A key aspect of understanding eukaryotic gene regulation will be the identification and analysis of proteins that bind mRNAs and control their function. Recently, a complex of seven Lsm proteins and the Pat1p have been shown to interact with yeast mRNAs and promote mRNA decapping. In this study we present several observations to indicate that the LsmI/Pat1 complex has a second distinct function in protecting the 3′-UTR of mRNAs from trimming. First, mutations in the LSM1 to LSM7, as well as PAT1, genes led to the accumulation of MFA2pG and PGK1pG transcripts that had been shortened by 10–20 nucleotides at their 3′ ends (referred to as trimming). Second, the trimming of these mRNAs was more severe at the high temperature, correlating with the inability of these mutant strains to grow at high temperature. In contrast, trimming did not occur in a dcp1Δ strain, wherein the decapping enzyme is lacking. This indicates that trimming is not simply a consequence of the inhibition of mRNA decapping. Third, the temperature-sensitive growth of lsm and pat1 mutants was suppressed by mutations in the exosome or the functionally related Ski proteins, which are required for efficient 3′ to 5′ mRNA degradation of mRNA. Moreover, in lsm ski double mutants, higher levels of the trimmed mRNAs accumulated, indicating that exosome function is not required for mRNA trimming but that the exosome does degrade the trimmed mRNAs. These results raise the possibility that the temperature-sensitive growth of the lsm1-7 and pat1 mutants is at least partially due to mRNA trimming, which either inactivates the function of the mRNAs or makes them available for premature 3′ to 5′ degradation by the exosome.
THE function of eukaryotic mRNAs is controlled by a variety of mRNA binding proteins. Recently, a complex of seven Lsm proteins (Lsm1p through Lsm7p) and the Pat1p were found to interact with yeast mRNAs and promote mRNA degradation by enhancing the rate of decapping (Bouveretet al. 2000; Tharunet al. 2000). These Lsm (Like-Sm) proteins were identified as a family of proteins that contain the “Sm motif” found in the Sm proteins (Hermannet al. 1995; Seraphin 1995). The Sm proteins are a family of small proteins that bind to the U1, U2, U4, and U5 snRNAs as a heptameric complex (Branlantet al. 1982; Liautardet al. 1982; Kambachet al. 1999). Sm proteins form a seven-member, doughnut-shaped structure through the interactions between the Sm motifs (Kambachet al. 1999). On the basis of the sequence similarities between the Sm and the Lsm proteins, combined with coimmunoprecipitation experiments (Mayeset al. 1999; Salgado-Garridoet al. 1999), Lsm proteins likely assemble into analogous heptameric ring structures. This view is supported by the finding that purified human Lsm2 to Lsm8 proteins form a seven-member ring structure (Achselet al. 1999). There are at least two functionally distinct Lsm complexes in yeast (for review, see He and Parker 2000). A nuclear Lsm complex consisting of Lsm2 through Lsm8 proteins is present in the nucleus, binds to the U6 snRNA, and functions in pre-mRNA splicing. In addition, a cytoplasmic Lsm complex consisting of Lsm1 through Lsm7 proteins and an additional protein (Pat1p) interacts with the mRNA decay machinery, facilitating the decapping step of mRNA degradation (Hatfieldet al. 1996; Bonnerotet al. 2000; Bouveretet al. 2000; Tharunet al. 2000).
Besides their functions in pre-mRNA splicing and promoting mRNA decapping, strains lacking Lsm proteins and the Pat1p have been reported to have additional phenotypes. For example, strains lacking Lsm1p or Pat1p, which are specific to the cytoplasmic Lsm complex, are viable but fail to grow at high temperature (Hatfieldet al. 1996; Wanget al. 1996; Boecket al. 1998; Tharunet al. 2000). In addition, some of these mutant strains accumulate mRNAs shortened at their 3′ end by 10–20 nucleotides. Specifically, a lsm1Δ mutant accumulates shortened mRNA species (Boecket al. 1998). Similar shortened MFA2 mRNA species have been noted in the lsm1, lsm5, lsm6, lsm7, and pat1 mutants (Bouveretet al. 2000). Time course experiments have demonstrated that these 3′ shortened species arise by degradation of the 3′ end of the mRNA following deadenylation (Boecket al. 1998; Schwartz and Parker 2000). We will refer to the specific removal of 10–20 nucleotides from the 3′ end of the mRNA as trimming.
