Endoplasmic reticulum proteostasis: a key checkpoint in cancer

Abstract

The unfolded protein response (UPR) is an intracellular signaling network largely controlled by three endoplasmic reticulum (ER) transmembrane proteins, inositol-requiring enzyme 1α, PRK-like ER kinase, and activating transcription factor 6, that monitor the protein-folding status of the ER and initiate corrective measures to maintain ER homeostasis. Hypoxia, nutrient deprivation, proteasome dysfunction, sustained demands on the secretory pathway or somatic mutations in its client proteins, conditions often encountered by cancer cells, can lead to the accumulation of misfolded proteins in the ER and cause “ER stress.” Under remediable levels of ER stress, the homeostatic UPR outputs activate transcriptional and translational changes that promote cellular adaptation. However, if the ER stress is irreversible despite these measures, a terminal UPR program supersedes that actively signals cell destruction. In addition to its prosurvival and prodeath outputs, the UPR is now recognized to play a major role in the differentiation and activation of specific immune cells, as well as proinflammatory cytokine production in many cell types. Given the numerous intrinsic and extrinsic factors that threaten the fidelity of the secretory pathway in cancer cells, it is not surprising that ER stress is documented in many solid and hematopoietic malignancies, but whether ongoing UPR signaling is beneficial or detrimental to tumor growth remains hotly debated. Here I review recent evidence that cancer cells are prone to loss of proteostasis within the ER, and hence may be susceptible to targeted interventions that either reduce homeostatic UPR outputs or alternatively trigger the terminal UPR.

The Unfolded Protein Response: A Signaling Pathway that Determines Cell Fate under ER Stress

the endoplasmic reticulum (ER) plays a major role in the synthesis, folding, and structural maturation of over one-third of all proteins made in the cell, including nearly all secreted proteins (1). After cotranslational translocation into the ER lumen, these proteins must be folded into their correct three-dimensional shapes and modified by ER-resident enzymes, such as chaperones, glycosylating enzymes, and oxido-reductases (94, 108). Despite the efforts of these protein-folding machines, at least one-third of all polypeptides translocated into the ER are improperly folded, and, for some proteins, the success rate is much lower (90). Incompletely folded forms are eliminated by quality control systems, including the ER-associated degradation (ERAD) pathway and autophagy (65, 67, 104).

When misfolded proteins in the ER accumulate above a critical threshold, a signal transduction pathway, called the unfolded protein response (UPR), is initiated to respond to this loss in ER proteostasis (39). In mammalian cells, the UPR is controlled by at least three ER transmembrane proteins, inositol-requiring enzyme 1α [IRE1α; also known as ERN1 (endoplasmic reticulum to nucleus signaling 1)], PRK-like ER kinase [PERK; also known as eukaryotic translation initiation (EIF) 2AK3], and activating transcription factor (ATF) 6α, each of which contains an ER luminal domain capable of directly or indirectly sensing misfolded proteins (87, 116). During homeostatic conditions, the luminal domains of IRE1α, PERK, and ATF6α are maintained in an inactive/monomeric state through association with an ER chaperone, called binding immunoglobulin protein [BiP; also known as GRP78 (78-kDa glucose-regulated protein) and HSPA5 (heat shock protein family A member 5)]. As BiP/GRP78 has a higher affinity for misfolded proteins, it is titrated away from the ER stress sensors when misfolded proteins begin to accumulate within the ER lumen, thereby releasing and priming the stress sensors for downstream signaling (82). The engagement of the ER luminal domain by misfolded protein then leads to changes in the oligomerization state of each sensor and activation of their associated downstream activities, thereby transducing a signal from the ER lumen into the cytoplasm (17, 28). The early outputs of the UPR attempt to restore homeostasis by increasing protein-folding capacity so the cell can continue to survive and function (114). However, if these adaptive responses are inadequate, these same ER sensors initiate an alternative response called the “terminal UPR,” which actively promotes cell death (73, 99).

