Proteasome-Mediated Degradation of Cotranslationally Damaged Proteins Involves Translation Elongation Factor 1A

Abstract

Rad23 and Rpn10 play synergistic roles in the recognition of ubiquitinated proteins by the proteasome, and loss of both proteins causes growth and proteolytic defects. However, the physiological targets of Rad23 and Rpn10 have not been well defined. We report that rad23Δ rpn10Δ is unable to grow in the presence of translation inhibitors, and this sensitivity was suppressed by translation elongation factor 1A (eEF1A). This discovery suggested that Rad23 and Rpn10 perform a role in translation quality control. Certain inhibitors increase translation errors during protein synthesis and cause the release of truncated polypeptide chains. This effect can also be mimicked by ATP depletion. We determined that eEF1A interacted with ubiquitinated proteins and the proteasome following ATP depletion. eEF1A interacted with the proteasome subunit Rpt1, and the turnover of nascent damaged proteins was deficient in rpt1. An eEF1A mutant (eEF1A^D156N) that conferred hyperresistance to translation inhibitors was much more effective at eliminating damaged proteins and was detected in proteasomes in untreated cells. We propose that eEF1A is well suited to detect and promote degradation of damaged proteins because of its central role in translation elongation. Our findings provide a mechanistic foundation for defining how cellular proteins are degraded cotranslationally.

Rad23 and Rpn10 can interact with multiubiquitinated proteins (6, 28, 38) and the proteasome (32), and several recent studies have indicated that they contribute to the degradation of ubiquitinated proteins by the proteasome (6, 9, 21, 24). Loss of both proteins results in temperature-sensitive growth, defects in proteolysis, and a delay in the G₂ phase of the cell cycle (22). We isolated yeast TEF1, a gene encoding the eukaryotic translation elongation factor 1A (eEF1A) (15), as a dosage suppressor of the cold (13°C) sensitivity of rad23Δ rpn10Δ (22). eEF1A promotes translation elongation through the binding and release of aminoacyl tRNAs, in a process that is coupled to GTP hydrolysis (15). In addition to its well-characterized role in translation elongation, in vitro studies showed that eEF1A could bind nascent as well as unfolded peptides and proteins (17, 23). eEF1A might possess a chaperone-like activity that prevents the aggregation of nascent polypeptide chains (4), since it could bind an unfolded protein but not a correctly folded counterpart (17). eEF1A could also stimulate the degradation of Nα-acetylated proteins (12). The isolation of eEF1A as a suppressor of rad23Δ rpn10Δ suggested that it performs a central role in monitoring the accuracy of protein synthesis. These studies also revealed an important function for Rad23 and Rpn10 in protein synthesis quality control.

A significant fraction of newly synthesized proteins is degraded cotranslationally (29, 33, 35). These nascent damaged proteins can be ubiquitinated while bound to the ribosome (31), demonstrating that there exists a close coupling between the pathways of protein synthesis and protein degradation. Previous studies showed that both stable and unstable proteins were susceptible to cotranslational degradation by the proteasome (35). However, the mechanism that permits recognition of nascent misfolded proteins by the ubiquitin conjugating system has not been determined. Additionally, it is unknown how ubiquitinated nascent proteins are rapidly transferred to the proteasome. Since eEF1A can bind nascent polypeptide chains after their release from the ribosome (17), it is possible that damaged polypeptide chains are escorted to the proteasome by eEF1A and other regulatory factors. Ubiquitin (Ub)-binding proteins have been shown to play an important role in subcellular protein trafficking (14). For instance, Rad23 can bind ubiquitinated proteins and the proteasome and was reported to play a role in the transfer of damaged proteins from the endoplasmic reticulum to the proteasome (25). It is conceivable that Rad23 might play a similar role in the translocation of damaged nascent proteins to the proteasome. In agreement with this hypothesis, we found that rad23Δ rpn10Δ is extremely sensitive to translation inhibitors, suggesting that Rad23 and Rpn10 promote the degradation of damaged translation products. Furthermore, we show that mutations in both eEF1A and the proteasome subunit Rpt1 can have an effect on cotranslational protein degradation.

MATERIALS AND METHODS

Plasmids and strains.

Plasmids that expressed glutathione S-transferase (GST)-Rad23, GST-UBA1, GST-UBA2, GST-UbL, and Pre1-FLAG were as described in reference 7. The GST fusion proteins were constructed in pCBGST1, as described in Schauber et al. (7, 32). GST-eEF1A, GST-S5a, and GST-Rpt1 were prepared in pGEX2TK from human cDNAs due to the instability of the yeast proteins. Pre1-FLAG was provided by J. Dohmen (University of Cologne) and was cloned in a 2μm plasmid, bearing a LEU2 selectable marker. The copper-inducible P_CUP1 promoter drives the expression of Pre1-FLAG, and the protein contains a FLAG epitope at the carboxy terminus. Yeast strains expressing eEF1A mutants were previously described (5, 8). Yeast strains lacking RAD23 and RPN10 were also described previously (22). Sensitivity to translation inhibitors was determined after spotting 10-fold dilutions of exponential-phase cultures and incubating at 24 to 30°C for 4 days. For studies with cim5-1/rpt1 (termed rpt1 hereafter), cultures were grown at 24°C and then transferred to semipermissive (30°C) or nonpermissive (37°C) temperatures. The expression of Met-β-galactosidase (βgal) and Ub-Pro-βgal was achieved by growing cells in 2% galactose.

Antibodies.

Polyclonal and monoclonal antibodies against ubiquitin and FLAG were purchased from Sigma Chemical Co. (St. Louis, Mo.). Monoclonal antihemagglutinin (anti-HA) antibodies were purchased from BabCo and human anti-Rpt1 antibodies were from Affiniti Inc., Devon, United Kingdom. We generated antibodies against yeast GST-Rpt1. Monoclonal anti-βgal antibodies were purchased from Promega (Madison, Wis.) and anti-His₆ antibodies from EMD Biosciences (La Jolla, Calif.). The antibodies were used at dilutions recommended by the manufacturers. Recombinant protein A Sepharose beads were obtained from Repligen and FLAG-agarose from Sigma. Antibodies against eEF1A were generated against the full-length protein in rabbits. Pab1 antibodies were kindly provided by S. Peltz (Robert Wood Johnson Medical School).

