|Home | About | Journals | Submit | Contact Us | Français|
The nonsense-mediated mRNA decay (NMD) pathway has historically been thought of as an RNA surveillance system that degrades mRNAs with premature translation termination codons, but the NMD pathway of Saccharomyces cerevisiae has a second role regulating the decay of some wild-type mRNAs. In S. cerevisiae, a significant number of wild-type mRNAs are affected when NMD is inactivated. These mRNAs are either wild-type NMD substrates or mRNAs whose abundance increases as an indirect consequence of NMD. A current challenge is to sort the mRNAs that accumulate when NMD is inactivated into direct and indirect targets. We have developed a bioinformatics-based approach to address this challenge. Our approach involves using existing genomic and function databases to identify transcription factors whose mRNAs are elevated in NMD-deficient cells and the genes that they regulate. Using this strategy, we have investigated a coregulated set of genes. We have shown that NMD regulates accumulation of ADR1 and GAL4 mRNAs, which encode transcription activators, and that Adr1 is probably a transcription activator of ATS1. This regulation is physiologically significant because overexpression of ADR1 causes a respiratory defect that mimics the defect seen in strains with an inactive NMD pathway. This strategy is significant because it allows us to classify the genes regulated by NMD into functionally related sets, an important step toward understanding the role NMD plays in the normal functioning of yeast cells.
Nonsense-mediated mRNA decay (NMD) is a highly conserved mRNA degradation pathway. Historically, NMD has been thought of as an RNA surveillance system whose role is to identify and rid cells of mRNA with premature termination codons and thus prevents accumulation of potentially harmful truncated proteins. However, more recently, it has become apparent that the NMD pathway of Saccharomyces cerevisiae has a second role regulating the decay of wild-type mRNAs.
Genome-wide transcription profiling has revealed that a significant number (estimated to be between 5 and 10%) of wild-type transcripts accumulate in yeast cells when the NMD pathway is inactivated (15, 28). These mRNAs that accumulate can be direct NMD substrates or could accumulate as an indirect consequence of inactivation of NMD. PPR1 and URA3 mRNAs represent examples of mRNAs that are directly and indirectly affected by inactivation of the NMD pathway, respectively. PPR1 mRNA is an NMD substrate because it is degraded more rapidly in cells with an active NMD pathway than those in which the NMD pathway has been inactivated (20). It encodes a transcription activator, and the genes activated by Ppr1 are up-regulated in cells with an inactive NMD pathway (18, 27). For example, URA3 is regulated by Ppr1. URA3 mRNA accumulates in cells with an inactive NMD pathway; however, URA3 mRNA has the same half-life in cells with active and inactive NMD pathways (27). Thus, accumulation of URA3 mRNA is due to increased transcription activation by Ppr1 as an indirect consequence of inactivation of the NMD pathway. To date, a limited number of natural NMD substrates have been identified. In addition to PPR1 mRNA, 12 wild-type mRNAs that are degraded by the NMD pathway have been identified (15, 20, 37, 41). Given the number of mRNAs that are affected by inactivation of the NMD pathway, it is very likely that additional wild-type mRNAs are direct NMD substrates.
The number of direct versus indirect NMD substrates in S. cerevisiae is controversial. Lelivelt and Culbertson (28) measured the half-lives of nine mRNAs whose abundance was increased in NMD mutants. None of these mRNAs has an altered half-life. This suggests that the majority of mRNAs that accumulate in NMD mutants may be indirect targets. He et al. (15) argue, on the other hand, that the majority of mRNAs that accumulate when NMD is inactivated are direct substrates. To resolve this controversy, it is important to identify direct versus indirect NMD substrates. This could be done by determining the mRNA half-lives of all of the potential NMD substrates in wild-type and NMD-deficient cells by using microarrays. However, this approach will miss low-abundance mRNAs, like PPR1 mRNA, that are below the threshold of detection (28). We have developed a complementary bioinformatics-based approach. Our approach involves using existing genomic and function databases to identify transcription factors whose mRNAs are elevated in NMD-deficient cells and the genes that they regulate. Using this strategy, we have investigated a coregulated set of genes. We have shown that NMD regulates accumulation of ADR1 mRNA, which encodes a transcription activator of genes for generation of acetyl coenzyme A (CoA) and NADH from nonfermentable substrates (42). Further, we propose that Adr1 also activates expression of ATS1.
