RBPWorld for exploring functions and disease associations of RNA-binding proteins across species

Abstract

RNA-binding proteins (RBPs) play key roles in a wide range of physiological and pathological processes. To facilitate the investigation of RBP functions and disease associations, we updated the EuRBPDB and renamed it as RBPWorld (http://research.gzsys.org.cn/rbpworld/#/home). Leveraging 998 RNA-binding domains (RBDs) and 87 RNA-binding Proteome (RBPome) datasets, we successfully identified 1 393 413 RBPs from 445 species, including 3030 human RBPs (hRBPs). RBPWorld includes primary RNA targets of diverse hRBPs, as well as potential downstream regulatory pathways and alternative splicing patterns governed by various hRBPs. These insights were derived from analyses of 1515 crosslinking immunoprecipitation-seq datasets and 616 RNA-seq datasets from cells with hRBP gene knockdown or knockout. Furthermore, we systematically identified 929 RBPs with multi-functions, including acting as metabolic enzymes and transcription factors. RBPWorld includes 838 disease-associated hRBPs and 970 hRBPs that interact with 12 disease-causing RNA viruses. This provision allows users to explore the regulatory roles of hRBPs within the context of diseases. Finally, we developed an intuitive interface for RBPWorld, facilitating users easily access all the included data. We believe that RBPWorld will be a valuable resource in advancing our understanding of the biological roles of RBPs across different species.

Graphical Abstract

Figure 1.

Refined RBP identification procedures enable comprehensive and high-confidence integration of RBP data across 445 species. (A and B) Bar charts illustrating the number of RBPs for the Top 9 species of animals (A) and Top 10 plants (B) collected in RBPWorld. (C and D) Bubble plots showing correlation between genomic complexity (chromosome ploidy, number of chromosomes, number of protein-coding gene and genome size) of animal species (C) or plants species (D) and the corresponding number of RBPs. Statistics were calculated as Pearson correlation coefficient. (E) Comparison of the number of RBPs collected by RBPWorld with those in other databases. (F) Density statistics of the number of hRBP homologous genes in all animals. Horizontal coordinates represent the number of homologous genes, while vertical coordinates indicate the density distribution. The total area under the curve is 1. (G) Venn plot showing distribution of RBPs identified through RBPome technology or RBD prediction. RBP, RNA-binding protein; RBPome, RNA-binding proteome.

Figure 2.

Exploring primary RNA targets of hRBPs and evolutionary conservation of RBP families. (A) A pie chart showing the distribution of primary RNA target types bound by RBPs that tend to bind specific RNA targets. (B) Statistics of RBP family numbers in different species. (C) Bar plot showing the number of RBPs with same RBD (RBP family) from RBPWorld with top 20 members. (D) Stacked bar chart showing the proportion of primary RNA targets of the top 15 RBP family members. (E) A pie chart showing the distribution of primary RNA target types bound by vRBPs that tend to bind specific RNA targets. (F) Heatmaps showing the Reactome pathway enrichment analysis result (left) and complex enrichment analysis result (right) of vRBPs. q-value of different pathway/complex were -log₁₀ transformed. RBD, RNA-binding domain; vRBP, hRBPs that interacting with RNA virus.

Figure 3.

hRBPs perform diverse and multifaceted roles in cellular processes. (A) Pie charts showing the distribution of RBPs with multi-functional roles beyond binding RNA (up) and those capable of forming diverse types of condensates (down). (B) Bar charts showing the distribution of mfRBPs with only one function alone or two (up), as well as the distribution of cRBPs with varying numbers of condensate sublocations (down). (C) Distribution of hRBPs with or without an RBD (canonic or non-canonic) in mfRBPs (left) and cRBPs (right). (D) The top five pathways from the Reactome pathway enrichment analysis for different classification of mfRBPs (left) and cRBPs (right). q-value of different pathway/complex were -log₁₀ transformed. (E) The composition of different kinds of mfRBPs and cRBPs binding to specific RNA types. hRBP, human RBP; mfRBP, multi-functional RBP; cRBP, condensate-associated RBP; E3, RNA-binding E3 ligases; DRBP, DNA- and RNA-binding protein; DUB, RNA-binding deubiquitinating enzymes; Transmembrane protein, RNA-binding transmembrane protein; Metabolic enzyme, RNA-binding metabolic enzyme; Kinase, RNA-binding kinase.

Figure 4.

Impact of perturbation of genes encoding RBPs on gene expression profiles and AS events. (A) Boxplot exhibiting the number of differentially expressed genes (DEGs) upon 512 perturbed RBPs binding to specific RNA types. (B) Pie chart showing the distribution of primary RNA targets for RBPs with over 2000 DAS events, also known as hyper-DAS RBPs, following perturbations of genes encoding for RBPs. (C) Sashimi plot illustrating a significantly altered AS event involving exon skipping upon RBP knockdown, with a P-value of 0.00201. The IncLevelDifference value is −0.432. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5.

hRBPs involved in Mendelian and somatic genetic diseases. (A) The overlap of genes encoding canonical RBPs (left), non-canonical RBPs (middle) and hRBPs (right) with somatic mutant genes causing disease and genes involved in Mendelian genetic disorders. Disease-associated hRBPs were annotated with Mendelian and somatic disease associations extracted from the Open Targets platform. (B) Heatmap depicting the compositional differences between sets of RBPs associated with Mendelian mutation diseases and those linked to somatic mutation disorders, highlighting differences in RBP types, moonlighting functions and primary RNA targets. (C) Therapeutic areas of disease-associated RBPs. Disease mutations with an association score >0.2 from Open Targets were summed for selected therapeutic areas for RBPs. The accumulated association scores were normalized for the amounts of protein in each category. The top 10 therapeutic areas are shown for each disease category, highlighting the highest association scores for Mendelian and somatic mutation diseases. Additional information on methods is available at http://www.hentze.embl.de/public/hRBPdiseases. (D) ‘Summary’ section screenshot of NPM1 gene from RBPWorld.