|Home | About | Journals | Submit | Contact Us | Français|
PUF proteins are eukaryotic RNA-binding proteins that repress specific mRNAs. The mechanisms and corepressors involved in PUF repression remain to be fully identified. Here, we investigated the mode of repression by Saccharomyces cerevisiae Puf5p and Puf4p and found that Puf5p specifically requires Eap1p to repress mRNAs, whereas Puf4p does not. Surprisingly, we observed that Eap1p, which is a member of the eukaryotic translation initiation factor 4E (eIF4E)-binding protein (4E-BP) class of translational inhibitors, does not inhibit the efficient polyribosome association of a Puf5p target mRNA. Rather, we found that Eap1p accelerates mRNA degradation by promoting decapping, and the ability of Eap1p to interact with eIF4E facilitates this activity. Deletion of EAP1 dramatically reduces decapping, resulting in accumulation of deadenylated, capped mRNA. In support of this phenotype, Eap1p associates both with Puf5p and the Dhh1p decapping factor. Furthermore, recruitment of Eap1p to downregulated mRNA is mediated by Puf5p. On the basis of these results, we propose that Puf5p promotes decapping by recruiting Eap1p and associated decapping factors to mRNAs. The implication of these findings is that a 4E-BP can repress protein expression by promoting specific mRNA degradation steps in addition to or in lieu of inhibiting translation initiation.
Precise regulatory mechanisms are crucial for the execution of gene expression programs and integration of signals. As intermediaries between genes and proteins, mRNAs are an important nexus for regulation. Posttranscriptional control of mRNA stability and translation is achieved through the concerted action of RNA binding factors, RNA decay enzymes, and the translation machinery (53, 62). Specific mRNAs are targeted for regulation by RNA binding factors that recognize sequences often found in the 3′ untranslated region (3′UTR). PUF proteins (Pumilio and Fem-3 binding factor) are one class of regulators (47, 73). PUF proteins are defined by a conserved RNA binding domain that mediates high-affinity binding to specific, 8- to 10-nucleotide, single-stranded RNA sequences (41, 48, 70, 76, 77). PUFs control diverse biological processes, including cell proliferation, development, fertility, and neurological functions (3, 7, 16, 17, 20, 36, 38, 39, 46, 52, 59, 75, 77). At the root of these functions lies the ability of PUFs to repress protein production from target mRNAs (47). The preponderance of evidence indicates that the major mechanism of PUF-mediated repression is by enhancing mRNA degradation (23, 47, 51, 71). In several cases, PUFs were shown to accelerate mRNA decay by removal of the 3′ polyadenosine tail (6, 21, 23, 24, 28, 51). PUFs were also reported to inhibit translation (9, 10, 26, 63, 72). A remaining challenge is to discover the corepressors and mechanisms of PUF-mediated repression.
Saccharomyces cerevisiae possess six PUFs, each of which bind a distinct set of mRNAs, dictated by their unique RNA binding specificities (22, 73). Puf4p and Puf5p/Mpt5p bind multiple mRNAs (22, 60) and share at least one well-characterized target, the mRNA encoding HO endonuclease, which catalyzes switching of the mating type (23, 28, 65). Puf4p and Puf5p were previously shown to bind specific sites in the HO mRNA 3′UTR and accelerate deadenylation and degradation of the message (23, 28). Deadenylation is essential for repression by Puf4p (28); however, Puf5p can still repress HO mRNA when deadenylation was blocked by deletion of the CCR4 gene (23), which encodes the major deadenylase (24, 67, 68). This finding indicated that Puf5p can repress by a second, deadenylation-independent mechanism (28). An additional corepressor(s) may be necessary for Puf5p activity.
We report here that a eukaryotic translation initiation factor 4E (eIF4E)-binding protein (4E-BP), Eap1p, serves as an essential corepressor for Puf5p. 4E-BPs are found throughout eukaryotes and are thought to inhibit translation by binding to the 5′ cap-bound initiation factor eIF4E, thereby blocking interaction with initiation factor eIF4G (62). 4E-BPs possess a conserved eIF4E binding motif, YXXXXLΦ (Φ indicates a hydrophobic amino acid, and X indicates any residue) (27, 42). 4E-BPs might globally reduce cap-dependent translation; however, specific examples demonstrate more-specialized roles (62). Two 4E-BPs have been identified in S. cerevisiae, Caf20p and Eap1p (1, 14). Both contain an eIF4E binding motif but are otherwise unrelated. Neither protein is essential for growth under standard conditions, but several mutant phenotypes have been described (8, 14, 32, 44, 45, 61). Eap1p was originally identified based on its ability to bind to eIF4E and was shown to compete with eIF4G (14). Therefore, like other 4E-BPs, Eap1p was proposed to repress by inhibiting translation initiation.
In this work, we show that Eap1p is required for Puf5p-mediated mRNA repression but is not necessary for Puf4p function. In contrast, Caf20p is fully dispensable for regulation by both Puf4p and Puf5p. Translational analysis demonstrates that Eap1p does not affect global translation nor does it inhibit polyribosome association of a Puf5p-targeted mRNA. Instead, we identify a novel activity of Eap1p to promote degradation of specific mRNAs, including a Puf5p target mRNA. Intriguingly, this activity is facilitated by the interaction of Eap1p with eIF4E. We find that deletion of the EAP1 gene causes deadenylated, capped mRNA to accumulate to substantial levels, indicating that Eap1p functions to promote removal of the mRNA's 5′ 7-methylguanosine cap. In accordance, coimmunoprecipitation experiments indicate that Eap1p associates with Puf5p and the decapping factor Dhh1p. Together, these results provide a new regulatory mechanism for a member of the diverse class of eIF4E-binding proteins, enhancement of mRNA decapping.
Yeast strains were obtained from Open Biosystems unless otherwise noted. The yeast strains used in this study were AGY111 (BY4741 MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0), AGY153 (BY4741 MATa eap1::KanR), AGY152 (BY4741 MATa caf20::KanR), AGY150 (BY4741 MATa puf5::KanR), AGY109 (BY4741 MATa puf4::KanR), and AGY151 (BY4741 MATa dhh1::KanR).
Plasmid ACG858 (YCp33 HIS3 HO 3′UTR) was previously described by Goldstrohm et al. (23). Plasmid ACG399 (YCp33 LacZ HO 3′UTR) was derived from ACG858 by replacing the HIS3 open reading frame (ORF) with the coding sequence for β-galactosidase (β-Gal). Plasmid ACG441 (YEp181 PUF5-T7) was previously described by Goldstrohm et al. (23). Plasmid ACG705 (p415 GPD PUF4-T7) was previously described by Hook et al. (28). Plasmid ACG137 (pACG1 NTB) contains the ADH1 promoter and 3′UTR. The ACG137 plasmid has a 2μ origin of replication and zeocin selectable marker. Plasmid ACG693 (pACG1 NTB-EAP1) was created by inserting the EAP1 open reading frame into KpnI and NotI sites in pACG1 NTB. To construct plasmid NB1 (YEp181 EAP1-FLAG), EAP1 was PCR amplified from S288C genomic DNA and cloned into XmaI sites of YEp181. The C-terminal FLAG epitope was added by inverse PCR using primers NB85/86. To construct plasmid NB2 (YEp181 EAP1 mt Y109A/L114A-FLAG), QuikChange PCR (Stratagene) was performed on plasmid NB1 with primers AG787/788 to mutate Y109A and L114A. To construct plasmid NB3 (pACG1-eIF4E-T7), the CDC33 ORF, encoding eIF4E, was PCR amplified from S288C genomic DNA and cloned into the KpnI and NotI sites of pACG1 plasmid. All plasmids were verified by restriction digestion and DNA sequencing.
