Endoplasmic reticulum stress responses in anticancer immunity

Abstract

The endoplasmic reticulum (ER) has a central role in processes essential for mounting effective and durable antitumour immunity; this includes regulating protein synthesis, folding, modification and trafficking in immune cells. However, the tumour microenvironment imposes hostile conditions that disrupt ER homeostasis in both malignant and infiltrating immune cells, leading to chronic activation of the unfolded protein response (UPR). Dysregulated ER stress responses have emerged as critical modulators of cancer progression and immune escape, influencing the initiation, development and maintenance of antitumour immunity. In this Review, we examine how tumour-induced ER stress reshapes the functional landscape of immune cells within the tumour microenvironment. We highlight recent discoveries demonstrating how ER stress curtails endogenous antitumour immunity and reduces the efficacy of immunotherapies. Furthermore, we underscore novel therapeutic strategies targeting ER stress sensors or UPR components to restore immune function and enhance cancer immunotherapy outcomes. Together, this provides a comprehensive overview of the interplay between ER stress responses and antitumour immunity, emphasizing the potential of UPR-targeted interventions to improve immune control of cancer.

Introduction

The endoplasmic reticulum (ER) has a vital role in diverse cellular processes, including lipid metabolism, calcium storage and the synthesis, folding and modification of transmembrane and secreted proteins. When the ER becomes ‘stressed’ due to the accumulation of misfolded or unfolded proteins, cells initiate the unfolded protein response (UPR), a coordinated signalling pathway aimed at restoring ER balance¹. The UPR, also known as the ER stress response, is governed by three sensors of ER stress: inositol-requiring enzyme 1α (IRE1α), protein kinase RNA-like ER kinase (PERK) and activating transcription factor 6 (ATF6). Signalling through these sensors leads to transcriptional, translational and post-translational reprogramming that helps re-establish the ER functionality¹ (Box 1).

Box 1 |. Overview of the UPR.

Protein synthesis, folding, and post-translational modification within the endoplasmic reticulum (ER) are tightly regulated processes that are essential for maintaining the cellular proteome and ensuring protein homeostasis (proteostasis)¹. However, several intrinsic and extrinsic factors and/or conditions can compromise proteostasis within this organelle, leading to ER stress. To counteract unfavourable cellular stress, cells activate the unfolded protein response (UPR), an evolutionarily conserved adaptive mechanism that restores optimal protein folding and trafficking within the ER lumen¹. In mammals, the UPR has evolved into a more intricate network of signalling pathways, with its three primary branches mediated by the transmembrane proteins inositol-requiring enzyme 1α (IRE1α), protein kinase RNA-like ER kinase (PERK) and activating transcription factor 6 (ATF6)¹. Each of these proteins features an ER luminal domain that allows them to detect a range of cellular stress signals.

Under normal conditions, the ER stress sensors IRE1α, PERK and ATF6 remain inactive due to their association with the ER chaperone binding-immunoglobulin protein (BiP, encoded by GPR78 and also called GRP78 or HSPA5) on their luminal side, which suppresses their activity. However, when the ER is under stress and its protein-folding capacity is overwhelmed, misfolded proteins accumulate, leading to the release of BiP from the UPR sensors. This dissociation triggers the activation of downstream signalling pathways of the UPR¹.

IRE1α

IRE1α is a central sensor of UPR, activated upon the accumulation of misfolded proteins in the ER. This activation triggers its dimerization and subsequent trans-autophosphorylation, leading to a conformational change that activates its RNase domain. The kinase activity of IRE1α not only enables its RNase function, but also engages downstream signalling pathways, including activation of the Jun N-terminal kinase (JNK) pathway and the nuclear factor-κB (NF-κB) pathway via interaction with tumour necrosis factor receptor-associated factor 2 (TRAF2). Activation of NF-κB promotes the expression of pro-inflammatory and survival genes, thereby linking ER stress to inflammation and cell fate decisions¹. Concurrently, the endoribonuclease activity of IRE1α mediates the unconventional splicing of X-box binding protein 1 (XBP1) mRNA, generating the active XBP1s transcription factor. XBP1s drives the expression of genes encoding ER chaperones, components of the ER-associated degradation (ERAD) system and lipid biosynthetic enzymes, collectively enhancing the capacity of the ER to manage protein folding and restore proteostasis¹. Additionally, IRE1α degrades selective mRNAs through regulated IRE1α-dependent decay (RIDD), a mechanism that can alleviate ER stress by reducing the translational burden but may also promote cell death under excessive stress conditions. Moreover, ER stress and IRE1α signalling intersect with metabolic pathways such as the hexosamine biosynthetic pathway (HBP), which influences protein glycosylation and broader cellular adaptation mechanisms. The interplay between these kinase and endoribonuclease functions enables IRE1α to orchestrate adaptive or apoptotic responses based on the severity and duration of ER stress.

PERK

PERK activation occurs in response to ER stress, leading to its dimerization and autophosphorylation. This activation enables PERK to phosphorylate several targets, including the eukaryotic initiation factor 2α (eIF2α), a key regulatory event that attenuates global protein synthesis, thereby reducing the influx of newly synthesized proteins into the ER and alleviating protein-folding stress. However, eIF2α phosphorylation also selectively enhances the translation of specific stress-responsive mRNAs, notably that of ATF4¹.

ATF4 functions as a central regulator of the integrated stress response, driving the expression of genes involved in amino acid metabolism, oxidative stress defence, autophagy and apoptosis. Among its critical downstream targets, ATF4 upregulates C/EBP homologous protein (CHOP), a key mediator of ER stress-induced apoptosis. CHOP promotes cell death when ER stress is severe or unresolved, in part by dysregulating cellular redox homeostasis and suppressing anti-apoptotic BCL-2 family proteins. Through these mechanisms, PERK signalling integrates adaptive and pro-apoptotic responses to maintain ER homeostasis or eliminate terminally stressed cells¹.

ATF6

ATF6 is a transmembrane ER stress sensor that undergoes regulated intramembrane proteolysis upon ER stress. In response to misfolded protein accumulation, ATF6 translocates to the Golgi apparatus, where it is sequentially cleaved by site-1 protease (S1P) and site-2 protease (S2P)¹. This processing releases its cytosolic, transcriptionally active form (ATF6p50), which then translocates to the nucleus. There, ATF6p50 drives the expression of genes that enhance ER protein-folding capacity, promote ERAD and support overall proteostasis¹. These target genes encode molecular chaperones, such as BiP, and components of the ERAD machinery, facilitating the clearance of misfolded proteins and maintaining ER homeostasis².

Harsh conditions and factors enriched in the tumour microenvironment (TME), such as hypoxia, nutrient restriction and accumulation of reactive oxygen species (ROS), can compromise the protein-folding capacity of the ER, leading to ER stress and sustained activation of the UPR in diverse tumour-resident cell types². Malignant cells often exploit this process to activate cytoprotective UPR pathways, allowing adaptation to metabolic and oxidative insults characteristic of the TME². For instance, UPR activation leads to increased glycolysis and lactate production, allowing cancer cells to thrive under hypoxia³. The UPR was also shown to enhance antioxidant protein expression, protecting cancer cells from oxidative stress-induced damage⁴. Hence, cancer cell-intrinsic ER stress responses have been shown to promote tumour development, progression, metastasis and treatment resistance^2,5,6. In addition, the TME perturbs ER homeostasis in infiltrating leukocytes, altering their metabolism, differentiation, effector capacity and antitumour activity, which ultimately facilitates immune evasion, resistance to immunotherapy and disease progression^2,7.

Despite the success of immunotherapy in some cancer types, resistance continues to pose a major challenge to achieving consistent and durable therapeutic responses^8,9. Emerging evidence indicates that the UPR contributes to this resistance by suppressing antigen presentation, impairing T cell activation, promoting T cell dysfunction and exhaustion, and establishing an immunosuppressive TME^10–14. New therapeutic strategies targeting the overactivation of ER stress sensors, particularly IRE1α¹⁵ and PERK¹⁶, have shown promise in preclinical cancer models by enhancing endogenous antitumour immunity and improving the efficacy of immune checkpoint blockade (ICB) and T cell-based immunotherapies. Of note, therapeutic targeting of these sensors has progressed to early clinical trials in patients with advanced solid tumours, including renal cell carcinoma, gastric cancer, metastatic breast cancer and small cell lung cancer, as well as relapsed refractory metastatic breast cancer, as an advanced line of treatment for recurrent malignancy (NCT03950570 and NCT04834778)^17–19. These recent developments suggest that modulating ER stress signalling or the UPR elements could open new therapeutic avenues to enhance antitumour immunity, overcome resistance to existing cancer therapies and potentially synergize with current immunotherapies.

The fundamental biology of an adaptive UPR has been recently and comprehensively reviewed²⁰. Hence, in this Review, we discuss the emerging drivers of ER stress in intratumoural immune cells. We also explore the evolving mechanisms by which dysregulated ER stress responses shape anticancer immunity and highlight novel UPR-targeted therapeutic strategies designed to enhance the effectiveness of cancer immunotherapy.

Disruption of ER homeostasis by the TME

The nutritional and metabolic composition of the TME is distinct from those of normal tissues, characterized by hypoxia, nutrient deprivation, acidosis and oxidative stress — all of which can compromise ER protein-folding capacity and cause relentless UPR activation. As the canonical factors that induce ER stress within the TME have been reviewed before^2,5, this section focuses on recently identified drivers of ER stress in solid malignancies that promote immune cell dysfunction.

Taurine restriction

Taurine is often referred to as a ‘semi-essential’ or ‘conditionally essential’ amino acid; however, it is technically a sulfonic acid, rather than a true amino acid, as it lacks a carboxyl group²¹. Unlike typical amino acids, taurine is not incorporated into proteins but instead exists circulating throughout various tissues and bodily fluids. This biodistribution allows taurine to access critical anatomical locations such as the heart, brain, retina and immune system²². Taurine has critical roles in antioxidant defences, calcium homeostasis, osmoregulation and immune modulation, making it crucial for diverse physiological processes²³. Taurine scavenges ROS through its role as an antioxidant. In this context, it interacts with hydrogen peroxide and hypochlorous acid, converting them into less harmful substances such as taurine chloramine²⁴. This conversion helps alleviate oxidative stress and the potential activation of the UPR by this process²⁵. Additionally, taurine stabilizes intracellular calcium levels by regulating the activity of calcium channels and transporters. This calcium homeostasis is key for the function of ER-resident protein chaperones that rely on consistent calcium levels to properly fold proteins and prevent the accumulation of misfolded proteins^26,27. Consequently, taurine has a pivotal role in supporting ER proteostasis, ensuring an optimal environment for protein-folding integrity and reducing the risk of ER stress and persistent UPR activation.

