Introduction

Proteins entering the secretory pathway of eukaryotic cells are modified and folded into their native structures in the endoplasmic reticulum (ER). Protein folding is an active process assisted by catalysts and chaperones, some of which are immunoglobulin heavy chain-lumen binding protein (BiP), calreticulin, calnexin and protein disulfide isomerase (PDI) (Gething 1999; Caramelo et al. 2004; Wilkinson and Gilbert 2004). This quality control system ensures that only properly folded proteins exit the ER (Trombetta and Parodi 2003). If protein modification and folding are disturbed, the accumulation of unfolded proteins causes ER stress. To recover, the cell has evolved an elegant mechanism, designated the unfolded protein response (UPR), which returns the protein folding process to its normal state (Harding et al. 2002).

During UPR, the state of protein folding in the ER is sensed and the expression of genes for protein folding enzymes and chaperones is stimulated, while secretory protein gene expression diminishes (Mori et al. 2000; Ron 2002; Martinez and Chrispeels, 2003; Trombetta and Parodi 2003). The transmembrane ER protein kinase, PERK, attenuates secretory protein mRNA translation (Harding et al. 2000; Harding et al. 2002). Finally, misfolded proteins are removed by conjugation to ubiquitin for proteasomal degradation (Travers et al. 2000; Bays et al. 2001; Kostova and Wolf 2003; Kamauchi et al. 2005). Together, these processes increase the folding capability and decrease the pool of nascent proteins that need to be folded in the ER (Mori et al. 2000).

The UPR signal transduction pathway has been extensively characterized in mammals and yeast. Some pathway homologs, such as BiP, IRE1, ATF6, calreticulin, and PDI, have been identified in plants. Excess unfolded proteins in the ER bind to BiP, which is released from the hybrid lumen sensor, protein kinase-ribonuclease, IRE1, leading to its dimerization and activation (Bertolotti et al. 2000; Koizumi et al. 2001; Noh et al. 2002; Okushima et al. 2002). In animals, IRE1 splices the X-Box binding protein (XBP1) mRNA (Yoshida et al. 2001), allowing synthesis of XBP1, a bZIP transcription factor (Lee et al. 2003; Kanemoto et al. 2005). In mammals, the release of BiP can also activate the ATF6 transcription factor, which mediates activation of UPR regulated genes (Lee et al. 2002). An ATF6-like activating transcription factor, AtbZIP60, has recently been found in plants (Iwata and Koizumi 2005). However, the XBP1 transcription factor and the translational regulator, PERK (Harding et al. 2000), have not been found in Arabidopsis, raising questions about the nature of the components and the role of translation in the UPR signaling pathway in plants (Kamauchi et al. 2005). Recently, it was found that salt stress signaling utilizes components of ER stress signaling (Liu et al. 2007a).

A key downstream component of UPR, which is present in all eukaryotes with sequenced genomes, is the highly abundant ER oxidoreductase, PDI (EC 5.3.4.1), PDI catalyzes thiol-disulfide interchange reactions that form, break, and isomerize disulfide bonds in substrate proteins (Noiva 1999). The classical PDI has four thioredxoin-like domains, a–b–b′–a′ (Darby et al. 1996; Gruber et al. 2006). The a and a’ are the thioredoxin catalytic domains that contain the active site, PWCGHCK (Edman et al. 1985). The b and b′ are two homologous domains with folded structures similar to thioredoxin, but without the active site or thioredoxin sequence homology (Lucero and Kaminer 1999). Recently, a genome wide analysis of gene expression in response to UPR in Arabidopsis found some PDI genes to be upregulated (Kamauchi et al. 2005). However, despite the importance of PDIs in UPR, the systematic analysis of PDI gene expression in response to ER stress has not, to date, been conducted in plants.

