Characterization of a general stabilizer element that blocks deadenylation-dependent mRNA decay.

mRNA degradation is a regulated process that can play an important role in determining the level of expression of specific genes. The rate at which a specific mRNA is degraded depends largely on specific cis-acting sequences located throughout the transcript. cis-Acting destabilizer sequences that promote increased rates of decay have been identified in several short-lived mRNAs. However, little is known about elements that promote stability, known as stabilizer elements (STEs), and how they function. The work presented here describes the characterization of a STE in the PGK1 transcript. The PGK1 stabilizer element (P-STE) has been delineated to a 64-nucleotide sequence from the coding region that can stabilize a chimeric transcript containing the instability elements from the 3'-untranslated region of the MFA2 transcript. The P-STE is located within the PGK1 coding region and functions when located in the translated portion of the transcript and at a minimum distance from the 3'-untranslated region. These results further support the link between translation and mRNA degradation. A conserved sequence in the TEF1/2 transcript has been identified that also functions as a STE, suggesting that this sequence element maybe a general stability determinant found in other yeast mRNAs.

Differences in the decay rates of individual mRNAs can have profound effects on the overall levels of expression of specific genes. A key to understanding how a specific mRNA is degraded requires characterizing the cis-acting elements and trans-acting factors that regulate its turnover. Studies in the yeast Saccharomyces cerevisiae have revealed several mRNA degradation pathways that can provide insights into how this process can be regulated. mRNA degradation might initiate by endonucleolytic cleavage (1), loss of the cap and subsequent 5Ј 3 3Ј exonucleolytic digestion (2), or deadenylation, which can be followed either by 3Ј 3 5Ј exonucleolytic degradation or decapping and 5Ј 3 3Ј exonucleolytic degradation (3)(4)(5)(6). However, many wild-type yeast mRNAs that have been investigated decay through the pathway that initiates with the shortening of the poly(A) tail to an oligo(A) form. Removal of the poly(A) tail triggers rapid decapping of the transcript by the Dcp1p and subsequent degradation of the body of the transcript by a 5Ј 3 3Ј exoribonuclease, Xrn1p (reviewed in Ref. 7).
cis-Acting sequences that modulate the decay rate of an mRNA have revealed several types of destabilizer elements (1, 8 -16). These instability elements have been identified by their ability to promote rapid decay when transferred to an appropriate location within inherently stable mRNAs or by mutational analysis. For example, the MFA2 mRNA is very unstable, with a half-life of 3 min (15). Previous studies have determined that the sequences responsible for the destabilization of MFA2 were localized to the 3Ј-UTR 1 and that they function by promoting increased rates of deadenylation, which is followed by decapping and 5Ј 3 3Ј decay (2). In the case of the MAT␣1 transcript, a 65-nt instability element (MIE) is located in the protein coding region of the transcript and requires translation in order to be functional (17). Interestingly, a given transcript can have multiple destabilizer sequences, suggesting that the process of targeting an mRNA for degradation can be redundant (reviewed in Ref. 18).
In addition to instability elements, a few examples of stabilizer elements that block rapid mRNA decay have been identified (19 -22). In mammals, a sequence in the 3Ј-UTR of the ␣-globin transcript slows down poly(A) shortening and rapid decay of the mRNA (23). In yeast, two types of sequences that promote specific stabilization of nonsense-containing transcripts have been described. Transcripts containing premature nonsense codons are rapidly degraded via the nonsense-mediated mRNA decay (NMD) pathway. Using the PGK1 and HIS4 transcripts, it was shown that rapid decay takes place only if the nonsense mutation occurs approximately within the first two thirds of the transcript and that 3Ј-proximal nonsense mutations are resistant to the NMD pathway (20,24). These results indicate that specific sequences in the last third of the PGK1 transcript confer resistance to NMD when translated. A second sequence that promotes stabilization of nonsense-containing transcripts was identified in the leader region of the GCN4 mRNA (21,25). This sequence inactivates the NMD pathway when positioned 3Ј of a nonsense codon.
A stabilizer sequence that inactivates the deadenylation-dependent decay pathway has been found in the stable PGK1 mRNA (19). The experimental approach involved constructing chimeric genes harboring the PGK1 gene fused with portions of an unstable mRNA. Studies with chimeric genes containing the PGK1 coding region and the MFA2 3Ј-UTR demonstrated that the MFA2 3Ј-UTR could not destabilize an intact PGK1 mRNA (19). Interestingly, the PGK1-MFA2 chimeric transcript was stabilized as an oligoadenylated form, suggesting that the stabilizer element protects the transcript from being rapidly decapped and/or from 3Ј 3 5Ј exonucleolytic attack (19). Similar results were obtained using the STE3 3Ј-UTR, also known to contain an instability determinant (11). Thus, it seems very likely that stabilizer elements occur in other stable transcripts.
