The major pathway of mRNA degradation in yeast occurs through deadenylation, decapping and subsequent 5′ to 3′ exonucleolytic decay of the transcript body. To identify proteins that control the activity of the decapping enzyme, which is encoded by the DCP1 gene, we isolated a high‐copy suppressor of the temperature‐sensitive dcp1‐2 allele, termed DCP2. Overexpression of Dcp2p partially suppressed the dcp1‐2 decapping defect. Moreover, the Dcp2 protein was required for the decapping of both normal mRNAs and aberrant transcripts that are degraded by the mRNA surveillance pathway. The Dcp2 protein contains a MutT motif, which is found in a class of pyrophosphatases. Mutational analyses indicated that the region of Dcp2p containing the MutT motif is necessary and sufficient for Dcp2p's function in mRNA decapping. The Dcp2p also coimmunoprecipitates with the DCP1 decapping enzyme and is required for the production of enzymatically active decapping enzyme. These results suggest that direct or indirect interaction of Dcp1p with Dcp2p is required for the production of active decapping enzyme, perhaps in a process requiring the hydrolysis of a pyrophosphate bond.
The stability of mRNA is an important determinant in gene expression (for reviews, see Ross, 1995; Caponigro and Parker, 1996; Jacobson and Peltz, 1996). An important goal will be to define the pathways of mRNA degradation and the mechanisms by which mRNA decay pathways are modulated. Recent results have indicated that two general pathways of mRNA decay function in the yeast Saccharomyces cerevisiae. The predominant decay pathway occurs via shortening of the poly(A) tail, followed by removal of the 5′ cap structure by the Dcp1 decapping enzyme (Beelman et al., 1996), which exposes the body of the transcript to Xrn1p‐dependent 5′ to 3′ exonucleolytic degradation (Decker and Parker, 1993; Hsu and Stevens, 1993; Muhlrad et al., 1994, 1995). A second general mRNA decay pathway occurs by deadenylation of polyadenylated mRNA to an oligo(A) tail length followed by 3′ to 5′ exonucleolytic decay of the transcript body (Muhlrad et al., 1995; Anderson and Parker, 1998).
Several lines of evidence indicate that these decay pathways identified in yeast are conserved in other eukaryotes. For example, deadenylated, decapped, full‐length mRNA species have been detected from murine liver cells (Couttet et al., 1997). Intermediates in mRNA decay that are shortened at their 5′ ends have also been identified in both plant and animal cells (Lim and Maquat, 1992; Higgs and Colbert, 1994; Gera and Baker, 1998). An important role for 3′ to 5′ mRNA decay in plants has also been demonstrated (Higgs and Colbert, 1994). Finally, the yeast proteins involved in these general mRNA decay pathways, such as Xrn1p, Ski2p, Ski6p/Rrp41p and Rrp4p, have homologs in higher eukaryotes (Dangel et al., 1995; Lee et al., 1995; Bashkirov et al., 1997; Mitchell et al., 1997).
Decapping is a key step in the 5′ to 3′ mRNA degradation pathway since it both precedes and permits the decay of the transcript body. Since individual mRNAs are decapped at different rates, decapping is also an important control point in mRNA decay (Muhlrad et al., 1994, 1995). Decapping also plays a critical role in the process of mRNA surveillance wherein aberrant nonsense codon‐containing mRNAs are very rapidly decapped and degraded in a 5′ to 3′ direction (Muhlrad and Parker, 1994; Hagan et al., 1995; Beelman et al., 1996). The product of the DCP1 gene has been shown to be necessary for mRNA decapping in vivo and sufficient for decapping in vitro, suggesting that the DCP1 gene encodes a yeast mRNA decapping enzyme (Beelman et al., 1996; LaGrandeur and Parker, 1998). Understanding how the activity of the DCP1 decapping enzyme is regulated in vivo will be critical for understanding the basis for differential mRNA decay rates. However, to date little is known about the factors that modulate the activity of this key enzyme.
To identify proteins that affect the activity of the DCP1 decapping enzyme we have performed a genetic screen. The design of this screen exploits the observation that loss‐of‐function mutations in the Dcp1p decapping enzyme or the Xrn1p cytoplasmic 5′ to 3′ exoribonuclease are synthetically lethal in combination with mutations in proteins that are required for normal 3′ to 5′ mRNA degradation, such as Ski2p, Ski3p, Ski8p or Rrp41p/Ski6p (Johnson and Kolodner, 1995; Anderson and Parker, 1998). One implication of this finding is that mRNA decay is required for cell growth. We have made the synthetic lethality of dcp1Δ ski8Δ double mutants conditional using a temperature‐sensitive allele of the DCP1 decapping enzyme, termed dcp1‐2 (Tharun and Parker, 1999). To then identify proteins involved in mRNA decapping, we isolated high‐copy suppressors of the temperature‐sensitive growth defect of the dcp1‐2 ski8Δ strain. Here we describe the properties of one such suppressor, referred to as DCP2, which suppressed the growth defect of dcp1‐2 ski8Δ mutants and is also required for decapping in vivo.
Identification of high‐copy suppressors of the conditional growth defect of dcp1‐2 ski8Δ mutants
To identify proteins involved in mRNA decay, we performed a genetic screen for high‐copy suppressors of dcp1‐2 ski8Δ mutants. The dcp1‐2 ski8Δ strain grows at 24°C, the permissive temperature for dcp1‐2 function (Tharun and Parker, 1999). However, the double mutant fails to grow at temperatures >30°C, which is the restrictive temperature for dcp1‐2 function (Tharun and Parker, 1999). The restrictive temperature for growth of the dcp1‐2 ski8Δ mutant correlates with the temperature at which mRNA decay is severely inhibited (data not shown), suggesting that the inability of this mutant to grow results from a block to mRNA decay. We therefore reasoned that suppressors that restore the ability of the dcp1‐2 ski8Δ double mutant to grow at the restrictive temperature would do so by restoring mRNA decay.