Taurine is also essential for the conjugation of primary bile acids, a crucial step in their synthesis. This conjugation with taurine transforms primary bile acids into their more soluble and less toxic forms, facilitating their function in fat digestion and absorption²⁸. Primary bile acids are synthesized in the liver from cholesterol and are subsequently conjugated with taurine or glycine, which impacts their properties and biological functions²⁹. Indeed, taurine-conjugated bile acids are more hydrophobic than the glycine-conjugated ones²⁹. This difference in hydrophobicity influences how bile acids are absorbed and interact with receptors in the gut²⁹. Microbial enzymes in the gut convert these conjugated bile acids into secondary forms, further diversifying their roles and effects in the body³⁰.

Unconjugated or primary bile acids, such as cholic acid and chenodeoxycholic acid, and secondary bile acids, such as deoxycholic acid and lithocholic acid, can cause ER stress and induce the UPR in cancer cells by disrupting the ER membrane and increasing the accumulation of misfolded proteins in this organelle³¹. However, when conjugated with taurine, primary bile acids lose their ER stress-inducing activity, attenuating UPR activation and decreasing pro-survival adaptations of cancer cells within the TME³¹ (Fig. 1a). Bile acids also modulate ER stress response pathways in tumour-infiltrating CD8⁺ T cells, shaping their function within the TME. In hepatocellular carcinoma, primary bile acids were found to induce oxidative stress, whereas the secondary bile acid lithocholic acid triggered ER stress, impairing T cell function and promoting their exhaustion³². Notably, genetic inhibition of bile acid conjugation in hepatocytes through deletion of the bile acid-CoA amino acid N-acyltransferase enhanced tumour-specific T cell responses and improved responses to anti-programmed cell death protein 1 (PD1) therapy³². Additionally, taurine-conjugated bile acids and ursodeoxycholic acid mitigated detrimental ER stress responses in T cells, preserving their function³². These findings highlight bile acid modulation, particularly through taurine conjugation, as a potential strategy to enhance antitumour immunity.

Fig. 1 |. Metabolic stressors disturbing ER homeostasis in the TME.

a, In cancer cells, unconjugated bile acids induce endoplasmic reticulum (ER) stress by disrupting ER homeostasis, whereas taurine-conjugated bile acids mitigate the unfolded protein response (UPR) pathway activation. Taurine restriction intensifies ER stress, activating the protein kinase RNA-like ER kinase (PERK) pathway and promoting apoptosis in cancer cells. In CD8⁺ T cells, unconjugated bile acids, particularly the secondary bile acid lithocholic acid, induces ER stress, which promotes T cell exhaustion and dysfunction in advanced hepatocellular carcinoma. When taurine is depleted from the microenvironment due to cancer cell uptake, CD8⁺ T cells exhibit exhaustion via PERK–activating transcription factor 4 (ATF4) signalling, impairing cytokine production and upregulating immune checkpoints. Question marks indicate that the mechanism remains unknown. b, Lipid imbalance in the tumour microenvironment (TME), including cholesterol and glucosylceramide accumulation, disrupts ER homeostasis. Cholesterol accumulation activates inositol-requiring enzyme 1α (IRE1α)–XBP1s, impairing CD8⁺ T cell cytotoxicity. Glucosylceramide induces lipid saturation in tumour-associated macrophages, activating detrimental IRE1α–XBP1s signalling that promotes T cell suppression. c, Carbon monoxide (CO) is often elevated in the TME due to upregulation of haem oxygenase-1, which is induced by oxidative stress and chronic UPR activation. While controlled CO administration enhances mitochondrial–ER communication (the dashed line is shown here to indicate indirect interaction), improving CD8⁺ T cell metabolic resilience and antitumour activity, high levels of CO drive immunosuppressive phenotypes. ARG1, arginase 1; BiP, binding-immunoglobulin protein (also known as GPR78); CHOP, C/EBP homologous protein; CTLA4, cytotoxic T lymphocyte protein 4; EGFR, epidermal growth factor receptor; eIF2α; eukaryotic initiation factor 2α; LAG3, lymphocyte activation gene 3; LPCAT3, lysophosphatidylcholine acyltransferase 3; PD1, programmed cell death protein 1; ROS, reactive oxygen species; TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIM3, T cell immunoglobulin and mucin-domain-containing-3.

Additional studies emphasize the importance of taurine in supporting the protective function of tumour-infiltrating CD8⁺ T cells^33–35. Hepatocellular carcinomas and pancreatic adenocarcinomas were found to exhibit high expression of the taurine transporter SLC6A6, enabling malignant cells to sequester taurine from the TME³³. This process restricts taurine availability to intratumoural CD8⁺ T cells, pushing them towards a dysfunctional state characterized by defective production of key antitumour cytokines such as interferon-γ (IFNγ) and tumour necrosis factor (TNF)³³. Mechanistically, taurine deprivation induced ER stress in CD8⁺ T cells, activating the transcription factor ATF4 through the PERK–Janus kinase 1 (JAK1)–signal transducer and activator of transcription 3 (STAT3) axis³³. ATF4 activation in this context caused upregulation of immune checkpoint molecules PD1, cytotoxic T lymphocyte protein 4 (CTLA4), T cell immunoreceptor with Ig and ITIM domains (TIGIT) and T cell immunoglobulin and mucin-domain-containing-3 (TIM3), promoting T cell exhaustion and defective cytotoxic function³³ (Fig. 1a).

A recent study in patients with lung cancer undergoing PD1 blockade revealed that elevated serum taurine levels were associated with a greater likelihood of response to immunotherapy³⁴. This correlation was attributed to the ability of taurine to enhance CD8⁺ T cell survival, proliferation and cytotoxic function, leading to increased secretion of IFNγ and TNF³⁴. Mechanistically, taurine was found to facilitate calcium signalling, oxidative phosphorylation and T cell receptor signalling, thus sustaining T cell activity under ER stress³⁴. By stimulating phospholipase C γ1 (PLCγ1)-mediated calcium and MAPK signalling pathways, taurine promoted metabolic and signalling adaptations necessary for T cell proliferation and function within the TME³⁴. Strikingly, combining taurine supplementation with ICB enhanced CD8⁺ T cell function and resulted in marked inhibition of B16 melanoma and 4T1 breast cancer growth and in mouse models³⁴. These findings collectively suggest that taurine not only supports T cell immunity but could also be harnessed to overcome ER stress-driven immune cell dysfunction in the TME, offering a novel strategy to enhance the effectiveness of cancer immunotherapy.

Lipid imbalance

Lipid dysregulation within the TME has a central role in ER stress induction. Although cholesterol is vital for cellular membrane integrity³⁶, the accumulation of intracellular cholesterol within the ER membrane can alter its fluidity and affect ER-resident enzymes and protein chaperones, causing ER stress and UPR activation³⁷. In preclinical studies, the 3-hydroxy-3-methylglutaryl (HMG) coenzyme A reductase inhibitor statin, which lowers extracellular cholesterol levels, was found to attenuate ER stress and enhance sensitivity to chemotherapeutic agents in CT26 mouse colon carcinoma cells³⁸. This effect was particularly strong in the context of KRAS mutant cancers, where statin-mediated inhibition of RAS prenylation causes ER stress, enhancing the immunogenicity of the cancer cells and sensitizing them to chemotherapy in syngeneic colorectal cancer and genetically engineered lung and pancreatic mouse models³⁸. However, persistent ER stress due to sustained cholesterol accumulation can overwhelm the UPR in cancer cells, shifting the response from adaptive to pro-apoptotic³⁹. Elevated cholesterol production in the melanoma TME was found to activate the transcription factor XBP1s in infiltrating CD8⁺ T cells, causing immune checkpoint upregulation and decreased effector capacity¹³ (Fig. 1b). Silencing Xbp1 expression in T cells or reducing systemic cholesterol levels was shown to restore CD8⁺ T cell antitumour function and delay melanoma progression in mouse models, highlighting the potential of cholesterol-lowering or ER stress-mitigating strategies to enhance T cell-based immunotherapy¹³.

Alterations in glucosylceramide levels in the TME can also provoke detrimental ER stress responses in infiltrating immune cells (Fig. 1b). Glucosylceramide, a sphingolipid metabolite enriched in the TME, is primarily derived from cancer cells and its accumulation is driven by dysregulated lipid metabolism and increased activity of glucosylceramide synthase⁴⁰. Elevated glucosylceramide levels in the TME result from heightened glycolipid biosynthesis in tumour cells, a process often exacerbated by hypoxia and oncogenic signalling pathways⁴¹. Once released, glucosylceramide is taken up by tumour-associated macrophages (TAMs), where it remodels the ER membrane by increasing lipid saturation⁴². This alteration triggers an unconventional ER stress response leading to sustained activation of the IRE1α–XBP1s pathway, which enhances the tumour-promoting attributes of TAMs⁴². ER lipid remodelling in TAMs is further amplified by the upregulation of lysophosphatidylcholine acyltransferase 3 (LPCAT3), a key enzyme that regulates ER lipid saturation (Fig. 1b). LPCAT3-mediated remodelling intensifies ER stress, reinforcing IRE1α–XBP1s signalling and sustaining TAM immunosuppressive functions⁴². Notably, this lipid-saturated ER environment establishes a self-perpetuating loop that maintains TAM-mediated immunosuppression⁴². Additionally, concurrent STAT3 activation cooperates with IRE1α–XBP1s signalling, amplifying pro-tumoural transcriptional programmes and enhancing TAM immunosuppressive properties⁴². This signalling convergence promotes TAM survival and induces the expression of arginase 1 (ARG1) (Fig. 1b), an enzyme that catabolizes arginine into urea and ornithine⁴³. The resulting depletion of arginine, an essential nutrient for T cell proliferation and function, fosters immune evasion and supports tumour progression⁴².

