Maintenance and regulation of mRNA stability of the Saccharomyces cerevisiae OLE1 gene requires multiple elements within the transcript that act through translation-independent mechanisms.

The Saccharomyces cerevisiae OLE1 gene encodes a membrane-bound Delta-9 fatty acid desaturase, whose expression is regulated by unsaturated fatty acids through both transcriptional and mRNA stability controls. In fatty acid-free medium, the mRNA has a half-life of 10 +/- 1.5 min (basal stability) that drops to 2 +/- 1.5 min when cells are exposed to unsaturated fatty acids (regulated stability). A deletion analysis of elements within the transcript revealed that the sequences within the protein-coding region that encode transmembrane sequences and a part of the cytochrome b5 domain are essential for the basal stability of the transcript. Deletion of any of the three essential elements produced unstable transcripts and loss of regulated instability. By contrast, substitution of the 3'-untranslated region with that of the stable PGK1 gene did not affect the basal stability of the transcript and did not block regulated decay. Given that Ole1p is a membrane-bound protein whose activities are a major determinant of membrane fluidity, we asked whether membrane-associated translation of the protein was essential for basal and regulated stability. Insertion of stop codons within the transcript that blocked either translation of the entire protein or parts of the protein required for co-translation insertion of Ole1p had no effect. We conclude that the basal and regulated stability of the OLE1 transcript is resistant to the nonsense-mediated decay pathway and that the essential protein-encoding elements for basal stability act cooperatively as stabilizing sequences through RNA-protein interactions via a translation-independent mechanism.

The Saccharomyces cerevisiae OLE1 gene encodes a membrane-bound ⌬-9 fatty acid desaturase, whose expression is regulated by unsaturated fatty acids through both transcriptional and mRNA stability controls. In fatty acid-free medium, the mRNA has a half-life of 10 ؎ 1.5 min (basal stability) that drops to 2 ؎ 1.5 min when cells are exposed to unsaturated fatty acids (regulated stability). A deletion analysis of elements within the transcript revealed that the sequences within the protein-coding region that encode transmembrane sequences and a part of the cytochrome b 5 domain are essential for the basal stability of the transcript. Deletion of any of the three essential elements produced unstable transcripts and loss of regulated instability. By contrast, substitution of the 3-untranslated region with that of the stable PGK1 gene did not affect the basal stability of the transcript and did not block regulated decay. Given that Ole1p is a membrane-bound protein whose activities are a major determinant of membrane fluidity, we asked whether membrane-associated translation of the protein was essential for basal and regulated stability. Insertion of stop codons within the transcript that blocked either translation of the entire protein or parts of the protein required for co-translation insertion of Ole1p had no effect. We conclude that the basal and regulated stability of the OLE1 transcript is resistant to the nonsense-mediated decay pathway and that the essential protein-encoding elements for basal stability act cooperatively as stabilizing sequences through RNA-protein interactions via a translation-independent mechanism.
Unsaturated fatty acids are essential components of membrane lipids that determine in large part the fluid properties of membrane bilayers. In eukaryotes, unsaturated fatty acids are synthesized by fatty acid desaturases that form double bonds in fatty acyl chains by an iron-requiring, oxygen-dependent mechanism. Because of the need to stringently control the acyl composition of membrane lipids, fatty acid desaturases are highly regulated enzymes that respond to a number of stimuli, including changes in environmental temperature, nutrient fatty acids, and cellular oxygen levels. In the yeast Saccharomyces cerevisiae, unsaturated fatty acids are formed by Ole1p, a membrane-bound ⌬-9 desaturase that converts long chain saturated fatty acyl-CoA substrates into monounsaturated species (1,2). Our laboratory has shown that the OLE1 gene is regulated by a number of mechanisms, including transcriptional and mRNA stability controls (3)(4)(5)(6)(7)(8).
Although most studies of gene regulation are focused on mechanisms that control the rate of transcription, mRNA degradation systems are now known to play a central role in the control of gene expression. The half-life of an mRNA species determines the number of times an individual transcript can be translated, which in turn affects the total amount of protein that can be produced by a gene at a given rate of transcription. This importance of mRNA degradation in determining the level of expression of specific genes is supported by studies that show that the decay rates of different mRNAs vary widely, and the half-lives of some transcripts have been shown to change in response to specific signals (5, 9 -11).
Only a few examples of regulated mRNA decay in response to external cues have been described in yeast. Two genes from S. cerevisiae have been shown to respond to the changes in carbon source that are associated with the diauxic shift during which cells adapt to glucose depleted conditions.
The SDH2 rate of mRNA turnover depends on the availability of glucose (12). When glucose is added to cells previously grown on the non-fermentable carbon source, glycerol, the halflife of the transcript drops from Ͼ60 min to ϳ5-7 min (10,12,13). By contrast, the TIF51A transcript is stabilized in glucosegrown cells, with a half-life of 20 min that drops to 7 min when cells are transferred to glycerol-containing medium (14).
The half-life of the OLE1 transcript is also regulated when cells are exposed to unsaturated fatty acids. Yeast cells readily import long chain (C 14 -C 18 ) acids and incorporate them into membrane lipids, resulting in the down-regulation of OLE1 transcription (4) and a rapid decrease in the half-life of its mRNA (5). In cells grown in fatty acid-free medium, OLE1 mRNA is a moderately stable species with a half-life of about 10 Ϯ 1.5 min (5), whereas exposure to unsaturated fatty acids for 15 min reduces its half-life to less than 2 min.
Progress has recently been made in understanding the gen-eral mechanisms by which the process of mRNA degradation occurs. In Saccharomyces, many transcripts appear to be degraded through a general pathway of mRNA decay that is initiated by deadenylation of the poly(A) tail (15)(16)(17)(18). The subsequent displacement of poly(A)-binding proteins that stabilize the binding of translation initiation factors to the 5Ј cap makes the transcript competent for decapping by the enzymes Dcp1p and Dcp2p. After removal of the 5Ј cap, the mRNA is rapidly degraded by the 5Ј 3 3Ј exonuclease, Xrn1p. Alternatively, some mRNAs can be degraded in a 3Ј-5Ј direction by exonucleases that are associated with the exosome complex (19). The rate of the deadenylation and decapping reactions for each transcript appears to be controlled by widely divergent sequence elements in different mRNAs (17), indicating that these common functions in the general decay pathway respond to a number of different regulatory mechanisms. Translation also appears to play an important role in governing the decay of some transcripts in the general pathway. For some mRNAs, sequences that promote efficient translation also stabilize the transcript (17,20). For example, mutants that interfere with ribosomal loading and translation initiation promote increased rates of deadenylation (17,18), and mutations that alter the context of the start codon can destabilize the highly stable PGK1 mRNAs.
