Signal transduction by IRE1-mediated splicing of bZIP50 and other stress sensors in the endoplasmic reticulum stress response of rice


(fax +81 29 838 8397; e-mail


The endoplasmic reticulum (ER) stress sensor IRE1 transduces signals by inducing the unconventional splicing of mRNAs encoding key transcription factors: HAC1 in yeast and XBP1 in animals. However, no HAC1 or XBP1 homologues have been found in plants, and until recently the substrate for plant IRE1 has remained unknown. This study demonstrates that the Oryza sativa (rice) OsbZIP50 transcription factor, an orthologue of Arabidopsis AtbZIP60, is regulated by IRE1-mediated splicing of its RNA. Despite the presence of a transcriptional activation domain, OsbZIP50 protein is not translocated into the nucleus efficiently in the absence of OsbZIP50 mRNA splicing. Unconventional splicing of OsbZIP50 mRNA causes a frame shift, which results in the appearance of a nuclear localization signal in the newly translated OsbZIP50. OsbZIP50 mRNA is spliced in a similar manner to HAC1 and XBP1 mRNAs; however, this splicing has very different effects on the translation products, a finding that shows the diversity of IRE1-related transcription factors in eukaryotes. In addition, the expression of OsbZIP50 is affected by ER stress sensor proteins OsIRE1, OsbZIP39 and OsbZIP60. ER stress-related genes differ with respect to their dependency on OsbZIP50 for their expression. The findings of this study improve our understanding of the molecular mechanisms underlying the plant ER stress response.


Endoplasmic reticulum (ER) stress is triggered by the accumulation of unfolded proteins in the ER lumen, which then activates the unfolded protein response (UPR) via intracellular signal transduction pathways (Ron and Walter, 2007). In plants, the ER stress response is involved in numerous cellular events, including environmental responses (Gao et al., 2008; Che et al., 2010; Liu and Howell, 2010b) and the massive synthesis of storage proteins required during seed development (Vitale and Ceriotti, 2004; Wakasa et al., 2011). It is also involved in the high-level production of recombinant proteins, such as pharmaceuticals, which are generated as secretory proteins in plants (Oono et al., 2010). However, early events in the plant ER stress response remain poorly understood at the molecular level (Liu and Howell, 2010b; Urade, 2007).

In yeast and animals, ER stress is sensed by IRE1, a transmembrane protein that exhibits kinase and endoribonuclease activities within its cytosolic domains (Ron and Walter, 2007; Hetz and Glimcher, 2009). The accumulation of unfolded proteins in the ER lumen leads to clustering of IRE1, autophosphorylation of the kinase domain and activation of the RNase domain (Ron and Walter, 2007; Hetz and Glimcher, 2009). In animals, activation of IRE1 is considered to depend upon the dissociation of the ER chaperone BiP from IRE1 by unfolded proteins (Ron and Walter, 2007; Hetz and Glimcher, 2009). On the other hand, the direct binding of yeast IRE1 to unfolded proteins results in the induction of IRE1 clustering (Gardner and Walter, 2011). In yeast and animals, activated IRE1 performs an unconventional cytoplasmic splicing of the mRNA, encoding the key transcription factors HAC1 and XBP1, respectively (Kawahara et al., 1997; Sidrauski and Walter, 1997; Yoshida et al., 2001; Calfon et al., 2002), which results in the appearance of an activation domain (Yoshida et al., 2001; Mori et al., 2000). In animals, ER stress is also sensed by the transcription factor ATF6, a transmembrane protein activated by ER stress-mediated proteolysis via site-1 and -2 proteases (Brown et al., 2000), as well as by protein kinase-like ER kinase (PERK), a transmembrane kinase that phosphorylates translation initiation factor eIF2, thereby reducing cellular protein synthesis as well as the protein loads entering the ER (Harding et al., 1999).

Although IRE1 orthologues and the Arabidopsis ATF6-like transcription factors, AtbZIP17 and AtbZIP28, have been identified as ER stress sensor proteins in plants (Koizumi et al., 2001; Okushima et al., 2002; Liu et al., 2007a,b), no PERK orthologue has been found. Arabidopsis AtbZIP60 was identified as a plant-specific ER stress-related transcription factor (Iwata and Koizumi, 2005), and some ER stress-induced genes show AtbZIP60-dependent expression (Iwata et al., 2008), which indicates that AtbZIP60 is a key factor in the plant ER stress response. Under unstressed conditions, there is no nuclear localization of the AtbZIP60 protein containing the predicted transmembrane domain (TMD; Iwata et al., 2008). However, a shorter AtbZIP60 protein is produced upon ER stress, and this product does relocate to the nucleus (Iwata et al., 2008). As AtbZIP60 shows a similar domain structure and behaviour to ATF6, it has been postulated that AtbZIP60 activation is regulated by proteolytic processing (Iwata et al., 2008).

Deng et al. (2011) and Nagashima et al. (2011) recently demonstrated that AtbZIP60 mRNA is a substrate of IRE1. IRE1 mediates the alternative splicing of AtbZIP60 mRNA. The spliced mRNA was not generated in IRE1 knock-out lines, and the point mutation of conserved bases of the intron interfered with IRE1-mediated mRNA splicing. The unconventional mRNA splicing led to a translational frame shift, resulting in the production of functional AtbZIP60 protein.

