Cellular response to unfolded proteins in the endoplasmic reticulum of plants

Authors


R. Urade, Division of Food Science and Biotechnology, Graduate School of Agriculture, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan
Fax: +81 774 38 3757
Tel. +81 774 38 3758
E-mail: urade@kais.kyoto-u.ac.jp

Abstract

Secretory and transmembrane proteins are synthesized in the endoplasmic reticulum (ER) in eukaryotic cells. Nascent polypeptide chains, which are translated on the rough ER, are translocated to the ER lumen and folded into their native conformation. When protein folding is inhibited because of mutations or unbalanced ratios of subunits of hetero-oligomeric proteins, unfolded or misfolded proteins accumulate in the ER in an event called ER stress. As ER stress often disturbs normal cellular functions, signal-transduction pathways are activated in an attempt to maintain the homeostasis of the ER. These pathways are collectively referred to as the unfolded protein response (UPR). There have been great advances in our understanding of the molecular mechanisms underlying the UPR in yeast and mammals over the past two decades. In plants, a UPR analogous to those in yeast and mammals has been recognized and has recently attracted considerable attention. This review will summarize recent advances in the plant UPR and highlight the remaining questions that have yet to be addressed.

Abbreviations
ATF

activating transcription factor

BiP

binding protein

bZIP

basic leucine zipper

eIF2α

initiation factor-2α

ER

endoplasmic reticulum

ERAD

ER-associated degradation

ERSE

ER stress response element

fl-2

floury-2

GFP

green fluorescent protein

GLS

Golgi body localization sequence

GPT

UDP-N-acetylglucosamine–dolichol phosphate N-acetylglucosamine-1-phosphate transferase

IRE1

inositol-requiring enzyme-1

PCD

programmed cell death

PDI

protein disulfide isomerase

PERK

interferon-induced dsRNA-activated protein kinase-related protein

S1P

site-1 protease

S2P

site-2 protease

UGGT

UDP-glucose–glycoprotein glucosyltransferase

UPR

unfolded protein response

UPS

ubiquitin-proteasome system

XBP-1

X-box binding protein 1

Introduction

The unfolded protein response (UPR) is a fundamental system common to unicellular organisms, plants, animals, and humans, and is conserved in all eukaryotic cells. However, there are differences in the molecular mechanisms underlying the UPR between organisms. In yeast, the UPR increases the folding and degradation capacities of unfolded proteins by inducing the expression of genes related to those capacities [1]. Inositol-requiring enzyme-1 (IRE1), an endoplasmic reticulum (ER)-transmembrane protein that is activated by ER stress, splices basic leucine zipper (bZIP) transcription factor HAC1 mRNA in a nonconventional manner [2,3]. HAC1 is translated from the spliced mRNA [4–6] and subsequently activates the transcription of a group of genes possessing UPR cis-activating regulatory elements in their promoter regions [7–9]. This pathway was the first example of a protein signal that is transduced from the ER to the nucleus, and this finding opened the door to investigation of the details of UPR signaling events.

In comparison with that of yeast, the UPR of mammalian cells is a much more complicated event, in which general attenuation of translation, apoptosis, and folding or degrading of unfolded proteins occurs [10–12]. The mammalian UPR is triggered by at least three ER stress sensors, including the mammalian ortholog of yeast IRE1 [13,14], activating transcription factor (ATF) 6 [15], and interferon-induced dsRNA-activated protein kinase-related protein (PERK) [16]. IRE1 is activated during ER stress and splices invalid mRNA, similar to yeast IRE1, into the mature X-box binding protein 1 (XBP-1) mRNA, a bZIP-like transcription factor [17–20]. XBP-1 is translated from the spliced mRNA and is translocated to the nucleus to regulate transcription of target genes. In addition, IRE1 independently mediates the rapid degradation of a specific subset of mRNAs due to their localization on the ER membrane and to the amino-acid sequence they encode [21]. This response could selectively halt production of proteins that challenge the ER and could make available the translocation and folding machinery for the subsequent remodeling process. In addition, IRE1 forms a trimeric complex with phosphorylated tumor necrosis factor receptor-associated factor 2, apoptosis signal regulating kinase 1 and the c-Jun N-terminal kinase and subsequently causes cell death [11,22,23]. ATF6 is an ER transmembrane protein that senses ER stress through its luminal domain, and then moves to Golgi bodies to be cleaved. The ATF6 cytosolic domain produced as a result of this cleavage event is released from the Golgi membrane into the nucleus, where it induces the expression of target genes [24–28]. PERK is an ER transmembrane protein that senses ER stress through its luminal domain and phosphorylates a specific serine residue of translation initiation factor-2α (eIF2α), resulting in general inhibition of translation [16,29]. Phosphorylation of eIF2α also stimulates translation of ATF4 [30], a bZIP-like transcription factor that induces the transcription of many amino-acid synthetic enzymes, amino-acid transporters, and antioxidation enzymes. ATF6 and ATF4 also stimulate the transcription of CHOP, a gene important for apoptotic cell death [31].

It has recently been shown that UPR signaling not only maintains the homeostasis of the ER, but also plays an important role in nutritional and differentiation programs in healthy and unstressed yeast and mammalian cells [11,32,33]. Furthermore, organ-specific UPR signaling pathways have been identified in mammalian cells [34–37]. These findings suggest that the UPR functions during normal processes as well as during emergency situations. The UPR pathways act cooperatively such that the fate of the cell depends on the balance between the individual UPR pathways. Therefore, disturbance of these functions causes malfunction of the ER transport machinery and defective UPR signaling, resulting in diseases such as neurodegenerative disorders, diabetes, and endocrine defects [11].

The UPR in plants is an important and constantly expanding topic. However, study of the plant UPR is a relatively new field, and its molecular details are only now becoming clear. Recent developments in this field will be explored in this review.

