|Home | About | Journals | Submit | Contact Us | Français|
Recent experiments have defined cytoplasmic foci, referred to as processing bodies (P-bodies), that contain untranslating mRNAs in conjunction with proteins involved in translation repression and mRNA decapping and degradation. However, the order of protein assembly into P-bodies and the interactions that promote P-body assembly are unknown. To gain insight into how yeast P-bodies assemble, we examined the P-body accumulation of Dcp1p, Dcp2p, Edc3p, Dhh1p, Pat1p, Lsm1p, Xrn1p, Ccr4p, and Pop2p in deletion mutants lacking one or more P-body component. These experiments revealed that Dcp2p and Pat1p are required for recruitment of Dcp1p and of the Lsm1-7p complex to P-bodies, respectively. We also demonstrate that P-body assembly is redundant and no single known component of P-bodies is required for P-body assembly, although both Dcp2p and Pat1p contribute to P-body assembly. In addition, our results indicate that Pat1p can be a nuclear-cytoplasmic shuttling protein and acts early in P-body assembly. In contrast, the Lsm1-7p complex appears to primarily function in a rate limiting step after P-body assembly in triggering decapping. Taken together, these results provide insight both into the function of individual proteins involved in mRNA degradation and the mechanisms by which yeast P-bodies assemble.
The regulation of mRNA translation and degradation is an important aspect of the control of eukaryotic gene expression. In eukaryotes, two major decay pathways initiate with deadenylation of the 3′ polyadenosine [poly(A)] tail, with the predominant cytoplasmic deadenylase being the Ccr4p/Pop2p/Not1-5p complex (reviewed in Meyer et al., 2004 ; Parker and Song, 2004 ). Deadenylation can lead to 3′ to 5′ degradation, but primarily is followed by removal of the 5′ end cap structure by the Dcp1p/Dcp2p decapping enzyme and 5′ to 3′ degradation of the body of the transcript by the exonuclease Xrn1p. Decapping is a key step of this process, because it precedes and permits the degradation of the body of the mRNA and represents the site of multiple control inputs.
The process of mRNA decapping and translation are mechanistically intertwined and appear to compete with each other, at least in yeast (reviewed in Coller and Parker, 2004 ). For example, decreasing translation initiation by a variety of means increases the rate of mRNA decapping (LaGrandeur and Parker, 1999 ; Muhlrad and Parker, 1999 ; Schwartz and Parker, 1999 , 2000 ). Conversely, inhibition of translation elongation leads to a significant decrease in the rate of mRNA decapping (Beelman and Parker, 1994 ). Moreover, coimmunoprecipitation experiments suggested that before decapping, the mRNA must exit translation and assemble into a translationally repressed messenger ribonucleoprotein (mRNP) complex capable of decapping (Tharun and Parker, 2001 ). These results indicate that before decapping mRNAs cease translation and assemble an mRNP containing the decapping machinery.
Additional evidence for a discrete population of nontranslating mRNPs has been that nontranslating mRNAs and the decapping machinery accumulate in discrete cytoplasmic foci called P-bodies (also referred as GW182 or Dcp bodies) (Bashkirov et al., 1997 ; Ingelfinger et al., 2002 ; Lykke-Andersen, 2002 ; van Dijk et al., 2002 ; Sheth and Parker, 2003 ; Cougot et al., 2004 ). P-bodies have now been observed in yeast, insect cells, nematodes, and mammalian cells and contain various proteins implicated in mRNA degradation, including the decapping enzyme (Dcp1p/Dcp2p), activators of decapping Dhh1p, Pat1p, Lsm1-7p, Edc3p, and the exonuclease Xrn1p (reviewed in Anderson and Kedersha, 2006 ; Eulalio et al., 2007 ; Sheth and Parker, 2007 ). Consistent with the reciprocal relationship between mRNA translation and degradation, the assembly of P-bodies is in a dynamic competition with translation (Teixeira et al., 2005 ; Sheth and Parker, 2007 ). Moreover, the mRNPs that accumulate in P-bodies have been suggested to be functionally involved in mRNA decapping (Sheth and Parker, 2003 ; Cougot et al., 2004 ), mRNA storage (Brengues et al., 2005 ; Bhattacharyya et al., 2006 ), general translation repression (Holmes et al., 2004 ; Coller and Parker, 2005 ), miRNA-mediated repression (Jakymiw et al., 2005 ; Liu et al., 2005 ; Pillai et al., 2005 ), nonsense-mediated decay (Unterholzner and Izaurralde, 2004 ; Sheth and Parker, 2006 ), and possibly viral packaging (Beliakova-Bethell et al., 2006 ). The existence of P-bodies as a discrete cytoplasmic compartment containing nontranslating mRNAs that can be either degraded or stored suggests that understanding the nature of the protein–protein and protein–RNA interactions that allow the assembly of both mRNPs containing the decapping machinery, as well as larger P-bodies visible by light microscopy, will be important in understanding the control of mRNA translation and degradation.
