Interplay between the unfolded protein response and reactive oxygen species: a dynamic duo

Abstract

Secretory proteins undergo modifications such as glycosylation and disulphide bond formation before proper folding, and move to their final destination via the endomembrane system. Accumulation of unfolded proteins in the endoplasmic reticulum (ER) due to suboptimal environmental conditions triggers a response called the unfolded protein response (UPR), which induces a set of genes that elevate protein folding capacity in the ER. This review aims to establish a connection among ER stress, UPR, and reactive oxygen species (ROS), which remains an unexplored topic in plants. For this, we focused on mechanisms of ROS production originating from ER stress, the interaction between ER stress and overall ROS signalling process in the cell, and the interaction of ER stress with other organellar ROS signalling pathways such as of the mitochondria and chloroplasts. The roles of the UPR during plant hormone signalling and abiotic and biotic stress responses are also discussed in connection with redox and ROS signalling.

Endoplasmic reticulum (ER) stress, ER oxidoreductase, H₂O₂, protein disulphide isomerase, reactive oxygen species, redox regulation, unfolded protein response (UPR)

Introduction

The endoplasmic reticulum (ER) is responsible for the synthesis of secretory proteins and is also a central organelle involved in many signalling and metabolic processes (Iwata and Koizumi, 2012). Adaptation to stressful environments requires re-programming of cellular protein production and profile (Kosová et al., 2011), which increase protein synthesis load. Under such conditions, unfolded or misfolded proteins accumulate in the ER lumen, causing ER stress. Abiotic stresses such as high temperatures and salinity, and biotic stresses such as bacterial pathogens are known to induce ER stress (Kørner et al., 2015; Bao and Howell, 2017). ER stress is perceived by the unfolded protein response (UPR), which includes recognition of abnormal/unfolded proteins, induction of protein folding capacity, and finally protein degradation processes (Iwata and Koizumi, 2012; Howell, 2013).

Recent findings suggest that ER stress is also intimately related to redox metabolism and reactive oxygen species (ROS) in animals (Ding et al., 2012; Cao and Kaufman, 2014; Gu et al., 2016), yeast (Liang et al., 2014; Weids and Grant, 2014), and plants (Ozgur et al., 2014, 2015; Uzilday et al., 2018). ROS have dual roles in cells; in low concentrations, ROS act as signal molecules, generally as secondary messengers (Mittler 2017) However, excess accumulation of ROS causes oxidative stress that results in DNA damage, lipid peroxidation, membrane instability, and protein oxidation (Mittler et al., 2012). Excess ROS can also cause overoxidation of regulatory thiols in proteins, disrupting the fine-tuned metabolic balance (Waszczak et al., 2015). Accumulation of ROS in a cell is defined by its rate of production and scavenging (Mittler, 2017). ROS can be scavenged via antioxidants which can be enzymatic or non-enzymatic (reviewed in Uzilday et al., 2015). Enzymatic antioxidants include superoxide dismutase (SOD), catalase (CAT), peroxidase (POX), ascorbate peroxidase (APX), and glutathione peroxidase (GPX), while major non-enzymatic antioxidants include tocopherols, carotenoids, ascorbate, and glutathione (GSH) in the plant cells (Mittler et al., 2004; Krieger-Liszkay et al., 2008). Among these, the thiol-based glutathione/glutathione disulphide (GSH/GSSG) redox couple is closely associated with protein folding that occurs in the ER since disulphide bonds are formed in this compartment via oxidative protein folding (Aller and Meyer, 2013). Besides its relationhip to thiol metabolism, the ER lumen also acts as a site of Ca²⁺ storage (Klüsener et al., 1995), which is a well-known up- or downstream mediator of ROS signalling (reviewed by Gilroy et al., 2016).

In animals and yeast, there is accumulated knowledge on the interaction among ROS, antioxidants, and UPR signalling (reviewed in Cao and Kaufman, 2014). Recently, the number of publications on ER stress and UPR signalling in plants has been increasing; these elucidate new components related to this phenomenon especially in regards to ER stress perception and signalling (Tajima et al., 2008; Iwata et al., 2009; Ruberti et al., 2015; Walley et al., 2015; Hossain et al., 2016). However, studies related to interaction among ER stress, ROS, and cellular redox in plants are very limited.

In this review, we will present an update about UPR and related signalling events and then focus on oxidative protein folding in the ER and changes in cellular redox during ER stress. Possible ROS–hormone interactions during ER stress, and abiotic and biotic factors that trigger UPR and their possible relationship to ROS signalling will be discussed.

Unfolded protein response: its perception, signalling, and consequences

Nascent polypeptides of secretory and membrane proteins translated by ribosomes enter the ER lumen and undergo folding and post-translational modification such as N-linked glycosylation and disulphide bond formation with the help of co-ordinated actions of ER-resident molecular chaperones and folding enzymes (Ellgaard and Helenius, 2003; Braakman and Bulleid, 2011). Binding protein (BiP) is a heat shock protein 70 (HSP70) family protein in the ER that binds to nascent polypeptide chains and prevents their aggregation. N-linked glycan chains are attached on asparagine residues of nascent polypeptide chains that are being inserted into the ER lumen. Calreticulin (CRT) and its membrane-bound isoform calnexin (CNX) are lectin-like chaperones that recognize N-linked glycan chains and retain unfolded glycoproteins in the ER. ER oxidoreductase 1 (ERO1) and protein disulphide isomerase (PDI) mediate efficient formation of disulphide bonds (discussed later). This protein folding and post-translational modification machinery ensures that only mature proteins are transported beyond the ER (see Fig. 1).

Fig. 1.

