Association of low-frequency genetic variants in regulatory regions with nonsyndromic orofacial clefts

Abstract

Genome-wide scans have shown that common risk alleles for orofacial clefts (OFC) tend to be located in noncoding regulatory elements and cumulatively explain only part of the heritability of OFCs. Low-frequency variants may account for some of the "missing" heritability. Therefore, we scanned low-frequency variants located within putative craniofacial enhancers to identify novel OFC risk variants and implicate new regulatory elements in OFC pathogenesis. Analyses were performed in a multiethnic sample of 1,995 cases of cleft lip with or without cleft palate (CL/P), 221 cases with cleft palate (CP) only, and 1,576 unaffected controls. One hundred and nineteen putative craniofacial enhancers identified from ChIP-Seq studies in craniofacial tissues or cell lines contained multiple low-frequency (0.01-1%) variants, which we genotyped in participants using a custom Illumina panel. Two complementary statistical approaches, sequence kernel association test and combined multivariate and collapsing, were used to test association of the aggregated low-frequency variants across each enhancer region with CL/P and CP. We discovered a significant association between CP and a branchial arch enhancer near FOXP1 (mm60; p-value = .0002). Additionally, we observed a suggestive association between CL/P and a forebrain enhancer near FOXE1 (hs1717; p-value = .001). These findings suggest that low-frequency variants in craniofacial enhancer regions contribute to the complex etiology of nonsyndromic OFCs.

Keywords: cleft lip; cleft palate; genetic association; orofacial cleft.

Figure 1:. Results of genetic association tests:

Quantile-quantile plots depict the observed log10-transformed p-values (y-axis) vs. the expected distribution of p-values (x-axis) under the null hypothesis of no association. Each point represents the evidence of genetic association for a specific craniofacial enhancer. The top panels show results for CP and the bottom panels show results for CL/P. Left panels show results of CMC scans, and right panels show results of SKAT scans. Horizontal dashed lines represent the threshold for suggestive association (p-value < 0.004). The horizontal dotted line represents the Bonferroni threshold for statistical significance.

Figure 2:. Genomic position and activity localization of associated enhancers:

(A) Position of the chromosome 3p13 craniofacial enhancer, mm60, and possible cis-regulatory target genes MITF and FOXP1. Chromatin states in normal human cell lines (dermal endothelium [HMEC], skeletal muscle [HSMM], epidermal keratinocytes [NHEK], and lung fibroblasts [NHLF]) are colored-coded with yellow and orange segments representing putative enhancer regions. The associated craniofacial enhancer, mm60, is active in the branchial arches at E11.5, and does not show enhancer chromatin states in cell lines. Other enhancers near FOXP1 (hs864, hs965) not associated with OFCs are active in the developing heart and limb, consistent with the expression pattern of FOXP1. (B) Position of the chromosome 9q22.33 craniofacial enhancer, hs1717, and possible cis-regulatory target gene FOXE1. The hs1717 enhancer can be subdivided into a smaller enhancer (hs1597), each showing activity in the forebrain. Enhancer activity of hs1717 is consistent with chromatin states in human cell lines.