Two Yeast PUF Proteins Negatively Regulate a Single mRNA*

mRNA stability and translation are regulated by protein repressors that bind 3′-untranslated regions. PUF proteins provide a paradigm for these regulatory molecules: like other repressors, they inhibit translation, enhance mRNA decay, and promote poly(A) removal. Here we show that a single mRNA in Saccharomyces cerevisiae, encoding the HO endonuclease, is regulated by two distinct PUF proteins, Puf4p and Mpt5p. These proteins bind to adjacent sites and can co-occupy the mRNA. Both proteins are required for full repression and deadenylation in vivo; their removal dramatically stabilizes the mRNA. The two proteins act through overlapping but non-identical mechanisms: repression by Puf4p is dependent on deadenylation, whereas repression by Mpt5p can occur through additional mechanisms. Combinatorial action of the two regulatory proteins may allow responses to specific environmental cues and be common in 3′-untranslated region-mediated control.

The PUF proteins are an evolutionarily widespread family of 3Ј-UTR-binding proteins (13). All PUF proteins examined to date either repress translation, induce decay, or both; their activity is correlated with poly(A) shortening (13). The yeast Saccharomyces cerevisiae possesses six PUF proteins (13). Each PUF is associated with many mRNAs in yeast lysates, suggesting that nearly 10 -15% of yeast mRNAs may be targets of PUF regulation (14).
Mpt5p, one of the yeast PUF proteins, binds to and regulates the mRNA that encodes HO endonuclease (7,15). HO protein controls mating-type switching by causing double-strand breaks that initiate a recombination event (16). The HO gene is tightly controlled transcriptionally in a cell type-specific manner to prevent aberrant switching and deleterious DNA breaks (17). Recently it has become clear that HO is also controlled post-transcriptionally (7,15,18). Mpt5p contributes to this control; indeed, mpt5 null mutants switch mating type aberrantly at high frequency (15). The observation that HO mRNA co-immunoprecipitated with another PUF protein, Puf4p, suggested that Puf4p might also participate in HO control (14).
PUF proteins from several organisms directly interact with Pop2p orthologs (7). In many organisms, Pop2p is a deadenylase, as is a related protein with which it interacts, Ccr4p (7, 19 -21). In yeast, Pop2p is required for repression and deadenylation mediated by Mpt5p; however, point mutations that disrupt its active site have no effect on PUF-enhanced deadenylation in vivo or in vitro (7,18). Instead, the Ccr4p enzyme catalyzes PUF-enhanced deadenylation; mutations in the Ccr4p active site eliminate regulated deadenylation (18). Pop2p is a bridge that links the PUF-RNA complex to the Ccr4p deadenylase. The same PUF-Pop2p connection simultaneously recruits other proteins involved in turnover and translational repression, including the decapping factor Dcp1p and the RNA helicase Dhh1p, which can repress translation (7).
In this report, we sought to determine whether Puf4p directly regulates HO expression. We combined in vivo and in vitro analyses to show that Puf4p and Mpt5p bind the HO 3Ј-UTR through distinct but adjacent binding sites and can do so simultaneously. Puf4p represses mRNA expression and enhances mRNA deadenylation and decay. The mechanisms of these effects overlap with, but are distinct from, those of Mpt5p. In particular, Puf4p repression activity is entirely dependent on the deadenylation activity of Ccr4p (which it recruits via Pop2p); in contrast, Mpt5p can repress by additional mechanisms. In the presence of either PUF protein, the Pop2p-Ccr4p complex is recruited to enhance deadenylation. The effects of the two proteins bound simultaneously to their cognate sites are additive and greatly enhance the efficiency and rate of deadenylation and decay. Together, they provide flexible opportunities for control.