CO

Carbon monoxide (CO), a by-product of haem degradation catalysed by haem oxygenase-1 (HO-1), is an intriguing regulator of immune cell function with paradoxical effects. Endogenously produced CO has key physiological roles, including vasodilation, anti-inflammatory signalling and cytoprotection by modulating oxidative stress and apoptosis through the regulation of guanylate cyclase and mitochondrial function⁴⁴. In the TME, CO levels are often elevated due to HO-1 upregulation, which is induced by oxidative stress via the activation of NRF2, a transcription factor that binds to the HO-1 promoter in response to ROS and hypoxia-induced signalling pathways⁴⁵. High CO levels disrupt mitochondrial respiration by binding to cytochrome c oxidase, leading to an accumulation of ROS, protein misfolding and robust activation of the UPR⁴⁶. This has been shown to facilitate tumour progression by promoting immunosuppression in the TME⁴⁷ (Fig. 1c). However, the evidence is correlative, and the precise mechanistic pathways linking CO-driven ER stress to immunosuppressive programming in specific leukocyte subsets remain incompletely defined. Further research is needed to elucidate how CO modulates ER stress responses and immune functionality across different immune populations in the TME, which could provide valuable insights into its potential as a therapeutic target in cancer immunotherapy.

By contrast, a recent study showed that controlled doses of CO can enhance ER–mitochondria communication in tumour-infiltrating T cells⁴⁸. Mechanistically, CO exposure transiently activated PERK, inducing a temporary and adaptive pause in protein synthesis. This allowed T cells to clear damaged mitochondria, restore mitochondrial integrity and ultimately enhance their antitumour function⁴⁸ (Fig. 1c). Conversely, in cancer cells, CO exposure induced metabolic stress and mitochondrial dysfunction, leading to apoptosis, suggesting a differential effect based on cell type and metabolic state⁴⁹.

Together, these studies suggest that whereas endogenous CO generation due to HO-1 dysregulation induces detrimental ER stress responses that promote immunosuppression, controlled exogenous CO administration can trigger an adaptive UPR that enhances T cell function in the TME. The divergent effects of CO underscore the need for further research to clarify its cell type-specific and dose-dependent regulatory functions. Understanding the balance between its immunosuppressive and immune-enhancing properties will be essential for leveraging its therapeutic potential, particularly in combination with immunotherapies targeting UPR signalling.

Immunoregulatory effects of ER stress in the cancer cell

The UPR influences cancer cell survival and fate in a manner dependent on the intensity and duration of signalling^50–52. Beyond its direct effects on tumour cells, recent studies have demonstrated that cancer cell-intrinsic ER stress responses can also modulate the function and behaviour of neighbouring immune cells within the TME, promoting immune evasion and tumour progression⁵³. Notably, chronic ER stress in cancer cells has been shown to impact antigen presentation, primarily by impairing the expression of major histocompatibility complex (MHC) class I molecules¹⁰. This occurs because ER stress-inducing conditions lead to a global reduction in protein translation, limiting the synthesis of key antigen-presenting machinery⁵⁴ (Fig. 2a).

Fig. 2 |. Immunomodulatory effects of ER stress in cancer cells.

The tumour microenvironment (TME), characterized by hypoxia, oxidative stress and nutrient deprivation, compromises the protein-folding capacity of the endoplasmic reticulum (ER), leading to the accumulation of misfolded proteins. This triggers activation of ER stress sensors protein kinase RNA-like ER kinase (PERK), inositol-requiring enzyme 1α (IRE1α) and activating transcription factor (ATF6) in cancer cells, which can modulate antitumour immunity. a, PERK activation phosphorylates eukaryotic initiation factor 2α (eIF2α), suppressing global translation and inhibiting major histocompatibility complex I (MHC-I) expression. The loss of PERK in melanoma cells induces SEC61β-dependent paraptosis, an alternative cell death pathway characterized by vacuolation and damage to the ER and mitochondria, and triggers immunogenic cell death (ICD), promoting long-term antitumour immunity through type I interferon (IFN-I) secretion by dendritic cells. IFN-I production promotes common monocyte precursor (cMoP) recruitment to the tumour and enhances their local differentiation into dendritic cells that prime antitumour CD8⁺ T cells. b, Activation of the IRE1α–XBP1s axis in breast cancer cells promotes expression of interleukin 6 (IL-6), IL-8, C-X-C motif chemokine ligand 1 (CXCL1), granulocyte–macrophage colony-stimulating factor (GM-CSF) and transforming growth factor-β (TGFβ). These cytokines facilitate the accumulation of cancer-associated fibroblasts and myeloid-derived suppressor cells in triple-negative breast cancer (TNBC) models, blunting immune surveillance. Generation of XBP1s by IRE1α also inhibits expression of natural killer (NK) group 2 member D (NKG2D) ligands, such as MHC-I polypeptide-related sequence A (MICA), by suppressing E2F1 that is responsible for MICA expression. Engagement of MICA would induce activating signals in NK cells and trigger NK cell-mediated cytotoxicity that eliminates cancer cells. Activation of IRE1α in melanoma cells has been shown to upregulate programmed cell death 1 ligand 1 (PDL1) expression. Similarly, in TNBC cells, ER stress was shown to enhance PDL1 stability through its interaction with the ER chaperone binding-immunoglobulin protein (BiP), but whether this is specifically mediated by IRE1α remains unclear (indicated with a dashed line). IRE1α–XBP1s also sustains expression of prostaglandin E synthase (Ptges, encoding microsomal prostaglandin E synthase-1 (mPGES1)) in lung cancer cells, enabling prostaglandin E2 (PGE₂)-driven immunosuppression in the TME. Beyond generating XBP1s, IRE1α also cleaves select mRNAs through regulated IRE1α-dependent decay (RIDD). This activity prevents double-stranded RNA (dsRNA) accumulation in docetaxel-treated TNBC cells, thus preventing Z-DNA-binding protein 1 (ZBP1)-mediated NOD-like receptor family, pyrin domain-containing 3 (NLRP3) activation and pyroptosis, a lytic, pro-inflammatory form of programmed cell death triggered by inflammasome activation. Inhibition of the IRE1α RNase domain enables pyroptosis, enhances CD8⁺ T cell infiltration, and sensitizes TNBC tumours to ICB and taxane-based chemotherapy. c, ATF6 activation in colon epithelial cells promotes a less structured mucosal layer and is therefore more permeable to bacterial penetration. This leads to microbiota dysbiosis, innate immune infiltration and inflammation, facilitating tumorigenesis. In mice expressing a constitutively active form of ATF6 (nATF6^IEC), this was shown to be mediated by myeloid differentiation primary response 88 (MyD88) and TIR-domain-containing adaptor inducing interferon-β (TRIF)-dependent signal transducer and activator 3 (STAT3) activation. This could be reversed by antibiotics, thus linking ATF6 activation to microbiome-driven immune modulation. Yet the current understanding of ATF6 in tumour-related immune regulation is limited. In mouse B16 melanoma and RAS-driven lung cancer cells, severe ER stress leads to calreticulin translocation from the ER to the cell membrane. The externalized calreticulin (ecto-CRT) is recognized by NK cells and enhances elimination of ER-stressed cancer cells, but the precise ER stress sensor mediating this process is unknown (indicated by a dashed line). HMGB1, high-mobility group box 1; moDC, monocytic-lineage inflammatory dendritic cell; TLR, toll-like receptor.

Although natural killer (NK) cells typically eliminate cells expressing low levels of MHC-I, tumour cells undergoing ER stress exhibit resistance to NK cell-mediated cytotoxicity^10,55. This resistance is mediated, at least in part, by the IRE1α–XBP1s signalling axis, which represses the expression of MHC-I polypeptide-related sequence A (MICA) in human melanoma cell lines (for example, MEL624)⁵⁶ (Fig. 2b). MICA serves as a ligand for NK group 2 member D (NKG2D), a major activating receptor on NK cells that triggers cytotoxic responses upon ligand recognition⁵⁷. By downregulating MICA, ER-stressed tumour cells reduce their susceptibility to NK cell-mediated elimination. Mechanistically, the activation of the IRE1α–XBP1s pathway under ER stress suppresses the expression of E2F1, a key transcription factor required for MICA induction⁵⁶. Although the repression of MICA has been well documented in in vitro cell culture models⁵⁶, it remains to be fully elucidated whether this mechanism operates similarly in vivo. Furthermore, it is important to clarify that whereas MICA and MICB are both ligands for NKG2D, the specific repression mediated by IRE1α–XBP1s in this context has only been demonstrated for MICA.

Nonetheless, recent findings indicate that severe ER stress can also lead to the translocation of calreticulin (CRT) from the ER lumen to the cell membrane, a phenomenon referred to as ectopic-CRT (ecto-CRT)⁵⁸. Ecto-CRT enhances tumour cell recognition by NK cells via the activating receptor NKp46, promoting NK cell-mediated killing in mouse models of B16 melanoma and RAS-driven lung cancer⁵⁸. Notably, ecto-CRT is also a hallmark of chemotherapy-induced immunogenic cell death (ICD)⁵⁹ (Fig. 2c).

Another key immunoregulatory mechanism by which cancer cells evade immune surveillance involves the upregulation of immune checkpoint ligands^60–62. In triple-negative BT-549 breast cancer cells, activation of the UPR has been shown to increase the expression of programmed cell death 1 ligand 1 (PDL1) in a glucose-regulated protein 78 (GRP78)-dependent manner⁶³. Mechanistically, GRP78 directly binds to PDL1 in ER-stressed triple-negative breast cancer (TNBC) cells, stabilizing its expression and thereby enhancing immune evasion⁶³ (Fig. 2b).

Beyond immune checkpoints, dysregulation of the UPR in cancer cells can enhance the expression of immunosuppressive lipid mediators and inflammatory cytokines, thereby shaping the tumour immune contexture and antitumour immune responses. For instance, in non-small-cell lung cancer (NSCLC), IRE1α–XBP1s signalling in tumour cells was found to drive malignancy and immunosuppression by sustaining prostaglandin E2 (PGE₂) production in the tumour⁶⁴. Using a novel computational pipeline to specifically quantify the percentage of the spliced XBP1 mRNA isoform (XBP1s) relative to total XBP1 from transcriptomic data available in The Cancer Genome Atlas database, the authors found that patients with low XBP1s levels had improved overall survival compared with those with high XBP1s expression⁶⁴. Importantly, an IRE1α-deletion-related, but not a PERK- or ATF6-dependent, gene signature correlated with immune infiltration and survival in patients with NSCLC⁶⁴. IRE1α ablation in tumour cells delayed disease progression and extended survival in mouse models of NSCLC by increasing infiltration by conventional type 1 dendritic cell (cDC1) and cDC2 populations while decreasing the abundance of polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs) in the tumour. Notably, although IRE1α ablation in cancer cells did not alter overall T cell infiltration, T cells residing in IRE1α-deficient lung tumours showed enhanced capacity to produce IFNγ and TNF, and were required for durable anticancer immunity⁶⁴. Mechanistically, IRE1α activation in cancer cells sustained expression of microsomal prostaglandin E synthase-1 (mPGES1), driving PGE₂-mediated immunosuppression in the lung TME⁶⁴. PERK activation was found in pancreatic cancer cells treated with the histone deacetylase inhibitor valproic acid, which led to the production of PGE₂ that compromised dendritic cell (DC) function while increasing the release of pro-tumorigenic interleukin-10 (IL-10) and IL-8 (ref. 65). These findings highlight that cancer cell-intrinsic ER stress responses coordinated by IRE1α or PERK signalling sustain PGE₂-driven immunosuppressive programs that facilitate tumour immune evasion.

