R EPORTS wild-type PGK1 mRNA was synthesized with a poly(A) tail of approximately 70 residues and Exosome-Mediated Recognition was subsequently deadenylated slowly (Fig. 2A) (12). In contrast, nonstop-PGK1 transcripts dis- and Degradation of mRNAs appeared rapidly without any detectable dead- enylation intermediates (Fig. 2B). In addition, in Lacking a Termination Codon a ski7⌬ strain, the nonstop mRNA persisted as a fully polyadenylated species for 8 to 10 min Ambro van Hoof,1,2* Pamela A. Frischmeyer,1,3 Harry C. Dietz,1,3 before disappearing (Fig. 2C). These data indi- Roy Parker1,2* cate that exosome function is required for rapid degradation of both the poly(A) tail and the One role of messenger RNA (mRNA) degradation is to maintain the fidelity of body of the mRNA. Based on these observa- gene expression by degrading aberrant transcripts. Recent results show that tions, we suggest that nonstop mRNAs are rap- mRNAs without translation termination codons are unstable in eukaryotic cells. idly degraded in a 3Ј-to-5Ј direction by the We used yeast mutants to demonstrate that these “nonstop” mRNAs are exosome, beginning at the 3Ј end of the poly(A) degraded by the exosome in a 3Ј-to-5Ј direction. The degradation of nonstop tail (13). transcripts requires the exosome-associated protein Ski7p. Ski7p is closely Two observations suggest a mechanism by related to the translation elongation factor EF1A and the translation termi- which nonstop mRNAs are specifically recog- nation factor eRF3. This suggests that the recognition of nonstop mRNAs nized and targeted for destruction by the exo- involves the binding of Ski7p to an empty aminoacyl-(RNA-binding) site (A site) some. First, nonstop mRNA degradation re- on the ribosome, thereby bringing the exosome to a mRNA with a ribosome quires that a translating ribosome reach at least stalled near the 3Ј end. This system efficiently degrades mRNAs that are the poly(A) tail, and most likely the 3Ј end of the prematurely polyadenylated within the coding region and prevents their ex- mRNA (2, 14). The simplest interpretation of pression. these data is that nonstop mRNAs are recog- nized when a ribosome reaches the 3Ј end of the mRNA biogenesis is a multistep process with a ski4-1 allele encodes a point mutation in mRNA. Such a recognition would be analogous certain frequency of errors, either due to inher- one of the core exosome subunits that spe- to the recognition of ribosomes with an empty A ent inaccuracies in transcription and processing cifically disrupts cytoplasmic 3Ј-to-5Ј deg- site by a tRNA-mRNA hybrid (tmRNA) in pro- or due to mutations in the DNA template. The radation of mRNA without affecting any of karyotes (15, 16). Second, the COOH-terminal cell has evolved mechanisms to rapidly degrade the other known functions of the exosome region of the Ski7 protein is closely related to aberrant mRNAs, such as unspliced pre- (9). The ski4-1 mutation stabilizes the non- the guanosine triphosphatases (GTPases) EF1A mRNAs, mRNAs with aberrantly long 3Ј un- stop-PGK1 mRNA at least sixfold (Fig. 1, and eRF3, including similarity in the GTPase translated regions (3ЈUTRs), and mRNAs with A and B). Exosome-mediated degradation domain (17–19). EF1A and eRF3 are translation premature translation termination codons (1). of normal cellular mRNAs requires the Recently, it has been found that eukaryotic exosome and two other factors (6, 9). One mRNAs that do not contain a termination codon factor is a heterotrimeric helicase complex are rapidly degraded (2). The rapid decay of of Ski2p, Ski3p, and Ski8p (6, 10). As these transcripts is referred to as nonstop mRNA shown in Fig. 1, C through E, Ski2, -3, and decay and requires translation of the mRNA (2). -8 are all required for nonstop mRNA deg- However, degradation of a PGK1 mRNA, from radation. The second factor required for which all in-frame termination codons have exosome-mediated mRNA decay is Ski7p, been removed (nonstop-PGK1), requires none and deletion of SKI7 also caused stabiliza- of the enzymes involved in the major pathway tion of nonstop mRNAs (Fig. 1F). Because for mRNA degradation, which occurs by dead- Ski2p and Ski7p localize to the cytoplasm enylation, decapping, and 5Ј-to-3Ј digestion (2– (10, 11), we interpret these observations to 5). This suggests that nonstop mRNAs might be indicate that nonstop mRNAs are degraded degraded by the exosome complex of 3Ј-to-5Ј 3Ј to 5Ј by the cytoplasmic exosome. exoribonucleases, the functions of which in- Given that the major deadenylase (Ccr4p) is clude 3Ј-to-5Ј degradation of mRNA in the cy- not required for nonstop decay (2) and that toplasm, nuclear processing of ribosomal RNA degradation occurs by the exosome, it is possi- and small nucleolar RNAs, and degradation of ble that the exosome both deadenylates and processing intermediates and stalled mRNAs in degrades nonstop