A second translation-dependent decay pathway involves the degradation of aberrant mRNAs. In that system, the introduction of premature stop codons into protein coding sequences destabilizes some transcripts (21). The termination of translation at the nonsense codon and the assembly of a surveillance complex triggers the rapid decapping and degradation of the mRNA (22). Transcripts that are degraded by the nonsense decay pathway require the presence of a destabilizing element 3Ј to the premature stop codon (23).
Elements within the 3Ј-untranslated region (UTR) 1 of many eukaryotic mRNAs also play an important role in governing mRNA stability. The decay rate of a number of regulated transcripts are controlled by AU-rich elements (AREs) located within the 3Ј-UTR, and mutations in those elements have been shown to reduce the rate of poly(A) degradation and the overall rate of decay (14). The role of AREs in mRNAs appears to be somewhat variable, suggesting that they may recognize more than one type of binding protein. An ARE in the yeast MFA2 transcript, for example, destabilizes the mRNA under all tested growth conditions, whereas the ARE in the glucose-sensitive TIF51A transcript acts as the primary determinant of its carbon source-regulated decay (14).
An important part of understanding how mRNA stability is controlled requires identifying and characterizing the RNA elements that regulate its turnover. Unlike DNA binding proteins, trans-acting RNA-binding proteins that stabilize or destabilize transcripts rarely recognize distinct nucleotide se-quences and instead bind to relatively long elements, which suggests that the context and secondary structure of the sequences play key roles in the process.
We previously determined that some elements involved in the OLE1 mRNA stability control mechanism are associated with the 5Ј-UTR of the transcript. Replacing the native 5Ј-UTR with sequences from the yeast GAL1 5Ј-UTR produces a functional OLE1 transcript with an intermediate half-life that is not regulated by unsaturated fatty acids (5). Those results suggested that the 5Ј-UTR of OLE1 mRNA contains sequence elements required for fatty acid-triggered destabilization. Although these elements appear to be necessary for the generation of a normal half-life, they do not appear to be sufficient to establish the normal control of basal and regulated stability of the mRNA.
In this paper, using a series of deletion and substitution mutations, we identify multiple, widely separated elements within the protein-coding sequence of the OLE1 transcript that are essential for its basal stability. These encode residues that are a part of the membrane-spanning regions of the protein and a part of its C-terminal cytochrome b 5 domain. Unexpectedly, the substitution of the 3Ј-UTR region of the transcript with that of the highly stable PGK1 gene does not stabilize the transcript or block its fatty acid-regulated stability. Given that the ⌬-9 desaturase is a major determinant of membrane fluid properties, we also investigated whether membrane-associated translation of the protein is involved in the mRNA stability regulation mechanism. Insertion of stop codons throughout the transcript, however, including those that block translation of the membrane-spanning regions failed to affect either basal or regulated stability of the transcript. We conclude that OLE1 mRNA stability is regulated by a novel system that is resistant to nonsense-mediated decay and the presence of AU-rich 3Ј-UTR elements. We further conclude that the membrane-associated translation of the Ole1p is not an integral part of the decay process.

Strains and Growth Conditions
S. cerevisiae strains used in this study are shown in Table I. Escherichia coli strains XL-BLUE (Stratagene, La Jolla, CA) and DH5␣ (Invitrogen) were used for DNA cloning and propagation of plasmids. E. coli cells were grown according to standard protocols (24,25).
Yeast cells were grown as previously described (5). The cells were grown at 30°C in complete synthetic medium plus 1% Tergitol Nonidet P-40 (Sigma) lacking appropriate amino acids (based on the selectable marker used in the incorporated plasmid). Typically, the carbon source was 2% glucose. However, when the cells contained the gene of interest fused to a galactose-inducible promoter, they were grown overnight in medium containing 2% raffinose. 2% galactose or 2% glucose was added to the growth medium when required to induce or repress transcription of genes under the control of the GAL1 promoter. Unsaturated fatty acids (obtained from Sigma) were added to the growth medium to a final concentration of 1 mM. Thiolutin (received as a gift from Pfizer) was added to a final concentration of 15 g/ml.

Plasmid Constructions
Plasmids used in this study are described in Table II. Most of the manipulations in the OLE1 protein-coding sequence ( Fig. 1) were made by PCR amplification and splicing through the splicing by overlap extension (SOE) technique (26). The constructs were verified by restriction analysis and DNA sequencing as required. Two yeast centromerebased plasmids, pGAL-OLE2.8 (5) and pGAL-OLE-207, were used as parent vectors to carry out several modifications within the OLE1 gene sequences. pGAL-OLE-207 is similar to pGAL-OLE2.8 but has OLE1 gene sequences starting from base Ϫ207 of the 5Ј-UTR and including the entire protein coding sequences and 1 kb of 3Ј-UTR fused to the GAL1 promoter through a 20-base pair linker region.
Transmembrane loops of OLE1 gene were replaced by amplification of the entire FAH1 TM region using primers with an incorporated AflIII site. Plasmid p(ϪB)HpaI was created by destroying the unique BstX1 site in plasmid pCRScriptSKϩ. This was followed by insertion of an HpaI fragment encompassing the entire OLE1 gene cloned into the SrfI site to generate plasmid pMV055. The previous PCR product was digested with AflIII and cloned into the NcoI sites of the OLE1 gene in plasmid p(ϪB)HpaI, replacing both of its TM elements with FAH1 TMs. An OLE1 gene fragment derived from p(ϪB)HpaI that contains all of the protein-coding sequences distal to TM1 and including 1 kb of 3Ј-UTR was then cloned adjacent to the FAH1 TM1 in plasmid pMV055 to generate pMV056. A SnaBI/PacI fragment from pMV056 containing the entire modified protein coding sequences was then cloned into the same sites in vector pGAL-OLE-207 to generate pMV057 (pGAL-OLE-FAH-tm1).