IRE1α is a bifunctional kinase/endoribonuclease (RNase) that uses kinase autophosphorylation as a rheostat of ongoing ER stress to determine cell fate under these conditions (Fig. 1) (30, 35). Remediable ER stress causes low-level, transient kinase autophosphorylation and dimerization/tetramerization that restricts IRE1α’s RNase activity to a single adaptive task: excising an intron in XBP1 (X-box binding protein 1) mRNA. Religation of spliced XBP1 shifts the open reading frame, and its translation produces the homeostatic transcription factor XBP1s (s = spliced) (10, 121), which upregulates genes encoding ER protein-folding and quality control components (53). Sustained, high-level autophosphorylation, however, causes higher order IRE1α oligomerization; under these conditions, IRE1α’s RNase relaxes its specificity and endonucleolytically degrades many mRNAs at the ER membrane that encode secretory proteins (30, 35, 40), a process called regulated IRE1α-dependent decay (RIDD). Depletion of cargo-encoding transcripts through RIDD may initially be protective by reducing the protein-folding burden on the ER, but the indiscriminate degradation of mRNAs at the ER eventually depletes transcripts encoding structural and enzymatic components of the ER protein-folding machinery. The net consequence of RIDD is that ER function actively deteriorates. Moreover, RIDD reduces the levels of select micro-RNAs (possibly by directly cleaving their precursors at the ER membrane), such as miR-17, which normally repress proapoptotic targets, such as the prooxidant protein TXNIP (thioredoxin-interacting protein), leading to their rapid upregulation (57, 110). Increased TXNIP leads to sterile inflammation and pyroptotic cell death (123). Thus the hyperactive IRE1α RNase induces a terminal UPR in which adaptive signaling through XBP1 mRNA splicing is eclipsed by proapoptotic signals.

Fig. 1.

Role of the unfolded protein response (UPR) in cancer. Tumors frequently encounter extrinsic stresses that compromise protein folding in the endoplasmic reticulum (ER), including hypoxia, glucose deprivation, lactic acidosis, oxidative stress, and inadequate amino acid supplies. Moreover, intrinsic stresses, such as oncogene activation, changes in chromosomal number, and increased glycolysis, can all lead to an upregulation in protein translation and additional demands on the secretory pathway. Furthermore, genomic instability and somatic mutations in client proteins of the secretory pathway can cripple their folding and lead to ER stress. In response to an accumulation of ER misfolded proteins, the UPR is initiated by three transmembrane ER proteins: inositol-requiring enzyme 1α (IRE1α; also known as ERN1), PRK-like ER kinase (PERK; also known as EIF2AK3), and activating transcription factor (ATF) 6α. At low levels of ER stress, the bifunctional IRE1α kinase/RNase dimerizes/tetramerizes to cleave a nonconventional intron from XBP1 mRNA, which upon religation encodes the XBP1s transcription that upregulates ER protein-folding and quality control components to promote adaptation. However, if hyperactivated by sustained ER stress, IRE1α oligomerizes, and its relaxed RNase activity endonucleolytically degrades many mRNAs, a process called regulated IRE1α-dependent decay (RIDD), at the ER membrane to cause cell death. PERK signaling downregulates Cap-dependent protein translation through phosphorylation of eIF2α, while upregulating the expression of the ATF4. In the presence of misfolded proteins, ATF6α translocates to the Golgi and is cleaved by the site 1 and site 2 proteases to release the p50 ATF6(N) transcription factor into the cytoplasm before migrating to the nucleus. Together with XBP1s, ATF6(N) increases transcription of targets that expand ER size and increase its protein-folding capacity, as well as that of the ER-associated degradation (ERAD) pathway. The combined outputs of the UPR can influence tumor growth at many levels, including cell survival, angiogenesis, inflammation, antigen presentation, invasion, and metastasis. [Adapted by permission from Macmillan Publishers Ltd: Nature Immunology (Ref. 77, Fig. 1), copyright 2014.]

