supplying preliminary data for targeted experimental validation. This would conserve resources (29).
An additional concern during analysis of large-scale genomic data is that of false-positive results (30). During cancer progression, one must also consider tissue specificity changeability. Lastly, because there are other tissue- and cell-specific factors, there is no guarantee that a predicted interaction will also occur in vivo.
In summary, we have shown that TFs represent an important mechanism for controlling miRNAs which contribute to the pathogenesis of RCC. With feed-forward and feedback mechanisms, the relationship between miRNAs and TFs is complex. The activation or repression of miRNAs by TFs can be used as potential therapeutic targets.
Acknowledgements
This article was supported by the Li Ka Shing Institute and the University of Toronto. This project was supported by kind funding for the principal investigator, Dr. Krizova, from St. Michael’s Hospital (Unity Health).
Footnotes
Authors’ Contributions
Study design: PY, RI, MP, AK. Data retrieval: PY, CB, ZK. Informatics analysis: PY, CB, ZK. Critical analysis: all Authors. Manuscript drafting and revisions: all Authors.
Conflicts of Interest
All Authors declare no conflicts of interest.
- Received December 25, 2021.
- Revision received March 17, 2022.
- Accepted March 25, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.