To replace OLE1 TM2, a fragment that contains the OLE1 5Ј-UTR and the N-terminal part of the OLE1 protein-coding sequence (759 bp) including TM1, was amplified using primers prMV043 and prMV051 from plasmid p(ϪB)HpaI. In a separate reaction, FAH1 TM2, part of the OLE1-coding sequence and the OLE1 3Ј-UTR were amplified using primers prMV052 and prMV048 from plasmid pMV055. The two PCR products were fused by the SOE technique followed by digestion with restriction enzymes XbaI/HindIII and insertion into the same sites in plasmid pGAL-OLE-207 to create pGAL-OLE-FAHtm2.
To replace both OLE1 TMs, a fragment that contains the OLE1 5Ј-UTR, the N-terminal part of OLE1 protein-coding sequence linked to FAH1 TM1, and part of the OLE1 sequence between the TM loops was amplified with primers prMV043 and prMV051 from plasmid pMV056. In a separate reaction, a fragment containing the remaining elements of the OLE1 intermembrane loop sequences, FAH1 TM2, the C-terminal part of the OLE1 protein coding sequence, and the OLE1 3Ј-UTR was amplified using primers prMV052 and prMV048 from plasmid pMV055. The two fragments were then fused by the SOE technique, digested with restriction enzymes XbaI/HindIII, and inserted into the same sites in plasmid pGAL-OLE-207 to create pGAL-OLE-FAHtm1&2.
Deletion of Sequences from the Start Codon to the Beginning of Transmembrane Loop 1 (Plasmid pMV088)-The OLE1 5Ј-UTR (220 bp) was amplified by PCR using primers prMV043 and prMV133 from plasmid pGAL-OLE2. 8. In a second reaction the OLE1-coding sequence beyond the start of TM regions (ϩ301 bp) and OLE1 3Ј-UTR (1 kb) was amplified using primers prMV132 and prMV048 from plasmid pMV071. Both PCR products were spliced by SOE and cloned into SrfI site in  pCRScriptSKϩ plasmid to produce pMV085. The XbaI/HindIII fragment (2451 bp) from plasmid pMV085 was then cloned into the same sites in pGAL-OLE-207 plasmid vector to generate pMV088.
Deletion of the Protein Coding Sequence between the Transmembrane Loops of OLE1 (pMV089)-A 716-bp fragment that contains the OLE1 5Ј-UTR (220 bp) and protein-coding sequence to the end of OLE1 TM1 was amplified by PCR using the primers prMV043 and prMV134 from plasmid pMV071. In a second reaction, the OLE1-coding sequence that includes TM2 and 3Ј-UTR (1 kb) was amplified using primers prMV135 and prMV048 from plasmid pGAL-OLE2.8. The PCR products were spliced by SOE, and the resultant product was cloned into the SrfI site in pCRScriptSkϩ plasmid to make pMV086. The XbaI/HindIII fragment (2453 bp) of pMV086 was then cloned into the same sites on pGAL-OLE-207 vector to generate plasmid pMV089.
Deletion of Protein-coding Sequence between Transmembrane Loop 2 and PacI Site-Primers prMV043 and prMV138 were used to amplify 220 bp of OLE1 5Ј-UTR and 918 bp of OLE1 protein-coding sequence until the end of TM2 using plasmid pMV071 as a template. In another PCR reaction, primers prMV139 and prMV048 were used to amplify 195 bp of OLE1 protein-coding sequences starting from PacI site to the stop codon and 1 kb of 3Ј-UTR using plasmid p(ϪB)HpaI as template. The PCR products were spliced by SOE and cloned into SrfI site in pCRScriptSkϩ plasmid to make pMV087. The XbaI/HindIII fragment of pMV087 (2306 bp) was then cloned into the same sites in pGAL-OLE-207 plasmid vector to generate pMV090.
Substitution of Ole1p-coding Sequence from PacI Site in the Cytochrome b 5 -like Region to the Stop Codon with GFP Protein-coding Sequences-The GFP protein-coding sequence from plasmid GFPmut3.1 was amplified by PCR using primers prMV072 and prMV074. In a separate reaction the OLE1-coding sequence beyond the PacI site in the cytochrome b 5 -like region to the stop codon and OLE1 3Ј-UTR was amplified with primers prMV075 and prMV048 from plasmid p(ϪB)HpaI. The PCR products were spliced by SOE and cloned into SrfI site in pCRScriptSKϩ plasmid to make PCR_GFPm (2). A PacI/HindIII fragment from plasmid PCR_GFPm (2) was ligated into same sites in plasmid pGAL-OLE2.8 to generate plasmid pGAL-OLE-GFPm (2).
The OLE1 protein-coding sequence from PacI site to the stop codon was also replaced by the GFP protein-coding sequence from plasmid pGFPuv (Clontech). A 716-bp fragment containing the GFP-coding sequence was amplified from pGFPuv with primers prMV068 and prMV069. In a separate reaction the OLE1 protein-coding sequences from the PacI site to the stop codon and OLE1 3Ј-UTR was amplified using primers prMV070 and prMV048 from plasmid p(ϪB)HpaI. The two PCR products were spliced by SOE and cloned into SrfI site in plasmid pCRScriptSKϩ plasmid to generate plasmid PCR_GFP1. PacI/ HindIII digestion of PCR_GFP1 released a 1716-kb fragment that was then cloned into same sites on pGAL-OLE2.8 to make plasmid pGAL-OLE-GFP1.
Linkage of TM1, TM2, and Cytochrome b 5 Essential Elements within the OLE1 5Ј-and 3Ј-UTR Regions-The HpaI fragment of OLE1 containing the entire mRNA-encoding sequence was digested with SalI/ PacI to remove elements of the OLE1 protein-coding sequence extending from the N-terminal SalI site to the PacI site within the cytochrome b 5 domain. PCR-amplified fragments encoding the TM1 and TM2 regions were fused by the introduction of a SalI restriction site in the forward PCR primer and a PacI site in the reverse primer to construct vectors pGAL-OLE1-TM1-cytb 5 , pGAL-OLE1-TM2 cytb 5 , and vector pE31, which contain all three essential elements of the protein-coding sequence.