Transcriptional regulation of UPR genes

The most prominent phenomenon induced by ER stress is transcriptional regulation of UPR genes. The induction of genes assumed to be related to the UPR in plant cells has been reported. Binding protein (BiP) is a representative UPR gene. BiP is induced in the presence of drugs that cause ER stress, such as tunicamycin [38–45]. Tunicamycin inhibits UDP-N-acetylglucosamine– dolichol phosphate N-acetylglucosamine-1-phosphate transferase (GPT), such that the initial step of the biosynthesis of dolichol-linked oligosaccharides is blocked [46]. Treatment with tunicamycin results in the inability of asparagine (N)-linked glycoproteins synthesized in the ER to be glycosylated. Transgenic Arabidopsis thaliana plants with a 10-fold higher level of GPT activity were resistant to tunicamycin at a concentration that was lethal to control plants [44]. Likewise, transgenic plants grown in the presence of tunicamycin have N-glycosylated proteins, and expression levels of BiP mRNA was lower than in control plants. These findings suggest that treatment with tunicamycin results in the generation of misfolded or unfolded proteins by inhibiting N-glycosylation and activation of the UPR. Transcription of BiP mRNA is activated by other drugs such as the proline analog azetidine-2-carboxylase, which is incorporated into nascent polypeptides and prevents their folding [47], and dithiothreitol, which inhibits formation of disulfide bonds on nascent polypeptides and prevents their folding [39].

Two comprehensive analyses of the transcriptome of A. thaliana during drug-induced ER stress have been performed using two kinds of DNA microarray methods. Martìnez & Chrispeels [48] performed experiments using an Affymetrix GeneChip with a 8297 probe set (7372 independent genes of the ≈ 27 000 protein-coding genes of A. thaliana). The UPR was induced by treating Arabidopsis plants with tunicamycin or dithiothreitol. Fifty-three genes were identified as up-regulated genes under ER stress, whereas 31 genes were identified as down-regulated genes. Kamauchi et al. [49] analyzed the transcriptome of Arabidopsis UPR genes by fluid microarray analysis of tunicamycin-treated plantlets. Using this method, target genes were cloned from selected fluid microarray beads [50], and 215 up-regulated genes and 17 down-regulated genes were identified. These genes were reanalyzed with functional DNA microarrays using DNA clones from the fluid microarray analysis. Together, 36 up-regulated genes and two down-regulated genes in all samples treated with the three drugs, tunicamycin, dithiothreitol or azetidine-2-carboxylase were recognized as UPR genes. The up-regulated UPR genes identified by the two research groups are shown in Table 1, and include ER chaperones, glycosylation/modification-related proteins, translocon subunits, vesicle transport proteins, and ER-associated degradation (ERAD) proteins. Most of these proteins are orthologs of the genes identified as being related to the UPR in yeast and mammalian cells [1,30,51–54]. In addition, genes related to the regulation of translation (P58IPK) [55] and apoptosis (BAX inhibitor 1) [56,57] were also identified as being up-regulated during the UPR in plants [49,58]. Phospholipid biosynthetic enzymes increase in expression in the maize (Zea mays) floury-2 (fl-2) mutant (described below) and soybean (Glycine max) suspension cultures when treated with tunicamycin [45], and, in yeast, a number of lipid metabolism-related genes are up-regulated by ER stress [1]. On the other hand, neither of the DNA microarray analyses of the Arabidopsis transcriptome described above detected any up-regulation of lipid metabolism-related genes, suggesting that additional experiments are needed to assess if phospholipid metabolism-related genes are related to the UPR in plant cells.

Table 1.   Genes up-regulated during ER stress. Data from [48,49] are combined. NEM, N-Ethylmaleimide; GST, glutathione S-transferase.
AGI geneDescriptioncis-Acting regulatory elementaReferences
  • a

     Numbers in parentheses show the number of elements on the promoter.

Protein folding
At1g09080BiP-likeERSE like (2), XBP1-BS-like48, 49
At5g28540BiPP-UPRE48, 49
At5g42420BiPXBP1-BS-like, P-UPRE49
At4g21180Similar to ERdj3 48
At5g61790Calnexin 1ERSE like, XBP1-BS-like48, 49
At5g07340Calnexin 2P-UPRE48, 49
At1g56340Calreticulin 1ERSE like49
At1g09210Calreticulin 2ERSE like, XBP1-BS-like48, 49
At4g24190AtHsp90–7ERSE like, XBP1-BS-like48, 49
At2g47470Similar to PDIERSE like, XBP1-BS-like48, 49
At1g77510Similar to PDIERSE like49
At2g32920Similar to PDI 48, 49
At1g04980Similar to PDIERSE like, XBP1-BS-like49
At5g58710AtCYP20-1 (cyclophilin ROC7)ERSE like, XBP1-BS-like49
Glycosylation/modification
At2g02810UDP-glucose/UDP-galactose transporterERSE like48, 49
At2g41490UDP-GlcNAc:dolichol phosphate
N-acetylglucosamine-1-phosphate transferase
ERSE like48, 49
At2g47180Putative galactinol synthaseXBP1-BS-like48
At2g41490GPTERSE like, XBP1-BS-like48
At4g15550UDP-glucose indole-3-acetate β-d-glucosyltransferase 48
Translocation
At5g50460SEC61 γ subunitXBP1-BS-like49
At1g29310Similar to SEC61 α subunitERSE like49
At2g34250Similar to SEC61 α subunit 49
At2g45070Similar to SEC61 β subunitXBP1-BS-like48, 49
At4g24920Similar to SEC61 γ subunitXBP1-BS-like48, 49
At1g27330Similar to SERP1/RAMP4ERSE like49
At1g27350Similar to SERP1/RAMP4ERSE like48, 49
At3g51980Similar to ER chaperone SIL 1ERSE like, XBP1-BS-like49
At5g03160P58IPKERSE like (2)49
At2g18190Putative AAA-type ATPase 48
At2g03120Similar to signal peptide peptidase 48
Protein degradation
At1g65040Similar to HRD1ERSE like49
At4g21810Similar to DER1 48, 49
At1g18260Similar to HRD3/SEL1ERSE like49
At5g35080Similar to OS-9 48
At2g46500Similar to ubiquitin 48
Vacuolar
At3g52190Similar to SP12p 48
At1g78920Similar to H+-pyrophosphatase 48
Translation   
At5g03160P58IPKERSE like (2)49
Vesicle trafficking
At3g07680Similar to Emp24pERSE like, XBP1-BS-like (2)49
At4g21730Similar to NEM-sensitive fusion protein 49
At1g11890Similar to vesicle trafficking proteinXBP1-BS-like48, 49
At1g62020Similar to coatomer α subunit 49
At1g09180Similar to SAR1BERSE like48
At4g01810Similar to SEC23pXBP1-BS-like48
PCD
At5g47120BAX inhibitor 1ERSE like49
Kinase
At1g08650Putative calcium-dependent protein kinaseERSE like48
Transcription factor 
At3g24050GATA-1 48
At1g56170Hap5b 48
At2g38470WRKY-33 48
At5g08790ATAF2ERSE like48
At5g59820Zat12 48
Stress protein 
At5g16660HSP-like (D2T2)ERSE like48
At1g67360Putative stress-related proteinXBP1-BS-like48
Unclassified
At2g25110Similar to stromal cell derived factor-2 48, 49
At5g09410Similar to anther ethylene-up-regulated
 calmodulin-binding protein ER1
ERSE like, XBP1-BS-like49
At4g12720Similar to growth factor protein with
 mutT domain
 48
At4g19880GSTERSE like48
At2g16060Similar to AHB1 48
At4g26400Putative ring zinc finger protein 48
At4g14430Carnitine racemase-like proteinERSE like48
At1g07670ER-type calcium transporter ATPase 4ERSE like, XBP1-BS-like48
At5g39580Peroxidase ATP24a 48
At4g10040Cytochrome cERSE like48

