|Home | About | Journals | Submit | Contact Us | Français|
eIF3 promotes translation initiation, but relatively little is known about its full range of activities in the cell. Here, we employed affinity purification and highly sensitive LC-MS/MS to decipher the fission yeast eIF3 interactome, which was found to contain 230 proteins. eIF3 assembles into a large supercomplex, the translasome, which contains elongation factors, tRNA-synthetases, 40S and 60S ribosomal proteins, chaperones, and the proteasome. eIF3 also associates with ribosome biogenesis factors and the importins-β Kap123p and Sal3p. Our genetic data indicated that the binding to both importins-β is essential for cell growth, and photobleaching experiments revealed a critical role for Sal3p in the nuclear import of one of the translasome constituents, the proteasome. Our data reveal the breadth of the eIF3 interactome and suggest that factors involved in translation initiation, ribosome biogenesis, translation elongation, quality control, and transport are physically linked to facilitate efficient protein synthesis.
The recruitment of the 40S ribosomal subunit to the mRNA start codon is thought to be the rate-limiting step in eukaryotic translation. This process requires the assembly of a ribonucleoprotein complex, which joins mRNAs with some 30 different polypeptides referred to as eukaryotic initiation factors (eIFs) (Hershey and Merrick, 2000). 40S ribosomes associate with the eIF2·GTP/Met-tRNA ternary complex, the multisubunit eIF3 complex, and several other eIFs to form the 43S pre-initiation complex. This complex then binds to a second protein assembly organized around eIF4G, resulting in the 43S initiation complex. eIF4G interacts with both the cap-binding protein eIF4E and the poly-A binding protein, thus presumably circularizing the mRNA. eIF4G also recruits the eIF4A helicase assisting the 43S complex in scanning along the mRNA. Once the start codon is identified, the 43S complex is converted into the 48S initiation complex, which forms a stable interaction with the initiator AUG. At this point, eIF2-bound GTP is hydrolyzed, leading to dissociation of eIFs, thus allowing the 60S ribosomal subunit to join for productive protein synthesis.
eIF3 is the most complex translation initiation factor and plays several important roles that were revealed by in vitro reconstitution experiments (Dong and Zhang, 2006; Hinnebusch, 2006). First, eIF3 binds to the 40S ribosome and facilitates loading of the eIF2·GTP/Met-tRNA ternary complex to form the 43S pre-initiation complex. Subsequently, eIF3 assists in recruiting mRNAs to the 43S complex, presumably involving the RNA recognition motifs found in some of its subunits. Lastly, eIF3 binding to the 40S ribosome prevents the joining of the 60S subunit until the start codon is identified, eIF2-bound GTP is hydrolyzed by eIF5, and all eIFs are released. While these discreet reaction steps were deciphered extensively in vitro, it remained unclear how they are coordinated in vivo in order to ensure efficient translation.
Whereas human eIF3 consists of 13 subunits, consecutively named eIF3a – m (Damoc et al., 2007; Unbehaun et al., 2004; Zhou et al., 2008), budding yeast contains only five stochiometric subunits, which are orthologs of human eIF3a, b, c, g, and eIF3i, and the substoichiometric eIF3j. These subunits may constitute a core complex, as all are essential for viability (Asano et al., 1997; Phan et al., 1998). In the fission yeast, Schizosaccharomyces pombe, eIF3 contains the same five core subunits, in addition to the non-core subunits eIF3d, e, f, g, h, i, and m (Akiyoshi et al., 2001; Bandyopadhyay et al., 2002; Burks et al., 2001; Crane et al., 2000; Dunand-Sauthier et al., 2002; Ray et al., 2008; Zhou et al., 2005). Two distinct eIF3 complexes were identified in fission yeast that contain an overlapping set of core subunits but are distinguished by the presence of the related eIF3e and eIF3m proteins (Zhou et al., 2005). The eIF3m containing complex appears to mediate the translation of the bulk of cellular mRNAs, whereas the eIF3e containing complex associates with a far more restricted set of mRNAs. Distinct eIF3 complexes may therefore contribute to mRNA specificity of translation.
eIF3 also has functions that are apparently independent of its role in translation initiation. For example, some eIF3 subunits interact with the 26S proteasome (Dunand-Sauthier et al., 2002; Hoareau Alves et al., 2002; Paz-Aviram et al., 2008; Yen et al., 2003b). The significance of this interaction was revealed in S. pombe, where deletion of the non-essential eIF3d/Moe1p and eIF3e/Yin6p confers a series of cellular phenotypes that indicate a defect in proteasomal protein degradation (Yen et al., 2003b). This defect was pinpointed to a role of these eIF3 subunits in the nuclear accumulation and assembly of the 26S proteasome. However, the molecular mechanisms underlying eIF3-directed proteasome localization and assembly remained unknown.
