Exome-wide Low-frequency Genetic Variants Contribute to Human Craniofacial Morphology

摘要

The genetic basis of normal-range variation in human facial traits is still poorly understood. Most studies to date have approached facial morphology via univariate measurements with genetic hypotheses centered on common genetic variants. Studies on other complex traits have identified fruitful associations with rare and low-frequency variants involved. To better understand the genomic architecture of facial morphology, we studied the influence of low-frequency coding variants on multi-dimensional facial shape phenotypes. A cohort of 2,329 European individuals were genotyped for approximately 245,000 coding variants on the Illumina Exome v1.2 array. Using three-dimensional facial images, we partitioned the full face into 31 hierarchically arranged modules to model global-to-local features, and generated multi-dimensional phenotypes representing the shape variation within each module. We used multivariate kernel regression (implemented in the MultiSKAT R package) to test the association between the multivariate facial phenotypes and exome-wide variants with frequencies <1% in a gene-based manner. After accounting for multiple tests, we identified eight significant genes (AR, CARS2, FTSJ1, HFE, LOC108783645, LTB4R, TELO2, NEC-TIN 1) as influencing the morphology of the cheek, chin, nose and philtrum region. These genes displayed a wide range of phenotypic effects, with some impacting the full face and others affecting only localized regions. Notably, NECTIN1 is a well-established craniofacial gene that underlies the etiology of a syndromic form of cleft lip and palate (MIM#225060). These results have expanded our understanding of the genetic basis of normal-range human facial morphology by implicating rare and low frequency coding variants in novel candidate genes.