Synthetic oligonucleotides were purchased from Integrated DNA Technologies. The synthetic oligonucleotides used in this study were as follows: NB64 (5′-TCCATTCCCGGGGTTTTAATGTATTGAAAATCACTTAGTTGTATATAGCC-3′), NB65 (5′-GCAAGGCCCGGGGCTTTCAGGCGCAGAAAACCTGAAAA-3′), NB85 (5′-GATGCGTAACGAGCGAGTACTTGACAGG-3′), NB86 (5′-TCACTTGTCATCGTCATCCTTGTAATCGATGTCATGATCTTTATAATCACCGTCATGGTCTTTGTAGTCTTTTATATTCTTTTTAGAG-3′), AG787 (5′-GCAATACAAGACATATGCAGCGTCCATGAATGAAGCGTATCATTTGAAACCATCTTTGGC-3′), AG788 (5′-GCCAAAGATGGTTTCAAATGATACGCTTCATTCATGGACGCTGCATATGTCTTGTATTGC-3′), AG1029 (5′-GCACGGTACCTATGTCCGTTGAAGAAGTTAGCAAG-3′), and AG1030 (5′-GCACGCGGCCGCTTACAAGGTGATTGATGGTTGAGGG-3′). The HO ORF riboprobe T7 transcription template PCR primers used were NB22 (5′-GGATCCTAATACGACTCACTATAGGGAGAACCTGCGTTGTTACCACAACTCTTATGA-3′) and NB23 (5′-AAGTGGTCACAAAACAAGAGAAGTTCCG-3′). The HO 3′UTR riboprobe T7 transcription template PCR primers used were NB92 (5′-GGATCCTAATACGACTCACTATAGGGAGACATCCAAAATATTAAATTTTACTTTTATTAC-3′) and NB98 (5′-GCAAGTATGTACCAGAAGCACGTGAA-3′). The LacZ 3′UTR riboprobe T7 transcription template PCR primers were NB92 (5′-GGATCCTAATACGACTCACTATAGGGAGACATCCAAAATATTAAATTTTACTTTTATTAC-3′) and NB115 (5′-CATCAGCCGCTACAGTCAACAGCAA-3′). The oligonucleotide used in HO 3′UTR RNase H cleavage was NB99 (5′-ATACAGTGATGACCGCTG-3′). The oligonucleotide used in LacZ 3′UTR RNase H cleavage was NB116 (5′-AACTGGAAGTCGCCGCGCCAC-3′). The RNR1 probes were pNB5 (5′-AAAGCACATTCCTTCAAGGTGTCGTAAATCCCCTCGATAGAGTCCTCCTTC-3′) and pNB6 (5′-AAGAGGACATTTGAGGTTTTGGAGTACCGGCATTGAACAACGTTGGAGAGG-3′). The RPL41A probe was pNB76 (5′-CCGCTTATTTGGATCTGGCTCTCACCTTCCGTCTCTTTCTCTTAAG-3′). The SCR1 probe was pNB79 (5′-CGCCTCCATCACGGGTCACCTTTGCTGAC-3′). The HO quantitative reverse transcription-PCR (qRT-PCR) primers were NB36 (5′-CCTCATAAGCAGCAATCAATTCTATCTAT-3′) and NB90 (5′-TTTAATTTCACCGTTAGCCATCAGAA-3′). The 18S rRNA qRT-PCR primers were NB69 (5′-ACGGAAGGGCACCACCA-3′) and NB70 (5′-CCACCCACAAAATCAAGAAAGAGCTCTC-3′).
Yeast growth assays were performed to detect repression by Puf4p and Puf5p as described by Goldstrohm et al. (23) with the following modifications. Wild-type yeast strain BY4741 or gene-specific deletion strains were transformed with the reporter gene YCp33 HOp HIS3-HO 3′UTR and either empty vector YEp181 or the PUF5 expression plasmid YEp181 PUF5 or p415 GPD PUF4. Colonies were isolated and grown to mid-log phase at 30°C, and the indicated number of cells was spotted onto selective minimal medium with or without histidine. The His3p competitive inhibitor 3-aminotriazole was added at a final concentration of 1 mM to increase stringency. For assays presented in Fig. 1C, the medium was supplemented with 300 μg/ml zeocin to select for pACG1 NTB EAP1 or the negative-control plasmid pACG1.
Wild-type or gene-specific deletion strains were transformed with the YCp33 LacZ-HO3′UTR reporter gene. Each strain was then grown to mid-log phase, and 8.9 × 107 cells (3 optical density at 600 nm [OD600] units) from each sample were harvested and resuspended in 100 μl of fresh medium. An equal volume of room temperature Beta-Glo (Promega) reagent was added to each tube. Samples were transferred to 96-well plate and incubated for 1 h at 25°C. Luminescence measurements were made using a GloMax Multi+ detection system (Promega). Specific signals were 2 orders of magnitude above the background level in wells measured with Beta-Glo reagent or medium or in empty wells. Each assay was performed with five biological replicates, and data are plotted as the mean value of relative light units with standard error of the mean.
Yeast cultures were seeded in 250 ml of the appropriate medium at an optical density at 600 nm (OD600) of 0.2 and grown to an OD600 of 0.8. The cells were rapidly harvested at 4°C, and all subsequent steps were carried out in a cold room. When indicated, cycloheximide (60 μg/ml) or EDTA (50 mM) was added to cultures, which were immediately poured into cold centrifuge bottles with one-third volume of crushed ice. The cells were pelleted by centrifugation at 3,200 × g for 5 min. The medium was decanted, and the cells were washed with 10 ml of ice-cold 50 mM Tris-Cl (pH 6.8), 100 mM NaCl, 30 mM MgCl2, and 50 μg/ml cycloheximide (or when indicated, 50 mM EDTA was added in lieu of cycloheximide). The cells were pelleted again and resuspended in 650 μl of ice-cold 50 mM Tris-Cl (pH 6.8), 100 mM NaCl, 30 mM MgCl2, and 50 μg/ml cycloheximide (or 50 mM EDTA was added in lieu of cycloheximide) with 20 U/ml RNasin, and 2× protease inhibitors (2 mM phenylmethylsulfonyl fluoride [PMSF], 100 μg/ml aprotinin, 100 μg/ml pepstatin, 100 μg/ml leupeptin) in 1.5-ml tubes containing 650 μl glass beads. Cells were lysed in a FastPrep (MP Biomedicals) 2 times for 60 s each time at 6.5 m/s. Cell debris and beads were removed from the extract by centrifugation for 5 min at 1,400 × g. The extracts were diluted 1/200, and the absorbance at 260 nm (A260) was measured with a NanoDrop spectrophotometer (Thermo Scientific). Twenty A260 units of extract were loaded onto each sucrose gradient. Ribosome runoff was performed as described above, except that cycloheximide was omitted from all steps.
To fractionate ribosomes and polyribosomes, 7 to 47% sucrose gradients were prepared using a Gradient Master (BioComp). Samples were applied to the top of each gradient. In Fig. 2, the gradients were centrifuged 2 h and 30 min at 28,000 rpm in an SW41 Ti rotor at 4°C. In Fig. 3, the gradients were centrifuged 4 h to resolve 40S, 60S, 80S, and polyribosome peaks. The gradients were fractionated using a Biologic DuoFlow system with an Econo Gradient peristaltic pump (Bio-Rad) at a rate of 1.75 ml/min while collecting 500-μl fractions. Gradient tubes were pierced with a gradient fractionation device (Brandel), and gradients were pumped from the bottom of the tube using Fluorinert. A260 readings were made continuously during fractionation using a Bio-Rad Quadtec spectrophotometer. Samples were stored at −80°C after fractionation. RNA was extracted from a total of 460 μl of each gradient fraction using the Maxwell RNA purification system and 16 cell LEV RNA purification kit (Promega). The percentage of each mRNA in each fraction was calculated relative to the total detected by Northern blotting in all fractions. Values from each fraction were represented as the mean value of multiple replicates. For Western blot analysis of fractions, 45 μl of each fraction was boiled 5 min in 1× SDS-PAGE loading buffer and then separated in a 4 to 12% SDS-polyacrylamide gel.