However, the role played by the alternative AtbZIP60 C-terminal region remains unclear, and there has been little characterization of signalling pathways involved in the IRE1-mediated ER stress responses.

This study shows that mRNA encoding OsbZIP50 (Nijhawan et al., 2008; Wakasa et al., 2011), an orthologue of AtbZIP60 in Oryza sativa (rice), is a substrate of rice IRE1. In addition, it provides new insight into the functional significance of the alternative OsbZIP50 C-terminal region, which is obtained by IRE1-mediated splicing. Furthermore, a framework for signalling pathways in the rice ER stress response is discussed, with specific reference to the relationship between OsbZIP50 and ER stress sensors.


ER stress induces unconventional splicing of OsbZIP50 mRNA

OsbZIP50 expression patterns were examined to determine whether or not its expression was similar to AtbZIP60. In unstressed rice plants, OsbZIP50 transcripts were detected in both aerial and root tissues (Figure 1a). Upon treatment with inducers of ER stress – dithiothreitol (DTT) or tunicamycin (Tm) –OsbZIP50 expression increased significantly (Figure 1b), and two different sizes of OsbZIP50 protein were detected, as observed previously with AtbZIP60 (Iwata and Koizumi, 2005; Iwata et al., 2008; Figure 1c). In unstressed plants, no OsbZIP50 accumulation was detected by immunoblotting with anti-OsbZIP50 antibody, whereas the presence of OsbZIP50 was clearly visible in ER-stressed plants (Figure 1c). To investigate the subcellular localization of OsbZIP50 under unstressed conditions, confocal microscopy and immunocytochemical analyses were performed on cells expressing OsbZIP50 with C-terminal fusions to green fluorescent protein (GFP) and haemagglutinin (HA), respectively (Figure 1d,e, respectively). Fluorescence signals were detected predominantly in the cytoplasm, although there was some overlap with the ER. In cultured cells expressing OsbZIP50-HA, OsbZIP50 was found primarily in the soluble fraction during non- and ER-stress conditions, and some OsbZIP50 was detected in the microsomal fraction (Figure 1f), suggesting some association with the ER membrane. In ER-stressed wild-type plants, immunoblotting detected two bands corresponding to OsbZIP50, and these were found primarily in the soluble fraction (Figure 1g).

Figure 1.

 Endoplasmic reticulum (ER) stress induces expression of OsbZIP50.
(a) RT-PCR analysis of OsbZIP50 transcripts in 8-day-old seedlings.
(b) Northern blot analysis of OsbZIP50 transcripts in rice seedlings following ER stress-inducing treatment.
(c) Immunoblot analysis of OsbZIP50 in rice seedlings after ER stress-inducing treatment.
(d) Confocal fluorescence imaging of cultured rice cells transiently expressing an OsbZIP50u-sGFP fusion protein. The signal peptide (SP)-mCherry-KDEL was used as an ER localization control. Scale bars: 10 μm.
(e) Confocal immunofluorescence microscopy of transfected protoplasts expressing OsbZIP50 with a C-terminal HA tag. OsbZIP50-HA and the ER were detected with anti-HA and anti-Calnexin antibodies, respectively. Secondary antibodies were labelled with fluorescent dye. Scale bars: 10 μm.
(f) Immunoblot analysis of proteins fractionated from cultured cells expressing OsbZIP50-HA. Protein extracts collected following DMSO or tunicamycin (Tm) treatments, were ultracentrifuged and separated into supernatant (S) and pellet (P) fractions.
(g) Immunoblot analysis of proteins fractionated from wild-type plants. Protein extracts collected following DTT treatments were ultracentrifuged and separated into the supernatant (S) and pellet (P) fractions. Arrowheads indicate OsbZIP50 proteins. Calnexin and cFBPase were used as controls for membrane and cytosolic proteins, respectively.

The OsbZIP50 protein with a smaller size that appeared following ER stress was initially considered to be a proteolytically processed product. However, we considered the possibility that OsbZIP50 may be processed at a pre-translational level, and that newly translated OsbZIP50 may be translocated to the nucleus upon ER stress. As shown in Figure 2a, the second intron of OsbZIP50 is located immediately upstream of the region encoding the predicted TMD. As alternative splicing might produce a translational frame shift that would change the C-terminal region of OsbZIP50, the region around the second intron was examined using reverse transcription-polymerase chain reaction (RT-PCR) analysis. In contrast to the single transcript found in unstressed plants, two unique PCR fragments were detected in ER-stressed plants (Figure 2b). Sequencing analysis revealed that the shorter fragment contained a 20-base deletion relative to the intermediately sized fragment detected in unstressed plants (Figure 2c). Additionally, the second fragment appeared to be a hybrid molecule comprising both the short and intermediate fragments (Figure S1). These results indicate that an unconventional splicing event occurs in response to ER stress. Although OsbZIP50 expression was induced by treatment with high concentrations of salt, DTT or Tm, this alternative splicing pattern was not detected in plants treated with high concentrations of salt (Figure S2).

Figure 2.

