Meta-analysis Reveals Genome-Wide Significance at 15q13 for Nonsyndromic Clefting of Both the Lip and the Palate, and Functional Analyses Implicate GREM1 As a Plausible Causative Gene

Abstract

Nonsyndromic orofacial clefts are common birth defects with multifactorial etiology. The most common type is cleft lip, which occurs with or without cleft palate (nsCLP and nsCLO, respectively). Although genetic components play an important role in nsCLP, the genetic factors that predispose to palate involvement are largely unknown. In this study, we carried out a meta-analysis on genetic and clinical data from three large cohorts and identified strong association between a region on chromosome 15q13 and nsCLP (P = 8.13 × 10(-14) for rs1258763; relative risk (RR): 1.46, 95% confidence interval (CI): 1.32-1.61)) but not nsCLO (P = 0.27; RR: 1.09 (0.94-1.27)). The 5 kb region of strongest association maps downstream of Gremlin-1 (GREM1), which encodes a secreted antagonist of the BMP4 pathway. We show during mouse embryogenesis, Grem1 is expressed in the developing lip and soft palate but not in the hard palate. This is consistent with genotype-phenotype correlations between rs1258763 and a specific nsCLP subphenotype, since a more than two-fold increase in risk was observed in patients displaying clefts of both the lip and soft palate but who had an intact hard palate (RR: 3.76, CI: 1.47-9.61, Pdiff<0.05). While we did not find lip or palate defects in Grem1-deficient mice, wild type embryonic palatal shelves developed divergent shapes when cultured in the presence of ectopic Grem1 protein (P = 0.0014). The present study identified a non-coding region at 15q13 as the second, genome-wide significant locus specific for nsCLP, after 13q31. Moreover, our data suggest that the closely located GREM1 gene contributes to a rare clinical nsCLP entity. This entity specifically involves abnormalities of the lip and soft palate, which develop at different time-points and in separate anatomical regions.

Fig 1. Regional association plot for the 15q13 region.

P-values for SNPs at 15q13 that were analyzed as part of the Ludwig 2012 meta-analysis (P_{nsCL/P_meta}) are plotted against their chromosomal position (hg19). Full data are provided in S1 Dataset. For each variant, color code denotes linkage disequilibrium to rs1258763, based on 1000genomes. After combination with data from replication I and II, the top variant rs1258763 (purple diamond; indicated by dotted line) reaches genome-wide significance. Plot was generated using LocusZoom [28].

Fig 2. Regional association plots for the 15q13 region in different types of nsCL/P.

In the imputed data of the Central European cohort, the 15q13 region was analyzed in the overall phenotype nsCL/P (A) and both subphenotypes, i.e. nsCLP (B) and nsCLO (C). For each SNP, the P-value is plotted against its chromosomal position (hg19). In nsCL/P and nsCLP, a highly associated cluster of SNPs in strong linkage disequilibrium is present, located between GREM1 and FMN1. The lowest P-value was observed for rs2600520 (purple diamond). In each panel, the top genotyped variant rs1258763 is marked by an open circle. For all other variants, color code denotes linkage disequilibrium to rs2600520, based on 1000genomes. Regional association plots were generated using LocusZoom [28].

Fig 3. Forest plot for association of rs1258763 and nsCL/P subphenotypes.

Subphenotype analyses of cleft lip and palate (nsCLP, black) and cleft lip only (nsCLO, grey) were conducted in the Ludwig 2012 meta-analysis data, the replication I and II cohorts, and in the combined analysis of the present study. Boxes represent point estimates of the relative risk for each of the four studies, with box sizes scaled according to the number of affected individuals. Lines indicate the extent of the confidence interval. These data illustrate the consistent association between rs1258763 and nsCLP across the various studies, and the presence of a narrow effect size range in the combined cohort. Please note that confidence intervals for nsCLO are larger due to the lower number of nsCLO patients. Informed by the specific expression of Grem1 in lip and soft palate development in the mouse embryo (see below) we also analyzed the effect size of the particular soft palate subphenotype (nsCLP_soft). The arrow indicates the point estimate for rs1258763 and nsCLP_soft.

Fig 4. Grem1 expression during mouse craniofacial development.

(A-E) Expression of Grem1 is visualized by X-Gal staining of heterozygous Grem1^LacZ whole mount embryos. (A) At E11.5, Grem1 is expressed in the dorsal part of the lateral nasal prominence (lnp). Stippled lines demarcate the nasal pits. (B) At E12.5, Grem1-positive domains are also detectable in the merging zones (arrowheads) of medial nasal prominences (mnp) and maxillary prominences (mxp). (C-G) Secondary palate development. (C) At E13.5, Grem1-positive domains are observed in the forming soft palate (sp). (D) At E14.5, the hard palate (hp) has formed while the Grem1-expressing shelves of the soft palate are not yet fused. (E) At E15.5, the soft palate has fused and Grem1 expression extends posterior to the pharynx (ph). Note the sharp anterior boundary of Grem1 expression in the soft palate (arrowheads). (F, G) Sections of whole mount stained embryos. (F) Cross section at the level indicated in (D) showing that Grem1 expression is restricted to the mesenchyme. (G) Cross section at the level indicated in (E) showing Grem1 expression in the soft palate, which separates the nasopharynx (np) from the oral cavity (oc). Additional abbreviations: a, anterior; l, lateral; m, medial; mdp, mandibular prominence; p, posterior. Scale bars: 500μm.

Fig 5. Analyses of the effect of Grem1 loss of function and of ectopic Grem1 protein on secondary palate development.

(A-D) Hematoxilin/Eosin staining of paraffin sections. (A,B) At the level of the posterior hard palate, the palatal shelves have fused and epithelial rests (arrowheads) are seen in both Grem1^+/+ (A) and Grem1^-/-(B) embryos at E14.5. At E15.5, the soft palate has fused in both Grem1^+/- (C) and Grem1^-/- embryos (D). (E-H) Organ culture experiments of secondary palatal shelves dissected at E13.5. (E) The area (A) between palatal shelves was measured at the onset and after 48 hours of culture. The presence of Grem1 protein led to an increase in the area (F), whereas the size of the area did not change in controls (G). Stippled lines demarcate the medial edges of the secondary shelves at the onset of culture. (H) The difference in area (A_δ = A_48hours−A_0hours) is significantly larger in the Grem1-treated palatal shelves compared to those of controls. Abbreviations: a, anterior; hp, hard palate; l, lateral; m, medial; np, nasopharynx; oc, oral cavity; p, posterior; sp, soft palate. Scale bars: 200μm in A-D, 500μm in E-G.