Limits...
Transcriptional dynamics reveal critical roles for non-coding RNAs in the immediate-early response.

Aitken S, Magi S, Alhendi AM, Itoh M, Kawaji H, Lassmann T, Daub CO, Arner E, Carninci P, Forrest AR, Hayashizaki Y, FANTOM ConsortiumKhachigian LM, Okada-Hatakeyama M, Semple CA - PLoS Comput. Biol. (2015)

Bottom Line: Surprisingly, these data suggest that the earliest transcriptional responses often involve promoters generating non-coding RNAs, many of which are produced in advance of canonical protein-coding IEGs.Consistent with this, we find that the response of both protein-coding and non-coding RNA IEGs can be explained by their transcriptionally poised, permissive chromatin state prior to stimulation.Our computational statistical method is well suited to meta-analyses as there is no requirement for transcripts to pass thresholds for significant differential expression between time points, and it is agnostic to the number of time points per dataset.

View Article: PubMed Central - PubMed

Affiliation: MRC Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh, United Kingdom.

ABSTRACT
The immediate-early response mediates cell fate in response to a variety of extracellular stimuli and is dysregulated in many cancers. However, the specificity of the response across stimuli and cell types, and the roles of non-coding RNAs are not well understood. Using a large collection of densely-sampled time series expression data we have examined the induction of the immediate-early response in unparalleled detail, across cell types and stimuli. We exploit cap analysis of gene expression (CAGE) time series datasets to directly measure promoter activities over time. Using a novel analysis method for time series data we identify transcripts with expression patterns that closely resemble the dynamics of known immediate-early genes (IEGs) and this enables a comprehensive comparative study of these genes and their chromatin state. Surprisingly, these data suggest that the earliest transcriptional responses often involve promoters generating non-coding RNAs, many of which are produced in advance of canonical protein-coding IEGs. IEGs are known to be capable of induction without de novo protein synthesis. Consistent with this, we find that the response of both protein-coding and non-coding RNA IEGs can be explained by their transcriptionally poised, permissive chromatin state prior to stimulation. We also explore the function of non-coding RNAs in the attenuation of the immediate early response in a small RNA sequencing dataset matched to the CAGE data: We identify a novel set of microRNAs responsible for the attenuation of the IEG response in an estrogen receptor positive cancer cell line. Our computational statistical method is well suited to meta-analyses as there is no requirement for transcripts to pass thresholds for significant differential expression between time points, and it is agnostic to the number of time points per dataset.

No MeSH data available.


Related in: MedlinePlus

Density plots of gene length against ts for early peak clusters.Grey contour lines indicate the projected ts for the completion of transcription accounting for gene length (a transcription rate of 60 bases/s is assumed [32]). (A) Early peak known IEGs (red symbols represent the underlying IEG CAGE cluster data). (B) Early peak known nucleotide binding genes and underlying data (blue symbols). (C) Travelling ratios for known IEGs and for early peak genes in MCF7 cells demonstrate promoter proximal pausing as the travelling ratio is shifts towards higher values. The intersection of IEGs and early peak genes (right-most plot) shows that the strong pausing effect seen for IEGs holds for those assigned the early peak signature.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4401570&req=5

pcbi.1004217.g003: Density plots of gene length against ts for early peak clusters.Grey contour lines indicate the projected ts for the completion of transcription accounting for gene length (a transcription rate of 60 bases/s is assumed [32]). (A) Early peak known IEGs (red symbols represent the underlying IEG CAGE cluster data). (B) Early peak known nucleotide binding genes and underlying data (blue symbols). (C) Travelling ratios for known IEGs and for early peak genes in MCF7 cells demonstrate promoter proximal pausing as the travelling ratio is shifts towards higher values. The intersection of IEGs and early peak genes (right-most plot) shows that the strong pausing effect seen for IEGs holds for those assigned the early peak signature.

Mentions: However, a weak positive correlation between gene length and ts could be shown for early peak genes by Pearson correlation (all early peak genes: R = 0.10, p = 2.8e-11; known IEGs: R = 0.11, p = 1.2e-3; nucleotide binding genes: R = 0.11, p = 3.8e-5). Fig 3A and 3B contrast the density of gene length vs ts for known IEGs assigned to the early peak signature with the densities of early peak nucleotide binding genes. Known IEGs were typically shorter in length and had lower ts than nucleotide binding genes (combined data from all four datasets). Surprisingly, Fig 3A demonstrates that short IEGs ∼ 1-5Kb in length were activated with broad range of kinetics, from the lowest to the highest switch time ts. Thus the typically short length of IEGs will decrease the time required for their transcription, but IEGs are not necessarily induced with equally rapid kinetics. The time at which short IEGs reach their transcriptional peak was up to three hours after the stimulus suggesting their activation rates coordinate their expression with diverse processes and pathways: late-acting IEGs are not delayed due to gene length. Further, many early peak genes not known to be IEGs fell within the range of characteristics of known IEGs: length from 1.2Kb-240Kb, ts less than 210 min.