Schematic drawing of the UPR pathway in plants. IRE1 and bZIP28 are sensors for ER stress. BiP is bound to the their luminal domains under unstressed conditions. When unfolded proteins accumulate in the ER, BiP is released from these domains, resulting in their activation. With its ribonuclease activity, activated IRE1 cleaved bZIP60U mRNA encoding bZIP60U protein that is anchored to the ER membrane with its transmembrane domain. To complete this cytoplasmic splicing, tRNA ligase (Rlg1) is considered to be necessary. The bZIP60S protein that loses the transmembrane domain is translocated to the nucleus where it functions as a transcription factor. IRE1 has an alternative role, which is cleavage of mRNA encoding proteins synthesized in the ER, such as secretory proteins (RIDD). bZIP28, another ER stress sensor, is translocated to the Golgi and then cleaved by proteases. S2P cleaves the transmembrane domain of bZIP28, and the cleaved bZIP28 (bZIP28n) is translocated to the nucleus and functions as a transcription factor co-operatively with bZIP60.

The ER folding machinery can be overwhelmed under unsuitable environmental conditions and results in the accumulation of unfolded proteins in the ER, a condition called ER stress (Iwata and Koizumi, 2012; Howell, 2013). ER stress occurs under various circumstances, including abiotic and biotic stress responses (discussed later), that require a large amount of secretory protein synthesis resulting in accumulation of newly translated, unfolded polypeptide chains in the ER. Pharmacological agents such as tunicamycin, an inhibitor of N-glycan chain synthesis, and DTT, a reducing agent that disrupts disulphide bond formation, both induce accumulation of unfolded polypeptide chains in the ER and therefore have been used to trigger ER stress (Martínez and Chrispeels, 2003; Iwata and Koizumi, 2005; Kamauchi et al., 2005).

The UPR signalling in plants is depicted in Fig. 1. Inositol-requiring enzyme 1 (IRE1) is the most conserved ER stress sensor in eukaryotes, including yeast, animals, and plants (Walter and Ron, 2011). IRE1 is an ER membrane protein with a sensor domain facing the ER lumen and kinase and ribonuclease domains in the cytosol. BiP binds to the sensor domain of IRE1 and keeps IRE1 in an inactive state. Under ER stress conditions, BiP binds to unfolded polypeptide chains, dissociating them from IRE1, which allows IRE1 to undergo autophosphorylation and oligomerization for activation (Kimata et al., 2004). Besides BiP-mediated regulation, it has been shown that IRE1 itself is also able to sense ER stress by directly binding to unfolded proteins (Kimata et al., 2007; Karagöz et al., 2017). Activated IRE1 mediates splicing of an mRNA encoding a bZIP transcription factor. In plants, bZIP60 mRNA is the target of IRE1 (Deng et al., 2011; Nagashima et al., 2011). After IRE1 cleaves bZIP60 mRNA, two bZIP60 RNA halves are ligated, probably by a tRNA ligase, Rlg1, to generate spliced bZIP60 mRNA (Nagashima et al., 2016). This splicing is different from the spliceosome-dependent splicing that occurs in the nucleus, and is therefore referred to as cytoplasmic splicing. Unspliced bZIP60 (bZIP60U) mRNA encodes a protein with a bZIP domain followed by a transmembrane domain that anchors bZIP60U protein in the ER (Iwata et al., 2008). Splicing removes an intron that is present before the RNA sequence encoding the bZIP domain and after the one encoding the transmembrane domain. Because this intron is 23 nucleotides in length, the spliced bZIP60 (bZIP60S) mRNA encodes a protein with the same bZIP domain and a newly translated amino acid sequence that is different from the transmembrane domain. Therefore, bZIP60S protein is able to translocate into the nucleus (Deng et al., 2011; Nagashima et al., 2011).

Plants have another ER stress sensor, bZIP28. bZIP28 is a transcription factor that is anchored in the ER membrane by virtue of the transmembrane domain, with a DNA-binding bZIP domain in the cytosol and a sensor domain in the ER lumen (Liu et al., 2007; Tajima et al., 2008). It has been reported that association/dissociation of BiP regulates the activity of bZIP28, as in the case of IRE1 (Srivastava et al., 2013). Binding of BiP to the luminal domain of bZIP28 retains it in the ER, without it being transported to the Golgi (Srivastava et al., 2012, 2013). Upon ER stress, BiP dissociates from bZIP28 and binds to unfolded proteins as in the case of IRE1, which allows bZIP28 to translocate to the Golgi where it is cleaved by site-2 protease, a metalloprotease with multiple transmembrane domains (Iwata et al., 2017). The cleavage occurs within the transmembrane domain of bZIP28 and allows the cytoplasmic portion of bZIP28 to be freed from the membrane and move to the nucleus.

Nuclear-localized bZIP60 and bZIP28 activate transcription of genes encoding proteins for the ER folding machinery through cis-elements, which are the ER stress response element (ERSE) and the UPR element (UPRE), on their promoters (Iwata and Koizumi, 2005; Liu et al., 2007; Liu and Howell, 2010). Although bZIP60 and bZIP28 activate a similar set of genes, the dependence of induction of ER stress-responsive genes on bZIP60 and bZIP28 differs to some extent. For instance, BiP3, one of the three BiP genes in Arabidopsis, is commonly used as a UPR marker gene and is highly dependent on the activity of bZIP60 because the induction of the BiP3 gene is greatly reduced in the bzip60 single mutant (Iwata et al., 2008). In contrast, inductions of BiP1 and BiP2 genes are only slightly affected in the bzip60 single mutant (Iwata et al., 2008). Transcriptional induction of ER stress-responsive genes by bZIP60 and bZIP28 is important for tolerance to ER stress because the bzip60 bzip28 double mutant exhibits high sensitivity to ER stress inducers such as tunicamycin and DTT (Sun et al., 2013).