The study of PDIs in plants requires a clear definition of the features that specifically constitute a PDI. In this report, we used stringent bioinformatic criteria to define and characterize a 12-member Arabidopsis PDI gene family (AtPDI). We then analyzed the expression of AtPDI genes in different tissues, in response to chemically induced ER stress and in UPR-signaling mutants. BiP2 gene expression served as an internal control verifying UPR activation. We used three chemicals, dithiothreitol (DTT), tunicamycin (Tm) and beta-mercaptoethanol (β-Me), which are known to induce the accumulation of unfolded proteins and activate ER stress and UPR (Martinez and Chrispeels 2003; Noh et al. 2003; Kamauchi et al. 2005). The reducing agents, DTT (dithiol) and β-Me (monothiol), break disulfide bonds and inhibit their formation (Jämsä et al. 1994; Back et al. 2005). Tm inhibits asparagine N-linked glycosylation (Takatsuki et al. 1971), which is a prerequisite for glycoprotein folding (Dorner et al. 1990). In Arabidopsis, DTT and Tm induce the expression of genes for protein folding enzymes and decrease secretory protein gene expression (Martinez and Chrispeels 2003). Here we found that the expression of a sub-group of six AtPDI genes was increased in response to ER stress. The transcription inhibitor, actinomycin D, significantly attenuated this response. We identified null mutants, Atire1-2 and Atbzip60, and investigated AtPDI gene expression in these mutants exposed to ER stress. We found that AtbZIP60 modulates the up-regulation of a sub-group of AtPDI genes in response to UPR. This work paves the way for extended studies of a gene family of fundamental importance to all eukaryotes, Besides providing insight into the UPR in plants, this work also highlights evolutionary differences in the UPR signaling between plants, yeast and mammals.

Materials and methods

Plant material and growth conditions

For chemical treatments, wild type Arabidopsis thaliana L. Heynh (ecotype Columbia) and T-DNA insertion mutant seeds were surface-sterilized, and placed in the dark for 2 days at 4°C. They were germinated in 50 mL of liquid Murashige and Skoog medium, containing 3% Suc (w/v), in a 16-h-light/8-h-dark cycle at 23°C with constant shaking (50 rpm) for 7 or 16 days. Arabidopsis seeds were also grown in soil in the same environmental conditions as liquid-cultured plants for 16 days for T-DNA insertion genotyping and 2 months for tissue dissection.

Chemical treatments

One-week-old liquid cultured Arabidopsis plantlets were treated with 5 μg/ml Tm or 20 mM β-Me for 0, 2, or 5 h as described by Martinez and Chrispeels (2003). Sixteen-day-old liquid-cultured plantlets were treated with 10 mM DTT for 0, 2, or 5 h as described by Martinez and Chrispeels (2003). In some experiments, 9-day-old liquid cultured plants were pre-treated with actinomycin D at final concentration of 25 μg/ml for 3 h. Then the plants were subjected to 5 μg/ml Tm for 0, 2 and 5 h. The control for Tm treatment was a final concentration of 1% DMSO, which was the solvent for Tm. After the chemical treatments, plants were harvested and total RNA isolated.

RNA isolation and RT-PCR

Total cell RNA was isolated as described (Hoffer and Christopher 1997) from whole seedlings or from roots (R), stems (S), rosette leaves (RL), cauline leaves (CL), and flowers (F) of 2-month-old Arabidopsis plants grown on soil. For reverse transcription PCR (RT-PCR), RNA was treated with RNase-free DNase (Promega, Madison, WI, USA) and used to produce the first-strand cDNA with M-MLV reverse transcriptase and oligo dT (Promega). 50 ng of total RNA was used to amplify the transcripts of each AtPDI, actin, calreticulin, and AtbZIP60 gene by RT-PCR for the indicated number of cycles in each figure. The cDNA mixtures used for PCR were the Bio-X-Act Short Mix (Bioline Inc., Taunton, MA, USA). The gene-specific primers used in the study corresponded to exon–exon junctions to avoid amplification of genomic DNA, and they are listed in Table 1. The PCR products were sequenced to confirm their identity.

Table 1 The gene-specific primers used this study
Full size table

RNA gel blot analysis

RNA was separated and transferred to nylon membrane as described (Chun et al. 2001). The 32P-labeled cDNA probe was made using a DECAprimer II labeling kit (Ambion, TX, USA). The cDNA inserts (Atpdi2 to Atpdi11) were generously provided by Dr. Danon (Weizmann Institute of Science). Hybridization and radioactive signal detection were performed as described (Christopher and Mullet 1994).