Transcriptional dynamics reveal critical roles for non-coding RNAs in the immediate-early response.

Aitken S, Magi S, Alhendi AM, Itoh M, Kawaji H, Lassmann T, Daub CO, Arner E, Carninci P, Forrest AR, Hayashizaki Y, FANTOM ConsortiumKhachigian LM, Okada-Hatakeyama M, Semple CA - PLoS Comput. Biol. (2015)

Density plots of gene length against ts for early peak clusters.Grey contour lines indicate the projected ts for the completion of transcription accounting for gene length (a transcription rate of 60 bases/s is assumed [32]). (A) Early peak known IEGs (red symbols represent the underlying IEG CAGE cluster data). (B) Early peak known nucleotide binding genes and underlying data (blue symbols). (C) Travelling ratios for known IEGs and for early peak genes in MCF7 cells demonstrate promoter proximal pausing as the travelling ratio is shifts towards higher values. The intersection of IEGs and early peak genes (right-most plot) shows that the strong pausing effect seen for IEGs holds for those assigned the early peak signature.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4401570&req=5

pcbi.1004217.g003: Density plots of gene length against ts for early peak clusters.Grey contour lines indicate the projected ts for the completion of transcription accounting for gene length (a transcription rate of 60 bases/s is assumed [32]). (A) Early peak known IEGs (red symbols represent the underlying IEG CAGE cluster data). (B) Early peak known nucleotide binding genes and underlying data (blue symbols). (C) Travelling ratios for known IEGs and for early peak genes in MCF7 cells demonstrate promoter proximal pausing as the travelling ratio is shifts towards higher values. The intersection of IEGs and early peak genes (right-most plot) shows that the strong pausing effect seen for IEGs holds for those assigned the early peak signature.
Mentions: However, a weak positive correlation between gene length and ts could be shown for early peak genes by Pearson correlation (all early peak genes: R = 0.10, p = 2.8e-11; known IEGs: R = 0.11, p = 1.2e-3; nucleotide binding genes: R = 0.11, p = 3.8e-5). Fig 3A and 3B contrast the density of gene length vs ts for known IEGs assigned to the early peak signature with the densities of early peak nucleotide binding genes. Known IEGs were typically shorter in length and had lower ts than nucleotide binding genes (combined data from all four datasets). Surprisingly, Fig 3A demonstrates that short IEGs ∼ 1-5Kb in length were activated with broad range of kinetics, from the lowest to the highest switch time ts. Thus the typically short length of IEGs will decrease the time required for their transcription, but IEGs are not necessarily induced with equally rapid kinetics. The time at which short IEGs reach their transcriptional peak was up to three hours after the stimulus suggesting their activation rates coordinate their expression with diverse processes and pathways: late-acting IEGs are not delayed due to gene length. Further, many early peak genes not known to be IEGs fell within the range of characteristics of known IEGs: length from 1.2Kb-240Kb, ts less than 210 min.

Bottom Line: Surprisingly, these data suggest that the earliest transcriptional responses often involve promoters generating non-coding RNAs, many of which are produced in advance of canonical protein-coding IEGs.Consistent with this, we find that the response of both protein-coding and non-coding RNA IEGs can be explained by their transcriptionally poised, permissive chromatin state prior to stimulation.Our computational statistical method is well suited to meta-analyses as there is no requirement for transcripts to pass thresholds for significant differential expression between time points, and it is agnostic to the number of time points per dataset.

View Article: PubMed Central - PubMed

Affiliation: MRC Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh, United Kingdom.

ABSTRACT
The immediate-early response mediates cell fate in response to a variety of extracellular stimuli and is dysregulated in many cancers. However, the specificity of the response across stimuli and cell types, and the roles of non-coding RNAs are not well understood. Using a large collection of densely-sampled time series expression data we have examined the induction of the immediate-early response in unparalleled detail, across cell types and stimuli. We exploit cap analysis of gene expression (CAGE) time series datasets to directly measure promoter activities over time. Using a novel analysis method for time series data we identify transcripts with expression patterns that closely resemble the dynamics of known immediate-early genes (IEGs) and this enables a comprehensive comparative study of these genes and their chromatin state. Surprisingly, these data suggest that the earliest transcriptional responses often involve promoters generating non-coding RNAs, many of which are produced in advance of canonical protein-coding IEGs. IEGs are known to be capable of induction without de novo protein synthesis. Consistent with this, we find that the response of both protein-coding and non-coding RNA IEGs can be explained by their transcriptionally poised, permissive chromatin state prior to stimulation. We also explore the function of non-coding RNAs in the attenuation of the immediate early response in a small RNA sequencing dataset matched to the CAGE data: We identify a novel set of microRNAs responsible for the attenuation of the IEG response in an estrogen receptor positive cancer cell line. Our computational statistical method is well suited to meta-analyses as there is no requirement for transcripts to pass thresholds for significant differential expression between time points, and it is agnostic to the number of time points per dataset.

No MeSH data available.


Related in: MedlinePlus