Tobias Fehlmann

Distribution of miRNA expression across human tissues.

We present a human miRNA tissue atlas by determining the abundance of 1997 miRNAs in 61 tissue biopsies of different organs from two individuals collected post-mortem. One thousand three hundred sixty-four miRNAs were discovered in at least one tissue, 143 were present in each tissue. To define the distribution of miRNAs, we utilized a tissue specificity index (TSI). The majority of miRNAs (82.9%) fell in a middle TSI range i.e. were neither specific for single tissues (TSI > 0.85) nor housekeeping miRNAs (TSI < 0.5). Nonetheless, we observed many different miRNAs and miRNA families that were predominantly expressed in certain tissues. Clustering of miRNA abundances revealed that tissues like several areas of the brain clustered together. Considering -3p and -5p mature forms we observed miR-150 with different tissue specificity. Analysis of additional lung and prostate biopsies indicated that inter-organism variability was significantly lower than inter-organ variability. Tissue-specific differences between the miRNA patterns appeared not to be significantly altered by storage as shown for heart and lung tissue. MiRNAs TSI values of human tissues were significantly (P = 10−8) correlated with those of rats; miRNAs that were highly abundant in certain human tissues were likewise abundant in according rat tissues. We implemented a web-based repository enabling scientists to access and browse the data (https://ccb-web.cs.uni-saarland.de/tissueatlas).

Distribution of microRNA biomarker candidates in solid tissues and body fluids.

ABSTRACT Small non-coding RNAs, especially microRNAs, are discussed as promising biomarkers for a substantial number of human pathologies. A broad understanding in which solid tissues, cell types or body fluids a microRNA is expressed helps also to understand and to improve the suitability of miRNAs as non- or minimally-invasive disease markers. We recently reported the Human miRNA Tissue Atlas (http://www.ccb.uni-saarland.de/tissueatlas) containing 105 miRNA profiles of 31 organs from 2 corpses. We subsequently added miRNA profiles measured by others and us using the same array technology as for the first version of the Human miRNA Tissue Atlas. The latter profiles stem from 163 solid organs including lung, prostate and gastric tissue, from 253 whole blood samples and 66 fractioned blood cell isolates, from body fluids including 72 serum samples, 278 plasma samples, 29 urine samples, and 16 saliva samples and from different collection and storage conditions. While most miRNAs are ubiquitous abundant in solid tissues and whole blood, we also identified miRNAs that are rather specific for tissues. Our web-based repository now hosting 982 full miRNomes all of which are measured by the same microarray technology. The knowledge of these variant abundances of miRNAs in solid tissues, in whole blood and in other body fluids is essential to judge the value of miRNAs as biomarker.

miRPathDB: a new dictionary on microRNAs and target pathways

In the last decade, miRNAs and their regulatory mechanisms have been intensively studied and many tools for the analysis of miRNAs and their targets have been developed. We previously presented a dictionary on single miRNAs and their putative target pathways. Since then, the number of miRNAs has tripled and the knowledge on miRNAs and targets has grown substantially. This, along with changes in pathway resources such as KEGG, leads to an improved understanding of miRNAs, their target genes and related pathways. Here, we introduce the miRNA Pathway Dictionary Database (miRPathDB), freely accessible at https://mpd.bioinf.uni-sb.de/. With the database we aim to complement available target pathway web-servers by providing researchers easy access to the information which pathways are regulated by a miRNA, which miRNAs target a pathway and how specific these regulations are. The database contains a large number of miRNAs (2595 human miRNAs), different miRNA target sets (14 773 experimentally validated target genes as well as 19 281 predicted targets genes) and a broad selection of functional biochemical categories (KEGG-, WikiPathways-, BioCarta-, SMPDB-, PID-, Reactome pathways, functional categories from gene ontology (GO), protein families from Pfam and chromosomal locations totaling 12 875 categories). In addition to Homo sapiens, also Mus musculus data are stored and can be compared to human target pathways.

