Endoplasmic reticulum stress differentially inhibits endoplasmic reticulum and inner nuclear membrane protein quality control degradation pathways

Endoplasmic reticulum (ER) stress occurs when the abundance of unfolded proteins in the ER exceeds the capacity of the folding machinery. Despite the expanding cadre of characterized cellular adaptations to ER stress, knowledge of the effects of ER stress on cellular physiology remains incomplete. We investigated the impact of ER stress on ER and inner nuclear membrane protein quality control mechanisms in Saccharomyces cerevisiae. We analyzed the turnover of substrates of four ubiquitin ligases (Doa10, Rkr1/Ltn1, Hrd1, and the Asi complex) and the metalloprotease Ste24 in induced models of ER stress. ER stress did not substantially impact Doa10 or Rkr1 substrates. However, Hrd1-mediated destruction of a protein that aberrantly engages the translocon (Deg1-Sec62) and substrates with luminal degradation signals was markedly impaired by ER stress; by contrast, Hrd1-dependent degradation of proteins with intramembrane degrons was largely unperturbed by ER stress. ER stress impaired the degradation of one of two Asi substrates analyzed and caused a translocon-clogging Ste24 substrate to accumulate in a form consistent with persistent translocon occupation. Degradation of Deg1-Sec62 in the absence of stress and stabilization during ER stress were independent of four ER stress–sensing pathways. Our results indicate ER stress differentially impacts degradation of protein quality control substrates, including those mediated by the same ubiquitin ligase. These observations suggest the existence of additional regulatory mechanisms dictating substrate selection during ER stress.

inner nuclear membrane (INM), which is physically continuous with the ER, where it promotes the destruction of nucleoplasmic and integral membrane proteins (14,15). Two additional E3 complexes, the transmembrane Asi complex (Asi1, Asi2, and Asi3) and the anaphase-promoting complex (APC), mediate protein quality control at the INM (16 -18).
Following the discovery of the UPR, additional ER stresssensing mechanisms have been identified that reduce the burden of aberrant proteins in the ER. The ER stress surveillance (ERSU) signaling pathway, mediated by the Slt2 mitogen-activated protein kinase, prevents inheritance of cortical ER containing aggregated proteins during ER stress (32,33). ER stress also promotes lysosomal destruction of ER-resident proteins by at least two mechanisms: the rapid ER stress-induced export (RESET) pathway, in which misfolded glycosylphosphatidylinositol (GPI)-anchored proteins are trafficked to the lysosome via the secretory pathway (34), and the activation of autophagy of ER subdomains (micro-ER-phagy) (35). In macro-ER-phagy, segments of ER are also targeted for lysosomal destruction (36,37); however, this mechanism has not been shown to be induced by ER stress. In the recently identified stressinduced homeostatically regulated protein degradation (SHRED) pathway, ER stress accelerates Ubr1-dependent degradation of misfolded cytosolic proteins and ER-localized proteins with misfolded cytosolic domains (38). In mammals, a subset of ER-targeted proteins exhibit translocational attenuation during ER stress, presumably as a preemptive measure to reduce the burden of unfolded proteins in the ER (39). This is likely mediated in part by the recently identified function of HRD1 in targeting secretory proteins prior to translocon insertion in a mechanism termed ER preemptive quality control (ERpQC) (40,41).
Despite the expanding catalogue of characterized ER stressresponse mechanisms, not all consequences of ER stress are known or understood. ER stress is a feature of several human diseases, including metabolic and neurodegenerative disorders, some forms of cancer, inflammation and immune syndromes, and mental illness (42)(43)(44)(45)(46). Prolonged ER stress leads to cell death, and at least one class of anticancer medications exerts its toxic effects on malignant cells by inducing ER stress (47,48).
In this work, we systematically analyzed the effect of ER stress on degradation of a panel of ER and INM quality control substrates. We found that ER stress did not substantially alter the degradation profiles of model substrates for Doa10 or Rkr1. However, despite well-documented induction of Hrd1 machinery by the UPR (2), degradation of model luminal and translocon-associated substrates of Hrd1 exhibited strong sensitivity to ER stress. By contrast, degradation of Hrd1 substrates with intramembrane degrons proceeded with normal kinetics in the face of ER stress, indicating Hrd1 is not broadly inhibited during stress. Furthermore, two Asi substrates demonstrated different sensitivities to ER stress. Divergent responses of ERAD and INMAD pathways mediated by the same ubiquitin ligase suggest novel layers of regulation of protein degradation by ER stress. Finally, a translocon-clogging substrate of Ste24 accumulated in a form consistent with prolonged translocon engagement. Impaired degradation of proteins that persistently engage the translocon is expected to curtail ER import of other proteins, thereby mitigating the impact of ER stress.