Two aspects of the assembly of P-bodies have emerged. First, it has been shown that P-bodies require mRNA for their formation and integrity (Teixeira et al., 2005 ). Second, coimmunoprecipitation and two-hybrid experiments have revealed a dense network of interactions between the components of the mRNA decapping and degradation machinery found in P-bodies (e.g., Hata et al., 1998 ; Coller et al., 2001 ; Ho et al., 2002 ; Fenger-Gron et al., 2005 ; Gavin et al., 2006 ; Krogan et al., 2006 ). However, the specific interactions that mediate the assembly of the translation repression and decapping machinery on mRNAs as well as the aggregation of individual mRNPs into larger P-bodies are largely unknown.
Several proteins have been described as affecting P-body assembly in yeast and mammalian cells. However, the competition between translation and P-body formation suggests that lack of a specific protein can affect P-body formation by either reducing the pool of nontranslating mRNAs, or by decreasing the aggregation of the nontranslating “P-body” mRNPs. For example, in mammalian cells, P-bodies are greatly reduced by knockdown of GW182, RCK/p54, RAP55, miRNA biogenesis in general, LSm4, Hedls/Ge-1, or 4E-T (Andrei et al., 2005 ; Ferraiuolo et al., 2005 ; Jakymiw et al., 2005 ; Pauley et al., 2006 ). However, for at least LSm4 knockdowns, P-bodies are restored when translation initiation is inhibited by arsenite (Kedersha et al., 2005 ). Additionally, depletion of GW182 relieves translational repression by miRNAs (Jakymiw et al., 2005 ; Liu et al., 2005a ; Meister et al., 2005 ; Rehwinkel et al., 2005 ; Behm-Ansmant et al., 2006 ). Moreover, 4-ET and RCK/p54 are known to function in translational repression (Andrei et al., 2005 ; Ferraiuolo et al., 2005 ). These observations argues that LSm4p, and possibly these other factors as well, are not required for P-body assembly per se, but instead contribute to P-body formation in mammalian cells by increasing the pool of translationally repressed mRNAs. Similarly, in yeast, Dhh1p and Pat1p are required for global translation repression of mRNAs, targeting mRNAs for decapping, and promoting their assembly into P-bodies (Coller and Parker, 2005 ). Strains lacking Dhh1p and Pat1p are defective in translation repression in response to glucose deprivation and amino acid starvation, mRNA decapping, and consequently in P-body formation (Holmes et al., 2004 ; Coller and Parker, 2005 ). In contrast, overexpression of Dhh1p or Pat1p results in inhibition of translation and induces P-body formation (Coller and Parker, 2005 ).
Given the limited understanding of how P-bodies assemble our goal in this work was to determine the requirement for numerous yeast proteins in P-body assembly and organization. To this end, we examined P-body formation and composition in strains defective in proteins known to accumulate in P-bodies. These experiments revealed specific dependencies in P-body assembly for individual proteins. In addition, this work argues that P-body assembly is redundant and no single known component of P-bodies is absolutely required for P-body assembly although Dcp2p and Pat1p can affect P-body assembly. Taken together, these results provide insight both into the function of individual proteins involved in mRNA degradation and the mechanisms by which yeast P-bodies assemble.
The genotypes of all strains used in this study are listed in Table 1. All strains have GAL1 upstream activating sequence-regulated PGK1pG and MFA2pG genes, as well as the LEU2 gene, collectively termed LEU2 pm, integrated at the CUP1 locus (Hatfield et al., 1996 ). Proteins were C terminal tagged with green fluorescent protein (GFP) following the PCR-based gene modification method described by Longtine et al. (1998) . These fusion proteins include the full-length protein and are all at least partially functional (Sheth and Parker, 2003 ). Yeast crosses were carried out using standard laboratory procedures. Strains were grown on standard yeast extract/peptone medium (YP) supplemented with 2% Dextrose (Glu) as carbon source. Strains were grown at 30°C.
Cells were grown to an OD600 of 0.3–0.35 in YPD. For observation, cells were washed once and resuspended in synthetic medium (SC) supplemented with amino acids and Glu and immediately observed. For glucose depletion, cells were washed in YP without Glu and resuspended in the same medium for 10 more minutes before being collected. Cells were then washed with SC plus amino acids without Glu, resuspended in the same medium, and examined. For observation at high cell density, cells were grown in SC to an OD600 of 1.0. Cells were then washed and resuspended in SC plus amino acids without Glu before observation. For in vivo 4′,6-diamidino-2-phenylindole (DAPI, Sigma) staining, 1 ml of cells was harvested by centrifugation and resuspended in 150 μl of medium containing 1.5 μl of DAPI (1 mg/ml). After shaking for 30 min at 30°C, cells were again pelleted, resuspended in 150 μl of media alone, and observed. Observations were made using Nikon PCM 2000 confocal microscope (Melville, NY) using a 100× objective with a 3× zoom using Compix software (Sewickley, PA). All images are a z-series compilation of 6–10 images in a stack.
