Resolving candidate genes of mouse skeletal muscle QTL via RNA-Seq and expression network analyses

BMC Genomics

Research Article

David A Blizard¹  David J Vandenbergh²  Jonathan MJ Derry³  Arimantas Lionikas⁴  Aivaras Ratkevicius⁴  Andrew M Carroll⁴  Caroline Meharg⁵ 

[1] College of Health and Human Development, The Pennsylvania State University, 16802, University Park, PA, USA;Department of Biobehavioral Health, 16802, University Park, PA, USA;Penn State Institute for the Neurosciences, 16802, University Park, PA, USA;Sage Bionetworks, 98109, Seattle, WA, USA;School of Medical Sciences, University of Aberdeen, AB25 2ZD, Aberdeen, UK;School of Medicine & Dentistry, University of Aberdeen, AB25 2ZD, Aberdeen, UK;Bioinformatics group, Max Planck Institute for Biology of Aging, D-50931, Köln, Germany;

关键词: Functional genomics;    QTL;    Skeletal muscle;    Gene expression;   

DOI  :  10.1186/1471-2164-13-592

 received in 2012-04-20, accepted in 2012-10-26,  发布年份 2012

来源: Springer

【 摘 要 】

BackgroundWe have recently identified a number of Quantitative Trait Loci (QTL) contributing to the 2-fold muscle weight difference between the LG/J and SM/J mouse strains and refined their confidence intervals. To facilitate nomination of the candidate genes responsible for these differences we examined the transcriptome of the tibialis anterior (TA) muscle of each strain by RNA-Seq.Results13,726 genes were expressed in mouse skeletal muscle. Intersection of a set of 1061 differentially expressed transcripts with a mouse muscle Bayesian Network identified a coherent set of differentially expressed genes that we term the LG/J and SM/J Regulatory Network (LSRN). The integration of the QTL, transcriptome and the network analyses identified eight key drivers of the LSRN (Kdr, Plbd1, Mgp, Fah, Prss23, 2310014F06Rik, Grtp1, Stk10) residing within five QTL regions, which were either polymorphic or differentially expressed between the two strains and are strong candidates for quantitative trait genes (QTGs) underlying muscle mass. The insight gained from network analysis including the ability to make testable predictions is illustrated by annotating the LSRN with knowledge-based signatures and showing that the SM/J state of the network corresponds to a more oxidative state. We validated this prediction by NADH tetrazolium reductase staining in the TA muscle revealing higher oxidative potential of the SM/J compared to the LG/J strain (p<0.03).ConclusionThus, integration of fine resolution QTL mapping, RNA-Seq transcriptome information and mouse muscle Bayesian Network analysis provides a novel and unbiased strategy for nomination of muscle QTGs.

【 授权许可】

CC BY   
© Lionikas et al.; licensee BioMed Central Ltd. 2012