 Endoplasmic reticulum (ER) stress induces unconventional splicing of OsbZIP50 mRNA.
(a) Structures of the OsbZIP50 transcription and translation products. Arrows indicate primers used for the RT-PCR analysis in (b). The second intron of OsbZIP50 is located immediately upstream of the region encoding the predicted transmembrane domain (TMD).
(b) RT-PCR analysis of OsbZIP50 transcripts in rice seedlings after treatment with water, dithiothreitol (DTT), dimethylsulphoxide (DMSO) and tunicamycin (Tm). Arrowheads indicate ER stress-specific PCR fragments.
(c) Sequences around the splice sites in PCR fragments. A 20-base deletion (positions 650–669) was detected in the ER stress-specific fragment.
(d) Structures for the unspliced and spliced forms of OsbZIP50. Splicing causes a frame shift, which results in the replacement of the C-terminal region.
(e) Comparison of native and recombinant unspliced (U) and spliced (S) OsbZIP50 forms. V indicates vector control. For SDS-PAGE fractionation under the same conditions, recombinant proteins expressed in Escherichia coli were mixed with extract from unstressed plants.

This ER stress-induced splicing event results in a frame shift within the OsbZIP50 C-terminal region. In the unspliced form, this region is predicted to contain a TMD (Figure 2d). Comparison between spliced and unspliced recombinant proteins expressed in Escherichia coli demonstrated that the short and long proteins detected in ER-stressed plants corresponded with the spliced (OsbZIP50s, 261 amino acids) and unspliced forms (OsbZIP50u, 304 amino acids), respectively (Figure 2e).

OsIRE1 mediates the unconventional splicing of OsbZIP50 mRNA

The spliced region of OsbZIP50 mRNA did not appear to exhibit typical characteristics of spliceosome-mediated splicing, which obey the GU-AG rule at splice junctions (Figure 3a). This spliced region was predicted to form the bipartite stem-loop structures observed previously in spliced regions of yeast HAC1, animal XBP1 and Arabidopsis AtbZIP60 mRNAs (Kawahara et al., 1997; Sidrauski and Walter, 1997; Yoshida et al., 2001; Calfon et al., 2002; Deng et al., 2011; Nagashima et al., 2011; Figures 3b and S4). These characteristic structures are reported to be target sites for the IRE1 endoribonuclease domain, which suggests that OsbZIP50 mRNA may be a candidate substrate for rice IRE1. To confirm that OsIRE1 is involved in the unconventional splicing of OsbZIP50, RNA interference was used to generate transgenic rice expressing reduced levels of OsIRE1 mRNA (Figure S3a). When wild-type plants were treated with DTT, the transcripts and translation products of the spliced form were found, whereas they were not detected in the knock-down lines (Figures 3c,d and S3a). Upon analysis of OsIRE1 overexpression lines (OsIRE1 OE lines; Figure S3b), in which constitutive activation of OsIRE1 is expected, the same unconventional splicing was observed in the absence of ER stress treatment (Figure 3c,d). These results strongly support the idea that OsbZIP50 mRNA is an OsIRE1 substrate. It is interesting to note that a bipartite stem-loop structure was also found in OsbZIP50 orthologue mRNAs from many plant species, including Physcomitrella patens (Figure S4). These findings suggest that the IRE1-mediated splicing mechanism of OsbZIP50 orthologues may be conserved among land plants.

Figure 3.

 OsIRE1 mediates an unconventional splicing of OsbZIP50 mRNA.
(a) Nucleotide and amino acid sequences around the splice sites in OsbZIP50 mRNA. Arrowheads and boxes in (a) and (b) indicate predicted cleavage sites and typical characteristics of IRE1 substrate, respectively.
(b) Comparison of predicted RNA structures around the splice sites of OsbZIP50, XBP1 and HAC1 mRNA.
(c) ER stress-induced splicing of OsbZIP50 mRNA in OsIRE1 knock-down (KD) and overexpression (OE) lines.
(d) Immunoblot analysis of OsbZIP50 in OsIRE1 KD and OE lines.

OsbZIP50s induces expression of the target gene using its N-terminal activation domain

To examine how OsbZIP50s functions as a transcription factor, it was expressed transiently in a suspension of cultured rice cells, followed by analysis of its activation of the OsBiP1 promoter, as OsBiP1 expression is known to be induced by ER stress (Oono et al., 2010; Wakasa et al., 2011). As shown in Figure 4(a,b), OsbZIP50s significantly activated the transcription from the OsBiP1 promoter, although OsbZIP50u also activated the promoter to some extent even in the absence of tunicamycin treatment. Yeast two-hybrid analysis showed that OsbZIP50s could function as a homodimer (Figure S5). Immunostaining with anti-HA antibody revealed that the C-terminally-tagged OsbZIP50s localized efficiently to the nucleus in the absence of ER stress treatment (Figures 4c and S6). By contrast, most of the C-terminally-tagged OsbZIP50u was detected outside of the nucleus (Figures 1d,e, 4c and S6). These results suggest that the spliced form of OsbZIP50 may be targeted to the nucleus, where it could be involved in the regulation of chaperone genes such as the BiPs.

Figure 4.