In this work, we have mapped and characterized the stability element in the PGK1 transcript (P-STE). The results presented here indicate that a 65-nt sequence from the PGK1 coding region promotes stabilization of a PGK1-MFA2 chimeric transcript. Furthermore, a larger region that contains this element also stabilizes a PGK1-STE3 chimera. The P-STE functions only when located in the translated portion of the transcript and not immediately proximal to the MFA2 3Ј-UTR. In addition, we have identified a sequence similar with the P-STE in the TEF1/2 transcript that promotes stabilization of a PGK1-MFA2 chimeric transcript. Taken together, these results indicate that we have identified a general stability determinant that can modulate the stability of mRNAs.
mRNA Decay Measurements, RNA Preparation, and Analysis-The various hybrid PGK1-MFA2 and PGK1-STE3 alleles were transformed into strains harboring the temperature-sensitive allele of the RNA polymerase II (rpb1-1) and the mRNA decay rates were determined by Northern analyses as described previously (20,24). All blots in Figs. 1, 2, 4, and 5 were probed with a 32 P-labeled fragment containing the entire MFA2 gene, and all blots in Fig. 3 were probed with a 32 P-labeled fragment containing the 3Ј-UTR of STE3. The results of these experiments were quantitated using a Bio-Rad model G-250 molecular imager or a Bio-Rad model GS-670 imaging densitometer. The mRNA abundances were normalized using the U3 RNA (31). The mRNA half-life measurements presented in each case are the average of at least three different experiments.
Plasmid Constructions-All plasmid constructions described below harbor a 291-base pair fragment containing the PGK1 signals required for transcription initiation. All numbering includes this sequence plus the coding region bases as detailed in Peltz et al. (20). The yeast centromeric plasmid pRIP was used as vector for all the alleles constructed and was cleaved with BamHI and HindIII. The MFA2 3Ј-UTR present in many of the constructs shown below was synthesized by polymerase chain reaction reaction using the appropriate oligonucleotides as primers and plasmid pRP455 (32) as template. The PGK1 fragments of the constructs shown in this study were derived from plasmids pRIPPGK(ϪAU)H 2 (3)UAA, pRIPPGK(ϪAU)AspUAA, pRIPPGK(ϪAU)H 2 (2)UAA, and pRIPPGK(ϪAU)H 2 (1)UAA (24) harboring the PGK1 gene with UAA stop codons at 5.6%, 39%, 55%, and 67% of the PGK1 coding region, respectively, or from pRIPPGK(ϪAU) (20), which harbors the wild-type PGK1 gene. In Fig. 1, constructs 1A-1D were prepared by digestion of the PGK1 UAA-containing alleles with HpaI, which cut a few nucleotides downstream of the UAA stop codon, and inserting the MFA2 3Ј-UTR at this position in the various alleles. The constructs shown in Fig. 1B were prepared by digestion of the PGK1 UAA-containing alleles with BglII, which cuts at 92% of the PGK1 coding region and insertion of the MFA2 3Ј-UTR at this position in the various alleles.
In Fig. 2, construct 2A is identical to 1C. To prepare construct 2B, plasmid pRIPPGK(ϪAU)H 2 (2)UAA was digested with HincII (at position 1138 of the PGK1 coding region) and HindIII (downstream of the PGK1 transcription terminator region) and inserting at that position the MFA2 3Ј-UTR. Construct 2C was prepared from construct 2B by substituting the 159 nt (nt 979 -1138) of the PGK1 coding region for 159 nt (nt 393-552) from the GCN4 leader region. Construct 2D was derived from plasmid pRIPPGK(ϪAU)H 2 (2)UAA by inserting the MFA2 3Ј-UTR into its unique BglII site (at 92% of the protein coding region) and deleting the 159-nt fragment from nt positions 979 -1138. To prepare construct 2E, plasmid pRIPPGK(ϪAU)AspUAA was partially digested with HincII (to cut only at position 1138 of the PGK1 coding region) and HindIII (downstream of the PGK1 transcription terminator region) and the MFA2 3Ј-UTR was inserted at that position. Construct 2F was prepared from construct 2E by substituting the 159 nt (nt 979 -1138) of the PGK1 coding region for 159 nt (nt 393-552) from the GCN4 leader region.
In Fig. 3, constructs 3A and 3B were prepared by digestion of plasmids pRIPPGK(ϪAU)H 2 (3)UAA and pRIPPGK(ϪAU)H 2 (1)UAA, respectively, with BglII (cuts at 92% of the PGK1 coding region) and insertion of the STE3 3Ј-UTR at this position. The fragment harboring the STE3 3Ј-UTR was synthesized by polymerase chain reaction using the appropriate oligonucleotides as primers and total yeast DNA as template.