Preparation of protein extracts and immunological methods.

Yeast strains containing plasmids were grown in synthetic medium, pelleted, and frozen at −70°C. Yeast cells were suspended in buffer A (50 mM HEPES, pH 7.5, 150 mM NaCl, 5 mM EDTA, and 1% Triton X-100 with the addition of protease inhibitors [including Pefabloc SC, leupeptin, aprotinin, antipain, pepstatin, and chymostatin]) and lysed by glass bead disruption. For gel filtration, protease assays, and isolation of intact 26S proteasome, 4 mM ATP was included in buffer A. Extracts were normalized to equal protein concentration (Bradford, Bio-Rad) and volume and applied directly to either glutathione Sepharose 4B (Pharmacia, Piscataway, N.J.) or anti-FLAG-M2 affinity agarose (Sigma), depending on the protein to be isolated. Beads were washed three times with buffer A, and the bound proteins were released by boiling and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to nitrocellulose, and subjected to immunoblotting.

ATP depletion.

Yeast cells expressing specific proteins (Pre1-FLAG, GST-UBA1, GST-UBA2, and GST-UbL) were grown to exponential phase and treated with 0.2 mM 2,4-dinitrophenol (2,4-DNP) and 20 mM 2-deoxy-d-glucose (2-DG) at 30°C for 1 to 2 h. Aliquots were removed during the course of the incubation for analysis, and equal amounts of protein (500 μg) were immunoprecipitated with FLAG-agarose (or glutathione-Sepharose). The levels of eEF1A, Rpt1, Met-βgal, and ubiquitin were determined by immunoblotting. To examine expression of these proteins in total extracts, 50 μg of protein was resolved in SDS-PAGE. Translation inhibitors were from Sigma.

Protease assays and stability measurements.

Total cell lysates from yeast cells were prepared as described above with the addition of 4 mM ATP, normalized to equal protein content, and assayed for chymotrypsin-like protease activity, with the fluorogenic substrate Suc-LLVY-AMC (Boston Biochem), using a Turner TD-700 fluorometer (Turner Designs, Inc., Sunnyvale, Calif.). Pulse-labeling measurements were described previously (27).

Gel filtration.

Lysates (containing 2 mg of protein) were separated by gel filtration chromatography using a Superose 6 10/30 HR column (Pharmacia) in 25 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 1 mM ATP, and 10% glycerol. One-milliliter fractions were collected and 300 μl was precipitated with 10% trichloroacetic acid, solubilized in SDS-containing sample buffer, and resolved by polyacrylamide gel electrophoresis.

RESULTS

Isolation of eEF1A as a suppressor of rad23Δ rpn10Δ.

Both low-copy-number (CEN) and high-copy-number (2μm) expression of eEF1A suppressed the temperatture-sensitive growth defect of rad23Δ rpn10Δ (22) (Fig. 1A). Because eEF1A performs an essential role in translation elongation, we investigated if Rad23 and Rpn10 were required for regulating protein synthesis. The growth of rad23Δ rpn10Δ was examined in medium containing the protein synthesis inhibitor cycloheximide, as well as hygromycin B and paromomycin, which are drugs that can cause premature release of polypeptide chains. Expression of eEF1A restored normal growth in rad23Δ rpn10Δ on medium containing each of these drugs (Fig. 1B), consistent with a role for Rad23 and Rpn10 in protein synthesis quality control.

FIG. 1.

eEF1A can suppress the sensitivity of rad23Δ rpn10Δ to protein synthesis inhibitors. (A) eEF1A (TEF1) was expressed in rad23Δ rpn10Δ on low-copy-number (CEN) and high-copy-number (2μm) plasmids. Ten-fold serial dilutions were spotted onto agar medium and incubated at the permissive (24°C) and semipermissive (30°C) temperatures. (B) Wild-type and rad23Δ rpn10Δ mutant strains were transformed with a control 2μm plasmid, or the same plasmid expressing eEF1A, and spotted on medium containing cycloheximide (0.001 mM), hygromycin B (0.2 mM), or paromomycin (1 mM). WT, wild type.

eEF1A interacts with the proteasome.

Rad23 and Rpn10 contribute to the recognition of ubiquitinated proteins by the proteasome (6, 9, 21, 37). A proteomic survey also revealed an interaction between eEF1A and the proteasome (36), although the significance or specificity of this binding was not determined. To investigate the mechanism of eEF1A suppression of rad23Δ rpn10Δ, we tested the hypothesis that its interaction with the proteasome could provide a way to bypass the defect of rad23Δ rpn10Δ and restore a link between protein synthesis and protein degradation. We immunoprecipitated Pre1-FLAG and confirmed that the 19S proteasome subunit Rpt1 was copurified (Fig. 2A). eEF1A was also copurified, but not in a control reaction that lacked Pre1-FLAG (see lanes 1 and 4). In the reciprocal experiment eEF1A was immunoprecipitated and Pre1-FLAG was successfully copurified (Fig. 2B). The interaction between eEF1A and the proteasome did not require Rad23 (Fig. 2A and B, lane 6), or Rpn10 (data not shown), consistent with its ability to suppress the defects of rad23Δ rpn10Δ. To further verify this interaction, we immunoprecipitated the proteasome using antibodies against a component of the 19S regulatory particle (Rpn8-V5), and in agreement with the previous results eEF1A and a 20S subunit, Pup1-HA, were detected. (Pup1-HA was expressed from a high-copy-number plasmid, and its apparent lower interaction with the proteasome does not indicate substoichiometric binding.)