 OsbZIP50s induces gene expression via its N-terminal activation domain.
(a) Structures of proteins expressed by effector plasmids in experiments (b) and (c). HA and His were used as C-terminal tags.
(b) OsbZIP50 activity in Oryza sativa (rice) protoplasts. The reporter plasmid contained the OsBiP1 promoter linked to GUS. GUS activity was then determined relative to conditions without the effector plasmid and tunicamycin (Tm). Means ± SDs are shown, n = 3.
(c) Immunofluorescence microscopy of OsbZIP50 in transfected protoplasts. Anti-HA antibody and a secondary antibody labelled with green fluorescent dye were used. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Rice Oc cells are usually multinucleate. Scale bars: 10 μm.
(d) Structures of proteins expressed by effector plasmids for the experiments in (e) and (f). In these constructs, the indicated segments of OsbZIP50s were fused to the GAL4-binding domain.
(e) The activities of OsbZIP50s segments in rice protoplasts. A construct containing the GAL4-binding sequence (UAS)-TATA-GUS was used as a reporter plasmid. GUS activities were determined relative to conditions with the effector construct (a) alone. Means ± SDs are shown, n = 3.
(f) Activities of OsbZIP50s segments in yeast cells. The genomic UAS-TATA-HIS3 reporter was used and three independent colonies of each transformant were grown on SC medium with or without His.

The roles played by the N- and C-terminal domains of OsbZIP50s were investigated next. To characterize the OsbZIP50s transcriptional activation domain, the activities of the N- and C-terminal regions were examined in rice and yeast cells. It is notable that, in contrast to HAC1 and XBP1, the OsbZIP50 activation domain is located within the N-terminal region, and is not affected by OsIRE1-mediated splicing (Figure 4d–f). Analysis of transcriptional activation in yeast cells suggests that plant-specific factors are not likely to be required for activation (Figure 4f). These results indicate that the transcriptional activation domain is also retained in OsbZIP50u, irrespective of differences in nuclear translocation efficiency.

The OsbZIP50s C terminus contributes to high-efficiency nuclear localisation

To elucidate the role played by the alternative C-terminal region, full-length OsbZIP50s and C-terminally truncated OsbZIP50 (OsbZIP50deltaC), which does not have the predicted TMD, were expressed in cultured rice cells. Although some nuclear localisation of GFP-OsbZIP50deltaC fluorescence was observed in a previous study (Wakasa et al., 2011), the fluorescence outside of the nucleus was not scrutinized. In this study, to exclude artificial effects, we used untagged versions of OsbZIP50s and OsbZIP50deltaC. Immunostaining with anti-OsbZIP50 antibody demonstrated that OsbZIP50s clearly localized within the nucleus (Figure 5a), whereas OsbZIP50deltaC was primarily detected outside the nucleus, although some nuclear localisation did occur (Figure 5a). This result was further confirmed using constructs with C-terminal HA tags that showed the same localisation pattern as OsbZIP50deltaC without tags (Figures S6 and S7). Thus, it appears that the C-terminal region contributes to the highly efficient nuclear localisation of OsbZIP50s. Interestingly, sequence analysis indicated that some of the spliced OsbZIP50 orthologues in plants had lysine- and arginine-rich C-terminal regions (Figure S4b). These data imply that the alternative C-terminal regions serve as nuclear localisation signals (NLS), which typically contain lysine and arginine residues (Makkerh et al., 1996). To confirm this possibility, we observed the subcellular localization of OsbZIP50s with GFP attached to an alternative C-terminal end, corresponding to the one generated by splicing. The fluorescence signals were clearly observed in the nucleus (Figures 5c and S8), indicating that the C-terminal region of OsbZIP50s acts as an NLS. Taken together, these results strongly suggest that ER stress activation of OsbZIP50 depends not only on the level of the gene products accumulated but also on the efficiency of nuclear localisation. Based on these findings, it is possible that OsbZIP50 is activated by a different mechanism than previously proposed for HAC1 and XBP1 (Figure 5c).

Figure 5.

 The OsbZIP50s C-terminus contributes to high-efficiency nuclear localization.
(a) Fluorescence imaging of cultured rice cells transiently expressing untagged OsbZIP50s or OsbZIP50deltaC. OsbZIP50 was immunostained with anti-OsbZIP50 antibody. Nuclei were stained with DAPI. Scale bar: 10 μm.
(b) Fluorescence imaging of cultured rice cells transiently expressing sGFP fusion protein. sGFP was fused to the C-terminal region of OsbZIP50s, as presented in Fig. S4b. mCherry fused to the SV40 NLS was used as a nuclear localization control. Scale bar: 10 μm.
(c) Proposed model for the OsbZIP50 activation mechanism. Accumulation of unfolded proteins in the ER lumen activates the OsIRE1 cytosolic endoribonuclease domain. Activated OsIRE1 cleaves OsbZIP50 mRNA, inducing unconventional splicing that results in a frame shift and alteration of the OsbZIP50 C-terminal region. The translated product of unspliced mRNA (OsbZIP50u) has an activation domain (AD) in its N-terminal region, but it does not effectively localize to the nucleus. On the other hand, the spliced form (OsbZIP50s) contains a nuclear localization signal (NLS) within the C-terminal region, and OsbZIP50s is effectively localized to the nucleus, where it induces the expression of ER stress-responsive genes.

