Fig. 5: A gradient of functional effects underlies phenotypic differences of VHL variants.
From: Saturation genome editing maps the functional spectrum of pathogenic VHL alleles
a, SNVs (n = 797) in exons 2 and 3 were assayed in HAP1–HIF1A–KO cells. Compared to previous data (top), variants were well-tolerated, independent of consequence. b, Function scores across isogenic lines showed no significant correlation (Spearman’s ρ = −0.06, P = 0.08). c, ‘Pathogenic’ and ‘likely pathogenic’ SNVs from ClinVar were grouped based on annotations in VHLdb21. SNVs associated only with type 1 VHL disease or ccRCC were deemed ‘type 1’ (n = 74 SNVs), whereas SNVs associated only with type 2 disease or predominantly pheochromocytoma were deemed ‘pheo-predominant’ (n = 29 SNVs, excluding SNVs associated with type 2B disease). The remaining pathogenic SNVs lacked unambiguous phenotypic data in VHLdb (n = 64 SNVs, ‘type unclear’). The boxplot shows function scores for SNVs in these categories, as well as for n = 2,033 other SNVs assayed, including variants either not deemed pathogenic in ClinVar or absent (one-way ANOVA, adjusted P = 0.00043 between ‘pheo-predominant’ and ‘type unclear’; ****P < 1.0 × 10−10 for all other comparisons; boxplot: center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; all points shown). d, Patients in the Freiburg VHL Registry with missense variants assayed by SGE (n = 321) were grouped based on the function class of their germline variant. Due to the high prevalence of p.Y98H in this cohort, an additional LOF2 group excluding p.Y98H was analyzed, as well. A Kaplan–Meier estimator was used to assess age-related ccRCC penetrance (log-rank test for significance).