All the constructs depicted in Figs. 4 and 5 carry the MFA2 gene under control of the PGK1 promoter. Constructs 4A-4C contain the PGK1 sequences from nt 789 to 1138, inserted at the unique BamHI site located 4 nt upstream of the MFA2 stop codon. In construct 4A, the PGK1 fragment is inserted in frame with the MFA2 sequence. In construct 4B, the PGK1 fragment contains a UAA stop codon at the position corresponding to nt 979 of the PGK1 sequence. In construct 4C, the stop codon is located at nt position 789 of the PGK1 sequence. Construct 4D was prepared by replacing the PGK1 sequences in construct 4A with PGK1 sequences from nt 302 to 652. Construct 5A contains the TEF1 sequences from nt 1080 to 1409 (numbering starts at the AUG of the coding region, and this segment includes 32 nt 3Ј of the TEF1 stop codon) inserted at the unique BamHI site located 4 nt upstream of the MFA2 stop codon in frame with the MFA2 coding sequence. Construct 5B was prepared by replacing the 190-nt P-STE (nt 789 -879) in construct 4B with the TEF1 sequences from 1117 to 1176, harboring the region of homology to the P-STE in frame with MFA2. This sequence is 100% identical between TEF1 and TEF2, the second gene encoding translation elongation factor 1A (eEF1A).

Deletion Analysis Demonstrates the Presence of a Stability Element within the First Two Thirds of the PGK1 Transcript-
The MFA2 gene encodes an inherently unstable transcript with a half-life of 3 min (15). The sequences responsible for the destabilization of the MFA2 mRNA have been localized to the 3Ј-UTR (33). Surprisingly, a chimeric PGK1-MFA2 mRNA, in which the PGK1 3Ј-UTR was replaced with that from MFA2 transcript, encodes a stable mRNA that lacks the poly(A) tail (19). This result is consistent with previous observations demonstrating that the poly(A)-deficient form of the PGK1 transcript is stable (34). These and other observations suggested that the coding region of the PGK1 gene contains a stabilizer element (P-STE) that blocks decapping of the mRNA when translated. Our goal was to identify and characterize the P-STE in the PGK1 transcript. The boundaries of the P-STE were initially defined by deletion analysis of the stable PGK1-MFA2 transcript. The deletions were constructed by inserting a termination codon at four different locations within the PGK1 protein coding region and deleting the region 3Ј of the termination codon. The MFA2 3Ј-UTR that includes the region required for rapid mRNA decay, as well as the cleavage and polyadenylation sites, was inserted immediately 3Ј of the termination codon (Fig. 1A). These alleles were transferred to yeast centromere plasmids and transformed into a upf1⌬ strain harboring the rpb1-1 temperature-sensitive allele of the RNA polymerase II. mRNA decay rates were determined by RNA blotting analyses (see "Materials and Methods"). Half-life measurements were performed in a upf1⌬ strain to eliminate any effect of the NMD pathway that could result from the introduction of premature termination codons. The results of these experiments indicated that transcripts harboring the amino-terminal 55% of the PGK1 coding sequences (up to nt position 979) were rapidly degraded, with half-lives of 3 to 6 min ( Fig. 1A, constructs 1A-1C). A hybrid transcript harboring 67.7% of the PGK1 coding sequences encoded a moderately stable mRNA with a half-life of 18 min, indicating the presence of a STE (Fig. 1A, construct 1D). Replacing the MFA2 3Ј-UTR with the PGK1 3Ј-UTR resulted in stabilization of all these transcripts, indicating that the instability observed is due to the MFA2 3Ј-UTR (20; data not shown). Taken together, these results indicate that there is a sequence within the first two thirds (67.7%) of the PGK1 transcript that blocks rapid degradation of the PGK1-MFA2 chimeric transcript.
