|Home | About | Journals | Submit | Contact Us | Français|
Regulation of mRNA turnover is an important cellular strategy for posttranscriptional control of gene expression, mediated by the interplay of cis-acting sequences and associated trans-acting factors. Pub1p, an ELAV-like yeast RNA-binding protein with homology to T-cell internal antigen 1 (TIA-1)/TIA-1-related protein (TIAR), is an important modulator of the decay of two known classes of mRNA. Our goal in this study was to determine the range of mRNAs whose stability is dependent on Pub1p, as well as to identify specific transcripts that directly bind to this protein. We have examined global mRNA turnover in isogenic PUB1 and pub1Δ strains through gene expression analysis and demonstrate that 573 genes exhibit a significant reduction in half-life in a pub1Δ strain. We also examine the binding specificity of Pub1p using affinity purification followed by microarray analysis to comprehensively distinguish between direct and indirect targets and find that Pub1p significantly binds to 368 cellular transcripts. Among the Pub1p-associated mRNAs, 53 transcripts encoding proteins involved in ribosomal biogenesis and cellular metabolism are selectively destabilized in the pub1Δ strain. In contrast, genes involved in transporter activity demonstrate association with Pub1p but display no measurable changes in transcript stability. Characterization of two candidate genes, SEC53 and RPS16B, demonstrate that both Pub1p-dependent regulation of stability and Pub1p binding require 3′ untranslated regions, which harbor distinct sequence motifs. These results suggest that Pub1p binds to discrete subsets of cellular transcripts and posttranscriptionally regulates their expression at multiple levels.
Gene expression in eukaryotes is a highly diverse process, involving regulation at both transcriptional and posttranscriptional levels (47, 70). The process of mRNA turnover is an important posttranscriptional control point that helps to modulate the cellular abundance of a transcript. A large number of clinically relevant transcripts exhibit regulated decay in response to cellular signals, and the deregulation of their decay rates directly correlates with disease states (58, 66, 70). In the yeast Saccharomyces cerevisiae, the principal mRNA degradation pathway initiates with the removal of the poly(A) tail, followed by either decapping and 5′→3′ exonucleolytic decay or 3′→5′ exosome-mediated degradation (10, 64). In addition to this pathway, aberrant mRNAs harboring premature termination codons are degraded by an alternate nonsense-mediated decay (NMD) pathway that functions to ensure quality control of gene expression. NMD is initiated when a premature termination codon is recognized during translation termination and stimulates rapid deadenylation-independent decapping, followed by 5′→3′degradation of the mRNA (4). Messages undergoing leaky scanning (69) or harboring upstream open reading frames (50, 59) represent some naturally occurring substrates that decay through this pathway.
The turnover of mRNAs is mediated by the interplay between a number of different cis-acting sequences localized in the target substrate and the various trans-acting factors that interact with them (70). In both yeast and in mammalian systems, there are a large number of RNA-binding proteins (RBPs) that can act as either enhancers or inhibitors of stability and translation efficiency (70). Many of these RBPs mediate their effects by selective binding to target motifs in the 3′ untranslated region (3′UTR). Examples include the UGU element for the yeast Puf proteins (32) and the AU-rich elements (AREs) interacting with the mammalian ARE-binding proteins such as HuR, Hsp70, TTP, BRF1, AUF1, and KSRP (7). In addition to the 3′UTR, cis-acting elements may also be present in the 5′UTR (9, 59) or in the coding region of transcripts (51, 62). In addition to modulating stability, many of these 5′- and 3′UTR elements can also actively regulate translation by interacting with specific factors. Examples of this type of regulation include the translational repression of the tumor necrosis factor alpha (TNF-α) mRNA by the ARE-binding protein T-cell internal antigen 1 (TIA-1)/TIA-1-related protein (TIAR) (55) or the control of ferritin mRNA translation through the iron response element-binding protein (20).
The poly(U)-binding protein (Pub1p) is a yeast homologue of the mammalian ELAV-like proteins HuR and TIA-1/TIAR (1, 43). Pub1p has been recently implicated as a regulator of cellular mRNA decay (59, 65). This abundant 51-kDa RBP containing three RNA recognition motifs (43) modulates the stability of targets that are degraded through at least two mechanistically distinct pathways. Similar to HuR, which stabilizes ARE-bearing transcripts degrading though the deadenylation-dependent pathway in mammalian cells (8, 48, 53), Pub1p can specifically bind to a chimeric yeast mRNA bearing the TNF-α-ARE and stabilize this transcript (65). Analogous to ARE-dependent regulation in mammalian systems, stabilization by Pub1p requires glucose, which activates the p38 MAP kinase pathway (65). Additionally, Pub1p has also been shown to selectively bind to a stabilizer element (STE) located in the 5′UTR of the upstream open reading frame (upstream ORF)-containing transcripts YAP1 and GCN4 and to prevent their turnover through the NMD pathway (59). These results demonstrate the Pub1p can bind to at least two classes of stability elements and modulate decay, based on cellular conditions.
Several lines of evidence also suggest that Pub1p may be involved in other aspects of mRNA metabolism. Both mammalian homologues of Pub1p, HuR and the TIA-1/TIAR, are involved in translational regulation. While HuR acts as a translational enhancer or repressor (42a), the TIA-1 and TIAR proteins are involved in ARE-mediated translational repression (55). Significantly, Pub1p has been shown to be associated with nonpolysomal mRNAs (1); moreover, it binds to the 3′UTRs of endogenous ARE-containing transcripts such as MFA2 and TIF51A but does not modulate their stability under any conditions analyzed (65; S. Vasudevan and S. W. Peltz, unpublished observations). Taken together, these results suggest that, in addition to regulating decay, Pub1p may also have a role in modulating translation efficiency. Furthermore, the TIA-1/TIAR proteins have also been implicated as activators of constitutive and alternative splicing (16, 18, 61). As Pub1p is equally abundant in both the nucleus and cytoplasm (1) and demonstrates interaction with Nab2p (31), a factor involved in mRNA processing and export (26), this additionally suggests that the protein might play nuclear roles, such as regulating the maturation and biogenesis of mRNAs.
The high cellular abundance of Pub1p and its ability to regulate the turnover of two known classes of transcripts makes it likely that Pub1p is a key trans-acting factor mediating the stability of multiple mRNAs. In addition, the glucose dependence of Pub1p function suggests that this protein has the potential to mediate large changes in gene expression in response to cellular conditions. As expression and transcript stabilities of a large fraction of genes have recently been demonstrated to be regulated by RNA-binding proteins (22) and can vary depending on cellular states such as changes in carbon source (14, 17), we hypothesize that Pub1p acts as a global regulator of gene expression engaged in coordinating the expression of a wide network of mRNAs. The goal of this work is to characterize the role of Pub1p in regulating mRNA stability and to identify classes of transcripts that associate with the protein.