Besides transcriptional induction of genes encoding ER chaperones and folding enzymes, IRE1 also mediates the destabilization of a wide range of mRNAs encoding secretory and membrane proteins in response to ER stress. This mechanism is called regulated IRE1-dependent decay of mRNA (RIDD) (Mishiba et al., 2013). IRE1-mediated degradation of mRNA occurs on the cytoplasmic side of the ER, on which the ribonuclease domain of IRE1 is present. RIDD-dependent mRNA destabilization reduces the amount of secretory and membrane proteins synthesized in the stressed ER. As in the case of transcriptional induction, RIDD plays a critical role in recovering from ER stress because Arabidopsis ire1 mutants exhibit much higher sensitivity to ER stress, indicating the role of RIDD in ameliorating ER stress (Nagashima et al., 2011; Mishiba et al., 2013). Interestingly, transcriptional induction of cytosolic HSP genes is observed in the ER-stressed ire1 mutants but not in the wild type, implying that RIDD deficiency during ER stress results in disturbance of the protein folding machinery in the cytosol (Sugio et al., 2009; Mishiba et al., 2013). Disruption of both IRE1 and bZIP28 activities is lethal in Arabidopsis, demonstrating that the UPR, consisting of transcriptional induction and RIDD, plays a critical role in normal growth and development in plants (Deng et al., 2013).

Oxidative protein folding in the ER

Besides the protein folding events mentioned above, the ER is also the site for post-translational modifications. One of the main post-translational modifications that occur in the ER lumen is the formation of disulphide bonds (R-S–S-R), which requires oxidation of thiol (S-H) groups. In the sense of interplay between ER stress and ROS, this function of ER presents a direct mechanistic link. Consistent with this function, the ER encounters a more oxidized milieu when compared with other cellular compartments. Measurements with redox-sensitive green fluorescent protein (GFP) demonstrated that the ER redox potential is ~100 mV higher than that of other cellular compartments (–300 mV) (Birk et al., 2013). Glutathione (GSH) is the major thiol–disulphide redox buffer in plant cells and, in addition to the GSH/GSSG ratio, the concentration of this molecule in a given compartment determines its redox potential. In other compartments of the cell, such as the cytosol, chloroplasts, or mitochondria, a very high GSH/GSSG ratio can be maintained by glutathione reductase (GR) with utilization of NADPH as a source of reducing power. However, the ER lacks GR activity, thus creating a more oxidized environment suitable for disulphide bond formation. Although there is an ample amount of GSSG in the ER lumen, cysteine disulphide bond formation proceeds slowly at the physiological pH of this compartment (pH 7.2–7.3) (Shen et al., 2013). It also requires specific mechanisms to maintain such an oxidized milieu within the reducing cellular environment. This is achieved by utilizing the oxidative power of the molecular oxygen (O₂) with the help of a series of proteins involved in thiol exchange. Thiol groups in target proteins are first oxidized by a thioredoxin-like protein, protein disulphide isomerase (PDI), forming a disulphide bridge. However, to catalyse a new round of reaction, PDI itself needs to be re-oxidized. This re-oxidation is achieved by ER oxidoreductase 1 (ERO1), which is a flavoenzyme tightly associated with the lumenal face of the ER membrane (Dixon et al., 2003). This mechanism is shown in Fig. 2. ERO1–PDI co-operation is vital for providing oxidizing equivalents to the ER lumen and maintains the ER oxidation state (Pollard et al., 1998). During this process, hydrogen peroxide (H₂O₂) is formed by ERO1 which is a direct outcome of electron flow from target proteins to O₂ (Fig. 2). PDI proteins are highly diverse in plants, which seems to be essential for specific interaction with a diverse set of target proteins. In terrestrial plants, six structurally different PDI subfamilies can be found, named PDI-A, -B, -C, -L, -M, and -S (Selles et al., 2011). For example, Arabidopsis contains six PDI-Ls (PDI1–PDI6), two PDI-Cs (PDI7 and PDI12), two PDI-Ms (PDI9 and PDI10), and one each of PDI-A, PDI-B (PDI8), and PDI-S (PDI11) isoforms (Lu and Christopher, 2008; Selles et al., 2011). Among these, members of PDI-L and PDI-M strongly localize within the ER (Yuen et al., 2013). The classical PDIs contain four domains named a-b-b'-a', in which a and a' represent thioredoxin (TRX) domains that usually display two cysteines included in a WCGHC active site motif. b and b' are structurally similar to TRX domains, but they do not contain the specific active site mentioned above. When compared with TRXs with the CxxC motif in their active site, PDIs possess a higher redox potential (around –150 mV), rendering them more suitable for disulphide bond formation (Selles et al., 2011). For the protein isomerase activity, PDIs must be in reduced form, and this is achieved by thiol–disulphide exchange with GSH in the ER lumen with transfer of two electrons (2GSH/disulphide bond) to PDI resulting in the formation of GSSG. PDIs reduced via GSH in this way can be used to repair mismatched disulphide bonds that can occur during ER stress. In yeast, GSH can move across the ER membrane via facilitated diffusion through the Sec61 translocon (Ponsero et al., 2017), which is controlled by the oxidation state of BiP (Kar2 in yeast) via ERO1 activity. Increased GSH in the lumen causes reductive activation of ERO1, triggering H₂O₂-dependent BiP oxidation resulting in inhibition of GSH transport. Therefore, in yeast, ER redox poise can be tuned by reciprocal control of GSH import and ERO1 activation. However, in plants, there is no information on such regulatory mechanisms, and this issue needs further scrutiny.

Fig. 2.

