Limits...
Comparative analysis of human and mouse expression data illuminates tissue-specific evolutionary patterns of miRNAs.

Roux J, Gonzàlez-Porta M, Robinson-Rechavi M - Nucleic Acids Res. (2012)

Bottom Line: In this comparative framework, we confirm some predictions of previously advanced models of miRNA evolution, e.g. that miRNAs arise more frequently de novo than by duplication, or that the number of protein-coding gene targeted by miRNAs decreases with evolutionary time.Together, our results refine the models used so far to depict the evolution of miRNA genes.They underline the role of tissue-specific selective forces on the evolution of miRNAs, as well as the potential co-evolution patterns between miRNAs and the protein-coding genes they target.

View Article: PubMed Central - PubMed

Affiliation: Department of Ecology and Evolution, Biophore, University of Lausanne, Switzerland.

ABSTRACT
MicroRNAs (miRNAs) constitute an important class of gene regulators. While models have been proposed to explain their appearance and expansion, the validation of these models has been difficult due to the lack of comparative studies. Here, we analyze miRNA evolutionary patterns in two mammals, human and mouse, in relation to the age of miRNA families. In this comparative framework, we confirm some predictions of previously advanced models of miRNA evolution, e.g. that miRNAs arise more frequently de novo than by duplication, or that the number of protein-coding gene targeted by miRNAs decreases with evolutionary time. We also corroborate that miRNAs display an increase in expression level with evolutionary time, however we show that this relation is largely tissue-dependent, and especially low in embryonic or nervous tissues. We identify a bias of tag-sequencing techniques regarding the assessment of breadth of expression, leading us, contrary to predictions, to find more tissue-specific expression of older miRNAs. Together, our results refine the models used so far to depict the evolution of miRNA genes. They underline the role of tissue-specific selective forces on the evolution of miRNAs, as well as the potential co-evolution patterns between miRNAs and the protein-coding genes they target.

Show MeSH
Expression divergence of miRNAs between human and mouse. (A) Boxplot of expression divergence of miRNA families in different tissues between human and mouse. The significance of expression divergence was assessed using a test developed by Audic and Claverie (see ‘Materials and Methods’ section). The P-values are corrected for multiple testing, and −log10 of the adjusted P-values is displayed on the x-axis. This allows to spread on a broad range the small adjusted P-values, which correspond to significant cases of expression divergence. Only families where expression counts were non  in both species were used in this analysis; see Supplementary Figure S3 for the analysis using the complete dataset. A vertical dashed line indicates the 20% FDR threshold. (B) Relation between the expression divergence score, −log10 (adjusted P-values), of miRNA families between human and mouse and their date of appearance in the genome. Darker dots in the plot result from the superposition of several data points.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3401464&req=5

gks279-F5: Expression divergence of miRNAs between human and mouse. (A) Boxplot of expression divergence of miRNA families in different tissues between human and mouse. The significance of expression divergence was assessed using a test developed by Audic and Claverie (see ‘Materials and Methods’ section). The P-values are corrected for multiple testing, and −log10 of the adjusted P-values is displayed on the x-axis. This allows to spread on a broad range the small adjusted P-values, which correspond to significant cases of expression divergence. Only families where expression counts were non in both species were used in this analysis; see Supplementary Figure S3 for the analysis using the complete dataset. A vertical dashed line indicates the 20% FDR threshold. (B) Relation between the expression divergence score, −log10 (adjusted P-values), of miRNA families between human and mouse and their date of appearance in the genome. Darker dots in the plot result from the superposition of several data points.

Mentions: It is possible that part of the differences results from our methodology, i.e. the low amount of data available for some tissues limits the power of the statistical test. Lowly significant results can either reflect a low divergence of expression, or a lack of statistical power to detect divergence. This lack of power is particularly marked for families which display no expression in a given HOG, for at least one of the two species. To take this into account, we tested only the subset of families in HOGs for which both species had at least one count (Figure 5A). As expected, due to the increased statistical power, the proportion of significant families increases in most HOGs (up to 70% for metanephros/kidney). Their ranking is also affected: testis now displays an intermediate pattern, similar to ovary and placenta, more consistent with the observations of Khaitovich et al. (60), although still at odds with Brawand et al. (61). Brain still displays an elevated rate of divergence, reflecting that there might be less purifying selection, or more positive selection, acting on expression patterns of miRNAs than on those of protein-coding genes in the brain (62). Of note, our test cannot differentiate between divergence due to relaxed purifying selection or due to positive selection.Figure 5.