Genotyping of T-DNA insertion mutants

The T-DNA insertion in each mutant was detected by PCR using a gene specific primer and a T-DNA left border primer, LBa1. The accurate insertion positions were determined by sequencing each PCR product using the LBa1 primer. The genotypes for T-DNA insertion were determined by PCR using two gene-specific primers.

Computer analysis of sequences

Clustal W (http://align.genome.jp/) was used for the multiple sequence alignments and for pairwise analysis of the phylogenetic relationship of the Arabidopsis PDI proteins. The pairwise analysis used a gap penalty of 10 and a gap extension penalty of 0.05 (http://align.genome.jp/). Based on the multiple sequence alignment of the 12 AtPDIs, an unrooted phylogenetic tree was constructed using “Neighbour Joining” algorithm (Saitou and Nei 1997) wthin the Phylo-win program (Galtier et al. 1996). A bootstrap analysis with 1,000 replicates was performed and the bootstrap values are shown in Fig. 1a. Signal P3.0 (http://www.cbs.dtu.dk/services/SignalP/) was used to predict the presence of signal peptides. TMHMM ver. 2.0 (http://www.cbs.dtu.dk/services/TMHMM-2.0/) was used to determine transmembrane domains. The subcellular localizations of AtPDIs were predicted by using TargetP1.1 (http://www.cbs.dtu.dk/services/TargetP/) and pSORT (http://wolfpsort.org/).

Fig. 1
figure 1

The phylogenetic relationship and primary structure of 12 AtPDIs. a The phylogenetic and pairwise analyses of predicted PDI gene products are shown resulting in three sub-groups. The bootstrap values are circled. b The 12 AtPDIs were grouped based on sequence and domain similarity. The predicted number of amino acids encoded by each cDNA is shown in parenthesis at the C-terminus. SP signal peptide, Trx thioredoxin catalytic domain, TMD transmembrane domain; a and a′, full thioredoxin domains with catalytic site and fold region; b and b′, thioredoxin fold domains lacking catalytic site

Full size image

Results

Sequence analysis of AtPDIs

Using the classical human PDI (HsPDI) as a standard (Darby et al. 1996; Tasanen et al. 1998), we defined 12 members of the PDI family, ranging in size from 361 to 597 amino acids (Fig. 1a, b; Table 2). Their sequences were aligned with HsPDI (Supplemental Fig. 1). Ten PDIs (AtPDI1, AtPDI2, AtPDI3, AtPDI4, AtPDI5, AtPDI6, AtPDI8, AtPDI9, AtPDI10, and AtPDI11) and HsPDI were anchored at a conserved thioredoxin domain proximal to the N-termini, designated classical domain a (Fig. 1b). The second thioredoxin domains of HsPDI, AtPDI1, AtPDI2, AtPDI3, AtPDI4, AtPDI5 and AtPDI6 were collinear and designated classical domain a′ (Fig. 1b). As with HsPDI, the fold domains b and b′ are located between domains a and a′ in these six plant sequences (Fig. 1b). Thus, HsPDI, AtPDI1, AtPDI2, AtPDI3, AtPDI4, AtPDI5 and AtPDI6 share the standard PDI domain architecture: a–b–b′–a′. The second thioredoxin domains of AtPDI9, AtPDI10, AtPDI11, and the single corresponding domains of AtPDI7 and AtPDI12, aligned together. These thioredoxin sites are closer to the N-termini in AtPDI9, AtPDI10, AtPDI11 than in AtPDI1, AtPDI2, AtPDI3, AtPDI4, AtPDI5, and AtPDI6. No homologous b or b′ fold domain could be found in AtPDI9, AtPDI10, and AtPDI11. Recently, the orthologs of Arabidopsis AtPDI11 in soybean, GmPDIS-1 and GmPDIS-2, were identified to have the domain organization of a–a′-ERp29c, without b or b’ domains (Wadahama et al. 2007).