Overview of interaction among ER stress, ROS, and part of thiol metabolism in the ER. ER stress causes Ca²⁺ release, which in turn can induce NADPH oxidase (encoded by RBOH genes) activity at the plasma membrane triggering ROS signalling (Ozgur et al., 2014; Gilroy et al., 2016). H₂O₂ can move into the cell through the plasma membrane via simple diffusion or via aquaporins (Appenzeller-Herzog et al., 2016). H₂O₂ after moving into the cell is thought to regulate regulatory thiols on proteins, which include transcription factors, to transduce the stress signal (Waszczak et al., 2015). On the other hand, under ER stress, the ERO–PDI system is induced to increase folding capacity, resulting in accumulation of H₂O₂. Evidence suggests that the ER membrane is permeable to H₂O₂ (Ramming et al., 2014), but involvement of aquaporins in this case is not clear (Appenzeller-Herzog et al., 2016). H₂O₂ accumulated through these sources can cause oxidative stress if ER stress is prolonged. As a response to this ER stress-mediated oxidative stress, plant cells induce antioxidant defence enzymes (Ozgur et al., 2014) and in particular regulate their glutathione metabolism (Ozgur et al., 2017), which is intimately related to thiol metabolism. Regulation of glutathione metabolism involves an increase in glutathione biosynthesis and also apoplastic degradation of glutathione via γ-glutamyltransferases (GGTs), which induces the gamma-glutamyl cycle (reviewed in Masi et al., 2015); however, the function of this response during ER stress is not clear.

Due to its role in production of H₂O₂ during oxidative folding, flavoprotein ERO1 seems to be crucial for oxidative metabolism of the ER. Therefore, understanding its function and regulation is required to establish a connection between ROS and ER stress. As mentioned, re-oxidation of PDIs is achieved by ERO1. ero1 mutants of yeast are unable to produce oxidizing equivalents in the ER lumen and are more sensitive to DTT (Frand and Kaiser, 1998; Pollard et al., 1998). Interestingly, GSH-deficient gsh1 mutants of yeast are not compromised in the ability to form disulphide bonds (Cuozzo and Kaiser,1999). This suggests that the GSH/GSSG couple maintains the oxidation state of the ERO1–PDI system and, if needed, acts during proofreading (repair of mismatched disulphides) of protein folding but is dispensable for disulphide bond formation. Most of the eukaryotes contain two genes for ERO (Zito et al., 2010), which is no exception for plants such as Arabidopsis (AtERO1 and AtERO2). In this case, a rare exception is rice, which has only one gene encoding an ERO-like homologue (Onda et al., 2009). Arabidopsis ERO1 and ERO2 have 20 amino acid hydrophobic sequences at the N-terminal side of the protein, which is predicted to be a transmembrane domain (Dixon et al., 2003). In rice, Onda et al. (2009) have experimentally shown that this transmembrane domain in the N-terminus anchors the protein to the ER membrane, while the C-terminus faces the ER lumen. Arabidopsis EROs contain 18 cysteine residues that can be grouped into clusters, and some of these cysteines can also be classified as structural, regulatory, or active site cysteines according to their functions (Appenzeller-Herzog et al., 2008; Sevier and Kaiser, 2008). EROs are regulated with the principle of negative feedback via their regulatory cysteines. If the ER lumen becomes oxidized, oxidation of these cysteine residues inhibits ERO1 activity to prevent further oxidation and, in contrast to this, reducing conditions activate the enzyme. This type of regulation is vital for preventing overoxidation of the ER lumen because, besides oxidation of thiols, H₂O₂ produced by ERO1 is also released to the ER lumen, further oxidizing the environment (Gross et al., 2006).

The redox status of the ER glutathione pool seems to be intimately linked to the regulation and formation of disulphide bonds. Previously, in Arabidopsis, it has been demonstrated that tunicamycin-induced ER stress induces glutathione accumulation in the cell and changes its redox state (Ozgur et al., 2014). Tunicamycin treatment also regulates glutathione biosynthesis and catabolism, and induces activities of some glutathione-related enzymes in Arabidopsis (Uzilday et al., 2018) (Fig. 2). Consistent with glutathione accumulation, genes responsible for glutathione biosynthesis, GSH1 (γ-EC synthetase) and GSH2 (glutathione synthetase), were induced with tunicamycin treatment; however, surprisingly, GGT1 (γ-glutamyl transpeptidase 1), which is responsible for apoplastic glutathione degradation, was also induced. This latter finding implies activation of the γ-glutamyl cycle. On the other hand, there were no changes in the cytoplasmic glutathione degradation pathway (OXP1) (Fig. 2). It has been suggested that these two glutathione degradation pathways (extra-cytoplasmic and cytoplasmic) have distinct physiological functions (Masi et al., 2015), and during ER stress only the extra-cytoplasmic pathway was induced. Even though GGT1 gene expression was induced, glutathione levels at the apoplast were increased but not decreased, indicating that this induction was related to salvaging excess GSSG accumulated at the apoplast. Another possible scenario for enhancement of GGT1 expression under ER stress is a compensatory expression caused by blockage of the secretory pathway. In this scenario, GGT1 proteins could not be moved to the apoplast, which would cause accumulation of GSSG, and accumulated GSSG at the apoplast, via an unknown mechanism, would signal to induce GGT1 expression. However, this hypothesis needs further testing. Previously, Tolin et al. (2013) demonstrated that silencing the ggt1 gene results in a constitutive alert response with effects related to stress response, antioxidant metabolism, and photosynthesis, implying a signalling role for GGT1 and apoplastic GSSG accumulation. Since there is no mechanism to reduce GSSG in the apoplast, these findings indicate that accumulation of GSSG in the apoplast due to ggt1 silencing activates defence-related signalling pathways. Therefore, it is tempting to speculate that activation of the UPR in part shares the same signalling branches with the γ-glutamyl cycle. Besides these close associations among protein folding, glutathione metabolism, and ER stress, it has also been demonstrated that gsh2-deficient Arabidopsis mutants showed disruption of ER morphology, where swollen bodies are observed in the ER, indicating a role for glutathione and its redox buffering property in the biogenesis and maintenance of this organelle.