Signal-transduction-related proteins such as protein kinases and transcription factors are also up-regulated during the plant UPR. WRKY33 and ATAF2 were identified as repressors of the signal-transduction pathway activated in response to pathogens [59,60]. Zat12 enhances the expression of oxidative-stress and light stress-response transcripts and plays a central role in reactive oxygen and abiotic stress signaling [61], implying that the UPR signal-transduction pathway connects other stress signaling pathways. Genes regulated by other transduction pathways connected with UPR signal transduction may eventually be identified as being either up-regulated or down-regulated after treatment with drugs that induce ER stress. The role of these genes under these circumstances remains to be elucidated in plants.

There are discrepancies in the identification and analysis of genes down-regulated during ER stress obtained from the two DNA microarray assays described above. Thirty-one down-regulated genes were identified using the Affymetrix GeneChip, and among them, 29 genes were predicted to encode proteins containing signal peptides. Lowering the threshold of detection from 2.5-fold to 2-fold inhibition increases this amount to 129 independent genes. Among these genes, 82% of the encoded proteins have signal peptides. On the other hand, only two down-regulated genes, vegetative storage proteins Vsp1 and Vsp2, were identified by the fluid microarray method. Both of these proteins also have a signal peptide. In mammalian cells, expression of abundant genes is repressed during ER stress depending on IRE1 but not on XBP-1. Repression of these genes is fast compared with expression changes mediated by XBP-1. Furthermore, functional signal sequences of proteins encoded by down-regulated genes are required for this repression event to occur. Taken together, it is possible that IRE1-mediated mRNA degradation occurs during cotranslational translocation [21]. The fact that more than 80% of the encoded proteins in Arabidopsis with down-regulated expression during ER stress have signal peptides raises the possibility that similar systems may function in plant cells.

In both DNA microarray analyses, only the genes that complied with certain restrictive criteria were designated UPR genes, implying that some UPR genes were missed during the analysis as a result of these criteria. Thus, genes expressed at very low levels might have been unintentionally eliminated from the analysis because of difficulty in assessing differences in their expression levels. For example, AtbZIP60, which was not designated a UPR gene by DNA microarray analysis, is induced in response to ER stress as detected by Northern blot and RT-PCR analyses [62]. It is expected that genes identified by the DNA microarray analyses will eventually be confirmed by other methods such as mRNA quantification and promoter analysis.

A pivotal role of the UPR is to maintain ER homeostasis. Therefore, the presence of mutated proteins that are unable to fold into their native conformation in the ER induces the UPR in an effort to restabilize the ER environment. Many examples of this phenomenon have been described in yeast and mammalian cells, and few examples have been found in plants. For example, maize high-lysine starchy endosperm (opaque) mutants are characterized by a decrease in the accumulation of storage proteins in the ER and by alterations in protein body morphology in their endosperm. The opaque mutants fl-2 and defective endosperm B30 have a defective signal peptide in the 24-kDa α-zein and the 19-kDa α-zein endosperm storage proteins, respectively. These mutant proteins are translocated into the lumen of the ER, but remain anchored to the membranes through the noncleaved signal peptide [63,64]. A decrease in the expression of α-zein is accompanied by an increase in the level of b-70, a water-soluble maize BiP ortholog associated with both the ER and protein bodies [64–70]. The increase in maize BiP mRNA and corresponding protein concentrations in mutants compared with those of wild-type maize was endosperm-specific and inversely proportional to changes in mutant zein synthesis [66]. The pattern of gene expression in normal and the seven opaque mutants o1, o2, o5, o9, o11, Mc and fl-2, protein synthesis of which is the molecular basis of the mutation, was assayed by profiling endosperm mRNA transcripts with an Affymetrix GeneChip containing more than 1400 selected maize gene sequences [71]. Compared with normal maize, alterations in the gene expression patterns of the opaque mutants were pleiotropic, where the expressions of BiP, protein disulfide isomerase (PDI), calreticulin, GRP94 and cyclophilin, and other physiological stress-related genes were increased in the opaque mutants. The transcriptional response in fl-2 may be induced by the UPR, as the change in the pattern of gene expression was restricted to the endosperm in which the mutant α-zein was synthesized. The expression pattern of o2 and fl-2 depends on the molecular basis of the mutation. It remains necessary to evaluate the relationship between the expression patterns and the molecular basis of each mutation in the other mutants before a complete understanding of how these mutants affect ER homeostasis in plants will be obtained.