We screened five genome equivalents of a yeast 2μ genomic DNA library for genes that, when overexpressed, would suppress the growth defect of the dcp1‐2 ski8Δ double mutant. In cases where growth at 34°C was restored, we showed plasmid dependence of the growth phenotype through isolating the plasmid, retransforming the parental dcp1‐2 ski8Δ double mutant, and retesting the growth phenotype. We identified 42 plasmid‐dependent suppressors. Four of these plasmids were found to contain the DCP1 gene and 34 were shown to contain SKI8. In addition, we identified three plasmids containing the PSU1/DCP2 gene (ORF YNL118C) as the only common open reading frame. Subsequent testing of only the PSU1/DCP2 gene demonstrated that this gene was sufficient to restore growth to the dcp1‐2 ski8Δ mutant, although not to wild‐type levels (Figure 1). Specifically, the dcp1‐2 ski8Δ mutant overexpressing DCP2/PSU1 has a growth rate of 190 min at 34°C compared with 90 min for an isogenic wild‐type strain. Finally, one plasmid was found that contained six open reading frames. Of these, YGL222C was shown to be the gene responsible for suppression (data not shown). Characterization of YGL222C will be reported in future work. Here we define an important role for PSU1/DCP2 in mRNA decapping. The PSU1/DCP2 gene has been identified previously as a high‐copy suppressor of a nuclear petite mutant (A.Tzagaloff, personal communication). However, based on the identification of a critical role of this protein in mRNA decapping presented here, we will refer to the PSU1 gene hereafter as the DCP2 gene.
Overexpression of Dcp2p partially rescues the mRNA degradation defect of the dcp1‐2 mutation
In principle, the rescue of the dcp1‐2 ski8Δ double mutant's growth defect by DCP2 overexpression could be accompanied by some restoration of either 5′ to 3′ or 3′ to 5′ mRNA degradation. To determine whether Dcp2p overexpression restored mRNA decay through either of these pathways we made use of a galactose‐inducible MFA2 mRNA that contains a poly(G) insertion in its 3′ UTR, termed MFA2pG (Decker and Parker, 1993). By blocking the exonucleases that degrade mRNA, poly(G) insertions can be used to trap intermediates in the degradation of mRNA (Vreken and Raue, 1992; Decker and Parker, 1993; Anderson and Parker, 1998). Importantly, the specific degradation intermediates that are trapped can be used to infer the directionality of mRNA decay (Figure 2A; see also He and Parker, 1999). Therefore, we introduced the DCP2 gene on a high‐copy number plasmid into either dcp1‐2 or ski8Δ mutants that also contained the MFA2pG mRNA and determined the mRNA decay phenotype of the resulting strains (see Materials and methods).
Dcp2p overexpression partially rescues the mRNA degradation defect of the dcp1‐2 mutation at the restrictive temperature (Figure 2, compare C and D). This effect is seen in the dcp1‐2 mutant overexpressing Dcp2p as a small but clear increase in the level of mRNA degradation intermediate corresponding to that produced from 5′ to 3′ mRNA decay, referred to as the poly(G)→3′ end fragment (Anderson and Parker, 1998). Consistent with only a partial restoration of 5′ to 3′ mRNA decay, Dcp2p overexpression did not restore production of the poly(G)→3′ end fragment to wild‐type levels. These results argue that only a low level of decapping is required for viability in combination with the ski8Δ, and are consistent with the observations that mutations in genes causing a partial block to decapping are not synthetically lethal with lesions preventing 3′ to 5′ decay (J.S.Anderson and R.Parker, unpublished observation). Dcp2p overexpression did not restore 5′ to 3′ mRNA degradation in a dcp1Δ mutant (Figure 2, compare E and F). This observation indicated that Dcp2p overexpression was not a bypass suppressor of dcp1 mutations. Finally, DCP2 overexpression had no detectable effect on the 3′ to 5′ mRNA decay defect of the ski8Δ mutant (data not shown), suggesting that Dcp2p overexpression does not significantly affect 3′ to 5′ mRNA decay.
Dcp2p is required for 5′ to 3′ mRNA degradation
Since Dcp2p overexpression affects 5′ to 3′ decay in a Dcp1p‐dependent fashion, we hypothesized that Dcp2p normally functions in the 5′ to 3′ mRNA decay pathway. To determine whether Dcp2p has a role in mRNA decay, we deleted the DCP2 gene from the genome through homologous recombination and assayed the phenotypes resulting from loss of Dcp2p. The Dcp2p is not required for viability in our strain background at any temperature tested from 18 to 37°C, though DCP2 has been reported to be an essential gene in other strain backgrounds (A.Tzagaloff, personal communication). However, the dcp2Δ strain grows extremely slowly at all temperatures, a phenotype comparable to that resulting from loss of the Dcp1p decapping enzyme (Beelman et al., 1996).
