|Home | About | Journals | Submit | Contact Us | Français|
The endoplasmic reticulum (ER) must contend with a large protein flux, which is especially notable in cells dedicated to secreting hormone-regulated gene products. Because of the complexity of the protein folding pathway and the potential for genetic or stochastic errors, a significant percentage of these nascent secreted proteins fail to acquire their native conformations. If these species cannot be cleared from the ER, they may aggregate, which leads to cell death. To lessen the effects of potentially toxic polypeptides, aberrant ER proteins are destroyed via a process known as ER-associated degradation (ERAD). ERAD substrates are selected by molecular chaperones and chaperone-like proteins, and prior to degradation most substrates are ubiquitin-modified. Together with the unfolded protein response, the ERAD pathway is a critical component of the protein quality control machinery in the ER. Although emerging data continue to link ERAD with human diseases, most of our knowledge of this pathway arose from studies using a model eukaryote, the yeast Saccharomyces cerevisiae. In this review, we will summarize the discoveries that led to our current understanding of this pathway, focusing primarily on experiments in yeast. We will also indicate links between ERAD and disease and emphasize future research avenues.
Protein biogenesis is an error-prone process mainly because of the complexity of the protein folding pathway. In addition, polypeptides may be encoded by mutated genes, which result in unstable proteins or kinetically trapped folding intermediates. Furthermore, cellular stresses or nutrient deprivation may lead to the incorporation of incorrect amino acids or compromise the acquisition of post-translational modifications. In each of these scenarios, toxic, aggregation-prone species can form, which may lead to cell death. Thus, cells cannot risk the threat that accompanies aberrant protein accumulation. To offset this threat, quality control mechanisms evolved, of which central mediators are molecular chaperones .
The synthesis of molecular chaperones is induced during cell stress to prevent polypeptide aggregation and to attempt to refold non-native proteins. If the native conformation cannot be achieved, molecular chaperones can deliver substrates to proteolytic enzymes. The prevalence of quality control-associated proteolysis is highlighted by the fact that perhaps 30% of newly synthesized polypeptides are destroyed in some cells . Moreover, defects in the proteolysis of cytoplasmic aggregates lead to neurodegeneration .
Quality control pathways also exist in eukaryotic organelles, such as the endoplasmic reticulum (ER). Under stress, ER quality control becomes essential . Because the ER is the first organelle encountered by secreted proteins, this compartment is stocked with enzymes that catalyse folding and post-translational modifications, such as protein disulphide isomerase, the signal peptidase and the oligosaccharyl transferase. The ER is further laden with molecular chaperones and chaperone-like lectins that prevent the generation of off-pathway folding intermediates [5,6].
If unmitigated, ER stress triggers a specific response that leads to a cascade of events that help protect the ER from stress. This cascade of events is known as the unfolded protein response (UPR). In mammals, the UPR is initiated by three ER transmembrane sensors, known as Ire1, PERK and ATF6, at least two of which may directly recognize unfolded proteins. The sensors are also regulated by an ER chaperone, known as BiP (also see below) [7,8]. On activation, downstream effectors of Ire1, PERK and ATF6 increase the synthesis of molecular chaperones, enzymes required for protein modification, components required for delivery to later compartments of the secretory pathway and lipid biosynthetic enzymes that increase the volume and surface area of the ER. Thus, the ER becomes enriched for factors that favour protein folding; protein synthesis is also slowed. In the event that the UPR cannot be resolved, an apoptotic programme is initiated. An increasing body of data indicates that ER stress may be linked to several human diseases, including diabetes .
Another manner in which the concentration of unfolded proteins in the ER can be lessened is via degradation. Several lines of evidence ultimately indicated that the cytoplasmic, 26S proteasome can degrade ER membrane and luminal proteins. The 26S proteasome is a multicatalytic protease that preferentially destroys polypeptides to which ubiquitin has been appended. The 26S proteasome is composed of two particles: a 20S `core' that harbours three pairs of distinct proteases and a 19S `cap' that shovels polypeptide substrates into the core in an ATP-dependent reaction. The 19S cap also regulates substrate entry into the core . The conjugation of ubiquitin onto proteasome substrates requires the consecutive action of an E1 ubiquitin-activating enzyme, an E2 ubiquitin-conjugating enzyme and an E3 ubiquitin ligase . Efficient proteasome targeting requires the conjugation of four ubiquitins – linked to one another – and the recognition by a series of ubiquitin receptors on the proteasome .