To clarify this issue and to further elucidate the functions of eIF3, we performed a high sensitivity mass spectrometry analysis of eIF3 complexes purified from S. pombe. This eIF3 interactome suggests an extensive repertoire of eIF3 roles in protein synthesis and degradation thus establishing a molecular link between these processes.
To affinity purify eIF3 complexes, two S. pombe strains were used that encode fully functional eIF3e or eIF3m modified at the endogenous genomic loci with protein A epitope tags preceded by a cleavage site for tobacco etch virus (TEV) protease (Zhou et al., 2005). Total cell lysate was absorbed to IgG coupled magnetic beads, and retained proteins were eluted by cleavage with TEV protease (Fig. 1A). Mock purifications of cell lysate devoid of any epitope-tagged protein were included as specificity controls. The eluates of a representative purification series were resolved by SDS-PAGE alongside fractions taken at various steps of the purification procedure. Approximately 30 – 40% of the TEV cleaved material was released from the beads into the supernatant as judged by Coomassie staining (Fig. 1B, compare lanes “TE” and “SE”). The gel also revealed an overlapping but not identical pattern of bands in the eIF3e and eIF3m complexes as previously described (Zhou et al., 2005). eIF3e is known to bind the proteasome (Yen et al., 2003b). To confirm the integrity of the purifications, we examined a 19S proteasome subunit, Rpn1p, and found that it co-purified with both the eIF3e and eIF3m bait (Fig. 1C).
Purifications of the eIF3e and eIF3m complexes were performed in triplicate. For the third series of purifications, cell lysates were treated with RNase A prior to affinity capture, in order to disrupt protein interactions that were mediated by RNA (Supplementary Fig. 1). Nine samples (triplicates of mock, eIF3e, eIF3m) were digested with trypsin and analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) on a high-sensitivity LTQ Orbitrap XL mass spectrometer. Each sample was analyzed 3 – 4 times to yield a total of 33 independent LC-MS/MS runs. The resulting mass spectra were searched against the S. pombe protein database using SEQUEST, and search results were subjected to probability-based filtering using ProteinProphet (Trans-Proteomic Pipeline; Institute for Systems Biology, Seattle, WA) for a false discovery rate of protein identification of ≤ 0.02. Altogether, we identified 3876 unique proteins, which equal 77.1% of the predicted S. pombe proteome. The majority of these proteins appear to be low abundance background, because they were represented by peptides that were detected in only a few runs, and because they had very low spectrum counts (< 3). The spectrum count of a protein is the cumulative number of times peptides defining that protein were selected for MS/MS scans. Spectrum counts therefore provide a semi-quantitative measure of relative protein abundance within and between samples (Liu et al., 2004). To obtain a high confidence set of specific and reproducible eIF3 interacting proteins, the eIF3 interactome, we subjected our entire dataset to background subtraction using the mock purifications and filtering based on spectrum counts (Supplementary Methods).
The high confidence eIF3 interactome comprised 230 proteins, which were consistently identified in all six independent purifications (Fig. 2A). As expected, the 10 known subunits of S. pombe eIF3 were the most abundant proteins retrieved (Fig. 2A, B). In addition, a novel protein homologous to human eIF3j (SPAC3A12.13c) was identified as a substoichiometric eIF3 subunit, a finding that is consistent with the observation that eIF3j undergoes regulated cycles of association and dissociation with holo-eIF3 in human cells (Miyamoto et al., 2005). Interestingly, we found two phosphorylation sites in eIF3j, which might be involved in this regulation (Supplementary Table 1). In summary, S. pombe eIF3 contains orthologs of all human subunits except eIF3k and eIF3l, which are absent from the S. pombe genome.
Next, we performed a comparative quantification of subunits identified in purifications of either the eIF3e or the eIF3m bait proteins. Since the masses of individual eIF3 subunits vary between 30.5 kDa (eIF3j) and 107 kDa (eIF3a), we adjusted spectrum counts to molecular weights and, for better comparison between samples, normalized the numbers to eIF3a (Supplementary Data File 1). The adjustments clarified the suggestion from raw spectrum counts that eIF3d and eIF3e are substoichiometric components of the complex purified via eIF3m (Fig. 2B, C). In contrast, the eIF3e bait co-purified a complex containing roughly stoichiometric amounts of all subunits, except eIF3j (Fig. 2C).