Analysis of mRNA degradation in the dcp2Δ mutant showed that Dcp2p was required for normal rates of decay of both stable and unstable transcripts (Table I). Specifically, loss of Dcp2p function resulted in an ∼2‐fold stabilization of the PGK1pG mRNA (Figure 3) and a >3‐fold stabilization of the MFA2pG mRNA (Figure 4). The dcp2Δ also caused a 2‐fold stabilization of the GAL1 and GAL7 mRNAs and a 4‐fold stabilization of the GAL10 mRNA compared with wild type. Moreover, the poly(G)→3′ end fragment that is normally produced upon 5′ to 3′ degradation of the PGK1pG mRNA is not detectable in dcp2Δ mutants, suggesting a complete block to 5′ to 3′ mRNA decay (Figure 3). Consistent with a complete block to 5′ to 3′ decay, the residual decay rates that we observed in the dcp2Δ strain were very similar to those seen in a dcp1Δ strain (Table I) and result from the transcripts now being degraded by the alternative 3′ to 5′ decay pathway (see Anderson and Parker, 1998). These observations indicated that Dcp2p is required for the 5′ to 3′ degradation of a variety of transcripts.
Dcp2p is required for 5′ to 3′ decay following deadenylation
To define the role of Dcp2p in 5′ to 3′ mRNA decay, it was important to determine at which step 5′ to 3′ decay is blocked in the dcp2Δ mutant. In principle, any of the steps of deadenylation, decapping, or 5′ to 3′ exonucleolytic decay could be affected. To distinguish between these possibilities, we first examined the steady‐state poly(A)‐tail distribution of full‐length MFA2pG mRNA. In wild‐type cells, the MFA2pG transcripts exist with a distribution of poly(A) tails from ∼75 nucleotides to an oligoadenylated length of ∼10–12 nucleotides, which is the length at which the transcript becomes a substrate for decapping (Decker and Parker, 1993). In mutants such as dcp1Δ or xrn1Δ, which inhibit mRNA decapping or 5′ to 3′ exonucleolysis, oligoadenylated, full‐length mRNA accumulates (Muhlrad et al., 1994; Beelman et al., 1996; Hatfield et al., 1996; Boeck et al., 1998). If loss of Dcp2p blocks decapping or 5′ to 3′ exonucleolysis, oligoadenylated mRNA species should be stabilized and therefore should accumulate in the dcp2Δ mutant.
To determine the steady‐state poly(A)‐tail distribution of full‐length MFA2pG mRNA in the dcp2Δ mutant, we analyzed RNA from time points following glucose repression of the galactose‐induced MFA2pG mRNA on polyacrylamide Northern gels to resolve polyadenylated and oligoadenylated mRNA species (Figure 4). Using this analysis, we found that the dcp2Δ mutant shifts the distribution of polyadenylated mRNA into the oligoadenylated pool at the early time points. Additionally, the oligoadenylated form of the mRNA persists at the later time points, suggesting that loss of Dcp2p function stabilizes the oligoadenylated form of the mRNA. The stabilization of oligoadenylated, full‐length mRNA species indicated that 5′ to 3′ mRNA decay in the dcp2Δ mutant was blocked following deadenylation, either at the decapping or 5′ to 3′ decay steps. Further evidence that loss of Dcp2p function does not significantly affect deadenylation was obtained in transcriptional pulse–chase experiments wherein mRNA from the dcp2Δ mutant underwent poly(A) shortening with essentially wild‐type kinetics (data not shown).
Oligoadenylated mRNA species are capped in dcp2Δ mutants
To distinguish between a role for Dcp2p in decapping or in 5′ to 3′ exonucleolysis, we performed immunoprecipitation experiments using antibody specific for the 5′ cap structure of the mRNA (see Materials and methods). In dcp1Δ mutants, which block mRNA decapping, the mRNA is immunoprecipitable using these anti‐cap antibodies (Figure 5; see also Beelman et al., 1996). However, in xrn1Δ mutants, which block 5′ to 3′ exonucleolysis following decapping, only polyadenylated mRNAs are immunoprecipitable (Figure 5; see also Muhlrad et al., 1994). Results for the dcp2Δ mutant showed that the oligoadenylated, full‐length MFA2pG mRNA from a dcp2Δ strain was immunoprecipitable using these antibodies (Figure 5), indicating that the deadenylated mRNA that accumulated in dcp2Δ mutants was capped. This finding indicated that Dcp2p is required for mRNA decapping in vivo.
Dcp2p is required for deadenylation‐independent decapping
Decapping of mRNA is a key step not only in the deadenylation‐dependent decay pathway but also in the deadenylation‐independent decay pathway, which acts to degrade aberrant mRNAs. In this pathway, aberrant mRNAs, such as those containing premature translation termination codons, are rapidly decapped and degraded independently of deadenylation (Muhlrad and Parker, 1994). The same DCP1 decapping enzyme that functions in the normal mRNA decay pathway is required for the decapping of aberrant nonsense codon‐containing mRNAs through the nonsense‐mediated decay pathway (Beelman et al., 1996). However, a second class of proteins that have non‐essential roles in deadenylation‐dependent decapping, such as Mrt1p, Mrt3p and Spb8p are not required for deadenylation‐independent decapping of aberrant mRNAs (Hatfield et al., 1996; Boeck et al., 1998). Since Dcp2p was essential for the decapping of mRNAs through the deadenylation‐dependent decapping pathway, an important question was whether Dcp2p was also required for deadenylation‐independent decapping of nonsense codon‐containing mRNA.