ER stress impairs degradation of a Hrd1 ERAD-T substrate
N-terminal fusion of the Deg1 degron from MAT␣2 to Sec62 converts the protein to a Hrd1 ERAD-T substrate (25). Following co-translational insertion of two transmembrane segments of Sec62, a portion of the N-terminal tail aberrantly translocates into the translocon (Fig. 1A). A disulfide bond forms between a portion of the Deg1-Sec62 N-terminal tail and the interior of the translocon, contributing to persistent channel engagement (25,49). After aberrant translocon engagement, Deg1-Sec62 becomes progressively modified by glycosylation, causing the protein to migrate as multiple species by SDS-PAGE ( Fig. 1B) (25). Deg1-Sec62 is also acetylated on its N terminus and on two internal lysine residues (50). Neither acetylation nor glycosylation are required for Deg1-Sec62 degradation (25,50). However, glycosylation serves as a visual indicator of aberrant translocon engagement. This aberrant translocon engagement triggers Hrd1-dependent degradation of Deg1-Sec62, although the mechanism by which Hrd1 recognizes the protein remains unclear (25).
To determine the impact of ER stress on ERAD-T, we analyzed the degradation of Deg1-Sec62 by pulse-chase in the presence and absence of the ER-specific stressor tunicamycin, which blocks N-linked glycosylation. In untreated cells, Deg1-Sec62 exhibited characteristic modification and degradation over time (Fig. 1C). Strikingly, treatment of yeast with 1 g/ml tunicamycin for 30 min prior to radioactive labeling of nascent protein profoundly impaired Deg1-Sec62 degradation. Observable post-translational modification (PTM) of Deg1-Sec62 was also strongly impaired. Increasing tunicamycin concentration to 10 g/ml completely abrogated observable Deg1-Sec62 PTM. The effects of tunicamycin were rapid. Degradation and PTM were strongly impaired when tunicamycin incubation was reduced to 10 min prior to radiolabeling, but not when tunicamycin was present only during pulse labeling (Fig. 1D). Similarly, treatment with the reducing agent dithiothreitol (DTT) at a concentration of 6 mM strongly impaired Deg1- Figure 1. ER stress impairs degradation of a Hrd1 ERAD-T substrate. A, schematic depiction of Deg1-Sec62 prior to (uninserted) and following (inserted) aberrant translocon engagement. Deg1-Sec62 consists of Deg1 (the N-terminal 67 amino acids from the yeast transcriptional repressor MAT␣2), a FLAG (F) epitope, the two-transmembrane protein Sec62, and two copies of the S. aureus protein A (PrA). Following translocon engagement, Deg1-Sec62 undergoes extensive PTM, including N-linked glycosylation, and is polyubiquitylated by Hrd1. The primary glycosylated asparagine residue is indicated as a blue circle. Ub, ubiquitin. B, virtual SDS-PAGE illustrates differential migration of Deg1-Sec62 based on glycosylation status. C, pulse-chase analysis of WT yeast expressing Deg1-Sec62 cultured in the presence of DMSO or tunicamycin at the indicated concentrations for 30 min. DMSO and tunicamycin were maintained at the same concentrations throughout pulse labeling. D, pulse-chase analysis of WT yeast expressing Deg1-Sec62 cultured in the presence of 10 g/ml tunicamycin for the indicated times (or DMSO for 30 min). Tunicamycin and DMSO were maintained at the same concentrations throughout pulse labeling. E, pulse-chase analysis of WT or hrd1⌬ yeast expressing Deg1-Sec62 cultured in the presence of 6 mM DTT (or no treatment) for 30 min. DTT was maintained at the same concentrations throughout pulse labeling. F, doa10⌬ hrd1⌬ yeast cells expressing Deg1-Sec62 were cultured in the presence 6 mM DTT, 10 g/ml tunicamycin, or DMSO control for 30 min. DTT, tunicamycin, and DMSO were maintained at the indicated concentrations throughout pulse labeling. Immunoprecipitated Deg1-Sec62 was incubated in the presence or absence of Endo H and calf intestinal phosphatase as indicated. G and H, cycloheximide chase analysis of WT yeast expressing Deg1-Sec62 cultured in the presence of 10 g/ml tunicamycin or DMSO (G) or 6 mM DTT or no treatment (H) for 1 h. Tunicamycin, DTT, or DMSO were maintained at the same concentration during incubation with cycloheximide. Deg1-Sec62 was detected with AlexaFluor-680 -conjugated rabbit anti-mouse antibodies. Pgk1 served as a loading control. Where indicated, the percentage of Deg1-Sec62 remaining at each time point is presented below the image. For cycloheximide chase experiments, Deg1-Sec62 signal intensity was normalized to Pgk1. The experiment depicted in C was performed three times with tunicamycin at 0 and 10 g/ml, and one time with tunicamycin at 1 g/ml. Experiments depicted in D and F were performed one time. Experiments depicted in E, G, and H were performed three times.

ER stress and ER/INM protein degradation
Sec62 turnover (Fig. 1E); DTT causes ER stress by countering the normally oxidizing environment of the ER lumen. DTT also reduced PTM of Deg1-Sec62, but its impact on PTM was less pronounced than that of tunicamycin.
To characterize the PTM impaired by ER stressors, we incubated pulse-labeled Deg1-Sec62 with endoglycosidase H (Endo H) or calf intestinal phosphatase. The electrophoretic mobility of Deg1-Sec62 from unstressed cells exhibited strong sensitivity to Endo H, consistent with N-linked glycosylation (Fig. 1F). By contrast, Deg1-Sec62 exhibited little if any change in mobility after treatment with phosphatase, which is consistent with failure to detect Deg1-Sec62 phosphorylation in a recent biochemical analysis (50). Loss of Endo H-sensitive Deg1-Sec62 species following tunicamycin treatment confirmed that the compound completely inhibited Deg1-Sec62 N-glycosylation. DTT reduced the extent of or delayed glycosylation.
The pulse-chase experiments described above evaluated the degradation and modification of nascent Deg1-Sec62. Treatment for 1 h with either tunicamycin or DTT also strongly stabilized and impaired PTM of the steady-state Deg1-Sec62 population in cycloheximide-chase experiments (Fig. 1, G and H). Taken together, our results indicate that two different forms of ER stress strongly impair degradation and PTM of nascent and steady-state pools of a Hrd1 substrate that aberrantly engages the ER translocon.

