A key step in ensuring the proper decoding of the genetic information is performed by the aminoacyl-tRNA synthetases (aaRSs). The accuracy of aminoacylation first depends on the correct recognition of the appropriate amino acid by an aaRS. However, the task of distinguishing between two very similar amino acids becomes challenging for aaRSs and can lead to errors. If noncognate amino acids are activated, aaRSs employ an editing mechanism to hydrolyze the misactivated aa-AMP (pretransfer editing) or misacylated aa-tRNA (posttransfer editing). The accuracy of charging the tRNA is therefore ensured at two levels: amino acid selection and editing of errors, leading to the term ‘double-sieve’ model.

Phenylalanyl-tRNA synthetase (PheRS) is an (αβ)2 heterotetramer that possesses both sieving mechanisms. Phenylalanine (Phe) is the cognate amino acid for PheRS; however, tyrosine (Tyr, 4-hydroxyphenylalanine) can also be charged at low frequency.

Lu et al. (2014) reported a double-sieving defective model for PheRS. Tyr503 of the lacZ protein is crucial for its β-galactosidase activity. When this codon is mutated to a Phe codon, β-galactosidase activity decreases to less than 1% under normal condition. In PheRS sieving-defective flies, noncognate Y is incorporated at the F503 position at higher frequency, restoring β-galactosidase activity.