PERK contains a cytosolic kinase that phosphorylates eukaryotic translation initiation factor 2α (eIF2α), which inhibits assembly of the eIF2-GTP-Met-tRNA ternary complex to inhibit its activity and thus slow down global protein translation (6, 60), thereby decreasing protein-folding demand in the ER (37). Moreover, PERK actively signals adaptation because, while the translation of most Cap-dependent mRNAs is inhibited, eIF2α phosphorylation favors the internal ribosome entry site-mediated translation of select mRNAs, such as that encoding ATF4 (also known as cAMP response element binding protein 2), a key adaptive output that transcriptionally upregulates genes involved in amino acid metabolism, oxidative stress resistance, and autophagy (36). While a pause in overall protein translation can be beneficial for stressed cells by providing extra time to fold a backlog of proteins, a protracted block in translation from sustained PERK signaling is inconsistent with survival. Therefore, ATF4 also increases the expression of the gene encoding CADD34, a major regulatory subunit of protein phosphatase 1, which results in the dephosphorylation of eIF2α and restoration of mRNA translation (16). However, prolonged PERK activation also upregulates the transcription factor CCAAT-enhancer-binding protein homologous protein/growth arrest- and DNA damage-inducible gene 153, which inhibits expression of anti-apoptotic BCL2 and increases expression of pro-apoptotic BCL2 family proteins to hasten cell death (63, 66). Hence, similar to IRE1α, PERK switches from homeostatic to terminal UPR under chronic ER stress (Fig. 1).

ATF6α is a type II transmembrane protein that has a bZIP transcription factor contained within its cytoplasmic tail. In the presence of misfolded proteins, ATF6α translocates to the Golgi and is cleaved by the site 1 and site 2 proteases to release the p50 ATF6(N) transcription factor into the cytoplasm before migrating to the nucleus (34). Together with XBP1s, ATF6(N) increases transcription of targets (e.g., BiP, GRP94, p58IPK/DNAJC3) that expand ER size and increase its protein-folding capacity, as well as that of the ERAD pathway (30). The contribution of ATF6α to cell death under conditions of ER stress is not well understood at the present time (Fig. 1).

The ER Stress of Cancer

Tumor cells often invade or metastasize into foreign environments where unfavorable conditions, such as hypoxia, glucose deprivation, lactic acidosis, oxidative stress, and inadequate amino acid supplies, compromise protein folding in the ER (51, 56, 61, 69). Moreover, the intrinsic stresses common to many tumor cells, including oncogene activation, changes in chromosomal number, and increased glycolysis can all lead to an upregulation in protein translation and additional demands on the secretory pathway (22, 88, 107). Furthermore, genomic instability and somatic mutations in client proteins of the secretory pathway can cripple their folding and lead to ER stress (42). Consistent with this, many studies have documented sustained, high-level activation of the UPR (IRE1α, PERK, and ATF6α) in a wide range of primary human solid tumors, including glioblastoma and carcinomas of the breast, stomach, esophagus, and liver (27, 69, 100). In contrast, somatic mutations in IRE1α or PERK have been found rarely (<1%) in these tumors (32), which in most cases seem to be loss of function (30, 120). For example, we discovered that cancer-associated mutations in IRE1α disable its apoptotic outputs (30), suggesting that some cancer cells may impair the terminal UPR to survive. The ER chaperones BiP/GRP78 and GRP94 are likewise overexpressed in a variety of cancer types (21, 27, 54, 100). However, despite the overwhelming evidence of ongoing ER stress and UPR activation in many forms of cancer, whether this ultimately inhibits or promotes tumor growth in patients remains an area of intense debate. Most of the evidence arguing that the UPR supports tumor growth comes from xenograft studies in mice in which genetically deleting one or more branches of the UPR or altering the expression of the ER chaperones influences the in vivo growth of tumor cells.