To examine the role of various proteins in P-body assembly, we analyzed the contributions and dependencies to P-body formation of Dcp1p, Dcp2p, Edc3p, Dhh1p, Pat1p, Lsm1p, Xrn1p, Ccr4p, and Pop2p. For each protein a series of strains was constructed lacking one protein and individually containing a GFP fusion version of the other proteins. The GFP fusion proteins are integrated at the relevant genomic locus under the control of the endogenous promoter, include the full length protein C terminal tagged to GFP, and are all at least partially functional (Sheth and Parker, 2003 ). Each strain was then first examined microscopically for accumulation of each protein in P-bodies in yeast grown in glucose-supplemented rich medium during the midlog growth phase, wherein P-bodies are small in wild-type cells (Teixeira et al., 2005 ). This provides an ideal condition to identify lesions that increase the size of P-bodies. In some cases, strains were also observed under the stress condition of glucose deprivation, which decreases translation initiation and leads to the accumulation of large P-bodies (Ashe et al., 2000 ; Teixeira et al., 2005 ). Because P-bodies are large in wild-type cells undergoing a response to glucose deprivation, this provides an opportunity to identify lesions that reduce or compromise P-body assembly. Finally, where relevant we examined the accumulation of some proteins in strains lacking two proteins. The important results are discussed below.
Several observations demonstrated that blocking the catalytic events of decapping or 5′ to 3′ exonuclease digestion led to increased P-bodies. For example, strains lacking the 5′ to 3′ exonuclease Xrn1p, accumulated Dhh1p, Pat1p, Lsm1p, Dcp1p, Dcp2p, and Edc3p in P-bodies even in midlog cultures where yeast P-bodies are generally small (Figure 1B, III–VIII). Similarly, during midlog growth the dcp1Δ strain showed high levels of accumulation of Dhh1p, Pat1p, Lsm1p, Dcp2p, Edc3p, and Xrn1p in P-bodies (Figure 1C, III–IX). These large P-bodies observed in xrn1Δ or dcp1Δ strains did not increase significantly when cells were subjected to glucose deprivation, by shifting the cells for 10 min to medium lacking glucose (data not shown and Figure 2B), which is consistent with the observation that xrn1Δ and dcp1Δ strains are unable to repress translation during glucose deprivation (Holmes et al., 2004 ; Coller and Parker, 2005 ).
Because the xrn1Δ blocks 5′ to 3′ degradation and the dcp1Δ blocks decapping, these observations indicate that defects in the catalytic steps of decapping or 5′ to 3′ degradation lead to increased P-bodies. Moreover, because all the proteins examined accumulated in P-bodies under these conditions, it argues that no other known component of P-bodies is strictly dependent on Xrn1p or Dcp1p for accumulation in P-bodies. This is consistent and extends earlier work showing an accumulation of Dhh1p in P-bodies in xrn1Δ or dcp1Δ strains (Sheth and Parker, 2003 ). Moreover, these observations are consistent with the observation that RNAi-mediated knockdown of the decapping enzyme Dcp2p in mammalian cells leads to an accumulation of P-bodies as assessed by the human ortholog of Dhh1p, RCK/p54, as well as hCcr4p, and hLSm1p (Andrei et al., 2005 ).
Interestingly, Ccr4p and Pop2p were also clearly observed to accumulate in P-bodies in the xrn1Δ, dcp1Δ, and dcp2Δ mutant strains (Figure 1, B, I and II, C, I and II, and D, I and II). Similarly, we observed that Ccr4p and Pop2p accumulated at low levels in P-bodies during glucose deprivation (see below, Figure 8). This indicates that Ccr4p and Pop2p can accumulate in yeast P-bodies. This is consistent with earlier observations that other components of the Ccr4p/Pop2p/Not complex, the Not1-5 proteins, are detected in P-bodies in dcp1Δ strains (Muhlrad and Parker, 2005 ) and that yeast Ccr4p can be observed in small foci during a water stress response (Sheth and Parker, 2003 ). In addition, mammalian and Drosophila Ccr4p also accumulate in P-bodies (Cougot et al., 2004 ; Temme et al., 2004 ; Andrei et al., 2005 ). These results indicate that the Ccr4p/Pop2p/Not1-5p complex accumulates in P-bodies in yeast as well as other eukaryotes. However, the limited amount of Ccr4p-GFP and Pop2p-GFP seen in P-bodies compared with other factors suggests that these proteins are present at reduced stoichiometry.
Strains lacking Dcp2p also showed increased P-bodies in midlog cultures with accumulations of Ccr4p, Pop2p, Dhh1p, Pat1p, Lsm1p, Edc3p, and Xrn1p in P-bodies (Figure 1D, I–V and VIII–IX), but with some differences from dcp1Δ or xrn1Δ strains. First, despite the increased P-bodies in dcp2Δ strains as judged by other protein components, Dcp1p-GFP was distributed throughout the cell and completely absent from P-bodies (Figure 1D, VI). Moreover, Western analysis showed that Dcp1p-GFP was still efficiently expressed in the dcp2Δ strain (data not shown). Because dcp1Δ and dcp2Δ strains have identical and complete blocks to decapping (Beelman et al., 1996 ; Dunckley and Parker, 1999 ) and all other proteins accumulate in P-bodies in the dcp2Δ strain, this observation argues that Dcp2p is required to recruit Dcp1p to P-bodies.