In this study we address the roles of the cytoplasmic Lsm complex in preventing mRNA trimming and the relationship of this function to its role in promoting mRNA decapping. We demonstrate that defects in any component of the cytoplasmic LsmI/Pat1p complex, but not defects in decapping per se, lead to trimming of multiple mRNAs in a process that is accelerated at higher temperatures. Moreover, inhibition of cytoplasmic 3′ to 5′ degradation by the exosome does not prevent trimming, but partially suppresses the temperature-sensitive growth of the lsm and pat1 strains and leads to the accumulation of higher levels of trimmed mRNAs. These results indicate that the LsmI/Pat1p complex has a distinct role in preventing mRNA trimming. They also suggest that the temperature-sensitive growth of lsm1-7 and pat1 mutants is at least partially due to mRNA trimming, which either inactivates the function of the mRNAs or makes them available for premature 3′ to 5′ degradation by the exosome.
MATERIALS AND METHODS
Media and yeast strains: Yeast media were prepared according to standard methods. Cells were grown in YEP rich medium or complete minimal (CM) drop-out medium to maintain plasmids. Most yeast strains, unless indicated, contained GAL1 upstream activating sequence (UAS) regulated MFA2pG and PGK1pG genes and were grown in medium containing 2% galactose to induce the transcription of reporter genes MFA2pG and PGK1pG. Conditional lsm3 or lsm4 mutants (GAL UAS-controlled LSM3 and LSM4) were transformed with a glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter-regulated MFA2pG plasmid, grown in CM drop-out medium containing 2% galactose, and then shifted to CM drop-out medium containing 2% dextrose for 14 hr to shut off the transcription of LSM3 or LSM4 (Mayeset al. 1999).
Yeast strains used in this study are listed in Table 1. Most strains used in this study are in the genetic background of yRP841, with mutations introduced either by transformation or by repeated backcrosses. The lsm3, lsm4, and lsm8-1 strains used for examining mRNA trimming are in a different genetic background and were compared to their isogenic wild types. All double mutants were constructed by crossing the respective single-mutant strains. PCR analysis was used to identify the lsm7Δ::HIS3 mutation. Primers oRP1046 (5′ CCT TGT TGT CGT ACT GTC 3′) and oRP1047 (5′ GGA CAC TGA GTT TCG AAA 3′) were used to amplify the LSM7 region. The HIS3 replacement of the LSM7 open reading frame changes the length of the PCR products. The ski4-1 mutation abolishes the recognition site for restriction endonuclease StyI. The ski4-1 allele was identified by PCR amplification using oRP943 and oRP944, followed by StyI as described previously (van Hoofet al. 2000).
Plasmid construction: The GPD promoter controlled MFA2pG plasmid (pRP1043) was constructed by homologous recombination. The GPD promoter region was PCR amplified with primer pairs oPR1045 (5′ GGA AAC AGC TAT GAC CAT GAT TAC GAA TTC GCA TTA TCA ATA CTC GCC 3′) and oPR1042 (5′ GTT ATT GTT GTA TGA AGA TGA TAG CTC GCT GGC TTG GTG TTT TAA AAC 3′) from vector PG-1 (Schena and Yamamoto 1988). The MFA2pG region was amplified using oPR1043 (5′ GTT AGT CTT TTT TTT AGT TTT AAA ACA CCA AGC CAG CGA GCT ATC ATC TTC 3′) and oRP1044 (5′ GTT TTC CCA GTC ACG AC 3′) from vector pRP1044 (pCD61). The two PCR products share a homologous region, while each of them has a region homologous to the vector pRP11. The two pieces of PCR products and linearized vector pRP11 were cotransformed into yeast to allow homologous recombination. The recombined plasmids were isolated from yeast cells, amplified in Escherichia coli, and confirmed by restriction enzyme digestion.
