inhibition of JNK or dominant-negative TRAF2 prevents autophagy induced by ER stress, suggesting that IRE1α controls autophagy through the recruitment and activating phosphorylation of TRAF2/JNK.

EIF2α phosphorylation by GCN2 and PKR is known to be involved in the mediation of autophagy during starvation and herpes simplex virus (HSV) infection, respectively. Similarly, phosphorylation of eIF2α by PERK has been suggested to mediate autophagy in response to ER stress induced by cytosolic accumulation of polyQ aggregates. In this system, autophagy is blocked by either expression of a dominant-negative form of PERK or the eIF2αA/A mutant protein, which cannot be phosphorylated. The PERK/eIF2α pathway is also required for the upregulation of the autophagy mediator ATG12 at both the mRNA and protein levels in response to ER stress.

Finally, calcium flux perturbations occurring during ER stress have been suggested to participate in autophagy signaling. Consistent with the proposed role of calcium, autophagy induced in MCF-7 cells and hepatocytes by thapsigargin, an inhibitor of the ER Ca2+- dependent ATPase, is blocked by the calcium chelator BAPTA-AM. Influx of calcium into the cytosol in response to ER stress leads to the activation of the AMP-activated protein kinase (AMPK) through the Ca2+/calmodulindependent protein kinase kinase-β (CaMKK-β). AMPK, in turn, negatively regulates the mammalian target of rapamycin (mTOR) protein, leading to the induction of autophagy. Recently, the kinase PKCθ has also been shown to mediate thapsigargin-induced autophagy in immortalized hepatocytes. Interestingly, this appears to be independent of mTOR activity, suggesting that multiple independent pathways may mediate induction of autophagy during the ESR.

4. THE ESR AND CELL DEATH

The ESR acts to restore normal ER homeostasis and therefore is cytoprotective. However, when a stress is so strong or persistent that ER dysfunction cannot be corrected, metazoan cells can initiate apoptosis, allowing the regulated destruction of cells that are irreparably damaged or a risk to the organism as a whole. A unified model for ER stress-induced apoptosis is only beginning to emerge, but recent interest in the field has generated an increasing amount of information (see Figure 6-2 for an overview).

Some core components of the ESR can function in ER stress-induced apoptosis as well. For example, mammalian Ire1 can activate JNK and other proapoptotic kinases such as ASK1, which may contribute to ER stress-induced apoptosis. ER stress inducers such as

tunicamycin, thapsigargin, or reducing agents, as well as the over-expression of IRE1α, induce the activation of JNK. IRE1 also interacts with TRAF2, an adaptor protein involved in the signaling pathways of proinflammatory cytokines such as tumor necrosis factor α and interleukin-1. This interaction can recruit ASK1 to form an IRE1/TRAF2/ASK1 ternary complex, which in turn can activate JNK. The functional importance of JNK activation in the ER stress pathway has not been fully explored, but ASK1–/– cells are partially resistant to ER stress-induced apoptosis, suggesting that JNK may promote apoptosis in this context.

CHOP has also been shown to promote apoptosis, and this effect can be blocked by BiP over-expression, suggesting that CHOP-activated apoptotic pathways are downstream from the ER. CHOP can transcriptionally downregulate the antiapoptotic protein Bcl-2 and upregulate DR5, a member of the death receptor protein family, two effectors that function in non-ER forms of cell death as well. Interestingly, CHOP also leads to a depletion of cellular glutathione and an increase of reactive oxygen species (ROS) in the ER, due in part to its induction of ERO1α, an ER oxidase. Interfering with ERO1α function reduces the accumulation of ROS in the stressed ER, leading to cytoprotection. This implies that ERO1α may be an important apoptotic effector downstream of CHOP. CHOP–/– MEFs are partially resistant to ER stress-induced cell death, although CHOP–/– mice are not resistant to lethal doses of tunicamycin, suggesting that other proapoptotic pathways are also at work.

The eIF2α phosphatase cofactors GADD34 and CReP also mediate apoptotic signaling downstream from the ER. During ER stress, GADD34–/– cells display persistent eIF2α phosphorylation and fewer misfolded protein aggregates in the ER lumen, suggesting that GADD34 function is proapoptotic. Indeed, GADD34–/– mice are resistant to the toxic effects of tunicamycin. Similarly, RNAi-mediated knockdown of CReP protects cells from a variety of stimuli, including ER stress. It seems likely that the cytoprotection provided by the loss of GADD34 or CReP is due primarily to increased eIF2α phosphorylation because enforcing eIF2α phosphorylation with a constitutively active PERK has a similar effect.

