Waste not, want not – transcript excess in multicellular eukaryotes

There is growing evidence that mammalian genomes produce thousands of transcripts that do not encode proteins, and this RNA class might even rival the complexity of mRNAs. There is no doubt that a number of these non-protein-coding RNAs have important regulatory functions in the cell. However, do all transcripts have a function or are many of them products of fortuitous transcription with no function? The second scenario is mirrored by numerous alternative-splicing events that lead to truncated proteins. Nevertheless, analogous to ‘superfluous’ genomic DNA, aberrant transcripts or processing products embody evolutionary potential and provide novel RNAs that natural selection can act on.

Introduction

Directed experimental and biocomputational efforts continue to expand the numbers of known small untranslated RNAs (utRNAs), such as small nucleolar RNAs (snoRNAs), micro RNAs (miRNAs), short interfering RNAs (siRNAs) or other tiny RNAs 1, 2. In addition, the cloning of full-length cDNAs has revealed several thousand large utRNAs 3, 4, 5 that resemble mRNAs in that they are polyadenylated, often spliced but lack substantial open reading frames (ORFs). Finally, various studies – most prominently microarray tiling 6, 7, 8, 9, 10, 11, 12, 13 – currently predict a plethora of additional transcripts. Hence, the complexity of utRNAs could easily exceed that of mRNAs.

Are all of these RNAs functional? My prediction is that thousands do indeed have cellular functions, including regulatory roles (e.g. as co-regulatory RNAs, including antisense transcripts 14, 15, 16, 17, 18), but by no means all of them. One way of distinguishing functional and non-functional utRNAs is to examine evolutionary conservation [19]. Recently, the targeted deletion of two large (1.5 and 0.8 million base pairs), non-genic loci was reported in mice that had apparently normal phenotypes [20]. Despite realizing our current limits of analysis, this came as a surprise, because the combined loci contain >1200 conserved sequences shared between mice and humans (>100 bp, >70% ungapped similarity). In addition, it is highly unlikely that neither locus is devoid of transcripts that densely map to other chromosomal regions 6, 7, 8, 10, 12, 13 and, thus, overlap at least some of the conserved regions. On the one hand, conservation, even of transcribed sequences, does not itself imply function. On the other hand, lack of conservation, or mosaic conservation, does not preclude function as we learned from untranslated H19 RNA, inactive X-specific transcripts (Xist), antisense X (inactive)-specific transcript (Tsix) or antisense Igf2r RNA (Air) [14]. This is merely an indication of the task lying ahead in our efforts to assign a function to thousands of utRNAs. However, without this knowledge, we cannot even hope to understand fully the workings of a cell or the molecular bases of many diseases.

Even if a large percentage of observed transcripts were to be explained by experimental shortcomings [13], the presence of thousands of non-functional RNAs (nfRNAs) seems hard to comprehend by Molecular and Cell Biologists. They tend to look at the cells and components of multicellular organisms much in the same way as those of unicellular organisms – as perfectly balanced entities wherein virtually every molecule has a ‘purpose’ or at least, if error-ridden, is neatly disposed of 21, 22. Therefore, it is not surprising that there are tendencies to try to assign a cellular function to all newly discovered RNA transcripts 15, 16, 18, 23. However, recent evidence points to RNAs that exist simply as cellular by-products. In yeast, conditional repression of the SER3 gene occurs by transcriptional interference initiated upstream. The resulting product of the SRG1 gene, the RNA itself, is non-functional [24].