Recurrent rhinovirus infections in a child with inherited MDA5 deficiency

Lamborn, Ian T.; Jing, Huie; Zhang, Yu; Drutman, Scott B.; Abbott, Jordan K.; Munir, Shirin; Bade, Sangeeta; Murdock, Heardley M.; Santos, Celia P.; Brock, Linda G.; Masutani, Evan; Fordjour, Emmanuel Y.; McElwee, Joshua J.; Hughes, Jason D.; Nichols, Dave P.; Belkadi, Aziz; Oler, Andrew J.; Happel, Corinne S.; Matthews, Helen F.; Abel, Laurent; Collins, Peter L.; Subbarao, Kanta; Gelfand, Erwin W.; Ciancanelli, Michael J.; Casanova, Jean-Laurent; Su, Helen C.

To explore further whether the IFIH1 variant found in our patient was disease-causing, we analyzed the mutation burden at this locus in the general population. The ExAC database contains 63 loss-of-function (nonsense, splice acceptor/donor, frameshift) and 392 nonsynonymous missense IFIH1 variants. Of the loss-of-function variants, five had MAF of 0.065% or higher, with homozygotes reported for most of these variants. Homozygotes for the three most frequent loss-of-function variants (rs35732034, rs35337543, and rs35744605) were observed at slightly higher than expected, frequencies in the Icelandic population (Sulem et al., 2015) . However, the earliest death of these homozygotes was recorded at 81 yr of age, and there were no childhood deaths from the offspring of heterozygous couples, which are predicted to be 25% homozygotes. No data were available regarding the early medical history of these homozygous individuals, including incidence of childhood HRV or other respiratory virus infections. Given these preexisting data, we determined whether nonsynonymous missense IFIH1 mutations that compromised function had similar population-level representations. Missense variants having MAF of 0.01% and higher were overexpressed and tested for ability to drive luciferase expression from an IFNB1 promoter after stimulation with intracellular poly(I:C), with loss-of-function (nonsense, splice acceptor/donor, frameshift) and two known gain-of-function variants included for comparison ( Fig. 10 and Table S3 ). Among the 38 missense variants we tested, all showed equivalent protein expression (unpublished data), whereas 11 exhibited >40% decreases in luciferase activity compared with wild-type MDA5, when stimulated by transfected poly(I:C). Although variants demonstrating more severely impaired luciferase activity tended to occur above the mutation significance cutoff, activity levels were generally not well correlated with CADD scores (Fig. 10 A; Itan et al., 2016) . The experimentally validated variants having decreased luciferase activity had MAF ranging from 0.01 to 1.1%, and were found in homozygote form in as many as 22 people ( Fig. 10 B and Table S3 ). Furthermore, the presence of complete loss-of-function homozygotes in the population was consistent with little purifying selection. This conclusion was also supported by the computed f-value, which estimates the proportion of nondeleterious nonsynonymous mutations remaining in the population and is inversely proportional to the level of purifying selection (Eilertson et al., 2012) . Specifically, the f-values for IFIH1 and the related RLR sensor DDX58 were higher than that for STAT1 (0.5968 and 0.4434, respectively vs. 0.1949), sug-gesting that the RLR sensors are individually nonessential for the control of most pathogenic viruses (Deschamps et al., 2016) . Thus, these data indicate that deleterious IFIH1 variants exist at rare frequency in the normal population. This implies that there are even rarer homozygous or compound heterozygous individuals, further supporting the role of MDA5 deficiency in our patient's HRV disease, which is a rare infectious phenotype. These data further suggest that other patients with severe HRV disease may carry bi-allelic deleterious variants in IFIH1.