Functional enrichment of alternative splicing events with NEASE reveals insights into tissue identity and diseases

Abstract

Alternative splicing (AS) is an important aspect of gene regulation. Nevertheless, its role in molecular processes and pathobiology is far from understood. A roadblock is that tools for the functional analysis of AS-set events are lacking. To mitigate this, we developed NEASE, a tool integrating pathways with structural annotations of protein-protein interactions to functionally characterize AS events. We show in four application cases how NEASE can identify pathways contributing to tissue identity and cell type development, and how it highlights splicing-related biomarkers. With a unique view on AS, NEASE generates unique and meaningful biological insights complementary to classical pathways analysis.

Keywords: Alternative splicing; Differential splicing; Dilated cardiomyopathy; Disease pathways; Functional enrichment; Multiple sclerosis; Platelet activation; Protein-protein interactions; Systems biology.

Fig. 1

Overview of NEASE. A Annotated exons are mapped to Pfam domains, motifs, and residues. The joint graph of PPIs, DDIs, DMIs, and co-resolved structure is used to identify the interactions mediated by these features. B For a list of exons/events, NEASE identifies interactions mediated by the spliced protein features and pathways that are significantly affected by those interactions. C NEASE provides a corrected p value, in addition to an enrichment score (NEASE Score) for every pathway (see the “Methods” section). The user can further focus on an individual pathway, where NEASE can prioritize genes and find new biomarkers. In this example, the gene G3 was not part of the enriched pathway A but it has the largest number of affected interactions with genes from the pathway

Fig. 2

Analysis of tissue-specific exons. A Heatmap and hierarchical clustering of standardized PSI values obtained from VastDB. The heatmap only shows events with a standard deviation of PSI values ≥ 20. The heatmap shows that clusters of exons upregulated in neural tissues and muscle/heart tissues are dominant (clusters C1 and C3). B NEASE analysis shows that 28% and 27% for both neural and muscle upregulated exons, respectively, are encoding protein features: domains, linear motifs, and residues. For these subgroups of events, the exact protein complexes involved can be identified, and NEASE enrichment can be performed. C, D Comparison between gene-level enrichment and NEASE enrichment for the two sets of exons

Fig. 3

NEASE visually highlights the impact of the AS regulation at the interactome level. The gray nodes represent proteins from the pathway and the red nodes represent genes with AS events. Red edges represent the affected interactions for the nodes with known DDIs, DMIs, or co-resolved structures. The visualization of the pathway “Synaptic vesicle cycle” from the KEGG database for the exons upregulated in the neural tissues shows that the splicing in the genes CLTA and CLTB is co-regulated and affects the interactions of the same complex. Similarly, NEASE highlights the importance of the domain ATP6V0A1 which is upregulated in neural tissues and binds seven proteins from the “Synaptic vesicle cycle” pathway

Fig. 4

A 15 % of differentially spliced exons, between reticulated and mature platelets, are known to encode protein features. For this subset of exons, NEASE enrichment can be performed. B Gene level enrichment of all differentially spliced exons in the Reactome database fails to capture the most relevant pathways. C In contrast, NEASE shows an enrichment of the GPCR downstream signaling and other related pathways that are well known to be important in platelet activation. D A further look at the genes driving the enrichment of the GPCR pathway shows the most relevant genes affected by AS