Two locus inheritance of non-syndromic midline craniosynostosis via rare SMAD6 and common BMP2 alleles

Andrew T. Timberlake, Jungmin Choi, Samir Zaidi, Qiongshi Lu, Carol Nelson‐Williams, Eric D. Brooks, Kaya Bilgüvar, Irina Tikhonova, Shrikant Mane, Jenny F. Yang, Rajendra Sawh‐Martinez, Sarah Persing, Elizabeth Zellner, Erin Loring, Carolyn Chuang, Amy Galm, Peter W. Hashim, Derek M. Steinbacher, Michael L. DiLuna, Charles C. Duncan, Kevin A. Pelphrey, Hongyu Zhao, John A. Persing, Richard P. LiftoneLife 2016 · DOI

TL;DR

This study reveals that non-syndromic midline craniosynostosis, a condition causing premature fusion of skull bones, is often caused by an epistatic interaction between rare, damaging mutations in the SMAD6 gene and a common genetic variant near the BMP2 gene. This two-locus model explains nearly all the observed phenotypic variation, highlighting how rare and common genetic factors can combine to drive disease.

What Problem Does This Paper Address?

Craniosynostosis, affecting about 1 in 2000 newborns, is the premature fusion of skull sutures, which can lead to brain compression and developmental issues. While syndromic forms of craniosynostosis have well-identified genetic causes, the genetic basis of non-syndromic midline craniosynostosis (affecting the sagittal and metopic sutures) has been elusive. This condition often appears sporadically, making it unclear whether it's due to de novo mutations, incomplete penetrance of rare variants, or complex genetic interactions. The paper aims to identify the genetic underpinnings of this common, non-syndromic form of craniosynostosis.

Methods

The researchers employed whole-exome sequencing on a cohort of 191 probands with non-syndromic midline craniosynostosis, including 132 parent-offspring trios. They used standard bioinformatics pipelines (GATK, TrioDeNovo) for variant calling and annotation. A key method was the use of MetaSVM to predict the functional impact of missense variants, classifying them as damaging (D-mis) or tolerated (T-mis). They also analyzed de novo mutation rates and the burden of rare, damaging variants across genes. To investigate the interaction between SMAD6 and BMP2, they performed genotype analysis for a known BMP2-associated SNP (rs1884302) and conducted transmission disequilibrium tests (TDT) and parametric two-locus linkage analysis.

Datasets

The study utilized exome sequencing data from:

  • 191 probands with non-syndromic midline craniosynostosis.
  • 132 parent-offspring trios from the craniosynostosis cohort.
  • Selected extended family members.
  • Control data from 3337 healthy European parents of children with autism (used for case-control burden analysis).
  • The Exome Aggregation Consortium (ExAC) database for allele frequency information.

No new benchmark datasets were introduced; rather, existing large-scale genomic datasets were leveraged for comparison and analysis.

Key Insights and Ideas

  • Epistasis is Key: The most significant finding is the demonstration of a two-locus epistatic interaction. A rare, damaging mutation in SMAD6 alone has a low penetrance, but when combined with a common risk allele near BMP2, it leads to a near-complete penetrance for craniosynostosis. This highlights how common variants can dramatically influence the risk conferred by rare variants.
  • SMAD6 as a Major Player: SMAD6, an inhibitor of BMP signaling, emerged as a critical gene. Rare damaging mutations in SMAD6 were found to be significantly enriched in affected individuals, both as de novo events and transmitted variants.
  • Incomplete Penetrance Explained: The striking incomplete penetrance of SMAD6 mutations (observed to be <60% in carriers without the BMP2 risk allele) is largely explained by the presence or absence of the common BMP2 risk allele.
  • Mechanism of Action: The study suggests that SMAD6 acts as a haploinsufficient inhibitor of BMP signaling, and increased BMP signaling (potentially driven by the BMP2 variant) accelerates suture closure.
  • Broader Implications: This finding has implications beyond craniosynostosis, suggesting that similar epistatic interactions between rare and common genetic variants may underlie other complex diseases.