Table 2 Genbank identifications, AtPDI features, predicted polypeptide sizes, and predicted presence of signal peptide (s.p.) or other targeting sequences such as chloroplast (chl), mitochondrion (mit), plasma membrane (p.m.) or other
Full size table

Ten PDIs are predicted to have a signal peptide at the N-termini (scores 67–100% probability, Table 2; Fig. 1b), which indicates that they are translocated over the ER membrane into the lumen and possibly further into the secretory pathway (Bendtsen et al. 2004). However, AtPDI1, AtPDI2, AtPDI4, and AtPDI10 also have secondary predictions for the chloroplast (Table 2). AtPDI7 and AtPDI8 lack signal peptides altogether and have scores (TargetP, pSORT) for entering the “other” subcellular locations or mitochondria (Table 2). AtPDI8 has a high score for the plasma membrane. The C-terminal ER retention signal, KDEL, is conserved in AtPDI1, AtPDI2, AtPDI3, AtPDI4, AtPDI5, AtPDI6, AtPDI9, and is partially conserved in AtPDI10 (KDDL), further suggesting that these are ER proteins. AtPDI8 has a KKXX-like tetrapeptide, KKED, at its C-terminus, which serves as an ER retention signal for a membrane protein (Vincent et al. 1998). Three of the proteins are predicted to have transmembrane domains (TMD). AtPDI8 has a TMD in the C-terminus, whereas AtPDI7 and AtPDI12 have them in both the N-and C-termini.

Pairwise comparisons were used to identify five pairs of closely related PDIs (Fig. 1a). The pairs are AtPDI1 and AtPDI2 (54% identity), AtPDI3 and AtPDI4 (76% identity), AtPDI5 and AtPDI6 (79% identity), AtPDI7 and AtPDI12 (73% identity) and AtPDI9 and AtPDI10 (76% identity) (Fig. 1). This suggests that the genes encoding the each pair of proteins may have arisen from a recent duplication event. The 12 PDIs fall into three divergent subgroups (Fig. 1a). AtPDI9 and AtPDI10, with AtPDI11 form the subgroup I, sharing the similar domain organization, but AtPDI9 and AtPDI11, AtPDI10 and AtPDI11 share overall low identity. Within subgroup II, the three PDI pairs all have the standard PDI domain structure a–b–b′–a′. AtPDI8 only has one thioredoxin domain and the standard b and b′ domains, and resides outside all three groups (Fig. 1a). The subgroup III consists of the PDI pair (AtPDI7 and AtPDI12), which have a single thioredoxin domain and two common TMDs. A gene for a thirteenth protein (At1g07960) containing a thioredoxin domain at the same position as AtPDI7 and AtPDI12 was found, but it was completely truncated at the c-terminus. This sequence was considered too divergent from the PDI family to be included, although it was a member of the thioredoxin superfamily.

The relative abundance of the AtPDI transcripts in seedlings and in different mature tissues

The relative transcript abundance of the 12 AtPDI genes, relative to actin mRNA as a reference control, was examined using RT-PCR of total cell RNA from whole 7-day-old seedlings (Fig. 2a). AtPDI11, AtPDI5, AtPDI9, AtPDI2, and AtPDI6 have moderately high transcript abundance (Fig. 2a), whereas the abundance of AtPDI3, AtPDI4, AtPDI12, and AtPDI7 mRNA was lower, with AtPDI3 being the lowest. BLAST searches for each AtPDI gene in the NCBI Arabidopsis EST database confirmed these observations (Fig. 2b). The numbers of ESTs for AtPDI11, AtPDI5, AtPDI2, and AtPDI6 were high, with 100–101 ESTs per gene. AtPDI9 has 63 ESTs, while AtPDI3, AtPDI4 and AtPDI12 had lower numbers (Fig. 2b) with AtPDI3 being the lowest (5 ESTs). We also performed RT-PCR analysis on total cell RNA from roots, stems, rosette and cauline leaves, and flowers of a mature plant (Fig. 2c). AtPDI genes were expressed in all organs, with the highest expression in flowers. This pattern is similar to that of the ER chaperon gene for calreticulin. In non-floral tissues, the AtPDI transcript levels did not differ notably.

