Functionality of intergenic transcription: an evolutionary comparison

Abstract

Although a large proportion of human transcription occurs outside the boundaries of known genes, the functional significance of this transcription remains unknown. We have compared the expression patterns of known genes as well as intergenic transcripts within the ENCODE regions between humans and chimpanzees in brain, heart, testis, and lymphoblastoid cell lines. We find that intergenic transcripts show patterns of tissue-specific conservation of their expression, which are comparable to exonic transcripts of known genes. This suggests that intergenic transcripts are subject to functional constraints that restrict their rate of evolutionary change as well as putative positive selection to an extent comparable to that of classical protein-coding genes. In brain and testis, we find that part of this intergenic transcription is caused by widespread use of alternative promoters. Further, we find that about half of the expression differences between humans and chimpanzees are due to intergenic transcripts.

Figure 1. Signal Intensities of Exonic and Intergenic Probes

The signal intensity range presented in the main figure covers 95% of all array probes with positive signal intensity. The insert shows the signal intensity range including the additional 4% of array probes. The x-axis shows the number of exonic and intergenic probes. Red indicates the proportion of probes we classified as expressed. Since there exists no empirically established cutoff for classifying tiling array probes into “expressed” and not expressed, we chose an arbitrary cut-off based on both absolute signal intensity and relative expression of perfect match and mismatch probes (Materials and Methods). The distribution presented here is based on an average of the four tissues. Taken separately, all tissues show very similar distributions (Figure S1).

Figure 2. Distribution of Expressed Probes among Exons, Introns, and Intergenic Regions

The distributions of expressed probes among exonic (blue), intronic (gray), and intergenic (red) regions in four tissues in humans and chimpanzees. The size of the circles reflects the number of probes (Table S2). All distributions represent an average of the two DNA strands, measured independently. There were no identifiable differences between expressed probe distributions on the two strands (Figure S2). “Total” indicates the distribution of all array probes, irrespective of their expression levels.

Figure 3. Overlap of Expressed Probes between Species

Shown is the overlap of expressed exonic (left, lighter shades) and intergenic (right, darker shades) probes between humans and chimpanzees in brain (B), lymphoblastoid cell line (C), heart (H) and testis (T) for the positive (left panel) and negative (right panel) chromosome strands. The horizontal lines inside the bars show the overlap expected by chance. The error bars represent 95% confidence intervals based on bootstrapping of 1,000 subsets of exonic and intergenic probes having the same number of probes from each category and the same signal intensity distribution.

Figure 4. Schematic Representation of Expression Diversity and Divergence in Humans and Chimpanzees in the Three Tissues

The expression was measured using either “classical” transcript-based arrays (upper row) or exonic probes on the tiling arrays (lower row). The trees are inferred from the mean of the squared difference of expression intensities of all detected probe sets [16] or all expressed exonic probes (Materials and Methods). Greater variation observed within species for the exonic probe expression is likely due to greater technical variation associated with tiling arrays measurements, caused by probe design limitations, and by the fact that tiling array measurements are based on signal probe intensity and not on the cumulative intensity of a set of probes.

Figure 5. Expression Divergence and Divergence-to-Diversity Ratio in the Three Tissues for Exonic and Intergenic Probes

Shown are the average expression divergence (A) and divergence-to-diversity ratio (B) between humans and chimpanzees in brain (yellow), heart (blue), and testis (red) for exonic and intergenic probes. The colored areas indicate 95% confidence intervals based on bootstrapping 1,000 subsets of exonic and intergenic probes, having the same number of probes from each category and the same signal intensity distribution. The darker shades indicate expression from the positive DNA strand, while the lighter shades indicate expression from the negative DNA strand. The symbols represent the mean value for each tissue on either the positive (▵) or the negative (○) strand, respectively.

Figure 6. Correlation between Signal Intensity Difference of Intergenic Probes and that of the Nearest Exon

Letters and colors indicate tissues (B, yellow—brain; C, green—cell line; H, blue—heart; T, red—testis). Correlation was calculated separately for probes located upstream (5′) or downstream (3′) from the nearest exon. The width of the bars is proportional to the number of the ENCODE regions showing significant correlations (Spearman correlation test, p < 0.05, corrected for multiple testing). The mean of the bars shows the mean correlation coefficient, while the bar borders represent a 75% confidence interval. The error bars depict a 95% confidence interval of the correlation coefficient, calculated by bootstrapping the list of intergenic probes within each region 500 times (Materials and Methods).

Figure 7. Proportions of Exonic and Intergenic Probes with Significant Expression Difference between Humans and Chimpanzees

The colors indicate the proportions of exonic (blue) and intergenic (red) probes. The figure represents an average of the probe numbers identified as differently expressed on the positive and on the negative strands. The size of the circles reflects the number of probes.