To therefore distinguish classes of Pub1p-regulated transcripts, we have used cDNA arrays to query the yeast genome for potential Pub1p targets. Our results measuring global mRNA stability in wild-type and isogenic pub1Δ strains under glucose conditions show that nearly 10% of all yeast mRNAs decay in a Pub1p-dependent manner. The majority of these messages, representing 9% of all transcripts, exhibit destabilization in a pub1Δ strain. The remaining 1% of mRNAs, by contrast, are significantly stabilized. Furthermore, through an independent affinity purification approach, we demonstrate that Pub1p can directly associate with 368 cellular transcripts, which represent 6% of the genome. Classification and analysis of these mRNA populations reveal discrete subsets of transcripts whose decay is either directly or indirectly dependent on Pub1p. In addition, we also find groups of transcripts that demonstrate Pub1p binding without affecting mRNA half-lives, suggesting that Pub1p controls other aspects of their metabolism such as translation or mRNA export. Furthermore, each of these subsets of genes can be organized into distinct functional categories, suggesting a role of Pub1p in diverse biological processes. We identify putative regulatory motifs in the UTR regions of the Pub1p-associated mRNAs and demonstrate that transcripts exhibiting altered decay are specifically enriched in U-rich motifs, suggesting a role of these elements in modulating mRNA turnover. Finally, analysis of two candidate transcripts, RPS16B and SEC53, demonstrates that Pub1p-dependent regulation of stability requires sequences in their 3′UTR.
Taken together, these results implicate Pub1p as a major trans-acting factor involved in cellular mRNA decay and suggest a novel role of Pub1p in regulating other posttranscriptional processes. Our work extends the observation of Grigull et al. measuring global decay using 1,10 phenanthroline (22) by revealing that distinct groups of Pub1p targets exhibit coordinate regulation; moreover, a direct correlation between binding and stability can be established for a subset of these messages. Furthermore, we show the presence of conserved elements in the Pub1p target and set the stage for uncovering how Pub1p binding can lead to the modulation of mRNA decay rates or other posttranscriptional processes (33, 40, 71). The results from this analysis join a growing body of studies measuring genome-wide stability and RNA-protein interactions (19, 23, 27, 34, 35, 52, 56) and reiterate the extensive role RNA-binding proteins play in posttranscriptional control of gene expression.
Oligonucleotides used in this study were synthesized and purified through standard desalting by IDT (www.idtdna.com). All restriction sites are underlined, and the names of primers used for quantitative real-time PCR are indicated in italics with the gene names and corresponding letter “f” and “r” to indicate forward and reverse primers. The primers were #418 (5′TATTCTAGATACCAACCTTAATGC), #502 (5′TATAAGCTTTTTCGAAACGCAGAATT), #920 (5′GAAGATCTCTTTATTTCCAGGTTGCTACG), #921 (5′ATAAGAATGCGGCCGCACGATCTGCAAATAAAAGAAATCG), #922 (5′GAAGATCTGAAAATTTCAAGATTTTTTCCA), #923 (5′ATAAGAATGCGGCCGCAATAAAGGAAAAAAGATACG), #924 (5′CTTTAAGTTAGAGTTTAATC), #925 (5′ ATGGATTCTTGTCGTTCTTT), #926 (5′TTGTTGCCTTGG CAAACTTCCCA), #927 (5′TTCGAGCAATAGTAAATTGAGTTTC),SEC53f(5′GGCCCAAACGTTTTAGATGA), SEC53r (5′CGAGCCAGTTGATGAAGGAT), RPS24Bf (5′GGTTGAAAAGGCTTCCAGACA), RPS24Br (5′TCAGCGTTACGACGAGCAA), PAB1f (5′TTCTGTTTCCGAAGCCCACTT), PAB1r (5′CCGGCTTCATGGTCGTTAAA), RPS27Bf (5′GTCAAATGCCCAGGTTGTTTG), RPS27Br (5′TTTGGCCTTACCACCGGTT), RPL17Af (5′CGTGGTTCCTACTTGCGTGTT), RPL17Ar (5′TGGTGGTCCAAAACTTGTTCC), RPL20Af (5′TCAACGAAGCTCATCCAACCA), RPL20Ar (5′AAGGTTTCGACGGCAGCAA), RPS16Bf (5′TGGTGGTGGTCATGTTTCTCA), RPS16Br (5′CGTTCTTGGATTGTTCGTCAAC), RPL9Bf (5′CGGTCCAAGAGGTACTTTGACC), RPL9Br (5′CAACGTGCTTTCTGTCACCGT), URA5f (5′CAGCGCCACTCAAACTGTTA), URA5r (5′GAGGCACCGTAGGTTTGAAG), PGK1f (5′TTCCAGAAAGGTCGATGGTC), PGK1r (5′GGGTTCTCCAAAGCCTTACC), PCL5 (5′GCCCTTAGAGCACAACGGTA), PCL5r (5′TGAAAGAGGCCGACTGATCT), URA6f (5′CCTGACCAAGTTTCCGTGAT), URA6r (5′TGCTCAGCACGTAGAAGGTC), TIF51Af (5′GATGTCAAGGCTCCAGAAGG), TIF51Ar (5′CGGTTCTAGCAGCTTCCTTG), GCN4f (5′ATGGGTTTCTCACCATTGGA), and GCN4r (5′CGAAGGGGTATCCTGTTTGA).
Saccharomyces cerevisiae strains used in this study are Y534 (PUB1: MATα ura3-52 his4-939am rpb1-1) (68) and its isogenic derivatives Y535 (pub1Δ MATα ura3-52 his4-939am rpb1-1 PUB1::kan), Y538 (pub1Δ: MATα ura3-52 his4-939am rpb1-1 PUB1::URA3), and Y536 (PUB1-TAP MATα ura3-52 his4-939am rpb1-1 PUB1-TAP). Strain Y538 was generated by transformation with PRNP1.3 (1). For generation of the pub1Δ (Y535) and PUB-TAP (Y536) strains, a PCR-based deletion strategy was employed (5). To generate strain Y535, PCR amplification of genomic DNA from strain Y537 harboring a deletion of the PUB1 gene (ATCC 4005344) was performed using primers #924 and #925 and knockouts isolated by Geneticin selection. To generate strain Y536 harboring a C-terminal fusion of the TAP tag cassette (57) to the PUB1 gene, a PCR-amplified fragment from strain SC0375 (Euroscarf) using primer pairs #926 and #927 was used, and transformants were selected on C-URA plates. All strains carry a temperature-sensitive point mutation in subunit I of RNA polymerase II (rpb1-1) (49), allowing rapid inhibition of transcription at the nonpermissive temperature of 37°C. All strains were cultivated in C-URA minimal medium containing 2% dextrose, following standard protocols (2).
mRNA decay rates of strains Y534 (PUB1) and Y535 (pub1Δ) were measured by microarray analysis as previously described in Wang et al. (68) with the following modifications. Briefly, 900 ml of cells was grown at 24°C to an optical density at 600 nm (OD600) of ~0.6. The cells were centrifuged and reconstituted in 110 ml of C-URA medium. An equal volume of prewarmed (50°C) medium was added, and the culture was rapidly transferred to 37°C to inhibit transcription. Aliquots of cells were removed at 0, 2, 4, 6, 8, 10, 20, and 30 min after temperature shift and rapidly harvested by filtration on nitrocellulose filters (Millipore), followed by freezing in liquid nitrogen. Each experiment was repeated three times using individual colonies and analyzed through independent microarray hybridization. For Northern analysis, the protocol was essentially the same; however, only 100 ml of culture was utilized and samples were collected at 0, 3, 6, 9, 12, 18, and 24 min after transcriptional arrest. Detection and visualization of transcripts by Northern blotting was performed as described in Duttagupta et al. (15). Transcription inhibition was confirmed by probing for the unstable transcript HTB1, which is not affected by Pub1p (59); each analysis was repeated at least three times to generate an average half-life.