Evidence for proteasome-mediated destruction of ER proteins arose from studies in both yeast and mammalian cells. First, disruption of UBC6, which encodes a cytoplasmic E2 ubiquitin-conjugating enzyme, rescued defects associated with a thermolabile subunit of the translocation channel, Sec61 . Second, an immature form of the cystic fibrosis transmembrane conductance regulator (CFTR) was degraded by the proteasome [14,15], a phenomenon that leads to the absence of this channel at the plasma membrane and results in cystic fibrosis (CF). Third, we developed an in vitro system in which the delivery of a mutated, yeast prepheromone from the ER to the cytoplasm was observed, and found that the exported polypeptide, known as pαF, was rapidly degraded . By adding lysates from proteasome mutant strains and through the application of small molecule inhibitors we discovered subsequently that pαF degradation was proteasome-mediated and termed this process ER-associated degradation (ERAD) . Fourth, concurrent with our efforts, Wolf and colleagues found that CPY*, a misfolded vacuole-targeted protein in yeast, was delivered from the ER to the cytoplasm for degradation . Fifth, Hampton and co-workers discovered that the proteasome was responsible for the sterol-regulated degradation of hydroxymethylglutaryl coenzyme A reductase (HMG-CoA-R) in yeast ; subsequent work from the Hampton laboratory indicated that sterols trigger a conformational change in HMG-CoA-R . And sixth – and also concurrent with these efforts – Ploegh and colleagues showed that human cytomegaloviral gene products trigger the dislocation and proteasome degradation of major histocompatibility class I molecules from the mammalian ER . These collective data indicated that soluble and integral membrane proteins in the ER can be delivered from the ER lumen and membrane and destroyed by the proteasome. The data also highlighted the fact that the ERAD pathway can be employed to regulate protein expression, and that basic elements of the pathway are conserved from yeast to humans .
As the number of ERAD substrates grew, they were classified into distinct groups. Proteins with lesions in the luminal space were referred to as `ERAD-L' substrates, proteins with lesions within the membrane were tagged `ERAD-M' substrates and proteins with lesions in the cytoplasm were called `ERAD-C' substrates . Based on studies in yeast, the classes were distinguished by their E3 ubiquitin ligase dependence . In addition, as the number of ERAD substrates grew, new links between ERAD and human diseases emerged [24,25]. For example, in antitrypsin deficiency, in a familial form of dementia and in hypofibrinogenaemia, disease arises because the ERAD machinery fails to keep pace with the synthesis of aggregation-prone mutated proteins [26–28]. In other cases, the substrate may remain soluble, but accumulation triggers a stress response. The stress response is potentiated if proteasome activity is inhibited. This may explain the successful application of proteasome inhibitors to treat multiple myeloma . Finally, a disorder can arise because ERAD is too efficient, and a functional, albeit mutated protein is prematurely destroyed. This scenario is encountered during the development of CF, which remains the most common, lethal inherited disease in North America.
Approximately one-third of the eukaryotic proteome is targeted to the ER . Consequently, ER-associated chaperones must distinguish kinetically trapped, unfolded polypeptides and proteins that have failed to mature from the bulk of proteins en route to acquiring their native conformations. Although the rules that govern substrate–chaperone interactions are still being established, work from our laboratory and others indicates that the Hsp70 and Hsp40 chaperones facilitate ERAD substrate selection and targeting . Hsp70 chaperones bind and release peptides with overall hydrophobic character. The affinity for peptides is highest when the Hsp70 is bound to ADP, but is lower when ATP is bound. Therefore, these chaperones bind and release their substrates concomitant with rounds of ATP binding, ATP hydrolysis and nucleotide exchange . Hsp40s serve as Hsp70 co-chaperones by virtue of the fact that they accelerate Hsp70's rate of ATP hydrolysis, which helps Hsp70s lock onto peptide substrates. Many Hsp40s also bind peptide substrates directly. This allows Hsp40s to deliver substrates to Hsp70s and then catalyse substrate hand-off [32,33]. Ultimately, a peptide is released from Hsp70 when the bound ADP is released and ATP rebinds. As a consequence of this cycle (figure 1), Hsp70–Hsp40 chaperone pairs can aid in maintaining polypeptide solubility and in facilitating protein folding, transport and modification.