RNAs were separated on 1.4% agarose-formaldehyde denaturing gels with 1× morpholinepropanesulfonic acid (MOPS) running buffer and transferred to Immobilon NY+ membranes (Millipore) using a downward transfer method. The membranes were UV cross-linked with a UVP CL1000. Membrane blocking and hybridization were performed using oligo-hyb or ultra-hyb hybridization buffers (Ambion). End-labeled oligonucleotide probes were hybridized in oligo-hyb buffer overnight at 42°C and washed two times for 30 min each time in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) with 0.5% SDS at 42°C. Body-labeled riboprobes for HO mRNA were hybridized in ultra-hyb buffer at 68°C and washed two times for 5 min each time in 2× SSC with 0.1% SDS, and then washed two times for 15 min each time in 0.1× SSC with 0.1% SDS at 68°C. The membranes were exposed to phosphor screens and scanned using a Typhoon Trio phosphorimager (General Electric).
Transcription shutoff using thiolutin (Pfizer) and RNA purification was performed as described previously (5). Specific Northern blot bands were quantitated on a Typhoon phosphorimager using ImageQuant TL software. Each time point was normalized to the SCR1 RNA in the same sample, and decay was calculated relative to the time of drug addition. mRNA half-lives were calculated with GraphPad Prism software using nonlinear regression and one-phase decay analysis of biological replicate samples.
Forty micrograms of total RNA was first treated with Turbo DNase (Ambion). RNA was precipitated with 1/10 volume of 3 M sodium acetate and 3 volumes of 100% ethanol at −20°C. The pellets were washed with 70% ethanol and resuspended in 24 μl buffer A (20 mM KCl, 1 mM EDTA), 20 pmol of cleavage oligonucleotide NB99, and either 5 μl of water or 5 μl of oligo(dT) (500 ng/μl). The samples were heated for 5 min at 90°C, cooled for 5 min at 65°C, and then cooled for 20 min at room temperature. To each sample, 30 μl of buffer B (40 mM Tris-Cl [pH 8.0], 56 mM MgCl2) was added, along with 1 unit of RNase H (NEB), and 1 unit of RNasin Plus (Promega). The samples were incubated for 1 h at 37°C. RNA was then precipitated with 1/10 volume of 3 M sodium acetate and then with 3 volumes of ethanol at −20°C. RNA pellets were washed with 70% ethanol and resuspended in denaturing polyacrylamide Northern loading buffer. After heating for 5 min at 95°C, the RNA was separated on a 6% polyacrylamide, 7 M urea denaturing gel and transferred to Immobilon NY+ with a transblotter (Bio-Rad). Northern blotting was performed as described above.
Twenty micrograms of total RNA extracted from wild-type and eap1 strains was used for each reaction. Where indicated, control samples were treated with 15 units of tobacco acid pyrophosphatase (TAP) (Epicentre) to remove 5′ cap structures using the supplied TAP reaction buffer for 1 h at 37°C. RNA was precipitated, pelleted, and washed with 70% ethanol. RNA was then digested with 10 units of Xrn1 (NEB) in reaction buffer with 1 unit of RNase Inhibitor Plus for 1 h at 37°C. The RNA was precipitated, washed with 70% ethanol, and resuspended in Northern loading buffer.
Yeast cells expressing FLAG-EAP1, or empty vector (mock) as bait, and T7-tagged prey were grown overnight in the appropriate medium. On the next day, yeast cells were seeded to an optical density, OD600, of 0.2 in 1 liter of medium and then grown to an OD600 of 0.8. The cells were then harvested, washed with TNMT250 (50 mM Tris-Cl [pH 8.0], 250 mM NaCl, 2 mM MgCl2, 0.1% Tween 20), pelleted, and stored at −80°C. Anti-FLAG M2 agarose (Sigma) and a 50-μl bed volume of beads were preequilibrated in 10 ml TNMT250 with 300 μg/ml denatured salmon sperm DNA and 500 μg/ml bovine serum albumin (BSA) for 1 h at 4°C. The beads were then washed twice for 10 min each time with TNMT250. The cell pellets were thawed on ice, suspended in an equal volume of TNMT250 with 10 μg RNase A (Fermentas), 100 units of RNase One (Promega), and 2× protease inhibitors. The cells were transferred to 15-ml tubes containing 700 μl acid-washed glass beads. Lysis was performed with a FastPrep (MP) with three 60-s pulses at 6.5 m/s. Cell debris was pelleted by centrifugation at 6,000 × g for 10 min, and the supernatant was transferred to a fresh tube and pelleted again by centrifugation at 12,000 × g for 10 min. The resulting lysate was precleared by incubation with IgG-agarose to remove nonspecific interactions. After preclearing, RNase-treated lysates were applied to anti-FLAG M2 affinity agarose (Sigma) and incubated for 2 h at 4°C. The beads were pelleted at 1,000 × g for 5 min, the supernatant was removed, and beads were washed 5 times for 15 min each time with 10 ml TNMT250. After the final wash, beads were transferred to a microcentrifuge tube. The beads were resuspended in 100 μl of TNMT250 and 150 ng of FLAG peptide (Sigma). Elution was performed 30 min at 4°C with end-over-end rotation. After incubation, the supernatant was passed through a Bio-Rad mini-spin column to remove beads. Eluates were then separated by 4 to 12% SDS-PAGE gels and probed with anti-FLAG antibody (Sigma) and anti-mouse horseradish peroxidase (HRP)-conjugated monoclonal antibody (Thermo Scientific), anti-T7 monoclonal antibody linked to HRP (Novagen), or antiactin monoclonal antibody (MP Biomedical).
Coimmunoprecipitation of mRNA with Eap1p was performed as described above with the following alterations. Yeast cells were lysed in TKNM140 buffer (40 mM Tris-Cl [pH 8.0], 140 mM KCl, 0.1% NP-40, 2 mM MgCl2, 40 units/ml RNasin Plus with 2× protease inhibitors). The beads were washed five times for 10 min each time in TKNM140 buffer, and then protein and RNA were eluted as described above with 150 ng of FLAG peptide.
Input and elution samples were first treated with 4 units of Turbo DNase (Ambion) in 1× Turbo buffer at 37°C for 30 min. RNA was precipitated with 1/10 volume sodium acetate and 3 volumes ethanol for 1 h at −20°C, and the pellets were washed with 70% ethanol. qRT-PCR was carried out as described previously (5). Briefly, HO cDNA was generated with GoScript reverse transcriptase (Promega) using 20 pmol of primer NB90. Amplification of PCR products was measured using GoTaq qPCR master mix (Promega) with 200 nM each primer in 50-μl reactions. A Bio-Rad CFX 96 C1000 real-time PCR instrument was used for all assays. Each immunoprecipitation was performed in triplicate, and the elution threshold cycle (CT) was normalized to HO input CT for each sample. Fold enrichment was then calculated relative to the mock immunoprecipitation samples using the ΔΔCT method (40, 57).
To identify corepressors necessary for PUF-mediated repression, we undertook a genetic approach using a reporter gene that is repressed by Puf4p and Puf5p (23, 28). The HIS3-HO 3′UTR reporter gene was created by replacing the open reading frame of HO with the auxotrophic marker gene HIS3 (Fig. 1A) (23). In wild-type cells, wherein Puf4p and Puf5p levels do not fully silence the reporter, introduction of the reporter confers histidine biosynthesis and thus growth on medium lacking histidine (Fig. 1B, wild-type [WT] strain). As previously demonstrated (23, 28), increased expression of Puf4p or Puf5p repressed HIS3-HO expression, thereby abrogating growth in the absence of histidine (Fig. 1B, wild-type strain with PUF5 and PUF4 expression plasmids). Importantly, repression depends on the PUF binding sites and the RNA binding activity of each PUF (23, 28).