Yeast Y6.4k cDNA spotted arrays containing 6,218 genes printed in duplicate and the Arabidopsis thaliana chlorophyll synthetase gene as a control were purchased from University Health Networks (Toronto, Canada) and used for all hybridizations. The chips were customized by the addition of 240 spots, each containing either PCR product 1, 2, or 3 from the SpotReport Alien cDNA Array Validation system (Stratagene). All controls were equally distributed in each of the 48 subarrays of the chip. Sample preparation for hybridization was performed essentially as described in Wang et al. (68). Briefly, 15 μg of total RNA was extracted from each time point by the hot-phenol method (37), digested with DNase1 (Promega), and reverse transcribed using random hexamers in the presence of Cy5-dUTP (Perkin-Elmer) and Superscript II reverse transcriptase (Invitrogen). As a reference, genomic DNA was extracted from the Y534 strain (PUB1) and digested overnight with DpnII, and 200 ng of the digested sample was labeled with Cy3-dUTP (Perkin-Elmer). For normalization of each time point, a cocktail of internal standards was prepared by pooling in vitro-transcribed RNA corresponding to the Arabidopsis thaliana chlorophyll synthetase gene (250 pg/μl) and Spot Report Alien PCR products 1 (150 pg/μl), 2 (100 pg/μl), and 3 (500 pg/μl). A final amount of 500 pg of the internal standard mixture was added to each of the labeling reaction mixtures prior to reverse transcription. Hybridization and washing were performed as described in www.microarrays.cam, and the washed slides were scanned with the Genepix 4000A scanner from Axon Instruments (Foster City, CA). Data were collected with the GENEPIX PRO 4.0 software and analyzed with Microsoft EXCEL and the statistical package R. Spots that had a signal-to-noise ratio of <1.2 were excluded from further analysis. Each subarray was normalized to its specific controls, and the Cy5/Cy3 ratio was plotted as a function of time after transcription arrest. A nonlinear least-squares model, as described in Wang et al. (68), was used to compute half-lives; the Shapiro-Wilk normality test (11) was used to assess confidence in these values. Half-lives with P values of >0.05 were considered reliable and selected for all analysis.
To profile mRNAs bound to Pub1p, strain Y536 was grown at room temperature in 1 liter of C-URA medium to an OD600 value of ~0.8. Strain Y534 was cultured in parallel and served as a mock control for the experiment. Extract preparation and RNA isolation were performed essentially as described in Gerber et al. (19).
For hybridization, 600 ng of affinity-purified RNA and 3 μg of total RNA were labeled with Cy5 and Cy3, respectively, by indirect incorporation using the amino-allyl labeling procedure (19). In both channels, an equal amount of the normalization mixture as previously described was added to serve as a labeling control. Slides were hybridized, washed, and scanned as detailed above. Spots flagged absent by GENEPIX PRO 4.0 were excluded from further analysis. For all viable spots, the log2 (Cy5/Cy3) ratio was calculated followed by computation of the Z-score (11). Each experiment was repeated three times. To balance the selectivity and sensitivity of the experiments a median Z-score of 1.5 or more corresponding to a confidence level of 87% was chosen as a threshold to select genes that were specifically enriched by affinity purification (enriched genes). To eliminate nonspecifically bound messages, similar analysis was applied to the mock affinity-purified messages. Genes corresponding to RNAs detected in the control chip were parsed from the list of enriched genes to yield a final set of 368 targets that significantly associated with Pub1p.
For classification of Pub1p targets gene ontology (GO) annotations were downloaded from www.geneontology.org (March 2004 version). Annotations were retrieved for Pub1p targets to calculate the gene frequency and the Y6.4K genome to calculate genome frequency. Specifically, the fraction of the Y6.4K genome that had reliable half-lives (stability array) or identifiable spots (TAP-tag array) were used to calculate the genome frequency.
Two-step quantitative reverse transcription-PCR (qRT-PCR) was performed to validate the half-lives and binding of target genes as obtained from microarray experiments. The TaqMan reverse transcription kit (Applied Biosystems Group) containing either random hexamers or a 1:1 mixture of hexamers and oligo(dT) primers was used to reverse transcribe RNA from time course experiments or Pub-TAP affinity-purified messages. For each gene, specific PCR primers were designed using Primer Express 3 software (Applied Biosystems), and amplification reactions were performed using DNA Engine Opticon (MJ Research) with the SYBR Green PCR Core kit (Applied Biosystems). Results were analyzed using Opticon Version 1.4 software and data normalized by the ΔΔCt method (39). Each qRT-PCR experiment was performed three times, and the data reported represent an average. Gene-specific PCR products obtained for the affinity-purified messages were furthermore analyzed on a 1% agarose gel and stained with ethidium bromide for visualization.
All DNA manipulations were carried out by standard protocols (2, 54). Plasmid p5046 has been described previously (65). The following plasmids were constructed for this study. p5125, which contained the MFA2 coding region fused to the 273 nucleotides (nt) of PGK1 3′UTR and terminator under the control of the ADH1 promoter, was generated as follows: a fragment containing the MFA2 ORF with the PGK1 3′UTR sequence was obtained by PCR using p5042 (65) as a template and primers #418 and #502. The PCR products were digested with XbaI and HindIII, and the ends were filled by Klenow polymerase. The fragment was then ligated to plasmid p5065 (15), which had been digested with XbaI and HindIII, followed by Klenow treatment. Plasmids p5126 and p5127 containing the MFA2 ORF fused to 350 nt of putative 3′UTR and the terminator regions from RPS16B (YDL083C) and SEC53 (YFL045C) were constructed as follows. The RPS16B and SEC53 3′UTR sequences were obtained by PCR using genomic DNA from the Y534 strain as template and primer pairs #922 and #923 for RPS16B and #920 and #921 for SEC53. PCR products were digested with BglII and NotI, ligated to plasmid p5125, and digested with BglII and NotI to remove the PGK1 3′UTR and part of the vector sequence. To obtain transcription runoff products for cross-linking analysis, the RPS16B and SEC53 3′UTR sequences generated by PCR were cloned into the pGEM-T Easy vector (Promega) to yield plasmids p5124 and p5131, respectively.
