Systematic characterization of indel variants using a yeast-based protein folding sensor

Gene variants resulting in insertions or deletions of amino acid residues (indels) have important consequences for evolution and are often linked to disease, yet compared to missense variants the effects of indels are poorly understood and ineptly predicted. To approach this issue, we developed a sensitive protein folding sensor based on complementation of uracil auxotrophy in yeast by circular permutated orotate phosphoribosyltransferase (CPOP). The sensor accurately reports on the folding of disease-linked missense variants and artificially designed proteins. Applying the folding sensor to a saturated library of single amino acid indel variants in human DHFR revealed that most regions which tolerate indels are confined to internal loops and the N- and C-termini. Surprisingly, indels are also allowed at a central α-helix. Several indels are temperature-sensitive and the folding of most indels is rescued upon binding to the competitive DHFR inhibitor methotrexate. Predictions using Rosetta and AlphaFold2 correlate with the observed effects, suggesting that most indels operate by destabilizing the native fold and that these computational tools may be useful for classification of indels observed in population sequencing. The human DHFR sequence was divided into five tiles. For each tile, we generated an indel library comprising all possible single residue deletions as well as insertion of a single glycine residue at each position in that tile. The DHFR sequence outside each tile in the five libraries was wild-type. The DHFR libraries were inserted into the CPOP folding sensor and cloned into pAG415. The libraries were transformed into a ura5Δura10Δ strain. Cells were plated on medium with and without methotrexate at 30 °C and 37 °C to select for folding of DHFR indel variants. Based on the relative change in frequency, we calculated a CPOP folding score of each DHFR indel variant.