ER stress can also directly activate well-known general regulators of mammalian apoptosis, including the Bcl-2 and caspase families of proteins. It has long been known that a pool of endogenous Bcl-2 resides in the ER membrane, and although Bcl-2 family members are thought to function principally at the mitochondrial outer membrane, there is ample evidence that they influence homeostasis and apoptosis from the ER as well. For example, variants of the antiapoptotic family

cytoplasm

APOPTOSIS

members Bcl-2 or Bcl-XL targeted specifically to the ER membrane can block apoptosis induced by pharmacological kinase inhibition or by proapoptotic Bcl-2 family members. Conversely, ER stress itself can upregulate or otherwise activate several BH3-only proapoptotic members of the Bcl-2 family, including Bim, BIK, and PUMA. Therefore, efferent signaling from the stressed ER can engage the Bcl-2 death machinery directly.

Recent studies have also demonstrated that the proapoptotic multidomain Bcl-2 family proteins, Bax and Bak, regulate ER stress-induced apoptosis. Bax–/–/ Bak–/– MEFs are remarkably resistant to many forms of apoptosis, including ER stress, implying that ER stress and other apoptotic signaling pathways converge on Bax and Bak. Interestingly, endogenous Bax and Bak regulate ER stress-induced apoptosis from both the mitochondrial and ER membranes. In the ER membrane, Bax and Bak are crucial for maintaining the resting level of lumenal calcium, probably through an interaction with the type 1 inositol trisphosphate receptor. As a result, Bax–/–/ Bak–/– cells display reduced calcium release from the ER in response to such stimuli as arachidonic acid and

oxidative stress, thereby attenuating apoptosis. However, in response to other ER stress stimuli, such as the ER- to-Golgi vesicle transport inhibitor brefeldin A, Bax and Bak must be present at both the ER and mitochondrial membranes for apoptotic execution to proceed normally. Therefore, Bax and Bak participate in signal integration between the ER and the mitochondria to influence cell survival choices from multiple locations within the cell. Indeed, more recently, Bax and Bak were found to interact directly with IRE1α in the ER to promote UPR signaling, demonstrating a directly molecular connection between the ESR and the core apoptotic machinery.

The caspase family of proapoptotic cysteine proteases also plays a critical role in ER stress-induced apoptosis. Caspase-12, a murine protein associated with the cytosolic side of the ER membrane, is activated by ER stress-induced apoptosis, but not by non-ER stimuli, and is required for cell death in response to both pharmacological ER stress and ER-targeted Bim. Caspase-12 can be activated by ER stress in several ways. For example, the cytoplasmic calcium-activated protease calpain can cleave and activate caspase-12 in response to

calcium flux from the ER, which is often triggered by ER stress. Caspase-12 can also autoactivate through a direct association with IRE1α and the adaptor protein TRAF2, though how the formation of this complex is regulated by ER stress is not yet clear. Caspase-12 has also been detected in high-molecular-weight complex with apoptosis-linked gene-2 protein and the ERAD mediator p97 (also referred to as valosin-containing protein). Interference with the formation of this complex protects cells from ER stress-induced apoptosis, presumably by blocking the activation of caspase-12. Because p97 is also involved in ERAD, it represents another protein that may mediate cross-talk between the prosurvival and proapoptotic pathways induced by ER stress.

Once activated, caspase-12 can initiate downstream apoptotic pathways. For example, ER stress can induce the activation of caspase-9 independent of Apaf-1, the usual mediator of caspase-9 activation. In addition, caspase-7 translocates to the ER in response to some apoptotic stimuli, and it has been proposed that caspase-7 can directly activate caspase-12. However, other experiments suggest that caspase-12 cleavage precedes caspase-7 cleavage under ER stress conditions, implying that the order of activation may be the opposite. In addition, glycogen synthase kinase (GSK) 3β may influence caspase-3 activation specifically during ER stress, although whether this is a direct effect or a farupstream event remains unclear.

It is worth noting that the role of human homologs of caspase-12 in ER stress-induced apoptosis has been controversial. However, it was recently shown that human caspase-4, which is 48% homologous to murine caspase12, is localized to the ER membrane and is specifically activated by and required for ER stress-induced apoptosis. These data suggest that caspase-4 is the human functional counterpart of murine caspase-12.