Insertion of Early Termination Codons into Full-length PGK1 Indicates That the P-STE Requires Translation up to 55% of the PGK1 Transcript-The function of several cis-acting instability elements requires ribosome translocation up to or through the element (Ref. 35; reviewed in Refs. 7, 36, and 37). Based on these observations, we explored whether translation is required for the P-STE to be functional. As a first test, translation was disrupted by inserting in-frame nonsense codons at four different locations within the PGK1 protein coding region and the decay rate of the PGK1-MFA2 mRNA was determined in a upf1⌬ strain (Fig. 1B, constructs 1E-1H). A PGK1-MFA2 chimeric gene was constructed by inserting the MFA2 3Ј-UTR into a unique BglII site located at 92.6% of the  20. The number at the top of the PGK1 schematic indicates the percentage of the PGK1 coding region that is translated. The MFA2 3Ј-UTR is depicted as a black square. mRNA half-lives were determined in the RY262 Ϫ strain (as described under "Materials and Methods") and are shown in the column at the right of the figure. The Northern blots for these alleles are shown below the schematic representation. The abundances of all the mRNAs were normalized to the abundance of the U3 small nucleolar RNA (data not shown). B, schematic representation of the PGK1-MFA2 alleles containing the complete PGK1 coding region. The translated portion of the PGK1 coding region is depicted as a dotted rectangle and the corresponding percentage indicated above. The length of the PGK1 coding region that is not translated is indicated by a white rectangle. The Northern blots for these alleles are shown below the schematics. PGK1 coding region. A premature stop codon was inserted at 5.6%, 39%, 55%, and 67.7% of the PGK1 coding region (Fig. 1B, constructs 1E-1H, respectively). Therefore, although the size of the transcripts is equivalent, the amount of the PGK1protein coding region that is translated is different in each allele. mRNA decay rates were determined in a upf1⌬ strain harboring the rpb1-1 ts allele of RNA polymerase II. The results demonstrated that, relative to translation of the entire PGK1 mRNA, translation termination after 5.6% or 39% of the PGK1 coding region, respectively, promoted destabilization of the PGK1-MFA2 transcript (half-life of 6 and 8 min, respectively; Fig. 1B, constructs 1E and 1F). However, location of premature nonsense mutations following 55% or greater of the PGK1 coding region remained stable with a half-life of greater than 25 min (Fig. 1B, constructs 1G and 1H). Therefore, translation of up to 55% of the coding region is sufficient to promote stabilization of the PGK1-MFA2 mRNA when the C-terminal part of the transcript is also present.
The Distance between the P-STE and the 3Ј-UTR Affects the Decay Rate of the mRNA-The results obtained above indicated that the MFA2 3Ј-UTR does not promote rapid decay if 55% (nt 979) or greater of the PGK1 protein coding region is translated (Fig. 1B). Interestingly, these results seem in conflict with the results shown in Fig. 1A, in which insertion of the MFA2 3Ј-UTR at 55% of the protein coding region promoted accelerated mRNA decay (Fig. 1, A (construct 1C) compared with B (construct 1G)). This raises the possibility that the P-STE is upstream of nt 979 (55%) and that either: (i) it requires a specific sequence 3Ј of a translation termination codon, or (ii) there needs to be a certain amount of sequence between the P-STE and the instability element in the 3Ј-UTR. To test which of these hypotheses is valid, the following constructs were prepared. In these constructs, a nonsense mutation at position 979 (at 55% of the PGK1 protein coding region) was inserted, and the amount of sequence immediately downstream of the termination codon was varied. In construct 2A (identical to construct 1C in Fig. 1A), the nonsense codon is immediately followed by the MFA2 3Ј-UTR. In construct 2B, the MFA2 3Ј-UTR was inserted at position 1138 of the PGK1 coding sequence and therefore it contains the 159 nt naturally located 3Ј of nt 979. Construct 2C is similar to construct 2B, except that the 159 nt 3Ј of the nonsense mutation correspond to a fragment from the GCN4 leader sequence. This construct serves as a control to determine whether a specific sequence 3Ј of the termination codon is required for the P-STE to function. In construct 2D, the MFA2 3Ј-UTR was inserted at position 1449 of the PGK1 coding region and the sequences between positions 979 and 1138 were deleted; therefore, it contains 311 nt of the 3Ј terminal PGK1 sequences between the nonsense codon and the MFA2 3Ј-UTR.
The decay rates of these hybrid transcripts were determined in a upf1⌬ strain as described above, and the results are shown in Fig. 2A. As shown in Fig. 1, the transcript in which the MFA2 3Ј-UTR is immediately downstream of the nonsense mutation was an unstable mRNA with a half-life of 6 min ( Fig.  2A, construct 2A). However, insertion of other sequences from the PGK1 or the GCN4 transcripts between the stop codon and the MFA 3Ј-UTR led to a 3-fold stabilization of the hybrid transcript ( Fig. 2A, constructs 2B, 2C, and 2D). This result suggests that there is no specific sequence requirement 3Ј of the termination codon. Taken together, these results indicate that the P-STE is located upstream of nucleotide 979 and that its activity requires positioning at least a minimal distance of 159 nt from the 3Ј-UTR.