Studies have shown that IRE1α’s homeostatic target XBP1 promotes tumor progression in models of triple-negative breast cancer (15), and that PERK supports tumor growth in murine models of mammary cancer by limiting oxidative damage (8), arguing that UPR signaling may be beneficial for tumor survival. Furthermore, deletion of IRE1α in a human glioma cell line resulted in reduced angiogenesis and decreased tumor growth when these cells were subsequently injected into mice (41). These findings suggest that, not only is the homeostatic UPR frequently activated in tumors, but it may be necessary for the survival and/or growth of the cancer cells under conditions that stress the ER.

The UPR in Immune Cell Development, Function, and Inflammation

The development of cancer and its response to therapy are strongly impacted by innate and adaptive immunity, which can either promote or diminish tumor growth (96). Over the past decade, it has become well-established that the UPR plays an essential role in the proper differentiation and function of many immune cell types (33). However, the UPR’s effects on the immune system are often ignored when discussing its role in cancer. This is because immunologists and cancer biologists have largely studied the UPR in either immune cells or tumor cells, respectively, but rarely cancer in an immunologically intact host. Therefore, the impact of the UPR on immune function needs to be considered and comprehensively studied in the context of cancer. As such, I will briefly review our current understanding of the role of the UPR in immunity and inflammation.

Given their function to secrete complex polypeptides in great abundance, plasma cells, neutrophils, and many other immune cells are particularly dependent on maintaining a large ER through the adaptive outputs of the UPR. For example, IRE1α and XBP1 are required for plasma cell development (50, 84, 91, 95, 106, 109), a differentiation process that involves the marked expansion of the ER necessary for high levels of antibody production and secretion. Dendritic cells show chronic IRE1α activation as evidenced by high levels of XBP1 splicing, and genetic deletion of Xbp1 results in significant reductions in the numbers of conventional and plasmacytoid dendritic cells (44, 45, 77). Eosinophils, macrophages, and Paneth cells are all highly dependent on the IRE1α/Xbp1 arm of the UPR for survival and/or function (reviewed in Ref. 33). In response to the absence of its homeostatic target XBP1, IRE1α becomes hyperactivated in many cell types, leading to increased RIDD and subsequent degradation of key secretory cargo products. For example, some of the decrease in immunoglobulin production in XBP1-deficient B cells seems to be a direct result of RIDD-mediated degradation of immunoglobulin-μ heavy-chain mRNAs rather than the absence of XBP1 transcriptional outputs (5). Likewise, dendritic cells subjected to Cd11c-Cre-mediated deletion of Xbp1 have defects in cross-presentation due to RIDD-dependent degradation of tapasin and other components of the cross-presentation machinery (77).

In addition to its importance in immune cell function, the UPR can be positively or negatively tuned by proinflammatory stimuli, such as cytokines and Toll-like receptor (TLR) ligands in many cell types. Cytokines can directly regulate and shape the UPR. Proinflammatory cytokines, including interleukin-1 (IL-1), IL-6, and tumor necrosis factor (TNF), can induce UPR activation in the liver, which seems to promote release of acute phase response products critical for fighting infection and tissue repair (122). On the other hand, IL-10 can temper inflammation-induced ER stress by preventing nuclear translocation of ATF6-p50 through a p38-MAPK-dependent mechanism (98). Pathogen sensors, such as TLRs, detect microbes to activate transcriptional programs that orchestrate immune responses against specific insults. TLR2 and TLR4 have been shown to specifically activate IRE1α and XBP1 splicing in macrophages where they are required for optimal and sustained production of proinflammatory cytokines (64).