FIG. 2.

eEF1A interacts with the 26S proteasome. (A) Pre1-FLAG was expressed in wild-type (WT) (lanes 2 and 5) and rad23Δ cells (lanes 3 and 6). Wild-type cells lacking Pre1-FLAG were also examined (lanes 1 and 4). Pre1-FLAG was immunoprecipitated (IP) and an immunoblot was probed with antibodies against eEF1A and Rpt1. (B) eEF1A was immunoprecipitated from extracts described in panel A, and an immunoblot was incubated with anti-FLAG antibody. (C) A yeast strain expressing epitope-tagged 20S subunit (Pup1-HA) and 19S subunit (Rpn8-V5) were grown to exponential phase and extracts incubated with V5 antibodies. An immunoblot was sequentially incubated with antibodies against eEF1A and HA. (D) Total cell extracts (2 mg) were resolved in Superose 6, and aliquots were examined by immunoblotting. The sedimentation positions of eEF1A, Pre1-FLAG, and Rpt1 were determined by immunoblotting. (E) Pre1-FLAG was immunoprecipitated from yeast cells expressing wild-type or mutant eEF1A proteins. The coprecipitation of eEF1A and Rpt1 with Pre1-FLAG was determined.

We resolved yeast proteins in Superose 6 and detected eEF1A in fractions that contained the proteasome subunits Rpt1 and Pre1-FLAG (Fig. 2D). Because the ribosome is a large complex that could conceivably cofractionate with the proteasome, we immunoprecipitated Pre1-FLAG from the Superose 6 fractions and confirmed that eEF1A was copurified (data not shown), consistent with the results in Fig. 2A to C. The sedimentation position of the proteasome was determined by measuring the hydrolysis of Suc-LLVY-AMC in each fraction. Peak levels were detected in fractions 7 to 13 (Fig. 2D).

eEF1A mutants were identified in a genetic search for alleles with reduced fidelity (5). However, the derivatives we characterized did not appreciably affect protein translation rates. To determine if the interaction with the proteasome was altered for well-characterized eEF1A mutants, we immunoprecipitated Pre1-FLAG and compared the copurification of Rpt1 and eEF1A mutant proteins (Fig. 2E). Equivalent levels of Rpt1 were purified from cells that expressed eEF1A^D156N, eEF1A^N153T, eEF1A^N153T/D156E, and eEF1A^E286K (lanes 6 to 10). Only the eEF1A^E286K mutant showed poor interaction with the proteasome (lane 10). However, this derivative was expressed at lower levels in the cell (lane 5).

eEF1A interacts with Rpt1 in the 19S regulatory particle.

Because eEF1A is expressed at high levels, it could form a nonspecific interaction with the proteasome. To address this concern we characterized eEF1A-proteasome interaction in yeast strains harboring defects in several ATPase subunits, including Rpt1 (cim5-1), Rpt4 (rpt4-1), and Rpt6 (cim3-1), as well as the non-ATPase subunits Rpn2 (sen3-1) and Rpn10 (rpn10Δ). The expression of eEF1A and Rpt1 was not altered in these strains (Fig. 3A; also data not shown). However, the copurification of eEF1A with Pre1-FLAG was specifically reduced in rpt1 (Fig. 3A, lane 9). In contrast, eEF1A-proteasome interaction was not altered in rpt4, rpt6, rpn2, and rpn10Δ (Fig. 3A; also data not shown). The failure of eEF1A to bind the proteasome efficiently in rpt1 is not due to reduced proteasome stability, because the copurification of the wild-type Rpt1 and mutant rpt1 proteins with Pre1-FLAG was similar (Fig. 3A, compare lanes 8 and 9).

FIG. 3.

eEF1A interacts with proteasome subunit Rpt1. (A) Pre1-FLAG was purified from wild-type (WT; lanes 1, 3, 6, and 8), rpn2 (lanes 2 and 7), rpt1 (lanes 4 and 9), and rpt6 (lanes 5 and 10), and the copurification of eEF1A with the proteasome was determined. The strain background for rpn2 differed from that harboring rpt6 and rpt1 mutations. (B) rpt1 was grown at 24, 30, and 37°C, and Pre1-FLAG was precipitated. The levels of eEF1A and Rpt1 were examined. Stability of the proteasome at high temperature (37°C) was unaffected, as determined by the constant levels of Rpt1 that was precipitated from the wild-type and rpt1 strains. (C) Rpt1 was isolated from Escherichia coli and 100 ng of the purified protein was resolved in the right lane. Rpt1 was applied to immobilized GST-eEF1A, GST-Rpn8, GST-S5a, and GST, and >90% of the input Rpt1 bound GST-eEF1A. (D) The positions of purified Rpt1 (panel 3) and GST-eEF1A (panel 2) in Superose 6 are compared to bovine serum albumin (panel 1). Ten-fold less Rpt1 and GST-eEF1A were combined (to minimize aggregation or nonspecific binding), incubated at 4°C for 2 h, and resolved in Superose 6 (panels 4 and 5). Fractions were incubated with glutathione-Sepharose (GST pulldown [GST-PD]) and Rpt1 was coprecipitated with GST-eEF1A (panels 6 and 7). Dashed lines indicate the position of free eEF1A. IP, immunoprecipitation.

Based on the reduced eEF1A-proteasome interaction in rpt1, we examined the binding at the nonpermissive temperature. Yeast cells were grown at 24°C for 18 h and then shifted to 24, 30, and 37°C. Protein extracts were prepared after 6 h incubation, and Pre1-FLAG was purified (Fig. 3B). Although the levels of Rpt1 and eEF1A did not change at higher temperature (see Extract panel), we detected ∼4-fold reduced coprecipitation (of eEF1A) with Pre1-FLAG at high temperature in rpt1, while the interaction increased ∼2-fold in wild-type cells (lower panel). Consistent with the results in Fig. 3A, proteasomes were not destabilized at high temperature in rpt1 (Fig. 3B).