These data extend and refine those of our previous study, employing lower sensitivity mass spectrometry, which suggested two distinct eIF3 complexes in S. pombe that are distinguished by the presence or absence of eIF3d and eIF3e (Zhou et al., 2005). Since neither eIF3d nor eIF3e are essential for global mRNA translation and cell viability (Akiyoshi et al., 2001; Bandyopadhyay et al., 2002; Bandyopadhyay et al., 2000; Chen et al., 1999; Crane et al., 2000; Yen and Chang, 2000; Zhou et al., 2005), the bulk of protein synthesis under normal growth conditions may be executed by a “global” eIF3 complex lacking these subunits, which is represented by the complex purified via the eIF3m bait (Fig. 2D). Consistent with this observation is our previous demonstration that this global eIF3 binds the majority of cellular mRNAs (Zhou et al., 2005). Under certain conditions, eIF3d and eIF3e may join the global complex to modulate its mRNA specificity. For example, these factors may facilitate the recruitment of mRNAs that are translated under conditions of cellular stress to which eif3d and eif3e mutants are sensitive (Akiyoshi et al., 2001; Bandyopadhyay et al., 2000; Crane et al., 2000; Yen and Chang, 2000).
Since an eIF3 complex containing apparently stoichiometric amounts of eIF3d and eIF3e can be readily purified from unstressed S. pombe cells (Fig. 1, Fig. 2) as well as from other organisms, these proteins may also carry out important, albeit non-essential, functions under normal growth conditions. The critical target mRNAs of complexes containing eIF3d and eIF3e in stressed and unstressed cells are currently unknown. Nevertheless, recent mass spectrometry analysis of the intact 13-subunit human eIF3 complex suggested that eIF3d and eIF3e are peripheral subunits that may undergo dynamic exchange (Zhou et al., 2008), although the signals that can trigger such exchange remain to be identified.
The eIF3 interactome also contained other initiation factors of the 43S initiation complex, including eIF2, eIF5A and B, eIF4A and eIF4G (Fig. 2A, ,3A,3A, Supplementary Table 2). Several other eIFs, including eIF1A, eIF2B, eIF5, and the cap binding protein eIF4E1 were also identified, although they did not pass the strict spectrum count thresholds in some of the six individual datasets (Supplementary Data File 1). Since none of these interactions was affected by RNase treatment of the cell lysate (Fig. 3A), S. pombe eIF3 appears to assemble into a RNA-independent complex with other initiation factors of the 43S complex, although we cannot entirely exclude the possibility that short remnants of mRNA resistant to complete RNase digestion contribute to the interactions.
eIF3 also associated with translation elongation factors, multiple tRNA-synthetases as well as ribosomal proteins (Fig. 2A). Remarkably, eEF1A, eEF2, and eEF3 were represented more abundantly in the eIF3 complex than other 43S subunits (Fig. 3B, Supplementary Table 2). In addition, we found eight tRNA-synthetases enriched in the eIF3 complex (Fig. 3C, Supplementary Table 2). These enzymes may constitute the S. pombe homolog of the multisynthetase complex (MSC) of mammalian cells (Dang, 1986). Moreover, we identified twenty four 40S ribosomal subunits and forty four 60S ribosomal proteins (Fig. 2A, Supplementary Table 2). Consistent with association of eIF3 being physiologically relevant, a previously published high-copy suppressor screen resulted in the isolation of many ribosomal proteins as suppressors of the growth defect of eif3eΔ cells (Supplementary Fig. 2 and (Sha et al., 2007)).
The finding that eIF3 forms stable associations with proteins involved in translation elongation is seemingly at odds with its role in initiation that was established in biochemical reconstitution experiments. Most notably, stable binding of eIF3 to 40S ribosomes was shown to inhibit the joining of the 60S subunit (Kolupaeva et al., 2005). Only upon the action of the eIF5B GTPase on the 48S complex, can 60S join the 40S subunit thereby releasing eIF3 (Unbehaun et al., 2004). Both of these observations are seemingly inconsistent with the interaction of eIF3 and 60S ribosomal proteins revealed here. The eIF3 interactome indicates that the multiple biochemical functions of eIF3 may be coordinated in a complex manner in vivo. For example, it is possible that eIF3 is only temporarily released upon 40S and 60S subunit joining followed by rebinding to the 80S complex, potentially promoted by protein post-translational modifications. Although our data do not prove that eIF3 associates with actively translating 80S ribosomes, a recent report provided experimental evidence for this phenomenon in budding yeast (Szamecz et al., 2008). eIF3 may subsequently orchestrate the recruitment of factors such as the MSC and eEFs, which are required for efficient elongation.