Two additional observations confirm Dcp2p is required for Dcp1p recruitment to P-bodies. First, even when dcp2Δ strains are undergoing a response to glucose deprivation, which increases Dcp1p-GFP accumulation in P-bodies in wild-type strains, Dcp1p-GFP is completely absent from P-bodies in a dcp2Δ strain (Figure 2A, IV). In contrast, Dcp2p-GFP accumulates in P-bodies independent of Dcp1p with or without glucose deprivation conditions (Figure 2B, II and IV). Moreover, other P-body markers still accumulate in P-bodies in a dcp2Δ with or without glucose deprivation conditions (Figure 1D and data not shown). Second, although Dcp1p-GFP accumulates in P-bodies to high levels in the xrn1Δ strain (Figure 2C, I), this accumulation is completely lost in the xrn1Δ dcp2Δ double mutant strain (Figure 2C, II). The requirement of Dcp2p for Dcp1p recruitment to P-bodies is consistent with the direct physical interaction between Dcp1p and Dcp2p in yeast (Steiger et al., 2003 ; She et al., 2004 , 2006 ) and with the observation that specific point mutations in residues of Dcp2p that disrupt the physical interaction between Dcp2p and Dcp1p prevent Dcp1p from accumulating in P-bodies (Decker and Parker, unpublished observation). These results indicate that Dcp1p is recruited to the P-body through interactions with Dcp2p.
A second interesting result observed with the dcp2Δ strain was that P-bodies were reproducibly smaller than in dcp1Δ or xrn1Δ strains (Figure 1, B–D). This can be easily seen by comparing the accumulation of Dhh1p, Pat1p, Lsm1p, or Xrn1p in dcp1Δ strains compared with dcp2Δ strains (compare Figure 1C, III–V and IX with 1D, III–V and IX). Because both dcp1Δ and dcp2Δ strains show absolute blocks to decapping (Beelman et al., 1996 ; Dunckley and Parker, 1999 ), this difference in P-body size must reflect a difference in the function of Dcp1p or Dcp2p in the formation of P-bodies independent of their catalytic role in decapping. Interestingly, the smaller size of P-bodies in the dcp2Δ strain relative to the dcp1Δ strain was not as pronounced when Edc3p was examined (Figure 1D, VIII), suggesting some difference in how Edc3p's accumulation in P-bodies is affected compared with other factors.
Two possible models can explain the difference between P-body size in the dcp1Δ and dcp2Δ strains, which we distinguished by the analysis of a dcp1Δ dcp2Δ double mutant. First, it could be that Dcp1p is also an inhibitor of P-body formation and that P-bodies are larger in the dcp1Δ strain compared with the dcp2Δ strain because of the loss of the Dcp1p inhibitory role on P-body formation. In this model, a dcp1Δ dcp2Δ double mutant is predicted to have large P-bodies similar to the dcp1Δ strain. Alternatively, Dcp2p might have an additional role in the assembly or maintenance of P-bodies, possibly through additional protein–protein interactions. In this model, a dcp1Δ dcp2Δ strain is predicted to be phenotypically similar to a dcp2Δ strain.
To distinguish these models, we examined the accumulation of Dhh1p and Pat1p in dcp1Δ dcp2Δ double mutants strains as compared with dcp1Δ and dcp2Δ single mutant strains. Strikingly, we observed that both Dhh1p-GFP and Pat1p-GFP accumulated in P-bodies in the dcp1Δ dcp2Δ double mutant to the same extent seen in the dcp2Δ single mutant (Figure 3, V and VI). This provides evidence that Dcp2p has some role in the assembly or maintenance of P-bodies. A likely possibility is that the multiple interactions that Dcp2p has with other components of P-bodies contribute to the assembly of the P-body mRNP or to the interactions that allow P-body aggregation into visible structures in the light microscope.
We also examined how Ccr4p and Pop2p affect the formation and composition of P-bodies in both midlog cultures and in response to glucose deprivation. We observed that ccr4Δ led to a small, but clear, decrease in the accumulation of Dcp2p-GFP and Edc3p-GFP in P-bodies in midlog cultures (Figure 4B, V and VI). Furthermore, we observed pop2Δ led to a small decrease in the accumulation Dcp1p-GFP and Dcp2p-GFP in P-bodies (Figure 4C, IV and V). This is consistent with the ccr4Δ and pop2Δ affecting deadenylation (Tucker et al., 2001 ) and previous work arguing that in midlog growth phase deadenylated mRNAs preferentially associate with P-body components such as Lsm1p (Tharun and Parker, 2001 ). In midlog cultures, the ccr4Δ or pop2Δ strains did not show a significant difference in the concentration of Dhh1p-GFP, Pat1p-GFP, Lsm1p-GFP, or Xrn1p-GFP in P-bodies (Figures 4B, I–III and VII, and and4C,4C, I-III and VII). However, because it is difficult to observe these proteins in P-bodies in midlog growth, we are unable to determine if the ccr4Δ or pop2Δ affects their accumulation in P-bodies under these conditions.