If a corepressor is required for PUF repression, deletion of its gene will result in loss of repression and thus growth on medium lacking histidine. We tested candidate genes with known roles in mRNA degradation and translational control. Deletion of one gene encoding a 4E-BP, EAP1, abrogated Puf5p repression but had no effect on Puf4p repression (Fig. 1B, Δeap1 strain with PUF5 versus PUF4). Therefore, EAP1 is necessary for Puf5p repression and dispensable for Puf4p function, indicating a unique mode of repression by Puf5p. Because Eap1p binds to eIF4E (14), our data suggest that Puf5p may repress translation using Eap1p as a corepressor. We also tested the second known yeast 4E-BP, CAF20. In contrast to Eap1p, Caf20p is not necessary for repression by either PUF4 or PUF5, as revealed by the ability of each PUF to repress in cells lacking the CAF20 gene (Fig. 1B, Δcaf20 strain). This indicates that Puf4p represses through a separate mechanism that does not require a 4E-BP.
If Eap1p is a Puf5p corepressor, then overexpression of Eap1p may also repress the HIS3-HO reporter. Indeed, when Eap1p was overexpressed by introducing an Eap1p expression plasmid into wild-type cells, the HIS3-HO reporter was repressed (Fig. 1C). Eap1p inhibitory activity was dependent on Puf5p, as Eap1p no longer repressed in a puf5 deletion strain (Fig. 1C). Together, these data demonstrate that Eap1p and Puf5p function together.
To further measure the contribution of Eap1p to Puf5p to repression, we created a reporter gene encoding β-galactosidase (β-Gal) controlled by the HO 3′UTR (LacZ-HO 3′UTR [Fig. 1D]). Protein expression from LacZ-HO was measured from an identical number of cells using a luminescence-based β-Gal activity assay in three genetic backgrounds: wild type, eap1 deletion, and puf5 deletion. If Eap1p and Puf5p function together to repress HO, then β-Gal activity should increase when each is absent. Indeed, deletion of PUF5 resulted in a 6.5-fold increase in β-Gal activity relative to the wild type (Fig. 1E), and deletion of EAP1 caused a 14.8-fold fold increase (Fig. 1E). For a positive control for comparison, mutation of both PUF binding sites (LacZ-HO 3′UTR mt [Fig. 1D]) resulted in a 6.2-fold increase (Fig. 1E) (28). These results indicate that Eap1p represses β-Gal synthesis or promotes its decay. To gain additional insight, the steady-state level of LacZ mRNA was measured by Northern blotting (see Fig. S1 in the supplemental material). Deletion of EAP1 increased LacZ-HO mRNA by 2.5-fold (Fig. 1F), suggesting that Eap1p represses LacZ mRNA synthesis or promotes mRNA degradation. Consistent with their role in promoting decay of HO mRNA, PUF binding site mutations or deletion of PUF5 caused a 2.3-fold increase in the reporter mRNA (Fig. 1F). Deletion of EAP1 increased the ratio of β-Gal activity to LacZ mRNA by 5.6-fold relative to wild-type cells (Fig. 1G), suggesting that the amount of protein synthesized per mRNA may increase. Together, our data demonstrate that both Puf5p and Eap1p repress a target mRNA and that Puf5p activity is dependent on Eap1p. The results show that Eap1p reduces both protein and mRNA expression, and given that 4E-BPs are generally thought to inhibit translation, we next investigated the effect of Eap1p on HO translation.
4E-BPs are proposed to inhibit translation by blocking interaction of eIF4E and eIF4G during initiation (62). In this context, we hypothesized that Puf5p may utilize Eap1p to inhibit HO translation. To measure the effect of Eap1p on the translation state of HO mRNA, we performed sucrose gradient ultracentrifugation to separate ribosome-bound and -unbound mRNAs. If Eap1p inhibits initiation, Eap1p should decrease the percentage of HO mRNA bound to ribosomes (ribosome occupancy) and reduce the number of ribosomes bound to HO mRNA. Therefore, deletion of EAP1 should increase the ribosome occupancy and density of HO mRNA.
To test these predictions, cell extracts from wild-type and eap1 deletion strains were separated on 7 to 47% sucrose gradients. Each gradient was fractionated while monitoring UV absorption to show separation of 80S ribosomes and polyribosomes (Fig. 2A and andB),B), corroborated by ethidium bromide staining of rRNAs (Fig. 2C). The chromatograms of wild-type and eap1 deletion strains were highly similar, indicating that Eap1p does not substantially alter global translation (compare Fig. 2A and andBB showing the WT and Δeap1 strains, respectively).
We next detected HO mRNA in the gradient fractions from three biological replicates to quantitate the HO translation state. In wild-type cells, 97% of total HO mRNA was in polyribosome-bound fractions (Fig. 2D and and2E,2E, WT, fractions 7 to 19). For comparison, the average ribosome occupancy for mRNAs in S. cerevisiae is 71% (2). HO mRNA was predominantly detected in fractions containing 3 or more ribosomes, steadily increasing in the fractions containing 7 or more ribosomes (Fig. 2D, WT fractions 9 to 19). Less than 1% of HO mRNA was found in ribosome-free fractions (Fig. 2D, fractions 1 to 3), and only 2% was present in fractions containing monoribosomes (Fig. 2D, lanes 1 to 9). For a control, polyribosomes were dissociated with EDTA. This treatment caused HO mRNA to shift from the bottom to the top of the gradient (see Fig. S2 in the supplemental material), consistent with HO being polyribosome associated. These results indicate that HO mRNA efficiently engages with ribosomes in wild-type cells, contradicting the prediction that Eap1p inhibits translation initiation of HO mRNA.
Next, the effect of Eap1p on HO translation state was investigated. Deletion of EAP1 did not alter the ribosome occupancy of HO mRNA; 98% associated with polyribosomes, nearly identical to the wild type (Fig. 2D and andE,E, Δeap1 strain, fractions 7 to 19). Like the wild type, less than 1% of HO mRNA was present in the ribosome-free fractions (Fig. 2D, Δeap1 strain, fractions 1 to 3). The ribosome density of HO mRNA actually decreased slightly in the eap1 deletion strain, with the peak density shifting from polyribosome fraction 16 in wild-type cells to the less-dense fraction 15 in eap1 deletion cells (Fig. 2D and andE,E, Δeap1 strain). These observations contradict the prediction that deletion of EAP1 would increase ribosome occupancy and density of HO mRNA.
The major difference observed between wild-type and eap1 deletion strains is a change in the abundance of HO mRNA, which increased by 1.7-fold in the eap1 deletion strain (Fig. 2D and andE,E, Δeap1 strain). We conclude that Eap1p does not inhibit translation initiation of HO mRNA but instead decreases the abundance of HO mRNA, suggesting an effect on HO mRNA synthesis or stability.
We also investigated the effect of Eap1p on the translation state of two mRNAs that are not PUF targets: the ribonucleotide reductase mRNA, RNR1, and the large ribosomal subunit protein L47 mRNA, RPL41A (22, 23, 60). Like HO, RNR1 is a low-abundance, cell cycle-regulated mRNA with a large ORF of 2,667 nucleotides (compared to HO ORF, which is 1,761 nt). The distribution of RNR1 mRNA was not altered by Eap1p (Fig. 2F). All RNR1 associated with polyribosomes in wild-type and eap1 deletion cells (Fig. 2F. compare fractions 12 to 19 from the WT and Δeap1 strains). RPL41A is an abundant mRNA with a 78-nucleotide ORF that engages, on average, one ribosome in wild-type cells (Fig. 2G, WT strain, peak fractions 4 to 6) (2). The ribosome association of RPL41A mRNA was not altered by deletion of EAP1 (Fig. 2G, Δeap1 strain). Therefore, EAP1 does not change the ribosome association of two mRNAs that are not targeted by PUFs.