RNA analysis: Temperature-shift experiments were performed by growing cells at 24° in media containing 2% galactose until early log phase (OD600 = 0.3). Then the culture was split into two, one kept at 24°, and the other shifted to 37°. After an hour, cells were harvested and immediately frozen in dry ice.
Total RNA isolation and Northern analysis were performed according to standard protocols (Caponigroet al. 1993). The oligonucleotides used as probes in the Northern analysis are: oPR140 (5′ ATA TTG ATT AGA TCA GGA ATT CC 3′) for MFA2pG; oPR141 (5′ AAT TGA TCT ATC GAG GAA TTC C 3′) for PGK1pG; oRP100 (5′ GTCTAGCCGCGAGGAAGG 3′) for signal recognition particle (SRP); oRP98 (5′ CTT GGA CCC GTA AGT TTC AC 3′) for GAL10; oRP1041 for the 5′ end → poly(G) fragment of MFA2pG (5′ CCA AAT TCC TAG ATC TCT TGG 3′); oPR154 to detect the 5′ end → poly(G) fragment of PGK1pG (Muhlrad and Parker 1994); oRP1048 for 25S rRNA (5′ CTA AGT CGT ATA CAA ATG 3′); oRP1050 for 18S rRNA (5′ GGA CGT AAT CAA CGC AAG 3′); and oRP924 for 5.8S rRNA (van Hoofet al. 2000).
RNase H reactions were done as previously described (Muhlrad and Parker 1992). The oligos used in RNase H reactions to reduce the size of mRNAs are as follows: oligo oRP70 (5′ CGG ATA AGA AAG CAA CAC CTG G 3′) for PGK1pG; oRP97 (5′ GTA TCT ACA AGG CTC GAT TG 3′) for GAL10; oRP1049 for 25S rRNA (5′ CAA TTC GCC AGC AAG CAC 3′); and oRP1051 for rRNA 18S (5′ CGC TTA CTA GGA ATT CCT C 3′).
Strains defective in the LsmI/Pat1p complex accumulate 3′ trimmed mRNAs: To examine how general the accumulation of 3′ shortened mRNAs is in strains defective in components of the LsmI/Pat1p complex, we examined the 3′ end of several mRNAs in a variety of yeast strains. These included strains defective in the LSM1 through the LSM7 genes, as well as a strain lacking Pat1p, which associates with the Lsm1–7p complex. In these experiments we examined the structure of the mRNAs at both 24° and after a 1-hr incubation at 37°.
Examination of the PGK1pG mRNA showed that full-length transcripts shortened from the 3′ end were seen in the lsm1Δ strains at 24° (Figure 1A). Since these trimmed PGK1 mRNAs are ~10 nucleotides smaller than the full-length mRNA (FL), we refer to this species as FL(3′-10). At 24°, the PGK1 FL(3′-10) species was not observed in the wild type, dcp1Δ, or in the lsm2ts, lsm5ts, lsm6Δ, and lsm7Δ mutants. In contrast, after a 1-hr incubation at 37°, the amount of the PGK1 FL(3′-10) increased in the lsm1Δ strain, and the PGK1 FL(3′-10) mRNA was detected in the lsm2ts, lsm5ts, lsm6Δ, lsm7Δ, and pat1Δ mutant strains. The increase in trimmed mRNA species at the high temperature suggested that this process is accelerated at the high temperature (see below).
The shortening of the 3′ end of the mRNA can also be observed on an intermediate of mRNA decay trapped by an 18-nucleotide poly(G) tract inserted in the 3′-untranslated region (UTR) of the PGK1pG mRNA. The poly(G) tract forms a strong tertiary structure and blocks exonucleolytic degradation of the reporter mRNAs (Vrekenet al. 1992; Decker and Parker 1993). This allows the accumulation of a mRNA fragment that has been decapped and degraded from 5′ to 3′ to the 5′ side of the poly(G) tract. Similar to previous results (Boecket al. 1998; Bouveretet al. 2000; Schwartz and Parker 2000), we observed that the PGK1 poly(G) fragment was also shortened by ~10 nucleotides at the 3′ end in the lsm2ts, lsm5ts, lsm6Δ, and lsm7Δ mutant strains at 37°, and in the lsm1Δ and pat1Δ strains at both temperatures, but not in the wild type at either temperature (Figure 1A). The levels of poly(G) fragment (FR) and trimmed poly(G) fragment [FR(3′-10)] were low in the lsm2ts and lsm5ts mutants and absent from the dcp1Δ strain due to the inhibition of decapping in these mutants (Figure 1A).