Signal transduction during the UPR

Transcription of genes related to the UPR is controlled by the specific transcription factor that binds to the cis-acting regulatory element on the promoter of a UPR gene. Many experiments have revealed the details of the signal-transduction mechanism by which yeast and mammalian cells adapt to ER stress [10,11,72,73]. In yeast, a 22-bp segment in the promoter of KAR2 (yeast BiP) was identified as the first regulatory element responding to ER stress [7–9], and the sequence CAGCGTG within this 22-bp segment was identified as the minimal regulatory element and named UPRE (UPR cis-acting regulatory element). HAC1 produced from mRNA spliced by IRE1 binds to the UPRE and induces the transcription of UPR genes [4,5]. In mammalian cells, bZIP-like transcription factors XBP1 [17–20], ATF6 [15], ATF4 [30], ATF3 [74], CHOP [75], nuclear factor-erythroid 2-related factor 2 [76], OASIS [35], CREB-H [36] and Tisp40 [37] function under ER stress. These transcription factors bind to one or more cis-acting regulatory elements and activate or repress the transcription of target genes. More than 10 types of cis-acting regulatory elements that respond to ER stress are known in mammals [11]. Among them, ER stress response element (ERSE) and ERSE-II are targets for both ATF6 and XBP1 [15,77–79]. ATF6 is constitutively synthesized as a type II transmembrane protein in the ER [24]. When the ER-membrane-bound precursors of ATF6 are cleaved by the serine protease site-1 protease (S1P) and the metalloprotease site-2 protease (S2P) in response to ER stress, the N-terminal halves become soluble transcription factors. These soluble factors are translocated into the nucleus and bind to ERSE and ERSE-II [24–28]. ERSE controls the expression of ER-localized molecular chaperones [80,81]. Transcription from another cis-acting regulatory element, XBP1-BS, is entirely controlled by XBP1, and induces expression of components of the ERAD system [80,81]. In plants, cis-acting regulatory elements that respond to ER stress have also been discovered. The soybean BiP paralog genes gsBIP6 and gsBIP9 have domains similar to ERSE and ERSE-II in their 5′ flanking sequences that are responsive to treatment with tunicamycin [82]. Similarly, a 24-bp sequence in the 5′ flanking sequences of Arabidopsis BiP is crucial for gene induction by tunicamycin [83]. This 24-bp sequence is called P-UPRE and contains two overlapping elements similar to mammalian ERSE-II and XBP-BS. Putative cis-acting regulatory sequences similar to ERSE, XBP1-BS, and P-UPRE are found at high frequencies (> 65%) in the 5′ flanking sequences of the Arabidopsis UPR genes identified by the DNA microarray analyses (Table 1).

Novel transcription factor AtbZIP60 has been identified as a member of the plant UPR signal-transduction pathway. To date, every transcription factor related to the UPR in mammals and yeast is bZIP-like. Hence, Iwata & Koizumi [84] analyzed transcripts of 75 putative bZIP transcription factors in the Arabidopsis genome. Among them, only AtbZIP60, a factor that is induced by treatment with tunicamycin, dithiothreitol and azetidine-2-carboxylase, activates transcription from P-UPRE and ERSE elements. The AtbZIP60 gene encodes a predicted type II transmembrane protein of 295 amino acids with an N-terminal bZIP DNA-binding domain, a putative transmembrane domain, and a 56-amino-acid small C-terminal domain (Fig. 1A). A truncated form of AtbZIP60 lacking the transmembrane domain (AtbZIP60 ΔC) fused with green fluorescent protein (GFP) localized to the nucleus. In other experiments, AtbZIP60 ΔC clearly activated both P-UPRE and ERSE-like sequences in a dual luciferase assay using protoplasts of cultured tobacco (Nicotiana tabacum) cells. Therefore, AtbZIP60 is considered to be a transcription factor responding to ER stress, where AtbZIP60 ΔC induces the expression of AtbZIP60 through ERSE-like sequences present in the promoter of AtbZIP60. In contrast, wild-type AtbZIP60 is unable to activate ERSE-like sequences and P-UPRE, probably because it is anchored to the membrane. This suggests that native AtbZIP60 may be released from the membrane into the cytosol during ER stress to act as a transcription factor in the nucleus (Fig. 2). In the Arabidopsis genome, the At4g20310 gene encodes a membrane protein analogous to S2P, but it remains to be confirmed whether AtbZIP60 is cleaved and released from the membrane during ER stress. In addition, no conserved sequence necessary for cleavage by S1P and S2P has been identified near the putative transmembrane domain of AtbZIP60, suggesting that it is possible that AtbZIP60 is released by an unknown intramembrane proteolysis event unique to plant cells.

Figure 1.

 Comparison of the primary structure of ATF6 and Arabidopsis bZIP60 (A) and of yeast IRE1, Arabidopsis IRE1-1 (AtIre1-1) and Arabidopsis IRE1-2 (AtIre1-2) (B). The black bar represents the region required for oligomerization. The dotted bars represent regions that interact with BiP. TAD, Transcriptional activation domain; TM, transmembrane domain; SP, signal peptide. Arrows indicate the positions cut by S1P and S2P.

Figure 2.

 Model of ER-stress signaling pathways in plants. Question marks indicate incompletely understood relationships.

It is not known how AtbZIP60 senses ER stress. Two Golgi body localization sequences (GLS1 and GLS2) were identified in the ER-luminal domain of ATF6 [85]. ATF6 localizes to the ER through interaction between GLS1 and BiP. In the absence of BiP, ATF6 is constitutively transported to the Golgi bodies. Thus, when unfolded proteins sequester BiP from GLS1 under ER stress, ATF6 is transported into the Golgi body to become a substrate for S1P and S2P. However, because the luminal domain of AtbZIP60 is much smaller than that of ATF6 (Fig. 1A), it remains unclear whether it functions as a sensor for ER stress in a manner similar to ATF6. Investigation into the cellular localization of AtbZIP60 will probably clarify these issues.