ER stress differentially impacts degradation of Hrd1 substrates
Previous work indicated that degradation of the soluble Hrd1 ERAD-L substrate CPY* (a misfolded variant of carboxypeptidase Y possessing the G255R mutation) is inhibited by ER stress (2). Thus, one possible explanation for impairment of Deg1-Sec62 degradation is that Hrd1 catalytic activity is impaired by such stress. We analyzed the impact of ER stress on the degradation of a panel of Hrd1 substrates (depicted in Fig. 2A). Consistent with earlier reports, HA-tagged CPY* was strongly stabilized by both DTT and tunicamycin (Fig. 2B). Similar to Deg1-Sec62, both treatments impeded PTM of CPY*-HA, which is N-glycosylated. DTT and tunicamycin also stabilized the transmembrane ERAD-L substrate Erg3-13myc (51) to a similar extent as HRD1 deletion (Fig. 2C).
We next evaluated the effect of ER stress on degradation of ERAD-M substrates 6myc-Hmg2 (3) and Pdr5*-HA (52). In contrast to ERAD-T and -L substrates, degradation of 6myc-Hmg2 (Fig. 2D) and Pdr5*-HA (Fig. 2E) was largely insensitive to DTT and tunicamycin; degradation of Deg1-Sec62 expressed in the same cells was markedly impaired during ER stress. These substrates were substantially stabilized by loss of Ubc7, the primary ubiquitin-conjugating enzyme that functions with Hrd1 (4). Finally, we evaluated the impact of ER stress on turnover of a self-ubiquitylating substrate (SUS-GFP) in which GFP and the Hrd1 RING domain are fused to a Myc-tagged transmembrane portion of Hmg1 (53,54). Neither form of ER stress impaired turnover of SUS-GFP (Fig. S1). Thus, stabilization of model ERAD-T and ERAD-L substrates by ER stress is not due to broad impairment of Hrd1 ubiquitin ligase activity.

ER stress does not impair Doa10-dependent degradation
We evaluated the impact of ER stress on ERAD substrates targeted by Doa10. The nascent population of the transmembrane Doa10 ERAD-C substrate Deg1-Vma12 (Fig. 3A) (21) exhibited rapid degradation in the presence of ER stress in pulse-chase experiments (Fig. 3B). By contrast, this model Doa10 substrate was strongly stabilized by point mutations in the Deg1 degron (Deg1*, F18S and I22T (55)) that prevent Doa10-dependent degradation (21). Similarly, ER stress only mildly perturbed degradation of the steady-state population of Deg1-Vma12. This contrasted sharply with the strong stabilization of Deg1-Sec62 in the same cells (Fig. 3C).
Deg1-sec62 † possesses a mutation in Sec62 (G127D of the fusion protein, equivalent to G37D of untagged Sec62) that substantially reduces aberrant translocation of its N-terminal tail (25,56,57). This variant is degraded predominantly by the Doa10 ERAD-C pathway. A minor subpopulation of this protein still undergoes aberrant translocation and retains Hrd1dependent degradation (25). Deg1-sec62 † was highly unstable, regardless of the presence of tunicamycin (Fig. 3D). Interestingly, a small fraction of Deg1-sec62 † was stabilized by tunicamycin; this likely reflects the fraction of Deg1-sec62 † that has become an ERAD-T (Hrd1) substrate by virtue of having undergone aberrant translocation.

ER stress does not impact Rkr1 substrate abundance
Vma12-K12-13myc is a model Rkr1 ERAD-RA substrate ( Fig. 4A) (8). In this construct, N-terminally FLAG-tagged Vma12 is followed, in sequence, by two copies of protein A, 12 lysine residues (K12), and a 13myc epitope. The positively charged K12 sequence triggers ribosome stalling, likely via ionic attraction to the negatively charged ribosome exit tunnel (12,58,59). A virtual SDS-PAGE of lysates from WT and rkr1⌬ cells expressing Vma12-K12-13myc is depicted in Fig. 4B. When Vma12-K12-13myc expressed in WT yeast cells is analyzed by SDS-PAGE and immunoblotting, two major populations are observed: a translationally stalled product (Vma12-K12) and the full-length translational read-through product (Vma12-K12-13myc). RKR1 deletion increases abundance of the translationallystalled protein relative to the full-length read-through product (8). Loss of RKR1 also results in the appearance of multiple species migrating more slowly than Vma12-K12. We speculate these species represent Vma12-K12 molecules that have been modified by the C-terminal addition of alanine and threonine (CAT tails) as observed for soluble translationally-stalled substrates of Rkr1 (60,61).
Incubation with ER stressors did not substantially alter Vma12-K12 mobility or abundance (Fig. 4C). We also have not observed an increase in the abundance of translationally stalled species of model soluble ER-targeted Rkr1 substrates in cells exposed to ER stress. 7 Our results suggest that ER stress does not inhibit Rkr1-dependent destruction of translationally stalled ERAD-RA substrates.

ER stress differentially impacts degradation of Asi substrates
We investigated the effect of ER stress on turnover of Erg11-FLAG and Vtc1-3HA, which are degraded via Asi-mediated INMAD upon mislocalization to the INM (Fig. 5A) (16,17). Whereas ER stress perturbed Erg11-FLAG steady-state abundance, its turnover rate was largely unaffected relative to the impact of ASI1 deletion (Fig. 5B). By contrast, ER stress stabilized Vtc1-3HA to a similar degree as loss of ASI1 (Fig. 5C). Thus, Asi substrates are modestly, but differentially, sensitive to ER stress.