Replacing the OLE1 3Ј-UTR with the PGK1 3Ј-UTR (Plasmid pGAL-OLE-PGK3Јutr)-The OLE1 protein-coding sequence extending from ϩ1 to ϩ1387 was amplified with primers prMV005/prMV032 from pGAL-OLE2.8 and the PGK1 3Ј-UTR (1 kb after the stop codon) was amplified by PCR with primers prMV034/prMV035 from genomic DNA (W3031A). Each of the PCR fragments was cloned separately into the SrfI site in pCRScriptSKϩ to generate plasmids pMV048 (containing the OLE1-coding sequence) and pMV049 (containing PGK1 3Ј-UTR). AatII/NotI digestion of pMV049 released a 1-kb fragment that was inserted into AatII/EcoRV sites in plasmid vector pMV048 behind the OLE1-coding sequence. The resulting plasmid pMV050 was then digested with XhoI and Klenow-filled to blunted end followed by PacI cut to release 2533-kb fragment. This fragment was inserted into HindIII (Klenow filled-in)/PacI sites in pGAL-OLE2.8 to generate plasmid pGAL-OLE-PGK3Јutr.
Mutation to Introduce GCG in Place of Translational Start Codon ATG (Plasmid pMV061) in OLE1 Gene-OLE1 start codon was mutated through PCR technique. Primers prMV059 and prMV060 are complementary and a three-base pair change compared with the original sequence was incorporated to facilitate the conversion of the ATG start codon to GCG. OLE1 5Ј-UTR (937 bp) was amplified using primers AGM116 and prMV060. In a separate reaction the OLE1-coding sequence and 3Ј-UTR was amplified using primers prMV059 and prMV032. The two PCR products were spliced by SOE followed by SnaBI/PacI digestion. This fragment was then cloned into the same sites in pGAL-OLE2.8 to make plasmid pMV061.
Introduction of a Stop Codon before TM Loop 1 in the OLE1-coding Sequence (Plasmid pMV071)-The PCR primers prMV082 and prMV083 are complementary to each other and include a three-base pair change in the sequence that will introduce a stop codon at residue ϩ323 before the start of TM1 in the OLE1 gene. The OLE1 5Ј-UTR (220 bp) and protein-coding sequence from ϩ1 to ϩ340 was amplified by PCR using primers prMV043 and prMV083. In a separate reaction the OLE1 protein-coding sequence from position ϩ301 to the stop codon was amplified by PCR using primers prMV082 and prMV048 from plasmid p(ϪB)HpaI. The two PCR products were spliced through SOE followed by digestion with SnaBI/HindIII and insertion into the same sites in plasmid pGAL-OLE2.8 to generate pMV071.
Introduction of a Stop Codon through a Frameshift Mutation in the Cytochrome b 5 -like Region in OLE1 Protein-coding Sequence-Plasmid pGAL-OLE2.8 was cut with PacI, filled-in by Klenow polymerase, and religated to produce pAM165. The resulting plasmid contains a frameshift mutation that introduces a stop codon between residues ϩ1353 to ϩ1355 within the cytochrome b 5 domain of Ole1p.
mRNA Decay Measurements mRNA decay rates were essentially determined as previously described (5). Transcription was arrested either by using transcriptional inhibitor thiolutin (15 g/ml) or through glucose repression of the yeast GAL1 promoter. RNA isolations and quantitative Northern blot analysis of cells after transcription arrest were performed as previously described (5).

Preparation of Radiolabeled Probes
For the detection of OLE1 mRNA, a radiolabeled DNA probe was made using a 1.0-kb EcoRI fragment from the OLE1-coding sequence isolated from YEp352-OLE4.8. The OLE1-FAH1 chimeric mRNAs were detected by probing against either one or both of the substituted FAH1 transmembrane loops. The DNA templates for FAH1 probe preparation were prepared by PCR. Primers used to amplify TM1 (506 bp) are prMV056 and prMV057, and primers used to amplify TM2 (501 bp) are prMV041 and prMV058. Probes for GFP detection were made from a template amplified by PCR from plasmid pGFPmut3.1 (Clontech). The primers used are prMV072 and prMV074. The DNA template for scR1 probe was amplified using the primers prMV016 and prMV017. For all experiments the internal control was provided by a 1.0-kb HindIII-KpnI fragment of the phosphoglycerate kinase (PGK1) gene isolated from plasmid pRIP1PGK (27) unless specifically stated. All DNA fragments were separated by agarose gel electrophoresis in 1ϫ TAE buffer and purified by Gene Clean II (Bio101). The DNA fragments were labeled to high specific activity with [␣-32 P]dATP (PerkinElmer Life Sciences) by the random primer extension method using NEBlot kit from New England Biolabs. Unincorporated nucleotides were removed from the sample using a Sephadex G-50 spin column. The specific activities of labeled probes were determined by liquid scintillation counting.

OLE1 mRNA Degradation under Basal and Regulated Conditions Involves Decapping by Dcp1p and Xrn1p-dependent Ex-
onuclease Digestion-Recent studies show that many yeast mRNAs are degraded by a general deadenylation-dependent pathway that involves removal of the 5Ј cap by the Dcp1p-Dcp2p complex and subsequent degradation by the 5Ј-3Ј exonuclease, Xrn1p (17,18,28). An important question was whether the basal and regulated decay of the native OLE1 transcript occurred through that pathway or through an alternative mechanism. Fig. 2 shows the effects of null mutations in the XRN1 and DCP1 genes on OLE1 mRNA decay. Panel A shows that in wild type cells the OLE1 transcript decays with a half-life of 10 min in fatty acid-free medium and 2.5 min on medium containing 1 mM linoleic acid (18:2) Sigma. In xrn1⌬ cells (panel B) the lifetime of the transcript is dramatically increased to 45 and 23 min under the same conditions, indicating that fatty acids continue to exert some regulation even though the rate of decay in the cells exposed to 18:2 is reduced by about 10-fold. By contrast, disruption of the DCP1 gene (panel C) stabilized the transcript under basal and fatty acidmediated conditions, resulting in half-lives under both circumstances that were greater than 50 min. Under the conditions of the experiment we could not detect significant differences between the stability of the OLE1 transcripts from dcp1⌬ cells grown under basal or regulated conditions, suggesting that the decapping reaction is an essential step in determining its basal and regulated stability.
