Main

Folding and maturation of proteins in eukaryotic cells that are destined for distal compartments of the secretory pathway occur in the endoplasmic reticulum (ER). Proteins that fail to acquire their proper structure are retained within the ER and eventually degraded to prevent toxification by the accumulation of misfolded proteins in the secretory pathway. This turnover is mediated by a mechanism known as ERAD, which may be described as a three-step process1,2,3,4. The first is the recognition of aberrant ER proteins and probably involves the action of ER-resident chaperones and folding enzymes as well as the glycosylation machinery5,6,7. In the second step, degradation substrates are dislocated from the ER into the cytosol. This retrograde transport involves the action of the Sec61 translocation channel, which also mediates the import of proteins into the ER6,8,9. Further factors have been identified in screens for yeast mutants deficient in ERAD10,11. These proteins have been proposed to participate in the dislocation process1,3,4, although their precise function is unknown. One of them, Hrd1/Der3, is of particular interest because it contains a RING-H2 finger domain, which is also found in a specific class of ubiquitin ligases12,13. In the final step, substrates are polyubiquitinated by the E2 enzymes Ubc6 and Cue1-associated Ubc7, and are proteolytically disintegrated by 26S proteasomes14,15,16,17. Yeast genetic data indicate that a non-functional ERAD mechanism does not affect growth. This has led to the assumption that further mechanisms may increase cellular tolerance to misfolded proteins when ERAD is abrogated. A known response to the accumulation of aberrant proteins within the ER is the upregulation of several ER chaperones by the UPR18. Increased formation of misfolded proteins activates the kinase/nuclease Ire1, which is located at the ER and/or the inner nuclear membrane19,20. Activated Ire1 subsequently triggers formation of the transcription factor Hac1, which in turn induces the expression of the UPR target genes21,22. UPR is thought to assist in refolding of denatured proteins in the ER by increasing the concentrations of chaperones in this compartment18.

Both ERAD and the UPR are found in many eukaryotic cells in a mechanistically similar form, although neither is essential for cell viability. Here we show that mutations affecting both pathways cause synthetic phenotypes in yeast. Furthermore, deletion of ERAD-mediating components results in significant UPR induction. We also show that genes encoding enzymes of the ubiquitin system and other ERAD components are under the control of the UPR. These observations indicate that UPR activation may not be restricted to ER-resident chaperones but may also account for the enhanced expression of other gene products involved in ER function. Our data also highlight the necessity of coordinated regulation of ERAD and the UPR.

Results

Ubc1 and Ubc7 contribute to CPY* turnover in a synergistic manner.

CPY* is a mutant lysosomal protein that is retarded in the ER lumen and rapidly degraded with a half-life of about 24 min ( ref. 11). It is stabilized in cells lacking Ubc7, as demonstrated in pulse–chase experiments15. However, quantitative analysis has shown that this stabilization is not complete (Fig. 1). To characterize the further components that contribute to CPY* proteolysis, we tested yeast mutants lacking certain ubiquitin-conjugating enzymes and found that CPY* was stabilized in cells lacking Ubc1 (Fig. 1). Ubc1 has previously been implicated in degradation of misfolded proteins in the cytosol and has been shown to be important for the outgrowth of spores23. Mutants lacking Ubc1 (Δubc1) degraded CPY* at a slightly increased rate (t1/2 = 45 min) compared to Δubc7 cells (t1/2 = 80 min), although stabilization was significantly stronger than in cells lacking Ubc6 (data not shown). As Δ ubc7 and Δubc6 mutants are epistatic with respect to CPY* degradation15, we were interested in the genetic relationship between UBC1 and UBC7. In double mutants lacking both enzymes, little degradation of CPY* was detected within 90 min of chase (t1/2 = 305 min). We therefore conclude that degradation of ER-lumenal proteins requires Ubc1 in addition to the previously described Ubc6/Ubc7 pathway.