The eIF3 interactome also contained 10 out of 17 subunits of the 19S proteasome regulatory particle (Fig. 2A, ,3D,3D, Supplementary Table 2). The remaining 7 subunits were also identified in some of the purifications, but did not pass abundance-based filtering in all of the six datasets (Fig. 3D). We also identified 6 subunits of the 20S proteasome, albeit at low levels (Supplementary Data File 1). This may be due to the fact that our purifications were performed in the absence of ATP, a condition that is likely to cause the 20S subunits to dissociate from the 19S subunits. These data are consistent with the previous demonstration that S. pombe eIF3e interacts with a subunit of the 19S lid (Yen et al., 2003b). Likewise, eIF3-proteasome interactions were reported in human cells and in plants (Dunand-Sauthier et al., 2002; Hoareau Alves et al., 2002; Paz-Aviram et al., 2008). Molecular chaperones of the HSP70/40 family and the CCT complex, which are known to mediate co-translational protein folding (Albanese et al., 2006; Fedorov and Baldwin, 1997), were also enriched in the eIF3 complex (Fig. 2A, Supplementary Table 2). Lastly, we identified the E2 ubiquitin-conjugating enzyme Ubc4p.
Proteasome recruitment to elongating ribosomes may mediate co-translational degradation of proteins that cannot be properly folded by chaperones. In order to prevent proteotoxicity, this tight spatial linkage may facilitate the rapid removal of the 30 – 50% of newly synthesized proteins that are co-translationally degraded by the proteasome (Schubert et al., 2000; Turner and Varshavsky, 2000). Ubc4p was previously shown to be involved in the ubiquitylation and proteasomal targeting of misfolded nascent proteins (Chuang and Madura, 2005; Seufert and Jentsch, 1990).
The protein interactions revealed here suggest that eIF3 organizes a series of protein complexes that coordinately perform diverse steps in mRNA translation. These interactions may establish cytoplasmic “translation factories” akin to the nuclear transcription factories that coordinate the synthesis and downstream processing of pre-mRNA (Calvo and Manley; Jackson, 2005; Pandit et al., 2008). Likewise, eIF3 appears to coordinate translation initiation, elongation, and quality control through forming an RNA-independent supercomplex with eIFs, MSC, eEFs, 40S and 60S ribosomes, and the proteasome, which we named the “Translasome”.
Notably, we have obtained independent evidence for the translasome upon biochemical purification of the budding yeast proteasome by column chromatography and native gel electrophoresis. Analysis of the 26S proteasome holoenzyme (Glickman et al., 1998) by LC-MS/MS revealed co-purification of the five eIF3 core subunits, twenty three 40S and thirty eight 60S ribosomal proteins, eEF1, and glutamyl-tRNA synthetase (Supplementary Table 3). Similar interactomes were recently encountered upon cross-linking of proteasome interacting proteins of budding yeast (Guerrero et al., 2008) and during purification of the human 26S proteasome (A. Kisselev, personal communication). These data strongly suggest that the core translasome is conserved across three eukaryotic species and is not an artefact of affinity purification of eIF3.
The eIF3 interactome also contained 22 proteins involved in ribosome biogenesis, many of which are components of the small subunit (SSU) processosome (Dragon et al., 2002) (Fig. 2A, Supplementary Table 2). These included U3 snoRNP subunits as well as several subunits of the U3 protein complex required for 35S pre-rRNA transcription (tUTP,) and the UTP-B and UTP-C complexes (Henras et al., 2008) (Fig. 3E). In addition, two proteins involved in modification of pre-rRNA were identified; Fib1p, which mediates 2'- hydroxyl methylation of ribose, and Gar1p, which catalyzes pseudouridylation (Fig. 3E). Furthermore, 7 helicases involved in the maturation of pre-40S and pre-60S ribosomes as well as Rrp12p and SPBC16H5.08c, two factors mediating their nuclear export were identified (Zemp and Kutay, 2007) (Fig. 3E).