miRTargetLink—miRNAs, Genes and Interaction Networks

Information on miRNA targeting genes is growing rapidly. For high-throughput experiments, but also for targeted analyses of few genes or miRNAs, easy analysis with concise representation of results facilitates the work of life scientists. We developed miRTargetLink, a tool for automating respective analysis procedures that are frequently applied. Input of the web-based solution is either a single gene or single miRNA, but also sets of genes or miRNAs, can be entered. Validated and predicted targets are extracted from databases and an interaction network is presented. Users can select whether predicted targets, experimentally validated targets with strong or weak evidence, or combinations of those are considered. Central genes or miRNAs are highlighted and users can navigate through the network interactively. To discover the most relevant biochemical processes influenced by the target network, gene set analysis and miRNA set analysis are integrated. As a showcase for miRTargetLink, we analyze targets of five cardiac miRNAs. miRTargetLink is freely available without restrictions at www.ccb.uni-saarland.de/mirtargetlink.

cPAS-based sequencing on the BGISEQ-500 to explore small non-coding RNAs

BackgroundWe present the first sequencing data using the combinatorial probe-anchor synthesis (cPAS)-based BGISEQ-500 sequencer. Applying cPAS, we investigated the repertoire of human small non-coding RNAs and compared it to other techniques.ResultsStarting with repeated measurements of different specimens including solid tissues (brain and heart) and blood, we generated a median of 30.1 million reads per sample. 24.1 million mapped to the human genome and 23.3 million to the miRBase. Among six technical replicates of brain samples, we observed a median correlation of 0.98. Comparing BGISEQ-500 to HiSeq, we calculated a correlation of 0.75. The comparability to microarrays was similar for both BGISEQ-500 and HiSeq with the first one showing a correlation of 0.58 and the latter one correlation of 0.6. As for a potential bias in the detected expression distribution in blood cells, 98.6% of HiSeq reads versus 93.1% of BGISEQ-500 reads match to the 10 miRNAs with highest read count. After using miRDeep2 and employing stringent selection criteria for predicting new miRNAs, we detected 74 high-likely candidates in the cPAS sequencing reads prevalent in solid tissues and 36 candidates prevalent in blood.ConclusionsWhile there is apparently no ideal platform for all challenges of miRNome analyses, cPAS shows high technical reproducibility and supplements the hitherto available platforms.

Clinical genetics and outcome of left ventricular non-compaction cardiomyopathy

Aims In this study, we aimed to clinically and genetically characterize LVNC patients and investigate the prevalence of variants in known and novel LVNC disease genes. Introduction Left ventricular non-compaction cardiomyopathy (LVNC) is an increasingly recognized cause of heart failure, arrhythmia, thromboembolism, and sudden cardiac death. We sought here to dissect its genetic causes, phenotypic presentation and outcome. Methods and results In our registry with follow-up of in the median 61 months, we analysed 95 LVNC patients (68 unrelated index patients and 27 affected relatives; definite familial LVNC = 23.5%) by cardiac phenotyping, molecular biomarkers and exome sequencing. Cardiovascular events were significantly more frequent in LVNC patients compared with an age-matched group of patients with non-ischaemic dilated cardiomyopathy (hazard ratio = 2.481, P = 0.002). Stringent genetic classification according to ACMG guidelines revealed that TTN, LMNA, and MYBPC3 are the most prevalent disease genes (13 patients are carrying a pathogenic truncating TTN variant, odds ratio = 40.7, Confidence interval = 21.6-76.6, P < 0.0001, percent spliced in 76-100%). We also identified novel candidate genes for LVNC. For RBM20, we were able to perform detailed familial, molecular and functional studies. We show that the novel variant p.R634L in the RS domain of RBM20 co-segregates with LVNC, leading to titin mis-splicing as revealed by RNA sequencing of heart tissue in mutation carriers, protein analysis, and functional splice-reporter assays. Conclusion Our data demonstrate that the clinical course of symptomatic LVNC can be severe. The identified pathogenic variants and distribution of disease genes-a titin-related pathomechanism is found in every fourth patient-should be considered in genetic counselling of patients. Pathogenic variants in the nuclear proteins Lamin A/C and RBM20 were associated with worse outcome.

Identification of miR-34a-target interactions by a combined network based and experimental approach.

Circulating miRNAs have been associated with numerous human diseases. The lack of understanding the functional roles of blood-born miRNAs limits, however, largely their value as disease marker. In a systems biology analysis we identified miR-34a as strongly associated with pathogenesis. Genome-wide analysis of miRNAs in blood cell fractions highlighted miR-34a as most significantly up-regulated in CD3+ cells of lung cancer patients. By our in silico analysis members of the protein kinase C family (PKC) were indicated as miR-34a target genes. Using a luciferase assay, we confirmed binding of miR-34a-5p to target sequences within the 3′UTRs of five PKC family members. To verify the biological effect, we transfected HEK 293T and Jurkat cells with miR-34a-5p causing reduced endogenous protein levels of PKC isozymes. By combining bioinformatics approaches with experimental validation, we demonstrate that one of the most relevant disease associated miRNAs has the ability to control the expression of a gene family.