Deletion and Substitution Analysis of OLE1 mRNA Reveals Sequences in the Protein-coding Region That Are Required for Basal and Regulated Stability-Previously we reported that the 5Ј-UTR of OLE1 mRNA is required for its basal and unsaturated fatty acid-mediated stability (5). This element, however, is insufficient to confer the wild type pattern of basal and regulated stability on heterologous mRNAs. To identify other essential elements that reside in the protein-coding or 3Ј-UTR regions of the mRNA we performed a deletion and substitution analysis of the entire transcript. The relationship of the structural domains of Ole1p to features within the protein-coding region of the OLE1 transcript that are relevant to those experiments are shown in Fig. 1. Ole1p is a bifunctional protein that consists of an N-terminal fatty acid desaturase domain fused to a C-terminal cytochrome b 5 domain. The protein is anchored to the membrane through two hydrophobic regions within the desaturase domain, and each hydrophobic element appears to traverse the membrane as a loop, so that the active sites of the two functional domains are located on the cytosolic side of the membrane.
Two initial test vectors that were constructed for these experiments contained deletions in sequences that encode the TM loop elements. TM loop 1 is positioned between amino acids 116 and 159, and TM loop 2 is positioned between amino acids 257 and 305 (Fig. 1) (2).
Plasmid pGAL-OLE-BstEII⌬ contains a 522-bp internal de-letion that encodes amino acids ϩ159 to ϩ333. The polypeptide encoded by that plasmid retained loop region 1, but the deletion removed all of the sequences positioned between the TM loops and TM loop 2 ( Figs. 1 and 3b). Plasmid pGAL-OLE-NcoI⌬ contained a 607-bp internal deletion that encodes residues from ϩ101 to ϩ303. The polypeptide that was encoded by that plasmid lacked both TM domains of the desaturase (Figs. 1 and  3c). Although these deletions did not alter the reading frames, neither plasmid would support growth of an ole1⌬ gene-disrupted strain on fatty acid-free medium, indicating that both encoded a non-functional protein (data not shown). Both vectors produced the mutant OLE1 transcripts under control of the GAL1 promoter, which is inducible by galactose and can be rapidly arrested by the addition of glucose. The mRNA decay measurements in wild type cells containing those plasmids revealed that they produced unstable transcripts, with half-lives of ϳ3 min in the presence and absence of unsaturated fatty acids (Fig. 4) compared with the native transcripts (inhibited with the RNA polymerase inhibitor, thiolutin), which exhibits the normal basal half-life of 10 min and a regulated half-life of 2.5 min.

Essential Elements for Basal and Regulated Decay within the OLE1 Protein-coding Region Include the Transmembrane-encoding Elements and Sequences Associated with the C-terminal
Part of the Cytochrome b 5 Domain-The results from the above deletion analysis suggested that specific elements within the transmembrane-encoding sequences might play a role in stabilizing the transcript and in the regulation of its stability. To test the possibility that the membrane-spanning sequences were essential for stability regulation, the OLE1 TM loop regions were replaced with non-homologous sequences (Ͻ30% nucleotide sequence identity and Ͻ21% peptide sequence identity) that encode membrane-spanning regions from the Saccharomyces FAH1 gene. Fah1p is a membrane-bound fatty acid biosynthetic enzyme that hydroxylates 26-carbon fatty acids (29). It is a bifunctional, cytochrome b 5 fusion protein that has similar structural features to Ole1p. In Fah1p the fused cytochrome b 5 2. Decay of the OLE1 transcript in wild type, xrn1⌬, and  dcp1⌬ strains. Phosphor images of Northern blots of total cellular RNA extracted from two parallel cultures containing no fatty acids or 1 mM 18:2 (added 15 min before transcription arrest) at indicated intervals after the addition of the inhibitor. A, wild type cells containing plasmid pGAL-OLE2.8 were induced by 2% galactose and inhibited by the addition of 2% glucose to both cultures. B, xrn1⌬ cells containing plasmid pGAL-OLE2.8 as described in A. Northern blots were probed with radiolabeled OLE1 and PGK1 sequences. C, phosphor images of Northern blots of the native OLE1 transcript in dcp1⌬ cells inhibited with thiolutin (15 g/ml). Total RNA was extracted from cells collected at the indicated times after transcription arrest and subjected to northern blotting. The blots were first probed with OLE1 gene sequences followed by PGK1 and scR1 sequences as internal controls .   FIG. 3. Chimeric GAL1 promoter and OLE1 gene constructs with transmembrane loop deletions. Diagrams showing the relevant elements of plasmid pGAL-OLE2.8, which encodes the wild type OLE1 mRNA sequences, and its derivatives that contain deletions in the transmembrane-encoding regions. The constructs are under the control of the GAL1 promoter, which includes upstream activation sequences (UAS), basal promoter elements, and 65 base pairs of the GAL1 5Ј-UTR. The GAL1 promoter sequences are fused to OLE1 sequences that include 220 bp of the 5Ј-UTR, protein-coding sequences, and 3Ј-UTR. a, plasmid pGAL-OLE2.8. b, plasmid pGAL-OLE-BstEII⌬ contains an in-frame deletion that removes the region between the TMs and TM2. c, plasmid pGAL-OLE-NcoI⌬ contains an in-frame deletion that removes TM1, TM2, and the intervening region.
Three chimeric plasmids were made by replacing one or both of the OLE1 transmembrane loops with the relevant FAH1 sequences (Fig. 5). Plasmids pGAL-OLE-FAHtm1 and pGAL-OLE-FAHtm2 contained replacements of the single transmembrane loop regions; plasmid pGAL-OLE-FAHtm1&2 contained substitutions in both regions.