As a final remark about oxidative protein folding, inhibiting this function of ER has been widely exploited in the literature to create ER stress by utilization of disulphide exchangers such as DTT; however, it should be considered that DTT also creates reductive stress (Trotter and Grant, 2002) not only in the ER, but possibly in all compartments of the cell such as chloroplasts and mitochondria. This would inevitably interfere with the redox-related signalling mechanisms (retro- or anterograde) or metabolic pathways that are regulated by redox-controlled enzymes in these compartments. Therefore, it can be concluded that use of such reducing agents is not suitable in experiments in which changes to redox regulation in response to the UPR or ER stress are investigated.

Reactive oxygen species production and signalling during ER stress

Although ERO activity is strictly controlled via regulatory thiols on the enzyme, under various stress conditions its expression and activity increase. This is especially prominent under ER stress, when the ERO–PDI machinery is induced (as a part of the UPR) to form the needed disulphides, resulting in an increase of ERO-mediated H₂O₂ production. However, with the increase of possible substrates, free thiols in the polypeptides, the probabilities of forming mismatched disulphides are also enhanced, further increasing the H₂O₂ production. Due to this, it has been suggested that ERO activity during ER stress may significantly contribute to oxidative stress in animals (Harding et al., 2003; Tu and Weissman, 2004), in yeast (Liang et al., 2014), and in plants (Ozgur et al., 2014).

As mentioned above, during ER stress, GSH is utilized to repair disulphides, causing a depletion of GSH in the ER lumen. This would cause a change in glutathione redox potential, with the possiblity of further amplifying the stress signal via Ca²⁺ release from the ER since the ER is the main storage site of Ca²⁺ in cells. In animals, this can be achieved by the ER-localized Ca²⁺ channel ryanodine receptor (RyR), which is controlled by oxidation of its redox-sensitive thiols. Thiol oxidation by ROS or GSSG activates while reducing agents, such as GSH, inhibit the activity of this channel (Feng et al., 2000). In addition, the ER-resident protein (ERp44), a thioredoxin-like motif-containing protein, is highly redox regulated and its chaperone activity can block the production of ERO1α and ERO1β (Anelli et al., 2002; Hisatsune et al., 2015). ERp44 also mediates Ca²⁺ signalling in the ER, and these findings suggested that ERO1α and ERp44 are crucial for Ca²⁺ homeostasis (Anelli et al. 2012). In plants, there is no homologue of ERp44, but a thiol-mediated Bri1-5 retention system similar to ERp44 has been suggested to act as a chaperone with a possible oxidative signalling role (Hong et al., 2008). Hong et al. (2008) also argued a need to identify the functional homologue of ERp44 in plants. Another study in yeast has shown that deficiency of a glutaredoxin (GRX6) shifts redox equilibrium of the ER lumen toward a more oxidized state, influencing the intracellular calcium homeostasis. Interestingly, grx6 mutants had reduced levels of Ca²⁺ in the ER, whereas Ca²⁺ accumulation occurs in the cytosol (Puigpinós et al., 2015). Findings mentioned above suggest that the ER GSH/GSSG ratio and its redox potential might be important regulators of Ca²⁺ signalling that originates from the ER.

ROS signalling is usually associated with NADPH oxidases (Suzuki et al., 2011). Unlike their mammalian homologues, plant NADPH oxidases that catalyse the production of O₂·^– at the plasma membrane have two EF-hand domains at their N-terminal region, and binding of Ca²⁺ to at least one of these EF-hand motifs is essential to trigger O₂·^– production (Takeda et al., 2008, Oda et al. 2010). Produced O₂·^–, in turn, can further activate Ca²⁺ channels acting in a manner of a positive feedback (Takeda et al., 2008). Indeed, it has been demonstrated that ionomycin, a Ca²⁺ ionophore, activates the NADPH oxidase activity of Arabidopsis thaliana and increases ROS production. Further, NADPH oxidase activation also depends on phosphorylation by Ca²⁺-dependent protein kinases (CPKs or CDPKs) and calcineurin B-like protein-interacting protein kinases (CIPKs) (Kobayashi et al., 2007; Asai et al., 2013). Recently, it has been demonstrated that ER stress induced by tunicamycin can rapidly (within 10 min) enhance RBOHD and RBOHF expression in Arabidopsis. Moreover, inhibition of NADPH oxidase activity with diphenyleneiodonium (DPI; a flavin inhibitor) decreased the H₂O₂ accumulation, which peaked after 30 min (Ozgur et al., 2014) (Fig. 2). These findings indicate the involvement of NADPH oxidase-mediated ROS signalling after a certain point following the onset of ER stress, which is most probably related to Ca²⁺ release from the ER. Interestingly, with an experimental design that involved electron transport chain inhibitors to produce ROS in mitochondria and chloroplasts specifically, it has also been shown that ROS production in these organelles differentially induces expression of UPR- and ER stress-related genes, with low concentrations of ROS being more efficient in induction of gene expression, consistent with a signalling role (Ozgur et al., 2015). Remarkably, enhanced expression of BiP3 and ERO1 was tightly associated with H₂O₂ but not ¹O₂, indicating that H₂O₂ has a signal role to induce ERO1 (Ozgur et al., 2015) (see Fig. 3 for changes in BiP3 expression with organellar oxidative stress). These findings support the fact that there is a ‘vicious cycle’ between ER stress and ROS production inducing each other in plants, similarly to the case of animals (Malhotra and Kaufman, 2007). In agreement with this hypothesis, expression of ZAT12, a master switch transcription factor responsive to oxidative stress, was up-regulated during ER stress (Martínez and Chrispeels, 2003).

