Identification of genes in lipid metabolism associated with white matter features in preterm infants

Abstract

BackgroundWe have previously identified an association between lipids and features of diffusion tensor imaging in preterm infants, highlighting the KEGG peroxisome proliferator-activated receptor pathway (PPAR) as the most highly ranked in association with a white matter phenotype. Here we applied a penalised linear regression model—the graph-guided group lasso (GGGL)—to select single nucleotide polymorphisms within functionally related genes associated with the imaging trait. The GGGL model utilises network-based prior knowledge to improve the selection of variables of interest. MethodsImages acquired with a 3T MR scanner were collected along with saliva for 72 preterm infants (mean gestational age 28 weeks [+4 days], mean postmenstrual age at scan 40 weeks [+3 days]). Salivary DNA was extracted and genotyped with HumanOmniExpress-12 arrays (Illumina, San Diego, CA, USA). Fractional anisotropy maps were constructed from diffusion tensor imaging, and tract-based spatial statistics were used to obtain a group white matter skeleton varying with degree of prematurity, adjusting for age at scan. GGGL was applied to the genes in the PPAR pathway. FindingsThe GGGL method selected five of the 69 genes in the PPAR pathway, and these were functionally related in terms of Gene Ontology Biological Process and linearly correlated with white matter fractional anisotropy (AQP7, ME1, PLIN1, SLC27A1, and ACAA1). Analysis of transcriptional regulation with the PASTAA algorithm indicated that ACAA1, AQP7, ME1, and SLC27A1 were jointly regulated by the EGR4 transcription factor (adjusted p=7·7×10−4). InterpretationOur GGGL analysis of genes in the PPAR pathway identified five highly ranked genes involved in neuronal growth, myelinogenesis, and nervous system response to inflammation, and jointly transcriptionally regulated. Since preterm infants are at increased risk of mental illness and cardiovascular morbidity in later life, this work uses an integrative data-driven strategy to propose a unifying mechanism for these systemic effects, suggesting a promising avenue for intervention. FundingMedical Research Council, National Institute for Health Research Biomedical Research Centre.