The half-lives of the mRNAs produced from the three constructs expressed in fah1⌬ strains were determined using RNA probes complementary to the inserted FAH1 sequences (Fig. 5). All three chimeric genes produced unstable mRNAs with halflives of 3-4 min in cells exposed to either fatty acid free or 18:2-supplemented growth medium. This strong correlation of the decay pattern with those from the transcripts generated from the TM loop deletion constructs further supports that hypothesis that sequences in the OLE1 transmembrane loop regions are essential elements for the basal and regulated stability of the transcript.
A stability-determining element was also uncovered within the cytochrome b 5 domain of the OLE1 protein-coding sequence when sequences between the PacI restriction site to the C terminus were replaced with GFP protein-coding sequences. Transcripts generated from two plasmid constructs, pGAL-OLE-GFP1 and pGAL-OLE-GFPm (2), showed that both mRNAs are very unstable and decay with a half-life of about 3 min in the presence and absence of fatty acids (Fig. 6).
By contrast deletion analysis revealed that the sequences that separate the stabilizing TM1, TM2, and cytochrome b 5 elements do not play a direct role in either basal or regulated mRNA stability. Plasmids pMV088, pMV089, and pMV090, respectively, contain deletions of the sequences from the start codon to TM1, the sequences between the transmembrane loops, and a sequence from TM2 to a PacI site within the cytochrome b 5 -like domain (Fig. 7). The half-life of transcripts produced from all three plasmids is 13-14 min in the absence of fatty acids and 4 min in the presence of 1 mM 18:2 (Fig. 7).
The Context and Spacing of the Essential Stabilizing Sequences Play a Role in Determining Basal and Regulated Stability-The essential stabilizing TM1, TM2, and cytochrome b 5 elements were tested by placing them between the OLE1 5Јand 3Ј-UTR regions to determine whether they were sufficient to establish the wild type pattern of mRNA decay. A series of vectors containing individual TM1, TM2 sequences, and the TM1 and TM2 elements fused together generated highly unstable and unregulated transcripts with half-lives of Ͻ3 min under basal and regulated conditions (data not shown). Then, to determine whether linkage of all of the essential elements was sufficient to exert the wild type pattern of control, we constructed vectors that contain all three stabilizing sequences placed between the 5Ј-and 3Ј-UTR regions (Fig. 8). Vector pE31 contains the fused TM1, TM2, and cytochrome b 5 regions Phosphor images of Northern blots of total RNA isolated from wild type strain DTY10 containing plasmids pGAL-OLEBstEII⌬ (a) and pGAL-OLENcoI⌬ (b). Transcription was induced by adding 2% galactose to the growth medium and arrested by adding 2% glucose and 15 g/ml thiolutin. The blots were probed with OLE1 gene sequences, stripped, and reprobed with PGK1 sequences used as an internal control. Two OLE1 transcripts are detected in each experiment. The longer form (A) is the native mRNA (inhibited by thiolutin), and the shorter form (B) contains deletions in the transmembrane loops regions (inhibited by glucose). c and d, plots of decay rates of deleted transcripts shown in panels a and b in the absence (q, t1 ⁄2 ϭ 2.5 min) and the presence of (E, t1 ⁄2 ϭ 2.5 min) of 1 mM 18:2. Data was normalized to the corresponding steady state levels of PGK1 and plotted as the log of percentage mRNA remaining versus time after transcription arrest. Lines through data points indicate 2nd order regression curves, r 2 Ն 0.9. flanked by the OLE1 5Ј-and 3Ј-UTR regions. Analysis of the chimeric transcript, however, revealed that they have a halflife of Ͻ3 min under both basal and regulated conditions. We also constructed and tested vectors in which the sequences that separate those elements were replaced by corresponding FAH1 gene sequences. Those chimeric transcripts also showed simi-lar instabilities under all conditions tested (data not shown). Taken together, these data indicate that the stabilizing elements within the protein-coding regions are necessary, but not sufficient, to determine the wild type levels of basal and regulated stability. Furthermore, although the individual regions that flank the TM and cytochrome b 5 elements are not essential for the wild type patterns of mRNA stability, the context and spacing of those intervening sequences appear to have some role in determining the wild type pattern of OLE1 mRNA decay, possibly by providing essential spacing between a minimum number of the essential elements or through secondary structure contributions to the transcript.
Translation of Ole1p and mRNA Stability Regulation-The essential mRNA-stabilizing elements in the protein-coding region of the OLE1 transcript could function in several different ways. The nucleic acid sequences could serve as a recognition sites for RNA-binding proteins that stabilize the mRNA. Alternatively, mRNA decay might be controlled by the translation of the nascent polypeptide in a mechanism analogous to the decay mechanism for ␤-tubulin mRNA (30). Given that sequences in the transcript that are essential for basal and regulated stability are located in regions that encoded the transmembrane elements, that mechanism might be triggered by the co-translational insertion of the nascent protein into the membrane in FIG. 5. Decay of chimeric OLE1 mRNAs containing FAH1 TM elements. a-d, schematic diagrams illustrating substituted sequences in each plasmid and corresponding decay patterns. All the constructs were under the control of GAL1 promoter elements (gray box) that include 57 bases of the GAL1 5Ј-UTR. The straight line represents OLE1 gene sequences, and the dashed line represents the vector sequences. The GAL1 promoter sequences are fused through a 20-base pair linker region to the OLE1 sequences at base Ϫ207. a, parent plasmid pGAL-OLE-207 containing wild type OLE1 mRNA-encoding sequences. b, plasmid pGAL-OLE-FAHtm1, 177 bp of the OLE1-coding sequence (from base ϩ302 to base ϩ479) that encodes TM1 was replaced with 250 bp of FAH1 TM1 sequences (from base ϩ575 to base ϩ825). c, in plasmid pGAL-OLE-FAHtm2, 144 bp (from base ϩ766 to base ϩ910) of OLE1 TM2-coding sequence was replaced with 126 bp of FAH1 TM2 coding sequence (from base ϩ853 to base ϩ979). d, in plasmid pGAL-OLE-FAHtm1&2, both the transmembrane domains were replaced with FAH1 TM sequences used in b and c. Transcription of the chimeric mRNAs was induced with 2% galactose and arrested with 2% glucose as described in Fig. 4. The chimeric RNAs were hybridized to probes made from the same FAH1 TM sequences that were substituted in each plasmid. There are two transcripts detected by the FAH1 probes. The longer one is the OLE1::FAH1 chimeric transcript, and the shorter one is a part of the FAH1 transcript fragment that is disrupted with the LEU2 gene. To test the hypothesis that translation through the transmembrane loop sequences is essential for basal and regulated mRNA decay, vectors were constructed that would produce full-length OLE1 mRNAs with stop codons inserted at different positions within the protein-coding sequence. Vector pMV071 contains a stop codon immediately before the TM loop 1-encoding region by replacing bases ACC with GTT. The resulting mRNA has unaltered sequences within the transmembrane loop regions and encodes a 107-residue soluble N-terminal polypeptide. Analysis of the decay kinetics of the mutant transcript in wild type cells showed that it exhibits half-lives of 10.5 and 3.5 min under basal and regulated conditions, which are identical to those of the native mRNA species (Fig. 9b). These results indicate that the mRNA sequences within the transmembrane loop regions and not the protein encoded by those elements are essential for basal and fatty acid-regulated stability.