Similar, but slightly different, results were obtained when the 3′ end of the MFA2 mRNA was examined in the same strains (Figure 1B). The trimmed full-length MFA2pG mRNA was not detected in the lsm2ts, lsm5ts, lsm6Δ, and lsm7Δ mutants at 24°, with only a small amount present at 37°. In contrast, a 3′ shortened full-length MFA2pG mRNA [FL(3′-10)] was detected in the lsm1Δ and the pat1Δ mutants, most notably at 37° (Figure 1B). No trimming of the full-length MFA2pG mRNA was observed either in the wild-type or the dcp1Δ mutant (Figure 1B). This latter observation suggests that simply blocking decapping does not account for the mRNA trimming. Two forms of the trimmed MFA2 poly(G) fragments, shortened by ~10 and 20 nucleotides, respectively, were clearly seen in the lsm1Δ, lsm6Δ, lsm7Δ, and pat1Δ mutants [Figure 1B, FR(3′-10) and FR(3′-20)]. The wild-type cells showed a low level of trimmed MFA2 poly(G) fragment FR(3′-10) at 37° (Figure 1B), indicating that high temperature makes the 3′-UTR of MFA2 poly(G) fragment accessible to trimming even in the wild-type cells.
We also used the MFA2pG transcript to determine whether strains defective in Lsm3p and Lsm4p would show trimming of mRNAs. To do this, we used cells in which Lsm3p and Lsm4p were expressed from a GAL promoter and were depleted following a shift to glucose media. Under this condition, MFA2 transcripts trimmed from the 3′ end appeared [Figure 1C, FL(3′-10)]. This indicates that the lsm3 and lsm4 mutations lead to trimming of the MFA2 transcript.
The above experiments indicate that lesions in the cytoplasmic LsmI/Pat1p complex lead to accumulation of mRNAs shortened at their 3′ ends. To test whether the nuclear Lsm complex was involved in this process, we examined the trimming phenotype in a lsm8-1 mutant, which is a partial loss of function allele in this gene (Pannoneet al. 1998). The Lsm8 protein is a unique member of the nuclear Lsm complex. There was no trimming of the MFA2 mRNA in the lsm8-1 mutant (data not shown). This result indicates that protecting the 3′-UTR of mRNAs is a distinct function of the cytoplasmic LsmI/Pat1p complex.
3′ mRNA trimming is not simply due to an inhibition of decapping: The above results document that defects in the LsmI/Pat1p complex lead to the accumulation of mRNAs shortened from the 3′ end. This effect is seen with the MFA2 and PGK1 mRNAs, and we have also seen similar results with the GAL10 mRNA (data not shown). In principle, the accumulation of the 3′ trimmed species could be caused in two different ways. First, because defects in the LsmI/Pat1p complex cause an inhibition of decapping, this defect in decapping could simply allow sufficient time for mRNAs to be trimmed at the 3′ end. However, this possibility is inconsistent with the observation that a dcp1Δ strain, which has a strong block to decapping, does not accumulate 3′ trimmed species (Beelmanet al. 1996; Figure 1, A and B). An alternative possibility is that the protection of the 3′ end of mRNAs from trimming is a second distinct function of the LsmI/Pat1p complex, independent of this complex's role in promoting decapping. To more rigorously distinguish between these possibilities, we examined the difference in mRNA trimming between LSM1 and lsm1Δ strains in a dcp1Δ background where decapping is completely inhibited. Since in a dcp1Δ background all decapping is inhibited, any difference between LSM1 and lsm1Δ strains must be due to an effect distinct from one affecting decapping. As shown in Figure 2, the MFA2 mRNA and the PGK1 mRNA were not trimmed in the dcp1Δ mutant. However, mRNAs were still trimmed in the lsm1Δ dcp1Δ mutant. On the basis of this result, we suggest that the LsmI/Pat1p complex has a distinct function in protecting the 3′ ends of mRNAs from some type of exo- or endonucleolytic degradation.
