Final Report Summary - TOAFOLNR (Exploring the dark matter of the human brain transcriptome: The origin and function of long non-coding RNAs)
Project context and objectives
The emergence of multicellular eukaryotes required complex gene regulatory networks to evolve so as to allow the spatially and temporally regulated transcription of thousands of genic loci. Transcripts lacking open-reading frames, non-coding RNAs, have been proposed to be an important yet largely unexplored class of transcriptional regulatory elements. In my fellowship I have used a powerful combination of in silico and in vitro approaches to characterise the transcriptional roles of non-coding RNAs, in particular those that are longer than 200 nucleotides and are intergenic (long intergenic noncoding RNAs, lincRNAs). I have addressed the following two questions: 1. What are the contributions of pseudogene derived-noncoding RNAs to post-transcriptional regulation? 2. How does transcriptional regulation by lincRNAs contribute to phenotypic variation?
Work performed and main results
Contribution of pseudogene-derived noncoding RNAs to post-transcriptional regulation
Negative regulation of transcript levels by small (22-25 nucleotides) non-coding RNAs (miRNAs) involves the recognition and binding of mature miRNA to miRNA response elements (MREs) often located in the 3' untranslated regions of target mRNAs. A miRNA can regulate large numbers of transcripts and target recognition is thought to lead to decreased miRNA concentration. It was recently demonstrated that competition for miRNA binding between transcripts sharing MREs (competitive endogenous RNAs, ceRNAs) can alter their abundance. This new layer of post-transcriptional regulation involves miRNA mediated crosstalk between numerous protein-coding and noncoding transcripts.
I have identified and am currently completing the characterisation of the retrotransposed gene copy (retrocopy) of a protein-coding member of a large multiprotein complex involved in transcriptional regulation of a number of loci. This retrocopy shares MREs with over 90 % of genes whose products form this protein complex. Importantly, changes in its expression influence the levels of many, but not all members, of this ceRNA network. Increased or decreased expression of miRNAs is sufficient to mimic some of the post-transcriptional effects of the retrocopy, indicating that indeed its post-transcriptional function is miRNA-mediated. By analysing the relative number of shared MREs and expression levels of different ceRNAs in this network I aim to shed light on the properties of transcripts that efficiently crosstalk. Furthermore, the multiprotein complex I am studying is involved in the regulation of several human disease genes. The detailed analysis of this retrocopy's contribution to the regulation of gene expression will provide insights into its contribution to abnormal human phenotypes.
An individual miRNA can contribute to the post-transcriptional regulation of ten-to-hundreds of non-homologous genes. This raised the question of whether the miRNA-mediated roles of an mRNA are indispensable per se. In particular, if the miRNA-mediated crosstalk between functionally related transcripts contributes to stabilising relative transcript levels this would make an mRNA's transcriptional regulatory functions less dispensable. To address this question and separate the protein-coding and miRNA decoy functions of transcripts, I considered the conservation of post-transcriptional roles by unitary pseudogenes. Pseudogenes exhibit significant sequence similarity to known protein-coding genes in the same or related genomes but contain mutations such as premature stop-codons or frame-shifting nucleotide insertion deletions that preclude their translation into functional proteins. Most such pseudogenes arose by duplication followed by pseudogenisation. Relatively few are 'unitary pseudogenes', those that derive not from duplication but from the lineage-specific acquisition of disrupting mutations in the coding sequences.
Investigating transcribed unitary pseudogenes allowed me to dissect their miRNA-mediated roles away from of their ancestral protein-coding functions. I identified 47 protein-coding genes, each present in the earliest eutherian ancestor, whose protein-coding capability was specifically lost during rodent evolution, hereafter termed unitary pseudogenes. Seventy percent of unitary pseudogene loci are expressed in mice and their transcripts exhibit expression profiles that are as well conserved in human tissues as genes whose protein-coding capacity has been retained. In mice, the maintenance of spatial expression patterns after loss pseudogenisation is associated with the preservation of genetic interactions and four-fold enrichment in conserved, between mouse and human, miRNA response elements. Experimental evidence for one such unitary pseudogene, mouse pbcas4, revealed it to conserve its ancestral role as a miRNA decoy. Our results indicate that miRNA-mediated functions are sufficient to selectively maintain transcription after loss of protein-coding potential and suggest that post-transcriptional regulation can be an indispensable function of mRNAs.