These findings suggest that eIF3 is also involved in various steps of ribosome biogenesis. Notably, ribosome biogenesis factors were previously found in eIF3 preparations of budding yeast obtained in systematic protein interaction studies (Gavin et al., 2006; Krogan et al., 2006). These included U3 snoRNP and UTP-C components as well as the nuclear export factor Rli1p. In addition, budding yeast eIF3j/Hcr1p has dual roles in translation initiation and 20S pre-rRNA processing (Valasek et al., 2001).
Beside proteins involved in protein synthesis, the eIF3 complex also consistently co-purified 42 metabolic enzymes belonging to the KEGG pathways central carbon, carbohydrate, amino acids, fatty acids and lipids, nucleotides, and vitamins and cofactors (Fig. 2A, Supplementary Table 2). This diverse set of functions and the fact that 81% of these proteins are among the 10% of the most abundant proteins in S. pombe ((Schmidt et al., 2007) and data not shown) implies that they may have co-purified as nascent polypeptides. On the other hand, two of the most abundant interactors, the redundant glyceraldehyde 3-phosphate dehydrogenases (GAPDH), Gpd3p and Tdh1p, are multifunctional enzymes associated with a wide variety of glycolysis-independent functions, including membrane fusion, phosphotransferase activity, nitric oxide sensing, and nuclear RNA export among others (Sirover, 2005). It is thus possible that GAPDHs are genuine translasome components.
The eIF3 purifications also contained substantial amounts of actin and the actin regulators Hob3p and Cpc2p/Rack1 (Fig. 2A, Supplementary Table 2). The latter was also isolated as an eIF3d/Moe1p binding protein in a previously reported yeast two-hybrid screen (Chen et al. 2000). Interactions of the protein synthesis machinery with cytoskeletal components have been recognized for decades (reviewed in (Hovland et al., 1996)). Ribosomes, eIF2, eIF4A and B, and eIF3 form cytochalasin D sensitive physical interactions with the cytoskeleton in HeLa cells (Howe and Hershey, 1984). Likewise, eEF2 and eEF1A are high affinity F-actin binding proteins (Bektas et al., 1994; Yang et al., 1990). The integrity of the filamentous actin network is critical to normal protein synthesis in mammalian cells (Stapulionis et al., 1997), presumably because it locally organizes components of the translation machinery (Liu et al., 1996). For example, human eIF3a interacts with actin during localization to the ER membrane (Pincheira et al., 2001). Interactions with actin may therefore be involved in localizing the translasome to particular cellular compartments.
eIF3 co-purified with several proteins involved in intracellular vesicle trafficking (Fig. 2A, Supplementary Data File 1) and cellular transport mechanisms, most notably membrane transport, mitochondrial transport, and nuclear transport (Fig. 4A). Importin-β molecules are critical for nuclear transport because they can, either independently or via importins-α, bind the cargo as well as the nuclear pore complex. There are 13 known importins-β in the S. pombe genome (http://www.sanger.ac.uk/Projects/S_pombe/; (Chook and Blobel, 2001; Chua et al., 2002)); however, only two of them, Kap123p and Sal3p, which are orthologs of budding yeast Kap123p and Kap121p/Pse1p, are components of the eIF3 interactome. We assessed the physiological relevance of the eIF3–importin-β interactions in a series of genetic experiments in eif3eΔ cells, which, unlike deletion mutants of most other eif3 genes, are viable. We first generated eif3eΔ/+ kap123Δ/+ and eif3eΔ/+ sal3Δ/+ diploid strains and induced them to sporulate. Tetrad analysis showed that kap123Δ eif3eΔ haploids divided only a few times after germination, indicating synthetic lethality (Fig. 4B). Although the sal3Δ eif3eΔ double mutant was viable, it grew more slowly than the individual single mutants, a growth defect that could be readily detected at elevated temperature (Fig. 4C). We selected three other non-essential importins-β that do not bind eIF3e (Kap111p, Kap114p, and Kap104p, (Chen et al., 2004)) but found no genetic interactions with eif3e (data not shown). Since eIF3 associates with both Kap123p and Sal3p, we tested for genetic interaction between these two genes. Indeed, sal3Δ kap123Δ double mutants grew very slowly (Fig. 4D). These data indicate that the association of eIF3 with Kap123p and Sal3p is critical for normal cell growth.
