Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2012 Feb 10:3:3.
doi: 10.1186/1758-907X-3-3.

Target gene expression levels and competition between transfected and endogenous microRNAs are strong confounding factors in microRNA high-throughput experiments

Takaya Saito  1 , Pål Sætrom
Affiliations

Target gene expression levels and competition between transfected and endogenous microRNAs are strong confounding factors in microRNA high-throughput experiments

Takaya Saito et al. Silence. .

Abstract

Background: MicroRNA (miRNA) target genes tend to have relatively long and conserved 3' untranslated regions (UTRs), but to what degree these characteristics contribute to miRNA targeting is poorly understood. Different high-throughput experiments have, for example, shown that miRNAs preferentially regulate genes with both short and long 3' UTRs and that target site conservation is both important and irrelevant for miRNA targeting.

Results: We have analyzed several gene context-dependent features, including 3' UTR length, 3' UTR conservation, and messenger RNA (mRNA) expression levels, reported to have conflicting influence on miRNA regulation. By taking into account confounding factors such as technology-dependent experimental bias and competition between transfected and endogenous miRNAs, we show that two factors - target gene expression and competition - could explain most of the previously reported experimental differences. Moreover, we find that these and other target site-independent features explain about the same amount of variation in target gene expression as the target site-dependent features included in the TargetScan model.

Conclusions: Our results show that it is important to consider confounding factors when interpreting miRNA high throughput experiments and urge special caution when using microarray data to compare average regulatory effects between groups of genes that have different average gene expression levels.

PubMed Disclaimer

Figures

Figure 1
Figure 1
Heatmaps show effects of ectopic miRNA regulation for sub-groups of 3' UTR length, 3' UTR conservation, and mRNA expression level. Three heat maps show -log (base 2) transformed P-values for (C) 3' UTR length, (D) 3' UTR conservation, and (E) mRNA expression. We added two cumulative density plots for Lim (A) and Selbach (B) to illustrate the multiple non-parametric tests for the sub-groups of 3' UTR lengths; 'All' is the cumulative density for all the genes measured in the experiment and represents the reference distribution. 'All' genes include both predicted miRNA target and non-target genes. Although the mRNA expression data (E) was cell type specific and for HeLa cells, we included the Linsley dataset (from HCT116 and DLD-1 colon tumor cells) in the heat map for comparison purpose (indicated with '*'). The color labels under the dendrogram represent green for microarray of transfection assay, gray for proteomics of transfection assay, and orange for both microarray and proteomics with inhibition assay. miRNA, microRNA; mRNA, messenger RNA; UTR, untranslated region.
Figure 2
Figure 2
Endogenous miRNAs tend to target genes with long 3' UTRs and exogenous miRNAs target highly expressed genes that had a small influence of endogenous miRNAs. Two cumulative density plots of the log-ratio values show the miRNA down-regulatory effects on sub-groups of (A) 3' UTR length with the Grimson dataset and (B) mRNA expression level with the Jackson dataset for ectopically expressed miRNA or siRNA target genes that were separated into T +Endo (T +E), T -Endo (T -E), NT +Endo (NT + E), and NT -Endo (NT -E). miRNA, microRNA; mRNA, messenger RNA; siRNA, small interfering RNA; UTR, untranslated region.
Figure 3
Figure 3
Microarrays but not proteomics are biased towards detecting differential expression among highly expressed genes. Cumulative density plots of log-ratio values for miRNA targets grouped by gene expression levels in (A) the Grimson and (B) the Selbach datasets. (C) Barplots show the ratio of the six sub-groups of mRNA expression levels subdivided by predicted exogenous and endogenous miRNA targeting in the Grimson and Selbach datasets for all genes ('All') and down-regulated genes ('Down-reg' P < 0.01; log ratio (lr) < -0.01). (D) Scatter plots show log2 enrichment of down-regulated genes compared with all genes for the six sub-groups of mRNA expression levels in all studied datasets. Lines and shaded grays show respectively linear fits and standard errors for the microarray (red dots) and proteomics (blue triangles) experiments; P-values (lower left) are unadjusted P-values from Pearson correlation tests. Data points based on a single gene were excluded. The regression lines show that in the microarray but not the proteomics experiments, down-regulated genes are enriched among highly expressed genes and that this enrichment depends on gene expression levels. miRNA, microRNA; mRNA, messenger RNA.
Figure 4
Figure 4
Genes without target sites for endogenous miRNAs show less dilution effects than does the complete set of potential targets. (A) The scatter plot shows the average log ratios for predicted miRNA and siRNA targets as measured by microarrays of 90 over-expression experiments (55 miRNAs and 35 siRNAs) as a function of the miRNAs' and siRNAs' total number of target sites. The line is based on a linear regression and indicates that there is a significant correlation between the total number of target sites and average log ratio (r = 0.37; P < 0.001). (B) The scatter plot shows the average log ratios for the subset of genes that have no predicted target sites for endogenous miRNAs as a function of the miRNAs' and siRNAs' total number of target sites (r = 0.22; P = 0.068). Only the 70 samples assayed in HeLa were included. In both plots, red circles represent miRNAs, and blue triangles represent siRNAs. miRNA, microRNA; siRNA, small interfering RNA.
Figure 5
Figure 5
Coefficients of a linear regression with eight factors. The dot plot shows the coefficients of the liner model with formula: -log ratio = ln3 + cs3 + exp +#site_m + #endo_m + #site_s + p_ma + e_oe. The dot size shows -log10 of the coefficient's P-value. Positive coefficients associate with miRNA down-regulation. miRNA, microRNA.
Figure 6
Figure 6
Coefficients of a linear regression with nine factors. The dot plot shows the coefficients of the liner model with formula: -log ratio = ln3 + cs3 + exp +#site_m + #endo_m + #site_s + p_ma + e_oe + ts_score. The dot size shows -log10 of the coefficient's p-value. Positive coefficients associate with miRNA down-regulation. miRNA, microRNA.
Figure 7
Figure 7
CpG-rich genes, non-developmental genes, and housekeeping genes appear to be strong miRNA targets in microarray experiments. We subdivided all RefSeq genes into sub-groups based on three different features: CpG frequency (CpG), and whether the genes were developmental (Dev) and housekeeping (HK) genes (see Methods). See Figure 1 for a description of the heat maps. miRNA, microRNA.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Ambros V. The functions of animal microRNAs. Nature. 2004;431:350–355. doi: 10.1038/nature02871. - DOI - PubMed
    1. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. 2004;116:281–297. doi: 10.1016/S0092-8674(04)00045-5. - DOI - PubMed
    1. Stefani G, Slack FJ. Small non-coding RNAs in animal development. Nat Rev Mol Cell Biol. 2008;9:219–230. doi: 10.1038/nrm2347. - DOI - PubMed
    1. Croce CM. Causes and consequences of microRNA dysregulation in cancer. Nat Rev Genet. 2009;10:704–714. doi: 10.1038/nrg2634. - DOI - PMC - PubMed
    1. Hwang HW, Mendell JT. MicroRNAs in cell proliferation, cell death, and tumorigenesis. Br J Cancer. 2007;96(Suppl):R40–44. - PubMed

LinkOut - more resources