TL;DR
This paper introduces exoRBase 2.0, an expanded database that catalogs mRNA, lncRNA, and circRNA found within extracellular vesicles (EVs) from various human biofluids. This resource aims to facilitate the identification of novel EV-derived RNA signatures for improved disease diagnosis and therapy by providing a unified platform for analyzing and visualizing these crucial biomolecules.
What Problem Does This Paper Address?
Extracellular vesicles (EVs) are tiny, membrane-bound sacs released by virtually all cells, carrying a diverse cargo of molecules, including various RNA species. These EV-derived RNAs (exLRs) are protected from degradation and can reflect the state of their originating cells, making them promising non-invasive biomarkers for diseases. While research has focused on small RNAs within EVs, there's a growing need to comprehensively understand the landscape of longer RNA molecules like mRNAs, lncRNAs, and circRNAs. Existing databases often focus on specific RNA types or a limited range of biofluids. This paper addresses the need for a centralized, comprehensive resource that integrates and analyzes exLR data from multiple human biofluids, enabling a deeper understanding of their functional roles and potential as diagnostic tools.
Methods
The researchers developed an improved exLR-seq analysis pipeline to process RNA-seq data from human biofluids. This pipeline involves:
- Quality Control and Preprocessing: Using FastQC for quality assessment and Trimmomatic for adapter trimming and low-quality read removal.
- Alignment: Aligning reads to the human reference genome (GRCh38) using STAR in a two-pass mapping mode.
- Quantification: Counting reads for mRNA and lncRNA genes using featureCounts and normalizing them to Transcripts Per Million (TPM). CircRNAs were identified and quantified using both ASJA and CIRI2 tools, with CPM (counts per million) used for normalization to account for sequencing depth variations.
- Annotation: Annotating exLRs using GENCODE version 29.
- Tissue Specificity: Calculating tissue specificity scores for mRNA and lncRNA genes based on the GTEx expression atlas.
- Pathway Enrichment Analysis: Employing the ssGSEA method to analyze enrichment scores of 11,536 MSigDB pathways.
- EV Origin Prediction: Using a modified EV-origin approach to predict the relative and absolute fractions of EVs derived from 16 human tissues and 23 blood cell types.
- Statistical Analysis: Utilizing Mann-Whitney U tests for differential analysis between healthy individuals and disease/tumor cohorts, with q-values (Benjamini-Hochberg adjusted) for groups with larger sample sizes.
Datasets
The study integrated a total of 905 RNA-seq data samples from EVs derived from:
- Blood: Samples from healthy individuals and patients with benign diseases, breast cancer (BRCA), coronary heart disease (CHD), colorectal cancer (CRC), esophageal squamous cell carcinoma (ESCC), glioblastoma multiforme (GBM), gastric cancer (GC), kidney cancer (KIRC), hepatocellular carcinoma (HCC), malignant lymphoma (ML), melanoma (MEL), ovarian cancer (OV), pancreatic adenocarcinoma (PAAD), and small cell lung cancer (SCLC).
- Urine: 16 samples.
- Cerebrospinal Fluid (CSF): 5 samples.
- Bile: 17 samples.
Additionally, they incorporated the GTEx project's RNA-seq analysis (V8 release) for annotating potential tissue origins of exLRs.
Key Insights and Ideas
- Unified View of EV Transcriptome: exoRBase 2.0 provides a single platform to explore mRNA, lncRNA, and circRNA within EVs across diverse biofluids, offering a more holistic view than previous, more specialized databases.
- Functional Interpretation: The integration of pathway enrichment analysis allows researchers to move beyond simply identifying exLRs to understanding their potential functional roles in different physiological and pathological states.
- Deciphering EV Heterogeneity: The EV-origin approach is crucial for understanding the complex origins of circulating EVs, helping to distinguish between tissue-specific and blood cell-derived EVs, which is vital for accurate biomarker interpretation.
- Enhanced Biomarker Discovery: By providing comprehensive annotation and expression profiles, the database significantly aids in the identification of novel exLR signatures that can serve as non-invasive biomarkers for early disease detection and therapeutic monitoring.
Key Results
- Expanded Data Content: exoRBase 2.0 contains data from 905 exLR-seq samples, a substantial increase from its predecessor.
- Increased RNA Species: The database now includes 19,643 mRNAs, 15,645 lncRNAs, and 79,084 circRNAs.
- Novel Biofluid Types: The inclusion of urine, CSF, and bile samples expands the scope of biofluids analyzed.
- Comprehensive Pathway and Origin Data: The database incorporates enrichment scores for 11,536 MSigDB pathways and predicted proportions for 39 tissue/blood cell origins.
- User-Friendly Interface: The redesigned web interface offers enhanced browsing, searching, and visualization capabilities, making the data more accessible to researchers.
Limitations and Future Work
- Low Abundance of EVs: The authors acknowledge that exLRs are often present at low levels in biofluids due to the limited abundance of EVs. They address this by providing expression frequencies alongside mean expression values.
- Sample Size for Certain Biofluids: The sample sizes for urine, CSF, and bile are relatively small, and the authors plan to focus on increasing these in future updates.
- Data Quality for Specific Cohorts: For some blood samples (ESCC, GBM, MEL), the low mapped read counts necessitated their use primarily for annotation rather than differential analysis. Future refinements will focus on improving sample selection criteria and potentially data quality.
- Future Expansion: The team plans to regularly update the platform and is considering adding exLR data from other biofluids like ascites, gastric fluid, and saliva. They also encourage researchers to contribute their exLR-seq data.
Summary updated