Proteasomes concentrate in the nucleus and at the nuclear membrane (Wilkinson et al., 1998). We have previously shown that eIF3e is required for proper proteasome accumulation in the nucleus (Yen et al., 2003b). As a consequence, eif3e mutants, like proteasome mutants, are hypersensitive to canavanine, an arginine analog, whose incorporation into nascent proteins triggers their removal through proteasomal degradation. To assess the significance of the physical and genetic interactions between eIF3e and Sal3p, we determined their possible cooperation in regulating proteasome function. As shown in Fig 5A, cells deleted for sal3, but not those deleted for kap111, which encodes an importin-β that did not co-purify with eIF3, were highly sensitive to canavanine (8 mg/L). A lower concentration of canavanine (4 mg/L) that only mildly affected the sal3Δ single mutant, still severely impaired the growth of the sal3Δ eif3eΔ double mutant. Taken together, these findings suggested that eIF3e cooperates with Sal3p in proper proteasome function, possibly by regulating its nuclear localization.
To test this possibility, the 19S lid subunit Rpn7p was tagged with GFP by homologous recombination resulting in a fusion protein, which was previously shown to be fully functional and integrated into the 26S proteasome (Sha et al., 2007). Rpn7p-GFP expressed from the endogenous promoter is thus suitable for monitoring the subcellular localization of the entire proteasome. In wildtype cells optically scanned across the mid-section by confocal microscopy, the Rpn7p-GFP signal was most highly concentrated at the nuclear membrane (Fig 5B). Rpn7p-GFP was also readily detectable in the nucleoplasm, but excluded from an area that is presumed to be the nucleolus. The phenotypes of sal3Δ eif3eΔ cells appear to be pleiotropic and can be devided into two groups. Approximately 70% of sal3Δ eif3eΔ cells resembled eif3eΔ single mutants, in which Rpn7p-GFP was nuclear, but not concentrated at the nuclear membrane (Fig. 5B; (Yen et al., 2003b)). The nuclei of these cells were deformed, a phenotype that can also be observed in eif3eΔ cells when maintained at low temperature. The remaining ~30% of sal3Δ eif3eΔ cells had normal nuclear morphology, but contained very little Rpn7p-GFP in the nucleoplasm (Fig. 5B). Cells with this deficiency were undetectable either in wildtype or the single mutants (Fig. 5B and data not shown). In summary, a significant proportion of sal3Δ eif3eΔ cells appear severely deficient in localizing proteasomes to the nucleus.
To directly assess the efficiency of proteasome nuclear accumulation, we photobleached the entire nucleus and then measured the reappearance of nuclear Rpn7p-GFP over time by quantifying the relative abundance of the GFP signal in the nucleus versus the cytoplasm. While Rpn7p-GFP nuclear accumulation was only slightly inhibited in the sal3Δ single mutant, it was substantially reduced in sal3Δ eif3eΔ cells belonging to the fraction of 30% that had very low nuclear Rpn7p-GFP (Fig. 5C). Inefficient accumulation in the nucleus would be expected to impair nuclear functions of the proteasome such as its role in DNA double strand break repair (Krogan et al., 2004; Tatebe and Yanagida, 2000). Indeed, sal3Δ eif3eΔ cells were far more sensitive to phleomycin, a chemical that induces DNA double-strand breaks, than either of the single mutants (Fig. 5D).
These results demonstrate that proper nuclear localization of the proteasome (Wilkinson et al., 1998) requires eIF3e in addition to Sal3p. The Sal3p-eIF3e cooperation may reflect the two step mechanism of proteasome nuclear accumulation, which requires (i.) passage through the nuclear pore, and (ii.) anchorage to the nuclear membrane by Cut8p (Takeda and Yanagida, 2005; Yen et al., 2003a). Sal3p appears to mainly regulate passage through the nuclear membrane, since the kinetics of Rpn7p-GFP accumulation in the nucleus was slower in sal3Δ cells than in wild type cells (Fig. 5C). eIF3e may primarily control nuclear retention, as we have previously shown that cut8Δ is synthetically growth deficient with eif3eΔ (Yen et al., 2003b).
As with sal3Δ cells, kap123Δ cells were hypersensitive to both canavanine (Fig 5A) and phleomycin (Fig. 5E), suggesting that Kap123p may play a role in proteasome nuclear import that is partially redundant with Sal3p. The lethality of kap123Δ eif3eΔ cells suggests that eIF3 may interact with Kap123p to mediate nuclear trafficking of additional cargos, such as ribosomes and ribosome biogenesis factors, whose nuclear functions may be essential for cell viability.
