In a recent publication in the European Journal of Human Genetics, Khajavi et al1 reviewed the physiological role of nonsense-mediated mRNA decay (NMD), its implications for human disease, and why knowledge of NMD is important to understand genotype–phenotype correlations in various genetic disorders. These authors stated that if a transcript carries a premature translation termination codon located more than 50–54 nucleotides upstream the last exon–exon junction, then NMD is triggered to prevent translation, thereby reducing the dominant-negative effect of the C-terminally truncated protein that could be produced.

However, it must be noted that if a premature translation termination codon is sufficiently AUG-proximal, it does not follow the ‘50–54 nucleotides boundary rule’.2, 3 For example, mutant human β-globin mRNAs with AUG-proximal nonsense mutations escape NMD.4 Nevertheless, the thalassemic phenotypes of heterozygotes and homozygotes for AUG-proximal nonsense mutations do not appear to be any more or less severe than in patients carrying more distal nonsense mutations that are effectively targeted to NMD. We have demonstrated that the observed NMD inhibition is determined by the proximity of the nonsense codon to the initiation AUG.2 In addition, the AUG-proximity effect may also operate in parallel with other modifying influences to inhibit NMD.3 Specifically, the occurrence of translation re-initiation 3′ to the short open reading frame may also contribute to the alleviation of the NMD effect.3, 5 These two parameters can independently contribute to the net overall NMD resistance of a nonsense-containing mRNA with implications for genotype–phenotype correlations in various human genetic disorders. Specifically, if translation re-initiation occurs in-frame downstream to the nonsense codon, it produces an N-terminally truncated protein that may or may not be functional. Some results have been reported, in which translation reinitiation at internal AUG codons modulates disease phenotypes.6, 7, 8, 9, 10, 11, 12 As a recent example, Sanchez-Sanchez et al6 have reported a novel mutation detected in the retinoblastoma 1 (RB1) gene in 10 individuals of an extended family, but only three of whom are affected by retinoblastoma. The mutation comprises a 23-basepair duplication in the RB1 first exon, producing a premature translation termination in exon 2, and no appreciable NMD is induced. Instead, transcription–translation in vitro assays revealed that alternative in-frame translation start sites involving methionine (Met) 113 and possibly Met233 are used to generate N-terminally truncated RB1 products, known and suspected to exhibit tumor suppressor activity. These results strongly suggest that modulation of disease penetrance in this family is achieved by translation re-initiation downstream to a nonsense codon.6

In conclusion, when analyzing the genotype–phenotype correlations in genetic disorders owing to nonsense codons, besides the fact that NMD may modulate the corresponding clinical outcome, it is also important to consider that some nonsense codons, although located more than 50–54 nucleotides upstream the 3′-most exon–exon junction of a transcript, do not induce NMD. Instead, translation re-initiation at downstream internal AUG codons may also modulate human disease phenotypes and should therefore be considered whenever unexpected phenotypes resulting from severe mutations located early in the coding region are observed.