Exposure effects? Relevance of dose level, time, no effect level, adverse effect level
Nigel Gooderham, Computational and Systems Medicine, Department of Surgery and Cancer, Imperial College London, UK
MiRNAs are an abundant class of genes in mammals that may control gene expression through translational repression and by induction of mRNA degradation. Since miRNAs regulate development, cell proliferation, apoptosis, and differentiation, they play a central role in processes that are potentially important in toxicology. Specifically, miRNAs have been shown to interact with cellular pathways that are relevant for carcinogenesis (Calin and Croce, 2006). Tumour-suppressor miRNAs may negatively regulate protein coding oncogenes, whereas oncogenic miRNAs may repress tumour suppressor genes. MiRNAs may also alter the epigenomics landscape by reprogramming a cell’s epigenome (Moazed, 2009). For instance, miR-29 has been observed to inhibit the expression of DNA (cytosine-5-)-methyltransferase 3 (Fabbri et al, 2007), and miR-101 regulates the histone methyltransferase EZH2 (enhancer of zester homolog-2) (Floris et al, 2015). Further, miRNA combinations may cooperate to regulate multiple proteins within cancer-important pathways.
Currently, genotoxic carcinogens can be identified reliably and quickly using a weight-of-evidence approach that takes into account structural information, in vitro assays, and in vivo DNA damage assays. By contrast, there are no short-term tests having received regulatory acceptance that allow predicting non-genotoxic carcinogenicity. To investigate if changes in miRNA expression may constitute early indicators of substance-induced carcinogenicity with adequate sensitivity and specificity and provide information on mechanisms of toxicity, a spectrum of liver carcinogens was investigated in 90-day oral toxicity studies in Fisher rats. These substances were applied at carcinogenic doses, whereas additional non-carcinogenic substances were applied at the respective maximum tolerated dose. In none of the test groups did tumours become evident by the end of the exposure period. Total RNA was extracted from the liver, labelled and profiled using the Agilent miRNA microarray platform (Agilent Technologies, USA). MiRNA expression was normalised to the 75th percentile, and miRNAs that were not detected in at least 50% of the samples of any test group were excluded from the evaluation. Thereby, 21 miRNAs were identified as differentially expressed, and hierarchical clustering revealed specific patterns of miRNA expression (Koufaris et al, 2012). These miRNAs appeared to regulate pathways that are frequently disrupted during chemical carcinogenesis or implicated in the progression or suppression of carcinogenesis; these same miRNAs had previously been found to be dysregulated in tissue-specific tumours. Bioinformatic analysis indicated that specific pathways (such as phosphoinositide 3-kinase or epidermal growth factor) were targeted and over-represented in the analysis. This points to the need to assess the biological plausibility of the predicted miRNA-regulated pathways in respect to cancer development.
Further, a 14-day rat oral toxicity study was conducted to investigate whether the tumour promoter phenobarbital elicits miRNA-related effects in a temporal and dose-dependent manner. Again, total RNA was extracted from the liver, labelled and profiled using the Agilent microRNA microarray platform. While there were no obvious statistically significant microarray responses during the first 7 days of treatment, within 14 days, clustering could be observed and a distinct dose-related time-dependent effect on the miRNA expression profiles could be shown (Koufaris et al, 2013). In summary, the mentioned 90-day and 14-day studies suggest that both genotoxic and epigenetic carcinogens can dysregulate miRNA expression to produce a ‘fingerprint’ that can be detected long before tumours develop in the treated animals. This miRNA ‘fingerprint’ appears to be compound, dose and temporally regulated, and details of the miRNA ‘fingerprint’ may offer insight into potential MoAs. Consequently, the concept that miRNAs are biomarkers of toxicity such as chemical-induced carcinogenesis is highly attractive and offers the potential of translational biomarkers of progression of the disease (Koufaris et al, 2013; Gooderham and Koufaris, 2014).
How was differential expression of miRNAs quantified? Which specific statistically measurable differences in miRNA expression were recorded? - Between 3-5 animals per treatment group were used. RNA quality was determined using an Agilent Bioanalyser, and expression was assessed using the Agilent miRNA microarray platform and microarray scanner. The data were collected using the Agilent Feature Extraction software, for miRNA hybridisation signals a threshold of 1 was set, and the signals were log-2 transformed and normalised to the 75th percentile using the GeneSpring GX software. One-way ANOVA with Benjamini-Hochberg multiple testing correction followed by Tukey’s test was used to identify dysregulated miRNAs. Targeted quantitative real-time polymerase chain reaction (qRT-PCR) was used to confirm differential miRNA expression between control and test groups.
Are there miRNA reference databases against which to anchor data, such as the ones collected during the mentioned studies? – At the Imperial College London, a number of bioinformatic packages and databases to interpret the miRNA data are applied. These include packages to predict miRNA targets (miRWALK, DIANA‑mt, miRanda, miRDB, RNAhybrid, PICTAR, PITA, RNA22 and Targetscan), the KEGG database and PANTHER (both open source) to pathway map predicted targets and Gene Expression Omnibus for archiving miRNA expression data.
Could similar investigations also be performed for other organs? – Yes, in addition to the liver, similar analyses were performed on the kidneys, intestinal tissue, mammary tissue, cardiac tissue, biological fluids and numerous cell lines and culture media. In the presented animal studies, blood samples were also analysed. However, the sensitivity of detecting miRNA in the blood was lower than in, e.g. the liver tissue.