To confirm these results we investigated if Rpt1 interacted directly with eEF1A. Rpt1 was purified from bacteria and applied to affinity matrices that contained GST-eEF1A, GST-Rpn8, GST-S5a, and GST. These studies were conducted with human proteins due to instability of the yeast counterparts. Following 4 h incubation at 4°C, the matrix was washed, and the bound proteins were resolved by SDS-PAGE. An immunoblot was examined and we discovered that Rpt1 interacted only with GST-eEF1A (Fig. 3C). Significantly, the binding was quantitatively efficient, as >90% of the input Rpt1 protein was recovered with GST-eEF1A. We showed earlier (Fig. 2D) that eEF1A and Rpt1 could be cofractionated in Superose 6. To verify a direct interaction between GST-eEF1A (70 kDa) and Rpt1 (51 kDa) we applied purified proteins on Superose 6 (Fig. 3D). GST-eEF1A (0.5 μg; second panel) was detected in the same position as bovine serum albumin (69 kDa; first panel). SDS-PAGE revealed extensive degradation of eEF1A, which is known to be highly unstable (26). The degradation appeared to occur primarily after the chromatography step. Surprisingly, purified Rpt1 (0.5 μg) migrated in a high-molecular-weight region of the chromatogram (50 to 300 kDa; third panel). Because certain members of this class of AAA-ATPases can form hexameric rings (∼300 kDa) (16), it is conceivable that purified Rpt1 can oligomerize. We combined 10-fold less GST-eEF1A and Rpt1 (compared to the amount used in panels 2 and 3), to minimize nonspecific interactions, and incubated the reaction at 4°C for 2 h. The sample was then resolved in Superose 6, and GST-eEF1A was detected in a higher-molecular-weight fraction (compare panels 2 and 4). Significantly, this low level of GST-eEF1A was protected from degradation (fourth panel), and the mobility of Rpt1 was also altered (compare third and fifth panels). Incubation of the fractions with glutathione-Sepharose resulted in the isolation of Rpt1 (seventh panel) and the copurification of GST-eEF1A (sixth panel). Collectively, these results demonstrate that eEF1A can interact directly with Rpt1.

eEF1A associates with ubiquitinated proteins following ATP depletion.

Simultaneous disruption of oxidative phosphorylation and reduction of glucose levels can deplete intracellular ATP, terminate translation prematurely, and release misfolded, nascent polypeptide chains (1, 10, 11). Given its critical role in translation elongation, eEF1A might recognize damaged proteins and link protein synthesis to cotranslational protein degradation. Since nascent chains can be ubiquitinated while still associated with the ribosome (31), we investigated if eEF1A could bind ubiquitinated proteins that are associated with the UBA1 domain in Rad23 (6). We confirmed that ubiquitinated proteins could be purified with GST-UBA1 from both treated and untreated cells (Fig. 4A, lanes 3 and 4). The same immunoblot was incubated with antibodies against eEF1A, and >20-fold increased association with GST-UBA1 was detected, specifically in extracts prepared from ATP-depleted cells (Fig. 4B, lane 8). Longer exposures revealed an interaction in the untreated cells as well, but no binding was detected with GST. In a 2-h time course study, ATP depletion resulted in progressively higher interaction between eEF1A and GST-UBA1 (Fig. 4C, GST pulldown). Pre1-FLAG was immunoprecipitated from the same extracts, and we observed ∼25-fold increased eEF1A-proteasome interaction (Fig. 4C, FLAG-IP panel). However, the copurification of Pre1-FLAG with GST-UBA1 (upper panel) and the copurification of Rpt1 with Pre1-FLAG (lower panel) decreased progressively, revealing the slow dissociation of the 19S and 20S particles when ATP was depleted.

FIG. 4.

eEF1A interacts with ubiquitinated proteins, following ATP depletion. (A) GST and GST-UBA1 were expressed in wild-type cells that were either untreated (−) or subjected to ATP-depleting conditions (+). Protein extracts were incubated with glutathione-Sepharose and precipitated material was resolved by SDS-PAGE and transferred to a nitrocellulose filter. The interaction between GST-UBA1 and high-molecular-weight Ub cross-reacting material [Ub_(n)] was confirmed in both extracts (lanes 3 and 4; arrow). Ubiquitinated proteins were not recovered with GST (lanes 1 and 2). (B) eEF1A was purified with GST-UBA1 (lane 8) only from ATP-depleted extracts (lane 8), although it was expressed at equivalent levels in both conditions (lanes 1 to 4). (C) GST-UBA1 was purified from untreated and ATP-depleted extracts, and the copurification of eEF1A and Pre1-FLAG was determined over time. Increasing amounts of eEF1A were bound to ubiquitinated proteins following ATP depletion, while progressively lower levels of Pre1-FLAG were detected with GST-UBA1. Control (Unt; time 0 and time 120 min) samples are present in lanes 1 and 8. Similarly, Pre1-FLAG was purified from the same extracts and the rapid association of eEF1A with the proteasome was confirmed. A detectable loss of Rpt1 over time was observed, consistent with slow dissociation of the 19S and 20S particles. (D) The increased association between eEF1A and the proteasome is also observed following treatment of cells with canavanine (100 μg/ml), neomycin (2 mM), and paromomycin (1 mM). The increased eEF1A-proteasome binding after 2 h was ∼8-fold in the presence of canavanine. Approximately twofold increased binding was observed in the presence of neomycin and paromomycin, respectively. (E) The interaction between eEF1A and various subfragments of Rad23 was examined in the absence (−) and presence (+) of ATP depletion. A strong interaction was detected with intact Rad23 (GST-Rad23; lane 4) and with GST-UBA1 (lane 6). In contrast, eEF1A was not recovered with GST or GST-UBA2, and very low levels were detected with GST-UbL. (F) Purified His₆-eEF1A (100 ng each in lanes 4 to 6) was incubated with 50 ng tetra-Ub, and eEF1A was immunoprecipitated using anti-His₆ antibodies. Although eEF1A was efficiently purified, tetra-Ub was not detected. Lanes 1 and 2 contained 10 and 50 ng of tetra-Ub. Lane 3 did not contain His₆-eEF1A. In lanes 4 to 6, the incubation was performed in the presence of 50, 150, and 250 mM NaCl. IP, immunoprecipitation.