Cytoplasmic yeast extracts were prepared from strains Y534 and Y535 as described in Czaplinski et al. (13). Plasmids p5008 (65), p5124, and p5131 were linearized using HindIII, NcoI, and SpeI, respectively. Linearized plasmids were transcribed using either T7 RNA polymerase (p5008 and p5131) or SP6 RNA Polymerase (p5124). UV cross-linking was performed essentially as described in Vasudevan et al. (65).
Since the exact 5′ and 3′ ends of the mRNA targets which demonstrate Pub1p binding have not been mapped, 200 nt of sequence downstream of the stop codon and 100 nt of the sequence upstream of the start codon were downloaded from the Saccharomyces Genome Database (SGD; June 2004 version) (30) and was searched for putative motifs in the sense strand. The lengths of the chosen sequences fall well within the predicted average lengths of the untranslated regions (21, 45). Sequences were downloaded from SGD, and motifs were constructed and analyzed using multiple expectation maximization for motif elicitation (MEME) and motif alignment and search tool (MAST) tools run on a local server (3). Motifs detected in at least 10% of the targets were selected for further analysis and motif logo constructed using WEBLOGO (12). To calculate the frequency of random occurrence of motifs, we used the Y6.4k genome, parsed for genes displaying poor signal quality.
Strain Y534 (PUB1) and Y538 (pub1Δ) were grown in YPD medium to mid-logarithmic phase (OD600 = 0.6) and plated for confluent growth in YPD plates. BBL 1/4-in.-diameter paper disks were placed on the center of the plates, and 10 μl of 500 mg/ml paromomycin or 800 mg/ml paromomycin was spotted onto each disk. Plates were incubated in 24°C for 48 to 72 h, and the drug sensitivity was measured by calculating the radius of inhibition of growth. Each assay was done in triplicate, and the values reported are an average of the three experiments.
The following data can be accessed as supplementary tables (see Tables S1, S2, and S3 in the supplemental material) from the ASM journal website: a complete data set of half-life values; complete lists of transcripts that are either regulated, not regulated, or bound by Pub1p; complete GO classifications of Pub1p targets exhibiting regulated stability, binding, and correlations of binding and stability; and MAST analyses of 5′- and 3′UTR motifs.
To identify the cellular repertoire of mRNAs whose stabilities are regulated by Pub1p under glucose growth conditions, we analyzed genome-wide expression profiles of PUB1 and its isogenic derivative pub1Δ through a time course analysis (Fig. (Fig.1A).1A). Both these strains harbor a temperature-sensitive mutation in subunit 1 of the RNA polymerase II gene, resulting in transcriptional arrest at the nonpermissive temperature of 37°C (49). To assess global decay rates, total RNA was isolated from both the wild-type and pub1Δ mutant strains at various times after transcription inhibition, and mRNA levels were monitored by hybridization to cDNA arrays (Fig. (Fig.1A).1A). Each time course was done in triplicate, and decay curves were generated for each independent experiment using a nonlinear least-squares model (68) (see Table S1a in the supplemental material).
The calculated global decay rates in the wild-type and mutant strains over three independent experiments demonstrated that the median half-life of transcripts in the PUB1 strain was 17.9 ± 0.2 min. This is closely correlated to the global median value of 20 min previously reported for overall decay by Wang et al. (68) and to the average global half-life of 19 min calculated by Holstege and colleagues (28). In contrast, to the half-life determined in the wild-type strain, the pub1Δ strain demonstrated a shorter median half-life of 12.7 ± 0.18 min, indicating a moderately enhanced rate of turnover. Furthermore, the similarity in strain background, experimental design, and method of data analysis between our study and that of Wang et al. (68) permits us to closely compare specific half-lives between the two analyses. Our results revealed a strong correlation with half-lives determined from Northern analysis (Pearson coefficient of 0.72) and measurements by Wang et al. (Pearson coefficient of 0.77) for a representative subset of 32 transcripts (68), which was previously used to estimate the confidence of the half-life values. Moreover, independent measurement of decay rates of one nonregulated transcript (HTB1) and two Pub1p-regulated targets, endogenous GCN4 (59) and the MFA2-TNF-α chimeric mRNA (65), demonstrates that the half-lives obtained from Northern analyses (Fig. (Fig.1B)1B) are in reasonable agreement to those determined from the array (Fig. (Fig.1C;1C; see Table S1a in the supplemental material). Figure Figure1C1C displays the decay profiles of HTB1 and GCN4 after combining data from the three independent microarray experiments. The median half-life of the MFA2-TNF-α chimeric mRNA was determined to be 20 ± 5.6 min in the wild type and 7 ± 3.3 min in the mutant (see Table S1a in the supplemental material). Thus, in accordance with previous studies (59, 65), the Pub1p target mRNAs exhibited altered stability between the wild-type and pub1Δ strains, with approximately ≥2-fold decreases in half-life in the mutant. Taken together, these results indicate that the global expression analysis of the PUB1 and the pub1Δ mutant strain can confidently estimate half-lives for a major fraction of the transcriptome and is consistent with the previously determined decay rates of several transcripts.
For comparison of half-lives between the wild-type and mutant strains, genes displaying reliable half-lives and consistent pattern of expression in at least two of three experiments were selected for further analysis. This approach allowed us to precisely identify a set of 1,547 genes whose physiological expression pattern between the wild-type and pub1Δ strain could be accurately measured. To detect mRNAs whose stabilities were differentially regulated, transcripts displaying a ≥2-fold difference in half-lives between the wild-type and pub1Δ strains were determined. Consistent with the role of Pub1p as a stabilizing RNA-binding protein, we obtained a set of 573 genes that were selectively destabilized in the pub1Δ strain compared to the wild-type strain (see Table S1b in the supplemental material). In contrast, we identified only 78 genes that were stabilized by the loss of PUB1 (see Table S1b in the supplemental material), suggesting that a major role of Pub1p in mRNA turnover is to protect transcripts from rapid decay. The remaining set of 896 genes did not display any changes in mRNA stability (Table S1b in the supplemental material), indicating that the requirement of Pub1p for regulating mRNA turnover is localized to specific sets of cellular transcripts. To therefore characterize those transcripts displaying Pub1p-mediated stabilization of mRNA half-lives, we proceeded to classify the 573 genes into biologically relevant groups by analyzing for significant shared GO terms (30, 38). Our analysis revealed that the Pub1p targets were significantly enriched in genes associated with biosynthetic activity and protein synthesis implying a biological role of the protein in these processes (Table (Table1;1; see Table S1c in the supplemental material). Significantly, we found approximately 15% of the regulated targets (78 genes) were involved in protein biosynthesis, 10% (53 genes) encoded structural components of the ribosome, and 13.5% (69 genes) participated in ribonucleoprotein complexes. Additionally, we also found genes involved in translational elongation and the regulation of translational fidelity (Table S1c in the supplemental material). We furthermore validated a subset of these enriched genes by measuring half-lives of 10 transcripts through qRT-PCR (Table (Table2).2). The half-lives of these mRNAs in the wild-type strain corresponded closely to values determined from previous studies using the rpb1-1 strain (data not shown) (28, 68). Comparison of the decay kinetics of RPS16B, RPS24B, RPS27B, RPL9B, RPL17A, RPL20A, PAB1, PCL5, URA6, and SEC53 in the wild-type and pub1Δ strains, demonstrated that all of these transcripts were significantly destabilized in the mutant strain displaying a consistent ≥2-fold difference in their half-lives. Our data therefore confirms and extends the observations of Grigull et al. (22), suggesting that regulation of mRNA decay through Pub1p has a significant effect on ribosomal biogenesis, by identifying and validating additional substrates of this pathway. Consistent with the hypothesis that Pub1p is involved in general translation, we find that the pub1Δ strain exhibits a fivefold increase in resistance to the translation error-inducing antibiotic paromomycin (Table (Table3),3), suggesting a possible role of Pub1p in modulating translation fidelity.
