Poster #4: Justin Blair MS

Systematic In Silico Mutagenesis of JAG1 Identifies Structural Hotspots Relevant to Missense Variant Interpretation

Justin J. Blair, MS¹; Tristan J. Hayeck, PhD¹; Markos Xenakis, PhD¹; Melissa A. Gilbert, PhD¹ ¹Children’s Hospital of Philadelphia (CHOP), Philadelphia, PA, USA

Clinical variant interpretation for missense variants is challenging for diagnostic labs, particularly when functional data is unavailable. Protein structure-based modeling can estimate the energetic consequences of amino acid substitutions and identify regions intolerant to variation, to augment variant interpretation. For JAG1, a gene encoding a Notch signaling ligand, pathogenic variants are most commonly protein-truncating and cause Alagille syndrome. However, nearly 700 JAG1 missense variants are documented as Variants of Uncertain Significance in ClinVar. To improve interpretation of JAG1 missense variants, we systematically characterized the predicted structural consequences of all missense variation across a structurally modeled region of JAG1 and compared two independent computational approaches. Methods: A three-dimensional model of JAG1 residues 31–333 was generated using AlphaFold for systematic in silico mutagenesis. Amino acid substitutions were evaluated using FoldX and Rosetta to estimate changes in protein stability (ΔΔG). Predictions were summarized by residue to identify mutation-sensitive positions and structural hotspots. Cross-method agreement was assessed using raw and normalized scores and relative variant rankings. Predictions were also compared with known and experimentally characterized variants. Results: Predicted structural effects were nonuniform, with discrete residues demonstrating broad intolerance to substitution. Conserved cysteine and glycine residues were among the most mutation-sensitive positions, with cysteine substitutions predicted to produce particularly large effects, consistent with their roles in disulfide-bonding. FoldX and Rosetta showed overall concordance and were also in agreement with known and experimentally characterized variants. Conclusions: Systematic structure-based mutagenesis provides a residue-level map of predicted mutation intolerance across JAG1, which could support missense variant interpretation. Concordance between independent modeling approaches and known variants supports the utility of structural modeling for variant prioritization, while method-specific discordance supports integrating multiple approaches. This approach can be applied to other disease-associated proteins where structural prediction could provide direct clinical utility in variant interpretation within standardized guidelines.