Figure 1: Degradation of CPY* is dependent on Ubc1 and Ubc7.
figure 1

Pulse–chase analysis of CPY* degradation. Wild-type, Δubc1, Δubc7 and Δubc1Δubc7 cells, all expressing CPY*, were lysed at the indicated times and immunoprecipitated with specific antibodies against CPY. Immunoprecipitions were separated by SDS–PAGE and analysed using a PosphorImager. a, Typical SDS–PAGE. b, Quantification of five experiments. Error bars representing s.d. for the individual strains do not overlap (data not shown).

Full size image

We next addressed the question of whether Ubc1 and Ubc7 function in pathways that diverge after the retrograde transport of proteolytic substrates, or whether both E2 enzymes are components of completely independent ERAD systems. To distinguish between these possibilities, we combined a null mutation in HRD1 with Δubc1 or Δubc7 alleles. Quantitative analysis showed that the degree of stabilization of CPY* in Δhrd1 cells was slightly lower than in the Δubc1Δubc7 double mutant (t1/2 = 180 min compared with 305 min). Δ hrd1Δubc1 or Δhrd1Δubc7 double mutants exhibited no further stabilizing effects in comparison to Δ hrd1 cells (Fig. 2). Similar results were obtained using a deletion allele of DER1 (data not shown). Thus, Δ ubc1 and Δubc7 deletions are epistatic with the Δ hrd1 deletion, whereas the Δubc1Δubc7 double mutation has synergistic effects. The most likely explanation for this observation is that Ubc1 and Ubc7 participate in distinct pathways that are both dependent on Hrd1.

Figure 2: Hrd1 is epistatic to Ubc1 and Ubc7 with respect to CPY* degradation.
figure 2

Pulse–chase analysis of CPY* degradation. The experiment was carried out as described for Fig. 1; wild-type, Δhrd1 , Δhrd1Δubc1 and Δhrd1Δ ubc7 strains, all expressing CPY*, were used. a, Typical SDS–PAGE. b, Quantification of five experiments. Error bars representing s.d. for the individual strains do not overlap (data not shown).

Full size image

The UPR is required for turnover of increased concentrations of ERAD substrates.

Δubc1Δubc7 mutants do not exhibit impaired growth despite the fact that ERAD is severely affected in these cells. One explanation for this is that misfolded proteins may rarely be found under normal growth conditions. Alternatively, other systems may exist that increase cellular tolerance to misfolded proteins in the absence of ERAD. One of such complementing mechanism may be the UPR. Loss of DER1 activity does not cause an increase in the UPR11, whereas deletion of other components mediating ERAD results in weak induction9,19,22. Nevertheless, we monitored the UPR using a β-galactosidase reporter under control of a hybrid promotor containing a UPR-response element in ERAD-defective cells (Fig. 3). In Δubc1 cells, the β-galactosidase concentration was comparable to that of wild-type cells. Intriguingly, cells lacking Ubc7 displayed a fivefold increase, and Δubc1Δubc7 mutants a tenfold increase, in β-galactosidase activity (Fig. 3a). Induction of the UPR in the Δ hrd1 mutant was fivefold, which is similar to that caused by the UBC7 deletion. These findings indicate that unfolded proteins may accumulate to a significant degree under normal growth conditions when ERAD is abrogated. Expression of endogenous levels of CPY* from the chromosomal locus prc1–1 did not significantly affect the UPR (Fig. 3a). However, UPR induction was more pronounced (up to 20-fold compared with the wild type) when CPY* was overexpressed, using a copper-inducible construct under the control of the CUP1 promotor, in cells lacking functional ERAD (Fig. 3b). This activation was caused by the accumulation of unfolded proteins in the ER and does not represent a previously proposed direct function of ERAD components on the half-life of Hac1 ( ref. 19), as it was not observed when ERAD mutations were combined with a deletion of the IRE1 kinase (data not shown). Moreover, a recent study22 has shown that enhanced UPR in Δubc7 cells is not caused by a change in the half-life of the transcription factor Hac1. Neither the addition of copper to the medium nor the use of CPY*-overexpressing plasmid without copper induction had any detectable effect on the UPR (data not shown). In all strains investigated, except those containing mutations in HAC1 or IRE1, the UPR could be fully induced by treatment with 3 mM dithiothreitol, which led to similar levels of β-galactosidase activity (60-fold relative to non-induced wild type; data not shown).