Web-based NGS data analysis using miRMaster: a large-scale meta-analysis of human miRNAs

Abstract The analysis of small RNA NGS data together with the discovery of new small RNAs is among the foremost challenges in life science. For the analysis of raw high-throughput sequencing data we implemented the fast, accurate and comprehensive web-based tool miRMaster. Our toolbox provides a wide range of modules for quantification of miRNAs and other non-coding RNAs, discovering new miRNAs, isomiRs, mutations, exogenous RNAs and motifs. Use-cases comprising hundreds of samples are processed in less than 5 h with an accuracy of 99.4%. An integrative analysis of small RNAs from 1836 data sets (20 billion reads) indicated that context-specific miRNAs (e.g. miRNAs present only in one or few different tissues / cell types) still remain to be discovered while broadly expressed miRNAs appear to be largely known. In total, our analysis of known and novel miRNAs indicated nearly 22 000 candidates of precursors with one or two mature forms. Based on these, we designed a custom microarray comprising 11 872 potential mature miRNAs to assess the quality of our prediction. MiRMaster is a convenient-to-use tool for the comprehensive and fast analysis of miRNA NGS data. In addition, our predicted miRNA candidates provided as custom array will allow researchers to perform in depth validation of candidates interesting to them.

Bias in recent miRBase annotations potentially associated with RNA quality issues

Although microRNAs are supposed to be stable in-vivo, degradation processes potentially blur our knowledge on the small oligonucleotides. We set to quantify the effect of degradation on microRNAs in mouse to identify causes for distorted microRNAs patterns. In liver, we found 298, 99 and 8 microRNAs whose expression significantly correlated to RNA integrity, storage time at room temperature and storage time at 4 °C, respectively. Expression levels of 226 microRNAs significantly differed between liver samples with high RNA integrity compared to liver samples with low RNA integrity by more than two-fold. Especially the 157 microRNAs with increased expression in tissue samples with low RNA integrity were most recently added to miRBase. Testing potentially confounding sources, e.g. in-vitro degraded RNA depleted of small RNAs, we detected signals for 350 microRNAs, suggesting cross-hybridization of fragmented RNAs. Therefore, we conclude that especially microRNAs added in the latest miRBase versions might be artefacts due to RNA degradation. The results facilitate differentiation between degradation-resilient microRNAs, degradation-sensitive microRNAs, and likely erroneously annotated microRNAs. The latter were largely identified by NGS but not experimentally validated and can severely bias microRNA biomarker research and impact the value of microRNAs as diagnostic, prognostic or therapeutic tools.

Paired proteomics, transcriptomics and miRNomics in non-small cell lung cancers: known and novel signaling cascades

High-throughput omics analyses are applied to elucidate molecular pathogenic mechanisms in cancer. Given restricted cohort sizes and contrasting large feature sets paired multi-omics analysis supports discovery of true positive deregulated signaling cascades. For lung cancer patients we measured from the same tissue biopsies proteomic- (6,183 proteins), transcriptomic- (34,687 genes) and miRNomic data (2,549 miRNAs). To minimize inter-individual variations case and control lung biopsies have been gathered from the same individuals. Considering single omics entities, 15 of 2,549 miRNAs (0.6%), 752 of 34,687 genes (2.2%) and 141 of 6,183 proteins (2.3%) were significantly deregulated. Multivariate analysis also revealed that effects in miRNA were smaller compared to genes and proteins indicating that expression changes of miRNAs might also have limited impact of pathogenicity. However, a new algorithm for modeling the complex mutual interactions of miRNAs and their target genes facilitated precise prediction of deregulation in cancer genes (92.3% accuracy, p=0.007). Lastly, deregulation of genes in cancer matched deregulation of proteins coded by the genes in 80% of cases. The resulting interaction network, which is based on quantitative analysis of the abundance of miRNAs, mRNAs and proteins each taken from the same lung cancer tissue and from the same autologous normal lung tissue confirms molecular pathological changes and further contributes to the discovery of altered signaling cascades in lung cancer.