Downloaded from genesdev.cshlp.org on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press mRNA destabilization triggered by premature translational termination depends on at least three cis-acting sequence elements and one trans-acting factor Stuart W. Pehz, 1 Agneta H. Brown, and Allan Jacobson 2 Department of Molecular Genetics and Microbiology, University of Massachusetts Medical School, Worcester, Massachusetts 01655 USA Nonsense mutations in a gene can accelerate the decay rate of the mRNA transcribed from that gene, a phenomenon we describe as nonsense-mediated mRNA decay. Using amber (UAG) mutants of the yeast PGK1 gene as a model system, we find that nonsense-mediated mRNA decay is position dependent, that is, nonsense mutations within the initial two-thirds of the PGKl-coding region accelerate the decay rate of the PGK1 transcript ~<12-fold, whereas nonsense mutations within the carboxy-terminal third of the coding region have no effect on mRNA decay. Moreover, we find that this position effect reflects (1) a requirement for sequences 3' to the nonsense mutation that may be necessary for translational reinitiation or pausing, and (2) the presence of an additional sequence that, when translated, inactivates the nonsense-mediated mRNA decay pathway. This stabilizing element is positioned within the coding region such that it constitutes the boundary between nonsense mutations that do or do not affect mRNA decay. Rapid decay of PGK1 nonsense-containing transcripts is also dependent on the status of the UPF1 gene. Regardless of the position of an amber codon in the PGK1 gene, deletion of the UPF1 gene restores wild-type decay rates to nonsense-containing PGK1 transcripts. [Key Words: Nonsense mutations; mRNA decay; translational termination; protein-coding region; UPF1 gene] Received April 21, 1993; revised version accepted June 22, 1993. To a first approximation, changes in the expression of nipulation, we have focused on the yeast Saccharomyces specific genes are manifested by changes in the steady- cerevisiae, developed simple and reliable procedures for state levels of individual mRNAs. Although such measuring mRNA decay rates, and begun to characterize changes are assumed generally to result primarily from the cis-acting sequences and trans-acting factors that differential transcription or RNA processing activities, regulate the rapid decay of inherently unstable mRNAs differences in the decay rates of individual mRNAs can (Herrick et al. 1990; Jacobson et al. 1990; Parker and also have profound effects on the overall levels of expres- Jacobson 1990; Heaton et al. 1992; Herrick and Jacobson sion of specific genes. Although the potential impor- 1992). As in similar studies in higher eukaryotes (for re- tance of mRNA stability as a mechanism for regulating view, see Peltz et al. 1991), we find that unstable yeast gene expression has been recognized (for reviews, see mRNAs contain cis-acting sequences that dictate their Ross 1988; Cleveland and Yen 1989; Atwater et al. 1990; instability and are capable of promoting rapid decay Hentze 1991; Peltz et al. 1991; Peltz and Jacobson 1993), when transferred to appropriate locations within stable the structures and mechanisms involved in the determi- mRNAs (for review, see Pehz and Jacobson 1993). For nation of individual mRNA decay rates have yet to be five different genes (i.e., MATal, HIS3, STE3, STE2, and elucidated. CDC4) we find that such "instabilty elements" can be To address the problem of mRNA stability in an or- found within the coding regions of the respective ganism amenable to both biochemical and genetic ma- mRNAs, an observation suggesting that the processes of mRNA translation and mRNA turnover may be inti- mately linked. This conclusion is supported further by tPresent address: Department of Molecular Genetics and Microbiology, experiments in yeast demonstrating that (1) the coding University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Piscataway, New Jersey 08854-5635 USA. region instability element from the MATal mRNA will 2Corresponding author. destabilize chimeric transcripts unless it is preceded by a GENES & DEVELOPMENT 7:1737-1754 91993 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/93 $5.00 1737 Downloaded from genesdev.cshlp.org on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press Peltz et al. nonsense codon that blocks translation through the ele- tations on the decay rate of the PGK1 transcript, a DNA ment (Parker and Jacobson 1990); (2) drugs and muta- tag was inserted into the 3'-untranslated region (UTR) tions that inhibit translational elongation also inhibit (3'-UTR) of the PGK1 gene. Therefore, the mRNA decay mRNA decay (Herrick et al. 1990; Peltz et al. 1992); and rates of wild-type and mutant PGK1 alleles could be (3) nonsense mutations in the URA3, URA1, HIS4, and monitored by RNA blot analysis, hybridizing with a ra- LEU2 genes accelerate the decay