Orthologs of IRE1 have been identified in Arabidopsis (AtIre1-1 and AtIre1-2) and rice (Oryza sativa) (OsIre1) [86–88]. Fusion proteins of AtIre1-1, AtIre1-2 or OsIre1 with GFP expressed in tobacco By2 cells localize to the perinuclear ER. The expression patterns of AtIre1-1 and AtIre1-2 have been examined with fusion genes of their promoter and a reporter gene. The expression of AtIre1-1 is restricted to certain tissues at specific developmental stages such as the apical meristem, the leaf margins where vascular bundles end, the anthers before pollen is formed, the ovules at an early stage of development, and the cotyledons immediately after germination. AtIre1-2 is generally expressed in plants. The C-terminal cytosolic domain of IRE1ps is conserved among a variety of organisms (Fig. 1B). The C-terminal halves of recombinant AtIre1-2 and OsIre1 have autophosphorylation activity. When Lys442 of AtIre1-2 was mutated to Ala, this activity was lost. The N-terminal luminal domains of AtIre1-1, AtIre1-2 and OsIre1 function as ER stress sensors in yeast cells, although the amino-acid sequences of these N-terminal domains are dissimilar from that of yeast IRE1. Thus, when chimeric genes were created by fusing the N-terminal domains of AtIre1-1, AtIre1-2 and OsIre1 with the C-terminal domain of yeast IRE1, and were introduced into a yeast ΔIre1 mutant, treatment with tunicamycin no longer inhibited growth, and treatments with tunicamycin or dithiothreitol induced the UPR [86,88].

Yeast and mammalian IRE1 function as a sensor to ER stress through a process involving homodimerization and autophosphorylation. The luminal domain has a BiP-binding site in a region neighboring the transmembrane domain, and dissociation and association of BiP with this domain regulates the activation of IRE1 [89–91]. Thus, IRE1 is inactive when its luminal domain is bound by BiP. Upon accumulation of unfolded proteins in the ER, BiP is competitively titrated from the luminal domain of IRE1 by the abundant unfolded proteins in the ER lumen, and IRE1 is activated. Structural studies of the luminal domains of yeast and human IRE1 show that dimerization of luminal domain monomers creates a major histocompatibility complex-like groove at the interface [92,93]. However, it remains unknown if plant IRE1 orthologs function as regulators of transcription during ER stress, but it is possible that BiP plays an important role in sensing unfolded proteins in the ER, as overexpression of BiP in tobacco cells results in a decrease in the UPR induced by tunicamycin [94].

Plant ER is different from animal ER, in that it is continuous throughout the entire plant by way of the plasmodesmata network [95]. Certain stress signals, such as an attack by a pathogen, are transmitted throughout the plant, giving rise to systematic induction of specific genes through this continuity of the ER. However, the UPR is restricted to the cells where the stress was initiated and cannot induce a systemic response in plants, as transcription of BiP mRNA was found to be restricted to leaves treated with tunicamycin [96].

Enhancing cellular quality control systems by the UPR

Folding

Folding of nascent polypeptides in cells is not as efficient as was once thought. More than 30% of the nascent polypeptides are assumed to be degraded as junk products before being folded into their proper conformation in the cytosol of animal cells [97]. Nascent polypeptides produced in the ER are presumed to undergo a similar fate. However, folding of polypeptides translocated into the ER lumen may fail more often than that of the polypeptides in the cytosol because these folding events require more complicated steps such as glycosylation and/or formation of disulfide bonds. Therefore, the UPR is considered to be weakly but constitutively activated and maintains the homeostasis of the ER even in apparently unstressed cells. In particular, developmental events associated with high secretory activity are predicted to induce the UPR [98,99]. The quality control of proteins includes the folding of nascent polypeptide chains into their native conformation, post-translational modifications important for proper folding, and the degradation of misfolded proteins and nonassociated subunit proteins. Enhancement of folding is accompanied by induction of ER-localized molecular chaperones and foldases (PDI-related proteins). In Arabidopsis, mRNA of BiP, the SIL1 homolog, cyclophilin, GRP94 and PDI-related proteins are up-regulated by the UPR as described above. BiP is best characterized by its role in protein folding and assembly [100,101]. In addition, BiP plays an essential role in maintaining the permeability barrier of the ER translocon during early stages of protein translocation [102], targeting misfolded proteins for proteasomal degradation [103,104], sensing ER stress [85,89], and contributing to the ER calcium stores [105]. Most of these functions require its ATPase activity, where in the ATP-bound state, BiP is in an ‘open’ form that binds and releases unfolded substrates rapidly. Hydrolysis of ATP drives it to the ADP-bound or ‘closed’ state, thus stabilizing its association with unfolded proteins. The release of ADP and the rebinding of ATP reopens the substrate-binding domain to release and fold the nascent protein. SIL1 is a cochaperone of BiP and regulates its ATPase cycle by stimulating ATP hydrolysis and accelerating the ADP–ATP exchange [106].

Proline can exist in either the cis or trans form in a polypeptide chain, and its orientation dramatically influences the secondary structure of the protein. Peptidyl-prolyl-cis-trans isomerases (cyclophilin) survey the status of the proline residues and rearrange them from the cis to the trans form to ensure proper folding of the nascent polypeptide chains. Twenty-nine genes encoding cyclophilin family members are present in the Arabidopsis genome, and five gene products are assumed to be targeted to the ER lumen with N-terminal signal peptides [107]. Among them, ATCYP20-1 is up-regulated during ER stress, and contains a domain essential for peptidyl-prolyl-cis-trans isomerase activity.

Four PDI-related genes are up-regulated during ER stress. PDI catalyzes the formation and rearrangement of disulfide bonds between correct pairs of Cys residues in nascent polypeptide chains in the ER [108]. PDI and related proteins are characterized by thioredoxin motifs within their primary structure [109,110]; Arabidopsis PDI-related proteins, the expression of which is induced during ER stress, have two of these motifs. A comprehensive search of the Arabidopsis genome identified 22 orthologs of known PDI-like proteins [111]. PDI purified from plants or recombinant PDI-related proteins expressed in Escherichia coli have protein disulfide oxidoreductase activity [38,112–116], and their importance in protein folding has been demonstrated in rice endosperm [117]. In endosperm of rice esp2 mutants lacking PDI, a precursor of the storage protein proglutelin forms aggregates with other storage proteins via interchain disulfide bonds within the ER lumen, whereas in wild-type rice, proglutelins are processed normally into acidic and basic subunits and accumulate in protein storage vacuoles. In soybean cotyledon, PDI-related proteins GmPDIS-1 (an ortholog of At2g47470) [116] associates with a precursor of the storage protein glycinin in the ER, suggesting that the PDI-related protein participates in glycinin folding. Yeast and mammalian PDI are activated by the FAD-dependent oxidases ERO1 and Erv2p [118–121]. Similarly, the Arabidopsis genome encodes an ERO1 homolog, At2g38960, and an Erv2p homolog, At1g15020 or At2g01270, but so far the plant varieties have not been characterized. Mammalian PDI not only folds polypeptides, but it also aggregates unfolded proteins via disulfide bonds for retention in the ER lumen [122], and reduces aggregated proteins before retro-translocation into the cytosol for degradation [123]. No evidence for the function of PDI proteins in plants has been reported.