Interestingly, in addition to the genes involved in protein biosynthesis, we also observed enrichment for genes involved in diverse metabolic pathways as well as messages encoding pseudogenes (1 transcript) and transposable elements (10 transcripts), which showed regulated stability between wild-type and pub1Δ strains (see Table S1c in the supplemental material). Taken together, these results demonstrate that Pub1p has a function in regulating diverse biological processes, primarily protein biosynthesis, as well as aspects of cellular metabolism.
In principle, Pub1p can regulate stability either by directly binding to the RNA of interest or through an indirect effect of modulating expression of another gene. To distinguish between these two possibilities, we wanted to isolate the fraction of transcripts whose stabilities were regulated by direct binding to Pub1p. To identify mRNAs associated with Pub1p, a PUB1-TAP fusion strain (Y536) was generated by insertion of the TAP tag (57) at the C terminal of the PUB1 gene at its chromosomal location. This strategy enabled the purification of a fusion protein while maintaining normal expression under its endogenous promoter (Fig. (Fig.2A).2A). To control for nonspecifically enriched mRNAs, the untagged wild-type strain (Y534) was used for mock purification. Extracts were prepared from both strains and subjected to RNA purification and hybridization protocols (Fig. (Fig.2B),2B), as described by Gerber et al. (19).
A total of 368 targets were found to be selectively enriched by this procedure with highly reproducible Z-score values of >1.5 and median standard deviation of 0.18 (Fig. (Fig.2C;2C; see Table S2a in the supplemental material). For positive internal controls, we were able to detect signal from GCN4 (Z-score = 1.8 ± 0.18) (Fig. (Fig.2C),2C), which has been previously demonstrated to bind to Pub1p through a STE sequence located in the 5′UTR (59). Additionally, we also detected strong hybridization to TIF51A (Z-score = 2.2 ± 0.2) (Fig. (Fig.2C),2C), coding for the translation initiation factor eIF5A, which is known to bind Pub1p but is not regulated at the level of mRNA stability under conditions tested (65; Vasudevan and Peltz, unpublished). Low intensity from the MFA2-TNF-α mRNA, which was encoded by plasmid p5046 (65), precluded the spot from any further analysis. Classification of these targets by gene ontology revealed functional themes similar to those observed with the 573 stability targets of Pub1p (see Table S2b in the supplemental material). Specifically, genes involved in protein biosynthesis and cellular metabolic pathways were highly enriched in the set of Pub1p-associated transcripts, indicating that these processes are controlled by direct interaction of Pub1p with the target transcripts. Taken together, our data demonstrate that 368 cellular targets representing about 6% of the genome bind to Pub1p specifically (Fig. (Fig.2C).2C). In addition, a majority of Pub1p targets can be categorized into biological pathways functionally similar to those of messages exhibiting altered stability, suggesting Pub1p regulates targets of these pathways through direct interaction.
To distinguish the fraction of Pub1p targets which demonstrated direct binding to Pub1p and in parallel exhibited differential stability between the wild-type and pub1Δ strain, we set out to map the overlap between the 573 regulated genes (see Table S1b in the supplemental material) and the set of 368 transcripts that bind directly to Pub1p (Table S2a in the supplemental material). Among the RNAs that bound directly to Pub1p, stability values could be reliably estimated for a set of 53 genes that displayed a ≥2-fold destabilization in the mutant strain compared to the wild-type. In contrast, only one gene (YPR128C) exhibited stabilization in the pub1Δ strain. As predicted from the similarity in GO categories (see Tables S1c and S2b in the supplemental material), all the 53 genes that were destabilized in the pub1Δ strain and demonstrated direct binding to Pub1p were involved in cellular metabolism and protein biosynthesis (Table (Table4;4; Table S2c in the supplemental material). From this analysis, we conclude that out of the global set of 573 mRNAs displaying decreased stability, Pub1p directly binds to 53 transcripts (designated class A targets). The remaining 520 genes are predicted to represent indirect targets of Pub1p under current experimental conditions or genes, which may demonstrate Pub1p binding in alternative growth conditions.
Because a significant portion of those mRNAs that bound Pub1p did not display changes in stability, we wanted to explore if these genes represented missing values in our analyses or demonstrated insignificant changes in stability. Comparison of the 368 Pub1p bound targets to the group of 896 mRNAs that had not exhibited a change in turnover in the pub1Δ strain revealed a set of 35 transcripts that bound Pub1p but were not regulated at the level of stability (designated class B targets). Moreover, we found that these 35 mRNAs were involved in biological pathways distinct from the transcripts that exhibited Pub1p-dependent decay and were specifically engaged in a variety of transport processes (Table (Table5;5; see Table S2d in the supplemental material). Significantly, the fact that several mRNAs bound to Pub1p under conditions where their stability was not regulated indicates that Pub1p may have a function in regulating aspects of mRNA metabolism other than stability.
The validity of our mRNA target identification scheme by microarray analysis was additionally tested by monitoring for the presence of several mRNAs from both class A and B in the pool of transcripts precipitated with the TAP-tagged Pub1p. As shown in Fig. Fig.3,3, qRT-PCR performed on mRNAs which displayed regulated stability between wild-type and pub1Δ strains such as GCN4, RPS16B, RPS27B, SEC53, and RPS24B (class A) could amplify these transcripts from the PUB-TAP-bound mRNA pool, while both RT negative and mock isolations did not give any detectable signal. Similarly TIF51A (class B), which is predicted to bind Pub1p (65) but is not regulated at the level of transcript stability (Fig. (Fig.2C;2C; see Table S1b in the supplemental material), could be distinctively identified, while PGK1 or the URA5 transcripts that were not anticipated to associate with Pub1p (see Table S2a in the supplemental material) failed to show any selective amplification. This analysis therefore validates the data obtained by microarray analysis of PUB-TAP bound mRNAs by revealing direct association with the protein.
