Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2013 Mar 1;41(5):2817-31.
doi: 10.1093/nar/gks1471. Epub 2013 Jan 17.

Interplay of microRNAs, Transcription Factors and Target Genes: Linking Dynamic Expression Changes to Function

Affiliations
Free PMC article

Interplay of microRNAs, Transcription Factors and Target Genes: Linking Dynamic Expression Changes to Function

Petr V Nazarov et al. Nucleic Acids Res. .
Free PMC article

Abstract

MicroRNAs (miRNAs) are ubiquitously expressed small non-coding RNAs that, in most cases, negatively regulate gene expression at the post-transcriptional level. miRNAs are involved in fine-tuning fundamental cellular processes such as proliferation, cell death and cell cycle control and are believed to confer robustness to biological responses. Here, we investigated simultaneously the transcriptional changes of miRNA and mRNA expression levels over time after activation of the Janus kinase/Signal transducer and activator of transcription (Jak/STAT) pathway by interferon-γ stimulation of melanoma cells. To examine global miRNA and mRNA expression patterns, time-series microarray data were analysed. We observed delayed responses of miRNAs (after 24-48 h) with respect to mRNAs (12-24 h) and identified biological functions involved at each step of the cellular response. Inference of the upstream regulators allowed for identification of transcriptional regulators involved in cellular reactions to interferon-γ stimulation. Linking expression profiles of transcriptional regulators and miRNAs with their annotated functions, we demonstrate the dynamic interplay of miRNAs and upstream regulators with biological functions. Finally, our data revealed network motifs in the form of feed-forward loops involving transcriptional regulators, mRNAs and miRNAs. Additional information obtained from integrating time-series mRNA and miRNA data may represent an important step towards understanding the regulatory principles of gene expression.