EXPERIMENTAL PROCEDURES
Strains-The yeast three-hybrid strain YBZ1 was described previously (22). The wild-type BY4742 yeast strain and isogenic strains with gene-specific deletions of CCR4 and POP2 were obtained from Open Biosystems. These deletion strains were created by PCR-mediated gene modification using the kanamycin/G418 resistance marker. The MPT5-TAP, PUF4-TAP, and POP2-TAP strains (Open Biosystems) were created in the S288C strain background by integrating a C-terminal tandem affinity purification (TAP) tag onto the coding sequence using PCR-mediated gene modification with a HIS3 marker.
Yeast Three-hybrid Assay-Three-hybrid assays were performed as described (23). PUF proteins fused with the Gal4 activation domain were expressed from pACT2 or pGADT7 plasmids. Full-length open reading frames were used for PUF1, PUF2, PUF3, PUF4, and PUF6. The Mpt5p open reading frame starts with methionine 26. DNA oligonucleotides were designed to express either the Puf4 BS or Mpt5 BS (sequence found in Fig. 1B) and cloned into the XmaI and SphI sites of pIIIA/MS2-2. Assays were performed in strain YBZ1. ␤-galactosidase activity was measured using the Beta-Glo (Promega) substrate as described (22).
Pop2p was cloned into a derivative of the pMAL vector (New England Biolabs) that contained a TEV protease cleavage site and an N-terminal hexahistidine epitope tag. The resulting maltose-binding protein (MBP)-Pop2p was purified from E. coli according to the manufacturer's protocol (New England Biolabs).
Electrophoretic Mobility Shift Assays-RNA binding assays were performed as described (24) using chemically synthesized RNAs (Dharmacon). The apparent K d , the concentration of protein at which half-maximal binding occurs, was calculated using Grafit 5 (Sigma). The modest difference in Mpt5p binding affinities between Fig.  1C and that reported in Fig. 2A are likely due to different protein preparations. The data and apparent dissociation constants reported in Fig. 2A are the statistically determined results of three independent binding experiments.
mRNA Decay Analysis-Transcriptional inhibition experiments were done using 20 g ml Ϫ1 of thiolutin (Pfizer) as described (7). RNA was purified by an acid-phenol-chloroform protocol at pH 4.6 and separated by 3.7% (v/v) formaldehyde, 1.0% (w/v) agarose gel electrophoresis. Transfer to nylon membrane (Ambion) was achieved by capillary action using 10ϫ SSC. Blots were probed with radiolabeled antisense riboprobes specific to genes listed. Repression Assays-The HIS3-HO 3Ј-UTR reporter gene contains the HIS3 coding sequence with 2 kbp of the HO promoter upstream and 1 kbp of the HO 3Ј-UTR downstream in plasmid YCp33. The "HO mt" reporter ( Fig. 4) has the UGU trinucleotides in the Puf4 and Mpt5 binding sites changed to ACA by site-directed mutagenesis of an otherwise wild-type HO 3Ј-UTR (sequence in Fig. 1). Wild-type yeast strain BY4742 or gene-specific deletion strains (ccr4 or pop2) were transformed with the reporter gene plasmid and either YEp181, YEp181 MPT5, p415-GPD, or p415-GPD-PUF4. Plasmid YEp181 MPT5 mutant was created by site-directed mutagenesis (Stratagene) of amino acids S454A and N455A in PUF repeat 7. Plasmid p415-GPD-PUF4 mutant was derived from p415-GPD-PUF4 by site-directed mutagenesis (Stratagene) of amino acids S799A and N800A in PUF repeat 7. CCR4 expression plasmid was created in the high copy vector pACG1-NT as described (18). The active site mutant CCR4 contains missense mutation E556A described by Chen et al. (19). For growth assay, colonies were isolated and grown to mid-log phase at 30°C and plated on minimal medium with or without histidine. The HIS3 competitive inhibitor 3-aminotriazole was added to increase stringency where indicated.