How does transcriptional regulation by lincRNAs contribute to phenotypic variation?
Although a number of long non-coding RNA catalogues in different species have been compiled, the question of their evolutionary persistence had never been properly addressed. Only 15 % of intergenic lincRNAs in mice or humans were found to have transcribed orthologous sequences in the other species. This might reflect turnover of transcribed loci, or it might imply that lincRNAs, which are often lowly expressed and tissue-specific, have transcribed orthologous sequences that remain undetected. Indeed, analysis of the transcription of three lincRNA loci across homologous regions of the mammalian and avian brain revealed that some lincRNAs can have conserved expression patterns. Understanding the extent of lncRNA locus conservation or turnover will be essential if we are to understand their contributions to transcriptional control and to evolutionarily conserved and lineage-specific functions.
I have deployed for the first time a direct, interspecies comparative approach to assess the extent of lincRNA turnover. In collaboration with Duncan Odom's group (Cambridge), I used a combination of chromatin signatures and total RNA sequencing to annotate lincRNAs expressed in the adult liver (a highly homogeneous and evolutionarily conserved organ) of three closely related rodents. The relatively small evolutionary distance between these species allowed the distinction of sequence turnover from transcriptional turnover. I found that during rodent evolution lincRNA expression is gained or lost at over five times the rate of protein-coding genes and that sequence constraint reflects transcriptional conservation. Perhaps more surprising was my observation that lineage-specific expression of lincRNAs is specifically and robustly associated with the increased expression of their neighbouring protein-coding genes. Our results indicate that turnover of transcribed lincRNA loci is an important determinant of lineage-specific transcriptional differences.
The emergence of multicellular eukaryotes required complex gene regulatory networks to evolve so as to allow the spatially and temporally regulated transcription of thousands of genic loci. Transcripts lacking open-reading frames, non-coding RNAs, have been proposed to be an important yet largely unexplored class of transcriptional regulatory elements. In my fellowship I have used a powerful combination of in silico and in vitro approaches to characterise the transcriptional roles of non-coding RNAs, in particular those that are longer than 200 nucleotides and are intergenic (long intergenic noncoding RNAs, lincRNAs). I have addressed the following two questions: 1. What are the contributions of pseudogene derived-noncoding RNAs to post-transcriptional regulation? 2. How does transcriptional regulation by lincRNAs contribute to phenotypic variation?
Work performed and main results
Contribution of pseudogene-derived noncoding RNAs to post-transcriptional regulation
Negative regulation of transcript levels by small (22-25 nucleotides) non-coding RNAs (miRNAs) involves the recognition and binding of mature miRNA to miRNA response elements (MREs) often located in the 3' untranslated regions of target mRNAs. A miRNA can regulate large numbers of transcripts and target recognition is thought to lead to decreased miRNA concentration. It was recently demonstrated that competition for miRNA binding between transcripts sharing MREs (competitive endogenous RNAs, ceRNAs) can alter their abundance. This new layer of post-transcriptional regulation involves miRNA mediated crosstalk between numerous protein-coding and noncoding transcripts.
I have identified and am currently completing the characterisation of the retrotransposed gene copy (retrocopy) of a protein-coding member of a large multiprotein complex involved in transcriptional regulation of a number of loci. This retrocopy shares MREs with over 90 % of genes whose products form this protein complex. Importantly, changes in its expression influence the levels of many, but not all members, of this ceRNA network. Increased or decreased expression of miRNAs is sufficient to mimic some of the post-transcriptional effects of the retrocopy, indicating that indeed its post-transcriptional function is miRNA-mediated. By analysing the relative number of shared MREs and expression levels of different ceRNAs in this network I aim to shed light on the properties of transcripts that efficiently crosstalk. Furthermore, the multiprotein complex I am studying is involved in the regulation of several human disease genes. The detailed analysis of this retrocopy's contribution to the regulation of gene expression will provide insights into its contribution to abnormal human phenotypes.