The ski mutations that cause defects in the 3′ to 5′ mRNA degradation suppress the temperature-sensitive growth of lsm and pat1 mutants: The observation that the 3′ trimming of mRNAs is increased at high temperature in the mutant strains suggests a possible explanation for the growth phenotypes of strains defective in the LsmI/Pat1p complex. Lsm1p and Pat1p, the two proteins specific for the cytoplasmic LsmI/Pat1p complex, are both required only for growth at 37°. [Deletion of other Lsm genes is either lethal (LSM2, LSM3, LSM4, LSM5, and LSM8) or also causes temperature-sensitive growth (LSM6 and LSM7), although the interpretation of these phenotypes is more complicated since these proteins are also components of a nuclear Lsm complex (Mayeset al. 1999).] The observation that the trimming of PGK1pG and MFA2pG mRNAs is increased at 37° suggests that the inability of at least the lsm1Δ and pat1Δ strains to grow at the higher temperature might be partially due to the enhanced mRNA trimming. A prediction of this hypothesis is that blocking the 3′ to 5′ degradation of mRNA might rescue the temperature-sensitive growth.
To test this hypothesis, we created double-mutant strains carrying the lsm1Δ and a second lesion affecting the 3′ to 5′ degradation of mRNA. For this analysis, we used lesions in the SKI2, SKI3, SKI4, SKI7, and SKI8 genes, all of which are required for the 3′ to 5′ degradation of mRNA following deadenylation by the exosome complex (Anderson and Parker 1998; van Hoofet al. 2000). The creation of these double mutants revealed two important observations. First, in contrast to other lesions that affect decapping, such as the dcp1Δ (Anderson and Parker 1998) or the dcp2Δ (Dunckleyet al. 2001), the lsm1Δ was not synthetically lethal with any of the ski mutations. Because the lsm1Δ is not a complete block to decapping, this observation is consistent with only a low level of mRNA degradation being required for viability (Dunckley and Parker 1999; Dunckleyet al. 2001). The second important observation was that the ski2Δ, ski4-1, ski3Δ, ski7Δ, and ski8Δ mutations partially suppressed the temperature sensitivity of the lsm1Δ mutant at 37° (Figure 3A; A. van Hoof and R. Parker, unpublished results). This suggests that the temperature-sensitive growth of the lsm1Δ strain is at least partially due to the 3′ to 5′ degradation of mRNAs.
To test the generality of the suppression of the lsm1Δ thermosensitivity by lesions in the SKI genes, we determined whether the growth defect at 37° caused by the pat1Δ, lsm2ts, lsm5ts, and lsm7Δ lesions could also be suppressed by defects in the SKI genes. For this analysis we used the ski4-1 allele, which is a point mutation in a core component of the exosome and has the strongest effect on mRNA turnover of the ski mutations (van Hoofet al. 2000). In this case, we observed that the lsm2ts ski4-1, lsm5ts ski4-1, and lsm7Δ ski4-1 double mutants could all grow at 37°, although not as well as a wild-type strain (Figure 3B). We also observed that the ski4-1 lesion could very weakly suppress the temperature-sensitive growth of the pat1Δ strain (Figure 3B). Taking these observations together, the ski mutations (ski2Δ, ski3Δ, ski4-1, ski7Δ, and ski8Δ) suppress the temperature-sensitive growth of the lsm and pat1 mutants.