The S. cerevisiae strains used in this study are listed in Table Table1.1. Unless otherwise stated, yeast strains were grown and maintained by standard methods (4). Adr1 was repressed in YP medium with 8% glucose (YP-8% glucose) and derepressed in YP medium containing 3% ethanol and 1% d-glucose (YP-ethanol). Adr1p repression and derepression media were made and used as described by Sloan et al. (39). YP medium with 2% galactose (YP-2% galactose) was used for induction of Gal4. YP-2% galactose medium was prepared according to a standard protocol (4). The plasmids for ADR1 overexpression, pRS314ADR1 (ADR1 TRP1 CEN6 ARSH4) and pMW5 (ADR1 TRP1 2μ), were generously provided by Elton T. Young. The plasmid for expression of ADR1 from a GAL10 promoter, pKD34 ( PGAL10-ADR1-TCYC1 TRP1 2μ), was generously provided by Kenneth Dombek. Plasmids were transformed into yeast strains by the high-efficiency Li acetate method (11).
mRNA steady-state levels and half-lives were measured as described by Kebaara et al. (19), with the exception of ADR1 mRNA expressed from the GAL10 promoter. The half-life of this mRNA was measured by first growing the cells in YP-2% galactose to activate transcription of the GAL promoter and then repressing the GAL promoter in YP-2% glucose as described by Parker et al. (34). The PCR primer pairs used for synthesis of DNA used for probe synthesis are in Table Table2.2. Northern blots were PhosphorImaged with a Storm PhosphorImager (Amersham Pharmacia Biotech, Inc., Piscataway, NJ). Analysis of mRNA levels was performed with ImageQuant Software, version 5.1, (Amersham Pharmacia Biotech, Inc., Piscataway, NJ) and Microsoft Excel (Microsoft Corporation, Redmond, WA). mRNA levels were normalized with ScR1 RNA, an RNA polymerase III transcript that is insensitive to NMD (29). mRNA half-lives were determined from graphs of percent mRNA remaining versus time. The graphs were prepared with SigmaPlot software, version 6.10 (SPSS Science, Chicago, IL). Steady-state mRNA levels and mRNA half-lives were averaged from a minimum of three independent experiments.
Identification of potential transcription factor binding sites in the ATS1 promoter region was performed with the Promoter Database of Saccharomyces cerevisiae (http://rulai.cshl.edu/SCPD). Genomic DNA sequences for Saccharomyces paradoxus, Saccharomyces mikatae, and Saccharomyces bayanus were obtained from the Saccharomyces Genome Sequencing at the Genome Sequencing Center (http://genomeold.wustl.edu/projects/yeast/). BLASTN was used to find the 5′ end of the ATS1 open reading frame (ORF) sequences in the S. paradoxus, S. mikatae, and S. bayanus genomic sequences. Five hundred base pairs of sequence upstream of the ATS1 ORF was recovered from all of the Saccharomyces yeast strains and aligned with CLUSTALW (http://www.ebi.ac.uk/clustalw/). The average n-fold increase for various mRNAs was obtained from the Nonsense-Mediated mRNA Decay database (http://184.108.40.206/default.htm) (28). A two-sided test was used to calculate P values for the all of data in the Lelivelt and Culbertson database with the null hypothesis that the mean transcript abundances of a particular gene are equal in the knockout and wild-type samples and the alternative hypothesis that the transcript abundances are not equal.
ATS1 mRNA is one of the several hundred mRNAs that accumulate in S. cerevisiae when the NMD pathway is inactivated (28). Northern blot analysis confirmed that ATS1 mRNA levels are elevated in upf1Δ cells relative to an isogenic UPF1 strain (Fig. (Fig.1A).1A). ATS1 mRNA accumulation is 1.9-fold ± 0.3-fold higher in upf1Δ cells than in isogenic UPF1 cells. No hybridization to RNA isolated from an ATS1Δ strain was seen. This strain lacks the ATS1 gene and is thus unable to synthesize ATS1 mRNA. This confirms that the correct mRNA was detected on the Northern blots. We conclude that NMD affects ATS1 mRNA accumulation.