To determine whether Dcp2p is required for nonsense‐mediated decay, we measured the half‐life of a mutant PGK1 mRNA that contains an early nonsense codon, termed B55TPGK1N103pG (Muhlrad and Parker, 1994), in both wild‐type and dcp2Δ yeast strains. Consistent with earlier work, in wild‐type cells the B55TPGK1N103pG mRNA has a half‐life of <2 min (Figure 6, top). The same mRNA in the dcp2Δ mutant was stabilized >20‐fold (Figure 6, bottom). This observation indicated that Dcp2p was required for the rapid, deadenylation‐independent decapping of nonsense codon‐containing mRNA. Interestingly, the Dcp2 protein was also identified previously in a two‐hybrid screen as interacting with the Upf1p (He and Jacobson, 1995; referred to as NMD1). Since the Upf1p is a critical factor in triggering the degradation of aberrant mRNAs during mRNA surveillance (for review see Jacobson and Peltz, 1996), this identifies a set of interactions linking premature translation termination and the mRNA decapping machinery, with possible mechanistic implications (see Discussion).
Dcp2p contains a MutT motif that is required for mRNA decapping
The observations above indicated that Dcp2p was required for mRNA decapping in vivo. An important issue was the role of the Dcp2p in mRNA turnover. Insight into a possible function for Dcp2p came from an analysis of the protein sequence. The Dcp2 protein contains a conserved 23 amino acid MutT motif in its N‐terminus (amino acids 134–156). This MutT motif is also conserved in several homologs of Dcp2p found in the databases (Figure 7). The MutT motif is found in a class of proteins that hydrolyze pyrophosphate linkages (for reviews see Koonin, 1993; Bessman et al., 1996). Moreover, mutagenic studies have identified specific residues within the MutT motif that are required for the catalytic activity of these enzymes (Lin et al., 1996; Safrany et al., 1998). The presence of a conserved MutT motif in Dcp2p suggested that this protein might also be involved in the hydrolysis of pyrophosphate bonds, of which the mRNA cap structure contains two. For this reason, we determined if the MutT motif was important for Dcp2p's function in decapping.
To determine if the region of Dcp2p containing the MutT motif was sufficient for Dcp2p‘s function in mRNA decapping we constructed two N‐terminally FLAG‐tagged deletion mutants of Dcp2p (see Materials and methods). One deletion, dcp2‐ΔC, consists of the FLAG peptide followed by the N‐terminal 300 amino acids, which contains the MutT motif and is the most highly conserved region of the Dcp2p. The second deletion, dcp2‐ΔN, consists of the FLAG peptide followed by the C‐terminal 670 amino acids of Dcp2p. Both deletions were expressed from the GPD promoter in a dcp2Δ background. Expression of either a full‐length FLAG–Dcp2p fusion protein or of the N‐terminal 300 amino acids completely rescued both the slow growth (data not shown) and mRNA decapping defect of the dcp2Δ mutation, as measured by the production of wild‐type levels of the poly(G)→3′ end fragment (Figure 8A). This result indicated that the N‐terminal 300 amino acids of Dcp2p, which contain the MutT motif, are sufficient to perform Dcp2p’s function in mRNA decapping. In contrast, the dcp2‐ΔN construct failed to rescue either the slow growth (data not shown) or mRNA decay defect of the dcp2Δ mutant (Figure 8A). Consistent with the dcp2‐ΔN protein being non‐functional, this polypeptide fragment is not stably expressed in vivo (see below; Figure 9B) and suggests that the N‐terminus of Dcp2p is required both for its decapping function and for the production of stable protein.
To determine if the MutT motif was required for mRNA decapping we introduced mutations into the Dcp2p MutT motif at key residues that have previously been shown to be essential for the catalytic activity of other MutT motif‐containing proteins (Lin et al., 1996; Safrany et al., 1998). Specifically, changing the C‐terminal‐conserved glutamic acid in the MutT motif to glutamine has been shown to reduce greatly the enzymic activity of both the canonical Escherichia coli MutT protein and the rat diphosphoinositol polyphosphate phosphohydrolase (DIPP). We therefore constructed different combinations of mutations incorporating this conserved glutamic acid residue, which corresponds to glutamic acid 153 in the Dcp2 protein. In one allele, dcp2‐1, we changed four of the conserved amino acids in the MutT motif to alanine (Arg148, Glu149, Glu152 and Glu153). As shown in Figure 8B, the dcp2‐1 allele was unable to promote the decapping of the MFA2pG mRNA as measured by the lack of production of the poly(G)→3′ end fragment. Similarly, the dcp2‐2 allele, which contains both an E152Q mutation and the key E153Q mutation, was non‐functional in mRNA decapping. The dcp2‐3 allele, containing the single E152Q mutation, was functional for decapping as shown by the production of near wild‐type levels of the poly(G)→3′ end fragment. Importantly, the single E153Q mutation in the dcp2‐4 allele completely eliminated Dcp2p‘s function in decapping. The effects of these mutations on decapping probably do not result from differences in the expression of the mutant proteins since epitope‐tagged versions of these proteins show the same decapping phenotypes and are expressed to similar levels (data not shown). Since the identical mutation in the homologous glutamic acid residue of E.coli MutT and rat DIPP greatly reduces the catalytic activity of those enzymes, this result with the E153Q mutation strongly suggested that the activity of Dcp2p’s MutT motif is required for mRNA decapping. This implies that Dcp2p may function in some manner as a pyrophosphatase that is required for the decapping of mRNA (see Discussion).
Purified Dcp2p lacks mRNA decapping activity itself in vitro but coimmunoprecipitates with active Dcp1p
Since Dcp2p was required for decapping and contained a functional MutT motif, a simple possibility was that the Dcp2p was a second decapping enzyme. In order to test this possibility, we purified the FLAG–Dcp2p fusion proteins described above from a wild‐type strain and assayed for their ability to decap a synthetic mRNA in vitro (Beelman et al., 1996; LaGrandeur and Parker, 1998; see also Materials and methods). To avoid any possible contaminating decapping activity from the Dcp1p decapping enzyme, we also purified the FLAG–Dcp2p proteins from dcp1Δ strains.
