(626.6) Use Of Weighted Gene Coexpression Network Analysis To Identify Connectivity Between Gut And Brain Gene Expression

Poster Board Number: E679

Tasnin Khan (David Geffen School of Medicine, UCLA), Asa Hatami (David Geffen School of Medicine, UCLA), Chunni Zhu (David Geffen School of Medicine, UCLA), Riki Kawaguchi (David Geffen School of Medicine, UCLA, David Geffen School of Medicine, UCLA), Swapna Joshi (David Geffen School of Medicine, UCLA), Han Chen (David Geffen School of Medicine, UCLA), Jill Hoffman (David Geffen School of Medicine, UCLA), Ivy Law (David Geffen School of Medicine, UCLA), Carl Rankin (David Geffen School of Medicine, UCLA), Varghese John (David Geffen School of Medicine, UCLA), Daniel Geschwind (David Geffen School of Medicine, UCLA, David Geffen School of Medicine, UCLA, David Geffen School of Medicine, UCLA), Charalabos Pothoulakis (David Geffen School of Medicine, UCLA), Elizabeth Videlock (David Geffen School of Medicine, UCLA)

Background: Alterations in the gut brain axis are being recognized as a pathogenic factor for an increasing number of diseases, including neurodegenerative diseases such as Parkinson Disease (PD). The extent to which gene expression profiling in the gut and the brain is co-regulated is not well understood. Weighted gene coexpression network analysis (WGCNA) uses unbiased hierarchical clustering to reduce gene expression profiling data to modules of highly correlated genes. 
Hypothesis: Gut-brain connectivity will be strongest for modules related to physiologic mechanisms with a systemic component, such as immune reactivity.

Methods: WGNCA networks were constructed for distal colon and striatum RNA seq data from mice overexpressing human wild type alpha synuclein (ASO, n=18) and wild type (WT, n=16) mice at 1 and 3 months. Mice with matched colon and striatum data (n=10/6 ASO/WT) were included in this analysis. Linear regression identified associations between colon and striatum modules. For this analysis, we controlled for both age and genotype as genotype- and age-associated modules could be correlated due to these covariates alone. Colon-striatum intermodular connectivity was visualized in Cytoscape. Overrepresented gene ontology (GO) terms in WGCNA modules were determined using the hypergeometric function in the GOstats package in R. Enrichment for gene signatures from single cell sequencing data was determined using the hypergeometric test against cell type signatures from the Panglao and Cellmarker databases.

Results: Selected associations are shown in Figure 1. There are strong correlations between colon and striatum modules that are enriched for terms related to the immune response. Modules most clearly related to the immune response are highlighted in yellow, but several other modules are also closely related to the immune response. For example, there is an association between the striatum module enriched for endothelial cells of the blood brain barrier, and the colon module enriched for goblet cells which produce the mucus layer of the epithelium, a key component of the innate immune system. 

Conclusions: Through analysis of matched colon and striatum samples, we can see correlations between gene expression in the colon and striatum. Most modules in this gut-brain network have some relevance to the immune system which is not unexpected. These results demonstrate the feasibility of “profiling” the gut-brain axis. A larger sample size would permit evaluation of disease-related changes in this profile.

This work was supported by a UCLA CURE Digestive Diseases Research Center (P30DK041301) Pilot and Feasibility Study Award to EJV, a Rapid Pilot Award from the UCLA Claude Pepper Older Americans Independence Center (P30AG028748) to EJV and the Vatche and Tamar Manoukian Division of Digestive Diseases at UCLA. EJV also received support from the NIH Loan Repayment Program (L30DK106759) and the UCLA Gastroenterology Training Grant (T32DK07180).



Figure 1: Correlations between colon and striatum modules. Colon and striatum modules are indicated by figures of mouse colon/brain. Red and blue arrows are positive/negative associations. Thickness is proportional to effect size. Only associations with adjusted p<0.05 are show. The direction of the arrow indicates the dependent variable in the regression model (double-headed arrows mean that both models were significant). Modules enriched for immune-related terms are shaded yellow for emphasis.