The NMD pathway can affect steady-state mRNA levels either directly or indirectly. Directly affected mRNAs are degraded by the NMD pathway. These mRNAs have a shorter half-life in yeast cells with a functional NMD pathway than in yeast cells with mutations in the UPF genes. Indirect effects on mRNA accumulation result when transcription of the gene encoding the mRNA is regulated by the product of another mRNA that is affected by the NMD pathway. Indirectly affected mRNAs have the same half-lives in yeast cells with a functional NMD pathway as yeast cells with mutations in the UPF genes. To determine if ATS1 mRNA is a direct or indirect target of NMD, we compared ATS1 mRNA half-lives in isogenic UPF1 and upf1Δ yeast strains (Fig. (Fig.1B).1B). The ATS1 mRNA half-life is 10.4 ± 2.1 min in UPF1 cells and 11.6 ± 1.9 min in upf1Δ cells. ATS1 mRNA half-lives are not significantly different in UPF1 and upf1Δ strains. Therefore, ATS1 mRNA is not a direct substrate of NMD. Rather, it accumulates as an indirect consequence of inactivation of the NMD pathway.
Because ATS1 mRNA is not degraded by the yeast NMD pathway, we hypothesized that NMD affects the accumulation of an mRNA encoding an ATS1 transcription regulator. We developed a strategy to identify ATS1 transcription regulators whose mRNAs are elevated in upf mutants relative to UPF yeast cells. Our overall strategy was to (i) map putative transcription factor binding sites in the promoter region of ATS1, (ii) determine whether the putative transcription factor binding sites are conserved in closely related Saccharomyces yeast strains, (iii) identify putative transcription factors of ATS1 whose mRNAs accumulate in NMD mutants, (iv) test whether ATS1 is regulated by the candidate transcription factors, and (v) determine if the transcription factor mRNA is directly or indirectly affected by NMD.
Putative transcription factors of ATS1 were identified by analyzing the 500 bp upstream of the ATS1 ORF for putative transcription factor binding sites with the Promoter Database of Saccharomyces cerevisiae (http://rulai.cshl.edu/SCPD). There are 13 putative binding sites for nine transcription factors in the ATS1 promoter region (Fig. (Fig.2A2A).
Potential transcription factor binding sites can also be identified by alignment of orthologous promoter regions from closely related Saccharomyces strains (8, 21). We aligned the sequences of the ATS1 promoter regions from S. cerevisiae, S. paradoxus, S. mikatae, and S. bayanus with ClustalW (http://www.ebi.ac.uk/clustalw/). The binding sites for four transcription factors, Adr1, Gcr1, SCB, and Gal4, are conserved in the ATS1 promoter (Fig. (Fig.2B).2B). Thus, ATS1 may be regulated by Adr1, Gcr1, SCB, and/or Gal4. Additional conserved sequences were observed, suggesting that ATS1 has additional transcription factor binding sites. This result is not surprising since the binding sites for only ~40% of the S. cerevisiae transcription factors are known (26).
To determine whether Adr1, Gcr1, SCB, or Gal4 might be regulated by NMD, we identified their genes and then found the average n-fold increase calculated from high-density oligonucleotide arrays for the corresponding mRNAs (Table (Table3;3; 28; http://220.127.116.11/default.htm). The average n-fold increase is a measure of mRNA levels in upf mutant strains relative to UPF yeast strains. The average GAL4 mRNA increase was 3.51-fold, indicating that this mRNA was elevated in NMD-deficient strains. The average increase in ADR1 mRNA was 1.20-fold, suggesting that this mRNA might be slightly elevated in NMD-deficient cells (Table (Table3).3). The GCR1 mRNA and the mRNAs encoding the subunits of SCB do not seem to accumulate in NMD mutants (Table (Table3).3). Further, the mRNA levels of the genes regulated by Gcr1 and SCB are similar in wild-type and NMD-deficient cells (data not shown). Gcr1 and SCB were not examined further.