Fig. 2
figure 2

The relative abundance of AtPDI transcripts within seedlings and among tissues. a The relative abundance of AtPDI transcripts (1–12) and for actin (Act) in 1-week-old liquid-cultured plantlets was determined by RT-PCR analysis for the indicated number of cycles. b the number of EST clones of each PDI gene in NCBI EST database is shown. c Levels of AtPDI mRNAs in different Arabidopsis tissues of 2 month-old plants determined by RT-PCR analysis. R root, S stem, RL rosette leaf, CL cauline leaf, F flower. Actin was used as internal control and the transcript levels for the ER stress genes, encoding calreticulin and AtbZIP60, were also analyzed

Full size image

The expression of AtPDI genes in response to ER stress

To study the expression of AtPDI genes in response to ER stress, we treated seedlings separately with three chemical inducers of ER stress: dithiothreitol (DTT), beta-mercaptoethanol (β-Me), and tunicamycin (Tm). The mRNA levels for the BiP2 gene, which is a known internal molecular marker of ER stress, and the AtbZIP60 gene, which is a novel transcription factor believed to be involved in ER stress, are induced by the chemical treatments (Fig. 3a). Similarly, the expression of AtPDI1, AtPDI5, AtPDI6, AtPDI9, AtPDI10, and AtPDI11 genes are increased in plants treated with the three chemicals (Fig. 3a). Because DTT (β-Me) and Tm induce ER stress by different mechanisms, the increases in mRNA abundance of AtPDI genes were very likely to be due to ER stress rather than nonspecific effects of the chemicals. In contrast, these reagents either slightly repress or have minimal effects on the expression of AtPDI2, AtPDI7, and AtPDI8. Due to low abundance, the mRNAs of AtPDI3, AtPDI4 and AtPDI12 were barely detected by RNA gel-blot analysis. Thus, RT-PCR was performed to determine their expression (Fig. 3b). Their transcript levels decreased, or had minimal changes, upon chemical treatments.

Fig. 3
figure 3

Levels of the AtPDI mRNA in response to treatment with ER-stress inducing chemicals, DTT, β-Me, and Tm, as determined by a, RNA gel-blot analysis or b, RT-PCR. The liquid-cultured plants were treated with the three chemicals for 0, 2 and 5 h. Analysis of transcripts for the BiP2 and AtbZIP60 genes were used as positive controls for the induction of UPR in response to ER stress. The analysis of rRNA was used as a loading control

Full size image

ER stress increases AtPDI gene transcription

To study whether the induction of the AtPDI genes was regulated at the transcriptional level, we applied actinomycin D to block gene transcription prior to inducing ER stress with Tm. The RT-PCR results revealed that, in the presence of transcription inhibitor, the induction of BiP2, AtbZIP60, AtPDI1, AtPDI5, AtPDI6, AtPDI9, AtPDI10, and AtPDI11 genes was significantly (>70%), but not completely, impaired. The effect was most notable at 2 h of Tm treatment, while the induction of some PDIs recovered partially at 5 h (Fig. 4). This experiment indicates that the increase in AtPDI mRNA levels in plants exposed to the chemical inducers of ER stress occurred in part at the transcription level.

Fig. 4
figure 4

Determination of the effects of the transcription inhibitor, actinomycin D, on the ability of Tm to induce AtPDI gene expression. RT-PCR analysis was conducted to measure the mRNA levels for key representative AtPDI genes, and the control genes, BiP2, AtbZIP60 and actin, in plants exposed to Tm to induce ER stress, after being pre-treated with actinomycin D (ActD)

Full size image

The expression of AtPDI genes in AtIRE1-2 mutants under ER stress

To determine whether the expression of the AtPDI genes is controlled by putative ER stress signal transduction pathway components, AtIRE1-1 and AtIRE1-2, we endeavored to identify their homozygous T-DNA insertion mutants. For the AtIRE1-1 gene, however, only heterozygous mutant plants were recovered after repeated selfing of heterozygous parents and followed by molecular diagnoses of the ire1-1 mutant (SALK_018150) (supplemental Fig. 2). We used RT-PCR to compare the mRNA levels of the AtIRE1-1 gene in the ire1-1 heterozygote relative to wild-type plants (Supplemental Fig. 2d). No difference in AtIRE1-1 transcript levels was observed.

