Global analysis of alternative splicing differences between humans and chimpanzees

Abstract

Alternative splicing is a powerful mechanism affording extensive proteomic and regulatory diversity from a limited repertoire of genes. However, the extent to which alternative splicing has contributed to the evolution of primate species-specific characteristics has not been assessed previously. Using comparative genomics and quantitative microarray profiling, we performed the first global analysis of alternative splicing differences between humans and chimpanzees. Surprisingly, 6%-8% of profiled orthologous exons display pronounced splicing level differences in the corresponding tissues from the two species. Little overlap is observed between the genes associated with alternative splicing differences and the genes that display steady-state transcript level differences, indicating that these layers of regulation have evolved rapidly to affect distinct subsets of genes in humans and chimpanzees. The alternative splicing differences we detected are predicted to affect diverse functions including gene expression, signal transduction, cell death, immune defense, and susceptibility to diseases. Differences in expression at the protein level of the major splice variant of Glutathione S-transferase omega-2 (GSTO2), which functions in the protection against oxidative stress and is associated with human aging-related diseases, suggests that this enzyme is less active in human cells compared with chimpanzee cells. The results of this study thus support an important role for alternative splicing in establishing differences between humans and chimpanzees.

Figure 1.

Strategies used to identify alternative splicing differences between humans and chimpanzees. (Left panel) In the comparative genomics strategy, regions including orthologous human and chimpanzee exons (red boxes) and the flanking 150 nucleotides of intron sequence were aligned. Nucleotide substitutions (indicated by Xs) in these regions were scored, and events displaying >5% substitution rates were analyzed for alternative splicing differences by RT–PCR assays using poly(A)⁺ RNA from human and chimpanzee tissues (see main text for details). (Right panel) In the quantitative alternative splicing microarray profiling strategy, labeled cDNA from the same poly(A)⁺ RNA samples were hybridized to an alternative splicing microarray designed to monitor inclusion levels of cassette-type alternative exons. Each alternative splicing event is monitored by a set of six oligonucleotide probes (black horizontal lines) (Pan et al. 2004). Predictions for alternative splicing differences between the corresponding human and chimpanzee tissues were validated by RT–PCR assays (see Table 1; Figs. 4, 5; Supplementary Fig. 1).

Figure 2.

Quantitative alternative splicing microarray profiling reveals the extent of alternative splicing differences between human and chimpanzee orthologous exons, as well as the degree of divergence between splicing patterns over different evolutionary time periods. (A) Color spectrum plots indicating the number and magnitude of alternative splicing differences between human and chimpanzee frontal cortex and heart tissues. The Y-axes indicate the number of alternative splicing events profiled; these are sorted according to the magnitude of the absolute value of the percent exon inclusion level (%in) difference between the human and chimpanzee tissue being compared. The magnitude of the percentage inclusion difference is indicated by the color scale on the right. (B) Cumulative distribution plot displaying the distribution of percentage inclusion differences when comparing microarray data for 217 conserved alternative splicing events between the following pairs of tissues: human and chimpanzee frontal cortex (blue line), human and chimpanzee heart (red line), human frontal cortex and mouse cortex (green line), and human and mouse heart (purple line). (C) Spearman correlation coefficients are shown for pairwise comparisons between alternative splicing levels (black numbers) and transcript levels (purple numbers) for the set of 217 orthologous genes analyzed in human, chimpanzee, and mouse tissues in B. Double arrows indicate the pairs of species compared. (Hs) Homo sapiens; (Pt) Pan troglodytes; (Mm) Mus musculus.

Figure 3.

Transcript and AS level differences between humans and chimpanzees involve largely distinct subsets of genes. The color spectrum plots compare splicing level differences and transcript level differences for the same set of genes expressed in the frontal cortex and heart. For each tissue comparison, the left column shows splicing differences (measured as the magnitude of percent inclusion difference, columns 1 and 3) and the right column measures differences in gene transcript level (measured as the magnitude of the hyperbolic arcsine [arcsinh] difference, columns 2 and 4). An hyperbolic arcsine difference of ∼0.4 corresponds to a 1.5-fold change in expression level, and a difference of ∼0.7 corresponds to a twofold difference in expression level.

Figure 4.

Examples of alternative splicing differences between humans and chimpanzees confirmed by semiquantitative RT–PCR and sequencing. RT–PCR assays were performed using primers specific for sequences in constitutive exons flanking the alternative exons predicted to be differentially spliced by the comparative genomic or alternative splicing microarray profiling analyses. Corresponding tissues from human (Hs) and chimpanzee (Pt) are indicated. Major splice isoforms that include and skip alternative exons (black boxes) are indicated on the right of each panel. Diagrams below each gel illustrate the predicted consequence of alternative splicing changes at the protein and/or transcript levels for TAF6, ADD3, the SR-repeat family protein splicing factor of 40 kDa (SFRS5/SRp40) and GSTO2 (refer to Table 1 and main text for details). Protein domains are labeled as follows: (H) head domain; (N) neck domain; (C) C-terminal tail region known to interact with spectrin; (RRM1) RNA recognition motif 1; (RRM2) RNA recognition motif 2; (RS) arginine/serine-rich domain; (GST) glutathione S-transferase domain. The stop sign indicates the insertion of a PTC. The asterisk indicates an additional TAF6 splice isoform detected in human but not chimpanzee tissue, as confirmed by sequencing.

Figure 5.

Comparisons of AS levels in RNA samples from macaque, human, and chimpanzee individuals. (A) RT–PCR assays were performed using rhesus macaque total RNA from frontal cortex and heart samples from a number of different individuals. RT–PCR primers were designed to anneal to the exons neighboring each alternative exon, resulting in the amplification of two products (the isoforms including and skipping the alternative exon, as indicated in each of the panels). Asterisks denote novel isoforms observed primarily in one species but not the other. The transcripts shown correspond to the ADD3 (top panel), GSTO2 (second panel), HVEM/TNFRS14 (third panel), and ARL3 (bottom panel) genes. The macaque is used as an outgroup to define the ancestral splicing pattern for each gene. Examples of chimpanzee lineage-specific splicing differences (ADD3 and ARL3 transcripts) and human lineage-specific splicing differences (GSTO2 and TNFRSF14) are shown (see also Supplementary Table 3 for additional examples). (B) RT–PCR experiments were performed using total RNA from each of the individual samples that were pooled for analysis in the comparative genomic and microarray experiments. The same alternative splicing events displayed for macaque experiments in A are also shown in B for comparison. The labels for each gel lane represent the macaque, human, and chimpanzee individuals listed in Supplementary Table 2.

Figure 6.

Increased skipping of exon 4 in GSTO2 transcripts reduces overall levels of functional GSTO2 protein in humans compared with chimpanzees. (A) RT–PCR experiments were performed to measure transcript abundance of individual myc-tagged GSTO2 isoforms transfected into HeLa cells. Primers were designed to anneal to exons neighboring the alternative exon in order to amplify isoforms including and skipping exon 4. β-Actin transcript levels were measured to normalize for input of total RNA between samples. (B) Western blotting experiments on the same transfected samples from A. Anti-c-myc antibodies were used to detect protein expression of individual GSTO2 splice variants. α-Tubulin protein levels were measured to control for loading input. (C) Measure of the relative protein/mRNA abundance for each of the transfected samples in A and B. Measurements represent the average of three independent transfection experiments, and standard deviations are shown. (D) Western blotting using samples from human and chimpanzee lymphoblastoid cells and antibodies to detect endogenous GSTO2. α-Tubulin protein levels were measured to control for loading input.