Steady-state ADR1 and GAL4 mRNA levels in upf1Δ and UPF1 yeast strains were determined by Northern blot analysis (Fig. (Fig.3).3). Both GAL4 and ADR1 mRNAs accumulate to higher levels in upf1Δ cells than in UPF1 cells. Two GAL4-specific bands of 2.9 and 1.8 kb were observed on the Northern blots (Fig. (Fig.3A).3A). These bands are specific for GAL4 because they were not present in the lane containing RNA from a gal4 deletion strain. The expected size of the GAL4 mRNA is 2.8 kb (25). Thus, the upper band is the expected size of the GAL4 mRNA and the lower band is a truncated GAL4 transcript. The lower band was not characterized further. The full-length GAL4 mRNA accumulation was 2.9-fold ± 0.2-fold higher in upf1Δ cells than in UPF1 cells. ADR1 mRNA also accumulated to higher levels in upf1Δ cells than in UPF1 cells (Fig. (Fig.3B).3B). We found a 2.6-fold ± 0.2-fold higher level of ADR1 mRNA accumulation in upf1Δ cells compared to UPF1 cells (Fig. (Fig.3B).3B). Thus, transcription activation of ATS1 by Gal4p and/or Adr1p could account for the increased ATS1 mRNA accumulation in upf1Δ cells. For this reason, we tested whether Gal4p and Adr1p regulate ATS1.
Gal4 is a transcription activator for genes controlling the metabolism of galactose and galactose disaccharides such as lactose (reviewed in reference 6). Gal4-dependent transcription of these genes is activated by galactose and strongly repressed by glucose. Thus, we expect the mRNA levels for Gal4-regulated genes to be higher in YP-2% galactose-grown cells than in YP-2% glucose-grown cells. To test if Gal4 controls ATS1 transcription, the effects of changes in Gal4 expression on ATS1 mRNA levels were compared (Fig. (Fig.4).4). We determined steady-state ATS1 mRNA levels in UPF1 GAL4 and upf1Δ GAL4 yeast cells grown in YP-2% galactose and YP-2% glucose. Steady-state ATS1 mRNA levels were 1.0-fold ± 0.0-fold and 0.4-fold ± 0.0-fold in UPF1 GAL4 yeast cells grown in glucose- and galactose-containing media, respectively. ATS1 mRNA levels were 2.3-fold ± 0.4-fold higher in upf1Δ cells than in UPF1 cells grown in glucose and 1.9-fold ± 0.2-fold higher in upf1Δ cells than in UPF1 cells grown in galactose (Fig. (Fig.4).4). As a control, we examined GAL1 mRNA levels in cells grown in YP-2% galactose and YP-2% glucose. GAL1 is a Gal4-regulated gene (6). As expected, GAL1 mRNA was undetectable in cells grown in YP-2% glucose and readily detectable in YP-2% galactose. Thus, we conclude that ATS1 mRNA does not increase under conditions that activate Gal4.
We also tested whether ATS1 mRNA levels were affected by loss of Gal4 function by examining ATS1 mRNA levels in GAL4 and gal4Δ yeast strains. ATS1 mRNA accumulation was 1.0-fold ± 0.0-fold and 0.7-fold ± 0.0-fold in GAL4 and gal4Δ yeast strains, respectively (Fig. (Fig.4).4). Thus, loss of GAL4 function causes a small decrease in ATS1 expression.
Based on these results, we conclude that Gal4 is probably not a transcription activator of ATS1 because ATS1 expression is not activated under conditions that activate Gal4 and loss of Gal4 results in only a small decrease in ATS1 expression. Thus, the NMD-dependent increase in GAL4 mRNA does not account for the accumulation of ATS1 mRNA in upf1Δ cells. We have not examined the basis of the Upf1p-dependent increase in GAL4 mRNA further.
The potential regulation of ATS1 by Adr1p was tested in two ways. First, steady-state ATS1 mRNA levels were determined for strains differing only in their ADR1 gene copy numbers (Fig. (Fig.5A).5A). Second, steady-state ATS1 mRNA levels were determined in yeast strains grown under conditions that repress and derepress Adr1, respectively (Fig. (Fig.5B5B).
