Landscapes of missense variant impact for human superoxide dismutase 1

Abstract

Amyotrophic lateral sclerosis (ALS) is a progressive motor neuron disease for which important subtypes are caused by variation in superoxide dismutase 1 (SOD1). Diagnosis based on SOD1 sequencing can not only be definitive but can also indicate specific therapies available for SOD1-associated ALS (SOD1-ALS). Unfortunately, SOD1-ALS diagnosis is limited by the fact that a substantial fraction (currently 26%) of ClinVar SOD1 missense variants are classified as "variants of uncertain significance" (VUSs). Although functional assays can provide strong evidence for clinical variant interpretation, SOD1 assay validation is challenging given the current incomplete and controversial understanding of SOD1-ALS disease mechanism. Using saturation mutagenesis and multiplexed cell-based assays, we measured the functional impact of over 2,000 SOD1 amino acid substitutions on both enzymatic function and protein abundance. The resulting "missense variant-effect maps" not only reflect prior biochemical knowledge of SOD1 but also provide sequence-structure-function insights. Importantly, our variant-abundance assay can discriminate pathogenic missense variation and provides new evidence for 41% of missense variants that had been previously reported as VUSs, offering the potential to identify additional people who would benefit from therapy approved for SOD1-ALS.

Keywords: ALS; Lou Gehrig disease; MAVE; SOD1; amyotrophic lateral sclerosis; deep mutational scanning; multiplexed assay of variant effect; superoxide dismutase; variant classification; variant-effect mapping.

Figure 1

Generation of SOD1 missense variant-effect maps (A) Overview of process to generate variant-effect maps. (B) SOD1 variant-effect maps measuring total enzymatic activity (top) and abundance (bottom). Box color indicates the WT residue (yellow); a substitution with damaging (blue), tolerated (white), or above-WT (“hyper”; red) functional score; or missing data (gray). Consensus tracks summarize the map scores by position. β strands are indicated by arrows, and α helices by loops.

Figure 2

Comparison between activity and abundance scores (A) Scatterplot representing the combination of activity and abundance score for each variant, divided into quadrants (defined by scoring either above or below a score of 0.5, the midpoint between nonsense and synonymous scores) as follows: neutral in both maps (white), only deleterious in the abundance map (teal), deleterious in both maps (blue), and only deleterious in the activity map (purple). (B) Structure of SOD1 (PDB:1HL5), with Cu²⁺ (orange)/Zn²⁺ (blue-gray) ions, in which each residue is “painted” according to the dominant quadrant for substitutions at that position. Positions Val32 and Arg80, which did not show a dominant quadrant type, were painted gray. (C) Map of missense variants painted by quadrant type for variants measured in both maps. Secondary structure is indicated by arrows for β strands and loops for α helices. The WT residue is shown in yellow, while gray indicates that the substitution was not well measured in at least one of the maps.

Figure 3

Modeling the effects of SOD1 missense variants on protein structure (A) Distributions of median functional scores (where each median is of missense substitutions at each residue position) are shown for total enzymatic activity (left) and abundance (right) maps, summarizing sets of positions where the WT residue is “buried” (below 20 Ų accessible surface area [ASA]), “metal-binding,” “surface” (above 35 Ų of ASA), and “interface” (at the homodimerization interface). Boxes correspond to interquartile range, and bold bars indicate medians of the positional medians. Whiskers correspond to minima and maxima. p values were calculated by Mann-Whitney U test, ∗∗∗∗p < 0.0001, ∗∗p < 0.001. (B) Structural model of SOD1, colored according to the median functional score of substitutions at each position, with positions tolerant to substitutions in white and positions intolerant to substitution in blue. (C) Venn diagram indicating whether amino acids found at SOD1 interfaces (either the homodimeric SOD1 interface or the SOD1-CCS interface with and without Zn²⁺ bound to SOD1) show for each map whether the variant was tolerant (white) or intolerant (blue) to substitutions. See material and methods for details on determination of threshold scores.

Figure 4

Genotype-phenotype analyses of SOD1 variants in both assays against a dataset of various phenotypes Scores for both activity and abundance maps (mean score for substitutions at each position) were compared with an assembled dataset phenotypes from people with ALS, including (A) protein half-life measurements (in mice), (B) enzyme specific activity, (C) protein abundance measurements, (D) age of ALS onset, and (E) disease duration. NS, not significant. Gray indicates 95% confidence interval for significant correlations. See material and methods for details on regression and significance calculations.

Figure 5

The SOD1 abundance map best distinguishes positive from negative reference variants and provides evidence for 41% of VUSs (A) Evaluation of precision (fraction of variants in the positive reference set that scored below each threshold functional score) vs. recall (fraction of positive reference variants with functional scores below threshold). More specifically, we used balanced precision such that precision values reflect performance in a setting where positive and negative sets contain the same number of variants for reference sets with variants obtained from Labcorp/gnomAD. Balanced precision vs. recall curves are shown for the SOD1 total enzymatic activity (purple) and abundance maps (black), as well as for the best-performing computational predictor (CPT; turquoise). Performance for three additional predictors is shown in Figure S13. Performance was summarized in terms of area under the balanced precision vs. recall curve (AUBPRC) and recall at a balanced precision of 80% (R80BP). Plot indicates the number of variants in the positive (P/LP), gnomAD negative (PB), and Labcorp negative (B) reference sets used here (see material and methods for details). (B) Currently clinically annotated variants and new evidence proposed based on calibrating our abundance map scores against the Labcorp Genetics reference sets. See main text and material and methods for details on the derivation of evidence strength.