Expression of OsbZIP50 is affected by OsbZIP50 and ER stress sensors

In Arabidopsis, ER stress-induced expression of AtbZIP60 decreased slightly in an AtbZIP28 knock-out line (Liu and Howell, 2010a). Therefore, we examined whether or not OsbZIP39 and OsbZIP60 (Nijhawan et al., 2008), rice orthologues of AtbZIP17 and AtbZIP28, could induce OsbZIP50 gene expression. As shown in Figure 6(a–c), reporter assays demonstrated that the cytosolic N-terminal regions of OsbZIP39 and OsbZIP60 (OsbZIP39deltaC and OsbZIP60deltaC) activated both the OsBiP1 and OsbZIP50 promoters. Furthermore, OsbZIP50 promoter activity was enhanced by OsbZIP50s (Figure 6c). These findings suggest that OsbZIP50 amplifies the ER stress signal by inducing its own transcription. Next, OsbZIP50 expression levels were examined in OsIRE1 knock-down (KD) and overexpression (OE) lines to ascertain whether OsbZIP50 self-activation depends upon OsIRE1 activity. Lower OsbZIP50 transcript levels were observed in OsIRE1 KD lines, even under ER stress conditions (Figure 6d), whereas elevated transcript levels were observed in OsIRE1 OE lines, even in unstressed conditions (Figure 6e). These results indicate that OsbZIP50 expression is transcriptionally regulated in an IRE1-dependent manner, and they demonstrate the relationship between the ER stress sensors and OsbZIP50 (Figure 6f).

Figure 6.

 Transcriptional regulation of OsbZIP50 during endoplasmic reticulum (ER) stress.
(a) Structure of modified OsbZIP39 and OsbZIP60. Like AtbZIP17 and AtbZIP28, OsbZIP39 and OsbZIP60 contain a transmembrane domain (TMD) and a putative site-1 protease (S1P) site. The cytosolic N-terminal regions (OsbZIP39deltaC and OsbZIP60deltaC) were used as active forms in (b) and (c).
(b) Effector activity of OsbZIP39deltaC and OsbZIP60deltaC in rice protoplasts. The OsBiP1 promoter linked to GUS was used as a reporter. All GUS activity is reported relative to activity in the absence of the effector plasmid; means ± SDs, n = 3.
(c) Activities of effector constructs OsbZIP39deltaC, OsbZIP60deltaC and OsbZIP50s, in rice protoplasts. The reporter plasmid contains the OsbZIP50 promoter region linked to GUS, and GUS activity is reported relative to activity in the absence of effector plasmid; means ± SDs, n = 3.
(d, e) OsbZIP50 mRNA levels in OsIRE1 KD lines after ER stress treatment (d) and in OsIRE1 OE lines under non-stress conditions (e). Results of quantitative RT-PCR analysis in three biological replicates; means ± SDs.
(f) Proposed model for the relationship between ER stress sensors and OsbZIP50. ER stress induces active forms of OsIRE1, OsbZIP39 and OsbZIP60. Proteolytically activated OsbZIP39 and OsbZIP60 induce the expression of ER stress-responsive genes, including OsbZIP50. Following OsIRE1-dependent activation, the active form of OsbZIP50 induces its own expression, amplifying the ER stress signal. The active form of OsbZIP50 also induces expression of ER stress-responsive genes such as OsBiP1. Induced chaperones then assist with protein folding in the ER lumen. When ER stress is resolved, active forms of OsIRE1, OsbZIP39 and OsbZIP60 are reduced, which causes amplification of active OsbZIP50 to cease.

DNA microarray analyses were used to characterize the expression of ER stress-inducible genes that are highly dependent upon OsbZIP50 and OsIRE1. To investigate the expression patterns of many genes that are induced by ER stress as far as possible, OsbZIP50 KD (Figure S9) and OsIRE1 KD lines were treated with DTT, the effect of which is expected to be broader than that of Tm, for 2 h, and then genes exhibiting reduced expression by more than threefold, compared with wild type, were examined further. Table S1 shows the difference in expression levels (fold change) of representative ER stress-inducible genes that show a strong dependency upon both OsbZIP50 and OsIRE1. For example, no expression could be detected for certain OsBiP paralogues (OsBiP2, OsBiP3 and OsBiP4) and a SAR1B-like gene, in both the OsbZIP50 KD and OsIRE1 KD lines (Figure 7). These results further support a working model in which the activation of OsbZIP50 requires the expression of OsIRE1 to fulfil its function as a transcription factor. When Tm was used to induce ER stress more specifically (than DTT), by preventing glycosylation of newly synthesized proteins in the ER (Martínez and Chrispeels, 2003), quantitative RT-PCR analysis of representative ER stress-responsive genes showed that the ER stress-induced expression of the SAR1B-like gene, and all OsBiP paralogues, except for OsBiP1, is almost completely dependent upon OsbZIP50 (Figure 8). Furthermore, ER stress-induced expression of Ero1 (Onda et al., 2009) and Fes1-like also showed a strong dependency upon OsbZIP50, which is consistent with the data obtained by DNA microarray analysis. On the other hand, expression of OsBiP1, Calnexin, PDIL1-1, ERdj3B-like and SDF2-like showed only partial dependence upon OsbZIP50 (Figure 8). Similar expression levels were observed for OsBiP1 in OsbZIP50 KD and wild-type plants treated for 2 h, whereas the expression levels of this gene were significantly reduced in OsbZIP50 KD lines treated for 4 h (Figure 8). These results suggest that OsbZIP50 shares some target genes with other transcription factors, which include OsbZIP39 and OsbZIP60.

Figure 7.

 Dependency of endoplasmic reticulum (ER) stress-induced gene expression upon OsbZIP50 and OsIRE1. RT-PCR analysis of several ER stress-inducible genes listed in Table S1. Transcripts were extracted from seedlings of wild-type Oryza sativa (rice), OsbZIP50 KD and OsIRE1 KD lines, after 2 h treatment with dithiothreitol (DTT).