We found that the ER luminal Hsp70, BiP (also known as Kar2 in yeast), partners with two luminal Hsp40s, Scj1 and Jem1, to maintain the solubility of the luminal ERAD substrates CPY* and pαF [34,35] (Table 1). We also discovered that the cytosolic Hsp70 Ssa1, which partners with cytosolic Hsp40s including Hlj1 and Ydj1, is involved in the degradation of the ERAD-C substrate, CFTR, in yeast [36,37]. In addition, we determined that small heat shock proteins (sHsps), which potently inhibit protein aggregation, are upregulated by CFTR expression in yeast. Consistent with this fact, yeast lacking the genes encoding two functionally redundant sHsps were unable to degrade CFTR. Moreover, we showed that human homologues of the yeast sHsps facilitate the degradation of the major disease-causing variant of CFTR and of another misfolded protein, the epithelial sodium channel, ENaC, in mammalian and vertebrate cells, respectively [38,39]. Like CFTR, ENaC degradation also requires the presence of two functionally redundant Hsp40 chaperones, but these chaperones reside within the ER lumen . Thus, ENaC exhibits some degradation requirements consistent with that of an ERAD-L substrate.
In contrast to the `pro-degradative' functions of the Hsp70–Hsp40s and the sHsps during ERAD, we discovered that the Hsp90 chaperone – which is thought to act at later stages in the protein folding pathway – stabilizes ERAD-C substrates . A different chaperone, Hsp110, helps stabilize another substrate from being degraded by the ERAD pathway in both yeast and mammalian cells . The substrate is an apolipoprotein, Apolipoprotein B, and the co-translational proteolysis of this factor is metabolically regulated in order to control the delivery of cholesterol, cholesterol esters and phospholipids from the liver. Together, these data indicate that chaperones either can facilitate folding or degradation in the ER. Increasing evidence also suggests that a given chaperone can play both roles, depending on the nature of the substrate .
After soluble ERAD substrates are selected, they must be ferried across the membrane. In contrast, membrane proteins must be stripped from the lipid bilayer, either prior to or during degradation; these events have been termed `retrotranslocation' or `dislocation', which will both be used in this review. It was consequently proposed that an ER channel is required for ERAD. Candidates for the pore included Sec61 [43–45] and the Derlins, an ER membrane protein family that binds factors linked to substrate selection, ubiquitination and retro-translocation [46,47]. It has also been proposed that integral membrane E3 ubiquitin ligases can function secondarily as protein-conducting channels: Hrd1, Doa10 and gp78 are composed of 5–14 transmembrane (TM) domains and complex with proteins required for retro-translocation and substrate processing [48–50]. The E3s may form oligomers, further increasing their potential to form channels . Various efforts implicate one or another of these candidates as contributing to channel activity. However, proof for a single channel is lacking. Thus, the channel might be composed of multiple proteins, it might form transiently and/or it might be dispensable if lipid microvesicles form in the ER independent of a channel and deliver membrane proteins from the organellar bilayer . For membrane proteins, it is also possible that the channel requirements depend on the overall hydrophobicity of the TM domain.
In nearly all cases ERAD substrates are ubiquitinated, and substrate retro-translocation requires the AAA protein, Cdc48 (p97 or VCP in mammals), which associates with Ufd1–Npl4 . The Cdc48–Ufd1–Npl4 complex in yeast (hereafter referred to as the `Cdc48 complex') binds ubiquitinated substrates and couples ATP hydrolysis with retro-translocation. One exception to this rule is pαF. pαF does not appear to be ubiquitinated , and consistent with this observation we discovered that retro-translocation is Cdc48-independent [54,55]. Instead, the 19S particle or cap of the proteasome is sufficient to extract pαF from yeast or mammalian microsomes. Similarly, the cholera toxin A1 chain, which is probably not ubiquitinated, is retrotranslocated independent of p97 function . A two-step model, with sequential action of Cdc48 and the 19S particle has also been proposed . However, by using an in vitro system in which the selection, ubiquitination and retrotranslocation of a membrane protein could be recapitulated, we showed an absolute dependence of Cdc48 for substrate extraction .