Furthermore, ablation of IRE1α in TNBC cells impaired their capacity to produce IL-6, IL-8, C-X-C motif chemokine ligand 1 (CXCL1), granulocyte–macrophage colony-stimulating factor (GM-CSF) and transforming growth factor-β2 (TGFβ2)⁶⁶. Subsequent studies showed that downregulation of these pro-inflammatory factors upon IRE1α inhibition in TNBC cells was associated with decreased frequency of cancer-associated fibroblasts and MDSCs in tumour beds⁶⁷ (Fig. 2b).

Malignant cells undergoing ER stress can further modulate myeloid cell function through a process termed ‘transmissible ER stress’^2,68. Exposure of mouse myeloid cells to secreted factors from prostate cancer, lung cancer or melanoma cells undergoing ER stress was found to activate their UPR. This process was associated with the induction of pro-tumorigenic and immunosuppressive functions^2,68. Specifically, DCs treated with supernatants from ER-stressed cancer cells exhibit upregulation of immunosuppressive mediators such as Arginase 1 and PGE₂, and decreased capacity to cross-present antigens to CD8⁺ T cells⁶⁹. Consequently, DCs conditioned in vitro with these supernatants acquire an immunosuppressive phenotype that, upon adoptive transfer into mice bearing B16.F10 melanoma, promotes tumour growth⁶⁹. Nonetheless, the mechanisms underlying transmissible ER stress are unknown.

Inactivation of ER stress sensors IRE1α or PERK in cancer cells has been shown to induce immunogenic antitumour responses by inducing pyroptosis or paraptosis, respectively. Beyond its XBP1s-inducing activity, the IRE1α RNase domain can cleave select mRNAs through regulated IRE1α-dependent decay (RIDD) and this process was reported to have a critical role in taxane-induced immunogenicity in TNBC⁷⁰. In this study, inhibition of IRE1α RNase activity using ORIN1001 caused accumulation of double-stranded RNA (dsRNA) within docetaxel-treated TNBC, both in genetically engineered mouse models and patient-derived xenograft models such as BCM5998 (ref. 70). IRE1α degraded the dsRNA induced by taxane treatment through RIDD, preventing Z-DNA-binding protein 1 (ZBP1)-mediated NOD-like receptor family, pyrin domain-containing 3 (NLRP3) inflammasome activation and, therefore, pyroptosis in TNBC cells⁷⁰ (Fig. 2b). Selective inhibition of IRE1α using ORIN1001 in p53-deficient 2153L TNBC tumours treated with docetaxel induced CD8⁺ T cell infiltration and activation⁷⁰. Consequently, ORIN1001 plus docetaxel sensitized these tumours to anti-PD1 therapy⁷⁰.

PERK inhibition in cancer cells was found to induce immunogenic cell death and enhance antitumour immunity. Specifically, genetic ablation or pharmaceutical inhibition of this ER stress sensor in melanoma cells induced SEC61β-dependent paraptosis and triggered the release of ATP, high-mobility group box 1 (HMGB1) and ecto-calreticulin (ExoCRT), all hallmarks of ICD⁷¹. In B16 melanoma tumours lacking PERK, flow cytometry analyses revealed an accumulation of common monocyte precursors (cMoPs) and monocytic-lineage inflammatory dendritic cells (MoDCs) within the TME⁷². Mechanistically, PERK deletion led to heightened type I interferon (IFN-I) signalling, which drove the differentiation of cMoPs into MoDCs. RNA-sequencing analysis of precursor cells from PERK-deficient B16 tumours demonstrated a distinct induction of IFN-I response genes and MoDC lineage commitment transcripts compared with wild-type controls⁷². Notably, blockade of the interferon-α/β receptor 1 (IFNAR1) using a neutralizing antibody restored tumour growth and prevented the expansion of cMoPs and MoDCs, confirming the requirement of IFN-I signalling in this process⁷². The IFN-I response facilitated CCR2-dependent trafficking of cMoPs into the tumour, where they locally differentiated into MoDCs that primed antitumour CD8⁺ T cells, ultimately enhancing long-term protective immunity⁷² (Fig. 2a). Intriguingly, in a different study, genetic deletion of the ER oxidoreductase-1α (ERO1α) in cancer cells of mice harbouring Lewis lung carcinoma, B16 or MC38 tumours impaired their IRE1α activation, causing a defective UPR that induced ICD and in turn enhanced the efficacy of PD1 blockade therapy⁷³.

Activation of ATF6 in colon epithelial cells has been implicated in promoting intestinal dysbiosis, which in turn disrupts innate immune homeostasis and contributes to tumorigenesis⁷⁴. Transgenic mice expressing a constitutively active form of ATF6 in the colon epithelium (nATF6^IEC) spontaneously develop adenomas in the large intestine⁷⁴. nATF6^IEC mice show less structured mucus layer and are therefore more permeable to bacteria penetration. Notably, tumour formation in this model was microbiota dependent, as treatment with broad-spectrum antibiotics completely abrogated ATF6-induced tumour initiation and progression⁷⁴. Although the direct effects of ATF6-driven microbiota alterations on immune responses were not explored in this study, the authors proposed that changes in bacterial composition may facilitate innate immune cell infiltration, promote tissue inflammation and drive TIR-domain-containing adaptor inducing interferon-β (TRIF)-dependent STAT3 phosphorylation, ultimately fostering a pro-tumorigenic microenvironment⁷⁴ (Fig. 2c).

Altogether, these findings underscore the role of cancer cell-intrinsic ER stress responses in remodelling the TME. UPR activation in tumour cells promotes immune evasion through multiple mechanisms, including upregulation of immune checkpoint ligands, impairment of antigen presentation, suppression of myeloid cell-mediated antitumour responses and disruption of microbiome homeostasis. Targeting specific UPR components in cancer cells thus presents a promising therapeutic strategy to reverse immune evasion and enhance the efficacy of immunotherapy.

ER stress signalling in intratumoural immune cells

Persistent ER stress responses cause major changes in the transcriptional profiles, metabolic states and functional activities of intratumoural immune cells. In this section, we review how ER stress signalling modulates antitumour immune responses by directly influencing the function of both innate and adaptive immune cells in the TME.

Innate immune cells

Several studies have demonstrated that ER stress and maladaptive UPR activation are commonly induced within various innate immune cell subsets in the TME^2,5,6,75. These dysregulated ER stress responses drive functional and pro-survival reprogramming in these cells, frequently pushing them towards immunosuppressive, pro-tumour phenotypes⁷⁵.

DCs.

Dysfunctional DCs residing in the ovarian cancer microenvironment demonstrated robust expression of ER stress markers and sustained activation of the IRE1α–XBP1s arm of the UPR compared with DCs isolated from non-cancerous tissue⁷⁶. These tumour-infiltrating DCs exhibited high levels of ROS, which promoted intracellular lipid peroxidation and generated by-products, such as 4-hydroxynonenal (4-HNE), that modified ER-resident chaperones and caused ER stress⁷⁶. Treatment with antioxidants or hydrazine derivatives that sequester 4-HNE prevented the induction of ER stress in DCs isolated from ovarian cancer-bearing mice exposed to cell-free ascites in vitro⁷⁶. Overactivation of IRE1α–XBP1s in tumour-infiltrating DCs caused uncontrolled accumulation of triglycerides within the cell and lipid droplet formation, impairing their antigen-presenting capacity⁷⁶ (Fig. 3a). Accordingly, conditional deletion or therapeutic silencing of Xbp1s in DCs enhanced their antigen-presenting function, promoting the activation of antitumour T cell responses that delayed ovarian cancer progression⁷⁶.

Fig. 3 |. ER stress responses in intratumoural immune cells.

a, Reactive oxygen species (ROS)-driven endoplasmic reticulum (ER) stress in tumour-infiltrating dendritic cells (DCs) leads to potent inositol-requiring enzyme 1 α (IRE1α)–XBP1s activation, causing dysregulated lipogenesis and lipid droplet accumulation that impairs antigen presentation to T cells. Abrogating IRE1α–XBP1s in tumour-infiltrating DCs enhances T cell activation and promotes antitumour immunity in ovarian cancer models. BAT3 deficiency in DCs exacerbates the unfolded protein response (UPR), suppresses co-stimulatory molecule expression and alters metabolic pathways to promote immunosuppression, which can be reversed by disabling IRE1α. Tumour-associated macrophages (TAMs) exhibit ER stress-driven activation of IRE1α–XBP1s, supporting tumour growth through the secretion of interleukin-4 (IL-4), IL-6, vascular endothelial growth factor (VEGFA), tyrosine-protein phosphatase non-receptor type substrate 1 (SIRPA) and thrombospondin 1 (THBS1). Protein kinase RNA-like ER kinase (PERK) activation in TAMs enhances serine biosynthesis by upregulating phosphoserine aminotransferase 1 (PSAT1) and promotes glycolysis through increased glucose transporter 1 (GLUT1) expression, facilitating their immunosuppressive polarization. Deleting IRE1α or PERK in TAMs reprogrammes their phenotype, reducing tumour progression and improving responses to immune checkpoint blockade (ICB). In neutrophils, ER stress activates IRE1α and activating transcription factor 6 (ATF6), reinforcing their pro-tumour phenotype and limiting early T cell responses by upregulating arginase 1 (ARG1) and prostaglandin E2 (PGE₂). Deleting IRE1α in neutrophils delays tumour growth and enhances the efficacy of programmed cell death protein 1 (PD1) blockade in primary high-grade serous ovarian cancer models. ER stress also increases Nectin2 and chemokine ligand 5 (CCL5) expression through transcriptional control, which was found to suppress antitumoural T cell responses. In monocytic-myeloid derived suppressor cells (M-MDSCs), PERK preserves mitochondrial DNA integrity, promoting immunosuppression. Ablation of PERK in MDSCs limits phospho-NRF2 antioxidant activity, therefore increasing the levels of ROS. The increased ROS triggers mitochondrial DNA leakage, activating the stimulator of IFN genes (STING) pathway and enhancing antitumour immunity through type I interferon (IFN-I) secretion. Activation of the PERK–ATF4–C/EBP homologous protein (CHOP) axis also upregulates C/EBPβ expression, which induces IL-6 secretion and downstream signal transducer and activator 3 (STAT3) phosphorylation. Phosphorylated STAT3 directly binds to the Arg1 promoter, inducing its expression and thereby regulating MDSC immunosuppressive function. IRE1α–XBP1s activation supports natural killer (NK) cell proliferation and mitochondrial function by upregulating MYC. This process sustains NK cell-mediated tumour control in mouse melanoma models. UPR dysregulation in tumour-associated myeloid cells therefore contributes to reduced antigen presentation, impaired cytotoxicity against cancer cells, increased T cell suppression and resistance to ICB. b, Tumour-induced ER stress provokes aberrant UPR activation in infiltrating T cells, altering their transcriptional, metabolic and functional profiles, ultimately compromising adaptive immunity. Restricted glucose availability or decreased glucose uptake causes persistent ER stress by limiting N-linked glycosylation of ER-resident proteins. This causes constitutive IRE1α–XBP1s signalling in ovarian cancer-infiltrating CD4⁺ T cells, which dampens glutamine uptake and blunts T cell mitochondrial respiration. In CD8⁺ T cells, cholesterol and tumour-derived factor uptake induces ER stress through IRE1α–XBP1s signalling (as demonstrated in ovarian tumours). This suppress transgelin 2 (TAGLN2), a cytoskeletal element essential for fatty acid-binding protein 5 (FABP5)-mediated lipid uptake and mitochondrial fatty acid oxidation (FAO), thus promoting T cell dysfunction. This also upregulates immune checkpoint molecule expression, including PD1, T cell immunoglobulin and mucin-domain-containing-3 (TIM3), T cell immunoreceptor with Ig and ITIM domains (TIGIT) and lymphocyte activation gene 3 (LAG3). Taurine deprivation in intratumoural CD8⁺ T cells activates PERK–ATF4 signalling, also upregulating immune checkpoint molecules, promoting T cell exhaustion. CTLA4, cytotoxic T lymphocyte protein 4; MHC-I, major histocompatibility complex I; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PMN-MDSC, polymorphonuclear myeloid-derived suppressor cell; RIDD, regulated IRE1α-dependent decay.

