Evolutionary dynamics of gene and isoform regulation in Mammalian tissues

Abstract

Most mammalian genes produce multiple distinct messenger RNAs through alternative splicing, but the extent of splicing conservation is not clear. To assess tissue-specific transcriptome variation across mammals, we sequenced complementary DNA from nine tissues from four mammals and one bird in biological triplicate, at unprecedented depth. We find that while tissue-specific gene expression programs are largely conserved, alternative splicing is well conserved in only a subset of tissues and is frequently lineage-specific. Thousands of previously unknown, lineage-specific, and conserved alternative exons were identified; widely conserved alternative exons had signatures of binding by MBNL, PTB, RBFOX, STAR, and TIA family splicing factors, implicating them as ancestral mammalian splicing regulators. Our data also indicate that alternative splicing often alters protein phosphorylatability, delimiting the scope of kinase signaling.

Figure 1. Conservation of expression signatures in all tissues and of alternative splicing signatures in some tissues

A) Clustering of samples based on expression values (FPKM) of singleton orthologous genes present in all 5 species (n=7713). Average linkage hierarchical clustering was used with distance between samples measured by the square root of the Jensen-Shannon Divergence (JSD) between the vectors of expression values. B) Clustering of samples based on PSI values of exons in singleton orthologous genes conserved to chicken, with alternative splicing detected in all individuals analyzed (n=489). Clustered as in (A). When the set of genes used in this analysis was clustered by gene expression rather than PSI values, tissue-dominated clustering was observed, as in (A) (fig. S15).

Figure 2. Exonic features associated with evolutionary change in alternative splicing

A) Above: PSI values for eef1d exon 3 across tissues and species analyzed. Below: Eef1d gene expression values. (Mean ± SD of 3 biological replicates). PSI values were calculated only for tissues with FPKM ≥ 5. Inset: exon structure of 5' end of eef1d gene (ENSMUSG00000055762). B) Below: Number of internal exons binned by the age of the inferred alternative splicing based on occurrence in ≥ 2 individuals. Above: The fraction of exons with length divisible by 3 and the mean and SEM of the tissue-specificity. C) Top: mean ± SEM of PSI values of exons binned by the phylogenetic extent of alternative splicing as in (B). Middle: mean ± SEM of 3' splice site scores of exons in each bin. Bottom: mean ± SEM of 5' splice site scores. Splice sites were scored using the MaxEnt model (31). *Indicates t-test p-value < 0.05. **Indicates P < 0.001. D) Fraction of regulatory 5mers in the downstream intron that differed between mouse and rat in exons binned by the phylogenetic extent of alternative splicing as in (B) (Mean ± SEM). *Indicates t-test P < 0.05 (t-test). **P < 0.001.

Figure 3. Tissue-specific regulatory motifs accumulate and are preferentially bound in broadly alternative exons

A) Mean ± SEM of Phastcons scores (using the placental mammals alignment, with mouse coordinates) in exons and flanking introns grouped by phylogenetic pattern of alternative splicing. Splicing pattern indicated by letters adjacent to colored bars, as in Fig. 2B. B) Top ten 5mers in broadly alternative exons relative to constitutive exons ranked by discrimination information (6). C) The conservation of all 5mers (6) compared with their discrimination information. All 5mers with discrimination information ≥ 0.001 bits are highlighted. UUUUU was an outlier in enrichment (0.011 bits) and is not shown. D) Fold enrichment (log₂) relative to constitutive exons in downstream region was plotted for 5mers with discrimination information ≥ 0.001 bits for exons grouped by phylogenetic breadth of alternative splicing. 5mers associated with known splicing regulators are shown in color, with mean of all 5mers in black. E) The fraction of introns containing MBNL1 CLIP-Seq clusters (19) was assessed in introns adjacent to exons with different phylogenetic patterns splicing, as in (A). F) As in (E), but grouped by presence/absence of a MBNL1 motif. The mean fraction ± SEM of 1000 bootstrap samples is shown. *P < 0.01 (binomial test). G) As in (E), but with exons sampled from each set to match the MBNL motif counts in the CQRM set. Mean ± SD of 1000 samplings is shown for the first 3 groups; observed mean is shown for the CQRM set. *P < 0.05, **P < 0.001.

Figure 4. Alternative splicing delimits the scope of protein phosphorylation

A) GO analysis of genes containing tissue-regulated exons whose splicing is conserved. B) Density of Phosphosite phosphorylation sites (top) or Scansite predicted phosphorylation sites (bottom) in exons grouped by alternative splicing status, tissue-specificity and splicing pattern conservation (6). Mean ± SEM is shown. C) Mean Scansite predicted phosphorylation site density in exons grouped by phylogenetic breadth of splicing. D) TJP1 exon 20 splicing has a higher switch score in tissues where Erk1 is expressed above median levels (shaded blue) than where it is expressed below median (shaded pink); KSI is defined as the difference between these switch scores. PSI value not calculated if TJP1 expression fell below a cutoff (e.g., cow spleen). E) Mean KSI values for kinase-exon pairs involving the sets of exons as in (B). Mean ± SEM is shown. Observed values (obs) were compared with controls in which PSI values in different tissues were randomly permuted (ctl). Comparisons marked (*) were significant by Mann-Whitney U test (P < 0.005).