A silver‐stained SDS–polyacrylamide gel of the purified FLAG–Dcp2p fractions showed that this affinity selection yielded proteins of the expected size for both the wild‐type Dcp2p and for the dcp2‐ΔC mutant (Figure 9A). A Western blot using anti‐FLAG antibody to detect the presence of the FLAG epitope confirmed that the proteins of the expected size were indeed the FLAG–Dcp2 fusion proteins. Two FLAG fusion proteins were present in the full‐length Dcp2p preparation. One species corresponds to ∼110 kDa, the predicted molecular weight of the full‐length FLAG–Dcp2 fusion protein. The second fusion protein has a mobility of 44 kDa. At present it is unclear if the smaller molecular weight protein represents an artifactual in vitro degradation product, or if this form of the protein represents a relevant in vivo processing reaction. The protein detected by the FLAG antibodies in the dcp2‐ΔC preparation corresponds to the predicted molecular weight of this fusion protein. No specific proteins were detectable by Western analysis in the preparation from strains expressing the dcp2‐ΔN fusion (Figure 9B). Additionally, we have been unable to detect expression of this fusion protein in crude cell extracts (data not shown). This indicates that the dcp2‐ΔN fusion protein is not stably expressed in vivo.
To determine if Dcp1p interacts with Dcp2p, we performed Westerns on the partially purified FLAG–Dcp2p fractions using antibodies specific for the Dcp1 protein. As shown in Figure 9C, Dcp1p interacts with the full‐length FLAG–Dcp2 fusion protein and with the dcp2‐ΔC fusion protein. Dcp1p is not present in the dcp2‐ΔN preparation, which contains no Dcp2p fusion protein (Figure 9B). The Dcp1p antibodies detect no band in either the full‐length FLAG–Dcp2p or the dcp2‐ΔCp fractions from the dcp1Δ strain (Figure 9C). This result indicated that the protein interacting with the full‐length Dcp2p and with the dcp2‐ΔC protein from the wild‐type strain was Dcp1p. It should be noted that the Dcp1p is not present at stoichiometric amounts when compared with the levels of purified Dcp2p. In part, this may result from the fact that the Dcp2p is overexpressed in this experiment to allow the purification of a significant amount of this polypeptide for biochemical assays (see below). However, two observations argue that this interaction is significant. First, the presence of Dcp1p in the fractions requires the presence of the Dcp2 protein (Figure 9C). Secondly, the suppression of dcp1‐2 with overexpression of Dcp2p is consistent with some type of interaction between these proteins (Figure 2). These results suggest that Dcp1p and Dcp2p interact physically in vivo, either directly or through interactions with additional proteins.
We next assayed for decapping activity of the purified Dcp2p preparations using an in vitro decapping assay described previously (Beelman et al., 1996; LaGrandeur and Parker, 1998). In this assay, the decapping of an in vitro‐transcribed mRNA containing a radiolabeled cap is monitored over time using PEI‐cellulose thin‐layer chromatography (TLC) to separate reaction products. Using this assay we found that both the purified full‐length Dcp2p and the purified dcp2‐ΔCp from the wild‐type strain were active for decapping, as shown by the production of the 7mGDP decapping product (Figure 9D). The purified dcp2‐ΔNp fraction containing neither Dcp1p nor FLAG–Dcp2p protein lacked detectable decapping activity. Importantly, full‐length Dcp2p and dcp2‐ΔCp purified from the dcp1Δ strain lacked any detectable decapping activity. This finding shows that, although decapping activity copurifies with Dcp2p, the activity is dependent upon the presence of the Dcp1p decapping enzyme. Thus, under these in vitro conditions, Dcp2p is not itself capable of decapping an mRNA (see Discussion).
Dcp1 protein levels are unaffected by loss of Dcp2p
Since Dcp2p was essential for decapping in vivo but does not appear to be a decapping enzyme in vitro, we explored other possible functions for Dcp2p in decapping. One possibility was that Dcp2p was required for the expression of Dcp1p, and that the defect in decapping that we observed resulted from a decrease or a loss of Dcp1p. To test this hypothesis, we determined whether dcp2Δ affected the expression of the Dcp1p decapping enzyme by Western analysis using anti‐Dcp1p antibodies. We found that Dcp1p expression was comparable between the wild‐type strain and the dcp2Δ mutant (Figure 10A). This observation indicated that the block to decapping seen in the dcp2Δ mutant was not the result of an effect on the level of expression of Dcp1p.
Dcp2p is required for the production of active Dcp1p
The observation that Dcp1p was present at wild‐type levels in dcp2Δ mutants suggested the possibility that the Dcp1 protein made in the dcp2Δ mutant was non‐functional, possibly due to the lack of a required modification. This possibility was based on the prior observations that Dcp1p is modified and that only a fraction of the protein purified from yeast cells is enzymatically active (LaGrandeur and Parker, 1998). To test the possibility that Dcp2p is required for the production of active Dcp1 decapping enzyme, we purified a FLAG‐tagged version of Dcp1p from wild‐type DCP2 and dcp2Δ strains and assayed for the ability of the purified Dcp1p to cleave a capped mRNA in vitro using the assay described above.