The P-STE Needs to Be Translated in Order to Be Active-The results described above suggested that the P-STE was located between positions 789 and 979 and that it required translation in order to be active. However, since the distance between the stop codon and the 3Ј-UTR is critical for the stabilization effect, it was possible that translation of the P-STE per se is not required for its activity. One possibility is that transcripts that contain the P-STE but harbor nonsense codons at positions less than 55% (Fig. 1B, constructs 1E and 1F) are not stabilized because the stop codon is too far away from the 3Ј-UTR. Two constructs were prepared to test this possibility. In construct 2E, a nonsense mutation was inserted at position 789 to avoid translation of the coding region between nt 789 and 979 (Fig. 2B, construct 2E). The MFA2 3Ј-UTR was inserted at a HincII site at nt 1138 of the PGK1 coding region. Therefore, all the sequences between the stop codon and the MFA2 3Ј-UTR were derived from PGK1. In construct 2E, the distance between the stop codon and the 3Ј-UTR is sufficient to allow the P-STE to function (compare with the distance between the stop codon and the 3Ј-UTR in construct 2D in Fig. 2A and construct 1G in Fig. 1B). As a control, a construct was used which is identical to construct 2E except that the 159 nt located immediately upstream of the MFA2 3Ј-UTR were replaced with a 159-nt region from the GCN4 leader sequence (Fig. 2B, construct 2F) known to have no effect on mRNA stability. The decay rates of the mRNAs encoded by these alleles were determined in a upf1⌬ strain as described above. The results indicated that both transcripts were unstable with a half-life of ϳ6 min. As shown above, changing the position of the stop codon to nt position 979 in otherwise identical constructions ( Fig. 2A, constructs 2B and 2C) resulted in stable transcripts. Taken together, these results indicated that the P-STE is located between positions 789 and 979 of the PGK1 coding region and that translation over this sequence is required for its activity.
A Region of PGK1 Containing the P-STE Functions to Promote Stabilization of a PGK1-STE3 Chimeric Transcript-The results presented above indicated that the P-STE promotes stabilization of the MFA2 mRNA, a transcript known to decay through the deadenylation-dependent decay pathway. Previous studies have shown that the 3Ј-UTR from the STE3 mRNA, encoding the a-mating factor receptor, promotes rapid deadenylation-dependent mRNA decay (11). Furthermore, the STE3 3Ј-UTR does not promote rapid decay when inserted after the complete PGK1 protein coding region (11). Thus, we hypothesized that the P-STE would also stabilize a PGK1-STE3 hybrid transcript. To test this possibility, the MFA2 3Ј-UTR from two of the PGK1-MFA2 chimeric transcripts was replaced with the 3Ј-UTR of the STE3 transcript and the half-lives of the corresponding mRNAs were determined (Fig. 3). Both constructs contain 92.6% (starting at the N terminus) of the PGK1 coding region followed by the STE3 3Ј-UTR. Construct 3A contains a stop codon at position 5.6% (nt 363) of the PGK1 protein coding region so that the P-STE is not functional. Construct 3B harbors a stop codon at position 67.7% (nt 1138) of the coding region; therefore, the P-STE is translated and functional. The decay rates of these transcripts were determined as described previously. The results indicated that insertion of a nonsense mutation, which terminates translation at 5.6% of the PGK1 coding region promotes destabilization of the transcript (Fig. 3, construct 3A). By contrast, a nonsense mutation that allows translation of up to 67.7% of the PGK1 coding region results in a stable mRNA (Fig. 3, construct 3B). These results parallel the results obtained with the PGK1-MFA2 transcripts (compare with Fig. 1B, constructs 1E and 1H) and suggest that the region containing the P-STE can function to override the instability caused by different instability determinants.
The P-STE Functions when Inserted within the MFA2 Transcript-The results described above indicate that we have identified a sequence in the PGK1 coding region that, when translated, promotes stabilization of a PGK1-MFA2 chimeric transcript. We next tested whether the P-STE could function in other heterologous transcripts. To test this possibility, we analyzed whether the P-STE was functional when positioned within the unstable MFA2 transcript. We inserted different DNA fragments spanning the P-STE into the MFA2 transcript, and the mRNA half-life of the resulting mRNAs was determined. The relevant features of the constructs used are shown in Fig. 4. A unique BamHI site located at the 3Ј end of the MFA2 coding region was used to insert the sequences from the PGK1 mRNA (constructs 4A-4D). Construct 4A contains the PGK1 coding region from nt 789 to 1138, which harbors the putative P-STE plus an additional 159 nucleotides inserted in frame in the MFA2 transcript. Construct 4B is essentially identical to construct 4A, except that it contains a stop codon at position 979, immediately downstream of the sequences corresponding to the P-STE. In construct 4C, the stop codon was inserted at position 789, precluding translation of the P-STE. As a control, a 350-nt fragment from the N-terminal region of the PGK1 transcript (nt 302-652) was inserted in the unique BamHI site of the MFA2 transcript (construct 4D). The decay rates of these transcripts were determined in a upf1⌬ strain as described above. The MFA2-PGK1 mRNAs containing the P-STE in frame with the MFA2 sequences (Fig. 4, constructs 4A and 4B) were stable with a half-life of 15 min. In contrast, transcripts in which the P-STE was either not translated (construct 4C) or containing the sequence from the N-terminal region of the PGK1 transcript (construct 4D), were rapidly degraded. All together, these results demonstrate that the 190-nt segment from PGK1 (nt 789 -979) functions as a general mRNA stabilizer element when positioned within the trans-lated portion of the transcript, and at a certain distance of the sequence corresponding to the MFA2 3Ј-UTR.