ER stress impairs degradation of a translocon-clogging substrate of Ste24
We analyzed the effects of ER stress on a second protein engineered to clog the ER translocon. This protein, dubbed "Clogger," consists of the soluble ER luminal Pdi1 protein followed, in sequence, by a rapidly folding variant of DHFR, three engineered glycosylation acceptor sequences, and an HA epitope ( Fig. 6A) (31). The Pdi1 signal sequence promotes posttranslational translocation. However, the rapidly folding DHFR causes a substantial fraction of the protein to clog the translo- in the presence of 6 mM DTT, 10 g/ml tunicamycin, or DMSO for 1 h. DTT, tunicamycin, and DMSO were maintained at the same concentration during incubation with cycloheximide. Asterisks in B and E denote nonspecific bands. Cells analyzed in D and E also expressed Deg1-Sec62 as a control for ER stress induction. CPY*-HA and Pdr5*-HA were detected with anti-HA antibodies. Erg3-13myc and 6myc-Hmg2 were detected with anti-Myc antibodies. Deg1-Sec62 was detected with AlexaFluor-680 -conjugated rabbit anti-mouse antibodies (D) or peroxidase anti-peroxidase antibodies (E). Pgk1 served as a loading control. Where indicated, the percentage of substrate remaining (normalized to Pgk1) at each time point is presented below the image. We note that, relative to other experiments, Deg1-Sec62 exhibited weaker stabilization in the absence of UBC7 in the cycloheximide chase presented in D; this may be related to differences in genetic background in the strains analyzed. Experiments depicted in B-D were performed three times. The experiment depicted in E was performed two times.  A, schematic of Vma12-K12-13myc, which consists of a FLAG epitope tag (F), the two-transmembrane protein Vma12, two copies of the S. aureus protein A (PrA), 12 lysine (K12) residues (depicted as sequential "ϩ" symbols), and a 13myc epitope tag. After insertion of the two transmembrane segments of Vma12, K12 triggers translational stalling (middle). This is resolved by predicted C-terminal addition of alanine and threonine (AT) residues (CAT tailing), Rkr1-mediated ubiquitylation, and degradation (left) or release from stalling and translation of 13myc (right). Ub, ubiquitin. B, virtual SDS-PAGE illustrates differential migration of translationally stalled Vma12-K12, CAT-tailed Vma12-K12, and read-through product Vma12-K12-13myc in WT and rkr1⌬ yeast lysates. C, yeast of the indicated genotypes harboring plasmids encoding Vma12-K12-13myc and Deg1-Sec62 (as a control for ER stress induction) or empty vectors were cultured in the presence of 6 mM DTT, 10 g/ml tunicamycin, or DMSO (Ϫ) for 1 h prior to lysis and separation by SDS-PAGE. Vma12-K12-13myc was detected using anti-Myc antibodies, which bind to both 13myc and protein A. Deg1-Sec62 was detected with AlexaFluor-680 -conjugated rabbit anti-mouse antibodies. G6PDH served as a loading control. Expression of Vma12-K12(-13myc) in rkr1⌬ cells may increase Deg1-Sec62 levels, suggesting stabilized ERAD-RA substrates may cross-inhibit ERAD-T. The experiment depicted was performed four times.

ER stress and ER/INM protein degradation
con. Because sequences upstream and downstream of the clogging moiety possess N-glycosylation acceptor sites, Clogger's ER insertion status can be assessed by comparing the relative abundance of differently migrating species (Fig. 6B). The fastest migrating species are nonglycosylated, cytosolic (uninserted) molecules. The slowest migrating species are fully glycosylated, completely translocated molecules. Species exhibiting intermediate migration are partially glycosylated, translocationally stalled molecules (31). The metalloprotease Ste24 cleaves the clogged form of this protein, thereby relieving translocon obstruction. Loss of Ste24 causes accumulation of faster migrating (i.e. clogged and cytosolic) forms of Clogger.
In the presence of DTT, faster migrating (presumably clogged and preinserted) Clogger species accumulated (Fig.  6C). Tunicamycin profoundly stabilized and impaired modification of Clogger. Thus, ER stress impairs degradation of Clogger and causes it to accumulate in a form consistent with persistent translocon engagement.

Deg1-Sec62 induces the UPR
ER stress impairs degradation of Deg1-Sec62 (Fig. 1) and Clogger (Fig. 6), two proteins that aberrantly engage the translocon. High-level expression of Clogger induces the UPR (31). To determine whether Deg1-Sec62 also induces the UPR, we transformed WT and ubc7⌬ yeast with an empty vector or a plasmid encoding Deg1-Sec62 driven by the strong glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter. These yeast expressed GFP under the control of the unfolded protein response element (UPRE). Fluorescence intensity, measured by flow cytometry, is therefore a readout of ER UPR induction. Deg1-Sec62 expression caused an ϳ2-fold induction of the UPR in ubc7⌬ cells, consistent with exacerbation of ER stress by persistent translocon engagement (Fig. 7).

