mRNA decay rates often increase when translation is terminated prematurely due to a frameshift or nonsense mutation. We have identified a yeast gene, UPF1, that codes for a trans-acting factor whose function is necessary for enhanced turnover of mRNAs containing a premature stop codon. In the absence of UPF1 function, frameshift or nonsense mutations in the HIS4 or LEU2 genes that normally cause rapid mRNA decay fail to have this effect. Instead, the mRNAs decay at rates similar to the corresponding wild-type mRNAs. The stabilization of frameshift or nonsense mRNAs observed in upfl-strains does not appear to result from enhanced readthrough of the termination signal. Loss of UPF1 function has no effect on the accumulation or stability of HIS4 + or LEU2 + mRNA, suggesting that the UPF1 product functions only in response to a premature termination signal. When we examined the accumulation and stability of other wild-type mRNAs in the presence or absence of UPF1, including MAT~I, STE3, ACT1, PGK1, PAB1, and URA3 mRNAs, only the URA3 transcript was affected. On the basis of these and other results, the UPF1 product appears to participate in a previously uncharacterized pathway leading to the degradation of a limited class of yeast transcripts. Nonsense mutations that generate a premature translational termination signal often reduce the steady-state accumulation of the corresponding mRNA (Brown 1989;Peltz et al. 1990). In a study of the yeast URA3 gene, it was shown that the extent of reduced mRNA accumulation depends on the position of the nonsense mutation (Losson and Lacroute 1979). Mutations near the 5' end of the transcript were shown to have a greater destabilizing effect than mutations near the 3' end. Furthermore, introduction of an amber tRNA suppressor restabilized ura3 nonsense mRNA, indicating that the turnover rate is determined in part by the relative efficiencies of termination versus readthrough of the stop codon. These studies suggested that the turnover rate of nonsense mRNA is probably related to some aspect of its translation rather than to a potential change in mRNA structure that might result from the presence of a nonsense mutation.Similar studies in higher eukaryotes have proven more difficult to interpret. In some cases, the introduction of a premature stop codon into a gene has been linked to 3Corresponding author. increased cytoplasmic turnover (Maquat et al. 1981;Barker and Beemon 1991). However, other studies suggest that nonsense mutations may cause changes in nuclear processing and/or transport, and these changes, rather than cytoplasmic mRNA degradation, may be primarily responsible for decreased steady-state mRNA levels (Humphries et al. 1984;Takeshita et al. 1984;Urlaub et al. 1989;Cheng et al. 1990).Here, we report the characterization of mutations in the yeast Saccharomyces cerevisiae that specifically stabilize mRNAs containing a premature translational termination signal. The mutations arose in a strain containing his4-38, a + 1 frameshift mutation near the 5' end of the HIS4 tr...