The ski mutations cause increased accumulation of trimmed mRNAs in a lsm1Δ mutant: To determine the mechanism by which the ski mutations were suppressing the lsm1Δ and pat1Δ temperature-sensitive growth we examined the 3′ ends of the MFA2pG and PGK1pG mRNAs in the lsm1Δ ski2Δ, lsm1Δ ski3Δ, lsm1Δ ski4-1, and lsm1Δ ski8Δ double mutants. This analysis led to two important observations. First, we observed that trimming of the PGK1pG and the MFA2pG mRNAs still occurred in the double mutants, because both trimmed full-length and trimmed poly(G) fragments of MFA2pG and PGK1pG mRNAs were detected at 24° (Figure 4) and 37° (data not shown). Therefore the ski2Δ, ski3Δ, ski4-1, and ski8Δ mutations do not prevent trimming in the lsm1Δ mutant. This implies that the ski mutations do not suppress the temperature-sensitive growth of lsm1Δ by preventing trimming. Another implication of this observation is that trimming does not require the SKI-dependent exosome activities and that there must be other nucleases that perform the trimming reaction.
A second important observation was that there were increased levels of trimmed MFA2pG and PGK1pG mRNAs and mRNA fragments in the double mutants as compared to the lsm1Δ alone (Figure 4, A and B). In addition, a second trimmed species shortened by ~20 nucleotides was more prevalent in the double mutant, both for the full-length mRNAs and for the mRNA fragment [FL(3′-20) and FR(3′-20), Figure 4, A and B]. This increased level of the trimmed mRNA fragments is consistent with the previous work demonstrating that these types of mRNA fragments are degraded 3′ to 5′ by the exosome (Anderson and Parker 1998). The increased levels of trimmed full-length mRNAs in the double mutants imply that the trimmed full-length mRNAs are, at least in part, being degraded 3′ to 5′ by the exosome (see discussion).
Interestingly, the ski2Δ (data not shown) and the ski4-1 (Figure 4) mutants also generated the trimmed poly(G) fragments like the lsm1Δ mutant. Two observations argue that the trimmed poly(G) species from the ski2Δ and the ski4-1 mutants have the same structures as those from the lsm1Δ mutant. First, both the PGK1 and the MFA2 trimmed poly(G) fragments migrated in the same way in the ski2Δ, ski4-1, and lsm1Δ mutants. Second, the lsm1Δ ski2Δ and lsm1Δ ski4-1 double mutants produce the same trimmed species as the single mutants. This suggests that when the competing 3′ to 5′ decay pathway mediated by the exosome is blocked, a low level of trimming can occur even in the presence of the LsmI/Pat1 complex.
The ski mutations inhibit the 3′ to 5′ degradation of trimmed mRNAs: The increased amount of trimmed species in the lsm1Δ ski double mutants argues that at least some of the trimmed full-length mRNAs are being degraded by the exosome.
Since this model predicts that mRNAs are being degraded 3′ to 5′ in the lsm1Δ strain, we should be able to detect a RNA fragment degraded to the 3′ side of the poly(G) tract (Anderson and Parker 1998). In addition, the production of this 3′ degraded mRNA should be dependent on the Ski proteins. To test this prediction, we looked at the 5′ end → poly(G) fragment of the PGK1pG mRNA generated by the 3′ to 5′ mRNA degradation using a probe hybridizing to the poly(G) tract and the 5′-flanking region (Muhlrad et al. 1994, 1995). The 5′ end → poly(G) fragment phenotypes were similar at 24° (Figure 5) and 37° (data not shown). Upon treatment with RNase H and oPR70, the ~60-nucleotides long 5′ end → poly(G) fragment was not detected in the wild-type cells (Figure 5). This is so because the majority of the mRNA is processed through the 5′ to 3′ mRNA degradation pathway and any 5′ end → poly(G) fragment produced is rapidly degraded through the 5′ to 3′ degradation pathway. There was no 5′ end → poly(G) fragment in the ski2Δ mutant (Figure 5), which is known to be defective in the 3′ to 5′ degradation. The 5′ end → poly(G) fragment accumulated in the lsm1Δ and dcp1Δ mutants (Figure 5), because it could not be efficiently degraded through the 5′ to 3′ degradation pathway. There was no 5′ end → poly(G) fragment in the lsm1Δ ski2Δ double mutants (Figure 5), indicating that the trimmed species cannot be degraded by the 3′ to 5′ degradation pathway. This implies that the ski2Δ mutation might suppress the temperature-sensitive growth of lsm1Δ by preventing the 3′ to 5′ degradation of the trimmed species (see discussion).
