What’s in a face?

This is a guest post by Professor Martyn Cobourne.  We let him choose a paper to review and he has reviewed the first basic science paper  ever for this blog.  Martyn was brought up in the same Worcester country village as me, but we never met (I have illustrated this blog with the view over our village).  It is clear that Martyn is very clever and he has done a great job of bringing some proper science to our blog.

I want to thank Padhraig Fleming and Kevin O’Brien for inviting me to write a guest blog post on a paper of my choosing. I have decided to discuss a manuscript recently deposited onto BioRxiv, an open-access preprint repository for the biological sciences, which allows authors to post their work and offers the opportunity for online comment and discussion. This paper has not, therefore been formally peer-reviewed. Still, it represents a fascinating large-scale international collaborative effort investigating the genetic basis of human facial variation, and a contributing author is Stephen Richmond, Professor of Orthodontics at Cardiff University.

Introduction

The human face is a complex structure that demonstrates almost infinite variation while conforming to the same basic organisational plan. Understanding how complex morphological structures such as the face are established during development remains a significant challenge in biology, and it will come as no surprise to orthodontists that genetics plays a vital role in this process. Over the last thirty years, many significant genes have been identified primarily through the analysis of mutant mice and linkage studies of Mendelian disorders in humans. However, the role of minor genetic variants, epigenetic influences and environmental factors has been more challenging to elucidate, mainly because it requires investigation at the population level. It is only relatively recently that the population-based genome-wide profiling techniques that are needed have become established. In this study, the authors have combined highly accurate phenotyping of facial variation with some of these sophisticated genetic analyses using several accessible international data-sets to identify regions of the genome associated with the facial variation.

What did they do?

The study utilised three-dimensional (3D) facial surface scans and genomic information derived from over 8000 individuals within three independent studies, two based in the USA and one found here in the UK – the Avon Longitudinal Study of Parents and their Children (ALSPAC) in Bristol. They began by mapping the facial scans of these individuals in great detail and using hierarchical clustering, were able to group regions of the faces into 63 correlated segments. They were then able to identify the significant phenotypic variation that existed in each facial segment and investigate patterns of association with this variation across the genomes of these individuals.

What did they find?

They found two key features associated with these global association patterns. First, the highest correlations were between segments of the same facial quadrant (lips, nose, upper face, etc.). Second, relationships between groups of segments from different quadrants also existed with some of these seeming to reflect shared embryologic origins of the segment groups (for example, the nose and upper face). A meta-analysis confirmed 203 genomic regions associated with the normal-range facial variation, and 103 of these were novel. What is interesting is that many genes in close approximation with these regions were highly enriched for processes and phenotypes associated with craniofacial development in both humans and mice. The top human phenotype was oro-facial clefting, which suggests a considerable overlap between genes involved in normal facial variation and those implicated in one of the most frequent human facial anomalies.

Moreover, many of these genes encode members of signalling pathways that are known to be involved in craniofacial development. This includes members of the WNT and TGF pathways. Some of these genes also have an essential role in limb development, which given the known associations between craniofacial and limb anomalies in many human syndromes, is perhaps not surprising.

What are the likely cell types and embryonic processes that these genomic regions influence? To answer this question, they investigated how these regions influence transcriptional activation in multiple different cell types, finding the strongest associations with cranial neural crest cells and dissected embryonic craniofacial tissues. These observations suggest that there is an embryonic origin for human facial variation that has influence at multiple time-points in development. In addition, they found particular enrichment of these genomic regions in enhancer regions – areas of the genome that regulate gene transcription, suggesting that genetic facial variation is significantly influenced by enhancer activity.

What about the known and novel genomic regions that were identified? Sixty-four of these regions harboured known craniofacial genes previously associated with human or animal craniofacial malformations, but not with normal facial variation – one of these being MSX1, a well-known cleft-causing gene in both humans and mice. However, fifty-three regions were associated with genes that had no previous association with craniofacial development, one exciting example being DACT1, which encodes an inhibitor of WNT signalling and seems to be particularly associated with variation in chin morphology.

Finally, these investigators hypothesised that because the human face demonstrates such complexity, it is likely that groups of genomic regions may also contribute to the same trait. Using structural equation modelling, a technique that uses multivariate analysis, they identified four genomic region combinations with significant pairwise interaction in specific facial regions. One of these was between PRDM16 and GLI3 – and associated with morphology of the premaxillary region. Interestingly, both these genes encode proteins involved in the Hedgehog signalling, a pathway that has strong associations with craniofacial development. Indeed, GLI3 mutations in humans and mice cause Greig cephalopolysyndactyly syndrome, a condition associated with abnormal facial morphology, including broadening of the nasal bridge.

What did I think?

This is a significant international study that has used complex phenotypic and genetic analysis to identify genomic regions of individuals with European ancestry that are associated with the normal facial variation. It has used data from several longitudinal studies, including one from the UK and will form a useful springboard for further investigations into this fascinating area of biology. It goes beyond the analysis of major loci that cause significant defects and begins to unravel the contribution of minor traits in normal variation. It is a slightly different study to those normally discussed on this blog site. Still, I thought it might be of interest to the orthodontic community, and it is great to see the involvement of an orthodontist!