Figure 8.

 Dependency of endoplasmic reticulum (ER) stress-induced gene expression on OsbZIP50. mRNA levels of representative ER stress-induced genes in wild-type (open bars) and OsbZIP50 KD (grey bars) seedlings after treatment with tunicamycin (Tm). Quantitative RT-PCR analyses of three or four biological replicates; means ± SDs are shown.

Taken together, these data demonstrate that ER stress signals are transduced separately by multiple ER stress sensors (IRE1, OsbZIP39 and OsbZIP60), and then integrated for functional activation of OsbZIP50 via transcriptional and post-transcriptional regulation mechanisms. Thus, depending upon the signalling input, OsbZIP50, OsbZIP39 and OsbZIP60 may function in different temporal phases with partial overlaps.


Recently, Deng et al. (2011) and Nagashima et al. (2011) used computer and microarray analyses as initial approaches to discover that Arabidopsis AtbZIP60 mRNA is a substrate for plant IRE1, and showed that the active product is generated via IRE1-mediated RNA splicing. This study demonstrates that IRE1 also affects OsbZIP50. In addition, it provides new insights into the role played by the C-terminal region of OsbZIP50s, which results from a frame shift generated by IRE1-mediated splicing (Figure 5d). Although IRE1-mediated RNA splicing is required to generate the transcriptional activation domains of yeast HAC1 and animal XBP1, the alternative C-terminal domain of OsbZIP50 serves as an NLS. The transcriptional activation domain of AtbZIP60 is also located in the N-terminal region (Iwata and Koizumi, 2005), suggesting that domain organizations and functions are essentially the same between OsbZIP50 and AtbZIP60. Therefore, the C-terminal region of the spliced form of AtbZIP60 may also contribute to highly efficient nuclear localization. Thus, it appears that although plants (monocot and dicot), yeast and animals use fundamentally similar mechanisms for IRE1-mediated RNA splicing, the translation products may perform different functions. This interspecies diversity reflects flexibility in the molecular evolution of IRE1-mediated signalling. In addition, ER stress-induced expression of OsbZIP50 was affected by OsIRE1, and by the ER stress sensor proteins OsbZIP39 and OsbZIP60 (Figure 6f). This OsbZIP50 regulation mechanism is fundamentally similar to that of XBP1 in animals (Yoshida et al., 2001). Based upon apparent conservation in higher eukaryotes, this unconventional splicing mechanism may be extremely effective for providing strict control of signal transduction from stressed ER. Finally, the results also demonstrated that ER stress-induced genes show different dependencies upon OsbZIP50. Taken together, the results from this study provide a framework for understanding the early stages of the plant ER stress response.

It has not yet been determined whether or not dissociation of BiP from IRE1 leads to IRE1 cluster formation in plants, which is the case in animals (Ron and Walter, 2007; Hetz and Glimcher, 2009). In this study, overexpression of plant IRE1 leads to unconventional RNA splicing in the absence of ER stress treatment (Figure 3c,d). This effect is also observed in animals (Yoshida et al., 2001), which suggests that plants and animals share similar IRE1 activation mechanisms. Arabidopsis contains two IRE1 paralogues, whereas rice IRE1 is encoded by a single gene. In Arabidopsis, the expression of IRE1a is restricted to specific tissues such as the apical meristem, leaf margins, anthers and cotyledon, whereas IRE1b is expressed constitutively (Koizumi et al., 2001). This expression pattern suggests that some characteristics of IRE1 differ between Arabidopsis and rice. Future comparative studies between Arabidopsis, rice and other plant species will promote a greater understanding of IRE1 orthologue function.

DNA microarray and RT-PCR analyses have shown a wide variety of dependencies of ER stress-induced gene expression upon OsbZIP50. In contrast to OsbZIP39 and OsbZIP60, which are probably only activated by proteolytic processing, the activation of OsbZIP50 requires de novo protein synthesis. Therefore, the activation of OsbZIP39 and OsbZIP60 is predicted to respond more quickly than that of OsbZIP50. In addition, OsbZIP50 expression is assumed to be partially controlled by OsbZIP39 and OsbZIP60, as shown in Figure 6. These points suggest that although OsbZIP39 and OsbZIP60 may be predominantly involved in regulating gene expression during the early stages of ER stress, they may collaborate with OsbZIP50 at later stages. As OsBiP1 expression is induced by OsbZIP50, OsbZIP39 and OsbZIP60, the observed expression pattern of OsBiP1 in OsbZIP50 KD lines may reflect their distinct activation patterns. The presence of candidate UPR cis-elements in the promoters of ER stress-responsive genes analysed here are listed in Table S2. Of particular interest is the expression pattern of five OsBiP paralogues induced by ER stress. With the exception of OsBiP1, these genes exhibited very low levels of expression in unstressed rice plants, and induction appeared to be almost completely dependent upon OsbZIP50s (Figures 7 and 8). Such conspicuous differences in regulation suggest that the roles played by OsBiP1 may differ to those of the other OsBiP paralogues during the ER stress response. BiP orthologues are known to be key regulators of the ER stress response in other eukaryotes, and elucidation of their roles in rice will lead to a clearer understanding of the rice ER stress response.

