Neighbours matter – local DNA sequence context may explain recurrent mutations in calmodulinopathy
S Noe Niikanoff Christiansen, L C Crotti, P J S Schwartz, S B Besenbacher, M N Nyegaard, M T O OvergaardAbstract
Background
CALM1, CALM2, and CALM3 encode an identical calmodulin protein that is strictly evolutionary conserved among all vertebrates. Missense variants in any of these genes causes calmodulinopathy, a disease characterised by ventricular arrhythmias and sudden cardiac death at a young age. Due to the evolutionary conservation of calmodulin, most observed missense variants are unique. Yet, the recurrence of certain variants, including the N98S missense mutation, is surprisingly high.
Purpose
To understand the cause of recurrent missense mutations in calmodulinopathy, we examined the link between neighboring nucleotides around a given variant and the predicted probability of de novo missense mutations.
Methods
We examined predicted mutation rates for all possible single-base missense variants in CALM1, CALM2, and CALM3 using three context-dependent mutation models for prediction of mutation rates and compared those with observed missense variants in the International Calmodulinopathy Registry.
Results
Among the 62 distinct missense variants in the International Calmodulinopathy Registry, ten were observed in more than one gene. The most frequent variant, N98S, was observed in 22 of 97 index patients (22.7%) and observed in CALM1 (11/35) and CALM2 (11/46), but not in CALM3 (0/16). The predicted mutation rate for N98S was consistently higher in CALM1 and CALM2 compared with CALM3 (3.5-, 3.6-, and 4.1-fold, respectively) across all three mutation models. Furthermore, the predicted mutation rate for N98S was substantially higher than for any other observed de novo variants with median fold increases of 5.3 (range: 1.9-18.7), 6.5 (range: 2.2-34.0), and 6.7 (range 2.0-19.4) across the three models.
Conclusions
Recurrent clinical observations of N98S in CALM1 and CALM2 but not in CALM3 can partially be explained by differences in predicted mutation rates driven by the local DNA sequence. Mapping predicted mutation rates across these genes can identify regions with fewer missense variants than expected, thereby suggesting functional importance and increased pathogenicity. Incorporating this information into variant interpretation may improve clinical risk assessment and allow a better tailored preventive strategy.