Integrative omics analysis reveals relationships of genes with synthetic lethal interactions through a pan-cancer analysis

Li Guo, Sunjing Li, Bowen Qian, You-quan Wang, Rui Duan, Wenwen Jiang, Yihao Kang, Yuyang Dou, Guowei Yang, Lulu Shen, Jun Wang, Tingming LiangComputational 2020 · DOI

TL;DR

This study integrates multi-omics data to map synthetic lethal interactions across various cancer types, revealing key gene relationships and their prognostic value. The findings highlight the potential of synthetic lethality as a therapeutic strategy and identify novel gene-miRNA interactions relevant to cancer.

What Problem Does This Paper Address?

Cancer treatment often faces challenges with drug resistance, especially in metastatic cases. Synthetic lethality, a concept where the simultaneous disruption of two genes leads to cell death while disrupting either gene alone is viable, has emerged as a promising therapeutic avenue. However, identifying and understanding these interactions, particularly across different species and molecular levels, is complex. This paper aims to bridge this gap by performing an integrative pan-cancer analysis to uncover the distribution, relationships, and potential clinical utility of synthetic lethal gene pairs.

Methods

The researchers employed an integrative omics approach, combining data from multiple sources and molecular levels. Key methods include:

  • Phylogenetic Conservation: Leveraging conserved synthetic lethal interactions between yeast and humans to identify candidate gene pairs.
  • Database Integration: Utilizing the SynLethDB database for human synthetic lethal interactions.
  • Pan-Cancer Analysis: Analyzing mutation and gene expression data across a wide range of cancer types from The Cancer Genome Atlas (TCGA).
  • Functional Enrichment Analysis: Using DAVID to understand the biological roles and pathways associated with identified genes.
  • Survival Analysis: Assessing the prognostic value of gene pairs by correlating their mutation and expression patterns with patient survival.
  • miRNA-mRNA Interaction Analysis: Investigating the regulatory roles of microRNAs (miRNAs) at the isomiR level, focusing on their interactions with synthetic lethal genes.
  • Network Construction: Building interaction networks (gene-gene and miRNA-mRNA) to visualize relationships and identify key players.
  • Randomization Tests: Used to statistically validate the significance of observed prognostic values.

Datasets

  • SynLethDB database: A resource for human synthetic lethal interactions.
  • The Cancer Genome Atlas (TCGA): Provided mutation data, gene expression profiles, and clinical data for various cancer types.
  • InParanoid 6: Used for identifying orthologous genes between yeast and humans based on phylogenetic conservation.
  • Genomics of Drug Sensitivity in Cancer (GDSC) database: Used for correlating gene pairs with drug responses.
  • starBase v2.0: Used for identifying miRNA-mRNA interactions.

Key Insights and Ideas

  • Cross-species conservation is a starting point, but human data is crucial: While conserved synthetic lethal interactions from yeast provide a foundation, further filtering based on human mutation and expression data is essential for identifying clinically relevant pairs. Many yeast-derived predictions were filtered out, suggesting that direct human data is more predictive for cancer applications.
  • Synthetic lethal genes are often involved in core cellular processes and cancer hallmarks: A significant portion of identified genes are linked to hallmarks of cancer, core essential genes, and have known roles in tumorigenesis, underscoring their importance.
  • Gene expression patterns are strong prognostic indicators: Gene pairs with dysregulated expression patterns showed a higher prognostic value than those with mutations alone, suggesting that expression levels might be more directly indicative of disease progression and therapeutic response.
  • miRNAs play a complex regulatory role: The study highlights the intricate network of miRNA-mRNA interactions, suggesting that these small regulatory RNAs can influence synthetic lethal interactions and contribute to cancer pathology.

Key Results

  • A total of 37,588 candidate gene pairs involving 7,816 genes were collected.
  • 49.74% of genes had 2-10 interactions, indicating a complex interaction landscape.
  • 22.93% of genes were involved in hallmarks of cancer, and 21.61% were identified as core essential genes.
  • Gene pairs with dysregulated expression patterns demonstrated higher prognostic values compared to random pairs, validated by randomization tests (p < 0.001 for two groups, p = 0.012 for three groups).
  • The miRNA-mRNA interaction network revealed that many synthetic lethal genes contributed to critical pathways like the cell cycle (seven of 12 genes) and cancer pathways (five of 12 genes).

Limitations and Future Work

  • IsomiR complexity: While the study analyzed miRNA-mRNA interactions at the isomiR level, it primarily focused on the most dominant isomiR. The authors acknowledge that other isomiRs can also be unexpectedly dominant and warrant further investigation.
  • Yeast predictions: The study found that many predicted gene pairs from yeast were filtered out. Future work could explore why these conserved interactions don't translate as effectively to human cancer contexts.
  • Further validation: The authors suggest that further systematic studies with more candidate data are needed to fully explore the potential applications of synthetic lethality in personalized anticancer therapeutics.
  • Combined therapies: The identified gene pairs and their interactions could serve as a basis for developing novel combined therapeutic strategies.

Summary updated

Abstract

Synthetic lethality is thought to play an important role in anticancer therapies. Herein, to understand the potential distributions and relationships between synthetic lethal interactions between genes, especially for pairs deriving from different sources, we performed an integrative analysis of genes at multiple molecular levels. Based on inter-species phylogenetic conservation of synthetic lethal interactions, gene pairs from yeast and humans were analyzed; a total of 37,588 candidate gene pairs containing 7,816 genes were collected. Of these, 49.74% of genes had 2-10 interactions, 22.93% were involved in hallmarks of cancer, and 21.61% were identified as core essential genes. Many genes were shown to have important biological roles via functional enrichment analysis, and 65 were identified as potentially crucial in the pathophysiology of cancer. Gene pairs with dysregulated expression patterns had higher prognostic values. Further screening based on mutation and expression levels showed that remaining gene pairs were mainly derived from human predicted or validated pairs, while most predicted pairs from yeast were filtered from analysis. Genes with synthetic lethality were further analyzed with their interactive microRNAs (miRNAs) at the isomiR level which have been widely studied as negatively regulatory molecules. The miRNA-mRNA interaction network revealed that many synthetic lethal genes contributed to the cell cycle (seven of 12 genes), cancer pathways (five of 12 genes), oocyte meiosis, the p53 signaling pathway, and hallmarks of cancer. Our study contributes to the understanding of synthetic lethal interactions and promotes the application of genetic interactions in further cancer precision medicine.
BibTeX
@inproceedings{guo2020integrative,
  author = {Li Guo and Sunjing Li and Bowen Qian and You-quan Wang and Rui Duan and Wenwen Jiang and Yihao Kang and Yuyang Dou and Guowei Yang and Lulu Shen and Jun Wang and Tingming Liang},
  title = {{Integrative omics analysis reveals relationships of genes with synthetic lethal interactions through a pan-cancer analysis}},
  booktitle = {Computational},
  year = {2020},
  doi = {10.1016/j.csbj.2020.10.015},
}
Figure