In contrast to AtIRE1-1, we identified three independent homozygous T-DNA insertion lines of the AtIRE1-2 gene. They were ire1-2-1 (SALK_010332), ire1-2-2 (SALK_018112), and ire1-2-3 (SALK_002316). Sequence analysis of PCR products revealed T-DNAs were inserted in different locations within the AtIRE1-2 gene (Fig. 5a). The results of RT-PCR using AtIRE1-2 specific primers confirmed that both ire1-2-2 and ire1-2-3 are null mutants (Fig. 5b). Although ire1-2-1 has transcripts, the levels are much lower level than in wild type (Fig. 5b). To determine the impact of the AtIRE1-2 mutant on the expression of AtPDI genes in plants exposed to ER stress, wild type and the three homozygous mutants were treated with Tm (Fig. 5c). Tm treatment induced the expression of six AtPDI genes and the two positive controls, BiP2 and AtbZIP60, to comparable levels in the mutant and wild type plants. Therefore, the disruption of AtIRE1-2 gene did not interfere with Tm-induced AtPDI gene expression.

Fig. 5
figure 5

The effect of disruption of the AtIRE1-2 gene on the expression of AtPDI genes in plants exposed to Tm treatment. a the schematic representation of AtIRE1-2 gene structure and T-DNA insertion sites of three homozygous mutant lines (ire1-1, ire1-2, and ire1-3). Exons are represented by grey boxes, while the UTRs are indicated by darker boxes. Introns are represented by lines. The positions of the translational start (ATG) and stop (TAA) codons are indicated. In ire1-2-1, the T-DNA is inserted at 148 bp upstream from the translation start codon, in the 5′UTR. Ire1-2-2 has the T-DNA within the fourth intron, at +2018 bp (a in ATG is +1). The T-DNA in ire1-2-3 is inserted at +2329 bp, in the fifth intron (Panel A). b the detection of AtIRE1-2 mRNA in three homozygous mutant lines and wild type by RT-PCR. The actin gene was used as an internal control. c RNA gel-blot analysis of AtPDI gene expression in the three AtIRE1-2 mutant lines and wild-type plants exposed to Tm treatment for 0, 2 and 5 h. Transcripts for the BiP2 and AtbZIP60 genes were analyzed as indicators of ER stress

Full size image

The Atbzip60 mutant affects AtPDI gene expression in response to ER stress

AtbZIP60 is reported to be a transcription factor involved in ER stress signaling in plants (Iwata and Koizumi 2005). To test whether AtbZIP60 influences AtPDI gene expression in response to ER stress, we identified a homozygous T-DNA insertion in the AtbZIP60 gene (Fig. 6a). To verify that AtbZIP60 gene expression was abolished in the Atbzip60 mutant, RT-PCR analyses were performed on total RNA from mutant and wild type plants using two sets of gene-specific primers (Fig. 6b). No AtbZIP60 mRNA was detected in the mutant when the forward primer, LP1 (upstream to the T-DNA insertion site), and reverse primer, RP, were used. However, a very faint band was obtained if the primer, RP, and the primer, LP2 (downstream to the T-DNA inserted site), were used (Fig. 6b). This indicated that a truncated AtbZIP60 mRNA was present in the mutant, although the mechanism of transcription is unknown. Nevertheless, the experiment confirmed that AtbZIP60 expression was severely impaired in the mutant (Fig. 6b).

Fig. 6
figure 6

Analysis of key representative AtPDI genes, and the control genes, BiP2, AtbZIP60 and actin, in the homozygous bzip60 SALK insertion mutant (bzip60Δ) and wild-type plants under ER stress. a The schematic representation of AtbZIP60 gene (At1g42990) and SALK T-DNA insertion site. Exons are in grey rectangles and intron is a black line. A black rectangle indicates the 3′ untranslated region. The positions of three AtbZIP60 gene primers LP1, LP2, and RP are also indicted. LP1 is upstream to the T-DNA insertion site, whereas, LP2 is downstream to the T-DNA insertion site. B, RT-PCR analysis to detect AtbZIP60 mRNA in the bzip60Δ insertion mutant and wild-type plants using different combinations of primers and for 25 and 50 PCR cycles. The actin gene was used as an internal control. c RNA gel-blot analysis of AtPDI gene expression in the Atbzip60 mutant and wild-type plants exposed to Tm or β-Me treatment for 0, 2 and 5 h. BiP2 gene was used as an indicator of ER stress. d Quantitation of the RNA gel-blot analyses of representative AtPDI genes and the BiP2 gene in the bzip60Δ mutant relative to wild-type plants exposed to 5 h Tm or β-Me treatments. The ratio of RNA levels in the mutant divided by the wild-type were graphed. The actin gene was used as an internal control and set as a value of one (dashed line). Values below the dashed line indicate a decrease in mRNA levels in the mutant. Values are the means ± standard errors of two independent experiments done in duplicate