Moreover, there is now abundant evidence that ER stress can directly turn on inflammatory pathways. For example, nuclear factor (NF)-κB, the master transcriptional regulator of multiple proinflammatory pathways, is strongly induced in response to ER stress. This is accomplished in part through IRE1α’s interaction with TNF receptor-associated factor 2 (TRAF2), which, upon IRE1α activation, leads to recruitment of IκB kinase and subsequent phosphorylation and degradation of IκB, such that NF-κB is released for translocation to the nucleus (43). PERK phosphorylation of eIF2α inhibits Cap-dependent protein translation, which, given the rapid turnover of IκB, leads to a greater NF-κB-to-IκB ratio (24, 105). In pancreatic β-cells and macrophages, ER stress induces activation of the NLRP3 (LRR and PYD domains containing protein 3) inflammasome to cause caspase-1 cleavage and IL-1β secretion. The mechanism by which ER stress leads to NLRP3 inflammation activation is at least in part through TXNIP, which is strongly induced in these cells downstream of IRE1α and/or PERK (9, 57, 76). Hence, in terms of their ability to activate the inflammation, ER misfolded proteins seem akin to damage-associated molecular pattern molecules.

ER stress and the UPR have also been found to impact the presentation of antigen on the cell surface (31, 81) and influence tumor antigenicity (83). For example, constitutive activation of XBP1 in tumor-associated dendritic cells has been shown to promote models of ovarian cancer progression by blunting anti-tumor immunity (20). Given the recent advances in immunotherapy for cancer, including checkpoint inhibitors, such as anti-programmed death 1 and anti-cytotoxic T-lymphocyte-associated antigen 4, targeting the UPR to increase tumor antigen presentation may be a promising strategy to use in combination with this approach (19).

From just these select examples, it is clear that the UPR plays a large role in immune cell function and tissue inflammation. Given the myriad ways that the immune system both defends against malignancy in some settings and promotes the tumor microenvironment in others (reviewed in Refs. 13, 34, 119), the contributions of the UPR to tumor immunology are currently unknown. This is especially true since most of the studies to date examining the UPR in preclinical models of cancer have been done using immunocompromised mice.

UPR Promotes Tumor Angiogenesis

One of the greatest challenges to the continued growth of a solid tumor is the ability to provide sufficient oxygen, glucose, and other nutrients to cells at its core. It is now well known that hypoxia and glucose deprivation stimulate neo-angiogenesis and cooption of nearby vessels through the upregulation of vascular endothelial growth factor (VEGF), fibroblast growth factor 2, and other pro-angiogenic signals (47, 118). There is strong evidence that the UPR impacts the transcriptional and posttranslational regulation of several pro-angiogenic factors (7, 80). All three arms of the UPR have been reported to strongly upregulate VEGF-A in response to ER stress (29), which protects rapidly growing tumor cells from hypoxia and promotes their survival (75, 79, 85, 86). The downstream mechanism by which the UPR stimulates angiogenesis is in part through direct binding of ATF4 to the VEGF-A promoter (117), as well as XBP1 transcriptionally upregulating VEGF-A and IL-6 expression, possibly through a hypoxia-inducible factor-1α (HIF-1α)-dependent mechanism (3, 15). The ER chaperone Bip/GRP78 also promotes cancer growth and chemoresistance through upregulating tumor angiogenesis (25, 55, 59, 112). Interestingly, VEGF has recently been reported to activate the UPR in endothelial cells in the absence of ER stress through a noncanonical mechanism involving phospholipase C and mTOR signaling (49). In such a way, VEGF seems able to induce its own expression through triggering the UPR in endothelial cells, a feedforward loop with the potential to strongly support angiogenesis (111).