Fig. 3.

Interaction between ROS accumulation in different cellular compartments (chloroplasts, mitochondria, and apoplasts) and induction of the UPR. Expression of BIP3, an ER stress-responsive gene, is shown as a heat-map upon exposure to chemicals that inhibit electron transport at mitochondria (rotenone=Rot) or chloroplasts [methyl viologen=MV and 3-(3,4-dichlorophenyl)-1,1-dimethylurea=DCMU] and exogenous H₂O₂ (data taken from Ozgur et al., 2015 and presented as log₂ base). Exogenous H₂O₂ and mitochondrial ROS induce the UPR. In chloroplasts, low levels of ROS (O₂·^–) produced via MV induce the UPR; however, high concentrations of MV down-regulate BIP3 expression. On the other hand, ROS (¹O₂) produced via DCMU treatment down-regulate BIP3 expression at all concentrations (Ozgur et al., 2015), showing a different mode of action for O₂·^– and ¹O₂. In addition, as a possible mode of interaction between the ER and chloroplasts, the membrane hemifusion-based model suggested by Mehrshahi et al. (2013) and reviewed in Mehrshahi et al., (2014) has been presented. In this model, inner leaflets of the chloroplast outer membrane and ER membrane are fused, revealing the possibility to transfer H₂O₂ or other redox active compounds between these organelles.

Besides ROS production related to the Ca²⁺ release and NADPH oxidase activity triggered by ER stress, the H₂O₂ itself produced at the ER via ERO activity should also be considered as a possible source of ROS signal. Appenzeller-Herzog et al. (2016) hypothesized that the vast amounts of H₂O₂ produced by ERO might be a component of signal transduction. To act as a signal, H₂O₂ should be able to move out of the ER. It is thought that the ER membrane is permeable to H₂O₂ because in animal systems when GPX8, the POX responsible for detoxification of H₂O₂ produced by ERO1α, is knocked out, H₂O₂ overflows to the cytosol (Ramming et al., 2014). So, this raises the question of how H₂O₂ moves across the membrane. The simplest answer would be the diffusion of H₂O₂ through the membrane since it has no charge like O₂·^–. Besides this, it has been suggested that aquaporins might drive this H₂O₂ transport (Appenzeller-Herzog et al., 2016). In HeLa cells, AQP8 was determined to be an efficient H₂O₂ transporter through the plasma and ER membrane (Bertolotti et al., 2013). In plants, proteins such as ZmPIP1 in maize (Zelazny et al., 2007) and AtNIP2;1 in Arabidopsis (Mizutani et al., 2006) were reported to localize at the ER membrane. However, the function of these aquaporins in transportation of H₂O₂ through the plant ER membrane still needs to be elucidated.

For ROS signals to reach their destined location, the site of production and the target location should be in close proximity in the cell. In this case, an important point regarding the signal role of ER-associated H₂O₂ is the physical contacts of this organelle with other organelles such as chloroplasts or mitochondria. For example, by utilizing a transorganellar assay, Mehrshahi et al. (2013) provided evidence for a chloroplast–ER membrane hemifusion model. The model suggests that the inner leaflets of the ER membrane and plastid envelope are fused to form a bilayer, which would allow diffusion of non-polar metabolites from one compartment to the other (Mehrshahi et al., 2013) (Fig. 3). Interestingly, in intact leaves of Arabidopsis and Nicotiana benthamiana, chloroplastic extensions, the stromules, retract and move in tandem to the neighbouring ER tubules, indicating a close physical contact between these two compartments (Schattat et al., 2011). Moreover, the ER is directly involved in the biogenesis of peroxisomes (Mullen and Trelease, 2006). Peroxisomes have central roles in oxidative metabolism especially in C₃ plants due to photorespiratory H₂O₂ production (Corpas et al., 2017). Similar to stromules, peroxules, dynamic extensions of peroxisomes, are also extended along paths defined by ER tubules. Interestingly, peroxules can be induced by low levels of hydroxyl radicals or H₂O₂ (Sinclair et al., 2009). Also a portion of the peroxisomal proteins are trafficked through the ER. For example, APX is transferred through the ER to peroxisomes (Mullen and Trelease, 2006), implying that severe ER stress can cause a decline in peroxisomal APX activity. These interactions provide evidence for the establishment of physical contact points where bidirectional movement of ions, lipids, metabolites, including ROS (especially H₂O₂), or redox compounds can occur. If this is the case, the ER can send and receive redox signals to and from other organelles via exchange of redox compounds. As an example, it has been demonstrated that there is a link between MEcPP, a signal metabolite that is released from perturbed chloroplasts, and ER stress. Increased MEcPP levels induced the expression of UPR genes such as bZIP60, possibly sensitizing the cell for episodes of increased stress (Walley et al., 2015). Similar to MEcPP, tocopherols, important non-enzymatic antioxidants, might be important retrograde signals due to their role in regulation of fatty acid metabolism from the ER and their interoperability between chloroplast and ER membranes (Allu et al., 2017; Mehrshahi et al., 2014).