Full size image

Wild-type and the Atbzip60 mutant were treated with Tm or β-Me, then the expression of key AtPDI genes was analyzed via RNA-gel blot hybridization (Fig. 6c). The expression of the AtPDI6, AtPDI9, AtPDI10, and AtPDI11 genes decreased in the Atbzip60 mutant relative to wild type under ER stress treatments, especially at 5 h Tm and β-Me treatments. Thus, the data of 5hr treatment was quantitated and represented as the ratio of RNA levels from mutant compared to wild type plants (Fig. 6d). In particular, the AtPDI9 mRNA levels in the mutant were 56% after the 5 h Tm treatment and ~34% after the β-Me treatment relative to wild type mRNA levels (Fig. 6c, d). However, Tm and β-Me has moderately opposite effects on AtPDI1 and BiP2 expression differed in the Atbzip60 mutant relative to wild type (Fig. 6c, d). Also, there was a ~20% increase in AtPDI2 and AtPDI8 expression in the mutant relative to wild type.

Discussion

Variation in the primary structure and organ expression of the PDI family

The Arabidopsis PDI gene family has been proposed to have a varied number of members (Meiri et al. 2002; Lemaire and Miginiac-Maslow 2004; Buchanan and Balmer 2005). Recently, a comprehensive genome-wide search and phylogenetic analysis of proteins that contain thioredoxin domains (thioredoxin superfamily), generated a clade of 22 Arabidopsis PDI-like proteins (Houston et al. 2005). However, two of them are quiescin-sulfhydryl oxidases and seven are adenosine 5′-phosphosulfate reductase-like proteins. Here we used a different approach, emphasizing the classical human and yeast PDI domain structure, a–b–b′–a′ (Wilkinson and Gilbert 2004), to identify 12 PDI genes in Arabidopsis, and gave weight to amino acid sequence identity beyond the thioredoxin domain, with allowances for minor divergence (Fig. 1; Table 2, Supplementary Fig. 1).

New putative functional domains were identified in the analysis. The12 PDIs differed in polypeptide length, presence of signal peptide and ER retention signal, and the number and positions of TRX and TMD domains. However, key pairs of AtPDIs were very conserved at the amino-acid level. They were AtPDI1 and AtPDI2, AtPDI3 and AtPDI4, PDI5 and PDI6, AtPDI7 and AtPDI12, and AtPDI9 and AtPDI10 (Fig. 1). PDIs are believed to have evolved from an ancestral thioredoxin enzyme (Buchanan 1991; McArthur et al. 2001), and the diverse domain structures generated through duplications and deletions (Kanai et al. 1998; Meiri et al. 2002). Individual PDIs in other eukaryotes vary in their target substrates (McArthur et al. 2001; Wilkinson and Gilbert 2004). The multiplicity of conserved PDIs in Arabidopsis, and the structural differences in divergent members, suggest that they could have both specialized and overlapping functions to facilitate adaptation to new biochemical needs or environments during evolution. Alternatively, PDI redundancy provides for genetic compensation, ensuring that mutation or loss of one gene will not have a deleterious impact on the plant. Finally, the overall high level of all AtPDI mRNAs in flowers suggests that they play important roles in reproductive development.

