Quantitative approaches to DNA-protein interactions using synthetic biology, high-throughout yeast one-hybrid assays and structural modelling of prote

PAX6 is a transcription factor, which binds to the genome via its paired and homeo DNA binding domains. The primary sequence of PAX6 is highly conserved throughout evolution. It functions as a master regulator of eye development. The gene encoding PAX6 is mutated in a variety of human eye malformations. Inactivation of one copy of the PAX6 gene in humans result in classical aniridia; typified by absence of the iris and visual impairment as a result of a developmental anomalies of the retina, lens and cornea. Multiple different causative missense mutations have been identified in the paired domain in individuals with classical aniridia. No definitely pathogenic missense variants have been identified affecting the homeo domain. It thus seems likely that interaction of paired domain with DNA is responsible for most of the developmental function of PAX6.
Our understanding of how PAX6 functions during development has been hampered by our limited knowledge of where it binds in the genome. Few bone fide PAX6 binding sites have been defined and those that have show remarkable diversity in the sequence of the binding site making it difficult to define a canonical motif.
Aims
This project will employ yeast one-hybrid (Y1H) technology (PMID: 22884952), which allows massively parallel analysis of the binding of different proteins to a defined DNA sequence. Within MRC HGU the Kudla lab use synthetic biology technology to construct libraries of plasmids encoding saturated amino acid substitutions in any peptide. The applicant will be involved in the creation of a paired domain library of all possible amino acid substitutions at each residue. Y1H will then be used to test the binding of all ~2600 variants of the PAX6 paired domain to three validated vertebrate PAX6 binding sites (PMID 12710953). These Y1H experiments will generate a quantitative measurement assumed to correlate with the DNA binding potential of each variant domain.
The computational component of this project will attempt to define the structural basis of the Y1H output. This will use the existing crystal structure of the PAX6 paired domain bound to DNA as a backbone to model the alterations in the quaternary structure for each mutation and reconcile this with the experimental measurement of paired domain function. For each of the three binding sites tested, four groups of variants in the paired domain will be of specific interest:
Those with DNA binding activity that is stronger that wild type (super-activators)
Those with reduced but detectable binding (weak activators)
Missense variants associated with human disease (pathogenic missense)
Missense variants found in the normal population (benign missense).
Using computational modelling and building on recent work (PMID:26490019) we will also explore the intriguing possibility that it is subtle distortions of the classic DNA helical shape rather than or in addition to base sequence that is recognised by the PAX6 paired domain.
We consider that this exciting project will allow us to identify structural elements in both the protein and DNA that mediate the affinity of DNA-protein interaction. Specifically, this project will improve:
Our understanding of the grammar of DNA-paired domain interactions
The differentiation of disease-causing and benign mutations in patients
The recognition of functional PAX6 binding sites in the human genome.