Hormone signalling is also tightly linked to redox regulation, and the role of ROS as a seconder messenger during hormone homeostasis is well established (reviewed in Kwak et al., 2006; Xia et al., 2015). Although the number of studies that deal with ROS and their interaction with ER stress in plants are limited, there are several studies that investigated the metabolism of different plant hormones under ER stress and their role in the induction of the UPR (Che et al., 2010; Chen et al., 2014; Nagashima et al., 2014; Meng et al., 2017). For example, Nagashima et al. (2014) found that UPR genes were up-regulated with exogenous application of salicylic acid (SA) via IRE1-mediated activation of bZIP60. Moreover, in Arabidopsis, CPR5, a negative modulator of the stress response hormone SA, physically interacts with bZIP60 and bZIP28. As a result of this interaction, SA moderates root growth via the IRE1–bZIP60 signalling pathway (Meng et al., 2017). SA is also considered as a positive regulator of bZIP transcription factors that are involved in the UPR (Meng et al., 2017). Correlation between SA and induction of the UPR should be expected due to a heavy secretory protein load during biotic stress that requires an adaptive mechanism such as the UPR. The role of ROS in the SA signalling and vice versa has been widely studied since the 1990s, and interaction between NADPH oxidases and SA signalling has also been demonstrated (Torres et al., 2005; Khokon et al., 2011). Therefore, SA can induce the UPR via both direct interaction with UPR components and ROS production via NADPH oxidases. In this case, ROS produced via NADPH oxidase might also contribute to further activation of the UPR for adapting cells to excess protein load.

Jasmonates are key regulators of the plant immune system and are generally related to wounding response. It has been demonstrated that wounding induces ER-derived bodies (Hara-Nishimura and Matsushima, 2003), indicating a change in ER metabolism, and methyl jasmonate, an active form of jasmonic acid, can regulate the production of these ER bodies in several plant species (Gotté et al., 2015). These ER bodies are thought to be important regulators of stress response (Gotté et al., 2015). Therefore, a possible connection between jasmonates and the UPR seems logical. However, Moreno et al. (2012) tested whether methyl jasmonate can induce the UPR and did not observe activated bZIP60 splicing with 30 µM methyl jasmonate treatment. Further investigations are needed to explain the metabolic relationship between jasmonates and ER stress during development of various types of biotic stresses.

Auxins are responsible for regulating diverse aspects of plant growth. However besides controlling growth, they are also vital for regulating the trade-off between growth and defence responses under stressed conditions (Park et al. 2007). This type of regulation can also be observed during ER stress. Chen et al. (2014) showed that auxin signalling is negatively regulated by ER stress, and ER-based auxin homeostasis is important for UPR activation. Inhibition of auxin receptors and transporters during ER stress was observed in the ire1 mutant, showing that IRE1 is required for the auxin responses (Chen et al. 2014). It is well known that ROS play a vital role in growth by loosening or stiffening the cell wall, and auxin induces ROS formation during root growth (Schopfer et al., 2002). These findings place IRE1 upstream of ROS, in which auxin perception and signalling act as the mediator. Therefore, it seems that UPR components can also regulate growth under non-ER stress conditions, which deserves further investigation.

Ethylene modulates various plant processes such as germination, senescence, and ROS production. Chen et al. (2002) localized an ethylene receptor ETR1 in the ER, and suggested a central role for the ER in ethylene signal transduction. Since ethylene regulates ROS signalling (reviewed in Xia et al., 2015) and its receptor is located in the ER, its relationship with the UPR needs further scrutiny.

In addition to the plant hormones mentioned above, brassinosteroids (BRs) that function in growth and development also regulate the response to environmental stress, and it has been shown that BR treatment confers plants with tolerance to salt and heat stresses (Kagale et al., 2007). BR insensitive 1 (BRI1) is a plasma membrane-localized BR receptor that triggers phosphorylation and initiates signal transduction from receptor kinases to transcription factors (Kim and Wang, 2010). Several studies demonstrated that BRI1 is involved in the UPR, endoplasmic reticulum quality control (ERQC), and endoplasmic reticulum-associated degradation (ERAD). bri1-5 interacts with CNX, BiP, and some components of ERAD (Hong et al., 2008). Che et al. (2010) also indicated that bZIP17 and bZIP28 can trigger BR signalling along with the UPR. ERAD component UBC32, a ubiquitin-conjugating enzyme, is involved in the BR-mediated salt stress response (Cui et al., 2012).

ER stress during biotic and abiotic stress and its possible interactions with ROS and cellular redox

In animals, ER stress and oxidative stress are inter-related during pathogenesis of several diseases such as diabetes, Alzheimer’s, or Parkinson’s disease (Malhotra and Kaufman, 2007; Hotamisligil, 2010; Zhang and Kaufman, 2008). In plants, environmental stress conditions such as drought, salinity, and especially heat stress can cause accumulation of unfolded and/or misfolded proteins leading to ER stress and induction of the UPR (Iwata and Koizumi, 2012; Howell, 2013). In addition, most of the proteins synthesized in response to pathogens are secretory proteins, therefore biotic stresses such as bacterial infection can also result in ER stress due to an increase in secretory load. In this section, the role of the UPR during salinity, high temperature, anoxia, and biotic stress will be discussed.

Liu et al. (2007) demonstrated that salinity causes ER stress in Arabidopsis, and the bZIP17 transcription factor, a component of the UPR, takes a role in the regulation of salt stress responses. Upon exposure to salt, AtbZIP17 is processed by site-1 protease and relocates to the nucleus to activate salt stress-responsive genes. Moreover, overexpression of bZIP60, another UPR component, induced salinity tolerance in Arabidopsis (Fujita et al., 2007). When Arabidopsis plants treated with salt were exposed to ER stress with tunicamycin, they were more sensitive to salt, and induction of UPR genes was synergistic when compared with individual treatments (Ozgur et al., 2014). It is thought that ERAD components including UBC32, HRD1/HRD3 complex, DER1, and SEL1 are needed for stress tolerance and ER lumen homeostasis. Among the mentioned proteins, the Arabidopsis hrd3a mutant was sensitive to tunicamycin-induced ER stress and, in addition, this mutant was also sensitive to salinity, demonstrating the need for an intact ERAD during salt stress (Liu et al. 2011).