In unstressed rice plants, we could not detect the endogenous OsbZIP50u protein, although some OsbZIP50u mRNA was detected. One possible explanation for this is that the sensitivity of the anti-OsbZIP50 antibody used for immunoblot analysis was insufficient for the detection of OsbZIP50u protein. Another possibility is that the accumulation of OsbZIP50u may be regulated translationally and/or proteolytically. In yeast HAC1, the intron that is spliced out by IRE1 serves as a translational repressor of unspliced HAC1 mRNA (Kawahara et al., 1997). On the other hand, OsbZIP50u protein was clearly detected under ER stress conditions, indicating that OsbZIP50u mRNA does not act as a translational repressor, as in the case of HAC1. The clear accumulation of OsbZIP50u protein under ER stress conditions suggests that it may play a significant role during the ER stress response. We demonstrate that exogenously expressed OsbZIP50u protein was localized to both cytosol and ER (Figure 1d,e), and was present in both soluble and microsomal fractions (Figure 1f,g), suggesting that OsbZIP50u may be a peripheral ER membrane protein rather than a transmembrane protein. The putative TMD of OsbZIP50u may act as a hydrophobic region for binding to the periphery of the ER membrane, as is the case for XBP1 (Yanagitani et al., 2009). A similar role for the predicted TMD of AtbZIP60 has been suggested by Nagashima et al. (2011). As reported for XBP1 (Yoshida et al., 2006; Yanagitani et al., 2009), OsbZIP50u and AtbZIP60 may be involved in the negative regulation of signalling or enhancement of IRE1-mediated splicing. Thus, the detailed characterization of OsbZIP50u will be an important subject for future investigations.

Recent reports have indicated relationships between the ER stress response and other environmental responses in plants (Che et al., 2010; Liu and Howell, 2010b). Elucidating the relationships between the ER stress response mediated by IRE1 and other biological processes will be a significant challenge for the future in plant science.

Experimental Procedures


The oligonucleotides used in this study are listed in Table S3.

Plant material and growth conditions

Oryza sativa L. cv. Kita-ake plants were grown on MS medium (1 × Murashige and Skoog salt mix, 3% sucrose, B5 vitamin and 0.25% Gelrite, adjusted with KOH to pH 5.8) at 25°C using a 16-h light/8-h dark cycle. For ER stress treatments, 8-day-old seedlings were incubated in liquid MS medium containing 2 mm DTT or 5 μg ml−1 tunicamycin. As negative controls for DTT and tunicamycin, equal volumes of water or DMSO (final concentration of 0.1%) were added, respectively.

Generation of transgenic plants

Knock-down lines for OsbZIP50 (OsbZIP50 KD) and OsIRE1 (OsIRE1 KD) were generated by RNA interference. For the expression of intron-containing hairpin RNA, OsbZIP50 and OsIRE1 cDNA regions were amplified by PCR and linked to the intron sequence for the rice aspartic protease gene (Kuroda et al., 2010). This construct was linked to the ubiquitin promoter and inserted into a modified pHGW binary vector (Wakasa et al., 2011). For OsIRE1 overexpression lines (OsIRE1 OE), OsIRE1 cDNA was amplified by PCR, linked to the ubiquitin promoter and inserted into a modified pHm43GW binary vector (Wakasa et al., 2011). Transgenic rice plants were generated by Agrobacterium-mediated transformation, and then the T1 and T2 generations were analysed.

RNA extraction and gene expression analysis

Total RNA was extracted from root tissues using the RNeasy Plant Mini Kit (Qiagen, Northern blot analysis was performed as described previously (Yamamoto et al., 2006). PCR-amplified fragments were used as DNA probes. For RT-PCR analysis, first-strand cDNA was synthesized from 0.8 μg of total RNA using the SuperScript III First-Strand Synthesis SuperMix for qRT-PCR (Invitrogen,, which includes both oligo (dT) and random hexamers, according to the manufacturer’s instructions. RT-PCR analysis was performed using 1/80 of the prepared cDNA. Quantitative and semi-quantitative RT-PCR was performed using SYBR® Premix Ex Taq™ (Takara, and TaKaRa Ex Taq™, respectively. For quantitative RT-PCR, the EF1a gene was used as an internal reference. For two-color microarray analyses, RNAs derived from young root tissues after treatment with 2 mm DTT for 2 h were labelled using an Agilent Low RNA Input Fluorescent Linear Amplification kit (Agilent Technologies, Aliquots of cRNAs (1 μg each) labelled with Cy3 or Cy5 were used for hybridization on a rice oligo microarray (44 k, custom-made; Agilent Technologies). To eliminate dye bias, dye-swap experiments using the same RNA samples were performed. Expression data were obtained from a single experiment. The microarray experiments were performed at the Rice Genome Resource Center in Tsukuba, Japan (

Preparation of recombinant proteins and generation of antibody

For the production of anti-OsbZIP50 antibody, the cDNA fragment encoding OsbZIP50 amino acids 1–216 was cloned into the pCold II vector (Takara). The recombinant protein containing an N-terminally fused hexahistidine tag was expressed in E. coli BL21 (DE3). The protein was then purified using an Ni Sepharose 6 Fast Flow column (GE Healthcare, and used to immunize rabbits. Serum was purified using the recombinant protein. Antibodies were used for immunoblot and immunostaining analyses. For comparison of recombinant protein and OsbZIP50 in plants, the full OsbZIP50u and OsbZIP50s open reading frames (ORFs), including stop codons, were cloned into the pET23d vector (Novagen, now EMD4Biosciences, and expressed in E. coli BL21 (DE3).