The cytoplasmic LsmI/Pat1p complex has a distinct function in stabilizing the 3′ terminus of mRNAs: Several lines of evidence suggest that the LsmI/Pat1p complex has a specific role in stabilizing the 3′ termini of mRNAs. Initially, this was suggested by the observation that defects in, or depletion of, Lsm1p to Lsm7p, or Pat1p lead to the accumulation of mRNA molecules trimmed into the 3′-UTR ~10–20 nucleotides (Figure 1 and Boecket al. 1998; Bouveretet al. 2000; Schwartz and Parker 2000). This effect is observed with several mRNAs, including the MFA2pG, PGK1pG, and GAL10 transcripts (Figure 1 and data not shown; Boecket al. 1998; Bouveret 2000; Schwartz and Parker 2000). Two observations indicate that the shortening of the mRNAs' 3′ end is due to the absence of the LsmI/Pat1p complex and is not a general consequence of the partial inhibition of decapping that also occurs in these mutants. First, dcp1Δ strains, which are completely blocked for decapping, do not accumulate the 3′ trimmed mRNAs for the MFA2, PGK1, and GAL10 mRNAs (Figure 1 and data not shown). Second, the 3′ trimmed mRNAs accumulate in a dcp1Δ lsm1Δ double mutant (Figure 2). On the basis of these observations, we argue that the LsmI/Pat1p complex has a distinct function to protect the 3′ termini of the mRNAs from a trimming reaction.
The mechanism by which the LsmI/Pat1p complex protects the 3′ termini of mRNAs from degradation is currently unclear. The simplest hypothesis is that this complex binds to the mRNAs in this region and sterically inhibits an exo- or endonuclease. This possibility is supported, but not proven, by two observations. First, it is known that the LsmI/Pat1p complex binds to mRNAs (Tharunet al. 2000). Second, the analogous nuclear Lsm complex is known to bind near the 3′ end of the U6 snRNA (Achselet al. 1999; Mayeset al. 1999). Moreover, the binding of the nuclear Lsm complex to the U6 snRNA appears to be required for the stability of the U6 snRNA (Cooperet al. 1995; Mayeset al. 1999; Pannoneet al. 2001). Specifically, overexpression of U6 snRNA can rescue the temperature-sensitive growth of mutants defective in the nuclear Lsm complex (Mayeset al. 1999; Pannoneet al. 2001). This raises the possibility that both the nuclear and the cytoplasmic Lsm complexes function to protect the 3′ ends of RNAs from degradative reactions.
Previous work has observed that inhibition of decapping by the addition of cycloheximide led to the accumulation of 3′ trimmed MFA2 and PGK1 mRNAs (Boecket al. 1998). This led the authors to the reasonable interpretation that the trimmed species arose because of an inhibition of decapping. However, this is inconsistent with our observation that a dcp1Δ strain does not accumulate trimmed species (Figures 1 and 2). There are several possible explanations for this apparent contradiction. For example, since the mechanism by which cycloheximide inhibits mRNA decapping is unclear, it may be that this drug indirectly affects the interaction of the LsmI/Pat1p complex with mRNAs. This would both inhibit decapping and lead to mRNA trimming. Alternatively, it may be that in different strain backgrounds, or for different mRNAs, the susceptibility of the mRNAs to trimming may be different. This is based on the observation that in our strain background we do not observe trimming of the MFA2 transcript following the addition of cycloheximide (Beelman and Parker 1994).