Pub1p is known to bind a range of sequences, including poly(U) (1), U-rich STE sequences in the 5′UTR of GCN4 and YAP1 (59), and AU-rich sequences from the TNF-α 3′UTR (65). We therefore wanted to determine if the global set of 368 mRNAs that bound Pub1p (Fig. (Fig.2C;2C; see Table S2a in the supplemental material) harbored any common sequence motifs. Additionally, we also wanted to distinguish if mRNAs that associated with Pub1p and displayed reduced stability (class A; 53 genes) or no change in stability in the pub1Δ strain (class B; 35 genes) were enriched for any conserved elements that could functionally distinguish these two classes.
We utilized MEME as a discovery tool (3) to query 100 nt of sequence upstream of the start codon and 200 nt of the sequence downstream of the stop codon of the Pub1p-associated genes for the presence of potential motifs. MEME results for the 368 Pub1p-bound mRNAs for which sequences could be retrieved revealed the presence of distinct 15 to 16 nucleotide sequence motifs in either the 5′- or 3′UTR regions (Fig. (Fig.4A).4A). The identified 3′UTR motifs showed high degree of similarity to ARE-like elements such as the AU- and U-rich sequences (70) and were also enriched in a less abundant but novel A-rich sequence (Fig. (Fig.4A).4A). Parallel analysis of the 5′UTR of the Pub1p global targets similarly revealed the presence of an A-rich motif, as well as a U-rich motif similar to the STE sequence (Fig. (Fig.4A).4A). All UTR motifs were revealed by MAST analysis to be statistically significant and enriched in our group of selected genes as opposed to random occurrence in the genome (see Table S3 in the supplemental material).
Significantly, MEME analysis for the transcripts that bound Pub1p and exhibited enhanced decay (class A) in the pub1Δ strain revealed an enrichment of the U-rich ARE-like 3′UTR element (Fig. 4B and C), whereas no selective motifs could be identified for those mRNAs that did not decay in a Pub1p-dependent manner (class B). The identified 3′UTR U-rich motif harbored a conserved U residue in position 12 and additionally showed a high degree of similarity to the HuR target motif (42) and the group III AU-rich elements (70). Moreover, the motif was identified in 50% of the Pub1p-regulated transcripts (class A), compared to a 9% genome frequency, thus representing an overall fivefold enrichment (Fig. (Fig.4B).4B). The high abundance of the U-rich motif in the 3′UTR of the regulated genes and the fact that Pub1p has been previously shown to recognize U-rich sequences are consistent with the hypothesis that these motifs represent binding sites of Pub1p and can potentially mediate Pub1p-dependent regulation of mRNA stability.
In yeast and in mammalian cells, 3′UTR sequences harboring regulatory elements such as the ARE are known to be a critical determinant of mRNA stability (70). We therefore wanted to determine if the 3′UTR sequences of class A transcripts harboring the predicted motifs could indeed be recognized by Pub1p. We selected the 3′UTR of RPS16B (YDL083C) and SEC53 (YFL045C) mRNAs, both of which were demonstrated to bind Pub1p (Fig. (Fig.2C2C and and3),3), harbored U-rich elements (Fig. (Fig.4C),4C), and additionally showed altered half-lives in the wild-type and pub1Δ mutant strains (Table (Table2).2). The putative 3′UTR sequences of these mRNAs were uniformly labeled with [α32P]UTP and UV cross-linking analysis performed in extracts prepared from either wild-type or pub1Δ strain as described previously (65). Radiolabeled TNF-α-ARE was used as a positive control, since Pub1p has been previously demonstrated to bind to this sequence (65). Our analysis showed that Pub1p, which migrates at ~60 kDa, bound specifically to the TNF-α-ARE and to the 3′UTR from RPS16B and SEC53 messages (Fig. (Fig.5)5) and was absent in extracts from the pub1Δ strain (Fig. (Fig.5,5, lane 14). As has been previously demonstrated for the Pub1p protein (59), this protein could be successfully competed off with either 100-fold excess of cold competitor or poly(U) (Fig. (Fig.5,5, lane 11) and was insensitive to poly(A) and poly(G) competition (Fig. (Fig.5,5, lanes 10 and 12). Taken together, these results show that Pub1p can directly and specifically bind to the 3′UTR sequences of RSPS16B and SEC53 mRNA containing specific ARE-like motifs. As both these transcripts exhibit altered turnover in the pub1Δ strain, our results suggest a role of the 3′UTR region of these transcripts in mediating Pub1p-dependent regulation of their stability.
To test whether the 3′UTR of targets exhibiting binding to Pub1p have a role in regulating mRNA stability, the half-life of a reporter gene harboring these regions was monitored to assess if Pub1p-dependent stability could be observed. To accomplish this, a chimeric plasmid was used in which the putative 3′UTR encompassing 350 nt downstream of the stop codon from either the SEC53 or the RPS16B gene was fused downstream of the MFA2 open reading frame under the control of the ADH1 promoter (Fig. (Fig.6A).6A). The stability of the MFA2 transcript was previously reported to be independent of the control of Pub1p (65), thus allowing us to use it as a sensor for measuring decay rates of the chimeras. Each of the fusion constructs was transformed into either PUB1 or pub1Δ, and the half-lives of the hybrid transcripts were monitored as described previously (15). As a control, the MFA2-PGK1 hybrid containing the 3′UTR from the PGK1 gene, which is predicted not to be regulated by Pub1p (59), was monitored and displayed no changes in stability between the two strains (Fig.6Bi). In contrast, both the MFA2-RPS16B and MFA2-SEC53 chimera displayed a stable half-life of 11 min and >24 min, respectively, in the wild-type PUB1+ strain. However, in the pub1Δ mutant strain, both these mRNAs were significantly destabilized, decaying with a half-life of 4 min for MFA2-16B and 11 min for the MFA2-SEC53 chimera (Fig. (Fig.6B).6B). Thus, the fusion transcript exhibiting differential decay kinetics could recapitulate the regulation in turnover as observed for the native transcript. These results therefore define RPS16B and SEC53 as two representative transcripts that demonstrate a direct correlation of Pub1p-dependent binding and stability. The results from this in vivo analysis furthermore demonstrate that the 3′-UTR region of RPS16B and SEC53 is adequate to respond to regulation by Pub1p and additionally has the ability to confer Pub1p-dependent regulation onto a heterologous transcript.