Immunoblot analysis

Protein extraction and immunoblot analyses were performed as described previously (Yamamoto et al., 2006). Total proteins were extracted from root tissues and fractionated on a 10% SDS-PAGE gel. CBB staining of SDS-PAGE was used as a loading control. For the immunoblot analysis of fractionated proteins, rice plants were treated with 2 mm DTT for 5 h. Root tissues were ground in liquid nitrogen and homogenized in extraction buffer (80 mm Tris-HCl, pH 8.0, 12% sucrose, 1 mm EDTA, 1 mm DTT and Complete Protease Inhibitor Cocktail; Roche, Cultured rice cells expressing OsbZIP50u containing C-terminal HA-tags, were treated with 0.1% DMSO or 5 μg ml−1 tunicamycin for 16 h. Following homogenization in extraction buffer, samples were centrifuged at 8000 g for 10 min, and then the supernatant was further centrifuged at 100 000 g for 1 h. The pellet was resuspended in an equal volume of SDS-PAGE sample buffer and used for immunoblot analysis. For the detection of OsbZIP50, HA tag, Calnexin and cFBPase, rabbit anti-OsbZIP50 antibody (this lab), mouse anti-HA antibody (Santa Cruz Biotechnology,, rabbit anti-Calnexin (Wakasa et al., 2011) and rabbit anti-cFBPase (Agrisera, were used, respectively.

Transient expression in cultured rice cell protoplasts

Rice Oc cells were used for transient expression assays, as described previously (Kawakatsu et al., 2009). For reporter assays, the 5′ flanking regions of OsBiP1(Os02g0115900) or OsbZIP50(Os06g0662700) were linked to GUS in the pBluescript II KS-based vector and UAS-TATA-linked GUS in the pUC19-based vector. Effector constructs were generated by cloning ORFs without stop codons (OsbZIP50u, OsbZIP50s, OsbZIP60deltaC; ORF1–242) into the vector Pro35S-ShΔ:2HA-6His (Kawakatsu et al., 2009) and linking segments of OsbZIP50s to GAL4 BD under the control of the 35S promoter. For the effector construct OsbZIP39deltaC, the ubiquitin promoter-driven OsbZIP39deltaC (ORF1–311), which contains a stop codon, was cloned into the binary vector p35SHPTAg7-43GW (Wakasa et al., 2006). A construct containing the LUC gene, under control of the ubiquitin promoter, was used as a transfection control. GUS activity was normalized against LUC activity. For immunostaining, rabbit anti-HA (Santa Cruz Biotechnology), mouse anti-HA (Santa Cruz Biotechnology), rabbit anti-OsbZIP50 and rabbit anti-Calnexin were used as the primary antibodies, and Alexa Fluor 488 goat anti-rabbit IgG (Invitrogen) and Alexa Fluor 350 goat anti-mouse IgG (Invitrogen) were used as secondary antibodies. For 4′,6-diamidino-2-phenylindole (DAPI) staining, samples were mounted with SlowFade Gold antifade reagent with DAPI (Molecular Probes, now Invitrogen). Epifluorescent microscopy was performed with a BX50 (Olympus, using NIBA, WU and WIG filter sets. Confocal microscopy was performed with an FV10i (Olympus). For the expression of untagged OsbZIP50s or OsbZIP50deltaC, the complete ORFs for OsbZIP50s and OsbZIP50deltaC (ORF1–217), including stop codons, were cloned into Pro35S-ShΔ:2HA-6His. For the expression of the sGFP fusion protein, the sequence encoding the C-terminal region of OsbZIP50s (indicated in Figure 5b) was fused to the C-terminus of sGFP and cloned into Pro35S-ShΔ:2HA-6His.

Identification of the transcriptional activation domain in yeast

Yeast strain YRG-2 was transformed with pGBKT7 (Clontech, plasmids containing DNA encoding segments of OsbZIP50s. The transformed yeast was grown on SC medium, as indicated in Figure S5.

Accession Numbers

Sequence data for the genes used in this study can be found in the Rice Annotation Project Database ( under the following accession numbers: OsbZIP50(Os06g0622700); OsIRE1(Os07g0471000); OsbZIP39(Os05g0411300); OsbZIP60(Os07g0644100); OsBiP1(Os02g0115900); OsBiP2(Os03g0710500); OsBiP3(Os05g0367800); OsBiP4(Os05g0428600); OsBiP5(Os08g0197700); Calnexin(Os04g0402100); PDIL1-1(Os11g0199200); ERdj3B-like (Os05g0156500); SDF2-like (Os08g0278900); Ero1 (Os03g0733800); Fes1-like (Os09g0512700); and SAR1B-like (Os06g0225000).


We thank Ms Y. Ikemoto, K. Miyashita, Y. Suzuki, M. Utsuno and Y. Yajima for technical assistance. This work was partly funded by the Ministry of Agriculture, Forestry and Fisheries of Japan (Genomics for Agricultural Innovation: research grant GMC0004 to FT) and by the Ministry of Education, Culture, Sports, Science and Technology of Japan (Grant-in-Aid for JSPS Fellows, 0803157 to YW; Grant-in-Aid for Young Scientists, 22688001 to TK).