Immunoprecipitations-Reactions were performed with TAP-tagged MPT5 and PUF4 strains carrying T7-tagged versions of Pop2p, Ccr4p, or green fluorescent protein expressed from vector pACG1 NTB as described (7). Lysates and pellets were treated with 5 units of RNase A and 200 units of RNase T1 (Ambion).
Deadenylation in Vitro-Deadenylation reactions were carried out with Pop2p-TAP complexes purified from yeast and recombinant PUF protein purified from E. coli as previously described (7). The reactions contained 10 ng of Pop2p-TAP

RESULTS
Two Different PUF Proteins Bind the HO 3Ј-UTR-The yeast PUF protein Mpt5p regulates HO mRNA by binding a specific sequence in its 3Ј-UTR, called the Mpt5 BS (Fig. 1A, and Refs. 7 and 15). A second potential PUF binding site lies 20 nucleotides upstream and conforms to a sequence found in 66% of mRNAs that co-immunoprecipitated with Puf4p (14). These data suggested that both Mpt5p and Puf4p might regulate HO mRNA.
We first examined the binding of Mpt5p and Puf4p to their respective sites using the yeast three-hybrid system (22,23). Each of the six yeast PUF proteins was individually fused to the Gal4p transcriptional activation domain and co-expressed with RNA "baits" containing either the predicted Puf4p binding site (Puf4 BS) or the Mpt5p binding site (Mpt5 BS). Protein-RNA interactions were detected by the production of ␤-galactosidase and assayed in yeast lysates (22). Only Puf4p bound the Puf4 BS (Fig. 1B), whereas only Mpt5p bound the Mpt5 BS (Fig. 1B). Mutations that changed UGU to ACA in each binding site eliminated binding (data not shown; see Fig. 1C).
To determine whether Puf4p and Mpt5p bind directly to the two sites, we used an electrophoretic mobility shift assay. Purified, recombinant GST-tagged Puf4p or Mpt5p was incubated with labeled RNAs comprising either the Mpt5 BS or Puf4 BS, and the resulting complexes were analyzed by electrophoresis (Fig. 1C). Mpt5p bound the Mpt5 BS with an apparent K d of 100 nM, consistent with our previous findings (7), whereas Puf4p bound the Puf4 BS with an apparent K d of 10 nM. Neither Mpt5p nor Puf4p bound its respective binding site when the critical UGU trinucleotides were mutated to ACA (Fig. 1C). We conclude that Puf4p specifically binds the newly defined Puf4 BS in the HO 3Ј-UTR.
Both PUF Proteins Can Bind Simultaneously-To quantify binding specificity, we compared the binding of Puf4p and Mpt5p to the two sites in vitro. Puf4p exhibited a 25-fold preference for its cognate site over its non-cognate site; Mpt5p displayed a 3.5-fold preference for its own site ( Fig. 2A).
To test whether Puf4p and Mpt5p can bind simultaneously to a single RNA molecule, we prepared a 5Ј-end-labeled RNA containing both the Puf4 and the Mpt5 binding sites (Fig. 2B). To test for co-occupancy, we performed gel shift assays using a saturating concentration of one recombinant PUF protein mixed with increasing concentrations of the other. To resolve the different complexes, the GST affinity tag was removed from Puf4p but left on Mpt5p; as a result, their complexes differed in mobility.
We next performed the reciprocal experiment, holding Mpt5p constant and titrating Puf4p. Mpt5p was added at saturating concentrations (500 nM, 5-fold above its K d for the Mpt5 BS); as expected, it bound the Mpt5 site to form Complex 4 and both sites to form Complex 6 (Fig. 2B, lane 9). Addition of Puf4p yielded Complex 5, containing Mpt5p, Puf4p, and the RNA (Fig. 2B, lanes 10 -15). At high Puf4p concentrations, Puf4p competed with Mpt5p, ultimately displacing it from the Puf4p BS; as a result, Complexes 2 and 3 increase in abundance at the expense of Complexes 4, 5, and 6 ( Fig. 2B, lanes 14 and 15). Mutation of the Puf4 BS from UGU to ACA abolishes all complexes except that of Mpt5p to the Mpt5p site (Complex 4). These data demonstrate that Mpt5p and Puf4p preferentially bind their respective sites, that they can co-occupy the HO 3Ј-UTR in vitro, and that they can compete for one another's binding sites.