Because ATP depletion could have unforeseen effects on eEF1A-proteasome binding, and proteasome stability, we examined the effect of protein synthesis inhibitors that cause translation errors. Yeast cells were incubated in medium containing canavanine, neomycin, or paromomycin. Protein extracts were examined, and increased association between eEF1A and the proteasome was observed (Fig. 4D), consistent with results obtained following ATP depletion (Fig. 4C). We note, however, that the magnitude and rate of increased eEF1A-proteasome binding was more moderate than when ATP was depleted. We believe this distinction reflects the different severity of the treatment regimens. The cells continued to grow (albeit poorly) in the presence of the translation inhibitors, while cell growth ceased following ATP depletion.

We determined if eEF1A could interact with full-length Rad23, because of its strong association with GST-UBA1 (Fig. 4B). We purified GST, GST-Rad23, GST-UBA1, GST-UBA2, and GST-UbL from yeast and determined that eEF1A was copurified with GST-Rad23. eEF1A was also associated with GST-UBA1 (Fig. 4E), which represents the primary binding site for ubiquitinated proteins in Rad23 (6). eEF1A did not bind GST or GST-UBA2, pointing to the specificity of the interaction, although low levels were purified with GST-UbL (lane 10). This minor level could be associated with ubiquitinated proteins that are bound to the proteasome. The ability to bind UBA1 and purified Rpt1 suggests that eEF1A interaction with ubiquitinated proteins and the proteasome are separable, although potentially coupled, events.

Both multiubiquitin chains and unfolded proteins present large hydrophobic patches on the surface (3). To determine if eEF1A interacted with multi-Ub chains, or hydrophobic domains in damaged proteins, we incubated His₆-tagged eEF1A with tetra-Ub and failed to observe an interaction (Fig. 4F). Similarly, eEF1A did not bind in vitro-ubiquitinated histone H2B (data not shown), suggesting that it interacts primarily with damaged proteins and not multi-Ub chains or artificial protein conjugates.

Newly synthesized proteins and eEF1A interact simultaneously with the proteasome.

It is difficult to detect ubiquitinated proteins in association with the proteasome because they are degraded rapidly. However, if the delivery of proteolytic substrates is significantly increased, as shown previously by overexpressing Rad23 (6), Ub cross-reacting material could be precipitated with Pre1-FLAG. Similarly, we investigated if high levels of translationally damaged proteins that are generated by ATP depletion would be detected in the proteasome. To specifically determine if a newly synthesized protein could be targeted to the proteasome, we examined the fate of Met-βgal, a stable protein (2) that can be degraded cotranslationally by the proteasome (35). We predicted that if ATP depletion generated high levels of damaged Met-βgal fragments they would be targeted to the proteasome. Yeast cells expressing Met-βgal and Pre1-FLAG were incubated with 2,4-DNP plus 2-DG, and protein extracts were characterized. As expected, the levels of Met-βgal in extracts was significantly reduced by ∼40 min after treatment was initiated (Fig. 5A, lane 9). We precipitated Pre1-FLAG and confirmed that eEF1A formed a rapid association with the proteasome (Fig. 5B, upper panel, lane 3; 5 min). Total eEF1A levels were unaffected following ATP depletion (Fig. 5B, lower panel), and eEF1A-proteasome interaction was constant in untreated cells (Fig. 5B, even-numbered lanes). We detected a rapid increase in the levels of Ub cross-reacting material in the proteasome, concurrent with the elevated eEF1A-proteasome binding following ATP depletion (Fig. 5C, upper panel). There was little or no change in the levels of proteasome-associated ubiquitinated proteins in untreated samples (even-numbered lanes). Total extracts were also examined, and increased levels of Ub conjugates were detected following ATP depletion (Fig. 5C, lower panel). This intriguing result shows that the reduction in ATP pools did not block the conjugation of Ub to cellular proteins, although translational damage was evident (see below).

FIG. 5.

eEF1A can bind a normally stable protein in the presence of translation damage. (A) A yeast strain expressing Pre1-FLAG and Met-βgal was incubated with 2,4-DNP plus 2-DG, and culture samples were examined at the indicated times. Protein extract (50 μg) was separated by SDS-PAGE, transferred to nitrocellulose, and incubated with antibodies against βgal. Lane 1 is an untreated sample. Even-numbered lanes represent the untreated samples, while odd-numbered lanes show the progressive inhibition of protein synthesis following ATP depletion. (The same protein extracts were examined in panels A to D, and the lane numbers indicated at the bottom of panel D correspond to all the panels.) (B) Pre1-FLAG was immunoprecipitated (IP) and the rapid copurification of eEF1A was confirmed (see odd lanes). However, eEF1A abundance in the extracts was unaffected (Extract). (C) The filter examined in panel B above was incubated with antibodies against Ub, and high-molecular-weight Ub cross-reacting material was detected in association with the proteasome only in ATP-depleted extracts [upper panel; arrow, Ub(n)]. Equal amount of protein extracts were also resolved and incubated with antibodies against Ub (lower panel). (D) The filter shown in panel B above was incubated with antibodies against Met-βgal, and increasing levels of full-length and truncated fragments were detected (arrow). (E) A direct interaction between eEF1A and Met-βgal, following ATP depletion, was confirmed by immunoprecipitating Met-βgal and reacting an immunoblot with antibodies against eEF1A. As noted above, the rapid interaction occurred primarily in ATP-depleted samples.

We immunoprecipitated Pre1-FLAG and detected a broad distribution of high-mobility βgal cross-reacting products in association with the proteasome, within 5 min (Fig. 5D). The generation of the truncated products, following ATP depletion, is indicative of prematurely terminated translation products that were not detected in untreated cells (even-numbered lanes). Although the synthesis of full-length Met-βgal was reduced drastically 60 min after ATP depletion (Fig. 5A, lane 11), the levels that were detected in association with the proteasome remained high (Fig. 5D, lane 11). Because βgal is a tetramer, undamaged monomers that are associated with damaged βgal subunits may also be translocated to the proteasome (20). The accumulation of βgal fragments in the proteasome could represent a failure to degrade proteins following prolonged ATP depletion. (We used 10-fold more protein for the immunoprecipitation in Fig. 5D.)

eEF1A can bind a stable protein following ATP depletion.