We next evaluated whether Eap1p associates with ribosomes. If Eap1p blocks translation initiation, then Eap1p would be expected to be found exclusively in translationally inactive, ribosome-free fractions of the sucrose gradient. Eap1p with a FLAG tag was expressed in the eap1 deletion strain, and extracts were fractionated by sucrose density gradients using conditions that separate ribosomal subunits, mono-, and polyribosomes (Fig. 3A). Three peaks of Eap1p were observed in the gradient fractions. At the top of the gradient, a peak was present in the ribosome-free fractions (Fig. 3A, fractions 1 and 2). A second peak cofractionated with 60S ribosomal subunit (Fig. 3A, fraction 5 and 6). A third major peak of Eap1p fractionated with polyribosomes (Fig. 3A, fractions 14 to 16). Interestingly, a slower-migrating Eap1p species was observed in the first two gradient fractions, perhaps the result of a posttranslational modification(s).
Eap1p binds eIF4E, and this interaction may mediate Eap1p association with polyribosomes. First, we assessed the distribution of eIF4E in the gradient fractions. Western blotting of eIF4E revealed that the protein was distributed throughout the gradient, with a major peak at the top of the gradient (Fig. 3A, middle panel, fractions 1 to 4) and a secondary peak corresponding to the 80S monoribosome (Fig. 3A, fractions 6 to 8). Like Eap1p, eIF4E was also present in polyribosome fractions (Fig. 3A, fractions 9 to 16). To test whether Eap1p binding to eIF4E is necessary for polyribosome association, the eIF4E binding motif, Y109XXXXL114, was mutated by introducing two alanine substitutions, Y109A and L114A, to create Eap1p mt (mt stands for mutant) (14). This mutant was expressed in the eap1 deletion strain, and then extract from these cells was fractionated on a sucrose gradient. The Eap1p mutant was detected only in the first three fractions at the top of the gradient (Fig. 3A, Eap1 mt, fractions 1 to 3), indicating that the association of Eap1p with polyribosomes is dependent on binding to eIF4E. To confirm that Eap1p mt no longer bound to eIF4E, wild-type Eap1p or Eap1p mt were immunoprecipitated. eIF4E was detected only in the wild-type Eap1p immunoprecipitate (Fig. 3B); therefore, the mutations disrupted interaction with eIF4E.
Two controls were performed to verify that Eap1p associated with polyribosomes. First, extracts were treated with EDTA, resulting in collapse of polyribosomes into 40S and 60S peaks (Fig. 3C) and causing Eap1p to fractionate predominantly at the top of the gradient (Fig. 3C, fractions 1 to 7). Importantly, the portion of Eap1p that fractionated with polyribosomes in wild-type cells (Fig. 3A, fractions 9 to 16) was greatly diminished (Fig. 3C, fractions 9 to 16). Second, ribosomes were allowed to elongate and “run off” by omitting cycloheximide (37), resulting in accumulation of 80S particles (Fig. 3D, fractions 6 and 7). If Eap1p were engaged in translating polyribosomes, then runoff should cause Eap1p to shift into lighter fractions. Indeed, Eap1p was detected only at the top of the gradient (Fig. 3D, fractions 1 and 2). Collectively, these data indicate that Eap1p associates with polyribosomes. This finding was unexpected and is not consistent with the model that Eap1p inhibits translation initiation. Instead, our findings indicate that Eap1p may promote a distinct mode of repression that remained to be discovered.
Puf5p promotes degradation of the mRNAs it targets (23, 24, 28, 60). Because Eap1p serves as a corepressor for Puf5p and HO (Fig. 2D) and LacZ-HO 3′UTR (Fig. 1F) mRNA levels increased in the eap1 deletion strain, we reasoned that Eap1p may affect mRNA decay. To test this hypothesis, mRNA decay rates were measured in the presence or absence of Eap1p. Cells were treated with thiolutin to inhibit transcription, and RNA samples were collected over time. Next, specific mRNAs were detected by Northern blotting (Fig. 4). In wild-type cells, the half-life of HO mRNA was 12 min (Fig. 4A), which is consistent with past measurements (23, 28). Deletion of EAP1 dramatically stabilized HO, increasing the half-life to 41 min (Fig. 4A). These results represent the first demonstration that EAP1 affects the rate of mRNA decay.
We next tested whether the interaction of Eap1p with eIF4E was required for acceleration of mRNA decay. To do so, the eap1 deletion strain was complemented with plasmid expressing either wild-type Eap1p (EAP1) or eIF4E binding-defective mutant (EAP1 mt). The HO mRNA half-life was then measured in each strain. When wild-type Eap1p was expressed in eap1 deletion cells, the HO mRNA half-life was 7.5 min (Fig. 4A, Δeap1 + EAP1), a substantial reduction relative to 41 min in the eap1 deletion strain and slightly shorter than the 12 min observed in wild-type cells. These results, summarized in Fig. 4E, support the conclusion that Eap1p promotes HO mRNA decay. Eap1p mt also accelerated decay of HO mRNA, resulting in a half-life of 20 min (Fig. 4A, Δeap1 + EAP1 mt). Thus, binding to eIF4E facilitates but is not absolutely required for Eap1p to enhance mRNA decay.
To determine whether the effect of Eap1p on mRNA stability was specific to PUF-regulated mRNAs, we analyzed two mRNAs that are not regulated by PUFs, RNR1 and RPL41A. RPL41A had a half-life of about 60 min that was not altered by deletion of EAP1 or complementation with wild-type or mutant EAP1 (Fig. 4B). In contrast, the half-life of RNR1 mRNA increased from 18 min in wild-type cells to 53 min in the eap1 deletion strain (Fig. 4C). Importantly, expression of wild-type Eap1p in the deletion strain restored the RNR1 half-life to 16 min (Fig. 4C, Δeap1 + EAP1). EAP1 mt also restored RNR1 decay to 22 min, albeit less effectively than wild-type EAP1 (Fig. 4C, Δeap1 + EAP1 mt). These results are summarized in Fig. 4F. The effect on RNR1 demonstrates that Eap1p promotes degradation of an mRNA that is not a known PUF target. These results reveal a novel activity of Eap1p to promote mRNA degradation. Importantly, Eap1p binding to eIF4E facilitates decay, but this interaction is not obligatory for enhanced mRNA degradation.