ERAD-T and its inhibition during ER stress are independent of characterized ER stress-responsive pathways
ERAD-T impairment during ER stress was surprising, given that the UPR increases expression of ubiquitin ligases (including Hrd1) and chaperone proteins involved in ER quality control (2). We tested the hypothesis that the UPR selectively inhibits ERAD-T during stress. In this case, restoration of ERAD-T during stress in cells lacking the UPR transducer Ire1 would be predicted. However, impairment of Deg1*-Sec62 degradation by ER stress was not mitigated by loss of Ire1. Furthermore, ERAD-T proceeded largely unimpeded in ire1⌬ cells (Fig. 8A). Failure to splice HAC1 mRNA confirmed UPR deficiency in ire1⌬ cells (Fig. 8B).
We investigated whether mediators of other ER stress-responsive pathways are required for Deg1*-Sec62 degradation or stabilization by ER stress. Deletion of the gene encoding the Slt2 kinase, which prevents transmission of stressed ER to daughter cells via the ERSU mechanism (32), did not stabilize Deg1*-Sec62 under nonstress conditions or prevent its stabilization by tunicamycin (Fig. 8A). In mammalian cells, where the RESET pathway was first characterized, the p24 family member Tmp21 mediates ER export of GPI-anchored proteins during ER stress (34). Loss of the Tmp21 homologue Emp24, which results in a p24-null phenotype in yeast (62), did not impair Deg1*-Sec62 degradation or ER stress-dependent stabilization (Fig. 8C). Finally, mutation of the SHRED mediator Ubr1, which ubiquitylates misfolded proteins during cellular stress (38), did not detectably alter Deg1*-Sec62 degradation kinetics in the presence or absence of ER stress (Fig. 8D).
We note that the variant of Deg1-Sec62 (Deg1*-Sec62) employed in experiments presented in Fig. 8 (and in Fig. 11, A and C) possesses point mutations in Deg1 (F18S and I22T). These alterations do not affect aberrant translocon engagement or Hrd1-dependent degradation (25). This construct has been used interchangeably with Deg1-Sec62 to investigate Hrd1-dependent ERAD-T (25, 50). B, cycloheximide chase analysis of yeast of the indicated genotypes expressing Erg11-FLAG or Deg1-Sec62 cultured in the presence of 6 mM DTT, 10 g/ml tunicamycin, or DMSO for 1 h. C, cycloheximide chase analysis of yeast of the indicated genotypes expressing Vtc1-3HA and Deg1-Sec62 cultured in the absence or presence of 6 mM DTT for 1 h. DTT, tunicamycin, and DMSO were maintained at the same concentration during incubation with cycloheximide. Erg11-FLAG was detected with anti-FLAG antibodies. Vtc1-3HA was detected with anti-HA antibodies. Deg1-Sec62 was detected with peroxidase anti-peroxidase antibodies. Pgk1 served as a loading control. The experiment depicted in B was performed two times. The experiment depicted in C was performed three times (with the exception of tunicamycin treatment, which was performed one time).

ER stress and ER/INM protein degradation Kar2 overexpression does not rescue Deg1-Sec62 degradation during ER stress
We tested whether overexpression of the multifunctional ER chaperone Kar2 suppresses inhibition of Deg1-Sec62 degradation during ER stress. We mildly overexpressed Kar2 by transforming WT yeast expressing Deg1-Sec62 with a centromeric plasmid encoding Kar2 (Fig. 9A). Western blot analysis confirmed a mild plasmid-dependent increase in Kar2 abundance. This mildly increased KAR2 gene dosage did not accelerate Deg1-Sec62 degradation in the presence of DTT.
To more dramatically increase KAR2 gene dosage, we transformed yeast expressing Deg1-Sec62 with a 2 plasmid expressing KAR2 from a yeast genomic library clone (63). Western blot analysis confirmed overexpression of Kar2 (Fig. 9B). In cells harboring the KAR2 plasmid, we observed the appearance of a second, more slowly migrating species of Kar2, which likely corresponds to cytosolic Kar2 precursor that has retained its signal peptide prior to translocation (64). The accumulation of immature Kar2 in cells expressing Kar2 from a 2 plasmid suggests an effective upper limit for ER-specific Kar2 overexpression. This moderately increased KAR2 expression also did not accelerate Deg1-Sec62 degradation in the presence of DTT.

ER stress does not alter membrane association of Deg1-Sec62
ER stress limits Deg1-Sec62 PTM. Because these modifications largely occur in the ER lumen, we considered the possibility that ER stress alters translocation and membrane association of Deg1-Sec62, thereby impairing its ability to be recognized or degraded by Hrd1. We subjected ER-derived microsomes prepared from nonstressed and DTT-stressed cells to a variety of conditions to analyze the membrane association of Deg1-Sec62 (Fig. 10). Deg1-Sec62 was solubilized by detergent (Triton X-100). Consistent with membrane association, Deg1-Sec62 was not substantially solubilized following incubation with sodium chloride (which solubilizes cytosolic peripheral proteins) or sodium carbonate (which solubilizes both luminal and cytosolic peripheral proteins). Incubation of cells with DTT did not markedly alter the membrane association properties of Deg1-Sec62. Therefore, impairment of Deg1-Sec62 degradation by ER stress is not due to a stress-dependent change in Deg1-Sec62 membrane association.

ERAD-T is not broadly stress-sensitive
ER stress can also be induced by membrane aberrancy caused by inositol depletion (65). We cultured exponentialphase cells expressing Deg1*-Sec62 in the presence or absence of inositol for 5 h. Inositol depletion induced the

ER stress and ER/INM protein degradation
UPR, as evidenced by increased UPRE-driven GFP expression (Fig. 11A). However, the rate of Deg1*-Sec62 degradation was not impacted by inositol limitation.
DTT and tunicamycin induce ER stress by specifically promoting misfolding of ER-localized proteins. We asked whether heat shock, which causes protein misfolding throughout the cell, impairs Deg1-Sec62 destruction. We performed cycloheximide chase experiments to analyze Deg1-Sec62 degradation in cells cultured at 30°C and shifted to 42°C for 1 h and for the duration of the chase. Exposure to elevated temperatures did not stabilize Deg1-Sec62 in WT cells (Fig. 11B). Therefore, conditions associated with global protein misfolding are not sufficient to impair Hrd1-dependent destruction of an aberrant translocon-associated protein.
Finally, we determined whether oxidative stress affects Deg1*-Sec62 turnover. The oxidant hydrogen peroxide induces the UPR in cultured myoblasts (66). Incubation of yeast in the presence of 0.4 mM hydrogen peroxide for 1 h did not affect Deg1*-Sec62 degradation (Fig. 11C). Induction of oxidative stress-responsive GFP-tagged Rtc3 (67) in a parallel culture confirmed hydrogen peroxide activity (Fig. 11D).