Autophagy is generally a protective response against cellular stresses, but under some circumstances, especially when apoptosis is blocked, it can also contribute to cell death (referred to as type II programmed cell death). In fact, it has been a subject of much debate whether autophagy plays a pro-death or prosurvival function during ESR. It is clear that autophagy is cytoprotective during yeast ESR. Under most circumstances, this seems to be the case in mammalian cells as well. For example, ER stress-induced caspase-12 activation and cell death are enhanced in ATG5-deficient MEFs and in MEFs treated with the type III PI3 kinase inhibitor 3- methyladenine (3-MA), a potent inhibitor of autophagy. In contrast, autophagy may contribute to the cell death process in apoptosis-resistant cells, as ATG5-deficient

MEFs expressing Bcl-XL are more resistant to ER stressinduced cell death than wild-type MEFs expressing BclXL. In addition, the influence of autophagy on the cell survival after ER stress may also depend on whether the cells have been transformed, as primary ATG5-deficient MEFs have been reported to be less sensitive to ER stressinduced cell death, and 3-MA protects nonimmortalized normal human colon cell line CCD-18C against ER stress toxicity.

5. THE ESR IN DEVELOPMENT AND TISSUE HOMEOSTASIS

In mammals, several components of the ESR are essential for organismal development and tissue homeostasis. As in yeast, this likely reflects the fact that normal physiologic fluctuations can strain the capacity of the ER such that cells need the ESR to cope with the secretory burden. As might be expected from this model, professional secretory cell types, such as plasma B cells and pancreatic β-islet cells, are among the most severely affected when the ESR is compromised.

Ire1α–/– mouse embryos die around day 10.5 of gestation as a result of unknown causes, but Ire1α–/– MEFs exhibit no obvious defect in the UPR. It may be that some embryonic cell types rely on IRE1α for UPR signaling even though fibroblasts do not, or that IRE1α is important to non-UPR functions in the embryo as well. On the other hand, mice deficient for IRE1β restricted to the intestines develop without obvious abnormalities but display increased sensitivity to colitis.

Further downstream in the UPR, XBP-1 is essential for fetal hepatocyte growth. XBP-1–/– embryos have hypoplastic livers with reduced cell proliferation and increased apoptosis and show reduced hematopoiesis that results in severe anemia and death. Whether this phenotype is due solely to the role of XBP-1 in the ESR is not yet clear. However, XBP-1 is also known to play an important role in plasma B cell differentiation, as XBP-1–/– lymphocytes transplanted into RAG-2 chimeric mice display a severe defect in the generation of plasma cells. XBP-1 coordinates a broad transcriptional program in the developing B cell, upregulating many genes in the secretory pathway and the physical expansion of the ER. These data suggest that the XBP-1-mediated component of the UPR is critical for the ability of certain secretory cell types to develop and function.

The translational control pathway of the ESR also plays a prominent role in development. PERK is highly expressed in the pancreatic acini that secrete digestive enzyme, and in the islets of Langerhans, which produce and secrete the polypeptide hormones insulin and glucagon. Active, phosphorylated PERK can be

detected in the wild-type pancreas, lung, liver, spleen, and thymus. Furthermore, the normal phosphorylation of eIF2α is lost in the PERK–/– pancreas and thymus, suggesting that translational control downstream of the ER normally operates in those organs. Consistent with this hypothesis, compensatory activation of IRE1α is detected in the pancreas of PERK–/– mice.

PERK–/– mice survive to birth but develop progressive disturbances in glucose metabolism and hypoinsulinemia. Although the neonatal pancreas of PERK–/– mice is largely normal, the number of insulin-producing cells decreases over time and is associated with a profound increase in apoptosis in the pancreas. Therefore, PERKmediated translational control is necessary to manage the processing and export of secreted proteins such as insulin under normal physiologic conditions, and inappropriate apoptosis and disease can occur when this aspect of the ESR is impaired.

Similar conclusions can be drawn from eIF2αA/A mice, which exhibit a more severe phenotype. At birth, eIF2αA/A mice are indistinguishable from their wildtype or heterozygous littermates, but most die within 18 hours as a result of hypoglycemia. Indeed, the induction of gluconeogenic enzyme genes in the liver is significantly reduced. Because fetal glycogen synthase expression is normally induced to promote the storage of maternal glucose as liver glycogen for survival during the early neonatal period, the reduced ability to store glycogen may contribute to the hypoglycemia in eIF2αA/A neonates. In addition, insulin levels in the eIF2αA/A pancreas are significantly lower than that of wild type.