Having established that Eap1p accelerates mRNA degradation, we next investigated which step of decay is affected. mRNA degradation generally initiates by removal of the polyadenosine tail (i.e., deadenylation). Once the polyadenosine tail is shortened to about 10 nucleotides (pA10), the mRNA is decapped and degraded in a 5′-to-3′ direction or by the alternative 3′-to-5′ decay pathway (11). Previous research demonstrated that Puf5p and Puf4p enhance deadenylation (23, 24, 28). Because Eap1p serves as a Puf5p corepressor, we speculated that it may affect deadenylation. To compare the rate of HO deadenylation in wild-type and eap1 deletion cells, transcription was inhibited with thiolutin, and RNA samples were collected over time. To resolve the poly(A) tail length, HO mRNA was cleaved with RNase H and a cDNA oligonucleotide to generate a 253-nucleotide, 3′ fragment with a poly(A) tail of up to 80 nucleotides (Fig. 5A). Products were resolved by denaturing polyacrylamide gel electrophoresis and detected by Northern blotting (Fig. 5B). In wild-type cells, HO 3′UTR mRNAs had poly(A) tails ranging from 80 to 10 adenosines (Fig. 5B, lane 2). Treatment with RNase H and oligo(dT) provided a marker for deadenylated HO mRNA (Fig. 5B, lane 1). The stable noncoding RNA, SCR1, served as a loading control (Fig. 5B). Following transcription shutoff, the tail was quickly shortened over the first 10 min to a length of about 10 nucleotides. This oligo-adenylated intermediate subsequently decayed with a 6-min half-life (Fig. 5B, lanes 3 to 7), consistent with our past observations (23). HO mRNA was deadenylated within 10 min in the eap1 deletion strain, exhibiting the same kinetics observed in wild-type cells (Fig. 5B, lanes 9 and 10). To more closely analyze the effect of Eap1p on HO deadenylation, we repeated the experiment using 3-min intervals (Fig. 5C). In wild-type and eap1 deletion cells, HO poly(A) tails progressively shorten at equivalent rates (Fig. 5C). Therefore, deadenylation does not appear to be affected by Eap1p. However, a major difference was observed in the eap1 strain: the oligo-adenylated intermediate accumulated and persisted throughout the time course with a half-life greater than 50 min (Fig. 5B, lanes 10 to 14, and Fig. 5C, lanes 9 to 14). This pattern of decay, wherein an oligo-adenylated mRNA accumulates, is identical to that caused by mutations in decapping factors (4, 13, 19, 66).
To further characterize the impact of Eap1p on HO mRNA decay, we detected the 5′ HO mRNA fragment by Northern blotting (Fig. 5B, HO 5′ fragment). In wild-type cells, the 5′ fragment began to rapidly disappear as the oligo-adenylated species appeared, consistent with the mRNA being degraded by the 5′ decay pathway (Fig. 5B, lanes 2 and 3). In contrast, when EAP1 was deleted, the 5′ fragment persisted throughout the time course (Fig. 5B, lanes 9 to 14). These results demonstrate that Eap1p accelerates degradation of HO mRNA at a step following deadenylation and preceding 5′ decay.
We next sought to determine whether the HO mRNA that accumulated in the eap1 deletion retained a 5′ cap or was decapped but not subsequently degraded by the processive, 5′-to-3′ exonuclease, Xrn1p. To assess the presence of the 5′ cap, RNA was treated with recombinant Xrn1p. Decapped mRNA is sensitive to degradation by Xrn1p, whereas capped mRNA is resistant. Ninety-five percent of HO mRNA in wild-type cells was resistant to Xrn1p, consistent with the presence of a 5′ cap (Fig. 5D, lanes 3 and 4). HO mRNA from eap1 cells was fully resistant to Xrn1p and thus retained a 5′ cap (Fig. 5D, lanes 7 and 8). Thus, deletion of EAP1 causes accumulation of capped HO mRNA. We also observed that the RNR1 mRNA that accumulated in eap1 deletion cells was also resistant to digestion with Xrn1p, indicating that it too remained capped (Fig. 5D). Two controls were performed to verify the Xrn1p assay. First, RNA was decapped with tobacco acid pyrophosphatase (TAP) and then incubated with Xrn1p, resulting in a 64% reduction of HO and RNR1 mRNAs (Fig. 5D, lanes 11 and 12) (5). Second, ethidium bromide staining the RNA samples demonstrated that Xrn1p degraded the uncapped 26S and 18S rRNAs (Fig. 5D, lanes 3 and 4, 7 and 8, and 11 and 12). As observed in Fig. 5B, the predominant HO mRNA decay intermediate in the eap1 strain at 20 and 40 min after transcription shutoff has an oligo-adenylate tail. We used Xrn1p sensitivity to determine whether this mRNA species possesses a 5′ cap. At 20 min, 76% of the HO intermediate was resistant to Xrn1p. At the 40-min time point, the HO intermediate was fully resistant to Xrn1p digestion (Fig. 5E, lanes 3 and 4). Removal of the 5′ cap with TAP and then digestion with Xrn1p resulted in decapping and destruction of 95% of the oligo-adenylated HO intermediate (Fig. 5E, lanes 5 and 6). Collectively, these results support the conclusion that deletion of EAP1 causes the accumulation of capped, oligo-adenylated HO mRNA. We conclude that Eap1p promotes decapping, a novel function for a 4E-BP.
We next asked whether Puf5p associates with Eap1p. FLAG-tagged Eap1p (Eap1-FLAG) was coexpressed with T7-tagged Puf5p (Puf5-T7) in wild-type cells. Eap1p was then immunoprecipitated with FLAG antibody resin, washed extensively, and specifically eluted with FLAG peptide. Eluates were then analyzed by Western blotting (Fig. 6A). As a negative control, a mock FLAG immunoprecipitation was performed on cells expressing only Puf5p-T7 (Fig. 6A). Puf5p was detected in the Eap1p immunoprecipitate (Fig. 6A). As a negative control, the actin protein, which is abundant in the input extracts, was not detected in the FLAG eluates (Fig. 6A). Because these extracts were extensively treated with both RNase A and RNase One to degrade RNA prior to immunoprecipitation (Fig. 6B), we conclude that Puf5p likely associates with Eap1p via protein interactions, not by a bridging RNA.
The observation that Eap1p enhanced decapping suggested that it may physically associate with the decapping machinery. The Dhh1 protein, a DEXD/H box helicase, is a well-known activator of decapping (11). T7-tagged Dhh1p coimmunoprecipitated with Eap1p-FLAG but was not present in mock immunoprecipitate (Fig. 6C). These extracts were also RNase treated prior to immunoprecipitation; therefore, association of Eap1p and Dhh1p is not dependent on RNA. This result provides a physical link between Eap1p and the decapping machinery. In addition to this physical interaction, the mRNA decay phenotypes caused by deletion of EAP1 (Fig. 5A) and DHH1 (see Fig. S3 in the supplemental material) (13) were remarkably similar; HO mRNA was stabilized (half-life [t1/2] of >50 min) and accumulated as an oligo-adenylated species.
Our data show that Eap1p participates in Puf5p-mediated degradation of HO mRNA. Puf5p binds directly to the HO 3′UTR and may recruit Eap1p to the message. Alternatively, Eap1p may associate with HO mRNA by binding to eIF4E. To test these models, we expressed FLAG-tagged Eap1p or the eIF4E binding-defective Eap1p mt and asked whether HO mRNA coimmunoprecipitated with each protein. To assess the role of Puf5p in recruiting Eap1p, wild-type Eap1p-FLAG was immunoprecipitated from the puf5 deletion strain. As a negative control, a mock immunoprecipitation was performed using extract from wild-type cells. The FLAG eluates were analyzed by Western blotting to confirm purification of Eap1p and Eap1 mt (Fig. 7A). HO mRNA was measured in the eluates by reverse transcription and quantitative PCR. HO mRNA was nearly undetectable in mock eluates, whereas it was enriched 1,400-fold in the Eap1p FLAG eluates (Fig. 7B). This demonstrates that Eap1p associates with HO mRNA.
Purification of Eap1p from a puf5 deletion strain reduced its association with HO mRNA by 35-fold relative to the wild-type strain (Fig. 7B, 49-fold enrichment); therefore, Puf5p facilitates Eap1p association with HO mRNA. We next tested the contribution of Eap1p interaction with eIF4E. No significant change was observed relative to wild-type Eap1p; HO mRNA was enriched 1,400-fold in the Eap1 mt FLAG eluates (Fig. 7B). As a negative control, the noncoding 18S rRNA was not enriched in these immunoprecipitates. These findings demonstrate that Eap1p associates with HO mRNA, mediated by Puf5p.