Key Results

  • SMAD6 Mutations: Thirteen probands (7%) carried damaging de novo or rare transmitted mutations in SMAD6.
  • Incomplete Penetrance: SMAD6 mutations showed striking incomplete penetrance, with only 24% of transmitting parents (excluding probands) exhibiting craniosynostosis.
  • BMP2 Interaction: In kindreds with SMAD6 mutations, individuals with the BMP2 risk allele (rs1884302) had 100% penetrance for craniosynostosis, while those without the risk allele had only 19% penetrance.
  • Two-Locus Model: A two-locus model combining rare SMAD6 variants and the common BMP2 risk allele provided an extremely strong fit to the data, with a maximum lod score of 7.37 and a likelihood ratio 1410-fold higher than single-locus models.
  • Prevalence: The SMAD6/ BMP2 genotype combination accounts for approximately 7% of non-syndromic midline craniosynostosis cases in this cohort, making it a frequent cause.

Limitations and Future Work

  • Sample Size: While substantial, the cohort size limits the precision of penetrance estimates and the ability to identify very rare genetic variants with lower effect sizes. The authors suggest that larger sample sizes will improve these estimates.
  • Mechanism Elucidation: While the study strongly implicates the SMAD6/BMP2 interaction and proposes a mechanism involving BMP signaling, further studies are needed to precisely delineate how these genotypes cause disease at a molecular level.
  • Other Genes: The authors acknowledge that de novo mutations in other genes likely contribute to a significant portion (~15%) of non-syndromic midline craniosynostosis cases, and suggest that sequencing larger cohorts will be necessary to identify these.
  • Neurodevelopmental Outcomes: The paper notes that some individuals with SMAD6 mutations exhibited developmental delay, suggesting that aberrant BMP signaling might also contribute to neurodevelopmental outcomes independent of suture fusion. This warrants further investigation.
  • Broader Disease Relevance: The authors propose investigating whether similar rare-common variant epistatic interactions are relevant to other complex, unexplained disorders.

Summary updated

Abstract

Premature fusion of the cranial sutures (craniosynostosis), affecting 1 in 2000 newborns, is treated surgically in infancy to prevent adverse neurologic outcomes. To identify mutations contributing to common non-syndromic midline (sagittal and metopic) craniosynostosis, we performed exome sequencing of 132 parent-offspring trios and 59 additional probands. Thirteen probands (7%) had damaging de novo or rare transmitted mutations in <i>SMAD6</i>, an inhibitor of BMP - induced osteoblast differentiation (p<10<sup>-20</sup>). <i>SMAD6</i> mutations nonetheless showed striking incomplete penetrance (<60%). Genotypes of a common variant near <i>BMP2</i> that is strongly associated with midline craniosynostosis explained nearly all the phenotypic variation in these kindreds, with highly significant evidence of genetic interaction between these loci via both association and analysis of linkage. This epistatic interaction of rare and common variants defines the most frequent cause of midline craniosynostosis and has implications for the genetic basis of other diseases.
BibTeX
@inproceedings{timberlake2016two,
  author = {Andrew T. Timberlake and Jungmin Choi and Samir Zaidi and Qiongshi Lu and Carol Nelson‐Williams and Eric D. Brooks and Kaya Bilgüvar and Irina Tikhonova and Shrikant Mane and Jenny F. Yang and Rajendra Sawh‐Martinez and Sarah Persing and Elizabeth Zellner and Erin Loring and Carolyn Chuang and Amy Galm and Peter W. Hashim and Derek M. Steinbacher and Michael L. DiLuna and Charles C. Duncan and Kevin A. Pelphrey and Hongyu Zhao and John A. Persing and Richard P. Lifton},
  title = {{Two locus inheritance of non-syndromic midline craniosynostosis via rare SMAD6 and common BMP2 alleles}},
  booktitle = {eLife},
  year = {2016},
  doi = {10.7554/elife.20125},
}
Figure