If Adr1 regulates ATS1, we expect ATS1 mRNA levels to increase in cells overexpressing ADR1 and decrease in adr1Δ cells because they lack Adr1. The ADR1 gene copy number was increased by transforming W303a (ADR1) with an ADR1 centromeric plasmid (Fig. (Fig.5A).5A). As a control, ADR1 mRNA accumulation was determined. ADR1 mRNA accumulation is 4.1-fold ± 0.1-fold higher in cells with additional copies of the ADR1 gene on a centromeric plasmid than in an isogenic ADR1 yeast strain which only expressed ADR1 from its normal chromosomal location (Fig. (Fig.5A).5A). Thus, we see an increase in ADR1 expression when the ADR1 gene copy number increases. ATS1 mRNA accumulation is 4.0-fold ± 1.5-fold higher in cells transformed with the ADR1 gene on a centromeric plasmid than in an isogenic yeast strain expressing only the chromosomal copy of ADR1 (Fig. (Fig.5A).5A). The effect of loss of Adr1 function on steady-state ATS1 mRNA levels were determined by measuring ATS1 mRNA levels in isogenic ADR1 and adr1Δ cells grown under Adr1-derepressing conditions (YP-3% ethanol-1% d-glucose; Fig. Fig.5B).5B). The relative ATS1 mRNA abundance was 0.65-fold ± 0.05-fold lower in adr1Δ cells than in the isogenic ADR1 cells.
If Adr1 regulates ATS1, we expect ATS1 mRNA levels to be higher under conditions that derepress Adr1 and lower under conditions that repress Adr1. Sloan et al. (39) showed that Adr1 is derepressed in cells in ethanol (YP-3% ethanol-1% d-glucose) and is repressed in cells grown in high glucose (YP-8% glucose; note that standard yeast growth medium contains 2% glucose). As a control for repression and derepression of Adr1, we examined ADH2 mRNA accumulation in cells grown under derepressing and repression conditions, respectively (Fig. (Fig.5B).5B). Adr1 positively regulates ADH2 by binding its promoter (10). ADH2 mRNA levels were difficult to detect in RNA prepared from cells grown under repressing conditions and readily detectable in RNA prepared from cells grown under derepressing conditions (Fig. (Fig.5B).5B). The accumulation of ATS1 mRNA in ADR1 cells grown under derepressing and repressing conditions was measured by quantitative Northern blot analysis (Fig. (Fig.5B).5B). ATS1 mRNA accumulation is 6.8-fold ± 2.0-fold higher in ADR1 cells grown under derepressing conditions than in ADR1 cells grown under repressing conditions.
Thus, ATS1 mRNA levels increase when ADR1 expression increases and when Adr1 is derepressed. ATS1 mRNA levels decrease when Adr1 is absent or when it is repressed. These results are consistent with Adr1 being a transcription activator of ATS1.
Adr1 has been shown to bind the promoters of 14 genes (42, 43). We examined the average n-fold increase calculated from high-density oligonucleotide arrays for the corresponding mRNAs (Table (Table4;4; 28, http://18.104.22.168/default.htm). Eight of the mRNAs have an average increase of 1.3-fold or greater, suggesting that these mRNA might be slightly elevated in NMD-deficient cells (Table (Table4).4). Four genes have an average increase of equal to or less than 1.15-fold. The average n-fold increase was not available for two of the genes. We confirmed that the mRNA for one Adr1-regulated gene, CTA1, accumulates by quantitative Northern blotting. CTA1 mRNA accumulation is 2.9-fold ± 0.5-fold higher in upf1Δ cells compared to that in UPF1 cells.
Initiation of nonsense mRNA decay depends on Upf1p, Upf2p, and Upf3p. Single, double, and triple deletions of the UPF genes have essentially identical effects on nonsense mRNA accumulation (3, 13). ADR1 mRNA accumulation depends on Upf1p (Fig. (Fig.3).3). We tested whether ADR1 mRNA accumulation also depends on Upf2p and Upf3p by measuring steady-state ADR1 mRNA levels in UPF, upf1Δ, upf2Δ, and upf3Δ yeast strains (Fig. (Fig.6A).6A). The steady-state ADR1 mRNA levels were 2.1-fold ± 0.1-fold and 1.8-fold ± 0.4-fold higher in upf2Δ and upf3Δ yeast cells relative to those in UPF yeast cells. The increase in steady-state ADR1 mRNA levels observed in upf2Δ and upf3Δ yeast cells is similar to the increase in steady-state ADR1 mRNA levels observed in upf1Δ yeast cells (2.1 ± 0.4). Thus, ADR1 mRNA accumulation is dependent on Upf2p and Upf3p, as well as Upf1p.