In the ccr4Δ strains in response to glucose deprivation stress, where decreases in P-bodies are more easily identified, Pat1p-GFP and Edc3p-GFP accumulated in P-bodies in a manner similar to wild-type cells (Figure 4B, IX and XIII), whereas Dhh1p-GFP, Lsm1-GFP, Dcp1p-GFP, Dcp2p-GFP, and Xrn1p-GFP accumulated but less pronouncedly compared with wild-type cells (Figure 4B, VIII, X–XII, and XIV). This slightly reduced accumulation of Dhh1p in P-bodies in these mutant may be due to the reduced levels of protein expression in ccr4Δ strains, as has been shown to be the case for Dhh1p (Sheth and Parker, 2003 ). In contrast, after glucose deprivation in a pop2Δ strain, Dhh1p, Pat1p, Lsm1p, Dcp1p, Dcp2p, Edc3p, and Xrn1p accumulated in P-bodies in a manner similar to wild-type cells (Figure 4C, VIII–XIV). Taken generally, these results indicate that during glucose deprivation, neither Ccr4p nor Pop2p has a large impact on P-body assembly. This is consistent with recent results indicating that poly(A)+ mRNAs are translationally repressed during glucose deprivation and accumulate in P-bodies (Brengues and Parker, in press).
The proteins Dhh1p, Pat1p, and Lsm1p are all found in P-bodies and function as general decapping activators (reviewed in Coller and Parker, 2004 ). To determine the roles of these proteins on P-body assembly we examined the accumulation of GFP tagged version of Dcp1p, Dcp2p, Edc3p, Xrn1p, Dhh1p, Pat1p, and Lsm1p in lsm1Δ, pat1Δ or dhh1Δ strains. Because results from mammalian cells indicate that the Lsm1-7p complex must be intact to localize to P-bodies (Ingelfinger et al., 2002 ), and lsm1Δ is sufficient to fully inactivate this complex with respect to decapping (Tharun et al., 2000 ), the lsm1Δ should be sufficient to address the contribution of the entire Lsm1p-7p complex to P-body formation. In addition, we examined both the affect of the deletions on P-bodies during midlog growth, where P-bodies are small and increases are easily seen, and during glucose deprivation, where P-bodies are large, and decreases in P-body assembly are more easily observed. These results are presented in Figure 5 and important observations are discussed below.
One important observation was that lsm1Δ strains showed increased accumulation of Dcp1p, Dcp2p, Edc3p, Xrn1p, and Dhh1p in P-bodies even in midlog growth cells (Figure 5B, I–V). Pat1p did not accumulate in P-bodies in midlog cultures in the lsm1Δ strain, perhaps because of an effect of the lsm1Δ on the nuclear-cytoplasmic distribution of Pat1p (Figure 5B, VI; see below). The accumulation of all P-body components except Pat1p in P-bodies in lsm1Δ strain is consistent with the decapping defect seen in lsm1Δ strains (Boeck et al., 1998 ; Bouveret et al., 2000 ; Tharun et al., 2000 ). Moreover, this observation argues that the rate-limiting step in decapping in lsm1Δ strains is after mRNAs assembly into an mRNP that can accumulate in P-bodies. This identifies an important function of the Lsm1-7p complex in triggering decapping after translation repression and accumulation of the mRNA in the P-body mRNP. Consistent with the partial block to mRNA decapping in the lsm1Δ strain, the accumulation of P-bodies is not as strong in the lsm1Δ strain compared with the dcp1Δ strain, where decapping is absolutely blocked (compare Figure 1C with with5B).5B). Moreover, in lsm1Δ strains, P-bodies still increased during glucose deprivation compared with nonstressed lsm1Δ cells (Figure 5B, VIII–XIII), which suggests that lsm1Δ strains do not have a maximal accumulation of P-bodies.
In contrast to the accumulation of P-bodies in the lsm1Δ strain, in general the pat1Δ and dhh1Δ strains showed either minor or no reduction of P-bodies during midlog growth, respectively, as assessed by Dcp1p, Dcp2p, Edc3p, and Xrn1p (Figure 5, C, I–IV, and D, I–IV). Interestingly, we did observe a minor increase in P-bodies during midlog growth for dhh1Δ and pat1Δ cells in a subpopulation of the cells (Figure 5). This suggests that Dhh1p and/or Pat1p can affect the rate of mRNA decay within P-bodies or the rate of mRNA exit from P-bodies in at least some cells. However, during glucose deprivation the pat1Δ strain and to a more modest effect the dhh1Δ strain showed a reduction in the amount of P-bodies formed (compare Figure 5A, VIII–XIV, with 5C, VIII–XIV and 5D, VIII–XIV). This is consistent with earlier results that dhh1Δ and pat1Δ strains are partially defective in translation repression during glucose deprivation (Holmes et al., 2004 ; Coller and Parker, 2005 ). Because both dhh1Δ and pat1Δ appear to affect translation repression similarly during glucose deprivation (Holmes et al., 2004 ; Coller and Parker, 2005 ), this suggests Pat1p may have a more significant role in P-body assembly and aggregation than Dhh1p (see below and Discussion).
