Workshop Report 32

Cause or Consequence? Links between changes in ncRNA expression and pathological phenotype

Jörg Hackermüller, Young Investigators Group Bioinformatics and Transcriptomics, Department of Molecular Systems Biology, Helmholtz Centre for Environmental Research – UFZ, Leipzig, Germany

Considering that mammalian cells are capable of producing a plethora of ncRNAs, the question arises to which extent non-coding transcription is functionally important. Are changes in ncRNA expression causal for diseased or adverse states or a consequence thereof? Even if the latter is the case, ncRNAs might still form a valuable pool for biomarkers of adversity or disease. However, if ncRNAs play in part causal roles, do we need to consider ncRNAs in mode-of-action (MoA) frameworks or AOPs, and, if so, is it already possible to define MoAs or AOPs where ncRNAs participate in key events? The example of interleukin 6 (IL-6) and signal transducer and activator of transcription 3 (STAT3)-controlled ncRNAs served to further elucidate these questions. Specific multiple myeloma (B cell malignancy) cells depend strictly on IL-6 and become apoptotic when IL-6 is not present. However, this cellular phenotype cannot be explained by an IL-6-enabled differential expression of protein-coding genes (Brocke-Heidrich et al, 2004). Instead, the miRNA miR-21 appears responsible for the anti-apoptotic effect of IL-6 and indeed shows strong differential expression in B cell malignancy cells (Löffler et al, 2007). Further, miR-21 has been observed to target the tumour suppressor genes ANP32A, PDCD4 and SMARCA4 (Schramedei et al, 2011). IL-6 and STAT3 have been observed to control a plethora of lncRNAs and to regulate a set of macroRNAs (including STAiRs; STAT3-induced RNAs) that are in part specific for multiple myeloma (Hackermüller et al, 2014). Some STAiRs seem to be tightly coupled with IL6 signalling via STAT3 and inherent components of this pathway.

Even though knowledge on such specific interactions is beginning to evolve, it can be difficult to ascertain the causality of the expression of a given ncRNA for several reasons: Its effects on the regulation of gene expression may depend on a given context. With the exception of some small RNA classes, limited conservation at the primary sequence level complicates tracing ncRNAs between model animals. Finally, ncRNAs have been found to interfere with pathways at multiple levels – with strong consequences on the entire pathway but little effect of individual interactions (e.g. Boll et al, 2013). Nevertheless, increasingly, the role of lncRNAs in toxicity testing is being addressed, such as the effects of chemicals on lncRNA H19 and other imprinted genes as well as Hox gene transcript antisense RNA (HOTAIR) (Croce, 2010; Bhan et al, 2014). These findings may form the basis for an AOP. Non-coding loci are also of increasing interest in environmental epidemiology. Deep methylome sequencing, i.e. an analysis of the methylation status of the genome, in a mother-child cohort study identified numerous differentially methylated regions of the DNA (DMRs) in children that are affected by their mothers’ smoking habits. The majority of DMRs was associated with noncoding targets and differential methylation was in part found to persist over years (Bauer et al, 2016).

In conclusion, to a larger part, the differential expression of ncRNAs may be a consequence of disease or adverse effect, but many short ncRNAs do play a causal role in disease. Also for a growing number of lncRNAs, changes in either expression or mutation have been found causal for disease. Even though only few lncRNAs have so far been associated with pathways of toxicity and knowledge on their role in the evolvement of apical effects is still limited, lncRNAs, and not only miRNAs, should be considered in AOP or MoA analyses.