Nonsense mRNAs are degraded by deadenylation-independent decapping, followed by 5′→3′ decay (32). Dcp1p and Xrn1p are required for decapping and 5′→3′ decay, respectively (5, 32). To determine if decapping and 5′→3′ decay are also required for ADR1 mRNA decay, we examined steady-state ADR1 mRNA levels in isogenic wild-type, dcp1Δ, and xrn1Δ yeast strains (Fig. (Fig.6B).6B). Steady-state ADR1 mRNA levels were 5.2-fold ± 2.4-fold and 4.9-fold ± 2.9-fold higher in dcp1Δ and xrn1Δ cells, respectively, relative to wild-type cells. Thus, ADR1 mRNA accumulation also depends on these same decay activities because ADR1 mRNA accumulates in a decapping mutant and a 5′→3′ exoribonuclease mutant.
ADR1 mRNA has two features that could target this mRNA for NMD. First, the ADR1 start codon is located in a suboptimal context for initiation of translation and it is followed by an out-of-frame AUG in the optimal context at +83 with respect to the first base of the ORF. Second, ADR1 mRNA has an unusually long 3′ untranslated region (UTR) (420, 590, and 810 nucleotides; 7). A suboptimal start codon context predisposes an mRNA for leaky scanning of the ribosome past the translation initiation codon. NMD is then triggered when termination occurs following initiation of translation at a downstream, out-of-frame AUG (41). The S. cerevisiae optimal start codon context is ANNAUGPuPuPu, where N is any base and Pu is an A or a G. The ADR1 start codon context is ACUAUGGCT. Further, the downstream out-of-frame AUG at +83 in the optimal context for initiation of translation. The average UTR length (5′ plus 3′ UTRs) of yeast mRNAs is 256 nucleotides (16). mRNAs with unusually long 3′ UTRs are substrates for NMD (2, 33). Furthermore, the ADR1 3′ UTR contains at least three potential downstream sequence elements (DSEs). DSEs are thought to function with premature translation termination to target mRNAs for NMD (35). Termination of translation upstream of a DSE targets an mRNA for NMD, while termination of translation downstream of a DSE does not. We hypothesized that leaky scanning and/or the unusually long ADR1 3′ UTR could make this mRNA an NMD substrate.
To test the possibility that ADR1 mRNA could be a direct target of the NMD pathway, we determined ADR1 mRNA half-lives in upf1Δ and UPF1 yeast cells. ADR1 mRNA levels were determined following inhibition of RNA polymerase II. RNA polymerase II was inhibited by shifting rpb1-1 cells to the nonpermissive temperature or with thiolutin. rpb1-1 is a temperature-sensitive allele that encodes an RNA polymerase II subunit. As a control, we examined CYH2 pre-mRNA levels following inhibition of transcription (Fig. (Fig.7B;7B; data not shown). Both treatments effectively arrested transcription, judging by the decrease in CYH2 pre-mRNA levels following arrest. Further, the CYH2 pre-mRNA levels decreased faster in the UPF1 cells than in the upf1Δ cells. This is consistent with previously published work (14).
The pattern of ADR1 mRNA levels following arrest of transcription is unusual. Initially, an increase in ADR1 mRNA levels lasting approximately 10 min in upf1Δ cells and approximately 15 min in UPF1 cells was observed. The amount of ADR1 mRNA then decreases with time. The half-lives are 42.3 ± 12.9 and 9.3 ± 4.0 min in UPF1 and upf1Δ cells, respectively. Interestingly, the pattern of ADR1 mRNA levels in these experiments was independent of the method used to arrest transcription because the pattern of ADR1 mRNA levels was the same when transcription was arrested with thiolutin (data not shown). Thus, ADR1 mRNA appears to actually be degraded faster in upf1Δ cells than in UPF1 cells following arrest of transcription by shifting rpb1-1 cells to the nonpermissive temperature or by treatment with thiolutin.