A second important result from these comparisons was that strains lacking Pat1p failed to accumulate Lsm1p in P-bodies with or without glucose repression (Figure 5C, VII and XIV). This argued that Pat1p is required for Lsm1-7p to be recruited to P-bodies. However, because the loss of Pat1p can affect the size of P-bodies during glucose repression, we verified this result under conditions where P-bodies were large even in a pat1Δ strain. To do this, we examined whether Lsm1p accumulates in P-bodies in a pat1Δ strain at high cell density, where P-bodies are large (Teixeira et al., 2005 ). Consistent with Pat1p being required for Lsm1p to enter into P-bodies, we observed that pat1Δ cells did not accumulate Lsm1p in P-bodies at high OD (Figure 6A, II). These observations demonstrate that Pat1p is required to recruit the Lsm1-7p complex to P-bodies.
A third observation from these comparisons was that in the lsm1Δ strain Pat1p accumulated in a large circular region of the cell that was similar in size the nucleus (Figure 5B, VI). Subsequent DNA staining with DAPI revealed that this circular region of Pat1p concentration colocalized with the nucleus (Figure 6B). This observation indicates that Lsm1p, and presumably the whole Lsm1-7p complex, plays a role in decreasing the concentration of Pat1p in the nucleus and increasing the cytoplasmic concentration of Pat1p. These results suggest that Pat1p is a likely nuclear-cytoplasmic shuttling protein. This conclusion is consistent with earlier work showing that Pat1p shows physical interactions with the mRNA export factor Crm1p (Jensen et al., 2000 ).
Despite the nuclear concentration of Pat1p in the lsm1Δ mutant, three observations suggest that some Pat1p is still in the cytoplasm and is functioning in decapping. First, even in the lsm1Δ strain during midlog growth, some Pat1p-GFP was observed distributed in the cytoplasm (Figure 5B, VI). Second, when lsm1Δ cells were glucose deprived some Pat1p was seen to accumulate in P-bodies (Figure 5B, XIII). Third, when P-bodies accumulate because of dcp1Δ, Pat1p-GFP is still present in P-bodies, even in the absence of Lsm1p, albeit at a reduced level compared with a dcp1Δ alone (Figure 6C, III). These observations indicate that Lsm1p is not absolutely required for Pat1p to enter P-bodies, although it clearly appears to decrease the amount of Pat1p seen in P-bodies, at least in part by affecting the distribution of Pat1p between the nucleus and the cytoplasm. Interestingly, although P-bodies accumulate due to xrn1Δ, Pat1p-GFP was no longer observed in P-bodies in the xrn1Δ lsm1Δ double mutant (Figure 6C, IV). This further suggests the idea that Pat1p/Lsm1-7p complex/Xrn1p form a functionally important complex, which is consistent with their previous biochemical copurification (Bouveret et al., 2000 ).
The above observation suggest a model for P-body assembly wherein Pat1p acts early in the process and is required to recruit the Lsm1-7p complex, which functions later to enhance decapping rate. A prediction of this model is that a pat1Δ lsm1Δ double mutant should show the phenotype of the pat1Δ single mutant strain and not the lsm1Δ strain. To test this prediction, we examined the accumulation of GFP tagged Dcp1, Dcp2, Edc3p and Xrn1p in pat1Δ lsm1Δ strains. For comparison, we also examined the same proteins in a dhh1Δ lsm1Δ strain.
A clear and significant observation was that during midlog growth the pat1Δ lsm1Δ strain showed little P-body accumulation of all four proteins similar to a pat1Δ strain (Figure 7, E–H) and unlike a lsm1Δ strain where P-bodies are increased (Figure 5B). This result provides additional evidence that Pat1p acts upstream of Lsm1p in the process of P-body assembly and mRNA decapping. In contrast, we observed that dhh1Δ lsm1Δ leads to increased concentration of Dcp1p, Dcp2p, Edc3p, and Xrn1p in P-bodies, because P-bodies increased in size and number compared with both wild-type cells and dhh1Δ or lsm1Δ single-mutant strains (Figure 7, A–D). This suggests that Dhh1p, like Lsm1p, can also play a role in the actual rate of decapping after assembly of the mRNP capable of aggregation in P-bodies.
In examining the function of Dhh1p, Pat1p, and Lsm1p in P-body assembly we also examined how they affected the recruitment of Ccr4p and Pop2p, subunits of the major cytoplasmic deadenylase, to P-bodies during glucose deprivation (Figure 8A). We observed that during glucose deprivation the amount of Ccr4p or Pop2p seen in P-bodies declined in dhh1Δ, pat1Δ or lsm1Δ strains compared with a wild-type strain, where small but clear accumulation of Ccr4p and Pop2p in P-bodies could be seen during glucose deprivation (Figure 8A, III, IV, VII, VIII, XI, XII, XV, and XVI). This suggests that the accumulation of Ccr4p and Pop2 in P-bodies requires Dhh1p, Pat1p, and Lsm1p. However, a limitation of this analysis is that the accumulation of Ccr4p and Pop2p in P-bodies in the wild-type strain is relatively low; thus it is difficult to make a robust conclusion from this experiment alone.