Yeast PUF proteins have a well-documented role in accelerating mRNA degradation (23, 24, 28, 29, 34, 51, 60, 69), and deadenylation plays an important role. Both Puf4p and Puf5p enhance deadenylation of HO mRNA in vivo and in vitro (23, 24, 28). Puf4p repression depends on both POP2 and CCR4 genes, which encode subunits of the Ccr4-Not deadenylase complex, and the catalytic activity of Ccr4p deadenylase (24, 28). That said, several clues indicate that additional mechanisms are utilized by specific PUFs. First, Puf5p represses a target mRNA even when deadenylation is genetically blocked by removal of the CCR4 gene (23). Further evidence of a deadenylation-independent mechanism was revealed from analysis of HO mRNA half-lives in different genetic backgrounds. HO mRNA is stabilized 10-fold when both PUF4 and PUF5 are deleted, whereas deletion of CCR4 stabilizes HO only 3-fold (23, 24, 28). While deadenylation of HO is blocked by the absence of Ccr4p, the mRNA is still degraded (24), indicating that another step of decay is promoted by Puf5p.
We hypothesized that Puf5p recruits additional corepressor proteins (23, 24, 28). In the present work, we provide strong evidence that the eIF4E-binding protein Eap1p participates in Puf5-mediated repression. Deletion of EAP1 blocks the repression by Puf5p (Fig. 1). In reciprocal tests, Eap1p represses an HO reporter mRNA, dependent on the PUF5 gene. Together, these results indicate that Eap1p is a Puf5p corepressor. The requirement of Eap1p is specific to Puf5p, as Puf4p repression is not EAP1 dependent (Fig. 1). Additionally, Eap1p and Puf5p associate with each other (Fig. 6) and HO mRNA (Fig. 7), biochemically connecting both regulators to the target mRNA.
Eap1p is one of two 4E-BPs encoded by the S. cerevisiae genome (1, 14). The second 4E-BP, Caf20p, was reported to associate with Puf4 and Puf5p in an RNA-dependent manner (15), though the function was not tested. The results of our functional analysis demonstrate that CAF20 is dispensable for repression by Puf4p and Puf5p (Fig. 1). Thus, a general 4E-BP function is not essential for PUF repression. Instead, our data indicate that Puf5p specifically utilizes Eap1p to elicit repression. We conclude that Puf4p and Puf5p exert their repressive effects through distinguishable mechanisms. Both PUFs enhance deadenylation, while Puf5p also promotes Eap1p-dependent repression (Fig. 8) (23, 24, 28). The features of Puf5p that confer Eap1p dependence are currently not known. While the conserved RNA binding domains of Puf4p and Puf5p bind to the Pop2p-Ccr4p deadenylase (Fig. 8) (23, 24, 28), unique domains of Puf5p may dictate specificity for Eap1p.
The physical association of Eap1p with Puf5p and the dependence of Puf5p for Eap1p interaction with HO mRNA support a model wherein Puf5p binds HO mRNA and recruits Eap1p (Fig. 8). How might Eap1p be recruited? We have not detected a direct protein interaction between Eap1p and Puf5p, which could be a purely technical issue, as both full-length proteins are difficult to purify. Alternatively, a factor that bridges the interaction may exist. Because Puf5p and Eap1p coimmunoprecipitate from extracts extensively treated with RNases, RNA is not a likely bridging factor (Fig. 6). Decapping factors may interconnect Eap1p and Puf5p, because decapping proteins associate with both regulators (Fig. 6) (23). Future biochemical analysis of Eap1p and Puf5p complexes may illuminate this aspect.
The discovery that Eap1p is necessary for Puf5p repression suggested a mechanism based on the existing model of 4E-BP molecular function: Puf5p recruitment of Eap1p may block translation initiation, thereby reducing loading of ribosomes onto target mRNAs. If accurate, then HO mRNA should inefficiently associate with ribosomes, resulting in accumulation of HO mRNA in the ribosome-free fractions of sucrose gradients. Our findings contradict this hypothesis. First, nearly all HO mRNA associates with polyribosomes in wild-type cells, indicating efficient translation (Fig. 2). HO mRNA remains associated with polyribosomes when EAP1 is deleted. Instead of the predicted increase in ribosome density, eap1 deletion slightly reduced the density of polyribosome-associated HO mRNA. These findings argue against Eap1p-mediated translational inhibition of HO. The predominant effect of eap1 deletion was to increase the total amount of HO mRNA and increase its half-life, pointing toward a role of Eap1p in promoting mRNA decay. The small reduction of HO ribosome density caused by the absence of Eap1p may reflect the accumulation of HO mRNA with short poly(A) tails in the eap1 deletion strain (Fig. 5).
If Eap1p inhibits translation initiation, then Eap1p would be predicted to be found in ribosome-free fractions at the top of the sucrose gradient. In contrast to this, a significant portion of Eap1p associates with polyribosomes (Fig. 3). Because polyribosomal mRNAs have undergone multiple rounds of initiation, polyribosome-associated Eap1p cannot have blocked initiation. Our data indicate that interaction of Eap1p with eIF4E mediates its association with polyribosomes. This conclusion is supported by the observation that eIF4E was distributed across the gradient, including polyribosome fractions (Fig. 3). Furthermore, mutations in Eap1p that disrupt binding to eIF4E cause a complete loss of Eap1p polyribosome association (Fig. 3). At this time, the functional significance of polyribosomal Eap1p remains unknown.
Eap1p is also present in two other peaks of the sucrose gradient. One peak, in the ribosome-free fractions at the top of the gradient, increases upon EDTA treatment, ribosome runoff, and mutation of the eIF4E binding motif (Fig. 3). These observations indicate that this pool of Eap1p is not bound to ribosomes or eIF4E. An Eap1p species with lower electrophoretic mobility is present in this peak, suggesting that Eap1p may be modified posttranslationally. This idea is supported by proteomic identification of multiple phosphorylation sites in Eap1p (56). Eap1p phosphorylation could block its interaction with eIF4E in a manner similar to other 4E-BPs (62). A third peak of Eap1p cofractionates with 60S ribosomal subunits. This peak shifts to the ribosome-free fractions upon ribosome runoff and mutation of the eIF4E binding motif; therefore, it might represent an intermediate in translation. Future biochemical analysis of Eap1p complexes will be necessary to understand their composition and functions.
Deletion of EAP1 does not alter the global translation state, as deduced from the identical chromatograms of polyribosomes from wild-type and eap1 deletion strains (Fig. 2) (32). This conclusion is supported by analysis of ribosome association of RNR1 and RPL41A mRNAs, which like HO, were not inhibited by Eap1p. That said, on the basis of the available data, we do not exclude the prospect that specific mRNAs may be translationally inhibited by Eap1p. In the study that identified Eap1p, an in vitro translation assay indicated that Eap1p inhibits translation (14); however, the role of eIF4E binding and impact on mRNA stability were not addressed. More recently, a microarray-based study found that deletion of EAP1 changed the translation state of 329 mRNAs by greater than 1.8-fold, as measured by the ratio of mRNA in polyribosomes to monoribosomes (15). Of these, the ratios of 176 mRNAs increased, suggesting that Eap1p inhibits their translation. In our analysis of HO mRNA, deletion of EAP1 caused a 1.6-fold increase in the polyribosome-to-monoribosome ratio; however, this change reflects the 1.7-fold increase in HO abundance. Moreover, eap1 deletion did not increase HO ribosome occupancy and density (Fig. 2). It remains to be investigated whether Eap1p and Puf5p function together to regulate additional mRNAs. The mRNAs affected by eap1 deletion did not correlate with those that coimmunoprecipitate with Puf5p (15). Cridge et al. also reported that deletion of EAP1 altered steady-state levels of 99 mRNAs more than 2-fold, and of these, 56 increased, hinting that Eap1p may affect the stability of additional mRNAs (15). Whether the observed effects of eap1 deletion on translation state and mRNA levels are direct remains to be established. Germane to the challenge of discerning direct Eap1p effects from indirect effects, our demonstration that an Eap1p-regulated mRNA (i.e., HO) coimmunoprecipitates with Eap1p provides proof of principle for future ribonomic approaches to identify Eap1p target mRNAs.