Figure 3: Disruption of ERAD causes induction of the UPR.
figure 3

a, Yeast cells of the indicated strains were transformed with a UPR–reporter construct and β-galactosidase activity was measured (see Methods). prc1–1 denotes expression of CPY*. Values are means ± s.d. from eight experiments and are expressed as fold activity relative to that in wild-type cells. b, The yeast strains in a were further transformed with plasmids pTX191 or pTX192; CuSO4 was added to the medium to induce overexpression of CPY* (CPY*oe) and β-galactosidase activity was measured. Wild type denotes an untransformed control. Values are means ± s.d. from six experiments and are expressed as fold activity relative to that in the untransformed control.

Full size image

We next investigated whether UPR induction influences the ability of ERAD to degrade misfolded proteins. We expressed constitutively active Hac1 ( HAC1i; ref. 19) in wild-type cells and determined the half-life of CPY*. No change in breakdown was observed (Fig. 4a; t1/2 = 22 min in wild-type cells compared to 19 min in cells transformed with HAC1i). We then measured CPY* turnover in the absence of a functional UPR. Degradation of endogenous levels of CPY* was not altered in cells lacking Ire1 (Fig. 4b; t1/2 = 24 min in wild-type cells compared to 29 min in Δire1 cells). However, overexpression of CPY* resulted in a prolonged half-life (t1/2 = 40 min). Efficient degradation of increased amounts of CPY* was strictly dependent on Ire1, as indicated by a markedly decreased CPY* turnover in Δ ire1 cells (t1/2 = 180 min). The breakdown of overexpressed CPY* paralleled that of the endogenous protein, as it was dependent on Ubc1, Ubc7, Hrd1 and proteasomal functions (data not shown).

Figure 4: Degradation of misfolded proteins under normal growth conditions does not require the UPR.
figure 4

a, The half-life of CPY* in wild-type and Δire1 cells, both expressing CPY*, was determined in pulse–chase experiments. Where indicated, CPY* was overexpressed from a CUP1 promotor (CPY*oe). b, CPY* turnover was monitored by pulse–chase analysis in CPY*-expressing cells. Where indicated, constitutively active Hac1 was expressed (HAC1i).

Full size image

Two principal conclusions can be drawn from these results. First, ERAD alone is sufficient to degrade endogenous unfolded proteins under normal growth conditions. However, it can be easily overloaded by overexpression of a single misfolded protein, as indicated by the induction of the UPR. Second, increased concentrations of misfolded proteins, which induce the UPR, are degraded by the ERAD system only when the UPR is functional.

UPR regulates expression of ERAD components.

On the basis of the observations described above, we investigated whether the accumulation of unfolded proteins in the lumen of the ER would also induce the expression of ERAD components. We treated cells with various concentrations of dithiothreitol for 3 h. Cell extracts were prepared, analysed by immunoblotting and quantified. As a control we used glyceraldehyde-3-phosphate dehydrogenase (GAPDH), which was unaffected by dithiothreitol treatment. We found that the steady-state concentrations of Ubc7 and of its membrane receptor Cue1 were increased. Calculation of the ratios of their concentrations to that of GAPDH revealed that these enhancements were 6-fold and 2.5-fold, respectively. Induction was mediated by the UPR, as cells lacking HAC1 or IRE1 (data not shown) showed virtually no change in protein levels after dithiothreitol treatment. Intriguingly, the amounts of other proteins involved in ERAD, such as Sec61, were also elevated in dithiothreitol-treated wild-type cells (data not shown), whereas the level of Ubc1 was unaffected (Fig. 5a).