The UPR and Cancer Metastasis

Recently, there have been a few studies suggesting that the UPR may play an important role in cancer cell metastasis, at least in preclinical models. The epithelial-to-mesenchymal transition that tumor cells often undergo before metastasis involves acquisition of a highly secretory phenotype and increased UPR signaling (26, 97). In breast cancer cells lines, PERK inhibition strongly diminished metastatic capacity, as assessed by lung tumor burden after tail-vein injection into an immunocompromised mouse (26). Through a mechanism involving activation of lysosome-associated membrane protein 3, ATF4 has also been reported to be essential for metastasis of breast cancer cells subjected to hypoxia (71, 72). In contrast, glioma tumor cells expressing a dominant-negative mutant of IRE1α showed reduced growth and angiogenesis in vivo, but increased invasiveness (3, 46). Hence, while loss of IRE1α signaling may halt primary tumor progression, it has the potential to induce a more invasive phenotype. In this model, IRE1α controls glioma cell migration in part through RIDD-mediated decay of the mRNA for SPARC (secreted protein acidic and rich in cysteine)/osteonectin, a matrix-associated protein that regulates the interaction between glioma cells and the extracellular matrix. When IRE1α’s RNase is inhibited, SPARC expression increases and promotes tumor cell invasion (23). The proinvasive phenotype that glioma cells acquire in response to IRE1α inhibition is reminiscent of that of some cancer types to direct angiogenesis inhibitors (such as those targeting VEGF) (93). However, we are only beginning to understand how the UPR impacts the ability of tumor cells to invade, metastasize, and survive in distant sites.

Interplay between UPR and Other Tumor-associated Signaling Pathways

In addition to its unique signaling outputs, the UPR communicates in complicated and poorly understood ways with many other signaling pathways that impact tumor development. The NF-κB and HIF-1α pathways are both signaling pathways well known to promote tumorigenesis in some settings that can be regulated by the UPR. For example, PERK-mediated inhibition of Cap-dependent protein translation leads to a relative reduction in IκB, promoting the translocation of NF-κB to the nucleus and transcription of its target genes (48). Through its association with TRAF2, IRE1α activation leads to IκB phosphorylation and degradation, which again results in NF-κB nuclear translocation and transcriptional activation (43). In a study involving triple negative breast cancer, IRE1α’s adaptive output XBP1 was shown to form a transcriptional complex with HIF-1α and regulate expression of HIF-1α targets via recruitment of RNA polymerase II (15). These few examples illustrate the intricate cross talk between the UPR and other signaling pathways that determine the fate of any particular tumor cell, and the difficulties in trying to ascribe a specific role of the UPR in cancer generally.

The UPR in Myeloma: It’s Complicated

To illustrate the complexity of UPR signaling in the setting of cancer, let us examine how our thinking about it has evolved over the years in one tumor type. The UPR has been most comprehensively studied as a potential target in myeloma (14), a highly secretory tumor composed of malignant plasma cells, whose development requires this pathway. In mice, IRE1α and its homeostatic target XBP1 are both required for the differentiation of B-lymphocytes into plasma cells (69, 70), illustrating a critical role for the UPR in the health of this secretory cell type. Interestingly, up to 50% of primary myelomas show unusually high levels of XBP1s (11). This led to the exciting hypothesis that, while IRE1α/XBP1 signaling is critical for normal plasma cell differentiation, constitutive signaling through this pathway might promote inappropriate plasma cell survival and permit the accumulation of additional myeloma causing mutations. In support of this notion, mice expressing a transgene of Xbp1s (which is missing the 26nt intron and hence requires no further processing by IRE1α) in B-lymphocytes develop a plasma cell malignancy closely resembling myeloma (11). There is also evidence to suggest that proteasome inhibition with bortezomib (Velcade), which is Food and Drug Administration approved as first-line therapy for myeloma, leads to myeloma cell death in part by preventing disposal of misfolded proteins through the ERAD pathway and thus triggering ER stress-induced apoptosis (52, 74). Myeloma cells that were knocked down for XBP1 reportedly showed enhanced sensitivity to bortezomib and 17-N-allylamino-17-demethoxygeldanamycin, (68), suggesting a critical role for XBP1 in myeloma cell growth. In 253 patients with newly diagnosed myeloma, those with low XBP1 spliced-to-unspliced ratios had a longer overall survival than those with high XBP1 spliced-to-unspliced ratios (56 mo vs. 40 mo; P = 0.03) (4). Finally, early pharmacological inhibitors of IRE1α were reported to have antitumor activity against myeloma cell lines injected into mice (68, 78); however, the specificity and off targets effects of these agents are not yet well understood (see Drugging the UPR below for more details). Together, these findings led to speculation not only that the homeostatic branch of IRE1α is frequently activated in tumors, but that it may be necessary for the survival and/or growth of the myeloma.