By systematically measuring global decay rates in the wild type and pub1Δ strain, we have explored the regulation of mRNA turnover mediated by the abundant mRNA binding protein Pub1p in the yeast Saccharomyces cerevisiae. Our results extend the previously characterized role of Pub1p as an mRNA-stabilizing factor, modulating the decay of two classes of mRNAs through distinct mRNA turnover pathways (22, 59, 65). Specifically, we find that nearly 9.2% of all cellular mRNAs are significantly destabilized by the loss of the PUB1 gene product in contrast to 1.3% of transcripts, which are stabilized in a pub1Δ strain (see Table S1b in the supplemental material). This result indicates that analogous to its mammalian homologue HuR, which functions to stabilize target mRNAs (42a), a significant role of Pub1p in regulating mRNA turnover is to protect transcripts from accelerated decay, through either direct or indirect control of their stability. Additionally, as a smaller subset of transcripts is stabilized by the absence of Pub1p, this furthermore suggests that the protein also has the capacity to mediate changes that help to promote mRNA turnover.
Mechanistically, the destabilization of transcripts in the pub1Δ strain can result from either direct interaction of Pub1p with its substrate or indirect effects of the protein on other processes influencing mRNA accumulation. To distinguish between these two possibilities, we have isolated and identified the repertoire of RNAs associating with Pub1p in vivo through microarray analysis. Our results surveying global mRNA targets of Pub1p reveal that 368 transcripts (Fig. (Fig.2C;2C; see Table S2a in the supplemental material), approximating 6% of the yeast genome is bound by Pub1p. We were able to accurately estimate half-life for about 24% of these transcripts and distinguished a set of 53 genes (class A) (see Table S2c in the supplemental material) that bind Pub1p and in parallel are stabilized by the presence of the protein. This result therefore establishes a direct correlation of binding and stability for a subset of Pub1p targets that could be precisely estimated in this study. A significant portion of mRNAs exhibiting regulated stability between wild-type and pub1Δ strains (Table S1b in the supplemental material) did not associate with Pub1p or did so transiently beyond the limits of detection. Similar limited overlaps have been previously reported for HuR targets, indicating both the complexity of genome-wide integrative approaches (41) and the potential for indirect effects of gene regulatory factors on downstream substrates. It is therefore likely that the stability of a considerable fraction of mRNAs is controlled by Pub1p through indirect mechanisms, probably through the modulation of trans-acting factors whose expression is directly regulated. The mRNAs that are stabilized by the loss of Pub1p (see Table S1b in the supplemental material) also appear to be indirect targets, since with the exception of one gene (YPR128C), the majority of these transcripts do not associate with the protein (see Table S2a in the supplemental material). Interestingly, we find that many of the Pub1p-bound genes encode mediators of posttranscriptional regulation such as PAB1, DED1, DHH1, and CAF20 (63), as well as members of the chaperone family such as SSA1 that have previously been implicated in regulating decay (15; see Table S2a in the supplemental material). Consequently, modulation of these effectors by Pub1p might selectively trigger cascades and mechanistically allow the regulation of stability of the indirect targets.
Comparison of the Pub1p-associated transcripts to mRNAs displaying no change in stability between the wild type and the pub1Δ strain also intriguingly identified a set of 35 genes (see Table S2d in the supplemental material) which bind Pub1p but are immune to regulation at the level of mRNA stability. This result suggests that expression of these targets is regulated by Pub1p through novel and uncharacterized mechanisms or alternatively requires additional condition-specific cofactors that may have a functional effect on mRNA decay. As Pub1p has been shown to cofractionate with a translationally inactive pool of cellular mRNAs (1) and its mammalian homologues, the TIA-1/TIAR and HuR proteins, are involved in translational regulation (42a, 44, 55), one likely role of Pub1p may be to act as a translational modulator of a subset of cellular transcripts. Interestingly, the TIA-1/TIAR proteins also function as cis regulators of alternate splicing (16, 18, 61), thus suggesting a possible role for Pub1p in mRNA processing. The presence of Pub1p in both the nucleus and cytoplasm (1), combined with the carbon-source sensitive function of the protein (65), additionally suggests that the protein can potentially regulate both nuclear and cytoplasmic processes under a variety of cellular conditions. Significantly, Pub1p was earlier shown to bind the TIF51A mRNA (65) whose stability is not altered in a pub1Δ strain (Fig. (Fig.2C2C and and3;3; see Table S1b in the supplemental material). This result is also consistent with an alternate function of Pub1p in modulating expression of this transcript. Interestingly, we also find that Pub1p binds the 3′UTR of both PAB1 and RPS27B mRNAs but cannot mediate regulation of mRNA stability exclusively through these sequences (R. Duttagupta and S. W. Peltz, unpublished observations). This furthermore supports the possibility that the association of Pub1p with mRNAs may control processes other than stability. Our analysis therefore distinguishes two distinct classes of Pub1p-associated transcripts that could be isolated in this study, based on the precision of measurements: those that exhibit Pub1p-dependent mRNA decay (class A) and those whose turnover appears to be independent of Pub1p (class B). Based on the intracellular abundance of Pub1p, the extent of each of these classes is likely to include a broader range of substrate mRNAs than could be determined in this analysis, thus indicating a major role of Pub1p in controlling cellular homeostasis.
It is known that in yeast and in higher eukaryotes, transcripts acting in the same biological pathway exhibit coordinated decay (29, 60, 68, 72), suggesting that mRNA dynamics are fine tuned to suit the biological requirements of the cell. This is also observed for protein-mRNA interactions where specific RNA-binding proteins bind to distinct subsets of transcripts sharing common physiological roles (19, 23, 27, 35). GO classification (38) of Pub1p targets shows that Pub1p regulates and associates with distinct classes of transcripts, the majority of which are involved in general translation (Tables (Tables11 and and4;4; see Tables S1c, S2b, and S2c in the supplemental material). Specifically, we find that most of the direct targets of Pub1p exhibiting altered turnover (class A) are involved in protein synthesis and protein metabolism, suggesting that control of mRNA stability is one of the mechanisms employed by Pub1p to regulate general translation (Table (Table4;4; see Table S2c in the supplemental material). Several of these mRNAs furthermore overlap with Pub1p targets identified using 1,10 phenanthroline as a transcriptional inhibitor (22; data not shown) thus substantiating an involvement of the protein in regulating substrates of select pathways. Interestingly, the expression of many genes involved in translation is regulated by changes in carbon source such as a shift to galactose (17) or under diauxic conditions (14), supporting the possibility that Pub1p might be a mediator of gene expression under these alternate cellular states. The paromomycin sensitivity of the PUB1 strain (Table (Table3)3) additionally supports a role of the protein in modulating translation, suggesting that Pub1p may enable rapid reprogramming of gene expression when cellular conditions change. Interestingly, several of the Pub1p-bound mRNAs, which do not display altered stability (target B), are involved in cellular transport (Table (Table5;5; see Table S2d in the supplemental material), suggesting that in addition to protein synthesis Pub1p regulates genes involved in the transport of metabolites through posttranscriptional mechanisms distinct from stability. We also find that in addition to these functional classes, Pub1p regulates the stability of retrotransposons (see Table S1c in the supplemental material), signifying that these elements may have evolved to take advantage of the stabilizing function of Pub1p.