To further support this observation we asked how dhh1Δ, pat1Δ or lsm1Δ affected the accumulation of Ccr4p or Pop2p in P-bodies in combination with xrn1Δ where the accumulation of Ccr4p and Pop2p in P-bodies is easily observed (Figure 1). We observed that the dhh1Δ, pat1Δ or lsm1Δ all reduced the accumulation of Ccr4p and Pop2p in P-bodies compared with xrn1Δ single mutant (Figure 8B, III–VIII). This is consistent with the observations during glucose deprivation and argues that the efficient recruitment of Ccr4p and Pop2p to P-bodies requires Dhh1p, Pat1p, and the Lsm1-7p complex.
This work begins to reveal the process by which P-bodies assemble and the possible roles of individual proteins within that assembly. One principle that emerges is that P-body assembly appears to be redundant, with no single protein being absolutely required for formation of P-bodies.
One protein that can affect P-body formation was Dcp2p. This was surprising because strains lacking Dcp2p show a complete block to decapping and therefore, like dcp1Δ strains, would be expected to accumulate very large P-bodies. However, dcp2Δ strains had reproducibly smaller P-bodies than dcp1Δ or xrn1Δ strains, although dcp2Δ strains showed enhanced P-bodies compared with wild-type strains (Figure 1). Moreover, because dcp2Δ strains have smaller P-bodies than a dcp1Δ strain as assessed by seven different proteins, it strongly argues that the difference is in the actual formation of P-bodies themselves and not simply a change in P-body composition. Finally, because dcp2Δ dcp1Δ strains are similar to dcp2Δ single mutant strains alone when assessed for P-body formation, it argues that Dcp2p is required for optimal P-body accumulation (Figure 3). Because Dcp2p shows direct physical interactions with Dcp1p, Dhh1p, Edc3, and possibly Pat1p (Steiger et al., 2003 ; She et al., 2004 ; Decker, Pilkington, and Parker, unpublished observation), the simplest possibility is that the multiple protein–protein interactions that Dcp2p participates in either stabilizes the P-body monomer mRNP or helps to provide cross-linking interactions between individual mRNPs, thereby contributing to aggregation of multiple mRNPs into a larger P-body. Thus, Dcp2p is not only the catalytic subunit of the decapping enzyme, but also has a second role in promoting P-body assembly or maintenance.
Several observations argue that Pat1p is likely to affect P-body formation in multiple manners. First, because Pat1p overexpression leads to inhibition of translation (Coller and Parker, 2005 ), one possible role of Pat1p is likely to be to inhibit translation initiation in some manner, thereby contributing to P-body formation by affecting the size of the pool of nontranslating mRNA, which can then aggregate into P-bodies. Second, because P-bodies are reduced in pat1Δ strains, even during glucose deprivation, where translation is markedly repressed even in a pat1Δ strain, it suggests that Pat1p also plays a role in assembly of P-bodies (Figure 5). The role of Pat1p in assembly is again likely to be through multiple protein–protein interactions that promote formation of the P-body monomer mRNP, and cross-linking between individual mRNPs. Finally, we suggest that Pat1p has a final role in promoting mRNA decapping after assembly of the P-body mRNP. This is suggested by both the requirement for Lsm1p for efficient decapping after P-body assembly, and the requirement for Pat1p for assembly of Lsm1p into P-bodies.
Our analysis of yeast P-bodies suggests that there are some clear dependencies in the assembly of specific components. For example, the assembly of Dcp1p in P-bodies is dependent on Dcp2p (Figures 1 and and2).2). This is consistent with the direct physical interaction between these components from yeast (Steiger et al., 2003 ; She et al., 2004 , 2006 ).
A second clear dependency is that Pat1p is required for the recruitment of Lsm1p to P-bodies (Figures 5 and and6).6). Because Lsm1p is a component of the Lsm1-7p complex and Lsm1-7p complex formation is required for its recruitment to P-bodies and for the Lsm1-7 complex to function in mRNA decay (Bouveret et al., 2000 ; Tharun et al., 2000 , 2005 ; Ingelfinger et al., 2002 ), this suggests that Pat1p is required for the recruitment of the entire Lsm1-7p complex into P-bodies. This is consistent with the physical interactions between Pat1p and the Lsm1-7p complex (Bouveret et al., 2000 ; Tharun et al., 2000 ).
Our results also argue that the Lsm1-7p complex has an important role after P-body assembly in the actual triggering of decapping. The key observation is that lsm1Δ strains show an accumulation of P-bodies, as judged by the subcellular distribution of Dcp1p, Dcp2p, Edc3p, Xrn1p, and Dhh1p (Figure 5). This is consistent with the observed defect in mRNA decapping in the lsm1Δ strains (Boeck et al., 1998 ; Bouveret et al., 2000 ; Tharun et al., 2000 ) and indicates that the decapping mRNP can assemble and aggregate in P-bodies in the absence of the Lsm1p, but is then deficient at actual decapping, and hence a pool of mRNAs accumulates in P-bodies. It should be noted that these results do not rule out a potential earlier role of the Lsm1-7p complex in translation repression or assembly of the monomer unit, they just reveal what becomes the rate-limiting step in the absence of Lsm1p.