The TEF1/2 Transcript Contains a Functional Stabilizer Element-We next determined whether stabilizer elements similar to the P-STE could be identified in other yeast transcripts. We carried out a computer search in which the 190-nt P-STE was compared against the complete S. cerevisiae nucleotide data base. The TEF1/2 transcripts, which encode identical forms of eEF1A, revealed the highest conservation, showing 70% identity in a 63-nt region (Fig. 5A). Interestingly, the conservation in most sequences identified was restricted to a ϳ60 -65-nt region in the central portion of the 190-nt P-STE (Fig. 5, A and B). Additional searches utilizing this conserved region revealed several shorter conserved sequences. Use of the 5Ј and 3Ј regions of the P-STE identified only one sequence from the CHC1 transcript with 78% identity in 40-nt overlap to the 3Ј region of the P-STE.
Since the TEF1/2 transcript shows the greatest homology with the P-STE, we tested whether it contains a stabilizer element. For this purpose, we inserted different TEF1 fragments spanning the P-STE homology region in frame into the MFA2 transcript, and the mRNA half-life of the resulting mRNAs was determined. The relevant features of the constructs used are shown in Fig. 5C. The unique BamHI site at the 3Ј end of the MFA2 coding region was used to insert the sequences from the TEF1 mRNA (constructs 5A and 5B). Construct 5A contains 330 nt from the TEF1 coding region inserted in frame in the MFA2 transcript. This fragment (nt 1080 -1409, ending 32 nt downstream of the stop codon) harbors the region of homology to the P-STE plus additional 5Ј-and 3Ј-flanking sequences. The 3Ј sequences provide the minimum distance defined from the PGK1 mRNA stabilizer element to the 3Ј-UTR. As shown above, this distance allowed the P-STE to be active. Construct 5B is identical to construct 4B and 4C (Fig. 4), except that the 190 nt corresponding to the P-STE were replaced with 65 nt from the TEF1 transcript harboring the region of homology to the P-STE, and which is 100% identical between TEF1 and TEF2. Therefore, in this construct the sequence 3Ј of the 65-nt region from TEF1 comes from PGK1. Constructs 4C and 4D serve as negative controls. The decay rates of these transcripts were determined in a upf1⌬ strain as described above. Both MFA2-TEF1 mRNAs (constructs 5A and 5B) were stable with a half-life of 15 min, identical to the half-life of the MFA2-PGK1 transcripts containing the 190-nt P-STE. These results indicate that the P-STE is a general stabilizer element and that a 65-nt region is sufficient for the stabilization effect. DISCUSSION Differences in decay rates of mRNAs were initially thought to be attributed to the presence or absence of specific destabilizer elements within the transcript. However, the identification of mRNA stabilizer elements in the yeast PGK1 and human ␣-globin mRNAs suggests that mRNA stabilization can also be an active process that requires the presence of specific sequences. In this scenario, the rate of decay of a specific mRNA would be mainly determined by the presence of specific sequences that function by increasing (instability elements) or decreasing (stability elements) the rates of poly(A) shortening and/or decapping and/or any of the steps in the decay pathway.
We have characterized a sequence in the coding region of the PGK1 transcript that functions as an mRNA stabilizer element (P-STE). The results demonstrated that the P-STE: (i) is contained within a 65-nt sequence; (ii) promotes stabilization of a PGK1-MFA2 or a PGK1-STE3 chimeric transcript; (iii) must be translated; (iv) must be located at least a minimal distance from the 3Ј-UTR to function; and (v) is conserved in other transcripts, i.e. the TEF1/2 mRNA.