Puf4p and Mpt5p Stimulate Degradation of HO mRNA-We analyzed the effect of Puf4p and Mpt5p on endogenous HO mRNA by comparing its half-life in wild-type versus puf4, mpt5, and puf4 mpt5 deletion strains (Fig. 3). Total RNA was prepared at various times after addition of thiolutin, which inhibits RNA polymerase II (25). HO mRNA half-life increased from 9 min in the wild-type strain, to 19 min in the puf4 mutant, and to 35 min in the mpt5 mutant. Most dramatically, the half-life of HO mRNA increased to more than 90 min in the puf4 mpt5 mutant strain (Fig. 3, A and B). Steady state levels of HO mRNA increased dramatically in the mpt5 (3-fold) and puf4 mpt5 mutants (4-fold). As a control, we analyzed an mRNA that does not interact with Mpt5p. The half-lives of neither ACT1 mRNA (Fig. 3C) nor SCR1 RNA (Fig. 3D) were unaffected by removal of PUF4, MPT5, or both. These results demonstrate that Puf4p binding site to ACA is indicated by an "X". B, Puf4p repression assays were performed using wild-type or mutant HO reporter genes with plasmids expressing wild-type or mutant PUF4 proteins or empty vector as indicated on the left. Yeast cells, the number of which is shown at the top, were spotted on media with (control) or without (ϪHis) histidine. 3-Aminotriazole was added to medium lacking histidine where indicated. C, Puf4p repression of the wild-type HO reporter was tested in wild-type (WT), pop2, and ccr4 yeast strains as indicated on the left. D, Puf4p repression of wild-type HO reporter gene in wild-type yeast carrying empty vector as a control or overexpressing wild-type CCR4 or catalytically inactive mutant CCR4 (CCR4 mt). E, Puf4p repression of wild-type HO reporter gene in ccr4 deletion strain carrying empty vector as a control or overexpressing wild-type CCR4 or catalytically inactive mutant CCR4 (CCR4 mt). and Mpt5p specifically destabilize HO mRNA and that the two proteins combine to enhance destabilization.
Puf4p Represses HO in Vivo and Requires Pop2p and Ccr4p-To test whether Puf4p represses expression in vivo, we used a reporter gene consisting of the HO promoter, HIS3 coding sequence, and the HO 3Ј-UTR (Fig. 4A) (7). When overexpressed, Puf4p repressed this reporter mRNA (Fig. 4B), largely preventing growth on selective media lacking histidine; Mpt5p behaves similarly (supplemental Fig. S1A) (7). Repression by Puf4p required the PUF binding sites in the RNA (Fig. 4B) and a functional RNA-binding domain in the protein (Fig. 4B). We conclude that Puf4p and Mpt5p repress the reporter mRNA in vivo by binding the HO 3Ј-UTR.
Puf4p-mediated repression also required Ccr4p, the enzyme responsible for deadenylation of HO mRNA in vivo (18); repression was virtually abolished in a ccr4 mutant (Fig. 4C). In contrast, Mpt5p-mediated repression largely persists in the ccr4 mutant (supplemental Fig. S1B) (7). We conclude that Puf4p repression via the HO 3Ј-UTR requires both the Pop2p and Ccr4p subunits of the major cytoplasmic deadenylase complex.