It was previously shown that Met-βgal could be degraded cotranslationally (35) (Fig. 5D). We determined that translation damage caused rapid recruitment of eEF1A to the proteasome and concurrent accumulation of ubiquitinated proteins in the proteasome. Although Met-βgal is a stable protein, significant levels were detected in the proteasome following ATP depletion. Based on these results, we investigated if eEF1A could bind Met-βgal. Protein extracts that were examined in Fig. 5A to D were also incubated with anti-βgal antibodies to precipitate Met-βgal and nascent βgal fragments. Consistent with the association of eEF1A with ubiquitinated proteins in ATP-depleted extracts (Fig. 4B), we detected a rapid interaction between eEF1A and Met-βgal following similar treatment (Fig. 5E). In contrast, the association in untreated cells was very low. As expected, in the reciprocal experiment, when eEF1A was immunoprecipitated, Met-βgal was precipitated only from ATP-depleted extracts (data not shown). The coordinated association of eEF1A, Met-βgal, and ubiquitinated proteins with the proteasome is consistent with the hypothesis that eEF1A plays a role in cotranslational protein degradation.

Rpt1 influences proteasome interaction with eEF1A.

The direct interaction between eEF1A and Rpt1 (Fig. 3) provided a mechanism for bringing together central players in translation quality control and proteasome-mediated degradation. Because eEF1A also interacted with Met-βgal after ATP depletion (Fig. 5E), we examined the interaction of this reporter protein with the proteasome in rpt1, which is defective in binding eEF1A (Fig. 3A). Low levels of eEF1A were detected in untreated wild-type and rpt1 proteasomes (Fig. 6A, lanes 1 and 7). As expected, ATP depletion resulted in increased eEF1A-proteasome binding in wild-type proteasomes (lanes 1 to 6) but lower interaction in rpt1 (lanes 7 to 12). We examined the levels of Met-βgal and observed a time-dependent increase in its interaction with the proteasomes in wild-type cells (Fig. 6B, lanes 1 to 6). Note also the accumulation of truncated βgal cross-reacting species following ATP depletion. In contrast, higher levels of Met-βgal were already present in untreated rpt1 proteasomes (Fig. 6B, compare lanes 1 and 7), consistent with the reduced performance of this mutant proteasome. Met-βgal levels in rpt1 proteasomes increased dramatically following ATP depletion and achieved maximal levels within ∼10 min. These results indicate that rpt1 proteasomes can efficiently recognize and bind damaged translation products. However, the defect of the rpt1 proteasome is expected to reduce the turnover of bound substrates, with a corresponding interference with the delivery of additional damaged proteins. To test this idea, we examined the levels of Met-βgal in extracts following ATP depletion and found that it was rapidly eliminated in wild-type cells (Fig. 6C, lanes 1 to 6). The almost complete loss of full-length Met-βgal could be due to the association of normal polypeptides with damaged subunits in the βgal tetramer. A recent study showed that proximity to the proteasome was sufficient for targeting proteins for degradation (18). In contrast, dramatically higher levels of Met-βgal and degradation fragments were already present in extracts prepared from rpt1 (Fig. 6C, lane 7). High levels of Met-βgal fragments remained detectable in the extract for the duration of the investigation. Although Met-βgal is normally not a substrate of the Ub-proteasome pathway, it can be degraded cotranslationally. Ubiquitinated derivatives of Met-βgal were detectable after ATP depletion (data not shown). The apparent difference in stability of Met-βgal observed in Fig. 5A and 6B is the result of different genetic backgrounds and also because the study in Fig. 6C was conducted at elevated temperature (37°C) to investigate the defect of rpt1.

FIG. 6.

rpt1 is defective in degrading nascent proteins. (A) Wild-type and rpt1 strains were subjected to ATP depletion conditions. In Fig. 5, yeast cells were grown at 30°C, while in the experiments described here, yeast cells were grown at the nonpermissive temperature for rpt1 (37°C for 6 h). Pre1-FLAG was precipitated and the levels of eEF1A in the proteasome were threefold lower in rpt1 than in the wild-type strain. (B) The same immunoblot was incubated with antibodies against βgal, and the level of Met-βgal in the proteasome was determined. (C) The levels of Met-βgal in cell extracts were determined. Note the much faster elimination of Met-βgal in this study (37°C) than in Fig. 5A (30°C). Essentially the same βgal cross-reacting bands were detected in the extract and in the proteasome (compare bands in panels B and C). (D) The levels of Ub cross-reacting material were investigated in wild-type and rpt1 proteasomes. Note that the film exposure for the wild-type samples (lanes 1 to 6) was 10 times longer than for the rpt1 samples (lanes 7 to 12). (E) The interaction between eEF1A and the proteasome was examined in other proteasome mutants (rpt4 and rpt6) following ATP depletion. WT, wild type; IP, immunoprecipitation.

We investigated if the defect of rpt1 proteasomes was related to a failure to efficiently ubiquitinate proteins. Proteasomes were immunopurified after ATP depletion, and high-molecular-weight Ub cross-reacting material was detected in wild-type cells (Fig. 6D). These levels increased significantly in the wild-type proteasome during the course of ATP depletion, consistent with their accumulation in cell extracts (Fig. 5C). In contrast, surprisingly high levels of Ub cross-reacting material were already detected in the rpt1 proteasome (Fig. 6D), consistent with the proteolytic defect in this mutant proteasome. (Significantly, the film exposure for lanes 1 to 6 is 10-fold greater than for lanes 7 to 12.) These results suggested that the localization of ubiquitinated substrates to the rpt1 mutant proteasome was not impaired, although a posttargeting defect caused accumulation of ubiquitinated proteins. Surprisingly, the levels of Ub cross-reacting material that were associated with the rpt1 proteasome decreased progressively during the course of the analysis (Fig. 6D, lanes 11 to 12). One interpretation of these results is that cotranslationally damaged substrates are not efficiently retained by the rpt1 proteasome. Alternatively, proteasome-associated de-ubiquitinating enzymes may be more active on the rpt1 proteasome and promote dismantling and release of ubiquitinated substrates.

