Systematically testing human HMBS missense variants to reveal mechanism and pathogenic variation

Abstract

Defects in hydroxymethylbilane synthase (HMBS) can cause Acute Intermittent Porphyria (AIP), an acute neurological disease. Although sequencing-based diagnosis can be definitive, ~⅓ of clinical HMBS variants are missense variants, and most clinically-reported HMBS missense variants are designated as "variants of uncertain significance" (VUS). Using saturation mutagenesis, en masse selection, and sequencing, we applied a multiplexed validated assay to both the erythroid-specific and ubiquitous isoforms of HMBS, obtaining confident functional impact scores for >84% of all possible amino-acid substitutions. The resulting variant effect maps generally agreed with biochemical expectation. However, the maps showed variants at the dimerization interface to be unexpectedly well tolerated, and suggested residue roles in active site dynamics that were supported by molecular dynamics simulations. Most importantly, these HMBS variant effect maps can help discriminate pathogenic from benign variants, proactively providing evidence even for yet-to-be-observed clinical missense variants.

Figure 1.

Generating and evaluating HMBS variant effect maps (A) Workflow for generating HMBS variant effect maps. (B) Correspondence between erythroid-specific and ubiquitous HMBS isoform functional scores. For reference, null- and WT-like scores are indicated with dashed red or green lines, respectively, while the blue line corresponds to a linear regression fit (R=0.96; P = 2.2×10⁻¹⁶). (C) Distributions of functional impact scores of nonsense (red), synonymous (green), and missense variants (gray) from the combined erythroid-specific and ubiquitous HMBS map (left). (D) Preview of full-sized combined HMBS map

Figure 2.

Identifying patterns of mutational tolerance Functional scores for each possible substituted amino acid (y-axis) at each active-site residue position (x-axis) responsible for: (a) altering cofactor binding; (b) PBG binding for pyrrole chain elongation; (c) hinge flexibility; (d) pyrrole stability; and (e) HMB release. For each substitution, diagonal bar sizes convey estimated measurement error in the corresponding functional score. Box color either indicates the wild-type residue (yellow), a substitution with damaging (blue), tolerated (white), or above-wildtype (‘hyper-complementing’, red) functional score, or missing data (gray).

Figure 3.

The “closed” and “open” conformations of the active site loop in WT HMBS and three HMBS variants – D61N, D61A, and the double mutant D61K K27D. The average distance (Å) between protein position 27 and 61 is shown, along with time (expressed as a percentage) in each conformation.

Figure 4.

Modeling the effects of HMBS missense variants on protein stability and structure (A) Comparison of functional impact scores (black) and predicted free energy change (ΔΔG; red) values of HMBS missense variants. Plotted values are averages within windows of five amino acid (AA) positions. (B) WT (top) and E250R variant (bottom) comparison. The E250R substitution repels R99 and opensa channel, which is exposed to solvent. PBG-G218 interaction is lost and replaced by a salt bridge between PBG and R195, which in turn disrupts D99-pyrrol interactions. For clarity, hydrogen atoms are not shown. Water molecules within 7 Å of E250 (or R250) and Q217 with larger than 50% occupancy are shown. (C) Structural model of HMBS; colored according to the median functionality score of substitutions at each position, along with a wireframe model of the tetrapyrrole (green), and noting residues located at the dimer interface. (D) Median functionality scores of variants at amino acid positions which were: (1) below 20% accessible surface area (ASA); (2) above 40% ASA; (3) at the dimerization interface with a threshold ΔASA of 1.0; and (4) active site residues required for polypyrrole assembly. P values were calculated by Mann-Whitney U test. Thick and thin bars correspond to median, upper, and lower quartiles, respectively. Light green and dark blue dashed lines correspond to WT-like and null-like scores, respectively.

Figure 5.

Performance of variant effect maps in distinguishing pathogenic from benign reference variants. Evaluation of precision (fraction of variants scoring below each threshold functionality score that are in the positive reference set containing pathogenic variants) vs recall (fraction of positive reference variants with functionality scores below threshold). Here, precision has been balanced to reflect performance in a balanced test setting where positive and negative sets contain the same number of variants. Balanced precision-recall curves are shown for erythroid-specific (green line) , ubiquitous (orange) and combined maps (pink). Performance is also described in terms of area under the balanced precision vs recall curve (AUBPRC) and recall at a balanced precision of 90% (R90BP).