It is known that selective mRNA degradation can be mediated by a number of different cis-acting sequences located in the mRNA substrate (70). The three previously characterized targets of Pub1p contain two distinct stability elements represented by either the 5′UTR stabilizer element or the 3′UTR AU-rich element (59, 65). To distinguish if the Pub1p-bound mRNAs were enriched in similar or novel motifs, we examined this distinct set of 368 mRNAs through a computational approach. Our results show that ARE-like, STE-like, and novel motifs can be identified in the putative 3′- and 5′UTRs of the endogenous Pub1p-bound targets (Fig. (Fig.4A)4A) and moreover are statistically enriched in this class (see Table S3 in the supplemental material).
For this study, we chose to focus on the 3′UTR motifs and identify a novel A-rich element and two ARE-like elements represented by either U-rich or AU-rich motifs in the set of Pub1p targets (Fig. (Fig.4A).4A). Of the ARE-like elements, the U-rich motif is particularly distinct as it is selectively enriched in the class A targets harboring mRNAs exhibiting regulated turnover (Fig. 4B and C). The high affinity of Pub1p for poly(U) sequences (1) and the similarity of the U-rich element to the HuR-binding motif in mammalian cells (42) suggest that the U-rich motif may have preferentially evolved to mediate Pub1p binding. We find several of our targets including RPS16B, SEC53 (Fig. (Fig.5),5), or RPS27B, PAB1, and OLE1 (data not shown) harbor the ARE-like elements and can specifically cross-link to Pub1p. This suggests that the element per se or its sequence context permits Pub1p binding and complex formation to mediate regulation involving that site.
Interestingly, the other ARE-like, AU-rich motif (Fig. (Fig.4A)4A) is very similar to the core efficiency element (EE), present in mRNAs associating with the shuttling hnRNP protein Nab4/Hrp1 (23, 25, 36). This element has been implicated in 3′ end processing by mediating proper poly(A) site selection (46). A comparison of seven yeast transcripts, harboring precisely mapped major poly(A) sites and putative EEs (24) with the Pub1p-associated mRNAs, reveal that for the ACT1 mRNA the AU-rich, ARE-like element directly overlaps with the EE for that gene (data not shown). Additional overlaps can be identified in ADH1, CYC1, and GCN4 mRNAs when the stringency of selection of the motifs is reduced. The possibility that the ARE-like motif may be involved in mRNA maturation is intriguing. As the 3′UTR of yeast mRNAs is relatively short (45), evolving multiple functions for a particular motif or having overlapping regulatory motifs may provide selective advantage by allowing coordinate control of diverse posttranscriptional events. We also find that the novel A-rich element isolated in our study (Fig. (Fig.4A)4A) displays a striking similarity to a 12-nt adenosine-rich sequence found in targets that associate with Nab2p (23), a factor involved in the control of polyadenylation and export (26). Since Pub1p does not bind adenylate sequences (Fig. (Fig.5,5, lane 10) and moreover biochemically interacts with Nab2p (31), this motif may therefore represent a secondary site for binding to Nab2p or other novel Pub1p-associated cofactors. The function of an mRNA thus may be modulated by multiple elements that combine regulation of different posttranscriptional events to control gene expression.
Theoretically, binding of Pub1p to regulatory sequences in the mRNA can either modulate stability or regulate other posttranscriptional events such as translation, 3′end processing or splicing. Based on the established regulatory role of the 3′UTR in controlling stability (70), we chose to test whether Pub1p binding to 3′UTR of targets harboring U-rich motifs can directly modulate transcript turnover for a set of candidate genes from class A which were demonstrated to decay in vivo in a Pub1p-dependent manner. Our results show that the 3′UTR from the RPS16B and SEC53 mRNAs is sufficient to confer Pub1p-dependent regulation of mRNA stability to a heterologous transcript (Fig. (Fig.6).6). This result supports our hypothesis that Pub1p mediates gene expression of a subset of transcripts through direct interaction with distinct regulatory sequences and furthermore identifies for the first time two endogenous targets of Pub1p whose decay is modulated specifically through their 3′UTRs.
Our data provide systematic evidence that Pub1p globally influences the posttranscriptional fate of mRNAs through distinct mechanisms, at least one of which is the regulation of mRNA stability. The mechanism of control of mRNA stability by Pub1p can be both direct and indirect. We show that for a subset of Pub1p targets, binding and stability can be directly correlated and is furthermore mediated through 3′UTR sequences (Fig. (Fig.55 and and6).6). The majority of Pub1p-regulated targets that could be isolated in this study, however, appear to be under indirect control. In principle, Pub1p can regulate stability by either altering mRNP conformation or regulating the enzymatic steps of the decay process. Our analyses reveal that PAB1 mRNA, coding for the poly(A)-binding protein, is rapidly destabilized in the pub1Δ strain (Table (Table2)2) with a concurrent moderate reduction in Pab1p steady-state protein levels (Duttagupta and Peltz, unpublished). As removal of Pab1p is a necessary step to triggering deadenylation and initiate decay (6), this result suggests that up-regulation of Pab1p by Pub1p could be a mechanism for regulating stability of a subset of transcripts that decay through the deadenylation-dependent decay pathway. In this model, in the absence of Pub1p, a decrease in the level of cellular Pab1p would result in rapid removal of poly(A) tails, resulting in destabilization of the target transcripts.
Interestingly, the presence of A-rich motifs in the UTRs of several Pub1p-bound mRNAs brings up the intriguing possibility that they might be Pab1p-binding sites or allow other cofactors such as Nab2p (23) to associate and modulate posttranscriptional events. The potential of Pub1p to modulate gene expression in response to altered cellular states (65) furthermore suggests that Pub1p targets identified in this study are a representative subset, based on specific cellular conditions. It is likely that the effect of Pub1p extends to a broader spectrum of physiological transcripts that exhibit regulation in differing conditions when the cellular effects of Pub1p become apparent.
In summary, results obtained from this study demonstrate that reprogramming of gene expression by Pub1p is mediated through posttranscriptional control of mRNA levels. Our results support a model where the coordinate control of stability and other posttranscriptional events is mediated by the dynamic interplay of cis-acting sequences and trans-acting factors, which finally determine the cytoplasmic fate of a message. Future experiments will focus on trying to determine the exact boundaries of the cis-acting elements and the mechanistic nature of Pub1p regulation on our repertoire of cellular targets.
We thank the entire CAG staff, especially Saleena Ghanny, Anthony Galante, and Virginie Aris, for assistance with microarray experiments and analysis. We thank the members of the Peltz lab for valuable comments and Andre Gerber for yeast strains and protocols. We also thank Yulei Wang for helpful discussions and Maurice Swanson for the kind gift of the PUB1 antibody.
This work is supported by a grant from the National Institutes of Health (GM 058276 and AI 057596) to S.W.P. C.J.W. was supported by the American Heart Association (award 0130470T).
†Supplemental material for this article may be found at http://mcb.asm.org/.