Revertant mosaicism has been reported in several inherited diseases, including the genetic skin fragility disorder epidermolysis bullosa (EB). Here, we describe the largest cohort of seven patients with revertant mosaicism and dystrophic EB (DEB), associated with mutations in the COL7A1 gene, and determine the underlying molecular mechanisms. We show that revertant mosaicism occurs both in autosomal dominantly and recessively inherited DEB. We found that null mutations resulting in complete loss of collagen VII and severe disease, as well as missense or splice-site mutations associated with some preserved collagen VII function and a milder phenotype, were corrected by revertant mosaicism. The mutation, subtype, and severity of the disease are thus not decisive for the presence of revertant mosaicism. Although collagen VII is synthesized and secreted by both keratinocytes and fibroblasts, evidence for reversion was only found in keratinocytes. The reversion mechanisms included back mutations/mitotic recombinations in 70% of the cases and second-site mutations affecting splicing in 30%. We conclude that revertant mosaicism is more common than previously assumed in patients with DEB, and our findings will have implications for future therapeutic strategies using the patient's naturally corrected cells as a source for cell-based therapies.

Mutations in the type XVII collagen gene (COL17A1) result in the blistering disorder non-Herlitz junctional epidermolysis bullosa (JEB-nH). The incidence of revertant mosaicism, also called "natural gene therapy", was identified in a cohort of 14 patients with JEB-nH caused by COL17A1 mutations in the Netherlands. Five different in vivo reversions, all correcting the germ-line COL17A1 mutation c.2237delG in exon 30, were found in four mosaic JEB-nH patients. The correcting DNA changes involved a wide variety of somatic mutations, from which an indel mutation (c.2228-101_2263+70delins15) and a large deletion of 2,165 base pairs (c.2227+153_2336-318del) have not been previously observed in patients with revertant mosaicism. Our results show that there is no preference for a repair mechanism. Moreover, revertant mosaicism was confirmed on a DNA level in 6 out of 10 generalized JEB-nH patients. Further, photo-material and clinical history of the other four generalized JEB-nH patients demonstrated that each patient has revertant skin patches. In contrast, revertant mosaicism was not detected in the four localized JEB-nH patients. The fact that so many, if not all, generalized JEB-nH COL17A1 patients have revertant patches offers opportunities for cell therapies in which the patient's own naturally corrected cells are used as a source.