The high-capacity calcium-binding proteins, calnexin (an ER transmembrane protein) [124,125] and calreticulin (an ER luminal protein) [126,127], are molecular chaperones in mammalian cells specific for unfolded N-glycosylated proteins [128]. The first step in the N-glycosylation of a protein is the transfer of a core glycan Glc3Man9GlucNac2 from a membrane-bound dolichol phosphate anchor to consensus Asn-X-Ser/Thr residues in the polypeptide chain. The glucose residues on the transferred core glycan are sequentially trimmed to Glc1Man9GlucNac2 by β-glucosidase I and β-glucosidase II. The monoglucosylated glycan on the polypeptide chain is trapped by calnexin or calreticulin to protect it from degradation, resulting in retention of the polypeptide in the ER for folding [129,130]. The monoglucosylated form of the unfolded protein shuttles through cycles of deglucosylation by β-glucosidase II and reglucosylation by UDP-glucose–glycoprotein glucosyltransferase (UGGT), which preferentially recognizes unfolded glucosylated glycoproteins [131]. This process is called the calnexin/calreticulin cycle, and is one arm of the quality control machinery in the mammalian ER. It is possible that interaction between monoglucosylated N-glycan with calnexin/calreticulin functions for the quality control of N-glycosylated proteins in plants, although the calnexin/calreticulin cycle remains to be elucidated in plants. However, circumstantial evidence supports the idea that the calnexin/calreticulin cycle is present in plant cells [132]. For example, it has been shown in in vitro translation systems with wheat germ extract and bean microsomes that the rate of phaseolin assembly is accelerated when a glucosidase inhibitor is included to stop glucose trimming of the N-glycan [133]. In this system, phaseolin with partially trimmed glycans was unable to assemble into trimers, probably because of being trapped by calnexin or calreticulin. In kaiware radish (Raphanus sativus), the glucosidase inhibitors castanospermine and deoxynojirimycin suppressed the growth of seedlings by inhibiting glucose trimming of the N-glycan [134,135], and, in Arabidopsis, homozygous deletion of β-glucosidase I by T-DNA tagging is lethal [136]. In potato, curled leaves and low yields have been reported when expression of the β-glucosidase II gene MAL1 was knocked-down by antisense RNA [137]. Furthermore, the knock-down of MAL1 caused an increase in the expression of BiP, suggesting the presence of ER stress. In Arabidopsis rsw3, a temperature-sensitive mutant of the β-glucosidase II β-subunit, some morphological abnormalities and growth impairments were observed [138]. As trimming glucose residues of N-glycan by β-glucosidase I and β-glucosidase II is a prerequisite for modification of the ER-type glycan to the complex glycan in Golgi bodies, it is possible that the impairment of this process is responsible for the adverse effects on plant morphology. However, this explanation may be unlikely, as neither growth inhibition nor reproduction defects have been observed in Arabidopsis mutants defective in GlcNAc-transferase I, which catalyzes the first modification reaction to the complex-type glycan [139].

UDP-glucose, the substrate for re-glucosylation of N-glycan by UGGT, is synthesized in the cytosol, indicating that a UDP-glucose transporter would be required for the calnexin/calreticulin cycle. AtUTr1 from Arabidopsis is an ER-localized membrane protein, the expression of which is induced by treatment with dithiothreitol [140], and is recognized as a UDP galactose/glucose transporter [141]. In addition, up-regulation of the ER chaperones, BiP and calnexin, has been observed in an AtUTr1 insertional mutant, suggesting that these plants may constitutively activate the UPR. Taken together, it is possible that the calnexin/calreticulin cycle discriminates between folded and unfolded glycoproteins in plant cells. In mammalian cells, the recognition of the unfolded glycoproteins by calnexin/calreticulin is coupled with the formation of disulfide bonds, where the PDI-related thiol-oxidoreductase, ER-60/ERp57, interacts with the P domain of calnexin or calreticulin to fold N-glycosylated proteins [142–144]. The amino-acid sequence of the P domain of plant calnexin and calreticulin is highly conserved compared with that of its mammalian counterparts [145,146]. However, it is not known whether plant calnexin or calreticulin cooperates with any plant PDI-related oxidoreductase to form disulfide bonds in N-glycosylated proteins.

Degradation of unfolded proteins

Unfolded proteins generated in the rough ER are predominantly degraded by ERAD in yeast and mammalian cells [147], requiring that the unfolded polypeptides be transported across the ER membrane into the cytosol via a translocon located on the ER membrane [148] to be degraded by the cytoplasmic ubiquitin-proteasome system (UPS) [149].

In plants, misfolded storage proteins generated in the ER are degraded by an unidentified system [150,151]. However, it has been proposed that both ERAD and a vacuolar system may degrade the unfolded proteins generated in the rough ER, although the details of this mechanism have not been established.

In plants, UPS-dependent and UPS-independent ERAD-like degradation have been observed. Ricin is a heterodimeric ribosome-inactivating protein that accumulates in castor beans (Ricinus communis). The mature ricin comprises a catalytic A chain and a B chain linked by a single disulfide bond. The ER-targeted A chain is degraded by a pathway that closely resembles ERAD when expressed in tobacco protoplasts in the absence of a B chain [152]. The degradation of ricin A chain is brefeldin A-insensitive and is inhibited by the proteasome inhibitor clasto-lactacystin β-lactone, resulting in the accumulation of ricin A chains. These stabilized ricin A chains are partly deglycosylated by a peptide–N-glycanase-like activity. Taken together, these results indicate that the ricin A chain behaves as a substrate of the ERAD where it is exported into the cytosol, deglycosylated, and degraded by the proteasome [153,154]. A mutant of barley (Hordeum vulgare) mildew resistance O protein-1 is also degraded by UPS-dependent ERAD in plants [155]. Individual mutant mildew resistance O protein-1 proteins with single amino-acid substitutions in its seven-transmembrane domain exhibit markedly reduced half-lives, are polyubiquitinated, and can be stabilized through inhibition of proteasome activity. When the mutant mildew resistance O protein-1 is transfected into Arabidopsis plants previously transfected with dominant negative mutants of the putative AAA ATPase AtCDC48A/p97 (a component of the ERAD machinery) [156,157], the degradation of the mutant mildew resistance O protein-1 is impaired. This strongly suggests that mildew resistance O protein-1 is an endogenous substrate of a UPS-dependent ERAD-related quality control mechanism in plants.