To determine whether repression by Puf4p requires the catalytic activity of Ccr4p, we tested Puf4p-dependent repression in wild-type and ccr4 deletion strains expressing either wild-type Ccr4p or a point mutant that disrupts its active site (Fig. 4, D and  E, CCR4 mt). In the wild-type strain, Puf4p represses with and without overexpressed Ccr4p, as expected (Fig. 4D). Overexpression of the Ccr4p mutant partially inhibited repression, indicating a dominant negative effect (Fig. 4D). In the ccr4 mutant strain, Puf4p repression was restored by expression of active Ccr4p, but not by the Ccr4p active site mutant (Fig. 4E). We conclude that Puf4p repression requires both Pop2p and an active form of Ccr4p.
Puf4p Associates with Deadenylation Factors-To determine whether Puf4p physically associates with deadenylase proteins, we performed co-immunoprecipitation experiments with yeast extracts (Fig. 5A). We used yeast strains expressing either TAPtagged Puf4p or TAP-tagged Mpt5p and a T7 epitope-tagged derivative of either the deadenylase subunits Ccr4p or Pop2p or, as a control, green fluorescent protein. Mock purifications from a wild-type, untagged strain carrying each T7-tagged test protein served as negative controls.
TAP-Puf4p and TAP-Mpt5p were affinity-purified and eluted from the beads, and the immunoprecipitates were analyzed by Western blotting. Pop2p and Ccr4p were physically associated with Puf4p and Mpt5p (Fig. 5A, lanes 5, 6, 8, and 9); green fluorescent protein was not (lanes 2 and 3). The co-immunoprecipitations of Pop2p and Ccr4p with Puf4p and Mpt5p were due to protein-protein interactions rather than co-occupancy of RNA, because they were unaffected by RNase A and T1 treatment.
Puf4p Directly Binds Pop2p-To determine whether Puf4p bound directly to Pop2p, we used purified recombinant Pop2p fused to maltose-binding protein (MBP-Pop2p). MBP-Pop2p was incubated with purified recombinant Puf4p, Mpt5p, and Ccr4p, each fused to GST and immobilized on glutathioneagarose beads. After extensive washing, the presence of MBP-Pop2p was determined using Western blotting. Immobilized Puf4p and Mpt5p bound MBP-Pop2p (Fig. 5B). As controls, Ccr4p bound Pop2p as expected, whereas GST alone did not. We conclude that Puf4p, like Mpt5p, binds directly to Pop2p. Because these experiments used only the RNA-binding domains of the PUF proteins, those regions are sufficient to bind Pop2p.
Puf4p Stimulates Deadenylation in Vitro-Puf4p repression requires both CCR4 and POP2, implying that Puf4p repression stimulates deadenylation. Consistent with this, deadenylation of HO mRNA is dramatically impaired in puf4 mpt5 double mutants compared with mpt5 alone (7). To study Puf4p effect on poly(A) shortening in detail, we examined Puf4p-stimulated deadenylation in vitro. Recombinant GST-Puf4p was purified from E. coli. As a substrate, we created a 5Ј-end-labeled HO 3Ј-UTR RNA containing the Puf4 BS with fourteen adenosine residues at its 3Ј-end (sequences in Fig. 6A). As expected, Puf4p bound to the HO 3Ј-UTR substrate with high affinity (apparent K d of 15 nM), and mutation of the UGU trinucleotide completely abrogated the interaction (Fig. 6A).
Pop2p complexes were purified from yeast using TAP tag affinity purification; both Pop2p and Ccr4p were enriched in this partially purified fraction as judged by Western blotting (7). The Pop2p complex alone removed the poly(A) tail in a concentration-and time-dependent manner and was capable of completely deadenylating the RNA at high concentrations (supplemental Fig. S2). Puf4p alone neither deadenylated nor destroyed the RNA (Fig. 6B, lanes 3-5). The Pop2p complex, added at low concentrations, showed modest deadenylation activity; 5% of the RNA was fully deadenylated after 120 min (lanes 6 -9). Addition of Puf4p to the Pop2p complex resulted in rapid deadenylation; fully deadenylated RNA was observed even at 30 min and comprised 85% after 120 min (lanes 10 -13). This Puf4p-dependent deadenylation was sequence-specific; a mutation in the HO RNA substrate that abolished Puf4p binding eliminated enhanced deadenylation by the two mixed components but did not affect the activity of the Pop2p complex alone (lanes 18 -21). In conclusion, Puf4p can stimulate deadenylation of an RNA substrate by directly recruiting the Pop2p deadenylase complex.