Our eIF3 interactome data suggest that protein complexes that are necessary for protein synthesis and degradation can form a supercomplex, which we named the translasome (Fig. 6). While surfaces of the individual protein complexes may mediate their interactions, the data from this and other studies suggest that the actin cytoskeleton may provide additional physical support. The translasome is proposed to spatially coordinate distinct steps of protein synthesis thus contributing to the efficiency of mRNA translation. The integration of proteasomes in translasomes could ensure translational fidelity by enabling the timely removal of abnormal nascent proteins.
Although the bulk of eIF3 and other translasome components are cytoplasmic at steady state, our data suggest that translasomes are dynamically localized within the cell because they associate with nuclear ribosome biogenesis proteins and with importins-β. In support of the possibility that eIF3 shuttles between the cytoplasm and the nucleus is the observation that eIF3e concentrates in the nuclei of mammalian cells in a cell cycle-dependent manner (Watkins and Norbury, 2004). The nuclear trafficking of eIF3e is in part mediated by a conserved leucine-rich NES at its N-terminus (Guo and Sen, 2000). Likewise, point mutations in the PCI domain lead to nuclear accumulation of eIF3e in HeLa cells (Sha et al., 2007). In S. pombe, eIF3e is readily detected in the nucleus, when eif3d/moe1 is deleted; conversely, eIF3d becomes nuclear, when eif3e is missing (Yen and Chang, 2000). Finally, budding yeast eIF3a, an essential eIF3 core component, was identified as an import cargo of Sal3p/Kap121p and Kap123p (Leslie et al., 2004).
These considerations raise the intriguing question of what role nuclear transit of eIF3 may play. Our findings suggest that one of these functions is to promote the proper nuclear accumulation of the proteasome. Whereas Sal3p and Kap123p mediate its transport, nuclear eIF3 appears to be involved in tethering the proteasome to the nuclear membrane through Cut8p (Yen et al., 2003b). Several lines of evidence suggest that eIF3 then disassociates from the proteasome before it can move on to its second nuclear function, which our interaction data suggest is in ribosome biogenesis. Firstly, eIF3 subunits do not assume the same nuclear rim association as the proteasome, arguing against the maintenance of stable eIF3-proteasome interactions in the nucleus. Secondly, the proteasome is apparently excluded from nucleoli (Fig. 5B) into which eIF3 would have to move in order to assist in ribosome biogenesis. We surmise that eIF3 may serve as an assembly platform for 90S pre-ribosomal particles in the nucleolus and facilitate nucleoplasmic pre-40S and pre-60S ribosome maturation. A mild 40S biogenesis defect was recently described in a budding yeast eIF3a partial loss-of-function mutant (Szamecz et al., 2008). Since eIF3 is directly bound to mature 40S ribosomes in the cytoplasm, it may be co-exported from the nucleus together with pre-40S particles, thus resulting in a short nuclear transit time. Consistent with this possibility is the presence of ribosome export factors in the purified eIF3 complexes.
Although we have no direct evidence for the proposition that eIF3 and the proteasome translocate into the nucleus as part of the entire translasome, we note that both Sal3p and Kap123p were previously shown to regulate the nuclear import of many ribosomal proteins (Leslie et al., 2004; Rout et al., 1997). The nuclear pore allows passage of molecules up to ~10 MDa in size (Gorlich and Kutay, 1999), and would therefore be able to accommodate the translasome with its estimated size of 7–8 MDa. If nuclear translocation of a fraction of intact translasomes occurred, this could explain the low levels of translation factors that have been observed in nuclei and at sites of transcription (Brogna et al., 2002; Iborra et al., 2004a; Lund and Dahlberg, 1998). It might even enable a modest level of nuclear translation and nonsense-mediated mRNA decay (Buhler et al., 2002; Iborra et al., 2001), although the widespread occurrence of these processes in nuclei remains a matter of debate (Dahlberg and Lund, 2004; Dahlberg et al., 2003; Iborra et al., 2004b), and references therein).
Cells were grown in either yeast extract medium (YEAU) or minimal medium (MM) with appropriate supplements (Chen et al., 1999). All experiments were with cells pre-cultured to early logarithmic phase (2–5 × 106 cells/ml). For growth experiments on plates, cells were serially diluted 1:5. To test sensitivity to DNA damage, phleomycin stock solution (5 mg/ml, Sigma) was prepared in DMSO, and controls included DMSO lacking phleomycin.
