Biased alternative polyadenylation in human tissues

Abstract

Background: Alternative polyadenylation is one of the mechanisms in human cells that give rise to a variety of transcripts from a single gene. More than half of the human genes have multiple polyadenylation sites (poly(A) sites), leading to variable mRNA and protein products. Previous studies of individual genes have indicated that alternative polyadenylation could occur in a tissue-specific manner.

Results: We set out to systematically investigate the occurrence and mechanism of alternative polyadenylation in different human tissues using bioinformatic approaches. Using expressed sequence tag (EST) data, we investigated 42 distinct tissue types. We found that several tissues tend to use poly(A) sites that are biased toward certain locations of a gene, such as sites located in introns or internal exons, and various sites in the exon located closest to the 3' end. We also identified several tissues, including eye, retina and placenta, that tend to use poly(A) sites not frequently used in other tissues. By exploring microarray expression data, we analyzed over 20 genes whose protein products are involved in the process or regulation of mRNA polyadenylation. Several brain tissues showed high concordance of gene expression of these genes with each other, but low concordance with other tissue types. By comparing genomic regions surrounding poly(A) sites preferentially used in brain tissues with those in other tissues, we identified several cis-regulatory elements that were significantly associated with brain-specific poly(A) sites.

Conclusion: Our results indicate that there are systematic differences in poly(A) site usage among human tissues, and both trans-acting factors and cis-regulatory elements may be involved in regulating alternative polyadenylation in different tissues.

Figure 1

Tissue-specific positional preference of poly(A) site usage. (a) Tissue-specific positional preference of type II genes. (b) Tissue-specific positional preference of type III genes. Distance to the median was calculated as (observed usage - median usage)/median usage. Only tissues with significant p values (<0.05, Chi-squared test) are shown here. 2F, 2M, and 2L are the poly(A) sites closest to the 5' end, middle, and closest to the 3' end in a type II gene, respectively. 3U and 3D are poly(A) sites located upstream of the exon closest to the 3' end and poly(A) sites in the exon closest to the 3' end in a type III gene, respectively.

Figure 2

Tissue-specific strong and weak poly(A) site usage. For each tissue, three bars represent the distance to the median usage according to three different cutoffs for the classification of strong and weak sites, for example, 60%, 75% and 90%. For each gene at a given cutoff, the poly(A) site with the percent of supporting ESTs above the cutoff was classified as a strong site. If there was a strong poly(A) site, other sites of the same gene were classified as weak sites. Only values for the weak sites are shown. Significant ones (p value < 0.05, Chi-squared test) are marked with asterisks.

Figure 3

Gene expression of polyadenylation factors. (a) Two-way hierarchical clustering of the U95Av2 data using 21 polyadenylation factors. (b) Two-way hierarchical clustering of the U133A data using 23 polyadenylation factors. See Table 1 for polyadenylation factors in each dataset. Tissues and genes that are consistently clustered together in both datasets are marked by gray lines. (c) Correlation of mRNA expression levels of 21 polyadenylation-related factors across 25 human tissues (upper diagonal, data from U133A; lower diagonal, data from U95Av2). Based on the scale displayed on top of the figure, small squares are colored to represent the extent of correlation between mRNA expression levels of the 21 genes in each pair of human tissues. DRG, dorsal root ganglion.

Figure 4

Boxplots of mRNA expression of several factors in brain and other tissues. (a) PTB, (b) U1A, (c) τCstF-64, (d) PC4, and (e) nPTB. All factors except nPTB were present in both the U95Av2 and U133A datasets. Brain tissues include amygdala, thalamus, caudate nucleus, fetal brain, and whole brain. Expression values lower in brain tissues than other tissues are in green; and those higher in brain tissues are in red.

Figure 5

Brain-specific over-represented cis elements. (a) Schematic of the four poly(A) regions investigated. (b) Identified cis elements in -100/-41. (c) Identified cis elements in -40/-1. (d) Identified cis elements in +1/+40. Logos are shown for cis elements. Under each logo is the percent of hits for the corresponding cis element in poly(A) sites preferred in brain tissues (red), poly(A) sites not preferred in brain tissues (green), and all poly(A) sites (black). In the graphs, the y-axis is the percent of hits, and the x-axis is the distance to the poly(A) site. Horizontal dotted lines are the average value, and vertical dotted lines are the -100, -40, +40, +100 nt positions.