Does Selection against Transcriptional Interference Shape Retroelement-Free Regions in Mammalian Genomes?

Willerslev E (2008) Does Selection against Transcriptional Interference Shape Retroelement-Free Regions in Mammalian Genomes? PLoS ONE 3(11): e3760. doi:10.1371/journal.pone.0003760 Does Selection against Transcriptional Interference Shape Retroelement-Free Regions in Mammalian Genomes? Tobias Mourier 0 Eske Willerslev 0 Rodolfo Aramayo, Texas A&M University, United States of America 0 Ancient DNA and Evolution Group, Department of Biology, University of Copenhagen , Copenhagen , Denmark Background: Eukaryotic genomes are scattered with retroelements that proliferate through retrotransposition. Although retroelements make up around 40 percent of the human genome, large regions are found to be completely devoid of retroelements. This has been hypothesised to be a result of genomic regions being intolerant to insertions of retroelements. The inadvertent transcriptional activity of retroelements may affect neighbouring genes, which in turn could be detrimental to an organism. We speculate that such retroelement transcription, or transcriptional interference, is a contributing factor in generating and maintaining retroelement-free regions in the human genome. Methodology/Principal Findings: Based on the known transcriptional properties of retroelements, we expect long interspersed elements (LINEs) to be able to display a high degree of transcriptional interference. In contrast, we expect short interspersed elements (SINEs) to display very low levels of transcriptional interference. We find that genomic regions devoid of long interspersed elements (LINEs) are enriched for protein-coding genes, but that this is not the case for regions devoid of short interspersed elements (SINEs). This is expected if genes are subject to selection against transcriptional interference. We do not find microRNAs to be associated with genomic regions devoid of either SINEs or LINEs. We further observe an increased relative activity of genes overlapping LINE-free regions during early embryogenesis, where activity of LINEs has been identified previously. Conclusions/Significance: Our observations are consistent with the notion that selection against transcriptional interference has contributed to the maintenance and/or generation of retroelement-free regions in the human genome. - Transposable elements are genetic elements that are capable of proliferating withinand even betweengenomes. The elements can be broadly divided into two classes [1]: Class I elements transpose via an RNA intermediate that is reverse transcribed to DNA. We henceforth refer to class I elements as retroelements. Class II elements transpose via a DNA intermediate. With a few recorded exceptions (e.g. refs [2,3]) retroelements are found in all eukaryotic genomes examined, and nearly half of the human genome sequence can be attributed to the activity of retroelements [4]. Recently, Simons and colleagues identified almost 1000 regions in the genomes of human and mouse of at least 10 kilo base pairs (kbp) in size with no transposable elements [5]. Such regions termed TFRs for Transposon-Free Regionswere found to be conserved among other mammals, and associated with microRNAs and genes encoding transcription factors [5,6]. The authors hypothesized that the TFRs encode regions of essential regulatory information that are intolerant to the insertion of transposable elements. The hypothesised selective disadvantage of transposable elements may in many cases be a result of disruption of the informational content of the sequence in which the transposable element is inserted. Yet, the transcriptional activity of transposable elements could be an additional contributor to the deleterious effects of transposable elements, which are presumably selected against in TFRs. This implies that retroelements are not just avoided in TFRs due to the insertion per se, but also to minimize spurious transcription from retroelements. I.e. it is not necessarily the insertion of a sequence that has a deleterious effect, but rather the subsequent transcriptional activity from the inserted sequence. Retroelements contain promoters and transcription factor binding sites necessary for their own transcription. Occasionally, the transcription may continue into adjacent regions. If these adjacent regions encode genes, the transcription may potentially result in transcripts containing both transposable element sequence and gene sequence [7,8], or for example, repress endogenous transcription of the neighbouring gene by promoter competition [9]. Transcriptional interference may potentially occur at different stages of transcription, of which some are experimentally verified and others are purely speculative (see [10] and references therein). Retroelements display a great divergence in transcriptional capacity and activity. Short interspersed elements (SINEs) contain a weak internal polymerase III promoter [11], usually not capable of initiating transcription by itself [12]. Further, the polymerase III generates only shorter transcripts. In contrast, long interspersed elements (LINEs) and Long terminal repeat (LTR) elements harbour polymerase II transcription start sites that are capable of transcribing into adjacent genomic regions [13,14]. LINEs even contain an additional promoter situated in the antisense orientation, which is known to transcribe neighbouring genes [15,16]. The difference in transcriptional features between different transposable elements predicts that the elements will differ in their capabilities in transcriptional interference of neighbouring genes. Firstly, polymerase II transcribed elements will be able to transcribe into adjacent genes, which is not expected for polymerase III transcribed elements. Secondly, as protein-coding genes and presumable microRNAs [17] are transcribed by polymerase II, promoter competition will exclusively be expected from transposable elements transcribed by this polymerase. Thirdly, any physical interaction between transcriptional complexes is expected to be most prominent from polymerase II transcribed elements, simply because these transcriptional complexes will move further along the genome. Consequently, the impact of transcriptional interference should be highest for LINEs and LTR elements, and we are thus able to test the hypothesis that transcriptional interference is contributing to the existence and maintenance of TFRs: Protein-coding genes and RNA genes that are sensitive to the deleterious effects of transcriptional interference should be enriched in genomic regions devoid of polymerase II transcribed transposable elements, whereas this should not be the case for regions devoid of polymerase III transcribed transposable elements. Which genes are then susceptible to the deleterious effects of transcriptional interference? Two conditions must be fulfilled. First, the precise regulation of the genes must be crucial to the organism, and second, the space and time (i.e. developmental stage (...truncated)