All importins-β mutants used in this study were kindly provided by D. Balasundaram (Chen et al., 2004), except as described below. To measure Rpn7p localization and import in sal3Δ or sal3Δ eif3eΔ background, strains SAL3U (sal3Δ) and Y6AR7GFP (yin6/eif3eΔ rpn7-gfp, (Sha et al., 2007)) were crossed to generate sal3Δ/+ eif3eΔ/+ rpn7-gfp diploid cells. These cells were then induced to sporulate, and tetrad dissection was performed to isolate rpn7-gfp, sal3Δ rpn7-gfp, eif3e/yin6Δ rpn7-gfp, and sal3Δ eif3eΔ rpn7-gfp cells, and these strains were named N7G, SAL3UN7G, Y6AN7G, and S3UY6AN7G. Strain KAP123C was generated by a PCR-based gene deletion method using ClonNat as the selectable marker (Gregan et al., 2006).
eif3e-proA and eif3m-proA cells (strains C648 and C617/1), and their WT parental cells (strain DS448/1) were described previously (Zhou et al. 2005). The affinity purification procedures were as described with minor modifications (Zhou et al. 2005). Briefly, cells were collected from 2 L cultures at OD595 = 0.6 and disrupted with glass beads in 5 ml lysis buffer (50 mM Tris-HCl pH 7.4, 140 mM NaCl, 0.5% Triton X-100, 10 µg/ml leupeptin, 10 µg/ml pepstatin, and 1mM phenylmethylsulfonylfluoride). Crude lysates were centrifuged twice at 14,000 rpm for 20 min at 4°C, and 100 mg cleared protein in a volume of 5 ml was mixed with 300 µl Dyna beads (Dynal Biotech) coupled to rabbit IgG (Jackson Immunochemicals). Mock purifications were performed in parallel with lysate from untagged cells. Beads were collected and washed 4 times in 5 ml lysis buffer. Beads were cleaved overnight with 300 U TEV protease (Invitrogen). Protein aliquots were separated by SDS-PAGE, and visualized by Coomassie Blue staining or by Western blotting with anti-Rpn1 antibody (1:1,000, (Yen and Chang, 2000)). To quantify protein levels in Western blots, fluorescently conjugated secondary antibodies were used, and the signals were measured by the Odyssey infrared imaging system (Li-COR Biosciences).
Proteins from TEV eluates were reduced, alkylated, and digested with trypsin using standard procedures (see Supplementary Methods for details). Peptides were desalted with a C18 cartridge (Waters). Peptides were dried, re-suspended in 0.1% trifluoroacetic acid/2.0% acetonitrile, and stored at 4 °C until LC-MS/MS analysis. The analyses used a Paradigm HPLC/autosampler (Michrom Bioresources, Inc.) and an LTQ OrbitrapXL mass spectrometer (Thermo Fisher Scientific). A C18 analytical column and an ADVANCE source (Michrom) were used. The RP HPLC gradient (solvent A = 0.1% formic acid; solvent B = 100% acetonitrile) consisted of 2% B to 5% B from 0 to 2.0 min and 5% B to 35 % B from 2.1 to 120.0 min. The MS/MS method was top-4, data-dependent; precursors were scanned in the Orbitrap and MS/MS scans were in the ion trap. Dynamic exclusion was enabled. Data was searched against an S. pombe protein database using Sorcerer™-SEQUEST® (SageN Research). Static alkylation of Cys, and differential Met oxidation, Lys ubiquitination, and Ser, Thr and Tyr phosphorylation were specified. QTools, which are in-house developed visual basic macros for automated spectral count analysis, were used to compute spectral counts of the proteins (Liu et al., 2004).
LC-MS/MS analyses of the budding yeast 26S proteasome were performed in a similar manner, and are described in Supplementary Methods.
Rpn7p-GFP localization was visualized by confocal microscopy as described in the Supplementary Methods section.
We thank D. Balasundaram (Singapore University) for providing materials, M. Petroski for reviewing the manuscript, and K. Motamedchaboki and A. Iranli for bioinformatics support. We also thank Mike Mancini and his staff at the Integrated Microscopy Core at the Dan Duncan Cancer Center for technical assistance. This work was funded by NSF grant 0920229 and NIH grant GM059780 to D.A.W. and by NIH grants CA90464 and CA107187 to E.C.C; ZS was supported by a pre-doctoral fellowship from the DOD (BC030443). LMB is funded through the NIH Center Grants 5 P30 CA30199-28 and 5 P30 NS057096. We also thank Glen and Judy Smith for their generous support.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.