Transient inhibition of general mRNA transcription by either genetic or chemical means induces a general stress response (12). As a part of this response, the mRNA levels for a subset of heat shock genes increase. Consistent with this, Adr1 is activated by growth in ethanol, which also induces stress responses (1). To begin to examine the basis for ADR1 mRNA accumulation in upf1Δ cells, we determined the half-life of ADR1 mRNA expressed from pKD34, which carries PGAL10-ADR1-TCYC1 in upf1Δ and UPF1 yeast cells (Fig. (Fig.7C).7C). The half-lives of the ADR1 mRNA expressed from this construct are 3.8 ± 0.1 and 3.7 ± 0.1 min in upf1Δ and UPF1 yeast cells, respectively. We can eliminate leaky scanning as a mechanism for targeting ADR1 mRNA for NMD because this mRNA has the same half-life in upf1Δ and UPF1 yeast cells and it contains the ADR1 translation initiation codon in its native context. However, we cannot distinguish targeting of ADR1 mRNA for NMD by a long 3′ UTR from an NMD-dependent change in ADR1 transcription because this mRNA lacks the long ADR1 3′ UTR (PGAL10-ADR1-TCYC1 has 34 bp of sequence downstream of the ADR1 ORF, followed by the CYC1 terminator).
Adr1 regulates genes involved in aerobic oxidation of nonfermentable carbon sources, including lactate (42). UPF1, UPF2, and UPF3 are required for full respiratory competence (9). upf mutants have a respiratory impairment because they grow poorly on medium containing lactate as a nonfermentable carbon energy source at 18°C. This correspondence suggests that altered Adr1 expression could account for the poor growth of upf mutants on lactate-containing medium. We tested this possibility by plating yeast cells overexpressing ADR1 on CEN and 2μ plasmids. As shown in Fig. Fig.8,8, W303a cells transformed with pMW5 (2μ-ADR1) grow slower on complete minimal medium lacking tryptophan and containing lactate than W303a cells transformed with pRS314. The reduced growth rate of W303a(pMW5) was specific for lactate because these transformants grew at the same rate as W303a(pRS314) on complete minimal medium lacking tryptophan and containing glucose. The reduced growth rate of W303a(pMW5) was specific for overexpression of ADR1 because lactate sensitivity was not seen in Research Genetics strain 3575 (adr1Δ). Interestingly, the upf1Δ strain (AAY320) grew as well as the isogenic UPF1 strain (W303a) in these experiments.
We have used a bioinformatics-based strategy to investigate a coregulated gene set. We have shown that NMD regulates accumulation of ADR1 mRNA, which encodes a transcription activator, and that Adr1 is probably a transcription activator of ATS1. The NMD-dependent regulation of ADR1 mRNA is physiologically significant because overexpression of ADR1 causes respiratory impairment. This strategy is significant because it allows us to classify the genes regulated by NMD into functionally related sets, an important step toward understanding the role NMD plays in the normal functioning of yeast cells. Further, this is a unique way to identify genes regulated by transcription factors.
Adr1 is a transcription activator, and three lines of evidence indicate that it activates expression of ATS1. (i) ATS1 has conserved Adr1 binding sites in its promoter region (Fig. (Fig.2),2), (ii) ATS1 mRNA levels correlate with ADR1 expression levels (Fig. (Fig.5A),5A), and (iii) conditions that affect Adr1 activity have a corresponding effect on ATS1 mRNA levels (Fig. (Fig.5B).5B). Our results are consistent with a global localization analysis in which intergenic microarrays were probed with DNA from chromatin bound by Adr1. Hybridization to the intergenic region of ATS1 is 1.544-fold by Adr1-bound DNA relative to background. This binding ratio may underestimate Adr1 binding to the ATS1 promoter because the putative Adr1 binding sites in the promoter region overlap the FUN30 ORF. Thus, only the Adr1-bound chromatin fragments that extend into the intergenic region between the ATS1 and FUN30 ORFs would hybridize to the DNA on the intergenic microarray. Based on our results and the results of Tachibana et al. (40), we propose that Adr1 activates ATS1 expression and that activation accounts, at least in part, for the UPF1-dependent effect on ATS1 mRNA accumulation.