mRNAs. This would be sur- Fig. 1. Nonstop mRNA degradation requires the nucleus (6–8). prising because normal mRNAs cannot be exosome and cytoplasmic exosome cofactors. To test whether the exosome functions deadenylated by the exosome (5). Alternatively, Nonstop-PGK1 mRNA stability was measured in wild-type (A), ski4-1 (B), ski2⌬ (C), ski3⌬ in nonstop decay, we first examined non- an unidentified nuclease may remove the poly- ⌬ ⌬ ⌬ stop decay in a ski4-1 strain of yeast. The adenylate [poly(A)] tail from nonstop mRNAs, (D), ski8 (E), ski7 (F), ski7- C(G), and ski7- ⌬N(H) strains. Each strain contained a URA3 followed by exosome-mediated decay. plasmid encoding the reporter gene and was 1Howard Hughes Medical Institute, 4000 Jones Bridge To investigate whether the exosome de- grown to early- to mid-log phase at 30¡C in Road, Chevy Chase, MD 20815, USA. 2University of grades the poly(A) tail of nonstop transcripts, media containing 2% galactose and lacking Arizona, Department of Molecular and Cellular Biolo- we performed transcriptional pulse-chase exper- uracil. Transcription of the reporter gene was 3 gy, Tucson, AZ 85721, USA. Johns Hopkins Universi- iments. In these experiments, transcription of inhibited by replacing the media with media ty, Institute for Genetic Medicine, School of Medicine, ϭ the reporter mRNA was induced briefly and was containing glucose (T 0 min) and aliquots Baltimore, MD 21205, USA. were taken thereafter. RNA was analyzed as followed by transcriptional repression, which *To whom correspondence should be addressed. E- described (9). The indicated half-lives are aver- mail: [email protected] (A.v.H.) or rrparker@ yielded a synchronous population of mRNA ages of at least two experiments and were u.arizona.edu (R.P.) whose fate could be monitored. For comparison, calculated after correction for loading (9). 2262 22 MARCH 2002 VOL 295 SCIENCE www.sciencemag.org R EPORTS factors that interact with the A site of the ribo- COOH-terminal GTPase domain does not play ing with 1 M NaCl (Fig. 3A) (22), suggesting a some when it contains a sense or nonsense a role in exosome-mediated degradation of nor- strong interaction between Ski7p and the exo- codon, respectively. The interaction of Ski7p mal mRNAs. First, the NH2-terminal domain, some. The nuclear form of the exosome con- homologs with the ribosomal A site suggests but not the COOH-terminal domain, is required tains one additional subunit, Rrp6p (23, 24). that the homologous domain of Ski7p may func- for viability under conditions in which exo- Purification of protein A–tagged Rrp6 did not tion to distinguish nonstop from normal mRNAs some-mediated decay is essential for viability result in copurification of Ski7p (22), which is by binding to the empty A site of ribosomes that (19). Second, the deletion of the NH2-terminal consistent with Ski7p being specific to the cy- have reached the 3Ј end of the mRNA. This part, but not the COOH-terminal part, of Ski7p toplasmic exosome. Recently, Araki et al.(11) hypothesis predicts that the COOH-terminal do- causes a dramatic decrease in the rate of exo- independently found that, when overexpressed, main of Ski7p is specifically required for non- some-mediated decay of normal mRNAs (19). the NH2-terminal part of Ski7p can coimmuno- stop decay but may not be required for exo- Both ski7 alleles stabilized the nonstop precipitate with the exosome. The finding that some-mediated degradation of normal mRNAs. reporter transcript (Fig. 1, G and H), indicat- Ski7p stably associates with the exosome To determine the function of the Ski7p do- ing that the COOH-terminal part of Ski7p through its NH2-terminal suggests a mechanism mains in exosome-mediated decay of nonstop functions in the nonstop mRNA degradation to recruit the exosome to nonstop mRNAs rec- and normal mRNAs, we generated yeast strains pathway. However, the COOH-terminal trun- ognized by the COOH-terminal of Ski7p. that express different deletion mutants of Ski7p cation of Ski7p has a smaller effect than To determine whether the interaction of (20). Two lines of evidence indicate that the either the NH2-terminal deletion or complete Ski7p with the exosome is biologically rel- NH2-terminal nonconserved domain of Ski7p is deletion of SKI7. This suggests that other evant, we examined whether mutations in necessary and sufficient for exosome-mediated factors may to some extent be able to substi- the exosome that disrupt all Ski7p-depen- degradation of normal mRNAs and that the tute for the COOH-terminal domain. Taken dent functions of the exosome also disrupt together, these results indicate that the NH2- Ski7-exosome interaction. Figure 3C shows terminal part of Ski7p plays a central role in that the ski4-1 mutation severely reduces exosome-mediated mRNA decay and that the the copurification of Ski7p with the exo- COOH-terminal domain plays a specific role some.