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

Mature microRNA regulation and host gene activation.(A) Expression of mature hsa-mir-6163 and transcriptional activation of two of its target IEGs FOSB and EGR3 in MCF7 cells in response to HRG. Data values are plotted as circles (median value is filled). (B) Median CAGE expression (black circles) of precursor miRNA and median mature miRNA expression (red triangles) for hsa-mir-320a, host lncRNA MIR155HG and mature hsa-mir-155, and for hsa-mir-21 in MCF7-HRG (three replicates, lines are a spline fitted to the data). For hsa-mir-320a the increase in CAGE expression is significant when comparing 0min and 210min and the decrease in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-155 the increase in CAGE expression of MIR155HG is significant when comparing 0min and 180min, and the increase in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-21 the increases in CAGE expression and in mature transcript levels are significant when comparing 0min and 80min (p ≤ 0.05 by t test).
© Copyright Policy
Related In: Results  -  Collection

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

pcbi.1004217.g005: Mature microRNA regulation and host gene activation.(A) Expression of mature hsa-mir-6163 and transcriptional activation of two of its target IEGs FOSB and EGR3 in MCF7 cells in response to HRG. Data values are plotted as circles (median value is filled). (B) Median CAGE expression (black circles) of precursor miRNA and median mature miRNA expression (red triangles) for hsa-mir-320a, host lncRNA MIR155HG and mature hsa-mir-155, and for hsa-mir-21 in MCF7-HRG (three replicates, lines are a spline fitted to the data). For hsa-mir-320a the increase in CAGE expression is significant when comparing 0min and 210min and the decrease in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-155 the increase in CAGE expression of MIR155HG is significant when comparing 0min and 180min, and the increase in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-21 the increases in CAGE expression and in mature transcript levels are significant when comparing 0min and 80min (p ≤ 0.05 by t test).

Mentions: Reasoning that miRNA-mediated repression will be reflected in the CAGE signals, either by direct action on mRNA or indirectly through transcriptional inactivation processes, we sought to establish a connection between the targets of mature miRNA that are assigned to the dip signature, and protein-coding genes with CAGE clusters assigned to the early peak signature in MCF7 cells stimulated with HRG. Making use of the TargetScan database of miRNA targets (version 6.2; http://www.targetscan.org) we found all targets for dip miRNAs. For 12 of the 15 dip miRNA present in TargetScan we observed a greater representation of miRNA targets in early peak genes than in a reference set of unregulated genes (namely, all protein-coding genes that could not be assigned to a kinetic signature in MCF7-HRG). The targets of seven of these miRNA were significantly overrepresented (by hypergeometric test): hsa-mir-139 (p = 2.6e-2); hsa-mir-224 (p = 1.1e-2); hsa-mir-522 (p = 1.5e-6); hsa-mir-548n (p = 9.8e-7), hsa-mir-676 (p = 3.2e-2), hsa-mir-3163 (p = 2.1e-7) and hsa-mir-3662 (p = 3.5e-3) and are candidate ID-miRs for MCF7 cells. The data for mature hsa-mir-3163 and two of its targets FOSB and EGR3 are shown in Fig 5A.


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)

Mature microRNA regulation and host gene activation.(A) Expression of mature hsa-mir-6163 and transcriptional activation of two of its target IEGs FOSB and EGR3 in MCF7 cells in response to HRG. Data values are plotted as circles (median value is filled). (B) Median CAGE expression (black circles) of precursor miRNA and median mature miRNA expression (red triangles) for hsa-mir-320a, host lncRNA MIR155HG and mature hsa-mir-155, and for hsa-mir-21 in MCF7-HRG (three replicates, lines are a spline fitted to the data). For hsa-mir-320a the increase in CAGE expression is significant when comparing 0min and 210min and the decrease in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-155 the increase in CAGE expression of MIR155HG is significant when comparing 0min and 180min, and the increase in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-21 the increases in CAGE expression and in mature transcript levels are significant when comparing 0min and 80min (p ≤ 0.05 by t test).
© Copyright Policy
Related In: Results  -  Collection

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

pcbi.1004217.g005: Mature microRNA regulation and host gene activation.(A) Expression of mature hsa-mir-6163 and transcriptional activation of two of its target IEGs FOSB and EGR3 in MCF7 cells in response to HRG. Data values are plotted as circles (median value is filled). (B) Median CAGE expression (black circles) of precursor miRNA and median mature miRNA expression (red triangles) for hsa-mir-320a, host lncRNA MIR155HG and mature hsa-mir-155, and for hsa-mir-21 in MCF7-HRG (three replicates, lines are a spline fitted to the data). For hsa-mir-320a the increase in CAGE expression is significant when comparing 0min and 210min and the decrease in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-155 the increase in CAGE expression of MIR155HG is significant when comparing 0min and 180min, and the increase in mature transcript levels is significant when comparing 0min and 240min (p ≤ 0.05 by t test). For hsa-mir-21 the increases in CAGE expression and in mature transcript levels are significant when comparing 0min and 80min (p ≤ 0.05 by t test).
Mentions: Reasoning that miRNA-mediated repression will be reflected in the CAGE signals, either by direct action on mRNA or indirectly through transcriptional inactivation processes, we sought to establish a connection between the targets of mature miRNA that are assigned to the dip signature, and protein-coding genes with CAGE clusters assigned to the early peak signature in MCF7 cells stimulated with HRG. Making use of the TargetScan database of miRNA targets (version 6.2; http://www.targetscan.org) we found all targets for dip miRNAs. For 12 of the 15 dip miRNA present in TargetScan we observed a greater representation of miRNA targets in early peak genes than in a reference set of unregulated genes (namely, all protein-coding genes that could not be assigned to a kinetic signature in MCF7-HRG). The targets of seven of these miRNA were significantly overrepresented (by hypergeometric test): hsa-mir-139 (p = 2.6e-2); hsa-mir-224 (p = 1.1e-2); hsa-mir-522 (p = 1.5e-6); hsa-mir-548n (p = 9.8e-7), hsa-mir-676 (p = 3.2e-2), hsa-mir-3163 (p = 2.1e-7) and hsa-mir-3662 (p = 3.5e-3) and are candidate ID-miRs for MCF7 cells. The data for mature hsa-mir-3163 and two of its targets FOSB and EGR3 are shown in Fig 5A.

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