Induction of the UPR might have a crucial role in perception of heat stress. Increased temperature stimulates the Ca²⁺ channels, and Ca²⁺ flux leads to activation of the signal transduction pathways including Ca²⁺ signalling, kinases, phosphatases, and ROS signalling. An increase in temperature also causes protein instability and/or problems in protein folding activating bZIP60 splicing by IRE1 and triggering the UPR (Deng et al., 2011; Mittler et al., 2012). During heat stress, in other compartments of the cell, a general UPR response independent of the ER is also induced due to protein denaturation (Mittler et al., 2012). As discussed before, during ER stress, Ca²⁺ release from the ER induces different signalling pathways including ROS signalling, suggesting that the UPR might act as a heat sensor to perceive and respond to high temperatures (Mittler et al., 2012). Similar to this, recently it has been proposed that the ER might also act as an oxygen sensor during anoxic conditions in plants (Schmidt et al., 2018). As mentioned above, the PDI–ERO system depends on the availability of O₂ for oxidizing power. Under anoxic conditions, in which O₂ is limited, the formation of disulphide bonds would be inhibited due to a decrease in ERO activity. As a result, unfolded proteins would accumulate and induce the UPR. In this case, an arm of the UPR might mediate adaptation to anoxic conditions with induction of specific transcription factors, which deserves further investigation.

Besides abiotic stresses, the UPR also takes place in response to biotic factors. Enzymatic degradation of the cell wall, mimicking pathogen attack, induced rapid BiP expression (Jelitto-Van Dooren et al., 1999). Consistent with this, ire1a ire1b double mutants and the bzip60 mutant were more susceptible to Pseudomonas syringae infection (Moreno et al., 2012). Moreover, the ability to induce systemic acquired resistance (SAR) of these mutants in response to SA treatment was decreased, which is in agreement with findings of Nagashima et al. (2014). NPR1 [non-expressor of pathogenesis related (PR) genes 1] is a master regulator of SAR, which regulates the expression of secretory proteins such as PDI, BiP2, CNX, CRT, and, in addition, antioxidant enzymes such as POXs (Wang et al., 2005), which might indicate a possible association among SAR, antioxidant defence, and the UPR. Since ROS act both up- and downstream of SA (Herrera-Vásquez et al., 2015), with our current knowledge it is not possible to discriminate the order of occurrence of SA-dependent SAR induction, ROS burst, and UPR induction in response to pathogen attack in plants.

Other reactive species similar to ROS are reactive nitrogen species (RNS) such as nitric oxide (NO) and peroxynitrite (ONOO^–), and production of these reactive species is closely related (reviewed by Del Río, 2015). Although it is not the main topic of this review, it is worth mentioning that RNS might have an important role in the induction of the UPR and ER stress, similar to ROS. In animals, there is direct evidence for a relationship between nitrosative stress and ER stress. For example, in diabetic hearts, a transcription factor that modulates cellular metabolism, oxidative stress, and immune homeostasis, also mediates nitrosative stress and ER stress responses (Guo et al., 2014). In addition, increased S-nitrosylation of PDI by transfer of NO to a thiol promoted protein misfolding and neurodegeneration (Uehara, 2007). Also, NO-mediated apoptosis is modulated by the UPR in pancreatic cells (Oyadomari et al., 2001). In plants, the relationship between nitrosative stress and ER stress also needs to be clarified.

Conclusion

In animal systems, ER stress has been extensively studied in connection with other metabolic processes such as ROS metabolism, due to their involvement in various diseases. However, in plants, research has been mainly focused on the elucidation of the ER stress perception mechanism and identification of various UPR and ERAD components (Iwata and Koizumi, 2005; Tajima et al., 2008; Liu and Howell, 2010; Nagashima et al., 2011; Srivastava et al., 2013). In terms of ROS–redox regulation–ER stress interaction, the knowledge we have in plants is still in its infancy when compared with animal and yeast systems. Similar to animal systems, in plants, ER stress increases oxidative load and can cause oxidative stress depending on the severity of ER stress (Ozgur et al., 2014). Moreover, ER stress triggers NADPH oxidase-mediated ROS signalling, induces antioxidant defence, and changes the redox status of glutathione (Ozgur et al., 2014; Uzilday et al., 2018). On the other hand, excess accumulation of ROS in different cellular compartments, especially the mitochondria, can induce the UPR (Ozgur et al., 2015). These findings indicate that there is a mutual relationship between ER stress and ROS accumulation in plants, which can be considered as a positive feedback mechanism. Supporting this idea, treatment with melatonin, a well-known antioxidant, can alleviate ER stress in Arabidopsis (Ozgur et al., 2017). In conclusion, knowledge from the literature suggests that ER stress, just like oxidative stress, is a secondary stress caused by a primary environmental constraint, and these two secondary stresses are inter-related to each other directly or via other signalling pathways. Some outstanding questions in this area are the following. (i) How does GSH move into the ER and how are its movement and redox state regulated in plants during ER stress? (ii) How does H₂O₂ transport occur across the ER membrane, what components are involved, and what is the role of organelle–organelle contact sites in transducing ROS and redox signals to and from the ER? (iii) What is the role of ROS during UPR–plant hormone interaction? Is it up or downstream of UPR; if ROS signals are disrupted would it affect hormone-induced UPR induction or UPR-mediated hormone signalling? (iv) What is the role of the UPR as a stress sensor during stresses such as high temperature and anoxia, in which protein folding is disrupted? What is the role of ROS in transducing this signal?

Acknowledgements

The research of authors from Ege University (RO, BU, and IT) was supported by The Scientific and Technological Council of Turkey (TUBITAK; grant no. 212T018). This work was also supported by The Japan Society for the promotion of Science KAKENHI (17K07450) to NK.

Author notes

These authors contributed equally to this work.