Plasmids containing other modifications of the protein-coding region showed that the context of the start codon or translation of the cytochrome b 5 region was not required for basal or fatty acid-mediated regulation of mRNA stability. A plasmid in which the ϩ1 ATG start codon was replaced with a GCG sequence (plasmid pMV061) and one in which a stop codon was introduced into the cytochrome b 5 domain (plasmid pAM165) both produced full-length, but non-functional transcripts that could not rescue the ole1⌬ strain (Fig. 9a). A kinetic analysis of both transcripts showed that they decay with the same basal and regulated half-lives as the wild type mRNA. (Fig. 9, c and  d). Thus, it appears that translation of the entire OLE1 polypeptide is not essential for basal or regulated stability of the transcript.
The 3Ј-UTR Region Is Not Essential for Basal or Regulated Stability of the OLE1 Transcript-Sequences in the 3Ј-UTR regions have previously been shown to be involved in mRNA stability regulation of many yeast and mammalian genes (15,20,31,32). An analysis of expressed sequence tag data indicates that the OLE1 3Ј-UTR region includes an essential "efficiency element" (UAUAUA) 252 bases downstream of the stop codon. Within that region are multiple ARE sequences. To determine whether these and other AREs are essential reguof total RNA isolated from pMV088/DTY10A (d) pMV089/DTY10A (e), and pMV090/DTY10A (f) strains. mRNAs from cultures induced by galactose and repressed by glucose were analyzed as described in Figs. 4 and 5.
FIG. 7. Decay of OLE1 mRNAs containing deletions in regions that flank the TM encoding sequences. a-c, schematic diagrams of vectors and corresponding decay patterns of mRNAs expressed from vectors pMV088, pMV089, and pMV090 derived from plasmid pGAL-OLE-207 (Fig. 5A). A stop codon was introduced at position ϩ323 before TM loop 1 in all three test vectors. a, plasmid pMV088 contains a deletion within the coding sequence immediately after the ATG start codon to the beginning of the transmembrane loop 1 (ϩ301 bp). b, plasmid pMV089 has the OLE1-coding sequence between the transmembrane loops (from ϩ475 to ϩ768 bp) removed. c, plasmid pMV090 has the coding sequence from transmembrane loop 2 (ϩ910 bp) to PacI site (ϩ1350 bp) removed. Shown are phosphor images of Northern blots FIG. 8. Linkage of the essential protein encoding elements within the 5-and 3-UTR does not produce a stable or regulated transcript. Phosphor images of Northern blots of total RNA isolated from wild type strain DTY10A containing a plasmid expressing a chimeric gene containing the OLE1 5Ј-UTR, the essential TM1, TM2, and cytochrome b 5 elements, and the OLE1 3Ј-UTR. Transcription of the chimera was induced by galactose and repressed by glucose as described in Fig. 4. latory sequences, we replaced 1 kb of the OLE1 sequence after the stop codon with 1 kb of PGK1 3Ј-UTR sequences. We have previously confirmed that the PGK1 transcript exhibits a halflife of 45 min and is not regulated by unsaturated fatty acids (5). We were surprised to find that the PGK1 3Ј-UTR did not increase the stability of the chimeric mRNA. The basal stability of the recombinant transcript was determined by the upstream OLE1 sequences, yielding a normal half-life of 13 min in fatty acid-free medium (Fig. 10). Furthermore, the chimeric transcript exhibited strong fatty acid-mediated regulation that was only partially attenuated by the substituted 3Ј-UTR, resulting in a 6-min half-life when cells are exposed to 18:2.

DISCUSSION
The short basal half-life of the OLE1 transcript and its fatty acid-regulated instability indicates that mRNA degradation is an important component of OLE1 gene expression (5). Although this system has common features with the terminal steps of the general decay pathway, its basal and regulated stability appear to be controlled by a system that differs significantly from previously described systems. The dramatic increase in the stability of the OLE1 transcript in dcp1⌬ and xrn1⌬ mutants (5) shows that under basal and regulated conditions degradation involves the removal of the 5Ј cap, which is then followed by 5Ј-3Ј digestion by the Xrn1p exonuclease. Although the Ͼ10-fold in increase of the OLE1 regulated halflife in xrn1⌬ cells indicates that Xrn1p digestion is the primary mode of OLE1 degradation under normal growth conditions, it appears that other nucleases also contribute to regulated OLE1 decay. This is supported by the observation that the OLE1 transcript decays about twice as fast in 18:2-treated xrn1⌬ cells as it does in xrn1⌬ cells growing in fatty acid-free medium. One candidate for this alternative degradation pathway might be the recently described yeast exosome, a complex of 3Ј-5Ј exoribonucleases (19,(33)(34)(35).
To understand how the half-life of OLE1 mRNA is determined, we attempted to identify the cis elements that control its stability. We previously reported that the OLE1 5Ј-UTR is necessary, but apparently not sufficient for the fatty acidmediated mRNA stability regulation (5).