Antigenic peptides were then reported to directly activate IRE1α in DCs, causing RIDD-driven downregulation of MHC-I molecules⁷⁷. Abrogating IRE1α function in DCs alleviated RIDD and increased their capacity to cross-present antigens to T cells⁷⁷. Systemic pharmacological inhibition of the kinase domain of IRE1α, using the small molecule G9668, improved antitumour responses and augmented the efficacy of checkpoint blockade in mouse models of breast cancer⁷⁷. Whether these therapeutic effects were mediated by improved intratumoural DC function was not established and warrants further investigation (Fig. 3a).

BAT3 is an ER chaperone involved in the quality control of newly synthesized proteins, and its deficiency was found to cause robust UPR activation in DCs, which impaired antitumour T cell responses⁷⁸. DCs isolated from the spleen and lymph nodes of MC38 tumour-bearing mice displayed reduced BAT3 expression compared with splenic DCs from naive mice⁷⁸ (Fig. 3a), indicating that tumours suppress the expression of this ER chaperone in distal DCs to co-opt their function. UPR activation driven by BAT3 deficiency altered DC metabolism, redirecting acetyl-CoA towards the production of immunosuppressive glucocorticoids and reduced expression of co-stimulatory molecules, which were restored upon treatment with the IRE1α RNase domain inhibitor 4μ8C (ref. 78) (Fig. 3a).

These studies demonstrate that IRE1α sustains the tolerogenic phenotype of DCs in cancer hosts and indicates that controlling abnormal IRE1α activation in these myeloid cells can enhance endogenous antitumour immunity and improve the efficacy of cancer immunotherapy.

Macrophages.

TAMs are an abundant myeloid subset in various cancers and their presence often correlates with poor patient survival and resistance to immunotherapy⁷⁹. TAMs from patients with colorectal cancer exhibit elevated expression of XBP1s, which was proposed to support the growth and metastasis of cancer cells⁸⁰. This tumour-promoting phenotype appears to be driven by XBP1s-dependent induction of several key molecules secreted by TAMs such as vascular endothelial growth factor A (VEGFA)⁸¹, IL-4 and IL-6, which drive macrophage polarization towards an immunosuppressive phenotype⁸² and correlate with poor prognosis in patients with cancer⁸³ (Fig. 3a). In addition, XBP1s regulates the expression of signal regulatory protein α (SIRPA), which interacts with the ‘don’t eat me’ signal, CD47, expressed by tumour cells that prevents phagocytosis and thrombospondin 1 (THBS1), an extracellular matrix glycoprotein that inhibits the antitumour immune response by interacting with CD47 and CD36⁸⁰ (Fig. 3a).

Activation of PERK has similarly emerged as a key driver to the immunosuppressive features of TAMs. PERK deletion prevented macrophages from becoming immunosuppressive and caused decreased serine biosynthesis due to downregulation of phosphoserine aminotransferase 1 (PSAT1), a process regulated by the transcription factor ATF4⁸⁴ (Fig. 3a). PSAT1 mediates serine biosynthesis, which supports mitochondrial fitness and balances the production of α-ketoglutarate (α-KG). This process is necessary for jumonji domain-containing protein-3 (JMJD3)-dependent histone demethylation, which helps reinforce TAM immunosuppressive function and proliferation⁸⁴. Moreover, deletion of PERK reprogrammed the metabolic profile of TAMs and limited their immunosuppressive actions, which delayed tumour growth, enhanced T cell-mediated antitumour responses and improved efficacy of anti-PD1 therapy⁸⁴. A recent study further highlighted the major role of PERK in regulating the glycolytic capacity of immunosuppressive TAMs in glioblastoma⁸⁵. Of note, PERK activation in TAMs isolated from glioblastoma-bearing mice was shown to promote expression of the glucose transporter GLUT1 and enhance glycolysis⁸⁵. PERK deletion in these TAMs abolished histone lactylation in genes encoding immunosuppressive factors such as IL-10, disrupting their ability to support immune evasion and tumour progression⁸⁵.

Neutrophils.

In most patients with cancer, tumour-associated neutrophils (TANs) exhibit an inherently immunosuppressive and pro-tumorigenic phenotype⁸⁶. A recent study using an autochthonous mouse model found that high-grade serous ovarian carcinoma (HGSOC) programmes neutrophils to inhibit T cell-mediated antitumour responses by activating IRE1α in infiltrating neutrophils⁸⁷. TANs showed heightened activation of ER stress response markers compared with their counterparts in non-tumour sites⁸⁷. Selective deletion of IRE1α in neutrophils delayed primary ovarian tumour growth and extended survival in HGSOC-bearing mice by enabling early T cell-mediated tumour control⁸⁷ (Fig. 3a). Notably, loss of IRE1α in neutrophils sensitized tumour-bearing mice to PD1 blockade, resulting in HGSOC regression and long-term survival in approximately 50% of treated mice⁸⁷. In pancreatic ductal adenocarcinoma, upregulation of ER stress-related genes, namely HSPA5, DDIT3 and ATF6 were found in both human and mouse TANs⁸⁸. TANs enhance tumour cells migration and invasion through the secretion of chemokine ligand 5 (CCL5), while inducing T cell exhaustion through Nectin2 upregulation. Treatment with the ER stress inhibitor 4-phenylbutyric acid (4-PBA) decreased CCL5 and Nectin2 expression, while impairing cancer cell migration and invasion⁸⁸ (Fig. 3a). These findings indicate that targeting neutrophil-intrinsic ER stress pathways enhances endogenous and immunotherapy-induced immune responses to control tumour progression.

MDSCs.

MDSCs, immature myeloid cells with potent T cell-suppressive functions, operate as key mediators of immune evasion in cancer^89–91. Activation of the UPR upon exposure to tumour-derived stressors enhances the immunosuppressive capabilities of MDSCs⁹². In tumour-associated MDSCs, PERK signalling preserves mitochondrial DNA integrity, promoting their suppressive function against antitumour T cells⁹³. Pharmacological inhibition or conditional genetic deletion of PERK in MDSCs led to the cytosolic release of mitochondrial DNA due to inhibition of NRF2 phosphorylation by PERK⁹³. The ensuing leakage of mitochondrial DNA into the cytosol activated the cyclic GMP-AMP synthase (cGAS)–stimulator of interferon genes (STING) pathway, which resulted in type I IFN secretion and the development of robust antitumour immunity⁹³ (Fig. 3a). Hence, PERK ablation reprogrammed MDSCs to facilitate T cell responses against malignant cells.

Deletion of the PERK downstream targets CHOP and ATF4 has also been shown to impair MDSC suppressive function as this downregulates C/EBPβ, phosphorylated STAT3 and IL-6⁸⁹ (Fig. 3a). Beyond PERK, IRE1α and ATF6 have also been found to modulate the functions of specific MDSC subsets. The ablation of IRE1α or ATF6 disrupted the suppressive function of PMN-MDSCs as they display improved antigen-specific responses and delayed tumour growth in a mouse model of MC38 colon carcinoma⁹⁴ (Fig. 3a). Intriguingly, this did not affect monocytic MDSCs⁹⁴.

NK cells.

Activated NK cells exhibit constitutive activation of the IRE1α–XBP1s axis, which has been shown to be essential for their proliferation but not for their survival⁹⁵. This effect is partially mediated by IRE1α-dependent upregulation of MYC, a key transcription factor regulating cellular metabolism and proliferation⁹⁵. Mechanistically, activation of the IRE1α–XBP1s–MYC axis supports mitochondrial function, which is critical for maintaining NK cell bioenergetic demands, sustaining ATP production and promoting oxidative metabolism, necessary for cytokine production and cytotoxic activity against tumour cells^95–98. Consequently, selective deletion of IRE1α in NK cells led to defective mitochondrial function, impairing their ability to control melanoma tumour progression⁹⁵ (Fig. 3a). This underscores the adaptive role of the UPR in NK cells, ensuring they can meet the metabolic and functional demands required for robust antitumour immunity. However, whether more aggressive malignancies induce maladaptive or aberrant UPR signalling in NK cells to compromise their protective function remains unknown. Since the connection between NK cells and the UPR has primarily been explored in melanoma^95–98, further studies are needed to determine whether the same effects are observed in other malignancies. Additional research is needed to understand how different tumour contexts manipulate the UPR in NK cells and whether targeting this pathway could alter NK cell-mediated immunotherapy.

