78. NETWORK BIOLOGY ALGORITHMS IDENTIFY BIOLOGICAL PATHWAYS UNDERLYING CIGARETTE SMOKING BEHAVIORS

• Published in: European neuropsychopharmacology Vol. 75; pp. S98 - S99
• Main Authors: Sullivan, Kyle, Miller, J. Izaak, Kainer, David, Lane, Matthew, Cashman, Mikaela, Garvin, Michael R., Townsend, Alice, Kruse, Peter, Quach, Bryan C., Willis, Caryn, Xu, Ke, Aouzierat, Bradley E., Johnson, Eric O., Hancock, Dana B., Jacobson, Daniel
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2023
• DOI: 10.1016/j.euroneuro.2023.08.183
• ISSN: 0924-977X

Genetic risk variants in nicotinic acetylcholine receptor subunit genes (e.g., CHRNA5-A3-Br) are well-known for contributing to cigarette smoking behaviors. Recent genome-wide association studies (GWAS) from the GWAS and Sequencing Consortium of Alcohol and Nicotine Use (GSCAN) have identified hundreds of other genome-wide significant loci contributing to the heritability of cigarette smoking behaviors. However, the biological pathways reflected by the full repertoire of cigarette smoking-associated variants are poorly understood. To better understand the biological pathways underlying smoking behaviors, we applied the network biology algorithms GRIN and Functional Partitioning. We used publicly available GSCAN GWAS summary statistics collected from European ancestry subjects for two traits: Smoking Initiation (SmkInit; 805,431 individuals) and Cigarettes Per Day (CigsPerDay; 326,497 individuals). We assigned genome-wide significant (p < 5e-8) single nucleotide polymorphisms (SNPs) to genes using 5 different methods: SNP-nearest gene, conventional MAGMA, and H-MAGMA using Hi-C data collected from dorsolateral prefrontal cortex (dlPFC), cortical neurons, or midbrain dopaminergic neurons. Using a multiplex network of 10 gene-gene network layers from distinct types of biological experimental evidence including two dlPFC-specific layers, we used GRIN to remove false positive genes based on the premise that correctly assigned genes would be highly interconnected in the networks, as determined by the random walk with restart (RWR) network traversal algorithm. We then used Functional Partitioning to identify functional groupings of smoking-associated genes based on RWR-based network topology, followed by gene set enrichment of biological pathways. We assigned 526 unique genes targeted by SNPs associated with CigsPerDay. Of these, GRIN retained 235 genes based on high network interconnectivity. We assigned 3013 unique genes targeted by SmkInit-associated SNPs, and GRIN retained 1329 highly interconnected genes. Using Functional Partitioning, we identified chromatin regulation as a common pathway involved in both smoking traits based on shared genes (e.g., NFAT5) and genes unique to each trait (SmkInit: DNMT3B, NFKB2; CigsPerDay: CYP2A6, DBH). Using the Functional Partitioning groupings, we also identified pathway enrichments for “regulatory T-cell differentiation” from the CigsPerDay genes, and “regulation of DNA-templated transcription” from the SmkInit genes. We identified both shared and unique genes and biological pathways associated with significant SNPs from cigarette smoking behaviors. Understanding which genes underlie heritability of cigarette smoking behaviors can identify druggable gene targets that may inform follow-up drug repurposing efforts to treat nicotine addiction.