Discussion
Our results reveal that ER stress differentially impacts ER and INM protein quality control proteolytic pathways. Model Doa10 ERAD-C and Rkr1 ERAD-RA substrates were largely unaffected by ER stress. Similarly, Hrd1-dependent degradation of proteins with intramembrane degrons proceeded with similar kinetics regardless of ER stress induction. However, destruction of three Hrd1 substrates (a soluble ERAD-L substrate, a transmembrane ERAD-L substrate, and an ERAD-T substrate) was specifically impaired by ER stress. Modification and turnover of a translocon-clogging substrate of Ste24 was also perturbed by ER stress. Finally, degradation of one of two tested Asi INMAD substrates was sensitive to ER stress.
During ER stress, translocation of a subset of proteins into the ER is slowed (39). Reduced ER chaperone availability correlates with and likely contributes to translocational attenuation, which is thought to be protective, reducing the load for the already burdened proteostasis machinery. Dampened translocon quality control during ER stress may be an additional adaptive mechanism in which undegraded channel-engaged proteins temporarily impede translocation of other polypeptides into the stressed ER.
Yeast and mammalian homologues of Hrd1 have been reported to physically interact with the translocon (41,68). Interaction of mammalian HRD1 with the translocon was reported to increase ϳ1.5-fold during ER stress (41), an observation that would not be predicted by our results. Increased association of HRD1 with the translocon correlates with HRD1-dependent turnover of ER-targeted proteins prior to their translocation via ERpQC. By preemptively promoting degradation of secretory proteins prior to translocation, ERpQC may protect the stressed ER. We therefore speculate that regulation of Hrd1 by ER stress reduces protein load in the ER by two mechanisms: 1) stimulation of preemptive Hrd1mediated targeting of secretory proteins prior to ER insertion, and 2) inhibition of Hrd1-dependent targeting of proteins that are already translocon-engaged.
Sensitivity of Deg1-Sec62 degradation to ER stress is consistent with an earlier observation that high-level expression of polyQ-expanded huntingtin protein both induces ER stress and impairs Deg1-Sec62 degradation (69). ERAD-T impairment is specific to particular subtypes of ER stress, as several different forms of stress expected to disrupt global proteostasis or ER homeostasis (elevated temperature, inositol limitation, and oxidative stress) did not impair Deg1-Sec62 degradation. Thus, misfolded proteins per se are likely not the direct signal impairing Hrd1-mediated destruction of translocon-associated proteins.
One potential trivial explanation for impairment of Deg1-Sec62 degradation is that ER stress prevents aberrant translocon engagement by Deg1-Sec62, which converts the protein into a Hrd1 substrate in the first place (25). We do not believe this is the cause for Deg1-Sec62 stabilization by ER stress for two reasons. First, Deg1-Sec62 becomes N-glycosylated in the presence of DTT (albeit in a delayed fashion), strongly suggesting that aberrant translocation of the fusion protein does occur. Second, mutations that prevent aberrant translocon engagement of Deg1-Sec62 cause a reversion of dependence of Deg1-Sec62 degradation from Hrd1 to Doa10 (25). Therefore, if ER stress blocks Hrd1-dependent degradation of Deg1-Sec62 by preventing translocon engagement, we would expect Deg1-Sec62 to become a Doa10 substrate and still be rapidly degraded; this is not observed. Consistently, we did not observe a difference in association of Deg1-Sec62 with ER membrane fractions in the presence or absence of ER stress.
Another possible explanation for impaired Deg1-Sec62 degradation is that ER stress prevents one or more PTMs

ER stress and ER/INM protein degradation
required for recognition by Hrd1. However, neither glycosylation nor acetylation are required for Deg1-Sec62 degradation (25,50). Although it remains possible that uncharacterized PTMs contribute to Deg1-Sec62 degradation, no available evidence suggests this is the case. The mechanism by which DTT delays modification of Deg1-Sec62 remains unclear. Redox perturbation may alter the structure or intermolecular interactions of enzymes required for glycosylation or

