Genotype-phenotype correlations in PAX9 mutations

The phenotypes of missense mutations in human PAX9 reflect subtle differences in the types and numbers of missing teeth. While they clearly affect molar development, some affect premolar and incisor development. Objective: The goals of this study were to correlate data from our structural and functional analyses of the mutant proteins with the phenotypes for the missense mutations. Methods: The number and types of missing teeth were noted for the six mutations. Using the crystal structure of the PAX6 paired domain bound with DNA, predictions of the structural consequences on the paired domain were made. A series of recombinant plasmids encoding the mutations were first generated. Next, we performed electromobility shift assays (EMSA) in order to determine effects on DNA-binding, immunocytochemistry to characterize the subcellular localization of the mutant proteins, and co-immunoprecipitation to ascertain effects on protein-protein interactions. Results: Sequence alignment analysis revealed that each mutation affects invariant amino acid residues in the paired domains of the Pax family in humans and across species. Our evaluation of the DNA-binding capacities of the proteins revealed that while the L21P, R26W, and R28P are unable to bind DNA, G51S, I87F, and K91G are able to form a complex with DNA. All of the proteins are stably expressed in mammalian cells 24 hours after culture, localize primarily in the nucleus and retain their ability to heterodimerize with Msx1. Conclusions: Our results suggest the less severe phenotypes are associated with those mutations which were able to bind to DNA and support the notion that the proteins may be the product of a hypomorphic allele, rather than an amorph. The lack of observable defects for three of the mutant proteins requires additional functional studies to determine the underlying pathogenic mechanisms. This research was supported by NIH grant R01 DE13668 to RDS, NIH K08-DE14237 to HK.