An important observation was that the purified FLAG–Dcp1p from dcp2Δ strains was inactive for decapping in vitro (Figure 10C). In contrast, the Dcp1p purified from the DCP2 strain was active in vitro. This suggested either that Dcp2p was required for the activation of Dcp1p, or that the purified FLAG–Dcp1p fraction from the dcp2Δ strain contained an inhibitor of decapping. The presence of a diffusible inhibitory molecule in the preparations from the dcp2Δ strain is unlikely, since the addition of active Dcp1p to these preparations yielded decapping activity (Figure 10C, last lane). It should be noted that, although additional proteins copurified with Dcp1p in these preparations, the additional proteins from the DCP2 and dcp2Δ strains were indistinguishable between Dcp1p preparations. This argues that the inactivity of Dcp1p from the dcp2Δ strain does not result from copurification of a polypeptide that is inhibitory to decapping. Therefore, these results suggested that the Dcp1p made in the dcp2Δ strain is inactive for decapping and imply that Dcp2p is required for the production of active Dcp1p decapping enzyme (see Discussion). Interestingly, addition of partially purified FLAG–Dcp2p to the inactive FLAG–Dcp1p did not restore decapping activity in vitro (data not shown), indicating that purified Dcp2p is not sufficient to reconstitute decapping activity with the inactive Dcp1p. This observation suggests either that additional proteins are required for Dcp2p to activate Dcp1p or that the purified fractions do not contain the substrate for Dcp2p, the hydrolysis of which may activate Dcp1p (see Discussion).
Dcp2p is required for mRNA decapping
Several observations showed that Dcp2p was required for mRNA decapping in vivo. First, both stable and unstable mRNAs were stabilized in dcp2 mutants (Table I). Secondly, dcp2Δ mutants accumulated oligoadenylated, full‐length mRNAs (Figure 4), indicating that the block to decay occurs following deadenylation. Thirdly, immunoprecipitation experiments with anti‐cap antibody showed that these deadenylated mRNAs accumulating in dcp2Δ mutants were capped (Figure 5). Fourthly, the Dcp2 protein was required for the decapping of mRNAs that contain premature translation termination codons (Figure 6). Since Dcp2p was required for both deadenylation‐independent and ‐dependent decapping, and no 5′ to 3′ decay products were observed in dcp2Δ strains (Figure 3), we conclude that Dcp2p, like the DCP1 decapping enzyme (Beelman et al., 1996), is required for all mRNA decapping in vivo and is therefore likely to be a critical component of the mRNA decay machinery. This is in contrast to other proteins such as Mrt1p, Mrt3p and Spb8p, which affect the efficiency of decapping, but are not absolutely required for decapping (Hatfield et al., 1996; Boeck et al., 1998).
Dcp2p is required for the production of active Dcp1p
The decapping defect seen in the dcp2Δ strains can be attributed, at least in part, to the absence of active Dcp1p. The critical observation was that although Dcp1p is present at normal levels in a dcp2Δ mutant (Figure 10A), when purified this Dcp1p is inactive for decapping in vitro (Figure 10C). This suggests that the inactive Dcp1p made in the absence of Dcp2p may be significantly altered in some way from the active Dcp1 protein made in the wild‐type strain (see below). The simplest interpretation of these observations is that Dcp2p is required to produce active Dcp1p enzyme and that this is the basis for the decapping defect seen in the dcp2Δ strains. However, it should be noted that we cannot rule out the possibility that Dcp2p has additional roles in mRNA decapping.
Dcp2p is likely to function as a pyrophosphatase
Consideration of the specific role of Dcp2p should include the inference that Dcp2p is likely to function as a pyrophosphatase. This was first suggested by the fact that Dcp2p contains a MutT motif, which has been described previously in a variety of pyrophosphatases (Koonin, 1993; Bessman et al., 1996; Safrany et al., 1998). Three observations suggest that this MutT motif is central to the function of Dcp2p in decapping. First, the region of Dcp2p containing the MutT motif and the MutT motif itself are conserved in several Dcp2p homologs in other species (Figure 7). Secondly, deletion analysis of the Dcp2 protein showed that the region of the protein containing the MutT motif was sufficient to rescue the decapping defect of the dcp2Δ mutation. Thirdly, based on prior work on other MutT‐like enzymes, mutation of the key glutamic acid residue in the Dcp2p MutT motif eliminated the function of Dcp2p in mRNA decapping (Figure 8). The simplest interpretation of these observations is that Dcp2p's role in decapping is to hydrolyze a specific pyrophosphate bond, which must be cleaved to allow decapping to occur. However, it should be noted that we cannot rule out other possible functions of the MutT motif, perhaps in a novel role as a protein phosphatase.
Possible mechanisms of Dcp2p function
Based on the points discussed above there are two critical, and possibly related, issues for understanding the role of Dcp2p in mRNA decay. First, what is the actual bond, pyrophosphate or other, cleaved by Dcp2p? Secondly, why is Dcp1p produced in an inactive form in dcp2Δ strains? One formal possibility was that Dcp2p is the decapping enzyme and that all Dcp1p preparations previously tested for decapping activity were contaminated with Dcp2p. Three observations make this possibility unlikely. First, separation of Dcp1p from other proteins by denaturing gel electrophoresis followed by renaturation of Dcp1p yields a protein capable of decapping an mRNA substrate (LaGrandeur and Parker, 1998). Secondly, fractionation of purified Dcp1p by size exclusion chromatography yielded a peak of active enzyme at 30 kDa, the expected size for Dcp1p (T.E.LaGrandeur and R.Parker, unpublished observation). Thirdly, we have not been able to demonstrate cleavage of an mRNA substrate by purified Dcp2p under standard (Figure 9) or a variety of other conditions (data not shown), wherein highly purified Dcp1p does have decapping activity (LaGrandeur and Parker, 1998). We interpret these observations to indicate that Dcp1p is a decapping enzyme and that Dcp2p is somehow required for its function.