However, recent studies have clouded this interpretation and suggested that the role of IRE1α and XBP1 in myeloma is more complicated than originally anticipated. Leung-Hagesteijn et al. conducted kinome- and genome-scale small interfering RNA studies in the KMS11 myeloma cell line treated with bortezomib (58). In this study, IRE1α was the kinase whose knockdown led to the greatest degree of bortezomib resistance and ranked in the top 1% of all genes required for bortezomib-induced death. In confirmatory studies, knockdown of IRE1α or its homeostatic target XBP1 was surprisingly well tolerated and led to bortezomib resistance in all six myeloma lines tested. When examined microscopically, the myeloma cells deficient for IRE1α or XBP1 were much smaller and resembled pre-plasmablasts. In tumor samples from patients with myeloma, this group found that Xbp1-deficient and pre-plasmablasts were resistant to bortezomib. Furthermore, these findings are consistent with recent genetic evidence from human myeloma samples. Whole exome sequencing of 20 myeloma tumors found two mutations in XBP1 (12). In contrast to the original notion that high XBP1 promotes myeloma, both of these XBP1 mutations turn out to be inactivating and present in treatment-refractory tumors (41, 58). Together, these findings suggest that reduced IRE1α/XBP1 signaling leads to a block in plasma cell maturation and resistance to proteasome inhibitors and calls into the question the potential benefits of using IRE1α inhibitors in this disease, at least in combination with proteasome inhibition.

Drugging the UPR

Given the evidence of UPR deregulation in cancer, there is a great deal of interest in the possibility of pharmacologically modulating its outputs as a strategy to control tumor growth. As the most upstream regulatory components of the UPR, the ER stress sensors IRE1α, PERK, and ATF6 are attractive candidates through which to control this pathway (62). In particular, the enzyme active sites of IRE1α and PERK are the most obvious target for the development of small molecule modulators of the UPR. A number of small molecules have been identified that directly bind and inhibit the RNase of IRE1α (18, 68, 78, 113). Most of the direct inhibitors identified to date contain a reactive electrophile that covalently binds IRE1α’s RNase active site, with the most successful pharmacophore being a salicylaldehyde. Inhibitors of this class include SFT-083010, MKC-3946, and 4μ8c. Several studies have found that these salicylaldehyde-based inhibitors form a Schiff based with a lysine (K907) in the RNase active site (18, 89). While these compounds inhibit IRE1α’s RNase activity in cell culture and several have shown some anti-myeloma activity in xenograft models, there are serious concerns about the selectivity of these compounds, given their likely reactivity against many intracellular proteins.

To develop more specific IRE1α inhibitors, our team recently developed first-in-class ATP-competitive IRE1α kinase inhibiting RNase attenuators (KIRAs) that bind into the kinase domain and allosterically inhibit IRE1α’s RNase (115). We later identified second-generation KIRA6 as an advanced KIRA capable of dose-dependently inhibiting endogenous IRE1α’s kinase activity, oligomerization, ER-localized mRNA decay, Xbp1 mRNA cleavage in vivo. When administered systemically, KIRA6 protect rodents against a form of diabetes caused by mutations in proinsulin that cripple its folding (Akita diabetes), and, when administered intravitreally, KIRA6 protects against ER stress-induced retinal degeneration (30). A group at Amgen subsequently published a series of potent and selective IRE1α kinase inhibitors (38), but these were not used in vivo.