The severe defects in glucose and insulin metabolism in PERK–/– and eIF2αA/A mice suggest that ER protein folding and the regulation of eIF2α phosphorylation are important homeostatic controls in glucose and glycogen metabolism. Because eIF2αA/A mice exhibit an earlier and more severe phenotype than PERK–/– mice, additional eIF2α kinases may participate in normal fetal and neonatal development. However, none of the single knockouts of the other three eIF2α kinases exhibits any abnormality in glucose metabolism, suggesting that two or more of these enzymes in combination may be responsible for liver glycogen regulation.

Interestingly, mice with targeted deficiencies in genes involved in ER stress-induced apoptosis, including caspase-12, CHOP, EF2K, and BIK/Blk, generally display no obvious deleterious phenotype, confirming that the main homeostatic function of ESR lies in its ability to adjust cellular function to accommodate increased secretory burden, rather than in mediation of cell death in response to ER stress.

6. THE ESR IN HUMAN DISEASE

Dysregulation of the ESR or the apoptotic pathways coupled to it have been implicated in myriad human diseases. Three broad classes of diseases involving ER stress – diabetes, neuronal ischemic injury, and viral infection – are addressed briefly here as examples.

ER stress can contribute to both insulin-dependent (type 1) and insulin-resistant (type 2) diabetes. Interestingly, defects in the PERK-dependent eIF2α phosphorylation pathway cause a disease in humans, known as Wolcott-Rallison syndrome (WRS), which is reminiscent of the phenotype of PERK–/– mice. WRS is a rare, autosomal-recessive disorder characterized by permanent neonatal or early infancy nonautoimmune type 1 diabetes, with epiphyseal dysplasia, osteoporosis, and growth retardation occurring later. Mutations in the eukaryotic translation initiation factor 2-alpha kinase 3 gene (EIF2AK3), which encodes the human homolog of PERK, are the underlying genetic defect in WRS families. WRS-associated mutations in EIF2AK3 abrogate the eIF2α kinase function of human PERK, confirming the involvement of the ESR in WRS. These findings have led to the suggestion that defects in other aspects of the ESR may contribute to type 1 diabetes, or that polymorphisms in the EIF2AK3 gene may predispose some nonWRS patients to develop type 1 diabetes.

More recently, ER stress has also been implicated in type 2 diabetes. Both genetic and diet-induced obesity in mice are associated with the activation of the ESR, possibly due to an increased burden on the cells of secretory organs. Interestingly, activation of the ESR antagonizes insulin signaling in both mouse models via the JNK-mediated phosphorylation of insulin receptor substrate 1. Furthermore, enforcing the expression of XBP-1 to increase UPR activation resulted in less ER stress and restored insulin sensitivity and proper glucose homeostasis. Intriguingly, “chemical chaperones” – or pharmacological agents that promote protein folding and ameliorate consequent ER stress – reduce hyperglycemia, tissue insulin sensitivity, and fatty liver disease in obese and diabetic mice. Therefore, ER stress may be a major upstream cause of diabetic symptoms in some contexts. Consistent with this notion, XBP-1+/– mice are more susceptible to obesity-induced diabetes than were wild-type mice. Mutations or obesity-induced dysfunction in the UPR component of the ESR may therefore be a risk factor for type 2 diabetes in humans as well.

Interestingly, recent work suggests that ESR also contributes to lipid and carbohydrate metabolism in the liver. XBP-1 is induced in the livers of mice fed a highcarbohydrate diet, in conjunction with genes involved

in fatty acid biosynthesis. Conversely, deletion of XBP- 1 in the liver results in a reduction of hepatic lipid biosynthesis and, secondarily, hypocholesterolemia and hypotriglyceridemia. Similarly, recent work also showed that eIF2α dephosphorylation in the liver enhances glucose tolerance and steatosis in mice. Although not directly linked to diabetes per se, these newly discovered functions of core ESR components indicate that the ESR may integrate signals from various nutritional cues in the body to control the physiologic response to dietary carbohydrates and lipids. Therefore, dysfunctional ESR signaling may contribute to a range of metabolic disorders.