Adaptive immune cells

T cells.

Tumour-induced ER stress has been shown to trigger aberrant UPR activation in infiltrating T cells, altering their transcriptional, metabolic and functional profiles⁹⁹. These changes compromise their capacity to eliminate malignant cells, fostering a permissive environment for tumour growth. For instance, chronic activation of the IRE1α–XBP1s pathway in ovarian cancer-infiltrating T cells weakens their metabolic fitness and effector functions^11,12. Specifically, tumour-derived factors present in ovarian cancer ascites were found to reduce the expression of the glucose transporter GLUT1 in activated CD4⁺ T cells. This process impairs glucose uptake and disrupts normal protein glycosylation in the ER, leading to persistent IRE1α–XBP1s activation in T cells¹² (Fig. 3b). Elevated XBP1s levels negatively regulate the abundance of glutamine transporters, thereby restricting T cell glutamine uptake – a critical, alternative fuel source for mitochondrial respiration under glucose-limited conditions¹² (Fig. 3b). Notably, ovarian cancer-bearing mice lacking IRE1α or XBP1s specifically in T cells exhibited delayed tumour progression and extended overall survival¹². This was accompanied by a transcriptional reprogramming of intratumoural CD4⁺ T cells leading to increased IFNγ production and upregulation of immune-stimulatory gene profiles in the TME¹².

Beyond CD4⁺ T cells, recent findings have highlighted the detrimental impact of IRE1α–XBP1s signalling in CD8⁺ T cells infiltrating ovarian tumours¹¹. In both human and mouse models of HGSOC, activation of the IRE1α–XBP1s pathway was shown to suppress the expression of transgelin 2 (TAGLN2), a cytoskeletal organizer that has a pivotal role in lipid metabolism within CD8⁺ T cells¹¹. This study found that TAGLN2 not only coordinates cytoskeletal processes but also supports the localization and function of fatty acid-binding protein 5 (FABP5), which is essential for lipid uptake and mitochondrial fatty acid oxidation. Mechanistically, XBP1s directly binds to the TAGLN2 promoter, inhibiting its transcription and thereby disrupting FABP5-mediated lipid metabolism in CD8⁺ T cells¹¹ (Fig. 3b). This study therefore established a novel link between ER stress responses and CD8⁺ T cell metabolism and function through repression of a cytoskeletal organizer. Notably, low expression of TAGLN2 correlated with a dysfunctional phenotype in both human and mouse intratumoural CD8⁺ T cells¹¹, suggesting that TAGLN2 deficiency may serve as a hallmark of T cell metabolic and effector malfunction in the TME. This observation extends beyond ovarian cancer, with emerging evidence indicating that reduced TAGLN2 expression is similarly observed in exhausted T cells across multiple tumour types¹¹. Future research could leverage TAGLN2 as a target to restore cytoskeletal integrity, potentially reinvigorating T cell metabolism and functionality under tumour-induced ER stress.

In a B16 mouse model of metastatic melanoma, cholesterol accumulation induced intratumoural CD8⁺ T cell exhaustion through activation of IRE1α–XBP1s¹³ (Fig. 3b). This upregulated immune checkpoints such as PD1, UD2B4, TIM3 and lymphocyte activation gene 3 (LAG3), impairing T cell function. Targeted inhibition of XBP1s in CD8⁺ T cells restored their antitumour activity in a cholesterol-rich TME¹³. Targeting XBP1s in CD8⁺ T cells reverses T cell exhaustion in cholesterol-rich environments, and assessing whether similar benefits occur in TMEs with different metabolic profiles could enhance the adaptability and effectiveness of this therapeutic strategy.

Active uptake of taurine by cancer cells results in reduced levels of taurine in the TME, which causes detrimental ER stress responses in infiltrating T cells³³ (Fig. 3b). Indeed, a recent study demonstrated, in mice harbouring gastric cancer (MFC) or melanoma (B16F10) cell lines, that intratumoural CD8⁺ T cells with limited taurine availability undergo ER stress responses resulting in T cell exhaustion that impairs antitumour T cell responses³³. Mechanistically, taurine deprivation caused ER stress-driven ATF4 induction that transactivated immune checkpoint genes PDCD1, CTLA4, TIGIT and TIM3, thereby impairing the ability of CD8⁺ T cells to mount a sustained antitumour response³³ (Fig. 3b). Taurine supplementation enhanced the antitumour activity of CD8⁺ T cells and increased the efficacy of cancer immunotherapy, as demonstrated in other studies^34,35,100. In addition to taurine, CO has been identified as another TME-associated molecule that affects T cell metabolism and function. Mechanistically, CO-mediated activation of PERK triggers autophagy, enabling T cells to metabolically adapt to prolonged ER stress, restoring mitochondrial function and bolstering antitumour responses⁴⁸. By reprogramming T cell metabolism to favour autophagy, CO creates a paradoxical situation where it both limits T cell exhaustion and enhances their persistence and effector function in the hostile TME. Autophagy enables T cells to manage and recycle cellular components, thus sustaining their viability and functionality. This mechanism helps prevent the energy depletion typically associated with T cell exhaustion, allowing them to persist and remain active against tumours. This finding suggests that manipulating autophagy by regulating the PERK pathway may offer a therapeutic avenue for improving T cell survival and efficacy in cancer.

C/EBP homologous protein (CHOP), a downstream transcription factor induced during ER stress via PERK signalling, has been shown to inhibit T cell function in mouse models of melanoma and ovarian cancer¹⁰¹. Mechanistically, ER stress-induced PERK–CHOP activation leads to the upregulation of ATF4, which directly represses the expression of T-bet, a major transcription factor required for the differentiation and effector function of cytotoxic CD8⁺ T cells¹⁰¹. This repression blunts the expression of crucial effector molecules such as granzyme B, perforin and IFNγ, impairing the tumour killing function of T cells and thus allowing immune evasion¹⁰¹. CHOP-mediated suppression of T-bet is a critical node in the ER stress–T cell dysfunction axis. Modulating CHOP activity may help restore T-bet expression and reinvigorate T cell-based therapies by overcoming T cell dysfunction in the TME. Furthermore, the cumulative effects of chronic ER stress result in mitochondrial exhaustion, which is marked by the loss of mitochondrial integrity and a reduction in ATP production in T cells¹⁰¹. In this context, recent studies found that the prolonged activation of UPR pathways, particularly the IRE1α–XBP1s and PERK branches, leads to mitochondrial dysfunction in CD8⁺ T cells, compromising their bioenergetic capacity and anticancer effector function^99,102.

A subsequent study demonstrated that sustained PERK–ERO1α activity drives oxidative protein folding, generating mitochondrial reactive oxygen species (mtROS), which were identified as a biomarker of ER-induced mitochondrial exhaustion in T cells¹⁰³. High mtROS levels were found to be specifically enriched in PD1⁺ tumour antigen-specific CD8⁺ tumour-infiltrating leukocytes (TILs), correlating with mitochondrial dysfunction and diminished antitumour capacity¹⁰³. One potential mechanism linking mtROS to increased PD1 expression involves oxidative stress-mediated activation of signalling pathways that upregulate PD1. Oxidative stress can trigger nuclear factor-κB (NF-κB) and other transcription factors that enhance PD1 transcription in T cells, serving as a mechanism to limit T cell activation and prevent excessive inflammation that could further damage the tissue¹⁰⁴. Further research is needed to understand how mtROS regulate PD1 expression in CD8⁺ TILs, which could guide more effective cancer immunotherapy strategies. Pharmacological inhibition of PERK or ERO1α, as well as genetic deletion of PERK in T cells, reduced oxidative stress, preserved mitochondrial energy reserves and enhanced T cell effector function. These T cells exhibited superior tumour control compared with conventional effector T cells, demonstrating the therapeutic potential of targeting this ER stress pathway in T cells to bolster antitumour immunity¹⁰³. Additionally, in both human and mouse tumours, glucose restriction in the TME was shown to activate the PERK-dependent eIF2α phosphorylation (p-eIF2α), leading to translational attenuation that weakened the potential of T cells to suppress mouse B16-F1-OVA melanoma tumour growth¹⁰⁵. Intriguingly, proteasome activity emerged as a key regulator of this pathway, mitigating p-eIF2α induction and preventing translation attenuation. Metabolic and pharmacological interventions targeting the proteasome enabled prolonged cytokine production, enhancing T cell-mediated tumour suppression in mouse solid tumours¹⁰⁵. These findings identify proteasome activity as a critical modulator of translational control in T cells, providing a promising therapeutic strategy to overcome PERK–p-eIF2α-induced intratumoural T cell dysfunction and improve the efficacy of tumour immunotherapy.

The role of T cell-intrinsic ATF6 in tumours is less understood compared with the other ER stress sensors. Multiple studies have shown the intrinsic tumorigenic role of ATF6 in cancer cells. For instance, in colon cancer cells and their associated stromal cells, ATF6 enhances autophagy and supports oncogenic microbial dysbiosis^74,106–108. However, the specific role of ATF6 in intratumoural T cell biology remains poorly understood. Further research is crucial to understand whether ATF6 influences T cell activity and fate in the TME.

B cells.

A recent study suggested that B cell function within tertiary lymphoid structures (TLSs) is crucial for mounting effective antitumour immunity¹⁰⁹. The formation and maintenance of TLSs are critical for orchestrating local immune responses, including the activation of B cells that produce tumour-targeting antibodies, yet how ER stress impacts B cell dynamics is unknown. One intriguing possibility is that metabolic restrictions common in the TME¹¹⁰, such as limited availability of glucose, glutamine or taurine, could also affect B cell function through ER stress pathways. For instance, the depletion of glucose or glutamine could impair N-linked glycosylation processes essential for proper antibody folding and maturation, thereby inducing ER stress in infiltrating B cells¹¹¹. Such stress might also lead to activation of the UPR and subsequent dysfunction of B cells, potentially compromising their ability to produce and secrete functional antibodies against tumour cells. Importantly, in B cells, IRE1α cleaves the mRNA encoding the secretory μ (μS) heavy chain of immunoglobulin via RIDD^112,113. This process was caused by IRE1α hyperactivation due to XBP1s deficiency, leading to accelerated degradation of μS mRNA and a consequent reduction in IgM synthesis⁹⁰. Yet, whether XBP1s-competent intratumoural B cells exhibit endogenous RIDD that decreases IgM synthesis remains to be established. A deeper exploration into the effects of ER stress in intratumoural B cell biology and TLS formation could offer insights into novel immunotherapeutic strategies, particularly for enhancing B cell responses in the TME.