Given the possible pyrophosphatase function of Dcp2p, a working hypothesis is that Dcp2p activates Dcp1p by cleaving a pyrophosphate bond that either relieves a specific inhibitory effect or provides an activation of Dcp1p function. For example, the Dcp2 protein could be required either for an activating covalent modification of Dcp1p or to prevent an inactivating modification of Dcp1p. In this view, the Dcp1p that is produced in dcp2Δ strains would never be properly modified and would therefore be inactive for decapping. However, we have been unable to observe any difference in mobility between wild‐type Dcp1p and Dcp1p purified from dcp2Δ strains by SDS–PAGE (data not shown), suggesting that if there is a relevant modification it is of low molecular weight. The Dcp1 protein is a phosphoprotein (LaGrandeur and Parker, 1998). However, the form of the phosphorylation is unknown and could consist of covalent linkage of a small pyrophosphate‐containing molecule. It may be that both active and inactive forms of Dcp1p are phosphorylated and that the enzymic activity of Dcp2p's MutT motif functions to alter the form of the phosphorylation, perhaps by releasing the covalently bound small molecule, thereby converting inactive Dcp1p to active Dcp1p. Alternatively, Dcp2p could be required for the hydrolysis and subsequent inactivation of a small inhibitor of Dcp1p, or for the production of a small molecule that activates Dcp1p. This model differs from the first only in that the pyrophosphate‐containing molecule need not be covalently linked to Dcp1p. Future work should resolve these issues.
Could Dcp2p be an important site for control of mRNA surveillance?
Several observations suggest that Dcp2p may be a critical site for the control of mRNA surveillance. First, Dcp2p is required for the rapid degradation of nonsense codon‐containing mRNAs. Secondly, Dcp2p is required for the production of active Dcp1p. Thirdly, Dcp2p was identified several years ago as a protein that showed two‐hybrid interactions with Upf1p (He and Jacobson, 1995; referred to as NMD1). The Upf1p is required for mRNA surveillance (Leeds et al., 1991) and part of its role is to function as an ATPase, whose ability to hydrolyze ATP is required for activation of decapping (Weng et al., 1996). Since we have shown an interaction between Dcp2p and the DCP1 decapping enzyme (Figure 9), the interaction between Dcp2p and Upf1p, which has been shown to associate with translation termination release factors (Czaplinski et al., 1998), provides a continuous set of interactions from the site of translation termination to the decapping enzyme. This suggests a possible mechanism wherein the hydrolysis of ATP by Upf1p would trigger Dcp2p to hydrolyze its as yet unknown substrate, which would then lead to activation of Dcp1p and decapping of the mRNA. Future experiments should be able to test the relevance of this specific hypothesis.
Is Dcp2p a multifunctional protein?
Recent results raise the possibility that Dcp2p/Psu1p may also have a function in transcription. The key observation is that small regions of Dcp2/Psu1p can activate transcription when tethered to DNA by fusion to a DNA binding domain (Gaudon et al., 1999). Moreover, depletion of Dcp2p/Psu1p led to a loss of ligand‐stimulated transcription. However, the in vivo significance of this finding remains unclear, and is probably unrelated to the role of Dcp2p in decapping, since the MutT motif of Dcp2p is required for decay and is not included within the regions that activate transcription when tethered to DNA.
Materials and methods
Plasmids and strains
Strains used in this study are listed in Table II. Strain yRP1340 was constructed in three steps. First, a portion of genomic DNA 3′ of the DCP1 gene was PCR amplified from plasmid pRP872 and cloned into the pBluescriptII KS+ vector using the BamHI and SacII sites. Next, the TRP1 gene was cloned into the BamHI site of this plasmid to generate plasmid pRP926. The dcp1‐2 allele was then PCR amplified from plasmid pRP872 and cloned into pRP926 using the ApaI and ClaI sites. This plasmid was then digested with ApaI and SacI to release the region of DNA containing the dcp1‐2 allele, TRP1 and the genomic DNA from the region 3′ of DCP1. This fragment was then transformed into yRP1070 and TRP+ transformants were screened at 24°C for loss of the URA3 marker. Genomic integration of the dcp1‐2 allele was confirmed by Southern analysis.
The dcp2Δ strain was constructed by PCR amplification of the regions flanking the DCP2 gene from plasmid pRP925. The PCR products were subcloned on either side of the TRP1 gene in a pBluescript vector. The resulting plasmid was digested to release the dcp2Δ cassette and transformed into a wild‐type diploid (yRP840 crossed to yRP841). Heterozygous transformants were identified using Southern analysis and dissected to generate strain yRP1346.
The FLAG–DCP2 fusions were constructed using in vivo recombination between a FLAG‐tagged DCP2 PCR product and BamHI–SalI digested pRP935 in strain yRP1346. Plasmids were rescued from the resulting transformants and sequenced to confirm that no mutations were introduced during PCR. The full‐length FLAG–DCP2 fusion PCR product was constructed through amplification of the DCP2 gene from pRP925 using universal PCR primers from Research Genetics (YNL118C). This product was then reamplified using oRP798 and oRP799. The dcp2‐ΔC allele was constructed through amplification of the N‐terminal 900 nucleotides of DCP2 from pRP925 using the Research Genetics YNL118C forward primer and oRP752. This PCR product was then reamplified using oRP798 and oRP799. The dcp2‐ΔN allele was constructed through amplification of the C‐terminal 2010 nucleotides of DCP2 from pRP925 using the Research Genetics YNL118C reverse primer and oRP751 followed by reamplification with oRP798 and oRP799.