Why might Adr1 regulate ATS1? In previous studies, Adr1 was shown to bind the promoters for 14 different genes involved in the generation of acetyl-CoA and NADH from nonfermentable substrates (42, 43; Table Table4).4). We showed that Ats1p interacts with Nap1p, a cytoplasmic protein that regulates the activity of the Cdc28p/Clb2p complex (38). Based on these results, we proposed that the interaction between Ats1p and Nap1p coordinates the microtubule state with the cell cycle. Cell size changes during growth in different environments. For instance, cells grown in ethanol are larger than glucose-grown cells (24). The increase in cell size seen in ethanol-grown cells is due to a delay in the cell cycle that is partially mediated by the tyrosine kinase Swe1, a negative regulator of the Cdc28-Clb complexes (24). Since Ats1p regulates the activity of the same complex, we propose that Adr1 regulates ATS1 as part of a mechanism to coordinate the cell cycle with the metabolic status of the cell. If this is the case, we predict that changes in ATS1 expression will have an effect on cell size. Consistent with this prediction, deletion of ATS1 results in larger cells (23, 38).
Here we have shown that ADR1 mRNA accumulation is regulated by NMD (Fig. (Fig.6).6). The NMD-dependent regulation of ADR1 mRNA is physiologically relevant because overexpression of ADR1 mRNA may partially contribute to the respiratory impairment of upf mutants. We show that overexpression of ADR1 from a 2μ plasmid causes respiratory impairment (Fig. (Fig.8).8). Overexpression of ADR1 may only partially explain the respiratory impairment of upf mutants because ADR1 mRNA accumulation is 2.6-fold ± 0.2-fold in a upf1Δ strain (Fig. (Fig.3B)3B) and we did not see respiratory impairment until ADR1 was overexpressed from a 2μ plasmid where ADR1 mRNA levels were elevated 9.9-fold ± 3.8-fold (data not shown). Alternatively, deletion of the upf genes in the strains used by de Pinto et al. (9) may have a more significant impact on ADR1 expression than in the W303a background. We have previously observed strain-dependent differences in the accumulation of mRNAs degraded by the NMD pathway (19). Future experiments will focus on determining whether ADR1 regulation by NMD is direct or indirect.
Our strategy complements existing microarray analyses of the effects of inactivation of NMD on global mRNA abundance in two ways. First, it allows us to sort the mRNAs affected by NMD into physiologically relevant, coregulated gene sets. Second, it uncovers low-abundance mRNAs at or below the threshold of detection on microarrays. For example, PPR1 mRNA is near the threshold of detection on microarrays (28). Our approach is applicable to other coregulated gene sets. Our strategy is limited to transcription factors with known DNA binding sites. As binding sites are identified for additional transcription factors, we expect to be able to identify additional NMD-regulated gene sets.
Several lines of evidence suggest that a role for NMD in the regulation of decay of select wild-type mRNAs is not unique to S. cerevisiae. Upf1 is essential for mammalian embryonic viability (30). NMD-deficient mouse embryos do not develop; instead, they are resorbed shortly after implantation and NMD-deficient blastocysts isolated at 3.5 days postcoitum commit apoptosis after a brief period of growth. And in Caenorhabditis elegans, NMD deficiency causes minor morphogenic abnormalities of the genitalia and reduced brood size (36). These effects probably reflect the failure both to rid the cells of mRNAs with premature termination codons and to down regulate natural substrates. As is seen in yeast, a significant percentage (4.9%) of physiologic transcripts are up-regulated mammalian cells depleted of Upf1 or Upf2 accumulate (31). A representative subset of the up-regulated transcripts with potential structural features that could cause premature termination of translation had longer half-lives in Upf1-depleted cells. Furthermore, recently Kim et al. (22) showed that mammalian Arf mRNA decay is dependent on Upf1 and Stau1, but not Upf2 or Upf3X. Stau1 binds the 3′ UTR of Arf1 mRNA and reduces its abundance. This suggests at least that mammalian Upf1 also functions in decay of wild-type mRNAs that lack an apparent premature termination codon, a substrate reminiscent of the yeast PPR1 mRNA.
We thank Mike Lelivelt for pointing out that steady-state ATS1 mRNA levels are higher in upf1Δ yeast strains than in the isogenic UPF yeast strains. We also thank Kenneth Dombek, Alan Jacobson, Phil James, Susan Wente, and Elton T. Young for yeast strains and plasmids used in this study. We are grateful to members of the Atkin laboratory for critical reading of the manuscript, helpful comments, and discussions.
This work is based upon work supported by the National Science Foundation under grants 9874516 and 0444333.
Any opinions, findings, conclusions, or recommendations expressed in this report are ours and do not necessarily reflect the views of the National Science Foundation.