In summary, our findings do not support a role of Eap1p in inhibition of translation initiation of Puf5p-targeted HO mRNA. Instead, our results argue that Eap1p elicits a mode of repression that is divergent from the canonical 4E-BP mechanism of translation inhibition.
We discovered a novel function for Eap1p in promoting mRNA degradation. Deletion of EAP1 stabilizes HO mRNA more than 3-fold. Conversely, overexpression of Eap1p enhances decay of HO mRNA beyond that observed in wild-type cells. Targeting of HO by Eap1p is likely directed by Puf5p, a hypothesis supported by their mutual interdependence for repression (Fig. 1) and the finding that association of Eap1p with HO mRNA depends on Puf5p (Fig. 7).
Enhancement of mRNA decay by Eap1p is not likely to be solely dependent on Puf5p, because we also observed an effect on RNR1 mRNA, which is not regulated by Puf5p (23, 60) and is not known to associate with other PUF proteins (22). The factors that control RNR1 mRNA remain to be discovered in future work. Other mRNAs, such as RPL41A, are unaffected by Eap1p, suggesting that Eap1p-enhanced mRNA decay may be restricted to specific messages.
Is the eIF4E binding activity of Eap1p necessary for mRNA degradation? An Eap1p mutant that cannot bind eIF4E exhibits half of the mRNA decay activity of the wild-type protein (Fig. 4); therefore, we conclude that binding to eIF4E facilitates, but is not essential for, Eap1p mRNA decay activity. Binding to eIF4E could promote decay by displacing eIF4G (14), destabilizing the eIF4E-5′ cap interaction and facilitating access of mRNA degradation enzymes (Fig. 8). This idea is supported by data showing that mutations in translation initiation factors, including eIF4E and eIF4G, increase mRNA decay (58). However, our data also indicate that this explanation in itself is not sufficient. First, the Eap1p mutant does not bind eIF4E (Fig. 3), nor does it associate with polyribosomes (Fig. 3), yet it still stimulates mRNA decay (Fig. 4), albeit with reduced efficiency. We interpret this as evidence that Eap1p promotes mRNA decay by another means, perhaps by affecting decapping (Fig. 8), as discussed below.
Degradation of yeast mRNAs typically initiates by shortening of the poly(A) tail to an oligo-adenylated length, followed by decapping (11, 25). Decapped mRNA is then rapidly degraded by Xrn1p. As Puf5p enhances deadenylation of HO mRNA, we examined the effect of Eap1p. Loss of Eap1p does not affect poly(A) removal, which occurs rapidly in both wild-type and eap1 deletion cells (Fig. 5). Instead, Eap1p dramatically affects the fate of the oligo-adenylated intermediate. In wild-type cells, this species is rapidly degraded, coincident with disappearance of the 5′ end of the mRNA (Fig. 5). When EAP1 is deleted, the oligo-adenylated mRNA is highly stabilized, as is the 5′ end of HO (Fig. 5). The oligo-adenylated HO that accumulates is resistant to Xrn1p and thus remains capped (Fig. 5). We conclude that Eap1p promotes decapping of HO mRNA. Our data support that HO is degraded by the 5′ decapping pathway; deletion of decapping factor genes greatly stabilizes HO mRNA, including the Δpat1 (t1/2 = 55 min, and Δdcp2 (t1/2 > 60 min) strains (unpublished data) and the Δdhh1 strain (t1/2 > 50 min) (see Fig. S3 in the supplemental material). Additionally, accumulation of deadenylated, capped mRNA species is identical to the effect of mutations in decapping factors, including Dhh1p (see Fig. S3), Pat1p, and the Lsm1-7p complex (4, 13, 19, 66). We note that deletion of genes encoding the 3′-to-5′ exosome complex did not increase HO mRNA levels (see Fig. S4 in the supplemental material) or abrogate Puf5p repression (23); therefore, the 3′ decay pathway does not impact HO mRNA degradation.
Eap1p associates with the decapping factor Dhh1p. This finding suggests that Eap1p recruits decapping factors or alters their activity (Fig. 8). Puf5p also associates with decapping factors, including Dhh1p and Dcp1p (23). These associations do not depend on RNA, indicating that the proteins were not simply tethered to the same mRNA, though the direct protein contacts remain to be delineated.
We propose a model of posttranscriptional regulation of HO mRNA that integrates past and new data (Fig. 8). Both Puf4p and Puf5p bind to their respective recognition sequence in the 3′UTR of HO and recruit the Ccr4-Not complex via direct contact with the Pop2p subunit, thereby enhancing deadenylation. Puf5p also recruits Eap1p and decapping factor Dhh1p and decapping enzyme Dcp2-Dcp1 to promote removal of the 5′ cap. Eap1p binding to eIF4E facilitates decay. This model provides a useful framework for future research on PUF repression.
Recent evidence indicates that mRNA decay can occur cotranslationally on polyribosomes (30, 31). It is tempting to speculate that decay of HO mRNA, promoted by Eap1p, might occur on polyribosomes; however, because eIF4E binding-defective Eap1p retains partial activity but does not associate with polyribosomes, ribosome association is unlikely to be an essential feature.
Our analysis does not exclude the possibility that PUFs block translation by additional mechanisms. Chritton et al. reported that the Puf5p inhibits translation of a capped, polyadenylated reporter mRNA in vitro (10). Eap1-mediated decapping may account for the observed regulation. Alternatively, Puf5p may have an independent direct effect on translation. Dhh1p can inhibit translation (12); therefore, Puf5p recruitment of Dhh1p may impact translation.
Multiple 4E-BPs have been identified in eukaryotes (1, 14, 18, 35, 42, 43, 49, 50, 54, 55, 64, 74). The finding that Eap1p is required for repression by Puf5p is reminiscent of other examples wherein an RNA-binding protein utilizes a 4E-BP to repress an mRNA (49, 50, 62, 64, 74). In these cases, translation inhibition is thought to be the means of repression. Our results showing that Eap1p enhanced decay and decapping were therefore unanticipated in the context of current understanding of 4E-BP repression. Furthermore, Puf5p may not be the only RNA binding factor to use Eap1p as a corepressor, as the yeast Vts1p protein employs Eap1p to enhance mRNA decay (C. Smibert, University of Toronto, personal communication).
Additional clues are emerging that implicate specific 4E-BPs as versatile regulators that can influence steps of gene expression other than translation initiation. The human eIF4E transporter was implicated in the destabilization of adenine uracil-rich element (ARE)-containing mRNAs (18). A recent analysis of the Drosophila 4E-BP, Cup, found that it represses mRNAs by specifically enhancing deadenylation (33). Interestingly, Cup subsequently stabilizes the deadenylated mRNA by blocking decapping (33). These findings suggest the exciting possibility that each 4E-BP has unique activities to control translation, localization, and degradation of distinct groups of mRNAs.
A.C.G. gratefully acknowledges support by Marvin Wickens, University of Wisconsin, during the initial phase of this research. We appreciate advice from Trista Schagat, Jamie Van Etten, Chase Weidmann, Nathan Raynard, and Joel Hrit. We thank Brad Hook, Daniel Seay, Jeff Coller, and May Tsoi for technical assistance and Eric Wagner for comments on the manuscript. Pfizer generously supplied thiolutin.
Nathan Blewett was supported by NIH Cellular and Molecular Biology training grant T32-GM007315.
Published ahead of print 13 August 2012
Supplemental material for this article may be found at http://mcb.asm.org/.