Modulating ER stress responses for cancer immunotherapy

Preclinical models

Among the key sensors of ER stress, IRE1α has emerged as a compelling therapeutic target due to its dual enzymatic functions — RNase and kinase activity — which promote tumour adaptation and immune evasion. Pharmacological inhibition of these domains has been shown to disrupt tumour survival mechanisms while also reshaping the tumour immune microenvironment. Small-molecule inhibitors of the IRE1α RNase domain, such as MKC3946, MKC8866 (also called ORIN1001), B-I09 and STF83010, have demonstrated antitumour activity in preclinical models^{53,66,113–116} (Table 1). MKC3946 enhances the efficacy of the proteosome inhibitor bortezomib and the HSP90 inhibitor 17-AAG in multiple myeloma¹¹⁷, whereas MKC8866 improves the response to the microtubule-stabilizing agent docetaxel in TNBC models⁵⁰. Moreover, a recent seminal study revealed that IRE1α acts as a crucial checkpoint that limits the immunostimulatory effects of taxane chemotherapy in TNBC by preventing dsRNA accumulation and subsequent pyroptotic cell death⁷⁰. Mechanistically, inhibiting the IRE1α RNase domain with ORIN1001 in TNBC cells exposed to taxane leads to accumulation of dsRNA within the cell, which subsequently activates NLRP3–GSDMD-dependent pyroptosis, a highly inflammatory form of cell death⁷⁰. This in turn triggers an immunogenic form of cell death that transforms previously immune-cold, PDL1-negative tumours into immunogenic PDL1-high tumours, rendering them more responsive to ICB⁷⁰. This approach of combining ORIN1001 with taxanes presents a promising strategy to convert immunoresistant tumours into ‘hot’ tumours, enhancing their susceptibility to immunotherapy. Importantly, however, these effects were restricted to taxanes, as other chemotherapeutic agents did not induce IRE1α-dependent degradation of dsRNA⁷⁰. Additional research is needed to determine whether suppression of IRE1α signalling may potentiate ICD induced by other cytotoxic agents such as anthracyclines, leading to improved immune infiltration and enhanced responses to ICB.

Table 1 |.

Antitumour effects of targeting ER stress pathways in preclinical cancer models

UPR target	Therapy (through inhibitions or modulation)	Effects in cancer cells	Effects in immune cells	Refs.
IRE1α RNase domain	MKC3946	Induces apoptosis in multiple myeloma cells via CHOP-dependent mechanisms	Insufficiently characterized	116
	MKC8866 (ORIN1001)	Sensitizes MYC-addicted breast and prostate tumours to therapy; reduces expression of tumorigenic cytokines (IL-6, IL-8, CXCL1, TGFβ2, GM-CSF) by breast cancer cells; induces dsRNA accumulation and pyroptotic cancer cell death; sensitizes prostate tumours to anti-PD1 therapy	Rescues TAGLN2 expression and lipid uptake in ER-stressed T cells	11,50,66,70
	B-I09	Sensitizes CARM1hi and MYC-transformed tumours to therapy	Insufficiently characterized	118,119
	STF83010	Sensitizes breast tumours to tamoxifen-mediated cancer therapy	Insufficiently characterized	70,146
	4μ8C	Insufficiently characterized	Enhances mitochondrial respiration and IFNγ production in CD4+ T cells exposed to ovarian cancer ascites; reduces lipid accumulation by ovarian cancer-associated DCs exposed to ascites; improves antigen presentation by BAT3-deficient DCs	12,76,78
IRE1α kinase domain	KIRA8	Restrains tumour growth and enhances response to chemotherapies	Insufficiently characterized	121,122
	G9668	Reduces tumour growth in multiple syngeneic models and improves the effects of anti-PDL1 treatment in mice bearing EMT6 tumours	Improves antigen cross-presentation by DCs	77
PERK	GSK2606414, GSK2656157, AMG44, LY-4, HC-504, NMS-812	Suppresses tumour growth; reduces blood vessel density	Improves T cell effector function	93,127,131
eIF2α	Salubrinal, ISRIB	Promotes translation of pro-apoptotic factors	Improves immune recognition	129–131
ATF6	Ceapins	Sensitizes cancer cells to ER stress-induced apoptosis	Insufficiently characterized	132
BiP	KP1339, HA15	Induces unresolved ER stress; triggers immunogenic cell death	Insufficiently characterized	133–141

ATF6, activating transcription factor 6; BiP, binding-immunoglobulin protein; CHOP, C/EBP homologous protein; CXCL1, C-X-C motif chemokine ligand 1; DC, dendritic cell; dsRNA, double-stranded RNA; eIF2α; eukaryotic initiation factor 2α; ER, endoplasmic reticulum; GM-CSF, granulocyte–macrophage colony-stimulating factor; IFNγ, interferon-γ; IL-6, interleukin-6; IRE1α, inositol-requiring enzyme 1α; PD1, programmed cell death protein 1; PDL1, programmed cell death 1 ligand 1; PERK, protein kinase RNA-like ER kinase; TAGLN2, transgelin 2; TGFβ2, transforming growth factor-β2; UPR, unfolded protein response.

B-I09, another selective IRE1α RNase inhibitor¹¹⁸, has been shown to improve responses to ICB in mouse models of HGSOC¹¹⁹. Of note, B-I09 was particularly effective in tumours overexpressing CARM1, a histone arginine methyltransferase linked to heightened ER stress and tumour progression¹¹⁹. CARM1 overexpression promotes dependency on IRE1α–XBP1s signalling, making these tumours especially vulnerable to IRE1α inhibition¹¹⁹. In these models, B-I09 treatment not only suppressed tumour growth but also synergized with anti-PD1 therapy, improving immune-mediated tumour clearance¹¹⁹.

Inhibiting the IRE1α kinase domain presents another therapeutic avenue. KIRA8, a small-molecule kinase inhibitor, selectively inhibits IRE1α autophosphorylation, thereby preventing downstream RNase activation¹²⁰. In multiple myeloma, KIRA8 preferentially reduced the viability of CD138⁺ plasma cells from both newly diagnosed and relapsed patients, while sparing non-malignant haematopoietic cells¹²¹. Preclinical studies using subcutaneous and orthometastatic mouse models of multiple myeloma demonstrated that KIRA8 enhances the efficacy of bortezomib (a proteasome inhibitor) and lenalidomide (an immunomodulatory drug that promotes anti-multiple myeloma immune responses)¹²¹. Beyond multiple myeloma, KIRA8 has also shown efficacy in TNBC and pancreatic neuroendocrine tumour models, where it suppresses tumour progression via IRE1α inhibition^3,122. Further studies are needed to determine whether targeting the IRE1α kinase domain affects antitumour immunity, particularly in the context of ICB.

PERK inhibitors, such as GSK2606414 and GSK2656157, have demonstrated efficacy in solid tumours, where they suppress tumour growth, reduce blood vessel density and restore T cell function¹²⁰. In immunogenic sarcoma models, PERK inhibition reprogrammed the TME, enhancing immune infiltration and improving the response to anti-PD1 therapy¹²⁰. However, systemic pancreatic toxicity and off-target effects have limited their clinical translation, necessitating the development of more selective PERK inhibitors.

AMG44, a next-generation PERK inhibitor, has been shown to delay tumour growth and modulate MDSC activity in the TME without affecting glucose levels or insulin secretion, thus reducing the risk of systemic toxicity⁹³. When administered as a monotherapy, the PERK inhibitor LY-4 exhibited potent antitumour activity in BRAF^V600E mutant melanoma and MYC-driven lymphoma models, demonstrating a favourable toxicity profile^123,124. Additional PERK inhibitors, including HC-5404 (NCT04834778)¹⁸ and NMS-812 (a dual PERK/GCN2 inhibitor) (NCT06549790)¹²⁵ are currently being evaluated in early phase clinical trials^19,126. Their ability to modulate intratumoural immune cell function and enhance ICB efficacy remains a critical area of investigation.

Under conditions of ER stress, PERK activation phosphorylates eIF2α, initiating the integrated stress response (ISR) – a key regulatory pathway that controls protein synthesis and immune cell function to maintain cellular homeostasis and adaptation to stress^{12 ,128}. Pharmacological inhibition of p-eIF2α using salubrinal has been shown to enhance antitumour immune responses in preclinical cancer models. Specifically, in human inflammatory breast cancer and ovarian cancer cell lines, salubrinal promotes the translation of pro-apoptotic mRNAs, inducing tumour cell death^129,130. Additionally, salubrinal has been demonstrated to upregulate MHC class I expression on tumour cells, enhancing CD8⁺ T cell recognition and cytotoxicity, which could improve tumour clearance in vivo¹³¹. However, further studies in syngeneic mouse models are needed to validate whether this effect extends to the TME and whether it enhances ICB therapy. Another small-molecule ISR inhibitor, ISRIB, reverses p-eIF2α-mediated translational suppression and has been shown to inhibit tumour progression in PTEN-deficient and MYC-driven prostate cancer models, leading to prolonged survival in tumour-bearing mice⁵². Although these findings highlight the therapeutic potential of ISR modulation, its role in regulating tumour-infiltrating immune cells remains underexplored. Future studies are needed to address how ISR inhibition influences DC function, T cell activation and myeloid cell differentiation within the TME.

ATF6 signalling has a key role in tumour adaptation to ER stress, and its inhibition has been explored as a potential therapeutic strategy. Ceapins, a novel class of small-molecule inhibitors, selectively block ATF6α activation without interfering with other UPR pathways such as IRE1α or PERK¹³². However, although ATF6 inhibition has been proposed as a strategy to disrupt tumour survival mechanisms, its role in antitumour immunity remains poorly understood and deserves further investigation. Targeting the binding-immunoglobulin protein (BiP), also known as glucose-regulated protein 78 (GPR78), a master ER protein chaperone upregulated in many cancers^133–136, may offer additional immunotherapeutic strategies. BiP inhibitors such as KP1339, a ruthenium-based compound, induce terminal ER stress responses and ICD in diverse cancer cell lines in vitro^137–139. HA15, another BiP inhibitor, induces unresolved ER stress leading to cancer cell death, particularly in melanoma models resistant to BRAF inhibitors¹⁴⁰. Additionally, therapies targeting cell-surface BiP, which is upregulated and relocated to the cell surface in tumour cells in response to ER stress — unlike in normal cells, where it remains within the cytoplasm and nucleus — show promise in cancer immunotherapy¹⁴¹. Synthetic chimeric peptides have been developed that combine BiP-binding motifs with molecules known to induce cell death¹⁴², which was shown to suppress tumour growth in xenograft mouse models of prostate and breast cancer¹⁴². This approach leverages the overexpression of BiP on cancer cells to selectively deliver cytotoxic agents, thereby inducing targeted apoptosis without harming normal cells¹⁴². Another study identified a cancer cell-specific O-linked carbohydrate moiety on cell surface-localized BiP¹⁴³. This glycosylation pattern is not found in normal cells, making it an ideal target for both immunotherapies and antibody-based therapies¹⁴³. In gastric cancer, exploiting this unique feature could enhance the specificity and efficacy of treatment, potentially leading to better patient outcomes with reduced side effects¹⁴³. A subsequent study demonstrated that chimeric antigen receptor (CAR) T cells engineered with BiP-binding peptide Pep42, targeting cell surface BiP, show promise as a potential therapy for glioblastoma, a cancer known for its poor prognosis and resistance to conventional treatments¹⁴⁴. In vitro co-culture experiments demonstrated that this CAR-T cells were able to specifically target and eliminate glioblastoma tumour cells and glioblastoma stem cells, accompanied by the release of IFNγ¹⁴⁴. Additionally, in a tumour xenograft model, these CAR-T cells effectively reduced the glioblastoma stem cell population and curtailed tumour growth¹⁴⁴.