Shortening of the mRNA 3′ end occurs by an unknown mechanism: Our results suggest that the shortening of the 3′ end observed in the lsm and pat1 mutant strains is different from the previously observed mRNA degradation processes. The 3′ to 5′ degradation of the mRNA body following deadenylation requires the exosome as well as the Ski2p, Ski3p, Ski7p, and Ski8p (Anderson and Parker 1998; van Hoofet al. 2000). We have demonstrated that the accumulation of the 3′ trimmed species is independent of these proteins (Figures 4 and 5). However, because the levels of the trimmed species are increased in the ski2Δ, ski4-1, lsm1Δ ski2Δ, lsm1Δ ski3Δ, lsm1Δ ski4-1, and lsm1Δ ski8Δ strains, the degradation of the trimmed species occurs, at least in part, by the normal exosome-mediated 3′ to 5′ decay pathway of mRNA. These results indicate that a different, as-yet-unidentified exo- or endonuclease is able to remove the 3′ terminal portion of the mRNA, but is generally inhibited from further degradation of the mRNA. This raises the interesting implication that there is a specific organization of the 3′ end of the mRNP that can allow mRNA trimming to only a limited extent. Moreover, because multiple mRNAs show the same behavior this would have to be a shared feature of mRNP organization.
Phenotypic consequences of lsm and pat1 mutations: Several observations suggest that at least part of the reason that the lsm1Δ and pat1Δ strains die at high temperature is inappropriate RNA degradation in a 3′ to 5′ direction. We observed that the temperature-sensitive growth of the lsm1Δ and pat1Δ strains could be at least partially suppressed by lesions in the known 3′ to 5′ cytoplasmic mRNA degradation machinery. There are two general possibilities for how the lesions in 3′ to 5′ mRNA decay machinery could suppress the temperature sensitivity of the lsm and pat1Δ strains. First, it could be that the cytoplasmic exosome degrades an as-yet-to-be-identified essential cytoplasmic noncoding RNA whose stability requires the Lsm proteins. To date, we have not seen any differences in the 3′ ends of stable cytoplasmic 5.8S, 18S, and 25S rRNAs and SCR1 RNA (a small cytoplasmic RNA that is a component of the SRP; Hann and Walter 1991) in mutants defective in the LsmI/Pat1p complex (data not shown). These observations are consistent with the possibility that the LsmI/Pat1p complex specifically protects the 10–20 nucleotides at the 3′ end of mRNAs.
An alternative possibility for the requirement for Lsm proteins for growth at high temperature is that the LsmI/Pat1p complex inhibits the trimming of certain mRNAs, whose function would be inactivated by 3′ trimming. In this view, the ski mutations would suppress the temperature sensitivity of the lsm and pat1 lesions by stabilizing the trimmed mRNAs. A prediction of the above hypothesis is that trimmed mRNAs might show increased rates of 3′ to 5′ degradation by the exosome. However, so far we have been unable to directly demonstrate such a change in the rate of 3′ to 5′ mRNA decay for the MFA2pG transcript in a strain where trimming is occurring (A. van Hoof and R. Parker, unpublished observation). This suggests that if trimming does lead to increased rates of 3′ to 5′ mRNA degradation it is a relatively small effect (<150% of the rate of full-length mRNA based on kinetic modeling; C. Cao and R. Parker, unpublished observations), or is more pronounced on specific mRNAs. In support of mRNA-specific effects of trimming, we have observed that some mRNAs are trimmed by >30 nucleotides into the 3′-UTR in lsm1Δ strains (W. Olivas and R. Parker, unpublished observation). In the extreme, if any essential mRNA was either trimmed into the coding region or subjected to deleterious degradation as a consequence of trimming, this would be sufficient to explain the temperature-dependent growth phenotypes. An alternative possibility is that trimming of mRNAs compromises the function of at least one essential mRNA in some other manner. For example, trimming might affect translation or localization of some mRNAs. In this view, the suppression would occur because stabilization of the trimmed species would allow for prolonged time for increased function of the transcript, thereby compensating indirectly for the defect in mRNA function. Future experiments examining the function and metabolism of the trimmed mRNAs should help to distinguish these possibilities.
We thank Ambro van Hoof and Sundaresan Tharun for their helpful comments and discussions. We thank C. J. Decker for the generous gift of the pCD61 plasmid. This work was supported by a grant from the National Institutes of Health (GM-45443) and funds from the Howard Hughes Medical Institute to R.P.
Communicating editor: S. Sandmeyer
- Received March 15, 2001.
- Accepted May 23, 2001.
- Copyright © 2001 by the Genetics Society of America