ER stress and ER/INM protein degradation
substrate molecules in a manner that precludes efficient modification.
Unperturbed degradation of a subset of Hrd1 substrates argues against complete Hrd1 inhibition during ER stress. We speculate that one or more proteins sense ER stress and specifically inhibit Hrd1-dependent degradation of translocon-associated proteins. Our experiments argue against roles for the UPR, ERSU, RESET, and SHRED stress-sensing pathways in impeding the destruction of translocon-associated proteins during stress, unless they function redundantly.
It is possible that an as yet unidentified ERAD-T co-factor functions in a novel ER stress-sensing mechanism and becomes limiting for translocon-associated protein degradation. Stressdependent changes in expression and complex association of Hrd1 cofactors have been observed (70). Another possibility is that one or more protein(s) sensitive to redox and glycosylation state mediate the effect of ER stress on degradation. Efforts to identify and characterize factors that regulate Hrd1-dependent degradation of translocon-engaged proteins are underway.
ER stress may similarly impair targeting of the transloconassociated clientele of Ste24. Accumulation of higher-mobility species suggests that the translocon-engaged form of Clogger is enriched in the presence of DTT. DTT-dependent changes in Clogger mobility were not observed in a previous investigation (31). This difference may be related to differences in treatment protocol between the two studies. In our experiments, cultures were incubated in the presence of 6 mM DTT for 1 h; in the experiments described in Ref. 31, yeast were incubated in the presence of 5 mM DTT for 4 h. It is possible that short-and long-term responses to ER stress differ.
Both Clogger and Deg1-Sec62 overexpression mildly induce the UPR (Fig. 7) (31). On its face, this runs counter to our suggestion that persistent translocon engagement protects against ER stress by reducing inward flux of nascent proteins. However, these two ideas can be reconciled with the following model. At basal levels of unfolded proteins in the ER, translocon clogging may increase stress by preventing inward movement of proteostasis machinery. By contrast, at higher levels of stress, nonspecifically stemming ER import may minimize the luminal burden of unfolded proteins.
Why does ER stress impair ERAD-L and not ERAD-M? Following Hrd1-mediated ubiquitylation, both ERAD-L and ERAD-M substrates must undergo protein extraction from the ER into the cytosol (retrotranslocation) prior to proteasomemediated degradation (71). ERAD-L and ERAD-M substrates have different requirements for retrograde transport, and ER stress may only impair ERAD-L retrotranslocation (54, 71).
Finally, ER stress impacts Asi complex substrates in subtly different ways. Induction of ER stress alters the steady-state abundance of Erg11-FLAG without dramatically changing its rate of degradation. By contrast, ER stress impedes turnover of Vtc1-3HA to a similar degree as deletion of ASI1. These results imply that Asi substrates may be degraded and regulated via distinct modalities reminiscent of the multitude of Hrd1-mediated degradation mechanisms.
Unperturbed degradation of several proteins in the presence of tunicamycin and DTT appears to contradict a previous report of global proteasome inhibition by ER stress (72). However, although the earlier study reported acute stabilization of a subset of unstable proteins, degradation of a metabolically-labeled cytosolic protein was observed to proceed with similar kinetics in the absence and presence of stress. Mild accumulation of this protein during stress was observed over a longer time course following transient proteasome inhibition. These results argue for minor, indirect effects of ER stress on global protein degradation. In a more recent proteomics investigation, ER stress accelerated the proteasomal degradation of a subset of physiological proteins, while stabilizing others (73), consistent with divergent effects of ER stress on different protein populations.

Yeast and plasmid methods
Yeast were cultured at 30°C in standard growth medium as described previously (49). Plasmids were introduced to yeast by the lithium acetate transformation procedure (74). Yeast strains and plasmids used in this study are presented in Tables  1 and 2, respectively. Yeast strains were generated by standard gene replacement methods and yeast mating, sporulation, and haploid selection (74,75). Plasmids pVJ343, pVJ411, and pVJ463 were constructed by subcloning BamHI/HindIII fragments containing Deg1-Vma12, Deg1*-Sec62, or Deg1-Sec62, respectively, from plasmids constructed in a previous study (25) into expression vectors with the desired promoter and auxotrophic marker gene (76,77).
For galactose induction of Clogger, which is driven by the GAL1/10 promoter, yeast were cultured overnight in selective medium containing 2% raffinose as the carbon source. Overnight cultures were diluted in fresh medium containing 4% galactose and cultured until cells reached mid-exponential growth.
For inositol limitation experiments, cells were cultured until they reached mid-exponential growth in medium containing inositol (prepared using yeast nitrogen base without amino acids). Cells were washed six times in medium lacking inositol (prepared using yeast nitrogen base without amino acids and inositol) and incubated in medium lacking inositol for 5 h. ER-derived microsomes were prepared from hrd1⌬ yeast expressing Deg1-Sec62 that had been cultured in the presence or absence of 6 mM DTT for 1 h. Microsomal fractions were incubated in the presence of water, 1% Triton X-100, 0.5 M sodium carbonate, or 0.5 M sodium chloride before being separated into pellet (P) and supernatant (S) fractions and solubilized. Deg1-Sec62 was detected with peroxidase anti-peroxidase antibodies. Cue1 and Cdc48 were detected by anti-Cue1 and anti-Cdc48 antibodies, respectively. The experiment depicted was performed two times (with the exception of sodium chloride treatment, which was performed one time).

ER stress and ER/INM protein degradation Pulse-chase analysis
Pulse-chase analysis was performed as described previously (25,78). Briefly, yeast cells were labeled with 20 Ci of Tran 35 S-label (MP Biomedicals) per 1 OD 600 unit of cells at 30°C for 10 min in medium lacking methionine and cysteine. Chases were performed in the presence of excess unlabeled methionine and cysteine. Deg1 fusion proteins were immunoprecipitated with anti-FLAG M2 affinity resin (Sigma) (Fig. 1, D-F) or sequential incubation with anti-Deg1 antibody (79) and recombinant protein A-cross-linked agarose A (Repligen; Figs. 1C and 3, B and D). Immunoprecipitated proteins were separated by SDS-PAGE. Gels were analyzed by autoradiography, using a Storm 860 Phosphorimager system and ImageQuant 5.2 software (Molecular Dynamics). Figures for which molecular weight markers are not available (Figs. 1, C, E and F, and 3D) portray experiments performed on variants of proteins analyzed multiple times elsewhere in this study in experiments that include molecular weight markers.

Endoglycosidase H and calf intestinal phosphatase treatment
Cells were radiolabeled as described for pulse-chase experiments, and FLAG-tagged Deg1-Sec62 was immunoprecipitated with anti-FLAG M2 affinity resin. After five washes of resin with wash buffer (150 mM NaCl, 50 mM Hepes, pH 7.5, 5 mM EDTA, 1% Triton X-100, and 0.1% SDS), two additional equilibrating washes were performed with NEBuffer 4 (New England Biolabs). The resin was resuspended in 50 l of NEBuffer 4 that had been supplemented with potassium acetate, pH 5.6, to a final concentration of 80 mM. 0.005 units of Endo H (Roche Applied Science), 10 units of calf intestinal phosphatase (New England Biolabs), both, or neither were added to the resin suspension. Samples were incubated at 37°C for 3 h with gentle mixing approximately every 10 min. 1ϫ Laemmli sample buffer was added, and samples were heated to 100°C for 5 min.

