Clinical phenotypes of MAGEL2 mutations and deletions

Abstract

Letter to the EditorAlthough it has long been known that Prader-Willi syn-drome (PWS) is caused by the loss of function of im-printed, paternally expressed genes in 15q11q13, thecontribution of the different genes within this region hasnot yet been completely resolved. Based on the identifi-cation of rare deletions affecting only the snoRNA genecluster SNORD116 it has been suggested that this is themajor locus [1-3]. Recently, Schaaf et al. have describedtruncating mutations of in four patients with abroad range of clinical phenotypes [4]. The authors con-clude that MAGEL2 loss of function can contribute toseveral aspects of the phenotype. While this maybe true, we think that the available data are not sufficientto justify this conclusion.We have recently seen a 3-year-old boy with a paternallyinherited deletion of~3.9 Mb that includes MAGEL2,but not the SNRPN/SNORD116 locus (Figure 1 andAdditional file 1: Figure S1). Apart from delayed motorskills, the boy is asymptomatic (for a detailed clinical de-scription see Additional file 1). This is the second individ-ual with a MAGEL2deletion who certainly does not havePWS; the first one was also described by our group [5].Here we offer an explanation for the apparently discrepantfindings, which is also important for deciphering the roleof candidate genes in and other contiguous genesyndromes.Usually, exome sequencing is performed to identify agene that is affected in several patients with the samedisease. The identification of such a gene is a strong in-dication that a mutation in this gene causes the disease.Schaaf et al. have started their study with a patient ofunknown clinical diagnosis, whose genome was investi-gated under a de novo model only. The other three pa-tients were identified by searching a clinical exome database. Apparently, an exome-wide analysis under diffe-rent genetic models was not performed in these patients.Therefore, the number of potentially pathogenic variantsin these patients is unknown. In this situation, it is difficultto prove causality, especially when there is no consistentphenotype (since each of the Holm's criteria for re-fers to a rather common and unspecific clinical sign, manypatients with diverse disorders fulfill some of them; theseshould not be called PWS phenotypes). The paternalorigin of theMAGEL2 mutations does not prove causal-ity, because the majority of point mutations occur dur-ing spermatogenesis. In summary, it is possible that theMAGEL2mutations are innocent bystanders and thatthe patients have autosomal recessive or X-linked reces-sive disease (note that all patients are male).Even if the mutations were causally relatedto the clinical phenotypes of the patients described bySchaaf et al., it is still possible that they do not contrib-ute to PWS, and there is a precedent for this. In fact,MAGEL2 is not the first protein-coding gene in the PWSregion found to be mutated. The first one is MKRN3(Figure 1d), which was found to be mutated in patientswith central precocious puberty [6]. In contrast to thesepatients, patients with typically have incomplete ordelayed puberty. The finding that MKRN3 loss of func-tion alone causes central precocious puberty, but not incombination with the loss of function of the SNORD116genes, indicates that the SNORD116 loss of functionis epistatic to MKRN3 loss of function, probably becausethe SNORD116 genes act developmentally upstream ofMKRN3.Another possibility is that there is leaky expression ofthe maternal allele in a subset of neurons inpatients with a paternal deletion (our patientand patients), but not in patients with a truncatingMAGEL2 mutation (the patients described by Schaafet al.). A precedent for this situation is the recent findingof stochastic loss of silencing of the imprinted Ndn/NDN allele [7]. These authors find weak expression of