Together, these preclinical findings suggest that targeting UPR components holds potential for enhancing the efficacy of cancer therapies, particularly in combination with immune-modulating strategies such as ICB. Ongoing clinical trials investigating UPR-modulating agents (Table 2) present an important opportunity to establish their role in human antitumour immunity and cancer immunotherapy. Selective inhibition or modulation of key UPR pathways has the potential to synergize with standard treatments, not only enhancing tumour cell sensitivity to therapy, but also reshaping the tumour immune microenvironment to improve protective immune infiltration and activation.

Table 2 |.

UPR modulators currently in clinical trials

UPR target	Agent	Clinical phase and trial info	Key findings	Refs.
IRE1α RNase domain	ORIN1001	Phase I/II (NCT03950570, NCT05154201)	Manageable safety profile with dose-limiting toxicities guiding dosing; combination therapy shows potential	147,148
PERK	HC-5404-FU, NMS-03597812	Phase I (NCT04834778, NCT05027594)	Improved selectivity and safety in solid tumours and multiple myeloma	18,126,149
BiP	KP1339	Phase I (KP1339/IT-139)	Triggers immunogenic cell death and achieves disease control in advanced solid tumours	137,150

BiP, binding-immunoglobulin protein; ER, endoplasmic reticulum; IER1α, inositol-requiring enzyme 1 α; PERK, protein kinase RNA-like ER kinase; UPR, unfolded protein response.

Further research is essential to define the precise immunological consequences of UPR targeting, optimizing strategies that exploit ER stress response pathways to boost antitumour immune responses while minimizing potential off-target effects. A deeper understanding of these mechanisms will be critical for the successful integration of UPR-targeted agents into immunotherapeutic regimens, ultimately advancing more effective and durable treatment options for patients with cancer.

Correlative studies and patient responses

Recent studies have highlighted the noteworthy impact of ER stress on patient responses to immunotherapy, particularly immune checkpoint inhibitors. Elevated ER stress responses within tumours have been associated with poor clinical outcomes. For instance, increased expression of XBP1s correlates with worse prognosis in patients with NSCLC⁶⁴ and higher levels of CHOP in intratumoural T cells were associated with unfavourable clinical outcomes across various cancers¹⁰¹. Conversely, patients with melanoma whose tumours expressed low levels of ER stress markers, such as reduced XBP1s, BiP and ATF4, responded better to CTLA4 blockade and demonstrated longer progression-free survival¹⁴⁵. Additionally, tumours from patients with melanoma that have increased expression of a PERK-related mRNA signature show lower survival and reduced response to ICB⁴⁸. These findings underscore the potential of using ER stress-related elements as predictive biomarkers of immunotherapy responsiveness. Monitoring these stress levels in patients could inform personalized treatment strategies, allowing for the identification of individuals who might benefit from additional therapeutic interventions aimed at modulating ER stress. Integrating UPR-targeting therapies in patients with high ER stress could enhance the effectiveness of immune checkpoint inhibitors. Thus, the assessment of ER stress levels or the magnitude of UPR activation could have a pivotal role in optimizing treatment approaches in cancer immunotherapy, emphasizing the need for further research into the mechanistic links between ER stress, the UPR and immune resistance in cancer.

Conclusions and future directions

Recent discoveries have established the UPR as a key determinant of immune regulation in the TME, influencing cancer progression, immune suppression and immunotherapy resistance. The UPR regulates key processes such as antigen presentation, immune cell metabolism and inflammatory signalling, collectively promoting immune evasion and tumour progression. Although emerging evidence highlights the potential of targeting UPR pathways to enhance antitumour immunity, several fundamental questions remain unresolved. A deeper understanding of how ER stress shapes immune dynamics within tumours and how its modulation can improve immunotherapies is essential for translating these insights into clinical applications.

One key question is the impact of ER stress on cancer stem cell maintenance and immune evasion. Cancer stem cells exhibit enhanced UPR activation, allowing them to survive under conditions of metabolic and proteotoxic stress. Whether the UPR actively facilitates immune evasion in cancer stem cells by impairing antigen presentation or modifying their interactions with immune cells remains unknown. Given that cancer stem cells are a major contributor to tumour relapse and resistance to therapy, determining how ER stress influences cancer stem cell immune recognition and clearance will be critical for developing more effective immunotherapies.

Another unresolved question is whether manipulation of ER stress pathways could enhance the efficacy of cancer vaccines by improving antigen processing and immune activation. ER stress regulates key antigen-presenting machinery, yet it remains unclear how it affects the presentation of vaccine-derived tumour antigens. Investigating these mechanisms could inform the design of next-generation cancer vaccines that leverage ER stress modulation to increase immune priming and enhance durable tumour immunity.

The role of ER stress in immune cell epigenetic modifications is also poorly understood. Stress-induced epigenetic changes can have lasting effects on immune cell fate and function, potentially influencing tumour-associated macrophage polarization, dendritic cell maturation and T cell exhaustion. Whether ER stress rewires the epigenetic landscape of immune cells within the TME, thereby promoting immune dysfunction or therapy resistance, remains an open question. Identifying these ER stress-induced epigenetic signatures could uncover new therapeutic targets to restore effective antitumour immune responses.

Similarly, the impact of ER stress on immune memory formation remains unclear. Robust and durable antitumour immunity requires long-term immune memory, yet tumours exploit chronic stress pathways to dampen recall responses and prevent immune-mediated clearance. Understanding how ER stress influences the differentiation and persistence of memory T cells will be essential for improving strategies that sustain antitumour immunity over time.

Another pressing challenge in the field is defining the role of ER stress as a mechanism of immunotherapy resistance. Many tumours exhibit chronic UPR activation, which may interfere with ICB, adoptive T cell therapies, and other immunotherapeutic interventions. Understanding how specific ER stress sensors contribute to therapy resistance could lead to novel approaches for overcoming resistance mechanisms and optimizing treatment responses. A particularly promising avenue is the development of immune cells engineered with resistance to ER stress-induced dysfunction, which could enhance the efficacy of TILs, CAR macrophages and CAR-T cells by improving their persistence and cytotoxicity within the TME.

Pharmacological inhibition of key ER stress regulators, such as IRE1α and PERK, presents an opportunity to reprogramme the TME and enhance antitumour immunity. Although IRE1α RNase inhibitors have been shown to enhance tumour immunogenicity and convert immune-cold tumours into ICB-responsive tumours, the long-term impact of IRE1α inhibition on immune homeostasis requires further investigation. Similarly, PERK inhibition has been proposed as a strategy to enhance T cell persistence and fitness; however, the systemic effects of long-term PERK inhibition remain unclear. Future studies should focus on integrating UPR-targeting strategies with immunotherapies in well-characterized preclinical models to determine their immunological consequences and safety.

To translate these insights into clinical applications, future research should also focus on defining context-specific effects of ER stress on immune cell function, leveraging single-cell transcriptomics, proteomics and spatial multiomics approaches to distinguish between stress-adaptive and immunosuppressive UPR pathways. A more comprehensive understanding of how UPR modulation influences immune fate decisions will be critical for the rational design of combination immunotherapies.

Another key priority is the development of predictive biomarkers to identify patients most likely to benefit from ER stress-targeting therapies. Given that UPR activity varies across tumour types and patient populations, biomarker-driven patient stratification will be essential for ensuring precision in clinical applications. Identifying biomarkers of ER stress in tumour biopsies or circulating immune cells could provide insights into therapy response prediction and resistance mechanisms, ultimately guiding personalized treatment approaches.

In conclusion, targeting ER stress response pathways represents a promising frontier in cancer immunotherapy, with the potential to reshape the TME, enhance immune responses and overcome therapy resistance. However, several critical knowledge gaps remain, including the role of ER stress in cancer stem cell immune evasion, immune memory formation, vaccine efficacy and therapy resistance mechanisms. Addressing these questions through multiomics integration, biomarker-driven stratification and rational drug combinations will be essential for harnessing the full therapeutic potential of UPR modulation. As our understanding of ER stress in immune regulation deepens, these insights will drive the development of next-generation cancer immunotherapies with improved efficacy and durability, ultimately transforming clinical outcomes for patients.

Glossary

Autophagy

A conserved lysosomal degradation pathway that recycles cellular components, regulating metabolism, stress adaptation and immune responses in health and disease.

Immunogenic cell death

(ICD). A regulated form of cell death that elicits an adaptive immune response by releasing damage-associated molecular patterns and promoting antigen presentation.

Major histocompatibility complex

(MHC). A group of cell surface molecules responsible for antigen presentation to T cells, enabling immune recognition and activation in both innate and adaptive immunity.

Paraptosis

A caspase-independent, non-apoptotic form of programmed cell death distinguished by cytoplasmic vacuolization and mitochondrial swelling, often linked to ER stress and protein aggregation.

Polymorphonuclear myeloid-derived suppressor cells

(PMN-MDSCs). A subset of immunosuppressive myeloid cells with neutrophil-like features that inhibit T cell activation and promote immune evasion within the tumour microenvironment.

Pyroptosis

A lytic, pro-inflammatory form of programmed cell death triggered by inflammasome activation, characterized by gasdermin-mediated membrane pore formation and cytokine release.

Tertiary lymphoid structures

(TLSs). Ectopic lymphoid aggregates that develop in non-lymphoid tissues during chronic inflammation or cancer, serving as local hubs for immune activation and adaptive responses.