Figure 5: Upregulation of ERAD components in response to dithiothreitol is dependent on an active UPR.
figure 5

a, Wild-type and Δhac1 cells expressing Myc-tagged Ubc7 (Ubc7–Myc) were treated with the indicated dithiothreitol concentrations for 3 h. Whole-cell extracts were prepared and subjected to immunoblotting (lower panel) with specific antibodies against the Myc epitope, Ubc1, Cue1 and GAPDH. Results of 2–5 experiments were quantified and normalized to GAPDH values (upper panel). b, Wild-type and Δire1 cells were treated with 4 mM dithiothreitol for 30 min; samples were subjected to northern blotting (see Methods). Values are means from three experiments and are expressed as the ratio of mRNA levels in induced cells to those in uninduced cells. u, uninduced cells; i, induced cells. Both Δire1 and Δhac1 cells were viable throughout the experiments, as verified by plating assays (data not shown).

Full size image

To demonstrate that this regulation occurs at the level of transcription, we quantified the mRNA concentrations of ERAD components in wild-type, Δ ire1 (Fig. 5b) and Δhac1 (data not shown) cells, under normal growth conditions or shortly after induction of the UPR by treatment with dithiothreitol. The levels of actin and Ubc1 mRNAs were unaffected by dithiothreitol treatment. Expression of UBC7 and CUE1 was increased, which is consistent with the results of our immunoblot analysis (Fig. 5a). In addition, we found that levels of mRNA for Hrd1, another ERAD component, were increased. The mRNA induction was completely dependent on Hac1 and Ire1. Together, these results indicate that the UPR may upregulate components of the ERAD system. However, the UPR is not required for standard expression of these proteins, which further implies the existence of a basal level of the ERAD system that is independent of UPR function.

Mutations in ERAD and the UPR cause synthetic phenotypes.

The perceived relationship between the UPR and ERAD indicates that both systems may complement each other in preventing damage to the secretory pathway. We tested this by constructing mutants that lack ERAD as well as the UPR. As was found for the respective single mutants, growth of Δhac1Δ ubc1, Δhac1Δubc7 and Δire1Δ hrd1 double mutants was not affected at 30 ˚C. Furthermore, the double mutant Δubc1Δubc7 exhibited no growth deficiency. In contrast, the triple mutant Δubc1Δubc7Δ hac1 exhibited a significantly reduced growth rate at normal temperature. When this mutant was shifted to the elevated temperature of 37 °C, it was unable to form colonies within 3 days of incubation. Furthermore, we observed severe growth impairment in Δhac1Δubc7 and Δ ire1Δhrd1 double mutants at 37 °C (Fig. 6).

Figure 6: Mutations affecting ERAD and the UPR cause synthetic lethal phenotypes.
figure 6

Exponentially growing cells of the indicated strains were spotted onto synthetic minimal plates in serial tenfold dilutions and incubated for 3 days at 30 °C or 37 °C.

Full size image

Although the above results are convincing, it must be remembered that elevated temperature is not an ER-specific form of stress. We therefore analysed the sensitivity of these strains to the glycosylation inhibitor tunicamycin. Exponentially growing cells were treated with 2 µg ml–1 tunicamycin for various time periods. They were then plated on rich medium without tunicamycin, to determine rates of survival (Fig. 7). Wild-type, Δ ubc7 and Δhrd1 cells were not affected by tunicamycin even when treated for 2 h (data not shown). Although Δire1 cells did not survive prolonged tunicamycin treatment, >50% of cells were still able to form colonies after 15 min of incubation. In contrast, Δ ire1Δubc7 and Δire1Δhrd1 double mutants were much more sensitive to tunicamycin — <5% of cells were viable after 15 min of incubation. Overall, these data demonstrate synthetic effects between ERAD and the UPR under both ER stress and normal growth conditions.

Figure 7: Δire1Δubc7 and Δire1Δ hrd1 double mutants are more sensitive to accumulation of unfolded proteins in the ER.
figure 7

Yeast cells of the indicated strains, growing exponentially in liquid minimal medium, were treated with 2 µg ml–1 tunicamycin for 15 min. Before and after treatment, equal amounts of cells were plated onto YPD plates.