An individual miRNA can contribute to the post-transcriptional regulation of ten-to-hundreds of non-homologous genes. This raised the question of whether the miRNA-mediated roles of an mRNA are indispensable per se. In particular, if the miRNA-mediated crosstalk between functionally related transcripts contributes to stabilising relative transcript levels this would make an mRNA's transcriptional regulatory functions less dispensable. To address this question and separate the protein-coding and miRNA decoy functions of transcripts, I considered the conservation of post-transcriptional roles by unitary pseudogenes. Pseudogenes exhibit significant sequence similarity to known protein-coding genes in the same or related genomes but contain mutations such as premature stop-codons or frame-shifting nucleotide insertion deletions that preclude their translation into functional proteins. Most such pseudogenes arose by duplication followed by pseudogenisation. Relatively few are 'unitary pseudogenes', those that derive not from duplication but from the lineage-specific acquisition of disrupting mutations in the coding sequences.
Investigating transcribed unitary pseudogenes allowed me to dissect their miRNA-mediated roles away from of their ancestral protein-coding functions. I identified 47 protein-coding genes, each present in the earliest eutherian ancestor, whose protein-coding capability was specifically lost during rodent evolution, hereafter termed unitary pseudogenes. Seventy percent of unitary pseudogene loci are expressed in mice and their transcripts exhibit expression profiles that are as well conserved in human tissues as genes whose protein-coding capacity has been retained. In mice, the maintenance of spatial expression patterns after loss pseudogenisation is associated with the preservation of genetic interactions and four-fold enrichment in conserved, between mouse and human, miRNA response elements. Experimental evidence for one such unitary pseudogene, mouse pbcas4, revealed it to conserve its ancestral role as a miRNA decoy. Our results indicate that miRNA-mediated functions are sufficient to selectively maintain transcription after loss of protein-coding potential and suggest that post-transcriptional regulation can be an indispensable function of mRNAs.
How does transcriptional regulation by lincRNAs contribute to phenotypic variation?
Although a number of long non-coding RNA catalogues in different species have been compiled, the question of their evolutionary persistence had never been properly addressed. Only 15 % of intergenic lincRNAs in mice or humans were found to have transcribed orthologous sequences in the other species. This might reflect turnover of transcribed loci, or it might imply that lincRNAs, which are often lowly expressed and tissue-specific, have transcribed orthologous sequences that remain undetected. Indeed, analysis of the transcription of three lincRNA loci across homologous regions of the mammalian and avian brain revealed that some lincRNAs can have conserved expression patterns. Understanding the extent of lncRNA locus conservation or turnover will be essential if we are to understand their contributions to transcriptional control and to evolutionarily conserved and lineage-specific functions.
I have deployed for the first time a direct, interspecies comparative approach to assess the extent of lincRNA turnover. In collaboration with Duncan Odom's group (Cambridge), I used a combination of chromatin signatures and total RNA sequencing to annotate lincRNAs expressed in the adult liver (a highly homogeneous and evolutionarily conserved organ) of three closely related rodents. The relatively small evolutionary distance between these species allowed the distinction of sequence turnover from transcriptional turnover. I found that during rodent evolution lincRNA expression is gained or lost at over five times the rate of protein-coding genes and that sequence constraint reflects transcriptional conservation. Perhaps more surprising was my observation that lineage-specific expression of lincRNAs is specifically and robustly associated with the increased expression of their neighbouring protein-coding genes. Our results indicate that turnover of transcribed lincRNA loci is an important determinant of lineage-specific transcriptional differences.