The FLAG–Dcp1p fusion was constructed in two steps. The FLAG–DCP1 fusion was PCR amplified from plasmid pRP801 using primers oRP311 and oRP312. This PCR product was then cotransformed into strain yRP1070 with BamHI–SalI digested pRP935. Plasmids resulting from in vivo recombination that complemented the dcp1Δ were rescued to E.coli using a yeast DNA isolation system (Stratagene), sequenced and subsequently transformed into yRP1346.
DCP2 missense mutations were constructed using oligonucleotide‐directed mutagenesis as previously described (Tharun and Parker, 1999). The template was the DCP2 gene, containing its native promoter and terminator elements in pRS416.
Genetic screening procedures
To screen for high‐copy suppressors of the dcp1‐2 ski8Δ mutant, strain yRP1345 was transformed with a yeast 2μ genomic DNA library (ATCC 37323). Transformants were selected at 24°C. Positive transformants were then replica plated and screened for growth at 24, 34 and 37°C. Plasmids were isolated from clones that supported growth at either 34 or 37°C, retransformed into yRP1345 and retested for growth at the restrictive temperature. The inserts from plasmids that conferred growth at high temperature were then sequenced.
RNA samples were prepared and isolated as previously described (Caponigro et al., 1993). Half‐lives were determined by quantitation of blots using a Molecular Dynamics PhosphorImager. Loading corrections for quantitation were determined by hybridization to the 7S RNA, a stable RNA polymerase III transcript.
Determination of dcp1‐2 suppression
Strains yRP840, yRP1341, yRP1355, yRP1356 and yRP1357 were grown at 24°C to mid‐log phase in selective media containing 2% galactose. Cultures were then shifted to 34°C and grown for an additional 1 h. Cultures were then pelleted and resuspended in 1/10 volume YEP medium. Dextrose was then added to 4% to repress transcription from the GAL1 upstream activating sequence. Aliquots of cells were harvested at various time points following repression of transcription. RNA was prepared as previously described (Caponigro et al., 1993).
Immunoprecipitation using anti‐cap antibodies
Anti‐cap immunoprecipitations were performed as described by F.He and A.Jacobson (manuscript in preparation). Briefly, 10 mg protein A–Sepharose CL4B was swelled in IPPH buffer (500 mM NaCl, 10 mM Tris pH 7.5, 0.1% NP‐40). The swelled protein A was washed three times and resuspended in 1 ml IPPH. Anti‐cap antibody (10 μl) (Munns et al., 1982) was then added to the beads and allowed to bind overnight at 4°C in the presence of 0.1 U/μl RNasin. The antibody‐coupled beads were washed five times with IPPL buffer (150 mM NaCl, 10 mM Tris pH 7.5, 1 mM EDTA, 0.05% NP‐40). A 10 μl aliquot of total RNA was added to 100 μl antibody‐coupled beads and incubated overnight at 4°C with 40 U RNasin. The supernatant was recovered and bound to fresh antibody‐coupled beads overnight. The two pellets were combined and resuspended in 180 μl IPPL. RNA was eluted during a 1 h incubation at room temperature using SDS at a final concentration of 1%. The solution was centrifuged to pellet the protein A–Sepharose and the resulting supernatant extracted with phenol/chloroform. Fractions were analyzed on a 6% polyacrylamide Northern gel.
Purification of FLAG fusion proteins
A 500 ml culture containing the appropriate FLAG fusion was grown to late log phase in selective medium containing 2% dextrose. Cells were washed once with 30 ml ddH2O, once with 30 ml TMN‐150 (10 mM Tris–HCl pH 7.7, 150 mM NaCl, 5 mM MgCl2, 0.5 mM PMSF, 0.5 mM DTT), and resuspended in 5 ml TMN‐150. A 2.5 ml aliquot of acid‐washed beads was added to the cell suspension and cells were lysed by 10 cycles of vortexing for 30 s and 45 s in ice water. The lysate was centrifuged at 18 000 g for 15 min at 4°C and the supernatant removed. This supernatant was spun at 36 000 g for an additional 45 min at 4°C. The clear lysate was then removed and glycerol added to 20%. For purification of the FLAG construct, 500 μl of the lysate was added to 100 μl anti‐FLAG M2 affinity gel (Kodak) pre‐equilibrated in TMN‐150. Binding was performed in batch overnight at 4°C. The affinity gel was washed five times with 1 ml TMN‐150 and the bound FLAG fusion protein eluted at 4°C in TMN‐150 containing 200 μg/ml FLAG peptide (Sigma).
The FLAG fusion protein preparations were analyzed by standard SDS–PAGE on a 10% gel (Laemmli, 1970). Protein size markers were purchased from Gibco‐BRL. The gels were silver‐stained using Silver Stain Plus reagents (Bio‐Rad).
Determination of Dcp1p expression
Strains yRP840, yRP1070 and yRP1346 were grown to an OD600 of 1.0. Whole cell extracts were prepared as previously described (Beelman et al., 1996). One hundred micrograms of total protein was run on a standard 10% SDS–polyacrylamide gel. Expression of Dcp1p was detected by Western analysis using polyclonal antibodies to the full‐length Dcp1 protein.
We gratefully thank all members of the Parker laboratory for helpful comments and discussions on the manuscript. This work was supported by the Howard Hughes Medical Institute and a grant to R.P. from the National Institutes of Health (GM4544).
- Copyright © 1999 European Molecular Biology Organization