In plants, several misfolded proteins are translocated across the ER membrane to the cytosol and degraded by an unknown UPS-independent system. The C-terminal extension mutant of phaseolin transfected into tobacco protoplasts is degraded very rapidly in a brefeldin A- and proteasome inhibitor-insensitive manner [158], suggesting that it is performed in a pre-Golgi compartment, probably in the cytosol. Likewise, when both endogenous and recombinant cell wall invertases are synthesized without their N-glycans in BY2 tobacco cells, they both degrade very rapidly [159]. This degradation does not occur in an acidic compartment and is also insensitive to brefeldin A and proteasome inhibitor. Furthermore, a fusion protein consisting of misfolded N-terminally truncated calreticulin with GFP is also retrotranslocated from the ER lumen to the cytosol and is subsequently degraded [160,161]. The dislocated fusion proteins accumulate in the nucleoplasm in a microtubule-dependent manner and are degraded very slowly by an unknown UPS-independent system. These UPS-independent ERAD-like degradations are unique in plants. However, any underlying molecular mechanism of the system remains unknown.

Some genes relevant to the translocation of misfolded proteins across the ER membrane into the cytosol are induced during ER stress in Arabidopsis (Table 1). SEC61 subunits form the specific translocon required for retro-translocation of misfolded polypeptides [162]. Stress-associated ER protein 1 (SERP1)/Ribosome-associated membrane protein 4 (RAMP4) orthologs are also up-regulated during ER stress. SERP1/RAMP4 interacts with the SEC61 α-subunit, the SEC61 β-subunit, and calnexin [163,164]. This complex stabilizes membrane proteins in the ER membrane through a translocational pausing mechanism [165]. P58IPK was previously implicated in translational control (described below). Recently, the novel role of mammalian P58IPK in the control of the translocation of newly synthesized polypeptides to the ER lumen was reported by Oyadomari et al. [166]. P58IPK associates with SEC61, recruits HSP70 chaperones to the cytosolic face of SEC61 and associates with translocating polypeptides during ER stress. In P58IPK-knockout mice, cells with a high secretory burden are markedly compromised in their ability to cope with ER stress. On the basis of these results, P58IPK is thought to be a key mediator of cotranslocational ER protein degradation, and probably contributes to ER homeostasis in stressed cells.

Genes that stimulate vesicle transport from the ER to the cis-Golgi are induced during ER stress in Arabidopsis (Table 1). Among them, EMP24, SAR1B and SEC23 are shown to make a complex with subunits of the COPII coat, which are key molecules for export of proteins from the ER, and promote transport of newly synthesized proteins from the ER into ER subdomains or Golgi in yeast [167–170]. Newly synthesized proteins that do not fold correctly in the ER are targeted for ERAD through distinct sorting mechanisms; soluble luminal ERAD substrates require ER–Golgi transport and retrieval for degradation, whereas transmembrane ERAD substrates are retained in the ER [169]. Retained transmembrane proteins are often sequestered into ER subdomains containing BiP. Sequestration and degradation of membrane proteins is disrupted in a mutant yeast strain lacking guanine-nucleotide exchange factor SAR1, SEC23 or SEC13 [170]. Therefore, it has been proposed that SAR1/COPII-mediated sorting of membrane proteins into ER subdomains is essential for its entry into the proteasomal degradation pathway. In plants, a similar sorting system for membrane proteins is presumed to function.

In the plant UPS-dependent ERAD system, HRD1 complex-like machinery may play an important role in the elimination of misfolded proteins. Putative orthologs of the constituents of a yeast ERAD system, HRD1, HRD3/SEL-1 L, DER1 and YOS9, are induced during ER stress in Arabidopsis (Table 1). These components constitute the HRD1 complex, which functions in recognition and ubiquitination of proteins with misfolded ER-luminal domains (ERAD-L) and proteins with misfolded intramembrane domains (ERAD-M) in yeast [171–173]. HRD1 is an E3 ubiquitin ligase, specialized for ERAD-L and ERAD-M, which catalyzes the final reaction of ubiquitination of misfolded proteins. HRD1 is stabilized by forming a complex with HRD3/SEL-1 L [174]. HRD3/SEL-1 L is a type I transmembrane protein equipped with a large luminal domain that recognizes proteins that deviate from their native conformation [173]. DER1 is a small, membrane-bound protein, the function of which remains unclear, but its deletion abolishes degradation of misfolded proteins in yeast [175]. Remarkably, maize DER1-like gene (Zm Derlins) is capable of functionally complementing a yeast DER1 deletion mutant [176]. YOS9 is a member of the OS-9 protein family and shows similarity to mannose-6-phosphate receptors. It is an essential component for degradation of misfolded ER-luminal glycoproteins [177], and specifically associates with misfolded ERAD substrates [171].

ERAD is considered to be the primary disposal route for unfolded and misfolded proteins, but growing evidence suggests a vacuolar role in protein quality control. Even in plants, the vacuolar system is involved in the degradation of misfolded proteins generated in the ER. Pimpl et al. [178] demonstrated that BiP is constitutively transported to the vacuole in a wortmannin-sensitive manner in tobacco, and that it could play an active role in this second disposal route for misfolded proteins. ER export of BiP to the Golgi apparatus is dependent on COPII. BiP is transported to the lytic vacuole via multivesicular bodies, which represent the plant prevacuolar compartment. When the plant is treated with tunicamycin, a subset of BiP-unfolded protein complexes is transported to the vacuole and degraded. As this degradation process is very rapid, the transported BiP–ligand complexes in the vacuole are not detected under normal circumstances. When the route from the Golgi apparatus to vacuoles is blocked in the presence of wortmannin, BiP–ligand complexes are secreted into the medium and are subsequently detected. In tobacco seeds, a misfolded phaseolin mutant is degraded in vacuole-derived organelles, protein storage vacuoles [179]. Vacuolar disposal may function with ERAD to maximize the quality control of proteins in the secretory pathway. It is not known whether the vacuolar function is enhanced by the UPR in plants.