Over the past several years, there have been efforts to find compounds that inhibit the ability of PERK to phosphorylate eIF2α. A team at GlaxoSmithKline (GSK) used structure-based design to develop a series of highly potent and selective inhibitors of PERK’s kinase domain, including GSK2606414 and GSK2656157 (2). Oral administration of GSK2606414 leads to reasonable central nervous system penetration and protects against preclinical models of neurodegeneration (70). Moreover, GSK2656157 administration showed anti-tumor effects in xenograft models of pancreatic adenocarcinoma and multiple myeloma in immune-compromised mice (2). However, much more work needs to be done regarding the potential benefits and risks of inhibiting PERK in vivo, because most of the inhibitors tested to date quickly cause pancreatic β-cell loss and diabetes. The β-cell toxicity observed phenocopies Wolcott-Rallison syndrome, a rare human diabetic syndrome caused by PERK mutations.

As an alternative strategy, small molecules have been found that attenuate eIF2α phosphorylation without inhibiting PERK per se. Recently, a symmetric bis-glycolamide named ISRIB (integrated stress response inhibitor) has been described that binds to and activates the guanine nucleotide exchange factor eIF2B (elongation initiation factor 2B), which in turn releases inhibition of protein translation caused by eIF2α phosphorylation (92, 101–103). ISRIB has been described to enhance memory formation in rodents through mechanisms that are still not completely understood (101). However, there are no reports yet of ISRIB being tested in models of cancer.

While by no means exhaustive, these examples of inhibitors against IRE1α and PERK illustrate emerging concepts about the therapeutic potential (and risks) of targeting the UPR to control cell fate under conditions of ER stress.

Concluding Remarks

The UPR is a highly conserved signal transduction pathway activated when cells are unable to keep up with the protein-folding demands on the ER. In response to the accumulation of misfolded proteins in the ER lumen, a condition called “ER stress,” the UPR initiates adaptive outputs that decrease the load and increase the capacity of the ER secretory pathway in an effort to restore homeostasis. However, if these corrective measures are not sufficient to put the protein-folding status of the ER back into balance, the UPR sends out proinflammatory and prodeath signals to cause cell demise. The loss of ER proteostasis is a common occurrence in many cancer types due to hypoxia, nutrient deprivation, proteasome dysfunction, sustained demands on the secretory pathway, and somatic mutations in its client proteins. Hence, UPR hyperactivation has been documented in numerous solid and hematopoietic malignancies, but whether this ultimately inhibits or promotes tumor growth in patients remains an area of intense debate. There has been mounting evidence over the past few years that the UPR supports tumor growth from xenograft studies in mice in which genetically deleting one or more branches of the UPR or altering the expression of ER chaperones influences the in vivo growth of tumor cells. Depending on the tumor model, the UPR has been implicated in regulating cell survival, angiogenesis, inflammation, antigen presentation, invasion, and metastasis. Taken together, these results argue that the UPR has strong promise as a therapeutic target in cancer, but much more preclinical studies are needed before we understand the benefits and risks of drugging the UPR in any particular tumor type. Fortunately, the development of better small molecule inhibitors of upstream UPR components has finally made these experiments feasible. In particular, the selective IRE1α and PERK inhibitors now available represent powerful tools for probing the role of the UPR in cancers and promising starting points for the development of therapeutics against this pathway.

GRANTS

This work was supported by an American Cancer Society Research Scholar Award, American Association for Cancer Research-Caring for Carcinoid Foundation Award, Harrington Discovery Institute Scholar-Innovator Award, and the National Institutes of Health Grants R01CA136577, R01DK095306, and U01DK108332.

DISCLOSURES

S.A. Oakies is a founder, equity holder, and consultant for OptiKIRA, LLC (Cleveland, OH).

AUTHOR CONTRIBUTIONS

S.A.O. prepared figures, drafted manuscript, edited and revised manuscript, approved final version of manuscript.