A computational report on Fibrocystin structure via AlphaFold2: considerations and limitations for ARPKD research

Autosomal Recessive Polycystic Kidney Disease (ARPKD) is a rare, genetic disease that often occurs during early childhood and can be lethal at its most severe presentation. ARPKD is caused by biallelic variants in PKHD1 which encodes Fibrocystin, a large receptor protein. The disease presentation and progression of ARPKD is highly variable, the reason for this is currently unknown. To address this, AlphaFold2 was employed to generate a model of Fibrocystin which was used to establish spatial relationships between compound heterozygous pairs in ARPKD patients. AlphaFold2 was used to generate a model of full-length human Fibrocystin which was assessed for structural similarities to known protein structures. Compound heterozygous variant pairs and disease outcomes were acquired from previously published work and annotated onto the three-dimensional Fibrocystin model to identify spatial relationships. AlphaFold2 predicted the full-length of Fibrocystin with varying confidence. Globally, average modelling confidence was low, with more confident modelling around the G8 and PBH1 domains. Confidence was lowest around the transmembrane region and intracellular tail, which modelled poorly. FoldSeek identified structural similarities to other proteins; however, these had similar structures to those already predicted present in Fibrocystin. No new structural elements were detected. Spatial relationships between compound heterozygous variants did not correlate with disease outcomes. There are limitations to AlphaFold2 which limits its ability to accurately model a protein like Fibrocystin. These limitations may hinder the tools application in determining genotype-phenotype relationships in ARPKD. Currently, Fibrocystin models produced by AlphaFold2 are unable to identify genotype-phenotype relationships.