Comparative analysis of human and mouse expression data illuminates tissue-specific evolutionary patterns of miRNAs.

Roux J, Gonzàlez-Porta M, Robinson-Rechavi M - Nucleic Acids Res. (2012)

Expression divergence of miRNAs between human and mouse. (A) Boxplot of expression divergence of miRNA families in different tissues between human and mouse. The significance of expression divergence was assessed using a test developed by Audic and Claverie (see ‘Materials and Methods’ section). The P-values are corrected for multiple testing, and −log10 of the adjusted P-values is displayed on the x-axis. This allows to spread on a broad range the small adjusted P-values, which correspond to significant cases of expression divergence. Only families where expression counts were non  in both species were used in this analysis; see Supplementary Figure S3 for the analysis using the complete dataset. A vertical dashed line indicates the 20% FDR threshold. (B) Relation between the expression divergence score, −log10 (adjusted P-values), of miRNA families between human and mouse and their date of appearance in the genome. Darker dots in the plot result from the superposition of several data points.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3401464&req=5

gks279-F5: Expression divergence of miRNAs between human and mouse. (A) Boxplot of expression divergence of miRNA families in different tissues between human and mouse. The significance of expression divergence was assessed using a test developed by Audic and Claverie (see ‘Materials and Methods’ section). The P-values are corrected for multiple testing, and −log10 of the adjusted P-values is displayed on the x-axis. This allows to spread on a broad range the small adjusted P-values, which correspond to significant cases of expression divergence. Only families where expression counts were non in both species were used in this analysis; see Supplementary Figure S3 for the analysis using the complete dataset. A vertical dashed line indicates the 20% FDR threshold. (B) Relation between the expression divergence score, −log10 (adjusted P-values), of miRNA families between human and mouse and their date of appearance in the genome. Darker dots in the plot result from the superposition of several data points.
Mentions: It is possible that part of the differences results from our methodology, i.e. the low amount of data available for some tissues limits the power of the statistical test. Lowly significant results can either reflect a low divergence of expression, or a lack of statistical power to detect divergence. This lack of power is particularly marked for families which display no expression in a given HOG, for at least one of the two species. To take this into account, we tested only the subset of families in HOGs for which both species had at least one count (Figure 5A). As expected, due to the increased statistical power, the proportion of significant families increases in most HOGs (up to 70% for metanephros/kidney). Their ranking is also affected: testis now displays an intermediate pattern, similar to ovary and placenta, more consistent with the observations of Khaitovich et al. (60), although still at odds with Brawand et al. (61). Brain still displays an elevated rate of divergence, reflecting that there might be less purifying selection, or more positive selection, acting on expression patterns of miRNAs than on those of protein-coding genes in the brain (62). Of note, our test cannot differentiate between divergence due to relaxed purifying selection or due to positive selection.Figure 5.

Bottom Line: In this comparative framework, we confirm some predictions of previously advanced models of miRNA evolution, e.g. that miRNAs arise more frequently de novo than by duplication, or that the number of protein-coding gene targeted by miRNAs decreases with evolutionary time.Together, our results refine the models used so far to depict the evolution of miRNA genes.They underline the role of tissue-specific selective forces on the evolution of miRNAs, as well as the potential co-evolution patterns between miRNAs and the protein-coding genes they target.

View Article: PubMed Central - PubMed

Affiliation: Department of Ecology and Evolution, Biophore, University of Lausanne, Switzerland.

ABSTRACT
MicroRNAs (miRNAs) constitute an important class of gene regulators. While models have been proposed to explain their appearance and expansion, the validation of these models has been difficult due to the lack of comparative studies. Here, we analyze miRNA evolutionary patterns in two mammals, human and mouse, in relation to the age of miRNA families. In this comparative framework, we confirm some predictions of previously advanced models of miRNA evolution, e.g. that miRNAs arise more frequently de novo than by duplication, or that the number of protein-coding gene targeted by miRNAs decreases with evolutionary time. We also corroborate that miRNAs display an increase in expression level with evolutionary time, however we show that this relation is largely tissue-dependent, and especially low in embryonic or nervous tissues. We identify a bias of tag-sequencing techniques regarding the assessment of breadth of expression, leading us, contrary to predictions, to find more tissue-specific expression of older miRNAs. Together, our results refine the models used so far to depict the evolution of miRNA genes. They underline the role of tissue-specific selective forces on the evolution of miRNAs, as well as the potential co-evolution patterns between miRNAs and the protein-coding genes they target.

Show MeSH