The PGK1 Transcript Contains at Least Two Different Stabilizer Elements-Studies of yeast mRNAs have revealed a certain degree of diversity in the structural determinants that dictate rapid mRNA decay of a specific transcript (Refs. 1, 3, 9 -12, 14, and 33; reviewed in Ref. 18). For example, at least two different regions within the STE3 transcript can stimulate rapid mRNA degradation (11). Presumably, the presence of multiple decay elements within a single mRNA allows for flexibility in how decay rates can be modulated. A similar picture is beginning to emerge from the studies of mRNA stability elements. Previous studies have identified a stabilizer element in the PGK1 protein coding region that inactivates the NMD pathway (18). The results presented here identify a second element in the PGK1 coding region (P-STE) that stabilizes mRNAs that degrade via 3Ј-UTR elements and the deadenylation-dependent decay pathway. Both of these elements need to be translated in order to be functional. However, they are not functionally redundant; the P-STE does not prevent the NMD pathway from functioning. For example, the presence of a premature termination codon at position 55% of the PGK1 protein coding region still promotes rapid mRNA decay through the NMD pathway (18,24). However, translation of this region fully activates the ability of the P-STE to stabilize mRNAs containing the MFA2 or STE3 3Ј-UTR. This result suggests that these stability elements block the activity of a different factor(s) in their respective decay pathways (see below).
Three other stabilizer elements have been characterized in addition to those identified within the PGK1 transcript. One stabilizer element has been identified in the yeast GCN4 mRNA (GCN4-STE; Ref. 21) and is functionally equivalent to a STE found in the YAP1 mRNA (Y-STE; Ref. 25). The third one was isolated from the human ␣-globin mRNA (38). The GCN4 -STE and Y-STE are located in the 5Ј-untranslated region of the GCN4 and YAP1 transcripts, respectively, and function to prevent degradation of the transcript through the NMD pathway. These two STEs can also function when inserted into the untranslated region of heterologous transcripts 3Ј of a nonsense codon. However, we have shown previously that the GCN4-STE does not stabilize wild-type transcripts containing the MFA2 3Ј-UTR (21). The ␣-globin mRNA stability element is a pyrimidine-rich sequence located within the 3Ј-UTR of the transcript. Several highly stable mRNAs contain a similar sequence motif in the 3Ј-UTRs, suggesting that it is a general determinant for stabilization of eukaryotic mRNAs (39 -41). The GCN4-STE, the Y-STE, and the ␣-globin stabilizer differ from the P-STE in that they are located in untranslated portions of the transcripts. Therefore, the identification of the P-STE is a novel type of stabilizer element that uses a different mechanism to stabilize wild-type transcripts. Sequence analysis of the P-STE demonstrates that it is U-rich but has no sequence conservation with the ␣-globin stabilizer element. Comparison of the P-STE sequence to the yeast data base at the nucleotide level identified regions of sequence identity in several genes (Fig. 5A). The most highly conserved of these regions is located in the TEF1/2 transcript, with 70% identity over a 65-nt region. Interestingly, the results presented here indicated that a fragment spanning the highly conserved 65-nt region from TEF1/2 region can function to prevent rapid degradation of transcripts containing the MFA2 3Ј-UTR. Therefore, the conservation at the nucleotide level corresponds with functional conservation, indicating that the P-STE is a general stabilizer element. Further investigations will determine whether other regions identified during the data base comparison, and which have less homology to the P-STE, can also function as stabilizer elements. These analyses will aid in the determination of what are the important nucleotides for the function of the P-STE.
The P-STE Functions through the 3Ј-UTR and Requires Translation-Several lines of evidence suggest that the P-STE functions through the 3Ј-UTR: (i) it requires a minimal distance from the 3Ј-UTR in order to function; (ii) it stabilizes transcripts harboring the MFA2 3Ј-UTR instability element, which is degraded by the poly(A) shortening-dependent pathway; and (iii) it can stabilize a chimeric transcript as part of residues 363-1138 of PGK1 with the STE3 3Ј-UTR. The results described for PGK1 above indicate that the P-STE requires translation elongation in order to promote stabilization of the mRNA. Preventing translation of the P-STE leads to a mRNA that is nearly 3-fold less stable when compared with an almost identical mRNA that allows translation of the P-STE. This is consistent with numerous reports that indicate the existence of a link between translation and mRNA decay (reviewed in Refs. 7, 36, and 37). The question remains as to what step(s) in the translation process is required for the instability/stability effect. Furthermore, the P-STE of PGK1 and TEF1/2 encode peptides that are 50% identical and 60% similar. Compared with the 27% identity and 31% similarity of an end-weighted gap analysis of both proteins, it may be possible that the protein sequence has an effect on P-STE function.