ER stress and the UPR regulate the transcription of a sub-group of AtPDI genes

We identified six genes (AtPDI1, AtPDI5, AtPDI6, AtPDI9, AtPDI10 and AtPDI11) that were up-regulated by UPR in response to well-documented chemical inducers of UPR: DTT, Tm, and β-Me (Martinez and Chrispeels 2003; Iwata and Koizumi 2005; Kamauchi et al. 2005). The BiP2 gene and, in some experiments, the calreticulin gene, served as internal positive controls to confirm that the chemicals were indeed inducing UPR. It is likely that these genes are up-regulated to increase the protein folding capacity in the ER. Such general folding catalysts will have a broader range of substrates, as do BiP and the classical PDI enzyme (Haugstetter et al. 2005). A sub-group of AtPDI genes (AtPDI2, AtPDI7, AtPDI12, AtPDI8) were not induced by UPR. There is precedence for a PDI containing an ER-retention signal leaving the ER and entering the chloroplast in Chlamydomonas (Levitan et al. 2005). The Arabidopsis AtPDI1 and AtPDI2 are most closely related to the Chlamydomonas sequence (RB60-PDI). Interestingly, despite their strong sequence similarity between AtPDI1 and AtPDI2, they have different responses to UPR. They may have functionally diverged from each other in terms of gene regulation and protein function.

The use of the transcription inhibitor (actinomycin D) and the promoter disruption mutant demonstrated that ER stress activated transcription of the six PDI genes. The inhibitory effect of actinomycin D was not 100%, especially in the longer Tm treatment (5 h, Fig. 4). The concentration of Actinomycin D may be insufficient to block transcription completely, or its activity diminishes during treatment. AtPDI2 and AtPDI7 were not induced by Tm, but there was slight changes in mRNA levels within experimental variation.

The AtbZIP60 is partially required for the induction of AtPDI genes upon ER stress

We sought to determine whether the induction of AtPDI gene transcription was mediated by UPR signal pathway components homologous to those in mammals. To test whether disruption of UPR signaling homologs, AtIRE1-1 or AtIRE1-2 (Koizumi et al. 2001), can affect the up-regulation of AtPDIs, we endeavored to identify their homozygous mutants. However, no homozygous AtIRE1-1 gene T-DNA mutants were present in seed from selfed heterozygous parents. Atire1-1 homozygosity could be lethal, and that AtIRE1-1 is an essential gene, however, more evidence is needed to investigate the possible lethality during reproductive development. Coincidently, the deletion of IRE1-α in the mouse produced embryonic lethality (Urano et al. 2000b). In contrast to AtIRE1-1, we were able to obtain AtIRE1-2 homozygous null mutants. The induction of AtPDI and BiP2 gene expression was not affected by the disruption of the AtIRE1-2 gene, which is similar to the response of the BiP gene in IRE1-β knockout mice (Bertolotti et al. 2001; Urano et al. 2000a). Therefore, AtIRE1-2 might not be involved in regulating AtPDI gene expression in response to ER stress in plants, or it is possible that other genes or pathways complement the mutation of the AtIRE1-2. In contrast, the Arabidopsis ATF6-like protein, AtbZIP60 (Lee et al. 2002; Iwata and Koizumi 2005) was involved in the up-regulation of at least four AtPDI genes in response to UPR. The expression of these genes was partially decreased in the Atbzip60 mutant. We hypothesize that the 20% increase in AtPDI2 and AtPDI8 gene expression in the mutant indicated that the AtbZIP60 might also control the expression of repressors of gene expression. The partial and contrasting effects suggest that additional complex UPR signaling components complement activation of PDI gene expression in the Atbzip60 mutant exposed to ER stress. Recently, a novel ATF6-like transcription factor, AtbZIP28 was identified to mediate UPR in Arabidopsis. It appears to be more involved in the early phases of Tm-induced ER stress signaling than AtbZIP60 (Liu et al. 2007b).

Overall, we found that the expression of a sub-group of six AtPDI genes was increased in response to ER stress, and the up-regulation was at least partially modulated by AtbZIP60. Our work not only paves the way for extended study of the PDI gene family, but also provides insight into UPR in plants and its relationship with UPR signaling in mammals and yeast. Further research is needed to define the factors and their interaction in the UPR signaling pathway regulating the AtPDI genes. The promoters for the UPR-up-regulated AtPDI genes will serve as useful tools to begin elucidating the transcriptional regulatory network.