Full size image

Discussion

The data presented here indicate that both Ubc1 and Ubc7 may contribute to polyubiquitination of ERAD substrates in a synergystic manner. The activities of Ubc1 and Ubc7 seem to represent the principal ubiquitin-conjugating steps in this process. Unlike single-gene disruptions, deletion of both enzymes results in complete stabilization of an ER degradation substrate. Surprisingly, cells with severely impaired ERAD exhibit no growth-impairment phenotype10,11,24,25 and no increased sensitivity to drugs that enhance the formation of ERAD substrates (data not shown). We therefore propose the existence of a ‘back-up’ system for ERAD that assists cells in dealing with misfolded ER proteins. A previously known mechanism, the UPR, is induced in ERAD-impaired cells, demonstrating the existence of aberrant proteins under normal conditions. Moreover, overexpression of a mutant ER protein enhances UPR induction not only in ERAD mutants but also in wild-type cells. We also observed that the turnover of low concentrations of an aberrant ER protein occurs independently of the UPR. In contrast, degradation of increased amounts of this protein (as a result of overexpression) is dependent on UPR function. In conclusion, these results indicate that the occurrence of protein misfolding in wild-type cells may be dealt with by ERAD alone. However, a slight increase in the abundance of unfolded proteins seems to overload the ERAD system as expressed under normal growth conditions.

We also observed upregulation of ERAD components in response to treatment with substances that cause accumulation of misfolded proteins in the ER. Obviously, the UPR not only confers cellular tolerance to increased amounts of aberrant ER proteins by promoting their folding18, but also contributes to the enhanced turnover of such substrates. Therefore, coordinated action of ERAD and the UPR is required to cope with unfolded ER proteins. Upregulation of Ubc7 in cells treated with cadmium ions has been reported25. As exposure to heavy metal ions has been shown to cause induction of the UPR26, it is feasible to speculate that these earlier observations represent a regulatory effect mediated by the UPR.

Consistent with our results, we observed a genetic link between ERAD and the UPR — a combination of mutations affecting both pathways caused increased sensitivity to tunicamycin compared with a single IRE1 mutation. This indicates that ERAD may also be involved in elimination of unfolded proteins as generated by tunicamycin, although mutations affecting only the ERAD system do not confer sensitivity to the drug. Moreover, mutations in the UPR and ERAD severely affect growth, emphasizing that these systems are compensatory and are both required for proper ER function under normal conditions.

The relationship between ERAD and the UPR parallels that between the cytosolic heat-shock response and the ubiquitin–proteasome pathway. Levels of cytosolic chaperones and specific components of the ubiquitin system are increased upon heat shock. Conversely, disruption of UBC4/UBC5 or inhibition of the proteasome causes a heat-shock response27,28. The same is true in the ER — cellular levels of ERAD components are raised when the UPR is activated and a defect in ERAD results in increased expression of ER folding enzymes by the UPR. In both cases, crosstalk between the degradative machinery and the chaperone system is required to prevent damage by unfolded proteins. Under normal growth conditions, basal expression of the degradative machinery is sufficient to eliminate misfolded proteins. However, once cells are subjected to stress caused by increased formation of aberrant proteins, further degradative capacities must be provided.

Methods

Plasmid and strain construction.

Yeast rich and minimal media were prepared as described29; standard genetic methods29 were used. Yeast strains used in this study are listed in Table 1. The XL1 Blue Escherichia coli strain was used; standard techniques29 were used for recombinant DNA work. Myc-tagged Ubc7 was expressed from an ARS/CEN plasmid as described17. For construction of the HRD1 deletion, the coding sequence was amplified by the polymerase chain reaction (PCR) and sequences from SpeI (–108) to HincII (+1,418) were replaced with the TRP1 marker gene (pJU165). To generate the Δhac1 yeast strain, the gene was amplified by PCR and the region from SpeI (+25) to EcoRI (+930) was replaced with the TRP1 marker gene. To obtain a deletion construct for IRE1 disrupted with the LEU2 marker gene (pJU341), an ApaI– SacI fragment derived from pCS135 (a gift from P. Walter) was cloned into pRS405. The URA3 disruption of IRE1 was effected using a construct provided by P. Walter. All deletion constructs were introduced into yeast cells as linearized vectors; correct integration was confirmed by PCR. A CPY*-overexpressing plasmid was generated by PCR amplification of the CPY* open reading frame and 3′ non-coding sequences from the start codon to a HindIII restriction site (–2,092), using pRS306prc1–1 ( ref. 11) as a template plasmid. The resulting fragment was fused to the CUP1 promotor, which was cloned by a PCR strategy from yeast genomic DNA using primers 5′-GGATCCCATTACCGACATTTG and 5′-GAATTCTTTATGTGATGATTGATTGATTG, respectively, and subsequently transferred into the yeast ARS/CEN vectors pRS313 and pRS314 (ref. 29), thereby generating plasmids pTX191 and pTX192, respectively. Constitutively active HAC1i was obtained after removal of the intron sequence19 by a PCR strategy, followed by cloning into plasmid pRS315 ( ref. 29) and sequencing.