Several reports demonstrate a link between ribosomal scanning and mRNA stabilization (13,(42)(43)(44). In one case the authors demonstrate that insertion of inhibitory structures in the 5Ј-UTR can modulate the activity of stability determinants present in the mRNA (13). This led to the conclusion that disruption of ribosomal scanning in the 5Ј-UTR rather than changes in translation initiation efficiencies per se modulate mRNA stability. However, the strategies used to block ribosomal scanning could also block translation initiation making the interpretation of these results difficult. Another approach utilized to investigate the relationship between translation initiation and mRNA decay involved using mutants in translation factors. The results of these experiments demonstrated that the mRNA decay rate of a subset of mRNAs can be modulated by mutations in several translation initiation factors (42, 44 -46). More recent results using chimeric MFA2-PGK1 mRNAs demonstrated that both the PGK1 translation start codon context and the coding region act together to increase the stability of this mRNA (43). These results also demonstrate that the initiation and coding sequences are required for efficient translation of the mRNA. These results led to the interesting proposition that the nature of the translation initiation complex may modulate the rate of decapping and decay (43). Studies of c-fos instability elements further support these links, and additionally the requirement for spacing between an element and the 3Ј-UTR (16).
Our results indicated that the 65-nt P-STE is sufficient to promote stabilization of the MFA2 transcript when inserted in the translated portion of the transcript. If, as suggested, the nature of the initiation codon is important for the stabilization effect, this result would indicate that the MFA2 initiation codon could act in combination with the P-STE to promote stabilization of the chimeric transcript. The apparent disagreement of this conclusion with the results reported previously could be due to the nature of the chimeric mRNAs utilized in each case. The constructs used in this study harbor all the sequences from the MFA2 transcript, whereas the previously reported constructs only contained the 5Ј-UTR and the first three codons of the MFA2 transcript followed by the PGK1 coding region. As for the PGK1 transcript, it is possible that efficient translation initiation at the MFA2 AUG requires sequences downstream that are not present in the PGK1 transcript (specific combination of sequences). Additionally, whereas the presence of the P-STE in the context of PGK1 stabilizes an mRNA with the STE3 3Ј-UTR (Fig. 3), prior work has shown elements 5Ј of the P-STE can affect mRNA stability (11). Further work is necessary to delineate which elements of PGK1 are fully responsible for effects on the STE3 3Ј-UTR chimeras and if stability elements can differentially affect different destabilizing 3Ј-UTRs.
Previous results suggested that the P-STE does not block the activity of instability determinants located in the translated portion of the transcript. A PGK1-MAT␣1 hybrid transcript in which the MIE was inserted downstream and in-frame with the P-STE was still rapidly degraded (8,17). The 65-nt MIE is located in the protein coding region and is composed of two domains: a 5Ј 33-nt domain, which has a high content of rare codons; and the 3Ј 32-nt domain, which is predominantly AUrich. The function of the MIE is dependent on ribosomal translocation up to the last rare codon, which coincides with the 3Ј end of a 15-nt sequence complementary to the 18 S rRNA (35). It has been proposed that the rare codons together with a rRNA-mRNA interaction induces a translational pause that allows recognition of the downstream AU-rich portion of the MIE promoting rapid turnover (8,18). The MIE promotes rapid mRNA decay by increasing the rates of deadenylation and/or decapping (13). An attractive possibility to explain why the P-STE does not prevent MIE-promoted degradation is that translation termination, but not translational pausing, is required for the activity of the P-STE. This model suggests that the effect of the P-STE would not be manifested until the translation termination cycle is repeated. Since the MIE promotes rapid mRNA decay during the translation elongation cycle, the P-STE will not affect how this sequence element functions.
A Model for the Function of the P-STE on mRNA Stability-Based on the observations described above, one possible mode to explain the P-STE function is based on observations that suggest that the poly(A) tail-Pab1p interaction with the 5Ј end of the mRNA is an inhibitor of decapping (reviewed in Refs. 7 and 47), suggesting that the P-STE may protect the transcript from being rapidly decapped and/or from 3Ј 3 5Ј exonucleolytic attack (19). During translation of the P-STE, a change on the translating ribosome allows the modification of a factor that binds to the ribosome during the initiation process of translation (or de novo binding of a factor). After translation terminates, the putative factor is delivered to the 3Ј-UTR, where it is able to interact directly with the 3Ј-UTR and/or associated factor(s). As a consequence, these factors re-establish or maintain the interaction between the 5Ј and 3Ј ends, preventing rapid decay. The distance required between the P-STE and the 3Ј-UTR would provide certain flexibility in the mRNP to allow for the proper interaction between the two factors. It is possible that the P-STE is competing for the same factor that binds to the 3Ј-UTR and that promotes decapping. However, the fact that the activity of the P-STE requires some distance from the 3Ј-UTR suggests some kind of steric hindrance, suggesting the interplay of more than one factor. Alternatively, if the P-STE functions as a protein sequence, such as in MAT␣1 (8,17), this may directly affect translocation by the ribosome, a question that requires further analysis. P-STE function may require trans-acting factors, interactions between the stability and instability elements, and effects on interactions between the proteins associated with the 5Ј and 3Ј ends of an mRNA. This system may provide a novel means for a genetic approach to dissect both stability and instability elements, and the interactions between the two.