An important implication of the function of the Lsm1-7p complex in triggering decapping after P-body formation is that this role of the Lsm1-7p complex could be used to alter the fate of mRNAs within P-bodies. Specifically, individual mRNAs that are destined for storage could simply be packaged into an mRNP that is lacking the Lsm1-7p complex and thereby have a reduced rate of mRNA decapping. Similarly, in certain biological contexts where mRNA storage is important, the Lsm1-7p complex may be lacking from P-bodies and thereby converts the fate of mRNAs within P-bodies into storage.
Our data also indicate that Pat1p is required for Lsm1p to assemble into P-bodies. The key observation is that P-bodies formed in pat1Δ strains lack Lsm1p, even under glucose deprivation (Figure 5) or high OD (Figure 6). This is also consistent with earlier work showing that the coimmunoprecipitation of Lsm1p with Dcp2p is greatly reduced in a pat1Δ strain (Tharun and Parker, 2001 ). The simplest model is that Pat1p directly interacts with Lsm1-7p complex and recruits it to P-bodies, although one anticipates that Pat1p might also facilitate the interaction of the Lsm1-7p complex with other components of P-bodies.
The phenotypes of lsm1Δ and pat1Δ strains suggest a model for these proteins function wherein Pat1p participates early in translation repression and the movement of mRNAs into P-bodies and the Lsm1-7p complex has a rate limiting role at a later stage to trigger decapping. The early role of Pat1p is suggested by the observations that overexpression of Pat1p inhibits translation (Coller and Parker, 2005 ) and that pat1Δ strains show reduced P-bodies even under glucose deprivation conditions (Figure 5). The late role of Lsm1p is indicated by the accumulation of P-bodies in the lsm1Δ strain. Moreover, epitasis analysis is consistent with Pat1p acting before Lsm1-7p because pat1Δ lsm1Δ double mutant strains show reduced P-bodies similar to pat1Δ strains (Figure 7).
Our results indicate that Pat1p is likely to be a nuclear-cytoplasmic shuttling protein and its distribution between the nucleus and the cytoplasm is affected by the Lsm1-7p complex. The key observation is that in lsm1Δ strains, Pat1p accumulates in the nucleus (Figures 5 and and6).6). Consistent with Pat1p being a nuclear-cytoplasmic shuttling protein, previous work has identified a two-hybrid interaction between Pat1p and Crm1p, which is involved in mRNA export (Jensen et al., 2000 ). In addition, Pat1p was identified in a proteomic analysis of the components of the penta-snRNP, which is a large nuclear complex involved in pre-mRNA splicing (Stevens et al., 2002 ). Interestingly, this nuclear localization is not unique to Pat1p in lsm1Δ strains, because we also detect a clear accumulation of Dhh1p in the nucleus in lsm1Δ strains under glucose deprivation conditions (Figure 5).
In principle, the accumulation of Pat1p in the nucleus in lsm1Δ strain could be explained in two manners. First, it could be that interaction of the Lsm1-7p complex with Pat1p in the nucleus is required for efficient export of Pat1p from the nucleus, perhaps in association with mRNAs. This model would imply that an mRNP structure similar to a P-body mRNP may form on some, or all, mRNAs in the nucleus and be transported with them to the cytoplasm. An alternative, and potentially overlapping, model is that interaction of Pat1p with the Lsm1-7p complex in the cytoplasm limits import of Pat1p into the nucleus. Future experiments should be able to distinguish between these models.
Pat1p might function in the nucleus in one of two manners. First, it could be that Pat1p plays a role in a nuclear process such as splicing or degradation of aberrant mRNAs in the nucleus. Alternatively, Pat1p may enter the nucleus to become assembled into nascent mRNPs before export to the cytoplasm, which might lead to those mRNAs entering the cytoplasm in a translationally repressed mRNP that is directly targeted to P-bodies. Such potential targeting of nascent mRNPs might occur on all mRNAs or might be more specific to subclasses of mRNAs, or biological contexts, where translation repression of nascent transcripts is particularly important. Interestingly, one example of such a context is in the biogenesis of maternal mRNAs, which are exported to the cytoplasm and then stored in maternal germ granules, which are related to P-bodies (Anderson and Kedersha, 2006 ; Sheth and Parker, 2007 ). Here it is worth noting that the Xenopus ortholog of Dhh1p is a nuclear-cytoplasmic shuttling protein (Smillie and Sommerville, 2002 ) and may play a role in packaging nascent mRNAs for direct targeting to storage particles. An interesting area of future work will be to understand the functional significance of P-body components shuttling between the nucleus and the cytoplasm.
We thank the members of the Parker lab for helpful discussions, especially Carolyn Decker, Denise Muhlrad, and Anne Webb for critical reading of the manuscript and technical support. We also thank Bettsy Valencia for help in strain construction and Guy Pilkington for colocalizing Pat1p with nuclear markers. We also thank the Department of Molecular and Cellular Biology and Carl Boswell for the use of the confocal facility. National Institutes of Health Grant (R37 GM45443) and funds from the Howard Hughes Medical Institute supported this work. D.T. was supported by Fundacao para a Ciencia e Tecnologia (SFRH/BD/2739/2000), Portugal.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-03-0199) on April 11, 2007.