Table 1 Yeast strains used in this study
Full size table

Antibodies.

Specific antibodies against the Myc epitope and against CPY were from Santa Cruz and Molecular Probes, respectively. Specific antibodies against Ubc1 and against GAPDH were supplied by S. Jentsch and G. Daum, respectively. Specific antibodies against Cue1 were as described17.

Pulse–chase experiments and immunoprecipitations.

Pulse–chase analysis was carried out as described14, except that overnight cultures were grown in synthetic minimal medium supplemented with 0.1% yeast extract. To overexpress CPY* from plasmids pTX191 and pTX192, respectively, cells were grown in synthetic minimal medium to early log phase and 0.1 mM CuSO4 was then added for 4 h. Addition of CuSO4 did not affect the turnover of endogenous CPY* (data not shown). Pulse labelling with [35S]Met/Cys (Amersham) and chase experiments were carried out at 30 °C. Cells were disrupted in cold lysis buffer (50 mM Tris–HCl pH 7.5 and 1% SDS) using glass beads. Lysates were diluted tenfold with cold IP dilution buffer (1.1% Triton, 165 mM NaCl, 50 mM Tris–HCl pH 7.5, 5.5 mM EDTA and 1 mM phenylmethyl sulphonyl fluoride) and subsequently cleared by two centrifugation steps at 16,000g. Immunoprecipitation and quantitative analysis were carried out as described14.

Preparation of cell extracts.

Exponentially growing yeast cells corresponding to 10 A600were treated with 0, 1, 2 or 4 mM dithiothreitol for 3 h. Cells were disrupted with glass beads as described above. Samples were heated to 65 °C in sample buffer and subjected to SDS–PAGE. Immunoblotting was carried out as described14, developed using enhanced chemiluminescence (Amersham) and quantified using a LAS1000CH Luminescent Image Analyser (Fuji).

Northern blotting.

Yeast cells were grown in synthetic minimal medium to exponential phase and, where indicated ( Fig. 5b), were treated with 4 mM dithiothreitol for 30 min. Cells corresponding to 5 A600 were disrupted with glass beads in 200 µl disruption buffer (50 mM HEPES pH 7.5, 100 mM LiCl, 1 mM lithium dodecyl sulphate (LiLDS), 10 mM EDTA and 5 mM dithiothreitol) and diluted with 1 volume of disruption buffer containing 900 mM LiCl. Samples were extracted twice with phenol/chloroform at 65 °C for 5 min and once with chloroform alone. mRNA was purified using Oligo(dT)25 magnetic beads (Dynal) according to the manufacturer’s instructions. Samples were run on a 1% agarose formaldehyde gel and blotted with Hybond N (Amersham). Probes were amplified by PCR and radiolabelled with [32P]α-dATP using a Random primed DNA-labelling kit (Boehringer). Hybridization and washing procedures for the blots were as described29; quantitative analysis was carried out as described for pulse–chase experiments.

UPR induction.

UPR induction was measured using a reporter construct harbouring a CYC1 promoter containing the UPR element fused to the lacZ gene30. In short, cells harbouring this plasmid were grown in synthetic minimal medium to mid-log phase and, where indicated (Fig. 3b), treated for 4 h with 0.1 mM CuSO4 to induce overexpression of CPY* from plasmid pTX191 or pTX192, respectively. Full induction of the UPR was achieved by adding 3 mM dithiothreitol for 4 h to the medium. β-galactosidase activity was then determined as described29.