Figures

Figure 1.
Figure 1.
Graphical representation of the computational workflow. In the first step, miRNA and mRNA array data were pre-processed and the quality was assessed using Partek® GS. A filtering step was included to select only expressed and detectable features. Co-expression analysis and identification of potential targets were performed directly on the paired miRNA-mRNA data. Differential expression analysis was performed using the ‘limma’ package of R/Bioconductor to identify significant differentially expressed mRNAs and miRNAs over time. Functional and pathway analysis together with identification of upstream regulators was carried out in IPA based on SDE molecules. The dynamic behaviour of gene regulation circuitry in response to IFN-γ stimulus was visualized as Circos plots.
Figure 2.
Figure 2.
Projection of dynamic changes of the transcriptome and the miRNome of melanoma cells after IFN-γ stimulation. (A) PCA visualizes the evolution of both data sets over time, with the vertical axis corresponding to the first principal component of mRNA data and the horizontal axis showing the principal component of miRNA data. The percentage of variability in the data sets represented by each axis is shown. The dots represent sample duplicates (complete arrays) for the indicated stimulation times. (B) The number of SDE mRNAs and miRNAs with FDR < 0.001 for each condition compared with the untreated control is shown. The maximum gradient was observed between 12 h/ctrl and 24 h/ctrl for mRNAs and between 24 h/ctrl and 48 h/ctrl for miRNAs.
Figure 3.
Figure 3.
Heat maps representing the time evolution of the top 100 mRNAs and all 65 differentially regulated miRNAs detected through multi-class ‘limma’ analysis (FDR < 0.001). Standardized expression values for each feature were reordered by hierarchical clustering, resulting in three pronounced clusters depicted on the right of each heat map. Each cluster contains member gene or miRNA profiles (grey lines) and mean expression values (dots). (A) The top 100 significantly expressed and annotated mRNAs fall into three main groups (cluster A, B, C). Names of TFs involved in IFN-γ signalling are marked in red. Clustering was supported by bootstrapping using the ‘clusterCons’ package of R. Altered cluster annotations were only observed for two genes (A2M and APOBEC3G), which after bootstrapping were more likely to belong to cluster C rather than B. (B) The majority of all SDE miRNAs belong to two clusters (a and c), with delayed up- or down-regulation after 24 h, respectively. Cluster b contains only two miRNAs (miR-125b* and miR-21*).
Figure 4.
Figure 4.
Dynamic changes in inferred functional categories based on SDE mRNAs and miRNAs. The minimum adjusted P-values for member functions were combined to illustrate dynamic changes in all enriched functional categories obtained from IPA analysis (‘n/s;: P > 0.05). The intensity of grey boxes represents scaled adjusted P-values (log-transformed) for each category: white—non-significant (>0.01), dark grey—smallest adjusted P-value for each category among the time points.
Figure 5.
Figure 5.
Representation of the top canonical pathway ‘interferon signalling’ detected when simultaneously analysing the mRNA and miRNA data sets with IPA. Parts related to IFN-α/β were removed, and SDE miRNAs targeting the detected genes of the pathway were added. Connections between miRNAs and their targets were established by IPA. Genes that were differentially expressed in at least one condition are marked with filled grey symbols. Expression changes at 3, 12, 24, 48 and 72 h with respect to untreated controls are shown as bar charts close to each molecule. For non-significant conditions, a line is shown instead of a bar. The last bar on the far right always corresponds to JII-treated control (72 h). Connections between main players of the signalling pathway are depicted as lines: relationships between miRNAs and target mRNAs are shown as thin lines, whereas relationships between TFs and target mRNAs are presented by thicker lines either indicating activation or repression.
Figure 6.
Figure 6.
Circos diagrams showing dynamical dependency of the transcriptome, represented in the form of top biological categories (purple) and of inferred upstream regulators: transcription regulators (TRs: brown) and miRNAs (blue). Time points for 3, 12, 24, 48 and 72 h and for the 72-h JII-treated control were compared with the untreated control, and the SDE molecules were analysed by ‘limma’, with contrasts (FDR < 0.001) as described in ‘Materials and Methods’ section. The width of the categories is related to the number of member mRNAs. A connection between a TR and a functional category means that this TF was detected as an activated or inhibited TR by upstream regulator analysis, and that its target genes grouping in the respective functional category were differentially expressed in at least one time point. A connection between a miRNA and a TR implies that this miRNA was SDE and was predicted to target the TR genes. Finally, a connection between a miRNA and a functional category indicates that one or several of its target genes of a differentially expressed miRNA belonged to the assigned category. The thickness of connecting lines illustrates higher number of targets of TRs or miRNAs within this functional category.
Figure 7.
Figure 7.
Graphical representation of regulatory sub-networks (extracted from IPA) includes three activated TRs (dark grey boxes), four genes (rounded boxes) and four miRNAs (ellipses). Activation time for each part of the sub-network is shown by arrows on top. ‘mir’ represents the immature form of miRNAs, whereas ‘miR’ denotes the mature form. Connections between molecules are presented with respect to experimental observations and IPA predictions. Black: molecules have correlated (anti-correlated in case of inhibition) expression profiles. Dotted arrows: mRNA profile of a target molecule is in concordance with the predicted activation state of the TF. Grey arrows: direct interaction was not observed, suggesting presence of cumulative effect of other regulators.

Similar articles

See all similar articles

Cited by 68 articles

See all "Cited by" articles

References

    1. Lee RC, Feinbaum RL, Ambros V. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell. 1993;75:843–854. - PubMed
    1. He L, Hannon GJ. MicroRNAs: small RNAs with a big role in gene regulation. Nat. Rev. Genet. 2004;5:522–531. - PubMed
    1. Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalian mRNAs are conserved targets of microRNAs. Genome Res. 2009;19:92–105. - PMC - PubMed
    1. Mack GS. MicroRNA gets down to business. Nat. Biotechnol. 2007;25:631–638. - PubMed
    1. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. 2004;116:281–297. - PubMed

Publication types

MeSH terms

Feedback