Puf4p and Mpt5p Collaborate to Greatly Accelerate Deadenylation in Vitro-To examine the combined affects of Puf4p and Mpt5p on the Pop2p complex, we performed an in vitro deadenylation assay in the presence of both PUF proteins with an RNA substrate containing both the Puf4 and Mpt5 binding sites (Fig. 7A). We used a low concentration of Pop2p complex to yield only slow deadenylation (Fig.  7A, lanes 3-5, and panel B). Each PUF protein stimulated deadenylation (Fig. 7A, lanes 6 -15). In the presence of both PUF proteins simultaneously, deadenylation was enhanced even more (Fig. 7A, lanes 16 -20). For example, after 15 min the Pop2p complex alone fully deadenylated only 4% of the substrate RNA (Fig. 7A, lane 3), and Puf4p and Mpt5p individually yielded fully deadenylated 34 and 58% of the substrate RNA, respectively (Fig. 7A, lanes 8 and 13); at the same time point, the two PUFs mixed together deadenylated virtually all of the substrate and in fact progressively degraded the body of the RNA (Fig. 7A, lane 18). The mixed PUF proteins alone had no effect on the substrate in the absence of deadenylase complex (Fig.  7A, lanes 21-25). We conclude that Puf4p and Mpt5p collaborate to accelerate poly(A) removal by recruiting the Pop2p deadenylase.

DISCUSSION
Our results permit three main conclusions, summarized in Fig. 8. First, two PUF proteins, Puf4p and Mpt5p, collaborate to decrease expression of a single mRNA. The additive effects of regulation are seen both in vivo and in vitro and result from interactions with adjacent binding sites in the 3Ј-UTR. Second, the mechanism of repression by Puf4p involves direct binding to Pop2p, recruitment of the Pop2p-Ccr4p complex, and the enzymatic activity of the Ccr4p deadenylase. Third, the activities of the two PUF proteins differ: Puf4p, but not Mpt5p, strictly requires deadenylation.
Is repression by the two PUF proteins entirely a consequence of deadenylation? Both Puf4p and Mpt5p enhance deadenylation of HO mRNA in vivo and in vitro (this report and Refs. 7 and 18). The conclusion that the modes of action of the two PUF proteins differ is based on the fact that repression via Puf4p requires both Pop2p and Ccr4p whereas Mpt5p stringently requires only Pop2p. Importantly, repression via Puf4p is abolished by mutations in the active site of Ccr4p, strongly suggesting that deadenylation itself is essential for Puf4p function. Mpt5p also requires Pop2p for repression but, in contrast to Puf4p, Ccr4p is not critical for Mpt5p-mediated repression (this report and Refs. 7,18). Yet deletion of Ccr4p stabilizes HO mRNA, effectively blocking deadenylation (18). Thus, in ccr4 mutants, fully adenylated mRNA accumulates and can be repressed by overexpressed Mpt5p. These findings demonstrate that Mpt5p elicits an additional, second form of repression that is deadenylation-independent (Fig. 8). The simplest interpretation is that the complexes recruited by the two PUF proteins differ. Partners of the PUF proteins functionally analogous to Nanos, Brat, GLD-3, and CPEB in metazoa may be critical (13).
Puf4p and Mpt5p act in concert to ensure that HO mRNA is efficiently deadenylated and repressed. Indeed, simultaneous mutation of both PUF binding sites abolishes regulation (Fig.  4B). Dual regulation provides not only efficient control but new regulatory opportunities as well; differential expression or regulation of the two PUF proteins under specific conditions could selectively enhance or depress HO expression.