Rpt1 specifically affects the interaction between eEF1A and the proteasome, as revealed by a direct interaction (Fig. 3). To further support this result, we examined this interaction in yeast strains harboring mutations in other AAA-type ATPase subunits. We found that ATP depletion resulted in rapid eEF1A-proteasome association in rpt4 and rpt6 mutant cells (Fig. 6E).

An eEF1A mutant confers hyperresistance to canavanine.

The incorporation of the arginine analog canavanine during protein synthesis can cause misfolding and degradation by the Ub-proteasome system. To determine if the functional state of eEF1A contributed to the turnover of misfolded proteins, we characterized a collection of mutants (5, 8) for their ability to grow in medium containing canavanine. We determined that eEF1A^D156N and eEF1A^N153T/D156E conferred hyperresistance to canavanine (Fig. 7A; also data not shown). These alleles contain mutations in the GTP-binding motif that reduced affinity for GTP (5). It is noteworthy that eEF1A^D156N does not have an appreciable defect in translation rate, although it reduced translation fidelity. Therefore, the canavanine resistance of cells expressing eEF1A^D156N is not caused by reduced incorporation of canavanine into polypeptide chains (5). Additionally, eEF1A^D156N was not defective in its interaction with aminoacylated tRNA, and consequently Arg-tRNA that is charged with canavanine should be bound to the eEF1A mutants with similar efficiency.

FIG. 7.

(A) The growth of wild-type (WT) and tef1Δ tef2Δ expressing eEF1A^D156N was determined in the presence of 2-μg/ml l-canavanine. Ten-fold dilutions were spotted, and growth was examined after 4 days. (B) GST-UBA1 was immunoprecipitated following ATP depletion, and the copurification of eEF1A and eEF1A^D156N was determined. eEF1A^D156N formed an efficient interaction with ubiquitinated proteins that were bound to the UBA1 domain of Rad23. A longer exposure (lower panel) revealed constitutive eEF1A^D156N interaction with ubiquitinated proteins in untreated cells (lane 3). (C) The stability of a proteolytic substrate, Ub-Pro-βgal, was determined in cells expressing eEF1A^D156N. Chase times are indicated (in minutes), and the stability (t_1/2) was determined over the initial 30 min. Note that the initial incorporation of ³⁵S label was similar in both strains (time 0). (D) The abundance of Met-βgal was examined in wild-type and tef1Δ tef2Δ expressing eEF1A^D156N, in the presence and absence of 2-DG and 2,4-DNP. Total protein extracts were resolved by SDS-PAGE, transferred to nitrocellulose, and incubated with anti-βgal antibodies. High levels of Met-βgal were detected in the wild-type strain, while dramatically reduced levels were present in eEF1A^D156N-expressing cells. In contrast, the levels of highly stable Pab1 were unchanged. The severely reduced level of Met-βgal is unlikely to be the result of a modest twofold reduction in translation rate in eEF1A^D156N-expressing cells but is probably due to reduced fidelity and rapid cotranslational degradation. (E) We examined the stability of Ub-Pro-βgal, a well-characterized substrate of the Ub-proteasome, to determine if eEF1A suppressed the proteolytic defect of rad23Δ rpn10Δ. Yeast cells were labeled with [³⁵S]Met plus [³⁵S]Cys and chased in medium containing excess unlabeled amino acids and cycloheximide. Unlike the Rpn10 gene (which, as expected, was also isolated as a suppressor of rad23Δ rpn10Δ), we detected no significant recovery of degradation of Ub-Pro-βgal.

Although eEF1A^D156N dramatically increased resistance to canavanine, its interaction with the proteasome was similar to the wild-type protein (Fig. 2E). We therefore investigated if the association of eEF1A^D156N with ubiquitinated proteins was altered. GST-UBA1 was purified from ATP-depleted yeast cells and the copurification of eEF1A^D156N was compared to wild-type eEF1A. We found that eEF1A^D156N could be purified with ubiquitinated protein (GST-UBA1) from both untreated and ATP-depleted cells (Fig. 7B). This result demonstrates that the interaction between eEF1A^D156N and both the proteasome and ubiquitinated proteins is unimpaired. In fact, increased interaction of eEF1A^D156N with ubiquitinated proteins was observed (see lower panel) and may reflect the presence of elevated translational errors, or an intrinsically higher affinity for damaged proteins. To determine if eEF1A^D156N affected protein degradation, we measured the stability of Ub-Pro-βgal, a well-characterized substrate of the Ub/proteasome pathway (Fig. 7C). We determined that the in vivo t_1/2 of this reported protein was reduced from ∼27 to ∼13 min. Because the incorporation of ³⁵S into newly synthesized Ub-Pro-βgal was similar in both strains (see time 0 min), we conclude that protein synthesis was unaffected in the strain expressing eEF1A^D156N. Both stable and unstable proteins can be eliminated cotranslationally (35). Since Ub-Pro-βgal is degraded by the Ub-proteasome pathway, it was unclear if its rapid elimination was due to the stimulation of a cotranslational targeting step by eEF1A^D156N. We therefore examined the stability of Met-βgal, a stable test protein. eEF1A^D156N mutant yeast cultures expressing Met-βgal were subjected to ATP depletion and the levels of Met-βgal that remained in the extract were measured (Fig. 7D). We detected much lower levels of Met-βgal in untreated eEF1A^D156N cells. However, following ATP depletion the reporter protein was eliminated rapidly in both cells. The significantly reduced steady-state levels of Met-βgal in eEF1A^D156N-expressing cells is likely to be caused by increased cotranslational degradation and not reduced rates of protein synthesis, as noted in Fig. 7C. The rapid, constitutive turnover of proteins by eEF1A^D156N could contribute to the 10,000-fold increased resistance in medium containing canavanine (Fig. 7A).