Activation of the ESR in acute ischemia/reperfusion (I/R) brain injury is also well documented. It has long been known that protein synthesis is inhibited in ischemic neurons on reperfusion and that translation gradually resumes in regions resistant to damage while remaining suppressed in vulnerable neurons. Phosphorylation of eIF2α and inhibition of translational initiation are also found in I/R-injured brains. Some canonical downstream consequences of eIF2α phosphorylation, such as the induction of ATF4, CHOP, and GADD34, are also observed, indicating that the translational control pathway of the ESR may be activated in toto. However, the protein synthesis block can persist even after eIF2α has been dephosphorylated, suggesting that other inhibitory mechanisms exist.

EIF2α phosphorylation in I/R injury is likely caused by a strong and rapid activation of PERK. Consistently, PKR–/–, HRI–/–, and GCN2–/– mice show no reduction in I/R-induced eIF2α phosphorylation, implying that PERK is solely responsible for this activity. However, the perinatal lethality of PERK–/– mice precludes testing them directly in an I/R model.

Exactly how I/R might activate PERK is not clear. Protein aggregates are observed inside I/R injured neurons, suggesting that perturbation of protein folding may play a role. Another possibility is that ROS accumulation in the ER leads to PERK activation. It has also been proposed that disturbances in ER calcium homeostasis, which are common in I/R injury, may cause ER stress in reperfused neurons. Indeed, inhibitors of ER calcium release, such as dantrolene or TMB-8, can protect cells from ischemic or excitotoxic injury. However, it is not yet clear whether eventual neuronal death is due to the loss of ER lumenal calcium per se or the resultant spike of calcium in the cytoplasm (or both).

More recently, I/R injury has been found to induce other components of the ESR. Although activation of ATF6 is not observed, IRE1 is activated and XBP-1 is spliced and translated soon after reperfusion. Furthermore, caspase-12 activation occurs in both transient and

permanent models of focal ischemia, suggesting that ERspecific apoptotic pathways are also engaged. Caspase12 activation might be an important, general proapoptotic event in I/R, as caspase-12–/– mice are resistant to cell death in a related model of cardiac ischemia. However, it is not yet clear whether the activation of other ESR pathways is cytoprotective or cytotoxic in I/R injury. Selective pharmacological manipulation of certain ESR components may help to resolve this question.

As obligate intracellular parasites, viruses enforce the production of large amounts of polypeptides by cellular machinery and therefore place a heavy burden on organelles sensitive to protein overload, such as the ER. In particular, certain classes like the flaviviruses, which include important human pathogens such as the hepatitis C, dengue, yellow fever, and West Nile viruses, use the ER as the primary site for polyprotein processing, envelope glycoprotein biogenesis, and virion formation. Such ER-tropic viruses can perturb normal ER homeostasis and engage the ESR. Indeed, the protein products of the vesicular stomatitis virus (VSV), Sindbis, rabies, hepatitis C virus (HCV), the neurovirulent murine FrCasE virus, and the measles virus (MV) are all found in specific complexes with ER lumenal chaperones. In some cases, interaction between the host chaperones and the viral proteins is essential for viral protein processing and virion assembly. In addition, a glut of viral glycoproteins in the ER may compete with cellular proteins for chaperones, leading to ER stress and the activation of the ESR.

In fact, ESR activation has been observed in many instances of viral infection. For example, Japanese encephalitis virus causes the hypertrophy of the ER membrane and induces such canonical UPR targets as BiP, calnexin, Grp94, and CHOP. Similarly, MV infection upregulates myriad ESR targets, including BiP, calreticulin, calnexin, Grp94, CHOP, and ATF4, implying the activation of both the UPR and PERK pathways. HCV also induces the cleavage and activation of ATF6, followed by upregulation of BiP and CHOP. Furthermore, the induction of BiP by HCV infection is dependent on the ESR elements in the BiP promoter, indicating an ESR-specific signal. Interestingly, the HCV E2 protein, which resides in the ER membrane, causes ER stress when expressed at low levels but, at higher levels, binds to and inhibits PERK, presumably to block eIF2α phosphorylation and allow viral protein synthesis to continue.

ER stress-induced apoptosis is also observed during viral infection. A cytopathic strain of bovine viral diarrhea virus (BVDV) activates PERK and eIF2α, upregulates proapoptotic molecules such as CHOP and caspase-12, and downregulates Bcl-2. Therefore, ER stress may be the primary proapoptotic stimulus that