ER stress and ER/INM protein degradation
were harvested, suspended in 200 l of 0.1 M NaOH, and incubated for 5 min at room temperature. Cells were pelleted, resuspended in 1ϫ Laemmli sample buffer, and heated to 100°C for 5 min. Lysates were subject to centrifugation to clear the preparations of insoluble material. The soluble fraction was separated by SDS-PAGE.
For experiments presented in Figs. 2E, and 5, B and C, and Fig. S1, yeast were lysed as described previously (82). Briefly, 2.5 OD 600 units of yeast were harvested. NaOH was added to a final concentration of 0.26 M, and ␤-mercaptoethanol was added to a final concentration of 0.13 M. Cells were incubated on ice for 15 min. To precipitate proteins, TCA was added to a final concentration of 5%. Proteins were collected by centrifugation at 4°C.
Protein pellets were resuspended in 50 l of TCA sample buffer (3.5% SDS, 0.5 M DTT, 80 mM Tris, 8 mM EDTA, 15% glycerol, and 0.1 mg/ml bromphenol blue) and incubated at 37°C for 30 min. Insoluble material was pelleted by centrifugation prior to electrophoretic separation.

Cycloheximide-chase analysis
Cycloheximide-chase experiments were performed as described previously (83). Briefly, yeast were grown to mid-exponential phase at 30°C, unless otherwise specified. Cells were concentrated to 2.5 OD 600 units/ml in fresh media. Cycloheximide was added to a final concentration of 250 g/ml. Aliquots of 2.4 OD 600 units of cells (950 l) were harvested at the indi-

Microsome preparation and protein-membrane association analyses
Yeast microsomal membranes were prepared essentially as described previously (10,25,49). 10 OD 600 units of cells were harvested, suspended in 1 ml of resuspension buffer (10 mM Tris, pH 9.4, and 10 mM DTT), and incubated at room temperature for 10 min. Cells were harvested by centrifugation, washed with spheroplast buffer (1 M sorbitol, 20 mM sodium phosphate, pH 7.5, 150 mM NaCl, and 2 mM DTT), and digested with 140 g of zymolyase 100T (MP Biomedicals)/10 OD 600 units of cells in spheroplast buffer for 20 min at 30°C. Spheroplasts were harvested (5 min at 600 ϫ g at 4°C) and washed in spheroplast buffer containing 20 g/ml pepstatin A, 1 mM EDTA, and 1ϫ EDTA-free Complete Protease Inhibitor Mixture (Roche Applied Science). Spheroplasts were centrifuged again (5 min at 600 ϫ g at 4°C), resuspended in fractionation buffer (200 mM D-mannitol, 20 mM sodium phosphate, pH 7.5, and 150 mM NaCl) with protease inhibitors, and lysed by vortexing in the presence of glass beads for three 30-s pulses (1 min on ice between pulses). Unbroken cells and cellular debris were pelleted by centrifugation (5 min at 600 ϫ g at 4°C), and the supernatant was used as the microsomal preparation. Microsomes were incubated with sodium carbonate (final concentration 0.2 M, pH 11), Triton X-100 (final concentration 1% v/v), sodium chloride (final concentration 0.5 M), or no additive. Samples were maintained on ice for 15 min with occasional vortexing and then centrifuged (15 min at 13,000 ϫ g at 4°C). Supernatants were transferred to fresh tubes and preserved as the soluble fraction. To ensure samples had equal detergent concentrations, Triton X-100 was added to all supernatants that had not been initially solubilized with Triton X-100 (final concentration 1%); an equal volume of water was added to samples that had been previously solubilized by Triton X-100. Pellets were washed with fractionation buffer containing protease inhibitors and resuspended in fractionation buffer with 1% Triton X-100. 1ϫ Laemmli sample buffer was added to all samples, prior to heating at 100°C for 8 min and separation by SDS-PAGE.

Western blotting
Proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride membrane via wet transfer at 20 V for 1 h, 70 V for 2.5 h, or 30 V for 8 h at 4°C. Membranes were blocked in a solution containing 5% skim milk in Tris-buffered saline (TBS: 50 mM Tris-base, 150 mM NaCl) at room temperature for 1 h or at 4°C overnight. Membranes were probed in a solution containing 1% skim milk in TBS with 1% Tween 20 (TBS/T) and the appropriate antibody. The membranes were incubated in the presence of antibodies for 1 h at room temperature followed by three 5-min washes in TBS/T.

Flow cytometry
Flow cytometry was performed using cells expressing the indicated GFP-tagged proteins. Cells were cultured until they reached mid-exponential growth and were subjected to experimental treatments, as indicated. Mean GFP fluorescence of 10,000 cells in synthetic-defined medium was measured using the MACSquant Analyzer X.

Analysis of HAC1 mRNA splicing
HAC1 mRNA splicing was assessed using previously described methods with modifications (85)(86)(87). RNA was extracted from 1 OD 600 unit of cells cultured to mid-exponential growth using RNeasy Mini (Qiagen) and DNA-free (Ambion) kits. cDNA was synthesized from 1 g of total RNA using the iScript cDNA synthesis kit (Bio-Rad). The resulting cDNA was used as a template for PCRs as described (85). PCR products were analyzed on a 1.5% agarose gel and confirmed by DNA sequencing.