rates of the mRNAs dioactive probe specific for only the tag sequence (see transcribed from these genes (Losson and Lacroute 1979; Materials and methods). Insertion of the DNA tag into Pelsy and Lacroute 1984; Leeds et al. 1991). The latter the PGK1 gene or into a pGAL--lacZ fusion neither altered phenomenon, nonsense-mediated mRNA decay, is the the decay rate of the PGK1 transcript nor the f~-galac- focus of this study. tosidase activity of the gene fusion when compared with Previous studies of nonsense-mediated mRNA decay the same genes lacking the tagged sequence (Jacobson et in yeast showed that (1) mRNA destabilization is linked al. 1990; S.W. Peltz, A.H. Brown, and A. Jacobson, un- to premature translational termination, because non- publ.). sense-containing URA3 mRNA is stabilized in a strain To investigate the relationship between the location of containing an amber suppressor tRNA (Losson and La- a nonsense mutation in the PGK1 protein-coding region croute 1979); (2) the extent of destabilization is position and its effect on mRNA half-life, a linker harboring an dependent, because 5' proximal nonsense mutations de- amber codon was inserted (in separate constructs) into stabilize transcripts to a greater degree than those that six restriction sites of the PGK1 gene (Fig. 1B). The mu- are 3' proximal (Losson and Lacroute 1979; Pelsy and tant alleles were transferred to yeast centromere plas- Lacroute 1984; Leeds et al. 1991; Peltz and Jacobson mids and transformed into yeast cells harboring the 1993); and (3) the products of the UPF1 and UPF3 genes rpbl-1 temperature-sensitive allele of RNA polymerase are involved in this degradative pathway, as mutations II (Nonet et al. 1987). The abundance and decay rates of in these genes stabilize mRNAs with nonsense muta- the wild-type and mutant PGK1 mRNAs were deter- tions without affecting the half-lives of most wild-type mined by RNA blotting analyses of RNA isolated at dif- transcripts (Leeds et al. 1991,1992; Peltz and Jacobson ferent times after inhibiting transcription by shifting the 1993). culture to the nonpermissive temperature (36~ The In this paper we have analyzed amber mutants of the results of these experiments indicate that 5' proximal yeast PGK1 gene to delineate further the cis-acting ele- nonsense mutations accelerate the PGK1 mRNA decay ments and trans-acting factors essential for nonsense- rate more than 3' proximal mutations, although the re- mediated mRNA decay. Our analysis focused on the fol- lationship is nonlinear (Fig. 1). Nonsense mutations that lowing four aspects of the nonsense-mediated mRNA de- terminate translation after 455% of the PGKl-coding cay pathway (1) the relationship between the physical sequence accelerate the PGK1 mRNA decay rate -12- position of a nonsense mutation and its effect on mRNA fold. A nonsense mutation that allows translation of turnover; (2) the identification of sequences, in addition 67% of the protein-coding region still decreases the to the nonsense codon, that are required for rapid mRNA PGK1 mRNA half-life fourfold, whereas nonsense muta- decay; (3) an understanding of the basis for the position tions inserted in the last quarter of the PGK1 transcript effect, that is, the resistance of 3' proximal nonsense had no effect (Fig. 1). As internal controls, we measured mutations to nonsense-mediated decay; and (4) the effect the decay rate of an unrelated, stable mRNA encoded by of mutations in the trans-acting factor, Upflp, on the the CYH2 gene and also measured the relative steady- half-lives of PGK1 mRNAs with nonsense mutations lo- state levels of mutant and wild-type PGK1 mRNAs. The cated at various positions within the coding region. decay rate of the CYH2 mRNA was essentially identical in all cells (tl/2 -- 40-44 min), regardless of the nature of their respective PGK1 alleles (Fig. 1B). Steady-state lev- Results els of the wild-type and nonsense-containing PGK1 alle- les were compared by RNA blot analysis of equal Destabilization of the PGK1 transcript is dependent amounts of RNA from the time zero points from each of on the position of a nonsense codon within the PGK1 the decay measurements, normalizing to the abundance protein-coding region of the CYH2 mRNA in each sample. The results of these The PGK1 gene was chosen for analysis because it en- experiments demonstrate that the mRNA abundance of codes an abundant, stable mRNA with a half-life of 60 each PGK1 nonsense allele is related directly to the de- min and, therefore, small changes in its decay rate are cay rates of the respective nonsense-containing mRNAs readily detected (Herrick et al. 1990; Parker and Jacobson (Fig. 1B). 1990; Heaton et al. 1992). The transcript of the PGK1 The product of the yeast UPF1 gene is required for the gene is -1400 nucleotides in length with a protein-cod- rapid turnover of mRNAs containing a premature trans- ing region of 1251 nucleotides (Hitzeman et al.