This study revealed that three elements within the proteincoding region of the transcript are also required for the maintenance of OLE1 basal stability, which agrees with findings by other laboratories that multiple elements are typically involved in the modulation of mRNA decay rates (31,36). Deletion or substitution of any of the three sequences makes the OLE1 transcript highly unstable, indicating that they act together to establish its basal stability. Furthermore, the unresponsiveness to unsaturated fatty acids of the transcripts that are modified in those regions suggests that these elements play a role in fatty acid-mediated regulation of mRNA stability.
Two stabilizing elements that consist of relatively long sequences have recently been identified in yeast mRNAs (23,37). A 65-nucleotide sequence from the coding region of the PGK1 gene promotes stabilization of a chimeric transcript containing the unstable MFA2 mRNA, but a larger region that contains the same element is required to partially stabilize a chimera containing the unstable STE3 gene (37). Similarly, deletion of a 68-nucleotide sequence located between micro-ORFs within the GCN4 mRNA makes the transcript susceptible to non- FIG. 9. Test of plasmids carrying mutant OLE1 genes with pre-mature stop codons. a, plasmids pMV071, pMV061, and pAM165 were transformed into ole1 HpaI⌬::LEU2, DTY-10A strain. The transformants were rescued on leucine, uracil drop-out SD plates supplemented with 16:1/18:1 fatty acids. These were then re-streaked onto leucine, uracil drop-out galactose plates with and without 1 mM 18:2 to test for desaturase function. b-d, mRNA decay patterns of mutant transcripts pMV071 (stop codon before TM1), pMV061 (mutant start codon), and pMV165 (stop codon within cytochrome b 5 region). sense-mediated decay. That element is the apparent binding site of the RNA-binding protein Pub1p (22). Neither the PGK1 nor the GCN4 element has significant homologies to the stabilizing sequences in the OLE1 protein-coding region, which suggests that the essential OLE1 sequences represent a different class of mRNA elements.
Previous studies show that for many mRNAs there is a strong linkage between translation and mRNA turnover (11,17,38). For example, translation of ϳ55% of the PGK1 mRNA is required for its stabilizing elements to be functional (37). Other studies of the PGK1 mRNA show that a critical step in the triggering degradation involves competition between the cap binding complex, which promotes translation initiation, and the decapping enzyme Dcp1p (17,18). The rate of decapping and degradation of that transcript is highly sensitive to its translation efficiency (18,20), and changes in the sequence context of the PGK1 start codon that reduce its rate of translation initiation dramatically destabilize the mRNA.
The studies presented here, however, indicate that the halflife of the OLE1 mRNA is remarkably insensitive to its translation. It is resistant to changes in the context of the native start codon to nonsense-mediated decay and is not destabilized by the insertion of stop codons placed throughout the proteincoding region. These results further demonstrate that the OLE1 mRNA stability determinants have novel features that differ from those identified with other yeast mRNAs.
Another type of translation-dependent decay mechanism involves the autoregulation of mammalian ␤-tubulin mRNA stability. That system involves a mechanism in which excess free ␤-tubulin subunits bind to the N-terminal residues of the nascent polypeptide as it emerges from the ribosome, destabilizing the mRNA (30). Although the exact molecular mechanism that triggers the rapid decay of the transcript is not known, an important feature of the regulation system is that the N terminus of the partially translated protein acts as a sensor that can trigger the degradation of the ribosome-bound mRNA.
The identification of essential stability elements in the OLE1 transmembrane loop regions suggested that the nascent Ole1p polypeptide sequences might play a similar role in regulating mRNA stability. During translation of its membrane-spanning regions, OLE1 polysomes must remain tightly associated with the endoplasmic reticulum membrane surface, and the association of the nascent Ole1p polypeptides with the lipid bilayer could offer a novel mechanism for controlling its stability through some type of membrane fluidity regulation mechanism.
The lack of response to stop codons placed throughout the OLE1 transcript, however, indicates that its degradation is not dependent on translation of the native Ole1p polypeptide and that translation of the essential stability elements is not essential for basal and regulated decay. These findings argue that the RNA sequences in the essential regulatory regions rather than the translated protein is the molecular basis of control over the half-life of the OLE1 transcript. These apparently serve as recognition sites for RNA-binding proteins that stabilize the transcript in the absence of fatty acids and promote rapid decapping and Xrn1p exonuclease digestion when cells are exposed to fatty acids.
A most surprising finding is that the OLE1 3Ј-UTR has little effect on the basal or regulated stability of the transcript. Most studies of eukaryotic mRNA stability indicate that elements in the 3Ј-UTR of transcripts are primary determinants of mRNA half-lives. A number of transcripts in yeast and mammalian cells have been shown to be tightly regulated through AU-rich elements that are present in their 3Ј-UTRs, and these both yeast and mammalian AU-rich elements promote deadenyla-tion-dependent decapping in yeast systems (14). The OLE1 3Ј-UTR contains a number of AU-rich sequence elements consisting of five pentameric (AUUUA) and two heptameric (UAUUUAU) consensus sequences within the UTR region. Thus, it was surprising to find that replacement of the OLE1 3Ј-UTR with the 3Ј-UTR of the highly stable PGK1 gene did not alter the basal stability of the transcript nor did it abolish the fatty acid-regulated decay.
This study describes the identification of multiple cis elements that are required for the unsaturated fatty acid-mediated regulation of OLE1 mRNA stability and demonstrates that this regulation system has novel features compared with those shown for other yeast mRNAs. Besides OLE1, only the yeast SDH2 (12) and TIF51A (14) genes have been shown to exhibit instability that is affected by changes in nutritional or physiological conditions. The Ole1p fatty acid desaturase plays a central role in the formation of membrane lipids and the control of the physical properties of membrane lipid bilayers, and it is regulated by a complex system of controls at both the levels of transcription and mRNA stability. Given the importance of eukaryotic fatty acid desaturases to the maintenance and assembly of cellular membranes as well as in the production of fatty acid derived signaling molecules, it is not surprising that multiple systems have evolved for the control of these enzyme systems. The identification and characterization of proteins that bind to the OLE1 elements and the determination of the role that their structure and context play in the regulation system should provide important insights into how this system contributes to the control of eukaryotic lipid metabolism.