The binning of ERAD substrates into the L, M and C classes helped to develop and test new hypotheses. Notably, two E3s in yeast, Doa10 and Hrd1, ubiquitinate well-characterized members of each class, and the ligases associate with a non-overlapping set of components [23,59,60]. (It is worth noting that the Cdc48 complex binds both assemblies.) Even though much has been learned from these efforts, caveats to the substrate classification system exist. Most importantly, the designations were derived from the analysis of a relatively small number of substrates. Given the expanded number of E3s that catalyse substrate ubiquitination in mammals it is unlikely that the rules for substrate classification will be universal . Third, our laboratory and others uncovered partially redundant requirements for Doa10 and Hrd1 [58,62]. Fourth, complete stabilization is rarely observed in yeast lacking Doa10 and/or Hrd1, suggesting that other E3s (or alternate degradative mechanisms) are in play. These other E3s linked to ERAD under special circumstances include Rsp5 and Hul5 [63,64]. And fifth, the choice of which pathway is employed – or the selection of an alternative quality control pathway – may reflect substrate abundance. For example, we discovered that an aggregation-prone ERAD substrate is destroyed by the autophagic pathway but only when overexpressed .
During or after ubiquitination – and probably while still bound to the Cdc48 complex – ERAD substrates interact with proteasomes and proteasome-associated factors . Some of the factors bind polyubiquitin chains, which facilitate delivery into the 20S core of the proteasome. Although most models depict the proteasome as the sole mediator of degradation, there is evidence that other pathways, such as delivery to the vacuole/lysosome  or autophagy (see above) contribute to substrate turnover. These alternate routes may become apparent because of substrate-specific interactions, substrate load, ubiquitin depletion or compromised proteasome activity. In fact, select neurological disorders may derive from the proteasome becoming overwhelmed, which leads to the accumulation of toxic aggregates or pre-aggregates . By employing a yeast model for antitrypsin deficiency, we found that reduced proteasome activity gives rise to differing effects on unique ERAD substrates . This may result from variable rate-determining steps in the degradation pathway for each substrate. Modulation of this step should give rise to the greatest effect on substrate stability, thus providing the best targets to treat ERAD-linked diseases.
As should be evident from this review, the ERAD pathway requires the functions of many components, and as the analysis of this pathway matures it is becoming increasingly clear that unique substrates require distinct components during degradation . Based on this fact, one might envision the design of pharmaceuticals that target the stability of a single ERAD substrate. On the other hand, the general function of the ER quality control machinery appears to be quite plastic, and attempts to chemically modulate the `proteostasis network' may provide a more general route to treat a variety of protein conformational diseases . Because new inhibitors of substrate ubiquitination are continuously being developed, we believe that synergistic effects may be uncovered when substrate-specific ubiquitination modulation is paired with more general proteostasis regulators (figure 2). Based on the growing number of diseases that have been linked to ERAD, the time is ripe to employ these combinatorial strategies.
One target for modulation of the ERAD pathway is pancreatic islet cells, which synthesize, fold and secrete large amounts of insulin. As might be anticipated, these cells exhibit ER stress and islets that are unable to mount a UPR are more susceptible to apoptosis than wild-type islets. In addition, chemical chaperones that facilitate protein folding lessen the severity of several metabolic defects in obese mouse models. Moreover, individuals with genetic diseases in which the UPR cannot be induced exhibit diabetes . Therefore, increasing ERAD, along with augmenting the folding capacity of the ER should provide a synergistic means to offset some forms of diabetes. This hypothesis is supported by the fact that the overexpression of BiP, an ER luminal chaperone that aids in the selection of ERAD substrates and helps fold secreted proteins, is sufficient to decrease hepatic steatosis in insulin-resistant obese mice . In the future, it will be exciting to determine whether small molecules can be developed that can mimic the effect of BiP overexpression.
Research on ERAD in the Brodsky laboratory is supported by grants from the National Institutes of Health (GM75061, DK65161 and HL58541) and the Cystic Fibrosis Foundation Therapeutics (BRODSK08XX0). We also acknowledge support from National Institutes of Health grant DK79307, which supports the Pittsburgh Center for Kidney Research.
Conflict of Interests The authors do not declare any conflict of interest relevant to this manuscript.