Combinatorial control by PUF proteins may be common. The 3Ј-UTRs of hunchback mRNA, a target of Drosophila Pumilio, and of COX17 mRNA, a target of yeast Puf3p, both possess multiple binding sites that are important for repression in vivo (8,26); gld-1 mRNA, a target of FBF in Caenorhabditis elegans, possesses two FBF binding sites in its 3Ј-UTR (27). Twenty-eight mRNAs co-immunoprecipitate separately with Puf4p or Mpt5p; however, each of these mRNAs appears to contain a single PUF protein binding site, implying competition between the two proteins (14). Co-regulation of mRNAs by multiple PUF proteins may be common; combinations of other PUFs regulate several other mRNAs. 3 Regulation of HO by Puf4p and Mpt5p demonstrates that two different PUF proteins act on a single mRNA in vivo. Complexes formed via a regulatory protein's interaction with a binding site in the 3Ј-UTR are common and a hallmark of 3Ј-UTR control (28). What is different here is that the different proteins bind two different sites to achieve full repression. The magnitude of repression achieved by the two proteins together (more than an order of magnitude change in mRNA half-life) is large, and greater than that achieved by either protein alone. 3 W. Olivas, personal communication. Puf4p and Mpt5p bind to their cognate sites with different affinities and with very different preferences for cognate over non-cognate sites (25versus 3.5-fold for Puf4p and Mpt5p, respectively ( Fig. 2A). Although it is difficult to accurately measure the low in vivo levels of Mpt5p and Puf4p, their intracellular concentrations were estimated to be ϳ29 and 17 nM, respectively (29). In this range, the cognate site of each protein would be partially occupied during standard growth conditions and neither protein would be expected to occupy the other's site. This situation provides robust opportunities for control. Under certain circumstances, one PUF protein might occupy only its own site, and in another, occupy both. Regulation could be influenced via specific protein partners as are involved in PUF regulation in C. elegans (NOS-3, GLD-3, and CPB-1, see Refs. 30 -32) and Drosophila (NOS and Brat, see Refs. 33,34). In addition, cell cycle and mitogen-activated protein kinases, with which Mpt5p interacts physically and genetically, could modulate repression activity (35). Importantly, the mechanism of repression will depend on which PUF is bound, providing additional versatility.
In many instances of nucleic acid polymerization or removal, the protein that possesses enzymatic activity is not intrinsically specific for a particular RNA or DNA template but gains that specificity by interacting with a separate protein. Our results show that Puf4p and Mpt5p provide sequence specificity to Ccr4p deadenylation activity. In their absence, the Ccr4p enzyme, as part of the Pop2p-Ccr4p complex, inefficiently deadenylates substrates independent of their sequence; in their presence, deadenylation becomes rapid and sequence-specific. It will be necessary to purify the relevant deadenylation complex to homogeneity to identify any essential components other than Pop2p and Ccr4p.
Deadenylation is the rate-determining step in mRNA decay, and deadenylation rates vary widely among mRNAs. Regulators such as RNA-binding proteins, and more recently small RNAs, can modulate deadenylation rates (36 -38). In principle, enhancement of deadenylation can be achieved either by increasing the catalytic rate of the enzyme or by increasing its rate of association with the substrate. PUF proteins almost certainly do the latter, because their effect on deadenylation is entirely dependent on the presence of a cognate binding sequence in the RNA (this study; Fig. 6B; Ref. 7). Once bound by a PUF protein, an RNA molecule is rapidly deadenylated. Pab1p, which is bound to poly(A) tails in vivo, may also influence the character of deadenylation by Ccr4p in vivo.
The off-rates of PUF proteins from RNA and from Pop2p are predicted to determine the rate and extent of deadenylation in vivo. Because repression by Puf4p strictly requires deadenylation, these parameters will also determine the efficiency of repression.