Other UPR in plants

The UPR is composed of three steps in mammalian cells: enhancement of the folding and degradation of unfolded proteins (described above), attenuation of translation, and apoptosis. ER stress causes translational arrest through phosphorylation of eIF2α (Ser51) by PERK, which senses ER stress through its luminal domain and leads to the degradation of ER-localized mRNAs by IRE1 [16,21,29]. In plants, however, a PERK ortholog has yet to be described, and an increase in phosphorylation of eIF2α (Ser51) and attenuation of translation has not been confirmed during ER stress [49]. Mammalian P58IPK is an inhibitor of PERK [180] which is induced at a later phase of ER stress by the XBP-1 signal transduction pathway [58]. Because deletion of P58IPK increases the amount of phosphorylated eIF2α in the cell [58], it is thought to function as a feedback regulator of translation in the later phase of ER stress. In Arabidopsis, the P58IPK gene is up-regulated and the phosphorylation of eIF2α (Ser51) is partially inhibited by ER stress [49], but translation as a whole is not affected. Induction of Arabidopsis P58IPK and a subsequent decrease in phosphorylation of eIF2α (Ser51) may increase the translational efficiency of unidentified gene(s). Alternatively, induction of P58IPK could be required for the cotranslocational degradation of ER proteins in an effort to maintain the homeostasis of the ER as described above.

The idea that programmed cell death (PCD) functions during the UPR in plants is supported by several lines of indirect evidence. van Doorn & Woltering [181] categorized plant PCD into three morphological types, including apoptotic-like PCD, autophagy, and nonlysosomal PCD. In cultured sycamore (Acer pseudoplatanus L) cells, treatment with tunicamycin induced apoptotic PCD, as indicated by nuclear morphology and DNA fragmentation [182,183]. In cultured soybean cells, inhibition of ER-type IIA Ca2+-pumps by cyclopiazonic acid induced ER stress and PCD [184]. However, the regulatory mechanism that underlies apoptotic-like PCD induced during ER stress remains unclear. Two apoptotic-like PCD-related genes, BAX inhibitor 1 [49] and Hsr203J[185], have been identified as UPR genes. BAX inhibitor 1 is a conserved integral membrane protein localized in the ER that is a pro-apoptotic member of the multidomain Bcl2 family [56,57]. In mammalian cells, BAX inhibitor 1 affords protection from apoptosis induced by ER stress by inhibiting the activation of BAX and its translocation to the mitochondria, by preserving the mitochondrial membrane potential, and by suppressing caspase activation [186]. Plant BAX inhibitor 1 is induced by stressors such as wounding and infection with pathogens [187]. It also suppresses fungal elicitor-induced apoptotic PCD in rice and barley [188,189]. Therefore, BAX inhibitor 1 is thought to be one of the key factors required for regulation of plant apoptotic PCD. However, BAX, Bcl2 and their relatives have not been found in plants, and the underlying mechanism of BAX inhibitor 1 remains unknown. The ERSE-like cis-acting regulatory element is found in the promoter region of Arabidopsis BAX inhibitor 1 gene (Table 1), suggesting that BAX inhibitor 1 may be induced by the AtZIP60 signal-transduction system during ER stress.

Hsr203J is a PCD-related serine hydrolase that is induced by ER stress and is traditionally used as a marker for PCD [190,191]. Accumulation of Hsr203J mRNA begins at 10 h and plateaus at 24 h after treatment with tunicamycin, whereas accumulation of BiP and PDI mRNA begins 2 h after treatment with tunicamycin [185]. This suggests that transcription of Hsr203J mRNA is induced by a signal-transduction system different from the UPR governing the induction of molecular chaperones during ER stress. Taken together, these data suggest that apoptotic PCD is induced in plants when ER homeostasis is not restored after stress.

Future perspectives

Plant ER is an extremely flexible and adaptable organelle, which differentiates into various types of organelle to cope with internal and external stresses and to contain the enormous number of proteins that are actively synthesized there [192–194]. Therefore, the UPR that is unique to plants is expected to function widely, although the molecular mechanisms underlying the UPR system in plants, animals, and yeast share common components. This is supported by the fact that a number of plant-specific genes are induced by ER stress, but the functional significance of their induction has not yet been established. Recent studies in yeast and mammals have highlighted the importance of the UPR in nutrient sensing and control of differentiation [11,32,33]. In diploid yeast, nitrogen starvation inhibits HAC1 splicing and induces pseudohyphal growth. As this phenomenon is repressed in strains defective in the UPR, the latter is thought to have an important underlying role in differentiation depending on nutritional conditions. Many data also support a role for the UPR in the control of nutritional and differentiation programs in the mammalian system. Under conditions of low glucose concentration, translation of proinsulin in pancreatic β-cells is repressed by activation of PERK, and the UPR controls the terminal differentiation of B-cells into antibody-secreting plasma cells. In plants, abundant unfolded storage proteins are loaded into the ER during seed development, where the UPR is presumed to enhance the ability of the ER to fold these proteins [195]. However, there is currently no experimental confirmation of this, and the role of the UPR in seed development remains to be explored in greater detail.

The ER stress-regulated genes identified by the DNA microarray analyses described in this review are valuable for understanding the plant UPR. However, these analyses may have identified either genes primarily regulated under the UPR or genes regulated by other signal-transduction systems cross-talking with the UPR. Isolation of mutants deficient in sensor proteins and transcription factors that function in UPR signal transduction will provide valuable tools for further study of the plant UPR.

Acknowledgements

The author thanks Dr Makoto Kito, Emeritus Professor of Kyoto University, for critical reading of the manuscript, valuable advice, and warm encouragement.

Ancillary