As noted earlier, eEF1A was isolated as a dosage suppressor of the growth defects of rad23Δ rpn10Δ (Fig. 1). eEF1A suppressed the extreme sensitivity of the double mutant to translation inhibitors, as well as cold and high temperatures. Recent studies suggested that Rad23 and Rpn10 could function as alternate receptors in the proteasome (9, 37). Notably, the degradation of a test protein, Ub-Pro-βgal, is abolished in rad23Δ rpn10Δ. We therefore questioned if eEF1A could suppress the protein degradation defect of rad23Δ rpn10Δ. We compared the stability of Ub-Pro-βgal in rad23Δ rpn10Δ, in the presence or absence of high levels of eEF1A. We determined that Ub-Pro-βgal degradation was not restored (Fig. 7E). However, as expected, coexpression of Rpn10 in rad23Δ rpn10Δ restored degradation of Ub-Pro-βgal. We speculate that the primary defect of rad23Δ rpn10Δ is related to a failure to eliminate proteins that contain translation errors, rather than the posttranslational degradation of mature proteins.

DISCUSSION

A link between protein synthesis and protein degradation has been reported previously (29, 33, 35). Our results are consistent with the hypothesis that eEF1A can bind damaged nascent proteins that are ubiquitinated and facilitate their delivery to the proteasome through an interaction with Rpt1 (Fig. 8). The ability of eEF1A to bind ubiquitinated proteins and the proteasome resembles the activities of Rad23 and Rpn10. However, while Rad23 and Rpn10 bind multi-Ub chains, eEF1A interacts with damaged proteins that are ligated to multi-Ub chains. A conceptually similar mechanism is involved in the translocation of ubiquitinated misfolded proteins from the endoplasmic reticulum to the proteasome (19, 25).

FIG. 8.

Model. The proposed model is consistent with published studies and the results described here. Nascent proteins that are misfolded are rapidly ubiquitinated and degraded by the proteasome. We suggest that eEF1A, Rad23, Rpn10, and the proteasome can affect cellular tolerance to translation inhibitors by regulating the degradation of nascent proteins. Because eEF1A binds ubiquitinated proteins and the proteasome after treatment with translational inhibitors, we propose that it can participate in the recognition and degradation of damaged nascent proteins.

The activity of eEF1A in translation elongation may endow it with unique properties that permit detection of damaged proteins. For instance, eEF1A was reported to bind nascent polypeptide chains, unfolded proteins, and hydrophobic peptides (17, 23). eEF1A also promoted the degradation of Nα-acetylated proteins in vitro (12). We determined that eEF1A could be purified with GST-UBA1 in ATP-depleted extracts (Fig. 4B), although UBA1 bound ubiquitinated proteins in both ATP-depleted and untreated cells. This finding shows that eEF1A interacts specifically with nascent proteins that are ubiquitinated following translation damage. We speculate that UBA-containing proteins, such as Rad23, may bind ubiquitinated nascent misfolded proteins and translocate them to the proteasome.

eEF1A formed a strong interaction with the proteasome following ATP depletion and in the presence of translation inhibitors. This interaction was specifically reduced in rpt1 but not in rpt4, rpt6, rpn2, and rpn10 strains. eEF1A-proteasome interaction in rpt1 was reduced further at the nonpermissive temperature, although the proteasome was not destabilized. Because the ATPase subunits in the 19S particle have distinct effects on proteolysis (30), it is possible that Rpt1 performs a specific role in the degradation of nascent polypeptide chains.

A role for ubiquitination in protein synthesis quality control was shown by the discovery of extensive ubiquitination of the L28 ribosomal protein (34). L28 is conjugated to a lysine-63 Ub chain, and expression of a mutant Ub (Ub-K63R) caused sensitivity to translation inhibitors (34). Under specific conditions, cells expressing Ub-K63R contained unstable polysomes and high levels of the 80S ribosome complex, consistent with a principal defect in translation initiation. However, the nature of the 80S complex is not well defined. Although it is generally believed to contain 40S and 60S subunits, and mRNA, it has also been speculated that the 80S species lacks mRNA, or that it contains a dimer of 40S subunits. Since multiple steps in proteins synthesis rely on the availability of energy, it is likely that ATP depletion interferes with numerous processes including initiation, elongation, and recycling. L28 lies near the peptidyltransferase center, and the ligation of a Ub-K63 chain promoted stabilization of a translation elongation complex. Therefore, the failure to ubiquitinate L28 could lead to the accumulation of stalled 80S complexes and premature release of polypeptide chains from the ribosome.

The expression of a Ub-K63R mutant protein caused slight resistance to hygromycin B and cycloheximide (34), unlike the severe sensitivity observed in rad23Δ rpn10Δ. Furthermore, translation rates were reduced in cells that expressed Ub-K63R. In contrast, eEF1A mutants that conferred hyperresistance to translation inhibitors did not alter translation rates significantly and had normal polysome profiles. The differences in translation rate, fidelity, polysome stability, and sensitivity to translation inhibitors, in strains expressing Ub-K63R and eEF1A mutants, suggest that they act at different steps in the process of protein synthesis quality control. A simple interpretation of these results is that L28 ubiquitination provides a mechanism for minimizing translational errors, while eEF1A functions to maximize the elimination of translation errors once they have occurred. These are synergistic pathways, and Ub-K63 modification of L28 is not expected to directly impinge upon the activities of eEF1A.

The depletion of intracellular Ub can also cause sensitivity to translation inhibitors (13). For instance, a ubp6 mutant is unable to recycle Ub efficiently and is sensitive to anisomycin and cycloheximide. Both drugs block translation and caused increased turnover of Ub, with resulting depletion of Ub pools (13). In contrast, hygromycin and paromomycin do not block translation but can promote misincorporation, which results in protein misfolding. rad23Δ rpn10Δ and rpt1 mutants are extremely sensitive to hygromycin, paromomycin, and neomycin, demonstrating that the cellular response to translation fidelity errors requires the Ub/proteasome pathway. In contrast, ubp6 was moderately resistant to hygromycin B and neomycin, indicating that these drugs do not affect the levels of free Ub. Collectively, our results, as well as several recent studies, have revealed a convergence of multiple quality control mechanisms that provide insight into the mechanism of cotranslation protein degradation.